Hypervelocity impact cratering on water ice targets at temperatures ranging from 100 K to 253 K



[1] Surface temperatures on the icy satellites decrease substantially with distance from the Sun. Some of the main surface features on these bodies are hypervelocity impact craters. Understanding the influence of ice temperature on impact crater morphology is thus important. Previous laboratory impact crater data on water (H2O) ice targets was mostly at relatively high temperatures (255 K) and studies at low temperatures are limited. In addition, the few low-temperature studies have been at much lower velocities than those considered hypervelocity. They are thus not representative of impact craters on icy satellites. Accordingly data is presented here studying the dependence on target temperature for hypervelocity impact cratering on H2O ice at velocities of 5 km s−1 using spherical, 1 mm diameter aluminium projectiles. In this study, the ice target temperature ranged from 100 K to 253 K. Results for crater diameter, depths and volume are given along with ratios of these data. The results showed strong dependencies with temperature for all the above parameters. The strongest dependencies were from crater depth and volume. However, not all these properties vary in the same way with temperature. As the temperature of the ice increased, volume and depth increased, but crater diameter decreased. While the overall size scale of the laboratory studies (mm–cm dimensions) is still not appropriate to planetary scales (m–km), the general behavior as a function of target temperature has been established.

1. Introduction

[2] Spectroscopic data from the satellites of Jupiter and the other outer planetary bodies have indicated that ice dominates the surfaces of many of the satellites beyond 4 AU from the Sun [Grundy et al., 1999]. A number of studies listed in [Hartmann, 1993] have either measured or calculated the surface temperatures of bodies in the Solar System. Data obtained from the Voyager spacecraft measured the mean surface temperatures of Europa, Ganymede and Callisto at 103, 107 and 122 K, respectively. Mean surface temperatures of the icy moons of Saturn have been calculated to be around 70 K and the mean surface temperatures of Pluto and Charon were found to be around 37 K when furthest from the Sun, rising to approximately 70 K when inside the orbit of Neptune. These results clearly show a diverse range of surface temperatures and therefore rheological conditions on the icy bodies in our solar system.

[3] A major surface feature observed on the icy bodies within our solar system, are craters formed from hypervelocity impacts. Studies of impact cratering have been carried out in the laboratory on nonporous targets of the most abundant form of ice, H2O, at temperatures 250 to 265 K, to try and understand the cratering process on icy bodies [Kawakami et al., 1983; Cintala et al., 1985; Iijima et al., 1995; Kato et al., 1995]. However, these target temperatures do not represent typical surface temperatures on the icy satellites. Few papers with experimental data have been written on cratering at such low temperatures. In Croft et al. [1979], data are presented for 2 shots onto ice targets at approximately 210 K and several onto targets at 270 K. However, in Croft et al. [1979] it was not possible to see any difference in crater size with such limited data sets. Lange and Ahrens [1987] investigated crater morphologies on H2O targets with surface temperatures of 81 and 257 K at velocities of ≤0.64 km s−1. They found that crater dimensions (depth, diameter, and volume) increased slightly at the higher temperature. Kato et al. [1992, 1995] also carried out a study on H2O targets at temperatures of 77 K and 255 K (at velocities ≤1.22 km s−1). Unfortunately, they found that at the lower temperature the craters were too large for their targets so crater size could not be obtained. Over a more limited temperature range, data is available for high-speed impact cratering in porous ice targets [Koschny et al., 2001]. For speeds of 1 to 4 km s−1, with 25 mg nylon projectiles and porous ice targets (bulk density typically 520 kg m−3) at 193 and 253 K, it was found that targets at the lower temperature had, on average, craters with greater volume. This in opposition to Lange and Ahrens [1987].

[4] Another major problem with all these studies is that none of them were in the hypervelocity regime. Crudely speaking this is usually assumed to be when impacts are at greater than a few km s−1, that is, at faster than the speed of the resultant shock waves. This results in extreme temperatures and pressures being generated at the impact site. Traditionally there has been very little laboratory impact work on ices in this hypervelocity regime, Croft [1981] was one of the very few, with reports of two impacts into ice at 6.2 km s−1. Recently, more work on hypervelocity impact craters into ices was conducted by Burchell et al. [1998] who studied impacts on CO2 ice and compared them to two shots on H2O ice. Shrine et al. [2002] determined how impact craters in H2O ice scale with velocity over the range 1 to 7 km s−1. In addition, Burchell et al. [2001] formulated impact damage equations, which predict crater sizes in ice over the kinematic regimes used in laboratory experiments. Again however, all the impacts were on H2O ice at approximately 255 K. A recent paper by Grey et al. [2001a] compared craters in H2O ice and ammonia hydroxide ice at 155 K, but did not look at the evolution of crater size with temperature.

