RETARDERS FOR CONCRETE, AND THEIR
EFFECTS ON SETTING TIME AND
SHRINKAGE
DECEMBER 1972 - NUMBER 51
BY
YASUHIKO YAMAMOTO
JHRP
JOINT HIGHWAY RESEARCH PROJECT
PURDUE UNIVERSITY AND
INDIANA STATE HIGHWAY COMMISSION
TECHNICAL REPORT STANDARD TITLE PAGE
1. Report No.
1. Government Accession No.
3. Recipient's Catalog No.
4. Title and Subtitle
RETARDERS FOR CONCRETE, AND THEIR EFFECTS
ON SETTING TIME AND SHRINKAGE
5. Report Date
December 1972
6. Performing Organization Code
7. Author(s)
Yasuhiko Yamamoto
8. Performing Organization Report No.
JHRP-51-72
9. Performing Organization Name and Address
Joint Highway Research Project
Civil Engineering Building
Purdue Universit
10. Work Unit No.
11. Contract or Grant No.
We
roue university
st Lafayette, Indiana 47907
HPR-l(lO) Part II
12. Sponsoring Agency Name and Addres*
Indiana State Highway Commission
100 N. Senate Avenue
Indianapolis, Indiana 46204
13. Type of Report and Period Covered
Interim Report
14. Sponsoring Agency Code
15. Supplementary Notes
Prepared in cooperation with Federal Highway Administration and
U. S. Department of Transportation
16. Abstract
Sixty- five pure chemicals and three proprietary retarders were
tested in the penetration test for setting time at concentrations
at or around 0.1% of the cement. The most effective retarders had
several, closely-grouped oxygen atoms in hydroxyl, carboxyl , or
carbonyl groups. Some of the retarders were used in the
fabrication of 0.2 x 0.2 x 4-in. bars of cement paste, which
were subsequently oven-dried and the shrinkage, non-evaporable
water content, and specific surface of the paste was determined.
The retarders caused a change in the shrinkage that was paralleled
by a change in the specific surface of the paste.
The contents do not necessarily reflect the official views or
policies of the Federal Highway Administration. This report does
not constitute a standard, specification, or regulation.
17. Key Words
Concrete, retarders, shrinkage,
mi cros tructure
18. Distribution Statement The contents Qf tnig
report reflect the views of the
author who is responsible for the
facts and the accuracy of the data
presented herein. *
19. Security CloniC (of this report)
20. Security ClaisiC (of this page)
21* No. of Paget
181
22. Price
Form DOT F 1700.7 (8-69)
Digitized by the Internet Archive
in 2011 with funding from
LYRASIS members and Sloan Foundation; Indiana Department of Transportation
http://www.archive.org/details/retardersforconcOOyama
Interim Report
RETARDERS FOR CONCRETE, AND THEIR EFFECTS ON SETTING
TIME AND SHRINKAGE
TO: J. F. McLaughlin, Director
Joint Highway Research Project
FROM: H. L. Michael, Associate Director
Joint Highway Research ProjeC
December 28, 1972
Project: C-36-47L
File: 4-6-12
The attached Interim Report titled "Retarders for
Concrete, and Their Effects on Setting Time and Shrinkage"
is submitted on the HPR Part II Research Study "Effect of
Retarders on Volume Changes in Portland Cement Concrete".
The Report has been authored by Mr. Yasahiko Yamamoto under
the direction of Professor W. L. Dolch.
The purpose of this work was to gain some insight into
effective molecular structure of retarders and to examine the
shrinkage behavior of cement pastes as affected by the
addition of retarders. Principle findings include a
determination that effective in retarder molecules are
hydroxyl, carbonyl and carboxyl groups and that retarders
cause a moderate increase in the shrinkage of cement paste.
The Report is submitted in partial fulfillment of the
objectives of this Study. It will also be forwarded to the
ISHC and FHWA for their review, comments and acceptance.
Respectfully submitted,
Harold L. Michael
Associate Director
HLM:ms
cc: W. L. Dolch
R. L. Eskew
W. H. Goetz
M. J. Gutzwiller
G. K. Hal lock
R. H. Harrell
M. L. Hayes
C. W. Lovell
G. W. Marks
R. D. Miles
J. W. Miller
G. T. Satterly
C. F. Scholer
M. B. Scott
J. A. Spooner
N. W. Steinkamp
H. R. J. Walsh
E. J. Yoder
Interim Report
RETARDERS FOR CONCRETE, AND THEIR EFFECTS ON SETTING
TIME AND SHRINKAGE
by
Yasahiko Yamamoto
Graduate Instructor in Research
Joint Highway Research Project
Project No.: C-36-47L
File No. : 4-6-12
Prepared as Part of an Investigation
Conducted by
Joint Highway Research Project
Engineering Experiment Station
Purdue University
in cooperation with the
Indiana State Highway Commission
and the
U.S. Department of Transportation
Federal Highway Administration
The contents of this report reflect the views of the author
who is responsible for the facts and the accuracy of the data
presented herein. The contents do not necessarily reflect the
official views or policies of the Federal Highway Administration
This report does not constitute a standard, specification, or
regulation.
Purdue University
West Lafayette , Indiana
December 28, 1972
Ill
ACKNOWLEDGEMENTS
The writer wishes to express deep appreciation to his
major professor, Dr. W. L. Dolch, for his kind guidance and
understanding during the course of this work. Dr. S.
Diamond and Dr. C. F. Scholer should be also appreciated
for their valuable suggestions and helps.
The discussion with Mr. D. N. Winslow was helpful and
meaningful in solving many difficulties. Some technical
phases of this work were done by the assistance of Mrs.
T. R. Brendel. The writer is thankful to them and also to
many other people who have helped him in many respects
toward the accomplishment of this work.
This research was sponsored by the Indiana Highway
Department and administrated by Joint Highway Research
Project of Purdue University. The writer wishes to thank
the authorities for the support of this work.
Finally, the writer is thankful to Dr. M. Kokubu and
Dr. H. Okamura, professor and associate professor, re-
spectively, University of Tokyo, Japan, for their continu-
ous encouragement in pursuing the advanced degree in the
United States.
TABLE OF CONTENTS
Page
LIST OF TABLES vii
LIST OF FIGURES viii
ABSTRACT xiii
CHAPTER I - INTRODUCTION 1
CHAPTER II - REVIEW OF LITERATURE 6
Interaction of Retarder with Cement 6
Effective Molecular Structure of Retarder and
Retardation Mechanism 13
Drying Shrinkage of Paste, Mortar, and Concrete
as Affected by Retarders 17
CHAPTER III - MATERIALS 21
Admixtures 21
Cement 2 7
Sand 2 9
Water 2 9
CHAPTER IV - EXPERIMENTAL WORK 30
A - Setting Time Measurements 30
General Remarks 30
Apparatus 31
Container 31
Penetration Resistance Apparatus 31
Mixer 31
Preparation of Mortar Specimens 33
Mixing Proportions and Size of Batch 33
Mixing Procedures 33
Casting and Storage of Specimens 34
Penetration Test Procedure 34
TABLE OF CONTENTS, cont.
Page
B - Measurement of Shrinkage of Cement Pastes ... 35
General Considerations 35
Apparatus 36
Mold for Cement Paste Specimens 36
Gauge Studs and Length Comparator 36
Mixing Apparatus 38
Preparation of Cement Paste Specimens 40
Mixing Proportion and Concentration of Retarders 40
Mixing, Casting, and Curing of Cement Paste
Specimens 42
Measurements of Length and Weight, Method of
Drying 44
C - Determination of Degree of Hydration of Cement
Pastes 46
Loss on Ignition Test 46
Derivation of Equations for Determining Non-
Evaporable Water and Evaporable Water 47
Assumptions 47
Notation 48
Derivation of Equations 50
Reduction of Data 51
D - Surface Area Measurements of Cement Pastes . . 52
General Remarks 52
Apparatus and Procedures of Measurement 53
E - Scanning Electron Microscope Observation ... 54
General Procedures 54
CHAPTER V - RESULTS 56
Infrared Spectra of Commercial Retarders .... 56
Non-Evaporable Water Content of Fully Hydrated
Cement Paste 56
Result of Setting Time Experiments 59
Commercial Retarders 59
Pure Chemicals 66
Effects of Retarders on the Drying Shrinkage of
Cement Paste at Various Stages of Hydration . . 72
TABLE OF CONTENTS, cont,
VI
Page
Drying Shrinkage of Mature Cement Pastes When
Dried at 50% Relative Humidity 7 9
Water Content of Cement Pastes 84
Specific Surface Area of Cement Pastes 87
Scanning Electron Microscopy of Cement Pastes . . 96
CHAPTER VI - DISCUSSION 109
Molecular Structure of Retarders 109
Effect of Retarders on the Hydration of Cement . 12 8
Effect of Retarders on the Shrinkage of Cement
Pastes 143
CONCLUSIONS 153
LIST OF REFERENCES 155
APPENDICES
Appendix A: Correction Curve for Warping of
Cement Paste Bars 161
Appendix B: Summary of Setting Time Experiments
When Pure Chemicals Were Added 163
Appendix C: Shrinkage of Vacuum Oven-Dried Cement
Pastes at Various Stages of Cement Hydration . 169
VITA 181
LIST OF TABLES
Table Page
1. List of Chemicals 22
2. List of Commercial Retarders 27
3. Composition and Properties of Lab Cement
No 319 28
4. Concentration of Retarders Added to
Cement Pastes 41
5. Non-Evaporable Water Content of Bottle Hydrated
Cement (W/C =10) 58
6. Summary of Relative Initial Setting Time of
Mortars When Pure Chemicals Were Added .... 6 7
Appendix
Tables
7. Summary of Relative Setting Time of Mortar
Samples When Pure Chemicals Were Added .... 164
Vlll
LIST OF FIGURES
Page
Figure 3
1. Calibration Curve for Proctor Penetrometer
(ASTM C403) 32
2. Mold for Cement Paste Specimens, Cement Paste
Sample, and Paste Caster
3. Length Comparator 39
4. Apparatus for Drawing Mixing Solution from
Funnel into Mixing Container 43
5. Schematic Diagrams of Cement and Cement Paste . 49
6. Infrared Spectra of Commerical Retarders ... 57
7. Penetration Resistance vs Elapsed Time for
Mortar with Retarder L 60
8. Penetration Resistance vs Elapsed Time for
Mortar with Retarder A 61
9. Penetration Resistance vs Elapsed Time for
Mortar with Retarder S 62
10. Effect of Concentration of Retarder L on
Setting Time of Mortar 63
11. Effect of Concentration of Retarder A on
Setting Time of Mortar
12. Effect of Concentration of Retarder S on
Setting Time of Mortar
13. Penetration Resistance vs Elapsed Time for
Mortar with 1, 3-Dihydroxy-2-Butanone 73
14. Concentration of Admixture vs Relative Initial
Setting Time of Mortar (No. 1) 74
15. Concentration of Admixture vs Relative Initial
Setting Time of Mortar (No. 2) 75
IX
LIST OF FIGURES, cont.
Figure Page
16. Shrinkage of Vacuum Oven-Dried Cement Pastes
at Various Stages of Hydration (Commercial
Retarders) 76
17. Shrinkage of Vacuum Oven-Dried Cement Pastes
at Various Stages of Hydration (Pure Chemicals) 77
18. Shrinkage vs Drying Time for Mature Cement
Pastes When Dried at 50% Relative Humidity
(Commercial Retarders) 80
19. Shrinkage vs Drying Time for Mature Cement
Pastes When Dried at 50% Relative Humidity
(Pure Acid Chemicals) 81
20. Shrinkage vs Weight Loss of Mature Cement
Pastes When Dried at 50% Relative Humidity
(Commercial Retarders) 82
21. Shrinkage vs Weight Loss of Mature Cement
Pastes When Dried at 50% Relative Humidity
(Pure Acid Chemicals) 83
22. Non-Evaporable Water vs Evaporable Water of
Cement Pastes (Commercial Retarders) 85
23. Non-Evaporable Water vs Evaporable Water
of Cement Pastes (Pure Chemicals) ....... 86
24. Non-Evaporable Water vs Curing Age of
Cement Pastes (Commercial Retarders) 88
25. Non-Evaporable Water vs Curing Age of
Cement Pastes (Pure Chemicals) 89
26. Specific Surface Area of Vacuum Oven-Dried
Cement Pastes When Expressed on the Basis of
Total Ignited Weight of Sample (No. 1) . . . . 90
27. Specific Surface Area of Vacuum Oven-Dried
Cement Pastes When Expressed on the Basis of
Total Ignited Weight of Sample (No. 2) .... 91
28. Specific Surface Area of Hydrated Portion of
Vacuum Oven-Dried Cement Pastes
(Commercial Retarders) 93
>:
LIST OF FIGURES, cont.
Figure Page
29. Specific Surface Area of Hydrated Portion of
Vacuum Oven-Dried Cement Pastes
(Pure Chemicals - No. 1) 94
30. Specific Surface Area of Hydrated Portion of
Vacuum Oven-Dried Cement Pastes
(Pure Chemicals - No. 2) 95
31. Unhydrated Cement Grains (X5000) 97
32. Cement Paste With No Admixture (X3000) ,
Age: 1 min 97
33. Cement Paste With No Admixture (X9000) ,
Age: 1 min 98
34. Cement Paste With Calcium Lignosulfonate
(X5000) , Age: 1 min 98
35. Cement Paste With Citric Acid (X3000) ,
Age: 1 min 99
36. Cement Paste With Sucrose (X5000) ,
Age: 1 min 9 9
37. Cement Paste With No Admixture (X5000) ,
Age: 5 min 101
38. Cement Paste With Calcium Lignosulfonate
(X5000) , Age: 5 min . . 101
39. Cement Paste With Citric Acid (X3000) ,
Age: 5 min 102
40. Cement Paste With Sucrose (X5000) ,
Age: 5 min 102
41. Cement Paste With Citric Acid (X3000) ,
Age: 10 min 103
42. Cement Paste With Citric Acid (X8000) ,
Age: 10 min 103
43. Cement Paste With Citric Acid (X3000) ,
Age: 3 0 min 104
XI
LIST OF FIGURES, cont.
Figure Page
44. Cement Paste With Citric Acid (X3000) ,
Age: 1 hr 104
45. Cement Paste With No Admixture (X5000) ,
Age: 1 hr 106
46. Cement Paste With Sucrose (X3000) ,
Age: 30 min 106
47. Cement Paste With No Admixture (X2000) ,
Age: 12.5 hr (degree of hydration: 24%) .... 107
48. Cement Paste With Glycolic Acid (X2000) ,
Age: 15.5 hr (degree of hydration: 26%) .... 107
49. Cement Paste With Retarder A (X2000) ,
Age: 14 hr (degree of hydration: 18%) 108
50. Cement Paste With Retarders (X2000) ,
Age: 13 hr (degree of hydration: 20%) 108
51. Retardation Ability vs Complexing Ability
of Chemicals 135
52. Effect of Continuous Vacuum Oven-Drying on
Shrinkage of Cement Pastes 145
53. Effect of Continuous Vacuum Oven-Drying on
Weight Loss of Cement Pastes 146
Appendix
Figures
54. Correction Curve for Warping of Cement Paste
Bars 162
55. Shrinkage of Vacuum Oven-Dried Cement Pastes
at Various Stages of Cement Hydration
(No Admixture) 17 0
56. Shrinkage of Vacuum Oven-Dried Cement Pastes
at Various Stages of Cement Hydration
(Retarder L) 171
57. Shrinkage of Vacuum Oven Dried Cement Pastes
at Various Stages of Cement Hydration
(Retarder A) 172
LIST OF FIGURES, cont.
Figure Page
58. Shrinkage of Vacuum Oven-Dried Cement Pastes
at Various Stages of Cement Hydration
(Retarder S) 173
59. Shrinkage of Vacuum Oven-Dried Cement Pastes
at Various Stages of Cement Hydration
(Glycolic Acid) 174
60. Shrinkage of Vacuum Oven-Dried Cement Pastes
at Various Stages of Cement Hydration
(Sucrose) 175
61. Shrinkage of Vacuum Oven-Dried Cement Pastes
at Various Stages of Cement Hydration
(3-Hydroxy-2-Butanone) 17 6
62. Shrinkage of Vacuum Oven-Dried Cement Pastes
at Various Stages of Cement Hydration
(Hydroquinone) 177
63. Shrinkage of Vacuum Oven-Dried Cement Pastes
at Various Stages of Cement Hydration
(CaCl2-2H20) 178
64. Shrinkage of Vacuum Oven-Dried Cement Pastes
at Various Stages of Cement Hydration
(Effect of Concentrations of Citric Acid) . . . 179
65. Shrinkage of Vacuum Oven-Dried Cement Pastes
at Various Stages of Cement Hydration
(effect of Water-Cement Ratio) 180
Xlll
ABSTRACT
Yamamoto, Yasuhiko. Ph.D., Purdue University,
December 1972. Retarders for Concrete, and Their Effects
on Setting Time and Shrinkage. Major Professor: W. L. Dolch.
The main purposes of this work are to gain some insight
into effective molecular structure of retarders and to
examine the shrinkage behavior of cement pastes as affected
by the addition of retarders.
A total of sixty five organic chemicals were tested
for their retarding abilities by means of setting time
experiment on mortar specimens by the penetration method.
The chemicals were added mostly at the concentration of
0.1% of cement by weight. Three commercial retarders that
are chemically classified in the three principal categories
were also tested at several concentrations. The mix pro-
portions of the mortars were kept constant; water-cement
ratio and sand-cement ratio being 0.5 0 and 2.75, respective-
ly. The effectiveness of the chemicals as retarders was
judged by comparing the initial setting time of a sample
with that of a control sample without the admixture.
Effect of the concentration of admixture was studied for
the commercial retarders and some chemicals. These results
XIV
were discussed in connection with the effective molecular
configurations of retarders and possible retardation
mechanisms.
Small cement paste bars (1/5 x 1/5 x 4 in.) were pre-
pared for shrinkage measurements. Admixtures used in this
part include five chemicals that were found to be retarders
in this work, the three commercial retarders, and CaCl-^H-O
(accelerator) . Concentration of each retarder was selected
in such a way that it gave 50% of additional retardation
compared with a control mortar. Water-cement ratio was
fixed at 0.40. Drying of the samples was conducted in a
vacuum oven at 105 to 110°C for 2 4 hours. Shrinkage was
compared at various degrees of cement hydration; the ages
of samples were as early as lOhr after mixing to 8 months,
at which time 90% or more of the cement was hydrated.
Determinations of specific surface area, non-evaporable
and evaporable water contents, and scanning electron micro-
scopic observation were made on cement paste samples to
explore whether or not some changes were induced in the
cement hydrates by the addition of retarders.
The following are main findings in this research.
1. Hydroxyl, carbonyl, and carboxyl groups are all
effective in retarder molecules. Strong retarder molecules
contain many of these groups in such a way that the oxygen
atoms are constrained to approach each other closely.
XV
Substitution of other more weakly electronegative groups
for these groups results in greatly reduced effectiveness.
2 . Retarders cause a moderate increase in the shrink-
age of cement pastes at ages when more than about half
hydration is attained in about 4 days. At lesser ages the
shrinkage may be increased or decreased.
3. When cement paste are severely dried of all their
evaporable water, they exhibit shrinkage behavior that
shows a minimum at about half hydration and a maximum
earlier. A similar change is observed in the specific
surface area of the hydrated cement. Some cause and effect
relationship exists between these quantities.
CHAPTER I - INTRODUCTION
Retarders are classified as type B or type D chemical
admixtures for portland cement concrete in ASTM Designation
C494. Their primary function is, as the name implies, to
retard the setting time of concrete. Many problems of hot
weather concreting, mass concrete, multi-lift concreting,
continuous placement, and others have been helped by the
use of retarders (1, 2)*.
As a secondary effect, many retarders reduce the water
requirement to produce a concrete of desired workability and
produce a concrete of higher strength with equal cement con-
tent. Some retarders, additionally, produce many small air
bubbles, much the same as those produced by air-entraining
agents in concrete mixtures. Hence, the use of retarders is
also beneficial to other properties of concrete such as
workability, strength, and durability (3) .
Commercial retarders are classified chemically into
three categories:
1. lignosulfonic acids and their salts
2. hydroxycarboxylic acids and their salts
3. carbohydrates (sugars)
* Numbers in parenthesis refer to references at the end of
this paper.
When these retarders are added to concrete mixtures,
some kind of physical and/or chemical reaction takes place
in such a way to retard the setting time. Little is known
of what the effective molecular structures of the retarders
are and how the retardation mechanism proceeds.
As will be shown in Chapter II, the most widely be-
lieved theory of the retardation mechanism is the so-called
"coating theory". The idea of the theory is basically as
follows: As soon as retarders are added they are adsorbed
at the solid-water interface of cement grains or their
hydration products. The sorbed retarder molecules block
or delay further contact of the unhydrated cement with
water and therefore retard the hydration of the cement.
The effective molecular structure of retarders, then,
has to be of such a configuration that the coating is most
efficiently attained. Most workers believe that large
chain molecules with many hydroxyl groups are essential to
effective retarders. The -OH groups are for bonding with
cement grains, presumably by hydrogen bonding to surface
oxygen atoms.
However, owing to the variety of both the compounds
in portland cement and molecular structures of retarders,
it is difficult to imagine that the theory mentioned above
is the only mechanism for the retardation.
Some workers have reported that retarders increase
the shrinkage of concrete (4). On the other hand, other
workers observed no increase in shrinkage with the addition
of retarders (3) . Because the shrinkage behavior of
concrete is extremely complex and is influenced by many
factors, it is not surprising to see this discrepancy.
Yet, it is of importance to establish basic information as
to whether retarders really change the shrinkage behavior
of concrete and, if they do, how much the change is and how
it should be considered in practice. If sound aggregates
are used in concrete, change in shrinkage behavior of the
concrete can be attributed to that of the cement paste in
the concrete. Therefore a detailed study on cement paste
should provide such information.
Hardened cement paste consists of capillary pores,
cement gel, and unhydrated cement grains (5) . Each of them
has a certain role in the shrinkage behavior of cement
paste. The extent to which these phases exist and, there-
fore, the structure of the cement paste depends on the
degree of hydration. Since retarders change the rate at
which a cement hydrates, a consideration of the magnitude
of these phases is appropriate when drying shrinkage of
retarded cement paste is compared with that of unretarded
paste.
Additionally, there is evidence that the morphology of
hydrated C,A* is altered significantly by the addition of
The usual abbreviations are used through the paper:
C = CaO, S - Si02, A = A1203, H = H20.
lignosulf onate , a constituent of some retarders (6). So
there is some possibility that retarders also change the
specific surface or crystal habit of the hydration products
of cement paste and thereby induce different shrinkage
behavior of the paste.
Hence, what seems important and necessary is to examine
the effect of retarders on the structure of the hydration
products at various hydration stages and to interpret the
observed shrinkage behavior of cement pastes in these terms.
An alternate hypothesis, based on the coating theory,
is that retarders act merely by changing the rate of, or
time of initiation of, the hydration reaction. A combina-
tion of the kinetic and the structural change theories is,
of course, also possible.
The present paper discusses the results of experiments
undertaken to gain information on the action of retarders.
By means of setting time experiments on mortar specimens,
many chemicals were examined for their effects as retarders.
Possible effective molecular structures of retarder mole-
cules are discussed. Small cement paste bars were made
with and without retarders. Their shrinkages were measured.
The results are compared at various degrees of hydration of
the cement, ranging from as early as 10 hours after mixing
with water to 8 months of age, at which time the pastes
were almost mature.
Determination of evaporable and non-evaporable water
contents, measurement of specific surface area, and scann-
ing electron microscopic observations were conducted to
examine the properties and the structure of the cement
pastes.
CHAPTER II - REVIEW OF LITERATURE
Interaction of Retarder with Cement
Blank, Rossington and Weinland (7) observed that the
adsorption from aqueous solutions of salicylic acid and
calcium lignosulf onate on C2S and C~S was negligibly small
and showed the following order of the adsorption on the
cement compounds from aqueous solution in 10 to 15 minutes,
calcium lignosulf onate:
C3A > C.AF > Type I portland cement > C3S, C„S > 0
salicylic acid:
C3A > C4AF > TyPe x Portland cement > C2S = C3S = 0
The adsorption on the compounds from ethyl alcohol solution,
however, was in the following order for salicylic acid:
C3S > C2S > Type I portland cement > C^A = C.AF = 0
From these results, they suggested that the negligible ad-
sorption on C2S and C~S in aqueous solution is probably a
result of the competitive adsorption of water, and that
certain modified properties of cement due to admixture addi-
tion are a result of the preferential adsorption of the
admixture by C,A and C,AF , thereby rendering them inactive
and allowing the main cementing compounds (C..S and C„S)
to control the hydration reaction.
