DO
158712
CQ
OSMANIA UNIVERSITY LIBRARY
CaU No. ^f/.
Author
This book should be returned on or before the date last mavKed below.
INORGANIC COLLOID
CHEMISTRY
BY PROFESSOR HARRY B. WEISER
INORGANIC COLLOID CHEMISTRY
Vol I The Colloidal Elements.
389 pages 6 by 9 54 figures Cloth
Vol II. The Hydrous Oxides and Hydroxides.
429 pagps 6 by 9 70 figures Cloth
Vol. III. The Colloidal Salts.
473 pages 6 by 9 71 figures Cloth
EDITED BY PROFESSOR WEISER
COLLOID SYMPOSIUM ANNUAL
Papers presented at the Seventh Symposium on Col-
loid Chemistry, Johns Hopkins University, June,
1929 300 pages 6 by 9. 127 figures Cloth
PUBLIbHED BY
JOHN WILEY & SONS, INC.
INORGANIC COLLOID
CHEMISTRY
BY
HARRY BOYER WEISER
Professor of Chemistry at The Rice Institute
VOLUME III
THE COLLOIDAL SALTS
NEW YORK
JOHN WILEY & SONS, INC.
LONDON: CHAPMAN & HALL, LIMITED
1938
COPYRIGHT, 1938, BY
HARRY BOYER WEISER
All Right* Reserved
This book or any part thereof must not
be reproduced in any form without
the written permission of the publisher
PRINTED IN U 3. A.
PRESS OF
BRAUN WORTH & CO INC.
BUII_DKRS OF BOOKS
BRIDGEPORT. CONN
PREFACE
This book is the last in a three-volume treatise on inorganic col-
loid chemistry which deals with the colloidal behavior of the ele-
ments and their inorganic compounds. Like the two preceding books,
The Colloidal Elements (1933) and The Hydrous Oxides and Hy-
droxides (1935), this volume, entitled The Colloidal Salts, is con-
cerned primarily with the contributions that have been made to the
theories and applications of colloid science from observations on
inorganic substances. The importance of the salts in the study of
colloidal phenomena is evidenced by the amount and character of the
work that has been carried on with this class of compounds during the
past decade. So significant are the results of the investigations that
a book on this subject written by the author and published by the
McGraw-Hill Book Company ten years ago had to be completely
revised and almost entirely rewritten in the preparation of the present
volume.
The plan of the book is similar to that followed in the first two
volumes of the series. After an introductory chapter dealing with
the general methods of formation of gels and sols of the salts, sepa-
rate sections are devoted in turn to the colloidal sul fates; colloidal
carbonates, phosphates, chromates, and arsenates; colloidal halides;
colloidal sulfides; colloidal ferrocyanides and f erricyamdes ; and col-
loidal silicates. The first portion of each section is concerned with a
critical survey of the conditions of formation and the general charac-
teristics of the individual salts in the colloidal state; and the second
portion, with the principles underlying their applications.
Among the examples of colloid chemical behavior that have re-
ceived special consideration in this volume are: the velocity of pre-
cipitation and the physical character of precipitates; the stability of
sols and the mechanism of the electrolyte coagulation process; ion
antagonism in colloid systems; the mutual coagulation process; ad-
sorption on ion lattices; adsorption indicators; the color of colloids;
the permeability of membranes; and the phenomena of thixotropy
and rheopexy.
The principles underlying the technical applications of the col-
loidal salts are illustrated by chapters on: plaster of Paris; lithopone
vi PREFACE
and other sulfide pigments; Prussian blue; the colloidal halides in
photography; the base-exchange phenomenon in silicate gels; the in-
organic colloids of the soil; and Portland and aluminous cements.
Following the practice of the earlier volumes of the treatise, the
attempt is made to render this critical survey of the colloidal behavior
of the salts as clear and concise as possible by the generous use of
section and paragraph headings in bold- face type. It is hoped that
this outline of the subject matter will contribute to the usefulness of
the book both for reference purposes and as a textbook for the study
of the principles of colloid chemistry in connection with a well-known
and widely diversified class of inorganic compounds.
August 10, 1937 HARRY BOYER WEISER.
CONTENTS
PAGE
PREFACE v
CHAPTER
I. INTRODUCTION 1
PART I
THE COLLOIDAL SULVATES AND RELATED
COMPOUNDS
II COLLOIDAL BARIUM SULFATE li
III. COLLOIDAL SULFATES OF LEAD AND STRONTIUM . . 49
"IV. COLLOIDAL CALCIUM SULFATE; PLASTER OF PARIS . . 61
V THE COLLOIDAL CARBONATES, PHOSPHATES, CHROMATES, AND
ARSENATEb . , . . 83
PART II
THE COLLOIDAL HAL1DES
VI. COLLOIDAL SILVER IODIDE 99
VII THE COLLOIDAL HALTDES OF SILVER, LEAD, AND MERCURY . . 126
VIII THE SIIVER HALIDES IN PHOTOGRAPHY 143
PART III
THE COLLOIDAL SULFIDES
IX. COLLOIDAL ARSENIC TRISULFIDE GENERAL PROPERTIES . . . 167
X. COLLOIDAL ARSENIC TRISULUDE: STABILITY OF SOL . . .181
XI. THE COLLOIDAL SULFIDES OF ANTIMONY, BISMUTH, TIN, AND LEAD 222
XII. THE COLLOIDAL SULFIDES OF COPPER, SILVER, GOLD, AND THE
PLATINUM FAMILY 234
XIII. THE COLLOIDAL SULFIDES OF ZINC, CADMIUM, AND MERCURY . 250
XIV. THE COLLOIDAL SULFTDES OF MANGANESE, NICKEL, COBALT, IRON,
AND THE RARER ELEMENTS 267
XV. LlTHOPONE AND OTHER SULFIDE PIGMENTS 280
viii CONTENTS
PART IV
THE COLLOIDAL FERROCYANIDES AND FERRICYANIDES
CHAPTER PAGE
XVI. COLLOIDAL FERROCYANIDES AND FERRICYANIDES: GENERAL PROP-
ERTIES 305
XVII COLLOIDAL COPPER FERROCYANIDE' THE SOL 315
XVIII. COLLOIDAL COPPER FERROCYANIDE AND FERRICYANIDE THE MEM-
BRANES 326
XIX PRUSSIAN BLUE AND RELATED PRODUCTS 343
PART V
THE COLLOIDAL SILICATES
XX. SILICATE SOLS AND GELS; THIXOTROPY . . 361
XXI. BASE EXCHANGE IN SILICATE GELS . . 388
XXII. THE INORGANIC SOIL COLLOIDS .... . 401
XXIII. CEMENT 427
INDEX OF AUTHORS 445
INDEX OF SUBJECTS . 463
THE COLLOIDAL SALTS
CHAPTER I
INTRODUCTION
Thomas Graham * in his fundamental paper "Liquid Diffusion
Applied to Analysis" showed that substances like certain inorganic
salts, sugar, and glycerol, which are readily obtained in a crystalline
form, will diffuse through water and certain membranes much more
rapidly than substances such as albumin, gelatin, and the hydrous
oxides which are gelatinous in character and are obtained in the form
of crystals with difficulty, if at all. Substances which diffused rapidly
were called crystalloids, and those which showed little or no diffusion
were termed colloids, from the Greek word for glue. On the basis
of this distinction most inorganic salts would not be classified as col-
loids; but Graham recognized that the classification was arbitrary.
The important investigations of von Weimarn 2 with a definitely crys-
talline inorganic salt, barium sulfate, confirmed the view that the
distinction between crystalloids and colloids is not tenable and that
we should speak of a colloidal state of matter, a concept first recog-
nized by Graham. A material is said to be colloidal when it is suffi-
ciently finely divided, the diameters of the dispersed particles ranging
from approximately 100 m/* to 1 m/*. 8 Since a large number of the
inorganic salts are readily obtained in the colloidal state of subdi-
vision, investigations with this class of substances have played an
important part in the development of colloid chemistry both theoretical
and applied. Attention will first be given to the general conditions for
forming colloidal inorganic salts.
iPhil Trans, 151, 183 (1861)
* Kolloid-Z., 2, 199, 230, 275, 301, 326, suppl. 2 LII; 3, 89, 282 (1908) ; 4, 27
(1909); "Grundzuge der Dispersoidchemie" (1911); "Zur Lehre von den Zus-
tanden der Materie" (1914).
8 Ostwald, Wo.: "Grundriss der Kolloidchemie" (1909).
2 INTRODUCTION
FORMATION or GELS AND SOLS OF INORGANIC SALTS
The two general methods for obtaining substances in the colloidal
state are (1) condensation, in which smaller particles, molecules, or
atoms are gathered into particles of colloidal dimensions, and (2)
dispersion, in which the material in mass is dispersed into particles
of colloidal dimensions.
Condensation Methods
Colloidal salts are obtained either in the form of gels gelatinous
precipitates and jellies or sols. Condensation methods are far more
important than precipitation methods in the preparation of both these
types of colloidal systems. Indeed, as we shall see, the most common
examples of dispersion methods of forming colloidal salts actually
start with aggregates of particles of colloidal dimensions which have
been obtained by a condensation process.
Condensation methods as applied to the formation of colloidal salts
usually involve precipitation as a result of chemical reactions in solu-
tion. Von Wcimarn 2 made the first systematic investigation of the
form in which substances precipitate as a result of reactions in solu-
tion and outlined the conditions under which sols and gels would be
expected to form. The von Weimarn theory has been considered
briefly in the earlier volumes of this work, 4 and its applicability and
limitations will be taken up in detail in the next chapter, which deals
with colloidal barium sulfate. For the present it need only be pointed
out that particles of colloidal dimension will precipitate when the
percentage supersaturation is sufficiently high at the moment precipi-
tation begins. The percentage supersaturation is defined by the ex-
pression (Q L)/L, where Q is the total concentration of the sub-
stance that will precipitate and L is its solubility. In general, a highly
insoluble substance precipitates in the colloidal state as a sol, a gelat-
inous precipitate, or a jelly depending on the concentration of the
reacting substances [(Q L)/L is large because L is very small].
Fairly soluble substances, on the other hand, come down in the gelat-
inous or jelly-like form only from highly concentrated solutions
[(Q L)/L is large because Q is very great].
Von Weimarn emphasized that substances thrown down in the form
of gels pass through the sol state at some stage of the process. This
means that a gel results from the agglomeration of a sol that may
have but a transient existence or that may be quite stable in the ab-
*Vol. I, pp. 1-4, 26-28 (1933); II, p. 3 (1935).
CONDENSATION METHODS 3
sence of excess electrolyte (Vol. II, p. 15). From this point of view,
the problem of the synthesis of stable sols is to create conditions such
that the sol state is maintained for an indefinite period. The von
Weimarn theory tells us only that very low solubility of the precipitat-
ing phase and high dilution favor sol formation. In actual practice,
precipitation in the form of particles of colloidal dimensions is ef-
fected and coagulation or agglomeration of the colloidal particles is
prevented in one of two ways: (1) keeping down the concentration
of coagulating electrolyte by choosing a suitable reaction or by dilu-
tion, or (2) adding protecting agents such as "water-soluble" protect-
ing colloids which are adsorbed on the surface of the particles and
prevent their coming in contact. The most typical reaction for the
preparation of metallic sols is reduction, for the preparation of hydrous
oxide sols is hydrolysis, and for the formation of sols of the in-
organic salts is double decomposition. The general methods of form-
ing salt sols together with some specific illustrations will be taken
up in order.
Double Decomposition in Which the Second Product Is a Non-
electrolyte. Since sols are coagulated by electrolytes above a certain
concentration, it follows that the preparation of stable sols by double
decomposition depends to a large extent on the second product of the
reaction. If the reaction is chosen so that the second product is a non-
electrolyte or a weakly dissociated electrolyte, the conditions will be
favorable to sol formation. The classical example under this heading
is the reaction between hydrogen sulfide and a solution of arsenious
acid: 2H 3 AsO 3 + 3H 2 S - As 2 S 3 '+ 6H 2 O. 5 The electrolyte con-
centration is quite low throughout the process, and the sulfide re-
mains dispersed as negatively charged particles stabilized by the prefer-
ential adsorption of S and HS~ ions. In this case the conditions
for sol formation are so favorable that the sol can be made as con-
centrated as 3 g of As 2 S 3 per 5 g of water.* Similarly, a stable sol
of mercuric sulfide is formed by the interaction of a saturated solu-
tion of mercuric cyanide and hydrogen sulfide 7 since hydrocyanic
acid, the second product of the reaction, is so weakly dissociated;
copper glycocoll treated in a similar way yields a stable cupric sulfide
sol. T
Double Decomposition in Which the Second Product Is an Electro-
lyte. Because of the coagulating action of electrolytes, a stable sol
8 Berzelms: "Lehrbuch der Chemie," 3rd ed. (1834).
Schulze: J. prakt Chem., (2) 25, 431 (1882).
T Lottermoser : J. prakt. Chem., (2) 75, 293 (1907).
4 INTRODUCTION
results when the second product of the double decomposition is an
electrolyte, under one of the following conditions: (1) very high
dilution, (2) one of the electrolytes contains an ion which is adsorbed
very strongly by the dispersed particles. As an example of the first,
sols of most of the metallic sulfides are obtained by the interaction
of very dilute solutions of the respective chlorides and hydrogen
sulfide. 8 The classical example of the second is the observation of
Lottermoser 9 that at sufficiently low concentrations silver halide sols
result from the interaction of silver nitrate and alkali halide provided
that one or the other of the reactants is in slight excess. With the
silver salt in excess the particles are positively charged by reason of
preferential adsorption of silver ion; with the halide in excess the
particles are negatively charged because of preferential adsorption of
halide ion. This stabilizing effect of the adsorbed ions is greater than
the coagulating action of alkali nitrate, the second product of the
metathetical reaction
Double Decomposition in the Presence of Strongly Adsorbed Non-
electrolytes. The precipitation of salts is frequently prevented from
going beyond the colloidal range and the dispersed phase kept in the
sol state by the addition of substances which are strongly adsorbed
such as the so-called protecting colloids and sugar or glycerol. Thus,
concentrated stable sols of silver halides and Prussian blue are ob-
tained by precipitation in the presence of gelatin. 10 Silver sulfide and
cadmium sulfide are stabilized in the sol state by gum arabic or
casein. 11 Silver chromate remains as a sol in the presence of sugar;
and most insoluble salts can be held in colloidal solution by sodium
"protalbinate" and "lysalbinate" which are formed by the saponifica-
tion of albumin. 12 Sodium chloride resulting from the interaction of
sodium malonic ester and chloroacetic ester in dry benzene is stabilized
in the sol condition by strong adsorption of one or more of the
organic reagents. 13 Barium sulfate thrown down from aqueous al-
cohol gives a gel which forms a sol on shaking with a larger quantity
of water. 14 Strong adsorption of alcohol doubtless helps to prevent
both coalescence and growth of the dispersed particles since a jelly
s Winssinger: Bull. soc. chim., (2) 49, 452 (1888).
J. prakt. Chem., (2) 68, 341 (1903) ; 72, 39 (1905) ; 73, 374 (1906).
">Lobry de Bruyn: Rec. trav. chim., 19, 236 (1900).
"Miiller and Artmann: Oesterr. Chem.-Ztg., 7, 149 (1904)
12 Paal: Ber., 39, 1436; Paal and Kiihn: 2859, 2863 (1906); 41, 51, 58 (1908).
"Michael- Ber., 38, 3217 (1905).
"Kato: Mem. Coll. Sci, Kyoto Imp. Univ, 2, 187 (1909).
DISPERSION METHODS 5
thrown down from aqueous solution does not give a stable sol on
shaking with water.
A special case under this heading is the sols in concentrated elec-
trolyte solutions. For example, colloidal sulfidcs and silver halides
may be prepared by double decomposition in concentrated sulfuric
acid or phosphoric acid. 15 The particles in such sols are without
charge; the stability results from strong adsorption of the dispersion
medium. It thus appears that a hydrophobic sol of a given charge
has two stable and two unstable zones depending on the electrolyte
concentration. With little or no electrolyte such sols are unstable;
with a small electrolyte concentration they are stable ; somewhat larger
concentrations cause flocculation ; and finally very high concentrations
again exert a stabilizing influence.
Replacement of Solvent. In this general method of sol synthesis,
a solution of the substance under consideration is poured into a
liquid in which the solute is so insoluble that it appears as a highly
dispersed phase. This procedure is of limited applicability in the
synthesis of sols of the inorganic salts, and one example will suffice.
Silver iodide is insoluble in water but is fairly soluble in a solution
of potassium iodide. 16 When a solution of the salt is poured into a
large volume of water, a silver iodide sol is formed stabilized by the
preferential adsorption of iodide. High dilution is essential to avoid
an excess of potassium iodide which would cause coagulation.
Oxidation. Certain salts have been obtained in the colloidal state
by the oxidation of the corresponding metallic sols. Thus silver halide
sols are obtained by the action of chlorine, bromine, and iodine, re-
spectively, on a silver sol 17 until the characteristic color of the metal-
lic sol is destroyed. The stability of the resulting halide sols is in-
creased by the presence of gelatin or ammonium citrate. A similar
procedure may be employed to prepare the more insoluble halide sols
of lead and mercury.
Dispersion Methods
Dispersion methods of preparing sols involve the washing out of a
precipitating agent, the addition of a peptizing agent, and disintegra-
15 Voet: J. Phys. Chem., 40, 307 (1936) ; Ostwald and Wannow: Kolloid-Z.,
76, 159 (1936).
"Hellwig: Z. anorg. Chem, 26, 157 (1900).
17 Lottermoser and Meyer: J. prakt. Chem., (2) 66, 247 (1897); 67, 543
(1898).
6 INTRODUCTION
tion by electrical or other means. Each of these general procedures
has been employed in preparing salt sols.
Washing Out of Precipitating Agent. When a precipitate con-
sists of highly dispersed primary particles that have agglomerated into
aggregates which settle out, it may be repeptized to the sol state by
washing out the electrolyte responsible for the agglomeration. It
should be emphasized that this procedure is applicable only if the ma-
terial to be peptized already exists in the colloidal state of subdivision
and if the particles do not grow as a result of solution of smaller grains
and growth of larger ones at their expense (Ostwald ripening). The
method has proved particularly useful in the direct preparation of
fairly pure hydrous oxide sols, and it is applicable to the preparation
of various kinds of salt sols. Thus zinc sulfide thrown down from
chloride solution by ammonium sulfide is peptized in part by washing
out the ammonium salt. 18 Copper ferrocyanide gel thrown down
from alkali ferrocyanide solution with a small excess of copper salt
is peptized if the copper salt is removed by washing. 19 By using
hydro ferrocyanic acid and cupric acetate to form the gel and washing
by the aid of the centrifuge, a sol of almost any desired purity may
bo prepared. 20 Silver halides precipitated rapidly in the cold and
washed at once by decantation undergo partial repeptization to the sol
state.
Addition of Peptizing Agent. If the primary particles of a col-
loidal precipitate have coalesced sufficiently, the coagulation is not
reversible by washing alone. Under such circumstances the particles
may be broken apart once more by the addition of a suitable peptiz-
ing agent. The Prussian blue gel is readily peptized by oxalic acid,
and freshly precipitated metallic sulfides are peptized by hydrogen
sulfide. Similarly, a freshly formed silver halide gel is peptized by
a small amount of either silver nitrate or alkali halide. Carbon
dioxide passed into a solution of barium oxide in methyl alcohol gives
at first a thick gel of barium carbonate which is peptized by the
addition of more gas, forming an opalescent sol. 21
Electrical Disintegration. The formation of sols by means of an
electric arc under water or organic liquid is applicable chiefly to the
preparation of metallic sols; but the method finds some application in
"Donnini: J. Chem. Soc., 66 (2), 318 (1894).
"Berkeley and Hartley: Phil Trans, 206A, 486 (1906).
zoWeiser and Milligan: J. Phys. Chem., 40, 1071 (1936).
91 Neuberg and Neumann: Biochem. Z., 1, 166 (1906) ; Neubcrg and Rewald:
Kolloid-Z, 2, 321 (1908).
DISPERSION METHODS 7
the preparation of sols of certain salts, chiefly sulfides, which conduct
the current fairly well. Von Hahn 22 prepared stable sols of galena,
molybdenite, and antimonite 2S by cathodic disintegration with a di-
rect-current arc as used by Bredig. 24 The dispersion of molybdenite
takes place only when the temperature exceeds 53 more or less, the
critical temperature varying slightly with different minerals. This
appears to be the first recorded example of a critical temperature in
the electrical synthesis of sols. The oscillatory discharge first used by
Svedberg 25 to synthesize sols gives sols of antimonite, sphalerite,
chalcocite, and molybdenite, whereas only coarsely dispersed systems
are obtained with iron pyrites and galena.
The general conditions for obtaining salts in the colloidal state
having been outlined, attention will be directed to the colloidal be-
havior of this important class of inorganic substances. It seems ad-
visable to begin with colloidal barium sulfate since this compound
illustrates so well the effect of the conditions of precipitation on the
nature and form of the resulting precipitate
Kolloid-Z. (Zsigmondy Festschrift), 36, 277 (1925)
23 Cf. Currie: J Phys. Chcm., 80, 205 (1926)
24 Z. Elektrochem, 4, 514 (1898) ; Bredig and Rerneck Z physik Chcm, 31,
258 (1899).
as "The Formation of Colloids," 19 (1921).
PART!
THE COLLOIDAL SULFATES AND RELATED COMPOUNDS
CHAPTER II
COLLOIDAL BARIUM SULFATE
THE PHYSICAL CHARACTER OF PRECIPITATED BARIUM SULFATE
The importance of barium sulfate in gravimetric analytical chem-
istry, is responsible for a large number of investigations into the con-
ditions of precipitation of the salt and how these influence its physical
character and properties. Von Wiemarn was the first to show that
barium sulfate can be prepared in the laboratory in a variety of
physical states from masses of distinct macrocrystals to transparent
jellies The extended investigations of von Wiemarn on this and
other related salts led to a theory of the precipitation process, the ap-
plicability and limitations of which will be considered in some detail.
Velocity of Precipitation and Growth of Particles
The Velocity Equations. In his systematic study of the conditions
which influence the form of precipitates, von Weimarn 1 calls atten-
tion to a number of factors which may determine their nature: the
solubility of the substance; the concentration at which precipitation
takes place; the latent heat of precipitation; the normal pressure at
the surface of the solvent ; the molecular complexity of the reactants ;
polymerization of the reacting molecules; viscosity of the reaction
medium ; adsorption ; the presence of dust particles ; and the extent of
agitation on mixing. Other important factors which he seems to have
overlooked arc the specific tendency of the reaction product to form
nuclei and the specific tendency to grow on nuclei. Von Wiemarn
recognized the impossibility of taking all these factors into account
and simplified the problem at the outset by considering only the first
two factors : the solubility of the precipitating substance and the con-
centration at which the precipitation starts. The precipitation process
is considered as taking place in two stages: the first stage, in which
iKolloid-Z., 2, 199, 230, 275, 301, 326, suppl 2 LII; 3, 89, 282 (1908); 4,
27 (1909); "Grundziige der Dispersoid Chemie" (1911); "Zur Lehre von den
Zustanden der Materie" (1914) ; Repts. Imp. Ind Research Inst, Osaka, Japan,
V, 8, 7 (1927) ; 9, 3 (1928) ; 12, 5 (1931).
11
12 COLLOIDAL BARIUM SULFATE
the molecules condense to invisible or ultramicroscopic nuclei; and
the second, which is concerned with the growth on the nuclei as a
result of diffusion. The velocity W at the important first moment
of the first stage of the process is given by :
w _ K Pto*on pressure = K Q-L = K P KU
precipitation pressure L L
where Q is the total concentration of substance that is to precipitate ;
L t the solubility of coarse crystals ; and K, a constant. P is the abso-
lute supersaturation, and P/L = U is the percentage supersaturation
at the beginning of precipitation. To take care of all the other factors
which may enter into the process von Weimarn introduces a "variable
multiplier" /, and the equation becomes :
(2)
The Nernst-Noyes equation gives the velocity, V, of growth on
nuclei :
V-j>0-(Q-L) (3)
where D is the diffusion coefficient ; S, the thickness of the adhering
film (length of the diffusion path); 0, the extent of surface; and
Q and L have the same significance as above.
Several facts may be interpreted qualitatively by the aid of these
equations: It will be seen that the velocity of the precipitation de-
pends not on the supersaturation P, but on the percentage super-
saturation P/L. Thus, with a given value of P (say, a few grams
per 100 cc), a very soluble substance, such as sodium chloride, will
deposit nothing at first and finally a few crystals may form ; but with
the same value of P, an almost insoluble substance, such as alumina or
silver chloride, will give an immediate gelatinous or curdy precipitate.
The difference is that the velocity of precipitation is much smaller in
the first case than in the second. On the other hand, if sodium
chloride is formed by the interaction of sodium ethylate or thiocyanate
and hydrochloric acid in a mixture of ether and amyl alcohol in which
sodium chloride is practically insoluble, the precipitate is curdy like
that of silver chloride.
Although the value of P is not in itself of primary importance in
determining the form of the precipitate, its value is not without in-
fluence, since quite different results are obtained depending on
VELOCITY OF PRECIPITATION AND GROWTH OF PARTICLES 13
whether a given value of U is obtained by a large P or by a small
L. In the first instance, a large amount of the dispersed phase will
be produced, and in the second, very little. Hence a large {/-value
resulting from a large value of P will, in general, give a gelatinous
precipitate or a jelly ; whereas a large C/-value resulting from a very
small L-value will give a large number of highly dispersed particles
a sol. It would appear, therefore, that the dispersed phase can be
made to separate in any desired form by suitable alteration of P or L
or of both.
Effect of Concentration of Reactants on the Precipitate Form.
For sparingly soluble substances, von Weimarn distinguishes five
stages with increasing supersaturation : If the supersaturation is
slight, no precipitation occurs inside of several years ; in the next stage
of higher supersaturation, perfect crystals appear in a relatively short
time; in the third stage, skeleton crystals and needles appear; in the
fourth stage, gelatinous precipitates are formed; and in the highest
stage of supersaturation, jellies result.
The observations of von Weimarn on barium sulfate illustrate the
effect of concentration of reactants on the form of a precipitate. The
concentration of this salt (0.0024 g/1 at 18) is fairly high compared
to that of alumina, say ; hence the values of P which can be obtained
are not large enough to give large values of U, using the ordinary
laboratory solutions of barium chloride or nitrate and alkali sulfate.
With these reagents, the rate of formation of particles is relatively
slow and their subsequent growth is rapid; hence the precipitate or-
dinarily obtained consists of fairly large crystals. But by making
use of more soluble salts, such as barium thiocyanate or iodide and
ammonium or manganese sulfate, it is possible to obtain the salt in any
form from large crystals to a clear jelly. In the actual experiments,
equivalent solutions should be mixed in equal volumes; it is neces-
sary, of course, to use correspondingly large volumes of the very
dilute solutions, otherwise there will not be a visible quantity of
barium sulfate to separate out. Strictly, the product of the volume
and concentration should be constant. The results of a series of ob-
servations are given in Table I. It is obvious from these data that
a barium sulfate sol cannot be obtained unless L is diminished since
increasing the value of P to the point where colloidal particles are
formed gives a gel. Actually Kato 2 obtained a sol in the presence of
alcohol in which barium sulfate is much less soluble than it is in water.
Kato also obtained much more stable gels by the interaction of
2 Mem. Coll. Sci., Kyoto Imp. Univ., 2, 187 (1909).
14
COLLOIDAL BARIUM SULFATE
TABLE I
EFFECT OF CONCENTRATION OF REACTANTS ON THE PHYSICAL CHARACTER OF
BARIUM SULFATE (L - 0.002 g/1)
Normality of
p
Ba(CNS)*
P
V = j
Nature of precipitate
and MnSO 4
L
00005
000
No precipitate in a year. Microcrystals would
to
to
to
be expected in a few years and macrocrystals
00014
006
3
from large amounts of solution.
000014
006
3
Slow precipitation at U - 8. Momentary sol
to
to
to
stage at U = 25. Complete separation in
0017
096
48
months to hours.
0017
096
48
Crystal skeletons and needles precipitate in a
to
to
to
few seconds at U = 48, beyond this, instan-
75
43.8
21,900
taneous precipitation. Crystals barely recog-
nizable at U - 21,900.
75
43 8
21,900
Precipitates which appear amorphous, form
to
to
to
immediately.
3
175 1
87,500
3
175 1
87,500
Clear cellular jellies.
to
to
to
7
409
204,500
barium and sulfate ions in alcohol-water mixtures. Similarly, Lehner
and Taylor s prepared stable gels by the interaction of dilute solutions
of barium chloride and sulfuric acid in selenium oxychloride in which
barium sulfate is almost entirely insoluble. Careful observations dis-
closed that selenium oxychloride is without action on polished sur-
faces of barite and does not penetrate barium sulfate crystals. The
evidence appears conclusive, therefore, that gelatinous barium sulfate
obtained in selenium oxychloride is merely barium sulfate made plas-
tic by adsorbed selenium oxychloride. This accords with the author's
view * that a gelatinous precipitate consists of very finely divided solid
particles which have adsorbed the liquid strongly. The greater stabil-
ity of the alcogels and selenium oxychloride gels, as compared with the
J. Phys. Chem., 28, 962 (1924).
*Weiser: Bogue's "Colloidal Behavior," 1, 389 (1924); cf. Vol. II, p
14.
THE PRECIPITATION LAWS 15
aqueous gels, results both from lower solubility of barium sulfate in
the organic liquids and from stronger adsorption of the organic liquids
than of water. This is not true of all organic liquids: when barium
sulfate is shaken with a mixture of water and heptane the salt con-
centrates in the dineric interface, indicating that the adsorption of the
two liquids is of the same order ; on the other hand, when the salt is
shaken with a mixture of heptane and selenium oxychloride, it con-
centrates in the selenium oxychloride layer, and when shaken with a
mixture of selenium oxychloride and 70% H 2 SO 4 , it collects in the
acid layer.
The Precipitation Laws
Von Weimarn 5 formulated the following precipitation laws as a
result of observations on the precipitation of various salts such as the
sul fates of barium, strontium, calcium, and silver from aqueous and
aqueous-alcoholic solutions :
1. With increasing concentrations of reacting solutions, the average
size of the precipitated individual crystals (not their aggregates)
passes through a maximum during the process of direct crystallization.
2. With increasing concentrations of reacting solutions, the size of
the crystals decreases continuously after the completion of the process
of direct crystallization.
3. For the same absolute concentrations (Q L) of reacting solu-
tions, other conditions being equal, the average size of the precipitated
crystals decreases with decreasing solubility of a substance.
4. With increasing viscosity of the dispersion medium, the average
size of the particles decreases.
The first three laws may be represented schematically in Fig. 1,
in which the effective concentration of reacting solutions (Q L) or
the velocity of precipitation is plotted against particle size for three
solubility values, L', L", and L'", which yield, respectively, macro-
scopic, microscopic, and ultramicroscopic crystals. The L^-curves repre-
sent the first law ; the /^-curves, the second ; and the positions of the
L^-curves and L-curves with respect to each other, the third. In
general, the L^-type of curve showing a maximum is obtained a short
time 15 to 30 seconds after the mixing, and the La-type a month
or more after the mixing. The L'-curves are obtained from a medium
in which the substance is fairly soluble, for example, barium sulfate
precipitated from hot hydrochloric acid solution; the Z/'-curves cor-
*Cf. Chem. Rev., 2, 217 (1926); Alexander's "Colloid Chemistry," 1, 27
(1926).
16
COLLOIDAL BARIUM SULFATE
respond, for example, to the precipitation of barium sulfate from aque-
ous solutions; and the //"-curves to the precipitation from aqueous
alcoholic solutions. With 40% alcohol and higher, the curves for
barium sulfate lie almost entirely in the ultramicroscopic zone.
Contrary to von Weimarn's contention, Oden, 6 working with barium
sulfate, showed the absence of a maximum in the concentration-size
of particle curve, and demonstrated that the degree of dispersion al-
ways diminishes as the concentration of reacting ions increases. By
imm
*200m/f
1mm
200m/*
(Q-L) , Velocity of Precipitation
FIG. 1. Schematic representation of von Weimarn's precipitation laws.
means of a special apparatus, Oden measured the distribution of vari-
ous sizes of particles of barium sulfate thrown down on mixing vary-
ing concentrations of ammonium sulfate and barium thiocyanate. He
found that aggregates of secondary particles are very often formed
with quite small particles: above 100 m/* scarcely any are obtained,
but below 20 m/t aggregates nearly always result if no special pre-
cautions are taken, such as the addition of citrate, to redisperse the
flArkiv Kemi, Mineral. Geol, 7, No. 26 (1920); Svedberg's "Formation of
Colloids," London, 94 (1921).
THE LAW OF CORRESPONDING STATES 17
aggregates. Oden attributed the difference between his observations
and those of von Weimarn to the failure of the latter to recognize the
existence of aggregates formed at certain concentrations, even though
he was looking for them. Oden's results are in agreement with those
for condensation in gases where the degree of dispersion increases
regularly as the concentration rises (Vol. 1, p. 6). On the other
hand, Bruzs 7 obtained evidence of the existence of a region of maxi-
mum grain size of barium sulfate with certain concentrations of
reagents, from measurements of the surface energy by a calorimetric
method.
It should be pointed out that von Weimarn calls all precipitates
crystalline but he believes liquids and gases to be crystalline, which
suggests that he is not using the term in its ordinary sense. Even
the most gelatinous precipitates of barium sulfate give the x-ray dif-
fraction pattern of heavy spar, 8 and it is probable that most salt pre-
cipitates are crystalline to x-rays. At the same time, the author does
not believe with von Weimarn that there is no such thing as an
amorphous state in nature ; but if one accepts von Weimarn's criterion
that the amorphous state must be completely without vectoriality, it is
probable that nothing completely amorphous exists. 9
The Law of Corresponding States
The Simplified Expression; Limitations. Von Weimarn recog-
nized that the velocity W of the first stage of precipitation cannot be
measured in actual practice, and that, in many cases especially interest-
ing in the synthesis of colloid systems, the velocity V of the growth
of particles cannot be determined. In due time, therefore, he intro-
duced a specific coefficient called the "precipitate form coefficient" or
"dispersity coefficient" N, which is given by the expression :
N = j.K ab .K ed . K bd -Kao-Z (4)
Lt
in which P/L is the percentage supcrsaturation, as in the velocity equa-
tion; Z, the viscosity; and KM, K cd , etc., represent the "physical and
chemical association" of the substances AB, CD, etc., which enter into
the reaction AB (in solution) + CD (in solution) = AC (precipitate)
7 J. Phys. Chem., 84, 621 (1930); Bruzs and Jankauskis: Chem. Abstracts,
24, 4197 (1930).
8 Von Weimarn and Hagiwara: Kolloid-Z., 88, 129 (1926).
9 Von Weimarn and Hagiwara: Japan J. Chem., 8, 15 (1926) ; von Weimarn:
Kolloid-Z., 44, 279 (1928).
18 COLLOIDAL BARIUM SULFATE
-f- AC (in solution) + BD (in solution). As a first approximation all
factors except P/L may be neglected, that is,
* = (5)
or taking into account all factors in addition to P/L
JV=/| (6)
in which / has the same significance as in Equation (2).
Now if N is taken as approximately equal to P/L, then for the dif-
ferent substances x t y, and z,
tf.-; tf*-^; and tf.-fs.
Li x Liy 1^9
If the character of the precipitate is to be the same, irrespective of
the chemical nature of the salt; in other words, if
N. = N v - N.
then
This is the simplest expression for von Weimarn's law of correspond-
ing states for the precipitation process, which says that, under corre-
sponding conditions of precipitation, the mean magnitude (expressed
in gram molecules) of the crystals of substances capable of precipita-
tion will be the same. In the form given in Equation (7) the so-called
law can hardly be regarded as a first approximation, even with sub-
stances that are related chemically. This is illustrated by the observa-
tions of Buchner and Kalff 10 for two series of related salts, as given
in Table II. The solubility L is given in equivalents per liter, and AT
is calculated, for the concentrations (in equivalents) employed, from
the expression N = P/L.
Considering the first group of salts, it will be seen that the nature
of the precipitate is essentially the same although the values of N
vary between 75 and 100,000. Von Weimarn observed the rapid for-
mation of definite crystals of BaSO 4 at values of N varying between
50 and 20,000 (Table I). The law of corresponding states in its
ioRec. trav. chim., 89, 135 (1920).
THE LAW OF CORRESPONDING STATES
19
simplest form would require that, under the conditions recorded in
Table II, calcium sulfate, calcium fluoride, and barium fluoride should
give definite crystals instead of clear jellies. Turning to the silver
halides, if von Weimarn's so-called law were applicable in its simplest
form, these salts should give jellies more stable than barium sulfate
under the conditions used; but instead, they give transparent drops
TABLE II
PHYSICAL STATES OF PRECIPITATES
Salt
L
N
State of precipitate
CaF 2
BaF 2
CaSO 4
BdS0 4 .
4X10- 4
18X10- 3
3X10-*
2X10- 5
3,400
75
140
100,000
Clear jelly, very stable
Clear jelly, very stable
Clear jelly, stable
Clear jelly, stable
AgCJ . .
AgBr
Agl . . . .
Pbl,
1x10-5
7X10-'
1 5X10- 8
4 8X10~ 3
700,000
8,000,000
300,000,000
360
Jelly, vary instable
Jelly, very instable
Jelly, very instable
Jelly, very instable
which cloud up and disintegrate in a few seconds into floes of an en-
tirely different physical character from the barium sulfate jelly.
Finally, the small value of N for lead sulfate would lead one to expect
the formation of a definitely crystalline precipitate; actually, a jelly
results temporarily.
An even more striking exception to the simplified formulation is
magnesium arsenate, which can be made to form a clear stiff jelly by
mixing quite dilute solutions of potassium arsenate and manganese
sulfate (p. 92). The value of L for the precipitate is so large that
precipitation is slow and quantitative precipitation impossible in the
dilute acid solution ; hence P/L = N is quite small.
Solubility and Crystal Size. It was observed by Hulett 11 and
confirmed by Dundon 12 that the addition of finely ground calcium sul-
fate or barium sulfate to saturated solutions of their respective salts
caused an increase in conductivity which rose to a maximum and then
decreased slowly, finally approaching that of the normally saturated
solution. This changing conductivity is usually attributed to changing
"Z. physik. Chem,, 37, 385 (1901).
" J. Am. Chem. Soc., 46, 2658 (1923).
20
COLLOIDAL BARIUM SULFATE
solubility with particle size. From measurements of particle size and
of solubility, it is possible to calculate the surface tension. Dundon
has done this for several substances using the equation of Dundon
and Mack. 13 The results are given in Table III, in which M stands
TABLE III
Diam-
Increase
Substance
M
*
M.V.
eter
in solu-
Temp
op
<r
Hardness
/*
bility %
BaS0 4 (H)
233
4 5
52
1
80
25
1250
2 5-3 5
BaS0 4 (D) . ..
233
4 5
52
2
90
30
3000
2 5-3 5
CaSO 4 -2H 2 O .
172
2 32
74 2
02-05
4.4-12
30
370
1 6-2
SrSO 4
184
3 96
46 4
25
26
30
1400
3.0-3 5
Ag 2 CrO 4
332
5 52
60 1
3
10
26
575
2 (approx.)
PbI 2 . . .
461
6 16
74 8
4
2
30
130
Very soft
PbF 2
245
8 24
29 7
3
9
25
900
2 (approx.)
CaF a . . .
78
3 18
24 6
0.3
18
30
2500
4
for the molecular weight; , the density of crystal; M.V., the molecu-
lar volume ; r, the radius of particle measured microscopically ; and <r,
the surface tension of particle.
Considering slightly soluble salts which are completely dissociated
and whose activity coefficients may be taken as unity, the relation be-
tween increase in solubility on the one hand and size of crystals and
surface tension on the other may be represented by the equation :
S r 2<r
in which S r is the solubility of particles with radius r\ and S is the
solubility of normal crystals.
By means of this equation and Dundon's data Kolthoff 14 calcu-
lated the ratio S r /S for three salts at a size r = 0.02, obtaining the
following: for BaSO 4 , 930; for Ag 2 CrO 4 , 4.0; for PbI 2 , 1.38. As-
suming that Dundon's data give at least the order of magnitude of
the surface tension of various crystals, it appears that the solubility
" J. Am. Chem. Soc, 45, 2479 (1923); cf. Ostwald: Z. physik. Chem., 84,
495 (1900); Freundlich: "Kapillarchemie," 144 (1909).
"J. Phys. Chem., 86, 860 (1932).
THE LAW OF CORRESPONDING STATES 21
of very small crystals of barium sulfate is about 1000 times greater
than that of larger crystals, whereas the solubility of silver chromate
increases only 4 times and that of lead iodide only 1.4 times under
the same conditions.
Kolthoff points out how the differences noted above will explain
the fact that substances of about the same solubility and precipitated
under analogous conditions give entirely different forms of precipi-
tates, contrary to von Weimarn's predictions. For example, silver
chloride and barium sulfate have solubility products of about the same
order of magnitude, but the former always comes down in floes and
the latter in the form of microcrystals under analytical conditions.
This can be explained from the difference in solubility with crystal
size for the two salts. Starting with such concentrations of the respec-
tive reacting ions that the macroscopic supersaturation of the salts are
the same, one obtains soft crystals 15 of the silver chloride whose solu-
bility is more or less independent of the crystal size and relatively
hard crystals of barium sulfate whose solubility varies greatly with
the crystal size. Since the solubility of the primary particles (nuclei)
of silver halide first formed is about the same as that of large crystals
and the solubility of the primary particles of barium sulfate is much
larger than that of large crystals, it follows that the solution will be
much less supersaturated with respect to the small particles of barium
sulfate on the one hand than of silver chloride on the other. From
Equations (2) and (3) we see that the velocity of formation of
nuclei and growth on nuclei increases with increasing supersatura-
tion; therefore the formation of nuclei is more spontaneous with the
silver salt than with the barium salt. The rapid formation of so
many nuclei of silver chloride soon exhausts the solution ; under these
conditions, little or no growth on nuclei takes place and the precipitate
is a flocculent, colloidal mass. With barium sulfate, much fewer nuclei
form under the same conditions ; these grow at the cost of ions left
in solution, and a microcrystalline mass results. Kolthoff 18 points out
further that Bottger's 17 relation between the sensitivity, E, of a pre-
cipitation reaction and the solubility, L, of the particles : E = L + V,
where V is the visibility of the particles, has no general validity. It
will hold only if the "micro" and "macro" solubilities of the substance
are approximately the same as those of the silver halides.
18 Cf. Reis and Zimmermann: Z. physik. Chem., 102, 299 (1922).
*Z. anal. Chem. Bottger Festschrift, 86, 34 (1931).
"Chem.-Ztg., 88, 1003 (1909); 86, 1097 (1912); Z. angew. Chem., 26, 1992
(1912) ; Gorski: Z. anorg. Chem., 81, 315 (1913).
22 COLLOIDAL BARIUM SULFATE
In this connection Balarew 18 questions whether the changing con-
ductivity (p. 19) on adding finely ground barium sulfate, say, to the
saturated solution of the salt is due primarily to varying solubility
with particle size. Balarew attributes the initial increase in conduc-
tivity to (1) the presence of barium chloride in the sulfate, (2) the
greater solubility of broken than of complete crystals, and (3) the
breaking up of atomic aggregates, The subsequent fall in conduc-
tivity could be caused partly by the crystallizing out of barium sulfate
on account of the solution of the chloride ; but the velocity of crystal-
lization is much greater than the rate of fall of conductivity so that
the latter is attributed to the slow restoration of equilibrium between
the complete and broken crystals. Although the greater solubility of
broken crystals may be the determining factor in the experiment re-
ferred to, there is no doubt that, below a certain size, the solution
pressure of smaller particles is greater than that of larger ones. Bala-
rew raises but does not settle the question as to whether Hulett's
experimental method gives a true measure of the change in solubility
with crystal size.
The Generalized Expression; Applicability and Limitations.
Von Weimarn's 19 explanation of the above discrepancies is, of course,
that the law of corresponding states for the precipitation process is
not the simple expression :
L. Ly L,
but
j . Px _ j m P. = j . ^ . .
J^x L'y "*
in which J rt J y , and J z are specific variable multipliers, the value for
any substance being "the product of all other factors (in addition to
P/L) which influence the crystallization process. These values must
be expressed by abstract numbers such that the values for P/L are
equivalent." 19 In other words, von Weimarn's equation for his so-
called law becomes quantitative and generally applicable by introduc-
ing "variable multipliers," handy wastebaskets, as it were, into which
are thrown all the variable factors known or unknown which have
not been evaluated. Von Weimarn also calls attention to the fact
" Z. anorg. Chem., 145, 122 (1925) ; 151, 68; 15ft, 170 (1926) ; 168, 213 (1927) ;
Kolloid-Beihefte, 30, 249 (1930) ; 82, 205; cf Stranski: 197 (1931).
"Kolloid-Beihefte, 18, 48 (1923).
THE LAW OF CORRESPONDING STATES 23
that the L-value employed in the calculation should not be the known
solubility in the pure solvent but some other value which corrects for
the effect of electrolytes in the surrounding solution. 20
Although it is possible to express facts fairly accurately by means
of such flexible formulas, it is doubtful whether anything is gained
scientifically by regarding formulations of this kind as quantitative
representations of natural laws. Von Weimarn evidently thinks so,
but his opinion is not shared generally. Bancroft 21 prefers to discard
the formulas altogether and state the whole thing from another point
of view. He points out that the mean size of the crystals is deter-
mined by the total amount of material crystallizing and the number
of nuclei. The really important thing, therefore, is the number of
nuclei which are formed under any given conditions. It is con-
tended, very properly, that factors other than percentage supersatura-
tion influence the number of nuclei formed. Thus the specific nature
of the substance, stirring, and temperature have a profound effect on
nuclei formation, and adsorption exerts a marked influence on the
growth of particles. 22 ' 28 Freundlich 2 * likewise does not believe that
the separation of a solid phase is generally and uniformly regulated by
its solubility and the supersaturation prevailing: "At the same degree
of supersaturation, the velocity of formation of nuclei, the velocity of
crystallization, and the ratio between the two are quite different for
different substances ; and they may be influenced in very different ways
by foreign substances, perhaps by the ions present during precipi-
tation."
Every analytical chemist is aware of the marked effect on the
physical character of precipitates of the presence of foreign ions in
solution. Barium sulfate comes down in a more finely divided state
when precipitated with sulfate in excess than when precipitated with
barium in excess. 25 Since, in general, any substance which is adsorbed
by a second will tend to peptize the latter, it follows that, other con-
ditions being the same, barium sulfate will come down most finely
divided when precipitated in the presence of those substances which
it adsorbs most strongly. 22 Now barium sulfate adsorbs its own ions
2 <> Von Weimarn: Kolloid-Z. f 2, 278 (1908); 32, 145 (1923).
21 J. Phys. Chem., 24, 100 (1920).
22Weiser: J. Phys. Chem., 21, 314 (1917).
23 Cf. von Weimarn: "Grundzuge der Dispersoid Chemie," 97 (1911).
2* "Kapillarchemie," 3rd ed., 2, 78 (1932).
2 *Foulk: J. Am. Chem. Soc., 18, 803 (1896); cf. Popoff: Ind. Eng. Chem.,
Anal. Ed, 2, 230 (1930).
24 COLLOIDAL BARIUM SULFATE
strongly, and hydrogen ions are adsorbed more strongly than most
cations; accordingly, when sulfuric acid is treated with barium chlo-
ride in excess, the precipitate tends to come down in a finely divided
state. This effect is sufficiently great to peptize the salt as a positive
sol under suitable conditions. The precipitate would come down in a
very finely divided form in the presence of a slight excess of sulfuric
acid, in which it is less soluble than in water, 26 were it not that the
strongly adsorbed hydrogen ion counteracts the effect of the adsorbed
sulfate. With potassium sulfate in excess the precipitate is finer than
with sulfuric acid in excess, since potassium ion is not strongly ad-
sorbed. As would be expected, the presence of hydrochloric acid
tends to give a coarser precipitate because of the solvent action of
the acid.
Von Weimarn's reply is that all the several specific factors other
than P/L are taken into account by means of the variable multiplier.
It seems rather unfortunate that von Weimarn attributed the criti-
cism of his point of view to misunderstanding owing to incomplete
knowledge of his work. It is more likely that people understand it
too well and so refuse to recognize the general validity of a so-called
law which is formulated by the use of a variable multiplier made up
of an indefinite number of unevaluated factors.
Although the separation of a solid phase is not regulated generally
and uniformly by its solubility and the prevailing supersaturation, it
should be especially emphasized that certain statements of von Wei-
marn have quite general validity. Thus, when the reacting solutes
are very dilute, the resulting solid precipitates in a definitely crys-
talline form. At low concentrations, the velocity of formation of nu-
clei is small and the few nuclei which do form grow so slowly at the
cost of the solute present in weakly supersaturated solution that large
crystals result. Conversely, when the reacting solutions are very con-
centrated, jellies are formed consisting of drops of the liquid sur-
rounded by the solid. The rate of formation of nuclei is extremely
high at the high concentration, so that at any point where the two
solutions touch there is formed immediately a feltwork of nuclei
which do not have time to crystallize further, and so yield a solid
film. Subsequently, the nuclei grow into crystals as the interacting
substances diffuse through the film. 27
In contradistinction to von Weimarn's views, Haber 28 is of the
*Hammett and Deyrup: J Am. Chem. Soc, 55, 1900 (1933).
**Cf. Freundlich: "Kapillarchemie," 2nd ed, 632 (1922).
2 Ben, 56B, 1717 (1922).
THE ANALYTICAL PRECIPITATE 25
opinion that the form of precipitates is influenced primarily by the
following two factors: aggregation velocity and orientation velocity.
When the limit of solubility is exceeded for a given substance, the
molecules or molecular aggregates tend to group together into larger
aggregates. The velocity of this process is a function of the super-
saturation ; hence, the higher the supersaturation, the less regular will
be the aggregates. The absolute concentration of the reacting ions
will also be of importance. By mixing the molecules in a more or
less arbitrary manner, instable aggregates are formed which lose en-
ergy and become oriented in a regular way in the crystal lattice. The
speed of this process is termed the orientation velocity. From Haber's
point of view the form of a given precipitate depends upon its rela-
tive velocity of orientation and of aggregation : if the supersaturation
is very high, the aggregation velocity will predominate and the result-
ing precipitate will be amorphous to x-rays; whereas, if the super-
saturation is sufficiently low, the orientation velocity predominates
and the precipitate is crystalline. A precipitate, amorphous to x-rays,
becomes crystalline during the aging process. Since the orientation
velocity varies with different substances, it follows that, at the same
degree of supersaturation, the form of a precipitate will vary with
different substances.
The Analytical Precipitate
As noted above, barium sulfate comes down in a less finely divided,
more readily filterable form when precipitated with excess sulfate.
But since the analyst is usually called upon to determine sulfate rather
than barium, the question of getting the precipitate in the proper form
for quantitative filtration has received considerable attention. A
readily filterable precipitate is ordinarily obtained by adding the
barium salt drop by drop with vigorous stirring to the boiling sulfate
solution containing a small amount of hydrochloric acid. Since the
solubility of barium sulfate is increased greatly by the presence of
acid, the concentration of the acid is ordinarily kept quite low. Mur-
mann 29 points out that complete precipitation results in the presence
of a relatively large amount of hydrochloric acid provided ethyl alco-
hol is added and the filtration carried out in the cold.
Although slow addition of barium chloride to the hot sulfate solu-
tion acidified slightly with hydrochloric acid ordinarily yields a granu-
lar, readily filterable salt, an occasional precipitate is obtained which
M Oesterr. Chem.-Ztg., 18, 227 (1911).
26 COLLOIDAL BARIUM SULFATE
is too fine to be retained, even by a close filter. It is common practice
to digest such a precipitate at the boiling point until it takes on the
desired physical character. The coalescence of barium sulfate pre-
cipitates by digestion is ordinarily attributed to the growth of larger
particles at the expense of smaller ones, but Trimble 30 showed that
the solution pressure of barium sulfate at about 100 ceases to be a
function of particle size for particles larger than about 2ji in diameter.
Oden 81 boiled a suspension of barium sulfate for 100 hours and ob-
served that the number of particles under 0.2/i in radius was decreased
only from 47% to 30%. The observed coalescence by digestion in
the mother liquor is apparently caused by the collection of the particles
into relatively larger clumps, followed by the cementing together of
the unit particles into aggregates which are retained by the filter.
Balarew 32 believes that the setting up of equilibrium in a suspension
of finely ground crystals of a difficultly soluble salt involves two kinds
of processes: molecular and submicronic solution together with pep-
tization and molecular and submicronic growth. Kolthoff 33 finds that
Ostwald ripening is of subordinate significance in the aging of pre-
cipitates at room temperature. Aging in the mother liquor consists
in the perfection of very imperfect crystals as a result of recrystalliza-
tion without always being accompanied by pronounced changes of the
external surface. It was found that all factors which decrease the
solubility of the precipitate inhibit the speed of the perfecting process ;
and all factors which increase the solubility promote recrystallization
and the perfecting process. It was concluded that the recrystallization
takes place in a liquid film around the primary particles and hence
that the speed of perfection is determined by the solubility in the
liquid film rather than the solubility in the bulk of the solution. To
speed up the growth of larger particles at the expense of smaller ones,
Krak 34 recommends pouring off the supernatant liquid and adding
10 cc of saturated ammonium acetate, the solvent action of which
causes the particles to grow to readily filterable dimensions. 35
80 J. Phys. Chem., 81, 601 (1927).
3i Svensk Kern. Tid, 32, 74, 90, 108 (1920) ; Chem. Abstracts, 15, 971 (1921).
"Kolloid-Z., 66, 51 (1934); Z. physik. Chem., B28, 78 (1935).
as Science, 84, 376 (1936); Kolthoff and Noponen: J. Am Chem Soc, 59,
1237 (1937).
s* Chemist-Analyst, 5, 26 (1912).
30sborne: J. Phys. Chem., 17, 629 (1913).
HISTORICAL 27
ADSORPTION BY PRECIPITATED BARIUM SULFATE
Historical
Because of the intrinsic importance of barium sulfate in quan-
titative analysis, its adsorptive power has been the subject of numerous
investigations. A quantitative study of the contamination of the pre-
cipitate formed by the addition of sulfuric acid to barium chloride
and the reverse was first carried out by Richards and Parker 36 and
by Hulett and Duschak. 87 The latter observed that barium chloride is
taken up not only during precipitation but also when finely divided
crystals are suspended in a solution of the salt. To explain the phe-
nomenon, they consider the possible formation of complex salts such
as BaClHSO4 and (BaCl) 2 SO4. 88 Schneider 39 investigated quanti-
tatively the contamination of barium sulfate by ferrous sulfate, and
Creighton 40 made a similar study of the contamination with alumi-
num sulfate. Both Schneider and Creighton regard the phenomenon
as an example of solid solution, whereas Richards 41 compared the
contamination with ferric sulfate to the occlusion of hydrogen by pal-
ladium which involves both compound formation and adsorption
(Vol. I, p. 218). The work of Kiister and Thiel 42 and of Korte, 48
who repeated and extended Schneider's experiments, indicates, how-
ever, that the contamination is an adsorption phenomenon. Indeed, an
adsorption mechanism of some sort seems to offer the only plausible
explanation of the fact that barium sulfate carries down all manner of
substances from either true 44 or colloidal 45 solutions; and most
investigators accept this explanation. 46 Thus from precise observa-
sX. anorg Chem , 8, 413 (1895).
3T Z anorg Chem., 40, 196 (1904).
38 Cf. Folm: J. Biol. Chem., 1, 131 (1906); Karaoglanow: Z. anal. Chem.
106, 129 (1936).
39 Z. physik. Chem., 10, 425 (1892); Glendmning and Edgar: Chem. News,
24, 140 (1871) ; Sloane: 44, 221 (1881) ; Jannasch and Richards: J. Prakt Cliem,
(2) 39, 321 (1889).
Z. anorg Chem., 68, 53 (1909).
41 B. anorg. Chem., 23, 383 (1900)
Z. anorg. Chem , 19, 97 (1899) ; 22, 424 (1900)
43 J Chem. Soc, 87, 1503 (1905).
"Patten: J. Am. Chem. Soc, 26, 186 (1903).
Vanino and Hartl: Ber., 87, 3620 (1904).
48 Wohlers' Z. anorg Chem., 69, 203 (1908); Allen and Johnston- J. Am
Chem. Soc., 82, 588 (1910) ; Johnston and Adams: 88, 829 (1911) ; Weiser: J.
Phys. Chem, 21, 317 (1917); Koelsch: Chem-Ztg., 43, 117 (1919); Oden: Arkiv
Kemi, Mineral. Geol., 7, No. 26, p 92 (1920); Dhar, Sen, and Chatterji: Kol-
loid-Z, 88, 29 (1923) ; cf., however, Smith: J. Am. Chem. Soc., 89, 1152 (1917) ;
Kolthoff and Vogelenzang: Z. anal. Chem., 68, 49 (1919).
28
COLLOIDAL BARIUM SULFATE
tions on the carrying down of a large number of cations by barium
sulfate Johnston and Adams 47 conclude "that this occlusion is a phe-
nomenon of adsorption at the surface of the grains of the precipitate
and that its amount depends upon (a) the composition of the original
solution, (b) the initial fineness of the precipitate, and (c) the amount
of recrystallization which has taken place." Although most people
subscribe to this general point of view there is a difference of opinion
as to the exact nature of the adsorption process and the factors which
influence it. The several mechanisms will be considered in the sub-
sequent paragraphs.
Adsorption of Cations
Adsorption of Cations of the Commoner Elements. The first
systematic study of the adsorption by barium sulfate of a series of
common cations was made by Johnston and Adams. In Table IV are
given the amounts of the several cations (expressed as the sulfate)
adsorbed by 1 g of BaSO 4 thrown down in the hot with a slight
excess of barium chloride from solutions 0.003 N in HC1 and con-
taining the metallic chlorides in the initial concentration A = and
TABLE IV
ADSORPTION OF CATIONS BY BARIUM SULFATE
Sulfate adsorbed
by 1 g of BaSO<
Metal
j.
1
B
Millicquivalent
Milhmol
Millicquivalent
Milhmol
Mg .
048
024
Li .
033
033
040
040
Na
0.029
029
043
043
K . .
033
0033
035
035
Al
051
017
060
020
Fe (ous) .
074
037
120
0060
Ni
062
031
100
050
Cu
082
041
100
050
Zn .
080
0040
100
050
Mn
128
064
160
080
Cd . .
160
080
180
090
J. Am. Chem. Soc., 33, 829 (1911).
ADSORPTION OF CATIONS 29
B = O.IN. The time of precipitation was 4 minutes, and the time of
standing 18 hours. The adsorption values obtained under these con-
ditions are of the same order of magnitude; nevertheless, they show
considerable variation. If the values are expressed in equivalents,
one can detect the tendency for ions of higher valence to be more
strongly adsorbed in accord with Schulze's rule. This tendency is
less marked when the adsorption values are expressed in mols. More-
over, when so expressed, the adsorption of trivalent aluminum is less
than that of either the univalent or bivalent ions. In this respect
aluminum appears quite different from trivalent lanthanum which
Frion 48 found to be adsorbed much more strongly than bivalent
magnesium.
Observations of adsorption on barium sulfate by Paneth 49 ' r>0 and
on the silver halides by Fajans 51 led to the conclusion that those ions
will be most strongly adsorbed by a heteropolar adsorbent whose
compounds with the oppositely charged ions of the ion lattice are
least soluble. This is known as Paneth-Fa Jans' rule. According to
this rule the adsorption should be in the order : Cu > Ni > Mg >
Cd>Mn>Zn; but the order actually obtained by Johnston and
Adams is quite different, as follows : Cd > Mn > Zn, Cu > Ni >
Mg > Al.
Although recognizing the applicability of Paneth-Fa Jans' rule
under certain conditions, Tezak 62 suggests that in the case at hand
the adsorbability should increase with the similarity of the constitu-
tion of the ions with one of the ions of the adsorbing lattice. Now
barium ion is very strongly adsorbed by barium sulfate, and it has
the same electron structure as xenon. Magnesium, on the other hand,
is the most weakly adsorbed of the bivalent cations, and it has the
configuration of neon. Since the adsorbability falls off in the scries:
xenon, krypton, argon, neon, it is deduced that the order of adsorp-
tion of the bivalent cations should be : Ba > Hg > Pb > Cd >
Mn > Zn > Cu > Ni > Mg in agreement with the above-mentioned
observations of Johnston and Adams. From the peptizing action on
secondary aggregates of barium sulfate, Tezak 53 deduces the order
of adsorption for two series of ions to be : Cd > Mn > Cu > Ni >
J. chim. phys., 7, 101 (1909).
"Physik. Z, 16, 924 (1914)
6Horovitz and Paneth: Z. physik. Chem., 89, 513 (1915).
Fajans and Beckcrath: Z physik Chcm., 97, 478 (1921)
Kolloid-Z, 69, 158 (1932).
53 Z physik. Chem., A176, 284; B32, 46, 52 (1936).
30 COLLOIDAL BARIUM SULFATE
Zn > Mg; and K > Na > Li, which is the reverse of the order of
the hydration of the ions in the two series.
The coprecipitation of ferrous sulfate with barium sulfate was at-
tributed by Giacalone and Russo 64 to both occlusion and adsorption
occurring simultaneously. It was found that the contamination at
varying concentrations may be expressed by Freundlich's adsorption
equation. Coprecipitation was prevented entirely by adding ammo-
nium thiocyanate to the solution having \% or less of ferric chloride
and warming to the point of decoloration, i.e., complete reduction,
before precipitating the barium with sulfuric acid.
Pincus and de Brouckere 55 have made extensive investigations of
the adsorption of a number of metallic halides by barium sulfate
which was precipitated, washed, kept in alcohol to prevent growth of
crystals, and finally dried at 100 before use. The star-shaped aggre-
gates below 10~ 5 cm in diameter were made up of ultramicroscopic
granules. The results of the adsorption studies indicate that the anion
and cation of the salts are adsorbed in equivalent amounts and that
the adsorption is reversible, the same equilibrium value being obtained
from both sides. Some typical adsorption curves are shown in Fig. 2,
in which the log of the adsorption, x, in mols per gram of adsorbent
is plotted against the equilibrium concentration. It will be seen that
all the isotherms consist of two rectilinear parts connected by a smooth
curve. In every isotherm the first part is an approximately straight
line which makes an angle of about 45 with the axes, indicating that
in sufficiently dilute solutions the adsorption is nearly proportional to
the electrolyte concentration. The second part of the curve, which
runs parallel to the abscissa axis, corresponds to saturation of the sur-
face. This course of the isotherm is explained by assuming that a
more or less discontinuous monomolecular adsorption layer is formed.
From this point of view the maximum amount that can be fixed on a
given surface would depend chiefly on the diameters of the adsorbed
particles, as the results seem to indicate. The amount of electrolyte
adsorbed on a positively charged surface appears to depend more on
the nature of the anion than on that of the cation; thus the valence
of the cation plays no marked role in the adsorption. De Brouckere
explains this by assuming that the anions are adsorbed directly to the
barium sulfate surface and an equivalent amount of cations are held
"Gazz. chim. ital., 66, 631 (1936).
"J. chim. phys., 26, 605 (1928); de Brouckere: 26, 250 (1929); 27, 543
(1930); Bull. soc. chim Belg, 39, 174 (1930); 46, 353 (1936); Ann. chim., 19,
79 (1933).
ADSORPTION OF CATIONS
31
by electrostatic attraction. On the other hand, with negatively
charged particles the cations are in contact with the surface. Experi-
ments carried out with lead and thallium halides show that the quan-
tity adsorbed is greater the less the solubility, in accord with Paneth-
Fajans* rule. Complex ions such as found in mercury and cadmium
halides and the I 3 anion are adsorbed to a larger extent than simple
ions. The adsorption of iron is 10 to 200 times stronger than the
cations of univalent and bivalent salts, which are not readily hydro-
Log Concentration, Mol per Liter
FIG 2. Adsorption of metallic halides by barium sulfate.
lyzed, not because of the trivalence of the iron but because the iron
in ferric chloride solutions is in the form of colloidal micelles which
are strongly and irreversibly adsorbed. 58
The solvent is adsorbed to some extent by barium sulfate
(p. 14), but the ratio KC1/H 2 O was found to be greater in the ad-
sorption film than in the solution. 87
Balarew and Koluschewa 58 claim that the adsorption mechanism
is not so simple as assumed by Tezak and de Brouckere. Thus the
56 De Brouckere: Bull. sci. acad. roy. Belg, 13, 827 (1928) ; Bull. soc. chim.
Belg., 88, 409 (1929).
"De Brouckere: Bull soc. chim. Belg., 41, 412 (1932); cf. Dumanskii:
Kolloid-Z., 66, 178 (1933).
Kolloid-Z. f 67, 203 (1934).
32 COLLOIDAL BARIUM SULFATE
order of cation adsorption is for the most part the reverse of that
deduced by Tezak provided the barium sulfate is precipitated in the
presence of a large excess of sulfate. Balarew 89 believes that most of
the adsorption is on the internal surface of the crystals and only a
relatively small part on the outermost part of the crystals. He points
out that, in a suspension of barium sulfate in potassium chloride solu-
tion, the following equilibrium is set up : 2KC1 + BaSC>4 & BaG 2 +
K 2 SO 4 . He concludes that potassium chloride is not adsorbed di-
rectly on the barium sulfate surface; on the contrary, the sulfate ions
are believed to constitute the inner portion of the adsorption layer
with the chloride ions farther from the surface. In support of this
he 60 claims that a precipitate thrown down in an almost saturated
solution of potassium chloride and containing 6.25% K 2 SC>4 and
3.11% H 2 O is free from adsorbed chloride. He believes that the
adsorption of cations follows Paneth-Fajans' rule in a given group of
the periodic system and that the adsorption of simple inorganic anions
follows the Hofmeister series.
Kolthoff 81 likewise criticizes de Brouckere's work on both experi-
mental and theoretical grounds. Kolthoff, like Balarew, believes that
most of the adsorption by crystalline precipitates takes place in the
fine cracks and capillaries which constitute the inner surface. It is
believed that in the first stage of the process the adsorbed ion dis-
places ions in the surface layer of the adsorbent. For the adsorption
of lead ion, say, the distribution coefficient K is given by the ex-
pression :
Pb (surface) Ba (surface)
~ K :
Pb (solution) Ba (solution)
Other factors which may enter into the process are: true adsorption
at active centers only; true adsorption followed by secondary precipi-
tation of the displaced lattice ion; exchange with a third kind of ion
already adsorbed on the precipitate; molecular adsorption; and acti-
vated adsorption in which the adsorbed substance may be ionized on
the surface. Kolthoff's views will be taken up in some detail in
the next chapter. The differences in the experimental results of
69 Z. anal. Chem., 72, 303 (1927) ; Kolloid-Beihefte, 80, 249 (1930) ; 32, 304;
S3, 310 (1931).
o Balarew: Kolloid-Beihefte, 30, 281 (1930).
Chem. Weekblad, 31, 230, 244, 251 (1934) ; Kolthoff and Noponen: J. Am.
Chem. Soc., 59, 1237 (1937); cf. reply by Pinkus: Bull. soc. chim. Belg., 44,
637; de Brouckere: 625 (1935).
ADSORPTION OF CATIONS
33
de Brouckere and Kolthoff are due in part at least to differences in
the nature of the adsorbent and variations in the experimental pro-
cedure. 62 The differences can be reconciled, if at all, only by each
investigator working with samples of adsorbent prepared by the
other. 63
Adsorption of Radium and Radioactive Cations. Germann 64 first
investigated the adsorption of radium by barium sulfate and showed
that the same laws which apply to the adsorption of ponderable masses
are applicable with an equal degree of accuracy to masses as small as
5 X 10~ 8 g adsorbed per gram of adsorbent. The observations were
made by adding a known amount of radium- free barium sulfate to
80 cc of a standard solution of radium barium chloride, which was
allowed to stand until adsorption equilibrium was established, after
which the barium sulfate was filtered off and the supernatant liquid
analyzed for the radium content by the emanation method. The data
are reproduced in Table V. These data are in accord with Freund-
TABLE V
ADSORPHON OF RADIUM BY BARIUM SULFATE
BaSO 4
employed = m
g
Equilibrium concentration
of Ra after adsorption = c
g/cc X 10"
Decrease in Ra concentration
clue to adsorption = x
g/cc X 10"
22 79
00
05
18 55
4 24
10
15 12
7 67
20
9 47
13 32
30
5 86
16 93
40
3 97
18 82
50
2 70
20 09
lich's adsorption formula xfm = ac 1/T1 , where x is the amount ad-
sorbed by mass m of adsorbent, c the equilibrium concentration, and
a and n constants, as evidenced by the fact that a straight line is
obtained by plotting log x/m against log c.
In addition to radium, the active substance adsorbed by barium
sulfate, when it is precipitated in solutions containing the radioactive
De Brouckere: Bull. soc. chim. Belg, 46, 279 (1936).
C/. Kolthoff: Bull. soc. chim. Belg., 46, 270 (1936).
* J. Am. Chem. Soc., 48, 1615 (1921).
34
COLLOIDAL BARIUM SULFATE
constituents of uranium ores, is chiefly ionium, 65 accompanied by a
relatively small amount of actinium, and not actinium alone as claimed
by Debierne. 66 Strong adsorption of ionium by barium sulfate is
indicated by the fact that the salt has a very great capacity for ad-
sorbing thorium, 67 the isotope of ionium.
Paneth 49 and Horovitz 50 - 68 observed that radium is strongly ad-
sorbed by barium sulfate and chromate whereas it is not adsorbed at
all by hydrous chromic oxide and silver chloride. Since radium sul-
fate and chromate are soluble whereas radium oxide and chloride are
not, Paneth was led to formulate what has come to be called Paneth-
Fajans' rule (p. 103).
More recently Imre 69 has studied the velocity of adsorption of
actinium, thorium B, radioactive lead, and radium on barium sulfate,
so
Lanthanum (Actinium)
Lead 01 onum B)
12
16
Time, Hours
FIG. 3. Velocity of adsorption of actinium and thorium B by barium sulfate.
confirming the observation of Paneth 50 - 88 that adsorption equilibrium
is established only after a considerable time. 70 This is illustrated in
Fig. 3. Although lead sulfate is less soluble than actinium sulfate,
lead ion is adsorbed less strongly than actinium ion which is not in
accord with Paneth-Fa Jans' rule ; instead the ion of higher valence is
the more strongly adsorbed, in accord with Schulze's rule.
65 Kammer and Silverman: J. Am. Chem. Soc.. 47, 2514 (192S).
66 Compt rend., 129, 593 (1899); 180, 906 (1900).
"Balcar and Stegemann: J. Phys. Chem., 82, 1411 (1928).
6 <> Paneth and Vorwerk: Z. physik. Chem., 101, 445 (1922).
Z. physik. Chem., A163, 262 (1931) ; AIM, 327, 343, 364 (1933) ; A171, 239
(1934); Z. Elektrochem., 88, 539 (1932).
70 C/. Kolthoff and Rosenblum: J. Am. Chem. Soc., 68, 116 (1936).
ADSORPTION OF CATIONS 35
The adsorption from a lead nitrate solution is believed by Imre 71
to take place in two stages: in the first, the surface of the barium
sulfate lattice is covered with a layer of adsorbed lead nitrate ; and in
the second, the adsorbed ion is embodied in the solid surface layer.
Kolthoff and MacNevin, 72 on the other hand, found that lead nitrate
is not appreciably adsorbed by aged barium sulfate and that the
measured adsorption of lead (ThB) is the result of an exchange be-
tween barium ions in the surface of the adsorbent and lead ions:
BaSO 4 + Pb++ -> PbSO 4 + Ba++
Surface Solution Surface Solution
virtually no nitrate ions being carried down (cf., however, p. 37).
By making certain assumptions, Imre derived the expression for
the coefficient of distribution, K, of ions between the adsorption sur-
face and the solution, which for lead ion on barium sulfate is:
K e (Epi>-EB & )/RT t j n w hich Epb and B a are the adsorption en-
ergy of lead ions and barium ions, respectively, on barium sulfate. The
adsorption energy E is calculated from the expression E =L/N, in
which L is the heat of solution of the salt and N the number of ions
in it. On such theoretical grounds, Imre calculated the value of A'
to be 0.64 at 20. From his experimental work on the change of
adsorption with temperature, of lead (ThB) on barium sulfate, Imre
calculated a value of K = 0.210.39. Kolthoff and, MacNevin made
direct measurements of the distribution of lead sulfate between water
and barium sulfate on small but well-aged crystals and found the
value of K=ie (Epl >- EB ti /RT =0.l2; and for the distribution be-
tween SQ% ethyl alcohol and barium sulfate, K 0.067. Kolthoff be-
lieves that the first step in the adsorption consists of a rapid exchange
between lead and barium ions in the surface of the adsorbent, followed
by a slow incorporation of the exchanged lead in the barium sulfate
lattice as a result of recrystallization. The lead adsorbed as a result
of the kinetic exchange is readily displaced by surrounding the crys-
tals with a large excess of barium salt. Although Kolthoff s views
are different in some respects from those of Imre, it would appear
that the general concept of both is similar in essential respects. 73
Kolthoff 74 found that the surface of aged precipitates may be de-
Z. physik. Chem., A171, 239 (1934).
72 J. Am. Chem. Soc., 68, 499, 725 (1936).
73 Cf. Imre: Z. physik. Chem., A177, 409 (1936).
Kolthoff: Chem. Weekblad, 81, 395 (1934); cf. Kolthoff and MacNevin:
J. Am. Chem. Soc., 58, 725 (1936).
36 COLLOIDAL BARIUM SULFATE
termined from adsorption of chemically identical ions. In such cases
the total amount adsorbed on the surface equals that which has entered
the lattice. This equality holds for ThB++ on lead sulfate but does
not hold for ThB++ or Ra++ on barium sulfate where the adsorption
is about double that in the crystal. The adsorption of radioactive ions
can therefore be applied to calculate the specific surface only for iso-
topic ions; hence the expression of Paneth and Thimann 76 is not gen-
erally applicable (p. SO).
Adsorption of Anions
When an alkali is precipitated with barium chloride in slight ex-
cess, the most usual analytical procedure, the determinations are too
low, since some sulfate is weighed as alkali sulfate and calculated as
if it were pure barium sulfate. 78 Opposed to this is the adsorption
of chloride probably as barium chloride, which tends to make the
analytical results too high. The latter effect manifests itself especially
in the precipitation of sulfuric acid by barium chloride and the reverse.
Ilulett and Duschak 37 ' 77 found it possible to obtain exact results in
such determinations by estimating the chlorine content of the pre-
cipitate and deducting the barium chloride equivalent from the weight
of the crude barium sulfate. Although the adsorption of chloride ion
by barium sulfate is appreciable, Mendelejeff 78 long ago showed it
to be small compared to that of nitrate. It is therefore somewhat
surprising that Kolthoff failed to observe any adsorption of nitrate
from a lead nitrate solution. Mendele Jeff's observation was confirmed
in the author's laboratory 79 in the course of an investigation of ad-
sorption of various ions by barium sulfate formed on mixing sodium
sulfate with a definite excess of the barium salts of the respective
ions. The extent of the adsorption was determined by direct analysis
of the washed precipitate. In Table VI the ions are arranged in the
order of equivalent adsorption, beginning with the most strongly ad-
sorbed ferrocyanide. The adsorption in mols and the solubility of the
"Ber, 57B, 1215 (1924).
7 C/. Hahn and Keim: Z. anorg. Chem., 206, 398 (1932).
"C/. Karaoglanow: Z. anal. Chem., 57, 77 (1918).
^Pogg. Ann., 55, 214 (1842).
Weiser and Shernck: J. Phys Chem, 23, 205 (1919); cf Ghosh and
Dhar: Kolloid-Z., 35, 144 (1924) ; Chakravarti and Dhar: 44, 63; 45, 12 (1928) ;
von Weimarn: Repts. Imp. Ind. Research Inst., Osaka, Japan, 12, 153 (1931);
Balarew, Koluschewa, and Totewa: Kolloid-Beihefte, 33, 299 (1932); Chao,
Hsiung, and Chu: J. Chinese Chem. Soc, 3, 325 (1935) ; Schneider and Rieman,
III: J. Am. Chem. Soc., 59, 354 (1937).
ADSORPTION OF ANIONS
37
several barium salts are also included in the table. From a considera-
tion of the absolute amount of the adsorption in each case, there is
little to suggest Schulze's rule ; for, although a tetravalent ion appears
to be adsorbed most strongly, four univalent ions are more strongly
adsorbed than trivalent ferricyanide. Furthermore, contrary to what
is implied in Schulze's rule, there is a wide variation in the amount
of univalent ions adsorbed, nitrate being carried down 150 times more
strongly than iodide (cf. p. 42).
TABLE VI
ADSORPHON OF ANIONS BY BARIUM SUL*ATE
Adsorption by 100 molt. BaSO 4
Solubility of
barium salts,
Anion
Ion in
milhmols per
excess
Gram
Gram
gram water
equivalents
mols
at 25
Ferrocyanicle
Ba++
13 20
3 30
07
Nitrate
Ba + ->
8 48
8 48
40
Nitrite
Ba + +
7 47
7 47
3 10
Chlorate
Ba ++
5 84
5 84
1 25
Permanganate
Ba^
2 85
2 85
1 93
Ferricyanide
Ba*'
2 70
90
Very soluble
Chloride
Ba+ +
1 76
1 76
1 78
Bromide
Ba- +
83
83
3 57
Cyanide
Ba+ +
31
31
4 25
Sulfocyanate
Ba +f
22
22
6 13
Iodide
Ba + +
06
06
5 43
Chloride
so 4
125
125
1 78
Chlorate
so 4
227
227
1 25
Permanganate
so 4
137
137
1 93
Considering the relationship between the solubility of the several
barium salts and the adsorbability of the respective anions by barium
sulfate, one can note a tendency for the anions of the less soluble
salts to be more strongly adsorbed, in accord with Paneth-Fa Jans'
rule. There are, however, numerous exceptions to this qualitative
statement. Thus nitrate ions are adsorbed to approximately the same
extent as nitrite ions although they are but one-eighth as soluble.
Kolthoff and MacNevin 80 determined the adsorption by barium sul-
w J. Am. Chem. Soc., 58, 1543 (1936).
38
COLLOIDAL BARIUM SULFATE
fate of a number of univalent anions from 50% ethyl alcohol. The
results are given in Table VII. The adsorption follows the Freund-
TABLE VII
SOLUBILITY AND ABSORBABILITY
Order of
Relative solu-
Order of
Barium salt
adsorbabihty
flX 10*
bility in 50%
ethyl alcohol
increasing
solubility
Bromate
1 80
1
1
Formate
91
81
3
Nitrate
69
33
2
Perrhlorate
59
1610
7
Chloride
54
223
4
Bromide
37
843
5
Iodide
35
1477
6
Thiocyanate
26
1663
8
lich adsorption isotherm, x/m = oc 1/n (for discussion see p. 101), and
the order of adsorbability is expressed in a X 10 4 . The order is the
same as in Table VI for ions common to both series but there is no
close parallelism between adsorbability and solubility. Formate is ad-
sorbed more strongly than nitrate although the nitrate is much less
soluble. Perchlorate likewise is very much out of place, its adsorp-
tion being similar to that of chloride and much greater than that of
bromide or iodide although its solubility in 50% ethyl alcohol is
greater than that of the halides. Karaoglanow 81 found that the con-
tamination by various anions is related to the solubility of the corre-
sponding barium salt and that of cations to the solubility of the corre-
sponding sulfate. He attributes the contamination to secondary
chemical precipitation rather than to adsorption.
Although chemically dissimilar ions of the same valence may show
a wide variation in the degree of adsorption, thus indicating that ad-
sorbability is a specific property of ions, the similarity in the adsorp-
tion of nitrate and nitrite ions, which are more nearly related chemi-
cally, suggests that the adsorption of ions by a given disperse phase
is determined by two factors: the valence of the ion, and the specific
adsorbability of the ion, which is influenced by the solubility of the
salt formed with the opposite ion of the crystal lattice, and the radius
"Z. physik, Chem., A178, 143 (1937).
ADSORPTION OF ANIONS 39
and deformability of the ion (cf. p. 128). By choosing a series
of ions of much the same general character, thus minimizing the
specific factor, it is possible to emphasize the effect of valence. This
is well illustrated by the series of cyanides, the order of adsorption
of which is: Fe(CN) 6 > Fe(CN) 6 > CN > CNS.
It should be pointed out that data on adsorption of various ions by
barium sulfate will be strictly comparable only when the conditions of
precipitation, the size of the particles of the adsorbent, the ions in
excess, etc., are constant. De Brouckere 82 found, for example, that,
under conditions of adsorption entirely different from those described
above, the order of adsorption of the halides is : I > Br > Cl. This
is the reverse of what would be predicted by Paneth-Fa Jans' rule
but is the order of decreasing size and deformability of the halo-
gen ions.
Referring to Table VI, it will be seen that the adsorption is very
much greater when the precipitation is carried out with barium ion in
excess than with sulfate ion in excess. This was attributed by Weiscr
and Sherrick to the well-known difference in the physical character
of the particles precipitated under the two sets of conditions; but
from observations on adsorption by calcium oxalate Kolthoff and
Sandell 83 concluded that, in general, the occlusion of foreign ions by
internal adsorption is always greater on precipitation in the presence
of an excess of a lattice cation than in the presence of an excess of a
lattice anion.
Contamination by Potassium Permanganate and Barium Nitrate.
The precipitation of barium sulfate in the presence of potassium per-
manganate is said by Grimm 8 * to give well-defined mixed crystals of
a red color which are not decolorized by being warmed with oxalate
for two months. Grimm and Wagner 85 obtained precipitates contain-
ing as much as 80 mol per cent potassium permanganate and showed
the relationship between the permanganate composition of the crystals
Ccr and the solution from which they separate C 9 to be given by the
expression Ccr = kC 9 Moreover, Wagner showed that the x-ray dif-
fraction lines of barium sulfate are displaced toward the lines of the
isomorphous potassium permanganate, in proportion to the perman-
ganate content of the crystals. The samples were prepared by pre-
ss J. chim. phys., 27, 543 (1930).
w J. Phys. Chem., 37, 443, 459 (1933).
8 *Z. Elektrochem., 80, 467 (1924); Naturwissenschaften, 15, 561 (1927); cf.
Huttig and Menzel: Z. anal. Chem., 68, 343 (1926).
w Z. physik. Chem., 182, 131 (1928); Wagner: B2, 27 (1929).
40 COLLOIDAL BARIUM SULFATE
cipitation at 50 in the presence of potassium permanganate, centri fug-
ing and washing with water, which gave compositions below 8 mol
per cent permanganate. Higher concentrations were obtained by
washing with acetone only; and the highest concentrations by not
washing at all. The samples were dried in vacuum at 110-120 C.
Similar observations with a sample of barium sulfate containing 15.7
mol per cent barium nitrate gave no apparent displacement of the
barium sulfate diffraction lines.
The above observations would appear to offer conclusive evidence
that potassium permanganate forms mixed crystals with barium sul-
fate. 86 Balarew and coworkers 87 are equally certain that the potas-
sium permanganate is held by adsorption on the inner surface of the
barium sulfate crystals. A product colored similarly to that of Grimm
is obtained by allowing potassium sulfate in saturated potassium per-
manganate to diffuse through a porous cup into barium chloride
solution saturated with potassium permanganate; but this product is
completely decolorized by washing with oxalate and at the same time
breaks up into minute prisms. Since mixed crystal formation should
not be prevented by a change in the velocity of crystallization, the
obvious conclusion is that the red precipitate is not homogeneous but
consists of minute prisms of barium sulfate colored by adsorbed per-
manganate. A freshly precipitated sample of barium sulfate takes up
permanganate that is not removed by prolonged contact with solutions
of sulfurous acid, oxalic acid, or hydrogen peroxide. 88 The amount
of permanganate taken up decreases with the age of the barium sul-
fate, and a well-aged sample is not colored at all in contact with the
permanganate solution. These and other facts support the view that
the permanganate is held by adsorption on the inner surfaces of
barium sulfate, and not in solid solution. The most convincing argu-
ment in support of the solid-solution theory is the apparent displace-
ment of the x-ray diffraction lines observed by Wagner; but Bala-
rew 89 considers this evidence to be misleading. The samples which
Wagner examined contained a large amount of potassium perman-
86 C/., also, Karaoglanow and Sagortschev: Z. anorg. Chem, 221, 369; 222,
249 (1935).
87 Balarew Z anorg. Chem., 166, 301 (1926); Balarew and Kaischew: 167,
237; 168, 154 (1927) ; Balarew and Kandilarow: 162, 344 (1927) ; Balarew: 169,
257 (1928) ; Balarew, Kaischew, and Srebrow 174, 295 (1928) ; cf. Koluschewa
and Scwrugowa: Kolloid-Z., 60, 141 (1932).
88 Cf. Gcilmami: Z. anorg. Chem., 169, 271 (1926); Kolthoff and Sandell:
J Phys. Chem., 87, 723 (1933).
* Balarew and Lukowa: Kolloid-Beihefte, 82, 304 (1931).
ADSORPTION OF ANIONS
41
ganate entrained in vacuoles in the precipitate. Balarew thinks that,
when the precipitates are dried in vacuum at 120, the excess per-
manganate crystallizes on the surface of the barium sulfate crystals,
making layers of crystals of two different compounds that will give
an x-radiogram slightly different from either. If this be true, then
8.880
0<
E
8.870
8860
5.460
5 5.450
fO
160
- 7. 150
7.140
1234
Weight Percent Contammant=100-% BaS0 4
FIG. 4 Effect of permanganate contamination on the lattice parameters of
barium sulfate crystals.
the x-radiogram of undricd samples of barium sulfate containing
permanganate should be identical with that of pure barium sulfate.
That they would not be exactly the same is indicated by Averell and
Walden's 90 results of an x-ray study of permanganate-contaminated
crystals prepared by Grimm and Wagner's procedure, in the absence
of all foreign ions except permanganate and hydrogen. In Fig. 4 are
80 J Am. Chem Soc., 59, 906 (1937).
42 COLLOIDAL BARIUM SULFATE
plotted the lattice parameters a fc c in Angstrom units against the
weight percentage of total contaminant (water + HMnO 4 ) present
in the crystals. The radii of the circles around the experimental
points are the probable errors determined by the least squares analysis
of the films. The lattice parameters vary more regularly with the
percentage of total contaminant than with the percentage of perman-
ganic acid alone, indicating that the expansion of the lattice is caused
by the entry of both permanganic acid and water in varying degrees.
Since the lattice parameters vary with increasing percentage of con-
taminant, the solid-solution character of the precipitates is indicated.
This does not mean that all or even the largest percentage of con-
taminant is held in solid solution. Since the increase in the lattice
parameters is small, it may be that most of the foreign material is
held by adsorption.
From a phase-rule study of the system BaSO 4 -KMnO 4 , Benrath
and Schackmann 91 found that no solid solution of the two constituents
exists in equilibrium with the mother liquor. There is, however, no
reason to believe that the precipitates prepared by Grimm and Wagner
are the result of heterogeneous equilibria but rather of kinetic proc-
esses localized at the surface of the growing crystal.
In this connection Walden and Cohen 92 made a precise x-ray study
of barium sulfate contaminated with barium nitrate and came to the
conclusion that a part at least of the nitrate was held in solid solution.
The data using calcium radiation are summarized in Fig. 5, which
shows the relation between the lattice parameters (accurate to less
than 0.01%) and the nitrate content of the crystals. This orderly
variation of the lattice spacing with the nitrate content, as well as an
observed increase in volume of the unit cell of the contaminated
barium sulfate, indicates that the foreign ions have actually entered
the lattice giving a solid solution. But, as already pointed out, the
formation of a solid solution does not exclude surface adsorption as
a contributing factor in the contamination. Indeed, Schneider and
Rieman, III, 93 attribute the contamination by nitrate chiefly to ad-
sorption and the contamination by nitrite chiefly to solid-solution for-
mation. In support of this they showed that: (1) the excess of lattice
ion influences the coprecipitation of nitrite ion much less than it does
that of iodide, bromide, chloride, nitrate, and chlorate ; (2) confirming
the observations of Weiser and Sherrick, all the above ions except
"Z. anorg. Chem., 218, 139 (1934)
J. Am. Chem. Soc., 57, 2591 (1935).
M J. Am. Chem. Soc., 59, 354 (1937).
ADSORPTION OF ORGANIC IONS
43
nitrite follow Paneth-Fajans' rule; (3) the crystal size is decreased
much less by the carrying down of nitrite than of the other ions; (4)
the nitrite ion is removed by digestion more slowly than the other
ions. In the light of these observations it is still an open question to
8.890
8.880
e<
E
8870
8860
5.448
5440
7.150
o<
S
7.140
X"
oo
<tf-
1234
Weight Percent Nitrate -100-% BaS0 4
FIG. 5. Effect of nitrate contamination on the lattice parameters of barium
sulfate crystals.
what extent the contamination by nitrate as well as by permanganate
is due to adsorption and to what extent to solid solution.
Adsorption of Organic Ions
Studies on the adsorption of organic substances by barium sulfate
contribute nothing especially new. Thus Rossi and Scandellari 94
showed that the adsorption of a colloidally dispersed dye like night
blue increased with increasing degree of dispersion of the adsorbent
and the adsorbed substance. Hel'd 95 showed that electrolytes influ-
ence the adsorption of nonylic acid according to well-established rules.
For example, a strongly adsorbed cation like barium increases the ad-
sorption of the nonylate anion, and a strongly adsorbed anion like
"Gazz chim. ital., 68, 261 (1933).
Hel'd and Dyachkov: Compt rend. acad. sci. U.S.S.R., 1, 193; Hcl'd and
Samokhvalov: 263 (1934); J. Phys. Chem. (U.S.S.R.), 6, 1210 (1935); Kol-
loid-Z., 72, 13 (1935).
44 COLLOIDAL BARIUM SULFATE
sulfatc cuts down the adsorption of the nonylate ion. The adsorption
of sodium laurate apparently gives a film two molecules thick on
barium sulfate. It is claimed that the first layer is held by chemical
attraction and the second by van der 5 Waals 1 forces. Flotation of
barium sulfate by sodium laurate was found to be at a maximum
when the first layer was saturated. The heat of adsorption of sodium
oleatc was estimated as +0.067 cal./g BaSO 4 . Bartell and Hersh-
berger 97 showed that barium sulfate is more highly wetted by organic
liquid than the sulfide pigments (p. 280).
ELECTRICAL PROPERTIES OF PRECIPITATED BARIUM SULFATE
Oyemant 98 studied the effect of barium ion on the clectrokinetic
potential, , at the interface of finely ground heavy spar and water.
The ^-potential was determined from electrocndosmotic data by substi-
tuting in the formula : "
where -q is the viscosity at room temperature, M the volume of liquid
transferred through the diaphragm in unit time, I) the dielectric con-
stant, / the current strength, and o- the specific resistance of liquid.
Since the solubility product of barium sulfate is KH , it follows that
/>Ba + pSO = 10, where />Ba is the negative logarithm of the barium
ion concentration. The sample of barium sulfate was negative in pure
water; at />Ba = 7, i.e., 0.001 N Na 2 SO 4> the cataphoretic velocity
and hence the f -potential was zero This, therefore, represents the
point of change of sign of the potential. The small positive potential
remains almost constant to />Ba = 4 then falls rapidly to a minimum
at />Ba = 2. Kruyt and Ruyssen 10 likewise showed that natural
bante possesses an appreciable negative charge in pure water. The
sign of the potential depends, however, on the extent of surface : nat-
ural barite of particle size 150ft* in diameter has a larger negative
potential than precipitated particles 20-40/* in diameter; and particles
around lO/i, have only a very slight negative potential which is changed
M Hel'd and Khainskn: Kolloid-Z, 75, 287 (1936).
wind. Eng. Chcm, 22, 1304 (1930)
M Z physik Clicm, 103, 260 (1922)
99 Cf. Freundhch: "Kapillarchemie," 2nd ed., 331 (1922).
10 Proc. Acad. Sci. Amsterdam, 37, 624 (1934); Oyama: J. Chem Soc.
Japan, 54, 582 (1933).
ADSORPTION OF ORGANIC IONS
45
to positive with a trace of barium chloride, because of the strong ad-
sorption of barium ion. The potential is always negative in sulfatc
solution on account of the strong adsorption of sulfate ion 1(U Some
data showing the relative effect of barium chloride and sodium sulfate
on the (-potential are given graphically in Fig. 6. Chlorides of the
bivalent metals and aluminum give a positive (-potential to the par-
ticles because of stronger adsorption of cations than of the anion. If
the (-potential is a measure of the adsorption of the cations, the
metals arrange themselves in the order : Al > Sr > Ca > Zn >
-20
+20
100 200 300 400
Electrolyte Concentration, Millimols per Liter
FIG 6. Effect of barium chloride and sodium sulfale on the (-potential of
barium sulfate
Mg > Cd. This is not the same as the order of adsorption given above
(p. 29).
It is a simple matter, by means of Congo red, to differentiate be-
tween colloidal particles or crystals that are positively charged owing
to preferential adsorption of barium ion and those that are negatively
charged owing to preferential adsorption of sulfate ion. 102 The posi-
tively charged particles adsorb the Congo red anion giving a red to
deep violet-red precipitate; the negatively charged particles are not
colored at all. At a pH of 3.3-3.5, a blue precipitate is obtained
Cf, also, Mattson- J Phys. Chem , 32, 1532 (1928).
102 Riegel and Riddson- Kolloid-Z , 64, 304 (1933).
46 COLLOIDAL BARIUM SULFATE
which consists of a mixture of the blue Congo acid and colorless crys-
tals of barium sulfate.
BARIUM SULFATE SOLS
Preparation
Von Weimarn obtained momentary sols of barium sulfate by mix-
ing solutions of barium thiocyanate and manganese sulfate of ap-
proximately 0.0008 normality. In spite of the dilution, the particles
agglomerate and settle out in a short time. Kato 103 diluted a
MH 2 SC>4 solution with twice its volume of alcohol and added it to
an equivalent amount of M Ba(C 2 H 3 O 2 )2 diluted with 5 times its
volume of alcohol. On evaporating the resulting gelatinous precipi-
tate and milky sol to dryness under reduced pressure below 40 , a
translucent casein-like residue was obtained which was readily and
completely dispersed in water to a stable fluorescent sol. The sol was
positively charged and showed the usual precipitation reaction towards
anions. Cations of high valency stabilized the sol, barium chloride
and nitrate precipitating it only at high concentrations. Thorne and
Smith 104 prepared a similar sol using methyl instead of ethyl alcohol,
and Recoura 105 mixed equivalent solutions of barium ethylate and
sulfuric acid in pure glycerol followed by dilution with water.
Since the charge on the barium sulfate particles becomes negative
when the sulfate ion concentration of the surrounding liquid is
above 0001 M, it should be possible to prepare a stable negative sol
in the presence of sulfate or other readily adsorbed anion. Such a
sol is obtained by adding a slight excess of 0.1 N K 2 SO 4 to 0.1
N BaCl 2 using as solvent 1 part of water to 5 parts of glycerol. 106
A negative hydrosol is also obtained by double decomposition of
barium salts and sulfates in the presence of sodium citrate. Such sols
are almost clear, and Spiller 10T believed them to be true solutions of
a double salt of barium sulfate and citrate. Nichols and Thies 108
showed, however, that the apparent solutions are merely negatively
charged dispersions of barium sulfate in which the particles are in an
"a Mem. Coll. Sci. f Kyoto Imp. Univ., 2, 187 (1909-10) ; Lott: J. Am. Pharm
Assoc., 17, 454 (1928).
"'Kolloid-Z, 42, 328 (1927).
105 Compt. rend., 146, 1274 (1908).
loewciser- J Thys. Chem., 21, 318 (1917).
107 J. Chem. Soc, 10, 110 (1858).
108 J. Am. Chem. Soc., 48, 303 (1926).
PROPERTIES 47
extremely fine state of subdivision. Quantitative estimations involv-
ing the precipitation of barium sulfatc can be made in the presence
of citrates, provided slightly more than enough acid is added to con-
vert the citrates into citric acid, thereby cutting down the concentra-
tion of citrate ion below the point necessary for peptizatton of barium
sulfate.
Feilmann 109 prepared a sol of barium sulfate by precipitation in
a slightly alkaline solution of commercial casein. Since the reaction
of the sol towards acid and alkalis is that of casein, it is obvious that
the colloidal barium sulfate is surrounded by a protecting film of
proteid. The hydrous oxides of iron, aluminum, chromium, and
thorium are adsorbed by colloidally dispersed barium sulfatc particles
giving stable sols.
Properties
The sols of barium sulfate are faintly opalescent to distinctly milky
in appearance, depending on the concentration and the size of par-
ticles. Bechhold and Hebler 110 investigated nephelometrically the
connection between turbidity and the size of suspended particles of
barium sulfate in various mixtures of ethyl alcohol and glycerol. For
suspensions containing the same amount of barium sulfate, but of
varying degrees of dispersion, the turbidity increases from 2.5 /*
downwards. The maximum turbidity is reached with particles of
about 800 mp, for white light, that is, the region of the extreme red.
Further reduction of particle size causes a marked decrease in tur-
bidity. Rayleigh's law, I = K(nv*/\*), in which / is the intensity of
the scattered light, n the number of particles per unit volume, v the
volume of the particles, X the wave length of the light, and K a con-
stant, holds only for the region of size below 800 m/*. If a standard
sol of known turbidity is available, it is possible to estimate the size
of amicrons and submicrons from the relation between turbidity and
particle size.
Owe ni showed that the turbidity of barium sulfate sols, as meas-
ured by the nephelometer, is different for particles of the same size
formed in glycerol- water and glycerol-alcohol mixtures. The differ-
ence is probably due to variations in crystalline form and in the re-
fractive indices of the salt thrown down in the different media. The
109 Chem. News, 98, 310 (1908).
"o Kolloid-Z, 31, 70 (1922).
* Kolloid-Z. f 82, 73 (1923).
48 COLLOIDAL BARIUM SULFATE
turbidity in such mixtures is a maximum with particles around 200 m/t
in reflected light, and around 1000 m/* in transmitted light.
Svedberg and Nichols 112 investigated the size and distribution of
size of particles by a centrifugal method for barium sulfate sol pre-
pared by the interaction of 0.1 N Ba(SCN) 2 and 0.1 N (NH 4 ) 2 SO 4
in the presence of potassium citrate.
112 J. Am. Chcm. Soc., 45, 2910 (1923) ; cf. Nichols and Liebe: Colloid Sym-
posium Monograph, 3, 268 (1925).
CHAPTER III
COLLOIDAL SULFATES OF LEAD AND STRONTIUM
COLLOIDAL LEAD SULFATE
Lead sulfate cannot be thrown down as a transparent jelly from
aqueous solution like barium sulfate, which is only one-fifteenth as
soluble It can be obtained in the sol state by mixing solutions of
lead nitrate and sulfuric acid as dilute as 0005 N. With a small ex-
cess of lead nitrate, the particles are positively charged by selective
adsorption of lead ions ; and, with a small excess of sulfuric acid, the
particles are negatively charged by selective adsorption of sulfate
ions. 1 In the absence of protecting colloids, the sol coagulates in a
short time ; but Leuze 2 prepared a stable sol by treating a mixed solu-
tion of lead acetate, sodium hydroxide, and sodium "protalbinate"
with a solution of sodium sulfate. On dialysis, the protected brown
sol which first formed became gradually opalescent and milky.
Colloidal Behavior of Lead Sulfate in the Storage Battery. In
the lead storage battery, both the spongy lead and lead peroxide plates
are formed in sulfuric acid solution. The first step in forming the
lead peroxide plate consists in the precipitation of colloidal lead sul-
fate which is negatively charged and so is carried to the lead anode.
The sulfate is held on the anode by the cataphoresis effect and to a
certain extent by adsorption, in a condition for the second step, the
oxidation to peroxide. If an alternating current is passed between
lead electrodes, lead sulfate formed at one moment is carried away
by cataphoresis on the reversal and at both electrodes, resulting in a
precipitate of lead sulfate at the bottom of the container. No ap-
preciable amount of lead peroxide is formed using a fairly rapid
alternating current, indicating that adsorption of lead sulfate by the
lead grid is of minor importance in comparison with the cataphoresis
effect. As is well known, this is not true in the slow reversals of
Mewett: J. Phys Chem , 33, 1024 (1929).
2 "Zur Kenntnis Kolloidaler Metalle und ihrer Verbindungen," Erlangen, 21
(1904).
49
50 COLLOIDAL SULFATES OF LEAD AND STRONTIUM
actual practice: when the spongy lead plate of the battery is con-
verted in part into lead sulfate on discharge, the sulfate is not carried
away by cataphoresis when the discharged plate is made cathode, but
is reduced back to spongy lead. Similarly, when a spongy lead plate
is formed by alternate oxidation and reduction, the sulfate formed
on the oxidation is reduced to lead when the current is reversed and
is not carried away by cataphoresis. The differences in behavior on
rapid reversals and slow reversals of the current indicate that lead
sulfate on the surface of lead changes with time so that the tendency
of the particles to move to the anode in sulfuric acid solution is
negligible compared to other factors. Since suspended lead sulfate
formed in sulfuric acid moves to the anode hours after its prepara-
tion, Jewett 1 believes the important factor to be an increased ad-
herence to the lead. It is probable that with time an oriented adsorp-
tion of lead sulfate on lead takes place with the result that the
sulfate is held more firmly to the lead, The relatively large crystals
of aged lead sulfate in the plates of a "sulfated" battery are very
difficult to oxidize and reduce at the respective electrodes.
ADSORPTION ON THE LEAD SULFATE LATTICE
Extent of Surf ace from Adsorption Measurements
An ionic precipitate such as lead sulfate is in kinetic equilibrium
with the surrounding solution, the speed with which the ions leave
the surface being equal to the speed with which the ions from the
solution deposit on the surface. Paneth * used this principle in deter-
mining the specific surface of lead sulfate from measurements of the
adsorption of thorium B, an isotope of lead. Paneth's fundamental
equation representing the distribution when kinetic exchange equi-
librium is established is:
Isotope (surface) __ Element (surface)
Isotope (solution) Element (solution)
from which he derived the following expression for calculating the
specific surface in terms of grams of lead on the surface of 1 g of
PbSO 4 , from data on the adsorption of thorium B :
... , ThB adsorbed in % g Pb in solution ,.
SpeClfic8 " rfaoe " 100 -ThB adsorbed in % , PbSO 4 (2)
'Physik. Z., 15, 924 (1914); Paneth and Vorwcrk: Z. physik. Chem., 101,
445 (1922).
EXTENT OF SURFACE FROM ADSORPTION MEASUREMENTS 51
In a given aged sample of PbSO4, Paneth and Vorwerk obtained the
value 6.4 X 10~ 4 g Pb for the specific surface, allowing 3 minutes'
shaking time for equilibrium to be set up. Equilibrium is not reached
entirely even after 30 minutes' shaking, probably because of the slow-
ness with which agglomerates are broken up by shaking. Kolthoff
and Rosenblum 4 found that an hour was required to set up kinetic
exchange equilibrium in aged samples and obtained a value for the
specific surface of 8.4 X 10~ 4 g Pb for a sample having the same
average size of crystals as Paneth and Vorwerk's sample. This moans
that there are (8.37 X 10~4)/207 X 6.06 X 10*3 = 24.5 X 10'? lead
ions in the surface of 1 g of PbSO 4 . Assuming that the crystals
are cubic and have a density of 6.23, there are 2.32 X 10 7 molecules
of PbSO 4 per cm, and the size of 1 molecule = 4.32 X 10 -** cm;
hence the surface per molecule is (4.32 X 10 -)2 = 18.6 X 10 10 -
The specific surface is therefore 18.6 X 10~i X 24.5 X 10" cms =
45 5 dm 2 . The area measured microscopically was found to be
33.4 dm 2 , confirming Paneth's observations that the two methods
show good agreement, considering the errors in the microscopic
measurements.
The adsorption of dyes such as Ponceau 2R, 5 methylene blue,
Ponceau 4R, 6 and wool violet 4BN 7 has been used as a measure of
the specific surface of lead sulfate. Of these, wool violet appears to
be especially suitable since the surface of lead sulfate ib satuiated at
relatively small concentrations of the dye in the supernatant liquid
and the lead salt of the dye is sufficiently soluble to permit a study
of the dye adsorption in the presence of a slight excess of lead ions
in solution.
The change in the specific surface with time can be followed by
means of adsorption measurements. Fresh precipitates of compounds
like lead sulfate consist of very imperfect crystals and possess an
internal surface of fine capillaries and canals which is much greater
than the external surface. On aging, the crystals rapidly become more
perfect which results in a rapid decrease in the internal surface,
whereas the external surface decreases but very slowly. Kolthoff and
Rosenblum 8 showed that the external surface may be estimated by
J. Am. Chcm. Soc., 55, 2656 (1933)
Paneth and Vorwerk: Z. physik. Chem , 101, 480 (1922); Paneth and
Thimann: Ber., 57B, 1215 (1924).
6 Kolthoff and Rosenblum: J Am. Chem. Soc, 55, 2664 (1933).
7 Kolthoff, von Fischer, and Rosenblum: J. Am Chem. Soc., 56, 832 (1934).
8 J. Am. Chem. Soc., 66, 1264, 1658 (1934); Phys. Rev., 45, 341 (1934).
52
COLLOIDAL SULFATES OF LEAD AND STRONTIUM
determining the adsorption of wool violet after short intervals of
shaking with the dye since the large size of the dye molecules does
not allow penetration into the fine capillaries of the internal surface.
The sum of the external and internal surfaces may be obtained by
measurements of the adsorption of thorium B from a radioactive lead
nitrate solution. The speed of distribution of the thorium B from
such a solution was found to be independent of the rate of shaking as
long as the precipitate was not allowed to settle. 9
20 40
Age, Minutes
FIG 7. Rate of decrease of specific surface of lead sulfate in the presence of a
slight excess of potassium sulfate and lead nitrate, respectively.
The specific surface of a precipitate freshly formed at room tem-
perature is very large, amounting to approximately 700 mg Pb per
g PbSO 4 . Since 1 g of PbSO 4 can contain, at most, 683 mg of lead
ions, it is obvious that the thorium B distributes itself uniformly
throughout the fresh precipitate in spite of the fact that the crystals
appear compact on microscopic observations.
The porous mass ages rapidly, the specific surface decreasing 20
to 40 times after one hour. Some observations on the rate of decrease
of specific surface are shown graphically in Fig. 7 for a sample pre-
cipitated with a slight excess of lead nitrate and one precipitated with
a slight excess of potassium sulfate. The final state of the aging
Kolthoff and Rosenblum: J. Am. Chem. Soc., 68, 121 (1936).
EXTENT OF SURFACE FROM ADSORPTION MEASUREMENTS S3
process is reached very slowly, requiring several days as evidenced by
microscopic observations of increasing particle size and by the de-
creasing adsorption of wool violet. The very rapid aging immediately
following the precipitation results in a perfection of the primary
crystals; on longer standing, a slow crystal growth occurs as a sec-
ondary process. The speed of aging is independent of the amount of
precipitate, the amount of lead in solution (the lead concentration
being constant), and of stirring. The rapid aging is not due to an
Ostwald ripening process in which larger particles grow at the ex-
pense of smaller ones since speed of agitation would influence such a
process. The recrystallization responsible for the aging and perfection
of the crystals apparently takes place at great speed in a liquid layer
around the particles. The fresh precipitate does not age in alcohol,
in which it is insoluble, and external crystallization does not take place
if the fresh precipitate is coated with an adsorbed film of wool violet. 10
Moreover, the adsorption of dyes such as Ponceau 4R and wool
violet inhibits greatly the speed of the kinetic interchange at the inter-
face lead sul fate-solution. As would be expected, the degree of per-
fection of crystals and the rate of aging are influenced by the con-
centration of the interacting solutions and by the temperature during
precipitation. Tn accord with the usual rule, a precipitate obtained
from 0025 Af solutions is much more perfect and coarser than one
obtained from 1 M solutions. 11
Bancroft and Harnctt 12 questioned J'aneth's method of determin-
ing the specific surface from measurements of the adsorption of dyes,
since this adsorption is a function of the apparent pll of the solution.
They showed, for example, that the adsorption of the basic dye
methylene blue by lead sul fate was greater the higher the pll value.
Similarly Kolthoff, von Fischer, and Kosenhlum found the adsorption
of the acid dye wool violet to be much greater from an acid medium
than from a neutral medium, and to increase with the acidity. These
examples are in accord with the rules proposed by Bancroft 11 and
demonstrated with several dyes by Briggs and Bull. 74 The variation
in the amount of dye taken up at varying pll may result in part from
the influence of the added acid or base on the dye itself, changing the
'<> Kolthoff and Roscnblum: J Am Chem. Soc, 57, 597, 607 (1935); 68,
116, 121 (1936); Thys Rev, 47, 631 (1935).
Kolthoff and Rosenblunv J. Am. Chem. Soc., 57, 2577 (1935).
12 Colloid Symposium Monograph, 6, 73 (1928).
i*J Phys Them., 18, 1, 118, 385 (1914); 19, 50, 145 (1915).
"J. Phys. Chem., 26, 845 (1922).
54 COLLOIDAL SULFATES OF LEAD AND STRONTIUM
soluble dye salt to insoluble dye acid as upon the addition of acid to
Congo red, or from the conversion of soluble salt to insoluble dye
base such as takes place on adding sodium hydroxide to methylene
blue. These complications merely emphasize the importance of taking
the pH value into account in adsorption studies with dyes.
In this connection, Kolthoff and Rosenblum 6 showed that, in
accord with the usual behavior (p. 106), the adsorption of the acid
dye Ponceau 4R by lead sulfate is increased by the addition of the
strongly adsorbed lead cations and is decreased in the presence of the
strongly adsorbed sulfate anions.
Mechanism of the Adsorption
Exchange Adsorption. As already pointed out, when an ionic
precipitate such as lead sulfate is brought in contact with a solution,
a kinetic equilibrium is set up between lattice ions in the surface and
foreign ions in the solution. For the adsorption of ThB on lead sul-
fate the distribution is given by Paneth's equation :
ThB (adsorbed) ^ Pb (surface)
ThB (solution) "" Pb (solution)
Let x = the amount of the element to be adsorbed, originally present
in the solution ; y = the amount adsorbed ; z = the amount remaining
in solution, then x = y + z J and = the amount of the element on
the surface of the powder. Substituting in the Paneth equation :
3r="; - * (4)
* y *
from which
r\ . ^
(5)
By means of this equation, Paneth and Vorwerk attempted to calculate
the amount of lead ions adsorbed on the surface of an aged lead
sulfate. For example, a preparation having a specific surface of 1.29
mg Pb per g PbSO 4 , adsorbed 0.5 mg Pb++ ions when suspended in
water (saturated PbSO4 solution) and 1.2 mg Pb++ ions when sus-
pended in a solution of Pb(NOs)2 0.005 N and above. These figures
would indicate that, in the suspension in water, about 40% of the
surface is covered by adsorbed lead ions, and that in 0.005 N
Pb(NO 3 ) 2 the surface is approximately covered with a monomolecular
layer of lead ions.
MECHANISM OF THE ADSORPTION 55
Kolthoff and Rosenblum 4 confirmed Paneth and Vorwerk's ob-
servations that the specific surface was the same in lead nitrate solu-
tions as in water and showed further that it was the same in potassium
sulfate solutions as in water. This could not be true if there were any
appreciable adsorption of lead and sulfate ions by the aged lead sul-
fate. It must therefore be concluded that the equation used by Paneth
and Vorwerk to calculate the adsorption of lead ions is not correct.
Kolthoff and Rosenblum point out that the distribution equation is :
ThB (surface) _ Total Pb (surface)
ThB (solution) "" Total Pb (solution) '
where Total Pb (solution) represents the concentration of the lead
analytically determined and Total Pb (surface) is the specific surface
determined as described above. An adsorption of lead ions from lead
nitrate solution would result in an increase in the concentration of
lead at the surface and hence an increase in the specific surface. Since
the specific surface undergoes little or no change in lead nitrate solu-
tions, it must follow that no measurable adsorption of lead from lead
nitrate takes place on the aged precipitate. The same applies to the
adsorption of sulfate from sodium sulfate solutions. There must be
some adsorption, although it is not measurable, since even aged lead
sulfate has a positive charge in lead nitrate solution and a negative
charge in sodium sulfate solution. This adsorption appears to be
limited to the edges and corners of the crystals and is therefore quite
small in aged precipitates. The adsorption would be expected to be
much larger in freshly formed precipitates with an irregular and
higher specific surface.
When an ionic precipitate is brought in contact with any solution
containing a strongly adsorbed ion, an exchange adsorption between the
lattice ion and the foreign ion takes place similar to that between lead
and ThB considered above. An exchange of this nature is not limited
by the similarity in size of the exchanging ion and the exchanged
lattice ion. Kolthoff and Rosenblum " showed that, in the adsorption
of the sodium salts of Ponceau 4R and wool violet, only the dye
anions are adsorbed, no sodium ions being removed from solution;
and for each adsorbed dye anion an equivalent amount of 'sulfate is
sent into solution. This recalls the adsorption of acid dyes by hy-
drous oxide mordants where the adsorbed dye anion displaces an
J. Am. Chem. Soc., 55, 851, 2664 (1933) ; 56, 834 (1934).
56 COLLOIDAL SULFATES OF LEAD AND STRONTIUM
equivalent amount of adsorbed chloride from the surface of the
hydrous oxide and no sodium ion is adsorbed. 10
Other Mechanisms. Although the adsorption of acid dyes by lead
sulfate is probably best interpreted as an exchange adsorption, other
mechanisms are possible. 17 For example, it may be argued that wool
violet is adsorbed as lead salt on the surface of lead sulfate, the lead
ions being furnished by the saturated solution of lead sulfate. The so-
lution is no longer saturated, and accordingly more lead sulfate will
go into solution until equilibrium is restored. The final result is that
equivalent amounts of lead and wool violet ions are removed from
the solution, leaving it with sulfate ions in excess. Another mechan-
ism which involves neither exchange nor adsorption of the lead salt of
wool violet assumes that a triple layer is formed, the dye ions being
truly adsorbed on the surface. 18 Since the thermodynamic potential
of the precipitate is scarcely changed by the adsorption, some lead
ions are adsorbed on the surface but most will be present as counter
ions.
In this connection, Kolthoff points out that, if an ion foreign to
the lattice ion gives a marked exchange with the lattice ion of the same
electrical sign in the surface, it cannot be concluded from the experi-
mental data that an equivalent adsorption of lattice ion and foreign
ion has taken place although equivalent amounts of lattice ion and
foreign ion are removed from solution. The foreign ion may dis-
appear by exchange and the lattice ion by precipitation in the form of
the adsorbent itself. For example, in the adsorption of the lead salt of
wool violet, the primary reaction is: PbSO 4 -f wool violet -> Pb-
wood violet + SO 4 . The exchanged sulfate ions are precipitated
as lead sulfate by the excess lead ions in solution. Taking the
analytical results, one might conclude that the lead salt of wool violet
is adsorbed when as a matter of fact the wool violet ions are removed
by exchange and the lead ions by precipitation. A similar behavior
is probably encountered in the adsorption of calcium iodate by calcium
oxalate. 19
COLLOIDAL STRONTIUM SULFATE
The solubility of strontium sulfate in water (0.00062 mol/1) is
more than twice as great as that of lead sulfate ; hence it is obtained
"Weiser and Porter: J. Phys. Chem., 31, 1824 (1927).
"Kolthoff: J. Phys. Chem., 40, 1027 (1936).
"Verwey: Kolloid-Z. f 72, 187 (1935).
"Kolthoff and Sandell: J. Am. Chem. Soc., 56, 2170 (1933).
MECHANISM OF THE PRECIPITATION PROCESS 57
in the colloidal state only by precipitation in alcohol-water solution. 20
Sols in 75% or more of alcohol, and containing 0.6 g SrSO 4 /l, possess
but a slight opalescence when first formed and are stable for months.
By working with concentrated solutions, gelatinous precipitates or
jellies are obtained consisting of interlocking bundles of needle-shaped
crystals.
Mechanism of the Precipitation Process
Oden 21 studied the mechanism of the precipitation of strontium
sulfate by measuring the conductivity of solutions which result from
mixing measured amounts of strontium hydroxide and sulfuric acid.
It was found that the first stage in the process, the formation of
nuclei, can be represented by the equation
n = const. C\ C^F,
in which n is the number of nuclei, C t and C 2 are the concentrations
of the reacting ions, t is the time, and a is a number greater than 3
and dependent on the number of ions in the unit cell. This confirms
von Weimarn's view concerning the relationship between particle size
and the concentration of reagents. In further accord with von Wei-
marn's theory, the second stage in the process consists in the growth
of nuclei which is followed by an aggregation of the primary particles
into secondary aggregates analogous to the coagulation of sols, and
finally by the growth of larger crystals at the expense of smaller ones.
The conductivity of the solution supersaturated with respect to the
precipitate was found to be constant at first for a considerable period
of time, the length of which is determined by the concentration of
the base and acid. After this interval, the conductivity decreases
gradually during the process of actual precipitation until it reaches a
minimum value corresponding to complete precipitation. This process
requires several hours in concentrations of 0.02 to 0.06 N but only a
few minutes in more concentrated solutions. The rate of growth of
single crystals (primary particles) in a supersaturated solution and in
a non-aggregated state (cf. p. 12) may be expressed by the equation
V k = K- O t ' (C t - C w ), where V* is the velocity of crystallization;
2 <> Von Weimarn: Repts. Imp. Research Inst, Osaka, Japan, 9, 13 (1928);
12, 117 (1931); cf., also, "Die Allgemeinheit des Kolloidzustandes," Dresden
(1925).
aiArkiv Kemi, Mineral. Geol., 9, No. 23 (1925); Ode*n and Werner: 9,
No. 32 (1926).
58 COLLOIDAL SULFATES OF LEAD AND STRONTIUM
O t> the total surface of the crystals at time t; C% and C m , the concen-
tration of the solution at time t and at saturation, respectively; and
K, a constant.
The observations and conclusions of von Weimarn and of Oden
concerning the mechanism of the precipitation process were confirmed
and extended by Lambert and Hume-Rothery, 22 who worked with
solutions varying in concentration between 0.01 and 3.47 N. Under
certain conditions, especially high concentrations and low temperatures,
the initial precipitate consists of needle-shaped crystals believed to be
a hydrate of strontium sulfate. On standing in the molten liquor the
needle crystals go over spontaneously into stable rhombic crystals of
anhydrous strontium sulfate which are also formed directly from
dilute solutions at low temperatures and from solutions not too con-
centrated at higher temperatures. The conclusion that the needles are
a hydrate and the rhombic crystals are the anhydrous salt is question-
able. This could be decided definitely by x-ray examination of the
two types of crystals.
Observations of the time which elapsed before a precipitate ap-
pears in the presence of different salts led to the conclusion that a
positively charged sol is formed as an intermediate stage in the precipi-
tation and that the sol precipitates more or less rapidly depending on
the precipitating or stabilizing effect of the other ions. For example,
other conditions being the same, a definitely increasing time is re-
quired for the appearance of the precipitate, in the presence of cations
in the order: Na < Mg < Al < H, which is the order of increasing
protecting power of cations for positive sols. Similarly, the order of
increasing velocity of precipitation of strontium sulfate in the pres-
ence of anions is : CH 3 COO > Cl > NOs, which is the order of the
coagulating power of these anions for positive sols. This point of
view accounts for the relative velocities of precipitation of the more
concentrated solutions; but, in dilute solutions, other factors such as
viscosity play an important part in determining the relative times for
the appearance of a precipitate. The size of the particles was found
to depend on two main factors: (1) the velocity of precipitation, in-
crease of which leads to smaller primary particles in accord with von
Weimarn's theory; and (2) the facility for rapid diffusion which
favors larger particles. Similar observations and conclusions were
made from a study of calcium sulfate. 28
J. Chem. Soc., 2637 (1926) ; cf. von Weimarn:929 (1927).
"Lambert and Schaffer: J. Chem. Soc., 2648 (1926).
MECHANISM OF THE PRECIPITATION PROCESS 59
Campbell and Cook 24 observed the fall in conductivity of super-
saturated solutions of strontium sulfate formed by mixing potassium
sulfate and strontium chloride solutions. Solutions down to and in-
cluding 50$> supersaturation crystallize spontaneously, but the con-
ductivity never falls to normal. Prior to the formation of microscopic
particles, von Weimarn's colloidal zone was observed, but not all the
colloidal particles appear to function as nuclei. Apparently a nucleus
once formed grows rapidly to considerable magnitude. The interest-
ing observation was made that after the degree of supersaturation has
fallen to 40-50%, further precipitation appears to stop or to become
very slow. 25 A solution appears to remain permanently supersaturated
at 30-40% supersaturation despite the presence of large particles. No
experiments were carried out with interacting solutions of sufficiently
high concentrations to give the needle-shaped crystals observed by von
Weimarn and by Lambert and Hume-Rothery.
ADSORPTION ON THE STRONTIUM SULFATE LATTICE
Since radioactive isotopes of strontium, calcium, and barium are
not known, Paneth's equation in its original form is inapplicable for
the estimation of the extent of surface of the alkaline-earth sulfates.
In respect to isomorphous substances such as lead sulfate and stron-
tium sulfate, it is assumed that the lead ions in a solution in which
strontium sulfate is suspended will enter into kinetic exchange with
the outer layer of strontium sulfate. At equilibrium the distribution
will be represented by the expression :
Pb (surface) __ Sr (surface) . .
Pb (solution) ~~ Sr (solution)
in which K represents the distribution coefficient of lead between the
solution and the surface of strontium sulfate. As already pointed out,
Kolthoff showed the applicability of an equation of this type to the
distribution of lead or chromate ions between the surface of barium
sulfate and solution (p. 32), and to the distribution of bromide ions
between silver chloride and solution. In such cases the foreign ion
fits the lattice of the adsorbent and its size is similar to that of the
ion replaced. An exchange of this nature is not limited to iso-
morphous ions or to ions of similar size as evidenced by the results
on the distribution of the dye ions between lead sulfate and solution
*J. Am. Chem. Soc. f 57, 387 (1935).
Cf. Dundon: J. Am. Chem. Soc., 48, 2658 (1923).
60 COLLOIDAL SULFATES OF LEAD AND STRONTIUM
(p. 55). An equation similar to (6) represents the distribution of
sulfate or barium ions between calcium oxalate and solution 19 and of
eosin 2e and wool violet 17 anions between silver chloride and solution.
For determining the extent of surface of strontium sulfate or
barium sulfate which are isomorphous with lead sulfate, the equation
of Paneth 27 takes the form :
, ThB (surface) Sr (surface) ,_
1^ i ^s - (7 1
ThB (solution) Sr (solution)
where k is the ratio of solubilities of the sulfates of thorium B and
strontium in salt solutions. The value of k was found to be 1/15
when the surface of strontium sulfate is determined from adsorption
of thorium B ; and 14 when the surface of barium sulfate is estimated
by the same method. The applicability of the procedure is evidenced
by the similarity between the values estimated microscopically and by
the radioactive method. The latter method is not applicable for the
determination of the extent of the surface of calcium sulfate since
the calcium salt is not isomorphous with the sulfates of thorium B,
lead, strontium, and barium.
Kolthoff and MacNevin 28 found that the simplest method of de-
termining the specific surface of barium sulfate is based on the de-
termination of the amount of wool violet adsorbed on the saturated
surface of the precipitates; a method based on the exchange between
chromate and sulfate 20 ions likewise gives good results; the thorium
B method is unsatisfactory except with well-aged precipitates.
a Kolthoff: Kolloid-Z. f 68, 190 (1934).
""Radio Elements as Indicators," McGraw-Hill Book Co, 65 (1928).
" J. Am. Chem Soc., 58, 725 (1936).
* Kolthoff and Noponen: J. Am. Chem. Soc., 59, 1237 (1937).
CHAPTER IV
COLLOIDAL CALCIUM SULFATE; PLASTER OF PARIS
The relatively high solubility of calcium sulfate in water precludes
the possibility of forming a stable hydrosol ; but a dilute fairly stable
sol may be prepared in an 86$? ethyl alcohol-water mixture. 1 The
gelatinous form of calcium sulfate is precipitated by the action of
sulfuric acid on a methyl alcohol solution of calcium oxide 2 or by
the addition of ethyl alcohol to an equal volume of a saturated solu-
tion of calcium sulfate. 8
Like strontium sulfate (p. 58), calcium sulfate appears to form a
positive colloid during the process of precipitation. From a given pair
of precipitants, calcium sulfate is thrown down directly either as the
dihydrate or as the hemihydrate, the determining condition being the
temperature of precipitation.*
Colloidal behavior of calcium sulfate is encountered in the highly
dispersed products plaster of Paris and the so-called "soluble an-
hydrite" formed by the dehydration of the dihydrate under suitable
conditions. These substances possess hydraulic properties, reacting
with water and setting to form a coherent mass.
THE SYSTEM CALCIUM SULFATE-WATER
The Existence of Calcium Sulfate Hemihydrate
Calcium sulfate is ordinarily believed to exist in three chemical
forms, the dihydrate CaSO 4 -2H 2 O; the hemihydrate CaSO 4 --
0.5H 2 O, commonly called plaster of Paris; and the anhydrous salt.
Two crystalline modifications of the anhydrous salt have been identi-
fied by x-ray diffraction analysis. One form obtained by igniting
gypsum fairly strongly gives the same orthorhombic pattern as the
mineral anhydrite. Since neither the synthetic product nor the min-
eral possesses hydraulic properties, this form is commonly called
insoluble anhydrite. A second crystalline form, hexagonal, obtained
by igniting gypsum or hemihydrate between 100 and 200, possesses
1 Von Weimarn: Repts. Ind. Research Inst, Osaka, Japan, 12, 117 (1931).
'Neuberg: Sitzber. preuss. Akad. Wiss., 820 (1907).
'Cavazzi: Kolloid-Z., 12, 196 (1913).
* Lambert and Schaffer: J. Chem. Soc., 2648 (1926).
61
62 COLLOIDAL CALCIUM SULFATE; PLASTER OF PARIS
hydraulic properties and was called by van't Hoff 5 soluble anhydrite.
