Skip to main content

Full text of "Principles & Techniques of Electron Microscopy volume 2"

See other formats


m — 



of Electron 

Biological Applications 
Volume 2 

Edteih M.A. Hayat 

Van Nostrand Reinhold Company 

Principles and 

Techniques of 

Electron Microscopy 

Biological Applications 
Volume 2 

Edited by 


Associate Professor of Biology 
Newark State College 
Union, New Jersey 


' New York/Cincinnati/Toronto/London/Melbourne 

Van Nostrand Reinhold Company Regional Offices: 
New York Cincinnati Chicago Millbrae Dallas 

Van Nostrand Reinhold Company International Offices: 
London Toronto Melbourne 

Copyright © 1972 by Litton Educational Publishing, Inc. 

Library of Congress Catalog Card Number: 70-129544 
ISBN: 0-442-25670-1 

All rights reserved. No part of this work covered by the copyright hereon may 
be reproduced or used in any form or by any means— graphic, electronic, or 
mechanical, including photocopying, recording, taping, or information storage 
and retrieval systems— without written permission of the publisher. 

Manufactured in the United States of America 

Published by Van Nostrand Reinhold Company 
450 West 33rd Street, New York, N.Y. 10001 

Published simultaneously in Canada by Van Nostrand Reinhold Ltd. 

15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 

Library of Congress Cataloging in Publication Data 

Hayat, M Arif, 1936- 

Principles and techniques of electron microscopy. 

Includes bibliographies. 

CONTENTS: v. 1-2. Biological applications. 

1. Electron microscope— Collected works. 
I. Title. 

QH212.E4H38 578'.4 70-129544 

ISBN 0-442-25670-1 (v. 2) 

It is a pleasure to 
dedicate this volume to: 

W. Bernhard, D. E. Bradley, L. G. Caro, Humberto 
Ferndndez-Mordn, Isidore Gersh, Cecil E. Hall, 
R. W. Home, Elizabeth Leduc, J. Liquier-Melward, 
Hans Moor, Daniel C. Pease, Fritiof S. Sjostrand, 
and Robert P. Van Tubergen. 


In this volume is presented a compilation of methods employed in the 
study of the structure, composition and location of cellular components. 
The primary aim of this book is to provide the reader with a compre- 
hensive discussion on the theory, principles and application of special 
techniques employed in electron microscopy (including high resolution). 
Since the information retrieved from an electron micrograph is largely 
dependent upon the preparatory procedures employed, a thorough 
understanding of the fundamentals governing these procedures is impera- 
tive, Also, the book provides a detailed description of the methods for 
the preparation of specimens. Limitations of the methods and potential 
research areas have been pointed out. For the convenience of the 
reader, references with complete titles are provided. 

In order to accomplish the above mentioned goals, it has been im- 
perative to enlist the aid of nine scientists, each a specialist on the par- 
ticular subject which he has contributed. As a consequence of this 
approach, an authoritative description of the most modern and reliable 
methods is compiled in this volume. I have been, indeed, fortunate in 
obtaining contributions from leading scientists in various parts of the 
world. The high quality of this volume attests to their efforts. It is with 
great pleasure and pride that I present to the readers these valuable con- 
tributions by outstanding scientists. 

The need for a book of this kind is well certified by the enthusiastic 
response of the scientists who were approached for a contribution. The 
importance of this up-to-date work becomes apparent when one con- 
siders that electron microscopy is not only advancing rapidly, but is also 
being increasingly employed as an aid to many diverse investigations in 
the biological and medical sciences. 

The preparation of a third volume under the same main title is in 
progress, and will contain a detailed discussion on the Electron Micro- 
scope, Electron Microscopy of Selectively Stained Molecules, High Reso- 
lution Dark-Field Electron Microscopy, In-Focus Phase Contrast Electron 
Microscopy, Ultracentrifugation, Morphometry, and Critical-Point Dry- 
ing. The first volume of the series presents a comprehensive discussion on 
Fixation, Embedding, Sectioning, Staining and Support Films. 

My first acknowledgement is to the able contributors from several 
countries who so generously expended their time and energy in con- 
tributing to this volume. During the preparation of this work, I have 


come to admire and respect highly the achievements of these contrib- 
utors who have reached eminence in their respective areas of specializa- 

I owe special thanks to Drs. R. W. Home and J. Jacobs for their 
help in the preparation of this volume. With pleasure I acknowledge 
the cooperation shown by Mr. George Narita, Editor, Van Nostrand 
Reinhold Company. For the patience and accuracy with which the 
manuscript was typed, I am grateful to Miss Madalin Petruzzelli, Miss 
Debie Long and Miss Christine Droppa. 

M. A. Hayat 
Union, New Jersey 





Lionel I. Rebhun 

Introduction 3 

General Principles 5 

Structure of Water and Ice in Cells 5 

Freezing and Viability 9 
Morphological Effects of Pretreatment Regimes on Living Cells 12 

Recognition and Avoidance of Intracellular Ice Crystals 13 

Dehydration of Frozen Cells 19 

Freezing Technology 23 

General Principles of Freezing in Boiling Fluids 23 

Application of Liquid Nitrogen and Fluoro- and Hydrocarbon 

Quenching Fluids 27 

Description of Freeze-Substitution Technique 32 

Quenching Fluid 33 

Collecting Cylinder 33 

Specimen Preparation 33 

Specimen Freezing 35 

Removal of Quenched Specimens 35 

Transfer into Freezer 35 

Substitution Fluid 35 

Fixation 36 

Final Substitution 36 

Embedding 36 

Description of Freeze-Drying Technique 36 

Specimen Freezing 38 

Preparation of Drying Chamber 38 

Transfer of Frozen Tissue to Drying Chamber 39 

Fixation 39 

Embedding 40 

Freeze-Thawing 40 


Appendix 41 

Obtaining Cold Quenching Fluid 41 

References 42 


James K. Koehler 

Introduction 53 

Scope of the Work 53 

Historical Background 54 

Freeze-Etching Methodology 55 

General Considerations 55 

Specific Methodology 56 

Pretreatment of the Specimen 56 

Freezing the Specimen 58 

Fracturing 59 

Etching (Ice Sublimation) 61 

Replication 63 

Cleaning the Replica 66 

Instrumentation 68 

Microtome-Type Freeze-Etchers 68 

Steere Freeze-Etching Module 69 

Self -Contained "Block-Type" Devices 70 

Ultrahigh Vacuum Devices 72 

"Double-Replica" Preparations 76 

"Broken Capillary Tube" Method 76 

Steere's Device 79 

Bullivant's Device 80 

Intepretation of Freeze-Etching Images 83 

Artifacts 83 

Interpretation of Cellular Components 88 

General Considerations 88 

Organelles 90 

Membrane Structure Considerations 92 

Membrane Particles 94 

References 95 



Rudy H. Haschemeyer and Robert J. Meyers 

Introduction 101 

Equipment Requirements 110 

Negative Staining 


Support Films 

Preparation of Negative Stains 

Embedding the Particles in Negative Stain 

Drop Method 

Spray Method 

Float Method 




Summary Comments on Negative Staining Procedures 


Photography of Negatively Stained Specimens 


Dark Room Technique 


Particle Dimensions and Molecular Weights 

Magnification Calibration 
Particle Dimensions 


Molecular Weight Determinations 



Identification of Specimen Contrast 


Resolution Requirements 
Distribution of Stain Around Particles 


Beam Damage and Contamination 


Anticipated Appearance of Idealized Images 
Optical Aids for Enhancing Periodic Detail 







W. J. Henderson and K. Griffiths 



Shadow Casting 


Electron Beam Evaporation 



Some Applications of the Shadow Casting Technique 163 

Replication Technique 164 

References 1 91 


R. Abermann, M. M. Salpeter, and L. Bachmann 

Principles 197 

Thin Film Formation 198 

Evaporation Methods 198 

Resistance Heating J 99 

Electron Beam Evaporation 200 

High Resolution Shadowing Materials 202 

Platinum-Carbon (Pt-C) 202 

High Melting Metals 203 

Artifacts 205 

Mounting of Specimens 205 

Specimen Contamination within the Evaporator 207 

Thermal Load 209 

Heat of Condensation 210 

Heat of Radiation 210 

Ion or Electron Bombardment 211 

The Problem of Resolution 211 

References 215 


M. M. Salpeter and L. Bachmann 

Introduction 221 

Conditions for Quantitation 222 

Resolution 223 

General Considerations 223 

Resolution Determination 226 

Sensitivity 236 

General Considerations 236 

Measuring Sensitivity 240 

Section Thickness 241 

Emulsion Thickness 243 


Analysis of Autoradiograms 245 

Conclusion 254 

Technical Considerations 255 

Specimen Preparation 255 

Preparing Collodion-Coated Slides 257 

Section Mounting 258 

Measuring Section Thickness 259 

Staining 259 

Intermediate Layer 260 

Emulsion Coating 262 

Ilford L4 263 

Kodak NTE 264 

Gevaert Nuclear 307 266 

Processing Autoradiograms 266 

Storage 266 

Development 267 

Developers for Ilford L4 267 

Developers for Kodak NTE 269 

Final Steps in Photographic Processing 269 

Stripping and Mounting Specimens 272 

Statistical Accuracy of Data 272 

References 274 

Index 279 

Contents of 

Volume 1 

1. Fixation 

2. Embedding 

3. Sectioning 

4. Staining 

5. Support Films 

6. Appendix 

Volume 3 

1. The Electron Microscope 

2. Electron Microscopy of Selectively Stained Molecules 

3. High Resolution Dark-Feld Electron Microscopy 

4. In-Focus Phase Contrast Electron Microscopy 

5. Electron Microscopic Evaluation of Subcellular Fractions 
Obtained by Ultracentrifugation 

6. Steriological Techniques for Electron Microscopic 

7. Critical-Point Drying Method 

Principles and Techniques of 
Electron Microscopy 

1 Freeze-Substitution 
and Freeze-Drying 

Lionel I. Rebhun 

Department of Biology, University of Virginia, 
Charlottesville, Virginia 


I he removal of a cell from its natural environment for the purposes of 
examination, e.g., in tissue culture studies, necessarily introduces an 
uncertainty into considerations of the nature of the cell prior to our 
intervention. If our methods of examination impose further modifications 
in the form of fixation or, in biochemical studies, homogenization, then 
we are clearly far from the initial state of the system and there is no 
guarantee that we shall be able to reconstruct it. We have necessarily 
introduced artifacts which are, however, absolutely necessary for any 
further process of analysis.* The sine qua non of whether an artifact is 
"good" or "bad" is whether we can understand the process by which it 
arises and whether the information obtained can be used in recon- 
structing the original system. It does little good to have pictures of great 
symmetry and beauty if we cannot relate them to what occurs in the 
living cell and if our only criterion for their acceptance is some aesthetic 

If, however, we can examine the results of the interaction of a process 
such as chemical fixation on the cell and from knowledge of the chem- 
istry and physics involved determine how the structures we see have 
arisen and what the structure of the cell must have been prior to the 
fixation, then it matters little what distortions occur in the process of 
artifact production. It is, therefore, the production of interpretable arti- 
facts which should be the goal of preparative techniques in electron 
microscopy and, in fact, of any investigative procedure. 

Chemical fixation offers a number of advantages in cell biology, not 
the least of which are a long tradition of use, a wide group of prac- 
titioners and an enormous number of fixatives tested. For a compendium 
of fixation methods, the reader is referred to Hayat (1972). Although 
many of the earlier cell biologists were interested in what effects were 
produced on cell structure by various fixation procedures (Wolman, 
1955; Baker, 1960), this analytical tradition has not continued with any 
great force in recent years. More recently, however, Hayat (1970) has 
discussed in detail the chemical reactions involved and the artifacts pro- 
duced during chemical fixation. It is accepted that chemical fixation can 
cause serious alterations of cell structures, some of which arise because 
of a necessary diffusion of different components of the fixative into the 

* No technique without artifacts is available, from which vital information for the 
mental reconstruction of a living cell can be extracted. 


cell at different rates and because of the fact that turbulence due to local 
dehydration by the fixative of proteins and other macromolecules, pH 
changes, local precipitation and crosslinking reactions can occur during 
this diffusion. Possible alterations arising from such processes may give 
rise to uninterpretable artifacts although these may be in the form of 
regularities which appeal to our aesthetic sense. 

Attempts to use other processes to prepare cells for microscopic 
examination have been undertaken, in part, to overcome these difficulties 
in interpretation. Rapid freezing is an obvious technique which may be 
used and which appears to have little in common with chemical methods 
of fixation. However, alterations of other sorts become apparent in the 
form of ice crystallization which freezing normally induces ( Rebhun and 
Sander, 1971a). Although it is possible that some minute separated cells 
such as yeast or bacteria may be frozen free of ice crystals ( Moor, 1964 
and 1969), in the usual case this is probably not possible for unmodified 
cells. We must, therefore, suppress ice crystal formation if we are to use 
the technique for routine work (although where ice crystals form, their 
size and which structures in the cell survive ice crystal formation are 
themselves of some interest). Such suppression requires modification of 
the cell water so that a sufficient number of cells escape ice crystal 
injury for useful observations. 

Alteration in the ability of water to form ice crystals generally in- 
volves partial dehydration or the utilization of an "antifreeze" substance, 
and it is necessary to justify such use by detailed examination of the 
effects of these agents on living cells. Other than studies of viability 
(which in reality do not yield information on this matter) there are 
few direct studies of the effects on living cells of agents such as dimethyl- 
sulf oxide (DMSO), ethylene glycol and dimethylformamide (Rebhun 
and Sawada, 1969). One may consult Smith (1961), Doebbler (1966) 
and Nash (1966) for common freeze protective agents. Both partial de- 
hydration and freeze-protective agents modify the cells in a controlled 
manner, and this factor must be kept in mind where results are viewed 
on the screen of the electron microscope. 

Since freezing techniques are not simple and their results have gen- 
erated some controversy, it is not useful to present techniques as a set 
of recipes for uncritical use; the criteria for acceptance of the technique 
must be set before the prospective user. The epistemological problem of 
interpreting each step in the preparatory procedure thus becomes an 
integral part of the total technique, which is, perhaps, a great advantage 
of freezing methods. 

In the following pages, an attempt has been made to analyze various 
aspects of the freezing process and its applications to the study of 


cellular ultrastructure. Since many problems remain unanswered, some 
of the directions which might lead to fruitful investigation have been 
pointed out. The discussion is limited to the processes of freeze-substitu- 
tion and freeze-drying, for the technique of freeze replication is discussed 
elsewhere in this volume. 


Structure of Water and Ice in Cells 

In order to investigate the conditions for the formation and suppression 
of ice in frozen cells, it is necessary to look at recent ideas of water and 
ice and the form in which these occur in cells. Most approaches to the 
study of the liquid state extrapolate from what is known of either the 
gaseous or the solid form of the substance (Kavanau, 1964; Eisenberg 
and Kautzman, 1969). Thus, known ice structures have played a very 
large role in the models proposed for the structure of liquid water. A 
recent model of water ( Frank and Wen, 1957 ) is based on the idea that 
hydrogen bonds in water may be cooperatively formed and broken so that 
in a region where several bonds form, a higher probability occurs for 
further formation of H-bonds. If one bond is broken in such a cluster, 
due to thermal fluctuations, the tendency is for others in its immediate 
vicinity to break and the result is a cooperative breakdown of the cluster. 
Water is thus conceived of as a flickering cluster liquid and the lifetime 
of each cluster, about 10 -11 sec, is the result of the two competing 
processes. A number of interesting properties of water have been ex- 
plained through this model. It has also been used to calculate properties 
of hydrophobic bonds in proteins (Scheraga, 1966). 

It has become clear, however, that this theory cannot account for 
much experimental data, particularly the high-angle X-ray scattering data 
from water (Narten and Levy, 1969), and it has been abandoned even 
by its original propounder (Frank, 1970). For a detailed discussion the 
reader is referred to Eisenberg and Kautzman (1969). The best model so 
far proposed in its ability to account for much of the experimental data 
including that from X-ray scattering appears to be in interstitial model 
in which an ice-like lattice is somewhat distorted and in which water 
molecules may occupy spaces within the cavities of the lattice ( Berend- 
sen, 1966; Narten and Levy, 1969; Frank, 1970), and hydrogen may bond 
to lattice oxygens. The lattice itself has been suggested to be a modified 
hexagonal ice lattice (Berendsen, 1966; Narten and Levy, 1969), or more 
likely, a mixture of a variety of ice structure with considerable and 
rapid distortion of lattice H-bonds. 


If the liquid structure so closely resembles that of the solid, the 
freezing of water would be expected to yield one of the crystalline forms 
and it would be expected that amorphous or vitreous (glass-like) water 
would be difficult to obtain. In confirmation of this, there do not appear 
to have been any reports that water may be frozen into an amorphous 
state starting from the liquid state irrespective of the freezing velocity or 
the size of the sample (Dowell and Rinfret, 1960; Delsemme and 
Wenger, 1970). Vitreous ice has been reported but only when pure 
water is condensed from the vapor state onto a cold surface. When such 
a film of frozen water is raised to about — 160°C, it can transform into 
cubic ice; the time for such transition decreases exponentially as the 
temperature approaches — 130°C. At this point and above, the stable ice 
form is hexagonal (at normal pressures) and if cubic ice has formed, it 
will transform into the hexagonal form. All this implies considerable 
mobility and possibility for phase transition at very low temperatures 
even though crystal lattices are maintained. 

Considering the ease with which water should and does crystallize 
into ice, it is quite surprising that pure samples can be supercooled to 
temperatures as low as — 40°C (Fletcher, 1970). Since ice forms around 
nuclei which may be of foreign material (heterogeneous nuclei) or of 
water itself (homogeneous nuclei) (Fletcher, 1970), these pure samples 
appear to have a dearth of homogeneous nuclei. This is difficult to under- 
stand if liquid water is built around an ice-like lattice, unless the lattice 
distortions render difficult the choice of what lattice to assume when 
transition to the solid state is imminent. 

Frozen solutions can be considerably more complex than frozen 
water. In particular, solutions of proteins can be frozen without forma- 
tion of detectable crystalline ice, but, at moderate freezing velocities, 
only when the concentration of proteins is 50% or greater ( Dowell et al., 
1962; Sterling, 1969). Solutions of glycerol can also form amorphous or 
vitreous bodies, and at moderate concentrations a frozen solution of 
glycerol will contain both amorphous and crystalline water. Only when 
concentrations of glycerol are higher than 73%, is it able to form bodies 
containing only vitreous ice (Luyet, 1970). Although the samples used 
were considerably larger than would normally be used in freezing of 
glycerol-impregnated cells for morphological studies, and equilibrium 
conditions (i.e., slow freezing) rather than kinetic conditions were in- 
vestigated, it is clear that pure vitreous ice is not easy to obtain even 
with antifreeze agents. Indeed, MacKenzie and Luyet ( 1962a ) were still 
able to see some evidence of ice formation in very thin (less than a 
micron) films of 20% gelatin frozen at rates in excess of 100,000°C/sec. 

The difficulty in freezing the ordinary cells free of ice crystals is 


apparent since cells contain approximately 20% total solids (Barer, 1956), 
which is less than the concentration required in model systems to sup- 
press ice crystal formation. This presupposes, however, that the water 
inside cells is similar to that in the model solutions. In other words, there 
is no special cellular state for water which would make model protein or 
glycerol systems unfit models for comparison to how cell water will 
behave in the freezing process. 

The state of intracellular water is a subject of some controversy. Ling 
(1965), Horowitz and Fenichel (1965), Cope (1969) and Troschin 
( 1958 ) are of the opinion that intracellular water is in some relatively 
rigid, if not crystalline, form, exists in a special relation to cellular 
macromolecules, and determines the ion uptake and specificity properties 
of cells. On the other hand, Dick (1966) and Olmstead (1966) believe 
that most cellular water is not special in any extraordinary manner. 

That there is some bound water in cells seems to be clear from a 
variety of physical measurements (Webb, 1965; Meryman, 1966), and a 
value of 10-15% of the total cell water seems reasonable from these 
studies. However, that the concept of bound water need not be invoked 
to explain the properties of systems whose anomalous properties were at 
one time thought to derive from some fraction of odd internal water 
seems clear in the light of recent work by Solomon ( 1971 ) . He showed 
that anomalies of osmotic pressure in mammalian red cells resulted from 
hemoglobin interaction rather than from bound "non-osmotic" internal 
water. These studies would suggest that most or even all of the water in 
cells is ordinary solvent water. In particular, there is no evidence for 
anomolous or polywater (Derjaguin, 1970; Page et al., 1970; Allen and 
Kolman, 1970) in cells, assuming that such water exists at all. Several 
recent studies oppose the existence of such water (Kurtin et al., 1970; 
Rousseau and Porto, 1970; Wentorf, 1970; Rapideau and Florin, 1970). 

Whatever the state of water is in normal cells, one must consider how 
cellular water behaves at very low temperature. Direct investigation of 
this problem is difficult since any technique for the microscopic observa- 
tion of cells which have been frozen requires further preparative steps 
after freezing. These steps include removal of water which could change 
the frozen cell so as to destroy the information that is sought. Methods 
such as X-ray diffraction or nuclear magnetic resonance, which can tell 
us something about the states of water, do not allow us to examine the 
water in individual cells, but only the average for the whole preparation, 
which includes extracellular water. 

Nevertheless, interesting results have been obtained with macro- 
molecular solutions which suggest that some water in frozen solutions 
retains liquid-like mobility. For instance, in collagen there is a fraction 


of water which does not form crystalline ice at — 50° C when the rest of 
the solution has frozen (Dehl and Hoeve, 1969; Dehl, 1970; Berendsen 
and Migchelson, 1965 and 1966). In an extensive study of several pro- 
teins and nucleic acids in solution, Kuntz et al. (1969) found that at 
— 50°C a fraction of water retained a liquid-like state (certainly of 
higher viscosity than ordinary water) amounting to about 0.4 g of water 
per g of protein and 1.7 g of water per g of nucleic acid. These quantities 
agree well with water of hydration as measured by other techniques and 
calculate to be enough for one or at most two layers of water surround- 
ing each cellular macromolecule. 

If the above results can be transferred to intact cells, they suggest 
that about 10-15% of cellular water (assuming cells to have about 20% 
of their weight as macromolecules ) does not freeze at low temperatures. 
It also strongly suggests that water associated with macromolecules in 
the cell is not in a state in which it can nucleate ordinary cellular water 
to form ice (since it does not freeze itself), and thus removes this large 
cell component from the possible sources of heterogeneous nucleating 
sites. This may help explain why cells are so easily supercooled ( Mazur, 
1966; Asahina, 1965 ) when one would expect a large number of potential 
heterogeneous nucleating sites to exist. 

Considering that water and relatively concentrated solutions of pro- 
teins and of potential antifreeze agents are difficult to freeze free of ice 
crystals and that most of the intracellular water is probably not in some 
special state in which it can avoid ice crystallization when the tempera- 
ture is lowered beyond the cell's ability to supercool, it seems unlikely to 
expect that most cells can be frozen free of ice crystals without the inter- 
vention of some prior modification of cellular water. This conclusion has 
been presented in the deductive manner in which the results of numer- 
ous experiments are usually reconstructed, but the reader should have no 
difficulty in recognizing that it is based on an intensive and long series 
of attempts to do the opposite of what is suggested as unlikely. The con- 
clusion is based on the results of considerable frustration in attempts at 
freezing cells free of ice crystals. The arguments in support of this con- 
clusion have been presented in greater detail in several recent papers 
(Rebhun, 1965; Rebhun and Sander, 1971a, 1971b and 1971c). 

The above conclusions have been based on the author's direct ex- 
perience with cells and tissues which must be handled in samples (liver, 
pancreas, small droplets of marine eggs, etc.) of the order of 0.1 to 
0.2 mm in linear dimension. However, it may be possible to freeze thin 
layers of cells in culture (Asahina et al., 1970) or very small droplets of 
yeast cells free of ice without prior modification of cellular water ( Moor, 
1964). In such cases heat transfer from the sample to the quenching bath 


may be very much enhanced by the smaller distances it traverses and 
by the possibility that in smaller cells water may behave differently from 
the way it behaves in larger cells. This may arise from special wall effects 
on freezing of water, as appear to occur in a variety of systems such as 
montmorrillite clays (Derjaguin, 1970), or because of the inhibition of 
ice formation in very small capillaries (Kuntz et al., 1969), which may 
be the same thing. 

A technique of freezing in which tissues are rapidly "squashed" 
against a rod held at liquid or freezing nitrogen temperatures may also 
attain vitreous freezing in the surface layers ( 10-15 (x, thick ) immediately 
in contact with the rod (Van Harreveld and Crowell, 1964). The pos- 
sibilities of compression and pressure effects on cells so treated are ob- 
vious but the technique might be of considerable use if these effects can 
be rationally understood in those cases where other methods, to be 
discussed below, cannot be used for one reason or another. 

Another technique of very rapid freezing which may yield useable 
ice-free regions of tissue has been reported by Bullivant (1970) and 
Monroe et al., (1968). It consists of a device to fire 22 gauge (0.39 mm 
inside diameter) hypodermic tubing at supersonic velocities through the 
tissue to be frozen and into liquid propane. It is difficult to imagine a 
more rapid dissection and transfer technique and the application to 
tissues undergoing rapid changes (heart muscle during contraction was 
the subject under investigation in the above papers) is clear and im- 

Freezing and Viability 

Given the difficulties of obtaining ice crystal-free freezing of unmodified 
cells for morphological purposes, it is reasonable to turn for any informa- 
tion available to a possible relation between the survival of frozen cells 
and preservation of morphology. Most of the work in this field indicates 
that no direct relation between the two exists (Mazur, 1966 and 1970). 
At the very best the survival of frozen cells can only be determined after 
the cells are thawed. However, considerable evidence exists to suggest 
that this process can seriously affect survival (Meryman, 1970; Farrant 
and Woolgar, 1970). 

Nevertheless, certain relatively well-supported conclusions can be 
drawn. The survival of cells appears to depend on two processes — avoid- 
ance of intracellular ice crystallization and avoidance of salt damage due 
to osmotic concentration of solutes in the cells in which crystalline ice 
formation does not occur. If one considers a suspension of cells during 
the freezing process, it will be seen that the extracellular solution com- 


mences freezing at some point and in doing so causes an increase in the 
concentration of solutes in the remaining solvent water (Farrant, 1970; 
Farrant and Woolgar, 1970). This results in an increase in osmotic pres- 
sure which will cause water to leave the cell. If the cell permeability is 
high, water can leave fast enough so that intracellular freezing does not 
occur because of rapid increase in intracellular concentration of salts, 
metabolites and macromolecules. If the permeability is not high, the 
water does not leave fast enough and the cellular contents will remain 
dilute enough to freeze at an appropriate temperature. 

Extensive evidence suggests that the formation of intracellular ice 
during freezing is fatal to survival of cells after thawing (Mazur, 1966 
and 1970; Mazur et al, 1970). In the case where no intracellular ice is 
formed, the concentration of salts which occurs during freezing may at- 
tain very high levels (4-5 M) (Farrant, 1970) and the cells may suffer 
salt damage including damage to membranes (Lovelock, 1953; Levitt 
and Dear, 1970). The longer the cells remain in contact with such high 
salt concentrations, the greater the probable damage. Thus, at slow 
freezing rates cells are likely to suffer salt damage which tends to de- 
crease as the freezing rate is increased ( less time in contact with the high 
salt concentration in the liquid state), but as still higher rates are 
reached, the cell suffers from intracellular ice crystallization. The two 
processes thus yield an optimum freezing rate for a given type of cell 
depending upon its permeability properties and its resistance to salt 

The above-mentioned considerations have been presented in detail 
by Mazur (1966) and appear capable of being utilized in the interpre- 
tation of most freeze-survival curves, although other suggestions to 
account for damage at low freezing rates have been proposed (Mery- 
man, 1970). The interaction of the parameters of cell permeability to 
water and sensitivity to salt damage yields very different curves for dif- 
ferent types of cells, and optima in freezing rates can vary several 
thousand fold, e.g., for mammalian stem cells from spleen and red blood 
cells (Mazur et al, 1970). 

The effects of freeze protective agents on survival are of considerable 
interest. Agents such as glycerol and DMSO do not generally increase 
survival of cells at freezing velocities higher than the optimum for un- 
protected cells (Mazur et al., 1970) and, in fact, in certain cells many 
agents actually significantly decrease the survival of cells ( Farrant, 1970 ) . 
The method by which these agents affect cell survival on the slow 
freezing rate side of the optimum point appears to be through protec- 
tion of the cells from the effects of high salt, a theory which has been 
given considerable experimental support (Farrant and Woolgar, 1970); 


for additional and alternative considerations, the reader is referred to 
Meryman (1970) and Mazur et al. (1970). Interestingly, agents such as 
sucrose, which can protect cells against freezing damage but do not 
penetrate them, appear to protect cells much more readily on the high 
freezing rate side of the optimum rate found in the absence of any pro- 
tective agent (Mazur, et al, 1970; Farrant, 1970). The reason for this 
(and for similar action of PVP) is not known but may be due to the 
inhibition of seeding of intracellular ice crystals through the surface 
membrane by ice crystals in the extracellular medium. 

From this very brief summary of recent results in the field of survival 
of cells which have undergone freezing, it can be seen that the use of 
survival data to justify employment of freeze protective agents for 
morphological studies requires more care than it is usually afforded. In 
particular, survival data by itself cannot justify utilization of, for example, 
glycerol to protect cells against freezing damage to structure in the rapid 
freezing process, because, in fact, the data provide evidence for increased 
damage (Farrant, 1970) after thawing of cells frozen too rapidly. In 
addition, in some types of cells (e.g., spleen stem cells) while the survi- 
val rate is greatly increased as glycerol concentration is increased, the 
optimal freezing rate for greatest survival is actually decreased (Leibo 
et al, 1970). This property makes the freezing of large samples of 
biological material much more feasible. 

These results make difficult the direct application of survival data to 
the interpretation of the state of water in frozen cells. The survival of 
frozen cells after thawing is no guarantee that morphology has been 
unchanged in the frozen state; it may only mean that the cells more 
readily overcome morphological damage when a protective agent is 
present. The difficulties of comparing survival data with morphological 
data are enhanced by the fact that most freeze survival studies are done 
with yeast, bacteria, red blood cells or cells in culture, and that most 
morphological studies have been done with organized tissues such as 
liver, pancreas or kidney. Other reasons for proceeding with morpholog- 
ical considerations independently of survival studies are: (1) cell survival 
studies generally use concentrations of protective agent in the range of 
5-15% (Mazur, 1966 and 1970), whereas morphological studies generally 
use considerably higher concentrations (20-60%) (Rebhun, 1965; Pease, 
1967a and b); (2) each cell type has its own optimal rate of survival 
with a protective agent (Mazur et al, 1970); and (3) no studies of the 
effects of freezing rate on survival are available for small samples of 
organized tissues. The question of morphological integrity is, therefore, 
independent of the question of viability, although, of course, the rela- 
tions of the two are of considerable interest and importance. 


On the other hand, the facts that partial dehydration is a natural 
accompaniment of freeze survival in many cells (Mazur, 1966) and that 
freeze protective agents do considerably enhance viability after freezing 
at certain rates lend considerable strength to the suggestion that the 
morphology of cells so treated may be of considerable help in recon- 
structing the model of the living cell. In other words, partial dehydration 
and treatment with freeze protective agents introduce ultimately inter- 
pretable artifacts which allow freezing methods to be applied to many 
cell studies. 

Morphological Effects of Pretreatment Regimes on Living Cells 

Since in most cells it is unlikely that ice crystal free freezing can be 
achieved without either partial water removal or suppression of ice 
formation by use of freeze protective agents, possible effects on the 
morphology of cells using these procedures are important enough to be 
investigated. For this and other considerations, the effects of hypertonic 
conditions and of various freeze protective agents on the in vivo mitotic 
apparatus (MA) have been examined (Rebhun and Sawada, 1969; 
Rebhun and Sander, 1971c). In the case of hypertonic sea water, the 
process of mitosis appears to be slowed down. If the MA is in metaphase 
it will remain at this stage and, depending upon the degree of hyper- 
tonicity, it will gradually break up over a period of 0.75-3 hr into a set of 
birefringent fibers which themselves gradually disappear (Rebhun and 
Sander, 1971c). 

There is also a considerable effect of freeze-protective agents such as 
DMSO, ethylene glycol ( glycerol penetrates marine eggs only with great 
difficulty and over long periods of time), formamide and dimethyl- 
formamide on the MA of marine eggs ( Rebhun and Sawada, 1969 ) . The 
first two agents cause a great increase in the size of this structure. The 
effect is rapid (3-5 min for completion) and reversible if the agent is 
removed within a period of 10-30 min. The increase in volume and 
birefringence of the MA is accompanied by a large increase in the num- 
ber of microtubules present in the structure, and an analysis of this 
effect suggests that the augmentation is due to mobilization of a pool of 
MA material including microtubule protein into the MA in a reversible 
manner (Rebhun and Bernstein, 1967). 

The other two protective agents have the opposite effect on the in 
vivo MA, namely, they disperse it so that birefringence disappears ( Reb- 
hun and Sawada, 1969). Although an ultrastructural analysis of this 
latter process has not yet been completed, the preliminary work by 
the author and by Nevo (1966) and Harris (1962) suggests that the 


MA microtubules are not broken down as would occur in a Colchicine- 
treated cell, but are separated from each other and become dispersed in 
the cytoplasm similar to what occurs in spindles after treatment with 
isopropyl N-phenylcarbamate (Hepler and Jackson, 1969). 

Although the results discussed above cannot be applied without 
reservation to other types of cells, it is clear that processes which are 
proven to allow some morphology to survive freezing and which protect 
some cells against freezing damage in the sense of increasing their sur- 
vival rates after thawing have profound effects, at least on the MA, 
and presumably on the microtubules which are an important part of its 
structure. While these effects are themselves of importance in the study 
of miotic processes, they suggest caution in accepting the results of rapid 
freezing, especially if results are at variance with the data obtained 
by other techniques. 

None of these results should be taken as a reason for abandoning this 
promising technique. They are meant only as precautions which if kept 
in mind may forestall unwarranted conclusions and preclude unneces- 
sary controversy. In a sense the very pitfalls and difficulties of the method 
may be its greatest strength in that it forces direct consideration of the 
basic epistemological problem in electron microscopy, that is, how the 
standards should be set up so that one may have confidence in the results 
displayed in the micrographs. 

It should be clear from the above discussion that the control of 
intracellular water is one of the most important of the variables in the 
freezing process, and that claims of ice crystal free freezing should be 
examined in great detail to see whether some accidental loss of water 
may not have occurred during the preparative steps in rapid freezing 
either prior to or during freezing itself. It should also be noted that while 
a number of authors have reported success in ice crystal free freezing 
of unmodified cells, their everyday results are often obtained with 
glycerol impregnated, or even fixed and glycerol impregnated cells. 
There is nothing to be criticized except the implication that ice crystal 
free unmodified cells are the usual fare. Let the prospective practitioner 
be cautioned that the facts are otherwise. 

Recognition and Avoidance of Intracellular Ice Crystals 

The appearance of cells which have been subjected to intracellular ice 
crystallization has been recently discussed in some detail (Rebhun and 
Sander, 1971a, 1971b and 1971c). Fig. 1.1 (marine egg) and Fig. 1.2 (rat 
pancreas) illustrate the appearance of cytoplasm in cells frozen so as 
to prevent accidental evaporation of water. A comparison of the walls 

"» 1 ' ■«*» "- ■ ■ >XM "> ^«i«*" <flMVaU ^<_ a*-**' „'"2«BB» -- i' m ■ -w- '■■■mmu- ■■■■'■■• ■ 

Fig. 1.1 Structure of an egg of the surf clam Spisula solidissima frozen in normal 
sea water in Formvar film sandwiches to prevent accidental evaporation of water. 
N = nucleus, ER = endoplasmic reticulum, AL = annulate lamellae (barely recog- 
nizable), V = vitelline membrane. Other particles are yolk, mitochondria, etc. 
Notice that no ice forms in the latter particles. X8.000. 

i J V.,'' 

> V 

a ••*■: 

■* .* 



■ " ■; j*: + ■> '■. :; > ; v*'#?/ • .*/ &■. ■ - ;.- - v Y *-.-;- t - . -■■ y* ■. 

Fig. 1.2 Nuclei and cytoplasm of cells from rat pancreas frozen so as to prevent 
any water loss. Ice crystal damage is recognizable in both nucleus and cytoplasm. 
ER and mitochondria (M) can be seen in the cytoplasm. X6.000. 



and spaces in the cytoplasm of these cells with those seen in a va- 
riety of model systems which have been frozen and either vacuum 
dried or solvent substituted (Luyet, 1966; Rapatz et al., 1966) leaves 
little doubt that the spaces were formerly occupied by ice crystals and 
the walls result from the displacement of material from the forming ice 
crystal. The size of the spaces varies from cell to cell in a given prepara- 
tion and in some preparations almost disappear from some regions of 
the cell. 

If marine eggs are frozen in sea water with increasing salt concentra- 
tion so that the cells shrink osmotically, the size of the spaces decreases, 
although much variability in size occurs in each preparation. When more 
than half of the water is removed from the eggs a morphology similar to 
that in Fig. 1.3 occurs in the cytoplasm of 10-50% of the cells, and as 
more water is removed the walls and spaces disappear from the nucleus 
as well. One interpretation of these results is that ice crystallization be- 
comes less probable as water is removed from cells by osmotic means 
and that when water content is reduced to one-half to one-third of 
normal, a significant number of cells escape nucleation and can be pre- 
pared free of ice crystal artifacts. 

However, if the results of many freezing experiments on undehy- 
drated cells are examined, a small percentage will be found which do not 
have the wall and space morphology which we ascribe to ice crystal 
formation and it is necessary to account for such cells. Often, ice crystal- 
free regions are immediately adjacent to other regions in which the ice 
crystal spaces are particularly large (Fig. 1.4). It is suggested that such 
appearances result from dehydration of the region without ice crystals by 
the region in which such crystals are forming as a result of osmotic pres- 
sure built up in the spaces between ice crystals. The osmotic dehydration 
of cells by water loss across the cell membrane to a region undergoing 
ice crystal formation outside the cell is the basis for the analysis of 
viability phenomena during freezing (Mazur, 1966), and there seems 
no reason to doubt that it can also occur within cells as large as marine 

It is concluded that ice crystallization can be suppressed in a certain 
proportion of cells from which a sufficient amount of water has been 
removed. Similar results are obtained in pancreas tissue which has been 
partially air dried for short periods of time (~lmin) (Fig. 1.5) and in 
liver cells which have been frozen in physiological solutions made hyper- 
tonic with sucrose (Fig. 1.6). In the latter the form and shape of the 
mitochondria and other subcellular inclusions appear to have suffered 
from shrinkage but there is little difficulty in recognizing these structures, 
and the fact that they are obtained by techniques other than the usual 


•\ A 


; .r* : *' 

" ? •■»£■■ :"^*«*>*w '- " "■ ** ■ A____ 

if. '" " ' ■' .'■*".'■ -''"/-'I 

HI j#_ _ *t ■ ■ ■ rfijiit.'i hj - fA ^» *» j 

ft ■ V •« ; • «£• j,, v&|lK*&Jm 

■&.*' ■ J ..VV.'" -■'■'■ j 



Fig. 1.3 Cytoplasm of a clam egg frozen from hypertonic sea water (2 1 /2 times 
normal). Ice crystals are not evident. Mitochondria (M), individual cisternae (C,) 
of the ER, ribosomes, and microtubules (M,) can all be readily recognized in 
cells frozen in this manner. Particle free regions containing dense material 
(possibly a matrix material) can be seen (PF) and show strong orientation proper- 
ties in the spindle and asters during mitosis (Rebhun and Sander (1971c)). X83.000. 


Fig. 1.4 Clam egg frozen from slightly hypertonic sea water (112%). Over 95% 
of the cells in such preparations are similar to those shown in Fig. 1.1; that is, they 
possess cytoplasm and nucleus extensively vacuolated with the spaces arising 
from ice crystal formation. Some eggs possess regions similar to those on the 
right hand side of the figure in which ice crystal formation is slight or non- 
existent. Such regions are generally immediately adjacent to regions possessing 
vacuoles arising from extensive ice crystal formation. A golgi body (G) occurs in 
the region of reduced ice formation. X12.000. 

chemical fixation holds hope that they may be of considerable use in 
cytochemical procedures. 

The second procedure for avoidance of ice crystal formation in cells 
makes use of the freeze protective agents such as glycerol, ethylene 
glycol, DMSO, etc. An extensive set of studies was initiated in the 
author's laboratory to examine conditions which might allow ice crystal- 
free freezing of liver and pancreas cells of rat. Concentrations of glycerol, 
DMSO and ethylene glycol from 5-30% in various saline solutions were 
used in freeze-substitution techniques by methods to be outlined below. 
The tissues were removed from rats under Nembutal anesthesia and 
transferred within 2 to 3 sec to solutions of the particular type to be 
investigated. They were cut into small pieces of no more than 0.2 mm 
on a side (controlled by examination of the cutting process with a 
microscope) and were then left in the solutions for the desired length 



Fig. 1.5 Part of cytoplasm from a sample of rat pancreas frozen after exposure 
to air for one minute. Extensive areas of the first three or so layers of cells in 
from the surface possess morphology similar to this. x32,000. 

of time from 5-30 min. The specimens were frozen and substituted using 
techniques described in detail below. 

The salt solutions used were various physiological salines and also 
salt solutions made up to mimic reported intracellular concentrations of 
various salts. In particular, both sodium predominant and potassium pre- 
dominant salines were used. Tissues were allowed to equilibrate at 0-2° C 
or at room temperature (22-25°C) prior to freezing. In the author's ex- 
perience with tissues frozen from over 50 rats ( about 500 samples ) con- 
siderable damage to tissues was found unless they were impregnated 
with a freeze protective agent for at least 15 min at room temperature 
and in solutions containing 25-30% glycerol (Fig. 1.7). In all other cases 
almost all of the cells possessed ice crystals or were full of vacuoles, 
highly swollen or very shrunken mitochondria and other subcellular in- 
clusions and were, in general, very distorted (Fig. 1.8). (See similar ex- 
periences with frozen replicated HeLa cells reported by Bererhi and 
Malkani, 1970.) 

At most 25% of the cells in each block showed the morphology seen 
in Fig. 1.7, and thus the block had to be carefully trimmed for study. 



Fig. 1.6 Portion of cytoplasm of rat liver cell frozen in Locke's solution with 
enough added sucrose to make the solution twice normal tonicity. Mitochondria 
are quite elongated. ER and Golgi are readily recognized. The spaces are pre- 
sume* location of glycogen. X30.000. 

Although regions of injury showed no consistent relation to the surface 
of the block, certainly some of the damage must have resulted from the 
process of cutting very small samples of tissue from large organs. At 
glycerol concentrations of 30% the cells began to resemble those seen 
after freezing in hypertonic solutions of sucrose. It is not unreasonable 
to expect that hyperosmotic conditions can result from high concentra- 
tions of glycerol. Micrographs produced from tissues impregnated with 
concentrations of glycerol up to 60% show clear signs of considerable 
shrinkage (Bullivant, 1965; Pease, 1966a and b; 1967a and b), and the 
success in freezing such cells without ice crystals undoubtedly derives 
from a combination of increased concentration of intracellular con- 
stituents and a reduction in the ability of water to form ice in the 
presence of glycerol. 

Dehydration of Frozen Cells 

The discussion so far has concentrated on the fundamental aspects of 
avoidance of ice crystal formation during freezing and not on the pro- 


Fig. 1.7 Rat liver frozen after impregnation with 25% glycerol in Locke's solution 
for 30 min at room temperature. About 20-25% of the cells in this block had 
morphology comparable to this. The others had ice crystals or various other 
forms of obvious damage (See Fig. 1.8). X1 7,000. 



Fig. 1.8 Rat liver cytoplasm from sample frozen under the same conditions as 
the sample in Fig. 1.7. Extensive vacuolization of the ER occurs as well as 
widened spaces in the mitochondrial cristae. X14,000. 

cedures by which cells are subsequently processed. The three most 
widely used methods for ultimately observing the results of freezing in 
the electron microscope are the use of solvent dehydration (Fernandez- 
Moran, 1960 and 1961; Rebhun, 1961; Bullivant, 1965), vacuum dehydra- 
tion (Gersh, 1956; Gersh et al, 1956; Sjostrand and Baker, 1958) and 
freeze replication, which is discussed elsewhere in this volume. 

In the first technique, organic solvents such as ethanol, acetone or 
tetrahydrofuran are used to remove ice from the frozen specimen at low 
temperatures (Rebhun, 1965; Rebhun and Sander, 1971a, 1971b and 
1971c; Pearse, 1968). The necessary time to accomplish this is not known 
with precision, but experiments with the rate of dissolution of frozen 
droplets of dye and NMR measurements (Van Harreveld et al, 1965) 
suggest that at —70° to — 80°C small pieces of tissue are substituted in 
several days, although we generally use up to two weeks. There have 
been suggestions that the nature of the solvent partially determines the 
morphological result obtained, but the author has seen no difference in 
his work that can be ascribed to the use of different solvents (see also 
Van Harreveld and Steiner, 1970). This may not extend to other prop- 
erties of the cells and any chemical differences which may be discovered 


in cells substituted with different fluids may be of importance for cyto- 
chemical purposes. 

The inclusion of fixatives such as osmium tetroxide has not made any 
important ultrastructural difference in our work but it docs offer a great 
convenience in being able to locate cells after plastic embedding. How- 
ever, it is possible that new cytochemical procedures may result from 
systematic exploration of the abilities of various fixatives to differentially 
stabilize certain structures or chemical groups at low temperatures which 
may then be revealed by later treatments. One of the possible outgrowths 
of this work may thus be an extension of normal cytochemical procedures 
to new methods employed at low temperatures. It is also possible that 
these considerations will hold for cells briefly fixed in glutaraldehyde or 
formaldehyde, impregnated with a protective agent and subsequently 
subjected to freeze-substitution. There is no basis for deciding whether 
ordinary dehydration will be similar or different from dehydration 
through the freezing route and very important differences may be found. 

The process of vacuum drying of frozen tissues appears to be a more 
exacting and morphologically damaging process than solvent substitu- 
tion. Reviews of earlier results may be obtained from Bell (1956), 
Pearse ( 1968 ) and Burstone ( 1969 ) . In experiments with model systems, 
MacKenzie ( 1965 ) found numerous morphological transformations as the 
drying process proceeded. Furthermore, ordinary freeze protective agents 
cannot be used, since they will concentrate as the drying process pro- 
ceeds, which will result in the frozen tissue ultimately sitting in a pool of 
protective agent. This may not apply if volatile protective agents are dis- 
covered (Meryman, 1969 and 1970), but would appear to be limiting at 
the present time. 

Embedding of vacuum dried material in plastic for subsequent sec- 
tioning would appear to impose some possible dangers to structure 
through the necessity of "wetting" the tissue with plastic (Hanzon and 
Hermodssen, 1960; Sjostrand and Baker, 1958). Fixation of frozen-dried 
cells with alcohol (Gersh et al., 1956) or osmium vapor (Eckert, 1969) 
has been used in part to overcome this difficulty. Despite these draw- 
backs, frozen dried tissues may offer considerable opportunity to explore 
the cytochemical potentialities of vapor reagents (for which a different 
set of reagents may be available than for ordinary fixation ) after suitable 
fixation with aqueous reagents. If the requirement for morprological 
fidelity to the living state is relaxed then one may focus on the possibly 
important cytochemical differences which fixed, frozen-substituted, or 
frozen-dried cells may offer. In this context the clearly shrunken cells 
which result from the freezing techniques of Pease ( 1966a and b; 1967a 
and b) may have an important unique cytochemical advantage in some 


In the above discussion it was assumed that the usual techniques for 
substitution or vacuum dehydration were employed. However, the ex- 
citing possibility arises from recent work with low temperature thin 
sectioning (Bernhard and Leduc, 1967; Leduc et at, 1967; Zotikov and 
Bernhard, 1970; Christensen, 1969) in that these methods may be applied 
not only to pieces of tissue, but also to slices of frozen tissues. In freeze 
substitution, frozen tissues or cells may be sectioned directly onto a 
substituting fluid or sections may be transferred to it subsequent to sec- 
tioning. This may result in great speedup of specimen preparation and 
total avoidance of plastic impregnation. In the case of freeze-drying it 
appears possible to section onto a dry knife and to dehydrate using a 
scheme of dry nitrogen dehydration ( Meryman, 1959; Christensen, 1969 ) . 
Avoidance of plastic impregnation may result in quite different cyto- 
chemical properties of cells. 


General Principles of Freezing in Boiling Fluids 

As has been described above, a given tissue may require quite different 
conditions for optimal preservation of morphology when compared to 
other tissues. These conditions may involve different freeze protective 
agents or dehydration conditions and different freezing rates for a given 
size sample. Since such technology has hardly been conceived let alone 
carried out, the author has attempted to utilize one freezing rate in his 
work to date and uses that as a base line. The rate chosen is, for historical 
reasons, the fastest available. How that may be attained, however, is 
itself a subject of some controversy. 

Liquid helium II has been suggested as a quenching fluid which 
should yield exceedingly high freezing rates because of its superfluid 
properties and exceedingly high velocity of heat conduction ( Fernandez- 
Moran, 1960). However, actual thermocouple measurements have not 
supported this suggestion (Bullivant, 1965). Furthermore, the liquid 
helium II component of liquid helium only exists below the lambda point 
for helium, which can be obtained by subjecting liquid helium to a 
vacuum of below 70 /a of mercury. Since the specimen must also be 
placed in a dry atmosphere prior to immersion in the liquid helium 
( Femandez-Moran, 1960) to prevent condensation of water vapor in 
the system, the chances for accidental "dry-freezing," i.e., partial dehy- 
dration of cells before quenching, are clearly great. 

For all of these reasons, and also because of the great expense in 
handling this material, attention was turned to other substances. Liquid 
hydrogen has certain properties which suggest that it might be an ex- 


cellent quenching fluid if used as a nucleate boiling fluid (see below), 
but the expense of setting up and safely handling the system led the 
author to abandon experimentation with this material. The author, there- 
fore, experimented with hydro- and fluorocarbons. A description of 
freezing rates attainable with these materials under a variety of condi- 
tions is given below. 

In order to measure and apply freezing rates obtained with these 
fluids, it was necessary to specify the conditions used during measure- 
ment, since, for instance, the freezing rates attainable with liquid nitro- 
gen can be increased 5-10 fold if liquid N 2 is used properly ( Cowley et 
at, 1961; Luyet, 1961; Moline and Glenner, 1964). Since similar increases 
in freezing rate might be attainable with proper use of fluids such as 
propane, a systematic program was instituted with this in mind. A resume 
of the physical principles involved is given below. 

In order to measure freezing rates, attention was focused on a model 
system rather than on measurements of biological material for two rea- 
sons. First, difficulty was encountered in constructing reproducible micro- 
thermocouples such as described by Stephenson ( 1960 ) and second, it is 
very difficult, if not impossible, to cut replicate samples of tissue such as 
liver; slight differences in geometry can cause great differences in the 
freezing rate. In addition, the possible influence of the thermocouples on 
the freezing rates of very small samples plus the great importance of the 
nature of the surface of the sample in freezing (see below) added 
further discouragement. For these reasons and because of the desire to 
evaluate a large number of freezing fluids and freezing conditions, a 
simple thermocouple ballasted with a short piece of hypodermic tubing 
was employed as our test object. 

For many of the fluids used in rapid freezing, the temperature of the 
object to be cooled exceeds the temperature to which the quenching 
fluid may be superheated, which results in bubble formation and thus 
boiling at the object surface. This is certainly true for liquids such as 
liquid nitrogen, liquid oxygen, Freon 14, Freon 13 and probably also for 
Freon 22 and propane. The formation of bubbles on a surface to be 
cooled has, however, many beneficial effects and, indeed, some of the 
highest heat transfer fluxes measured have been obtained with boiling 
fluids (Westwater, 1959). An excellent account of this phenomenon may 
be read in Westwater ( 1959 ) and a more technical account in Rohsenow 
and Choi (1961). 

Consider an object immersed in a fluid so that the temperature of the 
object may be varied. As its temperature is increased above that of the 
boiling point of the fluid, the fluid in the immediate vicinity of the object 
becomes superheated but may not become converted to a vapor, that is, 


it may not form bubbles. As the temperature continues to rise, bubbles 
start to form rapidly from certain localized defects in the object surface 
(pits) containing vapor, usually air to start with (Westwater, 1959). 
More and more sites participate in bubble formation as the temperature 
differential, AT, increases until, above a certain level, so many sites are 
involved that the bubbles coalesce to form vapor films which detach and 
reform rapidly so that the fluid is only intermittently in contact with the 
object surface. At still higher AT, a stable vapor film is formed and 
bubbles originate only at the boundary of the fluid and the stable film 
which coats the surface of the object. 

The heat transfer from the object to the fluid begins to increase 
greatly as the bubbles of fluid begin to form at the outset of the so- 
called nucleate boiling range (Westwater, 1959). The primary effect 
of bubble formation appears to be that of fluid agitation, that is, of 
bringing more cooling fluid to the surface as a result of the turbulence 
created (Rohsenow and Choi, 1961). At the point where bubbles co- 
alesce to form partial films, the heat transfer begins to go down dras- 
tically and the range of transition boiling is reached. Finally, the region 
of film boiling is attained where the efficiency of heat transfer is again 
lower, that is, heat transferred per unit temperature difference is lower. 
The point which separates nucleate boiling from transition boiling is 
known as the maximum nucleate boiling point, and at this point an 
approximate equation for the heat transfer, in the absence of gross fluid 
movement, is: 



where Q is heat removed, A is area, k is latent heat of vaporization, o- is 
surface tension, g is the acceleration of gravity (bubbles will not leave 
the object surface unless forced to do so) and p L and p Y are the liquid 
and vapor densities respectively (all measured in appropriate units). 
This equation assumes that the fluid is at its boiling point. Considerably 
greater fluxes are obtainable if the fluid is cooled below this point. 

So far, only passing mention has been made of the object surface, but 
since the set of pits, scratches, dirt, adsorbed gas, etc., will determine the 
number of bubble producing sites, it is clear that in any practical appli- 
cation of these results, the character of the surface must be carefully 

Cowley et al. ( 1961 ) obtained the very surprising results that coating 
the surface of an object with insulating material such as Vaseline or 
glycerol often greatly increases the cooling rate in liquid nitrogen. This, 
however, is not true in a fluid such as isopentane which is liquid at room 


temperature and will not, therefore, form surface vapor if an object at 
room temperature is immersed in it. The explanation appears to be that 
the insulator coat can maintain a temperature gradient between the 
object and the liquid nitrogen so that the outer surface is at or below the 
maximum nucleate boiling point, whereas its inner surface is at the tem- 
perature of the object. Since nucleate boiling will greatly increase heat 
transfer, greater cooling rates are obtained. The character and size of 
the object to be cooled determine the thickness of the layer of a given 
insulator needed for maximum heat transfer ( Luyet, 1961 ) . In addition, 
if nucleating sites in the form of powders are added to the surface, an- 
other increase in cooling rate results (Cowley et al., 1961; Luyet, 1961). 
Luyet ( 1961 ) has made an extensive study of this effect an d has obtained 
cooling rates in liquid nitrogen up to 10X those attainable with no surface 

Since liquid nitrogen is such a poor quenching fluid to start with, it 
seemed possible that other fluids might offer even greater bonuses in 
heat transfer. To this end the author performed experiments with over 
20 common hydrocarbons, fluorocarbons and mixtures as quenching fluids 
(Table 1.1) and the results are discussed below. 


X = latent heat 

Boiling point 

Melting point 

of vaporization 

















3-Methyl 1-butene 





2-Methyl 1-pentane 






























Freon 12 





Freon 13 





Freon 14 





Freon 22 





Freon 13 B-1 





Genetron 23 






Vinyl chloride 





Perfluoro propane 





"Matheson Gas Data Book (1961). 


Application of Liquid Nitrogen and Fluoro- and Hydrocarbon 
Quenching Fluids 

Our experimental system consisted of a copper-constantan thermocouple 
embedded in a 2.75 mm piece of #17 hypodermic tubing. A reference 
junction was maintained at 0°C in an ice-water bath. Cooling curves 
were recorded on Polaroid 146 L film with an oscilloscope camera at- 
tached to a Tekronix 502 oscilloscope. The latter was set to be triggered 
internally by the incoming signal, and to be sure that all cooling curves 
were comparable, the sweep was started only when the thermocouple 
reached 0°C, since a certain signal level had to be chosen so that the 
generator would not be triggered by noise or by variations in ambient 
temperature during the experimental day. In addition, the major infor- 
mation sought is the cooling velocity in the region from about 0°C to 
approximately — 79°C. The oscilloscope screen was calibrated by using 
room temperature, 0°C, -79°C (dry ice at its sublimation point) and 
— 198° C (boiling point of liquid nitrogen). 

Initially, the thermocouple was immersed in the quenching fluid 
either by hand or by dropping it in. It was found that cooling curves ob- 
tained by these methods often had at least two major rates of cooling — 
a slow initial rate and a higher later rate. It became clear that the slow 
initial cooling was due to cooling in the vapor above the quenching fluid 
and it was, therefore, necessary to traverse this region as rapidly as pos- 
sible. The simplest device found for this purpose was a rubber band 
operated rod holding the thermocouple which could be shot into the 
quenching fluid at varying rates and to different depths. This rig was 
used for all the work described below. 

The fluids used were either liquids at room temperature or were 
liquified by cooling in liquid nitrogen. Table 1.1 presents some physical 
constants for the fluids used in this study. The temperature of the fluid 
was generally held near its freezing point with liquid nitrogen and was 
recorded with a Brown recorder with a +50 to — 200 °C range. 

The insulator coatings used were glycerol, dibutyl-phthalate, Vase- 
line, Formvar and a number of oils. These substances were generally 
dissolved in an appropriate solvent to a known concentration and the 
solvent then allowed to evaporate. This controlled the thickness of the 
coating although this thickness was not known. 

Powders used as nucleating sites included copper powder, carbon 
powders (charcoals of different sources), protein powders, silica pow- 
ders, pollen grains, powdered rosin and activated alumina and molecular 
sieve, ground in a ball mill and sieved to different size classes. On the 
basis of considerable experience with liquid N 2 , Vaseline and dibutyl- 


Fig. 1.9 Oscilloscope tracings of temperature changes in a thermocouple im- 
mersed in quenching fluids under the conditions specified in the text and under 
the description of each curve. Time scale and temperature scale are indicated 
for each curve. A: Curve 1 is the cooling curve for the bare thermocouple im- 
mersed in boiling liquid nitrogen. The arrow labeled 1 points to the increase in 
cooling velocity representing the start of nucleate boiling. Curve 2 is a curve of 
the thermocouple cooling curve when the latter is coated with a thin layer of 
dibutylphthalate, DBPh (see text). Curve 3 shows the increase in cooling obtained 
when nucleating sites are added to the DBPh layer. In this case they were in 
the form of spores from the horsetail, Lycopodeum. Temperatures are indicated 
on the figure. Vertical lines are separated by 0.5 sec. B: Three curves showing 
the relative efficiencies of two different coatings in affecting cooling velocities. 
Curve 1 is in liquid nitrogen alone, curve 2 is with a coating of DBPh and curve 
3 with a coating of Vaseline (deposited from an ether solution). Time scale is 
0.5 sec between vertical lines. C: Three cooling curves showing the relative 
efficiency of two different nucleating sites attached to a DBPh coating. The time 
scale has been enlarged and the time between vertical lines is 0.1 sec. Curve 1 
is DBPh alone, curve 2 is DBPh with Lycopodeum dust and curve 3 is DBPh 
with Santacel, a silicaceous powder. D: Cooling curves with Freon 14, one with 


phthalate were used as insulators and Santacel (a silicaceous fluffy ma- 
terial) and graded Alcoa activated alumina were used as nucleating sites 
with all the rest of the quenching fluids. 

The increases in cooling velocity in liquid nitrogen with application 
of insulator coats can be seen in Fig. 1.9A. Curve 1 represents the cool- 
ing curve in liquid nitrogen with bare thermocouple. The knee (arrow) 
represents the transition to nucleate boiling with a great increase in 
cooling velocity. Curve 2 is the cooling curve obtained with a coating 
of dibutylphthalate and curve 3 with one of Vaseline. In curve 2 the 
knee is at a considerably higher temperature indicating that nucleate 
boiling has occurred much earlier with consequent increase of overall 
cooling velocity is obtained. Figure 1.9C shows the relative efficiency of 
pears to have been attained and nucleate boiling has already started as 
the cooling curve starts to be recorded. 

In Fig. 1.9B, curves 1 and 2 are similar to the corresponding ones in 
Fig. 1.9 A. Curve 3 was obtained with DBPh plus a coat of Lycopodeum 
powder as a source of nucleating sites. Again considerable increase in 
cooling velocity is obtained. Figure 1.9C shows the relative efficiency of 
two different nucleating powders applied to the same thickness of 
DBPh, with Santacel yielding a rate increase double that of Lycopodeum 

In general, our results confirm the experiences of Cowley et al. ( 1961 ) 
and Luyet (1961) with liquid nitrogen, namely, the thickness and type 
of insulator coat affect heat transfer rates to varying degrees, metal 
powders are considerably worse than nonmetallic powders as nucleating 
sites, different powders have very different abilities to effect nucleation, 
and no consistent correlation exists between similar materials (alumina 
and molecular sieve ( Linde ) ground to different sizes over a range from 
less than 74 fx, to over 177 [i. The greatest increases in cooling velocities 
(7 to 8 times greater than in liquid nitrogen alone measured at — 79°C) 
were obtained with coatings made with full strength DBPh dipped into 

an uncoated thermocouple (no coat) the second with a coating of DBPh. Space 
between two vertical lines represents 50 msec. Vertical scale (temperature) is 
expanded compared to previous three figures. E: Cooling curves in Genetron 
23 at — 151 °C. Two curves marked "no coat" indicate the reproducibility of the 
technique. Similarly the two curves marked DBPh show increased cooling rate 
with this coating and are reproductible. F: Three curves comparing cooling in 
liquid nitrogen (uncoated), liquid nitrogen with the thermocouple coated with 
DBPh and rosin (an efficient nucleating powder) and ethylene with a DBPh coal- 
ing on the thermocouple. The ethylene was at — 171°C. Measurement on a cooling 
curve recorded with a 20 msec/cm time base showed that the time to reach 
— 79°C with ethylene-DBPh was 7 times less than that needed to reach this 
temperature with the DBP-rosin coat in liquid nitrogen. 


