Biomarker evidence for the depositional environment of basinal UK Mississippian mudstones

Supplementary material: Tables of biomarker ratios and programmed pyrolysis data for each sample are available at https://doi.org/10.6084/m9.figshare.c.6988037

The Carboniferous geology of northern England provides a record of an ancient epicontinental seaway. The lithofacies recorded from the shelf, basin and deltas of the seaway has hosted hydrocarbon, metal and coal resource plays, which in turn has directed industrial endeavours as well as scientific enquiry across northern England (Waters et al. 2020a). Inspired by gas-in-place (GIP) estimates of 822–2281 TCF (Andrews 2013), the most recent industrial interest was focused on sources of unconventional gas from shale. The target source rocks were the Bowland Shale and Hodder Mudstone formations. Understanding the controls on organic matter type and its distribution in the prospective rocks has been identified as a key characteristic to determine the potential of unconventional resources (Andrews 2013). Different studies have identified differing controls on the organic matter type. Using programmed pyrolysis data, mixed Type II/III kerogens have been inferred from comparison of S2 and total organic carbon (TOC) content (Gross et al. 2015; Waters et al. 2020b). However, studies on kerogen kinetics and palynology have attested to Type II and I kerogen generative capability (Yang et al. 2016; Hennissen et al. 2017; Newport et al. 2020). Environmental factors controlling water column stratification, redoxclines and sediment source have been cited as generating variability in the potential of the formations by preferentially preserving organic matter (Gross et al. 2015; Emmings et al. 2020a); understanding these is vital to determine the Mississippian basinal setting and resources sourced.

The main unconventional target in the Carboniferous succession of northern England is the Bowland Shale Formation, defined as a grouping of the regional formations of the Holywell, Edale and Bowland Shale formations (Waters et al. 2009). The formation is described as dark-grey fissile to blocky hemipelagic mudstones, which transitioned with decreasing age from a calcareous to siliciclastic mineralogy (Waters et al. 2009, 2020b; Clarke et al. 2018). Other relevant formations include the Morridge formations, which have been used as analogue studies or alternative targets, despite these formations having a greater siliciclastic content (i.e. Könitzer et al. 2014; Gross et al. 2015; Słowakiewicz et al. 2015; Yang et al. 2016; Hennissen et al. 2017; Whitelaw et al. 2019). Data from microscale sealed vessel (MSSV) pyrolysis on the source rock demonstrated a promising Type II kerogen productive capability (Yang et al. 2016). Recent resource estimates derived from high-pressure water pyrolysis on Arnsbergian and Pendleian samples have reduced the GIP estimates to 140 ± 55 TCF, further limited by a 10% economic recovery for northern England (Whitelaw et al. 2019). The source rock facies were confined primarily within half-graben basins; however, deposition extended onto the shelves in association with maximum flooding surfaces (Waters et al. 2020b). This was calculated to have occurred in a 110 ka cycle for flooding surfaces with closely related faunal assemblages, and a 400 ka cycle is typically associated with more significant faunal change (Waters and Condon 2012). The combined influence of sea-level fluctuation with proximity to delta systems generated lateral and stratigraphic heterogeneity in the preserved organic fraction (Gross et al. 2015; Emmings et al. 2020b).

Emmings et al. (2020a) determined that the chemistry of the Pendleian water column in the Craven Basin had an oscillating palaeoredox zone that maintained euxinic conditions during transgressive-highstand sea-level states, becoming more oxidized with falling or low sea-level states. Similarly, Arnsbergian mudstones from the distal Widmerpool Gulf were deposited under anoxic conditions that extended to the photic zone during high sea levels, with low sea-level associated turbidite sediments associated with increased volumes of terrestrial organic matter, reduced salinity and more dysoxic to oxic conditions prevailing in the water column (Gross et al. 2015).

Oxic or aerobic water is classified as containing over 2 ml l⁻¹ of oxygen, representing well-ventilated water inducing no biological stress on fauna and enabling degradation of dispersed organic matter in sediment (Algeo and Li 2020). Anoxia represents the total lack of dissolved oxygen, conditions that are often associated with the biological use of sulfate in oxidation–reduction reactions (Algeo and Li 2020). Anoxic conditions are ferruginous if dissolved Fe²⁺ is present, or euxinic if H₂S is present instead (Algeo and Li 2020). Dysoxia is a commonly applied term in geoscience that relates to the impact of reduced oxygen levels (2–0.5 ml l⁻¹) to fauna distribution and diversity and organic enrichment (Tyson and Pearson 1991). Sub-oxic is used to describe water with very low (0.7–0 ml l⁻¹) oxygen concentrations, where the depositional environments lack significant bioturbation and macro-fossils (Breck 1974; Tyson and Pearson 1991) and reduction reactions involving nitrate, manganese or iron are facilitated by microbial organisms.

Biomarkers represent the biomolecular record of organisms existing within the depositional regime in which sedimentary rocks are deposited. Some biomarkers such as pristane/phytane (Pr/Ph) may be interpreted as indicative of palaeoredox conditions (Didyk et al. 1978). Published values for Pr/Ph ratios are shown in Figure 1 where oxic waters are represented by values over 3, dysoxic–sub-oxic as values between 2 and 0.8, while anoxic to euxinic values will be lower than 0.8. Other biomarkers such as aryl isoprenoids have been proven to represent the presence of water column euxinia that extends to the photic zone (Schwark and Frimmel 2004; Gross et al. 2015). The change of aryl isoprenoids has also been included in Figure 1. High levels of C₃₅ homohopanes have also been related to the presence of anoxia, as the hopanes are more likely to have originated from bacterial groups. Ts/Tm as a biomarker index may also demonstrate Eh effects at the sediment–water interface if it is anoxic (Peters et al. 2005).

In the current study, 201 samples from the Central Pennines and Craven Basin were analysed to develop a spatially distributed dataset on biomarker environmental indices, representing the first study to analyse organic geochemical trends across multiple boreholes dated from the Brigantian to the Arnsbergian–Chokierian boundary. To relate the study to other regional works, programmed pyrolysis was undertaken on all samples to understand the bulk geochemical source rock character of Mississippian mudstones. Biomarker results were produced from extractable organic matter (EOM) to delineate changes pertaining to salinity, water column anoxia and the provenance of organic matter across the basins.

