Thermal modeling of shock melts in Martian meteorites: Implications for preserving Martian atmospheric signatures and crystallization of high-pressure minerals from shock melts


  • Cliff S. J. Shaw,

    1. Department of Earth Sciences, University of New Brunswick, Fredericton, New Brunswick, Canada
    Search for more papers by this author
  • Erin Walton

    Corresponding author
    1. Department of Physical Sciences, MacEwan University, Edmonton, Alberta, Canada
    2. Department of Earth and Atmospheric Sciences, University of Alberta, Edmonton, Alberta, Canada
    • Department of Earth Sciences, University of New Brunswick, Fredericton, New Brunswick, Canada
    Search for more papers by this author

Corresponding author. E-mail:


The distribution of shock melts in four shergottites, having both vein and pocket geometry, has been defined and the conductive cooling time over the range 2500 °C to 900 °C calculated. Isolated 1 mm2 pockets cool in 1.17 s and cooling times increase with pocket area. An isolated vein 1 × 7 mm in Northwest Africa (NWA) 4797 cools to 900 °C in 4.5 s. Interference between thermal haloes of closely spaced shock melts decreases the thermal gradient, extending cooling times by a factor of 1.4 to 100. This is long enough to allow differential diffusion of Ar and Xe from the melt. Small pockets (1 mm2) lose 2.2% Ar and 5.2% Xe during cooling, resulting in a small change in the Ar/Xe ratio of the dissolved gas over that originally trapped. With longer cooling times there is significant fractionation of Xe from Ar and the Ar/Xe ratio increases rapidly. The largest pockets show less variation of Ar/Xe and likely preserve the original trapped gas composition. Considering all of the model calculations, even the smallest isolated pockets have cooling times greater than the duration of the pressure pulse, i.e., >0.01 s. The crystallization products of these shock melts will be unrelated to the peak shock pressure experienced by the meteorite.


Isolated regions of silicate glass containing a variety of microlites are found heterogeneously distributed throughout the groundmass of strongly shocked chondrite and achondrite meteorites (Dodd and Jarosewich 1979, 1982; Chen et al. 1996; Gillet et al. 2000; Malavergne et al. 2001; Xie et al. 2002, 2006; Beck et al. 2004). These features are called shock-melt veins or shock-melt pockets to indicate their origin via impact on the parent body. They are interpreted to have formed in local hot spots (up to 2500 K) by shock impedance contrasts or frictional melting along shear bands as shock waves traveled through heterogeneous, cracked, and/or porous materials (Langenhorst and Poirier 2000; Beck et al. 2004, 2007). The hot spots are distinct from the bulk rock in which the temperature increase, by shock compression and the nonadiabatic deposition of heat after decompression, was limited to a few hundred degrees (Sharp and DeCarli 2006). This study focuses on shock-melt veins and pockets (hereafter referred to simply as veins and pockets) in shergottites: mafic, permafic, or ultramafic igneous rocks from Mars having subophitic, porphyritic, or poikilitic textures (Walton et al. 2012). Shock melts are ubiquitous among shergottites, comprising up to 14 vol% of the host rock (Allan Hills [ALH] 77005; Treiman et al. 1994). Shock melts in shergottites are of particular interest because they host a nearly pure sample of the Martian atmosphere, defined by isotopic ratios and abundances of N2, CO2, and noble gases (Bogard and Johnson 1983; Marti et al. 1995; Walton et al. 2007).

In this study, we present a detailed analysis of the postshock thermal history of four shergottites using the 2D mode of the HEAT model developed by K. Wohletz (Wohletz et al. [1999] and The goals are twofold: (1) to resolve the discrepancy between shock-melt cooling times, i.e., the time required for cooling to the solidus of the melt, derived from previous calculations of heat flow and those from dynamic crystallization experiments, and (2) to assess the thermal history of natural meteorites with a range of shock-melt distributions and abundances. The models in this study provide refined estimates for the rate of meteorite cooling after a shock event.

Rationale for current study

Shock veins and shock-melt pockets comprise material that was locally melted (Fredriksson et al. 1963) and then cooled by conduction of heat to the surrounding host rock (Langenhorst and Poirier 2000; Leroux et al. 2000; Sharp et al. 2003; Xie et al. 2006). Calculations by Beck et al. (2007) indicate cooling rates for a 1 mm diameter shock melt of 5000 °C s−1 over the cooling interval 2500–500 °C, giving a cooling time for this interval of 0.2 s. This is considerably shorter than cooling rates of 0.2–0.3 °C s−1 determined by Walton et al. (2006) from dynamic crystallization experiments. These longer cooling times correspond to a cooling duration of 8−16 min to hours for the largest cm-size shock melts found in shergottites. Reaction textures between shock melts and host rock minerals are consistent with cooling times longer than those estimated by Beck et al. as indicated by experimental data (Walton and Shaw 2009). The cooling times of shergottite shock melts are of particular importance for sampling of Martian atmosphere. The longer cooling times of Walton et al. (2006) support diffusion of Martian atmosphere to the host rock, which has the potential to erase or modify that atmospheric signature, especially if the diffusion rates of the gases involved are significantly different.

