Abundances and implications of volatile-bearing species from evolved gas analysis of the Rocknest aeolian deposit, Gale Crater, Mars



The Sample Analysis at Mars (SAM) instrument on the Mars Science Laboratory (MSL) rover Curiosity detected evolved gases during thermal analysis of soil samples from the Rocknest aeolian deposit in Gale Crater. Major species detected (in order of decreasing molar abundance) were H2O, SO2, CO2, and O2, all at the µmol level, with HCl, H2S, NH3, NO, and HCN present at the tens to hundreds of nmol level. We compute weight % numbers for the major gases evolved by assuming a likely source and calculate abundances between 0.5 and 3 wt.%. The evolution of these gases implies the presence of both oxidized (perchlorates) and reduced (sulfides or H-bearing) species as well as minerals formed under alkaline (carbonates) and possibly acidic (sulfates) conditions. Possible source phases in the Rocknest material are hydrated amorphous material, minor clay minerals, and hydrated perchlorate salts (all potential H2O sources), carbonates (CO2), perchlorates (O2 and HCl), and potential N-bearing materials (e.g., Martian nitrates, terrestrial or Martian nitrogenated organics, ammonium salts) that evolve NH3, NO, and/or HCN. We conclude that Rocknest materials are a physical mixture in chemical disequilibrium, consistent with aeolian mixing, and that although weathering is not extensive, it may be ongoing even under current Martian surface conditions.

1 Introduction

The Mars Science Laboratory (MSL) rover Curiosity landed in Gale Crater at the base of Mount Sharp on 6 August 2012 (Coordinated Universal Time). The first sampling location was an aeolian deposit named Rocknest, a sand shadow that has accumulated in the lee of a group of rocks [Blake et al., 2013]. The deposit is armored with particles 1–3 mm in diameter and is covered with bright airfall dust deposits indicating that it is not presently active. The interior of the deposit is composed primarily of darker material with <150 µm size particles. This location was selected for two primary reasons: first, it contained fine-grained material that could be used to remove potential terrestrial contamination on the internal surfaces of the Collection and Handling for Interior Martian Rock Analysis (CHIMRA) system on the rover prior to delivery of samples to the Sample Analysis at Mars (SAM) instrument and second, Rocknest material is an aeolian feature representative of local and possibly global Martian soil (where the term “soil” indicates loose, unconsolidated materials that can be distinguished from rocks, bedrock, or strongly cohesive sediments and may or may not include organic materials). Because soils contain some fraction of Martian dust, which is globally distributed by the wind, Rocknest soil is geochemically relevant to global soils on Mars. Independent of arguments pertaining to the local versus global-mixing source for basaltic soils is the observation that Rocknest soil is chemically very similar to other basaltic soils measured elsewhere on Mars [Blake et al., 2013; Gellert et al., 2004; Yen et al., 2005]. Consequently, analysis of Rocknest samples by the SAM instrument suite on MSL has implications for both local and global soils.

The SAM instrument contains a quadrupole mass spectrometer (QMS), a gas chromatagraph system (GC), and a tunable laser spectrometer (TLS) for analysis of atmospheric and solid samples [Mahaffy et al., 2012]. Solid samples are heated and the gases released are analyzed with a mass spectrometer, a technique commonly called evolved gas analysis (EGA) that has been used to analyze volatile-bearing samples in terrestrial settings for many years [e.g., Carthew, 1955; Escardino et al., 2008; Grim and Rowland, 1942; Hyatt et al., 1958; Kissinger, 1956] and was used on Mars by the Viking and Phoenix landers [e.g., Biemann et al., 1977; Boynton et al., 2009]. Examples of volatile-producing reactions are thermal dehydration (e.g., hydrated sulfate losing its structural water), thermal dehydroxylation (e.g., Fe-oxyhydroxide losing its structural OH), and other thermal decomposition reactions such as a carbonate decomposing to a metal oxide and carbon dioxide. The volatile release temperature depends on the thermodynamics of the reaction and therefore the decomposition temperature and gases evolved can be used to constrain the initial sample composition. However, many factors, such as particle size [Archer et al., 2013] and interactions with other reactive species [Cannon et al., 2012], can affect decomposition temperatures to the point that the initial composition cannot be known definitively. Although EGA does not provide definitive mineralogy, the high sensitivity of mass spectrometers makes this a powerful tool to both confirm the presence of volatile-bearing species detected at high abundances and detect gases released by species at low (i.e., ppm) concentrations.

During analysis of Rocknest soil, the SAM instrument detected the release of four major volatiles, H2O, SO2, CO2, and O2, as well as other potential minor species including H2S, HCl, NH3, HCN, and NO across a broad temperature range (Figure 1). In this paper, we (1) detail the method used to determine evolved gas abundances, (2) report the detection of the major volatile species released during Rocknest EGA, expanding upon the initial report that included the major volatiles released from Rocknest samples (H2O, SO2, CO2, and O2) [Leshin et al., 2013], (3) give abundances for the minor volatile species H2S, HCl, NH3, HCN, and NO, (4) discuss the possible sources of the evolved gases with particular emphasis on H2O, CO2, and nitrogen-bearing species, and (5) discuss the implications for the sources of all of these gases being present in a single sample and what this sample can tell us about Mars in general. Companion papers go into greater depth on the possible sources of S-bearing gases [McAdam et al., 2014] and an oxychlorine phase as the primary source of O2 and chlorinated species detected by SAM [Glavin et al., 2013].

Figure 1.

The primary gases evolved during SAM thermal analysis plotted versus the sample temperature range over which they were detected. Evolved gas analysis (EGA) products are arranged in order of decreasing molar abundance starting at the top.

2 Experimental Methods

On sol 93(Ls 204, Mars Year 31), the Sample Acquisition/Sample Processing and Handling (SA/SPaH; described in Anderson et al. [2012]) system on MSL acquired a fifth scoop of Rocknest soil. This material was then passed through the 150 µm sieve pathway, with ~90% of the sample volume passing through the sieve (Figure 2) (L. Jandura, personal communication, 2013). This is consistent with the description of the Rocknest aeolian deposit as a primarily fine-grained sand deposit armored with larger particle size grains [Blake et al., 2013].

Figure 2.

Mastcam images of (left) the fifth scoop of Rocknest material that was analyzed by Sample Analysis at Mars (SAM) and (right) the coarse-grained material that remained in the scoop after sieving. Roughly 90% of the initial material passed through a 150 µm sieve, portions of which were analyzed by SAM and CheMin. Mastcam images 0093MR0608000000E1 and 0128ML0734000000E1. Image credit: NASA/JPL-Caltech/MSSS.

Between sols 93 and 117, four samples weighing 50 ± 8 mg (2σ standard deviation), all from the same scoop, were delivered to SAM for subsequent analysis (Table 1). The processing of this sample inside CHIMRA was expected to produce a homogenous mixture for delivery to analytical instruments.

Table 1. List of Rocknest Scoop 5 Acquisition and Analysis Datesa
ActionSol Number
  1. a

    The sample was acquired on Sol 93, Ls 204.

