Noble gas adsorption with and without mechanical stress: Not Martian signatures but fractionated air


Corresponding author. E-mail:


Abstract– Sample preparation, involving physical and chemical methods, is an unavoidable step in geochemical analysis. From a noble gas perspective, the two important effects are loss of sample gas and/or incorporation of air, which are significant sources of analytical artifacts. This article reports on the effects of sample exposure to laboratory air without mechanical influence and during sample grinding. The experiments include pure adsorption on terrestrial analog materials (gibbsite and olivine) and grinding of Martian meteorites. A consistent observation is the presence of an elementally fractionated air component in the samples studied. This is a critical form of terrestrial contamination in meteorites as it often mimics the heavy noble gas signatures of known extra-terrestrial end-members that are the basis of important conclusions about the origin and evolution of a meteorite. Although the effects of such contamination can be minimized by avoiding elaborate sample preparation protocols, caution should be exercised in interpreting the elemental ratios (Ar/Xe, Kr/Xe), especially in the low-temperature step extractions. The experiments can also be transferred to the investigation of Martian meteorites with long terrestrial residence times, and to Mars, where the Mars Science Laboratory mission will be able to measure noble gas signatures in the current atmosphere and in rocks and soils collected on the surface in Gale crater.


As in almost any other type of analysis, laboratory preparation is an unavoidable step in noble gas measurements of meteorites. In the most favorable case, it is reduced to mining a small piece on the order of 50–150 mg in weight from the main mass of a meteorite, wrapping it in metal foil, or placing it into a laser port and measuring it without further preparation. However, some projects ask for different protocols, e.g., when homogeneous splits are needed for parallel measurements. Also, some samples require more careful sample selection as they are weathered finds from hot or cold deserts (see Meyer [2012] for Martian meteorites). As for the homogeneous splits, preparing a fine powder before sample division is the standard method. Air contamination is introduced this way, as first clearly shown in a laboratory experiment by Niemeyer and Leich (1976). Comparing results from four laboratory treated samples with an untreated chip of lunar rock 60015, these authors found that all ground samples contained an “up to ten-fold increase in the trapped rare gas contents.” The incorporation furthermore caused elemental fractionation different from what was observed in water solution and (regular) adsorption (Niemeyer and Leich 1976). The increase in noble gas concentrations was accompanied by higher amounts of active gases that reached the mass spectrometer despite normal operation of the clean up getters. Water or acetone treatment after grinding appeared not to introduce detectable additional contamination but caused partial loss of the indigenous gases (Niemeyer and Leich 1976). In the analysis of a bulk sample from the Allende meteorite that had been ground in a ball mill, presence of adsorbed air Xe was demonstrated by deficiencies relative to untreated Allende in the light isotopes 124Xe to 130Xe (Srinivasan et al. 1978). As in the case of lunar rock 60015, the contamination persisted to high temperature degassing steps (600 °C and above in this experiment). Even gentle crushing of an aubrite (Pesyanoe) was observed to introduce considerable amounts of xenon, whereas the lighter noble gases did not show a significant atmospheric component (Garrison et al. 1988). Definite proof for the occurrence of the effect was obtained by Niedermann and Eugster (1992) who crushed lunar rock samples in atmospheres spiked with 86Kr and 129Xe. Noble gases were incorporated into the samples upon grinding and in dependence on grain size (i.e., surface area) and partial pressure of the noble gases. Xenon thereby was incorporated about an order of magnitude more efficiently than Kr (Niedermann and Eugster 1992).

As a further complication, not only gain, but also loss of noble gases is expected during grinding. This fact is employed by the use of crushing as a method to release noble gases for measurement. It was first carried out in the 1960s (see Rama et al. [1965] for citation of Russian work) and came into wider use in the 1980s (e.g., Kelley et al. 1986; Wiens 1988). Crushing is especially valuable in studies concerned with separating noble gases situated in fluid/volatile inclusions from those in lattice positions (e.g., Rama et al. 1965; Funkhouser et al. 1971; Kelley et al. 1986; Turner and Songshan 1992; Trieloff et al. 2000; Kendrick et al. 2001, 2006; Kendrick and Phillips 2009; Kamensky and Skiba 2011). For terrestrial samples two fields of study are especially important: noble gases in MORB and the study of diffusion parameters (e.g., Graham 2002; Kurz et al. 2009; Czuppon et al. 2010). In the case of Martian meteorites, crushing revealed a very special component that was not seen in analyses where the gas was released by other methods. The so-called EETV-component released by crushing the Elephant Moraine (EET) A79001 meteorite (Wiens 1988) accounted for 5% of the total gas and was interpreted to be released from the larger vesicles in the glass.

