Clay mineral formation under oxidized conditions and implications for paleoenvironments and organic preservation on Mars

Clay mineral-bearing locations have been targeted for martian exploration as potentially habitable environments and as possible repositories for the preservation of organic matter. Although organic matter has been detected at Gale Crater, Mars, its concentrations are lower than expected from meteoritic and indigenous igneous and hydrothermal reduced carbon. We conducted synthesis experiments motivated by the hypothesis that some clay mineral formation may have occurred under oxidized conditions conducive to the destruction of organics. Previous work has suggested that anoxic and/or reducing conditions are needed to synthesize the Fe-rich clay mineral nontronite at low temperatures. In contrast, our experiments demonstrated the rapid formation of Fe-rich clay minerals of variable crystallinity from aqueous Fe3+ with small amounts of aqueous Mg2+. Our results suggest that Fe-rich clay minerals such as nontronite can form rapidly under oxidized conditions, which could help explain low concentrations of organics within some smectite-containing rocks or sediments on Mars.

T housands of locations on Mars have rock units containing Fe/Mg-rich clay minerals [1][2][3][4][5][6] , which can both indicate the past long-term presence of liquid water important for potentially habitable environments, and also preserve organic matter adsorbed onto clay mineral surfaces and interlayers. A key question is the conditions under which these martian clay minerals formed, including both pH and oxidation state. Reducing and/or anoxic conditions might imply subsurface formation, buffered by a basaltic reservoir, or surface conditions in which the atmospheric composition was different than at present. Oxidized conditions would almost certainly be at or near the surface, made oxic by photochemistry and/or atmospheric escape. The precise details of clay mineral composition can serve as fingerprints to constrain the past environmental conditions under which they formed 6 . Previous work by Harder 7 , Decarreau et al. 8 , and Mizutani et al. 9 has shown that the precipitation of Fe-bearing clay minerals, even the dioctahedral Fe 3+ -rich smectite endmember, nontronite, occurs in solutions containing Fe 2+ under conditions that are anoxic and/or reducing (Eh <~−0.2 V) 7 . The initially formed Fe 2+ -bearing precipitates are subsequently oxidized to form nontronite 7,9 . The apparent requirement for initially anoxic and/or reducing conditions is because a divalent cation, e.g., Fe 2+ , is required to stabilize the octahedral layers and promote bi-dimensional (e.g., the a and b axis) growth of these clay minerals 7,10,11 . Low temperature experimental attempts to synthesize clay minerals from Fe 3+ -bearing solutions without a divalent cation have formed amorphous 7 and/or very poorly crystalline products 12 .
Here we test whether Fe/Mg clay minerals can form under oxidized conditions at temperatures in which terrestrial life may survive by performing clay mineral synthesis experiments. These experiments were designed to be relevant to potentially habitable environments on Mars, containing a range of chemical compositions, including both Fe and Mg end-members as well as intermediate compositions. We demonstrate the rapid formation of Fe-rich clay minerals of variable crystallinity from aqueous Fe 3+ in the presence of at least small amounts of Mg 2+ . Our results suggest that Fe 3+ clay mineral-bearing terrains may not necessarily be conducive to the preservation of organic matter on Mars.

Results
XRD and µXRD analyses. Fe-rich clay minerals, including nontronite 8,9 were rapidly synthesized within 2 days at 150°C, following 1 day at room temperature and within 60 days at 100°C following 1 day at room temperature from oxidized, Fe 2+ -free solutions, which contained at least 5% Mg (i.e., a molar ratio of Mg:Fe 3+ greater than or equal to 1:19 in the solution) ( Table 1).
X-ray diffraction (XRD) and synchrotron micro XRD (μXRD) analysis of synthesized Fe/Mg materials indicate broad low angle 001 peaks, which are characteristic of the large interlayer spacing of smectites (Fig. 1, and Supplementary Figs. 11-13, 24-25, 28-30, 33-35, 37, 39), particularly given the lack of other well-defined peaks that might indicate mica or a peak at a 7 Å spacing which could indicate 1:1 clay minerals 13 . Upon glycolation, synthesized products showed increasing expansion as the Fe 3+ :Mg ratio decreased from 19:1, with the 1:1 ratio material expanding to 17 Å, indicative of a smectite. The range of swelling properties of the precipitates is a characteristic of synthetic high-charge nontronites 8,9,14 and Fe/Mg smectites (Supplementary Note 1).