Santacell or powdered rosin. Rather than repeat this for all coatings, 
these plus occasional use of Formvar coats and graded alumina series 
was used for all further work with the hydro -and fluoro carbons in order 
to cut the total work to manageable size. 

In addition, in the initial work with each fluid, cooling rates were ob- 
tained visually by noting the point at which the cooling curve crossed 
the — 79°C horizontal line and calculating the cooling rate from the 
known sweep rate of the oscilloscope. Final photographs were obtained 
after the initial work was completed. The values in Table 1.2 are typical 
and are taken from photographs. The conclusions, however, are based 
on considerably more experience. In particular, many more mixtures were 
tried than are indicated in the table but abandoned if, as was always the 
case, they yielded cooling rates not greater than the fastest component 


Final Velocity 

("C/sec measured 


T (ave) of fluid 











Ethane plus DBPh and activated 







Ethylene plus thin layer of DBPh 






1/3 isopentane plus 2/3 propane 









2-methyl pentane 






Freon 12 



Freon 13 



Freon 13 plus vaseline and 




Freon 13-B-1 



Freon 14 



Freon 14 plus DBPh and rosin 



Freon 22 



Genetron 23 



Genetron 23 plus DBPh 



Vinyl chloride 



Perfluoro propane 



1/2: 1/2 F13: F22 



Azeotropic F-22-propane mixture 




in the mixture. No "slushes" of isopentane with liquid nitrogen were 
made as suggested by Moline and Glenner (1964), but the experience 
with mixtures and partially frozen quenching fluids suggested that the 
addition of liquid nitrogen is likely to decrease rather than increase 
freezing rates. 

The results of application of these methods are summarized in Table 
1.2. In general, fluids such as Freon 13 and Freon 14 are poor quenching 
fluids and their ability to cool can be enhanced with coatings and 
nucleating sites as in the case of liquid nitrogen (Fig. 1.9D). In the 
range of hydrocarbons tried we have obtained no cooling rates faster 
than those gotten with liquid propane or liquid propylene. Among the 
fluorocarbons, Genetron 23 gave cooling rates about equal to that of 
propane. Freon 22 gave cooling velocities about 70% of that of propane 
and Freon 12 was further behind. No increase in cooling velocity was 
obtained with coats or nucleating sites when propane, Freon 12 or Freon 
22 was used. With ethane and Genetron 23 increases were obtained 
with coatings and nucleating sites and the maximum velocities were 
equivalent to those seen with propane ( Fig. 1.9E ) . With ethylene, coat- 
ings and nucleating powders, consistently yielded rates that were faster 
than those obtained with propane. Figure 1.9F compares ethylene with 
liquid nitrogen and facilitated liquid nitrogen (coated plus rosin). 

It is clear, therefore, that the fastest rates attainable with common 
substances and mixtures used under conditions stated above are those 
with propane, propylene or Genetron 23 and that surface modifications 
may boost attainable rates by 10-30% for Genetron 23 and ethylene. It 
seems likely that the temperature difference between a room temperature 
object and the quenching fluids in the case of propane, Freon 22, etc., 
are such that nucleate boiling already takes place at the test object 
surface so that little extra cooling capacity can be expected from the 
coatings and added nucleating sites. Coatings and added nucleating sites 
may, in fact, act to cut down some heat transfer. In the case of ethylene, 
ethane and Genetron 23, the boiling points are ~40~60°C lower than 
for the other fluids so that incipient film boiling may decrease the initial 
cooling rate. In this case, addition of coatings and nucleating sites ap- 
pears to help, but, unfortunately, no dramatic increases in heat transfer 
fluxes occur as with liquid nitrogen. However, it is quite possible that in 
the cooling of larger specimens where greater heat fluxes are necessary 
(Moline and Glenner, 1964), coatings and nucleating sites may be of 
considerable use, even, perhaps, with propane or Freon 22. Finally, the 
wide variety of freezing rates available with the substances used may 
ultimately be of use in finding optimal freezing conditions for given 


Description of Freeze-Substitution Technique 

The work discussed above was started in the hopes of attaining cooling 
velocities in cells and tissues in excess of those available by ordinary 
quenching procedures. Even though the studies were carried out with a 
model system and not with tissue as the test object, the results offer little 
hope that quenching with ordinarily available fluoro- and hydrocarbons 
can give freezing rates significantly exceeding that available with pro- 
pane. Among the considerably safer fluorocarbons, Genetron 23 (fluor- 
form) appears to offer freezing rates equal to those of propane and, 
indeed, with coatings and nucleating sites, perhaps still greater. Freon 22 
is not far behind G-23 and has one considerable advantage over the 
latter, namely, its boiling point ( —40.8° ) which is high enough so that 


5ml serum vial - — 
with liquid nitrogen 
to receive frozen 

Freon 22 
inner beaker 

Beaker support ring 
and handle 

Dewar flask 

Millers silk held 
on brass cylinder 
with rubber band 

Nested beaker 
(1000ml, 600ml 
and 450 ml are 
convenient sizes) 


1 inch #29 


3mm loop of 75 micron 
tungsten wire (3 mm wire) 
coated with 1/2 micron 
Formvar film 

Fig. 1.10A Diagram of set-up used in rapid freezing. The specimen insertion rod 
assembly (Fig. 1.11B) is mounted so that specimens are injected into the quench- 
ing fluid in the brass cyclinder. When enough of these are collected they are 
transferred by long forceps through the cold vapor above the liquid nitrogen 
(approximately — 120°C) into a serum vial containing liquid nitrogen. This is 
loosely stoppered with the stopper-hypodermic assembly described in the text 
and transferred to a freezer or dry ice box where the substituting fluid may be 
added. B: Specimen support platform for handling of specimen and rapid in- 
sertion into the quenching fluid without introduction of a large warm mass (such 
as a forceps). 


it can easily be stored in a deep-freezer or dry ice box and may be used 
over and over again, after filtration at low temperatures through ordinary 
filter paper. In the author's laboratory, it is the quenching fluid of choice. 

It has been noted that the depth to which the thermocouple is im- 
mersed in the fluid influences the cooling rate so that this rate increases 
with penetration up to a certain distance. The presumed reason for this 
is that turbulence is created which increases freezing rates and also that 
the thermocouple moves to liquid which has not been warmed in its 
passage through the medium. 

On the basis of the above results, a technique for rapidly cooling 
tissue was developed which offers ease and flexibility in handling tissue 
and maximum cooling rates available with common cooling media. The 
basic set-up is as follows and is illustrated in Fig. 1.10A. 

Quenching Fluid 

The quenching fluid (usually Freon 22) is obtained as described in the 
Appendix and is poured into a beaker nested in two other larger beakers, 
the whole supported in liquid nitrogen in a Dewar flask. This arrange- 
ment gives enough heat loss so that Freon does not reach the liquid 
nitrogen temperatures and usually remains liquid between —150 and 
— 160°C, which is its freezing point. 

Collecting Cylinder 

A brass cylinder ~ 1.5 in in diameter and 2.5 in long is cooled in liquid 
nitrogen and supported in the quenching fluid. The bottom of the cylin- 
der is closed by a piece of #25 Millers silk (held on with string or with 
a rubber band) so that fluid can freely penetrate. A slowly driven 
Archimedes screw ( Bell, 1956 ) is used to circulate cooling fluid and pre- 
vent a temperature gradient, which otherwise will rapidly form (Fig. 

Specimen Preparation 

Tissue and cell suspensions must be rapidly injected into the quenching 
fluid in such a way as to avoid accidental evaporation of water, unless 
such evaporation is desired. In order to accomplish both goals, special 
tissue supports are utilized (Fig. 1.10B). One inch pieces of #29 hypo- 
dermic tube (80 /a inside diameter) are broken from longer pieces by 
snapping scored tubing as one would in breaking glass tubing; if the 
pieces are cut by other means, the lumen is generally compressed. A loop 


of 3 mil tungsten wire (75 fi in diameter) ~3 mm in diameter is inserted 
into a piece of the hypodermic tubing which is then compressed on the 
wire to hold it in place. Pieces of Formvar film ~0.5 /a thick are used to 
coat the loops and act as the specimen support platform. Specimens are 
placed on the film either in a moist chamber (Gersh, 1956) or under the 
surface of a solution of freeze protective agent, and are then coated with 
a second Formvar film to prevent evaporation. This is accomplished 
by transferring a film from a split copper loop onto the specimen con- 
taining platform loop (Fig. 1.11A). 


Split copper loop 
with coating film 





Rubber band for 
propelling specimen 
in quencing fluid 

support rod 

Pin to hold support 
rod(rod propelled 
on removal of pin) 

Rubber band for retracting 
specimen support rod 

Retaining screw for 
hypodermic needle 

Specimen support platform 

Fig. 1.11A Technique for coating a specimen on the support platform so as to 
prevent accidental drying during handling. A: Formvar film is transferred onto the 
specimen from a U-shaped copper loop. This is usually done in a moist chamber 
when tissue is frozen directly from the animal with no pretreatment. B: Appa- 
ratus used for rapid insertion of the mounted specimen. The mounting platform 
is inserted into a hypodermic needle mounted in the end of a hollow rod. Upon 
removal of the retaining pin, a rubber band shoots the rod toward the brass 
cylinder in Fig. 1.1 OA and inserts the specimen platform an inch or more below 
the surface of the quenching fluid. The momentum will cause the platform to 
continue into the fluid. At the same time the retracting rubber band is stretched 
and pulls the specimen support rod out of the fluid. 


Specimen Freezing 

The specimen and its support assembly are now inserted into a rod 
activated by rubber bands (similar to that used in the thermocouple 
work) which shoots the specimen into the quenching fluid and then 
rapidly withdraws (Fig. 1.11B). Similar supports were used by Mack- 
enzie and Luyet ( 1962b ) in their work on rapid freezing of gelatin films. 

Removal ot Quenched Specimens 

Once the requisite number of specimens has been quenched, the brass 
cylinder with cloth bottom into which it has been shot, may be lifted 
into the upper part of the Dewar flask and drained of Freon in the 
process. The specimens are then transferred into 5 ml serum bottles con- 
taining liquid nitrogen, which are suspended in the Dewar. The transfer 
is facilitated by a double forceps which allows the transfer at a distance 
and thus aides in prevention of frostbitten fingers (even so, lambskin 
gloves, similar to those used in cryostats, considerably increase the com- 
fort of these experiments ) . 

Transfer into Freezer 

The vials are now transferred to the freezer or dry ice chest and capped 
with rubber serum caps ( one should not linger long over a dry ice chest 
if fatal effects are not to interrupt the experiment). In order to permit 
egress of the vaporized nitrogen and transfer of the substituting fluid, 
a 2.5 in #20 gauge hypodermic needle is inserted through the thin 
rubber septum of the cap. No success will be obtained if a venting needle 
is omitted and for this we use a #25 guage 0.5 in needle. 

Substitution Fluid 

Storage of the substituting fluid is most convenient in 20 ml serum 
bottles using the same configuration as for the specimen bottles except 
that 4.5 in #20 gauge needles are used because of the length of the vial. 
In all cases, small corks are tightly inserted into the needles when fluid 
transfer is not in progress. This prevents accidental frost falling into the 
vial via the neeedle and, if a dry-ice chest is used, solution of C0 2 in the 
substituting fluid. The latter can have exasperating effects on the process 
of fluid transfer from one vial to the other. A small quantity of Vaseline 
or stopcock grease will aid in sealing the rubber stoppers to the vials. 
The transfer is accomplished by ordinary syringes with glass tips (steel 
lock tips are almost impossible to keep clear of some frost which acts 


like glue at low temperatures), precooled and kept in the substituting 


If osmium tetroxide is used in the substituting fluid, it must be dissolved 
at low temperatures, otherwise it will almost immediately form a black 
precipitate. To accomplish this, it is placed in an empty serum storage 
vial at room temperature, cooled to the desired temperature and then 
cold substituting fluid is inserted in the cold environment with glass tip 
syringes. The osmium tetroxide readily dissolves in acetone, methanol, 
ethanol, tetrahydrofuran and presumably most other potential substi- 
tuting fluids, and forms a yellowish solution stable for many months if 
kept in the cold (e.g., -70°C). 

Final Substitution 

The tissues are allowed to substitute for about two weeks and then the 
substitution fluid is changed (without osmium tetroxide if it has been 
used) just before raising the temperature of the vials. If osmium 
tetroxide has been used in the substitution fluid, the specimen will turn 
light brown in color; the substituting fluid does not alter color. 


The vials are allowed to reach room temperature directly without a step- 
wise increase in temperature. Van Harreveld et al. (1965), however, 
prefer a step-wise increase in temperature. When room temperature has 
been attained the tissue samples are treated as ordinary dehydrated 
specimens and are embedded in plastic. The micrographs in this chapter 
were obtained with the technique herein described. 

Description of Freeze-Drying Technique 

A second method (historically, the first) for the dehydration of frozen 
tissue consists of the vacuum sublimation of water vapor at low tem- 
peratures. Theoretical discussions of freeze-drying have been given by 
Meryman (1960), Rey (1960), Rowe (1960) and Stephenson (1960) 
and these papers should be consulted for design characteristics of 
equipment used in the technique. Basically, frozen tissue is transferred 
into a vacuum chamber which is maintained at the drying temperature. 
This temperature should be below that at which devitrification phenom- 


enon becomes important in frozen tissues; however, this temperature is 
not known with any degree of certainty. In the absence of exact knowl- 
edge, the drying temperature should be as low as is practical. This tem- 
perature is partly determined by the patience of the investigator since 
the vapor pressure of ice decreases very rapidly below temperatures of 
— 70° C, and it is essentially the removal of water vapor sublimated from 
the specimen which constitutes drying. 

According to Sjtistrand (1967) drying at -100°C for 7-10 days fol- 
lowed by a slow temperature change over a 2-3 day period to — 79°C 
and drying at this temperature for several more days is sufficient. 
Malhotra (1968) dries at -79°C for 7 days followed by several days at 
higher temperature, and Bondareff (1967) and Gersh et al. (1960) sug- 
gest drying at -45° C for a maximum of 48 hr. Clearly the user has a 
wide choice of drying times and temperatures and, to quote one of the 
practioners in the field, "the satisfactory temperature can only be de- 
termined by actual repeated experiments in connection with electron 
microscopic analysis at high resolution" (Sjostrand, 1967). 

The pressure in the freeze-drying apparatus should be such that the 
mean free path of residual gas is greater than the distance from the 
specimen to a cold finger in the instrument (if one is incorporated) in 
order to allow the most rapid access of vapor molecules to the trapping 
surface or the pump (Sjostrand, 1967; Stephenson, 1960). After the first 
water layers are removed, the rate of subsequent water removal is de- 
pendent upon the speed of water transport through the layers of the 
tissue which have already been dried. This transport is much slower 
than that from the specimen surface (Stephenson, 1960), and thus re- 
moval of water and thus from the specimen is less dependent upon the 
speed of transport from the specimen surface to the pump or a cold trap. 
Thus, the actual drying pressures are probably not very critical. How- 
ever, pressures of the order of 10- 3 Torr (Sjostrand, 1967; Malhotra, 
1968) or lower (Nagata et al, 1969) are generally used, which necessi- 
tates utilization of a diffusion pumped system. 

Two disadvantages of the freeze-drying technique must be brought 
to the attention of the prospective user. Because the majority of the 
freeze-protective agents in use for morphological studies are not as 
volatile as water, they cannot be used to protect against ice-crystal 
damage. During vacuum dehydration the tissue will progressively be im- 
pregnated with higher and higher concentrations of the agent, ultimately 
resulting in a freeze-substitution by the protective agent. In addition, 
since vacuum-dried tissue is not fixed, the introduction of artifacts 
especially during embedding is almost unavoidable. Considerable caution 
must therefore be taken in the use and interpretation of results obtained 


from freeze-dried specimens. Freeze-drying, perhaps more than freeze- 
substitution, is an experimental technique. 

The number and types of freeze-drying equipment are large and the 
majority of them appear to be custom made according to the individual's 
experience and bias. In the following discussion it is assumed that the 
equipment available to the user has these characteristics: (1) it can 
attain vacuum of the order of 10 -3 Torr or better; and ( 2 ) it has a side 
arm or other device ( Nagata et al., 1969 ) which may be closed off from 
and opened to the main drying chamber so that vapor fixatives or em- 
bedding medium may be introduced subsequent to vacuum drying. 
Cytochemical studies using vapor reagents which are introduced into the 
drying chamber via special side arms have been carried out by Gersh 
(1964) and Gersh et al, (1960). 

Specimen Freezing 

The same fundamental problems of specimen freezing so as to avoid ice 
crystal formation apply to freeze-drying and freeze-substitution. The same 
precautions as outlined for freeze-substitution apply here. Tissues must 
be obtained rapidly from animals, cut into very small samples, protected 
from evaporation (unless evaporation is specifically to be investigated) 
and frozen in the shortest possible interval after removal from the animal. 
In case tissues have been protected against accidental evaporation by 
means of Formvar films, the films must be removed as thoroughly as pos- 
sible since they will clearly act as an impediment to vacuum dehydration 
at low temperatures. In order to accomplish this, serum vials containing 
frozen specimens on support platforms are transferred to a shallow flask 
partially filled with liquid nitrogen. The support platforms are lifted out 
into the Dewar and specimens are removed with long forceps under 
liquid nitrogen. Much of the support film may be removed in this 
manner although it is not possible to consistently remove all of it. In 
most freezing techniques, protective films are not used so this problem 
is not encountered. 

Preparation of Drying Chamber 

The frozen tissue must be transferred into a cooled drying chamber so 
as to prevent its temperature from rising and to prevent condensation of 
vapor during transfer. The temperature of drying is at the discretion of 
the worker; however, — 79° C is recommended. The drying chamber 
should be evacuated and then immersed in a mixture of dry ice and 
acetone (or in a refrigerator if that is used in the equipment) and 


cooled for several minutes. Dry air may then be admitted to the drying 
chamber after the chamber is removed from the vacuum system, and it 
may then be covered to prevent moisture condensation during transfer 
of the specimen. Further preparation of the chamber is necessary if an 
embedding medium is to be introduced prior to specimen drying. This 
will be described below. 

Transfer of Frozen Tissue to Drying Chamber 

The tissue must be transferred to the drying chamber in such a way that 
it remains at a very low temperature and no frost condenses on it. Sev- 
eral methods have been devised to accomplish this. A simple method 
has been described by Sjostrand (1967) and involves transfer under 
liquid nitrogen. The frozen tissue as prepared above is transferred under 
liquid nitrogen into a small brass cylinder which has fine wire mesh on 
one face and which can be lifted and manipulated by a nylon thread 
attached at the other face. The wire mesh will allow drainage of the 
liquid nitrogen after transfer into the drying chamber. In order to keep 
the tissue under liquid nitrogen during transfer, the cylinder is placed 
into a brass cylinder to which a long rigid handle is attached. The tissue 
is thus located on the screen of the specimen-handling cylinder which is 
in the transfer beaker and under liquid nitrogen. 

The entire assembly is now lifted out of the Dewar in which manipu- 
lations have been taking place and quickly transferred into the dry 
atmosphere of the precooled drying chamber. The nylon handle of the 
specimen handling cylinder is used to lift this from the transfer beaker, 
simultaneously draining off the liquid nitrogen. The transfer beaker is 
now removed, the drying chamber reattached to the freeze dry apparatus, 
and the chamber is pumped maintaining its temperature with the dry 
ice-acetone mixture or with whatever temperature control system is used. 
Drying is continued according to the regime selected by the investigator. 


In many cases the dried tissue is fixed prior to embedding. This may be 
done by introducing osmium tetroxide vapors through the side arm of 
the drying chamber and is usually accomplished after disconnecting the 
chamber from the vacuum system either physically or by means of a 
valve. Fixation may be carried out either at low temperatures or after 
return of the apparatus to room temperature, and this can be accom- 
plished at low pressure or at atmospheric pressure. Formaldehyde, 
glutaraldehyde, alcohol or any other vapor fixative may be used for this 


purpose. Malhotra (1968) used osmium tetroxide at atmospheric pres- 
sure for 24 hr. Bondareff (1967) used osmium tetroxide at reduced pres- 
sure for ~lhr. The latter worker also postfixed the specimen with 
formaldehyde vapors by introducing paraformaldehyde crystals into the 
drying chamber under vacuum (by means of a side arm) and heating 
the whole chamber ( under vacuum but isolated from the pump ) at 80°C 
for 1 hr. This temperature is above the point at which the formaldehyde 
polymer breaks down to liberate formaldehyde vapors. 


If fixation is used, tissues may be removed from the drying chamber and 
embedded in the same manner as chemically fixed tissues. Unfixed tissues 
are harder to handle since they are highly hygroscopic. Two methods 
have been used. In the first method the embedding medium is degassed 
under vacuum and then poured into the drying chamber under vacuum 
and at temperatures selected by the investigator. The details will depend 
upon the type of side arm assembly available on the drying chamber. 
The tissue will sink in the medium as it becomes impregnated and may 
then be removed and finally embedded in the normal manner. 

In the second method the introduction of the embedding medium 
takes place prior to transfer of the specimen to drying chamber (p. 39 ) . 
In preparing the drying chamber, embedding medium is introduced and 
degassed under vacuum. After bubbling of the medium has become 
minimal, the chamber is cooled to the chosen dehydration temperature 
and specimens are introduced onto the frozen surface of the embedding 
medium. After vacuum sublimation of the water from the specimen has 
occurred, the drying chamber is raised above the melting point of the 
mixture and specimen impregnation takes place in vacuo at the desired 
temperature. After impregnation, the specimen is removed and embedded 
in the usual manner. This technique involves the least amount of ma- 
nipulation of the specimen once it is introduced into the chamber. How- 
ever, if vapor staining or fixation is to be used, it must be carried out 
below the melting point of the embedding medium. 


The success of freezing and thawing in the preservation of cells suggests 
that it may be a technique of some use in electron microscopy. It was 
indicated earlier that cells may be frozen slowly enough so that cellular 
water may pass through the membrane to join extracellular ice or that 
the freezing rate may be rapid enough to cause intracellular ice forma- 
tion. In the first case the cells are subjected to hypertonic conditions. 


Cells may survive these conditions if they are vitrified before significant 
mebrane damage occurs due to increased salt concentration. In the sec- 
ond case, survival rates are generally poor, although not in all cases 
(Sherman and Kim, 1967). 

It was also indicated earlier that penetrating cryoprotective agents 
appear to protect cells on the slow freezing rate side of the optimal 
survival point. Some ultrastructural work has been carried out on frozen 
thawed cells frozen in both conditions. Like the majority of the results 
in freezing studies, the specific effects vary according to the tissue. Baker 
(1962) froze mouse kidney in isopentane and then thawed cryostat sec- 
tions of this material in osmium tetroxide solutions at 4°C. He also 
"froze" tissue at — 18° C and again fixed thawed cryostat slices in osmium 
tetroxide at 4°C. He found that the isopentane frozen material had some 
mitochondrial swelling, blebbing of the outer nuclear membrane and 
some chromatin condensation, whereas the slowly frozen material looked 
very much like control tissue. Similar results were found with mouse 
liver by Trump et al. (1964). They froze large samples of liver slowly 
to — 79°C, thawed for 7 min at room temperature and 10 min at 0-4°C 
and then fixed in buffered osmium tetroxide. 

Comparison with isopentane- or propane-frozen smaller samples 
thawed by the same regime showed considerably greater damage in the 
cyptoplasm of faster frozen cells. The reverse, however, was seen in 
damage to the nucleus (Stowell et al, 1965). It is clear, therefore, that 
while the technique may be promising, much work must be done on 
optimal tissue size, freezing regime, thawing regime and fixation pro- 
cedure. What is remarkable is the degree of restoration of cellular struc- 
ture after ice crystal damage. It is possible that this technique may be 
useful in the study of reformation of cytoplasmic structures after variable 
degrees of dissociation caused by ice crystals of different sizes. 


Obtaining Cold Quenching Fluid 

Quenching fluids such as isopentane which are liquid at room tempera- 
ture may be poured into a beaker of the desired size and cooled directly 
while stirring by supporting the beaker in a Dewar flask containing 
liquid nitrogen. Certain gases such as Freon 12 or 22 may be purchased 
in small containers (ca. 1 pint) at local refrigeration supply houses and 
can be fitted with a valved spout. The cans may be cooled by partial 
immersion in liquid nitrogen and the liquid may then be poured into the 
assembly shown in Fig. 1.10A. Propane may be obtained from small 
propane torch tanks in a similar manner. Propane, however, is not rec- 


ommended as a quenching fluid for the beginner because of the possible 
danger of explosive mixtures forming. Propane will supercool below the 
condensation point for liquid oxygen, which is not recognized unless a 
relatively accurate thermometer is used. Such mixtures can be ignited 
even at very low temperatures by certain organic compounds or by high 
frequency sound. In addition, disposing of the propane after use is not 
the job for a novice. It is dangerous to store liquid propane in the types 
of containers available in biological laboratories. 

For other quenching fluids obtained from larger storage tanks, the 
following method has been found convenient though not rapid. The 
control valve outlet to the tank is fitted with ordinary vacuum pressure 
tubing. A Pasteur pippette is attached to the other end and is forced into 
a cork stopper which fits to a flask. The flask volume should be larger 
than the volume of quenching fluid desired (the author usually uses 
~ 300 ml of quenching fluid). A venting needle of at least 18 gauge is 
also inserted through the cork. The flask is immersed in liquid nitrogen 
and cooled until only mild bubbling occurs. The cooling and later heat 
transfer may be considerably enhanced by coating the flask with a layer 
of glycerine to which a nucleating powder is added. 

Gas is allowed to flow in at such a rate that only a small amount is 
emitted from the venting needle; this is recognized by the vapor con- 
densation it causes. As liquid forms in the flask, it aids in the condensa- 
tion of further gas and thus the process is speeded up. Approximately 
1 hr is required to obtain 300-400 ml of liquid by employing this pro- 

The author generally stores and reuses nonexplosive quenching fluids. 
They may be filtered in a freezer below their condensation point with 
ordinary filters and funnels, and can be stored in Pyrex reagent bottles 
with glass stoppers in the chamber used for substitution. 


The work described in this chapter was supported at various times by The 
National Institutes of Health, The National Science Foundation, and The 
American Cancer Society. 


Allen, L. C, and Kollman, P. A. (1970). A theory of anomalous water. 
Science, 167, 1443. 


Asahina, E. (1965). Freezing process and injury in isolated cells. Fed. Proc, 

24 (Suppl. 15), 183. 
Asahina, E., Shimada, K., and Hisada, J. (1970). A stable state of frozen 

protoplasm with invisible intracellular ice crystals obtained by rapid cooling. 

Exp. Cell Res., 59, 349. 
Baker, J. R. (1960). Cytological Techniques. Methuen and Col. Ltd., London. 
Baker, R. F. (1962). Freeze-thawing as a preparatory technique for electron 

microscopy. /. Vltrastruct. Res., 7, 173. 
Barer, R. F. (1965). Phase Contrast And Interference Microscopy In Cytology. 

In: Physical Techniques In Biological Research (Oster, G. and Pollister, 

A. W., eds.) Vol. 3, p. 29. Academic Press, New York. 
Bell, L. G. E. (1956). Freeze-drying. In: Physical Techniques In Biological 

Research (Oster, G. and Pollister, A. W., eds.) Vol. 3, p. 1. Academic 

Press, New York. 
Berendsen, H. J. C. (1966). Water structure in biological systems. Fed. Proc, 

25, 971. 
Berendsen, H. J. C, and Migchelsen, C. (1965). Hydration structure of 

fibrous macromolecules. Ann. N.Y. Acad. Sci., 125, 365. 
. (1966). Hydration structure of collagen and influence of salts. Fed. 

Proc, 25, 998. 
Bererhi, A., and Malkani, K. (1970). Etude des Cellule Hela en Culture par 

le Cryodecapage. /. Ultrastruct. Res., 32, 23. 
Bernhard, W., and Leduc, E. H. (1967). Ultrathin frozen sections. I. Methods 

and ultrastructural preservation. /. Cell Biol, 34, 757. 
Bondareff, W. (1967). Demonstration of an intercellular substance in mouse 

cerebral cortex. Z. Zellforsch., 81, 366. 
Bullivant, S. (1965). Freeze-substitution and supporting techniques. Lab. 

Invest., 14, 1178. 
. (1970). Present Status Of Freezing Techniques. In: Some Biological 

Techniques In Electron Microscopy (Parsons, D. F. ed.) p. 101. Academic 

Press, New York. 
Burstone, M. S. (1969). Cryobiology Techniques In Histochemistry, Including 

Freeze-Drying and Cryostat Techniques. In: Physical Techniques In Bio- 
logical Research (Pollister, A. W., ed.), Vol. 3(c), p. 1. Academic Press, 

New York. 
Christensen, A. K. (1969). A Way To Prepare Frozen Thin Sections Of Fresh 

Tissue For Electron Microscopy. In: Autoradiography Of Diffusable Sub- 
stances (Roth, L. J., and Stumpf, W. E., eds.) p. 349. Academic Press, New 

Cope, F. W. (1969). Nuclear and magnetic resonance evidence using D 2 

for structured water in muscle and brain. Biophys. J., 9, 303. 
Cowley, C. W., Timson, W. J., and Sawdye, T. A. (1961). Ultra rapid cooling 

techniques in the freezing of biological materials. Biodynamica, 8, 317. 
Dehl, R. E. (1970). Collagen: mobile water content of frozen fibers. Science, 

170, 738. 


Dehl, R. E., and Hoeve, C. A. J. (1969). Broad-line NMR study of H 2 and 
D 2 in collagen fibers. ]. Chem. Phys., 50, 3245. 

Delsemme, A. H., and Wenger, A. (1970). Superdense water ice. Science, 
167, 44. 

Derjaguin, B. V. (1970). Superdense water. Sci. Amer., 223, 52. 

Dick, D. A. T. (1966). Cell Water. Butterworth's, Washington, D.C. 

Doebbler, G. F. (1966). Cryoprotective compounds: review and discussion of 
structure and function. Cryobiohgy, 3, 2. 

Dowell, L. G., Moline, S. W., and Rinfret, A. P. (1962). A low temperature 
X-ray diffraction study of ice structures formed in aqueous gelatin gels. 
Biophys. Bioch. Acta, 59, 158. 

Dowell, L. G., and Rinfret, A. P. (1960). Low temperature forms of ice as 
studied by X-ray diffraction. Nature, 188, 1144. 

Eckert, H. (1969). Distribution Of Soluble Substances At The Ultrastructural 
Level. In: Autoradiography Of Diffusable Substances (Roth, L. J., and 
Stumpf, W. E., eds.) p. 321 Academic Press, New York. 

Eisenberg, D., and Kautzman, W. (1969). The Structure And Properties Of 
Water. Oxford Univ. Press, Oxford, England. 

Farrant, J. (1970). Mechanism Of Injury And Protection In Living Cells And 
Tissues At Low Temperatures. In: Current Trends In Cryobiology (Smith, 
A. U., ed. ) p. 139. Plenum Press, New York and London. 

Farrant, J., and Woolgar, A. E. (1970). Possible Relationships Between The 
Physical Properties Of Solutions And Cell Damage During Freezing. In: The 
Frozen Cell. A Ciba Foundation Symposium (Wolstenholme, G. E. W., and 
O'Conner, M., eds.) p. 51. J. and A. Churchill, London. 

Fernandez-Moran, H. (1960). Low-temperatures preparation techniques for 
electron microscopy of biological specimens based on rapid freezing with 
liquid helium II. Am. N.Y. Acad. Sci., 85,(2), 689. 

. ( 1961 ) . Lamellar Systems In Myelin And Photoreceptors As Revealed 

By High Resolution Electron Microscopy. In: Macromolecular Complexes 
(Edds, M. V., ed.) p. 113. Ronald Press, New York. 

Fletcher, N. H. (1970). The Chemical Physics Of Ice. Cambridge University 
Press, Cambridge, England. 

Frank, H. S. (1970). The structure of ordinary water. Science, 169, 635. 

Frank, H. S., and Wen, W. Y. (1957). Structural aspects of ion-solvent inter- 
action in aqueous solutions: a suggested picture of water structure. Disc. 
Farraday Soc, 24, 133. 

Gersh, I. (1956). The preparation of frozen-dried tissue for electron micros- 
copy. /. Biophys. Biochem. Cytol., 2, 37. 

. (1964). A cytochemical method for fixation and staining of lipids in 

frozen-dried mouse pancreas for study with light and electron microscopes. 
Histochemie, 4, 322. 

Gersh, I., Isenberg, I., Stephenson, J. L., and Bondareff, W. (1956). Sub- 
microscopic structure of frozen-dried liver specifically stained for electron 
microscopy. Anat. Rec, 128, 91. 


Gersh, I., Vergara, J., and Rossi, G. (1960). Use of anhydrous vapors in post- 
fixation and in staining of reactive groups of proteins in frozen-dried speci- 
mens for electron microscopic studies. Anat. Rec, 138, 445. 

Hanzon, V., and Hermodssen, L. H. (1960). Freeze-drying of tissues for light 
and electron microscopy. /. Ultrastruct. Res., 4, 332. 

Harris, P. (1962). The effect of mercaptoethanol on the mitotic spindle. Proc. 
5th. Intern. Cong. Electron Micros., 2, NN-1, (Breese, S.S., ed.) Academic 
Press, New York. 

Hayat, M. A. (1970). Principles And Techniques Of Electron Microscopy: 
Biological Applications, Vol. 1. Van Nostrand Reinhold Company, New 

. (1972). Basic Electron Microscopy Techniques. Van Nostrand Rein- 
hold Company, New York. 

Hepler, P. K., and Jackson, W. T. (1969). Isopropyl IV-phenylcarbamate 
affects spindle orientation in dividing cells of Haemanthis katherinae Baker. 
/. Cell Sci., 5, 727. 

Horowitz, S., and Fenichel, I. R. (1965). Diffusion and transport of organic 
nonelectrolytes in cells. Ann. N.Y. Acad. Sci., 125, 572. 

Kavanau, J. L. (1964). Water And Solute- Water Interactions. Holden-Day 
Inc., San Francisco. 

Kuntz, I. D., Brassfield, T. S., Law, G. D., and Purcell, G. V. (1969). Hydra- 
tion of macromolecules. Science, 163, 1329. 

Kurtin, S. L., Mead, C. A., Mueller, W. A., Kurtin, B. C, and Wolf, E. D. 
(1970). "Polywater": a hydrosol? Science, 167, 1720. 

Leduc, E., Bemhard, W., Holt, S. J., and Tranzer, J. P. (1967). Ultrathin 
frozen sections. II. Demonstration of enzymic activity. /. Cell Biol, 34, 773. 

Leibo, S. P., Farrant, J., Mazur, P., Hanna, M. G., and Smith, L. H. (1970). 
Effects of freezing on marrow stem cell suspensions: interactions of cooling 
and warming rates in the presence of PVP, sucrose or glycerol. Cryobiology, 
6, 315. 

Levitt, J., and Dear, J. (1970). The Role Of Membrane Proteins In Freezing 
Injury And Resistance. In: The Frozen Cell. A CIBA Foundation Sym- 
posium (Wolstenholme, G. E. W., and O'Connor, M., eds.) p. 149. J. and A. 
Churchill, London. 

Ling, G. (1965). The physical state of water in living cell and model systems. 
Ann. N.Y. Acad. Sci., 125, 401. 

Lovelock, J. E. (1953). The haemolysis of human red blood cells by freezing 
and thawing. Biochim. Biophys. Acta, 10, 414. 

Luyet, B. J. ( 1961 ) . A method for increasing the cooling rate in refrigeration 
by immersion in liquid nitrogen or in other boiling baths. Biodynamica, 8, 

. (1966). Anatomy Of The Freezing Process In Physical Systems. In: 

Cryobiology (Meryman, H. T., ed.) p. 115. Academic Press, New York. . 

. (1970). Physical Changes Occuring In Frozen Solutions During Re- 
warming And Melting. In: The Frozen Cell. A CIBA Foundation Symposium 


(Wolstenholme, G. E. W., and O'Connor, M., eds.) p. 27. J. and A. 

Churchill, London. 
MacKenzie, A. P. (1965). Factors affecting the mechanism of transformation 

of ice into water vapor in the freeze-drying process. Ann. N.Y. Acad. Sci., 

125, 522. 
MacKenzie, A. P., and Luyet, B. J. (1962a). Electron microscope structure of 

very rapidly frozen gelatin solutions. Biodynamica, 9, 47. 
. (1962b). A collodion sandwich film technique for the study of the 

growth of ice in very thin layers of aqueous solutions. Proc. 5th. Intern. 

Cong. Electron Micros., (Breese, S. S., ed.) 2, 2. Academic Press, New 

Malhotra, S. K. (1968). Freeze-Substitution And Freeze-Drying In Electron 

Microscopy. In: The Interpretation Of Cell Structure ( McGee-Russell, S. M., 

and Ross, K. F. A., eds. ) . Edward Arnold, Ltd., London. 
Mazur, P. (1966). Physical And Chemical Basis Of Injury In Single-Celled 

Microorganisms Subjected To Freezing And Thawing. In: Cryobiology 

(Meryman, H. T., ed.) p. 214. Academic Press, New York. 
. (1970). Cryobiology: the freezing of biological systems. Science, 168, 

Mazur, P., Leibo, S. P., Farrant, J., Chu, E. H. V., Hanna, M. G., and Smith, 

L. H. (1970). Interactions Of Cooling Rate, Warming Rate And Protective 

Additive On The Survival Of Frozen Mammalian Cells. In: The Frozen Cell. 

A CIBA Foundation Symposium (Wolstenholme, G. E. W., and O'Connor, 

M., eds.) p. 69. J. and A. Churchill, London. 
Meryman, H. T. (1959). Sublimation freeze-drying without vacuum. Science, 

130, 628. 
. (1966). Review Of Biological Freezing. In: Cryobiology (Meryman, 

H. T., ed.) p. 2. Academic Press, New York. 
. (1969). Aspects Of Freezing Injury Relevant To Autoradiography. In: 

Autoradiography Of Diffusible Substances (Roth, L. J., and Stumpf, W. E., 

eds.) p. 201. Academic Press, New York. 

(1970). The Exceeding Of A Minimum Tolerable Cell Volume In 

Hypertonic Suspension As A Cause Of Freezing Injury. In: The Frozen 

Cell. A CIBA Foundation Symposium (Wolstenholme, G. E. W., and 

O'Connor, M., eds.) p. 51. J. and A. Churchill, London. 
Moline, S. W., and Glenner, G. G. (1964). Ultrarapid tissue freezing in liquid 

nitrogen. J. Histochem. Cytochem., 12, 777. 
Monroe, R. G., Gamble, W. J., La Farge, C. G., Gamboa, R., Morgan, C. L., 

Rosenthal, A., and Bullivant, S. (1968). Myocardial ultrastructure in 

systole and diastole using ballistic cryofixation. /. Ultrastruct. Res., 22, 22. 
Moor, H. (1964). Die Gefrier-Fixation lebender Zellen und ihre anwendung 

in der Electron mikroskopie. Z. Zellforsch., 62, 546. 

. (1969). Freeze-etching. Int. Rev. Cytol., 25, 391. 

Nagata, T., Nawa, T., and Yokota, S. (1969). A new technique for electron 


microscope dry mounting radioautography of soluble compounds. Histo- 
chemie, 18, 241. 

Narten, A. H., and Levy, H. A. (1969). Observed diffraction pattern and pro- 
posed models of liquid water. Science, 165, 447. 

Nash, T. (1966). Chemical Constitution And Physical Properties Of Com- 
pounds Able To Protect Living Cells Against Damage Due To Freezing And 
Thawing. In: Cryobiology (Meryman, H. T., ed.) p. 179. Academic Press, 
New York. 

Nevo, A. C. (1966). Experimental study of H-bonding in the achromatic 
apparatus and in the chromosomes. 2nd. Inter. Cong. Biophysics, Vienna, 
Abstr. #274. 

Olmstead, E. G. (1966). Mammalian Cell Water. Lea and Febiger, Phila- 

Page, T. F., Jakobsen, R. J., and Lippincott, E. R. (1970). Polywater: proton 
nuclear magnetic resonance spectrum. Science, 167, 51. 

Pearse, A. G. E. (1968). Histochemistry. Theoretical And Applied. Little, 
Brown and Co., Boston. 

Pease, D. C. (1966a). The preservation of unfixed cytological detail by dehy- 
dration with "inert" agents. J. Ultrastruct. Res., 14, 356. 

. (1966b). Anhydrous ultrathin sectioning and staining for electron 

microscopy. /. Ultrastruct. Res., 14, 379. 

. (1967a). Eutectic ethylene glycol and pure propylene glycol as sub- 
stituting media for the dehydration of frozen tissue. /. Ultrastruct. Res., 21, 

. (1967b). The preservation of tissue fine structure during rapid freezing. 

J. Ultrastruct. Res., 21, 89. 

Rapatz, G. L., Menz, L. J., and Luyet, B. J. (1966). Anatomy Of The Freezing 
Process In Biological Materials. In: Cryobiology (Meryman, H. T., ed.) p. 
139. Academic Press, New York. 

Rapideau, S. W., and Florin, A. E. (1970). Anomalous water: characterization 
by physical methods. Science, 169, 48. 

Rebhun, L. I. (1961). Application of freeze-substitution to electron microscope 
studies of invertebrate oocytes. /. Biophys. Biochem. Cytol., 9, 785. 

. (1965). Freeze-substitution as a function of water concentration in cells. 

Fed. Proc, 24, (Suppl. 15), 217. 

Rebhun, L. I., and Bernstein, L. (1967). In vivo sperm aster augmentation by 
hexylene glycol-independence of protein synthesis. J. Cell Biol, 35, 111A. 

Rebhun, L. I., and Sander, G. (1971a). Electron microscope studies of frozen- 
substituted marine eggs. I. Conditions for avoidance of intracellular ice 
crystallization. Am. J. Anat., 130, 1. 

. (1971b). Electron microscope studies of frozen-substituted marine eggs. 

II. Morphology of ice crystal-free unfertilized eggs. Am. J. Anat., 130, 17. 
. (1971c). Electron microscope studies of frozen-substituted marine eggs. 

III. Structure of mitotic apparatus of the first meiotic division. Am. J. Anat., 
130, 35. 



Rebhun, L. I., and Sawada, N. (1969). Augmentation and dispersion of the 
in vivo mitotic apparatus of living marine eggs. Protoplasma, 68, 1. 

Rey, L. (1960). Theorie de la lyophilisation. II. Les diverses phases de la 
cryodessication. In: Traite de Lyophilisation (Rey, L., ed.). p. 39. Hermann, 
Saint-German, Paris, VI. 

Rohsenow, W. M., and Choi, H. Y. (1961). Heat, Mass And Momentum 
Transfer. Prentice-Hall, Inc., Englewood Cliffs, N.J. 

Rousseau, D. I., and Porto, S. P. S. (1970). Polywater: polymer or artifact? 
Science, 167, 1715. 

Rowe, T. W. G. (1960). The theory and practice of freeze-drying. Ann. N.Y. 
Acad. Sci., 85, 641. 

Scheraga, H. A. (1966). The effects of solutes on the structure of water and 
its implications for protein structure. Ann. N.Y. Acad. Sci., 125, 253. 

Sjostrand, F. S. (1967). Electron Microscopy Of Cells And Tissues. Vol. 1. 
Academic Press, New York. 

Sjostrand, F. S., and Baker, R. F. (1958). Fixation by freeze-drying for elec- 
tron microscopy of tissue cells. }. Ultrastruct. Res., 1, 239. 

Sherman, J. F., and Kim, K. S. (1967). Correlation of cellular ultrastructure 
before freezing, while frozen and after thawing in assessing freeze-thaw- 
induced injury. Cryobiology, 4, 61. 

Smith, A. U. (1961). Biological Effects Of Freezing And Supercooling. Phys- 
iological Society Monograph 9. The Williams and Wilkins Company. 

Solomon, A. K. (1971). The state of water in red cells. Sci. Amer., 224, 88. 

Stephenson, J. L. (1960). Fundamental Physical Problems In The Freezing 
And Drying Of Biological Materials. In: Recent Research In Freezing And 
Drying. (Parkes, A. S., and Smith, A. U., eds.) p. 122. Blackwell Scientific 
Publications, Oxford, England. 

Sterling, C. (1969). Gel-water relationships in hydrophilic polymers. X-ray 
diffraction of fresh and frozen gels. Cryobiology, 5, 393. 

Stowell, R. E., Young, D. E., Arnold, E. A, and Trump, B. F. (1965). Struc- 
tural, chemical, and functional alterations in mammalian nucleus following 
different conditions of freezing, storage and thawing. Fed. Proc, 24, SI 15. 

Troschin, A. S. (1958). Das Problem der Zellpermeabilitat. Fischer, Jena, 

Trump, B. F., Goldblatt, P. J., Griffin, C. C, Waravdekar, V. S., and Stowell, 
R. E. (1964). Effects of freezing and thawing on the ultrastructure of mouse 
hepatic parenchymal cells. Lab. Invest., 13, 967. 

Trump, B. F., Young, D. E., Arnold, E. A., and Stowell, R. E. (1965). Effects 
of freezing and thawing on the structure, chemical constitution, and func- 
tion of cytoplasmic structures. Fed. Proc, 24, S144. 

Van Harreveld, A., and Crowell, J. (1964). Electron microscopy after rapid 
freezing on a metal surface and substitution fixation. Anat. Rec, 149, 381. 

Van Harreveld, A., Crowell, J., and Malhotra, S. K. (1965). A study of 
extracellular space in central nervous tissue by freeze-substitution. J. Cell. 
Biol., 25, 117. 



Van Harreveld, A., and Steiner, J. (1970). Extracellular space in frozen and 
ethanol substituted central nervous tissue. Anat. Rec, 166, 117. 

Webb, S. J. (1965). Bound Water In Biological Integrity. Charles C. Thomas, 

Wentorf, R. H. (1970). Anomalous water: attempts of high pressure synthesis. 
Science, 169, 175. 

Westwater, T. W. (1959). Boiling heat transfer. Amer. Sci., 47, 427. 

Wolman, M. (1955). Problems of fixation in cytology, histology and histo- 
chemistry. Inter. Rev. Cytol., 4, 79. 

Zotikov, L., and Bernhard, W. (1970). Localisation au microscope electronique 
de l'activite de certaines nucleases dans de coupes a congelation ultrafines. 
/. Ultrastruct. Res., 30, 642. 

2 The Freeze-Etching 

James K. Koehler 

Department of Biological Structure, University of Washington, 
School of Medicine, Seattle, Washington 

Scope of the Work 

Louring the last two or three years, the use of, and interest in, freeze- 
etching as an ultrastructural preparative technique has increased dra- 
matically. The reasons for this growth include the realization among 
biologists that this method has unique advantages over conventional 
preparative methods together with the increasing availability of freeze- 
etching or fracturing equipment having greater reliability than was 
previously the case. 

This chapter should provide those in or about to enter the freeze- 
etching field with pertinent technical information concerning procedures, 
instruments and interpretation. It is also hoped that the present article 
will serve as a reference to those wishing a general background in this 
field which would enable them to more adequately assess the impact of 
freeze-etching results in their own areas of specialization. 

The basic procedures involved in the freeze-etching technique' have 
not significantly changed in the last few years. Thus although these steps 
will be reviewed in this treatment, no attempt will be made to discuss in 
detail those routine aspects of methodology which can be found in com- 
petent older reviews (Koehler, 1966; Moor, 1966; Bullivant, 1970). 
Emphasis will be given to those aspects of the technique which have 
been or seem likely to be improved by new approaches. Sections are 
therefore included on double replica preparations, the use of electron 
beam shadowing and ultrahigh vacuum devices. All of these areas are 
destined to grow and substantially add to the quality of freeze-etching 

No effort will be made to review the over-all biological significance 
of freeze-etching observations. These results will be mentioned only 
where they seem particularly pertinent to a discussion of specific tech- 
nical problems. A section is included, however, on the interpretation of 
freeze-etching images and sources of artifacts. The attached bibliography 
is not exhaustive and should be supplemented by the literature quoted in 
the above mentioned reviews. The Balzers High Vacuum Co. in Liech- 
tenstein maintains an extensive bibliography in the freeze-etching field 
which can also be consulted. 



Historical Background 

It is not generally appreciated that Hall (1950) initially suggested that 
the sublimation of ice in vacuo could be utilized to create a relief view 
of hydrated specimens. Hall employed the word "etching" to describe 
this process and examined silver halide and ice crystals by making 
replicas with chromium-shadowed and SiO-reinforced films. Provisions 
for producing fractures of the specimens prior to sublimation were in- 
corporated into an instrument used by Meryman and Kafig (1955) who 
were interested in the crystal morphology of ice undergoing recrystalliza- 
tion at low temperatures. Steere (1957) improved the procedure for 
fracturing the specimen with a cooled scalpel and applied the method 
successfully for the first time to biological systems, primarily virus 

Several years after Steere's experiments the first publication ap- 
peared from the laboratories of Moor et al. (1961) describing a "freezing 
ultramicrotome" which was utilized in vacuo to provide fi eeze-etching 
replicas. This rather sophisticated machine constituted a major departure 
from the earlier "more provisional" instruments in several important re- 
spects. Among other refinements, the Moor instrument contained a 
microtome, patterned after the yolk and gimbel design of Porter and 
Blum, built into the vacuum chamber. The knife of this microtome 
could be cooled to liquid nitrogen temperatures for the cutting steps. 
This instrument also made provisions for temperature control and 
measurement of the specimen. These features greatly increased the 
quality and reliability which could be attained with this type of prepara- 
tion. The Moor prototype served as the model for the commercial pro- 
duction of the Balzers freeze-etcher which will be further discussed in 
the section on instrumentation. 

A number of other instruments or components have appeared over 
the years capable of yielding freeze-etching or freeze-fracturing results. 
The most important of these are the Bullivant-Ames block device ( 1966 ) 
together with subsequent refinements and the modular freeze-etching 
collar devised by Steere (1969a). The reason for the development of these 
later instruments seems to have been a feeling that the Moor device was 
overly complex, bulky and expensive, and that much more modest units 
could do an equally competent job. Although many of these so-called 
"simplified" freeze-etching instrument are in use, it has yet to be demon- 
strated that they offer the flexibility and reliability of the microtome-type 
of instrument. 

It is of interest that much of the impetus for the development of 


freeze-etching methodology grew out of efforts to produce frozen- 
ultrathin sections. Only recently has some measure of success actually 
been demonstrated in the frozen-thin section field ( Christensen, 1967). 
It would seem natural that more comparative work will be forthcoming 
combining these techniques. 


General Considerations 

The biological specimens of interest must, of course, be frozen in a 
manner which minimizes ice crystallization damage or other artifacts 
connected with the freezing process. Part of the success that the freeze- 
etching method has enjoyed is undoubtedly related to the fact that the 
initial freezing process is its only major limitation with regard to the 
production of cryo-artifacts. Since the replicas are prepared while the 
specimens are maintained at — 100 °C or below, problems which arise due 
to thawing phenomena are essentially obviated. It is very often the 
thawing step which yields the most troublesome disturbance of normal 
morphology in preparative methods such as freeze-drying or cryosection- 
ing. Recrystallization phenomena involving ice crystal growth are known 
to occur even at temperatures below — 100°C in unprotected solutions 
(Meryman, 1957). It is beyond the scope of this chapter to go into a 
discussion of the complex physicochemical factors involved in the freez- 
ing and thawing damage which can occur in cells. Excellent treatments 
of this subject can be found elsewhere (Sherman, 1969) and also in this 
volume by Rebhun (chapter 1). 

However, it behooves the prospective freeze-etcher to be acquainted 
with the major sources of structural damage which can occur in the 
course of freezing procedures and to be cognizant of the fact that various 
cellular systems differ vastly in their ability to withstand freezing and 
thawing trauma. The ideal criterion would be the demonstration of 
viability in the cells or tissue under consideration after the procedure has 
been carried out. Indeed, in a limited number of cases it is possible to 
demonstrate high degrees of survival after freezing and thawing such 
cells as spermatozoa (Sherman, 1963) and red blood cells (Doebbler 
et al., 1966), particularly after the addition of cryoprotective agents. 

In the following paragraphs the basic procedures involved in pre- 
paring freeze-etch specimens will be discussed. The method can be 
divided into six steps which involve in sequence: pretreatment of the 
specimen with cryoprotectants, freezing the specimen, fracturing and 
etching, replication and cleaning the final replica. 


Specific Methodology 

Pretreatment of the Specimen 

The majority of the plant and animal cells which are investigated by 
freeze-etching cannot withstand the freezing process either in the sense 
of viability or in the preservation of their ultrastructure. It appears that 
the damage results from two major sources as suggested from the work 
of Mazur (1970) and others. One of these involves the production of 
intracellular nucleation sites for the growth of the ice crystals as the cells 
are cooled below the point of intrinsic supercooling. At high rates of 
cooling (several hundred °C/sec) such nucleation is massive, resulting 
in large numbers of very small intracellular ice crystals. Although not 
particularly disturbing at the light microscope level of resolution, such 
microcrystals greatly disturb normal ultrastructural morphology. As the 
cooling rate is decreased to slower values, the number of nucleation sites 
is reduced, resulting finally in a smaller number of rather large intra- 
cellular ice crystals. Although not necessarily detrimental to viability 
(Sherman, 1962), such large ice crystals greatly interfere with the assess- 
ment of normal ultrastructural information. 

In addition to the crystallization phenomenon, slower cooling rates 
produce dehydration of the cells due to the influence of extracellular 
ice and concomitant increases in extracellular electrolyte content. Thus, 
shrinkage of the cellular material is to be expected when employing slow 
cooling rates. The increased concentration of electrolytes in extracellular 
fluids during slow cooling is considered to be an important factor in pro- 
ducing lethality in some cell types exposed to slow freezing (Lovelock, 

The two factors over which the morphologist can exercise some con- 
trol during the preparation of specimens for freeze-etching are the 
cooling rate during freezing (to be discussed in the next section) and 
the pretreatment or incubation of the tissue before freezing with a 
cryoprotective agent. 

It was shown many years ago that glycerol has the ability to protect 
certain cell types from irreversible damage during freezing and thawing 
( Polge et at, 1949 ) . The list of such cryoprotective substances has grown 
to include a large number of additives. Two broad categories of cryopro- 
tectants can be distinguished. The intracellular or penetrating substances 
include ethylene glycol, certain sugars, dimethyl sulfoxide (DMSO) and 
glycerol. The extracellular, or nonpenetrating, additives include poly- 
vinyl pyrollidone and the dextrans (see review by Nash, 1966). 

The most reasonable explanation for the cryoprotective action of the 


glycols and glycerol involves their ability to form extensive hydrogen 
bonds with water, thus interfering with the tendency of water to 
crystallize. In the presence of glycerol, the freezing rate dependence of 
intracellular ice nucleation becomes much less critical and relatively slow 
rates of cooling can be utilized with such specimens without crystalliza- 
tion phenomenon becoming evident. 

By far, the greatest bulk of work in the freeze-etching field has been 
done with cells or tissues protected by incubation in a 10-20% glycerol 
solution. The time of incubation is, of course, a direct function of the 
diffusion constant for glycerol in the particular system under study. 
Unless this factor is known from previous studies, the incubation time 
must be determined empirically by studying the freeze-etch morphology 
as a function of incubation time. The incubation time is fairly important 
since one must allow an adequate time for diffusion to occur and yet not 
leave the tissue (especially unfixed) in the bath any longer than abso- 
lutely necessary to obviate post mortem changes. Diffusibihty of glycerol 
into cells varies widely even among similar cell types from various 
species. For example, sheep red blood cells require several hours of in- 
cubation whereas those of rat or human can be frozen successfully after 
just a few minutes treatment. 

The use of glutaraldehyde fixation prior to freezing has reduced the 
need for a very critical evaluation of this parameter since postfixation 
changes are kept to a minimum. It is in a sense unfortunate that so many 
investigators now routinely fix their specimens chemically before the 
freeze-etching procedure. One of the advantages of the method, after all, 
is that it can be utilized with fresh, unfixed and even living specimens. 
Although many workers now routinely use pretreatment regimens in- 
volving 30, 40 or even 50% glycerol solutions, it is advisable to minimize 
the concentration as much as possible consistent with good morphological 
results. One of the persistent criticisms of freeze-etching results pertains 
to the artifacts which might be contributed by the use of nonphysiolog- 
ical concentrations of additives such as glycerol. Quite adequate results 
can usually be achieved with pretreatment solutions containing 10-15% 

A number of investigators have had some success preparing freeze- 
etch specimens without using cryoprotectants. Such efforts have been 
largely restricted to very hardy cell types, such as bacteria or yeast (Van 
Gool et al, 1969; Moor and Miihlethaler, 1963), or to cells already par- 
tially dehydrated due to natural processes, such as spores ( Sassen et al, 
1967). Most animal cells have a water content too high for vitrification 
to occur even with very high rates of cooling. It is, however, reassuring 
that cells such as the yeasts show essentially the same freeze-etch mor- 


phology in both glycerol pretreated and untreated preparations (Moor 
and Miihlethaler, 1963). 

Freezing the Specimen 

After pretreatment the specimen must be frozen prior to introducing it 
into the vacuum evaporator. The usual procedure involves depositing a 
dense droplet of cell suspension or a small bit of tissue on a copper speci- 
men carrier, plunging the carrier into liquid Freon 22 cooled close to its 
freezing point (— 165°C) with liquid nitrogen and either transferring 
the specimen directly to the freeze-etcher or storing it in liquid nitrogen 
for later use. Many modifications of the precise methods of accomplish- 
ing the freezing step exist dependent upon the needs of the particular 
freeze-etching instrument. Thus the Bullivant-Ames (1966) or McAlear- 
Kreutziger (1967) block instruments (see the section on Instrumenta- 
tion) involve freezing the specimens in small boats, tubes or rolled up 
electron microscope grids. The freezing either occurs within the pre- 
assembled-chilled block apparatus or in the various holders which are 
then kept at liquid nitrogen temperatures for later use. 

By no means must all specimens be frozen in an ultrarapid manner 
in order to achieve good morphology. In the case of compact (naturally 
dehydrated) cells such as some typical spermatozoa, it has been shown 
that slow freezing (several degrees/min) is preferable with regard to 
survival of such populations (Polge, 1953). Freeze-etching studies on 
such cells (Koehler, 1966) have confirmed that such slow freezing (with 
glycerol pretreatment) results in the maintenance of all morphological 
features that can be recognized by other fine structural techniques. A 
simple means by which slow cooling can be accomplished is to mount 
or suspend a microscope slide in close proximity over the liquid Freon 
bath. The specimen carriers are then placed on the slide and, dependent 
upon the distance from the surface of the liquid Freon and the thermal 
conductivity geometry, slow cooling proceeds. Once solidification of the 
sample has occurred (recognized by opacity change), the specimens can 
simply be pushed into the bath proper for maximal cooling. Approximate 
cooling rates for such simple systems can be estimated from thermo- 
couple or thermister probe readings on an implanted sample. 

Some efforts have been expended toward achieving vitreous ice speci- 
mens in a reproducible manner without cryoprotectant additives. Moor 
and Riehle (1968) utilized specimens subjected to 2000 atmospheres of 
pressure (which lowers the freezing point of water to about — 20°C) 
combined with ultrarapid freezing in a liquid nitrogen stream. Such 
efforts are of academic interest and, considering the success of more 


routine procedures, they cannot be considered necessary for the achieve- 
ment of useful freeze-etching specimens. Storage of frozen samples in 
liquid nitrogen does not appear to alter the quality of preservation at 
least for periods of several weeks. 


The principle way in which freeze-etching preparations differ from 
standard replicating procedures is that internal cellular details are re- 
vealed in freeze-etching rather than only surface structures. This internal 
detail is exposed by means of cutting or fracturing the frozen specimen 
in some way prior to preparing the replicas. The exact manner of accom- 
plishing this fracturing is probably the most variable parameter among 
the various freeze-etching devices in current use. The microtome-type 
freeze-etchers (Balzers Model BA360M) employ a remote control, me- 
chanical advance microtome not unlike the early Porter-Blum ultra- 
microtome mounted in an upright position (Fig. 2.1). A micrometer 

Fig. 2.1 View of the commercially available microtome-type freeze-etching in- 
strument produced by Balzers (Model BA360M). This particular instrument is 
fitted with electron beam evaporator (arrow). Courtesy of the Balzers High 
Vacuum Co., Liechtenstein. 


screw advances the cutting edge and a fine thermal advance is available. 
For most specimens a fairly coarse advance is perfectly suitable and an 
extremely fine advance comparable to that used for thin sectioning is 

A good quality stainless steel razor blade, thoroughly cleaned with 
xylol or other organic solvents, is used as a knife. Special sharpening is 
not required; however, a microscopic examination of the cutting edge is 
advisable in order to reject blades with numerous defects. The Steere 
module ( Fig. 2.2A, B ) employs a long handled scalpel which is manipu- 
lated from outside the vacuum chamber through a universal seal. This 
device is obviously not as easily controlled with respect to cutting height 
or speed, but is considered sufficiently accurate for most preparations. 
In the McAlear-Kreutziger block marketed by the Ebtec Corp., (Fig. 
2.3) a preset spring-loaded trigger is released at the time of cutting 

Fig. 2.2A 


The Steere-Denton freeze-etching device. Courtesy of Dr. R. L. Steere 


Fig. 2.2B Schematic representation of the Steere unit. B: stainless steel specimen 
tube; C: copper or brass shroud; D: stainless funnels for liquid nitrogen supply; 
J: copper specimen stage; Q: electrodes for shadowing and carboning; S: scalpel 
for fracturing. See reference for full description. Courtesy of Dr. R. L. Steere 

which permits a stainless steel knife to make one pass through the frozen 
specimen prior to replication. This system has the advantage of sim- 
plicity and reproducibility, but suffers from a lack of flexibility in that 
the operator cannot, if he chooses, make more than one attempt to frac- 
ture a given specimen. The fractures are, furthermore, all of the same 
quality in terms of coarseness. Some specimens are displayed more 
advantageously if rather fine increments of fracture thickness are em- 
ployed. This is especially true if one wishes to obtain cross fractures 
through very small cellular entities such as bacteria. The coarse cutting 
most frequently gives surface reliefs of such objects rather than cross 

The Type II block freeze-fracturing apparatus (Bullivant et al., 1968) 
requires fracturing the specimen under liquid nitrogen with a cooled 
razor blade prior to transferring the assembled block apparatus to the 
vacuum chamber. While providing a very simple and rapid way of ob- 
taining a fractured specimen, this procedure suffers from some of the 
same constraints mentioned above. 