Although the examples listed above demonstrate the importance of contamination by air/fractionated air in the case of lunar samples and a variety of meteorites, only one such study has been performed specifically addressing the case of Martian meteorites. Elementally fractionated air (EFA) was found by examining the low-temperature steps from Martian meteorite noble gas measurements. Mohapatra et al. (2009) defined EFA as an air component that has the original isotopic signature of terrestrial air, e.g., in its xenon isotopes, but fractionated elemental ratios. In this study it was demonstrated that the 84Kr/132Xe ratio can be as low as 1, while it is approximately 15 in air dissolved in water (Ozima and Podosek 2002). This makes understanding air contamination an even more serious problem, because, given analytical uncertainties, the Martian interior component (Ott 1988) often cannot be distinguished from EFA. Thus, understanding the effects of air contamination is crucial. In the most serious cases, incorporated air can not only mimic Martian signatures, but even fully overprint the original signatures and, unless recognized, lead to erroneous interpretations. Therefore, this study evaluates the influences of air uptake during laboratory sample preparation and terrestrial weathering on Martian meteorite noble gas signatures. We carried out adsorption experiments with terrestrial minerals; and we are making use of Martian meteorite samples, which were ground for the purpose of obtaining same sample splits. Preliminary data of these experiments were reported earlier in abstracts (Schwenzer 2005; Herrmann et al. 2006; Schwenzer and Ott 2006). Taken together, as reported here, the measurements illustrate how fundamentally important precise knowledge of air contamination is in Martian meteorite research.

Samples, Experiments, and Measurement Procedures

In this study we used terrestrial analog materials (gibbsite, olivine) and a set of ground bulk Martian meteorites to study adsorption with and without mechanical stress.

Adsorption Experiments

Two samples were used to evaluate the noble gases introduced into a sample by adsorption alone: San Carlos olivine and a well characterized standard gibbsite. We chose olivine, because olivine is a component of many of the Martian meteorites and has high temperature stability. It serves as a proxy for the silicates in Martian meteorites. To find an analog material for smectitic clay that is stable throughout our experiment turned out to be more difficult, since there is no fine-grained silicate with sheet structure entirely stable at high temperatures. Finally, we settled for gibbsite, which consists of Al(OH)3 layers that are also part of many clay mineral structures. While gibbsite dehydrates between 200 and 700 °C, depending on grain size and heating rates (Gan et al. 2009), it is stable at comparably high temperatures and does not melt upon decomposition, unlike smectitic clay minerals. Also, very small grain size and high surface area seem to be retained (see below) despite the transformation. Hence, although it probably undergoes change at approximately 600 °C, gibbsite appeared the most suitable choice. The sample of San Carlos olivine was mildly crushed to reduce the grain size, while the gibbsite remained untreated. As the goal of the study was to evaluate the effect of pure adsorption, we first “cleaned” the samples by performing a three-temperature step measurement, after which the samples were removed from the vacuum system and exposed to air. After the exposure, samples were measured again in a four-temperature step experiment.

In detail, the samples were wrapped in Pt-foil, loaded into the sample holder and pre-degassed for 24 h at 130 °C in vacuum. In the first (“cleaning”) run, samples were degassed at 250, 400, and 800 °C, after which they were removed from the vacuum system. The Pt-foils were opened carefully on one side and—stored with samples inside—in an open sample container. After a waiting time (27 days for olivine and gibbsite #1, 16 days for gibbsite #2) the Pt-foils were closed, the samples were again loaded into the sample holder, pre-degassed as before, and gases were extracted at 250, 400, 600, and 800 °C. Obviously, one cannot expect that the samples degas fully at this temperature, since this may require melting. However, the rationale of our experiment was to single out the adsorbed gas. This component (as well as weakly bound gases from sites inside the sample) can be confidently assumed to be released during the first “cleaning run.” Therefore, the second run should only release gas that was freshly inherited from adsorption during the intermediate air exposure.

Between and after runs, several mineralogical checks were carried out: optical inspection was done after each experiment, and XRF (for the gibbsite) and measurements of the BET surface area were carried out after the second experiment. The latter were kindly performed by W. Debschütz (Clausthal). As for the olivine, optical inspection indicated no change after the first run. Also, the sample weight (99.56 mg before and 99.41 mg after the first run) remained virtually the same. Since we used only a few milligrams of sample for the experiment, the BET surface area of olivine was below our detection limit. However, in general, the active surface area of coarse-grained (about 0.25–0.8 mm, which largely covers the grain size used for our experiment) San Carlos olivine has been determined to be about 300 cm2 g−1 (Wogelius and Walther 1992; Brantley and Mellott 2000). Gibbsite, in contrast to olivine, underwent severe changes. We did two experiments, the first with 98.84 mg and re-exposure time of 27 days, the second with 39.87 mg and re-exposure time of 16 days. The sample weight decreased by about 25% during the “cleaning” run, in both experiments. The sample—originally finely powdered and of bright white color—became lumpy with some yellowish spots. After the second experiment we characterized the gibbsite using XRF and measured its BET surface. XRF measurements confirmed that the gibbsite had undergone structural changes, but a new phase could not be identified. This can be explained by the large transition area between gibbsite and α-Al2O3 observed by Gan et al. (2009). Critically important for our experiment is knowledge of the surface area, which increased from 3.77 m2 g−1 in untreated gibbsite (control sample) to 65.33 m2 g−1 in the heated sample.