Scoop 5 acquired93
Run 1 analysis93
Run 2 analysis96
Run 3 analysis99
Run 4 analysis117

Prior to evolved gas analysis of the Rocknest sample, a single quartz cup (of 59 total quartz cups in SAM) that had previously been heated to above 800°C to remove potential contaminants was rotated into position underneath the SAM solid sample inlet tube to receive a single portion of sieved sample from CHIMRA. After the sample was received, the quartz cup was sealed inside a SAM pyrolysis oven. The bottom of each quartz sample cup contains a porous quartz frit that allows He carrier gas to flow up through the sample for efficient removal of evolved gases during heating [Mahaffy et al., 2012]. Samples were heated to ~835°C at a ramp rate of 35°C/min at 25 mbar pressure with a model-derived flow rate of 0.8 standard cubic centimeters per minute (sccm) for the He carrier gas. Evolved gases are measured continuously by the QMS during thermal analysis and select temperature cuts are sent to the TLS or GCMS during each run.

2.1 Abundance Calculations

Evolved gas abundances on Mars are tied to analyses of samples of known quantity and composition on Earth. Before integration into the spacecraft, multiple samples of a carbonate and iron sulfate mixture were analyzed in the flight instrument under conditions identical to those described above. FeSO4•4H2O (Sigma-Aldrich, product # F8048, batch # 048K0680) and CaCO3 (calcite — Sigma-Aldrich, standard A.C.S. reagent, product #398101, batch # 16826BH) were mixed with fused silica (FS-120 from HP technical ceramics [see Conrad et al., 2012]) that had been crushed and sieved to collect the 20–150 µm size fraction and baked at 550°C for 2 h in air to remove organic contamination. During pyrolysis, these samples thermally decomposed and released H2O and SO2 from the FeSO4•4H2O and CO2 from calcite, shown in the reactions:

display math(1)

followed by

display math(2)


display math(3)

Thermal gravimetric analysis (TGA; described in more detail below) was conducted on the same material run in SAM to measure the completeness of these reactions. When run under SAM-like pressure, gas, and flow conditions, calcite thermal decomposition is complete by 835°C, with ±1% accuracy. The decomposition of FeSO4 in this sample is not as straightforward. Although FeSO4 decomposition begins at a lower temperature and should be complete by the end of the temperature ramp, a portion of iron sulfate decomposition overlaps with the onset of calcite decomposition. When FeSO4 decomposes in the presence of CaO (the CaCO3 decomposition product), the SO2 is scrubbed from the system by CaO and forms CaSO4 [e.g., Vamvuka et al., 2004]. To quantify the S remaining in each sample after pyrolysis, the residues were analyzed via laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS). The method executed 300 scans across the mass range, and the element menu included the following masses (and dwell times per scan): 29Si (1 ms), 30Si (1 ms), 32S (1 ms), 33S (10 ms), and 34S (5 ms). Medium resolution (m/Δm = 2000) detection parameters were implemented in order to resolve Si and S peaks from potential isobaric interferences, such as 12C16OH+ and 58Ni++ on mass station 29Si+ and 16O2+ on mass station 32S+. Each pyrolysis residue was analyzed multiple times (two to four replicate analyses) and bracketed with two reference materials, namely NIST SRM 610 (a calcium-sodium aluminosilicate glass doped with 61 trace elements) and USGS basaltic glass BHVO-2. Silicon (via 29Si and/or 30Si) served as the internal standard, thus allowing for quantification and a statistical analysis of the S measurements. The external precision of each sample analysis (including replicates) ranged from 5 to 14% (2σm). This analysis found that 7 ± 2% of the initial S was not released during thermal decomposition of the FeSO4. The moles of residual S measured were subtracted from the total moles of S in the initial sample, giving the moles S that were released in SO2 during thermal decomposition.

By knowing the initial masses of CaCO3 and FeSO4•4H2O weighed into SAM cups, verifying that carbonate decomposition proceed to completion, and measuring the residual S in the pyrolyzed sample, the number of moles produced during thermal analysis can be calculated. Integrating the number of counts per second measured by SAM for various mass/charge (m/z) values gives the total counts for a given gas release. Using these two values, we compute a mol/counts ratio that was used to calculate the molar abundance of CO2, H2O, and SO2 released from samples of unknown composition (Table 2). It is important to note that SAM has two pyrolysis ovens and each oven will have its own mol/counts numbers. The numbers reported in Table 2 apply only to samples run in pyrolysis oven 1 (oven 2 has not yet been used on Mars). This approach is relatively straightforward but there are complications related to detector saturation, differences in isotopic composition (notably D/H) between Mars and Earth, the unmonitored loss of water from FeSO4 · 4H2O at low pressure, and gas species not released from these minerals—factors that must be taken into account to correctly calculate molar abundances.

Table 2. List of m/z Value Used, How it Relates to the Primary Peak, and the mol/counts Value for Eacha
Gas Species (Major Molecular Species)m/z UsedTypemol/counts
  1. a

    These numbers only apply to samples run in oven 1—oven 2 has not yet been used on Mars. Furthermore, these values may change as detector sensitivity changes with time.

CO2 (m/z 44)12ionization fragment1.6 × 10−13 ± 7.9 × 10−16
22ionization fragment3.8 × 10−13 ± 4.6 × 10−14
45isotopologue2.3 × 10−13 ± 1.5 × 10−14
46isotopologue6.7 × 10−13 ± 3.5 × 10−14
H2O(m/z 18)19isotopologue1.8 × 10−12 ± 2.1 × 10−13
20isotopologue1.2 × 10−12 ± 4.6 × 10−14
SO2 (m/z 64)50isotopologue (of SO)3.8 × 10−13 ± 4.3 × 10−14
66isotopologue1.4 × 10−13 ± 2.5 × 10−14

2.1.1 Correcting for Detector Saturation

SAM uses a detector with a fixed (but large) dynamic range and is tuned to detect trace species; therefore, the most abundant gases saturate the detector and have no useful information with regards to the amount of gas that is evolved. During calibration runs, for the three gases being quantified (H2O, CO2, and SO2), m/z 17 and 18 saturated for H2O, m/z 44 for CO2, and m/z 48 and 64 for SO2. To circumvent this problem, we calculate mol/counts values for isotopologues of a given species or ion fragments formed in the detector (where “counts” is the number of counts per second integrated over the entire gas release peak). For example we can use m/z 45 and 46 (stable isotopologues of CO2) as well as m/z 12 (C formed by fragmentation in the detector) and m/z 22 (doubly ionized mass 44, i.e., m = 44 z = 2) in place of the saturated m/z 44. We verified that these m/z values reflect CO2 evolution and are not caused by another gas species present by looking at ratios between these m/z values to verify that they all track together, indicating that they are indeed isotopologues/fragments of CO2 and can therefore be used to compute CO2 abundance. Similar techniques were used to characterize mol/count values for H2O and SO2 (Table 2).

