Petrographic and organic geochemical study of the Eocene Kosd Formation (northern Pannonian Basin): Implications for paleoenvironment and hydrocarbon source potential

The Eocene Kosd Formation forms part of the Hungarian Palaeogene Basin. The coal measure of this formation was investigated using an 18m drill core from borehole W – 1. Petrographic and organic geochemical investigations (Rock-Eval pyrolysis, biomarker analysis) were performed in order to characterize the depositional environment, to determine the source of the organic matter within, and to assess the hydrocarbon generative potential. The presence of marine fossils, high TOC/S ratios and ash yields show that the deposition of the coal measure occurred in a marine delta with individual coal layers accumulating in low-lying, rheotrophic mires. The dis- tribution of land plant-derived biomarkers demonstrates that the peat-forming vegetation was dominated by angiosperms, but the relative contribution of gymnosperms varied through time. In addition to land plants, algae and aquatic macrophytes contributed to the biomass. This dense vegetation established a CO 2 -limited environment forcing aquatic plants to utilise HCO 3 − during photosynthesis. The marine environment, as well as the predominance of carbonate rocks in the hinterland, caused slightly alkaline conditions, which, together with reduced oxygen availability, stimulated sulphate-reducing bacterial activity and the microbial degradation of plant remains. Consequently, Kosd Formation coal is very rich in sulphur (up to 8.8%). Moreover, the coal contains vitrinite with a strong orange-brown ﬂ uorescence colour and swells strongly during pyrolysis. These features are typical for coals with marine in ﬂ uences. Vitrinite re ﬂ ectance, Tmax, and biomarker proxies indicate that the organic matter is thermally mature and that the Kosd coal reached the high volatile bituminous rank in the deep borehole (~2.6km depth). Rock-Eval parameters imply that the coal is gas- and oil-prone and reached the maturity threshold critical for ﬁ rst gas generation and the onset of oil expulsion.


Introduction
During the Eocene, the Mesozoic Tethys Ocean decayed into a series of intercontinental seas (Rögl, 1999). This new configuration of land and sea areas modified oceanic circulation and climate (Popov et al., 2001). In the Late Eocene, Europe formed an archipelago and was enclosed by a subtropical sea, where variations in sea level significantly affected the distribution of depositional environments .
The Hungarian and Slovenian Palaeogene Basin is a predecessor of the Pannonian Basin (Tari et al., 1993). During the Eocene, sedimentation in this basin was characterized by a generally transgressive nature, where depocenters shifted northeastwards through time (Báldi and Báldi-Beke, 1985;Kováč et al., 2016). Hence, coal formation started earlier in the area of the present-day Transdanubian Mountains (the Middle Eocene Dorog Coal Formation) than in the North Hungarian Mountains (the Upper Eocene coal-bearing Kosd Formation; Figs. 1, 2; Báldi-Beke, 2003a, 2003bGidai, 1978;Hámor-Vidó and Hámor, 2007). The Kosd Formation includes economic coal seams (Gidai, 1978). The Kosd coalfield (location is shown in Fig. 1) is registered in the national coal cadastre of Hungary (MGSH -Mining and Geological Survey of Hungary, 2019), however, mining of this subhttps://doi.org/10.1016/j.coal.2020.103555 Received 2 April 2020; Received in revised form 1 July 2020; Accepted 5 July 2020 bituminous coal ended in the 1930s (Némedi Varga, 2010). Bechtel et al. (2007) performed an exhaustive geochemical study on the Middle Eocene Dorog Coal finding that the seam originates from a topogenous mire and evolved within a peneplane coastal area covered with eutrophic swamps. The peat-forming vegetation at the time was predominated by angiosperms, which is characteristic of Eocene coals throughout central Europe (Bechtel et al., 2008). In contrast, factors controlling the depositional environment and coal facies of the Upper Eocene Kosd coal have not previously been investigated.
Maceral composition, petrography-based facies indicators and biomarker analysis became essential tools during the last decades for the reconstruction of paleoenvironments and peat-forming floral changes (e.g. Bechtel et al., 2003Bechtel et al., , 2007Gross et al., 2015;Sachsenhofer et al., 2010). Additionally, the stable carbon isotope composition of individual biological markers allows the identification of specific sources (Hayes et al., 1987). Primary producers fix inorganic carbon through photosynthesis, which leads to specific fractionation of carbon isotopes (Diefendorf and Freimuth, 2017;Holtvoeth et al., 2019). Ficken et al. (1998) found that accumulating organic matter experiences rapid microbial degradation and associated diagenetic changes in the top few centimetres of sediment. Other studies, however, have reported no significant changes in δ 13 C during early diagenesis (Li et al., 2017, and references therein). Thermal maturation of organic matter may nonetheless affect the carbon isotope composition (Bjorøy et al., 1992;Clayton, 1991;Rooney et al., 1998). The degree of isotope fractionation depends on the temperature, but the effect is limited to a range of 2‰ as maturity progresses (Clayton, 1991;Schoell, 1984). Therefore, biologically controlled isotope compositions can be used to identify precursor sources after diagenesis (e.g. Collister et al., 1994;Freeman et al., 1990;Rieley et al., 1991).  The current study is based on samples from borehole W-1, located about 50 km SE from the abandoned Kosd coalfield (Fig. 1), which drilled the Eocene succession beneath potential Oligocene hydrocarbon source rocks. Whereas the Oligocene source rocks have been investigated by several authors (e.g. Badics and Vető, 2012;Bechtel et al., 2012;Milota et al., 1995;Sachsenhofer et al., 2018), the Eocene Kosd Formation remained largely uninvestigated.