Since both the hexagonal and the orthorhombic forms can apparently
be obtained in both the hydraulic and non-hydraulic state, the terms
soluble anhydrite and insoluble anhydrite are misnomers. Balarew 6
and Ramsdell and Partridge 7 have suggested that the product formed
by dehydrating hemihydrate be called "dehydrated hemihydrate" to
distinguish it from the orthorhombic anhydrite. This terminology will
be used in this chapter.
Although calcium sulfate hemihydrate or plaster of Paris is gen-
erally included among the time-honored compounds, several papers
have appeared in recent years which question the individuality of
hemihydrate and the mechanism of its dehydration. Linck and Jung 8
deduced from dehydration data that plaster of Paris loses its water
after the manner of zeolites. Balarew, 9 at different times, has been
on all sides of the question. At first he claimed that the dehydration
isobar would indicate the water to be present in the form of a true
hydrate ; later he stated that the dehydration curves show the water
to be held in a new way "half-hydratic and half-zeolitic" ; finally he
concluded that the water is lost in the manner of zeolites. Gibson and
Holt 10 concluded from temperature-pressure curves that the water is
lost continuously as in a zeolite. Parsons n claimed that all the water
in gypsum may be lost by heating without the intermediate formation
of a hemihydrate.
Jung, 12 Ramsdell and Partridge, 18 and Caspari 14 ' 15 reported that
the x-ray diffraction patterns of calcium sulfate hemihydrate and its
dehydration product are identical, in agreement with the view that
the hemihydrate is a zeolite. Onorato 1G and Gallitelli 17 found the
B Van't Hoff, Armstrong, Hinrichsen, Weigert, and Just: Z. physik. Chem.,
46, 257 (1903).
Kolloid-Z., 48,63 (1929).
T Am. Mineral., 14, 59 (1929).
Z anorg Chem., 137, 407 (1924); cf. Krauss and Jorns: Tonind-Ztg., 64,
1467, 1483 (1930).
Z. anorg. Chem, 166, 258 (1926) ; Kolloid-Z., 48, 63 (1929) ; Balarew and
Koluschewa: 70, 288 (1935).
10 J. Chem. Soc., 638 (1933).
"Univ. Toronto Studies, Geol. Sen, No. 24, p 24 (1927).
Z. anorg. Chem., 142, 73 (1925).
"Am. Mineral., 14, 59 (1929) ; cf., also, Partridge and White: J. Am Chem.
Soc, 61, 360 (1929).
"Nature, 138,648 (1934).
"Proc. Roy. Soc. (London), 166A, 41 (1936).
"Periodico mineral. (Rome), 3, 138 (1932).
"Periodico mineral. (Rome), 4, 1, 132 (1933).
THE EXISTENCE OF CALCIUM SULFATE HEMIHYDRATE 63
hemihydrate and dehydrated hemihydrate to have the same type of
structure but recognized minor differences in the powder diagrams.
Feitknecht l8 likewise observed small differences in the two patterns
provided care was taken to prevent rehydration of the dehydrated
product, a precaution that Jung apparently overlooked.
Although the evidence cited above indicates that calcium sulfate
hemihydrate is not a true hydrate, dehydration and x-ray studies car-
ried out in the author's laboratory 19 lend strong support to the view
i
tf.U
1.5
i
OS
c
~ -^
V
OD
R
shydration
ihydretion
\lftrt rt
*>-o o
I 50 100 151
Temperature, Degrees C.
FIG. 8. Isobaric dehydration curve for CaS(V2H>O.
that the hemihydrate is a definite chemical individual and that the
process of dehydration is not zeolitic in character.
Isobaric Dehydration. Pure transparent selenite CaSO 4 -2H 2 O
was subjected to isobaric dehydration (vapor pressure 23.6 mm) tak-
ing special care to secure equilibrium at each temperature by holding
the sample for days or weeks, weighing at intervals until a constant
state was attained. The results for both dehydration and rehydration
are shown graphically in Fig. 8. The hemihydrate portion of the
"Helv. Chim. Acta, 14, 85 (1931).
"Weiser, Milligan, and Ekholm: J. Am. Chem. Soc., 68, 1261 (1936).
64 COLLOIDAL CALCIUM SULFATE; PLASTER OF PARIS
curve is enlarged in Fig. 9. The points on the curves represent the
average of four values which differ by not more than 0.006 mol of
water per mol of calcium sulfate. The dehydration isobars for the
hemihydrate are typical for a definite hydrate which adsorbs a small
amount of water and gives a dehydration product which likewise ad-
sorbs water. Since the hemihydrate and especially the dehydrated
hemihydrate are known to adsorb water, the evidence seems con-
clusive that when CaSO 4 2H 2 O is dehydrated isobarically a definite
hydrate CaSO4'0.5H 2 O which adsorbs a small amount of water is
UD
ffl 1
"H
U4
i
i
3
|02
k
1
70 105 140 175 21
Temperature, Degrees C.
FIG. 9. Isobaric dehydration curve for CaSO* 0.5H a O.
first formed and this in turn dehydrates in stepwise fashion to de-
hydrated hemihydrate, CaSO 4 , which adsorbs water that is not re-
moved completely at a vapor pressure of 23.6 mm until well above
the decomposition temperature. The dehydration is reversible, but
there is no indication that the water is present in the zeolitic form
unless one wishes to apply the term "zeolitic" to the very small amount
of water adsorbed first by the very finely divided hemihydrate and
later by the resulting dehydrated hemihydrate. The transition tem-
perature of minute crystals of hemihydrate * dehydrated hemihydrate
is approximately 97 20 at 23 mm ; and 107 at 971 mm. Nacken and
Fill 21 claim that "soluble anhydrite" is formed at 20 in dry air or
20 Cf. Balarew: Z. anorg. Chem., 166, 2S8 (1926) ; 168, 137 (1927).
Tonind.-Ztg. f 55, 1194 (1931); cf. Krauss: 65, 1222 (1931).
THE EXISTENCE OF CALCIUM SULFATE HEMIHYDRATE 65
in a vacuum. From solubility studies Partridge and White 22 found
the transition temperature of gypsum to hemihydrate to be 98 and
from gypsum to anhydrite to be slightly below 40, both in contact
with water. In accord with van't Hoff, plaster of Paris goes over
slowly to dehydrated hemihydrate in contact with water at 100.
X-ray Examination. In the light of the dehydration experiments,
one would expect the x-ray diffraction pattern for the hemihydrate to
differ somewhat from that of its dehydration product. This was
found to be true by comparing the patterns obtained with a Seemann
crystal analysis apparatus using K a Cu radiation and a precision Debye-
Scheerer camera. The films were calibrated by mixing a small amount
of pure nickelous oxide with the samples examined. The x-ray dif-
fraction data are represented diagramniatically in Fig. 10. It is ap-
1 l,ll
|GaSO;iH 2
1 1 I 1 1 1 1 ll 1 ill II 1
1 1 1 1
CaS0 4
Dehydrated
1 Hemihydrate
II i i i .In III .
FIG. 10. Diagrams of the x-ray diffraction patterns of CaSOi OSHaO and
CaSO* (dehydrated hemihydrate).
parent that the patterns of the two substances are similar, showing
that the structures arc similar; but the existence of definite differences
between the two indicates that the water molecules in the hemihydrate
occupy fixed positions in the lattice.
Caspari 15 failed to observe the small but significant differences on
comparing x-ray rotation diagrams of large crystals of the two sub-
stances; but he demonstrated conclusively that the crystals possess
a hexagonal structure. Weiser and Milligan 23 found the same charac-
teristic differences in the powder diagrams of macrocrystals that they
observed with microcrystals. Moreover, the dehydration isobar for
the macrocrystals was a step-curve like that in Fig. 9.
The crystal structure of gypsum is monoclinic. Davis 24 postu-
22 J. Am. Chem Soc., 51, 360 (1929); Toriumi and Kara: J. Chem. Soc.,
Japan, 66, 1051 (1934).
28 J. Am. Chem. Soc., 69, 1456 (1937).
2 J. Soc. Chem. Ind., 26, 727 (1907).
66 COLLOIDAL CALCIUM SULFATE; PLASTER OF PARIS
lated an unstable rhombic modification, but this has not been disclosed
by x-ray diffraction studies. Rohland 25 assumed the existence of
three anhydrous forms of calcium sulfate in addition to dehydrated
hemihydrate and anhydrite; but here, also, x-ray analysis has failed
to establish the identity of the alleged modifications. It appears there-
fore, that four and only four crystalline modifications of calcium sul-
fate have been identified : the dihydrate, monoclinic ; the hemihydrate,
hexagonal; dehydrated hemihydrate, hexagonal; and anhydrite, or-
thorhombic.
The Formation of Hydraulic Plasters. The most common gypsum
plasters are plaster of Paris and dehydrated hemihydrate. By refer-
ring to Fig. 8 it will be seen that gypsum may be burned to plaster
of Paris at as low a temperature as 65 at 23 mm vapor pressure;
but the time required under these conditions is excessive. From a
survey of the literature 28 it appears that, in technical practice, the
gypsum is burned at temperatures varying from 120 to 200 in either
a kettle or rotating kiln using coal, gas, or electric heat. 37 Since gyp-
sum loses water more readily at 200 than at 120, the heating must
be shorter at the former temperature to burn to the same end point,
provided the kilns are otherwise the same. This was demonstrated
experimentally by Keane 28 and is substantiated by Hursch's 2e state-
ment : "Commercially temperatures up to 200 or above are used but
for short periods of time." The best temperature for commercial
burning varies with the cost of fuel, the size and nature of the
plant, etc.
Since the inversion temperature for plaster of Paris-dehydrated
hemihydrate is 97 at 23 mm, it is probable that the original plaster
contains more or less of the completely dehydrated product, especially
when the burning temperature is high. But if the plaster is allowed
to stand in moist air for a time, the dehydrated hemihydrate hydrates
completely to plaster of Paris,
If the ignition is carried out for too long a time above 200, anhy-
drite is formed which hydrates very slowly 28 and is known as dead-
Z. anorg. Chem, 31, 437 (1902); 85, 194; 36, 332 (1903); cf. Lacroix:
Compt. rend, 126, 360, 553 (1898); Gaubert: 197, 72 (1933); Bull. soc. franc,
mineral., 67, 252 (1934).
28 C/. Keane: J. Phys. Chem., 20, 701 (1916); Rogers: "Manual of Indus-
trial Chemistry," 5th ed., 1, 374 (1931); Read: "Industrial Chemistry," 232
(1933) ; Hursch: Trans. Am. Ceram. Soc., 17, 549 (1915).
"Tupholme: Ind. Eng. Chem., News Ed., 18, 441 (1935).
28 Cf. Chassevent: Compt. rend., 194, 786 (1932); cf. Stratta: Industria
chimica, 9, 28 (1934).
THE EXISTENCE OF CALCIUM SULFATE HEMIHYDRATE 67
burned plaster. Jolibois 29 gives about 365 as the point above which
dead-burned plaster results. But at higher temperatures between 600
and 1000, a plaster known as flooring plaster or Estrich gypsum is
formed. This sets more slowly than plaster of Paris but gives a very
hard and highly resistant plaster. A product known as Keene's cement
is made by burning the gypsum at red heat, cooling it, impregnating
it with an alum solution, and igniting once more. The trace of alu-
mina is said to catalyze the setting process, but since plaster formed
at red heat will set without its presence, the action of alumina is not
known with certainty.
It is apparently necessary to use temperatures above 500 in pre-
paring flooring plasters, but this is probably determined in part by
the time of ignition. Muller says that a temperature of 800-900 is
best, and Glasenapp 30 recommends 900 although temperatures from
800-1300 give a satisfactory product.
Since gypsum ignited below 200 sets readily and that ignited
around 300-400 sets very slowly if at all, one might attribute the
difference primarily to crystal structure, the setting compound being
hexagonal dehydrated hemihydrate and the non-setting compound
being orthorhombic anhydrite. Although this is a factor, it is prob-
ably not the most important one. Thus, one can prepare dehydrated
hemihydrate which will not set at a little above 30 in contact with a
solution of sodium chloride, 5 and anhydrite which will set is pre-
pared at 800 to 1000. The difference in the setting power of dif-
ferent samples of dehydrated hemihydrate on the one hand, and anhy-
drite on the other, is probably due in large measure to the size and
physical character of the particles of the respective preparations. 31 In
support of this view, Keane 32 showed that gypsum calcined at 600
behaved like dead-burned plaster when the particles were O.OS mm in
diameter but set quite rapidly to a hard resistant mass when the
particle size was reduced by grinding to 0.005 mm in diameter. Gill 8S
found that both finely ground anhydrite and rough lumps of dead-
burned plaster were bound firmly together by gypsum crystals after
2 Bull. soc. chim., (4) 41, 117 (1927).
ao Z. angew. Chem, 07 III, 306 (1914); Tonind.-Ztg., 82, 1148, 1197, 1230
(1908).
81 Cf. Desch: Trans. Am. Ceram. Soc., 18, IS (1918-19).
38 J. Phys. Chem., 20, 701 (1916); Winterbottom : Bull. S. Australian Dept.
Chem., 7 (1917).
J. Am. Ceram. Soc., 1, 65 (1918).
68 COLLOIDAL CALCIUM SULFATE; PLASTER OF PARIS
standing covered with water for 6 years; and Hartner 34 confirmed
Keane's observation that natural anhydrite sets if ground to an ex-
tremely fine powder. Since anhydrite is 1.5 times as soluble as gyp-
sum, setting of anhydrite will take place if the particle size is reduced
to the point where the rate of solution and hydration is high enough
to give a supersaturated solution of gypsum. Electrolytes which in-
crease the solubility and rate of solution may be added to make the
anhydrite set. 85
The effect of particle size on the rate of solution of anhydrite has
been studied in detail by Roller. 86 Some data are shown graphically
A ihydnte
Gypsum
ot
r\
2 5 10 20
Diameter of Particles, Microns
50 100
FIG. 11. Effect of particle size on the rate of solution of anhydrite and gypsum.
in Fig. 11, in which the particle size is plotted against the dissolution
factor, which is the relative specific rate of solution of each fraction
referred to some fraction, usually the coarsest, taken as unity. Two
procedures were followed, one in which the anhydrite powder was
added dry to the water, and the second in which the powder was first
dispersed with a small amount of water. All conditions of a given
series of experiments were maintained constant except the particle
size. It will be seen that for the powder added dry the maximum
dissolution factor is at 8/*. Moreover, the dissolution factor is ap-
s* Z. angew. Chem., 38 I, 175 (1920); Weissenberger : Kolloid-Z, 82, 181
(1923).
"Budnikov: Compt. rend., 188, 387 (1926); Z. anorg. Chem., 166, 141
(1926); Sexton: Can. Mining J., 61, 247 (1930).
3 J. Phys. Chem., 86, 1133 (1931) ; 36, 1202 (1932).
THE EXISTENCE OF CALCIUM SULFATE HEMIHYDRATE 69
proximately the same with the dispersed and dry anhydrite for a
fraction in which the mean diameter of the particles is 26. S^. For
smaller particles it is less for the dispersed powder, until a flat inflec-
tion point is reached that lies near the particle size corresponding to
the maximum for the powder added dry, that is at 8/*. At the inflec-
tion point the rate of solution is, by definition of the solution factor,
strictly proportional to the surface exposed. With further decrease
in particle size beyond the inflection point, the dissolution factor of
the dispersed powder increases continuously.
The observed behavior is interpreted in the light of the theory
that solution takes place from active centers which involve the edges
and corners of a crystal. For the dry powder, the rate of solution is
determined by the structure of the sediment. For the initially dis-
persed anhydrite, the same situation holds for particles larger than 8/*.
Between 8/* and SO/* the enhanced rate of solution results from increas-
ing dispersity of the anhydrite ; with decrease in particle size to 8ft, the
physical interaction of the grains vanishes and the rate of solution is
proportional to the exposed surface. Below S/A, the increasing rate
with diminishing particle size is attributed solely to the effect of edges
and corners.
Because of the observed effects of size and physical nature of par-
ticles on the rate of solution of anhydrite, it is not immediately obvious
why one gets a non-setting plaster by burning at 30CM00 and a set-
ting plaster by burning at 800-900. Other things being equal one
would expect to get larger, more perfect, and more slowly soluble
crystals at the higher temperature, in opposition to the facts. A
logical interpretation of this behavior would be that an allotropic
modification of anhydrite is formed at the higher temperature; but
this is not so. It has been suggested that Estrich gypsum or flooring
plaster is basic sulfate resulting from the decomposition of the anhy-
drite ; 8T but this seems to be open to question first because the ex-
istence of a basic sulfate has not been established and second because
anhydrite decomposes only slightly below 1000 . 88 It is possible, how-
ever, that the decomposition is sufficient to cause some cracking or
shattering of the crystals which would increase the number of corners
and edges and thus hasten the rate of solution.
"Glasenapp: Tonind.-Ztg. f 82, 1148, 1197, 1230 (1908); Grengg: Z. anorg.
Chem., 90, 327 (1914); Gallo: Gazz. chim. ital., 44 I, 497 (1914).
38 Budnikov and Syrkin: Chem.-Ztg., 47, 22 (1923) ; cf. Cobb: J. Soc. Chem.
Ind, 29, 69 (1910).
70 COLLOIDAL CALCIUM SULFATE; PLASTER OF PARIS
THE MECHANISM OF THE SETTING OF PLASTER OF PARIS
The setting of plaster of Paris involves a definite chemical trans-
formation from the hemihydrate to the dihydrate of calcium sulfate.
Since the hemihydrate is 4.5 times as soluble as the dihydrate 80 at
room temperature, Le Chatelier 40 assumed the following mechanism
for the setting process : The hemihydrate first forms a saturated solu-
tion in water, then reacts to form the dihydrate giving a supersaturated
solution from which is deposited a compact mass of interlacing
needle-shaped crystals the set plaster. 41 The amount of water neces-
sary to bring the hemihydrate back to the fully hydrated condition is
much less than is necessary to dissolve it since a given amount of
solution supersaturated with respect to gypsum deposits crystals,
thereby releasing the water to dissolve another portion of hemihy-
drate the process continuing until the transformation to gypsum is
complete.
For more than a quarter of a century Le Chatelier's theory of the
mechanism of the setting of plaster of Paris was regarded as com-
pletely satisfactory. But in more recent years, it has been considered
by a number of investigators as inadequate to account for all the facts.
Following the lead of W. Michaelis 42 and Keisermann, 48 who ob-
served the formation of a jelly as well as of crystals in the setting of
Portland cement (p. 435), they have visualized the formation of some
kind of a jelly as an intermediate stage in the setting of plaster of
Paris. Thus, Cavazzi 44 observed that gypsum precipitated rapidly
from aqueous solution with alcohol gave a gelatinous mass from which
distinct crystals separated on standing. Without further evidence, he
concluded that there was probably an intermediate gel stage in the
setting of plaster.
Traube 48 was the next to suggest that colloidal behavior plays a
role in the setting process. It is a well-known fact that soluble salts
may have a marked effect on the setting rate, 48 some salts accelerating
30 Marignac: Ann. chim. phys., (5) 1, 274 (1874).
40 "Recherches experimentales sur la constitution des mortieres hydraliques,"
Paris (1887) ; cf., also, van't Hoff et al: Z. physik. Chem, 45, 257 (1903) ;
Rohland: Z. anorg. Chem., 31, 437 (1902); 36, 194; 86, 332 (1903); Jolibois
and Chassevent: Compt. rend, 177, 113 (1923).
" Cf., also, Chassevent: Ann. chim., 6, 244, 313 (1926) ; 7, 43 (1927).
"Chem-Ztg., 17, 982 (1893) ; Kolloid-Z., 5, 9 (1909) ; 7, 320 (1910).
"Kolloid-Beihefte, 1, 423 (1910).
"Kolloid-Z., 12, 196 (1913); cf. Neuberg and Rewald: 2, 354 (1908).
4 Kolloid-Z., 26, 62 (1919).
4 *Ditte: Compt. rend., 126, 694 (1898).
MECHANISM OF THE SETTING 71
it and others retarding it. To account for this behavior Rohland 4T as-
sumed, in accord with Le Chatelier's theory, that any salt which in-
creases the solubility of calcium sulfate will accelerate the setting
whereas any salt which decreases the solubility will retard the setting.
This explanation is inadequate since small amounts of soluble sul fates
decrease the solubility of gypsum and yet increase the rate of set.
And there are other exceptions. 48 Traube observed the effect of salts
on the time required for the plaster to attain a definite state of hard-
ness. He found cations to be especially important, the order of in-
fluence being the reverse of that in which they precipitate sols. This
led him to the conclusion, which is not obvious, that some kind of
colloidal behavior must be involved in the setting process.
Ostwald and Wolski* 9 likewise concluded from indirect evidence
that colloidal processes are probably involved in the setting of plaster
of Paris. Experimentally, they followed the rate of change in vis-
cosity of suspensions of plaster, varying the concentration of the sus-
pensions, the degree of dispersion, the temperature and the nature of
the medium, i.e., using solutions of various salts as well as pure water.
The theoretical deductions from the experimental data were promised
in a second paper which was never published. They merely state that
the viscosity data indicate a colloid process, "as for example, the
separation of perhaps only a thin gel layer as an integral part of the
setting process."
The guarded statement of Ostwald and Wolski that "perhaps only
a thin gel layer" was formed at some stage of the setting process,
together with their failure to discuss theoretically their viscosity data,
suggests that they were probably in doubt as to whether there was any
gel formation at all. In marked contrast, Baykov, 80 Neville, 51 and
Budnikov 52 came out definitely in support of the formation of a gel
as an intermediate stage in the setting process. Baykov reached this
conclusion as a result of a procedure which he claimed would give a
gypsum jelly. Five to ten grams of hemihydrate were mixed with
100 cc of water or 10% (NH 4 ) 2 SO 4 solution and shaken vigorously.
On stopping the shaking after a suitable time, the entire mass was
"Z. Elektrochem, 14, 421 (1906).
Haddon: J. Soc. Chem. Ind., 89, 165T (1520) ; 40, 1227 (1921) ; cf. Welch:
J. Am. Ceram. Soc., 6, 1197 (1923).
"Kolloid-Z., 27, 79 (1920); Neugebauer: 31, 40 (1922).
"Kolloid-Z., 42, 151 (1927); 44, 242 (1928).
"Compt. rend., 182, 129 (1926); cf. Budnikov: Kolloid-Z., 42, 151 (1927).
J. Phys. Chem., 80, 1037 (1926).
72 COLLOIDAL CALCIUM SULFATE; PLASTER OF PARIS
said to set to a "gelatinous mass presenting the appearance of a silica
gel." Weiser and Moreland 53 showed that this precedure did not give
a true gel but a network of relatively long crystal needles of gypsum.
The only resemblance between this cloudy, non-uniform entangling
mass of crystal needles and silica gel was that both stayed in an in-
verted beaker.
Since the hydration of plaster of Paris is an exothermic reaction,
the rate of the reaction may be followed by measuring the rate of
evolution of heat. This has been done by Cloez, 54 Emley, 55 Chasse-
vent, 66 Neville, 67 Budnikov, 68 Hansen, 59 and others. Starting with a
high grade of hemihydrate mixed with pure water and determining
the rise in temperature with time, an S-shaped curve is obtained. For
an interval of several minutes the temperature rises but slightly, after
which it goes up relatively rapidly to a maximum, and then falls off.
Neville observed that a so-called "initial set," 60 results before there is
any marked heat evolution. He concluded from this that the setting
takes place in two stages: (1) the formation of a gel or adsorption
complex between the plaster and water, a process accompanied by but
little heat effect ; and (2) the exothermic reaction between the plaster
and the adsorbed water, forming gypsum. At first he attributed the
observed contraction in volume to the initial step and the subsequent
expansion to the second step, but later 61 he concluded that the hydra-
tion which causes the initial contraction takes place throughout the
whole period but is masked for a time by the thermal expansion. 62
The action of salts on the rate of setting was attributed to their effect
on the adsorption of water to form a gel and subsequently to their
catalytic action on the reaction between hemihydrate and water.
Budnikov carried out thermometric observations on the rate of
setting of plaster under varying conditions, apparently quite inde-
pendently of the work of Neville, and reached similar conclusions as
to the mechanism of the process. There is a distinct difference in
63 J. Phys. Chem, 36, 1 (1932) ; Colloid Symposium Monograph, 9, 1 (1931).
"Bull. soc. chim., (3) 29, 171 (1903).
"Trans. Am. Ceram. Soc., 19, 573 (1917).
ee Ann. chim., 6,264 (1926).
"J. Phys. Chem., 30, 1037 (1926); cf., also, Neville and Jones: Colloid
Symposium Monograph, 6, 309 (1928).
"Kolloid-Z., 44, 242 (1928) ; Z. angew. Chem., 40, 408 (1927).
"Ind. Eng. Chem., 22, 611 (1930).
60 Cf. f however, Emley: Trans. Am. Ceram. Soc., 19, 573 (1917).
01 Neville and Jones: Colloid Symposium Monograph, 6, 309 (1928).
**Cf. Williams and Westendick: J. Am. Ceram. Soc., 12, 381 (1929).
MECHANISM OF THE SETTING 73
the form of Neville's time-temperature curves and those obtained by
Budnikov, who, apparently without knowing it, used a plaster con-
taining a large amount of soluble anhydrite. Accordingly, there was
a marked rise in temperature at the outset as a result of the hydration
of the anhydrite to hemihydrate. 54 ' 63 Budnikov goes a step further
than Neville and postulates the formation of a gel around the plaster
particles which protects them from the action of water, thereby pro-
ducing the induction period which varies in length depending on the
nature of the addition agents present. The period of induction is as-
sumed to be broken by crystallization of the enclosing gel which allows
the water to act again on the plaster. This theory deserves little con-
sideration, for, if the facts are as postulated, the disappearance of
the first gel layer would merely be followed by the formation of a
new one giving a second induction period, and so on, the process
being repeated indefinitely.
The arguments for gel formation as a step in the setting of plaster
of Paris may seem quite conclusive if taken collectively. On reflec-
tion it appears, however, that all the evidence of true gel formation is
indirect. No one has really observed the formation of a gel of gypsum
prior to the appearance of the interlacing crystals in the plaster pastes.
Chassevent, 56 independently of Neville or Budnikov, observed an ini-
tial inhibition period in the time-heat curves for the hydration of
hemihydrate. It probably never occurred to him to invoke the forma-
tion of a gelatinous adsorption complex to account for this period of
inhibition since he had previously observed an inhibition period in the
crystallization of gypsum from its supersaturated solution in the ab-
sence of nuclei. Indeed, he found that solutions of gypsum contain-
ing five times the saturation value did not start to crystallize for 28
minutes when particular care was taken to exclude nuclei. The effect
of salts on the rate of setting was also observed; he states that with
potassium sulfate this "accelerates the crystallization and diminishes
the time interval during which the instable saturated solutions of hemi-
hydrate remain without crystallization."
Hansen 59 likewise failed to find any direct evidence of gel forma-
tion and, independently of Chassevent, reached the same conclusion:
"the effect of foreign material upon the rate of precipitation from its
supersaturated solution appears to explain the ability of foreign ma-
terials to accelerate or retard the setting of calcined gypsum pastes."
Although one can offer no objections to the statements of either
Chassevent or Hansen, their conclusion in the last analysis is merely
68 Chassevent: Ann. chim., 6, 265 (1926).
74 COLLOIDAL CALCIUM SULFATE; PLASTER OF PARIS
that foreign substances affect the rate of set by influencing the rate at
which gypsum precipitates from its supersaturated solution. On the
other hand, they offer no explanation of the variation in behavior
with various substances. This will be considered in the following
sections.
Effect of Gypsum Nuclei on the Rate of Set
It is a well-known fact that samples of high-grade plaster of Paris,
free from any added accelerators or retarders, show considerable
variation in the time of set. In general a plaster which exhibits a
long period of inhibition before any marked rise in temperature takes
place is designated a slow-setting plaster, whereas one with a short
period of inhibition is referred to as a rapid-setting plaster. If the
period of inhibition is caused by the building up of an adsorption
complex "whereby the two reactants are brought into chemical con-
tact," as Neville assumes, then it is not obvious why different samples
prepared by similar procedures and having the same average particle
size should show such differences in the period of inhibition. On
the other hand, if the inhibition period is merely a phenomenon of
supersaturation, the variation in the length of the period with different
samples of plaster might well be caused by variation in the number
of gypsum nuclei in the samples. Chassevent M showed that the addi-
tion of gypsum to plaster hastened the time of set, and Hansen 59
found that the time of setting was appreciably shortened 65 if a plaster
paste was made with water shaken for 35 minutes with a small amount
of plaster which ordinarily attained its maximum temperature in 75
minutes. But the importance of the presence of gypsum nuclei in
plaster of Paris on the rate at which it sets has been pretty generally
overlooked, especially by everybody who has visualized gel formation
as a stage in the setting process.
To determine the effect of gypsum nuclei on the rate of set, Weiser
and Moreland 5S added varying amounts of a standard sample of gyp-
sum to a fixed amount of plaster of Paris and determined the setting
time by the thermometric method, using a Dewar flask calorimeter.
The "standard sample" was a high-grade commercial product previ-
ously heated for 2 hours at 130 and subsequently allowed to stand in
a moist atmosphere to transform into hemihydrate the dehydrated
hemihydrate formed during ignition. The time-temperature curves for
various mixtures as given in Table VIII are shown in Fig. 12.
"Ann. chim., 6,313 (1926).
C/. Wiggin's Sons Co.: Brit. Pat. 221,853 (1923).
EFFECT OF GYPSUM NUCLEI ON THE RATE OF SET 75
TABLE VIII
EFFECT OF GYPSUM NUCLEI ON THE RATE OF SET OF PLASTER OF PARIS
Substances mixed with 35 cc water
Time to attain
maximum
temperature,
minutes
Plaster of Paris
50 g
Grams gypsum
Added
Calculated
Sample heated 5 hours
Standard sample
0.0
0.01
0.05
10
0.25
0.50
1 00
00012
00092
0098
0.052
08
0.23
50
0.94
100
64
38.0
26.5
24.0
19.0
16
14.0
FIG. 12. Time-temperature curves obtained with plaster of Paris seeded with
varying amounts of gypsum.
From the form of the curve with a plaster to which no gypsum
was added, it will be noted that at first the temperature rises abruptly
76 COLLOIDAL CALCIUM SULFATE; PLASTER OF PARIS
approximately 1, probably as a result of the heat of wetting and of
the transformation of a small amount of dehydrated hemihydrate into
hemihydrate. The initial rise is followed by a "period of inhibition"
after which there is a gradual increase in rate of reaction until a
maximum temperature is attained. The sample heated for 5 hours
(curve not shown) which was quite free from nuclei, did not attain
the maximum temperature for 100 minutes, whereas the standard
sample showed a shorter inhibition period and reached the maximum
temperature in 65 minutes. The inhibition period was appreciably cut
down by adding nuclei of gypsum until, with 0.5 g in 50 g of plaster,
it was practically zero. When the time to attain the maximum tem-
perature is plotted against the weight of gypsum nuclei added, a
parabolic curve is obtained, which shows that the rate of set ap-
proaches infinity as the number of nuclei present approaches zero.
For the particular experimental conditions, the curve is represented
by the equation W = 1.52 X 10 5 f~ 4 5 5 , where W is grams gypsum
added and t is time to attain maximum temperature. Using this equa-
tion, calculated values were obtained for the amount of gypsum pres-
ent in the samples referred to in Table VIII.
The above observations indicate that the length of the inhibition
period is influenced to a marked degree by the amount of gypsum
nuclei in the plaster paste after the mixing with water is complete.
Hence the observed inhibition period appears to result from delayed
precipitation from a supersaturated solution owing to dearth of nuclei
rather than to the time necessary to form a gel or adsorption complex.
This was confirmed by a series of observations which show the effect
of stirring the plaster-water mixtures on the rate of set.
Effect of Electrolytes on the Rate of Setting
Since the setting of plaster of Paris involves the precipitation of
gypsum from its supersaturated solution, the effect of electrolytes may
be considered in the light of von Weimarn's theory of the precipitation
process. It will be recalled that the initial velocity of precipitation is
proportional to the percentage supersaturation, W = (Q L)/L =
P/L, where Q is the total quantity of substance that is to precipitate
and L is the solubility (p. 21). The solubility of plaster of Paris in
water is approximately 0.067 mol/1, and when this hydrates to gypsum
which has a solubility of but 0.015 mol/1, the percentage supersatura-
tion is (0.067 - 0.015 )/0.015 = 3.5 = U. The initial velocity of pre-
cipitation is proportional to U, that is, W = K 3.5. In view of the
EFFECT OF ELECTROLYTES ON THE RATE OF SETTING 77
relatively long period of inhibition following the mixing of pure plaster
of Paris with water, it is obvous that this percentage supersaturation
is insufficient to cause rapid precipitation of nuclei which must be
present in abundance for a rapid reaction to take place throughout the
mass. Now, if the addition of a foreign electrolyte cuts down the
period of inhibition, it follows that the percentage supersaturation of
the solution with respect to gypsum must be greatly increased. This
may be accomplished in one of two ways : either the solubility of the
hemihydrate is increased appreciably more than that of gypsum by
the presence of the foreign electrolyte, or the solubility of the gypsum
is decreased appreciably more than that of hemihydrate by the pres-
ence of the foreign electrolyte. In the first instance the value of
(Q L)/L = P/L, the percentage supersaturation, is increased be-
cause the value of Q t which is determined by the solubility of hemi-
hydrate, is increased proportionately more than L, the solubility of
the gypsum in the medium; and in the second, P/L is increased be-
cause L is decreased proportionately more than P in the given medium.
In other words, for a foreign electrolyte to change the initial rate of
formation of nuclei as compared to the rate of formation in water
alone, all that is necessary, other things being equal, is for the ratio
of the solubility of hemihydrate to the solubility of gypsum to be
greater or less than 4.5, the ratio of the solubilities in pure water.
There is, of course, no reason to expect a constant ratio of solubilities
in widely different environments, but direct evidence of such variation
in the solubility ratios is often quite impossible to get since solutions
of foreign electrolytes added to hemihydrate usually result in such
rapid precipitation of gypsum that the solubility of hemihydrate in the
solution cannot be determined accurately. On the other hand, the ob-
servations recorded in the subsequent paragraphs furnish strong indi-
rect evidence of the expected variation in the solubility and in one
case this has been evaluated directly.
The rate of growth on nuclei already present is given by the
Nernst-Noyes equation (p. 12), which states that the velocity of
growth is proportional to Q L, the absolute supersaturation. For a
given amount Q in solution, the velocity of growth is influenced
strongly by the solubility. If L is small so that Q - L is relatively
large, the growth of particles will be relatively rapid ; whereas, if L is
large so that Q - L is relatively small, well-formed crystals will build
up slowly. As is well known, for the growth of large well-formed
crystals, the Q L value must be very small and few nuclei must be
present on which precipitation takes place.
78 COLLOIDAL CALCIUM SULFATE; PLASTER OF PARIS
Effect of Ammonium Nitrate. Since ammonium nitrate is quite
soluble and the solubility of gypsum in a wide range of concentration
is known, 86 this salt will be used to illustrate the effect of electrolytes
on the rate of set as measured by the time to attain maximum tem-
perature. The results of a series of observations are shown graphi-
cally in Fig. 13. The same general procedure outlined above was fol-
lowed, using the same plaster. It is apparent that, for concentrations
012
2468
Concentration of NH,N0 3 , Normality
FIG. 13. Effect of ammonium nitrate concentration on the rate of set of plaster
of Paris.
of salt in the neighborhood of 1 to 2 N, the inhibition period is very
small and the rate of set quite rapid; with concentrations below 1 N,
the inhibition period and the time of set become gradually longer, ap-
proaching those of pure water. The same is true for higher concen-
trations as evidenced particularly by the behavior of 10 N solution
where the inhibition period and time of set are much longer than in
Cameron and Brown: J. Phys. Chem., 9, 210 (1905).
EFFECT OF ELECTROLYTES ON THE RATE OF SETTING 79
pure water. The U-shaped form of the rate-of-set curve -indicates
that, at low concentrations of nitrate solution, the ratio of the solu-
bility of hemihydrate to gypsum is sufficiently large that a high per-
centage supersaturation of gypsum obtains. This results in prompt
precipitation of nuclei, and, because of relatively high absolute super-
saturation, the reaction goes rapidly to completion, On the other hand,
in strong nitrate solution in which gypsum is quite soluble, Q L = P
is small and P/L is so small that the initial formation of nuclei and
the subsequent growth of crystals are greatly retarded. This inter-
pretation was confirmed in a striking way by comparing the size of
crystals formed in nitrate solutions of different concentrations.
For a complete quantitative formulation of the rate of formation
and the nature of the precipitate as it is affected by the solubility of
the substances concerned, it would be necessary to know the solubility
of plaster of Paris as well as of gypsum in varying concentrations of
ammonium nitrate. At low concentrations of nitrate the rate of trans-
formation of hemihydrate is too high to obtain such data ; but in 10 AT
nitrate, the rate of transformation is sufficiently slow that the solu-
bility of hemihydrate may be obtained fairly accurately. Weiser and
Moreland 58 found the solubility in 10 N nitrate to be 0.114 mol/1 as
compared with 0.089 mol/1 for the solubility of gypsum in the salt
solution. The ratio of solubilities is thus 1.3 as compared with 4.5 in
pure water, and (Q L)/L in the nitrate solution is 0.3 as compared
with 3.5 in pure water. These data furnish a quantitative basis for
the above explanation of the observed differences in behavior of plas-
ter of Paris in different solutions. 67
Effect of Various Salts. The effect of a number of electrolytes on
the rate of set of plaster of Paris is shown graphically in Fig. 14. It
will be noted that the curves are U-shaped, 88 except the one using
ammonium acetate (see below). The similarity of the curve with am-
monium sulfate to that with ammonium nitrate is especially interesting
since the solubility curve of gypsum in ammonium sulfate passes
through a minimum. This shows that the effect of a salt on the solu-
bility of gypsum is altogether insufficient to explain its effect on the
rate of set. 69 The important thing is the ratio of the solubility of plas-
ter to that of gypsum in a concentration of salt, since this ratio deter-
mines the initial percentage supersaturation which influences greatly
6T CY. Yamane: Sci. Papers Inst. Phys, Chcm Research (Tokyo), 18, 101
(1932).
8C/., also, Gibson and Johnson- J. Soc. Chem. Ind., 51, 25T (1932).
C/. Neville: J. Phys. Chem., 80, 1037 (1926).
80 COLLOIDAL CALCIUM SULFATE; PLASTER OF PARIS
the rate o*f formation of nuclei, and the absolute supersaturation which
determines the rate of growth. The form of the ammonium sulfate
curve indicates that the ratio of the solubility of plaster of Paris to
gypsum is high even in relatively low concentrations of ammonium
sulfate where the solubility of gypsum is less than in pure water.
X-radiograms of the crystals of widely different shapes obtained
in the action of plaster of Paris in solutions of the several electrolytes
46
Concentration of Salts, Normality
10
FIG 14. Effect of varying concentrations of several electrolytes on the rate of
set of plaster of Paris.
showed them to be identical in structure with gypsum formed in the
presence of water alone.
Unlike the other salts included in Fig. 14, ammonium acetate exerts
a marked retarding action on the rate of set. This would not be pre-
dicted from the von Weimarn theory since the rate is much slower
than would be expected from solubility relations and the prevailing
supersaturation. The reason for the retarding action was traced by
EFFECT OF ELECTROLYTES ON THE RATE OF SETTING 81
Weiser and Moreland 53 to the fact that gypsum does not deposit
readily from its supersaturated solution on gypsum nuclei formed in
the presence of ammonium acetate, probably because of an adsorbed
film of acetate on th'e surface of the crystals. The addition of gypsum
crystals formed in the absence of acetate, or of broken crystals formed
in the presence of acetate, induces a rapid rate of set. Ammonium
citrate and alkali borates and biborates 70 likewise exert a retarding
action. It would appear that the action of such salts is similar to
that of small amounts of glue and gelatin, the adsorption of which
on gypsum nuclei inhibits or prevents the growth of the crystals 71
and so delays or prevents the setting of the plaster. It is said that
the ancient Romans used blood to retard the rate of set of plaster
of Paris.
Gibson and Johnson 70 found that borates even in low concentra-
tion reduced the expansion on setting of plaster of Paris. It was sug-
gested that, by adjusting the proportion of potassium sulfate, an ac-
celerator, and borax, a solution could be made up which would give
any desired rate of set and a very low expansion. It thus appears
that the rate of setting, linear expansion, hardness, and compressive
strength may be controlled by the number of gypsum nuclei present,
the plaster-water ratio, the time of mixing, the presence of soluble
salts, and, to a limited extent, by the temperature. 72
The utility of plaster of Paris in the preparation of casts depends
on its power of filling the mold by the expansion on setting and so
taking a sharp impression of its surfaces. Although there is an ex-
pansion, the final volume of the crystallized gypsum is less than the
sum of the original volumes of the materials by about 7%, as can be
calculated from the respective densities of the substances concerned. 73
The apparent density of a plaster cast is, however, less than that of
the plaster and water constituting it, since the mass is porous, being
made up of groups of interlocking crystals with spaces between them.
The manner in which the crystals are laid down is probably the impor-
tant factor in filling the mold. Borates reduce the normal expansion
on setting by modifying the form of the interlocking crystals.
The initial contraction and subsequent expansion of plaster were
well known in Persia, for up to about 1880 this property was utilized
TO Gibson and Johnson: J Soc. Chim. Ind., 51, 25T, 399T (1932).
"Ostwald and Wolski: Kolloid-Z., 27, 78 (1920); Traube: 26, 62 (1919);
cf. Serb-Serbma and Dubinskh: J Phys. Chem., U.S.S.R, 5, 1186, 1190 (1934).
C/. Johnson: Trans. Ccram. Soc., 82, 284 (1933).
"Desch: "The Chemistry and Testing of Cement," 105, 108 (1911).
82 COLLOIDAL CALCIUM SULFATE; PLASTER OF PARIS
as a means of executing criminals. 74 The condemned individuals were
placed in hollow stone columns and surrounded with wet plaster of
Paris. At first they suffered but little inconvenience, but after a time
they showed signs of distress and screamed loudly; finally, paralysis
and death supervened. This doubtless occurred as the expansion of
the plaster compressed the thorax and abdomen.
Cf. J. Soc. Chem. Ind., 26, 737 (1907).
CHAPTER V
THE COLLOIDAL CARBONATES, PHOSPHATES, CHROMATES,
AND ARSENATES
COLLOIDAL CARBONATES
Formation of Gels and Sols
Neuberg 1 prepared sols of the carbonates of calcium, strontium,
barium, and magnesium by conducting carbon dioxide into solutions of
the oxides of the respective metals in methyl alcohol. For example,
from a solution of barium oxide in the alcohol, carbon dioxide threw
down a thick gel which was subsequently peptized, giving a trans-
parent sol of the consistency of thick collodion. On concentrating the
sol by evaporation in vacuum, it set to a jelly which was repeptized
by shaking with methyl alcohol. Evaporation to dryness gave a trans-
parent celluloid-like mass.
Buzagh 2 modified Neuberg's procedure by passing dry carbon di-
oxide for several hours into a suspension of the ignited oxides in
methyl alcohol and filtering off the residue. The clear sol could be
diluted with alcohol, ether, benzene, and chloroform without coagula-
tion. Acetone and carbon disulfide gave flocculent precipitates, and a
little water caused the sol to set to a jelly from which crystals of the
metallic carbonate separated in a short time. The dispersed phase of
the sols was found to be not the normal alkaline-earth carbonates but
the dimethyl carbonates of the metals. The latter salts were obtained
in a pure crystalline state by conducting carbon dioxide for several
hours into methyl alcohol containing the finely divided oxides in
suspension.
To prepare sols of the normal carbonates, Buzagh passed carbon
dioxide into a suspension of the metallic hydroxides in methyl alcohol
and filtered off the residue. The alcohols were converted to hydrosols
by dialyzing against water. A more satisfactory method of preparing
i Neuberg and Neimann: Biochem. Z. f 1, 166 (1906) ; Neuberg and Rewald:
Kolloid-Z., 2, 321 (1908); Neuberg: Sitzber. preuss. Akad. Wiss., 820 (1907).
"Kolloid-Z., 88, 222; 39, 218 (1926).
83
84 THE COLLOIDAL CARBONATES, ETC.
the hydrosols consists in decomposing the dimethyl carbonates of the
metals with a small amount of water. These sols contain, besides the
insoluble alkaline-earth carbonate, some soluble bicarbonate and methyl
alcohol which can be removed by dialysis.
Colloidal carbonates are readily prepared by the use of protecting
colloids. Thus, colloidal carbonates of mercury, 8 lead, 8 and silver 4
are formed in the presence of the so-called protalbinates and lysalbi-
nates of sodium; and colloidal calcium carbonate by precipitation in
the presence of albumin, gelatin, and peptone. 5
Constitution of Barium Carbonate Sol
Buzagh studied the constitution and properties of barium carbonate
sol prepared by decomposition of the dimethyl carbonate. To prevent
the sol from becoming basic, the dialysis was stopped before all the
bicarbonate was removed. In order to make the barium and carbonate
content exactly equivalent, sufficient barium hydroxide was added to
convert into carbonate the bicarbonate which was known to be present
from an analysis of the supernatant liquid after coagulating a sample
of the sol. Even when the amounts of barium and carbonate were
equivalent, the intermicellar liquid was found to contain bicarbonate
ion from hydrolysis of carbonate: CO 3 + H 2 O?=fcHCO 3 "~ +
OH~. The constitution of the colloidal particles was deduced to be
{ (BaC0 3 )x[Ba(OH) 2 ]n
paired with 2HCO 3 ~" ions. From the specific conductivity of a cer-
tain sol and its ultrafiltrate together with the mobilities of the colloidal
particles and HCO 3 - ions and the concentration of BaCO 3 , Ba++,
and HCO 3 ~, it was calculated that, when n = 1, x = 120. The con-
stitution of the particle was formulated [120BaCO 3 Ba(OH) 2 -
Ba+ + ] + +, associated with two HCO 3 ~ ions. The barium carbonate
content of the particles was found to decrease with dilution and to in-
crease with particle size, as one would expect.
Although the above method may give the number of single mole-
cules in a colloidal particle, it would be well to have an independent
confirmation by determining ultramicroscopically the number of par-
ticles in a given volume of a sol of known concentration. Since the
barium carbonate sols contain particles of widely varying size, it is
obvious that average values only will be obtained by either method.
8 Leuze : "Zur Kenntnis kolloidaler Metalle und ihrer Verbindungen," 21, 28.
Paal and Voss: Ber., 87, 3862 (1904).
"Sabbatani and Salvioli: Atti ist. Veneto sci., Pt. 2, 71, 1057 (1912).
COAGULATION OF BARIUM CARBONATE SOL
85
There is no objection to expressing the supposed composition of a
colloidal particle by a formula, but it should be recognized that col-
loidal particles, in general, are not complex ions of definite composi-
tion such as a cobalt amine cation or ferrocyanide anion. The barium
carbonate particle is a finely divided portion of solid that adsorbs the
bivalent barium ions more strongly than the univalent bicarbonate ions
and so assumes a positive charge. The number of barium carbonate
molecules in one of the solid particles will vary, of course, with the
size and density of the particle which, in turn, are determined by the
conditions of precipitation, the concentration, and the purity of the sol.
Coagulation of Barium Carbonate Sol
Barium carbonate sol is quite sensitive to the action of electrolytes.
Since the particles show considerable variation in size, the sol may be
v 2 4
Concentration of Electrolyte, Millimols per Liter
FIG. 15. Fractional coagulation of barium carbonate sol by electrolytes.
precipitated fractionally by the stepwise addition of electrolytes, espe-
cially those with high precipitating power such as alkali hydroxide,
sulfate, and iodate. It is claimed that alkali halides and nitrate which
precipitate only in relatively high concentrations do not cause frac-
tional coagulation. It is much more likely, however, that the frac-
tional coagulation with the latter salts is less marked and so was over-
looked. Some observations on the fractional coagulation of the sol
are given in Fig. IS. To S-cc portions of sol containing 25 g BaCO 3
per 1 was added 5 cc of electrolyte of varying concentrations. After
standing 10 minutes, the mixture was centrifuged and the barium
86
THE COLLOIDAL CARBONATES, ETC.
content of the supernatant sol was determined. It will be seen that
the order of precipitating power of the anions is : OH > SO >
CrO 4 > IO 3 . As usual, iodate, which is taken to be a univalent ion,
behaves more like a bivalent ion. 6
The close relationship between particle size and fractional coagu-
lation is illustrated by the observations given in Table IX.
TABLE IX
RELATION BETWEEN FRACTIONAL COAGULATION OF BACOs AND PARTICLE SIZE
K 2 SO 4 in
Average diameter
millimol/1
5 cc added to
BaCOa,
not coagulated
of particles in the
uncoagulated sol,
5 cc sol
m/t
147 2
198
2
143
178
4
128
148
6
73
89
0,8
28
77
COLLOIDAL PHOSPHATES
The slightly soluble orthophosphates of lead and the alkaline-earth
metals are precipitated in a gelatinous form by the interaction of
alkali phosphate and the metallic halides. Sols of these compounds
have not been prepared in the absence of protecting colloids. De Toni 7
made sols of calcium orthophosphate by mixing hot solutions of so-
dium phosphate and calcium chloride in the presence of gelatin, gum
arabic, blood serum, and starch. A sol is formed, also, by mixing
dilute solutions of phosphoric acid and calcium hydroxide in the pres-
ence of gelatin. Hatschek * obtained rhythmic bands of calcium phos-
phate by allowing calcium chloride to diffuse into a gelatin gel con-
taining normal sodium phosphate. Phosphate clays possessing the
properties of typical colloidal gels have been described by Elschner. 9
Colloidal silver orthophosphate is prepared by mixing 0.05 N
solutions of AgNO 3 with a slight excess of 0.05 N Na 3 PO 4 or
*Cf Weiser and Middleton: J. Phys. Chem., 24, 51 (1920).
'Kolloid-Z., 28, 145 (1921).
s Kolloid-Z., 27, 225 (1920).
Kolloid-Z., 31,94 (1922).
COLLOIDAL SILVER CHROMATE 87
Na 2 HPO4. 10 The sol precipitates slowly on standing, but it may be
stabilized by adding sodium "lysalbinate" or "protalbinate." *
Colloidal lead phosphate for use in cancer treatment 11 may be
made by adding sodium phosphate solution drop by drop to a boiling
solution of lead chloride containing gelatin, followed by centrifuging
to remove any unpeptized material. The colloidally dispersed lead
phosphate is said to be less toxic than lead acetate, colloidal lead, or
colloidal lead oxide. 12
COLLOIDAL CHROMATES
Colloidal Silver Chromate
Sols of silver chromate are formed by bringing together silver
13
nitrate and potassium chromate in the presence of gelatin or sugar.
On mixing the solutions in the presence of gelatin, the resulting solu-
tion is yellow for a time and then turns red. The cause of this be-
havior has been the subject of several investigations and is still open
to some question. Williams and MacKenzie, 14 Bolam and MacKen-
zie, 15 and Bolam and Donaldson 16 made electromotive force and con-
ductivity measurements which show that, prior to the appearance of
the red color, the activity of the silver ion in the yellow mixture is
very much higher than in a pure saturated solution of silver chromate
at the same temperature and that when the red color appears the
activity of the silver ion decreases simultaneously to a marked degree.
Desai and Nabar " found, further, from electrometric measurements
that the activity of silver ion in the mixture remains constant for some
minutes, then decreases rapidly, and finally becomes constant ; and that
the point at which the activity commences to fall corresponds in every
instance to the appearance of the red color. The obvious conclusions
from these observations is that the yellow mixture contains a highly
supersaturated solution of silver chromate and that the red color is
caused by the actual formation of the solid phase rather than by
coagulation of a highly dispersed solid phase already present.
"Lottermoser: J. prakt. Chem., (2) 72, 39 (1905).
"Bischoff and Blatherwick: J. Pharmacol., 81, 361 (1927).
C/. Kehoc and Thamann: J. Lab. Clin. Med., 19, 178 (1933).
"Lobry de Bruyn: Rec. trav. chim., 19, 236 (1900) ; Ber, 35, 3079 (1902).
"J. Chem. Soc., 117, 844 (1920).
"Trans. Faraday Soc., 22, 162 (1926); Bolam and Desai: 24, SO (1928).
18 Trans. Faraday Soc., 29, 864 (1933).
"Trans. Faraday Soc., 28, 449 (1932); Nature, 127, 628 (1931).
88 THE COLLOIDAL CARBONATES, ETC.
Dhar and collaborators, 18 on the other hand, made similar observa-
tions which indicate that two-thirds of the silver chromate is present
in the colloidal state from the start. The yellow mixture was believed
to be a sol in which the particles are negatively charged by adsorption
of chromate ions, and the red a less stable sol, in which the particles
are positively charged by adsorption of silver ions. Chatterji 19 failed
to observe a decrease in the electromotive force with time. In* accord
with this, later work of Naik, Desai, and Desai 20 indicates that what-
ever changes take place in the conductivity of the mixture occur im-
mediately on mixing the solutions and that there is no gradual decrease
in conductivity with changes in the color of the mixtures. It was
therefore concluded that the yellow mixtures contain some solid par-
ticles in a very highly dispersed condition, and that changes in color
result from growth and agglomeration of very fine particles into larger
ones.
These and other conflicting observations, both in different labora-
tories and in the same laboratory, may be due in part to differences in
reagents, especially the gelatin, in the />H-value of the mixtures, and
in the exact method of procedure. In this connection it has been found
that hydrolyzed gelatin produces more highly dispersed silver chro-
mate than unhydrolyzed gelatin. 21 It must be borne in mind, also,
that the inhibiting action of gelatin may be quite complicated, involving
one or more of the following: (1) prevention of the formation of
crystallization centers, thereby producing a highly supersaturated solu-
tion; (2) prevention of growth of particles, thereby giving a system
in which the particles may approach molecular dimension; and (3)
prevention of agglomeration of particles. It is a mistake to assume
that the inhibiting action gives rise to supersaturation only. By suit-
able adjustment of the temperature, amount of gelatin, pH of the
gelatin, and concentration of the reactants Khanolkar, Barve, and
Desai 22 found that the conductivity of the mixture (1) may not de-
crease at all while the color remains yellow, (2) may decrease some
time after the color change, (3) may not change at all in spite of the
color change.
" J. Phys. Chcm., 28, 41 (1924) ; Kolloid-Z., 64, 270 (1924) ; J. Indian Chem.
Soc., 5, 175 (1928) ; Trans. Faraday Soc., 23, 23 (1927).
iProc. Indian Sci. Congr., 10 (1932).
a J. Indian Chem. Soc., 11, 45; Desai and Naik: 59 (1934) ; J. Univ. Bom-
bay, 2 II, 90 (1933).
"Bolam and Desai: Trans. Faraday Soc., 24, 50 (1928) ; Ganguly: J. Indian
Chem. Soc., 3, 177 (1926).
Indian Acad. Sci., 4A, 468 (1936).
COLLOIDAL LEAD CHROMATE 89
The behavior of silver and lead iodide in gelatin solutions is similar
to that of silver chromate (p. 139).
The phenomenon of rhythmic precipitation was discovered in con-
nection with silver chromate when Liesegang 23 placed a drop of silver
nitrate on a glass plate coated with moist gelatin containing a small
amount of potassium dichromate and obtained the series of rings now
so well known. A similar experiment can be carried out in a test
tube, giving the so-called rhythmic bands. Numerous investigations
with silver chromate and many other substances have been carried out
with the end in view of determining the mechanism of the banding
process. The various theories have been considered elsewhere (Vol.
II, pp. 24, 180) and will not be repeated here. The inhibiting action
of gelatin on the precipitation of silver chromate must influence the
ring formation in gelatin jelly. This, however, is of minor importance
in a general theory of rhythmic banding since silver chromate will
precipitate rhythmically under suitable conditions in the absence of a
jelly.
Colloidal Lead Chromate
A lead chromate sol is formed by precipitation from dilute solu-
tions in the presence of gelatin. The precipitated salt is peptized to a
certain extent by a boiling concentrated solution of potassium nitrate. 24
If lead sulfate suspended in potassium nitrate solution is treated with
potassium chromate, lead chromate is thrown out in the form of a
greenish fluorescent sol. 28
Lead chromate is a valued yellow pigment known as chrome yellow.
The color varies from a light yellow to orange, depending on the con-
ditions of precipitation and the subsequent treatment. The canary
yellow product thrown down from a chromate solution with lead
acetate or nitrate changes to an orange on washing. To maintain the
desired yellow tone, the precipitation is carried out in the presence of
sulfate so that lead chromate and lead sulfate come down simul-
taneously. The role of the lead salt has been attributed by Habich 2fl
to the formation of a double salt, PbCrO 4 PbSO 4 or PbCrO 4 --
2PbSC>4, but tne existence of such compounds has not been estab-
23 Phot, arch., 37, 321 (1896); "Chemische Reaktionen in Gallerten," Dussel-
dorf (1898); Z. anal. Chem, 60, 82 (1911); Kollotd-Z, 9, 296 (1911); 12, 74,
269 (1913) ; 18, 76 (1915) ; Z. physik. Chem., 88, 1 (1914).
2 *Oeschsner de Coninck: Bull. acad. roy med Belg., 665 (1909).
25 Milbauer and Kohn: Chem.-Ztg., 46, 1145 (1922).
**See Amsel: Z. angew. Chem., 9, 613 (1896).
90 THE COLLOIDAL CARBONATES, ETC
lished. Jablczynski 27 attributes the stabilizing action of lead sulfate
to lead ions which cut down the hydrolysis of lead chromate thereby
preventing the formation of "chrome red," basic lead chromate. This
hypothesis appears untenable since lead sulfate exerts no stabilizing
action unless it is precipitated simultaneously with the chromate.
Moreover, lead sulfate is itself hydrolyzed appreciably, 28 and it is not
obvious how it could prevent the hydrolysis of the chromate. Gobel 29
suggests that the lead sulfate prevents the coalescence of the fine yellow
particles of lead chromate into larger particles which are darker in
color. 80 This plausible suggestion is treated lightly by Milbauer and
Kohn, 81 who conclude, from a microscopic examination of the crystals
and the action of the solvents on them, that the stable yellow pigment
is a solid solution of lead sulfate and lead chromate. Lead sulfate,
being more soluble than lead chromate, has a greater tendency to come
down in crystals from dilute solution. For this reason, it is argued
that the technical production of chrome yellow is carried out in very
dilute solutions and with continuous stirring in order to induce the
simultaneous precipitation of the substances as mixed crystals.
The conclusions of Milbauer and Kohn have been confirmed and
extended by Wagner 82 and by Quittner, Sapgir, and Rassudowa a3
using x-ray analytical methods. Under the usual conditions of pre-
cipitation lead chromate comes down in the form of extremely minute
rhombic grains which appear bright yellow. The rhombic lead chro-
mate is the least stable form and goes over into the more stable mono-
clinic modification which may possess an undesirable dark yellow color.
Larger and stabler crystals of the rhombic chromate are formed as
mixed crystals with lead sulfate under suitable conditions. The color
is not alone a question of crystal structure since mixed crystals of
nPbCrO4 mPbSO^ having a bright yellow color may be either
monoclinic or rhombic depending on the ratio of m:n. It is always
monoclinic if the mixed crystals contain more than 50% of PbCrO 4 .
The rhombic form may be stabilized by protecting colloids. For ex-
"Chem. Ind. (Ger.) f 31, 731 (1908).
"Dolezalek: Z. Elektrochem., 5, 533 (1899); 6, 557 (1900).
Chem.-Ztg., 28, 544 (1899).
sC/. Free: J. Phys, Chem., 18, 114 (1909); Bock: Farben-Ztg, 26, 761
(1920) ; Wagner and Keidel: 81, 1567 (1926).
"Z. physik. Chem., 91, 410 (1916); Chem.-Ztg., 46, 1145 (1922).
"Z. angew, Chem., 44, 665 (1931); Wagner, Haug, and Zipfel: Z. anorg.
Chem., 208, 249 (1932).
88 Z. anorg. Chem., 204, 315 (1932); Sapgir, Rassudowa, and Kvitner:
Lakokrasochnuyu Ind., Za, 1, 56 (1932).
COLLOIDAL LEAD CHROMATE 91
ample, in the precipitation of the mixed crystals, if lead nitrate is
added to a mixture of dichromate and aluminum sulfate (instead of
sulfuric acid) followed by careful post-precipitation of hydrous alu-
mina with alkali, the rhombic crystals are surrounded by a film of
hydrous oxide which inhibits the transformation to the rhombic form. 84
The tetragonal form of lead chromate can be obtained in the form
of orange-red mixed crystals with lead molybdate.
The products known as the "chrome reds" are bright red pigments
formed by adding basic lead acetate to a solution of alkali chromate.
Between these and the various "yellows," a number of different shades
of chrome orange may be prepared.
The basic lead chromates formed by precipitation in the presence
of alkali, or by the action of alkali on lead chromate, vary in color
with the alkali concentration. With relatively low alkali concentra-
tions there results first an orange-red color, then, with increasing alkali
concentration, a pure red uniform product, and finally, a non-uniform
product. Whatever the color, the product has the same composition
PbO PbCrO 4 and gives the same x-radiogram, 85 the crystals belong-
ing to the tetragonal system ; variations in color are caused by differ-
ences in particle size and shape. If the basic chromate is allowed to
stand in the mother liquor in contact with the air, a clear yellow double
salt K 2 CO 3 PbCrO 4 is formed, This may be converted into the
orange-red basic salt by the action of alkali. The relationships among
the three compounds may be represented as follows :
KOH KOH, KjCr0 4 , COj
PbCr0 4 > PbO-PbCr0 4 ( S K 2 C0 3 -PbCr0 4
Dark yellow Orange to red Ca(OH) a Clear yellow
The lead chromate colors are ordinarily used as pigments in paints,
but they may be precipitated directly on cellulose 86 or wool 8fl fibers
which adsorb them sufficiently strongly that the color is permanent.
Yoshida and Matsumoto 87 claim that the crystals so produced belong
to the tetragonal system.
"C/. Bock: Farben-Ztg., 82, 459 (1926).
38 Wagner and Schirmer- Z. anorg. Chem., 222, 245 (1935).
"Il'inskii and Kozlov: J. Russ. Phys.-Chem. Soc, 82, 665 (1930).
87 J. Soc. Chem. Ind., Japan, 88, Suppl. Binding, 113 (1935); Matsumoto:
89, 180 (1936).
92 THE COLLOIDAL CARBONATES, ETC.
COLLOIDAL ARSENATES
The Precipitated Salts
The addition of alkali arsenates such as KH 2 AsO 4 to solutions of
bivalent heavy metal salts gives precipitates that vary in form from
gelatinous to jelly-like depending on the conditions. According to the
von Weimarn theory (p. 11), mixing dilute solutions which interact
at once to give a highly insoluble precipitate should give a gelatinous
precipitate but not a jelly. The reason is evident when we consider
the impossibility of getting the instantaneous mixing of the solutions,
which is essential for uniform precipitation throughout the solution.
One part is precipitated before another is mixed with the precipitant,
and the homogeneity characteristic of a jelly is lost. Moreover, the
mixing itself will tend to destroy the jelly structure. The results are
therefore not unlike those obtained when a colloid, capable of forming
a jelly by slow precipitation, is coagulated too rapidly by the addition
of excess electrolyte (cf. Vol. II, p. 186). It would seem, however,
that precipitation of a hydrous substance as a result of double decom-
position might form a jelly instead of a gelatinous precipitate provided
the thorough mixing of the solutions could be effected before precipi-
tation began and provided the precipitation, once started, proceeded at
a suitable rate. Such conditions do not obtain as a rule ; but they are
quite possible theoretically. Thus the precipitation may be the result
of a step wise reaction, one step of which proceeds at a suitably slow
rate. It is further possible to have a reaction that proceeds very
slowly at low temperatures but with marked velocity at higher tem-
peratures. This would not only allow of mixing without precipitation
but would also enable one to control the subsequent rate of reaction
by a suitable regulation of the temperature.
Such a favorable combination of circumstances apparently obtains
when a manganese salt of a strong acid and KH 2 AsC>4 are mixed.
The latter salt ionizes thus: KH 2 AsO 4 ^K+ + H 2 AsO 4 - ; but on
account of the solubility of Mn(H 2 AsO 4 ) 2 no Mn++ ions are re-
moved from solution by interaction with H 2 AsO 4 ~. The latter ion,
however, undergoes secondary ionization to a slight degree as follows :
H 2 AsO 4 -**H+ +HAsO 4 ; and insoluble MnHAsO 4 is formed
in accord with the following equation: Mn+++HAsO 4 -
MnHAsO 4 . 88
Since the precipitation of MnHAsO 4 is accomplished by the for-
mation of an equivalent amount of free hydrogen ion in solution, an
z: Kolloid-Z., 14, 139 (1914).
THE PRECIPITATED SALTS
93
equilibrium is set up which prevents the complete precipitation of the
manganese. The amount of MnHAsC>4 formed, however, and the rate
of formation by the above process are apparently influenced to a
marked degree by the temperature, so that good jellies can be ob-
tained by mixing dilute solutions of the necessary salts in the cold and
allowing the mixture to stand at room temperature or warming to a
suitable temperature. This has been demonstrated with the arsenates
of manganese, cobalt, ferrous iron, cadmium, and zinc. 39 The results
with manganese arsenate are recorded in Table X. Solutions of N
TABLE X
PRECIPITATION OF MANGANESE ARSENATE
Electrolytes mixed, cc
Final
. .
volume
Nature of precipitate
MnCU
KII 2 AsQ 4
10
10
20
Firm jelly, almost clear
10
20
30
Firm jelly, almost clear
10
20
50
Firm jelly, perfectly transparent
10
20
100
Firm jelly, perfectly transparent
10
20
200
Soft jelly, perfectly transparent
10
20
300
No jelly
NaH 2 AsO 4
10
10
20
Cloudy jelly, not uniform
10
10
50
Cloudy gelatinous precipitate
MnCl 2 and KH 2 AsO 4 were prepared from freshly boiled water. This
precaution was necessary to prevent the formation of air bubbles in
the jelly. The solutions were cooled to and suitable amounts of
each were placed in 60-cc test tubes in the ratio shown in the table.
The solutions were diluted with cold water to the final volume shown
in column 3 or an aliquot part thereof. After rapid mixing, the solu-
tion was set aside for IS to 20 minutes and, if jelling had not begun,
the tube was warmed by dipping carefully into boiling water until
precipitation started and was then allowed to stand quietly. The
jellies obtained in this way are quite stable, showing little tendency to
cloud up and crystallize on standing in the cold; but on heating, crys-
tals of MnHAsO 4 are formed. Good jellies are not obtained with
s Weiser and Bloxsom: J. Phys. Chem , 28, 32 (1924).
94 THE COLLOIDAL CARBONATES, ETC.
Na 2 HAsO 4 since the precipitation is too rapid to allow time for mix-
ing and the formation of the jelly structure.
Ar senate Sols
The addition of excess ferric chloride to a normal solution of diso-
dium arsenate gives a sol which is said to be Fe 2 (HAsO4) 3 but which
may be, at least in part, an adsorption complex of hydrous ferric oxide
and arsenic oxide. This sol is precipitated as a jelly either by dialy-
sis 40 until the concentration of the stabilizing hydrogen ion is reduced
below a critical value or by the addition of a precipitating electrolyte. 41
The so-called aluminum arsenate sol prepared in the same way as the
iron sol exhibits similar properties.
The most important colloidal arsenates are the calcium and lead
salts which are dusted or sprayed on plants or trees for the purpose
of destroying insects. If the finely divided precipitated salts are sus-
pended in water they assume a negative charge. Since the surfaces
of leaves possess a like charge, the insecticide does not adhere to them
readily and is soon washed off by rain or dew. The particles of salt
adhere much more tenaciously if they are prepared in such a way that
they adsorb a positive ion which gives them a positive charge when
wet. Thus, a positively charged basic calcium arsenate made under
commercial manufacturing conditions, not specified, and tested in the
field was found to possess an adherence from 200 to 260% greater
than the ordinary precipitated salt. 42
The lead arsenate precipitated from lead nitrate solution with diso-
dium arsenate is the acid salt which does not become positively charged
under ordinary conditions. On the other hand, the basic salt thrown
down from lead acetate solutions readily assumes a positive charge. 43
The acid arsenate is assimilated to a greater degree than the basic
arsenate; hence the former possesses higher insecticidal efficiency. 44
The acid salt may be made adherent by coating the fine particles with
a film of lead oleate. 48 This is accomplished by suspending the salt in
"Grimaux: Compt rend., 98, 1540 (1884); Holmes and Rindfusz: J. Am.
Chem. Soc., 38, 1970 (1916) ; Holmes and Arnold: 40, 1014 (1918) ; Holmes and
Fall: 41,763 (1919).
Weiser and Bloxsom: J. Phys. Chem., 28, 26 (1924).
"Moore: Ind. Eng. Chem., 17, 465 (1925); J. Econ. Entomol, 18, 282
(1925).
Moore: Ind. Eng. Chem, 17, 466 (1925).
"Lovett: Oregon Agr. Expt. Sta. Bull., 169, 1 (1920); Chem. Abstracts, 16,
3718 (1921).
Van Leeuwen: J. Econ. Entomol., 18, 744 (1925).
ARSENATE SOLS 95
water containing a known amount of sodium oleate and adding an
equivalent amount of lead acetate with vigorous stirring. The dried
product is not wetted by water, but the suspension spreads evenly on
foliage and adheres firmly to the surface. The charge on the acid
arsenate particles may be rendered positive and adherent to negative
surfaces by the addition of a suitable amount of aluminum ion. 46
Woodman 47 demonstrated that leaves can be wetted by a spray
liquid provided the surface tension is reduced below a critical value
said to be in the neighborhood of 32 dynes/cm. 48 This can be accom-
plished cheaply by the addition of a small amount of soap. The maxi-
mum amount of spray liquid is retained at a critical surface tension ;
at higher tensions the wetting of the leaves is imperfect, and at lower
values the wetting power is not increased but the spreading power is
greatly augmented. After the surface tension has been reduced to
the point where the foliage is wetted perfectly, an increase in the vis-
cosity of the spray liquid causes an increase in the amount of liquid
retained as a film by the leaves; hence the use of gelatin in concentra-
tions of about 0.3% is advantageous. Moreover, gelatin 49 4T in con-
centrations between 0.2 and 0.5 % is a better stabilizer for lead arse-
nate sols than sodium caseinate, dextrin, starch, or soap.
"Hensill and Hoskins: J. Econ. Entomol., 28, 942 (1935); Hoskins and
Wampler: 29, 135 (1936).
J. Pomology Hort. Sci., 4, 78, 95, 184 (1925).
48 Cf., however, Robinson: J. Agr. Research, 31, 71 (1925).
"Brinley: J. Agr. Research, 26, 373 (1923).
PART II
THE COLLOIDAL HALIDES
CHAPTER VI
COLLOIDAL SILVER IODIDE
PRECIPITATED SILVER IODIDE
The silver halides are sufficiently insoluble that they are ordinarily
precipitated in the form of curds of minute crystals on mixing cold
solutions of alkali halide and silver nitrate. Silver iodide, which is
less soluble (9.7 X 10-* mol/1) than either the bromide (6.6 X 10~ 7
mol/1) or the chloride (125 X lO" 6 mol/1), comes down in the most
finely divided form under the same conditions of temperature and con-
centration of reacting solutions. All three halides are sufficiently
soluble that the colloidal precipitates which are first formed age rather
rapidly on standing to give agglomerates of crystals which are larger
and more perfect, and possess a lower adsorption capacity than the
freshly formed curds. The rate of aging of the salts is in the order:
Agl < AgBr < AgCl, which is the reverse of the order of the sta-
bility of the gels and sols of the halides.
The precipitation of silver halides in the absence of an excess of
either positive or negative ions is attributed by Lottermoser to coagu-
lation of the particles into larger secondary aggregates which settle
out. Jablczynski 1 objects to this mechanism of the precipitation on
the ground that it does not explain the velocity of increase of turbidity
in the presence of gelatin or other protecting colloids which inhibit
precipitation even in low concentrations. The view held by Jablczynski
is that the formation of a precipitate and the process of grain growth,
both in the presence and in the absence of gelatin, are entirely a process
of recrystallization by solution of the smaller grains and growth of
larger ones at their expense (Ostwald ripening) until the larger grains
are thrown down. This contention of Jablczynski is contradicted by
Sheppard and Lambert's 8 observations of reversible flocculation and
deflocculation of a part of the silver halides precipitated in the absence
1 Jablczynski, Fordonski, Frankowski, Lisiecki, and Klein: Bull. soc. chim,
(4) 88, 1392 (1923).
2 Colloid Symposium Monograph, 4, 281 (1926).
99
100 COLLOIDAL SILVER IODIDE
of gelatin, and by microscopic examination which reveals the presence
of secondary aggregates consisting of loosely adhering primary par-
ticles. In the absence of protecting colloids, it thus appears that ag-
gregation is the important factor in causing precipitation, although
some recrystallization does take place; whereas, in the presence of
protecting colloids, the coagulation is largely inhibited, at least until
high concentrations of the reactants are used. The grain growth ob-
served in the ripening of photographic emulsions is caused primarily
by a process of recrystallization from solution in which crystal aggre-
gation plays but a minor role. 8
The crystals of silver iodide thrown down in the presence of excess
Ag+ ion are cubic whereas those precipitated in the presence of ex-
cess I" ion are hexagonal. 4
Adsorption by Silver Iodide
Adsorption of Silver Ions and Iodide Ions. Lottermoser 5 first
called attention to the fact that sols are formed on mixing dilute solu-
tions containing silver ions and halide ions provided one or the other
is present in excess ; with silver ions in excess the particles are posi-
tively charged, and with iodide ions in excess the particles are nega-
tively charged. The charge results from the preferential adsorption
of the potential-determining ions : 6 silver with the positive sol and
iodide with the negative sol. Since the range and degree of stability
of negative sols are greater than those of positive sols (p. 114) it
would appear that iodide ion should be more strongly adsorbed than
silver ion. This is borne out by some observations of Lottermoser
and Rothe 7 on the adsorption of potassium iodide and silver nitrate,
respectively, by precipitated and thoroughly washed silver iodide. The
adsorption of iodide reaches a maximum which is considerably greater
than that of silver nitrate at the same concentration. The maximum
in the adsorption isotherm of potassium iodide is caused by growth of
the silver iodide crystals in the presence of the larger amounts of
potassium salt (cf. with the behavior of cadmium sulfide in the pres-
ence of hydrochloric acid, p. 256). Even in very low concentrations
iodide ion appears to be adsorbed more strongly than silver ion since
8 Sheppard: Colloid Symposium Monograph, 1, 346 (1923).
'Bloch and Moller: Z. physik. Chem., A162, 245 (1931); Kolkmeijer and
Hengel: Z. Krist, 88, 317 (1934).
J. prakt. Chem., (2) 72, 39 (1905).
Lange and Berger: Z. Etektrochem., 36, 171 (1930).
*Z. physik. Chem., 62, 359 (1908).
ADSORPTION BY SILVER IODIDE
101
Lange and Crane 8 showed that aged and very thoroughly washed sil-
ver iodide is negatively charged in contact with its saturated solution.
On the other hand, Lange and Berger 9 concluded from potentiometric
titration experiments that silver ions are more strongly adsorbed than
iodide ions at equal equilibrium concentrations. But as Kolthoff and
Lingane 10 point out, this conclusion was based on the erroneous as-
sumption that neither silver nor iodide ions are adsorbed at the po-
tentiometric end point where CAg* = Cr in solution. It was demon-
strated by Kolthoff and Lingane that the isoelectric point of freshly
precipitated silver iodide at room temperature is at a silver ion con-
centration of approximately
10~ 6 molar (/>Ag = 6) and
that, at the equivalence
potential, pAg ==/>! = 7.83,
the freshly formed precipi-
tate retains an excess of
adsorbed iodide correspond-
ing to 0.09% of the total
amount of iodide in the pre-
cipitate. This conclusion was
based on precision poten-
tiometric titrations of silver
with iodide from which the
adsorption of silver and
iodide ions was calculated
by comparing the experi-
mentally determined titra-
tion curve with the theo-
retical curve based on the
35 7 9 11
pAg Observed
FIG. 16. Adsorption of silver and iodide
ions by freshly precipitated silver iodide.
assumption that the precipitate has no adsorptive properties. The
results are shown graphically in Fig. 16.
In investigating the precision of the potentiometric silver iodide
titration, Lange and Berger showed that the adsorption of both silver
and iodide ions is represented by the expression dx = a dlog c instead
of by the Freundlich equation, dlog x = a dlog c, where x is the
amount adsorbed, a is a constant, and c is the equilibrium concentra-
*Z. physik. Chern., A141, 225 (1929); cf , also, Labes: Z. physik. Chem.,
116, 1 (1925); Kruyt arid van der Willigen: A139, 53 (1928); Mukherjee and
Kundu: J. Indian Chem Soc., 3, 335 (1926).
*Z. Elektrochem., 86, 171, 980 (1930).
10 J. Am. Chem. Soc., 68, 1524, 1528 (1936).
102 COLLOIDAL SILVER IODIDE
tion of the ion being adsorbed. Similarly, the adsorption data given
in Fig. 16 can be represented by the equation d x = a rflog c in the
/>Ag range between 4 and 7.2, but at higher values the adsorption is
less than corresponds with this equation, probably because of the
rapidity with which fresh precipitates of silver iodide age in contact
with potassium iodide. As we shall see, Verwey and Kruyt 11 found
that the equation applies to the adsorption of iodide by an aged silver
iodide sol over a much wider range than Kolthoff and Lingane found
with a fresh precipitate. Verwey and Kruyt could not measure di-
rectly the isoelectric point without flocculation of the sols, but with
the aid of the above equation they extrapolated to the point of zero
adsorption (isoelectric point) and found it to be at ^Ag = 6, in good
agreement with the value found directly by Kolthoff and Lingane.
On the other hand Kruyt 12 showed by endosmotic experiments that
fused silver iodide cannot be charged positively even with 0.001 N
AgNO 3 . It appears, therefore, that the zero point of the electrokinetic
charge depends on the nature of the silver iodide. In aged dialyzed
sols the charge is reversed at pAg = 6.
From exhaustive studies of the precision of the iodide-silver po-
tentiometric titration, Lange and Berger 9 concluded that the titration
should give results accurate to 0.01% if made under the most
favorable conditions. The titration was found to be altogether im-
practical when carried out at room temperature because of strong
adsorption of silver and iodide ions by the voluminous precipitate and
the consequent slowness with which a steady e.m.f. was attained.
These objectionable effects of adsorption were practically eliminated
by titrating at 90 where an aged precipitate of low adsorption ca-
pacity results.
Lange and Berger's results were confirmed and the accuracy of the
method established by Kolthoff and Lingane working with highly
purified reagents. In the slow titration of silver with iodide, the error
at room temperature amounts to 0.1% ; at 70, to 0.048% ; and at 90,
to 0.017%. By digesting the precipitate at 90 near the equivalence
point, and finishing the titration at room temperature, the error was
0.028%. In every precipitate an excess of iodide was adsorbed at the
equivalence point.
Honigschmid and Striebel 1S determined the atpmic weight of
iodine by means of silver iodide. One sample precipitated in nitric
11 Z physik. Chem., A167, 149 (1933) ; Verwey: Kolloid-Z., 72, 187 (1935).
12 Physik. Z. Sowjetunion, 4, 295 (1933).
i*Z. anorg. Chem., 208, 53 (1932).
ADSORPTION BY SILVER IODIDE 103
acid solution with excess silver ion, washed, dried, and fused, did not
contain enough excess silver to be detected.
Adsorption of Inorganic Ions and Paneth-Fajans' Rule. In an
earlier chapter (p. 34) attention was called to Paneth's 14 observation
that radium is strongly adsorbed by barium sulfate and chromate
whereas it is not adsorbed at all by chromic oxide or silver chloride.
Since radium sulfate and chromate are insoluble and radium oxide
and chloride are soluble, Paneth formulated the adsorption rule for
radio elements: ions will be relatively strongly adsorbed if the com-
pound with the oppositely charged ions of the crystal lattice is slightly
soluble. This rule was confirmed and extended by Fajans and his
coworkers 15 to include elements other than radio elements. The
original Paneth-Fajans rule may be stated as follows: ions will be
strongly adsorbed on an ion lattice which form a difficultly soluble or
weakly dissociated compound with the oppositely charged ion of the
lattice. An important example of this rule, the strong adsorption of
common ions by an ion lattice, was considered in the preceding sec-
tion. In further accord with the rule, Fajans and Beckerath found
that both thorium B and lead ions are quite strongly adsorbed by
silver halide made negative by preferential adsorption of halide ion,
and it is known that the halides of thorium B and lead are not easily
soluble. Moreover, the adsorption of thorium B is greater than that
of lead, and the halides of the former are less soluble than those of
the latter. Fajans and Beckerath showed further that thorium B and
lead ions are not adsorbed by a silver halide made positive by prefer-
ential adsorption of silver ion. This is because the adsorption of
thorium B at the concentration used is not sufficiently great to dis-
place the more strongly adsorbed silver ions.
Beekley and Taylor 18 investigated the relation between the solu-
bility of various silver salts and their adsorbability by silver iodide.
The effect of valence was eliminated by choosing salts with univalent
anions only. The results of their observations are represented in
Fig. 17. The number in parentheses following the formula of the salt
"Paneth: Physik. Z., 16, 924 (1914); Horovitz and Paneth: Z. physik.
Chem., 89, 513 (1915).
"Fajans and Beer: Ber., 46, 3486 (1913); Fajans and Richter: 48, 700
(1915); Fajans and Beckerath: Z. physik. Chem., 97, 478 (1921); Fajans and
Frankenburger: 106, 255 (1923) ; Fajans and Sterner: 126, 309 (1927) ; Fajans
and Erdey-Gruz: A168, 97 (1931).
16 J. Phys. Chem., 29, 942 (1925); cf. Mukherjee, Basu, and Mukherjee:
J. Indian Chem. Soc., 4, 459 (1927) ; Chakravarti and Dhar: Kolloid-Z., 46, 12
(1528).
104
COLLOIDAL SILVER IODIDE
is the relative solubility, silver bromate, the most insoluble salt, being
taken as unity. It will be seen that a quantitative relationship between
solubility and adsorbability is not approached. The most that can be
said is that the less soluble salts tend to be more strongly adsorbed
and the more soluble salts less strongly adsorbed. Among the more
soluble ones, silver nitrate, though nineteen times as soluble, is ad-
sorbed more strongly than silver chlorate ; and among the less soluble
ones, silver acetate, though eight times as soluble, is adsorbed much
more strongly than silver bromate. Moreover, the jump in adsorption
between the relatively insoluble and relatively soluble salts is not
O.OD8
0.004 0.008 0.012 0.016
Equilibrium Concentration, Millimol per Gram Solution
0.020
FIG. 17 Adsorption of silver salts by silver iodide. (Number in parentheses is
the relative solubility of the silver salts )
marked ; e.g., the difference in adsorption between bromate and acetate
is no greater than the difference in adsorption between bromate and
nitrate, although the ratio of solubilities of the first pair is 8.2 and
that of the second pair is 100.
From these observations and others already mentioned (p. 29) it
is obvious that solubility is not the only factor which determines the
degree of adsorption of ions on ion lattices and may not be the most
important one in certain cases. Fajans considers adsorption of ions
by a crystal to take place through the agency of the residual valences
of the ions on the surface of the lattice. Thus, when iodide ion is
adsorbed on silver iodide, it becomes attached to the silver ion and
ADSORPTION BY SILVER IODIDE 105
so becomes an integral part of the silver iodide lattice. In so doing
the adsorbed ion loses its hydration shell," the dehydration being ac-
companied by a marked energy change. From this point of view it
would follow that, the greater the affinity of an ion for water, in other
words the more highly hydrated it is, the less strongly it will be
adsorbed (p. 352). A third factor which must be taken into account
is the relative size of the adsorbed and the lattice ions, irrespective of
whether the ions are adsorbed directly on the surface or enter into
kinetic interchange with an ion on the surface (p. 54). A fourth
factor is the deformation which an ion undergoes in a field of force.
In general, the more readily deformable an ion the better it will be
adsorbed to the oppositely charged ion of the lattice. 18
In opposition to the Paneth-Fajans rule in its original form Hahn
and Imre 19 observed a number of cases in which a radio element is
not adsorbed even when it forms a difficultly soluble compound with
the oppositely charged ion of the lattice. For example, the failure
of thorium B to be adsorbed appreciably by positively charged silver
halide is cited as being in opposition to the rule since thorium B forms
sparingly soluble halides. Also oxalate ion is not appreciably ad-
sorbed by silver bromide and silver iodide, and chloride is not ad-
sorbed very strongly by silver iodide. Hahn attempted to make the
adsorption rule more general by the following formulation : an ion in
dilute solution is adsorbed on a precipitate if the surface of the pre-
cipitate has a charge opposite to that of the ion being adsorbed and
if the resulting compound is difficultly soluble in the solvent. It would
follow from this that a neutral surface would adsorb only ions com-
mon to the lattice or isomorphous with it. 20 In opposition to this
Fajans 21 showed that certain organic dye ions, thorium B cation, as
well as certain non-radioactive inorganic ions, may be concentrated on
an adsorbent which does not exhibit a charge opposite to that on the
ion. For example, thorium B is adsorbed by silver iodide, chromate,
oxalate, and phosphate even when positively charged by adsorption
of a slight excess of silver ion. Also, certain dye anions are adsorbed
on negatively charged silver halide and dye cations on positively
"Fajans: Naturwissenschaften, 9, 733 (1921); Z. Krist, 66, 321 (1928);
cf. Lange and Crane: Z. physik. Chem., A141, 225 (1929).
"Fajans: Z. Krist, 66, 321 (1928).
"Hahn: Ber., 69B, 2014 (1926); Naturwissenschaften, 14, 1196 (1926);
Hahn and Imre: Z. physik. Chem., AIM, 161 (1929).
w Hahn: Z. angew. Chem, 48, 871 (1930).
* Fajans and Erdey-Gruz: Z. physik. Chem., A158, 97 (1931).
106 COLLOIDAL SILVER IODIDE
charged silver halide. It should be recalled that a neutral surface is
seldom encountered except in the presence of some electrolyte; thus
we have seen that silver iodide forms an equivalent compound only
in the presence of a small excess of silver nitrate. In this connection
Fa jans and Erdey-Gruz have made an extended study of the effect
of the presence of various electrolytes on the adsorption of thorium
B by silver bromide, iodide, sulfide, iodate, chromate, oxalate, and
phosphate. The results show that strongly adsorbed cations cut down,
and strongly adsorbed anions increase, the adsorption of thorium B.
The adsorption on silver iodide, for example, is increased by the pres-
ence of anions in the order: I > CNS > Br > Cl, which is the same
as the order of adsorption of these ions on the silver halide. In the
presence of the strongly adsorbed hydrogen ion, the adsorption is less
than in neutral solution.
In the light of these and similar observations Fajans formulates
what may be termed the Paneth-Fajans-Hahn rules: (a) An ion is
strongly adsorbed on an equivalent compound of the salt type only
when it forms a difficultly soluble or weakly ionized compound with
the oppositely charged ion of the lattice, (b) The adsorption of a
cation is raised by adsorbed anions, that is, by charging the surface
negatively, and is lowered by adsorbed cations, that is, by charging
the surface positively, (c) The adsorption of anions is raised by
adsorbed cations and is lowered by adsorbed anions. The effects in
both (b) and (c) increase with increasing adsorption of the added
ions. These rules will probably hold in most cases of adsorption on
ion lattices provided one also takes into account the hydration, size,
and deformability of the adsorbed ions. There is nothing new about
rules (b) and (c). These facts were first recognized by Lachs and
Michaelis 22 and by Estrup 2S and have been worked out in detail for
the process of dyeing on fibers by Pelet-Jolivet, 24 by Bancroft, 25 and
by Briggs and Bull ; 28 and for the mordanting process on hydrous
oxide mordants by Weiser and Porter 27 (Vol. II, Chapter XVI).
The order of adsorption of cations by silver iodide was found by
Mukherjee and Kundu 28 to be: Ag > Al > Ba > Ca > K. With the
*'Z. Elektrochem., 17, 1 (1911).
2S Kolloid-Z., 11, 8 (1912).
a* "Die Theorie des Farbeprozesses," 94, 98, 119, 148 (1910).
J. Phys. Chem., 18, 1, 118, 385 (1914).
28 J. Phys. Chem., 26, 845 (1922)
*'J. Phys. Chem., 31, 1383, 1704, 1824 (1927); Weiser: 83, 1713 (1929).
M J. Indian Chem. Soc., 3, 335 (1926).
ADSORPTION BY SILVER IODIDE 107
exception of the common silver ion, the ion with the highest valence
is the most strongly adsorbed.
In this connection may be mentioned the observation of Frumkin
and Obrutschewa 29 that silver iodide exhibits a maximum adsorption
for caprylic alcohol at a silver ion concentration of a definite strength
corresponding to 0.16 volt as measured with a silver electrode against
a normal calomel electrode.
Carey Lea 80 claimed that freshly precipitated silver iodide adsorbs
iodine from an alcohol-water solution of the element, but Germann and
Traxler 81 found that this is not so. A thoroughly purified sample of
the salt does not decolorize a dilute iodine solution ; this results only
when the sample is not washed so as to remove adsorbed silver nitrate
which will react with iodine.
Adsorption of Dye Ions. The silver halides adsorb dyes both from
true and from colloidal solutions 82 and also adsorb organic colloids
such as casein, tannin, gums, and gelatin. 88 The adsorption of dyes is
of special interest and importance because of the use of dyes in the
sensitizing of the photographic plate (p. ISO) and the application to
adsorption indicators in analytical titrations. Since the principle of
titration by means of adsorption was worked out by Fajans in connec-
tion with silver bromide, the discussion of adsorption indicators will
be taken up in the next chapter (p. 132).
The adsorption of dye ions on the silver halides is, in general, in
accord with the Paneth-Fajans-Hahn rules. 21 Thus the silver salt
of the acid dye eosin is fifty times as soluble as the silver salt of
the acid dye erythrosin, and erythrosin is adsorbed by silver bromide
much more strongly than eosin. Moreover, the presence of electro-
lytes with a common cation and different anions cuts down the ad-
sorption of acid dyes in proportion to their adsorbability on the salt in
question. This is illustrated by the effect of potassium salts on the
adsorption of erythrosin by silver iodide as shown graphically in
Fig. 18. From the displacing power of the anions one would deduce
the order of adsorption to be : I > CNS > Br > Cl. This same order
would be deduced from the solubilities of the respective silver salts
except that thiocyanate is adsorbed more strongly than bromide al-
though silver thiocyanate is more soluble than silver bromide (p. 114).
"Biochem. Z., 182, 220 (1927).
so Am. J. Sci., (3) 88, 492 (1887).
ij. Am. Chem. Soc., 44, 460 (1922).
w Cf Luppo-Cramer: Kolloid-Z., 28, 90 (1921).
aaReinders: Z. physik. Chem., 77, 677 (1911).
108
COLLOIDAL SILVER IODIDE
The presence of silver nitrate increases greatly the adsorption of acid
dyes by the silver halides (p. 133).
Mechanism of the Adsorption. The adsorption phenomena above
mentioned were observed with precipitated and thoroughly washed,
hence with aged, silver iodide. The adsorption of a common ion by
such a precipitate is a true adsorption as distinct from an exchange
adsorption. Just as in the adsorption of thorium B on lead sulfatc,
Verwey and Kruyt 34 showed that iodide ion is not adsorbed over
the entire surface but only at certain active spots on the surface. In-
deed Kruyt 35 showed by endosmotic experiments that fused silver
iodide which has relatively few corners and edges cannot be charged
5 10 is ,20
Salt Concentration, Mol per Liter x 10
FIG. 18 Effect of potassium salts on the adsorption of erythrosm by silver
iodide.
positively even with 0.001 N AgNO 3 . Just as in the adsorption of
wool violet on lead sulfate, one may have exchange adsorption of the
foreign anion, say, followed by precipitation with a corresponding
amount of the lattice cation (p. 56).
In the adsorption of foreign ions, it is probable that an exchange
takes place between the foreign ion and the lattice ion or between the
foreign ion and the counter ion associated with another adsorbed ion.
In the adsorption of erythrosin anion on negative silver iodide, it is
probable that the dye anion enters into ionic exchange with iodide.
In the adsorption of erythrosin on silver iodide which is positively
charged by preferential adsorption of silver from silver nitrate, the
anion probably enters into exchange adsorption with the counter nitrate
Z. physik. Chem., A167, 149 (1933).
Z. Sowjetunion, 4, 295 (1933).
ADSORPTION BY SILVER IODIDE
109
ion; and in the adsorption of lead ion by silver iodide which is nega-
tively charged by preferential adsorption of iodide ion from hydriodic
acid, the lead ion enters into exchange adsorption with the counter
hydrogen ion. 86 As we shall see in the next chapter, the action of
adsorption indicators may be interpreted as exchange adsorption rather
than as direct adsorption.
Imre 37 studied the aging of silver halide precipitates, especially of
silver iodide, following the time course of the adsorption by means
55
Time, Hours
FIG 19. The time course of the adsorption of ( thorium B and actinium by
freshly precipitated silver iodide.
of the radioactive indicator method used in a study of the aging of
barium sulfate (p. 35). As radioactive indicators were used thorium
H, which forms insoluble lead iodide, and actinium and radium, each
of which forms an easily soluble iodide. Some typical observations are
shown graphically in Fig. 19. With thorium B which forms an in-
soluble compound, the adsorption value increases very slowly in spite
of the marked decrease in the specific surface with time ; on the other
hand, the adsorption value of actinium which forms a soluble iodide
falls off appreciably from the beginning. To account for this differ-
ence in behavior, it is assumed that the structure of the aging particles
opens up because of the crystallization tendency thus enabling an ex-
aeVerwey and Kruyt: Z. physik. Chem., A167, 312 (1933).
**Z. angew. Chem., 48, 875 (1930); Z. physik. Chem., A15S, 127 (1931);
cf. Kolthoff and Yutzy: J. Am. Chem. Soc., 59, 1215 (1937).
110 COLLOIDAL SILVER IODIDE
change reaction to take place in the immediate vicinity of the lattice.
This accounts for the increasing adsorption of thorium B with time
since the cations which form easily soluble halides remain in the outer
diffuse portion of the double layer. It would follow that the adsorp-
tion phenomenon on freshly formed, highly dispersed precipitates in-
volves two processes: the first process of momentary adsorption con-
sists in an exchange adsorption of ions that depends on the surface
charge and the valence of the adsorbed ion ; the second process con-
sists of a building up of molecules and resolution of the same at the
surface layer. In the latter process the velocity of the decrease of
surface and the insolubility of the adsorption compound are the im-
portant factors.
SILVER IODIDE SOLS
The Electrical Double Layer
Nature. The older concept of the double layer at the interface
solid-solution assumed it to consist of two "mono-ionic" layers of op-
posite charge touching each other at the boundary plane. 88 The mod-
ern theory due to Gouy S9 assumes that the total potential drop caused
by the double layer occurs in the outer liquid portion which consists
of a diffuse layer of ions. Stern 40 - 41 assumes that part of the ions
remain attached to the surface and part are free to move. He thus
takes into account the size and physical properties of the ions consti-
tuting the outer layer. The attached ions are held by electrostatic
and physical forces which differ for each ion, that is, each ion possesses
a specific adsorption potential. Extending Gouy's views, Smoluchow-
ski 42 assumes a continuous drop in potential in both the solid and
liquid phases. This was confirmed by calculations of Verwey 4S from
data obtained with a purified, well-aged silver iodide sol. The results
are represented diagrammatically in Fig. 20. Part of the potential
drop from E Q to E is in the liquid phase and part in the solid. The
part of the potential in the liquid phase f consists of (1) the potential
in the mobile portion of the outer layer, the -potential, and (2) the
38 Helmholtz : Pogg. Ann., 89, 211 (1853); cf. Perrin: J. chim. phys., 2,
601 (1904) ; 3, SO (1905).
89 J. Phys., (4) 9, 457 (1910); Ann. phys., (9) 7, 129 (1917); Chapman:
Phil. Mag., (6) 25, 475 (1913) ; Herzfeld: Physik. Z., 21, 28, 61 (1920) ; Debye
and Huckel: 24, 185, 305 (1923).
40 Z. Elektrochem., 80, 508 (1924).
"Cf, also, Janssen: Physik. Z. Sowjetunioa, 4, 322 (1933).
42 Graetz's "Handbuch der Elektrizitat und des Magnetismus," 2, 366 (1914).
"Rec. trav. chim., 68, 933 (1934).
THE ELECTRICAL DOUBLE LAYER
111
potential in the attached portion of the outer layer, ^ . Verwey
points out that the conditions for an aged silver iodide sol, represented
in Fig. 19, are relatively simple. The potential curve may be more
complicated, sometimes containing a maximum or a minimum. But
even if one disregards these complications, it is usually very difficult
to find a relation between the different parts of the total potential drop.
Such information would be particularly helpful since the stability of
lyophobic sols depends on the ^-potential which in turn is connected
with the potential in the other parts of the double layer.
Potential
Agl
E
Solution
FIG 20. Schematic curve for the potential perpendicular to the surface silver
iodide-solution, for a dialyzed silver iodide sol.
Formation. If one takes the relatively simple case of silver iodide
suspended in dilute hydriodic acid, a double layer is set up as a result
of the adsorption of iodide ion on the surface of the silver iodide. The
adsorbed iodide constitutes the inner portion of the double layer, and
the counter hydrogen ion the outer diffuse portion. For electrostatic
reasons, the negative charge like the equal positive charge in the solu-
tion is localized in the immediate neighborhood of the boundary layer
and the potential drop caused by the double layer is in the same
region. The potential difference between silver iodide and the solu-
tion is thus determined by a distribution of the common iodide ions
over both phases. Ions which both phases have in common, and which
are subjected to a distribution equilibrium to give the double layer, are
called potential determining ions, after Lange. 6 For a silver iodide
electrode at equilibrium, the potential E is given, in accord with the
classical work of Nernst and of van Laar, 44 by the equation:
RT
E - Eo In ci-
r
44 Van Laar: "Lehrbuch der theoretischen Elektrochemie" (1907).
112 COLLOIDAL SILVER IODIDE
Furthermore, if E is the total potential drop silver iodide-solution and
CQ the equilibrium concentration of iodide ions at the point of zero
charge, then
E _ Eo = ^i n *L_ 0.058 log ^
F ci- ci-
For a constant capacity of the double layer the amount of iodide ad-
sorbed x by 1 g of Agl is proportional to E EQ, hence
x = ai + 02 log ci-
in which a { and o 2 arc fixed by the location of the zero point of charge
and by the adsorption capacity which is given by the equation :
dx ,
or dx =
dlog CI-
AS already pointed out, the adsorption of potential-determining ions
such as iodide and silver on silver iodide is represented by the equa-
tion dx a dlog c instead of by the Freundlich equation dlog c =
a dlog c where x is the amount adsorbed per gram of adsorbent and c
is the equilibrium concentration. Verwey 45 refers to this taking up
of potential-determining electrolytes, with the formation of a double
layer, as a process of assimilation rather than as one of adsorption.
Gibbs 48 points out that an electrolyte lowers the surface tension when
it is positively adsorbed and raises the surface tension when it is nega-
tively adsorbed. Since the taking up of potential-determining electro-
lytes does not satisfy Gibbs' criterion, Verwey contends that it is
not an adsorption phenomenon. The difference is that one ion of the
potential-determining electrolyte goes from the solution onto the solid
phase causing a surface charge which theoretically could be effected
from within this phase by the application of an external electric cur-
rent. The surface tension is therefore not connected directly with the
accumulation of electrolyte in the boundary layer but only indirectly
as a result of the charge of the double layer thus formed ; hence the
process is not an adsorption. Similarly, it is argued that the exchange
of lattice and counter ions should not be called adsorption. True
adsorption of an electrolyte, according to Verwey, consists in the ac-
cumulation as a whole of an electrolyte in that part of the solution
which is nearest the surface; it is therefore not accompanied by a
change of the total boundary potential drop.
e Chem. Rev., 16, 363 (1935) ; Wis-Natuurkund. Tijdschrt., 7, 89 (1934).
46 "Thermodynamics," Longmans-Green (1906).
PREPARATION 113
The author prefers not to be as logical as Verwey and will con-
tinue to refer to the concentration of potential-determining ions at
surfaces and to the exchange of lattice and counter ions as adsorption
phenomena. True adsorption of indifferent electrolytes at ion lattices
in the sense of Verwey appears to be encountered rarely if at all. It
will be recalled that de Brouckere claimed to get equivalent adsorption
of cations and anions at a barium sulfate surface, but this is denied by
Kolthoff (p. 32). The taking up of ions from indifferent electrolytes
appears to be represented by the Freundlich equation dlog x =
a dlog c 47 rather than by the equation dx = a dlog c which holds for
the adsorption of potential-determining ions.
The essential difference between the building up of a double layer
by adsorption of potential-determining ions and the formation of a
double layer by adsorption of electrolytes is that, in the first, the
potential-determining ions are assimilated into the surface in accord
with the equation dx = a dlog c, whereas, in the latter, the ions are
merely oriented in the solution nearest the surface in accord with the
equation dlog c = a dlog c. It is assumed by Verwey that the build-
ing up of the primary double layer by assimilation of potential-deter-
mining electrolyte is required for the primary stability of sols. It fol-
lows from this that the adsorption of potential-determining ions only,
will peptize a substance; the adsorption of indifferent ions will not
cause the primary stability but will influence the stability of a sol once
formed.
Preparation
Peptization by Common Ions. From the preceding section it
would follow that electrolytes containing the common silver and iodide
ions would be the best peptizing agents for silver iodide sols. This
classical method of Lottermoser 48 consists essentially in adding a suit-
able excess of one of the reacting salts, silver nitrate and alkali halide,
to a suspension of the freshly precipitated silver halide. For example, 49
if 0.1 N AgNO 3 is allowed to drop from a buret into 500 cc of 0.002
N KI stirred by a motor stirrer, the first drop gives a green-yellow, per-
fectly transparent sol which becomes gradually more and more opales-
Janssen: Physik. Z. Sowjetunion, 4, 322 (1933) ; Kruyt and van der Willi-
gen- Kolloid-Z., 45, 307 (1928).
J. prakt. Chem., (2) 66, 241 (1897); 68, 341 (1903); 72, 39 (1905); 73,
374 (1906); Z. physik. Chem., 60, 451 (1907); 70, 239 (1910).
49 Lottermoser, Seifert, and Forstmann: Kolloid-Z. (Zsigmondy Festschrift),
36, 230 (1925).
114 COLLOIDAL SILVER IODIDE
cent with further additions of iodide until, at approximately the
equivalence point between iodide and silver, the opalescence increases
rapidly, the solution becomes cloudy, and an additional drop causes
flocculation of the silver iodide. If the process is reversed and 0.1
N KI is allowed to drop slowly into 004 N AgNO 3 , the phenomena
are similar except that the clouding takes place somewhat more
rapidly. Similar relations obtain not only with the other silver halides
but also with the closely related silver cyanide and silver thiocyanate.
Since the negative halide sols are readily coagulated by multivalent
cations and the positive sols by multivalent anions, the stability of sols
is much less if the interacting solutions contain a multivalent ion oppo-
site in charge to the stabilizing ion. Thus a negative sol can be
formed by adding 0.05 N AgNO 3 to 0.05 N KT solution, whereas, if
barium iodide or cadmium iodide is used, the solutions must be at
least as dilute as 0.01 N or sol formation does not occur. 80 With dif-
ferent univalent cations, the nature of the cation has little effect ; but
at higher concentrations, the stabilizing action diminishes in the order :
Li, Na, K, NH 4 ."
The order of stability of the negative sols, 52 stabilized by preferen-
tial adsorption of the respective anions from potassium salt in excess,
is as follows : Agl > AgBr > AgCl > AgCNS. A similar order ob-
tains for the positive sols prepared with silver nitrate in excess. This
order would be deduced from the solubility, indicating that the ten-
dency to precipitate is determined in large measure by the velocity
of growth of crystals rather than to agglomeration of particles. Com-
paring the positive and negative sols, the negative silver iodide is more
stable than the positive at all electrolyte concentrations whereas the
positive silver chloride sol appears to be more stable than the negative.
The greater stability of negative silver iodide sols is evidenced by the
following observations: (1) the positive sol is more cloudy and settles
out more rapidly than the negative in the presence of a like excess of
the stabilizing ions ; (2) the positive sol cannot be purified by dialysis
without agglomeration, whereas highly pure negative sols can be made
in this way; and (3) a freshly precipitated and washed silver iodide
gel is peptized in part by shaking with dilute potassium iodide solu-
tion, whereas silver nitrate does not peptize such a gel. The marked
stability of negative silver iodide was shown by Kruyt and van der
*Lottermoser: Kolloid-Z., 2, suppl. I, p. iv (1907).
"Basinski: Roczniki Chem., 14, 1017 (1934).
62 Cf. Basinski: Kolloid-Beihefte, 36, 257 (1932).
PREPARATION 115
Willigen 6S to be due to the asymmetric position of the zero point of
the electrokinetic potential.
To prepare pure concentrated sols of silver iodide, Kruyt and
Verwey 5 * mixed pure silver nitrate with a 10% excess of potassium
iodide, ammonium iodide, or hydriodic acid in such concentrations that
the resulting sol contained 80 millimols Agl/l, or less. This was
dialyzed at once in an electrodialyzer until quite pure and then con-
centrated by the process of "electrodecantation" M which consists in
removing the clear solution from around the cathode membrane, the
negatively charged particles having concentrated in the region of the
anode membrane. Whether one starts with an iodide or hydriodic
acid, the highly purified sol contains only hydriodic acid as stabilizing
electrolyte. In this way stable, highly purified sols containing as high
as 160 g Agl/kg of sol were prepared.
Sols containing 0.05 g Agl/l were made by electrolysis of 0.001
N KI solutions with a silver anode. 56
Peptization by Ions Other Than Common Ions. Positively charged
silver iodide sols are obtained only by adsorption of the common silver
ion. Negatively charged sols " 57 * 8 are formed, however, by the ad-
dition of a suitable excess of the following alkali salts : iodide, bromide,
chloride, cyanide, thiocyanate, phosphate, ferricyanide, and ferro-
cyanide ; but not by the following : fluoride, nitrate, chlorate, permanga-
nate, formate, carbonate, sulfate, chromate, dichromate, and hydroxide.
Since many of the amons in the second list form insoluble silver salts
and so should be absorbed, Kruyt concluded that adsorption of an
anion is insufficient in itself to form a negative sol; the important
thing is that the adsorbed anion should give an isomorphous silver salt
in order that the primary double layer which is essential for peptiza-
tion and stability may form. It appears, however, that the ion need
not be exactly isomorphous to effect peptization. For example, alkali
chloride and bromide which give cubic silver hahdes peptize silver
iodide which is hexagonal when negative. 4 The important thing is
that the adsorbed ion should fit the lattice sufficiently well that it will
be taken up by the lattice to a certain extent and so will be distributed
in both phases, in other words, will be a potential-determining ion.
Whenever ions fit the lattice to a certain extent, it is probable that an
S3 Kruyt and van der Willigen: Z. physik Chem., A139, S3 (1928).
"Z. physik. Chem., A167, 137, 149, 312 (1933).
"Pauli: Naturwissenschaften, 20, 551 (1932).
"Peskov and Saprometov: Kolloid-Z., 69, 181 (1934).
"Kruyt and Cysouw: Z. physik. Chem., A172, ,49 (1935).
"Basinski: Kolloid-Beihefte, 86, 258 (1932).
116 COLLOIDAL SILVER IODIDE
exchange of lattice ions takes place between the adsorbed ion and
iodide ion. This liberates iodide ion from the lattice which may be-
come the predominating potential-determining ion even with iso-
morphous ions. The failure of Kruyt and Cysouw 5r to detect iodide
in the ultrafiltrate from a sol peptized by excess alkali chloride does
not mean necessarily that adsorbed iodide ions are not the potential-
determining ions. The reversal of charge of a positive silver iodide
sol by sodium vanadate 5B is probably caused in part by exchange of
vanadate for iodide ions. A fairly large excess of phosphate displaces
sufficient iodide ion to peptize the sol. The dispersed particles consist
of a mixture of silver iodide with an amount of silver phosphate
equivalent to the amount of iodide set free. The addition of a small
amount of sodium hydrosulfide to an equivalent mixture of silver
nitrate and potassium iodide gives a silver iodide sol and a precipitate
of silver sulfide; potassium iodide equivalent to the silver sulfide is
the stabilizing electrolyte. Sodium para iodobenzoate behaves like
sodium phosphate. A precipitated and thoroughly washed silver iodide
is peptized by potassium iodide and cyanide but not by potassium
bromide and chloride ; the peptizing action of the cyanide results from
a reaction with silver iodide giving potassium iodide, the real peptizing
agent.
Protected Silver Iodide Sols. Because of the instability of Lotter-
moser's silver halide sols, Lobry de Bruyn 60 used gelatin and Paal
and Voss 61 used sodium lysalbinate as protecting colloids. Paal's
preparations are typical hydrophilic colloids, since they are not pre-
cipitated even by 5 volumes of saturated sodium chloride or an equal
amount of 10% Na 2 HPO 4 . Gutbier 62 prepared fairly stable sols by
treating with the respective halogen a silver sol formed by reduction
of silver nitrate with hydrazine hydrate in the presence of gum arabic.
Von Weimarn 68 used caoutchouc as protecting colloid in preparing sil-
ver halide sols in aromatic hydrocarbons.
Charge and Stability
Thanks to the comprehensive investigations of Verwey and
Kruyt, 54 ' e4 considerable information is available concerning silver
iodide sols, especially the aged and dialyzed sols.
Lottermoser and May: Kolloid-Z., 68, 168 (1932).
Rec. trav chim., 19, 251 (1900).
i Ber., 87, 3862 (1904)
* Kolloid-Z., 4, 308 (1909).
sj. Russ. Phys.-Chem.-Soc., 48, 1046 (1916).
* Verwey: Rec. trav. chim., 88, 933 (1934); cf., also, Gorokhovskii and
Protass: Z. physik. Chem., A174, 122 (1935).
CHARGE AND STABILITY 117
Sols prepared by Lottermoser's method followed by dialysis and
concentration by electrodecantation have a very large electrochemical
(colloid) equivalent. For a sol with particles having an edge ap-
proximately 40 m/4 in length, the particle charge was calculated to be
900 electrons and the "free" charge (from Pauli's conductivity
method) 330 electrons per particle as compared with a free charge of
28,000-58,000 electrons per particle for gold sols (Vol. I, p 75). By
boiling a pure sol with dilute hydriodic acid followed by dialysis, a
sol was obtained having a particle charge of 440 electrons and a free
charge of 220 electrons. The colloid equivalent was calculated to be
1500 on the basis of the total charge, and 2700 on the basis of the free
charge. It contained 16.7 g Agl/kg of sol; the concentration of
hydrogen ion was 0.05 milliequivalent and of the iodide ion 0.00021
milliequivalent per liter ; and yet it was quite stable, as we shall see.
From such observations, it follows necessarily that the double layer
is not distributed uniformly over the surface but at certain active
spots, corners, and edges The number of such active spots and edges
is large in a freshly formed precipitate; but, upon aging, the crystals
arc perfected so that there are fewer places that can take up iodide
ions. 68 With excessive aging, the double layer is concentrated in a
relatively few spots; hence the colloid equivalent is quite large.
Granting that the primary stability of a sol results from adsorption
of potential-determining ions, it is determined by the concentration
of these ions and the location of the zero point of the charge. Verwey
calculated, for the highly aged sol above, the total potential drop due
to free charges (E EQ) from the equation
RT . CQ
E-Eo = -ln-
where cr~ is the iodide ion concentration of the sol ( = 0.00021 milli-
equivalent) and CQ is the iodide ion concentration where E= EQ. The
latter value is difficult to determine accurately but it is less than
5 X 10" 10 ; hence log CQ/CI is approximately 3 and E E =
0.058 X ~3 or approximately -0.2 volt. Referring to Fig. 20, it will
be seen that the f -potential is only about one-third of the total potential
drop, the remainder occurring in the silver iodide phase and in the
layer of attached ions. The critical value of E EQ required for the
primary stability will thus be almost three times the critical {-potential.
Since the critical value of is frequently around 40 millivolts the
critical value of E - E for a dialyzed sol of silver iodide would be
Verwey: Proc. Acad. Sci. Amsterdam, 86, 225 (1933).
118 COLLOIDAL SILVER IODIDE
expected to be around -120 millivolts. Actually about -2 X 0.058
volt was required for the stability. In another highly dialyzed sol in
which rr- = 10~ 7 , the value of E - E was around -0.150 volt. But
since the electrolyte concentration is very small, a considerable portion
of this potential drop is the -potential to which the sol owes its stabil-
ity ; hence one may have a stable sol even when the double layer charge
is relatively low, provided the electrokinetic potential is above a critical
value.
A dialyzed negative silver iodide sol (#Ag, about 9) is stable but
flocculates when its pAg is raised to about 8 by titrating with very
dilute silver nitrate solution. As already noted, the zero point of
charge is about />Ag = 6. A positive silver iodide sol coagulates when
the concentration of free silver ions is lowered by dialysis to about
^Ag = 4. ee On diluting a positively charged sol with water,
the ^-potential decreases as a result of dilution of the potential-deter-
mining silver ion, and finally becomes negative. 58 ' 67 If the sol is di-
luted with ultrafiltrate from the sol, the ^-potential decreases to the
coagulation point but does not change sign. 88 Gorochovskii 69 found
the isoelectric point of silver iodide to depend greatly on the iodide
concentration of the sols; but his results are not conclusive since he
failed to correct for adsorbed silver or iodide ions on the surface of
the silver iodide.
Coagulation by Electrolytes
Comprehensive investigations have been made by Verwey and
Kruyt of the phenomena which take place during the coagulation of
dialyzed and aged negative silver iodide sols with hydriodic acid the
stabilizing electrolyte. The experiments were carried out with various
sols and the coagulating power deduced from the tendency to replace
hydrogen from the outer portion of the double layer is : Ce > UO 2 &
Pb > Ba > H > Cs > K. As usual, the ion with the higher valence
has the higher coagulating and displacing power. Similarly Basinski 70
found the order of coagulating power of cations to be : Th > Zr ;
Al > Ce > Fe+ + + ; Pb > Ba > Cu > Sn > Ca > Ni > Zn >
Co > Mn > Mg > Cd; Ag > H > NH 4 > K > Na > Li. Marked
fiC/. Schneller: Kolloid-Z., 71, 180 (1935).
C/., also, Lottermoser and Riedel: Kolloid-Z., 51, 30 (1930).
Basinski: Roczniki Chem., 18, 117 (1933).
69 J. Phys. Chem, 89, 465 (1935); Gorokhovskii and Protass: Z physik.
Chem., A174, 122 (1935) ; J. Phys. Chem. (U. S. S. R.), 7, 354 (1936).
"Roczniki Chem., 15, 430 (1935).
COAGULATION BY ELECTROLYTES
119
ion antagonism (pp. 216, 322) was observed in the precipitations with
mixtures of univalent with bi-, tri-, or tetravalent cations.
The above results were confirmed and extended by Kruyt and
van Gils' 71 observations of the effect of electrolytes of varying valence
on the mobility of the particles in a silver iodide sol. In Fig. 21 the
mobility in microns per second at 1 volt/cm is plotted against the
negative logarithm of the electrolyte concentration in mols per liter.
Log Concentration
FIG. 21. Effect of varying concentrations of electrolyte on the mobility of the
particles in a silver iodide sol.
The broken portion of the curves marks the zone of instability. The
coagulating power of the several cations, as measured by the lowering
of the mobility and hence of the ^-potential to the point of instability,
is in the order: Th > [Pt(Aein) 3 ] > Al > hexol > Ba > H > K.
Hexol (a hexavalent cobalt amine nitrate) lowers the mobility to zero
at a lower concentration than aluminum nitrate. For all salts with
multivalent cations, the coagulation of the negative particles takes place
at a mobility of approximately l/n/sec, ; but for salts with univalent
" Kolloid-Z., 78, 32 (1937) ; cf. Weiscr, Milligan, and Coppoc : J. Phys. Chem.,
42 (1938).
120 COLLOIDAL SILVER IODIDE
cations the critical mobility is much higher (cf. Vol. I, pp. 73, 181).
Potassium bromide containing the potential-determining bromide ion
increases the mobility at all concentrations employed. At sufficiently
high concentrations, thorium nitrate and hexol not only reverse the
sign of the charge on the particles but even give stable positive sols.
The point of maximum flocculation for thorium nitrate corresponds
to zero mobility hence to zero {-potential, but for hexol it occurs well
above the zero point. The explanation for this is not immediately
obvious. In this connection, some observations of Coppoc 71 in the
author's laboratory indicate that the zone of instability is quite wide
for the concentrated polydisperse sols of silver iodide prepared by the
method of Kruyt and Verwey. For example, more than 95% of a
typical sol was precipitated in 24 hours by 10 milliequivalents per liter
of BaQ 2 , but 20 milliequivalents per liter were required for complete
precipitation in this time.
Titration of Sols. In the coagulation of sols by electrolytes, it has
been observed that the ion having a charge opposite to that on the sol
is adsorbed in exchange with the counter ions of the double layer on
the particles. The adsorption and displacement have been followed
potentiometrically in the author's laboratory with a number of hydrous
oxide and salt sols during the stepwise addition of electrolytes to the
sols. The process has been referred to as "titration of sols." In sols
examined by the author (cf. pp. 208, 319) only a portion of the counter
ions were measurable potentiometrically in the original sol; the re-
mainder were displaced by adding electrolytes and were then measur-
able. With the aged silver iodide sols of Verwey and Kruyt, the
titration procedure was more complicated and less exact. In the first
place, the particle charge was much smaller in the iodide sols and the
concentration of counter hydrogen ions was correspondingly smaller.
Moreover, practically all the counter ions were measurable potentio-
metrically in the original sol ; hence the amount of hydrogen displaced
on adding electrolytes was inappreciable. The measurement of hydro-
gen ion exchanged for the added cation had to be determined in the
ultrafiltrate after the addition of the electrolyte. This introduces
errors which may be more or less serious depending on the conditions
(Vol. II, p. 58).
The procedure was as follows: 15- to 25-cc portions of sol were
weighed in a glass-stoppered flask and varying amounts of electrolyte
added, followed by reweighing. After 24 hours the uncoagulated
samples were ultrafiltered and the coagulated ones were centrifuged
and the supernatant solution decanted. The exchanged hydrogen ion
COAGULATION BY ELECTROLYTES
121
was determined colorimetrically or with the glass electrode. The ad-
sorption of uranium and cerium was estimated directly by colorimetric
procedures and that of lead with thorium B as radioactive indicator
(p. SO). The exchanged hydrogen ion may serve as an indirect
measure of the amount of cation taken up. The results of some ob-
servations with a number of different sols are shown graphically in
Figs. 22 and 23. From the form of the curves it appears that in cer-
tain cases the coagulation point (indicated by an arrow) occurs near
the point of maximum adsorption, whereas in others the maximum
adsorption is reached well below the coagulation concentration. The
latter behavior has not been encountered with sols investigated in the
Soli. 158gAg I/Kg Sol
SolII, 75gAgI/KgSol
Sol III, 20gAg I/Kg Sol
1234
Electrolyte Added, Millieq /Kg Sol
FIG 22. Adsorption of cations by silver iodide sol.
author's laboratory and calls for special consideration. The barium
adsorption estimated indirectly from />H measurements on Sol 3 can
be disregarded since the change in hydrogen ion concentration is too
small to measure accurately. The behavior of Sol 2 with UO 2 + +
cannot be disposed of in this way, but the possibility of the presence
of a stabilizing impurity from the membrane used in the dialysis has
not been excluded. In any event the very different behavior of UO 2 "*" +
toward Sol 1 and Sol 2 in this respect requires explanation. If con-
tamination of Sol 2 by the membrane is ruled out, another possibility
is that the size distribution of the particles in Sol 2 is quite different
from that in Sol 1. If most of Sol 2 is coagulated at low concentra-
tions (cf. p. 120) but the last trace only at relatively high concentrations
(taken as the precipitation value), and if Sol 1 does not exhibit such
marked stepwise coagulation, the difference in the behavior of the two
sols toward UC>2 ++ ion is accounted for. 71
122
COLLOIDAL SILVER IODIDE
Of special importance are the precise observations of the adsorp-
tion of lead ion by a dialyzed sol containing 159 g Agl/kg of sol and
the same diluted eleven times (Fig. 23). The dotted line gives the
curves for the diluted sol, again on a scale eleven times larger. It is
apparent that both curves have the same shape and nearly coincide.
The small difference is probably caused: (1) by the giving up of
iodide ion from the double layer by dilution, which lowers somewhat
the charge on the diluted sol ; and (2) by the fact that the dilution is
somewhat larger than elevenfold because of the volume occupied by
2 4
Beginning Concentration, Ci
FIG 23 Adsorption of lead ion by silver iodide sols of two different
concentrations.
the particles. These results show that the relative adsorption is inde-
pendent of the dilution or, what amounts to the same thing, to the
mass of the adsorbent. In this case x/m is not a function of c as in
the Freundlich isotherm but is a function of c/m, that is, x/m =
f(c/m). This same equilibrium applies to the base exchange in clays
and permutites (p. 396).
For the concentrated sol (Fig. 23) the coagulation point comes just
a little below the point of maximum adsorption. The corresponding
point of the curve for the diluted sol represents a concentration of
lead eleven times smaller, hence well below the zone of coagulation.
The line ab parallel to the straight line x y connects points of equal
concentration of free lead ion. It must be borne in mind that the two
sols are alike only in their primary particle size. The secondary ag-
gregates of the original sol are peptized to a greater or lesser extent
by dilution, the adsorption equilibria are changed, and the stability
toward electrolytes is modified.
COAGULATION BY ELECTROLYTES
123
Mechanism of the Coagulation by Electrolytes. To effect the
coagulation of colloidal particles, the ^-potential must be lowered below
a critical value which allows agglomeration of the particles into aggre-
gates that settle out. This lowering of the {[-potential is brought about
by a contraction of the double layer by adsorption or otherwise.
The adsorption mechanism may be represented diagrammatically in
Fig. 24. At some active edge or spot a double layer is set up with the
-TOW**"
H*
H +
H+
K*
AgAgAg-I- j
AgAgAg-I-
-AgAgAgAgA
I I I I I
H*
H*
!H*
_ _ : ii
Mill
H*
iill!
I-AgAgAgAgAg-
+++ 1 I HI
H*
"
H*
H*
H*
H*
H*
C.
H*
B
FIG. 24. Diagrammatic representation of the constitution of a portion of ^a
particle of silver iodide sol (A) before and (B} after the addition of eerie
nitrate.
potential-determining iodide ions constituting the inner layer and hy-
drogen ions the diffuse outer layer. In the sol under consideration
most of the hydrogen ions are measurable potentiometrically in the
original sol, but a few shown within the dotted line in Fig. 24a are
held so strongly by the inner layer that they do not affect the hydrogen
electrode. If this particle were ultrafiltered, none of the counter ions
would leave the particle entirely and enter the filtrate. On adding an
electrolyte such as eerie nitrate, the strongly adsorbed cerium ions dis-
124 COLLOIDAL SILVER IODIDE
place the hydrogen ion from the innermost portion of the double layer
and take up a position closer to the inner layer than the hydrogen ions,
Fig. 24b. This results in an increase in the hydrogen ion concentration
of the sol and, at the same time, in a reduction of the ^-potential by
thinning or contracting the double layer. The displaced hydrogen
which causes a decrease in the ^H value of the sol is not equivalent to
the cerium adsorbed, since most of the hydrogen corresponding to the
cerium taken up is measurable in the original sol. On the other hand,
if the sol is ultrafiltered at some point below the coagulation value,
the increase in the hydrogen ion concentration over that in the ultra-
filtrate from the original sol corresponds with the cerium ion adsorbed.
The cerium ion has therefore entered into exchange adsorption with
the counter hydrogen ion of the original sol ; the {-potential is lowered
by contraction of the double layer as a result of the exchange adsorp-
tion or, as Stern 40 would put it, by an increase in the charge of the
attached portion of the double layer at the expense of the unattached
portion.
Verwey 7a objects to the proposed adsorption mechanism of poten-
tial reduction on two grounds : first, he prefers not to refer to exchange
adsorption as adsorption; and second, and more important, in certain
cases with silver iodide sol, the maximum adsorption, which proved
to be about equivalent to the total hydrogen in the double layer, was
reached well below the precipitation value, and in other cases, floccula-
tion occurred before sufficient electrolyte was added to reach the maxi-
mum adsorption. To account for the maximum adsorption below the
precipitation value Verwey accepts in principle Muller's 73 theory of
coagulation and concludes that the lowering of the -potential is
"merely due to 'compression 1 of the diffuse outer layer and a subse-
quent increase in the capacity of this part of the double layer." The
essential difference between the point of view of Verwey and that of
the author appears to be that the contraction of the double layer or
compression of the outer layer is believed by the present author to
result from adsorption of precipitating ions whereas Verwey con-
siders that adsorption is neither a necessary nor a sufficient cause of
the potential reduction at the surface of the particles. It so happens
that in all cases investigated by the author (pp. 207, 318, also Vol. II,
pp. 76, 117, 142) adsorption plays the major role in bringing about the
contraction of the double layer and the consequent lowering of the
{-potential. It remains to be seen whether the behavior of the sols
Chem. Rev., 16, 363 (1935).
"Kolloid-Beihefte, 26, 257 (1928).
COAGULATION BY ELECTROLYTES 125
investigated in the author's laboratory is special and that of silver
iodide typifies the general behavior or whether the reverse is true.
Certainly the older view of Freundlich that coagulation is the result of
neutralization of the charge on the particles by equivalent adsorption
is no longer tenable. The important thing is the lowering of the -po-
tential since the charge in the mobile portion of the double layer may
be altered without lowering the {-potential." The lowering is accom-
plished by a contraction of the double layer resulting in whole or in
part by the precipitating ions being drawn closer (adsorbed) to the
attached inner layer. In general, the contraction is greater for ions of
varying valence in the order : trivalent > bivalent > univalent which
is the same as the usual order of adsorption and displacing power of
the ions (p. 205). This means that less of a trivalent ion needs to be
adsorbed than of a bivalent ion to lower the -potential to the coagu-
lation point. (Cf. Vol. II, p. 77.)
