CRYOBIOLOGY
27, 492510 (1990)
Physical Problems with the Vitrification GREGORY M, FAHY,2 JOSEPH SAUR, American
of Large Biological Systems’ AND
RO3ERT J. WILLIAMS
Red Cross Trartsplantation Laboratory, Jerome Holland Laboratory for the Biomedical 15601 Crabbs Branch Way, Rockville, Maryland 20855
Sciences,
Vitrification is an attractive potential pathway to the successful cryopreservation of mature mammalian organs, but modem cryobiological research on vitrification to date has been devoted mostly to experiments with solutions and with biological systems ranging in diameter from about 6 through about 100 km. The present paper focuses on concerns which are particularly relevant to large biological systems, i.e., those systems ranging in size from approximately 10ml to approximately 1.5 liters. New qualitative data are provided on the effect of sample size on the probability of nucleation and the ultimate size of the resulting ice crystals as well as on the probability of fracture at or below Tg. Nucleation, crystal growth, and fracture depend on cooling velocity and the magnitude of thermal gradients in the sample, which in turn depend on sample size, geometry, and cooling technique (environmental thermal history and thermal uniformity). Quantitative data on thermal gradients, cooling rates, and fracture temperatures are provided as a function of sample size. The main conclusions are as follows. First, cooling rate (from about 0.2 to about 2.PCimin) has a profound influence on the temperature-dependent processes of nucleation and crystal growth in 47~50% (w/w) solutions of propylene glycol. Second, fracturing depends strongly on cooling rate and thermal uniformity and can be postponed to about 25°C below T, for a 482-ml sample if cooling is slow and uniform. Third, the presence of a carrier solution reduces the concentration of cryoprotectant needed for vitrification (C,). However, the Cv of samples larger than about 10 ml is significantly higher than the Cv of smaller samples whether a carrier solution is present or not. 0 1990 Academic press, Inc.
A good deal of information has become available over the past several years about the vitrification and devitrification of aqueous cryoprotectant solutions and living systems. However, vitrification was developed as an alternative to freezing to solve the problems of human organ cryopreservation, whereas research to date has focused on small systems whose behavior does not reflect the problems of scaling up to the larger volumes of human organs. We now report that, unfortunately, just as the problems of freezing become more formidable when one proceeds from cells and tissues to entire organs, so also do the problems of vitrification. Although some
scaleup problems have to do with the greater complexity of organs as opposed to tissues, most of the problems specific to large biological systems arise simply because these systems are large. It is the latter type of problem that is described here. Most data were collected by placing solution samples of one of three different volumes into a low-temperature environment and recording events during cooling to below TB. Data were obtained on the volume and temperature dependence of cooling rate, thermal gradient magnitude, nucleation, ice-crystal size, fracturing, and fracture morphology. MATERIALS
Received November 10, 1989; accepted December 26, 1989. ’ Presented at the Symposium on Vitrification at the 26th Annual Meeting of the Society for Cryobiology, Charleston, South Carolina, June 1989. ’ To whom correspondence should be addressed.
Copyright 0 1990 by Academic Press, Inc. All rights of reproduction in any form reserved.
METHODS
Solurions. Solutions were made up to a total concentration of 50 or 47.6% (w/w) I ,2-propanediol (propylene glycol, PC) with or without inclusion of the carrier solution RPS-2’, which was identical to the
492 001l-2240/90 $3.00
AND
VITRIFICATION
OF LARGE
RPS-2 previously developed as a storage solution for rabbit kidneys (3, 6) except for the omission of calcium and magnesium. RPS-2= has the following composition: dextrose, 180 mM; KCl, 28.2 mM; NaHCO,, 10 mM; K2HP04, 7.2 mit4; reduced glutathione, 5 mM; adenine HCl, 1 mM. RPS-2= was incorporated into PG solutions as follows. First, the ingredients of 1 liter of RPS-2= were made up in water to a total volume of 200 ml, and this 200-ml volume was weighed, For 50% (w/w) PG, an equal weight of PG was added, and the total volume was then brought to 1 liter by the addition of a solution of 50% (w/w) PC in water to adjust the final concentration of RPS-2’ ingredients. For 47.6% (w/w) PG, the weight of PG added to the 200 ml of concentrated RPS-2= was 94% of the weight of the concentrate, and total volume was then brought to 1 liter by addition of 47% (w/w) PG in water. All solutions were filtered through Whatman No. 2 filter paper and degassed under an absolute pressure of about 200 -I 50 mm Hg for 0.5 to 1 hr (usually 30 min) before use. One liter of water held at the same pressure for 24 hr lost only 1.2% of its initial weight. After degassing, the solutions were placed in screw-capped bottles and stored in a 4°C cold room or at -20°C until they were used. Low-temperature glove box. The primary tool used in the present study is the lowtemperature glove box shown in Figs. 1 and 2. It consists of a modified Fisher ll391-100 glove box in which most of the left glove entry port has been replaced by a latchable hinged closure to permit access to the sample at cryogenic temperatures and to which has been added a window which does not fog despite internal temperatures as low as - 130 to - 160°C. The window consists of three acrylic wails separated by two insulating vapor spaces. The rear, middle, and front walls are 1, l/2, and ‘/z in. (2.54, 1.27, and 1.27 cm) in thickness, respectively. The rear vapor space is 11’2 in.
