High microscale variability in Raman thermal maturity estimates from shale organic matter

International Journal of Coal Geology 199 (2018) 1–9

Contents lists available at ScienceDirect

International Journal of Coal Geology journal homepage: www.elsevier.com/locate/coal

High microscale variability in Raman thermal maturity estimates from shale organic matter

T

⁎

Aaron M. Jubba, , Palma J. Botterella, Justin E. Birdwellb, Robert C. Burrussa, Paul C. Hackleya, Brett J. Valentinea, Javin J. Hatcheriana, Stephen A. Wilsonc a

Eastern Energy Resources Science Center, U.S. Geological Survey, Reston, VA 20192, USA Central Energy Resources Science Center, U.S. Geological Survey, Denver, CO 80225, USA c Geology, Geophysics, and Geochemistry Science Center, U.S. Geological Survey, Denver, CO 80225, USA b

A R T I C LE I N FO

A B S T R A C T

Keywords: Organic matter Petroleum source rocks Raman spectroscopy Thermal maturity Vibrational spectroscopy

Raman spectroscopy has recently received attention as a means to estimate thermal maturity of organic matter in petroleum generating source rocks to complement more traditional approaches such as vitrinite reflectance and programmed pyrolysis. While many studies have observed positive correlations between source rock thermal maturity and Raman spectral parameters, little attention has been given to the degree of variation in the Raman response across individual organic grains, especially for shales or mudrocks with highly dispersed organic matter. Here the spatial variation in Raman estimates of thermal maturity within individual organic grains is assessed from shales from the Boquillas, Marcellus, Niobrara, and Woodford Formations. The thermal maturity parameters extracted from Raman spectra can vary widely across distances of ≤5 μm within the same organic grain. These results illustrate the high degree of chemical heterogeneity inherent to the organic matter within these source rocks. Additionally, the spatial pattern of the Raman parameters, as revealed by 2D Raman mapping, imply that organic matter structure is influenced by associations with mineral surfaces within the surrounding rock matrix. Chemical heterogeneity and matrix effects directly impact the Raman response from these types of materials and thus the extracted thermal maturity estimate. These findings highlight the care which must be adopted when making Raman measurements of organic matter within source rock matrices, especially for samples which feature highly dispersed, heterogeneous organic matter as found in petroliferous mudrocks.

1. Introduction The thermal history of preserved organic matter (OM) within shale rocks is one of the key parameters required to evaluate the petroleumgenerating ability of these materials (Dow, 1977). The most common thermal maturity proxies are based on measurements of light (λ = 546 ± 10 nm) reflectance from vitrinite macerals or programmed pyrolysis of a small amount (~100 mg) of powdered rock sample (e.g., Peters, 1986; Hackley and Cardott, 2016). However, there are instances in which these traditional approaches are challenged. In some cases, formations contain source rocks with little or no vitrinite. Examples include pre-Devonian shales deposited prior to the evolution of woody plants (the source of vitrinite) as well as marine formations in which algal and microbial input are the dominate OM source. Similarly, organically lean or high maturity source rocks are difficult to analyze by programmed pyrolysis due to detection limit considerations (Peters, 1986).

⁎

Raman spectroscopy has been used since the late 1980s and early 1990s as a tool to evaluate the thermal history of OM in sedimentary rocks, e.g. (Beny-Bassez and Rouzaud, 1985; Jehlička and Bény, 1992; Wopenka and Pasteris, 1993). In the past 10 years, there has been increased interest in using this approach as a thermal maturity proxy because it is fast, does not rely on the presence of specific organic macerals, requires very little sample, and is potentially non-destructive. While many studies have reported correlations between source rock thermal maturities and Raman spectral parameters (Spötl et al., 1998; Kelemen and Fang, 2001; Beyssac et al., 2002, 2003; Rahl et al., 2005; Lünsdorf et al., 2014, 2017; Romero-Sarmiento et al., 2014; Wilkins et al., 2014, 2015, 2018; Zhou et al., 2014; Lünsdorf and Lünsdorf, 2016; Schmidt Mumm and Inan, 2016; Bonoldi et al., 2016; Lünsdorf, 2016; Lupoi et al., 2017; Sauerer et al., 2017; Schito et al., 2017; Xueqiu et al., 2017; Cheshire et al., 2017; Childress and Jacobsen, 2017; Baludikay et al., 2018; Henry et al., 2018; Khatibi et al., 2018), only a few studies have examined the effect of sample heterogeneity on the

Corresponding author. E-mail address: [email protected] (A.M. Jubb).

https://doi.org/10.1016/j.coal.2018.09.017 Received 5 August 2018; Received in revised form 24 September 2018; Accepted 25 September 2018 Available online 26 September 2018 0166-5162/ Published by Elsevier B.V.

International Journal of Coal Geology 199 (2018) 1–9

A.M. Jubb et al.

Table 1 Shale geologic reference materials (GRMs) information and bulk geochemical parameters. Total organic carbon (TOC) data from LECO TOC analyzer. Programmed pyrolysis data from Wildcat Technologies Hydrocarbon Analyzer with Kinetics (HAWK) where Tmax = temperature of S2 peak maximum, HI (hydrogen index) = (S2/ TOC) × 100, OI (oxygen index) = (S3/TOC) × 100, PI (production index) = S1/(S1 + S2). Formation

Sample ID

Basin

Long.

Lat.

Boquillas Woodford

ShBOQ ShWFD

Gulf Coast Anadarko

101.2111 97.1489

29.7028 34.3520

Marcellus Niobrara

ShMAR ShNIO

Appalachian DenverJulesburg

77.9884 105.2315

42.9787 40.2230

Age

TOC (wt%)

Late Cretaceous DevonianMississippian Middle Devonian Late Cretaceous

Programmed pyrolysis S1 (mg-HC/gRock)

S2 (mg-HC/gRock)

S3 (mg-CO2/gRock)

Tmax (°C)

HI

OI

PI

5.0 8.0

1.4 1.2

31 44

0.7 0.6

422 427

620 550

14 7.8

0.04 0.03

6.8 2.5

4.6 2.1

20 4.4

0.2 0.2

436 455

300 180

2.7 8.4

0.19 0.32

shale OM grains is influenced by associations with the surrounding mineral matrix (Baldock and Skjemstad, 2000; Kennedy et al., 2002). We thus hypothesize that: (i) variation of Raman thermal maturity values extracted across single OM grains of the same type will be low, (ii) the presence of spatially distinct chemical domains or localized organic-mineral interactions within single OM grains will be reflected in high variation for the Raman parameters, and (iii) the observed Raman variation will decrease as the thermal maturity of the shale GRMs increases and chemical heterogeneity is lost with the concomitant increase in OM aromaticity (Lünsdorf, 2016). To test these hypotheses we characterized the four shale GRMs using programmed pyrolysis and optical petrography and then evaluated the spatial variation of the extracted Raman parameters used for thermal maturity estimates across short distances (≤5 μm) within individual OM grains of the same type through Raman microscopy mapping. Our findings are discussed in the context of understanding chemical variation within source rock OM and highlight the care which must be adopted when using Raman to evaluate the thermal maturity of petroleum source rocks, especially for shales and mudrocks which feature highly dispersed, heterogeneous OM.

