Benzoic Acid as the Preferred Precursor for the Chlorobenzene Detected on Mars: Insights from the Unique Cumberland Analog Investigation

Each of the individual phases used in the analog were crushed using a mortar and pestle and sieved to <150 μm for XRD analyses on a Bruker D8 Discover X-ray diffractometer (Cu Ka radiation, l = 1.54059 Å) to check for trace minerals. Data were collected from 2 to 70 °2theta. The data were interpreted using the International Centre for Diffraction Data, Powder Diffraction File-4 (ICDD PDF-4) Minerals database and the Materials Data, Inc. (MDI) JADE software. The palagonite used in the CBA is a natural sample collected from Maunakea, on the Big Island of Hawaii, USA (HWMK101, R. Morris Collection; Morris et al. 2001); this sample contains significant X-ray amorphous material and also plagioclase, which is normal for a natural palagonite sample. The plagioclase component used in the CBA is a labradorite from Labrador, Canada, with very minor amounts of muscovite, quartz, and carbonate. The clay mineral used in the analog is a Fe-saponite from Griffith Park in Los Angeles, California, USA (Treiman et al. 2014). This particular saponite was chosen because CheMin analyses of the CB drill sample indicated that the smectite was a Fe-rich trioctahedral clay mineral, like a Fe-saponite. XRD analyses of this sample indicated that is contains minor quartz. The pyroxene phases of augite and pigeonite identified by CheMin analyses of the CB were grouped together and a natural sample of augite from Fone, Norway, was used to represent pyroxene in the CBA. The augite was not pure as it contained approximately 10 wt. % of quartz. The magnetite was also a natural sample, from Franklin, New Jersey, USA, and contained very minor (<3 wt. %) quartz. The enstatite was a natural sample purchased from Ward's Science and contained a very minor amount (<1 wt. %) of phyllosilicate. A laboratory synthesized akaganeite (DC11202008-11B-SC, R. Morris Collection) was also chosen for the analog to ensure a known composition. The microcline is a natural sample, again from Ward's Science and is dominated by the K-feldspar; however, it does contain <50 wt. % of a Na-rich feldspar, which is common for natural samples. The CheMin analyses of the CB drill sample indicated that both anhydrite and bassanite were present in the CB sample; the bassanite in the analog was substituted with anhydrite since bassanite dehydrates at low temperatures during SAM EGA and GCMS analyses and should not affect reactions occurring at higher temperatures in the SAM oven. The anhydrite was purchased from Sigma-Aldrich (Lot #MKBW9429V) and it contains minor (<2 wt. %) gypsum. The pyrrhotite is a pure, natural sample from Santa Eulaia, Mexico (SAEUMX01, R. Morris Collection). The olivine used for the CBA is a forsterite from San Carlos, Arizona, USA, and contained minor (<5 wt. %) enstatite. The hematite is a synthetic sample from Sigma Aldrich (Batch #0712BE) and contained minor (<5 wt. %) magnetite. The ilmenite was another natural sample from Ward's Science, from St. Urbain, Quebec, Canada, and contained minor amounts (<5 wt. % each) of pyrite and hematite. The halite used in the CBA was pure and was also a Ward's Science sample, from Carlsbad, New Mexico, USA. The last phase present in the CB drill sample as described in Vaniman et al. (2014) is quartz, present at approximately 0.1 wt. %. Since many of the other phases used to create the CBA already contained minor amounts of quartz, no additional quartz was added to the sample (Table 1).