2.1.2 Correcting for Earth/Mars Isotopic Differences

Although the differences in oxygen isotopic composition between the earth and Mars are small on an absolute scale (a δ18O of ~50 ‰ [Leshin et al., 2013] is small compared to other uncertainties in our measurements), the difference in D/H values (Mars D/H is 4–6× Earth [Leshin, 2000; Leshin et al., 1996; Novak et al., 2011]) must be taken into account. There are two ways in which this can be done, and both provide comparable results, adding confidence to the final value because the methods are largely independent. First, because m/z 17 and 18 from H2O saturate during the run, only m/z 19 and 20 can be used. The ratio of m19 to m20 of the terrestrial iron sulfate tetrahydrate can be measured directly from the three SAM runs. The mean 19/20 ratio for background corrected data, computed for the portion of the run where there was a significant water signal, is 0.63 ± 0.05 (1σ standard deviation of the average of three runs). This same ratio can also be computed for water released during the four Rocknest runs and the value is 1.19 ± 0.04 (1σ standard deviation of the average of the four runs). The (19/20_Rocknest)/(19/20_terrestrial_FeSO4•4H2O) ratio is thus 1.89 ± 0.09. Because the isotopic composition of m/z 20 is more strongly controlled by differences in the oxygen isotopic values (H216O18O) than enhanced deuterium (D2O) and the Mars/Earth oxygen isotopic differences are small on an absolute scale, we assume that the difference in the m/z 19/20 ratio is primarily caused by differences in m/z 19. Therefore, we divide the integrated counts of m/z 19 in Rocknest soils by the computed Earth:Mars 19/20 ratio in order to use the mol/counts number computed from terrestrial experiments, i.e., m19_counts(Rocknest)/(19:20_Rocknest/19:20_Earth) = m19_counts(Earth).

The other way of accounting for the D/H ratio between Mars and Earth is to estimate the m/z 18 count rate based on assumed terrestrial isotopic values to establish a mol/counts value for m18 and then do the same m/z 18 estimation on Mars using measured Martian isotopic values. For the Rocknest analyses, the H2O composition assumed for these abundance calculations was taken as the TLS result for Rocknest run #3 (which sent the largest portion of the H2O release during EGA to the TLS): δD = 5880 ± 60‰ and δ18O = 84 ± 10‰ [Leshin et al., 2013]. The calculated abundances for these two approaches agree within the uncertainty of the measurements, lending confidence to the robustness of the abundance calculation against effects from any potential O and H isotopic differences between Earth and Mars. The agreement of the numbers also strengthens the case for the accuracy of the Martian isotopic values measured by the TLS because, if they were incorrect, the abundance calculations based on the two approaches would not agree.

2.1.3 Water Loss From FeSO4•4H2O at low Pressure

One complication in computing water abundances using the SAM ground calibration data is the loss of water from FeSO4•4H2O when it is exposed to low relative humidity environments (e.g., low pressure). In addition to the few minutes at 25 mbar before the pyrolysis ramp began, samples were exposed to low pressure or vacuum for days/weeks as they sat inside SAM in a Mars simulation chamber before analysis. The rapid water loss when exposed to vacuum and unknown pressure history experience by the sample make estimating the amount of water in the sample at the beginning of the run extremely uncertain. Fortunately, the thermal decomposition behavior of FeSO4•4H2O provides a better way to quantify the amount of water lost. There are two discrete water release peaks seen in the decomposition of FeSO4•4H2O (Figure 3). The first water release begins as soon as the heating starts and is accompanied by a large endotherm (the large dip in the heat flow data between 50 and 200°C). The second water peak is seen just below 250°C and is associated with a strong exotherm (the coincident peak in the DSC data). This exotherm is likely due to the recrystallization of dehydrated FeSO4, with the energy released speeding up the water evolution from sample that was not yet completely dehydrated, causing the peak in the EGA data. Using instruments configured to run under SAM-like conditions, and analyzing the same material that was run in the SAM ground calibration experiments, we have shown that this second water peak releases a repeatable fraction of the initial water and can therefore be used to calibrate number of moles of H2O per integrated counts in the SAM instrument. The amount of water lost was reproducible when varying the amount of sample run, if the FeSO4•4H2O was run by itself or mixed with another material such as calcite or fused silica, or if the samples were run at different temperature ramp rates. However, it is possible that the amount of water loss during this exotherm could depend on other untested instrument/sample parameters. To address this possibility, the hydrated iron sulfate used in prelaunch runs will be analyzed by the SAM test bed, a functional equivalent to the SAM instrument on Mars [Mahaffy et al., 2012]. This work is rather involved and is beyond the scope of this paper. Until those analyses are complete, the reported uncertainty for H2O abundances includes the uncertainty in the amount of water present in the prelaunch FeSO4•4H2O runs.

Figure 3.

Thermal analysis of CaCO3/FeSO4•4H2O mixture. Data shown up to 350°C. This sample was run in a commercial instrument configured to match SAM instrument parameters. After an initial water release at 75–200°C, there is a secondary water release with an accompanying exotherm and small mass loss (in box). The amount of water lost from the sample from this event was repeatable at the 5% level (2σ). A similar water release in the 200–250°C temperature range is seen in SAM test data. The data shown here include the evolved gas, differential scanning calorimetry, and thermal gravimetric analysis.

This supporting thermal analysis work was conducted on a Netzsch STA 449 F1 Jupiter with simultaneous TG/DSC (thermal gravimetry/differential scanning calorimetry) capability at the NASA Johnson Space Center. The TGA has a resolution of 0.025 µg at a weighing range of 5 g. A Pfeiffer Thermostar GSD 320 with a 1–200 AMU range PrismaPlus quadrupole mass spectrometer is coupled to the TG/DSC for evolved gas analysis via a 1 mm inner diameter stainless steel tube. The samples are run at a pressure of 30 mbar using He as a carrier gas with a flow rate of 20 sccm. These parameters have been tuned so that the release profile of gases is a close match to what was measured by SAM in the calibration runs already discussed. Ordinarily, samples are heated at a ramp rate of 35°C/min, the ramp rate used in SAM prelaunch calibration as well as Rocknest runs. However, to accurately determine the mass loss in the second water release, the ramp speed was slowed to 1°C/min. For these runs, ~30 mg of the same FeSO4 • 4H2O that was run in SAM ground calibration tests was weighed in an alumina crucible using a Mettler-Toledo XP26 balance with an accuracy of ±0.01 mg. Samples were heated to >250°C at which point the small water release was complete (Figure 4). A second run of the same sample was done to insure repeatability. The remaining material (presumably FeSO4) was weighed in the balance after thermal analysis to verify that the starting material was the same for each run. The average weight loss for both samples was 31.37 ± 0.04% (1σ standard deviation from the mean), indicating that the initial material was the same and did not vary with amount of time exposed to lab air or the humidity on the day the sample was run.

Figure 4.

Thermal gravimetry results for the thermal analysis of the same FeSO4 · 4H2O that was run in SAM. The large mass loss between 40 and 150°C is the initial water lost from the sample. The second smaller mass loss is associated with an exotherm (likely a recrystallization as the mineral loses the last of its water) and a small, sharp H2O peak in the evolved gas data. The small offsets in the curves can be attributed to the differences in total sample size (~3 mg) as well as the slightly different amounts of time spent pumping on the sample before the ramp began, which again demonstrates the importance of using the smaller second peak to calibrate the SAM water abundance numbers.

The average mass loss for the second, sharp water release, expressed as a percent of the total initial FeSO4•4H2O mass was 0.95 ± 0.02% (also 1σ standard deviation), showing that the mass loss is repeatable and therefore can be used to determine a more accurate mol/counts number for water in the SAM calibration runs (amount of sample used in SAM calibration runs is listed in Table 3).