Seligmann and Greening (8) showed that one gram of
C^A could remove 99% of the sugar from 5 cc of a 1% sucrose
solution within 7 minutes. The presence of gypsum or
calcium hydroxide did not interfere. C^A also sorbed
considerable calcium lignosulf onate , whereas C^S and C.AF
sorbed little. It may be added that a rough calculation
using these results showed that the sugar adsorption on
unhydrated C-.A corresponded to a uniform coverage of about
150 molecular layers. The result is unrealistic, and it
would be adequate to suppose that the sucrose was adsorbed
by the hydration products of C~A instead of or in addition
to the unhydrated material as Diamond (9) observed with
salicylic acid.
Stein (10) observed that monomer organic anions
exerted a relatively small influence on C,S hydration, but
a pronounced effect on C,A hydration by forming calcium
double salts with C^A, similar to ettringite. Hansen (11)
has a similar concept that the hydration of C,A is retarded
by chemisorbing, rather than adsorbing, lignosulonate .
The chemisorption results from the formation of double
salts between the lignosulfonate and calcium ions in the
surface of cement minerals.
Mielenz and Peppier (12) , however, pointed out that
formation of compounds isostructural with sulf oaluminates
and containing such large anions as lignosulfonate would
be impossible. They suggested that retardation of C-.A
hydration by materials like lignosulf onates may better be
explained in terms of adsorption of lignosulf onate upon
C,A grains or precipitation of insoluble calcium ligno-
sulfonate around the grains of C^A.
The hydration of C-.A is widely believed to proceed
in two steps in the absence of gypsum:
2C-.A + (n+8)H •> C0AH0 + C.AH ; metastable phase
J z o 4 n / , , .
(hexagonal)
C0AH0 + C.AH ■* 2C_AH_ + (n-4)H ; stable phase
(cubic)
Daugherty and Kawalewski (13) examined the effect of many
chemicals on the relative amount of hydration products of
C-,A by quantitative x-ray diffraction analysis. They found
that, in general, organic acids permitted the hydration of
C-.A to proceed to C-,AHfi while sugars and their derivatives
blocked the hydration reaction at metastable C~AH8 and
C.AH . However, all the sugars and sugar derivatives
examined accelerated the initial hydration of CUA at the
concentration level of 0.25 mole % relative to C,A. They
speculated on the retardation mechanism of C-.A hydration:
"In concentrations of 0.3 mole % or less relative to
C3A the organic materials complex or chelate with
available calcium ions (and other cations) in solution
and on the surface of the CoA. These reactions might
affect the rates of solution of C3A and the solution,
precipitation or nucleation of the hydration products
in a manner which accelerates hydration.
After the available calcium is tied up, the
organic material is adsorbed on the surface of C^A.
This adsorption may be chelation with calcium in the
C3A,, complexation, or surface adsorption depending
upon the type of organic molecule
Organic compounds with more insoluble calcium
salts would be expected to form more closely packed,
more impervious sheaths around the C^A. This might
prevent H20 from further attacking the C3A. The
organic layers could prevent the intermediate products,
C2AHs an<^ c4AHn/ from transforming to C^AHg.
Retardation of hydration would be the result."
Young (6) made similar observation with lignosulf onate
by measuring refractive indices of the hydration products
of C^A. Lignosulfonate stabilized C4AH13 and C^AHg with
respect to C,AHg . When gypsum was present, the formation
of ettringite was not affected by the addition of ligno-
sulfonate, and the final hydration products were low sulfo-
aluminate, C„AHR , and C.AH..-.. Another interesting observa-
tion by Young is that there was no evidence of the trans-
formation of C_AH0 and C.AH, _. to C0AH, when C-.A, even in
the absence of lignosulfonate, hydrated at low temperature,
say less than 30°C.
Diamond (14, 15) analyzed salicylate bearing precipi-
tates observed in the C^A (or C,AH,) - salicylic acid-water
system. They were amorphous salicylate-aluminum complexes
and did not contain calcium.
Because the C^A and C,S phases in cement are responsible
for early hydration of the cement, the papers introduced so
far give an impression that the C\A is mainly concerned
with the retardation mechanism when retarders are added.
Forbrich (16), however, demonstrated that salicylic acid
delayed the time of maximum heat generation due to hydra-
tion of C3S by 8 to 10 hr relative to a reference mixture,
10
and that calcium lignosulf onate shifted the peak more than
27 hours. On the other hand, the hydration of C3A with
12% gypsum interground was accelerated by the addition of
salicylic acid, but was retarded remarkably when calcium
lignosulf onate was added.
Polivka and Klein (17) examined the setting time of
many cements influenced by various admixtures and showed
that retarders are more effective when used with cements of
low alkali and low C,A. Forbrich (16) also showed that
cements low in C,A and high in C,S were the ones retarded
to the greatest degree by either calcium lignosulf onate or
salicylic acid.
Bruere (18) showed that delayed addition of retarders
(citric acid and calcium lignosulf onate) produced a con-
siderably longer setting time of cement paste. This effect
was much greater in pastes made from ordinary portland
cement, which had a high C^A content, than in pastes made
from low heat cement, which had a low C^A content. Dodson
and Farkas (19) observed the same result. Bruere suggested
that this effect was due to competition between the gypsum
in the cement and organic retarders to combine with or
adsorb on C3A surfaces and explained further:
"When cement is pre-mixed with water for a few minutes,
gypsum has ample time to dissolve and coat the C3A.
Consequently, when the retarder is added to the pre-
mixed paste, the C3A is unable to adsorb it and a
large amount of retarder is available to retard the
silicate hydration reactions."
11
Seligmann and Greening (8) examined the effects of
sucrose and lignosulf onate on the hydration of various
cement constituents by x-ray diffraction and observed that
sucrose accelerated the reaction between gypsum and C^A;
but lignosulf onate had little, if any, acceleration effect.
Lignosulf onate inhibited the release of Ca (OH) 2 from C,S
and also suppressed the release of alkali into the liquid
phase. The alkali is supposed to react chemically with
the retarder to destroy its hydration inhibiting effect.
They concluded:
"Materials that cause retardation of set, such as
sucrose or lignosulf onates, can also produce a large
initial acceleration of the hydration reactions. This
acceleration may be caused by inhibition of lime
release by the silicates phases or by increased
reactivity of the C3A with gypsum, resulting from the
very high sorption of the additive in the C3A surface.
The early acceleration and high sorption can be avoided
by delayed addition of the retarder; the retarding
effect is then markedly increased."
Kalousek et a_l. (20) analyzed extracts from various
clinkers at 7 minutes and 2 hr after mixing and observed
that sucrose was the only added material (the others being
gypsum, CaCl- , calcium acetate, fluosilicic acid, TDA,
tannic acid, and triethanolamine) that increased the
basicity of the extract. This was observed not only by
increased extraction of alkali, but also, to a greater
extent, by almost complete conversion of dissolved sulfate
to insoluble products. Thus sucrose accelerated the early
hydration involving the alumina-bearing constituents of the
clinkers.
12
Tamas (21) assumed that retarders would behave in the
opposite manner to calcium chloride and suggested that
retarders formed an adsorbed layer on the grain of cement
silicates, causing de-activation of the reaction between
the silicate phase and water. He also suggested the
possibility of the formation of insoluble compounds between
retarders and the aluminates in cement.
An interesting result was shown by Ramachandran (22) .
He examined the adsorption and desorption isotherms of
calcium lignosulf onate from both aqueous and non-aqueous
solutions on C3S , hydrated C3S, and lime. It was the
hydrated CUS, but not C,S, that was responsible for a
perceptible amount of adsorption of calcium lignosulf onate
from the aqueous solution. Because the adsorption was
irreversible, he suggested surface complex formation
(chemisorption) involving the silicate surface, calcium
lignosulf onate, and water.
It appears that silicate phases in portland cement can
not be neglected in considering the interaction between
retarders and cement and, therefore, the retardation
mechanism in cement paste. Long ago, Steinour (23) said,
"Indeed, in cements with good contents of gypsum in which
setting is due to hydration of tricalcium silicate, it is
logical to assume that in order to obtain further retarda-
tion it is the reaction of the silicate that must be slowed
up."
13
From the review of literature in this section, the
following summary of information is meaningful:
1. C~A, C.AF, hydrated C,S , and, possibly, hydrated
CUA adsorb retarders from solution.
2. Retarders that have chemically different structures
act differently in their effects on cement hydra-
tion.
3. Not only the aluminate phase but also the silicate
phase in cement plays a significant role in the
setting process.
4. Retarders are more effective when they are added
to cement of low C-.A and low alkali content.
Effective Molecular Structure of Retarder and
Retardation Mechanism
Hansen (24) found that most of the organic compounds
that retarded the hydration of oil well cements had a common
structural feature of the H-C-OH group. He later suggested
that large anions of retarders could react by ionic bonding
with the calcium ions in the surface of cement minerals
and that OH groups react by hydrogen bonding with oxide ions
of the minerals. Thus the organic compounds block the
contact of cement granules with water molecules and so
inhibit hydration (11) .
Steinour (25) showed that retarders characterized by
undissociated OH groups retarded B-C-S hydration. Because
these retarders are effective at small concentrations, and
14
because the OH groups can form hydrogen bonds, he assumed
that they retard the hydration of cement by an adsorption
mechanism (23, 26).
Taplin (27) added a large number of inorganic and
organic chemicals to cement paste to find out the effective
molecular structure of retarders. He found that the group
HO-C-C=0 was usually present in organic substances that
retard the hydration of portland cement, and that the
effective molecules usually contain at least two oxygen
atoms each bound to a single but different carbon atom in
such a way that the oxygen atoms can approach each other.
However, there were some exceptions. He also showed that
location and arrangement of OH groups in molecules changed
the retardation power widely. His proposed retardation
mechanism is the adsorption of these chemicals (i) on the
surface of clinker minerals so as to protect them from
attack by water and (ii) on the surface of a coherent coat-
ing of hydration products so as to prevent transport of
materials to or from the clinker surface.
Danielson (28) made cement pastes with thirteen
different organic calcium salts and measured the setting
time of the pastes. He concluded that the retarding capa-
cities of the salts were related to the characteristics of
the added molecule in a way that suggests a confirmation of
the hypothesis proposed by Hansen and many other workers,
namely that retardation is due to adsorption of the mole-
cules on cement grains.
15
Previte (29) measured the end point of the dormant
period in the course of hydration of cement paste by iso-
thermal calorimetry and showed that saccharides of larger
molecular weight generally were more powerful as retarders
when they were added on an equimolar basis. However, the
rate of alkaline degradation was another factor affecting
the set retardation effectiveness. He concluded that the
effect of alkalinity in the aqueous phase of hydrating
cement reduces the set retardation efficiency of alkaline-
degradable saccharides.
Bruere (30) examined why the dissaccharide a, a-
trehalose had been found to possess only weak retarding
ability, while all other sugars studied were powerful set-
retarders. Only a, a-trehalose failed to produce any
trace of cuprous oxide in boiling Fehling's solution at
all alkali concentrations used, whereas all the other
sugars produced significant amounts of cuprous oxide in the
test solution. He postulated that the latter could hydro-
lyse significantly in the highly alkaline conditions in
cement paste to produce reducing sugars that in turn could
be converted to saccharinic acids containing the HO-C-C=0
group, and thus he supported Taplin's theory even in the
case of saccharides.
But according to Daugherty and Kawalewski (13) ,
a, a-trehalose is a powerful retarder for the hydration of
C-jA, and it retards the hydration to the same extent as
16
sucrose. They also found that OH groups tend to retard the
hydration of C,A and carbonyl (C=0) groups tend to accele-
rate the hydration of C^A, but the a-hydroxyl carbonyl
(HO-C-C=0) group itself did not appear to be important.
Koide (31) stated that carboxyl (COOH) group has always
retarded the hydration of CA (the principal component of
high-alumina cement) . The effect was especially marked when
the group was contained in a chelating compound (example;
EDTA) that produce sexadendates of high stability.
In a review, Young (32) postulated that organic re-
tarders adsorb on calcium hydroxide nuclei in cement pastes
and poison their future growth. Retardation is the result
of this delayed formation of calcium hydroxide. He further
explained that higher levels of calcium hydroxide super-
saturation are required to overcome the effects of the
retarders .
Summarizing this section, most workers believe that the
retardation of cement hydration is a result of the adsorp-
tion of the retarder molecules on the surface of cement
grains or hydration products. The adsorption may be by
hydrogen bonding. However not much is known of the effect
of the molecular structure of the retarders. The different
results for a, a-trehalose found by different workers
indicate that the behavior of retarders in cement paste
may not necessarily be the same as that in pure compound
systems.
17
Drying Shrinkage of 'Paste, Mortar, and Concrete
as Affected by Retarders
When retarders (or water reducing admixtures) are
added to concrete, the subsequent drying shrinkage is
generally thought to increase (4, 33). On the other hand,
Wallace and Ore (3) examined the test results reported by
many field laboratories and concluded that these admixtures
neither increase nor decrease the shrinkage of concrete.
This uncertainty of the effect of retarders on the shrinkage
of concrete arises from the many complicated factors that
affect shrinkage (34, 35). Tremper and Spellman (4)
demonstrated that the cumulative effects of some factors
on shrinkage could be expressed by the product, not the
sum, of individual effects. Hence, the true effect of
retarders on the shrinkage of concrete has to be examined
carefully so that indirect influences are avoided. How-
ever, only a few papers have been published concerning the
shrinkage behavior as affected by retarders.
Danielson (28) observed that the shrinkage of cement
paste in an atmosphere of 6 0% R.H. was, in general, increas-
ed by the admixtures studied. The increase was consider-
able when benzonate and salicylate, which were the only
aromatic salts examined, were added.
Tremper (36) dried mortar specimens that were 4 days
old for 4 days in an atmosphere of 50% R.H. At low SO^
contents of the cement, the admixtures (lignosulf onate and
hydroxylated carboxylic acid) increased drying
18
significantly. At higher S03 contents the trend was re-
versed, and in most cases the drying shrinkage decreased
relative to that of mortars containing no admixture. Thus
he suggested that the optimum gypsum content in cement
should be increased to minimize the shrinkage when ad-
mixtures are added.
In cement paste, gypsum reacts with C3A and forms the
high-sulfate form of calcium sulfoaluminate hydrate
(ettringite) in early stages of hydration. This reaction
is known to be expansive. According to Lerch (37) , for
cements of low alkali content, those of high C^A content
require larger additions of gypsum to give minimum subse-
quent shrinkage than do those of low C3A content. For
cements of the same C3A content, those of high alkali
content require larger addition of gypsum than do those of
low alkali content. Therefore, the observation by Tremper
could be related to some modified rate of hydration of
the cement caused by the retarders.
Another conceivable cause could be changes in the
structure of the hydration products. Young (6) observed
that no new hydrate was formed in C,A hydration by the
addition of calcium lignosulf onate, but a modification of
the crystal habit of the usual hydrates took place;
acicular structures were formed instead of hexagonal
plates of C-AHg and C.AH,3. When gypsum was present, the
formation of high sulfoaluminate was not affected by the
19
admixture. However, the final hydration products (low
sulfoaluminate, C„AH , and C.AH..-) were changed from their
usual hexagonal plate form.
Danielson (28) examined the properties of cement paste
affected by organic admixtures at an age of 28 days. The
amount of combined water and the total heat of hydration
were not altered significantly by the admixtures. The
compressive strength was also either unaffected or some-
what lowered. The reduction in Young's modulus was un-
expectedly small compared with the reduction in compressive
strength for all the pastes to which dibasic salts were
added. He attributed this result to a modified structure
of hydration products.
On the other hand, Seligmann and Greening (8) showed
that sucrose and lignosulf onate did not affect the nature
of the hydration reaction of cement; only the rate was
influenced. Prior and Adams (1) mentioned that differ-
ential thermal analysis studies of hydrated cement at the
ages of 7 and 2 8 days showed no significant change in the
identity or proportion of the hydration products of cement
with the addition of lignosulf onates.
By means of a scanning electron microscope, Schwiete,
Ludwig, and Sieler (38) examined the morphology of the
hydration products of cement pastes at the ages of 1, 7,
14, and 28 days. There was no difference observed among
them regardless of the addition of different kinds of
20
retarders. These workers also confirmed, by means of
x-ray diffraction, that retarders changed only the rate
of hydration of the cement.
21
CHAPTER III - MATERIALS
Admixtures
A total of sixty-five chemicals were tested in the
setting time experiments to examine the effects of the
molecular structure of the retarders. They included a
variety of organic chemicals - many aliphatic compounds,
aromatic compounds, chelating agents, dyes and others.
About one half were organic acids. Their names, chemical
formulae, and producer's names are listed in Table 1.
The three commercially available proprietary retarders
L, A, and S were used throughout the work. Each represents
one of the three main categories of organic retarders, i.e.,
calcium (also, sodium or ammonium) lignosulf onates , hydroxy-
carboxylic acids, and carbohydrates, respectively. Since
the commercial retarders were supplied in solution form,
their solids content was determined by oven-drying in the
usual way. Density and amount of solid content of each
commercial retarder are shown in Table 2. It should be
noted that concentrations of all retarders are, hereafter,
discussed in terms of weight of the solid component per unit
weight of cement in the mixture.
22
a
id
cu
E
o
u
e
<D
A
CJ
a
c
id
■d
m
M
Sh
B
0)
1
<D
Q)
CO
H
(0
01
W
<d
rH
in
■H
•H
w
rt!
w
&4
PL,
u u
p
-u
T3
t>
SH
0
0
<u
U
•
M
M
X
03
-*
<d
y
xj
U
£
X!
PQ
G
o
<J$
a
u
cj
•rH
■H
•H
*
■H
•H
•
H
H
•
rH
M
i-l
E-t
H
TJ
CJ
H
^3
>9
TJ
(0
rH
•
1
h
rH
•
S
<
s
X
<c
<
h)
I
M
O
fa
-H
Id
CJ
•H
g
J
U
a
X
X
a
co
CN
CN
CN
<N
a
o
o
o
o
o
D
CCj
a
CJ
a
cj
u
u
CM
m
<N
3
CN
CN
CN
r>J
ro
*»
CN
a
a
a
CN
O
X
o
^-,
. — .
^— >
a
a
X
H
X
P0
u
CJ
o
CJ
CJ
u
CN
(N
CN
o
2
w
U
•
a
X
a
y
a
X
(N
X
a
3S
eg
CN
(N
CN
o
u
o
o
o
o
o
IB
o
u
U
X
X
a
a
a
o
X
a
CJ
a
X
U
—
- — •
■ — ■
CJ
u
u
u
cj
u
u
CJ
id
u
CJ
as
X
X
a
X
a
a
a
a
X
X
nc
2
a
X
CN
CN
CN
<N
CN
CN
(N
(N
CN
CN
cn
CM
CN
CN
(N
o
O
o
o
o
o
o
O
O
o
o
o
O
O
o
cj
u
cj
u
cj
cj
u
u
CJ
u
cj
o
CJ
CJ
CJ
id
u
■H
H
0)
a;
CJ
*j-i
o
■p
(0
•H
CD
2
D1
•H
rH
— »
«—
*"*
§
a1
_
-a
•H
■H
aj
•H
TJ
r-i
-P
, — »
o
0
Tl
id
M
a1
rij
T3
•H
T5
^*v
■H
U
Q
•H
CJ
>i
•
rH
•H
' *
0
*C
8
cr
M
*C
CJ
'
■H
rH
in
-P
T3
•rH
CJ
r)
ri
"•^
oo
3
o
TJ
T3
n
•3
X>
■H
P
•-H
^-s
m
<
•rH
•H
TJ
•H
•H
■H
-P
(1)
U
13
i
o
O
•H
CJ
O
tJ
CJ
a)
o
<
■H
•d
2
u
<:
<
O
<
<
•H
<
o
id
O
•H
i
■H
<
O
id
rl
u
<
CJ
>i
c
u
u
CJ
CJ
<
CJ
0
0
-H
<
X
0
•H
•H
O
■H
•H
-H
0)
■P
H
H
o
0
rH
C
rH
•H
a
M
0
H
a
CU
J3
>i
•r(
o
n
<d
o
i
c
•H
id
-H
O
-H
m
CJ
X
>
•H
^
g
rJ
0
CJ
•p
0,
U
CJ
CJ
0
0
3
■P
>1
0
■P
H
CJ
3
-H
>i
>1
H
a
>1
M
o
a
-p
u
fm,
s
3
rH
3
rH
rH
CD
o
r^
>i
id
i
CD
m
1
CO
o
o
o
S
2
O
CM
J
a
X
Eh
"-*
23
>i
a
rtj
»
o
o
-P
O
u
X
o
a
u
•H
M
XI
CJ
■H
t3
X!
CJ
•H
>H
"8
u
X
0
c
•H
X!
cj
■H
1-1
T3
a
G
cs
M
nj
$
0)
X!
5
oi
05
01
<T5
•H
rd
W
fa
w
3
s
M
o
fa
H
n
u
-H
g
0)
Xj
u
O t>
o o
CN U
^ (N —
DC
o
u
b
CN
o
u
X
rsi
O
U
cn
CN
DC
o
u
w
o
u
B
O
cn
B
o
X
o
X
CN
o
o
X
CN
o
cj
o
as
u
X
o
(N
EC
u
X
o
X
u
X
CN
o
CJ
o
u
CN
CN
X
u
B B
<N CN
O O
u u
B
o
u
cn
X
u
X
CN
o
u
X
u
CN
K
u
B
CN
O
U
B
O
u
CN
B
CJ
X
CN
O
U
B
CN
O
u
B
CJ
M
U
X
o
o
X
CJ
CN
B
u
DC
o
B
CJ
m
X
u
DC
O
CN
DC
CJ
o
u
o
m
DC
CJ
DC
U
o
u
CN
DC
Z
DC
o
CN CJ
o
u
DC
O
CN
DC
CJ
DC
O
CN
DC
CM CJ
B B
O
DC
CJ
DC
O
CN
DC
U
x
u
DC
o
DC
U
o
u
m
DC
CJ
X!
EH
2
T>
■H
CJ
<
CJ
■
•H
0)
CT
^-*
rH
G
H
-0
•
>1
. — .
0
rH
T3
•H
a1
X
•
G
' — -
•-H
0
•rH
0
a1
rd
<X3
CJ
<
rH
XI
•H
a.
cu
13
•H
rt!
HQ"
u
H
O
G
•H
u
0
<#>
•rH
rd
* —
CD
M
0
CJ
<
o
•H
o
o
CJ
T3
a,
G
<c
•H
a
LO
<c
■H
<~^
a)
■H
i
rd
o
U
0
—
rl
•
p
e
CN
^— ^
P
CJ
•H
rd
■<-{
CJ
P
D1
rd
rd
1
•
3
■r|
0)
+>
&
T3
•H
0)
TJ
•H
r-\
P
>1
D1
CQ
u
rH
>h
0
•H
SH
c
■H
rH
0
a>
X
r)
1
id
ftj
rd
u
0
rd
rd
T}
U
' '
CJ
CJ
o
H
CN
■p
e
■P
Pi
T3
<c
+J
CU
■r|
<
>1
rd
u
' '
1
M
>i
>i
5?
•H
3
0
0
H
rH
>i
T3
>i
rd
X
X
CJ
CJ
rH
U
<
0
0
O
X
>i
G
X
H
0
0
0
«:
•H
tn
cu
■H
n
o
x;
-rl
0
n
M
M
a
0
1
CJ
C
rH
rH
u
■H
rl
u
1
T3
-0
-d
o
0
P
n
•H
0
rd
>1
T3
Q
Q)
T3
>i
>1
>i
•H
o
0)
•*
U
p
P
XJ
>i
1
O
>i
— ^
X!
X!
s
CJ
3
«
CN
■P
0
<D
P
B
CO
>i
DC
2
•H
•H
i
3
rH
i
•*
•H
rl
CJ
1
1
•*
rH
1
Q
Q
m
a
O
CM
ft
u
u
<
a
rH
O
ro
24
G
rd
Ck
g
o
u
G
X!