The Carboniferous basins of northern England began rifting in the Devonian, relating to collapse of the Caledonian Orogen following the closure of the Iapetus Suture (Fraser and Gawthorpe 2003). The Late Devonian–Carboniferous subduction of the Rheno-Hercynian oceanic crust contributed to additional thermal and magmatic input, generating crustal stretching and Dinantian rift basin formation (Leeder 1988; Timmerman 2004). Far-field orogenic influences contributed transtensional forces enabling periodic eastward extension through which compression is accommodated via ‘crustal escape’ or extrusion mechanisms (Maynard et al. 1997; Fraser and Gawthorpe 2003).

Rifting initiated over continental conditions, with subsidence enabling the formation of lacustrine and transgressive horizons within the Devonian sequences. By the Early Carboniferous, marine to periodically sub-aerial conditions were established, depositing the Carboniferous Limestone Supergroup of Waters et al. (2009). The earliest stratigraphy was dominated by regional shallow-water carbonates. This depositional environment persisted on the shelves while the basins rifted, forming ramp and slope architectures into the depocentres (Riley 1990; Fraser and Gawthorpe 2003). The margins would periodically collapse into the depocentres forming a framework of calciturbidites and other transport deposits interbedded with calcareous mudstones.

The Bowland Shale Formation and its diachronous counterparts were deposited under late synrift and early post-rift conditions, onlapping the carbonate margin sediments (Waters et al. 2020b). The source rock interval was confined to the basin depocentres during the Dinantian, but ‘drowning’ of palaeohighs enabled draping of the source rock onto the carbonate highs (Kirby et al. 2000; Waters et al. 2020b). Palaeowater depths of depocentres were interpreted to be >300 m by Fraser and Gawthorpe (2003).

Siliciclastic turbidites acted as precursors to basin infill by the deltaic Millstone Grit Group (Waters et al. 2020b). Infill of the basins occurred in a stepwise staged process as carbonate highs provided a barrier directing sedimentation through erosional valley systems (Fraser and Gawthorpe 2003; Waters et al. 2020b), generating diachroneity in the age at the top of the Bowland Shale. Within the Millstone Grit Group, periods of deltaic influence were broken by the deposition of organic-rich mudstones such as the Caton Shales (Clarke et al. 2018), extending source rock deposition into at least the late Arnsbergian (Waters et al. 2020b). Within the Widmerpool Gulf, Arnsbergian-age sediments are represented by the Morridge Formation, which includes distal turbidites and hemipelagic mudstones (Waters et al. 2009). Brigantian to Pendleian mudstones from the Craven Basin contain a relatively high detrital mineral fraction and an associated mixed Type II/III organic matter composition, particularly in the Upper Bowland Shale (Clarke et al. 2018). Palynological interpretation of their Widmerpool Gulf counterparts revealed a plethora of microbially degraded organic matter types and evidence of algal groups such as Botryococcus spp. (Könitzer et al. 2016; Hennissen et al. 2017; Hennissen and Montesi 2020).

Boreholes held at the National Geological Repository core store in Keyworth were studied during the project. Cores were selected based on whether they contained representative samples of Carboniferous mudstones. The cores were required to have been tied to the goniatite biostratigraphy of the UK Carboniferous, through either core logs available via the BGS GeoIndex, industry reports available through the UK Onshore Geophysical Library (UKOGL), academic papers or PhD theses. Core chips were taken where possible at regular intervals through the cores.

Core chips were washed with n-pentane to remove any surface labile hydrocarbons and contaminants. The samples were powdered using a ceramic mortar and pestle to an approximate size of <30 µm. A single powdered aliquot for the current study was produced, weighing c. 3–5 g. This was split as required for Rock-Eval 6, and solvent extraction ahead of biomarker analysis.

Programmed pyrolysis is a commercially available standard methodology for studying source rock quality and richness through the measurement of produced volatile products such as hydrocarbons, CO₂ and CO, from the staged heating of sample material (Espitalié et al. 1977; Lafargue et al. 1998). The measurements of pyrolytically released volatiles are often compared to determine the kerogen types and whether the rock is prone to producing gas or oil (Jarvie 2012; Peters et al. 2015). Furthermore, the temperature at which maximum hydrocarbon generation occurs (T_max) is known to relate to the maturity of samples as an alternative to vitrinite reflectance (Espitalié 1986). The use of the method formed a central focus of many Bowland Shale specific studies (i.e. Gross et al. 2015; Hennissen et al. 2017; Newport et al. 2020; Waters et al. 2020b); thus the analysis from this project provides a direct comparison to other regional studies. This project made use of the Rock-Eval 6 pyrolysis and oxidation capabilities hosted at the British Geological Survey to standardize and compare measurements with published results such as Waters et al. (2020b).

Powdered aliquots of samples were analysed using a Rock-Eval 6 analyser configured in standard mode, pyrolysis and oxidation as a serial process (as in Waters et al. 2020b). Powdered rock samples were heated isothermally and maintained for 3 minutes at 300°C to release free hydrocarbons. Subsequently, a heating ramp of 25°C per minute was used to increase the temperature from 300 to 650°C, cracking non-volatile organic matter in an inert atmosphere of N₂. Hydrocarbons released from this process were measured using a flame ionizing detector (FID), and CO with CO₂ were measured using an infrared (IR) cell. From the pyrogram FID and IR peaks, the S1, S2 and S3 parameters were given and used to calculate hydrogen index (HI), oxygen index (OI) and production index (PI) as per the equations in Lafargue et al. (1998). After pyrolysis, the samples were oxidized from 300 to 850°C to determine the residual carbon in the rock, which was used with data from the S2 and S3 peaks to derive the TOC content.

Around 1 to 1.5 g of powdered aliquot were added to clean test tubes, with 10 μl of a prepared internal standard of squalane and p-terphenyl; 4 ml of dichloromethane (DCM)/methanol 93:7 v/v were added to the samples, which were subjected to 5 minutes of sonication and centrifuged for 3 minutes at 1.5 rpm. The resultant supernatant was transferred into a new test tube and the process repeated four times per sample. Copper turnings were added to the supernatant and the volume was reduced and replaced with DCM:n-hexane 1:1 v/v mixture before the aliphatic and aromatic fractions were eluted using DCM:n-hexane 1:1 v/v by column chromatography using activated alumina powder. The fraction was further concentrated to 0.5 ml.