Table 3. Table of Test Identification Numbers (TID) Values Used for Pyrolysis Oven 1 mol/counts Calibration and the Amounts of CaCO3, FeSO4 · 4H2O, and Fused Silica (FS120) Used in Each Run
TIDCaCO3 Mass (mg)FeSO4•4H2O Mass (mg)

2.1.4 Estimating Gases not Released During CaCO3/FeSO4•4H2O Evolution

Finally, the techniques outlined are applicable for calculating abundances of H2O, SO2, and CO2 but are not applicable for other gases released during pyrolysis of Rocknest soils. To calculate the total moles of O2 released we need to know the ionization cross section of O2 in SAM relative to a gas for which we have already calculated a mol/counts ratio. This is accomplished by utilizing data from another prelaunch run where SAM analyzed an equimolar mixture of gas containing O2, CO2, N2, and Ar (test identification number [TID] 21976). The ratio of O2 (m/z 32) to various masses representative of CO2 (as stated earlier: m/z 12, 22, 45, and 46) was computed from this calibration gas run. Using these ratios, we turn a measured O2 count rate into an equivalent CO2 count rate, for which we have already calculated a mol/counts ratio. This allows us to compute O2 abundances based on the ionization efficiencies of O2 and CO2 in SAM.

For other minor species, no prelaunch calibration data exist, so literature values for relative ionization cross sections were used to convert count rate of a species of unknown mol/counts to the count rate of a species with known mol/counts (CO2 in this case). This technique adds additional uncertainty because differences in mass spectrometer systems (e.g., how the QMS is tuned) leads to different ionization efficiencies. This uncertainty is not included in the reported error in calculated abundances; however, typical variations among cross sections reported in the literature are less than 20%. Abundance estimates for HCl, H2S, NO, HCN, and NH3 use relative ionization cross sections reported by Fitch and Sauter [1983], Nakao [1975], and Rapp and Englander-Golden [1965]. The relative ionization cross section values used to convert counts of gases with unknown mol/counts numbers to gases with known values are listed in Table 4.

Table 4. Relative Ionization Cross Sections for Gas Species not Evolved During CaCO3/FeSO4•4H2O Decomposition
Gasm/z RatiosValueSource
  1. a

    The 27/44 value for HCN was estimated from a plot of ionization cross section relative to nitrogen versus the number of electrons/molecule in Nakao [1975] because the specific value of the relative ionization cross section of HCN was not reported.

O232/1244.9 ± 2.9Sample Analysis at Mars (SAM) equimolar gas mix
 32/4558.5 ± 5.8
NO44/301.12[Nakao, 1975]

3 Results and Discussion

The four primary volatiles released from the Rocknest samples were, in order of decreasing average molar abundance, H2O, SO2, CO2, and O2 (Table 5). These species are all present at levels much higher than instrument background, determined by a pyrolysis run of an empty cup prior to analysis of Rocknest samples, and they are unambiguously Martian in origin (with the exception of a very small amount of CO2 that could be due to combustion of terrestrial organics). Although the abundances differ slightly between runs, the shape of the H2O, CO2, and O2 evolved-gas profiles is fairly consistent run-to-run (Figure 5). The shape of the SO2 release, on the other hand, varies run-to-run, and the differences in calculated abundances are statistically significant (Table 5 and Figure 8). The apparent heterogeneity in S-bearing species is not currently understood.

Table 5. Molar Abundances (in µmol) and Weight % of the Major Volatiles Released From Rocknest Samplesa
 Molar Abundance (μmol)
  1. a

    The error for CO2, SO2, and O2 molar abundances is the 2σ standard deviation from the average of abundance numbers calculated from different m/z values. The uncertainty in the H2O abundance is due to uncertainty in prelaunch water calibration. The uncertainty on the weight percent values includes these errors as well as the 2σ uncertainty on the amount of sample delivered to SAM for each run, 50 ± 8 mg.

 Run 1Run 2Run 3Run 4
H2O43.3 ± 24.366.5 ± 37.354.5 ± 29.755.9 ± 30.9
SO22.9 ± 0.213.7 ± 1.921.7 ± 2.910.5 ± 1.4
CO28.3 ± 2.010.8 ± 2.610.1 ± 2.410.4 ± 2.5
O23.0 ± 0.45.1 ± 0.63.7 ± 0.43.7 ± 0.5
 Weight Percent
 Run 1Run 2Run 3Run 4
H2O1.6 ± 0.92.4 ± 1.42.0 ± 1.12.0 ± 1.2
SO3 equiv0.5 ± 0.12.2 ± 0.53.5 ± 0.71.7 ± 0.3
CO20.7 ± 0.21.0 ± 0.30.9 ± 0.30.9 ± 0.3
ClO4 equiv0.3 ± 0.10.5 ± 0.10.4 ± 0.10.4 ± 0.1
Figure 5.

A plot of the major molecular species of the most abundant volatiles released during thermal analysis of four aliquots of Rocknest scoop #5 materials delivered to SAM.

Other gas species that were detected at much lower abundances include HCl, H2S, NO, HCN, and NH3 (detection of NH3, NO, and HCN is preliminary pending deconvolution of other compounds that have overlapping m/z signals). The gases released indicate that the Rocknest samples are a complicated mixture of phases, the composition of which will take time to unravel and will require analysis of analog samples in laboratory instruments. We will discuss possible sources of these gases, treating them both individually and as possible products of interactions with other material present in the samples.

3.1 H2O

The water released from Rocknest materials, an average of 2.0 ± 1.3% across all four runs, is broadly consistent with what has previously been detected on Mars for similar samples. The GCMS instruments on both Viking landers detected ~1% water content in soils heated to 500°C [Biemann et al., 1977] and the Neutron Spectrometer aboard the Mars Odyssey spacecraft detected a water equivalent hydrogen abundance of ~2–4% in the Gale Crater region [Feldman et al., 2004]. The SAM results are also consistent with the measurement of 1.3% water equivalent hydrogen of Rocknest material made by the Dynamic Albedo of Neutrons (DAN) instrument on MSL [Jun et al., 2013].

The water release seen in Rocknest EGA data could be from a number of different sources. The broad nature of the release indicates that it is unlikely to be derived from one source and instead has contributions from multiple sources. This is consistent with the data from the Chemistry and Mineralogy (CheMin) instrument on MSL which performed X-ray diffraction analyses on different portions of the same material that was delivered to SAM. If the 1–3 wt.% water detected by SAM were all associated with one mineral, that mineral is likely to have been detected by CheMin, which has a detection limit of <3 wt.% [Bish et al., 2013; Blake et al., 2012]. In addition to the crystalline component of the sample, CheMin detected an X-ray amorphous component of the soil, comprising ~45 wt.% of the total mass [Bish et al., 2013; Blake et al., 2013]. If the amorphous component is the source of all the water detected by SAM, it would contain ~4 wt.% H2O.

The majority (80 ± 2.5% [1 sigma uncertainty]) of H2O was driven off in a broad release below 450°C, peaking at 250°C, during pyrolysis of the Rocknest material. The remainder of the water was released between 450 and 850°C and was similarly broad in nature (Figure 5). The lower temperature water release (<450°C) may be adsorbed, associated with an exchange cations (e.g., phyllosilicate, zeolite), structural H2O (e.g., gypsum, hydrated Mg-sulfates, and many other potential phases), structural OH (e.g., Fe-oxyhydroxides, jarosite, and many other potential phases), or occluded in glass or minerals. Adsorbed H2O and water associated with exchange cations will generally be evolved below 300°C; thus, adsorbed water and water associated with exchange cations cannot be the only source for the broad lower temperature water release up to 450°C.

