Provenance of altered carbon phases and impact history of the Stac Fada Member, NW Scotland

The Stac Fada Member (Stoer Group) is a ~1.2 Ga melt‐rich impact breccia preserved and intermittently exposed along the NW coast of Scotland. Using a combination of x‐ray diffraction and micro‐Raman spectroscopy, we identify potential coesite that is spatially associated with micron‐sized diamonds, as well as disordered carbon phases. Comparing the graphite G‐band of disordered carbon phases in the impact breccia to samples from underlying units indicates that most of the carbon in the Stoer Group was ultimately derived from the underlying Lewisian basement. Disordered carbon phases within the Stac Fada Member have been modified by mild heating within a hot ejecta blanket rather than shock pressure. We also report the first evidence for impact diamonds discovered within the Stac Fada Member. These diamonds have an average Raman shift of 1328.5 cm−1 and are present within both the impact breccia and the shocked gneiss clasts that are present in sandstones directly underlying the Stac Fada Member contact, and within sandstone rafts entrapped in the unit. These findings have implications for the timing of deposition of the Stac Fada Member, which must have occurred after ballistic ejection of Lewisian basement clasts during the impact event.


INTRODUCTION
The Stac Fada Member is an impact ejecta horizon exposed locally along a 50 km stretch of the NW Scottish coastline (Figure 1). The member is part of the Mesoproterozoic Bay of Stoer Formation, a unit of primarily fluvial channels and overbank muddy floodbasin lithologies that overlies the Clachtoll Formation, which predominantly comprises alluvial sandstones (Benton et al., 2021;Stewart, 2002). Both units were deposited as part of the~1.2 Ga Stoer Group rift basin succession deposited on the Lewisian gneiss basement (Ielpi et al., 2016;Stewart, 1982Stewart, , 1988. From 40 Ar/ 39 Ar dating of authigenic K-feldspars precipitated in degassing structures immediately post-deposition, the age of the Stac Fada Member is estimated at 1177 AE 5 Ma (Parnell et al., 2011).
The impact ejecta horizon varies in thickness from~4 to 12 m across different exposures, becoming generally thinner and more bedded toward the South (Simms, 2015). Texturally, the ejecta is a massive, poorly to moderately sorted, mud-rich matrix-supported conglomerate with cm-to dm-scale lithic clasts and intraclasts. At the Bay of Stoer locality, a folded raft of sandstone a few meters thick is hosted within the impact breccia ( Figure 1). A significant proportion of the Stac Fada Member appears to be reworked sediment with a composition and grain size similar to the underlying Bay of Stoer and Clachtoll formations, with detritus primarily derived from the basement Lewisian gneiss complex (Kenny et al., 2019). This is mixed with highly altered and partially devitrified vesicular glass which makes up 20%-30% by volume and is interpreted to be an impact melt (Amor et al., 2008). Previous work has identified shocked quartz (Amor et al., 2008;Osinski et al., 2020), a high-pressure zircon polymorph reidite (Reddy et al., 2015), and shock features in zircon (Kenny et al., 2019), directly linking the deposition of the Stac Fada Member to a hypervelocity impact event. However, evidence for shock within the member is notably sparse compared to other terrestrial hypervelocity impact deposits (Kenny et al., 2019;Osinski et al., 2020). At present, the location of the impact and, therefore, its distance from Stac Fada Member exposures, remains debated (Amor et al., 2019;Osinski et al., 2020;Simms, 2016). In addition, the mechanism that triggered the Stac Fada Member impactoclastic density current is uncertain. Suggestions include melt fuel coolant interaction, ballistic emplacement, and mixing of suspended ejecta with deeply sourced impact melt within a turbulent plume (Branney & Brown, 2011;Osinski et al., 2020;Simms, 2016). To investigate further how the Stac Fada Member was deposited, we carried out a detailed field study of the member as well as of the underand overlying units. We also carried out mineralogical analyzes of samples throughout the Stoer Group, allowing us to characterize sediment maturity and provenance for the Stac Fada Member. Using mineralogical and spectroscopic characterizations, we present evidence for (i) additional high pressure and previously unidentified phases associated with a hypervelocity impact in the Stac Fada Member, namely coesite and diamond and (ii) the origin and alteration of carbon-rich phases within the impactite.

SAMPLES AND METHODS
We collected samples during fieldwork campaigns in 2017 and 2022 from two localities with coastal exposures of the Stac Fada Member: Bay of Stoer [~NC03302850] and Second Coast [~NG92639112] (Figure 1). At the Bay of Stoer (and adjoining Clachtoll Bay, which we combine into this locality), we collected samples from the underlying Clachtoll and Bay of Stoer formations to allow for comparison of the Stac Fada Member with underlying lithologies. We also collected samples from the overlying Poll a' Mhuilt Member. Barring the Stac Fada Member, which represents an isochronous surface deposited in relation to a hypervelocity impact, samples from underlying units are from a variety of different depositional environments (Appendix 1). Samples from the Bay of Stoer Formation were collected from a crevasse splay sequence (labeled TAS in Figure 1) that represents repeated fluvial sediment packages that comprise most of the Bay of Stoer Formation at its type locality (Ielpi et al., 2016). Sample prefixes match the British Geological Survey preferred map codes for each unit.