Etching (Ice Sublimation) 

After completing the fracturing of the specimen, the topography of the 
surface should reflect the upward or downward excursions caused by 
preferential deviations from the theoretical fracture plane. Thus, final 
image contrast in the replica would be limited to those structures which 


Fig. 2.3 The Ebtec Corporation's freeze-etching unit patterned after the McAlear- 
Kreutziger design (1967). The base-plate and electrode assembly is permanently 
installed in a vacuum evaporator. The cylindrical freeze-etching module is de- 
mountable. Courtesy of the Ebtec Corporation. 

caused deviation from the theoretical fracture plane by the presence of 
membranes or other substances less easily cross fractured than the gen- 
eralized cytoplasmic matrix. Additional information can be obtained by 
allowing some of the ice to sublime away under vacuum conditions in 
order to expose intra- or extracellular substances which would otherwise 
be masked from view. This process has been termed etching, hence the 
name freeze-etching. This is to be contrasted with the procedure of 


Bullivant (1970) which does not utilize the sublimation step and is 
referred to as freeze-fracturing. 

Two conditions must be met if the etching process is to proceed suc- 
cessfully. Since the vapor pressure of ice varies strongly with tempera- 
ture (see Table I in Koehler, 1968), it is necessary to have good tem- 
perature measurement and control of the specimen stage. The drift in 
temperature over the normal etching times of one to several minutes 
must be minimal in order to prevent surges of sublimation or condensa- 
tion which might distort the specimen. At — 100°C the vapor pressure 
of ice is approximately 1 X 10 -5 Torr which corresponds to an "etching 
rate" of 23 A of depth per second. This is a reasonable rate for the ex- 
posure of nonetchable structures approximately 1000 to 1500 A above the 
surrounding ice level in one minute of etching time. Deviation from 
— 100°C by ±10°, however, results in an order of magnitude change in 
the etching rate. The temperature must, therefore, be accurately known 
at least to ±1°C in order to make a reasonable estimate of the required 
etching time. 

The other factor of critical importance during etching is the elimina- 
tion or prevention of contamination which would otherwise tend to con- 
dense on the cold surface of the exposed fracture before replication. In 
the microtome-type freeze-etchers (Moor et al., 1961) this is accom- 
plished by leaving the liquid nitrogen cooled knife holder in position 
over the specimen until just before replication, thus providing both a 
colder surface in close proximity for efficient ice sublimation and a pro- 
tective "cold trap" over the specimen. In the Steere (1969a) device, an 
outer "shroud" is placed around the specimen to prevent contamination, 
and in the block-type devices, the relatively long, cold pathways through 
the upper block trap most contaminants before they can reach the 

Many of the most useful interpretations of freeze-etch results are 
made by comparisons of nonetched and etched (or deeply etched) speci- 
mens. This aspect will be further explored in the section on Interpreta- 
tion. Despite the usefulness of the etching step in this procedure, a good 
deal of information can be obtained from nonetched specimens (just 
f reeze-f ractured ) as summarized by Bullivant (1970). Indeed, there 
appear to be several instances in which the sublimation step actually 
obscures detail present in the original fractured specimen as suggested 
for nuclear chromatin and basal lamina material (Bullivant, 1970). 


After exposing the fractured surface of the specimen and in those instru- 
ments so equipped allowing etching to proceed, the surface must be 


replicated for future viewing in the electron microscope. The replica 
technique has been in use in the fine structure field for many years and 
has a large number of variations. Standard works should be consulted 
for a general introduction to the principles and practice of these methods 
(Bradley, 1965). The constraints imposed by freeze-etching require a 
replication technique that takes a minimal time and imposes a negligible 
heat load on the specimen. The latter requirement is necessary to prevent 
micromelting phenomena on the specimen surface. Since the work of 
Moor et al. ( 1961 ) , the method of choice for freeze-etching has involved 
shadowing the specimen with a carbon-platinum film followed by a 
carbon "backing" film at normal incidence. As pointed out by Bradley 
(1959), the carbon-platinum combination is capable of giving very fine 
grain resolution in some cases bordering on a grainless deposit. Essen- 
tially all of the freeze-etching work done to date has employed the 
carbon platinum shadowing procedure. The notable exception to this is 
the recent advent of electron beam shadowing (Bachmann et ah, 1969) 
which will be discussed below. 

The carbon-platinum method can be carried out either by using pre- 
formed carbon-platinum pellets (Ladd Research Industries) or by pre- 
paring small platinum wire coils (.005 in in diameter), which are tightly 
applied around sharpened carbon rods as shown in Fig. 2.4. The geom- 
etry of the carbon-platinum electrode is quite critical to the quality of 
the resulting replicas. A distance of 15 cm is recommended from the 
specimen stage in order to minimize thermal effects. At this distance, 
about three inches of .005 in diameter platinum is used and provides a 
film giving adequate contrast. The maintenance of proper contact be- 
tween opposing carbon electrodes is also very critical for achieving a 
good replica. Some laboratories routinely plane off the electrode tips by 
lathe or rotate the electrodes relative to each other under a hand lens 
in order to provide the best possible contact. 

The spring tension must be adjusted so that firm contact is maintained 
during heating, but not so tightly that the tip breaks before the platinum 
has completely vaporized. The current should be regulated so that the 


■i»muvmmum\i«vtt»\wi!a.N! u»^v'aM^-^ ^te.^\touvto.«.»^^vuAftt^A^to,tt«.^^ 

Fig. 2.4 Typical configuration for carbon-platinum resistance evaporation. G is a 
graphite rod wile C is carbon. The arrow indicates the position of the platinum 
wire coil. 


platinum is vaporized in ten seconds or less. Heating of much longer 
duration can cause some micromelting of surface structures, resulting in 
a "pebbly" appearance. Despite strict adherence to all the precautions 
that can be taken in order to provide a good carbon-platinum evapora- 
tion, most workers admit that only one out of two or three evaporations 
are successful. The factors responsible for this rather poor yield are diffi- 
cult to pinpoint or control. Some of the fault lies with contact problems, 
spring tension variations, slight inhomogeneities in carbon, and other 
parameters not easily identified. 

These difficulties have resulted in the quest for a more reproducible 
method of shadowing freeze-etchings on the part of several investigators 
(Bachmann et al, 1969; Steere, 1969a; Zingsheim et al, 1970). The 
method which appears most promising is to employ electron beam 
evaporation. In this procedure, a small electron gun is mounted inside 
the vacuum chamber and its columnated beam is directed against an 
anode cup containing the metal to be vaporized. Although a number of 
commercial units employing this concept have been available for sev- 
eral years, the requirements for freeze-etching have necessitated a special 
effort to design a unit specifically for this application. Zingsheim et al., 
(1970) have described an electron beam shadowing device that appears 
to fit the necessary qualifications. This device is diagrammed in Fig. 2.5. 

The water-cooled unit minimizes thermal effects and a system of 
deflection plates is used to prevent ions produced during the vaporiza- 
tion from hitting the specimen. The source used is a tungsten-tantalum 
alloy produced by simply winding tantalum wire around a tungsten rod. 
Bachmann has claimed that the preparations made in this way show a 
higher resolution than comparable carbon-platinum preparations. This 
contention is open to some debate; however, the real advantage of the 
electron beam system is in its ease of operation and reproducibility. The 
kinds of problems encountered in resistance-heating evaporation are of 
no consequence since there are no contact points between electrodes to 
be controlled, and no spring tensions to be adjusted. Although the units 
available at the present time (the Balzers instrument, for example) are 
very expensive, it is hoped that their purchase and use will be com- 
pensated for by increased productivity. 

The second step in the double evaporation process, that of forming 
a continuous backing film of carbon over the shadowed preparation, has 
far fewer critical requirements than the shadowing step. Carbon is the 
material of choice evaporated in the usual way in which carbon coated 
grids are prepared. The evaporation must be commenced as soon as the 
shadowing is completed and should be carried out as quickly as possible. 
One obviously strives for a film which will be strong enough to hold the 


y shield 

Fig. 2.5 Electron beam evaporation source. A,-A„ are water-cooled feedthroughs 
mounted on flange (B). Electrons emitted by cathode (C) are focused by aperture 
(E) onto the tip of anode (D). Courtesy of H. P. Zingsheim ef a/. (1970), The In- 
stitute of Physics and The Physical Society. 

replica together during subsequent cleaning and handling, but not so 
heavy as to limit resolution. Resistance carbon evaporation is used in this 
step even when electron beam evaporation is employed for shadowing. 

Cleaning the Replica 

It is somewhat disconcerting that amongst all of the complex delicate 
processes involved in making freeze-etching preparations, some of the 
greatest difficulties are encountered in the final steps of releasing the 
replica from underlying tissue and removing traces of adhering debris. 
The carbon-platinum replicas are extremely fragile and brittle and, 
therefore, cannot be directly manipulated without danger of breakage. 

After removing the specimen from the vacuum coating unit and 
allowing it to thaw, it is floated onto the surface of an aqueous solution, 


similar in composition to the suspending medium in which the tissue was 
frozen. If the replica does not float free of the underlying tissue then one 
must suffer through the more difficult chore of cleaning the replica of 
adherent macroscopic material. It is the experience of most laboratories 
that pellets of free living cells or other particulates are the simplest 
materials to work with, in that replicas of such samples generally float 
free of underlying debris leaving only microscopic residue to be digested. 
Tissue blocks, on the other hand, very often adhere tightly to the replicas, 
thus complicating the procedure. In one's eagerness to pick away the 
tissue, the replica very often will shatter or otherwise become unusable. 
In order to combat this problem Wehrli et al., (1970) evaporated a 
covering layer of napthalene onto the finished replicas in order to 
strengthen them during subsequent processing. After cleaning the rep- 
licas, the napthalene is sublimed away at 70°C in an oven. 

Once the replica, with or without adherent tissue, is floating on the 
surface of a depression slide or small Petri dish, a cleaning solution 
replaces the original medium. Many concoctions have been utilized for 
the removal of cell debris from replicas. The most generally successful 
of these is ordinary household chlorine bleach ( 5% sodium hypochlorite ) . 
The generation of many small bubbles by this agent, especially when 
there is a large amount of tissue present, occasionally causes some diffi- 
culties including the disruption of the replica when large bubbles burst. 
Dichromate cleaning solution, strong sodium hydroxide, sulfuric or hot 
nitric acid and combinations of these have all been utilized for this pur- 
pose, as well as proteolytic enzymes. An empirical approach must be 
used to determine which cleaning agent does the best job for the par- 
ticular biological material under investigation. The length of time in 
contact with the cleaning solution must also be determined experi- 
mentally, since some substances are digested in a matter of minutes 
whereas others require prolonged treatment for hours or even days. 

Subsequent to cleaning, replicas are carefully rinsed in several 
changes of distilled water. This is most readily accomplished by picking 
the replicas up in a droplet suspended by a fine wire loop and placing 
it onto the surface of the fresh water. Alternatively, the bleach or other 
cleaning agent can be carefully removed by pipette from the depression 
slide and replaced a number of times with fresh distilled water. Finally, 
the replicas are picked up on a wire loop and transferred to coated or un- 
coated grids. If large mesh (150 or 100 mesh) grids are used, it is ad- 
visable to use a Formvar or parlodian film as coating. Fishing the floating 
replicas directly from the water surface with grids is not advisable due 
to a lack of control over where the replica will finally come to rest on 
the microscope grid. 



The three basic types of freeze-etching or freeze-fracture devices have 
already been mentioned in earlier sections. In the following discussion a 
further consideration of these instruments is provided with particular 
emphasis on design concepts and differences in "philosophy" on the part 
of proponents of these various instruments. Detailed specifications of 
particular instruments will not be dealt with since such information can 
easily be provided by the respective manufacturers. 

Microtome-Type Freeze-Etchers 

The prototype described by Moor et al. (1961) has served as the basis 
for the commercial development of the freeze-etching instrument by the 
Balzers Company (Liechtenstein) (Fig. 2.1). Such instruments (Model 
BA360M) have been in use for a number of years and have proven to 
be highly reliable and capable of providing excellent results. During this 
time these instruments have undergone only very minor revisions and 
remain the least modified of the various types of devices that will be 
described in this section. The primary considerations in the construction 
of this unit are to provide an internal cutting device of high accuracy 
having a fine mechanical advance and, when required, thermal advance 
capability. In addition, the specimen table was required to have not only 
a temperature measuring device, but an associated temperature control 
unit which could maintain a desired temperature for prolonged periods 
of time. Such temperature control is especially necessary for deep etching 
procedures which can provide important information additional to that 
obtained from nonetched or minimally etched specimens. Thus, the 
Balzers instrument incorporates both liquid nitrogen cooling and heating 
features in the specimen stage capable of temperature control to 

The microtome cutting device must also be cooled to a low tempera- 
ture in order to minimize surface melting of the specimen during cutting 
and in order to provide an efficient condensing surface during the etching 
step. The hollow microtome arm is filled with liquid nitrogen and 
allowed to cool to a temperature well below that of the specimen. The 
temperature (approximately — 180°C) of the microtome razor blade 
housing acts not only as a sink for sublimed ice from the specimen 
during etching, but also minimizes the access of vacuum space con- 
taminants to the exposed specimen surface. 

During a typical freeze-etching run, the specimen stage is first cooled 
to a low temperature (—110 to — 125°C) subsequent to obtaining a good 


high vacuum in the chamber. The vacuum is then released with dry 
nitrogen and the specimen mounted quickly on the cold stage. A 
vacuum is then once more pumped while simultaneously cooling down 
the microtome knife with liquid nitrogen. Ten or 15 min are required 
for this cooling step and the achievement of the operating vacuum of 
approximately 5 X 10 -6 Torr or better. The temperature of the speci- 
men is then adjusted to the desired value (usually about — 100°C) and 
several cuts are made across the specimen. The number of passes 
( usually 3-6 ) should be minimized in order to develop fewer knife edge 
imperfections. After the last pass, the microtome razor blade housing is 
left over the specimen and allowed to remain there until the predeter- 
mined etching time ( one or several minutes ) has elapsed. The specimen 
is then replicated, removed and cleaned as discussed previously. Etching 
is, of course, not mandatory and the specimen could be replicated imme- 
diately upon completion of the last cutting stroke. 

Replication is carried out in the usual manner with standard re- 
sistence-heated electrodes, although an electron beam evaporation device 
is now available for this unit as discussed in the section on replication. 

The author constructed a microtome-type freeze-etching device sev- 
eral years ago (Koehler, 1966) which serves in approximately the same 
manner as the Balzers instrument. Although a few dollars might still be 
saved by this approach, the short-term gain could well be negated by 
the frustration of obtaining reliable results from a home-made device. If 
a laboratory is interested in making a major investment in the freeze- 
etching field and has the space and personnel to utilize the instrument, 
the Balzers device must certainly be among the top contenders for con- 

Steere Freeze-Etching Module 

One of the original procedures for obtaining freeze-etching data was 
utilized by Steere (1957) in an investigation of virus crystals. A refine- 
ment of this device has been described by Steere (1969a) which em- 
bodies some of the basic elements of the Moor instrument but in a 
simplified fashion. The concept was to provide a freeze-etching device 
in which a sample could be fractured within the vacuum chamber, 
etched and replicated, but with less elaborate and less expensive com- 
ponents than those involved in the Balzers instrument. Thus, instead of 
a fine advance microtome, this unit (Fig. 2.2A) is fitted with a long 
handled scalpel operable through a "universal joint" vacuum seal. The 
specimen holder is in direct contact with a liquid nitrogen cold finger 
which is simply filled through an exterior funnel. A thermocouple is 


utilized to monitor specimen temperature and another liquid nitrogen- 
filled tube or "shroud" provides contamination protection and a cold 
sink around the specimen during the etching procedure. 

Problems associated with water vapor condensation during specimen 
introduction caused difficulties in earlier models of this instrument but 
these have apparently been overcome by various modifications of design 
including the use of a "double-shroud" system. The shadowing and car- 
boning electrodes are of the usual resistance type but of a special com- 
pact design permitting relatively simple charging and cleaning. An elec- 
tron gun shadowing unit is also available for the commercial model of 
the Steere module (Denton Vacuum, Cherry Hill, N. J.; Model DFE-3). 
Multiple specimens can be prepared with this device up to 6 or more in 
number including the double replicas described later. It is claimed that 
precision cutting of the specimen can be achieved with the scalpel ob- 
served under a dissecting microscope, but the usual preparations appear 
to be produced by a coarse motion described as "ice pick-like." 

The operation of this module is similar to the procedure employed 
with the Balzers instrument. The specimen tube is cooled to — 125°C 
and after opening the vacuum chamber, the object(s) is placed on the 
specimen table. After once again pumping a vacuum, the shroud is also 
cooled to liquid nitrogen temperatures and the specimen rotated so as 
to be covered by the cold shroud. The scalpel is in thermal contact with 
the shroud for cooling purposes as it does not have an independent 
source of coolant. The specimen is then exposed, fractured and returned 
to its shroud whereupon the specimen temperature is warmed to about 
— 100°C for etching. After the appropriate etching time, the specimen 
is once more exposed and replicated by means of the carbon-platinum 
and carbon-coating electrodes. The objects are then removed and 
cleaned in the usual way. A detailed account of the use and character- 
istics of this apparatus can be found in Steere ( 1969a ) . 

Self-Contained "Block-Type" Devices 

The notion of simplifying the Moor-Balzers type of operation was carried 
to an ultimate point by Bullivant and Ames (1966) several years ago. 
This device consisted simply of a brass block with a small hole drilled in 
the center for a specimen and a metal cover capable of being lifted away 
from the block by using a rotary motion winch in the vacuum evapora- 
tor. The block and cover are placed under liquid nitrogen in a suitable 
insulated container, a specimen is introduced into the hole, fractured 
with a cold razor blade and covered with the metal top. All of these 
operations are performed outside the vacuum chamber. 


The covered block is then transferred to the evaporator where a 
vacuum is pumped, the specimen uncovered and replicated in the usual 
manner with a double electrode system. Bullivant and Ames termed 
this process "freeze fracturing" since the very low temperature of the 
block did not permit ice sublimation to take place. In addition to the 
lack of etching capability, some difficulties with the build-up of con- 
tamination were experienced with this unit due to the relatively exposed 
nature of the fractured specimen during much of the operation and the 
very low temperature of the block, tending to encourage condensation. 
This difficulty was resolved by the construction of a new model. (Bulli- 
vant et al., 1968) which included another massive block just over the 
specimen block containing tunnels for the access of shadowing and car- 
boning materials to the specimen. This center piece was left in place 
during the coating operations and served to virtually eliminate con- 
taminants from the specimen surface. This type of block, still employed 
by Bullivant and others is termed the Type II freeze-fracturing device. 

Further refinements of the block concept have been attempted since 
Bullivant and Ames' original description. McAlear and Kreutziger ( 1967 ) 
have perfected a unit similar to the Bullivant Type II apparatus, but 
capable of fracturing the specimen after introduction to the vacuum 
chamber rather than before. This is accomplished by means of a 
spring-loaded upper block which is provided with a cutting device. Upon 
tripping the spring, the upper block rotates, thus fracturing the specimen 
and at the same time bringing the tunnels into alignment with the rep- 
lication electrodes. 

Various techniques have been attempted to introduce etching ca- 
pabilities into this unit and the most recent commercial model (Fig. 
2.3) of this device (C. W. French Division, Ebtec Corp., Bedford, Mass.) 
includes a resistance heater, thermocouple and control unit which pre- 
sumably "maintains preselected etching temperature." In contrast to the 
Balzers instrument, this controller must operate only with heating ca- 
pabilities and is dependent upon the initially cooled block to provide 
cooling if necessary to maintain the temperature constant during etching. 
It is not known at this time how effective this system actually is in 
providing etching (especially deep etching) in addition to fracturing 
results. Good and useful data has, however, been forthcoming from a 
number of laboratories employing block devices of this type and there 
is a substantial difference in initial cost in comparison to the Balzers 
instrument. Those laboratories interested in preliminary explorations in 
the freeze-fracturing field might find this feature attractive. 

A unit involving the same progression of steps as the original Bulli- 
vant- Ames (1966) method has recently been put on the market by the 


Leybold-Heraeus Co. of Cologne, Germany. The specimen is fractured 
under liquid nitrogen on a special holder which is encased by a tight 
fitting clamp. The entire assembly is then transferred to the vacuum 
chamber where further processing is similar to that described for other 
units except that the specimen can be rotated or tilted during the evap- 
oration process. Components for heating and cooling the specimen are 
also furnished so that freeze-etching as well as fracturing results can 
be obtained. 

Ultrahigh Vacuum Devices 

Virtually all of the results forthcoming in the freeze-etching field have 
been achieved with instruments operated with conventional vacuum 
technology. That is, the basic vacuum evaporator has been in the oil 
diffusion-oil forepump variety capable of attaining ultimate vacua of the 
10 -6 Torr range. Some interest has been expressed in developing freeze- 
etching instruments which utilize more modern ultrahigh vacuum com- 
ponents such as the device recently described by Kreutziger ( 1970 ) . Not 
only can these newer vacuum units provide a much lower ultimate pres- 
sure (of the order of 10 -9 Torr), but they also can be designed to 
operate without oil pumps or grease seals of any kind. The absence of oil 
contaminants would provide a much better environment for freeze- 
etching operations due to the tendency of such hydrocarbons to con- 
dense out on cold surfaces. Kreutziger's unit is in essence a sophisticated 
version of the C. W. French freeze-etch unit described earlier. Unfor- 
tunately, little information is given in the abstract concerning the vacuum 
unit per se. 

The author has recently completed the assembly of an ultrahigh 
vacuum unit containing a simplified freeze-fracture block device which 
will be briefly described below. The basic system consists of an NRC 
model 3401 table top unit (Figs. 2.6, 2.7) pumped by a titanium getter 
having three interchangeable filaments and a 400 1/sec orb-ion pump. 
This pump is similar to other ion pumps except that electrostatic fields 
are utilized rather than magnetic fields to modulate the motion of elec- 
trons and ions. Copper crush, foil and viton seals are used throughout the 
instrument. The roughing system consists of a Gast carbon vane turbine 
pump and two cryosorption pumps filled with Linde molecular sieve 

A Ladd double electrode unit has been incorporated into the system 
for making the replicas. The freeze-fracture device consists of two brass 
blocks (Fig. 2.8), the base of which contains a central cut-out and 
specimen hole. Fine polyethylene tubing about 1 mm in diameter and 


Fig. 2.6 NRC ultrahigh vacuum unit utilized for freeze-fracture work. Bell jar 

1 cm long provides convenient specimen holders for biological material. 
The top of the sliding block system has evaporation tunnels and a razor 
blade mounting. The blocks are charged with a specimen under liquid 
nitrogen, assembled and introduced into the vacuum chamber. After 
pumping a vacuum, the top block is pushed over the specimen thus 
fracturing it and aligning the evaporation tunnels for replication. At the 
moment, the linear motion feed-through seal is the only element of the 
system which utilizes high vacuum grease. It is hoped that this com- 




Fig. 2.7 Detailed view of the electrode assembly (Ladd Research Corp.) and 
freeze-fracture block in place in the NRC unit of Fig. 2.6. 



Fig. 2.8 The double block freeze-fracture unit disassembled. Note razor blade 
mounted on upper block and specimen tube (arrow) in lower block. 


ponent can be replaced by a bellows-type greaseless seal in the near 
future. A thermocouple is used to monitor the block temperature, but 
thus far no heater has been added to the block to enable etching tem- 
peratures to be reached. Actual tests with biological materials are fust 
being initiated with this system (Fig. 2.9) to see if superior results can 
be achieved attributable to cleaner vacuum conditions. 

L ? ■•** 

Fig. 2.9 Yeast cell prepared by freeze-fracturing in the apparatus described in 
Figs. 2.6-2.8. All magnification markers in electron micrographs represent one 
micron unless otherwise noted. 


'Double-Replica" Preparations 

The usual freeze-etch or freeze-fracture preparation allows visualization 
of only one-half of the fractured specimen, namely that portion still in 
contact with the cooling stage and positioned properly for replication. 
The previously adhering portion has been chipped away by the micro- 
tome, scalpel or other device employed for this purpose. It became evi- 
dent quite early in freeze-etching research that it would be extremely 
useful, for the purposes of interpretation, to be able to visualize both 
previously adhering halves subsequent to the fracturing procedure. For 
example, Branton (1966) has pointed out the apparent discrepancy con- 
cerning the numbers of particles and corresponding pits which should be 
visualized in freeze-etch membrane preparations. One would reason that 
the particles should be complimented by equal numbers of pits, assuming 
random processes to be in effect during the fracturing step. The usual 
finding, however, has been that the particles far outnumber the pits on 
most membrane faces. Where did the pits go? Or, alternatively, where 
did the particles come from? Obviously a capability of matching the 
replicated surfaces of both halves of the fractured specimen might be 
expected to answer such questions. 

To date, four separate groups within the short time span of about 
one year have published methods for obtaining double replica freeze- 
etching or freeze-fracturing preparations. All of these methods are 
rather similar in their approach and differ only to the extent that the 
devices were adapted to quite different freeze-etching instruments. In 
essence, the procedure involves the creation of a specimen holder which 
can be broken into two halves in a configuration which will expose the 
two corresponding fractured surfaces to the replication electrodes in the 
same plane. The replicas are then produced in the usual way and re- 
moved from underlying tissue prior to examination. The four devices thus 
far described will be discussed in more detail below in order of their 
appearance in the literature. 


This technique was adapted by Wehrli et ah, (1970) to the commer- 
cial Balzers microtome-type freeze-etcher. The specimen holder consists 
of a small funnel-topped capillary tube of thin metal construction (Fig. 
2. IOC). The tube is scored around its midpoint for reproducible break- 
age and is filled with cell suspension through the funnel. After freezing 
in the usual bath (liquid Freon 22), the tube is placed in a hole of the 
redesigned Balzers cold stage (Figs. 2.10A, B). A liquid nitrogen reser- 


10 d 

Fig. 2.10 Double replica adaptor for Balzers freeze-etching unit before fracturing 
(2.10A), after fracturing (2.10B). 1: microtome knife adaptor; 2: specimen holder; 
3: trough; 4: base piece; 5: object table of the freeze-etching unit. Detail of speci- 
men holder (2.1 0C). Modified adaptor for multiple double fracturing (2.1 OD). The 
object is cleaved by turning the upper specimen holder with the microtome knife 
arm. Cocked position shown on left, cleaved on right. Courtesy of E Wehrli ef a/ 
(1970) and K. Muhlethaler ef a/. (1970). 

voir surrounds the specimen in order to maintain uniformly low tem- 
peratures during the operation. After pumping a suitable vacuum, the 
arm of the microtome is used to break the top of the specimen tube away 
from the bottom half. The movement of the arm and positioning of the 
specimen is such that the free broken half falls upside down into a hole 
next to the original specimen. The hole is of such depth that the two 




i it-."*- 

Fig. 2.11 Double replica preparation of spinach vacuole membranes produced 
by the modified adaptor shown in Fig. 2.10D. Courtesy of Dr. E. Wehrli (un- 


reciprocal broken halves are in approximately the same plane and expose 
their free broken surfaces to the replicating electrodes. 

This same group has recently reported an improved technique 
(Muhlethaler et at., 1970) which apparently provides somewhat more 
reproducible results. In this modification, the specimen is frozen be- 
tween two gold disks which slip into a slotted holder on the specimen 
stage of the Balzers instrument (Fig. 2.10D). At the appropriate time, 
movement of the microtome arm trips the spring which forces the two 
slotted units apart, thus fracturing the specimens. Three pairs of disks 
can be fractured and replicated simultaneously. Excellent results are 
achieved with this unit, and it is expected that the device will soon be 
available commercially as an accessory for the Balzers instrument. The 
accompanying micrographs (Figs. 2. 11 A, B) illustrate the excellent re- 
sults obtainable with this technique using spinach cells. 

steere's device (steere and moseley, 1969, 1970) 

In this adaptation to the Steere modular freeze-etcher, the specimen 
is again contained in a small tube mounted on a hinged device which 
can be forced apart (Figs. 2.12, 2.13). The operation is similar to open- 
ing a book to its center page, thus exposing both faces to the electrode 
assemblies for replication. A modified double replica device (Fig. 2.13) 
allows for the preparation of up to 10 double replicas in a similar fashion. 
Figs. 2.14A, B illustrate portions of chicken red blood cells showing 
complimentary membrane faces. The two procedures thus far described 
involve fracturing the specimen into the two halves while inside the 
vacuum space of the freeze-etching instrument. 

bullivant's device 

Among the simplest procedures for accomplishing double replication 
is the adaptation of Chalcroft and Bullivant (1970) to the simple block 

Fig. 2.12 Drawing of the Steere double fracture device in the cocked (A) and 
fractured (B) configuration. See Fig. 2.13 (left) for schematic representation of 
mechanism of action. Courtesy of R. L. Steere and M. Moseley (1969). 



2 Jt 



Fig. 2.13 Schematic drawing of hinged specimen holders for double replica 
preparations (top). Photos of actual cap and holders in cocked position, fractured 
and from the shadow angle (bottom). Courtesy of R. L. Steere and M. Moseley 

type freeze-fracture device described earlier. The procedure is outlined 
in Fig. 2.15. In essence, a cylindrical specimen holder is cut in half and 
laid end to end (Fig. 2.15A). After freezing, the specimen is immersed 
in liquid nitrogen, pulled apart and rotated together to form a complete 
cylinder once more with the two complimentary broken surfaces side by 
side (Figs. 2.15 B, C). The cylinder is then inserted into the precooled 
specimen block (Fig. 2.15D) and assembled before being placed into 
the vacuum chamber. This procedure is open to the criticism that con- 
taminants might deposit on the broken surfaces long before they are 
replicated. However, the quality of the micrographs obtained is high, as 
shown in Fig. 2.16. 

Fig. 2.14 Complimentary pair of freeze-etchings prepared by the device illus- 
trated in Fig. 2.13. They show mouse red blood cells pretreated in 30% glycerol, 
etched 3 min at — 95°C. Note: shadows in these micrographs are black and re- 
versed from the other figures shown in this chapter. Courtesy of R. L. Steere 



Fig. 2.15 Method of breaking tissue to obtain double replica according to Chal- 
croft and Bullivant (1970). a: specimen holder in end-to-end position; b: side view 
(section) of holder before fracturing; c: after fracturing; d: lower block showing 
specimen insertion. Courtesy of J. P. Chalcroft and S. Bullivant (1970). 

A procedure very similar to that described by Chalcroft and Bullivant 
has been evolved by Sleytr (1970) for use with the Leybold Heraeus 
Co.'s freeze-etch instrument. The similarity even extends to the breaking 
of the specimen under liquid nitrogen outside the vacuum chamber, 
mounting in a specimen block and then transferring to the freeze-etcher 
for replication. An apparent advantage in the Sleytr system is the fact 
that 12 specimens can be prepared simultaneously. Faced with the 
reality of the only moderately reproducible carbon platinum shadowing 
process, however, the coating of 12 specimens with the same replication 
means that occasionally one has 12 failures instead of just one. 


Fig. 2.16 Double replica preparation made with procedures outlined in Fig. 2.15. 
The micrographs illustrate regions of gap junctions in mouse liver cell membranes. 
The reader is referred to the original paper for detailed discussion of these 
images. Courtesy of J. P. Chalcroft and S. Bullivant (1970). 



Every fine structural procedure for the examination of biological material 
is burdened by artifacts created during the preparation of the sample. 
The following discussion is not concerned with sources of error which 
are due to routine handling problems encountered with conventional fine 
structure techniques, but rather with difficulties related directly to the 
freeze-etching technique. The sources of artifacts encountered in con- 
ventional techniques for electron microscopy have been discussed in 
detail by Hayat (1970). 

The freezing process itself may introduce grossly disturbing elements 
into the tissue in the form of ice crystals. Intracellular ice crystallization 
can vary in severity from barely perceptible crystal nuclei to large clus- 
ters of crystals equal to the size of organelles which fill entire cells ( Fig. 
2.17). The latter situation, of course precludes obtaining any fine struc- 
tural information, but at least does not present any difficulty in the 



Fig. 2.17 A "cytoplasmic droplet" region of a rat spermatozoon exhibiting mas- 
sive dendritic ice crystals. Structural details are totally obscured. 

sense of causing an incorrect identification of the structure. As pointed 
out by Steere (1969b), the relatively straightforward recognition of ice 
crystals in freeze-etch replicas can be used as an assay of the adequacy 
of various cryoprotective treatments. 

Most materials require relatively high concentrations of glycerol 
( 10-15% or more ) coupled with rapid cooling rates in order to achieve 
ice crystal-free preparations. When slow cooling rates are employed as 
with some spermatozoa, a partial dehydration of the cells may occur 
causing some shrinkage. It is best to compare such specimens with those 
frozen very rapidly in order to determine the effect of cooling rate on 


The process of fracturing the specimens may also lead to difficulties 
in interpretation. Thus, just as knife marks are occasionally encountered 
in sections of plastic embedded material, they are also commonly found 
in replicas prepared by freeze-etching or fracturing (Fig. 2.18). Those 
areas of the replica most useful for examination are actually the regions 
which have not been "cut" per se by the knife, but rather where fracture 
has occurred along other preferential planes in the sample. In addition 
to the actual scratches present, surface detail is often obscured due to 
the smearing which apparently occurs due to frictional heating at the 
cut surface. For these reasons, regions where the knife has actually 


Fig. 2.18 Frog erythrocyte showing extensive knife scratches in the nuclear 
region (arrows). 


scraped the surface are generally worthless in terms of obtaining ultra- 
structural information. This is a type of artifact which should be appar- 
ent even to the relative newcomer in the field. 

A much more serious potential source of trouble is the possibility of 
plastic deformation of structures into abnormal configurations due to the 
fracturing process. Clark and Branton (1968) have shown quite con- 
vincingly that polystyrene latex particles can be deformed into elongated 
cylinders by cleavage forces at low temperatures. They also interpret 
their images to suggest that collagen fibrils in the extracellular space are 
capable of undergoing a similar deformation. Unless one has a priori 
knowledge of the natural shape of the structure in question, this type of 
artifact would be difficult to detect. It may be that objects cleaved at 
liquid nitrogen temperatures may undergo little or no plastic deforma- 
tion as compared to those processed at — 100°C. Comparisons of such 
samples may be useful when deformation is suspected. 

The time period just after producing the specimen fracture and prior 
to replication is another critical moment with regard to the production 
of artifacts. The freshly created surface is then accessible to the con- 
densation of contaminants from the vacuum space. Thus, precautions are 
taken in virtually all the devices described to protect the fractured 
specimen surface from such contaminants. Such protection, usually in 
the form of shielding maintained at low temperatures, is especially im- 
portant for prolonged etching periods. Steere (1969b) has illustrated 
the difficulties encountered when water vapor or carbon dioxide is inten- 
tionally introduced into the vacuum space. One of the most troublesome 
sources of contamination is the hydrocarbon or silicone vapors from the 
backstreaming of vacuum pumps. These components as well as others 
mentioned may condense out as small particulates which are virtually 
indistinguishable from bona fide membrane-associated particles which 
have been described as universal components of frozen-etched cells. 

The degree of contamination varies according to the technique em- 
ployed from essentially unrecognizable, minute quantities which are 
accepted together with inherent cellular structures as being authentic, to 
grossly contaminated specimens where there is no doubt about the 
nature of the particulates. The latter situation leads to no difficulty since 
the images are useless for fine structural analysis. More serious is the 
problem of deciding whether certain particulates in apparently good 
preparations are contaminants or not. Some guidelines are available for 
coming to such a conclusion. For example, long etching periods should 
result in larger quantities of condensed particulates than short or no 
etching times, whereas true membrane particles should not vary with 
etching time. The prevention of such contamination is also- possible by 


using low temperatures traps utilizing molecular sieve absorbents. The 
use of non-oil pumping systems is, of course, another solution to this 
particular contamination problem (see section on ultrahigh vacuum 
devices ) . 

The process of replication can itself produce changes in the exposed 
surface which may lead to erroneous results. If the shadowing process 
takes too long a time period, excessive surface heating can result in local 
melting and recrystallization on the specimen surface (Figs. 2.19, 2.20). 
Such effects are not necessarily obvious since the production of granular- 

Fig. 2.19 Frog erythrocyte showing "granulation" probably caused by excessive 
heating during the evaporation process. Effect is especially noticeable in cross 
fractured cytoplasm (C). N: nuclear membrane. 


Fig. 2.20 Well-preserved frog erythrocyte showing the lack of coarse granula- 
tion. Compare with Fig. 2.19. C: cytoplasm; N: nuclear membrane. 

ity may be restricted to specific regions of cellular compartments. The 
minimization of evaporation time is necessary to avoid this artifact. 

Interpretation of Cellular Components 

General Considerations 

Apart from the problems associated with the interpretation of the various 
artifacts which may occur in freeze-etchings, the process of relating the 
replica images to cellular entities remains an important challenge. The 


general principles involved in the evaluation of replicas also apply to 
freeze-etch preparations. For example, a knowledge of the shadow direc- 
tion enables one to conclude whether a given structure in the replica is 
elevated or depressed with reference to the general background. Thus, if 
the build-up of metal shadowing material (dark in positive print) on a 
given structure is similar to that of a known particulate nearby, the 
structure must be similarly elevated. If the distribution is reversed, then 
the structure is most likely depressed ( Figs. 2.21, 2.22 ) . 

It should be remembered that the orientation of the micrograph is 
of prime importance with respect to the qualitative visual impression of 
depth. Thus, although the analysis mentioned above with respect to 
shadow direction determination will always result in the correct evalua- 
tion of elevation, the visual impression may be reversed simply by 
rotating the image 180°. Because of this effect, most investigators pub- 
lish freeze-etching micrographs with the shadow direction from bottom 
to top. This orientation yields the correct visual impression. Many 
authors additionally supply an arrow on the micrograph indicating the 

Fig. 2.21 Chylomicrons from milk prepared by freeze-etching and printed with 
shadows (white) pointing up. Some particles (b) appear to protrude (convex) 
whereas others (a) seem broken out of the fracture (concave). 


111*—. .. - TOUBF ••*£mZf- •> • ir^t"^!,**^ ,- ;' ,p . -- f .' 


Fig. 2.22 Same micrograph as Fig. 2.21 but rotated 180° (shadows pointing down- 
ward). The (a) particles now appear to protrude whereas the (b) particles look 
depressed or broken out. 

shadow direction. The precise measurement of heights from shadow 
lengths is complicated by the fact that the fracture creates an uneven 
surface making an infinite variety of angles with the shadowing source. 
It has also become standard practice among freeze-etchers to utilize and 
publish positive prints produced directly from the original negatives 
without recourse to the use of "intermediate plates." Thus, shadows in 
these preparations appear light rather than dark. This reversal appears to 
have no effect on the visual quality of the product and has the benefit 
of doing away with unnecessary and sometimes deleterious photographic 


For purposes of discussion, cellular organelles can be categorized into 
those composed of or surrounded by membranes and those which are 
not membranous. The latter category includes structures such as fila- 


ments of various types and some inclusion bodies (e.g., glycogen, starch, 
and certain lipids). The identification of such structures requires that 
they be capable of being differentiated from the general cytoplasmic 
background in which they occur. Etching is an aid in exposing some of 
these structures, but, as mentioned earlier, the granularity produced in 
background cytoplasm may also obscure some classes of inclusions. A 
careful comparison of etched and nonetched specimens could be very 
helpful in making such identifications. Highly organized filaments such 
as those present in muscle appear to be easily visualized with the freeze- 
etching method ( Bertaud et at, 1968 ) . Many storage forms of lipid ap- 
pear to display a characteristic layered "onionskin" structure in freeze- 
etched preparations ( Frey-Wyssling and Muhlethaler, 1965). 

Spindle microtubule structure in yeast cells has been discussed in 
considerable detail by Moor (1967), however, microtubules in general 
have not been extensively investigated by this technique. It appears that 
in some systems they are rather difficult to demonstrate, particularly in 
unfixed cells. It may be that prefixation in an aldehyde may be justified 
to preserve cellular elements which are highly cold labile. 

Most of the structures of cytological interest within cells are mem- 
branous in nature or at least bounded by membranes. The process of 
low temperature fracturing appears to preferentially expose extended 
face views of these membranes. Most of the common cellular organelles 
are easily identified in such preparations. Thus the nucleus (Koehler, 
1968), Golgi body (Staehelin and Kiermayer, 1970) and mitochondria 
( Meyer, 1969 ) all exhibit characteristic properties of freeze-etch replicas. 
Both face views and cross-fractures of such organelles yield relatively 
unmistakable images. 

A much more formidable problem is involved in discriminating be- 
tween rough and smooth elements of the endoplasmic reticulum. The 
identification of ribosomes on ER membranes appears to be complicated 
by the observation that such membranes are not exposing their surfaces, 
but rather an internal plane (Branton, 1966). Thus, since ribosomes are 
known to be located on external surfaces of ER cisternae, they may be 
obscured by an overlying membranous sheet. Deep etching must be used 
in order to lower the level of ice on such membrane surfaces in order to 
demonstrate the presence of ribosomal particles (Wartiovaara and 
Branton, 1970). Hence, rather special analysis is sometimes required to 
definitively establish a given membranous structure as belonging to 
smooth or rough ER. Obviously, if independent information is available 
to aid in the identification (thin sections, negative staining, etc.), such 
precautions may not be necessary. 


Membrane Structure Considerations 

Although the general identification of membranes with respect to 
organelle type has not presented great difficulties, the precise determina- 
tion of which membrane surfaces or faces are being exposed has been of 
concern to freeze-etchers for several years. The idea that membrane 
surfaces were exposed by cleavage at low temperatures was seriously 
challenged by Branton (1966). In his membrane splitting hypothesis, 
arguments were put forth that suggested that the internal lipoidal layer 
of membranes would be a more likely place for a fracture plane to be 
generated. It was suggested in a previous review (Koehler, 1968) that 
the use of "topographic" markers such as conjugated ferritin in freeze- 
etching experiments could be utilized to specifically label certain struc- 
tures. The definitive experiments utilizing such methods to test the mem- 
brane splitting hypothesis have recently been reported by two groups 
of investigators. Pinto da Silva and Branton (1970) employed ferritin 
conjugated to rabbit red blood cell ghosts with toluene-2,4-diisocyanate. 
In nonetched preparations, no ferritin particles could be detected on 
the membrane faces. Similar preparations deeply etched (Fig. 2.23), 
however, revealed ferritin particles on the true membrane surfaces ex- 

Fig. 2.23 Rabbit red blood cell ghosts after freeze-etching. Ferritin-conjugated 
sites (arrows) are seen on membrane faces exposed by deep etching E but not 
on originally fractured face F. Inset shows position of membrane at higher mag- 
nification. Courtesy of P. Pinto da Silva and D. Branton (1970). 


Fig. 2.24 Part of red cell ghost membrane deeply etched to reveal ferritin par- 
ticles on the surface (arrows). The conjugated ferritin shows presumptive A 2 anti- 
gen sites. Courtesy of P. Pinto da Silva (unpublished). 

posed by the etching. A similar study utilizing fibrous actin as a mem- 
brane marker yielded the same results (Tillack and Marchesi, 1970). 

An extension of the ferritin studies has been carried out by Pinto da 
Silva et al., (1971) in which it was attempted to label specific antigen 
sites on red cells with ferritin conjugated immunoglobulins (Fig. 2.24). 
Although such experimental confirmations of the membrane splitting idea 
are clear-cut in the red blood cell system, it remains to be seen whether 
all membranes will always behave in this manner. Certainly, a deep 
etching trial would be advisable on any membrane system of particular 
interest to an investigator before automatically invoking the membrane 
splitting concept. Branton and Southworth ( 1967 ) have shown that some 
plant cell types may indeed yield surface images of the plasmalemma 
after fracturing and suggest that the nature of bonding (hydrophobic 
versus hydrophilic) at the cell surface may determine the location of 
the fracture. 


Membrane Particles 

Particles have been observed on virtually all membrane faces since the 
beginnings of freeze-etch methodology. Discounting those instances 
where contamination appeared to be the source, these particles (ranging 
from 70 to 150 A in size ) appear in many cases to be integral components 
of the membranes on or in which they reside. Figure 2.25 illustrates such 
particles associated with rat sperm middle piece; note the patches of 
close hexagonal packing ( arrow ) . In a number of instances such particles 
appear to play an important role in membrane function. For example, 
the quantasomes of chloroplasts have been characterized by many 






- "^*>T- 



•ST - '"^iSr* •** 

Fig. 2.25 Portion of rat sperm middle piece. A bit of plasma membrane bearing 
a cluster of closely packed particles (arrow) can be seen overlying the helically 
oriented mitochondria (M). 


methods in addition to freeze-etching (Park, 1966). These particles now 
have been established as having definite chemical activities and are 
accepted as existing in close association with other chloroplast mem- 
brane components. 

The close-packed particles associated with so-called "gap" junctions 
can be similarly demonstrated with thin sections, negative staining as 
well as freeze-etching (Fig. 2.16), and have been partially characterized 
chemically ( Goodenough and Revel, 1970 ) . There is no reason to believe 
that many of the other particulates seen in association with membrane 
faces do not also possess important metabolic and structural activities 
that we usually associate with membrane function. Indeed, Branton 
(1969) has generalized these findings and suggested that a quantitative 
relationship exists between the particle content of a given membrane 
system and its "activity" in a functional sense. Thus, it would appear 
that we have a spectrum of membrane types based on particulates char- 
acterized by freeze-etching with the nonparticulate myelin sheath in the 
most inactive category and chloroplast or mitochondrial (inner) mem- 
branes in the highly active category. A given membrane, plasmalemma 
for example, may also have highly active as well as relatively inactive 


A number of outstanding workers in the freeze-etching field have given gen- 
erously of their illustrative material, preprints and personal insights, resulting 
in far greater coverage than would otherwise have been provided in this work. 
Fear of omitting some prompts me to refrain from an extensive name listing. 
Those who read this chapter will certainly become familiar with the names of 
these individuals. 

Technical assistance in some of the author's work was provided by Mrs. 
Jaroslava Krivacek and parts of the freeze-fracture device described were 
machined by Mr. James Rankin. Expert secretarial services were performed by 
Mrs. Doris Ringer and portions of the author's research were supported by 
grants from the National Science Foundation (GB 8165) and United States 
Public Health Service (GM 16598). 


Bachmann, L., Abermann, R., and Zingsheim, H. P. (1969). Hochaufiosende 

Gefrieratzung. Histochemie, 20, 133. 
Bertaud, W. S., Rayns, D. G., and Simpson, F. O. (1968). Myofilaments in 

frozen-etched muscle. Nature, 220, 381. 
Bradley, D. E. (1959). High resolution shadow casting technique for the elec- 


tron microscope using the simultaneous evaporation of platinum and carbon. 
Brit. J. Appl. Phys., 10, 198. 

. (1965). Replica And Shadowing Technique. In: Techniques For Elec- 
tron Microscopy (Kay, D. H., ed.), F. A. Davis and Co., Philadelphia. 

Branton, D. (1966). Fracture faces of frozen membranes. Proc. Nat. Acad. 
ScL, 55, 1048. 

. (1969). Membrane structure. Ann. Rev. Plant Physiol, 20, 209. 

Branton, D., and Southworth, D. (1967). Fracture faces of frozen Chlorella 
and Saccharomyces cells. Exp. Cell Res., 47, 618. 

Bullivant, S. (1970). Present Status Of Freezing Technique. In: Some Biolog- 
ical Techniques In Electron Microscopy. (Parson, D. F., ed.) Academic 
Press, New York. 

Bullivant, S., and Ames, A. (1966). A simple freeze-fracture replication method 
for electron microscopy. /. Cell Biol., 29, 435. 

Bullivant, S., Weinstein, R. S., and Someda, K. (1968). The type II simple 
freeze-cleave device. J. Cell Biol., 39, 19A. 

Chalcroft, J. P., and Bullivant, S. (1970). An interpretation of liver cell mem- 
brane and junction structure based on observation of freeze fracture replicas 
of both sides of the fracture. 7. Cell Biol, 47, 49. 

Christensen, A. K. (1967). A simple, way to cut frozen thin sections of tissue at 
liquid nitrogen temperatures. Anat. Rec, 157, 227. 

Clark, A. W., and Branton, D. (1968). Fracture faces in frozen outer segments 
from the guinea pig retina, Z. Zellforsch., 91, 586. 

Doebbler, G. F., Rowe, A. W., and Rinfret, A. P. (1966). Freezing of mam- 
malian blood and its constituents. In: Cryobiology (Meryman, H. T., ed.), 
Academic Press, New York. 

Frey-Wyssling, A., and Muhlethaler, K. (1965). Ultrastructural Plant Cytology, 
pp. 167, 172. Elsevier Publishing Co., New York. 

Goodenough, D. A., and Revel, J. P. (1970). A fine structural analysis of 
intercellular junctions in the mouse liver. /. Cell Biol, 45, 272. 

Hall, C. E. (1950). A low temperature replica method for electron microscopy. 
/. Appl. Phys., 21, 61. 

Hayat, M. A. (1970). Principles And Techniques Of Electron Microscopy: 
Biological Applications. Vol. 1. Van Nostrand Reinhold Company, New 

Koehler, J. K. (1966). Fine structure observations in frozen -etched bovine 
spermatozoa. /. Ultrastruct. Res., 16, 359. 

. (1968). Freeze-etching observations on nucleated erythrocytes with 

special reference to the nuclear and plasma membranes. Z. Zellforsch., 
85, 1. 

. (1968). The Techniques And Applications Of Freeze-etching In Ultra- 
structure Research. In: Advances In Biol. Med. Physics (Lawrence J., and 
Gofman, J. W., eds.) Vol. 12, 1—84. Academic Press, New York. 

Kreutziger, G. O. (1970). The cryopump freeze-etch device. /. Cell Biol, 47, 


Lovelock, J. E. (1953). The mechanism of the protective action of glycerol 

against haemolysis by freezing and thawing. Biochim. Biophys. Acta, 11, 28. 
Mazur, P. (1970). Cryobiology: the freezing of biological systems. Science, 

168, 939. 
McAlear, J. H., and Kreutziger, G. O. (1967). Freeze-etching with radiant 

energy in a simple cold block device. Proc. 25th Ann. Meet. Electron Micros. 

Soc. Am., p. 116. Claitor's Publishing Division, Baton Rouge. 
Meryman, H. T. (1957). Physical limitations of the rapid freezing method. 

Proc. Roy. Soc. B, 147, 452. 
Meryman, H. T., and Kafig, E. (1955). The study of frozen specimens, ice 

crystals and ice crystal growth by electron microscopy. Naval Med. Res. 

Inst. Rept. NM 000018.01.09 13, 529. 
Meyer, H. W. (1969). Die Darstellung von Hefemitochondrien durch Gef- 

rieratzung. Z. Allg. Mikrobiologie, 9, 603. 
Moor, H. (1966). Use of freeze-etching in the study of biological ultrastruc- 

ture. Int. Rev. Exp. Pathol, p. 179. Academic Press, New York. 
. (1967). Der Feinbau der Mikrotubuli in Hefe nach Gefrieratzung. 

Protoplasma, 64, 89. 
Moor, H., and Miihlethaler, K. (1963). Fine structure in frozen etched yeast 

cells. J. Cell. Biol, 17, 609. 
Moor, H., Miihlethaler, K., Waldner, H., and Frey-Wyssling, A. (1961). A 

new freezing ultramicrotome. /. Biophys. Biochem. Cytol., 10, 1. 
Moor, H., and Riehle, U. (1968). Snap freezing under high pressure: a new 

fixation technique for freeze-etching. 4th Europ. Reg. Conf. Electron Micros., 

2, 33 (Bocciarelli, D. S., ed.) Tipografia Poliglotta Vaticana, Rome. 
Miihlethaler, K., Wehrli, E., and Moor, H. (1970). Double fracturing methods 

for freeze-etching. Proc. 7th Int. Cong. Electron Micros., 1, 449 (Favard, 

P., ed.) Societe Francaise De Microscopie Electronique, Paris. 
Nash, T. (1966). Chemical Constitution And Physical Properties Of Com- 
pounds Able To Protect Living Cells Against Damage Due To Freezing And 

Thawing. In: Cryobiology (Meryman, H. T. ed.), Academic Press, New 

Park, R. B. (1966). Subunits of chloroplast structure and quantum conversion 

in photosynthesis. Int. Rev. Cytol., 20, 67. 
Pinto da Silva, P., and Branton, D. (1970). Membrane splitting in freeze- 
etching. Covalently bound ferritin as a membrane marker. J. Cell Biol., 45, 

Pinto da Silva, P., Douglas, S. D., and Branton, D. (1971). Localization of A 

antigen sites on human erythrocyte ghosts. Nature (in press). 
Polge, C. (1953). The storage of bull semen at low temperatures. Vet. Res., 

65, 557. 
Polge, C, Smith, A. U., and Parkes, A. S. (1949). Revival of spermatozoa 

after vitrification and dehydration at low temperatures. Nature, 164, 666. 
Sassen, M. M. A., Remsen, C. C, and Hess, W. M. (1967). Fine structure of 

Penicillium megasporum conidiospores. Protoplasma, 64, 75. 


Sherman, J. K. (1962). Survival of higher animal cells after the formation and 

dissolution of intracellular ice. Anat. Rec, 144, 171. 
. (1963). Improved methods of preservation of human spermatozoa by 

freezing and freeze-drying. Fertil. Steril., 14, 49. 

(1969). Freeze-thaw induced structural changes in cells. /. Cryosurgery, 

2, 123. 

Sleytr, U. (1970). Die Gefrieratzung korrespondierender brunchhalften: ein 
neuer Weg zur Aufklarung von Membran strukturen. Protoplasma, 70, 101. 

Staehelin, L. A., and Kiermayer, O. (1970). Membrane differentiation in the 
Golgi complex of Micrasterias denticulata Breb: visualized bv freeze-etching. 
J. Cell Sci., 7, 787. 

Steere, R. L. (1957). Electron microscopy of structural detail in frozen biolog- 
ical specimens. /. Biophys. Biochem. Cytol., 3, 45. 

. (1969a). Freeze etching simplified. Cryobiology, 5, 306. 

. (1969b). Freeze etching and direct observation of freezing damage. 

Cryobiology, 6, 137. 

Steere, R. L., and Moseley, M. (1969). New dimensions in freeze-etching. 
Proc. 27th Ann. Meet. Electron Micros. Soc. Am., p. 202. Claitor's Publish- 
ing Div., Baton Rouge. 

. (1970). Modified freeze-etch equipment permits simultaneous prepara- 
tions of 2-10 double replicas. Proc. 7th Int. Cong. Electron Micros., 1, 451. 
Societe Francaise De Microscopie Electronique, Paris. 

Tillack, T. W., and Marchesi, V. T. (1970). Demonstration of the outer sur- 
face of freeze-etched red blood cell membranes. /. Cell Biol., 54, 649. 

Van Gool, A. P., Lambert, R., and Landelout, H. (1969). The fine structure 
of frozen etched Nitrobacter cells. Arch. Mikrohiol, 69, 281. 

Wartiovaara, J., and Branton, D. (1970). Visualization of ribosomes by freeze- 
etching. Exp. Cell Res., 61, 403. 

Wehrli, E., Miihlethaler, K., and Moor, H. (1970). Membrane structure as 
seen with a double replica method for freeze-fracturing. Exp. Cell Res., 59, 

Zingsheim, H. P., Abermann, R., and Bachmann, L. (1970). Apparatus for 
ultrashadowing of freeze-etched electron microscopic specimens. J. Phys. 
Scient. Inst., 3, 39. 

3 Negative Staining 

Rudy H. Haschemeyer 

Department of Biochemistry, 

Cornell University Medical College, New York, New York 

Robert J. Myers 

Rogosin Laboratory, Department of Biochemistry, 
Cornell University Medical College, New York, New York 


^■lectron microscopes with the capability of a point to point resolution 
of better than 3 A are now commercially available. Those in the biolog- 
ical sciences with an interest in macromolecules and other particulate 
material were obviously attracted to electron microscopy in the hope that 
this technique might provide structural information by direct visualiza- 
tion of particle detail. Biological specimens for the most part, however, 
contain atoms of low atomic number, which do not scatter to an ap- 
preciable degree electrons of the energy produced in the electron 
microscope. Electron scattering is a major contribution to image forma- 
tion in electron microscopy and it was immediately apparent that al- 
though moderately-sized particles supported on thin films could be seen 
in outline, imaging high resolution detail of particles by transmission 
electron microscopy would require artificial enhancement of contrast. 

Three major methods to achieve appropriate contrast to view par- 
ticulate material preceded the advent of negative staining. 

1. Plastic embedded pellets can be stained and sectioned by normal 
tissue section techniques (e.g., see Hayat, 1970 and 1972). This 
powerful technique has been of great value and will continue to 
play a prominent role in deducing structural and biological prop- 
erties of certain systems. It has the advantage that isolated extra- 
cellular particles can also be identified in cell sections for func- 
tional, structural and biochemical correlations. The technique often 
complements negatively stained results in elucidating structural 
detail, for example, in the study of viruses containing membranes 
as a part of their structure. 

2. Shadowing with heavy metals provides excellent contrast with 
even small molecules. It is still widely used today and is discussed 
in this volume. 

3. Positive staining of particles entails contrast gain by binding heavy 
metal ions to the molecules of interest, usually through charge 
interactions. The binding of uranyl ions to nucleic acids and 
nucleoproteins ( e.g., ribosomes and viruses ) is still popular. Heavy 
metal ion binding (e.g., phosphotungstic acid) to proteins has 
been extraordinarily useful in structural studies with various fibers, 
including such biologically significant examples as muscle, collagen 
and fibrin. 



These techniques, however, often proved to be deficient in revealing 
many details present at the subparticle level. The ability to "see" sub- 
particle details such as the subunit structure of viruses and proteins by 
negative staining constitutes the prime reason for its rapid development 
and widespread popularity today. Negative staining is also often pre- 
ferred in studies where the question of interest could be redundantly 
answered by another contrasting method, because negative staining is 
a comparatively simple technique. 

Negative staining effects were used with light microscopy as early as 
1892 by Welch and Nuttall, but were not reported in specimens ob- 
served in the electron microscope until 1954, by Farrant. The next year, 
Hall (1955) purposefully negatively contrasted bushy stunt virus with 
phosphotungstic acid at pH 4.6, correctly described the nature of the 
phenomenon, and recognized the future usefulness of negatively stain- 
ing particles. In 1956, Huxley demonstrated that tobacco mosaic virus 
( TMV ) is a hollow rod from the dark outline of phosphotungstate which 
had penetrated into the 40 A diameter central hole during negative 
staining, Brenner and Home ( 1959 ) developed a simple spray technique 
for routine negative staining and demonstrated that substructural detail 
of viruses could be contrasted by this method. A simpler procedure (the 
drop method) was introduced shortly thereafter by Huxley and Zubay 

The general principles of negative staining are easy to understand. In 
one procedure, for example, a small droplet of particles in dilute solution 
with a heavy metal salt, such as potassium phosphotungstate at near- 
neutral pH (KPT), is placed on a thin support film covering a micro- 
scope grid. The droplet is removed by touching it with filter paper, leav- 
ing behind a thin film of solution which dries rapidly. During the later 
stages of evaporation, the KPT solidifies into a smooth glassy film. The 
stain surrounds particles on the support film as the result of surface ten- 
sion interactions and, to a degree, penetrates into the open irregularities 
presented at the particle surface (Fig. 3.1). Final dehydration of the 
particle occurs after stain solidification (Johnson and Home, 1970). 
This accounts in large measure for the preservation effect of nega- 
tive staining when compared to structural alterations frequently observed 
when such particles are dehydrated without embedding in stain. In 
the electron microscope, the negatively stained particles appear as 
light areas or holes because of the low scattering power of the par- 
ticle compared to the dense surrounding stain. The frequently observed 
buildup of stain immediately surrounding the particle gives the impres- 
sion of a dark halo, and both top and bottom surface irregularities, sig- 
nificantly penetrated by stain, project irregular contrast detail from 



Support film 

Fig. 3.1 Schematic representation of a particle embedded in negative stain. See 
text for details. 

Fig. 3.2 Polyma virus negatively stained with 1% potassium phosphotungstate at 
pH 7.4. Partial degradation is observed because the sample was allowed to de- 
hydrate on the grid before negative stain was applied. The particle diameter is 
approximately 450 A. 

which substructure features can be deduced. The background density 
between particles varies with the amount of residual stain deposited on 
the support. These typical characteristics are illustrated in the micro- 
graph of polyoma virus shown in Fig. 3.2. 


The technique of negative staining enjoyed almost immediate popu- 
larity, and in over a decade of use has been successfully applied to a 
wide variety of biological systems. A few brief examples may help to 
orient the reader with respect to the type of problems which have been 
profitably investigated by negative staining. The greatest effort, as esti- 
mated from the number of published micrographs, has been directed 
toward the study of viruses. The subunit arrangement of many simple 
viruses possessing icosahedral or helical symmetry (Home and Wildy, 
1961; Casper and Klug, 1962) has been determined, including the 
polyoma virus shown in Fig. 3.2 ( see also reviews by Finch and Holmes 
( 1967 ) and Home ( 1967 ) ) . Negative staining has also provided a means 
for the identification and characterization of viruses containing mem- 
branes ( Fig. 3.3 ) and in deducing the functional role of morphologically 
distinct components in bacterial viruses. In addition, negative staining 
has been used to characterize viral products which indirectly aids in the 

Fig. 3.3 KPT stained mouse mammary tumor virus. The pleomorphic shape of 
the virus and the existence of surface projections are clearly visible. x1 60,000. 
Courtesy of M. J. Lyons. 





?fflBk, /J 




111 Mill 








Khr." TDiuMiLilliD 











#J> "• ; .*t8H 



Fig. 3.4 Comparison of bacteriophage / 2 (A) with spherical particles obtained from 
self-assembly of viral protein (B) stained with uranyl oxalate at pH 6.8. The centers 
of the particles are obscured due to penetration of the stain. The particle diam- 
eters are approximately 240 A. From P. Zelazo and Ft. Haschemeyer (1970). 



solution of biochemical and genetic problems (see the review by Kozloff 
(1968) on the large DNA bacteriophages). 

Negative staining has contributed to the characterization of the 
products of partial viral degradation, interaction of viruses with anti- 
body, and of in vitro self-assembly systems. Figure 3.4 shows the com- 
parison between the appearance of bacteriophage f 2 and self-assembled 
viral protein. Other examples of the use of electron microscopy in study- 
ing the assembly of simple RNA bacteriophage and of spherical plant 
viruses can be found in excellent reviews by Hohn and Holm ( 1970 ) and 
by Bancroft (1970), respectively. Most viruses have a sufficiently char- 
acteristic appearance so that they are readily identified by negative stain- 
ing even in unfractionated cell lysates (Fig. 3.5). The method is of value 
in purification procedures, both for a criterion of purity and for locating 
small amounts of virus in a fractionation scheme, for instance, by pre- 
paring grids from cesium chloride or sucrose density gradients. 