Because of the large number of possible sources of water in this temperature range and the broad temperature range over which the water signal is seen in SAM data, EGA alone cannot identify the water-bearing phases in the Rocknest material. However, based on chemistry and no evidence of crystalline water-bearing phases from CheMin, the water-bearing phases in Rocknest are most likely associated with the X-ray amorphous component. The composition of the amorphous component does provide some constraints on the water-bearing phases. The bulk chemical composition of the amorphous phase, calculated from APXS, CheMin, and MER data [Blake et al., 2013] on a H2O/OH-free basis is silica poor (SiO2 ~ 37 wt.%), has a high SiO2/Al2O3 ratio (~18) and is iron, sulfate, phosphate, and Cl rich (FeO + Fe2O3 ~ 23 wt.%, SO3 ~ 11 wt.%, P2O5 ~2 wt.%, Cl ~ 1.4 wt.%), This nonbasaltic bulk composition implies chemical weathering and incorporation of volatile elements, but it does not provide direct evidence for the speciation of those volatiles.

The evolution of chlorohydrocarbons and O2 during pyrolysis is an indicator of a perchlorate salt in Rocknest materials [Glavin et al., 2013]. Although CheMin did not detect a perchlorate salt, calculated perchlorate abundances based on O2 released during EGA are below CheMin detection limits (Table 5), leaving hydrated perchlorate as a possible water-bearing phase. Based on the temperature of the O2 release and the CaO content of the amorphous material, Ca-perchlorate is a good candidate for the perchlorate [Glavin et al., 2013]; however, Mg-perchlorate cannot be excluded as a candidate phase in Rocknest based on the MgO content of the amorphous component. Hydrated Ca-perchlorates (Ca(ClO4)2•4H2O or Ca(ClO4)2•8H2O) are predicted to be stable at the surface of Mars [Nuding et al., 2013]; hence, Ca-perchlorate is a candidate H2O-bearing phase in Rocknest materials. All of the H2O in Ca-perchlorate is driven off at temperatures below 450°C based on laboratory experiments performed in the Netzsch instrument described above on synthetic Ca(ClO4)2 run under SAM-like conditions.

As stated above, Si, Fe, and S are major components of the amorphous component. The elevated low angle background in the CheMin Rocknest X-ray diffraction patterns may result from an allophane-like or hisingerite-like phase [Bish et al., 2013]. These phases contain water in their structures and evolve this water below well below 450°C. Other Fe-bearing phases also release water below 450°C [Archer et al., 2013]. Spirit and Opportunity's Mössbauer Spectrometer detected nanophase iron oxides (npOx) in regolith materials (dust component) at Gusev crater and Merdiani Planum [Morris et al., 2004, 2008, 2006]. Nanophase materials are generally X-ray amorphous and may consist of any combination of superparamagnetic goethite (structural OH, evolves H2O), lepidocrocite (structural OH, evolves H2O), akaganéite (structural OH and Cl, evolves H2O and Cl-bearing species), schwertmannite (structural OH, SO4, and H2O, evolves SO2 and H2O), hydronium jarosite (structural H3O+, SO4, and OH, evolves SO2 and H2O), ferrihydrite (structural H2O/OH, evolves H2O), iddingsite (structural H2O/OH, evolves H2O), and the Fe3+ pigment similar to that found in palagonitic tephra (structural H2O/OH, evolves H2O). Water evolves from all of these phases below 450°C; hence, all are candidate H2O or OH-bearing phases in Rocknest. There are many other potential candidates that could also contribute to the water release below 450°C; however, the candidates listed above are leading candidates based upon chemistry, lack of crystallinity, and characterization of soils and dust on previous Mars surface missions.

A large number of candidate phases may also contribute to the evolution of H2O at higher temperatures (i.e., 450–850°C). Although the contribution of water from this high temperature region is much smaller than the low temperature region, the two mechanisms for release of water are dehydroxylation of octahedral sites in clay minerals and escape of occluded or trapped H2O in glass or minerals. Candidates that undergo dehydroxylation in this temperature region are smectites (e.g., nontronite from 400 to 600°C montmorillonite from 600 to 800°C, saponite near 700–800°C); chlorites and talc (from 750 to 850°C); serpentines (e.g., antigorite from 600 to 800°C); prehnite; and rock-forming minerals (e.g., epidote, amphiboles).

3.2 CO2

The CO2 released from the Rocknest soils is fairly consistent from run-to-run and can be deconvolved into contributions from five separate peaks (Table 6 and Figure 6). Peak fits were done in Origin Pro version 8.6 using the peak analyzer capability, with all peaks being fit to gaussians. Peak width is constrained based on the width of peaks of single gas releases (O2 for example) and raw data are background subtracted. The fifth peak is present at levels barely above the experimental uncertainty and more likely results from the slow removal of gas from the system, which does not correspond to an ideal gaussian as was modeled. Consequently, we do not discuss sources for the fifth peak.

Table 6. Percent of Total CO2 Released by Peak Number Across All Four Rocknest Runsa
Peak Number% of Total CO2 Release
  1. a

    Uncertainty is the 1σ standard deviation from the average over all four runs. Peak analysis done for m/z 12 data, an ion fragment of CO2.

Peak 118 ± 6
Peak 217 ± 6
Peak 332 ± 7
Peak 428 ± 7
Peak 54 ± 3
Figure 6.

Plot of (left) CO2 released and (right) temperature versus experiment time in Rocknest run 3. Five gaussian peaks are fit to the data with a good match seen in the resulting cumulative peak fit data.

The first peak likely results from the release of adsorbed CO2 from the sample and/or a contribution of combusted organic material that is most likely of terrestrial origin. The second peak evolves at a high enough temperature that the release of adsorbed CO2, which is generally released at temperatures <200°C [Meng-fei et al., 1997; Thompson et al., 2003; Yagi et al., 1997], is unlikely. The combustion of terrestrial organics is a likely source of this peak and could also contribute to others. Seven cups in the SAM instrument contain a mixture of N-methyl-N-(tert-butyldimethylsilyl)-trifluoroacetamide (MTBSTFA) and dimethylformamide (DMF), chemicals that will be used for derivatization experiments in SAM [Mahaffy et al., 2012]. Pyrolysis products of this molecule have been detected in SAM blank runs on Mars and are also seen in Rocknest analyses [Glavin et al., 2013; Leshin et al., 2013]. The concentration of one hydrolysis product of MTBSTFA, 1,3-bis(1,1-dimethylethyl)-1,1,3,3-tetramethyldisiloxane (m/z 147), drops off dramatically with the rise in O2 and peak 2 in the CO2 release, consistent with the combustion of this organic molecule. Laboratory work is currently underway to better quantify the amount of MTBSTFA seen in SAM EGA runs to constrain how much MTBSTFA can contribute to the CO2 release seen. At present, the m/z 147/O2/CO2 data are consistent with a terrestrial organic source for this CO2 peak. However, we cannot exclude a contribution from the combustion of Martian organics to these CO2 peaks.