The aims of this study are therefore to enhance understanding of the depositional environment and organic matter sources in this formation, and to estimate the hydrocarbon potential of deep Eocene Kosd coal. To achieve this goal, organic petrological and organic geochemical analyses were performed.

Geological setting
A series of sub-bituminous coalfields of Middle Eocene Dorog Coal Formation occur along the Transdanubian Mountains Bechtel et al., 2007). Eocene coal in the North Hungarian Mountains belongs to the Upper Eocene Kosd Formation ( Fig. 1; Gidai, 1978;Báldi-Beke, 2003a, 2003b. The diachronous development of depocenters followed the gradual northeastward trend of the Eocene transgression (Báldi and Báldi-Beke, 1985). In the western part of the Transdanubian Mountains, the initial transgression took place during the early Lutetian. The next transgression flooded the entire Transdanubian Range during the late Lutetian (the Dorog Coal Formation; Fig. 2). In contrast, for the eastern part of the Transdanubian Mountains and the North Hungarian Mountains, the transgression occurred during the Bartonian and Priabonian (the Kosd Formation; Fig. 2; Báldi-Beke, 2003a, 2003bKercsmár et al., 2015;Kováč et al., 2016).
The Kosd Formation crops out in small areas throughout the Buda Hills, the Left Side Blocks of Danube and the Bükk Mountains. In the subsurface, it extends towards the south into the Gödöllő Hills ( Fig. 1), where it was encountered by several boreholes (e.g. Bauer et al., 2016;Palotai and Csontos, 2010).
The Kosd Formation unconformably overlies the Mesozoic basement, which belongs to different mega-units. In the Buda Hills and the Left Side Blocks of Danube (Transdanubian Range Unit) the basement is represented by Ladinian to Norian platform carbonates (e.g. the Budaörs Dolomite Formation; the Dachstein Limestone Formation), and cherty basinal carbonates (e.g. the Mátyáshegy Formation; Haas and Kovács, 2012). In the Bükk Mountains (Bükk Unit), in addition to Ladinian to Norian platform carbonates (e.g. the Berva Limestone Formation; the Kisfennsík Limestone Formation) and basinal carbonates (the Felsőtárkány Limestone Formation), a deep marine Middle to Upper Jurassic sedimentary succession, with debris flows and turbidites, are present (the Mónosbél Group; Pelikán, 2005). The Mesozoic strata were uplifted during the Late Cretaceous and as a consequence of the subaerial exposure intense karstification occurred (Haas and Kovács, 2012).
Depending on the palaeotopography of the karst surface, the thickness of the Kosd Formation varies considerably and reaches 244 m in borehole Noszvaj-1 (Nv-1; the location is shown in Fig. 1; Less, 2005). The Kosd Formation was described in detail in two wells ; Varbó-75 [V-75]; Fig. 1) by Gidai (1978) and Less (2005). In both wells, the Kosd Formation is characterized by two different sedimentary facies. The lower facies contains fossil-and carbonate-free variegated clay with different amounts of rock debris, deposited in a terrestrial environment. The overlying sediments include claystone, siltstone with varying carbonate content, marl and thin coal layers (Gidai, 1978;Less, 2005). The thickness of the lower and upper sedimentary sequences is 63 and 12 m, respectively, at V-75 well (Less, 2005) and 9 and 123 m, respectively, at K-20 well (Gidai, 1978). The coal is intercalated into calcareous claystone and marl at V-75 and K-20 wells, respectively, whereas the direct underlying sediments are characterized by varicoloured clay with carbonate debris. The overlying sediments grade into Miliolina-bearing calcareous marl. Gidai (1978) and Less (2005) noted an upward transition from a freshwater environment, indicated by gastropods (Melanopsis sp.) to a shallow marine, lagoonal environment marked by the presence of nannoplankton (Isthmolithus recurves) and foraminifers (Quinqueloculina, Nummulites sp.). Moreover, a proximal coastal environment is supported by mangrove vegetation, represented by Nypa palm pollens (Kvaček, 2010, and references therein;Rákosi, 1978). Coaly layers were noted by Gidai (1978) and Less (2005) in boreholes K-20 and V-75, but were not investigated in detail.