[5] A first look at how hypervelocity impact crater morphology systematically varies with target ice temperature was reported by Grey et al. [2001b] for the temperature range 152 to 253 K where crater diameter, depth and volume were analysed in 6 impact events. In that preliminary study, crater diameter was not found to have any temperature dependence but crater depth and volume did. However, these targets still had temperatures well above that found on the icy bodies in the solar system.

[6] Accordingly, in this present paper we further advance the study of hypervelocity impact cratering in low temperature ices in the laboratory. More data has been obtained, almost doubling the number of shots in the previous temperature range (152–253 K) and taking new data down to 100 K target ice temperature. A correspondingly more detailed analysis has been carried out and the results are compared to the known changes in ice properties of ice as temperature is lowered. Thus for the first time, the change in morphology of hypervelocity impact craters in ice has been determined as a function of ice temperature.

2. Experimental Procedure

[7] The ice targets used in this work were created from distilled water, which was boiled to reduce its dissolved gas content. The water was then cooled to 3 K above freezing, and siphoned into a container that consisted of a thermally conductive bottom and insulated top and sides. This was then placed in a standard freezer. Heat sources in the upper portions of the container caused the water to freeze from the bottom up. Ice was created in this method so to allow any dissolved gas remaining in the water to escape and to allow the ice once formed to expand naturally without restriction. Stresses within the ice were therefore kept to a minimum, see Grey [2000] for more details. Polycrystalline Ih H2O ice targets, 10 cm in-depth and 18 cm in diameter were created using this method. The targets were stored in a freezer at 253 K. The low target temperatures required in this work were obtained by placing the ice in a large volume CO2 freezer (capable of reaching 130 K), and cooling gradually to a selected temperature. For the lowest temperatures used in this work, the cooled targets from the CO2 freezer were cooled further in a liquid nitrogen freezer to approximately 100 K. The cooling typically took place over a period of days to prevent undue stressing of the targets.

[8] The hypervelocity impacts were obtained using a two-stage light gas gun [Burchell et al., 1999]. For this experiment, the projectiles were 1 mm diameter Aluminium (type 2017) spheres. During a shot the target chamber was evacuated to a pressure of (46 ± 19) Pa so that projectiles did not slow in flight. The speed of the projectiles was measured by their passage through two laser light curtains and was measured in each shot. Note that this method of calculating velocities is accurate to ±1%. The mean projectile speed for the programme was (5.07 ± 0.23) km s−1, thus giving a mean impact energy of (17.9 ± 1.3) J.

[9] During a shot the ice targets were placed in the gun's target chamber. This was cooled with a liquid nitrogen cold plate. A temperature sensor was fixed to the target so that the temperature was continually recorded. Tests showed that the variation in temperature between where the sensor was positioned on the target and the centre of the front face where the impact occurred, was less than 1 K.

3. Results and Discussion

3.1. Cratering Experiments

[10] The shot programme consisted of 14 shots covering a ice target temperature (T) range of 100 to 253 K. The details of the shots are given in Table 1. The impacts produced craters with a relatively deep central pit surrounded with a shallow terrace and no raised crater rim. This is typical of impacts in brittle materials on laboratory scales, for example, see Croft [1981] or Burchell et al. [1998, 2001] for hypervelocity impact craters on ices, Gault [1975] for impacts on rocks, and Burchell and Grey [2001] for impacts on glass. Images of craters at various temperatures are shown in Figure 1. There is a striking change in morphology from 218 K to 100 K. At 253 K the crater morphology in general (and in particular the terraces) is very coarse but as the temperature decreased the craters became smoother, the pit less significant and the terrace more prominent. The crater edge also became a lot smoother. Overall the craters became relatively shallower at lower temperatures.

Figure 1.

Images of H2O ice craters and respective profiles as a function of temperature caused by 1 mm Al. projectiles, fired at 5.07 km s−1. The scale bar in each crater image is 10 mm.

