Mineral Identification and Preliminary Mapping of Organic Compounds in Serpentinite-Related Lithologies Using μFTIR

Serpentinization, the water-rock reaction forming serpentine mineral assemblages from olivine and pyroxene-bearing protoliths, can co-occur with the production of hydrogen, methane, and diverse organic compounds. Serpentinization is regarded as a geologic process important to the sustainability of the deep biosphere and the origin of life. Serpentinization has been found to support metabolisms involving anaerobic COoxidation, acetogenesis, reduction of sulfur compounds, H2 oxidation, CO2 fixation, magnetite reduction, and fermentation. There is on-going research to develop a method that can visually depict mineral associations in serpentinites with serpentinization-related organics. In this report, we describe new findings, using μFTIR to map minerals and associated organics related to serpentinization. To do this, we identified, confirmed, and documented FTIR wavenumber regions linked to (I) serpentinization-associated minerals and embedded organics, (II) carbonate-associated minerals, and (III) a polysaccharide proxy for EPS. The findings of this study indicate that organic presence with a mineral background can be detected using reflection mode-μFTIR (R-FTIR) and transmission mode-μFTIR (T-FTIR). Implications of this study include increased prospects for use of FTIR in interrogating polished rock surfaces to determine the association of organics to mineral phases and boundaries in serpentinization-related lithologies.

. Abiotic processes, such as the water-rock reaction of serpentinization, fuel and sustain microbial communities in sediments and rocks exposed at the seafloor and in the subsurface of the ocean crust (Kelley 2005). Other such abiotic drivers of life in the deep subsurface include radiolytic H 2 production (Blair et al. 2007) and hydrocarbon 'cracking' (Horsfield et al. 2006). Diverse chemosynthetic metabolisms are employed by the deep biosphere, taking advantage of these abiotic processes, including H 2 oxidation, SO 4 2reduction, CH 4 oxidation, methanogenesis, O 2 reduction, organic carbon degradation, and Mn reduction (Orcutt et al. 2013).
Serpentinization, the water-rock reaction forming serpentine mineral assemblages from olivine and pyroxene-bearing protoliths, can co-occur with the production of hydrogen, methane, and diverse organic compounds (McCollom and Seewald 2013).
Natural waters impacted by serpentinization are often Ca 2+ -or Mg 2+ -rich (Neal 1984;Paukert et al. 2012), thus appropriate for carbonate precipitation, including in ophiolite groundwater flow systems and travertine-producing seeps/springs. Serpentinization is regarded as a geologic process important to the sustainability of the deep biosphere ) and the origin of life (Russell et al. 2010;Sleep et al. 2011).
Carbonate minerals result also from spring deposits fed by serpentinizing waters at continental sites; serpentinite-associated travertines are well known (Barnes and Oneil 1971;Flinn and Pentecost 1995) with implications for Mars-related serpentinization (Szponar et al. 2013) and CO 2 sequestration via travertine formation (Kelemen and Matter 2008;Paukert et al. 2012). These travertines have the potential to lock in organic material and preserve it over time, recording changing community characteristics. Coregistered mineral and organic analytical data are needed to test for robust detection and discernment of closely associated mineral and organic phases: careful application of Fourier Transform Infrared (FTIR) Spectroscopy has potential to answer this need.
µFTIR relies upon coupling classic petrographic microscopy with FTIR Spectroscopy. µFTIR is used in a wide range of environmental studies on microbial ecology (Wenning et al. 2002;Igisu et al. 2006Igisu et al. , 2009, and in recent projects, µFTIR has been used to identify bacteria even down to strain level (Igisu et al. 2012)

METHOD
We prepared specimens as polished solid wafers and/or homogenized powders.
We identified, confirmed, and documented FTIR wavenumber (cm -1 ; aka IR frequency) regions linked to (I) serpentinization-associated minerals and embedded organics, (II) carbonate-associated minerals, and (III) a polysaccharide proxy for EPS. In all cases, we referenced observed IR spectra to published findings (Hunt and Salisbury 1974;Lafuente et al. 2015), supported by X-ray diffraction results. ATR-FTIR is able to analyze powders, solids, and liquids with little sample preparation. The sample is placed on a crystal window (we use diamond), and an IR beam is directed onto the crystal, and attenuates through the crystal as an evanescent wave, which comes into contact with the sample at each bounce; the resulting wavelengths are read by the detector as an

Sample preparation
For µFTIR (R-and T-FTIR), we polished peridotite, serpentinite, and carbonate samples from various locations (Table 1), with polishing protocol as described previously (Lowenstern and Pitcher 2013). In brief, wafers were ground to < 500 µm thickness for R-FTIR analysis and to <100 µm thickness for T-FTIR while fixed to a round glass slide, and polished using increasingly fine polishing papers and diamond paste suspensions. For ATR-FTIR and X-ray diffraction (XRD), we powdered splits of these samples, and passed through a 100-mesh (150 µm pore size) sieve to work with a standardized <150 µm size fraction. More detail is provided below.