The natural samples used in the CBA (e.g., labradorite, enstatite, and augite) were also characterized with an EGA to assess the presence of trace volatile evolving phases present below XRD detection limits. A SAM-like EGA of selected individual CBA components was performed with either a Hiden Analytical HPR-20 quadrupole mass spectrometer (QMS) coupled upstream to a custom-built manifold and pyrolysis oven or on a Setaram Labsys Evo Thermogravimeter/Differential Scanning Calorimeter coupled to a Pfieffer OmniStar QMS. For samples analyzed on the Hiden system, evolved gases were monitored as samples were heated at 20°C minute⁻¹ from ∼50°C to ∼1000°C, under SAM-like He pressures (∼30 mbar) and gas flow conditions (∼1 mL minute⁻¹ ). The manifold lines between the oven and the mass spectrometer (MS) were heated to approximately 100°C to mitigate for organics condensing in the lines before reaching the MS and to reduce water background in the lines. Target mass-to-charge (m/z) values focused on were 45 to represent CO₂ (the m/z for the ¹³CO₂ isotopolog because the main m/z 44 for CO₂ is often saturated), 20 to represent H₂O (the m/z for the H₂¹⁸O isotopolog because the main m/z at 18 for H₂O is often saturated), and 66 to represent SO₂ (the m/z for the ³⁴SO₂ isotopolog, in the case that the main m/z at 64 for SO₂ is saturated). For samples analyzed on the Seteram system, evolved gases were monitored as samples were heated at a SAM-like ramp rate (modeled after temperature profiles from flight experiments) from 60°C to 850°C, under SAM-like He gas pressure of approximately 25 mbar in the pyrolysis oven, and with a He carrier gas flow rate of 5 mL minute⁻¹ modified for this system. Mass-to-charge values up to m/z 200 were monitored. Due to our strategy of choosing the purest minerals out of multiple samples available for making the analog, no phases in addition to those identified by the XRD analyses were identified by EGA in the minerals used in the CBA.

All individual components of the mixture were characterized by pyrolysis GCMS to assess qualitatively the background organics present in the mineral matrices. Samples were loaded in precombusted quartz tubes and analyzed in a pyroprobe 5200 (CDS Analytical Inc.) coupled to a Trace gas chromatograph (GC), ISQ TM QMS (ThermoFisher). A first pyrolysis step to 500°C allowed us to characterize semi-volatile adsorbed organics and was followed by a control pyrolysis GCMS run at the same pyrolysis temperature. The second heating step to 950°C (followed by a control, similar to the first step) allowed for the detection and characterization of refractory adsorbed organic compounds as well as organics released during the breakdown of a mineral structure. GCMS separation was performed with an Agilent DB-5ms column (length of 30 m, internal diameter of 0.25 mm, and film thickness of 0.25 μm) heated at 5°C minute⁻¹ from 30°C to 200°C and held for 30 minutes at this temperature, with a helium carrier gas flowing at 1 mL minute⁻¹ . Detection was carried out with MS in full scan mode with a mass range of 5–555 u. A 10 minutes cleanup run at 300°C was performed between each analysis and blank.

The results show that most of the minerals are free of organics or contain only minor traces of them, both quantitatively and qualitatively. The labradorite, augite, and magnetite contain traces of aromatic compounds released at both low and high temperatures and long-chain hydrocarbons at low temperatures. Benzene, styrene, and naphthalene are the most abundant traces. Akaganeite is the most contaminated mineral, with the presence of aromatic compounds including some benzoic and phthalic acids, as well as chlorobenzonitrile at low temperatures. At high temperatures, the Cl released from the decomposition of the mineral chlorinates the organic background to form a low abundance of chlorobenzenes. Overall, although some minerals contain traces of organic compounds, they are considered clean in the sense that the contamination is present at traces levels only, is well characterized, and is not known to interfere with the further experiments and interpretations.

Each of the 14 individual phases were weighed out and mixed using a mortar and pestle to create a homogeneous analog sample of 500 mg (Figure 2). That mixture was weighed out into 12 portions of approximately 40 mg each and distributed to members of the SAM, CheMin, and ChemCam teams for analyses.