The Kosd Formation grades upward into shallow marine platform carbonates of the Szépvölgy Limestone Formation (Fig. 2). The presence of the corallinacean algae and the frequently-occurring monospecific Nummulites fabianii indicate deposition on the inner shelf, while a diverse fauna including orthophragminas in the upper part of the formation, marks a transition to an outer shelf environment during the deposition of Szépvölgy Limestone (Less, 2005). Continuing basin subsidence caused deposition of the shallow bathyal Buda Marl Formation, in low-oxic environments across the Eocene-Oligocene transition (Ozsvárt et al., 2016). The Tard Clay Formation accumulated in oxygen-depleted conditions (Bechtel et al., 2012;Ozsvárt et al., 2016), and the Kiscell Clay Formation was deposited in a more oxygenated environment (Bechtel et al., 2012). A sequence of siliciclastic and carbonate platform sediments terminates the Palaeogene succession, which is followed by thick Neogene deposits (Less, 2005;Kercsmár et al., 2015).
The coal measures of the Kosd Formation in the former Kosd coalfield ( Fig. 1) are 5 to 32 m thick and include three seams (Kubacska, 1925;Gidai, 1978;Némedi Varga, 2010). The lower seam is 0.5 to 2.5 m thick and was exploited between 1904 and 1931 via an underground mine approximately 130 m below the surface. The coal reaches the sub-bituminous stage in the shallow mine. Based on contemporary data (Papp, 1913), the coal includes 3-4 wt% moisture and the ash yield varies from 6 to 20 wt%. The elemental composition of the coal is 56-67 wt% carbon, 4-6 wt% hydrogen, 1 wt% nitrogen and 5-6 wt% sulphur (Papp, 1913). According to Némedi Varga (2010), the average calorific value is 19 MJ/kg and the potential reserves were estimated as 15 Mt.

Samples
Borehole W-1, drilled in the early 2000s in the southern part of the Gödöllő Hills by MOL Plc. (Fig. 1), penetrated the Kosd Formation from 2415 m to 2791 m. A drill core, representing the upper section of the coal measures in the Kosd Formation, was recovered from 2587 m to 2605 m. In total, 35 samples (Table 1) were obtained from this drill core, each of them was selected based on lithological differences and is representative of a 20 cm interval. In order to avoid contamination, about 0.5 cm of the outer rim of each core was removed. Moreover, to prevent outlier readings, samples containing pyrite aggregates or nodules were not selected during sample collection. Pyrite crystals in the macroscopic range were hand-picked from the samples prior to pulverization.

Petrographic analysis
Fifteen core samples, including all coal and coaly shale samples as well as selected mudstone samples, were prepared as whole-rock polished blocks for maceral analyses. Each sample was cut perpendicular to the bedding plane, embedded in epoxy resin and polished according to standard procedures (ASTM International, 2015). Organic matter was identified by reflected white-and UV-fluorescent light microscopy using a Leica DM4P microscope equipped with a 50× oil-immersion objective and a point counter equipped with an OptiScan fully automated scanning stage. Maceral analyses were performed using a single scan method (Taylor et al., 1998) considering at least 1500 individual points to assess the minimum 500 counts of macerals. The terminology and classification of macerals used in this study are based on the ICCP system (ICCP, 1998(ICCP, , 2001Pickel et al., 2017).
Mean random reflectance (%Rr) of telovitrinite was measured according to the methods of Taylor et al. (1998), and was performed under monochromatic (546 nm) light using an Olympus BX41 microscope equipped with a 50× oil-immersion objective and optical standards of Buehler Ltd. Reflectance values were processed by image analysis (Taylor et al., 1998).

Organic geochemical analysis
Powdered rock samples were analysed in duplicate for total sulphur (TS) and total carbon (TC) contents using an ELTRA Helios CS-580A analyser. Samples pre-treated by hot and diluted H 3 PO 4 were used to determine total organic carbon (TOC) content. Total inorganic carbon (TIC = TC -TOC) was used to calculate calcite equivalent (CEq) percentages (CEq = TIC × 8.34 [%]). Ash yields were determined according to standard procedures (ASTM International, 2018).
Based on the results of Rock-Eval pyrolysis and the diverse lithologies observed, ten samples (#20 to #29) representing the depth interval between 2599.0 and 2602.4 m were chosen for solvent extraction. The upper-and lowermost samples represent typical low-TOC intercalating lithologies, whereas the others are coaly shale and coal samples together with their adjacent sediments.