Velocity of Coagulation. Jablczynski TB studied the velocity of
coagulation of silver halide sols both in the presence and the absence
of protecting colloids and found the process to take place in accord
with Smoluchowski's theory of rapid coagulation (Vol. I, p. 92).
This furnishes no evidence as to whether the precipitation is due pri-
marily to aggregation of primary particles or to crystallization, since
both processes follow the same law for the determination of number
of particles. 2 Schneller 66 showed that, with low concentrations of
stabilizing electrolyte, the coagulation is due chiefly to an agglomera-
tion of sol particles and the velocity decreases with progressive coagu-
lation, whereas, with increasing concentrations, the process of recrys-
tallization becomes more and more pronounced and the velocity attains
an almost constant value.
In the slow agglomeration process (Vol. I, p. 94) in the presence
of a small amount of coagulating electrolyte, the mobility of the par-
ticles increases somewhat. Kruyt and de Haan 7fl explain this by as-
suming that the {-potential is not lowered uniformly over the surface
of the particles. Under these conditions, coalescence takes place at
the point of lowest potential, giving agglomerates with a high potential
on the outside which may have a higher mobility than the original
particles.
The role of silver iodide in photographic emulsions will be con-
sidered in Chapter VIII.
"C/. Bull and Gortner: J. Phys. Chem,, 86, 309 (1931).
"Bull. soc. chim., (4) 86, 1277, 1286 (1924); 89, 1322 (1926); 48, 159
(1928) ; cf. Fromherz: Z. physik. Chem., Bl, 324 (1928).
"Kolloid-Z, 61, 61 (1930).
CHAPTER VII
THE COLLOIDAL HALIDES OF SILVER, LEAD, AND MERCURY
COLLOIDAL SILVER BROMIDE AND CHLORIDE
The general methods of formation and characteristics of gels and
sols of silver bromide and chloride are so similar in essential respects
to those of silver iodide that the former will not be given the detailed
consideration afforded the latter in the preceding chapter. Because of
increasing solubility in the order : Agl < AgBr < AgCl, the colloidal
behavior and stability decrease in the same order; hence silver bro-
mide and chloride have been used much less than silver iodide in the
study of colloidal phenomena. In this section it is found convenient
to consider the bromide and chloride together. Some reference will
also be made to the thallous halides.
. THE PRECIPITATED SALTS
In the absence of electrolytes that exert an appreciable solvent
action the formation of silver halide suspensions takes place in two
stages: (1) an initial very rapid formation of primary colloidal par-
ticles, and (2) the slower coagulation of the primary particles to give
fairly highly dispersed flocculent precipitates. Grain growth by Ost-
wald ripening and by coalescence of crystals l takes place as a sec-
ondary process. Greene and Frizzell 2 have studied the two stages
in the precipitation of silver chloride 8 by means of a specially designed
photronic nephelometer which enabled them to measure the opalescence
of the salt suspension from the moment precipitation begins. Visible
particles are formed in the initial stage in a few hundred ths of a
second, and this stage is virtually complete in 2 seconds or less. With
increasing concentration of precipitant, the initial opalescence increases
1 Cf. Sheppard and Lambert: Colloid Symposium Monograph, 4, 281 (1926) ;
6, 265 (1928) ; Kolthoff and Yutzy: J. Am. Chem. Soc., 68, 1215, 1634 (1937).
2J. Am Chem. Soc., 68, 516 (1936); Greene: 66, 1269 (1934).
'Tczak: Bull. soc. chim. roy. Yougoslav, 4, 137 (1933); Kober: Ind. Eng.
Chem., 10, 556 (1918).
126
ADSORPTION OF INORGANIC IONS 127
to a maximum and then falls off rapidly. The maximum is reached
with a lower concentration of silver nitrate in excess than of hydro-
chloric acid in excess. Variations in the rate of mixing and the rate
of stirring are without measurable influence. The change in opales-
cence with concentration of reactants is caused by variation in par-
ticle size as the concentration of the reagent in excess is increased,
the size of the first formed primary particles diminishing in accord
with von Weimarn's second law (p. 15). This conclusion is sup-
ported by observations with a centrifuge and an ultramicroscope and
by measurements of the color of the suspensions and the coagulating
effect of nitric acid on them.
The lattices of the fresh precipitates of silver chloride and bromide,
like those of silver iodide, have a marked adsorption capacity for many
ions, especially dye ions. This is of importance particularly in the
photographic process and in the quantitative estimation of the haliclcs
using what have been termed adsorption indicators.
Adsorption of Inorganic Ions
Since silver bromide and chloride age rather rapidly, the amount
of adsorption per gram of adsorbent is usually rather small. To make
accurate estimations of the order of adsorption of common inorganic
ions under such conditions, one may use a radioactive indicator such
as thorium B and determine the effect of various electrolytes on the
adsorption of thorium B. This has been done on silver halide ad-
sorbents, especially silver bromide, by Fajans 4 and Hahn. 5 More re-
cently King and Greene a have made a systematic investigation of the
effect of alkali and alkaline-earth bromides on the adsorption of tho-
rium B by silver bromide. The salt is negatively charged in contact
with bromide ions and will, therefore, adsorb thorium B cations. The
amount of this adsorption will decrease in proportion to the nature
and amount of other adsorbable cations in the solution. Some results
given graphically in Fig. 25 show that the several ions cut down the
adsorption of thorium B and are therefore adsorbed in the order:
Cs > Rb > K > Na > Li, at all concentrations measured. Since all
factors including the valence of the cation were held constant except
the cation present in the system at equilibrium, there must be some
connection between the order of adsorption and some properties of
4 Fajans and Erdey-Grflz : Z. physik. Chem., A158, 97 (1931).
5 Hahn and Itnrc: Z, physik. Chem., AIM, 161 (1929); Imre: AU6, 41
(1930); Hahn: Z. angew. Chem., 48, 871 (1930).
J. Phys. Chem, 87, 1047 (1933) ; King and Romer: 87, 663 (1933).
128 COLLOIDAL HALIDES OF SILVER, LEAD, AND MERCURY
the ions. From Table XI, which gives some properties of the several
cations, it appears that the most readily adsorbed ion (1) forms the
TABLE XI
SOME PROPERTIES OF ALKALI IONS
Adsorption of
Heats of
Solubilities
Ionic
Alkali
ThB in % at
hydration
of bromide
size
ion
8 millimols/1 of
cal./g
at 20 C
A
alkali bromide
mols/1000 g H 2 O
Li
73 3
120
20 9
70
Na
70 8
92
8 14
1.00
K
61
72
5 69
1 33
Rb
53 8
68
6 95
1.52
Cs
38 7
62
5 78
1.70
most insoluble compound with the opposite ion of the crystal lattice,
(2) is the least hydrated, and (3) is the most easily deformable or
polarized on forming a compound. These three factors are considered
8 12 16
Concentration of MBr, Millimols per Liter
FIG. 25. Effect of alkali bromides on the adsorption of thorium B by silver
bromide.
ADSORPTION OF INORGANIC IONS
129
by Fajans to be of prime importance in determining the adsorption of
similar ions.
Experiments similar to the above were carried out by King and
Pine 7 to show the effect of various anions and cations on the adsorp-
tion of thorium B by thallium bromide and iodide. Contrary to Hahn
and Imre's 8 observations, they found that thallium bromide and iodide
like silver iodide are always negatively charged in contact with their
saturated solutions, and adsorb thorium B. The "neutral powders"
obtained by Hahn and Imre must have been precipitated with excess
thallium ion and not thoroughly washed. Since the adsorption of vari-
ous anions will increase the adsorption of thorium B in proportion to
their own adsorbability, the following order of adsorption of anions
on the thallium salts was obtained : I > CrO 4 > CNS > Br > PO 4 >
Cl > C 2 O4. This order would be deduced from the solubility of the
thallium salts as given in Table XII with the exception of thiocyanate
TABLE XII
ADSORPTION OF ANIONS BY THALLOUS IODIDE
Ion
milliequivalcnts/1
Increase in adsorption
of ThB in %
Solubility of the salt
millimols/1
!-
58
18
CrO 4
51
57
CNS-
2
12
Br-
1 9
1 7
P04
7
ci-
-5
13
CiO"
4 + but decreasing
53
more rapidly thfin Cl
which is again more strongly adsorbed than bromide even though the
bromide forms the more insoluble thallium salt (cf. p. 107). This
may be caused by the greater deformability of the larger complex ion.
Cations displace thorium B in the order: Tl > Ag > Cu > H.
The adsorption isotherm of cupric ions by silver bromide was de-
termined by Luther, 9 who was investigating the mechanism of the
photographic desensitizing action of copper salts. The copper salt
* J. Phys. Chem., 37, 851 (1933).
Z. physik Chem., AIM, 168 (1929).
Trans. Faraday Soc., 19, 394 (1923).
130 COLLOIDAL HALIDES OF SILVER, LEAD, AND MERCURY
was added to a suspension of the silver halide and the adsorption meas-
ured by a catalytic method based on the catalytic action of cupric ions
on the reaction between persulfate and thiosulfate in the presence of
iodide ions. The adsorptive power of the salt after coagulation was
less than when suspended. The maximum adsorption of the coagu-
lated compound was approximately 1 mol Cu++ per 400 mols AgBr.
Adsorption of Dyes
Studies of the adsorption of dyes on silver bromide and chloride
confirm the results previously given for the adsorption on silver iodide
(p. 107). For example, Wulff and Seidl 10 found that silver and thal-
lium ions increase the adsorption of the acid dye resorcin in the order :
Ag > Tl ; and anions cut down the adsorption in the order : Br >
Cl > SO 4 > PO 4 , CO 3 , B 4 O 7 . The last three ions have little or no
effect. The hydrogen ion concentration influences the adsorption of
dyes in the usual way, increasing hydrogen ion concentration favoring
the adsorption of acid dyes and cutting down the adsorption of basic
dyes. Some typical qualitative observations made by Sheppard, Lam-
bert, and Keenan " are given in Table XIII. It will be noted that the
TABLE XIII
EFFECT OF pH AND OF EXCESS OF COMMON ION ON DYE ADSORPTION
Dye
Nature of dye
pu
Adsorption in the presence of
Excess Br~
Excess Ag +
pinacynol
Basic
5
7 5
+
+ +
+ +
Dichlorofluorescein . .
Acid
5
7 5
-
+ +
+
basic dye pinacynol is adsorbed in an acid medium and in the presence
of silver ion. This is attributed by Sheppard to the formation of
complex ions; but another explanation is possible (p. S3). In any
event, it is a common phenomenon. Thus erythrosin anion is adsorbed
i Z. wiss. Phot, 28, 239 (1930).
J. Phys. Chem., 86, 174 (1932); Colloid Symposium Monograph, 9, 174
(1932).
ADSORPTION OF DYES
131
by silver bromide appreciably both in the presence of an excess of
silver ion (positive AgBr) and of bromide ion (negative AgBr) ; on
silver iodide, however, the dye anion is adsorbed only when silver ion
is in excess. The effects of silver and bromide ions in excess on the
adsorption by silver bromide of the acid dye eosin and the basic dye
phenosafranin are given graphically in Fig. 26. 4 It is apparent that
the adsorption of eosin is very small at the equivalence point but rises
sharply with increasing concentration of silver ion ; on the other hand,
AgN0 3 , Mo(/L x id" 9 (Eosin)
KBr<-->AgN0 3 ..
Electrolytes, Mo/L x 10 (Phenosafranin)
12
FIG. 26 -Effect of excess silver and bromide ions on the adsorption of dyes by
silver bromide. (1) 50 cc eosin containing 10-* tnol/1 with 1 g AgBr. (2) 15 cc
phenosafranin containing 1.6 X 1Q-" mol/1 with 0.2 g AgBr.
the adsorption of the basic dye increases greatly in the presence of
bromide ion and is cut down, but not to zero, in the presence of an
appreciable excess of silver ion. The behavior of the basic dye methyl
violet is similar to that of phenosafranin."
Quantitative observations of the number of bromide ions required
to adsorb one dye ion at saturation are summarized in Table XIV from
the results of several investigators. For purposes of comparison,
similar adsorption data are given for the inorganic ions thallium and
silver. In every instance the adsorption is still far from being a mono-
molecular adsorption. This is in line with the view that the adsorp-
tion is not uniform but takes place only at the more active spots on
the surface of the crystals.
"Hodakow: Z. physik. Chem., 127, 43 (1927).
132 COLLOIDAL HALIDES OF SILVER, LEAD, AND MERCURY
TABLE XIV
ADSORPTION OF DYES ON SILVER BROMIDE
Dye
Ratio at saturation
of the surface
dye ion : bromide ions
Investigator
Methylene blue in neutral solution
Resorcinate in 1 N NaOH
05 mol/1
1 .8
1-17
Wulff and Seidl *
WulfT and Seidl *
025mol/l
.01 mol/1
Tl+inO 1 JVNaOH
Orthochrome T pH = 5 5
Pinacynol pH = 6 8 .
Erythrosin
Ag+
1:25
1:16
1:30
1 :2 3
1 : 1 69
1:3
1:60
Wulff and Seidl *
Wulff and Seidl *
Wulff and Seidl *
Sheppard and Crouch f
Sheppard, Lambert, and
Keenan J
Walker, O. J.
Fajans and Franken-
burger ||
* Z. wiss. Phot., 28, 239 (1930).
t J. Phys. Chem., 32, 751 (1928).
J J. Phys. Chem., 36, 174 (1932); Colloid Symposium Monograph, 9, 174 (1931)
I C/. Fajans and Brdey-Grfz. Z. physik. Chem., 158, 105 (1930).
|| Z. Blektrochem., 28, 499 (1922); Z. phywk. Chem., 105, 273 (1923).
Adsorption Indicators
Definition, and Use. Since silver halides, when positively charged
by adsorption of silver ions, adsorb and deform certain dye anions
strongly, and when negatively charged by adsorption of halide ions
adsorb certain dye cations strongly, the dyes even in low concentration
give rise to an intensely colored adsorption layer. Fajans 13 and his
coworkers have shown that dyes may be employed under suitable con-
ditions as indicators in argentometry. Since the color change at the
end point is due to adsorption, Kolthoff I4 called the dyes adsorption
indicators. The following experiment demonstrates the principle on
which the functioning of an adsorption indicator is based. 13 - 16 About
3 mg of the sodium salt of eosin is added to a liter of distilled water.
"Fajans and Hassel: Z. Elektrochem., 29, 495 (1923) ; Fajans and Steiner:
Z. physik. Chem., 126, 309 (1927); Fajans and Wolff: Z. anorg. Chem, 187,
221 (1924); Hassel: Kolloid-Z., 84, 304 (1924).
"Kolthoff and Furman: "Volumetric Analysis," 111 (1928).
"Kolthoff: Chem. Rev., 16, 87 (1935).
16 Cf. Fajans: "Radio Elements and Isotopes: Chemical Forces and Optical
Properties of Substances," 96 (1931).
ADSORPTION INDICATORS 133
The salt is dissociated and partly hydrolyzed imparting to the solution
a greenish fluorescence and a yellowish red color in transmitted light.
The addition of 2 cc of 0.1 N AgNO 3 causes no appreciable color
change since the solubility product of silver eosinate is not exceeded.
On adding 0.5 cc of 0.1 N alkali bromide, an intense color change to
red or red-violet occurs and the fluorescence disappears. The highly
dispersed silver bromide adsorbs silver ions and, simultaneously, an
equivalent amount of eosin ions which are deformed, the process caus-
ing a marked change in color. On further addition of bromide, more
silver bromide results and the color deepens until the equivalence point
is passed, whereupon the excess bromide ions displace the adsorbed
eosin ions which return to the solution. The silver bromide particles
are decolorized thereby and the solution takes on its original color and
fluorescence. The color change is completely reversible.-
If a basic dye such as Rhodamine 6G is substituted for eosin, the
process is the reverse of the above. In the presence of excess silver
ions, these are adsorbed primarily and prevent the adsorption of the
dye cation; but as soon as a slight excess of bromide is present, the
latter ions are adsorbed and attract an equivalent amount of dye
cations which give an intense red color to the silver bromide sus-
pension.
Mechanisms of the Action of Adsorption Indicators. Fajans' in-
terpretation of the mechanism of the action follows from the above
experiments : A dye anion is not adsorbed by a silver halide as long
as an excess of halide ions is present in the supernatant solution; with
a slight excess of silver ions part of the latter are adsorbed which
results in an equivalent adsorption of the dye anions. The reverse
process takes place with a dye cation as adsorption indicator. The
adsorption usually causes deformation of the dye ion with an intense
color change which marks the equivalence point. It is not essential
that a color change occur at the end point. When the indicator
is unadsorbed, the supernatant solution has the color of the dissolved
dye and the precipitate is while (AgCl) or yellowish (AgBr and Agl) ;
when strong adsorption of the dye takes place, the solution is decol-
orized and the precipitate takes on the color of the dye. Taking ad-
vantage of these facts Berry and Durrant 1T used tartrazin as an indi-
cator in titrating silver with bromide or chloride, and Lang and Mes-
singer 18 used diphenylamine blue to titrate strongly acid chloride with
silver nitrate, under suitable conditions.
1T Analyst, 65, 613 (1930).
18 Ber, 88B, 1429 (1930).
134 COLLOIDAL HALIDES OF SILVER, LEAD, AND MERCURY
According to Fajans' mechanism, a dye anion is adsorbed as a sec-
ondary process following the primary adsorption of silver ion, and a
dye cation is adsorbed as a secondary process following the primary
adsorption of halide ion on the silver halide surface. If this were
strictly true, a dye anion should not be adsorbed in the presence of
excess halide ion and a dye cation should not be adsorbed in the pres-
ence of excess silver ion. As a matter of fact, acid dyes are fre-
quently adsorbed to some extent even in the presence of excess halide
and basic dyes in the presence of excess silver. Eosin, for example, is
useless as an adsorption indicator in the titration of chloride since the
dye ion is adsorbed by silver chloride from the beginning of the titra-
tion in spite of the presence of excess chloride ions; hence the end
point is not sharp in titrating chloride. On the other hand, it is en-
tirely satisfactory in titrating bromide and iodide. Similarly, erythro-
sin may not be used as an indicator in titrating silver, chloride, or
bromide but is satisfactory in titrating iodide. The adsorption of
phenosafranine and methyl violet in the presence of both excess silver
and excess bromide has been considered above (p. 131). In spite of
this behavior, phenosafranine is a good adsorption indicator in titrating
silver with bromide; and methyl violet may be used in titrating silver
with chloride. The essential thing is that the dye anion should not be
adsorbed very strongly in the presence of ions of the same sign but
should be highly adsorbed in the presence of a slight excess of ions of
opposite sign, in order that a marked color change may result at or near
the equivalence point.
It has been generally assumed that the adsorption of dye ions in
the presence of lattice ions of the same sign is due to a primary adsorp-
tion of both the lattice and the dye ions. Kolthoff 19 claims, however,
that the adsorption of dye ions under these conditions is wholly or in
part an exchange adsorption between the dye ion and the lattice ion
at the surface of the particles, just as was claimed for the adsorption
of dyes by lead and barium sulfate (p. 32). For example, if silver
chloride containing neither an excess of adsorbed silver nor adsorbed
chloride (equivalent body) is shaken with fluorescein, a slight adsorp-
tion of the dye anion (Fin-) takes place and an equivalent amount of
chloride appears in solution. This may be represented by the replace-
ment equation:
Ag+Cl- + Fln-*=* Ag+Fln- + Q-
Surface Solution Surface Solution
(reddish)
"Kolthoff: Kolloid-Z., 68, 190 (1934); Chem. Rev., 16, 87 (1935); cf. Kolt-
hoff and Yutzy: J. Am. Chem. Soc, 58, 1215, 1634 (1937).
ADSORPTION INDICATORS 135
In this instance the replacement of chloride by fluorescein is slight;
but to quote Kolthoff: "in the presence of a slight excess of silver,
however, the chloride ion concentration decreases sharply, favoring a
marked replacement of chloride ions in the surface by dye ions, re-
sulting in a reddish color of the precipitate." It is apparent that
Kolthoff considers the dye adsorption in any event to be an exchange
adsorption between the dye ions and the lattice ions. With eosin
(Eos-) and silver chloride the exchange is represented by the
equation :
Ag+Cl- + Eos-<=* Ag+Eos- + Cl-
Surface Solution Surface Solution
(red)
As we have seen, the action takes place to such an extent, even in the
presence of excess chloride, that the dye is useless as an adsorption
indicator for chloride titration. With bromide and iodide, on the other
hand, this effect is not sufficiently pronounced to interfere with the
titration.
In titrating silver with bromide using the basic dye phenosafranine
(Phen+) as indicator, the following exchange adsorption of lattice
ions is assumed:
Ag+Br- + Phen+ * Phen+Br" + Ag+
Surface Solution Surface Solution
(red)
And in titrating chloride and bromide with mercurous nitrate using
bromphenol blue (Bp B-), it is believed to be: 20
Hg 2 Cl 2 + 2Bp B- <r Hg 2 (Bp B) 2 + 2C1~
Although Kolthoff has demonstrated the existence of what appears
to be an exchange of lattice ions in the adsorption of dyes on the
silver halides, he seems to have taken an extreme position in claiming
that, under the conditions specified, the dye adsorption is essentially
an exchange adsorption between dye ions and the surface lattice ions.
In the presence of excess silver nitrate, a silver halide adsorbs silver
ions preferentially, nitrate ions being the counter ions in the diffuse
outer layer. If a dye such as fluorescein is present under these con-
ditions it seems altogether probable that most if not all the adsorption
is the result of an exchange between the dye anion and the counter
nitrate ions. A similar view seems to be held by Verwey, 81 who says
2 <> Kolthoff and Larson: J. Am. Chem. Soc., 56, 1881 (1934).
"Kolloid-Z., 72, 187 (1935).
136 COLLOIDAL HALIDES OF SILVER, LEAD, AND MERCURY
that the action of adsorption indicators rests fundamentally on an ex-
change of counter ions, a phenomenon which always takes place when
counter ions of the right sign are in the double layer. He believes
further that a pure or direct adsorption of the dye ion can take place
which complicates the relations in the double layer. This pure ad-
sorption can be mistaken for a lattice ion exchange which, in favorable
cases, corresponds almost exactly with it. Whenever this pure ad-
sorption (or lattice ion exchange adsorption) is large in comparison
with the counter ion exchange adsorption, the dye cannot be used as
indicator.
In any titration it is important to get the color change at the
equivalence point. As Verwey points out, if an adsorption indicator
is used which enters into counter ion exchange only, with no direct
adsorption on the precipitate, the color change should take place ex-
actly at the isoelectric point; if some direct adsorption comes in, the
change will take place on one side or the other of the isoelectric point.
It follows from this that silver iodide with isoelectric point of ^Ag = 6
is titrated with greatest exactness using a suitable dye cation rather
than anion as indicator, since a weak adsorption of the former will
shove the isoelectric point more nearly to the equivalence point.
Applicability of Adsorption Indicators. Even when a suitable in-
dicator is found for a given titration, various factors must be taken
into account for its successful application. 15 Since the adsorption of
the indicator is confined to the surface of the precipitate, the method
is applicable, in general, only when the surface is large. The silver
halides age very rapidly after flocculation thereby decreasing the total
surface of the precipitate ; hence the titration is less exact in the pres-
ence of electrolytes which induce flocculation of the halide sol much
before the end point. For this reason an excess of acid will interfere
with the detection of the end point. Furthermore, hydrogen ions may
exert a specific effect on the indicator. Fluorescein, for example, is a
weak acid, similar in this respect to phenolphthalein, and hydrogen
ions will remove the indicator ions by driving back the dissociation of
the dye. Fluorescein must therefore be used in neutral solution. Dyes
such as eosin, diiodofluorescein, bromphenol red, metanil yellow, etc.,
will work in weakly acid solution, and dichlorofluorescein may be used
in titrating chloride at a pH smaller than 4. M
The concentration of the solution to be titrated is important in de-
termining the applicability of an indicator. Bromides and iodides can
be titrated with eosin as indicator in solutions even more dilute than
^Kolthoff, Lauer, and Sunde: J. Am. Chem. Soc., 51, 3273 (1929).
ADSORPTION INDICATORS 137
0.001 N. Chlorides cannot be titrated with fluorescein as indicator at
dilutions greater than 0.005 N since the more soluble silver chloride
tends to separate in a coarser form than the bromide or iodide. More-
over, the adsorbability of fluorescein is much less than that of chloride.
Dichlorofluorescein is so much more strongly adsorbed than fluorescein
that it may be used in titrating chloride solutions as dilute as 0.001 N,
Under proper conditions it is possible to determine an ion in the
presence of another which also yields a slightly soluble salt with the
reagent. Thus Fajans and Wolff 2S titrated iodide in the presence of
chloride with diiodofluorescein, dimethyl diiodofluorescein, and rose
bengal as indicators; and Kolthoff estimated small amounts of iodide
in the presence of much chloride by adding ammonium carbonate and
using eosin as adsorption indicator.
The application of adsorption indicators in quantitative analysis is
confined chiefly to the titration of halides with silver or mercurous
mercury, or the reverse. In qualitative analysis the adsorption-indi-
cator principle is used in the colorimetric detection of several metals,
for example: magnesium with titan yellow; beryllium with curcumin
and 1,2,5,8-oxyanthraquinone; and aluminum, with the latter reagent
and with alizarin, purpurin, and aurintricarboxylic acid.
HYDROSOLS
Like the corresponding silver iodide sols the negative sols of silver
bromide, chloride, and thiocyanate formed in the presence of a slight
excess of the respective anions are more stable than the positive sols
formed in the presence of excess silver ions, Wereide M describes the
preparation of dilute silver halide sols by electrolysis of the corre-
sponding acid using silver electrodes and either alternating or direct
current. Stable sols may be prepared by precipitating in the presence
of gelatin 25 or gum arabic. 26 It is of interest in this connection, that
concentrated chloride solutions can be titrated quite accurately by
Mohr's method using chromate as indicator, provided 5-10 cc of 0.1%
agar is added to the chloride solution before adding the silver ion.
The effect of the agar is to prevent the balling up of the casein-like
precipitate and thus to give a much sharper end point. 27
as z anorg. Chem, 187, 221 (1924).
**Z. Physik., 41,864 (1927).
"Chatterji and Dhar: J. Indian Chem. Soc., 7, 177 (1930).
2* Van der Widen and Witteboon: Pharm. Weekblad, 72, 1037 (1935).
27 Lottermoser and Lorenz: Kolloid-Z., 68, 201 (1934).
138 COLLOIDAL HALIDES OF SILVER, LEAD, AND MERCURY
The results of the work of Jablczynski and coworkers 28 and of
Schneller 29 on the kinetics of the coagulation of silver halide was con-
sidered in part in the silver iodide chapter (p. 125).
Because of the opalescence imparted to water in which a small
amount of silver halide is suspended, very small amounts of silver 80
or of chloride 81 have been determined by nephelometric titration to
the so-called "equal opalescence" end point. This is an important
operation in precise atomic-weight measurements where the stoichio-
metric ratio is determined between a pure compound furnishing chlo-
ride or bromide ions and pure silver. Johnson 8a has investigated in
detail the effect on the equal opalescence end point of various factors
such as the order of titration, the presence of extra compounds in the
solution, shaking and cooling, etc. In general, it was found that under
the proper conditions the method meets the accuracy demanded in
atomic-weight investigations provided shaking and cooling of the ana-
lytical systems are avoided. These operations in the presence of a
variety of foreign compounds tend to leave in the supernatant liquid
an excess of chloride equivalent to several tenths of a milligram per
liter. Whether this is real or virtual, the effect represents a source of
constant error which, if permitted to operate in an atomic-weight
titration, would tend to give a low value for the calculated atomic
weight. The fluctuations in the observed results may be ascribed to
some combination of the following factors : adsorption of ions by silver
chloride during peptization and coagulation; adsorption on the pre-
cipitate ; differences in the coagulating action of the two precipitating
agents ; the peptizing and coagulating action of the foreign compound.
THE COLLOIDAL HALIDES OF LEAD
With the exception of the fluoride, the halides of lead are much
more soluble than the corresponding halides of silver and so are less
"Bull. soc. chim., (4) 83, 1392 (1923); 85, 1277, 1286 (1924); 89, 1322
(1926) ; 48, 159 (1928) ; 47, 50 (1930).
29 Kolloid-Z., 71, 180 (1935); cf. Jablczynski and Jaszczolt: Roczniki Chem.,
9, 111 (1929).
w Richards and Wells: Am. Chem. J., 81, 235 (1904) ; Wells: 85, 99 (1906) ;
Richards: 85,510 (1906).
"Lamb, Carlton, and Meldrum: J. Am. Chem. Soc., 42, 259 (1920); Kolt-
hoff and Yutzy: 55, 1915 (1933).
82 J. Phys. Chem., 85, 540, 830, 2237, 2581 (1931) ; 86, 1942 (1932) ; 89, 781
(1935); Johnson and Low: 86, 2390 (1932); cf. Briscoe, Kikuchi, and Peel:
Proc. Roy. Soc. (London), 188A, 440 (1932); Scott and Hurley: J. Am. Chem.
Soc., 59, 1297 (1937).
LEAD HALIDES 139
readily obtained in the colloidal state. The solubilities in millimols
per liter at 25.2 are: PbCl 2 , 38.80; PbBr 2 , 26.28; and PbI 2f 1.S8. 83
From the von Weimarn theory one would not expect any of these salts
to be thrown down in the colloidal state from aqueous solution. Ac-
tually the most soluble lead halide, the chloride, precipitates as a slimy
mass which runs through a fine-pored filter paper, when a molar solu-
tion of sugar of lead is mixed with a 2 M solution of sodium chlo-
ride. 84 In a systematic investigation with the three lead halides, von
Weimarn found that the size of the precipitated particles varies in-
versely as the solubility when the precipitation takes place under the
same conditions of supersaturation. 85
Although unprotected hydrosols are quite instable, von Weimarn
obtained fairly stable sols in aqueous alcohol by carrying out the pre-
cipitation preferably in the presence of an excess of lead ions. As
would be expected the stability of the aqueous alcosols follows the
order of decreasing solubility in water : PbI 2 > PbBr 2 > PbG 2 .
Stable hydrosols of lead iodide are obtained with the aid of gelatin
as protecting colloid. 86 Reinders 87 mixed 10 cc each of 0.1 N
Pb(C 2 H 3 O 2 ) 2 and KI with 80 cc of 0.05% gelatin and obtained an
orange-yellow sol with a beautiful silky luster. The particles were
hexagonal plates which exhibited a distinct double refraction when
examined with crossed nicols. With 0.39& gelatin, the sol formed more
slowly and the particles were much finer; but after some weeks the
gradual growth of the crystals caused precipitation of the sol. This
sedimentation may be prevented almost entirely by using sodium "pro-
talbinate" as the protecting colloid : 88 5 cc of a 5% solution of this
protector were heated with IS g of a 20% solution of Pb(C 2 H3O 2 ) 2 .
The resulting lead "salt" was dissolved in sodium hydroxide and
treated with potassium iodide giving a sol which is reddish brown in
transmitted light and greenish yellow in reflected light. After purifi-
cation by dialysis, the preparation may be evaporated to dryness with-
out destroying the sol- forming property of the residue.
Conductivity measurements on mixtures of lead acetate and potas-
sium iodide in the presence of agar show that the agar keeps the lead
iodide in a supersaturated solution and inhibits greatly the precipita-
33 Von Ende: Z. anorg. Chem. f 26, 159 (1901).
3* Van der Velde: Chem.-Ztg, 17, 1908 (1893).
88 Von Weimarn, Chen, Kida, and Yasuda: Kolloid-Z. f 42, 305 (1927).
* 6 Lobry de Bruyn: Z. physik. Chem, 29, 562 (1899).
"Kolloid-Z., 21, 164 (1917) ; cf. Lachs: J. phys. radium, 3, 125 (1922),
88 Leuze: "Zur Kenntnis kolloidaler Metalle und ihrer Verbindungen," 32
(1904).
140 COLLOIDAL HALIDES OF SILVER, LEAD, AND MERCURY
tion of the salt 89 (p. 89). That sol formation is prevented is evi-
denced by the absence of a color change on mixing the solutions. The
precipitation of lead iodide in agar gels is likewise slowed down and
the form of the crystals is influenced by the purity of the agar.
Large, beautifully formed crystals are obtained by allowing a potas-
sium iodide solution to diffuse slowly into a silica gel containing lead
acetate. 40 Rhythmic bands result in silica gel in the presence of citric
or tartaric acids which decrease the grain size of the iodide.* 1
THE COLLOIDAL HALIDES OF MERCURY
Mercurous Halides
The precipitates formed on mixing the respective alkali halides
with mercurous nitrate are fine crystalline powders. These salts have
been used by Hahn and Imre 5 in conjunction with the corresponding
silver salts, in their investigations of the mechanism of the adsorption
process at ion lattices (p. 109) .
Stable hydrosols of mercurous halides have been prepared only in
the presence of protecting colloids. On mixing 0.05 M solutions of
NaCl and slightly acidified HgNO 3> strongly double refracting needle-
crystals of mercurous chloride, 10/* in length, are formed. In the
presence of 0.3% gelatin, the growth of the crystals is inhibited to
such an extent that a stable sol results which is milky bluish-white in
reflected light and exhibits distinct double refraction when examined
with crossed nicols. 42
If a dilute solution of a mercurous salt is treated with a stannous
salt, a sol is formed consisting of hydrous stannic oxide with adsorbed
mercury. Such a sol is converted into a mercurous halide sol sta-
bilized by stannic oxide, by treating with the amount of halogen cor-
responding to the reaction 2Hg + X 2 -> 2HgX or with the amount of
mercuric salt corresponding to the reaction Hg + HgX 2 - 2HgX. 48
Very stable mercurous halide sols are prepared for therapeutic use by
the interaction of a mercurous salt and alkali halide in the presence of
albuminous bodies, followed by dialysis. 44 By careful evaporation of
8Bolam: Trans. Faraday Soc., 24, 463 (1928); 28, 133 (1930).
"Holmes: J. Phys. Chem., 21, 718 (1917); J. Franklin Inst., 184, 758
(1917).
Isemura: J. Chem Soc., Japan, 58, 58 (1937).
"Reinders: Kolloid-Z, 21, 165 (1917).
Lottermoser: J. prakt. Chem., (2) 57, 485 (1898).
"Galerosky: Pharm. Ztg., 230 (1904); Chemische Fabrik von Hayden:
Chem. Zentr., II, 1757 (1905).
MERCURIC HALIDES 141
the sol or by coagulation with alcohol, a gel results which is peptized
by water, alcohol, ether, or benzene. The gel thrown down by an acid
is repeptized by thorough washing with water containing a little alkali.
Any germicidal or fungicidal action of mercurous halide sols is un-
doubtedly due primarily to the ions present and not to the colloidal
particles. 45
Mercuric Halides
Mercuric chloride and bromide are too soluble to form hydrosols
but a fairly stable iodide sol might result in the presence of a suitable
protecting colloid. Charitschkov 46 made mercuric chloride sols by the
action of hydrogen chloride on a solution of mercuric "naphthenate"
in benzene, toluene, or light petroleum.
The transition rhombic Hgl (yellow) ^octahedral Hgl (red) is of
interest especially in connection with the effect of protecting colloids
on the process. The transition temperature for small crystals is usually
given as 127, but Cohen and Bredee 47 place it at 125.1 to 128.4.
Kohlschutter 48 showed that the transition from the red to the yellow
form was sharp at 129-130, using large single crystals. At room tem-
perature the yellow form, prepared by subliming the red at a low
pressure or in a stream of indifferent gas, changes slowly from lemon
yellow to bright yellow and then to red, the change being more rapid
for large crystals than for small. The transformation was found to be
autocatalytic in both directions. 49
On mixing solutions of mercuric chloride and potassium iodide at
room temperature, the yellow form, identical with that obtained by
heating the red, 50 appears first; but this goes over promptly to the
stable red modification. If gelatin 61 is added to the solutions before
mixing, the yellow form is stabilized probably by an adsorbed film of
gelatin on the surface, and goes over only very slowly into the red.
Egg albumin has a similar action, but agar has little or no effect in
inhibiting the transformation, indicating that agar is not adsorbed
strongly by mercuric iodide. The change from the red to the yellow
Wedekmd and Bruch: Biochem. Z, 208, 279 (1929).
*J Russ. Phys.-Chem. Soc., 52, 97 (1920); 122 (2), 827 (1922)
"Z. physik. Chem , Bodenstein Festband, 481 (1931).
^Kolloid-Beihefte, 24, 319 (1927) ; cf. Coppock: Nature, 188, 570 (1934).
*Benton and Cool- J. Phys. Chem , 86, 1762 (1931).
wjolibois and Fouretier: Compt. rend., 197, 1322 (1933).
51 Friend: Nature, 109, 341 (1922); Sameshima and Suzuki: Bull. Chem.
Soc., Japan, 1, 81 (1926); Sameshima: 8, 189 (1928); Kisch: Kolloid-Z., 49,
433 (1929).
142 COLLOIDAL HALIDES OF SILVER, LEAD, AND MERCURY
form is likewise inhibited by the presence of gelatin, a temperature
well above the ordinary transition point being necessary to effect the
transformation.
The yellow form of the iodide which precipitates first from an
alcoholic solution starts to go over to the red form within IS minutes.
The transformation begins at an edge of the crystal and spreads
parallel thereto to the opposite edge. 52
"Coppock: Nature, 183, 570 (1934).
CHAPTER VIII
THE SILVER HALIDES IN PHOTOGRAPHY
PHOTOCHEMICAL DECOMPOSITION OF SILVER HALIDES
Cause of the Darkening
The darkening of silver chloride in the light was probably dis-
covered by Schulze in 1727. Scheele * likewise observed it and was
the first to show that chlorine is liberated in the process and that
violet rays are more effective than red or green rays in producing the
phenomenon. The darkening is caused by the liberation of finely di-
vided metallic silver which colors the salt violet to brown to black.
The photochemical decomposition with the accompanying discolora-
tion is superficial ; hence the loss in weight due to vaporization of the
halogen liberated with the silver is slight when the halide is insolated
in mass. 8 By the aid of a microbalance Hartung 8 showed that thin
films of the halides are almost completely decomposed into silver and
the halogen on prolonged exposure to sunlight. Thus it was found
that more than 96% of the bromine was expelled by insolating thin
films of silver bromide in a vessel containing a suitable halogen ab-
sorbent, which was evacuated to 0.001 mm before sealing. Under
similar conditions, 83% of the iodine was expelled from, silver iodide
and 9\% of the chlorine from silver chloride. If the vessel was filled
with hydrogen instead of air before evacuating, the percentage of
chlorine expelled from silver chloride was 95, and that of iodine
from silver iodide, 92. Hartung showed conclusively that the decom-
position products are always silver and the halogen; 4 there is no indi-
cation of the formation of a sub-halide.
*Cf. Mellor: "Comprehensive Treatise on Inorganic and Physical Chem-
istry," 3, 391 (1923).
*Cf. Baker: J. Chem. Soc., 81, 728 (1892); Richardson: 69, 536 (1891);
Koch and Schrader: Z. Physik, 6, 127 (1921).
Phil. Mag., (6) 43, 1056 (1922) ; J. Chem. Soc., 121, 682 (1922) ; 127, 2691
(1923); 129, 1349 (1926).
*C/. Koch and Kreiss: Z. Physik, 32, 384 (1925) ; Ehlers and Koch: 3, 169
(1920).
143
144 THE SILVER HALIDES IN PHOTOGRAPHY
In this connection attention should be called to the existence of an
optimum concentration at which chlorine attacks silver most readily,
the rate of chlorination decreasing rapidly as the concentration of
halogen in the surrounding air increases. This recalls the behavior
of phosphorus, which is not attacked by pure oxygen at room tem-
perature probably owing to the formation of a thin protecting film
of oxide. 5
Mechanism of the Process
Sheppard and Trivelli 6 and Fajans, 7 at about the same time and
independently, proposed the following reaction scheme to represent
the primary photochemical change in the production of the latent
image (p. 153), where represents an electron:
(a) Br~ 8 = Br (bromine atom)
(b) Ag + + = Ag (silver atom)
Eggert and Noddack 8 first investigated the photochemical decompo-
sition of silver halide with the purpose of determining whether Ein-
stein's photochemical equivalence principle is applicable. From the
results of their observations it was concluded that the primary reac-
tion agrees with this principle in the sense that, for each quantum of
light absorbed, one silver atom is produced, thus :
(a) Br~ + hv - Br + 6
(b) Ag+ + =Ag
Eggert and Noddack's experiments were carried out with gelatin-silver
bromide plates, the gelatin in which was supposed to absorb most of
the shorter waves. Weigert 9 claims that this is not the case and that
Eggert and Noddack's conclusions are therefore not justified. More
recently, however, the applicability of the photochemical equivalence
principle has been verified by Hilsch and Pohl 10 and by Feldman and
*Cf. Weiser and Garrison: J Phys. Chem., 26, 61, 349, 473 (1921).
a Phot. J., 61, 403 (1921); Sheppard and Vanselow: J. Phys. Chem, 83, 331
(1929).
7 Fajans and Bcckerath: Chem-Ztg., 45, 666 (1921); Fajans and Franken-
burger: Z. Elektrochem., 28, 499 (1922).
Sitzber. preuss. Akad. Wiss., 81, 631 (1921).
9 Z physik. Chem, 90, 499 (1921); Z. Physik, 18, 232 (1923).
10 Z. Physik, 64, 612 (1930).
OPTICAL SENSITIZATION
145
Stern " with the silver halides in the absence of gelatin. Some results
are summarized in Table XV in which JVAgBr/AQ fe the quotient ob-
tained by dividing the entire number of molecules of AgBr present,
^AgBr, by the number of absorbed quanta AQ; and , the quantum
yield, is given by AW/A0, where A AT is the number of halogen ions
TABLE XV
QUANTUM YIELD IN THE PHOTOCHEMICAL DECOMPOSITION OF SILVER HALIDES
Wave length
Preparation
of light
*y AgBr
***
Investigator
?
~*Q
Photographic plate
436
5X10 3
Approx. 1
Eggert and Nodclack
AgBr (crystals)
AgCl (crystals)
4751
405 /
4X10'
Approx 1
Ililsrh and Pohl
AgBr (precipitated)*
436
1X10
1
Held man and Stern
AgBr (crystals)
365
2X10'
4
Feldman and Stern
* Suspended in a dilute solution of NaNOjj as halogen acceptor.
set free and A Q has the same significance as above. From these and
similar data Feldman and Stern conclude that the quantum yield in
the decomposition of silver chloride and bromide with sodium nitrite
solution as halogen acceptor is unity J2 within the limits of experi-
mental error (less than 10%) ; and that the quantum yield in the for-
mation of the latent image on the photographic plate is probably in
the neighborhood of unity. 13 This does not mean that the whole proc-
ess of electron transfer is as simple as the above equations would
indicate. 1 *
Optical Sensitization
Silver halides are normally photosensitive only in their absorption
region in the blue violet. This sensitivity can be increased, however,
by various processes of what has been termed optical sensitization. The
11 Z. physik. Chem., B12, 449, 467 (1931); B26, 45 (1934); cf. Plotnikov:
Phot Korr,67, 199 (1931).
C/. Mutter: Z. wiss. Phot, 26, 193 (1929).
"Witt x-rays, each absorbed quantum appears to liberate approximately
1000 atoms of stiver. Eggert and Noddack: Z. Physik, 43, 222 (1927); 81,
796 (1925)
"C/. Shcppard and Vanselow: J. Phys. Chem, 83, 250 (1929).
146 THE SILVER HALIDES IN PHOTOGRAPHY
best known optical sensitizers are certain dyes that sensitize the silver
halide for an extended spectral region which is not the same as the
absorption spectrum of the dye in ordinary solvents but which is prob-
ably identical with the absorption of the dye-silver halide combina-
tion. 15 Other optical sensitizers are: adsorbed silver ion, colloidal
silver, and silver sulfide. Because of the colloidal behavior involved,
the mechanism of the optical sensitization process will be considered
in some detail.
Sensitization by Adsorbed Silver Ion. In 1863 Vogel lfl called at-
tention to the fact that silver bromide precipitated in the presence of
excess silver ions changes color in light much more rapidly and
strongly than the salt formed in the presence of excess bromide ions.
The classical explanation of this action is that silver nitrate acts as a
bromine acceptor which removes the bromine liberated by the light in
amount equivalent to the silver, in accord with the equation : AgNO 3 +
Br 2 + H 2 O - AgBr + HOBr + HNO 3 . By removing bromine, the
recombination with silver is prevented; hence the photochemical de-
composition proceeds faster and further in the presence of the ac-
ceptor. This explanation doubtless holds in part, but Fajans and
coworkers " suggest that silver nitrate not only acts as a bromine ac-
ceptor but in some way influences the primary light process also. An
effect of adsorbed ions on the primary light process appears to be the
only way of accounting for the marked increase in photochemical
sensitivity of zinc sulfide in the presence of zinc chloride 18 (p. 288).
To test his theory Fajans 19 determined the absorption curve for
samples of silver bromide in contact with silver and bromide ions re-
spectively. Since, in accordance with the Grotthus-Dreaper law, light
is photochemically active only when it is absorbed, it follows that every
influence which affects the primary photochemical process would alter
the absorption curve. Some actual observations of the absorption on
carefully prepared silver bromide-gelatin emulsions are shown graphi-
cally in Fig. 27, from which it is apparent that the silver body absorbs
light much more strongly than the bromide body. Adsorbed thallium
ions likewise increase the light absorption.
C/. Sheppard: Chem. Rev., 4, 319 (1927).
"Pogg. Ann, 119, 497 (1863).
Fajans and Beckerath: Z. physik Chem., 97, 478 (1921); Fajans and
Frankenburger: 106, 255 (1923) ; Z. Elektrochem, 28, 499 (1922).
"Weiser and Garrison: J. Phys. Chem., 31, 1242 (1927).
"Fajans, Fromherz, and Karagunis: Z. Elektrochem., 88, 548 (1927);
Fromherz: Z. physik. Chem., Bl, 324; Fromherz and Karagunis: 345 (1928).
OPTICAL SENSITIZATION
147
An absorption band for solid substances exhibits a maximum and
two branches that fall more or less steeply toward longer and shorter
wave lengths respectively. In the case of silver bromide it was pos-
sible to study only the branch descending to longer wave lengths, and
the curves of Fig. 27 do not show whether the observed effect indicates
a raising of the whole band, in other words an increase of the absorp-
tion over the whole spectrum, or a shifting of bands toward the longer
wave lengths, or a combination of the two.
3981
AlnA
4466 5011
5623
3.65 3.70
Log A
FIG. 27. Influence of the adsorption of silver and bromide ions on the absorption
of light by silver bromide sol.
The energy e of a quantum transferred to a single molecule, ion,
or atom stands in the following relation to the wave length A, the fre-
quency v, and the velocity of light c
The absorption of light of shorter wave length may thus cause the
greater chemical change ; but the number of absorbed quanta, that is,
the amount of absorbed light and hence the number of photochemically
altered particles, is also of importance in determining the course of
the photochemical change. If the observed effect of the adsorbed
silver ions in the case at hand consists in raising the whole absorption
band, this would indicate that the amount of absorbed light is in-
creased, in other words, the number of elementary processes is in-
148
THE SILVER HALIDES IN PHOTOGRAPHY
creased, without altering their character. On the other hand, if there
is a spectral shift of the whole band, it would follow that the amount
of energy required by the process is altered. If the latter represents
the facts, then, as Fig. 27 shows, the adsorbed silver ions cause a
shifting toward the region of longer wave lengths which means that
the amount of energy required to transfer an electron from a bromine
ion to a silver ion has been decreased by the adsorbed silver ions.
Fajans and Karagunis 20 made absorption measurements on silver
iodide where the whole absorption band is available for study. The
results are shown graphically in Fig. 28. With this salt also was ob-
served an increase in absorption of the silver body in comparison with
3548
5011
+ 0.4
Agl Alone
1 o Withtf/20AgCI0 4
2 With N/2Q Kl
3 WithN/20NaCI0 4
3.55
360
Log X
3.65
3.70
FIG. 28. Influence of adsorbed ions on the absorption of light by silver iodide.
normal silver iodide and the iodide body. The maximum in the ab-
sorption band was scarcely shifted, however, the whole curve of the
silver body being raised fairly uniformly in comparison with that of
normal silver iodide. These observations indicate that there is no
change in the energy requirements but only an increase in the number
of absorbing molecules or ions. A silver iodide surface coated with
adsorbed silver ions absorbs a larger portion of the incident light than
an equal amount of a normal surface; hence more of the former is
decomposed. It is probable but not proved that adsorbed silver ion on
20 Z. physik. Chem., B5, 385 (1929); Naturwissenschaften, 17, 274 (1929).
OPTICAL SENSITIZATION 149
silver bromide likewise raises rather than shifts the absorption band.
The effects are so pronounced that they are easy to demonstrate. 21
For example, it was found under given experimental conditions that
silver bromide in the presence of potassium bromide solution was not
noticeably colored except by violet light whereas the same salt in con-
tact with silver nitrate was colored more or less throughout the whole
range of the visible spectrum down into the red. A similar behavior
was observed with silver chloride.
These observations indicate strongly that light absorption by the
silver halides with the concomitant chemical change is influenced
markedly by adsorbed silver ions. This raises the question as to how
adsorption of ions influences the absorption of light. Fajans inter-
prets the phenomenon as a result of the deformation of ions 22 which
can be represented first as a polarization or induced dipole formation
and second as a distortion of the electron orbits of the atom. In any
event, the change in photochemical sensitivity is associated in some
way with disturbance of the space lattice of the surface of the ab-
sorbing crystals. 28
The optical sensitivity of a silver halide increases with the p\l
value, being more pronounced above ^H = 7.5. This has been at-
tributed by Fajans to the adsorption of OH~ ions 24 which possess a
lower electron affinity than halide ions and to the deforming action of
the silver ions of the lattice on the adsorbed ions.
Sensitization by Colloidal Silver. In 1867 Becquerel 25 reported
that silver chloride papers after exposure to blue-violet or white light
gave an enhanced visible image on further exposure to yellow and
red rays. It was demonstrated by Luppo-Cramer 2e that this is due to
direct optical sensitization by colloidal silver. The sensitization of
ordinary photographic plates by treating with bisulfite and washing
with alkaline water 27 is probably due both to colloidal silver and to
21 Fajans and Frankenburger : Z. Elektrochem., 28, 499 (1922); Frankcn-
burger: Z. physik. Chem., 105, 273 (1923); Steiner: 125, 275; Fajans and
Steiner: 307 (1927).
22 Fajans: Z. Krist, 61, 29 (1925); Z Elektrochem., 84, 506 (1928); Fajans
and Karagunis: Z. physik. Chem., B5, 385 (1929).
2 C/. Sheppard and Vanselow: J. Phys. Chem., 88, 250, 331 (1929); 84,
2719 (1930); Hevesey: Z. physik. Chem., 101, 337 (1922); Gudden and Pohl:
Z. Physik, 18, 42 (1923); Pohl: Naturwissenschaften, 14, 214 (1926).
24 Cf. Rabinovich and Bogdassaryan : Z. wiss. Phot, 82, 97 (1933).
2 "La Lumiere," Paris, 176 (1867).
*Phot Korr., 46, 269, 339, 579 (1909) ; 47, 21 (1910).
2 *Capstaff and Bullock: Brit. J. Phot, 87, 719 (1920).
ISO THE SILVER HALIDES IN PHOTOGRAPHY
adsorption of hydroxyl ions, as noted above. Treatment with thiosul-
fate likewise produces sensitization, 28 probably because of the forma-
tion of colloidal silver sulfide. 28
Fajans attributes the optical sensitization of colloidal silver, like
that of adsorbed silver ion, to change in the absorption of light as a
result of deformation of silver halide by adsorbed silver. Sheppard, 29
on the other hand, attributes the sensitizing effect primarily to sensi-
tized photoelectric electron emission by silver amicrons and ultrami-
crons, rather than to ionic deformation. 80
Sensitization by Dyes. Two principal classes of dyes have been
found especially useful as optical sensitizers : (a) phthaleins, such as
erythrosin and eosin, which are acid dyes; and (b) cyanins, such as
carbocyanins and isocyanins, which are basic dyes. According to Ban-
croft 81 an optical sensitizer in photography, in cases in which fluores-
cence is barred, is a "colored substance which is adsorbed by silver
bromide, which does not bleed into gelatin sufficiently to form a color
screen, and which is either a powerful enough reducing agent to pro-
duce a latent image with silver bromide when activated by light or is
converted by light into a reducing agent sufficiently powerful to pro-
duce a latent image with light."
The above definition seems to cover the ground practically but it
leaves unanswered the mechanism by which the adsorbed dyes effect
the sensitization. It has been suggested that the adsorbed dye mole-
cules are excited by the light and transfer their absorbed energy by
second-order collisions to silver halide or halide ion similar to the
photochemical sensitization of gas reactions by excited mercury
atoms. 82 This may be represented by the following scheme, where D
stands for dye and D B x, the excited dye.
D B x+Vr--+D + Er + 6
A g + + e -* Ag (silver atom)
If this be true the same molecule of dye could assist in the decompo-
sition of an indefinite number of silver halide molecules. The process
M Cf. Sheppard, Wightman, and Trivelli: J. Franklin Inst, 204, 491 (1927).
J. Franklin Inst, 210, 587 (1936).
so Cf., also, Eggert and Noddack: Z. Physik, 31, 922 (1925).
Bancroft, Ackerman, and Gallagher: J. Phys. Chem., 86, 154 (1932);
Colloid Symposium Monograph, 9, 154 (1931).
82 Cf. Sheppard: J. Franklin Inst., 210, 587 (1930); Ind. Eng. Chem., 22,
555 (1930).
OPTICAL SENSITIZATION 151
cannot be a collisional transfer only, since Leszinski 8S showed that the
absorption of light by one molecule of erythrosin gives at least 20
silver atoms of visible decomposition of silver halide. Sheppard and
Crouch " observed two conditions of adsorption of the basic dye ortho-
chrome T to silver bromide : one, an approach to monomolecular ad-
sorption when the dye is in true solution ; and a second, multimolecular
adsorption when the solution is so concentrated that the dye is in part
in colloidal solution. The maximum effect in optical sensitization was
found well below the level of concentration necessary for the forma-
tion of a complete monomolecular layer, indicating that the sensitiza-
tion is effected in monomolecular patches of relatively few dye mole-
cules. If the collision hypothesis were correct it would seem that any
strongly adsorbed dye should act as a sensitizer for its own absorption
band. But, as already pointed out, many strongly adsorbed dyes do
not sensitize at all and only two groups are actually useful technically.
A second hypothesis is that the adsorbed dye gives up electrons on
absorption of light in their own absorption band and that these elec-
trons reduce the silver ions, the residual organic radical at the same
time undergoing further chemical change. From this point of view
the anion of an acid dye adsorbed to a silver ion might easily transfer
its valence electron to the silver ion, giving a silver atom; but it is
less obvious how the adsorbed cation would lose an electron to silver
ion. Yet, so far as is known, the sensitizing behavior of both classes
of dyes is quite similar.
Since the sensitizing dyes are usually bleached or oxidized in light
and the change is accelerated by silver salts both in solution and as
solid adsorbents for the dyes, it is possible that the sensitizing action
consists in a photoreduction by the dye. Kogel and Steigmann 85 sug-
gest that the dye in light acts as a reducing agent by way of hydrogen
released from water. This like the other theories is without quantita-
tive experimental support.
PHOTOGRAPHIC EMULSIONS
The photographic emulsions, so called, are suspensions of the light-
sensitive silver halides in a colloidal medium such as collodion or gela-
tin. Layers of silver bromide prepared by sedimentation of the sols
M Z. wiss. Phot, 24, 261 (1926).
8 * J. Phys. Chem., 32, 751 (1928).
Z. wiss. Phot, 24, 18 (1926).
152 THE SILVER HALIDES IN PHOTOGRAPHY
have been used for photographic investigations M and for plates sensi-
tive to the ultraviolet ; 87 but, for most purposes, the sols formed in the
presence of gelatin are employed. Since the properties of silver halide
sols vary considerably depending on whether they are formed in the
presence of a slight excess of soluble silver salt or in the presence of
a slight excess of alkali halide, it has been found practicable to dis-
tinguish two classes of silver halide emulsions.
Technical emulsions formed in the presence of excess soluble silver
salt, usually silver nitrate, include "wet collodion," which is used ex-
tensively in photomechanical work, and collodion emulsions for print-
ing out. Silver ions adsorbed on the halide ions of the silver halide
lattice sensitize the primary photochemical decomposition of the salt,
and, as already noted, the silver nitrate may also act as a chemical
sensitizer by combining with bromine liberated by the primary light
process.
The group of emulsions formed in the presence of excess alkali
halide comprise both positive and negative emulsions for development.
The excess soluble halide influences the physical character of the pre-
cipitated salt chiefly by acting as a solvent in the "ripening" process
which will be considered in a later paragraph. In the preparation of
most negative emulsions, soluble iodide up to 5% of the silver halide
is added to the alkali bromide so that the resulting silver bromide con-
tains a small amount of silver iodide. Emulsions used for photo-
graphic papers may consist of nearly pure silver bromide, of mixtures
of silver bromide and chloride, or, in some instances, of pure silver
chloride.
The precipitation of the silver halide must be carried out under
carefully controlled conditions in order to prevent flocculation of the
particles and to ensure uniform dispersion throughout the gelatin. The
action of gelatin as a protecting colloid tends to give a fine-grained
uniform precipitate which always consists of crystalline particles. The
adsorption of gelatin to the crystals is so strong that a monomolecular
layer exists on them even after digestion in boiling water for several
hours. 88 In commercial positive emulsions, the particles seldom exceed
0.3/A in diameter; those in negative emulsions, 3 to 4ft. Actually, there
is usually a wide variation in grain size. Sheppard and his co-
se Herschell : Hunt's "Researches on Light," London, 66 (1859); Schaum:
Eder's Jahrbuch Phot, 74 (1904) ; Weisz: Z. physik. Chem., 54, 322 (1906).
" Schumann: Sitzber. Akad. Wiss. Wien, Abt Ha, 102, 994 (1893).
M Sheppard, Lambert, and Keenan: J. Phys. Chem., 86, 174 (1932); Colloid
Symposium Monograph, 9, 174 (1931).
THE LATENT IMAGE 153
workers 89 have determined the proportion of grains of different size
in given emulsions and have thereby established the grain-size fre-
quency, that is, the relative proportion of grains of different sizes in,
say, 1000 grains. The statistical curves of most emulsions can be ex-
pressed by probability formulas of the exponential or (modified)
Gaussian type. The distribution is of particular importance because
of the relation between grain size and sensitivity in a given emulsion.
THE LATENT IMAGE
The action of light on the silver halides gives products that vary
widely in color, depending on the nature and physical character of
the halide and the time of exposure to the light. These so-called
photo-halides 40 have been found to be colloidal dispersions of silver in
silver halide (Vol. I, p. 121).
In photography with development, the exposures of the sensitive
plate to light are insufficient to cause any microscopically visible change
in the silver halide. Yet some change must take place since the appli-
cation of suitable reducing agents causes the so-called "latent image"
to develop. The nature of the latent image has been a subject of
controversy for a long time. Owing to the very slight change in silver
halide on short exposure, the latent image was thought to be a physical
or allotropic modification of silver halide. Namias 41 assumed poly-
merization; Hurter and Driffield, 42 depolymerization ; Bredig, 43 me-
chanical disintegration ; Jones, 44 a labile form ; and Bose, 45 mechanical
strain. Bancroft 4G points out that all these assumptions, and the fur-
ther one of von Tugolessow 47 that the latent image is an oxidation
product, are untenable since all the phenomena of the latent image can
be duplicated by immersing the plate in a solution of a weak reducing
agent such as sodium arsenite. 48 It would appear, therefore, that the
latent image is some reduction product of silver bromide. At one time
sWightman, Sheppard, and Trivelli: J. Phys Chem., 27, 1, 141 (1923).
"Lea: Am. J Sci., (3) 83, 349 (1887).
""Chimie photographique," 102, 110 (1910)
"Phot. J, 22, 149 (1898).
"Eder's Jahrbuch Phot, 18, 365 (1899).
""Science and Practice of Photography," 1, 383 (1904).
"Phot. J. f 26, 146 (1902).
"Trans. Faraday Soc, 19, 243 (1923).
"Phot. Korr., 40, 594 (1903).
"Bancroft: J. Phys. Chem., 14, 294; Perley: 689 (1910); Clark: Brit. J.
Phot, 69, 462 (1922).
154 THE SILVER HALIDES IN PHOTOGRAPHY
this was believed to be either a sub-halide 49 or a series of sub-halide*
of varying composition. 50 These assumptions appear unfounded since
no one has been able to prepare any sub-halide derived from the chlo-
ride, bromide, or iodide of silver, and it is improbable that any exists. 51
A second assumption is that the latent image consists of an infini-
tesimally small quantity of metallic silver which acts as a germ or
nucleus 58 facilitating the reduction of the surrounding silver halide.
This view has been criticized on the grounds that the latent image
shows none of the reactions of metallic silver, 68 does not exhibit the
potential of colloidal silver, 54 and does not account for the facts of de-
velopment and solarization (image reversal). 55 To get around these
difficulties, the latent image was assumed to consist of a phase of vari-
able composition with silver chloride the end term, that is, a solid
solution probably of silver in silver halide. 58 It is much more probable
that the latent image is an early stage of photo-halide formation and
is, therefore, a colloidal solution of metallic silver dispersed in silver
halide 57 and possibly adsorbed thereby. 55 ' 46 This view seems to ac-
count best for all the reactions of the latent image including the effect
of oxidizing agents and the acceleration in reduction shown in develop-
ability. Thus, the latent image is destroyed, so far as chemical devel-
opment is concerned, by treating with chromic acid (2% CrO$ -f- 1%
H 2 SO 4 ). On the other hand, if the plate is considerably over-ex-
posed, and the silver halide subsequently removed by sodium thiosul-
fate, it is possible to develop an image by an acid silver developer
such as silver nitrate plus ferrous sulfate and acetic acid. Luppo-
Cramer 58 attributes this behavior to the formation of colloidal silver
inside the grain by prolonged exposure. Under ordinary conditions,
Luther: Z. physik. Chem., 80, 680 (1899).
MTrivelli: Chem. Weekblad, 7, 321, 350, 381, 404 (1910) ; 8, 101 (1911) ; 0,
232, 248 (1912).
Baur : Z. physik. Chem , 46, 613 (1903) ; Reinders: 77, 213, 356, 677 (1911).
"Wi. Ostwald: "Lehrbuch allgemeinen Chemie," 2nd ed., 2, 1078 (1893);
Abegg: Arch. wiss. Phot, 1, 268 (1899); Lorenz: Z. Elektrochem., 7, 277
(1900).
w Luppo-Cramer: Phot. Korr., 88, 145 (1901).
"Sheppard and Mees: Proc. Roy. Soc. (London), 76A, 217 (1905); 78A,
461 (1907); Luppo-Cramer: Kolloid-Z., 2, 103, 135 (1908).
"Bancroft: J. Phys. Chem., 17, 93 (1913).
M Luppo-Cramer: "Das latente Bild" (1911); Reinders: Z. physik. Chem.,
77,213 (1911).
57 Abegg: Arch. wiss. Phot., 1, 268 (1899); Lorenz: Z. Elektrochem., 7, 277
(1900).
""Photo. Probleme" (1907).
THE LATENT IMAGE 155
these exposed particles are protected by silver halide and so cannot
initiate reduction with a developer after the surface latent image is
removed by an oxidizing agent. On the other hand, if the protecting
silver halide is first removed, the nuclei are uncovered and can act as
centers for the deposition of silver from a supersaturated solution.
Luppo-Cramer prepared synthetic photo-halides that behave in the
same way as the latent image. If the excess silver is not removed
from the photo-halides by the use of too strong an oxidizing agent,
they are readily developable ; but treatment with chromic acid destroys
this property. It thus appears that the latent image does show certain
of the reactions of metallic silver. There is, of course, no reason why
colloidal silver protected by silver halide should show the same elec-
trical potential as silver in mass.
One objection to considering the latent image as a photo-halide is
that the latter is highly colored, whereas the former never is, under
ordinary conditions. The difference is apparently merely a question of
the amount and degree of dispersity of the colloidal silver. If plates
are exposed to the action of x-rays before exposure to ordinary light,
the x-ray image develops as a pinkish coloration in contrast with the
gray-green tone of the untreated halide. 59 Luppo-Cramer 60 attributes
this difference to the higher degree of dispersity of the silver in the
grains exposed to x-rays. The effect of the x-rays is to produce a much
larger number of nuclei per grain and hence a larger number of
smaller particles with the resulting pink tone. This is in line with the
observation that the color of silver hydrosols passes from yellow
through brown, red, purple, and blue to gray with increasing particle
size.
Since it is now generally agreed that the latent image consists of
minute traces of metal, the chief question that remains is its origin.
One view assumes that it is already present before the light acts, the
function of the exposure being to change its physical state. 81 Renwick
suggests that the colloidal silver is present originally as a "negative
sol" which is coagulated by light to a "neutral gel" that can catalyze
the chemical development. Weigert, 61 on the other hand, assumes that
the exposure causes micellar deformation or reorientation of pre-exist-
ing specks of colloidal silver, with or without a sensitizer, thereby
59 Luppo-Cramer: "Die Rontgenographie," 29 (1909).
eophot Korr., 47, 337, 527 (1910).
"Weigert: Z. physik. Chem, B3, 377 (1929); Weigert and Luhr: Z. wiss.
Phot, 28, 312 (1930) ; Schmidt and Pretschner: 25, 293 (1928) ; 27, 36 (1929) ;
28, 30, 35, 111 (1930).
156 THE SILVER HALIDES IN PHOTOGRAPHY
producing a more active nucleus for catalyzing the action of the
developer.
The second and more probable theory of latent-image formation
assumes that the silver is released by the photochemical decomposition
of the silver halide according to the same primary reaction as the
measurable decomposition (p. 144). In support of this, it has been
demonstrated that the spectral sensitivities of the silver halides are
determined by the halide, a fact established by the behavior of mix-
tures, 62 whereas, on the basis of Weigert's assumption, the sensitivities
should be determined by the specific silver adsorption complex with-
out any reference to the halide. In this connection Sheppard 88
showed that desensitizing plates by means of chromic acid or other
oxidizing agent does not change the relative spectral sensitivity dis-
tribution although greatly reducing its absolute value.
Furthermore, the observations of Toy and Harrison M on the photo-
conductivity effect of silver halide crystals, that is, the change in
conductance on illumination, indicate that the primary light action
consists in the release of valency electrons from the bromide ions which
are changed into bromine atoms that react with other atoms and mole-
cules such as those in the neighboring gelatin. A permanent change
thus takes place with the formation of metallic silver as one of the
reaction products.
Further reference to the nature of the latent image will be made
in the next section.
PHOTOGRAPHIC SENSITIVITY
In an earlier section, attention has been given to the optical sen-
sitivity of the silver halides and the methods of varying it. Photo-
graphic sensitivity or developable sensitivity (latent image) is a dif-
ferent effect which will be given special -consideration.
Sheppard 85 defines the sensitivity of a photographic emulsion by
the exposure in candle-meter-seconds (product of intensity of light
and time of exposure) necessary to give a certain normal negative
with chemical development behind a sensitometer table. Practically,
silver bromide emulsions may be prepared which range from the Lipp-
"Huse and Meulendyke: Phot. J, 66, 303 (1926).
63 Colloid Symposium Monograph, 3, 76 (1925).
Proc. Roy. Soc. (London), 127A, 613, 629 (1930) ; cf., also, Toy and
Edgerton: Phil. Mag., (6) 48, 947 (1924).
Colloid Symposium Monograph, 1, 346 (1923).
THE SENSITIVITY SUBSTANCE 157
mann type (speed 1) to the high-speed type (speed 150,000). A
number of conditions are operative in determining the sensitivity:
Ripening
The speed of an emulsion of the Lippmann type may be increased
by ripening in the sense of digestion of the emulsion at temperatures
up to 80 in the presence of silver bromide solvents such as ammonia
or soluble bromide. Since the ripening process consists chiefly in the
dissolution and reprecipitation of silver bromide, 85 ' 68 with the conse-
quent coarsening of grain, it has been generally assumed that a coarse
grain is more sensitive than a fine one. This is only partly true, since
grains of the same size may differ in sensitivity when present in dif-
ferent emulsions. Thus, mere ripening of an emulsion of the Lipp-
mann type might increase the speed five hundredfold or more, but it
will not give an emulsion of the high-speed type. Such emulsions are,
in general, prepared with relatively high concentrations of the reacting
silver salt and alkali bromide, low gelatin concentration, high tempera-
ture of mixing, and retarded addition, so that the dispersity of grain
is determined at mixing and is altered but little by ripening. 87 The
extended investigations of Sheppard and his collaborators on the grain-
size distribution in a large number of different emulsions disclose
that though coarseness of grain is a necessary condition, it is not a
sufficient condition 65 for high speed. In general, the sensitivity in-
creases statistically with the size of grain in one and the same emul-
sion, but with different emulsions there is no necessary relation be-
tween average grain size and sensitivity. The absence of a definite
correlation between grain size and sensitivity suggests that some other
factor is operating.
The Sensitivity Substance
It is a well-known fact that a much faster plate is obtainable with
gelatin than with dry collodion. The formal way to account for this
is to assume that gelatin is a better sensitizer than collodion since
gelatin will react with bromine more readily than collodion does and
hence is a better depolarizer. This cannot be the whole explanation
since different gelatins show marked differences in their sensitizing
action. For the same reason, it is not permissible to attribute the
66 Sheppard and Lambert: Colloid Symposium Monograph, 4, 281 (1926);
cf. f however, Renwick: Phot. J., 64, 324 (1924).
67 Trivelli and Sheppard : "The Silver Bromide Grain of Photographic Emul-
sions," Van Nostrand, 104 (1921).
158 THE SILVER HALIDES IN PHOTOGRAPHY
superiority of gelatin over collodion to the possibility of producing
larger individual crystals of silver bromide in the former medium than
in the latter.
Some years ago Luppo-Cramer 88 showed that the sensitivity of
ripened photographic plates was diminished appreciably by treatment
with chromic acid before exposure, followed by thorough washing.
Luppo-Cramer accounted for this behavior by postulating the existence
of "Reifungskeime" or sensitive nuclei of highly dispersed colloidal
silver. 89 Sheppard 70 and his collaborators showed that practically
every kind of photographic emulsion could be desensitized in this way
and pointed out the relation between this phenomenon and the destruc-
tion of the latent image by chromic acid and other oxidizing agents.
It was shown further that the smaller grains were reduced relatively
more in sensitivity than the larger ones by the chromic acid treat-
ment. 71 It was evident, therefore, that the silver halide grain contains
a sensitivity substance other than silver bromide which increases pho-
tographic sensitivity and is destroyed by chromic acid.
A fundamental advance which made possible the isolation and
identification of the sensitivity substance was made at the Eastman
Kodak Company by R. F. Punnett. He succeeded in obtaining from
a gelatin which yielded highly sensitive emulsions an extract which
could be added to emulsions of low sensitivity and thereby increase
both their speed and density-giving power. A systematic search for
this sensitizing material showed it to consist for the most part of cer-
tain organic sulfur-containing substances, such as mustard oil and
bodies derived therefrom. 68 The action of these may be illustrated by
taking allyl mustard oil as an example. This compound reacts with
ammonia as follows :
NHCsHs
C 3 H 6 CNS + NH 3
the thiocarbamide formed reacting with silver halide in excess to give
a relatively insoluble silver salt :
y NHC 3 H 5
+ AgBr - Br C^-S Ag
NH 2 X NH 2
""Phot. Mitteilung," 328 (1909).
C/., also, Renwick: J. Soc. Chem. Ind., 89, 1S6T (1920).
TO Sheppard, Wightman, and Trivelli: J. Franklin Inst, 196, 779, 802 (1923).
"C/., also, Clark: Phot. J., 68, 230 (1923).
THE SENSITIVITY SUBSTANCE 159
These compounds do not appear to be the actual sensitizers, for, on
treating with slightly alkaline solutions, silver sulfide is formed in
some such way as the following:
Br C^-S Ag + AgBr = Ag 2 S + 2HBr +
N NHC 3 H 5
Since definite sensitization is obtained only after treatment of the
emulsion at sufficiently high />H value to produce Ag 2 S, it is probable
that this compound is the effective sensitizer.
The existence in gelatin of varying amounts of sulfur bodies which
yield the sensitizer, silver sulfide, accounts for the facts that gelatin
emulsions are more sensitive than collodion emulsions and that gelatin
emulsions made with different samples of gelatin, but otherwise simi-
lar, may exhibit marked differences in sensitivity. The action of
chromic acid in cutting down the sensitivity substance is due some-
times, perhaps, to destruction of the mustard oil impurity, but more
generally to destruction of silver sulfide nuclei.
Mechanism of the Sensitization. Trivelli and Sheppard 72 first
showed that the photochemical decomposition of the octahedral sur-
faces of silver bromide crystals does not take place simultaneously
over the entire surface of the crystal but starts at isolated points.
Similarly, Hodgson 7S found that, if the development of exposed single
grains was carried out with a dilute developer and interrupted at an
early stage before the grain was completely reduced, the development
was observed to commence at one or more isolated points on the grain.
Svedberg 74 and Toy 7B showed these sensitivity centers to be statisti-
cally distributed among the grains in a purely haphazard manner.
Further, it appeared to be sufficient for a grain to have one such devel-
opment center in order to be completely developable while the chance
of a grain having one such center at a given exposure to light in-
creased with the size of the grains in one and the same emulsion. As
already noted, Sheppard, Wightman, and Trivelli 76 showed that treat-
ment of an emulsion with chromic acid, prior to exposure, desensitizes
72 "The Silver Bromide Grain of Photographic Emulsions," Van Nostrand,
83 (1921) ; Phot J. f 63, 334 (1923) ; J. Phys Chem., 29, 1568 (1925) ; cf., also,
Lorenz and Eitel: Z. anorg. Chem., 01, 57 (1915).
"J. Franklin Inst, 184, 705 (1917).
"Phot J., 62, 180, 310 (1922).
78 Phil. Mag, (6) 44, 352 (1922) ; Trans. Faraday Soc., 19, 290 (1923).
76 J. Franklin Inst, 196, 674 (1923).
160 THE SILVER HALIDES IN PHOTOGRAPHY
the smaller grains to a greater extent than the larger ones. The sensi-
tivity spots therefore must exist prior to exposure and are not pro-
duced either by exposure 77 or by development. It now appears that
these nuclei are silver sulfide particles 78 which exist at certain points
in the crystal lattice of the bromide, causing a localized concentration
of the photochemical effect whereby the same light energy which would
have produced silver atoms dispersed about the silver halide crystal
produces the same number immediately contiguous to the silver sulfide,
thereby affording a nucleus for development for a correspondingly
smaller exposure.
From this point of view the latent image consists of colloidal
silver on nuclei of silver sulfide. It has been suggested 79 that the
silver sulfide nuclei are surrounded by halos of deformed ions in the
silver halide lattice in such a way that light falling on the grain is
oriented toward the centers and the photochemical reduction of the
halide takes place in their immediate vicinity. It is supposed that a
center makes a grain developable when it reaches a certain size and
the orienting effect of a center grows with the formation of silver, in
other words, is autocatalytic. On this theory, the greater sensitivity
of the larger grains in the same emulsion is a consequence both of the
increased chance that a larger grain has a larger sulfide nucleus and
of the increased mass of silver salt available to afford oriented photo-
product. Disintegration by chromic acid is due primarily to the de-
struction of the silver sulfide nuclei which are active in initiating the
localized decomposition. Any residual sensitivity after chromic acid
treatment is probably due to nuclei which are protected by the silver
halide.
An interpretation of the photolytic process in the silver halide lat-
tice from the point of view of the quantum mechanics of crystals has
been given by Webb. 80
Action of Silver Iodide
It has been pointed out that silver iodide up to $% is usually
present in silver bromide emulsions; more than 5% is detrimental since
it reduces the developability. Now the crystal structure of silver bro-
"C/, however, Silberstein: Phil. Mag, (6) 44, 257, 956 (1922).
"Sheppard: Phot. J., 49, 380 (1925).
Sheppard, Trivelli, and Loveland: J. Franklin Inst, 200, 51 (1925);
Trivelli: 204, 649 (1927); Sheppard and Trivelli: Phot. Korr., 64, 145, 173,
242, 273 (1928) ; Sheppard Phot. J., 68, 397 (1928).
8 J. Optical Soc. Am., 26, 367 (1936).
ACTION OF SILVER IODIDE 161
mide is simple cubic of open type, whereas that of silver iodide is
tetrahedral of a much more compact form. X-ray analysis 81 shows
that the presence of silver iodide in the silver bromide lattice enlarges
the lattice and consequently induces changes in the interatomic forces.
This produces a condition of strain which renders the mixture more
light sensitive than the simple bromide lattice. Fundamentally, the
iodide ion is an optical sensitizer it loses an electron for a lower
quantum number. The effect is, perhaps, still further enhanced by
the presence of the foreign colloidal particles of silver sulfide within
the lattice.
An examination of certain high-speed iodobromide emulsions by
Ren wick and his collaborators 82 disclosed the presence of more iodide
in the larger grains than in the smaller ones. It has been inferred
from this that the apparent relation of size of grain to sensitivity in
the same emulsion may be caused entirely by the greater proportion
of iodide in the larger grains. Sheppard and Trivelli 83 disproved this
assumption by showing that the larger grains of a silver bromide emul-
sion free from iodide are markedly more sensitive on the average than
the smaller grains. Steigmann 84 reports that the addition of colloidal
silver iodide to emulsions exerts a restraining action on the ripening
of the emulsion both in regard to the size of grain and to the sensi-
tizing action of the gelatin.
DEVELOPMENT
The development of the exposed photographic plate consists essen-
tially in the conversion of silver ions to metallic silver. Technically,
two types of development are distinguished: (1) "Physical" develop-
ment, so called, in which the silver which builds up the developed
image on the colloidal silver nuclei of the latent image is obtained
from soluble silver salt in the developing solution. Owing to the high
concentration of silver ions, the reduction potential is low and reducers
are employed in acid solution. (2) "Chemical" development, in which
silver is formed by reduction of the solid silver halide grains them-
selves, the reduction being catalyzed by the silver nuclei of the latent
siWilsey: J. Franklin Inst, 200, 739 (1923); Slater-Price: Phot J., 75,
447 (1935).
82 Renwick and Sease: Colloid Symposium Monograph, 2, 37 (1925); Phot.
J., 64, 360 (1924); Baldsiefen, Sease, and Renwick: 66, 163 (1926); cf Luppo-
Cramer: Z wiss. Phot, 27, 9 (1929).
a J. Franklin Inst, 208, 829 (1927).
"Phot. Ind., 27,375 (1929).
162 THE SILVER HALIDES IN PHOTOGRAPHY
image. In this case, higher reduction potentials are employed, the so-
lutions being alkaline. 85
The commonly held theory of the development of the latent image
is that the developer is a reducing agent sufficiently powerful to reduce
exposed silver bromide which has a nucleus in each grain on which
the deposition of silver can take place, but not powerful enough to
reduce unexposed silver bromide. 86 Bancroft 58 points out that this
theory is in itself inadequate since measurements of the electromotive
force of developers and of silver bromide before and after exposure
fail to substantiate the theory.
In general 4e a photographic developer must be a fairly strong re-
ducing agent, but such substances as stannous chloride, w-aminophenol,
formaldehyde, and gallic acid do not give a satisfactory negative. All
developers will develop gelatin-free silver bromide readily, irrespective
of whether the silver salt has been exposed to light. Most developers
will give negatives after very short exposures and positives after very
long ones; but some give positives even with very short exposures.
With a given developer, the exposure necessary to cause a change
from negative to positive varies with varying concentration of the de-
veloper. Some developers work rapidly and others slowly, the differ-
ence not always being a question of reducing power, although this is
important.
Since a negative may result with one developer and a positive with
another, it is apparent that the potentials of exposed and unexposed
silver bromide are not the only factors. If this were true, a negative
would always be obtained with one exposure and a positive with an-
other, irrespective of the developer. The developer must therefore
have functions other than those of reducing agents with different
strengths. To account for the varying behavior, Bancroft postulates
that the developers are selectively adsorbed at the surface of the grain
giving markedly varying concentrations at these surfaces. If the de-
veloper is adsorbed much more strongly by exposed silver bromide
than by the unexposed halide, the former will develop more rapidly,
giving a negative; if the conditions are reversed, a positive will result;
and if there is little difference in the adsorptions, the development will
be more or less uniform over the whole plate. As Bancroft points out,
the adsorption theory enables one to predict the three types of fairly
Cf. Sheppard and Mees: "Theory of the Photographic Process"; Nietz:
"The Theory of Development/' Van Nostrand (1922).
8C/., for example, Nietz: Phot. J. f 60, 281 (1920).
DEVELOPMENT 163
strong reducing agents actually encountered, but it does not necessarily
mean that the developers behave in this way.
Sheppard and Meyer 87 are likewise of the opinion that adsorption
of the reducer to the silver halide grain is the first phase of develop-
ment. They suggest that the free silver ions of the crystal lattice hold
ions of the reducer by electroaffinity, the* adsorption complex thus
formed breaking down in the presence of the nuclei of colloidal silver.
Selective adsorption doubtless plays an important role also in the effects
of certain dyestuffs, some of which accelerate development and others
of which desensitize the halides to such an extent that development
can be carried out in strong yellow or even white light. 88
Bancroft believes that the selective adsorption which he postulates
involves more or less peptization and that there exists a transition
from weakly peptizing, strongly reducing to strongly peptizing, weakly
reducing developers, the furthest examples of the latter giving direct
solarized or reversed images. In support of his view, Bancroft 4e suc-
ceeded in destroying the developability of an exposed plate by washing
with elon (metol) containing no alkali, so as to exclude reduction.
This was also observed by Luppo-Cramer and confirmed by Shep-
pard, 89 who, however, questions whether this is truly a peptization of
the latent image and suggests that it may be due to slight etching of
the surface layers of the silver halide by the reducers which form
soluble complexes with silver halides. In the same connection, Shep-
pard points out that the production of direct positives by development
is associated with the formation of "dichroic fogs," colored forms of
colloidal silver produced when a solvent for silver halide is present in
the developing solution.
A modification of the adsorption theory of development has been
put forward by Rabinovich. 90 According to this, the silver nuclei of
the latent image themselves selectively adsorb the developing agent.
Mention should also be made of the work of Reinders 91 and of Beu-
kers 92 which has brought forward new quantitative evidence for the
importance of the oxidation-reduction potential in photographic de-
velopment.
J. Am. Chem. Soc. f 42, 690 (1920).
88 Luppo-Cramer : "Negativentwicklung bci hellem lichte (Safraninver-
fahren)," Liesegang's Veriag., M. Eger, Leipzig (1921).
89 Bogue's "Theory and Applcation of Colloidal Behavior/ 1 2, 776 (1924).
9 Z. wiss. Phot, 88, 57; Peisakhovich : 94 (1934).
91 J. Phys. Chem., 88, 783 (1934).
92 "Fotografisch Outwikkerlaars," Thesis, Delft (1936).
PART III
THE COLLOIDAL SULFIDES
CHAPTER IX
COLLOIDAL ARSENIC TRISULFIDE: GENERAL PROPERTIES
THE PRECIPITATED SALT
Physical Character
Arsenic trisulfide is thrown down as a citron-yellow flocculent mass
or gel by passing hydrogen sulfide into a solution of arsenic trioxide
acidified with hydrochloric acid. Spring 1 claimed to get a definite
hydrate, As 2 S 3 6H 2 O, by drying the gel at 20 in air having a rela-
tive humidity of 70%. The specified volume of the alleged hydrate
was greater than that of the sum of its constituents ; hence water was
removed by applying 6000-7000 atmospheres pressure. In spite of
the analysis and the behavior under pressure, it is unlikely that Spring's
preparation was other than arsenic trisulfide with adsorbed water, the
composition of which like that of any hydrous precipitate could be
varied continuously by changing the conditions of drying.
Evidence of the colloidal character of the precipitated sulfide is
furnished by the observation that the freshly formed gel is decomposed
appreciably by water, giving hydrogen sulfide and arsenious acid,
whereas the gel previously heated to 125 is more stable. 2 When
carried out carefully, the estimation of arsenic as arsenic trisulfide can
be done with quantitative accuracy, 8 but Schmidt 4 claims that the pre-
cipitate contains traces of As(SH) 3 and As 2 O 3 , the respective errors
produced thereby being in opposite directions and so equalizing each
other.
Films of arsenic trisulfide formed by the action of hydrogen sulfide
on the surface of arsenic trichloride solutions are said to be hydrophilic
and to have a layer structure 180-200 molecules thick. 5
iZ. anorg. Chem., 10, 185 (1895).
2 Clermont and Frommel: Compt. rend, 87, 330 (1878) ; Chodounsky: Chem.
Zentr., I, 569 (1889); cf. Cross and Higgin: Ber., 16, 1195 (1883).
"Puller: J. Chem. Soc. f 24, 586 (1871); Friedheim and Michaelis: Z. anal.
Chem., 84, 505 (1895).
Arch. Pharm., 266, 45 (1917); Chem. Abstracts, 11, 3005 (1917).
"Demenev and Mokruschin: J. Phys. Chem. (U. S. S. R.), 7, 763 (1936).
167
168 COLLOIDAL ARSENIC TRISULFIDE: GENERAL PROPERTIES
Color. The citron-yellow precipitate thrown down with hydrogen
sulfide from acid solution dries to an impalpable yellow powder. In
addition to the yellow product Winter 8 prepared what he called a red
allotropic modification of the trisulfide by freezing or evaporating the
hydrosol, or by coagulating the hydrosol by a salt or acid followed by
drying. The observations and conclusions of Winter were opened to
question by more recent work of Semler 7 and of Bhatnagar and Rao. 8
Semler was unable to prepare a red sulfide except by coagulating the
arsenic trisulfide sol with an electrolyte which was believed to give a
red sulfarsenite ; thus a red preparation was obtained by coagulating
the sol with barium chloride but not with sodium chloride. Bhatnagar
and Rao claimed that the red preparation was not the trisulfide As 2 S 3
but more nearly the disulfide As 2 S 2 . 9
Winter's view that the yellow and red sulfides are allotropic modi-
fications resulted from the observation that the red preparation changed
to yellow on standing for a long time at room temperature or by pro-
longed heating at 150-160. He could not establish the existence of
an inversion point, however, and the color change is exactly the reverse
of that observed by Borodowski, 10 who reported that the yellow form
was stable up to a temperature of 170 where it changed to red.
Semler's conclusion, that the red color obtained by coagulating ar-
senic trisulfide sol with barium chloride results from the formation of
red Ba2As 2 S 5 , lacks experimental justification. In the first place,
there is no direct evidence of the formation of this compound on add-
ing barium chloride to a purified arsenic trisulfide sol ; and in the next
place, the color of barium sulfarsenite is apparently not such as to
mask the yellow color of the sulfide with red. Berzelius " describes
the sulfarsenite of barium as a gummy reddish-brown mass which dis-
solves completely in water, giving a colorless solution. Nilson, 12 on
the other hand, characterizes the pyrosulfarsenite, Ba 2 As 2 S 5 H 2 O, as
grayish green in color, changing to indigo-blue when allowed to remain
in contact with the mother liquor for some time. The blue color is
probably due to colloidal sulfur dispersed in the salt. 13 The ortho-
sulfarsenite is described as slightly soluble pale yellow prisms which
Z. anorg Chem , 48, 228 (1905).
'Kolloid-Z,84, 209 (1924).
s Kolloid-Z., 88, 159 (1923).
C/. Bhatnagar: J. Phys. Chem., 86, 1803 (1931).
10 Chem. Zentr., II, 297 (1906).
"Pogg. Ann., 7, 142 (1826).
" J. prakt. Chem., (2) 12, 295 (1875) ; 14, 145 (1876) ; 18, 93 (1877).
"C/. Mellor: "Inorganic and Theoretical Chemistry," 9, 296 (1929).
PHYSICAL CHARACTER 169
would have little effect on the yellow color of arsenic trisulfide. Fi-
nally, if one admits that barium sulfarsenite is formed on adding
barium chloride to arsenic trisulfide sol and that the color of the sulfar-
senite is red, there is no justification for assuming that sufficient
amount will be retained by the sulfide to impart a distinct red color to
it. It has been demonstrated that but 0.1 milliequivalent of barium is
adsorbed per gram of arsenic trisulfide during coagulation of the tri-
sulfide sol (p. 203). The adsorbed barium is probably associated
chiefly with HS" and S ion on the sulfide and not with sulfar-
senite anion. Even if one makes the improbable assumption that all
the barium is present as Ba 2 As 2 S 5 , there would be but 1 mol of sulfar-
senite to 160 mols of trisulfide.
Bhatnagar and Rao's contention that the red sulfide is not As 2 S 3
but As 2 S 2 , or colored by As 2 S 2 , is based largely on the results of an
analysis of a red product formed by coagulating an As 2 S 3 sol, freed
from hydrogen sulfide and kept in the dark. It is claimed that As 2 S 2
is formed in accord with the following equation : As 2 S 3 + H 2 O ->
As 2 S 2 H 2 S + O.
Bikerman " obtained a red precipitate by prolonged boiling of a
yellow arsenic trisulfide organosol in nitrobenzene. Merely heating the
sol above 100 changed it to red, the yellow color returning at a lower
temperature with no definite point of transformation. It is obvious
that, in a non-aqueous medium, As 2 S 2 could not be formed in accord
with the scheme of Bhatnagar and Rao nor could a red thioarsenite
form in the absence of both metal and hydrogen ion.
An investigation by Weiser 15 of the cause of the variation in color
indicates that it is due neither to allotropy nor to differences in com-
position. In Table XVI are summarized some observations on the
color of precipitates that were filtered, washed, and dried in a vacuum
desiccator over sulfuric acid. In part (a) of the table, the precipi-
tates, 0.435 g, were obtained by conducting hydrogen sulfide into 50 cc
of solutions of As 2 O 3 containing various electrolytes; and in part (b)
the same amounts of precipitates were thrown down from a hydrosol
with various electrolytes. From these and similar observations, start-
ing with solutions of Na 2 HAsO 4 , it was found that the color of ar-
senic trisulfide varies continuously from yellow through orange-yellow,
orange, red-orange to red, depending on the conditions of precipitation.
The lighter shades always result by direct precipitation, and the darker
shades by coagulation of the sol. There is little if any difference in
i*Z. physik. Chem., 118, 261 (1925).
"Weiser: J. Phys. Chem., 34, 1021 (1930).
170 COLLOIDAL ARSENIC TRISULFIDE: GENERAL PROPERTIES
the color of the sulfide precipitated directly in the presence of barium
chloride and that thrown down in the presence of potassium or am-
monium chloride. The sulfide precipitated in the presence of hydro-
chloric acid is the familiar yellow irrespective of whether barium chlo-
ride is present or absent. The orange-red precipitate formed by coagu-
lation of the sol is the same shade irrespective of the nature of the
electrolyte, indicating that the color is not caused by a colored sulfar-
senite such as Semler assumed.
TABLE XVI
COLOR OF ARSENIC TRISULFIDES
(a) Precipitated directly with H a S
Total volume 50 cc
(6) Precipitated from sol
Total volume 50 cc
As,0a
14g/l
cc
N/2
electrolyte
Color
As 2 S. sol
10 9 g/1
cc
N/2
electrolyte
Color
25
25
25
25
25
25
25
LiCl, 10 cc
NaCl, 10 cc
KC1, 10 cc
NH 4 Cl f 10 cc
BaCU, 10 cc
HC1, 10 cc
HC1, 5 cc +
BaCl 2 , 10 cc
Orange-yellow
Orange-yellow
Orange-yellow
Orange-yellow
Orange-yellow
Yellow
Yellow
40
40
40
40
40
40
40
LiCl, 10 cc
NaCl, 10 cc
KC1, 10 cc
NH 4 C1, 10 cc
BaClt, 10 cc
CaCl,, 10 cc
HC1, 10 cc
Orange-red
Orange-red
Orange-red
Orange-red
Orange-red
Orange-red
Orange-red
(slightly darker)
The flocculent precipitate obtained by direct precipitation dries to
a yellow powder whereas the more gelatinous mass thrown down from
the sol dries to a red horny mass or glass. Both give x-ray diffraction
patterns consisting of three bands in the same position as the sharpest
lines in the diffraction pattern for orpiment, As 2 S 3 ; both analyze for
As 2 S 3 , and there is no indication of the existence of any appreciable
amount of As 2 S 2 in the sulfide as ordinarily obtained or in the prepara-
tion made according to Bhatnagar and Rao's directions.
The marked differences in the physical character of the yellow and
red preparations are probably the underlying cause of the variations
in color. These differences are readily accounted for: When hydrogen
sulfide is conducted into the hydrochloric acid solution of arsenious
oxide, the trisulfide is thrown down in the form of minute amorphous
particles, discrete or collected together into loose clumps which disin-
PREPARATION 171
tegrate on drying, giving an impalpable powder. On the other hand,
when the gas is conducted into an arsenious oxide solution free from
foreign electrolytes, the sulfide appears in the sol state as hydrous
ultramicroscopic particles. As has been emphasized repeatedly, 18 the
rapid coagulation of hydrous sols favors the formation of a gelatinous
precipitate or jelly in which the particles become stuck together or
oriented into aggregates forming an enmeshing network which entrains
liquid. This is what happens when hydrochloric acid or other elec-
trolyte is added to arsenic trisulfide sol in excess of the precipitation
concentration. The gelatinous clumps formed in this way do not disin-
tegrate into a powder on drying, but the minute particles coalesce into
a glassy mass that appears orange-red rather than yellow. The color
of a transparent chip of the dry sulfide formed from the sol is similar
to the color of a thick layer of the relatively concentrated sol viewed
by transmitted light.
If the arsenic trisulfide is yellow when in the form of an impal-
pable powder and red in the form of a horny glassy mass resulting
from coalescence of ultramicrons, it would follow that sufficient grind-
ing of the red mass should yield a yellow powder. This was confirmed
experimentally. Similarly, the red sulfide was rendered yellow by dis-
integrating the glassy mass below the sintering temperature. Heating
the yellow sulfide in the neighborhood of 175 causes it to sinter, con-
tract, and assume a permanent orange to brown color, depending on
the temperature and the duration of the heating.
The red sulfide is stable in the dark and is not affected by light
when thoroughly dry. The combined action of light and moisture
causes a superficial chemical disintegration of the red sulfide, coating
it with a yellow film of sulfide and sulfur.
ARSENIC TRISULFIDE SOL
Preparation
Hydrosols. More than a century ago Berzelius 1T called attention
to the formation of a yellow solution when hydrogen sulfide is passed
into a pure aqueous solution of arsenious acid. Since arsenic trisul-
fide precipitated out slowly, Berzelius suggested that the apparent solu-
tion was probably a suspension of transparent particles. The colloidal
"Weiser: J. Phys. Chcm., 26, 402 (1922); Weiser and Bloxsom: 28, 26
(1924) ; (Vol. II. p. 15).
""Lehrbuch der Chemie," 3rd ed, 3, 65 (1834); cf. Graham-Otto: "Lehr-
buch der Chemie/' 2, 863 (1840).
172 COLLOIDAL ARSENIC TRISULFIDE: GENERAL PROPERTIES
nature of the yellow liquid was first definitely recognized by Schulze, 18
who prepared stable sols of widely varying concentrations by passing
hydrogen sulfide into arsenious acid solution and removing die excess
hydrogen sulfide by boiling or by washing with hydrogen. The limit-
ing concentration of sol obtained in this way is not determined by the
solubility of arsenic trioxide in water. Thus Schulze prepared a sol
containing 37.5% As 2 S 3 by adding small amounts of arsenic trioxide
intermittently to a sol into which hydrogen sulfide was bubbled con-
tinuously ; and Boutaric and Vuillaume 1B obtained one containing 300 g
As 2 S 3 /l by a similar procedure. Picton 20 prepared a sol containing
5 g As 2 S 3 /l by dissolving arsenic trioxide in a solution of potassium
acid tartrate or of sodium hydroxide before adding hydrogen sulfide
and then purifying by dialysis.
As would be expected, the sols prepared under different conditions
contain particles of widely varying sizes. 21 In a quantitative study
of the effect of conditions of sol formation on particle size, Boutaric
and Vuillaume 22 found the particles to be larger the greater the con-
centration of arsenic trioxide, the higher the temperature, the slower
the current of hydrogen sulfide, and the larger the excess of hydrogen
sulfide. Most modifications of Schulze's procedure have been made
with the end in view of obtaining stable sols containing particles of
uniform size. Better results are obtained by allowing a dilute solution
of arsenic trioxide to drop at a slow uniform rate into hydrogen sulfide
water through which the gas is bubbled continuously. 23 The most sat-
isfactory procedure for preparing dilute sols has been worked out by
Freundlich and Nathansohn : 24 SO to 100 cc of cold saturated As 2 O 3
solution are diluted to 200 cc and mixed with 100 cc of a solution
containing 1 cc of saturated H 2 S solution. After a light-yellow colora-
tion appears, the mixture is diluted to 1 1 with a hydrogen sulfide solu-
tion ten times as strong as the above. Finally, the solution is saturated
with the gas, the excess being washed out with hydrogen. The sol
gives a faint light cone in the ultramicroscope, but there are no ultra-
is j prakt. Chem., (2) 25, 431 (1882).
is J. chim. phys., 21, 247 (1924); Boutaric and Simonet: Bull. acad. roy.
med Belg., (5) 10, 150 (1924).
20 J Chem. Soc. f 61, 137 (1892).
21 Picton: J. Chem Soc., 81, 140 (1892) ; Linder and Picton: 67, 63 (1895);
Biltz: Nachr. kgl Ges. Wiss. Gottingen, 2, 1 (1906).
2*Compt. rend., 178, 938 (1924).
as Linder and Picton: J. Chem. Soc., 67, 63 (1895); Kruyt and van der
Spek: Kolloid-Z., 26, 1 (1919).
2* Freundlich and Nathansohn: Kolloid-Z., 28, 258 (1921).
PREPARATION 173
microns. The success of this method of preparation probably lies in
the fact that high concentrations of the reacting solutions are avoided
and the saturation with hydrogen sulfide is delayed until a large num-
ber of arsenic trisulfide nuclei are present.
Concentrated sols which are relatively monodisperse are best pre-
pared by concentrating dilute monodisperse sols by Pauli's process of
electrodecantation (p. 115). Thus Pauli and Laub 85 conducted hy-
drogen sulfide into 18 1 of dilute As 2 O 3 consisting of 15 1 of 50 times
diluted, plus 3 1 of 15 times diluted, saturated As 2 O 3 solution; and
concentrated the resulting dilute sol by electrodecantation to a strength
of 43.2 g As 2 S 3 /l. In spite of the relatively high concentration, the
resulting sol was pure, perfectly clear, and of a beautiful orange-red
color.
On account of the very low solubility of arsenic trisulfide (2.1 X
10- fi mol/1), 26 the transformation to the sol state is quantitative in
the presence of a slight excess of hydrogen sulfide. 27 Since the salt
hydrolyzes to a certain extent, washing the sol with hydrogen for too
long a time results in some decomposition, hydrogen sulfide being car-
ried off and a corresponding amount of arsenious acid remaining in
solution. 28
On mixing 5 cc of 0.1 N H 3 AsO 3 with 5 to 10 cc of 0.01 N H 2 S
in 300 cc of pure water, there results a clear solution which turns
yellow suddenly after a few seconds. If diluted still more, the solution
will remain colorless in the dark for several days, although a trace of
dilute acids causes it to turn yellow immediately. Since the removal
of hydrogen sulfide from the colorless solution with a stream of hy-
diogen takes place very slowly, Vorlander and Haberle 29 conclude
that hydrogen sulfide and arsenic trioxide react in very dilute solution
forming an instable molecular compound which is hydrolyzed to a cer-
tain extent into the original substances. Semler 30 suggests that this
compound may be thioarsenious acid. Peskov 31 makes the more prob-
able assumption that the colorless solution contains ordinary arsenic
trisulfide, the absence of color being due to two factors: the ex-
25 Pauli and Laub: Kolloid-Z., 78, 295 (1937).
2Weigel: Z. physik. Chem, 68, 293 (1907).
"Krister and Dahmer: Z. anorg. Chem., 88, 105 (1902) ; 84, 410 (1903).
28 Ghosh and Dhar: Kolloid-Z., 86, 129 (1925); Krestinskaja and Jakow-
lewa: 66, 187 (1933).
29 Ber., 46, 1612 (1913).
s KolIoid-Z, 84, 213 (1924).
81 J. Russ. Phys.-Chem. Soc., 46, 1619 (1914); J. Chem. Soc., 108 (2), 429
(1915).
174 COLLOIDAL ARSENIC TRISULFIDE: GENERAL PROPERTIES
tremely small magnitude of the primary particles, and, more especially,
the complete individuality of the particles. Agglomeration of the fine
particles into larger secondary aggregates which appear yellow is pre-
vented in the colorless solution owing to the protecting effect of a
relatively large excess of arsenious acid.
Organosols. Bikerman 14 prepared sols of arsenic trisulfide in ni-
trobenzene and in acetoacetic ester by dissolving dry AsQ 3 in the
purified and dried solvents and passing dry hydrogen sulfide through
the respective solutions. Attempts to make sols in acetone and aniline
by the same procedure were not successful. A sol in aniline was ob-
tained by adding realgar (As 2 S 2 ) and sulfur in the right proportions
to aniline followed by heating the mixture above the melting point of
sulfur.
Ostwald and Wannow 82 obtained sols of arsenic trisulfide in acetic
acid, acetic anhydride, and chloroacetic acid as well as in highly con-
centrated sulfuric acid and phosphoric acid by dissolving arsenic tri-
oxide in the respective solvents and conducting in hydrogen sulfide.
Since such sols appear to be uncharged the stability results from
strong adsorption of the dispersing medium (p. 242).
Composition and Constitution
Linder and Picton S3 recognized the necessity for a slight excess of
hydrogen sulfide for the stability of an arsenic trisulfide hydrosol and
therefore concluded that the colloidal sulfides are polymerized hydro-
sulfides such as 16As 2 S 3 H 2 S. Duclaux S4 proposed the general for-
mula (nAs 2 S 3 ^H 2 S) as a basis for explaining coagulation by elec-
trolytes as a purely chemical process. It is common practice to wash
out the excess hydrogen sulfide with hydrogen. As we have seen, re-
moval of too much of the gas, either by washing or by oxidation,
results in hydrolysis of the sulfide with the formation of arsenious
acid more or less of which is adsorbed by the colloidal particles. Ac-
cordingly, if the washing has not been carried far enough, the precipi-
tate from the sol will be high in sulfur, whereas if the washing is car-
ried too far or if the sol is aged, especially in the light (p. 176), it
will contain excess arsenic as As 2 O 3> H 3 AsO 3 , or H 3 AsO 4 . 15 ' 85 Pauli
s'Kolloid-Z., 76, 159 (1936); cf. Voet: J. Phys. Chem., 40, 307 (1936).
88 J. Chcm. Soc. f 61, 114 (1892).
8"Les Colloides," Paris (1920).
a* Cf. Murphy and Mathews: J. Am. Chem. Soc, 46, 16 (1923); Gazzi:
Zymologica, 2, 1, 10 (1927); Krestinskaja and Jakowlewa: Kolloid-Z., 66, 187
(1933).
GENERAL PROPERTIES 175
and Semler 88 consider H 2 As 2 S 4 to be the stabilizing electrolyte and
the sol to be a strongly dissociated complex acid to which they assigned
the formula (*As 2 S 3 -^StHg As 2 S 4 H)~ + H+. This formula
was made to fit the specific case where but one of four hydrogens is
displaced on coagulating the sol with barium chloride, the remaining
three appearing in the supernatant solution after coagulation. Con-
sidering the sols as strongly dissociated complex electrolytes, attempts
have been made to apply the limit laws of Debye-Huckel, 37 but they
hold, if at all, only over a narrow concentration range. 88 As a matter
of fact, it is now quite generally recognized that the composition of the
sol varies with the method of preparation, the excess of peptizing
electrolyte, the age, and the treatment to which it is subjected.
There appears no good reason for attempting to assign a formula to a
mixture of such variable composition. Formed in the presence of hy-
drogen sulfide the particles adsorb H 2 S, and they also adsorb S
and HS~ ions which form the inner portion of the double layer with
H+ ions constituting the diffuse outer portion. The constitution may
be represented diagrammatically as shown in part A of Fig. 37, p. 212.
The particles of an aged sol also contain more or less arsenious acid ; a5
and the presence of thioarsenite in small amounts is neither excluded
nor established. Kargin and Klimovitzkaja 89 believe that arsenious
acid is the stabilizing electrolyte in the sol rather than hydrogen sulfide
or thioarsenious acid; but this is improbable especially with hydrogen
sulfide in excess.
General Properties
Color. The color of arsenic trisulfide sol varies from red-orange
to citron-yellow, depending on the size of the secondary particles and
the concentration. A sol with very small particles appears more orange
than one of the same concentration containing large particles. A con-
centrated sol always appears yellower than a dilute one, probably be-
cause it is likely to contain larger aggregates. If aggregation is pre-
vented by following the procedure of Pauli and Laub (p. 173), even
a concentrated sol possesses an orange color. Because of the marked
Kolloid-Z., 84, 145 (1924); cf., also, Pauli and Laub: 78, 295 (1937).
"Physik. Z, 24, 185, 305 (1923); Debye: Trans. Faraday Soc., 28, 334
(1927).
8 *Audubert: Compt. rend, 196, 210, 306 (1932) ; cf. Hartley: Trans. Faraday
Soc., 81, 31 (1935).
M J. Phys. Chem. (U.S.S.R), 5, 969 (1934).
176 COLLOIDAL ARSENIC TRISULFIDE: GENERAL PROPERTIES
opalescence and coloring power of the colloidal particles, 1 part of
As 2 S 3 in 100,000 parts of water can be detected in thin layers.
Boutaric and Vuillaume 40 studied the absorption spectrum of a sol
by means of the Fery spectrophotometer. Designating the intensity
of the incident radiation A by 7 and the intensity of the radiation
after traversing the absorbing medium by /, then, if the sol behaves
as a turbid medium, log 7 // should vary inversely as the fourth power
of A in accord with Rayleigh's law, where the suspended particles are
small compared with A; and log 7 // should vary inversely as some
power n of A for larger particles, n being less than 4 and correspond-
ingly less as the particles are larger. The absorption curve of a sol
containing 6.2 g As 2 S 3 /l did not follow either of these laws but showed
a regular increase in n from n = 33 for A = 6400 to n = 12 for
A = 5300 with n = 4 at about A = 6200. The absorption thus appears
to be the resultant of two phenomena : an absorption by diffusion obey-
ing Rayleigh's law, and a selective absorption caused by the reflection
of incident rays from the surface of the particles. The latter absorp-
tion should vary with the extent of total surface, diminishing for a
constant weight of the sulfide as the size of the particles is increased.
In accord with this view, it was found that prolonged boiling, which
increases the particle size, decreases the selective absorption. Bou-
taric 41 showed further that the proportion of polarization of diffused
light from the sol decreased with increasing particle size and increased
with dilution of the sol.
Action of Light. Although an arsenic trisulfide sol is quite stable
when first prepared, the stability decreases on standing, especially in
the light. Thus Freundlich 42 observed a decrease in concentration of
approximately 15% when a sol was allowed to stand for a year in a
closed vessel. Dumanskii 48 placed a sol containing 65 g/1 in a vessel
1 m long and 2 cm in diameter, and observed the rate of fall of the
particles. The velocity was approximately 0.031 cm per day over a
period of 4 years. Since care was not taken to exclude light during
the period of observation, it is probable that the settling was due indi-
rectly to the agglomerating effect of light which resulted in the forma-
tion of particles sufficiently large to overbalance the effect of Brown-
ian movement.
Compt. rend, 177, 259 (1923) ; cf. Menon: Kolloid-Z., 78, 9 (1936).
"Boutaric and Tourneur: Compt. rend., 193, 1011 (1931); cf Lange: Z
physik. Chem., 182, 1 (1928).
Z. physik. Chem., 44, 129 (1903).
Kolloid-Z., 88, 98 (1925).
GENERAL PROPERTIES 177
The destabilizing action of light on arsenic trisulfide sol was first
observed by Young and Pingree. 44 Freundlich and Nathansohn 84
attribute this to the photochemical oxidation of the hydrolysis product
hydrogen sulfide to colloidal sulfur and pentathionic acid, accom-
panied by a reaction between the hydrogen sulfide and pentathionic
acid which serve as the stabilizing electrolytes for arsenic trisulfide
and sulfur, respectively. 45 Removal of the stabilizing electrolytes in
this way produces a decrease in the charge on the particles and the
consequent precipitation. The electrical conductivity increases to a
constant value on exposure to light, the rate of change increasing
somewhat with diminishing concentration of sol. Murphy and
Mathews 48 attribute this to the building up of a concentration of the
thionic acid sufficient to serve as the stabilizing electrolyte for the col-
loidal sulfur, the reaction between hydrogen sulfide and thionic acid
then proceeding at such a slow rate that equilibrium between the sev-
eral components of the system is maintained and further change in the
conductance is prevented. The rate of change] of conductivity in-
creases somewhat with diminishing concentration of electrolyte, prob-
ably owing to the increased photochemical activity of the sol per unit
mass of arsenic trisulfide brought about by the greater dispersity of
the more dilute sol. Pauli and Laub 25 believe that the sulfur comes
from the photochemical decomposition of sulfo complexes which they
assume to exist in the sol.
The appearance of a sol changes on exposure to light not only
because of agglomeration of particles but also because it is trans-
formed into a complex mixture. 47
Density and Viscosity. The density of arsenic trisulfide sols varies
linearly with the concentration 48 up to about 9%, above which it in-
creases more rapidly. 49 For concentrations up to 3.6 % the index of
refraction is also a linear function of the concentration. The refrac-
tive index is apparently independent of the degree of dispersion. 80
The density of the colloidal particles of arsenic trisulfide has been
**J. Phys Chem., 17, 657 (1913); cf., also, Boutaric and Maniere Bull,
acad. roy. med. Belg., (5) 10, 571 (1924)
Freundlich and Scholz: Kolloid-Beihefte, 16, 234 (1922).
*J. Am, Chem. Soc., 45, 16 (1923).
47 Cf Weiser: J. Phys. Chem., 84, 1021 (1930) ; Krestinskaja and Jakowlewa:
Kolloid-Z , 65, 187 (1933) ; Joshi, Barve, and Desai- Current Sci, 3, 105 (1934).
"Under and Picton: J. Chem. Soc, 67, 71 (1895); Wintgen: Kolloid-
Beihefte, 7, 266 (1915).
"Boutaric and Simonet: Bull, acad, roy. med. Belg, (5) 10, 150 (1924).
6 Lifschitz and Beck: Kolloid-Z,, 26, 10 (1920).
178 COLLOIDAL ARSENIC TRISULFIDE: GENERAL PROPERTIES
estimated by Dumanskii 51 by applying Einstein's 52 formula for the
viscosity of suspensions of rigid spheres,
, = , (1 + kf) (1)
where y is the viscosity of the sol, VJQ the viscosity of the pure liquid, /
the volume of the suspended particles, and k a constant which Ein-
stein at first took to be unity but which he later made 2.5. From vis-
cosity measurements, / can be determined, and, if the concentration of
the suspended particles c is known, d the density of the particles fol-
lows from the expression
'-i < 2 >
Since the particles adsorb water, the value of c cannot be determined
analytically. This is obtained indirectly from the density of the sol d t ,
the density of the medium d w , and /. Thus
but since
3=/ (2)
Taking the value of k in Einstein's equation as unity, Dumanskii
finds the density of the particles of colloidal As 2 S 3 to be 1.50 dt 0.02.
Later he makes k = 2.5 and so concludes that /, from Equation (2),
is 2.5 times larger than it should be, and so calculates d to be 3.75 .
0.12. Dumanskii seems to have completely overlooked the fact that, if
/ is 2.5 times larger than it should be in his original equation, this will
change the calculated value of c also. Had he calculated d from Equa-
tons (1) [k = 2.5], (2), and (3), as he says he did, the value of d
would come out to be 1.90. This value would appear to be much
nearer correct than 3.75, since the particles with their adsorbed water
must be less dense than the crystalline mass of As 2 S 3 whose density
is 3.45. There is, however, no reason to believe that density determi-
nations involving the use of Einstein's equation are any more than first
approximations, at best. As already mentioned, Einstein first made
" Kolloid-Z., 9, 262 (1911) ; 12, 6; 18, 222 (1913).
"Ann. Physik, (4) 19, 289 (1906); 84, 591 (1911); Kolloid-Z., 27, 137
(1920).
CHEMICAL PROPERTIES 179
k 1 and later changed it to 2.5 ; Bancelin M found k to be 2.9 for a
suspension of gamboge, and Hatschek 54 used k = 4.5 -According to
these formulas, the viscosity is independent of the degree of dispersion,
which is not the case. Thus Oden found the viscosity of sulfur sols
in which the particles have a diameter of 10 m/* to be 5Q% greater
than with sols in which the sulfur particles have a diameter of 100 nip
(Vol. I, p. 336). Hatschek 56 attributes this to the presence of an ad-
sorbed layer of water on the particles which increases the effective
volume of the smaller particles much more than the larger ones. Bou-
taric 19 finds the value of k to vary continuously with dilution, ap-
proaching 2.5 as the dilution approaches infinity.
From the density of the sol and water, together with the concen-
tration determined analytically, Dumanskii calculates the density of
the arsenic trisulfide in the particles, from Equation (3) above, to be
3. II. 57 By a similar procedure Burton and Currie 58 find the density
of arsenic trisulfide in the colloidal state to be the same as the density
in mass.
The original Einstein equation for the viscosity of a suspension of
rigid spheres has been modified by Smoluchowski * for the viscosity,
T} 8 , of a sol:
where k is the conductivity of the system; r, the radius of the par-
ticles ; D, the dielectric constant ; and , the electrokinetic potential of
the particles. Tendeloo 60 has used this equation with some success in
measuring the mean size of the particles in different arsenic trisul-
fide sols.
Chemical Properties
Ozone acts on arsenic trisulfide sol giving arsenic acid. 61 Stannous
chloride reduces it to As2S2i which is similar in color to antimony tri-
sulfide ; hence in qualitative analysis arsenic is sometimes mistaken for
"Compt. rend., 162, 1382 (1911).
"Kolloid-Z, 7,301 (1910).
Z. physik. Chem, 80, 709 (1912).
Kolloid-Z., 11, 280 (1912).
87 Cf. Wintgen: Kolloid-Beihefte, 7, 251 (1915).
88 Trans. Roy. Soc. Can., Ill 16, 109 (1922).
M Kol!oid-Z., 18, 190 (1916).
oKolloid-Z., 41, 290 (1927).
"Riesenfeld and Haase: Z. anorg. Chem, 147, 188 (1925).
180 COLLOIDAL ARSENIC TRISULFIDE: GENERAL PROPERTIES
antimony in the presence of tin. 82 Peskov 63 studied the velocity of
solution of colloidal arsenic trisulfide in sodium hydroxide or sodium
carbonate and found the process to be subject to the lyotropic influ-
ence of the cations of neutral salts in the order: NH 4 < Li <Na <
K < Rb < Cs. The velocity of solution does not take place either as
a purely heterogeneous or molecular process but appears to be a hetero-
geneous process modified by the large surface of the small particles
and by Brownian movement.
If arsenic trisulfide sol is taken into the body in any way, it is pre-
cipitated in a granular form. 84 Injected intramuscularly or subcu-
taneously it appears in the tissues as minute granular particles. On
intravenous injection, there is rapid coagulation which may result in
small emboli in the capillaries, especially of the lungs. The sulfide
produces a more marked pharmacological effect on the smallest lung
capillaries and alveoli than any other organic or inorganic compound.
Meneghetti suggests that for this reason non-lethal doses may have a
therapeutic value in some diseases of the lungs. Traces of arsenic
trisulfide sol have been found to increase liver autolysis, but higher
concentrations inhibit the process. 85
azEhrenfeld- Ber., 40, 3962 (1907).
"Kolloid-Z, 32, 163, 238 (1923).
* Meneghetti Biochcm. Z., 121, 1 (1921) ; Foa and Aggazzotti: 19, 1 (1909).
65 Ascoh and Izar: Biochem. Z., 6, 192 (1907).
CHAPTER X
COLLOIDAL ARSENIC TRISULFIDE: STABILITY OF SOL
The hydrosol of arsenic trisulfide behaves for the most part as a
typical hydrophobic sol. It therefore owes its stability primarily to
the adsorption of the potential-determining S and HS- ions de-
rived from hydrogen sulfide, the stabilizing electrolyte. This chapter
will deal with the various factors which influence the stability of the
sol, especially the effect of foreign electrolytes.
COAGULATION BY ELECTROLYTES
The Critical [-potential
The addition of electrolytes to sols usually lowers the electrokinetic
or ^-potential which is calculated from the migration velocity, u t of
the particles moving under a potential H in a medium of viscosity rj
and dielectric constant D by means of Freundlich's * equation :
Hydrosols. The concept of a critical ^-potential, above which the
sol is relatively stable and below which it coagulates rapidly, follows
from observations of Powis 2 on negative arsenic trisulfide sol. In
Table XVII are given the -potentials at the surface of arsenic trisul-
fide particles when sufficient amounts of the several electrolytes are
added to cause rapid coagulation. It is significant that the critical
potential is the same in the presence of all multivalent cations, the
coagulating ions for negative sols. For the univalent potassium ion,
the calculated -potential is much higher than for the multivalent ions.
This is not a question of the univalence of potassium ion, as evidenced
by some data of Briggs, 3 shown graphically in Fig. 29, in which the
i "Kapillarchemie," 2nd ed., 331 (1922).
*J. Chem. Soc., 109, 734 (1916).
J. Phys. Chem., 34, 1326 (1930); cf. Ghosh: J. Chem. Soc., 2693 (1929).
181
182 COLLOIDAL ARSENIC TRISULFIDE: STABILITY OF SOL
TABLE XVII
CRITICAL {--POTENTIAL OF PARTICLES IN AsiSi SOL
Electrolyte
Concentration
milliequivalents/1
{-potential
millivolts
KCl
40
44
Bad*
1
26
AlCli
15
25
Th(NO,),
Th(NOs)2
20
28
27
26
Th(NO,) 2 .
40
24
mobility, rather than the -potential calculated therefrom, is plotted
against the electrolyte concentration for potassium chloride, new fuch-
a. Potassium Chloride
20
40 60
s
x.
c. Stryc!
mine Nitrate
3
1
V
^
0.04 0.08 0.1
b. New Fuc ism Chloride
0.04 0.08 0.12
v
d. Ban
jm Chloride
V
s
Concentration of Electrolyte, Millimols per Liter
FIG. 29. Effect of electrolytes on the mobility of the particles in arsenic
trisulfide sol.
sin chloride, strychnine nitrate, and barium chloride. In this connec-
tion McBain 4 showed Equation (1) to be inadequate for calculating
* J. Indian Chem. Soc. (Ray Commemorative Volume), 67 (1933) ; cf. t how-
ever, Audubert: J. chim. phys., 30, 89 (1933) ; also, Muller : Kolloid-Beihefte, 28,
257 (1928) ; (Vol. II, p. 68).
THE CRITICAL S-POTENTIAL 183
the f -potential of the particles and suggested the wisdom of using the
M- value which gives a direct quantitative expression of all electro-
kinetic phenomena rather than the corresponding (-potential calculated
by means of an equation which does not include all the variables. As
would be expected, the mobility of the sol particles changes on
dialysis. 8
Referring to the arrows which mark the point of rapid coagula-
tion it is apparent that the univalent dye and strychnine cations, and
bivalent barium, cause coagulation at approximately the same mobility
(or -potential). On the other hand, the mobility at the coagulation
point for potassium chloride is considerably higher than that for the
electrolytes which coagulate in low concentration, and the course of
the mobility-concentration curve is S-shaped. In some cases the mo-
bility curve with salts of the KCl-type increases from the start to a
maximum and then decreases; 6 in others, it decreases from the start; T
and in still others, it decreases to a minimum and then increases. 8
With potassium ferrocyanide as precipitating electrolyte, coagulation
takes place when the mobility is higher than that of the original sol. 6
Mukherjee and coworkers 8 found differences in the form of the
curves depending on the method of preparation and the dilution of
the sol.
Powis attributed the apparently anomalous behavior of electrolytes
that coagulate in high concentrations to a salting-out effect which
causes coagulation when the calculated (-potential is higher than it is
for electrolytes which coagulate in low concentration. Kruyt e postu-
lated a change in the dielectric constant to account for the unusual
behavior of negatively charged sols toward salts with univalent cations
of the KCl-type. With weak solutions of electrolytes it is permissible
to assume that D is approximately the same as for water, but with
stronger solutions D is lower than for pure water, giving a (-potential,
Bjoshi, Barve, and Desai: Proc. Indian Acad. Sci, 4A, 590 (1936).
Kruyt and van der Willigen: Z. physik. Chem., 180, 170 (1927).
TFreundlich and Zeh: Z. physik. Chem., 114, 84 (1925); Mukherjee and
Chaudhury: J. Indian Chem. Soc, 2, 296 (1925); 4 f 493 (1927).
Mukherjee, Chaudhury, and Bhattacharya : J. Indian Chem. Soc., 0, 735
(1928); Mukherjee and Ganguly: 7, 465 (1930); Mukherjee, Chaudhury, and
Rajkumar: 10, 26 (1933) ; Mukherjee: Nature, 122, 960 (1928) ; Mukherjee and
Chaudhury: Science and Culture, 1, 111 (1935).
Kruyt, Roodvoets, and van der Willigen : Colloid Symposium Monograph,
4, 304 (1926) ; Kruyt and van der Willigen: Z. physik. Chem., 180, 170 (1927) ;
Kruyt and Briggs: Proc. Acad. Sci. Amsterdam, 82, 384 (1929); Ivanitskaja
and Proskurnin: Kolloid-Z., 89, 15 (1926).
184 COLLOIDAL ARSENIC TRISULFIDE: STABILITY OF SOL
calculated from Equation (1), that is higher than it should be. In
other words, the critical potential does not depend on a definite w-value
but on the value of u/D. In accord with this, it has been found 10
that in very dilute solutions of potassium chloride, for example, the
value of D decreases slightly, then immediately increases as the con-
centration of the solution increases, the minimum being at a concentra-
tion of 4-8 millimols per liter. This would account for the S-shape
of the mobility-concentration curve for potassium chloride, as found
by Briggs.
Mukherjee and coworkers 8 and Ghosh 11 believe that an increase
in dielectric constant alone would not account for the observed be-
havior of salts of the KCl-type with arsenic trisulfide sol unless an
improbably high value were assigned to it. The anomalous behavior
may be due in part to the stabilizing action of adsorbed chloride and
ferrocyanide ions which opposes the precipitating action of potassium
ion. It is further assumed that the force of attraction between the
particles may vary inversely with the distance according to a power
higher than that involved in the electric repulsion due to the presence
of the electric double layer. 11
Because of the anomalous effect of certain coagulating* electrolytes
on the mobility of negative sols, Mukherjee and coworkers contend
that one should not speak of a critical mobility or a critical -potential
This point of view seems a bit extreme under the circumstances. In
most cases electrolytes effect coagulation at a critical mobility or -po-
tential. The fact that we do not know enough at present to interpret
adequately the behavior with certain salts that coagulate in relatively
high concentrations is no reason to abandon the whole concept of a
critical mobility or ^-potential. Indeed, for some reason the anomalous
behavior seems to be confined to negative sols. Thus Ghosh 12 found
that a positive ferric oxide sol becomes instable when the potential
falls to about 32 millivolts using such widely different electrolytes as
potassium chloride, sulfate, oxalate, and f erricyanide ; sodium hy-
droxide; and aniline sulfate (cf. also p. 419).
Organosols. Bikerman 18 investigated the change in the ^-potential
on adding electrolytes to organosols of arsenic trisulfide in nitrobenzene
lOFurth: Physik Z., 26, 676 (1924) ; Pechhold: Ann Physik, 83, 427 (1927) ;
Walden and Werner: Z. physik. Chem., 129, 389 (1927); Hellman and Zahn-
Physik. Z. f 27, 636 (1926).
"Ghosh: J Chem Soc., 2693 (1929); cf Briggs- J Phys. Chem, 34, 1326
(1930).
J. Chem. Soc., 2693 (1929).
is Z. physik. Chem, 116, 261 (1925).
THE CRITICAL ^-POTENTIAL
185
and ethyl acetoacetate. Some observations with two ethyl acetoacetate
sols are given in Table XVIII and shown graphically in Fig. 30. It will
TABLE XVIII
EFFECT OF ELECTROLYTES ON THE {--POTENTIAL OF THE ARSENIC TRISULFIDE
PARTICLES IN ETHYL ACETOACETATE ORGANOSOL
Electrolyte
Concentration
millimols/1
^-potential
millivolts
00123
85
0036
80
FeCl,
038
110
52
53
422
40
700 (coagulation value)
30
0011
78
0113
65
Cu(CH,0 8 ) 2
220
53
660
38
900 (coagulation value)
31
134
78
1 280
46
N(C 8 H 7 ) 4 I
3 (approximate coag-
ulation value)
Fe CI 3
Cu(C 6 H 9 3 ) 2
O N(C 3 H 7 ) 4 I
0.5 1.0 3
Concentration of Electrolyte, Millimols per Liter
FIG 30. Effect of electrolytes on the ^-potential of arsenic trisulfide particles
dispersed in ethyl acetoacetate.
186 COLLOIDAL ARSENIC TRISULFIDE: STABILITY OF SOL
be seen that a given molar concentration of electrolytes decreases the
{-potential of the particles in the order: FeCl 3 > Cu(C 6 H 9 O3)2 >
N(C 3 H 7 )4l. The precipitation value (indicated by an arrow) in mil-
limols per liter for FeQ 3 is 0.7; for Cu(C 6 H 9 O 3 ) 2 , 0.9; and for
N(C 3 H 7 ) 4 I, in the neighborhood of 3. If all the values were given
in equivalents, the curve for the iron salt would be above that for the
copper salt and the precipitation value for the former would be below
that for the latter in accord with Schulze's valency rule. The critical
{-potential is in the neighborhood of 30 millivolts for all electrolytes.
