Journal of Microscopy, Vol. 161, Pr 2, February 1991, pp. 327-335. Received 6 July 1990; revised and accepted 10 October 1990
ADONIS002227209100024C
Early results using high-resolution, low-voltage, low-temperature SEM
by J A M E SB. P A W L E YP, A U LW A L T H E RS, H I A N - J I U NS H I H *and M A R E KM A L E C K IIntegrated , Microscopy Resource, 1675 Observatory Dr., Madison, WI 53706, and *Biochemistry Department, University of Wisconsin, Madison, WZ 53706, U . S . A .
KEY
w O R D S . Low-voltage SEM, freezing, cryo-SEM, high-resolution SEM.
SUMMARY
Recent advances in the design of the scanning electron microscope (SEM) column, such as the coupling of a field-emission gun to a low-aberration immersion lens and the availability of a high-stability cryo-transfer stage, make low-temperature, lowvoltage SEM (LTLVSEM) possible at very high resolution. We have used this combination to obtain results with uncoated biological specimens. The trichocyst from a Paramecium was used as a test specimen to observe the shrinkage of this structure as the temperature is raised from 170 K to room temperature following freeze-drying. High-magnification stereo images were obtained of trichocysts that had been prepared by freezing, freeze-substitution and critical-point drying and which were subsequently viewed by LTLVSEM to reduce beam damage and contamination. INTRODUCTION
Freeze-fracture, and freeze-etch were the earliest techniques to provide structural information of biological surfaces at macromolecular resolution (Steere, 1957; Moor 81Muhletaler, 1963). Later, when SEM surface images became available (Oatley et al., 1965), it was found that, although lower in resolution, they were still useful for three major reasons: (1) specimen preparation was easier; (2) the surfaces revealed were not limited to those formed by random fracture and were continuous over large areas, simplifying the correlation of high-resolution structural data from different areas of heterogeneous specimens; (3) as the specimen itself was viewed, rather than a replica, it could undergo elemental analysis or subsequent correlative study by other microscopic techniques.
Early efforts to combine these two surface imaging techniques led to pioneering work in which the frozen surface was viewed directly in the SEM without coating (Thornley, 1960): eventually a number of commercial systems were developed to coat the freeze-fracture surface and then introduce it directly onto an SEM cold stage
c) 1991 The Royal
Microscopical Society
327
328
J. B. Pawley et al.
(Hayes & Koch, 1975; Nei & Fujikawa, 1977). Equipment to facilitate this procedure has been developed by a number of individuals and manufacturers as reviewed by Beckett & Read (1986) and Read & Jeffree (1991). While this approach has had considerable success, particularly on botanical specimens (Echlin, 1971; Walther et al., 1990) and on specimens to be studied by X-ray microanalysis (Zierold, 1983), most studies were limited to relatively low resolution and magnification compared to freeze-etch (Pawley et al., 1978; Muller et al., 1986; Wergin & Erbe, 1989; Fujikawa et al., 1990). Early failure to obtain high-resolution results was primarily due to four factors: (1) mechanical instability in the available SEM cold stages; (2) obscuration of surface detail by condensation of contaminants from the vacuum onto the cold specimen; (3) low contrast of fine image details at the high beam voltages used; (4) gross radiation damage to the ice/sample matrix (caused in part by the high beam current needed to overcome the low contrast). Recent improvements in instrumentation allow these problems to be overcome. The purpose of this paper is to give a brief outline of these improvements and to give some examples of the advantages that they offer to increase the effectiveness of cryo-SEM. The improvements discussed include: (1) field emission (FE) gun SEMs with immersion lenses: these instruments permit 3-4-nm probes at beam voltages (V,)down to 1.5 kV where contrast is high, radiation damage is low, and specimen charging is less severe; (2) high-stability, side-entry SEM cold stages; (3) improved equipment for the low-temperature transfer and coating of cryofractured specimens. EARLY CRYO-SEM
The earliest efforts to view biological specimens at low temperature in the SEM were performed at low V,. This was done chiefly to avoid charging artefacts, but other advantages such as increased contrast were also noted (Thornley, 1960). Unfortunately the early instruments were incapable of high-resolution performance at 1-3 kV so work at high magnification was impossible. The perceived solution to this problem was to develop an apparatus that allowed the cryo-fracture surface to be coated with metal or carbon and thereby permit the use of higher beam voltages. This approach was tried by several groups divided about equally between those interested in X-ray microanalysis (Zierold, 1983) and those interested in topographical imaging (Pawley et al., 1978; Wergin & Erbe, 1989; Herter et a)., 1991; Read & Jeffree, 1991). It is the latter aspect of cryo-SEM use that is of concern here. Cryo-SEM techniques simplified the task of finding the corresponding parts of complementary replicas because the entire area of each fracture surface could clearly be seen and, in addition, the recognition of matching areas could be assisted by reversing the leads on a scanning coil so that the images of both sides have the same handedness (Pawley et al., 1978). However, these advantages did not compensate for the markedly inferior resolution of the early SEM images. Early cold stages had problems with drift and vibration and the low contrast of aqueous specimens, only lightly coated with metal, made focusing difficult when viewed at 20-30 kV. Finally, with a beam penetration of tens of micrometres, gross radiation damage to the aqueous matrix beneath the metal coating was almost inevitable whenever high-magnification focusing was desired. The phenomenon of gross damage of organic materials embedded in ice was later studied in depth by Talmon (1984).