Raman response from probed OM (Beyssac et al., 2003; Lünsdorf et al., 2014; Henry et al., 2018). Furthermore, to our knowledge, there have been no reports on the variability of the Raman response from within individual organic grains embedded in shale matrices, although recent work employing coupled atomic force microscopy and infrared spectroscopy suggest this variability may be limited (Yang et al., 2017). Raman spectra of source rock OM are typically dominated by two broad peaks centered at ~1600 cm−1 and 1350 cm−1, generally assigned to the E2g CeC vibrations of sp2 carbon atoms in aromatic rings and to A1g vibrational modes which arise due to disorder and defects in the aromatic lattice, respectively (Wang et al., 1990; Ferrari and Robertson, 2001; Sauerer et al., 2017). These peaks are referred to as the G and D bands in analogy to the Raman spectrum of graphitic materials where G stands for graphite and D stands for disorder. The Raman spectrum of single crystalline graphite does not feature a D band (Tuinstra and Koenig, 1970; Wang et al., 1990). Due to resonance enhancement effects between aromatic ring moieties within carbonaceous materials and the excitation wavelengths used in most Raman measurements (Ferrari and Robertson, 2001; Jorio et al., 2011) the strong signal from the G and D peaks can obscure the signal from other Raman active chemical groups within the Raman probe volume. Hence, efforts to extract correlations to thermal maturity from the Raman spectra of source rocks have focused exclusively on the analysis of these two peaks. Analysis of the G and D peaks is generally approached by spectral deconvolution through iterative peak fitting (Wopenka and Pasteris, 1993; Beyssac et al., 2002, 2003; Rahl et al., 2005; Wilkins et al., 2014; Lünsdorf and Lünsdorf, 2016; Sauerer et al., 2017), although several authors have pointed out that this approach can be subject to operator biases (Lünsdorf et al., 2014; Wilkins et al., 2014; Lupoi et al., 2017; Henry et al., 2018). Raman thermal maturity estimates are then made by empirical correlation between the extracted Raman peak parameters and another thermal maturity proxy such as vitrinite reflectance or Tmax. The present research assesses the spatial variation in common Raman parameters used to estimate thermal maturity from individual organic grains within four proposed U.S. Geological Survey shale geologic reference materials (GRMs) from the Boquillas, Marcellus, Niobrara, and Woodford Formations. These GRMs are representative of several important U.S. shale plays, vary in thermal maturity from immature through the oil window, and feature highly dispersed OM embedded in a fine-grained mineral matrix characteristic of sedimentary mudrocks. While the Raman response can vary between different OM types present in shale source rocks due to compositional differences amongst them and their own innate heterogeneity (Beyssac et al., 2003; Lünsdorf et al., 2014; Henry et al., 2018), variation of the Raman response from within individual OM grains of the same type (e.g., kerogen, solid bitumen, etc.) should be greatly reduced. However, differences in Raman signal from within individual OM grains could reflect spatial variation in the chemical domains present across micron scales and/or may also imply that the local molecular structure across

2. Materials and methods 2.1. Samples The four shale samples analyzed in this study are all proposed U.S. Geological Survey GRMs relevant to U.S. unconventional petroleum plays (https://www.eia.gov). The purpose of these proposed GRMs is to serve as quality control and assurance standards for mineralogical and organic and inorganic geochemical studies on shale and mudrock systems. General sample information is provided in Table 1 and is taken from Birdwell and Wilson (2017). The Boquillas sample (ShBOQ) is representative of the Late Cretaceous (Cenomanian-Turonian) Boquillas Shale of the Gulf Coast basin and was collected from a road cut on US90 west of Del Rio in Val Verde County, Texas. The Marcellus sample (ShMAR) is representative of the Middle Devonian Marcellus Shale in the Appalachian Basin and was collected at the Oatka Creek outcrop in the village of Le Roy in Genesee County, New York. The Niobrara sample (ShNIO) is representative of the Late Cretaceous Niobrara Shale in the Denver-Julesburg Basin and was collected at the CEMEX quarry near Lyons in Boulder County, Colorado. The Woodford sample (ShWFD) is representative of the Devonian-Mississippian Woodford Shale in the Anadarko Basin and was collected at a road cut on I-35 near Ardmore and Springer in Carter County, Oklahoma. All samples were supplied as crushed rock chips from the U.S. Geological Survey Reference Material Project (https://crustal.usgs.gov/geochemical_ reference_standards/).

2

International Journal of Coal Geology 199 (2018) 1–9

A.M. Jubb et al.

surface for all three lasers was ≤1 μm. The phonon band at 520.7 cm−1 from crystalline silicon was used to calibrate all Raman spectra. In a Raman mapping measurement, a desired OM grain was identified within a shale fragment ROI on the sample briquette surface. Coordinates for the individual grains were noted to allow for later correlative petrographic, fluorescent, and SEM measurements of the same region (Fig. 1). A Raman map was then collected across the ROI encompassing the OM grain with a programmable xyz stage where a Raman spectrum was collected every 2–5 μm in a raster pattern. A minimum of 100 Raman spectra were analyzed per sample ROI. Thermal alteration of the OM in the shale ROIs by the laser beam was checked by comparing optical micrographs of the sample surface taken prior to and after the Raman measurements. This is critical as the thermal alteration of carbonaceous OM will strongly affect the Raman response from these materials, especially for the low-maturity shales under study here. Raman spectra were fit with a sum of four Lorentzian peaks between 500 and 2300 cm−1 with a cubic baseline using the IGOR Pro software suite (Fig. 2). No Lorentzian peak or baseline parameters were fixed during the spectral fitting routine. The choice of a Lorentzian line shape, as opposed to a Gaussian, Voigt, or pseudo-Voigt profile, for the peak fit was arrived at by considering the inherent Lorentzian nature for the harmonic oscillator model of vibrations (Meier, 2005). The four Lorentzian peaks used to fit the Raman spectra correspond to the G peak and the D1, D3, and D4 peaks previously reported (Beyssac et al., 2002, 2003; Cheshire et al., 2017; Schito et al., 2017); fitting the GRM Raman spectra with 2, 3, or 5 Lorentzian peaks resulted in less reproducible fitting results. Several common Raman thermal maturity proxies extracted from the spectra fitting parameters include the full-width-athalf-maximum (FWHM) of the G band (Wopenka and Pasteris, 1993; Zhou et al., 2014), the separation of the G and D1 peak centers in wavenumber space, referred to as the Raman band separation (RBS) (Kelemen and Fang, 2001; Sauerer et al., 2017), and the intensity ratio of the D1 band over the G band (Beyssac et al., 2002, 2003). These parameters were chosen for the current study as they are widely reported Raman thermal maturity proxies, however, other Raman thermal maturity proxies have also been suggested (Lünsdorf et al., 2014, 2017; Wilkins et al., 2014, 2015, 2018; Lünsdorf and Lünsdorf, 2016). Of special note when extracting RBS values from Raman spectra is the well documented effect of laser wavelength on the peak position for the D bands (Wang et al., 1990; Matthews et al., 1999; Ferrari and Robertson, 2001; Ferrari, 2007; Sauerer et al., 2017). The dispersion of the D peaks results in greater RBS values when longer laser wavelengths are used for excitation as the position of the G peak is unaffected by the excitation laser wavelength. For the four shale GRMs the dispersion of the D1 band was examined from a representative OM location using 473 nm, 532 nm, and 785 nm excitation (Fig. 3a) and the dispersion exhibited by the samples shows good agreement with previous literature values for the laser wavelength dependence of the D1 band position (Sauerer et al., 2017). The dispersion of the D1 peak exhibits approximately linear behavior with the change in excitation wavelength which indicates that the RBS values determined with 473 nm excitation reported here are ~5–10 cm−1 lower than if the 532 nm laser had been used for the Raman mapping experiments, although the degree of RBS variation with excitation wavelength will also depend on the thermal maturity of the sample (Fig. 3b).