We selected and powdered two samples from the Stac Fada Member for analysis using powder x-ray diffraction (XRD), TASF_22.08 [Bay of Stoer: NC03312856] and TASF_22.16 [Second Coast: NG92639112]. Bulk samples were crushed, from which we manually picked "matrixrich" and altered "glass-rich" grains under a binocular microscope, before powdering all fractions by hand using an agate pestle and mortar, resulting in three powders for each sample (Table 1) Powdered samples were prepared by mixing~0.1 g of the powder with~1 mL of amyl acetate, using a pestle and mortar. The resultant slurries were transferred to glass microscope slides and air-dried. Slides were analyzed by powder XRD on a Bruker D2 Phaser diffractometer at the University of Manchester, equipped with a Lynxeye XE-T detector that had an axial 2.5°Soller slit and anti-scatter screen. A CuK a1 source at 30 kV and an intensity of 10 mA was used to generate x-rays. Samples were scanned from 5°to 70°2h, with a step size of 0.04°and a count time of 0.4 s per step. Raw XRD spectra (Appendix 2) were processed using Match!-Phase Analysis using Powder Diffraction, Version 3 (Putz & Brandenburg, 2014), comparing the collected experimental data to standards from the Crystallography Open Database (Gra zulis et al., 2009(Gra zulis et al., , 2012Vaitkus et al., 2021). For consistency, the same reference phases were employed for each sample. Weight percentage abundances for each mineral phase were estimated semi-quantitatively using an inbuilt reference-intensity ratio (RIR) method based on the assumption that the non-crystalline amorphous fraction of the sample was not a significant fraction of the bulk. As nearly all glass phases within the Stac Fada Member have been either devitrified or altered to crystalline material , this is a valid assumption. We did not attempt to estimate RIR uncertainty in this study but quote accessory minerals with abundances smaller than 1 wt% as <1 wt% to reflect potential error at low abundances where the signal-tonoise ratio (SNR) is low. Data are provided in full in Table 1. A total of 23 thin sections were cut from 17 of the collected samples (Appendix 1). The samples were attached to glass slides and ground to 30 lm thickness. Although a synthetic polishing paste was used, silicon carbide in the form of carborundum (also known as moissanite) has a two diagnostic Raman peaks (Nasdala et al., 2016), allowing for easy differentiation of genuine "geological" diamonds from possible contamination related to sample preparation. Micro-Raman spectroscopy was carried out using a Horiba Xplora Plus instrument at the University of Manchester, collecting point spectra using a green laser light with a wavelength of 532 nm, a power of~5-10 mW, and 5 s exposure. Spectra were background corrected and calibrated against a polystyrene reference material as described in ASTM E1840-96 (2014). Raman hyperspectral maps were collected with pixel sizes of 1.5-15 lm using a laser beam size of 1.5 lm. The characteristic peaks for graphite and diamond were fitted using a Gaussian function to calculate peak height, full width at half maximum (FWHM), intensity, and Raman shift (Appendices 3 and 4). We quote Raman peak positions with AE1.5 cm À1 calibration errors as in Itoh et al. (2019) with the assumption this range includes possible uncertainties in the Gaussian fit. A SNR was calculated as the square of the reciprocal of the coefficient of variation (SNR = l 2 /r 2 ; where l = peak height, r = standard deviation). A Savitzky-Golay smoothed profile (window size = 9, polynomial order = 3) was subtracted from the raw data to provide a noise profile, from which standard deviation was calculated. Data are provided in full in Tables 2 and 3.

Composition and Mineralogy of the Stoer Group
Powdered rock samples for the Stoer Group analyzed by XRD show a common mineral assemblage of quartz, albite, and microcline with accessory biotite, chlorite-group minerals (either clinochlore or chlorite), and hematite (Table 1, Figure 2). The samples from the Bay of Stoer Formation (TAS_22.01, TAS_22.02, and TAS_22.03) represent a sediment sequence that has previously been interpreted as a muddy floodplain, channel sandstones, and coarse flood-deposited levee deposits within a broader floodplain environment which featured fluvial channels and avulsion events (Ielpi et al., 2016). Biotite and chlorite-group mineral abundances decrease up this~3.5 m thick sediment package. Averaging across these three samples to create a representative mineralogy of the Bay of Stoer Formation provides a near-identical mineral assemblage to that seen in the Clachtoll Formation (TAT_22.01) (Table 1), including a~1:1 weight percent ratio of quartz to albite. This same ratio is seen in the sample from the Poll a' Mhuilt Member, which shares a comparable mineralogy other than the addition of~15 wt% dolomite ( Table 1).