Martian Bulk Rock Samples

In the course of our systematic study of Martian meteorites (Schwenzer 2004), we measured noble gases (He, Ne, Ar, Kr, and Xe) in nine bulk samples: the basaltic shergottites Shergotty and Zagami; the lherzolitic shergottite Allan Hills (ALH) A77005; the olivine phyric shergottite EETA79001 (Lith. A); the nakhlites Nakhla, Lafayette, and Governador Valadares; the orthopyroxenite ALH84001; and the chassignite Chassigny. For classification and further information see, e.g., McSween (1994), Goodrich (2002), and Treiman (2005). The bulk measurements were part of a consortium study carried out on same-sample aliquots. To achieve same-sample splits, samples were homogenized by grinding them to analytical grade powder (<60 μm). In this consortium study our neon data complemented elemental (Jochum et al. 2001) and radionuclide data (Bastian 2003).

The atmospheric component introduced by grinding does not affect the helium and neon isotopic signatures (Schwenzer 2004). We used them to calculate the cosmic ray exposure ages obtained from cosmogenic 3He and 21Ne, which are consistent with results from Ar-measurements on non ground samples of the nakhlite meteorites (Korochantseva et al. 2011). The helium data, moreover, provide key insights into helium loss due to shock metamorphic overprint (Schwenzer et al. 2008). On the other hand, air contamination is visible in the heavy noble gas—Ar, Kr, and Xe—data. In the present article we discuss the implications of the heavy noble gas data for evaluating the influence of the atmospheric component from air exposure during laboratory sample preparation.

Mass Spectrometry

All samples were measured in the noble gas laboratory of the Max-Planck Institute in Mainz as described by Schelhaas et al. (1990) and especially for the samples of the consortium study by Schwenzer (2004). As for the bulk meteorites, about 50 mg of ground bulk sample was wrapped in nickel foil, loaded into the sample tree, and degassed under vacuum at 130 °C for 24 h to remove adsorbed terrestrial gas. The samples were measured in two temperature steps, at 800 and 1800 °C. The gases were cleaned by Ti- and Al-Zr getters, separated into four fractions (He+Ne, Ar, Kr, Xe) and measured in the order Ne, He, Ar, Kr, Xe. Noble gas blanks were somewhat variable, typical values (in cm3 STP) being 36Ar = 4 × 10−12, 40Ar = 1.3 × 10−9, 84Kr = 1.8 × 10−13, and 132Xe = 7.1 × 10−14. The samples for the adsorption experiment were wrapped in platinum foil. Pre-measurement degassing and noble gas measurement procedures were similar as for the bulk meteorite samples. A typical 1600 °C blank at the time of the adsorption experiment measurements gave 36Ar = 5×10−12, 84Kr = 2×10−13, and 132Xe = 1×10−13 (in cm3 STP units). Data were corrected for mass spectrometric interferences using standard correction protocols (Schelhaas et al. 1990; Mohapatra et al. 2009).

Results and Discussion

The results are summarized in Table 1 (ground Martian meteorites; data from Schwenzer 2004) and Table 2 (adsorption experiment). Figures 1–3 display the results on ground Martian meteorites, Figs. 4 and 5 show the adsorption experiments.

Table 1.   Martian meteorites, ground bulk samples (totals). 36Artr, 84Kr, and 132Xe in 10−12 cm3 STP/g. Artr was calculated with trapped 38Ar/36Ar of 0.244 and cosmogenic 38Ar/36Ar of 1.5 (Wiens 1988).
Meteorite 36Artr 84Kr 132Xe 129Xe/132Xe 36Ar/132Xe 84Kr/132Xe
  1. 1Highly uncertain because 36Ar and 38Ar are almost entirely cosmogenic. Only approximate values are listed.

ALHA770051811.6 ±72.9107.8 ±7.329.03 ±0.951.122 ±0.01562.4 ±3.23.71 ±0.28
ALH840011975.8 ±86.6202.3 ±10.544.77 ±1.241.377 ±0.00744.1 ±2.34.52 ±0.27
(Lith. A)
547.1 ±30.743.64 ±2.3213.61 ±0.631.062 ±0.01840.2 ±2.93.21 ±0.23
Chassigny1208.9 ±85.256.01 ±2.8913.44 ±0.881.130 ±0.03390.0 ±8.74.17 ±0.35
(∼2089)186.90 ±4.4722.34 ±0.751.131 ±0.022∼943.89 ±0.24
Lafayette(∼460)162.34 ±3.2928.91 ±2.321.130 ±0.015∼162.16 ±0.21
Nakhla(∼900)149.82 ±2.3814.68 ±2.071.249 ±0.030∼603.39 ±0.51
Shergotty1267.1 ±121.957.46 ±3.1715.38 ±0.731.165 ±0.02082.4 ±8.83.74 ±0.27
Zagami932.2 ±72.860.78 ±3.2521.43 ±0.761.074 ±0.01438.7 ±3.32.52 ±0.16
Table 2.   Adsorption experiment, 36Ar, 84Kr, and 132Xe in 10−12 cm3 STP/g.
Olivine 1 Original sample
T 36 Ar err 84 Kr err 132 Xe err 36 Ar/ 132 Xe err 84 Kr/ 132 Xe err
  1. – indicates temperature steps with gas releases below blank limits. We assume no significant amount of sample gas to be released in those steps. Because different noble gases show different release patterns, we sum over the entire experiment.

