- Email: [email protected]
The physical behavior of solid water at low temperatures and the embedding of electron microscopical specimens
Ultramicroscopy 16 (1985) 151-160 North-Holland, Amsterdam
151
T H E P H Y S I C A L B E H A V I O R OF S O L I D W A T E R AT L O W T E M P E R A T U R E S AND T H E E M B E D D I N G OF E L E C T R O N M I C R O S C O P I C A L S P E C I M E N S H.-G. H E I D E and E. Z E I T L E R Fritz - H a b e r - l n s t i t u t der M a x - P l a n c k -Gesellschaft, F a r a d a y w e g 4 - 6, D - I 0 0 0 Berlin 33, G e r m a n y
Received 19 December 1984
Modifications of solid water and their transitions are described as they relate to cryo electron microscopy. In particular, the various amorphous states (amorphous polymorphs) as they exist below 100 K are extensively investigated. The "high-density" modification exhibits a lower viscosity than the "low-density" form. Differences are also observed in the mechanism of void formation due to electron irradiation: in the high-density form, voids are formed - not, however, in the low-density form. Together with the reaction to radiation damage, the physical properties of amorphous solid water are discussed with respect to embedding of organic specimens. Finally, the conditions and pitfalls associated with preparation of thin and entirely vitrified ice layers by shock-freezing are described.
1. Introduction It has been the dream of biologists to embed their specimens in amorphous solid water, thereby preserving them in their native state, whose structure can then be revealed by electron microscopy. After early attempts it became obvious that more of the physical nature of ice and its behavior had to be known before a successful procedure could be devised. As soon as our cryo objective, allowing specimen temperatures from 6 to 300 K, was completed, we embarked upon the investigation of ice layers. Although this study is far from complete, we feel that the results obtained thus far are worthwhile for the practitioner, because consideration of the facts found will assure successful specimen preparation. Furthermore, some of these results might be of interest in water research. Two methods for preparing ice layers of, say, 10-100 nm thickness have been employed namely, condensation of water vapor on carbon films within the cryo objective at various rates, and shock-freezing of thin water layers outside the microscope and subsequent cryo transfer. It is this second method which is commonly attempted, with more or less success, and we shall lay down at the end of this paper the rules and recommendations
derived by reasoning based on the measured physical data. In the following, for lack of a simple and convenient term for solid amorphous water, we sometimes call it "ice", although in a strict sense this should be used only for crystalline water. Whenever the structure is important, we shall use the appropriate term.
2. Four ice modifications in four temperature ranges Ice can exist in twelve different modifications, some only under extreme conditions. Under electron microscopy conditions, only four modifications prevail: hexagonal crystalline ice [Ih], cubic crystalline ice [Ic], amorphous low-density water [H20(as, 1)] and amorphous high-density water [H20(as, h)]. But only the vitreous amorphous kinds are desirable, because here those forces which can occur during crystallization and may damage the specimen are absent. The two amorphous forms (low density: 0.94; high density: 1.1 g / c m 3) were first described by Narten, Venkatesh and Rice [1], and later by Sceats and Rice [2]. Polymorphic transitions can occur at various temperatures, with or without the influence of
0304-3991/85/$03.30 © Elsevier Science Publishers B.V. (North-Holland Physics Publishing Division)
152
H.-G. Heide, E. Zeitler / Phys'ical behavior of solid water at low temperatures
irradiation. The transitions themselves can be investigated quantitatively by means of electron diffraction. Results of such investigations - for example, the electron doses required for effecting changes - have been reported in a previous paper [3]. The important facts are briefly reiterated in this introduction. Above 100 K, crystallization of an amorphous layer, that is, the transition H20(as, 1) ~ ice Ic, sets in, without irradiation. The time this transition takes becomes shorter at higher temperature; at 160 K it occurs practically instantaneously. Any electron irradiation shortens the transition times. The reverse reaction, that is, the amporphization of crystalline ice
Table 1 Ice in four t e m p e r a t u r e ranges T (K)
8-20
as,h
•
QS,,~
•
[c
•
[h
•
30 -70 0\
80 - 9 0
> 100
o
.)
i
°) i
,¢'
•
e- irradiated: as, h
Ic [h behavior ; as, h
f,v
QSj "~
lc
ice Ic --, H20(as, h),
f
(v)
(f I , ( n )
(f),n r
V
r
V
can take place below temperatures of 70 K; however, only when electron radiation is applied. The doses required for this transformation become smaller the lower the temperature. At 8 K, already 2 e / A 2 suffice to effect this transition. Under the same conditions low-density water also changes into high-density water,
Stable or m e t a s t a b l e m o d i f i c a t i o n s are m a r k e d by solid circles in the broad t e m p e r a t u r e range at which they exist. Open circles denote the unstable forms. A r r o w s indicate the transitions between states. Flow behavior and the formation of voids u n d e r irradiation are denoted by the following abbreviations: f - flowing; r rigid; v void formation; n no voids. For details see text.
H 2 0 ( a s , 1)--, U20(as, h);
(indicated by arrows) takes place. The following three sections are subdivided in the same manner.
and, for example, at 30 K a dose of 5 e/f~ 2, quite common in practical electron microscopy, completes the transition. The reverse transition, H20(as, h) --, HzO(as, 1), sets in spontaneously with or without irradiation, when the specimen is warmed to at least 70 K. After having mastered the preparation, identification and transformations of the various ice forms, we could proceed to study their behavior and peculiarities in the microscope. Conspicuous differences were found in the mechanical flow behavior of the various ice forms and their disposition to form voids during electron bombardment. For presenting our results (see table 1) we have chosen four temperature ranges such that by changing from one range to the next only one event - that is, one transformation of ice into another form
3. Structure
3.1. 8 to 20 K Without irradiation, all four polymorphs are stable (or metastable) in this temperature range. By stable, we mean that the forms can survive for at least several hours without any change. Under irradiation, however, every form changes immediately into HzO(as, h), which in turn remains unchanged for several hours, with or without radiation, suggesting that this form might be the intrinsically stable modification of water at those very low temperatures. Although we have only measured to 8 K, we believe that the observed data would also hold at 4 K.