Cooling times of shock melts also have implications for the pressure at which shock melts crystallize and cool below the solidus. This is important because the mineral assemblages that crystallize within shock melts, when compared with phase diagrams obtained from static high-pressure experiments, can be used to constrain the pressure conditions of crystallization (see discussions in Sharp and DeCarli 2006; Gillet et al. 2007). How the crystallization pressure relates to the shock history of the meteorite will depend on two factors: the shock duration, defined as the time lag between the arrival of the initial shock wave and the production of the release wave, and the quench time of the melt (Xie et al. 2006). Constraints on the shock duration in one shergottite, Zagami, have been obtained by studying trace-element concentrations in liquidus aggregates of K-hollandite in a shock-melt pocket (Beck et al. 2005). This method assumes that the trace-element partitioning took place during the shock pulse so the time required for a trace element to diffuse from the melt into the K-hollandite can be used to calculate a minimum value for the duration of shock pressure. Based on their measurements of Cs, Ba, and Rb the equilibrium shock pressure duration was found to be of the order of 10 ms (0.01 s). Similar time estimates for the shock duration are derived based on formation and preservation of high-pressure phases in Chassigny (Fritz and Greshake 2009). If the shock duration exceeds the quench time, crystallization occurs at the peak pressure and the mineral assemblage that crystallizes will be directly related to the peak shock pressure experienced by the meteorite. If the shock duration is shorter than the quench time, only part of the cooling path will be at high pressure with remainder occurring after the passage of the release wave. Finally, if the quench time is much longer than the shock duration the shock melt will remain molten after pressure release to crystallize a mineral assemblage whose formation conditions are unrelated to the shock-pressure conditions.

Petrography of shock melts in shergottites

Shock-melt veins are easily observed in polished sections as black to brown veins cutting across the entire meteorite sample. Their widths vary from 1–2 μm up to several millimeters and they may be interconnected or occur as single, straight features. Shock-melt pockets are rounded or amoeboid features, varying in size from <1 mm to several cm in apparent diameter. We have selected four nonbrecciated, igneous Martian meteorites for our models: Los Angeles (Warren et al. 2004), Dar al Gani (DaG) 476 (Zipfel et al. 2000), DaG 1037 (Russell et al. 2004) and Northwest Africa (NWA) 4797 (Walton et al. 2012). These meteorites were chosen as they cover a wide range of shock-melt distribution and geometry, even among paired samples (DaG 476 and DaG 1037). NWA 4797 is the least complex in terms of the distribution of shock melts, as it contains only a single shock-melt vein. Los Angeles lacks veins but has a number of pockets of different sizes, some of which are isolated from other shock melts while others are within a millimeter or two of adjacent pockets. DaG 476 and 1037 are the most complex in terms of the variation in size and distribution of shock melts. DaG 476 contains regularly distributed pockets with a very thin vein whereas DaG 1037 has one thick and one thin vein, and a range of pockets sizes from less than 1 mm to 4 by 8 mm. Detailed microtextures of the shock melts in each meteorite sample are shown in Fig. 1.

Figure 1.

BSE images of the Los Angeles (a, b), DaG 476 (c), DaG 1037 (d) and NWA 4797 (e, f). a) A small shock-melt pocket in Los Angeles (center) has crystallized a high-pressure mineral assemblage of stishovite + glass (Chennaoui Aoudjehane et al. 2005). b) A larger mm-size shock-melt pocket in the same thin section as shown in (a) contains a mineral assemblage (olivine + pyroxene + alkali glass + iron sulfide spheres; Walton and Spray 2003). c) The contact between a mm-size shock-melt pocket in DaG 476 and the basaltic host rock is shown. Elemental exchange has occurred between plagioclase in the host rock and neighboring pyroxene as evidenced by the diffuse contact between these two minerals. d) The contact between the mm-size shock-melt vein in the DaG 1037 and the basaltic host rock. Reaction textures and diffusive exchange between neighboring igneous minerals are observed within a zone approximately 500 μm from the shock vein/host rock contact. e) The mm-size shock-melt vein in the NWA 4797 has crystallized a mineral assemblage of olivine + pyroxene and iron sulfide spheres in alkali-rich glass. f) Plagioclase within the host rock is a highly vesiculated glass with flow textures. SMP = shock-melt pocket, SMV = shock-melt vein, pl-glass = plagioclase glass, ol = olivine, px = pyroxene.

Analytical methods

High-resolution images of the polished surface of each meteorite were acquired using a Zeiss EVO MA LaB6 filament scanning electron microscope (SEM) in backscattered electron (BSE) mode at the University of Alberta. BSE images were acquired using a Si diode detector under conditions of 20 kV (accelerating voltage) and 5–8 mm (working distance). ImageJ, an image analysis software program, was used to determine the modal abundance of shock melts in each sample, as well as to measure vein and pockets dimensions from BSE images.