■Q 1 -^ 




Fig. 3.5 Phosphotungstate embedded lysate of cells infected with polyoma virus. 
The characteristic virions are easily recognized against the background of cellular 
debris. The virus is better preserved than the particles shown in Fig. 3.2 because 
dehydration was carried out in the presence of the stain. 



Fig. 3.6 Rat liver free polyribosomes negatively stained with 2% uranyl acetate 
on carbon coated collodion film. From Y. Nonomura, G. Beobel and D. Sabitini, 
J. Mol. Biol., in press, with permission of the publisher. 


Negative staining has proven effective in determining subunit stoi- 
chiometry and probable symmetry of oligomeric proteins ( Haschemeyer, 
1970), and in the study of enzyme-complexes (Williams et al., 1967). 
The significant contribution of negative staining to our understanding 
of structural components is clearly demonstrated in the study of muscle 
components (Huxley, 1969). The technique has been successfully applied 
to subcellular particles such as ribosomes (Fig. 3.6) and to membrane- 
bounded cells and cellular subfractions (Muscatello and Home, 1968). 

It should be mentioned that the resolution required to confidently 
interpret a micrograph varies over a wide range. For example, the gen- 
eral shape of a bacterium and the sine-like conformation of bacterial 
flagella are clearly seen even in poorly resolved photographs at low 
magnification (Fig. 3.7). Figure 3.8 shows bacterial flagellum contamin- 
ating a preparation of ovine glutamine synthetase. The photograph is of 
poor resolution due to underfocus (actually the first of a focal series), 
but the periodicity of the flagellum is clearly seen and the general 
characteristic shape of the ezyme is sufficienty resolved to be used for 
identification in a purification scheme or for a gross criterion of purity. 

. j, ■ ■ ■ . . .. v "■■ ■ *. . 

*" * 'A -Tni' •*«# i V£_* ^T J %fJO* .-*■ ' ''JtJk'-^.i; •< • ? , Bl 

; F^ff^Bs.' i 


''■ '■' ,'* .■ 4" ' ' .' ■' *.'* •;** " "' V 1 ' 

k£' £>',*.- > «" Mi \-.i 

Fig. 3.7 Micrograph of a bacterium and its flagellum contrasted with uranyl 
acetate at pH 4.5. X20,000. 



■^''■^'^W^i '*' -^ -••'-•*•• ••'•«'• •«!*?" 


Fig. 3.8 Higher magnification micrograph of the flagellum shown in Fig. 3.7. The 
bacterium was a contaminant in a preparation of ovine glutamine synthetase. The 
limited resolution of this micrograph can be deduced from the large size of phase 
grain resulting from underfocus. The enzyme particles are about 100 A across. 
Further discussion in text. 

The background grain superimposed on particle detail, however, negates 
the interpretation of subunit structure. 

Many of these diverse systems to which negative staining methods 
have been applied are characterized by problems of isolation, handling 
and interpretation which are too unique to be discussed here. The 
scope of this chapter will be limited to the consideration of methods 
which have general applicability and to specific details which must be 
taken into account to interpret high resolution details of subcellular 
structures. Most of these methods and details are discussed specifically 
in relation to viruses and macromolecules, but many can be applied to 
other biological specimens as well. 



Electron microscope requirements for obtaining satisfactory high resolu- 
tion micrographs from negatively stained specimens are often more 
stringent than those for the average section. For such work, it is desirable 
to have a microscope capable of resolving better than 10 A. Most modern 
electron microscopes are readily capable of such resolution, but constant 
vigilance by the operator is required to maintain high performance 
through proper alignment and cleanliness of the column components. 

Generally, there is a mandatory requirement for an anticontamination 
device in the microscope and for a double condenser system to provide 
small spot illumination of the specimen. Both of these features are neces- 
sary for the routine recording of fine details, which are quickly obscured 
by contamination and beam damage. Pointed filaments improve resolu- 
tion and contrast, and "self-cleaning" objective apertures decrease the 
time required for cleaning and frequent compensation. Proper focus and 
compensation for astigmatism is an absolute requirement for obtaining 
all high resolution micrographs. Since phase contrast (Thon, 1967; 
Heindenreich, 1967) passes through a minimum at focus and has a sym- 
metrical appearance when the microscope is stigmatic ( Wischnitzer, 
1972), it is highly desirable to be able to observe phase contrast directly 
on the screen of the electron microscope at high magnification. The ease 
with which this can be accomplished varies considerably depending 
upon the microscope (even at the same magnification and current 
density). It is suggested that this feature should be an important con- 
sideration when purchasing a new instrument for use with negative 
staining. Although the authors lack personal experience with image 
intensification systems, it appears as if some are available which permit 
observation of phase contrast even at very low illumination and would 
therefore be of great value. As will be noted later, a tilting stage can be 
used to great advantage for structural studies on some particles, such as 
icosahedral viruses. 

Ancillary equipment requirements include adequate darkroom facil- 
ities and a good vacuum evaporator. The latter is needed for the pro- 
duction of carbon support films and should approach a vacuum of 10~ 5 
Torr. The vacuum evaporator should be subject to minimal contamina- 
tion with oil vapors in the chamber and should be kept clean. 


A number of alternate experimental procedures are available for the 
preparation of negatively stained specimens. In this section we present 


some of those which have proven most useful in our laboratory. We 
believe that one or more of the combinations suggested here will be 
successful for obtaining quality micrographs of a majority of negatively 
stained particles. 

Support Films 

The preparation of thin support films and procedures for mounting them 
on grids have been described by Hayat (1970) in Vol. 1 of this series 
and by Bradley ( 1965 ) . The advantage of thin carbon films as a support 
for electron microscopy of high resolution detail will be discussed later. 
Here we present the modifications routinely employed in our laboratory 
for the production of carbon films. They differ from those given by 
Hayat ( 1970 ) primarily in that the carbon is evaporated onto freshly 
cleaved mica instead of glass. We doubt that this difference is significant, 
but have not performed controlled comparisons which permit a decisive 

A sheet of mica is cut into approximately a 4 cm square which is then 
cleaved along a crystal plane. This is accomplished with the aid of a 
sharp blade (e.g., a scalpel). Once the cleavage starts, it continues with 
gentle parting pressure of the blade. Scratching of the cleaved surface 
with the blade itself should be avoided. The mica is placed about 10 cm 
below the carbon rods in the evaporator, which has been prepumped to 
about 2 X 10 -5 Torr prior to opening. The chamber is evacuated and 
the evaporation of carbon is carried out as described by Hayat (1970) 
as soon as a sufficient vacuum is attained (e.g., about 5 X 10 -5 ). 

The chamber is opened and the mica is picked up with forceps and 
gently scored on the carbon surface into squares a little larger than the 
grid diameter. The weight of a scalpel without pressure is adequate to 
score the carbon without cutting into the mica itself. The mica with 
carbon side up is gently lowered onto the surface of distilled water at an 
angle of no more than 15° with the water's surface. The edge of the mica 
is allowed to touch the surface of the water for about 15 sec without im- 
mersion, during which time surface tension forces liquid between the 
carbon layer and the mica. The carbon layer floats off onto the water 
surface as the mica is slowly pushed under the surface and finally 
dropped to the bottom of the vessel. 

Grids are held at the edge with forceps, lowered below the water 
surface, and brought up under the carbon squares, which then adhere to 
the grid as it is lifted out of the water. The corners of carbon film which 
may wrap around the grid are removed by touching the bottom of the 
grid to filter paper while they are still wet. The grids, which are ready 


to use as soon as they dry, are prepared fresh each day as needed. Even 
when used immediately, some variation in the spreading of specimen and 
stain may be noted between different batches of grids, presumably re- 
flecting minor, uncontrollable differences in the amount of oil vapors or 
charge on the carbon surface. 

Similar contamination probably accounts for the difficulty sometimes 
encountered in stripping the carbon from the mica surface. More often 
than not, the very next preparation will strip properly. If the problem 
persists, cleaning the vacuum chamber and/or changing the vacuum 
pump oil may alleviate the difficulty. It should be noted that with many 
evaporators, including some with liquid nitrogen traps, the major source 
of oil vapor contamination of the chamber occurs during prepumping 
with the forepump and a good filter system in this line should prove 
very advantageous. Brenner and Home (1959) noted that superior re- 
sults were obtained with evaporation plants fitted with mercury rather 
than oil pumps. 

The collodion, Formvar, carbon coated plastic, and perforated support 
films used in our laboratory are prepared essentially as described by 
Hayat (1970). 

Preparation of Negative Stains 

The properties of a good negative stain include nonreactivity with the 
specimen, high solubility, high density (i.e., high scattering power for 
electrons), and a high melting point to prevent melting or sublimation 
in the electron beam. A good stain should have sufficiently small mole- 
cules to penetrate into particle irregularities that one desires to see, and 
the solidified film of stain should present a smooth nongranular appear- 
ance. The properties of several potential negative stains are discussed 
and reviewed by Valentine and Home (1962) and by Home (1967). 
The preparation of some particularly useful negative stains is described 

Uranyl oxalate is a favorite stain in our laboratory for negatively 
staining small proteins (and viruses as well) above the isoelectric point. 
It provides higher contrast and greater penetrability into intramolecular 
crevices than do the phosphotungstate stains. Granularity in thinly 
stained areas is negligible. The stain is prepared essentially as originally 
described by Mellama et al. ( 1967 ) . A solution containing 0.5% ( 12 mM ) 
uranyl acetate and an equimolar amount ( 12 mM ) of oxalic acid is ad- 
justed to a pH between 6.5 and 6.8 with dilute NH 4 OH. The original 
workers used concentrated ammonia, but we find that local over-titration 
(which results in an insoluble precipitate) is more easily avoided by 


titrating with dilute base. Titration is completed with reasonable speed 
by moderate stirring; an opaque container is used because of the light 
sensitivity of the salt. The stain is then divided into many small aliquotes, 
quick frozen, and stored at — 30° C where it is stable for at least a year. 
The solution is thawed just prior to use. Mellama et al. (1967) report 
that the stain has a useful pH range of 5-7 and that the stain is stable 
below pH 6 for up to 48 hr at 4°C. 

Phosphotungstic acid may be titrated to a pH near neutrality with 
normal KOH or NaOH to provide a negative stain (often termed KPT 
and NaPT, respectively) of proven utility and reliability. The neutral 
stain is often called simply phosphotungstate, without specifying whether 
it is the sodium or potassium salt. We have detected no differences be- 
tween the two salts, although preservative advantages of one over the 
other have occasionally been reported. We have employed the stain at 
concentrations of 0.5-2% and a pH range of 6-8; 1% solution at pH 7 is 
the "standard" stain in our laboratory. Phosphotungstate is stable for long 
periods of time if care is taken to prevent a pH drop by dissolution of 
carbon dioxide; otherwise, pH readjustment of the unbuffered stain will 
be required. Phosphotungstate has the advantage that relatively high 
concentrations of nonvolatile salts (particularly phosphate buffers) can 
be tolerated in negative embedding without detrimental effects. The 
stain may also be advantageous for preservation of structural details in 
some systems. 

Several excellent negative stains with properties similar to phos- 
photungstate have also been routinely used. One of these, sodium sili- 
cotungstate, was often employed in the phenomenally successful studies 
of the late Dr. R. C. Valentine (see Haschemeyer (1970) for further 
comments and original references). Ammonium molybdate (useful pH 
range, 7.0-7.4) may also prove advantageous for some studies. It has 
provided one-sided images of a virus (Nagington et al, 1964) and is 
often used to negatively stain catalase crystals (which are insoluble in 
the stain) for magnification calibration. A recent study by Muscatello 
and Home (1968) suggests that ammonium molybdate is a markedly 
superior stain for contrasting the elementary structure of membrane- 
bounded systems such as microsomes, kidney mitochondria and cell 
membranes from red blood cells. These authors note that the tonicity 
of this stain (e.g., a 1% stain is roughly comparable to 0.12 M sucrose) 
can be adjusted over a substantial range to match the requirements for 
stabilizing a particular membrane structure. 

Negative staining of particles below the isoelectric point with uranyl 
acetate is highly recommended (Van Bruggen et al., 1960). Background 
granularity is essentially absent in micrographs of specimens stained 


with uranyl acetate, and details are seen with remarkable contrast. This 
stain also spreads more reliably over a wider range of particle concentra- 
tion than do some of the other negative stains. We use uranyl acetate at 
concentrations of 0.2-0.5%. The freshly dissolved stain has a convenient 
pH of about 4.5, but it may be modified by titration to encompass a pH 
range from below 4 to about 5.5. Uranyl acetate is prepared shortly 
before use (15-30 min are required to solubilize the salt) and is 
stable for a few hours in the dark. Uranyl formate (Leberman, 1965) 
may be preferred over the acetate salt for structural studies where 
greater stain penetration into narrow particle irregularities is required. 
Upward titration of the uranyl formate (the pH is about 3.5 when not 
titrated ) with ammonium hydroxide to a pH of 4.5-5.2 is recommended. 

Embedding the Particles in Negative Stain 

Drop Method 

The drop method (in the form of one of several variations) is the sim- 
plest procedure for embedding the specimen in negative stain. This 
method is the one most frequently used in the authors' laboratory and, 
with a few exceptions, generally produces results which are as good as, 
or better than, more complicated methods. The standard procedure for 
staining above the isoelectric point with phosphotungstate, ammonium 
molybdate, silicotungstate oxalate, and uranyl oxalate is as follows: 

1. Grids mounted with plastic or carbon supports are picked up by 
the edge with self-clamping forceps and placed on a flat surface 
with the support side up. 

2. The specimen is diluted with water (or low concentration buffer) 
to the appropriate concentration. If a nearly ideal concentration 
cannot be estimated from experience, several grids may be pre- 
pared from tenfold serial dilutions of an original dilution thought 
to be still too concentrated (for enzymes and most viruses, 0.1 
mg/ml will readily fulfill the "too concentrated" criterion). The 
ideal concentration depends on the specimen, the type of stain 
used and its concentration, and the nature of the support film. It 
is emphasized that macromolecule concentrations used for electron 
microscopy are quite low compared to those encountered in many 
other physical studies. Dissociation of enzymes and viruses may 
result at these low concentrations and/or as the result of the pH 
employed, or by interaction of the particles with the stain. Both 
phosphotungstate and uranyl oxalate are essentially unbuffered at 


a pH near neutrality, so that it may be necessary to titrate the 
sample to the appropriate pH even after marked dilution. This is 
particularly true for samples at a pH above 7.0, using uranyl 
oxalate if precipitation in the stain is to be prevented. 

3. A small droplet of the sample is placed on the support to form a 
"bead" nearly extending to the edge of the grid. A fine-tipped 
micropipette containing 2-5 /A of sample is usually ideal for this 

4. The droplet is touched with the torn edge of Whatman #1 filter 
paper to remove most of the excess liquid, and a drop of the stain 
is immediately applied before the residual film of sample can dry. 

5. After from a few seconds to 1 min ( 30 sec is standard ) the droplet 
of the stain is mostly removed by touching it with torn filter paper. 
The residual film quickly dries and is then ready for observation 
in the electron microscope. 

The particles finally observed include some which become physically 
attached to the grid in step 4 and some which were still in solution in 
the residual film after removal of excess stain in step 5. Consequently, the 
number of particles observed can be markedly influenced by the time 
the sample is in contact with the grid in step 4. The authors occa- 
sionally use durations of as long as 30 min for samples which are so 
dilute that a proper particle distribution could not be directly obtained. 
In addition to a concentration series, a time series (in step 4) is some- 
times convenient to obtain proper distribution of particles. 

The sample may also be mixed with the stain and applied as a single 
drop. The choice of this alternative depends upon the stability of the 
particles in the stain environment. Certain particles are less stable in the 
stain, and they may precipitate or there may be a marked interaction 
(through ion binding) of the particles with the stain. For other particles, 
particularly those which are unstable at low ionic strength, direct mixing 
with the stain can be advantageous. If higher particle concentrations are 
used, the premixed sample and stain can be rapidly applied to the grid 
and the excess withdrawn immediately. 

To a degree, the possible deleterious effects of a stain can be moni- 
tored by measurements of biological activity and physical properties 
(e.g., sedimentation velocity). Because various negative stains may ex- 
hibit different effects when admixed in solution, the best approach is 
through experimentation (see also Home, 1967). 

Staining of particles below their isoelectric point with uranyl acetate 
or uranyl formate may often be accomplished as described above, par- 
ticularly if the sample is first brought to the appropriate pH (e.g., by 


dialysis or dilution into low concentrations of ammonium acetate at the 
proper pH). The problem often encountered is that many proteins can- 
not readily pass through the isoelectric point without precipitation, or 
the protein may aggregate, dissociate, or denature at low pHs. The 
authors have routinely stained specimens ( especially oligomeric enzymes ) 
available in solution at near neutral pH, with uranyl acetate directly on 
the grid. The diluted sample, still near neutral pH but in low concentra- 
tion buffer as the result of dilution, is placed as a 2-3 fil drop on the grid 
and allowed to remain there for about 4 min. During this time, many 
particles adhere to the grid. 

An equal sized drop of uranyl acetate is then added to the sample 
drop and the excess liquid is removed with filter paper after 4-5 min, as 
described above. During this stage, the pH changes to near that of the 
stain, and the particles, both free in solution and adhering to the grid, 
are titrated. Aggregates may form, but large areas of properly stained 
and spaced particles are still available for microscopy, presumably con- 
sisting of particles which adhered to the grid during the first step. 
Aggregates of sufficient size, which could settle under gravity, are pre- 
vented from depositing on the support film by turning the grid over so 
that the droplet is hanging down. Surprisingly, the simple procedure de- 
scribed above has always produced well-stained specimens of the systems 
studied in the authors' laboratory. The advantageous spreading of the 
stain achieved with this procedure is not entirely understood. 

The drop method may also be used for samples dissolved in solvent 
systems which are inappropriate for negative staining, such as those con- 
taining high concentrations of nonvolatile salts or sucrose. In this case, 
the sample is applied and left in contact with the grid for a sufficient 
duration to permit particles to adhere to the grid. The grid is then 
"washed" with several droplets of appropriate solvent (or stain solution 
itself) by withdrawing successive drops with filter paper and applying a 
new drop before drying can occur. The same principle can be applied to 
"on grid reactions" such as macromolecular associations. The first macro- 
molecule is placed on the grid and the second is permitted to react only 
with those particles which adhered to the support after intermediate 
washing steps. 

Occasionally, the end result of a methodical investigation to optimize 
conditions for staining acquires an almost magical air (Moore et al., 
1970). We mention this since there are an almost infinite number of small 
procedural variations which could be attempted to improve results on 
a particular system. In our experiences with small enzymes and viruses, 
we have seldom found it necessary to depart from the standard pro- 
cedure outlined above. 


Finally, we should like to call attention to the fact that partial dis- 
ruption or changes in tertiary structure in particles during negative stain- 
ing can occasionally be advantageous. A desirable effect is sometimes 
obtained by allowing the sample to totally dehydrate on the grid before 
applying the stain. 

Spray Method 

Several spraying devices, suitable for various types of samples, have been 
described. Diagramatic representations of typical apparatus and a more 
detailed discussion may be found in the reviews by Home ( 1965a and b). 
The low pressure Vaponefrin glass nebulizer ( commonly used as a throat 
spray) used by Brenner and Home (1959) in the first method described 
for negative staining could be conveniently applied in a routine manner. 
They mixed equal aliquots of a virus sample with 2% potassium phos- 
photungstate at pH 7.5, placed it in the nebulizer, and sprayed the mix- 
ture onto grids as fine droplets 5-20 /x in diameter. 

A high-velocity spray gun of the type designed by Backus and Wil- 
liams ( 1949 ) may also be used for negative staining. Only small volumes 
are required, which may be applied as a premixed volume of sample and 
stain, or the stain may be drawn into the capillary first, followed by an 
air space, and then the sample. The latter method results in an over- 
lap spray pattern that assures minimal contact of stain with sample 
(Hoglund, 1968; Haschemeyer, 1968; see also the cross-spray method of 
Fernandez-Moran, 1962). 

The high-pressure spray gun produces droplet patterns in such a 
manner that some drops are completely contained within the area of a 
single grid square. This permits the observation of all particles contained 
in that volume, a feature which is used advantageously for quantitative 
determinations of particle ratios (see below). It is thought that the 
droplets produced by the high-pressure spray gun dry in a fraction of a 
second. This procedure could therefore be useful to prevent various time- 
dependent structural changes that might be manifested during the 
slower stain concentration and dehydration process encountered with 
the drop method. 

Micrographs of particles applied with either high- or low-pressure 
spray guns showed no improvement in the preservation of detail over the 
drop method for systems studied in our laboratory, although improved 
results might be expected for other systems. However, one advantage of 
the high pressure spray method is that the preferential adherence to the 
grid of one type of particle over another type of particle in a mixture 
cannot occur (Lubin, 1969; Haschemeyer, 1970). It is otherwise quite 


possible for a "sticky" contaminant of minor proportions to be accorded 
undue significance in samples negatively embedded by the drop pro- 

Float Method 

This method is a variation of that described for the drop procedure with 
intermediate washing. The sample is applied to the grid either as a drop 
or by floating the grid upside down on a small volume of sample (e.g., 
a few drops on a watch glass ) . After an appropriate duration ( dependent 
upon particle concentration ) to permit particles to adhere to the support 
under the environmental conditions of the applied solvent, the grid is 
transferred to float on a drop of the staining solution. The stain diffuses 
to the support surface, and free particles diffuse out into the large stain 
volume rather than to the grid. The perturbing influence of the solvent 
change on the particles is thus minimally reflected in the final micro- 
graph. Clearly, some particles initially adhering to the grid will be 
dislodged or modified when encountering the stain environment. There- 
fore, quantitative information concerning particle dissociation-associa- 
tion, etc., should be viewed with caution. However, we believe this 
method to be the best available for studies on environmental perturba- 
tions with refractory solvent systems (e.g., high concentrations of urea, 
guanidine-HCI, ammonium sulfate, etc.). 

Parsons ( 1963 ) has described a surface-spreading technique for form- 
ing a layer of lysed cells or cell components on the surface of a solution 
of 1% KPT. The grid is then touched to the surface or floated on it and 
excess stain removed with filter paper as detailed above. 


Markedly superior preservation of tertiary and quaternary structure of 
particles subjected to the hostile influences encountered during negative 
staining is sometimes noted after chemical fixation. The formation of 
covalently-linked intra- and inter-subunit crosslinkages imposes geomet- 
rical constraints to structural modification. The chemistry of some useful 
reagents, notably reaction with low concentrations (0.5-2%) of gluta- 
raldehyde and its polymers at neutral pH, has been discussed by Hayat 

An important consideration in fixing particles is the possibility of 
forming inter-particle crosslinks. This is usually minimized by the fact 
that fixation is applied for limited durations (e.g., less than 30 min) to 
extremely dilute samples. If this is not possible, for example, in the 


application of fixation to prevent molecular dissociation at the dilution 
required for successful microscopy, fixation may be modified by dis- 
charging unreacted ends of gluteraldehyde with an excess of appropriate 
small molecules such as lysine. 

Alternatively, a bifunctional reagent such as dimethylsuberimidate 
(Wold, 1967; Davies and Stark, 1970) may be used. This reagent may 
be preferred because: (1) the reagent preserves the positive charge at 
lysine residues to reduce conformation changes, and (2) any unreacted 
ends of the bifunctional reagent "self-destruct" by hydrolysis at a con- 
venient rate. We have found that the degree of inter-subunit crosslinking 
optimally introduced into several oligomeric enzymes is essentially the 
same for dimethylsuberimidate and glutaraldehyde ( Trotta, Meister, and 
Haschemeyer, unpublished results ) . Chemical fixation to prevent quater- 
nary disruption ought to be further improved by using bifunctional 
reagents with a longer "reach." 


In the previous section we outlined some of the more useful alternatives 
which may be used for negative staining. Even though the list is far from 
complete, the number of combinations and modifications include some 
which most microscopists seldom implement, except under unusual cir- 
cumstances. A cursory evaluation of published results indicates that most 
laboratories emphasize a single procedure for negative staining. We note, 
for example, that almost all types of viruses, including bacteriophages, 
rod-like plant viruses, icosahedral viruses, membrane-bound mammalian 
viruses, etc., are reasonably well preserved and show subparticle detail 
when stained with neutral potassium phosphotungstate on carbon grids 
with the drop method. On the other hand, in a series of outstanding 
structural studies on icosahedral viruses, A. Klug and his associates sug- 
gest that uranyl acetate is preferred according to the standard published 
procedure (Finch and Klug, 1966). 

Clearly, the adoption of a single standard procedure which serves to 
answer the questions posed is all that is required. In the case of icosa- 
hedral and rod shaped viruses, bacteriophages of many kinds, and even 
membrane-bound viruses, substructural detail can be observed when 
they are stained by almost any of the combinations described in the pre- 
vious section. In the case of the 12 subunit enzyme glutamine synthetase 
in E. coli, studied in detail by Valentine et al. (1968) and Haschemeyer 
( 1968 ) , the stacked hexagon structure can be seen by any of the staining 
procedures discussed. At the opposite extreme have been the repeated 
attempts in our laboratory to obtain convincingly interpretable micro- 


! V.»- '•&£&" "3E_ Sv -4 sffi^ 

MHSPrai- .- i'-sv, 

A B 

Fig. 3.9 Aspartate-/3-decarboxylase negatively stained with A: uranyl acetate, 
pH 4.5; B: potassium phosphotungstate pH 7.2; C: uranyl oxalate, pH 6.8. The 
mean particle diameter is 130 A. 

graphs of the blood protein fibrinogen. To date, none of the techniques 
discussed and their variations have provided an entirely satisfactory 


The novice to negative staining must, therefore, recognize that the 
success encountered with this technique may range from the "cannot 
fail" variety to the "impossible" systems. Often, intermediate cases are ob- 
served in which one staining procedure is preferrable for preservation or 
for displaying significantly improved details. Also, it is possible that 
differences in details observed with alternate methods of staining can 
provide additional information to aid interpretation, as has been sug- 
gested in our study of ovine glutamine synthetase ( Haschemeyer, 1968 ) . 
Even redundant information, provided by several procedures, increases 
confidence in the results. The enzyme aspartate-/3-decarboxylase, for 
instance, presents a significantly different appearance when stained, with 



Fig. 3.9 (continued). 

uranyl acetate, KPT, or uranyl oxalate (Fig. 3.9). All other staining vari- 
ations attempted yielded results which were essentially identical to one 
of the above. We were able to present a reasonable interpretation of the 
results (Bowers et ah, 1970) only after the uranyl oxalate studies were 
completed and, at best, could offer only an ad hoc explanation as to why 
the enzyme appeared differently under various staining conditions. 

Such observations have prompted us to adopt a protocol for the in- 
vestigation of new systems, which we present below. Such a protocol 
ought to be modified by a priori knowledge of stability, interactions, etc., 
which may be unique to the system under investigation. 

Grids are coated with thin carbon supports to permit drift free, high 
resolution studies if required. Since there is a relationship between the 
total object thickness (including the support) and ultimate resolution 
(Cosslett, 1971), it is advantageous to keep the carbon film as thin as 
possible. A thickness of 60-100 A is suitable for most work. Thinner sup- 
ports may require sample deposition by the spray method or the addi- 
tional support of a holey film or net (Bradley, 1965). As a diagnostic aid 


for poor grids and in some cases for deteriorated stain solutions, col- 
lodion or Formvar grids are prepared to differentiate between spreading 
vs. specimen problems. The plastic supports are much less stable than 
carbon, but they have the advantages of greater reproducibility and 
uniform spreading because they are more hydrophilic. For a procedural 
control one should prepare specimens having good spreading properties. 
Bacteriophage T2, tobacco mosaic virus, and E. coli glutamine synthetase 
are used for this purpose in our laboratory. 

Embedding is carried out using the drop method with uranyl acetate, 
uranyl oxalate, or potassium phosphotungstate. In preparing grids, one 
must be aware that this process cannot be rigidly controlled. Therefore, 
grids identically handled may occasionally show few stained areas for 
observation or abnormally thick "puddled" areas, even though duration 
and concentration variables were optimal. Examination of two or three 
grids identically prepared is of value in limiting the number of erroneous 
conclusions which might be drawn from results that reflect abnormal 
stain spreading or particle distribution. 

Fixation is routinely attempted during the initial stages of a detailed 
study to see if the results of negative staining can be improved in this 
way. As noted earlier, care must be taken to correctly interpret the 
existence of possible artifacts, such as interparticle crosslinking. 



Equipment requirement have already been discussed and the general 
operation of the electron microscope is reviewed in Volume 3. Special 
requirements which must be met for interpretation of high resolution 
detail are discussed in some detail in the interpretation section. Here, 
we briefly summarize some typical steps followed in our laboratory in 
order to obtain micrographs of negatively stained specimens which are 
to be studied in detail at a later time. 

The electron microscope is operated at 60-100 kV with a 35-50 (jl ob- 
jective aperture, double condenser illumination, and with the anticon- 
tamination device in place. The grid is exposed to the beam at low 
magnification and low beam current, and is scanned until a square is 
located that appears to contain a significant area of thinly spread stain 
(Fig. 3.10). An edge of this area is inspected at high magnification to 
ascertain that the contrast, preservation and distribution of particles is 
sufficiently interesting to warrant photography. The beam spot size 
should be such that adjacent areas are not illuminated during visual 


Fig. 3.10 Low magnification micrograph of a grid square exhibiting good stain 
distribution. The specimen is bacteriophage T 2 (barely visible as white dots sur- 
rounded by a dark halo of thicker stain) embedded in PTA at pH 7.0. X2.500. 

examination. The magnification is now adjusted to the level to be used 
for photography. 

Although lower magnification photographs are desirable to obtain 
large fields of view (provided that photographic enlargement need not 
be excessive, as discussed below), we more routinely choose a high 
magnification consistent with observing phase contrast near focus with 
binocular viewing of the fluorescent screen. The distance between plates 
and the screen varies among instruments, so that the final magnification 
desired depends upon the type of electron microscope used. Phase con- 
trast is observed in a nonstained area of support film. With a little prac- 
tice (photographically confirmed) the operator can learn to locate the 
focal position, representing a few tenths of a micron under focus, with 
reasonable reproducibility, since at focus phase grain is of minimum 
size and contrast. At the same time, adequate compensation for astig- 
matism is confirmed. 

For highest resolution, beam damage and contamination to the 
specimen must be limited by minimal exposure to electrons. Therefore, 
visual observation and focus readjustment at the area to be photographed 


should be avoided. We assume, therefore, that the area of focus adjust- 
ment was sufficiently close to the area to be photographed that focal 
changes in moving the stage to the new area will be minimal. On our 
microscopes, stage drift is essentially absent, and thermal drift is also 
absent when using thin carbon grids. Consequently, the area to be 
photographed is selected, brought into the beam based on its supposed 
desirability from examination of a nearby area, and is immediately pho- 
tographed, either as a single "hit or miss" shot with minimal beam 
damage, or as a focal series obtained as quickly as possible ( see also the 
discussion and references given in the interpretation section of this re- 
view.) The above procedure is repeated on a sufficient number of grid 
areas from 2 or 3 preparations until enough photographs are obtained at 
the required resolution to demonstrate that the appearance of the par- 
ticles is representative. 

A considerably more relaxed procedure can often be adopted since 
most subparticle details are not obscured by a 0.5-2 min exposure to the 
beam. Under such circumstances, there is ample time to search for 
optimal areas by visual examination and for refocusing. The use of 
plastic supports may even be possible in areas where thermal drift stops 
in a reasonable time period. Clearly, the use of low beam currents would 
also be helpful during microscopy. In the procedure presented here, the 
beam current is chosen as the lowest value consistent with observation 
of the phase image. In-focus photographs can be obtained with less 
illumination, although less frequently, unless an appropriate image in- 
tensification system is available. 

Dark Room Technique 

The photographic plate should be dimensionally stable (glass or Estar 
base) and should have a fine grain and a low fog level. Several ideally 
suited products are commercially available and the manufacturer's in- 
structions for exposure and development should be followed. Good dark 
room technique is an absolute requirement for the microscopist, since the 
final evaluations and published photographs are made from enlarged 
prints. We cannot detail the darkroom procedures here, but do call atten- 
tion to several factors which are not always self-evident. 

1. Enlargers with point source of illumination give better contrast 
and resolution. 

2. Enlargements up to 4 or 5 times are ideal. Higher enlargements 
should be checked for the appearance of photographic grain and 
marked aberrations. 


3. Prints should have a full range of tones. In particular, photographs 
should show details of background granularity. Printing on paper 
of such high contrast that some gray tones which are visible on 
normal contrast paper are lost should be viewed with caution. 

4. Exact dimensions are best obtained from the plate itself, since 
many enlargers manifest significant lens aberrations, particularly 
near the periphery. 

For additional information, the reader is referred to Wischnitzer (1972). 


Magnification Calibration 

The determination of linear dimensions presupposes that the instrument 
magnification factor is accurately known. Approximate values of the 
magnification are generally provided by the manufacturer based on 
theoretical or practical measurements of lens strength, etc., and are 
applicable when lens currents are properly adjusted or measured. For 
critical work, this estimation is not sufficiently accurate, and further 
calibration against a standard object of known dimensions is required. 
This calibration may be obtained from photographing the calibration 
object on a separate grid but with all microscope settings (voltage, lens 
currents, etc. ) identically controlled. Such a procedure is termed external 
calibration, which is satisfactory for instruments where hysteresis in lens 
magnification is either absent or can be circumvented. This is accom- 
plished by introducing new specimens without turning off the high 
voltage and lens currents, and by positioning the grids in the column in 
a reproducible fashion so that negligible magnification alterations are 
introduced by specimen changes. 

If the calibrating object is admixed with the sample under investiga- 
tion for simultaneous photography, the calibration is termed internal. 
The ideal calibration standard is the internal one which exhibits a 
periodicity of known magnitude that is independent of staining condi- 
tions, and is conveniently observable at all magnifications desired. In the 
absence of such an ideal standard, we employ beef-liver catalase crystals 
for magnification calibration in our laboratory (Ferrier and Murray, 
1966; Luftig, 1967, 1968; Cox and Home, 1968; Murray, 1968; Wrigley, 
1968). The crystals exhibit a convenient half -period of about 84 A when 
negatively stained with ammonium molybdate, in which they remain in- 
soluble. Rapid sample deposition permits the use of catalase crystals as 
an internal standard with phosphotungstate and silicotungstate stains as 


well, provided that dissociated catalase molecules are not confused with 
the specimen of interest. We noted no differences in spacing of the 
crystals in these stains compared to embedding in ammonium molybdate 
(external calibration), as done for original calibration (Murray, 1968). 
Similar results were obtained with uranyl acetate staining as described 
by Luftig (1967, 1968). Rapid dissociation of the crystals in uranyl 
oxalate prevented their use as internal standards when this stain is used. 
Dow polystyrene latex spheres and diffraction-grating replicas may 
be used as calibration standards at lower magnifications (Hall, 1966; 
Sjostrand, 1967). The latter is now available as suspended fragments, so 
that both standards may be used for internal calibration. Some investiga- 
tors use well characterized molecules as internal standards (e.g., the 
interparticle separation of TMV rods in tachtoids). Suitable dispersed 
ultrathin filaments of asbestos ( which show a period of 7.3 A ) have been 
suggested as a suitable internal standard for high magnification work by 
Femandez-Moran et al. (1966). The characteristic 395 A repeat of mag- 
nesium (or calcium) tropomyosin tachtoids observed by Caspar et al. 
(1969) appears to be a promising calibration standard. 

Particle Dimensions 

Measurement of isolated particle dimensions from micrographs should 
be approached with considerable caution ( Haschemeyer, 1970). Drying 
artifacts are probable and edge detail of molecular domains is obscured 
by stain to a variable extent. The most reliable subunit and particle 
dimensions are often obtained from center-to-center distances between 
particles or between subunits if their geometrical orientation relative to 
the beam is known. 

In order to relate measured dimensions to molecular weights, it is 
necessary to assume that the particles or subunits are packed in a 
manner that can be related to a known partial volume, and that the 
geometry is evident. For example, the volume is calculated on the basis 
that the particle is a close packed sphere of the measured diameter. 
Clearly, high precision molecular weights cannot be expected from such 
measurements. Nonetheless, the data is useful, for example, for com- 
paring electron microscopic calculations with known molecular weights 
to decide if objects observed in the micrographs represent the real par- 
ticle, or if they are more likely aggregates or dissociation products. 

Precise measurement of distances observed in specimens which ex- 
hibit periodic structures are often much more meaningful than those on 
isolated particles. The interpretation is often peculiar to the system under 
study and will not be discussed here. Whenever precise dimensions are 
required, accurate calibration of the magnification is essential, and the 


measurements should be made on the plate instead of on a print, prefer- 
ably with the aid of a good optical microcomparator. 

Molecular Weight Determinations 

A number of average molecular weights can in principle be determined 
by electron microscopy in two ways : ( 1 ) counting the number of particles 
of known concentration (e.g., in mg/ml, separately determined) con- 
tained in a known volume of solution; (2) determining the ratio of the 
number of unknown particles to the number of standard particles of 
known molecular weight within an arbitrary volume. For the latter 
method, the weight concentration ratio of unknown and standard par- 
ticles must be already known with precision, a problem which itself is 
far from trivial (Casassa and Eisenberg, 1964). Several useful papers 
on subjects related to particle counting were presented at a symposium 
on "Quantitative Electron Microscopy" ( Bahr and Zeitler, 1965 ) . 

A major difficulty is encountered in electron microscopic determina- 
tion of particle counts or particle count ratios due to differential quanti- 
tative adherence of different particles to the substrate film. We therefore 
cannot recommend procedures in which particle interaction with the grid 
can be a major factor. Also, there is no way in which a sufficiently small, 
precisely known volume can be deposited on the grid to permit total 
viewing within a grid square. 

The recommended method for determining particle weights in the 
electron microscope is, therefore, the one described by Backus and Wil- 
liams (1949) using the high-pressure spray gun to apply the sample. The 
entire field contained within the small droplets may be examined to 
determine the ratio of the number of particles of interest to the number 
of admixed standard particles. The requirements for successful applica- 
tion of this method are that the unknown and standard particles are 
clearly distinguishable and that aggregation does not interfere with 
counting each particle in the droplet. For greatest precision, similar 
quantities of similar-sized standard and unknown particles should be 
used. A sufficient number of droplets must be examined to insure a 
statistically significant ratio. This method is not easily applied to small 
particles such as enzymes, but may well be one of the best methods for 
determining the molecular weight of larger particles such as viruses, pro- 
vided that structural integrity is maintained during embedding. 


The obvious intent of negative staining is to obtain micrographs which 
show contrast details that may be used to deduce structural aspects of 


the specimen which characterized it in solution. The difficulty encoun- 
tered in such interpretation may vary from the trivial to the impossible, 
depending upon the specimen and which of its characteristics one is 
attempting to identify. Only general interpretative aids are presented 
here, and we recommend that whenever possible, the novice will find 
the greatest reward in examining literature articles which include electron 
microscopic studies on specimens resembling as closely as possible those 
of his interest. 

Identification of Specimen Contrast 

It is of prime importance to first ascertain which contrast components in 
micrographs of negatively stained particles are related to differences in 
scattering power within the domain of the embedded specimen (speci- 
men contrast). For the most part, contrast variations due to other 
causes, such as contour phenomenon, phase image or granularity of the 
negative stain (spurious contrast), may be readily identified in areas 
adjacent to the particles of interest. Reliable interpretation of specimen 
contrast is generally not possible unless the extent of contrast has dimen- 
sions greater than those observed for background granularity in nearby 
areas devoid of particles. Images produced by interference effects of 
certain periodic structures are sometimes exceptions to this rule, but will 
not be further considered in this chapter. 

Inexperienced microscopists are to be warned that the superposition 
of true specimen contrast and spurious contrast often suggests apparent 
detail, temptingly consistent with possible expectations. A prerequisite 
for identifying specimen contrast is that it persists in through-focal 
series. The degree of apparent contrast differences may, however, be 
affected. Nevertheless, this condition is not sufficient for identifying 
specimen contrast; it would not, for example, distinguish particle details 
so small that they become apparent only very near focus. The relative 
ease with which interpretations can be made on near-focus micrographs 
(e.g., with symmetrical phase grain separated by 5-7 A) and the re- 
liability of such a result, makes the optimal employment of a superbly 
tuned electron microscope highly desirable and, for some purposes, 

Resolution Requirements 

Most reasonably aligned modern electron microscopes are capable of 
readily demonstrating a point to point resolution on thin carbon supports 
in the neighborhood of 5 A. Thin carbon support films (cast on freshly 
cleaved mica) themselves show amorphous detail with a dimension of 


several angstroms, as is convincingly demonstrated with the scanning 
electron microscope developed by Crewe and Wall (1970), where phase 
contrast is absent. Some popular negative stains such as phosphotungstate 
and silicotungstate have dimensions estimated to be of the order of 8-9 A 
although uranyl stains are considerably smaller (4r-5A) (Hoglund, 1968). 
These factors combined with the usual thickness of a negatively stained 
specimen (150 A or more) limit possible resolution of specimen detail 
to values which are seldom if ever below 7 A. The question as to whether 
or not any meaningful contrast variation of 7 A may be expected from 
negatively stained particles will be discussed later in this section. Clearly, 
ideal conditions will be required for such resolution, since resolution 
in most specimens is limited to values nearer 20 A by inherent noise 
in the electron image (see the discussion by several authors in the 
recent symposium organized by Huxley and Klug (1971). 

In order to achieve a demonstratable 5-7 A resolution on a micro- 
graph, it is necessary to have the microscope compensated for astigma- 
tism and achieved focus, both to within about 500 A (typically at 80 kV). 
On several fine commercially available microscopes, this is to the nearest 
stop of the most sensitive focus control. Defocus to 0.2 (jl and 0.8 fx, re- 
sults in phase image size of about 10 and 20 A respectively, thereby 
limiting meaningful resolution to comparable dimensions (Haydon, 1968). 

It follows from the above discussion that the resolution of possible 
structural detail is first determined by the degree to which focus is 
achieved. At a point very near focus, the molecular dimensions of the 
stain employed become the limiting factor in resolution. We estimate this 
limit to be about 7 A for the most favorable of the more popular nega- 
tive stains, such as uranyl formate. Not surprisingly, such estimates vary 
somewhat among different investigators and most likely depend upon the 
model used for the estimate. It is well known, for example, that several 
heavy metal atoms in close proximity on two dimensional image projec- 
tion are required to achieve the necessary contrast to "see" them by 
direct transmission electron microscopy (Hall, 1966). 

Suppose we wish to contrast a "groove" between protein subunits 
similar to that schematically illustrated in Fig. 3.11. The groove contains 
an excess of 6 heavy metal atoms in cross section, which is probably suf- 
ficient to permit contrasting, even in relatively thick specimens embedded 
in stain; overlapping scattering of adjacent sections would also favor 
seeing them. For the sake of argument, we will imagine that these 
molecules are 5 A in diameter and that visible contrast could not have 
been achieved with less penetration. Furthermore, we will assume that 
the positive identification of contrast difference required focusing to 

One now has the interesting situation that penetration to 5 A detail 


Fig. 3.11 Cross section of a hypothetical 15 A "grove" in a particle contrasted 
with stain whose shape approximates to a 5 A sphere. See text for details. 

was required for the 7 A resolution of a groove 15 A across, two of which 
could not be resolved unless separated by about 15 A (aside from the 
latter point, the subunit size itself is not relevant ) . The definition of what 
resolution was attained at the specimen level in such a hypothetical case 
obviously entails some semantic difficulty. In general, this type of con- 
sideration is more relevant to negative staining than point to point resolu- 
tion of specimen detail, for it appears that intrapolypeptide detail, for 
example, has seldom if ever been resolved ( Haschemeyer, 1970). 

A real example where the above considerations are likely to play an 
important role is in resolving the 23 A spacing between adjacent turns in 
the helical superstructure of tobacco mosaic virus ( TMV ) . Although the 
superposition of top and bottom staining (see next section) contributes 
greatly to the difficulties in seeing the helix by negative staining, one is 
nonetheless struck by the fact that very few of the many micrographs of 
TMV (and other rod shaped viruses with similar structure) published 
show any helical detail. 

Finch and Holmes (1967) have reviewed the negative staining of 
TMV and conclude that with phosphotungstate staining the 23 A period 
is only occasionally seen as a serration at the edge, while the uranyl 
formate, the basic helix as well as higher order helices are frequently 
observable. Staining with uranyl acetate gives intermediate results. Finch 
and Holmes suggest that these differences result primarily from the 
depth to which the negative stain can penetrate into the surface grooves. 
Uranyl formate will penetrate to an inner radius of 60 A ( the maximum 
radius of TMV is 90 A), thereby providing more dramatic contrast dif- 


ferences between the groove and adjacent areas, characterized by pro- 
truding protein subunits. The depth of penetration of uranyl acetate and 
of PTA has not been established. 

In a previous review ( Haschemeyer, 1970) it was estimated that 
about 7 A resolution on a micrograph was required to observe the TMV 
helix in a convincing fashion. This point of view varies from that ob- 
served by others and, on re-examination of our through-focal photo- 
graphs of TMV stained with uranyl acetate, we agree that lesser focus 
is required (10-15 A) to observe the groove, but maintain that a similar 
extent of contrast detail could not be positively identified as specimen 
contrast if it were not for the fact that the structure of this virus is a 
periodic one that is already well established. 

Clearly, the limits at which such substructure can be identified also 
depend upon factors other than stain penetration; such factors include 
the thickness of the support film, depth of embedding, maximizing con- 
trast by using pointed filaments, proper exposure, and good darkroom 
techniques. On the basis of the TMV results, it seems that resolution 
of the 15 A groove in the model system discussed above would be only 
marginally probable. 

In summary, the identification of specimen detail characterized by 
contrast changes of limited extent requires a demonstrated resolution 
on the micrograph (usually obtained from grain size in adjacent areas 
void of particles) which has dimensions less than that of the contrast 
extent. Except in those rare instances where high point-to-point resolution 
can be seen in negatively stained particles, the specification of specimen 
resolution will usually be of vague significance. Many problems are 
readily solved by negative staining methods at poor resolution (e.g., is 
a plant virus rod-like or spherical, does a bacteriophage have a tail or 
not, or is a preparation of bacterial flagella free of bacteria themselves?). 
The highest resolution possible is obtained with negative stains such as 
uranyl formate, which are characterized by smaller size and hence in- 
creased penetrability. Less penetrating stains such as phosphotungstate 
are often used advantageously in systems where extensive penetration 
might obscure certain surface detail. An example of this is the penetra- 
tion into interior domains of a virus, thereby obscuring capsomer struc- 
ture. Extensive penetration is also undesirable in cases where higher 
resolution might render gross observations more difficult. 

Distribution of Stain Around Particles 

The manner in which negative stain surrounds a particle is an important 
consideration in the interpretation of negatively stained images. If the 
stain penetrates into surface irregularities on both the near side (the 


side closest to the support film) and the far side of the particle, the 
resultant image will be the superposition of detail on the top and bottom 
surfaces. A one-sided image, on the other hand, will contrast one surface 
only to give a result more nearly like that which would be expected 
when looking at the surface of a model of the particle. Such a result 
would generally be more easily interpreted, although two-sided images 
have proven particularly advantageous when studying certain virus par- 
ticles with a tilting stage. Regrettably, there seems to be no way to 
obtain one- or two-sided images at will for most specimens, and it is often 
difficult to ascertain the degree to which stain has enclosed both sides 
of a particle. 

Our knowledge of stain distribution around particles is limited 
largely to data obtained on several viruses. A particularly dramatic ex- 
ample of the appearance of one- and two-sided images was obtained by 
Nagington et al. (1964) during their study of orf virus. In this case, the 
structural features observed by negative staining with potassium phos- 
photungstate contrasted details on both the near and far side of the 
virus, but embedding with ammonium molybdate produced images 
which were one-sided. We are aware of no other case where a one-sided 
image can be assured by appropriate choice of staining conditions. 

It is intuitively reasonable to suppose that a two-sided image will 
always be obtained ( except perhaps for very large particles ) for particles 
embedded in stain which has solidified across holes in the substrate 
film. This was found to be the case with turnip yellow mosaic virus nega- 
tively stained with uranyl acetate (Finch and Klug, 1966). A small 
degree of flattening (10-20%) of the particles in a direction perpendicular 
to the plane of the hole was noted. 

Finch and Holmes (1967) review experiments which give informa- 
tion on the topology of staining for several other virus particles, each 
embedded by the same procedure. Some of these results are summarized 
below. Using optical transform techniques, it was established that TMV, 
when contrasted with uranyl formate, was usually completely embedded 
in negative stain but showed considerable variation in the development 
of contrast and the depth of stain penetration on the two sides. Rabbit 
papilloma virus stained with phosphotungstate was shown to be con- 
trasted always on the near side (by noting the shift of an identifiable 
capsomer with respect to the particle periphery upon tilting the grid) 
with a variable degree of contrasting on the far side ranging from equal 
contrast, indicative of complete enclosure, to so little contrast on the far 
side that the image was essentially one-sided (about 1% of the virions). 
Substantial flatteneing of the virus particles was observed, except when 
they were arranged in close-packed arrays. Uranyl acetate-contrasted 


turnip yellow mosaic virus appeared considerably flattened and was 
primarily contrasted on one side, a result differing from that obtained 
over holes mentioned previously. 

If there is any generalization which can be drawn from the limited 
examples on the determination of stain distribution available, it must be 
that the results are largely unpredictable and depend primarily upon 
some unknown surface properties of the embedded particle. It is ex- 
pected that contrast will usually be obtained from the near side of the 
particle, but the microscopist must anticipate contrast from the far side 
to reflect any possibility from an equal contribution to essential absence. 
Even adjacent particles may have their far side contrasted to a variable 

Beam Damage and Contamination 

When specimens are exposed to the electron beam they become coated 
with decomposition products from residual vapors present in the micro- 
scope column (Hall, 1966). The phenomenon is readily visualized by 
lateral growth at projections, as in the gradual closing of holes in a 
holey support film. The effect is obvious on flat surfaces after demag- 
nification, for a comparative darkening is observed in the area of the 
grid subjected to an intense electron beam. A minimum result of this 
deposition is a reduction of contrast and resolution similar to that ex- 
pected with thicker support films. The use of an anticontamination 
device placed near the specimen and cooled in liquid nitrogen substan- 
tially reduces contamination on the specimen. Some instruments con- 
tain an adjustable "cold finger" which can be easily moved nearer or 
farther away from the specimen. It can be shown with a holey specimen 
that holes can be either enlarged by placing the device too near the 
specimen, or decreased in size with larger separation. The separation 
between grid and cold finger can be adjusted so that no change in hole 
diameter is observed, indicating a balance between contamination rate 
and the rate at which sublimation of the support specimen occurs due to 
irradiation (Haydon, 1969). Therefore, even if an anticontamination 
device is effective in reducing net scattering of the specimen to zero, 
beam damage to the specimen may still result in substantial alteration 
of fine details. In negative staining, one usually need not be concerned 
with damage to the particle itself, for the image is produced by the con- 
trast provided by the glassified stain originally outlining the particle 
during specimen mounting. Beam damage incurred by interaction with 
the stain to a degree which may result in obliteration of detail has not 
been extensively studied. A recent report by Williams and Fisher (1970), 


however, suggests that beam damage is a much more important con- 
sideration than we had previously supposed. They showed the basic 
helical structure of TMV in remarkable detail on micrographs taken with 
minimal exposure of selected specimen areas to the beam current. 
Marked deterioration of image quality was observed after as little as 30 
sec exposure to the beam. In view of the difficulty normally encountered 
in resolving helical detail, as discussed in the previous section, it is par- 
ticularly notable that the contrasting medium employed was phos- 

If beam damage is in general even partially as critical as is suggested 
by Williams and Fisher (1970), a different routine procedural sequence 
for photography from that previously recommended ( Haschemeyer, 
1970) will be required. It was previously recommended that the easiest 
way to obtain the highest quality results would be from a through-focal 
series. Even if preliminary focus is achieved on an adjacent area with 
small-spot double-condenser illumination, our instruments have no pro- 
vision for taking the several through-focal photographs at the speed 
which would be required. Instead, it would become necessary to focus 
with small-spot double-condenser illumination on one area of the grid, 
and then either shift or spread the beam, or move the specimen to an 
adjacent area for immediate photography. Williams and Fisher (1970) 
have suggested the manner in which this is accomplished on their 
microscope; alternative methods will be required for other instruments. 

In all cases, a thermally stable support film (thin carbon) will be 
required, and focus alterations resulting from beam or sample movement 
must be negligible. In addition, the operator will be required to judge 
critical focus, so that a reasonable percentage of photographs will be 
within the required range. The ease with which this can be accomplished 
depends upon critical observation of phase contrast at relatively high 
magnification and varies among instruments, as discussed earlier. The 
problem should be dramatically reduced if phase contrast may be ob- 
served with a high quality image intensification system. 

Beam damage will be a problem of negligible proportions for most 
studies for which the highest resolution is not required. Through-focal 
series will retain their value in such investigations, for it is usually more 
important to avoid identifying spurious contrast with specimen detail 
than it is to fail seeing detail at the extreme limit of resolution. 

Anticipated Appearance of Idealized Images 

When the microscopist has taken a sufficient number of quality micro- 
graphs of negatively stained particles, he often wishes to ascertain what 


structural features of the particle account for the detail observed in a 
majority of the images. In this section we largely exclude discussing the 
interpretation of gross structural features and concentrate on the quater- 
nary structure of particles which possess apparent symmetry. Further 
details may be found in reviews on electron microscopy of viruses ( Finch 
and Holmes, 1967; Home, 1967) and of oligomeric proteins (Hasche- 
meyer, 1970). 

The orientation attained by particles with respect to the support film 
after negative staining can vary from highly selective to nearly random. 
This final orientation probably depends to a large degree upon attractive 
interactions between the particle and the support film. In techniques 
such as the float method, which are followed by the removal of particles 
unattached to the grid by washing, the retention of the particles is due 
to favorable collisions which cause the particles to be immobilized on 
the support film. If the attractive forces are sufficiently great for each of 
several different sites on the particle, a distribution of orientations is 
achieved. Non-random distributions result if only a few sites are capable 
of interaction, or if the process is reversible (i.e., the activation energy 
barrier for release can be overcome), or one orientation has a larger 
attractive interaction either in magnitude per site or in the total number 
of sites which can interact with the support. 

In our experience, it seems that many of these interactions are not 
readily reversible, since we obtain a similar distribution of particles after 
floating the grid on a particle-free medium for varying lengths of time. 
The deposition of particles from bulk medium during rapid drying (e.g., 
with the spray gun method) may sometimes result in more random 
orientation, but in at least one case of preferred orientation with ovine 
glutamine synthetase ( Haschemeyer, 1968), the result was independent 
of the method of sample application. 

A somewhat naive analogy suggests that the particles are buffeted 
around with an ever increasing frequency of collisions with the support 
as the tidal wave (to the particles) of aqueous stain solution recedes. 
When favorable collisions occur, attachment takes place with results 
similar to those obtained by the float procedure, except that kinetic 
reversal with durations greater than a few tenths of a second cannot 
occur. Such reversal might result in a larger number of particles oriented 
with more favorable energy. Although the particles are subject to tre- 
mendous shearing and surface tension forces during the last moments 
of their aqueous existence, there is little evidence to suggest that these 
forces dislodge many particles from the grid to change their orientation. 

If a substantial percentage of molecules assume their final orientation 
by positive attraction with the surface support, as suggested above, 


mechanical analogies for the expected distribution of orientations will 
not necessarily be valid. By mechanical analogy, a protein dimer whose 
shape approximates two spheres in contact would always lie with its axis 
parallel to the support. This will, in fact, be one likely possibility in nega- 
tive staining, since that orientation permits both spheres to contact the 
support, thereby achieving twice the stabilization afforded by a single 
contact of that kind. 

It is not unreasonable to suppose, however, that other sites on the 
dimer exist with sufficient attraction for the support to "freeze-in" alter- 
nate attachment orientations, or even that other sites exist which have 
greater than twice the attractive interaction for the film than any of those 
which can simultaneously contact the support with both subunits. The 
orientation of the molecular axis for the negatively stained dimer with 
respect to the electron beam might therefore assume any value from 
0° to 90°. The appearance of the two-sided images would include those 
suggestive of a 1-, 2-, and 3-subunit structure. 

The unpredictability of particle orientation is illustrated with several 
oligomeric enzymes (see Haschemeyer, 1970, for further details and 
original references). For example, the 12-subunit stacked-hcxagon struc- 
ture of glutamine synthetase from E. coli will lie flat (along the 6-fold 
axis) or stand on edge with either 2 or 4 subunits in contact with the 
support, while the 10-subunit stacked-pentamer structure of arginine 
decarboxylase showed no clear on-end presentations. The tetramer of 
tryptophanase has 4 subunits with a square planar configuration, which 
probably represents D 2 symmetry. The dimeric presentations frequently 
observed with this protein are unlikely to be dissociation products, and 
therefore represent edge-on views of the tetramer. We note with in- 
terest the possibility of obtaining nearly random orientations of small 
protein molecules embedded in stain deposited across holes in the sup- 
port film. 

As the asymmetry of a particle increases, preferred orientations which 
maximize the number of possible contacts with the support film become 
increasingly probable. In the extreme example illustrated by TMV 
(3000 A by 190 A rods), virtually all particles are embedded with the 
long axis of the rod parallel to the support. Extensively degraded virus 
particles with more modest axial ratios, however, are frequently observed 
with end-on presentations. In the case of the icosahedral viruses con- 
taining many quasi-equivalent subunits, the orientation of individual 
particles appears to be more random and the probability of viewing the 
particle along any of its symmetry axes is more nearly equal. Here, the 
differences in the number of contacts in any orientations are minimized, 
due to a flattening of the particle on the support. This provides con- 


siderably more capsomer-support-film contacts than would be expected 
based on a rigid mechanical model. 

Preliminary attempts to interpret negatively stained micrographs of 
oligomeric molecules should emphasize the analysis of those presenta- 
tions of molecules which appear to possess a rotational axis of symmetry. 
For most particles (see Finch and Klug, 1967a, for the interpretation of 
the important class of viruses with skew structures) characterized by 
point group symmetry, the two-sided image viewed along an n-fold axis 
will itself appear to have n-fold symmetry or m-fold symmetry, where m 
represents an integer multiple of n. The frequent appearance of m- 
rather than n-fold symmetry results from the fact that even though the 
molecule contains chemically asymmetric protomers, these subunits none- 
theless appear symmetrical (e.g., spherical or with an apparent infinite 
symmetry axis ) at the limited resolution obtained. 

The tentative identification of presentations along the various sym- 
metry axes of the particle, coupled with an estimate of the number of 
subunits within the particle obtained from other physical and/or chem- 
ical procedures, usually permits the investigator to limit the choice of 
possible models because of symmetry considerations. The reasoning in- 
volved in narrowing the choice of models to a manageable few is beyond 
the scope of this chapter. Details are found in several reviews (Home 
and Wildy, 1961; Caspar and Klug, 1962; Finch and Holmes, 1967; 
Haschemeyer, 1970). 

Once a tentative model has been found, it is necessary to decide 
whether or not it can account for a majority of well-preserved images 
observed in the negatively stained micrographs. If the mental visualiza- 
tion of possible two-sided images that would be obtained from various 
orientations of the model is too difficult (as will be the case with most 
individuals for relatively complex structures ) , two major approaches are 
possible. An obvious and relatively simple one is based on the fact that 
only a few point groups exist for particles containing asymmetric units. 
There is an excellent possibility that the particle under investigation has 
marked geometrical similarity and appearance in negative stain to one 
already characterized in detail from another laboratory. Such direct com- 
parison will frequently be sufficient for the investigator to propose a 
reasonable model. 

If no satisfactory comparison with published results is possible, the 
microscopist will find it advantageous to obtain negatively stained analog 
images of the proposed model. Realistic analog images were obtained by 
Caspar (1966) by radiographing a proposed model of a virus particle 
embedded in plaster of Paris, topologically cast according to the antici- 
pated distribution of negative stain around the real virus particle. The 


major difficulty with Caspar's method is the inability to recover the 
model from the cast. Since multiple projections of several models are 
often desirable, an extensive construction effort is required. 

A more rapid procedure has been used in our laboratory for routine 
procurement of analag images of protein oligomers. A small model is 
constructed from material transparent to X-rays. These materials include 
balsa wood for arbitrary shapes and small cork buttons to represent 
spherical subunits. The model is lightly tacked in the desired orientation 
to a plastic lid which is placed over a container filled with a radio- 
opaque liquid (Hypaque). Contrast can be estimated by fluoroscopy, 
and the desired effect is obtained by adjusting the intensity of X-irradia- 
tion and/or by altering the opacity of the fluid. The X-ray photograph 
is then taken. The model is easily broken loose from the lid, remounted 
in another orientation, and rephotographed. The analog images of as- 
partate-/?- decarboxylase shown by Haschemeyer (1970) were obtained 
by this method. We estimate that after a little experience, the entire 
process could be completed within an hour. 

The microscopist with computer availability should consider ob- 
taining computed superposition patterns as an alternative to the X-ray 
analog method. Finch and Klug ( 1967b ) devised a program in which the 
model coordinates of a proposed virus structure as supplied to the com- 
puter, and the calculated superposition pattern corresponding to any 
desired orientation is displayed on CRT output, and photographed. 
Evaluation of many models in various orientations is thus possible in a 
short period of time. The elegant results obtained are quite superior to 
the shadowgraphs of geodetic stick models previously used to represent 
2-sided images of viruses. Both procedures are reviewed, with examples 
by Finch and Holmes (1967). 

A program to calculate analog images under more realistic conditions 
has been applied to oligomeric proteins ( Bowers et al., 1970 ) . Computa- 
tional details are outlined by Bowers (1971). Photographs from the 
CRT output approximate continuous tonality from white to black, and 
provisions for nonuniform staining may be incorporated into the pro- 
gram. Extended contacts not penetrated by stain may also be simulated. 
Computed results on a dodecameric model with tetrahedral symmetry 
are illustrated in Fig. 3.12. Rather sophisticated reprogramming will be 
required to extend this computational procedure in a manner that can 
be conveniently applied to virus structures with many subunits. 