Early, high-resolution cryo-L VSEM
329
There are two aspects to the question of whether or not the radiation dose to a bulk SEM specimen is a strong function of beam voltage (V,):total dose and near-surface dose. The total dose to a bulk specimen is proportional to the energy deposited by the beam. It can be calculated by first determining the power in the beam, P = IbV, where I b is the beam current with perhaps a small correction for power losses due to the emission of backscattered electrons. To turn this into radiation dose (the SI unit of absorbed dose is the gray = 1 J/kg), it is necessary to assume a beam dwell time and some estimate of the mass contained in the irradiated volume. Because the secondary electron (SE) coefficient drops with V,,the I, needed to produce an image of a certain quality increases with V, as does total power deposition. In a biological specimen viewed at high V,, however, much of this power is deposited well below the surface and the resulting damage may be more likely to distort the whole cell rather than severely modify the actual surface region itself (Pawley & Erlandsen, 1989). A consideration of near-surface damage is somewhat more complex because, as Joy (1987) has pointed out, energy deposition near the surface per beam electron is inversely proportional to V,. However, to a first approximation, the SE signal is proportional to the near-surface energy deposition, and so surface radiation damage per emitted secondary electron is not a strong function of beam voltage. On the other hand, when the energy deposition needed to make an image is considered, contrast becomes important. The higher contrast of small surface features characteristic of low-voltage SE signals means that fewer such electrons need to be collected from the specimen and that, as in the case of the total dose, the near-surface radiation dose necessary to produce a certain image quality is usually less at low voltage. The low image contrast and the practical limitations imposed on early biological cryo-SEM by radiation damage strongly suggested a return to the low-voltage operation pioneered by Thornley where both problems are much less severe (Pawley, 1984; Pawley & Erlandsen, 1989). Unfortunately, more than a decade would pass before the electron optical systems needed to produce a sufficiently small beam at low voltage became commercially available. RECENT D E V E L O P M E N T S
In 1986 two new instruments which coupled an FE source with a low-aberration, immersion-type final lens were introduced (Nagatani & Saito, 1986; Kersker et al., 1989). With a beam diameter of c. 0.5 nm at 30 kV and 3.5 nm at 1.5 kV, electron optical considerations were no longer a significant limitation when viewing most biological specimens. In addition to improved electron optical performance, these instruments feature eucentric, side-entry goniometer stages similar to those found on the TEM. As a result they could accept, with only slight modification, the high-resolution cryo-transfer stages previously developed for the TEM. Finally, the vacuum systems of these instruments was a great improvement over that found on conventional SEMs. In some instruments the specimen was almost completely surrounded by a liquid nitrogen cryo-shield (Kersker et al., 1989) while others used sophisticated dry pumping systems, crushable metal seals and anti-contaminators to reduce hydrocarbon contamination (Nagatani & Saito, 1986). These changes, in conjunction with the use of the oil-free molecular drag pump in place of the standard mechanical roughing pump, have reduced hydrocarbon contamination from the vacuum to an almost negligible level (Pawley & Erlandsen, 1989). In response to these developments, the potential of LTLVSEM needs re-evaluation. The prime advantage of LTLVSEM is that uncoated material can be examined. This means that not only can cryo-prepared specimens be observed directly without the need for cryo-coating equipment, but also that specimens whose surfaces are changing
330
J . B. Pawley et al.
Fig. 1. Uncoated, freeze-dried cell on gold foil substrate at (a) 15kV, (b) 5kV, (c) 3kV, and (d) 1.5kV. (Scale bars = 7 . 5 / i m ,2.0pm). At the high voltages no details of the cell surface are visible, but at 3 and 1.5 kV the surface can be seen t a be covered with contaminating crystallites.
with time, such as those undergoing freeze-drying, can also be observed. In addition, LTLVSEM can provide a reference image that can be used either to assess the effect of the coating on the specimen itself or its effect on the SEM image. This paper describes our preliminary uses of LTLVSEM to observe uncoated surfaces, to monitor the freeze-drying process and to optimize the observation conditions of specimens prepared at room temperature by reducing their susceptibility to sampleborne hydrocarbon contamination. Figure 1 shows four views of two parts of the same 3T3 cell viewed at 15,5,3 and 1.5 kV. The cell had been grown on a gold foil and rapidly frozen by plunging into
Early, high-resolution cryo-L V S E M
331
liquid ethane held at its melting point. It was then viewed in the Hitachi S-900 on a Gatan (Pleasanton, CA) model 636 cryo-transfer stage at 150 K after the surface ice had been removed by freeze-sublimation at 170 K in the microscope. Before the cell was frozen, it had been briefly rinsed to remove the growth medium but, as can be seen from the lower two images, this rinse was insufficient and the surface is obscured by a coating of microcrystallites. Charging artefacts are not present in any of the images. At the two higher beam voltages, penetration is sufficient for beam-induced conductivity to carry charge to the gold foil while, at the lower voltages, the combined secondary/backscattered coefficient evidently is close to unity. At 15 kV, and to a lesser extent at 5 kV, scratch marks in the foil can be seen through the cell process whereas details of the cell surface are invisible. At 3 kV, the scratch marks become invisible but the cell surface still remains somewhat blotchy. We attribute this blotchiness to variations in the mass thickness of the subsurface of the cell process because, at a beam voltage of 1.5kV which provides a better approximation of a true surface image, the blotchy appearance disappears. SHRINKAGE DURING FREEZE-DRYING