XRD analyses indicate variable crystallinity, with the material containing a Fe:Mg ratio of 1:1 having the highest degree of crystallinity based on the basal reflection (Fig. 1). The broad peaks in powder XRD and μXRD indicate that the precipitates formed in this study remained less crystalline than some wellcrystalline Fe/Mg-rich clay mineral standards. However, clay minerals in natural terrestrial environments are often less crystalline than reference clay mineral standards 15 , and in some cases exhibit lower crystallinity than the precipitates in this study 16 .
XRD analyses indicate that the b-dimension for these samples was~1.54 Å. This is comparable to the previous work of Russell and Clark 17 , documenting 060 spacings of 1.512 to 1.535 Å for natural nontronites, and to previous measurements of synthetic nontronites that have 060 spacings of~1.54 Å 8,9 . These values have been shown to result from the presence of Fe in both the octahedral and tetrahedral sites, which increases the b-dimension of the clay minerals 10,17 .
μXRD analyses also suggest the presence of ferrihydrite and potentially brucite in the synthesized materials ( Supplementary  Fig. 42). Although there is some overlap in the diffractions between ferrihydrite and brucite, the increased intensities of the putative brucite in Mg-rich experiments is consistent with the presence of brucite ( Supplementary Fig. 42).  Table 3) for all samples formed under oxidized conditions are also consistent with the ferric mineral nontronite (standard NAu-1), which was also analyzed and used as a reference in this investigation. The 100-Fe control (subsequently oxidized) had slightly larger isomer shifts and quadrupole splitting than the other experiments (Supplementary Table 3 and Supplementary Fig. 8), but the quadrupole splitting is still below that of Fe 2+ -bearing clay minerals 20 . In addition, our Mössbauer analysis of the 100-Fe control exactly replicated the Mössbauer analysis by Mizutani et al. 9 of their clay mineral precipitates, which were also shown to be dominated by Fe 3+ . The difference between the 100-Fe control and the other samples may result from the fact that the 100-Fe control initially contained Fe 2+ rather than Fe 3+ , with resulting differences in Fecoordination.
Visible near-infrared and infrared spectroscopy. Syntheses using pure Fe 3+ solutions with no Mg resulted in an amorphous precipitate ( Fig. 1, Supplementary Figs. 20 and 21) with visible near-infrared (VNIR) and infrared (IR) spectra distinctly different from Fe/Mg clay minerals (Fig. 2). Our synthesized Fe/Mg clay minerals, when compared by VNIR and IR at orbital instrument spectral resolution to other synthetic materials, terrestrial analogs, and martian Fe/Mg smectites 22 , are very similar (Fig. 2). With increasing Mg concentration, a band shift from the 2Fe-OH band (2.285 μm) towards the 3Mg-OH bending and stretching vibrations at 2.315 μm occurs 14,23,24 . The structural OH overtone located at~1.410 μm and the fundamental OHstretch between 2.760 and 2.800 μm also display a systematic shift (Supplementary Figs. 14-17) 14,24 .
The synthetic clay minerals appear red/brown in color (Supplementary Figs. 9,10,18,19,22,23,26,27,31,32,36 and 38) possibly due to tetrahedrally-coordinated Fe 3+ (common in terrestrial nontronites 17,25 ), but more likely resulting from the presence of small amounts of Fe-oxyhydroxides observed in the µXRD and inductively coupled plasma-optical emission spectroscopy (ICP-OES) results 25,26 , which are also common in terrestrial nontronites 27 . We note that the 50 Mg experiment (the most crystalline of the mixed cation precipitates) produced an absorption at~650 nm, which contributes to the green color in nontronite.
The presence of charge transfer absorptions and the position of electronic transitions are consistent with ferric clay minerals rather than ferrous for all precipitates in this study (Supplementary Figs. 14 and 16). Broad vibrational absorptions in the VNIR (Supplementary Figs. 14-17), indicate that the precipitates formed in this study remained less crystalline than some wellcrystalline Fe/Mg-rich clay mineral standards.
