Towards a kerogen-to-graphite kinetic model by means of Raman spectroscopy

Raman spectroscopy on carbonaceous material has become the most used geothermometer in Earth science studies. However, its application in very different settings, associated with different heating rates, demands an updated review to understand the different paths of maturation that can occur in the geological record. A comparison between organic matter matured under slow (diagenesis and regional metamorphism) or fast (around shallow intrusion or under artificial pyrolysis) heating rates highlight a mismatch between Raman spectra at the same thermal maturity. Such differences are probably due to the interplay of kinetics and strain, and highlights that classical kinetics based on vitrinite reflectance is not appropriated when Raman parameters are used. In the controversial application to faults and shear zones a review of existing literature indicates that a distinction is needed between strain and frictional heating effects since they lead to different spectra evolution. The effect of strain enhances organic matter aromaticity, while data from frictional heating experiments show strong analogies with charcoal spectra even if the kinetics of the process still need to be understood. All this evidence emphasizes that Raman spectroscopy is a powerful tool to describe the aromatization experienced by carbon material in most of the natural conditions and is a good candidate for the development of a universal geothermometer based on a new kinetic model for Earth and planetary sciences.


Introduction
Carbonaceous material (CM) dispersed in rocks derives mostly from the diagenetic alteration of the remains of organisms incorporated in sedimentary rocks during progressive burial. The residue left after the generation and expulsion of hydrocarbons (i.e., kerogen) becomes more enriched in carbon and rearranges into more condensed compounds, via the so-called aromatization process, progressing toward high temperatures. The increase of aromatization in CM is probably the most sensitive geological process to temperature variations and, therefore, has been widely applied as a geothermometer. Organic matter-based geothermometry is generally applied to studies focussed on the thermal state of the crust during diagenesis and metamorphism (Allen and Allen, 2013;Beyssac and Rumble, 2014) or to geologically fast heating (from here onwards called as 'transient heating') processes such as frictional heating on faults planes during earthquakes (Kitamura et al., 2012;Sheppard et al., 2015;Rabinowitz et al., 2020), wildfires (Marynowski and Simoneit, 2009), pyroclastic flows (Scott and Glasspool, 2005) and meteorite impact/atmospheric entry (Amari et al., 1990;Parnell et al., 2005Parnell et al., , 2011. Notably, graphite can also form by precipitation from volcanic organic-rich hydrothermal fluids (Lindgren and Parnell, 2006) or by metamorphic fluids derived from mineral reactions (Rumble, 2014) including decarbonation in subduction zones (Galvez et al., 2013).
CM Geothermometry has utilized methods such as elemental analysis, Rock-Eval Pyrolysis and vitrinite reflectance since the middle of the last century. However, in the last thirty years Raman spectroscopy has become a promising tool for the analysis of CM since it provides great accuracy at the microscale, avoiding time-consuming preparation.
Diagenetic/metamorphic and frictional heating trends in Raman data have been already reviewed by Henry et al. (2019a), but in light of Fig. 1. Diagram showing metamorphic zones, coal ranks, vitrinite reflectance and corresponding Raman spectra. Correlation with metamorphic ranks and vitrinite reflectance is from Hunt (1995). Correlation of Raman spectra and deconvolution are after Schito et al., 2017, Lahfid et al., 2010and Beyssac et al., 2002. Bands nomenclature is after Henry et al. (2019a). Abbreviations: Sub. Coals-subituminous coals; HVBChigh volatile bituminous coals; MVBCmedium volatile bituminous coals; LVBMlow volatiles bituminous coals: R omax: maximum vitrinite reflectance under oil immersion. the exceptional publishing rate on this topic we saw the need of a further review, to compare: 1) the coalification and graphitization trends under different thermal conditions (i.e., diagenesis/regional metamorphism versus contact metamorphism or pyrolysis) and 2) the CM aromatization degree under transient heating conditions that can be found in nature (i. e., wildfire or frictional heating). The main aim of this work is to suggest how a Raman-based kinetic model for CM can be derived in the future, applicable for most natural heating conditions.