Freeze-drying was the first method used for removing the liquid phase from biological specimens without producing surface-tension artefacts (reviewed by Steinbrecht & Muller, 1987). Its widespread use has been limited by the difficulty of initially freezing the specimen without producing ice-crystal artefacts and by the fact that many biological structures collapse when the bound water is removed as the specimen temperature goes above about 210K (MacKenzie, 1972; Gross, 1987). T o re-evaluate the process of freeze-drying, we have used the trichocyst as our test specimen. T h e trichocyst is a small (1 x 3pm), ordered protein structure, contained in a membrane-bound organelle that lies just below the cell membrane of paramecia (Bannister, 1972). When triggered by Ca2+ release, the trichocyst expands to about nine times its original length in a few milliseconds. T h e structures of the condensed and the expanded forms have been well studied by negative-strain and freeze-fracture deep-etch techniques (Peterson et al., 1987). We were attempting to extend these studies in the SEM using gold-labelled monoclonal antibodies to specific structural proteins; however, our critical point-dried (CPD) preparations were not comparable to those imaged in freeze-fracture deep-etch. In particular, the diameter of the extended CPD trichocysts was smaller when viewed in the SEM. T h e trichocyst makes a good test specimen because it is small enough to be easily frozen, large enough to be interesting and, in addition, many structural studies exist to provide a basis for comparison (e.g. Peterson et al., 1987). T o determine the stage at which the shrinkage occurred, the freeze-drying process was observed by LTLVSEM. T o provide good thermal and electrical conductivity, 3-mm discs punched from gold foil were initially used as specimen supports (Fig. 2). However, the uncoated specimens showed very poor contrast against the gold background unless the gold foil had first been coated with a 100-pm layer of conductive carbon paint. T h e disc was incubated with a small droplet of the trichocyst suspension for about 2min, then washed on three droplets of distilled water, frozen by plunging into liquid ethane, and stored in liquid nitrogen. The frozen specimens were then loaded into the Gatan cryo-transfer stage, introduced into the microscope and freeze-dried at 170 K until the surface of the carbon became visible ( 30 min). During the freeze-drying process, images were recorded at 1.5 kV, 150K. Although contamination was not a problem because of the low specimen temperature, we noted some indication of beam damage. For this reason, astigmatism
-
332
J . B . Pawley et al.
Fig. 2. Uncoated freeze-dried trichocysts on a gold foil at 1.5 kV showing charging artefacts that made high-magnification images impossible (scale bar = 5 pm).
and focus were adjusted on an adjacent structure and the specimen was only viewed at low magnification unless an image was being recorded. Figure 3 shows a series of images of a trichocyst as it first emerged from the receding ice surface (Fig. 3a), and after it had been warmed to room temperature and then re-cooled to 150 K to record the image (Fig. 3b). Figure 3(c) was recorded immediately after Fig. 3(b). Small differences between Fig. 3(b) and 3(c) are probably attributable to beam damage while the larger changes between Fig. 3(a) and 3(b) may be due to alterations in the structure of the trichocyst as the bound water is removed (MacKenzie, 1972; Gross, 1987).
C R Y 0 OBSERVATION OF FREEZE-SUBSTITUTED
MATERIAL
The shrinkage evident in Fig. 3 and the unavailability of an apparatus for metal coating the freeze-dried specimens at low temperature prompted us to prepare specimens by freezing/freeze-substitution/criticalpoint-drying (Walther & Hentchel, 1989). Because carbon paint is not stable in the organic solvents used in this process, the trichocysts were prepared on discs cut from 2-mm carbon rods. Otherwise, initial preparation was the same as that for freeze-drying except that the wash steps were omitted.
Early, high-resolution cryo-L V S E M
333
Fig. 3. Shrinkage of a freeze-dried trichocyst as the temperature is raised from 170K where freeze-drying had occurred (a), to room temperature (and returned to 150K to make the micrograph, b). ( c ) Made immediately after (b) and shows some additional shrinkage, probably caused by beam damage (scale bar = 200nm).
Freeze-substitution and critical point-drying were carried out as described by Miiller er al. (1980) and by Walther & Hentschel (1989):the freeze-substitution medium consisted of methanol containing 0.5'),, osmium tetroxide, 3",,glutaraldehyde and 3",,water. Samples were freeze-substituted in a Balzers FSU 010 for 7 h at 183 K, 5 h at 21 3 K, and 5 h at 243 K. After washing with cold methanol (243K), samples were warmed to room temperature, washed with ethanol and critical point-dried (Anderson, 1951;Ris, 1985)using ethanol and carbon dioxide. Samples were coated with platinum by ion beam-sputtering (Franks er al., 1980) to a thickness of less than 2nm. Figure 4 is a stereo micrograph of the specimen prepared in this manner and recorded at a magnification of x 100,000 at 1.5 kV. T h e images were recorded using the cold stage at 150 K to reduce contamination. T h e ability to produce a usable stereo image is an indication that beam damage to the specimen during the recording of the first member of the pair was not great. CONCLUSION
We have outlined a few applications of LTLVSEM and have demonstrated that it can be useful for viewing uncoated specimens at up to 50,000 x at 1.5kV during and immediately after freeze-drying. Although these images demonstrate a higher resolution than has generally been possible in the past, a further improvement is evident if the specimen can be coated. These early results are a foretaste of what will be possible from the new S E M instruments when a suitable cryo-coating mechanism is developed. ACKNOWLEDGMENT
This work was supported by NIH Grant DRR-570 to the Integrated Microscopy Resource, Madison, Wisconsin.
334
J . B . Pawley et al.
Fig. 4. High-magnification stereo-pair of trichocyst prepared by freeze-substitution/criticalpoint-drying and coated with about 2 nm of Pt by ion beam-sputtering. As the images were made with the specimen held at 150K, no apparent differences occurred between the images that can be attributed to deposition of contamination on the specimen surface during the process of making the first member of the pair (scale bar = 1OOnm).