Finally, we wish to mention two experimental procedures which are 
frequently used to add confidence to the choice of one model over an- 
other. The first entails a search for conditions of negative staining that 
promote the formation of linear or 2-dimensional arrays, or of small 


Fig. 3.12 Negative stain simulations of a compact dodecomeric model pos- 
sessing tetrahedral symmetry. A: Successive views of the model from left to 
right along a 3-fold axis, a pseudo-3-fold axis, a 2-fold axis, a pseudo-2-fold 
axis and a pseudo-5-fold axis, respectively. B: Negative stain analog simula- 
tion of the various principal regular orientations assuming spherical subunits and 
a stain sheet thickness 4 times the subunit diameter. Note that the images along 
the 2-fold and 3-fold axes are indistinguishable from those of their respective 
pseudo-axes. C: Simulation as in B, but with isoiogous dimers linked by a 
sphere of diameter 0.9 that of the subunits. D: Simulation as in C, but with stain 
penetration of the hollow core prevented. 

crystals. This can be accomplished by employing abnormally high par- 
ticle concentrations and varying the pH. The advantage of such arrays 
is that they will often exhibit a preferred interaction of locations dictated 
by symmetry considerations, thus providing a large number of images 
from identically oriented particles which can be interpreted with in- 
creased confidence. 

Secondly, the orientation of particles with respect to the electron 
beam may be altered with special tilting stages that can be rotated 
through large angles (30°^15 o ). Thus it is possible to view the same 


particle along more than one symmetry axis and to align other particles, 
more arbitrarily oriented, along a symmetry axis. The value of such 
studies with viruses can hardly be overstressed, and individuals ex- 
amining virus particles can profitably consult the review of Finch and 
Holmes (1967) for additional details. A tilting stage was recently em- 
ployed in the study of a small enzyme as well (Josephs, 1971), and it 
should prove generally profitable to obtain tilted images on small pro- 

Optical Aids for Enhancing Periodic Detail 

The optical-transform method is an elegant procedure for extracting 
periodic detail from electron micrographs. An optical diffraction pattern 
of the photographic image is obtained with the aid of an appropriate 
optical diffractometer. The diffraction pattern contains information com- 
ponents which arise from periodic detail on both sides of the particle and 
from nonperiodic background detail and noise. The diffraction "spots" 
arise from periodic detail only, and this part of the diffraction image can 
be recombined to give a real image with a lens system by masking off 
light originating from nonperiodic components in the plane of the dif- 
fraction pattern. 

With some knowledge of the anticipated form of such diffraction pat- 
terns, it is often possible to additionally mask out image contributions 
arising from one side of the particle (a comparatively easy task for 
helical periodicity) and reconstruct a real image which primarily shows 
the details of one-sided periodic structure only. The one-sided images of 
TMV obtained in this manner by Klug and DeRosier ( 1966 ) display the 
quaternary details of the helical symmetry with astonishing clarity. This 
type of image reconstruction is receiving ever increasing attention and 
will doubtless assume a position of prominence in the future study of 
appropriate structures. 

Other methods of enhancing apparent periodic detail in electron 
micrographs may collectively be termed artificial superposition methods. 
Markham et al., (1963) first suggested that periodic detail could be 
enhanced with respect to background granularity and structural distor- 
tions by the repetitive superposition of images rotated by 360° /n for an 
n-fold symmetry axis, or translated by the distance d, for linear repeat. 
The numerical values of n and d are determined when maximal periodic 
enhancement is obtained from a series of superposition patterns in which 
the assumed values of n and d are appropriately varied. 

An alternative procedure is to form a superposition pattern from 
several different particle images, presumably reflecting identical molec- 





Fig. 3.13 A: 11 S protein aggregates of bacteriophage f 2 . B: Superposition pat- 
tern of the aggregates shown in A. The particles, which measure about 90 A by 
45 A, are thought to be composed of 6 trimer clusters (Zelazo and Haschemeyer, 
1969). The superposition photograph does not, in this case, substantially strengthen 
the arguments for the proposed structure. (Zelazo and Haschemeyer, unpublished 


ular orientation (Fig. 3.13). Again, background granularity and minor 
particle distortions will be averaged out, while quaternary detail is 
enhanced. In our view, these artificial superposition methods should be 
used with extreme caution, for false apparent symmetry will be identified 
in many cases for reasons already discussed. For further discussion, the 
reader is referred to Finch and Holmes (1967). Superposition photo- 
graphs for particles where symmetry is already established in some 
instances provide pleasing results for the purpose of illustration and may 
be advantageous for determining molecular and particularly inter-subunit 


Electron microscopy of negatively stained specimens has proven effective 
for the study of a wide range of problems since 1959, and will un- 
doubtedly continue to provide major contributions for some time to 
come. Developments in staining procedures and instrumentation over 
the past decade rapidly led to significant improvements in obtaining and 
interpreting negatively stained micrographs. The most significant ad- 
vantage to future microscopists, as we see it, is the ability to compare 
results with previously interpreted systems and to take advantage of 
more complicated, seemingly magical, handling procedures, uniquely 
effective at preserving structure or presenting details on particular 

Further applications of 3-dimensional reconstruction methods are 
likely to produce spectacular results on applicable systems. The recent 
3-dimensional reconstruction of T 4 phage tail ( DeRosier and Klug, 1968 ) 
and of F-actin and F-actin "decorated" with myosin fragments (Moore 
et al., 1970) provide ample justification for this high hope. 

Another major advance for the examination of negatively stained par- 
ticles can be expected when specimens are examined with the high 
resolution scanning microscope developed in Crewe's laboratory ( Crewe 
and Wall, 1970; Crewe, 1971). Although the major application of the 
instrument may be directed elsewhere, including the examination of 
unstained specimens, the contrast advantages obtained from the signal of 
scattered electrons or from various signal ratios as compared to the trans- 
mitted beam may add a new dimension to negative staining. Less pene- 
tration of stain into smaller crevices can be successfully contrasted, and 
the use of lower density negative stains should provide new choices for 
optimizing structural preservation. 



Some results and conclusions presented in this chapter are based on research 
conducted in the authors' laboratory that was supported by grants from the 
Public Health Service, the National Science Foundation, and the John Hart- 
ford Foundation. 


Backus, R. C, and Williams, R. C. (1949). The use of spraying methods and 
of volatile suspending media in the preparation of specimens for electron 
microscopy. /. App. Phys., 21, 11. 

Bahr, G. F., and Zeitler, E. H. (1965). Proceedings of a symposium on 
"Quantitative Electron Microscopy." Lab. Invest., 14, 729. 

Bancroft, J. B. (1970). The self-assembly of spherical plant viruses. Adv. 
Virus Res., 16, 99. 

Bowers, W. F. (1971). The study of enzyme quaternary structure: L-aspartate- 
/3-decarboxylase. Thesis, Cornell University, Ithaca, N.Y. 

Bowers, W. F., Czubaroff, V. B., and Haschemeyer, R. H. (1970). Subunit 
structure of L-aspartate-/3-decarboxylase from Alcaligenes faecalis. Biochem- 
istry, 9, 2620. 

Bradley, D. E. (1965). The morphology and physiology of bacteriophages as 
revealed by the electron microscope. /. Roy. Micros. Soc, 84, 257. 

Brenner, S., and Home, R. W. (1959). A negative staining method for high 
resolution electron microscopy of viruses. Biochem. Biophys. Acta 34, 103. 

Casassa, E. F., and Eisenberg, H. (1964). Thermodynamic Analysis Of Multi- 
component Solutions. In: Advances In Protein Chemistry. Vol. 19, (Anfinsen, 
C. B. Jr., Anson, M. L., Edsall, J. T., and Eichards, F. M., eds.), p. 287. 
Academic Press, New York. 

Caspar, D. L. D. (1966). An analog for negative staining. /. Mol. Biol, 15, 

Caspar, D. L. D., Cohen, C, and Longley, W. (1969). Tropomyosin: crystal 
structure, polymorphism and molecular interactions. J. Mol Biol, 41, 87. 

Caspar, D. L. D., and Klug, A. (1962). Physical principles in the construction 
of regular viruses. Cold Spring Harbor Symp. Quant. Biol, 27, 1. 

Cosslett, V. E. (1971). Electron Microscopy. In: Physical Techniques In 
Biological Research, 2nd edition, Vol. 1A, 71 (Oster, C, ed.). Academic 
Press, New York. 

Cox, R. W., and Home, R. W. (1968). Accurate calibration of the magnifica- 
tion of the A.E.I.E.M. 6B/2 electron microscope using catalase crystals. 
Proc. 4th Europ. Reg. Conf. Electron Micros. Rome, 1, 579. 

Crewe, A. V. (1971). A high-resolution scanning electron microscope., Sci. 
Amer., 224, 26. 


Crewe, A. V., and Wall, J. (1970). A scanning microscope with 5 A resolution. 
J. Mol. Biol, 48, 375. 

Davies, G. E., and Stark, G. R. (1970). Use of dimethyl suberimidate, a 
crosslinking reagent, in studying the subunit structure of oligomeric pro- 
teins. Proc. N.Y. Acad. Set., 66, 651. 

De Rosier, D. J., and Klug, A. (1968). Reconstruction of three dimensional 
structures from electron micrographs. Nature, 217, 130. 

Farrant, J. L. (1954). An electron microscope study of ferritin. Biochem. 
Biophys. Acta 13, 569. 

Fernandez-Moran, H. (1962). Cell-membrane ultrastructure. Low-temperature 
electron microscopy and X-ray diffraction studies of lipoprotein components 
in lamellar systems. Circulation, 26, 1039. 

Fernandez-Moran, H., Van Bruggen, E. J., and Ohtsaki, M. (1966). Macro- 
molecular organization of hemocyanins and apohemocyanins as revealed by 
electron microscopy. J. Mol. Biol., 16, 191. 

Ferrier, R. P., and Murray, R. T. (1966). Low-angle electron diffraction. /. 
Roy. Micros. Soc, 85, 323. 

Finch, J. T., and Holmes, K. C. (1967). Structural Studies Of Viruses. In: 
Methods In Virology, Vol. 3. (Maramorosch, K., and Koprowski, H., eds.), 
p. 351. Academic Press, New York. 

Finch, J. T., and Klug, A. (1967a). Arrangement of protein subunits and the 
distribution of nucleic acid in turnip yellow mosaic virus. II. Electron 
microscope studies. /. Mol. Biol., 15, 344. 

Finch, J. T., and Klug, A. (1967b). Structure of broad bean mottle virus. I. 
Analysis of electron micrographs and comparison with turnip yellow mosaic 
virus and its top component. /. Mol. Biol, 24, 289. 

Hall, C. E. (1955). Electron densitometry of stained virus particles. /. Biophys. 
Biochem. Cytol, 1, 1. 

Hall, C. E. (1966). Introduction To Electron Microscopy, 2nd ed. McGraw- 
Hill Book Company, New York. 

Haschemeyer, R. H. (1968). Electron microscopy of enzymes. Tran. N.Y. 
Acad. Sci., 30, 875. 

Haschemeyer, R. H. (1970). Electron Microscopy Of Enzymes. In: Advances 
In Enzymology, Vol. 33 (Nord, P. F., ed.), p. 71. John Wiley and Sons, Inc., 
New York. 

Hayat, M. A. (1970). Principles and Techniques of Electron Microscopy: 
Biological Applications, Vol. I. Van Nostrand Reinhold Company, New York. 

Hayat, M. A. (1972). Basic Electron Microscopy Techniques. Van Nostrand 
Reinhold Company, New York. 

Haydon, G. B. (1968). On the interpretation of high resolution electron micro- 
graphs of macromelecules. /. Ultrastruct. Res., 25, 349. 

Haydon, G. B. (1969). Visualization of substructure in ferritin molecules: an 
artifact. /. Microscopy, 89, 251. 

Heidenreich, R. D. (1967). Electron phase contrast images of molecular de- 
tails. Siemens Rev., 34, 4. 


Hoglund, S. (1968). Electron microscopic studies on some virus and immuno- 
globulin components. Arkiv Kemi, 28, 505. 

Hohn, T., and Hohn, B. (1970). Structure and assembly of simple RNA 
bacteriophages. Adv. Virus Res., 16, 43. 

Home, R. W. (1965a). The Examination Of Small Particles. In: Techniques 
For Electron Microscopy, 2nd ed. (Kay, D. H., ed.), p. 311. Davis, Phila- 

Home, R. W. (1965b). Negative Staining Methods. In: Techniques For Elec- 
tron Microscopy, 2nd ed. (Kay, D. H., ed.) p. 328. Davis, Philadelphia. 

Home, R. W. (1967). Electron Microscopy Of Isolated Virus Particles And 
Their Components. In: Methods In Viorology. (Maramorosch, K., and 
Koprowski, H., eds.), p. 521. Academic Press, New York. 

Home, R. W., and Wildy, T. (1961). Symmetry in virus architecture. Virology, 
15, 348. 

Huxley, H. E. (1956). Some observations on the structure of tobacco mosaic 
virus. Proc. Europ. Reg. Conf. Electron Micros., Stockholm, p. 280. Aca- 
demic Press, New York, 1957. 

Huxley, H. E. (1969). The mechanism of muscular contraction. Science 164, 

Huxley, H. E., and Klug, A. (1971). A discussion on new developments in 
electron microscopy with special emphasis on their application in biology. 
Phil. Trans. Roy. Soc. Lond., B. 261, 1. 

Huxley, H. E., and Zubay, G. (1960). Electron microscope observations on the 
structure of microsomal particles from Escherichia coli. J. Mol. Biol., 2, 10. 

Johnson, M. W., and Home, R. W. (1970). Some observations on the relative 
dehydration rates of negative stains and biological objects. /. Microscopy, 91, 

Josephs, R. (1971). Electron microscope studies on glutamic dehydrogenase: 
subunit structure of individual molecules and linear associates. /. Mol. Biol., 
55, 147. 

Klug, A., and De Rosier, D. J. (1966). Optical filtering of electron micro- 
graphs: reconstruction of one-sided images. Nature, 212, 29. 

Kozloff, L. M. (1968). A Biochemistry Of The T-even Bacteriophages Of 
Escherichia coli. In: Molecular Basis Of Virology. (Fraenkel-Conrat, H., ed.), 
p. 435. Van Nostrand Reinhold Company, New York. 

Leberman, R. (1965). Use of uranyl formate as a negative stain. J. Mol. Biol., 
13, 606. 

Lubin, M. (1969). Observation on the structure of RNA polymerase and its 
attachment to DNA. J. Mol. Biol, 39, 219. 

Luftig, R. B. (1967). An accurate measurement of the catalase crystal period 
and its use as an internal marker for electron microscopy. /. Ultrastruct. 
Res., 20, 91. 

Luftig, R. B. (1968). Further studies on the dimensions of viral and protein 
structures using the catalase crystal internal marker technique. J. Ultrastruct. 
Res., 23, 178. 


Markham, R., Frey, S., and Holls, G. J. (1963). Methods for enhancement 
of image detail and accentuation of structure in the electron microscope. 
Virology, 20, 88. 

Mellema, J. E., and Van Bruggen, E. F. J., and Gruber, M. (1967). An 
assessment of negative staining in the electron microscopy of low molecular 
weight proteins. Biochim. Biophys. Acta, 140, 182. 

Moore, P. B., Huxley, H. E., and De Rosier, D. J. (1970). Three-dimensional 
reconstruction of F-actin, thin filaments, and decorated thin filaments. /. 
Mol. Biol, 50, 279. 

Murray, R. T. (1968). Accurate calibration of image magnification by low 
angle electron diffraction. Proc. 4th Europ. Reg. Conf. Electron Micros, 
Rome, 1, 581. 

Muscatello, U., and Home, R. W. (1968). Effect of the tonicity of some 
negative-staining solutions on the elementary structure of membrane-bound 
systems. J. Ultrastruct. Res., 25, 73. 

Nagington, J., Newton, A. A., and Home, R. W. (1964). The structure of Orf 
virus. Virology, 23, 461. 

Parsons, D. F. (1963). Mitochondrial structure: two types of subunits on 
negatively stained mitochondrial membranes. Science, 140, 985. 

Sjostrand, F. S. (1967). Electron Microscopy Of Cells And Tissues, Vol. 1. 
Academic Press, New York. 

Thon, F. (1967). On the defocusing dependence of phase contrast in the elec- 
tron optical images. Siemens. Rev., 34, 13. 

Valentine, R. C, and Home, R. W. (1962). An Assessment Of Negative Stain- 
ing Techniques For Revealing Ultrastructure. In: The Interpretation Of 
Ultrastructure. (Harris, R. J. C., ed.), p. 263. Academic Press, New York. 

Valentine, R. C, Shapiro, B. M., and Stadtman, E. R. (1968). Regulation 
of glutamine synthetase. XII. Electron microscopy of the enzyme from 
Escherichia coli. Biochemistry, 7, 2143. 

Van Bruggen, E. F. J., Wiebenger, E. H., and Gruber, M. (1960). Negative- 
staining electron microscopy of proteins at pH values below their isoelectric 
points. Its application to hemocyanin. Biochim. Biophys. Acta, 42, 171. 

Welch, W. H., and Nuttall, H. F. (1892). A gass-producing bacillus capable 
of rapid development in the blood-vessels after death. Bull. Johns Hopkins 
Hosp., 3, 81. 

Williams, C. R., Oliver, R. H., Henney, H .R., Mukherjee, B. B., and Reed, 
L. J. (1967). Alpha-keto acid dehydrogenase complexes. VI. Dissociation 
and reconstitution of the dihydrolipoyl transacetylase of Escherichia coli. 
J. Biol. Chem., 242, 889. 

Williams, R. C, and Fisher, H. W. (1970). Electron microscopy of tobacco 
mosaic virus under conditions of minimal beam exposure. /. Mol. Biol., 52, 

Wischnitzer, S. (1972). The Electron Microscope. In: Principles And Tech- 
niques Of Electron Microscopy: Biological Applications, Vol. 3 (Hayat, 
M.A., ed.), Van Nostrand Reinhold Company, New York. 


Wold, F. (1967). Bifunctional Reagents. In: Methods In Enzymology, Vol. XI. 
(Hirs, C. H. W., ed.), p. 617. Academic Press, New York. 

Wrigley, N. G. (1968). The lattice spacing of crystalline catalase as an in- 
ternal standard of length in electron microscopy. /. Ultrastruct. Res., 24, 454. 

Zelazo, P. O., and Haschemeyer, R. H. (1969). Specific dissociation of bac- 
teriophage f 2 protein to an lis component. Biochemistry, 8, 3587. 

. (1970). Spherical protein shell formation from an lis subunit of bac- 
teriophage of f 2 . Science, 168, 1461. 

4 Shadow Casting 
and Replication 

W. J. Henderson and K. Griffiths 

Tenovus Institute for Cancer Research, The Welsh 

National School of Medicine, The Heath, Cardiff, United Kingdom 



I he techniques of surface replication and shadow casting were two of 
the earliest procedures used for the examination of specimens by electron 
microscopy. The replication technique provides a means of obtaining a 
thin film of electron-transparent material in such a manner that it 
corresponds to the surface structure of the specimen under investigation. 
The surface topography of an electron-opaque specimen can therefore be 
studied by examination of its replica in the electron microscope. Shadow 
casting has now become a recognized integral part of the replication 
technique essentially being the oblique deposition of an electron-dense 
material onto the replica in order to improve contrast and to permit 
assessment of the three-dimensional nature of the specimen. The contrast 
of the poorly visible fine structure can therefore be increased. The tech- 
nique of shadow casting has also been used as a procedure for the 
estimation of the size of various particles, including tissue organelles 
and viruses. 

Until recently, the replication procedure has found few applications 
in the field of the biological sciences since the conventional ultrathin 
sections of biological specimens have generally been adequate for most 
investigations of intracellular structures. Currently however, there is 
increasing interest in the nature of cell and tissue surfaces and the scan- 
ning electron microscope is now being widely used for such studies, 
although the resolution of this instrument does not yet approach that of 
the modern transmission microscope. The application of the replication 
technique to the study of biological materials is thus of considerable 


The shadow casting technique was originally developed by Williams and 
Wyckoff (1946) to increase the contrast of the electron microscope 
image. Subsequent application of the procedure indicated that shadowing 
provided a relatively accurate assessment of the size of the various par- 
ticles under investigation. Furthermore, the three-dimensional image 
obtained was found to be of considerable value in the interpretation of 
the surface characteristics of such particles. 

The technique basically involves the oblique deposition in vacuo of a 
thin film of electron-dense material onto the surface of the specimens 



Support film < 


Fig. 4.1 The shadowing technique. 

Direction of 

Shqdowing material 
- Specimen 


Fig. 4.2 A sodium chloride crystal shadow cast to display the surface step 
(arrow) approximately 600 A in depth. X1 15,000. 


under investigation (Fig. 4.1). This angled deposition leaves an area on 
the support membrane free of electron-dense material and, as an electron 
image, will resemble a shadow of the particle being studied. When micro- 
graphs of shadowed specimens are printed as positive, the shadow will 
appear to be white. Figure 4.2 illustrates this, and shows, using a crystal of 
sodium chloride, the accuracy with which surface characteristics can be 
evaluated. Similarly, micrographs can also be printed as negatives, the 
shadow appearing dark and the specimen apparently illuminated with 
white light. An example of this is shown in Fig. 4.3, which demonstrates 
the shadowing of a polyvinyl chloride paste polymer with a gold/palla- 
dium alloy. Certain high density metals, such as gold, chromium, gold/ 
palladium alloys or platinum, were found to be suitable shadowing 
materials producing good contrast with a very thin coating, yet display- 
ing little of their own structure. The degree of resolution obtained with 

Fig. 4.3 Polyvinyl chloride particle. The build-up of the shadowing material 
(45°) on the particles is shown. Negative print. X1 5,000. 


the electron microscope must, of course, depend upon the thickness of 
this coating and upon the metal vised. Metals with the highest melting 
points, which require high voltages for evaporation, provide the best 
resolution. Under the electron beam of the microscope, metals with low 
melting points, although more easily evaporated, tend to form aggregates 
of crystallites within the structure of the film and are consequently un- 

A number of procedures have been devised by which metals can be 
evaporated onto the specimen. A small loop of a metal with a low melt- 
ing point can be attached to a coiled tungsten filament (0.5 mm diam- 
eter) connected to the high current terminals of the vacuum coating 
unit. Alternatively, a 1.0 mm diameter, straight tungsten wire can be 
used for the evaporation of the materials with higher melting points. A 
"boat" prepared from thin molybdenum foil and laid across the terminals 
of the unit, and upon which the metal is placed, also offers a viseful 
means by which the material can be evaporated onto the specimen. 
Higher resolution has been recently obtained from the use of an electron 
beam source for the evaporation of high melting point metals. A wider 
range of materials can be evaporated using such a source, and this pro- 
cedure will be described later in more detail. 

A variety of metals may therefore be effectively used with the shadow 
casting technique. The efficiency of this technique and the ultimate reso- 
lution attained must also depend, however, on the degree of vacuum 
which can be maintained during the evaporation procedure (Holland, 
1956). A number of different commercial units, which are available in 
various regions of the world, can attain the high vacuum required for 
the procedure. A liquid nitrogen cooling system established between the 
vacuum chamber and the diffusion pump can be of particular assistance 
in maintaining these high vacuum conditions, although such conditions 
can only really be achieved with an efficient rotary pump. The higher the 
vacuum, the finer is the metal deposit on the surface of the specimen. A 
considerable loss of vacuum can be caused by the degassing of the 
tungsten filament when heated, although such an effect can be almost 
eliminated by pre-heating to a temperature at which the metal is seen 
to melt and flow along the filament. The filament is then cooled and 
high vacuum re-established before a final re-heating for the evaporation 
of the metal onto the specimen. 

Effective shadow casting requires the specimen to be mounted at a 
reasonable distance from the evaporation source in order to avoid heat 
radiation damage. A distance of 12 cm is generally considered adequate, 
but this varies according to the specimen under examination. The angle 
of shadowing (Fig. 4.1) also depends upon the nature of the particle 



'*■ •: :-saa 

* ■<: 

Fig. 4.4 Atmospheric pollution particles shadowed at a low angle of 23° to 
demonstrate the size of the finer particles present. X1 6,000. 

being studied; the larger the specimen, the larger will be the angle ( Hall, 
1956). Reduction of the shadowing angles down to approximately 14° 
is necessary for shallow structures, since the smaller the object, the lower 
must be the angle of deposition in order to form a recognizable shadow. 
Fig. 4.4 shows the shadowing of atmospheric pollution particles where 
the angle of deposition was extremely narrow (23°), and this has pro- 
vided another example of a positive print with a white shadow. Some 
authors offer equations relating the mass of metal evaporated to both the 
thickness of the resultant film and the shadowing angles ( Bradley, 1967 ) , 
although it must remain doubtful whether they possess any actual prac- 
tical value. Magnesium oxide crystals, shadowed at an angle of 45°, are 
shown in Fig. 4.5. The dark arrows indicate the direction of deposition, 
and the figure illustrates one of the difficulties associated with the inter- 
pretation of these micrographs. Reflected material, rebounded from the 
jig holding the specimen ( dotted arrow ) can be seen to have produced a 
shadow on the side of the crystal on which the material has been de- 


Fig. 4.5 Magnesium oxide crystals. The shadowing (45°) direction is indicated 

by the arrow ( -»); the display of the secondary shadow ( >) is due 

to the specimen being too close to the support holders; and the shadowing 
material has been reflected from this support to form another shadow in a direc- 
tion opposite to the initial cast. X1 4,000. 

A relatively simple method can be employed to estimate the thickness 
of the metal or carbon film which has been deposited. A small piece of 
white glazed porcelain, approximately 2.5 cm X 2.5 cm, is placed in the 
vacuum unit with a smear of vacuum oil covering one quarter of the 
surface area. The surface of the porcelain is placed at the same level as 
the specimen and facing the evaporation point. The metal or carbon 
deposited is controlled by the applied voltage, and the thickness may be 
estimated by the color of the deposition on the area of the porcelain not 
covered with vacuum oil. The area smeared with the vacuum oil will re- 


main white. The thicker the deposit the browner it will appear in rela- 
tion to the oil-smeared white porcelain. The initial point of difference 
between the clear and oil-smeared area will indicate a carbon thickness 
of approximately 50 A. The recently introduced quartz crystal thickness 
monitor can, however, provide an extremely accurate measurement of 
film thickness and will doubtless be preferred in future. Hayat (1970) 
has described the methods for the deposition of various types of films. 

The amount of material to be evaporated must depend upon the 
required resolution and the type of specimen which is to be examined. A 
list of materials with their various filament requirements is shown in 
Table 4.1. For the evaporation of gold/palladium alloy (60:40%) a coiled 
tungsten filament (0.5 mm diameter) is used together with a loop of the 
material 3-4 mm in length and 0.3 mm in diameter. Gold/palladium 
tends to give a coarse granular deposit of the type seen in the back- 
ground of Fig. 4.6, which illustrates the shadowing of a spirochaete. 
Deposition of this gold/palladium alloy has now been superceded by the 
use of platinum which gives a much finer film. Platinum, which requires 
a higher current for evaporation, is used with a 1.0 mm diameter straight 
tungsten filament. Chromium, zirconium and silicon monoxide, although 
used extensively in the early days of shadow casting, have been found 
of little value in modern procedure. 

Depending on the problem under investigation, many valuable varia- 
tions of the shadow casting technique can be applied. In one, the portrait 
technique, the coated specimen is turned through 180° and a further 
film, approximately 20% of the original deposit, is applied (Fig. 4.7). 
This technique allows the surface topography of what was the initially 
shaded aspect of the specimen to be examined for special characteristics. 

Conical shadowing, another variation of the technique, has been 
used in the biological field for the study of fibrous specimens and, in 
particular, nucleic acid molecules (Kleinschmidt and Lang, 1962; 
Kleinschmidt et al, 1962). The procedure essentially involves the oblique 
deposition of the coating while the specimen is being rotated ( Heinmets, 
1949). Whereas contrast is obtained in the usual procedure from the 
difference between coated and uncoated areas, in the conical shadowing 
technique the contrast depends upon the thickness distribution of the 
evaporated film. 

High resolution studies have been aided by the development of the 
platinum/carbon shadowing technique of Bradley (1959). The platinum 
and carbon are evaporated simultaneously to form an extremely fine non- 
granular film which does not produce aggregates of crystallites under the 
electron beam. Commercial rods or pellets are available with standard 
proportion of carbon and platinum, although a simple method involves 
placing a small loop of platinum wire (0.3 mm diameter, 4.0 mm length) 





















c c: 




E E E 
E E E 
6 t-: o 



Q. «> 

| 6 



CO (0 

T3 CD 


s § 

£ o 

CO » 


„ o 
c -° c 
<b _ a, 
E I E 

^ E c 

£ cd E 

_co x> _ca 

= ^ 5 = '^ 

<b -a 2 <» T) 

■- - ■ - £ a) 

_C0^_C0 CO Xl CO *- Xl CO 3 X! 

'o o o -2 '5 JJ ■= o ca « .2 'o o a 
O 2 O ii-OUJ u. o OcLUj OS UJ 



• ||* E E 

E S E E S ffl 

>- XI _ca 3 xi 

E c 5= (U C 

2 C j= E € - 

■fi — o -5 .>• o <-» o m 































































: — 
















E ^ 
























- . * s.- *-7. zJ* Xrt. ,«-\ •? : *J 



- »?.! 


Fig. 4.6 Spirochaete bacteria shadowed with gold/palladium 60/40. The 
granular grain is displayed. X24.000. 


over the points of the carbon rods which are extensively used for routine 
preparation of support membranes for electron microscopy. Application 
of a current to the carbon rods, with the development of the necessary 
temperature, results in the simultaneous evaporation of platinum and 
carbon. One disadvantage of this simple method, as opposed to the 
use of the commercial platinum/carbon rods, is that the amount of 
platinum evaporated is less easily controlled, and too high a platinum/ 
carbon ratio can lead to recrystallization of platinum within the film 
under heavy electron bombardment. However, since the beam intensities 
of electron microscopes differ widely, the capacity to determine the com- 
position of the platinum/carbon deposit, and to vary it depending on 
the available electron microscope, is of considerable value. 

Further improvement in resolution is also possible by introducing 
an aperture cut in a large screen placed between the carbon/platinum 

■ Direction of 
second shadow 

Initial shadow 

shadow material 

Direction of initial shadow 

Initial shadow material 

Fig. 4.7 The portrait technique. After the initial deposition of the shadowing 
material, the specimen is turned through 180° and a further deposit made of the 
shadowing material, approximately 20% in thickness of the initial material. 




Shield — ^^ * i< / X Direction of 

_ . _ . shadow 

2 mm. aperture 



Fig. 4.8 The platinum/carbon technique. Platinum and carbon are evaporated 
simultaneously. An aperture introduced between the source and the specimen 
increases the degree of resolution. 

source and the specimen (Fig. 4.8) (Bradley, 1959). A collimator used 
for a similar purpose has been described by Hibi ( 1952 ) . The aperture 
or the collimator improves the background and provides sharper defini- 
tion of the shadow. Furthermore, the evaporation source can be closer 
to the specimen if an aperture is used, since the screen tends to protect 
the specimen from the radiated heat. A distance of not less than 4.0 cm 
is, however, recommended. 

The use of carbon alone for shadowing has been disappointing, since 
the thickness of the carbon film required to achieve sufficient contrast 
lowers the resolution. A variety of other high melting point metals have 
been used in the shadow casting technique, but the difficulties in apply- 
ing sufficient voltage to evaporate these metals from a filament have 
prevented any real practical application. 

Tungsten produces a very fine background structure, but the long 
period for which the specimen must be exposed in order to deposit suffi- 
cient material for reasonable contrast is often damaging to the specimen. 
Tungsten oxide offers high resolution but the results are not consistent. 
Uranium wire can be satisfactorily evaporated if initially immersed in 
nitric acid. There are a few applications of these procedures to the study 
of biological materials, and Bradley ( 1967 ) has described in detail those 
that have been of value. However, these types of metals with high melt- 
ing points do produce the finest background structure within the cast, 
and techniques for their application using an electron beam source have 
been developed by Bachmann (1962). Iridium, tantalum, uranium and 
tungsten have been successfully evaporated by this procedure (Fig. 4.9). 

Fig. 4.9 Electron beam source evaporation. A: The surface of a lysolecithin 
crystal shadowed with tungsten/tantalum by electron beam bombardment. x370,000. 
B: Platinum/carbon shadowing on the lysolecithin crystal surface. The granular 
deposition contrasts with the finer structure of A. (Courtesy of Balzer High Vacuum 
Co., Ltd.) X370.000. 

'**■■ €X v*V ?'•'■•• :** -<$•&•■•./ ! 




The production of a fine metal deposit by the use of an electron beam 
source is also aided by the fact that there is no decrease in vacuum 
during the evaporation procedure, a characteristic of the normal filaments 
when heated. 


The technique of electron beam evaporation uses the kinetic energy of 
the electrons to produce the heat necessary to evaporate the metal or 
carbon. The material to be evaporated, which acts as the anode, is 
placed in a crucible and the electron beam focused on it. Electrostatic 
or magnetic fields can be used to focus the electron beam, but the elec- 

Fig. 4.10 Balzers vacuum coating unit, a small compact instrument which em- 
ploys both conventional and electron beam source evaporation facilities. 


trostatic field has the advantage of requiring a simple power supply, and 
the space necessary for the equipment is similar to that required for 
ordinary filament evaporation. For carbon evaporation, the carbon rod 
itself can be used without a crucible, although the rod can also act as 
a crucible for evaporation of metals if a slight depression is made in the 
end surface of the rod. The metal is then placed in the depression. Most 
modern electron beam units (Fig. 4.10) employ two crucibles, one for 
carbon deposition and the other for the shadowing metal. The crucibles 
can be controlled from outside the work chamber, so that the vacuum 
does not have to be broken to switch from one crucible to the other, and 
after carbon evaporation the crucible containing the metal is moved 
under the electron beam source, or vice versa. The target must be brought 
close to the emitter source for maximum efficiency. The electron beam 
evaporation equipment of the Balzer High Vacuum Co., for example, 
incorporates a magnetic baffle to minimize damage to the surface of the 
specimen being studied by stray electrons. Furthermore, replacement of 
the diffusion pumps by turbo-molecular pumps guarantees a completely 
oil-free vacuum. 


Except for its use in the sizing of bacteria and viral particles, there have 
been few applications of the shadow casting technique to the study of 
biological materials. Small particles can be difficult to identify on the 
screen of the electron microscope, and prior to the development of the 
negative staining technique, shadow casting provided one of the few 
means of locating them. Figure 4.11, for example, shows a micrograph of 
lambda bacteriophage shadowed with platinum. The shape of bacteria 
and viruses can also be investigated, and Fig. 4.12 shows a high mag- 
nification micrograph of lambda bacteriophage clearly illustrating the 
characteristic hexagonal head. The length of the tail can also be accu- 
rately assessed. 

When sizing particles which have been coated by the shadow casting 
technique, a measurement across the particle at right angles to the 
shadow direction should be made, since the particle dimensions can be 
considerably increased with the deposition of the material on the side 
directly facing the source. Again, before the use of thin-section staining 
procedures, the shadow casting technique was of some value in assessing 
the relationship of a high magnification study of a particular area of a 
thin section of tissue to a low power study of the specimen. Shadow cast- 
ing of the thin section can yield an electron micrograph with contrast 
sufficient to display the cellular organelles (Fig. 4.13). 


... .. -" .-.-■•; .v 'SWM|fKft^^sr'.-: . :,"-• • r." ;.. -\ :• ••'•;; &■■'.%, 

bUf . ■- "i-" . ■ '■ ^^jM 


*J5» •'"''' .^•^".'^■'Sr, 

Fig. 4.11 Lambda phage. Small particles with little electron dense contrast can 
be detected using the shadow casting technique. The lambda phage particles are 
highlighted against the background substrate. X60.000. 


The replication technique was developed for the study of electron 
opaque materials, such as metals or glass fibers, which are unsuitable for 
direct or thin-section evaluation with the electron microscope. The rep- 
lica, consisting of a thin film of an electron-transparent material corres- 
ponding to the surface structure of the specimen under investigation, can 
be studied in the microscope to produce information otherwise unattain- 

The surface topography of steel was investigated by Mahl (1940) in 
the first replication studies which made use of collodion. Polyvinyl 
formaldehyde, a plastic referred to as Formvar, was later employed by 



i^'^V-- ■"•'.'■ .-"■• :'•:.'••'.•: '•■•:'.'«*• ":>i>i".'-™ 

hf ■ A .■.,-.*"... i ■* 

Fig. 4.12 Lambda phage. The shadow casting technique allows the distinct 
hexagonal head to be displayed in the image; the size of the phage head and 
length of tail can also be accurately assessed. X180.000. 

Schaeffer and Harker ( 1942 ) for replication, and has been considered for 
a number of years to be the most reliable and consistent medium which 
could be used. Formvar can be dissolved in chloroform or ethylene 
dichloride, which when spread over the surface of the material to be 
replicated evaporates leaving a thin film of the plastic over the specimen. 
The rougher the nature of the specimen surface, the higher is the con- 
centration of the Formvar solution which must be applied, and a useful 
range is from 0.2 to 5.0% (w/v). Although this procedure has been 
widely used for the study of many metallic structures, its application to 
biological tissue has proved of little value, most probably due to the 
softer nature of such material. 

Although the use of carbon as a material for replication was intro- 
duced by Konig and Helwig in 1951, a thin film of carbon being deposited 
on the specimen by ionic decomposition of toluene vapor, the technique 
was never found to be satisfactory. However, a major development was 
the technique of carbon evaporation in vacuo by Bradley ( 1954 ) , which 
allowed the production of a replica, not only accurate in surface detail, 
but also stable under normal electron bombardment conditions. This 
procedure has proved to be of considerable value for the study of a 
variety of surface structures, but has not been extensively applied to the 
investigation of biological materials until the recent work undertaken in 
our laboratories and described later in some detail. 


Fig. 4.13 Liver section (methacrylate embedding). Unstained section for high 
magnification study which was shadow cast after evaluation to show the relation- 
ship between the high and low magnification of the organelles. X8.000. 

A number of different surface structures have been studied by the 
replication technique and the procedures employed in an attempt to pro- 
duce an accurate replica have varied depending on the nature of the 
material. A single stage replica (Fig. 4.14) is produced by the deposition 
of the replicating material directly onto the specimen, and the replica is 
examined in the electron microscope after stripping from the specimen. 
Moreover, as described earlier, the contrast of the replica can be in- 
creased by shadow casting. 

The plastic Formvar, deposited onto tissue surfaces, is difficult to 
remove. Generally, another plastic, Bedacryl, is deposited on the top of 


, A A_ 

I I * Specimen 

WUmf VMM////////// ^w^fa 

■ Replication material 




Fig. 4.14 Single stage replica technique. 

the Formvar and Scotch tape applied to the solidified Bedacryl. The 
plastic layers can then be removed and the Scotch tape and Formvar 
separated by dissolution of the Bedacryl in acetone. The Formvar replica 
may be studied directly in the electron microscope but, because of the 
limited contrast available, it is invariably shadowed in order to produce 
a reasonable image. However, thin plastic films tend to become charged 
under normal electron bombardment. This, together with the heat dis- 
sipated from the electron beam, causes instability and drifting of the 
image during evaluation. A procedure developed to overcome this is to 
deposit a carbon layer perpendicular to the plane of the plastic on the 
surface of the replica. The Formvar may then be dissolved and the 
second carbon impression mounted on a grid and obliquely shadowed 
with a metal to achieve reasonable contrast. 

Alternatively, the initial Formvar replica is shadowed and a carbon 
layer deposited before separation of the Bedacryl and Scotch tape. The 
Formvar may then be dissolved in chloroform or ethylene dichloride, 
depending on the original solution and the replica mounted on the grid. 
Although this technique has been invaluable in the study of metallic or 
similar surfaces, results on biological samples have been unsatisfactory. 
The rapid evaporation of the solvent and the resultant deposition of the 
plastic Formvar film, when applied to soft pliable biological tissues, 
results in distortion with the consequent formation of artifacts. Some 
degree of dehydration of certain soft tissues, either in alcohol or in vacuo, 
often provides a more rigid surface. This technique, however, described 


originally by Wolf (1948) for the replication of endothelial tissue, has 
not been found to yield consistently useful results by others in this field. 
Another technique for the replication of tissues such as muscles and liga- 
ments using collodion (Reed, 1960) has been described in full by 
Bradley ( 1967 ) , but would appear to be a complicated and tedious pro- 
cedure for routine application. 

The development of carbon evaporation in vacuo by Bradley ( 1954 ) 
provided a significant step forward in replication studies. A single-stage 
carbon replica procedure can be applied to easily dissolved specimens, 
and Fig. 4.15 shows the evaporation of carbon onto a sucrose crystal, 

Fig. 4.15 Sucrose crystal. Direct carbon replica. x3,000. 


which was then dissolved in water and the replica floated onto the grid. 
Only specimens with fine surface structures are suitable for study by this 
direct replication technique since the carbon film is easily fractured 
when the specimen is dissolved. Furthermore, if biological tissues are 
being studied, the chemical reagents necessary to dissolve the organic 
material can cause fracture of the replica and severe damage to the 
surface structure. 

The technique which has been found of most value consists essen- 
tially of obtaining first an impression of the specimen in a plastic material 
(the double stage procedure, Fig. 4.16) and then of carbon deposited 
onto this impression. Dissolution of the plastic and examination of the 
carbon replica has produced excellent results, and Bradley ( 1956, 1957 ) 
has published some extremely good photomicrographs of replicas of bac- 
terial specimens, pollen grains, viruses and various lower forms of plant 
life. A number of multi-stage techniques have been developed and used, 
one, for example, by Haanstra ( 1954 ) , but the more stages involved, the 
greater are the chances of producing changes in the surface topography 
with resultant artifact formation. Some of these have been described in 
detail by Bradley ( 1967 ) for application to metallic surfaces but, again, 
little has been undertaken in the biological field. A number of replication 

y\ a. 




S -t Evap orated 

Plastic cast 



Fig. 4.16 Double stage replica technique. 






Barnes, R. B., Burton, C. J. and 

Scott, R. G. (1945) 
Dinichert, P. and Kellenberger, E. 

Konig, H. and Helwig, G. (1951) 

Haefer, R. (1954) 

Haanstra, H. B. (1954) 

Hibi, T. and Yada, K. (1954) 

Bradley, D. E. (1955) 

Schreil, W. and Schleich, F. (1955) 

Bradley, D. E. (1956) 

Bradley, D. E. (1958) 

Biological tissue 
and pulp fiber 


Biological tissue 

Biological fibers 

Biological fibers 

Biological virus 
Protein crystals 



of toluene) 
of toluene 
Methyl methracry- 
late (polymerized 
in situ) evapor- 
ated metals 
Methyl methracry- 
late (Polymerized 
in situ) evapor- 
ated metals 
Evaporated carbon 
Silicon monoxide 
Carbon replicas 

techniques which have been applied to biological tissues are listed in 
Table 4.2. 

The use of a water soluble plastic, polyvinyl alcohol (PVA), by 
Powell et al, (1954) and Pooley and Henderson (1966) has been found 
to be invaluable for the replication of tissue, and has formed the basis 
of all the replication work on biological tissue from this laboratory. Since 
PVA is water soluble, it is conveniently handled as a solution. The slow 
solidification of the solution in contrast to the evaporation of the 
chlorinated hydrocarbons permits a higher degree of accuracy of the 
replicated surface structure. The water base for the PVA causes less 
damage to the tissue surfaces and can be easily removed from the tissue 
without causing disintegration of the specimen or damage to the replica. 
Moreover, the thickness of the PVA film can be varied to suit the rough- 
ness of the surface under study. 

The ease of handling PVA and the simple stripping procedure makes 
it possible to produce a number of preliminary shippings to remove 
loose material, thereby ensuring a clean tissue surface before taking the 
final replica for evaluation. Polyvinyl alcohol films are not suitable for 
direct study in the electron microscope by themselves. However, evapo- 


ration of platinum and carbon onto replicas, with the subsequent dissolu- 
tion of the PVA and mounting of the carbon/platinum cast on the grid, 
has been found very satisfactory. The PVA can be removed by floating 
the replica in a hot water bath at approximately 97°C for 15 min. The 
replica is then mounted on the electron microscope grids and is ready 
for microscopic investigation. 

The tissue to be examined by this procedure is embedded in acetone- 
softened sheets of cellulose acetate ( 0.5 mm thick, 25 X 25 mm 2 ) mounted 
on glass microscope slides for ease of handling (Fig. 4.17) (Henderson, 
1969). Fresh unfixed tissue can be embedded directly, whereas tissue 
which is softer and moist can be dehydrated in vacuo to ensure sufficient 
rigidity. Paraffin in routinely fixed histological specimens is removed by 
immersion in xylene before ethanolic dehydration and embedding in the 
cellulose acetate. Care is necessary to prevent the tissue from being com- 
pletely immersed in the cellulose acetate. This is then allowed to harden, 
and scalpel trimming of the tissue may then be required to provide a 
better surface for replication. The tissue is outlined with strips of Scotch 

(ZeMu/osa AeeZ-at-a. . 

-Embedded Tssua.. 

a.. £>. 

H/sto/agico/ Thin Sect/or? an S//de.^, ^- Seoteh Topa. 



s\..\>> • 


d. ~ 

J l 


Scohch Tb/oa.- 

£*-■■■■ Sa/oJica 


Fig. 4.17 Acetate/PVA replication technique. (Courtesy of Royal Microscopical 


Fig. 4.18 Galena ore replica. A thin film of PVA is retained on the carbon/ 
platinum cast to provide additional support. X15.000. 

tape in order to form a shallow well and to aid removal of the replica. A 
10% (v/v) solution of PVA is applied to cover the tissue. A 5% solution 
can be applied to smoother surfaces. Slow solidification is allowed to 
take place overnight, but if a faster result is required, the slide may be 
placed on a hot plate. Artifacts, however, can be produced by the warm- 
ing procedure, through degassing of the tissue with the consequent pro- 
duction of bubbles in the replica. On hardening, the PVA film replica of 
the tissue surface is stripped off and successive strippings through the 
embedded tissue can be taken to allow a stepwise study of the specimen. 
For the study of very rough surfaces it can be advantageous, when dis- 
solving the PVA, to leave a very thin film backing the platinum/carbon 
cast. This will support the replica and prevent its fragmentation, without 
loss of definition of the electron microscope image. Figure 4.18 shows an 
example of this procedure applied to a surface of Galena ore, the depth 
of the surface structures and sharp features being evident. It is important 
to realize that small pieces of tissue and any embedded foreign particles 


may also be extracted by the PVA film during the stripping procedure. 
As will be described later, the structure of such particles and their com- 
position can be further investigated using electron diffraction or micro- 
analysis. The location of particles in situ can also be assessed, especially 
if adjacent thin tissue sections, stained for normal routine histology, are 
also being evaluated at the same time by optical microscopy. 

A number of reports on the application of this extraction-replication 
technique have described the surface structure of tissues and the identi- 
fication of particles extracted from a variety of tissues ( Henderson, 1969; 
Henderson et al., 1969; Henderson et al., 1970; Henderson et at, 1971). 

Figure 4.19 shows an example of the degree of resolution obtainable 
by this replication technique. The surface replication of collagen fibers 
in tissue from a tumor of the cervix is shown, clearly illustrating the 
characteristic periodicity of the 640 A banding of the fibers. The replica- 
tion pattern of striated muscle, with its typical band-like structure, is 
seen in Fig. 4.20. Both specimens were taken from paraffin-embedded 
histological sections. Figure 4.21 shows replicated smooth muscle tissue 

Fig. 4.19 Collagen fibers. The replica displays the 640 A banding of the fibers 
(from a tumor of the cervix). X20.000. 


Fig. 4.20 Striated muscle. The H and A bands are displayed by the replication 
technique. X8.000. 

from human hypertrophic prostrate embedded directly into cellulose 
acetate without fixation. Accurate reproduction of the muscle blocks can 
be seen, especially when related to Fig. 4.22, which shows a normal elec- 
tron microscope thin section of the tissue. A higher magnification of this 
smooth muscle replica (Fig. 4.23) shows the individual fibers of the 
muscle band. 

A number of studies have been concerned with surface characteristics 
of normal human erythrocytes and with those from patients with various 
blood disorders. Most have made use of either interference or scanning 
electron microscopy and the reports would suggest that great care is 


Fig. 4.21 Hypertrophic prostatic tissue. The tissue was embedded directly into 
the acetate block without fixation, before the application of PVA. The muscle 
blocks and their fine fibers can be compared with the micrograph illustrated in 
Fig. 4.22. X5.000. 

necessary to prevent the formation of artifacts during the preparation 
of the cells. Kayden and Bessis (1970) described the crenated appear- 
ance of red blood cells by interference microscopy after washing in 
normal saline. This effect could be eliminated if the cells were resus- 
pended in indigenous plasma. They suggest that the normal routine 
blood smear on a slide, which has been dried, fixed and stained, shows 
the most artifacts. Scanning electron microscopy studies (Clarke and 
Salsbury, 1967) showed that red blood cells have an extremely pro- 
nounced central depression producing the characteristic "doughnut-like" 
appearance on the scanning micrographs, although the extent of the 


SI i : '< 

Fig. 4.22 Hypertrophic prostatic tissue. The thin section micrograph shows the 
muscle blocks, which can be compared with the same tissue replicated in Fig. 
4.21. X4.000. 

central depression is not nearly so pronounced on interference micros- 

In this laboratory, the replication technique has been applied to the 
study of blood cells in various states of preparation (Henderson, 1971). 
Figure 4.24 shows replicated cells from a blood specimen of a premature 
infant of 30 weeks gestation. The specimen was smeared onto a slide 
and allowed to dry naturally before the slide was applied to the cellulose 
acetate sheet in the usual way. The smooth outer membrane is evident, 
but also clearly seen is the crenated cell surface which appears to be 
more prevalent in infant blood than in an adult specimen examined by 
this procedure. These microprojections have previously been described 


Fig. 4.23 Hypertrophic prostatic tissue. Higher magnification of the replicated 
muscle blocks to show their individual fibers. x12,000. 

in relation to various blood disorders by a number of research groups 
(Oski et al, 1964; Smith et al, 1964; Grahn et al, 1968; Kayden and 
Bessis, 1970). The doughnut-like appearance is not evident, although a 
slight concave depression can be observed in some cells. A replication of 
blood cells from a histological thin section, studied after removal of the 
paraffin-wax, is shown in Fig. 4.25. The smooth outer membrane is pre- 
served, but cell distortion can be seen, the angular contours presenting 
an appearance characteristic of normal electron microscopy ultrathin 
section preparations in which erythrocytes are present. This may be due 
to the fixation procedures used. 

It is interesting that with replication studies of erythrocytes, a num- 
ber of PVA strippings may be required to remove material loosely at- 
tached to the cell surface before the smooth surface membrane is rep- 
licated. Continuous stripping removes the outer membrane to show the 
internal coarse granular structure (Fig. 4.26). This granular appearance 
has also been described by other groups ( Lewis et al, 1968; Stuart et al, 
1969) using ion-etching techniques with the scanning electron micro- 
scope. A similar granular appearance is obtained when red cell ghosts, 


Fig. 4.24 Premature human blood cells. The replica was obtained from a slide 
on which the blood smear had been allowed to dry naturally, without any fixation 
procedures. The smooth outer membrane is evident; also a number of crenated 
cells are shown. x3,000. 

isolated by homogenation and fractionation of red cells, are collected and 
replicated (Fig. 4.27). The rough structure, quite different from the 
outer membrane surface (Fig. 4.24), may well represent the inner mem- 
brane surface. 

Two further applications of the replication technique to biological 
specimens are seen in Figs. 4.28 and 4.29, showing, respectively, a rep- 
licated rabbit sperm and the typically granulated phagocyte with its 
characteristic pseudopodia. 



Fig. 4.25 Histological preparation of blood cells. The paraffin was removed 
before application of the replication technique. The smooth outer membrane re- 
mains, but some distortion of the cells is evident. This is similar to the angular 
appearance displayed in normal thin section micrographs of blood cells, but differs 
from that seen with the replication procedure shown in Fig. 4.24. x3,000. 

The technique of ultrathin sectioning of biological tissue for electron 
microscopic evaluation has never been successful when applied to bio- 
logical tissue in which hard particles or crystalline deposits are known to 
be found. The hard particles tend to adhere to the ultramicrotome knife 
edge, tearing the section and causing it to disintegrate during the sec- 
tioning process. Furthermore, crystalline deposits often break up and 
are then deposited in positions other than those in which they were 
originally located. The replication technique has been found to be ex- 





Fig. 4.26 Human blood cell. The replica was obtained from a slide on which the 
blood smear had been allowed to dry naturally. After embedding in acetate, re- 
peated strippings of the PVA casts removed the outer membrane to display the 
granular structure of the interior. X10.000. 

tremely valuable, not only in studying the presence and localization of 
various crystalline deposits, but also in their identification. The technique 
has been applied to a study of osteoarthritic joints and also to tissue from 
the supraspinatus tendon. Schumacher (1968) has investigated synovial 
membrane from patients with what is referred to as pseudogout by thin- 
sectioning technique. He reported the presence of crystalline particles 
in the tissue and also the appearance of spaces within the section which 
he considered had been sites of crystal particles. In this laboratory, 
material from patients with osteoarthritic conditions was obtained direct 
from the operating theater and embedded directly into the cellulose 
acetate without any fixation procedures. Crystalline deposits were ex- 
tracted with the initial PVA cast and, at the same time, subsequent strip- 
pings also provided replicas of the deposited crystals in situ. Figure 4.30 
illustrates a replica and extracted crystalline particles from the cartilage 
of an osteoarthritic knee. The extraction of the crystalline particle in a 
natural state without any fixation procedures renders it suitable for 


Fig. 4.27 Replica of red cell ghosts showing the rough structure which is in 
contrast to the fine surface displayed on the outer surface. Possibly represents 
the inner membrane surface. X1 6,600. 

analysis by the electron microscope microanalyser (AEI EMMA-4) in 
order to determine its composition. Such analyses have been undertaken 
to ascertain the calcium: phosphorus composition of the crystalline mate- 
rial. Crystalline deposits in the osteoarthritic knee can be substantial and 
our studies show aggregations of individual particles, many microns in 
length, both in the synovial fluid and attached to the synovial membrane 
(Fig. 4.31). 

The supraspinatus tendon is a common site for deposition of a chalk- 
like material which, at surgery, is extruded like toothpaste. The material 
comprises an aggregation of very fine crystals (200-2000 A in length) 
which loses its crystalline form under the electron beam in the same 
manner as the larger crystals (up to 6 /j. ) from osteoarthritic joints. 
Ultrathin section studies showed that in this material from the tendon 


SoEL ! 

— H 








Fig. 4.28 Rabbit spermatozoon, replicated from a vaginal smear. The acrosomal 
cap and the axial filament complex are shown; the background granulation is 
caused by the seminal plasma. X5.000. 

some of the very fine crystals appeared to be membrane bound (Fig. 
4.32). Application of the extraction-replication procedure to this material 
produced a micrograph (Fig. 4.33) complementary to the pattern ob- 
served in the thin-section micrograph (Fig. 4.32). High magnification 
studies showed the crystals in the thin section and the replica to corre- 
spond in size and shape, both displaying the same instability under the 
electron beam. Schumacher (1968) suggested that primary intracellular 
precipitation of crystalline material can take place with these clinical 
conditions and the studies briefly described above, using the ultrathin 
sectioning and extraction-replication techniques, would tend to support 
this hypothesis. It also demonstrates the usefulness of the replication 
technique as a complement to the normal ultrathin section studies of 
biological tissues. 

The replication technique has been extensively applied to other 
studies of foreign particles within biological tissues. Figure 4.34 shows 









Hil l 



Fig. 4.29 Phagocyte. The replica was made from a culture; the granular interior 
and pseudopodia are evident. x4,000. 

a replica of an asbestos fiber located in lung tissue from a patient with 
pneumoconiosis. After a number of stoppings, the ferritin body seen 
surrounding the asbestos is partly removed to display the surface char- 
acteristics of the fiber. Thus the procedure may be applied to the study 
of tissue from subjects with occupations in which there are accepted 
carcinogenic hazards to the staff. Figure 4.35 shows the surface char- 
acteristics of a crystalline material found in a specimen of lung from 
a man involved in the manufacture of abrasive materials. The surface 
structure of the particle can be compared with replicated standard refer- 
ence materials involved with the industry with a view to establishing its 
origin. The fact that much of the crystalline material is also extracted 


Fig. 4.30 Osteoarthritis. Cartilage was removed from an osteoarthritic knee and 
replicated without fixation procedures. The crystal deposition is shown and ex- 
tracted crystalline particles are adhering to the cast. X10.000. 

with the replica allows the application of further analytical procedures. 
For example, certain crystalline material will produce a characteristic 
diffraction pattern, seen in Fig. 4.36A, which was obtained from a talc 
particle embedded in the capillary wall of tissue from a squamous cell 
carcinoma of the cervix. The replica showing the extracted particle is 
seen in Fig. 4.36B; the characteristic decorative effect produced on the 
talc particle during the procedure is of interest. This is caused by aggre- 
gation of the platinum atoms as they impinge on the crystal surface, seen 
more clearly in the higher magnification of the crystal (Fig. 4.36C). The 
decoration effect is described in detail in another chapter in this book. 
Anthophyllite, of the asbestos group of materials, is the only other 


Fig. 4.31 Crystals attached to the synovial membrane. X1 0,000. 

crystalline material known to display this decorative pattern, and it may 
be significant that anthopyllite is converted to talc by natural geological 
metamorphic processes. Use of the EMMA-4 for the analysis of these 
crystals has confirmed that the silicon/magnesium composition of the 
material is similar to that of talc standards and emphasizes the potential 
value of the extraction-replication technique in association with electron 
microscope micro-analysis. If it is considered that the composition of the 
shadowing metal may interfere, the deposition of carbon alone will allow 
the micro-analysis procedure to be carried out in the extracted particles. 
Similar analyses (Fig. 4.37) have been performed on asbestos fibers 

Fig. 4.32 Supraspinatus tendon. Ultrathin section micrograph of chalk-like ma- 
terial, removed at surgery, displaying very fine crystal deposits, some membrane 
bound, (see Fig. 4.33). X8.000. 

Fig. 4.33 Supraspinatus tendon. Application of the replica technique to the 
chalk-like deposits demonstrated in Fig. 4.32. The fine crystals appear to be 
membrane bound. x8,000. 


Fig. 4.34 Pneumoconiosis. A replica of an 
asbestos body located within a section of 
lung tissue. Repeated strippings of the 
PVA cast has removed the ferritin coating 
to display the surface of the asbestos fiber. 
X8.000. (Courtesy of the Royal Microscopi- 
cal Society.) 

Fig. 4.35 (below) Pneumoconiosis. Histo- 
logical section of lung replicated to dis- 
play a foreign particle with a distinct sur- 
face structure, which could be related to 
the industrial material to which the subject 
was exposed. X8.000. 


(A) Diffraction pattern from a talc particle. 

(B) The replica showing extracted particles embedded in the wall of a capillary. 



(C) The characteristic decoration pattern is displayed on the talc crystals, when 
platinum is deposited onto the crystal surface in vacuo. A number of lattice planes 
are also evident. X32.000. 

Fig. 4.36 Squamous cell carcinoma of the cervix. The replica was cast from 
the histological slide of the tumor tissue. Talc particles are shown embedded in 
the tissue. A number of procedures could be employed to provide evidence that 
the particles were talc. 

extracted from mesothelioma tissue (Fig. 4.38) and identified by their 
shape as crocidolite or amosite (Henderson et al., 1970). An adequate 
knowledge of the appearance of standard reference materials is, of 
course, essential in order to attempt to identify crystalline material by 
surface or structural characteristics. Thus the extraction-replication tech- 
niques can be successfully used for the study of foreign particles such as 
kaolin, mica, coal, forge dusts and the various types of asbestos localized 
in lung tissues. Skin, bone and dental structures have also been found to 
be particularly susceptible to investigation by this method. 

It would now seem that after a period of relative inactivity the field 
of replication studies is rapidly developing with the advent of new in- 
strumentation and the application of the technique to the study of 






Probe currant 

Proba dla 

X Ray 30O- 




-*9 ° 


" r»bo ' 


Mto ' r9aO 

AOP Crystal 

X Ray 30O 

P ^ 2 oo. 

Probe current 
IO _8 A 

Wavelength ~* * 
Spectrum of Iron. ( t* ) 

7-OIO 7<SSO" TCivO" wSo 7-lVo 7-2IO 7i50 

Wavelength — > A 
Spectrum of Silicon. (Si) 

Fig. 4.37 X-ray spectra of iron and silicon in an asbestos fiber extracted from a 
mesothelioma and studied with EMMA-4. LiF: lithium fluoride crystal; ADP: am- 
monium dihydrogen phosphate. 

biological materials. The use of procedures providing much higher resolu- 
tion can be further evaluated using the technique of specimen tilting in 
the electron microscope. Tilt angles of up to 120° can be achieved and 
the procedure should allow the investigator a new parameter for the 
interpretation of surface structure. Application of laser optical diffrac- 
trometry to the recorded image will also offer a new exciting approach 
which could be invaluable in the future development of these studies. 


The authors gratefully acknowledge the generous financial support of the 
Tenovus Organization. They are also grateful to Mr. D. E. Evans, Department 
of Geology, National Museum of Wales, for the natural minerals which were 
required for reference purposes and would like to thank their colleagues in 
the Welsh National School of Medicine, Cardiff, for their ready cooperation in 
supplying biological tissues: Professor A. C. Turnbull, Department of Obstetrics 
and Gynaecology, and Dr. C. A. F. Joslin, Velindre Memorial Centre for 
Cancer Research, for tissue from the ovary and cervix; Professor P. Gray, 
Department of Child Health, for the blood specimens; and Mr. H. Richards, 
University Hospital of Wales, for the samples of material from arthritic pa- 
tients. The specimen of lung tissue containing the abrasive materials was 
kindly provided by Dr. E. N. Willey, Palms of Pasadena Hospital, St. Peters- 



Fig. 4.38 Mesothelioma. Extracted asbestos fibers displayed on the replica from 
a histological section; the fibers present themselves for microanalysis techniques. 

burg, U.S.A. Finally, they are indebted to Mrs. M. Lewis for her care in 
typing and checking the manuscript. 


Bachmann, L. (1962). Shadow-casting using very high melting metals. Proc. 
5th Inter. Cong. Electron Micros, Philadelphia. 1, FF-3. 

Barnes, R. B., Burton, C. J., and Scott, R. G. (1945). Electron microscopical 
replica techniques for the study of organic surfaces. /. Appl. Phys., 16, 730. 

Bradley, D. E. (1954). Evaporated carbon films for use in electron micros- 
copy. Brit. }. Appl. Phys. 5, 65. 

. (1955). A replica technique for "reflexion" electron microscopy. Brit. 

}. Appl. Phys., 6, 191. 

. (1956). Uses of carbon replicas in electron microscopy. /. Appl. Phys., 

27, 1399. 