Analysis was conducted using an Agilent Technologies 7890A gas chromatography (GC) system coupled to a 5975C mass spectrometer (MS). The GC injector was run in a splitless mode (1 μl) with the column flow rate set at 1.1 ml min⁻¹. A J&W scientific DB-5MS capillary column was used for compound separation and helium employed as a carrier gas. The GC oven was started at 40°C for 2 minutes; this was raised at a rate of 5°C min⁻¹ to 310°C and held for 14 minutes. Mass spectra were acquired in electron impact mode in the scan range of 50–500 a.m.u. The data was qualitatively and quantitatively studied using the Mass Hunter workstation software package (Agilent Technologies Inc. 2016). Biomarkers were identified and indices were determined by comparative investigation of peak areas for each chromatograph produced. A list of identified biomarkers is provided in Table 1.

X-ray diffraction (XRD) analysis was completed using a PANalytical X'Pert Pro Alpha-1 system hosted at the Natural History Museum, London. The unit was set up with a cobalt source tube that recorded a diffraction angle range from 4° to 90°. For each analysis, 1 g of powdered sample material was backloaded into metal holders. Mineral peaks were picked using the XPERT Highscore Plus software package (Degen et al. 2014), and samples were quantified utilizing Profex open source (Döbelin and Kleeberg 2015) XRD Rietveld refinement software.

The 201 samples included in the current study were taken between nine boreholes in northern England, either at equivalent ages or a range of depths to determine variation in biomarker trends temporally and spatially. The distribution of cores v. time is illustrated in Figures 2 and 3. From core material dated to the Arnsbergian or younger, 141 samples were taken, mostly focusing on the Carsington Dam Reconstruction-C4 (n = 53) and the Wray boreholes (n = 47). The Pendleian to Brigantian had a total of 60 samples taken between four boreholes. The distributions of samples are shown in Figures 2a and 3a showing well locations, and Figures 2b and 3b demonstrating sampling locations against sedimentary logs.

In the Wray geothermal borehole, the E_2b marine band is marked by the identification of Cravenoceras subplicatum at c. 120.45 m (Brandon et al. 1998). Beneath the E_2b marine band, the Wards Sandstone representing a delta top would have eroded any coeval representation of the E_2a3 Eumorphoceras yatesae marine band. The underlying siltstone facies represent the Close Hill Siltstone and may contain the E_2a2b marine band of Brandon et al. (1995). This borehole represents Arnsbergian strata in the Craven Basin, allowing correlation with datasets from the Widmerpool Gulf equivalents.

The Croft House borehole sited upon the Askrigg Block encountered similar strata to the Wray geothermal borehole. The E_2b marine band is attributed to 40–42 m depth as in Brandon et al. (1995). This is positioned above the Red Scar Grit and E_2a equivalent shales that relate to the Wards Sandstone and Close Hill Siltstones of the Wray geothermal borehole.

In the Carsington Dam Reconstruction-C4 borehole, the E_2b marine band is recognized through the presence of Cravenoceratoides edalensis from 22.52 m, with cravenoceratid fragments first appearing from 26.67 m. The E_2a3 marine band is present in this borehole and was logged by Könitzer et al. (2014) from c. 27.9 to 33.85 m. Both the E_2a2a and E_2a2b marine bands were first correlated between the Carsington Dam locality, the Duffield borehole and the Bowland basin by Brandon et al. (1995); Könitzer et al. (2014) logged an unidentified E_2a2 marine band from 52.6 to 53.1 m in the borehole. Based on published logs this is likely to be the E_2a2b marine band (Brandon et al. 1995). At the Carsington locality, the borehole records no delta top sandstone and represents a distal setting with consistent marine conditions influenced by siliciclastic turbidites (Könitzer et al. 2014). Rempstone-1 was also logged and studied, which represents E₂ mudstones in a turbiditic setting.

The logged section of the Ipstones Edge borehole contains the Isohomoceras subglobosum H_1a1 and the E_2c2–4 Nuculoceras nuculum marine bands (Chisholm et al. 1988; Morton et al. 2015). The lower section of the borehole is divided by the Hurdlow Sandstones, which are described as a turbiditic unit found in the Goyt Trough and western Widmerpool Gulf (Morton et al. 2015). The Hurdlow Sandstones overlie an undated mudstone, which is tentatively assigned to the E_2b3 zone.

Waters et al. (2020b) highlighted that, in the Widmerpool Gulf, only samples below the E_1b2 Tumulites pseudobilinguis marine band can be considered as being representative of the Bowland Shale Formation and that strata younger than that horizon should be considered as representative of the Morridge Formation. To test similarities of conditions within the Widmerpool Gulf, the Lees Farm borehole was studied. From borehole reports, the Lees Farm borehole has the E_1a marine band sited at 12.7 m depth, recognized by Emstites leion at 11.18 m depth. Below this at 23.30 m depth, Neoglyphioceras sp. represents P₂ zone Brigantian age strata diachronous with the Lower Bowland Shale. The Wardlow Mires boreholes were sited on the Derbyshire High, recording strata from the E_2aEumorphoceras bisulcatum zone to the P₂ zones represented by Neoglyphioceras sp. and Sudeticeras sp.

The Marl Hill-4 borehole has been extensively studied, and the biostratigraphic positioning is based on the resulting publications (see Emmings et al. 2019, 2020a, b). This puts the E. leion E_1a zone at c. 17 m depth, marking the boundary of the Upper and Lower Bowland Shale as they appear in the Craven Basin. Cominco-S3 provides an equivalent representation of this boundary at 13.72 m depth for comparison. At 31.50 m depth in the Cominco-S3 borehole, the P_2b to P_2c boundary was interpreted based on the upper observation of Neoglyphioceras sp. representing the Middle Brigantian.

The whole dataset of samples has a mean average TOC content of 3.14%, which ranges from 0.42% in the Croft House borehole to 20.16% in the Wardlow Mires core. Splitting the sample set into regions gives an average TOC content of 3.08% in the Craven Basin and 3.20% in the Central Pennines. In terms of the cores sampled in the Craven Basin, Cominco-S3 had the lowest average at 2.63% and Croft House had the highest average TOC at 4.37%. Comparatively, in the East Midlands, Ipstones Edge had the lowest of sampled wells at 1.80%, while Rempstone-1 had the highest mean TOC content of 5.74%. Figures 2 and 3 show TOC contents for the different boreholes; comparatively, the values show an overall greater abundance of TOC in the Widmerpool Gulf in comparison to the Craven Basin.