Olivine 2 Sample after first measurement and 4 weeks of air exposure
T 36 Ar err 84 Kr err 132 Xe err 36 Ar/ 132 Xe err 84 Kr/ 132 Xe err
Gibbsite 11 Original sample, first run
T 36 Ar err 84 Kr err 132 Xe err 36 Ar/ 132 Xe err 84 Kr/ 132 Xe err
Gibbsite 12 Sample after first measurement, first run, and 4 weeks of air exposure
T 36 Ar err 84 Kr err 132 Xe err 36 Ar/ 132 Xe err 84 Kr/ 132 Xe err
Gibbsite 21 Original sample, second run
T 36 Ar err 84 Kr err 132 Xe err 36 Ar/ 132 Xe err 84 Kr/ 132 Xe err
Gibbsite 22 Sample after first measurement, second run, and 4 weeks of air exposure
T 36 Ar err 84 Kr err 132 Xe err 36 Ar/ 132 Xe err 84 Kr/ 132 Xe err
Figure 1.

 Comparison of ground samples with literature data of the respective Martian meteorite. Percentages are calculated from sums of all measured T-steps as excess of the ground sample over the literature values. Closed symbols are 84Kr, open symbols are 132Xe. Literature data are from: Bogard et al. (1984), Nagao (1987), Ott (1988), Ott et al. (1988), Swindle et al. (1989, 1995), Miura et al. (1995), Murty and Mohapatra (1997), Bogard and Garrison (1998), Busemann and Eugster (2002), Mathew and Marti (2001, 2002).

Figure 2.

84Kr/132Xe versus 36Ar/132Xe of literature data from, to our knowledge, untreated samples (as in Fig. 1 and Table 3; gray symbols) and ground samples of this study (black symbols).

Figure 3.

136Xe/132Xe versus 129Xe/132Xe of ground (this study) and untreated (literature) Martian meteorites. Literature data from Becker and Pepin (1984), Swindle et al. (1986, 1995), Bogard et al. (1989), Miura et al. (1995), Murty and Mohapatra (1997), Mathew et al. (1998), Gilmour et al. (2001), Mathew and Marti (2001, 2002), and Ocker and Gilmour (2004).

Figure 4.

 Adsorption experiments (for details see text) were carried out on gibbsite and olivine. Shown are the elemental ratio of 84Kr/132Xe vs 132Xe amounts for the initial measurement (before heating) and the second measurement after the heated, i.e., measured sample was exposed to air. Amount of 132Xe in cm3 STP/g.

Figure 5.

36Ar/132Xe versus 84Kr/132Xe of three laboratory experiments. For reference, terrestrial air (off the diagram diagonally to the upper right, Swindle 2002), elemental ratios as found in terrestrial water (open star, Ozima and Podosek 2002) and Martian interior (open cross, Ott 1988) as well as elementally fractionated air (EFA, small white dot next to the open cross) are plotted. Note that EFA falls onto the symbol for Martian interior. Circles show the various adsorption experiments.

Adsorption Experiment

For both, the olivine and the gibbsite, a larger amount of 84Kr and 132Xe were found in the re-exposed sample compared to the original one (Table 2, Fig. 4). For olivine, the measured values can be compared to each other in a straightforward way, because no mineralogical or weight changes were observed after heating during the first measurement. Olivine therefore adsorbed about twice as much 84Kr as the original sample contained. The increase in 132Xe is a factor of about 5. For gibbsite, which lost about 25% of its weight during the first heating, the case is more complicated because of the structural change and the large increase in active surface area: 3.77 m2 g−1 in the unheated versus 65.33 m2 g−1 in the heated sample. The original increase in 84Kr (146 × 10−12 cm3 STP/g before and 6424 × 10−12 cm3 STP/g after; Table 2) and in 132Xe (16 × 10−12 cm3 STP/g before and 1737 × 10−12 cm3 STP/g after; Table 2) has to be scaled to this increase in surface area. Concentrations per surface area are 38.5 × 10−12 cm3 STP/m2 (before) and 98.3 × 10−12 cm3 STP/m2 (after) for 84Kr and 4.3 × 10−12 cm3 STP/m2 (before) and 26.6 × 10−12 cm3 STP/m2 (after) for our first gibbsite experiment. The values for the second gibbsite experiment are similar. The resulting factors of increase are 2.5 (1.8 for the second experiment) and 6 for 84Kr and 132Xe, respectively. This result matches the increase observed for olivine rather closely. We also observed—especially in gibbsite—loss of significant amounts of other reactive volatiles (removed by exposure to the Ti- and Al-Zr getters) during the first measurement, which was detected by the pressure increase in the system during heating of the sample. In the second set of measurements, i.e., after pre-degassing by the first measurement, the pressure increase during the heating phase was considerably less, suggesting that the pressure in the first run stemmed from water and other volatiles incorporated into the sample during mineral formation. In addition to the structurally bound water in gibbsite, many of the potential adsorption sites (e.g., grain boundaries) must have been occupied by these volatiles; and since they were removed during the first set of measurements, a fresh, clean surface was exposed to air during the adsorption phase. In laboratory air, especially when compared to water in hydrous mineral formation, the competition is less severe, which may lead to an increased adsorption of noble gases on fresh surfaces. Bernatowicz et al. (1984) suggest that this effect may account for the discrepancy between laboratory measured adsorption constants and observed natural noble gas concentrations. The effect may also explain the gain in noble gases upon grinding, because the breakup of grains exposes fresh and dry surfaces to air, just as it happens after the cleaning steps. In fact, this heating and cleaning may have altered the surface in a similar manner as described by Bernatowicz et al. (1982) who observed Henry constants that were higher by an order of magnitude after heating a feldspar sample as compared to crushing it. The authors discuss two possible explanations for the observation: either crushing may have opened up fractures and therefore the new surfaces were not truly pristine; or crushing released volatiles contained in the crystals, which were then competing for adsorption sites. They favor the second option, because a (volatile free) lunar sample did not show any difference between being crushed and being heated. Martian meteorites may have less volatiles stored in them than Bernatowicz’s labradorite, which would minimize “self-contamination” of surfaces. Heating under vacuum, however, will remove any volatiles that are released, e.g., due to breaking open fluid inclusions. We therefore expect both, the crushed Martian meteorites and the heated samples, to present fresh, clean surfaces to the air they are exposed to.