SYSTEMS
493
thick while the front vapor space is 1 in. thick. Before each experiment both vapor spaces are purged with N, gas bled from a liquid nitrogen tank’s pressure valve. The glove box is equipped with a copper heat exchange tube that distributes liquid nitrogen to the rear and upper walls (see Fig, 2) and terminates in a nozzle (not visible in Figs. 1 or 2) that sprays cold nitrogen gas and occasionally some remaining liquid nitrogen into the internal space of the glove box so as to ensure vigorous mixing of the atmosphere in the box, The internal temperature is controlled by regulating liquid nitrogen flow with a manual needle valve and is measured by means of both thermocouples and two pentane thermometers (Fisher 15-038), one suspended in the vapor overlying the sample (visible in Fig. 1) and one positioned on the shelf behind the window (Fig. 2). The walls of the box are insulated with blocks of 3.5-cm-thick styrofoam. The sample is supported on a ring stand (Fig. 2) positioned inside an internal aluminum tray (just visible in Fig. 2) that is present to collect any unevaporated liquid nitrogen. The entire glove box is placed in a similar tray (Fig. 1) that collects water from the external walls of the box after melting of the frost that deposited on the walls of the glove box during the experiments. Cooling procedure. Previously refrigerated solutions were poured into prechilled sample boxes in a 4°C cold room. They were then transferred to the low-temperature glove box (initially at room temperature), the glove box was closed, and cooling was initiated by opening the valves leading to the liquid nitrogen tank. The environmental (nitrogen vapor) temperature was set to about 20°C below Tg (i.e., to ca. - 128°C) and kept at this value until an average sample temperature approximately 4°C above Tg was attained, at which point the environmental temperature was usually raised about IO to 15°C to minimize the chances of fracture as the average sample temperature approached Tg. After approxi-
494
FAHY,
SAUR,
AND
WILLIAMS
FIG. I. Low-temperature glove box used to provide desired ambient temperatures without frost condensation on the sample and with full sample visibility for photography. Liquid nitrogen input line (L) is visible at lower right. Liquid nitrogen vaporizes as it travels through this line and continues through the copper heat exchange line in a zig-zag pattern across the back of the glove box and then up the sloped ceiling, finally emerging in a flattened, splayed nozzle which discharges the resulting vapor into the chamber to stir the atmosphere therein (nozzle not visible). Entire box is placed on blocks of foam insulation which in turn are located in a drip collector. Camera on tripod and blue backdrop for photography are visible. Sealed cracks are present on the front of the glove box and were caused by the contraction of the plastic during cooling, Side portholes were secured by metal rods to prevent the mild pressure inside the box from dislodging them. Several holes (not visible) were provided on the sides and top front of box to allow liquid nitrogen vapor to escape. Thermouple leads are seen emerging from the top. For further description, see text.
mately 30 min of annealing at about Tg, the environmental temperature was lowered until fracturing was observed. The fmai cooling rate varied but was usually established by maintaining a constant temperature difference of about 8-10°C between the environment and the center of the sample box. There were at least two runs for each combination of volume and solution composition. Photography. Solutions were photographed using a Nikon S-mm f13.5 macro lens and a Nikon F3 camera mounted on a tripod, The f stop was set to 11 for the Ek-
tachrome 200 film, with a nominal exposure time of 1/60th of a second. A Vivitar 285 automatic flash attachment was used. Photographs were taken whenever required rather than at fixed temperature intervals to document nucleation, crystal growth, and fracturing adequately. Several photographs were taken for each selected temperature to insure optimal focusing and to examine the effects of different illumination angles. The most successful approach was to illuminate from behind using light reflected from the back wall after being directed through the frosted side wall of the
VITRIFICATION
OF LARGE SYSTEMS
495
FIG. 2. Low-temperature glove box as seen from the side, showing the side portholes, the internal heat exchange coils, the internal spill tray, the internal sample tripod, and the blue background for photography. The retaining rods for the far porthole cover are visible, as is a shelf located just in front of the sample tripod. In use, the shelf supported a second pentane-in-glass thermometer. Finally, pressure relief holes are visible at upper left and right comers. For further discussion, see text.
box, but it was also necessary to take additional photographs at the same time in which the sample was illuminated from the front at a glancing angle to ensure that all ice particles could be detected. Sample containers. The sample containers are described in Fig. 3 and Table 1. The containers were made in the shape of boxes to allow photography through an optically flat surface. The sample boxes were designed to have two compartments, a lower compartment with a defined temperature field which was the main object of study and an upper compartment whose purpose was to act as a buffer between the lower compartment and the air-solution interface. During cooling, considerable volume contraction of the test solutions takes place. To prevent intrusion of the air-liquid interface into the lower compartment, additional test solution was added to the up-
per compartment during cooling. The extra solution was prechilled to about - 20°C in a 30- or 60-ml syringe used to introduce the solution. This procedure tended to introduce ice into the upper compartment and introduced thermal transients into the upper compartment, but did not significantly affect the lower compartment. In the case of the 1412-ml container, the upper chamber consisted of a tunnel (CD in Fig. 3) cut through a thick block of plastic. The 1412-ml box was selected to have a width equal to the diameter of a 600-ml beaker (which is capable of containing a human kidney) and a length twice the width. The 482-ml box was selected to match the dimensions of a human kidney, and the 46-ml box was selected to be more than large enough to contain a rabbit kidney and its associated cannulae. Thermometry. The circled numbers in
FAHY,
SAUR,
AND
WILLIAMS
UH I
LH 1
FIG. 3. Schematic of sample container layout indicating relative positions of temperature sensors and dimensions of the three boxes used in the study. Sample containers consisted of two compartments, an upper “buffer” compartment and a lower “isolated” compartment, both of which were fitted with removable lids. The internal, lower lids were perforated to permit upper chamber fluid to flow into the lower chamber as volume contraction took place in the latter (perforations not indicated in the figure). The right portion of the figure indicates a cross-sectional view of the lower chamber as seen from above [i.e., probe 8 is located at the front of the sample (near the camera)]. The dimensions LH, UH, IL, IW, and CD refer to the “lower height,” “upper height,” “inner length,” “inner width,” and “channel diameter” as indicated. The quantitative values of these dimensions and the coordinates of the probe positions are given in Table I, The coordinates O,O,bshown in the left figure indicate that the coordinates IL and LH are zero at the left lower comer of the box and that the coordinate IW cannot be seen in that figure. Similarly, the coordinates 0,&O at the right indicate that the coordinates IL and IW at the left front comer of the box are zero, but that the coordinate LH cannot be depicted in that figure. The volumes of the three lower compartments were 46,482, and 1412 ml and the respective volumes of the upper compartments were 73, 114, and 20 ml.