Raman spectrum and fitting approaches
The Raman spectrum of CM comprises the two main D (disorder) and G (graphite) bands and other minor bands in the so-called first-order region between 1000 and 2000 cm − 1 (Fig. 1). Other bands in the secondorder region between 2900 and 3200 cm − 1 become visible only through graphitization. The G band is related to the E 2g symmetry in-plane vibration of carbon atoms in graphene sheets and is active in crystalline graphite at 1582 cm − 1 (Tuinstra and Koenig, 1970;Reich and Thomsen, 2004;Pimenta et al., 2007). Two hypotheses on the origin of the D band at 1350 cm − 1 , active in disordered graphite, relate it to a double resonant Raman scattering process (Pócsik et al., 1998;Reich and Thomsen, 2004;Pimenta et al., 2007), or alternatively to the ring-breathing vibration in the graphite subunit or polycyclic aromatic compounds (PAHs) (Negri et al., 2004;Castiglioni et al., 2004;Lünsdorf, 2016).
The strong signal of these bands is due to the resonant effect between the aromatic compounds in CM and the wavelength of the most used excitation lasers (Ferrari and Robertson, 2000). This effect, however, implies that other non-resonant species appear at much lower intensities (up to 10 − 5 ) and are totally overwhelmed by the resonant species (Marshall et al., 2010). Therefore, Raman technology on CM is mainly insensitive to aliphatic or oxygenated groups that other techniques, like Fourier Transform Infrared Spectroscopy (FT-IR) or Nuclear Magnetic Resonance (NMR), can detect.
The main changes in Raman spectrum in the low temperature zone (i. e., coalification) are related to the shift in the D band position (Lünsdorf, 2016;Schito et al., 2017) due to the conversion of the conjugated linear PAHs into more compact (i.e., circular) structures . Also, the minor bands that lie between 1100 and 1300 cm − 1 and between 1400 and 1500 cm − 1 disappear, possibly because of a decrease of methyl groups in alkane chains (Ferralis et al., 2016;Hackley and Lünsdorf, 2018). In metamorphosed samples, the progressive disappearance of the D bands and the shift of G band toward its graphitic position at 1582 cm − 1 define the path of graphitization (Beyssac et al., 2002).
Parameters determined after spectrum deconvolution have been used in early works to determine the metamorphic grade (Pasteris and Wopenka, 1991;Wopenka and Pasteris, 1993;Spötl et al., 1998) and finally formalized as RSCM geothermometer by Beyssac et al. (2002) in the paleo-temperatures range between 330 and 650 • C. The RCSM method was later extended to the lower temperatures of metamorphism by Lahfid et al. (2010) and Lünsdorf et al. (2017) and it is now widely used as a thermal maturity tool in diagenesis (Guedes et al., 2010;Liu et al., 2013;Wilkins et al., 2014;Schito et al., 2017;Schito and Corrado, 2020;Henry et al., 2019a. The most recent review, compiled by Henry et al. (2019a), demonstrates that the Raman parameters that better correlate against thermal maturity and/or metamorphic paleo-temperatures are: distance between the D and the G bands (Raman Band Separation -RBS); the full width at maximum height of the G band (FWMH-G); the ratios of the D and G bands full width at maximum height (FWMH-D/FWMH-G), area (aD/aG) and peak intensities (R1). Parameters derived from the area ratio between the bands beneath the region around the D and the G bands are also generally used (RA2, R2, Fig. 1) in metamorphism (Beyssac et al., 2002: Lahfid et al., 2010 and in diagenesis (Schito et al., 2017).
Different fitting approaches can lead to misleading interpretation when they are based on a single parameter (e.g., RBS, R1, etc.). Thus, for the purpose of this work, spectra will be shown together with trend parameters. Spectra have been redrawn from published works and normalized with respect to the height of the G band.

Methods and heating regimes
Carbon is preserved in the geological record in its reduced form as the product of diagenetic and metamorphic transformations during burial (i.e., kerogen and graphite), charring (charcoals) and hydrothermal precipitation. Excluding the latter, the kinetics of such transformations is mainly driven by temperature and time while pressure starts to affect only above the diagenesis -low metamorphism boundary. The evolution of temperature through time varies according to different geological processes. It can be constant for million years during diagenesis and regional metamorphism, or for hundreds or a few thousands of years around an intrusion in the shallow crust. On the other hand, there are a variety of processes in which heating can be considered as instantaneous (from a geological point of view), lasting from days in the case of wildfires (Theurer et al., 2021) up to few seconds during seismic frictional heating (Kuo et al., 2017) on fault planes or due to the thermal shock of meteorite impact (Brolly et al., 2017).
The characteristics of intrusions, host rocks, range of Ro% and temperatures and fitting approach for Raman spectra used by different authors are summarized in Table 1.

Artificial maturation
Artificial maturation experiments can be classified on the basis of pyrolysis approaches used by the authors (summarized in Table 2). Kelemen and Fang (2001) first provided both isothermal and nonisothermal artificial pyrolysis experiments using an ultrahigh vacuum on a type I and II kerogen. Hackley and Lünsdorf (2018) and Khatibi et al. (2019) explored the low-maturity stages of solid bitumen, heating their samples with hydrous pyrolysis (HP) according to the method of Lewan (1983) in a temperature range between 300 and 360 • C, where the maximum temperatures were kept for 72 h and the pressure was lower than 20 MPa. The same system was used by Birdwell et al. (2021), Hackley et al. (2022) and Sanders et al. (2022).

Table 1
Intrusions' and host rocks' properties together with Ro% and temperature ranges and fitting approach used by different authors dealing with CM Raman evolution around intrusions. n.d. -not determined. Ro% and T ( • C) ranges include minimum and maximum.

Table 2
Experimental condition and organic matter type (including single macerals when specified) for artificial pyrolysis experiments. *For IFORS method see Lünsdorf and Lünsdorf (2016). ** Zhou et al. (2014) do not refer to number of bands but simply refer to Beyssac et al. (2003) that use 4 to 2 bands depending on the coal rank. In Ro %/BRo%/Rg% range column bitumen reflectance values (Bro%) are italicised and graptolite reflectance (Rg%) underlined.  Bonoldi et al. (2016) use an open-hydrous system to maturate type I, II and III kerogen samples to temperatures up to 360 • C under a pressure of 30 MPa. Du et al. (2014) compare parameters from three different pyrolysis methods: semi-open pyrolysis, and closed systems with a gold tube or glass tube. In the semi-open system, volatile pyrolyzates derived from primary cracking are removed from the reaction system at each pyrolysis step and then organic material is re-heated (Takahashi and Suzuki, 2017). Samples for semi-open pyrolysis are muddy source rocks representing a type II kerogen heated between 280 and 560 • C for 72 h under a constant pressure of 80 MPa. In the closed system, the authors study isolated kerogen in a temperature interval between 280 and 480 • C for 72 h under a constant pressure of 45 MPa using a gold tube, while closed glass tube system experiments are performed between 320 and 500 All experimental settings are summarized in Table 2.