A comparison of the data obtained with ethyl acetoacetate and nitro-
benzene organosols shows that the critical {-potential calculated from
Equation (1) is practically independent of the dielectric constant of
the medium. Thus the values of D for the two media differ by 130%,
whereas the difference in the calculated value of is only 2Q%. These
results furnish additional support to the view that the {-potential, in-
stead of the charge on the surface of the particles, determines sol sta-
bility (cf. p. 419).
Precipitation Values of Electrolytes
Procedure. Several methods have been employed from time to
lime for determining the relative precipitating power of electrolytes.
Since the process is transitory, it is necessary to compare the electro-
lyte concentrations which produce the same velocity of coagulation.
The rate of coagulation has been followed by means of viscosity meas-
urements on alumina sol ; 14 by colorimetric observations on gold sol ; 1B
and Congo rubin sol ; 10 with a spectrophotometer on gold sol and ar-
senic trisulfide sol ; 17 and with a turbidity meter on arsenic trisulfide
sol. 18 In an arsenic trisulfide sol, the viscosity-time curve on adding
an electrolyte such as potassium chloride first falls and then rises
rapidly as the coagulation proceeds. 19 The times for attaining a given
viscosity might be taken as a measure of the coagulating power of dif-
ferent electrolytes. For the most part, however, investigators com-
pare the concentrations which just cause complete precipitation in a
i*Gann- Kolloid-Beihefte, 8, 64 (1916).
"Hatschek: Trans. Faraday Soc, 17, 499 (1921).
"Liiers: Kolloid-Z., 27, 123 (1920).
"Mukherjee and Majumdar: J. Chem. Soc., 126, 785 (1924) ; Boutaric and
Vuillaume: Compt. rend., 172, 1293 (1921).
"Ghosh: J. Indian Chem. Soc., 9, 591 (1932).
"Joshi and Viswanath: J. Indian Chem. Soc, 10, 329; Joshi and Menon:
599 (1933).
PRECIPITATION VALUES OF ELECTROLYTES 187
definite period of time, not less than 2 hours. As Freundlich 20 has
shown, these precipitation values have a definite meaning from the
standpoint of coagulation since with certain concentrations there al-
ways exists a constant maximum coagulation velocity, the velocity of
so-called instantaneous coagulation, 21 and hence the coagulations are
comparable when this velocity is attained.
In determining the precipitation values of sols a uniform procedure
as regards stirring should be followed. Thus, if an arsenic trisulfide
sol is shaken continuously, appreciable flocking will result in a given
time with a concentration of electrolyte that will cause no flocking
whatsoever without stirring. 22 It has been observed repeatedly 2S that
the critical concentration of electrolytes does not cause agglomeration
of the particles into a clump unless the mixture is shaken. Appar-
ently what happens is that the potential on the particles is reduced to
the critical value, but, instead of agglomerating into a clump, the indi-
vidual particles with their film of adsorbed water coalesce to a loose
jelly structure that is readily broken up by stirring. If a concentra-
tion of electrolyte close to the critical value is used and the sol is
allowed to stand quietly for a day or two, the surface of the precipi-
tate as it settles appears to be a fairly strong, translucent, mobile film,
strikingly similar in appearance to that of a copper ferrocyanide mem-
brane. Since the stability of arsenic trisulfide sols decreases on aging 24
it is advisable to obtain all the values of a given series within a com-
paratively short time interval if comparable values are desired.
Schulze's Rule. Some observations of Freundlich 25 on the pre-
cipitation of arsenic trisulfide sol are given in Table XIX. Except
for the univalent organic ions, the salts with ions of the same valence
fall in fairly well-defined groups, the univalent ions precipitating in
highest concentration and the trivalent ions in lowest concentration,
in accord with the rule proposed by Schulze 26 from similar observa-
2 Kolloid-Z., 28, 163 (1918).
2 *Smoluchowski: Z. physik. Chem., 92, 129 (1917).
22 Freundlich and Basu: Z. physik. Chem, 115, 204 (1925); Freundlich and
Kroch: 124, 155 (1926); Boutaric and Vuillaume: Compt. rend., 174, 1351
(1922); cf. Wo Ostwald: Kolloid-Z, 41, 71 (1927).
28 Weiser: Colloid Symposium Monograph, 4, 369 (1926).
24 Freundlich and Schucht: Z. physik. Chem., 80, 566 (1912); Ghosh and
Dhar: Kolloid-Z., 88, 134 (1925) ; Krestinskaja and Jakowlewa: 44, 141 (1928) ;
Krestinskaja:86, 58 (1934).
2 Z. physik. Chem., 78, 385 (1910); cf., also, Schilow: 100, 425 (1922);
Bach: J. chim. phys., 18, 46 (1920).
2 J. prakt. Chem., (2) 26, 431; 27, 320 (1883).
188 COLLOIDAL ARSENIC TRISULFIDE: STABILITY OF SOL
TABLE XIX
PRECIPITATION VALUES OF ELECTROLYTES FOR ARSENIC TRISULFIDE SOL
Electrolyte
Precipitation
value
millimols/1
Electrolyte
Precipitation
value
millimol/1
KC 2 H,O 2
110
Morphine chloride
425
KCH0 2
86
Crystal violet
165
LiCl (a) ...
58 4
New fuchsin .
114
NaCl (a)
51
MgCl 2 (b)
717
KNOa (a)
50
MgS0 4
0810
KC1 (a)
49 5
CaCl 2 (6)
649
K 2 S0 4 /2
65 6
SrCl 2 (6)
635
NH 4 C1 (a)
42 3
BaCl 2 (6) .
691
HC1
30 8
Ba(NO,) 2 (6)
687
H,S0 4 /2
30 1
ZnCl 2 (&)
685
Guanidine nitrate
16 4
U0 2 (N0 8 ) 2 (&)
642
Strychnine nitrate
8
Quinine sulfate
240
Aniline chloride
2 52
A1C1, .
093
Al(NOs), ... .
095
Al 2 (S0 4 ),/2
096
Ce(NO,) 8 .
080
Ce 2 (S0 4 ),/2
092
tions on arsenic trisulfide and antimony trisulfide sols. It should be
emphasized, however, that ions of the same valence do not behave
alike and that the behavior of the alkaloid and dye cations follows no
rule as to valence. Although the exceptions to Schulze's rule are so
numerous that it should be looked upon merely as a qualitative state-
ment Freuncllich and Schucht 27 have used the principle with some
success in confirming the valence of certain ions. Thus the results
recorded in Table XX indicate the trivalence of the rare earths and of
indium. In a similar way, Galecki 28 showed that beryllium behaved
like a bivalent metal as we now know that it should. The similarity
in precipitating power of multivalent elements having the same charge
is more striking in Table XX than in Table XIX, probably because of
the marked similarity which exists among the rare-earth elements.
Burton and Bishop 29 report that eerie nitrate behaves like a salt with
a trivalent cation and not like one with a tetravalent cation. Since
"Z. physik. Chem., 80, 564 (1912).
" Z. Elektrochem., 14, 767 (1908).
wj. Phys. Chem., 24, 701 (1920).
PRECIPITATION VALUES OF ELECTROLYTES 189
TABLE XX
PRECIPITATION VALUES OF SOME LESS COMMON ELECTROLYTES FOR ARSENIC
TRISULFIDE SOL
Electrolyte
Precipitation
value
millimol/1
Valence of
cation
Strontium nitrate . .
54
2
Xantho cobalt sulfate
55
2
Purpurco cobalt chloride
55
2
Aluminum sulfate
075
3
Yttrium chloride .
073
3
Cerium nitrate
075
3
Neodymium ammonium nitrate
080
3
Praseodymium ammonium nitrate
079
3
Samarium chloride
083
3
Europium chloride
092
3
Gadolinium chloride .
080
3
Dysprosium chloride
086
3
Erbium chloride .
064
3
Indium nitrate
082
3
Luteo cobalt chloride .
082
3
Roseo cobalt chloride
120
3
Ce(NO 3 ) 4 is said to be non-existent, it is probable that they were
using the trivalent salt which is sometimes erroneously called eerie
nitrate. It is interesting that the slow transformation of bivalent pur-
pureo cobalt chloride to the trivalent roseo salt can be followed by
the gradual increase in precipitating power.
Traube's Rule in Coagulation. From investigations on the coagu-
lation of arsenic trisulfide sols by different amine salts, Freundlich 30
found the precipitating power to increase regularly with the addition of
CH 2 groups in accord with Traube's rule, which states that the capil-
lary activity, that is, the lowering of the surface tension of water and
the adsorption by solid adsorbents, increases regularly with the addi-
tion of CH 2 groups in an homologous series. This is illustrated by
observations on two different sols prepared in the same way and of
approximately the same concentration: (a) 0.475 g/1, and (b) 0.465
g/1 (Table XXI). The addition of each successive pair of CH 2
groups causes a marked falling off in the precipitation value. This
30 Freundlich and Birstein* Kolloid-Beihefte,
Slottman: Z. physik. Chem., 129, 305 (1927).
, 95 (1926) ; Freundlich and
190 COLLOIDAL ARSENIC TRISULFIDE: STABILITY OF SOL
TABLE XXI
COAGULATION OF ARSENIC TRISULFIDE SOLS BY HOMOLOGOUS AMINE CHLORIDES
Sola
Sol 6
Electrolyte
Precipitation
Activity
Precipitation
Activity
value, x
factor
value, x
factor
millimols/1
X n /X n+ i
millimols/1
X n /X n+ i
NH 4 C1 . . .
35 00
51
C 2 HNH,C1
18 20
1.9
17 5
3
(C a H.) 2 NH 2 Cl. .
9 96
1 8
5 3
3 3
(C 2 H,),NHC1
2 78
3 5
1 5
3 5
(C 2 H.)4NC1
89
3 2
85
1 8
decrease is probably not directly proportional to the number of CH 2
groups, but the variation from a constant ratio in a series is no greater
than the variation between one sol and another. Similar observations
were made on a hydrous ferric oxide sol with a series of sodium salts
of fatty acids. Also, the difficultly soluble sodium fumarate was found
to have a much higher coagulating power for ferric oxide sol than the
more readily soluble isomeric sodium malonate. This accords with
the usual view that the least soluble substance is the most capillary
active, that is, the most strongly adsorbed.
Ostwald's Activity Coefficient Rule. Because of the limitations of
the rules of electrolyte coagulation which attempt to formulate the con-
ditions for reducing the f-potential on the particles to a critical value,
Ostwald 31 introduced the principle that the dispersion medium rather
than the micelles should be in a corresponding (in the simplest case
identical) physical chemical state for coagulation to take place. It is
argued that coagulation should take place at the same activity coeffi-
cient of the precipitating ion (/+ for negative sols and /- for positive
sols), irrespective of the salt employed. For negative sols such as
arsenic trisulfide the cation activity coefficient is defined, in accordance
with the Debye-Huckel theory, by a relation of the form
- log/+ - 0.5 (s+) 2 Vu/n+
in which s is the valence of the cation, n+ the number of cations in
the molecule (this is not in the equation of Debye and Htickel), and
u is the ionic strength which is given by the expression
K = 0.5
si Kolloid-Z., 78, 301 (1935).
PRECIPITATION VALUES OF ELECTROLYTES
191
where m+ and m~ are the molar concentrations (molarities) of the
cation and anion, and 2+ and *- are the valences of the respective
ions in the coagulating electrolyte. As an illustration of the applica-
bility of the rule /+ = constant, we may examine Table XXII which
TABLE XXII
COAGULATING POWER AND ACTIVITY COEFFICIENT OF ELECIROLYTKS
Salt
type
No. of
each type
Salts considered
from Table XIX
Precipitation value
mol/1 (average)
Activity coefficient
of cation (/+)
1-1
5
Salts marked (a)
0502
77
h-2
1
K 2 S0 4 /2
0656 (0 0328)*
78 (0 83)*
2-l a
7
Salts marked (6)
000672
82
2-2
1
MgSO
000810
77
3-1,
2
A1C1 8 ; A1(N0 8 ),
000093
78
3z-2 8
1
Ce,(S04)3
000092
82
* For the precipitation value of K 2 SCV2 Ostwald used 0.0328 mol/1 instead of Freundhch's
value, 0.0656.
Ostwald constructed from Freundlich's data 88 in Table XIX. The
figures under the heading "Salt Type" are the valences of the respec-
tive ions. Other tables of a similar nature were obtained from coagu-
lation data on arsenic trisulfide sol and other sols, by various investi-
gators. A comparison of the data on arsenic trisulfide sol shows /+
to be constant within a few per cent for the work of a given investi-
gator and to vary only between 0.68 and 0.83 when applied to the work
of different investigators. This might seem to establish the general
validity of the rule; but its limitations call for special attention. In
the first place the rule is admittedly applicable only to approximately
neutral inorganic salts. Confining ourselves to Tables XIX and XXII,
we see that Ostwald has chosen to omit the simple organic salts, the
dyes, the alkaloids, sulfuric and hydrochloric acid, and eerie nitrate,
because the inclusion of any one of them would make the rule look
less satisfactory. Omitting the organic compounds may be justifiable,
but it is not obvious why some of the inorganic compounds should be
omitted. In other tables of data, sulfates of bivalent cations like mag-
nesium sulfate give values of f + which are much too low to obey the
rule. This is explained by the tendency of the ions to form complex
88 These data are taken from Freundlich: Z. physik. Chem., 78, 385 (1910)
and not 44, 135 (1903), as stated by Ostwald.
192 COLLOIDAL ARSENIC TRISULFIDE: STABILITY OF SOL
salts with arsenic trisulfide. But magnesium sulfate fits in all right in
Table XXII and is therefore included. For the five salts of the type
1-1 that are lumped together, there is a maximum variation of 16%
above and below the mean coagulation value. It is questionable
whether it is permissible to average values that vary so widely. Po-
tassium formate (0.086) and hydrochloric acid (0.031) might have
been averaged with the other five without affecting the result materi-
ally, but Ostwald very properly did not do so( cf. p. 188). In calcu-
lating the /+-value for potassium sulfate, Ostwald used 0.0328 mol/1
for the precipitation value of K 2 SO 4 /2 instead of Freundlich's value
of 0.0656 mol/1. In spite of this, the calculated /+-value was in line
with the others. One must accept with reservations any rule that will
iron out data that are in error by lOO^fc. Finally, there appears to be
no reason why the /+ -values of univalent precipitating ions should be
the same as the /+ -values of multivalent precipitating ions since only
a small percentage of univalent ions is adsorbed at the precipitation
value whereas a large percentage of most multivalent ions is adsorbed
at the precipitation value.
Based on the above principles a new valency rule was formulated
which requires that the reciprocal of the coagulating molarities for
the valence types 1-1, ! 2 -2, ! 3 -3, ! 4 -4, 2-l 2 , 2-2, 3-l 3 , 3 2 -2 3 , 3-3,
4-l 4 , 6-l 6 stand in the ratio 1 : 1.5 :2:2.5: 48:64:436:608: 729:
2560 : 27,200. That this is not in agreement with observed results may
be due in part to inaccuracies in the data but is probably due chiefly
to some neglected factors. The averages for the many investigations
on arsenic trisulfide sol agree only in order of magnitude with this set
of ratios. In certain special cases the above valency rule reduces to
an earlier rule of Freundlich and Ostwald (p. 207).
Effect of Concentration of Sol. The precipitation value of a given
arsenic trisulfide sol changes with the concentration of the sol, a fact
observed first by Mukopadhyaya. 83 Later Kruyt and van der Spek 84
showed that, on dilution, the precipitation value of potassium chloride
increased, that of barium chloride decreased slightly, and that of
aluminum sulfate decreased greatly. This was confirmed and extended
by Burton and coworkers, 20 ' 85 who formulated the rule that, in general,
the precipitation value of univalent precipitating ions increases with
the dilution; that of bivalent ions is almost constant and independent
as J. Am. Chem. Soc., 37, 2024 (1915).
3*Kolloid-Z, 26, 3 (1919); Mukherjee and Sen: J. Chem. Soc., 115, 462
(1919); Mukherjee and Papaconstantmou : 117, 1569 (1920).
85 Burton and Maclnnes: J. Phys. Chem., 25, 517 (1921).
PRECIPITATION VALUES OF ELECTROLYTES
193
of the sol concentration; and that of trivalent ions varies almost di-
rectly with the sol concentration. Weiser and Nicholas 8e showed that
Burton's rule is not generally applicable since with many sols, espe-
cially those of the hydrous oxides, the precipitation value of electro-
lytes decreases with dilution of sol irrespective* of the valence of the
precipitating ion.
The results of some observations of Weiser and Nicholas are
shown graphically in Fig. 31 in which the percentage concentration
of sol is plotted against the ratio of each precipitation value for a given
electrolyte to that of the strongest colloid (6.7 g/1). Three factors
"oo
75 50 25
Cpncentration of Sol. Percent
FIG 31 Effect of the concentration of arsenic trisulfide sol on its stability
toward electrolytes.
appear to determine the effect of dilution of sol on the precipitation
value: (a) the smaller number of particles in the weaker sol requires
less electrolyte to lower the -potential to the critical value; (b) the
decreased chance of collision requires a lower potential and hence more
electrolyte to effect coagulation in a given time ; 87 and (c) the stabiliz-
ing effect of the ion having the same charge as the sol. 88 Of these
three factors, (a) predominates with electrolytes containing strongly
adsorbed multivalent precipitating ions that cause precipitation in low
ej. Phys. Chem, 25, 742 (1921); Takamatsu: Kolloid Z., 88, 229 (1926);
Rabinovich and Dorfmann, Z. physik Chem., 181, 313 (1927); Boutaric and
Perreau: J chim. phys, 24, 498 (1927); cf. Pauli and Laub: Kolloid-Z., 78,
295 (1937).
a'Kruyt and van der Spek: Kolloid-Z., 26, 3 (1919).
as Weiser and Nicholas: J. Phys. Chem., 26, 742 (1921).
194 COLLOIDAL ARSENIC TRISULFIDE: STABILITY OF SOL
concentrations; (b) can be disregarded as a rule; and (c) is of pri-
mary importance if the precipitation concentration is high, as is usual
with univalent precipitating ions. The tendency for the precipitation
value of electrolytes with univalent cations to increase with dilution is
indicated by a rise in* the mobility of the particles as a given sol is
diluted. This effect cannot be due to arsenious acid resulting from
increased hydrolysis of arsenic trisulfide in the more dilute sol since the
addition of arsenious acid has a sensitizing effect by lowering the
mobility. 89 Rossi and Marescotti 40 attribute the difference in the be-
0.90
J0.80
50.70
0.60
10 20
Concentration of Sol, Grams per Liter
FIG. 32. Influence of the concentration of arsenic trisulfide sol on the activity
coefficient of cations at the precipitation value.
havior of arsenic trisulfide sol from that of ferric oxide sol to the
greater increase in the degree of dispersity which the former under-
goes on dilution. 41
Ostwald 42 has applied his activity coefficient rule to the effect of
dilution on the coagulation value. The /+ -values were calculated from
the molar precipitation values for various dilutions of sol and these
were plotted against the sol concentrations. As an illustration, Ost-
wald's data calculated from Burton and Maclnnes' observations on a
sol containing initially 27 g As 2 S3/l, are given in Fig. 32. The data
89 Mukherjee and Ganguly: J. Indian Chem. Soc, 7, 465 (1930) ; Mukherjee,
Chaudhury, and Palit: 10, 27, 713 (1933); Mukherjee: Kolloid-Z., 68, 159
(1930).
Gazz. chim. ital., 09, 313 (1929); 60, 993 (1930).
41 C/., also, Gerasemov and Urzhumskii : J. Russ. Phys.-Chem. Soc. f 61,
393 (1929).
Kolloid-Z., 76, 39 (1936) ; Wannow and Hoffmann: 80, 294; Ostwald: 304
(1937).
PRECIPITATION VALUES OF ELECTROLYTES 195
for aluminum chloride do not fit in for some unknown reason ; those
for eerie nitrate do not fit for the obvious reason that there is no such
compound as Ce(NO 3 ) 4 (cf. p. 189). The recorded values for lan-
thanum sulfate were multiplied by 2 on the presumption that Burton
and Maclnnes had worked with an equivalent instead of a molar solu-
tion. From these and similar findings with more dilute sols used by
Kruyt and van der Spek and Weiser and Nicholas, Ostwald concluded
that the activity coefficient rule is a limiting rule which holds only at
high concentrations of sol. Hence the course of the precipitation
value-sol concentration curve is always determined by the magnitude
of the increasing divergence from the constant /+-value on dilution of
sol. Since the /+ -values diverge so widely with a sol as dilute as
3.63 g As 2 S 3 /l (Fig. 32) it is obvious that the fairly constant /+-
values recorded in Table XXII are quite accidental since Freundlich's
sol contained but 1.8 g/1.
In an important investigation, the results of which were published
while this volume was in press, Wannow and Hoffmann 42 determined
the precipitation value of KC1, MgQ 2 , and LaQ 3 on arsenic trisulfide
sols varying in concentration from 18.7 to 0.018 g As 2 S 3 /l. The /+-
concentration of sol curves for the three cations were found to cut
each other at a concentration of 4.7 g As 2 S 3 /l. The theoretical sig-
nificance of Wannow and Hoffmann's results were discussed by
Ostwald. 42
Effect of Anions on the Precipitation Value. Although it is often
assumed that one is justified in neglecting the effect of anions in the
precipitation of negative sols and of cations in the precipitation of
positive sols, it is evident that this will be true only if the adsorption
of the ion having the same charge as the sol the stabilizing ion is
negligible either because of its nature or the dilution. Although this
effect may be slight in certain cases it is quite marked in others as
illustrated by some observations of Freundlich, 48 Ghosh and Dhar, 44
and Weiser and Nicholas 88 summarized in Table XXIII. Pauli and
Valko 45 suggest that the activity rather than the concentration of
potassium ion in the several salts should be compared. This has been
done by Ostwald, and his /+ -values are included in the table. These
values are about the same with the chloride and sulfate but they are
lower with the ferrocyanide, indicating that the anion may influence
the precipitating power of an electrolyte independent of its effect on
Z. physik. Chem., 44, 152 (1903).
"Kolloid-Z, 86, 129 (192S).
"Elektrochemie der Kolloide," 193 (1929).
196 COLLOIDAL ARSENIC TRISULFIDE: STABILITY OF SOL
the activity of the cation. With potassium citrate containing the
strongly stabilizing citrate ion, the calculated /^-values are much
lower; but it is probably not permissible to apply the Debye-Huckel
theory to this salt.
TABLE XXIII
PRECIPITATION VALUE OF POTASSIUM SALTS FOR ARSENIC TRISULFIDE SOLS
[Precipitation values (*) in mol/1 of cation]
Sol 1
Sol 2
Sol 3
Electrolyte
1 8g/l
3 7 g/1
5 Og/1
X
/*
X
f +
X
f +
KC1
0495
77
085
72
0332
81
K a S0 4 /2 . ...
0656
78
100
73
00435
82
K 4 Fe(CN)./4, .
185
68
0712
79
K,C,H,0 7 /3
2400
63
270
62
Mukherjee and Chaudhury 46 estimated the time required for the
same concentration of various salts to make the sol so cloudy that a
heated filament a definite distance away was just obscured. From the
results obtained, it was concluded that the influence of anions is negli-
gible except in the case of the so-called complex ions like bcnzoate,
ferrocyanide, and salicylate which exert a marked effect. Actually,
their data show the precipitating power of iodate and nitrate to be less
than that of chloride and bromide. Oxalate admittedly has a lower
precipitating power than sulfate which can hardly be attributed to dif-
ference in the complexity of the ions. It appears, therefore, that
Mukherjee's observations support rather than refute the contention
that the effect of the anion cannot be neglected, even though his pro-
cedure probably gives a less accurate measure of precipitating power
than carefully determined precipitation values. Following the same
general method of procedure Ostwald 47 demonstrated the effect of
anions by determining the rate of precipitation of various organic and
inorganic acids as a function of their />H values. The />H-time curves
do not coincide, indicating that the anions have varying stabilizing
effects or that the different ions or molecules have varying dehydrating
powers.
J. Chem. Soc., 126, 794 (1924).
Kolloid-Z., 40, 201 (1926); cf. Boutaric and Perreau: Bull. sci. acad. roy.
Belg., 14, 666 (1928).
PRECIPITATION VALUES OF ELECTROLYTES 197
Acclimatization. Arsenic trisulfide sol like hydrous ferric oxide
sol (Vol. II, p. 72) requires less electrolyte to cause precipitation when
added all at once than when added stepwise through a relatively long
interval of time. 48 For example, 40 on rapid addition, 2.05 cc of 0.02 N
SrCl 2 and 2.50 cc of 0.5 N KC1 were required to coagulate 20 cc of
As 2 S3 s l whereas, on slow addition of the electrolytes in 0.1 -cc
portions over a period of approximately 36 hours, the amounts re-
quired were 2.50 and 2.70 cc, respectively. This phenomenon is known
as "acclimatization," meaning that the colloid becomes acclimated to its
surroundings when the electrolyte is added slowly so that more is re-
quired to produce a given result. In the author's opinion the term is
a misnomer. Sols which exhibit this behavior usually undergo frac-
tional precipitation to a certain extent during the stepwise addition of
electrolyte. 60 This fractional precipitation not only removes electrolyte
from solution by adsorption but also alters the stability of the sol by
decreasing the concentration. From this point of view, the excess of
electrolyte required for a given slow rate of precipitation is determined
by (o) the extent to which the sol undergoes fractional coagulation,
(b) the adsorbing power of the precipitate, (c) the adsorption of the
precipitating ions, and ( d) the effect of dilution of sol on the precipita-
tion value (see above). Krestinskaja and Jakowlewa 51 attribute the
so-called acclimatization of arsenic trisulfide sol toward barium chlo-
ride to irreversible hydrolysis of the salt giving arsenious acid and
hydrogen sulfide which react with barium to give slightly dissociated,
slightly soluble salts in larger amounts the more slowly the electrolyte
is added. The objection to this explanation is that it will not hold
for potassium chloride.
Charge Reversal: Irregular Series. Electrolytes with strongly ad-
sorbed ions which precipitate in low concentrations frequently reverse
the charge at concentrations above the precipitation value, and at still
higher concentrations, precipitate the reversed sol once more (Vol. I,
pp. 88, 126, 191). Charge reversal and the phenomenon of the so-called
irregular series were observed by Lottermoser and May 52 with arsenic
trisulfide sol using aluminum chloride, ferric chloride, and thorium
*s Freundlich : Z physik. Chem , 44, 143 (1903); cf. Dumanskii and Solin:
Kolloid-Z, 59, 314 (1932).
Weiser: J. Phys. Chem, 26, 399 (1921).
BO Cf. Burton and Annetts- J. Phys. Chem., 86, 48 (1931); Colloid Sym-
posium Monograph, 8, 48 (1931).
Kolloid-Z, 44, 141 (1928); cf Krestinskaja: 74, 45 (1936).
B * Kolloid-Z., 68, 168 (1932).
198 COLLOIDAL ARSENIC TRISULF1DE: STABILITY OF SOL
nitrate as precipitating electrolytes. Boutaric and Semelet M failed to
get a charge reversal with thorium nitrate, probably because they used
too strong a sol. It is necessary for charge reversal not only for the
reversing ion to be strongly adsorbed but also for the coagulation
velocity to be sufficiently slow to avoid coagulation before the reversal
takes place. Since the velocity of coagulation drops with dilution, it
is possible to get charge reversal without coagulation with a dilute sol
and coagulation without charge reversal with a more concentrated
one.
Freundlich and Buchler B4 studied the charge-reversing property of
a number of dyes and alkaloid hydrochlorides that have a high pre-
cipitating power on arsenic trisulfide sol. The results of some of
their observations are shown in Table XXIV, in which P- is the
precipitation values of the dyes and alkaloids for the original negative
TABLE XXIV
CHARGE REVERSAL OF ARSENIC TRISULFIDE SOL BY DYES AND ALKALOIDS
Basic dye or alkaloid
P-
Cw
P +
02
1
9 5
Malachite green
Methylenc blue
Chrysoidin
Crystal violet . .
05
10
0.25
1 3
1
2
2 5
3
35
40
3
3
006
06
30
007
10
15
004
12
5
010
26
5.0
010
24
4.0
sol, Cmo* is the concentration which gives the maximum positive {-po-
tential for the reversed sol, and P+ is the precipitation value of sodium
or potassium chloride for the reversed sol, all in millimols per liter.
In general the most strongly adsorbed cations give a reversed sol with
the highest positive {-potential and with the greatest stability. Ost-
wald " has attempted to apply "activity coefficient rules" to the charge-
reversal process with limited success.
63 Boutaric and Semelet: J. chim. phys., 26, 195 (1929).
"Kolloid-Z., 82, 305 (1923).
"Kolloid-Z., 76,297 (1936).
ACTION OF NON-ELECTROLYTES 199
It is of interest that the charge on arsenic trisulfide organosols can
be reversed like that on hydrosols by adding an excess of a salt with a
strongly adsorbed cation. 56 Thus a nitrobenzene sol containing 0.6 g
As 2 S 3 /l is precipitated by 0.11 millimol FeQ 3 /l; but 2.8 millimols
FeQ 3 /l gives it a charge of +55 millivolts.
In the light of the above observations and the further one that
stable sols may be obtained in highly concentrated electrolytes (p. 174),
it would follow theoretically that three zones of stability and three
zones of instability may exist for a given sol with a single electrolyte
in varying concentrations. 57 *
Action of Non-electrolytes
The addition of non-electrolytes to sols usually decreases their sta-
bility toward electrolytes. The sensitization is attributed by Ostwald 8
and Cassuto 59 to a decrease in the dielectric constant of the medium.
In line with this, Keeser 60 found arsenic trisulfide sol to be stabilized
by the addition of urea and glycocoll both of which increase the dielec-
tric constant of water. 61 Freundlich 62 suggests that the sensitization
results from the lowering of the charge by adsorption on their surface
of the organic non-conductor, which has a dielectric constant appre-
ciably lower than that of water. Thus, the charge e on a particle is
given by
d
where is the potential difference of the double layer at the surface
of a spherical particle of radius r, D the dielectric constant, and d
the thickness of the double layer. From this, it follows that a decrease
in D will lower e and, hence, the precipitation value.
The dielectric constant of the non-electrolyte is not the only factor
which influences the stability since sugar apparently acts as a stabilizer
for sols 68 and one is not justified in ascribing the difference between
alcohol and sugar to the dielectric constant. That sugars are adsorbed
"Bikerman: Z. physik. Chem., 115, 261 (1925).
"Ostwald and Wannow: Kolloid-Z., 76, 159 (1936).
""Grundriss der Kolloidchemie," 441 (1909).
W'Der Kolloide Zustand der Materie," 152 (1913).
<"Biochem. Z., 157, 166 (1925); cf. Ghosh and Dhar: J. Phys. Chem., 29,
668 (1925).
"Furth: Ann. Physik, 70, 63 (1923).
2"Kapillarchemie," 2nd ed., 637 (1922).
w Ghosh and Dhar: J. Phys. Chem., 29, 668 (1925).
200 COLLOIDAL ARSENIC TRISULF1DE: STABILITY OF SOL
by sulfide sols is evidenced by a loss in rotatory power when they
are dissolved in the sols instead of in water. 6 * In certain cases, the
variation in adsorption with concentration of sugar can be formulated
quite accurately by the Freundlich equation. 85
Kruyt and van Duin aa found the effect of non-electrolytes on the
precipitation value of electrolytes for arsenic tri sulfide sol to be deter-
mined by the nature of the precipitating ion. with univalent and tri-
valent ions, it was lowered ; and with bivalent ions it was raised. For
any given electrolyte, the change in precipitation value was independent
of the dielectric constant of the non-electrolyte employed. The effect
of the non-electrolyte is determined to some extent by the amount
present. For example, if one dilutes an arsenic trisulfide hydrosol
with propyl alcohol or the propyl alcosol with water, the stability of
the sol toward calcium nitrate is greatest when the propyl alcohol
content is approximately 25% * 7
From a study of this behavior in the author's 68 laboratory the sen-
sitizing action of non-electrolytes was attributed, at least in part, to the
displacing of a stabilizing ion or the cutting down of the adsorption
of a precipitating ion. These two actions are antagonistic, and so the
precipitation value may be increased, decreased, or remain unchanged
in the presence of a non-electrolyte. It was found, for example, that
the adsorption of barium ion is cut down by the presence of phenol and
the precipitation value is increased by the presence of phenol. This
means that the sol is sensitized in the sense that less barium must be
adsorbed to reduce the potential to the coagulation point; the higher
precipitation value in the presence of phenol results from the cutting
down of the adsorption of barium by the non-electrolyte.
In general the influence of foreign non-electrolytes on the stability
of lyophobic sols can be accounted for at least qualitatively by con-
sidering: 89 the effect on (1) the dielectric constant of the medium,
(2) the viscosity, (3) the degree of ionization of the electrolytes pres-
ent, and (4) the selective adsorption of the precipitating and stabilizing
*Bhatnagar and Shrivastava: J. Phys. Chem., 28, 730 (1924).
"Prasad, Shrivastava, and Gupta: Kolloid-Z, 87, 101 (1925)
Kolloid-Beihefte, 5, 270 (1914) ; cf., also, Boutaric and Semelet: Rev. gen
colloides, 4, 268 (1926); Lachs and Chwalinski: Z. physik. Chem., A169, 172
(1932).
"Bikerman: Kolloid-Z., 42, 293 (1927); Janek and Jirgensons: 41, 40
(1927).
fl8 Weiser: J. Phys. Chem., 28, 1254 (1924) ; cf., also, Janek and Jirgensons-
Kolloid-Z., 41, 40 (1927).
Weiser and Mack: J. Phys. Chem., 84, 101 (1930).
KINETICS OF THE ELECTROLYTE COAGULATION 201
ions by the dispersed particles. All these factors are measurable and
are known to have more or less influence on the stability. Chaudhury 70
believes that the change in interfacial tension between the dispersed
particles and the surrounding medium will have an important bearing
on the tendency of the particles to coalesce. Although this may be
true, it appears inadvisable to attach too much importance to a factor
that cannot be measured, until it has been demonstrated that the above-
mentioned measurable factors are inadequate to explain the observed
facts.
Keeser 71 found that salts of cholesterol and lecithin sensitize ar-
senic trisulfide sols toward electrolytes. The activity of the sols of
cholesterol esters in cutting down the stability of negative sols is re-
duced as the length of the carbon chain in the homologous scries is
decreased. The sensitizing action of the potassium and sodium salts
of cholesterol sulfuric acid depends on the solubility of the salt. The
constant ratio in the organism between free cholesterol and its esters
may have an important bearing on the physical state of the body col-
loids. If poisons are removed from the body by combining with free
cholesterol, such as happens in the removal of saponin, the normal
physical condition of the cell colloids could be disturbed in the event
that the resulting product has the properties of a cholesterol ester. This
point of view may serve to account for the fact that an increase in
the amount of cholesterol in the organism is always followed by an
increase in the fatty acid phosphatides.
Kinetics of the Electrolyte Coagulation
The velocity of the slow coagulation of arsenic trisulfide sol has
been followed by means of both a spectrophotometer and an opacity
meter. 72 Using the former technique Mukherjee and Majumdar 73
found that Smoluchowski's theory (Vol. I, p. 89) applies only in the
initial coagulation process, a limiting stage of coalescence being ulti-
mately reached where the equations fail completely. This stage was
reached more rapidly with lower concentrations, when the coagulation
rate was slower. The results were explained by assuming, as Freund-
lich 74 does, that the coalescence is reversible or irreversible according
to the magnitude of the ^-potential of the particles. At sufficiently low
j. Phys. Chem., 82, 1485 (1928).
'iBiochem. Z., 154, 321 (1924); 167, 166 (1925).
72 Dumanskii and Shershnev: J. Russ. Phys.-Chem. Soc., 62, 187 (1930).
7 J. Chem. Soc., 126, 785 (1924).
7 *Kolloid-Z., 28, 163 (1918); cf. Kruyt and van Arkel: 82, 29 (1923).
202 COLLOIDAL ARSENIC TRISULFIDE: STABILITY OF SOL
concentrations of electrolyte it is reversible, but is irreversible at higher
concentrations when the {-potential is small. The limiting stage results
when there is an equilibrium between the rates of breaking up of the
aggregates and of coalescence. From similar experiments on arsenic
trisulfide sol, using potassium chloride as precipitating electrolyte,
Jablczynski 75 first obtained results for slow coagulation in good agree-
ment with Smoluchowski's theory, provided the sol was not freed from
hydrogen sulfide; but later 78 he found that the Smoluchowski equation
had to be modified in order to make it applicable. 77
The rate of coagulation of arsenic trisulfide sol is decreased by both
visible and ultraviolet light ; but the velocity is speeded up if a fluores-
cent substance like fluoresceine or eosin 78 is added to the sol before
exposure to light. The effect is due to the fluorescence of the added
dye since other colored compounds which do not fluoresce are without
influence.
For every sol there appears to be a critical temperature above which
it will coagulate (Vol. I, p. 190). The addition of increasing amounts
of electrolytes not only lowers the critical temperature of stability but
also greatly increases the velocity of coagulation at high temperatures. 79
Mechanism of the Coagulation Process
Adsorption of Cations. The adsorption of cations by the arsenic
trisulfide particles during the coagulation of sol was first reported by
Linder and Picton. 80 Later Whitney and Ober 81 investigated the
phenomenon and claimed that equivalent amounts of various cations
were adsorbed by the same sol. This work was extended by Freund-
lich, 82 who at first observed an equivalent adsorption of ions of varying
valence even when different sols were employed. Freundlich and co-
workers 88 found a similar relationship with hydrous oxide sols
"Bull soc chim., (4) 86, 1277 (1924).
Jablczynski: Kolloid-Z., 64, 164 (1931).
"C/., also, Joshi and Prabhu. J. Indian Chem. Soc., 8, 337 (1931).
"Boutaric and Bouchard: Compt. rend, 192, 95 (1931); Bull. soc. chim.,
(4) 61, 757 (1932); cf. Pospelov and Pospelova: J. Phys. Chem. (U.S.S.R.),
6, 52 (1934)
Burton, Deacon, and Annetts: Trans. Roy. Can. Inst, (4) 18, pt. 1, 33
(1931).
ao J. Chem. Soc., 67, 64 (1895).
"Whitney and Ober: J. Am. Chem. Soc., 23, 842 (1901).
sa Kolloid-Z, 1, 321 (1907); Z. physik. Chem., 78, 408 (1910).
83 Freundlich and Ishizaka: Kolloid-Z., 12, 232 (1913); Gann: Kolloid-
Beihefte, 8, 73 (1916).
MECHANISM OF THE COAGULATION PROCESS
203
which led Freundlich to conclude that, in general, equivalent amounts
of varying precipitating ions are adsorbed at the precipitation value
with both positive and negative sols. This was not confirmed by the
author with hydrous oxide sols 8 * even for ions of the same valence ;
and, with arsenic trisulfide sols, 85 the adsorption values for the alkaline-
earth cations were found to be similar but not identical for the same
sol and to vary widely with different sols. Freundlich reached the same
conclusion as the result of his most recent investigations. 86 The ad-
sorption values obtained by the various investigators are collected in
Table XXV. Although the tabulated results show considerable varia-
TABLE XXV
ADSORPTION OF CATIONS DURING THE PRECIPITATION OF ARSENIC TRISULFIDE SOL
Cation
Cone,
of sol
g/1
Adsorp-
tion
m.eq./g
Observer
Cation
Cone,
of sol
g/1
Adsorp-
tion
m.eq./g
Observer
Ba
6 4
116
We ser *
0.074
Freundlich
Sr....
6 4
107
New fuchsin
1 1
076
ii
Ca
6 4
093
UO, . .. .
4 14
0088
i
Ba
11 8
060
Ce . .
4 14
069
i
Sr.
11 8
056
Ba
1.55
21
Freundlich ||
Ca
11 8
043
Zn .
1 55
1 50
Ba .
21 5
072
Ni . ...
1 55
1 22
Sr
21 5
069
it
In..
56
Ca .
21 5
073
ii
Te .
037
Ba
10
110
Whitney and
Th
.
27
Obert
New fuchsin
0.08
Sr
10
0082
ii
Methylene
Ba ..
10.0
100
<
blue
13
Ba
3 33
086
Linder and
Picton i
* J. Phys. Chem., 29, 955 (1925).
f J. Am. Chem. Soc.. 23, 842 (1901).
t J. Chem. Soc., 67. 64 (1895).
i Kolloid-Z., 1, 321 (1907); Z. phytilc. Chem., 73, 408 (1910).
|| Freundlich, Joachimaohn, and Bttisch: Z. phyaik. Chem., A141, 249 (1929)
tion from equivalent adsorption, it should be pointed out that the
maximum variation does not exceed 20% until one reaches the most
"Weiser and Middleton: J. Phys. Chem,, 24, 53, 648 (1920).
Weiser: J. Phys. Chem., 29, 955 (1925).
Freundlich, Joachimsohn, and Ettisch: Z. physik. Chem., A141, 249 (1929).
204 COLLOIDAL ARSENIC TRISULF1DE: STABILITY OF SOL
recent data reported by Freundlich. Unfortunately these data are
probably the least accurate. Since Freundlich worked with a dilute
sol and used potentiometric methods where such methods have not
been accepted, it is probable that the results for the adsorption of zinc,
nickel, and indium are inaccurate and should have been omitted from
the table as were the values for sodium and hydrogen.
There are two main reasons why the adsorption values are not
equivalent at the precipitation value: (1) even if the adsorption of
equivalent amounts lowers the potential to the critical value, adsorption
by the agglomerating particles takes place in varying amounts depend-
ing on the nature and concentration of the electrolyte; (2) less of a
strongly adsorbed ion needs to be adsorbed in order to reduce the
potential to the critical value. 87
Neglecting the effect of the stabilizing ions of electrolytes, the
precipitating ions which are most readily adsorbed will precipitate in
the lowest concentration. From this it follows that the greater pre-
cipitating power of ions of higher valence results from their relatively
greater adsorbability. Moreover, among ions of the same valence, the
one with the greatest adsorption capacity will effect precipitation in
the lowest concentration. Finally, the variation among electrolytes
with the same precipitating ion and different stabilizing ions is caused
in part by differences in the adsorbability of the latter.
Let us consider an ideal case fulfilling the following requirements
(1) the amounts of precipitating ions which must be adsorbed to lower
the -potential to the critical point are equivalent and depend only on
the number of charges they carry; and (2) the effect of the stabilizing
ion is a logarithmic function of its concentration, that is, follows the
usual adsorption isotherm. Now if the amounts adsorbed in mols of
the several ions which are proportional to the reciprocal of the valence
are plotted against the precipitation concentration in millimols per liter,
a curve is obtained like that shown in Fig. 33.
As is well known, a much higher molar concentration of a univalent
ion is necessary to effect coagulation than of a multivalent ion. It is
unnecessary to assume with Freundlich that the equivalent amounts
are adsorbed from equimolar solutions. The isotherm of Fig. 33 is
the curve of equivalent adsorption necessary for decreasing the poten-
tial to the critical value and must not be confused with the total amount
that may be adsorbed on coagulation.
One might expect to approach most nearly the ideal conditions
"Weiser: J. Phys. Chem., 86, 1, 1368 (1931); Weiser and Gray: 86, 2178
(1932).
MECHANISM OF THE COAGULATION PROCESS
205
referred to in the preceding paragraph by working with strongly ion-
ized salts having a common stabilizing ion and precipitating ions vary-
ing in valence but so similar chemically that the adsorption in mols
will be proportional to the reciprocal of the valence. The common
stabilizing ion will influence the adsorption of the precipitating ion
Concentration, Millimols per Liter
FIG 33. Adsorption isotherm for cations of varying valence (diagrammatic).
logarithmically in proportion to the concentration. Such conditions are
actually realized approximately in the precipitation of arsenic trisulfide
sol by cobalt amines with cations varying in valence from 1 to 6. The
average precipitation values for a series of such salts as observed by
Matsuno 8S are given in Table XXVI. In these experiments a rela-
TABLE XXVI
PRECIPITATION VALUES OF COBALT AMINES FOR ARSENIC TRISULFIDE SOL
Precipitation values (average)
Valence of
No. of salts
millimols/1
cation
examined
Observed
Calculated
1
7
5330
5130
2
5
360
333
3
6
66 6
66 6
4
4
18 8
22
6
1
4 2
4 3
* J. Coll. Sci. Imp. Univ. Tokyo, 41, Jfll, 1 (1921).
206 COLLOIDAL ARSENIC TRISULFIDE: STABILITY OF SOL
tively small amount of sol was coagulated by an excess of cobalt salt so
that the amount adsorbed may be neglected in comparison with the
precipitation value. In Fig. 34 the precipitation values are plotted
against the reciprocal of the valence, giving an isotherm similar to the
diagrammatic one. In the same figure are plotted also the logarithms
of the precipitation values against the logarithms of the re-
ciprocal of the valence. An approximately straight line is ob-
1.00
Precipitation Concentration of Electrolytes, Millimols per Liter
.0 1000 2000 3000 4000 5000 6000
1 2 3
Log Precipitation Concentration of Electrolytes
FIG. 34. Relationship between the valence of cobalt amine cations and the
precipitation concentrations for arsenic trisulfide sol.
tained, from the slope of which one may calculate the precipitation
values and compare with the observed values 89 (see Table XXVI).
Freundlich and Zeh 90 confirmed the observations of Matsuno, by a
study of the effect of the various cobalt amines on the cataphoretic
mobility of arsenic trisulfide. Similar observations on hydrous ferric
oxide in the presence of complex cyanide ions of varying valence re-
vealed as usual a specific adsorption factor in addition to valence. Thus
the precipitating power of Au(CN)4~ appears to be twelve times as
great as Au(CN) 2 ~ and that of Fe(CN) e only twice as great as
Pt(CN) 4 .
If the adsorption theory as outlined by Freundlich were generally
applicable then, as Ostwald 91 points out, the following relationship be-
89 Cf Freundlich: Kolloid-Z., 31, 243 (1922).
Z. physik. Chem., 114, 65 (1924).
'i Kolloid-Z., 26, 28, 69 (1920).
MECHANISM OF THE COAGULATION PROCESS 207
tween the valence and the precipitating power of an ion should hold :
....._!. 2.. a....
r* s, /* i . z o
Cl C2 63
where C lf C 2 , and C 3 are the precipitation values of electrolytes with
univalent, bivalent, and trivalent precipitating ions and n is a constant.
This recalls the older formula of Whetham : 92
1.1.1... = i:*. se 2...
Ci 2 3
where the C's have the same significance as above and x is a constant.
Formulas of this kind fit the experimental data so roughly that they
are of questionable use (p. 192).
Titration of Sols. Whitney and Ober 81 observed that barium, but
not chloride is adsorbed during the precipitation of arsenic trisulfide
sol with barium chloride and that hydrochloric acid equivalent to the
adsorbed barium is present in the supernatant solution. This was
believed to be a case of hydrolytic adsorption, but Rabinovich 9 ? showed
it to be an exchange adsorption in which hydrogen ions in the outer
portion of the double layer are exchanged for barium ions. The hydro-
gen ion concentration of the sol may be determined conductomctri-
cally, 94 but the method is of doubtful accuracy especially in sol-electro-
lyte mixtures. Weiser and Gray 05 adopted the more satisfactory pro-
cedure of potentiometric analysis with the glass electrode. Determina-
tion of the change in hydrogen ion concentration on the stepwise addi-
tion of coagulating electrolyte has been called "titration of sol." The
method was as follows: To 10-cc portions of sol containing 10 g
As 2 S 3 /l were added, in a mixing apparatus, definite amounts of elec-
trolytes diluted to 5 cc. The pH. value of each sol was first determined,
after which the pU of the water and electrolyte were brought to this
value with hydrochloric acid ; thus, the observed displacements of hy-
drogen ion were real and not due to dilution effects. A typical set of
data using barium chloride is given in Table XXVII and presented
92 Phil. Mag., (S) 48, 474 (1899); cf. Robertson: "Die physikalische Chemie
der Proteine," Dresden, 94 (1911).
*Z. physik. Chem., 116, 97 (1925).
w Pauli and Semler: Kolloid-Z., 84, 145 (1924); Rabinovich: Z. physik.
Chem., 116, 97 (1925) ; Rabinovich and Dorfmann: 181, 313 (1928).
9 J. Phys. Chem., 86, 2796 (1932).
208 COLLOIDAL ARSENIC TRISULFIDE: STABILITY OF SOL
graphically in Fig. 35 together with the titration curves for the chlo-
rides of strontium, calcium, and aluminum. The coagulation point for
each electrolyte is indicated by an arrow.
2 4 60 2 4 6
0.004 AT Electrolyte Added, Cc
FIG. 35. Titration curves of arsenic trisulfide sol with various electrolytes.
The adsorption values for the cations above the precipitation value
were determined on 100-cc portions of sol to which were added definite
amounts of electrolyte diluted to 50 cc, thus giving data comparable to
the titration data.
MECHANISM OF THE COAGULATION PROCESS
209
TABLE XXVII
TlTRATION OF ARSENIC TRISULFIDE SOL WITH BARIUM CHLORIDE
004 tfBaCl* added
cc
[H +]X10' in
solution
[II + ] X 10
displaced
Electrolyte added
equivalents X 10*
43 7
5
52 5
8 8
13 3
1
57 5
13 8
26 7
1 5
63 1
19 4
40
2
69 2
25 5
53 3
2 5
70 8
27 1
66 7
3
72 5
28 8
80
5
74 1
30 4
133 3
A set of curves showing the displacement of hydrogen ion by
cations of varying valence is given in Fig. 36. The marked displacing
action of aluminum ion, especially at low concentrations, is in line with
the higher precipitating power of aluminum chloride, whereas the rela-
234
0.004 N Electrolyte Added, Cc
FIG 36. Displacement of hydrogen ion from arsenic trisulfide particles by
cations of varying valence.
tively weak displacing action of ammonium ion is in accord with the
relatively low precipitating power of ammonium chloride. It will be
noted that the ratio of the precipitation value, in millimols per liter, of
barium chloride to aluminum chloride is 2.1 in the sol under considera-
tion containing 10 g/1 as compared with 7.4 in the sol containing but
210 COLLOIDAL ARSENIC TRISULFIDE: STABILITY OF SOL
1.8 g/1 (Table XIX). This supports the rule that the precipitation
value varies approximately as the concentration of the sol for trivalent
ions and is almost independent of the concentration of the sol for
bivalent ions. Obviously, precipitation rules which presume to give
the ratio of the precipitating power of ions of varying valence (p. 207)
must take into account the way in which the precipitation value is in-
fluenced by the concentration of sol.
The curves in Fig. 35 show that, in the sol under consideration,
somewhat more than one-third of the total hydrogen ion in solution
after coagulation was displaced from the particles, the remainder being
measurable potentiometrically in the original sol. The adsorption of
the several cations was approximately twice as great as the hydrogen
displaced from the diffuse outer layer, meaning that a portion of the
adsorbed cation corresponds to hydrogen ion in the outer layer that is
measurable in the original sol. The adsorption was from 14 to 22%
less than the total hydrogen ion in solution after coagulation. It was
a fortuitous combination of circumstances that gave Whitney and Ober
a filtrate having the same amount of hydrogen ions as the amount of
barium adsorbed.
Rabinovich 96 concluded from conductometric measurements that
equivalent amounts of barium, say, are adsorbed and hydrogen ion dis-
placed in the first stage of the coagulation process, which is followed
by a second stage, the agglomeration of the particles. It is probable
that the observed equivalence between adsorption and hydrogen ion
displacement results from the limitations in his experimental method.
In any event only the part of the counter ions in the diffuse outer
layer which are not measurable in the original sol can be displaced.
What Rabinovich doubtless means is that barium goes into the outer
layer in exchange for an equivalent amount of hydrogen ; but this could
be determined definitely only by measuring the />H of the sol and of
the ultrafiltrate from the sol after the addition of electrolyte, as Ver-
wey and Kruyt did with silver iodide sol (p. 120) . Rabinovich is con-
cerned because precipitation does not follow immediately after the
completion of the exchange adsorption. As a matter of fact, depending
on the conditions the coagulation may take place before, after, or at
the point where the displacement of hydrogen ion is complete. It is a
question not of the amount of hydrogen displaced from the innermost
portion of the outer layer but of sufficient contraction of the double
Rabinovich and Wasseliev: Kolloid-Z., 60, 268 (1932); Rabinovich and
Fodimann: Z. physik. Chem., AIM, 255 (1931); Rabinovich and Kargin:
A143, 21 (1929).
MECHANISM OF THE COAGULATION PROCESS 211
layer as a result of adsorption of the added cation to lower the potential
on the particles to the critical value. Referring to Fig. 36 it will be
seen that the precipitation concentration for barium chloride is above
the value necessary for displacing all the hydrogen, whereas, with
aluminum chloride, coagulation takes place a little before all the hydro-
gen is displaced. With the hydrous oxide sols the author found re-
peatedly (Vol. II, pp. 73, 114, 142) that coagulation occurred before
all the counter chloride ions in the innermost portion of the outer layer
were displaced by precipitating anions. Moreover, fewer chloride ions
were displaced at the precipitation value of a trivalent anion than of a
bivalent one. All of which means that the displacement of the counter
ions by the precipitating ion is of secondary importance. As a rule,
most of the counter ions will be displaced before the adsorption of the
precipitating ions is sufficient to lower the potential to the critical value.
It would be altogether surprising, however, if the adsorption were al-
ways sufficient at the point where the displacement was complete. The
fact that it is not with barium chloride as precipitant for arsenic tri-
sulfide sol led Rabinovich to postulate the existence of two phases in
the coagulation. The author has no objection to considering potential
lowering and agglomeration as two separate phases of the coagulation
process, but not for the reasons advanced by Rabinovich.
The order in which the several chlorides displace hydrogen ion
from the arsenic trisulfide particles is : Ai > Ba, Sr > Ca > NH 4 .
This is likewise the order of precipitating power of the several electro-
lytes and the order of adsorption of the several cations at concentra-
tions below the precipitation value.
To account for the observed behavior, the constitution of the hy-
drous sol may be represented diagrammatically as shown in Fig 37A,
S-- and HS- being the potential-determining ions which constitute
the inner portion of the double layer with hydrogen ions in the diffuse
outer portion. The hydrogen ions not measurable potentiometrically in
the original sol are represented between the particle and the dotted
line.
On adding an electrolyte such as barium chloride to the sol the
strongly adsorbed barium ions enter the outer layer, displace hydrogen
ions, and take up a position relatively closer to the inner layer than
hydrogen ions, as shown diagrammatically in Fig. 37J3. This con-
traction of the double layer, and increase in charge density of the outer
layer, result in a lowering of the potential on the particles to the point
where collisions result in partial coalescence and agglomeration. Be-
cause of the stronger adsorption of the aluminum ions, the thickness
212 COLLOIDAL ARSENIC TRISULFIDE: STABILITY OF SOL
of the double layer is less than with bivalent barium ions. Accord-
ingly, less aluminum needs to be adsorbed to reduce the potential to
the coagulation point, and precipitation takes place with lower con-
centrations of aluminum salts than of barium salts. With the weakly
adsorbed ions such as ammonium, a relatively high concentration is
necessary to increase the charge density in the outer layer and so lower
the potential to the coagulation point. Because of the weaker adsorp-
FIG. 37 Diagrammatic representation of the constitution of a particle in an
arsenic trisulfide sol (A) before and (B) after the addition of barium chloride.
tion of ammonium than of aluminum and barium, the displacement of
hydrogen by ammonium from the portion of the outer layer repre-
sented within the dotted line is much less than with an equivalent
amount of the multivalent ions.
As already pointed out, the adsorption of multivalent cations at
their precipitation value is much greater than the hydrogen displaced.
The reason is obvious, since a part of the adsorbed ion takes the place
of hydrogen in the outermost portion of the diffuse layer.
The mobility of ammonium ion is retarded in the presence of
arsenic trisulfide. The observed effects were attributed to exchange
adsorption and the action of the diffuse double layer. 97
"Bikerman: Trans. Faraday Soc., 33, 560 (1937).
HYDROPHOBIC SOLS
213
MUTUAL ACTION OF SOLS
Hydrophobic Sols
When suitable amounts of two sols of opposite sign are mixed,
complete mutual coagulation takes place. This is ordinarily attributed
to the mutual discharge of the electrically charged particles of opposite
sign with subsequent agglomeration into clumps that settle out. The
observations of Biltz 98 are commonly cited to show that the action
is determined only by the amount of the charge on the particles and
not at all by their nature." Thus a comparison of the precipitating
action of a series of sols is said to disclose that, whereas the optimum
amount of positive sols required to precipitate negative sols varies,
0*+Sols
100* -Sols
FIG 38. Range of mutual coagulation of oppositely charged sols (Biltz).
the order is always the same. That this is not true is evident from
Biltz's results when given in the form of a diagram as in Fig. 38. 100
The zone of complete mutual coagulation is shown in black. The posi-
tively charged hydrous oxide sols are arranged in- the order of their pre-
cipitating power for the three negative sols ; obviously the orders are
not the same. Similar experiments with three hydrous oxide sols and
a series of negative sols including arsenic trisulfide were carried out
wfier., 37, 1095 (1904).
9'Freundlich: "Kapillarchemie," 402 (1909); Thomas: Bogue's "Colloidal
Behavior," 1, 325 (1924).
100 Weiser and Chapman: J. Phys. Chem, 86, 543 (1931) ; Bancroft: 19, 363
(1915) ; cf. Joshi and Pannikkar: J. Indian Chem. Soc., 11, 797 (1934).
214 COLLOIDAL ARSENIC TRISULFIDE: STABILITY OF SOL
by Weiser and Chapman, taking special precautions to determine
sharply the zone of mutual coagulation. The results represented dia-
grammatically in Fig. 39 show a wide variation in the order of pre-
cipitating power of the negative sols for the three positive sols. Wint-
gen and Lowenthal 101 state the generally accepted view in another
way when they say that the mutual precipitation of oppositely charged
sols is a maximum when the concentrations of the sols expressed in
equivalent aggregates are the same, that is, when equal numbers of
charges of opposite sign are mixed. This rule was likewise found
Fe 2 3
Cr 2 o 3
8np a
,Fe(CN
CuFe(CN)
A B 8,
Congo
Cu.Fel
Cu,Fe(ONJ
100* 4- Sols
0*-Sols
FIG 39. Range of mutual coagulation of oppositely charged sols.
not to hold when a highly dispersed sol of one sign is mixed with a
coarser sol of opposite sign.
Lottermoser 102 observed that the most nearly complete coagulation
of positively charged silver iodide containing a slight excess of silver
nitrate, and negatively charged silver iodide containing a slight excess
of potassium iodide, was obtained when the excess of silver nitrate in
one sol was just equivalent to the excess of potassium iodide in the
other. This suggests that interaction between the stabilizing ions is
the cause of the mutual coagulation of oppositely charged sols. In
line with this, Freundlich and Nathansohn 103 found colloidal arsenic
trisulfide sol and Oden's sulfur sol to be instable in the presence of
101 Z. physik. Chem., 109, 391 (1924); cf. Lottermoser and May: Kolloid-Z,
68, 61 (1932).
10* Kolloid-Z., 8, 78 (1910).
"a Kolloid-Z., 28, 258 (1920) ; 29, 16 (1921) ; cf. Chernstitskaya and Kargin:
J. Phys. Chem (U.S.SR.), 9, 461, 471, 481 (1937).
HYDROPHOBIC SOLS 215
each other. Since both sols are negatively charged, this instability
cannot be due to mutual electrical neutralization but was found to
result from interaction between the stabilizing electrolytes of the two
sols, hydrogen sulfide and pentathionic acid. Following up the above
suggestion, Thomas and Johnson 104 attribute mutual coagulation in
other cases primarily to chemical interaction of the stabilizing electro-
lytes in the sols. This cannot be generally true since mutual coagula-
tion takes place very frequently where interaction between stabilizing
agents is a remote possibility. 105 For the behavior of colloidal silver
and arsenic trisulfide in contact with each other see Vol. I, pp. 131-132.
Boutaric and Dupin 106 followed the course of the mutual coagula-
tion of ferric oxide and arsenic trisulfide sols by measuring the rate
of increase in opacity with a spectrophotometer. As would be ex-
pected, a certain amount x of one must be added before any coagula-
tion takes place, above which the velocity increases until it becomes
instantaneous, then falls oft again, and finally, when more than the
quantity y is added, there is no further coagulation. Decreasing the
particle size, or increasing the proportion of hydrogen sulfide in the
arsenic trisulfide sol, does not affect x for ferric oxide sol but it in-
creases y. The longer the ferric oxide sol is dialyzed the lower the
y-value for arsenic trisulfide, but x is not altered much. The values
of x and y are consistent whether one adds arsenic trisulfide sol to
ferric oxide sol or the reverse. These observations of the effect of
varying purity of sol on the zone of mutual coagulation have been
confirmed and extended by Weiser and Chapman. To account for
the behavior of various combinations it was concluded lor that the pre-
cipitating power of positive sols for negative sols is not determined
exclusively by the magnitude of the ^-potential of the particles in the
respective sols. Other factors that may come in are : (a) mutual ad-
sorption of oppositely charged particles that may be independent of
their potential resulting from adsorbed ions, (b) the presence of pre-
cipitating ions as impurities in the sols, and (c) the interaction be-
tween stabilizing ions.
From the mobility measurements on mixtures of positively and
negatively charged sols with and without the addition of electrolytes,
"4 J. Am. Chem. Soc., 46, 2532 (1923).
*S*f 10 examples in Weiser and Chapman: J. Phys. Chem., 86, 543
(1931).
we Bull. soc. chim., 48, 44, 1059 (1928).
^ Weiser and Chapman: J. Phys. Chem., 36, 543 (1931); 86, 713 (1932);
cf. Bahl: Kolloid-Z, 69, 60 (1932).
216 COLLOIDAL ARSENIC TRISULFIDE: STABILITY OF SOL
Hazel and McQueen 108 conclude that mutual coagulation is due pri-
marily to mutual adsorption of oppositely charged particles with a
consequent unequal redistribution of the total charges around the
particles.
Hydrophilic Sols
Although gelatin is usually regarded as a protecting colloid for sols,
cutting down the precipitating action of electrolytes, Billitzer 109
showed that sols of the sulfides of arsenic and antimony are precipi-
tated by small amounts of gelatin. In larger amounts, it adsorbs the
arsenic trisulfide, converting the whole into a stable positive sol. The
sensitization of arsenic trisulfide sols toward electrolytes by the addi-
tion of small amounts of gelatin has been observed repeatedly. 110 * 1J1
Peskov 112 claims, however, that highly purified gelatin sols are unable
to cause coagulation of other sols unless the former are concentrated
and contain large particles. Negatively charged gelatin in small con-
centrations may precipitate negative arsenic trisulfide probably by ad-
sorbing the peptizing ion and thus reducing the potential. 118 Sugden
and Williams llx claim that gelatin sensitizes arsenic trisulfide sol at
low concentrations and even coagulates it without more than traces of
electrolyte being present. Miiller and Artmann 114 report that gum and
casein are better protecting colloids for arsenic trisulfide sol than are
glue and isinglass. Sodium and potassium soaps protect the sol in
the following order : linoleate > oleate > palmitate > stearate >
myristate > laurate. 113
ION ANTAGONISM IN THE COAGULATION OF SOLS
Linder and Picton, 118 in their early work on the coagulation of
sols by electrolytes, observed that the precipitating action of mixtures
of two electrolytes for arsenic trisulfide sol is approximately additive
provided the precipitating power of each is of the same order of
"8 j. Phys. Chem., 37, 553, 571 (1933).
109 z. physik. Chem., 51, 145 (1905).
nBoutaric and Perrcau: Compt. rend, 181, 511 (1925) ; Ghosh and Dhar:
Kolloid-Z., 41, 229 (1927).
"i Sugden and Williams: J. Chem. Soc., 2424 (1926).
i"J. Russ. Phys.-Chem. Soc., 49, 1 (1917).
i" Peskov and Sokolov: J. Russ. Phys.-Chem. Soc, 68, 823 (1926).
11* Chem. Zentr., I, 1388 (1904).
115 Papaconstantinou : J. Phys. Chem., 29, 319 (1925); cf. Bhatnagar, Prasad,
and Bahl: J. Indian Chem. Soc., 2, 11 (1925).
"J. Chem. Soc., 67, 67 (1895).
ION ANTAGONISM 217
magnitude, whereas the precipitating action may rise appreciably above
an additive relationship if the electrolytes vary widely in their pre-
cipitating power. For example, the precipitating action of mixtures of
strontium chloride and barium chloride are nearly additive, whereas
the addition of potassium chloride increases rather than decreases the
precipitation concentration of strontium chloride. This cannot be at-
tributed to a decrease in the dissociation of strontium chloride by po-
tassium chloride because other potassium salts, such as the nitrate,
give similar results. Since the so-called ion antagonism was not ob-
served with gold sol or with von Weimarn's sulfur sol, but was ob-
served with Oden's sulfur sol which is hydrous, Freundlich and
Scholz 117 conclude that the hydration of the sol and of the precipi-
tating ion are of primary importance in producing ion antagonism and
so in determining whether the precipitating action of mixtures shall
be additive or above the additive value (Vol. I, p. 323). This leads
to the deduction that arsenic trisulfide sol is a hydrophile sol although
it is not usually so considered ; and to the suggestion that the behavior
of colloids with mixtures is a suitable means of determining to what
extent the stability is influenced by hydration. The general accuracy
of these conclusions is rendered questionable by some observations on
the precipitation of chromic oxide sol 118 by mixtures of electrolytes
having widely diff erent precipitating power, such as potassium chloride
and potassium sulfate. Although this sol is highly hydrous, the pre-
cipitation values of mixtures are somewhat less than additive instead
of being considerably above the additive values as the theory of
Freundlich and Scholz would predict. One may be quite certain,
therefore, that hydration is not the only factor in bringing about the
phenomenon of ion antagonism and may be a relatively unimportant
one in certain cases.
As a result of an investigation of this anomalous behavior, in
1921, 119 it was concluded that, in the simultaneous adsorption by solids
from mixtures of two electrolytes having no ion in common, the
most readily adsorbed cation and anion are taken up most and the
other pair least readily; whereas, from mixtures having one ion in
common, the oppositely charged ions are each adsorbed less than if
the other were absent, but the most readily adsorbed ion is displaced
the least. This is illustrated in Table XXVIII, 85 which gives the pre-
11T Kolloid-Beihef te, 16, 267 (1922); cf., also, Neuschlosz: Pfluger's Arch.,
181, 17 (1920); Dorfmann: Kolloid-Z., 46, 186 (1928); 52, 66 (1930).
118 Weiser: J. Phys. Chem, 28, 232 (1924).
"'Weiser: J. Phys. Chem., 25, 665 (1921).
218 COLLOIDAL ARSENIC TRISULFIDE: STABILITY OF SOL
cipitation values of barium chloride and the alkali chlorides for arsenic
trisulfide sol together with the adsorption of barium ion during the
precipitation of the sol with barium chloride alone and when mixed
with a constant amount of the several alkali chlorides. It is apparent
that univalent ions cut down the adsorption of barium in the order:
Li < Na < K < H. Since under otherwise constant conditions, one
TABLE XXVIII
ADSORPTION OF BARIUM ION BY ARSENIC TRISULFIDE
Barium
Electrolytes added to 100 cc sol
adsorbed
Precipitation values
Total volume 200 cc
m.eq./g
milhmols/1
30 cc 02 N BaCla
058
BaCh
1 37
30 cc 02 N Bad, + 30 cc 5 N LiCl .
019
LiCl
88 7
30 cc 02 N Bad* + 30 cc 5 N NaCl
014
NaCl
73 5
30 cc 02 N BaCU + 30 cc 5 N KC1.
009
KC1
63 7
30 cc 02 N BaCU + 30 cc 5 N HC1 .
007
HC1
52 5
would expect the adsorption of a given cation to be cut down by the
presence of a second cation in proportion to the adsorbability of the
latter, it follows that the order of adsorbability of the univalent ions
is : H > K > Na > Li. This is exactly the same as the order one
would deduce from the precipitation values of the salts (Table
XXVIII) assuming that the salt containing the most readily adsorbed
cation precipitates in lowest concentration.
It should be pointed out in passing that the results recorded in
Table XXVIII furnish almost conclusive proof that univalent cations
are adsorbed less strongly 12 than bivalent barium.
Since there would appear to be a mutual cutting down of the ad-
sorption of ions of like charge when sols are coagulated by mixtures,
two factors in addition to hydration may influence the precipitating
action of mixtures- 119 (1) the antagonistic effect of each precipitat-
ing ion on the adsorption of the other, and (2) the stabilizing influence
of the ions having the same charge as the sol.
To illustrate the antagonistic action, some observations with mix-
120 Cf. t however, Dhar, Sen, and Ghosh: J. Phys. Chem., 28, 457 (1924);
Freundlich, Joachimsohn, and Ettisch: Z. physik. Chem., A141, 249 (1929).
ION ANTAGONISM
219
tures of lithium and barium chloride are given in Table XXIX. 121
Since it makes a difference whether the salts are added simultaneously
or separately with an intervening time interval, 122 the former pro-
cedure was used. In the precipitation experiments, 10 cc of the sol
containing 22.3 g As 2 S 3 /l were mixed with the volume of electro-
lytes indicated in the table, in a total volume of 5 cc. For the sake of
accuracy, the adsorption experiments were carried out with 12.5 times
the amounts of electrolyte and sol that were used in finding the pre-
TABLE XXIX
PRECIPITATION OF As 2 S a SOL WITH MIXTURES OF
AND THE ADSORPTION OF BA++ ION
AND LiCL
5 N LiCl
taken
01 N Bads to complete coagulation
Adsorption of
Ba ++ ion
cc
Taken
Calculated
Difference
g/mol AsjSi
cc
cc
%
4 05
4 03
....
....
00
4.50
1 310
50
4 50
3 54
27
971
00
4 25
1 250
1 00
4 25
3 03
38
841
00
3 76
1 145
2 00
3 76
2 03
84
0.520
3 00
2 25
1 03
118
cipitation values. The adsorption data are the average of two or
more determinations in each case.
The above results show the marked influence of lithium ion on the
adsorption of barium ion. Thus, at the precipitation concentration of
a mixture containing but one-eighth the precipitation value of lithium
chloride alone, the adsorption of barium ion is lowered more than
25%, whereas from a mixture containing one-half the precipitation
value of lithium chloride alone the adsorption of barium ion is de-
creased more than 50%. At the same time the presence of barium
ion unquestionably influences the adsorption of lithium ion. This
J. Phys. Chem., 28, 232 (1924); SO, 20 (1926); cf., also, Bou-
taric and Perreau: Compt. rend, 180, 1337; Boutaric and Manure: 1841 (1925) ;
Dumanskii and Vinnikova: J. Phys. Chem. (U.S.S.R.), 5, 133 (1934).
122 Freundlich and Tamchyna: Kolloid-Z., 58, 288; Buzagh, Freundlich, and
Tamchyna: 294 (1930).
220 COLLOIDAL ARSENIC TRISULFIDE: STABILITY OF SOL
mutual action must have some effect in raising the precipitation values
of certain mixtures above the additive value.
Since alkali cations cut down the adsorption of barium ion in
varying degrees, one would expect the precipitation value of barium
chloride to vary with the nature of the alkali ion. That such is the
case is shown graphically in Fig. 40, in which the concentration of
barium chloride is plotted against that of the univalent chloride at the
precipitation value of the mixture. In the same figure, the precipita-
tion concentrations of mixtures of potassium and sodium chloride are
20 40 60 80 100
AfCI, Millimols per Liter
FIG 40 Coagulation of arsenic trisulfide by mixtures of electrolytes
given to show the additive relationship which obtains with mixtures of
similar precipitating power.
Associated with the cation antagonism is the effect of the chloride
ion which cannot be ignored at the relatively high concentrations of
alkali chloride employed. The similarity in form of the curves in
Fig. 40 to that of the mobility curves frequently obtained on adding
alkali halides to negative sols (p. 182) suggests a common explana-
tion. 128 In most cases Dhar 124 and his collaborators rule out the in-
123 Cf. Moyer and Bull: J. Gen. Physiol., 19, 239 (1935) ; Bull and Gortner:
J. Phys. Chem., 86, 700 (1931); Vester: "Het Antagonistic der Electrolyten
bij de Uctolokking van Sunspensoiden," Amsterdam (1935).
is* j. Phys. Chem., 28, 313, 457, 1029 (1924); 29, 435, 517, 659 (1925);
Kolloid-Z., 84, 262 (1924) ; 36, 129 (1925) ; Sen and Mehrotra: Z. anorg. Chem.,
142, 345 (1925) ; see, also, Mukherjee and Ghosh: J. Indian Chem Soc, 1, 213
(1924); cf. t however, Ghosh, Bhattacharya, and Dhar: Kolloid-Z., 38, 145
(1926).
ION ANTAGONISM 221
fluence of cation antagonism in increasing the precipitation value of
certain salt pairs above the additive value, and attribute the effect en-
tirely to the stabilizing action of anions. Certainly this effect pre-
dominates if one of the electrolytes contains a potential-determining
ion, for example in the precipitation of negative copper ferrocyanide
sol with mixtures of potassium ferrocyanide and barium chloride
(p. 324). Freundlich has swung around to this point of view but he
emphasizes also the importance of hydration of the particles as well
as the hydration of the ions involved. Although the author was the
first to point out the necessity of taking into account the effect of the
ion having the same charge as the sol, he still believes that the antago-
nism between ions of the same sign, as recorded in Table XXIX,
must be a contributing factor in certain cases. Ostwald 125 has shown
recently that cation antagonism follows with certain salt pairs when
the cation activity coefficients of the binary mixtures are substituted
for concentrations.
The phenomena considered above have been observed chiefly with
negative sols, but Freundlich 126 observed a similar behavior with ar-
senic trisulfide sol rendered positive by malachite green. For the
results of observations with ferrocyanide sol see p. 322.
"5 Ostwald and Hoffmann Kolloid-Z , 80, 186 (1937).
"6 Freundlich and Tamchyna. Kolloid-Z. f 53, 288 (1930).
CHAPTER XI
THE COLLOIDAL SULFIDES OF ANTIMONY, BISMUTH, TIN,
AND LEAD
COLLOIDAL ANTIMONY TRISULFIDE
The Precipitated Salt
Antimony trisulfide occurs in nature as black orthorhombic crys-
tals known as stibnite. The alchemist "Basil Valentine" prepared an
amorphous red sulfide by subliming stibnite with ammonium chloride.
In this way, antimony trichloride and ammonium sulfide were first
formed and reacted on cooling to give the original compounds, the
trisulfide separating as a red powder. Glauber and also Lenery
speak of the solution of stibnite in caustic alkalis and the subsequent
precipitation of a red powder but it was not until 1714 that particular
attention was called to the preparation. In that year Simon, a Car-
thusian monk, was reported to have restored another monk to health
by the administration of the powder prepared by a German apothe-
cary, a disciple of Glauber. Simon gave to the powder the name
alkermes or kermes mineral by which it is still known. The original
method of formation consisted in boiling the crude sulfide with alkali
until a clear solution was obtained from which the kermes separated.
The preparation was at first believed to be a compound of antimony,
sulfur, and alkali, but it is now known to be antimony trisulfide with
more or less antimony trioxide and adsorbed alkali. The presence of
impurities was objectionable for its use as a therapeutic agent, and so,
in later modifications of the original process, the solution and subse-
quent precipitation of the sulfide were carried out in the absence of
air and any trioxide was extracted with tartaric acid.
By passing hydrogen sulfide into a slightly acidified antimonous
salt solution, the trisulfide is thrown down as a hydrous mass varying
in color from golden yellow to orange-red, depending on the condi-
tions of precipitation. When obtained from chloride solution it ad-
sorbs chloride, 1 and the precipitate dried at 105 usually contains a
lYoutz: J. Am. Chem. Soc., 80, 975 (1908).
222
ANTIMONY TRISULFIDE SOLS 223
small amount of sulfur, probably from the oxidation of adsorbed hy-
drogen sulfide. The precipitated compound adsorbs water quite
strongly. Prolonged drying over sulfuric acid is reported to give a
dihydrate, Sb 2 S 3 2H 2 O, 2 but this composition is doubtless an acci-
dental result of the conditions of drying. When dried at 100 it still
retains adsorbed water which is given up gradually on heating to
higher temperatures. 3 In order to remove all the water in the quanti-
tative estimation of antimony as trisulfide, 4 it is necessary to heat the
amorphous precipitate in a current of carbon dioxide until it goes over
to the black crystalline form which adsorbs very little water.
If a dilute acid solution of antimony trichloride is treated with a
solution of sodium thiosulfate, a bright red sulfide is obtained which
has been called crimson sulfide, antimony cinnabar, and antimony ver-
milion. The color and composition depend on the method of forma-
tion, but it always consists of antimony trisulfide with more or less
sulfur. The precipitation is quantitative in the presence of as low
acid concentration as 0.17-1 N HC 2 H 3 O 2 , whereas for quantitative
precipitation of arsenic 8-10.5 N H 2 SO 4 is required. 8 The various
precipitated sulfides are said to be amorphous to x-rays ; 6 but if heated
to temperatures around 200 they are converted into black crystalline
Sb 2 S 3 having the same crystal structure as the mineral stibnite. The
effect of method of preparation on the physical character and color
of antimony trisulfide and the factors influencing the transformation
to the definitely crystalline form will be considered in the chapter on
the sulfide pigments (p. 299).
Antimony Trisulfide Sols
Formation. A very dilute yellow antimony trisulfide sol is obtained
by passing hydrogen sulfide into a saturated solution of antimony
tnoxide in water. More concentrated sols are prepared by conduct-
ing hydrogen sulfide into a solution of potassium antimony tartrate
or of the oxide dissolved in tartaric acid. 7 The maximum concentra-
tion of antimony salt that can be employed will give a sol containing
approximately 5 g Sb 2 S 3 /l; if higher salt concentrations are used the
*Ditte Compt. rend., 102, 212 (1886).
De Bacho- Ann. chim. applicata, 12, 143 (1919).
*Fresenius-Cohn: "Quantitative Chemical Analysis," 1, 397 (1908).
"Kurtenacker and Furstenau: Z. anorg. Chem., 216, 257 (1933).
'Currier J. Phys. Chem., 80, 236 (1926).
'Schulze: J. prakt. Chem., (2) 27, 320 (1883); cf. Heyer: Crell's "Chem.
Ann ," 2, 227. 321,493 (1785).
224 COLLOIDAL SULFIDES OF ANTIMONY, BISMUTH, TIN, LEAD
sulfide precipitates. The most satisfactory method of preparing the
sol * consists in allowing 200 cc of a tartar emetic solution to drop
slowly into a like volume of water saturated with hydrogen sulfide
through which the gas is kept bubbling. The excess hydrogen sulfide
may be removed by a current of hydrogen or by boiling. Boiling is
not recommended, however, since the size of the particles is increased,
the sulfide is hydrolyzed appreciably, 9 and the stability of the sol is
reduced. The sols may be purified to a certain extent by dialysis, but
if the process is continued too long all the sulfide separates out. Just
as with arsenic trisulfide sol (p. 173), Pauli, Kolbl, and Laub 8 puri-
fied and highly concentrated antimony trisulfide sols by the process of
electrodecantation.
To increase the stability of the sols so that they can be sterilized
for therapeutic purposes, Wolvekamp 10 uses the sodium salts of pro-
talbinic and lysalbinic acids as protecting colloids. Soap 11 is also ad-
sorbed by antimony trisulfide and acts as a protecting colloid. Utzino 12
ground antimony trisulfide in a colloid mill with grape sugar, obtain-
ing sols which appear to show a maximum stability not at the finest
state of subdivision. Other conditions being the same, this result
cannot be correct. The adsorption of such sugars as arabinose, mal-
tose, and levulose at varying concentrations by antimony trisulfide sol
has been followed ia by means of the polarimeter, since adsorbed sugar
does not affect the plane of polarized light. In certain instances the
increase in adsorption with increasing concentration of sugar can be
represented quite accurately by the Freundlich equation.
Properties. The color of antimony trisulfide sols varies widely
with the dilution, as indicated in Table XXX. As with the corre-
sponding arsenic sol, it makes a difference whether a dilute sol is
prepared directly or whether a concentrated sol is diluted; the diluted
sol always contains larger particles and so appears more cloudy in
reflected light. 14 The absorption spectrum of the sol is a continuous
band from the violet to the blue. 15
Antimony trisulfide sol is negatively charged owing to adsorption
Ber., 37, 1097 (1904); Pauli, Kolbl, and Laub: Kolloid-Z., 80, 175
(1937).
'Elbers: Chem.-Ztg., 12, 355 (1888).
iU S. Pat. 1,412,438 (1920).
"Bhatnagar, Prasad, and Bahl: J. Indian Chem. Soc., 2, 11 (1925).
Kolloid-Z., 32, 149 (1923).
18 Prasad, Shrivastava, and Gupta: Kolloid-Z, 37, 101 (1925).
"Biltz and Geibel- Nachr. kgl. Ges. Wiss. Gottingen, 2, 141 (1906).
"Linder and Picton: J. Chem. Soc., 81, 133 (1892).
ANTIMONY TRISULFIDE SOLS 225
TABLE XXX
INFLUENCE OF CONCENTRATION ON THE COLOR OF SB 2 S SOLS
Ratio H 2 O/Sb 2 S 8
Color
200
400
600
1,000
10,000
100,000
1,000,000
Deep red; clear in transmitted light, cloudy brown-red by
reflected light
Raspberry red
Deep yellow-red
Reddish yellow to yellow
Yellow like dilute ferric chloride
Very light wine yellow
Faint yellow in layers at least IS cm thick
of hydrosulfide and tartrate ions. When not dialyzed, it is more stable
toward electrolytes than the corresponding arsenic trisulfide sol, but
the stability can be varied through wide limits by varying the duration
of the dialysis. On account of the similarity between the two sols,
one might expect the order of electrolytes arranged according to their
precipitating power to be approximately the same for both, as the ex-
periments show. 18 Moreover, the effect of concentration of sol on
the precipitation values of electrolytes containing cations of varying
valence is similar to that for arsenic trisulfide sol. 17 The stability of
antimony trisulfide sol decreases with age, and there is always more or
less precipitation of the sulfide together with sulfur after standing for
a long time. It is probable that the sol undergoes a photochemical
decomposition similar to that observed with arsenic trisulfide sol
(p. 176).
Jablczynski followed the velocity of coagulation of antimony trisul-
fide sol using a spectrophotometer. At first 18 the results seemed to
be in accord with the Smoluchowski equation, but later 19 it was found
necessary to modify the equation to fit the observations (cf. p. 201).
Joshi 20 followed the course of the coagulation viscosimetrically and
likewise showed that it did not follow Smoluchowski's equation. Ob-
16 Schilow: Z. physik. Chem., 100, 425 (1922); Iwanitzkaja, Orlowa, and
Schilow: Kolloid-Beihefte, 18, 1 (1923).
"Ghosh and Dhar: Kolloid-Z., 86, 129 (1925).
Jablczynski and Jedrzejowska : Bull. soc. chim., (4) 87, 608 (1925).
" Jablczynski: Kolloid-Z., 64, 164 (1931).
20 Joshi and Prabhu: J. Indian Chem. Soc., 8, 11 (1931) ; Joshi and Narayan:
(Ray Commemorative Volume), 41 (1933); Joshi and Nanjappa: 11, 133
(1934).
226 COLLOIDAL SULFIDES OF ANTIMONY, BISMUTH, TIN, LEAD
servations of the effect of changing the wall area in glass vessels indi-
cate that the coagulation process is autocatalytic. 21
COLLOIDAL ANTIMONY PENTASULFIDE AND TETRASULPIDE
Berzelius 22 first reported that antimony pentasulfide is formed by
the action of hydrogen sulfide on a solution of antimonic acid. Bun-
sen 23 showed this to be true, provided the precipitation is effected by
adding rapidly an excess of a saturated solution of hydrogen sulfide
to the acid. This procedure was recommended by Bunsen for the
quantitative estimation of antimony and for the separation of anti-
mony and arsenic. Bunsen's observations were not confirmed by
Wilm, 24 Mourlot, 25 and Thiele, 26 who found the precipitated sulfide to
contain sulfur easily extractable with a sulfur solvent. Brauner 27
and later Bosek 28 did much toward clearing up the matter by showing
that fairly pure pentasulfide is formed only when Bunsen's directions
are followed exactly. Care must be taken to get all the antimony in
the pentavalent state and to add the hydrogen sulfide solution rapidly
in the cold to the antimonic solution containing 10 to 20% HC1. If
the precipitation is accomplished by conducting hydrogen sulfide slowly
into a solution containing but little hydrochloric acid, a part of the
pentavalent antimony is reduced and the resulting precipitate is a
mixture of pentasulfide, trisulfide, and sulfur. 29 In general, the re-
duction was found to be greater the slower the stream of hydrogen
sulfide and the higher the temperature.
Schurmann and Bohm 30 obtained a fairly pure salt by passing
hydrogen sulfide into a solution of antimony pentachloride containing
12 to 15% free HC1 at a temperature not above -20. The pre-
cipitate contained 15 to 30$> sulfur which was extracted with carbon
disulfide.
Pure antimony pentasulfide prepared by Bunsen's method is a rich
orange in color when dried. Currie finds that heating has little visible
effect on the compound until temperatures around 100 are reached,
21 Cj Desai: Current Sci., 1, 376 (1933)
22p gg. Ann., 7, 2 (1826); cf., however, Rose: 107, 186 (1859).
as Ann., 192, 305 (1878).
2* Z. anal. Chem., SO, 428 (1891).
2Compt. rend., 128, 54 (1896).
28 Ann., 268, 371 (1891).
27 Brauner and Tomicek: J. Chem. Soc., 08, 145 (1888).
28 J Chem. Soc., 67, 515 (1895).
ae Currie- J. Phys. Chem, 80, 205 (1926).
30Kautschuk, 6, 70, 91, 136 (1930).
ANTIMONY PENTASULFIDE 227
when a dulling of the color is noted. At 135 the pigment darkens ap-
preciably to a brown, and at 150 traces of black appear, indicating
decomposition with the formation of black trisulfide. Carbon disul-
fide, chloroform, carbon tetrachloride, benzene, and toluene are with-
out influence on the pure pentasulfide in the cold and have little effect
at the boiling points of the several solvents. 31 Heating the sulfide
causes it to lose sulfur to carbon disulfide in amount depending on
the temperature. If heated between 70 and 75 it loses 1 atom of
sulfur, leaving antimony tetrasulfide, Sb 2 S 4 , which starts to decom-
pose slightly at 105 and rapidly at 155 to 160 giving Sb 2 S 3 . 32 Light
acts slowly on the orange compound, converting it into black tri-
sulfide. 88
It is claimed that the best method of preparing antimony pentasul-
fide is to treat a solution of Schlippe's salt, Na 3 SbS 4 -91I 2 O, with
dilute acids. 34 The product is never pure, containing, according to
Klenker, Sb 2 S 5 , Sb 2 S 3 , and S in varying amounts, depending on the
experimental conditions. This view was called in question by Kirch-
hof ss and by Short and Sharpe, 36 who claimed that the so-called golden
sulfide of antimony is the tetrasulfide, Sb 2 S 4 . This question was set-
tled by Currie 29 ' 87 who examined the golden precipitates obtained from
thioantimonate under different conditions. Equal samples of the sev-
eral preparations were treated with increasing amounts of sulfur sol-
vent and allowed to stand at constant temperature until equilibrium
was reached. The concentration of sulfur in the equilibrium solution
was plotted against the composition of the residue expressed in milli-
grams of S in excess of Sb 2 S 3 , giving curves which indicate that the
decomposition of thioantimonate by dilute acids yields no antimony
pentasulfide but a mixture of sulfur and a solid solution of sulfur and
antimony tetrasulfide.
Heating trisulfide with sulfur yields neither tetrasulfide nor penta-
sulfide. 29 Moreover, the trisulfide does not take up sulfur when kept
in contact with sulfur solvents saturated with sulfur. It appears,
31 Klenker: J. prakt. Chem., (2) 59, 159 (1899) ; Esch and Balla: Chem-Ztg,
28, 595 (1904).
32 Klenker: J. prakt. Chem., (2) 59, 150 (1889); Luff and Porritt: J. Soc
Chem. Ind, 40, 273T (1921); Dubosc: "Le Caoutchouc et la Gutta Percha,"
8886, 8958 (1916).
ssBrauner: J. Chem. Soc., 87, 528 (1895).
s*Abegg: Handbuch anorg. Chem., 8 (3), 620 (1907).
ss Z. anorg. Chem., 112, 67; 114, 266 (1920).
38 J. Soc. Chem. Ind., 41, 1097; cf., however, Twiss: 1717 (1922).
37 Cf. Hansen: Angew. Chem., 45, 505, 521 (1932).
228 COLLOIDAL SULFIDES OF ANTIMONY, BISMUTH, TIN, LEAD
therefore, that the equilibria Sb 2 S 6 * Sb 2 S 3 + 2S and
Sb 2 S 3 + S can be approached from one direction only, the side cor-
responding to the higher sulfide.
Antimony trisulfide and the tetrasulfide from antimonate appear
amorphous to x-rays. 88
A sol of antimony tetrasulfide from antimonate has been prepared
using sodium "protalbinate" as protecting colloid. This sol may be
sterilized by boiling and is recommended as a veterinary remedy in
place of the antimonate. 89
The golden sulfide is widely used as a pigment for rubber and as
an assistant in the vulcanization process (p. 302).
COLLOIDAL BISMUTH TRISULFIDE
The Precipitated Salt
Bismuth trisulfide is thrown down as a brownish black hydrous
mass on passing hydrogen sulfide into a solution of bismuth chloride
that is not too strongly acid. No precipitate is obtained if the con-
centration is greater than \2% HC1 and is incomplete in solutions
stronger than 7.7% . 40 The sulfide is also precipitated from a bismuth
chloride solution by sodium thiosulfate 41 and by thioacetic acid. 42 Ad-
sorbed water is not removed completely from the sulfide until a tem-
perature of approximately 200 is reached. Bismuth ion may be
precipitated and weighed quantitatively as Bi 2 S 3 provided the sulfide
is heated for 1 hour in a stream of hydrogen sulfide. 43 At this tem-
perature the apparently amorphous powder goes over rapidly to the
crystalline form. 44 Bismuth trisulfide is much less soluble in alkali
sulfides than the corresponding sulfides of arsenic and antimony, 45 but
the solubility is not inappreciable. 46
Schneider 4T claimed to get a hydrate of bismuth monosulfide, BiS -
H 2 O, by treating a solution of bismuth tartrate, stannous chloride, and
"Currie: J. Phys. Chem, 80, 237 (1926).
8 Wolvekamp: U. S. Pat. 1,412,438 (1920).
4 Ramachandran: Chem. News, 131, 135 (1925).
4l Vortmann: Monatsh., 7, 418 (1886).
"Tarugi: Gazz. chim. ital, 27 I, 316 (1897).
Moser and Neusser: Chem.-Ztg., 47, 541, 581 (1923)
"Spring: Z. physik. Chem., 18, 553 (1895).
"Stone: J. Am. Chem. Soc., 18, 1091 (1896).
"Ditte: Compt rend., 120, 186 (1895); Stillman: J. Am. Chem Soc., 18,
683 (1896).
"Pogg Ann., 97, 480 (1856).
THE PRECIPITATED SALT 229
potassium hydroxide with hydrogen sulfide; but Vanino and Treu-
bert 48 showed that the alleged compound was a mixture of hydrous
bismuth trisulfide and bismuth. There appears to be no conclusive evi-
dence of the existence of any sulfide of bismuth other than the tri-
sulfide. 49
Bismuth Trisulfide Sol
A sol results on passing hydrogen sulfide into a dilute solution of
bismuth salt and dialyzing to remove excess electrolyte. Winssinger 50
employed a dilute bismuth nitrate solution treated with acetic acid and
dialyzed 36 hours. The sol could be boiled without precipitation but
was quite sensitive to the action of electrolytes. In thin layers the
color was reddish brown and gave an absorption spectrum extending
from the violet to the green and faintly into the red. Ozone precipi-
tated the sol but was without action on the sulfide. 51
COLLOIDAL STANNIC SULFIDE
The Precipitated Salt
Stannic sulfide is precipitated as a yellow gel by the action of hy-
drogen sulfide on an acidified solution of stannic chloride. The pre-
cipitate contains no stannous sulfide but may contain hydrous stannic
oxide owing to hydrolysis of the chloride when the solution is quite
dilute or but slightly acid. 82 It is probable that any oxide carried down
would be converted to sulfide by prolonged action of hydrogen sulfide,
since the gas converts a suspension of freshly formed stannic oxide
into sulfide. 58 Stannic oxide sol peptized by hydrochloric acid is coagu-
lated by hydrogen sulfide, but in this case the conversion to stannic
sulfide is quite slow M because of the physical character of the oxide.
By the action of hydrogen sulfide on stannic chloride under suitable
conditions, Sisley and Meunier 65 claim to get a hydrate SnS 2 2H 2 O
Ber., 82, 1079 (1899).
Atcn: Z. anorg. Chem., 47, 386 (1905); Herz and Guttmann: 08, 63
(1907); 66, 422 (1908); cf. f however, Pelabon: J. chim. phys., 2, 321 (1904);
Compt. rend, 1ST, 648, 920 (1903).
"Bull. soc. chim., (2) 49, 452 (1888).
"Riesenfeld and Haase: Z. anorg. Chem., 147, 188 (1925).
"Storch: Monatsh., 10, 260 (1889).
"Scheerer: J. prakt. Chem., (2) 8, 472 (1871).
"Jorgensen. Z. anorg Chem., 28, 140 (1901); Barfoed: J. prakt. Chem.,
101, 368 (1867).
w Bull. soc. chim., (4) 51,939 (1932).
230 COLLOIDAL SULFIDES OF ANTIMONY, BISMUTH, TIN, LEAD
which is transformed to SnS 2 H 2 O at 120 ; the evidence for the ex-
istence of the alleged hydrates is altogether inadequate.
The adsorbing power of stannic sulfide renders it unsuitable for
the quantitative analysis of tin in certain solutions. Thus cobalt and
nickel are carried down by the gel, the amount adsorbed varying in-
versely as the hydrogen ion concentration of the solutions. 88 The gel
also adsorbs phosphate so strongly B7 that quite low results are obtained
when the phosphate is estimated after the removal of tin as sulfide. 58
If a stannic sulfide gel is allowed to remain in contact with am-
monia, a solution results from which a voluminous white precipitate
is obtained on acidification. 59 The same substance is formed by digest-
ing stannic sulfide with ammonium carbonate solution, filtering, and
acidifying. The dark red solution of stannous sulfide in ammonia be-
comes colorless on standing in the air, and from this, also, the white
voluminous gel is precipitated by an acid. When freed from sulfur
by washing with carbon disulfide the precipitate analyzes for hydrous
stannic oxysulfide, Sn 2 S 3 O #H 2 O. The freshly formed gel is readily
soluble in ammonium carbonate, but, when aged by drying, it loses this
property to some extent. If a dilute solution of the gel in ammonium
carbonate is acidified with sulfuric acid and shaken until most of the
carbon dioxide is removed, there results a very fine flocculent mass
with a blue tinge. On washing with water, this gel is peptized, giving
a sol with an acid reaction. Schmidt attributes the acidity of the sol
to the presence of an acid, Sn 2 S 3 O H 2 O or S(SnS'OH) 2 ; but it
seems more likely that the acid reaction is caused by the presence of
some sulfuric acid which cannot be washed out before peptization of
the gel takes place.
Stannic Sulfide Sol
A sol of stannic sulfide, first described in 1839, 80 was prepared by
the action of sulfur dioxide on a dilute acidified solution of stannous
chloride. More recently it has been prepared by passing hydrogen
sulfide into the sol of hydrous stannic oxide formed by dialysis of
stannic chloride, 61 or by peptization with tartaric acid. 62 The most
58 Auger and Odmot: Compt rend, 178, 710 (1924).
"Kikuchi: J. Chem Soc. Japan, 48, 329 (1922).
58 Lord: Chem. News, 118, 254 (1919) ; cf. Chandelier Bull. soc. chim. Belg.,
38, 255 (1929).
" Schmidt: Kolloid-Z., 1, 129 (1906).
"Hering: Ann Pharm., 29, 90 (1839).
Schneider: Z. anorg. Chem., 6, 83 (1894).
62 Dumanskii and Btmtin: J. Russ Phys.-Chem. Soc., 61, 279 (1929).
THE PRECIPITATED SALT
231
satisfactory method of preparation probably consists in peptizing the
freshly formed gel by thorough washing.
COLLOIDAL LEAD SULFIDE
The Precipitated Salt
The brownish black precipitate of lead sulfide thrown down by
hydrogen sulfide from slightly acid, neutral, and alkaline solutions of
lead salts is said to be amorphous to x-rays, but electron diffraction
studies show it to consist of extremely minute crystals. 68 Sloat and
Menzies 64 measured the adsorption by the precipitated sulfide from
0.01 M solutions of several salts, for the purpose of investigating the
relation between adsorption and lattice dimensions as well as solu-
bility of the adsorbed substance. The results are given in Table XXXI.
TABLE XXXI
VARIATION OF ADSORPTION WITH LATTICE DIMENSIONS AND WITH SOLUBILITY
Salt
Adsorption
mols/g PbSXIO 10
Sum of ionic radii
k
Mol fraction solu-
bility at 25
LiBr
40
2 745
281
NaBr
43
2 97
141
NH 4 Br . .
62 7
3 329
127
RbBr.
82 7
3 43
111
CsBr....
84
3 67
0949
KBr.
126 7
3 285
0897
(PbS)
2 985
. . ..
It is apparent that the amount adsorbed does not depend on the ability
of the adsorbed material to fit the space lattice of the adsorbent (cf.
p. US). Sodium bromide which has lattice dimensions nearest those
of lead sulfide is, with the exception of lithium bromide, adsorbed the
least. Ammonium bromide which belongs to the body-centered cesium
chloride arrangement is adsorbed less than cesium bromide, in spite of
the fact that ammonium bromide is capable of assuming the sodium
chloride arrangement like lead sulfide and is a better fit than cesium
bromide. Moreover, the amount adsorbed does not depend on the
ability of the salt to orient on lead sulfide since the amount adsorbed
63 Natta: Congr. intern, quim. pura aplicada, 9th Congr. Madrid, 2, 177
(1934).
"J. Phys. Chem., 85, 2022 (1931).
232 COLLOIDAL SULFIDES OF ANTIMONY, BISMUTH, TIN, LEAD
by the three unoriented salts, lithium, ammonium, and cesium bro-
mides, is of the same order as the amount adsorbed by the other three
salts which are known to be oriented. 85
Comparing adsorption and mol fraction solubility of the salts, it
will be seen that the amount adsorbed increases as the solubility falls
off, indicating that in this instance solubility is a more important factor
in adsorption than the lattice dimensions are.
By conducting hydrogen sulfide over many metallic salt solutions,
films 66 of the various sulfides are formed on the surface (p. 235).
The maximum thickness of such films of lead and bismuth sulfide on
various concentrations of the nitrate solutions is 1500-2000 A and is
greater the more concentrated the solutions. With increasing thick-
ness of film, the color changes from an olive brown to a red-brown.
Immig and Jander 6T showed that extremely small amounts of lead,
cadmium, copper, silver, and bismuth can be estimated by conducto-
metric titration with hydrogen sulfide using microchemical technique.
Lead Sulfide Sol
Lead sulfide sol may be prepared directly by the action of hydrogen
sulfide on dilute solutions of lead salts, 50 especially the tartrate. 62 The
use of the sol for the colorimetric estimation of small amounts of lead
has been recommended. 68 A fairly stable sol results from the cathodic
disintegration of galena ; 69 but if sols of a high degree of stability are
desired, some protecting colloid must be used. Lefort and Thibault 70
added hydrogen sulfide to a solution of lead acetate containing gum
arabic ; Menegazzi T1 used such protecting colloids as peptone, white of
egg, and starch paste; and Chistoni and Milanesi 72 used denatured
protein in preparing a sol which proved unsatisfactory for cancer
treatment.
Brooks 7S prepared a stable sol of lead sulfide by the action of hy-
drogen sulfide on lead acetate solution containing 0.5% gelatin. The
and Menzies: J. Phys. Chem., 35, 2005 (1931).
fl Mokruschin, Ginsburg, and Demjanova: Kolloid-Z., 75, 10 (1936).
<"Z. Elektrochem., 48, 207 (1937).
as Lucas: Bull. soc. chim., (3) 15, 39 (1896) ; Williams: J. Soc. Chem. Ind.,
25, 137 (1906) ; Ewan: 28, 10 (1909) ; Harcourt: J. Chem. Soc, 97, 841 (1910).
69 Von Hahn: Kolloid-Z. (Zsigmondy Festschrift), 86, 277 (1925).
J. pharm. chim., (5) 6, 169 (1882).
J. Chem. Soc., 110 (1), 452 (1916).
Arch, farmacol. sper., 46, 147 (1929) ; Orestano: Boll. soc. ital. biol. sper.,
7, 263 (1932).
"J. Phys. Chem., 32, 1717 (1928).
LEAD SULFIDE SOL 233
conversion to sol was 100$? with a 0.25% solution of lead as the
acetate; 50% with a 1.0% solution; and 3% with a 1.75% solution.
This behavior was attributed to a decrease in the velocity of nuclei
formation in more concentrated acetate solutions while the velocity of
growth on nuclei remained unchanged (cf. p. 81). At 15, the rate
of combination of lead sulfide in the sol state and phosphate ions was
found to be proportional to the surface of the unchanged particles,
indicating that the reaction takes place between the surface and ad-
sorbed phosphate ions, the adsorption of which is independent of the
phosphate concentration over a wide range.
Lewis and Waumsley 7 * made a deep brown opalescent sol by al-
lowing lead to remain in contact with a solution of caoutchouc in 90%
commercial' benzene containing a small amount of carbon disulfide.
The carbon disulfide acts on the lead, giving lead sulfide, which is
protected by the caoutchouc. Any zinc or lead sulfide formed during
the vulcanization of rubber in the presence of the metallic oxides is
kept in the colloidal state by the rubber. 75
** Kolloid-Z., 11, 39 (1912); J. Soc. Chem Ind., 31, 518 (1912).
"Martin and Davey: J. Soc Chem. Ind., 45, 174T (1926).
CHAPTER XII
THE COLLOIDAL SULFIDES OF COPPER, SILVER, GOLD, AND
THE PLATINUM FAMILY
COLLOIDAL CUPRIC SULFIDE
The Precipitated Salt
Copper sulfide precipitated at room temperature from cupric sulfate
solution with hydrogen sulfide is almost pure CuS, 1 but after aging for
7 days under hydrogen sulfide solution it contains about 2% Cu^S. If
the precipitation is carried out at the boiling point, it contains approxi-
mately 5% Cu 2 S, the amount increasing slightly on aging at room
temperature. This behavior accounts for the variation in the com-
position of the precipitate reported by different people. 2 Sauer and
Steiner 3 -* believe that the product formed by the interaction of a
cupric salt and hydrogen sulfide at room temperature is not cupric
sulfide but cuprous sulfide plus an equivalent amount of sulfur and
that cupric sulfide results only on heating the mixture. The experi-
mental evidence for this opinion was derived from colorimetric ob-
servations of suspensions obtained by the interaction of both cuprous
and cupric salts with hydrogen sulfide under varying conditions. Be-
cause of the similarity in color of cupric and cuprous sulfide, this evi-
dence appears inconclusive. Kolthoff and Pearson 5 report that pre-
cipitates obtained with cupric salt and hydrogen sulfide, both in the
hot and in the cold, give the same x-ray diffraction pattern. This was
confirmed in the author's laboratory, and it was demonstrated further
that the common pattern was from cupric sulfide and not from an
equivalent mixture of cuprous sulfide and sulfur.
iKoithoff and Pearson: J. Phys. Chem., 86, 642 (1932) ; cf. Feigl Z anal.
Chem., 72, 32 (1927); Rossing: Z. anorg. Chem., 26, 413 (1900); Jordis and
Schweizer: Z. angew. Chem., 23, 577 (1910); Coppock: Chem. News, 73, 262
(1896) ; 78, 231 (1897) ; Antony and Lucchesi: Gazz. chim. ital, 19, 545 (1889).
aXhomsen: Ber., 11, 2043 (1878); J. prakt. Chem., (2) 19, 4 (1879);
Brauner: Chem News, 74, 99 (1896); Abel: Z. anorg. Chem, 26, 411 (1901).
Kolloid-Z, 72, 41 (1935).
<C/. Fischbeck and Dorner: Z. anorg. Chem., 182, 228 (1929).
* J. Phys. Chem., 36, 549 (1932).
234
THE PRECIPITATED SALT 235
Films e of copper sulfide from 1 to 25 molecules thick form on a
copper sulfate solution in contact with hydrogen sulfide. Because of
their metallic luster, films 2-3 molecules thick are visible and films
5-10 molecules thick are readily observed. The thinnest films are
golden yellow and crystalline; thicker films appear dark brown to
black. Hydrogen sulfide will not penetrate a film more than 60-80 A
in thickness. The films appear to be laminar systems composed of a
succession of single monomolecular layers with adsorbed ions on their
surfaces. 7 They are hydrophilic on the side next to the solution and
hydrophobic next to the air. The contact angle of copper sulfide and
water is zero in the absence of air. 8 If a fragment of copper or other
metal above copper in the E.M.F. series is placed on the copper sul-
fide film, a secondary film of copper of the order of magnitude of
20 A in thickness is formed below the sulfide. 9 The phenomenon is
electrolytic, the following cell being set up : Cu | CuSO 4 solution | Cu.
Very finely divided copper sulfide, which dissolves in sulfuric acid
giving hydrogen sulfide, 10 is formed by placing copper and sulfur in
contact in a slightly acid solution of copper salt. 11 The color of the
resulting precipitate is described as blue. 19 A precipitate subjected to
600 atmospheres pressure likewise has a blue metallic luster. 18 The
mineral is black, bluish black, indigo blue, dark violet, or brown in
appearance.
Rhythmic bands of cupric sulfide are formed when copper ion is
allowed to diffuse into a gelatin jelly containing ammonium polysul-
fide. 14 The bands are white in color at the outset, but they gradually
become darker, changing through green to brown. It seems unlikely
that cupric sulfide should be white, and the initial white precipitate is
doubtless sulfur ; it is not observed if ammonium sulfide is substituted
for the polysulfide.
Liesegang" placed a strong sodium chloride solution containing
Mokruschin and Demjanova: J. Phys. Chem. (U.S.SR.), 5, 1092 (1934);
Kolloid-Z, 72, 261 (1935) ; Mokruschin: 70, 48 (1935).
7 Mokruschin, Demjanova, and Konyaev: J. Phys. Chem. (U.S.SR), 8,
100; Mokruschin and Vilesova: 640 (1935).
SDeWitt: J. Am. Chem. Soc. f 67, 775 (1935).
"Devaux: Compt. rend., 201, 1305 (1935); 202, 368; Cayrel. 926 (1936).
"Rieder: Z. Elektrochem., 8, 370 (1902).
"Wicke: Ann., 81, 241 (1852); Garelli: Rec. trav. chim., 42, 818 (1923).
"Pdabon: Bull. soc. chim., (4) 51, 377 (1932).
"Spring: Bull, acad. roy. mid. Belg., (3) 6, 492 (1883)
"Hausmann: Z. anorg. Chem., 40, 123 (1904).
"Z. angew. Chem, 86, 229 (1923).
236 COLLOIDAL SULFIDES OF COPPER, SILVER, GOLD, PLATINUM
a little sodium sulfide in the bottom of a beaker; over this was placed
a more dilute sodium chloride solution; and finally, a layer of very
dilute cupric sulfate solution. After a time, the copper ion diffusing
downward and the sulfur ion upward came in contact and formed a
thin layer of cupric sulfide which gradually settled down. A process
similar to this is believed to account for the copper sulfide deposit in
certain lakes. In the salt water at the bottom of the lake, hydrogen
sulfide is formed by the decay of organisms and by sulfur bacteria.
During a flood, fresh water containing a little copper ion flows over
this salt water. By diffusion, there is formed gradually a layer of cop-
per sulfide which settles slowly to the floor of the lake.
Contamination by Zinc. Precipitated copper sulfide adsorbs hydro-
gen sulfide strongly, a circumstance which led Linder and Picton ie
to conclude that hydrosulfides such as 7CuS H 2 S and 9CuS H 2 S are
formed. The sulfide thrown down by alkali sulfides is always con-
taminated more or less by adsorbed alkali salt. 17
When copper sulfide is precipitated in the presence of zinc, the zinc
is carried down in amounts depending on the concentration of zinc
salt, the acid concentration, and the temperature. Kolthoff and Pear-
son 1S have shown that this contamination is not the result of co-
precipitation, 10 direct adsorption, mixed crystal formation, or solid
solution formation. 20 On the contrary, the zinc sulfide precipitates
after the copper sulfide has been quantitatively formed ; the contamina-
tion is therefore essentially a process of post-precipitation. The cop-
per sulfide promotes the precipitation of zinc sulfide by virtue of its
fine state of subdivision and the presence of an adsorbed layer of
hydrogen sulfide on its surface. For example, when a mixture of
copper and zinc sulfate is treated with hydrogen sulfide at room tem-
perature in 0.36 N H 2 SO 4 and filtered immediately after quantitative
precipitation of the copper, all the zinc remains in solution. On the
other hand, if the precipitate is allowed to stand before filtration, zinc
separates on the precipitate as zinc sulfide, its amount increasing with
iJ Chem. Soc., 61, 120 (1892).
"Murmann: Monatsh., 17, 706 (1896); Scheringa. Pharm. Weekblad, 57,
1294 (1920).
18 J Phys Chem, 36, 549 (1932); Kolthoff and Moltzau- Chem. Rev, 17,
293 (1935); cf Kolthoff and van Dijk: Pharm. Weekblad, 59, 1351 (1922).
10 Balarew, Gantschew, and Srebrow: Z. anorg. Chem., 165, 192 (1927);
Balarew and Kaischew 167, 237 (1927); Balarew: Kolloid-Beihefte, 30, 249
(1930) ; Z anal. Chem, 102, 408 (1935) ; Feigl: 66, 25 (1924) ; Z. anorg Chem.,
157, 269 (1926) ; Bottger and Druschke: Ann., 453, 315 (1927).
*> Scheringa: Pharm. Weekblad, 55, 431 (1918) ; 57, 1294 (1920).
THE PRECIPITATED SALT 237
the time. Similarly when a zinc sulfate solution is brought in contact
with precipitated copper sulfide more or less zinc sulfide precipitates on
the copper sulfide surface.
Other things being equal, the amount of zinc sulfide post-precipi-
tated in a given time decreases with increasing acidity. 21 This accounts
for the analytical separation being made in strongly acid solution. 22
For the same acid normality, less post-precipitation was observed with
hydrochloric acid than with sulfuric, because of the higher activity of
hydrogen ion in the hydrochloric acid solution.
Other things being equal, copper sulfide precipitated at high tem-
peratures was found by Kolthoff and Pearson to favor the post-pre-
cipitation of zinc much more than that thrown down at room tempera-
ture ; hence, for the best separation, a higher acidity is necessary when
the precipitation is effected at high temperatures. The sulfide formed
at room temperature is black and easy to filter, whereas that formed
at 100 is greenish in color, slimy, and hard to filter. If copper sulfide
precipitated at room temperature is allowed to age under hydrogen
sulfide before any zinc solution is added, its promoting effect on the
precipitation of zinc is increased. Moreover, it assumes the same ap-
pearance as the sulfide formed at 100, changing from black to greenish
and becoming rather slimy and hard to filter.
To account for the above-mentioned variations in physical character
which are responsible for the varying degrees of post-precipitation, it is
necessary to know the combined effect of acidity, hydrogen sulfide con-
centration, and temperature on the primary and secondary particle size
of the precipitate. The fact that the cold-precipitated salt aged under
hydrogen sulfide assumes the properties of the hot-precipitated salt
indicates (1) that the primary particle size does not vary widely with
the temperature of precipitation and (2) that the cold-precipilatcd
salt forms dense aggregates having a lower effective surface which is
increased on standing by the peptizing action of hydrogen sulfide.
X-ray analysis disclosed that the different precipitates had the same
crystal structure, but the work should be repeated to determine from
the width of the diffraction lines whether the primary particle size
varies appreciably with the conditions of precipitation.
The increased reactivity of zinc ions to form zinc sulfide at the
21 Cf. Bottger and Druschke: Ann., 463, 315 (1927).
"Larsen: Z. anal. Chem , 17, 312 (1878); Berglund: 22, 184 (1883); cf.
Baubigny: Compt rend, 94, 1183, 1251, 1473, 1595 (1882); 96, 34 (1883); 106,
751, 805 (1887); 107, 1148 (1888); 108, 236, 450 (1889); Glixelli: Z. anorg.
Chem., 66, 297 (1907).
238 COLLOIDAL SULFIDES OF COPPER, SILVER, GOLD, PLATINUM
surface of copper sulfide is probably the result of the greater tendency
of adsorbed hydrogen sulfide to ionize as compared with hydrogen sul-
fide in the bulk of the solution. Cystine, thiophenol, and thiobarbituric
acid accelerate the precipitation of zinc sulfide alone but they strongly
retard the post-precipitation on copper sulfide, probably because they
are so strongly adsorbed that they displace or prevent the adsorption
of hydrogen sulfide.
Finely divided substances such as glass powder, silica gel, barium
sulfate, charcoal, sulfur, talcum, aluminum oxide, and filter paper like-
wise promote the precipitation of zinc sulfide. Kolthoff and Pearson
cite this behavior to show the general character of the surface effect
upon the speed of formation of a precipitate from a supersaturated
solution.
Metallic Conduction of Copper Sulfide. Solid cupric sulfide is a
relatively good conductor of electricity, Badeker 28 finding the specific
resistance at ordinary temperature to be 0.000125 when that of copper
is 0.0000017 and that of bismuth is 0.0001 IS. 24 Thus cupric sulfide is
almost as good a conductor as bismuth. Hittorff 25 and Bodlander and
Idaszewsky 26 observed no migration of copper ions during the passage
of the current and concluded therefore that the salt conducts like a
metal and not like an electrolyte. Cuprous sulfide is likewise a con-
ductor, but the bulk of the evidence indicates that it conducts electro-
lytically. 27 Bodlander attributes the slight conductivity of cuprous
sulfide at room temperature to the presence of a small amount of cupric
sulfide. He states further that, at 110, cuprous sulfide becomes an
electrolyte but the electrolysis produces cupric sulfide at the anode, the
metallic conductivity of the cupric gradually replacing the electrolytic
conductivity of the cuprous sulfide.
A completely satisfactory explanation of metallic conduction in
compounds such as cupric sulfide is not yet available. Trumpler 28
measured the potential of a number of solid conducting salts against a
saturated solution containing the negative component of the salt in the
23 Ann. Physik, (4) 22, 749 (1907).
2 *Giebe: Dissertation, Berlin (1903).
2sp og g Ann., 84, 1 (1851).
2 Z. Elektrochem., 11, 161 (1905).
2' Hittorff: Pogg. Ann., 84, 1 (1851); Kohlrausch: Wied. Ann., 17, 642
(1882); Bidwell: Phil. Mag., (5) 20, 328 (1885); Monch: Neues. Jahrb. Min-
eral. Geol., 20, 365 (1905) ; von Hasslinger: Monatsh., 28, 173 (1907) ; Trumpler:
Z. physik. Chem., 99, 9 (1921) ; Tubandt, Eggert, and Schibbe: Z. anorg. Chem.,
117, 1 (1921).
2 Z. physik. Chem., 99, 9 (1921)
THE PRECIPITATED SALT 239
free and ionic state. From the different behavior of metal-like and
electrolytic conductors with respect to the influence of the negative
component on the potential, a fundamental difference in internal struc-
ture is deduced. Thus it is assumed that, in pure metallically conduct-
ing compounds, the space lattice points are occupied by atoms or mole-
cules but not by ions ; hence such compounds appear to be non-polar,
in contradistinction to electrolytic conductors. This assumption may
be true, but it is not very helpful since it offers no explanation of
the fact that metallic conductivity obtains with only a few salts. More-
over, it does not account for the very much greater metallic conduc-
tivity of cupric sulfide than of other salts of the same kind. Since
Bridgman 29 has prepared a modification of phosphorus possessing
metallic properties, it may be that a similar form of sulfur exists and
that cupric sulfide is a compound of copper with this metallic form of
sulfur.
The Electrocapillary Phenomenon of Becquerel. If copper nitrate
solution is placed in a cracked test tube immersed iri a beaker of
sodium sulfide solution, keeping the two liquids at the same level, cop-
per sulfide first forms in the cracks and is followed by the appearance
of crystals of copper on the copper nitrate side and of a yellow layer
of solution containing polysulfide on the other side. Becquerel, 30 who
first observed this phenomenon, believed it to be connected in some way
with electroendosmose. 31 Thirty years after Becquercl's discovery
Braun 82 passed a high-potential current through capillaries such as
cracked glass, using heavy metal solutions, and observed a deposition
of metal on the side of the capillary turned toward the positive pole.
This phenomenon, called "electrostenolysis," is similar in certain re-
spects to the Becquerel phenomenon, and Coehn 83 believes it to be
closely connected with electroendosmose.
It is well known that colloidal particles will migrate in an electric
field in a direction depending on their charge; this is known as cata-
phoresis. If the particles are kept stationary, in other words, if a por-
ous diaphragm of the particles separates the anodic and cathodic solu-
tions an impressed potential will cause the liquid to move in a direction
*>J Am. Chem. Soc., 36, 1344 (1914); 88, 609 (1916).
soCompt. rend, 64, 919, 1211; 65, 51, 62, 720 (1867) ; 66, 77, 245, 766, 1066;
67, 1081 (1868); 71, 197 (1870); 74, 1310 (1872); 76, 245 (1873); 78, 1081; 79,
82, 1284 (1874) ; 80, 585 (1875) ; 82, 354 (1876) ; 84, 145; 86, 169 (1877).
81 Cf., however, Ostwald: Z. physik. Chem, 6, 71 (1890).
2 Wicd. Ann., 42, 450 (1891) ; 44, 473 (1892).
3 Z. Elektrochem., 4, 501 (1898); Z. physik. Chem., 26, 651 (1898); Holmes:
J. Am. Chem. Soc., 86, 784 (1914).
240 COLLOIDAL SULFIDES OF COPPER, SILVER, GOLD, PLATINUM
depending on the sign of the diaphragm charge which, in turn, de-
pends on the nature of the ion preferentially adsorbed by the dia-
phragm. 84 This phenomenon is called electroendosmose. The inner
walls of the capillary in contact with the liquid are coated with an
electrical double layer. With a glass capillary, the stationary layer on
the glass is negatively charged as a rule, owing to preferential adsorp-
tion of anions, whereas the movable layer in the liquid contains an
excess of cations and so is positively charged. Under the influence of
an electric current of sufficient voltage, the positively charged movable
side of the double layer is torn away, leaving the end of the capillaries
toward the anode charged negatively. On these cathodic points metal
will deposit only in small amounts as a rule ; first, because of the rela-
tively small charge on the capillary wall, and second, because the mi-
nute deposit takes part in the conduction, one end becoming anode and
losing as fast as the cathode end gains. But, as Coehn points out, the
quantity of metal can grow under certain conditions: first, if the metal
is noble; second, if an insoluble compound, especially a peroxide, is
formed at the anode side; and third, if the discharged anion, instead
of dissolving the metal, oxidizes "ous" to "ic" salts in the solution.
Freundlich 35 points out that in Recquerel's experiment the two
sides of the glass capillary will be oppositely charged by contact with
the oxidizing and reducing solutions on the two sides. Now, as al-
ready noted, copper sulfide is first formed in the capillary, and this salt
is a fairly good metallic conductor. Local currents will therefore be
formed which flow in one direction through the capillaries and in the
opposite direction through the copper sulfide as well. Just as in elec-
trostenolysis, an appreciable amount of metal can form on the cathodic
portion of the capillary provided the conditions referred to above
obtain.
Cupric Sulfide Hydrosols
The sol of cupric sulfide was first mentioned by Wright, 86 who
prepared it by treating the precipitated salt with insufficient potassium
cyanide for complete solution, filtering and washing the residue which
J chim. phys., 2, 601 (1904); Bethe and Toropoff: Z physik.
Chcm, 88, 686 (1914) ; 89, 597 (1915) ; Girard; J. chim phys, 17, 383 (1919) ;
Girard and Platard: Compt. rend., 178, 1212 (1924); Gyemanf Kolloid-Z.,
28, 103 (1921).
as "Kapillarchemie," 2nd ed, 371 (1922); Kolloid-Z., 18, 11 (1926); Freund-
lich and Sollner- Z physik. Chem., A138, 349 (1928); A152, 313 (1931); cf.
Bikcrman: A163, 451 (1931)
aJ. Chem. Soc., 43, 163 (1883).
CUPRIC SULFIDE HYDROSOLS 241
was promptly peptized. Spring 3T first obtained a stable sol by wash-
ing the precipitated sulfide with hydrogen sulfide water until peptiza-
tion was complete. The sol was brown when dilute and black with
a slight greenish fluorescence when concentrated. It withstood boiling
without precipitation but was quite sensitive to the presence of salts.
The amorphous residue obtained on evaporating to dryness had the
appearance of a black varnish. It was not peptized by washing even
when the drying was done at ordinary temperatures. Linder and
Picton 16 suspended hydrous cupric oxide in water through which hy-
drogen sulfide was passed until the oxide was converted into sulfide
and peptization was complete, 5 days being required. Young and
Neal 38 started with suspended copper carbonate instead of the hydrous
oxide.
Muthmann and Stutzel * 9 made a fairly stable sol by passing hy-
drogen sulfide into a solution of potassium cupri thiosulfate, K 2 Cu-
(S 2 O 3 ) 2 . Lottermoser * obtained a very much more satisfactory
preparation by treating a saturated solution of copper glycocoll with
hydrogen sulfide. When the procedure was carried out in the cold,
the sol was clear brown in color and was made up of very finely
divided particles; when carried out in the boiling solution, the color
was a deep olive green and the particles were distinctly larger. The
brown sol formed in the cold was changed to green by heating. The
presence of glycocoll has a marked stabilizing influence on the sol,
just as succinimide protects hydrous cupric oxide sol formed by hy-
drolysis of copper succinimide. 41
Stable sols result by the interaction of cupric salt and hydrogen
sulfide in the presence of gelatin 42 or gum arabic 3 as protecting col-
loid. In accord with Lottermoser's observation, sols formed in the
cold are brownish in color and change to greenish on heating. Since
the sols formed by the action of hydrogen sulfide on cuprous salt in
the presence of gum arabic are brownish and do not change their color
on heating, Sauer and Steiner 3 conclude that the brown sol obtained
with a cupric salt in the cold is an equivalent mixture of cuprous sul-
fide and sulfur and not cupric sulfide. As already indicated, this
question may be settled definitely by obtaining x-ray diflFraction pat-
s' Ber, 16, 1142 (1883).
as J. Phys Chem, 21, 14 (1917).
39 Her, 31, 1734 (1898).
* J. prakt. Chem., (2) 75, 293 (1907).
"Ley: Her, 88, 2199 (1905).
42 Menegheth : Boll soc. ital. biol. sper., 4, 613 (1929).
242 COLLOIDAL SULFIDES OF COPPER, SILVER, GOLD, PLATINUM
terns of the respective products. Warming a mixture of protected
copper sol and sulfur sol gives a stable cupric sulfide sol. 48
The removal of the excess hydrogen sulfide from sols prepared
by the method of Spring, or of Linder and Picton, decreases their
stability appreciably. 44 On the other hand, an excess of hydrogen sul-
fide is said to decrease the precipitation value of certain electrolytes
for the sol. 45 The relative precipitating powers of potassium chloride,
calcium chloride, and aluminum chloride are in the approximate ratio
1:39:875.
The sulfides of copper and mercury are obtained in the sol state
by subjecting suspensions of the salts to the action of the silent elec-
tric discharge. 46 The peptizing action is attributed to hydrogen sulfide
formed by reduction of the metallic sulfides.
Peskov 47 observed the same lyotropic influence of the cation of
neutral alkali salts on the solution of colloidal cupric sulfide as already
noted with colloidal arsenic trisulfide (p. 180).
Cupric Sulfide Organosols
Organosols of cupric sulfide are obtained by passing hydrogen sul-
fide into dilute solutions of copper salts in an organic solvent. Nau-
mann 48 obtained a pyridine sol by the action of dry hydrogen sulfide
on a dilute solution of copper chloride in pyridine; Lottermoser 40
made an alcosol from a solution of copper glycocoll in absolute alcohol
and an ether sol from a solution of copper acetoacetic ester in ether;
and Errera 49 prepared an alcosol from a solution of copper acetate in
absolute alcohol. In general, the organosols are most stable when
quite dilute, but Errera's alcosol was quite stable although it contained
38.7 g CuS/1.
Errera observed the time required to coagulate alcosols of mer-
curic sulfide, cupric sulfide, and platinum by the addition of liquids of
varying dielectric constant, Z). One cubic centimeter of the alcosol
containing 38.7 g CuS/1 was added to 6 cc of the several liquids and
the observations made as recorded in Table XXXII. Experiments were
also carried out with mixtures of the several liquids with varying
*3Sauer and Steiner: Kolloid-Z., 72, 35 (1935).
"Young and Neal: J. Phys. Chem., 21, 18 (1917).
Mukherjee and Sen: J. Chem. Soc., 115, 461 (1919).
Miyamoto: J. Chem. Soc. Japan, 56, 1359 (1935); Kolloid-Z., 74, 32
(1936).
Kolloid-Z., 82, 24, 163 (1923).
"Ben, 87, 4612 (1904).
*' Kolloid-Z., 82, 240 (1923).
CUPRIC SULFIDE ORGANOSOLS
243
TABLE XXXII
ACTION OF LIQUIDS ON CUPRIC SULFIDE ALCOSOL
Liquid
D
Observations
Water
81
Milky in 30 min., no further change in 127 hr.
Nitrobenzene .
Methyl alcohol . ..
Ethyl alcohol
35.5
33
25.0
No change in 127 hr.
No change in 127 hr.
No change in 127 hr.
Acetone
21.0
No change in 127 hr.
Isobutyl alcohol
Isoamyl alcohol . . .
Amyl acetate
18 6
5 7
5 6
No change in 127 hr.
No change in 127 hr.
No change in 127 hr.
Chlorobenzene ....
Chloroform . .
Ether
Benzene
5.57
4 95
4 36
2 29
Partial coagulation in 8 min,; complete in 3 5 hr.
Partial coagulation in 96 hr.; incomplete in 126 hr
Partial coagulation in 10 min.; complete in 3 5 hr.