Zoom In Zoom Out Reset image size

SAM-like EGA experiments were performed on an Agilent 5975T LTM-GC/MSD coupled to a Frontier Lab 3030D pyrolyzer and an AS-1020E auto-shot sampler. The samples (Table 2) were weighed out in precombusted Frontier Lab stainless steel Eco-Cups, which were then dropped into the pyrolysis oven by the autosampler in a programmed sequence of EGA runs. In each sample run, the oven was held at 75°C for 25 minutes, then ramped at 35°C minute⁻¹ to 1050°C, and then held for a further six minutes. The sample cups sat in a quartz liner within the pyrolysis oven during heating; a new liner was used for this work to ensure that chlorine contamination from previous runs was not present. Pyrolysis was conducted under 30 mbar helium with a 100:1 split directed to the mass spectrometer via a transfer line held at 135°C. The mass spectrometer scanned a 2–200 u range.

VNIR reflectance spectra were measured using the Spectrophotometer with cHanging Angles for the Detection Of Weak Signals (SHADOWS) spectro-gonio-radiometer available at the Laboratoire de Planétologie de Grenoble (IPAG, France); details of the instrumentation can be found in Potin et al. (2018). Bidirectional reflectance spectra are measured with a photometric accuracy below 1% in the 0.5–3.0 μm range with a sampling interval of 10 nm. These spectra can be compared to the VNIR spectra normalized obtained by Mastcam and ChemCam on board MSL. LIBS analyses were also attempted. However, due to the small mass available for the sample, difficulties arose in focusing the laser to the sample at the bottom of the sample holder, and the plasma shock wave generated by the LIBS ablation under Mars atmosphere quickly ejected the sample from the holder. Unfortunately, the measurements did not lead to exploitable results.

Laboratory EGA. The EGA of the unspiked CBA produced a minor benzene peak at 380°C, which was accompanied by small organic fragments (m/z 27, 39, 41, and 43) from the known contamination, and the onset of broad CO₂ and H₂O releases, which peaked at 450°C. The traces for H₂O, CO₂, and benzene are shown in Figure 4(a). A small CO₂ peak was seen at 615°C, which likely originated from decomposition of the carbonate impurity known to be present in the labradorite sample used in making the analog (Table 1). As the organic contamination produced both aliphatic and aromatic components during pyrolysis, it was important to determine if these contaminants could react with magnesium perchlorate to generate detectable levels of chlorobenzene.

Zoom In Zoom Out Reset image size

The addition of 2 wt. % magnesium perchlorate to the CBA increased the oxidation of benzene, resulting in a significantly weakened benzene peak and a low-temperature CO₂ peak developing at 370°C (Figure 4(b)). The CO₂ peak at 615°C became more prominent due to the lower-temperature release of the CO₂ from the organic contaminants in the presence of the perchlorate. Crucially, no chlorobenzene (m/z 112) was detected during the EGA run.

A subsample of the benzoic acid standard was pyrolyzed to confirm that m/z 105 was the major fragment during the SAM-like EGA. When benzoic acid was mixed with the CBA and Mg-perchlorate (CBA-BA-MgP-2), an m/z 105 peak was detected at 435°C, which was accompanied by benzene and chlorobenzene, and a sharp CO₂ peak developed at 450°C (Figure 4(c)). The simultaneous release of benzoic acid and chlorobenzene suggests that Cl₂ initiated partial benzoic acid chlorination. HCl (not shown) began to evolve at the same temperature as benzoic acid, benzene, and chlorobenzene. The benzene release continued to high temperatures and increased after perchlorate decomposition was complete, producing a second peak at 575°C. The extended benzene release was associated with a broad CO₂ shoulder and was suggestive of decarboxylation of the remaining benzoic acid. An additional CO₂ peak occurred at 485°C and was associated with the O₂ release from the perchlorate, indicating oxidation.