. (1957). Some carbon replica techniques for the electron microscopy of 

small specimens and fibres. Brit. J. Appl. Phys., 8, 150. 

(1958). Simultaneous evaporation of platinum and carbon for possible 

use in high resolution shadow-casting for electron microscopy. Nature, 181, 

. (1959). High-resolution shadow-casting techniques for the electron 

microscope using the simultaneous evaporation of platinum and carbon. 

Brit. J. Appl. Phys., 10, 198. 
. (1967). Replica And Shadowing Techniques. In: Techniques For Elec- 
tron Microscopy. (Kay, D., ed.) Blackwell Scientific Publications, London. 
Clarke, J. A., and Salsbury, A. J. (1967). Surface ultramicroscopy of human 

blood cells. Nature, 215, 402. 
Dinichert, P., and Kellenberger, E. (1948). Nouvel appareil pour les procedes 

d'empreientes. Experientia, 4, 407, 449. 
Grahn, E. P., Dietz, A. A., Stefani, S. A., and Donnelly, W. J. (1968). Burr 

cells in hemolytic anemia and cirrhosis. Amer. J. Med., 45, 78. 
Haanstra, H. B. (1954). The use of surface-active materials in electron 

microscopy, particularly for surface replicas. Proc. 3rd Inter, Cong. Electron 

Micros, London, p. 489. 
Haefer, R. (1954). Formation of thin layers in a self -maintained high-vacuum 

discharge and its application in electronmicroscopy specimen technique. 

Proc. 3rd Inter. Cong. Electron Micros, London, p. 446. 
Hall, C. E. (1956). Visualization of individual macromolecules with the elec- 
tron microscope. Proc. Nat. Acad. Sci., 42, 801. 
Hayat, M. A. (1970). Principles and Techniques of Electron Microscopy: 

Biological Applications. Vol. 1. Van Nostrand Reinhold Company, New 

Heinmets, F. (1949). Modification of silica replica techniques for study of 

biological membranes and application of rotary condensation in electron 

microscopy. /. Appl. Phys., 20, 384, 
Henderson, W. J. (1969). A simple replication technique for the study of 

biological tissues by electron microscopy. /. Microscopy, 89, 369. 
Henderson, W. J. (1971). Application of an extraction replication technique 

for the study of biological materials. Micron 2, 250. 
Henderson, W. J., Gough, J., and Harse, J. (1970). Identification of mineral 

particles in pneumoconiotic lungs. J. Clin. Path., 23, 104. 
Henderson, W. J., Harse, J., and Griffiths, K. (1969). A replication technique 

for the identification of asbestos fibres in mesothelioma. Europ. J. Cancer, 5, 

Henderson, W. J., Joslin, C. A. F., Turnbull, A. C, and Griffiths, K. (1971). 

Talc and carcinoma of the ovary and cervix. /. Ohstet. Gynaeco. Brit. 

Commonw., 78, 266. 
Hibi, T. (1952). A new device for metallic shadow-casting. Rev. Sci. Instrum., 

23, 383. 
Hibi, T., and Yada, K. (1954). High resolution and their application. /. 

Appl. Phys., 25, 712. 


Holland, L. (1956). Vaccum Deposition Of Thin Films. Chapman and Hall, 

Kayden, H. J., and Bessis, M. (1970). Morphology of normal erythrocyte and 
acanthrocyte using Nomarski optics and scanning electron microscopy. 
Blood, 35, 427. 

Kleinschmidt, A. K., and Lang, D. (1962). Intrazellulare desoxyribonuclein- 
saure von bakterien. Proc. 5th Inter. Cong. Electron Micros, Philadelphia, 
2, 0-8. 

Kleinschmidt. A. K., Lang, D., Jackerts, D., and Zahn, R. K. (1962). Prepara- 
tion and length measurements of the total desoxyribonucleic acid content 
of T2 bacteriophage. Biochim. Biophys. Acta, 61, 857. 

Konig, H., and Helwig, G. (1951). Aber dunne aus Kohlenwasser stoffen 
durch Elektronenoder IonenbeschuB gebildete Schichten. Z. Phys., 129, 491. 

Lewis, S. M., Osborne, J. S., and Stuart, P. R. (1968). Demonstration of an 
internal structure within the red blood cell by ion etching and scanning elec- 
tron microscopy. Nature, 220, 614. 

Mahl, H. (1940). A plaster-cast method for supermicroscopical investigation 
of metal surfaces. Metallwirtsch, 19, 488. 

Oski, F., Nathan, D. G., Sidel, V. W., and Diamond, L. K. (1964). Extreme 
hemolysis and red cell distortion in erythrocyte pyrovate kinase deficiency. I. 
Morphology, erythrokinetics and family enzyme studies. New Eng. J. Med., 
270, 1023. 

Pooley, F. D., and Henderson, W. J. (1966). PVA as a replication medium 
for electron microscopy. Res. Techn. lustrum., 18, 6. 

Powell, A. S., Lebras, L. R., Bobalek, E. G., and Fisher, W. (1954). An im- 
proved replica technique for electron microscopy of paint films. /. Appl. 
Phys., 25, 757. 

Reed. Personal Communication to Bradley, D. E. (1967). Replica And Shadow- 
ing Techniques. In: Techniques For Electron Microscopy. (Kay, D., ed.) 
Blackwell Scientific Publications, London, p. 111. 

Schaeffer, V. J., and Harker, D. (1942). Surface replicas for use in the elec- 
tron microscope. /. Appl. Phys., 13, 427. 

Schreil, W., and Schleich, F. (1955). Elektronenmikroskopische untersuchun- 
ger on fixierten und lebenden mikroorganismen unter anwendung der ab- 
druckverfahren. Z. Hygiene, 141, 576. 

Schumacher, H. R. (1968). The synovitis of pseudogout: electron microscopic 
observations. Arihrit. Rheuma., June, 426. 

Smith, J. A., Longergan, E. T., and Sterling, K. (1964). Spur cell anemia: 
hemolytic anemia with red cells resembling acanthocytes in alcoholic 
cirrhosis. New Eng. }. Med., 271, 396. 

Stuart, P. R., Osborn, J. S., and Lewis, S .M. (1969). A new approach to red 
blood cell structure. Med. Biol. Ill, 19, 202. 

Williams, R. C, and Wyckoff, R. W. G. (1946). Applications of metallic 
shadow-casting to microscopy. /. Appl. Phys., 17, 23. 

Wolf, K. (1948). II netodo del micro rilievoel istologia plastica. Minerva Med., 
39, (II), 19. 

5 High Resolution 

R. Abermann* 

Physikalisch-Chemisches Institut, 
Universitat Innsbruck, Austria 

M. M. Salpeter 

School of Applied and Engineering Physics and 
Section of Neurobiology and Behavior, 
Cornell University, Ithaca, New York 

L. Bachmann 

Institut fur Technische Chemie, 
Technische Universitat, Munich, Germany 

"Present address: School of Applied and Engineering Physics, Cornell University, 
Ithaca, New York. 


'hadowing was probably the first successful method for the study 
of biological material with electron microscopy. Pictures of shadowed 
viruses were used to demonstrate the capabilities of early electron 
microscopes. Since then, particle shadowing has to a large extent 
been replaced by negative staining, while surface replication is losing 
much of its utility with the advent of scanning electron microscopy. 
However, with the introduction of freeze etching, there has been a re- 
newed interest in high resolution surface replication of biological mate- 
rials. Fortunately, the method of shadow casting has made considerable 
progress in the meantime. It has improved its capacity for attaining high 
resolution and now stands as a true competitor to some alternative 
methods, even to negative staining. In this chapter, an attempt has been 
made to discuss the technical aspects as well as the capabilities and 
limitations of biological applications of high resolution shadowing. No 
effort was made to deal in any detail with the general aspects of shadow 
casting since they are the subject of chapter 4. 

Shadow casting is intended to depict surface geometry. When a 
pointed light source shines on a surface, any unevenness is revealed by 
shadows and variations in brightness. By analogy, unevenness in a sur- 
face should also be revealed by a metal evaporated from a point source, 
casting shadows and forming a film whose thickness is dependent on the 
amount of material and the angle of incidence. Why then is high resolu- 
tion shadowing a problem? 

First, it is obvious that one cannot reveal a structure by a film much 
thicker than the size of the structure itself. However, since a 10 A film of 
heavy metal already provides sufficient contrast, one should expect to 
resolve structures in this size range. Yet this is not the case. If one looks 
with high magnification at a metal film of this thickness, the real problem 
becomes apparent. The film is not a film at all, but merely a collection of 
isolated particles. If one could reduce this granularity, making more but 
smaller dots per area, the resolution of the replica would be improved in 
the same manner as high quality print is superior to newsprint. 

There are also secondary problems which can degrade the informa- 
tion obtainable from a shadowed specimen. These include specimen 
changes due to contamination in the vacuum system or due to heat 
damage during shadowing. Furthermore, the final replica can be affected 
by recrystallization and oxidation in air or during stripping, chemical 
damage during cleaning, or beam damage in the electron microscope. 



Finally, the interpretation of highly magnified shadowed specimens might 
be misleading unless shadowing is differentiated from another phenom- 
enon known as preferred nucleation or decoration. For a clearer under- 
standing of these problems, it might be useful to have a quick look at 
thin film formation. 


The shadowing beam consists of atoms of the evaporated metal (non- 
atomic vapor occurs in carbon ) . On reaching the specimen surface, these 
atoms release their excess energy until they are in thermal equilibrium 
with the surface. During thermal accommodation the atoms will be 
highly mobile on the surface and, unless the surface is at a very low 
temperature, the atoms retain enough energy to continue to migrate 
until they are either trapped by a strong binding site provided by the 
specimen or become stabilized by combining with other atoms or atom- 
clusters of the evaporated material. The still very small but already stable 
clusters of atoms, commonly called nuclei, then continue to grow during 
subsequent evaporation. 

It is obvious that if these clusters are far apart they cannot provide 
information with high resolution. Thus, shadowing resolution is coupled 
to the surface mobility of the shadowing material which modifies the 
initial statistical distribution of the impinging metal atoms. The main 
problem, therefore, is to decrease this surface mobility. 

It has been shown that the surface mobility at a given temperature 
depends primarily on a specific binding energy of the atoms of the 
shadowing material with itself and with the specimen surface (for a 
detailed discussion of thin film formation, see Mayer, 1955; an article 
by Bassett et al., 1959, which is also illustrated with electron micrographs 
may be easier for biologists). It follows from these considerations that 
shadowing resolution can be increased by cooling the specimen, by using 
shadowing material of high melting point (i.e., strong internal binding) 
and, wherever possible, by choosing an optimum specimen surface ( e.g., 
substrate for particle shadowing, see Fig. 5.1). A completely different 
approach to suppressing surface mobility is the simultaneous evaporation 
of a high density, lower melting metal, and a substance with very high 
melting point, such as carbon (Bradley, 1958a and b, 1959, 1960; Knoch 
and Konig, 1956). 


In addition to producing films capable of high resoluton, evaporation 
methods must also fulfill the following conditions: (1) In order to obtain 


Fig. 5.1 Solution containing particles and buffer dried onto mica, shadowed 
with tantalum-tungsten and backed with carbon. The granularity of the shadowing 
material is nonuniform on the patchy background of dried salt. 

sharp shadows, the effective size of the source must be small compared 
with the source specimen distance; (2) To avoid contamination and 
specimen changes, evaporation time should be short, and especially in 
freeze etching should be only ~10 sec. The two most commonly used 
evaporation methods, namely resistance heating and electron bombard- 
ment, will be discussed below.* 

Resistance Heating 

The present standard method for shadowing is resistance heating. Prac- 
tically all metals with a melting point below 2000° K can be evaporated 
simply by passing a heavy current through a support wire, boat or rod 
made of tungsten, molybdenum, carbon, etc. Details are given in chapter 
4 and in standard texts on techniques for electron microscopy (Hall, 

'Vacuum arc evaporation of refractory metals has also been reported ( Nagakura 
et al., 1966 ) but the structure of the films obtained has not yet been studied at high 


1965; Reimer, 1967; Bradley, 1965). High melting (refractory) metals 
cannot practically be evaporated by resistance heating for lack of a 
suitable supporting material. Such a support must be a good electrical 
conductor, and its vapor pressure must be considerably lower than that 
of the material to be evaporated. Finally, it must be mechanically stable 
at the temperature of evaporation and not be destroyed either by melt- 
ing or by the formation of low-melting alloys. 

For the high-melting metals there is no material available which has 
the above-stated properties. Tungsten, which has the highest melting 
point, readily forms alloys with all the other metals mentioned, an affect 
which is a well-known nuisance when evaporating platinum. Carbon, 
which does not melt, cannot be used as a support because it has a vapor 
pressure comparable with, or higher than, the materials to be evaporated. 
Finally, evaporation of high-melting metals by passing the current di- 
rectly through the materials is not feasible because the vapor pressures 
below their melting points are not high enough for adequate sublimation 
(Nesmeyanov, 1963; Honig and Kramer, 1969). 

In spite of these considerations, several attempts at evaporation of 
high-melting metals by resistance heating have been reported (Hibi, 
1952; Baylev, 1962; Hart, 1961 and 1963). These attempts involved 
extremely long evaporation times (~10 min to 10 hr) with resultant 
contamination and heat damage problems. The evaporation time with 
resistance heating can be shortened only by extremely unsuitable shad- 
owing geometry (extended sources, short distances). Therefore, in high 
resolution shadowing, resistance heating seems to be applicable only to 
the simultaneous evaporation of platinum and carbon. 

Electron Beam Evaporation 

An effective method for the evaporation of high-melting metals is elec- 
tron beam heating, a method which has long been used for the produc- 
tion of very clean deposits in thin films technology. It was adapted for 
shadowing specimens for electron microscopy as early as ten years ago 
( Westmeyer and Lorenz, 1960; Bachmann et al., 1960; Bachmann and 
Hayek, 1962; Bachmann, 1962). 

The basic principle involves the directing of an electron beam onto 
the material to be evaporated. As the schematic drawing in Fig. 5.2 
shows, a rod of the evaporating material serves as the anode. It pro- 
trudes only a few mm from a cooled anode holder. When the beam is 
focused onto its tip, melting it to a drop, a steep temperature gradient 
arises in the shadowing material. This temperature gradient prevents 
the drop from melting back, while at the same time allowing its 


focusing aperture 

front aperture 

ion deflection plates 

Fig. 5.2 Schematic diagram of an electron-beam evaporator. The cathode-fila- 
ment is a loop made of tantalum or tungsten wire, or ribbon. It is heated with 
the low voltage power supply. The electrons leaving the hot filament are ac- 
celerated towards the tip of the anode and melt it to a drop. This drop is the 
actual shadowing source. The focusing aperture prevents the electrons from hit- 
ting the upper parts of the anode or the anode-holder. The opening in the front 
shield is so small that radiation from the filament cannot reach the specimen. 
For outgasing prior to the shadowing, the front shield and the focusing aperture 
can be set to a high potential and heated up by the electron beam. The de- 
flection plates in front of the evaporator prevent the bombardment of the specimen 
with charged particles. One of the plates is at anode potential, the other one at 
cathode potential. For convenience, we keep the cathode at ground potential. 
Evaporator can be seen in ch. 2. 

face to be heated above the melting temperature, and thus yielding high 
evaporation rates. 

Given this relatively simple mechanism and the potentially higher 
shadowing resolution when using refractory metals, why is electron beam 
evaporation still not a standard method for shadowing? The complaint 
from users of early commercial units was primarily specimen damage. 
However, it has recently been shown that this damage was at least par- 
tially caused by charged particles generated in electron beam evaporators 
(Abermann and Bachmann, 1969; Bachmann et al., 1969). This was 
demonstrated by the fact that these authors could eliminate damage on 
freeze-etched specimens by filtering out the charged particles with an 
electric field. Although this principle has since been used also in com- 
mercial units, we feel that other prerequisites for optimum shadowing 
are still often neglected. Before shadowing it should be possible to out- 
gas (by heating) all parts of the evaporator which will heat up during 
evaporation. This means that during the actual shadowing no part of the 
evaporator should get hotter than during the outgasing. Furthermore, the 
heat generated during outgasing, as well as during evaporation, must be 
efficiently dissipated. 


Comparing the two methods of evaporation, resistance heating versus 
electron bombardment, we conclude that electron beam evaporation is 
more versatile since it can also be applied to high-melting metals. Higher 
thermal load leading to specimen damage is not inherent in this method. 
It is, however, a more elaborate and expensive technique than resistance 


Platinum-Carbon (Pt-C) 

As discussed above, the aim of high resolution shadowing is to obtain 
deposits with no granularity and with high electron scattering. In an 
attempt to solve this problem, Bradley (1959) introduced simultaneous 
evaporation of platinum and carbon. The carbon, having a high melting 
point, is intended to prevent the migration and crystal formation of the 
platinum atoms, which in turn give the film the required density, un- 
obtainable by carbon alone. The Pt-C shadowing can be performed in 
various ways. The technique originally proposed by Bradley involved 
the evaporation of mixed platinum-carbon electrodes. Kranitz and Seal 
(1962) proposed the evaporation of Pt-C pellets between carbon elec- 
trodes. Such mixtures are now commercially available (Ladd Research 
Industries ) . 

In other widely used variants of the method, platinum wire is 
wrapped around the touching tips of two carbon electrodes ( Jayme and 
Hunger, 1958; Moor, 1959; see also Reimer, 1967, pp. 303 ff.; Bradley, 
1965; Koehler, in this volume, Ch. 2 ) . Finally, Pt-C can also be evaporated 
by electron bombardment (Westmeyer and Lorenz, 1960; Bachmann et 
at., 1960). Recently this has been applied to freeze etching, and higher 
reproducibility as compared to resistance heating has been claimed 
(Moor, 1970). 

Platinum-carbon shadowing certainly was a great improvement over 
the then existing techniques. However, general experience and systematic 
studies ( Reimer and Hermann, 1962 ) have since shown that it is difficult 
to obtain reproducible deposits with an optimum ratio of platinum to 
carbon. While layers with too high a platinum content have a granularity 
comparable to pure platinum deposits, those with too low a platinum 
content lack contrast unless a film thickness incompatible with high 
resolution is used. Such thick films do not show any granularity even 
at high magnifications. High resolution is thus frequently fallaciously 
assumed, even though the absence of granularity unless related to film 
thickness is obviously no proof for high resolution of a replica. 




(Nesmeyanov, 1963: Honig and Kramer, 1969) 



Melting point, 

Vapor pressure 

at melting point, 



at which vapor 

pressure is 

10 1 Torr' 









1.1 X 10" 2 





2.7 X 10" 1 





6X10" 3 





3.3 X 10" 2 




~4,000 (?) 
(at high 


"For high evaporation rates from small sources the vapor pressure should be in the 10- 1 Torr range. 

High Melting Metals 

The high melting, so-called refractory metals (melting point >2500°K), 
which have frequently been used for shadowing, are iridium, molyb- 
denum, tantalum and tungsten. Some of the physical properties relevant 
to their use as shadowing materials are listed in Table 5.1. 

One way to evaporate all these metals by electron bombardment is to 
use rods of these metals as the target ( anode ) . The metal then evaporates 
from the hanging drop which is formed at the anode tip. Iridium can be 
evaporated better and more cheaply by wrapping a thin wire of this 
metal around the tip of a tungsten rod. In the electron beam, a drop of 
iridium-tungsten alloy is formed from which mainly the iridium evapo- 
rates because of its much higher vapor pressure. In the same way almost 
pure films of all other metals with even higher vapor pressures (e.g., 
platinum ) can be formed. 

A tantalum-tungsten alloy which presently appears to be a very 
promising shadowing material can similarly be evaporated from a 
tantalum wrapped tungsten rod. However, in this case the deposit con- 
sists roughly of two-thirds tantalum and one-third tungsten, in accordance 
with their relative vapor pressures (Abermann and Bachmann, 1969). 
The evaporation source can be used several times without changing the 
anode. X-ray fluorescence analysis has shown that the tantalum content 
of the deposit decreases with subsequent evaporations. However, no 
adverse effect on film fine structure or chemical stability was observed 
when the same source was used for five evaporations (Zingsheim et al., 


As already stated, the graininess of the deposits generally decreases 
with increasing melting temperature. Thus, tungsten has the finest grain. 
Unfortunately, thin deposits of this material are rather unstable chemi- 
cally, and therefore can be used only to shadow preparations which need 
no subsequent wet steps (e.g., particle shadowing or post-shadowing of 
replicas ) . Iridium, on the other hand, is chemically most stable. It should, 
therefore, be used when extreme chemical procedures for cleaning the 
replica ( such as cleaning with hydrofluoric acid ) are needed ( Fig. 5.3 ) . 
Although it has the coarsest grain of the high melting metals, it is still 
considerably finer than that of platinum. The tantalum-tungsten alloy 
combines a very fine grain with a fairly high chemical stability, e.g., 
replicas can float on 70% sulfuric acid for days without adverse effects 
on the shadowing film.* 

It is very likely that other as yet untried alloys may show even more 
favorable properties. An interesting extension of Bradley's basic principle 
would be the simultaneous evaporation of a high-melting metal and 
carbon (or some other "mobility inhibitor") from two independent 

In conclusion, two basic materials for high resolution shadowing 
are at present available: platinum-carbon and the high-melting metals. 
The platinum-carbon has the primary advantage of simple application 
by resistance heating. It is also chemically very stable and therefore has 
become the standard replicating material for freeze etching. However, 
the density of the replicating film is, even at its best, lower than that of 
a film consisting of pure metal. The required films are therefore always 
thicker. The greatest disadvantage is the lack of its reproducibility, so 
that there is always a risk of having either too high or too low a carbon 
content. Beam damage in the electron microscope at high intensity is also 
occasionally reported. 

High melting metal films have the advantage of high density, good 
reproducibility, and almost unlimited stability in the beam of the elec- 
tron microscope. The disadvantages in their use are the elaborate evapo- 
ration equipment necessary and the low chemical stability in some in- 
stances. Figures 5.4 to 5.6 illustrate the applicability of shadowing with 
high-melting metals mainly on biological specimens. Corresponding 
samples of platinum-carbon were omitted, since practically any high 
magnification picture from the freeze etching literature can be used for 

"Chemical stability of the shadowing material by no means guarantees the absence 
of morphological changes with time (Jacobs and Pashley, 1962; Bachmann and 
Hilbrand 1966; Jaeger et. al., 1969). It is therefore recommended that shadowed 
specimens not be stored for extended periods before being investigated in the EM., 
unless their morphological stability in air has been checked. 


Fig. 5.3 Replica of moon dust, shadowed with iridium at room temperature and 
backed with carbon. The particle was dissolved in 30% hydrofluoric acid. Note 
the sharpness of the long shadows on the substrate due to the small ratio of 
source size (3 mm^), to source-specimen distance (200 mm). The insert shows 
the granularity of iridium at higher magnification. 

this purpose. In addition, we wanted to avoid the danger of unfairly 
downgrading an established method by illustrating its limitations with 
pictures which might not be of the optimum quality attainable. 


In order to take advantage of improvements in shadowing materials, 
greater care must be expended on specimen preparation to avoid arti- 
facts due to extraneous factors. Some such factors are discussed below. 

Mounting of Specimens 

Shadowing, unlike negative staining, highlights the surface of the speci- 
men. Therefore, all the nonvolatile components of the solution, i.e., 
buffers as well as any contaminants, are visualized together with the 


Fig. 5.4 Alanine transfer RNA from yeast shadowed with tantalum-tungsten at 
an angle of 40". A: monomeric fraction (molecular weight, 25,000); B: hydrogen 
bonded dimers; C: buffer blank. (For details, see Yoshikami and Abermann, 1971.) 
To insure minimum contamination, the specimen in the vacuum system was kept 
at room temperature while surrounded by liquid nitrogen cooled walls. The finer 
grain of the shadowing material allows one to use a steeper than usual shadow- 
ing angle (40°). This is advantageous since the distortion in particle size and 
shape due to pile up of shadowing material is minimized. 

dried specimen particles (Fig. 5.1). It is thus essential to wash off all the 
salts before allowing the specimen to dry. 

A much used application of shadowing has been in the study of 
DNA. However, the extreme length of these molecules and the increased 
diameter due to a cytochrome c envelope ( Kleinschmidt, 1971) allow 
them to be visualized, even against a very coarse background. As the 
particles under study approach a spherical shape and their size ap- 
proaches the resolution limit of shadowing, background and dirt prob- 
lems may prevent their identification, let alone speculations regarding 
their fine structure. 


Fig. 5.5 Freeze-etched yeast cell shadowed with Ta/W at -100°C to demonstrate 
absence of heat damage. At this low magnification Ta/W replicas have the same 
appearance as Pt/C specimen. (From Bachmann ef a/., 1969.) 

Specimen Contamination within the Evaporator 

A point of increasing concern has been the elimination of artifacts pro- 
duced by contamination of the specimen in the vacuum system ( Bradley, 
1960; Bayley, 1962; Deamer et al, 1970; Sella et al., 1970). To under- 
stand vacuum contamination, it might be helpful to look at what happens 
during pump down. First, the system is pre-pumped by a mechanical 
pump, and the trouble can begin at this stage. As many vacuum manuals 
emphasize, to avoid backstreaming of forepump oil, one should not pre- 
pump a system below approximately 0.1 Torr (100 (jl). 

When the system is opened to the diffusion pump, the pressure 
should drop rapidly to the 10- 5 Torr range. At this pressure, the main 
constituent in the residual gas (provided we are dealing with a fairly 
leak-free system) is water vapor desorbed from the walls of the vessel. 
Subsequent pumping results mainly in a reduction of this water pressure. 

At the same time, due to the nature of oil diffusion pumps, there 
exists an almost constant pressure of oil vapor in the 10~ T Torr range. 
Since this is only about 1-10% of the total pressure, it cannot be moni- 
tored by the vacuum gauge. However, a pressure of lO" 7 means that 


(PllkS'dLi&r i ' 



Fig. 5.6 Freeze-etched catalase crystal in aqueous suspension, deep-etched and 
Ta/W shadowed at -100°C. Insert: Pt/C replica of same specimen. The peri- 
odicity of the crystal is well revealed in the Pt/C replica, partially due to decora- 
tion. Steps in the surface and single molecules are much better depicted in the 
Ta/W specimen. For comparison, high magnification pictures of Pt/C and Pt- 
shadowed, dried catalase are given in Hall (1950) and Labaw (1967). (From 
Bachmann ef a/., 1969.) 

every surface element within the vacuum is hit by oil molecules with a 
high enough frequency to potentially form a monolayer every 10 sec. 
Prolonged pumping therefore results in increased oil contamination of 
the surface. This can be prevented by the use of cold traps or baffles, 
usually cooled with liquid nitrogen. There can also be no oil-backstream- 
ing when dry-type pumps ( sublimation, ion, cryo, or turbo pumps, etc. ) 
are used instead of oil diffusion pumps. 

Unfortunately, backstreaming is not the only source of contamination; 
other sources are oil, grease or dirt in the system (e.g., from gaskets, 
finger prints or previous maltreatments with oil) which can saturate a 
vacuum system for a long time. Unless all these sources of contamination 
are eliminated, merely changing to more expensive pumps will not 


solve the problem although the meters might indicate a better vacuum. 
For a recent discussion of pumps and contamination relevant to the 
problems of this chapter, see Santeler (1971). 

A perfect way to avoid contamination would be to surround the 
specimen completely with surfaces which are colder than the specimen 
itself, because contaminants stick preferentially to the coldest surfaces, 
serving as the basis for cryopumping. This principle has rarely been used 
for shadowing except in freeze-etching, where it is approximated by 
placing the cold knife over the newly cleaved specimen surface. A freeze- 
fracture device where specimen shielding by cold surfaces is even more 
elaborate has been described by Bullivant et al., (1968) and Bullivant 

When the specimen temperature drops to — 100° C or lower in a con- 
ventional vacuum system (e.g., in freeze-etching), even the residual 
water-vapor will become a source of contamination by ice condensation 
on the specimen surface (vapor pressure of ice is 1 X 10~ 5 Torr at 
— 100°C). It is this maximized contamination of the coldest surface in the 
system which prevents the uncritical use of lowering substrate tempera- 
tures as a way of decreasing surface mobility, and thus the granularity 
of shadowed films. 

Another major source of contamination is dirt from various parts of 
the evaporator. We have already emphasized the need for outgasing, as 
well as the application of cold apertures. Finally, we would like to warn 
against the use of brass parts often found in homemade evaporator elec- 
trodes. Small parts (screws, pins, washers) which are not in good con- 
tact with the cool bulk easily become overheated and act as a source 
of zinc vapor. 

Thermal Load 

As already mentioned, most investigators who have tried shadowing 
biological materials with high-melting metals were discouraged by ad- 
verse effects on the specimens, which were interpreted as heat damage. 
Thus developed a widespread belief that heat damage is more severe 
with electron bombardment than with conventional shadowing. However, 
theoretical and experimental studies indicate that this is not necessarily 
so. During vacuum evaporation, the energy flow to the specimen is 
caused: (1) by the difference in the heat content of the vaporized 
shadowing material and its condensed film ("heat of condensation"); 
( 2 ) by the thermal radiation of the source ( heat of radiation ) ; and ( 3 ) 
by a flow of charged particles when electron beam evaporates are used 


(ion or electron bombardment) (Bachmann et al., 1969, and Zingsheim 
et al, 1970a and b). 

Heat of Condensation 

The heat of condensation depends primarily on the shadowing material. 
However, it is about the same for equal thickness tantalum-tungsten and 
platinum-carbon layers and amounts for all practical cases to less than 
5% of the total load. 

Heat of Radiation 

As equation (5.1) shows, the intensity (7) of the radiation reaching the 
specimen depends on the source temperature ( T ) and on the ratio of 
source radius (r) to source-specimen distance (R) 


2 Eo-r* (5.1) 

(E — total emissivity; cr = Stefan-Boltzmann constant.) Thus the heat 
of radiation can be effectively decreased by decreasing the ratio r to R, 
either by using a smaller source or by moving the specimen further away. 
Of course, doing this would also decrease the deposition rate of the 
shadowing material which is effected to the same extent as the radiation 
intensity. One can, however, compensate for this by raising the source 
temperature, thus increasing the vapor pressure and the evaporation 
rate.* Although the radiation intensity is then also increased by the 
fourth power of the temperature, fortunately, in the temperature range 
of interest, the increase in vapor pressure greatly exceeds the increase in 
radiation intensity. For instance, if the drop diameter is changed from 
3 mm to 2 mm or if the source is placed 30 cm away from the specimen 
instead of 20, then, both the heat of radiation and the deposition rate 
would decrease by 55%. Vapor pressure data show that by raising the 
temperature of tantalum-tungsten by 100°C, one can restore the original 
deposition rate, while raising the radiation energy by only 10%. Thus, 
by using a more favorable shadowing geometry, the final heat load due 
to radiation has been cut to one half without prolonging the shadowing 

An approach to the total elimination of radiation energy is the appli- 
cation of rotating apertures (Horn, 1962). We feel, however, that in- 
creasing the specimen source distance and decreasing the source size is 
at present simpler, and for most practical purposes, sufficient. 

"Vapor pressure data for metals are conveniently presented in a recent paper by 
Honig and Kramer (1969). 


Ion or Electron Bombardment 

Electron beam evaporators, especially when badly designed, cause ion 
or electron bombardment of the specimen. The energy released at the 
specimen can be several times higher than the heat of condensation, but 
in most cases will still be lower than the heat of radiation. However, the 
extent of specimen damage is determined not only by the amount of 
energy reaching the specimen but also by where this energy is released. 
Since the ions and electrons are completely absorbed within a very thin 
surface layer of the specimen, they can be relatively more harmful than 
the thermal radiation which, although carrying more energy, has a much 
deeper penetration. As already mentioned, this flux of charged particles 
was most likely the cause of the early discouraging results with electron 
beam evaporation in freeze-etching. Fortunately, it is easy to deflect 
these particles and thus completely eliminate this source of specimen 

In summary, contrary to common belief, the thermal load connected 
with shadowing depends less on the shadowing material than on the 
evaporation conditions. It has been experimentally verified that the heat 
flux during platinum-carbon shadowing with one of the commercial 
evaporators is higher than that of tantalum-tungsten shadowing with 
a properly designed electron beam evaporation gun. Even platinum 
shadowing using an electron gun was shown to result in a lower heat 
flux than platinum-palladium evaporation by resistance heating (Bach- 
mann et al., 1969; Zingsheim et al, 1970b). 

Thus, the main objection to shadowing with high-melting metals is no 
longer valid. Calculations and experiments have further shown that for 
minimum thermal load the evaporator should be operated at the highest 
possible temperature, allowing a small ratio of source size to specimen 
distance. In this respect, resistance heating is much more restricted by 
the need of a supporting material than electron bombardment. Of course, 
all calculations on the thermal load are based on the assumption that the 
only hot surface that the specimen sees is the actual vapor source. To 
fulfill these conditions, it is necessary to protect the specimen by placing 
a cooled aperture in front of the specimen. As already stated, the same 
aperture would also help prevent specimen contamination. However, in 
many evaporators no such shield is provided. 


Discussions of shadowing methods will ultimately lead to the question, 
what is their resolution? It seems, at least to us, that there is as yet no 


straightforward answer to this question. The difficulties involved are 
explained below. 

Considerations of shadowing resolution are frequently obscured by 
the phenomenon known as decoration. We have already seen that the 
atoms of the shadowing material migrate on the specimen surface until 
they become stabilized. Local variation in the physico-chemical prop- 
erties of a surface can influence the nucleation or stabilizing probability. 
This decoration effect was first used by Bassett (1958a and b) and Sella 
et ah, (1958) for the study of crystal surfaces. When a cleavage surface 
of sodium chloride was evaporated at right angles with a very thin 
"film" of gold (mean thickness 3-10 A), there was a characteristic 
deposition pattern. Gold crystals were linearly arranged like beads on a 
string in some areas and statistically distributed in others (Fig. 5.7). 
Bassett could show that the lines depicted steps in the crystal surfaces; 
some of which were only one half unit cell in height, i.e., 2.8 A. Since a 
depth resolution of atomic dimension can already be obtained by the use 
of a "low resolution material" such as gold, this can hardly be improved 
by using higher-melting metals. However, since the latter materials form 
deposits with smaller crystallites, which are closer together, the lateral 

Fig. 5.7 Cleavage steps on cleavage surface of sodium chloride, decorated 
with gold. (After Bassett, 1958a.) 


resolution of decoration replicas can be improved. For instance, the 
distance between two steps which are depicted as two separate lines can 
be much smaller in the case of platinum than in that of gold. 

Although decoration is most obvious on replicas of crystal surfaces, 
this phenomenon is by no means confined to such specimens. As a matter 
of fact, it is very difficult, if not impossible, to obtain replicas which are 
completely free from decoration. Thin film formation always begins with 
nucleation, and wherever nucleation occurs there is always a possibility 
of preferred nucleation since the object of investigation will hardly ever 
be a homogeneous, uniform, flat surface. Decoration on biological speci- 
mens has often been observed but has not yet been systematically in- 
vestigated. It is especially pronounced in freeze-etched replicas of re- 
peating structures such as virus-crystals, catalase crystals, or broken-up 
myelin (Fig. 5.6). 

We have seen that the "resolution" obtainable by gold-decoration, for 
instance, on sodium chloride is considerably better than that expected by 
shadowing. Why, then, is it necessary to distinguish between the two 
phenomena when talking about resolution? Since the information re- 
vealed by the two methods is radically different, failure to distinguish 
between them might result in fallacious and misleading interpretations. 
From shadowing one obtains information on geometry or relief (relative 
heights or distances of elevated regions). Decoration reveals local dif- 
ferences in binding strengths between the evaporated material and the 
surface. An additional complication arises from the fact that decoration 
might be enhanced by true shadowing due to a pile up on the original 
decorating particles. Shadow lengths could then give highly erroneous 

In order to distinguish shadowing from decoration, it might be neces- 
sary to determine the extent of decoration on a certain specimen. This 
can be done by evaporating a specimen at right angles, thereby prac- 
tically avoiding all shadows (decoration is visualized optimally when 
the evaporated film thickness is ~5A, i.e., only Yz to % of that used in 
shadowing ) . 

In shadowing one looks for actual shadows, their lengths and how the 
structure depends on the shadowing direction. If the shadowing angle is 
steeper than the slope of a structure, shadowing also causes variation in 
film thickness and not just a presence or absence of grains. This gives an 
overall soft appearance to the replica. Pure decoration, on the other 
hand, gives a much greater all-or-none effect and absence of the charac- 
teristic shadow. 

An approach to test true shadowing resolution is the one suggested 
by Reimer and Schulte ( 1966 ) . The test specimen is made of a uniform 


and noncrystalline material (germanium, carbon), in which decoration 
effects seem to be small. When this specimen, which consists of two 
diverging steps of controllable height, is shadowed, one can determine 
the minimal step height (step resolution) as well as the minimal distance 
between steps (lateral resolution) which can be detected (Fig. 5.8). 
With such a specimen, the following resolution values were obtained: 
Reimer and Schulte (1966) found for platinum-carbon a step resolution 
of 24 A, and a lateral resolution of 90-120 A. Abermann and Bachmann 
(1969) similarly found platinum-carbon step and lateral resolutions of 
20 A and 90 A respectively; and for tantalum-tungsten these values 
were 10 A and 40 A respectively. 

Since the test specimen is a product of shadowing and, therefore, does 
not have the idealized geometry of Fig. 5.8, it itself could already be 
limiting. With a more suitable specimen, higher resolutions may be de- 
monstrable. The lateral and step resolution of Reimer's specimen both refer 
to extended linear structures. These values are proportional to, but not 

Fig. 5.8 Schematic diagram of resolution specimen adapted from Reimer and 
Schulte (1966). A cube of magnesium oxide, set on mica, is shadowed from 
two slightly different directions with carbon forming two partially overlapping 
layers. The thicknesses of the two carbon films was measured by a quartz 
crystal oscillator. The shadowing material whose resolution was to be tested 
was then evaporated at right angles to the shadowing direction of one of the 
carbon layers. To determine the step resolution (i.e., the minimum layer thickness 
resolvable) several specimens were made varying the film thickness of carbon. 
The lateral resolution was determined by the minimum distance at which the 
two sides of the wedge shaped steps can be seen as two. Being a product of 
shadowing, the edges of this specimen will not be sharp on an atomic scale. 
For testing very high resolution such a specimen can itself become the limiting 


identical with, the resolution for other structures. For instance, for small 
spherical particles the resolution is expected to be poorer. Nevertheless, 
the actually measured values are less impressive than resolution values 
frequently claimed without adequate definition. For a meaningful assess- 
ment of different shadowing techniques, it is essential to compare their 
capabilities using identical test specimens. 


Abermann, R., and Bachmann, L. (1969). Elektronenmikroskopische Beschat- 
tung mit hoher Auflosung. Naturwissenschaften, 56, 324. 

Abermann, R., and Yoshikami, D. (1972). High resolution shadowing of trans- 
fer RNA. P.N.A.S. to be published. 

Bachmann, L. (1962). Shadow-casting using very high-melting metals. Proc. 
5th. Inter. Cong. Electron Micros., 1, FF3. 

Bachmann, L., Abermann, R., and Zingsheim, H. P. (1969). Hochauflosende 
Gefrieratzung. Histochemie, 20, 133. 

Bachmann, L., and Hayek, K. (1962). Beschattung elektronenmikroskopischer 
Praparate mit hochstschmelzenden Metallen. Naturwissenschaften, 49, 153. 

Bachmann, L., and Hilbrand, H. (1966). Sinterung von Aufdampf schichten 
aus Silber und Gold unter dem Einfluss von Adsorptions schichten. In: 
Basic Problems in Thin Film Physics. Proc. Intern. Symp., Clausthal— 
Gbttingen. (Niedermayer R. and Mayer H., eds.) p. 77. Vandenhoeck and 
Ruprecht, Gbttingen. 

Bachmann, L., Orr, W. H., Rhodin, T. N., and Siegel, B. M. (1960). Deter- 
mination of surface structure using ultra-high vacuum replication. J. Appl. 
Phys., 31, 1458. 

Bassett, G. A. (1958a). A new technique for decoration of cleavage and slip 
steps on ionic crystal surfaces. Phil. Mag. Series, 8, 1042. 

. (1958b). Nucleation of evaporated metal layers on single crystal sub- 
strates. Proc. 4th. Inter. Cong. Electron Micros., 1, 512. 

Bassett, G. A., Menter, J. W., and Pashley, D. W. (1959). The Nucleation, 
Growth, and Microstructure of Thin Films. In: Structure and Properties of 
Thin Films. (Neugebauer, C. A., Newkirk, J. B., and Vermileyea, D. A., 
eds.) John Wiley & Sons, Inc., New York. 

Bayley, S. T. (1962). A study of shadowing for high resolution electron 
microscopy. Proc. 5th. Inter. Cong. Electron Micros. 1, FF2. 

Bradley, D. E. (1958a). Simultaneous evaporation of Pt and carbon for pos- 
sible use in high resolution shadow-casting for the electron microscope. 
Nature, 181, 875. 

. (1958b). A new approach to the problem of high resolution shadow- 
casting: the simultaneous evaporation of Pt and carbon. Proc. 4th. Inter. 
Cong. Electron Micros., 1, 428. 

. (1959). High-resolution shadow-casting technique for the electron 


microscope using the simultaneous evaporation of Pt and carbon, Brit. J. 
Appl. Phys., 10, 198. 

. (1960). Study of background structure in platinum/ carbon shadowing 

deposits. Brit. J. Appl. Phys., 11, 506. 

(1965). Replica and Shadowing Techniques. In: Techniques for Elec- 

tron Microscopy. (Kay, D. H., ed.) p. 96. Davis Company, Philadelphia. 

Bullivant, S. (1970). Present Status of Freezing Techniques. In: Some Biolog- 
ical Techniques in Electron Microscopy. (Parson, D. F., ed. ) p .127. Aca- 
demic Press, New York. 

Bullivant, S., Weinstein, R. S., and Somenda, K. (1968). The Type II simple 
freeze cleave device. /. Cell Biol., 39, 19a. 

Deamer, D. W., Leonard, R., Tardieu, A., and Branton, D. (1970). Lamellar 
and hexagonal lipid phases visualized by freeze-etching. Biochim. Biophys. 
Acta, 219, 47. 

Hall, C. E. (1950). Electron microscopy of crystaline catalase. J. Biol. Chem., 
185, 749. 

. (1966). Introduction to Electron Microscopy. McGraw-Hill Book Co., 

Inc., New York. 

Hart, R. G., (1961). Surface structure of tobacco mosaic virus. /. Mol. Biol., 
3, 701. 

. (1963). A method for shadowing electron microscope specimens with 

tungsten. /. Appl. Phys., 34, 434. 

Hibi, T. (1952). On the metallic shadow-casting using a nozzle system. /. 
Appl. Phys., 23, 957. 

Honig, R. E., and Kramer, D. A. (1969). Vapor pressure data for the solid 
and liquid elements. RCA Rev., 30, 285. 

Horn, H. R. F. (1962). Thermische Bedampfung ohne Licht-und Warmestrah- 
lungsbelastung des Objektes. Proc. 5th. Inter. Cong. Electron Micros., 1, A9. 

Jacobs, M. H., and Pashley, D. W. (1962). The changes which occur in the 
structure of a thin surface deposit when it is detached from its substrate. 
Proc. 5th. Inter. Cong. Electron Micros. 1, DD4. 

Jaeger, H., Mercer, P. D., and Sherwood, R. C, (1969). The effect of ex- 
posure to air on silver and gold films deposited in ultra-high vacuum. 
Surface Science 2, 349. 

Jayme, G., and Hunger, G. (1958). Eine elektr. opt. Studie iiber das Verhalten 
der Zellulose-Mikrofibrillen bei der Trocknung. Mikroskopie, 13, 24. 

Kleinschmidt, A. K., (1971). Electron microscopic studies of macro-molecules 
without appositional contrast. Phil. Trans. Roy. Soc. Lond. B., 261, 143. 

Knoch, M., and Konig, H. (1956). Strukturlose Platinabdrucke biologischer 
Objekte. Z. wiss. Mikr., 63, 121. 

Kranitz, M., and Seal, M. (1962). A modified Pt-carbon replica technique. 
Proc. 5th. Inter. Cong. Electron Micros., 1, FF7. 

Labaw, L. W. (1967). Ox liver catalase crystal structure by electron micros- 
copy. J. Ultrastruct. Res. 17, 327. 

Mayer, H. (1955). Physik Dunner Schichten. Teil II, Kap. I: Elementar- 


prozesse beim Schichtaufbau, pp. 1-110. Wissenschaftliche Verlagsgesell- 

schaft, Stuttgart. 
Moor, H. (1959). Platin-Kohle-Abdruck-Technik angewandt auf den Feinbau 

der Milchrohren. /. Ultrastruct. Res., 2, 393. 
. (1970). High resolution shadow casting by the use of an electron gun. 

Proc. 7th. Inter. Cong. Electron Micros., 1, 413. 
Nagakura, S., Kikuchi, M., Aihara, K., and Oketani, S. (1966). Vacuum-arc 

evaporation of highly refractory metals and its applications. Proc. 6th. Inter. 

Cong. Electron Micros., 1, 267. 
Nesmeyanov, A. N. ( 1963 ) . Vapor Pressure of the Chemical Elements.. Elsevier 

Publishing Company, New York. 
Reimer, L. (1967). Elektronenmikroskopische Untersuchungs-und Prapara- 

tionsmethoden. 2. Aufl. Springer-Verlag Berlin, Heidelberg, New York. 
Reimer, L., and Hermann, W. (1962). Dichte-und Teilchengrossenbestim- 

mungen an Platin-Kohle-Mischschichten fur die elektronenmikroskopische 

Schragbeschattung. Naturioissenschaften, 49, 296. 
Reimer, L., and Schulte, C. (1966). Elektronenmikroskopische Oberflachenab- 

drucke und ihr Auflosungsvermogen. 'Naturioissenschaften, 53, 489. 
Santeler, D. J. (1971). The use of diffusion pumps for obtaining ultraclean 

vacuum environments. /. Vacuum Sci. Technol., 8, 299. 
Sella, C, Conjeaud, P., and Trillat, J. J. (1958). Nouvelle Methode d'Etude 

par microscopie electronique de la structure superficielle des faces de 

clivages d'halogenures alcalins. Proc. 4th. Inter. Cong. Electron Micros., 1, 

Sella, C, Martin, J. C, Escaig, J., Desforges, M. (1970). Ombrage a Haute 

Resolution en Ultra Vide. Proc. 7th. Inter. Cong. Electron Micros. 1, 303. 
Westmeyer, H., and Lorenz, E. (1960). Die Herstellung von Aufdampfschich- 

ten vermittels kleiner elektronenstossgeheizter Dampfquellen. Optik., 17, 244. 
Yoshikami, D., and Abermann, R. (1971). Qualitative and quantitative analy- 
sis of tRNA with the electron microscope. Proc. 29th Meet. Electron Micros. 

Soc. Amer., p. 434. 
Zingsheim, H. P., Abermann, R., and Bachmann, L. (1970a). An ultrashadow- 

ing-unit for the freeze-etching technique. ]. Physics E: Sci. Instru., 31, 39. 
. (1970b). Shadow casting and heat damage. Proc. 7th. Inter. Cong. 

Electron Micros., 1, 411. 


Supported by U. S. Public Health Service Grants GM 10422 from the 
Institute of General Medical Sciences and NS 09315 from the Institute of 
Neurological Diseases and Stroke; and by a grant from the Deutsche For- 

6 Autoradiography 

M. M. Salpeter 

School of Applied and Engineering Physics 
and Section of Neurobiology and Behavior, 
Cornell University, Ithaca, New York 

L. Bachmann 

Institut fur Technische Chemie, 
Technische Universitat, Munich, Germany 


l""lectron microscope (EM) autoradiography is rapidly coming into its 
own as a cytochemical tool. It is employed to localize and quantify 
radioactive material at submicroscopic levels. The higher autoradio- 
graphic resolution, although still far from that of the electron microscope 
itself, provides a more accurate localization of intracellular radioactivity 
than is possible with the light microscope technique. Nevertheless, very 
often the depth of focus and the high resolution of the electron micro- 
scope are themselves of even greater importance. The developed grains 
in the emulsion are distinctly seen and are in focus at the same time as 
the biological tissue. Grain counting for quantitative work is thereby 
made easy. The cytological details which are resolved in the electron 
microscope facilitate the identification of tissue structures which can 
then be correlated with developed grain density to establish the most 
likely source of radioactivity. 

Both electron microscope and light microscope autoradiography are 
based on the same physical principles since both use a thin layer of 
photographic emulsion closely applied to the specimen as a radiation 
detector. However, in EM autoradiography, the larger magnification and 
higher resolution, the scaling down of the section and emulsion thick- 
nesses, and the need for extra fine grained emulsions to obtain optimum 
autoradiographic resolution, accentuate some of the difficulties familiar 
to the light microscope autoradiographer while introducing new ones 
unique to itself. 

The basic technique involves the preparation and mounting of a 
specimen section on a support, coating it with emulsion, storing it for 
exposure, developing it, and analyzing the resultant autoradiograms. 
Numerous reviews are now available which cover all these steps in 
detail (Caro, 1964 and 1966; Salpeter and Bachmann, 1965; Salpeter, 
1966; Granboulan, 1963 and 1965; Rogers, 1967 and 1971; Neumann, 
1969; Bachmann and Salpeter, 1972; Budd, 1971; Jacob, 1971); see also 
"Technical Considerations," pp. 255. 

One technique frequently used, either as initially described or in 
some modified form, is that of Caro and Van Tubergen ( 1962 ) ( see also 
Moses, 1964; Granboulan, 1963) and Caro (1969). According to this 
technique, the section is placed on a collodion-coated specimen grid and 
then coated with emulsion formed in a wire loop. An alternative pro- 
cedure is to dip the grid into liquid emulsion (Hay and Revel, 1963). 



The main problem with any procedure in which a liquid emulsion is 
applied directly over a tissue section already placed on a specimen grid 
is that the grid, even when coated with collodion, provides a nonuniform 
substrate. One way to overcome the absence of a flat surface is to use a 
stripping film as recently suggested by Williamson and van den Bosch 
(1971). The need for a flat substrate was emphasized by Pelc et al., 
(1961) and Budd and Pelc (1964); they proposed a special "membrane 
technique" to attain it. This is similarly accomplished by the "flat sub- 
strate" procedure used first by Liquier-Melward (1965) and in modified 
form by the authors (Salpeter and Bachmann, 1964). A detailed step-by- 
step description of this method has been reported ( Salpeter, 1966; Bach- 
mann and Salpeter, 1972). 

Since this procedure will serve as the basis for the discussion to 
follow, it is briefly outlined below. A specimen section, placed on a 
collodion-coated microscope slide, is stained and then carbon and emul- 
sion coated. After the appropriate exposure, the specimen is developed 
and stripped onto a water surface. Grids are placed over the sections and 
the entire specimen sandwich picked up. 

Because of the availability of a vast literature concerning auto- 
radiographic technique, in this chapter an attempt has been made to 
discuss only those aspects which seem to be particularly troublesome or 
are not adequately covered by others. 


It is usually accepted that quantitative analyses of autoradiograms are 
necessary to answer the question: "How much radioactivity is present in 
the specimen?" Yet to answer "how much" is rarely meaningful without 
simultaneously answering "where," which itself involves even more 
sophisticated quantitation. For a complete answer to both these ques- 
tions, one needs to know both the efficiency (sensitivity) of the tech- 
nique (i.e., how many developed grains are produced by one radioactive 
decay) and the resolution (i.e., the expected grain distribution around a 
point source within a specimen). Once such a distribution is known for 
a point source, one can by simple integration obtain it also for extended 
sources, since any extended source is merely a collection of point sources. 
Finally, one needs enough developed grains for an adequate statistical 

The main problems in making autoradiography a quantitative and 
reproducible method arise from the large number of experimental factors 
which influence the efficiency and the grain distribution. The photo- 
graphic process is used near its limit of resolution and sensitivity, and 
the fine-grained emulsions and developing techniques are more sensitive 


to small variations in experimental conditions. As currently used, section 
and emulsion layers are thinner than the range of the radiation. Variation 
in the thickness of the specimen thus produces variations in the quantita- 
tive results. Therefore, measurements of section and emulsion thickness 
are of primary importance and will be discussed in some detail below. 
However, it is first appropriate to look at the general problems of resolu- 
tion and sensitivity. 


Genera/ Considerations 

Numerous reliable theoretical and experimental treatments of the prob- 
lem of resolution in autoradiography have been published (Caro, 1962; 
Pelc, 1963; Granboulan, 1963; Caro and Schnos, 1965; Rogers, 1967). The 
authors' views on the subject have been developed in detail in a series 
of publications ( Bachmann and Salpeter, 1965; Salpeter, 1966; Bachmann 
et al, 1968; Salpeter et al, 1969; Salpeter and Salpeter, 1971). None of 
these treatments will be repeated here in detail since we assume a gen- 
eral acquaintance with the problem. 

It is generally accepted that autoradiographic resolution is limited by 
the photographic process and by the source-specimen geometry. When 
radiation interacts with a silver halide crystal it can form a latent image 
(speck of reduced silver) (for a discussion, see Hamilton and Urbach, 
1966). One has to assume that the latent image can form anywhere 
within a penetrated silver halide crystal, independent of the site of the 
interaction, and that there can be more than one latent image per 
crystal. The greatest possible distance between a latent image and the 
site where the radiation hits the silver halide crystal is, therefore, equal 
to the diameter of the crystal. This distance could be greater if there 
happens to be energy transfer between crystals. 

Even in electron microscope autoradiography, a latent image must 
be developed in order for it to become visible. Development means that 
silver is deposited around the latent image. The silver can grow around 
the latent image in the form of irregular filaments, as clusters of silver 
specks, or as round particles similar in size to the silver halide crystal. In 
most cases, however, one has no way of knowing the point of contact 
between the latent image and final developed grain. The largest possible 
distance of a latent image to the center of its developed grain is therefore 
half the diameter of the developed grain. Thus, resolution is theoretically 
improved by decreasing the size of both the silver halide crystal of the 
emulsion and the developed grain. 


The second cause of limited resolution is the spread of radiation from 
the source to the site of latent image formation, or the "source detector 
geometry." Electrons are emitted from a source in all directions and thus 
very few travel in a direction perpendicular to the emulsion layer. An 
electron emitted at an angle will travel in the tissue until it reaches the 
emulsion. This distance travelled will be like the hypotenuse in the 
triangle depicted in Fig. 6. IB. If nothing would deflect or stop it, the 
length travelled would depend upon the distance between the source 
and the emulsion, which is statistically related to the thickness of the 
section. Once in the emulsion, if again not deflected or stopped, the elec- 
tron would continue to travel until it reached the outer surface. Latent 
images could be formed anywhere along its path in the emulsion. Since 
the majority of the electrons is emitted at an angle, an increase in 
emulsion and section thickness would invariably mean an increase in the 
horizontal distance from the source at which a latent image could be 

One method of decreasing the extent of radiation spread would 
therefore be to decrease the thicknesses of both the section and the 
emulsion. Another method of preventing a large radiation spread from 
the source would be to shorten the range of the electron. A latent image 
cannot be formed at a distance from the source beyond the range of the 




Fig. 6.1 Schematic diagrams for possible paths of electrons from a radioactive 
source in this biological material. A represents no scattering in the emulsion 
(high energy electrons), and B represents possible paths due to scattering (low 
energy electrons). In both cases, the electron leaves the source at a grazing 
angle (0) to the emulsion surface. For the unscattered (case A), the path length 
of the electron in the emulsion is extremely long compared with the thickness 
of the emulsion. This results in a high probability of forming developed grains 
at considerable distances from the source. For case B, scattering upward both 
decreases the path length and tends to return the electron towards the source. 
Scattering downward could lead to an increased path length, but (due to the 
initial small angle <p) often leads to the loss of the electron out of the emulsion. 
The net effect of scattering is thus to decrease the relative number of developed 
grains with distance from the source. 





















With energy (KV) With range* (fi) 

Isotope % Electrons larger than larger than 

H 1 


•Range in emulsion whose ratio of silver halide to gelatin is that of a close packed llford L4 
layer is approximately Va that in methacrylate (mean density ,^,3). 

electron.* Range is affected by the energy of the electron and by the 
density of the material through which it travels. The lower the energy, 
the shorter will be the range; and, for the same energy, the higher the 
density, the shorter the range (Table 6.1). For instance, the range in 
emulsion is roughly one quarter of that in methacrylate. 

Under electron microscope conditions the difference between the 
resolution of low and high energy electrons (i.e., tritium and C 14 ) stems 
primarily from the scattering conditions in the emulsion. The section 
thicknesses used are such that for both energy electrons there is essen- 
tially no scattering in the tissue, where the electrons can be considered to 
follow a straight line path. In the emulsion however the scattering of 
tritium electrons is considerable while that of C 14 is much less (Fig. 6.1). 
Thus, with the low energy electrons most of the radiation spread occurs 
within the section and not so much in the emulsion. Section thickness is 
thus the greater limiting geometric factor in electron microscope auto- 
radiographic resolution. Although this fact has been frequently em- 
phasized ( Caro, 1962; Pelc, 1963; Bachmann and Salpeter, 1965; Salpeter 
et dl., 1969), it is surprising that some workers still consider only the 
range in the emulsion as the relevant factor in limiting resolution. 

In conclusion since the photographic ( E p ) and geometric ( E g ) factors 
limit resolution independently, their total effect (E t ) can be considered 
by the following formula (Bachmann and Salpeter, 1965): 

E t =±((E p ) 2 + (E g ) 2 )V2 (6.1) 

It can be seen that if either the geometric or photographic factor is con- 
siderably larger than the other, it will dominate the final result. 

'Range should be distinguished from path length. Path length defines the total 
distance travelled, while range is the straight line distance travelled in the emitted 
direction. Path length is thus decreased only by energy loss, but range by both 
energy loss and scattering. 


Resolution Determination 

The resolving potential of a method can be expressed fully by describing 
the expected grain distribution around radioactive sources. Caro (1962) 
did this using H 3 -thymidine labeled bacteriophages (see also Caro and 
Schnos, 1965). The present authors obtained the distribution of devel- 
oped grains around a well-defined radioactive line source 400 A wide 
under systematically varying conditions (Bachmann et al, 1968; Salpeter 
et al, 1969; Salpeter and Salpeter, 1971). The source was made by sec- 
tioning at right angles a thin film of H 3 or C 14 polystyrene that had been 
mounted on the smoothed surface of an Epon halfblock. The experi- 
mental data were then arithmetically extrapolated to a point source and 


6 800 

S 600 




Grey Section 
Kodak NTE.-Dektol 

125 225 325 425 525 625 725 825 925 1025 1125 1225 


Distance (In I0A Units) 

Fig. 6.2A Histogram of grain distribution around the resolution specimen (line 
source) for a grey section, monolayer of Kodak NTE, Dektol development. The 
grains are added consecutively with increasing distance from the source, and 
thus each histogram column gives the total number of grains within this distance 
from the source. The distance from the source within which 50% of the grain 
fell (i.e., HD) is marked by the thin arrow at left; 10 times that distance is marked 
by thick arrow at right. It can be seen that when using integrated information 
to get a distance containing a given fraction of total grains (i.e., a "probability" 
distance), it is essential to count grains for enough distance from the source so 
that increasing the distance does not alter the results significantly (i.e., at least 
4-5 HD). We have used 10 HD. 


then to a more useful variety of other sources. Both density distributions 
(grains/ units area) and integrated distributions (total grains added 
consecutively) were described. 

It is, however, also useful to pick from this full distribution one 
arbitrary but well defined single value which is related to, but not 
identical with, a minimum distance which can actually be resolved. 

In EM autoradiography two different, equally meaningful points on 
the distribution curves have been used. One is based on the concept 
inherited from light optics which defines resolution as twice the distance 
from a point source at which the density of developed grains has dropped 
to 50% of its value over the source. It is obtained by determining grains 
per unit area at different distances from the source. This will be called 

3 4 

Distance /HD 

Fig. 6.2B Integrated distribution for a line and point source and density curve 
for a point source. Arrows indicate distance from the source of common values 
used to denote resolution. (Single arrow at HD, double arrow at 2 x 50% 
density, and dotted arrow at "50% probability" for point source, HR.) If one 
uses probability circles, they would come from the integrated distribution of the 
point source. It can be seen that the radius of a circle of 50% probability is 1.7 
HD (dotted arrow); that of 75% is 3.2 HD, and that of 95% is 7.3 HD. (The 75% 
and 95% probability distances are marked X.) In most autoradiograms, therefore, 
the use of circles of higher probability than at most 75% would entail radii 
larger than the average distance between radioactive sources. The overlapping 
information would make such large circles of little practical use. 





















O O 
O O 
00 o 

Q Q 




o I 

m o 






Q Q 

I 9 

o , _ 

o I o 
co co 

T- 04 



o o o o o o o 


CO ■* ■* CO ■* ■* o 

to (0 (/) 

■D T3 < 
g X <2 X » w ^ 

>> >% o 

c c C5 

a> a) w 

Q. Q. 

O O o o 

O O o o * 


O O V 1- ,- ,- * 

O O I I o 

co co o o o o o 

o o o o o 

o o o o CO 

z ra 

•g a 

o ~ 

T3 i± 




























— Kodak NTE, Dektol.grey section, H 3 

— Ilford L4, MX , gold section, H 3 
Ilford L4.MX, gold section, C 14 



6000 8000 10,000 

Fig. 6.3 Relative density distributions with distance from the source for three 
autoradiographic preparations of the test specimen differing in photographic and 
geometric conditions. All three are coated with monolayers of emulsion. The 
increase in spread of the distributions is consistent with their HD values (see 
Table 6.2). 

the density definition and was used by Caro and co-workers ( Caro, 1962; 
Caro and Schnos, 1965). 

The second called the integrated definition, is the distance from a 
developed grain with a 50% probability of containing the source ( Bach- 
mann and Salpeter, 1965). This is an integrated value since it is equiva- 
lent to the radius of a circle around (i.e., distance from) a source which 
contains 50% of the total developed grains. This is obtained by cumula- 
tively counting all grains with increasing distance from a source. 