The noble gas elemental composition of the gibbsite sample changes dramatically. The original samples show 36Ar/132Xe and 84Kr/132Xe ratios that scatter considerably and fall into the vicinity of atmospheric noble gases dissolved in water (Fig. 5). The adsorbed component, however, is similar in all samples ranging from approximately 40–48 for 36Ar/132Xe and from 3 to 5 for 84Kr/132Xe. This is much lower than in the original sample and—for Kr and Xe—resembles the adsorption values found upon grinding. Note that, like described in the case of grinding (Niemeyer and Leich 1976; Niedermann and Eugster 1992), the adsorbed component is not predominantly released in the first temperature steps, but a large fraction of it degasses at higher temperature (Table 2). One explanation is that heating created fresh surfaces that are as active (i.e., have an order of magnitude higher Henry constants and higher surface energies; Bernatowicz et al. 1982) as freshly crushed surfaces. Volatiles of any sort will take advantage of that and thus be strong competitors for noble gas adsorption. Fluids, i.e., strong competitors, occur in mineral formation environments to a much higher degree than in dry laboratory air, but may also be released upon heating or grinding of crystals that host fluids in inclusions or on grain boundaries (Bernatowicz et al. 1982).

Ground Martian Meteorites

We will now compare our results on pure adsorption to adsorption under the mechanical stress of grinding. Unfortunately, seen in their entirety, the elemental abundances in the ground samples do not display a clear picture of the effect that grinding may have on Martian meteorites (Fig. 1). Comparing the ground samples with untreated samples (literature values, Table 3) by calculating the excess amounts of 84Kr and 132Xe in the ground samples, spans a range from apparent loss to an excess as large as 650%. A definite “gain,” i.e., a higher Xe abundance in the ground sample than in the reference literature samples, can be observed for Shergotty, Nakhla, Governador Valadares, ALH84001, and Zagami. EETA79001 (Lith. A) spans a wide range from apparent deficits to apparent excesses, whereas Lafayette, ALHA77005 and Chassigny apparently experienced “losses.” The literature database, however, is small, and, moreover, there is considerable scatter among the untreated (to our knowledge) reference samples. Especially in the case of finds—and depending on sampling location—alteration of the meteorite’s surface and the presence of internal cracks may cause air being present in an untreated sample. Although this effect may offer an explanation for the “loss” (i.e., lower Xe in the ground specimen) in the finds Lafayette and ALHA77005, it does not apply to Chassigny, which is a fall. A potential reason for “loss” rather than “gain” in Chassigny may be the fact that it consists almost exclusively of olivine (see below).

Table 3. 36Ar, 84Kr, and 132Xe in Martian meteorites from the literature. 36Ar in 10−9 cm3 STP/g, 84Kr, and 132Xe in 10−12 cm3 STP/g; “bulk chips” refer to cases in which the authors only state that they took small pieces of the sample, but did not specify any further treatment. Our samples of Lafayette (Ott et al. 1988) were obtained from the Field Museum in Chicago and were slightly crushed to reduce grain size but they were not ground.
Meteorite 36Artr 84Kr 132Xe 36Ar/132Xe 84Kr/132XePreparation as stated by the authorSource
  1. 1Pers. comm. T. Swindle, August 2010.