Fig 3 and the coordinates in Table 1 indicate the positions of the thermocouples. For technical reasons recordings were not always made at all positions in a given sample box, but consistent values for average sample temperature and average sample cooling rates from run to run were nevertheless obtained, In most experiments the thermocouple wells consisted of O-054-in. (1.37mm)-inner-diameter, 0.072-in. (1.83mm)-outer-diameter metal tubing (No.15 gauge hypodermic tubing), and were welded shut at the immersed ends. Thermocouples were made of 28-gauge copperconstantan thermocouple wire (Revere Corp., Wallingford, CT) and were connetted directly to a temperature logger and plotter (DianaChart, Inc., Saddlebrook, NJ) which recorded temperatures both as continuous graphs and digitally every 5
min. The temperature distribution in the sample container was also recorded at the time each set of photographs was taken. Average sample temperature was estimated arbitrarily by averaging the front-to-back, some of the side-to-side, and, when available, the upper and lower probe temperatures. Cooling rates were estimated at particular temperatures by dividing the temperature interval bracketing the temperature of interest by the time interval over which the bracketing temperature decrement occurred. Cryomicroscopy ning calorimetry
and
differential
scan-
(DSC). Direct observations of ice-crystal growth rates were made in the case of 50% (w/w) FG in water using a Linkam cryomicroscope (MicroDevices, Inc., Jenkintown, PA). To accomplish this, the sample was cooled at lO”C/min to
VITRIFICATION
497
OF LARGE SYSTEMS
TABLE I Sample Container Dimensions and Probe Coordinates Probe coordinates (mm)
Container dimensions
(mm)
482 ml
46 ml
1412 ml
Dimensio&
46 ml
482 ml
1412ml
Probe
IL
LH
IW
IL
LH
IW
IW IL LH UH CD
28 40 40 60 NP
50.8 82.6 114.3 25.9 NP
86.2 171.2 98.4 38.1 28.0
I 2 3 4 5
5 NP= 20 NP 35
20 NP 20 NP 20
14 NP 14 NP 14
5 NP 41.3 NP 77.6
57 NP 57 NP 57
25.4 NP 25.4 NP 25.4
6 7
12.5 27.5
5 35
14 14
25.5 57.1
5 25.4 109.3 25.4
20 NP NP 20
20 NP NP 20
5 NP NP 23
41.3 41.3 41.3 41.3
8 9 10 11
57 57 57 5-l
5 10.2 35.6 45.8
IL
LH
IW
NP NP NP 42.8 50.8 43.1 85.6 50.8 43.1 128.4 50.8 43.1 NP NP NP 42.8 25.4 128.4 76.2
43.1 43.1
85.6 50.8 85.6 50.8 85.6 50.8 85.6 50.8
5 20.9 65.3 81.2
a Coordinates given in the directions of IL, LH, IW, where (0, 0,O) is the left lower front comer and (IL, LH, IW) is the right upper rear comer. Probes 1-5 are arranged from left to right, 6 and 7 from bottom to top, and 8-l 1 from front to back. b See Fig. 3 for definitions. ’ Not present.
- 50°C and then at IWrnin to - 1WC, then warmed at S”C/min to -70°C. The sample was held at -70°C and scanned until an ice crystal was found, at which time a videotape was made of the growing crystal. After data collection at -7O”C, the temperature was lowered to - 80°C and then to - 90°C for further observations. Growth rate was measured by measuring crystal boundary displacement per unit time (km/m@ directly off the video monitor on replaying the videotape. Precision of the measurement was about &0.7% at - 70°C; accuracy and/or variations in apparent rate measured at different places on the interface were I l-14% at - 70°C, 8% at - SO”C,and 22% at - 90°C. The rates reported are rates at the fastest growing crystal face. Rates on more slowly growing faces ranged from 53 to 66% of the fastest observed rates. Glass transition temperatures were determined by DSC (Perkin-Elmer DSC-4, Norwalk, CT) under computer control.
(w/w) PG in either water (heavy, discontinuous curve) or RPS-2= (thin, continuous curve) at a cooling rate of 1”Clmin. The presence of the carrier solution raised Tg by about l”C, from about - 108 to about - 107”C, and possibly increased the size of the heat capacity change at the transition, For both solutions, the glass transition was completed near - 117°C. These results provide a number of reference points for our subsequent studies and for comparison with the work of Boutron and Kaufmann (2).
Figure 5 gives the average sample cooling rates for all runs, expressed as a function of volume and the average temperature of the sample. Cooling rate varied strongly with both temperature and volume. For 46-ml samples, cooling rate at -30°C was near 2.5Wmin and fell to about 0.7”Cimin near Tg, while cooling rates of the 482-ml samples were about half those of the 46-ml samples down to about - 12O”C, and cooling rates of the two 1412-ml samples averaged a RESULTS little over half those of the 482-ml samples Figure 4 shows DSC cooling runs on 50% down to about -90°C. These results point
498
FAHY,
Tg-107B°C
-
-
SAUR,
AND
environmental temperature remains close to Tg during cooling. As will become clear from the data which follow, the latter restriction is necessary to avoid fracturing of the sample. Figure 6 gives the averaged temperature differences from wall to center (lower portions of panels A-C) and from wall to opposite wall (upper portions of panels A and B) for each volume as a function of temperature at the center of the sample. The wall to opposite wall data provide a measure of the uniformity of the temperature field surrounding the sample. Significant temperature differentials existed not only in the samples themselves but in their environment due to the early stage of development of our equipment and techniques. Since the upper probe was beneath the upper reservoir, a top to bottom uniformity measurement was less meaningful than a top to center temperature difference measurement (Fig. 6C). Not surprisingly, larger temperature differences were observed for larger samples, some differences approaching 25°C. The temperature differentials in the
Tg~106XS0C
Ii
0.501
I -90
I -110
-120
TEMPERATURE
-100
WILLIAMS
Co0
FIG. 4. Calorimeter scans on cooling of 50% (w/w) PG in water (heavy, discontinuous curve) or in RPS-2= (light continuous curve). The temperature of onset of the glass transition is raised by about 1°C by the presence of the carrier solution. Baselines and signal slopes were estimated by eye due to the noise in the record which precluded accurate estimations by the DSC’s computer.
out an inevitable and unfavorable reality of large-volume vitrification experiments: it is impossible to avoid steadily declining cooling rates while approaching T, if the 30
o 46ml
2 r 25 e
l
482ml
* 1412ml 20
,I
-20
-40
-60
-80
-100
-120
-140
-160
TEMPERATURE (‘C) FIG. 5. Dependence of average sample cooling rate on sample volume and average temperature. Data from each experiment plotted independently. Ambient temperature was set at a value of about 20°C below T. and held until average sample temperature fell close to T,. This was necessary to prevent surface fracturing before vitrification of the center of the samples could occur, and necessarily resulted in steadily declining cooling rates. Unlike a small sample, which can be cooled at a constant rate to below TB, large samples must necessarily be cooled at steadily decreasing rates as indicated in this figure.