CM maturation due to frictional heating or strain effects
Many authors have tried to quantify the frictional heating on a fault plane (Hirono et al., 2015;Ito et al., 2017;Kouketsu et al., 2017;Kuo et al., 2018aKuo et al., , 2018bNakamura et al., 2015) or the effect of the additional strain in aseismic shear zones on CM ordering (Kedar et al., 2020Muirhead et al., 2021). In both cases, similar approaches were utilized, analysing samples in the less deformed host rocks, moving progressively toward subzones of increasing deformation (e.g., damage zone, breccia zone and gouge zone). Samples' preparation include thin sections, concentrated CM and clay-size residue deposited on glass slides. Some authors perform artificial maturation experiment for comparison with natural cases. Experiments include pure (isothermal) friction, transient heating experiments or the application of both . Pure friction experiments (Furuichi et al., 2015;Hirono et al., 2015;Kuo et al., 2017;Kirilova et al., 2018;Fan et al., 2020) are performed using a rotary shear apparatus under room temperature conditions, and normal stresses that vary between 1 and 3 MPa (Hirono et al., 2015;, 2.7 and 13 MPa (Ito et al., 2017;Kuo et al., 2017) or 20-60 MPa (Fan et al., 2020) (Table 3).
Those authors who aim to reproduce the heating (Table 3) induced on the fault slip plane during earthquakes use a combination of heating rates (10 • C/min up to 100 • C/s). Even if the authors are aware that the typical heating rate on a slip plane can be tens to hundreds of • C per seconds (Kaneki et al., 2016), most of the experiments are performed between 10 and 50 • C/min (Hirono et al., 2015; and only  perform experiments at a maximum of 100 • C/s. At the same conditions, Ito et al. (2017) were able to artificially reproduce pseudotachylytes under high normal stresses, reaching temperatures between 850 and 1100 • C (as Table 3 Experimental conditions for frictional and transient heating.

Authors
Charring methods Starting materials Fitting approach (n of bands) determined by the evidence of illite dehydration and decomposition), while Kuo et al. (2014Kuo et al. ( , 2017 demonstrate, by high slipe rate experiments, that high temperature along slip zone could facilitate the formation of graphite whose presence can be consider as a seismic indicator. Notably, in rotary shear experiments, temperatures have been always measured by means of thermocouple. However, this method can have large uncertainty as outlined by a recent work from Aretusini et al. (2021) that show the use of optical fibres significantly improve the accuracy. Unfortunately, such technology has not been yet used in experimental sheared CM. The relatively low heating rate experiments are performed using a calorimeter apparatus under Ar gas flow in order to remove the products of pyrolysis (i.e., semi-open system), while for fast heating experiment a tube furnace is used under vacuum condition to replicate anoxic condition of faults at depth.
Morga (2011) perform high-temperature heating experiments to investigate the reactivity of coals rich in semifusinite and fusinite. These macerals, similarly to vitrinite, derived from continental plant with the difference that they suffered oxidation due to alteration in the atmosphere (dehydration and weathering), burning, fungal and bacterial decomposition (among others, see Coal I.C., 2001) before to be embedded in the sediments.
Samples from Schito et al. (2022) derive from charred plant material entombed in pyroclastic sediments and the maximum temperatures experienced are calculated from charcoal reflectance (Scott and Glasspool, 2005;Pensa et al., 2018) or from partial thermal remanent magnetization (pTRM; Pensa et al., 2015).

Comparing natural and artificial maturation
The main issue with artificial maturation is to verify if Raman spectra show the same characteristics with naturally matured samples at the same thermal maturity. This question has been addressed by Kelemen and Fang (2001) RBS is the most used parameter (Fig. 2a), while the FWHM-G was presented only in one study and was thus discarded in Fig. 2. Kelemen and Fang (2001), Zhou et al. (2014), Hackley and Lünsdorf (2018) and Hao et al. (2019) compare the width and intensity ratios (Fig. 2 b,d), while Kelemen and Fang (2001), Hackley and Lünsdorf (2018) and Hao et al. (2019) use the area ratio of the D and G bands (aD/aG parameter) that is introduced in Fig. 2c. The first works that compare natural and artificial maturation in kerogen and solid bitumen (Kelemen and Fang, 2001;Zhou et al., 2014;Bonoldi et al., 2016) claim no systematic differences (Fig. 2 a, Zhou et al., 2014;Bonoldi et al., 2016) exists or, alternatively, that differences can be detected by the RBS parameters but not by the others (Fig. 2b,