To map the distribution of shock melts the individual thin sections (Los Angeles, DaG 476, and NWA 4797) or polished tile (DaG 1037) were photographed in transmitted light (thin sections) or reflected light (tile). This provided a low magnification overview of the entire sample showing the size, shape, and spatial distribution of shock melts in 2D. The photographs were then imported into a commercial image software program to produce a binary map of each section, dividing the samples into shock melt (black) and host rock (white) (Fig. 2). The characteristics of each thin section examined, including the width/length ratio of shock melts and their volume% abundance (determined by manual point counts on the thin section + tile photographs) are given in Table 1.

Table 1. General characteristics of shock melts in the four shergottite samples modeled
MeteoritesShock meltsWidth/length ratio% AbundanceHigh-PLow-P
PocketsVeinsRangeAverageShock meltaMinMin
  1. n.a. = not applicable, NWA 4797 contains a single shock melt vein.

  2. a

    %Abundance of shock melts measured from point counts on thin sections.

  3. b

    DaG 476 and DaG 1037 are possibly paired meteorite specimens (Russell et al. 2004).

Los Angeles0.5–1.0 0.85.4
DaG 476b0.4–
DaG 1037b0.1–1.10.713.8?
NWA 47970.14n.a.10.3
Figure 2.

Binary maps showing the size, shape, location, and distribution of shock melts (black) and host rock (white) in four shergottite meteorites. 1 mm2 grids were then assigned to each black area on the sketches to model the thermal history of these natural meteorites during shock.

The cooling history of each meteorite sample was modeled using the HEAT program of Wohletz ( and Wohletz et al.1999) in 2D mode. We have also tested a 3D cooling model. Since the thermal conductivity is isotropic, cooling times in a 3D model are the same as those in the 2D models. However, we recognize that in the natural samples, it is possible for pockets or veins not in the plane of the 2D model to affect cooling rates. The 2D finite element model applied to the meteorite has a minimum grid size of 0.01 m, which is too large to model the meteorite samples. However, the results of the model are easily scaled to allow us to examine shock-melt pockets with an area as small as 1 mm2. It is possible to model even smaller pockets, but the required increase in the number of cells in the model increases the calculation time by a factor of 1000. A limit of 1 mm was reasonable based on the observed pocket and vein size and on the constraints of the model. To accommodate the thinner shock melts in the meteorite samples (100 μm), we applied a linear extrapolation of our data.

For each sample, the distribution of shock melts was mapped onto a grid (Fig. 2). For the purposes of modeling we ignored pockets and veins with length or width significantly less than 1 mm. For example this omits the thin shock vein and several shock-melt pockets from DaG 476. For all four meteorites we assumed a basaltic bulk composition with a rock density of 3000 g cm−3, thermal conductivity of 1.8 W m−1 K−1 (Murase and McBirney 1973) and heat capacity of 1000 J kg−1 K−1 at 800 K (Bouhifd et al. 2007). The temperature of the host rock was taken to be 500 °C and assumed to be homogeneous (i.e., the same conditions as the Beck et al. [2007] calculation). The shock melts are assumed to be of approximately basaltic composition with density of 2725 g cm−3, thermal conductivity of 2 W m−1 K−1, and heat capacity of 1500 J kg−1 K−1 (Murase and McBirney 1973; Bouhifd et al. 2007). Thermal conductivity in melts varies from 1.3 W m−1 K−1 at 900 °C to 2.3 W m−1 K−1 at 1500 °C (Murase and McBirney (1973). Using the minimum experimentally determined conductivity (1.3 W m−1 K−1) rather than our chosen value of 2 W m−1 K−1, results in a 7.5% increase in cooling time, whereas using the maximum value gives a 1.7% decrease in cooling time. The melt temperature was taken as 2500 °C, i.e., just above the solidus of basalt at 20–25 GPa (Wang and Takahashi 1999; Hirose and Fei 2002), corresponding to the same parameters modeled by Beck et al. (2007). This may be an underestimate of the initial temperature as we have no reliable indicator of the actual peak temperature. In all models we assumed that there was no convection in the melt and no P–T dependency of the solidus. Since modeled cooling times are small (seconds to minutes) radioactive decay heat is insignificant and is ignored. The latent heat of crystallization was taken into account (see Wohletz et al. [1999] for details on the calculation), as there is some quench crystallization in the pockets and veins (Walton and Shaw 2009).

Beck et al. (2007) carried out simple calculations of cooling rate in which they examined cooling from 2500 °C to 500 °C. We have chosen 900 °C as the end point of cooling for our models since at this temperature diffusion of even the fastest moving component (e.g., Na, Zhang et al. 2010), will be too slow to have any significant effect on the shock melt composition, even if cooling is three orders of magnitude slower than calculated by Beck et al. (2007).