VITRIFICATION
OF LARGE
direction IW were less than the differentials from lateral wall to center and from bottom wall to center despite similar ambient temperature uniformity in these directions. This is presumably because IW was the smallest dimension of the sample. Panel D of Fig. 6 relates the average sample temperature used in Fig. 5 to the temperature recorded at the center of the box, thereby allowing the remaining panels of Fig. 6 to be related to Fig. 5. The temperature differences of Fig. 6 were divided by the distances separating the relevant probes (from Table I) to obtain average thermal gradients, and the results are shown in Fig, 7. As in Fig 6, the largest average difference between the 46- and 482ml samples was in the vertical (B-C) direction. The fracturing temperatures are given in Table 2. As noted in Table 2, fracture temperature was highly dependent upon cooling conditions (cooling rate and thermal gradient) and could be depressed to surprisingly low temperatures depending on the sample volume. An interesting relationship was observed between the temperature of fracturing and the morphology of the fractures: the lower the fracture temperature, the more finely subdivided the sample became, i.e., the more extensive was the observed fracturing, This effect can be seen in Fig. 8, which shows fracture patterns seen for 482-ml samples which fractured at - 113°C (A), - 124°C (B), and - 134°C (C). Similar patterns were seen for the other volumes studied. Fractures in the 1412-ml samples, which formed at the highest temperatures (Table 2), resembled those of Fig. HA, while the fractures in the 46-ml samples, which formed at the lowest temperatures (Table 21, were always very similar to those of Fig. SC. The final and for the present purposes most important information is contained in Figs, 9 through I 1, which show the evolution of ice-crystal number and size as a function of temperature, sample volume,
SYSTEMS
499
and solution composition. Figure 9 shows the effect of size on nucleation and crystal growth in 50% (w/w) PG in water. Despite a lack of evidence for crystallization during cooling at l”C/min in a typical DSC sample, ice is clearly visible in Figure 9 at all sample volumes, including the 46-ml volume, which sustained a cooling rate in excess of l”C/min down to about - 90°C. The number and size of ice crystals formed increased dramatically as volume was increased to 482 and to 1412 ml. Ice crystals which nucleated at higher temperatures (clearly by a heterogeneous mechanism) were able to grow to rather large sizes only in the 482and 1412-ml volumes: evidently the cooling rate difference between the 46- and 483-ml samples was enough to prevent extensive crystal growth in the smaller samples. The rate of appearance of visible ice crystals increased as temperature fell to about -70 to -90°C. Below these temperatures, no additional visible ice crystals formed. At higher temperatures, crystal growth rate visibly decreased as temperature decreased, suggesting that new crystals might be forming below -70 to -90°C but might be unable to grow to visible size at these temperatures within the time of observation. Cryomicroscopic observations supported these macroscopic observations: directly measured ice crystal growth rates in 50% (w/w) PG in water were found to be 28.0 km/min at -7O”C, 4.84 km/min at -80°C and 0.48 km/min at -90°C. Given visibility of crystals on the order of 0.1 mm across, any new crystals should become visible in 4 min at - 70°C 21 min at - 80°C and 208 min at - 90°C. Another interesting point illustrated by Fig. 9 is the volume contraction of the solution during cooling. At the lowest temperatures shown, menisci have penetrated into the lower compartments of the two largest containers despite efforts to prevent such intrusion. The menisci occur in the centers of the samples, since this is where temperature is the highest and viscosity is the lowest, In the case of the 46-ml sample (Fig.
500
FAHY, SAUR, AND WILLIAMS CENTER TEMPERATURE
-20
-40
-60
-80
-100
-120
-140
Back @3-C) . . . . . to. . Center .._____..._ 1
-20
-40
CENTER TEMPERATURE -60 -80 -100
-120
-140
FIG. 6. Effects of volume and average temperature on temperature differences from the sample surface to the sample center (points connected by dotted lines in panels A and 3; ah data in panel C) or from wall to opposite wall (points connected by solid lines in panels A and B). Ba-C = temperature at the back of the container {probe 11) minus the remperature at the center of the container (prabe 3) ; F-Ba = front temperature {probe 8) minus back temperature {probe 1I), and so on. R = right (probe S), L = left (probe I), T = top (probe 7), and B = bottom (probe 6). Because the top temperature probe was not at the external wall near the vapor environment but was instead at the upper portion of the lower compartment, panel C shows both B - C (dotted lines) and T - C (solid lines) rather than 3 - C and T - B. Error bars for 1412-ml samples in lower portion of panel A were omitted for clarity. These standard errors (n = 2) were in the range of *lO”c. Horimntal standard errors (not shown) of ~2.5”C pert& to all points. Panel D: relationship between center temperatures of Fig. 6 and the average temperatures of Fig. 5.
SC), the gas cavities in the lower chamber just under the lower lid appear to have resulted from cavitation of the solution (bubble nucleation and growth), indicating the
large magnitude of the tensile stresses being developed. Figure 10 illustrates the effects of incorporating a standard renal carrier solution,
VITRIFKATION
OF LARGE
501
SYSTEMS
Topto Center -WI _--.w.:. .. P
_ T
-20 -40 -60
-100 -80 CENTER TEMPERATURE
-40
-60
-80 -100 CENTER TEMPERATURE FIG. 6-Continued
RPS-2=, in our 50% (w/w) PG solution. The presence of the carrier solution profoundly suppresses the number and size of visible ice crystals in 482- and 46-ml samples (1412-ml samples were not examined). This is hardly surprising given that the presence of RPS-2= raises total solute content from 50% (w/w) to something on the order of 54% (w/w). Figure 10B is included to demonstrate the lack of visible change associated with solidification of the sample (compare with Fig. lOA). Figure 11 shows the effect of partially compensating for the presence of RPS-2=
-120
-120
-140
-140
by reducing PG concentration to 47.6% (w/ w). The results appeared to be better than those obtained with 50% PG in water, as expected. Thus, the roughly 4% solute represented by RPS-2= is slightly more effective than 2.4% (w/w) cryoprotectant in its ability to facilitate vitrification. The implication is that carrier solution solutes significantly lower the concentration of cryoprotectant required for vitrification, as previously noted (4). DISCUSSION
This study was undertaken due to the
FAHY. SAUR, AND WILLIAMS
LLI 8
Back
-8 0.0
to Center
E -2.0 sLLI -3.0 LLI
-1.0
-4.0
a
-5.0
e E
-6.04
2 -4 -8
LLI r
l-
-12 -16 -20 .20
-40
-60
CENTER
-80
-100
-120
-740
TEMPERATURE
FIG. 7. Average thermal gradients through the samples. The temperature differences shown in Fig. 6 were divided by the probe separation distances (obtainable from Table 1) to obtain average thermal gradient in “C/cm. For B - C and T - C gradients, the vertical distance between the probes was used for calculating thermal gradient, not the actual probe separation distance.
very practical need to know how large volumes of concentrated cryoprotectant solutions behave with respect to fracture temperature, cooling rate, thermal gradient, ice nucleation, and crystal growth so that it can become possible to avoid fracturing and de-
vise means of coping with nucleation and growth. These data are important not only for understanding events during cooling but also for modeling devitrification, which ultimately requires knowledge of ice crystal growth rates as a function of temperature.