Trends against Ro%
Artificial pyrolysis reproduces geologically fast heating at relatively low lithostatic pressure (i.e., low burial) resembling conditions that can be found during source rocks maturation around shallow intrusions, so data were reviewed together in Fig. 3.
As already stressed by Henry et al. (2019a), Raman parameter trends are strongly influenced by fitting approach, laser wavelength and starting material, so that only general trends can be depicted. In detail, the use of different laser wavelengths is known to affect the position of the D band (Quirico et al., 2005;Lünsdorf, 2016) and, consequently, the RBS value. Looking at Tables 1 and 2 and at Henry et al. (2019a) the green (514 to 533 cm − 1 )laser is the most used in literature. Assuming negligible differences between 514 cm − 1 and 533 cm − 1 lasers, however, in Fig. 3, samples from Hackley and Lünsdorf (2018), Sanders et al. (2022), Birdwell et al. (2021) and Kelemen and Fang (2001) should be considered with caution since the authors use a 488 cm − 1 , 473 cm − 1 and 632 cm − 1 lasers respectively. This implies that, taking as a reference measures made by a green laser, RBS is slightly underestimate for Hackley and Lünsdorf (2018), Sanders et al. (2022) and Birdwell et al. (2021) and overestimate in Kelemen and Fang (2001). As well, it should be considered that many works dealt with solid bitumen, whose reflectance (BRo% in Table 2), is usually lower than Ro% at the same thermal maturity. For these samples a slight shift toward higher values on the x-axis should be considered. The response of different macerals to heat-induced aromatization measured by Raman has been issued only by Birdwell et al. (2021). Analysing AOM, solid bitumen, liptodetrinite and vitrinite from the same samples, the authors conclude that different macerals progress along similar trend but can respond to a greater or lesser degree to the same level of thermal energy. From their preliminary data, solid bitumen seems to react faster to heating then vitrinite and AOM but more data are still needed to confirm this conclusion.
In Fig. 3, artificial pyrolysis and samples matured around shallow intrusions have been plotted together and compared with trends already depicted by Henry et al. (2019a) for samples matured in diagenesis and under regional metamorphism. New trends found in this work are marked with dotted arrows while Henry et al.'s (2019a) trends with orange-shaded arrows.
RBS values (Fig. 3a), for artificial maturation and shallow intrusions (triangles and circles) show an increase between about 0.4 to 4.0% Ro that seems to slow down after 2% Ro (Fig. 3a). This trend outlines a net mismatch with respect to the general trend detected by Henry et al., (2019a) for regional metamorphism (see arrows in Fig. 3a). At the highest maturities, the decrease of RBS values after 5% Ro marks the onset of the graphitization process for deeper intrusion samples (squares in Fig. 3a).
Triangles and circles in Fig. 3b mark a constant decrease of FWHM-G values during coalification at high heating rates (dotted arrow) but without the net decrease observed in regional metamorphism as marked by the change in slope of the coloured arrow between 2 and 3-Ro%. The mismatch is further evidenced by the lower values of deep intrusions samples (squares) between about 4 and 6 Ro% (Fig. 3b).
The wD/wG parameter shows an up and down trend with a turning point between 2 and 3% Ro and stable values after about 6% Ro (Fig. 3c). Even if they are not so easily interpreted (question mark in Finally, no significant trends can be depicted for the R1 parameter (Fig. 3d) where values do not significantly change and lie at the lower limit of the orange arrow. An increase of R1 values at the onset of graphitization after 5% Ro was only detected for deep intrusion samples.
The mismatches outlined by the dotted and the yellow lines in Fig. 3 indicates a retarded aromatization trend for intrusion and artificial samples with respect to those matured under regional metamorphism. This is further highlighted by spectra in Figs. 3 e and f.
The mismatch, in particular between shallow intrusions and regional metamorphism/deeper intrusions, suggest differences can be mainly found at the highest thermal maturities. However, data for artificially matured samples in Hackley and Lünsdorf (2018), Khatibi et al. (2019) and Hao et al. (2019) in Fig. 2 show that Raman parameters can start to behave differently also at low thermal maturities, perhaps in the material studied by these authors (i.e. solid bitumen and graptolites). This suggests a kinetic effect on Raman spectra evolution, different from the one governing reflectance increase.
It should be noted that source rocks matured around intrusions experience different heating regimes according to their distance to the magmatic source. According to numerical simulations (Galushkin, 1997;Muirhead et al., 2012;Wang, 2012;Iyer et al., 2018) very high (up to 600 • C) maximum temperatures near the intrusion are maintained for a few months or years, while, at a distance equivalent to the intrusion thickness, they are kept for several hundreds of years. This implies that samples far from the intrusion probably experienced heating rates more similar to diagenetic conditions and that is likely why they better fit with the general trends of Fig. 3.
It should also be considered that in sedimentary basins, reflectance values higher than 3.5% (Fig. 3) generally correspond to very high burial depths where tectonic stress (i.e., pressure) can influence metamorphism of organic matter (Suchy et al., 1997;Barzoi, 2015). It is thus possible that the interplay of heating rate and pressure conditions could be the cause of the mismatch between the two trends.
In general, vitrinite reflectance and Raman parameters are related but they don't measure the same things. Reflectance is known to increase as a function of the refractive and absorption indexes that are related on the amount of delocalized electrons (Van Krevelen and Te Nijenhuis, 2009), while Raman spectrum directly give information on the aromaticity of the material. Carr and Williamson (1990) already show that a non-linear correlation between reflectance and aromaticity (measured by means of NMR) exist in kerogen, because the two methods respond differently to structural changes occurring during maturation in diagenesis and low metamorphism. Different rates were particularly observed for Ro% higher than 3% where the growth of aromatic sheets increase reflectance anisotropy (Carr and Williamson, 1990).
Results from the present review, suggest that under high heating rates and low tectonic stress the rate of aromaticity increase detected by Raman is lower than that under low metamorphic condition at the same vitrinite reflectance (i.e., thermal maturity).
Moreover, to understand the decoupling at lowest maturities, Hackley and Lünsdorf (2018) proposed that the retardation could be attributed to a greater abundance of methyl-substituted aliphatic hydrocarbons that probably hamper the growth of aromatic clusters in pyrolyzed samples. Whatever the cause, data shown in Figs. 2 and 3, indicate that samples with the same Ro% values, but matured under different conditions, have different Raman spectra. This confirm that the molecular changes that drive Ro% increase are not the same and suggest their correlation vary according to different heating conditions.