Partial coagulation in 96 hr.; incomplete in 126 hr.
Xylene . .. .
Hexane ....
Carbon tetrachloride
2 5
1 85
2 25
Incomplete coagulation in 7 min.; complete in 3 5 hr.
Complete in 1 min.
Incomplete in 1 min. ; complete in 3 5 hr.
quantities of ethyl alcohol. The addition of liquids having a dielectric
constant greater than the dispersing medium has no coagulating effect
on the sol, whereas liquids with a dielectric constant smaller than that
of the dispersing medium coagulate the sols, and usually, the smaller
the dielectric constant, the greater is the coagulating power of the
added liquid. In general, if the dielectric constant of the dispersed
phase is greater than that of the dispersing liquid, the addition of a
second liquid of lower dielectric constant increases the difference be-
tween the dielectric constant of the two phases, thereby rendering the
sol more stable. On the other hand, if the dispersed phase has the
lower dielectric constant and if the dielectric constant of the medium
is further decreased, two cases exist: (1) the difference in dielectric
constant is slight, the initial additions decreasing the stability to a
minimum followed by an increase in stability with further additions ;
(2) the dispersing phase may have a very low dielectric constant, addi-
tions lowering the stability both by decreasing the difference between
the dielectric constant of the media and by lowering the charge on
the submicrons. Cupric sulfide alcosol is negatively charged both
alone and when mixed with ten times its volume of benzene. Agglom-
eration of the colloidal particles is always accompanied by the distinct
color change from dark brown to a bright olive green.
244 COLLOIDAL SULFIDES OF COPPER, SILVER, GOLD, PLATINUM
COLLOIDAL CUPROUS SULFIDE
A brown sol of cuprous sulfide is said to form on mixing a cuprous
ammonium chloride solution with hydrogen sulfide when the ratio of
copper to sulfur in the interacting solution is 2 to 1. A similar sol
results by the interaction of copper and sulfur sols in the proper pro-
portions ; with sulfur in excess the first product is cuprous sulfide. 3 ' 43
All 1 these preparations should be examined by x-rays to determine
whether they are chiefly cuprous sulfide or cupric sulfide.
When a solution of sodium cupri thiosulfate is acidified, there is a
period of induction followed by the appearance of what is described
as red cuprous sulfide sol. 50 This subsequently precipitates, giving a
blood-red gel which turns successively brown, deep brown, greenish
black, and finally black. The length of the induction period depends
on the nature and concentration of the acid and the concentration of
sodium cupri thiosulfate. The chief cause of the phenomenon is the
formation of colloidal sulfur, which inhibits the precipitation process.
The use of ozonized hydrochloric acid, nitric acid, or perchloric acid,
which oxidize colloidal sulfur, cuts down the induction period. With
a constant amount of acid and sodium cupri thiosulfate, the presence
of increasing amounts of sodium thiosulfate decreases the induction
period up to a certain point and then increases it. The hastening is
attributed to the coagulating influence of the added salt on the col-
loidal sulfur; and the subsequent retarding at higher salt concentra-
tions, to the destruction of acid.
Svedberg 51 prepared a sol of cuprous sulfide in isobutyl alcohol
by electrical disintegration of the mineral, copper glance.
COLLOIDAL SILVER SULFIDE
The Precipitated Salt
Silver sulfide is precipitated as a black amorphous powder by the
action of hydrogen sulfide or a soluble sulfide on a solution of silver
oxide or salt. The precipitate from neutral silver nitrate solution al-
ways contains adsorbed sulfur which cannot be removed by carbon
disulfide. 52 Hantzsch 53 claims to have prepared silver disulfide,
Ag 2 S 2 , as a brown amorphous powder by adding a solution of sulfur
60 Sambamurty : Proc. Sci. Assoc. Maharajah's College, Vizianagaram,
Dec. 10 (1922) ; Chem. Zentr., I, 1414 (1923).
""Die Methoden zur Hersteltung Kolloider Losungen," 490 (1909).
2 Kohlschutter and Eydmann: Ann., 890, 347 (1912).
63 Z. anorg. Chem., 19, 104 (1899).
SILVER SULFIDE SOL 245
in carbon disulfide to a solution of silver nitrate in benzonitrile. If
solvents other than benzonitrile are used, silver sulfide is obtained.
Since the failure of carbon disulfide to extract sulfur from the alleged
disulfide is the chief evidence in support of its chemical individuality,
there may be some question whether it is a true compound or a solu-
tion of silver sulfide and sulfur.
Silver sulfide is formed by the action of moist hydrogen sulfide on
metallic silver. According to Cans 54 the reaction proceeds rapidly
at first, slows down after 200 hours, increases again after 500 hours,
and then continues at approximately constant velocity. Tt is suggested
tentatively that the decrease in velocity after 200 hours is caused by
the formation of a protecting layer of amorphous sulfide which is sub-
sequently ruptured or crystallizes, allowing the surface action to pro-
ceed more rapidly. This hypothesis should be tested experimentally
by examining the protecting layer of sulfide by means of electron dif-
fraction. Silver sulfide has been synthesized by the action of pure
sulfur vapors on pure silver for the purpose of estimating the atomic
weight of sulfur. 55
Unlike cupric sulfide, which is a pure metallic conductor, /J-silver
sulfide stable below 179 is a mixed conductor, about 80% of the
current being carried by silver ions and 20% being conveyed as in a
metallic conductor; the a modification, stable above 179, is a pure
electrolytic conductor. 56
Silver Sulfide Sol
Preparation. The most stable hydrosols of silver sulfide are ob-
tained by precipitating the salt in the presence of protecting colloids
such as gum arabic, 57 casein, 58 albumin, dextrin, gelatin, and glue. 59
Paal and Voss 00 prepared very stable sols using a salt of protalbinic
or lysalbinic acid as protecting colloid. One gram of sodium "pro-
talbinate" in 15 cc of water is treated with an equivalent amount of
silver nitrate; the precipitate of silver "protalbinate" is suspended in
water and treated with ammonium sulfide. The sol of silver sulfide
<"Z physik Chem, 109,49 (1924).
55 Homgschmid : Z. Elektrochem., 86, 689 (1930); Honigschmid and Sacht-
Icben: Z. anorg. Chem., 195, 207 (1931).
fifl Tubandt, Eggert, and Schibbc- Z. anorg. Chem., 117, 1 (1921) ; cf Trump-
ler- Z. physik. Chem., 99, 9 (1921).
57 Lefort and Thibault: J. pharm. chim , (5) 8, 169 (1882)
"Muller and Artmann: Oesterr. Chem.-Ztg, 7, 149 (1904).
B9 Stiasny: Gerber, 38, 124 (1907); Chem Zentr., II, 489 (1907).
>Ber., 57, 3862 (1904).
246 COLLOIDAL SULFIDES OF COPPER, SILVER. GOLD, PLATINUM
which results is dialyzed to remove excess electrolyte and evaporated
to dryness in a vacuum desiccator. The residue is readily peptized
by water, giving a stable sol of any desired strength.
Freundlich and Nathansohn 61 obtained a silver sulfide sol by mix-
ing Carey Lea's silver sol with a sulfur sol. Both sols are negatively
charged, and the interaction takes place at the surface of the micelles.
The transformation is accompanied by changes in color : brown, wine
red, violet, steel blue, greenish blue, greenish brown, and finally a
pale yellowish brown. The color changes result from the presence of
varying amounts of silver, sulfur, and silver sulfide, in the micelles, as
in the photo-chlorides (Vol. I, p. 121). A mixture of a solution of
hydrogen sulfide and colloidal silver goes through similar color changes,
finally giving colloidal silver sulfide.
Color. A stable sol of silver sulfide may be obtained without the
use of protecting colloids by adding hydrogen sulfide or sodium sulfide
to a dilute solution of silver nitrate. If the solution is as dilute as
0.002 N, a stable yellow sol results fl which passes through a range of
colors from yellow to green on adding electrolytes. 63 Some observa-
tions are recorded in Table XXXIII. Five-cubic-centimeter portions
TABLE XXXIII
COLOR CHANGES IN THE COAGULATION OF SILVER SULFIDE SOL BY ELECTROLYTES
Salt
Concen-
tration
m.eq./l
Color
Salt
Concen-
tration
m.eq./l
Color
KC1
KC1
KC1
60
80
90
Yellow
Ice blue
Blue
KNO.
KNOs
MgSO 4
100
120
10
Violet
Brown-red
Yellow
KC1
HC1
HC1
HC1
HC1
KNO . . .
KNO
100
80
100
120
140
60
80
Violet
Yellow
Blue
Violet
Dull green
Dull green
Ice blue
MgSO 4 . ..
MgS0 4 . . . .
MgSO 4
A1 2 (S0 4 ), ....
Al a (S0 4 ), . .
Al,(S0 4 )i ....
Al a (SO 4 ), .
20
40
80
1 5
2 4
3.0
3.6
Violet
Blue
Dull green
Leaf green
Blue
Red
Dull green
6i Kolloid-Z. f 29, 16 (1921).
Oden: Nova Acta Regiae Soc. Sci. Upsaliensis, (4) 3, 4 (1913); cf.
Winssinger: Bull. soc. chim., (2) 49, 452 (1888) ; Lottermoser : J. prakt. Chem.,
(2) 72, 39 (1905).
Von Hahn: Kolloid-Z., 27, 172 (1920).
SILVER SULFIDE SOL 247
of the colloid were mixed with 5 cc of electrolyte of varying concen-
trations, expressed in milliequivalents per liter (m.eq./l) in the final
volume, and the color noted after standing 5 minutes. The color is seen
to depend on the concentrations of electrolyte between fixed limits;
thus, for potassium chloride no change takes place below 60 m.eq./l,
and the color is always brownish red above 110 m.eq./l. The stabil-
ity at the above stages of agglomeration is not great, and complete
coagulation results in a few hours. The precipitate is always dark
brown, but, if shaken with the solution from which it separates, the
suspension appears bluish from the blue and violet sols and reddish
from the red and dull green sols.
The age of the sol has a marked influence on the variety of the
color changes during the flocculation of the sol; e.g., S hours after
preparation no blue or green colors can be obtained, and after 7 hours,
no violet color. If allowed to stand 24 hours the sols change directly
from yellow to reddish brown.
The results indicate that the micelles of the sol increase in size
spontaneously at least during the first 24 hours. Secondary particles
of a wide variety of sizes may be produced by partial agglomeration
of fresh sols with varying amounts of electrolytes. The difference in
size of the secondary aggregates is probably the chief cause for the
wide variety of colors observed. It is a mistake, however, to conclude
a priori that the variation in color is caused entirely by the size of the
micelles. Thus, colloidal gold sols fl4 can be obtained which are red,
violet, or blue by transmitted light, and colloidal silver sols 6fi which
are yellow, red, or blue ; in general, the micelles of the blue sols are
larger than the red, but this is not always true (cf. Vol. I, p. 79).
The appearance of the violet color in the flocculation of fresh sols
is so sharp that von Hahn 66 compared the stability of sols prepared
under different conditions by means of the "violet value" which was
taken arbitrarily as the concentration of electrolytes which is just suffi-
cient to produce a certain shade of violet after 5 minutes' standing (44,
according to Ostwald's color scale). The stability of the sol was
found to increase with (1) length of time of treatment with hydrogen
sulfide, (2) more rapid passing of hydrogen sulfide through the solu-
tion, and (3) higher temperatures of formation. In general, the op-
timum conditions for stability are those which cause the maximum
rate of precipitation and hence the greatest degree of dispersion. Von
* Faraday: Phil. Trans, 147, 145 (1857).
65 Lea: Am. J. Sci., (3) 87, 476 (1889).
Kolloid-Z., 29, 139 (1921).
248 COLLOIDAL SULFIDES OF COPPER, SILVER, GOLD, PLATINUM
Hahn's observations agree with those of Boutaric and Vuillaume on
arsenic trisulfide sol already referred to (p. 172).
The "violet value" for different electrolytes gives the order of their
precipitating power. From Table XXX1H the ratio A1 2 (SO 4 ) 3 :
MgSO 4 : KC1 : HC1 is approximately 1 : 10: 50: 70.
Pieroni 67 prepared a pyridine organosol of silver sulfide by treating
a solution of silver nitrate in pyridine with hydrogen sulfide dissolved
in pyridine.
COLLOIDAL SULFIDES OF GOLD
Gold monosulfide is* formed by passing hydrogen sulfide into a
solution of aurous cyanide in potassium cyanide and then acidifying
with hydrochloric acid. The finely divided steel-gray mass must be
washed with water containing dilute hydrochloric acid ; otherwise it is
peptized, forming a perfectly clear sol. 08 When dried it yields a
brownish-black powder that cannot be peptized. A sol is obtained
also by saturating with hydrogen sulfide a solution of aurous cyanide
in potassium cyanide, 69 adding a little hydrochloric acid, and warming
carefully until a brown turbidity is produced. On dialysis a stable
deep-brown sol results which can be separated from the heavy undis-
solved sulfide by decantation and filtration.
A so-called auro-auric sulfide, Au 2 S 2 , is said to form when hydro-
gen sulfide is passed into a cold neutral solution of auric chloride ; 70 71
but Gutbier and Durrwachter, 72 unable to obtain the compound by this
method, question its individuality. Whatever the substance may be,
it is readily obtained in colloidal solution. Winssinger 73 added hydro-
gen sulfide to a solution of auric chloride as nearly neutral as possible
and secured a sol containing 0.55 g of sulfide per 1 which was stable
for several weeks. Schneider 69 made a more concentrated sol by
treating the freshly precipitated sulfide with insufficient potassium cya-
nide or ammonium polysulfide for complete solution, suspending the
residue in water, and dialyzing. After the electrolyte concentration
was reduced below a critical value, the suspended particles were pep-
tized, giving a stable sol that was clear in both transmitted and re-
ef Gazz. chim. ital, 48 I, 198 (1913).
es Hoffmann and Kruss Ber., 20, 2369 (1887).
Schneider: Ber., 24, 2241 (1891).
^o Hoffmann and Kruss: Ber., 20, 2204 (1887); Ditte: Compt. rend, 120,
320 (1895).
Antony and Lucchesi: Gazz. chim. ital., 19, 545 (1889).
Z. anorg. Chem., 121, 266 (1922).
"Bull, soc chim., (2) 49, 452 (1888).
SULFIDES OF GOLD 249
fleeted light. The precipitate formed by agglomerating the sol was
reddish brown when moist and black when dry.
An alcosol was prepared by mixing the hydrosol with three times
its volume of alcohol, followed by dialyzing against absolute alcohol
to remove the water. The alcosol was similar in appearance to the
hydrosol but more stable.
On account of the alleged instability of Au 2 S 3 in the presence of
water, this salt was first prepared in the dry way or precipitated from
non-aqueous media. Thus Antony and Lucchesi 71 passed dry hydro-
gen sulfide over lithium auric chloride at 10, and Hofmann and
Hochtlen 74 precipitated the compound as a dark brown mass by the
interaction of auric chloride and hydrogen sulfide in absolute ether.
Gutbier and Durrwachter 72 claim to get Au 2 S 3 instead of Au 2 S 2 by
passing a rapid stream of hydrogen sulfide into an aqueous solution
of gold chloride at 2. With a slow stream of hydrogen sulfide at
100 the trisulfide first formed reacts with the excess of gold chloride
giving metallic gold. The precipitate obtained under intermediate con-
ditions is said to be a mixture of Au 2 S 3 and Au, no Au 2 S 2 being
formed by any modification of the procedure. These observations
raise the question of the individuality of Au 2 S 2 and suggest that the
alleged Au 2 S 2 precipitate and sol referred to in a preceding para-
graph are mixtures of Au 2 S 3 and Au. An Au 2 S 3 sol may be pre-
pared in concentrated phosphoric acid (p. 174).
THE COLLOIDAL SULFIDES OF THE PLATINUM FAMILY
Of the sulfides of the platinum family of elements, PdS, OsS 4 ,
Ir 2 S 3 , and PtS 2 are known to form colloidal solutions. The sols of
palladium monosulfide, osmium tetrasulfide, and platinum disulfide are
obtained by conducting hydrogen sulfide into dilute solutions of pal-
ladous chloride, osmium tetroxide, and platinic chloride, respectively. 78
Of these sols, platinum disulfide is the most readily formed and the
most stable. Indeed, the tendency of the salt to go into the colloidal
state is so great that electrolytes such as magnesium chloride must be
added to the solution before conducting in hydrogen sulfide in the esti-
mation of platinum as sulfide. 75 Freshly formed iridium sesquisulfide,
thrown down from a solution of the oxide by hydrogen sulfide, is
carried into colloidal solution by thorough washing. 78
^Ber, 37,245 (1904).
"Ivanov: J. Russ. Phys.-Chem. Soc., 48, 527 (1916); Gaze: Chera. Zentr,
I, 464 (1913).
Berzelius: "Lehrbuch dcr Chemie," 3rd ed., 8, 222 (1834).
CHAPTER XIII
THE COLLOIDAL SULFIDES OF ZINC, CADMIUM, AND MERCURY
COLLOIDAL ZINC SULFIDB
The Precipitated Salt
Zinc sulfide is obtained by precipitating neutral, slightly acid, or
alkaline solutions of zinc salts by hydrogen sulfide or alkali sulfides.
Four gels formed at room temperature by precipitating normal zinc
sulfate with hydrogen sulfide in the presence of (1) sodium hydroxide,
(2) ammonium hydroxide, (3) acetic acid, and (4) no foreign elec-
trolyte, were all found by Levi and Fontana * to be crystalline to x-rays,
giving the sphalerite diffraction pattern. This was confirmed and
extended in the author's laboratory by Milligan and Ekholm starting
with various zinc salts : sulfate, nitrate, chloride, bromide, iodide, and
acetate. The different behavior of zinc sulfide precipitates obtained
during analytical procedures depends on varying states of aggregation
since the primary particle size is approximately the same irrespective
of the ^H value of the solution from which the sulfide separates. Di-
gestion for several days on the water bath gives preparations that
show sharp interference lines in the x-ray diffraction pattern. The
crystals are submicroscopic, for Allen, Crenshaw, and Merwin 2 found
no microscopic evidence of crystalline structure in any gel precipitated
and digested under ordinary laboratory conditions. Microscopic crys-
tals are formed only by prolonged digestion of the gel in platinum tubes
with sodium sulfide or sulfuric acid in the presence of an excess of
hydrogen sulfide at temperatures varying from 200 to 350. The crys-
tals obtained by digestion with sodium sulfide are always the common
sphalerite or blende whereas both sphalerite and wurtzite are formed
in the digestion with acid. At a given temperature the amount of
wurtzite formed is greater the higher the acid concentration, and at a
given acid concentration the amount of sphalerite is greater the higher
the temperature.
lAtti accad. Lincei, (6) 7, 502 (1928); cf. Bohm and Niclassen Z. anorg.
Chem., 182, 1 (1924).
* Allen, Crenshaw, and Merwin: Am. J. Sci, (4) 84, 351 (1912).
250
THE PRECIPITATED SALT
251
Owing to the solubility of gelatinous zinc sulfide, the salt is not
precipitated completely unless the hydrogen ion concentration is rather
low. Allen, Crenshaw, and Merwin 8 found, however, that precipita-
tion takes place on long standing from solutions which are quite
strongly acid. Some data represented graphically in Fig. 41 show that
precipitation is complete in 10 days or less from 3 AT H 2 SO4 anc ^ * s
98.596 complete from 0.4 N H 2 SC>4. The time interval before pre-
cipitation starts increases with increasing concentration of acid. In
spite of the slowness with which the precipitate forms, the product is
12
"08
02
06 10 14
H 2 S0 4 after Precipitation, Normality
18
FIG 41. Influence of free acid on the precipitation of zinc sulfide.
always a gel containing no microscopically visible crystals; on the
other hand, the gel obtained by slow precipitation or by aging con-
tains much larger particles than a fresh, rapidly formed gel and is
much less soluble. 4
In analytical practice the best hydrogen ion concentration for com-
plete precipitation is between approximately pH = 2 and pH = 3. 5
At much lower />H values the precipitation is too slow or is incom-
plete, and at values higher than />H = 3 the precipitation is so rapid
a Am. J. Sci., (4) 84, 355 (1912) ; Glixelli: Z. anorg. Chem., 55, 297 (1907) ;
Kolthoff and van Dijk: Pharm. Weekblad, 69, 1351 (1922).
Bruner and Zawadzki: Bull. acad. sci. Krakow, 296 (1909); Z. anorg.
Chem., 65, 136 (1910); Bruni and Padoa: Atti accad. Lincei, (5) 14 II, 525
(1905); Glixelli: Z. anorg. Chem., 65, 297 (1907); Krokowski: Roczniki Chem.,
18, 561 (1933) ; Kolthoff and Moltzau: Chem. Rev., 17, 293 (1935).
"Fales and Ware: J. Am. Chem. Soc., 41, 488 (1919).
252 COLLOIDAL SULFIDES OF ZINC, CADMIUM, AND MERCURY
that the slimy gel produced is very difficult to filter. Higher acid con-
centrations can be used, of course, provided that the hydrogen sulfide
pressure is increased. 6 Quantitative precipitation is usually carried
out in solutions of acetic acid (containing sodium acetate), 7 tartaric
acid, 8 formic acid, 9 thiocyanic acid, 10 or dilute mineral 1 acids. The
precipitates are frequently so gelatinous that filtration is difficult
or slow. Jeffreys and Swift X1 obtained a dense granular mass by pre-
cipitation at 60 to 90 at a />H value of 1.6 maintained by a sulfate-
acid sulfate buffer; the precipitation was quantitative in 25 to 40
minutes. Similarly, Mayr 32 obtained a granular precipitate by carry-
ing out the process in a chloroacetic acid-acetate buffer at a pH value
of 2.6. Schilling 1S claims to get a more granular precipitate from a
solution containing benzene sulfonic acid.
Zinc sulfide is carried down from strongly acid solution by the
second group sulfides, copper (p. 236), mercury (p. 262), bismuth,
lead, 14 and tin. 14 Kolthoff claims that this phenomenon is the result
of post-precipitation with the first three salts (cf. p. 256). Similarly
zinc sulfide is contaminated by the sulfides of manganese, cobalt, nickel,
and iron, 15 when precipitated in their presence. It is an open question
to what extent these examples of "induced" precipitation are post-
precipitation effects and to what extent they result from coprecipitation
or mixed crystal formation. Schnasse 10 who worked with manganese
and zinc mixtures obtained evidence from x-ray analysis which indi-
cated mixed crystal formation. Under the conditions of precipitation
employed, a miscibility gap was indicated between 20 and 84 atomic
per cent of manganese; precipitates with compositions corresponding
to the gap were found to consist of two separate phases. In any
event, the contamination may be reduced to a minimum in actual
Bruni and Padoa: Atti accad Lmcei, (5) 14 II, 525 (1905)
'Villiers- Compt. rend., 108, 236 (1889).
s Alt and Schulze- Ber , 22, 3259 (1889).
'Pales and Ware- J. Am. Chem. Soc, 41, 487 (1919).
10 Zimmermann Ann, 199, 1 (1819).
11 J. Am. Chem. Soc, 64, 3219 (1933).
12 Z. anal. Chem., 92, 166 (1933); 96, 273 (1934); cf Frers- 95, 1, 113, 138
(1933).
"Chem-Ztg., 86, 1352 (1912).
"Lassieur: Chimie & Industrie, special No. 153 (1932).
"Funk: Z. anal. Chem., 46, 93 (1907); cf Kato: J. Chem. Soc Japan, 64,
867 (1933) ; 66, 293, 1148 (1934) ; Kling, Lassieur, and Lassieur: Compt rend.,
180, 517 (1925); Ruff: Z. anorg Chem., 186, 387 (1930).
Schnasse: Z. physik. Chem., B20, 89 (1933).
ZINC SULFIDE SOL 253
analytical procedures by a suitable adjustment of the concentration,
temperature, and pVL of the solution to be analyzed.
A zinc sulfide gel is obtained by the action of sodium thiosulfate
on a solution of zinc salt in the cold or at 100 , 2 and by the alternating-
current electrolysis of sodium thiosulfate solution with zinc elec-
trodes. 17 It is also formed by direct-current electrolysis of an am-
monium chloride solution using a zinc cathode and an anode of zinc
coated with sulfur. 18 Liesegang rings of both zinc and cadmium sul-
fides can be formed by precipitation in gelatin or agar jelly. 10
The freshly formed gel of zinc sulfide contains a great deal of
adsorbed water. If dried under suitable conditions, compositions cor-
responding to hydrates such as 2ZnS H 2 O, 3ZnS H 2 O, 4ZnS H 2 O,
3ZnS 2H 2 O, and ZnS H 2 O can be obtained; but it is probable that
each one is merely a hydrous sulfide whose composition is the acci-
dental result of the conditions of drying.
Owing to the solubility of zinc sulfide in acid solution, fibers im-
pregnated with the salt may be employed in the microchcmical detec-
tion of a number of metals which give colored sulfides. 20 Chamot and
Cole 2l dipped swollen wool fibers alternately in zinc acetate, which is
fairly strongly adsorbed by the fibers, and sodium sulfide, thereby
obtaining fibers which are sensitive to 0.001 mg of copper. To make
the test, a drop of the unknown solution is placed on an object glass
and acidified with a drop of dilute hydrochloric acid; in the drop is
placed a small piece of the impregnated wool fiber which is subse-
quently examined under the microscope for color change.
Zinc sulfide is a valued white pigment, both alone and when mixed
with barium sulfate to give the commercial product known as litho-
pone The properties and application of zinc sulfide pigments will be
considered in a later chapter (p. 281).
Zinc Sulfide Sol
A gel of zinc sulfide freshly formed in the cold is easily peptized
by removing the adsorbed salts by washing. 82 Even when precipitated
"Le Blanc and Schick: Z. physik. Chem., 46, 213 (1903).
is Griffith: Ger. Pat 332,199 (1921); Gmelin: "Handbuch anorg. Chem./ 1
(8) 32, 196 (1924); cf., also, Jacolliot: French Pat 415,605 (1910); Chem.
Abstracts, 6, 1884 (1912).
i^Daus and Tower: J. Phys. Chem., 33, 605 (1929).
aoEmich and Donau: Ann, 361, 432 (1907).
"Ind. Eng. Chem, 10, 48 (1918).
"Donnini: Gazz. chim. ital., 24 I, 219 (1894).
254 COLLOIDAL SULFIDES OF ZINC, CADMIUM, AND MERCURY
hot and allowed to stand several hours, the quantitative analyst must
have ammonium sulfide or ammonium chloride in the wash water to
prevent sol formation. For obtaining more concentrated sols, the gel
is thrown down from an ammoniacal solution with hydrogen sulfide,
washed by decantation, and suspended in water through which hydro-
gen sulfide is passed until peptization is complete. 28 This procedure
may be modified by suspending freshly precipitated hydrous zinc oxide
in water into which hydrogen sulfide is conducted until the oxide is
completely transformed into an opalescent zinc sulfide sol. The excess
hydrogen sulfide may be removed by heating to boiling, but prolonged
heating decreases the stability and may cause precipitation. In the
latter event, repeptization can be effected by treating with hydrogen
sulfide, but the process requires more time with the aged gel than with
a freshly formed one. 24 The peptizing action of hydrogen sulfide in-
creases with the pressure up to 1.5 to 2 atmospheres, above which it
apparently decreases slightly. 25
A pale opalescent sol is formed by the action of NaHS on a dilute
solution of zinc sulfate. 26 Muller 27 was unsuccessful in an attempt to
prepare a stable sol by the interaction of ammonium sulfide and zinc
sulfate in the presence of a large amount of glycerol; the increased
viscosity slows down the agglomeration, but the sulfide settles out after
a few days. The stability of the sol is increased enormously if pre-
pared in the presence of protecting colloids, such as gum arabic 28 and
gelatin. 29
A dilute sol appears clear in transmitted light, and a concentrated
sol possesses an orange-red color and shows a bluish fluorescence. The
gel which precipitates from the sol either spontaneously or by the
action of electrolytes is not a hydrosulfide such as 7ZnS H 2 S or
12ZnS - H 2 S 80 but is zinc sulfide with adsorbed hydrogen sulfide, the
stabilizing electrolyte in the sol.
"Wmssinger: Bull, soc chim., (2) 49, 452 (1888).
"Villiers: Compt. rend., 120, 149, 188 (1895).
Young and Goddard: J. Phys. Chem., 21, 1 (1917).
Thomsen: Ber., 11, 2044 (1878); von Zotta- Monatsh, 10, 807 (1889)
w Chem.-Ztg., 28,357 (1904).
MLefort and Thibault: J. pharm. chim., (5) 6, 169 (1882).
Alexander: U. S. Pat. 1,259,708 (1918).
80 Linder and Picton: J. Chem. Soc., 61, 114 (1892).
THE PRECIPITATED SALT 255
COLLOIDAL CADMIUM SULFIDE
The Precipitated Salt
A highly hydrous gel of cadmium sulfide is precipitated by the
action of hydrogen sulfide or alkali sulfides on a cold solution of
cadmium salt. The tendency to crystallize is more marked with cad-
mium sulfide than with the corresponding zinc salt; hence the gel
formed at room temperature gives fairly sharp x-ray interference
rings. 81 Microscopic crystals result by precipitation or digestion at
higher temperatures. A gel precipitated from a solution containing
2 g CdSO 4 in 20 cc of 30% H 2 SO 4 gave crystals 0.5 mm long, identi-
cal with the mineral greenockite, after digestion in the mother liquor
for 3 days at 180. M Similarly, a gel digested at 150 to 200 with
ammonium sulfide which has a slight solvent action gave large crystals
similar to greenockite. 83
In the absence of x-ray diffraction methods of examination, Allen,
Crenshaw, and Merwin were unable to determine the structure of the
extremely minute crystals of the freshly precipitated salt and were
forced to grow them to a size that could be observed optically, either
by heating the dry powder or by digesting the mass under pressure at
high temperatures. This always yielded hexagonal crystals like
greenockite ; but Bohm and Niclassen 8 * showed by x-ray diffraction
methods that the yellow precipitate thrown down from cadmium sul-
fate solution was cubic, being similar to cubic zinc blende. This was
confirmed by Ulrich and Zachariasen, 85 who showed that cubic or
0-CdS (a = 5.820 A) was precipitated from a saturated cadmium
sulfate solution by hydrogen sulfide ; on heating this yellow cubic sul-
fide to 700-800 in the presence of sulfur vapor, hexagonal or -CdS
resulted.
Milligan S6 in the author's laboratory confirmed and extended these
observations (p. 295). Cubic or 0-CdS chiefly was obtained from hot
solutions of cadmium sulfate and from both hot and cold solutions of
acidified sulfate and nitrate. On the other hand, hexagonal or -CdS
was thrown down chiefly from cadmium chloride, bromide, and iodide,
aiHaber: Ber., 55B 9 1730 (1922).
az Allen, Crenshaw, and Merwin: Am. J. Sci , (4) 84, 362 (1912); cf.
Baubigny: Compt. rend., 142, 577 (1906).
"Stanek: Z. anorg. Chem., 17, 117 (1898).
a*Z. anorg. Chem., 182, 1 (1924).
Z. Krist., 82, 260 (192S).
a Milligan: J. Phys. Chem., 88, 797 (1934).
256 COLLOIDAL SULFIDES OF ZINC, CADMIUM, AND MERCURY
both hot and cold and with and without the presence of the correspond-
ing acid.
For complete precipitation of cadmium sulfide, the solution must
be saturated with hydrogen sulfide since prolonged action of acids,
even in the cold, dissolves appreciable amounts of the salt. 37 The
equilibrium constant of the reaction CdSO 4 + H 2 S *=* CdS + H 2 SO 4
from solubility determinations in approximately M H 2 SO 4 containing
a very little H 2 S is: k = 6.6 X 10- for the sulfide from CdQ 2 and
4.6 X 10- 7 for the sulfide from CdSO 4 . Approaching the equilibrium
from the left by precipitation studies, the constant is k = 16 X
10 6.38 The variation from a constant value is the result of differ-
ences both in particle size and in crystal structure.
A gel of cadmium sulfide is also obtained by the action of sodium
thiosulfate on a solution of cadmium salt at the boiling point; 89 by
the electrolysis of a solution of sodium thiosulfate containing a little
sodium chloride, with a cadmium anode and an indifferent cathode;
and by the alternating-current electrolysis of 10% Na 2 S 2 O 3 solution
with cadmium electrodes. 40 A more satisfactory electrolytic method
appears to be the electrolysis of a sodium sulfate solution with a cad-
mium anode and a cathode consisting of a mixture of copper sulfide
and sulfur. 41 The electrolytic methods have been suggested for the
technical formation of cadmium yellow and will be referred to again
in the chapter on the sulfide pigments.
Purity: Adsorption of Chloride. Precipitated cadmium sulfide is
always contaminated by anions in the solution from which it sepa-
rates. The determination of cadmium as sulfide, especially in the
presence of sufficient hydrochloric acid to ensure its separation from
zinc, has long been regarded as a useless method because of the chlo-
ride carried down with the precipitate. Treadwell 42 attributes the
contamination to the formation of a double salt, CdS CdCl2, in vary-
ing amounts depending on the acid concentration and the temperature
and pressure at which the precipitation is carried out. Egerton and
Raleigh 4S believe the precipitate to have a constant composition when
"Stall: J Am. Chem Soc, 23, 512 (1901); Bruni and Fadoa: Atti accad.
Lmcei, (5) 14 II, 525 (1905).
S8 Bruner and Zawadzki: Bull acad sci. Krakow, 296 (1909).
s'Donath: Z. anal. Chem, 40, 141 (1901).
* Richards: Trans. Am Electrochem. Soc., 1, 221 (1902).
"Lorcnz: Z anorg. Chem, 12, 442 (1896); Bernfeld: Z. physik. Chem., 20,
46 (1898).
Treadwell-Hall : "Analytical Chemistry," 2, 191 (1912); cf. f also, Hulot:
Bull. soc. chim., (4) 41, 313 (1927).
J. Chem. Soc., 123, 3019 (1923).
THE PRECIPITATED SALT
257
thrown down at 80 from a solution containing 4 cc of concentrated
HQ in 100 cc, and washed with an unspecified, definite quantity of
water. Under these conditions, the precipitate is said to contain 8.16%
of the alleged salt, CdS CdCl2 ; hence, in determining cadmium quan-
titatively by weighing the sulfide, the molecular weight of the precipi-
tate is taken to be 147.4 instead of 144.47 for pure cadmium sulfide.
Since conclusive evidence of the existence of a definite double salt
CdS CdG 2 is lacking, Weiser and Durham 44 suggested that the con-
tamination of the sulfide gel is due to adsorption of cadmium chloride
in varying amounts depending on the conditions. To test this hypothe-
sis, hydrogen sulfide at room temperature was passed into a definite
volume of solution containing a constant amount of cadmium chloride
and varying amounts of hydrochloric acid. The precipitate was fil-
10
25'
0.1
0.2 0.3 0.4 0.5
Concentration of HCI, Normality
0.6
0.7
FIG 42 Adsorption of chloride by cadmium sulfide.
tered on a Gooch crucible, washed until the wash water gave no test
for chloride, and then analyzed for the chloride content. The results
are shown graphically in curve A of Fig. 42. The filtrates were
tested for completeness of precipitation by rendering them alkaline
with ammonia and saturating with hydrogen sulfide. With the most
acid solution a faint yellow coloration was noted but no precipitate.
The amount of cadmium required to produce this coloration was found
to be negligible as compared with the total amount of precipitate.
An attempt was made to repeat the above observations at 80 as
recommended by Egerton and Raleigh, but precipitation was found
to be far from complete. Accordingly, the several solutions were
J. Phys. Chetn., 32, 1061 (1928).
258 COLLOIDAL SULFIDES OF ZINC, CADMIUM, AND MERCURY
heated to 80 , removed from the source of heat, and hydrogen sulfide
conducted through continuously while they cooled down to room tem-
perature. The precipitates were washed and analyzed for chloride
with the results shown graphically in curve B of Fig. 42.
Referring to curve A, it will be seen that the amount of chlorine
in the precipitate does not increase continuously with the concentra-
tion of hydrochloric acid but exhibits a maximum. This maximum
corresponds to a visible change in the physical character of the pre-
cipitate from a flocculent to a distinctly granular structure. The curve
is thus a typical adsorption curve showing a maximum that results
from a physical change in the adsorbent.
Curve B for the second series of experiments lies under curve A,
as would be expected since the temperature at which the precipitation
starts is higher, hence the precipitate is more granular and the adsorp-
tion is less. With increasing concentration of acid, the adsorption of
chloride increases so that the first part of curve B is very similar to
that of curve A, and for the same reason. But when the acid reaches
a concentration in the neighborhood of 0.3 N the amount of cadmium
sulfide precipitated in the hot decreases with a corresponding increase
in the amount precipitated at lower temperatures. This means a
larger amount of finely divided particles and a correspondingly greater
adsorption which rises to a second maximum. This behavior would
be difficult to explain on the basis of double salt formation but is
readily accounted for by considering the contamination as a case of
adsorption. The final concentration of acid recommended by Egerton
and Raleigh was approximately O.S N. This might appear to account
for the constancy of composition of their precipitates since curve B
is relatively flat when the HC1 concentration is in the neighborhood
of 0.5 N. However, the chlorine content under these conditions cor-
responds to 12.2% of the alleged double salt instead of to 16.5% as
calculated from Egerton and Raleigh's results; the difference is, of
course, in the conditions of the precipitation.
Precipitated cadmium sulfide may carry down with it the sulfides
of other elements. Evidence from x-ray analysis indicates that the
coprecipitation effect with mercury 45 and manganese 48 results from
mixed crystal formation.
Color. In general, the color of cadmium sulfide gel is light yellow
when thrown down from cold solutions of low cadmium content by
"Bottger and Druschke: Ann., 463, 315 (1927); Ahrens: Dissertation,
Leipzig (1933) ; Kolthoff and Moltzau: Chem. Rev., 17, 293 (1935).
"Schnasse: Z. physik. Chem., B20, 89 (1933).
CADMIUM SULFIDE SOL 259
hydrogen sulfide or by alkali sulfides, whereas the color is a deep
orange when precipitated from hot acid cadmium solutions with hy-
drogen sulfide or by boiling cadmium solutions with an excess of
sodium thiosulfate. Microscopic crystals likewise vary in color from
clear yellow to orange. Because of the importance of cadmium sulfide
as a pigment, 47 the factors influencing the color will be considered in
some detail in the chapter on the sulfide pigments (p. 295).
Cadmium Sulfide Sol
Preparation. A sol of cadmium sulfide is best prepared by pre-
cipitating an ammoniacal solution of cadmium sulfate or cadmium
chloride 48 with hydrogen sulfide, washing the precipitate by decanta-
tion, and suspending in water through which hydrogen sulfide is con-
ducted until peptization is complete. The excess hydrogen sulfide can
be removed by boiling.
Sols may be obtained directly by conducting hydrogen sulfide into
00004 AT CdSO 4 49 or 0.025 N Cd(C 2 lI 3 O 2 ) 2 . BO The stability of the
preparations may be increased greatly by the presence of casein, gum
arabic, 81 soaps, 52 and probably by sugar, 58 which is adsorbed by the
colloidal particles.
A benzine sol of cadmium sulfide may be prepared by triturating
the gel with heavy oil and shaking the mixture with the light petroleum
distillate. 64 Oleates may be used as protecting colloids in the prepara-
tion of zinc and cadmium sulfide sols in benzene, toluene, and linseed
oil. 65
Properties. The sol prepared by Frost's method is a beautiful
golden yellow in transmitted light and is somewhat fluorescent in re-
flected light. Dilute sols are quite stable for a considerable time, but a
preparation containing as much as 11 g/1 precipitates in 24 hours. As
in all sulfide sols, the particles are negatively charged owing to prefer-
ential adsorption of S and HS~~ ions. The ratios of the precipi-
47 See Bugden: Continental Met. & Chem. Eng., 2, 109 (1927).
Prost: Bull, acad roy. mid. Belg., (3) 14, 312 (1887); J. Chem. Soc.,
64, 653 (1888).
*Meldrum: Chem. News, 79, 170 (1899); Vanino and Hartl: Bcr, 37,
3622 (1904).
5Bialek: Roczniki Chem., 14, 1499 (1934).
51 Muller and Artmann: Oesterr. Chem -Ztg., 7, 149 (1904).
"Bhatnagar, Prasad, and Bahl: J. Indian Chem. Soc., 2, 11 (1925).
"Prasad, Shrivastava, and Gupta: Kolloid-Z., 87, 101 (1925).
6 *Van Dorp and Rodenburg: Chem. Weekblad, 6, 1038 (1909).
B5 Bechhold and Szidon: Kolloid-Z. (Zsigmondy Festschrift), 86, 259 (1925).
260 COLLOIDAL SULFIDES OF ZINC, CADMIUM, AND MERCURY
tating powers of the sul fates of potassium, aluminum, and cadmium
are 1 : 140: 150. The high precipitating power of cadmium salts doubt-
less results from the removal of the stabilizing electrolyte, hydrogen
sulfide. Plant and animal fibers are 1 colored by colloidal cadmium
sulfide ; but in this form the color is not adsorbed sufficiently strongly
to serve as a dye.
Bialek 50 followed the coagulation of cadmium sulfide sol with a
spectrophotometer, and found that the process could be represented
by the equation K(t t s ) = In (jr/1 jr), where K is the velocity
constant; x, the relative degree of turbidity; t, the time; and t 9 , the
half-time period of coagulation. The process was autocatalytic, larger
particles hastening the coagulation of smaller ones. The coagulating
powers of several cations followed the usual order: Li < Na < K <
NII 4 < Rb < Cs; Ca < Sr < Ba.
A study of the viscosity and rigidity of copper ferrocyanide and
cadmium sulfide sols ce disclosed that some sols possess true rigidity,
that is, they support small stresses permanently without yielding. The
sols were found to exhibit a linear relationship between stress and
strain, behaving like perfectly elastic bodies under the experimental
conditions.
COLLOIDAL MERCURIC SULFIDE
The Precipitated Salt
Formation. Black mercuric sulfide is thrown down in a hydrous
form by the action of excess hydrogen sulfide on a mercuric salt in an
acid or neutral solution. Even when freshly precipitated at 0, the
sulfide gives an x-radiogram which shows that the crystals possess the
cubic structure similar to zinc blende and identical with the mineral
mela-cinnabarite.
Black mercuric sulfide is also obtained by the action of mercuric
chloride on a solution of sodium thiosulfate at concentrations in which
the ratio of mercuric chloride to sodium thiosulfate lies between 2 : 3
and 1:4; at higher thiosulfate concentrations, red mercuric sulfide is
precipitated. 2 The black gel is formed by the alternating-current elec-
trolysis of hot sodium thiosulfate solution with mercury electrodes. 57
Since the alternating current causes the mercury to vibrate, the surface
of the mercury is kept free from precipitated sulfide; hence, the cur-
rent efficiency is much higher than obtains in the alternating-current
McDowell and Usher: Proc. Roy. Soc. (London), 131A, 409 (1931)
"Weiser: ] Phys. Chem., 22, 77 (1918).
THE PRECIPITATED SALT
261
electrolysis with zinc and cadmium electrodes under similar conditions.
Moreover, the mercury electrodes are uniform and the experimental
conditions are therefore reproducible. In Fig 43 the current efficiency
at varying concentrations of electrolyte and at varying current den-
sities is shown for a 60-cycle alternating current. The marked falling
off in the efficiency with thiosulfate solutions stronger than 35^ is
due to the rapid action of the hot electrolyte on the mercury, forming
a film which prevents the rhythmical pulsations of the surface.
Red mercuric sulfide or cinnabar is the stable modification at all
temperatures up to its sublimation point, about 580. It is formed
15 25 35 45
Concentration of Na,S 2 3 , Percent
55
FIG 43 Efficiency of the electrolysis of sodium thiosulfate solutions at nieicury
electrodes with a 60-cycle alternating current.
from the black gel by digesting the gel with alkali or ammonium sul-
fide. The process consists in the solution of the black regular form
and the subsequent precipitation of the less soluble red hexagonal
modification. The red sulfide precipitates directly when Hg(SH)CNS
is boiled with concentrated NH 4 CNS or when hydrogen sulfide is con-
ducted into a warm mercuric salt solution in the presence of acetic
acid and excess NH 4 CNS or CS(NH 2 ) 2 . 58 The interaction of mer-
curic chloride and concentrated sodium thiosulfate gives a red sulfide
which Allen, Crenshaw, and Merwin took to be a third allotropic
modification; but x-radiograms proved it to be identical with cinna-
58 Venkataramaiah and Rao. J. Sci. Assoc. Maharajah's Coll, 1, 41 (1923);
Chem. Abstracts, 18, 626 (1924); cf. Alvisi: Atti accad. Lmcei, (5) 7 II, 97
(1898).
262 COLLOIDAL SULFIDES OF ZINC, CADMIUM, AND MERCURY
bar. 58 When properly prepared, red mercuric sulfide furnishes a valu-
able red pigment known as vermilion (p. 297).
Contamination. When mercuric chloride is precipitated with hy-
drogen sulfide or alkali sulfide, the precipitate may be first white, then
yellow, and finally black. The white precipitate is 2HgS HgQ 2 , the
black is HgS, and the yellow, a mixture of the two. 60 If a small
excess of sodium sulfide is used as precipitant, a sol is formed which
is yellow at first and then changes to black. 61 In the precipitation of
alkali halides (X = Cl, Br, or I), Smith and Semon 62 suggest that five
consecutive reactions take place with the formation of four intermedi-
ate compounds: [Hg(SH) 2 Hg]X 2 , [HgSHHg]X 2 , [Hg(SHg) 2 ]X 2 ,
and [Hg(SHg) 2 ] SH 2 . Fenimore and Wagner 68 believe that the high
results which they obtain in the estimation of mercury as sulfide may
result from the incomplete conversion of the alleged intermediate com-
pounds into HgS. It will be recalled that Weiser and Durham at-
tributed the contamination of cadmium sulfide with chloride to ad-
sorption.
Mercuric sulfide is contaminated by third-group sulfides when
precipitated in their presence even when the acidity is such that mer-
curic sulfide alone would be expected to precipitate. Moreover, a
cadmium solution so strongly acid that no precipitate results with
hydrogen sulfide will precipitate a great deal of cadmium on the addi-
tion of a mercury solution. The explanations that have been advanced
to account for the induced precipitation have already been given
(p. 2S2). The contamination of mercuric sulfide by zinc has received
special consideration by Kolthoff, 64 who attributes the phenomenon to
post-precipitation. Freshly precipitated mercuric sulfide on which zinc
sulfide is post-precipitated is quite as black as the pure compound, but
it is much more slimy and difficult to filter. Aging of the black sulfide
cuts down its tendency to promote the precipitation of zinc. A certain
"aged" condition is reached more quickly at higher than at lower tem-
peratures, possibly because of an increased tendency toward transfor-
mation to the cinnabar structure at higher temperatures. The same
explanation doubtless accounts for the greater promoting action of
the sulfide aged in acid solutions.
"Kolkmeijer, Bijvoet, and Karssen: Rec. trav. chim., 43, 894 (1924).
"Jolibois and Bouvier: Compt. rend., 170, 1497 (1920).
"Morosow: Kolloid-Z., 86, 21 (1925).
"J. Am. Chem. Soc., 46, 1325 (1924).
* J. Am. Chem. Soc., 68, 2453 (1931).
*Chem. Weekblad, 81, 102 (1934); Kolthoff and Molzau: Chem. Rev., 17,
293 (1935); J. Phys. Chem., 40, 779 (1936).
MERCURIC SULFIDE SOLS 263
At higher acidities, the amount of zinc sulfide entering a mercuric
sulfide precipitate in a given time falls off much more slowly with
increasing acidities than would be expected from solubility relations.
With freshly precipitated mercuric sulfide and a certain degree of
acidity, an appreciable quantity of zinc enters the solid phase within
the first 30 minutes or so after saturating with hydrogen sulfide, after
which the precipitation takes place more slowly to a point correspond-
ing to the solubility of well-aged zinc sulfide. At lower acidities the
presence of a very small amount of mercuric sulfide will lead to the
precipitation of relatively large amounts of zinc sulfide within a period
of time such that no precipitate would result from a solution of zinc
salt alone; at higher acidities (2 N H^SC^), the amount of zinc sul-
fide entering the precipitate in a given short time of shaking is approxi-
mately proportional to the amount of fresh mercuric sulfide present.
Keeping the acidity and the amount of mercuric sulfide constant, the
amount of zinc carried down is approximately proportional to the
amount of zinc salt in solution over a limited range.
Kolthoff attributes the results above described to (1) adsorption
phenomena and (2) true precipitation, in varying degrees depending on
the acidity. During the initial period, adsorption is the primary proc-
ess at both high and low acidities; and at high acidities, where the
rate of true precipitation is very slow, adsorption phenomena are like-
wise responsible for the contamination. The mechanism is as follows :
mercuric sulfide adsorbs sulfide ions primarily with hydrogen ions the
counter ions in the double layer; zinc ions in contact with the sulfide
enter into exchange adsorption with hydrogen ions thus contaminating
the precipitate with zinc sulfide. The presence of strychnine and
aluminum ions which will undergo the same exchange adsorption phe-
nomena will cut down the adsorption of zinc ions. Hydrogen ions in
high concentration will likewise cut down greatly the adsorption of
zinc ions by exchange. This accounts both for the smaller adsorption
of zinc at high acidities and for the fact that the contamination is
approximately proportional to the zinc ion concentration at high
acidities.
Mercuric Sulfide Sols
Preparation. The hydrosol of mercuric sulfide is easily prepared
by washing the black loose precipitate with water and suspending in
water through which a rapid current of hydrogen sulfide is passed.' 65
If the gel has aged until it possesses a metallic appearance with a
eBFreundlich and Schucht: Z. physik. Chem., 86, 641 (1913).
264 COLLOIDAL SULFIDES OF ZINC, CADMIUM, AND MERCURY
grayish tinge, peptization is accomplished with difficulty if at all.
Excess hydrogen sulfide may be removed by boiling, but it is better to
wash with hydrogen. The concentrated sol is deep black whereas the
dilute sol is brown with a greenish fluorescence in reflected light.
A sol results on conducting hydrogen sulfide into a cold saturated
solution of mercuric cyanide. 86 Owing to low ionization of hydro^
cyanic acid, it has only a slight precipitating action; but the sol is
more stable when the impurity is removed, preferably by distillation
under reduced pressure in an atmosphere of hydrogen sulfide. An
alcosol can be prepared in a similar way by conducting dry hydrogen
sulfide into an alcoholic solution of mercuric cyanide 67 and dialyzing
against pure alcohol, using membranes of parchment or collodion. 68
Weiser and Mack 89 obtained sols in methyl, ethyl, and w-propyl alco-
hols and in acetone by suitable modifications of this procedure.
Lottermoser 70 prepared a hydrosol by adding sodium sulfide drop-
wise to 1 cc of saturated mercuric chloride until the precipitate just
redissolved and pouring the resulting solution of alkali mercurisulfide
into 1 1 of distilled water. It is also formed by the action of the silent
electric discharge on a suspension of mercuric sulfide (p. 7).
The sodium salts of lysalbinic and protalbinic acids may be used
as protecting colloids in preparing a stable sol for medicinal purposes. 71
When injected intravenously, the sol is quite rapidly coagulated and
becomes attached to the tissues. If the injection is subcutaneous, the
sulfide remains in the subcutaneous tissue; but if it is injected intra-
venously, the salt appears in the liver, spleen, bone marrow, and lungs,
from which it is absorbed slowly. Its administration is said to facili-
tate the healing of syphilitic lesions, 72 but it is not a substitute for
arsenicals. 78
Coagulation and Adsorption. The adsorption of cations during the
precipitation of negative mercuric sulfide 7 * is less than with the more
hydrous arsenic trisulfide sol. Freundlich and Schucht 65 determined
MLottermoscr: J. prakt. Chem., (2) 75, 293 (1907).
<"Errera- Kolloid-Z., 82, 240 (1923); cf. Volkov and Glazman: Chem. Ab-
stracts, 30, 5483 (1936).
"Ostwald and Wolski: Kolloid-Z., 28, 228 (1921).
w J. Phys. Chem., 84, 86, 101 (1930).
70 Kolloid-Z., 66, 188 (1934); Lottermoser and Hessling: 75, 184 (1936).
"Wolvekamp: Brit. Pat 188,772 (1921).
"Quattrini: Chem. Abstracts, 10, 640 (1916); Sabbatani: 11, 1206 (1917).
"Gennerich: Am. J. Syphilis, 16, 198 (1932).
^Kruyt and van der Willigen: Z. physik. Chem., ISO, 170 (1927); (cf.
this volume, p. 203).
MERCURIC SULFIDE SOLS
265
the precipitation values for a number of salts and the adsorption of
cations at the precipitation value in an attempt to show that equivalent
amounts of all cations are adsorbed at this concentration ; but since
the adsorption values vary all the way from 0.004 to O.OSO m.eq./g
(Table XXXIV), the results disprove the assumption. The precipita-
TABLE XXXIV
ADSORPTION OF CATIONS DURING THE PRECIPITATION OF MERCURIC SULFIHK SOLS
Adsorption at pre-
Precipitation
Concentration
Cation
cipitation value,
value,
of colloid,
milhequivalent
millimols
grams per liter
NH<
050
10 20
13 74
Ag
020
28
11 74
New magenta
008
097
13 74
Brilliant green
004
048
8 38
Auramme
Oil
094
10 05
Methylene blue
007
097
14 96
Ba
044
510
8 29
CulCu(NOi)jl
030
150
8 26
Cu(CuSO 4 ).
022
260
14 43
Ce
012
082
10 45
tion values of new magenta and methylene blue are identical, and the
adsorption values are almost the same ; on the other hand, the precipi-
tation value of auramine is but 3% lower than that of new magenta,
and the adsorption value is Z7% higher. Moreover, the precipitation
value of brilliant green is only one-half that of new magenta, and the
adsorption value at this concentration is but half as much instead of
being the same or larger in accord with Freundlich's conclusions from
experiments on the adsorption of organic cations by arsenic trisulfide.
There is no doubt that the anion has some effect, but this cannot ac-
count for the precipitation value of cupric sulfate being 75% higher
than for cupric nitrate and the adsorption of copper from nitrate solu-
tion being 25% greater than from sulfate solution.
The cause of the variation from equivalent adsorption has been
considered in some detail in an earlier , chapter (p. 204) ; but no men-
tion was made of the much greater variation with mercuric sulfide than
with arsenic trisulfide. When mercuric sulfide sol is first flocculated
it is quite gelatinous but it becomes granular rapidly and gives up a
part of the adsorbed electrolyte. It is altogether probable that the
266 COLLOIDAL SULFIDES OF ZINC, CADMIUM, AND MERCURY
physical character of the precipitated sulfide varies appreciably with
different electrolytes, thus accounting for the greater variation from
equivalent adsorption by mercuric sulfide than by arsenic trisulfide.
Reversal of Adsorption. The course of the adsorption reversal of
mercuric sulfide has been followed by Freundlich. 75 Portions of 250 cc
each of a sol were treated with 25 cc of auramine of varying con-
centrations. The mixtures were stirred continuously, and at inter-
vals samples were withdrawn, freed from suspended sulfide, and an-
alyzed colorimetrically. The S-shaped form of the time-desorption
curves indicates that the adsorption reversal and, hence, the coarsening
of the particles are autocatalytic processes. Furthermore, the velocity
of the processes rises rapidly with increasing concentration of the co-
agulating dye. The mechanism is about as follows: When sufficient
electrolyte is added to a sol, the charge on the colloidal particles is
reduced below the critical value necessary for agglomeration. If the
electrolyte concentration is near the precipitation value, the primary
particles of the floes still possess a slight charge which tends to keep
them from coalescing; but with crystalline particles which are not very
hydrous, like those of mercuric sulfide, the low residual charge is not
sufficient to prevent a gradual coalescence and packing together of the
particles, thereby giving a diminished surface and a lessened adsorption.
The rate of coalescence increases with diminishing residual charge and
hence is greater the higher the concentration of precipitating electrolyte
up to the point where charge reversal would come in. The apparent
autocatalytic course may be the result of two processes : a slow initial
formation of secondary aggregates from primary particles and a more
rapid packing together of the larger aggregates.
"Freundlich and Schucht: Z. physik. Chem., 85, 660 (1913); Freundlich
and Hase. 89, 417 (1915).
CHAPTER XIV
THE COLLOIDAL SULFIDES OF MANGANESE, NICKEL, COBALT,
IRON, AND THE RARER ELEMENTS
COLLOIDAL MANGANOUS SULFIDE
Ammonium sulfide and monosulfides of the alkali metals precipi-
tate from a solution of manganous salt, a rose or flesh-colored gel of
manganous sulfide which oxidizes readily in the air, assuming a brown
tint. The precipitation is very slow or incomplete from dilute solu-
tions unless a salt such as ammonium chloride is present. 1 It is prob-
able that this salt acts by preventing sol formation. Citrates and tar-
trates prevent the precipitation of manganous sulfide by ammonium
sulfide, and citrates stop the precipitation by alkali sulfides as well. 2
It is altogether likely that these strongly adsorbed anions act by form-
ing a stable negative sol.
The rose-colored sulfide appears to be the stable form when pre-
cipitated with alkali sulfides; but when thrown down with ammonium
sulfides it frequently assumes a green color. Conditions favorable for
the transformation have been stated by a number of people. 8 The
various methods have been analyzed critically by Mickwitz and Lande-
sen.* As a result of this analysis and some observations of their own
the following conclusions were reached: (1) The transformation from
rose to green manganous sulfide never takes place when the precipita-
ijoulin: Ann. chim. phys., (4) 30, 275 (1873).
2 How Chem. News, 19, 137 (1869).
aFresenius: J. prakt Chem., 82, 267 (1861); Muck: Z. Chem., 12, 580, 629
(1869) ; Classen: Z. anal. Chem., 8, 370 (1869) ; 16, 319 (1877) ; DeClermont and
Guiot. Bull. soc. chim., (2) 27, 353 (1877); Meineke: Z. angew. Chem., 1, 4
(1888); Mourlot: Compt. rend, 121, 202 (1895); Raab and Wesslcy: Z. anal.
Chem., 42, 433 (1903); Olsen, Clowes, and Weidmann: J. Am, Chem. Soc.,
26, 1622; Olsen and Rapelje: 1615 (1904); Villiers: Compt. rend, 159, 67
(1914); Fisher: J. Russ. Phys.-Chem. Soc., 46, 1481 (1914); Seeligmann: Z.
anal. Chem., 58, 594 (1914); 64, 104 (1915); Hahn: Z. anorg. Chem., 121, 209
(1922).
*Z. anorg. Chem., 181, 101 (1923); Landesen: 198, 277 (1930); J. Phys.
Chem., 36, 2521 (1932).
267
268 COLLOIDAL SULFIDES OF MANGANESE, NICKEL, COBALT, ETC.
tion is effected with alkali sulfides. (2) The transformation from rose
to green manganous sulfide never takes place except in the presence of
free ammonia. (3) There are two rose sulfides of manganese. One,
which will not turn green, is precipitated by NH 4 HS in the absence
of free ammonia ; its composition may be represented by the formula
H 2 Mn 3 S4 or 3MnS H 2 S. A second rose sulfide, which turns green
spontaneously, is precipitated in the presence of free ammonia; its
composition may be expressed by the formula NH 4 HMn 3 S 4 or
3MnS NH 4 HS. In the next section it will be shown that all three
of these conclusions are open to question.
Transformation from Rose to Green Manganous Sulfide
Effect of Excess Alkali Sulfide and of Ammonia. The rose sulfide
may be transformed into green in the presence of an excess of alkali
B
V
08 1.2
Excess Na 2 S.Mols per Liter
1.6
357
5.24JVNH 4 OH, Cc Added
FIG 44. Effect of the concentration of (A) sodium sulfide and (J?) ammonium
hydroxide on the rate of transformation from rose to green manganous sulfide.
sulfide slowly at room temperature and rapidly at the boiling point;
free ammonia is not necessary for the transformation. 5 Some ob-
servations of the rate of transformation of 0.435 g of the rose sulfide
to green at the boiling point, in the presence of varying amounts of
sodium sulfide, are shown graphically in part A of Fig. 44. Under the
conditions of the experiment, the rose is transformed into the green in
a short time, only if the alkali sulfide concentration is greater than
approximately 0.5 M.
"Weiser and Milligan: J. Phys. Chem., 85, 2330 (1931); 36, 2840 (1932).
TRANSFORMATION 269
In part B of Fig. 44 are given the results of observations similar
to the above on the effect of ammonium hydroxide on the rate of
transformation at room temperature of the rose sulfide precipitated
by NH 4 HS in slight excess (005 mol/1). The U-shaped curve signi-
fies that an optimum concentration of ammonia exists which is most
favorable for rapid transformation. With too little ammonia no
change takes place in a reasonable time; with too much ammonia, the
rate is retarded. It was found also that free ammonia has little or no
effect on the transformation when but little excess NH 4 HS is added.
Moreover, the greater the excess of NH 4 HS [(NII 4 ) 2 S], the greater
the amount of ammonia which must be added in order to give the criti-
cal mixture for the most rapid rale of transformation. Finally, the
greater the excess of ammonia, the higher must be the concentration of
NH 4 HS [(NH 4 ) 2 S] to inhibit or prevent the change. The more
rapid the mixing, other things being equal, the more rapid the rate of
transformation. 8
A probable explanation of the above behavior is as follows : The
change from the flocculent rose precipitate, consisting of aggregates of
finely divided particles, to the denser green granules appears to be
favored by dissolution of the rose and reprecipitation of the more in-
soluble 7 green form having a different crystal structure. The rose
sulfide appears to be almost insoluble in an excess of NH 4 HS, and, in
view of the lower solubility of manganous sulfide than of manganous
hydroxide, the former is probably but very slightly soluble in am-
monium hydroxide. On the other hand, rose manganous sulfide is
somewhat soluble in (NH 4 ) 2 S. 8 Accordingly the addition of am-
monium hydroxide to a given excess of NH 4 HS gives an optimum
concentration of (NH 4 ) 2 S which is most favorable for the transfor-
mation. With too little ammonia the change is very slow, and with
too much ammonia the adsorption is sufficiently great to form a film
around the pai tides and so to inhibit the transformation. The rate
of mixing is important since it determines the form of the precipitate.
The precipitate obtained by slow mixing is a fairly dense floe that is
not greatly protected by adsorbed ammonia, at least when the ammonia
concentration is not too high. On the other hand, rapid mixing gives
a highly dispersed precipitate, the individual particles of which adsorb
ammonia and are protected thereby so that the rate of change is greatly
retarded.
Cf. Fisher: J. Russ Phys-Chem Soc, 46, 1481 (1914).
7 Weigel: Z physik. Chem., 08, 294 (1907).
sAbegg: "Handbuch anorg Chemie," 4 (2), 724 (1913)
270 COLLOIDAL SULFIDES OF MANGANESE, NICKEL, COBALT, ETC.
Tower 9 obtained Liesegang rings of manganese sulfate in gelatin,
agar, and silica jellies. Unfortunately he does not state whether the
transformation from rose to green was prevented by the hydrophilic
colloid.
X-ray Analysis. X-ray analysis 6 of the light green sulfide ob-
tained with sodium sulfide and the dark green preparation obtained
with ammonium sulfide disclosed that the two are identical. The crys-
tals of the so-called a-sulfide are cubic of the sodium chloride type;
the value of the lattice constant, a , for the crystals is 5.20 A. 10 The
structure is identical with that of the mineral alabandite.
The rose or j8-sulfide exists in two modifications, the usual hex-
agonal modification of the wurtzite type and a cubic modification of
the zinc blende type which was overlooked by Weiser and Milligan
and found by Schnaasc. 10 The rose cubic form is obtained in strong
alkaline solution, hence by precipitation with sodium sulfide; and the
hexagonal appears from weaker alkaline solution, hence by precipita-
tion with ammonium sulfide. From manganese acetate solution and
ammonium sulfide one obtains chiefly the cubic or chiefly the hexagonal
rose sulfide, depending on the acetate concentration; the hexagonal
crystals predominate from dilute solutions and the cubic from almost
saturated solutions. The lattice constant of the /J-hexagonal form is
3.976; of the 0-cubic form, 5.600.
A red or orange manganese sulfide is said to precipitate from a
weakly acid solution of a manganese salt by the prolonged action of
hydrogen sulfide. 11 Olsen and Rapelje 12 claim that the ordinary flesh-
colored sulfide is a mixture of the red sulfide with a gray form. Ac-
tually, x-ray analysis shows the gray sulfide to be a mixture of green
a-sulfide and the red /^-modification.
Manganese, cadmium, and zinc sulfides all may assume both the
zinc blende and wurtzite structure. The two modifications of cadmium
sulfide form a complete series of mixed crystals with the respective
modifications of /3-manganese sulfide. On the other hand, with the
sulfides of zinc and manganese a miscibility gap extending from about
20-84 atomic per cent manganese is indicated. 18 Precipitates in this
9 Tower and Chapman: J. Phys. Chem., 86, 1474 (1931); Tower: 40, 599
(1936).
10 Cf. Schnaase: Z. physik. Chem., B20, 89 (1933); Naturwissenschaften,
20, 640 (1932).
"Volker: Ann., 69, 27 (1846); Fisher: J. Russ. Phys.-Chem Soc, 46, 1481
(1914).
12 J. Am. Chem. Soc, 26, 1615 (1904).
"Schnaase: Z. physik. Chem., B20, 89 (1933).
THE PRECIPITATED SALT 271
gap consist of two separate phases having compositions corresponding
to the above limits. The mixed crystals are chiefly of the zinc blende
type, but some having a wurtzite structure were also observed.
The evidence from adsorption studies and from x-ray analysis of
the precipitates argues against the view of Landesen that the rose
sulfide which turns green spontaneously is an ammonium sulfo salt
of the composition NH 4 HMn 3 S 4 or any definite compound other than
manganous sulfide. Landesen 14 attributes the failure to find his al-
leged complex compounds by x-ray analysis methods to the fact that
they are instable except in contact with the mother liquor and are
therefore decomposed by the washing and drying operations incident
to preparing the samples for analysis. This question could be settled
definitely by x-ray analysis of the fresh moist gels obtained under vary-
ing conditions.
A disulfide of manganese having the pyrite structure may be pre-
cipitated from aqueous solutions at high temperatures (around 160 ). 15
In the x-ray diagrams from certain mixtures of manganese and cad-
mium sulfides, Schnaase observed lines corresponding to the higher
sulfide of manganese and suggested that excess sulfur not removable
by carbon disulfide from certain manganous sulfide precipitates may
be present as manganese disulfide.
COLLOIDAL NICKEL SULFIDE
The Precipitated Salt
It is a well-known fact that nickel and cobalt sulfides are not pre-
cipitated by hydrogen sulfide in dilute hydrochloric acid solution, and
yet the precipitates from alkaline solution are insoluble in dilute acid
after standing for a time. Thiel and coworkers 18 1T found that dif-
ferent methods of preparation and treatment of nickel sulfide gave
products that varied greatly in their solubility in hydrochloric acid.
They were thus led to assume the existence of three so-called modi-
fications designated as a-, 0-, and y-forms having more or less definite
solubilities in hydrochloric acid of a given strength. From Thiel's
point of view the a-form of a given preparation is all the sulfide which
is quickly and easily soluble in cold 2 N HC1 ; the 0-sulfide is the part
soluble in hot 2 N HC1 ; and the y-sulfide is the remainder. It is prob-
"Z. anorg Chem, 216, 114 (1933).
"Senarmont: Ann. chim phys., (3) 80, 140 (1850).
18 Thiel and Ohl: Z. anorg. Chem, 61, 3% (1909).
"Thiel and Gessner: 86, 1 (1914) ; cf. Herz: 27, 390 (1901).
272 COLLOIDAL SULFIDES OF MANGANESE, NICKEL, COBALT, ETC.
able that these three alleged isomeric forms of nickel sulfide are merely
stages in the continuous transformation from the soluble to the in-
soluble state. There is no inversion point between any two of the al-
leged isomers, and it is quite likely that any number of substances with
intermediate properties could be formed. As frequently observed with
the hydrous oxides, the very finely divided soluble and readily peptized
particles of the newly formed precipitate go over gradually and con-
tinuously into larger and denser particles which are less soluble and
less readily peptized. This change in physical character is always more
rapid in contact with a medium which possesses a slight solvent action.
Cone, Renfrew, and Edelblute 18 19 found the initial rate of solu-
tion in a given concentration of hydrochloric acid to depend on the con-
centration of hydrogen sulfide in the system. This is probably con-
nected with the adsorption of hydrogen sulfide, which is greater for
nickel sulfide than for cadmium, lead, and iron sulfides, and increases
with the hydrogen sulfide concentration. Because of this adsorption,
nickel sulfide precipitates contain more or less excess sulfide. The op-
timum pH value for the analytical precipitation of nickel as sulfide is
4.4, which may be maintained by an ammonium acetate-acetic acid
buffer. 20 ,
Nickel Sulfide Sols
Unlike the sulfides of iron and cobalt, nickel has a marked tendency
to form colloidal solutions, especially in the presence of an excess of
ammonium sulfide. Every analyst is familiar with the dark brown
coloration which forms gradually in a nickel salt solution to which
ammonium sulfide is added. Many investigators have attributed this
coloration to the formation of a sulfo salt 21 resulting from the pres-
ence of an excess of sulfur in the ammonium sulfide reagent. This
view was proved to be erroneous by Thiel and Ohl, 16 who obtained the
familiar brown-colored solution using ammonium sulfide entirely free
from polysulfide. They attributed the gradual development of the
brown color to polymerization of nickel sulfide with the ultimate for-
mation of a sol. As already indicated, it is likely that the phenomenon
results from gradual growth of the sulfide particles rather than from
is J. Am Chem. Soc, 67, 1434 (1935).
Cf. Middleton and Ward: J. Chem. Soc, 1459 (1935).
"Haring and Westfall: J. Am. Chem. Soc, 52, 5141 (1930).
siLecremer: Chem.-Ztg., IS, 431, 449 (1889); de Koninck and Ledent: Z.
angew. Chem., 5, 203 (1891); de Koninck: Compt. rend., 120, 735 (1895);
Villiers: 119, 1208, 1263 (1894); Antony and Magri: Gazz. chim. ital., 31 II,
265 (1901).
NICKEL SULFIDE SOLS 273
the formation of polymers. There is no doubt, however, that the color
results from nickel sulfide in colloidal solution. Thorne and Pates M
proved this conclusively when they ultrafiltered out the brown color
completely from a sol prepared with the ordinary laboratory reagents.
An investigation of the factors influencing the formation of the
brown sol was made by passing hydrogen sulfide directly into a nickel
ammonium hydroxide solution, thereby avoiding the presence of any
polysulfide. When the ammonia concentration is 100 g/1, a sol is
formed with a nickel concentration between 2.0 and 0.01 g/1. Some
precipitate results on standing if the nickel concentration exceeds
1.0 g/1; below 0.1 g/1 no coloration is observed. The range of nickel
concentration between which a sol is formed is determined by the am-
monia concentration. For a sol containing 0.3 g of the metal per liter,
the ammonia concentration may be reduced to 45 g/1 ; for 0.2 g nickel,
10 g ammonia per liter; and for 0.1 g nickel, 5 g ammonia per liter.
In every case, the sol must be kept saturated with hydrogen sulfide
to prevent any precipitation ; if the hydrogen sulfide is removed com-
pletely, precipitation results, but the precipitate is repeptized in part by
adding more hydrogen sulfide even if the sulfide is allowed to age for
several months. Moreover, the black residue obtained by taking the
sol to dryness in a vacuum desiccator is partly repeptized by suspend-
ing in water and treating with ammonia and hydrogen sulfide.
The sol is quite stable in the absence of air. Boiling to remove
hydrogen sulfide causes precipitation, but the excess gas may be washed
out with hydrogen or the excess ammonia removed in a vacuum desic-
cator without precipitation. The particles are negatively charged, but
alkali salts exert no precipitating action even in high concentration.
Salts with multivalent cations cannot be used since they react either
with the ammonia or the hydrogen sulfide.
To prevent the formation of the nickel sulfide sol in qualitative
analysis, Pan and Tang 23 recommend neutralizing the test solution
with nitric acid and making just alkaline with ammonia before con-
ducting in the hydrogen sulfide.
Stable brown sols of nickel sulfide are not obtained in the presence
of an excess of ammonium polysulfide ; instead, a black precipitate is
formed which cannot be washed without decomposition. 24 Analysis of
an unwashed sample indicates a composition between NH 4 NiS 4 and
NH 4 NiS 5 ; but the constitution is indefinite. It may be a sulfo salt
" Kolloid-Z., 88, 155 (1926).
** Nanking J., 4,33 (1934).
"Ephraim: Ber, 66B, 1885 (1923).
274 COLLOIDAL SULFIDES OF MANGANESE, NICKEL, COBALT, ETC.
like the copper compound NH 4 CuS 4 , 25 the ordinary nickel sulfide with
adsorbed polysulfide, or a higher sulfide of nickel.
Villiers 28 developed a test for nickel which depends upon the
formation of the brown sol. The nickel salt solution is treated with
tartaric acid and a slight excess of sodium hydroxide. On passing in
hydrogen sulfide, a part of the nickel sulfide is peptized, giving the
characteristic dark brown color. Cobalt ions are precipitated as cobalt
sulfide by this treatment except when a trace is present and this sulfide
is likewise peptized, giving a brown sol. 2T Tower showed that tartrate
is essential for the stability of nickel sulfide sols prepared in alkali
tartrate solution, complete precipitation resulting on removal of the
organic ion by dialysis. A sol formed in the presence of glycerin 28 is
likewise thrown down by dialyzing out the organic stabilizing agent.
COLLOIDAL COBALT SULFIDE
The gel and sol of cobalt sulfide are obtained by the same pro-
cedures used for nickel sulfide. As already noted, the tendency of the
gel to be peptized by ammonium sulfide is slight, but a dilute sol re-
sults by passing hydrogen sulfide into a cobalt solution containing an
alkali tartrate.
From colorimetric analysis of nickel and cobalt (M) sulfide sols
using a selenium cell, Mickwitz 29 concludes that the sols formed
from dilute solutions with sodium sulfide have the composition
M (OH) (SH) ; with excess hydrogen sulfide, the solid phase in the co-
balt sulfide sol is believed to be Co(SH) 2 . 19 The evidence for these
conclusions is altogether inadequate. It is probable that, if the sols were
examined directly by x-ray diffraction methods as described by Weiser
and Milligan, 80 they would give the pattern of MnS and CoS, respec-
tively.
The quantitative precipitation of cobalt as sulfide is accomplished
at a />H of 3.9 maintained by an ammonium acete-acetic acid buffer. 81
Riegel 82 describes the phenomena connected with the precipitation of
cobalt sulfide bands in silica gel.
'Gluud: Ber, 66B, 952 (1922); Gluud and Miihlendyck: 66B, 899 (1923).
2Compt. rend., 119, 1263 (1894) ; 120, 46 (1895).
27 Tower: J. Am. Chem. Soc., 22, 501 (1900); Tower and Cooke: J. Phys
Chem., 26, 728 (1922); cf. Dumanskii and Buntin: J. Russ. Phys.-Chem. Soc.,
81, 279 (1929).
"Muller and Artman: Oesterr. Chem.-Ztg., 7, 149 (1904).
Z. anorg. Chem., 196, 113 (1931).
*>J. Phys. Chem., 40, 1095 (1936).
"Haring and Leatherman: J. Am. Chem. Soc., 62, 5135 (1930).
" J. Phys. Chem., 86, 1674 (1931).
COLLOIDAL IRON SULFIDES 275
COLLOIDAL IRON SULFIDES
The black precipitate of ferrous sulfide thrown down from a cold
ferrous salt solution with ammonium sulfide is quite finely divided and
may be peptized in part by thorough washing with cold water. The
sol is rather instable, as is the one formed by passing hydrogen sulfide
into a very dilute solution of a ferrous salt. 83 The stability is very
much increased if the reaction is carried out in the presence of
glycerin, 28 gelatin, or sugar. Sabbatani S4 investigated the pharmacologi-
cal action of such sols when administered to dogs and rabbits. Since
the particles were protected by an adsorbed film, the preparations were
inactive so long as they remained colloidal.
Ferric sulfide is apparently formed 85 both by the action of am-
monium sulfide on a ferric salt and of ammonium polysulfide on a
ferrous salt. The precipitate behaves toward ammoniacal zinc solution
as if it were Fe 2 S 3 J but toward mercuric chloride it reacts as follows :
FeS FeS 2 + 2HgQ 2 - 2HgS + S + 2FeQ 2 . It is suggested that
ferric sulfide may exist either as Fe 2 S 3 or as FeS FeS 2 ; this is termed
valence isomerism. With freshly precipitated hydrous ferric oxide and
ammonium sulfide, Fe 2 S 3 results chiefly ; whereas, with ferric tartrate
and ammonium sulfide, a double salt, Fe 2 S 3 (NH 4 ) 2 S, is said to
form. 86 The accuracy of these statements should be checked by x-ray
analysis of the products.
A sol of ferric sulfide is said to form by passing hydrogen sulfide
into a dilute solution of hydrous ferric oxide in tartaric acid. On
adding to a sodium sulfide solution either ferric chloride or ferrous
sulfate followed by oxidation with sodium or hydrogen peroxide a
dark green sol is obtained which gives a dark green gel on coagulating
with sodium chloride. 37 Although the ratio of iron to sulfur in the gel
may be represented by Fe 2 S 3 , it is believed that the sol consists of
dispersed thioferrites. The color of the sulfur waters in Yellowstone
National Park is attributed to the dispersed iron compound, whatever
it may be.
ss Winssinger: Bull. soc. chim., (2) 49, 452 (1888).
"Atti accad Lincei, (S) 32 II, 326, 473 (1923); 88 I, 8; II, 228 (1924).
s'Feigl: Z. anal. Chem., 72, 32 (1927).
36 Feigl and Backer: Z. anal. Chem., 74, 393 (1928); cf. f however, Pearson
and Robinson: J. Chem. Soc., 814 (1928).
"Casares: Anales soc. espan. fis. quim., 81, 638 (1933).
276 COLLOIDAL SULFIDES OF MANGANESE, NICKEL, COBALT, ETC.
COLLOIDAL INDIUM TRISULFIDE
Yellow indium trisulfide is precipitated by conducting hydrogen sul-
fide into a solution of an indium salt which is neither too acid nor too
concentrated. 38 If indium hydroxide is suspended in water through
which hydrogen sulfide is passed, a golden-yellow negative sol of the
sulfide is produced. 30 This sol may be boiled to remove excess hydro-
gen sulfide without flocculation, but it is quite instable in the presence
of electrolytes, especially those with strongly adsorbed cations.
COLLOIDAL THALLOUS SULFIDE
Thallous sulfide, T1 2 S, is precipitated in a finely divided hydrous
condition when hydrogen sulfide or ammonium sulfide is added to an
alkaline or acetic acid solution of a thallous salt. Incomplete precipi-
tation of the hydrous sulfide results when hyposulfite is added to a
thallous salt solution; formed in this way the mass is reddish brown
at first, then violet, and finally black. 40 If the cold solution from which
the salt separates contains a trace of free sulfuric acid, it comes down
as microscopic tctrahedra. 41 The sol of thallous sulfide is prepared by
passing hydrogen sulfide into a very dilute solution of a thallous salt
and dialyzing the product 33
Thallic sulfide, T1 2 S 3 , is not precipitated by the action of hydrogen
sulfide on an aqueous solution of a thalhc salt, for, if formed at all,
it is reduced instantly to thallous sulfide and free sulfur. 42
COLLOIDAL SULFIDES OF GERMANIUM
Germanous sulfide, GeS, is precipitated as a hydrous mass when
hydrogen sulfide is passed into a solution of germanous chloride. 43 If
the precipitation is carried out in the hot, the sulfide is dark red in color
and granular; if precipitated in the cold, it is light yellow and gelatin-
ous. The granular product is definitely crystalline, but the gelatinous
product is said to be amorphous. 44 It is probable, however, that x-ray
aswinkler: J prakt Chem, 94, 1 (1865); 102, 273 (1867); cf Thiel and
Luckmann Z. anorg Chem, 172, 353 (1928).
seLmder and Picton- J Chem Soc, 61, 134 (1892).
*Brunck: Ann, 336, 281 (1904).
"Hebberling: Ann., 134, 11 (1865).
"Strecker- Ann., 136, 207 (1865).
"Winkler. J. prakt. Chem., (2) 34, 177 (1886); 36, 177 (1887).
** Dennis and Hulse- J. Am. Chem Soc., 62, 3553 (1930); cf. t also, Dennis
and Joseph: J. Phys. Chem, 31, 1716 (1927); Pugh: J. Chem. Soc., 2369
(1930).
COLLOIDAL MOLYBDENUM SULFIDES 277
analysis would show the gel particles to have an orthorhombic crystal-
line structure 46 identical with the darker granular preparation. If the
crystalline monosulfide is dissolved in potassium hydroxide and this
solution neutralized with hydrochloric acid, the monosulfide conies
down in a finely divided state 4e which hydrolyzes readily in water or
moist air. The freshly formed precipitate of germanous sulfide is
readily peptized by washing, giving an orange-red to brown sol which
is quite stable if kept out of contact with air. The sol is negatively
charged and is readily flocculated by acids and by salts with multivalent
cations.
A study of the absorption band spectrum of germanous sulfide dis-
closed the germanium isotopes of atomic weight 76, 74, 72, and 70. 4r
Germanic sulfide, GeS 2 , is thrown down by hydrogen sulfide from
strongly acid solutions of germanium dioxide. An aqueous solution
of the dioxide gives no precipitate with hydrogen sulfide, and the pre-
cipitation is incomplete from a slightly acidified solution ; Johnson and
Dennis 48 found the precipitation to be very nearly complete in 24-48
hours. The sulfide comes down at first in a very finely divided form
giving the suspension a milky, almost colloidal, appearance; but, on
standing, the white precipitate agglomerates to a flocculent form that
is peptized by thorough washing to give an opalescent negative sol. The
salt is hydrolyzed to some extent even in the cold ; hence, when the sol
is flocculated by heavy-metal salts, the coagulum has the color of the
heavy-metal sulfides. In boiling water, complete hydrolysis takes place,
giving a solution of GeO 2 . Advantage may be taken of this property
in the analytical determination of germanium. 49
COLLOIDAL MOLYBDENUM SULFIDES
The trisulfide of molybdenum, MoS 3 , is precipitated when hydro-
gen sulfide is passed into a concentrated solution of a molybdate fol-
lowed by the addition of hydrochloric acid. It is also formed by boil-
ing the molybdate of an alkali metal for a short time with ammonium
sulfide and then precipitating with dilute mineral acid. Winssinger 83
prepared the sol by adding to a dilute solution of potassium sulfo-
6Zachariasen: Phys. Rev., 40, 917 (1932).
* Johnson and Whcatley: Z. anorg. Chem., 216, 273 (1934).
* 7 Shapiro, Gibbs, and Laubengayer. Phys. Rev., 40, 354 (1932).
J. Am. Chem. Soc., 47, 790 (1925) ; Dennis and Papish: 43, 2131 (1921) ;
cf. Johnson and Wheatley: Z. anorg. Chem., 216, 273 (1934) ; Winkler: J. prakt
Chem., (2) 84, 228 (1886).
"Muller and Eisner: Ind. Eng. Chem., Anal. Ed., 4, 134 (1932).
278 COLLOIDAL SULFIDES OF MANGANESE, NICKEL, COBALT, ETC.
molybdate slightly more than enough acid to liberate the sulfide and
dialyzing. The transparent brown sol was quite stable; but if the
preparation was not purified by dialysis it coagulated in a few hours.
Hydrous molybdenum pentasulfide is precipitated as a brownish-
black mass by reducing with zinc an ammonium molybdate solution
containing 20% H 2 SO 4 and passing in hydrogen sulfide. 50 No sol of
this sulfide has been described.
Svedberg 81 prepared a sol of molybdenum disulfide in isobutyl alco-
hol by passing an oscillating discharge between electrodes of molyb-
denite ; in a similar way, von Hahn 82 prepared a blue hydrosol. A sol
was also formed by cathodic disintegration of the molybdenum min-
eral. 83 '
COLLOIDAL TUNGSTEN TRISULFIDE
The trisulfide of tungsten, WS 3 , is precipitated by adding acid
to a solution of tungsten trioxide in ammonium sulfide or by saturat-
ing an aqueous solution of a tungstate with hydrogen sulfide and acid-
ifying. The precipitate is much more finely divided than the corre-
sponding molybdenum salt and runs through the filter during washing.
If the washed sulfide is suspended in boiling water, it is largely pep-
tized, forming a brown sol. Winssinger prepared the sol by adding
slightly more acid to a sodium thio tungstate solution than is necessary
to free the sulfide. The color of the sol changes from a bright orange-
red to a dark brown without agglomerating.
COLLOIDAL SELENIUM SULFIDE
Several substances have been described as compounds of selenium
and sulfur, but they have all proved to be mixtures of the two ele-
ments, 84 which are known to form several series of mixed crystals. 85
It is possible, however, that the sol formed by conducting hydrogen
sulfide into an aqueous solution of selenium dioxide 88 may contain
60 Kruss: Ann., 225, 1 (1884); Mawrow and Nikolow: Z. anorg. Chem,
96, 188 (1916).
8i "Die Methoden zur Herstellung kolloider Losungen," 2nd ed, 490 (1909).
82 Kolloid-Z. (Zsigmondy Festschrift), 36, 277 (1925)
63 Muller and Lucas: Z. Elektrochem., 11, 521; Le Blanc: 813 (1905).
* Divers and Schimidzu: Chem. News, 51, 199 (1885).
B Ringer: Z. anorg Chem., 82, 183 (1902); Retgers: Z. physik. Chem., 12,
583 (1893).
86 Gutbier: Z. anorg. Chem., 32, 292 (1902); Gutbier and Lohmann: 42,
325 (1904); 43, 384 (1905).
COLLOIDAL TELLURIUM SULFIDES 279
some selenium sulfide. In any event, the preparation is always referred
to as colloidal selenium sulfide. The sol possesses a bright yellow color
in reflected light and red in transmitted light. It is very stable even
in the presence of electrolytes, but by adding hydrochloric acid and
boiling, or by exposure to light, a plastic red gel is precipitated. 57 Since
selenium sol is red, it is possible that the red color of the coagulum is
due to colloidal selenium dispersed in plastic sulfur. It should be
possible to settle this point by x-ray analysis or by an ultramicroscopic
examination of the gel.
COLLOIDAL TELLURIUM SULFIDES
Both tellurium disulfide, TeS 2 , and tellurium trisulfide, TeS 3 , are
said to exist in the sol state when prepared and kept at low tempera-
tures. 88 The disulfide sol is obtained by passing hydrogen sulfide into
a cold solution of a tetravalent tellurium salt. During the gradual
formation, the sol changes in color from yellow, orange, and reddish
brown to black with a bluish opalescence. If the reagents are pure the
sol is quite stable ; it is precipitated by freezing as reddish-brown floes
which are repeptized after thawing.
The sol of so-called tellurium trisulfide is formed by passing hydro-
gen sulfide into a cold dilute solution of telluric acid. The dialyzed
preparation appears perfectly clear in transmitted light with a light blue
to gray-violet color; but in reflected light, it is cloudy and gray in color.
Like the disulfide sol, it is quite stable and possesses similar properties.
The similarity in behavior of the two sols would seem to indicate
that both are primarily mixtures of colloidal tellurium and colloidal
sulfur.
"Von Hahn: Kolloid-Z., 27, 172 (1920).
"Gutbier: Z. anorg. Chem., 32, 292 (1902).
CHAPTER XV
LITHOPONE AND OTHER SULFIDE PIGMENTS
GENERAL CHARACTERISTICS OF PIGMENTS
Ordinary paints consist of three essential parts: the pigment, the
medium or vehicle, and the drier or siccative. Pigments are finely
divided insoluble powders ; when mixed with suitable media they form
paints. For oil paints, the media are usually vegetable drying oils, of
which linseed oil is by far the most important and most commonly used.
For water-color paints, the vehicle consists of such substances as
honey, glycerin, dextrin, and aqueous sols of the gums. The driers
are oxygen carriers which catalyze the oxidation of the oil ; among the
most important driers are suitable compounds of metals such as lead,
manganese, and cobalt which have more than one valence. Certain
pigments, notably litharge and white lead, possess the necessary sicca-
tive properties and a special drier is not required. In the pyroxylin
lacquers, which dry by evaporation of a volatile solvent, the nitrocellu-
lose is dispersed in such media as amyl acetate, n-butyl alcohol, and
glycol esters.
An ideal pigment should be quite stable chemically and should not
react with any material with which it is likely to come in contact, either
in the vehicle or on the surface which it covers. Actually, few pig-
ments fulfill these ideal requirements, for few substances are capable
of standing the action of light and the atmosphere, pure and impure,
for indefinite periods of time. Although it is obvious that no pigment
is absolutely proof against destruction, it is equally true that all pig-
ments are quite durable under certain conditions. The chemical char-
acteristics which a pigment must possess are determined, therefore,
by the conditions to which it will be subjected. In general, the most
valuable pigments are to be found among the metallic oxides and the
metallic salts of fairly strong acids.
In addition to the chemical requirements, a pigment must possess
the desired color, covering power, hiding power, and oil-adsorption
value. 1 Since the physical state of the pigment, including the particle
1C?. Bartell and Hershberger: Ind. Eng. Chem, 22, 1304 (1930).
280
PREPARATION AND GENERAL PROPERTIES 281
size, has such a profound effect on the above characteristics, one must
approach this subject from the standpoint of colloid chemistry. Many
of the general principles involved are well illustrated by the colloidal
sulndes.
ZINC SULFIDE AND LITHOPONE
Preparation and General Properties
Lithopone. Lithopone is a white pigment consisting of an intimate
mixture of barium sulfate and zinc sulfide prepared in a special way.
X-ray analysis of the material showed it to be a non-homogeneous mix-
ture and not a solid solution or a compound. 2 The first patent was
taken out for its preparation by De Douhit in France in 1850, 8 and it
was given the name lithopone by Boulez in 1877. The process pat-
ented by Orr in 1874 consists in the double decomposition of barium
sulfide and zinc sulfate in solution, followed by igniting the sulfide-
sulfate precipitate and quenching in water. Before ignition, the pre-
cipitate is useless as a pigment, having very little covering power or
body. Heating changes its physical character: first, by dehydrating
the zinc sulfide ; second, by rendering it brittle so that fine grinding is
possible; and third, by increasing the density and thereby increasing
the body of the pigment.
Since a suitable pigment cannot be prepared by grinding barium
sulfate and zinc sulfide together, Mann 4 concludes that lithopone is not
a mere mixture. A preparation having the properties of lithopone is
formed by mixing a positively charged sol of barium sulfate with a
negatively charged sol of zinc sulfide. Mann regards the mechanism of
lithopone formation to be a mutual precipitation of oppositely charged
colloids. Although mutual precipitation of colloidal barium sulfate
and colloidal zinc sulfide gives a mixture with the desired physical char-
acter, it seems unnecessary to postulate the initial formation of op-
positely charged colloidal particles to account for the nature of the
product resulting from direct metathesis. The simultaneous condensa-
tion of barium sulfate and zinc sulfide molecules from solution will of
itself give a more intimate mixture than could result from mutual
precipitation of the sols, and it is probable that the former process
gives the better product. One can obtain a more intimate mixture of
barium and strontium carbonates by simultaneous precipitation than
could result from grinding the two powders together. Like lithopone,
a Wood: J. Soc. Chem. Ind., 49, 300r (1930).
8 Cooper: Chemistry & Industry, 46, 552 (1927).
* Colloid Symposium Monograph, 3, 247 (1925).
282 LITHOPONE AND OTHER SULFIDE PIGMENTS
the product resulting from simultaneous precipitation will differ in
physical character from the mechanically mixed product ; but the quali-
ties of the simultaneously precipitated carbonates would not be at-
tributed to mutual precipitation of oppositely charged particles. More-
over, the difficulty of preparing a positively charged hydrosol of barium
sulfate precludes the possibility of its formation in the presence of an
excess of either sulfate or sulfide ion. 5
The ignition of lithopone in the air results in partial oxidation of
the zinc sulfide to zinc oxide. Since the presence of much zinc oxide
decreases the covering power of the pigment, the conditions of ignition
are adjusted so as to prevent undue oxidation and the product is
quenched in water. 8 Heating in the absence of oxygen has been sug-
gested; 7 but, as will be pointed out later, the formation of a small
amount of oxide on the surface of the zinc sulfide particles is an ad-
vantage in increasing the stability of the pigment to light. If the pig-
ment contains too much oxide and is heated too high, it goes off-color,
becoming yellowish instead of the desired pure white. 8 A "voluminous"
lithopone has been prepared having an apparent volume 20% greater
than ordinary types ; it has higher oil-adsorption power, and paints con-
taining it carry less pigment and have a lower hiding power. 9 Its lack
of normal properties is the result chiefly of insufficient calcination. 10
Theoretically, equivalent solutions of barium sulfide and zinc sul-
fate produce 29.S<fo ZnS and 70.5% BaSO 4 . The composition of the
commercial products is sometimes modified either by the addition of
barytes or of zinc sulfide prior to the ignition. The tinctorial strength
and hiding power of the pigment are reduced by adding barytes and
are increased by adding zinc sulfide.
When properly prepared, 11 lithopone is one of the most important
white pigments for paints and enamels and for compounding with rub-
ber and linoleum. It is pure white, very fine in texture, and has the
same tinctorial strength as zinc oxide and greater hiding power. 12
B Weiser: J. Phys. Chem., 21, 314 (1917).
O'Brien: J. Phys. Chem., 19, 113 (1915).
T Griffith: Brit. Pat. 3,864 (1877); Ostwald and Brauer: Ger. Pat. 202,709
(1908).
*Cf. Meyer: Ger. Pat. 246,021 (1908); Erase: 254,291 (1909); Englemann:
264,904 (1912); Eibner: 324,646 (1918); Steinau: Chem.-Ztg., 44, 974 (1920);
46, 741, 1238 (1921).
Hebberling: Farben-Ztg., 88, 619 (1933) ; Lins: Farbe u. Lack, 101 (1933).
"Seidel: Farbe u. Lack, 127 (1933).
11 Cf. t for example, Flynn, Stutz, and Schertzinger : U. S. Pat. 2,007,527
(1935) ; Flynn: 1,882,072; Hooey: 1,896,312; Stutz and Depew: 1,914,563 (1933).
"Toch: "Chemistry and Technology of Mixed Paints," 26 (1925).
PREPARATION AND GENERAL PROPERTIES 283
Unlike white lead, it is non-poisonous, is unaffected by sulfurous gases,
and is stable in every medium for paints except those of high acidity. 18
It mixes easily with oils and other colors. It is insoluble in water,
ammonia, and alcohol, and is practically fireproof. 14
Zinc Sulfide. Zinc sulfide alone is sometimes employed as a pig-
ment under the name metal white or zinc white, although the latter
term is more often applied to the oxide of zinc. When properly pre-
pared, the sulfide possesses good hiding power and higher durability
than lithopone 15 and, like zinc oxide, is not darkened by the action of
sulfur or sulfur vapors. A large number of patents have been taken
out for the technical production of the pigment; thus, a fine white
product is obtained by the action of sodium sulfide on an alkaline solu-
tion of sodium zincate. 16 A powder which does not mat together on
drying results if the precipitation is carried out at 195 under pres-
sure. 17 Good products are reported from the action of sulfur 18 or
of carbon disulfide 1D on an alkaline solution of zinc oxide. It is also
obtained from zinc blende or other zinc-bearing ores ; 20 indeed, zinc
blende has been finely pulverized and used directly without further
treatment. 21 It is also formed in the dry way by bringing together the
vapors of the two elements and by calcining zinc sulfate under suitable
conditions. 22 Work and Odell M showed that the development of pig-
ment value measured by covering power and tinting strength is asso-
ciated with growth of particles to visible range of sizes by suitable
ignition ; the maximum pigment value results by heating at 600.
A satisfactory pigment must be stable in the light and must possess
a pure white color. When formed by precipitation methods, the ad-
sorbed water can be removed only at relatively high temperatures. On
ignition in the air, the pigment assumes a yellow color which has been
attributed to the formation of an oxysulfide 24 but which is probably
"Morrell and Waelc: "Rubber, Resins, Paints, and Varnishes," 118
14 Scott: "White Paints and Painting Materials," 237.
Nelson: Am. Paint J., IB, No. 28, 20/ (1931).
^De Stuckle: Ger. Pat. 171,872 (1906).
"Goldschmidt and Sohn: Ger. Pat 262,701 (1912).
^Pipereaut and Vila: Ger. Pat. 223,837 (1907).
"Desachy: Brit. Pat. 126,627 (1919).
aoThwaites: Ger. Pat. 222,291 (1908) ; Brit. Pat. 9,175 (1912) ; J. Soc. Chem.
Ind, 31, 431 (1912); Clerc and Nihoul: Ger. Pat. 381,423 (1920).
'iRichter- Farben-Ztg., 29, 728 (1924).
"Pipereaut and Helbronner: U. S. Pat. 1,443,077 (1922); Helbronner: Brit
Pat. 148,351 (1920); Comment: U. S. Pat. 1,374,435 (1921).
2 3 Ind, Eng. Chem., 25, 411, 543 (1933).
2 *De Stuckle: J. Soc. Chem. Ind., 80, 96 (1911).
284 LITHOPONE AND OTHER SULFIDE PIGMENTS
zinc oxide, since the latter is known to be yellow when sintered. 85 The
color may be removed 24 by heating with NH 4 HS at 155, by treating
with hydrogen sulfide in the presence of hydrofluoric acid, or by heat-
ing with 2% H 2 SO 4 . 28
DARKENING OF ZINC SULFIDE PIGMENTS IN LIGHT
Unless special precautions are taken in the manufacture of zinc
sulfide pigments, they blacken in the sunlight and become white again
in the dark. Attention was first called lo this phenomenon by Phip-
son, 27 who observed that a gatepost painted white with lithopone
turned dark during the day and became white again at night. The
barium sulfate in lithopone is without influence or plays but a minor
role in the process since zinc sulfide alone will exhibit the same phe-
nomenon. Because of the importance of this behavior from both a
theoretical and technical standpoint, the cause of the blackening and
the mechanism of the process will be considered in some detail.
Cause of the Darkening
Phipson, who first investigated the darkening of lithopone, claimed
that it was not caused by the presence of impurities which form black
sulfides, and so was led to attribute the discoloration to the presence
of a new element, similar to lanthanum, which he named actinium.
This hypothesis was disproved by Cawley, 28 who concluded, by a
process of elimination, that the black color resulted from a small
amount of finely divided zinc. The first experimental evidence to sup-
port this assumption was obtained in Bancroft's laboratory by
O'Brien, 6 who brought some of the blackened lithopone in contact with
a ferric alum-potassium ferricyanide solution and obtained a blue col-
oration such as would be expected in the presence of metallic zinc.
Later, Durst 29 showed that a blackened lithopone became permanently
black if brought in contact with a solution of nobler metals such as
copper and lead, probably owing to the displacement of the heavy
metal from solution by metallic zinc. It should be pointed out, how-
ever, that this evidence is not altogether conclusive since salts of noble
metals may react with zinc sulfide directly. Thus, silver chloride re-
acts quantitatively with precipitated zinc sulfide in accordance with the
2Farnau: J. Phys. Chem., 17, 639 (1913).
*Koetschet and Meyer: U. S. Pat. 1,001,415 (1911).
2r Chem. News, 48, 283; 44, 73 (1881).
as Chem. News, 44, 51, 167; cf. Orr: 12 (1881).
2 Z. angew. Chem., 86, 709 (1922).
CONDITIONS FOR PHOTOCHEMICAL DECOMPOSITION 285
reaction: ZnS + 2AgCl -ZnCl 2 + Ag 2 S (black). 80 All doubt as to
the cause of blackening was removed by Job and Emschwiller, 81 who
obtained several centigrams of zinc by the action of the light from a
quartz mercury lamp on a light-sensitive zinc sulfide suspended in water
in a quartz vessel. The metal evolved hydrogen from acids and dis-
placed copper from copper sulfate. Simultaneously with the libera-
tion of zinc, sulfur was formed which was extracted with carbon di-
sulfide and subsequently crystallized from the solution. Besides the
primary products, zinc and sulfur, a small amount of zinc thionate and
hydrogen were formed, the latter probably resulting from the slight
decomposition of water by the colloidal zinc.
In opposition to the view that the darkening is due to colloidal zinc,
Eibner 32 claims that the discoloration is caused by the presence of
metals which form black sulfides. This view is altogether untenable :
first, because chemically pure zinc sulfide is blackened by light ; 3a and
second, because the addition to lithopone of a metallic salt such as
lead acetate has no appreciable influence on the tendency to darken in
light. 3 *
Conditions for Photochemical Decomposition
Crystal Structure. Zinc sulfide precipitated from ammoniacal solu-
tion with hydrogen sulfide or ammonium sulfide consists of minute
cubic crystals corresponding to zinc blende or sphalerite. Under or-
dinary conditions the precipitated sulfide is stable but is rendered light-
sensitive by ignition under such conditions that wurtzite is formed. A
second requirement for light sensitivity is the presence of an excess
of water. Specially prepared, chemically pure zinc sulfide 3B will
darken, so that impurities are not essential to the process, although cer-
tain salts, especially soluble zinc salts, increase the light sensitivity.
Indeed, an unignited zinc blende formed slowly from slightly acid solu-
tion will darken on exposure to light in contact with a zinc chloride
solution. 36
The much greater light sensitivity of wurtzite was recognized al-
most 50 years ago by Cawley, 87 who pointed out that zinc blende will
30 Jander and Stuhlmann: Z. anal. Chem., 60, 308 (1921).
s i Compt rend, 177, 313 (1923).
32 Farben-Ztg., 27, 3378 (1922); Chem.-Ztg, 47, 13 (1923)
saLenard: Ann. Physik, (4) 68, 553 (1922).
3*Maass and Kempf : Z. angew. Chem, 86, 294 (1923).
35 Tomaschek: Ann Physik, (4) 65, 189 (1921).
*Weiser and Garrison: J. Phys. Chem., 31, 1237 (1927).
"Chem. News, 68.88 (1891).
286 LITHOPONE AND OTHER SULFIDE PIGMENTS
not darken in ultraviolet light. This conclusion was confirmed by
Schleede 88 from observations with pure precipitated zinc sulfide
thrown down from alkaline solution. When ignited below 850 the
sulfide was not darkened by long exposure to quartz ultraviolet light,
and an x-radiogram showed it to consist of the cubic crystals of sphal-
erite. Ignition at 1150 (35 below the melting point of wurtzite)
gave a product with the maximum light sensitivity, and an x-radiogram
showed it to be hexagonal wurtzite. Ignition at 1000 gave a mixture
of both blende and wurtzite which darkened less readily than pure
wurtzite (see second paragraph below). The presence of copper, man-
ganese, or cadmium in amounts necessary to cause phosphorescence had
no effect on the light sensitivity.
By carrying out the ignition in the presence of a flux, Schleede
found the ignition temperature to be of secondary importance. Thus,
when the sulfide was ignited at as low a temperature as 750 in the
presence of potassium chloride, an x-radiogram showed the formation
of some wurtzite and even glass ultraviolet light caused darkening.
Since Schleede did not know the mechanism of the darkening process,
he attributed the light sensitivity in the presence of chloride to the
formation of mixed crystals of wurtzite and the halogen. Washing out
the chloride destroyed the sensitivity to glass ultraviolet light but did
not affect the action toward quartz ultraviolet light. Ignition with both
chlorides and bromides gave light-sensitive products, but ignition with
fluorides, phosphates, and borates gave light-stable preparations. This
is in line with O'Brien's 6 findings, that the addition of phosphates,
ferrocyanides, borates, cyanides, or bicarbonates to lithopone prevented
the darkening or decreased it to an appreciable extent.
Although wurtzite content and light sensitivity of zinc sulfide fre-
quently go hand in hand, it was found at the New Jersey Zinc Com-
pany 89 that certain zinc sulfides with very high wurtzite content are
very light-resistant. It was suggested, therefore, that it was not the
wurtzite form or the sphalerite form which was light-sensitive but
was the unstable zones at the boundaries between two dissimilar
crystals.
Moisture. Cawley, 87 * 40 who suggested that the darkening of zinc
sulfide was caused by metallic zinc and who first pointed out that or-
dinary zinc blende is non-sensitive to light, likewise was the first to
recognize the importance of the presence of moisture for the black-
88 Z. physik. Chem., 106, 391 (1923).
89 Private communication from Dr. Howard M. Cyr.
"C/. O'Brien: J. Phys. Chem., 19, 126 (1915).
MECHANISM OF THE DARKENING PROCESS 287
ening. Lenard 41 and Schleede 42 called attention to the fact that more
than a trace of moisture is necessary. Indeed the blackening is more
marked when the surrounding air is supersaturated with moisture than
when it is saturated, and the effect is still more pronounced when the
sulfide is covered with water. The reason for this will be discussed
in a later section.
Relation to Phosphorescence. Since the action of light on phos-
phorescent zinc sulfide is usually accompanied by blackening 48 which
disappears in the dark, some investigators conclude that the phenomena
of luminescence and darkening are intimately related. Thus Job and
Emschwiller 81 ' 44 give phosphorescence as one of the requirements for
a light-sensitive sulfide. Lenard showed, however, that a chemically
pure, non-phosphorescent sulfide will blacken in the light and that a
fairly dry phosphor will glow without darkening, the latter pheno-
menon manifesting itself only in the presence of an excess of water
vapor. Moreover, the blackening of a number of zinc phosphors was
found to require a shorter wave length of light than was needed to
excite phosphorescence. Thus the darkening appeared suddenly at a
wave length of 334 m/* while intense phosphorescence maxima were
observed by radiations of 430 and 360 m/*, which caused no darkening.
It appears, therefore, that the two phenomena are not necessarily re-
lated although they are produced simultaneously by proper excitation.
As is well known, the darkening of the silver halides by light is not
accompanied by phosphorescence.
Mechanism of the Darkening Process
Since zinc sulfide which has not been ignited will not blacken
ordinarily, Cawley suggested that the ignition results in the formation
of some zinc oxide with which the remaining zinc sulfide reacts in the
light, as follows : ZnS + 2ZnO - SO 2 + 3Zn. This view is untenable,
since ignition of lithopone under conditions favorable for forming a
film of zinc oxide over the sulfide particles gives a light-stable product,
whereas removal of the zinc oxide film from such a preparation by
heating with an acid restores the light sensitivity. 6 - 84 Furthermore,
chemically pure zinc sulfide is darkened by light. Maass and Kempf 84
believed that the darkening is occasioned by the following reaction:
"Ann. Physik, (4) 68, 572 (1922).
Z. physik. Chem., 108, 390 (1923).
"Lenard: Ann. Physik, (4) 31, 652 (1910); Baerwald: 89, 849 (1912);
Tomaschek: 66, 195 (1921).
"Loeb and Schmiedeskamp : Proc. Natl. Acad. Sci. U. S., 7, 202 (1921).
288 LITHOPONE AND OTHER SULFIDE PIGMENTS
2ZnS - ZnS 2 '+ Zn. This is likewise untenable since sulfur instead of
the hypothetical zinc disulfide is formed in the process.
Lenard explained the necessity for ignition by postulating the for-
mation of polymerized molecules, (ZnS)*, which were assumed to
blacken owing to the "liberation or loosening of zinc atoms from the
molecular union." The subsequent discoloration in the dark was at-
tributed to the recombination of the loosened zinc and sulfur atoms.
This mechanism is not satisfactory since it is based on some assump-
tions of doubtful accuracy and since it does not accord with all the
experimental observations. In the first place, the formation of poly-
merized molecules of zinc sulfide has not been proved. Moreover, it
is questionable whether a "loosened" atom of zinc would cause dark-
ening, and it is known definitely that free atoms of zinc are formed.
Finally, Lenard's assumption that the blackening-discoloration process
is reversible is not in accord with the experimental facts to be recounted
in the next section.
Effect of Soluble Zinc Salts. It has been known for a long time
that a soluble silver salt, such as silver nitrate, increases the light sensi-
tivity of silver bromide (p. 146), and Cawley and O'Brien observed a
very marked increase in sensitivity of lithopone in the presence of
soluble zinc salts.
The sensitizing action of silver nitrate on silver bromide has been
accounted for in part by assuming that the soluble salt acts as a bromine
acceptor reacting with the bromine in accord with the equation:
Br 2 + AgNO 3 + H 2 O - AgBr + BrOH +HNO 3 ; but no such
mechanism can be assumed for the sensitizing action of zinc chloride
on zinc sulfide. In the light of Fajan's observation on the effect of
adsorbed silver ions on the photosensitivity of silver halides (p. 148),
it seems altogether probable that adsorbed zinc ions would increase the
photosensitivity of zinc sulfide by increasing the number of elementary
processes. This means that, if the surface were covered with adsorbed
zinc ions, then an equal amount of zinc sulfide will absorb a larger
portion of the incident light and therefore more zinc sulfide will be
decomposed.
Since a salt always shows a strong tendency to adsorb its own ions,
there is little doubt that zinc ions will be preferentially adsorbed at a
zinc sulfide surface, just as silver ions are adsorbed at a silver halide
surface. From this point of view, any factor which favors the forma-
tion of zinc ions in the immediate region of the surface of zinc sulfide
will tend to increase its sensitivity toward light. As a matter of fact,
soluble zinc salts have a pronounced sensitizing action, whereas in-
MECHANISM OF THE DARKENING PROCESS 289
soluble zinc salts have little effect ; moreover, ignition in the presence
of a small amount of chloride or bromide which forms soluble zinc
salts favors the blackening, whereas ignition with fluorides, phosphates,
or borates which form insoluble salts retards or prevents blackening. 8
The Role of Water. From the above consideration it follows that
the role of water in the photochemical decomposition of zinc sulfide is
merely that of an ionizing solvent for the sulfide and for adsorbed zinc
salts, yielding zinc ions which are adsorbed on the surface of the sul-
fide lattice and sensitize it. As has been pointed out, natural zinc blende
and precipitated blende are ordinarily light-stable whereas wurtzite
is decomposed by ultraviolet light. The difference in behavior is read-
ily understood when one recalls that wurtzite is 4.5 times as soluble as
blende. 46 This means not only that the stability of the wurtzite lattice
is the smaller but also that it yields more readily the zinc ions which
may play such an important part in the darkening process.
Since a solution of a suitable acid or zinc salt increases the sensi-
tivity of wurtzite enormously, it seemed likely that precipitated zinc
sulfide might be made to darken under suitable conditions. This
proved to be true. 46 A 20% solution of recrystallized zinc sulfate was
treated with ammonia short of precipitation, and a stream of specially
purified hydrogen sulfide 47 was passed into the solution very slowly
until precipitation ceased. In this way, fairly large crystals were
formed in the presence of an excess of zinc ion since the precipitation
was incomplete, stopping when the hydrogen ion concentration became
too high. The sulfide, after being freed from excess sulfate, was ex-
posed to quartz ultraviolet light in contact with zinc chloride, and
prompt blackening resulted. A sample of the gelatinous sulfide, pre-
cipitated rapidly with ammonium sulfide, did not blacken in the pres-
ence of zinc chloride. A distinct crystal structure is therefore essential
to light sensitivity.
Since the preferential adsorption of zinc ions sensitizes the sulfide,
it seemed possible that the presence of a salt such as sodium sulfide,
the anion of which is more strongly adsorbed than the cation, would
stabilize the sulfide. Actually, the sensitivity of zinc sulfide to light was
decreased enormously in the presence of sodium sulfide. For example,
a sensitive sulfide covered with water was blackened by a 2-minute
exposure to quartz ultraviolet light whereas the same preparation cov-
ered by a sodium sulfide solution as dilute as 0.02 N showed no signs
Gmelin: "Handbuch anorg. Chem.," 8th ed., 32, 201 (1924).
Weiser and Garrison: J. Phys Chem., 31, 1242 (1927).
"Lenz: Z. anal. Chem., 22, 393 (1883).
290 LITHOPONE AND OTHER SULFIDE PIGMENTS
of blackening after a 30-minute exposure to quartz ultraviolet light of
the same intensity. Sodium sulfate and borax likewise have a stabiliz-
ing influence. In general, any salt with a readily adsorbed anion will
tend to stabilize the sulfide. Nishizawa * 8 found that the sulfide was
stabilized by glycerol, hydroxides, and the salts and esters of tartaric
and polyhydroxystearic acids. The stabilization resulted from the
strongly adsorbed anions of these compounds ; but this was not under-
stood by Nishizawa.
In addition to its action as an ionizing solvent, water may be as-
sumed to have a purely mechanical effect, forming a film around the
liberated zinc, thereby preventing its oxidation by oxygen, ozone, or
the liberated sulfur. This effect must be slight, however, since other
liquids which wet either zinc or sulfur are without influence on the
darkening. Since an excess of water is essential for darkening, it is
improbable that the liquid plays a catalytic role similar to that in the
thermal decomposition of ammonium chloride. 49 Maass and Kempf 8 *
postulated a reducing action of nascent hydrogen formed by photo-
chemical decomposition of the required water. 50 This assumption seems
far fetched since either nascent oxygen 81 or hydrogen peroxide " will
be formed simultaneously and will neutralize any effect of hydrogen.
It was further suggested that formaldehyde, 88 formed by the action
of light on moist carbon dioxide, accelerates the reduction. This view
is likewise untenable since the blackening goes on in the absence of
carbon dioxide." Since Maass and Kempf were the first to suggest
that adsorbed zinc ion might sensitize zinc sulfide in the same way that
adsorbed silver ion sensitizes silver bromide, one is at a loss to know
how they happened to overlook the true role of water in the darkening
process.
The Decolorization Process
The decolorization of blackened zinc sulfide or lithopone takes place
in the dark only in the presence of oxygen or some oxidizing agent
such as chlorine, ozone, or hydrogen peroxide. It is obvious, therefore,
that the process is only partly reversible if at all, the decolorization in
*8 j. Tokoyo Chem. Soc., 41, 1054 (1920) ; Chem. Abstracts, 15, 1407 (1921) ;
Brit Pat. 156,971 (1919).
Baker: J. Chem. Soc., 66, 611 (1894).
"Berthelot and Gaudechon: Compt. rend., 160, 1690; 161, 395 (1910)
"Thiele: Ber., 40, 4914 (1907).
"Kernbaum: Compt. rend., 162, 1668 (1911).
"Berthelot and Gaudechon: Compt. rend., 160, 1169, 1327, 1517, 1690 (1910).
"Weiser and Garrison: J. Phys. Chem., 31, 1239 (1927).
THE DECOLORIZATION PROCESS 291
the air being due to the oxidation of the finely divided metal to white
zinc oxide or basic carbonate. 6 As already mentioned, Lenard's 38 view
is that the photochemical process is reversible. This was based on his
observation that a sulfide thrice darkened and allowed to whiten ap-
peared to be as sensitive as the original preparation. Apparently,
Lenard started out to prove that the reaction is reversible or he would
not have been content with three repetitions. Such a small amount of
decomposition takes place that the darkening and decolorization must
be repeated a number of times before a marked decrease in sensitivity
is noted. Phipson reported that his gatepost, painted with lithopone,
became alternately dark in the daytime and white at night for a long
time, but at last it remained white. The permanent white color on
prolonged aging was probably caused by a protecting film of oxide or
basic carbonate. There is apparently no oxidation of the zinc sulfide
to zinc sulfate. 55
Convinced that the photochemical process is reversible, Lenard as-
sumed that discoloration in the presence of oxygen, chlorine, ozone,
or hydrogen peroxide is caused by the catalytic action of the oxidizing
agent on the recombination of zinc and sulfur. This view is untenable,
since it is generally known that zinc reacts more readily with chlorine
or ozone than with sulfur. Lenard recognized this condition but got
around it by postulating that the zinc atoms which cause the darkening
are merely "loosened," whereas we know definitely that the blacken-
ing is caused by free zinc.
PREVENTION OF DARKENING OF ZINC SULFIDK PICMKNTS
From a technical standpoint, it is of particular importance to
prepare lithopone under such conditions that it is not appreciably dis-
colored by light. From what has been said in the preceding section the
following general rules may be deduced: (1) carry out the necessary
ignition so that light-sensitive wurtzite is not formed; (2) avoid the
presence of soluble zinc salts which increase the light sensitivity; (3)
add a salt with a highly adsorbed anion ; (4) treat th pigment so as to
produce a film of oxide or other light-stable solid around the zinc sul-
fide particles. A more recent development is the use of small amounts
of cobalt salts (p. 293).
Theoretically, the simplest procedure would seem to consist in igni-
tion of the pigment below the temperature where wurtzite is formed.
o Wolff: Z. angcw. Chem., 87, 333 (1924).
292 LITHOPONE AND OTHER SULFIDE PIGMENTS
This is apparently practiced by the New Jersey Zinc Company. 58 It
is claimed that the lithopone should be precipitated in the presence of
chloride in order to get a commercial product with the desired covering
power and oil-adsorbing qualities. Variation in the amount of chloride
has a marked influence on the ignition temperature which must be em-
ployed to get the desired physical character. But, as already shown,
the presence of chloride decreases the stability of the pigment toward
light, 57 probably owing to the formation of some soluble zinc chloride
during ignition. The addition of 0.005% of cobalt salt to the zinc sulfate
solution before treating with barium sulfide is said to prevent the action
of chlorine compounds in causing instability. 58 Extended observations
disclose that the ignition temperature curves of covering power and
oil adsorption on the one hand, and of light stability on the other, cut
each other. Hence, to obtain a lithopone that is at once the most light-
stable and has the highest covering power, the amount of chloride and
the ignition temperature should correspond to the point of intersection
of the curves. Thus, for a lithopone precipitated from a zinc sulfate
solution (density 1.16) containing not more than 2 g of chlorine per
liter, the best ignition temperature is between 700 and 800. Under
these conditions it is probable that the chief product is sphalerite. A
rigid control of the ignition temperature is essential for a uniform
product. 59 It would seem that the addition of a salt with a strongly
adsorbed anion would destroy the light sensitivity resulting from the
formation of any wurtzite.
Attention has been called to the spontaneous formation of a pro-
tecting film around the sensitive zinc oxide particles by repeated dark-
ening and discoloration in air. It is, of course, impractical to form a
protecting film in this way, but most of the earlier methods for obtain-
ing light-proof lithopones involve the addition of an oxidizing agent
or some salt that will yield an insoluble protecting film. 6 ' 37 - 60 More
recent recommendations involve a similar principle. For example,
6Breyer, Croll, and Farber: U. S Pat. 1,411,645 (1922); Brcycr and
Farber- 1,446,637 (1923).
5T Roches: Rev. chim. ind., 31, 109 (1922); Stemau: Chem.-Ztg., 45, 741
(1921).
*C/ Sapgir and Davuidovskaya : J. Chem. Ind. (U.SSR.), No. 3, 44
(1933) ; Chem Abstracts, 27, 5557 (1933).
5 Singmaster, Breyer, and Farber: U. S. Pat. 1,411,646; 1,411,648 (1922)
aoAlberti: Chem. Zentr., II, 651 (1906); Steinau: I, 1593; Ostwald and
Brauer: II, 1707 (1908); Allendorff: I, 116 (1909); Erase. Ger. Pat 254,291
(1909).
PREVENTION OF DARKENING 293
Kuzell 61 sprays lithopone in the air to oxidize the zinc sulfide super-
ficially to zinc oxide. Thus, by igniting lithopone so that little wurtzite
is formed, or by protecting the light-sensitive sulfide with a strongly
adsorbed anion or a non-sensitive film, products are obtained which
appear to meet all the technical requirements.
The use of cobalt salts for increasing the light resistance of zinc
sulfide pigments calls for special consideration. The author is indebted
to Dr. Howard M. Cyr of the New Jersey Zinc Company for the
following information :
Jantsch and Wolski 62 pointed out that the presence of a cobalt
compound in a zinc sulfide pigment has a beneficial effect on light re-
sistance. The quantity required is given as 0.002 to 0.05% figured as
cobalt metal based on the metallic zinc content. In the United States
patent the amounts are erroneously specified as 0.02 to 0.5%. Too
large quantities are undesirable because of the tinting effect of the
cobalt sulfide formed. The cobalt is added preferably as a soluble salt,
such as the sulfate, to the zinc liquor, but it is effective also when
added to the precipitate before heating. In place of cobalt, salts of
nickel, copper, or iron may be used but the quantities of these salts
necessary for improvement in the light resistance give sufficient black
sulfide to produce noticeable darkening.
X-ray diffraction studies at the New Jersey Zinc Company dis-
closed that cobalt sulfide formed a solid solution with zinc sulfide. It
was observed also that the presence of cobalt decreased the concentra-
tion of wurtzite produced under given muffling conditions and, at the
same time, decreased the light sensitivity. This is another illustration
of correlation between wurtzite content and light sensitivity; but as
already pointed out (p. 286), this correlation does not always exist.
By exposing zinc sulfide to light of controlled wave length, it was
found that a certain zone in the ultraviolet was most effective in caus-
ing darkening, whereas other bands of longer wave length prevented
darkening. These two effects usually counterbalanced each other in
normal daylight but on certain hazy days the darkening effect pre-
dominated.
A comparison of cobalt-treated zinc sulfide pigments with the un-
treated pigments disclosed that the rate of darkening of both was the
same. On the other hand, the rate of bleaching was found to be greatly
accelerated by the presence of the cobalt compound. Since cobalt is a
01 U. S. Pat. 1,399,500 (1922); cf. t also, Nishizawa: J. Tokoyo Chcm. Soc.,
41, 1054 (1920); Brit. Pat. 156,971 (1921).
2Ger. Pat 435,840 (1923); U. S. Pat. 1,693,902 (1924).
294 LITHOPONE AND OTHER SULFIDE PIGMENTS
good oxidation catalyst, its acceleration of bleaching is in agreement
with the theory that the bleaching process consists of a reoxidation of
metallic zinc.
The addition of cobalt to lithopone according to Jantsch and Wol-
ski's patent is used by many producers of the product in this country
and in Europe.
CADMIUM SULFIDE PIGMENTS
Cadmium sulfide is used under conditions where the cheaper, bright
yellows are not satisfactory. 68 Its main uses are in the coloring of vul-
canized rubber, as a brilliant artists' color, and for switch and target
enamels. It is unaffected by exposure to light, heat, and city atmos-
pheres and mixes easily with any paint vehicle.
Formation
The yellow sulfide called "cadmium yellow" is obtained by pre-
cipitating cold solutions of low cadmium content with hydrogen sulfide
or by precipitating cadmium solutions with the alkali sulfides. The
deep orange pigment known as "cadmium orange" is precipitated from
hot strongly acid solutions with hydrogen sulfide or by long boiling of
cadmium solutions with sodium thiosulfate or alkali sulfide. Richards
and Roepper fl4 have patented processes for preparing cadmium yellow
by electrolysis of sodium thiosulfate solution with a cadmium anode
and an indifferent cathode, and by alternating-current electrolysis of
thiosulfate solution with cadmium electrodes. The latter process is too
inefficient to be of value, 85 but the former has been improved to the
point where it may find some technical application. Fischer 6e substi-
tuted for the indifferent cathode one consisting of a mixture of equal
parts of sulfur and copper sulfide, which has a relatively low resist-
ance; the current efficiency of Fischer's process is 100%. At a cur-
rent density of 0.05 ampere/cm 2 , both yellow and orange sulfide are
formed ; and at 0.005 ampere/cm 2 , only the yellow is obtained. Fink
and Grosvenor 8T used sodium sulfide as the electrolyte with cadmium
stick anode and iron cathode surrounded by ferrous sulfide as the
source of sulfur. Sacher 68 questions the applicability of the electro-
3Toch: "Chemistry and Technology of Mixed Paints," 73 (1925); cf.
Bugden: Continental Met. & Chem. Eng., 2, 109 (1927).
* Richards: Trans. Am. Electrochem. Soc., 1, 221 (1902).
White: Trans. Am. Electrochem Soc., 9, 305 (1906).
eZ. Elektrochem., 31, 285 (1925) ; Lorenz: Z. anorg. Chem., 12, 442 (1896).
* Trans. Am. Electrochem. Soc., 58, 475 (1930).
esFarbe u. Lack, 246 (1931).
COLOR 295
lytic method as compared with the batch precipitation method for
technical manufacture.
A very satisfactory cadmium lithopone or cadmipone 69 may be
prepared by the interaction of cadmium sulfate and barium sulfide.
Color
The variation in color of the pigment was attributed to adsorbed
impurities by Follenius 70 and to the existence of allotropic modifica-
tions of different density and crystal structure by Buchner 71 and
Klobukow. 78 Microscopic examination of pigments of various colors
led Allen, Crenshaw, and Merwin 73 74 to conclude that there is but one
crystalline form, the color variation being caused by difference in the
size and physical character of the particles. In accord with this view
Muller and Loffler obtained products which gave the same x-radio-
gram but which varied in color from yellow to red by precipitating
cadmium sulfate solutions with hydrogen sulfide in the presence of
different amounts of sulfuric acid. The crystal structure of Muller
and Loffler's colloidal preparations was cubic, whereas that of Allen,
Crenshaw, and Merwin's microscopic preparations, formed by diges-
tion at elevated temperatures, was hexagonal like the mineral green-
ockite.
From the above considerations it would follow that both crystalline
forms may be either yellow or red depending on the particle size. The
question is whether the colloidal, precipitated forms are always cubic
in structure. This was settled by Milligan, 79 who precipitated various
cadmium salts at 30 and 100 with and without the addition of the
corresponding acids, noted the color of the resulting products, and
determined the crystal structure of each by x-ray analysis. The results
which are given in Table XXXV show conclusively that either the cubic
or hexagonal modifications may be yellow or red (or orange), depend-
ing on the conditions of formation and treatment. All samples appear
more or less orange upon grinding the dry aggregates to a powder.
The cubic form is usually obtained from sulfate or nitrate solution,
especially from hot or acid solutions; the hexagonal modification is
"Ward: J. Oil and Colour Chem. Assoc., 10, 4 (1927).
fZ. anal. Chem., 18, 417 (1874).
Chem.-Ztg., 11, 1087, 1107 (1887).
J. prakt. Chem., (2) 89, 414 (1889).
73 Allen, Crenshaw, and Merwin: Am. J. Sci., (4) 84, 341 (1912).
74 C/., however, Egerton and Raleigh: J. Chem. Soc., 128, 3019 (1923).
" Z. angew. Chem., 46, 538 (1933).