R-FTIR
We used a Thermo Nicolet iS50 FTIR spectrometer coupled with a Continuum IR microscope to map minerals in R-FTIR (King and Larsen 2013), using a MCT-A (mercury cadmium telluride) detector and KBr beamsplitter, with 100x100 µm beam aperture, 64 scans, and 4 cm -1 resolution. Background data were collected on polished gold. We collected sample points on individual minerals on each polished wafer to provide representative spectra of each mineral. We also mapped serpentinite and travertine wafers with autofocus at each sample point in R-FTIR using the Omnic Atlµs software, and produced frequency heat maps to differentiate regions with different bond characteristics. For maps, we used a 25x25 µm beam aperture, scanning sample points at 25 µm intervals, with 128 scans and 4 cm -1 resolution at each sample point.

T-FTIR
We used the same instrument, detector, beamsplitter, and mapping settings as R-FTIR to map a CROMO serpentinite wafer (SRP-1) using T-FTIR.

Attenuated total reflectance FTIR
We analyzed powdered splits of rock/mineral samples via ATR-FTIR (Lowenstern and Pitcher 2013), using the same unit, same settings, and diamond window.
Background data were collected on air.

XRD
We confirmed mineral identification of powdered splits via X-ray diffraction using Olympus Terra XRD unit (Blake et al. 2012) outfitted with a Co tube. Operating parameters for this instrument are set to 250 exposures. Peak identification was facilitated using the XPowder (http://www.xpowder.com/) peak-matching software.

Incubation experiment
Serpentinite-hosted well water was collected directly from the scientific monitoring well at the Coast Range Ophiolite Microbial Observatory (CROMO) Quarry Valley 1,1 well (Cardace et al. 2013) in January, 2016, by pumping water into a cleaned polypropylene slide staining jar, fitted with polished serpentinite samples attached to glass slides. The closed jar remained at ambient laboratory temperature (~21ºC) for 3 weeks, followed by 4 weeks in a standard refrigerator to inhibit growth, prior to analysis.
Incubated wafer surfaces were analyzed by R-FTIR with standard reflection mode settings as above, and re-analyzed after surface cleaning with isopropanol.

Limit of detection assay for polysaccharides
We mixed pulverized serpentinite matrix (<150 µm fraction) obtained during CROMO drilling (Cardace et al. 2013)  analytical balance, and homogenized by shaking prior to analytical work. Xanthan gum mixtures were analyzed by ATR-FTIR, with 128 scans and 8 cm -1 resolution (inverse centimeters, also called wavenumbers, is equivalent to the infrared frequency).
In addition, we made a homogenous 2 g/L solution of xanthan gum and DI water and applied drops onto a polished serpentinite wafer (CROMO surface serpentinite sample). After a film formed, the polysaccharide-coated wafer was analyzed on and off the xanthan gum film with R-FTIR using standard reflection mode settings with the background collected on polished gold.