2.2. Bulk geochemistry Bulk geochemical measurements were collected from small amounts of crushed rock sample using a LECO C744 Carbon Analyzer and a Wildcat Technologies Hydrocarbon Analyzer with Kinetics (HAWK) following the manufacturer's instructions. The total organic carbon (TOC) of the samples was determined using the LECO instrument (carbonates removed) while the programmed pyrolysis parameters (e.g., S1, S2, S3, etc.) were collected with the HAWK. For all samples internal laboratory standards were run for quality assurance purposes (https://energy.usgs.gov/GeochemistryGeophysics/ GeochemistryLaboratories.aspx). 2.3. Organic petrography and reflectance measurements The four GRMs samples were prepared for petrographic analysis according to ASTM method D2797 (ASTM, 2011) and all subsequent measurements, i.e., reflectance, optical petrography, fluorescence and scanning electron microscopy imaging, and Raman microscopy, were collected from individual rock fragments embedded in the petrographic briquettes. Reflectance measurements were collected per ASTM method D7708 (ASTM, 2014) with a Leica DM400 microscope equipped with LED illumination and monochrome camera detection using the DISKUSFOSSIL program by Hilgers Technisches Buero. All reflectance measurements were calibrated with a YAG standard (%R0 = 0.908%). Individual shale fragments were imaged with white light under oil immersion using a Zeiss AxioImager and with UV light using a Leica DMR for each sample in order to identify the OM present and select the regions of interest (ROIs) for Raman mapping (Fig. 1). The Leica DMR microscope used to collect the fluorescent images for each sample was equipped with a filter assembly consisting of a 340–380 nm band pass filter in the excitation channel and a 425 nm long pass filter in the detection channel. Backscattered electron images of each sample ROI were collected on a Hitachi SU5000 variable pressure field emission scanning electron microscope (FE-SEM) operated at 70 Pa, 15 keV accelerating voltage, 30–40 spot intensity (unitless), and a 10 mm working distance. All FE-SEM images were collected post-Raman analysis in order to avoid any alteration of the OM by the electron beam that could possibly impact the Raman response from the interrogated OM. The correlated white light, fluorescence, and FE-SEM images for a representative Raman mapping ROI from each sample are shown in Fig. 1. 2.4. Raman microscopy All Raman measurements were collected for individual shale ROIs encompassing both organic and mineral grains using a Horiba Xplora Plus Raman microscope system. Previous work has shown that the Raman spectra of carbonaceous materials can be influenced by the mechanical polishing that occurs during petrographic briquette preparation (Beyssac et al., 2003; Lünsdorf, 2016) and polishing impacts different organic matter types to a varying degree (Hackley et al., 2018); however, as all Raman spectra generated here are used for intrashale comparisons within OM grains of the same type, we assume that any polishing impacts on OM structure will be similar. The Raman microscope was equipped with three excitation lasers with emission wavelengths of 473 nm, 532 nm, and 785 nm. All Raman spectra used for the mapping experiments were collected using the 473 nm laser with perpendicular polarization through a 100× objective (0.9 NA) with 50–500 μW power at the sample surface, 1–10 s acquisition times, and 3 co-averaged scans. Additional Raman spectra were acquired using the 532 nm and 785 nm lasers with similar settings in order to determine the spectral dispersion of the D1 band exhibited by each shale GRM. The 473 nm laser was chosen for the mapping experiments as it provided spectra with higher signal-to-noise levels due to the λ−4 dependence of Raman scattering. The beam diameter at the sample

3. Results and discussion 3.1. Bulk geochemistry The bulk geochemical parameters for all four shale GRMs are given in Table 1. The GRMs feature moderate TOC levels ranging from 2.5% (ShNIO) to 8.0% (ShWFD). Programmed pyrolysis results show Tmax values from 422 °C to 455 °C in the order ShBOQ < ShWFD < 3

International Journal of Coal Geology 199 (2018) 1–9

A.M. Jubb et al.

Fig. 1. White light (a, d, g, and j), FE-SEM (b, e, h, and k), and fluorescence (c, f, i, and l), micrographs of ShWFD (a-c), ShBOQ (d-f), ShMAR (g-i), and ShNIO (j-l) samples in regions where Raman microscopy mapping was performed. White light, FE-SEM, and fluorescence images for each sample were collected on similar fields of view. Dashed white box in each micrograph corresponds to approximate Raman microscopy mapping region encompassing both OM and mineral grains. Saturated bright spots in white light and FE-SEM micrographs correspond to pyrite framboids which fluoresce red under UV light.

within these samples and their marine depositional environment where Type III kerogen input is minimal (Hackley and Cardott, 2016).