Comparisons of XRD spectra between the bulk Stac Fada Member samples from Bay of Stoer (TASF_22.08) and Second Coast (TASF_22.16) (Appendix 5) reveal a similar mineral assemblage dominated by quartz, albite, and microcline with lesser amounts of chlorite-group minerals and accessory hematite (Table 1). Both bulk Stac Fada Member samples have relatively higher abundances of quartz compared to the underlying Stoer Group, and a higher ratio of quartz to albite of~2.5:1. Analyzing "matrix-rich" and altered "glass-rich" fractions allows a basic assessment of where minerals are distributed within the impact breccia. Compared with the  Figure 2). Compared with the bulk Stac Fada samples, the spectra of "matrix-rich" samples are nearidentical with only slight peak broadening (Table 1; Figure 2).

Identification of Coesite
Semi-quantitative analysis of XRD spectra from Stac Fada Member samples from Second Coast suggests trace amounts of coesite, a high-pressure polymorph of quartz (Table 1, Figure 2). The transition from quartz to coesite alters the structure to monoclinic with a 10% reduction in unit cell volume, causing a change in characteristic peak position from~26.6°2h to~29.0°2h in XRD spectra, and for the main Vs(Si-O-Si) stretching mode in Raman spectra from 467 to 521 cm À1 , for quartz to coesite, respectively (Hemley, 1987;Zinn et al., 1997). A 2h peak at 29.08°in XRD spectra for samples of the Stac Fada Member at Second Coast correlates with a prominent coesite peak at 29.01°, as well as subsidiary peaks at 29.00°and 29.20°which cannot be individually resolved. We suggest that part of the estimated~5 wt% abundance of coesite results from the presence of illite, which has a weak 2h peak at~29° (Grathoff & Moore, 1996), but was below detection limits. The "glass-rich" fraction displays the strongest coesite peak, with trace abundances of coesite also found in the bulk and "matrix-rich" fractions (Table 1). No peak for coesite was observed in the Stac Fada Member sample from Bay of Stoer ( Figure 2).
Only one~5 lm coesite grain was identified using micro-Raman spectroscopy in the studied thin sections, with a characteristic peak at 521 cm À1 (Figure 3a-c). Although this grain gives a weak signal and is superimposed with anatase (TiO 2 ; a polymorph of rutile), its spatial association with diamond within a shocked clast is compelling (Figure 3a,b). Overlap exists between the coesite 521 cm À1 peak and the~512 cm À1 band for A 1g symmetry in anatase ( Figure 3c). However, the coesite peak is unlikely to be an orientation artifact from anatase, given that potential B 1g (~397 cm À1 ) and E g (~639 cm À1 ) modes in anatase have not increased proportionally (Ghiribelli et al., 2002). We note that disseminated anatase is seen heterogeneously throughout the Stac Fada Member and is not uniquely associated with coesite.

Impact Diamonds and Disordered Carbon Phases
We identified diamonds via micro-Raman spectroscopy using the first-order band at~1332 cm À1 ( Figure 3d) associated with characteristic F 2g vibrations (Prawer & Nemanich, 2004). A combination of Raman hyperspectral imagery and reflected light microscopy constrained the size of diamonds to~2-10 lm in diameter ( Figure 3b). A Gaussian function was fitted to each spectral peak to measure Raman shift and FWHM (Table 2). We detected a total of six crystalline diamonds with discernible micro-Raman spectra in three thin sections of the Stac Fada Member (Figure 4). In addition, we identified four crystalline diamonds within graphite domains in angular gneiss cobbles hosted in the folded sandstone raft present at the Bay of Stoer (Figures 1, 4, and 5a,b) and in a gneiss cobble emplaced in the sandstones that directly underlie the Stac Fada Member at Second Coast (see also Simms, 2015) (Table 2; Figures 1, 4, and 5c,d). Most diamonds exhibited a strong optical fluorescence from the 532 nm laser which made them difficult to study using micro-Raman spectroscopy. Approximately one in ten diamonds located via microscopy in our study produced a discernibly crystalline signal, with an estimate of~20 diamonds per thin section. Diamonds were more common in the impact breccia at the Bay of Stoer compared to Second Coast but had a comparatively less intense micro-Raman signal ( Figure 4a, Table 2). At both localities, diamonds are only present within the Stac Fada Member and were not identified in any of the under-or overlying formations. Diamonds have an average FWHM of 24 AE 12 cm À1 , a micro-Raman shift of~1328 AE 3 cm À1 , and a typical size of 2-10 lm. Although distributed heterogeneously, diamonds are primarily located in the breccia matrix and are always associated with masses of disordered carbon (Figure 3b,d). We define disordered carbon phases as those that feature two Raman modes: a D-band at 1300-1345 cm À1 which originates from a sp3 and sp 2 hybridized carbon, as well as graphite G-band at~1560-1600 cm À1 (Beyssac et al., 2003;Ferrari & Robertson, 2000), a schema also used to describe disordered kerogens (e.g., Kouketsu et al., 2014). A shoulder at~1200-1250 cm À1 can be attributed to a D sub-band while a broad sub-band at~1450 cm À1 cannot be mapped onto existing peak labeling criteria ( Figure 3d). However, this same~1450 cm À1 band is observed ubiquitously throughout the Bay of Stoer Formation sediments suggesting this spectral characteristic is not unique to the Stac Fada Member or may be an artifact. The disordered carbon phase does not have a discernible grain size but form~5 to 30 lm masses that are disseminated throughout the matrix of the impact breccia.