H.-G. Heide, E. Zeitler / Physical behavior of solid water at low temperatures
3.2. 30 to 7 0 K
In this transitional range the phenomenon of an amorphous polymorphism appears obviously. Without irradiation, Ic and H20(as, 1) remain stable (or better, metastable). H20(as, h) transforms spontaneously into the low-density form with increasing temperature to a larger fraction. If radiation is applied, the low-density form converts into the high-density form - the more, the lower the temperature. The diffraction diagrams show certain intermediate states between high density and low density. They can also be observed in highdensity water which was produced by condensation at 30 K or in irradiated crystalline ice (Ic) at 20 to 60 K. In the Discussion, section 7, we shall come back to these intermediate states and offer an explanation within the frame of proposed structure models. 3.3. 80 to 90 K
Only in this temperature range are both Ic and H20(as, 1) stable during long times of observation and under assault by higher radiation doses ( > 1 0 - 6 e n m - 2 ) . The high-density form does not exist. These facts are important to know if one is to check the success of the vitrification of a preparation. (At lower temperatures the Ic will be amorphized almost instantaneously, whereas at higher temperatures an amorphous layer will crystallize). 3.4. 100 to 160 K
Above 100 K only crystalline ice is stable (Ic metastable in comparison to hexagonal ice).
4. Flow behavior
So far we have been able to discern the various amorphous solid states of water only by means of their diffraction patterns, but the observation of the motion patterns points up distinct differences which offer an aspect to characterize the physical behavior of the various modifications. The find-
153
ings are described for the same temperature ranges as in the preceding section on Structure. 4.1. 8 to 20 K
The first observations were made on thin selfsupporting films of H20(as, h) which showed holes or cracks. If the rim of such an opening is irradiated by a fine electron probe, the resulting electric charging then produces forces which, in turn, cause growth of little icicles (fig. 1). This growth of icicles is remarkably vigorous. A rapidly flowing transport of ice, sometimes showing fine surface waves, to the tip of the icicles can be made out. These icicles may vibrate and even tear off. The whole scenario of movement appears utterly blithe. Similar observations can be made on H20(as, h) layers supported on films, if these have a coarse granular structure. If such layers are irradiated, all the contours flow together, the grains disappear, and the irradiated area assumes a uniform grey appearance. 4.2. 30 to 70 K
In this transitional temperature range the flow behavior remains as the one just described, but at around 70 K the flow seems to be somewhat retarded. 4.3. 80 to 9 0 K
Under electron bombardment icicles grow on the rims of self-supporting films of H20(as, 1) also. In contrast to the earlier-described movements, this flowing is viscous, sluggish and slower. The icicles are not as fine, and the surface granulation appears more rough. In the many experiments the impression is intensified that the amorphous highdensity form flows more readily, like a liquid, than the amorphous low-density form. Also, in layers on supporting films, the disappearance of a coarse texture or the closing of holes occurs a little more slowly in the amorphous low-density than in the high-density form. Cubic ice does not flow in this temperature interval. Even bizarre and sharp-edged forms re-
H.-G. Heide. E. Zeitler / Physical behavior c (sohd water at low temperatures
154
Fig. 1. "Icicles" as they grow at the rim of thin amorphous solid water layers driven by charging in the beam. (a, b) H20(as, h) at 9K, very vigorous. (c, d) HzO(as, 1) at 75 K, somewhat more sluggish and viscous flow. Magnification 16,000 ×.
main unchanged in the electron beam (although mass is lost). 4.4. 100 to 160 K
In this range ice Ic does not flow either.
tion calculated under the assumption that the primary processes are direct knock-on impacts. The role of electron excitations in the bulk of the ice is uncertain. The foamy bubble formations which sometimes occur in wet organic specimens or aqueous solutions are unrelated to the void formation in pure ice (see section 7).
5. Void formation 5.1. 8 to 20 K
The production of cavities by radiation, mainly in metals, has been an object of study for the last twenty years. The same radiation damage occurs also in ice [4]. The mechanism for this phenomenon is understood as follows: The atomic displacement processes caused by the primary irradiation produce Frenkel defect pairs whose interstitials migrate to the surface, whereas the vacancies agglomerate and form voids. The experimental cross-section agrees with the theoretical cross-sec-
In amorphous high-density water void formation is very pronounced. Dose values between 500 and 1000 e / A 2 produce voids with irregular shapes and various sizes. With prolonged irradiation the size and number of the voids increase (see fig. 2), but only to such an extent that the distance from one void to another, or to the surface, is greater than about 200 A. (In a 100 A thick layer the void formation has never been seen, because the
H.-G. Heide, E. Zeitler / Physical beha~ior of solid water at low temperatures
155
Fig. 2. Void formation in HzO(as , h) at 9 K caused by irradiation with 100 kV electrons. (a) (h) 20, 1000, 1700, 2700, 3600, 4600, 5500, 6700 e/,~ 2, respectively. Initial layer thickness ca. 1000 A. Magnification 90,000 X.
156
H.-G. Heide, 1:2 Zeitler / Pt(vsical behat,ior of solid water at low temperatures
vacancies migrate directly to the surface.) After an irradiation of 1600 e / A 2 the voids assume a spherical shape, and their number per volume no longer increases. The total void volume amounts to about 3% of that of the ice layer. We have calculated that at 100 keV one elctron produces about 5 × 10 ~7 Frenkel pairs per cm 3. From this (and the molecule density of 3.6 × 1022 H 2 0 / c m 3) follows that an exposure of 2000 e/,~ 2 displaces 3% of all H 2 0 molecules. More radiation does not produce any additional voids. The ice layer itself is etched away by the electron beam at a rate of 80 , ~ / C cm 2 [3]. During this process voids are released through the surface, but without exhibiting any indentation. One might surmise that as soon as the bubble breaks, the remaining layer flows together to form a plane surface. As the void's diameter can never be larger than the thickness of the layer, the size of the voids reduces during observation in the same way as does the thickness of the ice layer. 5.2. 30 to 7 0 K
In this transitional region the void formation is seen rather irregularly - sometimes as single voids, sometimes as colonies. The void formation seems to become rarer to the same degree that amorphous high-density water transforms into the low-density form when the temperature is raised. 5.3. 80 to 90 K
In a pure H20(as, 1) layer, void formation does not occur. Irradiation just removes the ice with the well known uniform manner and rate. In crystalline ice layers, however, void formation has been observed. 5.4. 100 to 160 K
If the ice layers consist of large enough single crystals, these crystals also show void formation. Should the layers be composed of small microcrystals with many surfaces, however, void formation does not occur. Neither does void formation occur if the nucleation is prohibited: this is the case in very pure and fresh crystals, but not in condensed layers which always contain foreign rest
gas molecules. Furthermore, we assume that in ice crystals, just as in metals, the presence of dislocations will influence the migration and agglomeration of point defects.