ShMAR < ShNIO indicating that the thermal maturity of these samples ranges from immature conditions into the oil window (Peters, 1986; Peters and Cassa, 1994). The HI and OI indices decease with an increase in Tmax values except for the ShNIO sample's OI value. The PI values for these samples increases with increasing Tmax values. Plotting these samples into pseudo van Krevelen space (i.e., HI vs OI) indicates that the OM in the shale GRMs show Type II kerogen evolution. The kerogen evolution relationships between HI and Tmax suggest that the shale OM is composed primarily of Type II kerogen (ShWFD and ShBOQ) or Type II/III kerogen mixtures (ShMAR and ShNIO) (Peters, 1986; Peters and Cassa, 1994). Although a Type II/III kerogen mixture is suggested for the ShMAR and ShNIO samples, this signature is most likely due to OM conversion of Type II kerogen given the prevalence of solid bitumen

3.2. Organic petrography The ShBOQ sample consists of an immature to marginally mature marl and contains dispersed vitrinite and inertinite fragments as well as amorphous organic matter (AOM) schlieren with intimate association to the mineral matrix. This sample is weakly fluorescent and appears reddish in white light. Foraminifera are abundant and their chambers are typically filled with carbonate and minor pyrite (solid bitumenfilled chambers are sparse). Some granular micrinite associated with AOM schlieren and development of grey reflecting surfaces, mostly 4

International Journal of Coal Geology 199 (2018) 1–9

A.M. Jubb et al.

Fig. 2. Representative Raman spectrum from the ShNIO sample (black trace) following normalization to the peak intensity at ~1600 cm−1 and spectral deconvolution with four Lorentzian profiles (red traces, bottom panel) and cubic baseline (grey dashed trace, middle panel). Composite fit is indicated by solid red trace in middle panel. Fit residual are provided in top panel (red markers). Peak centers of the four fit Lorentzians are indicated by vertical grey lines. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

unexpelled solid bitumen, indicates kerogen conversion to petroleum. This sample is petrographically similar to previously described samples of Boquillas Formation source rocks, although solid bitumen filled foraminifera chambers (Hackley et al., 2015) are less common. The specific ROI analyzed for this GRM (Fig. 1d-f) shows an oil-prone degraded AOM lamella closely associated with the mineral matrix. The ShWFD sample consists of abundant well-preserved thin-walled telalginite (Leiosphaeridia) possessing brilliant organic fluorescence where the occasional telalginite is silicified. The groundmass contains copious dispersed fluorescent AOM and solid bitumen (with little or no fluorescence) with only rare dispersed inertinite fragments present. The ShWFD is similar in organic petrographic description and thermal maturity to Woodford mudstone studied by Ko et al. (2018) and a Woodford sample studied by Hackley and Lewan (2018). The specific ROI analyzed for this GRM (Fig. 1a-c) includes solid bitumen dispersed amongst quartz and pyrite grains. The ShMAR sample possesses abundant pyrite framboids amongst a groundmass of solid bitumen that occurs in lamellae on the bedding planes. There are rare dispersed inertinite fragments (20–30 μm) and euhedral carbonates (20–30 μm) throughout the sample. This sample exhibits significantly red-shifted organic fluorescence. The sample is similar to previous organic petrographic descriptions of Marcellus samples (Ryder et al., 2013). The specific ROI analyzed for this GRM (Fig. 1g-i) includes a large (~5 × 25 μm) solid bitumen domain with a euhedral carbonate inclusion on the bottom right hand side of the domain. The ShNIO sample features foraminifera filled with recrystallized carbonate (with reddish fluorescence) and pyrite. The lamellae of finegrained solid bitumen-rich zones show no fluorescence and there are finely dispersed inertinite fragments and rare granular micrinite present. The sample is similar in organic petrographic description and thermal maturity to Niobrara Formation source rocks from the CEMEX quarry reported previously (Hackley and Cardott, 2016). The specific

Fig. 3. (a) Position of the D1 peak center for a representative OM Raman spectrum from each shale GRM as a function of excitation wavelength, i.e., 473 nm, 532 nm, and 785 nm excitation. The Raman spectra for each excitation wavelength were collected from the same location on the respective GRM. Data are markers and the solid black trace corresponds to the linear regression fit to all data. Error bars correspond to precision uncertainty of the peak center position at the one sigma level. The linear regression equation relating the D1 peak location to the excitation wavelength is provided in the inset. (b) The Raman band separation as a function of excitation wavelength for the same representative Raman spectra yielding the D1 values evaluated in panel (a). Error bars correspond to the propagated uncertainty at the one sigma level from the D1 and G peak position standard deviations.

ROI analyzed for this GRM (Fig. 1j-l) is just below a large inertinite fragment and features highly dispersed solid bitumen within the mineral matrix.

3.3. Reflectance measurements Reflectance measurements were collected from vitrinite macerals for the ShBOQ sample and from solid bitumen for the other three GRM samples due to the absence of vitrinite within these rocks. The measured reflectance values are provided in Table 2 along with the standard deviation of the measurement. The values range from 0.40% (ShWFD) to 0.99% (ShNIO), in general agreement with the thermal maturity trend from the programmed pyrolysis Tmax values. For the two lowest maturity samples, ShBOQ and ShWFD, there is disagreement between the measured reflectance and the Tmax value, i.e., the ShBOQ sample has a higher reflectance than the ShWFD sample but a lower Tmax. This could be attributed to either the high organic sulfur content of ShBOQ kerogen (Little et al., 2012) and/or the inherent differences 5

International Journal of Coal Geology 199 (2018) 1–9

A.M. Jubb et al.

Table 2 Shale geochemical reference materials (GRMs) petrographic and Raman parameters for the D1 and G peaks. Reflectance (%Ro) values represent the mean while Raman parameter values represent the median. All stated uncertainties represent plus or minus one standard deviation. Sample ID

ShBOQ ShWFD ShMAR ShNIO a b c

Reflectance

Ramanc

n

%Ro

n

D1 Location (cm−1)

D1 FWHM (cm−1)

D1 Amplitude

G Location (cm−1)

G FWHM (cm−1)

G Amplitude

RBS (cm−1)

25 25 40 25

0.54 0.40 0.73 0.99

110 140 112 100

1368 1360 1360 1366

95 ± 32 113 ± 23 117 ± 48 143 ± 13

979 ± 420 1736 ± 692 396 ± 176 2643 ± 1131

1588 1591 1603 1612

86 81 60 52

1509 ± 589 3333 ± 1258 889 ± 289 4343 ± 2047

219 231 243 246

± ± ± ±

0.07a 0.07b 0.09b 0.09b

± ± ± ±

12 7 11 4

± ± ± ±

2 1 2 2

± ± ± ±

7 3 7 3

± ± ± ±

(D1/G)Amp

12 7 11 2

0.632 0.548 0.450 0.553

± ± ± ±

0.141 0.091 0.123 0.143

Vitrinite reflectance. Solid bitumen reflectance. Raman values from entire ROI.