45
g
CJ
rd
c3
rH
o
U
rd
<^»
rd
■-a
M
5
0)
-H
•rH
e
n
g
0)
>H
•
x:
M
M
-p
•
(D
+3
•
x!
0>
00
u
10
TJ
TJ
U)
u
.*
[fl
u
CO
-V
rd
•
•H
rH
rH
rd
•
rd
ITS
•
■H
03
W
s
&4
<
rt!
H
a
ca
W
s
Cn
03
n3
CJ
■H
s
<D
x
u
a
o
u
a
rH
rsi
o
rH
r*i
B
CM
<N
(*1
X
a
a
U
a
»-N
X
U3
CN
u
o
a
u
a
u
CJ
(N
CM
CM
X
X
a
o
x
o
o
a
o
o
1
,—,
^->
.-^
o
o
o
u
2
u
CNJ
u
* —
^r
u
o
X
a
X
■■»
^r
^r
"»
O
<N
X
• •
x
^~.
(N
1
o
o
O
a
a
X
a
u
a
u
X
u
X
a
in
—
— '
1 — ■
U5
•x>
vO
U3
CM
u
X
u
r»l
o
u
O
**
^r
«3"
u
CJ
tj
u
X
o
u
X
^^
CM
o
•-t
X
X
a
CM
<N
CM
•
u
u
'•
o
rH
X
u
rH
VO
kO
U3
o
o
O
o
PO
ro
(N
CM
u
u
n
X
O
U
u
Z
3
Z
X
IB
a
a
a
—
* —
X
U3
I
i
i
1
!
1
1
u
u
u
u
u
u
u
U
g
0
a
£
0
a,
e
2
U1
■H
*"*
_
^_^
rH
O1
•
•
•H
o1
D1
•H
TJ
0)
• — "
^
rH
rH
•H
TJ
•
— *
"~-*
0
■H
0>
ty
<
g
p
•rl
rH
QJ
«j
rd
rH
o
G
u
a
-P
"«-»
•H
0)
rH
0
■H
0
CD
Q
-P
o
-H
o
•H
o
rH
1
rd
p
TJ
rH
rH
r-i
N
a
rd
0
■>*
M
rl
0)
a)
O
0
0
c
0
0
XI
>.
TJ
M
c
G
c
G
G
0)
M
-P
o
r-i
>i
XJ
<tf
rH
0
0)
0)
0)
£>
CU
CD
O
1
a
-P
■P
0
G
X!
XJ
G
>i
H
o
H
0)
>i
c
G
H
•H
o.
04
Q*
■A
>i
<
<
G
rH
u
0)
CD
•H
0
3
0
0
0
0
x:
CD
rd
CD
A
01
o
X!
cr
M
n
U
)H
-P
rH
rH
-P
M
rd
0
M
O
0
-P
p
-P
TJ
01
>1
>1
3
0
-p
■
u
0
CD
u
•H
•H
•H
>i
2
A
rH
a
H
G
rr
o
03
-p
TJ
z
z
Z
a
I
-P
H
i
X!
CD
«-
3
d)
id
>i
1
1
1
i
Z
W
<
CN
U
Oi
<N
CO
a
u
a
g
o
04
s
25
§
cj
G
x:
X
H
G
X!
s
G
(3
o
U
CO
u3
03
o
id
u
s
M
M
•H
■H
u
x:
E
■H
•H
i
CD
-P
d)
d)
H
n
01
o
•
-U
M
H
X
CO
a;
^
T3
T3
^
w
CJ
co
T3
rH
CO
0!
to
m
nl
H
HI
rO
■H
■
Id
H
|
00
PQ
w
CQ
«
rt!
kC
CQ
Pm
£
W
<
w
to
u
•H
g
0)
J3
u
a
CN
CN
cm
O
O
O
U
u
u
•
^-.
■>*
'tf
X
X
X
X
X
o
o
U3
*£>
CN
tN
—
u
U
o
CB
X
•
•
U
CJ
u
o
o
in
u->
LO
X
X
X
x
X
1
1
UD
^x>
UD
a u
u
o
tN
X
o
CN X X
X X CN CN
O cn O O
CM U CJ
X n ro cn cn
vo X X X X
U «) io \o io
cj u cj cj
n .
ro ^-* ro co 0*1 n
cj o o o o o
O 2 X X X X
U — -" -" "— —
"* I I i I I
X vo ci in i/"> vo
VO •**«.* »
U 1" CN 01 f Tf
O
X CN «H <H n CN
CN
«l
O
o
u
CN
X
CJ
X
CN
8
CN
X
CJ
g
3
■H
'-,
-a
1
D1
•H
3
co
•H
rH
Q
' "*
O
-H
>
T)
^-^
a)
0
T3
•H
•
G
N
•H
U
o«
0
G
CJ
<
•H
rH
G
0)
d)^
Xt d)
rt!
o
■—'
13
x:
>1-4J
0
•H
•H
&
X t0
^-^
rJ
T3
0
rH
O
0
rH
0 u
e
+>
■H
N
0
<
■p
0
U T3
D
<u
U
a
tJ
X!
ai
C
T3 >1
-H
o
rt!
ai
•H
0
O
o
t)
•H
nrj
>iX!
B
id
J3
0
o
•H
(0
•H
rH
CJ
•H
X! 0
•rH
o
1?
<
r-\
rH
>i
O
0
3
O
H G
CO
u
•H
<
d)
*
<
rH
rH
<
W 0
•H
4J
H
0
u
Tl
0
rH
&
eh a
Q
0
>1
M
■H
f-\
g
M
o
nt
0
CJ
i *-»
' — '
rH
0
T3
0
>1
1
■d
■H
tP
M
■H
VO
■H
•H
>i
N
N
>i
M
0
0
rH
*
<
M
H
X
a
G
1
X
O
u
rH
rH
^r
H
■P
nj
1
a)
a>
(-}
i
•H
>l
x;
ai
»
Q
•H
w
CU
CD
0)
a
o
04
Pi
a.
CJ
CN
W
2
rd
2;
CN
o
CJ
CN
X
H
CN
CN
X
o
m
2
CN
o
CJ
CN
X
u
X
CN
O
U
(N
X
CJ
26
-p
o
o
k.
r-\
0)
r-l
CD
XX
e
a
<u
EH
2
H
CD I
^ I
CO 1
•H i
En
(1)
-P
(0
c
H
0
^
rH
O
0
nJ
CO
rH
0
m
CD
•H
in
o
o
m
e
r-1
CT>
CO
iH
m
0
rH
ro
*tf
r-
3
E
r~
m
•5T
<N
p
CN
<N
<N
^f
u
0
-H
O
CD
0)
CD
CD
•H
rH
>1
>1
>1
>i
M
fCJ
Q
Q
Q
Q
W
U
27
Table 2 - List of Commercial Retarders.
Retarders
Density
Residue*
(g/cc)
(g/cc)
Name
Chemical Classification
L
Calcium Lignosulf onate
1.147
0.370
A
Hydroxycarboxylic Acid
1.152
0.413
S
Carbohydrate (Sugar)
1.175
0.376
*Weight of residue was obtained by oven-drying retarder
at 110°C.
In the shrinkage experiments, five reagent grade
chemicals that had been found to be retarders were selected
and used along with the three commercial retarders. The
pure compounds were glycolic acid, citric acid, 3-hydroxy-
2-butanone, sucrose, and hydroquinone. Reagent grade
calcium chloride was also included to investigate the
effects of an accelerator.
Cement
The cement used was an ASTM Type I, Lab. No. 319, the
properties of which are shown in Table 3. Since a great
number of setting time measurements on mortar specimens
were conducted for more than a year, some cements had
changed slightly during storage, and it was found that the
cements thus partially hydrated and possibly carbonated
had slower setting properties. This difficulty was
28
Table 3 - Composition and Properties of Lab Cement No. 319
Compound Composition
c3s
C2S
C3A
C4AF
CaSO.
Total 96.94
Chemica
1 Analysis
(%)
sio2
21.64
A1203
5.29
Fe2°3
2.19
CaO
65.36
MgO
0.92
so3
2.40
Na20
0.08
K20
0.41
Lo s s on
Ignition
1.80
Free Lime 1.08
Insoluble Residue 0.25
Physical Tests
Normal Consistency (ASTM C187) 24.5 (%)
Expansion (ASTM C151) 0.136 (%)
Setting Time; Gilmore (ASTM C266) :
Initial 2:55
Final 4:15
Fineness; Blaine (ASTM C2 04) 3870 (cm2/g)
Paste False Set(ASTM C451) 93.5 (%)
Air Entrained (ASTM C185) 8.3 (%)
Compressive Strength (ASTM C109) :
1 day 1550 (psi)
3 day 3130 (psi)
7 day 4600 (psi)
28 day 6450 (psi)
Tensile Strength (ASTM C190) :
1 day 205 (psi)
3 day 345 (psi)
7 day 430 (psi)
28 day 520 (psi)
56.
.03
19.
.86
10.
.31
G.
.66
4,
,08
29
overcome by running a blank test on a plain mortar as a
control once a month or whenever cement was taken from a
new sack.
For shrinkage measurements of paste samples and for
some chemical experiments only one sack of cement was
selected and stored in sealed glass bottles.
Sand
The sand used in the preparation of mortar specimens
for setting time determinations had a specific gravity of
2.63, 1.33% absorption, and a fineness modulus of 2.26 and
2.45 for the two lots used. These were masonry mortar
sands with more than 99% passing the No. 8 sieve, because
it was thought that large sand particles might result in
erroneous penetration values when a small needle was used.
In use, these sands were oven-dried for 24 hr at 110°C
and stored in a temperature controlled room at 21°C.
Water
In the setting-time experiments, tap water was used to
prepare the mortar specimens. The water was put in a poly-
ethylene bottle and stored in the same temperature
controlled room.
Deionized or distilled water was used for all of the
other experiments in this paper. These include chemical
analyses and preparation of cement paste specimens.
30
CHAPTER IV - EXPERIMENTAL WORK
A - Setting Time Measurements
General Remarks
The setting times were determined by the Proctor
penetration method, ASTM Designation: C403. This method
involves obtaining a mortar sample by wet-screening concrete
to remove the coarse aggregate and is time-consuming and
requires large amounts of materials. It has been shown (39)
that the setting time of concrete can be determined from
that of a directly-mixed mortar if the mortar is made with
the same water-cement ratio, sand-cement ratio , and materials
as the concrete. Therefore, it was decided to measure the
penetration resistance of directly-mixed mortar as the
method of evaluating chemicals as retarders. Mix pro-
portions corresponding to those of a practical concrete
were chosen for the mortar. Initial setting time and final
setting time are, according to C403, defined as the elapsed
times to reach penetration resistances of 500 psi and 4000
psi, respectively. This definition has been used in this
work.
All materials were stored in a temperature-controlled
room at 21°C, and all operations were conducted in the same
room.
31
Apparatus
Container
One-gallon cardboard containers, with a diameter of
6.75 in., were used to contain the mortar; this represents
a slight modification of Method C403.
Since the containers are not as rigid as metal con-
tainers, it was suspected that the value of penetration
resistance might vary with the location that was probed in
the surface of the sample. However preliminary experiments
showed that the differences among the measurements were
negligibly small if the test was conducted within a circle
located 1 inch in from the side of the container.
Penetration Resistance Apparatus
A hand- loaded Proctor apparatus was used. The spring
was calibrated by means of a platform scale. The calibra-
tion curve is shown in Figure 1.
Mixer
A model A200 Hobart Mixer was used. The slow speed
was determined to revolve the paddle at a rate of 94 rpm,
with a planetary motion of 47 rpm. The medium speed re-
volved the paddle at a rate of 180 rpm, with a planetary
motion of 90 rpm.
Total capacity of the mixing bowl was 20 liters, and
an optimum mix volume was felt to be about 7 liters.
32
•
IPO
100
-
/
80
/
■ — ■
Pc
i
fiO
OU
/
■si
4.0
/
N
p
?o
/
7
\
II
/
/
0
| |
I
I
i
i
20 40 60 80 100
Reading on Rod,Pe(lb)
120
Figure 1 - Calibration Curve for Proctor Penetrometer
(ASTM C403)
33
Preparation of Mortar Specimens
Mixing Proportions and Size of Batch
Constant mix proportions were maintained throughout
the experiments. The water to cement ratio was 0.50, and
the sand to cement ratio was 2.75, by weight. These are
typical of ratios used in concrete mixtures. The mortar
thus proportioned was plastic and workable. Since the sand
was oven-dried, the actual water to cement ratio of the
mortars was a little lower than 0.50.
In preliminary experiments it was found that three
values of penetration resistance from three samples had a
range of less than 5% of the average value, in every case.
It was therefore determined to prepare a mix of 6*2/3
liters, and two test samples were prepared from this batch.
The reagent chemicals used as admixtures were added
at a concentration of 0.1% of the cement by weight in most
cases, although sometimes others were used to determine the
influence of concentration. For the three commercial
retarders, various concentrations were tested depending
on each manufacturer's suggested dosage.
Mixing Procedures
The admixture was added to the mixing water and stirred
until dissolved completely. Some relatively insoluble sub-
stances required heating for dissolution. The solution
thus prepared was placed in the mixing bowl, and the
34
cement was added. They were mixed at the slow speed for
3 0 seconds. The mixing operation was interrupted for about
5 sec during which time the sand was added. Mixing was
resumed at slow speed for 30 sec and, then for 1 min at
medium speed. Then the mixing was stopped for 1 minute.
During this period, all the mortar on the blade of the
mixer and the wall of the bowl was scraped down with a
spatula. The mixture was again mixed at medium speed for
an additional 1 minute. Mortars thus prepared had flow
values of 110 to 140%, according to ASTM Method C87,
depending on the admixture used.
Casting and Storage of Specimens
The mixed mortar was placed in the two containers and
was compacted by rodding 2 5 times. The surface of the
specimen was then approximately leveled with a trowel, and
the side of the container was tapped lightly with the
tamping rod. The distance between the finished surface of
the mortar sample and the top edge of the container was
approximately 1 inch. The sample and container were covered
with a damp cloth, and the lid was placed on.
Penetration Test Procedure
ASTM C4 03 was followed exactly. Six to seven penetra-
tion resistance determinations were made on most specimens.
Some samples did not reach the final setting time within
24 hr; in such instances the test was stopped after 24 hr,
35
mainly because there was no location left for further
penetration measurements on the surface of the sample.
A needle was selected for each penetration of such a
size that the load on it was between 40 and 90 lb, a
range in which the most reliable values were obtained.
B - Measurement of Shrinkage of Cement Pastes
General Considerations
Because of the impermeability of cement paste and its
consequent slowness of drying, it was desired to use small
specimens for the shrinkage experiments to minimize both
the time to reach drying equilibrium and the moisture (and
therefore, strain) gradients within the specimen. Homo-
geneity and reproducibility are required for such small
specimens.
A possibly troublesome effect with small specimens is
carbonation of the cement paste by reaction with atmos-
pheric CO„ , which can induce more shrinkage than drying (35)
Another consideration is that cement paste continues to
hydrate for a comparatively long period of time. This con-
tinued hydration changes the structure of the paste and,
therefore, affects the shrinkage behavior. Since many re-
tarders modify the early hydration of cement, this effect
of continued hydration is of significant importance in this
study. With these considerations in mind, the following
apparatus and techniques were employed.
36
Apparatus
Mold for Cement Paste Specimens
A special mold was made of Teflon so that the paste
specimens could be demolded without using any parting medi-
um, such as oil. Thus the surfaces of the specimens were
free from any contamination, so that uniform drying could
be attained. The mold is shown in Figure 2 together with
a cement paste specimen.
A total of ten specimens could be made at one casting.
The size of the specimen was 1/5 x 1/5 x 4 inch. A small
hole was made through the centers of the opposite end
blocks to accommodate the gauge studs for the length change
measurements.
Some specimens had to be demolded at about 7 to 10 hr
after casting. In these cases a thin polyethylene sheet
was placed between the teflon mold and the specimen. In
spite of this effort, about half of the specimens were
broken on demolding.
Gauge Studs and Length Comparator
Gauge studs were needed to measure the length changes
of the cement paste bars. The studs used were the ends
clipped from empty ball-point pen refills made of stainless
steel. These were inserted in the end blooks with the
ball-portion outward so that after casting about 3 mm of
the shaft was embedded in the paste specimen. The ball
37
4
u
<D
-H
W
03
U
<D
+J
W
id
C
0)
a,
e
rtJ
w
a)
-p
to
d<
4J
c
a>
e
u
01
C
0)
g
•H
u
a)
Cu
w
<D
+J
0)
nj
IX
■P
G
0)
i
u
H
O
en
-H
38
portions proved to be ideal as gauge studs and showed no
deformation, to 0.0001 in., within themselves.
The length comparator is shown in Figure 3. All
elements except the base plate were stainless steel. The
dial gauge provided a maximum stroke of 0.3 in., and the
smallest subdivision was 0.0001 inch. Since the length of
the cement paste bars was 4 in., a length change of 0.0025%
could be determined. A standard bar was also made of stain-
less steel to compensate the effects of temperature fluctua-
tion on the gauge distance of the apparatus. It was enclosed
in a rubber tube to eliminate heat transfer from the hand
to the bar during handling.
Mixing Apparatus
A vacuum mixing container was made from stainless steel
tubing. A stainless steel plate was welded to one end of
the tube, and a circular slit was made on the other end to
accommodate an O-ring. The O-ring provided the seal between
the container and the lid. The lid was attached to the
container by means of four clamping screws. On the side of
the container was inserted a vacuum-tight stainless-steel
needle valve, through which the container was evacuated
and through which a measured amount of water was introduced
into the container. The inside dimension of the container
was 3x4 inch. The size was so determined that a paste
made from lOOg of cement could be mixed efficiently and
39
Figure 3 - Length Comparator
40
uniformly by a shaking action. The shaking action was
provided by a Red Devil paint shaker.
Preparation of Cement Paste Specimens
Mixing Proportion and Concentration of Retarders
It was desirable to use a water-cement ratio as close
as possible to that normally used in concrete mixtures. A
ratio of 0.40, however, was found to be the maximum allow-
able for a cement paste to be mixed and cast without experi-
encing a significant amount of bleeding; this ratio was
employed throughout.
The three commercial retarders (L, A, and S) and five
pure chemicals (citric acid, glycolic acid, hydroquinone,
3-hydroxy-2-butanone, and sucrose), each representing a
certain group of retarders, were used. Their concentrations
were determined on the basis of the results of setting time
experiments on mortars. The relationship between concentra-
tion of retarder and initial setting time of the mortar
containing it was plotted and the concentration that gave
50% of additional retardation to a reference mortar contain-
ing no admixture was established from the curve. The values
obtained are given in Table 4. Additionally calcium
chloride dihydrate (CaCl2 • 2H20) , an accelerator of set, was
used at a concentration of 1% of the cement by weight.
41
Table 4 - Concentration of Retarders Added to Cement Pastes,
Retarder
Concentration
(percent of cement)
Commercial Retarders
L
A
S
0.260
0.065
0.100
Pure Chemicals
Citric Acid
Glycolic Acid
Hydroquinone
3-Hydroxy-2-
Butanone
Sucrose
0.050 (0.076)*
0.135
0.080
0.400
0.050
* This is the concentration which gives 100% of additional
retardation to a reference mortar. Others give 50%
of additional retardation.
42
Mixing, Casting, and Curing of Cement
Paste Specimens
The cement was placed in the mixing container. The lid
was attached and the four clamping screws were tightened.
The container was evacuated on an aspirator for 10 to 15
minutes. During this time, 8 0 ml of deionized water were
poured into a graduated separatory funnel and the required
amount of admixture, if necessary, was dissolved in it.
The evacuated mixing container was closed with the needle
valve and then was removed from the aspirator. The solution
was then drawn from the funnel into the container by open-
ing the valve. The valve was closed again when the required
amount of the solution, 4 0 ml in most cases, had been
introduced into the container. The apparatus is shown in
Figure 4.
The container was then placed on the shaker and shaken
for 5 minutes. This method yielded an extremely uniform
paste mixture that contained no entrained air.
After the shaking the paste was removed from the con-
tainer and placed in a "cake-decorator" type of apparatus
made of a pyrex glass tube with a squeezable ball attached
to one end (Figure 2) . The cement paste was cast in the
mold from the paste caster. First the bottom of one gauge
stud was filled; and the upper part next. The other end
was then filled in the same way, and the rest of the mold
was filled continuously from one end to the other. The
paste was cast slightly above the surface of the mold, and
43
Figure 4 - Apparatus for Drawing Mixing Solution from
Funnel into Mixing Container
44
was later sliced off at the time of demolding to eliminate
any effects of either bleeding or carbonation. The mold
was then placed on a vibrating table (paper jogger) and
was vibrated briefly. This general technique produced uni-
form specimens with a minimum amount of bleeding. Finally,
the specimens in the mold were covered with damp cloths
arranged so that they did not come in direct contact.
The specimens that were tested at an age of 4 days or
later were demolded at 2 days. The other specimens were
demolded at 7 to 22 hr after the mixing depending on the
kind of admixture added or the desired age of specimens.
The specimens were then immersed in closed containers in
a saturated calcium hydroxide solution.
Measurements of Length and Weight,
Method of Drying
At the desired age the specimens were measured for
length change, weight change, and degree of hydration. The
specimen was taken from the curing bottle and was placed
in the length comparator in the standard manner . The
length was recorded in the wet state. After the measure-
ment, the sample was returned in the storage bottle.
Shortly thereafter, the sample was again taken from
the bottle, the studs and the body of the paste bar were
wiped with tissue paper, and the weight was measured rapidly
on an analytical balance, to the nearest 0.1 mg.
45
The specimen was then dried in a vacuum oven controlled
at 105 to 110 °C, which was continuously evacuated on a
vacuum pump. This was a severe drying condition and un-
realistic for concrete in the field. It was used to shorten
the time of drying, to eliminate carbonation, and to mini-
mize the effects of continuing hydration of the cement. At
the end of the drying time the specimen was taken out of
the oven and was cooled to room temperature in a desic-
cator over anhydrous magnesium perchlorate.
The specimen was then weighed. The length change was
measured with the Ames-dial comparator. This sequence was
based on the observation that the weight of the dried
sample was very sensitive to exposure to air, whereas the
length was not especially so. Some very young specimens
had warped longitudinally. However, deflection at mid-
length (due to warping) was less than 0 . 5 mm for most
specimens. Calculation showed that errors caused by such
a warping were less than an experimental error of length
measurement, and no correction was made. Only four speci-
mens warped more than 0.5 mm at the center; they were
corrected for lengths, using the curve shown in Appendix A.
Shrinkage was calculated with respect to the water-saturated
length.
After all measurements, the specimen was broken into
three approximately equal lengths. The middle third was
used immediately for the determination of non-evaporable
46
water, and the other two portions were stored in a vial
out of contact with air.
Some relatively mature paste specimens were tested
after drying in air of 50% relative humidity at room temper-
ature. These specimens were placed in a desiccator over a
saturated salt solution of Mg(NO_)2. A chemical sorbent
for CO- was also placed in the desiccator, which was then
evacuated. Changes of length and weight were measured at
various times in the same manner described above.
C - Determination of Degree of Hydration of Cement Pastes
Loss on Ignition Test
The degree of hydration of the cement in the paste
was determined by measuring the amount of chemically com-
bined, or non-evaporable, water. Direct determinations
were made by the Penfield method (4 0) , which has certain
drawbacks. It was decided to conduct a loss on ignition
test and to calculate the amount of non-evaporable water
from the result. The derivation of the equation used is
shown in the next section.
The loss on ignition was determined in the usual way,
using approximately lg samples of the crushed, oven-dry
paste in platinum crucibles. The heating was in a muffle
furnace to constant weight at about 1050°C.
47
Derivation of Equations for Determining Non-Evaporable
Water and Evaporable Water
Loss on ignition has sometimes been considered a
measure of the degree of hydration of cement paste. But
this idea is misleading. Even an unhydrated cement has a
loss on ignition of around 1%, because the loss on ignition
includes not only non-evaporable water, but also carbon
dioxide from carbonated compounds.