The S2 results from the total data give an average value of 5.52 mgHC g⁻¹ rock. In the Craven Basin, the average value decreases to 2.92 mgHC g⁻¹ rock, ranging from 1.70 mgHC g⁻¹ rock in the Wray borehole to a high average of 5.61 mgHC g⁻¹ rock in Marl Hill-MHD4. In the Central Pennines, the mean S2 is 7.60 mgHC g⁻¹ rock, ranging from 1.86 mgHC g⁻¹ in the Ipstones Edge borehole to 25.53 mgHC g⁻¹ rock in Rempstone-1.

T_max results vary across the sample set between 421°C in the Ipstones Edge borehole and 464°C in the Cominco-S3 borehole. The mean value across all samples is 438°C. T_max was found to vary significantly within boreholes; for example, the Wardlow Mires core has a minimum T_max of 427°C at 8.71 m measured depth (MD) to 452°C at 105 m MD (Fig. 3b). Such a variation in T_max results is assumed to be incorrect for a standard geothermal gradient, which is attributed to lower S2 values associated with turbiditic and deltaic samples as in Figures 2b, c and 3b, c.

An average HI value of 144 mgHC g⁻¹ TOC was calculated for the dataset (Figs 2c & 3c), split between a mean of 196 mgHC g⁻¹ TOC in the East Midlands samples and 88 mgHC g⁻¹ TOC in Lancashire–Yorkshire. The Wray borehole had the lowest HI at 66 mgHC g⁻¹ TOC (Fig. 2b), while Rempstone-1 had the highest mean at 403 mgHC g⁻¹ TOC (Fig. 3b). Plotting HI against T_max (Figs 2c & 3c) to determine maturity and kerogen type demonstrates a distribution of data points in the immature to mid oil window, varying between Type II and III kerogen classifications. From the S3 data, OI was calculated for all samples with a mean value of 19. Within the dataset ten samples have OI values over 40 mgCO₂ g⁻¹ TOC.

Both pristane and phytane were detected in all samples; the ratio of these ranged from 0.62 to 6.92 with a mean value around 1.55, which correlates with values previously published from the Craven Basin (Emmings et al. 2020a). Most samples were distributed within values of 0.8–1.8 (Figs 4b & 5b). Comparing the isoprenoids in Figures 4a and 5a to their n-alkane counterparts n-C₁₇ and n-C₁₈, most samples have a ratio of less than 1. Samples mostly from the boreholes Croft House and Ipstones Edge plot above 1, alongside many of the Lees Farm borehole samples. The majority of samples fall into a mixed organic matter type and a mixed oxidizing–reduced setting according to those ratios.

The aliphatic content of the samples' EOM contained a suite of n-alkanes from n-C₁₁ to n-C₃₅, although the higher abundances generally ranged from n-C₁₃ to n-C₃₃. The percentile distribution of n-alkanes is orientated towards C_11–20, with C_21–25 representing the second highest distribution. This is reflected in calculated carbon distributions such as the terrigenous to aquatic ratio (TAR) in which most data points range below a ratio of 1 (Figs 4b & 5b). Carbon preference index (CPI) results range within the boreholes as shown in Figure 4b but modally plot close to a value of 1 for the Brigantian–Pendleian samples and are slightly elevated for Arnsbergian sample material (Fig. 5b), indicating a slight odd preference consistent with additional terrigenous organic matter.

The 191 terpane chromatogram is dominated by hopanes. Tricyclic and tetracyclic terpanes are also present, as shown in Figure 6, but were not quantified. 17α-trisnorhopane (Tm) and 18α-trisnorneohopane (Ts) were detected in all samples with a slight preference towards Tm in the peak heights. On average the C₃₀ 17α-21β-hopane homologue has preference over C₂₉αβ, while the C_31–35 homohopane series has a stepwise downward trend through the homologues from C₃₁ to C₃₅. This produced an average C₃₅ homohopane index of 7% rising to a maximum of 16% in the Carsington Dam Reconstruction-C4 borehole and a minimum of 4% in the Wray borehole. Comparing hopane homologue ratios shows a mean C₃₁R/C₃₀ of 0.30, a mean C₃₅/C₃₄ of ratio of 0.31 and for C₂₉/C₃₀, 0.77. Hopane isomerism shows a preference in the chromatograms towards the S isomer; this is reflected in the 22S/(22S + 22R) indices, which average 0.59 for C₃₁, 0.57 for C₃₂ and 0.54 for C₃₃. C₃₅ hopane isomerism averages significantly lower than the other indices owing to low concentrations and thus is not interpreted further. C₃₄ isomerism is higher, averaging 0.64 for the whole sample set.

Rearranged hopanes were also observed in trace quantities. The C₂₉ 18α(H)-30-norneohopane (C₂₉Ts) coeluted with C₂₉ 17α-21β-hopane and was quantified by the area of the peak shoulder effect produced. The C₃₀ 18α(H)-30-neohopane (C₃₀Ts) was identified by its relative retention indices to the C₂₉Ts shoulder and C₃₀αβ hopane. C₃₀ and C₂₉ 17α-diahopanes were widely observed. Trace doublet peaks between the C₃₂–C₃₅ homohopane doublets (Fig. 6) may represent the presence of a diahopane homologue series. Additionally, a small peak representing a trace amount of C₂₈ 28-norhopane was commonly observed eluting between C₂₉ diahopane and C₂₉αβ/Ts; it was confirmed through a m/z 355 extracted ion chromatogram (EIC).

In addition to the hopane series, gammacerane was also widely detected, with a relatively small peak area (Fig. 6) in the samples analysed. The gammacerane index (GI) was calculated as 10 × gammacerane/(gammacerane + C₃₀ hopane). The average value for the sample set is c. 1. The range of values includes a maximum of 2.68 and minimum of 0.34; both values were recorded from the Carsington Dam Reconstruction-C4 borehole, but most samples across the entire set distribute between 0.5 and 1.5 GI. GI varies stratigraphically; enhanced levels occur around marine bands found in Cominco-S3, Marl Hill-4, Ipstones Edge and Carsington Dam Reconstruction-C4.