J. Phys. Chem., 88, 797 (1934).
296
LITHOPONE AND OTHER SULFIDE PIGMENTS
TABLE XXXV
COLOR AND CRYSTALLINE FORM OF PRECIPITATED CADMIUM SULFIDE
Cadmium
salt
used
0.12V
Without the addition of acid
In the presence of added acid
At 30
At 100
At 30
At 100
Color
Crystalline
form
Color
Crystalline
form
Color
Crystalline
form
Color
Crystalline
form
Sulfate
Nitrate
Chloride
Bromide
Iodide
Yellow
Yellow
Yellow
Yellow
Orange
Hexagonal*
Hexagonal
Hexagonal*
Hexagonal*
Hexagonal
Yellow
Yellow
Yellow
Yellow
Orange
Cubic
Hexagonal*
Hexagonal
Hexagonal
Hexagonal
Yellow
Yellow
Yellow
Orange
Orange
Cubicf
Cubicf
Hexagonal
Hexagonal
Hexagonal
Red
Red
Red
Red
Orange
Cubic
Cubic
Hexagonal
Hexagonal
Hexagonal
* Trace of cubic 0-cadmtum sulfide.
t Trace ot hexagonal a-cadmium sulfide.
usually thrown down from chloride, bromide, or iodide solutions.
Small amounts of the cubic form sometimes occur in essentially hex-
agonal precipitates ; this is probably more or less accidental, depending
upon slight variations in the conditions of precipitation. The variation
in color must be attributed to differences in the physical character of
the precipitate such as particle size, the nature of the surface of the
particles, and the state of aggregation. Difference in particle size alone
will not suffice to explain all the known facts, since large crystals of
greenockite are yellow, although red (or orange) particles are usually
larger and more granular than yellow ones. The view that color varia-
tions result chiefly from differences in physical character of the samples
is in agreement with the observation that the red particles obtained
from hot acid solutions possess less adsorptive capacity than the yellow
particles formed in the cold (p. 257).
In discussing the cause of differences in color, it must be borne in
mind that, in the last analysis, the actual color is determined by the
relative amounts of light transmitted and reflected. 78 For example,
greenockite absorbs all the blue and part of the green of the spectrum
and transmits the remainder. When the grains are in masses having
diameters of 0.2 to 1.0 mm and are bounded by bright faces, a large
amount of blue light is reflected directly by resonance and small
amounts of red, orange, yellow, and green are reflected after passing
through the surface layer of the crystals. The combined effect of all
the reflected light is a lustrous dark yellow to yellowish^ green. Simi-
larly, plane-faced bright crystals having diameters of but 0.01 mm or
less reflect about the same amount of blue by resonance but they re-
COLOR 297
fleet much more of the light which penetrates the surface, the resulting
color being a pure yellow. A mass of crystals of the same size with
dull faces have a light yellow-brown color. The grains of powdered
crystals usually have bright but not plane surfaces and give a brilliant
orange color, for there is less direct reflection and much of the light
finally reflected from the interior has penetrated deeper and thus lost
more green and yellow than a powder having plane-faced fragments.
With the extremely minute submicroscopically crystalline or amor-
phous particles, there is more absorption in the yellow and green and,
by transmitted light, the color appears orange-yellow in films 0.01 mm
thick. A powder with grains 0.0001 to 0.001 mm in diameter is bright
yellow with a tinge of orange, and one with grains 0.004 to 0.007 mm
in diameter, or compact aggregates of smaller granules, is bright
orange.
VERMILION
Vermilion is the synthetic red hexagonal modification of mercuric
sulfide which corresponds to the mineral cinnabar. The native product
does not make a satisfactory pigment, for the impurities dull the color
and it does not possess the desired physical character. The pigment
is synthesized by dry and wet processes. The former, which is used
most frequently, consists essentially in the formation of the black sul-
fide by direct union of mercury and sulfur, and its subsequent con-
version into the red variety by a process of sublimation. Two dry
processes are generally recognized : the Dutch and the Chinese. 77 The
product prepared by the Chinese process is celebrated for its fine color
which inclines to a carmine. At one time it was thought that the Chi-
nese employed a wet method, but this was not the case. 78 The Dutch
and Chinese processes are essentially the same, any differences in phy-
sical character or color resulting from the care exercised in the sub-
limation process.
The wet process depends on the fact that alkali and ammonium
sulfides and polysulfides dissolve the black sulfide which subsequently
precipitates as the less soluble red form. 79 Rise in temperature and
excess sulfur favor the process. 80 The most satisfactory method of
77 Hurst and Heaton : "A Manual of Painters' Colors, Oils, and Varnishes,"
5th ed, 163 (1913).
78 Chem. News, 60, 77 (1884); J Soc. Chem. Ind., 1, 95 (1882).
79 Brunner: Pogg Ann., 15, 593 (1829); Firmenich: Dinglers Polytech J,
172, 370 (1864).
80 Stanek: Z. anorg. Chem., 17, 117 (1898); Christy: Am. J. Sci., (3) 17,
453 (1879); Ippen: Z. Krist., 27, 110 (1897).
298 LITHOPONE AND OTHER SULFIDE PIGMENTS
preparation is to heat the black sulfide at 100 in a closed vessel. 81 The
black modification is more readily soluble in concentrated sodium or
potassium sulfide than in ammonium sulfide. The pigment prepared
with alkali sulfide in the presence of excess mercuric salt is much
darker than the vermilion powder formed with ammonium sulfide.
The dark product consists of crystals sufficiently large to be easily
recognized by the naked eye. When the larger crystals are ground up
fine, the color is scarcely distinguishable from the ammonium sulfide
preparation.
It is difficult to remove completely the adsorbed sulfide and sulfur
from vermilion prepared in the wet way, and these impurities are detri-
mental to the product. For this reason, Picton and Linder 82 heated the
sol in a closed vessel from 160 to 170 for several hours until the de-
sired shade of red was produced. Instead of starting with the black
sulfide, Liebig 88 heated freshly prepared infusible white precipitate,
NH 2 HgCl, with ammonium polysulfide at 45; and Hausamann 84
heated a solution of the white precipitate in concentrated sodium thio-
sulfate.
In addition to its use as a pigment in paint, vermilion is employed
in making Chinese red ink and for coloring porcelain, paper, candles,
etc. Most of the numerous temples throughout China are painted red
with vermilion, since the Chinese look upon this as a lucky color.
ARSENIC SULFIDE PIGMENTS
Two arsenic compounds, the trisulfide As 2 S 3 and the disulfide
As 2 S 2 , have been used as pigments. The trisulfide known as King's
yellow is a brilliant but fugitive and extremely poisonous pigment
which was extensively used before the introduction of the similarly
colored lead and barium chromate. The pigment is prepared by pre-
cipitation from an arsenious acid solution with hydrogen sulfide or by
subliming a mixture of arsenic trioxide and sulfur. The sulfide is
found native as the mineral orpiment which is sometimes ground and
used as a pigment.
The disulfide occurs in nature as the mineral realgar. The com-
mercial red arsenic glass or ruby sulfur is an artificial disulfide pre-
pared by mixing arsenical pyrites and common pyrites in such propor-
81 Allen, Crenshaw, and Mcrwin: Am. J. Sci., (4) 84, 367 (1912).
82 Brit. Pat. 5,120 (1892).
as Ann, 5, 239; 7, 49 (1833).
*Ber., 7, 1747 (1874).
ANTIMONY TRISULFIDES 299
tion that the mixture contains about 15% of arsenic and 27% of sulfur.
Such a mixture is then sublimed and the resulting product is melted
with sulfur to give it the proper color. This so-called ruby sulfur is a
red glassy mass consisting of arsenic disulfide and sulfur in varying
amounts. 85 It was formerly used as an orange pigment, but, like the
trisulfide, it is no longer employed to any extent.
ANTIMONY SULFIDE PIGMENTS
Antimony Trisulfides
Classification. Antimony trisulfide pigments are usually divided
into two classes: the natural and the artificially prepared varieties.
The first class includes black crystalline stibnite and a brick-red trisul-
fide called metastibnite which is said to be amorphous ; 86 the second
class includes the precipitated sulfides which are sometimes used as
pigments. The precipitated trisulfides have been classified into two
subgroups : those thrown down in the presence, and those thrown down
in the absence, of hydrogen sulfide. The first group includes the tri-
sulfides precipitated by hydrogen sulfide directly and by the action of
acids on the antimonates and thioantimonates. The color is usually
some shade of orange, and the group may be designated as the "anti-
mony oranges." The pigments are sometimes called the "antimony
goldens," but the former term is preferable since antimony pentasulficle
is called the golden sulfide of antimony. The second group of trisul-
fides is formed by the interaction of an antimonous salt and sodium
thiosulfate. 87 The color is usually some shade of red, and the group
may be designated the "antimony crimsons."
Factors Influencing the Color. Currie 88 made an extended study
of the various factors influencing the color of antimony sulfide pig-
ments. The interaction of dilute solutions of sodium thiosulfate and
antimony trichloride at room temperature gives a light yellow pre-
cipitate which yields a fine yellow powder when washed and dried in
vacuum. On heating at 100 to 110 in a hot-air oven, the color changes
steadily through various shades of orange to a uniform red and finally
to a rich crimson. Raising the temperature to 150 to 170 causes the
color to change through a series of crimson shades to maroon, finally
becoming uniformly black. The presence of salts, chlorides especially,
"Roscoe and Schorlemmer: "Treatise on Chemistry," 5th ed. f 1, 708 (1920).
M Becker: Proc. Am. Phil. Soc., 20, 168 (1888).
"Cf. Long: J. Am. Chem. Soc., 18, 342 (1896).
J. Phys. Chem., 80, 223 (1926).
300 LITHOPONE AND OTHER SULFIDE PIGMENTS
and of acids, lowers the temperature at which the change to the black
modification is complete. 89 In the presence of 9 N HQ, the trans-
formation at the boiling point takes place quite rapidly; 90 but the
process goes on slowly even at room temperature. Two grams of the
orange compound in 5-cc portions of 12 N t 7 N, and N HC1 are trans-
formed completely into the black crystals in 0.5 day, 1 day, and 10. S
days, respectively; phosphoric, acetic, and sulfuric acids have little
effect within two months. 91 A 0,5-g sample of the orange sulfide in
contact with water was found by Lehrman 92 to change completely to
black in 10 months. Anions hasten the change in the order:
S > water > NO 3 > Cl > SO 4 > C 2 H 3 O 2 ; and cations in the order:
H > water > Na > NH 4 .
The wide variation in color from yellow through orange, crimson,
and maroon to black may be due entirely to variation in the size of the
particles, the larger particles possessing the darker shades. Since sul-
fides tend to give negatively charged sols, the pigments formed in the
presence of readily adsorbed anions are more highly peptized and
hence are lighter in color than those precipitated in the presence of
strongly adsorbed cations which favor coagulation into denser aggre-
gates possessing a darker color. Moreover, the particles precipitated
in the presence of gelatin are smaller and lighter in color than those
obtained under similar conditions in the absence of a protecting colloid.
Antimony Crimsons. The yellow and crimson trisul fides appear
to be amorphous to x-rays, but the final black product has a crystalline
structure identical with the natural stibnite. The maroon shades are
intimate mixtures of crimson with black sulfide; various maroons can
be synthesized by mixing the black and crimson powders. The den-
sities of the sulfides change in exactly the same order as the colors:
the yellow pigments have a density of approximately 4.10 to 4.12 ; the
crimson, 4.12 to 438; and the black, 4.6 to 4.S. 98 The density of the
maroon shades is usually about 4.4 to 4.5 ; the approximate density
may be calculated by comparing the color of the maroon pigment with
a mixture of known amounts of crimson and black sulfides of known
density.
Stibnite and the artificial black sulfides are formed under condi-
8 De Bacho: Ann. chim. applicata, 12, 143 (1919).
Mitchell: Chem. News, 67, 291 (1893).
i Wilson and McCrosky: J. Am. Chem. Soc., 48, 2178 (1921).
w J. Phys. Chem., 36, 2763 (1931).
MCurrie: J. Phys. Chem., 80, 232 (1926); cf. Kirchhof: Z. anorg. Chem.,
114, 266 (1920).
THE GOLDEN SULFIDE OF ANTIMONY 301
tions which would give relatively large crystals. Currie disintegrated
stibnite in an electric arc under water according to Svedberg's method,
obtaining a sludge of finely divided yellow crystalline particles. The
fact that the crystalline sulfide may be either yellow or black, depend-
ing on the size of the particles, supports Berthelot's 9 * contention that
the transformation from one color to another involves no measurable
heat effect. Currie was unable to prepare a crystalline crimson pig-
ment, the larger particles of amorphous trisulfide alone giving the
crimson color. It is possible, however, that the natural red meta-stib-
nite which is said to be amorphous would prove to be crystalline if
examined with x-rays.
Antimony Oranges. The trisulfides, as ordinarily precipitated by
hydrogen sulfide from a solution of trivalent antimony, are light golden
yellow and may be dried in vacuum to give a rich golden color. Unlike
the "antimony crimsons," heating to 105 to 110 causes little or no
change in color whereas further heating to 150 to 170 causes the pig-
ment to change through varying shades of brown to the black modi-
fication without showing any signs of the crimson color. The brown
colors were found to be mixtures of amorphous golden with crystalline
black trisulfide in varying proportions. Darkening through the crimson
and maroon shades is prevented by adsorption of hydrogen sulfide on
the surface of the golden particles which keeps them from coming into
intimate contact and coalescing to the darker shades before the trans-
formation to the black crystals takes place.
Both the orange and crimson trisulfide possess good hiding power
and mix well with oil. They cannot, of course, be employed with
alkaline vehicles, but they are quite stable in light and in the air.
Crimson antimony is the most valuable red pigment employed in color-
ing rubber. 95
The Golden Sulfide of Antimony
The term golden sulfide of antimony is frequently applied to anti-
mony pentasulfide, but, as we have seen (p. 227), it is usually a solid
solution of the tetrasulfide and sulfur. In general, the influence of
temperature in the neighborhood of 100 on the golden sulfide is slight,
a uniform darkening being the most noticeable effect. Above 115,
however, the tetrasulfide decomposes rapidly to trisulfide and sulfur,
94 Compt. rend., 189, 97 (1904); cf. t however, Guinchant and Chretien: 189,
51 (1904).
**Cf. Bierer: Chem. Age (N. Y.), 28, 194 (1920).
302 LITHOPONE AND OTHER SULFIDE PIGMENTS
further heating producing color changes like those of all trisulfides pre-
cipitated in the presence of hydrogen sulfide.
Golden sulfide of antimony is a valued pigment for rubber goods,
and a number of patents have been granted for its commercial produc-
tion. 06 In the rubber industry "golden sulfide" is applied to a fairly
wide range of products varying in shade from a golden yellow to a
deep orange and in composition from a nearly pure antimony trisulfide
to a mixture containing a relatively high percentage of tetrasulfide. 97
In addition to its value as a pigment, antimony sulfide is superior to
iron oxide pigment as a compounding ingredient. Mixes 98 in which
antimony sulfide is used possess strength and aging qualities " superior
to similar mixes containing iron oxide.
MOSAIC GOLD
Mosaic gold is the name given to the pigment stannic sulfide which
was extensively used in the eighteenth century as a gold coloring mat-
ter in paints. It was prepared then, as now, by subliming a mixture
of tin amalgam, sulfur, and ammonium chloride. For example, on
heating 18 parts of tin amalgam containing 6 parts of mercury with
6 parts of ammonium chloride and 7 parts of sulfur, the ammonium
chloride, mercuric chloride, and stannous chloride sublime, leaving the
pigment stannic sulfide in the form of beautiful golden yellow trans-
lucent scales. 100 At the present time it is employed as a bronzing
powder for wood, metal, wallpaper, and gypsum plaster.
**E.g, Chaillaux: Brit. Pat. 151,422 (1919); Stark: U. S. Pat. 1,414,836;
1,415,127 (1922); Bezzenberger : 1,528,394 (1925); Wilson: Can. Pat. 252,563
(1925).
"Cf. Luff and Porritt: J. Soc. Chem. Ind., 40, 2757 (1921).
08 The term "mixing" is applied to the operation by which sulfur and other
materials are incorporated with rubber.
Anderson and Ames: J. Soc. Chem. Ind., 42, 1367 (1923); Woodward:
Rubber Age, 1, 99 (1917).
"<>Woulfe: Phil. Trans., 61, 114 (1771).
PART IV
THE COLLOIDAL FERROCYANIDES AND FERRICYANIDES
CHAPTER XVI
COLLOIDAL FERROCYANIDES AND FERRIC YANIDES :
GENERAL PROPERTIES
The interaction of dilute solutions of metallic salts and alkali or
alkaline-earth f erro- and ferricyanides gives precipitates that vary in
physical character from highly gelatinous to flocculent; with concen-
trated solutions jellies result. The highly hydrous gels adsorb the
common anions so strongly that the precipitates are always contam-
inated with the corresponding alkali or alkaline-earth salts, especially
when the precipitation is accomplished with an excess of the latter.
In some instances definite double salts are formed, but because of their
high adsorption capacity they are seldom pure.
HEAVY-METAL FERROCYANIDES
Adsorption of Water
The metallic ferrocyanides are frequently assigned formulas which
suggest that they form definite hydrates. Copper ferrocyanide, for
example, is said to contain 10 1 molecules of water when in equilibrium
with 5% H 2 SO 4 ; 7, 2 when air dried; and 6 1 or 3 8 when dried over
concentrated H 2 SO 4 . Lowenstein 4 showed, however, that a prepara-
tion containing 10 molecules of water at 25 loses water continuously
without the formation of a definite hydrate. 6 Similarly, zinc ferrocya-
nide has been assigned 3, 6 4, 7 5, 8 6, 9 and 7 3 molecules of water when
1 Wyrouboff: Ann. chim. phys., (5) 8, 444 (1876); Lowenstein: Z. anorg.
Chem., 68, 125 (1909).
2 Rammelsberg: Pogg. Ann., 73, 80 (1848).
8 Hartung: Trans. Faraday Soc., 15 (3), 160 (1920) ; cf. f also, Tinker: Proc.
Roy. Soc. (London), 93A, 268 (1917).
*Z. anorg. Chem., 63, 125 (1909).
8 C/., also, Katz: Koninkl. Akad. Wetenschappen, Amsterdam, Verst, 31,
542 (1923).
Schindler: Phil. Mag., 36, 71 (1810).
Miller: J. Am. Chem. Soc., 18, 1100 (1896).
scumming and Good: J. Chem. Soc., 1924 (1926).
BLuckow: Chem.-Ztg., 16, 836 (1892).
305
306 COLLOIDAL FERROCYANIDES AND FERRICYANIDES
dried under varying conditions; but Lowenstein showed that the gel
loses ail its water continuously in the presence of 97% K^SO*. In
the light of these typical results, one should be slow to accept the for-
mulas for the alleged hydrates of the metallic ferrocyanides. It is
altogether likely that some of them form definite hydrates under cer-
tain conditions, but in every instance their acceptance should be based
on a phase-rule study of the ferrocyanide-water system and not on an
analysis of a product prepared and dried in a special way.
Adsorption of Ferrocyanide Ion
The strong adsorption of ferrocyanide ions by ferrocyanide gels is
well illustrated by the taking up of potassium, sodium, and hydrogen
ferrocyanides by copper ferrocyanide. 10 Solutions of the ferrocyanides
and cupric chloride were prepared such that SO cc of the former were
exactly equivalent to 100 cc of the latter, and, on mixing the respective
amounts, theoretically 0.5 g of Cu 2 Fe(CN) 6 was formed. To portions
of the copper chloride solution were added varying amounts of the
ferrocyanides in a total volume of 250 cc, and after standing 24 hours
the supernatant solutions were analyzed for excess ferrocyanide. A
typical set of observations is given in Table XXXVI, and all the data
TABLE XXXVI
CARRYING DOWN OF K 4 FE(CN)e BY Cu 2 FE(CN) 6
Cc solutions mixed (in 250 cc)
Equilibrium
K 4 Fe(CN)
concentration
carried down
K 4 Fe(CN),
mol/mol
CuCU
K 4 Fe(CN),
millimols/1
Cu a Fe(CN) 6
100
50
0000
100
60
200
100
70
400
100
75
069
0478
100
80
231
560
100
90
949
638
100
100
1 924
673
100
140
6 271
736
100
140*
22 076
0855
* Doable the concentration of previous solution.
: J. Phys. Chem., 84, 335 (1930); Miiller, Wegelin, and Kellerhoff:
J. prakt. Chem., (2) 86, 82 (1912).
ADSORPTION OF FERROCYANIDE ION
307
are shown graphically in Fig. 45. These results show the marked
tendency of the freshly formed gel to carry down ferrocyanides, espe-
cially the potassium and sodium salts. Indeed, this tendency is so
great that copper is present in the supernatant solution when equiva-
lent amounts of copper and alkali ferrocyanides are mixed. With
0.8
8 16
Equilibrium Concentration, Millimols per Liter
FIG. 45. Adsorption of ions by copper ferrocyanide gel.
hydro ferrocyanic acid, on the other hand, pure Gi2Fe(CN)0 is ob-
tained by mixing equivalent solutions.
Recognizing the alkali impurity in copper ferrocyanide gels, Du-
claux suggests that they should be represented by the general formula
Cu w KnFe(CN) 6 , where m + n/2 = 2. He believes that potassium
ferrocyanide combines with the copper salt to form a series of double
salts ; but the evidence for this point of view is inconclusive. It seems
308 COLLOIDAL FERROCYANIDES AND FERRICYANIDES
just as likely that the gel is copper ferrocyanide with potassium ferro-
cyanide adsorbed in varying amounts depending on the relative con-
centrations of the salts and the conditions of precipitation. This does
not preclude the formation of definite double salts under certain con-
ditions; but the composition of the precipitated gel can be varied con-
tinuously over such a wide range that one is not justified in concluding
that a double salt is formed simply from the analysis of an amorphous
mass.
From the curve in Fig. 45, it might be argued that a definite double
salt is formed of some such composition as 5Cu 2 Fe(CN) 6 -2K 4 Fe-
(CN) 6 and that the upper portion of the curve represents the ad-
sorption of K 4 Fe(CN) 6 by the double salt. Although this is possible,
the available evidence is inadequate to establish the existence of a
definite double salt of this formula. Fordham and Tyson 1X obtained a
characteristic electron diffraction pattern for copper ferrocyanide
membranes prepared by the interaction of copper sulfide and potassium
ferrocyanide. This proves that the precipitate is definitely crystalline
but it does not establish whether the precipitate is copper ferrocyanide
or a double salt. Milligan in the author's laboratory, had previously
secured an x-ray diffraction pattern for the gel which was found to
correspond essentially to the electron diffraction pattern of Fordham
and Tyson. A systematic study by analytical, x-ray, and electron dif-
fraction methods of the gels formed under various conditions will be
necessary to establish their composition (p. 310). Numerous definitely
crystalline alkali and alkaline-earth cupric ferrocyanides have been de-
scribed ; 12 but the only ones Messner 1S was able to prepare have the
simple formula X 2 +CuFe(CN) 6 , where X+ = K, Na, and NH 4 , or
X++CuFe(CN) 6 where X++ = Mg, Ca, Sr, and Ba.
Titration of Ferrocyanide Solutions
Procedure and General Results. Attempts have been made to de-
termine the composition of the precipitated ferrocyanide gels by elec-
trometric u and conductometric 15 titration. The most recent work of
11 J. Chem. Soc., 483 (1937).
C/. Bolley: Ann., 106, 228 (1858); Wonfor: Jahresbcr., 233 (1862);
Wyrouboff: Ann. chim. phys., (5) 8, 444 (1876).
Z. anorg. Chem., 8, 368 (1895).
"Bichowsky: Ind. Eng. Chem., 9, 668 (1917) ; Treadwell and Weiss: Helv.
Chim. Acta, 2, 680 (1919) ; Treadwell and Chervet: 5, 633 (1922) ; 6, 550 (1923),
"Kolthoff: Z. anal. Chem., 62, 209 (1923) ; Kolthoff and Verzijl: Rec. trav.
chim., 48, 394 (1924); cf. Ibarz and Fey to: Anales soc. espan. fis. quim., 34,
823 (1936).
TITRATION OF FERROCYANIDE SOLUTIONS
309
this kind was done by Britton and Dodd, 18 who made conductivity
measurements at 25 on mixtures of heavy-metal salts and potassium
f errocyanide : first, in the form of direct conductometric titration of
125 cc of 0.02 M salts with 0.1 M K 4 Fe(CN) 6 ; and second, on similar
mixtures of reactants after they had stood in a thermostat until equi-
librium was set up. In Fig. 46 are given the curves constructed from
10 15 20
0.1 Jf K 4 Fe (CN) 6> Cc
25
30
F IG 46. Titration curves for ferrocyanide solution with various metallic salts.
data corresponding to equilibrium conditions and the horizontal lines
which represent the specific conductivity of potassium sulfate formed
as a result of the equation: 2MSO 4 + (1 + X)K 4 Fe(CN) 6 -
M 2 Fe(CN) 6 XK 4 Fe(CN) 6 + 2K 2 SO 4 .
The values of X[= K 4 Fe(CN) 6 ] deduced from the conductivity
data, and from analysis of the precipitates formed in the presence of
wj. Chem. Soc., 1543 (1933).
310 COLLOIDAL FERROCYANIDES AND FERRICYANIDES
varying excess of potassium ferrocyanide, are summarized in Table
XXXVII. The direct analysis of the precipitates is of little use in
TABLE XXXVII
POTASSIUM FERROCYANIDE IN FERROCYANIDE GELS
[X - mol K 4 Fe(CN) 6 per mol J/ 2 Fe(CN)]
From conductivity measurements
From
M
First break
Second break
analysis
of gel
CcO lJlfK 4 Fe(CN)
X
CcO !JlfK 4 Fe(CN)
X
X
Cu
14
12
18 7
52
52-0 71
Zn
16 4
31
71-0 78
Cd
23
84
59-0 90
Co
22 3
78
50-0 74
Ni
19.7
058
23 4
087
56-1 22
Mn
23
0.84
83-0 92
Pb
12 5
very small
Ag
8 3
33
33
estimating the exact composition since they are peptized before they
can be washed free from entrained salt. Hence the values in the last
column of the table represent only very rough approximations.
Copper Ferrocyanide. In the titration of copper sulfate, normal
cupric ferrocyanide with a little adsorbed copper sulfate was obtained
whenever the precipitation was accomplished in the presence of excess
copper. When precipitation of copper is complete the composition of
the precipitate is Cu 2 Fe(CN) 6 '0.12K 4 Fe(CN) 6 . Further addition
of ferrocyanide causes no appreciable change in the conductivity, indi-
cating that all the ferrocyanide is taken up until the composition is
Cu 2 Fe(CN) 6 0.4-0.5K 4 Fe(CN) 6 . Thereafter some ferrocyanide re-
mains in the solution and the conductivity rises. These results confirm
the findings of Weiser shown in Fig. 45 ; but neither series of experi-
ments establishes the formation of a definite double salt. Britton and
Dodd are inclined to believe that the second break in their conduc-
tivity curve indicates the formation of a compound 2Cu 2 Fe(CN) 6 --
K 4 Fe(CN) 6 which adsorbs ferrocyanide strongly; but, since copper
ferrocyanide gel adsorbs potassium ferrocyanide so strongly, the value
of X = 0.52 at the second break cannot all be combined ferrocyanide.
TITRATION OF FERROCYANIDE SOLUTIONS 311
The value of X (combined) might be 0,4, giving a compound
5Cu 2 Fe(CN) 6 -2K 4 Fe(CN) 6 , with adsorbed K 4 Fe(CN) 6 , but as
already pointed out the experimental evidence is hardly sufficient to
justify this conclusion.
Zinc Ferrocyanide. The compound Zn 2 Fe(CN) 6 , like the corre-
sponding copper salt, results by precipitation with zinc ion in excess
or by the interaction of equivalent amounts of zinc ion and hydro-
ferrocyanic acid. Zinc may be estimated volumetrically by titration
with potassium ferrocyanide in neutral or acid solution. The follow-
ing reaction is said to take place: 2K 4 Fe(CN) 6 + 3Zn++ -K 2 Zn 3 -
[Fe(CN) 6 ] 2 + 6K+, with the formation of a definite double salt. Re-
ferring to Table XXXVII it will be seen that the observed equilibrium
concentration of K 4 Fe(CN) Q at the break in the curve is less than
corresponds to the above formula. In the rapid conductometric titra- .
tion, the break was observed at a point corresponding to the formula.
This indicates that the precipitate is not a definite salt of the above
composition but is either zinc ferrocyanide with adsorbed potassium
ferrocyanide or a double salt containing less potassium ferrocyanide
than the above, together with adsorbed potassium ferrocyanide. To
take care of possible variations in the composition of the precipitated
gel and so to obtain accurate results in the estimation of zinc, it is
essential not only that the conditions be rigidly controlled but also that
they be exactly the same as in the standardization of the ferrocyanide
solution against zinc.
Zinc may be estimated by potentiometric titration 1T with potassium
ferrocyanide in the absence of much sodium, rubidium, cesium, am-
monium, magnesium, calcium, copper, cadmium, manganese, and iron.
At the end point the zinc is all precipitated as the alleged double salt
K 2 Zn 3 [Fe(CN) fl ] 2 .
Uranium Ferrocyanide. Tetravalent uranium gives a red precipi-
tate of UO 2 Fe(CN) 6 . If alkali ferrocyanide is in excess a sol is
formed which is useful in the colorimetric determination of small
amounts of uranium. 18 Potentiometric titrations of uranyl nitrate and
acetate with potassium ferrocyanide indicate that the normal salt is
obtained with the acetate and a potassium uranyl double salt with the
nitrate. 19
"Bichowsky: Ind. Eng. Chem, 9, 668 (1917) ; Treadwell and Weiss: Helv.
Chim. Acta, 2, 694 (1919); Kolthoff: Rcc. trav. chirn., 41, 425 (1922); Kol-
thoff and Verzijl: 43, 380 (1924); Z. anorg. Chem,, 182, 318 (1923); Miiller:
128, 125 (1923) ; Saito: Chem. Abstracts, 24, 1595 (1930) ; 26, 4480, 5860 (1931).
18 Cf. Tissier and B&iard: Compt. rend. soc. biol., 99, 1144 (1928).
"Atanasiu: Bui. Chim. Soc. Roman*, Stiinte, 80, 77 (1927).
312 COLLOIDAL FERROCYANIDES AND FERRICYANIDES
Lead Ferrocyanide. Unlike the gelatinous ferrocyanide of copper,
zinc, cadmium, cobalt, nickel, manganese, and uranyl, lead ferrocyanide
is more granular, adsorbs little potassium ferrocyanide, and forms no
double salts. 20 The precipitate gives an electron diffraction pattern
indicating that it consists of simple cubic crystals larger than 200 A
in length. 11 The reaction between lead and ferrocyanide ions may be
used for the conductometric or potentiometric estimation of lead ; and
for the titrometric determination of ferrocyanide ions using sodium
alizarin sulfonate as an adsorption indicator (p. 132). 21
Silver Ferrocyanide. The precipitate appears to come down first
as Ag 4 Fe(CN) 6 , 22 going over rapidly and completely into insoluble
KAg 3 Fe(CN) 6 23 with the required excess of potassium ferrocyanide.
In conclusion it appears that the normal heavy-metal f errocyanides
may be prepared pure by adding solutions of alkali or alkaline-earth
f errocyanides to an excess of a solution of heavy-metal salt or by dis-
solving the hydroxides or carbonates of the metals in hydro ferrocyanic
acid. With excess soluble ferrocyanide the gels are always contam-
inated by adsorbed ferrocyanide, but, under suitable conditions, defi-
nite double salts may be formed. As a rule it is impossible to decide,
on the basis of existing evidence, whether a given gel is an adsorption
complex or a definite double salt. The only hope of settling this ques-
tion, with highly adsorptive gelatinous precipitates, rests on whethei
the gels formed under varying conditions are crystalline to x-rays and
so lend themselves to x-ray analysis.
Rhythmic Precipitation
By allowing alkali ferro- and ferricyanide solutions to diffuse into
jellies containing heavy-metal salts, most of the common ferro- and
ferricyanides have been precipitated in the form of rhythmic bands.
Holmes 24 obtained copper ferrocyanide in silica; Dounin and Schem-
jakin, 25 copper and silver f errocyanides in agar; and Chatter ji and
Dhar, 28 the ferro- and ferricyanides given in Table XXXVIII. Two
20 Cf., also, Kolthoff: Z. anal. Chem., 62, 211 (1923); Muller and Kogert:
75, 237 (1928).
21 Burstein: J. Russ. Phys-Chem. Soc., 59, 521 (1927).
28 Kolthoff: Z. anal. Chem., 62, 210 (1923); Kameyama and Gorai- J Soc.
Chem. Ind. f Japan (Suppl.), 80, 6B (1927).
"Steyer: Z. anal. Chem, 74, 106 (1928) ; Budnikov: 73,433 (1928)
"J. Am. Chem. Soc., 40, 1187 (1918).
2 5 Kolloid-Z M 47, 335 (1929).
*Kolloid-Z., 40, 97 (1926).
SOLS
313
types of rings were observed : type I rings were sharp with precipitate-
free spaces between; and type II were alternate layers of coagulated
and colloidally dispersed salt. By decreasing the concentration of dif-
fusing electrolytes, the number of type I rings is decreased and the
number and clearness of the type II rings are increased. The number
TABLE XXXVIII
RHYTHMIC BANDS OF FERRO- AND FERRICYANIDES
Jelly
Type of ring
Ferrocyanides of
Ferricyanidcs of
Gelatin
Gelatin . ...
AGrar.
I
II
II
Cu, Zn
Pb, Ag, Co, Ni, Ba,
Fe +++ , Sn
Mn, Ag, Cu, Ni, Zn
Co, Zn, Mn
Ag, Ni,Fe- +
Starch
II
Co, Ni, Ag
of rings is increased by exposure to light which has a coagulating
action on the dispersed salt.
Sols
Gels of the heavy-metal ferrocyanides are easily obtained in the
form of stable hydrosols by washing out the excess of electrolytes
from the gel or by the addition of a suitable excess of alkali ferro-
cyanide and dialyzing. By either procedure the sols are negatively
charged and are quite stable both because of strong adsorption of the
tetravalent potential-determining ions and the low solubility which re-
tards crystal growth.
HEAVY-METAL FERRICYANIDES
The solubility of the heavy-metal ferricyanides is greater than
that of the corresponding ferrocyanides; hence the precipitates are
more crystalline, less gelatinous, and have a lower adsorption capacity
(cf. p. 335). The normal salts may be obtained relatively pure by the
interaction of alkali ferricyanide and metallic salts especially if the lat-
ter are kept in excess. In the titration of salts of silver, copper, cad-
mium, cobalt, and nickel with potassium ferricyanide by the conducto-
metric method, a break occurs corresponding to complete precipitation
of the respective normal salts, 27 whereas with zinc the break occurs $%
"Kolthoff: Z. anal. Chem., 62, 215 (1923).
314 COLLOIDAL FERROCYANIDES AND FERRICYANIDES
below the calculated value for the normal ferricyanide and with man-
ganous salts the titration is erratic probably because of their reducing
action. Normal zinc ferricyanide conies down almost pure when pre-
cipitated in the hot with alkali ferricyanides. 28
The double salts with alkali and alkaline-earth ferricyanides are
usually definitely crystalline, and the ratio of heavy metal to alkali
or alkaline-earth metal is usually a small number.
The salts are too soluble to form stable sols. Copper ferricyanide,
probably the most insoluble of the common salts, gives a negative sol
which is stable for only a few days. 29
Special chapters will be devoted to colloidal copper ferro- and ferri-
cvanide and to Prussian blue.
28Cuta- Collection Czechoslov. Chcm. Commun., 1, 538 (1929).
*Luppo-Cramer: Kolloid-Z., 1, 353 (1907).
CHAPTER XVII
COLLOIDAL COPPER FERROCYANIDE: THE SOL
The formation, composition, and general properties of precipitated
copper ferrocyanide have been considered in the preceding chapter.
The hydrous gel like most heavy-metal ferrocyanides is a mordant
more especially for basic dyes. 1 The black silver image in an ordinary
photographic print may be converted into a red one, consisting in part
of copper ferrocyanide, by immersing the print in a copper toning
bath which is essentially a sol or solution of copper ferrocyanide in an
alkaline salt of citric, tartaric, or oxalic acid, together with other salts. 2
Owing to the mordanting action of the ferrocyanide dye, toned images
may be obtained by immersing the copper-toned print in an acid solu-
tion of basic dye such as thioflavine, victoria green, methyl green,
methylene blue, chrysoidine, or methyl violet. 8
FORMATION OF SOL
A stable brown sol of copper ferrocyanide results on mixing dilute
solutions of copper salt and potassium ferrocyanide with the latter in
slight excess. 4 More concentrated sols may be prepared in this way
by the use of protecting colloids such as gelatin or by peptizing the
freshly formed gel with potassium oxalate, 4 ammonium sulfate, 5 or,
preferably, with potassium ferrocyanide. 6 A pure stable, but rather
dilute, sol results on rapid, thorough washing of the gel by the aid of
the centrifuge, followed by dialysis. 7
Because of the marked tendency of copper ferrocyanide to carry
down alkali ferrocyanides, a sol free from alkali salt is best prepared
by peptization with dilute hydroferrocyanic acid of the gel thrown
down with copper acetate and the acid. 8 To prepare an easily peptiz-
1 Clark: Eastman Kodak Co., Abridged Publications, 2, 130 (1915-16).
8 Ferguson: Phot. J., 133 (1900).
8 Crabtree: J. Franklin Inst, 186, SIS (1918).
4 Graham: Phil. Trans. 151, 183 (1861); cf. Gurchot: J. Phys. Chem., SO,
90, (1926).
"Pappada: Kolloid-Z., 9, 136 (1910) ; Gazz. chim. ital., 41 II, 470 (1911).
Chakravarti and Dhar: Kolloid-Z., 42, 124 (1927).
'Weiser: J. Phys. Chem., 80, 1530 (1926).
Weiser and Milligan: J. Phys. Chem., 40, 1071 (1936).
315
316 COLLOIDAL COPPER FERROCYANIDE : THE SOL
able gel it is essential to avoid an excess of copper ion both during the
initial precipitation and during the repeated centrifugal washing neces-
sary to remove the acetic acid. The secret of the method is to have
ferrocyanide ion in very slight excess but not in sufficient amount to
prevent the throwing down of the gel with the centrifuge. When
properly prepared 10 to 20 g of Cu 2 Fe(CN) 6 are readily peptized by
1 1 of 0.002 N H 4 Fe(CN) 6 to give a clear red sol in which all the
ferrocyanide ion is adsorbed.
Ferrocyanide sols of copper, zinc, and iron in methyl alcohol may
be prepared by the interaction of alcoholic solutions of the respective
metallic chlorides and hydroferrocyanic acid followed by washing the
gels with methyl alcohol until they are peptized, giving clear sols. 9
GENERAL PROPERTIES
The negatively charged copper ferrocyanide sols are quite stable,
probably because of the high valence of the potential-determining ion.
When they are formed in the presence of an appreciable excess of
ferrocyanide, their stability toward the chlorides of potassium, barium,
and aluminum is decreased but little by boiling; dialysis increases their
stability toward the coagulating electrolytes. Chaudhury 10 has studied
the change in the mobility of the particles of sols prepared in different
ways on dilution and on adding electrolytes both in the presence and
in the absence of non-electrolytes. It is difficult to draw any definite
conclusions from the observations because of the complexity of the
sols employed. The work should be repeated with a sol as nearly
monodisperse as it can be made by centrifugal f ractionation ; and hy-
droferrocyanic acid should be used in its preparation in order to avoid
the presence of the large excess of adsorbed alkali ferrocyanides when
these salts are employed.
The relative viscosity of a sol containing 8.5 g Cu 2 Fe(CN) 6 /l was
1.03. 6 This was increased appreciably by adding electrolytes because
of agglomeration of the highly hydrous particles into larger aggregates.
COAGULATION OF SOL BY ELECTROLYTES
Coagulation by Single Electrolytes
Duclaux " first studied the precipitating action of various cations
on a copper ferrocyanide sol. A given volume of sol containing
Weiser and Mack: J. Phys. Chem., 84, 86 (1930).
10 J. Indian Chem. Soc., 10, 431 (1933); Chaudhury and Chatterjee: J.
Phys. Chem., 38, 244 (1929).
J. chim. phys., 5, 29 (1907).
COAGULATION BY SINGLE ELECTROLYTES
317
9.6 X 10- 8 gram atom of K+ required for coagulation the following
gram equivalents X 10" 6 of the several cations: Ag, 6.6; Cu, 3.4;
Al, 5.8; Fe+ + +, 6.2; UO 2 , 1S.O; Ba, 48.0; Mg, 98; K, 240. It is
claimed that these precipitation values are approximately equivalent,
irrespective of the valence, and are of the same order as the potassium
ion concentration, provided one considers the first four ions in the
above list and disregards the last four. On the basis of these and
similar data on hydrous ferric oxide sol, he considers the precipitation
process to be a definite stoichiometric chemical action, a double de-
composition of the ordinary type. This conclusion is not in accord
with the facts, even if one were permitted to disregard the last four
ions. The precipitation concentration expressed in equivalents is dif-
ferent for each ion and has no necessary connection with the potassium
ion concentration. In accord with the usual rule, the precipitating
power of an electrolyte is determined by the adsorbability of the pre-
cipitation ion, which is, in general, greater the higher the valence, and
by the adsorbability of the ion having the same charge as the sol. This
is illustrated by some of the observations of Sen 12 recorded in Table
XXXIX. In part A are given the precipitation values of salts with
cations of varying valence. It will be seen that the elements divide
TABLE XXXIX
PRECIPITATION CONCENTRATION (X) OF ELECTROLYTES FOR Cu 2 FE(CN)6 SOL
(Millimols/1)
A
Salts with cations of varying valence
B
Potassium Salts
Salts
X
Salts
X
KC1 .
35 6
92 5
0.445
458
0.538
760
058
0.034
0.038
KNO 8
28 7
27 5
47 5
47 5
80
95
170
205
260
NaCl
KBr
Ba(C*H,O a )a
K*SO 4
BaCla
K*HP0 4
SrCli
K*CrO 4
MgSO 4
KsC 4 H 4 O0
A1,(S0 4 ),
Ce(NOi)i
KC S O 4
KsFe(CN)e
Th(NO)4
K 4 Fe(CN)e
J. Phys. Chem., 29, 517 (1925).
318 COLLOIDAL COPPER FERROCYANIDE : THE SOL
themselves into three well-defined groups depending on the valence. 18
In part B are given the precipitation values of potassium salts ar-
ranged in decreasing order of precipitating power, which is the order
of increasing adsorption of the anions, on the assumption that the most
strongly adsorbed anion has the most pronounced stabilizing action.
The phenomenon of "acclimatization" (p. 197) was observed with
chlorides of the alkali metals. 14
Frankert and Wilkinson 18 determined the acidity or alkalinity de-
veloped on shaking solutions of potassium salts with powdered copper
ferrocyanide. The hydrogen ion concentration varies from slightly
acid with the weakly adsorbed chloride to highly alkaline with the
strongly adsorbed ferrocyanide. The order of adsorption of anions
based on the acidity developed is: Fe(CN) 6 > HPO 4 >
OH > Fe(CN) > SO 4 > Cl > NO 3 . It will be noted that the
order is similar to the one deduced from precipitation data for ions
common to both series. It is probable that the series would be more
nearly alike for the two sets of experiments if the same sample of
copper ferrocyanide were employed.
At this point it may be mentioned that uranyl ferrocyanide sol is
similar in appearance and in general properties to copper ferrocyanide
sol. 16 The molar precipitating power of cations for the sol is : Th >
Al > UO 2 > Ag > Ba[Cl] > Ba[OH] > H > NH 4 > Na." With
all common electrolytes the precipitation concentration decreases with
dilution of sol (cf. p. 193). Dilution affects the stability by disturb-
ing the adsorption equilibria of potassium and ferrocyanide ions. When
the ferrocyanide content of the sol is kept constant the precipitation
concentration varies but little with the content of UO 2 Fe(CN) 6 .
Mechanism of the Coagulation Process
Titration of Sols. 8 Electrolytes were added stepwise to sols pre-
pared by peptization of copper ferrocyanide gel with hydroferrocyanic
acid, and the change in concentration of the counter hydrogen ions
was followed with the glass electrode. This process is called titration.
The method was as follows : 20-cc portions of sol were mixed with a
definite quantity of electrolyte diluted to 5 cc and the mixtures allowed
is Cf. Pappada- Gazz. chim ital, 41 II, 470 (1911).
"Boutaric and Berthier: Bull. soc. chim., (5) 3, 696 (1936).
"I. Phys. Chem., 28, 651 (1924).
"Ghosh and Dhar: J. Phys. Chem., 31, 187, 649 (1927).
" Chatter jee: Kolloid-Z., 62, 214 (1930); J. Indian Chem. Soc., 12, 671
(1935).
MECHANISM OF THE COAGULATION PROCESS
319
to stand 4 hours before determining the hydrogen ion concentration.
A typical series of experiments with barium chloride as precipitating
electrolyte is given in Table XL. In the last column is recorded the
adsorption of barium ion at and above the precipitation value.
TABLE XL
TlTRATION OF Cu 8 FE(CN)e SOL WITH
02 WBaCh added
to 20 cc of sol.
Total volume 25 cc
[H+] X 10*
in solution
[H + ] X 10'
displaced
[Ba++] X 10
added
[Ba++] X 10 3
adsorbed
1 29
5
1 50
21
4
1
1 68
39
8
1 5
2 03
74
1 2
2
2 10
81
1 6
2 5
2 19
90
2
3
2 19
90
2 4
2 22
4
2 19
90
3 2
5
2 19
90
4
2 84
For comparing the action of different electrolytes a sol was titrated
with the chlorides of barium, strontium, and potassium, and the ad-
sorption of barium and strontium was determined at and above their
respective precipitation values. The results are summarized in Fig.
47. The precipitation concentration is indicated by an arrow.
The behavior of the sol toward electrolytes is so much like that of
sulfur sol " and the hydrous oxide sols (Vol. II, pp. 73, 114, 142) that
it can be explained by a similar mechanism. The potential-determining
ions are the common ferrocyanide ions, and the counter ions are hy-
drogen ions. In this respect it is similar to Verwey and Kruyt's silver
iodide sol (p. 115) in which the potential-determining ions are the
common iodide and the counter ions are hydrogen. On account of the
small ionization constant of the fourth hydrogen ion in hydro ferro-
cyanic acid, it is probable that, even in the dilute solutions employed,
the anion is chiefly HFe(CN) 6 (represented by R in Fig.
47) rather than Fe(CN) 6 . In any event, the negative ions may
be assumed to form the inner portion of a double layer surrounding
the hydrous particles of cupric ferrocyanide, as represented diagram-
is Weiser and Gray: J. Phys. Chem., 89, 1163 (1935).
320
COLLOIDAL COPPER FERROCYANIDE : THE SOL
matically in Fig. 48A. The counter ions are hydrogen ions which form
the diffuse outer portion of the double layer. Some of the hydrogen
ions are held so strongly (adsorbed) by the attractive force of the
inner layer that they are not detected by a hydrogen electrode, whereas
others, because of a relatively higher kinetic energy, exert sufficient
osmotic repulsive force against the attraction of the adsorbed ferro-
cyanide ions that they are a part of the intermicellar solution and thus
influence the hydrogen electrode. These are represented in the dia-
0.02 N Electrolyte Added, Co
FIG 47. Adsorption of precipitating cations and the displacement of hydrogen in
the titration of copper ferrocyanide sol with the chlorides of barium, strontium,
and potassium.
gram beyond the dotted line. On adding an electrolyte such as barium
chloride to the sol, the bivalent barium ions arc attracted more strongly
by the inner layer than the counter hydrogen ions, as shown diagram-
matically in Fig. 48B, and the thickness of the double layer is reduced.
At the same time, some adsorbed hydrogen ions are displaced and are
detected in the intermicellar solution. This contraction of the double
layer or compression of the outer layer resulting from stronger ad-
sorption of barium ions than hydrogen ions under the prevailing rela-
tive concentrations, causes a lowering of the (-potential on the par-
ticles; when this is reduced sufficiently, coagulation takes place. The
displacement of adsorbed hydrogen ions by adsorbed barium ions is far
from an equivalent displacement, since most of the adsorbed barium
ions correspond to hydrogen ions which are in the intermicellar solu-
MECHANISM OF THE COAGULATION PROCESS
321
tion and so are measurable potentiometrically in the original sol. It is
not at all surprising that in certain instances all the adsorbed hydrogen
ions are displaced before sufficient barium ions are adsorbed to reduce
the {-potential to the coagulation point.
Since most of the added barium ions are adsorbed at concentra-
tions slightly above the coagulation value, it would follow that most if
B
FIG. 48. Diagrammatic representation of the constitution of a particle in a
copper ferrocyamcle sol (^) before and (B} after the addition of barium
chloride.
not all the added barium ions are adsorbed below the coagulation value.
Since the total barium adsorbed above the coagulation value is some-
what greater than the total hydrogen ions in the sol, it follows that, to
a certain extent, both barium ions and chloride ions are adsorbed in
equivalent amounts from the barium chloride solution. The observed
phenomena are almost identical with those using the positive hydrous
oxide sols of iron, aluminum, and chromium in which the counter ions
are chloride ions and the precipitating electrolyte is potassium sulfate.
The behavior of strontium chloride is similar to that of barium chlo-
ride, whereas potassium chloride has a lower precipitating power, be-
cause potassium ions are less strongly adsorbed and displace hydrogen
less strongly than the bivalent ions at the same concentration.
322 COLLOIDAL COPPER FERROCYANIDE : THE SOL
From the above observations it is concluded that the lowering of
the f-potential of the cupric ferrocyanide particles on the addition of
electrolytes is caused by the contraction of the double layer or, if pre-
ferred, by a compression of the outer layer resulting from adsorption
of the added cations. At all concentrations the cation adsorption is
much greater than the hydrogen-ion displacement, since more of the
counter ions in the diffuse layer are in the intermicellar solution. The
adsorption is for the most part an exchange phenomenon, in which the
cations carried down are in exchange with the counter hydrogen ions
of the diffuse outer layer. The apparent difference in behavior between
the sols investigated by the author and the silver iodide sol studied by
Verwey and Kruyt (p. 124) is that in the former the potential reduc-
tion results from adsorption of precipitating ions, whereas in the latter
adsorption may not be essential for all the potential reduction.
Some observations of Bhatnagar 19 may be of importance for the
theory of the electrolyte coagulation process. Various sols were pre-
pared, including copper ferrocyanide, and coagulated by electrolytes
with precipitating ions of varying valence. By means of a special elu-
triating device, it was found that, for sols of one particle size, the
ratios of the elutriating velocities for the particles of the coagula ob-
tained by salts with univalent, bivalent, and trivalent precipitating ions
are, 1:2:3. A close relationship is thus indicated between size of
particles formed by agglomeration and the valence of the precipitating
ion. With spherical particles falling in a viscous liquid, the velocity of
fall, which corresponds to the elutriating velocity, is proportional to the
square of the radius, that is, to the extent of surface of the particles.
If Bhatnagar's observations are correct it follows, therefore, that the
magnitude of the charge present on the precipitating ion will determine
the extent of surface of particles formed by coagulation of sols. Un-
fortunately, the experimental procedure is open to so many possible
sources of error, and the observations are so incomplete, that any de-
ductions based on the recorded data would seem to be premature.
Coagulation by Mixtures of Electrolytes: Ion Antagonism
The phenomena observed in the precipitation of arsenic trisulfide
sols by salt pairs were attributed primarily to: (1) ion antagonism be-
tween certain cations in the sense that each cuts down the adsorption
of the other and (2) the stabilizing effect of anions which opposes the
precipitating action of the cations (p. 216). In some systems it is pos-
Bhatnagar, Mathur, and Shrivastava: J. Phys. Chem., 28, 387 (1924).
COAGULATION BY MIXTURES OF ELECTROLYTES 323
sible to eliminate the first effect entirely. Thus Sen, 20 working with
copper ferrocyanide sol, observed an increase in precipitation value of
potassium chloride and of barium chloride in the presence of potassium
ferrocyanide. It is quite obvious with mixtures of potassium chloride
and potassium ferrocyanide, that cation antagonism cannot come in,
since both precipitating cations are the same. Mukherjee and Ghosh ai
observed a similar behavior with mixtures of sodium benzoate and
sodium chloride on arsenic trisulfide sol, and the author confirmed
this with potassium ferrocyanide and potassium chloride on the same
sol. The determining factor in these systems is the stabilizing action
of the relatively strongly adsorbed benzoate and ferrocyanide, re-
spectively.
Consider Graham's ferric oxide sol, which owes its stability to
preferential adsorption of hydrogen ion derived from hydrolysis of
ferric chloride. The stability of this sol falls off as the hydrogen ion
concentration is decreased by dialysis, and if the dialysis is continued
long enough all the sol will precipitate; conversely, if hydrochloric acid
is added to a highly purified sol, the stability toward all electrolytes will
increase. Similar stabilization would be expected on adding ferric
chloride, aluminum chloride, or lanthanum nitrate, as Freundlich and
Wosnessenski 22 have shown. With the relatively insoluble Pean de
St. Gilles sol, a maximum in the stability is reached on adding hydro-
chloric acid, and at a suitable concentration coagulation takes place. 28
Similarly, the stability of copper ferrocyanide sol falls off with decreas-
ing concentration of the stabilizing ferrocyanide ion. On adding potas-
sium ferrocyanide to a highly purified sol, the stability towards all
electrolytes should increase to a maximum and then fall off as the
coagulating action of potassium ion begins to predominate. This is
exactly what Sen observed with two electrolytes of widely varying
precipitating power. The precipitation value of both potassium chlo-
ride and barium chloride is increased to a maximum that lies above
the value for either electrolyte alone. 24
The observations of Sen have been confirmed and extended with a
highly purified copper ferrocyanide sol. 26 The results with mixtures
20 J. Phys. Chem., 29, 354 (1925); J. Indian Chem. Soc., 3, 81 (1926).
21 J. Indian Chem. Soc., 1, 213 (1924).
22 Kolloid-Z. f 83, 222 (1923).
"Weiser: J. Phys. Chem., 26, 665 (1921).
"Sen: J. Phys. Chem., 29, 517, 539 (1925); Sen and Mehrotra: Z. anorg.
Chem, 142, 345 (1925).
^Weiser: J. Phys. Chem., SO, 1531 (1926); cf. Chatterjee: J. Indian Chem.
Soc., 12, 671 (1935).
324
COLLOIDAL COPPER FERROCYANIDE : THE SOL
of potassium ferrocyanide and both potassium chloride and barium
chloride are shown graphically in Fig, 49. It will be seen that the
addition of potassium ferrocyanide causes the precipitation value of
potassium chloride to mount sharply to a value more than six times
that of pure chloride. This indicates that the purity of the original
sol with respect to ferrocyanide ion is quite high. After the maximum
is reached, the curve bends sharply and then follows an almost linear
course, as Sen also observed. This means that, if one takes an impure
copper ferrocyanide sol, the precipitation values of mixtures of potas-
sium ferrocyanide and chloride will be approximately additive on ac-
200
1.00
10.75
= 0.50
B 0.25
'0 12.5 25 37.5 50 62.5
K 4 Fe(CN) 6l Millimols per Liter
FIG. 49. Precipitation of copper ferrocyanide sol with mixtures of electrolytes.
count of the absence of cation antagonism. A similar type of curve
should result with mixtures of barium ferrocyanide and barium chlo-
ride. This was found to be true so far as the linear part of the curve
is concerned ; but the sharp maximum could not be detected with cer-
tainty since the precipitation concentrations of the chloride and ferro-
cyanide of barium were so close together.
Turning to the potassium ferrocyanide-barium chloride curve, con-
trary to Sen's observations, one is impressed with the very great simi-
larity to the curves obtained with alkali-alkaline earth salt pairs on
arsenic trisulfide sol (p. 220) . Since in the latter system both cation
antagonism and the stabilizing action of the anion play a role, it is
possible that the difference between the potassium ferrocyanide-
barium chloride curve, and the potassium ferrocyanide-potassium
COAGULATING ACTION OF ALCOHOLS 325
chloride curve with copper ferrocyanide sol, is caused by the absence
of cation antagonism in the first instance and its presence in the
second.
It should be emphasized, however, that the initial increase in pre-
cipitation value of barium chloride in the presence of potassium ferro-
cyanide results primarily from the effect of ferrocyanide ion. Thus,
the precipitation values of mixtures of potassium chloride and barium
chloride fall slightly below the additive value at the lower concentra-
tions of potassium chloride, but at higher concentrations of potassium
chloride the values are slightly greater than additive. The same is
true for mixtures of potassium chloride and cupric chloride and of
potassium chloride and cupric sulfate. 28 For arsenic trisulfide sol the
antagonism between potassium and barium ferrocyanide is greater than
between the corresponding chlorides ; and a similar condition doubtless
exists with copper ferrocyanide sols.
Coagulating Action of Alcohols
Copper ferrocyanide sol is precipitated by methyl alcohol in rather
high concentration. Gurchot 27 offered two possible explanations of
this phenomenon : first, that the coagulation is brought about by a de-
crease in surface tension ; and, second, that the negative charge on the
particles is reduced below the critical value by the selective adsorption
of alcohol. The second explanation is regarded as the more probable
since surface-tension decrease would tend to cause peptization rather
than agglomeration. Gurchot implies that the reduction in the nega-
tive charge of the ferrocyanide particles is caused by selective ad-
sorption of the positive radical of the alcohol ; but, because of the slight
polarity of the alcohol molecules, it seems likely that this effect will
be negligible. A more probable explanation is that alcohol decreases
the adsorption of the stabilizing ferrocyanide ion, either by displacing
the latter or by cutting down the ionization of the salt in the inter-
micellar liquid (p. 200). In accord with this, Sen 28 showed that the
purer the sol the greater is its stability toward alcohols. The co-
agulating power of a series of alcohols is in the order: methyK
ethyl < propyl < butyl.
2 C/., however, Gurchot. J. Phys. Chem., SO, 98 (1926).
*rj. Phys. Chem, 80, 99 (1926).
2 <J. Indian Chem. Soc., 2, 289 (1925).
CHAPTER XVIII
COLLOIDAL COPPER FERROCYANIDE AND FERRIC YANIDE:
THE MEMBRANES
Copper ferrocyanide and ferricyanide furnish such good examples
of semipermeable membranes that their behavior in this respect will
receive special consideration.
SEMIPERMEABLE MEMBRANES
Definition
A membrane is said to be semipermeable when it permits one con-
stituent of a solution usually water to pass through and does not
allow diffusion of the other constituents or one or more of the other
constituents. The plasma membranes of cells are natural membranes
of this kind. Traube 1 was the first to recognize a similar semipermea-
bility in certain gelatinous precipitates, the so-called precipitation mem-
branes of which copper ferrocyanide is the most familiar example.
Such a membrane is obtained by holding a solution of copper sulfate
in a glass tube which is subsequently immersed in a solution of potas-
sium ferrocyanide. The film of copper ferrocyanide formed at the
junction of the two solutions will allow water to pass through but not
dissolved substances such as sugar and certain salts.
If a solution is covered with a layer of the pure solvent, diffusion
of the dissolved substances takes place until the concentration through-
out the system is the same. The force driving the dissolved substance
from the more concentrated to the less concentrated solution until equi-
librium is attained is termed osmotic pressure. The existence of this
force may be demonstrated by placing a solution in a vessel with semi-
permeable walls closed except for a capillary tube, and immersing the
vessel in pure solvent. Since the septum is, by definition, permeable to
the solvent and impermeable to the solute, and since equilibrium will
be attained only when the concentrations on both sides are equal, it
follows that the solvent must pass through the membrane and dilute
*Arch. Anat. Physiol., 86 (1867).
326
THEORIES OF SEMIPERMEABILITY 327
the more concentrated solution. This manifests itself by a rise of
liquid in the capillary tube. An unsupported membrane of copper
ferrocyanide ruptures so easily that it is not suitable even for qualita-
tive observation of osmotic pressure. To get around this difficulty,
Pfeffer * deposited the ferrocyanide on the inside of a porous cup by
allowing the ions to diffuse into the supporting medium from opposite
sides. A more satisfactory method devised by Morse 8 consists in
driving the ions into the walls by means of an electric current. Mem-
branes supported in this way have withstood pressures as high as 130
atmospheres and have proved to be nearly ideal as regards semiper-
meability.
Theories of Semipermeability
Atomic Sieve or Ultrafilter Theory. To account for the action of
semipermeable membranes, Traube conceived of them as atomic or
molecular sieves through which progressively larger molecules diffused
with increasing difficulty. Thus a copper ferrocyanide membrane was
believed to be an ultrafilter which contains pores large enough for the
small molecules of water to get through but too small to allow the
larger sugar molecules to pass. This view was shown to be inade-
quate by the observations of Tammann, 4 Walden, 6 I. Traube, 8 Barlow, 7
and others 8 who made comparative tests on a number of semiper-
meable membranes with various diffusing substances. If the action is
that of a sieve or ultrafilter, it should be possible to arrange the mem-
branes in a series in the order of their permeability. On the contrary,
it was found that a membrane quite impermeable to most substances
may be more permeable to some than a membrane which, in general,
possesses high permeability. Moreover, the impermeability of a mem-
brane such as rubber to water and its permeability to the much larger
molecules of benzene and pyridine cannot be accounted for on the sieve
theory. Finally the most recent evidence indicates that a membrane
such as copper ferrocyanide is not a static system but a dynamic one
2 "Osmotische Untersuchungen," Leipzig (1877).
s Morse and his collaborators: Am. Chem. J., 26, 80 (1901); 28, 1 (1902);
29, 173 (1903) ; 84, 1 (1905) ; 86, 1, 39 (1906) ; 87, 324, 425, 558; 38, 175 (1907) ;
89, 667; 40, 1, 194, 266, 325 (1908) ; 41, 1, 92, 257 (1909) ; Berkeley and Hartley:
Proc. Roy. Soc. (London), 78A, 436 (1904); Phil. Trans., 206A, 481 (1906).
*Z. physik. Chem., 10, 255 (1892).
*Z. physik. Chem., 10, 699 (1892).
Phil. Mag., (6) 8, 704 (1904).
T Phil. Mag., (6) 10, 1 (1905); Findlay and Short: J. Chem. Soc., 87, 819
(1905).
BRahlenberg: J. Phys. Chem., 10, 169 (1906).
328 FERRO- AND FERRICYANIDE MEMBRANES
capable of undergoing reversible permeability under suitable conditions
(p. 338).
The Solution Theory. This theory of the action of the semiperme-
able membrane postulates that a membrane is permeable to such sub-
stances as dissolve in it and impermeable to those that do not. This
view was anticipated by Liebig 9 as early as 1848 when he said: "The
volume changes of two miscible liquids which are separated from each
other by a membrane depend upon the unequal wetting or attraction
which die membrane exerts on the two liquids." The first experi-
mental work with the object in view of testing their theory was carried
out by Lhermite 10 in 1855. In a test tube he placed some water, above
this a thin layer of castor oil, and above this a layer of alcohol. In the
course of a few days the alcohol had passed through the castor oil to
the water, leaving but two layers in the tube. Turpentine was substi-
tuted for the oil with the same results. Again, when a layer of chloro-
form was separated from a layer of ether by a layer of water, the ether
passed through the water to the chloroform. As a result of similar
observations on eight different combinations of this kind, Lhermite
reached the conclusion that substances which pass through membranes
first dissolve hi them. This mechanism has been supported by the
work of numerous investigators. 11 Thus Kahlenberg found that ben-
zene, toluene, and pyridine, which are soluble in rubber, diffuse through
rubber, whereas water which is insoluble in rubber does not pass
through. Moreover, trichloroacetic acid passes through a rubber mem-
brane when dissolved in benzene but only very slowly when dissolved
in water. Kahlenberg assumed not only a solution of the liquids in
membranes to which they are permeable but a kind of loose chemical
union as well.
Although solution in the membrane may be a necessary and suf-
ficient criterion for semipermeability in certain systems, Bigelow 12 and
Bartell" demonstrated conclusively that osmotic effects can be ob-
'"Ursachen der Saftebewegung," Braunschweig (1848); Ann., 121, 78
(1862).
10 Ann. chim. phys., (3) 48, 420 (1855).
" Graham: Phil. Trans., 144, 177 (1854) ; 151, 183 (1861) ; Nernst: Z. physik.
Chem., 6, 37 (1890); Tammann: 10, 255 (1892); Overton: 22, 189 (1897);
Barlow: Phil. Mag., (6) 10, 1 (1905); Kahlenberg: J. Phys. Chem., 10, 169
(1906).
J. Am. Chem. Soc., 29, 1576, 1675 (1907); 31, 1194 (1909); Bigelow and
Robinson: J. Phys. Chem., 22, 99, 153 (1918).
i J. Phys. Chem., 16, 659 (1911); 18, 318 (1912); J. Am. Chem Soc., 86,
646 (1914); 38, 1029, 1036 (1916).
MEMBRANE AND SUGAR SOLUTION 329
tained with inert materials where neither solution nor chemical reac-
tion can take place. Thus porous cups served as semipermeable mem-
branes when the pores were sufficiently fine or when they were clogged
to a certain extent with such substances as barium sulfate, lead
chromate, or lead sulfate. Moreover, finely divided materials such as
silica, carbon, and metallic copper, silver, and gold acted as semiper-
meable membranes when compressed into discs containing very fine
pores. The limits of the pore diameters between which osmosis can
take place are not definitely known, but they undoubtedly vary from
substance to substance. Bartell found the upper limit to be around
0.9 /A with unglazed porcelain clogged with various materials. This is,
of course, much too large to represent the dimensions of molecular
interstices which Traube believed to be essential for semipermeability
in membranes.
The Adsorption Theory. The osmotic phenomena observed with a
distinctly porous, non-soluble membrane, such as a clogged porous
plate, are doubtless the result of negative adsorption as suggested by
Mathieu 14 and Nathansohn 15 and emphasized by Tinker 18 and Ban-
croft. 17 If a solid adsorbent takes up relatively more of a solvent
than of the dissolved substance, we have negative adsorption and the
solution becomes more concentrated. Mathieu observed this phe-
nomenon with a number of solutions using porous plates, membranes,
or capillary tubes as adsorbents. With normal solutions, the concen-
tration in certain capillary tubes was found to be as low as one-tenth
normal. The difference in concentration increases with decreasing
radius of the capillary tubes, and Mathieu concludes that with suffi-
ciently fine capillaries water alone would be adsorbed. The importance
of this for the theory of semipermeable membranes was recognized
especially by Bancroft and by Tinker who gave experimental support
to the theory by his work with copper f errocyanide and sugar solutions.
Sufficiently strong and irreversible positive adsorption by a mem-
brane may render it semipermeable under special conditions (p. 332).
The Copper Ferrocyanide Membrane and Sugar Solution
Since a gelatinous precipitate consists of myriads of particles en-
meshed in a network which entrains liquid (Vol. II, p. 9), one might
expect the copper ferrocyanide membrane to be porous in the sense
"Ann. Physik, (4) 9, 340 (1902).
" Jahrb. wiss. Botan., 40, 431 (1904).
"Proc. Roy. Soc. (London), 92A, 357 (1916); 93A, 268 (1917).
" J. Phys. Chem., 21, 441 (1917).
330
FERRO- AND FERRICYANIDE MEMBRANES
that a very fine porous plate is porous and to be semipermeable because
of strong negative adsorption. Tinker 16 examined these artificial
membranes microscopically and arrived at the conclusion that they are
granular in character 18 with particles having diameters between 100
and 1000 m/*. It is probable that the primary particles are much
smaller than this and that what he was measuring was the diameter
of secondary particles from partial agglomeration. Indeed, the electron
diffraction pattern of the membrane obtained by Fordham and Tyson
(p. 308) indicates that the crystal size is not larger than 100-150 A.
In any event, the important thing is that, independently of Mathieu,
Nathansohn, and Bancroft, he concluded that negative adsorption in
such a porous system is the primary cause of the semipermeability.
In support of this view, he demonstrated that cane sugar which ordi-
narily does not diffuse through a copper ferrocyanide membrane is
adsorbed negatively by the salt. Ten-gram samples of finely divided
copper ferrocyanide were shaken with sugar solutions of various
strengths and the change in concentration determined polarimetrically.
From this, the amount of water adsorbed was calculated. The results
with four samples dried in different ways and to different degrees are
given in Table XLI and shown graphically in Fig. 50. Correction was
TABLE XLI
ADSORPTION OF WATER FROM CANE SUGAR SOLUTIONS BY Cu 8 FB(CN)e
Water adsorbed by 100 g of dry Cu 2 Fe(CN) 6
Strength
1
2
3
4
%
Dried
com-
pletely
Dried over H 8 SO
10g-0689gH a O
Dried in air at 80
for 3 hours
10g= 1.38gHO
Not dried
lOg = 3 14gH 2 O
5
27
10
18 2
23 9
22 7
24.5
20
15.0
....
22.5
20.1
40
13 4
18
22 2
200
60
13 2
14 7
20.8
14.0
made for any water present in the adsorbent before immersing in the
solution. It will be seen that sample 3, which had been dried in air,
C/., also, Gurchot: J. Phys. Chem., 80, 99 (1926).
MEMBRANE AND SUGAR SOLUTION
331
adsorbs the most Water, and sample 1, which was completely dehy-
drated in a hot oven, adsorbs the least. Complete dehydration is evi-
dently accompanied by partial coalescence with consequent decrease in
specific surface.
Tinker believes that the completely dried ferrocyanide first hy-
drates to Cu2Fe(CN)e 3H 2 O which continues to adsorb water. Since
the amount of water in the compound appears to be determined
entirely by the conditions of drying, it is probable that no definite
hydrate exists (p. 305). So far as the theory of the semipermeable
10 20 30 40 50 60
Grams C J2 H 22 O n in 100 Grams H 2
FIG. SO. Adsorption of water from sugar solutions by copper ferrocyanide gel.
membrane is concerned, it is of course immaterial whether the salt
forms a hydrate under any conditions.
Hartung 19 objects to Tinker's conclusion that the impermeability
of copper ferrocyanide to sugar results from negative adsorption, on
three counts : First, the adsorbent used by Tinker was thoroughly dried
copper ferrocyanide powder and not the hydrous gel which constitutes
the osmotic membrane. Second, Tinker showed merely that water was
adsorbed by the powdered salt more strongly than sugar; but he did
not show that sugar was not adsorbed at all. Finally Hartung him-
self found that potassium sulfate is adsorbed by copper ferrocyanide
more strongly than potassium chloride and the sulfate passes the mem-
brane less readily than the chloride. All these objections will be an-
swered in subsequent sections.
"Trans. Faraday Soc, 15 (3), 160 (1920).
332 FERRO- AND FERRICYANIDE MEMBRANES
The Copper Ferrocyanide Membrane and Ferrocyanide Solutions
Although the impermeability to sugar of a copper ferrocyanide
membrane may be accounted for by the observed negative adsorption,
it is not immediately obvious why the membrane should be impermeable
to potassium ferrocyanide. That ferrocyanides do not pass the mem-
brane was pointed out by Tammann, Walden, and Donnan 20 and con-
firmed by Weiser. 81 Parchment membranes impregnated with copper
ferrocyanide do not allow the diffusion of dilute solutions up to N
K 4 Fe(CN) 6 into isotonic sugar solutions, and pure water is forced
through a membrane when a 0.1 JV K 4 Fe(CN) 6 solution is put under
pressure in an ultrafiltration apparatus. The greater tendency of the
membrane to allow relatively high concentrations to pass probably
results from partial coagulation of the colloidal film thereby opening
up cracks at weak points in an imperfectly formed septum.
An explanation of the above-mentioned behavior assumes with Col-
lander 22 that the copper ferrocyanide membrane is an ultrafilter or
sieve which screens out the relatively large ferrocyanide ion. The
objection to this view is that the membrane is permeable to much larger
ions and molecules such as phosphomolybdate, so that the screening
action is not that of simple ultrafiltration. A second hypothesis is that
ferrocyanides like sugar are negatively adsorbed by copper ferrocya-
nide. This view is contrary to the well-known strong positive adsorp-
tion of ferrocyanide ion by the gels. Finally it is assumed that a
gelatinous membrane is impermeable to its own ions. This lacks ex-
perimental support since the membrane is permeable to copper ion
which is much less strongly adsorbed than ferrocyanide.
Referring to Fig. 45 it will be seen that ferrocyanides are adsorbed
by copper ferrocyanide in the order: K 4 Fe(CN) 6 > Na 4 Fe(CN) 6 >
H 4 Fe(CN)g. The differences would be less marked if it were demon-
strated that double salts are formed with the alkali salts (p. 308).
But, in every instance, the adsorption is so strong that within a certain
range the adsorption is practically irreversible. Therein would seem to
lie the explanation of the impermeability of the membrane to ferro-
cyanide. The adsorption is so strong that the fixed walls of the pores
hold chains of oriented molecules extending from a monomolecular
film on the surface throughout the pore solution. The practically ir-
reversibly adsorbed ferrocyanide is sufficient to saturate the pore water
20 Donnan and Allmand: J. Chem. Soc, 106, 1941 (1914); Donnan and
Garner- 116, 1313 (1919).
21 J. Phys. Chem., 84, 335 (1930).
a Kolloid-Beihef te, 19, 72 (1924).
MEMBRANE AND SUGAR SOLUTION 333
so that no more can enter from the side of the membrane in contact
with the solution; and, since the adsorption necessary to saturate the
pore solution is not reversible, no f errocyanide will pass into water or
sugar solution on the opposite side. Another way of stating the con-
dition is that ferrocyanide ion is adsorbed so strongly and irreversibly
that the actual solution in the pores is practically pure water.
Copper chloride and potassium sulfate are likewise positively ad-
sorbed by copper ferrocyanide (Fig. 45), but the membrane is per-
meable to them since the relatively weak adsorption is almost com-
pletely reversible throughout the entire concentration range. Between
one extreme, that of potassium ferryocyanide which will not pass the
membrane owing to strong adsorption that is almost irreversible within
a certain range, and the other extreme, that of copper chloride which
passes the membrane because of relatively weak adsorption that is
completely reversible, there are an indefinite number of gradations in
the degree of permeability with salts as a result of positive adsorption,
The Copper Ferricyanide Membrane and Sugar Solution 23
Since a copper ferricyanide membrane is impermeable to cane
sugar, it would follow from analogy with the corresponding ferro-
cyanide that sugar would be negatively adsorbed by the salt from
aqueous solution. To test this, copper ferricyanide was precipitated
from sugar solution and the change in concentration of the sugar was
determined before and after the precipitation. The procedure was as
follows: The calculated amount of CuCl2'2H 2 O to give 5 g of
Cu 3 [Fe(CN) 6 ] 2 was dissolved in SO cc of water in one container of
a mixing apparatus, 24 and an equivalent amount of K 3 Fe(CN) 6 to-
gether with a definite weight of sugar was dissolved in 100 cc of water
in a second container. After thorough mixing, followed by centri-
fuging, the concentration of sugar in the supernatant solution was
determined in a Reichert Soleil-Ventzke saccharimeter. For compari-
son, a determination was made of the concentration of a sugar solution
containing the same weight of sugar in the same volume of water
as above, together with KC1 equivalent to 5 g of Cu 3 [Fe(CN) 6 ] 2 .
The results of two experiments with different concentrations of sugai
are given in Table XLII. The saccharimeter reading in Ventzke de
grees is an average of 20 readings made under carefully controlled
conditions. It will be seen that the concentration of sugar is greater
"Weiser: J. Phys. Chem., 34, 1826 (1930).
2*Weiser: J. Phys. Chem., 34, 340 (1930).
334
FERRO- AND FERRICYANIDE MEMBRANES
TABLE XLII
ADSORPTION OF H 2 O FROM CANE SUGAR BY Cu 8 [FE(CN)]2
Substances mixed in
grams
Cu,[Pe(CN).]j
precipitating
Saccha*
nmeter
reading
HaO adsorbed by
Cu[Pe(CN) tf ]s
CuClj-2HjO
K s Fe(CN) 6
KC1
Cane
sugar
H,0
g
degrees
Ventzke
g/g
mols/mol
4. 1624
5 3570
8.0
150
5
19 75
0.0
3.6395
8
150 88*
19.30
666
22 6
4. 1624
5 3570
13.0
150
5.0
31 48
0.0 ,
0.0
3 6395
13.0
150 88*
31.05
476
16.2
* 4. 1624 g CuCla<2H*O contains 88 g H 2 O.
in the solution from which the gel separated; from this increase the
adsorption of water given in the last column of the table was calculated.
The results of the above experiments show conclusively that water
is adsorbed strongly relatively to sugar by precipitated copper ferri-
cyanide, the extent of the adsorption being greater the more dilute the
solution. Since this behavior is similar to that observed by Tinker
with dry copper ferrocyanide in sugar solutions, it would seem that
Hartung's criticism of the experimental procedure employed by Tinker
is not valid.
Although the above results furnish strong evidence in support of
the view that the impermeability of copper ferricyanide to sugar is the
result of marked negative adsorption, the case would be even stronger
if it could be shown that no sugar at all is adsorbed by the precipitated
gel. Unfortunately it is difficult to determine the presence of a trace
of sugar in the presence of an excess of copper ferricyanide, but the
following experiment indicates that little or none is adsorbed : A gram
of the ferricyanide gel precipitated in the presence of sugar was washed
repeatedly by the aid of the centrifuge until the wash water was free
from sugar, using 0.5 N KC1 solution which prevented peptization of
the gel. The suspended gel was digested for a day in dilute hydro-
chloric acid, a procedure which should invert any cane sugar that
might be present. A sample was then subjected to Fehling's test by
dissolving the gel in alkaline sodium tartrate solution and heating; the
results were negative.
Copper Ferricyanide Membrane and Electrolyte Solutions 23
Adsorption of Electrolytes. The adsorption of potassium ferri-
cyanide, sulfate, and chloride by copper ferricyanide was determined
MEMBRANE AND ELECTROLYTE SOLUTIONS
335
by a procedure similar to that used with copper ferrocyanide (p. 306).
The results are shown graphically in Fig. SI, in which the adsorption
is plotted on a scale twice as large as that in Fig. 45. Potassium ferri-
cyanide is adsorbed by copper ferricyanide gel less strongly than potas-
sium ferrocyanide is by the corresponding ferrocyanide gel. Moreover,
the adsorption of ferricyanide is not completely irreversible even at
very low concentrations, although the adsorption curve approaches
0.25
0.20
Equilibrium Concentration, Millimols per K 3 Fe(CN) 6 per Liter
0.05
K 2 S
25 50 75 100
Equilibrium Concentration, Millimols K 2 S0 4 or KCI per Liter
FIG. 51. Adsorption of potassium salts by copper ferricyanide gel.
quite close to the #-axis. The adsorption of sulfate is considerably
greater than that of chloride in accord with Hartung's observations
with copper ferrocyanide gel.
Permeability Experiments. Since sufficiently marked positive ad-
sorption of a solute by a membrane may render the membrane imper-
meable to the solute, one would expect that, other things being equal,
the more strongly a solute is adsorbed by a membrane, the more slowly
it would diffuse through. It is of course impossible to realize ideal
experimental conditions, since the size, extent of hydration, and mo-
bility of the ions and molecules are specific; nevertheless the rate of
diffusion of potassium ferricyanide, sulfate, and chloride through a
336
FERRO- AND FERRICYANIDE MEMBRANES
copper ferricyanide membrane is roughly inversely as their adsorba-
bility by the gel.
Some typical observations are recorded in Table XLIII. Thimbles
of parchment were impregnated with copper ferricyanide using special
care to obtain a perfect septum. The membranes were suspended to
the bottom of a stopper which fitted a 250-cc bottle. After IS cc of
TABLE XLIII
DIFFUSION OF SALTS THROUGH Cu8[FE(CN) 6 ]2 MEMBRANES
In cup
Outside cup
%
Membrane
No
Salt
at start
after 24 hours
diffused
through
ccO 04JUT
g
cc 04 M
g
membrane
1
KC1
15
1118
12.75
0950
85
2
KC1
15
1118
12 20
0910
81 3
1
K 2 S0 4
15
1400*
5 74
0536*
38 3
2
K 2 S0 4
15
1400*
5 13
0479*
34 2
1
K,Fe(CN),
15
1975
68
0089
4 5
2
K,Fe(CN) 6
15
1975
64
0083
4 2
* Calculated in BaSOj equivalent.
electrolyte was placed in the thimble it was hung in ISO cc of sugar
solution isotonic with the electrolyte, and allowed to remain 24 hours.
Duplicate experiments were carried out, first with potassium chloride,
then with the sulfate, and finally with the ferricyanide, using the same
thimbles for each series of observations. The amount of diffusion was
determined by analysis of the outside solution. It will be seen that
the very strongly adsorbed ferricyanide diffuses quite slowly, the
weakly adsorbed chloride quite rapidly, and the sulfate occupies an
intermediate position. It appears therefore that the presence of an
adsorbed solute in the membrane retards the rate of diffusion of that
solute into and through the pores.
Potassium ferricyanide was found to diffuse through a cadmium
ferricyanide membrane much more readily than through a copper fer-
ricyanide membrane. There are two reasons for this : first, the adsorp-
tion is more readily reversible at low concentrations; and second, the
salt is less gelatinous and hence gives a membrane with larger pores.
SUMMARY 337
Summery
The above experimental results support the theory that a membrane
will be impermeable to a dissolved solute: (1) provided that it exhibits
sufficiently strong negative adsorption that the adsorbed film of pure
solvent fills the pores full, or (2) provided that it exhibits sufficiently
strong positive adsorption that the pores are filled with a network of
oriented chains of adsorbed solute molecules to the point where no
more can enter, within the range that the adsorption is practically irre-
versible.
The adsorption theory of the action of the semipermeablc mem-
brane is opposed to the view that the membrane merely acts as a sieve
or ultrafilter with pores sufficiently small that dissolved molecules
above a certain size are held back while smaller ones can pass through.
Nevertheless, pore size is quite as important for the true semiper-
meable membrane which functions by an adsorption mechanism as it is
for the the true ultrafilter which functions as a sieve without the inter-
vention of adsorption phenomena. For example, when a solute fails
to pass a membrane because of negative adsorption, the pores must be
sufficiently small that the adsorbed film of solvent fills the pores full,
otherwise the solute may pass through the center of the pores. In
other words, a membrane of a given composition may be impermeable
to a solute because of negative adsorption if the pores are small enough
and may be permeable to the same solute in spite of negative adsorp-
tion if the pores are too large. This does not mean that the porous
membrane becomes a molecular sieve or filter when the pores become
small enough that the adsorbed film completely fills them. On the con-
trary, it means that a porous membrane which does not exhibit marked
adsorption for a solvent ordinarily acts as an ultrafilter or sieve allow-
ing molecules in solution to pass but holding back particles of colloidal
dimensions; but if there is some adsorption of the solvent, and the
pores of the membrane are made sufficiently fine, the membrane may
become impermeable to certain molecules in solution because of nega-
tive adsorption.
In his classical experiments on ultrafiltration Bechhold 25 used col-
lodion which exhibited so little negative adsorption, relatively, that it
did not change to a semipermeable membrane at any pore diameter
employed. On the other hand, dried collodion membranes with very
small pores may not be true molecular sieve membranes or ultrafilters
so Z. physik Chcm., 60, 257 (1907).
338 FERRO- AND FERRICYANIDE MEMBRANES
as is generally assumed. 26 Collander 27 considers such membranes to
be molecular sieves through which solutes diffuse at a rate inversely
proportional to the molecular size. But he finds that the velocity of
diffusion is not determined exclusively by the molecular size and sug-
gests casually that the discrepancies may be caused by solution and
adsorption processes in the membrane. Grollman 28 likewise found
that the sieve-like action of a collodion membrane is influenced by a
layer of adsorbed liquid on the pore walls.
REVERSIBLE PERMEABILITY OF MEMBRANES
A copper ferrocyanide membrane in water is a negatively charged
colloidal film analogous to a sol. Since pore size is as important in a
true semipermeable membrane as in a true ultrafilter, it follows that
the permeability of a copper ferrocyanide membrane to sugar will be
increased by the presence of any contiguous solute which causes it to
undergo partial agglomeration. This is apparently what happens in
certain instances cited by Bancroft and Gurchot. 29 For example, Bar-
low 80 found that alcohol makes a copper ferrocyanide membrane per-
meable to sugar. Since sugar is insoluble in alcohol, this permeability
cannot be the result of increased solubility of sugar in the membrane.
Similarly, Czapek 81 observed the exosmosis of tannin from cells of
Echvueria which were exposed to the action of various alcohols; the
critical concentration for methyl alcohol was about \S% ; for ethyl,
10-11%; for propyl, 4-5%; for butyl, 1-2%; and for amyl, 0.5%.
Since the above solutions possess a surface tension about 68% that for
pure water, Czapek suggested that the surface-tension lowering was
responsible for the exosmosis. In view of the fact that tannin is in
colloidal solution, it is more likely that the exosmosis was due to partial
coagulation of the cell membrane by the alcohols. In support of this
view, Gurchot showed that copper ferrocyanide membranes are ren-
dered permeable to sugar by suitable concentrations of the several
alcohols. 82 Since a sol of copper ferrocyanide is precipitated by alcohol
26 Cf. Michaelis: Colloid Symposium Monograph, 6, 135 (1927); J. Gen.
Physiol., 8, 33 (1925); Michaelis and Perlzweig: 10, 575; Michaelis, Ellsworth,
and Weech: 671 (1927); Michaelis and Fujita: Biochem. Z., 161, 47 (1925).
* 7 Soc. Sci. Fennka Commentationes Biol., (6) 2, 1 (1926).
"J. Gen. Physiol., 9, 813 (1926).
89 J. Phys. Chem., 28, 1279 (1924); Gurchot: 80, 83 (1926).
o Phil. Mag., (6) 10, 1 (1905); 11, 595 (1906).
"Ber. deut. botan. Ges., 28, 159 (1910).
*C/., also, Sen: J. Indian Chem. Soc. ( 2, 289 (1925).
REVERSIBLE PERMEABILITY 339
(p. 325), it is reasonable to suppose that the dilute alcohol solutions in
contact with the membranes cause partial agglomeration, giving pores
so large that the film of adsorbed water does not fill them completely
and so allows sugar to pass. Salts such as sodium chloride and calcium
chloride likewise render the membranes permeable to sugar. The
critical concentration was about 2% for sodium chloride and below \%
for calcium chloride which contains the more strongly adsorbed bi-
valent ion. Below the critical concentration, the salts will dissolve in
the adsorbed water layer and will pass through the membrane by os-
mosis without coagulating it. But when the salt concentration is suf-
ficiently high to neutralize the negative charge on the ferrocyanide
below a critical value, partial agglomeration gives pores through which
both the salts and sugar can pass by diffusion. When a dilute solution
of copper sulfate was placed on one side of a membrane and one of
potassium ferrocyanide on the other side, the membrane did not be-
come permeable to sugar in the presence of alcohol or salts, the mem-
brane-forming reagents preventing the formation of effective pores.
This observation suggested that a membrane rendered permeable to
sugar by alcohols could be made impermeable once more by the addi-
tion of a suitable peptizing agent. Gurchot claims to have accom-
plished this reversal by the use of copper sulfate. It was assumed that
copper ion was the peptizing ion from analogy with the behavior of
the silver halides ; but since copper ferrocyanide is usually negative it
is quite as likely that sulfate ion which is fairly strongly adsorbed is
the effective one.
Although the permeability of a copper ferrocyanide membrane for
sugar will be increased by partial agglomeration and be restored by
repeptization, it seems to the author that such a mechanism need not
be invoked in all cases to account for change in permeability. The
degree of semipermeability in a given membrane is determined by the
extent of the negative adsorption and the size of the pores. If the
negative adsorption is not sufficiently marked, or if the pores are too
large, the pores will not be filled completely with the adsorbed film
and so will be more or less permeable. Now it is altogether probable
that the extent of the negative adsorption will be influenced by the
magnitude and sign of the charge on the colloidal particles and the
nature of any adsorbed ions. If this be true, it follows that the pres-
ence of electrolytes will influence the amount of the negative adsorp-
tion and so will change the permeability altogether apart from any
agglomeration or peptization. The important thing, in any case, is
that a membrane such as copper ferrocyanide is a dynamic system
340 FERRO- AND FERRICYANIDE MEMBRANES
which varies in permeability with the nature of the surrounding me-
dium and which is capable of reversible permeability under suitable
conditions. Since it is fairly well established that the cell is surrounded
by a semipermeable membrane comparable in certain respects to copper
ferrocyanide, this concept enables one to account for a number of ap-
parently contradictory facts in connection with the permeability of
living cells.
The composition of the cell wall is not known, nor is it known what
makes it semipermeable, let alone how the permeability is altered.
There can be no doubt, however, that the cell, in pursuing its metabolic
functions, does change its permeability in some way. Overton 88 as-
sumes that living cells are surrounded by a lipoid film. This in itself
would account for a great many of the permeability phenomena, but
such a film would not let water through and it is known that water
does get through. Moreover, a great many acid dyes which are in-
soluble in lipoids penetrate certain cells readily. More striking still is
the fact that the cells are normally impermeable to fruit sugar, cane
sugar, and other carbohydrates, the amino acids, the acid amides, and
many other substances which are foodstuffs for the cells and must get
into the cells from the outside. It is this condition which leads Hober 84
to remark : "What the cell can use it shuts out, and what it cannot use
it lets in."
The only way to account for these apparently contradictory facts
is to assume that the cell may be permeable to a given substance under
certain conditions and impermeable under others. Collander 85 con-
tends that protoplasm probably acts as an ultrafilter toward substances
which are not soluble in lipoids; but in doing so he has neglected to
distinguish clearly between a semipermeable membrane and an ultra-
filter and has failed to take into account the important fact that under
suitable conditions cells are both permeable and impermeable to the
same substance. Moreover, Collander reasons from analogy with the
behavior of copper ferrocyanide membranes with the membrane-form-
ing reagents on either side. He thus takes it for granted that cells
always have the membrane-forming reagents on either side of them.
This assumption is probably erroneous; in any event, it goes well
beyond our present knowledge.
as Z. physik. Chem., 22, 189 (1897) ; Pflugers Arch., 92, 115, 261 (1902).
M "Physikalische Chcmie der Zelle und der Gewebe," Sth ed, Part I, 503
(1922).
wKolloid-Beihefte, 10, 72 (1924); 20, 273 (1925).
MEMBRANE EQUILIBRIA 341
THE THEORY OF MEMBRANE EQUILIBRIA
Donnan's theory of membrane equilibria (Vol. 1, p. 315) deals with
the equilibria resulting when a membrane separates two electrolytes
containing one ion which cannot diffuse through the membrane. Don-
nan showed that, if one starts with two completely ionized electrolytes,
NaCl and Na7? f separated by a membrane impermeable to the ion R,
equilibrium will be established only when the product of the concen-
tration of sodium and chloride ions has the same value on both sides
of the membrane, thus
[Na+]i X [Cl-]i - [Na+] 2 X [Cl~] 2
where [Na+]! and [Cl-lj are the molar concentrations of sodium
and chloride ions on one side % of the membrane and [Na+] 2 and
[Cl~] 2 are corresponding concentrations on the opposite side of the
membrane. When equilibrium is reached, if the concentrations in one
solution are #Na and xC\, and in the other (y + #)Na, ^Q^ an( j 2 R f
the equation of products becomes
X2 = y(y + z)
It is obvious that x must be greater than y so long as z has a finite
value; hence the concentration of chloride ion in the first solution
must be greater than in the second, whereas the concentration of so-
dium ion must be greater in the second than in the first. This gives
rise to a potential difference E across the membrane which is repre-
sented by the equation
77 RT * *
E=3 _ log -
in which R is the gas constant, T the absolute temperature, and F the
faraday.
The accuracy of these deductions has been established by investi-
gations on a number of systems, for example: (1) with solutions of
potassium chloride and lithium chloride using a layer of amyl alcohol
as a membrane; 86 (2) with solutions of Congo red and sodium chlo-
ride using a parchment membrane; 86 and (3) with solutions of potas-
sium and sodium chloride and potassium and sodium ferrocyanide,
respectively, using a copper ferrocyanide membrane, 20 especially when
Donnan and Harris: J. Chem. Soc, 99 (1), 1554 (1911).
342 FERRO- AND FERRICYANIDE MEMBRANES
activities are substituted for concentrations. 87 It is of interest to
inquire into the nature of the action of the membrane in each instance.
In the first example, the amyl alcohol acts as a non-porous semiper-
meable membrane in which the lithium chloride dissolves and passes
through and the potassium chloride does not. In the second example,
the parchment functions as an ultrafilter or dialyzing membrane, the
concentration, osmotic, and electrical effects being caused by the in-
ability of the large ionic micelle of the colloidal electrolyte to diffuse
through the pores. In the third example, the copper ferrocyanide acts
as a semipermeable membrane for ferrocyanide ion according to the
mechanism described in this chapter.
s'Kameyama: Phil. Mag., 00, 849 (1925).
CHAPTER XIX
PRUSSIAN BLUE AND RELATED PRODUCTS
The most important of the iron cyanogen derivatives are the various
blue hydrous substances comprised under the general names Prussian
or Berlin blue and Turnbull's blue, discovered by Dippel in Berlin in
the first decade of the fifteenth century. The blue products are impor-
tant pigments, and the ease with which they may be obtained in the
sol state has led to their use in the investigation of colloid chemical
phenomena.
PRUSSIAN BLUE GELS
Formation and Composition
The common blue iron cyanogen compounds are obtained: (1) by
the interaction of ferric salts and ferrocyanides, (2) by the interaction
of ferrous salts and ferricyanides, (3) by the oxidation of ferrous
ferrocyanide, and (4) by the reduction of ferric ferricyanide. The
freshman is usually told that the product of the first reaction is ferric
ferrocyanide, Fe 4 + + + [Fe(CN) 6 ] 3 or Prussian blue, and the
product of the second reaction is ferrous ferricyanide, Fe 3 + + [Fe-
(CN)e]2 r Turnbull's blue. It is known, however, that on
mixing either a ferric salt and a ferrocyanide, or a ferrous salt and a
ferricyanide, the following equilibrium is set up before precipitation
begins :
Fe+++ + Fe(CN) 6 ^Fe++ + Fe(CN) a
This equilibrium is disturbed by the formation of a precipitate, the
composition of which is determined by the solubility relations which
exist. As a result of extended investigations, Hofmann 1 concluded
that all the blue gels are ferrocyanides, derivatives of hydroferro-
cyanic acid in which the hydrogen is either wholly replaced by ferric
iron or partly by ferric iron and partly by another metal. Hofmann
* Ann., 887, 1 (1904) ; 840, 267; 842, 364 (1905) ; 862, 54 (1906) ; J. prakt.
Chem., (2) 80, 150 (1909) ; cf. t also, Skraup: Ann., 186, 385 (1877) ; Vorlander:
Ber., 46, 181 (1913); Bowles and Hirst: J. Oil and Colour Chem. Assoc., 9,
153 (1926).
343
344 PRUSSIAN BLUE AND RELATED PRODUCTS
believed that, when alkali (X) ferro- or ferricyanides are employed,
the resulting hydrous gels were Fe 4 + + + [Fe(CN) 6 ] 3 or
XFe+++[Fe(CN) 6 ] or a mixture of the two. Midler 2 con-
firmed and extended the work of Hofmann by mixing various solutions
and determining the course of the reaction by indirect analysis of the
precipitates and by electrometric titration. The procedure consisted
in mixing definite volumes of Fe+ + + and Fe(CN) 6 or
Fe++ and Fe(CN) 6 of known composition and analyzing the
supernatant solution after precipitation to determine the ratio of com-
plex bound iron to iron not present in a complex, and the ratio of
Fe++ to Fe + + +. From such data the composition of the several
precipitates was calculated. On adding K 4 Fe(CN) 6 to FeCl 3 in
gradually increasing proportions the following compounds appear
to form in order: Fe 4 + + + [Fe(CN) 6 ] 3 , KFe++ + [Fe-
(CN) 6 ] , and K 2 Fe+ + [Fe(CN) 6 ] . When the ratio
of FeG 3 to K 4 Fe(CN) 6 is 4 or more to 3, the product is the normal
salt Fe 4 [Fe(CN) 6 ] 3 , but in other proportions the product is a mix-
ture of this compound with the alleged double salts. Similar observa-
tions on adding increasing amounts of K 3 Fe(CN) 6 to FeQ 2 solution
indicate the formation of the following substances in order: KFe 2 ++ -
Fe 6 + + + [Fe(CN) 6 ] 5 , KFe++Fe 3 + + + [Fe(CN) 6 ] 3 ,
KFe+ + + [Fe(CN) ? ] , and Fe 4 + + + [Fe(CN) 6 ] 3 . In-
termediate compositions are assumed to be mixtures.
When the ratio of complex cyanide to iron in solution is 1 : 1 the
product can be represented by the formula KFe[Fe(CN)e] which is
formed in accord with tfie equations:
K4Fe(CN) fl + FeCl 3 -> KFe[Fe(CN) 6 ] + 3KC1
K3Fe(CN) 6 + FeCl 2 -> KFe[Fe(CN) 6 ] + 2KC1
The alleged double salt KFe[Fe(CN) 6 ] is peptized by water and is
called soluble blue. Since the product has essentially the same compo-
sition whether one starts with a ferric salt and ferrocyanide, or a
ferrous salt and ferricyanide, there would appear to be no distinction
between the so-called "soluble" Prussian blue and "soluble" Turnbull's
blue. Reihlen and Zimmermann 8 considered the blue products to be
'Mullet and Stanisch: J. prakt. Chem., (2) 79, 81; 80, 153 (1909); 84, 353
(1911) ; Muller: 90, 119 (1914) ; Miiller and Lauterbach: 104, 241 (1922).
a Ann., 461, 75 (1926); cf. Justin-Mueller: Bull. soc. chim, (4) 49, 1285
(1931).
FORMATION AND COMPOSITION
345
polynuclear structures rather than salts of hydroferrocyanic acid, but
this is denied by Davidson and Welo. 4
Referring to the above equations, it will be seen that analogous
solutions would result if 1 mol KCl is added to the second. Miiller
and Lauterbach titrated electrometrically : (a) FeQ 3 with K 4 Fe-
(CN) 6 ; and (&) the reverse of (a) ; (c) FeCl 2 with K 3 Fe(CN) 6 in
Ratio Fe(CN) 6 Fe Mixed (Curves 1 and 2)
05
K 4 Fe(CN) 6 withFeCI 3 +KCI
Jw-
50
2 1
Ratio Fe (CN) 6 - Fe Mixed (Curves 3 and 4)
0.67
FIG. 52. Electrometric titration curves for iron salts and complex iron cyanides.
the presence of 1 mol KCl; and (d) the reverse of (c). The four
titration curves shown in Fig. 52 all cut at a single point which indi-
cates (1) that the precipitates corresponding to this point have the
same composition for all reactions and (2) that the precipitates formed
with other mixtures vary more or less in composition.
Although the work of Hofmann and of Miiller has gone a long
* J. Phys. Chem., 82, 1191 (1928).
346 PRUSSIAN BLUE AND RELATED PRODUCTS
way toward clearing up the chemistry of the complex iron-cyanogen
compounds, there is still considerable doubt as to whether such alleged
compounds as KFe 2 + + Fe 6 +++[Fe(CN) 6 ] 5 actually exist.
One cannot emphasize too strongly that the analysis of a gel, either
by a direct or an indirect method, does not give sufficient data to
establish the gel as a definite compound. 6 Electrometric titrations,
likewise, cannot give conclusive results when a gel with a strong ad-
sorption capacity is precipitated during the process. Miiller believes
that the mixtures of varying composition are solid solutions of normal
salts and the alleged double salts. It is useless to speculate on this
point until we know whether the alleged double salts are definite com-
pounds or adsorption complexes of Fe4[Fe(CN)g]3 and alkali ferro-
cyanide. It is altogether probable that these questions could be settled
by systematic x-ray and electron diffraction studies. In this connec-
tion Levi e claims that the x-radiograms of the so-called Prussian blue
and Turnbull's blue are the same. A definite crystalline compound
having the formula KFe+ + + [Fe(CN) 6 ] H 2 O is obtained
by diluting a hydrochloric acid solution of Prussian blue. 7 It is unlike
the gel in that it is not peptized either by water or oxalic acid and has
a violet or purple color from which it derives its name, Williamson's
violet. 8 X-ray analysis might disclose whether the larger crystals of
Williamson's violet are the same in composition and structure as the
much smaller crystals of Prussian blue.
Tarugi 9 claims from analysis of numerous blues that a percentage
of oxygen is present in all of them that cannot be attributed to ana-
lytical errors. The result was confirmed in various ways with highly
accurate analytical procedures and with reciprocal control. This is
not at all surprising when the blue is made by the interaction of ferro-
cyanide and ferric chloride, since the iron salt is partly hydrolyzed
giving colloidal hydrous ferric oxide which will be carried down by
the gel. Indeed Vorlander 10 showed that this hydrolysis of ferric
salts causes the reaction with ferrocyanide to proceed at a rate suffi-
ciently slow to measure. In this connection, mention should be made
C/ Bhattacharya and Dhar: Z. anorg. Chem., 213, 240 (1933).
Giorn. chim. ind. applicata, 7, 410 (1925).
'Hofmann, Heine, and Hdchtlen: Ann., 887, 1 (1904).
a Williamson: Ann., 67, 225 (1846); Hofmann, Heine, and Hochtlen- 887,
1 (1904).
Gazz. chim. ital, 55, 951 (1925); cf. Baudish and Bass: Ber., 65B, 2698
(1922).
10 Ber., 48, 181 (1913); Kolloid-Z., 22, 103 (1918).
FORMATION AND COMPOSITION 347
of Reitstiotter's 11 observation that the addition of a small amount of
ferric salt to the test tube containing the precipitate thrown down from
an alumina sol by the required amount of ferrocyanide does not give
Prussian blue until after an appreciable interval of time. This is ex-
plained by the very strong adsorption of ferrocyanide by alumina
which removes the former completely from the field of action. If
another strongly adsorbed anion is added to the alumina either before
or after coagulation, the ferrocyanide is displaced in part and the time
required for the appearance of Prussian blue is diminished appreciably.
"Soluble" and "Insoluble" Blue Products. With K 4 Fe(CN) 6 in
excess and Fe+ + +, or with K 3 Fe(CN) 6 in excess and Fe++, a gel
results which is readily peptized by water and is called soluble Prus-
sian blue. According to Mtiller and Lauterbach it is chiefly KFe +++ -
[Fe+ + (CN) 6 ] with adsorbed water. With K 4 Fe(CN) 6 and Fe+ + +
in excess a blue gel is obtained having the composition Fe 4 + + + [Fe++-
(CN)els which is not easily peptized by water and is therefore
called insoluble Prussian blue. A similar product prepared with
K 3 Fe(CN) 6 and Fe++ is said to have the formula Fe++[Fe+++-
(CN) 6 ] 2 or KFe++[Fe+ + + (CN) 6 ] 3 and is called in-
soluble Turnbull's blue. Insoluble Prussian blue may be distinguished
from the corresponding Turnbull's blue with ammonium fluoride which
decolorizes the latter but not the former. 12 As we shall see (p. 349),
the terms "soluble" and "insoluble" as applied to the blue products
are misnomers.
Precipitates in Gels. An experiment in diffusion which involves
the formation of Prussian blue is Alexander's so-called "patriotic tube"
experiment. A test tube is filled two-thirds full of a slightly alkaline
agar sol containing enough phenolphthalein to turn it pink, and a little
potassium ferrocyanide. After the agar has set to a solid, a dilute
solution of ferric chloride is carefully poured on top. The ferric ion
forms with the ferrocyanide a slowly advancing band of blue before
which the more rapidly diffusing hydrochloric acid spreads a white
band as it discharges the pink color of the indicator. In a few days
the tube is about equally banded in red, white, and blue. 18 Creighton 14
obtained rhythmic bands of Prussian blue by passing an electric cur-
" Kolloid-Z., 21, 197 (1917); Freundlich and Reitstotter: 28, 23 (1918);
Vorlander: 22, 103 (1918).
"Szebellldy: Z. anal. Chem,, 75, 165 (1928).
C/. Holmes: "Laboratory Manual of Colloid Chemistry," 11 (1925).
" J. Am. Chem. Soc, 86, 2357 (1914); cf. Peskov and Saprometov: Kol-
loid-Z., 69, 181 (1934).
348 PRUSSIAN BLUE AND RELATED PRODUCTS
rent through a dilute solution of potassium ferrocyanide and sodium
chloride in an agar jelly contained in a tube 70 cm long and 2 cm in
diameter.
Applications
Commercial Prussian blue is a very important blue pigment. What
is said to be the finest variety, known as Paris blue, is obtained by
mixing potassium ferrocyanide and ferrous sulfate, followed by oxida-
tion of the white ferrous ferrocyanide by chlorine. 15 The use of potas-
sium salts is said to be important, the color being less satisfactory
when sodium salts are employed. Eibner and Gerstacker 18 claim that
blues with a high potassium content possess the technically desirable
greenish shade whereas those poor in potassium are dull with a violet
tinge. It is not obvious why sodium salts should yield a pigment in-
ferior in color to potassium salts. Since the variation in color is
determined largely by the physical character of the precipitate, there is
no apparent reason why a proper control of the conditions of precipi-
tation should not yield a satisfactory product with sodium salts. Wil-
liams 17 claims that potassium salts may be replaced by ammonium salts
with satisfactory results.
In commercial pigments, Prussian blue is diluted with starch, heavy
spar, gypsum, zinc white, or burned and ground kaolin. The white
constituent is first pulverized and then added to a paste of Prussian
blue. The relative degree of subdivision of the white and colored con-
stituents is quite important. If a very fine powder is shaken with a
moderately coarse one, the former tends to coat the latter instead of
filling the voids between the coarser material. In one experiment,
Briggs 18 mixed 0.032 g of Prussian blue with 10 g of dolomite which
passed a 40-mesh sieve and did not pass a 100-mesh sieve; and in a
second experiment, the same amount of Prussian blue was mixed with
10 g of dolomite all of which would pass a 200-mesh sieve. In the
first experiment, the resulting mixture was a deep blue and in the
second it was practically white although the percentage compositions
were identical.
C/. Thorpe: "Dictionary of Applied Chemistry," 2, 448 (1921)
"J. Soc. Chem. Ind., 81, 1041 (1912).
17 "The Chemistry, Manufacture, and Estimation of Cyanogen Compounds,"
120 (1915).
" J. Phys Chem., 22, 216 (1918); cf. Fink: 21, 32 (1917).
FORMATION 349
PRUSSIAN BLUE SOLS
Formation
Prussian blue gel precipitated with a small excess of ferrocyanide
is called "soluble" Prussian blue because it is so readily peptized to a
clear sol by washing. The gel formed in the presence of a slight excess
of iron is not peptized by washing and so is designated "insoluble"
Prussian blue. As already pointed out, these terms are misnomers.
The "soluble" blue is peptized and not dissolved in water, and the
freshly formed "insoluble" blue is carried into the sol state by thorough
washing with the aid of the centrifuge, especially if a trace of potas-
sium ferrocyanide is added to the wash water. 19 A sol is formed also
by peptization of the gel with oxalic acid, 20 followed by dialysis. Un-
like oxalic acid, the neutral oxalates of potassium, sodium, and am-
monium dissolve the Prussian blue gel, giving a soluble potassium iron
oxalate with a green color. 21 Hence the test for ferric iron by the
Prussian blue test loses its reliability in the presence of neutral ox-
alates. The oxalic acid sol was once employed as an ink but it has
now been replaced by the blue aniline colors.
Clear blue sols result directly on mixing potassium ferrocyanide
and ferric chloride, provided that the ferrocyanide is present in suitable
excess and that the concentration is not too great. Zsigmondy 22 ti-
trated 30 cc of 0.1 M K 4 Fe(CN) 6 with 0.1 M FeCl 3 : with 26 cc of
the latter solution the sol was clear blue; with 27 cc, partial precipi-
tation was observed; and with 29 cc, precipitation was complete. A
stable sol resulted on mixing 0.2% K 4 Fe(CN) 6 with \% less than the
amount of Q.5% FeCl 3 required for precipitation, followed by dialysis.
The addition of an excess of potassium ferrocyanide to a ferric
chloride solution gives a blue-green to light-green sol ie which is not a
definite green double salt or other compound. Instead, it is a mixture
of very finely divided Prussian blue with a small amount of adsorbed
colloidal hydrous ferric oxide ; a similar green sol results on mixing a
dilute sol of ferric oxide with Prussian blue. The addition of alcohol,
hydrochloric acid, or neutral salts, including an excess of potassium
19 Bachmann: Z. anorg Chem., 100, 77 (1917).
20 Stephen and Nash. Ann. Pharm, 84, 348 (1840); Karmarsch: J. prakt.
Chem, 20, 175 (1840); Graham: Phil. Trans., 151, 183 (1861); Reindel: J.
prakt Chem., 102, 38, 255 (1867).
"Kohn: Monatsh., 43, 373 (1923); Kohn and Benczer: 44, 97 (1923); cf. t
however, Buzagh: Kolloid-Z., 44, 156 (1928).
""Lehrbuch Kolloidchemie," Leipzig, 5th ed., 2, 160 (1927).
350 PRUSSIAN BLUE AND RELATED PRODUCTS
ferrocyanide, to the green sol throws down the usual deep blue pre-
cipitate.
In this connection attention may be called to the brown sol of
ferric ferricyanide which is formed by the interaction of aqueous solu-
tions of ferric chloride and potassium ferricyanide. 28 Ultramicroscopic
examination shows that the particles increase in size gradually and
finally deposit a green substance which is probably a mixture of ferric
ferricyanide and Prussian blue. A similar green substance is formed
by adding ferric chloride to potassium ferricyanide solution contain-
ing a little potassium ferrocyanide. Williams 24 believes the gel to
be a double salt of ferric ferricyanide and potassium ferric ferroty-
anide which he formulates Fe 2 9 + ++ K 3 [Fe + +Fe2 + + + (CN) 18 ]9-
210H 2 O. Like so many alleged complex compounds, the gel in ques-
tion is doubtless nothing more than a mixture which can have any
composition you like, depending on how it is made.
General Properties
Prussian blue sols are composed of non-spherical particles 25 which
are negatively charged as a rule owing to preferential adsorption of
ferrocyanide ion. Bachmann showed, however, that the particles are
positively charged in the presence of a suitable small excess of the
strongly adsorbed ferric ion. The behavior is thus analogous to that
of the silver halides which assume a negative charge in the presence of
excess halide ion and a positive charge in the presence of excess silver
ion. The positive Prussian blue sol precipitates slowly on standing but
is completely repeptized by shaking. The stability of the sol is in-
creased by the presence of small amounts of gelatin. 26
The size of the colloidal aggregates in a sol falls off with increasing
concentration of the peptizing ferricyanide ion. Bechhold " found
that a Prussian blue sol is held back by an ultrafilter which allows a
Bredig platinum sol to go through. The addition of the protecting col-
loid sodium "lysalbinate" to Prussian blue sol peptizes the latter suffi-
ciently to allow it to pass through a filter that stops it before. Nistler 28
estimated the particle radius of a commercial, soluble Prussian blue to
be 114 X lO- 8 cm.
"Haller: Kolloid-Z. f 20, 76 (1917).
24 "The Chemistry, Manufacture, and Estimation of Cyanogen Compounds/'
142 (1915).
"Szegvari: Z. physik. Chem., 112, 277 (1924).
*Lobry de Bruyn: Rec. trav. chim., 19, 236 (1900).
Z. physik. Chem., 60, 257 (1907).
"Kolloid-Beihefte, 31, 1 (1930).
COAGULATION BY ELECTROLYTES 351
The sol is very stable at room temperature but is coagulated by
boiling and by visible and ultraviolet light and 0-rays but not by a-, y-,
or x-rays. 29 The flocculation time is proportional to the concentration
and to the thickness of the layer of sol exposed. The optimum wave
length for most rapid coagulation is in the neighborhood of 420 m/*.
Immediate and complete coagulation takes place on subjecting the sol
to a pressure of 2000 atmospheres. 30
Gatterer 81 has determined the density, viscosity, electrical conduc-
tivity, freezing point, and solvent action for carbon dioxide and acety-
lene, of sols formed by peptization with varying concentrations of so-
dium ferrocyanide.
Prussian blue sol is hydrolyzed by boiling, giving a negatively
charged sol of hydrous ferric oxide. 82
Coagulation by Electrolytes
Action of Salts. Prussian blue sols are precipitated readily by
electrolytes, especially those with multivalent cations. Pappada 88
found the order of precipitating power to be : Fe, Al, Cr > Ba, Cd >
Sr, Ca>H>Cs>Rb>K>Na>Li. Lachs and Lachman 84
investigated the precipitating power of salts with cations of the same
valence. The results are given in Table XLIV. On the basis of these
and similar observations on colloidal antimony pentoxide, it was con-
cluded that the adsorption of an ion which determines its precipitating
power is closely related to its degree of hydration. According to
Fajans, 85 the hydration of an ion may be regarded as the formation
around the ion of a polarized water envelope consisting of dipoles, the
process being accompanied by a positive heat effect. From this point
of view, one would expect the adsorption of the ions to be accom-
panied by partial dehydration, the extent of which will be determined
by the heat of hydration of the ions. Since both the heat of hydration
and the amount of hydration decrease in the series from lithium to
"Lederer and Hartleb: Kolloid-Z, 62, 42 (1933); Schoras: Bcr. f 3, 11
(1870); Ghosh and Dhar: J. Phys. Chem, 80, 1564 (1926); cf , however, Roy
and Dhar: 84, 122 (1930).
w Wilson and Poulter: Proc Iowa Acad. Sci, 86, 295 (1929).
81 J. Chcm. Soc., 299 (1926); cf. t also, Chakravarti and Dhar: Kolloid-Z.,
42, 124 (1927).
s* Hazel and Sorum: J. Am. Chem. Soc., 52, 1337 (1930).
38 Kolloid-Z. f 6, 83 (1911); cf. Yajnik and Bhatia: J. chim. phys., 22, 589
(1925).
8 *Z. physik. Chem., 128, 303 (1926).
85 Fajans and Beckerath: Z. physik. Chem., 07, 478 (1921).
352
PRUSSIAN BLUE AND RELATED PRODUCTS
cesium and from magnesium to barium, it follows that the adsorptive
power and coagulative action should increase in this direction, as the
results show (p. 218).
TABLE XLIV
COAGULATION OF COLLOIDAL PRUSSIAN BLUE SOL BY ELECTROLYTES
Soil
Sol 2
Electrolyte
Precipitation
value
Electrolyte
Precipitation
value
millimols/1
millimols/1
KNO,... . .
40
KNO
80
LiNO,.
130
Mg(NO,) 2
5 7
CsNO,
7
Ca(NO,) a
4 5
Li 2 S0 4 /2
1000
Sr(NO,) 2
2 9
K 2 S0 4 /2
72
Ba(NO,) 2
1 1
Cs 2 S0 4 /2
7
In order to get a more exact insight into the relationship between
adsorbabihty and heat of hydration, the heat of adsorption in solution,
U can be regarded as the small difference between the heat of adsorp-
tion in vacuum, U v , and the heat of hydration, W, that is, U a =
U v W. According to Fajans 38 the heat of adsorption in vacuum is
inversely proportional to the sum of the radius of the charged ions,
r l9 and of the adsorbed discharged ions, r 2 ; that is,
n + f*
Born 87 points out that the heat of hydration includes the dielectric
constant of the solution and as a first approximation is inversely pro-
portional to the ratio of the adsorbed ions r 2 ; that is,
Now if KI, r lt and K 2 are assumed to be constant, then
rr ^ ^ 2
U s - -
s Verhandl. deut. physik. Ges., 21, 549, 709, 714 (1919) ; Naturwissenschaften,
9, 733 (1921).
3*Z. Physik, 1, 47 (1920).
COAGULATION BY ELECTROLYTES
On differentiation this becomes,
353
, + 3-
r 2 ) 2 - Kr 2 2
The last value comes out to be positive in the case under consideration.
This means that, although the two values /^ 1 /(r 1 + r 2 ) and / 2 /r 2
decrease with increasing radius of the adsorbed ions r 2l the difference
with increasing r 2 is always positive. Accordingly, the heat of adsorp-
tion must increase in the ion series from lithium to cesium and from
magnesium to barium. Some data on the heat of hydration of alkali
cations and that of adsorption of the corresponding nitrates by charcoal
are given in Table XLV. It will be seen that the heat of hydration
TABLE XLV
HEAT OF HYDRATION OF ALKALI IONS AND HEAT OF ADSORPTION OF ALKALI NITRATE
ON CARBON
Heat of hydration of alkali ions
Molecular heat of adsorption of salts
by carbon
Ion
Heat of hydration,
calories per gram ion
Salt
Heat of adsorption per mol,
relative values
Li
Na
K
Rb
Cs
120
92
72
68
62
LiNO,
KNO,
CsNO,
8 90
10 40
11 72
falls as one goes down the series, whereas the heat of adsorption of
LiNO 3 , NaNO 8f and KNO 3 is in the ratio 8.9: 10.4: 11.7. Since the
heat of adsorption is a measure of the adsorbability, the increase in
precipitating power from lithium to cesium is accounted for.
Although these considerations indicate the existence of a qualitative
relationship between adsorption and heat of hydration in the cases
under consideration, it should be emphasized that this is but one of a
number of factors which determine adsorbability. Indeed, it is prob-
able that heat of hydration of ions plays but a minor role in many in-
stances. We have seen that the solubility of the salt which an ion may
form with an adsorbent appears to be the important factor in the ad-
sorption of certain ions (pp. 37, 103).
3S4 PRUSSIAN BLUE AND RELATED PRODUCTS
In this connection, Mukherjee ** attributes the variation in adsorb-
ability in a series of ions such as the alkalis to differences in their
mobilities. The ions adsorbed on the particles exert an attraction on
the ions of opposite sign in the solution. The latter will be held firmly
so that they cannot migrate provided that their kinetic energy remains
below a value U, which is necessary to separate the ion from the oppo-
sitely charged point of the surface. The value U, as heretofore, is
the work of adsorption or its equivalent, the heat of adsorption. Ac-
cording to Mukherjee, it can be calculated from the expression
in which HI is the valence of the adsorbed ion, w 2 the valence of the
ion of opposite charge in the solution, e the charge on the electron, D
the dielectric constant of water, and x the distance between the centers
of the ions at the position of minimum distance. The value of x is
therefore equal to the sum of the radii of the ions or (ri + fjj)'
Born * 9 showed, however, that the radius of the alkali metals increases
in the series from lithium to cesium ; hence, the value of U a calculated
from the above equation must decrease from lithium to cesium, which
is contrary to the observations of Lachs and Lachman. Moreover, the
view of Mukherjee is unsound theoretically, since adsorption repre-
sents an equilibrium condition which would not be determined solely
by the mobility of the ions. 40
The precipitation values of electrolytes decreases with dilution of
sol irrespective of the valence of the precipitating ion. 41 When the
activity coefficients of the cations at their respective precipitation
values, /+, are plotted against the sol concentration, the curves tend
to converge at higher concentrations (10 g/1). 41 (Cf. pp. 194, 318.)
Dilution with alcohol likewise sensitizes the sol toward electrolytes. 42
Action of Acids. Weir * 8 found the />H of the supernatant liquid
at the coagulation point of hydrochloric, sulfuric, and citric acids to be
approximately constant at a value of 1.9; for acetic acid, it was 2.7;
and for oxalic acid which exerts a strong peptizing action in dilute
ss Phil. Mag., (6) 44, 321 (1922).
w Z. Physik, 1, 221 (1920).
C/. Zsigmondy: "Kolloidchemie," 5th ed., 1, 200 (1925).
"Weiser and Nicholas: J. Phys. Chem., 26, 742 (1921); Wolshin: J. Russ.
Phys.-Chem. Soc., 42, 863 (1910); Lederer: Kolloid-Z., 76, 54 (1936).
"Kawamura: Hokkaido J. Med. Japan, 7, 287 (1929).
J. Chem. Soc., 127, 2245 (1925).
MUTUAL ACTION OF SOLS 355
solution, it was 0.9 to 0.8. Kugel, 44 studied the effect of varying con-
centrations of hydrochloric and sulfuric acid on the rate of agglomera-
tion of the sol. The ^H value was determined conductometrically just
after the acid was added and at the completion of the coagulation; and
the adsorption of hydrogen ion was calculated. The curves showing
the effect of />H value on the rate of coagulation for the two acids are
similar but they do not coincide. The adsorption curves are quite
similar from />H = 2 to 5, but the sulfuric acid curve is slightly lower
than the hydrochloric acid curve. This is probably experimental error
since sulfate is in general more strongly adsorbed than chloride and
hence, as a rule, sulfuric acid is more strongly adsorbed than hydro-
chloric acid.
Action of Mixtures of Electrolytes. Investigations of the precipi-
tating action of salt pairs on Prussian blue sol led to results and con-
clusions similar to those already described for copper ferrocyanide sol
(p. 323). Rabinerson 45 showed the existence of an antagonistic action
with the pairs KCl-BaCl 2 , NaCl-BaCl 2 and BaCl 2 -K 3 Fe(CN) 6
which was attributed to the stabilizing action of the anions. Ghosh
and Dhar 46 likewise found a strong antagonistic action with the salt
pair KCl-BaCl 2 ; on the other hand, with the pairs HC1-KC1 and
HNO 3 -KNO 3 , the acids exert such a marked sensitizing action that
the precipitation values of mixtures falls well below the calculated
additive values.
Mutual Action of Sols
If one sol is treated with a second the first will be stabilized, sensi-
tized, or completely or partially precipitated, depending on the nature
and relative concentrations of the two sols. Ghosh and Dhar 47 report
that gelatin and tannic acid sensitize Prussian blue in low concentra-
tions but stabilize it in higher concentrations. Since gelatin in acid or
neutral solutions is a positive sol, small amounts would be expected to
exert a sensitizing action on negative Prussian blue sol. Tannic acid,
on the other hand, is a negatively charged sol and, in the absence of
impurities, should have a stabilizing effect at all concentrations.
If a small amount of a positively charged sol is added to a large
amount of a negatively charged sol, the former may be stabilized by
adsorption on the latter, which, in turn, will be sensitized to some ex-
"Kolloid-Z., 51, 240 (1930).
Kolloid-Z., 42, SO (1927).
J. Phys. Chem., 29, 659 (1925) ; 80, 842 (1926).
"Kolloid-Z., 44, 218 (1928).
356
PRUSSIAN BLUE AND RELATED PRODUCTS
tent. Complete mutual coagulation takes place when the potential of
each sol is lowered below the point of stability. It is claimed that this
happens when electrochemical equivalent amounts of each sol are pres-
ent; 48 but other factors must be taken into account (p. 215). The
addition of electrolytes to the sol will widen the range of mutual coagu-
lation as illustrated by Rabinerson 49 with positive ferric oxide sol and
negative Prussian blue sol. Some typical results are shown diagram-
matically in Fig. 53, in which the percentage composition of the mix-
100
100
o-
100
100
1
B
IV. 35
ONaCI
V,
V.50
DNaCI
""-%
VI. IOC
ONaCI
OOFe 2 3 75 50 25 Q
OPrussian25 50 75 10
Blue
)Fe 2 3 75
Prussian 25
Blue
FIG. 53 Mutual action of ferric oxide sol and Prussian blue sol in the absence
and in the presence of sodium chloride. (Sodium chloride concentration in
millimols per liter )
tures is plotted as abscissa against the percentage amount of coagula-
tion as ordinate, without the addition of sodium chloride and in the
presence of five different concentrations of the salt. The several un-
broken curves show the effect of sodium chloride in the amounts indi-
cated, on the mutual action of the sols. The dotted curves show the
extent of coagulation by the sodium chloride of pure ferric oxide sols
of the different concentrations. The ferric oxide sol is the more sensi-
tive to the action of electrolytes, sodium chloride causing clouding of
"Cf Lottermoser and May: Kolloid-Z. f 68, 61 (1932).
Kolloid-Z., 89, 112 (1926).
MUTUAL ACTION OF SOLS 357
the 0.1% sol at a concentration of 200 millimols/1 and complete coagu-
lation at 500, whereas a 0.05% Prussian blue sol requires 400 milli-
mols/1 for clouding and 1000 for complete coagulation.
In the absence of sodium chloride, complete mutual precipitation
takes place sharply with approximately 33% Fe 2 O 3 and 67% Prussian
blue. Both to the right and left of this point the amount of precipita-
tion falls off, decreasing to zero with about 90% Fe 2 O3 an d 9% Prus-
sian blue on the left, and with 28.5% Fe 2 O 3 and 71% Prussian blue
on the right. The presence of sodium chloride in the mixture of sols
increases the range of complete precipitation both to the left and to
the right. This effect is more marked with excess ferric oxide than
with excess Prussian blue since the former sol is more sensitive to the
action of sodium chloride. When the concentration of salt reaches
500 millimols/1, the left portion of the curve becomes parallel with the
abscissa whereas the right portion takes this position only after the
concentration has been raised to 1000 millimols.
The observations show quite clearly the relationships among sensi-
tization, mutual coagulation, and protective action. It is obvious that
within a certain range each sol sensitizes the other since an amount of
sodium chloride which will not coagulate either sol alone will effect
complete coagulation of a mixture that appears stable. For example,
sols containing 95.24% Fe 2 O 3 + 4.76% Prussian blue and 90.91%
Fe 2 O 3 4- 9.09% Prussian blue do not precipitate in the absence of
sodium chloride but 100 millimols/1 will precipitate both mixtures,
although this amount of salt does not cloud either sol taken separately.
The protecting action of Prussian blue is also illustrated by the curves.
For example, the dotted curves in IV disclose that the 0.1% Fe 2 O 3
sol is clouded by 250 millimols/1 of sodium chloride whereas the 0.2%
sol is completely precipitated. On the other hand, the 0.02% sol is
not even rendered cloudy by this amount of salt when Prussian blue
is present in the ratio of 71.4% to 28.6% of Fe 2 O 3 . In 1 general, in
the region of electrolyte concentrations which lies between the coagula-
tion value of the two sols of opposite sign, the more sensitive sol is
protected by an excess of the more stable sol.
The gelatinous precipitates of the positively charged hydrous oxides
or hydroxides of iron, aluminum, magnesium, thorium, and the rare
earths 50 as well as barium sulfate 81 and plastic clay M adsorb Prussian
soWedekind and Fischer: Ber, 60B, 541, 544 (1927).