Table 1. Target Temperature, Impact Velocity, and Crater Dimensions for All Craters
Temperature, KVelocity, km s−1Crater Diameter, mmCentral Pit Diameter, mmCrater Depth, mmTerrace Depth, mmVolume, cm3
253 ± 15.3939.97 ± 2.2912.0 ± 0.58.94 ± 0.035.16 ± 1.077.67 ± 0.03
253 ± 15.2645.35 ± 6.2915.0 ± 0.59.51 ± 0.035.25 ± 1.147.21 ± 0.03
237 ± 15.0249.00 ± 1.2112.0 ± 0.58.95 ± 0.074.70 ± 0.564.75 ± 0.02
228 ± 15.0938.53 ± 0.3111.0 ± 0.58.94 ± 0.065.07 ± 0.445.67 ± 0.02
218 ± 15.0142.14 ± 1.1611.0 ± 0.58.41 ± 0.054.29 ± 0.315.25 ± 0.02
208 ± 15.0541.80 ± 0.587.0 ± 0.59.72 ± 0.095.14 ± 0.754.89 ± 0.02
190 ± 15.1745.55 ± 0.647.0 ± 0.58.24 ± 0.064.94 ± 0.364.18 ± 0.02
180 ± 14.9640.58 ± 0.588.0 ± 0.57.12 ± 0.103.82 ± 0.724.47 ± 0.02
174 ± 15.4744.55 ± 0.5410.0 ± 0.57.15 ± 0.084.32 ± 0.504.40 ± 0.02
164 ± 14.9248.95 ± 0.538.0 ± 0.57.13 ± 0.083.92 ± 0.472.77 ± 0.01
152 ± 15.0244.45 ± 1.3911.0 ± 0.56.18 ± 0.093.85 ± 0.582.55 ± 0.01
140 ± 15.2849.87 ± 2.886.0 ± 0.57.52 ± 0.144.42 ± 0.563.40 ± 0.01
110 ± 14.6854.58 ± 0.216.0 ± 0.55.88 ± 0.093.86 ± 0.472.70 ± 0.01
100 ± 14.7055.33 ± 1.025.0 ± 0.55.45 ± 0.304.29 ± 0.182.84 ± 0.01

[11] The results of measurements of crater diameter (CD), central pit diameter (PD), crater depth (Cd), mean terrace depth (Td) and crater volume (Vc) are given in Table 1. The diameters were taken from scaled images of the craters. The uncertainties on the diameters (mean error of ±1.16 mm in CD and ±0.5 mm for in PD) are not measurement errors, but represent the slightly irregular outlines of these features. These are typical values for such craters. Crater depth was obtained by measuring depth profiles across each crater. Examples are shown in Figure 1. The depth at the deepest point is the crater depth and is uncertain typically to ±0.06 mm (based upon repeated measurements on each crater). The mean terrace depth is also obtained from the depth profiles and has a mean error of ±0.58 mm. This larger value than for crater depth is due to irregularities in the terrace depth inside a crater. Crater volume (Vc) was obtained by packing micro diameter glass spheres into the crater and measuring their mass. This method was accurate to ±0.4% of the calculated volume. All measurements were made at room temperature in the laboratory immediately after the targets were removed from the gun. Because of the speed of the measurement taking, no problems were encountered with melting or frost.

[12] In addition to the direct measurements, a number of ratio analyses were conducted on the craters. The main ratio usually quote in impact studies is crater depth/diameter (Cd/CD = CC). In addition we also studied terrace depth/crater diameter (Td/CD = TC) and terrace depth/crater diameter (Td/Cd = Tc). Values for all these are given in Table 2. These are used to aid study of the evolution of crater morphology with ice temperature.

Table 2. Target Temperature and Ratio Data for All Craters
Temperature, KCrater Depth/Crater Diameter RatioTerrace Depth/ Crater Diameter RatioTerrace Depth/Crater Depth Ratio
253 ± 10.22 ± 0.010.13 ± 0.030.58 ± 0.12
253 ± 10.21 ± 0.030.12 ± 0.040.55 ± 0.12
237 ± 10.18 ± 0.010.10 ± 0.010.53 ± 0.07
228 ± 10.23 ± 0.010.13 ± 0.010.57 ± 0.05
218 ± 10.20 ± 0.010.10 ± 0.010.51 ± 0.04
208 ± 10.23 ± 0.010.12 ± 0.020.53 ± 0.08
190 ± 10.18 ± 0.010.11 ± 0.010.60 ± 0.05
180 ± 10.18 ± 0.010.09 ± 0.020.54 ± 0.11
174 ± 10.16 ± 0.010.10 ± 0.010.60 ± 0.08
164 ± 10.15 ± 0.010.08 ± 0.010.55 ± 0.07
152 ± 10.14 ± 0.010.09 ± 0.020.62 ± 0.10
140 ± 10.15 ± 0.010.09 ± 0.020.59 ± 0.08
110 ± 10.11 ± 0.010.07 ± 0.010.66 ± 0.09
100 ± 10.10 ± 0.010.08 ± 0.010.79 ± 0.07