In hardened cement paste, most bulk water is evaporable
at relatively low temperature. However, some water mole-
cules are sorbed on cement gel with a differentiating
degree of bond strength. As a result the distinction be-
tween chemically-combined (non-evaporable) water and
physically sorbed (evaporable) water is possible only on
an empirical basis. In this study, therefore, they were
defined as follows:
"Evaporable Water": Water lost from a sample during
vacuum oven drying at 105-110°C
for 24 hours.
"Non-Evaporable Water": Water remaining in a sample dur-
ing vacuum oven drying of 2 4 hr
at 110°C and lost on drying at
about 1050°C.
Assumptions
In the derivation of these equations it was assumed
that chemically combined water or carbon dioxide (due to
inadvertent exposure to atmosphere) in the "dry" cement
are not lost on vacuum oven drying to 110 °C and that any
further carbonation during mixing, curing, and drying of
48
the paste is negligible. In any case the pastes prepared
for this study had carbonated only slightly.
Notation
Schematic diagrams for "dry" cement and for cement
paste are shown in Figure 5. The notations are:
We : weight of evaporable water in cement paste
Wn* : weight of chemically combined water after
mixing
Wn : total weight of non-evaporable water in
cement paste
Wc : total weight of cement
Wee : weight loss of cement when dried in the
vacuum oven for 24 hr at 110°C
Wclw : weight of chemically combined water in
cement before mixing
Wclc : weight of chemically combined carbon dioxide
in cement before mixing
Wcl : (Wclw + Wclc)
Wch : weight of cement hydrated in cement paste
Wcu : weight of cement unhydrated in cement paste
Wei : ignited weight of cement at 1050°C
(=Wch + Wcu)
Wl : total weight loss on ignition of cement
before mixing
Wli : weight loss on ignition of cement paste after
being dried in the vacuum oven for 24 hr
at 110°C
Wss : weight of water saturated surface dry cement
paste
Wod : weight of the same sample after dried in the
vacuum oven for 24 hr at 110°C
49
■*
S8M
►
P°M
MM !3M
"M
ID
•
£
3
^
c
U
JJ
£
£
£
£
I
1
I
\
I -
«
3M
• IM '
l»M
i
i i
-P
ui
id
&<
4J
C
d)
E
cu
u
Ti
c
TO
c
e
0)
u
H-l
o
u
en
to
■H
Q
+J
TO
£
0)
A
o
w
-H
50
Derivation of Equations
Immediately obtainable relations are,
Wc = Wei + Wcl + Wee = Wei + Wl (1)
Wcl = Wclc + Wclw (2)
Wss = Wei + Wcl + Wn* + We (3)
Wod = Wei + Wcl + Wn* (4)
Wli = Wn* + Wcl (5)
From eq. (5) ,
Wli m Wn* + Wcl . Wn*_ = Wli _ Wcl .,.
Wei " Wei Wei Wei Wei
Combining eqs. (3) and (4), and dividing by Wei,
We _ Wss - Wod __ Wss - Wod Wod
Wei Wcl " Wod x WcT
i Wss ^ \ , t L Wcl x Wn* x ._,
= ( rr-j - 1 ) X ( 1 + - — r + rr — i- ) (7)
Wod Wei Wei '
Substituting eq . (6) into eq . (7) ,
We = , Wss _ Wli .
WcT ( Wod X ) ( X + WcT > (8)
We = We Wei _ We (Wc-Wl)
Wc Wei Wc Wei Wc
=(Wss_l) ( ! + Wli } (1_W1} (
v Wod x ; K •"• WcT ; l x Wc ; ^'
Dividing both terms of eq. (2) by Wei,
Wcl
Wcl "
Wclc
Wei
L Wclw
+ Wei
#
Wclw
Wcl Wclc
Wei
Wc i Wc i
Wcl
WcT
Wc lc Wc
Wc Wei
(10)
51
Since Wn = Wn* + Wclw,
Wn _ Wn* + Wclw
Wci " Wci
Substituting eqs. (6) and (10) into eq. (11) ,
(11)
Wn
WcT
Wli Wclc Wc
WcT Wc x WcT
Wli
WcT
Wn
Wc
Wn Wci Wn ,,
WcT x Wc WcT u
Wl
Wc
W£l£ (1 + ^K)— (12)
Wc VJ- WcT; vxz;
) (13)
It should be noted that Wci includes also the part of
cement which has been carbonated before mixing with water
and, therefore, is not available for hydration. Calcula-
tion, however, showed that the true Wn/Wci and Wc/Wci did
not differ by more than 1% from the calculated values from
eqs. (8) and (12). Because the experimental error of loss
on ignition test amounts to the calculative error, no
further consideration was made.
Reduction of Data
The following equations were derived and used to
reduce loss on ignition data in this study.
Wn _ Wli _ Wclc ,, Wl . ,..
WcT WcT Wc u WcT u '
Wn _ Wn M Wl. nt-v
Wc" " WcT (1 Wc0 (15)
We _ ,Wss _ .. n Wli, n _ Wl. ,,,.
Wc" " (Wod 1} (1 + WcT5 (1 Wc"5 (16)
52
Wli/Wci can be determined directly from the loss on
ignition test, as can Wl/Wci and Wl/Wc. Wclc/Wc was
determined by a common method (41) . The value of Wl/Wc
varied from 0.0196 to 0.0203 during the shrinkage study.
(Wclc/Wc) x ( 1 + Wl/Wci ) ranged only from 0.01091 to
0.01092. Wss and Wod were obtained from the specimens used
for the length change measurements. In calculating Wss and
Wod, the weight of the gauge studs was, of course, sub-
tracted from the total weight of the specimen. Degree of
hydration was expressed by dividing Wn/Wci of a sample by
that of fully hydrated cement paste.
D - Surface Area Measurements of Cement Pastes
General Remarks
Water vapor and nitrogen are the two gases commonly
used for measuring the surface area of a solid by the
adsorption method. These gases, however, give surface
areas of a different order of magnitude for cement paste.
Typical BET surface areas of a mature cement paste by water
2 2
vapor and by nitrogen sorption are 2 00 m /g and 2 0 m /g,
respectively. Because the surface area is one of the
important factors of a model of cement paste structure
there has been a critical controversy between some workers
concerning the value of surface area and the method of
measuring it (42, 43, 44, 45). This controversy is yet
53
unresolved. In this study, since only relative values were
needed, the water vapor adsorption value was chosen.
Apparatus and Procedures of Measurement
Four glass desiccators were prepared with solutions of
glycerin and deionized water to control the vapor pressure
of water. The relative humidities ranged from 10 to 32 per
cent. The concentration of glycerin in the solution, and
hence the relative humidity, was determined by the measure-
ment of the refractive index of the solution. Approximately
1 liter of each solution was placed in the desiccator and
continuously stirred with a magnetic stirrer.
The cement paste samples, which had been vacuum-oven
dried for the shrinkage measurements, were ground in a
mortar and pestle so that most passed a No. 60 sieve. The
powdered samples were placed on watch glasses and were re-
dried in the vacuum oven for an additional hour at 110 °C.
After drying, the samples were placed in weighed weighing
bottles. Lids were placed while the samples were still hot.
The bottles were cooled to room temperature in a desiccator
and weighed. The weight of the sample was about lg.
The bottles were placed in the desiccator of the lowest
humidity and their lids were removed. The desiccator was
closed and evacuated on a vacuum pump. Adsorption was
allowed to continue for 48 hours. Then the desiccator was
opened, and the bottles were covered with the lids. They
were weighed and replaced in the desiccator of the next
54
lowest humidity, and the process was repeated. The refr-
active index of the solution was tested after the bottles
were taken out of the desiccator, and the temperature was
recorded.
The adsorption isotherms were interpreted in the usual
way by the BET Theory (46, 4 7). The rectified form of the
equation was used and the data were analyzed by least
squares; the calculation was made by a computer. Specific
surface area was expressed on the basis of ignited weight
of the cement paste sample.
E - Scanning Electron Microscope Observation
General Procedures
The model JSM-U3 scanning electron microscope, Japan
Electron Optics Laboratory Co. Ltd., was used for the
observation of the samples.
At the desired age, the sample was oven-dried at 110°C
and mounted on a brass sample holder. The sample was stored
in a desiccator over anhydrous magnesium perchlorate until
the mounting glue dried. It was necessary to coat the
sample, usually only with carbon, to get a clear image.
The sample was then placed in the microscope and the image
of the sample was first observed on a TV screen at a
relatively low magnification. Various locations of the
sample were viewed in detail by changing the magnification,
55
and a desired representative area was determined on the
screen. Then the image was switched from the TV screen to
the probe screen of the microscope and examined further in
detail. Finally, photomicrographs were taken with a
Polaroid camera. The scanning speed at this time was
slowed down to 50 seconds.
56
CHAPTER V - RESULTS
Infrared Spectra of Commercial Retarders
The three commercial retarders were representative of
the three main categories in current use. To identify them
more precisely, a small amount of each was oven-dried at
110 °C, and KBr pellets were made and infrared spectra
obtained in the usual way with a Beckman DK-10 spectrophoto-
meter. Figure 6 shows the results. The interpretations of
individual absorption band follow the assignments given by
Halstead and Chaiken (48) . That marked L is a ligno-
sulfonate retarder, A is a hydroxycarboxylic acid retarder,
and S is a modified sugar or carbohydrate, probably with
some other materials in it that remain unidentified.
Non-Evaporable Water Content of Fully Hydrated
Cement Paste
The non-evaporable water content of fully hydrated
cement was needed to calculate the degree of hydration of
the cement paste samples. The value was determined by
hydrating the cement as follows: 45 g of the cement and
450 ml of distilled water were placed in a clean 16-ounce
glass bottle and a lid was placed on it with a teflon sheet
as a lid-liner. The bottle was rotated continuously on a
57
Figure 6 - Infrared Spectra of Commercial Retarders
58
roller mill for 10 days. The mixing was then changed to
magnetic stirring until the time of test. At this time
the sample was vacuum filtered on paper, and the residue
was placed in the vacuum oven at 105 to 110°C. Loss on
ignition of the oven-dry sample was then determined by
furnace drying at 1050°C.
The results of the determination conducted at hydration
times of 4 and 10 weeks are shown in Table 5. Most of the
cement had hydrated in 4 weeks; only a little further
hydration was observed from 4 weeks to 10 weeks. Hence,
the values at the age of 10 weeks were employed as those
of the fully hydrated cement in this work.
Table 5 - Non-Evaporable Water Content of Bottle-
Hydrated Cement (W/C = 10) .
Age
(weeks)
Loss on Ignition
Wli/Wci (%)
Non-Evaporable Water
Wn/Wci (%)
4
10
19.81
20.59
18.72
19.50
59
Result of Setting Time Experiments
Commercial Retarders
Figures 7 , 8 , and 9 show the results of setting time
experiments for the three commercial retarders when they
were used at different concentrations, expressed as weight
percent of the retarder-solids based on the cement. In
these figures, values of penetration resistance are plotted
against elapsed time after initial contact of the cement
with mixing water (plus admixture) . The general expectation
that the higher the concentration of retarder , the longer
the setting time, is apparent. However, it is noteworthy
that the retarders A (hydroxycarboxylic acid) and S (carbo-
hydrate) tend to accelerate the initial hydration of
cement when they were added at the highest concentration
whereas the retarder L (lignosulfonate) retards all stages
of the hydration evenly up to 4000 psi even at higher
concentration .
In order to compare the effects of concentration of
retarders at different penetration levels, Figures 10, 11,
and 12 were made from the same results. It is observed
that relative retardation of the mortars is smaller at
higher penetration resistances. This is an important and
necessary property for properly retarded concrete. Another
interesting point is that the increases in setting time at
the penetration resistance levels of 500, 1000, and 4000
psi are in close agreement and the retardation at 500 psi
60
4000
6 8
Elapsed Time (hr)
Figure 7 - Penetration Resistance vs Elapsed Time for
Mortar with Retarder L
61
4000
1000
- 500
in
a
a>
u
c
o
4-
</>
0)
a:
0)
c
6 8
Elapsed Time (hr)
Figure 8 - Penetration Resistance vs Elapsed Time for
Mortar with Retarder A
62
/ '
/
4000
1 /
/
/X
/
/
/
X
/
./
/
/
/ /
1000
/ / /
/ /
X
/
/
/ ,
/ V
/ c\"6
7
/
*
°
"55 500
Q.
/
/
- / 11
/ /
o /
01
u
c
2
w
1
#/ /
i i
/ /
<.
1
c
/
o
i /
/
Z 100
c
1
i /
y
- / 1
Q.
50
7 /
/ /
/ /
//
- y
is
- /
10
i
i
1
1
1
6 8
Elapsed Time (hr)
10
Figure 9 - Penetration Resistance vs Elapsed Time for
Mortar with Retarder S
63
0.40
0.30
;£
0.20
S
o
c
o
0.10
100
130 160 190
Relative Retardation (%)
220
Figure 10 - Effect of Concentration of Retarder L on
Setting Time of Mortar
64
0.15
0.10
c
o
c
o
O
0.05
100
130 160 190
Relative Retardation (%)
210
Figure 11 - Effect of Concentration of Retarder A on
Setting Time of Mortar
65
£
c
o
u
0.15
0.10
0.05
100
130 160 190
Relative Retardation (%)
220
Figure 12 - Effect of Concentration of Retarder S on
Setting Time of Mortar
66
level appears to be a mean retardation in the resistance
range of 100 to 4000 psi. Since initial setting time
corresponds approximately to the last time when revibra-
tion of concrete can be made, the result would indicate
that initial setting time is a practical and convenient
standard point for evaluating retarders.
Manufacturer's suggested dosages are also shown in
each figure by a shaded range. These dosages produce in-
creases in setting time of roughly 30 to 90% for all the
retarders. The resulting range of retardation is typical
of that of concrete (49) . This result implicitly affirms
the adequacy of the penetration resistance test as con-
ducted on mortar specimens in this work. In the same
figures, it should be noted that the lignosulf onate retarder
(L) required more than twice as much dosage as the others
(A and S) did to obtain a certain amount of retardation.
Pure Chemicals
Table 6 summarizes the initial setting times of mortar
specimens incorporating many chemicals at a concentration
of 0.1% of the cement by weight. For some of the chemicals
that were found to be a weak or ineffective retarder at a
concentration of 0.1%, other results of tests at a con-
centration of 0.5% are also shown. T500 is defined as
the initial setting time of each sample relative to that
of a control sample, and was obtained as follows: A
relationship between elapsed time and penetration resistance
67
C
<a
A
£
to
U
-P
u
£
<+h
o
CO
g
•H
En
D>
G
•<H
•P
-M
a>
CO
rH
id
•H T>
4J Q)
•H -O
M i$
QJ 0)
> H
•rl Q)
-PS
rH W
0) rH
« id
o
o e
>i.G
u u
nj
i a)
3 3
w a<
i
a)
id
Eh
05
U
-H
e
X!
U
<H
o
a
o
•H
+>
id
u
•P
G
a)
o
c
o
U
o
m
Eh
\
o
o
m
Eh
O
o
Eh
o
m
Eh
\
o
o
in
Eh
o
o
LO
EH
id
u
■H
e
(D
X!
U
CM rH
LD
*3"
t^ rH
en <x>
rH rH
yj oo
a> (N
n in o
(T> o o
o o
o o
<J » O N H
o o o o o
<N O O rH
O rH rH O
r-i r-t rH rH
O n
O rH
rH r-\
r~- r-» <n o •** oo
cn o o o in p»
rH (N rH r-i rH <N
O -P
O 3
w cj
o
-H 0
a. o
0)
c
•H
o
>1
rH rH
O CJ)
o
•H
■P
<D
CJ
id
o
•P
QJ
P
M
>i
2!
O
T3
•H
0
0
-H
u
>i T3
u
•rH
>
M
>1
O 0<
>i
X
o
>1
-p
3
CO
I
2
I
>i
X
o
n
»
i
a
•rl
(< -rl
CJ
<
p
c
CD
g
•H
Sh
0)
a,
x
<D
a>
x:
-p
m
o
■H
g
•H
0)
X!
■P
C
■H
0)
C
•H
e
u
<D
P
cu
ra
o
p
d)
CP
M
o
LO
Eh
\
O
o
nj in
rH En
0 l+H
o o
p
68
u
m
o
G
O
•H
+1
id
u
p
g
<D
u
c
8
*
*
m
o
CN
\o
vo
r-i
rH
en
CN
CN
O
o\
<J\
o
en
o
a\ cn
rH O
CN O
oo en
n en
en en
o o
vo cx>
en ao
o iH o
o
o
m
vo cn r-
^ o in
rH rH CN
n co
rH CN
m
CN
cn r^
o o
p
u
rH
-Q
E->
0
•r|
e
a)
.G
u
-d
•r)
a
u
■H
<D
rH
G
TJ
>1
0
T>
•H
X
s
•H
O
5
TJ
u
<
a*
0)
0)
■H
<
TJ
M
0
G
TJ
U
U
r)
id
<u
u
O
-H
a)
TJ
<
u
•rl
U
o
TJ
a,
C
g
p
•H
■H
c
*i
•H
0)
•H
i
ifl
(0
id
O
u
rl
0
H
p
E
(N
P
G
p
<
■H
id
•H
u
P
(0
<0
1
d
O
HI
0)
p
04
TJ
-rl
QJ
T»
rH
P
>i
CQ
•rl
o
0
r-i
M
0
•H
H
g
•rt
0
OJ
X
1
Q,
id
•H
(Tj
nj
M
U
id
nj
73
o
0
O
0
CM
0
o
H
£
■P
Cu
no
<
p
0.
■H
«e
>1
rrj
u
1
iH
p
rd
>i
>i
>i
•H
3
0
U
rH
^i
>i
TJ
a)
>1
04
QJ
P
X
X
X
U
u
H
M
<
O
o
o
X
>,
G
X
<-l
O
rl
0
0
0
rti
•H
cn
Qi
■H
TJ
0
A
•H
0
>,
<
rrj
M
H
H
c
0
1
0
G
rH
r-i
rl
■H
rl
lH
JZ
H
'O
-0
TJ
u
0
4J
n
•rl
O
cd
>1
TJ
Q
0)
TJ
p
rH
1
>i
>1
>i
•H
o
<1>
«.
H
■P
p
x:
>1
1
u
>i
a>
>1
^~.
.g
.g
a
O
d
*
CN
p
0
0)
■p
a
fl
>!
SB
X
J^
3
•H
•H
i
3
H
1
*
•H
u
o
2
i
**
r-i
1
i
P
Q
Q
ro
2
o
CN
rH
u
u
<
3
r-\
O
m
2
w
69
xi
u
o
a
o
-H
P
03
-P
c
<D
0
G
o
u
o
LT)
en en
o o
CN O
en m
CN 00
O rH O
O O O
o o
o o
iH rH
<N CO
O O
o
o
m
r» a\ (n
o o o
ro (N
O (N
c
o
0
XI
rd
E-
U
e
TS
TJ
•H
•H
O
U
r-t
<c
<
0
CD
O
o
T3
■H
0)
.h
C
■H
■rH
•H
Q
P
0
0
o
'O
0
rH
0
rH
I
(0
■p
•H
rH
H
rH
N
•H
N
O
<
o
**
n
•H
T)
0)
0
0
O
c
U
c
T)
x:
x:
^
Tf
!H
CD
c
G
c
c
CD
<
0)
•H
0
o
0
rH
>i
x:
c
rH
0
0)
0)
0)
-Q
X)
U
u
•-H
u
1
K
p
nj
o
c
x:
x;
43
>.
u
>i
<
r-i.
rH
rH
0>
>i
-P
c
<-*
•rl
0.
0.
Oi
X
•H
X
<
CD
<c
g
H
n
G
CD
•H
0
3
o
0
o
0
rH
0
U
T3
<D
id
CD
CD
CO
o
x:
D1
u
!J
u
u
>1
n
•H
•~i
C
rH
-P
u
ffl
04
0
M
u
0
-p
-p
TJ
u
T3
0
>1
(fl
>1
3
0
-P
1
M
0
0)
M
•H
•H
•H
>1
•H
>i
N
N
2
>-\
CO
rH
G
*3"
O
CO
■p
T)
2
2
2
K
rH
a
C
C
i
>-i
1
x:
CD
*
2
$
m
>i
1
1
1
1
rd
i
0)
0)
XI
<
CM
u
&
CN
W
&
u
35
e
0
0,
g
W
04
CQ
«
Q
70
C
o
o
rH
m
o
in
Eh
Ln
rH
N.
1
o
1
1
i
1
rH
1
1 1 1
I
1
i
o
•
•
o
■H
rH
<*>
m
rH
m
Eh
nj
•
o
-H
o
e
<D
o
,c
o
<N
«*
u
in
Eh
1
H
1
1
i
1
CO
H
1
1 1 1
i
I
i
m
0
a
0
•H
-M
o
(0
in
u
H
in
"3"
>>
r^
CO
o
CN
CO
n tj< r^
CO
CO
^
-p
\
o
o
H
o
o
m
O
cn
O O Cn
cn
en
a\
c
o
a)
o
rH
rH
H
rH
rH
rH
rH
o
rH rH o
o
o
o
o
dP
m
a
rH
Eh
0
•
u
O
o
o
CO
r^
Cn
rH
r->
r^
en
i*
O <J3 O
CN
m
CO
in
O
rH
rn
m
m
rH
CO
cn
o cn rH
o
o
rH
EH
H
rH
CN
rH
CN
<N
rH
rH
rH rH
rH
rH
rH
^-^
<D
-P
03
u
TJ
>1
J3
0
a
0
^-*
2
£
•H
T3
nd
■H
0
0
01
rH
<
■H
in
a
o
o
*— '
•H
-H
0)
a
<D
o
T3
P
0)
G
N
■H
<TJ
43
O
c
O
c
O
c
>i
<:
o
Eh
0
<4H
rH
a.
X
^->.
•H
,—s *-H *-*
^— ,
O
3
0
rH
o
£
-P
cn n co
r^
rd
0)
-p
0
M
3
<D
co in rH
o
rH
0
<D
c
tJ
-H
o
cn co m
r»
CO
c
U
TJ
■H
T3
>i
•a
rd
* % N*^
—
tn
(0
•H
rH
u
■H
Xi
0
■H
a)
•H
>1
O
0
3
O
-H
(0
M
moo
in
g
hH
X
<!
rH
rH
r<
M
-H
P
m 00 rl
in
0
0
rH
Cn
EH
a
O
■H fO tJ«
r^
P.
g
u
u
fl
0
o
1
—
rH
r» in ^r
CN
.C
3
TJ
■H
tn
u
•H
vo
■H
CN <N CN
T
o
•H
>1
rH
0
0
rH
V
rtj
M
0
o
K
0
u
■H
H
T
En
p
a) a) oj
OJ
•H
rH
1
•H
>1
rC
rrj
".
Q
•H
>i >t >i
>i
U
id
0
CU
o<
a<
O
CN
w
2
Q Q Q
Q
a
u
71
was drawn for each sample in the same way as was done for
the commercial retarders, and an initial setting time was
determined. This time was then expressed in terms of a
percentage of that of an appropriate control sample without
admixture. In the same way, T50, T100, and T4 0Q0, which
are relative setting times at the penetration resistance
of 50, 100, and 4000 psi, respectively, were determined.
They are to be found in Appendix B. A tendency, similar
to that observed with the commercial retarders, was noted
- namely the relative setting time at 500 psi, T500, is
about the same value as the corresponding T100 and T4000.
T50 is different from the other relative setting times for
some chemicals. If T50 is smaller than the others, it
indicates that the chemical accelerated, relatively speak-
ing, the very early hydration of cement. The ratio
T500/T50 was used for this indication in Table 6, i.e., the
larger the ratio the more the reactions were relatively
accelerated. Strong retarders such as tartronic acid,
d-tartaric acid, mucic acid, gluconic acid, 2-ketoglutaric
acid, citric acid, sucrose, and 2 ,4 ,6-trihydroxybenzoic
acid have large values of T500/T50.
Again, calcium lignosulf onate is not a strong re-
tarder. Glycolic acid, methyl glycolate and a-hydroxyacet-
amide are strong retarders at higher concentrations. Only
l,3-dihydroxy-2-propanone showed an unusual setting when
it was added at the rate of 0.5%. It accelerated early
72
hydration of the cement to reach a penetration resistance
of 2 00 psi in 2 hr , but extremely retarded the later
hydration. This behavior was similar to that observed in
the hydration of a Type I cement clinker without any gypsum
addition (Figure 13) .