Both density and integrated definitions here refer to point sources. 
Related values (both density or integrated) can be obtained for extended 
sources — the actual value being dependent on the shape of the source. 
We have recently obtained one integrated value experimentally for a 
line source (i.e., the distance from a line within which 50% of the grains 
lie), and have termed this distance HD (half distance) (Figs. 6.2 A, B). 
As will be seen later, all these definitions are interchangable once the 
grain distribution is known. Experimental HD values are given in Table 
6.2. They are consistent with the above stated predictions. The spread of 


IS 4 4.5 5 
Distance /HO 

Fig. 6.4 Average experimental grain density values from a line source for 
tritium, normalized in units of HD. Experimental distributions ranged from Kodak 
NTE with a grey section (HD 800) to ILford L4 with a gold section (HD 1650) 
(HD values from Table 6.2). The experimental values fit closely to a theoretical 
function (smooth solid curve) (Bachmann and Salpeter, 1965; Salpeter ef a/., 
1969), which will be referred to as the "universal curve." 

developed grains around the radioactive source is greater (i.e., higher 
HD) with increasing section and emulsion thickness, and energy of the 
electrons ( geometric factors ) , as well as with increasing silver halide and 
developed grain size (photographic factors). Furthermore, changing any 
one factor alone produces a relatively small overall effect — a situation 
consistent with equation 6.1. 

The increased spread of developed grains as a function of the geo- 
metric and photographic parameters can also be seen from the grain 
density (grains/unit area) distribution illustrated in Fig. 6.3. A signifi- 
cant observation is that although differing in width, the general shapes 
of these distributions are roughly the same. Because of this, all the dis- 
tributions can be made to fit a single universal curve merely by re- 
normalizing them and tabulating the x-axis in units of their own HD 
(Fig. 6.4). 

Universal curves for different types of sources were then obtained by 
numerical integration (Salpeter et al., 1969) (Figs. 6.5A, B, C). With 


I I 

Distance/ HD 

Fig. 6.5A-D Representative extrapolated families of universal curves (from Sal- 
peter et a/., 1969). 

6.5A Density functions for hollow circles. The x-axis represents distance in units 
of HD to the nearest point on the circumference of the circular source. Positive 
values indicate regions outside the circles; negative values indicate regions in- 
side the circle; the zero represents points on the circumference. Each curve is 
labeled by the radius of the circular source in units of HD. Density of developed 
grains is in each case normalized to unity on the circumference of the circular 
source. The limiting case of a circular source of zero radius is, of course, a 
point source. For finite distances from the circumference, a hollow circular source 
of infinite radius behaves like a line source. The density distributions for a point 
source and a line source are marked p or O and 1 or infinity (°°), respectively, 
and denote the full potential range of density distributions for hollow sources. 

extended sources, both the distance from and the size of the labeled 
structure have to be in units of HD. It was considered appropriate to 
approximate sources that would be found in biological specimens and 
circular and band sources were decided on, labeled either only at the 
periphery or uniformly throughout. The curves for any sources in between 
(e.g., ovals or those with nonuniform distribution of radioactivity) can 
be obtained by combining two or three existing distributions. 

From the above discussion it is clear that HD has a dual use. It gives 
a value of resolution which is meaningfully related to a minimum distance 
that can be resolved (Fig. 6.6), and it provides a normalizing value 


-I I 

Distance/ HO 

6.5B Density functions for circular sources with uniform radioactivity throughout, 
i.e., solid discs. In the limit of zero radius the solid disc (like the hollow circle) 
behaves as a point source (marked O or p). However, at finite distances from 
its circumference, a solid disc source of infinite radius behaves like a semi- 
infinite half plane with its edge at x = 0. This curve is marked »ore (for edge). 

which, when used as a unit of measure for distance, causes the grain 
distributions due to a given source to fit the same universal curve inde- 
pendent of the resolution of the autoradiograms. 

From the universal distributions (curve 0, Figs. 6.5A, C), one can 
see that the several resolution values discussed above are interchange- 
able; for example, twice the 50% density is 1.5 HD and the 50% prob- 
ability circle for a point source is 1.7 HD. It is, thus, in principle, 
irrelevant which resolution value is chosen to describe a method. In 



1 1 1 1 1 1 T 


1 1 1 1 1 1 1 r~. 

Disc ^^ - 





/ 1 « 

0.6®~- . - 


/ ..••* 

i 0.5 




°°o \ 

: °° \v 

/ •' 2 -' 

/ • * 

* 0.3 

S°o o \\ 

/ s 4 - -^ 


°°o t*. 1 

•7/ ^ 

.*// • ^oOOOOOOQ 


o v.» 
o >&• 




*° . i . i ■ i i i 




Distance/ HD 

6.5C Relative number (integrated) curves for uniformly labeled solid circular 
sources (discs). Curves are again labeled by the radius of the source in units 
of HD. The total number of grains (relative number 1.0) from any source is the 
sum of all grains, combining those over the source with those outside. The data 
are, however, presented separately for inside versus outside, counting in each 
case from the circumference. For instance: for a source of radius 4 HD, 61% 
(0.61 relative number) of total grains is over the source. Also 21% is found be- 
tween the circumference and 2 HD outside the source. (See x axis at +2.) Thus 
over the source plus an annulus of 2 HD outside the source, one finds 82% 
of all the grains due to that source. With increasing source size, the number 
of grains inside the source increases relative to that on the outside. (See Table 
6.4A.) For more detailed information on integrated distributions, see Salpeter 
et a/., (1969). 

obtaining experimental values, practical considerations can determine 
the choice as long as the general interrelations are known.* 

"In the authors' laboratory the choice of a line rather than a point source was a 
matter of experimental convenience. A line was easier to make and provided a larger 
number of grains for statistical analyses. Since it was possible to extrapolate back to 
a point source arithmetically, it still allowed a conventional treatment of the data. 
Similarly, for the single value of resolutions, the authors chose to rely on the 
integrated value "half distance" in which developed grains were added consecutively 
(Fig. 6.2A). Experimentally it is easier to obtain a high statistical accuracy for an 
integrated value than for a density value. The statistical accuracy of any determina- 
tion is ztN 1 ' 2 , where N is the total number of grains involved in that determination 
(see p. 272). Thus, in the integrated resolution value the statistical accuracy derives 


I ' r 

0.8 L 


Distance/ HD 

6.5D Integrated curve for solidly labeled band sources. The positive x-axis 
represents successive distances outside the band (including both sides), and the 
negative x-axis represents successive distances inside the band approaching the 
center line. Each curve is labeled according to the half width of the band. The 
limit for zero width is a single line. Since such a band of zero width has no 
"inside," all relative numbers are outside the source and, therefore, are plotted 
only on the positive x-axis, normalized to unity. Again, as the width of the source 
increases the relative number of grains outside it decreases. Note different rela- 
tive distributions inside versus outside these band sources compared with circu- 
lar sources in Fig. 6.5C and Table 6.4B. 

from the total number of developed grains counted within the HD distance. When 
obtaining a density value experimentally, however, one has to determine density of 
many individual small distance intervals from the source. The accuracy of any 
single density value and thus of the final resolution value, therefore, depends upon 
the number of grains within this small distance unit. 

On the other hand, a disadvantage in obtaining a meaningful integrated value 
of resolution is that the grains must be counted over a distance from the source 
which is considerably larger than the HD value, i.e., a large "cut-off distance" must 
be used ( Fig. 6.2A ) . This is so since most of the grains related to a source must be 
collected before one can speak of half these grains. This has been emphasized in 
previous publications (Salpeter et al., 1969; Salpeter and Salpeter, 1971), but is one 
of the commonest errors encountered in the literature related to a "half distance" or 
"probability circle" determination of resolution. The problem of "cut-off distance" 
does not exist in obtaining a density resolution value. Once the distance has been 
reached at which the density has dropped to one-half or to that of any other pre- 
determined value, no further grain counting is necessary. 





n. -o Hollow circle 

\ • ootid disk 












\ \ 




I \ 

\ V 

\ \ 












1 1 

>= , 

-4-3-2-1 I 2 3 4 

Inside -•— — *■ Outside 

Fig. 6.6 Normalized density distributions for two circular sources of 0.5 and 4 
HD radius labeled either uniformly throughout (solid line) or only at the circum- 
ference (dotted line). These curves emphasize the relation betwen HD and actual 
resolvable distance. With a circular source which has a radius less than 1 HD, 
one cannot distinguish the distribution of radioactivity. (For the ratio of grain 
density between the center and periphery of extended sources, see Salpeter 
et a/., 1969, Fig. 16). 

Using C 14 the authors found that the general shape of the grain 
distribution was slightly altered, although the universal curves still are 
applicable (Fig. 6.7) (Salpeter and Salpeter, 1971). From these C 14 
curves and from theoretical considerations, one can see that with higher 
energy electrons a particularly prominent feature of the grain distribu- 
tions is a long flat tail. With energies above that of C 14 , HD, as now 
defined, will no longer give an appropriate resolution value unless some 
modification, such as the subtraction of the tail grains, is applied (for 
discussion, see Salpeter and Salpeter, 1971). 

From a practical consideration, resolution should not be allowed to 
remain in the realm of esoteric discussion. It should not merely provide 
a value which is put in the introduction to a paper which uses electron 
microscope autoradiography, yet remains irrelevant to what follows. 
Resolution (i.e., information on expected grain distributions) can be put 


4 5 

Distance /HO 

Fig. 6.7 Average normalized experimental density distributions for C 14 labeled 
line source (solid curve) compared with the universal curve for tritium (dashed 
curve). Cross bars represent maximum range of experimental deviations from 
the means. The C 14 distribution is almost indistinguishable from the universal 
distribution for tritium up to about 3 HD, beyond which point the "tail" of the C" 
curve is systematically above that for tritium. The fit is deemed good enough to 
allow the use of the families of tritium curves for analyzing C 14 autoradiograms. 

to work in extracting information from autoradiographs. A brief discus- 
sion on how this can be accomplished is provided in the section on 


General Considerations 

The sensitivity of an autoradiographic method can be defined as the 
ratio of developed grains in an autoradiogram to the radioactive dis- 
integrations which occurred in the specimen during exposure. This is fre- 
quently expressed as per cent efficiency, i.e., grains/decays X 100%. 

The sensitivity of the method depends mainly upon the properties of 
the emulsion, the developing technique and the type of radiation. It is 
also influenced by absorption and scattering of the radiation in the 
specimen, in the emulsion and in any intervening layers; by chemical 
interaction between specimen and emulsion ( chemography ) ; and by 
latent image fading during prolonged exposure periods. 



Thickness" % Electrons 

Material' A emerging**" 

(density 1.1) 

H 3 














*The methacrylate was assumed to be comparable to a biological specimen where the isotope is 
randomly distributed throughout, emitting electrons at all angles. The "% emerging" from such 
a section represent those which are not "self absorbed." 

"Thickness refers to vertical distance. Since electrons are emitted in all directions their range 
has to be considerably larger than the thickness in order to emerge above. (See Table 6.1) 
***The values for H 3 in methacrylate from 500 to 5000 A have recently been confirmed experi- 
mentally (Salpeter and Szabo, 1972). One can see that with H 3 more of the self absorp- 
tion (~13%) in 1000 A of methacrylate occurs within the first 500 A and relatively less «10%) 
between 500 and 1000 A. This means that if in testing sensitivity for autoradiography the source 
(e.g., radioactive gelatin or methacrylate) is between 500 and 1000 A thick, the sensitivity value 
obtained is applicable to all sections in practical use in EM autoradiography. 
The problem of absorption with C 14 for EM autoradiography is negligible. 

Self absorption of radiation within the section will cause a decrease 
in autoradiographic sensitivity. This depends upon the density and the 
thickness of the specimen (Table 6.3). For tritium electrons in metha- 
crylate (density 1.1), it can be predicted theoretically that a loss due 
to self absorption in 1,000 A thick sections will be less than 20%, and the 
loss due to increase in thickness from 500 A to 1,000 A will be less than 
10% ( see also Perry, 1964 ) . Vrensen ( 1970 ) has disputed this prediction, 
claiming much higher self absorption in methacrylate sections ranging 
in thickness from 500 to 1,000 A. However, Salpeter and Szabo (1972) 
have recently verified experimentally that the self absorption in metha- 
crylate sections increasing in thickness from 500 to 1,000 A is not experi- 
mentally detectable with current sensitivity techniques. An alternative to 
Vrensen's interpretation is given in the discussion on the effect of radia- 
tion dose on sensitivity. 

Fixed and stained material might have densities about a factor of 
two higher than methacrylate, however (Huxley, 1957), and these 
densities may vary considerably among tissue components (Maurer and 
Primbsch, 1964). At present no experimental study is available which 
has considered the effect of varying tissue densities on sensitivity at the 
electron microscope autoradiographic level. 

Emulsion thickness will influence sensitivity until the thickness sur- 
passes the range of the radiation. As already discussed, most of the 
emulsion layers used in autoradiography are thinner than the range 
even of tritium electrons and thickness variations are thus expected to 
influence sensitivity. 


The radiation to which the emulsion is exposed can also influence 
sensitivity in various ways. Most obvious is the effect due to the overlap 
of developed grains (more than one developable latent image in an area 
the size of one developed grain). This is possible wherever developed 
grains are larger than silver halide cystals. An overlap of developed grains 
will result in a lower sensitivity since two developed grains in the same 
area might be counted as one. Fortunately, an overlapping of developed 
grains becomes obvious to the observer and can be guarded against by 
shortening the exposure period. 

Less obvious is the multiple hit problem. For a given dose, the 
probability of having more than one electron hit a silver halide crystal 
increases with the area occupied by that silver halide crystal. Since a 
crystal of Kodak NTE emulsion occupies roughly an area of .003 yu. 2 
versus .01 fx 2 iox a crystal of Ilford L4, the probability of multiple hits in 
the finer grained emulsion is considerably less. 


T 1 1 1 1 r 

~i r 

Gold- Elon Ascorbic 







7 8 9 

Decoys //i 




Fig. 6.8A % Efficiency (left scale) or Sensitivity (right scale) plotted as a func- 
tion of radiation dose (decays/^ 3 ). Monolayers of Ilford L4 emulsion were ir- 
radiated with tritium using a layer of radioactive gelatin of known thickness. All 
the data came from the same batch of emulsion. (Similar curves have been 
obtained with other emulsion batches and with radioactive methacrylate as 
source.) There is a definite decrease of sensitivity with increasing radiation dose. 
The probability of multiple hits within this range of radiation is < 5%. The effect 
appears strongest with MX and least with gold latensification EA3. 


Multiple hits have been considered the only dose-dependent influence 
on sensitivity. Recently, however, a more puzzling effect of radiation 
dose on sensitivity of Ilford L4 emulsion has been observed (Figs. 6.8A, 
B). There appears to be a decreasing sensitivity with increasing dose 
even in a range where the probability of multiple hits is negligible ( Sal- 
peter and Szabo, 1972). The extent of this effect depends upon the 
developer used and appears to be more prominent in high solvent de- 
velopers. No such effect has been seen with Kodak NTE developed with 

Other important factors to consider in connection with sensitivity are 
chemography (chemical interaction between section and emulsion) and 
latent image fading (the destruction of latent images during exposure). 
For further discussion on this subject, the reader is referred to Bachmann 
and Salpeter ( 1967 ) and Rogers ( 1967 ) . Experimental values for several 
emulsion-developer combinations are included under "Technical Con- 


1 1 1 1 

1 1 1 1 

1 1 1 1 





Gold- p. 
- - J^°"__ 4scor6/ c 

— -. 




1 1 1 1 

1 1 1 1 

I 1 1 1 

- 1/3 

- 1/4 

1/5 ;1 


1/6 % 

- 1/7 


- 1/20 

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 I.I 1.2 1.3 1.4 1.5 

^ • ,2 

Fig. 6.8B Same data as in 6.8A, but plotted as a function of grain density. 
This presentation of the data allows one to correct for the dose dependence in 
quantitative autoradiography. The sensitivity (or efficiency) applicable for con- 
verting the experimental grain density for any given organelle to that of radio- 
active decays can be read directly from the curves. 


Measuring Sensitivity 

To some extent, the effect on sensitivity of changes in various parameters 
can be predicted on the basis of physical data. Because of the com- 
plexity of the problem, however, experimental determinations for com- 
parison with theoretical considerations are essential. In order to measure 
sensitivity, a calibration method must be used in which all the factors 
that influence sensitivity do this to the same extent as in actual auto- 
radiography. If no single calibration specimen can accomplish this, in- 
formation from different ones should be combined. 

One wants to know first what the sensitivity of an emulsion layer is, 
without the effect of exposure time or specimen chemistry. This is useful 
information especially for each new batch of emulsion, and for a single 
batch after varying periods of shelf life. Secondly, one wants to test the 
effect of prolonged exposure periods, since frequently EM autoradiograms 
"cook" for as long as one year. Exposure atmosphere and chemical inter- 
action between tissue and emulsion can then express themselves. 

A simple method for testing sensitivity employs sections of radio- 
active plastic material which is available from commercial suppliers. It 
permits, primarily, the determination of sensitivity as a function of sec- 
tion thickness as well as the comparison of different developers and 
emulsions (Kopriwa, 1967; Vrensen, 1970; Salpeter and Szabo, 1972). 
Recently Miiller ( 1971 ) has suggested the use of an infinitely thick layer 
of tritiated polymer as the source for calibrating sensitivity. By using 
this specimen, the problem of thickness variation causing different de- 
grees of self-absorption can be avoided. The problem with such a speci- 
men, however, is that the energy spectrum of electrons reaching the 
emulsion is altered, and is thus not comparable with that from the thin 
sections actually used in EM autoradiography. 

The radioactive plastic sections, coated with emulsion cannot be used 
to test the influence of chemography resulting from the interaction of 
fixed and stained sections with emulsion layers. A test specimen consisting 
of a thin radioactive layer of measured thickness (Bachmann and Sal- 
peter, 1967) on a glass slide which can be clamped in contact with a 
photographic emulsion for a given exposure time and then separated can 
be used to assess chemography as well as latent image fading. 

In order to assess latent image fading, the emulsion layers are ex- 
posed, separated and either developed immediately or stored in different 
atmospheres for varying lengths of time before development. For the 
effect of tissue chemistry on sensitivity, a nonradioactive tissue section 
variously fixed or stained can be coated with the emulsion layer, which is 
then irradiated from above with the radioactive test specimen. The 


specimen can then again be developed either immediately or after con- 
siderable storage periods. The sensitivity of the emulsion overlying the 
section can then be compared with that of the emulsion over the glass. 
Background must always be determined separately and subtracted. 

Using a thin radioactive gelatin layer, Salpeter and Szabo (1972) 
have recently demonstrated that there is little (~20%) latent image 
fading after twelve months' exposure when using Ilford L4 emulsion 
stored in the refrigerator and developed with Microdol X. This was true 
even with heavily stained, Epon-embedded sections. Dektol-developed 
Kodak NTE, on the other hand, even though stored in helium showed a 
decrease in sensitivity and an increase in background after approximately 
three months. This effect was more marked with stained sections. A 
carbon layer between sections and emulsion was necessary with the finer 
grained emulsion (NTE) to maintain the sensitivity for even this long. 

It is essential for quantitation to recalibrate the method occasionally, 
especially when a new batch of emulsion or a new developer is used. 
We have found systematic fluctuations in sensitivity of ~20% with dif- 
ferent batches of emulsion and, as Fig. 6.8 indicates, sensitivity para- 
meters depend strongly upon the developers used. Furthermore, certain 
developers contain unstable components such as gold chloride (used 
in gold latensification ) whose short shelf life even in sealed vials can 
drastically affect sensitivity. 

In conclusion, in spite of the complexity, sensitivity data indicate that 
autoradiographic results can give absolute quantitation with a reliability 
well within a factor of two, and considerably better than that if special 
care is taken (~ ±20-30%). 

Section Thickness 

From the earlier discussion it is clear that for both sensitivity and resolu- 
tion determinations one of the factors of primary importance is the 
knowledge of section thickness. The standard procedure for judging 
overall section thickness is by the interference color of the section while 
floating on water. Peachey (1958) and Bachmann and Sitte (1958) have 
published scales relating the interference colors of sections with their 
measured thickness. Recently Williams and Meek ( 1966 ) have suggested 
that judgements by interference colors may not be sufficiently accurate. 
The reasons for this are that the eye does not notice thickness differences 
within the section and the overall judgement varies from one person to 
the other. 

In order to eliminate this problem it is necessary to actually measure 
each section. Such measurements can be made with incident light inter- 


ferometers when the section is mounted on a flat substrate. In this 
method, once the sections are on the collodion-coated slide, but before 
they are coated with the emulsion, the measurement does not need any 
additional steps in specimen preparation. A Nomarski type interferometer 
for incident light can be used. This attachment is available for the 
Reichert Zetopan microscope. The displacement of the interference lines 
by the section is measured with a filar eyepiece ( Fig. 6.9 and "Technical 
Considerations"). The method gives quick results (~1 min per measure- 
ment) and thickness variations within a section show up. For measuring 
sections several thousand Angstroms-thick with the Nomarski interfer- 
ometer, one should use white rather than monochromatic light. Then, the 
displacement of the zero or low order fringe can easily be recognized, 
even if the displacement is several half-wavelengths. 

Each investigator can decide for himself by the variability encoun- 
tered on a few measured sections whether he needs to measure each 

Fig. 6.9 Image as seen in Nomarsky interferometer used in measuring section 
thickness. The section is mounted on a microscope slide. The double appearance 
of the section is caused by the interferometer. The thickness (t) of the section 
is obtained from the following formula: t = d/D x Vz x. (D is the distance be- 
tween the interference lines on a flat surface, d is the extent of the offset caused 
by the section, X is the wavelength of the light.) In the above case D = 5mm, 
d = 1mm; and the measurements were done in green light (\ = 5470A). Thus 
the section thickness is Vs x 2735 = 5500A. (The decision on which way the 
line is offset has to be made in white light by observing the displacement of 
the zero order (black) interference line. Note that on thick sections the displace- 
ment d can be much larger than D.) 


section or only use spot checks from time to time. We have found that 
thickness judgements based on interference colors are usually accurate 
to ±100 A. When an Epon block is particularly soft there could be a 
systematic error due to greater compression immediately after cutting. 
For this reason, it is recommended to do interferometric spot checks of 
at least 20% of the sections before coating with emulsion. With an inci- 
dent light interferometer, one cannot measure the thickness of sections 
mounted directly on grids, which is a problem for those investigators 
who do not use a flat substrate method. This can be accomplished in 
transmitted light, however, using an interference microscope. 

For light autoradiography, one frequently uses 0.5-1.0 /a sections. To 
determine how reliable ultramicrotome settings at this range are, we 
measured such sections cut with the LKB ultratome III, and the Huxley 
Mark I microtomes. The error in the thickness was considerable when 
using the LKB; the Huxley microtome was much more reliable. It is, 
however, not possible to make any quantitative statements when using 
sections in this thickness range without calibration measurements. This 
problem is thus considerably greater at the light microscope level than at 
the electron microscope level, where interference colors are a reasonable 
criterion for section thickness. 

Emulsion Thickness 

Many coating procedures result in fairly uniform emulsion films but the 
use of an external criterion to check the emulsion thickness over the sec- 
tion is still necessary. First, one has to decide what is an optimum emul- 
sion layer and agree on a definition of a monolayer. In early electron 
microscope autoradiographic studies randomness of silver halide distribu- 
tion was used to demonstrate "good layers" or "monolayers." Caro ( 1962 ) 
emphasized the need for close packed rather than random layers, and 
pointed out that large spaces of gelatin between the silver halide de- 
creased both sensitivity and resolution. The effect on sensitivity is ob- 
vious. Electrons which travel through the emulsion in gelatin cannot be 
registered as a latent image. 

The effect on resolution is less obvious. This is due to the fact that 
silver halide has a higher scattering capacity than gelatin. Thus, if an 
electron enters into gelatin it will continue to travel in a straight line 
until it leaves the emulsion or hits a silver halide crystal where it is reg- 
istered. On the other hand, if it hits a silver halide crystal on entering 
the emulsion, it will be scattered from the straight line path and is less 
likely to see a neighboring silver halide crystal. This was called "shield- 
ing" by Caro (1962). The positive effect of scattering on resolution is 


similarly operating in low energy versus higher energy electrons and is 
illustrated in Fig. 6.1. 

What, then, is an optimum layer? One obviously wants it closely 
packed with minimum gelatin spaces. Since an emulsion consists of spheri- 
cal silver halide crystals which are not uniform in size, such a closely 
packed layer invariably entails some overlap. We call such layers single 
layers or monolayers. Vrensen (1970) also emphasizes the advantage of 
slightly overlapped layers. The question arises whether there is any 
advantage in multilayering emulsion. It is clear that if the range of the 
electron exceeds the thickness of a single layer of silver halide, then 
multilayering the emulsion would increase sensitivity but decrease resolu- 
tion. The effect on resolution of multilayering can be seen in Table 6.2. 

With tritium the increase in sensitivity and decrease in resolution was 
experimentally verified for Kodak NTE emulsion layers increasing in 
thickness from 700 to 1,400 A (Bachmann and Salpeter, 1967, Salpeter 
et al., 1969). For Ilford L4 we previously reported a 60% increase in 
sensitivity when thickness was increased from 1,300 to 2,700 A. We now 
believe that this rise was partly due to the dose dependence ( Fig. 6.8 ) , 
since we used a lower radiation dose for the thicker layers. We now use 
layers which are slightly overlapped and measure about 1,700 A. With 
these layers, when doubling the thickness, no significant increase in 
sensitivity was found experimentally. It is, therefore, concluded that with 
tritium radiation, emulsion thickness above about 1,700 A will no longer 
critically affect either sensitivity or resolution. With C 14 and S 35 radiation 
there still is a linear increase in sensitivity with emulsion thickness up to 
2,800 A (Salpeter and Salpeter, 1971). 

A reliable criterion must, therefore, be available for determining 
emulsion layering immediately over the section. If the tissue sections to 
be coated are mounted on collodion-filmed glass slides, determining 
emulsion thickness is relatively easy. When viewed in white light, thin 
emulsion layers on glass slides show interference colors which depend 
upon their thickness. The underlying collodion film and the section 
do not influence these interference colors since their refractive index 
is very similar to that of glass. The thicknesses of different emulsion 
layers were measured with a Nomarski interferometer and correlated 
with their interference color Fig. 6.17 (Bachmann and Salpeter, 1967). 

A close packed slightly overlapped layer of Ilford L4 which we now 
use as the optimum Ilford "monolayer" is deep purple or on the fringe 
of blue. Because of its density such a layer cannot be viewed directly in 
the electron microscope, but can be observed when its gelatin is stained 
with PTA and the silver halide removed in fixer. For centrifuged NTE, a 
pale gold layer is a "monolayer." For a detailed description of how 


emulsion interference colors can be judged over individual sections, refer 
to the Technical Considerations section (below). 

Different methods for emulsion coating have been described above 
involving dipping, dropping, looping and even specially designed ma- 
chines. In the hands of different investigators each method can give 
satisfactory layers. If the interference color criterion is used to determine 
the acceptability of layers, and if this is rechecked each time tissues are 
coated, the detailed method used is irrelevant and many are equally 


How can one extract the maximum information from an autoradiogram 
and with what degree of confidence can this be accomplished? The prob- 
lem of analysis of autoradiograms is at present the least clarified aspect 
of autoradiography. This may stem from the fact that too frequently one 
confuses density of developed grains with concentration of label. The 
two are obviously related but are certainly not identical. Figure 6.5B 
serves to remind us of that fact. By definition, the label or radioactivity is 
uniformly distributed inside these circular structures with none present 
outside; the grain density, on the other hand, rises towards the center 
and drops gradually outside. 

It is, therefore, helpful to think of analysis in two parts: (1) the 
gathering of information regarding the distribution of developed grains; 
and (2) relating these data to the distribution of radioactivity, while 
having an understanding of the limits of any quantitative assessment. 
The quantitative accuracy frequently depends upon the questions asked 
and the geometry of the specimen. 

Developed grain analyses were formerly carried out primarily qualita- 
tively. There is no doubt that this type of analysis can produce very 
useful information when the radioactive region is isolated or of large size 
relative to the resolution. For example, in Fig. 6.10 it is clear that the 
developed grains (radioactivity) are associated with the apical zone of 
the cell, and in Fig. 6.11 with the region containing dense filaments. This 
type of analysis does not allow one to determine amount or distribution 
of the label, however. Similarly, in Fig. 6.12 the nucleolus is obviously 
labeled, but no statement regarding the localization of radioactivity in 
the cytoplasm is possible. Furthermore, without quantitation, one could 
not determine whether the few nuclear grains are due to radiation spread 
from the nucleolus or due to low levels of extra-nucleolar radioactivity. 

Ross and Benditt (1965) first emphasized the need for quantitative 
analysis of developed grain distributions and the use of grain densities 



[;«£. 'so 

ft -I 



4 ' 


<■ . ! 




I* 1K 


fc : -... «M 



fk: J 


i Jlfci 


■ ■'■ v 4,i ^i'SiyRT " 

f~ ••' V'vSi,.' • ' ' -Ww^miF^'' £ -•'•■-'' l 


Fig. 6.10 Proximal tubule cells from rat kidney 10 minutes after perfusion with 
I 125 . Developed grains are in apical zone of cell which contains numerous vacuoles. 
Ilford L4 emulsion, D 19 developer, 12,000x. (From A. B. Maunsbach, J. Ultr. Res. 
15:1966, Fig. 7). 


?«* «£.: 




Fig. 6.11 Epidermal cell in regenerating limb of adult newt after injection of 
a H-leucine. Label is seen primarily over bundles of cytoplasmic filaments, flford 
L4 emulsion, paraphenylenediamine developer, 22.500X. 



i v--*-! 

Fig. 6.12 Nerve cell after injection of H 3 -l-histidine. Label appears randomly 
scattered in cytoplasm, whereas nucleolus is heavily labeled. Ilford L4 emulsion, 
Microdol X developer, 18,000x. 




^ [|if 


1 ; nm| 


— .u-fr^ — 

,' 1 



■' ' i : 

' ^ ,» JP I j*~ 



] h^ : y4— 

• '"' "-T ' *— " 

>■ ■ i 

■-'■■ 1 ■*■ :■■"• 


r --q-.-j ... 


i^ — i — .} — — 

lL. s "BISK IP '^"^ 


i jt 

fi' r ' 




— .- 


^ L • 




.„ J-- — i 


! ;** 


• •T 

r^r — 7 

F — ;— 

' m 

"ian — ~"~ 





1 -*^ 


>»•■ v 

■~-4. — 

■ r 

it- :-' 

1 «* 


f M 


, $., 'pie/* *. 

I 1 

1_ ] ] 


1 -WlW 

HEP wr* »' 
■'■ -1— 




i a— 

fcv> i 


:\ t. 

• . j V-v 

* i ' 


■ ;! 

• 1 


1 ' f&- 


i^%-; '• * 



■■i 9 


si! * s 


Fig. 6.13 Method for determining grain density: One can locate grain centers, 
and compare with grid points associated with any given organelles (for illustra- 
tion, see Ross and Benditt, 1965; Faeder and Salpeter, 1970). Alternatively, one 
can use "probability" circles (Bachmann and Salpeter, 1965; Salpeter, 1968; 
Williams, 1969; Nadler, 1971). Circles of set radius around developed grains are 
used to localize the grains and these are compared with circles of equal size 
drawn around random points. Cartilage cell labeled with H 3 proline — Kodak NTE, 
Gold-EA. Picture from Salpeter (1968). 



(grain per unit area) for that purpose. For different organelles in an 
autoradiogram the number of developed grains overlying an organelle 
was divided by the area occupied by this organelle. In order to get an 
area estimation, a grid of random points can be placed over the auto- 
radiogram and the relative number of points over each organelle tabu- 
lated. If the grid is calibrated for points per /j, 2 , the area of the organelles 
can be stated in absolute terms. Cutting out regions of the autoradio- 
grams and weighing them is another method frequently used for obtain- 
ing area estimations. 

Several authors have tried to include the developed grains due to 
radiation spread immediately outside an organelle. Such a procedure is 
most useful for relatively small structures. This can be accomplished by 
using the "probability circles" ( Bachmann and Salpeter, 1965 ) which are 
drawn around developed grains. If any part of such a circle is found over 
a structure, the grain is at least partly associated with it. Area correc- 
tions then have to involve similarly-sized random circles. For the applica- 
tion of such an analysis the reader is referred to Salpeter ( 1968 ) , Wil- 
liams (1969) and Nadler (1971) (Fig. 6.13). In special cases, where the 
geometry of the specimen is favorable, a full grain density distribution 
can be drawn around selected structures (Salpeter, 1969; Budd and 
Salpeter, 1969; Israel et al, 1968; Plattner et al, 1970). 

Fig. 6.14A-D Taken from various publications, illustrate one way of applying 
the universal curves for determining the distribution of radioactivity in autoradiog- 
raphy. One can for instance ask whether a structure is labeled; whether the 
developed grains outside the structure are due to a local low level of radioactivity 
present or merely due to radiation spread; and what the distribution of radio- 
activity within the structure is. As these figures illustrate no a priori knowledge 
on any of these aspects is needed. 

One first obtains the HD value applicable to ones autoradiogram from Table 
6.1. Then one plots an experimental grain density histogram going inside and 
outside the structure. Since the limiting membrane, or edge, is usually easy to 
see (or define), we have chosen to use it as the starting zone (0 point) for the 
distribution. The distance of all developed grains (centers) are measured from 
the edge (+ outside; — inside) and tabulated in histogram form, as the number 
of grains within progressive distances (histogram columns). It is convenient to 
use multiples of HD (.5-1) for the distance units. Identical tabulation has to be 
performed for grid points. To obtain enough grains and points for a statistically 
reliable analysis, data have to be collected from many representative micrographs 
randomly obtained. When the total number of grains per histogram column is 
divided by the total number of random points the data convert to a density dis- 
tribution. The SE of each histogram column is determined by the formula for 
the SE of a ratio governed by Poisson statistics, (see Appendix C; From the 
formula % SE = ± (1/VN) x 100, we see that the accuracy with 100 grains is 
±10%.) It is useful to have more points than grains so that the grains become 
the limiting factor in accuracy. 

The density is then normalized to 1 at the edge by dividing all the densities 
by that of the column. The experimental distribution can then be compared 
with that expected for different types of sources. 


2.0 r 


-2.4 -1.2 

< — cytoplasm' 

Distance /HD 

6.14A From Israel et al. (1968). The experimental grain density (histogram) fits 
one expected from an extended band source uniformly labeled. 

Once the gathering of developed grain information is accomplished, 
this has to be related, with a known degree of accuracy, to the distribu- 
tion of radioactivity. It is this operation which is the least trivial aspect 
of autoradiography. One needs to know what grain distributions to ex- 
pect from different distributions of radioactivity, and then to see which 
one of these match best with one's own experimental results. 

The families of universal curves (Salpeter et al., 1969) provide just 
this information on expected grain distributions for representative sample 
distributions of radioactivity. They are not a method of analysis but 
should be considered more as a set of constants. They contain informa- 
tion to be used in relating one's experimental grain concentrations to 





£ 1.0 





-2 -I 




6.14B From Budd and Salpeter (1969). The experimental histogram fits one 
expected for a labeled circular source with no radioactivity within 3HD outside 
it. The distribution inside was not uniform, however, since it did not rise as 
high as expected from a uniformly labeled source. Thus more radioactivity is 
present at the periphery than at the center. 

possible distributions of radioactivity (a series of examples is given in 
Figs. 6.14A-C). The information is finally derived from the "goodness of 
fit " between the two distributions. No a priori knowledge of the real 
distribution of radioactivity is required. 


<o 0.6 








Outside ■ 
Distort ce/HD 



■ .8 



- .6 

- .4 



- .2 

- 1 


- 8 



■ 6 




■ .4 


- .2 










inside outside 

Distance/ HD 

6.14C From Budd and Salpeter (1969). The experimental density histogram does 
not fit one expected from a specifically labeled structure. There was as much 
radioactivity outside as inside. 

6.14D From Plattner ef a/. (1970). The experimental density histogram fits one 
expected of a circular source labeled throughout. The experimental histogram 
outside was however incompatible with one merely due to radiation spread. There 
was therefore some radioactivity outside the tested structure. The dashed line 
represents distribution expected from radioactivity only at periphery. 

It should be emphasized that the information contained in the uni- 
versal curves does not only apply to grain distributions tabulated in the 
form of grain density histograms, but also applies to comparisons be- 
tween grain densities or relative number of grains, in two areas. For 
instance, the integrated curves (Figs. 6.5C-E) show what fraction of the 
total grains due to a source will fall over it (or within a ring of certain 
thickness around it) and what fraction will fall outside that area. It can 
be seen that these fractions change with the size of the labeled structure. 
The relative number lying over a small structure will be considerably 
lower than that for a large structure. A direct, noncorrected comparison 
of grain density between small and large structures will thus always favor 
the larger structure; Table 6.4 ( a and b ) illustrates this point. 

The few examples given above were included to alert investigators 
against blindly converting grain density to specific activity of label 





Grains (centers) 

% Grains 






over structure + 

circled grains/ 


HD = 1600 





random circles 



















1 /4 






"Data obtained from integrated curves (Fig. 5c). 
"Obtained by drawing probability circles, 1HD in radus, around centers of developed grains 
and considering all grains whose circles fall on the structure. 






(half thickness) 


% Grains over Structure 


for HD = 1,600 A 

Disc Solid band 


5 /i 

90 95 



60 80 


3,200 A 

25 45 



2.5 15 

•Data from Figs. 5c and d. 

without a consideration of some of the subtle factors which lie between 
the two. For a more complete discussion of analysis of autoradiograms 
see Salpeter and McHenry (1972). 


One subject not directly related to electron autoradiographic technique, 
and therefore not to be treated in detail here, is the extraction and dis- 
placement of radioactive material during processing. The problem of 
water-soluble electron microscope autoradiography is awaiting the de- 
velopment of thin frozen sectioning techniques. The current state of this 
art has been reported (Roth and Stumph, 1969; Stirling and Kinter, 1967). 
Meanwhile, there have been attempts to assess the extent of extraction 
of radioactive materials when employing modifications of conventional 
preparatory procedures; an introduction to this area is given by Williams 
( 1969 ) . Proteins and nucleic acids have never been a major problem. 
Certain lipids, on the other hand, are extracted (Stein and Stein, 1968 


and 1971; Idelman, 1964) but not radically redistributed during dehydra- 
tion (Hedley-Whyte et al, 1969; Darrah et al, 1971). Hayat (1970) has 
discussed in detail the loss of cellular materials during preparatory pro- 

A problem in reverse was pointed out by Peters and Ashley (1967), 
who were concerned with the artifactual retention of amino acids during 
fixation with glutaraldehyde. They were interested in detecting early 
stages of protein synthesis, where even a small amount of retained amino 
acids presents a large percentage error. Unfortunately, their data were 
not presented in absolute terms and thus the extent of the artifact cannot 
be assessed. 

Recently, the retention of physiologically meaningful amino acid 
pools was studied using autoradiography in connection with neuro- 
transmitter action in invertebrate nervous systems ( Faeder and Salpeter, 
1970; Orkand and Kravitz, 1971). No major tissue background was seen 
in these studies. 

It appears that electron microscope autoradiography using a photo- 
graphic film as the radiation detector has technically been pushed close 
to its limit. A marked improvement can come about only with the intro- 
duction of a radical innovation such as an intracellular radiation detector. 

On the other hand, information on fixation and retention of radio- 
active materials is likely to increase in the near future, as is the interest 
in absolute quantitation. In certain instances, when irreversible binding 
of radioactive material can be demonstrated, absolute quantitation is 
obviously feasible (Salpeter, 1969). Yet, even when extraction of radio- 
activity is unavoidable, it is useful to express one's results in absolute 
units (Salpeter and Faeder, 1971). This can, for instance, provide esti- 
mates regarding the ability of a given cell to concentrate and retain 
material from its incubation medium. 

The useful biological applications are already too numerous to list. 
Thus, although limited in many respects, autoradiography has proven to 
be an indispensable adjunct to biochemical and finestructural studies 
and is able to provide quantitative information at the subcellular level, 
unattainable by any other means. 


Specimen Preparation 

Figures 6.15A and B illustrate some of the preparatory procedures em- 
ployed in different laboratories. For reasons given in the main text, we 
prefer the "flat substrate" method which is outlined in greater detail 














^ ; 






Fig. 6.15A Steps in "flat substrate" specimen preparation developed for quan- 
titative electron microscope autoradiography. From Salpeter and Bachmann: 

a. Ribbons of sections are placed on collodion coated slides and allowed 
to dry. 

b. Sections are stained in a tightly closed petri dish by drops of stain placed 
over the individual ribbons; after the staining period the stain is flushed off with 
distilled water. 

c. The stained sections are vacuum coated with a thin layer of carbon. 

d. Liquid emulsion, which is kept at 60°C, 
then drained and dried in a vertical position; 
emulsion is equally satisfactory. 

is dropped over the section and 
dipping the specimen into liquid 


Fig. 6.15B Alternative to the flat substrate method. 

a. Sections are placed directly onto microscope grids which are attached to 
a glass slide. 

b . Emulsion is applied in pre-gelled form with a wire loop (Caro and Van- 
Turbergen, 1962; Caro 1969) or 

b . The slide is dipped into melted emulsion (Hay and Revel, 1963). 

c. The grid remains attached to the microscope slide during exposure and 
subsequent processing. Staining is usualy performed after photographic process- 
ing is complete. 

below. For an alternative flat substrate method, the reader is referred 
to Budd and Pelc (1964). 

Preparing Collodion-Coated Slides 

Frosted end glass slides should be carefully but not chemically cleaned. 
To clean, the slides are rubbed gently in "lakeseal" detergent, and 
then rinsed in distilled water. Drying the slides is an important step. 

e. The final specimen sandwich. 

f. Slides are developed in a series of beakers, with distilled water rinse be- 
tween successive chemicals, the sequence being: developer, 3% acetic acid 
stop bath for 10 seconds, non-hardening fixer (20% sodium thiosulfate + 2.5% 
potassium metabisulfite) for 1 minute, three rinses in distilled water. 

g. Specimen sandwich is stripped onto a water surface, and grids are then 
placed over the tissue sections. 

h. A procedure for picking up the specimen from the water surface by suction 
onto a moist filter paper applied over a filter plate. 


They should not be allowed to air dry since this may cause any con- 
taminants from the water to form a film on the slides. It is recommended 
that they be wiped dry with Kleenex; lens tissues or Kim Wipes are not 
recommended as they appear to contain a water soluble "glue" which is 
transferred to the wet slides. Once dry, the slides are blown free of lint 
with a Mini-Duster and then dipped into ~0.7% collodion in amyl acetate 
and dried vertically. If the collodion is bought as a solution in chloro- 
form, it is recommended that the solvent be removed by drying in a 
fume hood. 

A 2% (weight by volume) solution of collodion can be made in amyl 
acetate; the collodion takes at least overnight to dissolve. This solution 
can be stored indefinitely. Before use, an ~0.7$ dilution is made and 
filtered several times through the same filter paper. The diluted solution 
has a shelf-life of about one month. The repeated filtering is to prevent 
dirty collodion films which may ruin otherwise valuable autoradiograms. 

The collodion backing is thicker than that customarily used for 
routine grid supports. The reason for this is that it can withstand the 
treatments and storage involved in autoradiography and still be stripped 
from the slide at the end of the process. To judge whether the collodion 
film is of the right thickness, strip one onto water. It should exhibit a 
silver interference color; a light gold color indicates an unnecessarily 
thick film. 

Section Mounting 

Ribbons of thin sections from tissue embedded in either methacrylate 
or epoxy resin are sectioned in the conventional manner. Only ribbons 
of uniform thickness should be used. If dirt is a problem, it helps to 
transfer the sections from the knife boat to a beaker of clean distilled 
water before finally transferring them to the slide. They are transferred 
with a sharpened applicator stick, a glass rod, or a metal or plastic loop 
to a drop of distilled water or 10% acetone on a collodion-coated slide. 
Care must be taken not to tear the collodion film, for it will tear easily 
if touched while wet. (If a film is accidently damaged and the tear is 
at least a few mm from the sections, the specimen can sometimes be 
salvaged by applying a minute amount of amyl acetate to the film at the 
torn area. ) The drop of water is then drawn away with the applicator. As 
the drop is moved, the ribbon of sections settles onto the collodion with 
very little water around it. Once at a distance from the sections, the drop 
can be removed with a thin strip of filter paper or just shaken off. This 
precaution is taken in order to prevent any dirt from the drop drying 
on or around the sections. The location of the sections should be indi- 


cated on the back of the slide by a circle inscribed with a diamond 

Measuring Section Thickness 

At this stage or any time before emulsion coating, thickness can be 
measured with the Nomarsky type incident light interferometer (Fig. 
6.9). The slide is placed on a piece of black glass with a drop of water 
between the slide and the support. The water drop prevents any light 
reflection from the bottom side of the slide from decreasing the crispness 
of the interference lines. Thickness is measured by a displacement of 
these interference lines. The direction of the displacement has to be 
determined in white light by observing the Zero-order interference line 
(which always appears black). The measurement is then done in mono- 
chromatic light using a filar eyepiece. Accuracy of thickness measure- 
ments for 1000 A sections is =tl0%. 


Specimens prepared for autoradiography have very low contrast unless 
certain special treatment is introduced. The majority of the procedures 
for improving contrast involve specimen manipulation, either staining, 
gelatin removal or both, after the photographic process is complete (for 
review, see Moses, 1964). The most frequently used method for post- 
staining is that recommended by Hay and Revel ( 1963 ) which employs 
lead stain according to Karnovsky ( 1961 ) through the developed emul- 
sion. We have not been successful in getting consistent results with 
these methods. 

Sections can be stained before coating with emulsion. The slide 
bearing the sections to be stained is placed in a Petri dish, a few drops 
of the stain are placed over the sections, and the dish is covered. Only 
clean, precipitate-free, centrifuged stain should be used and it should 
not be allowed to evaporate during the staining period. After staining, 
the drops of stain should be flushed off quickly and gently with distilled 
water. Alternatively, the stain can be rinsed off by repeated dipping of 
the slide into distilled water in a series of beakers. To prevent "de- 
staining" during photographic processing, a thin carbon layer (~50 A) 
should be evaporated over the stained section, thus being interposed 
between the stained section and the emulsion (Koehler et al., 1963; 
Salpeter and Bachmann, 1964). 

Prestaining has the major disadvantage that it does not produce speci- 
mens with as high a contrast as does poststaining. Staining may also 


extract radioactive compounds, and may affect the emulsion despite the 
carbon layer interposed between stained section and emulsion. Finally 
prestaining with uranyl acetate makes subsequent stripping of the col- 
lodion film more difficult. 

Uranyl acetate and/ or lead are the most commonly used stains. We 
have used the former most frequently in a 2% aqueous solution for 3-4 hr, 
and have found it reliable for many tissues, although the contrast is 
lower than that obtained with lead. Lead staining (Karnovsky, 1961; 
Reynolds, 1963) may result in dense deposits on specific components 
such as chromatin, cellular filaments, mitochondrial membranes and extra- 
cellular collagen. The deposits appear to result from certain interactions 
during photographic processing. Figure 6.16 illustrates this effect. 

Lead stained sections, which without emulsion are perfectly clean, 
show dense, selectively located deposits when coated with emulsion and 
then processed. Not all tissues are equally affected. The inexperienced 
worker may confuse such deposits with developed grains if fine grained 
developers are used. If lead staining or any new staining procedure is to 
be used, it is essential to test the appearance of the autoradiograms 
either by using nonradioactive tissue or by processing a prepared auto- 
radiogram immediately after emulsion coating but without any exposure 
period. Any emulsion serves equally well for this purpose. 

In view of the limitations of all the above methods, we have tried 
the following post-staining procedure with promising results. After a 
specimen, which has been processed by the flat substrate procedure, is 
dried onto its microscope viewing grid (Fig. 6.15, A, h) the thick col- 
lodion film is removed by floating the grid face down on amyl acetate, 
while leaving the processed emulsion intact. (The collodion is the top 
layer of the specimen sandwich on the grid.) The section is thus de- 
nuded and can easily be stained by floating the grid on drops of stain. 

Intermediate Layer 

A thin (50-100 A) carbon layer should be evaporated over the section 
before emulsion coating. National SPK spectroscopic carbon, which can 
be obtained from Fisher Scientific Corp., is recommended. The carbon 
layer serves several functions. If prestraining is used it tends to protect 
the stained section from being destained by the developing fluids and 
to protect the emulsion from the stained section. It also provides a sur- 
face with uniform physical and chemical properties over which an even 
emulsion layer can be formed. 

We have found that the carbon layer is less essential to protect the 
sensitivity of the more stable Ilford L4 emulsion, especially when de- 

v0 ^*Iv 

W'" a. 


Fig. 6.16 Illustration of artifact which can be produced during processing of 
autoradiograms in which the section had been pre-stained with lead. 

A. Non-radioactive section stained with lead (Reynolds, 1963,) coated with 
the emulsion (llford L4) and developed immediately (with MX). Note dense de- 
posits, especially in mitochondria, which could be mistaken for fine developed 

B. Section similarly stained but without emulsion, x 15,800. 


velopment is with large grain developers. The thickness of a carbon 
layer can be estimated from its color; a 50-100 A layer has a light grey 
shade. It should be noted that with certain poorly designed evaporators 
the intense heat generated during evaporation of the carbon makes the 
subsequent stripping of the specimen especially difficult. 

Emulsion Coating 

An early method for forming pre-gelled emulsion layers on wire loops 
was recommended by Caro and Van Tubergen (1962). See also Caro 
1969. If the section to be coated is placed directly on a metal microscope 
grid, such a method is essential. It cannot, however, be used with either 
Kodak NTE or Gevaert 307, which do not form pre-gelled layers in a 
loop. Although the "loop" method is used routinely in many laboratories, 
our experience has been that when the emulsion becomes completely 
dry it tends to break in the loop. The application onto the tissue is there- 
fore frequently performed with a slightly wet emulsion. Pile-up of silver 
halide crystals at grid wires can thus still occur during drying. A pre- 
ferred method for coating sections placed directly on grids is the strip- 
ping film method of Williamson and van den Bosch ( 1971 ) . 

If the surface to be coated is of uniform physical properties, as is the 
case when the flat substrate technique is used, very uniform emulsion 
layers can be obtained when emulsions are applied either in liquid or 
pre-gelled form. Since layers of different thicknesses exhibit different 
interference colors (Fig. 6.17), they can be used as an aid in selecting 
the dilution of an emulsion which will yield the desired laver. It is clear 
that interference colors can be seen only in white light and not under 
darkroom safe light conditions. When making emulsion dilutions, it is 
essential, therefore, to coat test slides which are then tal<en out into the 
light to check whether the right dilution has been reached and whether 
the layer with the desired interference color covers that area of the slide 
on which the sections are placed. Interference colors can be seen most 
clearly when viewed from the back of the slide. 

Since the draining speed and, thus, the thickness of the emulsion 
depend upon the substrate, it is necessary to use test slides which are 
coated with the same substrate as will be used for the final auto- 
radiographic preparation; in the procedure given here it would be a 
carbon coated collodion film on glass. Once a dilution is reached which 
gives the desired interference color in white light, this color can be 
judged also under safe light conditions. This is so because different inter- 
ference colors then appear as shades of grey, and if one knows approxi- 
mately the region of the slide containing, for instance, a purple layer 
(i.e., tightly packed monolayer of Ilford L4), then its boundaries with the 





500 — 
1000 — 

1500 — 
Emulsion ar 


— 500 

— 1000 

— 1500 




, COWff? ,., 


^sias!?^' i^^^^^^s^^^^^^^S 

^d section thickness in A 

Fig. 6.17 Interference color — thickness scale for sections and emulsions. Inter- 
ference colors of sections on water and of emulsions on glass slide. 

thinner gold layer and thicker blue layer can be seen as lighter and 
darker shades of grey respectively. One can thus determine if the de- 
sired layer is in fact over the sections which are located by the ring 
scratched on the underside of the slide. Sample emulsion layers can be 
seen in Fig. 6.18. 

ilford l4 

The Ilford L4 emulsion is diluted (1 g to 4 ml) with distilled water, 
melted in a warm water bath at 40-50 °C, and gently stirred. This will 
give too thick a layer and will have to be diluted further. The final dilu- 
tion is determined empirically. After cooling to room temperature, test 
slides are coated with emulsion after successive dilutions ( adding 1 ml of 
water at a time) until the desired interference color when examined in 
white light is obtained. 

In order to coat, the slides can be dipped into the emulsion, and 
drained and dried in a vertical position. Alternatively, emulsion can be 
dropped onto the slide with a medicine dropper which is held hori- 
zontally. The excess is drained back into the beaker of emulsion. 
The dropper method is recommended when the amount of emulsion is 


limited since this method can be applied with very small quantities. 
We do not agree with Granboulan (1965) that dipping is basically 
superior to dropping. The "loop" method of Caro and Van Tubergen 
( 1962 ) is equally successful with the flat substrate method. If the inter- 
ference color is used as a criterion for judging the layer, then the final 
result is indistinguishable independent of the coating procedure used. 
Irrespective of the method used, the draining and drying should be 
carried out while the slide is in a vertical position if a uniform layer is 
to be obtained. 

The final dilution will depend upon the choice of the dropping or 
dipping method; on the rate of withdrawal of the slide from the emul- 
sion in the case of dipping, and the length of time the emulsion is on 
the slide before draining in the case of dropping. We have tried the 
semi-automatic coating apparatus designed by Kopriwa ( 1966 ) with 
limited success. Using a dilution of 1:2.5 and a speed of 84 mm/min, ex- 
cellent, purple monolayers can be obtained, although the method is not 
always reproducible. Vrensen (1970) has recommended using the 
Kopriwa machine with the above dilution of the Ilford emulsion, and the 
same withdrawal speed of 84mm/min. 

Ilford L4 is the emulsion of choice for most routine EM auto- 
radiography. It has a generally low background, long shelf life, and 
extremely high stability over long periods of exposure, up to at least one 
year ( Salpeter and Szabo, 1972 ) . 


As presently supplied, the Kodak NTE emulsion still contains too much 
gelatin to make monolayers of closely packed silver halide crystals. In 
order to reduce the gelatin content, 4 g of emulsion are first dissolved in 
40 ml distilled water in a water bath at 45-60°C. The warm emulsion is 

Fig. 6.18 — A. Layer of centrifuged Kodak NTE emulsion, interference color, 
silver; measured thickness, 600A. 

B. Slightly overlapped layer of centrifuged Kodak NTE emulsion. Interference 
color, gold; measured thickness, 1000A. As "monolayer" of Kodak NTE we now 
use a pale gold interference color and thus between A and B in thickness. 

C. Layer of Gevaert Nuclear 307 emulsion. Interference color, silver to gold. 
Such uniform layers of this emulsion were not obtained consistently. 

D. Layer of Ilford L4 emulsion. Interference color, purple; measured thickness, 
1500A. (The emulsion gelatin was stained with phosphotungstic acid and the 
emulsion layer then fixed. The images of the silver halide crystals are thus 
negatively stained ghosts. A close-packed layer of Ilford L4 is otherwise difficult 
to photograph without grossly disturbing the silver halide crystals due to the 
high beam intensity necessary for adequate illumination.) (From Salpeter and 
Bachmann, 1965.) As "monolayer" we now use a deep purple interference color 
(1700A) and thus a layer slightly more overlapping than shown here. 



then centrifuged at ~ 14,000 G for lOmin until the supernatant is clear. 
To facilitate the separation of the silver halide from the gelatin, the rotor 
of the centrifuge is first heated. This is done by allowing the centrifuge 
to rotate slowly for about 20 min while a hot air gun is directed at its 
rotor. If a small cart holds the centrifuge, the rotor can be heated outside 
the darkroom and then the whole apparatus wheeled into the darkroom 
just before use. 

After centrifugation, the bottom of the centrifuge tube is chilled 
briefly by dipping it in ice water and the supernatant discarded. The 
remaining concentrated emulsion is reheated in the hot water bath and 
resuspended in warm distilled water; l-2ml/g of the original emulsion 
used. The final dilution is determined empirically depending upon the 
emulsion thickness desired, i.e., by the interference colors of emulsion 
layers on test slides after successive dilutions as described for Ilford L4. 
Coating can be accomplished by applying the liquid emulsion with a 
medicine dropper over the specimen held horizontally. The dropping 
method is used with Kodak NTE because of the low volume of emulsion 
obtained after centrifugation. The emulsion is then drained off the slide 
and air-dried vertically. 

The NTE emulsion gives higher resolution (Table 6.2) and because 
of the smaller silver halide crystal size (~500A) is not as subject to 
multiple hits. On the other hand, it has lower over-all sensitivity ( Bach- 
mann and Salpeter, 1967; Salpeter and Salpeter, 1971), is not stable for 
exposure periods above 3 months, and seems to be more subject to 
chemography from prestained sections. 


Although this emulsion has been successfully used by Granboulan 
(1963), Kopriwa (1967, Vrensen (1970) and others, even after re- 
peated attempts using different batches of emulsion we have not had 
consistent results in preparing uniform layers. We, therefore, do not feel 
qualified to make recommendations regarding its use. 

We have no experience with any other emulsions described in the 
literature, such as the Japanese emulsions used by Uchida and Mizuhira 

Processing Autoradiograms 


Dried autoradiograms are stored in sealed black slide boxes. In order to 
avoid latent image fading which occurs with the Kodak NTE emulsion 


when specimens are stored in air during exposure, storage in helium is 
recommended; any other inert gas will do as well. For this emulsion, pre- 
liminary experiments with storage in the freezer suggest that this may 
be an alternative to helium. In that case, longer developing times (e.g., 
double) may be needed to offset loss in sensitivity. 

The Ilford L4 emulsion does not exhibit the same fading and can 
be stored in the refrigerator during exposure. With both Kodak NTE 
and Ilford L4, the specimens must be thoroughly dried before storing. 
The addition of Drierite to the storage containers is essential. 


Cleanliness is an important consideration during all the steps in develop- 
ment. All solutions should be made with distilled water. A distilled water 
rinse is used between consecutive steps. Slides are developed one or two 
at a time in 30-50 ml beakers at room temperature (23-24°C) unless 
otherwise recommended in the detailed descriptions below. All sensitivity 
values reported below will be for radiation from H 3 . 


1. Microdot X (Pelc et al, 1961; Caro and Van Tubergen, 1962) is a 
reliable standard developer. Three min development results in developed 
grains 2,000-^,000 A in diameter (Figs. 6.12 and 6.15C). Sensitivity is 
~%-/4o; background levels are low (~0.25 grain/100/x. 2 ). One disad- 
vantage is the recently described dose dependence (Figs. 6.8A and B). 
(Salpeter and Szabo, 1972). 

2. D 19 is also frequently used. It gives higher background (~1.3 
grain/100 (J? ) but also higher overall sensitivity ( ~ % ) ( Kopriwa, 1967; 
Vrensen, 1970; Salpeter and Szabo, 1972). (Fig. 6.8A). Development for 
2 min at 20 °C results in grains which are rather elaborate and slightly 
larger than those yielded by Microdol X (Figs. 6.10 and 6.15F). "Dose 
dependence" is lower than with Microdol X. 

3. Gold Latensification. The procedure called "gold latensification" 
which deposits gold on latent images and thereby increases their de- 
velopability, enhances sensitivity by a factor of 2 or 3. It was described 
by James (1948) and James et al. (1948), and was applied to auto- 
radiography for Kodak NTE (Salpeter and Bachmann, 1964) and 
adapted for Ilford L4 (Wisse and Tates, 1968). The concentration of 
gold used and the time the specimen is exposed to it effect the size of the 
final developed silver grains. 

The sensitizing gold thiocyanate solution is made as follows: a 2% 
stock solution of gold chloride ( AuCl 3 HCl • 3H 2 ) is made in distilled 


water ( which had been boiled with a few drops of bromine added ) and 
subsequently cooled. It may be stored in a plastic bottle for approxi- 
mately one month. Before using, 1 ml of the 2% stock solution is diluted 
in 100 ml brominated boiled distilled water. (Wisse and Tates 1968 
recommend that the pH be adjusted slowly to 7.0 with NaOH. We have 
found this pH adjustment to be unnecessary). Potassium thiocyanate 
(0.25 g) and potassium bromide (0.3 g) are added, and the solution made 
up to 500 ml with boiled distilled water. This solution should be used 
within 1-8 hr after making. 

Gold latensification has been used in autoradiography in combination 
with an Elon-ascorbic acid developer of Hamilton and Brady (1959) 
(Salpeter and Bachmann, 1964; Wisse and Tates, 1968). The size and 
shape of the final developed grain can be manipulated by adding Na 
sulfite without affecting the sensitivity to any marked extent (Salpeter 
and Szabo, 1972). Elon-ascorbic acid developer, without Na sulfite (EA), 
forms developed grains consisting of clusters of silver specks. (Fig. 
6.19G). The addition of Na sulfite to the developer (EAS) causes a 
filamentous developed grain to grow (Fig. 6.19D). 

EA (for punctate grains) is made by dissolving each of the following 
chemicals consecutively in 150ml boiled distilled water: 0.225 g Elon 
(Metol), 1.5 g ascorbic acid, 2.5 g sodium borate (borax) and 0.5 g potas- 
sium bromide. The solution should be made up to 500 ml with boiled 
distilled water. This developer in conjunction with 5 min of gold latensifi- 
cation has a background of ~2 grains/ 100 jit 2 and a sensitivity of ~%. 

EAS ( modification for filamentous grains ) is made by dissolving each 
of the following chemicals consecutively in 150 ml boiled distilled water: 
0.45 g Elon, 1.5 g ascorbic acid and 2.5 g sodium borate, 0.5 g potassium 
bromide and 7.5 g sodium sulfite. The solution should be made up to 
500 ml with boiled distilled water. This developer in conjunction with 
5 min of gold latensification has a considerably lower background, even 
lower than Microdol X (~0.04 grains/100 fi 2 ). Dose dependence is 
greatly reduced and overall sensitivity is high (~%) (Fig. 6.8). 

The gold latensification sensitizer and developers are used as follows: 

Gold (at20°C) 

5 min 

Distilled water 


EA (at20°C) 

8 min 

EAS (at20°C) 

4 min 

Nonhardening fixer 

1 min 

Distilled water 

30 sec in each of three changes 


4. p-Phenylenediamine is a physical developer recommended for 
electron microscope autoradiography by Caro and Van Tubergen ( 1962 ) 
and Caro (1964) and yields developed grains of 500-700 A. With 
1 min development at 24°C, these grains are compact rather than fila- 
mentous and obstruct the underlying fine structure of the specimen less 
than do the Microdol X grains (Figs. 6.11 and 6.19B). The developer is 
made by dissolving, in a water bath at 60°C, 1.1 g p-phenylenediamine 
in 100 ml of a 12.6% solution of sodium sulfite. It is then cooled and 
filtered. The developer is unstable and should be made fresh on the day 
of use. For adequate sensitivity with this developer, an intermediate 
carbon layer is essential (Bachmann and Salpeter, 1967). Background 
tends to be more variable than with the other developers. Sensitivity, 
which is '~%5, can be raised to % by gold latensification. 