Idealized Isolated Shock-Melt Pockets and Shock Veins

In the ideal case, i.e., the conditions used by Beck et al. (2007) in their model, shock melts will cool by conduction to the background temperature of the meteorite with no thermal interference from the cooling of nearby melt pockets or veins. The times calculated for cooling, over the range 2500 to 900 °C, of such isolated pockets and veins exhibiting different areas and different width/length ratio (Table 2, Fig. 3) represent minimum cooling times. In this ideal case the thermal gradient between the shock melt and host rock is at its steepest, thus yielding the most rapid quench for a given area/volume of melt. The minimum cooling time for the smallest pockets modeled (1 mm2) is 1.2 s and there is a linear increase in cooling time with area (Fig. 3). Rectangular pockets show marginally faster cooling than equidimensional pockets because of their larger perimeter/area ratio.

Table 2. Cooling times for shock-melt pockets and shock veins of different areas and geometries
Area (mm2)Cooling times (seconds)
DAG 476LADAG 1037NWA 4797Isolated square SMPIsolated rectangular SMPSingle SMVa
  1. SMP = shock-melt pocket; SMV = shock-melt vein.

  2. LA = Los Angeles.

  3. a

    Single SMV calculated for a width of 1 mm appropriate for the NWA 4797 and DaG 1037 meteorites.

11.24 2.50    
21.94 2.19  1.81 
22.83 2.40    
23.08 2.60    
33.274.943.40   2.68
3 7.15     
3 5.73     
4  3.90 3.37 3.16
4  3.90    
4  4.10    
4  4.60    
4  4.10    
6  6.10  5.184.08
6  7.09    
6  44.57    
6  14.08    
6  6.30    
7   4.43 4.43 
8  12.10  6.484.67
9  10.71 7.95  
12  12.89  10.28 
24  32.18    
Figure 3.

Ideal cooling times for square and rectangular shock-melt pockets and a 1 mm wide shock-melt vein of various lengths compared with calculated cooling times for shock melts in four shergottites, Los Angeles, DaG 476, DaG 1037, and NWA 4797.

One mm wide shock veins in which the width to length ratio is less than 1/3 show significantly faster cooling than the square and rectangular pockets (Table 2, Fig. 3). For such veins we find that the cooling time becomes constant at a width/length ratio of 1/15.


Northwest Africa 4797

The 1 mm vein in NWA 4797 is the only melt-filled structure in the sample (Fig. 2). Cooling of this vein therefore shows the same cooling time as the ideal system for the observed length of 7 mm (Fig. 3). Even if this is only a fraction of the actual length of the vein, the calculated cooling time of 4.5 s is only 0.8 s less than the maximum possible time calculated for an ideal isolated vein (Table 2).

Los Angeles

The Los Angeles sample contains two 1 mm2 and three 3 mm2 shock-melt pockets (Fig. 2). The 1 mm2 pocket give cooling times <0.1 s longer than the ideal time for an isolated pocket (1.24 s versus 1.17 s) indicating that there is no significant effect on cooling time from the other shock-melt pockets in this part of the sample (Fig. 3, Table 2). All three larger pockets give cooling times of 5 to 7 s, i.e., 1.8 to 2.6 times longer, compared to the cooling time of an isolated pocket of the same size (Fig. 4, Table 2). The region between two closely spaced 3 mm2 pockets is heated well above the solidus and could have suffered postshock melting or diffusive modification (Fig. 5).

Figure 4.

Effect of adjacent melt on cooling of 1 and 3 mm2 pockets in DaG 476 and Los Angeles.

Figure 5.

Effect of adjacent shock-melt pockets on the igneous host rock. A small patch of initially solid material between two large, closely spaced pockets in the Los Angeles sample (red shaded area in Fig. 2) would be heated above the solidus. The region remained above the solidus temperature for approximately 5.5 s.

Dar al Gani 476

The model DaG 476 sample contains 11 discrete shock-melt pockets (Fig. 2). The actual sample also has a thin shock-melt vein (1–100 μm wide) and several pockets with areas significantly less than 1 mm2. As these are smaller than the minimum grid size in the HEAT model they are ignored here. The consequences of this are minimal; the additional heat contained in these pockets would increase cooling times slightly but their omission has no significant effect on the results presented. Two of the seven 1 mm2 pockets cool to 900 °C at nearly the same rate at the idealized isolated pockets (Fig. 3, Table 2). However, the remaining five pockets take approximately 20% longer (1.42 versus 1.17 s) to reach 900 °C than an ideal (isolated) pocket. These slower cooling regions are within 1–2 mm of another pocket so that the temperature gradient during cooling is decreased (Fig. 4). The three 2 mm2 pockets also show significant variation in cooling time. The most isolated of these pockets cools in 1.9 s, which is very close to the ideal cooling time. The remaining two are near to each other and to a 1 mm2 pocket. The interference between the thermal haloes of these closely spaced pockets extends cooling to 900 °C by 1 and 1.3 s (compared to an isolated pocket of the same size). The cooling time of the single 3 mm2 pocket is also extended due to interference of its thermal halo with those of nearby pockets.