Trends against Temperature
New trends found in Fig. 3 against Ro% can be also converted into paleotemperatures (Fig. 4) assuming the shift from Raman "geothermometer" to a "thermal maturity indicator" when moving from regional metamorphism toward shallow intrusions or artificial maturation. Regarding shallow intrusions, this issue has been faced only by few authors (Chen et al., 2017;Mori et al., 2017;Li et al., 2020) with still some problems, as outlined below.
Paleotemperatures around intrusions can be calculated by using different approaches. For studies on deep plutons, a thermal model calibrated by means of mineral assemblage in the host rock is probably the best approach (Aoya et al., 2010;Hilchie and Jamieson, 2014). In shaly formations, or in coals around shallow intrusions, reflectance measurements are mainly used. In general, few attempts (Chen et al., 2017;Li et al., 2020) have been reported in the literature to provide a correlation between Raman parameters and temperatures around shallow intrusions, and they both use empirical correlations to convert Ro% values into temperatures. Li et al. (2020) converted their data by means of the equation of Barker and Pawlewicz (1994), obtaining values between 80 and 350 • C, while the reflectance values of Chen et al. (2017) are converted by means of the Bostick and Pawlewicz (1984) equation and range between about 130 and 500 • C.
However, it is advisable to use Ro% data to calibrate a thermal model, rather than directly converting them into temperature by means of empirical equations (Galushkin, 1997;Aarnes et al., 2011;Muirhead et al., 2012;Wang, 2012;Iyer et al., 2013Iyer et al., , 2017Mori et al., 2017) since, as outlined by Wang et al. (2008), Ro% conversion using the Barker and Pawlewicz (1994) equation in contact metamorphism can lead to underestimation by >200 • C near the contact with the intrusion.
Pyrolysis temperatures can also be compared since the residence time of most of the experiments is similar (72 h; Fig. 4). Artificial maturation is generally performed between 300 and 600 • C while some work (Zhou et al., 2014) extends the range up to 750 • C. Fig. 4a shows that the RBS parameter always shows a, more or less, linear increase with temperature (dotted arrow). Hydrous pyrolysis (Hackley and Lünsdorf, 2018;Khatibi et al., 2019;Hackley et al., 2022;Sanders et al., 2022) follows a similar trend for data of Li et al. (2020) (assuming their temperatures are underestimated). Even if based on fewer data, Figs. 4b,d shows that the "fast heating" trend (dotted arrows) roughly follow the same trend depicted by Henry et al. (2019a) but shifted toward higher temperatures. Moreover, Fig. 4e clearly shows the difference in R2 pattern for deep (Aoya et al., 2010;Beyssac et al., 2019) and shallow intrusions (Chen et al., 2017;Mori et al., 2017).
Comparing the new trends highlighted in Fig. 4 a-d with the general trend for coalification/graphitization (respectively orange and blue arrows), we suggest that pyrolyzed and shallow intrusion CM follow only the coalification trend and thus never achieve proper graphitization. Given this, the R2 parameter can be misleading if not used only to describe graphitization (Fig. 4 e).
Notably, especially in contact metamorphism, graphite precipitation from carbon-rich hydrothermal fluids (see Rumble, 2014 for a review) is common. This process, however, is not directly related to the metamorphic maximum temperatures rather to the redox redox state, pressure and temperature conditions of the fluids (Huizenga, 2011) and has thus not been considered in this review. Even if mainly observed at high metamorphic ranks (Luque and Rodas, 1999;Luque et al., 2009), fluid-precipitate-graphite at the contact with shallow intrusion has also been reported (Lindgren and Parnell, 2006).