5. Phenidone. Lettre and Paweletz (1966) have recommended a 
phenidone developer consisting of 1.5% ascorbic acid, 0.25% phenidone, 
0.6% potassium bromide, 1.3% potassium carbonate, 20% sodium sulfite, 
and 6% potassium thiocyanate. The developed grains are nonfilamentous 
and easy to count. In our hands, this formula gave too high a back- 
ground, 4-5 grains/100 fx. 2 , to be acceptable. 


1. Dektol. The standard developer for the Kodak NTE emulsion is 
Dektol (diluted 1 to 2 with distilled water). Developing for 2 min at 
24°C results in 800-1,200 A developed grains which are either filamentous 
or appear as a cluster of small, unconnected spheres or rods (Fig. 6.15A); 
with this procedure the background is <~3 grains/ 100 \i? and sensitivity 

2. Gold latensification and Elon-ascorbic acid. The sensitizing gold 
thiocyanate solution and Elon ascorbic solutions are prepared and used as 
described for Ilford L4 (pages 000). Gold latensification using EA de- 
velopment (Salpeter and Bachmann, 1964) has proved unreliable. Pre- 
liminary studies, however, indicate that gold latensification followed by 
EAS development at 24° C has a sensitivity of ~Vi 8 and lower back- 
ground than that produced by Dektol, i.e., ~2 grains/100 fi 2 . This 
promises to be a more reliable development for NTE. 

Final Steps in Photographic Processing 

After development by any of the solutions described above, slides are 
rinsed in distilled water, dipped in a 3% acetic acid stop bath for 15 sec, 
again rinsed in distilled water, and then fixed for 1 min. To facilitate the 

| j M|lg 






K£5¥EpaB* r s*?.. A 

Fig. 6.19 Illustration of different developed grains. X35.000. 
A — E. Radioactive "hot line"; F-G. random irradiation. 
A. Kodak NTE, developed with Dektol (2 min). 



&w. 4u 





B. Ilford L4, developed with paraphenylenediamine (1 min). 

C. Ilford L4, developed with Microdot X (3 min). 

D. Ilford l_4, developed with gold latensification-Elon Ascorbic (EAS; 4 min). 

E. Kodak NTE, developed with gold latensification-Elon Ascorbic (EA; .8 min). 

F. Ilford L4, developed with D19 (2 min). 

G. Ilford L4, developed with gold latensification-Elon Ascorbic (EA; 8 min). 



final stripping of the specimen, a nonhardening fixer is recommended. 
This type of fixer consists of 20% sodium thiosulfate (hypo) and 2.5% 
potassium metabisulfite. After fixing, the slides are rinsed in several 
changes of distilled water. 

Stripping and Mounting Specimens 

If the flat substrate method is used, the specimen sandwich, consisting of 
collodion film, section, carbon layer and developed emulsion, is stripped 
onto a water surface after photographic processing. Metal grids are 
placed over the individual ribbons, which can be identified on the float- 
ing film by their interference colors, and the sandwich is picked up. One 
method of picking up the specimens is seen in Fig. 6.15A. 

Stripping the specimen from the slide can be a problem especially if 
a long staining period has been used (e.g., uranyl acetate for 3hr). On 
the other hand, unstained specimens may float off prematurely if agitated 
roughly during developing and subsequent processing. If stripping is 
difficult, the following precautions should be observed: (1) the collodion 
film may be made slightly thicker; ( 2 ) after photographic processing the 
slide should be left in the last distilled water rinse for 15 mi n to 2 hr and 
should not be allowed to dry before stripping; (3) a nonhardening fixer 
should be used; (4) stripping should be done with the aid of a dissecting 
microscope or large magnifying glass; and ( 5 ) a teasing needle should be 
used to start and guide the process. 

Finally, a few drops of 10% hydrofluoric acid can greatly facilitate 
stripping. The slide is scored just below and above the region containing 
the sections. The slide is immersed into the stripping water which is 
drawn up to the lower score mark; any loose film is discarded. At that 
time 2-3 small drops of 10% hydrofluoric acid are released into the strip- 
ping bath near the scored slide. This loosens the film and allows easy 
stripping. A drop of the acid placed over the scored film (some dis- 
tance from the section) just prior to stripping may also work. 

Statistical Accuracy of Data 

Finally, we must emphasize that EM autoradiography is a statistical 
technique. Data regarding both developed grains and random points ( for 
obtaining relative area) follow Poisson statistics. A few guidelines will 
be helpful. 

If one has collected a certain number (N) of developed grains or 
random points, then the sampling error (SE) of each one of these values 
is ±\/N, and the percent error is (y/N/N) X 100%. These formulae 
apply only to the initial statistical samples (i.e., the developed grains and 


random points). Thus the statistical (sampling) accuracy of any final 
converted data (e.g., molar concentration) has to be determined from 
the raw data. Once obtained, however, the % error is applicable to any 
converted value. Because of this general applicability, the formula for 
the % error is the more useful. 

Whenever one has a ratio of two values (Ni/N 2 ) where each comes 
from an independent Poisson distribution (e.g., grains /points ) , the gen- 
eral formula for the % error is 

Since the SE of Ni is ±\/N, this last formula simplifies to: 
±\Zl/N-i + 1/N 2 X 100. For example, if the total number of grains col- 
lected is G and the total number of points collected is P, the grain density 
D = (G/P) has a % error of 


^ + I x 100. 

Similarly, the ratio of two densities (D1/D2) has a % error of 
or simplified 

± ^(SE|^ + (SE^ x 10Q 

*Jk + k + k + k x 


A specific illustration follows. Assume that in an autoradiographic 
experiment, we want to determine the radioactivity of a certain organelle 
after incubating the tissue with H 3 -leucine. We photograph many areas 
of the specimen randomly and then analyze the autoradiographs. A grid 
with random points is placed over these autoradiographs and the num- 
ber of points falling over different organelles tabulated. Let us assume 
that 60 points overlie the selected organelle. We also tabulate developed 
grains. We may tabulate only developed grains (i.e., grain centers) 
which fall over the organelle (in this case our final answer will 
have to be corrected for scattered grains as discussed in the section on 
resolution and in Figs. 6.5C and D). Alternatively we may do a 
density distribution, and collect, as belonging to our organelle, all the 
grains scattered from it. In either case, let us assume that we have 
counted 30 grains as belonging to our organelle. We thus have a grain 

density of 30/60 = .5 with a % error ± LL + _L x 100 = 22%. Had the 

\30 60 


sample been 300 grains and 600 points, the density would still be .5, but 
the % error only ±7%. 

Now, let us assume a sensitivity of 1/10 (i.e., 10 decays/developed 
grain) and a grid of 1 random point per 2/u, 2 . Our grain density then 
represents a decay density of 2.5 decays///. 2 of organelle surface, or, with 
a section thickness of 1000 A, 2.5 decays/0. 1/ji?. If we had an exposure 
time of 30 days, this is 9.6 X 10 6 decays/cc of organelle/sec. (The % 
error still remains 22% or 1% respectively.) 

We can now ask how much radioactivity was taken up into the 
tissue and retained during the processing for EM autoradiography. The 
following simple calculation can be done. Since 1 Curie = 3.7 X 10 10 
decays/sec, our organelle had retained 

9.6 X 1Q« decays/cc/sec lQ-*C/cc 

3.7 X 10 10 decays/sec/Curie ' 

If the specific activity of our leucine bath was 5C/mole, we have retained 
a 5.2 X 10 -2 Molar concentration of leucine, or 3.13 X 10 19 molecules/cc. 
(This of course refers to all the exogenous molecules, not just the radio- 
active ones. ) The statistical error still remains either 22% or 7% depending 
on the original sample size. 

We use the term retained advisedly because we want to emphasize 
that this represents only that amount of leucine taken up from the bath, 
which was retained in our organelle during processing. It also does not 
reflect any existing endogenous pool. The statistical sampling error dis- 
cussed here does not refer to any additional sources of error such as 
variations in section thickness, or fluctuations in sensitivity values. These 
latter would contribute to the overall accuracy of the result again as the 
square root of the sum of the squares. 


Supported by U.S. Public Health Service Grant GM 10422 from the Institute 
of General Medical Sciences, NS 09315 from the Institute of Neurological 
Diseases and Stroke and a Career Development Award K3NB-3738 from the 
Institute of Neurological Diseases and Stroke. We wish to thank Mrs. Frances 
McHenry for her help, support and friendship. 


Bachmann, L., and Salpeter, M. M. (1965). Autoradiography with the electron 
microscope: a quantitative evaluation. 1Mb. Invest. 14, 1041. 


. (1967). Absolute sensitivity of electron microscope autoradiography. 

/. Cell Biol, 33, 229. 

(1972). Electronenmikroskopische Autoradiographie. In: Methodens- 

ammlung der Elektronen mikroskopie (Schimmel, G. and Vogell, W., eds.). 

Wissenschafti Verlags ges Stuttgart. 
Bachmann, L., Salpeter, M. M., and Salpeter, E. E. (1968). Das Auflosungs- 

vermogen electronenmikroskopische Autoradiographien. Histochemie, 15, 

Bachmann, L., and Sitte, P. (1958). Dickenbestimmung nach Tolansky an 

Ultradunnschnitten. Microskopie, 13, 289. 
Budd, G. C. (1971). Recent developments in light and electron microscope 

radioautography. Int. Rev. Cytol, 31, 21. 
Budd, G. C, and Pelc, S. R. (1964). The membrane method of electron 

microscope autoradiography. Stain Technol., 39, 295. 
Budd, G. C, and Salpeter, M. M. (1969). The distribution of labeled 

norepinephrine within sympathetic nerve terminals studied with electron 

microscope autoradiography. /. Cell Biol., 41, 21. 
Caro, L. G. (1962). High resolution autoradiography. II. The problem of 

resolution. ]. Cell Biol, 15, 189. 
. (1964). High-Resolution Autoradiography. In: Methods In Cell Physiol- 
ogy. (Prescott, D. M., ed.) Vol. 1, p. 327. Academic Press, New York. 
(1966). Progress In High Resolution Autoradiography. In: Progress In 

Biophysics And Molecular Biology. Vol. 16, p. 171. Pergamon Press, New 

(1969). A common source of difficulty in high resolution radioautogra- 

phy. Science, 41, 918. 

Caro, L. G., and Schnos, M. (1965). Tritium and phosphorus-32 in high 
resolution autoradiography. Science, 149, 60. 

Caro, L. G, and Van Tubergen, R. P. (1962). High resolution auto- 
radiography. I. Methods. /. Cell Biol, 15, 173. 

Darrah, H. K., Hedley-Whyte, J., and Hedley-Whyte, E. T. (1971). Radio- 
autography of cholesterol in lung. An assessment of different tissue processing 
techniques. J. Cell Biol, 49, 345. 

Faeder, I. R., and Salpeter, M. M. (1970). Glutamate uptake by a stimulated 
insect nerve muscle preparation. /. Cell Biol, 46, 300. 

Granboulan, P. (1963). Resolving power and sensitivity of a new emulsion in 
electron microscope autoradiography. }. Roy. Micr. Soc, 80, 165. 

. (1965). Comparison Of Emulsions And Techniques In Electron Micro- 
scope Radioautography. Symp. Int. Soc. Cell Biol, 4, 43 (Leblond, C. P., 
and Warren, K. B., eds.). Academic Press, New York. 

Hamilton, J. F., and Brady, L. E. (1959). Electrical measurements on photo- 
graphic emulsion grains. I. Dark conductivity. /• Appl. Phys., 30, 1893. 

Hamilton, J. F., and Urbach, F. (1966). The Mechanism Of The Formation 
Of The Latent Image. In: The Theory Of The Photographic Process, 3rd 
ed. (Mees, C. E. K., and James, T. H., eds.) p. 87. Macmillan Company, 
New York. 


Hay, E. D., and Revel, J. P. (1963). Autoradiographic studies of the origin 
of the basement lamella in Ambystoma. Dev. Biol., 7, 152. 

Hayat, M. A. (1970). Principles And Techniques Of Electron Microscopy: 
Biological Applications. Vol. 1. Van Nostrand Reinhold Company, New York. 

Hedley-Whyte, E. T., Rawlins, F. A., Salpeter, M. M., and Uzman, B. G. 
(1969). Distribution of cholesterol- 1,2H 3 during maturation of mouse 
peripheral nerve. Lab. Invest., 21, 536. 

Huxley, H. E. (1957). Some aspects of staining of tissue for sectioning. /. 
Micros. Soc, 77, 30. 

Idelman, S. (1964). Modification de la technique de Luft en vue de la con- 
servation des lipides en microscopie eletronique. /. Microscopie, 3, 715. 

Israel, H. W., Salpeter, M. M., and Steward, F. C. (1968). The incorporation 
of radioactive proline into cultured cells: interpretations based on radio- 
autography and electron microscopy. /. Cell Biol, 39, 698. 

Jacob, J. (1971). The practice and application of electron microscope auto- 
radiography. Inte. Rev. Cytol., 30, 91. 

James, T. H. (1948). The site of reaction in direct photographic development. 
II. Kinetics of development initiated by gold nuclei. /. Colloid Set., 3, 447. 

James, T. H., Vanselow, W., and Quirk, R. F. (1948). Gold and mercury 
latensification and hypersensitization for direct and physical development. 
Phot. Soc. Am. J., 14, 349. 

Karnovsky, M. J. ( 1961 ) . Simple methods for "staining with lead" at high pH 
in electron microscopy. /. Biophys. Biochem. Cytol., 11, 729. 

Koehler, J. K., Miihlethaler, K., and Frey-Wyssling, A. (1963). Electron micro- 
scope autoradiography. An improved technique for producing thin films and 
its application to H 3 -thymidine-labeled maise nuclei. /. Cell Biol, 16, 73. 

Kopriwa, B. M. (1966). A semiautomatic instrument for the radioautographic 
coating technique. /. Histochem. Cytochem., 14, 923. 

— . (1967). The influence of development on the number and appearance 

of silver grains in electron microscope radioautography. /. Histochem. 
Cytochem., 15, 501. 

Lettre, H., and Paweletz, N. (1966). Probleme des elektronenmikroskopische 
autoradiographie. Naturwissenschaften, 11, 269. 

Liquier-Milward, J. (1956). Electron microscopy and radioautography as 
coupled techniques in tracer experiments. Nature, 177, 619. 

Maunsbach, A. B. (1966). Absorption of I 125 -labeled homologous albumin by 
rat kidney proximal tubule cells. /. Ultrastruct. Res., 15, 197. 

Maurer, W., and Primbsch, E. (1964). Grosse der /3 Selbstabsorption bei der 
H 3 -Autoradiographie. Expt. Cell Res., 33, 8. 

Moses, M. J. (1964). Application of autoradiography to electron microscopy. 
/. Histochem. Cytochem., 12, 115. 

Miiller, G. ( 1971 ) . Zur Standardisierung Der Elektronenmikroskopischen Auto- 
radiographie. Expt. Cell Res., 65, 386. 

Nadler, N. J. (1971). Interpretation of grain counts in electron microscope 
radioautography. /. Cell Biol, 49, 877. 


Neumann, D. (1969). Technik der elektronenmikroskopiscen Autoradiographic 

in der Biologic Teil I. Acta Histochem., 33, 191. 
Orkand, P. M., and Kravitz, E. A. (1971). Localization of the sites of 7-amino- 

butyric acid (GABA) uptake in lobster nerve-muscle preparations. J. Cell 

Biol, 49, 75. 
Peachey, L. D. (1958). Thin sections. I. A study of section thickness and 

physical distortion produced during microtomy. J. Biophys. Biochem. Cytol, 

4, 233. 
Pelc, S. B. (1963). Theory of electron autoradiography. J. Roy. Micros. Soc, 

81, 131. 
Pelc, S. B., Coombes, J. D., and Budd, G. C. (1961). On the adaptation of 

autoradiographic techniques for use with the electron microscope. Exp. Cell 

Res., 24, 192. 
Perry, B. P. (1964). Quantitative Autoradiography. In: Methods In Cell 

Physiology. (Prescott, D. M., ed.) Vol 1, p. 305. Academic Press, New 

Peters, T., and Ashley, C. A. (1967). An artifact in radioautography due to 

binding of free amino acids to tissue by fixatives. /. Cell Biol., 33, 53. 
Plattner, H., Salpeter, M. M., Saltzgaber, J., and Schatz, G. (1970). Promito- 

chondria of anaerobically grown yeast. Proc. Nat. Acad. Sci., 66, 1252. 
Beynolds, E. S. (1963). The use of lead citrate at high pH as an electron- 
opaque stain in electron microscopy. /. Cell Biol., 17, 208. 
Bogers, A. W. (1967). Techniques Of Autoradiography. Elsevier Publishing 

Co., New York. 
. (1971). Becent developments in the use of autoradiographic techniques 

with electron microscopy. Phil. Trans. Roy. Soc. 261, 159. 
Boss, B., and Benditt, E. P. (1965). Wound healing and collagen formation. 

V. Quantitative electron microscope radioautographic observations of proline- 

H 3 utilization of fibroblasts. /. Cell Biol, 27, 83. 
Both, L., and Stumpf, W. E. (1969). Autoradiography Of Diffusible Sub- 
stances. Academic Press, New York. 
Salpeter, M. M. (1966). General Area Of Autoradiography At The Electron 

Microscope Level. In: Methods In Cell Physiology. (Prescott, D. M., ed.) 

Vol. 2, p. 229. Academic Press, New York. 
. (1968). H 3 -proline incorporation into cartilage: electron microscope 

autoradiographic observations. /. Morph. 124, 387. 
-. (1969). Electron microscope radioautography as a quantitative tool in 

enzyme cytochemistry. II. The distribution of DFP-reactive sites at motor 
endplates of a vertebrate twitch muscle. J. Cell Biol, 42, 122. 

Salpeter, M. M. and McHenry, F. A. (1972). Electron miscroscope auto- 
radiography: analyses of autoradiograms. In: Electron Microscope Tech- 
niques in Biology, (Koehler, J. H., ed. ) Springer Verlag, Berlin, in press. 

Salpeter, M. M., and Bachmann, L. (1964). Autoradiography with the electron 
microscope. A procedure for improving resolution sensitivity and contrast. 
J. Cell Biol, 22, 469. 


. (1965). Assessment of Technical Steps In Electron Microscope Auto- 
radiography. Symp. Int. Soc. Cell Biol, 4, 23. (Leblond, C. P. and Warren, 
K. B., eds.). Academic Press, New York. 

Salpeter, M. M., Bachmann, L., and Salpeter, E. E. (1969). Resolution in 
electron microscope radioautography. /. Cell Biol., 41, 1. 

Salpeter, M. M. and Faeder, I. R. (1971). The Role of the Sheath Cells in 
Glutamale Uptake by Insect Nerve Muscle Preparations. Prog, in Brain 
Res. 34, 103. 

Salpeter, M. M., and Salpeter, E. E. (1971). Resolution in electron microscope 
radioautography. II. Carbon 14 . J. Cell Biol, 50, 324. 

Salpeter, M. M., and Szabo, M. (1972). Sensitivity in electron microscope 
radioautography using Ilford L4 emulsion: the effect of radiation dose. /. 
Histochem. Cytochem., (In Press). 

Stein, O., and Stein, Y. (1968). Lipid synthesis, intracellular transport and 
storage. III. Electron microscopic radioautographic study of the rat heart 
perfused with tritiated oleic acid. /. Cell Biol, 36, 63. 

. (1971). Light and electron microscopic radioautography of lipids: tech- 
niques and biological applications. Advances in Lipid Research, 9, 1. 

Stirling, C. E., and Kinter, S. (1967). High resolution radioautography of 
galactose- 3 !! accumulation in rings of hamster intestine. /. Cell Biol, 35, 585. 

Uchida, G. and Mizuhira, V. (1970). Electron Microscope Autoradiography 
with Special Reference to the Problems of Resolution. Arch, histol. Jap., 
31, 291. 

Vrensen, G. F. J. M. (1970). Some new aspects of efficiency of electron 
microscopic autoradiography with tritium. /. Histochem. Cytochem., 18, 278. 

Williams, M. A. (1969). The Assessment of Electron Microscopic Auto- 
radiographs. In: Advances In Optical and Electron Microscopy. (Barer, R., 
and Cosslett, V. E., eds.) Vol. 3, p. 219. Academic Press, New York. 

Williams, M. A., and Meek, G. A., (1966). Studies on thickness variations in 
ultrathin sections. J. Roy. Micros. Soc, 85, 337. 

Williamson, J. R. and van den Bosch, H. (1971). High Resolution Auto- 
radiography with Stripping Film. /. Histochem. Cytochem. 19, 304. 

Wisse, E., and Tates, A. D. (1968). A gold latensification Elon ascorbic acid 
developer for Ilford L4 emulsion. Proc. 4th. Europ. Reg. Conf. Electron 
Micros., Rome. 

Author Index 

Abermann, R., 64-66, 95, 98, 
201, 203, 206-208, 210, 
211, 214, 215, 217 

Aihara, K„ 199, 217 

Allen, L. C, 7, 42 

Ames, A., 54, 58, 70, 71, 96 

Arnold, E. A., 41, 48 

Asahina, E., 8, 43 

Ashley, C. A., 255, 277 

Bachmann, L., 64-66, 95, 98, 
160, 191, 200-204, 207, 
208, 210, 211, 214, 215, 
217, 221-223, 225, 226, 
228-231, 233-235, 239, 
240, 241, 244, 251, 259, 
264, 266, 268, 269, 274, 
275, 277, 278 
Backus, R. C, 117, 127, 143 
Bahr, G. F., 127, 143 
Baker, J. R., 3, 21, 22, 43 
Baker, R. F., 41, 43, 48 
Bancroft, J. B., 106, 143 
Barer, R. F., 7, 43 
Barnes, R. B., 170, 191 
Bassett, G. A., 198, 212, 215 
Bayley, S. T., 207, 215 
Bell, L. G. E., 22, 33, 43 
Benditt, E. P., 245, 249, 277 
Berendsen, H. J. C, 5, 8, 43 
Bererhi, A., 18, 43 
Bernhard, W., 23, 43, 45, 49 
Bernstein, L., 12, 47 
Bertaud, W. S., 91, 95 
Bessis, M., 175, 177, 192 
Bobalek, E. G, 170, 193 
Bondareff, W., 21, 22, 37, 40, 

43, 44 

van den Bosch, H., 222, 278 

Bowers, W. F., 121, 138, 143 

Bradley, D. E., 64, 95, 96, 

111, 121, 143, 155, 157, 

160, 165, 168, 169, 170, 

191, 198, 200, 202, 207, 


Brady, L. E., 202, 207, 268, 

Branton, D., 76, 86, 91-93, 
95-98, 207, 216 

Brassfield, T. S., 8, 9, 45 
Brenner, S., 102, 117, 143 
Budd, G. C, 221, 222, 252, 

253, 257, 267, 275, 277 
Bullivant, S., 9, 19, 21, 23, 

43, 46, 52, 54, 58, 61, 63, 

70, 71, 80, 82, 83, 96, 209, 

Burstone, M. S., 22, 43 
Burton, C. J., 170, 191 

Caro, L. G., 221, 223, 225, 

226, 229, 243, 257, 262, 

264, 267, 269, 275 
Casassa, E. F., 127, 143 
Caspar, D. L. D., 104, 126, 

137, 143 
Chalcroft, J. P., 80, 82, 83, 96 
Choi, H. Y., 24, 25, 48 
Christensen, A. K., 23, 43, 55, 

Chu, E. H. V., 10, 11, 46 
Clark, A. W., 86, 96 
Clarke, J. A., 175, 192 
Cohen, C, 126, 143 
Conjeaud, P., 212, 217 
Coombes, J. D., 222, 267, 277 
Cope, F. W., 7, 43 
Cosslett, V. E., 121, 143 
Cowley, C. W., 24-26, 29, 43 
Cox, R. W., 125, 143 
Crewe, A. V., 129, 142-144 
Crowell, J., 9, 36, 48 
Czubaroff, V. B., 121, 138, 


Darrah, H. K., 255, 275 
Davies, G. E., 119, 144 
Deamer, D. W., 207, 216 
Dear, J., 10, 45 
Dehl, R. E., 8, 43, 44 
Delsemme, A. H., 6, 44 
De Rosier, D. J., 116, 140, 

142, 144-146 
Derjaguin, B. V., 7, 9, 44 
Desforges, M., 207, 217 
Diamond, L. K., 177, 193 
Dick, D. A. T., 7, 44 
Dietz, A. A., 177, 192 

Dinichert, P., 170, 192 
Doebbler, G. F., 4, 44, 55, 96 
Donnelly, W. J., 177, 192 
Douglas, S. D., 93, 97 
Dowell, L. G., 6, 44 

Eckert, H., 22, 44 
Eisenberg, D., 5, 44 
Eisenberg, H., 127, 143 
Escaig, J., 207, 217 

Faeder, I. R., 249, 255, 275, 

Farrant, J. L., 9, 10, 11, 44- 

46, 102, 144 
Fenichel, I. R„ 7, 45 
Femandez-Moran, H., 21, 23, 

44, 117, 126, 144 
Ferrier, R. P., 125, 144 
Finch, J. T., 104, 119, 130, 

132, 135, 137, 138, 140, 

142, 144 
Fisher, H. W., 133, 134, 146 
Fisher, W., 170, 193 
Fletcher, N. H., 6, 44 
Florin, A. E., 7, 47 
Frank, H. S., 5, 44 
Frey, S., 140, 146 
Frey-Wyssling, A., 54, 63, 64, 

68, 91, 96, 97, 259, 276 

Gamble, W. J., 9, 46 

Gamboa, R., 9, 46 

Gersh, I., 21, 22, 34, 37, 38, 

44, 45 
Glenner, G. G., 24, 31, 46 
Goldblatt, P. J., 41, 48 
Goodenough, D. A., 95, 96 
Gough, J., 173, 189, 192 
Grahn, E. P., 177, 192 
Granboulan, P., 221, 223, 263, 

266, 275 
Griffin, C. C, 41, 48 
Griffiths, K., 173, 192 
Gruber, M., 112, 113, 146 

Haanstra, H. B., 169, 170, 

Haefer, R., 170, 192 



Hall, C. E., 54, 96, 102, 126, 

129, 133, 144, 155, 192, 

199, 208, 216 
Hamilton, J. F., 223, 268, 275 
Hanna, M. G., 10, 11, 45, 46 
Hanzon, V., 22, 45 
Harker, D., 165, 193 
Harris, P., 12, 45 
Harse, J., 173, 189, 192 
Hart, R. G., 200, 216 
Haschemeyer, R. H., 105, 

108, 113, 117, 119-121, 

126, 130, 131, 134-138, 

141, 143, 144, 147 
Hay, E. D., 221, 257, 259, 

Hayat, M. A., 3, 45, 83, 96, 

100, 111, 112, 118, 144, 

157, 192, 255, 276 
Haydon, G. B., 129, 133, 144 
Hayek, K., 200, 215 
Hedley-Whyte, E. T., 255, 

275 276 
Hedley-Whyte, J., 255, 275 
Heidenreich, R. D., 110, 144 
Heinmets, F., 157, 192 
Helwig, G., 165, 170, 193 
Henderson, W. J., 170, 171, 

173, 176, 189, 192, 193 
Henney, H. R., 108, 146 
Hepler, P. K., 13, 45 
Hermann, W., 202, 217 
Hermodssen, L. H., 22, 45 
Hess, W. M., 57, 97 
Hibi, T., 160, 170, 192, 216 
Hilbrand, H., 204, 215 
Hisada, J., 8, 43 
Hoeve, C. A. J., 8, 44 
Hoglund, S., 117, 129, 145 
Hohn, B., 106, 145 
Hohn, T., 106, 145 
Holland, L., 154, 193 
Holls, G. J., 140, 146 
Holmes, K. C, 104, 130, 132, 

135, 137, 138, 140, 142, 

Holt, S. J., 23, 45 
Honig, R. E„ 200, 203, 210, 

Horn, H. R. F., 210, 216 
Home, R. W., 102, 104, 108, 

112, 113, 115, 117, 125, 

132, 135, 137, 143, 145, 

Horowitz, S., 7, 45 

Hunger, G., 202, 216 
Huxley, H. E., 102, 108, 116, 

129, 142, 145, 146, 237, 


Idelman, S., 255, 276 
Isenberg, I., 21, 22, 44 
Israel, H. W., 250, 251, 276 

Jackerts, D., 157, 193 
Jackson, W. T., 13, 45 
Jacob, J., 221, 276 
Jacobs, M. H., 204, 216 
Jaeger, H., 204, 216 
Jakobsen, R. J., 7, 47 
James, T. H., 267, 276 
Jayme, G, 202, 216 
Johnson, M. W., 102, 145 
Josephs, R., 140, 145 
Joslin, C. A. F., 173, 192 

Kafig, E., 54, 97 

Karnovsky, M. J., 259, 260, 

Kautzman, W., 5, 44 
Kavanau, J. L., 5, 45 
Kayden, H. J., 175, 177, 193 
Kellenberger, E., 170, 192 
Kiermayer, O., 91, 98 
Kikuchi, M., 199, 217 
Kim, K. S., 41, 48 
Kinter, S., 254, 278 
Kleinschmidt, A. K., 157, 193, 

206, 216 
Klug, A., 104, 119, 129, 132, 

137, 138, 140, 142-145 
Knoch, M., 198, 216 
Koehler, J. K., 52, 58, 63, 69, 

91, 92, 96, 202, 259, 276 
Kollman, P. A., 7, 42 
Konig, H., 165, 170, 193, 198, 

Kopriwa, B. M., 240, 264, 

266, 267, 276 
Kozloff, L. M., 106, 145 
Kramer, D. A., 203, 210, 216 
Kranitz, M., 202, 216 
Kravitz, E. A., 255, 277 
Kreutziger, G. O., 58, 60, 62, 

71, 72, 96, 97 
Kuntz, I. D., 8, 9, 45 
Kurtin, B. C, 7, 45 
Kurtin, S. L., 7, 45 

Labaw, L. W., 208, 216 
La Farge, C. G., 9, 46 

Lambert, R., 57, 98 
Landelout, H., 57, 98 
Lang, D., 157, 193 
Law, G. D., 8, 9, 45 
Leberman, R., 114, 145 
Lebras, L. R., 170, 193 
Leduc, E., 23, 43, 45 
Leibo, S. P., 10, 11, 45, 46 
Leonard, R., 207, 216 
Lettre, H., 269, 276 
Levitt, J., 10, 45 
Levy, H. A., 5, 47 
Lewis, S. M., 177, 193 
Ling, G., 7, 45 
Lippincott, E. R., 7, 47 
Liquier-Milward, J., 222, 276 
Longergan, E. T., 177, 193 
Longley, W., 126, 143 
Lorenz, E., 200, 202, 217 
Lovelock, J. E., 10, 45, 56, 97 
Lubin, M., 117, 145 
Luftig, R. B., 125, 126, 145 
Luyet, B. J., 6, 15, 24, 26, 29, 
35, 45-47 

MacKenzie, A. P., 6, 22, 35, 

Mahl, H., 164, 193 
Malhotra, S. K., 36, 37, 40, 

46, 48 
Malkani, K., 18, 43 
Marchesi, V. T., 93, 98 
Markham, R., 140, 146 
Martin, J. C, 207, 217 
Maunsbach, A. B., 246, 276 
Maurer, W., 237, 276 
Mayer, H., 198, 216 
Mazur, P., 8-12, 15, 45, 46, 

56, 97 
McAlear, J. II., 58, 60, 62, 

71, 97 
McHenry, F. A., 254, 277 
Mead, C. A., 7, 45 
Meek, G. A., 241, 278 
Mellema, J. E., 112, 113, 146 
Menter, J. W., 198, 215 
Menz, L. J., 15, 47 
Mercer, P. D., 204, 216 
Meryman, H. T., 7, 9, 10, 11, 

22, 23, 36, 46, 54, 55, 97 
Meyer, H. W., 91, 97 
Migchelsen, C, 8, 43 
Misuhira, V., 266, 278 
Moline, S. W., 6, 24, 31, 44, 



Monroe, R. G., 9, 46 

Moor, H., 4, 8, 46, 52, 54, 

57, 58, 63, 64, 67, 68, 76- 

79, 91, 97, 98, 202, 217 
Moore, P. B., 116, 142, 146 
Morgan, C. L., 9, 46 
Moseley, M., 79, 80, 98 
Moses, M. J., 221, 259, 276 
Mueller, W. A., 7, 45 
Muhlethaler, K., 54, 57, 58, 

63, 64, 67, 68, 76-79, 91, 

96-98, 259, 276 
Mukherjee, B. B., 108, 146 
Miiller, G., 240, 276 
Murray, R. T., 125, 126, 144, 

Muscatello, U., 108, 113, 146 

Nadler, N. J., 249, 250, 276 
Nagakura, S., 199, 217 
Nagata, T., 37, 38, 46 
Nagington, J., 113, 132, 146 
Narten, A. H., 5, 47 
Nash, T., 4, 47, 56, 97 
Nathan, D. G., 177, 193 
Nawa, T., 37, 38, 46 
Nesmeyanov, A. N., 200, 203, 

Neumann, D., 221, 277 
Nevo, A. C, 12, 47 
Newton, A. A., 113, 132, 146 
Nuttall, H. F., 102, 146 

Ohtsaki, M., 126, 144 
Oketani, S., 199, 217 
Oliver, R. H., 108, 146 
Olmstead, E. G., 7, 47 
Orkand, P. M., 255, 277 
Orr, W. H., 200, 202, 215 
Osborn, J. S., 177, 193 
Oski, F., 177, 193 

Page, T. F., 7, 47 
Park, R. B., 95, 97 
Parkes, A. S., 56, 97 
Parsons, D. F., 118, 146 
Pashley, D. W., 198, 204, 

215, 216 
Paweletz, N., 269, 276 
Peachey, L. D., 241, 277 
Pearse, A. G. E., 21, 22, 47 
Pease, D. C, 11, 19, 22, 47 
Pelc, S. R., 222, 223, 225, 

257, 267, 275, 277 
Perry, R. P., 237, 277 

Peters, T., 255, 277 
Pinto da Silva, P., 92, 93, 97 
Plattner, H., 250, 253, 277 
Polge, C, 56, 58, 97 
Pooley, F. D., 170, 193 
Porto, S. P. S., 7, 48 
Powell, A. S., 170, 193 
Primbsch, E., 237, 276 
Purcell, G. U., 8, 9, 45 

Quirk, R. F., 267, 276 

Rapatz, G. L., 15, 47 
Rapideau, S. W., 7, 47 
Rawlins, F. A., 255, 276 
Rebhun, L. I., 4, 8, 11, 13, 

16, 21, 47, 48 
Reed, L. J., 108, 146, 168 
Reimer, L., 200, 202, 213, 

214, 217 
Remsen, C. C, 57, 97 
Revel, J. P., 95, 96, 221, 257, 

259, 276 
Rey, L., 37, 48 
Reynolds, E. S., 260, 261, 277 
Rhodin, T. N., 200, 202, 215 
Riehle, U., 58, 97 
Rinfret, A. P., 6, 44, 55, 96 
Rogers, A. W., 221, 239, 277 
Rohsenow, W. M., 24, 25, 48 
Rosenthal, A., 9, 46 
Ross, R., 245, 249, 277 
Rossi, G., 37, 38, 45 
Roth, L., 254, 277 
Rousseau, D. I., 7, 48 
Rowe, T. W. G, 36, 48, 55, 

Ryans, D. G., 91, 95 

Salpeter, E. E., 223, 225, 226, 
228, 230, 231, 233-235, 
244, 251, 266, 275, 278 
Salpeter, M. M., 221-223, 
226, 228-231, 233-235, 
237, 239-241, 244, 249- 
255, 259, 264, 266-269, 
Salsbury, A. J., 175, 192 
Saltzgaber, J., 250, 253, 277 
Sander, G., 4, 8, 12, 13, 16, 

21, 47 
Santeler, D. J., 209, 217 
Sassen, M. M. A., 57, 97 
Sawada, N., 4, 12, 48 
Sawdye, T. A., 24-26, 29, 43 

Schaeffer, V. J., 165, 193 
Schatz, G., 250, 253, 277 
Scheraga, H. A., 5, 48 
Schleich, F., 170, 193 
Schnos, M., 223, 226, 229, 

Schreil, W., 170, 193 
Schulte, C, 213, 214, 217 
Schumacher, H. R., 180, 193 
Scott, R. G., 170, 191 
Seal, M., 202, 216 
Sella, C, 207, 212, 217 
Shapiro, B. M., 119, 146 
Sherman, J. F., 41, 48 
Sherman, J. K., 55, 56, 98 
Sherwood, R. G., 204, 216 
Shimada, K., 8, 43 
Sidel, V. W., 177, 193 
Siegel, B. M., 200, 202, 215 
Silte, P., 241, 275 
Simpson, F. O., 91, 95 
Sjostrand, F. S., 21, 22, 37, 

39, 48, 126, 146 
Sleytr, U., 82, 98 
Smith, A. U., 4, 48, 56, 97 
Smith, J. A., 177, 193 
Smith, L. H., 10, 11, 45, 46 
Solomon, A. K., 7, 48 
Somenda, K., 61, 71, 96, 209, 

Southworth, D., 91, 93 
Stadtman, E. R., 119, 146 
Staehelin, L. A., 91, 98 
Stark, G. R., 119, 144 
Steere, R. L., 54, 60, 61, 63, 

65, 69, 70, 79, 80, 84, 86, 

Stefani, S. A., 177, 192 
Stein, O., 254, 278 
Stein, Y., 254, 278 
Steiner, J., 21, 49 
Stephenson, J. L., 21, 22, 24, 

36, 37, 44, 48 
Sterling, C, 6, 48 
Sterling, K., 177, 193 
Steward, F. C, 250, 251, 276 
Stirling, C. E., 254, 278 
Stowell, R. E., 41, 48 
Sturart, P. R., 177, 193 
Stumpf, W. E., 254, 277 
Szabo, M., 228, 237, 239- 

241, 264, 267, 268, 278 

Tardieu, A., 207, 216 
Tates, A. D., 267, 268, 278 


Thon, F., 110, 146 
Tillack, T. W., 93, 98 
Timson, W. J., 24-26, 29, 43 
Tranzer, J. P., 23, 45 
Trillat, J. J., 212, 217 
Troschin, A. S., 7, 48 
Trump, B. F., 41, 48 
Turnbull, A. C, 173, 192 

Uchida, G., 226, 278 
Urbach, F., 223, 275 
Uzman, B. G., 255, 276 

Valentine, B. C, 112, 119, 

Van Bruggen, E. F. J., 112, 

113, 126, 144, 146 
Van Gool, A. P., 57, 98 
Van Harreveld, A., 9, 21, 48, 

Vanselow, W., 267, 276 
Van Tubergen, B. P., 221, 

257, 262, 264, 267, 269, 

Vergara, J., 37, 38, 45 
Vrensen, G. F. J. M., 237, 

240, 244, 264, 266, 267, 

Waldner, H., 54, 63, 64, 68, 

Wall, J., 129, 142, 144 
Waravdekar, V. S., 41, 48 
Wartiovaara, J., 91, 98 
Webb, S. J., 7, 49 
Wehrli, E., 67, 76-79, 97, 98 
Weinstein, B. S., 61, 71, 96, 

209, 216 
Welch, W. H., 102, 146 
Wen, W. Y., 5, 44 
Wenger, A., 6, 44 
Wentorf, B. H., 7, 49 
Westmeyer, H., 202, 217 
Westwater, T.W., 24, 25, 49, 

Wiebenger, E. H., 113, 146 
Wildy, T., 104, 137, 145 
Williams, C. B., 108, 146 
Williams, M. A., 241, 249, 

250, 254, 278 
Williams, B. C, 117, 127, 

133, 134, 143, 146, 151, 

Williamson, J. R., 222, 262, 

Wischnitzer, S., 110, 125, 146 
Wisse, E., 267, 268, 278 
Wold, F., 119, 147 
Wolf, E. D., 7, 45 
Wolf, K., 168, 193 
Wolman, M., 3, 49 
Woolgar, A. E., 9, 10, 44 
Wrigley, N. G., 125, 147 
Wyckoff, B. W. G., 151, 193 

Yada, K., 170, 192 
Yokoto, S., 37, 38, 46 
Yoshikami, D., 206, 215, 217 
Young, D. E., 41, 48 

Zahn, B. K., 157, 193 
Zeitler, E. H., 127, 143 
Zelazo, P. O., 105, 141, 147 
Zingsheim, H. P., 64-66, 95, 

98, 201, 203, 207, 208, 210, 

211, 215, 217 
Zotikov, L., 23, 49 
Zubay, G, 102, 145 

Subject Index 

Acetone, 21, 36, 38, 39 

Ammonium molybdate, 113, 114, 125, 

126, 132 
Antifreeze, 4, 8 
Antigen, 93 
Artifacts (Imperfections), 3, 4, 15, 37, 

52, 55, 57, 69, 71, 80, 83, 85-88, 108, 

110, 122, 126, 167, 169, 172, 175, 

179, 205, 255, 261 
Autoradiograms, 221, 222, 227, 236, 245, 

250, 254, 258, 260, 261, 266, 273 

Bacteria, 4, 11, 57, 61, 108, 109, 131, 

159, 163, 169 
Bacteriophage, 106, 119, 122, 123, 131, 

141, 163, 226 
Bedacryl, 166, 167 

Buffers, 113, 114, 116, 199, 205, 206 
Bushy stunt virus (BSV), 102 

14 C, 225, 226, 235, 237, 244 

Carbon, 64-66, 70, 71, 82, 110-112, 
157-159, 198, 199. See also Shadow- 
ing materials 

Carbon dioxide, 35 

Carbon-platinum shadowing, in freeze 
etching, 64 

Caspar method, 137, 138 

Cell debris, removal from replica, 67 

Cellulose acetate, 171, 174, 176, 180 

Cesium chloride, 108 

Chemography, 236, 239, 240, 266 

Chloroplast, 94, 95 

Chromatin, 41 

Cleaning agent, 67 

Collagen, 7, 86, 100, 173, 260 

Collodion, 107, 112, 164, 168, 221, 222, 
242, 244, 256-258, 260, 262, 272 

Condensation, 71, 86 

Conical shadowing, 157 

Contamination, 63, 71, 118, 123, 133, 
197, 199, 207-209, 211, 258 

Contrast, 64, 100, 102, 110, 114, 122, 
124, 127-134, 142, 150, 153, 157, 160, 
163, 164, 166, 167, 197, 200, 202, 259, 

Crosslinking, 4, 118, 122 

Cryostat, 35, 41 

Cytoplasm, 13-19, 21, 41, 62, 87, 88, 91 

D19, 228, 246, 267, 271 
Decoration. See Nucleation 
Dehydration, 4, 15, 21-23, 37, 38, 40, 

56, 58, 84, 102, 106, 117, 167, 171, 

Dektol, 226, 228, 239, 241, 269, 270 
Diffusion, 3, 4, 57 
Diffusion pump, 37 
Dimethylformanide, 4, 12 
Dimethylsuberimidate, 119 
Dimethyl sulfoxide, 4, 10, 12, 17, 56 
DNA, 106, 206 

Double replica, 79, 80, 82, 83, 169 
Drop method, 102, 114, 116-118, 122, 

Dry ice, 35, 39 

Electrolytes, 56 

low energy, 225-244 

high energy, 225-244 
Electron beam shadowing, 201, 211 

in freeze etching, 64, 65, 69 

in shadowing, 160-163, 200-202 
Embedding (embedded specimens), 22 

23, 36, 40, 85, 102, 113, 118, 122, 126, 

127, 129, 131-133, 136, 137, 171-173, 

175, 180, 189 
Emulsion, 221, 223, 225, 229, 236, 238, 

241-245, 259-263, 266 

liquid, 232, 256 

thickness, 223, 224, 230, 237, 243, 
259, 262, 266 

layers, 223, 224, 240 
Endoplasmic reticulum, 14, 16, 19, 21, 

Enzymes, 67, 108, 113, 114, 116, 119, 

121, 127, 136, 140, 208, 213 
Etching, 54 

Ethenol, 21, 22, 36, 39, 167 
Ethylene glycol, 4, 12, 17, 30, 31, 56, 57 
Evaporation, 72, 87, 88, 102, 111, 112, 

154, 159, 162, 163, 168, 198, 199 
Extraction, 254, 255 

Ferritin, 92, 93, 183, 187 
Fibers, 100, 185, 187, 190 
Fixation, 3, 4, 38-41, 57, 118, 119, 122, 
175, 177, 178, 180, 255 



Flat substrate, 222, 243, 255-257, 260, 

262, 264, 272 
Float method, 118, 135 
Fluorocarbons, 24, 26, 27, 30-32 
Formaldehyde, 22, 39, 40 
Formvar film, 14, 27, 34, 38, 67, 112, 

122, 164-167 
Fracturing, 54, 59, 71, 76, 79, 80, 82, 

85, 86, 90 
Freeze damage, 11, 13, 20, 22 
Freeze drying, 5, 22, 36-38 
Freeze-etching, 202, 204, 206, 208, 209, 

211, 213 

definition of, 161-163 
Freeze fracture, definition of, 71 
Freeze-protective agents, 4, 10-12, 17, 

18, 22, 23, 34, 37, 41, 55-58, 63, 84 
Freeze replication, 5, 21 
Freeze substitution, 5, 17, 22, 32, 38 
Freeze thawing, 40, 41 
Freezing point, 27, 58 
Freon, 24, 26, 28, 30-33, 35, 41, 58, 76 
Frozen cells, 5, 7, 9 
Frozen sectioning, 254 

Gelatin, 6, 34 

Genetron, 23, 31, 32 

Gevaert, 262, 264, 266 

Glutamine synthetase, 108, 109, 119, 

120, 122, 135, 136 
Glutaraldehyde, 22, 39, 57, 91, 118, 119, 

Glycerol, 6, 7, 10-13, 17-20, 25, 27, 42, 

56-58, 80, 84 
Glycogen, 19, 91 
Gold-EAS, 228, 249, 269, 271 
Gold film. See Shadowing materials 
Gold latensification, 238, 267, 241, 268, 

269, 271 
Golgi, 17, 19, 91 
Grain, in autoradiography, 221, 222, 224, 

226, 227, 229-235, 238, 239, 245, 

246, 249-251, 253, 254, 261, 262, 267, 

268-270, 272-274 
Grids, 58, 65, 67, 102, 103, 108, 111, 

112, 114, 118, 119, 121-125, 127, 

132-135, 167, 169, 171, 221 

3 H, 225, 237, 247-249, 273 

3 H-thymidine, 226 

Heavy metals, 100, 102, 129, 197 

HeLa cells, 18 

Helium II, 23, 267. See also Quenching 

High resolution, 37, 109-111, 121, 123, 

131, 142, 154, 157, 190, 197, 198, 202, 

204, 221 
Holey film, 133 
Homogenization, 3 
Hydrocarbons, 24, 26, 27, 30-32 
Hydrogen, 23 

Ice, 5, 6, 8-10, 19, 21, 27, 32, 33, 35, 

37, 38, 40, 56, 57, 62, 63, 68, 84, 91 
Ice crystallization, 4, 7-20, 37, 41, 54- 

57, 83, 84 
Ice lattice theory, 5, 6 
Ilford L4, 225, 228, 230, 238, 239, 241, 

244, 246-248, 260-264, 266, 267, 

269, 271 
Incubation, 57 
Interference colors, 241-245, 258, 259, 

262-264, 266, 272 
Intracellular radioactivity, 221, 255 
Isobutane, 26, 30 
Isoelectric point, 112-116 
Isopentane, 25, 26, 30, 31, 41 
Isopropyl N-phenylcarbamate, 13 

Kidney, 11, 41, 113, 246 

Knife, 23, 54, 61, 63, 69, 77, 85, 179 

Kodak NTE, 226, 228, 230, 238, 239, 

241, 244, 249, 262, 264, 266, 267, 


Latent image, 223, 224, 236, 238-241, 

Lead staining, in autoradiography, 259- 

Lipids, 91, 92, 254 
Liquid nitrogen, 24-29, 31-33, 35, 38, 

39, 41, 54, 58, 59, 61, 63, 68-70, 72, 

73, 80, 82, 86, 112, 133, 154, 206, 

Liquid oxygen, 24, 42 
Liver, 8, 11, 15, 17, 19-21, 24, 41, 83, 

107, 125, 166 
Loop method, 262, 264 

Macromolecules, 100, 109, 114, 116 
Membranes, 10, 11, 14, 15, 40, 41, 62, 
76, 78, 79, 83, 86-88, 90-95, 100, 

108, 113, 119, 176-182, 185, 186, 250 
"Membrane tchnique," 222 

Mercury, 23 

Methacrylate, 225, 237, 238, 258 
Mica, 111, 112, 128, 189, 199, 214 
Microdol X, 241, 248, 267-269, 271 
Micromelting, 65 

Microtome, 54, 59, 63, 68, 69, 76, 77, 
79, 179 


Microtubules, 12, 13, 16, 90, 91 
Mitochondria, 14-16, 18, 19, 21, 41, 91, 

94, 113, 260, 261 
Mitotic apparatus (MA), 12, 13 
Mitosis, 16 
Moon dust, 205 
Multiple hit, 238, 239 
Muscle, 9, 91, 100, 108, 168, 173-177 
Myelin, 95, 213 
MX, 228, 261 

Napthalene, 67 

Negative staining, 91, 95, 100-142 
Nitrogen, 9, 23, 35, 69 
Nomarski interferometer, 241, 244, 259 
Nucleation (decoration), 198, 212, 213 
Nucleic acids, 8, 100, 157, 254 
Nucleoproteins, 100 

Nucleus, 14, 15, 17, 41, 63, 85, 87, 88, 

Oscilloscope, 27, 28, 30 

Osmium tetroxide, 22, 36, 39, 40, 41 

Osmolarity (tonicity), 113 

Pancreas 8, 11, 14, 15, 17, 18 

Paraformaldehyde, 40 

Parlodion film, 67 

Particulates, 67, 86, 89, 95, 100 

pH 4, 102, 103, 105, 108, 112-118, 120, 

123, 139, 268 
Phase contrast, 110, 123, 129, 134 
Phase transition, 6 
Phosphotungstic acid (PTA), 100, 102, 

113, 123, 131, 244, 264 
Phosphotungstate, 102, 106, 112 114, 

125, 129-132, 134 
Photographic process, 222, 223, 225, 

229, 230, 259, 260, 272 
Plant cells, 93 
Plasmalemma, 93, 95, 113 
Platinum, 64-66, 70, 82, 153, 157-159 
Pollen grains, 27, 169 
Polywater, 7 
Polyvinyl alcohol, 170-173, 175, 177, 

180, 187 
Porcelain, 156, 157 
Portrait technique, 157 
Post mortem changes, 57 
Potassium phosphotungstate (KPT), 102- 

104, 113, 117-122, 132 
p-Phenylenediamine, 228, 247, 269, 271 
Pressure, 37 

Propane, 24, 26, 30-32, 41, 42 
Propylene, 31 

Protective agents. See freeze Protective 

Proteins, 4-8, 12, 27, 100, 102, 106, 108, 

112, 116, 120, 129, 131, 135, 136, 

138, 140-142, 254, 255 
Pumps, 72, 86, 87, 111, 112, 154, 163, 


Quantasomes, 94 

Quantitative autoradiography, 221-223, 

239, 241, 245, 255, 256 
Quenching baths, 8, 23, 24, 26, 27-35, 

41, 42 

Radioactive material, 221, 226 

Radioactivity, 222-224 

Rapid freezing, 4 

Red blood cells, 7, 10, 11, 55, 57, 79, 

80, 87, 88, 92, 93, 113, 174-181 
Replication, 54, 55, 59, 61, 63, 64, 66, 

67, 69-72, 76, 79, 80, 82, 86, 87, 89, 

170, 171 
Resistance heating, 202, 211 
Resolution, 66, 100, 110, 124, 128 

in autoradiography, 221, 223, 225, 
227, 231, 232, 234, 235, 243, 244, 
Ribosomes, 16, 91, 100, 108 
RNA, 106, 206 
Rosin, 31 
Rotating apertures, 210 

^S, 244 

Salt, 10, 18 

Seawater, 12 

Sections, 55, 60 

Sectioning, 68 

Section thickness, 224, 225, 237, 240- 

243, 259, 263, 274 

Sensitivity, 223, 236-238, 240, 241, 243, 

244, 266, 269, 274 

Shadowing, 64-66, 70, 71, 82, 87, 89, 

90, 100, 150, 163 
Shadowing, angle of, 154-156, 159, 190, 

206, 213 
Shadowing materials 

Carbon, 160, 162, 163, 165, 167-169, 
171, 185, 199, 200, 202, 205, 214, 
222, 256, 260, 262, 272 

Chromium, 153, 158 

Gold, 153, 157-159, 212, 213 

Iridium, 160, 202, 204, 205 

Molybdenum, 203 

Palladium, 153, 157-159 

Platinum, 153, 157-159, 171, 184, 
189, 200, 204, 213 


Platinum-carbon, 157-159, 160, 163 
171, 172, 202, 204, 208, 210, 211, 
Platinum-palladium, 211 
Tantallum, 158, 160, 201, 202 
Tantallum-tungston, 199, 203, 204 

206, 210, 211, 214 
Tungsten, 158, 160, 211, 201, 203, 204 
Uranium, 160 
Shrinkage, 19, 22, 56, 84 
Silver halide, 54, 223, 225, 230, 238, 

243, 244, 262, 264, 266 
Single-stage replica, 166-168 
Sodium phosphotungstate, 113 
Sodium silicotungstate, 113, 125, 129 

Circular, 231-233, 252-254 
Extended, 222, 229, 231 
Line, 231, 233, 236 
Point, 222, 226, 227, 231-233 
Specimen, 60, 61, 63-65, 68-74 76 77 
79, 80, 82, 84-87, 110, 112, 116, 122,' 
125, 126, 128-134, 142, 150, 153-157, 
160, 165, 167, 168, 172, 176, 183, 
190, 197-199, 201, 205, 209, 210, 214 
Specimen stage, 61, 63, 68, 69, 79 
Speciment support film. See Support film 
Spermatozoa, 55, 58, 84, 94, 178 182 
Spleen, 10, 11 
Spores, 57 

Spray gun, 117, 127, 135 
Spray method, 117, 121 
Positive staining, 40, 100 
Stripping film, 222, 262, 272 
Sublimation, 112 
Sublimation, ice of, 54, 61-63 67 71 

133, 200 ' 

Substitution fluid, 35, 36 
Sucrose, 11, 15, 19, 113, 116 
Support films, 34, 38, 54, 64, 65, 100, 
102, 103, 110-112, 114-116, 118, 121 
123, 124, 127, 128, 131-137, 152, 159 
197, 199, 200, 202 

Tendon, 180, 181, 186 
Tetrahydrofuran, 21, 36 
Thawing, 55, 56, 66, 113 
Thermocouple 24, 27-29, 33, 35 58 69 

71, 75 ' ' 

Tobacco mosaic virus (TMV) 102 130- 

132, 134, 136, 140 
Tritium, 225, 230, 236-238, 244 
Tungston, 65, 199. See also Shadowing 

material, tungston 
Tungston filament, 154 
Turnip yellow mosaic virus (TYMV) 

132, 133 

Unfixed tissues, 40, 57 91 171 174 

175, 180, 184 ' 

Uranium salts, 100, 105, 129 
Uranyl acetate, 107, 108, 112-116 119- 

122, 126, 130-132, 260, 272 ' 
Uranyl formate, 114, 115, 129-132 
Uranyl oxalate, 112, 114, ,115, 120-122 


Vacuoles, 17, 18, 21, 246 

Vacuum, 15, 22, 23, 36-40, 42 54 58 
60-62, 65, 66, 68, 69, 71-73 75 77- 
80, 82, 86, 87, 110-112, 154, ' 162 
167, 168, 189, 197, 199, 206-209, 256 

Vapors, 22-26, 36-40, 42, 63 70 86 
112, 133, 198, 200, 207, 209 211 ' 

Virion, 106, 132 

Virus, 54, 69, 100, 102-106, 108-110 
112-114, 116, 117, 119, 122, 127 
130-132, 135-138, 140, 150, 163, 169 
197, 213 

Vitrification, 36, 41 

Water, 5-8, 10-14, 16, 17, 19 23 27 
33, 36, 37, 40, 57, 65-67, 86, 'ill,' 169 
Wire loop, 221, 257, 262 

Yeast, 4, 8, 11, 57, 75, 91, 206, 207 


The production of a fine metal deposit by the use of an electron beam 
source is also aided by the fact that there is no decrease in vacuum 
during the evaporation procedure, a characteristic of the normal filaments 
when heated. 


The technique of electron beam evaporation uses the kinetic energy of 
the electrons to produce the heat necessary to evaporate the metal or 
carbon. The material to be evaporated, which acts as the anode, is 
placed in a crucible and the electron beam focused on it. Electrostatic 
or magnetic fields can be used to focus the electron beam, but the elec- 

Fig. 4.10 Balzers vacuum coating unit, a small compact instrument which em- 
ploys both conventional and electron beam source evaporation facilities. 




; •■ ; 


■.■■.■■ ■: ,:: - ^ 






Fig. 2.6 NRC ultrahigh vacuum unit utilized for freeze-fracture work. Bell jar 

1 cm long provides convenient specimen holders for biological material. 
The top of the sliding block system has evaporation tunnels and a razor 
blade mounting. The blocks are charged with a specimen under liquid 
nitrogen, assembled and introduced into the vacuum chamber. After 
pumping a vacuum, the top block is pushed over the specimen thus 
fracturing it and aligning the evaporation tunnels for replication. At the 
moment, the linear motion feed-through seal is the only element of the 
system which utilizes high vacuum grease. It is hoped that this com- 



■?:■■■ ; ■•...: 


! 5"* f : "" 

Fig. 2.7 Detailed view of the electrode assembly (Ladd Research Corp.) and 
freeze-fracture block in place in the NRC unit of Fig. 2.6. 

: ■■ . ■ ■ - . :■■■ 

.:,■ ■ "■■ •• ■■•■•■ ,:■!■■ i' : .,:-.■■' ' : ,':■ ,' . : ,-. : 


Fig. 2.8 The double block freeze-fracture unit disassembled. Note razor blade 
mounted on upper block and specimen tube (arrow) in lower block. 


-U H 

:::::.:;..' ; : ,:: 

Fig. 2.2B Schematic representation of the Steere unit. B: stainless steel specimen 
tube; C: copper or brass shroud; D: stainless funnels for liquid nitrogen supply; 
J: copper specimen stage; Q: electrodes for shadowing and carboning; S: scalpel 
for fracturing. See reference for full description. Courtesy of Dr. R. L. Steere 

which permits a stainless steel knife to make one pass through the frozen 
specimen prior to replication. This system has the advantage of sim- 
plicity and reproducibility, but suffers from a lack of flexibility in that 
the operator cannot, if he chooses, make more than one attempt to frac- 
ture a given specimen. The fractures are, furthermore, all of the same 
quality in terms of coarseness. Some specimens are displayed more 
advantageously if rather fine increments of fracture thickness are em- 
ployed. This is especially true if one wishes to obtain cross fractures 
through very small cellular entities such as bacteria. The coarse cutting 
most frequently gives surface reliefs of such objects rather than cross 

The Type II block freeze-fracturing apparatus ( Bullivant et al., 1968 ) 
requires fracturing the specimen under liquid nitrogen with a cooled 
razor blade prior to transferring the assembled block apparatus to the 
vacuum chamber. While providing a very simple and rapid way of ob- 
taining a fractured specimen, this procedure suffers from some of the 
same constraints mentioned above. 

Etching (Ice Sublimation) 

After completing the fracturing of the specimen, the topography of the 
surface should reflect the upward or downward excursions caused by 
preferential deviations from the theoretical fracture plane. Thus, final 
image contrast in the replica would be limited to those structures which 


Fig. 2.3 The Ebtec Corporation's freeze-etching unit patterned after the McAlear- 
Kreutziger design (1967). The base-plate and electrode assembly is permanently 
installed in a vacuum evaporator. The cylindrical freeze-etching module is de- 
mountable. Courtesy of the Ebtec Corporation. 

caused deviation from the theoretical fracture plane by the presence of 
membranes or other substances less easily cross fractured than the gen- 
eralized cytoplasmic matrix. Additional information can be obtained by 
allowing some of the ice to sublime away under vacuum conditions in 
order to expose intra- or extracellular substances which would otherwise 
be masked from view. This process has been termed etching, hence the 
name freeze-etching. This is to be contrasted with the procedure of 



Biological Applications, Volume 1 

By M. A. Hayat, Associate Professor of Biology, Newark Stale College. Union, N. J, 
4l2pages, illustrated, 6x9. / 

Helps giveja sound working know (edge of the essential-biochemical concepts under- 
lying modern pfeparatory procedures for electron microscopy. Fully details the 
chemistry of the intorac-.'on of important i ■': agents wiih cellular substances to permit 
accurate interpretations of the electron micrographs of any specimen which has 
been subjected lo fixation, dehydration, embedding, sectioning, staining, and elec- 
tron bombardment. Many new techniques are explained. 

By M. A. Hayat. 144 pages, 5Vj x 8Yj, illustrated, 

Answers the Jong-felt need for a time-saving handbook ol basic, step-by-step pro- 
cedures involved in sparing biological specimens tor electron microscopy. De- 
■scribes newest, simplest, and best methods in fixation, embedding, sectioning, 
staining, tissue storage, and the preparation ol buffers. Included here — Out not 
found in other available source books — are procedures for isolation of certain 
cell components; perfusion fixation of lung, liver, and muscle; staining with iodine, 
silver, thorium, iron, vanadium, indium, lanthanum, alcian blue, and ruthenium red. 


Edited by William J, Adelman. Jr., University at Maryland School of Medicine. 

528 pages. 190 illustrations, 6x9. 

Eighteen contributing authors — each an expert in his field — present different 
methods for studying excitable membranes, the book coders, for instance, voltage 
clamp techniques; modeling concepts as they apply to any field of science; the 
analytic approaches used in low angle x-ray crystall&graphic analysis of lipid films; 
aid computer reconstruction of membrane' currents and the nerve impulse. This 
last section includes complete FORTRAN programs, from which direct access is 
provided to analytical solutions of Hodgkin-Huxley nerve equations. 


ACS Monograph Series. 373 pages. 6 x 8V u, illustrated. 

Provides an in-depth review of the biochemistry or bacterial viruses, giving par- 
ticular emphasis to the metabolism of the virus-infected cell. Presents supporting 
information in related areas of phage structure, viral genetics, and microbial phys- 
iology. Covers T-even and other phages: sigma factors and the control of trans- 
scription; effects of phage infection on cell membranes; and much more. 





> CD 


o «a 

o .— 

Van Nostrand Re|nhoJd Company 
450 West 33rd Street, New York 10001