"Wohlers: Z. anorg. Chem., 59, 203 (1908).
"Rohland: Z. anorg. Chem., 77, 116 (1912).
358 PRUSSIAN BLUE AND RELATED PRODUCTS
blue when shaken with the sol, but the negative hydrous oxides of ti-
tanium, tungsten, and tantalum do not adsorb the color.
Hydrous ferric oxide treated with hydrocyanic acid gives a black
gel which is an irreversible adsorption complex of Prussian blue and
ferric oxide. 88 It is believed that the hydrocyanic acid first reduces
some of the ferric oxide to ferrous oxide which dissolves in the excess
acid to give hydro ferrocyanic acid which then reacts with hydrous fer-
ric oxide giving Prussian blue that is adsorbed by the remaining oxide.
A black gel containing as much as 10.5% Prussian blue was prepared
in this way.
eaWedekind and Fischer: Ber., 60B, 541 (1927); Wedekmd: Naturwisscn-
schaften, 16, 163 (1927).
PARTY
THE COLLOIDAL SILICATES
CHAPTER XX
SILICATE SOLS AND GELS; THIXOTROPY
The most familiar examples of silicates in the colloidal state are
colloidal clays and claylike bodies such as bentonite. The laboratory
preparation of silicate gels and sols may be accomplished in some in-
stances by double decomposition of sodium silicate and salts of the
heavier metals. The nature of the product obtained by this process
will depend not only on the heavy metal entering into the reaction but
also, to a considerable extent, on the nature of the sodium silicate solu-
tions. Accordingly some attention must be given to the constitution
and properties of such solutions. 1
SODIUM SILICATE SOLUTIONS
When reference is made to solutions of sodium silicate one usually
thinks of water glass which is not a solution of a definite chemical
individual but is a variable system of the components sodium oxide,
silica, and water. In commercial silicate solutions the ratio Na 2 O:
SiO 2 may vary from 1:1 to 1:4; sodium metasilicate in which the
ratio is 1:1 is on the market as a crystalline powder. From earlier
measurements of conductivity 2 and freezing point, 8 it was concluded
that dilute solutions of sodium silicate are practically completely hy-
drolyzed, yielding sodium and hydroxyl ions and colloidal hydrous
silica, but no silicate ions. More recent work indicates, however, that
dilute silicate solutions are hydrolyzed much less than formerly sup-
posed and that the percentage hydrolysis is quite low in concentrated
solutions. Moreover, the solutions appear to contain silicate ions
which, rather than hydroxyl ions, appear to be adsorbed by colloidal
silica giving the latter a negative charge. Some of the experimental
evidence in support of these conclusions will now be given.
especially Vail: "Soluble Silicates in Industry," Am. Chem. Soc.
Monograph, No 46 (1928).
*Kohlrausch: Z. physik Chem., 12, 773 (1893).
'Loornis: Wied. Ann., 60, 523 (1897).
361
362
SILICATE SOLS AND GELS; THIXOTROPY
Conductivity
Harman 4 prepared pure crystalline sodium metasilicate, Na 2 SiO 3 -
9HgO, and from a solution of this salt he made solutions of varying
ratios of Na2O : SiC>2 by removing the alkali electrolytically. 5 Solu-
tions in which the ratio Na 2 O : SiO 2 was 2 : 1 and 1 : 1 were found
to be excellent conductors, 6 and solutions in which the ratio was 1 : 2,
1 : 3, and 1 : 4 were good conductors in dilute solution but abnormally
low in concentrated solution. This is illustrated graphically in Fig.
54A, in which the equivalent conductivity of solutions of various con-
1.1 1.2 1-3
Ratio Na 2 : S!0 2
1.1 12 13
Ratio Na 2 : St0 2
FIG. 54. Curves for (A) equivalent conductivity and (B) percentage hydrolysis
against the NaO : SiOs ratio for solutions of various concentrations.
centrations is plotted against the Na 2 O:SiO 2 ratio. Considering
dilute solutions, it will be seen that the conductivity of a caustic soda
solution to which silica has been added decreases linearly and rapidly
until the ratio 1 : 2 is reached, where there is a sharp change in direc-
tion, the conductivity again falling regularly and linearly but not so
rapidly as the less siliceous solutions. The sharp change in direction
*J. Phys. Chem., 29, 1155 (1925).
Spencer and Proud: Kolloid-Z., 31, 36 (1922); Lottermoser: 80, 346;
Kroger: 16 (1922).
C/., also, Kohlrausch: Wied. Ann., 47, 756 (1892); Z. physik. Chem., 12,
773 (1893); Kahlenberg and Lincoln: J. Phys. Chem., 2, 77 (1898).
HYDROLYSIS 363
at the ratio 1 : 2 is taken to indicate the existence of a definite salt in
solution. Since Na 2 SiO 3 9H 2 O is a well-defined crystalline salt, the
absence of a break in the curve at 1 : 1 indicates that the salt is appre-
ciably hydrolyzed in dilute solution and that the linear decrease in con-
ductivity is caused by the gradual disappearance of the mobile hydroxyl
ion. The conductivity for ratios up to 1 : 2 is much greater than could
result alone from hydrolysis into sodium hydroxide and colloidal hy-
drous silica. Above the ratio 1:2 where the hydroxyl ion concen-
tration is very low, the sodium ion accounts for only about one-half
of the observed conductivity. To account for the remainder, the ob-
vious way is to postulate the existence of silicate ions 7 with mobilities
ranging from 40 to 60. It may be mentioned at this point that the
equivalent conductivity calculated from freezing-point, hydroxyl ion,
and sodium ion measurements agrees well with conductivity data deter-
mined experimentally.
The curves for more concentrated solutions exhibit three changes
of direction which might be taken to indicate the presence in solutions
of a salt with the Na 2 O : SiO 2 ratio 2 : 1 as well as 1 : 1 and 1 : 2. But
since there is no independent evidence of the existence of salts with
ratios other than 1 : 1 and 1 : 2, it is probable that other ratios are mix-
tures of these with sodium hydroxide or hydrous silica, as the case
may be. To account for the relatively low conductivity of concentrated
solutions above the ratio 1 : 2, Harman suggests that either the extent
of ionization is too low, or aggregate or colloid formation takes place.
There is no doubt of the existence of colloidal particles in the more
siliceous solutions. 8
Hydrolysis
The hydrolysis of silicate solutions of varying concentrations and
of varying Na 2 O:SiO 2 ratios was determined electrometrically by
Bogue, 9 whose observations were confirmed in all essential respects by
Harman. 10 Harman's results are shown graphically in Fig. 545, in
which the percentage hydrolysis calculated from the hydroxyl ion con-
centration as given by hydrogen electrode measurements is plotted
against the Na 2 O : SiO 2 ratio. It will be seen that these data show
the hydrolysis to be far from complete. Thus a 0.01 N solution of
Na 2 SiO 3 is hydrolyzed to the extent of 27 &% whereas ratios of 1 : 3
TC/., also, Main: J. Phys. Chem., 80, 535 (1926).
Stericker: Chem. & Met. Eng., 26, 61 (1921).
' J. Am. Chem. Soc., 42, 2575 (1920).
"J. Phys. Chem., 80, 1100 (1926).
364 SILICATE SOLS AND GELS; THIXOTROPY
and 1:4 at the same concentration show only 1.5% hydrolysis. More-
over, the degree of hydrolysis falls off with increasing concentration
for all ratios.
It is obvious that the correctness of the conclusions concerning the
percentage hydrolysis depends upon the accuracy of the determination
of the hydroxyl ion concentrations. It is assumed, of course, that the
hydrogen electrode measurements give a true measure of the hydroxyl
ion present in the system. This is probably not the case if a part of
the hydroxyl ions are adsorbed by colloidal particles present in the sys-
tem. It has been demonstrated that adsorbed chloride ion will not
give a test with silver nitrate u and its effect on a chlorine electrode
will be negligible. Similarly, it is extremely improbable that adsorbed
hydroxyl ion would behave toward a hydrogen electrode in the same
way as a free hydroxyl ion. The values of the hydrolytic dissociation
determined from electromotive force measurements on a colloidal sys-
tem, such as certain sodium silicate solutions, will be correct only if no
hydroxyl ion is adsorbed. As will be shown in the next section, it is
not possible to account for the high osmotic activity of dilute sodium
silicate solutions if it is assumed that the silica exists only as colloidal
silica which has adsorbed hydroxyl ion. Accordingly, it may well be
that Bogue's and Harman's determinations of percentage hydrolysis
are approximately correct.
Osmotic Activity
The osmotic activity of various sodium silicate solutions has been
calculated from measurements of the vapor pressure, 12 boiling point,"
and freezing point. 14 In Fig. S5A are plotted Harman's data for the
molecular lowering of the freezing point, A/w, against the Na 2 O :
SiC>2 ratio for solutions of varying concentrations; and in Fig. 55/?,
the corresponding activity coefficients as calculated by Harman. 15 It
will be seen that ratios 2 : 1 and 1 1, and to a lesser degree 1 : 2, ex-
hibit a high degree of osmotic activity especially in dilute solutions.
Moreover, ratios 1 : 3 and 1 : 4 show a relatively high osmotic activity
"Ruer: Z. anorg. Chem , 48, 85 (1905).
12 Harman- J Phys. Chem., 30, 917 (1926); Bennett: 31, 890 (1927).
13 Cann and Cheek. Ind Eng. Chem., 17, 512 (1925); Cann and Gilmore:
J. Phys Chem, 32, 72 (1928).
"Kahlcnberg and Lincoln: J. Phys. Chem., 2, 77 (1898) ; Thompson: Vail's
"Soluble Silicates in Industry/' 42 (1928); Harman: J. Phys. Chem, 31, 359
(1927).
15 Cf., however, Randall and Cann: J. Am Chem, Soc. f 50, 347 (1928).
OSMOTIC ACTIVITY
365
in dilute solutions as compared with the low value in concentrated
solutions.
If the van't Hoff factor i is calculated for the solutions in which
the Na 2 O : SiO 2 ratio is 1 : 1 using the expression i = A/w X 1.8S8, 10
a fairly high degree of hydrolysis or ionization is indicated, even in
the more concentrated solutions. In a 0.01 N solution, the value for i
comes out to be 3.87, which lies between 3, expected if total ionic dis-
11 12 13
Ratio Na 2 Si0 2
11 12 1:3
Ratio Na 2 Si0 2
FIG 55 Curves for (A) molecular lowering of the freezing point and (#)
activity coefficient against the Na 2 O SiOa ratio for solutions of various
concentrations
sociation takes place, and 4, the consequence of total hydrolytic disso-
ciation and ionization of the hydroxide formed, assuming in the latter
event that all the silica is colloidal and without influence on the other
constituent of the solution If the laws of ideal solutions are assumed
to apply in this instance, the percentage hydrolysis is 3 87/4 X 100 =
97%, which agrees well with the figure calculated from conductivity
measurements of both Harman and Kohlrausch 2 and from the freez-
ing-point measurements of Loomis. 3
If the extent of hydrolysis is determined by the concentration of
sodium ions, 17 and if the OH~ ions are assumed to be adsorbed to a
16 The molecular lowering of an ideal substance at infinite dilution.
"Harman: J. Phys Chem., 30, 922 (1926).
366 SILICATE SOLS AND GELS; THIXOTROPY
large extent by colloidal silica so that the OH- ion concentration
agrees with the percentage hydrolysis found by electromotive force
measurements, Harman calculates i for 0.005 M Na 2 SiO 3 to be 2.53
as compared with 3.87 from freezing-point measurements. This non-
agreement, wfcich is clearly outside the bounds of experimental error,
furnishes definite evidence against OH~~ ion adsorption. Harman de-
duced therefore that the 1 : 1 ratio is Na 2 SiO 3 which undergoes both
hydrolytic and ionic dissociation giving Na+, OH, and SiO 3 ions,
and H 2 SiO 3 most of which is not colloidally dispersed.
From similar considerations, Harman concludes that the solutions
in which the Na 2 O : SiO 2 ratio is 1 : 2 do not contain the salt Na 2 Si 2 O 5
but rather NaHSiO 3 which behaves like Na 2 SiO 3 in giving rise to
Na+, OH-, and HSiO 3 - ions together with H 2 SiO 3 . Electro-
metric titration of sodium silicate solutions likewise indicates that
silicic acid is a dibasic acid giving salts NaHSiO 3 and Na 2 SiO 3 . 18 19
The 1 : 3 and 1 : 4 ratios, on the other hand, do not appear to be defi-
nite salts but rather to consist of complex aggregates and ionic mi-
celles which are assumed tentatively to have the composition [mSiO 3 ,
nSiO 2 og] m .
The ultramicroscope definitely discloses the presence of colloidal
particles in the more siliceous mixtures. Since some silicate ions exist
in such solutions, the colloidal particles derive their negative charge
from adsorption of these ions giving what may be termed colloidal ions
or ionic micelles ; but such colloidal particles differ in no essential way
from colloidal silver iodide particles stabilized by preferential adsorp-
tion of I"" ions or of colloidal gold particles stabilized by preferential
adsorption of AuCl 2 ~~ ions.
Transport Numbers of the Ions
The transport numbers of the ions in certain silicate solutions de-
termined by Harman 20 using the Hittorf method are given in Table
XLVI. These results indicate that a fair proportion of the current,
at least one-half in the most siliceous mixtures, is carried by the silica.
Harman recognized two possibilities : either the silica exists as ions, or
hydroxyl ions are adsorbed on colloidal particles giving the necessary
charge to the silica. Since the view that the silica exists chiefly as
colloidal particles that have adsorbed hydroxyl ions is incompatible
"Harman: J. Phys. Chem., 31, 616 (1927).
"Stericker: Vail's "Soluble Silicates in Industry/ 1 54 (1928); Joseph and
Oakley: J. Chem. Soc., 127, 2815 (1925).
20 J. Phys. Chem., 80, 359 (1926).
DIFFUSION.
367
TABLE XLVI
TRANSPORT NUMBERS OF IONS nf SODIUM SILICATE SOLUTIONS
Ratio
#OH =
NaiO+SiO*
#Na
^silicate
l-(tfNa+tfsilicate)
1:1
37
0.16
047
1:2
041
0.41
18
1:3
0.43
0.46
0.11
1 :4
0.48
59
with the high osmotic activity of dilute silicate solutions, it is concluded
that the silica must exist as ions. It does not follow from this, how-
ever, that all the silicate exists as simple SiO 3 or HSiO 3 - ions
free or adsorbed by colloidal silica. It is possible if not altogether
probable that a portion at least of the SiO 3 ions are present as true
colloidal micelles consisting of agglomerates of ions which have a
definite composition. Such colloidal ions or micelles possess a mo-
bility comparable to that of single ions.
Diffusion
The existence of silicate ions and of crystalloidal H 2 SiO 3 in silicate
solutions is further indicated by diffusion experiments of Harman 13
and Ganguly 21 using membranes of collodion and parchment paper.
For example, most of the silica in 0.3 N solutions containing Na 2 O and
SiO 2 in the ratio 1 : 1 and 1 : 2 was diffusible. Moreover, two-thirds
of the silica in 0.3 N and one-third of the silica in 1 N solutions of
the 1 : 4 ratio diffused through the membranes. In very dilute solutions
of all ratios, the silica behaved as if it were in true solution. This
was confirmed by testing for silicate ion colorimetrically, 22 using a test
which depends on the formation of greenish yellow silicomolybdate. 28
Although the work of Harman 24 and others has furthered our
knowledge of the nature of sodium silicate solutions, one is left with
the impression that an important factor in the behavior of such solu-
tions is not taken into account, namely the effect of aging. Mylius and
Groschuff 2e are of the opinion that, at the moment of its formation
"J. Phys. Chem., 31, 407 (1927).
2 Harman: J. Phys. Chem, 31, 622 (1927).
"Dienert and Waldenbulcke : Bull. soc. chim., (4) 33, 1131 (1923).
**Cf. J. Phys. Chem., 32, 44 (1928).
"Ber., 39, 116 (1906).
368 SILICATE SOLS AND GELS; THIXOTROPY
from water glass, silicic acid exists as such in a molecular solution
which passes unchanged through a dialyzing membrane; and that the
colloidal state results from polymerization of the acid with the split-
ting off of water. In commenting on this view some years ago, 28 it
was suggested that newly formed silicic acid is colloidal but the pri-
mary particles are too finely divided to be stopped by the membrane
employed; but, after a time, these very highly dispersed particles
coalesce to form larger primary particles with the loss of adsorbed
water as a result of the decrease in specific surface. In the light of
Harman's work, it now appears that silicic acid, as such, exists in solu-
tion as Mylius and Groschuff assumed ; moreover such a solution goes
over in time to the colloidal state and the particles of colloidal silica
agglomerate and age slowly but continuously, 27 * 28 approaching crys-
talline SiO 2 as a limit. 29 ' 2T So far as the author is aware, the experi-
ment has not been tried, but one might expect to find ultramicrons even
in dilute solutions of the definite salt Na 2 SiOa, if the solutions were
allowed to stand for a long time. Since everybody recognizes the exist-
ence of colloidal silica in more siliceous solutions, it is obvious that the
system when first prepared is not in equilibrium, attaining the condition
only after long standing, if at all. The tendency of colloidal silica par-
ticles to coalesce into larger aggregates, and finally to gel, is clearly
recognized by Vail 30 as one of the outstanding characteristics of sili-
cate solutions containing colloidal silica.
SILICATE GELS
When solutions not too dilute of sodium silicate and the salt of a
heavy metal are allowed to interact, a gelatinous precipitate is obtained
which varies in composition depending on the Na 2 O : SiO 2 ratio in the
silicate solution, the nature of the metal, the concentration of the solu-
tions, and the relative proportions of the reacting materials. Jordis
and collaborators 81 have made a careful study of the reactions between
solutions of sodium metasilicate and copper sulfate, ferrous sulfate,
and ferric chloride, with the following results :
"Weiser: "The Hydrous Oxides," 194 (1926).
"67 Schwarz and Stowener: Kolloid-Beihefte, 19, 171 (1924)
2* C/. Schwarz and Liede: Ber., 63B, 1509, 1680 (1920) ; Schwarz Kolloid-Z ,
28, 77 (1921); Grundmann: Kolloid-Beihefte, 18, 197 (1923).
20 Cf Bachmann- Z anorg. Chem , 100, 1 (1917); Zsigmondy-Spear
"Chemistry of Colloids," 137 (1917).
s J. Soc. Chem. Ind., 44, 214T (1925).
si Jordis and Hennis: J prakt Chem, (2) 77, 240 (1908) ; Jordis: Z angew.
Chem., 21 II, 1982 (1908).
SILICATE GELS 369
Copper metasilicate occurs in nature and may be prepared in a
definitely crystalline form by placing cupric nitrate and potassium sili-
cate on opposite sides of a permeable membrane which allows slow
diffusion 32 and at the same time holds the colloidal constituents in the
silicate solution. On rapid mixing of equal amounts of 0.1 M solutions
of copper sulfate and sodium metasilicate, the resulting gel contained
all the copper but between 3.5 and 4.75% of the silicate remained in
solution. When the amount of silicate was doubled, the additional
silicate did not all remain in solution, which contained but 43.8% in-
stead of the expected 54%. Equal amounts of N CuSO 4 and N Na 2 O -
2SiO 2 gave a gel which contained all but 4.5% of the total SiO 2 . The
gel developed microscopic green crystals, probably of CuSiO 3 , on
standing. Since the aqueous suspension of copper hydroxide turns
black on boiling and that of the gel does not, it is claimed that the
latter consists essentially of copper silicate rather than a mixture of
hydrous copper hydroxide and silica. This does not follow since the
blue copper hydroxide is stabilized by various substances (Vol. II,
p. 151) and silica may well be one of them.
Ferrous sulfate behaves toward silicate solutions like copper sul-
fate, and some ferrous silicate may form. Ferric and aluminum salts,
on the other hand, give mixtures of the respective hydrous metallic
oxides and hydrous silica.
Britton M has followed electrometrically the course of the reaction
when 100 cc of solutions of various metallic salts were treated with a
solution of Na 2 O 2.16SiO 2 that was 0.051 M with respect to NaOH.
In Table XLVII are given the solutions titrated, the />H value and
the amount of silicate required to start precipitation, and the ^H value
at which the respective hydrous oxides or hydroxides precipitate. The
titration curves arc given in Fig. 56. The point of beginning precipita-
tion is indicated by a short vertical line cutting the curve. In each
reaction 39 cc of alkali silicate give alkali corresponding to 100 cc of
salt solution. The curves lying beyond 39 cc give the effect of increas-
ing amounts of sodium silicate on the />H value of the mother liquor.
Some measure of the amounts of silica remaining in solution and hence
of the amounts carried down with the respective metallic oxides or
hydroxides was obtained by following the />H change when 0.051 M
Na 2 O 2.16SiO 2 was added to 139 cc of solution containing the same
concentration of neutral sodium salt as the original mother liquors, but
with different amounts of silica. Thus, curve I in Fig. 56 gives the
32Becquerel. Compt. rend., 67, 1081 (1868).
33 J. Chem. Soc., 425 (1927).
370
SILICATE SOLS AND GELS; THIXOTROPY
change in pU when the sodium silicate is added to 139 cc of solution
containing 39 cc of sodium silicate exactly neutralized with hydro-
TABLE XLVII
TlTRATION OF METALLIC SALTS WITH 051 M NA 2 O 2 . 16SlOj
Solutions titrated
100 cc
Precipitation began
Corresponding
oxide or
hydroxide
precipitates
atpH
atpH
with
cc of silicate
01 M ZrCl 4
01 M ThCU
3.98
3.50
4 04
5 31
5.25
7 35
9 50
10 07
35.0
30
5.0
20.0
1
1
1
3
1 86
3.50
4 14
5 69
5 20
8 41
10.49
067 M AlaCSOOs
002MBeSO 4
02 M ZnSO 4
02 M MnCU
02 M MgSO 4
02 M CaClj
0.3.
10
20 30 40
0.051 M Sodium Silicate, Cc
50
FIG. 56. Electrometric titration curves for metallic salts with 0.051 M
Na.O-2.16SiO*
SILICATE GELS 371
chloric acid. This curve would have been obtained if the oxides or
hydroxides alone had precipitated and all the silica had remained in
solution. Curve II shows the variation in pH which would have been
produced if the metasilicates had been precipitated and the remainder
of the silica had been left in solution. And curve III corresponds to
the change in />H which would have occurred if all the silica in the
39 cc had been precipitated. Except for the curve for calcium, the
portion of all curves corresponding to an excess of sodium silicate lies
between curves I and II ; hence the amounts of silica carried down by
the various precipitates were smaller than that required to form the
normal metasilicates (cf. p. 363). Britton interprets the calcium curve
to mean that calcium silicate is formed, the high />H value after the
addition of 39 cc of sodium silicate resulting from (a) hydrolysis of
the calcium salt and (b) hydrolytic adsorption of silicate from the
excess of reagent. All the other precipitates were believed to be
mixtures of oxides rather than combinations to give silicates, but the
possibility of some silicate formation with certain metals has not been
excluded. In general, one might expect the gel to contain considerable
silicate with salts of the alkaline earths ; 34 some silicate with salts of
copper, cobalt, and nickel ; 3B and little or no silicate with salts of ferric
iron, chromium, and aluminum. With the last group of salts, the gela-
tinous mass will probably consist almost entirely of hydrous oxides of
the trivalent metal and hydrous silica.
The gels formed on mixing water glass and solutions of the chlo-
rides of aluminum, chromium, calcium, copper, and nickel readily part
with the metallic oxide, leaving pure silica on treating the dried prepa-
rations with hydrochloric acid. 36 By this procedure a silica gel with
superior adsorbing qualities may be prepared.
Lloyd 87 prepared from fuller's earth an aluminum silicate gel
which adsorbs alkaloids strongly from either neutral or acid solutions ;
the adsorbed alkaloid may be recovered by treatment with an alkaloid
solvent. It has been suggested that the material be called Lloyd's
reagent. 88
The base-exchange phenomena in silicate gels both natural and syn-
thetic will be considered in Chapter XXL
3* Jordis and Kanter: Z. anorg. Chem., 86, 90, 344 (1903) ; 42, 419 (1904) ;
48, 48, 314 (1905) ; Le Chatelier- "Recherches exp6rimentales sur la constitution
des mortieres hydraliques," Paris (1887).
8B Schwarz and Mathis: Z. anorg. Chem., 126, 55 (1923).
86 Holmes and Anderson: Ind. Eng. Chem., 17, 280 (1925).
ST U. S. Pat. 1,048,712 (1912).
aWaIdbott: J. Am. Chem. Soc., 86, 837 (1913).
372 SILICATE SOLS AND GELS; THIXOTROPY
The "Silicate Garden"
When small crystals of various readily soluble salts such as cobalt
nitrate, copper sulfate, ferrous sulfate, nickel sulfate, manganese sul-
fate, zinc sulfate, or cadmium nitrate are dropped into a silicate of soda
solution of the right concentration and alkalinity, growths resembling
plant-shoots spring up, giving rise to the so-called "silicate garden,"
"artificial vegetation," or "colloidal forest." 89 When the soluble crys-
tal comes in contact with the silicate solution, it is quickly surrounded
by a semipermeable siliceous gel. Diffusion then takes place through
the gel membrane producing pressure inside the envelope which is dis-
torted or broken depending on the rigidity of the structure which, in
turn, depends on the concentration of the silicate solution and the na-
ture of the metallic salt. The water which passes through the membrane
dissolves more salt, and a new wall is formed around the extruded
solution. This process continues, giving rise to a great variety of
curious shapes and forms which differ markedly with different metals ;
for example, hair-like filaments result with cadmium salts, and thick
fungoid growths with nickel salts. The development of the plant-like
shoots is more rapid in dilute than in concentrated silicate solutions
up to the point where the gel is not sufficiently rigid to retain its form,
and is more rapid in solutions where the Na 2 O to SiO 2 ratio is high
than in the more siliceous solutions. The gels consist of the hydrous
oxides of the metals and hydrous silica 40 or of a mixture of these with
hydrous metallic silicates.
SILICATE SOLS
A stable silicate sol is best prepared by shaking a soft clay with
water preferably in the presence of small amounts of electrolytes with
strongly adsorbed anions such as alkali hydroxide, carbonate, or silicate
which exert a marked peptizing or deflocculating action. The nega-
tively charged particles of the clay sol are readily flocculated by elec-
trolytes with strongly adsorbed cations. The properties of the clay
sols will be discussed in Chapter XXII.
Ota and Noda 41 synthesized what they called silicate sols by the
addition of a dilute water glass solution to dilute solutions of salts of
3Bottger: J. prakt Chem., 10, 60 (1837); Mulder: 22, 41 (1841); Dollfus:
Compt. rend., 143, 1148 (1906); Ross: Proc. Roy. Soc, New S. Wales, 44,
583 (1910); J. Chem. Soc., 102 (2), 49 (1912); Krug: U. S. Pat 1,584,779
(1926).
Fordham and Tyson: J. Chem. Soc., 483 (1937).
J. Sci. Agr. Soc (Japan), 268, 287 (1924).
FACTORS DETERMINING THIXOTROPIC BEHAVIOR 373
copper, silver, magnesium, zinc, aluminum, titanium, manganese, fer-
rous iron, and cobalt. Although a part of the colloidal material in
such preparations may be the metallic silicate, such salts will hydrolyze
to a greater or lesser extent, giving a mixture of the hydrous oxide
of the metal and hydrous silica. It seems altogether unlikely, for exam-
ple, that any titanium silicate should exist in contact with water.
The colloidal mixtures of silicates and hydrous oxides exert an
enzymic action like an oxidase when examined by the indophenol re-
action. The order of activity of the silicate sols is : Ag > Cu > Co >
Ti > Al > Mn > Zn > Mg > Fe++. Powdered kaolin, talc, and ser-
pentine do not exhibit this action. This behavior recalls the oxidasc-
like action of colloidal manganese dioxide toward guiac tincture, hydro-
quinone, etc. (Vol. II, p. 332).
THIXOTROPY
If a suitable amount of electrolyte is added to a fairly strong ferric
oxide, chromic oxide, or alumina sol and the mixture is allowed to
stand quietly, it will set to a jelly which is no more cloudy than the
original sol. If the resulting jelly is shaken, a sol is re-formed which
will set again on standing; and the process may be repeated as often
as desired. This isothermal reversible sol-gel transformation has been
termed 42 thixotropy which is derived from the Greek and means lit-
erally "to change by touching." Since the phenomenon is often en-
countered among the hydrous oxide systems, frequent reference is
made to it in Volume II of this series.
The reversible transformation of a true inorganic sol to a true gel
is, as Freundlich 43 points out, a specially characteristic and limiting
case of thixotropy. This term also includes some related phenomena
where the sol and gel may not be sol and gel in the true sense of the
words, i.e., contain only ultramicroscopic particles. For example, sus-
pensions of clay and claylike substances such as bentonite, which con-
tain much larger particles, may exhibit marked thixotropic behavior
under certain conditions. Because of the importance of this behavior
both in theory and in application, it will be considered in some detail.
Factors Determining Thixotropic Behavior
Mutual Attraction and Repulsion. The particles of a sol do not,
in general, come in contact, because the magnitude and the like sign
of the f -potential exert a repelling action which keeps them apart.
"Peterfi: Arch. Entwicklungsmech. Organ., 112, 660 (1927).
48 "Thixotropy," Hermann et Cie, Paris, 3 (1935).
374
SILICATE SOLS AND GELS; THIXOTROPY
But, if the potential on the particles is lowered by the careful addition
of a coagulating electrolyte, a point is reached where the repelling ac-
tion is just insufficient to prevent their adhesion when standing quietly.
If the sol is sufficiently strong, the adhering particles form an en-
meshing network structure which entrains the liquid phase, giving a
jelly. Shaking breaks up the jelly structure again, giving the instable
sol which will set once more on standing. Since a concentrated sol
may be rendered thixotropic by adding electrolyte just short of coagula-
tion, one should expect to find a close relationship between thixotropy
of concentrated systems and the coagulation of more dilute systems.
This is illustrated by the data for iron oxide sols given in Table
XLVIII. 44 As a measure of the thixotropy of the concentrated sol
TABLE XLVIII
COMPARISON OF THIXOTROPY AND COAGULATION
Electrolyte
Q
52.6gFe 2 0,/l
C e
4 92 g FeaOa/1
NaCl
45
325
KC1
KBr
45
62
350
500
NaOH
Na 2 S04
Na 2 CaO 4
18
12
9 5
6 5
3 25
2.75
Ks-citrate
7.0
1.7
(52.6 g Fe 2 O 3 /l) is given the concentration Ct in millimols per liter of
the several electrolytes which will cause it to set to a jelly in 400 sec-
onds ; and as a measure of the coagulation concentration of the dilute
sol is given the concentration C e in millimols per liter which will cause
complete coagulation. It will be seen that for both phenomena the
order of electrolytes is the same and that the valence of the anion has
the usual marked effect on the positive sols.
The time of set is a measure of the time required for a sufficiently
large number of particles to come in sufficiently close contact that the
attractive forces cause them to adhere into an enmeshing network that
entrains the liquid phase. It is still an open question 45 whether in the
"Freundlich and Sollner: Kolloid-Z., 46, 348 (1928).
"Cf. Heller: Kolloid-Z., 60, 125 (1930); Compt. rend., 202, 61, 1507 (1936).
FACTORS DETERMINING THIXOTROPIC BEHAVIOR 375
set jelly all the particles are fixed and motionless or whether the co-
herence of only a fraction of the particles furnishes a network that en-
doses a more dilute sol phase. The latter view seems to be favored
theoretically/ 6 but microscopic and ultramicroscopic observations on a
thixotropic ferric oxide jelly indicate that all the particles are motion-
less. 47 However this may be, the phenomena of thixotropy, coagula-
tion, and adhesion can be explained only by assuming far-reaching at-
tractive forces. * 8 As an illustration Freundlich 49 made the following
calculation: Bentonite appears to consist of rod-shaped or plate-like
particles having the average dimensions of 1 : 0.1 : 0.01 /*. In a thixo-
tropic paste made by mixing 1 g of bentonite with 7 cc of water, one
particle will bind 4.7 X 10 8 water molecules. Assuming that all par-
ticles contribute to the solidification of the sol, 50 the average distance
between two particles would be 100 mp. Under such circumstances the
attraction which holds the particles together cannot be classical van der
Waals forces since, according to the theory, these decrease with the
sixth power of the distance and so operate only at very small distances,
not greater than the dimensions of a molecule. An extension of this
theory by London, 51 based on the quantum mechanics, suggests that
the attractive force between two molecules would also decrease with a
high negative power of the mutual distance r, e.g., proportional to
(1/r) 6 ; but that the forces have an additive behavior. On the basis
of London's theory Kallmann and Willstatter M calculated that the
attraction between large particles containing a large number of mole-
cules may operate at much greater distances, e.g., they may decrease
with the negative second power of the mutual distance.
In hydrophobic sols, the repulsive forces resulting from the double
layers of ions surrounding the particles decrease according to e~ tr t
where k, a constant used by Debye in his theory of strong electrolytes,
is a function of the concentration of the electrolyte present and the
valence of the ions.
From these considerations the relationship between adhesion, co-
agulation, and thixotropy may be represented diagrammatically as in
C/. Usher: Proc. Roy. Soc. (London), 125A, 143 (1929); Kuhn: Z.
physik. Chem., A181, 26 (1932); Bary: Compt. rend., 196, 183 (1933).
Szegvari and Schalek: Kolloid-Z., 82, 318; 88, 326 (1923).
C/. Buzdgh: Kolloid-Z., 51, 105, 230; 52, 46 (1930).
Kolloid-Z., 46, 289 (1928).
w Cf. Hauser: Kolloid-Z., 48, 57 (1929).
61 Z. physik. Chem., Bll, 222 (1931); Trans. Faraday Soc., 88, 8 (1937).
62 Naturwissenschaften, 20, 952 (1932).
376
SILICATE SOLS AND GELS; THIXOTROPY
Fig. 57, constructed by Freundlich 5S from calculations of Rubin. The
abscissa r is the mutual distance between two particles, and the ordinate
l<r 2 M/L -4-10T 3 M/L 6-10- 3 M/L|O10-lM/L
-40 mv. AC-6mv. I
0.4 0.8 1.2
Distance between Two Particles, r x 10" 4 Cm
1.6
FIG. 57 Curve showing the relationship between adhesion, coagulation, and
thixotropy.
is the potential energy of the repelling forces + E and the attracting
forces E. The curve of attraction extends with small but appre-
BS "Thixotropy," 13 (1936); cf. Hamaker: Rec. trav. chim., 85, 1015 (1936);
66, 1 (1937).
FACTORS DETERMINING THIXOTROPIC BEHAVIOR 377
ciable values up to about 1 /*. The upper four continuous curves repre-
sent the potential energy of repelling forces at four different concen-
trations of potassium chloride. The distance to which such forces are
effective diminishes with increasing electrolyte concentration. To test
whether an equilibrium exists between attracting and repelling forces,
the corresponding values of potential energy are shown in the broken
curves 1, 2, 3, and 4. If a curve of this kind has a minimum, it means
that an equilibrium exists between the two forces for the value of r
which corresponds to the minimum. The absence of a minimum in
curve 1 shows that the repelling forces predominate and a stable sol
exists. The minimum value in curve 2 is not well defined, but in curve
3 it lies at a value of r = 0.5 /*. Accordingly, a large amount of liquid
exists between the particles, and the equilibrium system whose particles
are motionless because of the balance between the repelling and attrac-
tive forces constitutes a thixotropic gel. In curve 4 the minimum value
is found at such a small value of r that but little liquid is enclosed be-
tween the particles, as in the coagulum from a sol.
The relationships among sol stability, thixotropy, and electrolyte
coagulation are apparently more complicated than the potential curves
of Fig. 57 would suggest. Hamaker M recognizes four different types
of potential curves considered as a superposition of London-van der
Waals attraction and electric repulsion. The curves shown diagram-
matically in Fig. 58 represent energies, and consequently the slope of
the curves represent forces. Corresponding curves of similar shapes
may be drawn in which the slopes represent energies. In curve 1 of
Fig. 58 the attractive force predominates over the repulsive force what-
ever the distance between the particles; in curve 2, the reverse of 1,
the electric repulsion is stronger everywhere than the London-van der
Waals attraction ; in curve 3, intermediate between 1 and 2, the repul-
sive force is greatest when the particles are separated by a large dis-
tance but the attraction predominates when they are close together ; and
in curve 4, the reverse of 3, the attraction is strongest at large distances
and the repulsion predominates at small distances.
The conditions represented by curve 3 probably obtain with most
sols, but curve 2 may represent the conditions in special cases. This
point of view enables one to account for reversible and irreversible
coagulation: Starting with a stable sol for which curve 3 holds and
adding sufficient electrolyte, the electric repulsion is reduced and the
curve assumes the shape of curve 1 which causes rapid coagulation.
By dialyzing the flocculated sol, the potential is restored to its original
"Rec. trav. chim., 56, 1015 (1936); 56, 1 (1937).
378
SILICATE SOLS AND GELS; THIXOTROPY
value but the gel will not be restored to its original stable condition
since the energy minimum F will keep the particles in contact in the
same way as the maximum S kept them apart. On the other hand, by
starting with a stable sol for which curve 2 is valid and adding electro-
lytes, the curve may assume the form of type 3 and finally of type 1
which leads to flocculation. Upon dialyzing the floe, the same set of
FIG. 58. Types of potential curves.
curves is passed in the reverse direction with the ultimate peptization
of the coagulum.
Curves 3 and 4 represent intermediate cases between 1 and 2, but
it cannot be predicted which will be valid in a given system. Referring
to Fig. 57 again, it will be seen that the transition between the lowest
and highest type of curve is given by a set of curves of type 4. Ha-
maker points out that Freundlich has failed to recognize the possi-
FACTORS DETERMINING THIXOTROPIC BEHAVIOR 379
bility of an equivalent though somewhat more complicated curve 3.
If this shape of curve is also taken into account, it greatly complicates
the consideration of the whole system. Thus, curves of either type 4
or type 3 may explain thixotropy. Those of type 3 may be useful also
in accounting for thixolability, thixotropic systems of very low stability,
such as has been observed with an alumina system. 55
Hydration. The formation of a water hull around the particles is
believed by Freundlich 58 to be of far more significance for thixotropic
behavior than is an equilibrium between electrical and London-van der
Waals forces. In support of this, thixotropic pastes have been made
of mercaptobenzothiazol and such liquids as benzene, toluene, and gaso-
line where electrical influences are largely absent. 67 The thixotropy is
attributed to adsorption of a rather thick layer of liquid, giving quasi-
fluid particles which occupy enough volume so that interference is suf-
ficient to induce gelation. Kistler 58 assumes some form of oriented
anisotropy of the water, probably chains of water molecules which ex-
tend out from the surface of each particle and tend to link it with
neighboring particles.
Although hydration, or more generally solvation, may influence
thixotropic behavior in certain instances, Hauser 59 80 ' 61 and collabora-
tors reject the water-hull theory of thixotropy in bentonite- water sys-
tems on several grounds : ( 1 ) Since gelation occurs with very dilute sus-
pensions (less that \% in some instances) the hydration theory would
require the formation of water hulls around the particles many mole-
cules thick, which seems unlikely. (2) Bentonite gels possess a tensile
strength which is difficult to explain on the basis of particles with ad-
sorbed layers of water touching each other. 62 (3) The rate of gelation
increases with increasing temperature, contrary to the expected be-
havior if water hulls are formed, since hydration usually falls off with
rising temperature. (4) The addition of alcohol, a dehydrating agent
Freundlich and Bircumshaw: Kolloid-Z, 40, 21 (1926).
"Freundlich and Schalek: Z. physik. Chem., 108, 153 (1923); cf. Ostwald:
Kolloid-Z, 260 (1928); Hauser: 48, 57 (1929); Deutsch: Z. physik. Chem.,
A150, 161 (1930); Werner: Ber., 62B, 1525 (1929); cf., also, Buzagh: Kol-
loid-Z., 51, 105 (1930).
"Recklinghausen: Kolloid-Z., 60, 34 (1932).
"J. Phys. Chem., 85, 815 (1931).
"J. Rheol, 2,5 (1931).
Hauser and Reed: J. Phys. Chem., 41, 911 (1937).
"Broughton and Squires: J. Phys. Chem., 40, 1041 (1936).
82 Lewis, Squires, and Thompson: Trans. Am. Inst. Mining Met. Engg.,
118, 1 (1936).
380 SILICATE SOLS AND GELS; THIXOTROPY
for hydrous colloids, changes the setting time of a bentonite suspension
but slightly instead of greatly increasing it, whereas addition of water
in equal amounts by volume increases indefinitely the time of setting.
Size and Shape of Particles. The particle size must be below a
certain limit for a system to show thixotropic behavior. 68 With Soln-
hofen slate, for example, Freundlich and Juliusburger 6 * found that a
certain percentage of the particles must have a diameter of 1 /* and
less; if the mass contained particles with diameters of 10 p, and more,
the system was not thixotropic. Similarly, centrifuged suspensions of
bentonite will give dispersions which show thixotropy at concentrations
of 1 % or lower, whereas ordinary bentonite from which the larger par-
ticles have not been removed by centrifuging must have a concentration
above 4%. M
The shape of the particles is likewise of great importance for thixo-
tropic behavior. Hauser 68 contends that a non-spherical shape is es-
sential for the manifestation of true thixotropy. In line with this, it
appears that the particles are rod-shaped in the thixotropic sols of
vanadium pentoxide, 67 benzopurpurin, cotton yellow GX, 68 dibenzoyl
cystine, 69 barium malonate, 70 and iron oxide ; 71 and plate-like in the
thixotropic pastes of bentonite 62 61 and clay. 72 Moreover, a drum's
alumina sol (Vol. II, p. 10S) is thixotropic 73 only when it exhibits
streaming double refraction, a phenomenon resulting from orientation
of non-spherical particles. On the other hand, an alumina sol may be
prepared which is thixotropic but which does not show streaming dou-
ble refraction, possibly because a lesser asymmetry of particles is neces-
sary for thixotropy than for streaming double refraction.
The gelation of bentonite is visualized by Lewis, Squires, and
Thompson 62 61 as follows : the particles are assumed to be flat plates
fi3 Freundlich "Thixotropy" (1935); Russell: Proc Roy Soc (London),
154A, 550; Russell and Rideal: 540 (1936).
* Trans Faraday Soc, 30, 333 (1934); 31, 769 (1935).
05 Cf. Broughton and Squires . J. Phys Chem., 40, 1041 ; Hauser and Reed
1169 (1936); 41,911 (1937).
eJ. Rheol.,2,5 (1931).
* Freundlich and Schalek: Z. physik. Chem., 108, 153 (1924); Jochims-
Kolloid-Z., 41, 215 (1927); Rabinerson: 68, 305 (1934).
8Kuhn and Erdos: Kolloid-Z., 70, 241 (1935).
"Papkova-Kwitzel: Kolloid-Z, 69, 57 (1934).
'oZocher and Albu: Kolloid-Z., 46, 27 (1928).
"Kandelaky: Kolloid-Z., 74, 200 (1936).
Jeppesen: Kolloid-Z., 57, 175 (1931).
73 Freundlich and Bircumshaw: Kolloid-Z., 40, 19 (1926); Aschenbrenner :
Z. physik. Chem., 127, 415 (1927).
THIXOTROPIC BEHAVIOR 381
which become loosely packed in three-dimensional random orientation,
edge touching edge in such a way that movement is impossible and a
solid "house of cards" gel structure is built up. Since gelation takes
place at concentrations of bentonite as low as 0.1%, it is necessary for
interference between particles to assume that their length and width is
much greater than their thickness, the ratio between these dimensions
being of the order of 1000-2000 to 1. Ultramicroscopic examination
fails to support the assumption that such ratios exist. 74 The view that
bentonite particles consist of plates of microscopic thickness and
macroscopic length and width was put forward by Wherry 75 on micro-
scopic grounds although direct microscopic evidence is lacking. The
results of x-ray analysis 76 indicate that bentonite clays consist of silica
and gibbsite layers separated by water, the spacing due to this separa-
tion depending on the water content of the bentonite. In the presence
of a large excess of water it is possible that the plates actually break
away from each other and behave as plates of molecular or colloidal
dimensions. The view that gelation is due to mechanical packing of
plate-like particles might account for gelation of concentrated bentonite
suspensions, but it is inadequate to explain the phenomenon in ex-
tremely dilute suspensions.
THIXOTROPY OF BENTONITE AND CLAYS
Thixotropic Behavior
Bentonite is a claylike material probably of volcanic origin found
largely in Wyoming. The mineral montmorillonite is its chief con-
stituent. This may be formulated A1 2 O 3 CaO 5SiO 2 ," but it con-
tains exchangeable cations other than calcium (magnesium, sodium,
potassium, etc.). Electrodialysis of bentonite removes most of the
metallic cations and replaces them with hydrogen, giving a hydrogen
bentonite. The direction of the streaming double refraction of ben-
tonile suspensions 78 is changed from negative to positive 79 by electro-
dialysis, indicating that the exchangeable cations are a part of the crys-
tal structure of the bentonite. 80
After hydrogen bentonite is dried, it does not give thixotropic sus-
Cf Hauser and Reed: J. Phys. Chem., 41, 911 (1937)
"Am. Mineral., 10, 120 (1925).
76 Marshall: Science Progress, 80, 422 (1936).
"Ross and Shannon: J. Am. Ceram. Soc., 9, 77 (1926).
"Buzdgh: Kolloid-Z., 47, 223 (1929).
^Bradfield and Zocher: Kolloid-Z. f 47, 223 (1929).
so Cf. t however, Hofmann, Endell, and Wilm: Z. Krist., 86, 340 (1933).
382 SILICATE SOLS AND GELS; THIXOTROPY
pensions or pastes, but the dried clay may be converted into the original
thixotropic bentonite by treating with solutions containing the cations
removed by electrodialysis. Bentonite swells in water and aqueous
solutions, a phenomenon closely related to its thixotropic behavior. A
hydrogen bentonite or an alkali-metal bentonite swells much less than
a natural bentonite.
As already pointed out, suspensions containing as little as \% of
a natural bentonite are thixotropic provided an optimum particle size
is obtained by removing the larger particles with the centrifuge. In-
deed, Hauser and Reed 81 - 80 report evidences of gel structure at con-
centrations as low as 001-0.05% by weight in the low colloidal range.
The most common method of measuring thixotropy is to determine the
time which elapses after shaking the suspension in a sealed tube until
it no longer runs on inverting the tube. Since thixotropic systems ex-
hibit plastic flow, that is, they possess a definite yield value, the shape
and size of the tube will influence the setting time. 61 In order to follow
the rate of gelation, Freundlich and Rawitzer 82 and Price- Jones 83 de-
termined the tensile strength and Rroughton and Squires 61 measured
the viscosity by the falling-ball method. The latter found that the set-
ting time obtained by the inverted-tube method gives comparative re-
sults provided care is taken in manipulation and the diameter of the
tube is kept constant. To eliminate the effect of variation of procedure
in inverting the tube, Hauser and Reed 60 designed an inverting ap-
paratus which they call a physical pendulum thixotrometer. The ad-
vantage of a procedure such as the falling-ball method is that it gives
the whole of the gelation-time curve and not simply one point.
Effect of Electrolytes. As we have seen, the addition of electro-
lytes to hydrous oxide sols is essential to induce thixotropic behavior,
and the valence of the ion opposite in charge to that on the sol is a
predominating factor. Electrolytes must be added to hydrogen ben-
tonite suspensions also in order to render them appreciably thixotropic,
but the two systems are not comparable. Freundlich 84 believes that
thixotropy of bentonite compounds is correlated with their degree of
dissociation. Bentonite may be considered as a colloidal electrolyte
81 J Am. Chem Soc, 58, 1822 (1936).
82 Freundlich and Rawitzer- Kolloid-Beihefte, 26, 231 (1927); Hauser:
Kolloid-Z, 48, 57 (1929).
83 J. Oil Colour Chem Assoc, 17, 305 (1934).
8 * Freundlich : Kolloid-Z, 46, 290 (1928); Freundlich, Schmidt and Lindau:
Z ph>sik Chem. (Bodenstein-Festband), 333 (1931); Kolloid-Beihefte, 36, 43
(1932); Freundlich: "Thixotropy," 34 (1935).
THIXOTROPIC BEHAVIOR
383
consisting of large anionic micelles and small exchangeable cations.
Only very weak thixotropic behavior is observed with the presumably
weakly dissociated hydrogen bentonite whereas marked thixotropy
exists with the presumably highly dissociated alkali bentonites formed
by neutralizing the hydrogen compound with alkali hydroxides. Strong
dissociation is believed to favor the hydration of the particles which
Freundlich considers so important for inducing thixotropy,
Hauser and Reed 88 separated natural bentonite into fractions of
varying particle size by means of the supercentrifuge, and determined
the influence of particle size on the rate of set. The results of some
observations are shown graphically in Fig. 59 for 0.85% suspensions
Concentration of Bentonite, 85 i
Temperature. 25 C,
1.0
10 100 1000
Thixotropic Setting Time, Minutes
10.000
100,000
FIG 59
-Influence of particle size and alkali concentration on the thixotropic
setting time of bentonite suspensions.
having the following average diameters : fraction I, 14.3 m/x ; fraction
IT, 20.3 rm*; fraction III, 28.1 m/*; and fraction JV, 33.8 m/t. The
suspensions were H-bentonites to which varying amounts of potassium
hydroxide were added. In every instance, at a given concentration of
base, the time required to form a gel of a given strength increases with
decreasing particle size. Hauser and Reed claim that pH as such ap-
pears to have no particular effect on the thixotropy of bentonite (cf.,
however, Fig. 60).
The close relationship between thixotropy and the amount of water
bound or enclosed by the particles under various conditions is shown
by Freundlich's observations on hydrogen bentonite pastes to which
* 5 J. Phys. Chem., 40, 1161 (1936); 41, 911 (1937).
384
SILICATE SOLS AND GELS; THIXOTROPY
potassium hydroxide is added, Fig. 60. The abscissas are the log-
arithms of the hydroxide concentration. Curve 1 represents the setting
times, curve 4 the swelling capacities, and curve 3 the sedimentation
volumes. It is obvious that the optimum thixotropic behavior coin-
cides with the maximum degree of swelling of the bentonite and the
minimum volume of the sedimented material. Furthermore, curve 2,
which gives the change in />H value, shows that optimum thixotropic
behavior lies in the region of the neutral point and weak alkalinity.
200
500 1000
10 25 50 100
KOH, Millimols per Liter
FIG. 60. Relationship between alkali concentration and various properties of
thixotropic bentonite suspensions.
Effect of Temperature. Increasing the temperature of bentonite
suspensions increases greatly their rate of gelation. 81 ' 60 The same
equation which was found by Schalek and Szegvari 8e to apply to iron
oxide sols is applicable to bentonite suspensions: log C t = At + B,
where C t is the setting time, t the temperature in degrees C, and A
and B are constants. The temperature coefficient A is affected by a
change in the diameter of the containing vessel.
Rheopexy. Thixotropic sols containing distinct rod- or plate-like
particles which set spontaneously rather slowly may be made to set
rapidly by rolling the container gently between the palms of the hands
or by tapping it gently in order to facilitate the orientation of the par-
ticles. For example, a vanadium pentoxide sol which sets spon-
taneously in 60 minutes solidified after the rolling procedure in 15
seconds. Freundlich and Juliusburger 87 have named this phenomenon
w Kolloid-Z., S3, 326 (1923).
87 Trans. Faraday Soc., 31, 920 (1935); Juliusburger and Pirquet: 32, 445
(1936).
THIXOTROPIC BEHAVIOR
385
rheopexy from the Greek "pectos" meaning curdled or solidified. They
failed to observe rheopexy with bentonite, but Hauser and Reed " eo
showed that a very finely divided fraction of 1.3% bentonite which set
spontaneously in 25 minutes solidified in 15 seconds after gentle tap-
ping of the container. The effect of gentle mechanical action on the
rate of gelation is well illustrated by Hauser and Reed's observations
on suspensions of varying particle size, shown graphically in Fig. 61.
The suspensions contained 0.85% H-bentonite to which 76.5 millimols
KOH/1 were added. The points on the thixotropic curve were ob-
tained by allowing the suspensions to stand quietly. To obtain each
100
50
20
3 10
0.5
0.2
0.1
uxotropic Curvi
10
20 30 40
Average Equivalent Spherical Diameter, i
50
FIG. 61. Influence of particle size on the thixotropic and rheopectic setting time
of bentonite suspensions.
of the points on the rheopectic curve, the tube, 100 mm in length,
which held the suspension was grasped between the thumb and third
finger and was made to oscillate through an amplitude of 15-20 on
each side of the vertical at the rate of 250 complete oscillations per
minute. This treatment always enormously increased the rate of set ;
in extreme cases the mechanical motion produced gels of a given
strength several hundred times faster than if the suspension were not
moved.
Ultramicroscopic Observations. Hauser and Reed 80 - 88 observed
the ultramicroscopic changes responsible for larger-scale thixotropic
behavior. On the addition of 9 milliequivalents KOH/g of dilute ben-
tonite suspension, the individual particles were seen to cluster up form-
ing particle clouds which in turn form a loose mass of secondary clus-
ss Cf. Hauser: Kolloid-Z., 48, 57 (1929).
386 SILICATE SOLS AND GELS; THIXOTROPY
ters intermeshed with channels of freely moving liquid. The particles
forming the clouds showed no Brownian movement, but a few single
particles remained in free motion in the channels. The application of
any strong shearing force broke up the clouds and all the particles again
showed Brownian motion; and the whole process was repeated on
standing quietly. Gentle tapping increased the rate of formation of
the clouds into more sharply defined secondary clusters rheopetic be-
havior. Increasing the quantity of electrolyte caused the more rapid
formation of aggregates of denser structure.
Behavior of Ordinary Clay. Concentrated suspensions of ordinary
clay, kaolin, etc., behave somewhat like bentonite suspensions, but both
the swelling capacity and thixotropy are much less marked. Buzagh 78
claims that suspensions of kaolin are thixotropic in the presence of
alkali but not in pure water. Tamamushi 89 found that a 33>$% kao-
lin is thixotropic in a neutral salt solution such as potassium chloride.
Alkali solutions favor thixotropy in most clays, but some plastic clays
become thixotropic when mixed with a suitable amount of pure water. 72
The particles in such thixotropic clays appear to be plate-like in shape
and swell when placed in water.
Some Applications of Thixotropic Clays
Casting. The application of thixotropic clays for molding purposes
should be useful in view of the facts that the volume does not change
appreciably during the sol-gel transformation and thixotropy is usually
closely allied to isothermal plasticity. 90 A concentrated thixotropic clay
suspension will fill a mold full and may be separated from it after it has
solidified. In clay casting, it appears likely that the setting of a thixo-
tropic paste is the primary effect, the drying action of the plaster of
Paris mold being of secondary importance. In practice the mixtures
are always weakly alkaline, a condition favorable for thixotropy in clay
pastes. Evidence is not yet available as to whether the optimum condi-
tions for thixotropy and clay casting are similar. 91
Thixotropic Latex. Preserved latex or concentrates of latex show
no tendency to become thixotropic. The mixtures are slightly alka-
line, and when bentonite is added in suitable amount it renders the
whole mass thixotropic. 92 Such mixtures may be employed as a co-
agulant dip or for casting in molds. As an application of the former,
>J. Chem. Soc Japan, 57, 132 (1936); cf. Kimura: 56, 1346 (1935).
Freundlich- "Thixotropy," 18 (1935); McMillen: J. Rheol., 3, 163 (1932).
01 Cf. Stephenson: J. Am Ceram. Soc, 10, 924 (1927).
Pat. 342,469 (1928).
SOME APPLICATIONS OF THIXOTROPIC CLAYS 387
a mold the shape of a rubber glove is dipped into the mixture which has
been liquefied by stirring; it is then pulled out at such a rate that the
mass has solidified when the form has just left the dip. The rubber
mixture covers the form in a uniformly thick layer without any sus-
pending drops. After the dipping process, the latex is dried and
vulcanized. The problem of mixing large charges equally has retarded
the technical application of the procedure.
Drilling Fluids. The application of thixotropic behavior on a large
scale is found in the so-called "drilling fluids" used in drilling for
petroleum. 93 In this operation the drill passes through layers of clay,
quartz, etc. If the boring fluid is water alone, the mixture in the hole
will be a concentrated suspension of clay, etc., which may easily settle
down for one reason or another and "freeze" the drill. This may be
prevented by using as a boring fluid a thixotropic suspension of ben-
tonite which sets, at the worst, to a soft gel which is easily liquefied
again. i
One of the important functions of a drilling fluid is to seal off gas
formations by the hydrostatic head of the fluid column. For this pur-
pose weighted drilling fluids are employed which not only furnish a
pressure exceeding that of the gas but will penetrate the formation to
a slight extent, thus preventing entrance of gas by diffusion or solution.
Ambrose and Loomis calculate that, if the gas pressure is 1500 Ib./sq.
in. at 2750 ft., the drilling fluid weight per gallon must be at least
10.46 Ib./gal. and should be at least 2 Ib./gal. heavier to furnish a rea-
sonable excess pressure. Clay suspensions of this density are entirely
too viscous for pumping; hence a suitable drilling fluid is made by
adding 2-3% of bentonite to mixtures of substances such as barium
sulfate and hematite whose suspensions alone have a low viscosity.
"Baroid" is a technical mixture of this kind. Just as with latex, the
thixotropic bentonite tends to impart thixotropy to the whole mass.
The effect of />H on the stability of such suspensions as "Baroid" has
been worked out by Ambrose and Loomis and applied to actual drilling
practice.
The foreman may test the thixotropic properties of a drilling fluid
just as was done with the rubber glove, above mentioned. The hand is
dipped into the well-stirred liquid and drawn out slowly, observing
whether the solidified mass covers the fingers uniformly without any
drops hanging down.
3Lawton, Ambrose, and Loomis: Physics, 2, 365; 3, 185 (1932); Ambrose
and Loomis: 4, 265 (1933); Ind. Eng Chem, 26, 1019 (1933).
CHAPTER XXI
BASE EXCHANGE IN SILICATE GELS
Gazzari 1 in 1819 made the interesting and important observation
that clay decolorizes liquid manure and retains soluble substances
which are given up subsequently to growing plants. This was perhaps
the first work on base exchange or exchange adsorption, but it re-
mained for Way 2 in 1850 to elucidate the significance of Gazzari's
observation. Way originated the instructive experiment of allowing
potassium chloride to percolate through a column of soil which was
found to take up potassium but not chlorine and to liberate another ele-
ment, chiefly calcium, in place of the adsorbed potassium. Similarly
Way found that ammonium was taken up by the soil in exchange with
calcium. Thus the ingredients, potassium and ammonium, which are
indispensable for plant growth are retained in the soil and prevented
from leaching, at the expense of the common element calcium. Later,
Eichorn 3 showed that among the natural hydrated double silicates sev-
eral bases are mutually interchangeable. Such substances are now
called zeolites. An important example of this class of substances is the
synthetic permutite (from permutare, to change) first prepared by
Gans * by fusing 3 parts of kaolin, 6 of sand, and 12 of sodium car-
bonate, followed by leaching the vitreous mass with water. Cans'
preparation is a highly porous, almost amorphous gel which corre-
sponds approximately to the formula Na 2 O A1 2 O 3 2SiO 2 ' ^H 2 O ; it
may be similar in constitution to the naturally occurring crystalline
zeolite of the same composition.
BASE-EXCHANGE PHENOMENA
Permutites
If water containing calcium or magnesium chloride is allowed to
percolate through a column of sodium permutite, there results a solu-
*C/. Sistini: Landw. Vers.-Stat., 16, 409, 411 (1873).
2 J. Roy. Agr. Soc Engl., 11, 313 (1850).
'Fogg. Ann., 106, 126 (1858).
* Jahrbuch Kgl. Preuss. Geol. Landesanstalt, 26, 179 (1905) ; 27, 63 (1906) ;
cf. Riedel: Ger. Pat. 186,630 (1906); 200,931 (1907); Gans-Riedel: U. S. Pat.
914,405 (1908).
388
PERMUTITES 389
tion of sodium chloride free from calcium or magnesium, and calcium
or magnesium permutite is formed. The original permutite is restored
by percolating a concentrated solution of sodium chloride through the
alkaline-earth permutite, whereby the calcium or magnesium is re-
placed by sodium. This base exchange is the basis of the permutite
process for softening hard water. If sodium or calcium permutite is
treated with a solution of a manganese salt, a manganese permutite is
produced ; and if a solution of potassium permanganate is employed, a
potassium-manganese permutite results, covered with a finely divided
layer of a higher oxide of manganese. This product is employed to
remove iron from water as well as for oxidizing organic matter and
bacteria in water. Regeneration is accomplished by treatment with a
solution of potassium permanganate.
In many cases, the exchange of cations takes place almost quanti-
tatively. Thus a sodium permutite in contact with a moderately con-
centrated silver nitrate solution was found by Gunther-Schulze 5 to
have exchanged 96.5$> of its sodium for silver in a day's time. The
extent and rate of the exchange depend in large measure on the physi-
cal character of the permutite; a freshly prepared sample exchanges
bases more completely and quickly than a dried one. 6 Permutites con-
taining but one base can be prepared by leaching the ordinary samples
for several months with solutions of the chloride of the metal desired.
Hydrogen permutite is prepared by electrodialysis of the ordinary prod-
uct as described on page 405.
A special siliceous gel known as "doucil" is made 7 by mixing solu-
tions of sodium silicate and sodium aluminate, drying, and washing to
remove soluble sodium salts. The dried gel has the approximate com-
position Na 2 O A1 2 O 3 5SiO 2 corresponding to the natural crystalline
zeolite, phillipsite. Like the zeolites and the artificial permutites, doucil
may be employed to soften water and to recover either alkali or alka-
line-earth metals from dilute solutions. 8 Unlike most base-exchange
silicates, doucil has a very minute gel structure possessing pores of
ultramicroscopic dimensions. It therefore exhibits a relatively high
exchange capacity for a given mass. "Its high capacity and quick re-
generation fits it especially for small domestic softening units although
Z physik Chem., 89, 168 (1915); cf. Bacon: J. Phys. Chem., 40, 747
(1936).
6 Beutell and Blaschke: Centr. Mineral. Geol., 142 (1915).
'Wheaton- Brit. Pat 177,746 (1922); U. S. Pat. 1,586,764 (1926).
Vail: Trans. Am. Inst. Chem. Engrs., 18 (2), 119 (1924).
390
BASE EXCHANGE IN SILICATE GELS
it has given excellent results in large industrial units especially designed
to take advantage of its properties." 9
Among the numerous observations on the base-exchange phenom-
enon in permutites special attention is called to the work of Wieg-
4 8 12 16 20
Salt Concentration, Milliequivalents per Liter
24
FIG 62. Isotherms for base exchange with ammonium-permutite and hydrogen-
permutite.
ner 10 and Jenny. 11 - 12 Some typical data of Jenny are shown graph-
ically in Fig. 62 for an NH 4 -permutite and a H-permutite. The ab-
Vail: J. Soc. Chem. Ind., 44, 2147 (1925).
"J. Landw., 60, 111, 197, 223 (1912); Kolloid-Z (Zsigmondy Festschrift),
86, 341 (1925).
"Jenny and Wiegner: Kolloid-Beihefte, 23, 428 (1926); Kolloid-Z., 42,
268 (1927).
"Jenny: Mo. Agr. Expt. Sta. Research Bull. 162 (1931); J. Phys. Chem.,
86, 2217 (1932).
CLAY 391
scissas represent the percentage displacement of ammonium and hydro-
gen respectively, and the ordinates, the equilibrium concentration of the
several electrolytes in milliequivalents per liter.
From these and similar data with other permutites and salts, it ap-
pears that the capacity of the ions to enter a gel and displace ammo-
nium (or hydrogen) is: Li < Na < K < Rb < Cs < H and Mg <
Ca < Sr < Ba. Similarly, the ease of replacement of the several ions
from alkali permutites by ammonium is : Li > Na > K > Rb > Cs.
The more highly hydrated ions like Li+ and Na+ are prevented by
their water envelope from coming as close to the oxygen atoms of the
permutite particle as the less hydrated ions. The electric charge is the
same for ions of the same valence, but the force of attraction is in-
versely proportional to the square of the distance between charges;
hence the large and voluminous lithium and sodium ions have the
lowest displacing power for ammonium and are the most easily dis-
placed by ammonium.
Hydrophilic colloids retard the base-exchange process 1S in the
order : gelatin > gum arabic > Irish moss > soil colloids > dextrin >
humic acid. The retarding action results chiefly from mechanical ef-
fects such as blocking of the pores of the zeolite.
Clay
The exchangeable bases in the soil are found chiefly in the colloidal
or clay fraction. This may be extracted from the soil by a suitable
method (p. 401) and subjected directly to base-exchange studies. Since
a clay may contain various exchangeable cations it is better to work
with a pure basic clay which may be prepared by adding the proper
amount of the hydroxides of sodium, potassium, ammonium, etc., to
H-clay formed by electrodialysis (for details see p. 405).
To illustrate the base exchange in clays, some results of Jenny " 1JJ
for an NH 4 -Putnam clay and a H-Putnam clay are shown in Fig. 63.
From these and similar experiments the adsorption (intake) of uni-
valent cations for Putnam clay was found to be : Li = Na < K < H ;
and for bentonite clay Na g Li ^ K < H ; the displacement of the
adsorbed ions takes place in the reverse order.
Although a comparison of the exchange adsorption curves for clays
and permutites would indicate an analogous behavior, the two classes
"Saner and Ruppert: Kolloid-Z., 78, 71 (1937).
" J. Phys. Chem, 36, 2217 (1932).
" Cf. Marshall and Gupta: J. Soc. Chem. Ind. f 52, 433T (1933).
392
BASE EXCHANGE IN SILICATE GELS
of substances are really quite different in certain respects: The per-
mutites are feebly crystalline or amorphous, and have a loose structure
from which practically all their cations may be readily replaced by
neutral salt solutions ; the clays are definitely crystalline, the structure
is less porous, and only a part of the bases present are readily replaced
12 18 24 30
Salt Concentration, Milliequfvalents per Liter
FIG. 63 Isotherms for base exchange with ammonium-Putnam clay and
hydrogen-Putnam clay.
by salt solutions. The maximum exchange capacity of a permutite
may be as high as 400-500 milliequivalents per 100 g (m.eq./100 g)
depending on the method of preparation, whereas for unground soil
colloids and bentonite it is seldom more than 100 m.eq./lOO g and
usually much less. The base-exchange capacity of permutites is altered
but little by grinding whereas that of the soil colloids may be increased
CLAY
393
greatly by reducing the size of the individual particles. 18 To illustrate,
in Table XLIX are summarized some observations of the effect of
grinding in a ball mill on the base-exchange capacity of (1) the zeolite,
natrolite; (2) a permutite; (3) bentonites; and (4) soil colloids. Sam-
ples under (3) and (4) were calcium saturated, before grinding, by
TABLE XLIX
EFFECT OF GRINDING ON BASE-EXCHANGE CAPACITY
Before grinding
After grinding
Material
Particle size
M.eq./lOOg
Time, hr.
M.eq./100 g
Natrolite
48
74 5
72
108 5
Permutite
225
48
225
Bentonite . .
I/*
126
72
238
Beidellite
IjLl
50
48
200 5
Yolo Soil Colloid
I/*
67
48
166
Cecil Soil Colloid ... .
I/*
17
48
151
Redding Soil Colloid . .
IM
35
48
122 5
leaching with calcium acetate solution. The base exchange is expressed
in m.eq./100 g.
Since grinding increases the base-exchange capacity, it might appear
that the distinction between exchangeable and non-exchangeable ions
in the soil colloids is only apparent and depends on the accessibility of
the ions as determined by the state of subdivision of the particles.
Actually, both the accessibility and the strength and nature of the at-
tractive force by which the ion is held to the crystal lattice play a role.
Kelley 17 found by x-ray analysis that grinding may actually destroy
the crystal lattice to a certain extent. Kelley's and Jenny's work indi-
cates that all the potassium and sodium and possibly all the calcium in
various classes of silicates are replaceable, provided that the particles
are made sufficiently small to allow access to these ions. 18 The situa-
tion with magnesium, however, is more complex : the octahedral mag-
nesium 19 is probably not replaceable stoichiometrically by alkali and
i* Kelley, Dore, and Brown: Soil Sci., 31, 25 (1931); Kelley and Jenny:
41, 367 (1936).
Trans. Intern. Congr. Soil Sci., 3rd Congr., Oxford, 3, 88 (1935)
18 Cf. van der Meulen: Rec. trav. chira., 64, 107 (1935).
" Pauling: Proc. Natl. Acad. Sci. U. S., 16, 123 (1930).
394 BASE EXCHANGE IN SILICATE GELS
ammonium, especially when OH- ions are a part of the octahedra.
On the other hand, magnesium can replace other replaceable cations
reversibly
The loss of ammonia from ammonium permutite, bentonite, and
clay increases linearly with the temperature. 20 The loss is practically
complete at 300 for the permutite ; 450 for the bentonite ; and 500
for the clay.
MECHANISM OF THE BASE-EXCHANGE PROCESS
The pioneer work of Way on base exchange led him to believe that
the process was essentially chemical in nature although he recognized
certain abnormal characteristics, such as the speed of exchange, in-
fluence of temperature, and existence of the lyotropic series, which are
not readily accounted for on a purely chemical basis. Liebig 21 took the
position that the phenomenon was essentially physical in character,
thus starting a controversy that has not yet been settled to the satis-
faction of everybody. It is known that the magnitude of the exchange
depends on the concentration of the salt solution used in the leaching.
A part of the controversy concerning the nature of the base-exchange
process has been precipitated by the use of solutions of only moderate
concentration, omitting very dilute or highly concentrated systems.
Some of the attempts that have been made to express the experimental
results mathematically are described in the following paragraphs.
Equations Based on the Law of Mass Action
Cans * observed the similarity between base-exchange equilibria and
those governed by the mass law and developed the following equation
based on the lattei, for ions of the same valence:
JT-
(m-n - x)(g - x)
where K is the equilibrium constant; m, the amount of the exchange
complex in grams; , the total amount in mols of exchangeable bases
in the complex; g, the total amount of the displacing ion in solution;
and x, the amount taken up.
By making use of different assumptions, other mass-action equa-
*oBottini: Kolloid-Z., 78, 68 (1937).
ai Ann., M, 373 (1855).
EQUATIONS BASED ON THE LAW OF MASS ACTION 395
tions have been set up by Anderegg and Lutz, 22 Kerr, 23 and Vanselow. 24
Marshall and Gupta 2B showed that the various formulations are inade-
quate to represent the experimental data. Kielland 26 has interpreted
some of Marshall and Gupta's data in terms of the mass law by taking
into account the so-called activity coefficients of the zeolite compo-
nents. 27 This mathematical analysis of the data for exchange reactions
with H- and Tl-Putnam clay and with H- and Tl-bentonite led to the
conclusion that with both materials a definite intermediate compound
is formed of the composition 2HZ T1Z where Z stands for the anion
portion of the substance. The improbability of the formation of a
definite double salt with any clay, and of the same double salt with a
Putnam clay and a bentonite, renders Kielland's analysis of question-
able value.
Rothmund and Kornfeld 28 point out that Gans' equation is a spe-
cial case of the more general mass-law equation :
~ = *7T
2 C/2
where r j and c 2 are the concentrations of the two ions in the solid
phase and C^ and C 2 are the concentrations of the corresponding ions
in the liquid phase. According to this equation, which applies through
a limited range, the equilibrium is independent of the volume and there-
fore remains unchanged whether liquid be added or removed. The
general relationship between the amount taken up and the equilibrium
concentration of the solution is given more nearly by the empirical
expression :
where c and C have the same significance as above, and k and n are
constants. This equation corresponds to that for the adsorption iso-
therm. Moreover, the value of the exponent lies between 0.3 and 0.7,
as in many typical cases of adsorption.
22 Soil Sci, 24,403 (1927).
23 J. Am. Soc. Agron., 20, 309 (1928).
2* Soil Sci., 33, 95 (1932).
2 * J. Soc Chem. Ind., 62, 433F (1933).
aeTids Kjemi Bergvesen, 15, 74 (1935); J Soc. Chem Ind, 64, 232T
(1935).
Cf. Randall and Cann: Chem. Rev., 7, 369 (1930); Holler: Kolloid-
Beihefte, 46, 1 (1937).
" Z. anorg. Chem., 103, 129 (1918); 108, 215 (1919); cf. Ramann and
Spengel: 95, 115 ,(1916); 105, 81 (1918); Ramann and Junk: 114, 90 (1920).
396 BASE EXCHANGE IN SILICATE GELS
Equations Based on Adsorption
The similarity in the form of base-exchange and ordinary adsorp
tion curves led Wiegner 29 to conclude that the process is an exchange
adsorption. In support of this he found that the exchange process is
fairly accurately described by Freundlich's familiar equation x/m =
k c 1/n , where x/m is the amount adsorbed per gram of adsorbent ; c is
the equilibrium concentration; and k and n are constants. Later
Jenny u modified the equation as follows :
m
to take care of the fact that the ionic exchange is independent of the
dilution (cf. p. 122). In the modified equation a represents the origi-
nal concentration of the added salt.
This empirical expression applied by Jenny to both permutites 11
and clays 14 was found to be quite satisfactory especially in dilute to
moderate concentrations with ions of low atomic weight. In a series
of experiments with different amounts of electrolyte, the constancy of
k is generally much better than 1/n which often shows a regular change
with concentration. With a given clay and various cations, the values
of k are closely related to the hydration of the cations and the mean
value of 1/n is controlled chiefly by the valences of the two cations
involved in the exchange.
Since the exchange reaches a maximum value at sufficiently high
concentrations of electrolyte, the parabolic equation of Freundlich will
not apply at the higher concentrations. To meet this difficulty, Vageler
proposed the equation :
x-S
where y is the amount taken up per gram of substance ; x, the equiva-
lents of salt added per gram of adsorbent ; S t the maximum exchange
capacity (saturation capacity) ; and C t the half value, that is, the con-
centration x at which 50% of S is exchanged. This hyperbolic equa-
tion, which is identical in form with the adsorption equation of Lang-
muir (Vol. I, p. 198), is more satisfactory than the Wiegner equation
at high concentrations. 25 - 80 Over a considerable range it represents the
2 J. Landw., 60, 111, 197 (1912); Kolloid-Z. (Zsigmondy Festschrift), 86
341 (1929); cf. Rabinerson: 62, 157 (1933).
Kottgen: Z. Pflanzenernahr. Dungung Bodenk., 82, 320
EQUATIONS BASED ON ADSORPTION 397
data more accurately provided the soil-water ratio is maintained con-
stant during the experiments. 31
Pauli 82 has given proof that the application of the law of mass
action to colloidal electrolytes leads to an equation corresponding to
Langmuir's adsorption equation. To arrive at this simple expression,
the surface is regarded as being in equilibrium with one kind of cation
only and the assumption is made that all ions which can dissociate
from the surface have equal chances of doing so. Neither of these
conditions is fulfilled in base-exchange processes hence the limited
applicability of the Vageler-Langmuir equation.
Jenny 33 has set up a simple base-exchange model and, on the basis
of kinetic concepts, has formulated an equation by the aid of statistical
methods. He considers a planar surface which contains a definite
number of attraction spots per unit area. If the ions, atoms, or mole-
cules which are initially adsorbed are designated by b and those which
are added to function as exchanging particles by w, then, at equilib-
rium, the number of cations w adsorbed or released is given by the
expression :
_ + (s + N) rfc V(s + N) 2 - 4sN(l - F/F 6 )
W 2(1 -
where N is the amount of electrolyte (number of ions) added initially;
s, the saturation capacity; and V w and F&, the volumes of the oscil-
lating spaces of the adsorbed ions.
The applicability of Jenny's adsorption equation is illustrated by
some data shown graphically in Figs. 64 and 65 obtained with a purified
Putnam clay. In the experiments, 7.5 g of the clay containing 4.50
m.eq. of adsorbed ions were treated with various amounts of the sev-
eral chlorides in a total volume of 500 cc, and the number of cations
in the supernatant solution was determined. The exchange values ex-
pressed in terms of saturation capacities (4.50 = 100% ) are plotted
as abscissas against the electrolyte expressed in terms of S, the satura-
tion capacity (e.g., 4S = 4 X 4.50 m.eq.). The lines were calculated
with the aid of Jenny's equation, and the observed values are shown as
dots, circles, etc. It is apparent that the kinetic equation describes
quite well the position and trend of the curves over a considerable con-
centration range. The agreement between theory and experiment is
almost perfect in some instances, but in others systematic deviations
31 Greene: Trans. Intern. Congr. Soil Sci., 3rd Congr, Oxford, 1, 63 (1935).
aapauli-Valko: "Elektrochemie der Kolloide," 108 (1929).
sa J. Phys. Chem, 40, 501 (1936).
398
BASE EXCHANGE IN SILICATE GELS
2 3
Electrolyte Added, 5
FIG 64 Exchange isotherms for ammonium-Putnam clay and various univalent
cations.
100 r
o - o Mg-Clay+BaClz
o oCa-Clay+BaCI 2
Ca-Clay+MgClz
2 3
Electrolyte Added, S
FIG 65 Exchange isotherms for magnesium-Putnam clay and barium chloride
and for calcium-Putnam clay and magnesium and barium chlorides.
EQUATIONS BASED ON ADSORPTION 399
seem to occur. The deviations are explained on the basis of structural
peculiarities of the colloidal particles and of extreme variations in the
properties of the participating ions. Jenny's adsorption equation has
the advantage over others in being a developed rather than an em-
pirical expression.
In the light of the above survey it would seem that the base-ex-
change process should be considered as an exchange adsorption phe-
nomenon rather than as a solid solution phenomenon governed by the
mass law. From base-exchange studies with glauconite, Austerweil 34
concluded that the distribution of ions between the zeolite and the solu-
tion takes place in accordance with the distribution law or law of
extraction, the solid taking the part of a non-miscible liquid. A corre-
lation was found between the extraction formula and Freundlich's ad-
sorption equation.
APPLICATION OF THE BASK-KXCIIANGE PROCESS
The phenomenon of base exchange was first observed in connection
with the fixation of fertilizer material by soils, and this is still its most
important application. A direct consequence of investigations with
clays was the invention of the zeolite or pcrmutite process of water
softening to which reference has already been made. Similarly, a
knowledge of the principles of base exchange has contributed to the
development of modern sugar-refining methods, and one may expect
the further application of these principles in such industrial processes
as the stabilization of technical emulsions and sols, the tanning of
leather, and the manufacture of milk products.
It is now becoming quite generally recognized that base exchange
plays an important role in the weathering of aluminum silicates which
results in the formation of soil The first step in this process appears
to be an ionic exchange in which the hydrogen ions of water and car-
bonic acid replace cations on the surface of the silicate minerals. It
is known that clays with the same SiO a : A1 2 O 3 ratio may vary widely
in properties because of differences in the nature of the exchangeable
cations which are present. Indeed, Gedroiz 35 has proposed a scheme
of soil classification based on the nature of the ions adsorbed on the
clay particles. 36 The investigations of Gedroiz, 35 de'Sigmond, 37 Kel-
Bull. soc. chim, (4) 51, 729 (1932); (5) 3, 1782 (1936).
35 Kolloid-Bcihcf te, 29, 149 (1929).
38 Cf. Bradfield: Trans. Intern. Congr. Soil Sci., 3rd Congr., Oxford, 2,
134 (1935)
87 Soil Sci., 21, 455 (1926).
400 BASE EXCHANGE IN SILICATE GELS
ley, 88 Burgess, 30 and others on sodium clays have shown that the poor
physical condition of alkali soils is caused by, or associated with, a
high content of exchangeable sodium and a relatively low salt content.
The reclamation of such soils involves the replacing of sodium by cal-
cium ions on an enormous scale (cf. p. 412).
The physical as well as the chemical properties of clays may be
determined in large measure by the nature of the exchangeable ions.
We have seen how the water content of clays will be influenced by
the hydration of the exchangeable cations; in the next chapter atten-
tion will be called to the role played by the exchangeable ions in the
phenomena of plasticity, flocculation, and deflocculation of colloidal
clays.
as J. Am. Soc. Agron, 22, 977 (1930).
30 Burgess and McGeorge: Ariz. Agr. Expt. Sla. Tech. Bull., 15, 359 (1927)
CHAPTER XXII
THE INORGANIC SOIL COLLOIDS
Schlosing 1 in 1874 prepared aqueous suspensions of clay soils
which were allowed to stand undisturbed for long periods. In the
course of time several distinct layers settled out leaving a sol which
contained particles so small that they were invisible in the highest-
powered microscope. By this method it was demonstrated that soils
contain more or less material in the colloidal state of subdivision.
The inorganic colloidal material in the soil is generally called clay.
The International Society of Soil Science in 1913 accepted the follow-
ing classification of soil particles on the basis of diameters of particles
expressed in millimeters: gravel, 20-2; coarse sand, 2-0.2; very fine
sand, 0.2-0.02 ; silt, 0.02-0 002 ; clay, < 0.002. Colloidal clay is that
fraction of the clay whose particles have an effective diameter less than
100 m/A (0.0001 mm). No lower limit is set on the diameter of the
clay particle, but experimental evidence places it as low as 10-20 m^. 2
COMPOSITION OF COLLOIDAL CLAY
Colloidal Content of Soils
It is difficult, if not impossible, to separate all the colloidal matter
from a soil. The earlier investigators merely rubbed up the soil with
a considerable amount of water and estimated as colloidal matter the
amount that remained suspended for a given length of time. Schlosing *
was of the opinion that the material which remains longest in suspen-
sion differs essentially from material which does not remain suspended
so long and so estimated the colloid content of soils to be only 0.5 to
1.59&. 8 Ehrenberg and Given 4 arrived at a similar conclusion. Hil-
iCompt. rend, 70, 1345 (1870); 78, 1276; 79, 376, 473 (1874).
2 Bradfield : J. Phys Chem., 35, 360 (1931) ; Colloid Symposium Monograph,
8, 360 (1931).
3(7/. Ehrenberg: "Die Bodenkolloide," Dresden, 99 (1922).
Kolloid-Z, 17, 33 (1915).
401
402 THE INORGANIC SOIL COLLOIDS
gard 5 and Williams 6 reported much higher percentages based on the
amount of material that does not settle in a 24-hour period.
Since the amount of soil that will remain suspended depends on
the degree of peptization of a gel and the time of settling, methods of
estimating the colloid content of soils based on such procedures 7 are
necessarily inaccurate. Other methods that have been employed are
based on determination of the adsorption capacity of the soil for mala-
chite green, 8 water, and ammonia. Gile 9 and his coworkers deter-
mined the adsorption capacity of a sample of soil and of the colloidal
material extracted from the soil, and from these data calculated the
percentage colloidal matter. Aftei correcting for the possible altera-
tion in adsorptive capacity of the colloid produced by extraction, the
percentages of colloidal matter indicated by adsorption of malachite
green, water, and ammonia showed fairly good agreement among them-
selves l and with the percentages estimated gravimetrically and micro-
scopically. Rouyoucos 11 estimated rapidly the colloidal material in a
soil suspension by means of a specially made hydrometer. As would
be expected, the colloidal content of different soils varied widely. As-
suming that all particles less than 1 /* in diameter are colloidal, the
sandy soils contain but a few per cent of colloids, whereas the loam
soils may contain 15-25%, and the clayey soils 40 to SO and up to
colloidal matter.
Extraction of the Colloidal Fraction
The separation in quantity of the colloidal fraction of soils was
made possible by the development of continuous-flow centrifuges
capable of developing a centrifugal force of 40,000 times gravity. The
method of procedure developed independently by Moore, Fry, and
Middleton 12 in the Bureau of Soils of the U. S. Department of Agri-
culture and by Bradfield 13 is essentially as follows . 14 The soil, prefer-
6 Am. J Sci, (3) 6, 288, 333 (1873); "Soils," New York, 333 (1919)
Forschr Gcbietc Agnkultur-Physik, 18, 225 (1895).
7 Scales and Marsh. Ind. Eng. Chem, 14, 52 (1922).
s Ashley U S. Geol Survey, Bull, 388, 65 (1909).
9 Gile, Middleton, Robinson, Fry, and Anderson : U. S. Dept. Agr. Bull.
1193 (1924).
ioC7 Davis. J. Am. Soc Agron., 17, 277 (1925).
*iSoil Sci., 23, 319 (1927), 26, 365, 473; 26, 233 (1928); cf , however,
Slegcl: J. Am. Ceram. Soc., 11, 185 (1928).
"Ind Eng. Chem., 13, 527 (1921).
is Mo Agr. Expt. Sta. Research Bull. 60 (1923).
i* Bradfield- Alexander's "Colloid Chemistry," 3, 569 (1931).
ANALYSIS OF THE COLLOIDAL FRACTION
403
ably fresh from the field, is agitated vigorously with 3 to 5 times its
weight of water for several hours and allowed to settle from 1 to 10
days. The suspended material is then siphoned off and passed through
the centrifuge at a rate sufficient to deposit all the non-colloidal ma-
terial in the bowl. The colloidal fraction is concentrated by passing
through the centrifuge at a slower rate 14 or by the aid of a Pasteur-
Chamberlain filter. Under favorable conditions the sol prepared in this
way is stable almost indefinitely. The colloidal matter dries in the air
to a hard, brittle mass which consists of secondary aggregates that may
be as difficult to disperse as the original sol. The binding power of
the clay when mixed with sand, molded into briquettes under high
pressure, and dried, is greater than that of Portland cement.
Analysis of the Colloidal Fraction
Before Electrodialysis. The chemical composition of the colloidal
fraction of soils seems to depend more on the environment under
which it is found than on the nature of the parent material. This is
illustrated by Table L which gives Bradfield's 14 analyses of the col-
TABLE L
CHEMICAL COMPOSITION OF COLLOIDAL CLAY
Memphis
Robert s-
ville
Q
Boone C
i*
Wabash
Sharkey
1
1*
Manon
Cherokee
Average
H 2
105 C
10 60
10 86
11 28
12 30
12 66
12 58
12 8?
12 46
12 17
12 44
10 69
11 90
Volatile
matter
11 76
13 38
12 47
11 88
11 34
11 16
11 51
12 43
13 37
11 83
14 09
12 29
SiO 2
50 00
49 81
48 77
SO 54
51 34
SO 11
52 78
51 18
50 22
52 15
48 79
50 11
A1 2 8
28 21
29 23
29 90
28 86
28 00
24 51
24 19
25 59
27 65
26 51
30 42
27 55
PeaOa
3 53
3 72
4 05
4 45
4 53
3 39
4 73
5 17
4 74
5 51
3 80
4 33
CaO
1 14
94
23
73
29
1 75
1 23
99
92
44
35
82
MgO
1 84
1 51
1 37
1 42
1 54
2 41
2 56
1 80
1 52
1 7S
1 29
1 73
K 2
1 68
1 25
80
1 01
1 25
1 87
1 63
1 34
85
1 06
1 03
1 2S
Na 2 O
600
000
120
071
301
602
854
318
449
302
S5S
379
MnO 2
056
018
007
028
021
027
023
025
014
018
017
023
loidal clay extracted from Missouri soils of diversified types with re-
spect to origin, topography, age, and source of parent material. These
clays show great uniformity in their content of SiO 2 , A1 2 O 3 , Fe 2 O 3 ,
and combined water. In contrast, Robinson and Holmes, 15 who ana-
lyzed the colloidal fraction from 45 different soils from various parts
15 U. S. Dept. of Agr. Bull. 1311 (1924).
404
THE INORGANIC SOIL COLLOIDS
of the United States, observed a much wider variation in composition
as follows: SiO 2 , 31.84 to 55.44%; A1 2 O 3 , 16.42 to 38.28%; Fe 2 O 3 ,
4.66 to 16.67% ; and combined water, 3.33 to 16.56%.
After Electrodialysis. Reference to Table L discloses that various
amounts of alkali and alkaline-earth metals are present in the colloidal
clays, the total amount on the average for these clays being almost one
equivalent for every molecule of ammonia. In the presence of water
the finely divided clay hydrolyzes to a certain extent, hydrogen ions
from water replacing the alkali or alkaline-earth cations. Under nat-
ural conditions, this process is favored by the presence of carbonic
acid. In the laboratory, the replacement process can be accomplished
TABLE LI
ANALYSES OF ELECTRODIALYZED COLLOIDAL MATERIAL FROM A SERIES OF CLAYS
Putnam
Putnam
+ H 2 2
Susque-
hanna
Boone
Sharkcy
Rock
River
Ben-
tonite
Chey-
enne
Ben-
tonite
Finest
Chey-
enne
Ben-
tonite
Loss on ignition
17.75
17 60
19 35
17 78
17 31
16 36
17 97
18 94
SiO 2
47 84
47 86
45 66
45 73
50 58
57 01
57 15
56 25
A1 2 0,
23 75
23 11
22 68
23 57
20 21
20 58
18 37
19 03
Fe 2 8
7 62
8 25
8 14
9 90
6 98
3 61
3 00
2 97
FeO
29
08
31
23
31
09
09
12
MnO .
02
01
02
03
01
00
00
TiO a
52
52
1 02
48
54
12
29
22
CaO
11
08
0.10
08
10
11
14
08
MgO
1 49
1 22
1 95
1 14
2 03
1 94
2 86
2 59
K a O
1 16
1 20
70
72
1 44
02
02
02
Na 2 0.. . .
07
05
05
10
11
02
00
02
c
977
184
667
566
591
055
198
073
P .
04
13
07
Trace
Trace
Trace
Trace
Trace
S ..
014
010
014
015
025
019
005
Oil
Sum* ....
100 67
100 11
100 05
99 77
99 67
99 89
99 90
100 25
SiOs/AUO,..
3 42
3 52
3 41
3 28
4 24
5 30
5 28
5 01
f
104 9
91
117 6
78 4
139 3
102
148 4
133 4
t
70
75.
80
70
80
90
100
110
40.0
45 2
42 6
47
36 5
46 8
40 3
45
* Carbon is included in loss on ignition.
t Non-exchangeable bases. Milliequivalents.
I Exchangeable cations (approximate). Milliequivalents.
8 [Exchangeable/ (exchangeable + non-exchangeable)] X 100.
MINERAL CONSTITUENTS 405
much more rapidly and completely by electrodialysis 18 which removes
the replaced cations continuously. The colloidal clay is placed in the
middle section of a three-compartment cell between membranes which
are permeable to the ions hydrolyzed off the particles but impermeable
to the colloidal clay ions themselves. Distilled water is flowed slowly
through the outside compartments in which are suspended suitable elec-
trodes (platinum for anode and nickel for cathode) and a potential of
100-200 volts is applied. The cations appear in the form of their hy-
droxides in the cathode chamber, and the anions, which are the clay
micelles, migrate to the anode membrane and accumulate there; any
soluble acid which can pass the membrane is removed.
Only a part of the total bases, usually from 25 to 50% f are removed
by electrodialysis. This is shown by Schollenberger's analyses of clays
electrodialyzcd by Bradfield, Table LI. It will be seen that the calcium
and sodium ions are almost completely removed from all samples
whereas the potassium and magnesium are more firmly held (cf. p.
393). If a clay from which no more base can be removed by electro-
dialysis is allowed to deposit on the anode membrane, the water in
which it was suspended will be neutral whereas the clay paste will have
a pR value of 2.0 to 3.5.
CONSTITUTION OF COLLOIDAL CLAY
Mineral Constituents
Clays are secondary products formed as a rule by the weathering
action of water, carbonic acid, etc., on feldspar and feldspathic min-
erals. In view of the manner in which the clays are formed and laid
down, it is not surprising that they should show considerable variation
in composition. The chemical composition being known, the question
naturally arises as to the constitution of the clays. Of the several
possibilities, the bulk of the evidence supports the view that the col-
loidal fraction of soils consists of new secondary minerals with which
may be mixed small amounts of the separate oxides or the primary
minerals. Another theory which used to be regarded with consider-
able favor assumes that the weathering influences transform the hy-
drated silicates into the hydrous oxides of iron, aluminum, and silicon
together with soluble salts of sodium, potassium, and calcium which
are adsorbed in part by the hydrous oxide gels. A third possibility is
that the clay consists merely of colloidal fragments of the coarser min-
eral from which it was derived.
"Bradfield: Proc. 1st Intern. Congr. Soil Sci, 2, 264 (1927).
406 THE INORGANIC SOIL COLLOIDS
The strongest evidence against the view that the colloidal fraction
of soils consists of mixtures of the hydrous oxides of iron, aluminum,
and silicon with adsorbed salts, or of fragments of the parent minerals,
is furnished by x-ray diffraction analysis of clays by Wherry, Ross,
and Kerr, 17 Hendricks and Fry, 18 Pauling, 19 Kelley, Dore, and Brown, 20
Hofmann, Endell, and Wilm, 21 and others. 22 Although the diffraction
patterns are not always as sharp as they should be for precise analysis,
they reveal that the bulk of the material in the soil colloids is crystal-
line, the most common constituents being secondary "clay minerals" of
two types as follows: (1) the halloysite type which includes halloysitc
(Al 2 O 3 -2SiO 2 4H 2 O), 23 metahalloysite, 23 kaolinite, 24 nacrite, 25 and
dickite; 2fl (2) the prophyllite type which includes prophyllite (A1 2 O 3 -
4SiO 2 H 2 O) and talc, montmorillonite (MgO-Al 2 O 3 5SiO 2 wH 2 O,
the major constituent of most bentonites), beidellite (A1 2 O 3 3SiO 2 -
H 2 O, which gives the same x-ray diffraction pattern as montmoril-
lonite 21 ), and nontronite 27 (Fe 2 O 3 3SiO 2 H 2 O). Notable excep-
tions are the laterites which contain, in addition to aluminosilicates,
hydrous a-A! 2 O 3 - H 2 O (diaspore) and -Fe 2 O 3 H 2 O (gothite) ; and
certain bauxite clays which contain hydrous y-A! 2 O 3 H 2 O. The pri-
mary soil minerals such as quartz, micas, and feldspars are present
only in very small amounts. This indicates strongly that the soil col-
loids do not result from mere reduction in size of the original mineral
with alterations caused by surface weathering, but from a secondary
synthesis of the degradation products of the primary minerals.
The absence of the primary minerals in the colloidal state is prob-
ably attributable to their instability in the presence of water, carbonic
acid, and air. Formed under conditions of high temperature and pres-
17 Colloid Symposium Monograph, 7, 191 (1930).
"Soil Sn, 29, 457 (1930)
"Proc Natl Acad Sci U. S, 16, 123, 578 (1930)
20 Soil Sci, 31, 25 (1931)
21 Z angew Chem., 47, 539 (1934); cf Hofmann and Bilkc. Kolloid-Z,
77, 238 (1936)
22 Cf Jacob, Hofmann, Loofmann, and Maegdefrau Bciheftc Z Ver
deut Chem, 21, 11 (1935); Trans Intern Congr Soil Sci, 3rd Congr, Oxford,
1, 88 (1935); Dubrisay and Trillat- Rev. gen. colloides, 8, 1 (1930); for a
summary see Marshall* Z. Krist, 91, 433 (1935); J Phys Chem, 41, 935
(1937).
"Mehmel- Z Krist., 90, 35 (1935).
2 *Gruner: Z Krist, 83, 75 (1932)
"Gruner: Z Krist, 85, 345 (1933).
2 Gruner. Z. Krist, 83, 394 (1932).
"Gruner- Am. Mineral, 20, 475 (1935).
WATER 407
sure, 28 they break down when exposed to weathering agents, forming
new compounds which are more stable under the new conditions. The
rate at which these processes go on increases so rapidly with decreasing
particle size that a particle of colloidal dimensions, below 100 m/u, can
exist as a primary mineral for a short time only. Since the rate at
which large particles are reduced to colloidal size is slower than the
rate at which the colloidal particles of the primary minerals are de-
stroyed, it follows that the modal fraction of the size- frequency curve
of the primary minerals will tend to remain well above the colloidal
range. On the other hand, the modal fraction of the clay minerals
must he within the colloidal range since they are seldom present in the
fine sand or silt fractions.
Further evidence against the hypothesis that colloidal clays are mix-
tures of hydrous oxides with adsorbed salts is the failure to prepare
synthetic clays which possess more than a superficial resemblance to
natural clays. To illustrate, the permutites which have base-exchange
properties analogous in certain respects to clays and which have been
suggested for use as "models," 29 differ from clays in many respects :
As we have seen, the clays give an x-ray diffraction pattern whereas
the permutites are less highly hydrated and are almost amorphous to
x-rays. Moreover, practically all the cations of permutites are replaced
at ordinary temperatures by the cations of neutral salt solutions
whereas only 30-50% of the cations of clays are readily replaced by
treating with neutral salts. Also, an electrodialyzed clay is a much
stronger acid with a much greater capacity to absorb bases than an
electrodialyzed permutite. Finally, the adaptation of mineralogical
methods to the determination of mean refractive index, mean density,
and double refraction of particles smaller than 100 m/*, ao and recent
investigations of the base-exchange phenomenon in clays 31 (p. 391),
furnish strong confirmatory evidence in support of the view that the
colloidal fraction of soils consists essentially of aluminosilicates rather
than of varying mixtures of the hydrous oxides.
Water
The water content of soil colloids is either adsorbed water or water
of crystallization, that is, contains OH~ ions as a part of a crystal
2867 Ewell and Insley J Research Natl. Bur. Standards, 15, 173 (1935).
20 CY Wiegner: J Landw., 60, 110, 197 (1912).
so Marshall. Trans Faraday Soc, 26, 173 (1930); Z. Krist., 90, 34 (1935);
J. Phys Chem., 41, 935 (1937).
*"Cf., for example, Kelley and Jenny: Soil Sci., 41, 367 (1936).
408
THE INORGANIC SOIL COLLOIDS
lattice structure. Drying air-dried samples at 1 10 results in a loss of
8-12% of adsorbed water depending on the vapor pressure and tem-
perature at which they are air dried. This adsorption is almost com-
pletely reversible even when the dehydration is accomplished at higher
temperatures. Indeed, Kelley, Dore, and Brown 20 found that ben-
tonite clays can be heated to 350 without any appreciable change in
their base-exchange capacity or x-ray diffraction pattern; and Kelley,
Jenny, and Brown 82 called "adsorbed water" that proportion of the
total water which comes off at 400 from such clays, provided that
eReddng(a)
o Redding (6)
o Sierra
Cecil
squehanna A
Susquehanna B
Susquehanna C
Susquehanna D
200 400 600
Temperature, Degrees C.
200 400 600
Temperature, Degrees C.
FIG 66. Dehydration curves for typical soil colloids.
there is no evidence of the loss of crystal water at this temperature
Heating from 400 to 700 causes an additional loss of water with a
complete and irreversible loss of colloidal properties 3S including base-
exchange capacity. 84
Kelley, Jenny, and Brown compared the dehydration isobars for a
number of minerals of known structure with those for typical soil col-
loids. Some isobars for soil colloids are given in Fig. 66A and B.
32 Soil Sci. f 41,259 (1936).
"Brown and Montgomery: Bur. Standards Tech. Paper 21 (1913).
84 Cf. the behavior of zeolites, Milligan and Weiser: J, Phys. Chem., 41,
1029 (1937).
WATER
409
From such curves the amount of adsorbed and crystal lattice water in
several clays is deduced as given in Table LI I. Unlike minerals of
known structure the soil colloids lose their lattice water at lower tem-
peratures, but it is not certain whether this is caused by particle size
alone, or by structural differences, or by both. It seems most likely
that both factors come in since the soil colloids are, in general, some-
what different from the pure clay minerals. Thus, of the two major
classes of soil colloids the first resemble in some measure kaolinite and
TABLE LII
ADSORBED AND CRYSTAL LATTICE WATER IN SOIL COLLOIDS J2
(Anhydrous basis)
Water in per cent
Soil colloid
Total
Adsorbed
Crystal lattice
Cecil . ...
18 04
4 74
13 30
Sierra
18 19
4 81
13 38
Redding . .
19 40
8 01
11 39
San Joaquin . .
14 54
5 41
9 13
Placentia .
18 03
5 74
12 29
Yolo .
21 83
12 12
9 67
Susquehanna A
19 12
10 61
8 51
Susquehanna B
21 07
14 01
7 06
Susquehanna C (10-20 in.) . .
20 75
13 39
7 36
Susquehanna D (10ft.).
21 08
14 12
6 96
Susquehanna E (11 ft.)
29 29
23 61
5 68
Putnam .
15 30
7 25
8 05
halloysite and the second appear to be related to but are not identical
with beidellite (p. 406). From a comparison of dehydration isobars
for the soil colloids and known minerals, Kelley, Jenny, and Brown
deduce that the composition of the surfaces of the former is similar
in all cases, consisting of Si-O-Si planes and possibly OH planes on
which the water is adsorbed. 85 The possibility of some so-called
broken-bond adsorption on linked tetrahedra surfaces is not excluded,
but such surfaces are largely absent in the soil colloids.
The addition of water to a colloidal clay dried at 105 is accom-
panied by a considerable heat of adsorption. Bouyoucos 8e proposed
ss Cf. Marshall: J. Phys. Chem., 41, 935 (1937).
36 Science, 60, 320 (1924).
410 THE INORGANIC SOIL COLLOIDS
the phenomenon of heat of wetting as a means of estimating the col-
loidal content of the soil. As would be expected, the colloids of dif-
ferent soils vary widely in their heat of wetting, owing to the difference
in their physical character. Heating to 750 is said to decrease the
heat of wetting of soils to zero ; but this cannot be strictly true, since
the adsorptive capacity of ignited soils may be 30 to 50% of the value
before ignition. Because of this loss of adsorptive capacity on ignition,
Alway 87 questions the reliability of the water-adsorption method of
estimating the colloid content of the soil. This seems to be beside
the point, since one might reasonably expect the coalescence accom-
panying ignition to decrease materially the amount of colloidal matter.
The validity of the water-adsorption method depends primarily on
whether non-colloidal material in unheated soil adsorbs an appreciable
amount of water under the conditions of determination.
The total water-holding capacity of a soil is influenced to a con-
siderable extent by the height of the soil column and by the mode of
packing of the particles; but the colloidal content is by far the most
important factor in determining the moisture-holding capacity. Bou-
youcos 38 found that some ordinary clays will hold as much as 7S%
water as compared to only 20% in some coarse sands.
Not only do the colloidal particles adsorb and conserve water for
times of drouth, but the freezing point of water is lowered very appre-
ciably when it is adsorbed. 39 As in the case of the hydrous oxide gels, 40
a part of the adsorbed water is not frozen until the temperature is
reduced several degrees below zero. This is doubtless of importance
in preventing complete desiccation of the soil by freezing and the con-
sequent destruction of the soil bacteria.
PHYSICAL PROPERTIES OF COLLOIDAL CLAY
General Properties
The Silica-Sesquioxide Ratio. In Table LIII 14 are summarized
the general properties of a few typical clays investigated by Anderson
37 Colloid Symposium Monograph, 3, 241 (1925) ; Puri, Crowther, and
Keen: J. Agr. Sci, 16, 68 (1925).
8 Colloid Symposium Monograph, 2, 132 (1924); cf. King: Wis. Agr.
Expt. Sta., Sixth Kept, 189 (1889); Alway and McDole: J. Agr. Research,
9, 27 (1917).
30 Bouyoucos and McCool: Mich. Agr. Expt. Sta., Tech. Bull., 31 (1916);
36 (1917); Parker: J. Am. Chem. Soc., 48, 1011 (1921).
<Foote and Saxton: J Am. Chem. Soc., 88, 588 (1916).
GENERAL PROPERTIES
411
TABLE LIII
SUMMARY OF PROPERTIES OF A SERIES OF COLLOID \L CLAYS
Kind of clay
Property
Susque-
Sassa-
Pallon
Sharkey
Marshall
hanna
fras
Norfolk
Aragon
Molecular ratio SiOg/AzOa
3 62
3 11
2 73
1 99
1 85
1 60
55
Sp gr. in H 2 O
2 766
2 718
2 627
2 715
2 748
2 708
Av. diam. of particles ran
102
91
106
141
128
129
Surface M/g ...
21 3
24 2
21 5
15 7
17
17 1
No. of particles/g X 10 - ...
680
960
613
263
335
322
.
Heat of wetting, cal /g ...
17 5
16 3
14 6
5 3
9 8
7 6
8
H 2 O absorbed over 30% H 2 SO 4 g/g
178
160
130
052
114
084
089
Moisture equivalent %
120
94
72
67
62
62
(Bcn-
tomte)
Vol. in H 2 O cc/g
2 5
1 9
1 8
2
1 8
1 8
9
Relative vis. of 2% sols
1 33
1 14
1 23
1 27
1 33
1 12
Exchbl Ca+ Mg+ Na+ K, m.eq./g.
1 08
69
53
09
23
09
pH .
8 2
6 8
7 1
5 6
5 7
5 3
Organic matter % . ...
1 79
3 83
7 94
1 86
1 88
2 18
5 96
and Mattson. 41 As a result of numerous observations, 42 it has been
found that a number of physical properties of clays may be correlated
with the contents of the major constituents as expressed by the ratio
of silica to alumina plus ferric oxide. Thus, reference to Table LIII
shows a significant correlation between this ratio and the heat of
wetting of a series of oven-dried clays. Moreover, swelling, viscosity,
dispersibility, heat of wetting, adsorption of bases and basic dyes, and
base exchange are all manifested to a greater extent by clays with a
high silica-sesquioxide ratio than by those in which this ratio is low. 41 ' 48
Clays with a high ratio are also more electronegative, as indicated by
the quantity of aluminum chloride or basic dye solution required to
neutralize the negative charge on the particles. 44 The particles with a
high proportion of silica and bases remain electronegative in acid as
well as in neutral and alkaline solutions, whereas those with a high
proportion of sesquioxides are amphoteric and become positive in acid
solutions. In accord with this behavior, clays high in sesquioxides
U. S. Dept. Agr., Bull. 1452 (1926).
C/. Anderson: J. Agr. Research, 28, 927 (1924); Gile, et at : U. S. Dept
Agr., Bull. 1193; Robinson and Holmes: 1131 (1924); Anderson and Mattson:
1452 (1926) ; Science, 62, 114 (1925).
Matron J. Am. Soc. Agron., 18, 458, 510 (1926); Proc 1st Intern.
Congr. Soil Sci., 2, 185 (1927).
"Mattson: J. Phys. Chem., 82, 1532 (1928).
412 THE INORGANIC SOIL COLLOIDS
adsorb anions such as chloride and sulfate from acid but not from neu-
tral solutions. 45 Clays with a high proportion of silica give stronger
acids when saturated with hydrogen ions. 48
Although the above-mentioned observations indicate the existence
of a relationship between the colloidal behavior of clays and their
silica-sesquioxide ratio, it should be emphasized that this is only quali-
tative and that the physical-chemical constants of a clay cannot be pre-
dicted with any degree of accuracy from the observed ratio. This is
readily understood when it is recalled that in certain clays all the silica
and alumina may be present in an aluminosilicate combination and in
others a part may exist as free hydrous alumina or silica having dif-
ferent properties.
The relationship between composition and colloidal properties is
well illustrated by the varying behavior of the clay soils of the humid,
temperate regions and those of the tropics. A soil of the first type
containing 50% or more of clay drains slowly and cannot be plowed
without "puddling" for several days after a heavy rain, whereas a red
soil of the second type with a similar content of colloidal matter can be
plowed within a few hours after a tropical rainstorm. The non-plastic
tropical clays are formed under conditions of high temperature and
heavy rainfall accompanied by luxuriant plant growth, all of which
intensify the weathering processes. This may result in the removal of
most of the silica, giving a soil consisting largely of the hydrous sesqui-
oxides of iron and aluminum. Because of their brick-red color such
soils have been called laterites.
Exchangeable Cations. In the preceding chapter, brief attention
was given to the importance of the nature of the exchangeable cations
on the properties of soils (p. 399). The undesirable acid soils of the
humid regions possess a low "degree of saturation with bases," and the
poor alkali soils of the arid regions possess a high content of exchange-
able sodium. Between these two extremes are found the most desir-
able soils, those which are comparatively rich in exchangeable calcium
and usually contain a reserve of calcium carbonate so that they remain
largely saturated with calcium.
The saturation capacity of a soil is defined as the sum of the ex-
changeable bases and exchangeable hydrogen. The acid soils of the
humid region have resulted from the gradual displacement of more or
less of the desirable calcium by hydrogen, and the alkali soils of the
arid regions from the displacement of calcium by sodium. The low
*5Mattson: Proc. 1st Intern. Congr. Soil Sci, 2, 199 (1927).
"Baver and Scarseth: Soil Sci., 31, 159 (1931).
GENERAL PROPERTIES 413
productivity of the so-called alkali soils of western United States is
undoubtedly connected with the high proportion of exchangeable so-
dium in the colloid fraction, which renders the clay more highly dis-
persed, more slowly permeable, and, in extreme cases, almost imper-
meable to water and air. As Bradfield 47 points out, the removal of
soluble salts by leaching with water is not sufficient to restore such
soils to productivity; on the contrary, if the soil contains no calcium
or magnesium carbonate, such leaching may make conditions worse.
To regain the normal physical conditions of such soils, the essential
thing is to restore the normal calcium-sodium ratio. If the soil contains
a large reserve of calcium carbonate, irrigation with adequate drainage
will eventually bring about normal conditions. 48 If calcium must be
added, time will be saved by using the more soluble gypsum in place of
calcium carbonate. In a soil containing some calcium carbonate, the
most economical way to restore the normal calcium-sodium ratio con-
sists in adding sulfur, which is oxidized to sulfate by soil organisms.
Acid soils must be treated with lime or calcium carbonate.
Because of the importance of the concept "degree of saturation
with bases" for the characterization of soils, many methods have been
proposed for determining this value. The results vary with the method
of procedure and are not always comparable. This raises the question
as to what is meant by a base-saturated soil. Bradfield 49 answers the
question in a logical way by defining a soil saturated with bases as one
which has reached equilibrium with a surplus of calcium carbonate and
the partial pressure of carbon dioxide existing in the atmosphere, and
at a temperature of 25.
In support of the above definition Bradfield showed: (1) that the
amount of calcium taken up is independent of the amount of excess
added; (2) that the amount adsorbed is the same regardless of the
direction or method by which the equilibrium is approached; and (3)
that the total amount of base adsorbed is practically independent of
the amount and nature of the bases originally present. Moreover, the
proposed saturation point represents a natural transition point in physi-
cal and chemical properties which are of particular significance in soil
science. Thus clays saturated with calcium coagulate in the form of
large stable granules which give calcareous soils their well-known im-
proved structure. As soon as excess calcium carbonate is removed
the percolating waters charged with carbonic acid bring about a re-
Alexander's "Colloid Chemistry," 3, 587 (1931)
**Kelley and Brown: Soil Sci, 20, 477 (1925).
*Proc. 2nd Intern. Soc. Soil Sci., A, 63 (1933).
414 THE INORGANIC SOIL COLLOIDS
placement of calcium ions with hydrogen ions, the initial step in soil
degradation.
Plasticity
Plasticity is defined as the property of a material which enables it
to change its shape without cracking when subjected to pressure, the
new shape being retained when the deforming stress is removed. Fine
dry sand is not plastic, but it becomes plastic when moist since the
solid adsorbs the liquid giving thin films which bind the particles to-
gether while still allowing them to move relatively to one another.
Moreover, the tendency of liquid surfaces to coalesce causes any break
to heal. Hence an adsorbed liquid film may be all that is necessary to
render a mass plastic in the ordinary sense.
In the clay industry, the term plasticity connotes not only that the
clay can be molded under pressure but also that it will harden under
the application of heat, forming a rock-like mass. From this point of
view wet sand alone is not plastic but requires the presence of some
binding material of a gelatinous character. Rohland 60 considers plas-
tic clay to consist of very minute, non-plastic grains or cores sur-
rounded by gelatinous films of material which are saturated with water
when the maximum plasticity is developed. With an excess of water
above that for maximum plasticity, the clay becomes "sticky"; and
with a still larger excess, a non-plastic sol or clay "slip" is formed. If
the clay is dried the gelatinous matter shrinks, becomes hard and horny,
and the plasticity disappears only to reappear when the clay is moist-
ened again.
The gelatinous binding material in the colloidal fraction of soils is
chiefly hydrous aluminosilicates together with more or less hydrous
alumina and silica. Since such substances tend to lose their power of
taking up water and re-forming a gel after they have once thoroughly
dried out, it is necessary to postulate the presence of something which
prevents the gel-forming process from becoming non-reversible. Ban-
croft and Jenks 81 give experimental evidence to support the view that
a gelatinous film can be produced and maintained by the presence to-
gether of at least one salt which tends to peptize the clay or some por-
tion of the clay and of at least one salt which tends to coagulate the
same portion of the clay.
If one accepts the hypothesis that the gelatinous character of a clay
results from aging in the presence of both peptizing and flocculating
eo Z. anorg. Chetn., 31, 158 (1902).
wj. Phys. Chem., 29, 1215 (1925).
PLASTICITY 415
agents, it follows that a non-plastic clay is one which contains but little
salts or which has not developed plasticity because of too uniform
conditions. The "rotting" of dug clay which increases its plasticity is
probably due to the moisture and temperature variations which in-
crease the amount of gelatinous material.
Since a gelatinous body consists of very finely divided particles that
have adsorbed water strongly, it follows that a mixture of hydrous
oxides or a hydrous aluminum silicate will continue to impart plasticity
to a mass, even after drying repeatedly, provided the primary particles
are prevented from coalescing or from agglomerating into dense ag-
gregates. As already noted, such coalescence or agglomeration is in-
hibited by a suitable peptizing agent. One would expect a protecting
colloid to exert a similar influence. Since humus doubtless possesses
the properties both of a peptizing agent and a protecting colloid, it is
probable that it plays an important role in the development and mainte-
nance of plasticity in many clays. 82 Ries 83 showed that the addition
of a \% solution of tannin to a clay noticeably increased the plasticity
and at the same time deflocculated or peptized the larger aggregates;
in addition, the tensile strength of the clay was doubled by this treat-
ment. Acheson 4 increased the plasticity of clays by adding tannin
and alkalis until the hydrous mass was completely deflocculated, after
which sufficient acid was added to coagulate the colloids to a pasty
mass. Everyone knows how Pharaoh increased the burdens of the
Israelites by withholding from them the straw which they used in mak-
ing bricks. It is possible, but not proved, that the straw served as a
source of tannin rather than as a binding material and that the bur-
den imposed consisted in their having to make bricks with less plastic
material.
Since, from the above point of view, the important thing for plas-
ticity is relatively fine particles coated with a film of colloidal matter
which adsorbs water strongly, it should be possible to increase the
plasticity by grinding. It is very difficult indeed to grind particles to
ultramicroscopic dimensions; nevertheless, the plasticity of certain
clays both soft and hard has been increased appreciably by prolonged
grinding. 05 Johnson and Blake 58 claim to have made a non-plastic
Cf. Keppeler: Chem.-Ztg, 86, 884 (1912).
5s Trans. Am. Ceram. Soc., 6, 44 (1904).
"Trans. Am Ceram. Soc., 6, 31 (1904).
55 Searle: Brit. Assoc. Advancement Sci., 3rd Rept. on Colloid Chem., 131
(1920); Walker: J. Am. Ceram. Soc, 10, 449 (1927).
"Am. J. Sci., (2) 98, 351 (1867).
416 THE INORGANIC SOIL COLLOIDS
china clay plastic by this means ; but they attributed this to the flatten-
ing out of the particles rather than to disintegration.
People who do not regard the presence of suitable colloidal matter
as a necessary criterion for the plasticity of clays, as the ceramist uses
the term, usually attribute the phenomenon to some purely chemical
characteristics of the molecules of a clay substance, 57 or to plate-like
structure of the particles. 88 - 56 The evidence for the existence of a
true clay is lacking; but one may be justified in attributing plasticity
solely to the plate-like or lamellar structure which the particles possess.
Bradfield B0 sees no reason why it is necessary to postulate an ex-
traneous colloidal material to account for the plasticity of clays con-
taining from 40 to 60% colloidal clay which is known to consist of
highly hydrated plate-shaped crystals having a variable spacing be-
tween certain groups of sheets. Bradfield compares the difficulty of
separating two glass plates with a film of water between them with
that of two spheres in contact with each other at a single point and
with an annular ring of water around the point of contact; and con-
cludes that the plate system if reduced to colloidal dimensions would
probably give a much more plastic mass than a system of spheres.
Certainly the shape of the particles of colloidal clay would be expected
to influence the plasticity of clay, but it is still an open question whether
the shape of the hydrated particles is alone sufficient to produce the
phenomenon.
Flocculation of Clay Sols
Clay sols are negatively charged and are usually flocculated by low
concentrations of neutral salts. The valency rule may apply fairly well
at least for certain sols. Hall and Mouson, 60 for example, determined
the precipitation values of various chlorides, sul fates, and nitrates on
a clay sol and found the order of cations to be : H, Al > Ca, Ba, Mg>
K > Na ; and the stabilizing power of the anions to be : OH > SO 4 >
NO 3 > Cl. 61 Bradfield 62 showed, however, that the valency rule can
have no general validity since the flocculation value may vary widely
5T Cf Asch and Asch "The Silicates in Chemistry and Commerce/ 1 London
68 Van Bemmelen's "Godenboek," 163 (1910); Biedermann and Herzfeld:
Rics' "Clays," 123 (1914); Cook: N. J. Geol. Survey, 287 (1878); Vogt:
Compt. rend., 110, 1199 (1899) ; Haworth: Mo. Geol. Survey, 11, 104; Wheeler:
106 (1896).
69 Private communication.
60 J. Agr. Sci., 2,251 (1907).
61 C/. Kermack and Williamson: Proc. Roy. Soc. Edinburgh, 46, 59 (1925).
62 Soil Sci., 17,411 (1924).
FLOCCULATION OF CLAY SOLS 417
with slight changes in (1) hydrogen ion concentration, (2) concentra-
tion of sol, and (3) nature of the exchangeable cations. The important
findings of Bradfield have been confirmed and extended by Wiegner, 68
Baver, 64 Mattson, 65 and especially by Jenny and Reitemeier, 66 whose
work will be considered in some detail.
^-potential, Exchange, and Flocculation Data for Clay Sols. The
migration velocity, /i, of the particles of various clay sols prepared
from natural Putnam clays (cf. p. 391) was measured with an ultra-
microscope in an open cataphoresis cell and the -potential in millivolts
calculated from the average of a number of measurements, using the
Freundlich equation (p. 396). The percentage release of the cations
from the multivalent clays was determined after treating with potas-
sium chloride, and that of the univalent clays was calculated from ex-
change values with NH 4 -clay. Since potassium and ammonium ions
are equally well adsorbed and released, the data are comparable quanti-
tatively. Flocculation values were obtained both from KC1 and for
MCI where M is the cation common to the clay. The data are given in
Table LIV. The amount of electrolyte added to the system is ex-
pressed in terms of "symmetry values, S," that is, in multiples of the
number of milliequivalents of adsorbed ions in the system. The mag-
nitude of the exchange expressed in percentage of the symmetry con-
centration equal to one (S = 1) is called the symmetry value.
Effect of the Charge and Size of Adsorbed Cations on the -poten-
tial. The influence of charge and size of the adsorbed cations on the
f -potential of clay particles is shown graphically in Fig. 67 from Jenny
and Reitemeier's data (Table LTV). Comparing rare gas type ions of
equal size, the potential tends to be lower as the valence of the ion is
higher. The comparison must be restricted to ions of equal size since
certain uni-clays (e g, Cs-clay) have lower potentials than some of the
bi-clays (eg., Mg-clay). Comparing rare gas type ions of equal val-
ency, the -potential is higher the smaller the adsorbed cation. This
agrees with Wiegner's view that the most highly hydrated ions (in
general small ions are strongly hydrated) cause the highest ^-potential.
^-potential and Ionic Exchange. Jenny visualizes clay particles as
plate-shaped crystals which hold adsorbed ions on their surfaces. Be-
cause of heat motion and Brownian movement, the ions are not at rest
but oscillate (p. 397) and, at some times, may be a considerable dis-
3Kolloid-Z. (Zsigmondy Festschrift), 86, 341 (1925).
* Baver. Mo. Agr. Expt. Sta. Research Bull 129 (1929)
'a Mattson: Soil Sci, 28, 179, 221 (1929).
J. Phys. Chem., 89, 593 (1935).
418
THE INORGANIC SOIL COLLOIDS
tance from the wall. The cation of an added electrolyte may slip
between the negative wall and the positive oscillating ion, the former
becoming adsorbed and the latter remaining in solution as an ex-
TABLE LIV
f-POTENTIALS, EXCHANGE ADSORPTION, AND FLOCCULATION OF CLAYS
Flocculation values
Sol
/*
Temp.
f
(milli-
%-release
(symmetry
5-concentrations
C
volts)
values)
KC1
MCI
Li-clay
3 45
30 3
58 8
68
21 6
26
Na-clay
3 41
28 9
57 6
66 5
11 2
14 8
K-clay.
56 4
48 7
7 8
7 8
NH 4 -clay
3 48
33 2
56
50
5 4
4 9
Rb-clay
3 25
30 3
54 9
37 4
Cs-clay
3.02
30 3
51 2
31 2
5 6
4 2
H-clay
2 84
30 3
48 4
14.5
1 5
36
Mg-clay
3 18
30 3
53 9
31 32
2 9
69
Ca-clay
3 27
32 8
52 6
28 80
3
55
Sr-clay
3 06
30 3
51 8
25 76
2 6
Ba-clay . .
3 01
30 3
50 8
26 75
2 3
La-clay
2.74
21 2
45 5
13 96
86
{50 4
Th-clay
3 11
32 2
47 1
1 85
0.60
M.B.*-clay
2 57
33 7
40 5
* Methylene blue
1.0 1.2
Ionic Radii, A
FIG. 67. Effect of charge and size of adsorbed cations on the ^-potential of
colloidal clay particles.
FLOCCULATION OF CLAY SOLS
419
changed ion. From this point of view, the more loosely an ion is held,
the greater the average distance of oscillation and the greater the likeli-
hood of replacement ; and conversely, the more closely the cation sticks
to the surface, the less readily will it be replaced. Since the average
distance of oscillation corresponds directly to the average thickness of
the double layer, it would follow that clays with high (-potentials
should have adsorbed ions which are easily exchangeable. A compari-
son of the (-potential with the percentage release of adsorbed ions
(columns 4 and 5 of Table LIV) supports this deduction. Since
thorium is so difficult to replace, it is probable that the observed (-po-
tential of Th-clay is much too high.
(-potential and Flocculation Values. The close relationship be-
tween (-potential and flocculation value is again illustrated by the data
of Table LTV shown graphically in Fig. 68 in which the precipitation
10
'30 35 40 45 50 55 60
^-Potential, Millivolts
FIG 68. Flocculation values of uni- and poly-clays by potassium chloride
values of potassium chloride are plotted against the (-potential. Ob-
viously, it takes much more electrolyte to coagulate a clay with a high
(-potential than one with a low potential. The curve cuts the ^r-axis
well above zero in accord with Powis' concept of a critical potential
for flocculation (p. 181). For the systems under consideration the
value appears to be 42 millivolts. That this value must be approx-
imately correct is evidenced by the behavior of methylene blue clay
which has a potential of 40.5 millivolts and therefore flocculates com-
pletely without the addition of electrolyte.
Flocculation Values and Exchange Adsorption. The flocculation
values, F, and exchange adsorption values, E, of various electrolytes
420
THE INORGANIC SOIL COLLOIDS
for three different clays are given in Table LV, after Jenny and Reite-
meier. Both values are expressed in symmetry values 5 1 (p. 417). With
the exception of H+ ion, the valency rule applies, but only in the sense
that the ion with the highest valence precipitates in the lowest concen-
tration. Ions of the same valence, especially univalent ions, show a
wide variation in precipitation value. Whenever the flocculating elec-
trolyte contains a foreign cation, exchange adsorption always takes
TABLE LV
RELATION BETWEEN FLOCCULATION VALUES OF ELECTROLYTES AND EXCHANGE
ADSORPTION
NH 4 -clay
Ca-clay
H-clay
Electrolyte
F(S)
E(S)
F(S)
E(S)
F(S)
E(S)
LiCl
8
32
4 8
13 08
2 9
6 6
NaCl
8
33 5
4 5
12 74
2 7
6 2
KC1
5 4
51 33
3
28 80
1 5
14 5
NH 4 C1
4 9
50
2 5
29 35
1 3
RbCl.
4 7
62 56
1 87
43 85
1 17
28 20
CsCl .
2 8
68 78
1 17
50 83
73
39 73
HC1
98
84 89
55
77 80
36
50
MgCU
1 22
65 44
59
47 53
47
15 78
CaCli
1 27
63 56
55
50
47
26 89
BaCl 2 ... .
1 16
71 67
55
52 96
35
23 78
Ld(N0 8 )s
90
47
18
ThCl 4
75
80 89
36
80 24
16
place. Comparing the flocculation and ionic exchange data of Ta-
ble LV, it appears that those ions which are weakly adsorbed have
high flocculation values, whereas ions which are taken up strongly by
exchange, precipitate in low concentrations. This may be shown
graphically by plotting the flocculating power F p , which is the recip-
rocal of the flocculation value, against the exchange adsorbability, Fig.
69. The resulting curves may be represented by the empirical
equations :
F p = 0.1400- 00332 * (for Ca-clay)
F p = 0.284<r 00431 * (for H-clay)
In other words, the coagulating effect of an ion increases exponentially
with its exchange adsorbability. The position of the curves for the dif-
FLOCCULATION OF CLAY SOLS
421
ferent clays emphasizes the fact that the flocculation value of a given
cation is determined in large measure by the nature of the exchangeable
ion initially present in the clay.
If the clay contains a highly hydrated univalent ion such as Na-
clay, the {-potential is high since the ions are loosely bound and oscil-
late through considerable distances. This means that the effective
width of the double layer is great or the degree of dissociation is high.
The flocculation value of an electrolyte for such a clay varies widely
with the properties of the coagulating ion. If it is highly hydrated and
univalent, such as Li+ ion, the extent of the exchange adsorption will
be relatively small and the change in {-potential will be altered but
F
Th-Clay
, Ca-Clay
20
40 60
Adsorbabihty, Percent
100
FIG. 69 Relationship between the exchange adsorhabihty of univalent ions and
their coagulating power for clays
little. The necessary lowering of the potential will be produced by a
decrease in thickness of the double layer or a decrease in dissociation, 65
which requires large amounts of electrolyte (15-20 S). On the other
hand, if the precipitating ion is univalent but not hydrated, such as
Cs+ ion, exchange adsorption predominates strongly, giving the par-
ticle a new outer layer which is so strongly attracted by the inner layer
that the {-potential drops sharply. Very little electrolyte is required to
repress the new outer layer to the critical distance characteristic of the
critical potential; hence the flocculation values are small, of the order
of IS.