The most common steranes in the EOM analysed were the C₂₇–C₂₉ sterane homologues (Fig. 7). Of the steranes quantified there was a preference in the samples towards C₂₉, which contributes a mean 43% of the total C_27–29 sterane homologue content. The C₂₈ sterane accounts for the lowest percentile of the steranes analysed. C₂₇ has a mean of c. 35% of the total sterane homologue content. In the ternary plots in Figures 4a and 5a the samples show a central distribution between C₂₇ and C₂₉, which fits into a mixed planktonic/bacterial/land plant organic matter zone.

Comparison of the m/z 217 and 259 EICs in Figure 7 demonstrated that a number of peaks represent diasteranes, or a coeluted mix of steranes and diasteranes. Affected sterane peaks observed include but are not limited to 27αββR, 27αααS and 28αββR. The coelution of compounds reduced the accuracy of quantification, distorting indices reported from those peak areas. For the purposes of testing the impact of clay mineralogy on organic matter alteration, the ratio of 27βα (S + R) diasterane to 27αααR sterane was used in place of conventional ratios. This was due to the fact these peaks have been influenced the least by coelution. For comparison of sterane abundances the αααR sterane peak was used.

Some lower mass peaks visible in the m/z 217 EIC were tentatively assigned to the C_22–33 sterane/diasterane homologues, but were not interpreted in the current study. After the C₂₉ααα20R sterane there are a number of peaks indicative of C₃₀ steranes (Fig. 7). Interpretation of the C₃₀ sterane peaks was tentative as the peaks coeluted with other compounds; thus they were not quantified for the current study.

Peaks representing a suite of aryl isoprenoid compounds with carbon numbers from C₁₁ to C₂₇ were recognized through comparison of the m/z 133 and 134 EICs as in Figure 8. Peak areas for some of these compounds were quantified to produce the aryl isoprenoid ratio (AIR) after Schwark and Frimmel (2004), calculated as (C_13–17)/(C_18–22). Of the 200 samples analysed through GC-MS methods, only 123 samples had aryl isoprenoid peaks. None of the samples from the Warlow Mires-1 and Cominco-S3 boreholes had quantifiable peaks, and only four samples from the Croft House core and 12 from Wray were quantifiable. The AIR across all measured samples ranges from 0.16 in the Ipstones Edge borehole to a maximum of 16.78 in the Carsington Dam Reconstruction-C4 borehole. Marl Hill and Lees Farm boreholes made up the lower range values, while Carsington Dam Reconstruction-C4 and Rempstone-1 contributed the most to the higher range. The precursor molecule for the aryl isoprenoids was not confirmed through isotope ratios; thus a β-carotane precursor cannot be ruled out. Neither β-carotane nor isoreniaretane were identified in any samples, and probably remain within the kerogen structure owing to the low maturity of material sampled.

As can be seen in Figure 8, the m/z 133 EIC shows a range of additional peaks alongside the aryl isoprenoids. Mass chromatograms were checked against other aromatic compounds, finding matches through the m/z 204 and 205 EICs showing coelution of C₁₅ isohexyl alkylaromatics. Additionally, through the m/z 246 EIC phenylalkanes were observed, C₁₈ phenylalkane in particular.

The XRD analysis, as summarized in Figure 9, showed a primarily clay to silicate rich composition. Two samples from Carsington Dam Reconstruction-C4 plotted as being carbonate rich. The distribution of data points matches those published by Hennissen et al. (2017). Two mineralogical populations were determined. Arnsbergian samples from Rempstone-1, Wray and Carsington Dam Reconstruction-C4 were found to commonly contain a clay-rich and carbonate-poor mineralogy. The second population is made of samples from Carsington Dam Reconstruction-C4, Lees Farm-1 and Marl Hill-MHD4; this plots under a siliceous mudstone category and matches average data for the Barnett Formation.

The clay mineral type comprises a mixture of expandable clays, classified through Profex as smectites combined with illite. Attempts to distinguish maturity levels were not possible owing to mixed layering of the expandable clay fraction. Kaolinite was commonly observed in the samples, but quantification was complicated as the peaks were shared with chlorite, which was mostly present in Arnsbergian-aged samples.

Carbonate mineralogy changes across the sample set dependent on location. Carsington Dam Reconstruction-C4 contained a mix of calcite and ankerite, which was also observed by Hennissen et al. (2017). Lees Farm was found to contain peaks correlating to dolomite. Marl Hill Moor-MHD4 contained only calcite. The Wray borehole only had calcite in two samples, which correlated to the marine band in the top of the cored section. Trace to 5% concentrations of siderite were observed in samples from Rempstone-1 and in turbidite facies belonging to the Wray borehole.

Pyrite was commonly measured in the samples. The highest pyrite contents were recorded in Marl Hill Moor-MHD4, which had a range from 1 to 11.64%. Pyrite was commonly associated with samples originating from marine facies. Very little pyrite was recorded in the Wray borehole, consistent with its common terrestrial kerogen classification. Anatase was also commonly observed in most samples but in low to trace abundances (<5%).

The average TOC contents for the sampled cores may be defined as ‘good’ to ‘very good’ (Peters 1986), with the average values above the proposed cut-off of 2% by Andrews (2013). According to the S2/TOC biplots in Figures 2c and 3c only 40 samples fell below the 2% cut-off. However, the volume of TOC that can generate hydrocarbons has a greater significance to source rock potential than the presence of any TOC.

Correlation of TOC content and S2 (Figs 2c & 3c) shows that the samples from each basin distribute slightly differently. The Bowland basin samples have a significantly lower S2 to TOC correlation than the samples from the East Midlands. Overall, the majority of data points align with published data, with samples analysed in this study matching argillaceous/terrestrial organo-facies (Newport et al. 2018; Emmings et al. 2020a). The Arnsbergian-aged samples (Fig. 3c) closely matched trends from the Duffield and Carsington Dam Reconstruction-C3 and C4 boreholes.