Representative portions of powdered rock samples were extracted using a Dionex ASE 350 accelerated solvent extractor. A dichloromethane solvent was used at confined conditions of 75°C and 110 bar (full details of this procedure are given in Gross et al., 2015). The saturated compounds were further separated into normal alkanes and branched-cyclic alkanes for compound-specific isotopic analyses using an activated molecular sieve (Merck, 500 pm pore space), cyclohexane, and a cyclohexane-n-pentane (12:88) solution.
Saturated and aromatic hydrocarbon fractions were analysed using a gas chromatograph equipped with a 30 m DB-5MS fused silica capillary column (i.d. 0.25 mm; 0.25 μm film thickness) and coupled to a ThermoFisher ISQ mass spectrometer. The measuring process followed is described in Gross et al. (2015).
Stable carbon isotope measurements of n-alkanes and acyclic isoprenoids were performed on selected samples using a Trace GC instrument attached to a ThermoFisher DELTA-V isotope ratio mass spectrometer via a combustion interface (GC isolink, ThermoFisher). MDmeasured depth; dbdry basis; TICtotal inorganic carbon content, TOCtotal organic carbon content, CEqcalcite equivalent, TSsulphur content, TOC/ TSratio of total organic carbon versus total sulphur content, S1free hydrocarbons, S2hydrocarbons generated during Rock-Eval pyrolysis, HIhydrogen index, BIbitumen index, QIquality index, PIproduction index, Tmaxtemperature of maximum hydrocarbon generation.
CO 2 was injected at the beginning and the end of each analysis in order to perform instrumental calibration. The GC column and temperature program used were the same as above. Stable isotope ratios are reported in delta notation (δ 13 C; Coplen, 2011) relative to the Vienna Pee Dee Belemnite (V-PDB) standard (δ 13 C = [(δ 13 C/δ 12 C) sample /(δ 13 C/ δ 12 C) standard − 1]). Delta notation is expressed in parts per thousand or per mil (‰). The analytical error was better than 0.2‰.

Lithology
Based on drill cuttings and well logs (gamma ray, density, porosity) the Kosd Formation is 376 m thick and includes two coal horizons at depths between 2594 and 2649 m (Fig. 3a). The lower and upper coal horizons are about 9 m and 36 m thick, respectively. The sediments between the coal horizons consist of grey sandstone, sandy siltstone, silty claystone, and conglomerate. Non-coal layers within the coal horizons comprise grey, sandy-, argillaceous-, calcareous-and coaly siltstones. Individual coal beds are 0.1 to 2.1 m thick and include coal and coaly shale, and intercalations of siltstone and claystone (Fig. 3a).
The studied drill core (2587-2605 m depth) represents the upper portion of the coal horizon I (Fig. 3a). Its base is located in the upper part of a 2 m-thick coal bed at a depth of 2605 m. The core includes coal layers, which are black, consolidated and hard. Two different lithotypes can be distinguished: (i) a generally bright, finely banded clarain coal, which is brittle, exhibits uneven fracture and occurs as 5 to 50 cm thick layers; and (ii) a stratified coaly shale, including intercalations of thin (< 1 cm) siltstone and claystone layers, in which the thickness varies from 5 to 40 cm. The interseam sediments are dominated by sandy siltstone, which is variably calcareous and argillaceous. Original sedimentary structures are not visible due to bioturbation, however, slumps are recognizable. Two distinct intervals with shell fragments are observed (2596.5-2598.0 m; 2601.5-2602.0 m). Microscopic investigation revealed that the siltstone is rich in fossils, including mollusc, brachiopod, echinoid fragments and miliolid foraminifers (Figs. 3b, c).
Tmax values range from 431 to 448°C (Table 1). The HI ranges from 36 to 291 mg HC/g TOC (Table 1). In Tmax vs. HI plots (Figs. 4a, b), samples can be subdivided into two populations, reflecting those with less or more than 150 mg HC/g TOC. The HI of coaly lithotypes varies from 229 to 291 mg HC/g TOC (avg. 252 mg HC/g TOC). The HI in the interseam sediments is from 36 to 289 mg HC/g TOC (avg. 87 mg HC/g TOC), however, higher readings (> 150 mg HC/g TOC) were recorded at samples #6, #25 and #27, which are adjacent to coals (Table 1). BI and QI were calculated for coals and coaly shales, and vary from 30 to 37 mg HC/g TOC and from 259 to 323 mg HC/g TOC, respectively (Figs. 4c, d; Table 1). An important observation noted during Rock-Eval pyrolysis is that coaly samples undergo stronger swelling, i.e. their volume increases significantly during heating.