SEM and EDS analyses. The scanning electron microscopy and energy dispersive spectrosocpy (SEM/EDS) analysis of the precipitates showed that all precipitates were extremely fine-grained, and most of the crystallites were below the resolution of the SEM, making characterization of the morphology difficult (Supplementary Figs. 40 and 41). This observation is in agreement with the XRD data, which indicated a poorly crystalline/fine grained product (also characteristic of terrestrial clay minerals). No other secondary phases were detected with SEM and EDS. However, the elevated concentrations of Fe relative to Si (as determined by EDS) within the samples are consistent with nano phase Fe-oxyhydroxides within the precipitates. If nano phase Fe-oxyhydroxides are present within the sample, they would almost certainly be below the resolution of the SEM/EDS analysis.
EMP and ICP-OES analyses. The chemical composition of the precipitates was analyzed by electron microprobe (EMP) (Supplementary Table 4), and ICP-OES (Supplementary Table 5). EMP and ICP-OES analyses indicate Fe and Mg enrichment relative to silica (Supplementary Tables 4 and 6). ICP-OES analyses were used to calculate proposed mineral formulae and secondary phases as described in the methods and reported in Supplementary Table 6. Most of the conducted experiments (e.g., the 100-Fe control, 100-Fe 3+ , 5-Mg 95-Fe 3+ , and 15-Mg 85-Fe 3+ precipitates) have Fe:Mg ratios calculated for the octahedral layer that are >1:1, and because the Fe is in the ferric state with the majority of the octahedral layer filled with trivalent cations,  Table 6), and therefore may be more similar to a trioctahedral smectite than a dioctahedral smectite. Because all experiments showed some degree of expansion, the di-octahedral precipitates are referred to as nontronite and the tri-octahedral precipitates as saponite/ stevensite.

Discussion
In this study, synthesis experiments conducted under oxidized conditions in which a divalent cation was present in a concentration of at least 5% led to the precipitation of clay minerals, whereas solely trivalent cations in solution led to a relatively  amorphous product. Previous work by Harder 7,11 has indicated that the divalent cation Fe 2+ was required for the precipitation of Fe-rich clay minerals, and more recent work by Baldermann et al. 10  The ranges in chemical composition used in our experiments and common in natural environments 17,28,29 can cause gradients in crystallinity. XRD analyses of our precipitates indicate variable crystallinity, with the material containing a Fe:Mg ratio of 1:1 having the highest degree of crystallinity based on the basal reflection (Fig. 1). Upon glycolation, synthesized products similarly showed increasing expansion as the Fe 3+ :Mg ratio decreased from the most Fe-rich precipitates, with the 1:1 ratio material expanding to 17 Å, indicative of classic smectite behavior. Previous work by Grauby et al. 14 synthesizing the nontronitesaponite series at higher temperatures over a longer time period does not report a significant gradient in crystallinity along the Fe 3+ -Mg compositional gradient comparable to that which we see in our experiments. However, although all precipitates from Grauby et al. 14 were crystalline, based on our examination of their published data, the most Fe-rich and Mg-rich samples were less crystalline, where the Fe-rich sample had decreased crystallinity indicated in the 0k0 reflection, and the Mg-rich sample had decreased crystallinity along the 00l (e.g., the c-axis). Intermediate Fe:Mg ratio experiments, in contrast, were more crystalline. Therefore, our examination of their results suggests a similar trend in crystallinity with changes in composition to that observed in our experiments.
The VNIR spectra of our precipitates (Fig. 2a) indicate that materials with varying degrees of crystallinity such as those precipitated in our experiments and common in natural terrestrial soils 15,16 may produce spectra similar to well-crystalline clay minerals with comparable chemical compositions (Fig. 2c). Our synthesized Fe/Mg clay minerals, when compared by VNIR and IR at orbital instrument spectral resolution to martian Fe/Mg smectites 22 , are also very similar (Fig. 2b). Milliken et al. 30 have also shown that the poorly/nano crystalline material hisingerite has VNIR spectral properties similar to nontronite. X-ray amorphous or poorly/nano crystalline phases are a significant component of the samples analyzed by MSL at Gale Crater 31 , and our results indicate that materials on Mars currently identified as clay minerals from orbit may also include poorly/nano crystalline materials.