Because the CBA contains akaganeite and halite, benzoic acid was added to the analog and pyrolyzed without perchlorate to determine if chlorobenzene could be generated by a reaction between benzoic acid and chlorides. The CBA + benzoic-acid mixture (CBA-BA) produced a small peak for m/z 112 at 500°C, in contrast to the strong m/z 112 peak evolved at 435°C by CBA-BA-MgP-2 (Figure 4(d)). The CBA-BA-MgP-2 chlorobenzene peak exhibited a prominent shoulder that occurred at the same temperature as the minor chlorobenzene formed by chlorides and benzoic acid in the CBA-BA sample. Mg-perchlorate forms an MgO residue during pyrolysis, but in some studies, MgCl₂ has also been observed (Acheson & Jacobs 1970). The presence of MgCl₂, in addition to akaganeite and halite, could explain the significant chlorobenzene shoulder seen in CBA-BA-MgP-2, while the reaction with perchlorate is responsible for the major peak at 435°C.

TB EGA and comparison with Mars data. EGA trends for several major volatiles from CBA-BA-MgP-1 and CBA-BA-MgP-2 were similar to each other and broadly similar to the EGA data from the run of the CB drill sample in SAM on Mars (Figure 5). CB drill sample EGA traces were extensively discussed ahead of time (Freissinet et al. 2015; Sutter et al. 2017). CBA-BA-MgP-1, CBA-BA-MgP-2 and CB all show a broad evolution of H₂O less than ∼500°C and a subtle high temperature evolution near ∼700°C–750°C. They also all exhibit a relatively sharp CO₂ peak occurring at temperatures between 200°C and 400°C, but CBA-BA-MgP-1 and CBA-BA-MgP-2 show those peaks followed by a lesser CO₂ evolution at higher temperatures. In CBA-BA-MgP-2 there is a significant CO₂ evolution centered around 500°C, overlapping with benzene evolution at similar temperatures. Both CBA-BA-MgP-1 and CBA-BA-MgP-2 developed m/z 78 and (low intensity) m/z 77 peaks near 500°C, attributed to benzene evolution. CB evolved a substantial O₂ peak, attributed to the thermal decomposition of Mg-perchlorate in that sample (black trace, Figure 5(a)). Though CBA-BA-MgP-1 and CBA-BA-MgP-2 had perchlorate added, no O₂ evolution was observed. The control sample made by doping inert and organically clean fused silica with Mg-perchlorate (Control-MgP) did show an O₂ evolution, however. This sample also showed very small amounts of CO₂ (possibly from oxidation of trace organic compounds by the large amount of evolved O₂) and H₂O (from dehydration of the perchlorate salt).

Zoom In Zoom Out Reset image size

The H₂O evolved below ∼500°C, from CB and CBA samples, derives from the amorphous material, dehydroxylation of akaganeite, dehydration of perchlorate salt, and any interlayer H₂O present in the samples' Fe-saponite, though both in SAM and the TB, the interlayer of the smectite clay minerals should have been largely dehydrated. The higher temperature H₂O evolution results from the dehydroxylation of the Fe-saponite in the CB and CBA samples. The sharp CO₂ peak <400°C in CB and CBA samples may result from a combination of oxidation of organic materials by evolved O₂ from perchlorate and perhaps some oxygen from evolved H₂O, in addition to CO₂ from the oxidation of the benzoic acid in the case of CBA. In both the CB samples and the CBA samples, the CO₂ is evolved at approximately the same temperature as O₂ evolution starts to rise (in the case of CB) or would be expected to rise (in the case of CBA-BA-MgP-1 and CBA-BA-MgP-2). In the CBA samples, O₂ would be expected to evolve at the same temperature as it evolves from the Control-MgP samples, though no distinct peak of O₂ was detected. We think no O₂ was detected from the CBA samples because it was all consumed by reaction with organics such as the benzoic acid and trace organics present in the CBA's component minerals to produce CO₂ before O₂ could be detected by the MS. In both the CB sample and CBA, it is likely that this oxidation of the organic compounds in the sample by O₂ was incomplete since, in both cases, enough remained to form the chlorobenzene and dichlorobenzene detected by the GC. Also, in the case of the CBA-BA-MgP samples, some benzene was observed at higher temperatures and presumably was a decarboxylation product of benzoic acid. Carboxylic acids have been shown to decarboxylate at high temperatures when adsorbed or embedded into a mineral matrix (Glavin & Bada 1998). We demonstrated in the laboratory that phthalic acid undergoes a first decarboxylation to benzoic acid at 520°C and a second decarboxylation to benzene at 560°C when heated with a SAM-like ramp in presence of montmorillonite. A similar decarboxylation of phthalic acid was observed in presence of hematite (Francois et al. 2016). Thus, the higher temperature CO₂ near 500°C coincident with the benzene evolution can either result from thermal decarboxylation of the benzoic acid, ongoing oxidation of the benzoic acid by the tail end of the O₂ peak, and/or the H₂O that continued to evolve at those temperatures. The non-detection of benzoic acid and chlorobenzene in the TB EGA runs, as opposed to its detection in laboratory EGA, may be explained by the lower sensitivity of the TB detector compared to a commercial MS. Additional occurrence may happen in the TB, such as a lower transport rate of heavy molecules throughout the lines.