We fitted a Gaussian function to the G-band of disordered carbon phases from samples throughout the Stoer Group to compare carbon-rich phases in the Stac Fada Member with those of under-and overlying sediments ( Figure 6). For all lithologies barring Stac Fada Member, most spectra have a consistent G-band peak position at~1575-1583 cm À1 (Figure 6). One carbon-rich area from the Clachtoll Formation has a peak position of 1596 cm À1 and is associated with a carbonate band at~1085 cm À1 (Table 3). This bimodal distribution has been previously reported for carbon-rich phases from the Stoer Group, with peaks at~1575-1580 and~1600 cm À1 corresponding to graphitic and kerogenous precursors, respectively (Brasier et al., 2019;Muirhead et al., 2017) (Figure 6). Disordered carbon within the Stac Fada Member shows a bimodal distribution within two G-band ranges at 1566-1580 and 1594-1596 cm À1 , similar to carbon-rich phases in the Stoer Group overall, but with a small shift in peak position by >2 cm À1 to lower wavenumbers and a broadening of the FWHM by >20 cm À1 (Figure 6). Note that TASF_22.05 is a sandstone sample from directly under the base of Stac Fada Member at Bay of Stoer and is not part of the impactite (Figure 1). The presence of hematite throughout the studied lithologies prevented proper fitting of disordered carbon peaks because a broad band at 1320 cm À1 overlaps the disordered carbon Dband. This, along with the~1450 cm À1 sub-band which does not map onto peak labeling criteria, prevented proper fitting of the D-band that might have allowed for both better characterization and potential calculation of peak thermal exposure (e.g., Kouketsu et al., 2014).

Bulk Sample Composition
Analysis of XRD data confirms that the mineralogy of the Stoer Group is dominated by felsic minerals, and is mostly derived from the local basement Lewisian gneiss FIGURE 2. X-ray diffraction spectrum for each sample, with comparison to standards from the RRUFF database (https:// rruff.info/). The location of the different samples in the Stoer Group stratigraphy is shown in Figure 1. Characteristic peaks are labeled (dol = dolomite, qz = quartz, ab = albite, mi = microcline, bi = biotite, cl = chlorite group [predominantly clinochlore], he = hematite, co = coesite). (Color figure can be viewed at wileyonlinelibrary.com.) (Rainbird et al., 2001). Weight percent ratios of quartz to albite are consistent with an arkose composition (Table 1; Figure 2), suggesting that the sediments originated from a predominantly physical rather than chemical weathering regime (Nesbitt & Young, 1996), which would otherwise cause a trend toward stable minerals and a quartzdominated lithology. This could lend evidence to a hypothesis that deposition of the Stoer Group occurred shortly after a glaciation event (Stephen & Hambrey, 1996), or that sediments were derived from a dry desert environment (Young, 1999). Young (1999) investigated the provenance of the Stoer Group using elemental analysis with a chemical index of alteration (based on molar proportions of oxides) as a proxy to suggest that chemical weathering within formations of the Stoer Group was inhibited, especially when compared to overlying Torridon Group sediments. Our mineralogical data are in agreement with this hypothesis. Although a lack of terrestrial vegetation during the Mesoproterozoic is a potential cause for reduced chemical weathering in rocks from this time (e.g., Baars et al., 2008), we would expect a mineralogical difference between lower Bay of Stoer Formation sandstones and mudstones within the Poll a' Mhuilt Member, which represent differing sedimentary facies typically associated with different sediment maturity (Boggs, 2009)  FIGURE 3. Compilation of Raman data for high-pressure phases associated with a shocked clast: (a) Reflected light micrograph of the shocked clast, with inset showing location of (b), a false-color hyperspectral image (pixel size = 1.5 lm). Brightness of color intensities corresponds to intensity of characteristic peaks for each phase. White polygons correspond to averaged masks, which are plotted for coesite (c) and carbon-rich phases (d). Additional spectra in (c) are reference spectra from the RRUFF database (https://rruff.info/). (Color figure can be viewed at wileyonlinelibrary.com.) sample suggests a similar sediment source to that of the lower Bay of Stoer Formation. The additional~15 wt% dolomite (Table 1) is likely due to the sample being collected from a horizon that included frequent carbonate laminations (NC03222861) and might not be representative of the entire mudstone succession. There was no obvious truncation of the Stac Fada Member at the Bay of Stoer type locality (Figure 1 Comparison of bulk Stac Fada Member compositions with XRD spectra from the underlying Bay of Stoer and Clachtoll formations confirms that the majority of crystalline material within Stac Fada Member is derived from reworking of underlying arkose sandstones (Stewart, 1990). Semi-quantitative estimates suggest approximately 62 wt% quartz, 30 wt% albite, 1.5 wt% microcline, and trace amounts of biotite. This differs moderately from the lower Bay of Stoer and Clachtoll formations, which approximate to 42 wt% quartz, 42 wt% albite, 5 wt% microcline, and 3% biotite ( Table 1). The Stac Fada Member is compositionally more mature (i.e., containing fewer unstable minerals such as feldspars and micas, a proxy typically used to infer greater sediment transport distance) reflected in a higher abundance of quartz and loss of biotite. This can be explained by a number of different hypotheses, including (i) highly energetic erosion and transport of preexisting material within the impactoclastic density current increased compositional maturity by preferentially eroding weakening minerals; (ii)  Highly altered and devitrified impact glasses are likely to be the biggest contributor to the threefold increase in chlorite-group minerals within the "glass-rich" fractions for both Stac Fada Member exposures compared to their respective bulk samples. In contrast to the bulk Bay of Stoer sample, the Second Coast bulk sample contains~10% less albite and has a higher proportion of clinochlore and potentially illite within the glass-rich fraction (Table 1; Figure 2). This is best explained by the preservation of the impact glasses; at Second Coast, impact glasses are less devitrified than at the Bay of Stoer and have instead been altered to poorly crystalline clay minerals (Table 1; Figure 2). Broadening of XRD spectral peaks in the matrix-dominated samples is most likely a result of a domain size, with smaller overall grain size creating broader FWHMs (Ung ar, 2000).

Identification of Coesite
We identified coesite using XRD, as a weak peak at 29.08°2h, in both the bulk sample from Second Coast (Figure 2), and with a stronger signal in the "glass-rich" fraction that is either due to higher coesite abundance or a relatively weaker quartz signal that otherwise masks the spectra. Semi-quantitative RIR analysis also suggests trace amounts of coesite within the matrix fraction (Table 1). Using micro-Raman on thin sections from Second Coast, we found coesite preferentially associated with a shocked crystalline clast. A weak coesite spectral peak at 521 cm À1 associated with a~3 lm grain within poorly crystalline clay-rich material was observed in close association with impact diamond (Figure 3). Coesite is formed at pressures of >~2.8 GPa and temperatures >600°C (Zhang & Zhang, 2021); it is, therefore, expected within the range of shock pressures estimated for the Stac Fada Member of 12-20 GPa, with possible localized pressures in excess of $ 30 GPa (Amor et al., 2008;Osinski et al., 2020;Reddy et al., 2015).

Reworking and Alteration of Carbon Phases
Using micro-Raman spectroscopy on samples from the Stac Fada Member, we identified two carbon-rich phases: (i) crystalline grains of impact diamonds~2-10 lm in diameter with a peak position of~1328 cm À1 (Figure 4) and (ii) disseminated heterogeneously distributed disordered carbon, with G-band positions that cluster either at 1566-1580 cm À1 or 1595-1596 cm À1 (Figure 6). These phases likely represent inclusion of disordered carbon into Stac Fada reworked from the underlying lithological units, and partial transformation of some of this disordered carbon into diamond during a hypervelocity impact.
The presence of two populations of disordered carbon (Figure 6b) is in agreement with previous work on the Stoer Group (Brasier et al., 2019;Muirhead et al., 2017), though the G-bands in this study show considerably broader FWHM, which we attribute to differences in measurement instrumentation. Note that only one kerogenous spectrum ( Figure 6) was collected in association with a 1085 cm À1 carbonate peak and could represent biogenic material. Alternatively, the two populations of carbon were noted by Parnell et al. (2021) for supracrustal metasedimentary units hosted in the Lewisian classified by their Raman D/G peak ratios and carbon isotopic composition, though neither of these are kerogenous.