REFERENCES Anderson, T.F. (1951) Techniques for the preservation of the three-dimensional structures in preparing specimens for the electron microscope. Trans. N Y Acad. Sci. 13, 130-133. Bannister, L.H. (1972) T h e structure of trichocysts in Paramecium rerraurelia. 3. Cell Sci. 11, 899-929. Beckett, A. & Read, N.D. (1986) Low-temperature scanning electron microscopy. Ultrastructure Techniques /or Microorganisms (ed. by H.C. Aldrich and W.J. Todd), pp. 45-86. Plenum Press, NY. Echlin, P. (1971) T h e examination of biological material at low temperatures. Scanning Electron Microsc. I, 225-232. Franks, J., Clay, C.S. & Peace, G.W. (1980) Ion beam thin film deposition. Scanning Electron Microsc. I, 155-1 62. Fujikawa, S . , Suzuki, T. & Sakurai, S. (1990) Use of a manipulator for continuous observation of frozen samples by cryoscanning electron microscopy and freeze replicas. Scanning, 12, 99-106. Gross, H. (1987) High resolution metal replication of freeze dried specimens. Cryotechniques in Biological Electron Microscopy (ed. by R. A. Steinbrecht and K. Zierold), pp. 205-215. Springer-Verlag, Berlin. Hayes, T . L . & Koch, G. (1975) Some problems associated with low temperature micromanipulation in the SEM. Scanning Electron Microsc. I, 3542. Herter, P., Tresp, G., Hentschel, H., Zierold, K. & Walther, P. (1991) High-resolution scanning electron microscopy of frozen hydrated and freeze-substituted kidney tissue. 3. Microsc. 161, 375-385. Joy, D.C. (1987) LVSEM. Electron Microscopy and Microanalysis I987 (ed. by L. M. Brown), Insr. of Phys. Con/. Series 90, 175-180. Kersker, M., Nielsen, C., Otsuji, H., Miyokawa, T . & Nakagawa, S. (1989) The JSM-890 ultra high resolution scanning electron microscope. Proc. Electron Microsc. SOC.A m . 47, 88-89. MacKenzie, A.P. (1972) Freezing, freeze drying, and freeze substitution. Scanning Electron Microsc. II, 173-1 80.
Early, high-resolution cryo-L VSEM
335
Moor, H. & Miihlethaler, K. (1963) Fine structure in frozen etched yeast cells. 3. Cell Biol. 17, 609-628. Miiller, T., Guggenheim, T., Diiggelin, M. & Liiond, G. (1986) On line cryopreparation and cryomicroscopy in SEM with SCU 020. Proc. X I 1 Int. Cong. Electron Microsc. 3 (ed. by H. Imura, S. Maruse and T. Suzuki), 2233-2234. Japanese Society of Electron Microscopy, Tokyo. Miiller, M., Marti, T . & Kriz, S. (1980) Improved structural preservation by freeze-substitution. Proc. V I I Eur. Cong. Electron Microsc. 2, 720-721. Nagatani, T. & Saito, S. (eds) (1986) Instrumentation for ultra high resolution scanning electron microscopy. Proc. Int. Cong. Electron Microsc. 111, 2201-2208. Nei, T . & Fujikawa, S. (1977) Freeze-drying process of biological specimens observed with a scanning electron microscope. 3. Elecrron Microsc. 20, 202-203. Oatley, C.W., Nixon, W.C. & Pease, R.F.W. (1965) Scanning electron microscopy. A&. Electronics Electron Phys. 21, 181-247. Pawley, J.B. (1984) Low voltage scanning electron microscopy. 3. Microsc. 13, 387-410. Pawley, J.B. & Erlandsen, S.L. (1989) The case for low voltage SEM of biological samples. Scanning Microsc. Suppl. 3, 163-178. Pawley, J., Hayes, T.L. & Hook, G. (1978) Preliminary studies of coated complementary freeze-fractured yeast membranes viewed directly in the SEM. Scanning Electron Microsc. 11, 683-690. Peterson, J.B., Heuser, J.E. & Nelson, D.L. (1987) Dissociation and reassociation of trichocyst proteins: biochemical and ultrastructural studies. 3. Cell Sci. 87, 3-25. Read, N.D. & Jeffree, C.E. (1991) Low-temperature scanning electron microscopy in biology. 3. Microsc. 161, 59-72. Ris, H. (1985) The cytoplasmic filament system in critical point dried whole mounts and plastic embedded sections. 3. Cell Biol. 100, 14741487. Steere, R.L. (1957) Electron microscopy of structural detail in frozen biological specimens. 3. Biophys. Biochem. Cytol. 3, 45-60. Steinbrecht, R.A. & Miiller, M. (1987) Freeze substitution and freeze drying. Cryotechniques in Biological Electron Microscopy (ed. by R. A. Steinbrecht and K. Zierold), pp. 149-172. Springer-Verlag, Berlin. Talmon, Y. (1984) Radiation damage to organic inclusions in ice. Ultramicroscopy, 14, 305-316. Thornley, R.F.M. (1960) Recent developments in SEM. Proc. IV Eur. Cong. Electron Microscopy (ed. by A. L. Howink and B. J. Split), pp. 173-176. Walther, P. & Hentschel, J. (1989) Improved representation of cell surface structures by freeze substitution and backscattered electron imaging. Scanning Microsc. Suppl. 3, 201-21 1. Walther, P., Hentschel, J., Herter, P., Miiller, T. & Zierold, K. (1990) Imaging of intramembranous particles in frozen-hydrated cells (Saccharomyces cerevisiae) by high resolution cryo-SEM. Scanning, 12, 300-307. Wergin, W.P. & Erbe, E.F. (1989) Increasing the versatility of an EMscope SP2000A sputter cry0 system on a Hitachi S-570 scanning electron microscope. Scanning, 11, 293-303. Zierold, K. (1983) X-ray microanalysis of frozen hydrated specimens. Scanning Electron Microscopy. Scanning Electron Microsc. 11, 809-826.