6. Radiation damage The radiation damage in ice which results in a loss of matter has recently been described extensively [3]. It could be shown that amongst all possible mechanisms, those ionization processes which take place in the surface layer contribute predominantly to the removal of the ice. Below 90 K the rate of removal is constant, amounting to one monolayer H 2 0 per dose of 25 e / f ~ 2 at 100 kV. The question of whether the removal rates in amorphous and in crystalline ice are the same could not answered experimentally up to now. The difficulty stems from the fact that the actual surface in a grainy microcrystalline layer cannot be readily controlled, whereas the surface of an amorphous layer is a rather defined entity. However, after we succeeded in preparing larger plates of single crystals, we could perform the comparative measurements, with the following results: At temperatures between 75 and 90 K, the mass loss rate, through electron irradiation, for crystalline ice (lh) is about two-thirds that for amorphous solid water. The rate seems to depend on the crystallographic orientation of the ice surfaces. More exact measurements are hampered by the migration of bend contours across the irradiated field of obserw~tion. Because of its great importance for ice embedding of specimens, one observation should be mentioned, and this is in connection with carbon film [3,5] since no other material has been used: When the ice is removed entirely by the electron beam, the carbon film support also shows radiation damage (about 6 ,~ thinning at 9 K). When the removal is effected partly by electrons and the remainder by thawing, the carbon support shows no radiation effect. We interpret this finding as the action of radicals which can react only when they are free and in contact with the carbon film. Fixed within the bulk of ice, they cannot escape. This etching effect depends, of course, on the
tL-G. Heide, E. Zeitler / Pltvsical behavior of solid water at low temperatures
support to be etched. More experiments are required, and Talmon [6] has performed some in this direction. Based on our observations, two points should be made. One must clearly distinguish between ice embeddings which are completely sealed and those which have gaps or interfaces (for example, between microcrystals) or incomplete coverage of particles protruding from a thin layer. Such specimens might be "frozen-hydrated" but not truly embedded. The second point is that in amorphous solid water irradiation cannot create new ice surfaces: because of its viscosity, the solid amorphous water will always close the gaps around particles. This can be seen, for example, when small polystyrene latex spheres are embedded in amorphous solid water at about 10 K. As long as they are completely embedded, no shrinkage of the particles can be observed even at radiation doses high enough to produce a marked mass loss (increase in density and decrease in volume) when those particles are n o t embedded.
7. Discussion
The various behavior patterns of the two amorphous forms of solid water are interesting in many aspects. For example, it is surprising that in the amorphous high-density water voids are formed below 30 K, whereas in the low-density species no such voids occur, although the temperature is higher (70 to 100K). These findings pose the question about the mechanism which, below 30 K, permits the migration of defects and nucleation for cavity clusters. Also, why are these voids stable in the rather fluid matrix of high-density amorphous water? On the other hand, what hinders the mobility of defects and the agglomeration of cavities (or promotes recombination) in the low-density amorphous water? In contrast to that, voids are formed in cubic ice at the very same temperature (ca. 85 K). (Perhaps here the thermal mobility of OH ions helps to explain the void formation.) The observed differences in the flow behavior of the amorphous solid water begs the question of whether the structure proposals can explain the
157
difference in the viscosity of the two amorphous forms. For both amorphous forms of solid water, structure models have been given [1,2] which are plausible and compatible with X-ray diffraction data. In both cases the nearest-neighbor distance between oxygen atoms is 2.76 A, and on the average in a tetrahedral coordination. If one starts with the structure of Ic but a greater variation for the second-neighbor distance distribution ( - 4 . 5 A), one arrives at a randomized network of hydrogen bonds, which is characterized through minor distortions of the hydrogen bonding the distortions stemming from the variation of the angle ( - 1 0 9 °) between three oxygen molecules. This model is called the randomized network model (RNM), and it is considered valid for HzO(as, 1). A model for the H20(as, h) can be derived readily from the above model by putting, in about every second cavity position, an "interstitial water molecule" into the ice lattice. In addition to the 2.76 and 4.5 A distances, a set of second O - O distances of 3.25 A and a significantly smaller than 109 ° O - O - O angle results. A corresponding structure model (RNM) can also be derived for amorphous water (as, h) as a randomized version of ice II or ice III. Most remarkable is the close similarity of the structure to that of liquid water. From the considerations about the observed intermedate stages between (as, h) and (as, 1) (section 3.2), a very likely explanation follows. One merely has to assume that varying fractions (0-45%) of cavity positions are occupied by interstitial water molecules. In any case, the observed "liquidity" of the high-density water seems to parallel its structural similarity to water at room temperature. The consequences for ice embedding are not clear in all aspects as yet. Certainly the remarkably low viscosity of both amorphous solid waters is advantageous for the avoidance of dangerous mechanical stresses which, in turn, would lead to artefacts. Furthermore, the capability of flowing prevents the creation of gaps around the embedded particle, thus enhancing the radiation protection. Although it is very unlikely, it is not known whether the transformation from low-density to high-density amorphous water will injure the embedded
158
H.-G. Heide, E. Zeith, r / Pto,stcal hehat#or ~f sohd water at low temperatures"
specimen. Another important point would be the clarification of whether the conditions which further or prevent void formation also affect the ice-embedded specimen. At this point it is important to control the formation of voids with the often observed "bubbling", an event which occurs in aqueous solutions but not in pure water, at various temperatures. For example, in a vitrified solution of glycerol in water such bubbles are formed at 80 K after a very short irradiation. They grow quickly and finally form a very dense network. The vigorous foaming reflects probably a viscous flowing under the driving force of gaseous radiolytic products. After all the water has been lost by radiolysis, a honeycomb-like skeleton of carbonaceous material remains. At very low temperature, where high-density water prevails (say at 10 K), only a few bubbles can be observed. and these disappear under prolonged irradiation. Possibly this is another example for the difference between the two amorphous forms. However, it could also indirectly be due to a difference in the separation between water and solute.
plunged into a coolant by means of a guillotine-like device (fig. 3). One problem arising here is to choose the right moment to release the catch for the tweezers to fall. The moment is right when the water layer has, by evaporation, reached a thickness that is thin enough to be transparent to the electrons in the microscope but not completely dried up. This difficulty can be coped with by applying the method of Adrian et al. [8]: A droplet of water or of the suspension to be investigated is applied to a naked copper grid and blotted off with blotting paper. Observation under a stereo microscope of the water film's evaporation shows the film to rupture. The right instant for triggering the guillotine is that moment just before the rupture occurs. If suspensions or solutions are to be applied, they should have the lowest possible concentration, and the blotting should be controlled carefully. Important is that the copper grids have a hydrophilic surface. According to our experience, chemical cleaning with solvents or a short etching in an ammonia/water/alcohol mixture are most recommendable.