Raman shift in faulted rocks: Difference between seismic and aseismic slip
Among all the uses of the Raman spectroscopy geothermometer, the application to faults and shear zones is the most controversial, and there is still an open debate on the impact on Raman spectral shift .
The researchers mentioned in Section 3.4 all work on low maturity kerogen or graphitic carbon, while Nakamura et al. (2015) and Kirilova et al. (2018) study the Raman ordering of graphite from cataclasite to slip plane, and in frictional experiments respectively.
These works evaluate different Raman parameters, but the most used are the R1 ratio and RBS for natural heating (Fig. 5 a,b) and experimental frictional heating (Fig. 5 c,d).  Hirono et al. (2015) and Mukoyoshi et al. (2018) work on low mature kerogen on both sides of main thrusts, the first in Chelungpu Fault (Taiwan) and the second in the Emi thrust and Kure out-of-sequence thrust (OST) in Japan. Hirono et al. (2015), find a slight increase in the R1 values and more marked increase in the RBS values (Fig. 5 a,b) toward the slip zone, and also after artificial shearing. An increase of the R1 ratio (Fig. 5 a) is also observed on graphitic carbon (i.e., carbon that suffered at least greenschist metamorphism) from Monti Romani (Italy) and Internal Rif (Morocco) by Muirhead et al. (2021) and by Ito et al. (2017). Working on zones of concentrated aseismic strain, other authors (Kuo et al., 2017 in low slip rate experiments; Kuo et al., 2018aKuo et al., , 2018bKedar et al., 2020Kedar et al., , 2021Muirhead et al., 2021) find a decrease of the R1 values (Fig. 5 a).
Apart from the uncertainties arising from the use of different Raman parameters, different fitting approaches (laser is mainly 514-532 cm − 1 , see Table 3) or different preparation (thin section, CM concentration, clay-size deposited gouges on glass slides), most of the confusion is because authors deal with different carbonaceous material precursors (i. e., amorphous carbon, graphitic carbon or graphite) and usually do not discriminate between the effect of strain or frictional heating. Much work to date has focussed on understanding the increase of temperature on fault planes, with some experiments trying to reproduce the process (Hirono et al., 2015;Furuichi et al., 2015;Kuo et al., 2017;Ito et al., 2017, Mukoyoshi et al., 2018. Nevertheless, frictional and transient heating experiments should be compared only with samples collected on the slip plane (Ito et al., 2017) while enhanced maturation within the fault's damage zone (sensu lato) should be ascribed to the effect of strain, fluid circulation or the interplay of both . This is the case of the Raman shift of amorphous carbon (Fig. 5), ascribed to aseismic shear strain by Kedar et al. (2020Kedar et al. ( , 2021 and Muirhead et al. (2021) and possibly in the damage, breccia and gouge zones of Hirono et al. (2015). From Fig. 5 the evolution of Raman parameters in these works is not straightforward, showing a simultaneous R1 low shift and RBS high shift ( Fig. 5; Kedar et al., 2020;Muirhead et al., 2021). However, the R1 parameter in low diagenesis is known to have an uncertain trend (Schito et al., 2017;Schito and Corrado, 2020;Henry et al., 2019a), so it cannot definitively point to a thermal maturity increase or decrease. On the other hand, R1 and RBS increases observed in both natural and experimental samples strongly suggest enhanced maturation due to the effect of strain (Fig. 5).
On graphitic carbon in low-grade metamorphic conditions (Fig. 5), R1 increases in natural samples , suggesting a strain-related ordering, while in Kuo et al. (2017Kuo et al. ( , 2018aKuo et al. ( , 2018b simultaneous R1 decrease, and RBS increase reflect a similar behaviour of amorphous carbon described above. In graphite, on the other hand, R1 increases in both natural and experimental examples, suggesting an ordering decrease (Fig. 5;Nakamura et al., 2015;Kirilova et al., 2018) likely due to mechanical disruption of the graphite lattice (Kirilova    et al., 2018). It is worth noting that in metamorphism, the R1 parameter first increases at greenschist facies and then decreases up to the graphite stage as shown by the blue arrow in Fig. 5. This behaviour is probably the cause of the apparent discrepancies in R1 in graphitic carbon (Fig. 5, Kuo et al., 2017Kuo et al., , 2018aKuo et al., , 2018bMuirhead et al., 2021). Fig. 5, highlights that there are instances in the literature of strainrelated induced maturation in both amorphous and graphitic carbon. Of particular interest is the work of Fan et al. (2020) that show the shift of the RBS parameter because of the increase of effective normal stress in friction isothermal experiments (Fig. 5 c,d) as already outlined by Maslova et al. (2012). This also seems to confirm the role of strain in the mismatch between spectra at the same maturity observed in Figs. 2 and 3.
Looking at the Raman shift in experiments, it can be observed that frictional heating by rotary shear, generally yield higher R1 values after shearing (Fig. 5 c;Furuichi et al., 2015;Hirono et al., 2015;Ito et a., 2017;Fan et al., 2020), except in the work of Kuo et al. (2017). Also, an increase in the RBS values (Fig. 5 a) is shown in shear experiments except from Hirono et al. (2015). On the other hand, the increase of the R1 and RBS values reported for graphite (Fig. 5 c,d) show how both brittle deformation in fault zones (Nakamura et al., 2015) or shearing at experimental sliding velocities reduce the ordering of graphite. The authors thus suggest caution when using the RSCM geothermometers in active tectonic settings.
Almost all the authors that performed frictional and/or transient heating experiment to calculate temperatures (Fig. 6), only the R1 and aG/aD parameters are used. As shown in Fig. 6a, R1 and aD/aG values remain mainly constant at low temperature, starting to increase at about 400 • C for low heating rates and after 800 • C heating rates higher than 1 • C/s. As well, the aD/aG parameter seems to only slightly increase up to 600 • C while between 600 and 1000 • C it shows a rapid increase and decrease for higher temperatures (Fig. 6 b). These trends have been already identified by Henry et al. (2019a) (coloured arrows in Fig. 6 a,  b).
Looking at Fig. 6, some interesting issues raise. First of all, a flat initial R1 ratio is observed, similarly to what observed in diagenesis/ metamorphism but at lower temperatures. In addition, the R1 parameter starts to raise at progressively higher temperatures depending on the heating rate conditions: samples from Hirono et al. (2015) burned at 10 • C/min (red and purple squares) start to increase at lower temperatures than samples burned at 50 • C/min (about 1 • C/s) from diamonds) and orange circles) and samples heated at 100 • C/s ; blue circles) only slightly increase after 900 • C. Ito et al. (2017) do not specify their heating rate but state they achieve maximum temperatures of >1000 • C in a few seconds to properly simulate frictional heating during earthquakes (Kaneki et al., 2016).
Accordingly, the enhanced maturation due to seismic frictional heating implies very high temperatures reached in a very short amount of time and is thus comparable to a flash pyrolysis (Muirhead et al., 2012;Ito et al., 2017;Aubry et al., 2018) leading to the trends and spectra shape of Fig. 6. Spectra drawn in Fig. 6 c, in particular, show high intensities in the valley between the D and G bands that start to drop after 1000 • C (Fig. 6 c). These spectra are different from those show in Figs. 2 and 3 suggesting a very different evolution path. The most outstanding evidence from these experiments is the apparent "disordering" shift in spectra in Fig. 6e. Here, spectra that already achieved a low-metamorphic ordering seem to transform into very disordered ones at high temperatures. The use of the term ordering is thus here misleading since the Raman shift is showing a "jump" from a diagenetic/ metamorphic path up to another at higher heating rates.