Figures 14 and 15 show the effects of concentration of
admixture on the relative initial setting time of mortar
specimens. It is seen that strong retarders can delay the
setting of mortars almost indefinitely even at a relatively
low concentration while some other chemicals accelerate the
setting when added at higher concentration.
The results seem to divide themselves into two groups.
In Figure 14, whatever the magnitude of the retardation,
additional increments of concentration produced an in-
creasingly larger relative retardation, i.e., the curves
are concave downward. In Figure 15, on the other hand
the opposite is true; the instances showed less additional
effect at higher concentrations. Indeed, sometimes the
effect was reversed.
Effects of Retarders on the Drying Shrinkage of
Cement Paste at Various Stages of Hydration
Figures 16 and Figure 17 show, respectively, the
shrinkage of the cement paste bars as affected by the
addition of commercial retarders and selected pure chemi-
cals when the pastes were dried in a vacuum oven at 105
to 110°C for 24 hours. The shrinkage values were plotted
73
4000
4 6
Elapsed Time (hr)
Figure 13 - Penetration Resistance vs Elapsed Time for Mortar
with 1, 3-Dihydroxy-2-Butanone
74
(%) UO!JDJ4Ua3UOO
u
id
4J
H
o
2
m
o
<u
6
■H
Eh
tr>
c
•H
-U
■P
<v
CO
m
■H
+J
■H
C
o
a
o
■H
■P
id
M
-P ~
o •
C O
O 2
U —
l
3
en
■H
75
(%) uojiD-uuaouoo
u
<c
-p
u
o
s
m
o
<u
e
en
C
■H
+J
+->
<D
Q)
>
■H
■P
id
h
QJ
«
>
(1)
+J
X
H
e
c
0
•H
+J
nJ
■P — *
<D
O •
c o
O 2
I
•H
fa
76
o c
c
o
I
Q
CM
CO
d
CD
d
o
CVI
d
(%) d6D>|UU4s
4-1
o
co
<D
CP
fl
4-)
CO
Dl
O
•H
i-l
>
rd
QJ
+J
ui
<C
a,
-p
c
0)
g
0)
u^-
m
Tj U
«U (1)
•H TJ
Q «J
I -P
a> «
>
O rH
(0
e-H
3 O
u a)
nj g
> 5
o
4-1 U
o <—
a) c
<7> o
(0 H
AJ -P
C «J
•H ^
U T)
X >1
I
(1)
M
77
00
<o <!•
(VI
o
b 6
(%) a6D>|uuqs
O
<D
Cn
id
■p
CO
W
0
■H
M
rd
>
m
CO
QJ
4J
[fl
(0
cu
+1
a
u
-d
CD
•rH
5-1
Q —
I (0
C r-i
Q) as
> O
O -H
e
e a)
3 x:
a u
u
aj a)
> n
4-1 Ol
o —
<u c
a> o
c m
■H U
U T3
jC >,
CO ffi
I
r-
H
a)
>-i
tn
-H
78
against the degree of hydration of the samples, determined
as previously explained. For clarity, data points are not
shown in the figures. They are to be found in Appendix C.
For citric acid, however, data points only are shown in
Figure 17 because of the lack of enough points to draw a
complete curve.
At the degree of hydration of about 50%, which roughly
corresponds to an age of 2 days, all the samples shrank
practically an equal amount, approximately 0.8%. The
cement paste alone, with no admixture, always had a smaller
shrinkage than the others at this point and at most later
ages. At the degree of hydration of 85 to 90%, the rate
of increase in shrinkage was reduced for all pastes except
that with calcium chloride.
When the degree of hydration was between 2 0 and 4 5%
the shrinkage behavior of the samples was changed signifi-
cantly. All hydroxycarboxylic acids such as retarder A,
glycolic acid and citric acid increased the shrinkage
considerably. Sucrose had an effect about equal to that of
these acids, but the carbohydrate retarder (S) , conversely,
minimized the shrinkage. The cement paste with lignosulfo-
nate retarder (L) shrank more or less in the same way as
did the cement paste with no admixture, but showed less
shrinkage at early ages. Addition of 3-hydroxy-2-butanone
or hydroquinone , which have no acid group, but only hydroxy 1
or carbonyl groups, resulted in a smaller shrinkage of
79
cement paste at this early age. Calcium chloride increased
the shrinkage at all stages of hydration and showed the
largest shrinkage of all hydrations higher than 50%.
When either the concentration of citric acid was in-
creased enough to result in a relative retardation of 200%
or the water-cement ratio of plain paste was changed from
0.40 to 0.43, the shrinkage of the pastes was practically
unchanged. These results are to be found in Appendix C.
Drying Shrinkage of Mature Cement Pastes
When Dried at 5 0% Relative Humidity
Shrinkage of 10-months-old (i.e. almost mature) samples
was measured over a period of 50 days in an atmosphere of
50% relative humidity. The results are shown in Figures 18
and 19. Each data point is the average of two separate
measurements. The range of duplicate measurements was
usually within 3% of the average. The commercial retarders
and most of the chemicals of the hydroxycarboxylic acid type
increased the shrinkage consistently. However, all the ad-
mixtures also increased the degree of hydration of cement
paste at this age, and the samples did not reach the same
degree of hydration. Instead, there is a tendency that a
sample of higher degree of hydration shrinks more, as
would be expected.
The relationships between shrinkage and weight loss
of the samples are shown in Figures 20 and 21. If the
80
C3
m
-p
ai
1-1
a
c
X!
M
U <D
(%) aBD^uuijs
3 -p
-m a;
S
rH
O -H
m o
u
a> a>
S g
•H s
M o
u
c
•H >i
>i-P
U -H
Q TJ
-H
oi e
> 3
w
a)
Oi 0)
m >
C -P
■H <d
x: a)
co a;
i
ai
M
■H
81
>5
Q)
O
<J>
(0
<*■
.£>
Q>
N-
fO
0)
£
o)
<fr
*
10
N
c
10
N
K
s
S
"O
<J
lA
<
O
u
E
O
<
■o
c
<
u
C
o
a*
u
<
u
o
a*
c
O
u
o
l_
E
o
+-
o
>>
•*-
.n
9)
z
CD
u
1-
*
<
■
¥
•
I
<
I -
c
I
1
o>
|
to
i
I
<i
*
•
<
o
m
O
a>
E
i-
c
-M
C
a>
e
^
0)
w
u
rH
id
<D
u
SH
■rH
3
e
+J
01
m
x:
g
u
n
T.
0
•rH
£ QJ
■H Vh
(%) aBo^uuqs
C
H >,
>i-P
M -H
Q T>
•H
o) e
> 3
QJ
t7> <D
<a >
^ -H
C -P
•H (0
rH rH
x: a)
i
en
•H
82
Weight Loss (%)
% 0-2.
o
3C
C
0.3
0.4
o ;
i
4
S
9
i
i
l
i
-
•
• ■
•
-
• • •
•
1 *>. •
>• xo
•
-
• .* No Admixture
x : Retarder L
o : Retarder
• : Retardei
■ A
- S
Figure 20 - Shrinkage vs Weight Loss of Mature Cement Pastes
When Dried at 50% Relative Humidity
(Commercial Retarders)
83
Weight Loss (%)
2 4 6
S. 0.2
o
it
c
in
0.3
0.4
1
I
1
I
-
1*
X
°i o
-
-
• : No Admixture
o i Glycolic Acid
x • Citric A**'^
• : Tartaric
A : Ketomal
Acid
onic Acid
Figure 21 - Shrinkage vs Weight Loss of Mature Cement Pastes
When Dried at 50% Relative Humidity
(Pure Acid Chemicals)
84
weight loss in these figures is replaced by drying time in
Figures 18 and 19, the relative position of each data point
in both sets of figures appears to be the same. This
similarity may be caused either by differences in the
amount of hydration products produced or some change in
the structure of the cement paste induced by the addition
of the retarders. The cement paste with retarder S, how-
ever, showed extraordinarily large shrinkage when a certain
amount of weight was lost. In this case, some changes in
composition or structure of hydration products would be
more expectable.
Water Content of Cement Pastes
Figures 22 and 23 show the relationship between evapo-
rable water and non-evaporable water contents of cement
paste samples. These water contents are expressed in per-
centage of original (i.e. dry) weight of cement, for ease
of comparison. The original water-cement ratio of the
samples was 0.40. The reason why the sum of Wn and We is
greater than this is the increased water content brought
about by self-desiccation and the efficient curing of
these samples.
Each cement paste with a certain admixture shows its
own relation in some different way from others. However,
as indicated for the plain cement paste by solid lines,
the data point could have scattered in a wide range if more
measurements had been made. Yet, most of the points fall
85
15
o
c
\
o
\°* • \
Orininnl
-
\ •• \
\ *• \
\ ° \
\ # \
W/C*40%
•
-
0 \
\ • *\
\ \ x
\ o ^
V\ i
-
w • [no MumixTure
x .'Retarder L
o : Retarder A
• : Retarder S
\f \.
* ••
\°\
1 1 1 1
1 1
i i i i
30
35
We/Wc (%)
40
Figure 22 - Non-Evaporable Water vs Evaporable Water of
Cement Pastes (Commercial Retarders)
86
c
35
We/Wc (%)
Figure 23 - Non-Evaporable Water vs Evaporable Water of
Cement Pastes (Pure Chemicals)
87
in the region that the two lines for the control samples
enclose. This could indicate that retarders do not change
the structure of cement paste significantly.
In Figures 24 and 25, the non-evaporable water content
is plotted against curing time of the samples. The solid
line represents the relation for the cement paste without
admixture. It can be seen that retarded degree of hydration
of a cement by the addition of retarders at the concentra-
tion that results in relative set delay of 150% seems to be
recovered completely in 2 to 4 days. After that, even some
increase in the degree of hydration is observed for most
pastes with retarders.
Specific Surface Area of Cement Pastes
The results of the water vapor adsorption experiments
were used to calculate the specific surfaces of the paste
samples. The calculation was done by means of the BET
theory in the usual way and with the assumption of a unit
area for the water molecule of 10.6 A . The results are
shown in Figures 2 6 and 27, on a basis of the ignited
weight of the total sample. The values shown are averages
of two measurements, which agreed to about 10%. As is
expected, the larger the degree of hydration, the higher
the specific surface area, and the relation is almost linear.
However, there is no quantitatively noticeable difference
observed among cement pastes with different admixtures.
88
O
CM
iO o
m
0)
u
cn
■H
Pq
89
o
0>
<
03
OJ
+J
03
05
4J
G
o
S-i
0)
u
o
0)
en
c
•H
M
U
>
U
0)
4-1
nj
s ~
OJ rH
>
W oj
I u
c 3
o &
I
u
Cn
•H
90
•
: NO
Admixture
250
: Retarder L
: Retarder A
o
•
Retarder S
• O
5 200
A
Citric Acid
Citric Acid
(200% Relative
Retardation)
»
,****
*
Specific Surface Area
O oi
O O
—
•
•
« x *
•
• •
•
1
X*
•
w •
X
o
•so
X
o
0
i
i
1
1
l
20 40 60
Degree of Hydration (%)
80
100
Figure 26 - Specific Surface Area of Vacuum Oven-Dried
Cement Pastes When Expressed on the Basis of
Total Ignited Weight of Samples (No. 1)
91
300
250
o»200
o
o
CO
o
CO
150
100
50
No Admixture
Glycolic Acid
Sucrose
3- Hydroxy -2- Butanone
Hydroquinone
CaCl2-2H20
0*#
O
V
•a, o
A o
V
20 40 60
Degree of Hydration (%)
80
100
Figure 27 - Specific Surface Area of Vacuum Oven-Dried
Cement Pastes When Expressed on the Basis of
Total Ignited Weight of Samples (No. 2)
92
When specific surface area was expressed in terms of
ignited weight of the hydrated portion of the sample,
rather than the total sample, interesting curves were
obtained. They are shown in Figures 28, 29, and 30. These
curves are similar to those of the shrinkage results, shown
previously. The samples that showed larger shrinkage at
earlier age have also hydration products with larger surface
areas at this age. At the degree of hydration of about 5 5
to 60% where the shrinkage curves showed a minimum, curves
for surface area also show it, for the most part. Maximum
specific surface area is attained when 85 to 90% of cement
is hydrated, except for the sample with calcium chloride,
and the surface area is reduced to some extent on further
hydration. In the shrinkage curves, the rate of increase
in shrinkage was also reduced at this degree of hydration.
The cement paste with calcium chloride showed a strong in-
crease in shrinkage at this stage and does so also in the
surface area curve. The curves for sucrose and saccharide
retarder (S) have a similar trend after more than 35% of
the cement is hydrated. At earlier ages, however, sucrose
increases surface area of hydration products remarkably
as it did also the shrinkage of cement paste containing it.
The only exception was the relatively young cement pastes
with 3-hydroxy-2-butanone; they had hydration products of
higher surface area, but shrank less than the others.
93
2 -i < w
X
'5 *- i» u
t t> D b
TJ P T) TJ
o a* at a*
zcKtt:
TT
o •
I
II I
* o •
8
j i i
>>
0>
l_
Q
O
O
(B/ lu) D9JV dOD^jns Di^pads
o
o
O
m
o
lO
CM
CM
r0
0)
■H
u
a
i
c
0)
>
o
g
0
id
>
O
a
o
■H
P
S-i
O —
a< en
P U
rfl (0
l-i -P
t> a)
>.«
rH
4-1 rtj
O H
O
(0 >-i
QJ d)
o
0) u
o —
(0
<+-! U)
^ a)
p -p
co en
rfl
U Pn
•H
4-1 P
•H C
U QJ
a) e
Qj QJ
LO CJ
I
0)
1-1
tn
94
TJ
0)
■H
M
Q
C
Q)
>
O
e
3
3
0
nj
>
m
0
^-s.
a*
c
0
•H
c
-M *-N
o
U r-i
+-
0
0
o< •
L.
o
T3 S3
I
+J l
M—
flj
o
>-l 03
TJ -1
<l>
>. m
p
ffi o
rr>
•H
qj
4-1 g
Q
O (1)
H-\
e
o
a)
.r:
03
u
(D
H
<D
<t
M
3
<D
&<
o
— -
Ifl
<H
01
M
0)
3
+J
CO
W
HJ
u
f_
•iH
4H
4J
■H
C
u
(U
(1)
g
ft
1)
C/)
u
o
^ (6/ cu) D3J\/ aDD^jns ajjpads
c
1)
(Jl
o o o
m o in
96
Scanning Electron Microscopy of Cement Pastes
Unhydrated cement grains seen through the scanning
electron microscope are shown in Figure 31. Cement graim
of various sizes ranging from 0.1 to more than 15 microns
can be seen. Their surface texture looks smooth at this
magnification. Figures 32 - 50 show pictures of hydrated
cement pastes. Samples for Figures 32 - 46 were cast on
glass, and the surface that faced the glass was observed
after its removal by drying shrinkage. Concentrations of
additives in these samples are those that yield a relative
retardation of 200% in mortars. Figures 47 - 50 are photos
of broken surfaces of samples that had been used for shrink-
age measurements. Age shown is elapsed time or curing time
before the sample was started to be dried in the oven.
At an elapsed time of 1 min, addition of retarders
resulted in remarkable differences in the morphology of
the hydration products. This can be seen in Figures 32 -
36. When no admixture was added, hydration products were
observed on the surfaces of the grains. With calcium
lignosulf onate added, the appearance of the product was
changed; some particles were long and some were spherical.
Their frequency seemed to be lower and their size larger.
When citric acid was added, many needle-like crystals were
seen. The needles were mostly observed to grow in a way
so that several radiated from a point located either
between large cement grains or sometimes from small grains
97
Figure 31 - Unhydrated Cement Grains (X5000)
t i 1
Figure 32 - Cement Paste with No Admixture (X3000)
Age: 1 min
98
Figure 33 - Cement Paste with No Admixture (X9000)
Age: 1 min
Figure 34 - Cement Paste with Calcium Lignosulfonate (X5000)
Age: 1 min
99
Figure 35 - Cement Paste with Citric Acid (X3000)
Age: 1 min
Figure 36 - Cement Paste with Sucrose (X5000)
Age : 1 min
100
of cement. EDAX analysis showed that these needles were
composed of some silica, alumina, and sulfur and a large
amount of calcium. It should be noted in the same figure
that the surfaces of most of the cement grains remained
relatively unchanged. In the sample containing sucrose
irregularly shaped and rounded hydration products were seen
on the surfaces of cement grains.
Photos of cement pastes at the age of 5 min are shown
in Figures 37 - 40. More hydration had occurred in all
samples. Most of the cement grains were covered with hydra-
tion product in the case of neat cement paste, and a hexa-
gonal plate, which is probably a lime crystal, can be seen.
The cement paste with calcium lignosulf onate still retained
some rod-like projections. The long needles observed in
the 1 minute citric acid sample tended to be fewer and
seemed to be precipitated on the surfaces of the cement
grains.
Figures 41 - 44 show that the long needles observed in
citric acid samples at early ages disappeared as time
elapsed. In the 10 minute sample, it is seen that most of
the shortened needles were gone or transformed into rounded
products on the cement grains. These round products look
like those observed in 1 minute samples containing calcium
lignosulf onate or sucrose. In 30 minute and 1 hour samples
containing citric acid, no needle was seen, and the
101
Figure 37 - Cement Paste with No Admixture (X5000)
Age: 5 min
Figure 38 - Cement Paste with Calcium Lignosulf onate (X5000)
Age : 5 min
102
Figure 39 - Cement Paste with Citric Acid (X3000)
Age: 5 min
iL. i &
Figure 40 - Cement Paste with Sucrose (X5000)
Age: 5 min
103
Figure 41 - Cement Paste with Citric Acid (X3000)
Age: 10 min
Figure 42 - Cement Paste with Citric Acid (X8000)
Age: 10 min
104
Figure 43 - Cement Paste with Citric Acid (X3000)
Age: 30 min
Figure 44 - Cement Paste with Citric Acid (X3000)
Age: 1 hr
105
transformed products continued to spread over the surface
of the cement grains.
Figure 45 is a photo of a cement paste with no ad-
mixture after 1 hour. Cement grains appear to be covered
by many grainy hydration products. The sucrose-containing
sample at 30 min is shown in Figure 46 and is similar to
the 1 hour sample containing citric acid.
Figures 47 - 50 show broken surfaces of cement pastes
with and without retarders at ages of about half a day.
The degree of hydration is also indicated. No noticeable
difference is observed among the figures.
No photo is shown for cement pastes older than these
samples in this thesis, because there was no notable
difference observed regardless of the presence of retarders,
Additionally, at ages later than 4 days, strong growth of
lime crystals obscured the observation.
106
Figure 45 - Cement Paste with No Admixture (X5000)
Age: 1 hr
Figure 46 - Cement Paste with Sucrose (X3000)
Age: 30 min
107
Figure 47 - Cement Paste with No Admixture (X2000)
Age: 12.5 hr
(degree of hydration: 24%)
Figure 48 - Cement Paste with Glycolic Acid (X2000)
Age: 15.5 hr
(degree of hydration: 26%)
108
Figure 49 - Cement Paste with Retarder A (X2000)
Age: 14 hr
(degree of hydration: 18%)
>~\. * i
Figure 50 - Cement Paste with Retarder S (X2000)
Age: 13 hr
(degree of hydration: 2 0%)
109
CHAPTER VI - DISCUSSION
Molecular Structure of Retarders
One of the purposes of this study was to gain more
understanding of the aspects of molecular structure that
make a substance an effective retarder. The previous work
of others (27) has shown that organic retarders frequently
contain hydroxyl groups and that one of these is frequently
in the a position (i.e., on the carbon atom adjacent) to a
carbonyl group as, for example, in glycolic acid,
CH2(OH)C02H, or
CHo-C-0H
i ^ ii
OK O
It is true that most of the strong retarders used in
this work do have this structure. Those with relative
retardations of more than 200% when tested at 0.1% con-
centration (by weight of the cement) are:
Tartronic acid HO-C-CH-CO_H
OH
Tartaric acid HO-C-CH-CH-CO-H
* l l *
OH OH
Mucic acid HO-C-CH-CH-CH-CH-CO-H
* I I I I z
OH OH OH OH
Gluconic acid CH--CH-CH-CH-CH-CO-H
I * l I I l *
OH OH OH OH OH
110
Citric acid
Pyrogallol
CO^H
H02C-CH2-C-CH2-C02H
OH
Gallic acid
2 ,4 ,6-Trihydroxybenzoic acid
C02H
HO
^^
OH
Sr
Sucrose CCH, . Oc-0-C,,Hn ,Oc
6 11 5 6 11 5
Pyruvic acid CHo-C-C0oH
J II 2
0
and a-Ketoglutaric acid H02C-CH2-CH2-C-C02H
Of these, all are acids except sucrose and pyrogallol
In connection with the latter compound it should be noted
that hydroquinone
OH
OH
Ill
almost fell into the class of compounds listed above, with
its relative retardation of 198%, and is also not an acid.
The compounds listed above contain hydroxyl groups, usually
many of them, and almost always, at least in aliphatic
compounds, there is one in the a position to the C = 0
of the carboxyl group. The only exceptions to these
generalizations are the last two compounds in the list,
pryuvic acid and 2-ketoglutaric acid. These two have a
carbonyl group in the a position to the carboxyl group.
This carbonyl group perhaps hydrates, to some extent, in
solution to form a 2,2-dihydroxyl compound stabilized by
the strong electron-withdrawing tendency of the adjacent
carboxyl group. Another example of this condition is
glyoxylic acid, which exists as the hydrate, even in the
solid state. So these a-carbonyl compounds may qualify
with respect to the hydroxyl group criterion, even though
it is not apparent from the conventional formula.
The importance of the hydroxyl group is further
emphasized by the changes that occur when it is exchanged
for other groups. The relative retardation (T500) of
glycolic acid
CH--CO„H
OH
is 127%, a moderately strong retarder and the simplest
example of an a-hydroxy acid. But if the hydroxyl group
is exchanged for a sulfhydryl group, forming mercaptoacetic
112
acid
CH,.-C0oH
SH
the T500 drops to 113% and for glycine
CHo-C0-H
I l 2
NH2
it is 100%, indicating no retarding action at all.
On the other hand, the presence of the hydroxyl group
in the a position to the carboxyl group is not, in itself,
sufficient for a compound to be a retarder. This conclusion
is drawn from the insignificant retarding abilities of
lactic acid
CH-.-CH -CO_H
OH
a-hydroxybutyric acid CH3-CH2-CH-C02H
OH
and mandelic acid,
-CH-C02H
all of which are a-hydroxycarboxylic acids.
If the reason for the frequent efficacy of an
a-hydroxy acid as a retarder is the activation of the
hydroxyl group in the 3,4 position with respect to the
double bond, then other compounds with a similar structure
should behave similarly. Allyl alcohol
CH0=CH-CH0
Z I 2.
OH
113
has no retarding ability; neither does 2-butene-l, 4-diol,
CH~-CH=CH-CH0
OH OH
but both have hydroxyls activated by the double bond. On
the other hand, a carbonyl group does seem to have some
effect, even if it is not part of a carboxyl group, as in
3-hydroxy-2-butanone
CH,-C-CH-CH0
3 ii I 3
0 OH
which has a T500 of only 107% at a concentration of 0.1%
but 173% at a concentration of 0.5%. Another example is
1, 3-dihydroxy-2-propanone
CH--C-CH-
I 2 II l 2
OH 0 OH
which has a value of 125% at 0.1%, but seems to act
atypically, with a value of 97%, i.e. slight acceleration,
at 0.5% concentration. The curve is, however, unusual.
See Figure 13. These last two examples cannot be said to
be strong retarders, although there is some action.
The presence of many hydroxyls is not, of itself,
sufficient to confer retarding abilities on a molecule, as
is shown by the result that glycerine
CH--CH-CH0
I 2 I I 2
OH OH OH
and pentaerythritol
114
OH
l
CH0
I *
H0-CHo-C-CHo-0H
Z | z
CH0
i *
OH
do not retard. Sometimes, however, strong retarders have
only hydroxyls as functional groups, as has already been
shown by the examples of hydroquinone and pyrogallol.