8. Preparation by shock-freezing The preparation of ice layers by condensation of water vapor in situ is most convenient and without problems, provided a specimen-cooling device is available which permits temperature control, separately, of the specimen part and of the surrounding cooling chamber [7]. The investigations in the electron microscope are then restricted to the ice per se. If, however, a suspension of biological particles is to be embedded in ice, the ice layer prepared from water rather than from vapor must be produced outside the microscope. In doing so, the water must be vitrified completely, and in the case of a solution a separation into salt and ice must be avoided. This, in turn, requires extremely high rates of freezing. In the following we shall set forth the conditions which are necessary to achieve this total vitrification of solutions (and of pure water as well). We set out with the usual procedure in which the copper grid carrying the preparation is. together with the tweezers which hold the grid,
'1 > 1.5 m/s
200 K !! 160K!!
!! !!!
110K -----
/
---~
77K LN2
--
Fig. 3. Schematic of a device for shock-freezing in liquid propane. N u m b e r s refer to a typical t e m p e r a t u r e distribution.
H,-G. Heide, E. Zeitler / Physical behatHor of solid water at low temperatures
Neither boiling liquid nitrogen nor supercooled liquid nitrogen is suitable as a coolant. Even in LN 2 at 64 K, water crystallizes cubically. At 77 K, large crystals of hexagonal ice are found using this coolant. The reason is that even specimens of extremely low mass become covered by isolating vapor (Leidenfrost effect) to slow down the rate of freezing. For this and other reasons to be expounded later, we use and recommend liquid propane. In order to demonstrate their importance, we shall deal with every single step of the shock-freezing method. It starts with the free fall of the specimen through the cold gas atmosphere above the coolant. An estimate indicates that at a free fall speed of 2 m / s , corresponding to an initial height of 20 cm, the thin water layer can freeze within a few mm, assuming a temperature of 120 K for the nitrogen gas layer. It is obvious that the freezing rate produced by a gaseous coolant is lower than that produced by a liquid coolant. Hence this first cooling step is critical and might be a source of errors unless care has been taken to guarantee that the fall speed is high and that the gas layer to be penetrated is sufficiently thin. The next step begins when the tweezers enter the coolant. The tweezers should be of stainless steel with long and pointed tips. They should be stopped, without rebound, after a few millimeters descent into the coolant. Here it is important to distinguish two phases: first, the phase of freezing; and second, the phase of ensuing cooling. Suppose that the water layer has been shock-frozen in a temperature range between 273 and 230 K and has assumed the desired amorphous structure. During further cooling, down to 110 K, this structure can still crystallize spontaneously and spoil. The reason for this is the rather high nucleation probability between 240 and 140 K; and therefore this dangerous range has to be traversed with the highest cooling rate possible. From all this follows that one cannot expect an amorphous product if the coolant itself has a temperature above 110 K. This is the physical reason why we opt for propane, whose melting point is at 85.4 K (boiling point 231 K, heat capacity 2.2 J / g . K). Convinced that the low temperature of the coolant is of utmost importance, we control the propane tern-
159
perature, by means of a thermocouple and a Joule heater, to 88 _+ 2 K. The next step, removal of the specimen from the liquid propane, is not without pitfalls either, because the thin specimen has to be brought into the relatively warm atmosphere, in which undesirable crystallization can take place (see fig. 3). We avoid this by first submerging the little container, which holds the liquid propane with the tweezers in it, up to its rim into the surrounding liquid nitrogen. Only then are the tweezers, with grid, taken out, immediately submerged in the liquid nitrogen and transferred, by means of a ladle, into a second Dewar. There the specimen is held for a short time as close above the liquid nitrogen level as possible, while the remainder of the liquid propane is removed from the tweezers and grid by blotting paper. All further manipulations are those of a routine cryo transfer namely, insertion into the specimen holder (also under liquid nitrogen), transfer of the specimen holder in a protection chamber ("diving bell"), replacement of the liquid nitrogen by dry nitrogen gas, and insertion of the cold holder in the objectcooling device. During all steps of this transfer a temperature rise above 110 K must be avoided. If all these precautions and conditions are observed, a thin layer of amorphous water of "low density" will be produced. The use of free layers of frozen water - that is, without support - has its advantages, but also its disadvantages. If a support film is not present, it is much easier and more reliable to determine the proper moment for the water layer to be catapulted into the coolant. An additional advantage is the certainty that the suspended particles are entirely embedded in solid water. As an example, see the micrograph of embedded ribosomes in fig. 4. A disadvantage, in the case of self-supporting layers, is caused by the electrical and physical behavior of amorphous solid water: if holes or cracks in the layer are present, they lead to instabilities during microscopy, on account of charging and flowing of the specimen. In general, disturbances due to charging are rarer when the embedding layer is supported by a conducting film. Finally, we mention that the actual temperature which such a layer achieves when the local electron
160
1t.-G. Heide. E. Zeitler / Pl~vsi('al hehat,ior of ~ohd water at low temperatures
Fig. 4. Ribosomal subunits of E. coli embedded in a self-supporting layer of amorphous solid water H20(as, h). Temperature 20 K. Magnification 200,000 X. (Micrograph courtesy of E. Boekema and J. Jhger.)
beam heating is considered can be easily calculated [7], at which the heat conductivity can conservatively be assumed to be 1 × 10 5 × T2 W cm 1 K 1 ( T i n kelvin).
9. Conclusion The behavior and the transformations of the various forms of solid water under the experimental conditions of cryo electron microscopy are manifold. The understanding of this behavior is essential in comprehending the behavior of the embedded specimen under irradiation. There exist marked differences between an embedding in "flowing" amorphous water (that is, between 4 and 30 K) and one in microcrystalline ice above 70 K. Observation of the physics of ice and of amorphous water will render the practicing biologists more likely to achieve an appropriate embed-
ding and thus a successful investigation of the native specimen.
Acknowledgement Thanks are due Mr. J. J~.ger for his skillful and engaged help during the course of the experiments.
References [1] A.H. Narten, ('.G. Venkatesh and S.A. Rice, J. Chem. Phys. 64 (1976) 1106. [2] M.G. Sceats and S.A. Rice, in: Water, Vol. 7, Ed. F. Franks (Plenum, New York, 1982) oh. 2. [3] H.G. Heide, Untramicroscopy 14 (1984)271. [4] P,N.T. Unwin and J. Muguruma, J. Appl. Phys. 42 (1971) 3640: Phys. Status Solidi (a) 14 (1972) 207. [5] H.G. Heide, Ultramicroscopy 7 (1982) 299. /6] Y. Talmon, Ultramicroscopy 14 (1984) 305. [7] H.G. Heide, Uhramicroscopy 6 (1981) 115: Ultramicroscopy 10 (1982) 125. [8] M. Adrian, J. Dubochet, J. Lepault and A.W. McDowall, Nature 308 (1984) 32.