TABLE 2 Fracture Temperatures Volume (ml)
individual fracture temperatures (“C)
Mean
46 482 1412
-134.4, -136.7, -139.3, -142.3, -143.3, -I46.6,
27, 492510 (1990)
Physical Problems with the Vitrification GREGORY M, FAHY,2 JOSEPH SAUR, American
of Large Biological Systems’ AND
RO3ERT J. WILLIAMS
Red Cross Trartsplantation Laboratory, Jerome Holland Laboratory for the Biomedical 15601 Crabbs Branch Way, Rockville, Maryland 20855
Sciences,
Vitrification is an attractive potential pathway to the successful cryopreservation of mature mammalian organs, but modem cryobiological research on vitrification to date has been devoted mostly to experiments with solutions and with biological systems ranging in diameter from about 6 through about 100 km. The present paper focuses on concerns which are particularly relevant to large biological systems, i.e., those systems ranging in size from approximately 10ml to approximately 1.5 liters. New qualitative data are provided on the effect of sample size on the probability of nucleation and the ultimate size of the resulting ice crystals as well as on the probability of fracture at or below Tg. Nucleation, crystal growth, and fracture depend on cooling velocity and the magnitude of thermal gradients in the sample, which in turn depend on sample size, geometry, and cooling technique (environmental thermal history and thermal uniformity). Quantitative data on thermal gradients, cooling rates, and fracture temperatures are provided as a function of sample size. The main conclusions are as follows. First, cooling rate (from about 0.2 to about 2.PCimin) has a profound influence on the temperature-dependent processes of nucleation and crystal growth in 47~50% (w/w) solutions of propylene glycol. Second, fracturing depends strongly on cooling rate and thermal uniformity and can be postponed to about 25°C below T, for a 482-ml sample if cooling is slow and uniform. Third, the presence of a carrier solution reduces the concentration of cryoprotectant needed for vitrification (C,). However, the Cv of samples larger than about 10 ml is significantly higher than the Cv of smaller samples whether a carrier solution is present or not. 0 1990 Academic press, Inc.
A good deal of information has become available over the past several years about the vitrification and devitrification of aqueous cryoprotectant solutions and living systems. However, vitrification was developed as an alternative to freezing to solve the problems of human organ cryopreservation, whereas research to date has focused on small systems whose behavior does not reflect the problems of scaling up to the larger volumes of human organs. We now report that, unfortunately, just as the problems of freezing become more formidable when one proceeds from cells and tissues to entire organs, so also do the problems of vitrification. Although some
scaleup problems have to do with the greater complexity of organs as opposed to tissues, most of the problems specific to large biological systems arise simply because these systems are large. It is the latter type of problem that is described here. Most data were collected by placing solution samples of one of three different volumes into a low-temperature environment and recording events during cooling to below TB. Data were obtained on the volume and temperature dependence of cooling rate, thermal gradient magnitude, nucleation, ice-crystal size, fracturing, and fracture morphology. MATERIALS
Received November 10, 1989; accepted December 26, 1989. ’ Presented at the Symposium on Vitrification at the 26th Annual Meeting of the Society for Cryobiology, Charleston, South Carolina, June 1989. ’ To whom correspondence should be addressed.
Copyright 0 1990 by Academic Press, Inc. All rights of reproduction in any form reserved.
METHODS
Solurions. Solutions were made up to a total concentration of 50 or 47.6% (w/w) I ,2-propanediol (propylene glycol, PC) with or without inclusion of the carrier solution RPS-2’, which was identical to the
492 001l-2240/90 $3.00
AND
VITRIFICATION
OF LARGE
RPS-2 previously developed as a storage solution for rabbit kidneys (3, 6) except for the omission of calcium and magnesium. RPS-2= has the following composition: dextrose, 180 mM; KCl, 28.2 mM; NaHCO,, 10 mM; K2HP04, 7.2 mit4; reduced glutathione, 5 mM; adenine HCl, 1 mM. RPS-2= was incorporated into PG solutions as follows. First, the ingredients of 1 liter of RPS-2= were made up in water to a total volume of 200 ml, and this 200-ml volume was weighed, For 50% (w/w) PG, an equal weight of PG was added, and the total volume was then brought to 1 liter by the addition of a solution of 50% (w/w) PC in water to adjust the final concentration of RPS-2’ ingredients. For 47.6% (w/w) PG, the weight of PG added to the 200 ml of concentrated RPS-2= was 94% of the weight of the concentrate, and total volume was then brought to 1 liter by addition of 47% (w/w) PG in water. All solutions were filtered through Whatman No. 2 filter paper and degassed under an absolute pressure of about 200 -I 50 mm Hg for 0.5 to 1 hr (usually 30 min) before use. One liter of water held at the same pressure for 24 hr lost only 1.2% of its initial weight. After degassing, the solutions were placed in screw-capped bottles and stored in a 4°C cold room or at -20°C until they were used. Low-temperature glove box. The primary tool used in the present study is the lowtemperature glove box shown in Figs. 1 and 2. It consists of a modified Fisher ll391-100 glove box in which most of the left glove entry port has been replaced by a latchable hinged closure to permit access to the sample at cryogenic temperatures and to which has been added a window which does not fog despite internal temperatures as low as - 130 to - 160°C. The window consists of three acrylic wails separated by two insulating vapor spaces. The rear, middle, and front walls are 1, l/2, and ‘/z in. (2.54, 1.27, and 1.27 cm) in thickness, respectively. The rear vapor space is 11’2 in.