Comparing the G-band from disseminated disordered carbon phases within Stac Fada Member samples shows a poorly defined trend of a >2 cm À1 decrease in Raman shift and broadening of the FWHM by >20 cm À1 compared to the bimodal distribution of disordered carbon in samples of the underlying Bay of Stoer and Clachtoll formations (Table 3; Figure 6). This trend suggests that the Stac Fada Member includes reworked carbon phases from the underlying Stoer Group sediments, or that altered carbon-rich phases were similarly derived from the Lewisian basement in both the Stac Fada Member and the underlying formations. The Lewisian gneiss in NW Scotland includes supracrustal outliers of amphibolite and granulite facies metasediments which host carbonaceous pelites, schists, and limestones with graphitic carbon (Parnell et al., 2021;Wheeler et al., 2010). Although the closest supracrustal section to Stoer in the modern day is Gairloch, similar metasediments (deposited in the mid-Paleoproterozoic) may have been exposed within the Lewisian at the time and site of the impact. Thus, the source of disordered carbon in the Stoer group could be re-erosion of organic matter from these supracrustal metasedimentary outliers (Parnell et al., 2021). We also expect graphitic carbon as an accessory phase within the Lewisian high-grade metamorphic rocks (Cartwright & Barnicoat, 1987). Because we have identified impact diamonds in situ within ballistic gneiss clasts, this necessitates a carbonaceous source material for conversion (Figure 4).
A preliminary experimental shock study (Wickham-Eade & Burchell, 2017) and previous work on kerogens exposed to high pressures (Huang et al., 2010) suggest that increasing pressure cause the G-band to become narrower and move to higher wavenumbers, which is the opposite to what we observed for disordered carbon in the Stac Fada Member (Figure 6). Alternatively, a combination of impact erosion processes, causing a reduction in grain size to sub-micron scales (Ferrari et al., 2004), and thermal heating may explain the observed Gband broadening and shift to lower wavenumbers. A temperature excursion would be expected either during the impact event or as remnant heat immediately postdeposition, which has been estimated at 200°C (Parnell et al., 2011) and not exceeding 675°C (Irving & Runcorn, 1957;Stewart, 2002). We hypothesize that remnant heat following deposition of the impactoclastic density current is the dominant cause for the difference in graphite G-peak position and width in the Stac Fada Member compared to its precursor. Other factors should also be considered, however, including oxidation of the carbon phases (Brolly et al., 2016) due to bulk changes triggered from partial melting during the impact, and/or interaction with post-depositional hot fluids, as suggested to have occurred at the Ries crater (Engelhardt, 1972;Lukanin & Kadik, 2007).
A series of studies of the micro-Raman spectra of diamonds have identified statistical trends associating formation mechanisms to the diamond F 2g band position and FWHM. Terrestrial diamonds formed under static pressure have a single first-order Raman band at 1332 cm À1 with a narrow FWHM of~2-4.5 cm À1 (Knight & White, 1989;Miyamoto et al., 1993;Solin & Ramdas, 1970). Diamonds produced during shock are shifted to smaller wavenumbers and characterized by a broader FWHM (Miyamoto et al., 1993). Within Stac the Fada Member, Raman shifts for the diamond F 2g band are situated at 1332 cm À1 or lower and have a mean of 1328.5 AE 3.0 cm À1 (n = 10) (Figure 4a), which is within the expected range for shock-produced impact diamonds. Diamonds in Stac Fada also have FWHM between ca. 12 and 49 cm À1 which could potentially have resulted from shock deformation-induced peak widening (Zamyatin, 2022). These may also indicate inclusion of defects of hexagonal packing known as the diamond polytype lonsdaleite. However, we could not detect the characteristic~1305 cm À1 A 1g vibrational mode at of lonsdaleite in our spectra (Ovsyuk et al., 2019), implying shock is the dominant cause for peak widening in the Stac Fada diamonds.
The Lewisian gneiss complex is a potentially favorable lithology to host diamonds which are often associated with ancient sub-cratonic roots. However, only one kimberlite intrusion within the gneiss complex has been identified (near Ben Hope, NW Scotland) with no confirmed diamonds (Smith et al., 2008). Peak pressures and temperatures for the central region of the Lewisian gneiss complex are estimated at 8-10 kbar and 900-1000°C (e.g., Feisel et al., 2018;Johnson & White, 2011), which is below the graphite to diamond transition that occurs at ~40 kbar at~1000°C (e.g., Day, 2012). Mesoproterozoic palaeoplacers (i.e., sedimentary rocks hosting diamonds eroded from primary sources) are possible if diamondiferous intrusions existed, but no palaeoplacers have been thus far discovered in the Lewisian gneiss complex (Leake, 1995;Smith et al., 2008). We exclusively identified diamonds in samples of the Stac Fada Member and directly underlying gneiss clasts, and not in any of under-and overlying formations, despite all thin sections having been prepared the same way. Based on all these observations, we are confident that diamonds identified using Raman spectroscopy in the Stac Fada Member have an impact origin.