151
T H E P H Y S I C A L B E H A V I O R OF S O L I D W A T E R AT L O W T E M P E R A T U R E S AND T H E E M B E D D I N G OF E L E C T R O N M I C R O S C O P I C A L S P E C I M E N S H.-G. H E I D E and E. Z E I T L E R Fritz - H a b e r - l n s t i t u t der M a x - P l a n c k -Gesellschaft, F a r a d a y w e g 4 - 6, D - I 0 0 0 Berlin 33, G e r m a n y
Received 19 December 1984
Modifications of solid water and their transitions are described as they relate to cryo electron microscopy. In particular, the various amorphous states (amorphous polymorphs) as they exist below 100 K are extensively investigated. The "high-density" modification exhibits a lower viscosity than the "low-density" form. Differences are also observed in the mechanism of void formation due to electron irradiation: in the high-density form, voids are formed - not, however, in the low-density form. Together with the reaction to radiation damage, the physical properties of amorphous solid water are discussed with respect to embedding of organic specimens. Finally, the conditions and pitfalls associated with preparation of thin and entirely vitrified ice layers by shock-freezing are described.
1. Introduction It has been the dream of biologists to embed their specimens in amorphous solid water, thereby preserving them in their native state, whose structure can then be revealed by electron microscopy. After early attempts it became obvious that more of the physical nature of ice and its behavior had to be known before a successful procedure could be devised. As soon as our cryo objective, allowing specimen temperatures from 6 to 300 K, was completed, we embarked upon the investigation of ice layers. Although this study is far from complete, we feel that the results obtained thus far are worthwhile for the practitioner, because consideration of the facts found will assure successful specimen preparation. Furthermore, some of these results might be of interest in water research. Two methods for preparing ice layers of, say, 10-100 nm thickness have been employed namely, condensation of water vapor on carbon films within the cryo objective at various rates, and shock-freezing of thin water layers outside the microscope and subsequent cryo transfer. It is this second method which is commonly attempted, with more or less success, and we shall lay down at the end of this paper the rules and recommendations
derived by reasoning based on the measured physical data. In the following, for lack of a simple and convenient term for solid amorphous water, we sometimes call it "ice", although in a strict sense this should be used only for crystalline water. Whenever the structure is important, we shall use the appropriate term.
2. Four ice modifications in four temperature ranges Ice can exist in twelve different modifications, some only under extreme conditions. Under electron microscopy conditions, only four modifications prevail: hexagonal crystalline ice [Ih], cubic crystalline ice [Ic], amorphous low-density water [H20(as, 1)] and amorphous high-density water [H20(as, h)]. But only the vitreous amorphous kinds are desirable, because here those forces which can occur during crystallization and may damage the specimen are absent. The two amorphous forms (low density: 0.94; high density: 1.1 g / c m 3) were first described by Narten, Venkatesh and Rice [1], and later by Sceats and Rice [2]. Polymorphic transitions can occur at various temperatures, with or without the influence of
0304-3991/85/$03.30 © Elsevier Science Publishers B.V. (North-Holland Physics Publishing Division)
152
H.-G. Heide, E. Zeitler / Phys'ical behavior of solid water at low temperatures
irradiation. The transitions themselves can be investigated quantitatively by means of electron diffraction. Results of such investigations - for example, the electron doses required for effecting changes - have been reported in a previous paper [3]. The important facts are briefly reiterated in this introduction. Above 100 K, crystallization of an amorphous layer, that is, the transition H20(as, 1) ~ ice Ic, sets in, without irradiation. The time this transition takes becomes shorter at higher temperature; at 160 K it occurs practically instantaneously. Any electron irradiation shortens the transition times. The reverse reaction, that is, the amporphization of crystalline ice
Table 1 Ice in four t e m p e r a t u r e ranges T (K)
8-20
as,h
•
QS,,~
•
[c
•
[h
•
30 -70 0\
80 - 9 0
> 100
o
.)
i
°) i
,¢'
•
e- irradiated: as, h
Ic [h behavior ; as, h
f,v
QSj "~
lc
ice Ic --, H20(as, h),
f
(v)
(f I , ( n )
(f),n r
V
r
V
can take place below temperatures of 70 K; however, only when electron radiation is applied. The doses required for this transformation become smaller the lower the temperature. At 8 K, already 2 e / A 2 suffice to effect this transition. Under the same conditions low-density water also changes into high-density water,
Stable or m e t a s t a b l e m o d i f i c a t i o n s are m a r k e d by solid circles in the broad t e m p e r a t u r e range at which they exist. Open circles denote the unstable forms. A r r o w s indicate the transitions between states. Flow behavior and the formation of voids u n d e r irradiation are denoted by the following abbreviations: f - flowing; r rigid; v void formation; n no voids. For details see text.
H 2 0 ( a s , 1)--, U20(as, h);
(indicated by arrows) takes place. The following three sections are subdivided in the same manner.
and, for example, at 30 K a dose of 5 e/f~ 2, quite common in practical electron microscopy, completes the transition. The reverse transition, H20(as, h) --, HzO(as, 1), sets in spontaneously with or without irradiation, when the specimen is warmed to at least 70 K. After having mastered the preparation, identification and transformations of the various ice forms, we could proceed to study their behavior and peculiarities in the microscope. Conspicuous differences were found in the mechanical flow behavior of the various ice forms and their disposition to form voids during electron bombardment. For presenting our results (see table 1) we have chosen four temperature ranges such that by changing from one range to the next only one event - that is, one transformation of ice into another form
3. Structure
3.1. 8 to 20 K Without irradiation, all four polymorphs are stable (or metastable) in this temperature range. By stable, we mean that the forms can survive for at least several hours without any change. Under irradiation, however, every form changes immediately into HzO(as, h), which in turn remains unchanged for several hours, with or without radiation, suggesting that this form might be the intrinsically stable modification of water at those very low temperatures. Although we have only measured to 8 K, we believe that the observed data would also hold at 4 K.