The third and fourth CO2 peaks represent 60 ± 10% of the total CO2 released and are consistent with two different carbonate phases, or a single phase with a bimodal particle size distribution. These two CO2 releases peak at ~390°C and ~475°C, which is more than 50°C lower than traditional Fe, Mg, or Ca-bearing carbonate decomposition temperatures, even accounting for the reduced pressure under which these samples were run [Archer et al., 2013; Sutter et al., 2012]. However, it is possible that the low peak decomposition temperatures can be explained by particle size effects. The sample particle size can lower the decomposition temperature of a calcium carbonate by almost 200°C when going from millimeter to submicron size particles [Archer et al., 2013]. If the CO2 detected in peaks 3 and 4 is all derived from one or more Fe, Mg, or Ca-bearing carbonate phases, such phases would be present in Rocknest samples at the 1–2% level, right at the CheMin detection threshold. Because CheMin did not detect carbonates in the crystalline component of Rocknest fines [Bish et al., 2013], the CO2-bearing phase is either a crystalline phase present in amounts below CheMin detection limits or makes up part of the X-ray amorphous material, the latter being consistent with the low decomposition temperature.

A likely source of Rocknest carbonates is the Mars global dust. Evidence from Mars Global Surveyor TES (Thermal Emission Spectrometer) data in the thermal infrared suggests that a small amount (2–5 wt.%) of Mg-rich carbonate may be present globally in Martian fines [Bandfield et al., 2003]. The mean particle size of the global dust component has been estimated to be from <10 µm down to <3 µm [Bandfield et al., 2003; Lemmon et al., 2004; Tomasko et al., 1999], and individual grain sizes could be even smaller if the dust particles are aggregates of smaller grains. These small particle size carbonates are consistent with the low decomposition temperatures seen in Rocknest materials though further work is necessary to determine the particle size required for different carbonates to reproduce Rocknest results.

In addition to particle size effects, interactions with other minerals during pyrolysis could also lead to lower carbonate decomposition temperatures. It has been shown previously that the presence of hydrated Mg(ClO4)2 can lead to the early onset of carbonate decomposition [Cannon et al., 2012]. When magnesium perchlorate decomposes to MgO, a significant amount of chlorine is released. This chlorine then reacts with water vapor in the system to form HCl, which can in turn react with carbonate, leading to carbonate decomposition at lower temperatures. However, the current leading candidate for the perchlorate in Rocknest materials is Ca-perchlorate [Glavin et al., 2013], which first decomposes to CaCl2. Calcium chloride, in turn, decomposes at higher temperature (>500°C), releasing Cl gas. Therefore, the decomposition of calcium perchlorate should not lead to the onset of carbonate decomposition below 500°C. However, preliminary lab results show that Ca-perchlorate may, in fact, lead to early carbonate decomposition and is under further investigation.

Therefore, although Ca(ClO4)2 is the current favored perchlorate species at Rocknest, we cannot rule out the presence of other perchlorate species, such as Mg-perchlorate, or the possibility of other reactions during pyrolysis that might lead to the early onset of carbonate decomposition. A possible signature of interaction between materials in Rocknest soils is seen in the data for H2O, CO2, and O2 in run 2. While the rise in H2O abundance seems uniform across the entire temperature profile, the rise in CO2 abundance comes almost entirely from a rise in the third peak and is associated with a rise in O2 released at the same temperature. Based on this O2/CO2 correlation, it seems clear that the release of these gases is correlated to some degree, but the mechanism causing this has not yet been identified.

3.3 O2

The oxygen release peaks around 380–390°C and is consistent with the decomposition of an oxychlorine salt such as a perchlorate. The simultaneous release of chlorinated hydrocarbons supports this hypothesis. This result is discussed in detail in a companion article [Glavin et al., 2013] and will not be reiterated here. We simply note that the O2 release is also important to the sample as a whole because it might interact with other releases such as CO2 (discussed above), SO2 (below), and Fe2+-bearing solids (detected by CheMin [Bish et al., 2013]).

3.4 SO2

The SO2 story is discussed in detail in a companion paper that identifies the potential source of the evolved SO2 as sulfate, sulfite, or sulfide minerals in Rocknest soils [McAdam et al., 2013]. However, an important point regarding SO2 evolution that is mentioned here because it affects O2 abundances is that it is potentially the product of oxidation of a reduced sulfur species in Rocknest samples. If the S is present in an iron sulfide such as pyrite (FeS2) or pyrrhotite (Fe(1 − x)S), then the oxidation during pyrolysis would consume oxygen released from a perchlorate-like source, leading to lower amounts of evolved O2. The possible presence of a reduced sulfur species in the Rocknest sand shadow is supported by the detection of H2S during EGA. However, it is also possible that the H2S is produced by a reaction of H2 (seen in Rocknest EGA) with SO2. If all of the oxygen in the detected SO2 was originally from an oxychlorine species, the O2 abundance would be a factor of ~4 higher (based on SO2 evolution averaged over all four runs). This would result in a Cl wt.% of 0.54 ± 0.18, which would imply that most or all of the Cl in Rocknest soils (0.61 ± 0.02 wt.% based on APXS analysis [Blake et al., 2013]) could be present in the form of perchlorate.

3.5 Less Abundant Volatiles

In addition to the species discussed above, SAM detected releases of H2S, HCl, NO, HCN, and NH3 during EGA (Table 7 and Figure 7). The H2S and HCl detected are almost certainly of Martian origin because they are present at levels about 10× background and no discrete release was seen in a SAM blank run before ingesting Rocknest samples. The NO, HCN, and NH3, on the other hand, might be products of terrestrial background in the SAM instrument, and work is under way to compare the quantities of these compounds detected with what is known to be present in the instrument. Both MTBSTFA and DMF, the derivatization agents seen in SAM blank runs, contain N and C and could be the source of the NO (the oxygen is likely Martian), HCN, and NH3. Each of these species is discussed in more detail below.

Table 7. Molar Abundances (in nmol) for Minor Gases Released From Rocknest Materialsa
 Molar Abundance (nmol)
 Run 1Run 2Run 3Run 4
  1. a

    Reported error is the 2σ standard deviation based of the counts/mol ratio for m/z 44. Integrated counts for these minor species are converted to counts of m/z 44 to calculate abundances. Overlapping signals at a given m/z make it difficult to compute multiple integrated count values for each species, as was done with the major gases released. Weight % values for these species are not included because they could be derived from a number of different possible species in the soil.

HCl5 ± 217 ± 745 ± 2080 ± 36
H2S41 ± 1857 ± 2580 ± 3566 ± 29
NH3670 ± 540860 ± 630780 ± 5501050 ± 740
NH3 (blank run)650 ± 400
NO173 ± 77250 ± 111190 ± 85177 ± 79
HCN39 ± 1757 ± 2549 ± 2243 ± 19
Figure 7.

EGA curves of the minor species released from the third SAM analysis of Rocknest soils. Data are shown for m/z 15 (potential NH3 fragment) in both the Rocknest run and the pre-Rocknest blank, m/z 27 (HCN), m/z 30 (NO), m/z 34 (H2S), and m/z 36 (HCl).

3.5.1 H2S

An interesting feature of the Rocknest samples is the observation of both oxidized and reduced sulfur species in the evolved gases. This topic is discussed in greater detail in a companion paper [McAdam et al., 2013]. The amount of H2S released ranged from ~40 to 80 nmol. The largest release of H2S was observed in run #3, which also had the highest ratio of SO2/H2S. As stated previously, the run-to-run variability of the sulfur-bearing phases is not yet understood.