samples, i.e., ShBOQ and ShWFD, than for the two higher maturity shales, i.e., ShMAR and ShNIO. This peak is commonly observed for Raman spectra of petroleum source rocks within the literature (Beyssac et al., 2002, 2003; Schmidt Mumm and Inan, 2016; Cheshire et al., 2017; Lupoi et al., 2017; Sauerer et al., 2017; Schito et al., 2017), but is often disregarded or attributed to the D2 peak (Cheshire et al., 2017; Sauerer et al., 2017). Here this peak was not included in any analyses due to its uncertain origin. The spectra in Fig. 4 represent the average of all Raman spectra collected across each representative ROI; i.e., Raman spectra from nonorganic/mineral grains within the ROI have been included. The G and D peaks are clearly the primary feature of the spectra, absent are peaks attributable to the non-organic mineral grains in the ROIs (e.g., ~1000 cm−1 SieO peak) even though these grains comprise ~50% of some ROIs (Fig. 5a). The absence of these signals highlights the magnitude of the resonance enhancement of aromatic CeC vibrations, which results in the G and D bands as the only clearly assignable spectral features. Furthermore, the Raman spectra from every spot examined for the GRMs contained this aromatic carbon signal. This is illustrated in Fig. 5 where contour maps corresponding to extracted Raman parameters for the G and D1 peaks from a ShMAR ROI are shown. This sample ROI was chosen due to the presence of the large, continuous solid bitumen domain (Fig. 5a). Similar results are observed for other ROIs within this sample as well as for the other three GRM samples. The outline of the solid bitumen domain from Fig. 5a is overlaid on the contour maps shown in Fig. 5b-h highlighting that areas within the ROI that do not correspond to this OM domain feature signal attributed to aromatic carbon groups. Indeed, from Fig. 5b and e it is seen that the strongest “carbon” signal within this specific ROI comes from outside of the solid bitumen particle. This observation not only emphasizes how the resonance enhancement of the aromatic CeC modes can obscure the Raman signal from other chemical moieties but it also indicates that care must be taken to properly identify OM regions prior to Raman analysis in order to avoid interpreting “carbon” signal from non-organic regions when contrasting Raman thermal maturity relationships. The origin of the observed aromatic carbon signal from non-organic regions within the shale materials under study here is unclear, but could indicate the presence of ubiquitous microscopic and/or nanoscopic OM throughout these shales (e.g., solid bitumen-filled nanopores or abiotic reduced carbon coatings (Steele et al., 2012)) or possibly hydrocarbon staining of the mineral matrix following petroleum migration, although this is unlikely for the Woodford and Boquillas samples given their low maturity level (Tables 1 and 2). While resolving this issue is beyond the scope of the current study, it is critical for understanding the Raman signal from shale and mudrock source rocks and is a focus of on-going work within our laboratory. Beyond illustrating the ubiquity of the G and D peaks across the ShMAR sample surface, the contour maps shown in Fig. 5 demonstrate the profound variation in the G and D1 peak parameters across this

between measuring vitrinite reflectance (ShBOQ) and solid bitumen reflectance (ShWFD) (Hackley and Lewan, 2018). Regardless, both the reflectance and Tmax values for these two shales indicate that they are in an immature stage of petroleum generation (Dow, 1977; Peters, 1986; Hackley and Cardott, 2016). 3.4. Raman microscopy The average Raman spectra from the ROI indicated in Fig. 1 for each shale GRM in the first-order carbon band region (~500–2300 cm−1) are shown in Fig. 4. The median values for the G and D1 peak parameters for each GRM data set and the corresponding median RBS and D1/G ratio parameters are provided in Table 2. These spectra are dominated by strong peaks at ~1350 cm−1 and ~1600 cm−1, characteristic of carbonaceous materials, and feature a broadly fluorescent background typical for Raman spectra of OM within low maturity source rocks. As the thermal maturity of the GRM samples increases, Table 2, the signalto-noise ratio for the G and D peaks correspondingly increases due to the increase in aromatic moieties within the OM and the decrease in the fluorescent background. For all four GRMs a peak is also observed at ~1740 cm−1 which is more prominent for the two lower maturity

Fig. 4. Average Raman spectra collected for the Woodford (red trace), Boquillas (blue trace), Marcellus (black trace), and Niobrara (green trace) Formation shale GRMs from the Raman mapping regions indicated in Fig. 1. The number of spectra averaged are indicated in the legend and range from n = 100 to n = 140. The spectra have been normalized to the G peak intensity at ~1600 cm−1 and are offset vertically for clarity. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) 6

International Journal of Coal Geology 199 (2018) 1–9

A.M. Jubb et al.

Fig. 6. Raman band separation versus (D1/G)Amp. ratio extracted for each Raman spectrum collected for the four shale GRM Raman ROIs indicated in Fig. 1 (empty circles) and the median values for each data set (filled circles). Error bars on the median values correspond to one standard deviation of the extracted Raman band separation and amplitude ratio values. Median RBS and D1/G intensity ratio values are provided in Table 2. Reflectance values for each shale GRM are provided in the figure legend.

Raman band separation (RBS) and the D1/G intensity ratio, are plotted for every Raman spectrum collected from the representative ROI from the four shale GRMs indicated in Fig. 1. The RBS and D1/G intensity values are generally self-consistent in that higher RBS values co-occur with lower D1/G intensity ratio values, both of which correlate to higher thermal maturities. This self-consistency is the source of the striking linear trend exhibited within the data sets shown when these two parameters are plotted against each other. There is large variation for both the RBS and D1/G parameters extracted from Raman spectra across the GRM sample ROIs and their enclosed OM grains, which is evident in Fig. 6. The degree of variation for these two parameters is greater than previous reports examining the effects of sample heterogeneity on the Raman response from carbonaceous materials (Lünsdorf et al., 2014; Henry et al., 2018), although these studies focused primarily on extracted kerogen and AOM. This provides indirect support that mineral associations could influence the Raman response from these types of OM, however Beyssac et al. (2003) suggested that estimation of an average heterogeneity for preserved OM with Raman may not be possible. Here the ShBOQ, ShMAR, and ShWFD samples exhibit similar variation for the RBS and D1/G parameters while the ShNIO sample shows tighter grouping, especially for the RBS proxy. This observation is consistent with the higher thermal maturity of the ShNIO sample and implies that the chemical heterogeneity within the ShNIO OM is lower than is present for the other three GRMs. While the variation in the RBS and D1/G intensity ratios is large, the median values (Fig. 6 and Table 2) for both of these thermal maturity proxies do exhibit systematic differences and show some degree of agreement with the Tmax and vitrinite/solid bitumen reflectance maturity estimates. The median RBS values in particular match the thermal maturity order determined by programmed pyrolysis, i.e., Tmax. Qualitative agreements such as this reaffirm that Raman is able to distinguish between materials with different thermal histories to a degree, as has been noted by many previous authors. However, the standard deviations for the median RBS and D1/G intensity ratio proxies demonstrate that these proxies are not capable of an accurate