H.-G. Heide, E. Zeitler / Physical behavior of solid water at low temperatures
3.2. 30 to 7 0 K
In this transitional range the phenomenon of an amorphous polymorphism appears obviously. Without irradiation, Ic and H20(as, 1) remain stable (or better, metastable). H20(as, h) transforms spontaneously into the low-density form with increasing temperature to a larger fraction. If radiation is applied, the low-density form converts into the high-density form - the more, the lower the temperature. The diffraction diagrams show certain intermediate states between high density and low density. They can also be observed in highdensity water which was produced by condensation at 30 K or in irradiated crystalline ice (Ic) at 20 to 60 K. In the Discussion, section 7, we shall come back to these intermediate states and offer an explanation within the frame of proposed structure models. 3.3. 80 to 90 K
Only in this temperature range are both Ic and H20(as, 1) stable during long times of observation and under assault by higher radiation doses ( > 1 0 - 6 e n m - 2 ) . The high-density form does not exist. These facts are important to know if one is to check the success of the vitrification of a preparation. (At lower temperatures the Ic will be amorphized almost instantaneously, whereas at higher temperatures an amorphous layer will crystallize). 3.4. 100 to 160 K
Above 100 K only crystalline ice is stable (Ic metastable in comparison to hexagonal ice).
4. Flow behavior
So far we have been able to discern the various amorphous solid states of water only by means of their diffraction patterns, but the observation of the motion patterns points up distinct differences which offer an aspect to characterize the physical behavior of the various modifications. The find-
153
ings are described for the same temperature ranges as in the preceding section on Structure. 4.1. 8 to 20 K
The first observations were made on thin selfsupporting films of H20(as, h) which showed holes or cracks. If the rim of such an opening is irradiated by a fine electron probe, the resulting electric charging then produces forces which, in turn, cause growth of little icicles (fig. 1). This growth of icicles is remarkably vigorous. A rapidly flowing transport of ice, sometimes showing fine surface waves, to the tip of the icicles can be made out. These icicles may vibrate and even tear off. The whole scenario of movement appears utterly blithe. Similar observations can be made on H20(as, h) layers supported on films, if these have a coarse granular structure. If such layers are irradiated, all the contours flow together, the grains disappear, and the irradiated area assumes a uniform grey appearance. 4.2. 30 to 70 K
In this transitional temperature range the flow behavior remains as the one just described, but at around 70 K the flow seems to be somewhat retarded. 4.3. 80 to 9 0 K
Under electron bombardment icicles grow on the rims of self-supporting films of H20(as, 1) also. In contrast to the earlier-described movements, this flowing is viscous, sluggish and slower. The icicles are not as fine, and the surface granulation appears more rough. In the many experiments the impression is intensified that the amorphous highdensity form flows more readily, like a liquid, than the amorphous low-density form. Also, in layers on supporting films, the disappearance of a coarse texture or the closing of holes occurs a little more slowly in the amorphous low-density than in the high-density form. Cubic ice does not flow in this temperature interval. Even bizarre and sharp-edged forms re-
H.-G. Heide. E. Zeitler / Physical behavior c (sohd water at low temperatures
154
Fig. 1. "Icicles" as they grow at the rim of thin amorphous solid water layers driven by charging in the beam. (a, b) H20(as, h) at 9K, very vigorous. (c, d) HzO(as, 1) at 75 K, somewhat more sluggish and viscous flow. Magnification 16,000 ×.
main unchanged in the electron beam (although mass is lost). 4.4. 100 to 160 K
In this range ice Ic does not flow either.
tion calculated under the assumption that the primary processes are direct knock-on impacts. The role of electron excitations in the bulk of the ice is uncertain. The foamy bubble formations which sometimes occur in wet organic specimens or aqueous solutions are unrelated to the void formation in pure ice (see section 7).
5. Void formation 5.1. 8 to 20 K
The production of cavities by radiation, mainly in metals, has been an object of study for the last twenty years. The same radiation damage occurs also in ice [4]. The mechanism for this phenomenon is understood as follows: The atomic displacement processes caused by the primary irradiation produce Frenkel defect pairs whose interstitials migrate to the surface, whereas the vacancies agglomerate and form voids. The experimental cross-section agrees with the theoretical cross-sec-
In amorphous high-density water void formation is very pronounced. Dose values between 500 and 1000 e / A 2 produce voids with irregular shapes and various sizes. With prolonged irradiation the size and number of the voids increase (see fig. 2), but only to such an extent that the distance from one void to another, or to the surface, is greater than about 200 A. (In a 100 A thick layer the void formation has never been seen, because the
H.-G. Heide, E. Zeitler / Physical beha~ior of solid water at low temperatures
155
Fig. 2. Void formation in HzO(as , h) at 9 K caused by irradiation with 100 kV electrons. (a) (h) 20, 1000, 1700, 2700, 3600, 4600, 5500, 6700 e/,~ 2, respectively. Initial layer thickness ca. 1000 A. Magnification 90,000 X.
156
H.-G. Heide, 1:2 Zeitler / Pt(vsical behat,ior of solid water at low temperatures
vacancies migrate directly to the surface.) After an irradiation of 1600 e / A 2 the voids assume a spherical shape, and their number per volume no longer increases. The total void volume amounts to about 3% of that of the ice layer. We have calculated that at 100 keV one elctron produces about 5 × 10 ~7 Frenkel pairs per cm 3. From this (and the molecule density of 3.6 × 1022 H 2 0 / c m 3) follows that an exposure of 2000 e/,~ 2 displaces 3% of all H 2 0 molecules. More radiation does not produce any additional voids. The ice layer itself is etched away by the electron beam at a rate of 80 , ~ / C cm 2 [3]. During this process voids are released through the surface, but without exhibiting any indentation. One might surmise that as soon as the bubble breaks, the remaining layer flows together to form a plane surface. As the void's diameter can never be larger than the thickness of the layer, the size of the voids reduces during observation in the same way as does the thickness of the ice layer. 5.2. 30 to 7 0 K
In this transitional region the void formation is seen rather irregularly - sometimes as single voids, sometimes as colonies. The void formation seems to become rarer to the same degree that amorphous high-density water transforms into the low-density form when the temperature is raised. 5.3. 80 to 90 K
In a pure H20(as, 1) layer, void formation does not occur. Irradiation just removes the ice with the well known uniform manner and rate. In crystalline ice layers, however, void formation has been observed. 5.4. 100 to 160 K
If the ice layers consist of large enough single crystals, these crystals also show void formation. Should the layers be composed of small microcrystals with many surfaces, however, void formation does not occur. Neither does void formation occur if the nucleation is prohibited: this is the case in very pure and fresh crystals, but not in condensed layers which always contain foreign rest
gas molecules. Furthermore, we assume that in ice crystals, just as in metals, the presence of dislocations will influence the migration and agglomeration of point defects.