ALHA770058.9655830329.571.84Bulk chips Miura et al. (1995)
10.055830333.001.84Not reported Nagao (1987)
ALH840011.2917119.964.828.59Not reported, likely untreated pieces Bogard and Garrison (1998)
1.0011415.962.897.17Not reported, likely untreated pieces Bogard and Garrison (1998)
1.3818424.656.107.48Not reported, likely untreated pieces Bogard and Garrison (1998)
1.2515724.151.956.51Bulk chips Miura et al. (1995)
1.6317527.659.096.34Bulk chips Miura et al. (1995)
1.4218528.150.506.59Bulk chips Murty and Mohapatra (1997)
8.7759.011.5762.615.13Bulk chips Swindle et al. (1995)
1.7315628.061.915.60Bulk chips Mathew and Marti (2001)
Chassigny0.8153.443.218.701.24Bulk chips Ott (1988)
2.2483.747.347.411.77Bulk chips Mathew and Marti (2001)
1.8181.844.740.391.83Bulk chips Mathew and Marti (2001)
0.8961.746.119.281.34Bulk chips Mathew and Marti (2001)
EETA790010.4011821.019.055.62Not reported, likely untreated pieces Bogard et al. (1984)
1.3029.04.7276.606.17Not reported, likely untreated pieces Bogard et al. (1984)
1.8920.531.160.870.66Not reported, likely untreated pieces Bogard et al. (1984)
Governador Valadares12.021.06.21935.483.39Bulk chips1 Swindle et al. (1989)
Lafayette 98.432.4 3.04Small chips Ott et al. (1988)
 56.520.4 2.77Small chips Ott et al. (1988)
Nakhla0.9039.611.577.833.44Bulk chips Ott (1988)
0.8834.98.2107.734.28Bulk chips Ott (1988)
0.8640.913.364.663.08Bulk chips Ott (1988)
13.724.09.01522.222.67Not reported Busemann and Eugster (2002)
 1.0555.712.783.254.40Most likely untreated pieces of bulk rock Mathew and Marti (2002)
Shergotty0.7023.75.3132.084.47Not reported, likely untreated pieces Bogard and Garrison (1998)
1.2466.612.797.875.24Bulk chips Ott (1988)
1.2853.78.8144.806.07Bulk chips Ott (1988)
Zagami5.5043.03.31666.6713.03Not reported, likely untreated pieces Bogard et al. (1984)
7.7017910.0770.0017.99Small pieces from a larger, untreated sample Ott et al. (1988)

Grinding of our samples was done under laboratory atmosphere. Upon grinding, the active surface area of a sample is increased. Furthermore, grinding increases the temperature of the sample by friction heating, and applies mechanical stress on the individual grains. The combination of these factors causes terrestrial air to be the most likely contaminant introduced by the sample preparation. To estimate the relative and absolute amounts of contaminating air to be expected we will first look at the adsorption process in detail. For this, it seems safe to assume that the application of Henry’s law is possible, because krypton and xenon are rare (Ozima and Podosek 2002). Air contains 0.114 × 10−3 vol. % Kr and 8.7 × 10−6% vol. Xe, which at P = 1.013 atm corresponds to p(Kr) of 1.155 × 10−6 atm, and p(Xe) of 8.81 × 10−8 atm. The influence of temperature on adsorption coefficients in geologic materials within the range of 0–25 °C is less severe than the influence of material properties, but in most geologic material Xe adsorbs to a greater extent than does Kr (Ozima and Podosek 2002). Deciding which adsorption coefficients for Kr and Xe are applicable to the grinding experiment is not straightforward because of the dependence on the material.

Comparing available data for shale, basalt and quartzite (see Ozima and Podosek 2002, their Table 3.2; data from Podosek et al. 1981) reveals that fine-grained clay adsorbs more and fractionates Kr from Xe more severely than do basalt and quartzite (Table 4). Especially for the fine-grained clay, the effect is a shift in the 84Kr/132Xe ratio from terrestrial atmosphere toward lower values. From this, the expected 84Kr/132Xe ratio for analytical grade powder is about 2.1, which is in the range of what is observed in our samples, independent of the original 84Kr/132Xe ratio as reflected by the literature values (Fig. 2) or the gain or loss factor (Fig. 1). In addition, it has been shown that the increase of the surface area is only one aspect of the enhanced adsorption by crushing-generated surfaces: crushing may furthermore cause an order of magnitude increase in the Henry constant (for the case of plagioclase see Bernatowicz et al. 1982) and may produce more energetic surfaces compared to the uncrushed sample (Bernatowicz et al. 1982). Thus, most likely adsorption contamination by terrestrial air has led to the elemental ratios observed in our ground samples and has overprinted the Martian signatures.

Table 4.   Henry constants, amounts of Kr and Xe and 84Kr/132Xe ratios of adsorbed air. All numbers are taken from Ozima and Podosek (2002), their table 2.2, p. 39). T in °C, Henry constant in cm3STP g−1 atm−1, and N in cm3STP g−1. VA shale = volcanic ash shale, CR basalt = Columbia River basalt, and U qtz = Uncomphagre quartzite. f is the fractionation factor that describes the enrichment of Xe relative to Kr.
rockTH (Kr)N(Kr)H (Xe)N(Xe)Kr/Xe 84Kr/132Xef
U qtz03 * 10−43.46 * 10−105 * 10−54.41 * 10−1278.61670.167
CR basalt00.283.23 * 10−70.76.17 * 10−85.2411.12.50
VA shale00.434.97 * 10−75.64. 94 * 10−71.012.1313.0
VA shale250.212.43 * 10−72.52.20 * 10−71.102.3311.9