The amount of extractable organic matter (EOM) varies between 31 and 212 mg/g TOC and yields of EOM are usually higher in intercalating sediments (Table 3). EOM is dominated by polar compounds (34-50 wt%), except in samples #20 and #29, where the saturated compounds are prevalent (38 and 43 wt%, respectively). The saturated and aromatic hydrocarbon contents ranging from 5 to 43 wt% and 14 to 28 wt%, respectively. The amount of asphaltene is between 9 and 31 wt % (Table 3). Aside from the general dominance of polar compounds, coaly lithotypes are found to be characterized by similar average percentages of aromatic hydrocarbons and asphaltenes (avg. 24 and 25 wt %, respectively), and the quantity of saturated hydrocarbons is low (avg. 5 wt%). In contrast, interseam lithologies are characterized by similar amounts of saturated and aromatic hydrocarbons (avg. 24 and 22 wt%, respectively) and a lower amount of asphaltenes (avg. 13 wt%; Table 3).
Mean random vitrinite reflectance was found to range from 0.67 to 0.78 (Table 2).
The stable carbon isotopic composition (δ 13 C) of n-alkanes, Pr and Ph is similar for all studied samples (Fig. 9). With increasing chain length, δ 13 C values of the n-C 15-19 and n-C 21-25 alkanes of selected samples show a decline of about 1‰, respectively. In contrast, longchain alkanes (n-C 27+ ) show a reversed trend and increasing δ 13 C with carbon number. Pr and Ph exhibit similar isotopic compositions, but Ph is observed to be slightly depleted in 13 C in comparison to Pr (Fig. 9).

Hopanoids
Hopanoids are important constituents of the non-aromatic cyclic triterpenoids (Table 3). The measured hopanoid patterns are characterized by the occurrence of 17α,21β(H)-and 17β,21β(H)-type hopanes from C 27 to C 35 with C 28 hopanes being absent. The predominant hopanoids in most samples are 17α,21β-C 30 and 17β,21α-C 29 hopane. The 17α,21β(H)-type homohopanes show a dominant pattern of exponential decrease in peak height with increasing carbon number. In most samples, a series of C 32to C 35 -benzohopanes was identified in the aromatic hydrocarbon fractions. In all investigated samples, the ratios of steroids to hopanoids are between 0.06 and 0.15 (Table 4).

Sulphur-aromatic compounds
The sulphur-containing aromatic compounds identified include dibenzothiophenes and benzonaphthothiophenes, both of which occur in considerable quantities in the investigated samples (Table 3) The DBT/P ratios (Hughes et al., 1995) range between 0.74 and 1.04 ( Fig. 11; Table 4).

Source of organic matter
The Kosd coal is generally dominated by vitrinite and inertinite subgroup macerals (Table 2), which are derived from woody tissues of herbaceous and arborescent plants (ICCP, 1998). Petrography-based facies indicators have been determined for the coaly lithotypes. The vegetation index (VI; Table 2) contrasts macerals of forest affinity with those of herbaceous and marginal aquatic affinity (Calder et al., 1991). All studied samples of Kosd coal, except #21 and #34, are characterized by VI values below 3 ( Fig. 6; Table 2), indicating a predominance of herbaceous peat-forming flora.
Long-chain lipids are typically attributed to higher terrestrial plants (Eglinton and Hamilton, 1967), whereas mid-chain n-alkanes are reported in aquatic macrophytes and Sphagnum (Bingham et al., 2010;Dehmer, 1995;Ficken et al., 2000;Nott et al., 2000), and short-chain nalkanes are identified predominantly in algae and other microorganisms (Cranwell, 1977;Cranwell et al., 1987). Hence, it is surprising that the studied coaly samples (TOC > 20 wt%) are characterized by significantly higher amounts of short-chain n-alkanes than long-chain nalkanes (Table 4). For example, the clean coal #21 (TOC 78.4 wt%; ash yield: 5.2 wt%) contains the highest relative percentage of short-chain n-alkanes (max. 46%) and the lowest relative percentage of long-chain n-alkanes (min. 12%; Table 4). A shift towards shorter n-alkanes may be caused by advanced maturity (e.g. Radke et al., 1980), but this effect alone cannot explain the pattern observed since some low-TOC samples also contain high percentages of long-chain n-alkanes (e.g. #23; 31%). The high amounts of short-chain n-alkanes in coal samples are related to the strong fluorescence of vitrinite macerals, high S1 and BI values (Table 1; Fig. 4c) and may be caused by either migration (Littke et al., 1990) or bacterial activity. Whereas relative amounts of long-and short-chain n-alkanes vary significantly, mid-chain n-alkanes are found at high, relatively constant proportions (27-34%).