3.5.2 HCl

All four Rocknest runs produced HCl, primarily observed in the EGA traces above 300°C (Figure 7). The ultimate source of the HCl is likely the oxychlorine phase (e.g., perchlorate) in Rocknest materials. The release of HCl after O2 released from perchlorate decomposition is typical of Ca-perchlorates and is one of the reasons Ca2+ is the preferred cation associated with the perchlorate anion. As stated above, Ca-perchlorate decomposes to CaCl2, not CaO. Mg(ClO4)2, on the other hand, decomposes to MgO and the Cl is released at the same time as the O2. X-ray diffraction studies of perchlorate decomposition have shown that CaCl2 can subsequently react with water to form exotic Ca–Cl–O phases, releasing HCl in the process [Cannon et al., 2012].

The amount of HCl detected in Rocknest runs increased steadily from ~5 to 80 nmol from runs 1 to 4. This increases in HCl abundance is not correlated with an increase (or decrease) of any other species yet identified and is not understood. It is possible that HCl was building up over time in SAM with each successive run adding to the HCl background. However, a subsequent blank run had no HCl signal in this temperature range so this explanation might not hold; additional laboratory experiments are required to explore the possible involvement of other Cl-bearing phases.

3.5.3 Nitrogen-bearing Species

Several masses that may be associated with nitrogen-bearing compounds were identified in Rocknest EGA data, with the most likely species being NH3 (m/z 14 and 15), NO (m/z 30), and HCN (m/z 27). Discovery of indigenous nitrogen-bearing compounds in either oxidized or reduced form would be significant, as solid nitrogen compounds have yet to be found on Mars and are necessary for life as we know it on Earth. However, this detection is preliminary for a few reasons. First, all of these m/z values have interferences with other possible compounds. For NH3, we can only use m/z 14 (N) and 15 (NH) because 16 and 17 have interferences with OH (17) and O (16), which are present at much higher levels as fragments of more abundant species. While it is possible that the 14 and 15 detected in EGA data are from NH3, it is equally possible that they are from methane fragments (CH2 and CH3), which have a similar 14/15 fragmentation pattern. The signal at m/z 30 could be NO but it could also be an organic fragment such as CH2O and we cannot differentiate between these two species with the EGA data alone. Similarly, the signal at m/z 27 could be HCN or it could be another organic fragment. Furthermore, even if they are present, the HCN and other potential N-bearing species could not be Martian at all and could simply be from terrestrial derivatization agents MTBSTFA and DMF, both nitrogen-containing compounds, known to be present in SAM. With these important caveats stated, we will discuss possible sources for N-bearing compounds potentially present.

Ammonium (NH4+) compounds have been proposed as a sink for nitrogen on the Martian surface [Mancinelli and Banin, 2003]. Based on the assumption that the mass 14 and 15 signal is from NH3, we calculate that ~650 nmol of NH3 were released in the pre-Rocknest blank run, likely from the terrestrial background in the SAM instrument. When comparing the m/z 14 and 15 evolution pattern from the blank run to Rocknest runs (Figure 7), the release pattern is different but the overall abundance is the same within uncertainty (Table 7). We believe that the different release pattern is caused by interactions between the terrestrial background materials and gases released (O2 for example) from Rocknest soil.

Another nitrogen-bearing compound, NO, has been proposed as the source of the m/z 30 signal [Navarro-González et al., 2013], which shows a broad release across temperatures ranging from ~150 to 650°C, although other possible compounds such as formaldehyde may also contribute, as previously mentioned. No detection of m/z 30 was observed in the blank run conducted prior to the analyses of the Rocknest samples. There are three possible sources for the observed NO release: (1) oxidation of terrestrial nitrogenated organics known to be present in SAM such as MTBSTFA and DMF, (2) decomposition of nitrates in the presence of perchlorates, and (3) oxidation of ammonium salts by perchlorates.

One explanation for the release of NO at temperatures below 500°C, based on work with terrestrial soils, could be the oxidation of nitrogenated organics [Navarro-González et al., 2009]. At present, there is no clear evidence of Martian organics in Rocknest materials [Glavin et al., 2013; Leshin et al., 2013]. However, a fraction of this NO could originate from the known terrestrial sources of nitrogenated organics such as MTBSTFA and DMF. Laboratory experiments have shown that when MTBSTFA is heated in the presence of Ca-perchlorate, NO and N2O are released [Navarro-González et al., 2013]. Current estimates of the amount of MTBSTFA that could be present in each pyrolysis run range from 4 to 300 nmol based on estimates from GCMS and EGA data [Glavin et al., 2013]. Given the observed abundance of 170–250 nmol of NO released during Rocknest pyrolysis runs, it is possible that all detected NO is from instrument background.

Nitrates are another potential source for the NO released during thermal analysis of Rocknest samples. Although nitrates generally decompose into NO and O2 at temperatures between 560 to 900°C [Stern, 1972], if nitrates are mixed with perchlorates, thermal decomposition can occur at significantly lower temperatures, which could account for the m/z 30 released from the Rocknest materials (Figure 7) [Navarro-González et al., 2013].

A third explanation for an NO signal is the oxidation of ammonium salts when heated in the presence of perchlorates [Navarro-González et al., 2013]. As stated previously, there is evidence for NH3 in pre-Rocknest blank runs but the release pattern changes during Rocknest sample analysis. If NH3 is present at the upper limits of our reported uncertainty in the blank run, there is enough background N in the system to account for the observed NO release, which would be from the reaction of NH3 (background) and O2 (from Martian perchlorate decomposition).

A contribution from HCN (m/z 27) is also present in the EGA data as a broad release from ~130° to 300° C (Figure 7) and is not seen above background levels in the blank run. While it is possible that this peak is a fragment from another compound, the m/z 27/26 ratio for Rocknest runs is ~6.5 ± 1.0 (1σ standard deviation) and the literature value for the 27/26 ratio from HCN is 6 (from the NIST11 NIST mass spectra library). This close match makes HCN a likely candidate for the m/z 27 peak seen in Rocknest data with an abundance of about 50 nmol over all four runs (Table 7). However, this low temperature evolution of HCN may be associated with pyrolysis of the derivatization reagents MTBSTFA and DMF in the presence of perchlorate [Glavin et al., 2013; Stern et al., 2013] and preliminary lab work using commercial EGA-MS instrumentation to pyrolyze MTBSTFA and DMF shows evolution of m/z 26, 27, at low temperatures [Glavin et al., 2013; Stern et al., 2013].

3.6 Run-to-Run Variability

By plotting the abundance of a particular species for a particular run divided by the average abundance for that species over all four Rocknest runs, we see that the variation in abundances for H2O, CO2, and O2 are the same within experimental uncertainty (Figure 8). This means that the run-to-run variation for these gases can either be explained by the uncertainty in abundance calculations or a spread in the amount of sample delivered to SAM for each run. The SO2 release, however, does not follow this pattern, does not seem to track with any other gas released, and is unexplained.

Figure 8.

Plot of the run-to-run variability of (top) four major and (bottom) five minor species evolved from Rocknest soils. The abundance for a given species from a particular run is divided by the average for that species over all four runs. Error bars include the 2σ uncertainty in both the molar abundance and amount of sample delivered to SAM.

Plotting the minor species in the same way (Figure 8, bottom) shows that, for the most part, the run-to-run variability is within experimental uncertainty. The HCl signal, on the other hand, varies at a statistically significant level, increases steadily with each run. Looking at the individual gases in more detail, we see that the variability in H2S abundance is similar to that of SO2 (Figure 8). NH3 abundance does not seem to track with any other gas released but is fairly consistent over the four runs. NO and HCN track together and closely follow the H2O, O2, and CO2 release pattern, supporting the idea that NO and HCN are produced by reactions between terrestrial and/or Martian C/N sources that are reacting with other materials in the sample. Overall, the run-to-run variability of most of the gases released is small, demonstrating the good repeatability of the SAM Rocknest measurements.