Now if the original clay contains non-hydrated univalent ions such
as lithium, or polyvalent ions such as thorium, the {-potential is low to
422 THE INORGANIC SOIL COLLOIDS
begin with. The addition of an electrolyte such as lithium chloride will
produce a small exchange which will tend to increase the thickness of
the double layer and thus raise the ^-potential. 67 The peptizing action
is finally overbalanced by the repressing effect of more electrolyte, and
flocculation results with concentrations of medium magnitude (2-10 5")
but somewhat higher than the common ion values. Finally, if the co-
agulating ion is similar in nature to the one on the clay originally, ionic
exchange again assumes greater proportions without greatly modify-
ing the ^-potential. Flocculation is occasioned chiefly by the necessary
decrease in dissociation at low electrolyte concentrations, often less
than 1 S.
In the flocculation of acid clay sols with bases, the first step in the
process is the neutralization of the acid followed by the coagulation
of the resulting clay salt. The precipitation value of salt-base mix-
tures such as KC1-KOH and CaCl 2 -Ca(OH) 2 involves not only the
coagulating action of the cation but also the neutralizing and peptizing
influence of hydroxyl ion. 88
Deflocculation of Clays
The colloidal matter of a soft clay is deflocculated on shaking with
water but the process is greatly facilitated by the presence of electro-
lytes with strongly adsorbed anions such as alkali hydroxide, carbonate,
and silicate ; and is retarded or prevented by electrolytes with strongly
adsorbed cations. The deflocculating power of calcium hydroxide is
less marked than that of the alkali hydroxides because of the relatively
strong precipitating action of calcium ion. 89 Comber 70 attributes the
abnormal flocculating power of calcium hydroxide above a certain con-
centration to its coagulating action on emulsoid matter such as silica
that has a stabilizing action on the clay sol. 71 Clays which contain ap-
preciable amounts of soluble salts, such as the sulfates of calcium and
magnesium, are difficult to deflocculate because of the precipitating ac-
tion of the cation. On the other hand, clays containing protecting col-
loids such as humus are easily deflocculated. In technical practice,
*iCf. Baver: Mo. Agr. Expt. Sta. Research Bull., 129 (1929).
"C/. Bradfield: J. Phys. Chem., 32, 202; Mattson: 1532 (1928); Oakley:
Nature, 118, 661 (1926).
fl'Cy. Joseph and Oakley: Nature, 117, 624 (1926).
j. Agr. Sci., 10, 426 (1920); 11, 450 (1921); 12, 372 (1922); Nature, 118,
412 (1926).
"C/., also, Hardy: J. Phys. Chem., 80, 254 (1926) ; Kermack and William-
son: Nature, 117, 824 (1926) ; Proc. Roy. Soc. Edinburgh, 47, 202 (1927).
DEFLOCCULATION OF CLAYS
423
tannin is sometimes added to facilitate the deflocculating action of the
electrolytes and to increase the stability of the slip (p. 415).
The />H value at which clay particles attain a maximum charge is
lower for sodium silicate solutions than for sodium hydroxide or
*0 4 8 12 16 20
Na 2 Added, Percent
FIG. 70. The action of deflocculating agents on a Florida kaolin.
sodium carbonate. This is illustrated by Fig. 70 72 in which the rela-
tive time of flow of a kaolin slip through an efflux viscosimeter is
plotted against the sodium oxide content of the several deflocculating
"McDowell: J. Am. Ceram. Soc., 10, 225 (1927).
424 THE INORGANIC SOIL COLLOIDS
solutions. The greater deflocculating action of the silicate solutions
probably results from the presence of colloidal silica which adsorbs
sodium ions, thus disturbing their equilibrium with the hydroxyl ions
and allowing the hydroxyl ions to be adsorbed more readily by the clay
particles than they would be from a sodium hydroxide or carbonate
solution. It is therefore necessary to specify the deflocculating agent
employed when giving the />H value at which a clay slip possesses its
maximum fluidity. Silica sol alone does not serve as a deflocculating
agent for clay.
Adsorption of Anions
The adsorption of univalent ions such as nitrate and chloride by
soil colloids is relatively weak. Indeed Mattson 73 observed negative
adsorption of various anions by clays in the order: Cl = NO 3 <
SC>4 < Fe(CN) 6 . This does not mean that none of the anions
is adsorbed but only that the solvent is more strongly adsorbed than
the anions. Actually, the adsorption of an anion such as nitrate is so
weak that a nitrate fertilizer should be applied only as needed since
any excess will be leached out by drainage water. Phosphate, on the
other hand, is retained by the soil colloids under suitable conditions that
have been summarized by Bradfield, Scarseth, and Steel : 74 At pH
values from 2 to 5, the retention is attributed to the gradual dissolution
of iron and aluminum and their reprecipitation as the phosphates. At
/>H 4.5 to 7.5, the phosphate appears to be adsorbed on the surface
of the clay particles. The adsorption proceeds rapidly, and a fairly
sharply defined saturation value is reached that is independent of the
concentration of the added phosphate and seems to bear no relationship
to the base-exchange capacity. At />H 6-10, no adsorption takes place
unless bivalent cations such as calcium are present. 75 The bivalent
cation may act by the formation of an insoluble phosphate or by in-
creasing the adsorption of the anion, or both phenomena may enter in.
The further possibility that phosphate ion replaces hydroxyl, silicate,
or other groups from the surface of the aluminosilicate particle has
not been excluded.
Chemical Reactions
The action of hydrogen clays with inorganic bases might be treated
under base exchange, but since a definite chemical neutralization is
"Soil Sci., 28, 179 (1929).
7 * Trans. Intern. Congr. Soil Sci, 3rd Congr., Oxford, 1, 74 (1935).
75 Cf. van Bemmelen: "Die Absorption" (1910); Rostworowski and Wieg-
ner. J. Landw., 60, 223 (1912).
CHEMICAL REACTIONS
425
involved it seems better to consider such processes under the heading
of chemical reactions.
By prolonged electrodialysis of a clay, practically all the exchange-
able bases are removed and all soluble acids whose anions are capable
of passing a parchment membrane are taken out. If such a "hydrogen
saturated" clay is allowed to deposit on the anode membrane, the paste
may have a p H value of 2-3.5 whereas the clear water in which it was
dispersed may be almost neutral. The />H value of such clay acids
e
1 Bentonite (Rock River)
2 Bentonite (Cheyenne)
3 Putnam Clay
4 Silicic Acid
SNaOH
~0 40 80 120 160 180
Milliequivalents per 100 Grams
FIG 71 Typical titration curves for electrodialyzed colloidal clays.
varies with the sol concentration in much the same way as that of a
weak acid, like acetic, varies with the concentration. 78
Titration of Clay Acids. Bradfield has titrated electrodialyzed
clay acids by means of the hydrogen electrode TT and the glass elec-
trode. 78 Some typical curves obtained by titrating 100 cc of the clay
sols with 0.1 N NaOH are shown in Fig. 71, together with the curve
"Bradfield. J. Phys. Chem , 28, 170 (1924)
"Bradfield. Proc 1st Intern Congr. Soil Sci , 4, 858 (1927); J. Phys.
Chem., 28, 170 (1924); 35, 367 (1931); Colloid Symposium Monograph, 8,
367 (1931).
"Naftel, Schollenbcrger, and Bradfield. Soil Research, 3, 222 (1933).
426 THE INORGANIC SOIL COLLOIDS
for electrodialyzed silicic acid. It is apparent that the silicic acid is
much weaker than the clay acids; the former exhibit practically no
buffer action whereas the latter are usually well buffered at pH = 5-
6.5 and hence are of about the same strength as the first hydrogen of
carbonic acid. Comparing the apparent dissociation constants, />K,
of silicic acid and a typical clay acid on the basis of the mass law equa-
tion : pll = pK + log salt/acid, at the point where the acid is half
neutralized (ie , where />H =/>K), it is found that the pK value of
silicic acid is 10-11 in agreement with the accepted values whereas the
"apparent />K" value of Putnam lay is 5.6; of Rock River bentonite
3.8; and of Cheyenne bentonite, 3.6 and 5.9. The term "apparent pK
value" was used since in the above treatment it was assumed that the
clay behaves as a simple monobasic acid giving only one break, whereas
it is probably a complicated polybasic acid the hydrogen ions of which,
in different positions and possibly in different groupings on the surface
of the colloidal particle, have somewhat different dissociation tenden-
cies. However, the majority of them seem to have values of the same
order of magnitude as a simple acid having a />K value approximately
that of the apparent pK. value. One complicating condition is that the
apparent pK value obtained from titration curves is influenced notice-
ably by the nature of the base employed. 64
The neutralization of the exchangeable hydrogen in clays is about
complete at />H values varying from about 7.5 with calcium, barium,
and magnesium hydroxides to 8.5 with sodium and lithium hydroxides.
At higher />H values, the aluminosilicate minerals are decomposed to a
certain extent, forming simpler aluminates and silicates. The result-
ing alkali decomposition products are either in molecular solution or
approach the molecular state of subdivision whereas the alkaline-earth
compounds are less highly dispersed than the alkali compounds but are
finer than the original clay. 79
70 Cf Bradfield: Alexander's "Colloid Chemistry," 3, 581 (1931).
CHAPTER XXIII
CEMENT
The term cement, as ordinarily used at the present time, refers to
mortars which possess the properly of hardening in water as well as in
air. Attention has already been given to plaster of Paris (p 61 ) and
to hydraulic mortars in which magnesia or zinc oxide is the most im-
portant constituent (Vol. 11, pp. 183, 189). This chapter will deal with
Portland cement and related products.
PORTLAND CEMENT
The need for a cementing material to bind sand and small stones
together was ecogmzecl from the time man started to build. In some
of their constructions, the Assyrians and Babylonians are known to
have used moistened clay which was probably the first cementing ma-
terial ever used for building purposes. Such a binder is not sufficiently
durable or hard for building massive structures, and the next develop-
ment appears to have been the discovery by the Egyptians of the
cement now known as plaster of Paris, which was mixed with sand to
make the mortar used in the construction of the Pyramids. The dis-
covery that the application of heat to certain rock minerals, such as
gypsum, would give a cementing substance was later utilized by the
Greeks in making lime from limestone or marble. The Greeks prepared
some very satisfactory mortars by mixing lime with sand and volcanic
earths known as pozzolana. The development of pozzolana mortars
was brought to a high state of perfection by the Romans, as evidenced
by many of their imposing structures which still exist. The so-called
Roman or pozzolana cements were similar in many respects to the
modern Portland cement.
The art of cement making declined with the fall of Rome and was
not advanced until 1756 when John Smeaton discovered that a clayey
limestone found in Cornwall would give a hydraulic lime, when burned.
This product was mixed with pozzolana to prepare the mortar used in
constructing the Eddystone lighthouse. Because of the scarcity of
427
428 CEMENT
pozzolana, which is found only in a few volcanic regions, subsequent
investigations were carried out in an attempt to produce an artificial
Roman cement. The invention of a satisfactory process is attributed
to Joseph Aspdin of Leeds, who took out a patent in 1824 for making
a cement by heating an intimate mixture of limestone and clay at the
temperature ordinarily used in burning lime. It had been the custom in
preparing limes to heat argillaceous limestones in a stack furnace under
conditions such that some parts of the charge were sintered. These
sintered lumps were discarded because they were difficult to grind.
Aspdin, however, ground some of them and found the resulting powder
to possess superior cementing properties. To this product, Aspdin gave
the name Portland cement, since its color, after hardening, was similar
to that of Portland stone, a famous English building stone. Aspdin's
original cement was not what is now known as Portland cement, since
the temperature of burning was not high enough ; but a year later, in
1825, the importance of heating the mass to incipient fusion was recog-
nized. From this beginning, more than a century ago, there has de-
veloped the modern Portland cement industry.
The importance of Portland cement in our present-day civilization
is difficult to overestimate. Starting with the turn of the century, the
industry began to grow by leaps and bounds until in the year 1936 one
hundred million barrels of cement were manufactured. This vast pro-
duction is occasioned by the diversity of its important applications
chiefly in the form of concrete, a mixture of cement, gravel, and sand.
The importance of concrete construction reveals itself in no more strik-
ing way than in the interlacing ribbon of rock which constitutes the
vast highway system. Concrete roads, including the bridges, tunnels,
and reinforced embankments, have made possible the full utilization of
the automobile in communication and in transportation. Moreover,
concrete, the ideal plastic rock, has found application in the fabricating
of all kinds of structures from small homes to massive buildings, and
gigantic monoliths such as Boulder Dam. The strength and rigidity of
rein forced-concrete buildings provides safety from such destructive
agencies as earthquake, fire, and tornado, which have leveled less
sturdy forms of construction. Nor need beauty be sacrificed to utility
in modern concrete fabrication. An outstanding example of artistry
in concrete is the magnificent Baha'i temple in Chicago, the dome of
which is "a shelter of cobweb interposed between earth and sky, struck
through and through with light." * This building was done by Earley,
*Cf. Bogue: Ind Eng. Chem., 27, 1312 (1935).
PORTLAND CEMENT 429
who also made one of the first great statues from concrete, Lorado
Taft's "Fountain of Time" in Chicago.
Much of the progress in the utilization of cement has been made
possible by the investigations of scientists into the constitution of the
material and the properties of its individual components. Because of
the application of this knowledge it is no longer necessary to cure
a concrete slab 21 days before use ; a concrete road may now be made
which may be safely opened to traffic two to three days after laying.
Modern research has likewise improved the volume constancy of con-
crete and has thereby minimized the cracking and crazing of structures.
Moreover, the scientist has shown how a cement may be manufactured
which is resistant to the attack of sul fates and sea water. Finally,
Portland cement with a low heat of hydration has been developed
which enables the engineer to pour rapidly huge volumes of concrete
without the generation of so much heat within the mass that cracks
develop on cooling. Boulder Dam was made possible as a result of this
scientific achievement.
The importance of scientific investigations on this highly practical
material is emphasized by Bogue, 1 as follows: "Research on Portland
cement and concrete has enabled the engineer to design structures with
greater assurance of durability, of security, and of a wider field of
usefulness. It has placed in the hands of the architect a material with
which he may fabricate the most massive monoliths or which he may
mold into the most delicate tracery of ornamentation and to which he
may impart any tone or tint of the rainbow. It has had an important
part in the development of an incomparable network of highways,
great and inspiring edifices, and economical homes which are durable
and secure abodes of lasting beauty. Thus has research upon Portland
cement contributed, in this industrial age, to the satisfaction of life."
It is to an account of some of the most important scientific results of
this research that attention will now be given.
MANUFACTURE
Portland cement is produced by heating a mixture of rock- forming
materials containing suitable amounts of aluminum, calcium, and sili-
con together with small amounts of iron and magnesium. In the early
stages of the development of the industry, the method of procedure
employed in making a satisfactory cement was determined by the
method of trial and error. Now, it is known that certain definite com-
pounds impart the desired properties to cement and that a uniform
430 CEMENT
product made up of these compounds results only when the raw ma-
terial containing calcium, aluminum, and silicon in rather definite pro-
portions is ground to a fine powder and the very intimate mixture
heated to a minimum temperature.
Typical raw materials employed in cement manufacture are lime-
stone and clay, both of which are found in large deposits of uniform
composition. In some places, there exist deposits of clayey limestone,
called cement rock, containing all three of the essential constituents;
but, as a rule, either limestone or clay must be added to get the desired
composition for good Portland cement. The alumina and silica are
sometimes derived from blast-furnace slag and the calcium oxide from
sea shells. From whatever source the material is obtained, the separate
constituents are mixed in the proper proportions and thoroughly pul-
verized. If the raw materials are rocks, the grinding is commonly car-
ried out in the dry way. On the other hand, soft materials, such as
marl and clayey mud which are gathered by dredging operations, are
usually ground wet and are kept suspended until dried out in the kiln.
The burning process is carried out in cylindrical kilns, 100 to 400
feet in length and 6 to 12 feet in diameter, built of steel plates and lined
with highly refractory material. The drums are held in a slightly in-
clined position by friction rollers and are rotated slowly. The process
is continuous, the raw mix entering at one end of the kiln and the ce-
ment clinker leaving it at the other. The heat is derived from pulver-
ized coal, fuel oil, or gas which are blown into the lower end of the
kiln by compressed air, giving a flame IS to 40 feet in length. The
time of passage through the kiln is from 1.5 to 3 hours, during which
the raw material is subjected to a gradually increasing temperature that
reaches a maximum of 1425 to 1500. In the first stage of the burn-
ing process, the raw material is thoroughly dried ; in the second stage,
carbon dioxide and organic matter are driven off ; and in the final stage,
the alumina, silica, and lime react to form the cement clinker. The
clinker consists of partially sintered masses of particles from 0.5 to 6
cm in diameter. After the addition of a small amount of gypsum which
regulates the rate of setting, the particles of clinker are ground to a fine
powder, the surface area of which ranges from about 1400 to 1900
cm 2 /g ; this is the Portland cement of commerce.
Ordinary Portland cement prepared as described above hardens
rather slowly, ten days or more being required for a concrete slab to
cure properly. Rapid-hardening Portland cement is now prepared in
much the same manner as the ordinary product except that the propor-
tion of lime may be increased somewhat, the raw materials are more
CONSTITUTION
431
carefully mixed and ground, the burning is carried out at a somewhat
higher temperature, and the clinker is much more finely ground. The
rapid-hardening cement contains more tricalcium silicate (p. 438) than
the ordinary cement.
COMPOSITION AND CONSTITUTION
Composition
The results of the analysis of a few typical cements are given in
the left-hand portion of Table LVI. Small amounts of K 2 O and
TABLE LVI
COMPOSITION AND CONSTITUTION OF SOME TYPICAL CEMENTS
Composition per cent
Free
Constitution per cent
Type of cement
CaO
CaO
MgO
A1 2 8
Pe a 8
SiO 8
SO,
C 4 AP
C 3 A
C 3 S
C 2 S
Ordinary Portland
63 9
2 5
5 8
2 9
21 1
1 7
8
9
10
52
21
Rapid-hardening Portland
64 5
2 3
6
2 8
20
2 3
1 5
9
11
60
12
White Portland .
64 7
9
4 2
4
23 5
1 9
1 5
1
10
44
34
Iron ore or Erz .
64 8
7
1 7
8 2
23 3
1 1
Portland-blast furnace
60 2
3 5
7 S
1 4
22 6
1 3
Eisen-Portland
46 8
2.2
8
2 7
31 6
2 3
6
(PeO)
Ti0 2
Aluminous
40 1
7 7
37 5
5 6
6 6
6
2
Na 2 O are also present in most cements. The analysis gives, of course,
only the percentage amounts of the several components and does not
indicate the nature of the compounds.
Since more than 90% of the average Portland cement consists of
calcium, aluminum, and silicon, referred to the oxides, it is reasonable
to suppose that its properties result chiefly from compounds of these
three constituents. As a matter of fact, Richardson 2 demonstrated
that a good Portland cement can be made by starting with lime, silica,
and alumina in the pure state.
Constitution
Many workers have been concerned with the constitution of Port-
land cement since Le Chatelier 8 published the results of his classical
investigations more than a half-century ago. In most of the work,
2 Cement, 6, 314 (1904).
3 "Experimental Researches on
(1887), translated by Hall (1905).
the Constitution of Hydraulic Mortars"
432 CEMENT
the evidence offered in support of the alleged reactions which take place
during the burning process, and of the compounds formed, is not con-
vincing since the criteria used to define a compound were either in-
definite or insufficient. The solution of many questions connected with
the constitution and setting of Portland cement has been brought about
by the thorough systematic investigations carried out in the Geophysical
Laboratory and the National Bureau of Standards. Rankin and
Wright* in their pioneer work in 1915 made a complete phase-rule
study of the ternary system CaO-Al 2 O 3 -SiO 2 which necessitated
the investigation of about 1000 different compositions and fully 7000
3AI ? 3 -2Si0 2
CaO ' ^ ^^ffffffamilff//ff//MS^ ^^A. A| Q
FIG 72. Final products of the crystallization of lime, alumina, and silica. P is
the Portland cement zone and ^4, the aluminous cement zone.
heat treatments and microscopical examinations. The results of these
observations are summarized in the triangular concentration diagram
shown in Fig. 72. In this diagram, the pure components are repre-
sented by the apexes of the triangle; the binary mixtures, CaO-
A1 2 O3, Al 2 O 3 -SiO 2 , and SiO 2 -CaO, respectively, by points on the
three sides ; and ternary mixtures, by points within the triangle. Each
side of the triangle is divided into 100 parts, and all compositions are
given as percentage weights of the components. The lines within the
large triangle divide it into 14 small triangular spaces which enclose
all possible mixtures of the three components whose compositions are
*Am. J. Sci., (4) 39, 1 (1915); Shepherd, Rankin, and Wright: Ind. Eng.
Chem., 3, 211 (1911) ; Rankin: 7, 466 (1915).
CONSTITUTION 433
represented by the apexes of the respective triangles. The region of
Portland cement falls in the shaded area within the triangle which indi-
cates the three solid phases at equilibrium to be tricalcium silicate, di-
calcium silicate, and tricalcium aluminate.
Since Portland cement contains other components than lime,
alumina, and silica, phase-rule studies had to be made of systems other
than CaO-Al 2 O 3 -SiO 2 . Among those which have received special
consideration are systems containing the components MgO, 5 Fe 2 Os, 6
K 2 O, 7 and Na 2 O. 8 The results of these comprehensive studies indi-
cate with considerable certainty that Portland cement under equi-
librium conditions consists chiefly of 3CaO SiO 2 , 2CaO SiO 2 ,
3CaO A1 2 O 3 , 4CaO A1 2 O 3 Fe 2 O 3 , and MgO.
In opposition to the above conclusion it has been claimed: (1) that
the lime in excess of that necessary to form dicalcium silicate with the
silica enters into solid solution with the dicalcium silicate; 9 (2) that
3CaO SiO 2 does not exist, and that the chief constituent of clinker is
8CaO A1 2 O 3 2SiO 2 ; 10 (3) that 3CaO A1 2 O 3 forms a solid solution
with the silicates; 11 (4) that the clinker contains relatively large
amounts of uncombined lime. 12 These views have been found un-
tenable ; and the results of phase equilibria, chemical, and microscopical
studies have been confirmed in all essential respects by the application
of x-ray analysis technique to the problem. 18 Taken together, the re-
sults indicate: (1) that the most abundant constituents of Portland
cement clinker are 3CaO SiO 2 and 0-2CaO SiO 2 ; (2) that, in ad-
dition, the following compounds are normally present : 3CaO A1 2 O 3 ,
4CaO A1 2 O 3 Fe 2 O 3 , and MgO; and (3) that free CaO is not nor-
mally present in amounts as great as 2.5%.
Bogue 14 has shown that it is possible to calculate the approximate
equilibrium constitution of a Portland cement clinker from the oxide
sRankin and Merwin: J. Am. Chem. Soc., 38, 568 (1916); Am J. Sci.,
(4) 46, 301 (1918) ; Ferguson and Merwin: 48, 81 (1919) ; Hansen and Brown-
miller. (5) 15, 225 (1928) ; Hansen: J. Am Chem. Soc., 60, 2155 (1928).
6 Hansen and Bogue- J. Am. Chem. Soc., 48, 1261 (1926); Hansen, Brown-
miller, and Bogue: 60, 396 (1928); Lea and Parker: Trans. Roy. Soc. (Lon-
don), 2S4A, 1 (1934).
'Brownmiller: Am. J. Sci., (5) 29, 260 (1935).
8 Brownmiller and Bogue: Am. J. Sci., (5) 28, 501 (1932).
'Dyckerhoff: Zement, 16, 731 (1927).
iJanecke: Z. angew. Chem., 74, 428 (1912).
"Kuhl: Zement, 18, 512 (1924).
"Nacken: Zement, 18, 1017 (1927).
is Brownmiller and Bogue: Am. J. Sci., (5) 20, 241 (1930).
"Ind. Eng. Chem., Anal. Ed., 1, 192 (1929).
434 CEMENT
and free lime analysis. The calculated constitution of some typical
cements is given in the right-hand portion of Table LVI. In this table
C 4 AF = 4CaO A1 2 O 3 Fe 2 O 3 ; C 3 A = 3CaO A1 2 O 3 ; C 3 S = 3CaO-
SiO 2 ; and C 2 S = 2CaO SiO 2 .
It must be emphasized that the above statements concerning the
constitution of cement refer to equilibrium conditions only. Lea and
Parker 15 and Dahl le have pointed out the possibility and outlined the
results of disequilibrium conditions during the cooling of the clinker.
For example, some of the liquid formed during clinkering which
amounts to 25-30% may solidify as undercooled liquid or glass. 17 In
the same connection, Insley 18 and Schwiete and Bussem 19 have ob-
served by x-ray and optical methods that detectable and sometimes im-
portant amounts of other materials may be present in solid solution in
the principal constituents ; and Insley found by a petrographic analysis
of commercial clinker that the relative amounts of the compounds, as
observed by that method, may differ widely from the amounts cal-
culated from chemical analysis according to Bogue's method. More-
over, in a study of the solidus and liquidus relations in the area CaO-
4CaO A1 2 O 3 Fe 2 O 3 -CaO A1 2 O 3 , McMurdie 20 found that 3CaO -
A1 2 O 3> 5CaO 3A1 2 O 3 , and CaO A1 2 O 3 all take Fe 2 O 3 into solid solu-
tion up to 2.5%; and that 4CaO A1 2 O 3 Fe 2 O 3 takes up between 3
and 5% of the calcium aluminatcs into solution.
SETTING AND HARDENING
When finely pulverized Portland cement is mixed with water, a
plastic mass results which becomes solid in the course of a few hours.
This process, which is called setting, is followed by a gradual increase
in strength or hardening of the mass. Although ordinary Portland
cement becomes very hard in the course of a week or ten days and
the rapid-setting variety within a few days, the strengths may be in-
creased over a period of years.
According to Le Chatelier, 8 the setting and hardening of Portland
cement consists in the dissolution in water of the anhydrous silicates
and aluminates, which subsequently become hydrated. Since the hy-
drates are less soluble than the anhydrous salts, the solutions become
"Dept Sci. Ind Research (Brit.), Build. Research Tech. Paper 16 (1935).
"Rock Products, 26 (Dec. 10, 1932).
i* See McMurdie: J. Research Natl. Bur. Standards, 18, 475 (1937).
18 J. Research Natl. Bur. Standards, 17, 353 (1936).
lOTomnd-Ztg., 66, 801 (1932).
20 J Research Natl. Bur Standards, 18, 475 (1937).
ACTION OF WATER ON PORTLAND CEMENT COMPOUNDS 435
supersaturated with respect to the former and deposit an entangling
mass of needles, thereby giving the cement its characteristic hardness.
This theory of the hardening process was not questioned until Mi-
chaelis 21 recognized the formation not only of crystals but also of a gel
which increased gradually in amount until it filled the interstices be-
tween the crystalline needles as well as those between the cement par-
ticles. The cementing gel was supposed to be calcium monosilicate, and
the crystals tricalcium aluminate and calcium hydroxide. According
to this hypothesis, the cement particles and crystals become embedded
in a common sheath of gelatinous substance which imparts ,a degree of
hardness that could not be attained by the felting of crystalline needles
alone.
More or less successful attempts were made to distinguish the vari-
ous products of hydration of Portland cement by the use of organic
dyes which stain colloidal and zeolitic minerals selectively. 22 Such ex-
periments led Blumenthal 28 to conclude that crystalline monocalcium
silicate and tricalcium aluminate were among the first products of hy-
dration and that a gelatinous silicate forms subsequently. Although
this method of attack gave some helpful information, a systematic in-
vestigation of the setting and hardening process was possible only
after the constitution of cement had been established. The essential
constituents of cement being known, investigations were made by
Klein, Phillips, and Bates 2 * and by Rankin 2B of the action of water
on each constituent in turn. These observations have been confirmed
in most respects and extended by Lerch and Bogue, 26 - 2T who investi-
gated first the reactions of the cement compounds with an excess of
water and later the behavior of the compounds both individually and
collectively when mixed with water in proportions similar to those used
with cements in concrete.
Action of Water on Portland Cement Compounds
Water added to dry Portland cement compounds is probably first
adsorbed on the surface of the particles and later reacts either by
21 Kolloid-Z. f 5, 9 (1909); 7, 320 (1910); Chem-Ztg, 17, 982 (1893).
22 Keisermann Kolloid-Beihefte, 1, 423 (1910).
23 Silikat-Z., 2,43 (1914).
24 Klein and Phillips: Bur. Standards, Tech Paper 43 (1914); Bates and
Klein: Tech. Paper 78 (1916); Phillips: J. Am. Ceram Soc, 2, 708 (1919).
2sj. Franklin Inst, 181, 747 (1916); Rankin and Wright: Am J Sci., (4)
39, 1 (1915).
20 Lerch and Bogue: J. Phys. Chem, 31, 1627 (1927).
27Bogue and Lcrch: Ind. Eng. Chem, 26, 837 (1934).
436 CEMENT
hydrolysis or by hydration. The product of a hydrolytic reaction may
take up water either by adsorption or in the form of a definite hydrate,
and a definite hydrate may adsorb water giving a hydrous hydrate. In
Fig. 73 26 some observations on the amount of water fixed by adsorp-
tion or hydrate formation or both, calculated as percentages of the
original anhydrous material, are plotted against time in days, for Port-
land cement compounds. The particles were ground to pass com-
pletely through a No. 100 sieve and 90% through a No. 200 sieve. In
general, an amount of water equal to 50% of the weight of the ce-
menting material was employed.
The compressive strengths of J % 6 by 1 in. cylinders of the several
compounds at ages up to 1 year are shown in Fig. 74. 27 The com-
pressive strength is expressed in pounds per square inch. It is ap-
parent from these curves that most of the compressive strength of
Portland cement comes from the silicate constituents.
Tricalcium Aluminate. When 3CaO A1 2 O 3 is mixed with water,
a gelatinous hydrous material is first formed which sets so rapidly that
it is almost impossible to make test pieces. The rapid reaction is ac-
companied by a rise in temperature and loss of water as steam which
favors the flash set. On adding more water and working rapidly,
the paste does not again assume a rapid set. With limited amount of
water in the paste, it is converted into and remains a submicroscopically
crystalline gel during the first 24 hours at least ; with more water, the
microcrystals grow fairly rapidly ; it never attains a high compressive
strength. The final products of the hydration consist of fluffy isotropic
aggregates of very fine crystalline grains having the composition
3CaO A1 2 O 3 6H 2 O, 28 with more or less adsorbed water. A hexag-
onal hydrate containing a larger amount of water may be formed at
lower temperatures, but it tends to revert to the isotropic form. The
isotropic grains of hexahydrate give a distinct x-ray diffraction pat-
tern which contains none of the lines characteristic of anhydrous
3CaOAl 2 O 3 , 5CaO-3Al 2 O 3 , or Ca(OH) 2 . Kiihl and Thuring 29
claim that, when 3CaO-Al 2 O 3 is hydrated in the presence of sat-
urated calcium hydroxide, tetracalcium aluminate is formed, and
Lafuma 80 concludes that hydrated tetracalcium aluminate and hy-
drated dicalcium aluminate result. Bogue and Lerch were unable to
2* Thorvaldson, Grace, and Vigfusson: Can. J. Research, 1, 36, 301 (1929);
cf. Pulfrich and Linck: Kolloid-Z, 34, 117 (1924); Duchez- Rock Products,
27, No. 18, 62 (1924).
2Zement, 18, 243 (1924) ; Roller: Ind. Eng. Chem, 26, 669, 1077 (1934).
"Ciment, 174 (1925).
ACTION OF WATER ON PORTLAND CEMENT COMPOUNDS 437
identify these new phases either by microscopic or x-ray methods in a
mixture of tricalcium aluminate and 10% lime which was allowed to
stand 2 years.
40
30
0-20
10
" 7 28 90 180 270 360
Time, Days
FIG 73 Water fixed by pure cement compounds after varying time intervals.
12000
3CaO-Al 2 3
180
Time, Days
FIG. 74 Comparison of compressive strengths of cement compounds after
varying time intervals.
Tricalcium aluminate mixed with the calcium silicates in the ab-
sence of a retarder tends to raise the early strength and lower the late
438 CEMENT
strength of the mixtures. Bogue and Lerch suggest that the effect of
aluminate on the increase in early strength may be due to a decrease
in the amount of water available to the 3CaO SiO 2 ; and the later re-
duction in strength may be caused by the formation of tricalcium
aluminate hydrate which gives an open structure and thereby prevents
optimum contact of the hydrating calcium silicate granules.
Beta-dicalcium Silicate. 0-2CaO SiO 2 takes up water very
slowly. At the end of a month a slight etching accompanied by the
formation of an amorphous layer around the edges was noted. The
amount of this amorphous material increased gradually at the expense
of the crystalline silicate, but a large amount of unhydrated material re-
mained even after 2 years. Not more than a trace of calcium hydrox-
ide could be detected, and no other new crystalline phase appeared.
The actual amount of water bound was 0.7% at 1 day and 12% after
1 year.
y-dicalcium silicate reacts much less rapidly even than the beta
compound.
Tricalcium Silicate. 3CaO-SiO 2 is the only one of the three
major constituents which reacts with water within a reasonable time
to give a mass comparable to Portland cement in hardness and strength.
In a paste containing 50% by weight of water, the compound was hy-
drolyzed to the extent of 15% (i e., the reaction 3CaO SiO 2 +
#H 2 O==3Ca(OH) 2 + SiO 2 #H 2 O had proceeded 15% to comple-
tion), leaving an amorphous material of lower basicity than the original
silicate. The hydrolysis had reached 23% in 7 days and 26% in 6
months ; no further change was observed in 2 years. The silicate had
fixed over 8% of water in 1 day and 19% at the end of a year (Fig.
73). Subtracting the water taken up as Ca(OH) 2 gives the water held
by the 3CaO SiO 2 residue. The hydrous hydrolysis products of both
3CaO SiO 2 and 2CaO SiO 2 appear to be entirely amorphous to
x-rays when formed at ordinary temperatures.
The hydrolytic reactions of both silicates proceed more rapidly and
further in an excess of water than in pastes that are allowed to set to
a hard mass. 26 * 2T The final decomposition product of 3CaO SiO 2 in
an excess of saturated limewater approaches the composition 3CaO -
2SiO 2 with a percentage hydrolysis of approximately SO. Assuming
this to be the equilibrium composition, the 3CaO SiO 2 paste under
consideration hydrolyzed to but 53% of the equilibrium value in 2
years. In a similar way it was found that -CaO SiO 2 pastes hy-
drolyzed to but 3.7% of the equilibrium value in 2 years.
The compressive strength developed by the calcium silicates in
ACTION OF WATER ON PORTLAND CEMENT COMPOUNDS 439
Portland cement is probably caused chiefly by the gel of amorphous
hydrous calcium silicate together with a felt of needle-like crystals of
calcium hydroxide. Referring to Figs. 73 and 74 it will be seen that
the increases in strength on aging are not proportional to the hydrolysis
of the silicates since 2CaO SiO 2 hydrolyzes but little up to 1 to 2 years,
yet it develops high strength in that time. Moreover, the strength in-
creases are not proportional to the total combined water since neither
calcium silicate shows any appreciable change in fixed water between
1 and 2 years whereas the strength, especially of the 2CaO SiO 2 mass,
increases during that period. During earlier periods there is some cor-
relation between the rate of development of strength and the water
taken up by the two silicates. For Portland cements of widely vary-
ing composition, Lea and Jones S1 found a fair relation between the
compressive strength of concrete and the water fixed in the set cement.
The curve obtained by plotting combined water against the tensile
strength is concave to the strength axis, indicating that the strength
increases at a rate somewhat more than directly proportional to the
amount of water taken up.
Tetracalcium Aluminum Ferrite. The compound 4CaO A1 2 O 3 -
Fe a O 3 takes up water rapidly, giving a gelatinous impervious layer
around the particles which retards the further penetration of the water.
The compound fixed 25% of water in 1 day, increasing to 29% in 1
year (Fig. 73). A hydrolytic reaction takes place with the formation
of a crystalline product which gives the same x-ray diffraction pattern
as 3CaO A1 2 O 3 6H 2 O, and an amorphous product of unidentified
constitution which Bogue and Lerch believe to be hydrous CaO -
Fe 2 O 3 . The compressive strength of the set compound is less even
than that of 3CaO A1 2 O 3 , and, when mixed with calcium silicates, the
compressive strength of the mixtures is reduced.
Portland Cement. When water is added to Portland cement con-
taining the four compounds considered above, the initial set results
from the formation of a gel with 3CaO A1 2 O 3 and to a certain extent
with 4CaO A1 2 O 3 Fe 2 O 3 . Pulfrich and Linck 82 emphasized the im-
portance of gel formation in the initial set by showing that mi-
croscopically visible crystallization does not take place at the out-
set in the presence of the amount of water used in technical practice.
Their observations were made in glycerol solutions in order to get
the necessary dilution for microscopic examination, and the gly-
cerol may have inhibited the crystallization. This, however, merely
J. Soc. Chem. Ind., 64, 63T (1935).
w Kolloid-Z., 84, 117 (1924).
440 CEMENT
emphasizes the contention that setting is not necessarily occasioned
by the formation of microscopically visible needles. The gel of
tricalcium aluminate, which is largely responsible for the original
set, reacts with water to give microscopic crystals of 3CaO A1 2 O 3 -
6H 2 O; and the gel from the ferrite gives ultimately both 3CaO-
A1 2 O 3 6H 2 O -and an amorphous ferrite, possibly CaO Fe 2 O 3 .
Whereas the initial set results primarily from the formation of
tricalcium aluminate gel, the subsequent fairly rapid increase in co-
hesive strength and hardness is due in large measure to the hydration
of 3CaO SiO 2 and, to a lesser degree, of 2CaO SiO 2 with the libera-
tion of a calcium silicate gel in which are embedded needle crystals
of calcium hydroxide. Rapid-hardening Portland cements contain
more 3CaO SiO 2 than the ordinary variety and are ground finer to
allow more ready access of water. It is a pity that dicalcium silicate
does not hydrate more rapidly since it is formed at a lower temperature
than tricalcium silicate and yields ultimately a higher percentage of the
important binding silicate gel.
To avoid overheating when enormous volumes of concrete are
poured rapidly, as in the construction of Boulder Dam, it was necessary
to develop a specification which placed definite limits on the computed
compound constitution of the Portland cement, in order to ensure a low
heat of hydration. These are as follows : S3 4CaO A1 2 O 3 Fe 2 O 3 not
over 20% ; 3CaO A1 2 O 3 not over 7% ; 2CaO SiO 2 not over 65% :
and 3CaO SiO 2 not over
Action of Calcium Sulfate
As already pointed out in connection with the manufacture of Port-
land cement, gypsum is added in amounts up to 2-2.5% before grind-
ing, in order to retard the time of set. The marked influence of gyp-
sum is due largely to the removal of calcium aluminate as the insoluble
calcium sulfoaluminate 3CaO A1 2 O 3 3CaSO 4 31H 2 O. 34 This de-
lays the formation of the gel of 3CaO A1 2 O 3 6H 2 O until all the sul-
fate is removed and so retards the initial set.
Gypsum also tends to counteract the effect of 3CaO A1 2 O 3 6H 2 O
in lowering the compressive strength of the calcium silicates. A pos-
33 Robertson: Ind. Eng. Chem., 27, 242 (1935).
"Kuhl: Zement, 18, 362 (1924); Lerch, Ashton, and Bogue: J. Research
Natl. Bur. Standards, 2, 715 (1929); Bogue and Lerch: Ind Eng. Chem., 26,
837 (1934); cf., however, Klein and Phillips: Bur. Standards, Tech. Paper 43
(1914); Phillips: J. Am. Ceram. Soc., 2, 708 (1919); Rohland: Kolloid-Z., 4,
223 (1909); 8, 251 (1911); Roller: Ind. Eng. Chem., 26, 669, 1077 (1934)
ACTION OF OTHER SALTS 441
sible explanation of this effect is that the structure of the set paste
has time to be established by the reacting tricalcium silicate before any
appreciable amount of 3CaO A1 2 O 3 6H 2 O, with its weak and open
structure, can form. 35 Gypsum also counteracts the influence of 4CaO -
A^Oa Fe 2 C>3 in lowering the compressive strength of the calcium
silicates. This effect may likewise be associated with the formation not
only of calcium sulfoaluminate but also of calcium sulfoferrite, thereby
cutting down the amount of amorphous hydrous ferrite that can be de-
posited on the calcium silicate granules.
Action of Other Salts
The addition of most salts influences the time of set to a greater or
lesser degree, but the available data 38 are often conflicting. Some salts
retard the set, and others accelerate it ; another group of salts retards
the rate of set when present in small amounts and accelerates it when
larger amounts are used. Furthermore, the effect produced often
varies with the composition of the cement employed. Gadd 36 investi-
gated the action of a large number of compounds including the car-
bonate, nitrate, chloride, sulfate, borate, and hydroxide of sodium,
ammonium, aluminum, zinc, cobalt, and chromium. He found that
nitrates had little effect on the rate of set whereas all other compounds
except gypsum and plaster of Paris accelerated it. In view of the
influence of salts on jelly formation (Vol. II, p. 15), it is not surprising
to find that their presence has an effect on the rate of set of cement,
which may be altogether independent of any chemical action. One
would expect the presence of foreign salts 37 to have a retarding or ac-
celerating action depending on whether they have a coagulating or pep-
tizing action on the colloids formed by the action of water on the
cement particles.
Various mixtures are sold under trade names for accelerating the
rate of set. These are composed of aqueous solutions of various salts
such as carbonates, aluminates, silicates, or mixtures of various chlo-
rides, such as those of calcium, aluminum, and sodium. One such accel-
erating material called "Cal" is prepared by the action of lime and cal-
cium chloride in water. The accelerating action probably results from the
5Kuhl: Zement, IS, 362 (1924).
36 Gadd: Brit. Portland Cement Research Assoc, Pamphlets, 1 (1922);
Thomas: Dept. Sci. Ind. Research (Brit.), Build. Research Special Kept. 14
(1929); Forsen: Zement, 19, 1130 (1930); Biehl: 17, 487 (1928).
"C/. Benson, Newhall, and Tremper: Ind. Eng. Chem, 6, 795 (1914).
442 CEMENT
precipitation of a calcium chloraluminate of the composition 3CaO . -
A1 2 O 3 CaQ 2 ' 18H 2 O, 80 * 88 with an accompanying decrease in pH
value. This reduction in />H accelerates the hydrolysis of the silicates
and so hastens the hardening process. Platzmann, 89 on the other hand,
attributes the action mainly to the hygroscopicity of calcium chloride
which, by absorption of moisture during the first few weeks, prevents
the shrinking and cracking of the cement and protects it from too rapid
a loss of moisture.
IRON ORE AND PORTLAND BLAST-FURNACE CEMENTS
Iron ore or Erz cement is manufactured near Hamburg, Germany,
in much the same way as ordinary Portland cement except that the
clay or shale in the original mixture is replaced by iron ore. The ferric
oxide content is relatively higher and the alumina relatively lower
(Table LVI) than in ordinary cement; but the chief hydraulic con-
stituent in both products is tricalcium silicate. The area occupied by
cements rich in iron in a triaxial diagram of the system CaO-SiO 2 -
Fe 2 O 3 is in nearly the same position as that of Portland cement in the
system CaO-SiO 2 -Al 2 O 3 . 40 Cements rich in iron contain 2CaO -
Fe 2 O 3 in addition to the four principal constituents of the ordinary
product. This 2CaO Fe 2 O 3 is in solid solution with 4CaO A1 2 O 3 -
Fe 2 3 .
Portland blast-furnace cement is a finely ground mixture of or-
dinary Portland cement clinker and granulated blast-furnace slag. The
British standard specifications call for not less than 35% clinker nor
more than 65% slag. In Germany two varieties of this type of cement
are recognized by specification: Eisenportland cement containing not
more than 30% granulated slag; and Hochoven cement containing not
more than 85% granulated slag.
Portland blast-furnace cements are said to be superior for sea-water
construction, 41 possibly because the added slag unites with any free
lime, thereby preventing it from forming calcium hydrosilicates or from
acting with the magnesium sulfate of sea water to give such compounds
ss Cf, however, Kuhl and Ullrich: Zement, 14, 859, 880, 898 (1925); Gass-
ner: Chem.-Ztg, 48, 157 (1924).
3 Zement, 10, 499 (1921); 11, 137 (1922); Chimie & Industrie, 7, 943; 8,
614 (1922).
*<>Kuhl- Zement, 10, 361, 374 (1921).
"For an account of the resistance of concrete to natural destructive agen-
da see Lea and Desch: "The Chemistry of Cement and Concrete," 350 (1935).
ALUMINOUS CEMENT 443
as magnesium hydroxide * 2 or calcium sulfoaluminate, 48 which are ac-
tive in producing cracks. It is claimed that such cements resist the ac-
tion of sul fates much more than ordinary Portland cement. 44 It is
probable, however, that the chief importance of cements made with
slag lies in the facility with which the waste product of an important
industry is transformed into a commercially salable article. In any
event, the present sul fate-resisting Portland cements which have a low
content of 3CaO A1 2 O 3 are quite as durable in sulfate waters as Port-
land blast-furnace cements.
ALUMINOUS CEMENT
Cements with a high alumina content were developed 46 because of
the serious difficulties which were experienced in France owing to the
decomposition of concretes in sea water and especially in earth con-
taining large amounts of sul fates, chiefly gypsum. Cements in which
the alumina content is equal to or greater than that of the silica content
are known commercially as "aluminous," "fused," or "electro fused"
cements. They are produced by fusion, because calcium aluminates
soften readily, and clinkering is very difficult. 46
The aluminous cement zone in the system CaO-Al 2 O 3 -SiO 2 4T is
shown at A in Fig. 72. The compounds which may be present are
therefore CaO A1 2 O 3 , 5CaO-3Al 2 O 3 , 3CaO-Al 2 O 3> 3CaO-5Al 2 O 3 ,
2CaO A1 2 O 3 SiO 2 , and 2CaO SiO 2 . Of these we have seen that
3CaO A1 2 O 3 possesses slight hydraulic properties, and the same is
true for 2CaO A1 2 O 3 SiO 2 . 2CaO SiO 2 is a good cementing ma-
terial but develops strength slowly. Of the other compounds SCaO
3A1 2 O 3 sets very rapidly indeed, whereas both 3CaO SA1 2 O 3 and
CaO A1 2 O 3 set slowly but harden rapidly developing great strength
in 24 hours. The evidence indicates that CaO A1 2 O 3 is by far the
most important ingredient in aluminous cement ; hence a pure alumin-
ous cement should contain 35.45% CaO and 64.55% A1 2 O 3 , but the
42 Lewis: Engineering, 109, 626 (1920); Gary: Mitt. Materialprufungsamt,
37, 12 (1919).
"Grim. Zement, 12, 297, 307, 317, 326 (1924).
"Probst and Dorsch: Zement, 18, 292, 338 (1929).
* 5 C/ Bied: "Recherches industrielles sur les chaux, ciments et mortiers,"
Paris (1926).
"Bied: Tech moderne, 14 f 508 (1922); Rev. met., 19, 759 (1922)
*i Bates: Bur. Standards Tech. Paper, 197 (1921); Endell: Zement, 8, 319
(1919); Berl and Loblem: 15, 642 (1926); Solacolu: 22, 17, 33, 114, 191, 311
(1933) ; Kuhl and Ideta: 20, 261 (1931) ; Richter . 21, 445 (1932).
444 CEMENT
commercial products vary in composition within 30-45% A1 2 O 3 ,
35-45% CaO, 5-12% SiO 2 , and 5-12% Fe 2 O 3 .
The reactions which take place in the setting and hardening of
aluminous cements are still the subject of controversy. Le Chatelier, 48
Lafuma, 49 Ktihl and Thiiring, 50 and Koyanagi 51 are of the opinion that
the hydration of aluminous cement can be represented essentially by
the equation :
2 (CaO A1 2 O 3 ) + 10H 2 O - 2CaO A1 2 O 3 7H 2 O + A1 2 O 3 3H 2 O
Assarsson, 5 * on the other hand, concludes from observations in the
presence of limited amounts of water that the main reaction is the
formation of monocalcium aluminate gel together with quite limited
amounts of dicalcium aluminate and hydrous alumina. As in Portland
cement, the ferric iron compounds doubtless give hydrous CaO Fe 2 O 3 .
As we have seen, the investigations which led to the production of
aluminous cements were stimulated by the need for a product which
was resistant to sulfur-bearing waters. These investigations were suc-
cessful; under ordinary temperature conditions, aluminous cements
resist the attack by sul fates and sea water to a greater extent than any
other constructional cement with the possible exception of the sulfate-
resisting Portland cements having a low 3CaO A1 2 O 3 content. Other
advantages claimed for aluminous cements are greater earlier strength,
and the higher temperature developed on setting, usually sufficient to
allow normal hardening even in severe weather. Contrary to what is
frequently reported, the chief advantage of aluminous cements is not
their rapid hardening, which is equally characteristic of some Portland
cements, but their marked resistance to chemical action.
Aluminous cements are now manufactured in many European
countries, notably France and England; in the United States; and in
French Indo-China. The chief drawback to their wide commercial use
is the lack of a widely distributed supply of bauxite and the consequent
high cost of raw material.
Ciment, 82, 82 (1927).
49 "Recherches sur les aluminates de calcium/' Paris (1932).
"Zement, 13, 109, 243 (1924).
si Concrete, 40 (8), 40 (1932).
"Zement, 28, 15 (1934).
INDEX OF AUTHORS
ABEGG, 154, 227, 269
ABEL, 234
ACHESON, 415
ACKERMAN, 150
ADAMS, 27-9
AGGAZZOTTI, 180
AHRENS, 258
ALBERTI, 292
ALBU, 380
ALEXANDER, 254, 347
ALLEN, 27, 250-1, 253, 255, 260-1,
298
ALLENDORFF, 292
ALLMAND, 332, 341
ALT, 252
ALVISI, 261
ALWAY, 410
AMBROSE, 387
AMES, 302
AMSEL, 89
ANDEREGG, 395
ANDERSON, K, 302
ANDERSON, J. A , 371
ANDERSON, M. S., 402, 410-1
ANNETTS, 197, 202
ANTONY, 234, 248-9, 272
VAN ARKEL, 201
ARMSTRONG, 62, 67, 70
ARNOLD, 94
ARTMANN, 4, 216, 245, 259, 274-5
ASCH, D., 416
ASCH, W., 416
ASCHENBRENNER, 380
ASCOLI, 180
ASHLEY, 402
ASHTON, 440
ASPDIN, 428
ASSARSSON, 444
ATANASIU, 311
ATEN, 229
AUDUBERT, 175, 182
295-6,
AUGER, 230
AUSTERWEIL, 399
AVERELL, 41
BACH, 187
BACH MANN, 349-50, 368
DE BACHO, 223, 300
BACKER, 275
BACON, 389
BADECKER, 238
BAERWALD, 287
BAHL, 215-6, 224, 259
BAKER, 143, 290
BALAREW, 22, 26, 31-2, 36, 40-1, 62, 64,
236
BALCAR, 34
BALDSIEFEN, 161
BALLA, 227
BANCELIN, 179
BANCROFT, 23, 53, 106, 150, 153-4, 162-3,
213, 329-30, 338, 414
BARFOED, 229
BARLOW, 327-8, 338
BARNETT, 53
BARTELL, 44, 280, 328-9
BARVE, 88, 177, 183
BARY, 375
BASINSKI, 114-5, 118
BASS, 346
BASU, I. Y , 103
BASU, S. K, 187
BATES, 435, 443
BAUBIGNY, 237, 255
BAUDISCH, 346
BAUR, 154
BAYER, 412, 417, 422, 426
BAYKOV, 71
BECHHOLD, 47, 259, 337, 350
BECK, 177
BECKER, 299
BECKERATH, 29, 103, 144, 146, 351
445
446
INDEX OF AUTHORS
BECQUEREL, 149, 239, 369
BEEKLEY, 103
BEER, 103
VAN BEMMELEN, 416, 424
BENARD, 311
BENCZER, 349
BENNETT, 364
BENEATH, 42
BENSON, 441
BENTON, 141
BERGER, 100-2, 111
BERGLUND, 237
BERKELEY, 6, 327
BERL, 443
BERNECK, 7
BERNFELD, 256
BERRY, 133
BERTHELOT, 290, 301
BERTHIER, 318
BERZELIUS, 3, 168, 171, 226, 249
BETHE, 240
BEUKERS, 163
BEUTELL, 389
BEZZENBERGER, 302
BHATIA, 351
BHATNAGAR, 168-70, 200, 216, 224, 259,
322
BHATTACHARYA, A. K., 220, 346
BHATTACHARYA, A S. f 183-4
BIALEK, 259-60
BICHOWSKY, 308, 311
BIDWELL, 238
BIED, 443
BlEDERMANN, 416
BlEHL, 441
BlERER, 301
BIGELOW, 328
BIJVOET, 262
BIKERMAN, 169, 174, 184, 199, 200, 212,
240
BILKE, 406
BlLLTTZER, 216
BILTZ, 172, 213, 224
BIRCUMSHAW, 379-80
BlRSTEIN, 189
BlSCHOFF, 87
BISHOP, 188, 192
BLAKE, 415-6
BLASCHKE, 389
BLATHERWICK, 87
BLOCK, 100, 115
BLOXSOM, 93-4, 171
BLUMENTHAL, 435
BOCK, 90-1
BODLANDER, 238
BOGDASSARYAN, 149
BOGUE, 363-4, 428-9, 433-6, 438-40
BOHM, J., 250, 255
BOHM, W., 226
BOLAM, 87-8, 140
BOLLEY, 308
BORN, 352, 354
BORODOWSKI, 168
BOSE, 153
BOSEK, 226
BOTTGER, R., 372
BOTTGER, W, 21, 236-7, 258
BOTTINI, 394
BOUCHARD, 202
BOULEZ, 281
BOUTARIC, 172, 176-7, 179, 186-7, 193,
196, 198, 200, 202, 215-6, 219, 248, 318
BOUVJER, 262
BOUYOUCOS, 402, 409-10
BOWLES, 343
BRADFIELD, 381, 399, 401-3, 405, 410, 413,
416, 422, 424-6
BRASE, 282, 292
BRAUER, 282, 292
BRAUN, 239
BRAUNER, 226-7, 234
BREDEE, 141
BREDIG, 7, 153
BREYER, 292
BRIDGMAN, 239
BRIGGS, D. R, 181, 183-4
BRIGGS, T. R., 53, 106, 348
BRINLEY, 95
BRISCOE, 138
BRITTON, 309-10, 369, 371
BROOKS, 232
DE BROUCKERE, 30-3, 39, 113
BROUGHTON, 379-80, 382, 384
BROWN, B. E., 78
BROWN, G. H., 408
BROWN, S. M., 393, 406, 40&-9, 413
BROWNMILLER, 433
BRUCH, 141
INDEX OF AUTHORS
447
BRUNCK, 276
BRUNER, 251, 256
BRUNI, 251-2, 256
BRUNNER, 297
DE BRUYN, 4, 87, 116, 139, 350
BRUZS, 17
BUCHLER, 198
BUCHNER, E. H , 18
BUCHNER, G. f 295
BUDNIKOV, 68-9, 71-3, 312
BUGDEN, 259, 294
BULL, A. W., 53, 106
BULL, H. B , 125, 220
BULLOCK, 149
BUNSEN, 226
BUNTIN, 230, 232, 274
BURGESS, 400
BURSTE1N, 312
BURTON, 179, 188, 192, 194-5, 197, 202
BUSSEM, 434
BuzAcH, 83-4, 219, 349, 375, 379, 381,
386
CAMERON, 78
CAMPBELL, 59
CANN, 364, 395
CAPSTAFF, 149
CARLTON, 138
CASARES, 275
CASPARI, 62, 65
CASSUTO, 199
CAVAZZI, 61, 70
CAWLEY, 284-8, 292
CAYREL, 235
CHAILLAUX, 302
CHAKRAVARTI, D. N , 315-6, 351
CHAKRAVARTI, M. N., 36, 103
CHAMOT, 253
CHANDELLE, 230
CHAO, 36
CHAPMAN, D. L., 110
CHAPMAN, E. E., 270
CHAPMAN, T. S., 213-5
CHARITSCHKOV, 141
CHASSEVENT, 66, 70, 72-4
CHATTERJEE, 316, 318, 323
CHATTERJI, A. C, 88, 137, 312
CHATTERJI, N. G., 27
CHAUDHURY, S. G., 183-4, 196, 201, 316
CHAUDHURY, S. P., 183-4, 194
CHEEK, 364
CHEN, 139
CHERNSTITSKAYA, 214
CHERVET, 308
CHISTONJ, 232
CHODOUNSKY, 167
CHRETIEN, 301
CHRISTY, 297
CHU, 36
CHWALINSKT, 200
CLARK, A B , 315
CLARK, W., 153, 158
CLASSEN, 267
CLERC, 283
CLERMONT, 167
CLOEZ, 72-3
CLOWES, 267
COBB, 69
COEHN, 239-40
COHEN, E, 141
COHEN, M. U., 42
COHN, 223
COLE, 253
COLLANDER, 332, 338, 340
COMBER, 422
COMMENT, 283
CONE, 272
CONINCK, DE, OECHSNER, 89
COOK, E J. R , 59
COOK, G. H., 416
COOKE, 274
COOL, 141
COOPER, 281
COPPOC, 119-21
COPPOCK, 141-2, 234
CRABTREE, 315
CRANE, 101, 105
CREIGHTON, 27, 347
CRENSHAW, 250-1, 253, 255, 260-1, 295-
6,298
CROLL, 292
CROSS, 167
CROUCH, 132, 151
CROWTHER, 410
CUMMING, 305
CURRIE, J. E, 179
CURRIE, L. M., 7, 223, 226-8, 299-301
CUTA, 314
448
INDEX OF AUTHORS
CYR, 286, 293
CYSOUW, 115-6
CZAPEK, 338
DAHMER, 173
DAUS, 253
DAVEY, 233
DAVIDSON, 345
DAVIS, R. O. E., 402
DAVIS, W. A., 65
DAVUIDOVSKAYA, 292
DEACON, 202
DEBIERNE, 34
DEBYE, 110, 175, 375
DECLERMONT, 267
DEISZ, 92
DEMENEV, 167
DEMJANOVA, 232, 235
DENNIS, 276-7
DEPEW, 282
DESACHY, 283
DESAI, B. N , 87-8, 177, 183, 226
DESAI, H. N , 88
DESCH, 67, 81, 442
DETONI, 86
DEUTSCH, 379
DEVAUX, 235
DEWITT, 235
DEYRUP, 24
DHAR, J M , 137
DHAR, N. R., 27, 36, 88, 103, 173, 187,
195, 199, 216, 218, 220, 225, 312, 315-6,
318, 346, 351, 355
DlENERT, 367
VAN DIJK, 236, 251
DIPPEL, 343
DITTE, 70, 223, 228, 248
DIVERS, 278
DODD, 309-10
DOLEZALEK, 90
DOLLFUS, 372
DONALDSON, 87
DONATH, 256
DONAU, 253
DONNAN, 332, 341
DONNINI, 6, 253
DORE, 393, 406, 408
DORFMANN, 193, 207, 217
DORNER, 234
VAN DORP, 259
DORSCH, 443
DE DOUHIT, 281
DOUNIN, 312
DREAPER, 146
DRIFFIELD, 153
DRUSCHKE, 236-7, 258
DUBINSKII, 81
DUBOSC, 227
DUBRISAY, 406
DUCHEZ, 436
DUCLAUX, 174, 316
VAN DUIN, 200
DUMANSKII, 31, 176, 178-9, 197, 201,
219, 230, 232, 274
DUNDON, 19-20, 59
DUPIN, 215
DURHAM, 257, 262
DURRANT, 133
DURRWACHTER, 248~9
DURST, 284
DUSCHAK, 27, 36
DYACHKOV, 43
DYCKERHOFF, 433
EARLEY, 428
EDELBLUTE, 272
EDGAR, 27
EDGERTON, 156
EGERTON, 256-8, 295
EGGERT, 144-5, 150, 238, 245
EHLERS, 143
EHRENBERG, 401
EHRENFELD, 180
EIDNER, 282, 285, 348
EICHORN, 388
EINSTEIN, 144, 178-9
EISNER, 277
EITEL, 159
EKHOLM, 63, 250
ELBERS, 224
ELLSWORTH, 338
ELSCHNER, 86
EMICH, 253
EMLEY, 72
EMSCHWILLER, 285, 287
VON ENDE, 139
ENDELL, 381, 406, 443
ENGLEMANN, 282
INDEX OF AUTHORS
449
EPHRAIM, 273
ERDEY-GRUZ, 103, 105-7, 127, 131
ERDOS, 380
ERRERA, 242, 264
ESCH, 227
ESTRUP, 106
ETTISCH, 203, 218
EWAN, 232
EWELL, 407
EYDMANN, 244
FAJANS, 29, 103-7, 127, 131-4, 137, 144,
146, 148-9, 351-2
FALES, 251-2
FALL, 94
FARADAY, 247
FARBER, 292
FARNAU, 284
FEIGL, 234, 236, 275
FEILMANN, 47
FEITKNECHT, 63
FELDMAN, 144-5
FENIMORE, 262
FERGUSON, C, 315
FERGUSON, J E , 433
FEYTO, 308
FILL, 64
FlNDLAY, 327
FINK, 294, 348
FlRMENICH, 297
FISCHBECK, 234
FISCHER, H., 357-8
FISCHER, P, 294
VON FISCHER, 51, 53
FISHER, 267, 269-70
FLYNN, 282
FoA, 180
FODIMAN, 210
FOLIN, 27
FOLLENIUS, 295
FONTANA, 250
FOOTE, 410
FORDHAM, 308, 312, 330, 372
FORDONSKI, 99
FORSEN, 441
FORSTMANN, 113
FOULK, 23
FOURETIER, 141
FRANKENBURGER, 103, 132, 144, 146, 149
FRANKERT, 318
FRANKOWSKI, 99
FREE, 90
FRERS, 252
FRESENIUS, C. R, 223
FRESENIUS, R , 267
FREUNDLICH, 20, 23-4, 44, 125, 172,
176-7, 181, 183, 187-9, 191, 195, 197-9,
201-4, 206, 213-4, 217-9, 221, 240, 246,
263-4, 266, 323, 347, 373-6, 378-80,
382-4,386
FRIEDHEIM, 167
FRIEND, 141
FRION, 29
FRIZZELL, 126
FROMHERZ, 125, 146
FROM MEL, 167
FRUMKIN, 107
FRY, 402, 406, 411
FUJITA, 338
FUNK, 252
FURMAN, 132
FURSTENAU, 223
FURTH, 184, 199
GADD, 441
GALECKI, 188
GALEROSKY, 140
GALLAGHER, 150
GALLJTELLI, 62
GALLO, 69
GANGULY, P. B , 88, 367
GANGULY, S. C., 183-4, 194
GANN, 186, 202
CANS, 245, 388, 394
GANTSCHEW, 236
GARELLI, 235
GARNER, 332, 341
GARRISON, 144, 146, 285, 289-90
GARY, 443
GASSNER, 442
GATTERER, 351
GAUBERT, 66
GAUDECHON, 290
GAZE, 249
GAZZARI, 388
GAZZI, 174
GEDROIZ, 399
GEIBEL, 224
450
INDEX OF AUTHORS
GEILMANN, 40
GENNERICH, 264
GERASEMOV, 194
GERMANN, 33, 107
GERSTACKER, 348
GESSNER, 271
GHOSH, B. N , 181, 184, 220, 323
GHOSH, D. N., 186
GHOSH, S., 36, 173, 187, 195, 199, 216,
218, 220, 225, 318, 351, 355
Gl AC ALONE, 30
GIBBS, R. C , 277
GIBBS, W. E, 112
GIBSON, 62, 79, 81
GIEBE, 238
GILE, 402, 411
GILL, 67
GILMORE, 364
VAN GILS, 119
GINSBURG, 232
GlRARD, 240
GIVEN, 401
GLASENAPP, 67, 69
GLAUBER, 222
GLAZMAN, 264
GLENDINNTNG, 27
GLIXELLI, 237, 251
GLUUD, 274
GMELIN, 253, 289
GOBEL, 90
GODDARD, 254
GOLDSCHMIDT, 283
GOOD, 305
GORAI, 312
GOROKHOVSKII, 116, 118
GORSKI, 21
GORTNER, 125, 220
GOUY, 110
GRACE, 436
GRAHAM, 1, 171, 315, 328, 349
GRAY, 204, 207, 319
GREENE, C. H, 126
GREENE, H., 397
GREENE, U. T., 127
GRENGG, 69
GRIFFITH, 253, 282
GRIMAUX, 94
GRIMM, 39-42
GROLLMAN, 338
GROSCHUFF, 367-8
GROSVENOR, 294
GROTTHUS, 146
GRUN, 443
GRUNDMANN, 368
GRUNER, 406
GUDDEN, 149
GUINCHANT, 301
GUIOT, 267
GUPTA, R. S., 391, 395-6
GUPTA, S., 200, 224, 259
GURCHOT, 315, 325, 330, 338-9
GUTBIER, 116, 248-9, 278-9
GUTTMANN, 229
GYEMANT, 44, 240
DE HAAN, 125
HAASE, 179, 229
HABER, 24, 255
HABERLE, 173
HABICH, 89
HADDON, 71
HAGIWARA, 17
HAHN, F. L., 36, 267
HAHN, O., 105, 127, 129, 140
VON HAHN, 7, 232, 246-8, 278-9
HALL, V. J., 416
HALL, W. T , 256, 431, 434
HALLER, 350
HAMAKER, 376-8
HAMMETT, 24
HANSEN, C. J., 227
HANSEN, W. C, 72-4, 433
HANTZSCH, 244
HARA, 65
HAROOURT, 232
HARDY, 422
HARING, 272, 274
HARMAN, 362-8
HARRIS, 341
HARRISON, 156
HARTL, 27, 259
HARTLEB, 351
HARTLEY, E. G. J , 6, 327
HARTLEY, G. S., 175
HARTNER, 68
HARTUNG, 143, 305, 331, 334
HASE, 266
HASSEL, 132
INDEX OF AUTHORS
451
VON HASSLINGER, 238
HATSCHEK, 86, 179, 186
HAUG, 90
HAUSAMANN, 298
HAUSER, 375, 379-85
HAUSMANN, 235
HA WORTH, 416
HAZEL, 216, 351
HEATON, 297
HEBBERLING, H, 282
HEBBERLING, M., 276
HEELER, 47
HEINE, 346
HELBRONNER, 283
HEL'D, 43-4
HELLER, 374
HELLMAN, 184
HELLWIG, 5
HELMHOLTZ, 110
HENDRTCKS, 406
HENGEL, 100, 115
HENNIS, 368
HENSILL, 95
HERING, 230
HERSCHELL, 152
HERSHBERGER, 44, 280
HERZ, 229, 271
HERZFELD, 110, 416
HESSLING, 264
HEVESEY, 149
VON HEYDEN, 140
HEYER, 223
HlGGIN, 167
HlLGARD, 401-2
HILSCH, 144
HINRICHSEN, 62, 67, 70
HIRST, 343
HlTTORFF, 238
HOBER, 340
HOCHTLEN, 249, 346
HODAKOW, 131
HODGSON, 159
VAN'T HOFF, 62, 65, 67, 70
HOFFMANN, K., 194-5, 221
HOFFMANN, L., 248
HOFMANN, K. A., 249, 343-6
HOFMANN, U., 381, 406
HOLMES, H. N., 94, 140, 239, 312, 347,
371
HOLMES, R. S., 403, 411
HOLT, 62
HONIGSCHMID, 102, 245
HOOEY, 282
HOROVITZ, 29, 34, 103
HOSKINS, 95
How, 267
HSIUNG, 36
HUCKEL, 110, 175
HULETT, 19, 22, 27, 36
HULOT, 256
HULSE, 276
HUME-ROTHERY, 58-9
HURLEY, 138
HURSCH, 66
HURST, 297
HURTER, 153
HUSE, 156
HUTTIG, 39
IBARZ, 308
IDASZEWSKY, 238
IDETA, 443
IL'INSKII, 91
IMMIG, 232
IMRE, 34-5, 105 109, 127, 129, 140
INSLEY, 407, 434
IPPEN, 297
ISEMA, 140
ISHIZAKA, 202
IVANITSKAJA, 183
IVANOV, 249
IWANITZKAJA, 225
IZAR, 180
JABLCZYNSKI, 90, 99, 125, 138, 202, 225
JACOB, 406
JACOLLIOT, 253
JAKOWLEVA, 173-4, 177, 187, 197
JANDER, 232, 285
JANECKE, 433
JANEK, 200
JANKAUSKIS, 17
JANNASCH, 27
JANSSEN, 110, 113
JANTSCH, 293-4
JASZCZOLT, 138
JEDRZEJOWSKA, 225
JEFFREYS, 252
452
INDEX OF AUTHORS
JENKS, 414
JENNY, 390-1, 393, 396-7, 407-9, 417,
420
JEPPESEN, 380, 386
JEWETT, 49-50
JlRGENSONS, 200
JOACHIMSOHN, 203, 218
JOB, 285, 287
JOCHIMS, 380
JOHNSON, C. R., 138
JOHNSON, E B., 277
JOHNSON, L., 215
JOHNSON, R. N., 79, 81
JOHNSON, S W , 415-6
JOHNSON, W. C , 277
JOHNSTON, 27-9
JOLIBOIS, 67, 70, 141, 262
JONES, C, 153
JONES, R E., 439
JONES, H. C, 72
JORDIS, 234, 368, 371
JORGENSEN, 229
JOHNS, 62
JOSEPH, A. F , 366, 422
JOSEPH, S. M , 276
JOSHI, C. B., 177, 183
JOSHI, S. S , 186, 202, 213, 225
JOULIN, 267
JULIUSBURGER, 380, 384
JUNG, 62-3
JUNK, 395
JUST, 62, 67, 70
JUSTIN-MUELLER, 344
KAHLENBERG, 327-8, 362, 364
KAISCHEW, 40, 236
KALFF, 18
KALLMAN, 375
KAMEYAMA, 312, 342
KAMMER, 34
KANDELAKY, 380
KANDILAROW, 40
KANTER, 371
KARAGUNIS, 146, 148-9
KARAOGLANOW, 27, 36, 38, 40
KARGIN, 175, 210, 214
KARMARSCH, 349
KARSSEN, 262
KATO, H., 252
KATO, Y., 4, 13, 46
KATZ, 305
KAWAMURA, 354
KEANE, 66-7
KEEN, 410
KEENAN, 130, 132, 152
KEESER, 199, 201
KEHOE, 87
KEIDEL, 90
KEIM, 36
KEISERMANN, 70, 435
KELLERHOFF, 306
KELLEY, 393. 399-400, 406-9, 413
KEMPF, 285, 287, 290
KEPPELER, 415
KERMACK, 416, 422
KERNBAUM, 290
KERR, H. W , 395
KERR, P. F., 406
KHAINSKII, 44
KHANOLKAR, 88
KTDA, 139
KlELLAND, 395
KIKUCHI, 138, 230
KIMURA, 386
KING, 410
KING, J. F., 127, 129
KIRCH HOF, 227, 300
KISCH, 141
KlSTLER, 379
KLEIN, A. A , 435, 440
KLEIN, M., 99
KLENKER, 227
KLJMOVITZKAJA, 175
KLING, 252
KLOBUKOW, 295
KOBER, 126
KOCH, 143
KOELSCH, 27
KOETSCHET, 284
KOGEL, 151
KOGERT, 312
KOHLRAUSCH, F., 361-2, 365
KOHLRAUSCH, W., 238
KOHLSCHUTTER, H. W., 141
KOHLSCHUTTER, V., 244
KOHN, K., 89-90
KOHN, M., 349
KOLBL, 224
INDEX OF AUTHORS
453
KOLKMETJER, 100, 115, 262
KOLTHOFF, I. M , 20-1, 26-7, 32-7, 39-
40, 51-6, 59-60, 101-2, 109, 113, 126,
132, 134-8, 234, 236-7, 251-2, 258,
262-3, 308, 311-3
KOLUSCHEWA, 31, 36, 40, 62
DE KONINCK, 272
KONYAEV, 235
KORNFELD, 395
KORTE, 27
KOTTGEN, 396
KOYANAGI, 444
KOZLOV, 91
KRAK, 26
KRAUSS, 62, 64
KREISS, 143
KRESTINSKAJA, 173-4, 177, 187, 197
KROCH, 187
KROGER, 362
KROKOWSKI, 251
KRUG, 372
KRUSS, 248, 278
KHUYT, 44, 101-2, 108-9, 113-tf, 118-20,
125, 172, 183, 192-3, 195, 200-1, 210,
264,322
KUGEL, 355
KUHL, 433, 436, 44(M
KUHN, G., 4
KUHN, W., 375, 380
KUNDU, 101, 106
KURTENACKER, 223
KUSTER, 27, 173
KUZELL, 293
KV1TNER, 90
VAN LAAR, 111
LABES, 101
LACK MAN, 351, 354
LACKS, 106, 139, 200, 351, 354
LACROIX, 66
LAFUMA, 436, 442, 444
LAMB, 133
LAMBERT, B., 58-9, 61
LAMBERT, R. H, 99, 125-6, 130, 132,
152, 157
LANDESEN, 267, 271
LANG, 133
LANGE, B., 176
LANGE, E., 100-2, 105, 111
LARSEN, 237
LARSON, 135
LASSIEUR, A., 252
LASSIEUR, M., 252
LAUB, 173, 175, 177, 193, 224
LAUBENGAYER, 277
LAUER, 136
LAUTERBACH, 344-5, 347
LAWTON, 387
LEA, F. M , 433-4, 439, 442
LEA, M. C, 107, 153, 247
LEATHERMAN, 274
LE BLANC, 253, 278
LE CHATELIER, 70, 371, 431, 434, 444
LECRENIER, 272
LEOENT, 272
LEDERER, 351
VAN LEEUWEN, 94
LEFORT, 232, 245, 254
LEHNER, 14
LEIIRMAN, 300
LENARD, 285, 287-8, 291
LENERY, 222
LENZ, 289
LERCH, 435-6, 438-40
LESZINSKI, 151
LEUZE, 49, 84, 139
LEVI, 250, 346
LEWIS, E. H., 443
LEWIS, E. W., 233
LEWIS, W. K., 379-80
LEY, 241
LHERMITE, 328
LIEBE, 48
LIEBIG, 298, 328, 394
LIEDE, 368
LIESEGANG, 89, 235
LlFSCHITZ, 177
LINCK, 62, 436, 439
LINCOLN, 362, 364
LINDAU, 382
LINDER, 172, 174, 177, 202-3, 216, 224,
236, 241-2, 254, 276, 298
LINGANE, 101-2
LINS, 282
LlSIECKE, 99
LLOYD, 371
LOBLEIN, 443
LOEB, 287
454
INDEX OF AUTHORS
LOFFLER, 295
LOHMAN, 278
LONDON, 375
LONG, 299
LOOFMAN, 406
LOOMIS, A. G, 387
LOOMIS, E. H , 361, 365
LORD, 230
LORENZ, R., 154, 159, 256, 294
LORENZ, W., 137
LOTT, 46
LOTTERMOSER, 3-5, 87, 99-100, 113-4,
116-S, 137, 140, 197, 214, 241-2, 246,
264, 356, 362
LOVELAND, 160
LOVETT, 94
Low, 138
LOWEN STEIN, 305-6
LOWENTHAL, 214
LUCAS, M, 232
LUCAS, R, 278
LUCCHESI, 234, 248-9
LUCKMANN, 276
LUCKOW, 305
LUERS, 186
LUFF, 227, 302
LUHR, 155
LUKOWA, 40
LUPPO-CRAMER, 107, 149, 154-5, 158,
161, 163, 314
LUTHER, 129, 154
LUTZ, 395
McBAiN, 182
McCooL, 410
MCCROSKY, 300
McDoLE, 410
MCDOWELL, C. M., 260
MCDOWELL, S. J., 423
McGEORGE, 400
McMiLLEN, 386
MCMURDIE, 434
MCQUEEN, 216
MAASS, 285, 287, 290
MACINNES, 192, 194-5
MACK, E., 20
MACK, G. L., 200, 264, 316
MACKENZIE, 87
MACNEVIN, 35, 37, 60
MAEGDEFRAU, 406
MAORI, 272
MAIN, 363
MAJUMDAR, 186, 201
MANIERE, 177, 219
MANN, 281
MARESCOTTI, 194
MARIGNAC, 70
MARSH, 402
MARSHALL, 381, 391, 395-6, 406-7, 409
MARTIN, 233
MATHEWS, 174, 177
MATHIEU, 329-30
MATHIS, 371
MATHUR, 322
MATSUMOTO, 91
MATSUNO, 205
MATTSON, 45, 410-2, 417, 421-2, 424
MAWROW, 278
MAY, 116, 197, 214, 356
MAYR, 252
MEES, 154, 162
MEHMEL, 406
MEHROTRA, 220, 323
MEINEKE, 267
MELDRUM, R., 259
MELDRUM, W. B , 138
MELLOR, J. W., 143, 168
MENDELEJEFF, 36
MENEGAZZI, 232
MENEGHETTI, 180, 241
MENON, A. S., 176
MENON, T. M., 186
MENZEL, 39
MENZIES, 231-2
MERWIN, 250-1, 253, 255, 260-1, 295-6,
298, 433
MESSINGER, 133
MESSNER, 308
VANDER MEULEN, 393
MEULENDYKE, 156
MEUNIER, 229
MEYER, E., 5
MEYER, G., 163
MEYER, J. C. A., 282, 284
MICHAEL, 4
MICHAELIS, L., 106, 338
MICHAELIS, P., 167
INDEX OF AUTHORS
455
MICHAELIS, W., 70, 435
MICKWITZ, 267, 274
MIDDLE-TON, A. W , 272, 274
MIDDLE-TON, E. B., 86, 203
MIDDLETON, H. E. f 402, 411
MILANESI, 232
MlLBAUER, 8-90
MILLER, 305
MILLIGAN, 6, 63, 65, 119-21, 250, 255,
268, 270, 274, 295, 315, 318, 408
MITCHELL, 300
MIYAMOTO, 242
MOKRUSCHIN, 167, 232, 235
M0LLER, 395
M6LLER, 100, 115
MOLTZAU, 236, 251, 258, 262
MONCH, 238
MONTGOMERY, 408
MOORE, C. J. f 402
MOORE, W , 94
MORELAND, 72, 74, 79, 81
MOROSOW, 262
MORRELL, 283
MORSE, 327
MOSER, 228
MOURLOT, 226, 267
MOUSON, 416
MOVER, 220
MUCK, 267
MUHLENDYCK, 274
MUKHERJEE, A., 103
MUKHERJEE, J. N., 101, 103, 106, 18$-4,
186, 192, 194, 196, 201, 220, 242, 323,
354
MUKHERJEE, S N., 194
MUKOPADHYAYA, 192
MULDER, 372
MULLER, 67
MULLER, A., 4, 216, 245, 254, 259, 274-5
MULLER, E., 278, 306, 312, 344-7
MULLER, F, 311
MULLER, H., 124, 182
MULLER, J. H., 277
MULLER, W. J., 295
MURMANN, 25, 236
MURPHY, 174, 177
MUTHMANN, 241
MUTTER, 145
MYLIUS, 367-8
NABAR, 87
NACKEN, 64, 433
NAFTEL, 425
NAIK, 88
NAMIAS, 153
NANJAPPA, 225
NARAYAN, 225
NASH, 349
NATHANSOHN, 172, 177, 214, 246,
329-30
NATTA, 231
NAUMANN, 242
NEAL, 241-2
NEIMANN, 6, 83
NELSON, 283
NERNST, 111, 328
NEUBERG, 6, 61, 70, 83
NEUGEBAUER, 71
NEUSCHLOSZ, 217
NEUSSER, 228
NEVILLE, 71-4, 79
NEWHALL, 441
NICHOLAS, 193, 195, 354
NICHOLS, J. B., 48
NICHOLS, M. L., 46
NICLASSEN, 250, 255
NIETZ, 162
NIHOUL, 283
NIKOLOW, 278
NILSON, 168
NISHIZAWA, 290, 293
NISTLER, 350
NODA, 372
NODDACK, 144-5, 150
NOPONEN, 26, 32, 60
OAKLEY, 366, 422
OBER, 202-3, 207, 210
O'BRIEN, 282, 284, 286-9, 291-2
OBRUTSCHEWA, 107
ODELL, 283
ODEN, 16-7, 2fr-7, 57-8, 179, 246
ODINOT, 230
OHL, 271-2
OLSEN, 267, 270
ONORATO, 62
ORESTANO, 232
ORLOWA, 225
ORR, 281, 284
456
INDEX OF AUTHORS
OSBORNE, 26
OSTWALD, Wi., 20, 154, 239, 282, 292
OSTWALD. Wo., 1, 5, 71, 81, 174, 187,
190-1, 194-<>, 198-9, 206, 221, 264, 379
OTA, 372
OTTO, 171
OVERTON, 328, 340
OWE, 47
OYAMA, 44
PAAL, 4, 84, 87, 116, 245
PADOA, 251-2, 256
PALIT, 194
PAN, 273
PANETH, 29, 34, 36, 50-1, 54-5, 60, 103
PANNIKKAR, 213
PAPACONSTANTINOU, 192, 216
PAPISH, 277
PAPKOVA-KWITZEL, 380
PAPPADA, 315, 318, 351
PARKER, F. W. 410
PARKER, H. G., 27
PARKER, T. W. 433-4
PARSONS, 62
PARTRIDGE, 62, 65
PATES, 273
PATTEN, 27
PAULI, 115, 173-5, 177, 193, 195, 207,
224, 397
PAULING, 393, 406
PEARSON, E. A., 234, 236-7
PEARSON, T. G., 275
PECHHOLD, 184
PEEL, 138
PEISAKHOVICH, 163
PfcLABON, 229, 235
PELET-JOLIVET, 106
PERLEY, 153
PERLZWEIG, 338
PERREAU, 193, 196, 216, 219
PERRIN, 110, 240
PESKOV, 115, 173, 180, 216, 242, 347
PETERFI, 373
PFEFFER, 327
PHARAOH, 415
PHILLIPS, 435, 440
PHIPSON, 284, 291
PICTON, 172, 174, 177, 202-3, 216, 224,
236, 241-2, 254, 276, 298
PIERONI, 248
PINE, 129
PlNGREE, 177
PINKUS, 30, 32
PlPEREAUT, 283
PlRQUET, 384
PLATARD, 240
PLATZMANN, 442
PLOTNIKOV, 145
POHL, R., 149
POHL, R W,, 144
POPOFF, 23
PORRITT, 227, 302
PORTER, 56, 106
POSPELOV, 202
POSPELOVA, 202
POULTER. 351
Powis, 181, 183
PRABHU, 202, 225
PRASAD, 200, 216, 224, 259
PRETSCHNER, 155
PRICE- JONES, 382
PROBST, 443
PROSKURNIN, 183
PROST, 259
PROTASS, 116, 118
PROUD, 362
PUGH, 276
PULFRICH, 436, 439
PULLER, 167
PUNNETT, 158
PURI, 410
QUATTRINI, 264
QUITTNER, 90
RAAB, 267
RABINERSON, 355-6, 380, 396
RABINOVICH, 149, 163, 193, 207, 210-1
RAJKUMAR, 183-4, 194
RALEIGH, 256-*, 295
RAMACHANDRAN, 228
RAMANN, 395
RAMMELSBERG, 305
RAMSDELL, 62
RANDALL, 364, 395
RANKIN, 432-3, 435
RAO, 168-70, 261
RAPELJE, 267, 270
INDEX OF AUTHORS
457
RASSUDOWA, 90
RAWITZER, 382
READ, 66
RECKLINGHAUSEN, 379
RECOURA, 46
REED, 379-85
REIHLEN, 344
REINDEL, 349
REINDERS, 107, 139-40, 154, 163
REIS, 21
REITEMEIR, 417, 420
REITSTOTTER, 347
RENFREW, 272
RENWICK, 157-8, 161
RETGERS, 278
REWALD, 6, 70, 83
RICHARDS, J W , 256, 294
RICHARDS, T W , 27, 138
RICHARDSON, A, 143
RICHARDSON, C, 431
RICHTER, F., 103
RICHTER, H , 443
RICHTER, P. W, 283
RlDDSON, 45
RlDEAL, 380
RIEDEL, J. D., 388
RiEDEL, W., 118
RlEDER, 235
RIEGEL, 45, 274
RIEMAN, 111, 36, 42
RIES, 415
RIESENFELD, 179, 229
RlNDFUSZ, 94
RINGER, 278
ROBERTSON, G. R., 440
ROBERTSON, T. B., 207
ROBINSON, C. S., 328
ROBINSON, P. L, 275
ROBINSON, R. H., 95
ROBINSON, W. O., 402-3, 411
ROCHES, 292
RODENBURG, 259
ROGERS, 66
ROHLAND, 66, 70-1, 357, 414, 440
ROLLER, 68, 436, 440
ROMER, 127
ROODVOETS, 183
ROSCOE, 299
ROSE, 226
ROSENBLUM, 34, 51-5
Ross, C. S., 381, 406
Ross, W. J. C., 372
Rossi, 43, 194
ROSSING, 234
ROSTWOROWSKI, 424
ROTHE, 100
ROTHMUND, 395
ROY, 351
RUBIN, 376
RUER, 364
RUFF, 252
RUPPERT, 391
RUSSELL, 380
Russo, 30
RUYSSEN, 44
SABBATANI, 84, 264, 275
SACHER, 294
SACHTLEBEN, 245
SAGORTSCHEV, 40
SAITO, 311
SALVIOLI, 84
SAMBAMUHTY, 244
SAMESHIMA, 141
SAMOKHVALOV, 43
SANDELL, 39-40, 56, 60
SAPGIR, 90, 292
SAPROMETOV, 115, 347
SAUER, 234, 241-2, 244, 391
SAXTON, 410
SCALES, 402
SCANDELLARI, 43
SCARSETH, 412, 424
SCHACKMANN, 42
SCHAFFER, 58, 61
SCHALEK, 375, 379-80, 384
SCHAUM, 152
SCHEELE, 143
SCHEERER, 229
SCHEMJAKIN, 312
SCHERINGA, 236
SCHERTZINGER, 282
SCHIBBE, 238, 245
SCHICK, 253
SCHILLING, 252
SCHILOW, 187, 225
SCHINDLER, 305
SCHIRMER, 91
458
INDEX OF AUTHORS
SCHLEEDE. 286-7
SCHLOSING, 401
SCHMIOT, E., 167
SCHMIDT, F. W., 230
SCHMIOT, H. H., 155
SCHMIDT, O., 382
SCHMIEDESKAMP, 287
SCHNASSE, 252, 258, 270
SCHNEIDER, E. A., 27, 230, 248
SCHNEIDER, F., 36, 42
SCHNEIDER, R., 228
SCHNELLER, 118, 125, 138
SCHOLLENBERGER, 405, 425
SCHOLZ, 177, 217
SCHORAS, 351
SCHORLEMMER, 299
SCHRADER, 143
SCHUCHT, 187-8, 263-4, 266
SCHULZE, GUNTHEB, 389
SCHULZE, H., 3, 143, 172, 187, 223
SCHULZE, J., 252
SCHUMANN, 152
SCHURMANN, 226
SCHWARZ, 368, 371
SCHWEIZER, 234
SCHWIETE, 434
SCOTT, A. F., 138
SCOTT, W. G., 283
SEARLE, 415
SEASE, 161
SEELIGMANN, 267
SEIDEL, 282
SETDL, 130, 132
SEIFERT, 113
SEMELET, 198, 200
SEMLER, 168, 173-5, 207
SEMON, 262
SEN, K. C , 27, 218, 220, 317, 323-5, 338
SEN, N. N, 192, 242
SENARMONT, 271
SERB-SERBINA, 81
SEWRUGOWA, 40
SEXTON, 68
SHANNON, 381
SHAPIRO, 277
SHARPE, 227
SHEPHERD, 432
SHEPPARD, 99-100, 125-6, 130, 132, 144-
6, 149-54. 156-63
SHERRICK, 36, 39, 42
SHERSHNEV, 201
SHIMIDZU, 278
SHORT, A., 227
SHORT, F. C., 327
SHRIVASTAVA, 200, 224, 259, 322
DE'SlGMOND, 399
SlLVERMAN, 34
SlLVERSTEIN, 160
SIMON, 222
SIMONET, 172, 177, 179
SlNGMASTER, 292
SlSLEY, 229
SISTINI, 388
SKRAUP, 343
SLATER-PRICE, 161
SLEGEL, 402
SLOANE, 27
SLOAT, 231-2
SLOTTMAN, 189
SMEATON, 427
SMITH, C. G., 46
SMITH, G. McP., 262
SMITH, G. P., 27
SMOLUCHOWSKI, L., 110
SMOLUCHOWSKI, M., 179, 187, 202
SOKOLOV, 216
SOLACOLU, 443
SOLIN, 197
SOLLNER, 374
SORUM, 351
SPEAR, 368
VAN DER SPEK, 172, 192-3, 195
SPENCER, 362
SPENGEL, 395
SPILLER, 46
SPRING, 167, 228, 235, 241-2
SQUIRES, 379-80, 382, 384
SREBROW, 40, 236
STANEK, 255, 297
STANISCH, 344
STARK, 302
STEEL, 424
STEGEMANN, 34
STEIGMANN, 151, 161
STEINAU, 282, 292
STEINER, D., 234, 241-2, 244
STEINER, W., 132, 149
STEPHEN, 349
INDEX OF AUTHORS
459
STEPHENSON, 386
STERICKER, 363, 366
STERN, A., 144-5
STERN, O, 110, 124
STERNER, 103
STEYER, 312
STIASNY, 245
STILLMAN, 228
STONE, 228
STORCH, 229
STOWENER, 368
STRANSKI, 22
STRATTA, 66
STRECKER, 276
STRIEBEL, 102
DE STUCKLE, 283-5
STUHLMANN, 285
STULL, 256
STUTZ, 282
STUTZEL, 241
SUGDEN, 216
SUNDE, 136
SUZUKI, 141
SVEDBERG, 7, 16, 48, 159, 244, 278
SWIFT, 252
SYRKIN, 69
SZEBELLEDY, 347
SZEGVARI, 350, 375, 384
SZIDON, 259
TAFT, 429
TAKAMATSU, 193
TAMAMUSHI, 386
TAMCHYNA, 219, 221
TAMMANN, 327-8, 332
TANG, 273
TARUGI, 228, 346
TAYLOR, H. G, 14
TAYLOR, H. S., 103
TENDELOO, 179
TEZAK, 29, 31, 126
THAMANN, 87
THIBAULT, 232, 245, 254
THIEL, 27, 271-2, 276
THIELE, H., 290
THIELE, J., 226
THIES, 46
THIMANN, 36, 51
THOMAS, A. W., 213, 215
THOMAS, W. N., 441
THOMPSON, L , 364
THOMPSON, W. 1 , 379-80
THOMSEN, 234, 254
THORNE, 46, 273
THORPE, 348
THORVALDSON, 436
THURING, 436, 444
THWAITES, 283
TINKER, 305, 329-31, 334
TlSSTER, 311
TOCH, 282, 294
TOMASCHEK, 285, 287
TOMICEK, 226
TORTUMI, 65
TOROPOFF, 240
TOTEWA, 36
TOURNEUR, 176
TOWER, 253, 270, 274
TOY, 156, 159
TRAUBE, 70, 81, 326-7, 329
TRAXLER, 107
TREADWELL, F. P., 256
TREADWELL, W. D,308, 311
TREMPER, 441
TREUBERT, 229
TRILLAT, 406
TRIMBLE, 26
TRIVELLI, 144, 150, 153-4, 157-61
TRUMPLER, 238, 245
TUBANDT, 238, 245
VON TUGOLESSOW, 153
TUPHOLME, 66
Twiss, 227
TYSON, 308, 312, 330, 372
ULLRICH, 442
ULRICH, 255
URZHUMSKTI, 194
USHER, 260, 375
UTZINO, 224
VAGELER, 396
VAIL, 361, 368, 389-90
VALENTINE, 222
VALKO, 195, 397
VALKOV, 264
VANINO, 27, 229, 259
VANSELOW, A. P., 395
460
INDEX OF AUTHORS
VANSELOW, W., 144-5, 149
VAN DER VELDE, 139
VENKATARAMAMIAH, 261
VERWEY, 56, 102, 10&-10, 112-3, 115-8,
120, 124, 135-6, 210, 322
VERZIJL, 308, 311
VESTER, 220
ViGJUbSON, 436
VILA, 283
VILESOVA, 235
VILLJERS, 252, 254, 267, 272, 274
VlNNIKOVA, 219
VlSWANATH, 186
VOET, 5, 174
VOGEL, 146
VOGELENZANG, 27
VOGT, 416
VOLKER, 270
VORLANDER, 173, 343, 346~7
VORTMANN, 228
VORWERK, 34, 50-1, 54-5
Voss, 84, 87, 116, 245
VUILLAUME, 172, 176, 179, 186-7, 248
WAELE, 283
WAGNER, E C , 262
WAGNER, G , 39-42
WAGNER, H , 90-1
WALDBOTT, 371
WALDEN, G H., 41-2
WALDEN, P., 184, 327, 332
WALDENBULCKE, 367
WALKER, O. J , 132
WALKER, T. C , 415
WAMPLER, 95
WANNOW, 5, 174, 194-5, 199
WARD, A M , 272, 274
WARD, H. W. D., 295
WARE, 251-2
WASSELIEV, 210
WAUMSLEY, 233
WAY, 388, 394
WEBB, 160
WEDEKIND, 141, 357-8
WEECH, 338
WEGELIN, 306
WEIDMANN, 267
WEIGEL, 173, 269
WEIGERT, 62, 67, 70, 144, 155
VON WEIMARN, 1-3, 11-3, 15-9, 21-4,
36, 46, 57-9, 61, 116, 139
WEIR, 354
WEISER, 2, 6, 14, 23, 27, 36, 39, 42, 46,
56, 63, 65, 72, 74, 79, 81, 86, 93-4,
106, 119-21, 144, 146, 169, 171, 174,
177, 187, 193, 195, 197, 200, 203-4,
207, 213-5, 217-9, 234, 257, 260, 262,
264, 268, 270, 274, 282, 285, 289-90,
306, 310, 315-6, 318-9, 323, 332-4, 354,
368, 408
WEISS, 308, 311
WEISSENBERGER, 68
WEISZ, 152
WELCH, 71
WELLS, 138
WELO, 345
WERETDE, 137
WERNER, D , 57
WERNER, H , 379
WERNER, O., 184
WESSLEY, 267
WESTENDTCK, 72
WESTFALL, 272
WHEATLEY, 277
WHEATON, 389
WHEELER, 416
WHERRY, 381, 406
WHETHAM, 207
WHITE, A II , 62, 65
WHITE, G R, 294
WHITNEY, 202-3, 207, 210
WICKE, 235
WIEGNER, 390, 396, 407, 417, 424
VAN DER WlELEN, 137
WIGGIN'S SONS Co , 74
WIGHTMAN, 150, 153, 158-9
WILKINSON, 318
WILLIAMS, A. M., 87
WILLIAMS, F J., 72
WILLIAMS, G W., 232
WILLIAMS, H. E., 348, 350
WILLIAMS, M. f 216
WILLIAMS, W. R., 402
WILLIAMSON, A. W., 346
WILLIAMSON, T. M., 416
WILLIAMSON, W. T. H, 422
VAN DER WlLLIGEN, 101, 113-5, 118, 183,
264
INDEX OF AUTHORS
461
WlLLSTATTER, 375
WILM, D., 381, 406
WILM, T,226
WlLSEY, 161
WILSON, N.R., 302
WILSON, R, 351
WILSON, S., 300
WINKLER, 276-7
WINSSINGER, 4, 229, 232, 246, 248-9,
254, 263, 275-8
WINTER, 168
WlNTERBOTTOM, 67
WlNTGEN, 177, 179, 214
WlTTEBOON, 137
WOHLERS, 27, 357
WOLFF, 132, 137, 291
WOLSHIN, 354
WOLSKI, 71, 81, 264, 293-4
WOLVEKAMP, 224, 228, 264
WONFOR, 308
WOOD, 281
WOODMAN, 95
WOODWARD, 302
WORK, 283
WOSNESSENSKI, 323
WOULFE, 302
WRIGHT, F. E, 432, 435
WRIGHT, L. T., 240
WULFF, 130, 132
WYROUBOFF, 305, 308
YAJNIK, 351
YAMANE, 79
YASUDA, 139
YOSHIDA, 91
YOUNG, 177, 241-2, 254
YOUTZ, 222
YUTZ, 109, 126, 134, 138
ZACHARTASEN, 255, 277
ZAUN, 184
ZAWADZKI, 251, 256
ZEH, 183, 206
ZlM HERMANN, C., 252
ZlMMERMANN, L, 21
ZlMMERMANN, W, 344
ZlPFEL, 90
ZOCHER, 380-1
VON ZOTTA, 254
ZSIGMONDY, 349, 354, 368
INDEX OF SUBJECTS
Acclimatization, 197
Adsorption, see under several salts
Adsorption indicators, 107, 132-7
applicability, 132-3, 136-7
definition, 132-3
mechanism of action, 133-6
exchange adsorption, 134-5
primary adsorption, 133
Alternating current electrolysis, 256,
260-1
Aluminous cement, 443-4
advantages, 443-4
Aluminum arsenatc sol, 94
Anhydrite, see Calcium sulfate, an-
hydrous
Antimonite sol, 7
Antimony crimson, pigment, 223, 227-8,
299-301
Antimony golden, pigment, 227-8,
301-2
Antimony cfrangc, pigment, 223, 227-8,
299-302
Antimony pentasulfide, 226-7, 301
pigment, 226-7, 301
Antimony red, pigment, 223, 227-8,
299-302
Antimony sulfides, 213-4, 222-8, 301-2
Antimony tetrasulfide, 226-8, 301-2
pigment, 227-8, 301-2
precipitated, 227-8> 301-2
sol, 228
Antimony trisulfide, 21&-4, 222-8, 299-
302
pigment, 222-3, 227-8, 299-302
color, 222-3, 299-301
factors influencing, 299-300
precipitated, 222-3, 227-8, 299-301
Antimony trisulfide sol, 213-4, 223-6
coagulation, 225-6
velocity, 225
color, 224^-6
electrodecantation, 224
Antimony trisulfide sol, formation,
223-4
mutual coagulation, 213-4
precipitation by electrolytes, 225
properties, 224-6
Arsenates, 92-5
Arsenic disulfide, 298-9
pigment, 298-9
Arsenic trisulfide, 3, 167-221, 298-9
pigment, 168-71, 298-9
precipitated, 167-71, 298-9
color, 16&-71, 298-9
physical character, 167-71
Arsenic trisulfide sol, 3, 171-221
acclimatization, 197
action of light, 176-7
action of non-electrolytes, 199-201
adsorption of cations, 202-12, 217-21
charge reversal, 197-9
with alkaloids, 19&-9
with dyes, 198-9
coagulation by electrolytes, 181-213
adsorption and precipitating
power, 200-12
anomalous behavior with KC1-
types, 18&-4
critical ^-potential, 181-6
ion antagonism, 216-21
kinetics, 201-2
mechanism, 202-13
effect of valence, 204-7
precipitation values, 186-99
effect of anions, 195-6
effect of sol concentration, 192-5
of cobalt amines, 204-7
Ostwald's rule, 190-6
Schulze's rule, 187-9, 204-7
Traube's rule, 189-90
velocity, 201-2
coagulation by mixtures of electro-
lytes, 216-21
coagulation by sols, 213-6
463
464
INDEX OF SUBJECTS
Arsenic trisulfide sol, color, 176-7
composition, 174-5
constitution, 174-5, 211-2
density, 177-9
displacement of hydrogen ions, 207-
12
electrodecantation, 173
formation, 171-4
ion antagonism, 216-21
irregular series, 197-9
monodisperse, 173
mutual coagulation, 213-6
hydrophihc, 216
hydrophobic, 213-6
organosol, 174, 184-6
fc-potential, critical, 186
polydisperse, 172
precipitation with mixtures of elec-
tiolytes, 216-21
properties, 175-80
chemical, 179-80
stability, 181-221
therapeutic value, 180
titration, 207-12
velocity of solution, 180
viscosity, 177-9
(-potential, critical, 181-6, 199
Artificial vegetation, 372
Auric and aurous salts, see Gold
Barium carbonate, 4, 83-6
gel, 4, 83
sol, 83-6
coagulation, 85-6
constitution, 84-5
Barium fluoride, jelly, 19
Barium sulfate, 6, 11-48
alcogel, 14
jelly, 13-4, 19
sol, 6, 13-4, 46-8
preparation, 46-7
properties, 47-8
turbidity, 47
Barium sulfate, precipitated, 11-46
adsorption by, 28-43
mechanism, 29-36, 3SM1
of anions, 36-43
of cations, 28-36
effect of constitution, 29-30
Barium sulfate, precipitated, adsorp-
tion by, of commoner ele-
ments, 27-33
of organic ions, 43-44
of radioactive cations, 33
of radium cations, 33
types, 32-3
velocity, 34-5
analytical, 25-6
particle size, effect of digestion, 26
contamination, 28-43
by barium nitrate, 42-3
by potassium premanganate, 39-
43
x-ray study, 40-3
crystal size and solubility, 19-22
electrical properties, 44-6
(-potential, 44-6
growth of particles, 11
law of corresponding states, 17-25
Ostwald ripening, 26
physical character, 11-26
effect of adsorption, 23
effect of concentration of react-
ants, 13-5
form, 13-27
pigment, see Lithopone
precipitation laws, 15-7
velocity equations, 11-3
velocity of precipitation, 11-25
equations, 11-3
Habeas theory, 24-5
von Weimarn's theory, 11-24
Baroid, 387
Base exchange, 388-400
application of process, 399-400
effect of grinding, 393
equations, 396-7
adsorption, 396-9
mass action, 394-5
isotherms, 390-3, 397-9
mechanism, 394-9
phenomena, 3S&-94
Becquerel phenomenon, 239-40
Bentonite, 381-6
thixotropy, 381-6
effect of electrolytes, 382-4
effect of temperature, 384
rheopexy, 384-5
INDEX OF SUBJECTS
465
Bentonite, thixotropy, ultramicro-
scopic changes, 385-6
Beta-dicalcium silicate, 427-43
Bismuth trisulfide, 228-9, 232
precipitated, 228-9
sol, 229
Cadmium arscnate jelly, 93
Cadmium ferricyanide, 313
Cadmium ferrocyanide, 309-10, 312-3
rhythmic bands, 312-3
Cadmium sulfide, 4, 100, 232, 253, 255-
9, 294-7
organosol, 259
Cadmium sulfide, pigment, 259, 294-7
color, 295-7
crystal structure, 296
preparation, 294-5
Cadmium sulfide, precipitated, 100,
232, 253, 255-9, 294-7
adsorption of chloride, 256-8
color, 258-9
crystal structure, 255-6, 296
purity, 256-8
rhythmic bands, 253
Cadmium sulfide sol, 4, 259-60
formation, 259
properties, 259-60
Calcium arsenatc, insecticide, 94
Calcium carbonate, sol, 83-4
protecting colloids, 84
Calcium fluoride, 19-21
jelly, 19
particle size and solubility, 19-21
Calcium orthophosphate sol, 86
Calcium silicate, 371
Calcium sulfate, 15, 19-21, 61-82, 436-
42
hydraulic plasters, 66-79
jelly, 19
particle size and solubility, 19-21
precipitated, 15
sol, 61
Calcium sulfate, anhydrous, 61-70
anhydrite, 66-9
particle size, 68-9
soluble, 66-9
setting, 67-8
solution, 68-9
Calcium sulfate, anhydrous, dehy-
drated hemihydrate, 65-6
x-ray examination, 65-6
Calcium sulfate, dihydrate, 61-82,
436-42
in Portland cement, 436-42
isobaric dehydration, 63-5
Calcium sulfate, hemihydrate, 61-7,
70-S2
dehydrated, 65-6
x-ray examination, 65-6
isobaric dehydration, 63-5
setting, 70-82
colloidal phenomena, 70
mechanism, 70-82
rate, 74-6
effect of electrolytes, 76-7
effect of gypsum nuclei, 74-6,
81
effect of salts, 77-81
retarders, 80-1
von Weimarn's theory, 76-7
utility, 81-2
x-ray examination, 65-6
Calcium sulfate-water system, 61-6
Carbonates, 4, 83-6
Casting clays, 386
Cement, 66-9, 427-44
aluminous, 443-4
blast-furnace Portland, 442-3
Erz, 442-3
Estrich gypsum, 67, 69
hydraulic plasters, 66-9
iron-ore, 442-3
Keenc, 67
plaster of Paris, 66-82
Portland, 427-43
action of calcium sulfate, 436-42
action of other salts, 441-2
composition, 431, 435-40
compounds, action of water on,
435-40
beta-dicalcium silicate, 438
tetracalcium aluminum fcrrite,
439
tricalcium aluminate, 436-8
tricalcium silicate, 438-9
constitution, 431-4
phase-rule study, 433
466
INDEX OF SUBJECTS
Cement, Portland, hardening, 434-42
importance, 428-9
manufacture, 429-31
raw materials, 430
setting, 434-42
Portland blast-furnace, 442-3
Chromates, 4, 20-1, 87-92
Chrome red, 89-91
Chrome yellow, 89-91
color and particle size, 89
Clay, 391-426
acids, 404-5, 425-6
titration, 425-6
adsorbed water, 407-10
adsorption, exchange, 417-22
of anions, 424
base-exchange, 391-400, 412-3; see
also Base exchange
isotherms, 397-9
bentonite, see Bentonite
casting, 386
charge, 417
coagulating power, 420-2
colloidal fraction, 401-5
analysis, 403-5
after electrodialysis, 404-5
before electrodialysis, 403-4
extraction, 402-3
composition, 401-5
constitution, 405-10
mineral constituents, 405-7
water content, 407-10
x-ray analysis, 406-7
crystal lattice water, 409
deflocculation, 422-4
dehydration curves, 408
difference between pcrmuhtes and,
392-3
exchange adsorption, 417-22
exchangeable cations, 412-3
flocculation values, 416-22
and exchange adsorption, 419-22
and ^-potential, 417-22
ionic exchange, 417-8
mineral constituents, 405-7
plasticity, 414-6
effect of grinding, 415-6
effect of humus, 415
effect of particle size, 416
Clay, properties, 410-26
chemical reactions, 424-6
general, 410-4
effect of silica-sesquioxide ratio,
410-2
physical, 410-24
saturation capacity, 412-3
thixotropy, 386-7
applications, 386-7
casting, 386
drilling fluids, 387
latex, 386-7
water content, 407-10
crystal lattice, 409
^-potential, critical, 416-22
and flocculation data, 417-22
and ionic exchange, 417-19
Cobalt arsenate jelly, 93
Cobalt ferricyanide, 312-3
rhythmic bands, 312-3
Cobalt ferrocyanide, 309-10, 312-3
rhythmic bands, 312-3
Cobalt silicate, 371
Cobalt sulfide, 274
colonmetnc analysis, 274
Colloidal forest, 372
Copper metasihcate, 369
Copper silicate, 369, 371
Crystal size and solubility, 19-22, 26
Cupnc ferricyanide, 314, 333-8
gel, 333-8
adsorption of electrolytes, 334-5
and electrolyte solutions, 334-6
and sugar solutions, 333-4
sol, 314
lupric ferricyanide membranes, 334-8
diffusion of salts, 336
negative adsorption of sugar, 333-4
semipermeable, 334-8
Cupric ferrocyanide, 6, 213-4, 309-12,
315-34, 337-8
gel, 6, 306-8, 326-34, 337-8
adsorption of ions, 306-8, 310-11,
332-3
adsorption of water, 305-6, 330-1
and ferrocyanide solutions, 332-3
and sugar solutions, 329-31
rhythmic precipitation, 312-3
INDEX OF SUBJECTS
467
Cupric ferrocyanide membranes, 326-
34, 337-8
negative adsorption of sugar, 330-1
reversible permeability, 338-40
semipermeable, 326-34, 337-8
Cupric ferrocyanide sol, 213-4, 309-12,
315-25
coagulation, 316-25
by alcohols, 325
by single electrolytes, 316-8
by mixtures of electrolytes, 322-5
mechanism, 318-22
constitution, 320-2
formation, 315-6
ion antagonism, 332-5
mutual coagulation, 213-4
properties, 316
titration, 318-22
Cupric sulfide, 3, 234-43
films, 235
organosol, 242-3
dielectric constant and stability,
242-3
sol, 240-3
Cupric sulfide, precipitated, 234-40
Becquerel phenomenon, 239-40
color, 235
contamination by /inc, 236-8
metallic conduction, 238-9
rhythmic bands, 235
Cuprous sulfide sol, 244
Donnan's theory of membrane equili-
brium, 341-2
Doucil, 389
Drilling fluids, 387
Electrodecantation, 115, 173, 224
Electrodialysis of clay soils, 403-5
Emulsions, photographic, see Photo-
graphy
Estrich gypsum, 67, 69
Exchange adsorption, 32, 35-6, 54-6,
120-2, 134-5, 207-12, 309-10,
318-22, 417-22
Ferri- and ferrocyanide membranes,
326-38
Ferric ferrocyanide, see Prussian blue
Ferric sulfide, 275
Ferricyamdes, 312-4, 333-8
Ferrocyanide solution, titration, 308-12
Ferrocyanides, 4-6, 213-4, 305-59
adsorption of ferrocyanide ion, 306-8
gel, 313
rhythmic precipitation, 312-3
sol, 308-13
titration, 308-12
Ferrous arsenate, 93^4
jelly, 93
sol, 94
Ferrous ferricyamde, sec Prussian
blue
Ferrous sulfide, 275
Galena sol, 7
Germanic sulfide, 277
Germanous sulfide, 276-7
Gold sulfides, 248-9
AuS, 248
Au,S*, 248-9
organosol, 249
sol, 248-9
Au,S 3 , 249
Golden sulfide of antimony, 301-2
Gypsum, see alw Calcium sulfate,
dihydrate
Estrich, 67, (.9
ignition, 66-8
in cement, 436-42
Haber's theory of precipitation pro-
cess, 24-5
Hahdes, 4-6, 19-21, 89, 99-166
Hydraulic plasters, 66-9
Estrich gypsum, 67, 69
formation, 66-9
Keene cement, 67
Indicators, adsorption, see Adsorption
Indium tnsulfide, 276
Inorganic soil colloids, see Clay soils
Ion antagonism, 216-21, 322-5, 355
Iridium sesquisulfide, 249
Iron sulfides, 275
Keene cement, 67
King's yellow, 298
468
INDEX OF SUBJECTS
Latent image, see Photography
Latex, thixotropic, 386-7
Lead arsenate, 94-5
insecticide, 94-5
sol, 94
Lead bromide, 138-9
sol, 138-9
protected, 139
Lead carbonate sol, 84
Lead chloride, 138-9
sol, 138-9
protected, 139
Lead chromate, 89-91
pigment, 89-S1
sol, 89
Lead ferrocyamde, 309-10, 312-3
rhythmic bands, 312-3
sol, 309-10, 312
Lead fluoride, 19-21
jelly, 19
particle size and solubility, 19-21
Lead halides, 19-21, 89, 138-40
Lead iodide, 19-21, 89, 139-40
jelly, 19
particle size and solubility, 19-21
rhythmic bands, 140
sol, 89, 138-40
protected, 139
Lead orthophosphatc sol, 86
Lead phosphate, 87
sol, 87
cancer treatment, 87
Lead sulfate, 49-56, 59-60
jelly, 49
sol, 49
Lead sulfate, precipitated, 49-56
adsorption on the lattice, 50-6
effect of H, 53
exchange, 54-6
mechanism, 50-6
Paneth's equation, 50-6, 59-60
extent of surface, 50-4
change with time, 51-4
in storage battery, 49-50
Lead sulnde, 231-3
films, 232
organosol, 233
precipitated, 231-2
adsorption, 231-2
Lead sulnde, precipitated, adsorption,
variation with lattice dimen-
sions, 231
variation with solubility, 231
sol, 232-3
Liesegang phenomenon, 312-3; see
also rhythmic bands, under
the several salts
Lithopone, 280-3, 288-91; see also
Zinc sulnde pigment
darkening, see Zinc sulnde pigment
decolorization process, 290-1
formation, 281-3
mechanism, 281-2
importance, 282
properties, 281
scnsitization by zinc salts, 288-9
Lloyd's reagent, 371
Magnesium carbonate sol, 83
Manganese disulfide, 271
Manganous arsenate, 92-4
jelly, 92-4
sol, 92-4
von Weimarn's theory, 92-3
Manganous ferricyanide, rhythmic
bands, 312-3
Manganous ferrocyanide, 309-10, 312-3
rhythmic bands, 312-3
Manganous sulnde, rose and green,
267-71
crystal structure, 270-1
transformation, 268-71
effect of alkali sulfides, 26&-71
effect of ammonia, 26&-71
mechanism, 269
x-ray analysis, 270-1
Membrane equilibrium, theory, 341-2
Membranes, 334-52
cupric ferri- and ferrocyanide, see
Cupric
reversible permeability, 338-40
of cell wall, 340
of cupric ferrocyanide, 338
semipermeable, 326-38
definition, 326-7
theory, 327-29, 337-8
adsorption, 329
atomic sieve, 327, 337
INDEX OF SUBJECTS
469
Membranes, semipermeable, theory,
solution, 328-9
ultrafilter, 327, 337
Mercuric carbonate sol, 84
Mercuric chloride, 141-2
Mercuric halides, 141-2
Mercuric iodide, 141
Mercuric sulfide, 3, 242, 260-6, 297-8
organosol, 242
pigment, 262, 297-8
Mercuric sulfide, precipitated, 260-3,
297-8
color, 261-2
contamination by third-group sul-
fides, 262-3
crystal structure, 261-2
formation, 260-2
by alternating-current electroly-
sis, 260-1
Mercuric sulfide sol, 3, 242, 263-6
adsorption of cations, 264-5
adsorption reversal, 266
coagulation by electrolytes, 264-5
formation, 263-4
Mcrcurous halides, 141-2
Mineral constituents of soils, 405-7
Molybdenite sol, 7
Molybdenum disulfidc, 278
Molybdenum pentasulfide, 278
Molybdenum trisulfide, 277-8
Mosaic gold, 302
Mutual coagulation process, 213-6,
355-8
Negative adsorption, 330-5, 337
Nephelometric titration, 138
errors in, 138
Nernst-Noyes equation, 12, 77
Nickel ferricyanide, 312-3
rhythmic bands, 312-3
Nickel ferrocyanide, 309-10, 312-3
rhythmic bands, 312-3
Nickel silicate, 371
Nickel sulfide, 271-4
precipitated, 271-2
forms, 271-2
velocity of solution, 272
sol, 272-4
colorimetric analysis, 274
Nickel sulfide, sol, stability, 273-4
test for nickel, 274
Optical sensitization, see Photography
Orpiment, 298
Osmium tetrasulfide, 249
Ostwald ripening, 6, 26, 53, 99-100, 126
Ostwald's activity coefficient rule,
190-6
Palladium monosulfide, 249
Paneth-Fajans' rule, 29, 31-8, 103-8,
353
Halm's modification, 105
Patriotic tube, 347
Permutites, 388-400, see Base ex-
change
difference between clay and, 392
isotherms, 390-3
Phosphates, 86-7
Photochemical decomposition, 143-51,
285-9
of silver halides, 143-51 ; see also
Photography
of zinc sulfide, 285-94; see also Zinc
sulfide
Photography, silver halides in, 100,
143-63
development, see below, latent
image
latent image, 144, 153-7, 160-3
development, 161-3
adsorption theory, 162
chemical theory, 162
nature, 153-6
origin, 155
optical sensitization by adsorption,
147-51
of bromide ions, 147-8
of colloidal silver, 149
of dyes, 150-1
of silver ions, 147-9
photochemical decomposition, 143-
51
cause of darkening, 143-4
mechanism, 144-5
photographic emulsions, 100, 143-
63
470
INDEX OF SUBJECTS
Photography, silver halides in, photo-
graphic emulsions, ripening,
100, 157-9
sensitivity substance, 157
types, 152
photographic sensitivity, 157-61
action of silver iodide, 160-1
mechanism, 159-61
ripening of emulsions, 157-9
sensitivity substance in emul-
sions, 157
Pigments, 89-91, 168-71, 223, 226-8,
254, 259, 262, 280-302, 348
antimony, 223, 227-8, 299-301; see
also Antimony and Anti-
mony sulfides
arsenic, 168-71, 298-9; see also
Arsenic sulfides
barium sulfate, 280-3, 288-91; see
also I-ithopone
cadmium sulfide, 294-7; see also
Cadmium sulfide
characteristics, general, 280-1
chrome red, 89-91
chrome yellow, 89-91
King's yellow, 298
lead chromate, 89-91
lithoponc, 280-3, 288-91; see also
Lithopone
mercury sulfide, 262, 297-8
mosaic gold, 302
orpiment, 298
Prussian blue, 348
realgar, 298
stannic sulfide, 302
vermilion, 297-8
preparation, 297-8
zinc sulfide, 254, 283-94, see also
Lithopone and Zinc sulfide
Plaster of Pans, see Calcium sulfate,
hemihydrate
in cement, 440-2
Plasters, hydraulic, see Cement and
Hydraulic plasters
Plasticity of clays, see Clay
Platinum disulfide, 249
Platinum family sulfides, 249
Portland cement, see Cement
Potentiometnctitration,iodide-silver,102
Precipitation process, 2-3, 15-25, 57-9;
see also von Weimarn's
theory
mechanism, 57-9
velocity, 11-25
Prussian blue, 4, 6, 34S-58
pigment, 348
Prussian blue gel, 6, 343-9
applications, 348
composition, 343-7
formation, 343-7
"insoluble," 347, 349
rhythmic bands, 312-3, 347-8
"soluble," see Prussian blue sol
Prussian blue sol, 4, 347-58
adsorption of, 357-8
adsorption of ions by, 352-4
and heat of hydration, 352-4
effect of mobility, 354
coagulation by electrolytes, 351-5
acids, 354-5
effect of concentration, 354
mixtures, 355
salts, 351-4
formation, 349-50
ion antagonism, 355
mutual coagulation, 355-8
Paneth-Fajans' rule, 353
properties, 350-1
Realgar, 298
Rheopexy, 384-5
Rhythmic precipitation, 312-3 , see also
under the several salts
Rubber, lead sulfide in vulcanization,
233
Schulze's rule, 29, 34, 37, 187-9, 204-7
Selenium sulfide, 278-9
Silicate, 361-444; see also Clay
beta-dicalcium, 427-43
cement, see Cement
garden, 372
sol, 372-3
thixotropy, see Thixotropy
tricalcium, 427-43
Silicate gel, 368-72, 388400
artificial vegetation, 372
base-exchange, 388-400; see also
Base exchange
INDEX OF SUBJECTS
471
Silicate gel, colloidal forest, 372
composition, 368-71
electrometric titration, 370-1
silicate garden, 372
Silicate solutions, sodium, 361-8
boiling point, 364-6
conductivity, 362-3
diffusion, 367-8
freezing point, 364-6
hydrolysis, 363-4
titration, 369-71
transport numbers of ions, 366-7
vapor pressure, 364-6
Silicates, 361-444; see also Clay
Silver bromide, 19, 99-110, 126-30, 317-
8 ; see also Silver halides
jelly, 19
sol, 137-8
Silver bromide, precipitated, 100-10,
127-30
adsorption by, 100-10, 127-30
of dyes, 107, 130-7
as indicators, 107, 132-7
of inorganic ions, 127-30
of thorium B, 127-30
desensitization by copper salts, 129
Silver carbonate sol, 84
Silver chloride, 19, 126-31, 137-8, see
also Silver halides
in nephelometric titration, 138
errors in, 138
jelly, 19
sol, 137-8
Silver chloride, precipitated, 100-10,
127-30
adsorption by, 100-10, 127-30
of dyes, 107, 130-7
as indicators, 107, 132-7
of inorganic ions, 127-30
Silver chromate, 4, 20-1, 87-9
precipitated, 20-1, 87-9
particle size and solubility, 20-1
rhythmic bands, 89
sol, 4, 87-9
precipitation process, 87-8
Silver ferncyamde, 312-3
rhythmic bands, 312-3
Silver ferrocyanide, 312-3
rhythmic bands, 312-3
Silver ferrocyanidc, sol, 309-10, 312
Silver halides, 4-6, 19, 89, 99-138, 143-
73; see also the bromide,
chloride, and iodide
in photography, see Photography
photographic emulsions, 100, 143-63
precipitated, 99-110, 126-30
adsorption by, 100-10, 127-30
Ostwald's ripening, 99-100, 126
sol, 137-8
Silver iodide, 5, 19, 89, 99-122, 161-2;
see also Silver halides
cubic, 100, 115
hexagonal, 100, 115
in photographic emulsions, 100, 161-2
jelly, 5, 19
Silver iodide, precipitated, 99-110,
115-6, 132-7
adsorption by, 100-10
mechanism, 108-10
of actinium, 109
of dye ions, 104, 107-8, 132-7
as indicators, 107, 132-7
of erythrosin, 108
of inorganic ions, 103-7
of iodide ions, 100-3
of iodine, 107
of potential determining ions,
100-3, 115-6
of silver ions, 100-3
of thorium B, 105, 109-10
Paneth-Fajans' rule, 103-8
crystallization, 100, 115
isoelcctnc point, 101-2, 118
precipitation mechanism, 99
reversible flocculation and defloc-
culation, 99
Silver iodide sol, 5, 89, 101-2, 110-25
adsorption by, 115, 118-25
mechanism, 115
of cations, 120-2
of lead ion, 121-2
of thorium B, 121-2
charge and stability, 116-8
coagulation, 118-25
by electrolytes, 118-25
mechanism, 123-5
velocity, 125
electrical double layer, 110-3
472
INDEX OF SUBJECTS
Silver iodide sol, electrical double
layer, formation, 111-3
nature, 110-1
isoelectric point, 101-2, 118
mobility of particles, 119-20
negative, 114
potential reduction, 124-5
mechanism, 124
preparation, 113
peptization, 113-6
by common ions, 113-5
by other ions, 115-6
protected, 116
stability, 114-8
titration, 120-2
Silver orthophosphate sol, 86
Silver sulfate, precipitated, 15
Silver sulfide, 4, 232, 244^8
precipitated, 244-5
sol, 4, 245-8
coagulation, 246-7
color, 246-8
formation, 245-6
"violet value," 247-8
Silver thiocyanate sol, 137-8
Sodium chloride sol, 4
Sodium silicate, 361-8; see also Sili-
cate gels, sols, and solutions
Soil colloids, inorganic, see Clay
Sols, see also under the several salts
general methods of formation, 2-7
condensation, 2-5
dispersion, 5-7
titration, 120-2, 207-12, 309-12, 318-
22
Sphalerite, 285
Stannic sulfide, 229-31, 302
pigment, 302
precipitated, 229-30
sol, 230-1
Strontium carbonate sol, 83
Strontium sulfate, 15, 19-21, 56-60
jelly, 19
particle size and solubility, 19-21
sol, 57
Strontium sulfate, precipitated, 15,
57-60
adsorption on the lattice, 59-60
Strontium sulfate, precipitated, extent
of surface, 60
mechanism of precipitation, 57-9
size of particles, 58
von Weimarn's theory, 57
Sulfates, 6, 11-81, 436-42
Sulfides, 3-4, 100, 167-304
Tellurium disulfide, 279
Tellurium trisulfide, 279
Tetracalcium aluminum f errite, 427-43
Thalhc sulfide, 276
Thallous bromide, 129-30; see also
Thallous iodide
Thallous iodide, 129-30
adsorption by, 129-30
of dyes, 129-30
of inorganic ions, 129
of thorium B, 129-30
Thallous sulfide, 276
Thixotropy, 373-87
and adhesion, 374-9
and coagulation of sols, 374-9
and electrolyte coagulation, 376-7
factors determining, 373-86
hydration, 379-30
mutual attraction and repulsion,
373-7
particle size and shape, 380-1,
384H3
of bentonite, 381-6
effect of electrolytes, 382-4
effect of temperature, 384
rheopexy, 384-5
ultramicroscopic changes, 385-6
of clays, 381-7
applications, 386-7
Baroid, 387
clay castings, 386-7
drilling fluids, 387
latex, 386-7
Traube's rule, 189-90
Tricalcium aluminate, 427-44
Tricalcium silicate, 427-43
Turnbull's blue, see Prussian blue
Tungsten trisulfide, 278
Uranium ferrocyanide, 311, 318
sol, 318
INDEX OF SUBJECTS
473
Vermilion, 297-8
"Violet value/' 274-8
Water, action on cement, 435-^40
Water content of soils, 407-10
von Weimarn's theory of the precipi-
tation process, 2-3, 11-25, 57,
76-7, 92-3, 127
law of corresponding states, 17-25
generalized expression, 22-5
limitations, 23-5
simplified expression, 17-22
limitations, 19-22
precipitation laws, 15-7, 127
Wurtzite, 285
Zinc ar senate jelly, 93
Zinc ferri cyanide, 312-4
rhythmic bands, 312-3
Zinc ferrocyanide, 213-4, 309-13
adsorption of ferrocyanide, 310-1
adsorption of water, 305
mutual coagulation, 213-4
rhythmic bands, 312-3
sol, 309-11
Zinc sulfide, 6, 236-9, 250-4, 283-94
Zinc sulfide pigment, 253-4, 283-94;
see also Lithopone
crystal structure, 285
darkening in light, 284-94
cause, 284-5
mechanism, 287-8
moisture in, 286-7, 289-90
prevention, 291-4
action of cobalt salts, 293-4
rules, 291-4
relation to phosphorescence, 287
sensitization by water, 289-90
sensitization by zinc, 288-9
decolorization process, 290-1
formation, 283-4
photochemical decomposition, 285-94
conditions for, 285-7
Zinc sulfide, precipitated, 250-3
contamination by second-group
sulfides, 252-3
crystal structure, 250
rhythmic bands, 253
solubility, 251
Zinc sulfide sol, 253-4
protected, 254