ADONIS002227209100024C
Early results using high-resolution, low-voltage, low-temperature SEM
by J A M E SB. P A W L E YP, A U LW A L T H E RS, H I A N - J I U NS H I H *and M A R E KM A L E C K IIntegrated , Microscopy Resource, 1675 Observatory Dr., Madison, WI 53706, and *Biochemistry Department, University of Wisconsin, Madison, WZ 53706, U . S . A .
KEY
w O R D S . Low-voltage SEM, freezing, cryo-SEM, high-resolution SEM.
SUMMARY
Recent advances in the design of the scanning electron microscope (SEM) column, such as the coupling of a field-emission gun to a low-aberration immersion lens and the availability of a high-stability cryo-transfer stage, make low-temperature, lowvoltage SEM (LTLVSEM) possible at very high resolution. We have used this combination to obtain results with uncoated biological specimens. The trichocyst from a Paramecium was used as a test specimen to observe the shrinkage of this structure as the temperature is raised from 170 K to room temperature following freeze-drying. High-magnification stereo images were obtained of trichocysts that had been prepared by freezing, freeze-substitution and critical-point drying and which were subsequently viewed by LTLVSEM to reduce beam damage and contamination. INTRODUCTION
Freeze-fracture, and freeze-etch were the earliest techniques to provide structural information of biological surfaces at macromolecular resolution (Steere, 1957; Moor 81Muhletaler, 1963). Later, when SEM surface images became available (Oatley et al., 1965), it was found that, although lower in resolution, they were still useful for three major reasons: (1) specimen preparation was easier; (2) the surfaces revealed were not limited to those formed by random fracture and were continuous over large areas, simplifying the correlation of high-resolution structural data from different areas of heterogeneous specimens; (3) as the specimen itself was viewed, rather than a replica, it could undergo elemental analysis or subsequent correlative study by other microscopic techniques.
Early efforts to combine these two surface imaging techniques led to pioneering work in which the frozen surface was viewed directly in the SEM without coating (Thornley, 1960): eventually a number of commercial systems were developed to coat the freeze-fracture surface and then introduce it directly onto an SEM cold stage
c) 1991 The Royal
Microscopical Society
327
328
J. B. Pawley et al.
(Hayes & Koch, 1975; Nei & Fujikawa, 1977). Equipment to facilitate this procedure has been developed by a number of individuals and manufacturers as reviewed by Beckett & Read (1986) and Read & Jeffree (1991). While this approach has had considerable success, particularly on botanical specimens (Echlin, 1971; Walther et al., 1990) and on specimens to be studied by X-ray microanalysis (Zierold, 1983), most studies were limited to relatively low resolution and magnification compared to freeze-etch (Pawley et al., 1978; Muller et al., 1986; Wergin & Erbe, 1989; Fujikawa et al., 1990). Early failure to obtain high-resolution results was primarily due to four factors: (1) mechanical instability in the available SEM cold stages; (2) obscuration of surface detail by condensation of contaminants from the vacuum onto the cold specimen; (3) low contrast of fine image details at the high beam voltages used; (4) gross radiation damage to the ice/sample matrix (caused in part by the high beam current needed to overcome the low contrast). Recent improvements in instrumentation allow these problems to be overcome. The purpose of this paper is to give a brief outline of these improvements and to give some examples of the advantages that they offer to increase the effectiveness of cryo-SEM. The improvements discussed include: (1) field emission (FE) gun SEMs with immersion lenses: these instruments permit 3-4-nm probes at beam voltages (V,)down to 1.5 kV where contrast is high, radiation damage is low, and specimen charging is less severe; (2) high-stability, side-entry SEM cold stages; (3) improved equipment for the low-temperature transfer and coating of cryofractured specimens. EARLY CRYO-SEM
The earliest efforts to view biological specimens at low temperature in the SEM were performed at low V,. This was done chiefly to avoid charging artefacts, but other advantages such as increased contrast were also noted (Thornley, 1960). Unfortunately the early instruments were incapable of high-resolution performance at 1-3 kV so work at high magnification was impossible. The perceived solution to this problem was to develop an apparatus that allowed the cryo-fracture surface to be coated with metal or carbon and thereby permit the use of higher beam voltages. This approach was tried by several groups divided about equally between those interested in X-ray microanalysis (Zierold, 1983) and those interested in topographical imaging (Pawley et al., 1978; Wergin & Erbe, 1989; Herter et a)., 1991; Read & Jeffree, 1991). It is the latter aspect of cryo-SEM use that is of concern here. Cryo-SEM techniques simplified the task of finding the corresponding parts of complementary replicas because the entire area of each fracture surface could clearly be seen and, in addition, the recognition of matching areas could be assisted by reversing the leads on a scanning coil so that the images of both sides have the same handedness (Pawley et al., 1978). However, these advantages did not compensate for the markedly inferior resolution of the early SEM images. Early cold stages had problems with drift and vibration and the low contrast of aqueous specimens, only lightly coated with metal, made focusing difficult when viewed at 20-30 kV. Finally, with a beam penetration of tens of micrometres, gross radiation damage to the aqueous matrix beneath the metal coating was almost inevitable whenever high-magnification focusing was desired. The phenomenon of gross damage of organic materials embedded in ice was later studied in depth by Talmon (1984).