It remains unclear whether disordered carbon associated with impact diamonds is from later diamond alteration and inversion, or represent residual graphitic carbon phases from which the diamonds formed (Hough et al., 1995). For the Stac Fada Member, we propose the latter given the disordered carbon associated with impact diamonds possess the same enigmatic~1450 cm À1 band seen throughout disordered carbon within the Stoer Group sediments, from which diamonds are likely derived (Figure 3d). Experimental studies have verified the shock synthesis of microdiamonds from graphitebearing gneiss, followed by regraphitization creating disordered carbon phases (Kenkmann et al., 2005), which may be analogous to what we observe within Stac Fada. As reported by Lapke et al. (2000) and Palchik and Vishnevsky (2010), who studied impact diamonds within highly shocked clasts in suevite from the Ries impact crater in Germany, we note that the majority of diamonds (~90%) exhibit strong optical fluorescence. Compared with those from the Stac Fada Member, impact diamonds from Ries show a wider range of wavenumbers and a consistently broader FWHM ( Figure 4) (Lapke et al., 2000). Comparing the spectral characteristics of diamonds from the Stac Fada Member with two types of impact diamond from the Kara impact structures (Pay-Khoy, Russia) (Shumilova et al., 2018(Shumilova et al., , 2020 shows a close correlation with their "sugar-like" diamonds, which they describe as rounded polycrystalline diamonds~0.1 lm in size ( Figure 6). These differ significantly from after-organic diamonds which have a lower micro-Raman shift (1318-1324 cm À1 ) and may be more akin to the diamonds seen at Ries (Figure 6). Importantly, a lower volume of pre-impact organic life during the Mesoproterozoic (Brasier et al., 2019;Wellman & Strother, 2015) could mean the volume of carbon-rich impact products may differ significantly between the Stac Fada Member and the much more recent Ries and Kara impact craters, dated at~14.8 and 70.3 Ma, respectively, when a more modern biosphere existed (Schmieder et al., 2018;Schwarz et al., 2020;Trieloff et al., 1998). As noted in Shumilova et al. (2020), natural diamonds called "karite" are present at the Kara astrobleme, suggesting the conversion of organic life at the time of impact to diamond is possible. Such forms of carbon within Stac Fada may be possible, particularly if the impact sampled supracrustal metasedimentary units hosted in the Lewisian which are rich in microbial carbon (Parnell et al., 2021).
From disordered carbon, we detect a graphitic precursor and a rare kerogenous potential biotic precursor ( Figure 6); as such, one would expect two types of impact diamonds derived from high pressure alteration of either. However, given the low abundance of shocked phases discovered, in agreement with earlier studies (Kenny et al., 2019;Osinski et al., 2020), impact diamonds formed after kerogen within the Stac Fada Member must be exceedingly rare.

The Utility of Raman Spectroscopy for Impact Studies
Raman spectroscopy is a very useful tool for the discovery and characterization of impactite deposits (i.e., lithologies which contain a shock-metamorphosed mineralogy) given the technique is sensitive to shocked polymorph phases where a structural rather than a compositional change has occurred. Examples of shock minerals identified with Raman in the literature include reidite, the high pressure polymorph of zircon Wittmann et al., 2006;Zamyatin, 2022); maskelynite, a shock product from plagioclase (Unsalan & Altunayar-Unsalan, 2020); coesite, a high pressure polymorph of quartz (Glass & Fries, 2008;Yin et al., 2021); as well as impact diamonds, as shown in this study and previous literature (Kenkmann et al., 2005;Lapke et al., 2000). Raman bands are sensitive to the deformation of mineral phases due to the high temperature-pressure impact conditions, generally causing a widening of the peak FWHM (Gucsik, Zhang, et al., 2004;Zamyatin, 2022). Within shocked zircons, Raman hyperspectral mapping has proven useful to assess the spatial variation in shock damage to visualize heterogeneous crystal lattice orientations, metamictization, and high-pressure mineral phase inclusions (Zamyatin, 2022). Evidence for shock has also been identified within known impact craters in minerals such as gypsum (Brolly et al., 2017) and feldspar (Pickersgill et al., 2021;Xie et al., 2020). The latter can be used to estimate the temperature and pressure of shock events (Yin & Dai, 2020). As shown in this study, disordered carbon also has the potential to record pressure-temperature excursions associated with impact events and if properly calibrated, it could be a useful tool as a geothermometer (e.g., Beyssac et al., 2002;Kouketsu et al., 2014;Lahfid et al., 2010). Thus, in addition to discovering and characterizing shocked minerals, Raman spectroscopy can also be used to quantitatively assess the shock conditions that occurred during impact events.