Dar al Gani 1037

DaG 1037 shows the most complex distribution of shock-melt pockets and veins, and consequently the most complex cooling history (Fig. 2). This meteorite sample contains abundant pockets that vary in size and spatial distribution, in addition to a 1 mm wide vein cutting across the entire sample. The three 1 mm2 pockets cool to the basalt solidus in slightly longer than the time for an ideal pocket; 1.4 s compared to 1.2 s. All remaining pockets take significantly longer to cool than isolated pockets of the same size. The main retarding influence on cooling is the large shock vein that runs through the center of the sample; the hottest part of which takes approximately 1.75 min to reach 900 °C. For example, cooling times for the six 6 mm2 pockets take from 6 to 44 s depending on proximity to the shock-melt vein compared to an ideal time of 4–5 s for isolated pockets.

In both DaG samples (476/1037) the smallest shock-melt pockets experience some reheating (e.g., Fig. 6), although this does not extend back above the solidus (900 °C). The interference effects of adjacent pockets in DaG 1037 are illustrated in Fig. 7. Even after 1 s there is some interference between the thermal aureole of the vein and largest pocket. By 1.8 s, even small adjacent pockets show overlapping thermal aureoles and by 9.5 s only two areas are unaffected by heat from the adjacent pockets and veins. By 22 s there is a single thermal aureole within the model meteorite.

Figure 6.

Cooling history of the shock vein and two small shock-melt pockets in DaG 1037 (see Fig. 2 for locations). The veins show a regular cooling path. The veins cool to the solidus along the same path but show different degrees of reheating as the thermal halo from the cooling vein migrates outward.

Figure 7.

Cooling of DaG 1037 showing how the central shock vein affects the cooling of the adjacent melt pockets. Units on the X and Y scale are in cm.


Summary of Results

We have defined the distribution and cooling history of shock melt with both vein and pocket geometry in four shergottite samples: Los Angeles, DaG 476, DaG 1037, and NWA 4797 using the HEAT model of Wohletz et al. (1999). Although the temperature difference between the shock melt and the host rock is an important factor in determining the rate at which the shock melt cools, our study demonstrates that other parameters are also important in governing the cooling rate, specifically the size, shape (equidimensional, rectangular and vein geometry), and spatial distribution of the pockets and veins. All shock melts cool conductively to the background temperature of the host rock. However, the thermal gradient is largest in the case where shock melts are not influenced by the thermal haloes of other nearby veins or pockets. These isolated shock melts therefore undergo the most rapid quench. This behavior is seen in NWA 4797, which contains a 7 mm long, 1 mm wide vein. This vein cooled to 900 °C in 4.5 s. The other three samples have cooling times from 1.4 to 100 times longer than in the ideal (isolated) system. This deviation is the result of interference between the thermal haloes of nearby pockets, and in the case of DaG 1037, a large vein, which decreases the thermal gradient.

Previous Estimates of Cooling Times

Beck et al. (2007) modeled the cooling history of shock-melt pockets for the same melt and host rock temperatures used in this study. In their study, cooling times were derived using a single thermal diffusivity for the host rock and shock melt. Calculations from this 1-D model predict that a 1 mm shock-melt pocket will cool to 900 °C in 0.2 s. Additional estimates of cooling times for shergottite shock melts by Walton et al. (2006) were based on comparisons of crystal shape between natural samples and experimental charges produced by controlled cooling. The rates predicted from these experiments are significantly slower. A 1 mm shock-melt pocket took 8–12 min to cool to approximately 990 °C; we attribute this to the nature of the starting material. The crystallization experiments were performed using synthetic glasses whose composition was based on natural shock-melt pockets in shergottites. These glasses were prepared by two fusions of decarbonated oxide—carbonate mixtures at 1600 °C. This method produces homogeneous, crystal-free glass. The use of such homogeneous starting material extends cooling times because of the necessity of developing nuclei in the melt (e.g., Lofgren [1980] and references therein). Thus, more time is required to produce particular textures (skeletal versus equant, euhedral crystal shapes) than would be the case in nuclei-rich natural shock melt.

The thermal models developed in this study are considered to more closely approximate the time needed for cooling and partial crystallization of shock melts in natural meteorite samples. This is because we take into consideration the size, geometry, and spatial distribution of shock melts in heavily shocked natural meteorite samples. The calculations by Beck et al. (2007) were for a very simplified system which did not take into account differences in density, heat capacity, and thermal conductivity between the melt and host, and therefore led to cooling times for shock melts that were too short. We calculate a cooling time for a shock-melt pocket identical to that of Beck et al. (2007) that is between six and seven times longer. The experimental conditions used to constrain shock-melt cooling times by Walton et al. (2006) were also too simple, yielding unrealistically long cooling times. The most significant result from our models is the long cooling times we calculate due to thermal interference between adjacent pockets and veins. These longer cooling times call the conclusion of prohibition of chemical diffusion of atmospheric gas phases in pockets made by Beck et al. (2007) into question. If diffusion carried on for longer than calculated in their model, there may indeed be modification of the composition of the pocket during cooling. This is addressed in the following section.