Charcoal Raman spectra
The application to charcoal geothermometry is probably the most recent advance, and has shown attractive potential application in archaeological, paleo-environmental and geological studies (Deldicque and Rouzaud, 2020;Mauquoy et al., 2020;Theurer et al., 2021Theurer et al., , 2022Schito et al., 2022) as well as for industrial application (Surup et al., 2019). Such large spectrum of applications results in a wide range of temperatures used in calibration experiments, up to almost 2000 • C.
Most of the results shown in Fig. 7 derive from artificial heating (except Schito et al., 2022).
As shown by the dotted arrow in Fig. 7a, the RBS parameter shows an increase from about 200 to 1100 • C and a further increase is also observed at higher temperatures by Surup et al. (2019), Zickler et al. (2006) and Urban et al. (2003). It is worth noting that Yamauchi and Kurimoto (2003) performed their Raman experiments using a red (632 cm − 1 ) laser and a power output of 220 mW. As discusses above, this imply an overestimation of D band related parameter (Kouketsu et al., 2014;Lünsdorf, 2016) and thus imply RBS shift toward higher values with respect to data from all the other authors that use a green (514 or 532 cm − 1 ) laser. Moreover, thermal degradation due to the high laser intensity for low temperature samples cannot be excluded, as evidenced by the work of Henry et al. (2018).
The FWHM-G parameter also shows a continue decrease over the whole temperatures range (Fig. 7b), while an increase of the wD/wG values up to 800 • C was observed by Yamauchi and Kurimoto (2003) and Schito et al. (2022) that is the opposite of what Paris et al. (2005) and Theurer et al. (2021) observed (Fig. 7c). Finally, the R1 values show an increase only from 500 to 1200 • C, followed by a decrease up to 1800 • C.
The "transient heating" Raman parameters trends (dotted arrows) in Figs. 6 and 7 can sometimes resemble those from Figs. 2 a-d and 3 a-b but this can be highly misleading since spectra evolution is entirely different as shown in Figs. 2 e, 3 c, 6c and 7e.
Spectra shown in Fig. 7e point out that only low-temperature (perhaps below 600 • C) spectra are similar to those observed in diagenetic samples (Fig. 2) or in pyrolysis experiments (see Fig. 3). Conversely, high temperatures spectra are all characterized by high intensities in the valley between the D and G bands and mostly resemble those ones from Fig. 6. This is because of similar experimental conditions. Accordingly, in these transient heating trend (charcoalification) an increase of the D band intensity with respect to the G band is observed, up the R1 ratio reaches a value of about 1 at 1200 • C (Fig. 7 d,  e). After that, a decrease of the valley intensities, after perhaps 1600 • C, led to spectra's shape similar to those of greenschist facies. Thermal graphitization is never achieved in these experiments but it is known from literature to be generally achieved at temperatures higher than 2200 • C (Bonijoly et al., 1982).