Early, high-resolution cryo-L VSEM
329
There are two aspects to the question of whether or not the radiation dose to a bulk SEM specimen is a strong function of beam voltage (V,):total dose and near-surface dose. The total dose to a bulk specimen is proportional to the energy deposited by the beam. It can be calculated by first determining the power in the beam, P = IbV, where I b is the beam current with perhaps a small correction for power losses due to the emission of backscattered electrons. To turn this into radiation dose (the SI unit of absorbed dose is the gray = 1 J/kg), it is necessary to assume a beam dwell time and some estimate of the mass contained in the irradiated volume. Because the secondary electron (SE) coefficient drops with V,,the I, needed to produce an image of a certain quality increases with V, as does total power deposition. In a biological specimen viewed at high V,, however, much of this power is deposited well below the surface and the resulting damage may be more likely to distort the whole cell rather than severely modify the actual surface region itself (Pawley & Erlandsen, 1989). A consideration of near-surface damage is somewhat more complex because, as Joy (1987) has pointed out, energy deposition near the surface per beam electron is inversely proportional to V,. However, to a first approximation, the SE signal is proportional to the near-surface energy deposition, and so surface radiation damage per emitted secondary electron is not a strong function of beam voltage. On the other hand, when the energy deposition needed to make an image is considered, contrast becomes important. The higher contrast of small surface features characteristic of low-voltage SE signals means that fewer such electrons need to be collected from the specimen and that, as in the case of the total dose, the near-surface radiation dose necessary to produce a certain image quality is usually less at low voltage. The low image contrast and the practical limitations imposed on early biological cryo-SEM by radiation damage strongly suggested a return to the low-voltage operation pioneered by Thornley where both problems are much less severe (Pawley, 1984; Pawley & Erlandsen, 1989). Unfortunately, more than a decade would pass before the electron optical systems needed to produce a sufficiently small beam at low voltage became commercially available. RECENT D E V E L O P M E N T S
In 1986 two new instruments which coupled an FE source with a low-aberration, immersion-type final lens were introduced (Nagatani & Saito, 1986; Kersker et al., 1989). With a beam diameter of c. 0.5 nm at 30 kV and 3.5 nm at 1.5 kV, electron optical considerations were no longer a significant limitation when viewing most biological specimens. In addition to improved electron optical performance, these instruments feature eucentric, side-entry goniometer stages similar to those found on the TEM. As a result they could accept, with only slight modification, the high-resolution cryo-transfer stages previously developed for the TEM. Finally, the vacuum systems of these instruments was a great improvement over that found on conventional SEMs. In some instruments the specimen was almost completely surrounded by a liquid nitrogen cryo-shield (Kersker et al., 1989) while others used sophisticated dry pumping systems, crushable metal seals and anti-contaminators to reduce hydrocarbon contamination (Nagatani & Saito, 1986). These changes, in conjunction with the use of the oil-free molecular drag pump in place of the standard mechanical roughing pump, have reduced hydrocarbon contamination from the vacuum to an almost negligible level (Pawley & Erlandsen, 1989). In response to these developments, the potential of LTLVSEM needs re-evaluation. The prime advantage of LTLVSEM is that uncoated material can be examined. This means that not only can cryo-prepared specimens be observed directly without the need for cryo-coating equipment, but also that specimens whose surfaces are changing
330
J . B. Pawley et al.
Fig. 1. Uncoated, freeze-dried cell on gold foil substrate at (a) 15kV, (b) 5kV, (c) 3kV, and (d) 1.5kV. (Scale bars = 7 . 5 / i m ,2.0pm). At the high voltages no details of the cell surface are visible, but at 3 and 1.5 kV the surface can be seen t a be covered with contaminating crystallites.
with time, such as those undergoing freeze-drying, can also be observed. In addition, LTLVSEM can provide a reference image that can be used either to assess the effect of the coating on the specimen itself or its effect on the SEM image. This paper describes our preliminary uses of LTLVSEM to observe uncoated surfaces, to monitor the freeze-drying process and to optimize the observation conditions of specimens prepared at room temperature by reducing their susceptibility to sampleborne hydrocarbon contamination. Figure 1 shows four views of two parts of the same 3T3 cell viewed at 15,5,3 and 1.5 kV. The cell had been grown on a gold foil and rapidly frozen by plunging into
Early, high-resolution cryo-L V S E M
331
liquid ethane held at its melting point. It was then viewed in the Hitachi S-900 on a Gatan (Pleasanton, CA) model 636 cryo-transfer stage at 150 K after the surface ice had been removed by freeze-sublimation at 170 K in the microscope. Before the cell was frozen, it had been briefly rinsed to remove the growth medium but, as can be seen from the lower two images, this rinse was insufficient and the surface is obscured by a coating of microcrystallites. Charging artefacts are not present in any of the images. At the two higher beam voltages, penetration is sufficient for beam-induced conductivity to carry charge to the gold foil while, at the lower voltages, the combined secondary/backscattered coefficient evidently is close to unity. At 15 kV, and to a lesser extent at 5 kV, scratch marks in the foil can be seen through the cell process whereas details of the cell surface are invisible. At 3 kV, the scratch marks become invisible but the cell surface still remains somewhat blotchy. We attribute this blotchiness to variations in the mass thickness of the subsurface of the cell process because, at a beam voltage of 1.5kV which provides a better approximation of a true surface image, the blotchy appearance disappears. SHRINKAGE DURING FREEZE-DRYING