The instrumentation for Raman spectroscopy is relatively cheap, has the potential to collect data with little-to-no sample preparation, and is effectively nondestructive if run at low laser power. As such, it is a valuable tool for surveying geological samples. These same arguments are applicable to the inclusion of Raman on the payloads of planetary exploration robots, capitalizing on the advantage of quick and unambiguous phase identification for a broad spectrum of silicate minerals and organics (Qu et al., 2021). For example, Raman spectroscopy is included as the SHERLOC instrument on the NASA Perseverance rover (Bhartia et al., 2021). Given that the surface of Mars and other airless planetary bodies have a significant cover of surface regolith generated by impact gardening (Cao et al., 2022;McKay et al., 1991;Szalay et al., 2019), developing an understanding of impact processes via Raman spectroscopy of impact sites on Earth is an important analogue for rover or lander-based investigations carried out elsewhere in the Solar System (Wang et al., 1995;Xie et al., 2021). This is in addition to an ever-increasing application of Raman spectroscopy for the characterization of possible biosignatures (e.g., Jorge Villar & Edwards, 2006).

Implications for the Deposition of the Stac Fada Member
The existence of impact diamonds in gneiss cobbles within the rafted sediment embedded in the Stac Fada Member (Bay of Stoer) (Figure 5a,b) suggests that shocked-modified gneiss clasts were excavated and ejected ballistically into soft sediment during the initial excavation stage of the impact. The impact process also triggered an impactoclastic current that ripped up and transported sandstone rafts (Amor et al., 2019;Simms, 2015). We infer that the gneiss cobbles within the sandstone directly underlying the Stac Fada Member exposure at Second Coast (Figure 5c,d) were also emplaced ballistically, given their angularity, the presence of impact diamond, and their exclusive deposition in the horizon directly underlying the erosive base of the Stac Fada impactite (Figure 5c). Unlike the mechanism proposed by Branney and Brown (2011) that the Stac Fada Member impactoclastic density current was derived from the impact ejecta curtain, our interpretations align with a twostage mechanism. Basement material was first excavated and thrown out ballistically by a hypervelocity impact before being overlain by an impactoclastic density current initiated afterwards, likely from the collapse of a hot impact plume. This mechanism has been proposed for other melt-rich impact breccias (commonly known as suevites), notably to explain the high abundance of melt clast typically sourced from the center of hypervelocity impact craters (e.g., St€ offler et al., 2013). Compared with other terrestrial melt-rich impact breccias, a higher volumetric proportion of melt paired with the limited abundance of shock indicator minerals within the Stac Fada Member has been used to infer the target rocks were rich in volatiles . Compositionally immature arkose sediments identified in this study suggest that sediments for the lower Stoer Group were sourced from an arid environment (see also Young (1999), Stephen and Hambrey (1996)). A volatile-rich target lithology could be possible if (i) the timing of the impact corresponded to a flooding event, (ii) the impactor hit some form of surface water, or (iii) the volatile-rich sediments are not represented by modern-day exposuresa potential explanation for differing compositional maturity within the Stac Fada Member compared to rest of the Stoer Group. To further decipher whether the Stac Fada Member represents a single-or double-ejecta layer deposit Simms, 2015), a better understanding of the proximity of present-day exposures to the crater is still needed to test the comparative lack of lithic breccias associated with ballistic sedimentation observed in most other terrestrial craters. However, given that material ejected ballistically from an impact has a maximum range that scales with the size of the impact crater (e.g., Housen & Holsapple, 2011), it may now be possible to use the size and distribution of diamondbearing ballistic Lewisian gneiss clasts to estimate a lateral distance of modern-day Stac Fada Member exposures from the crater.

CONCLUSIONS
This study adds to the inventory of high-pressure minerals present within the Stac Fada Member impact deposit, with the discovery of impact diamond and tentative evidence for the presence of coesite, a shock polymorph of quartz. Raman spectroscopy characterization of disordered carbon phases suggests an ultimate origin for diamonds from graphitic material derived from the underlying Lewisian basement, and much rarer kerogenous material. These carbon-rich phases have been altered within the Stac Fada Member to smaller Raman shift and broader peak widths, suggesting that thermal modification and a reduction in grain size was more significant than modification from impactrelated high pressures. The similarity between disordered carbon phases in the Stac Fada Member and in the underlying sandstones suggests that the impact breccia reworked and included underlying material from the Stoer Group and basement Lewisian gneiss. This provenance is further confirmed by XRD mineralogy data which shows similar mineral phases. However, a high ratio of quartz to albite compared to underlying Stoer Group lithologies suggests additional sedimentary processing is required within the Stac Fada Member impactoclastic density current to account for increased compositional maturity, in addition to the alteration of precursor carbon phases. The presence of impact diamond-bearing Lewisian gneiss clasts in sandstones embedded in, and directly underlying, the Stac Fada Member suggests that the impactoclastic deposition of Stac Fada occurred after ballistic deposition of the ejecta. Deposition of the Stac Fada Member may have been triggered by the collapse of the impact plume, similarly to mechanisms proposed for other melt-bearing impact breccias.