It is possible to have good retardation without the
presence of the alcoholic hydroxyl group, as with 1,2,3-
propanetricarboxylic acid
CH- — CH — CH-
I ^ I I *
C02H C02H C02H
which had a T500 of 139, a value higher than that of
glycolic acid. At the same time, it should be recognized
that if the hydroxyl is not absolutely necessary, it is
usually helpful in promoting retarding ability of a
molecule; if the tertiary hydrogen in the above compound
is replaced by a hydroxyl group, giving citric acid, the
retarding ability is greatly promoted.
Other good examples of this effect involve the
dicarboxylic acids. The acids malonic, succinic, glutaric,
and adipic have only slight retarding ability at 0.1%
concentration, although if five times as much is used,
malonic and succinic acids have a respectable action. But
the addition of hydroxyls greatly increases the retarding
ability. At a concentration of 0.1% succinic acid
115
H02C-CH2-CH2-C02H
has a T500 of only 110%. The addition of only one hydroxyl
to make malic acid
H02C-CH2-CH-C02H
OH
causes a great increase in a T500 of 198%. A second
hydroxyl gives tartatic acid
H0-C-CH-CH-C0oH
2. | | 2
OH OH
which has a T500 of 254% and is a powerful retarder.
Glutaric acid
H02C-CH2-CH2-CH2-C02H
has a T500 of also only 110%, but 2-ketoglutaric acid
H0oC-CH--CHo-C-C0„H
0
which may, as previously mentioned, exist as
OH
I
H02C-CH2-CH2-C-C02H
OH
in solution, has a T500 of 292%, one of the highest values
measured. A similar effect is observable with pyruvic acid
CH-,-C-C0oH
.3 || 2.
or
OH
I
CH„-C-C0oH
3 | 2
OH
116
which is a strong retarder with a T500 of 207%, whereas
the mono-hydroxy counterpart, lactic acid
CHo-CH-C0_H
j l Z
OH
has no retarding ability at all at 0.1% concentration and
not much at 0.5% (T500 = 118%).
Sometimes, however, more hydroxyls seem to be too
much of a good thing. For example, dihydroxytartaric acid
OH OH
I I
H0~C - C - C - C0oH
OH OH
is not as good a retarder (T500 = 166%) as tartartic acid
(T500 = 254%) , although it is still a strongly active sub-
stance. The same effect is observed with the hydroxy
devivatives of malonic acid,
H02C-CH2-C02H
which has a T500 of only 102%. Addition of one hydroxyl
group gives tartronic acid
H02C-CH"C02H
OH
which is a powerful retarder with a T500 of 278%. A
second hydroxyl gives the hydrate of ketomalonic acid
OH
I
H0oC-C-C0oH + Ho0 > H0_C-C-C0oH
Z || Z Z Z Z
0 OH
which is still a strong retarder, with a T500 of 154%, but
is not nearly so strong as tartronic acid.
117
From the above it is apparent that the hydroxyl group
is frequently found in retarder molecules and is frequently
a to a carbonyl, but that this structure alone, or the
plain hydroxyl alone, is not sufficient in itself to confer
a high retardation ability on a molecule.
Influences of molecular structure are observable also
with respect to the different retarding abilities of
position isomers. Salicylic acid, which has been recognized
as a set retarder for some time, has a T500 of 122%. It
is o-hydroxybenzoic acid.
C02H
0
But the meta and para isomers
and
OH "OH
had practically no retarding ability. Of the three
dihydroxybenzenes, resorcinol, the meta isomer,
OH
is not a retarder, but the ortho isomer, catechol
OH
OH
118
is a fairly good retarder, with a T500 of 127%, and the
para isomer, hydroquinone ,
Nr
is better yet, with a T500 of 198%. It is, of course,
usual for ortho and para isomers to be alike in chemical
properties and different from the meta compound. Another
example is the trihydroxybenzenes. Phloroglucinol,
1,3 ,5-trihydroxybenzene,
OH
HO
k^H
has a T500 of 131%, a fairly good retarder, but pyrogallol,
1,2 ,3-trihydroxybenzene ,
OH
/%, OH
^^
OH
is much stronger, with a T500 of 239%.
From this discussion certain generalities can be
inferred empirically with respect to the molecular
structures that characterize retarders. They are frequently
carboxylic acids, but need not be necessarily. They
frequently contain many hydroxyl groups, but need not
necessarily. The hydroxyl is frequently in the a position
119
to the carboxyl. And some compounds containing most of
these features are not especially good retarders.
All the really strong retarders used here are highly
oxygenated substances. The oxygen atoms can be in hydroxy 1,
carboxyl, or carbonyl groups; all seem to be effective,
although some are, no doubt, more so than others. The
oxygen atoms are also usually fairly well spread out across
the molecule. That is, significant portions of the molecule
devoid of oxygens seem to lessen the retarding ability as,
for example, with the dicarboxylic acids, lactic acid, and
a-hydroxybutyric acid. The strong retarders, furthermore,
usually contain several oxygen atoms that can, by virtue of
their location on the chain and rotations about single
bonds, approach each other closely and form a "cluster" of
oxygen atoms. In this connection, it should be remembered
that chain structures are kinked at tetrahedral angles, so
the usual formula as written is not a particularly clear
indication of the real spatial relationships in the
molecule.
The foregoing notions, of course, merely repeat to
some degree the older postulates of Hansen (24) , Steinour
(25), and especially Taplin (27).
Deserving perhaps of special comment is the seemingly
great strength of the a-carbonyl group. All substances
tested that have carbonyls adjacent to carboxyl groups -
120
pyruvic acid, ketoraalonic acid, and 2-ketoglutanic acid,
were strong retarders.
One can go a little further and say "Why oxygen?"
Why will not other groups function just as well? But it
seems that the oxygen atom is one form or another is almost
uniquely required for the retarding action. For example,
glycolic acid
CHo-C0oH
OH
is a fairly good retarder, but mercaptoacetic acid, glycine,
and monochloroacetic acid in which the hydroxyl has been
exchanged for -SH, -NH2 , and -CI, respectively, are all
poor or completely ineffective as retarders.
The properties of the oxygen atom that seem most likely
to be of consequence are its high electronegativity, its
consequent ability to participate in strong hydrogen bond-
ing with other electronegative atoms, its small size, and
its presence in the molecules of both the liquid phase of
concrete and the surfaces of the anhydrous cement minerals
and of their hydration products.
It may be that one important function of the highly
electronegative oxygens is to bring about electron shifts
in the retarder molecules and induce charge polarizations
that could determine the adsorption characteristics of
the molecules on the charged solids present.
121
For example, consider the differences between glycolic
acid,
CH~-C0oH
OH
lactic acid,
and pyruvic acid,
CH0-CH~C09H
OH
CHo-C-C0oH
J ii I
0
It should first be emphasized that in the relatively
highly basic medium of cement paste or concrete (about
pH=12.5) the carboxyl hydrogen atoms of all these weak
acids will be dissociated, leaving the anion as the species
really present in the aqueous phase.
In glycolic acid it can be postulated that small local
positive charges are generated on the carbons or the
hydrogens by the polarizing power of the various oxygens.
Lactic acid is an extremely weak retarder, regardless of
the fact that it has only a methyl group attached to the
second carbon in place of the one hydrogen atom of glycolic
acid. On the contrary, pyruvic acid which has one oxygen
attached to the second carbon by a double bond, instead of
the hydroxyl group of lactic acid, retarded the hydration
of cement strongly. As has already been noted, the pyruvic
acid may hydrate in aqueous solution to give the dihydroxy
compound with both groups on the alpha carbon atom.
122
In pyruvic acid the double bonded oxygen or the
oxygens in the hydroxyl groups, as the case may be, may
pull the electrons around the third carbon toward the second
carbon and polarize the hydrogen atoms on the third carbon
atom. This trend may be further promoted by the oxygens
in the acid group. This kind of polarizing effect on the
hydrogens in the methyl group would be expected little
for the lactic acid molecule, although the hydrogen of the
second carbon could be subjected to the effect, mainly by
the double bonded oxygen in the acid group and partially
by the oxygen in the hydroxyl group.
When one hydrogen of the third carbon of lactic acid
is replaced by another methyl group, the result is a-
hydroxybutyric acid, which was not a retarder. However,
if the end methyl group of lactic acid or a-hydroxybutylic
acid is replaced by an acid group, the strong retarders
tartronic acid and malic aaid, respectively are made.
Possibly all the hydrogens of the latter two chemicals
could have the aforementioned polarizing effect. Hence,
it appears that perhaps the movement of electrons should
be considered in seeking the effective molecular structure
of retarders.
The question of the relative locations of hydrogen
atoms and atoms of high electronegativity (usually oxygen)
arises next. The dicarboxylic acid compounds - malonic,
succinic, glutaric and adipic acids - have an acid group
123
at each end of the molecule, but no hydroxy 1 groups. The
former two are weak retarders when added at relatively high
concentrations, while the latter two are not retarders at
all. If the differences in retardation powers of these
chemicals are caused by the polarizing effects of the
carboxyl groups on nearby hydrogens, the result would
indicate that at most only two hydrogens attached to the
second carbons could have the influence to some extent.
Following this logic, it can be concluded that the
greater retarding abilities of the compounds in which
hydroxyl or carbonyl groups lie along the chain between
the acid groups (tartronic, ketomalonic, malic, tartaric,
dihydroxytartaric, 2-ketoglutaric, and mucic acids) are
due, at least partly, to the increased polarization effects
of these groups on the other atoms in the chain, effects
which are lacking in the simple acids.
For the oxygen in the hydroxyl group, the extent of
this effect is found by comparing the above result and the
molecular structures of the chemicals already referred to.
Lactic acid was an extremely weak retarder. 3-Hydroxy-
propionic acid and a-hydroxybutyric acid were not retarders.
Hence, the hydroxyl oxygen atom could have this effect on
only the one hydrogen atom attached to the same carbon.
The hydrogen atom in the hydroxyl group would be, of
course, also polarized.
124
If these ideas are applied to the glycine molecule,
the reason why it is not a retarder may be also explained.
The nitrogen of glycine has two hydrogens bonded to it,
but it has lower electronegativity than oxygen. Therefore,
the nitrogen may not have enough ability to polarize even
one hydrogen bonded to the second carbon atom. Additionally,
the polarizing effect per hydrogen which is bonded directly
to the nitrogen would be less than half of that of the
hydrogen in hydroxyl group, and the resulted polarization
might not be strong enough to permit the substance to act
as a retarder. The small retarding ability of mercapto-
acetic acid, compared with glycolic acid, can be also
explained by the considerably lower electronegativity,
therefore the polarizing effect, of sulfur than oxygen.
The discussion so far given is confined to several
chemicals. The idea can be generalized; that is, the
molecular configuration of effective retarders is that
atoms in the molecule are arranged in such a way to
polarize as many hydrogen atoms as strongly as possible.
Oxygen atoms are mainly responsible for the polarizing
process because of their high electronegativity and low
valence. However, it should be noticed that the effective
configuration does not necessarily require carboxylic acid
groups. If the idea is correct, it should be able to be
generalized, at least somewhat. The modest retarding
abilities of 1, 3-dihydroxy-2-propanone
125
I 2 II I '
OH O OH
and 3-hydroxy-2-butanone
CH^-CH-C-CH-
J I II 3
OH 0
are examples of such generalization.
The rationality of these ideas is further supported
by the fact that the electrogenativities of hydrogen,
carbon, sulfur, nitrogen, and oxygen are 2.1, 2.5, 2.5,
3.0, and 3.5, respectively. The larger the difference in
electronegativity between two atoms, the more ionic is the
bond. Therefore, stronger polarizing effect would be
expected for O-H combination than for N-H or S-H
combination.
These ideas are, furthermore, subject to semiquanti-
tative verification. For example, 2 , 4-pentanedione
CH-.-C-CH^-C-CH-,
J II l ,| J
0 0
and ethyl acetoacetate
CH,-C-CH0-C-0-C0Hc
3 || 2 || 2 5
0 0
are both mild retarders, as would be expected from the
strong polarizations of only the methylene hydrogens.
Generally, little retardation is expected unless more than
two-thirds of the hydrogens in a molecule are polarized
evenly throughout the molecule and the polarizing effect
is strong enough.
126
However, it should be noticed that the relative
effectiveness of some compounds, e.g. tartaric, dihydroxy-
maleic, and dihydroxytartaric acids, can not be explained
by these ideas. Also, ketomalonic acid should have been
a stronger retarder than tartronic acid, because the middle
double bonded oxygen of ketomalonic acid will be opened
in water and will form two hydroxyl groups. These apparent
contradictions were probably caused by the many other
physical and chemical influences of chemicals in cement
paste.
What has been said previously has dealt with the
aliphatic compounds tested. The aromatic compounds also
show many of the same features, as well as interesting
effects of structural isomerism mentioned earlier. The
strong aromatic retarders are all highly oxygenated and the
strongest retarders, gallic acid
pyrogallol
and 2,4 ,6-trihydroxybenzoic acid
127
OH
all have groups of oxygen atoms that approach each other
closely. In this connection, it should be recalled that
most of the hydrogen atoms of strong aliphatic retarders
are polarized strongly by many oxygen atoms and that these
oxygen atoms are usually located in such a way that they
also approach each other closely.
Benzoic acid is not a retarder, nor is its meta or
para hydroxy acid, but salicylic acid is a moderate re-
tarder. Since the hydroxy group of salicylic acid is
located on the second carbon from the acid group, it might
seem that it would be analogous to the non-retarding
3-hydroxypropionic acid. But the latter is a much more
flexible molecule with configurations in which the hydroxyl
and carboxyl are relatively removed from each other,
whereas in salicylic acid the two groups are constrained
to proximity and also tied together by their intramolecular
hydrogen bond.
The trihydroxybenzenes are good retarders, but pyro-
gallol, in which the hydroxyls are in adjacent positions
on the ring, is a much stronger retarder than is phloro-
glucinol, in which they are separated. Converting either
to its acid improves the retardation ability still further.
128
Gallic acid (3 , 4 , 5-trihydroxybenzoic acid) is an even
stronger retarder than is 2 , 4 , 6-trihydroxybenzoic acid.
The three dihydroxybenzenes present an apparent
exception to the generalization involving contiguous
hydroxyls or oxygen atoms. The ortho isomer, catechol, is
a fair retarder, while the meta isomer, rescoicinol, is
not active at all. But the para isomer, hydroquinone, in
which the hydroxyls are separated to their greatest extent,
is the best of all, with a T500 of 198%.
The findings of this study agree generally with those
of Steinour (25, 26) with respect to the importance of
hydroxyl groups and of Taplin (27) with respect to the
importance of the hydroxyl adjacent to a carbonyl group.
But, as mentioned earlier, glycerin and pentaerythritol
are not retarders in spite of many hydroxyl groups.
Propanetricarboxylic acid is a good retarder with no
alcoholic hydroxy groups; citric acid is very strong with
only one. Lactic and mandelic acids fulfill Taplirfs
criteria, yet they are not good retarders, and so on.
There are so many exceptions that it is clear the older
generalizations must leave out important aspects of the
problem.
Effect of Retarders on the Hydration of Cement
In an effort to explain some of the ways in which
retarders influence the hydration of portland cement,
129
some facts concerning the process need to be reviewed
briefly.
The principal active components of portland cement at
early stages of hydration are CUS and C^A. Through the
hydration of C^S, calcium silicate hydrate is formed and
lime is released into solution in the form of Ca++ and OH .
C,A, on the other hand, reacts with sulfate and calcium
ions from the solution to form ettringite, C,A- 3CS 'H-.- .
In a heat evolution curve of cement hydration, the
initial rapid hydration is followed by a dormant period,
perhaps a couple of hours long, and the second heat
generation, which is mainly due to additional hydration of
C,S, occurs at the end of the dormant period (37, 50).
According to Verbeck (50) , the initial setting time of
cement paste lies somewhere between the end of the dormant
period and the time when the maximum peak of the second
heat generation is reached. Previte (2 9) obtained the same
result.
Powers (5) has discussed the details of the processes
he considered to be taking place up to the decline of the
heat evolution rate after the second peak. This is the
period in which we are interested from the standpoint of
the setting process and the influence of retarders thereon.
The dormant period is thought to be due to a great
diminution of the hydration reaction brought about by a
coating of hydration products that limits the access of
130
water to the underlying anhydrous grains. For the C~A the
coating is presumably ettringite. For the C~S, with
which we are more concerned here, the coating is an
initially-formed C-S-H gel. Powers (5) considered the
reality of these ideas to be substantiated by various
electron microscope studies of the appearance of the paste
soon after mixing (52).
Powers considered the end of the dormant period and
the beginning of the second heat evolution peak to be
caused by a breakup of the coatings of hydration products
that would allow an increase in the rate of hydration to
take place. Indeed, if the dormant period is caused by
such a coating, its removal, at least partially, is
logically necessary for an increase in the hydration. The
reason given for the breakup of the coating is the genera-
tion of an osmatic pressure by the concentration difference
between the solutions outside the coating in the bulk
aqueous phase and inside the coating adjacent to the
anhydrous cement minerals.
In his recent review, Young (32) has classified the
various possible modes of action of retarders into four
categories. These are (i) adsorption, (ii) complexation,
(iii) precipitation, and (iv) nucleation. The applicability
of the results of the present study to these ideas will be
discussed in reverse order.
131
One possible action is an influence on the nucleation
of crystalline calcium hydroxide, changing the time at
which macroscopic lime crystals are observed, which is
about the same time as initial set. If this inhibition of
crystallization and consequent supersaturation of lime in
solution can inhibit further reaction of cement with water,
then it may be important to the setting process. Young
(32) considers this as perhaps the most important mode of
action of retarders. It is well-known that small amounts
of some dyes, surfactants, and other solutes can profoundly
influence the habit and growth rate of crystalline phases
(53) . The most recent study of this effect in the phases
of interest here is that of Berger and McGregor (54) who
found that some of the substances studied here altered the
crystal size and habit of the CH formed during C-.S hydra-
tion. They used, however, concentrations ten times those
usually used in this study.
The only results developed here that apply directly
to this aspect are those in which the four dyes were
tested. These dyes were selected from the list of Buckley
(53, p. 558) as being substances that strongly modify the
habit of some growing crystals. They show some influence
on the habit of CH grown slowly by interdif fusion of the
ions (55) . The change in the habit would have been caused
by the effect that Young called "adsorption of organic
compounds on the calcium hydroxide nuclei". But they show
132
negligible action as retarders. This is, of course, no
proof that the substances that were strong retarders did
not act in this way.
The precipitation idea is that retardation is brought
about by some precipitation of, for example, the calcium
salt of the retarder substance on the surface of the
anhydrous grains. Young places no great importance on
this idea. The applicability of this idea to the data of
this study is referred to next in the discussion of
complexation .
The complexation theory envisages the combination of
the agent as some sort of complex, most likely with calcium
or aluminum. It is true that many of the substances tested
form complexes of greater or less stability with calcium
or other metal ions. Furthermore, two of the substances
that complex calcium strongly, ethylene diamine tetracetic
acid (EDTA) and nitrilotriacetic acid (NTA) are fairly good
retarders, although the larger dosage of EDTA did not
result in as great an increase in retardation as was the
case for the better retarders (Figure 15) . Young (32)
pointed out, there is no correlation between the stability
constants of the calcium complexes of these substances and
their potency as a retarder.
Any effect may even be in the opposite direction.
The review of literature in Chapter II showed the retarders
(lignosulfonate, sucrose, salicylic acid) generally
133
accelerated the very initial hydration of cement, but
retarded subsequent hydration and transformations of some
hydration products. The reason for the initial accelera-
tion of cement hydration due to the addition of retarders
has to be sought in some changes in ion concentration in
the aqueous phase of cement paste. A few minutes after
mixing cement with water, ions in the solution of cement
paste would be saturated or supersaturated with respect to
possibly each ion concentration. Any consumption of these
ions would accelerate the hydration of the phases that can
supply the ions into the solution. The consumption could
be attained by any process such as precipitation or
crystallization of hydration products, or complex forma-
tion of the ions with some chemicals.
It should be recalled that molecules of strong re-
tarders contain many highly electronegative groups such as
hydroxyl, acid, and carbonyl. These groups could complex
metal ions in the aqueous phase of cement paste (13, 56) .
Whatever the states of the complexes, this complex forma-
tion would accelerate the hydration of cement paste,
especially in the very early stages of the hydration,
because more free organic molecules are available for
complexing.
As a rough indication of the complexing ability of
these materials under the conditions present in a cement-
itious system, the following test was performed. A
134
0.124M NaOH solution, which has an alkalinity equivalent
to the aqueous phase of the usual cement paste, was prepared.
In 100 ml of the solution, 0.2g of the retarder chemical was
dissolved in a beaker. This gave a concentration in the
solution equal to that in a 0.5 w/c paste. The solution
was continuously stirred by a magnetic stirrer and was
titrated with a 0.5M CaCl2 solution until the first
precipitate was detected. The result, volume of CaCl-
solution necessary to give a precipitate, was plotted in
Figure 51 against the relative initial setting time of the
corresponding mortar specimen (w/c=0.50) in which the
retarder has been added at a concentration of 0.1% of the
cement.
In the blank test, the first precipitate of Ca(OH)_
was observed when the calcium chloride solution consumed
was 0.9, 1.1, 0.9, 1.0, 0.8, 1.0, and 0 . 9 ml for several
tests, the average being 0.94 ml. The chemicals that
required more CaCl2 solution for the titration that this
blank amount must have complexed some calcium ions in the
solution. The chemicals which show stronger complexing
ability such as sucrose, gluconic acid, citric acid, and
catchol would presumably induce more acceleration during
the initial hydration of the cement, as mentioned earlier.
When some other chemicals were added, on the other
hand, precipitates were observed earlier than in the case
of the blank solution. These precipitates, presumably,
135
+->
■H
Cn
c
■H
X
<u
H
a,
E
o
u
>
4J
•H
<
C
H
"J
yt
rrj
-p
Pi
m q- ro c\j —
(|tu) uoudjjij. JOi papaaN uojjnios 2iddo
136
are not lime, but are insoluble calcium salts or complexes
of these chemicals. Calcium lignosulfonate would fall in
this category (19) .
In any event, Figure 51 shows that many substances of
strong "complexing" ability, as evidenced by the consumption
of a relatively large amount of the CaCl- solution, are
also strong retarders. But there are also many exceptions
and the results are only qualitative. Also, as implied
earlier in this section, any such agreement may be acci-
dental if the main effect of complexation is to accelerate
the very early reactions.
The discussion so far has emphasized the importance
of the C-jS and has made little mention of the aluminates.
In a cement that is normally-retarded with gypsum, the
formation of ettringite acts to retard the otherwise too-
rapid reaction of the C~A. The aluminates, and especially
the hydrated aluminates, have been shown to react with
certain retarders to form adsorbed phases and/or complexes
(7, 9, 14, 15). Such reactions may change the rate of
consumption of various species, e.g. sulfate, and are
important and complicated in their own right. But for
the purposes here, it is considered that the main influence
of the aluminates is to tie up more or less of the ad-
mixture in one way or another and so make less of it
available to do whatever it does in connection with the
silicate hydration process. The extent to which this
137
occurs with most of the retarders used in this study has
not been investigated. It is, however, known that re-
tarders would be more effective when used with cement of
low C3A (16, 17) .
After and/or during the initial acceleration of cement
hydration, retarders have to act in some way to retard
subsequent cement hydration. Change in ion concentration
in the aqueous phase of cement paste would probably not be
concerned with the retardation mechanism any more. Hence,
the mechanism has to be sought in direct interaction between
remaining anhydrous cement grains or hydration products of
cement and retarder molecules or complexes thereof.
Additionally, the interaction must proceed in such a way
that the effective molecular structures of retarders are
most efficiently utilized for the retardation.
The adsorption, or "coating", theory was the earliest
put forward to account for the influence of retarders on
cement hydration. It was assumed that the retarder in some
way became affixed to the cement grains and prevented their
contact with the water in the same way that the ettringite
on the C^A and the calcium silicate hydrate on the sili-
cates were presumed to act. This idea is consistent with
the results of calorimetric experiments that showed re-
tarders to extend the dormant period, i.e. delay the onset
of the second heat evolution peak and spread it out over
a longer time and with a smaller maximum value of heat
138
release rate (10, 16, 57) . These results were for both
Portland cements and C~S. Although the experiments were
not extensive in terms of the admixtures used, they form
the basis for the belief that, whatever else they may do,
retarders delay set by delaying the chemical reactions
between cement and water.
Several possible ways in which retarders might form
such a coating or film have been suggested.