We compared the CBA with a Rocknest (RN) analog and with the CB and RN measurements made by MSL. The RN windblown deposit is the best-characterized Martian soil to date, with precise mineralogical measurements done by MSL (Blake et al. 2012; Bish et al. 2013; Achilles et al. 2017). The similarity with soils analyzed at Gusev Crater and Meridiani Planum could imply globally similar basaltic materials deposited locally. The mineralogy of RN (26 wt. % plagioclase, 20 wt. % pyroxene, 13 wt. % olivine, 2 wt. % magnetite, and 1 wt. % hematite; Bish et al. 2013; Achilles et al. 2017) was used to generate the RN analog: the Mars Global Simulant-1 (MGS-1; Cannon et al. 2019). The 35 wt. % amorphous phase is simulated by 22.9 wt. % of basaltic glass, 5.0 wt % of hydrated silica, 4.0 wt. % of Mg-sulfates, 1.7 wt. % ferrihydrite, and 1.4 wt. % Fe-carbonate (Cannon et al. 2019). The simulant grain size is <1 mm and its spectral properties have been favorably compared to both rover and orbiter observations of Mars soils.

Figure 6 presents the VNIR spectra obtained for the two analogs compared to Mastcam's observations of RN scoop 4 and ChemCam's observations of RN scoop 2 Kenyon soils (sol 97) and the observations of the CB drill hole tailings (Johnson et al. 2015; Wellington et al. 2017). The three types of measurements follow the same trends. The CBA exhibits greater reflectance >600 nm than the RN analog, which is in agreement with Mastcam and ChemCam measurements. The CBA also presents a less prominent ferric absorption near 535 nm than the RN analog, which may be due in part to the different materials used to represent their amorphous components. The ChemCam passive spectra (Johnson et al. 2015) of the Martian targets also confirm this assessment, while presenting variations in overall albedo likely resulting from a combination of dust cover, surface texture, and photometric effects.

Zoom In Zoom Out Reset image size

The greater band depth at 1.9 μm in the CBA compared to the RN analog is peculiar, since CB and RN are supposed to have similar amounts of H₂O based on SAM measurements (∼2 wt. %; Sutter et al. 2017). Note that the water content of the CBA was not constrained, for practical reasons. While we took care to keep the samples under low-humidity conditions, the depth of the 1.9 μm band may indicate that some mineral phases present in the CBA sample (e.g., smectite and palagonite) are more hydrated than their Martian counterparts. This could also be related to variations in grain size of the hydrated phases in the mixtures (Cooper & Mustard 1999).

The bands around 1.4 and 1.9 μm in the CBA spectrum are mainly due to the presence of the ferrian saponite griffithite used in the mixture (Figure 6). The metal-OH bands associated with smectites around 2.3 μm are not obvious but do exist in the spectra, also indicating the presence of the ferrian saponite in the mixture (Treiman et al. 2014). This is a reminder that clay minerals within complex mixtures may not be easy to detect from orbit, even at the ∼20 wt. % level and without dust cover.