Confirmation of applicability of R-FTIR, ATR-FTIR and T-FTIR to serpentinites and travertines
Figures 3 through 8 (and appendix figures 1-9) show the bulk ATR-FTIR spectra and R-FTIR spectra of the minerals identified in each of the samples (Table 1); minerals identified by cross-referencing spectra are listed in parentheses.
In general, we determined major components of the mineralogy using ATR-FTIR with reference to the RRUFF database (Lafuente et al. 2015). ATR-FTIR mode was able to resolve the minerals in highest concentration in the powder splits. For example, serpentinite from New Zealand was analyzed in ATR-FTIR and R-FTIR; for this sample, cm -1 ). Sometimes, however, when there are multiple minerals represented in the spectra, identifying all of the minerals can be difficult due to peak overlap. For example, the ATR-FTIR spectrum for olivine (OL-1; Figure 7) shows strong olivine (forsterite) peaks, however, there is a peak at 3680 cm -1 , which may indicate the presence of serpentine.
Given this peak, it is likely that the shoulder of the strongest forsterite peak at 930 cm -1 is due to serpentine as well; the presence of lizardite would also account for the constructive interference at 600 cm -1 since this peak occurs in both forsterite and lizardite.
R-FTIR is useful in observing minerals individually on a polished surface, whereas ATR-FTIR identifies the minerals of highest concentration in the bulk powder.
R-FTIR provides spectra with similar shape and relative peak height to the ATR-FTIR reference spectra (as in Lafuente et al., 2015). There is, however, common occurrences of A principle component analysis (using JMP ® ) of ATR-FTIR spectra was conducted by including all wavenumber and intensity data (7000+ data points) from each sample ( Figure 12). Principle component 1 (PC 1) appears to separate spectra based on overall intensity of the spectrum; for example, MAG-1 pointing left has a much higher spectrum than other samples (Appendix Figure 6). PC 2 seems to separate the spectra by peak location; therefore, the samples are separated by lithologic type: ultramafics are With R-FTIR frequency heat maps, we are able to parse FTIR data using peaks that are diagnostic for specific minerals (Figure 10 &11). Frequency heat maps of serpentinite from CROMO (id SRP-1), based on the diagnostic peaks for serpentine at 976 cm -1 (Si-O) and for magnetite at 680 cm -1 ( Figure 10B and C, respectively), highlight areas of highest intensities (absorbance) at those wavenumber regions in bluer tones, and areas of lower intensities in redder tones. The heat map of 976 cm -1 effectively highlights the magnetite-rich region in red, which does not have a distinguishable peak or elevated absorbance at 976 cm -1 (that is, serpentine does not co-occur with magnetite at this scale)  Figure   11). The black mineral region has low intensity carbonate peaks at 1410 cm -1 and 884 cm -1 with the strongest peak at 1538 cm -1 , which is related to the ferrocyanide staining on this sample (C-N and/or H-N), used to flag ferrous iron-rich regions in this thin-section.
The red-shaded mineral region has low intensity peaks representing Si-O and a strong peak at 1556 cm -1 . The clear area to the left is a pore in the sample and has low intensity glue and glass peaks in this region.
A frequency T-FTIR area map of a serpentinite (SRP-1) doubly polished wafer was created to assess the ability of T-FTIR to detect embedded organics with a strong mineral background. With this map, we found an increase in intensity in C-H bond wavenumber regions (~2900 cm -1 ) and in O-H, and amide (Amide 1: C=O, Amide 2: CNH) wavenumber regions (~1600 cm -1 ) ( Figure 13D) near areas rich in magnetite. By conducting a PCA heat map, in the Omnic Atlµs software, we could select the peak area regions on which to base the PCA. Choosing the 2900 cm-1 (3000-2830 cm-1) and 1600 cm-1 (1800-1500 cm-1) peak areas for PCA provided a first principle component (PC1) that controls 92% of the variance, and is based on regions that have higher peak areas at both the 2900 cm-1 and 1600 cm-1 wavenumber regions ( Figure 13C); a PCA heat map of PC1 highlights all regions on and surrounding magnetite ( Figure 13B). By extracting a line map from the area map (red line in Figure 13A extracted as a heat map in Figure   13D), we observe an overall spectral elevation when the IR beam is overlying magnetite (the black area about half way up the line), which is a result of magnetite being opaque; and we can also see increases of the 2900 cm -1 and 1600 cm -1 regions at and near the magnetite and tapering off away from the magnetite vein.

Incubation Experiment
R-FTIR data were collected on a transect of a CROMO serpentinite sample (SRP-1) after incubation in naturally occurring groundwaters (from CROMO Quarry Valley 1,1 scientific monitoring well; Figure 14A) The line map was taken before ( Figure 14B) and after ( Figure 14C) cleaning with isopropyl alcohol; all sample points on the transect show very similar spectra before cleaning and again after cleaning; therefore, one representative spectrum was selected for before and after cleaning. Before cleaning the surface, the spectra show a strong serpentine signal with a broad, low intensity background noise from 1200 cm -1 to 2500 cm -1 . After cleaning the surface, the spectra show a strong serpentine signal with little background noise, likely as a result of the instrument warming up over time.