Ar and Xe Diffusion

Shock recovery experiments have demonstrated that hypervelocity impact provides a viable mechanism for implanting a sample of ambient gases in melts produced during shock, without elemental or isotopic fractionation (Wiens and Pepin 1988; Bogard et al. 1989). This results from diffusion in a high-pressure environment. The high-pressure gas would diffuse into the locally molten regions of the meteorite (shock melts). If this meteorite then cools from its postshock temperature in a low-pressure environment, it would be expected to lose some of its radiogenic gas (Davis 1977). Here, we assess the potential for shock melts to lose some of their noble gases implanted during shock due to the extended cooling times predicted by this study. These calculations assume that all the Ar and Xe are initially contained within the shock melt. This assumption is validated by spatially resolved argon isotope measurements performed on neutron-irradiated samples of several shergottites including Los Angeles, Zagami, ALHA77005, and NWA 1950 (Walton et al. 2007, 2008). These studies show that Martian atmosphere, traced using 40Ar/36Ar as an isotopic fingerprint (as measured by Viking landers; Owen et al. 1977), is specifically sited within the shock melts with little or no atmospheric component in host rock minerals.

There are good data for Ar diffusion in a range of melts (Nowak et al. 2004). These authors show that at 1623 K, Ar diffusivity can be related to the degree of polymerization of the melt. Walton et al. (2006) give bulk compositions of three shock-melt pockets whose NBO/T (ratio of nonbridging to tetrahedral oxygens in the melt; a measure of polymerization calculated using Mysen et al. 1985) is 1.2–3.7. Using the relationship determined by Nowak et al. (2004) and an NBO/T of 1.8 we calculate DAr to be 2.15 × 10−10 m2 s−1 at 1623 K.

The diffusivity of xenon is less well defined and there are no data for Xe in basalt. However, Roselieb et al. (1995) and Spickenbom et al. (2010) present data on both Ar and Xe diffusion in jadeite melts. Extrapolation between the two data sets indicates that Xe diffusion is approximately 5 times faster than that of Ar. If the same relation holds for basalt then the diffusivity of Xe is 1.1 × 10−9 m2 s−1 at 1623 K.

The percentage of each gas remaining in shock-melt pockets of various sizes can be calculated using the spherical diffusion couple solution (Zhang 2008) and the cooling times calculated from the HEAT models. Calculations were made for isolated shock-melt pockets with a diameter of 1, 2, and 3 mm in a hypothetical meteorite. For these models the diffusivity was taken to be that at 1350 °C rather than at the peak temperature of 2500 °C. The pockets are at this peak temperature for only a fraction of the cooling period; using the diffusivity at this temperature would give unrealistically long diffusion profiles. The value at 1350 °C is a compromise, which recognizes that we can only average the diffusivity over the cooling time. The jadeite data of Roselieb et al. (1995) suggest that as temperature increases, the diffusivity difference between Ar and Xe increases by an order of magnitude between 1350 and 1600 °C and a further factor of 20 from 1600 to 2200 °C, assuming that the experimental data can be reliably extrapolated to such high temperatures. Our estimates of diffusivity are conservative and will give a minimum amount of fractionation. In all the models the concentration at the center of the spherical region is not significantly altered from the initial concentration (Fig. 8).

Figure 8.

Evolution of the Ar/Xe ratio in shock-melt pockets of radius 1, 2, and 3 mm over the modeled cooling times (see text for details). Note that the Ar/Xe ratio in a 1 mm wide vein is significantly higher after cooling than in a spherical shock-melt pocket of the same radius. The inset shows the calculated compositional profile for Xe in a 1 mm radius shock-melt pocket.

The results of the calculations show that for small pockets (1 mm wide) 2.2% Ar and 5.2% Xe would be lost from the pocket over the 1.5 s cooling time. This would result in a small increase in the Ar/Xe ratio, to approximately 1.02, of the glass over that originally trapped (Fig. 8). However, if the cooling time is increased to 5 s due to the effects of nearby veins or pockets, the longer diffusion time leads to significant fractionation of Xe from Ar and the Ar/Xe ratio increases rapidly to 1.08. The behavior of Ar and Xe in a 1 mm wide 7 mm long vein, which cooled over 4.5 s (NWA 4797), is different. In this case, diffusion is modeled using a 1 dimensional solution and the resulting loses in Ar and Xe are 7 and 15.6%, respectively, leading to an Ar/Xe ratio of 1.1.