δ 13 C-values of n-alkanes in the selected samples ( Fig. 9) are consistent with previous studies performed on coal samples (e.g. Gross et al., 2015;Schoell et al., 1994;Simoneit et al., 1995;Schwarzbauer et al., 2013;Tuo et al., 2003). Variations of carbon isotopic compositions of individual compounds are within 1‰, indicating a positive correlation between sample sets (Sofer, 1984). δ 13 C-values also show a decreasing trend, becoming more depleted in 13 C with increasing chain length in all samples (Fig. 9), a phenomenon often observed in coals (e.g. Doković et al., 2018;Schoell et al., 1994;Schwarzbauer et al., 2013;Tuo et al., 2003). The long-chain n-alkane compounds, however, show an opposite tendency, getting more enriched in 13 C with increasing chain length (Fig. 9). Both long-chain n-alkane production and the biosynthetic fractionation of carbon isotopes vary among plant types (Diefendorf et al., 2011;Lockheart et al., 1997) and studies have pointed to 13 C enrichment with chain length in recent terrestrial plants (e.g. Collister et al., 1994;Diefendorf et al., 2011;Diefendorf et al., 2015;Lockheart et al., 1997;Mead et al., 2005). Nevertheless, the effect of thermal maturation on the carbon isotopic composition of organic matter has already been considered by several authors (e.g. Bjorøy et al., 1992;Rooney et al., 1998;Schoell, 1984). Diefendorf et al. (2015) found that the catagenic stage of maturation can lead to variations in the magnitude and direction of the change in δ 13 C-values according to species and chain length. Therefore, the effect of thermal maturation on the δ 13 C-values measured cannot be ruled out in the present study.
The δ 13 C-values correspond to common carbon isotope values of C3 plants and freshwater algae (e.g. Close, 2019;Diefendorf and Freimuth, 2017;Holtvoeth et al., 2019;Lamb et al., 2006;Meyers, 1997;O'Leary, 1981). Terrestrial plants utilise atmospheric CO 2 during photosynthesis, whereas aquatic photoautotrophs use dissolved CO 2 [AQ] or, in the case of alkaline or CO 2 -limited conditions, can fix HCO 3 − as the inorganic carbon source for photosynthesis (Holtvoeth et al., 2019;Lamb et al., 2006;Meyers, 1997). The assimilation of HCO 3 − leads to 13 C enrichment (Aichner et al., 2010;Lamb et al., 2006;Meyers, 1997). The intermediate-weight n-alkanes are isotopically heavier than long-chain nalkanes ( Fig. 9), moreover, δ 13 C isotopic differences could be more explicit prior to the thermal alteration. Furthermore, low-molecular weight n-alkanes are characterized by 13 C enrichment relative to midchain n-alkanes (Fig. 9). However, the carbon isotopic composition of lake-derived organic matter is typically indistinguishable from the surrounding watershed (Meyers, 1997). Nevertheless, the contribution of algae to the carbon isotope compositions measured is also suggested by the presence of lamalginite.
The predominance of C 29 steroids in coals is consistent with a dominant origin of organic matter from vascular plants ( Fig. 10; Table 4; Huang and Meinshein, 1979), which also explains the low amount of regular steranes (Doković et al., 2018;Volkman, 1986). Besides the vascular plants, the constant proportion of C 28 steranes and the occurrence of perylene has been suggested to indicate a contribution from wood-degrading fungi (Marynowski et al., 2013). Hopane derivatives occur in the investigated samples, in which the bacteriohopanepolyols have been suggested to be the most probable biological precursors, and have been identified both in bacteria and in some cryptogams (e.g. moss, ferns; Ourisson et al., 1979;Rohmer, 1993;Talbot et al., 2016). The occurrence of C 32 to C 35 benzohopanes (Hussler et al., 1984) in the aromatic hydrocarbon fractions also suggests that bacteriohopanetetrol was a significant constituent of the precursor biomass. Furthermore, both the low sterane to hopane ratio and the presence of drimane-type sesquiterpenoids suggest microbially reworked organic matter (Alexander et al., 1983(Alexander et al., , 1984Tissot and Welte, 1984).
Sesquiterpenoids and diterpenoids are present in the analysed sample sets (Table 3). The biological precursors of cadalane-type sesquiterpenoids are widely distributed throughout all conifer families (Simoneit et al., 1986;Otto et al., 1997, and references therein). Eudesmane is also a non-specific indicator of higher land plants (Alexander et al., 1983), hitherto missing in species of Araucariaceae and Taxaceae (Otto and Wilde, 2001). Labdane-type compounds are the most common diterpenoids in conifers and have been described in all families except in Cephalotaxaceae (Otto and Wilde, 2001). More than 100 pimarane-and isopimarane-type diterpenoids have been detected in species of Pinaceae, Taxodiaceae, Araucariaceae, and Cupressaceae (Sukh Dev, 1989). Phyllocladane-type diterpenoids are widespread among coniferales families, except in Pinaceae (Otto and Wilde, 2001). Abietane-type diterpenoids can be generated by the transformation of pimarane (Wakeham et al., 1980), the progressive thermal alteration of phyllocladane (Alexander et al., 1987), or the dehydrogenation of abietic acids (Peters et al., 2005). Alkylated phenanthrenes have been genetically related to abietic-and pimaric acid (Alexander et al., 1995;Laflamme and Hites, 1978;Radke et al., 1998). Based on the composition of diterpenoids present in the studied samples, Cupressaceae genera presumably contributed to the arboreal vegetation.