Our results can also help interpret the chemical composition of Fe/Mg-rich materials from orbit. Increasing concentrations of Mg (from 0 to 100%) show a band shift from the 2Fe-OH band (2.285 μm) towards the 3Mg-OH bending and stretching vibrations at 2.315 μm 14,23,24 . A systematic shift is also evident in the structural OH overtone located at~1.410 μm and the fundamental OH-stretch between 2.760 and 2.800 μm (Supplementary Note 1) 14,24 . In combination, these absorption features can be used to interpret the ratio of Fe to Mg in clay minerals observed from orbit on Mars (Fig. 2).
Synthesis experiments can also help us better interpret past aqueous conditions on Mars, including the chemical composition, the oxidation state, and the duration and temperature of alteration. Increasing the duration of ageing in experiments (corresponding to the time of water-rock interaction) increases the crystallinity of the precipitates 10,32 . Warmer temperatures over the same duration also increase crystallinity 8 . Thus, either shorter durations of water-rock interaction and/or colder temperatures may be implicated if Fe-rich clay minerals on Mars are, like some of our synthesized materials, poorly crystalline. In addition, the presence of more poorly crystalline materials is also consistent with a lack of significant water-rock interaction after their precipitation 33 .
Seeking possible signs of life, including organic biosignatures, is a crucial part of the Mars Exploration Program. Our work demonstrated that dioctahedral Fe-rich clay minerals, such as nontronite, of variable crystallinity can form rapidly under oxidized conditions in the presence of at least small amounts of Mg. Oxidized conditions would not favor the preservation of organic matter in the primary environment of clay mineral precipitation. Since Fe-rich smectites containing Mg are widespread on Mars 6,34-37 , these results suggest that clay mineral-bearing terrains may not always be conducive to the preservation of organic matter, and could help explain low concentrations of organic compounds in some smectite-bearing rocks explored by the Curiosity rover at Gale Crater 38,39 .

Methods
Mineral synthesis. In order to better understand clay mineral-forming environments on Mars and the potential for the preservation of organic matter, we performed synthesis experiments of clay minerals with a range of chemical compositions. Experiments were performed under oxidized and basic conditions ( Table 1). Initial silica-containing solutions were made acidic to dissolve Fe 3+ -and Mg-sulfates, and were then made basic by the addition of NaOH. This process is likely analogous to the interaction of acidic, ferric sulfate-containing solutions, such as those proposed on Mars 40 , with silica-containing minerals, resulting in an increased solution pH. The solutions were then incubated at two different temperatures to test the effects of temperature. In addition, a synthetic nontronite control was synthesized using standard methods from Fe 2+ -containing solutions 8,9 . Detailed experimental methods are provided immediately below.
Following a method similar to that of Mizutani et al. 9 , 4.35 g of sodium metasilicate was dissolved in 420 mL of 18.2 MΩ water. The solution was then acidified with 0.5 M H 2 SO 4 to a pH of~3. For the synthetic nontronite control only, 4.2 g of sodium dithionite was dissolved in the solution to maintain reducing conditions. Various concentrations (Table 1 and Supplementary Tables 1 and 2) Table 1), or 100°C in high-density polyethylene bottles for 60 days (Supplementary Table 2). At the end of each experiment, the pH of the solution was measured with an S20 SevenEasy pH meter using a two-point calibration. Samples were then vacuum-filtered to obtain the solid particulate using a Pyrex 47 mm, Microfiltration all-glass assembly with a 300 mL funnel and 47 mm, 0.45 μm membrane disc filters (as recommended by Moore and Reynolds 16 and following the methodology of Mizutani et al. 9 ). Samples were then dried invacuo in a desiccator. Filtration might favor the collection of larger crystallites/ particulates, as any crystallite/particle smaller than 0.45 µm may pass through the filter, which might increase the observed overall crystallinity of the precipitates. Table 1, we added ferric sulfates to the experiments, and ferrous sulfates to the nontronite control. The oxidation state of the reagents was confirmed by measuring them by XRD (Supplementary Figs. 1 and 2), and the oxidation state of the iron within our synthesized precipitates using SMS. We measured four samples using SMS: the clay mineral standard NAu-1 and experiments 100-Fe Control, 15-Mg 85-Fe, and 50-Mg 50-Fe -further analytical details are given below. All SMS analyses of the precipitates showed the sample to be 100% ferric, with any ferrous Fe present below the detection limit of SMS (see SMS results section and Supplementary Figs. 3-8 and Supplementary Table 3). The minimum oxidation state of the solutions based on Fe concentrations and an extremely conservative upper limit of the potential Fe 2+ contamination in our initial ferric starting materials of 3% is 0.856 V, and ferrous Fe is extremely rapidly oxidized under these high pH, oxidized conditions 42,43 .