3.2. Analysis

[13] The data for crater diameter and pit diameter are graphed versus target ice temperature in Figure 2. Note that on all figures where error bars are not visible they are smaller than the size of the symbol used for the datum. A power law fit to the data for crater diameter (Figure 2a) yielded:

equation image

where crater diameter is in mm, temperature in Kelvin and R is the regression coefficient of the fit. The greatest dependence on temperature was below around 160 K. Above this temperature, crater diameter was almost constant. This explains why earlier work [Grey et al. [2001a, 2001b] reported a near constant crater diameter with temperature (CD = (42.19 ± 1.73) mm) over the temperature range 152–253 K. This is compatible with the present data set in that temperature range, but is seen to be inappropriate as temperature is lowered further.

Figure 2.

Crater diameter and pit diameter as a function of temperature.

[14] By contrast, pit diameter (Figure 2b) increased as the target became warmer, and a fit yielded

equation image

Again diameter is in mm and temperature in K. Although there is scatter on the data, the trend is very clear, the pit is a less dominant feature of the crater as temperature is lowered.

[15] Crater depth is shown in Figure 3a. A power law fit yielded a power of temperature compatible with one. Therefore a linear fit was carried out (solid line). This gave

equation image

where Cd was in mm and temperature in K. The fit shown as a dashed line in Figure 3a is discussed later in the text. The terrace depth is shown in Figure 3b. Note that the two data points at 253 K overlap, so one has been displaced slightly along the temperature axes for display purposes. This has also been done in any subsequent figure similarly affected. A fit yielded

equation image

where terrace depth is in mm and temperature in Kelvin.

Figure 3.

Crater depth and terrace depth as a function of temperature.

[16] Crater volume is shown in Figure 4. A fit yielded

equation image

where crater volume is in cm3 and temperature in Kelvin. The craters were significantly smaller in volume at lower temperatures.

Figure 4.

Crater volume as a function of temperature.

[17] As stated, in order to better understand the crater morphology of the craters, a number of ratio analyses were conducted. The data for these is shown in Figure 5. A fit of crater depth/diameter (Figure 5a) yielded,

equation image

Above approximately 200 K the ratio appears roughly constant (given the scatter on the data) at (0.21 ± 0.02). This is compatible with previous reports on crater depth/diameter ratios in ice targets [e.g., Croft, 1981; Burchell et al., 2001]. However, the data shows a strong decrease in this ratio, (i.e., increased shallowness), as the ice temperature decreases. So much so that at 100 K this ratio has halved from its value above 200 K. From the data for the crater depth and diameter individually this is seen to be due to changes in both depth and diameter as temperature is lowered (the former decrease, the latter increases).

Figure 5.

Crater diameter, pit and terrace ratios.

[18] In Figure 5b the terrace depth/crater diameter is shown. A fit yielded,

equation image

Although the ratio is indicating a relatively shallower crater at lower temperatures, it should also be remembered that overall crater depth is being reduced even faster. This is illustrated in Figure 5c, where terrace depth/crater depth is shown. A fit is not shown to the data in Figure 5c. Above 130 K there is only a very weak dependence of this ratio on temperature, with an average value of (0.56 ± 0.03). However, at lower temperatures this ratio increases sharply, reaching (0.79 ± 0.07) at 100 K.

3.3. Discussion

[19] The pattern of behaviour of the evolution of crater size and morphology with decreasing ice temperature is complex. The different components of the crater behave differently. The central pit decreases in both observed diameter and total depth (equal to the crater depth) as the temperature of the ice target lowered. By contrast the total crater diameter is roughly constant over a wide temperature range, and then starts to increase slightly at the lowest temperatures. Meanwhile, due to the increasing shallowness of the craters, the crater volume decreases as temperature is lowered.