Our interpretation of the organic matter source from the S2–TOC plots in Figures 2c and 3c suggested that the samples were mostly composed of either ‘mixed’ type kerogens or Type III and IV. These kerogen types would not fit with the required standards for a shale play in the UK (Andrews 2013). HI–OI plots show that many of the samples plot close to the axis intercept for both sample sets, indicating a low generative potential (Fig. 2). This demonstrated some relationship to facies whereby samples from turbidite and hemipelagite facies had the lowest potential. Marine band facies were always found to have the highest HI to OI ratio. Data points typically follow the maturation trend lines for either Type II or Type III source rocks. Only a few samples exceed HI values of 300 mgHC g⁻¹ TOC, which would make for a suitable shale gas target (Andrews 2013).

The poor generative potential represented by HI or S2 values demonstrated that the TOC content present had a low pyrolytic potential. Palynological analysis of the Carsington Dam Reconstruction-C4 borehole (Könitzer et al. 2016) interpreted that dispersed organic matter was distributed by suspension settling of granular amorphous material. The organic matter has been reported to represent biotic degraded organic matter with observation of aliphatic-rich material supported in faecal matter (Emmings et al. 2019). Newport et al. (2020) interpreted the Bowland Shale Formation of the Marl Hill-11 borehole to have a high oil yield from primary generation based on relative proportions of Type I and II kerogens. Modelling and pyrolysis experiments on kerogen extracts from the Widmerpool Gulf showed high primary generation and expulsion; however, secondary gas generation was regarded as a significant contribution to the hydrocarbon yield (Yang et al. 2016). Hennissen and Montesi (2020) observed fluorescence and Type I macerals such as Botryococcus sp. in Arnsbergian-aged rocks, which would support the argument that Carboniferous shales have a high primary oil producing capability. The above studies did not report significant quantities of inertinite matter, which could be regarded as a candidate for enhanced TOC contents in low S2 yield samples.

Information about the environment of deposition extracted from programmed pyrolysis data is uncertain, presenting difficulties with estimating kerogen type and generative potential. Studying the molecular representation of organisms (biomarkers) is a means to determine the environmental conditions under which a source rock was deposited.

Using Pr/Ph as a palaeoredox index (Didyk et al. 1978), the mean average of the sample set (1.55) falls within the ‘dysoxic’ window. Comparison of these acyclic isoprenoids and their n-alkane counterparts (Figs 4a & 5a) supports the dysoxic interpretation. The mean of the dataset is similar to averages from the Marl Hill-4 borehole (Emmings et al. 2020a), the Duffield borehole (Gross et al. 2015) and Malton-4 (Słowakiewicz et al. 2015), highlighting that reducing conditions were prevalent across the Carboniferous seaways present in northern England. Predominance of C₃₁ over C₃₅ hopane homologues seen from chromatographs can indicate sub-oxic to oxic conditions, although the ratio is impacted by thermal maturity and mineralogy (Peters and Moldowan 1991).

Identification of trace amounts of aryl isoprenoids supports the evidence that reducing conditions were present and contributed to the water column chemistry (Fig. 10). Reducing conditions detected via low Pr/Ph supported more stable photic zone euxinia, evidenced by a prevalence of long-chain aryl isoprenoids in Figure 10 (Schwark and Frimmel 2004). Evidence of combined dysoxia and euxinia refers to zonation in the water column. During deposition, a thicker euxinic water column would have been present, representing poor ventilation of bottom waters. Organic matter produced by surface productivity was decomposed by sulfate-reducing bacteria at the sediment–water interface, recognized from samples at Marl Hill-4 (Emmings et al. 2020a). Samples representing more oxidizing environments in Figure 10 showed a predominance of shorter-chain aryl isoprenoids indicating that at locations where they could be detected, photic zone anoxia was likely to be episodic and organic matter was exposed to oxidative degradation for extended periods of time (Schwark and Frimmel 2004). Group 1 represents marine facies, i.e. marine bands, hemipelagite and carbonate slope, all associated with indicators of persistent euxinia extending to the photic zone. Turbidite facies form Group 2 on Figure 10, representing episodic photic zone anoxia consistent with ventilation of the water column during lowstand sea levels (Gross et al. 2015; Emmings et al. 2020a). Ventilation of the water column enabled degradation of the aryl isoprenoid chains (Schwark and Frimmel 2004) probably during the same periods that contributed to microbial sediment–water interface effects as in Emmings et al. (2020b).

Gammacerane supports the interpretation of a stratified water column (Sinninghe Damste et al. 1995); however, the values do not vary significantly through the dataset and no major peaks were detected from the 191 EICs. Gammacerane is commonly associated with bacteriovorous ciliates living beneath chemoclines (Harvey and McManus 1991; Sinninghe Damste et al. 1995) often connected with, but not limited to, hypersaline environments (Peters et al. 2005). The concentrations reported in the current study do not provide evidence for hypersaline conditions. Other influences must have generated stratification, i.e. thermal layering (Emmings et al. 2020a). GI v. Pr/Ph in Figure 4a shows that there is a slight relationship between Pendleian–Brigantian reducing conditions and water column salinity. Most samples shown in Figure 5a suggest that dysoxic saline conditions prevailed during the Arnsbergian.

From 51.6 m in the Carsington Dam Reconstruction-C4 log there is a significant excursion of GI values. This is following from the proposed location of the E_2a2b marine band at 52.6–53.1 m (Brandon et al. 1995; Könitzer et al. 2014). Gross et al. (2015) provided no data or interpretation on that minor cycle. The excursion coincides with the onset of low sea-level conditions and deposition of distal turbidites potentially connected to the uppermost Duffield Turbidite (Aitkenhead 1977). Könitzer et al. (2014) described bioturbated and lenticular facies indicating enhanced sediment–water interface oxygenation, proven here by relatively higher Pr/Ph. The excursion is interpreted as being representative of sea-level fall and the development of a substantial oxygenated freshwater lens across saline bottom waters in the Widmerpool Gulf. This is the only significant salinity stratification observed from the gammacerane data in the current study. Aryl isoprenoids were detected in this layer, with a relatively high AIR, ranging from 3.05 to 8.33, plotting in Group 2 on Figure 10. Any anoxia present would have been highly episodic with a ventilated water column; anoxia would have been driven by productivity in the surface-water layer (Emmings et al. 2020b).