SYSTEMS
493
thick while the front vapor space is 1 in. thick. Before each experiment both vapor spaces are purged with N, gas bled from a liquid nitrogen tank’s pressure valve. The glove box is equipped with a copper heat exchange tube that distributes liquid nitrogen to the rear and upper walls (see Fig, 2) and terminates in a nozzle (not visible in Figs. 1 or 2) that sprays cold nitrogen gas and occasionally some remaining liquid nitrogen into the internal space of the glove box so as to ensure vigorous mixing of the atmosphere in the box, The internal temperature is controlled by regulating liquid nitrogen flow with a manual needle valve and is measured by means of both thermocouples and two pentane thermometers (Fisher 15-038), one suspended in the vapor overlying the sample (visible in Fig. 1) and one positioned on the shelf behind the window (Fig. 2). The walls of the box are insulated with blocks of 3.5-cm-thick styrofoam. The sample is supported on a ring stand (Fig. 2) positioned inside an internal aluminum tray (just visible in Fig. 2) that is present to collect any unevaporated liquid nitrogen. The entire glove box is placed in a similar tray (Fig. 1) that collects water from the external walls of the box after melting of the frost that deposited on the walls of the glove box during the experiments. Cooling procedure. Previously refrigerated solutions were poured into prechilled sample boxes in a 4°C cold room. They were then transferred to the low-temperature glove box (initially at room temperature), the glove box was closed, and cooling was initiated by opening the valves leading to the liquid nitrogen tank. The environmental (nitrogen vapor) temperature was set to about 20°C below Tg (i.e., to ca. - 128°C) and kept at this value until an average sample temperature approximately 4°C above Tg was attained, at which point the environmental temperature was usually raised about IO to 15°C to minimize the chances of fracture as the average sample temperature approached Tg. After approxi-
494
FAHY,
SAUR,
AND
WILLIAMS
FIG. I. Low-temperature glove box used to provide desired ambient temperatures without frost condensation on the sample and with full sample visibility for photography. Liquid nitrogen input line (L) is visible at lower right. Liquid nitrogen vaporizes as it travels through this line and continues through the copper heat exchange line in a zig-zag pattern across the back of the glove box and then up the sloped ceiling, finally emerging in a flattened, splayed nozzle which discharges the resulting vapor into the chamber to stir the atmosphere therein (nozzle not visible). Entire box is placed on blocks of foam insulation which in turn are located in a drip collector. Camera on tripod and blue backdrop for photography are visible. Sealed cracks are present on the front of the glove box and were caused by the contraction of the plastic during cooling, Side portholes were secured by metal rods to prevent the mild pressure inside the box from dislodging them. Several holes (not visible) were provided on the sides and top front of box to allow liquid nitrogen vapor to escape. Thermouple leads are seen emerging from the top. For further description, see text.
mately 30 min of annealing at about Tg, the environmental temperature was lowered until fracturing was observed. The fmai cooling rate varied but was usually established by maintaining a constant temperature difference of about 8-10°C between the environment and the center of the sample box. There were at least two runs for each combination of volume and solution composition. Photography. Solutions were photographed using a Nikon S-mm f13.5 macro lens and a Nikon F3 camera mounted on a tripod, The f stop was set to 11 for the Ek-
tachrome 200 film, with a nominal exposure time of 1/60th of a second. A Vivitar 285 automatic flash attachment was used. Photographs were taken whenever required rather than at fixed temperature intervals to document nucleation, crystal growth, and fracturing adequately. Several photographs were taken for each selected temperature to insure optimal focusing and to examine the effects of different illumination angles. The most successful approach was to illuminate from behind using light reflected from the back wall after being directed through the frosted side wall of the
VITRIFICATION
OF LARGE SYSTEMS
495
FIG. 2. Low-temperature glove box as seen from the side, showing the side portholes, the internal heat exchange coils, the internal spill tray, the internal sample tripod, and the blue background for photography. The retaining rods for the far porthole cover are visible, as is a shelf located just in front of the sample tripod. In use, the shelf supported a second pentane-in-glass thermometer. Finally, pressure relief holes are visible at upper left and right comers. For further discussion, see text.
box, but it was also necessary to take additional photographs at the same time in which the sample was illuminated from the front at a glancing angle to ensure that all ice particles could be detected. Sample containers. The sample containers are described in Fig. 3 and Table 1. The containers were made in the shape of boxes to allow photography through an optically flat surface. The sample boxes were designed to have two compartments, a lower compartment with a defined temperature field which was the main object of study and an upper compartment whose purpose was to act as a buffer between the lower compartment and the air-solution interface. During cooling, considerable volume contraction of the test solutions takes place. To prevent intrusion of the air-liquid interface into the lower compartment, additional test solution was added to the up-
per compartment during cooling. The extra solution was prechilled to about - 20°C in a 30- or 60-ml syringe used to introduce the solution. This procedure tended to introduce ice into the upper compartment and introduced thermal transients into the upper compartment, but did not significantly affect the lower compartment. In the case of the 1412-ml container, the upper chamber consisted of a tunnel (CD in Fig. 3) cut through a thick block of plastic. The 1412-ml box was selected to have a width equal to the diameter of a 600-ml beaker (which is capable of containing a human kidney) and a length twice the width. The 482-ml box was selected to match the dimensions of a human kidney, and the 46-ml box was selected to be more than large enough to contain a rabbit kidney and its associated cannulae. Thermometry. The circled numbers in
FAHY,
SAUR,
AND
WILLIAMS
UH I
LH 1
FIG. 3. Schematic of sample container layout indicating relative positions of temperature sensors and dimensions of the three boxes used in the study. Sample containers consisted of two compartments, an upper “buffer” compartment and a lower “isolated” compartment, both of which were fitted with removable lids. The internal, lower lids were perforated to permit upper chamber fluid to flow into the lower chamber as volume contraction took place in the latter (perforations not indicated in the figure). The right portion of the figure indicates a cross-sectional view of the lower chamber as seen from above [i.e., probe 8 is located at the front of the sample (near the camera)]. The dimensions LH, UH, IL, IW, and CD refer to the “lower height,” “upper height,” “inner length,” “inner width,” and “channel diameter” as indicated. The quantitative values of these dimensions and the coordinates of the probe positions are given in Table I, The coordinates O,O,bshown in the left figure indicate that the coordinates IL and LH are zero at the left lower comer of the box and that the coordinate IW cannot be seen in that figure. Similarly, the coordinates 0,&O at the right indicate that the coordinates IL and IW at the left front comer of the box are zero, but that the coordinate LH cannot be depicted in that figure. The volumes of the three lower compartments were 46,482, and 1412 ml and the respective volumes of the upper compartments were 73, 114, and 20 ml.