Freeze-drying was the first method used for removing the liquid phase from biological specimens without producing surface-tension artefacts (reviewed by Steinbrecht & Muller, 1987). Its widespread use has been limited by the difficulty of initially freezing the specimen without producing ice-crystal artefacts and by the fact that many biological structures collapse when the bound water is removed as the specimen temperature goes above about 210K (MacKenzie, 1972; Gross, 1987). T o re-evaluate the process of freeze-drying, we have used the trichocyst as our test specimen. T h e trichocyst is a small (1 x 3pm), ordered protein structure, contained in a membrane-bound organelle that lies just below the cell membrane of paramecia (Bannister, 1972). When triggered by Ca2+ release, the trichocyst expands to about nine times its original length in a few milliseconds. T h e structures of the condensed and the expanded forms have been well studied by negative-strain and freeze-fracture deep-etch techniques (Peterson et al., 1987). We were attempting to extend these studies in the SEM using gold-labelled monoclonal antibodies to specific structural proteins; however, our critical point-dried (CPD) preparations were not comparable to those imaged in freeze-fracture deep-etch. In particular, the diameter of the extended CPD trichocysts was smaller when viewed in the SEM. T h e trichocyst makes a good test specimen because it is small enough to be easily frozen, large enough to be interesting and, in addition, many structural studies exist to provide a basis for comparison (e.g. Peterson et al., 1987). T o determine the stage at which the shrinkage occurred, the freeze-drying process was observed by LTLVSEM. T o provide good thermal and electrical conductivity, 3-mm discs punched from gold foil were initially used as specimen supports (Fig. 2). However, the uncoated specimens showed very poor contrast against the gold background unless the gold foil had first been coated with a 100-pm layer of conductive carbon paint. T h e disc was incubated with a small droplet of the trichocyst suspension for about 2min, then washed on three droplets of distilled water, frozen by plunging into liquid ethane, and stored in liquid nitrogen. The frozen specimens were then loaded into the Gatan cryo-transfer stage, introduced into the microscope and freeze-dried at 170 K until the surface of the carbon became visible ( 30 min). During the freeze-drying process, images were recorded at 1.5 kV, 150K. Although contamination was not a problem because of the low specimen temperature, we noted some indication of beam damage. For this reason, astigmatism
-
332
J . B . Pawley et al.
Fig. 2. Uncoated freeze-dried trichocysts on a gold foil at 1.5 kV showing charging artefacts that made high-magnification images impossible (scale bar = 5 pm).
and focus were adjusted on an adjacent structure and the specimen was only viewed at low magnification unless an image was being recorded. Figure 3 shows a series of images of a trichocyst as it first emerged from the receding ice surface (Fig. 3a), and after it had been warmed to room temperature and then re-cooled to 150 K to record the image (Fig. 3b). Figure 3(c) was recorded immediately after Fig. 3(b). Small differences between Fig. 3(b) and 3(c) are probably attributable to beam damage while the larger changes between Fig. 3(a) and 3(b) may be due to alterations in the structure of the trichocyst as the bound water is removed (MacKenzie, 1972; Gross, 1987).
C R Y 0 OBSERVATION OF FREEZE-SUBSTITUTED
MATERIAL
The shrinkage evident in Fig. 3 and the unavailability of an apparatus for metal coating the freeze-dried specimens at low temperature prompted us to prepare specimens by freezing/freeze-substitution/criticalpoint-drying (Walther & Hentchel, 1989). Because carbon paint is not stable in the organic solvents used in this process, the trichocysts were prepared on discs cut from 2-mm carbon rods. Otherwise, initial preparation was the same as that for freeze-drying except that the wash steps were omitted.
Early, high-resolution cryo-L V S E M
333
Fig. 3. Shrinkage of a freeze-dried trichocyst as the temperature is raised from 170K where freeze-drying had occurred (a), to room temperature (and returned to 150K to make the micrograph, b). ( c ) Made immediately after (b) and shows some additional shrinkage, probably caused by beam damage (scale bar = 200nm).
Freeze-substitution and critical point-drying were carried out as described by Miiller er al. (1980) and by Walther & Hentschel (1989):the freeze-substitution medium consisted of methanol containing 0.5'),, osmium tetroxide, 3",,glutaraldehyde and 3",,water. Samples were freeze-substituted in a Balzers FSU 010 for 7 h at 183 K, 5 h at 21 3 K, and 5 h at 243 K. After washing with cold methanol (243K), samples were warmed to room temperature, washed with ethanol and critical point-dried (Anderson, 1951;Ris, 1985)using ethanol and carbon dioxide. Samples were coated with platinum by ion beam-sputtering (Franks er al., 1980) to a thickness of less than 2nm. Figure 4 is a stereo micrograph of the specimen prepared in this manner and recorded at a magnification of x 100,000 at 1.5 kV. T h e images were recorded using the cold stage at 150 K to reduce contamination. T h e ability to produce a usable stereo image is an indication that beam damage to the specimen during the recording of the first member of the pair was not great. CONCLUSION
We have outlined a few applications of LTLVSEM and have demonstrated that it can be useful for viewing uncoated specimens at up to 50,000 x at 1.5kV during and immediately after freeze-drying. Although these images demonstrate a higher resolution than has generally been possible in the past, a further improvement is evident if the specimen can be coated. These early results are a foretaste of what will be possible from the new S E M instruments when a suitable cryo-coating mechanism is developed. ACKNOWLEDGMENT
This work was supported by NIH Grant DRR-570 to the Integrated Microscopy Resource, Madison, Wisconsin.
334
J . B . Pawley et al.
Fig. 4. High-magnification stereo-pair of trichocyst prepared by freeze-substitution/criticalpoint-drying and coated with about 2 nm of Pt by ion beam-sputtering. As the images were made with the specimen held at 150K, no apparent differences occurred between the images that can be attributed to deposition of contamination on the specimen surface during the process of making the first member of the pair (scale bar = 1OOnm).