A shock-melt pocket with a width of 2 mm and a cooling time of 4 to 6 s loses 1.7–2.2% Ar and 4.1–5% Ar, respectively. The Ar/Xe ratio increases slightly to 1.02–1.03 during cooling. For pockets with a protracted cooling history (e.g., the 2 by 3 mm pocket in DaG 1037), which takes approximately 45 s to cool to 900 °C (7–11 times the minimum rate), the amount of Ar and Xe lost increases to 6.6 and 13.9%, respectively, and the Ar/Xe ratio increases to 1.08 (extrapolated from the fit to the 2 mm pocket curve in Fig. 8). Pockets with widths of 3 mm giving cooling times of 10–12 s lose approximately 2% Ar and 4.4 to 4.9% Xe and show a similar increase in Ar/Xe ratio to the 2 mm wide pocket.

For the larger pocket (3 mm) there is less variation of the Ar/Xe ratio over the cooling time (Fig. 8b). This suggests that even though the cooling time is longer, larger, isolated pockets are more likely to preserve the original trapped gas composition than smaller pockets that have elevated cooling times due to thermal effects of nearby shock melts.

Implications for the Crystallization Pressure of Shock Melts

Minerals stable at high temperatures and pressures are found within or adjacent to shock melt veins and pockets in shergottites, e.g., omphacite, jadeite, stishovite, akimotoite, tuite, hollandite, amorphized (Mg,Fe)SiO3 pervoskite, magnesiowüstite, wadsleyite, ringwoodite, and Ca,Na-hexaluminosilicate (Langenhorst and Poirier 2000; Beck et al. 2004; Fritz and Greshake 2009; Imae and Ikeda 2010; Miyahara et al. 2011). These minerals form by crystallization from a silicate liquid during shock or by solid-state phase transformation. High-pressure mineral assemblages that crystallize during the shock pressure pulse can be used to constrain shock conditions and durations, by combining the crystallization assemblages within the shock vein with their phase equilibria determined from static high-pressure experiments (see discussions in Sharp and DeCarli 2006; Gillet et al. 2007). In contrast, high-pressure minerals formed by solid-state mechanisms are less reliable as indicators for shock conditions compared with shock-melt crystallization products because their formation requires over-stepping of the phase boundary (in pressure) for nucleation to occur (Sharp and DeCarli 2006).

The mineral assemblage that crystallizes within the shock melts, and which can be used to constrain shock P–T conditions experienced by the meteorite, is dependent on the shock duration and quench time (see discussion in the Rationale for Current Study section). Based on the cooling times modeled in this study using HEAT (see the Results section), and a 0.01 s shock pulse duration (Beck et al. 2005; Fritz and Greshake 2009) the following observations can be made of shock melts in shergottites:

  1. The mm-size pockets and veins, ubiquitous among the modeled shergottites, cool in timeframes greater than 1 s, even in those shock melts completely isolated from the thermal effects of nearby shock melts. The crystallization assemblage of these shock melts will be unrelated to the peak shock pressure experienced by the meteorite.
  2. Scaling our models to examine 100 μm wide shock-melt pockets observed in DaG 476 and DaG 1037 samples, but not included in our model calculations, decreases cooling times by a factor of 102 which is sufficient to allow cooling from 2500 to 900 °C over approximately 0.01 s. The pressure stability of the crystallization products within these smaller 100 μm size melts, when compared with their experimentally determined pressure stability, can be used to place constraints on the peak shock pressure experienced by the meteorite. The quench will be faster for 100 μm vein with a width/length ratio of less than 1/5, as this geometry cools more quickly compared with equidimensional pockets.
  3. When shock-melt regions are close together, even those that are submillimeter-sized will have cooling times that are longer than that required for crystallization of high-pressure phases during the shock pressure pulse.
  4. The search for a pristine sample of Martian atmosphere should focus on the center of large shock-melt pockets whereas only the smallest shock-melt regions will crystallize high-pressure phases related to the peak shock pressure experienced by the meteorite.

Application to Shock Melts in Shergottites

Our models make testable predictions regarding the distribution, size, and shape of shock melts in shergottites that will represent true high-pressure melts (i.e., crystallize in <0.01 seconds). Although a detailed study on the distribution of high-pressure minerals in the modeled shergottites is out of the context of this study, several earlier-published works, while not focussing specifically on the distribution of high-pressure phases versus shock-melt size, do support our predictions. Walton and Spray (2003) investigated shock-melt pockets in the same Los Angeles stone 1 thin section modeled in this study (Fig. 2). The mm-size shock-melt pockets contain schlieren-rich glass with Fe-sulfide spheres, and dendritic olivine, merrillite, titanomagnetite, and plagioclase. No high-pressure compositional equivalents such as wadsleyite, ringwoodite, lingunite, or tuite were encountered. The mm-size melt pockets are also vesicle-rich indicating low confining pressures, consistent with their crystallization during or after pressure release (Walton and Spray 2003). One smaller shock-melt pocket in the Los Angles thin section (approximately 300 μm thick) was devoid of vesicles and contained needle-shaped stishovite that crystallized from the melt (Fig. 1a; see also Walton and Spray 2003; Chennaoui Aoudjehane et al. 2005). Similarly, the mineralogy and composition of the minerals that crystallized within the 1 mm thick vein in the NWA 4797 thin section record a lower pressure mineral assemblage of olivine + pyroxene + alkali glass + Fe-sulfide spheres (Walton et al. 2012). Transmission electron microscopic investigation of thin (1–100 μm thick), black shock veins in the Zagami shergottite by Langenhorst and Poirier (2000) discovered a number of high-pressure minerals such as stishovite, lingunite, akimotoite, and amorphous silicate perovskite. Although Zagami was not specifically modeled in this study, the results of Langenhorst and Poirier (2000) do show that high-pressure phases occur exclusively in thin veins in some shergottites.