Most samples have a Pr/n-C₁₇ and Ph/n-C₁₈ ratio of <1, which in this case was interpreted as being representative of an algal or microbial short-chain alkane origin, as was suggested for the Duffield borehole by Gross et al. (2015). Such a trend could be generated by maturity, but would not be expected within early oil window samples. Analysis of the distribution of the n-alkanes further supports the aquatic organic constituent source. The TARs reported in Figures 4b and 5b are <1 indicating an aquatic origin. Meanwhile the CPI is c. 1 for many samples, indicative of marine conditions and no preference in chain length. Locations such as Wray, Croft House and Rempstone have slightly higher CPI value distributions (>1), which would be consistent with a terrestrial organic contribution represented by the deltaic derived detrital sandstones logged from the core material available. The tentative observation of C₃₀ sterane peak areas in the samples is further evidence for a marine organic source (Moldowan et al. 1985; Peters 1986).

Predominance of the C₂₉ sterane isomers over other sterane groups in the samples was interpreted to be representative of terrestrial organic matter (Volkman 1986). However, the samples relate to a planktonic/bacterial/land plant window suggestive of organic mixing. Some of the potential aromatic molecule candidates present alongside the aryl isoprenoids on the m/z 133 EIC could be representative of isohexyl alkylbenzenes such as C₁₅H₂₄ 1,3,4-trimethyl-2-(4-methylpentyl) benzene. Isohexyl alkylbenzenes are indicative of higher plant or microbial input (Wang et al. 1990; Ellis et al. 1996), both of which are viable explanations based on the organic macerals representing bacterial degradation of terrestrial matter described by Könitzer et al. (2016).

In agreement with observations from Gross et al. (2015), the general organic source was mixed. Algal and bacterial markers such as short-chain alkanes, tricyclic terpanes and C₂₇ steranes provide a strong indicator of organic matter akin to Type II kerogens. This interpretation is consistent with palynological observation (Könitzer et al. 2016; Hennissen et al. 2017; Newport et al. 2020) and interpretation of the source rock activation energies (Yang et al. 2016; Newport et al. 2020). C₂₉ steranes present evidence of terrestrial organic matter that was biotically decomposed as it settled through a stratified water column. Samples from the Widmerpool Gulf show greater variation in the ratio of C₂₇ to C₂₉ than their counterparts in the Craven Basin (Figs 4a & 5a). Extracted steranes from the Craven Basin express a slight preference towards C₂₉, which is probably generated by proximity to the Pennine Delta front and the resultant consistency of terrestrial organic matter influx and preservation.

Based on the average T_max trends for the boreholes in Figures 2b and 3b, the samples analysed in this study are early oil to peak oil mature. This is supported by low vitrinite reflectance (% R_o) values for locations such as 0.55% R_o at Rempstone-1 (Coombes et al. 1986; Hennissen and Gent 2019), a calculated % R_o of 0.5–0.8 at the Carsington Dam locality (Henry et al. 2018) and similarly the Marl Hill locality's calculated % R_o of 0.8 (Herrmann et al. 2018). Some boreholes exhibited anomalous T_max ranges that were inconsistent with the geothermal gradient and burial depth range of the sampled interval; for example, Cominco-S3 in Figure 2b has a T_max range of 20°C in the upper 20–30 m of the cored section. As the samples are within the oil window, they are not interpreted directly as a shale gas resource and should not exhibit such high maturities.

Biomarkers also provide maturity data, through effects such as preferential thermal degradation or isomerism of compounds. Figure 11 shows biomarker maturity indices plotted against T_max, revealing a range of either correlating or contradicting relationships. Firstly, the hopane inventory (Fig. 11a) of the samples had reached hopane isomer equilibrium, indicating that the source rocks were at an oil expulsion maturity level. Moretane ratios were also determined to have achieved full maturity with no apparent relationship to T_max. However, Ts/Tm (Fig. 11c) shows no T_max relationship and a low to moderate maturation level.

Influences from the environment of deposition are the most likely cause of these trends. The moretane ratios indicate a predominance of C₃₀ hopane over moretane. Rullkötter and Marzi (1988) observed high moretane values in hypersaline rocks, while Isaksen and Bohacs (1995) observed enhanced moretane ratios in highstand as opposed to lowstand tracts, which was attributed to higher plant contributions. It would be reasonable to expect such a trend in the Carboniferous mudstones. However, the generation of moretane can also be attributed to bacteria in oxidizing sediments as well as diagenetic effects under certain clay types (Quirk et al. 1984; Uemura and Ishiwatari 1995; Wang 2007; French et al. 2012).

A predominance of Tm over Ts may relate to either immaturity, oxidizing environments or variations in clay influence (McKirdy et al. 1983; Waples and Machihara 1990; Peters et al. 2005; French et al. 2012). Figure 11c shows no relationship between T_max and Ts/Tm, highlighting the low maturity of the samples and emphasizing that alternative influences must have a control on the ratio. Ts/Tm has a complex relationship with Pr/Ph. In the Wray geothermal borehole (Fig. 5b), Ts/Tm increases above the E_2b marine band indicating enhanced reducing conditions at the sediment–water interface. This is likely to be connected to sulfate reduction during flooding of shelf or delta proximal areas as suggested by Emmings et al. (2020a). In the Widmerpool Gulf at the Carsington Dam Reconstruction-C4 locality, samples above the E_2b marine band have a reduced Ts/Tm correlating with reduced Pr/Ph consistent with a switch to carbonate pH effects in porewater (Fig. 1) (Moldowan et al. 1994). The Pendleian–Brigantian samples in Figure 3b show elevated Ts/Tm ratios consistent with widespread Eh effects in porewater during the Mississippian. Carbonate influences are demonstrated in the lower 10 m of Wardlow Mires.

Figure 12a highlights three populations of samples based on the relationship of Ts/Tm and diasteranes. Group 1 represents anoxic marine conditions whereby organic matter is deposited under euxinic conditions with a moderate flux of detrital clays during marine band and highstand sea-levels. Group 2 represents deposition under dysoxic to oxic conditions with samples commonly representing prodelta depositional conditions. Group 3 was defined by the presence of slope, hemipelagite and marine band conditions, representing a setting in which marine organic matter is deposited under euxinia alongside clay input.