Hansen (11) suggested that the surfaces of C,S (and
C~A) would be principally Ca++ or 0 ions, because these
are larger and more numerous than ions of silicon and
aluminum and, therefore, the latter would be screened by
the former. Organic acids could presumably be attached
by ionic attraction between Ca ions in the surface and the
anionic carboxylate groups. Another suggested possibility
(25) involves the fact that most retarder molecules contain
hydroxyl groups, which are known to form hydrogen bonds
with highly electronegative atoms such as oxygen atoms in
the surfaces of the cement grains. Such bond could also be
a mode of attachment of the retarder molecule.
Young (32) has also suggested that the attachment
could be via some complex between surface atoms and the
retarder molecules.
The work done by Blank et. al (7) , however, showed
that adsorption of retarders on C\S from aqueous solution
was negligible. Recently the same result was obtained with
139
more realistic concentrations of materials by Diamond (9) .
These results cast considerable doubt on any mechanism of
retardation that involves adsorption of the agent on the
anhydrous C.,S.
On the other hand Ramachandran (22) showed that it was
hydrated C.,S, not C3S, that was responsible for a percept-
ible amount of adsorption of calcium lignosulfonate from
an aqueous solution. Diamond (9) obtained the same result
with salicylic acid. The details of the surface structure
of the C-S-H gel are unknown. Only speculation can be
o
based on analogies with the structure of 11 A tobermorite
(58) with which the gel is thought to have many similarities,
It seems safe to say that the surface is probably composed
of either oxygens or hydroxide groups and is partly covered
with calcium ions abstracted from the solution phase that
give it an overall net positive charge.
The formation of a discrete film of retarder molecules
on the anhydrous minerals is conceivable. Indeed, the
difficulty lies in the idea of too thick a film. If one
assumes a concentration of retarder of 0.1% of the cement,
a molecular weight of 100, and a coverage per molecule of
°2 2
50 A , then a cement with a fineness of 3000 cm /g would
have a layer of retarder ten molecules thick, which is
unrealistic in light of the known processes of adsorption
from solution.
140
If one considers the situation of adsorption on the
hydrated phases, the results are more believable. Some
dispute exists concerning the thickness of the hydrate
layer covering the cement grains during the dormant period.
Estimates range from several angstroms to several hundred
(59) . If one assumes 5% of the cement is hydrated during,
or at the end of, the dormant period, and if the specific
2
surface of the hydrated product is 2 00 m /g, then the
retarder would be sorbed on only 25% of this material.
It was found in this work that the chemicals effective
as retarder have molecular configurations in which most of
the hydrogen atoms around a molecule are strongly polarized
by many electronegative groups. Stronger polarization of
hydrogen atoms would lead to stronger adsorption on charged
surfaces of the hydration products of cement. More numer-
ous polarized hydrogen atoms would result in more efficient
coverage of the molecules on the hydration products.
Further, the most efficient retarders have several,
and usually many, oxygens closely grouped and so constrained
either by frequency along the chain or rigidity of a ring
structure. These oxygens can be in hydroxy 1, carbonyl,
or carboxyl groups. These groups can participate in bond-
ing to surfaces of the hydration products by hydrogen bond-
ing to surface oxygens in the solid hydrate or by ionic
attraction to calcium sorbed on the surfaces. A close
grouping as well as a large number of such groups in the
141
retarder molecule would presumably lead to stronger bond-
ing of the molecule to the solid.
The questions still remains of the mode of action of
retarder s. Assuming, as seems likely, that they are sorbed
out of solution onto the initially-formed C-S-H gel, how
do they thereby delay the hydration and set?
In view of the probability of less than a monolayer
being formed on the solid hydrate surfaces, it seems un-
likely that the sorbed molecules act as a physical barrier
to the passage of water inward to the unreacted minerals.
There could, of course, be some kind of partial "plugging"
action that could retard the passage of some water,
especially since the good retarders have a highly oxygen-
ated structure that could exert some binding influence on
adjacent water molecules by hydrogen bonding and thus
generate an effectively larger structure with a better
blocking action.
Another possibility may be more likely. At least it
satisfys the known facts. As previously mentioned, if the
dormant period is caused by a "coating" of C-S-H gel on
the silicate surfaces (the aluminates being controlled by
the gypsum) , then this coating must partially break up
for increased reaction to occur and bring about the onset
of the second heat peak. Whatever the force impelling the
breakup of the coating, if the coating is mechanically
stronger it would resist the breakup longer, and when it
142
occurred it would be less extensive and result in a "milder'
second heat peak. Such a result would be consistent with
the observed calorimetric data. So it is postulated that
the retarder molecules sorb on the initially formed hydrate
material and that some of the molecules will sorb in areas
where the gel particles are already in close contact and
are coherent. The molecular structure of good retarder s is
such that they have regions (the highly oxygenated or
charged parts of the molecule) that can sorb over a large
part of the molecule and that presumably do so relatively
strongly. If they sorbed onto two adjacent gel particles
the resulting bonding would, according to elementary
calculations, result in a considerable increase in the
resistance to rupture by whatever forces ultimately break
up the coating. No doubt many of the molecules would sorb
on only the surface of one particle of gel, but some of
the pore spaces are so small, even in a mature gel, that
it is reasonable to envisage such a bridging action of the
retarder molecule and a consequent increase in the mechani-
cal strength of the coating. It is recognized that this is
only a postulate, which requires much more work to estab-
lish or refute, but at least it fits most of the known
facts.
From the discussion of the retardation mechanisms
above, it appears that strong retarders not only accelerate
the initial hydration of cement by removing possibly more
143
calcium ions or other metal ions from the aqueous phase of
cement paste, but also have molecular configurations that
can either cover or bridge the hydration products on cement
grains more efficiently. Even if a chemical has an ap-
parently effective molecular structure, it may not act as
strongly as expected unless it can accelerate the initial
cement hydration and unless the molecular adsorption on
the hydration products on cement grains proceeds effici-
ently. Examples may be mucic, dihydroxy tartaric or
dihydroxymaleic, and ketomalonic acids that are consider-
ably weaker retarders than gluconic, tartaric, and tar-
tronic acids, correspondingly, regardless of their similar
molecular structures. These are the chemicals that seem
to have precipitated when only little CaCl- was added in
alkaline solution (Figure 51) . Either only a little complex
formation or less efficient action for adsorption would be
expected for these chemicals.
Effect of Retarders on the Shrinkage of Cement Pastes
As shown in Figures 16 and 17, the addition of retar-
ders altered the shrinkage of cement paste significantly at
earlier ages and increased the shrinkage to some extent at
later ages, when the samples were dried in the vacuum oven
for 2 4 hours. However, the general trend of the relation
between degree of cement hydration and shrinkage was the
same for all the samples. An early peak of shrinkage
occurred at about 2 5% hydration of the cement, and then the
144
shrinkage decreased to a minimum of 0.7-0.8% at about 60%
hydration. Thereafter the shrinkage increased parabolically
as the paste became more mature.
It would be normally expected that a longer period of
cement hydration would induce larger shrinkage of cement
paste, because larger amounts of C-S-H gel, which is the
component causing the shrinkage, are produced. The in-
creased shrinkage can be also explained by the reduction of
the amount of unhydrated cement grains, which act as a
"microaggregate" and restrain the shrinkage of the cement
gel (60) . The latter idea is similar to the shrinkage-
restraining action of aggregates in concrete (61) .
However, these ideas do not explain the existence of
the first peak observed in most of the shrinkage curves.
Part of the effect could be due to the slightly greater
"efficiency" with which the young and immature pastes were
dried by being in the vacuum oven for a constant time of
24 hours. The pastes continued to loose a little water and
to shrink appreciably when they were kept in the oven for
longer times; this effect was greater for the more mature
pastes (Figures 52 and 53) . Also, the less mature pastes
probably had a considerably lower elastic modulus than the
more mature ones, an effect that would result in greater
deformations for a given driving force of shrinkage. It
should be noted here that the data points were closely
grouped about the curves and that the effects being
145
in
4->
o
in
1
l
i
1
o
+)
o
6
X
1
1
1
1
i
A
in
o
U
>u
0
0)
c
i\
1
CO
■H
M
-G
CO
i\
o
G
0
1 1
>
_
_
in
i
_
C
T •
■H
I ',
m 9
—
>1
w.
° 1
■o |
-
o
m 3
a
i
\ \ 1
CO
c
1 II
<u
01
>
E
o
\
1 > 1
£
§
—
\
• -
i
6
\ * ii 1
» C "v. 1
c
CVJ >»
l_
3
o
Id
>
5i
» o *
« £
i. «. >
Q
(fl
0
c
' T ^ o^ v
»' \ * 4.
—
-U
A \ O
G
0
u
(0 | \ »
4-1
o \ — \ o
in
0
4J
o
*1
4-i
4-1
\ \
W
\ l
CM
1
\
EN
ID
J_,
i
1
i
1
0)
u
Cn
)
cvj cp *; o
Pn
— -
O
O
ci
(%) aODMUUMS
146
(%) ss<n igOia^
147
described are real, rather than the result of experimental
variability.
According to Powers (62) , menisci in which hydrostatic
tension develops can not form at humidities below about
45%, and all the evaporable water in cement paste at 45%
R.H. exists as adsorbed water on the solid phases. In
explaining the shrinkage of cement pastes at humidities
below the vapor pressure at which menisci vanish, Mills
(63) ascribed the shrinkage to the reduction of swelling
pressure in gel water which is adsorbed on the cement gel.
Verbeck and Helmuth (35) , on the other hand, said that
water held by cement paste below 30% R.H. is not "adsorbed
water" but rather is in a state similar to that of inter-
layer water in clays. So there is some question about the
exact state of the water in pastes at relatively low
humidities. However, it is known that only those water
molecules that are held tightly by the solid phase are left
when the paste is severely dried. Therefore, the measured
shrinkage should be directly related to the properties of
the solid phases, and especially of the hydration products.
Since the amount of unhydrated cement is constant at a
certain degree of hydration, it appears that the observed
difference in shrinkage due to the addition of retarders
has to be attributed to some modification of the properties
of the hydration products.
148
According to other workers, retarders do not alter the
composition or identity of the hydration products, but
change only the rate of cement hydration (1, 8). The
scanning electron microscope did not show any noticeable
difference among the hydration products of cement pastes
with and without retarders after more than 2 0% of the cement
hydrated. Therefore, the only conceivable modifications by
retarders would be either some morphological changes in the
microstructure of the hydration products or a difference in
the relative amounts of hydration products produced from
each compound of cement, or both. Whatever the reason is,
specific surface area measurement on the hydration products
should reveal some of the evidence, because the shrinkage
would have been caused by removal of water molecules from
their surfaces.
As shown in Figures 28, 29 , and 30, the retarders, in
fact, changed the specific surface areas of the hydration
products, and the change was frequently closely related to
the shrinkage behavior of corresponding cement paste. For
example, those retarders which induced the larger shrinkage
at earlier ages increased the specific surface area at the
ages and the subsequent change was also similar to that in
shrinkage. The consistently larger shrinkage of the cement
paste with calcium chloride would be well explained by the
consistently larger specific surface area of the hydration
products.
149
It is of interest to note that for the control paste
with no admixture the minimum in the shrinkage curve coin-
cided well with that in the surface area development curve,
at about 60% hydration. The maximum, however, occurred
somewhat earlier for the shrinkage curve, at 25% as opposed
to 45% for the surface area. They would not, of course, be
expected to be exactly similar, because of changing second-
ary affects, such as the elastic modulus.
For the commercial retarders, it may be significant
that at later ages all admixtures gave pastes of slightly
higher shrinkages than the control, whereas all except S
gave slightly smaller areas. For the pure chemicals the
relationship is equally mixed. At intermediate maturities
of the paste, near the minimum shrinkage, many of the pure
compounds resulted in significantly larger specific
surfaces, while causing only a small increase in the
shrinkage.
For the younger pastes, one would expect a closer cor-
respondence between shrinkage and specific surface because
their structures are less influenced by later, secondary,
processes. This seems to be the case for A and S, although
L gave a somewhat higher surface, but slightly smaller
shrinkage. All the pure chemical gave pastes of higher
surface area in the immature pastes, and all also gave
higher shrinkage except 3-hydroxy-2-butanone and hydroquinone,
It may be significant (Figure 25) that hydroquinone seems to
150
have been atypical in that it did not slow down the hydra-
tion rate, as determined by W , but did retard the set.
When the sample bars were demolded, the one containing
hydroquinone seemed to have a hardened core and a still-
soft exterior. The reasons for this anomalous behavior are
not known. The unusual shape of the curve for sucrose
should also be noted.
In all matters involving the specific surface area, it
should be remembered that different batches of samples
hydrate somewhat differently, even though the mixing and
curing conditions were as similar as possible. The data
for the control pastes were obtained from five different
batches, and they show such a variability. Therefore
caution should be observed in drawing conclusions from small
variations in this parameter.
In general, however, at the younger ages the shrinkage
behavior changes in a manner similar to the specific sur-
face, which is logical.
It should be recalled, that the adsorption process of
retarder molecules is not necessarily the same for all the
retarders. Such differences may have caused kinetic changes
in the subsequent hydration of each cement compound in
these samples.
Retarders also increased the shrinkage of mature cement
pastes (10 months old) when the samples were dried at 50%
R.H. (Figure 18 and 19). Menisci can exist in small pores
151
of the pastes in this case, and the drying would be equiva-
lent to a severe condition for normal concrete members in
the field. In interpreting these kinds of shrinkage data,
however, one can not forget that the delayed hydration of
cement due to the addition of retarders normally recovers
in 2 to 3 days, and retarders rather accelerate the sub-
sequent hydration (see Figures 24 and 25) . Then, it would
be apparent from the previous discussion that, if the
comparison is made at an equal age, retarders generally in-
crease the shrinkage of older cement paste. The comparison
at an equal age would have a practical meaning. But the
result may not be necessarily caused by the modified
property of hydration products by retarders but could be
caused simply by the difference in the degree of cement
hydration.
The relationships between shrinkage and weight loss of
the same samples revealed that the relation was more or less
the same as that of the control sample for all of the
samples except for the sample with retarder S. Hence, it
can be said that the observed increase in the shrinkage of
the samples are mostly caused by the difference in the
degree of cement hydration, for drying at 50% R.H.
Although the determination of evaporable water content
of cement paste was made carefully as possible, the
results scattered in a relatively wide range, and no infor-
mation was obtained as to why only retarder S increased
152
the shrinkage. It should be recalled that retarder S in-
creased the specific surface area of the hydration products
remarkably at later stage of the hydration. Presumably,
the effect of this increased specific surface area would
have remained even at the intermediate humidity. This
would, in turn, indicate a possibility that the modified
property of hydration products by retarders may affect the
shrinkage of cement paste in a relatively severe drying
condition.
153
CONCLUSIONS
The following conclusions seem reasonable. They are
based on the materials used and the tests performed.
1. Strong retarders have a molecular composition that
includes many oxygen atoms constrained to approach each
other closely. Hydroxyl, carboxyl, and carbonyl are all
effective, and carbonyl seems especially strong in its
influence. Many a-hydroxy acids are retarders, but the
presence of other groups can exert a strong effect reducing
the retarding action. The oxygen-containing groups can be
presumed to exert polarizing influences that could con-
tribute to relatively strong adsorption onto the solid
surfaces of concern.
2. More weakly electronegative atoms on the molecule
do not have the same effect as oxygens.
3. When cement pastes are strongly dried of all their
evaporable water, they exhibit shrinkage behavior that
shows a minimum at about half hydration and an earlier
maximum. This behavior is also exhibited by the change in
specific surface area of the hydrated cement; the hydrate
gains appreciable surface area between about 60% and 80%
hydration, and thereafter remains relatively unchanged.
154
Some cause and effect relationship exists between the
specific surface area of a paste and its shrinkage poten-
tial, although other factors, such as restraint by un-
hydrated cement, are probably also operative.
4. Retarders cause a moderate increase in the shrink-
age of cement pastes at ages greater than about half hydra-
tion. At lesser ages the shrinkage may be either increased
or decreased.
5. The changes in shrinkage brought about by retarders
can be partially ascribed to those brought about in specific
surface areas of the hydrated cement, but other effects,
probably related to the chemical species present, are also
important.
LIST OF REFERENCES
155
LIST OF REFERENCES
(1) M. E. Prior, and A. B. Adams, "Introduction to Pro-
ducers' Papers on Water -Reducing Admixtures and Set-
Retarding Admixtures for Concrete" ASTM STP No. 2 66,
1960, pp. 171-179.
(2) R. C. Mielenz, "Water-Reducing Admixtures and Set-
Controlling Admixtures for Concrete: Uses; Specifica-
tions; Research Objectives" ASTM STP No. 266, 1960,
pp. 218-233.
(3) G. B. Wallace and E. L. Ore, "Structural and Lean Mass
Concrete as Affected by Water-Reducing, Set-Retarding
Agents" ASTM STP No. 266, pp. 38-94.
(4) B. Tremper and D. L. Spellman, "Shrinkage of Concrete ■
Comparison of Laboratory and Field Performance"
Highway Research Record, Pub. 1067, No. 3, 1963,
pp. 30-61.
(5) T. C. Powers, "Some Physical Aspects of the Hydration
of Portland Cement" Journal of the PCA Research and
Development Laboratories, No. 3, Vol. 1, Jan. 1961,
pp. 47-56.
(6) J. F. Young, "Hydration of Tricalcium Aluminate with
Lignosulfonate Additives" Magazine of Concrete
Research, Vol. 14, No. 42, Nov. 1962, pp. 137-142.
(7) B. Blank, D. R. Rossington, and L. A. Weinland,
"Adsorption of Admixtures on Portland Cement" Journal
of the American Ceramic Society, Vol. 46, No. 8,
Aug. 1963, pp. 395-399.
(8) P. Seligmann and N. R. Greening, "Studies of Early
Hydration Reactions of Portland Cement by X-Ray
Diffraction" Highway Research Record, No. 62, 1964,
pp. 80-105.
156
(9) S. Diamond, "Interactions Between Cement Minerals and
Hydroxycarboxylic-Acid Retarders: I, Apparent Adsorp-
tion of Salicylic Acid on Cement and Hydrated Cement
Compound" Journal of the American Ceramic Society,
Vol. 54, No. 6, June 1971, pp. 273-276.
(10) H. N. Stein, "Influence of Some Additives on the
Hydration Reactions of Portland Cement II. Electro-
lytes" Journal of Applied Chemistry, Vol. II,
Dec. 1961, pp. 482-492.
(11) W. C. Hansen, "Action of Calcium Sulfate and Ad-
mixtures in Portland Cement Paste" ASTM STP No. 266,
1960, pp. 3-25 and 36-37.
(12) R. C. Mielenz and R. B. Peppier, Discussion of the
paper by W. C. Hansen, ASTM STP No. 266, 1960,
pp. 35-36.
(13) K. E. Daugherty and M. J. Kawalewski, Jr., "Effect of
Organic Compounds on the Hydration Reactions of Tri-
calcium Aluminate" Proceedings of the Fifth Inter-
national Symposium on the Chemistry of Cement,
Tokyo, 1968, Vol. IV, pp. 42-51.
(14) S. Diamond, "Interactions Between Cement Minerals and
Hydroxycarboxylic-Acid Retarders: II, Tricalcium
Aluminate-Salicylic Acid Reaction" Journal of the
American Ceramic Society, Vol. 55, No. 4, April 1972,
pp. 177-180.
(15) S. Diamond, "Interactions Between Cement Minerals
and Hydroxycarboxylic-Acid Retarders: III, Infrared
Spectral Identification of the Aluminosalicylate
Complex" Journal of the American Ceramic Society,
Vol. 55, No. 8, Aug. 1972, pp. 405-408.
(16) L. R. Forbrich, "The Effect of Various Reagents on the
Heat Liberation Characteristics of Portland Cement"
Proceedings of the American Concrete Institute,
Vol. 37, Nov. 1940, pp. 161-181.
(17) M. Polivak and A. Klein, "Effect of Water-Reducing
Admixtures and Set-Retarding Admixtures as Influenced
by Portland Cement Composition" ASTM STP No. 266,
1960, pp. 124-139.
(18) G. M. Bruere, "Importance of Mixing Sequence When
Using Set-Retarding Agents with Portland Cement"
Nature, Vol. 199, July 1963, pp. 32-33.
157
(19) V. H. Dodson and E. Farkas, "Delayed Addition of Set
Retarding Admixtures to Portland Cement Concrete"
Proceedings, ASTM 64, 1962.
(2 0) G. L. Kalousek, C. H. Jumper, and J. J. Tregoning,
"Composition and Physical Properties of Aqueous
Extracts From Portland Cement Clinker Paste Containing
Added Materials" Journal of Research of the National
Bureau of Standards, Vol. 30, March 1943, pp. 215-255.
(21) F. C. Tamas, "Acceleration and Retardation of Portland
Cement Hydration by Additives" Highway Research Board,
Special Report 90, 1966, pp. 392-397.
(22) V. S. Ramachandran, "Interaction of Calcium Ligno-
sulfonate with Tricalcium Silicate, Hydrated Tri-
calcium Silicate, and Calcium Hydroxide" Cement and
Concrete Research, Vol. 2, 1972, pp. 179-194.
(23) H. H. Steinour, Discussion of the paper by W. C.
Hansen, ASTM STP No. 266, 1960, pp. 25-33.
(24) W. C. Hansen, "Oil Well Cements" Proceedings of the
Third International Symposium on the Chemistry of
Cement, London, Cement and Concrete Association,
1952, pp. 598-627.
(25) H. H. Steinour, Discussion of the paper by W. C.
Hansen, Proceedings of the Third International
Symposium on the Chemistry of Cement, London, Cement
and Concrete Association, 1952, pp. 627-631.
(26) H. H. Steinour, "The Setting of Portland Cement"
A Review of Theory, Performance and Control, Research
Department Bulletin 98, PCA, 1958.
(27) J. H. Taplin, Discussion of the paper by H. E. Vivian,
Proceedings of the Fourth International Symposium on
Chemistry of Cement, National Bureau of Standards,
Monograph 43-11, 1960, pp. 924-925.
(28) U. Danielson, "Studies of the Effect of Simple Organic
Admixtures on the Properties of Cement Paste" Inter-
national Symposium on Admixtures for Mortar and
Concrete, Burssels, Aug. -Sept., 1967, pp. 59-68.
(29) R. W. Previte, "Some Insight on the Mechanism of
Saccharide Set Retardation of Portland Cement" Cement
and Concrete Research, Vol. 1, 1971, pp. 301-316.
158
(30) G. M. Bruere, "Set-Retarding Effects of Sugars in
Portland Cement Pastes" Nature, Vol. 212, Oct. 1966,
pp. 502-503.
(31) S. Koide, Discussion of the paper by K. E. Daugherty
and M. J. Kawalewski Jr., Proceedings of the Fifth
International Symposium on the Chemistry of Cement,
Tokyo, 1968, Vol. IV, p. 52.
(32) J. F. Young, "A Review of the Mechanisms of Set-
Retardation in Portland Cement Pastes Containing
Organic Admixtures" Cement and Concrete Research,
Vol. 2, 1972, pp. 415-433.
(33) R. C. Mielenz, "Use of Surface-Active Agents in
Concrete" Proceedings of the Fifth International
Symposium on the Chemistry of Cement, Tokyo, 1968,
Vol. IV, pp. 1-29.
(34) A. M. Neville, "Hardened Concrete: Physical and
Mechanical Aspects" American Concrete Institute
Monograph, No. 6, 1971, pp. 119-141.
(35) G. J. Verbeck and Helmuth, "Structures and Physical
Properties of Cement Paste" Proceedings of the Fifth
International Symposium on the Chemistry of Cement,
Tokyo, 1968, Vol. Ill, pp. 1-32.
(36) B. Tremper, Discussion of the paper by G. B. Wallace
and E. L. Ore, ASTM STP No. 266, 1960, pp. 94-96.
(37) W. Lerch, "The Influence of Gypsum on the Hydration
and Properties of Portland Cement Pastes" Research
Department Bulletin 12, PCA, March 1946.
(38) H. E. Schwiete, U. Ludwig, and K.(iSeiler, "Die
Wirkungsweise von Erstarrungsverzogerern unter
besonderer Berucksichtigung silikof luorhaltiger
Verzogerer" Institue fur Gesteinshiittenkunde , 1969.