Redox condition of experiments. As shown in Supplementary
Characterization and analytical techniques. Synthesized materials were analyzed with powder XRD, synchrotron μXRD, SMS, VNIR Spectroscopy, SEM with EDS, EMP analysis, and ICP-OES.
Powder X-ray diffraction. The powder XRD data were collected for the synthesized samples using a PANalytical X′Pert Pro X-Ray Diffractometer at 40 kV and 20 mA using CuKα radiation in the X-Ray Fluorescence and X-Ray Diffraction Laboratory (XXL) at the University of Nevada, Las Vegas. The analyzed samples were prepared in a similar manner to the glass slide method illustrated by Moore and Reynolds 16 in which samples were lightly ground in the presence of ethanol. The slurry was then poured onto a silicon wafer and dried at 60°C. Patterns were taken under air dried conditions, as well as after treatment with ethylene glycol vapor for a minimum of 24 h to measure expansion and therefore identify the specific clay mineralogy.
Clay minerals are generally fine-grained, crystalline, hydrous phyllosilicates 16 . Of these minerals, the smectite and the vermiculite group are defined by their ability to expand and shrink when exposed to heat, water and polar organic compounds (vermiculite to a lesser extent than smectites) 16 . If expansion was observed, a trait reserved for smectites and vermiculites, the clay mineral was described as a high-charge smectite to encompass both smectite and vermiculite properties, and if the clay mineral expanded to ≥ 17 Å that clay mineral was described as a smectite.
Synchrotron micro X-ray diffraction. Select aliquots of powdered sample were placed in 1 mm capillary tubes and examined with μXRD. The μXRD was conducted at the 16-ID-D beamline of the Advanced Photon Source in the Argonne National Laboratory using monochromatic radiation with a wavelength of 0.860250 Å and also at beamline 12.2.2 of the Advanced Light Source (ALS). The beam was focused to a 30 × 40 μm 2 spot at the sample position. A MAR165 area detector was used for collecting the diffraction data. Sample detector distance and geometric distortions were determined based on a CeO 2 standard using GSE-ADA 43 . The diffraction patterns were integrated using Fit2D 44 .
Synchrotron Mössbauer spectroscopy. In order to determine the oxidation state of the Fe within the precipitates, SMS was performed on selected samples (NAu-1, 100-Fe control, 15-Mg 85-Fe 3+ , and 50-Mg 50-Fe 3+ ) at the 16-ID-D beamline of the Advanced Photon Source at Argonne National Laboratory using monochromatic radiation of 14.4 kV. Collection times were a minimum of 1 h, using a 10 μm steel foil to determine center shifts ( Supplementary Fig. 3). In order to determine Mössbauer parameters, fitting was performed using the software CONUSS 21 . As with previous Mössbauer studies of nontronite, spectra were modeled with two Mössbauer sites 18,20 . The octahedral sites of clay minerals contain four oxygen ligands and the remaining two are hydroxyls, which may be in either cis or trans configuration 20 and therefore require two Mössbauer sites.
Visible near-infrared and infrared spectroscopy. VNIR and IR reflectance spectra were measured for each sample, using an Analytical Spectra Devices (ASD) VNIR spectrometer and a Fourier Transform Infrared Spectrometer, respectively. Spectra were taken over the range of 0.4-2.5 μm and 2.5-25 μm. The spectra of our samples are similar to spectra measured by Observatoire pour la Mineralogie, l'Eau les Glaces et l'Activité (OMEGA), the Compact Reconnaissance Imaging Spectrometer for Mars (CRISM), the Thermal Emission Spectrometer (TES), and Mini-TES, and are therefore comparable to remote and ground-based observation of the martian surface 42 . VNIR band centers were determined from continuum-removed spectra.
SEM and EDS analyses. SEM was performed using a JEOL JSM-5610 scanning electron microscope equipped with an Oxford ISIS Energy Dispersive Spectrometer (EDS) capable of semi-quantitative chemical analysis. Additional SEM analyses were performed using a LEO 1550 SFEG scanning electron microscope, equipped with an energy dispersive X-ray spectrometer. Samples were carbon-coated and examined for morphology and chemical composition. SEM/EDS analyses were conducted at the University of Nevada, Las Vegas Electron Microanalysis and Imaging Laboratory (EMiL) and at Stony Brook University's Materials Characterization Laboratory.