[20] When attempting to interpret these results in terms of the material properties of the target, difficulties quickly emerge. The obvious property to use is material strength. It is common to use tensile strength at low strain rate as the key single strength parameter to characterise targets in impacts. However, this relies on several assumptions: first that the ratio of compressive to tensile strength is constant and second that results of cratering are independent of stain rate (except perhaps to within an overall normalisation factor than can be absorbed elsewhere in any fit to data). However, neither of these assumptions is fully justified for ice. For example, Lange and Ahrens [1983] showed that the tensile strength of ice at 260 K was strongly strain rate dependent (with dynamic strength at strain rates of 104 s−1 significantly greater than static strength). Further, they reported that the ratio of compressive to tensile strengths at 260 K was also strain rate dependent.

[21] In addition the mechanism leading to crater widening by spallation of material from the surface is difficult to elucidate satisfactorily. The growth of cracks at very high strain rate is a field of study in itself. The traditional picture of pile up of dislocations at grain boundaries leading to crack development is not applicable due to the short timescales involved. Instead grain boundary sliding is considered important [Schulson, 1999]. As the ice strength continues to increase as the target reduces in temperature, we can suppose that relatively more of the impact energy is subjected onto the trans-granular cleavage. This latter effect contributes to widening of the craters (i.e., increase in diameter) while the former decreases the depth, causing the shallower craters. Note that in this model the central pit in the crater (formed around the impact point), should decrease in diameter as target strength increases, as its growth is not determined by crack propagation but by fragmentation of material and its movement (similar to excavation of crater depth). This would lead to a decrease of pit diameter as ice temperature is lowered. In the data the pit becomes increasingly less prominent in agreement with the model, until it has almost completely disappeared at a target temperature of 100 K.

[22] Arakawa and Maeno [1992] investigated the rheological properties of polycrystalline ice Ih over a wide range of temperatures (100 to 263 K) and low strain rates (4 × 10−3–4 × 10−6 s−1). They found that the brittle fracture strength was strongly temperature dependent. Strictly, they found that the logarithm of strength varied inversely with ice temperature. However, as noted by Lange and Ahrens [1987], the strength of ice only changes by a factor of two when the temperature is lowered from 255 K to 81 K. Thus, over the range of temperatures considered here, a linear relation is a good approximation between ice strength and the inverse of the temperature. While Arakawa and Maeno [1992] noted that the absolute magnitude of strength at any temperature was strain rate dependent, the rate of change of strength was roughly similar at all the strain rates they investigated. If the linear dependence of strength on the inverse of temperature is assumed applicable at the strain rates in hypervelocity impacts, then it is possible to investigate some aspects of the dependence of crater shape on ice strength. For example, it has long been assumed that in hypervelocity impact cratering, depth is a function of projectile penetration into the target. There is thus little difference between depths in ductile and brittle materials except that which comes from the initial material density and strength. Here we ignore the slight change in initial ice target density (≈2%) over the temperature range used and thus obtain that the change in crater depth is due to the increase in ice strength, which, following the argument above is linearly related to decreasing temperature. Further, it has often been observed [Shanbing et al., 1994] that an increase in crater depth depends on a change in target strength to the power of −1/3. Thus for ice a fit to temperature to the power 1/3 is equivalent to a fit to strength to the power −1/3. Fits to the data here suggested a linear rather than power law fit as most appropriate. However, a fit of crater depth (Cd) versus temperature (e.g., strength) to the power 1/3 yielded

equation image

(units of mm and K) and is shown as a dashed line in Figure 3a. Over the range of temperatures used the result is almost indistinguishable from the linear fit. Thus while we have the qualitative result that crater depth increases as material strength decreases in a fashion compatible with that expected, we cannot definitively demonstrate that the quantitative dependence is as predicted. Also, Schulson [1999] noted that tensile strength decreased with increased grain size. Within more coarsely grained ice, the cracks propagate immediately upon nucleation. In this experiment, grain size was kept constant, however, its effect on trans-granular cleavage and therefore crater size and morphology as the target temperature is reduced, is not fully understood.