The ratio of C₂₇ diasteranes to 27αααR steranes corelates positively to the full range of T_max values (Fig. 12b). As opposed to a maturity relationship, the corelation is representative of a mineral matrix effect on programmed pyrolysis results (Espitalié 1986; Peters 1986). The influence interpreted in the current study involved adsorption of free hydrocarbons onto mineral surfaces, decreasing pyrogram peaks and leading to a T_max shift to high temperatures or poor quality S2 peaks (Espitalié et al. 1984; Hu et al. 2014; Yang and Horsfield 2020). Research has suggested that samples with TOC contents over 2 wt% (Dembicki 1992) and over 0.2 mgHC g⁻¹ rock S2 values (Peters 1986) should not be susceptible to the mineral matrix effect. However, data in the current study and palynological data originally observed by Könitzer et al. (2016) demonstrated that not all of the TOC is necessarily generative owing to biotic decomposition, and the generative fraction can therefore be overwhelmed by the mineral surfaces.

Different facies types relate to diasterane/T_max trends. Turbidite facies representative of prodeltaic deposition have a strong positive correlation representative of the influence of clay minerals on pyrolysis results. Facies associated with terrestrial organic matter such as deltaic shelf sediments have the highest T_max results, which was also observed in Emmings et al. (2020a). Marine bands and hemipelagite facies show a lower influence by clay minerals on T_max maturity; the trends tend to reverse at boundaries between facies in Figures 3 and 4.

Hopane isomers, in particular the C₃₁ and C₃₂ homologues typically studied in the samples, have reached equilibrium (Fig. 11a, c) (Peters et al. 2005). The C₃₁ isomer ratios in Figure 12c positively correlate with the diasterane ratios used. Fully isomerized hopane homologues in immature sediments have been observed in other basins, i.e. by Moldowan et al. (1992) or ten Haven et al. (1986). Laboratory experimentation has in the past demonstrated high hopane isomer ratios that can be ‘reset’ during laboratory pyrolysis (Peters et al. 1990), suggesting that high values can be altered by expulsion. Hopane release is in part controlled by the breaking of sulfur bonds from kerogen (Köster et al. 1997), the nature of which may complicate the ratio. Corelation with diasteranes is indicative of a mineral catalytic effect or relates to the environment of deposition in a stratified water column that enabled sulfur incorporation into organic mediums. Emmings et al. (2020a) reported sulfidic conditions and shallow precipitation of pyrite framboids under highstand sea-level conditions; the fully isomerized hopanes in marine band and hemipelagite facies associated with diasteranes may be indicative of sulfur bond influences relating to the observations. In the current study, pyrite concentrations were found to be highest in the marine band or hemipelagite facies, further supporting this. In the case of turbidite facies, the observed ratio most probably relates to clay catalysis of organic matter. This is supported as turbidite facies plot with the highest clay mineral percentages and diasterane ratios (Fig. 12d). Hemipelagite type mudstones plotted with a lower clay to diasterane relationship, while marine bands were more distributed in Figure 12d, reinforcing that Eh to pH effects have a greater influence for those facies.

Studies employing equations to calculate % R_o equivalent from T_max (using the linear regression modelling) need to consider if the T_max result is likely to be increased by the mineral matrix effect giving inaccurate maturity trends. Studies such as Andrews (2013) specifically incorporated conversions to establish burial curves for the Upper and Lower Bowland Shale. To counter mineral matrix effects, Yang et al. (2016) recommended use of kerogen isolates, but the potential of a diagenetic mineral matrix effect on oil window organic matter means that the compositional kinetic pathways of yields from kerogen isolates may be inaccurate and not reflective of the true timing of generation v. expulsion.

Maturity results for the Bowland Shale and its diachronous counterparts have been largely misinterpreted in the literature, and care should be taken when using these to aid petrophysical calculations as in Hennissen and Gent (2019). Owing to the advanced ratios of certain biomarkers, the samples described in the current study are in the early to mid oil window with an equivalent vitrinite maturity of 0.7 to 0.8% R_o (Peters et al. 2005). Low T_max results in marine band facies were generated by mineral retention effects, and high values in low sea-level facies were influenced by the refractory oxidized organic macerals preserved.

The water column conditions as represented by EOM within the Mississippian basins of northern England are characterized by a continuous dysoxic layer between the sediment–water interface extending euxinia into the photic zone. Influx of terrigenous organic matter and settling of marine organic matter into the basin is proposed to have acted as a feedstock for a bacterial food web, leading to the preservation of compounds related to these groups, i.e. terpanes, hopanes and gammacerane. The high nutrient load into the basins encouraged the establishment of reducing conditions and switching to sulfate reduction. The kerogen type interpreted from the EOM is a mixed type, but hydrocarbon generative potential at marine band and highstand sea-level periods would be akin to Type II kerogens despite the mixing.

A mineral matrix effect influences both naturally and pyrolytically matured samples from the current study. Authors using programmed pyrolysis to interpret source rock hydrocarbon prospectivity or to calculate the original parameters should be aware that key values deriving S1 and S2 peaks will be reduced. Further work on hydrous, anhydrous and programmed pyrolysis exploring the role of pyrolysable v. residual carbon in hydrocarbon generation will be required to deduce the importance of this effect for this particular formation. The results of such studies have application in the wider exploration of deltaic-derived source rocks where detrital minerals and organic matter are deposited and matured together.

The work contained in this paper contains work conducted during a PhD study undertaken as part of the Natural Environment Research Council (NERC) Centre for Doctoral Training (CDT) in Oil & Gas. Samples were provided by the British Geological Survey (BGS), with special thanks given to the Core Store for provision of facilities, time and understanding. Finally, the two anonymous reviewers along with editors Bhavik H. Lodhia and Joseph F. Emmings are thanked for their comments and suggestions during the preparation of the work.

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.

MJES: conceptualization (lead), formal analysis (lead), methodology (lead), writing – original draft (lead); MAS: funding acquisition (lead), methodology (supporting), project administration (supporting), resources (lead), supervision (lead), writing – review & editing (lead); JSW: methodology (equal), writing – review & editing (supporting); AJF: funding acquisition (equal), project administration (equal), supervision (equal), writing – review & editing (equal); CHV: resources (supporting).

This work was supported by the Natural Environment Research Council (NERC) Centre for Doctoral Training (CDT) in Oil & Gas [grant number NEM00578X/1] and completed as part of the Natural Environment Research Council funded Unconventional Hydrocarbons in the UK Energy System project [grant number NE/R018065/1].

All data generated or analysed during this study are included in this published article (and its supplementary information files).