6. Radiation damage The radiation damage in ice which results in a loss of matter has recently been described extensively [3]. It could be shown that amongst all possible mechanisms, those ionization processes which take place in the surface layer contribute predominantly to the removal of the ice. Below 90 K the rate of removal is constant, amounting to one monolayer H 2 0 per dose of 25 e / f ~ 2 at 100 kV. The question of whether the removal rates in amorphous and in crystalline ice are the same could not answered experimentally up to now. The difficulty stems from the fact that the actual surface in a grainy microcrystalline layer cannot be readily controlled, whereas the surface of an amorphous layer is a rather defined entity. However, after we succeeded in preparing larger plates of single crystals, we could perform the comparative measurements, with the following results: At temperatures between 75 and 90 K, the mass loss rate, through electron irradiation, for crystalline ice (lh) is about two-thirds that for amorphous solid water. The rate seems to depend on the crystallographic orientation of the ice surfaces. More exact measurements are hampered by the migration of bend contours across the irradiated field of obserw~tion. Because of its great importance for ice embedding of specimens, one observation should be mentioned, and this is in connection with carbon film [3,5] since no other material has been used: When the ice is removed entirely by the electron beam, the carbon film support also shows radiation damage (about 6 ,~ thinning at 9 K). When the removal is effected partly by electrons and the remainder by thawing, the carbon support shows no radiation effect. We interpret this finding as the action of radicals which can react only when they are free and in contact with the carbon film. Fixed within the bulk of ice, they cannot escape. This etching effect depends, of course, on the
tL-G. Heide, E. Zeitler / Pltvsical behavior of solid water at low temperatures
support to be etched. More experiments are required, and Talmon [6] has performed some in this direction. Based on our observations, two points should be made. One must clearly distinguish between ice embeddings which are completely sealed and those which have gaps or interfaces (for example, between microcrystals) or incomplete coverage of particles protruding from a thin layer. Such specimens might be "frozen-hydrated" but not truly embedded. The second point is that in amorphous solid water irradiation cannot create new ice surfaces: because of its viscosity, the solid amorphous water will always close the gaps around particles. This can be seen, for example, when small polystyrene latex spheres are embedded in amorphous solid water at about 10 K. As long as they are completely embedded, no shrinkage of the particles can be observed even at radiation doses high enough to produce a marked mass loss (increase in density and decrease in volume) when those particles are n o t embedded.
7. Discussion
The various behavior patterns of the two amorphous forms of solid water are interesting in many aspects. For example, it is surprising that in the amorphous high-density water voids are formed below 30 K, whereas in the low-density species no such voids occur, although the temperature is higher (70 to 100K). These findings pose the question about the mechanism which, below 30 K, permits the migration of defects and nucleation for cavity clusters. Also, why are these voids stable in the rather fluid matrix of high-density amorphous water? On the other hand, what hinders the mobility of defects and the agglomeration of cavities (or promotes recombination) in the low-density amorphous water? In contrast to that, voids are formed in cubic ice at the very same temperature (ca. 85 K). (Perhaps here the thermal mobility of OH ions helps to explain the void formation.) The observed differences in the flow behavior of the amorphous solid water begs the question of whether the structure proposals can explain the
157
difference in the viscosity of the two amorphous forms. For both amorphous forms of solid water, structure models have been given [1,2] which are plausible and compatible with X-ray diffraction data. In both cases the nearest-neighbor distance between oxygen atoms is 2.76 A, and on the average in a tetrahedral coordination. If one starts with the structure of Ic but a greater variation for the second-neighbor distance distribution ( - 4 . 5 A), one arrives at a randomized network of hydrogen bonds, which is characterized through minor distortions of the hydrogen bonding the distortions stemming from the variation of the angle ( - 1 0 9 °) between three oxygen molecules. This model is called the randomized network model (RNM), and it is considered valid for HzO(as, 1). A model for the H20(as, h) can be derived readily from the above model by putting, in about every second cavity position, an "interstitial water molecule" into the ice lattice. In addition to the 2.76 and 4.5 A distances, a set of second O - O distances of 3.25 A and a significantly smaller than 109 ° O - O - O angle results. A corresponding structure model (RNM) can also be derived for amorphous water (as, h) as a randomized version of ice II or ice III. Most remarkable is the close similarity of the structure to that of liquid water. From the considerations about the observed intermedate stages between (as, h) and (as, 1) (section 3.2), a very likely explanation follows. One merely has to assume that varying fractions (0-45%) of cavity positions are occupied by interstitial water molecules. In any case, the observed "liquidity" of the high-density water seems to parallel its structural similarity to water at room temperature. The consequences for ice embedding are not clear in all aspects as yet. Certainly the remarkably low viscosity of both amorphous solid waters is advantageous for the avoidance of dangerous mechanical stresses which, in turn, would lead to artefacts. Furthermore, the capability of flowing prevents the creation of gaps around the embedded particle, thus enhancing the radiation protection. Although it is very unlikely, it is not known whether the transformation from low-density to high-density amorphous water will injure the embedded
158
H.-G. Heide, E. Zeith, r / Pto,stcal hehat#or ~f sohd water at low temperatures"
specimen. Another important point would be the clarification of whether the conditions which further or prevent void formation also affect the ice-embedded specimen. At this point it is important to control the formation of voids with the often observed "bubbling", an event which occurs in aqueous solutions but not in pure water, at various temperatures. For example, in a vitrified solution of glycerol in water such bubbles are formed at 80 K after a very short irradiation. They grow quickly and finally form a very dense network. The vigorous foaming reflects probably a viscous flowing under the driving force of gaseous radiolytic products. After all the water has been lost by radiolysis, a honeycomb-like skeleton of carbonaceous material remains. At very low temperature, where high-density water prevails (say at 10 K), only a few bubbles can be observed. and these disappear under prolonged irradiation. Possibly this is another example for the difference between the two amorphous forms. However, it could also indirectly be due to a difference in the separation between water and solute.
plunged into a coolant by means of a guillotine-like device (fig. 3). One problem arising here is to choose the right moment to release the catch for the tweezers to fall. The moment is right when the water layer has, by evaporation, reached a thickness that is thin enough to be transparent to the electrons in the microscope but not completely dried up. This difficulty can be coped with by applying the method of Adrian et al. [8]: A droplet of water or of the suspension to be investigated is applied to a naked copper grid and blotted off with blotting paper. Observation under a stereo microscope of the water film's evaporation shows the film to rupture. The right instant for triggering the guillotine is that moment just before the rupture occurs. If suspensions or solutions are to be applied, they should have the lowest possible concentration, and the blotting should be controlled carefully. Important is that the copper grids have a hydrophilic surface. According to our experience, chemical cleaning with solvents or a short etching in an ammonia/water/alcohol mixture are most recommendable.