Cooling History of Shock Melts in Shergottites Versus Chondrites

Shock metamorphism is ubiquitous among all meteorites (e.g., Stöffler et al. 1991). However, the shock history of Martian meteorites is very different from those meteorites (e.g., chondrites) derived from small planetary bodies. Most chondrites are breccias, reflecting multiple-impact processing within the asteroid belt early in the history of the solar system (Bogard 1995). Laser probe 40Ar-39Ar dating of shock veins in the Peace River L6 chondrite (containing minerals stable at high pressure) yielded an age of 450 ± 30 Ma (McConville et al. 1988), and therefore tied vein formation to disruption of the L-chondrite parent body in a major collision event at approximately 400–500 Ma (Anders 1964). In contrast, most Martian meteorites are coherent igneous rocks that have been strongly shock metamorphosed but not brecciated (with the exception of monomict breccia ALH 84001, a ˜4.5 Ga ungrouped Martian orthopyroxenite; Treiman 1995). Shergottites have been ejected from the near surface of Mars in five discrete impact events over the past 0.7 to 20 Ma (Nyquist et al. 2001). Studies of shock duration have shown that the pressure pulse in chondrites (approximately 1 s) is much greater than in Martian meteorites (10 ms) (Beck et al. 2005; Fritz and Greshake 2009). This will strongly affect the mineral assemblages that crystallize from shock melts. If the duration of the shock pulse is longer (e.g., 1 s), as it is for chondrites, then the shock melt is more likely to crystallize during shock compression, forming minerals that record a crystallization pressure directly related to the shock P–T history of the host rock. Indeed this is the case for several chondrites including L6 chondrite Yamato 791384, in which shock veins up to 2 mm thick have crystallized a majorite-pyrope solid solution indicative of shock pressures 18–23 GPa (Ohtani et al. 2004).


Propagating shock waves causes melting at local hot spots. These melts quench crystallize to glass + crystals that preserve high- or low-pressure mineral assemblages depending on the cooling times of the melts and the shock duration. We have defined the distribution of shock-melt veins and pockets in Martian meteorites NWA 4797, Los Angeles, DaG 476, and DaG 1037, and we use a finite element model to calculate their cooling history. The thermal models developed in this study are considered to more closely approximate the time needed for cooling and partial crystallization of shock melts in natural meteorite samples. This is because we take into consideration the size, geometry, and spatial distribution of shock melts in heavily shocked natural meteorite samples. Our results demonstrate that the quench process is more complex than simple conductive cooling to a cooler host rock. Other parameters are also important in governing the rate of shock-melt quench, specifically the size, geometry, and spatial arrangement of these hot spots within the host rock. We draw the following conclusions from the results of our thermal models:

  1. Cooling times of shock melts may be extended 1.4 to 100 times by the interfering thermal haloes of nearby shock melts. This decreases the thermal gradient between shock melt and host rock which increases the overall cooling time.
  2. Shock-melt veins having the same area as shock-melt pockets cool more quickly due to their increased surface area with the cooler host rock.
  3. There may be diffusive loss of atmospheric gases during extended cooling times from high postshock temperatures; however, the center of larger, isolated shock-melt pockets are more likely to preserve the original trapped gas composition than smaller pockets that have elevated cooling times due to thermal effects of nearby veins and pockets.
  4. Only smaller pockets and veins (approximately 100 μm thick) that are isolated from the thermal effects of other hot spots within the host rock will quench before pressure release. In these, since crystallization occurs at the peak pressure and temperature, the pressure stability of the mineral assemblage within the shock melt is related to the pressure conditions of the impact event. This is especially relevant to those meteorites experiencing a short shock duration such as the shergottites.


This work has been funded by NSERC Discovery Grant RES0007057 awarded to E. W. and NSERC Discovery Grant 249939 awarded to C. S. Helpful comments provided by Zhidong Xie and an anonymous reviewer, as well as those of the associate editor, Ed Scott, improved the quality of the final manuscript.

Editorial Handling

Dr. Edward Scott