Kinetic effects and future perspectives
The mismatches of Raman parameters against Ro% discussed in Section 4.1 indicate that these analytical approaches measure different processes. Both measures correlate to the increase of aromatic organic matter at increasing temperature, but at non-linear rates as already outlined by Carr and Williamson (1990). This review outlines that differences in maturation rates are enhanced when compared different heating regimes and that at increasing heating rates aromaticity increase proceed at a slower rate than vitrinite reflectance increase (Figs. 2 and  3). It is also possible that the spatial higher resolution of the laser spot in Raman spectroscopy is able to detect more precise characteristics of the molecular arrangement then those that lead to the reflectance increase on the surface of the organic material. This implies that, if Raman measurements are to be used in thermal modelling, the use of Ro% kinetics is not appropriate,and a new model based on Raman parameters needs to be developed. It has been argued that heating rate and strain can influence spectral evolution and thus their effects should be included in the kinetic model.
Additionally, in studies to date, the most used kinetic model for organic matter maturation has been the Sweeney and Burnham (1990) model (EASYR O %) that has been recently revised by Burnham (2019) to include models for different kerogen types. Recent alternatives have been also proposed by Nielsen et al. (2017) and Wood (2018). These models allow the user to reproduce an increase of vitrinite reflectance as a function of temperature and time by using an Arrhenius equation and assuming that organic matter maturation can be simply treated by considering four independent parallel reactions that in turn eliminate water and carbon dioxide during early maturation and heavy to light hydrocarbons in the later stages. In the EASYR O % model of Sweeney and Burnham (1990) and Burnham (2019), the Arrhenius equation is based on a single frequency factor and a distribution of activation energies for each one of the parallel reactions. Activation energies rise, up to known values at the metagenetic stage. After that, the turbostratic structure of anthracite starts to form, and in the graphitization process of natural anthracite the same kinetic parameters cannot be used, since lithostatic pressure starts to play a fundamental role in promoting the accumulation of strain energy needed to mechanically reorientate and align the aromatic units until defect-free aromatic sheets of triperiodic graphite form (Bonijoly et al., 1982;Bustin et al., 1995;Lünsdorf, 2016). It has been demonstrated that the role of pressure is to lower the activation energies required for the graphitization process and a kinetic control for graphitization, in which the role of pressure is considered, has been provided by Nakamura et al. (2017Nakamura et al. ( , 2020, but can be applied only in high metamorphism. The next step in this area of research will be to understand when the role of pressure must be included at the boundary between high diagenesis and low metamorphism. Other outcomes on this topic, derive from the experiments made by Huang et al. (2010) using a sapphire anvil cell connected to the Raman apparatus. Their data show a transient effect of differential pressure of Raman spectra that disappear after the stress is released. This is interesting evidence, that, together with the work of Khatibi et al. (2018) on the relation between CM Raman spectrum and the Young's modulus can pave the ground for future studies.
This review has also demonstrated that there are natural conditions of heating, such as charring on fault planes or in wildfires, where the evolution of Raman spectra follow different paths. Such an effect can also be described by means of a kinetic approach. Based on experiments performed mainly in the 1990s, it was observed that artificial graphitization at room temperature requires temperatures higher than 1700 • C (depending on the precursor material) and activation energies too high to be achieved in natural conditions (Ross and Bustin, 1990;Ross et al., 1991). Given this, we can assume that the activation energies necessary to generate the charcoals (spectra shown in Figs. 5e-i and 7 c-g) are higher than those needed to generate graphitic carbon under metamorphic conditions. In this way the temperature-driven "apparent" disordering trend observed in Fig. 6e can be explained and the coexistence of graphitic carbon and charcoal-like carbon could be predicted in a faulted rock. Fig. 8. Schematic Raman spectra evolution as a function of temperature, time and strain. Not in scale. The interrupted arrow in the diagenesis /regional metamorphism path highlight the switch of the time scale from hours-days-years up to million years. The red colour indicate the path that can be modelled with a T-t kinetics, the blue those can be modelled by using a T-t-P kinetics. *Graphitization at room conditions cannot be achieved under natural conditions. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) The evolution of carbonaceous material in natural conditions can be summarized as a function of their temperatures, the effects of time (consequently the heating rate) and those of strain (i.e. pressure). All these factors act on the activation energy and three main paths of Raman evolution can be recognized accordingly (Fig. 8): 1) a low temperature path where coalification and then graphitization are mainly driven by the temperature and the strain during metamorphism that keep low activation energies as demonstrated by Nakamura et al. (2017Nakamura et al. ( , 2020. Note that the boundary at which strain starts to have an effect is still not known but will probably lie at the low metamorphism onset; 2) a similar path but at higher temperatures and on different time scale, that represents coalification around (shallow) intrusions and in artificial pyrolysis 3) a high temperature/high heating rate (also implying high activation energy) path where high temperature charcoals can be formed as a consequence of transient heating as in wildfires or friction along fault slip planes.
The red or blue colours in Fig. 8 indicate the kinetics that would be able to describe the all these paths showing schematically where it could a T-t or a T-t-P based kinetics.
Based on these assumptions, calibrating a kinetic model with Raman parameters would be able to describe almost every condition of carbonaceous material maturation that can be found in natural systems.

Conclusions
A new updated review on the behaviour of Raman spectrum of CM under different maturation conditions highlights the need for a kinetic model for different heating rates that can be found in natural systems. The main conclusions are that: 1) There is a mismatch between Raman spectra at the same maturities between fast and slow heating rates; 2) Such differences are probably due to the interplay of different (other than Ro%) kinetics and pressure; 3) Both strain and frictional heating can affect CM spectra in highly deformed rocks, nevertheless, they need to be distinguished since they follow a different path; 4) Charcoals and artificial frictional heating experiments generate similar spectra. However, the kinetics of the process need to be understood; 5) Building a kinetic model based on Raman spectra would provide a universal geothermometer for CM in geological studies.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability
Data will be made available on request.

7Acknowledgements
This work was funded by the School of Geosciences, University of Aberdeen. Stimulating discussion with Sveva Corrado and Thomas Theurer greatly enriched this work. The Editor Shuhab Khan, the reviewer Aaron Jubb and two anonymous reviewers are kindly acknowledged for their comments that significantly improve the original version of the manuscript.