Fig 3 and the coordinates in Table 1 indicate the positions of the thermocouples. For technical reasons recordings were not always made at all positions in a given sample box, but consistent values for average sample temperature and average sample cooling rates from run to run were nevertheless obtained, In most experiments the thermocouple wells consisted of O-054-in. (1.37mm)-inner-diameter, 0.072-in. (1.83mm)-outer-diameter metal tubing (No.15 gauge hypodermic tubing), and were welded shut at the immersed ends. Thermocouples were made of 28-gauge copperconstantan thermocouple wire (Revere Corp., Wallingford, CT) and were connetted directly to a temperature logger and plotter (DianaChart, Inc., Saddlebrook, NJ) which recorded temperatures both as continuous graphs and digitally every 5
min. The temperature distribution in the sample container was also recorded at the time each set of photographs was taken. Average sample temperature was estimated arbitrarily by averaging the front-to-back, some of the side-to-side, and, when available, the upper and lower probe temperatures. Cooling rates were estimated at particular temperatures by dividing the temperature interval bracketing the temperature of interest by the time interval over which the bracketing temperature decrement occurred. Cryomicroscopy ning calorimetry
and
differential
scan-
(DSC). Direct observations of ice-crystal growth rates were made in the case of 50% (w/w) FG in water using a Linkam cryomicroscope (MicroDevices, Inc., Jenkintown, PA). To accomplish this, the sample was cooled at lO”C/min to
VITRIFICATION
497
OF LARGE SYSTEMS
TABLE I Sample Container Dimensions and Probe Coordinates Probe coordinates (mm)
Container dimensions
(mm)
482 ml
46 ml
1412 ml
Dimensio&
46 ml
482 ml
1412ml
Probe
IL
LH
IW
IL
LH
IW
IW IL LH UH CD
28 40 40 60 NP
50.8 82.6 114.3 25.9 NP
86.2 171.2 98.4 38.1 28.0
I 2 3 4 5
5 NP= 20 NP 35
20 NP 20 NP 20
14 NP 14 NP 14
5 NP 41.3 NP 77.6
57 NP 57 NP 57
25.4 NP 25.4 NP 25.4
6 7
12.5 27.5
5 35
14 14
25.5 57.1
5 25.4 109.3 25.4
20 NP NP 20
20 NP NP 20
5 NP NP 23
41.3 41.3 41.3 41.3
8 9 10 11
57 57 57 5-l
5 10.2 35.6 45.8
IL
LH
IW
NP NP NP 42.8 50.8 43.1 85.6 50.8 43.1 128.4 50.8 43.1 NP NP NP 42.8 25.4 128.4 76.2
43.1 43.1
85.6 50.8 85.6 50.8 85.6 50.8 85.6 50.8
5 20.9 65.3 81.2
a Coordinates given in the directions of IL, LH, IW, where (0, 0,O) is the left lower front comer and (IL, LH, IW) is the right upper rear comer. Probes 1-5 are arranged from left to right, 6 and 7 from bottom to top, and 8-l 1 from front to back. b See Fig. 3 for definitions. ’ Not present.
- 50°C and then at IWrnin to - 1WC, then warmed at S”C/min to -70°C. The sample was held at -70°C and scanned until an ice crystal was found, at which time a videotape was made of the growing crystal. After data collection at -7O”C, the temperature was lowered to - 80°C and then to - 90°C for further observations. Growth rate was measured by measuring crystal boundary displacement per unit time (km/m@ directly off the video monitor on replaying the videotape. Precision of the measurement was about &0.7% at - 70°C; accuracy and/or variations in apparent rate measured at different places on the interface were I l-14% at - 70°C, 8% at - SO”C,and 22% at - 90°C. The rates reported are rates at the fastest growing crystal face. Rates on more slowly growing faces ranged from 53 to 66% of the fastest observed rates. Glass transition temperatures were determined by DSC (Perkin-Elmer DSC-4, Norwalk, CT) under computer control.
(w/w) PG in either water (heavy, discontinuous curve) or RPS-2= (thin, continuous curve) at a cooling rate of 1”Clmin. The presence of the carrier solution raised Tg by about l”C, from about - 108 to about - 107”C, and possibly increased the size of the heat capacity change at the transition, For both solutions, the glass transition was completed near - 117°C. These results provide a number of reference points for our subsequent studies and for comparison with the work of Boutron and Kaufmann (2).
Figure 5 gives the average sample cooling rates for all runs, expressed as a function of volume and the average temperature of the sample. Cooling rate varied strongly with both temperature and volume. For 46-ml samples, cooling rate at -30°C was near 2.5Wmin and fell to about 0.7”Cimin near Tg, while cooling rates of the 482-ml samples were about half those of the 46-ml samples down to about - 12O”C, and cooling rates of the two 1412-ml samples averaged a RESULTS little over half those of the 482-ml samples Figure 4 shows DSC cooling runs on 50% down to about -90°C. These results point
498
FAHY,
Tg-107B°C
-
-
SAUR,
AND
environmental temperature remains close to Tg during cooling. As will become clear from the data which follow, the latter restriction is necessary to avoid fracturing of the sample. Figure 6 gives the averaged temperature differences from wall to center (lower portions of panels A-C) and from wall to opposite wall (upper portions of panels A and B) for each volume as a function of temperature at the center of the sample. The wall to opposite wall data provide a measure of the uniformity of the temperature field surrounding the sample. Significant temperature differentials existed not only in the samples themselves but in their environment due to the early stage of development of our equipment and techniques. Since the upper probe was beneath the upper reservoir, a top to bottom uniformity measurement was less meaningful than a top to center temperature difference measurement (Fig. 6C). Not surprisingly, larger temperature differences were observed for larger samples, some differences approaching 25°C. The temperature differentials in the
Tg~106XS0C
Ii
0.501
I -90
I -110
-120
TEMPERATURE
-100
WILLIAMS
Co0
FIG. 4. Calorimeter scans on cooling of 50% (w/w) PG in water (heavy, discontinuous curve) or in RPS-2= (light continuous curve). The temperature of onset of the glass transition is raised by about 1°C by the presence of the carrier solution. Baselines and signal slopes were estimated by eye due to the noise in the record which precluded accurate estimations by the DSC’s computer.
out an inevitable and unfavorable reality of large-volume vitrification experiments: it is impossible to avoid steadily declining cooling rates while approaching T, if the 30
o 46ml
2 r 25 e
l
482ml
* 1412ml 20
,I
-20
-40
-60
-80
-100
-120
-140
-160
TEMPERATURE (‘C) FIG. 5. Dependence of average sample cooling rate on sample volume and average temperature. Data from each experiment plotted independently. Ambient temperature was set at a value of about 20°C below T. and held until average sample temperature fell close to T,. This was necessary to prevent surface fracturing before vitrification of the center of the samples could occur, and necessarily resulted in steadily declining cooling rates. Unlike a small sample, which can be cooled at a constant rate to below TB, large samples must necessarily be cooled at steadily decreasing rates as indicated in this figure.