Assessment of FTIR-based resolution of surface films/embedded organic loads related to serpentinites: Constraining the limit of detection for a representative polysaccharide
Xanthan gum (exopolysaccharide of Xanthamonas campestris) serves in this study as a proxy for biologically produced exopolysaccharides, and data confirm that incipient biofilm formation can be tracked using FTIR. Xanthan gum represents several of the organic bonds that would be found if a biofilm developed on the surface of a rock ( Figure 15), including O-H, C-H, C=O, Carboxylate groups, acetate groups and C-O (Osiro et al. 2011). We used R-FTIR to contrast relevant spectral regions on a serpentinite wafer with and without xanthan gum film ( Figure 16). Data collected on the xanthan gum film show medium intensity peaks associated with xanthan gum and strong intensity peaks associated with serpentine in the underlying rock matrix ( Figure 16C).
When compared to data collected on serpentinite without the xanthan gum film, the intensity of the serpentine peaks was greater (higher) on the xanthan gum-free area; the intensity of the serpentine peaks was lesser (damped) on the xanthan gum-coated area. A PCA of data points along a transect from on the xanthan gum film ( Figure 16A point #1) to off the xanthan gum film ( Figure 16A point #29) separate spectra that are on and off the xanthan gum film ( Figure 16B).
For the set of synthetic mixtures with increasing xanthan gum proportions in pulverized serpentinite, analyzed using ATR-FTIR, we used an O-H angular deformation peak at 1600 cm -1 to determine the presence of xanthan gum because 1) there are no overlapping mineral peaks in this region, and 2) the peak is narrower and higher in intensity when compared to other peaks free of overlapping mineral peaks. Thus, the 1600 cm -1 peak can be detected most easily at low concentrations. When observing the entire spectra ( Figure 17A), the 1600 cm -1 peak emerges at 5 wt. % xanthan gum; the pure serpentinite powder and low xanthan gum concentration mixtures have a peak at 1630 cm -1 with a shoulder peak at 1555 cm -1 , both of which may relate to organics/biomass intrinsic to the rock matrix itself ( Figure 17B). Spectral data for mixtures with concentrations below 20 wt. % xanthan gum maintain a clear signal for serpentine in that region; however, at 0.1 wt. % xanthan gum, a change in peak shape is distinguishable from lower concentrations, where the baseline is smoother and the 1630 cm -1 peak is shifted towards lower wavenumbers. Peak height, area and location were recorded for every mixture (Figure 18-20). There is constructive interference at this region by other bonds at 1630 cm -1 ; the peak height still increases as the peak shifts from 1630 cm -1 to 1600 cm -1 with increasing xanthan gum concentration at a constant rate (R 2 = 0.9947; Figure 18A). Even at low concentrations (at least by 1 wt %), there is a consistent increase in peak height in the 1630 cm -1 to 1600 cm -1 region ( Figure 18B).
Peak area of the 1775-1500 cm -1 region also increases at a constant rate (R 2 = 0.99405; Figure 19A), and again, at low concentrations (at least by 1 wt %), there is a consistent increase in peak area ( Figure 19B). Peak location, as previously mentioned, shifts towards lower wavenumbers with increasing xanthan gum concentration in a logarithmic rate (R 2 = 0.79656; Figure 20A). There are outliers, such as 0.05 and 0.1 wt %, that occur in all three measurements; the reason may be that this range of concentrations is at a threshold with increased xanthan gum so that it experiences constructive interference with nearby peaks, causing a baseline shift, and thereby shifting all of the measurements.
It also may be the case that the aliquot of serpentinite for that sample started with fewer organics.

ATR-FTIR
ATR-FTIR, a method that is widely available, provides a bulk reading of dominant minerals in powdered rock/mineral samples. ATR-FTIR is able to differentiate the serpentine (~980 cm -1 and 3700 cm -1 ), olivine (~860 cm -1 ), pyroxene (multiple sharp peaks between 1060 cm -1 and 620 cm -1 ), and magnetite (~660 cm -1 ) individually with ease based on relative peak shape and location (Table 2). However, it is difficult to distinguish the subcategories of minerals (i.e., the serpentine minerals, antigorite and lizardite) because the bonds and their relative abundances (on which FTIR data are based) are very similar. ATR-FTIR is also able to distinguish between the carbonate (~1400, cm -1 870 cm -1 and 730 cm -1 ) and silicon-rich layers (1020 cm -1 ) in travertine (Table 2; Figure 11), although, again, differentiating the carbonate minerals presents a challenge because the bonds and their relative abundances are similar.
The main issue encountered with ATR-FTIR is the ability to distinguish peak overlap of different minerals in the bulk-powdered sample, in addition to its overall decreased sensitivity compared to µFTIR. One needs to be familiar with relative mineral peak location and shape to identify multiple minerals within the bulk, powdered reading.
As a result of this pilot study, it is clear that we need to develop a standardized, robust statistical treatment of peak locations that can help us interpret complex samples.