The air contamination becomes evident in the isotopic signatures of the ground samples compared to their untreated counterparts. On a plot of 136Xe/132Xe versus 129Xe/132Xe all ground samples, except Shergotty, plot in a small area close to air, whereas the untreated samples cover a much wider range, including the display of some fissiogenic 136Xe (Fig. 3). Our conclusion is in agreement with observations on lunar samples (Niemeyer and Leich 1976; Niedermann and Eugster 1992) and the Allende chondrite (Srinivasan et al. 1978). However, compared to these cases, it is considerably more and critically important to the study of Martian meteorites because of the unique similarity of the contaminant with the Martian interior component. A particular case is the 129Xe/132Xe isotopic ratio, which is often used as an indicator for the presence of Martian gas components. With gas amounts small, analytical precision of Xe analyses of Martian meteorites is often poor and data do not allow distinguishing between Martian interior (129Xe/132Xe = 1.03 ± 0.02; Ott 1988) and air (0.9832 ± 0.0012; Ozima and Podosek 2002). Fissiogenic 136Xe has been observed previously in other Martian meteorites (Mathew et al. 1998, 2003; Mathew and Marti 2001, 2002). If the contamination problem is not recognized in such a case, this then can lead to misinterpretation of the results.

Looking at one of the common elemental ratio plots in Martian meteorite noble gas research (84Kr/132Xe versus 36Ar/132Xe; Fig. 2) demonstrates the case: The Martian meteorite measurements of untreated samples (literature values) span a wide range and trend toward Martian atmosphere. The ground samples, on the other hand, cluster in a much smaller area (36Ar/132Xe between about 40 and 100, 84Kr/132Xe between 2 and 5) independent of any differences they may have had before the grinding procedure.

The three samples that show deficits of krypton and xenon are Lafayette, ALHA77005, and Chassigny. Lafayette, which belongs to the group of nakhlites, is the one with the highest olivine abundance (about 17%) in this study (Treiman 2005) and contains approximately 10% of iddingsite. As nakhlites do not contain shock melt, iddingsite may appear a likely host phase from which trapped gas could have been lost. This, however, does not fit with the results for Governador Valadares, which contains a similar amount of iddingsite, and thus would be expected to be similar to Lafayette, contrary to observation. A possible reason may be enhanced exposure of Lafayette to air prior to collection, e.g., through cracks or find location conditions and terrestrial residence time. Nakhla is an observed fall, and Governador Valadares may have been collected on the morning after its fall. In contrast, Lafayette is a find with a terrestrial age of about 3000 yr (Treiman [2005] and references therein) and was actually collected as a glacially striated stone in the first place. Whatever the reasons, it is important to keep in mind that among the nakhlites only for Nakhla there are multiple measurements of untreated samples (Ott 1988; Busemann and Eugster [2002] as well as Mathew and Marti [2002] report Kr and Xe-data; while Gilmour et al. [1999, 2001] report Xe only), while for Governador Valadares (Swindle et al. 1989) and Lafayette (Ott et al. 1988) there are two reference measurements only (Table 3). This is a problem, of course, because of sample inhomogeneities, and the small amount of sample actually used for noble gas analysis.

The second meteorite to display “loss” is ALHA77005, an olivine rich (∼60% olivine; Treiman et al. 1994) shergottite, which also contains a considerable amount of shock glass (∼14%), whereas the third one is Chassigny, a dunite (about 92% olivine, Floran et al. 1978). In Chassigny melt inclusions in olivine are described that contain amphiboles with “substantial hydrogen” and some fluorine (Floran et al. 1978). Grinding may have led to loss of those volatiles and of the noble gases that may have accompanied them. It appears questionable, however, whether this can account for the entirety of the loss, because melt inclusions make up only 0.3% of the rock. The case is different for ALHA77005, which contains a considerable amount of shock glass. Shock glass is known as the carrier of Martian atmosphere as found in EETA79001 (Bogard and Johnson 1983; Bogard and Garrison 1998), which may have been lost from this phase in principle.

Loss, however, is not very likely to have occurred under mild crushing (grinding) conditions as employed here. This is indicated by numerous experiments on terrestrial samples, which have been crushed to extract the noble gases in fluid inclusions, and subsequently heated to understand the signatures of the minerals enclosing these (e.g., Kelley et al. 1986; Turner and Songshan 1992; Kendrick et al. 2001, 2006; Kendrick and Phillips 2009; Kamensky and Skiba 2011). It also fits with the observation that the ground samples have retained their 3He and 4He and compare well to literature data (see Schwenzer et al. 2008), and it is in accordance with “true” crushing experiments on Martian meteorites as well. Crushing EETA79001 (Wiens 1988) revealed the special EETV-component that was interpreted to be located in vesicles, while heating of the sample after the crushing experiment showed Martian meteorite noble gas signatures as seen in other heating experiments of this and other Martian meteorites. While hints for a possible redistribution of Martian components during crushing have been seen (Mohapatra et al. 2009), no evidence for losses has been found in the same experiment. Thus, the major concern with mildly crushed samples is not loss of noble gases from lattice sites but rather contamination by introduced air. It is also noteworthy that both Chassigny and ALHA77005 (with apparent “losses”) join in the narrow area of the ground samples in the elemental ratios plot (Fig. 2), which argues for a common noble gas component introduced into them independent of what they may have lost.