(39) J. Okabe, K. Nakajima, and T. Yoshihara, " Invert iga-
tions on the Method of Test for Setting Time of
Concrete Containing Water Reducing Admixture"
Proceedings of the Fifth International Symposium on
the Chemistry of Cement, Tokyo, 1968, Vol. IV, pp. 58-64
(40) H. S. Washington, "The Chemical Analysis of Rocks"
John Wiley & Sons, Inc., 1930, pp. 242-245.
(41) P. G. Jeffery, "Chemical Methods of Rock Analysis"
Pergamon Press, 1970, pp. 169-170 (Method of Shapiro
and Brannock) .
159
(42) R. F. Feldman and P. J. Sereda, "A Model for Hydrated
Portland Cement Paste as Deduced from Sorption-Length
Change and Mechanical Properties" Materiaux et
Constructions, Vol. 1, No. 6, 1968, pp. 509-520.
(43) R. F. Feldman, "The Flow of Helium into the Interlayer
Spaces of Hydrated Portland Cement Paste" Cement and
Concrete Research, Vol. 1, No. 3, May 1971, pp. 285-
300.
(44) S. Brunauer, I. Odler, and M. Yudenfreund, "The New
Model of Hardened Portland Cement Paste" Highway
Research Record, No. 328, 1970, pp. 89-107.
(45) R. F. Feldman, "Assessment of Experimental Evidence
for Models of Hydrated Portland Cement" Highway
Research Record, No. 370, 1971, pp. 8-24.
(4 6) S. J. Gregg and K. S. W. Sing, "Adsorption, Surface
Area, and Porosity" Academic Press, 1967, pp. 36-41.
(47) T. C. Powers and T. L. Brownyard, "Studies of the
Physical Properties of Hardened Portland Cement Paste"
Research Department Bulletin 22, PCA, March 1948,
pp. 469-504.
(48) W. J. Halstead and B. Chaiken, "Water-Reducing Re-
tarders for Concrete - Chemical and Spectral Analyses"
Highway Research Board Bulletin 310, 1962, pp. 33-55.
(49) W. E. Grieb, G. Werner, and D. O. Wolf, "Tests of
Retarding Admixtures for Concrete" Highway Research
Board Bulletin 310, 1962, pp. 1-32.
(50) G. Verbeck, "Cement Hydration Reactions at Early
Ages" Research Department Bulletin 189, PAC.
(51) H. E. Schwiete, U. Ludwig, and P. Jager, "Investiga-
tions in the System 3CaO«Al203-CaS04-CaO-H20" Highway
Research Board, Special Report 90, 1966, pp. 353-367.
(52) L. E. Copeland and E. G. Schulz, Discussion to the
paper by Ake Grudemo, Proceedings of the Forth Inter-
national Symposium on Chemistry of Cement, Washington,
1960, Vol. II, pp. 648-655.
(53) H. E. Buckley, "Crystal Growth" John Wiley & Sons
Inc., 1951, p. 571.
160
(54) R. L. Berger and J. D. McGregor, "Influence of Ad-
mixtures on the Morphology of Calcium Hydroxide
Formed During Tricalcium Silicate Hydration" Cement
and Concrete Research, Vol. 2, No. 1, 1972, pp. 43-55.
(55) W. L. Dolch, Private Communication.
(56) L. L. Alt, " Spectrophotometry Determination of Gluco-
nic Acid and Its Salts" Analytical Chemistry, Vol. 27,
No. 5, May 1955, pp. 749-751.
(57) H. N. Stein, "Influence of Some Additives on the
Hydration Reactions of Portland Cement I. Non-Ionic
Organic Additives" Journal of Applied Chemistry,
Vol. 11, Dec. 1961, pp. 474-482.
(58) H. F. W. Taylor, "The Chemistry of Cement", Vol. I,
Academic Press, 1964, pp. 187-189.
(59) K. Fujii and W. Kondo, "Hydration of Tricalcium
Silicate in a Very Early Stage" Proceedings of Fifth
International Symposium on the Chemistry of Cement,
Tokyo, 1968, Vol. II, pp. 362-371.
(60) G. Verbeck, "Shrinkage of Concrete - Applied Research
Studies" PCA Annual Meeting, 1961.
(61) G. Pickett, "Effect of Aggregate on Shrinkage of
Concrete and Hypothesis Concerning Shrinkage"
Proceedings of the American Concrete Institute,
Vol. 52, No. 5, Jan. 1956, pp. 581-590.
(62) T. C. Powers, "Physical Properties of Cement Paste"
Research Department Bulletin 154, PCA.
(63) R. H. Mills, "Effects of Sorbed Water on Dimensions,
Compressive Strength and Swelling Pressure of Hardened
Cement Paste" Highway Research Board, Special Report
90, 1966, pp. 84-111.
APPENDICES
APPENDIX A
161
APPENDIX A
Correction Curve for Warping of Cement Paste Bars
162
M
rt
cq
0)
-p
tn
m
&4
-P
c
i
U
>H
O
en
c
■H
a
S
U
o
a)
>
M
U
c
o
•H
-p
o
Q)
M
M
0
u
(%) n/8
APPENDIX B
163
APPENDIX B
Summary of Setting Time Experiments When
Pure Chemicals were Added
Setting times are expressed in terms of T50, T100,
T500, and T4000 which are the relative retardations at the
penetration resistances of 50, 100, 500 and 4000 psi,
respectively, to the retardation of plain mortar with no
admixture.
164
C
c
0
<u
-H
X!
P
!2
rrj
M
CO
+J
0)
c
rH
0>
CU
o
e
B
rd
o
CO
u
H
rd
-P
u
o
s
m
O
0>
e
•H
Eh
en
C
-rH T3
-P <D
-P TJ
0> TJ
CO <
o> <a
> u
-H 0)
4J S
m
iH CO
0) rH
tf (0
o
m -h
O S
0)
>iXi
M u
03
e oi
I n
3 3
]
r-
0)
rH
XI
rd
E-<
x:
u
r^
(N
U>
m
rH
VD
VO
V£>
m
o
r-i
n
CT>
rH
O
o
CN
o
o
CM
00
o
O
LT)
rH
rH
rH
rH
rH
<-i
rH
<-{
rH
rH
H
rH
co
o
CO
r>
r^
CN
o
o
OJ
o
rH
CN
o
O
o
rH
rH
rH
rH
rH
OJ
<H
H
m r» cr>
(N LT) LT>
r-
Cft
O
O
CTi
o
OH
LO
l£>
O
o
<T>
m
M
en
o
o
rH
o
cm
O
rH
(N
o
o
o
in
CO
o
^r
rH
rH
rH
rH
rH
rH
<-\
rH
CN
rH
rH
rH
rH
CN
CN
vo
o
ro
(N
LD
fN
CO
ro
m
o
rH
CN
en
o
en
in
-U
C
0!
U3
6
JD
-H
H
Vj
0)
a
x
Ci
in
<D
io
Xi
>H
-p
4-1
0
P
o
■H
~T
g
H
-H
rH
<D
x;
■P
0)
c
•H
B
rH
0)
p
0)
"O
O
P
|
0)
rH
^— ^
TJ
•H
rd
01
■H
T3
rH
■p
U
O
tJ
<d
<
Ul
•H
O
M
•H
t;
O
o
T3
TJ
o
Q
H
<r;
p
-H
>i
•rl
— -
O
O
SC
rH
<
U
en
rtfi
"*— '
-P
T3
•rl
T,
u
•H
H
•rH
O
T3
0
U
T3
TJ
•H
•H
rd
c
Tj
T3
TJ
-H
•H
m
-<
•H
■H
O
0)
-P
o
T)
•rH
-H
■H
-P
O
TJ
i
o
o
<
rH
J-i
•H
•H
o
9
T>
o
0)
<
•H
TJ
2
O
<
<
rd
rd
-P
U
«!
ri:
-H
<
o
O
■H
I
•H
U
e
+J
rd
<
O
rd
U
<
CJ
>i
c
u
u
•H
>i
>i
13
o
o
<
u
0
■H
<
X
0
•H
-H
M
X
X
rH
o
-H
•r)
■H
0>
-P
rH
CJ
0
rH
c
rH
rrj
o
0
rd
•H
a
H
O
■-«
c
&
>1
•H
o
rH
rd
0
2
-P
rH
H
-P
c
-H
m
■H
O
•H
fd
X
>
•H
-d
e
H
U
T3
TJ
0)
0
O
-p
0.
o
o
o
o
3
-P
>1
0
p
rd
>1
>l
«
rH
O
3
-rl
>l
>1
M
>1
M
O
ac
■p
M
— ^
M
x:
x:
2
3
rH
^
rH
H
0
rH
>i
rd
i
d)
rd
1
1
•H
•H
CO
O
o
U
s
O
04
a
a
«
Eh
^^
T>
Q
Q
Ht
165
I
a
■H
+1
+J
<D
CO
o
o
CN
r^
o
(N
<Ti
^
•-i
m
in
O
in
o
en
ro
00
o
o
EH
i-l
(N
n
CM
iH
CN
rH
rH
id
u
•H
e
CD
cn m en o
O O O rH
rH rH rH rH
•H
O
<
U
-H
c
o
•H
0i
0
M
0. TJ
0
>i H
•H
O
•H
rH cr>
vo ro
(N rH
H3
•H
U
CJ
■H
M
nJ
-P
rH
a
o
n
O I
-P ro
0)
« <"M
I
(N rH
T3
•H
<
O
•H
rH
>1
J8
>H
Id
O
•H
H
-p
<D
C
id TJ
o
0>
-P
id
rH
o
0
>1
rH
a
o
I
o
■*r
U3
(N
00
o
o
o
O
O
O
CN
cm
rH
rH
rH
f-i
r-i
o
0) n
■H I
6 N
m
r~
CD
r-~
o
o
CO
ci
o
00
CN
o
O
o
o
o
o
O
O
CD
r-i
r-i
rH
rH
rH
rH
<-t
rH
ro
(1)
c
g
id
a a)
>1
CQ
•H
X
1
cu
o
IN
o
n
1
M
-a
>1
0,
>i
X
rH
A
0
>1
•H
Q
£
A
1
>1
01
ro
1
|
rH
C>
z;
■p
ro
-p
CD
o
id
o
-p
0)
o
rH
>1
A
-P
w
o
■H
Q
i
«*
rH
I
01
c
0)
■p
CQ
I
cm rsi
o
o
T3 -H
H
A
-p
>i
M
01
r<5
+J
G
0)
en
00 00 00
166
o
o
r-
00
r~
CTl
CM
CO
CM
CM
n
rH
r^
00
r-
Oi
rH
r-
in
CM
CM
o
o
O
o
<N
o
o
O
o
o
rH
n
CO
LO
EH
H
rH
iH
iH
rH
rH
rH
r-i
rH
rH
■H
rH
rH
CM
rH
CM
o
o
^"
VO
00
o>
CM
CM
co
00
CO
m
CM
LO
rH
CM
CT>
o
o
O
O
CM
o\
o
o
o
o
Id
Eh
O O
CO CM
O CM
H rH
00 00
rH 00
o
rH
in
a>
en
rH
CO
r-
00
in
T
O
CO
m
iH
rH
o
o
O
O
CM
en
o
o
o
o
Eh
rH
CM
rH
tH
rH
rH
rH
rH
rH
rH
rH
CM <T> <T>
CO ^ CO
<7>
[—
\o
rH
O
ro
CM
CM
rH
-0
•H
3
o
•H
TJ
'd
d)
0
T3
•H
•H
G
N
•H
O
U
o
G
u
<
<<
G
d)
d)
XI
<
O
u
T>
x:
>i
u
•H
■H
•H
CU
X
■H
0
T3
0
rH
O
o
rH
o
p
rH
rH
rH
N
•H
N
0
<
p
0
u
d)
d>
0
0
0
G
O
G
TJ
X!
d)
T5
G
T3
u
G
G
G
G
d)
<
<U
■H
0
o
o
•H
■H
n
>i
IT)
O
d>
d)
0)
X)
XI
U
O
•H
rd
u
rH
U
•H
.g
■H
g
£
XJ
X!
>1
U
fr
<c
rH
rH
>i
<
O
3
o
•H
rH
rH
-H
(X
Gi
CU
X
•H
<
d)
X
rH
H
<
M
4-1
0
3
O
0
0
0
rH
o
0
T)
o
0
rH
CT>
Eh
o
X!
0»
M
rH
u
U
-H
V4
•H
rH
G
M
•H
(0
0
o
1
rH
U
o
-P
■P
p
n
O
TJ
0
>1
«J
13
rH
tn
!h
•H
U3
■H
<D
W
•H
•H
•H
>i
>i
>i
N
N
S
>i
+J
0
0
H
».
<
U
-P
tJ
2
Z
2
ac
rH
m
G
S
I
IB
u
M
rH
rH
T
Eh
(Tj
>i
1
1
1
1
(0
i
0)
rJ
1
-H
>i
x\
flj
•.
Q
H
U
03
s
0
Cu
s
CO
04
«
m
Q
O
£U
&
CU
O
CM
W
2
167
en
a
-H
+>
+J
<U
M
Ol
"*
(N
Ol
rH
>X>
<Ti
Ol
H
<7\
O
CM
r-l
rH
rH
o
VD
O
(N
n
O
Ol
rH
O
o
rH
•-t
rH
rH
O
n
rH
o
•V
in
o
en
(N
o
o
(N
rH
rH
rH
>-{
rH
r~
CM
CO
-*
m
00
en
CT>
H
O
o
CM
>-t
rH
rH
rH
-p
nj
c
EH
o
4H
X.
H
.. — *
>~N
^^
.*— N
O
3
CM
CO
CO
r^
«J
(0
co
in
rH
o
rH
0
CM
co
m
t*>
03
c
*
>-^
*~^
*~*
QJ
•H
LD
o
o
in
g
J
-d
a\
CO
H
m
O
a>
<-i
ro
rf
r»
H
£
3
r»
in
■^
CM
x:
3
fl
OJ
CM
CM
"tf
o
0
■H
O
•H
-P
QJ
<D
0)
<u
•H
rH
C
>i
>i
>>
>i
H
nJ
0
a
Q
a
Q
w
U
o
168
CD
e
■H
H
Di
C
-H
P
-H
0>
o
o
o
n
m
<T>
CM
M
O
tn
(N
CM
r»
00
o
■*T
1—
v£>
ro
•«»
^<
n
<n
*
o»
CM
rH
r-i
<J*
a\
■K
*
CT*
in
r~
O
O
rH
r-
H
.H
iH
r-i
rH
iH
rH
r-\
r-i
rH
rH
r-i
rH
o
o
in
r~
o
V£>
00
00
00
CN
ro
in
r-«
ro
r-
(N
T
in
•«*
n
o
*
Cft
CN
o
rH
as
<j\
r-i
*
CTi
r-
r^
o
o
EH
■H
rH
rH
rH
rH
r-t
m
r-i
rH
r-i
r-i
o
10
00
^o
ro
<N
■^
r»
ro
CTl
00
*r
<o
rr
CM
<T>
r-
o
*tf
cn
ro
rH
(N
CA
<j\
(N
<*
«*
00
r-
o
O
O
r^
rH
■>*
r-i
rH
r-i
ro
ro
r-i
r-t
rH
rH
rH
rH
CM vo
VO 00
CN (N
v
c
o
c
rd
T3
0.
<L>
-d
-H
0
C
•H
O
0)
M
O
Q)
o
<
tJ
cu
fi
P
rH
rf!
O
p
■H
1
CN
rd
P
rd
-P
0
(U
o
•H
(0
ro
1
3
0)
Q
4J
H3
TJ
T>
•H
P
rH
P
>l
m
U
1
rd
T)
•H
•H
•rl
-P
0)
0
0)
X
i
rd
"5T
lH
n
•H
O
O
o
0)
o
•a
o
0
0
CN
0
>.
TJ
•rt
U
<
<
<
o
id
•H
>1
ffl
n
1
-P
r-i
>i
O
<
rd
M
o
Q)
r-i
r-i
>1
TJ
>1
<U
1
K
<
U
U
o
o
O
<
C
0
O
8
>1
X
o
OJ
u
•H
■H
•H
0)
p
rH
•H
T3
Xi
0
<:
c
rH
o
•H
C!
M
H
fl
a
Xj
o
M
r-i
r-i
M
-H
u
0)
rd
-H
C
•H
rB
0
■H
ft]
O
•rl
a>
rd
>1
TJ
Q
TJ
H
p
u
M
0
O
■P
O
O
o
0
P
o
-P
J5
>1
1
>i
>i
3
0
P
<
r-i
O
3
>i
>i
M
C
o
>i
(1)
-P
EC
ro
33
£1
m
r-i
O
EH
2
3
r-i
rH
rH
(1)
O
ro
r-i
O
d)
1
*
1
+J
i
Xi
•H
Q
t/>
O
o
O
a
s
►H-
O
<
S
a
r-i
ro
W
CM
u
(U
W
APPENDIX C
169
APPENDIX C
Shrinkage of Vacuum Oven-Dried Cement Pastes
at Various Stages of Cement Hydration
The following abbreviations are used in this part.
h: hour(s)
d: day(s)
m: month (s)
Each data point is the average of two separate measurements,
The range of duplicate measurements was usually about 1% of
the average.
170
o
E
m
10
H
J CVJ
.X)
o
H
,x:
CVJ
m
o
CD
-8 3
o
a*
£
O
c
4
m
CM
JZ
— IO
OS
o
CVJ
to
<u
en
to
-p
w
to
O
-H
M
>
P
to
ai
-p
to
td
cu
•p
c
■o
>»
I
«*- '
1
a o
>~
O
e c
3 o
3 -h
o -P
> W
•o
«W >i
O 35
Q) -P
cn G
a) a)
M E
c <D
H u
in
CN
a)
d
d
o
CVJ
d
(%) dOo^uuns
tn
-H
171
CM
00 CD
d d
(%) dfiD^UULIS
0)
(U
CJ>
nj
-P
3
o
■H
>
+J
HJ
U)
0)
+i
nj
o<
-p
c
0)
e
Q) ^
u ^
oj a>
a us
i -u
c a)
> ~
o
9 o
3 -H
O -P
> n
TJ
>4-4 >1
O tc
<D +J
tJ> C
fO a)
a: e
c 3
-H u
x: m-i
w o
Q)
u
0
-H
8
172
o
o
0)
tp
0)
3
o
o
CD
• M
IT)
>
m
en
CI
-p
in
03
Oh
O
(0
8
-P
C
0)
6
c
o
<D —
U <
■»-
e
0) dJ
u
I
•H T3
Q nl
1 -M
H-
o
C 0)
o
Si
O
<*
l_
o>
e c
O)
3 0
u
3 H
O 4J
rfl rfl
> H
4-1 >1
O EC
<D -P
nj 0)
o
C 0)
(VJ
-H U
f-0 0
C\J
00
d
<0
d
d
CM
d
(%) afiDHuuLis
173
m
4)
Cn
rfl
+J
W
w
O
u
>
m
U)
Q)
+J
to
A,
+J
C
<D
e
0) —
U CO
Ti U
(U Q)
l-l ^
Q id
i -u
c <u
<u «
> —
e c
o o
O -P
(0 (Tl
> M
^H >i
O K
QJ -P
rtj 0)
•H U
LO o
i
(%) eDDHUUMs
o
o
174
e
T3
CD
C\J
TJ
O
o
oo
CVJ
«fr
.c
<*
CO
if)
"O
I
Q
8
U]
0J
fl
-P
U]
=1
0
■H
M
d
>
nj
en
a)
+j
en
it)
c —
QJ TJ
0) u
u <
tJ o
M o
Q O
I >i
C rH
0) u
> —
o
e c
^ o
3 -H
U -P
rtf rtf
> n
TJ
M-l >i
o ac
0> +J
Cn C
aj <U
m e
C QJ
•H U
O) o
I
CVJ
CO
d
6
o
6
(%) a6D))uuLjs
175
c\j
00 (£>
d d
o
CM
d
(%) aBoHuugs
CO
CD
tr>
<d
-P
co
in
o
■H
>
in
0)
+J
en
10
Cu
■M
C
0)
e
CD
u
t> —
CD CD
•H CO
M O
Q U
I O
C 3
<D c/i
> —
O
3 o
3 H
O 4J
> H
T3
4-1 >t
O X
CD -M
en c
m cu
x 6
c 3
•H U
.C <4-l
co o
I
o
D
)-l
cn
•H
176
o
o
01
0)
en
TO
-P
w
V)
3
o
0
CD
-H
M
>
-P
rj
01
<u ^
-P Q)
U) C
(0 o
ft G
O
. — ,
-P -P
(0
i}8
G 3
g i
c
Q) <N
o
O 1
*-
>1
p
13 X
I
<1J O
-H U
U t3
Q >i
1 33
o
G 1
(1) rn
> "— '
O
0)
O
tf
c»
6 G
0i
3 o
Si
Q
3 H
U -P
«J (fl
> U
>W >i
0 33
<U -P
D> G
flj <D
* 6
C (D
-H U
,g 4-1
w o
(%) aGo><uui|s
177
■o
00
CO
« TJ
O
*— ?
V.
c
&
ro
£
m
OJ
x.
r-
ro
c
g
E
TO
I
2 ?
OJ
00 <0 *
6 O <5
CM
O
(%) dBD^uugs
0)
(1)
tr
n3
+1
W
[fl
o
-H
>
-P
w
0)
4J
CO
m
c
T3
<1)
■H
M
Q
I
a
a> a:
o
e c
3 o
- o -P
> u
TJ
O X
,* e
c a)
•h u
w
en o
i
ai
3
Co
178
w
<D
Cr>
«J
to
01
o
-H
M
«
>
-P
id
y>
0)
-p
en
m
cu
c
OJ
g
u o
CM
Q) <N
•H •
H <N
Q rH
I u
C <Tj
<D CJ
> —
O
e c
3 o
3 -H
U -M
nj «J
> U
•n
<W >i
O SB
<D -M
* e
C <D
•H U
M
XI m
to o
i
(%) aBojuuMs
179
x O
x
%
E
m
E
ro
00
CM
X5
O
X)
8
c
c
o
o
o
a
T»
■o
l_
i_
o
o
4-
•♦-
fl»
a>
a:
rr
a>
a»
>
>
♦-
■»—
o
o
a>
a*
C£
cr
o
in
<J*
o
o
CM
o
X
to
?
CM
w ^-,
CD TD
Cn-H
fl u
P<
w
o
0) -H
3 U
O -P
-H -H
u u
flj
> 4-1
o
+J
(0 in
C
01 o
<D -m
-p -P
Ul ra
(0 M
04 P
J*
P <u
•***s
c u
c
o
<d c
g 0
CD U
D
u
t.
U-l
"D
t> o
>.
a)
X
•H -P
n o
*-
a a)
O
i u-i
L.
c ^
CD w
o»
o
01
a
6. c
3 o
3 -H
O P
m m
> n
TJ
<« >1
O K
CD P
Cn C
(0 CD
M 6
C CD
H U
5 >«
c/l 0
CM
CD
O
CD
C>
O
CM
O
CD
U
•H
(%) a6o>|UjJi|s
8
180
9
ro
o
O
II
n
*e
$
*
I
X
tn
(D
Cn
m
-u
en
U)
O
o
o
CD
'"5
•H
Wl —
(TJ 0
> H
-P
-P «j
rt «
0) -P
Q) C
-P 0)
«J S
ITJ 3
Cn U
1
O
g
-P >-l
CO
C <D
fl) -P
S
•»-
o
H
<W
■D
T3 O
>.
0)
X
■H -P
1
01
Q
8
C\J
CO
d
CD
d
d
CVJ
d
H o
Q 0)
I M-l
C <w
> —
o
e c
3 o
3 H
O -P
as m
> i-i
o w
a; -p
tp c
(0 Q)
-* E
g a)
-H u
u
.c <«
CO o
i
LO
Q)
U
Di
•H
(%) aBD^uugs
VITA
181
VITA
Yasuhiko Yamamoto was born on May 1, 1943 in Korea.
He is a citizen of Japan.
After completing early education at Kofu First High
School in Yamanashi prefecture, he entered Yamanashi
University where he received B.S.C.E. degree in March 1966
In April 1966, he became a graduate student of
University of Tokyo and received M.S.C.E. degree in March
1968. He continued his study at the same university
toward Doctor of Science degree until he came to the
U.S.A. to study at Purdue University in September 1969.