R-FTIR is capable of differentiating adjacent minerals and individual grains
greater or equal to the aperture size of the IR beam (between 25x25 µm to 150x150 µm).
The higher resolution afforded by R-FTIR data allows direct probing of sample characteristics. However, there may be peak shifting issues with R-FTIR; one must consider the possibility of constructive interference of the presence of other minerals in the sample point region. The shape and relative peak locations will likely handle peak shifting; however, a correction (Kramer's Kronig correction), may be very useful addressing minor peak shifts (Spragg 2013).
Frequency heat maps are useful in highlighting specific aspects of rich spectral data sets, as when searching for a mineral that has a distinct spectral peak relative to adjacent minerals. If spectra have a low S:N, low intensity trace mineral or organic peak variations may not be resolvable, even with this method.
For microbiology, R-FTIR is a useful tool in observing changing organic bonds across spatial boundaries with high resolution, specifically to analyse microbial biofilms grown on varying nutrient conditions (Chen et al. 2013). Other environmental studies have used R-FTIR to observe phases in rocks, such as coal and shale (Chen et al. 2015) at the same scale, and finer resolution of organic compound type (full identification and possibly compound-specific isotopic characteristics) will strengthen the interpretive power of this analytical approach.

T-FTIR as a tool for detecting embedded organics at mineral boundaries
Microbiology studies that involve T-FTIR typically include stamping or drying a film of bacterial colonies on an IR transparent plate (Igisu et al. 2012;Faghihzadeh et al. 2016). In other environmental studies, thin wafers of rocks are analyzed to observe volatile inclusions, such as H 2 O and CO 2 in rocks (Lowenstern and Pitcher 2013).
Similarly to this study, another study has used T-FTIR to observe prokaryotic microfossils embedded in chert (Igisu et al. 2006). This study combines these T-FTIR utilities to observe naturally occurring biogenic organics, such as lipid and protein microbial debris in rocks related to serpentinization, a process that fuels life on Earth.

FTIR spectroscopy as a tool for resolving organic loads on/in Earth materials
It is unclear whether short duration incubations (~3 weeks' duration) produce enough adhered biomass/organic materials to be well resolved by FTIR approaches.
However, analyses of synthetic mixtures of serpentinite and xanthan gum do reflect changing intensity of the 1600 cm -1 peak, a peak related to polysaccharides (Osiro et al. 2011) in a wavenumber region unhindered by mineral peak interferences. We show that xanthan gum is detectable in the range of expected organic load in serpentinites (total carbon @ 0.001 to 0.1 %, equivalent to 10 to 1000 ppm total carbon, respectively) (Alt et al. 2012). FTIR thus has the capability to detect low, geologically relevant concentrations of organic material in natural serpentinites. Diagnostic FTIR peaks for organics as carbohydrates, DNA, lipids, and proteins (Maquelin et al. 2002;Osiro et al. 2011;Igisu et al. 2012) should be used in regions where mineral peaks do not overlap: the strongest minerals peaks are typically in the 1400-500 cm -1 range -identifying the minerals first and then searching for lower-intensity biomarker peaks in other wavenumber regions allows for best accuracy. Statistical treatment of xanthan gum mixtures' peak locations and intensities can help us interpret natural serpentinites.

CONCLUSIONS
The findings of this study indicate that organic presence with a mineral background can be detected using FTIR. Implications for this study include the ability to detect and map polished rock surfaces to determine the association of organics to mineral phases and boundaries in serpentinization-related lithologies. Future directions include analyzing polished rock and mineral wafers before and after incubation in serpentinitehosted natural waters in situ, differentiating signatures of surface biofilms (via R-FTIR) from embedded/preserved organics (via T-FTIR), and cross-referencing organic-rich regions in natural samples with other analytical techniques (e.g., Raman spectroscopy).
As a result of this study, we confirm the applicability of FTIR-based techniques in microscale investigations of organics in ultramafic and carbonate rocks, and we establish the need for an integrated database covering mineral and organic compounds.                     Figure 19. A) Peak area of the 1600 cm -1 region (1775-1500 cm -1 ) from 0 -100 wt. %

LIST OF FIGURE CAPTIONS
x.g. in powdered serpentinite matrix is presented. The best-fit line provides an R 2 of 0.99405. B) Peak area of the 1600 cm -1 region (1775-1500 cm -1 ) from 0 -5 wt. % x.g in powdered serpentinite matrix is presented as a log function. Figure 20. Peak location of the larger peak in the 1600 cm -1 region (1775-1500 cm -1 ) from 0 -100 wt. % x.g. in powdered serpentinite matrix is presented. The best-fit logarithmic curve provides an R 2 of 0.79656. B) Peak location of the larger peak in the 1600 cm -1 region (1775-1500 cm -1 ) from 0 -5 wt. % x.g in powdered serpentinite matrix is presented as a log function.