Microprobe. Analysis by EMP-WDS was carried out on a Jeol JXA-8900 microprobe at the University of Nevada, Las Vegas EMiL laboratory on polished epoxy mounts. Analysis conditions were 20 keV and 10 nA using a 10 μm beam.
The difficulties encountered when analyzing clay minerals by EMP analysis and the low totals observed (Supplementary Table 4) have been previously documented 46 . Clay minerals may undergo dehydration under desiccation (during the carbon coating process and within the vacuum chamber of the microprobe itself) but retain structural OH, which may contain~5 weight percent H 2 O 46 . In addition, EMP analysis relies on a relatively flat sample surface; this can be difficult to achieve due to etching and pitting during the polishing process, which was also observed by Treiman et al. 46 . Clay minerals are also highly porous, which may contribute to low totals observed by EMP analysis 46 (Supplementary Table 4).
ICP-OES analyses. The chemical composition of the precipitates was determined by ICP-OES after acid digestion. Acid digestion of the samples was achieved using previously published methods 47 described as follows. Samples were mixed with lithium borate (1:5 ratio), and heated in an oven at 975°C for 10 min, shaken (to ensure proper mixing), and then heated for an additional 10 min. Upon removal from the oven, the mixture was poured directly into 100 mL of 1 M nitric acid, and stirred for 1 h. The solution was then diluted with additional 1 M nitric acid to the desired concentration, below the point of silica polymerization, and then samples were stirred overnight to ensure homogeneity and total dissolution. ICP-OES analyses were performed at Stony Brook University. Elemental concentrations (Al, Ca, Fe, Mg, Na, Si, and Ti) within the solutions were determined with an iCAP 6300 radial view Inductively Coupled Plasma-Optical Emission Spectrometer (Supplementary Table 5). Standards were matrix matched with nitric acid for elemental analysis. Proposed chemical formulae for the synthetic Mg-Fe clay minerals and the estimated amounts of the presumed secondary phases ferrihydrite and brucite were calculated for each of the 150°C experiments from the bulk composition of the precipitates (Supplementary Table 5), and are presented in Supplementary Table 6. In order to generate mineral percentages and mineral formulae, we assume that the interlayer charge is 0.25 (for a O 10 (OH) 2 half unit cell) for all samples that expanded to the classic~17 Å upon glycolation (100-Mg and the 50-Mg 50-Fe 3+ ). An interlayer charge of 0.75 is used for samples that did not expand to~17 Å upon glycolation 8 . Fe and Mg were added at the ratio determined through ICP-OES to the octahedral and tetrahedral sites to balance the required charge. Because Mg does not reside within tetrahedral sites 16 , Mg was given priority when assigning Mg and Fe to the octahedral sites. The remaining Fe not present in the octahedral layer was then placed within the tetrahedral sites. The remaining Fe and Mg not assigned to the clay mineral structure was portioned into the secondary phases observed by μXRD (e.g., ferrihydrite and brucite). The resulting Fe:Mg ratio of the bulk clay minerals is within 1% (molar ratio) of the initial solution chemistry. The lower ratios of Si to Fe and Mg may have resulted from the incomplete solubilization of Si, as HF was not added during the digestion 47 .
The dioctahedral Fe 3+ end-member of the smectite group is nontronite, whereas the trioctahedral Mg end-member is either saponite or stevensite, where the charge is derived from tetrahedral and octahedral substitution, respectively 16 . However, significant Fe and Mg substitution can occur between these two end-members, leading to a semi-solid solution in Fe/Mg-smectites and other clay minerals 14 . In addition, previous work by Grauby et al. 14 also suggests that di-and trioctahedral domains may occur within the same crystallite. The di / trioctahedral nature of the precipitates of this study were assigned based on the molar ratio of di-and trivalent cations within the octahedral layer calculated as described above (e.g. if the octahedral layer is dominated by ferric Fe the clay mineral is dioctahedral, whereas if it is dominated by Mg it is trioctahedral).
Data availability. The authors declare that the data supporting the finding of this study are available within the paper and its supplementary materials, and/or from the corresponding author S.R.G. upon reasonable request.