Fig. 5. (a) White light micrograph of ROI from ShMAR sample with large (~5 × 25 μm) solid bitumen domain. Same ROI as Figs. 1g-i. (b-g) Contour maps showing intensities of extracted Raman parameters across the sample ROI for the G and D1 peaks: (b) amplitude of G peak; (c) location of G peak center; (d) full width at half maximum (FWHM) of G peak; (e) amplitude of D1 peak; (f) location of D1 peak center; and (g) full width at half maximum (FWHM) of D1 peak. (h) Raman band separation (RBS) for sample ROI determined by subtracting the D1 peak position (panel f) from the G peak position (panel c). For panels b-h outline of solid bitumen domain is indicated. Scale is the same for all panels.

shale ROI surface. For example, Fig. 5f shows a variation in the D1-band position of ~50 cm−1 across the OM grain. Considering the relative stability of the G-band position (Fig. 5c), a shift in the D1-band position of ~50 cm−1 corresponds to previously reported OM thermal maturities spanning immature conditions into the dry gas window (Cheshire et al., 2017; Sauerer et al., 2017). This variation appears greatest along the interface between the solid bitumen domain and the mineral matrix. While the delineation between organic and mineral regions is not readily apparent across the entirety of the ShMAR ROI shown in Fig. 5, this observation may suggest that close association between OM and mineral surfaces can influence the Raman response from the interrogated OM. Differences in the Raman response for OM near the edge of the grain may be the result of local molecular structure differences in mineral-associated OM compared to OM residing in the bulk of the grain, as mineral associations are known to impact the molecular structure of complex organic compounds (Baldock and Skjemstad, 2000). Furthermore, here the variation in the Raman response is not limited to differences between organic versus non-organic regions, but there is large variation within the same OM domain. These variations directly translate into large standard deviations for any derived Raman thermal maturity proxies (e.g., Fig. 5h) from these spectra, as shown in Fig. 6 where two common Raman proxies for thermal maturity, the 7

International Journal of Coal Geology 199 (2018) 1–9

A.M. Jubb et al.

estimate of thermal maturity, e.g., the median ShWFD RBS value is indistinguishable statistically from the ShBOQ and ShMAR values, although the deviations exhibited by the ShNIO sample are smaller (Table 2). The impact of this observation is apparent if the shale GRM RBS and reflectance values are compared against similar data from previous studies. Literature values for correlated RBS and reflectance measurements determined for OM from petroleum source rocks (Zhou et al., 2014; Schmidt Mumm and Inan, 2016; Lupoi et al., 2017; Sauerer et al., 2017; Schito et al., 2017; Henry et al., 2018; Khatibi et al., 2018) and one coal (Xueqiu et al., 2017) are compared to the median RBS and mean reflectance values determined for the four shale GRMs here (Fig. 7). The RBS values displayed in Fig. 7 exhibit wide scatter between data sets, most likely due to the different types of OM examined and the lack of a consistent excitation wavelength used to collect the Raman spectra, although intra-data set scatter is lower. Regardless, from Fig. 7 it is observed that for the three least mature GRMs (ShWFD, ShBOQ, and ShMAR) the RBS standard deviations encompass nearly the entire range of reflectance measurements reported in these previous studies. The ShNIO RBS standard deviations are smaller and this result is in better agreement with the previously reported data displayed. The failure of the RBS proxy to accurately distinguish between the ShWFD, ShBOQ, and ShMAR samples is entirely related to the variation in the Raman response across the interrogated sample ROIs and this result casts a critical light on the ability of a Raman approach to generate accurate thermal maturity estimates for shales with maturities less than ~1% Ro.

and non-organic mineral regions, where mineral regions also exhibited spectral responses traditionally associated with aromatic carbon moieties, as well as within organic matter (OM) regions comprised of the same type of OM (e.g., solid bitumen). This is in contrast to the results of Yang et al. (2017), who showed little signal variation from shale organic regions using combined atomic force microscopy-infrared spectroscopy. The greater variance in shale OM observed in this study may be the result of fundamental differences between the analytical methods used or inherent sample differences as a New Albany Shale sample, as investigated by Yang et al. (2017), was not examined here. Additionally, programmed pyrolysis and vitrinite or solid bitumen reflectance measurements were collected from these materials to better constrain their thermal history and suggest the thermal maturity for these samples span immature conditions into the oil window. While the Raman thermal maturity proxies determined in this study show qualitative agreement with the programmed pyrolysis and reflectance proxies, the high variability of the Raman response from these materials directly translates into large uncertainties for the Raman proxies explored here, especially for the low maturity samples from the Boquillas, Marcellus, and Woodford Formations. The source of the uncertainty in the Raman response from these materials could possibly be due to mineral effects which may drive differences in local OM chemical structure for compounds in intimate contact with mineral surfaces (Baldock and Skjemstad, 2000; Kennedy et al., 2002) or contribute interfering mineral fluorescence artifacts to the Raman signal, or this may reflect the inherent chemical heterogeneity present within OM from immature source rocks (Beyssac et al., 2003). Minor uncertainties in the measured Raman spectra for the Niobrara shale sample suggest that OM chemical heterogeneity is lost for samples with higher thermal maturity. This degree of uncertainty effectively prohibits an accurate determination of the thermal history for the Boquillas, Marcellus, and Woodford Formation shales using the RBS or D1/G proxies. Beyond exploring the natural heterogeneity present for these shales, these results urge caution when using Raman spectroscopy to investigate OM within highly heterogeneous shales and cast a critical light on the ability of this approach to accurately estimate thermal maturities of low maturity petroleum source rocks. First, as aromatic carbon signal was observed for nearly every shale GRM spot where a Raman signal was collected, it is imperative that a careful selection and identification of OM type be carried out before Raman analysis, in contrast to recent studies highlighting the ability of Raman spectroscopy to interrogate samples with minimal sample preparation (Lupoi et al., 2017; Sauerer et al., 2017). Otherwise, signal from regions that are mainly comprised of non-organic materials could be attributed to source rock OM. Second, the high degree of Raman variability exhibited within OM grains, as was observed for three of four shales studied here, indicates that many locations within the chosen OM region of interest need to be analyzed in order to fully capture a representative Raman spectrum from the sample. However, this variability will directly translate to any extracted Raman thermal maturity proxy possibly hindering statistically significant differentiation between samples, as was observed here for the Boquillas, Marcellus, and Woodford Formation samples. Ultimately these results highlight the chemical complexity present within shale OM, which is reflected in the Raman response from these materials. This complexity provides motivation for additional efforts to more fully develop Raman spectroscopy as a robust tool for the investigation of thermal history for petroleum-generating source rocks, especially for materials such as shales and mudrocks which contain highly dispersed, heterogeneous OM.