The di-/(di-+ triterpenoids) ratio can be used to estimate the relative contribution of gymnosperms and angiosperms to organic matter (Bechtel et al., 2003). Very low ratios (≤0.2; Table 4) indicate that angiosperm-derived biomarkers predominate in all analysed samples of the Kosd Formation. These results are consistent with previous observations on Eocene coal deposits throughout central Europe (Bechtel et al., 2007(Bechtel et al., , 2008 and are supported by palynological data from throughout the Transdanubian Mountains, which further indicates an angiosperm-dominated vegetation (Kvaček, 2010, and references therein;Rákosi, 1978). Although the vegetation was dominated by angiosperms, a continuous upward increase in the di-/(di-+ triterpenoids) ratio from 0.06 to 0.20 (Table 4) indicates an increasing relative contribution of gymnosperms during deposition of the studied depth interval (2602.4-2599.0 m).

Depositional environment
GWI AC values (> 10) in the studied section suggest that the deposition of the coaly lithotypes occurred in a low-lying mire with rheotrophic conditions (Fig. 6; Table 2; Calder et al., 1991). The presence of inertinite and liptinite in the investigated Kosd coal implies the existence of a paleomire influenced by the fluctuation of the water table (Diessel, 1992;Eble et al., 2019), allowing the accumulation of lamalginite during high-, and the oxidation of plant tissues during low water table conditions. The n-alkane isotope profile shows a generally declining pattern (Fig. 9), which has previously been assigned to fluvio-deltaic origins (e.g. Bjorøy et al., 1991;Cortes et al., 2010;Dzou and Hughes, 1993;Murray et al., 1994;Odden et al., 2002;Tuo et al., 2003;Wilhelms et al., 1994). The Pr/Ph vs. DBT/P plot also indicates a fluvio-deltaic environment ( Fig. 11; Hughes et al., 1995). The composition of regular steranes ( Fig. 10; Table 4; Huang and Meinshein, 1979) suggests a transition between shallow marine and deltaic paleoenvironments, which provided an appropriate setting for the preservation of higher plant triterpenoids (Strachan et al., 1988).
The sulphur content of the analysed samples has a general tendency towards higher readings with increasing organic carbon content (Table 3). Variations in sulphur content can be explained by an influx of saline water towards the paleomire (Petersen and Ratanasthien, 2011) or by changes in pH (Markič and Sachsenhofer, 1997). Considering the occurrence of marine miliolid foraminifers (Fig. 3c), an influx of marine water is obvious. In addition, the low TOC/TS ratios (Table 1) support marine influence (Berner, 1982(Berner, , 1984. Denudation of Mesozoic carbonate rocks underlying the Kosd Formation may have enhanced the concentration of bicarbonate ions. Moreover, the chemical weathering in the hinterland provided nutrients to the paleovegetation. Alkaline conditions control both, bacterial decomposition of plant remnants and reduction of sulphates by sulphate-reducing bacteria (Markič and Sachsenhofer, 1997). In the case of available reactive iron, the transformation of sulphide phases into densely packed framboidal aggregates of pyrite crystals is a prevalent process driven by bacterial sulphate reduction (Casagrande, 1987;Hámor, 1994;Maclean et al., 2008;Morad, 1998;Sweeney and Kaplan, 1973). Nevertheless, there is a poor fit between the TS and pyrite contents, which indicates the presence of organic sulphur. Organic sulphur-bearing compounds (e.g. dibenzothiophenes) have been detected in considerable concentrations ( Table 3). The prevalence of organic sulphur has also been described for Eocene coals by Hámor-Vidó and Hámor (2007). The marine influence and the presence of mobile Ca 2+ and HCO 3 − ions in the continental runoff are confirmed by the local elevation of calcite equivalent percentages (at #2, #8, #9, #11 and #33) in the interseam sediments (Table 1).