VITRIFICATION
OF LARGE
direction IW were less than the differentials from lateral wall to center and from bottom wall to center despite similar ambient temperature uniformity in these directions. This is presumably because IW was the smallest dimension of the sample. Panel D of Fig. 6 relates the average sample temperature used in Fig. 5 to the temperature recorded at the center of the box, thereby allowing the remaining panels of Fig. 6 to be related to Fig. 5. The temperature differences of Fig. 6 were divided by the distances separating the relevant probes (from Table I) to obtain average thermal gradients, and the results are shown in Fig, 7. As in Fig 6, the largest average difference between the 46- and 482ml samples was in the vertical (B-C) direction. The fracturing temperatures are given in Table 2. As noted in Table 2, fracture temperature was highly dependent upon cooling conditions (cooling rate and thermal gradient) and could be depressed to surprisingly low temperatures depending on the sample volume. An interesting relationship was observed between the temperature of fracturing and the morphology of the fractures: the lower the fracture temperature, the more finely subdivided the sample became, i.e., the more extensive was the observed fracturing, This effect can be seen in Fig. 8, which shows fracture patterns seen for 482-ml samples which fractured at - 113°C (A), - 124°C (B), and - 134°C (C). Similar patterns were seen for the other volumes studied. Fractures in the 1412-ml samples, which formed at the highest temperatures (Table 2), resembled those of Fig. HA, while the fractures in the 46-ml samples, which formed at the lowest temperatures (Table 21, were always very similar to those of Fig. SC. The final and for the present purposes most important information is contained in Figs, 9 through I 1, which show the evolution of ice-crystal number and size as a function of temperature, sample volume,
SYSTEMS
499
and solution composition. Figure 9 shows the effect of size on nucleation and crystal growth in 50% (w/w) PG in water. Despite a lack of evidence for crystallization during cooling at l”C/min in a typical DSC sample, ice is clearly visible in Figure 9 at all sample volumes, including the 46-ml volume, which sustained a cooling rate in excess of l”C/min down to about - 90°C. The number and size of ice crystals formed increased dramatically as volume was increased to 482 and to 1412 ml. Ice crystals which nucleated at higher temperatures (clearly by a heterogeneous mechanism) were able to grow to rather large sizes only in the 482and 1412-ml volumes: evidently the cooling rate difference between the 46- and 483-ml samples was enough to prevent extensive crystal growth in the smaller samples. The rate of appearance of visible ice crystals increased as temperature fell to about -70 to -90°C. Below these temperatures, no additional visible ice crystals formed. At higher temperatures, crystal growth rate visibly decreased as temperature decreased, suggesting that new crystals might be forming below -70 to -90°C but might be unable to grow to visible size at these temperatures within the time of observation. Cryomicroscopic observations supported these macroscopic observations: directly measured ice crystal growth rates in 50% (w/w) PG in water were found to be 28.0 km/min at -7O”C, 4.84 km/min at -80°C and 0.48 km/min at -90°C. Given visibility of crystals on the order of 0.1 mm across, any new crystals should become visible in 4 min at - 70°C 21 min at - 80°C and 208 min at - 90°C. Another interesting point illustrated by Fig. 9 is the volume contraction of the solution during cooling. At the lowest temperatures shown, menisci have penetrated into the lower compartments of the two largest containers despite efforts to prevent such intrusion. The menisci occur in the centers of the samples, since this is where temperature is the highest and viscosity is the lowest, In the case of the 46-ml sample (Fig.
500
FAHY, SAUR, AND WILLIAMS CENTER TEMPERATURE
-20
-40
-60
-80
-100
-120
-140
Back @3-C) . . . . . to. . Center .._____..._ 1
-20
-40
CENTER TEMPERATURE -60 -80 -100
-120
-140
FIG. 6. Effects of volume and average temperature on temperature differences from the sample surface to the sample center (points connected by dotted lines in panels A and 3; ah data in panel C) or from wall to opposite wall (points connected by solid lines in panels A and B). Ba-C = temperature at the back of the container {probe 11) minus the remperature at the center of the container (prabe 3) ; F-Ba = front temperature {probe 8) minus back temperature {probe 1I), and so on. R = right (probe S), L = left (probe I), T = top (probe 7), and B = bottom (probe 6). Because the top temperature probe was not at the external wall near the vapor environment but was instead at the upper portion of the lower compartment, panel C shows both B - C (dotted lines) and T - C (solid lines) rather than 3 - C and T - B. Error bars for 1412-ml samples in lower portion of panel A were omitted for clarity. These standard errors (n = 2) were in the range of *lO”c. Horimntal standard errors (not shown) of ~2.5”C pert& to all points. Panel D: relationship between center temperatures of Fig. 6 and the average temperatures of Fig. 5.
SC), the gas cavities in the lower chamber just under the lower lid appear to have resulted from cavitation of the solution (bubble nucleation and growth), indicating the
large magnitude of the tensile stresses being developed. Figure 10 illustrates the effects of incorporating a standard renal carrier solution,
VITRIFKATION
OF LARGE
501
SYSTEMS
Topto Center -WI _--.w.:. .. P
_ T
-20 -40 -60
-100 -80 CENTER TEMPERATURE
-40
-60
-80 -100 CENTER TEMPERATURE FIG. 6-Continued
RPS-2=, in our 50% (w/w) PG solution. The presence of the carrier solution profoundly suppresses the number and size of visible ice crystals in 482- and 46-ml samples (1412-ml samples were not examined). This is hardly surprising given that the presence of RPS-2= raises total solute content from 50% (w/w) to something on the order of 54% (w/w). Figure 10B is included to demonstrate the lack of visible change associated with solidification of the sample (compare with Fig. lOA). Figure 11 shows the effect of partially compensating for the presence of RPS-2=
-120
-120
-140
-140
by reducing PG concentration to 47.6% (w/ w). The results appeared to be better than those obtained with 50% PG in water, as expected. Thus, the roughly 4% solute represented by RPS-2= is slightly more effective than 2.4% (w/w) cryoprotectant in its ability to facilitate vitrification. The implication is that carrier solution solutes significantly lower the concentration of cryoprotectant required for vitrification, as previously noted (4). DISCUSSION
This study was undertaken due to the
FAHY. SAUR, AND WILLIAMS
LLI 8
Back
-8 0.0
to Center
E -2.0 sLLI -3.0 LLI
-1.0
-4.0
a
-5.0
e E
-6.04
2 -4 -8
LLI r
l-
-12 -16 -20 .20
-40
-60
CENTER
-80
-100
-120
-740
TEMPERATURE
FIG. 7. Average thermal gradients through the samples. The temperature differences shown in Fig. 6 were divided by the probe separation distances (obtainable from Table 1) to obtain average thermal gradient in “C/cm. For B - C and T - C gradients, the vertical distance between the probes was used for calculating thermal gradient, not the actual probe separation distance.
very practical need to know how large volumes of concentrated cryoprotectant solutions behave with respect to fracture temperature, cooling rate, thermal gradient, ice nucleation, and crystal growth so that it can become possible to avoid fracturing and de-
vise means of coping with nucleation and growth. These data are important not only for understanding events during cooling but also for modeling devitrification, which ultimately requires knowledge of ice crystal growth rates as a function of temperature.
TABLE 2 Fracture Temperatures Volume (ml)
individual fracture temperatures (“C)
Mean
46 482 1412
-134.4, -136.7, -139.3, -142.3, -143.3, -I46.6,