4. Conclusions

Author contributions

The results presented here show high variability in the Raman response across short (< 5 μm) spatial distances for shale materials from the Boquillas, Marcellus, Niobrara, and Woodford Formations. High variability in the Raman response was observed both between organic

AMJ, PJB, and RCB conceived the research; JEB selected collection locations for the GRMs, assisted with collection, coordinated bulk geochemical analyses, and supplied the samples for this work; AMJ performed Raman and fluorescence analyses; PCH and JJH prepared

Fig. 7. Median Raman band separation (RBS) values versus the measured reflectance values for each shale GRM studied here (black markers – Sample IDs inset) along with similar data from eight other studies reporting Raman data for carbonaceous OM (Zhou et al., 2014; Schmidt Mumm and Inan, 2016; Lupoi et al., 2017; Sauerer et al., 2017; Schito et al., 2017; Xueqiu et al., 2017; Henry et al., 2018; Khatibi et al., 2018). Error bars on shale GRMs data correspond to one standard deviation. Laser wavelengths used to generate each data set are provided in the legend as the RBS values will vary with excitation wavelength, see Fig. 3.

8

International Journal of Coal Geology 199 (2018) 1–9

A.M. Jubb et al.

samples and performed petrographic analyses, BJV performed SEM analyses; SAW participated in GRM collections and processed the raw materials; all authors contributed to writing the manuscript.

Kelemen, S.R., Fang, H.L., 2001. Maturity trends in Raman spectra from kerogen and coal. Energy Fuels 15, 653–658. Kennedy, M.J., Pevear, D.R., Hill, R.J., 2002. Mineral surface control of organic carbon in black shale. Science 295, 657–660. Khatibi, S., Ostadhassan, M., Tuschel, D., Gentzis, T., Bubach, B., Carvajal-Ortiz, H., 2018. Raman spectroscopy to study thermal maturity and elastic modulus of kerogen. Int. J. Coal Geol. 185, 103–118. Ko, L.T., Ruppel, S.C., Loucks, R.G., Hackley, P.C., Zhang, T., Shao, D., 2018. Pore-types and pore-network evolution in Upper Devonian-lower Mississippian Woodford and Mississippian Barnett mudstones: Insights from laboratory thermal maturation and organic petrology. Int. J. Coal Geol. 190, 3–28. Little, J., Formolo, M.J., Lewan, M.D., 2012. Correlating oils in Turonian-Cenomanian source rocks using hydrous pyrolysis and organic sulfer compounds. Am. Assoc. Pet. Geol. Search Discov. 10453. http://www.searchanddiscovery.com/pdfz/documents/ 2012/10453little/ndx_little.pdf.html. Lünsdorf, N.K., 2016. Raman spectroscopy of dispersed vitrinite — Methodical aspects and correlation with reflectance. Int. J. Coal Geol. 153, 75–86. Lünsdorf, N.K., Lünsdorf, J.O., 2016. Evaluating Raman spectra of carbonaceous matter by automated, iterative curve- fitting. Int. J. Coal Geol. 161, 51–62. Lünsdorf, N.K., Dunkl, I., Schmidt, B.C., Rantitsch, G., von Eynatten, H., 2014. Towards a higher comparability of geothermometric data obtained by Raman spectroscopy of carbonaceous material. Part I: evaluation of biasing factors. Geostand. Geoanal. Res. 38, 73–94. Lünsdorf, N.K., Dunkl, I., Schmidt, B.C., Rantitsch, G., von Eynatten, H., 2017. Towards a higher comparability of geothermometric data obtained by Raman spectroscopy of carbonaceous material. Part 2: a revised geothermometer. Geostand. Geoanal. Res. 41, 593–612. Lupoi, J.S., Fritz, L.P., Parris, T.M., Hackley, P.C., Solotky, L., Eble, C.F., Schlaegle, S., 2017. Assessment of thermal maturity trends in Devonian–Mississippian source rocks using Raman spectroscopy: Limitations of peak-fitting method. Front. Energy Res. 5. Matthews, M.J., Pimenta, M.A., Dresselhaus, G., Dresselhaus, M.S., Endo, M., 1999. Origin of dispersive effects of the Raman D band in carbon materials. Phys. Rev. B 59, 6585–6588. Meier, R.J., 2005. On art and science in curve-fitting vibrational spectra. Vib. Spectrosc. 39, 266–269. Peters, K.E., 1986. Guidelines for evaluating petroleum source rock using programmed pyrolysis. Am. Assoc. Pet. Geol. Bull. 70, 318–329. Peters, K.E., Cassa, M.R., 1994. Applied source-rock geochemistry. In: Magoon, L.B., Dow, W.G. (Eds.), The Petroleum System. From Source to Trap. American Association of Petroleum Geologists, Tulsa, pp. 93–120. Rahl, J.M., Anderson, K.M., Brandon, M.T., Fassoulas, C., 2005. Raman spectroscopic carbonaceous material thermometry of low-grade metamorphic rocks: Calibration and application to tectonic exhumation in Crete, Greece. Earth Planet. Sci. Lett. 240, 339–354. Romero-Sarmiento, M.F., Rouzaud, J.N., Bernard, S., Deldicque, D., Thomas, M., Littke, R., 2014. Evolution of Barnett Shale organic carbon structure and nanostructure with increasing maturation. Org. Geochem. 71, 7–16. Ryder, R.T., Hackley, P.C., Trippi, M.H., Alimi, H., 2013. Evaluation of thermal maturity in the lowmaturity Devonian shales of the northern Appalachian Basin. Am. Assoc. Pet. Geol. Search Discov. 10477, 67. http://www.searchanddiscovery.com/ documents/2013/10477ryder/ndx_ryder.pdf. Sauerer, B., Craddock, P.R., Aljohani, M.D., Alsamadony, K.L., Abdallah, W., 2017. Fast and accurate shale maturity determination by Raman spectroscopy measurement with minimal sample preparation. Int. J. Coal Geol. 173, 150–157. Schito, A., Romano, C., Corrado, S., Grigo, D., Poe, B., 2017. Diagenetic thermal evolution of organic matter by Raman spectroscopy. Org. Geochem. 106, 57–67. Schmidt Mumm, A., Inan, S., 2016. Microscale organic maturity determination of graptolites using Raman spectroscopy. Int. J. Coal Geol. 162, 96–107. Spötl, C., Houseknecht, D.W., Jaques, R.C., 1998. Kerogen maturation and incipient graphitization of hydrocarbon source rocks in the Arkoma Basin, Oklahoma and Arkansas: a combined petrographic and Raman spectrometric study. Org. Geochem. 28, 535–542. Steele, A., et al., 2012. A reduced organic carbon component in Martian basalts. Science 337, 212–216. Tuinstra, F., Koenig, J.L., 1970. Raman spectrum of graphite. J. Chem. Phys. 53, 1126–1130. Wang, Y., Alsmeyer, D.C., McCreery, R.L., 1990. Raman spectroscopy of carbon materials - Structural basis of observed spectra. Chem. Mater. 2, 557–563. Wilkins, R.W.T., Boudou, R., Sherwood, N., Xiao, X., 2014. Thermal maturity evaluation from inertinites by Raman spectroscopy: the “RaMM” technique. Int. J. Coal Geol. 128–129, 143–152. Wilkins, R.W.T., Wang, M., Gan, H., Li, Z., 2015. A RaMM study of thermal maturity of dispersed organic matter in marine source rocks. Int. J. Coal Geol. 150–151, 252–264. Wilkins, R.W.T., Sherwood, N., Li, Z., 2018. RaMM (Raman maturity method) study of samples used in an interlaboratory exercise on a standard test method for determination of vitrinite reflectance on dispersed organic matter in rocks. Mar. Pet. Geol. 91, 236–250. Wopenka, B., Pasteris, J.D., 1993. Structural characterization of kerogens to granulitefacies gaphite: Applicability of Raman microprobe spectroscopy. Am. Mineral. 78, 533–557. Xueqiu, H., Xianfeng, L., Baisheng, N., Dazhao, S., 2017. FTIR and Raman spectroscopy characterization of functional groups in various rank coals. Fuel 206, 555–563. Yang, J., Hatcherian, J., Hackley, P.C., Pomerantz, A.E., 2017. Nanoscale geochemical and geomechanical characterization of organic matter in shale. Nat. Commun. 8, 1–9. Zhou, Q., Xiao, X., Pan, L., Tian, H., 2014. The relationship between micro-Raman spectral parameters and reflectance of solid bitumen. Int. J. Coal Geol. 121, 19–25.