[23] The peak shock pressure in the target at the moment of impact can be estimated. Here, we use the method of Mizutani et al. [1990] who relate the peak pressure (PO) in the isobaric core under the impact point to the volume of the core (given by VP, the volume of the projectile) and the late stage effective energy (LE) via

equation image

where LE is given by

equation image

Here m is projectile mass, v is impact speed and CO and s are the material parameters from the linear shock wave speed equation. For high shock pressures, Iijima et al. [1995] obtained that CO was 1.660 km s−1 and s was 2.05 for temperate ice. This gives a typical peak impact pressure in the impacts here of 50 GPa. Lange and Ahrens [1987] estimated that their (lower speed) impacts on ice produced peak pressures in the range 0.2 to 0.5 GPa at 255 K and 0.2 to 0.7 GPa at 81 K. Similarly Kato et al. [1992, 1995] estimated that their impacts on ice (again at <1 km s−1) involved peak pressures of 0.14 to 0.68 GPa. at 255 K and 0.2 to 1 GPa at 77 K. There is thus a difference in almost 2 orders of magnitude in the peak pressures in the hypervelocity impacts on ice reported here, and those in the lower speed data reported previously, although it should be noted that the earlier experiments were consistent with each other. This has implications when considering the high-pressure phase diagram for ice. The earlier work at 255 K would have shocked the ice to either phase III or V depending on the exact impact speed. At a temperature of 80 K, both experiments would have shocked the ice to phase IX. Here at both 100 and 255 K the appropriate high-pressure phase would be ice VIII. There may thus be a fundamental difference between low and high-speed impact experiments on ice, namely the high pressure phase of the ice involved.

[24] To test if there is any significant influence of the shock pressure on the resultant crater, we show Figure 6 where crater volume is shown versus ice temperature. The solid line is the fit to the data here and also shown are predictions from the results of earlier experiments at lower speeds, extrapolated to the impact energy here (18 J). At low temperatures it can be seen that the extrapolated result of Lange and Ahrens [1987] agrees well with the data here. Similarly at high temperatures the extrapolation of the data of Kato et al. [1995] for impacts of aluminium on ice also agrees well with the data presented here. The result from Lange and Ahrens [1987] at 257 K is much lower than the prediction here or from Kato et al. [1995]. However, we note that the data for crater volume versus impact energy obtained by Lange and Ahrens [1987] at 257 K consists of just 4 events and extrapolations should not therefore be taken with great confidence. Indeed, in the work of Lange and Ahrens [1987], the data sets for crater volume versus kinetic energy obey power laws, but the fitted exponents for the 81 and 257 K data are significantly different. Hence at the extrapolated energy here the predicted crater volumes at the two temperatures are almost equal, even though the data upon which they are based show a clear difference.

Figure 6.

Crater volume versus ice temperature.

4. Conclusion

[25] The first comprehensive study of evolution of crater size and shape for hypervelocity impacts in water ice targets of widely differing temperatures has produced some interesting results. A key feature of the work was the systematic variation of the target ice temperature and the observation of impact craters at regular intervals over the temperature range studied (100 to 253 K). While some properties such as crater depth and volume decreased by factors of 2 or 4 respectively as the ice target temperature is lowered, the crater diameter was relatively insensitive to the ice temperature until the target was <160 K. These results show that while hypervelocity impacts at temperatures of 250 to 265 K can give valuable insights into crater morphologies, a more accurate understanding achievable in the laboratory is obtained using target temperatures similar to that found on the surfaces of the icy satellites. The results here demonstrate that insights gained from previous relatively high temperature ice (255 K) targets may be misleading, in not only that absolute crater size has changed at lower temperatures, but also the ratio of various quantities (e.g., depth/diameter) changes. In addition, at lower temperatures the craters have a smoother, more regular appearance. It is also important to know not only the properties of the craters at low temperatures, but also the evolution with temperature, as not only are the ice temperatures on bodies such as Europa, Ganymede, Callisto, etc., low, but they also vary from body to body and this has to be allowed for in any comparative study of craters on these various bodies. Attempts have been made to understand how the crater size and shape depends on how the tensile and compressive strengths alters with temperature; however, much more research is required on ice strengths at these strain rates. What can however be said, is that although low and high speed experiments have involved significantly different peak pressures during impact, and hence different high pressure phases of ice, the data from the various experiments shows a good degree of consistency when extrapolated to equal energy.


[26] The light gas gun is operated under a Grant from the Particle Physics and Astronomy Research Council (UK). We thank M. Cole and N. Shrine for assisting us with the firing of the gun.