In conclusion, we find air contamination far beyond pure adsorption being introduced into all our samples. In all our experiments the elementally fractionated terrestrial air remains in the sample to much higher temperatures than expected for simply adsorbed atoms. From this it becomes clear that simple adsorption is not sufficient to explain the observations, and a more tightly bound state has to be assumed. In turn, the persistence of elementally fractionated air above the first temperature step also means that care has to be taken with the interpretation of sample results. Natural incorporation of elementally fractionated air (Mohapatra et al. 2009) and any fractionation observed in the course of this study drives the elemental ratios of the incorporated gas to lower 36Ar/132Xe and 84Kr/132Xe ratios than found in air or surface water. For the interpretation of Martian meteorites, this matters, because the component “Martian interior” has low 84Kr/132Xe and 36Ar/132Xe ratios of about 1.2 and about 19, respectively (Ott 1988). At the same time the 129Xe/132Xe ratios of the Martian interior component and air are only moderately different (1.03 ± 0.02 [Ott 1988] versus 0.9832 ± 0.0012 [Ozima and Podosek 2002]). Bulk measurements of Martian meteorites commonly fall in the range between Martian atmosphere and such a low 36Ar/132Xe, 84Kr/132Xe, 129Xe/132Xe component (Fig. 2), and if the assumption (hope) that any air contamination is removed in the first temperature step does not hold true, this may lead to significant interpretation problems. Even though both end member compositions are fairly well defined, it is not possible to determine in which ratio the Martian interior component is mixed with Martian atmosphere, or even if the Martian interior component is present at all in a given bulk sample. Resorting to additional isotopic ratios that allow a distinction between the Martian interior and fractionated air components (e.g., 136Xe/132Xe) may be an obvious solution. But it may not always be feasible, due to the lower concentration of those isotopes and consequently more difficult measurements.

Our experiments have also shown that the amount of contaminating gas introduced into the samples may be large enough to overwrite the preexisting pristine signature of the sample. Bulk samples after grinding (except for Lafayette, which seems to be a special case, as discussed above) fall into a limited range centered around 36Ar/132Xe about 60 and 84Kr/132Xe about 3.5, which notably is rather similar to what we found for air adsorption on “cleaned” surfaces of both olivine and gibbsite (Fig. 6). Since the observed contamination may make it impossible to detect a meaningful Martian signal in the data for Martian meteorites, care has to be taken in sample preparation. In addition, natural weathering may have introduced a large amount of elementally fractionated air even before the samples were collected. In small samples weathering can be pervasive, as we found in our study of Yamato Martian meteorites, and the Martian noble gas signature may become seriously diluted (Schwenzer et al. 2009). In weathering, additional processes are at work and their investigation will be the subject of a separate publication. We note, however, that although it may be tempting to identify the host phase of such contamination and to attempt removing it by leaching, when trying to “clean” a sample from terrestrial contamination care needs to be taken to not introduce just as much as is removed (Schwenzer et al. 2005; Schwenzer and Ott 2006; Ott 2008).

Figure 6.

84Kr/132Xe versus 36Ar/132Xe of three laboratory experiments and the ground Martian meteorites. References for Martian interior and atmosphere and terrestrial air are as in Fig. 5.

Last but not least, the processes observed in laboratory experiments can be important for future spacecraft missions to Mars and for the interpretation of the fractionated noble gas component in the nakhlite Martian meteorites and on Mars. Spacecraft developments may lead to in situ Ar-dating on Mars (Bogard 2009; Talboys et al. 2009) and potentially the analysis of other noble gases, too. The first measurement of Martian noble gas signatures on Mars since Viking will be carried out by the spacecraft currently cruising to Mars: the rover Curiosity. Mars Science Laboratory carries the “Sample Analyzer at Mars” (SAM) instrument, which includes a mass spectrometer capable of measuring noble gases in the atmosphere and—through degassing up to ∼1000 °C—in Martian solids (Mahaffy 2008). On the Martian surface weathering through wind, ice, and—in the Martian past—water has shaped the surface and the individual grains. It is this surface which will be in reach of the instrument. Therefore, in order for a proper interpretation of the obtained results, it will be critically important to understand potential incorporation mechanisms of past and present Martian atmosphere into those samples.


Acknowledgments— Thanks go to W. Hofmeister, Mainz, for providing the gibbsite sample. M. Trieloff, Heidelberg, carried out the phase determinations; and W. Debschütz, Clausthal, provided the BET surface measurements. Thanks to S. Flude and S. P. Kelley, Milton Keynes, for interesting discussions and comments on an earlier version of the manuscript. Comments by J. D. Gilmour, T. Swindle, and M. Trieloff on an earlier version of the manuscript are acknowledged; and reviews by K. Marti and S. V. S. Murty improved the presentation of the manuscript. We thank T. Jull for his supportive editorial handling of the manuscript. We acknowledge the support by the German Science Foundation (Grants Jo305-1/1, Ot 171/3-4, and Schw1232/1-1).

Editorial Handling— Dr. A. J. Timothy Jull