In the studied samples, Pr and Ph exhibit similar carbon isotopic compositions, and are enriched in 13 C relative to terrestrial plant-derived n-alkanes, arguing for chlorophyll as their common source ( Fig. 9; Collister et al., 1992;Hayes et al., 1990;Hayes, 1993). Confined pyrolysis studies of coal samples support a maturation tendency of Pr/Ph (Monthioux and Landais, 1989;Radke et al., 1980). Therefore, the interpretation of redox conditions during sedimentation based strictly on Pr/Ph ratios (Didyk et al., 1978) is limited in the current study due to interferences resulting from thermal maturity. Nevertheless, the occurrence of des-A-triterpenes suggests dysoxic conditions (Jacob et al., 2007, and references therein). Furthermore, in immature terrigenous organic matter, perylene has been related to the activity of wood-degrading fungi and indicates reducing conditions in subsurface sediments during early diagenesis (Marynowski et al., 2013). Although the studied coals are mature, the presence of framboidal pyrite, as well as perylene, indicates eogenetic origin and locally reducing conditions (Sweeney and Kaplan, 1973).
Whereas Gidai (1978) and Less (2005) have mentioned the presence of coaly layers in wells K-20 and V-75, they did not investigate the organic material. Nevertheless, the new results on well W-1 agree well with the sedimentological and palaeontological data reported by these authors. Despite the strongly varying thickness, all three wells show a general trend from non-marine fine-grained rocks with debris flow deposits to brackish and marine sediments. Coaly layers occur within the transition zone. According to Less (2005), the upper marine part of the Kosd Formation in V-75 represents a lagoonal environment dominated by fine-grained rocks. Based on the lithological and faunal description provided by Gidai (1978), a similar environment can be assumed for well K-20. In contrast, the studied interval in well W-1 includes sandstones and conglomerates representing a delta environment. Moreover, the total thickness of the Kosd Formation in well W-1 (376 m) is higher than in any other well (cf. Less, 2005).

Hydrocarbon generation potential
The remaining hydrocarbon generation potential of the Kosd Formation samples can be characterized using TOC contents and the amount of free (S1) and generated hydrocarbons (S2). TOC contents of the analysed samples are generally greater than 0.5 wt% ( Table 1). The obtained HI values suggest that type III and II-III kerogen is dominant, nevertheless, type IV kerogen also occurs (#12, #16 and #20; Fig. 4a; Table 1). The coaly lithotypes and their adjacent intercalating sediments have excellent and fair-good petroleum potential, respectively (Peters, 1986;Peters and Cassa, 1994). In total, 69% of the analysed samples have HI values between 50 and 200 mg HC/g TOC and are gasprone, whereas 22% possess oil-and gas-prone kerogen (Peters and Cassa, 1994).
Following the approach of Sykes and Snowdon (2002), coaly samples were plotted onto diagrams of HI, BI and QI versus Tmax (Figs. 4b-d). These diagrams suggest that all coaly samples are gas-and oil-prone (Fig. 4b) and have passed the rank threshold for oil generation (BI; Fig. 4c) and expulsion (QI; Fig. 4d), furthermore, have reached the maturity threshold for the first gas generation (BI; Fig. 4c). Fig. 4c emphasizes the high BI of the coal samples, which is linked to abundant short-chain n-alkanes and probably causes the strong fluorescence observed in vitrinite macerals.

Conclusion
The investigation of the coal measure of the Eocene Kosd Formation in northern Pannonian Basin has yielded important new results regarding its depositional environment, organic matter source and hydrocarbon potential: (1) The coal measure evolved in a marine deltaic environment. The accumulation of peat-forming vegetation in a low-lying, rheotrophic mire was affected by fluctuations of the water table.
(2) Slightly alkaline conditions and the depletion in dissolved oxygen noted in the sediments promoted the reduction of sulphates by sulphate-reducing bacteria and bacterial decomposition of plant remains. In addition to the high sulphur contents observed (max. 8.8%), the orange-brown fluorescence colour of vitrinite and its strong swelling during pyrolysis are typical indicators of marineinfluenced coals.
(3) The peat-forming flora was dominated by land plants with varying contributions of algae and aquatic macrophytes. Similar to other Eocene coal seams, angiosperms predominated over gymnosperms.
An upward increase in the relative contribution of gymnosperms (e.g. Cupressaceae) to the biomass is observed at depth between 2599.0 and 2604.4 m. The organic matter in the paleomire was highly reworked by microbial processing. Dense flourishing vegetation established a CO 2 -limited environment forcing aquatic plants to utilise HCO 3 − during photosynthesis.
(4) Vitrinite reflectance, Tmax, and biomarker indices denote that organic matter in the Kosd Formation (as identified in deep borehole W-1) is thermally mature and suggests that the Kosd coal reached high volatile bituminous rank in the research area. This is higher than the rank of the sub-bituminous coal in the shallow Kosd coalfield of the North Hungarian Mountains. This advanced maturity influenced the molecular and isotopic composition of hydrocarbons. (5) Rock-Eval pyrolysis results indicate that the coaly samples studied are gas-and oil-prone and have reached the maturity threshold for first gas generation and the onset of oil expulsion.

Declaration of Competing Interest
The authors declare there is no conflict of interest.