REFERENCES Anderson, T.F. (1951) Techniques for the preservation of the three-dimensional structures in preparing specimens for the electron microscope. Trans. N Y Acad. Sci. 13, 130-133. Bannister, L.H. (1972) T h e structure of trichocysts in Paramecium rerraurelia. 3. Cell Sci. 11, 899-929. Beckett, A. & Read, N.D. (1986) Low-temperature scanning electron microscopy. Ultrastructure Techniques /or Microorganisms (ed. by H.C. Aldrich and W.J. Todd), pp. 45-86. Plenum Press, NY. Echlin, P. (1971) T h e examination of biological material at low temperatures. Scanning Electron Microsc. I, 225-232. Franks, J., Clay, C.S. & Peace, G.W. (1980) Ion beam thin film deposition. Scanning Electron Microsc. I, 155-1 62. Fujikawa, S . , Suzuki, T. & Sakurai, S. (1990) Use of a manipulator for continuous observation of frozen samples by cryoscanning electron microscopy and freeze replicas. Scanning, 12, 99-106. Gross, H. (1987) High resolution metal replication of freeze dried specimens. Cryotechniques in Biological Electron Microscopy (ed. by R. A. Steinbrecht and K. Zierold), pp. 205-215. Springer-Verlag, Berlin. Hayes, T . L . & Koch, G. (1975) Some problems associated with low temperature micromanipulation in the SEM. Scanning Electron Microsc. I, 3542. Herter, P., Tresp, G., Hentschel, H., Zierold, K. & Walther, P. (1991) High-resolution scanning electron microscopy of frozen hydrated and freeze-substituted kidney tissue. 3. Microsc. 161, 375-385. Joy, D.C. (1987) LVSEM. Electron Microscopy and Microanalysis I987 (ed. by L. M. Brown), Insr. of Phys. Con/. Series 90, 175-180. Kersker, M., Nielsen, C., Otsuji, H., Miyokawa, T . & Nakagawa, S. (1989) The JSM-890 ultra high resolution scanning electron microscope. Proc. Electron Microsc. SOC.A m . 47, 88-89. MacKenzie, A.P. (1972) Freezing, freeze drying, and freeze substitution. Scanning Electron Microsc. II, 173-1 80.
Early, high-resolution cryo-L VSEM
335
Moor, H. & Miihlethaler, K. (1963) Fine structure in frozen etched yeast cells. 3. Cell Biol. 17, 609-628. Miiller, T., Guggenheim, T., Diiggelin, M. & Liiond, G. (1986) On line cryopreparation and cryomicroscopy in SEM with SCU 020. Proc. X I 1 Int. Cong. Electron Microsc. 3 (ed. by H. Imura, S. Maruse and T. Suzuki), 2233-2234. Japanese Society of Electron Microscopy, Tokyo. Miiller, M., Marti, T . & Kriz, S. (1980) Improved structural preservation by freeze-substitution. Proc. V I I Eur. Cong. Electron Microsc. 2, 720-721. Nagatani, T. & Saito, S. (eds) (1986) Instrumentation for ultra high resolution scanning electron microscopy. Proc. Int. Cong. Electron Microsc. 111, 2201-2208. Nei, T . & Fujikawa, S. (1977) Freeze-drying process of biological specimens observed with a scanning electron microscope. 3. Elecrron Microsc. 20, 202-203. Oatley, C.W., Nixon, W.C. & Pease, R.F.W. (1965) Scanning electron microscopy. A&. Electronics Electron Phys. 21, 181-247. Pawley, J.B. (1984) Low voltage scanning electron microscopy. 3. Microsc. 13, 387-410. Pawley, J.B. & Erlandsen, S.L. (1989) The case for low voltage SEM of biological samples. Scanning Microsc. Suppl. 3, 163-178. Pawley, J., Hayes, T.L. & Hook, G. (1978) Preliminary studies of coated complementary freeze-fractured yeast membranes viewed directly in the SEM. Scanning Electron Microsc. 11, 683-690. Peterson, J.B., Heuser, J.E. & Nelson, D.L. (1987) Dissociation and reassociation of trichocyst proteins: biochemical and ultrastructural studies. 3. Cell Sci. 87, 3-25. Read, N.D. & Jeffree, C.E. (1991) Low-temperature scanning electron microscopy in biology. 3. Microsc. 161, 59-72. Ris, H. (1985) The cytoplasmic filament system in critical point dried whole mounts and plastic embedded sections. 3. Cell Biol. 100, 14741487. Steere, R.L. (1957) Electron microscopy of structural detail in frozen biological specimens. 3. Biophys. Biochem. Cytol. 3, 45-60. Steinbrecht, R.A. & Miiller, M. (1987) Freeze substitution and freeze drying. Cryotechniques in Biological Electron Microscopy (ed. by R. A. Steinbrecht and K. Zierold), pp. 149-172. Springer-Verlag, Berlin. Talmon, Y. (1984) Radiation damage to organic inclusions in ice. Ultramicroscopy, 14, 305-316. Thornley, R.F.M. (1960) Recent developments in SEM. Proc. IV Eur. Cong. Electron Microscopy (ed. by A. L. Howink and B. J. Split), pp. 173-176. Walther, P. & Hentschel, J. (1989) Improved representation of cell surface structures by freeze substitution and backscattered electron imaging. Scanning Microsc. Suppl. 3, 201-21 1. Walther, P., Hentschel, J., Herter, P., Miiller, T. & Zierold, K. (1990) Imaging of intramembranous particles in frozen-hydrated cells (Saccharomyces cerevisiae) by high resolution cryo-SEM. Scanning, 12, 300-307. Wergin, W.P. & Erbe, E.F. (1989) Increasing the versatility of an EMscope SP2000A sputter cry0 system on a Hitachi S-570 scanning electron microscope. Scanning, 11, 293-303. Zierold, K. (1983) X-ray microanalysis of frozen hydrated specimens. Scanning Electron Microscopy. Scanning Electron Microsc. 11, 809-826.