Notes The authors declare no competing financial interest. All data used is available either within the manuscript or within the associated data release available free of charge at https://doi.org/10.5066/P9D5ASO0. Acknowledgements Ryan J. McAleer (USGS) and an anonymous reviewer are thanked for their reviews which greatly improved this manuscript. This work was supported by the Energy Resources Program of the U.S. Geological Survey. Use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government. References ASTM, 2011. D2797 Standard Practice for Preparing Coal Samples for Microscopical Analysis by Reflected Light. ASTM International, West Conshohocken, PA. ASTM, 2014. D7708–14 Standard Test Method for Microscopical Determination fo the Reflectance of Vitrinite Dispersed in Sedimentary Rocks. ASTM International, West Conshohocken, PA. Baldock, J.A., Skjemstad, J.O., 2000. Role of the soil matrix and minerals in protecting natural organic materials against biological attack. Org. Geochem. 31, 697–710. Baludikay, B.K., et al., 2018. Raman microspectroscopy, bitumen reflectance and illite crystallinity scale: comparison of different geothermometry methods on fossiliferous Proterozoic sedimentary basins (DR Congo, Mauritania and Australia). Int. J. Coal Geol. 191, 80–94. Beny-Bassez, C., Rouzaud, J.-N., 1985. Characterization of carbonanceous materials by correlated electron and optical microscopy and Raman microspectroscopy. Scan. Electron Microsc. 1, 119–132. Beyssac, O., Goffé, B., Chopin, C., Rouzaud, J.N., 2002. Raman spectra of carbonaceous material in metasediments: a new geothermometer. J. Metamorph. Geol. 20, 859–871. Beyssac, O., Goffé, B., Petitet, J.-P., Froigneux, E., Moreau, M., Rouzaud, J.-N., 2003. On the characterization of disordered and heterogeneous carbonaceous materials by Raman spectroscopy. Spectrochim. Acta 59, 2267–2276. Birdwell, J.E., Wilson, S.A., 2017. Development of new shale geochemical reference materials: Results from Boquillas shale interlaboratory testing and other status updates. In: Geological Society of America Annual Meeting, (Paper No. 69-8, Denver, CO). Bonoldi, L., Di Paolo, L., Flego, C., 2016. Vibrational spectroscopy assessment of kerogen maturity in organic-rich source rocks. Vib. Spectrosc. 87, 14–19. Cheshire, S., Craddock, P.R., Xu, G., Sauerer, B., Pomerantz, A.E., McCormick, D., Abdallah, W., 2017. Assessing thermal maturity beyond the reaches of vitrinite reflectance and Rock-Eval pyrolysis: a case study from the Silurian Qusaiba formation. Int. J. Coal Geol. 180, 29–45. Childress, L.B., Jacobsen, S.D., 2017. High-pressure high-temperature Raman spectroscopy of kerogen: Relevance to subducted organic carbon. Am. Mineral. 102, 391–403. Dow, W.G., 1977. Kerogen studies and geological interpretations. J. Geochem. Explor. 7, 79–99. Ferrari, A.C., 2007. Raman spectroscopy of graphene and graphite: Disorder, electron–phonon coupling, doping and nonadiabatic effects. Solid State Commun. 143, 47–57. Ferrari, A.C., Robertson, J., 2001. Resonant Raman spectroscopy of disordered, amorphous, and diamondlike carbon. Phys. Rev. B Condens. Matter 64, 1–13. Hackley, P.C., Cardott, B.J., 2016. Application of organic petrography in north American shale petroleum systems: a review. Int. J. Coal Geol. 163, 8–51. Hackley, P.C., Lewan, M., 2018. Understanding and distinguishing reflectance measurements of solid bitumen and vitrinite using hydrous pyrolysis: implications to petroleum assessment. Am. Assoc. Pet. Geol. Bull. 102, 1119–1140. Hackley, P.C., et al., 2015. Standardization of reflectance measurements in dispersed organic matter: results of an exercise to improve interlaboratory agreement. Mar. Pet. Geol. 59, 22–34. Hackley, P.C., Valentine, B.J., Hatcherian, J., 2018. On the petrographic distinction of bituminite from solid bitumen in immature to early mature source rocks. Int. J. Coal Geol. 196, 232–245. Henry, D.G., Jarvis, I., Gillmore, G., Stephenson, M., Emmings, J.F., 2018. Assessing lowmaturity organic matter in shales using Raman spectroscopy: Effects of sample preparation and operating procedure. Int. J. Coal Geol. 191, 135–151. Jehlička, J., Bény, C., 1992. Application of Raman microspectrometry in the study of structural changes in Precambrian kerogens during regional metamorphism. Org. Geochem. 18, 211–213. Jorio, A., Dresselhaus, M., Saito, R., Dresselhaus, G.F., 2011. Raman Spectroscopy in Graphene Related Systems. WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany.

9