4 Implications

These results represent the first analysis with new instrument tools of a Martian soil that contains a truly global component (dust) as well as a component similar to other soils analyzed on Mars that may be representative of the bulk crust [Blake et al., 2013; Taylor and McLennan, 2009]. Because of the compositional similarity of Rocknest materials to soils analyzed on Mars by previous missions, and the fact that these soils are exposed to the same alteration processes, SAM and CheMin analyses of Rocknest materials have implications for soils on Mars in general.

4.1 Water

The results from the ChemMin analysis of the Rocknest materials show no hydrated crystalline phases indicating that the phases hosting water are either below the detection limit or are X-ray amorphous [Bish et al., 2013]. However, the evolved gas analysis provided by the SAM instrument provides detection of several phases that might be the sources of the water in the Rocknest materials.

First, the CheMin instrument has detected a significant amount of amorphous material in Rocknest soils [Bish et al., 2013]. The detection of 1–3 wt.% water by the SAM instrument, most of which is likely to be associated with the amorphous component, strongly suggests that there is a component of the Rocknest soil that formed via low temperature aqueous weathering. The broad temperature range of the water release allows for the possibility of contributions from sulfates, perchlorates, and/or carbonates, in addition to water associated with the amorphous component, representing a more heterogeneous combination of processes. This is not unexpected for a Rocknest-like soil that is a physical mixture of global dust and other aeolian materials sourced over a wide area. Therefore, the broad release profile that we observe might suggest that the weathering and hydration did not occur in situ and could represent weathered Martian materials from a variety of different times and places.

4.2 Carbonates in Rocknest Materials

The higher temperature (peaks 3 and 4) CO2 evolved during Rocknest suggests the presence of carbonates in Martian soils. If we assume that Martian global dust contains 5 wt.% Mg-rich carbonates [Bandfield et al., 2003] and we assume that peaks 3 and 4 in the SAM CO2 EGA data are from carbonate decomposition, the global dust would need to make up ~55% of the Rocknest sample to explain all of the carbonate detected. Using an estimate of 2 wt.% carbonates, the lower limit from the Bandfield et al. work increases this number. Given the makeup of Rocknest soils, it is unlikely that it is composed of such a high component of global dust [Blake et al., 2013]. Therefore, if the global dust component of the <150 µm portion of Rocknest material is <55%, the remaining aeolian material must also contain carbonate as well.

Rocknest carbonate could be the result of active carbonate formation on present-day Mars [e.g., Garenne et al., 2013], or could be from local or possibly distant ancient sources, a question that might be addressed by constraining the particle size of carbonates in Rocknest samples by matching lab data of carbonates of different particle sizes to Rocknest EGA results. In addition to carbonates in the global dust and in Rocknest samples, a calcium or magnesium carbonate was detected in the soils at the Phoenix landing site in the Martian northern plains at abundances of 3–5 wt.% [Boynton et al., 2009; Sutter et al., 2012]. The detection of similar amounts of carbonates in soils at two disparate locations on Mars suggests that whatever the carbonate formation or distribution mechanism is, it is global in nature.

4.3 Minor Species

4.3.1 H2S and HCl

The detection of H2S in Rocknest soils is important because it reveals the presence of a reducing agent in the soil. A reduced S or H-bearing species in a soil with perchlorates and sulfates is interesting from a redox perspective and also has implications for habitability (from a chemical energy perspective). The high temperature evolution of HCl from Rocknest samples is indicative of the presence of perchlorates or chloride salts in Rocknest materials. The detection of perchlorates in Rocknest soils has important implications for the global distribution of perchlorates on Mars. Perchlorates can complicate the detection of organic molecules in Martian samples using pyrolysis extraction techniques [Glavin et al., 2013] and have implications regarding habitability [Coates and Achenbach, 2004] and the possibility of liquid brines on Mars [Chevrier et al., 2009; McEwen et al., 2011].

4.3.2 Nitrogen-bearing Species

If determined to be Martian in origin, the nitrogen-bearing species detected by SAM are the first in situ detection of nonatmospheric N on Mars. The only evidence of Mars surface nitrogen before the MSL mission has been the possible detection of nitrates in two Mars meteorites [Grady et al., 1995]. Orbital near infrared spectroscopy has not detected surface nitrates. The Phoenix Lander's Wet Chemistry Laboratory (WCL) was equipped with a nitrate ion selective electrode but the presence of soil perchlorate interfered with detection of any nitrate [Hecht et al., 2009]. Additionally, the Phoenix TEGA mass spectrometer did not detect any simultaneously evolved NO and O2 that would have indicated the thermal decomposition of nitrate in the Phoenix soils.

These detections have the potential to enhance the understanding of the Martian nitrogen cycle. Only the atmospheric concentration (~0.15–0.2 mbar N2) has been known for Mars [Mahaffy et al., 2013; Owen et al., 1977]. The detection of Rocknest nitrogen can provide valuable contributions to understanding the total Martian N inventory. Such information can be used test the validity of various nitrogen cycle models that estimate nitrogen concentrations [e.g., Manning et al., 2008].

Finally, up to this point, all major elements required for life (C, S, O, P, H, and N) have been detected in biologically available materials except for N. Because laboratory experiments have shown that the Martian atmospheric N2 level is too low for biological fixation [Klingler et al., 1989], the detection of nitrogen-bearing species in Rocknest EGA data would have profound implications for habitability if the detection from indigenous martial surface material is confirmed.

5 Summary and Conclusions

The evolved gas analysis of the Rocknest aeolian deposit reveals a complex mixture of volatile-bearing phases with important implications for Mars. They contain 1–3 wt.% water and slightly lower amounts of carbonate, perchlorate, and sulfur-bearing species. Rocknest samples contain a mixture of both oxidized (perchlorates) and reduced (sulfides or H-bearing) species as well as minerals formed under alkaline (carbonates) and acidic (sulfates) conditions. The comingling of these minerals implies a physical mixture in chemical disequilibrium, revealing some of the history of Rocknest materials: namely, that the minerals in Rocknest soils are from different sources that have not encountered sufficient water and/or temperatures necessary to reach chemical equilibrium. This soil is consistent with being a physical mixture by aeolian processes on a relatively cold and dry Mars.

The oxygen release, consistent with the presence of an oxychlorine species such as perchlorate [Glavin et al., 2013] and at similar abundances to what was detected by the WCL instrument on the Phoenix lander [Hecht et al., 2009], implies that perchlorate formation may be a global process on Mars and that perchlorates might be present at similar levels in all Mars soils. The minor species detected on Mars and quantified in this paper give a glimpse of other important chemical species present in Martian soils, and further work is underway to confirm if they are indeed Martian in origin.


NASA provided support for the development of SAM. Data from these SAM experiments will be archived in the Planetary Data System (pds.nasa.gov) in 2013. Essential contributions to the successful operation of SAM on Mars and the acquisition of this data were provided by the SAM development, operations, and test bed teams. P.D.A. acknowledges support from the NASA Postdoctoral Program, administered by Oak Ridge Associated Universities through a contract with NASA. D.P.G. acknowledges funding support from the NASA ROSES MSL Participating Scientist Program.