8. Preparation by shock-freezing The preparation of ice layers by condensation of water vapor in situ is most convenient and without problems, provided a specimen-cooling device is available which permits temperature control, separately, of the specimen part and of the surrounding cooling chamber [7]. The investigations in the electron microscope are then restricted to the ice per se. If, however, a suspension of biological particles is to be embedded in ice, the ice layer prepared from water rather than from vapor must be produced outside the microscope. In doing so, the water must be vitrified completely, and in the case of a solution a separation into salt and ice must be avoided. This, in turn, requires extremely high rates of freezing. In the following we shall set forth the conditions which are necessary to achieve this total vitrification of solutions (and of pure water as well). We set out with the usual procedure in which the copper grid carrying the preparation is. together with the tweezers which hold the grid,
'1 > 1.5 m/s
200 K !! 160K!!
!! !!!
110K -----
/
---~
77K LN2
--
Fig. 3. Schematic of a device for shock-freezing in liquid propane. N u m b e r s refer to a typical t e m p e r a t u r e distribution.
H,-G. Heide, E. Zeitler / Physical behatHor of solid water at low temperatures
Neither boiling liquid nitrogen nor supercooled liquid nitrogen is suitable as a coolant. Even in LN 2 at 64 K, water crystallizes cubically. At 77 K, large crystals of hexagonal ice are found using this coolant. The reason is that even specimens of extremely low mass become covered by isolating vapor (Leidenfrost effect) to slow down the rate of freezing. For this and other reasons to be expounded later, we use and recommend liquid propane. In order to demonstrate their importance, we shall deal with every single step of the shock-freezing method. It starts with the free fall of the specimen through the cold gas atmosphere above the coolant. An estimate indicates that at a free fall speed of 2 m / s , corresponding to an initial height of 20 cm, the thin water layer can freeze within a few mm, assuming a temperature of 120 K for the nitrogen gas layer. It is obvious that the freezing rate produced by a gaseous coolant is lower than that produced by a liquid coolant. Hence this first cooling step is critical and might be a source of errors unless care has been taken to guarantee that the fall speed is high and that the gas layer to be penetrated is sufficiently thin. The next step begins when the tweezers enter the coolant. The tweezers should be of stainless steel with long and pointed tips. They should be stopped, without rebound, after a few millimeters descent into the coolant. Here it is important to distinguish two phases: first, the phase of freezing; and second, the phase of ensuing cooling. Suppose that the water layer has been shock-frozen in a temperature range between 273 and 230 K and has assumed the desired amorphous structure. During further cooling, down to 110 K, this structure can still crystallize spontaneously and spoil. The reason for this is the rather high nucleation probability between 240 and 140 K; and therefore this dangerous range has to be traversed with the highest cooling rate possible. From all this follows that one cannot expect an amorphous product if the coolant itself has a temperature above 110 K. This is the physical reason why we opt for propane, whose melting point is at 85.4 K (boiling point 231 K, heat capacity 2.2 J / g . K). Convinced that the low temperature of the coolant is of utmost importance, we control the propane tern-
159
perature, by means of a thermocouple and a Joule heater, to 88 _+ 2 K. The next step, removal of the specimen from the liquid propane, is not without pitfalls either, because the thin specimen has to be brought into the relatively warm atmosphere, in which undesirable crystallization can take place (see fig. 3). We avoid this by first submerging the little container, which holds the liquid propane with the tweezers in it, up to its rim into the surrounding liquid nitrogen. Only then are the tweezers, with grid, taken out, immediately submerged in the liquid nitrogen and transferred, by means of a ladle, into a second Dewar. There the specimen is held for a short time as close above the liquid nitrogen level as possible, while the remainder of the liquid propane is removed from the tweezers and grid by blotting paper. All further manipulations are those of a routine cryo transfer namely, insertion into the specimen holder (also under liquid nitrogen), transfer of the specimen holder in a protection chamber ("diving bell"), replacement of the liquid nitrogen by dry nitrogen gas, and insertion of the cold holder in the objectcooling device. During all steps of this transfer a temperature rise above 110 K must be avoided. If all these precautions and conditions are observed, a thin layer of amorphous water of "low density" will be produced. The use of free layers of frozen water - that is, without support - has its advantages, but also its disadvantages. If a support film is not present, it is much easier and more reliable to determine the proper moment for the water layer to be catapulted into the coolant. An additional advantage is the certainty that the suspended particles are entirely embedded in solid water. As an example, see the micrograph of embedded ribosomes in fig. 4. A disadvantage, in the case of self-supporting layers, is caused by the electrical and physical behavior of amorphous solid water: if holes or cracks in the layer are present, they lead to instabilities during microscopy, on account of charging and flowing of the specimen. In general, disturbances due to charging are rarer when the embedding layer is supported by a conducting film. Finally, we mention that the actual temperature which such a layer achieves when the local electron
160
1t.-G. Heide. E. Zeitler / Pl~vsi('al hehat,ior of ~ohd water at low temperatures
Fig. 4. Ribosomal subunits of E. coli embedded in a self-supporting layer of amorphous solid water H20(as, h). Temperature 20 K. Magnification 200,000 X. (Micrograph courtesy of E. Boekema and J. Jhger.)
beam heating is considered can be easily calculated [7], at which the heat conductivity can conservatively be assumed to be 1 × 10 5 × T2 W cm 1 K 1 ( T i n kelvin).
9. Conclusion The behavior and the transformations of the various forms of solid water under the experimental conditions of cryo electron microscopy are manifold. The understanding of this behavior is essential in comprehending the behavior of the embedded specimen under irradiation. There exist marked differences between an embedding in "flowing" amorphous water (that is, between 4 and 30 K) and one in microcrystalline ice above 70 K. Observation of the physics of ice and of amorphous water will render the practicing biologists more likely to achieve an appropriate embed-
ding and thus a successful investigation of the native specimen.
Acknowledgement Thanks are due Mr. J. J~.ger for his skillful and engaged help during the course of the experiments.
References [1] A.H. Narten, ('.G. Venkatesh and S.A. Rice, J. Chem. Phys. 64 (1976) 1106. [2] M.G. Sceats and S.A. Rice, in: Water, Vol. 7, Ed. F. Franks (Plenum, New York, 1982) oh. 2. [3] H.G. Heide, Untramicroscopy 14 (1984)271. [4] P,N.T. Unwin and J. Muguruma, J. Appl. Phys. 42 (1971) 3640: Phys. Status Solidi (a) 14 (1972) 207. [5] H.G. Heide, Ultramicroscopy 7 (1982) 299. /6] Y. Talmon, Ultramicroscopy 14 (1984) 305. [7] H.G. Heide, Uhramicroscopy 6 (1981) 115: Ultramicroscopy 10 (1982) 125. [8] M. Adrian, J. Dubochet, J. Lepault and A.W. McDowall, Nature 308 (1984) 32.