The organic chemistry of coal is a complex product of the various progenitors for the individual macerals. During peatification, diagenesis, and subsequent coalification processes, the original vegetation is decomposed, altered, and ultimately transformed into coal. South African coals occurring in the Main Karoo Basin (MKB) are generally rich in inertinite with variable proportions of vitrinite, a relationship which may be considered to be the inverse of that observed in most Northern Hemisphere coals (Falcon 1986; O’Keefe et al. 2013). The formation of the inertinite-rich, Karoo coals need to be understood in terms of the original material, and the origin pathways along which the inertinite and vitrinite macerals evolved.

Most vitrinite macerals are derived from the humification and gelification of woody tissues composed of lignin, cellulose, and a hemicellulose fraction (Teichmüller 1989; van Krevelen 1993; Hatcher and Clifford 1997; ICCP 1998; O’Keefe et al. 2013). In addition to the wood itself, other vitrinite macerals, such as corpogelinite and gelinite, are derived primarily from cell wall secretions and the contents thereof, or alternatively, from humic gels infilling cells, pores, and fissures (ICCP 1998; O’Keefe et al. 2013). The origin of inertinite macerals is variable for different macerals belonging to this group (van Krevelen 1993; ICCP 2001), derived through multiple origin pathways (Hower et al. 2011a, 2013; O’Keefe et al. 2013; Moroeng et al. 2018a, b, c). However, there is some general consensus that some macerals of this group, such as semifusinite and fusinite, formed through the charring of woody plant matter (Austen et al. 1966; Scott 1989, 2002, 2010; Diessel 1992, 2010; ICCP 2001; Glasspool 2003a, b; Hower et al. 2013; O’Keefe et al. 2013; Moroeng et al. 2018a, b, c). Other authors contend that semifusinite may also form through the aerial oxidation of parechymatous and xylem tissues of plants (Falcon 1986; Falcon and Snyman 1986; Hagelskamp and Snyman 1988; Snyman 1989; Taylor et al. 1998; ICCP 2001). Other inertinite macerals form through the degradation of plant matter by organisms (macrinite), and from the remains of the organisms (funginite) (Moore et al. 1996; Moore and Shearer 1997; ICCP 2001; Hower et al. 2009, 2011b), or from the partially digested excretion and/or regurgitation products (so-called copromacrinite) of those organisms (Hower et al. 2011a, 2013; O’Keefe et al. 2013; Valentim et al. 2016). Inertodetrinite represents reworked plant matter that would have produced primary inertinite macerals, and secretinite reflects extreme oxidation or charring of resin and humic gels (ICCP 2001; O’Keefe et al. 2013).

The organic chemical composition of coal, with coal being a product of most kingdoms of life (Hower et al. 2009, 2011b), is thus informed by: (1) the various botanical and zoological precursor materials; (2) processes that altered the matter during peatification, diagenesis, and subsequent coalification processes; (3) the level of coal maturity; as well as, (4) the depositional environment in its entirety, what Hower et al. (2011b, p. 187) term the “total environment”.

Various analytical techniques have been used in an attempt to elucidate the chemistry of the maceral components of coal. Although numerous studies discuss the pyrolysis-gas chromatography mass spectrometry results for various macerals, and especially vitrinite-rich coals of the Northern Hemisphere (e.g., Meuzelaar et al. 1984; Senftle et al. 1986; Senftle and Larter 1987, Nip et al. 1988, 1992; Eglinton et al. 1990, 1991; Boreham and Powell 1991; Powell et al. 1991; van Krevelen 1993, and references therein; Hartgers et al. 1994; Veld et al. 1994; Han et al. 1995; Odden and Barth 2000; Davidson 2004, and references therein; Sun and Horsfield 2005; Liu and Peng 2008; al Sandouk-Lincke et al. 2014; Zieger et al. 2018), studies using inertinite-dominated South African coals are comparatively sparse.

The main purpose of the present study is to examine differences and/or similarities in the organic molecular chemistry of iso-rank bituminous coals, a vitrinite-rich and an inertinite-rich sample, using a pyrolysis gas chromatograph (py-GC) equipped with a thermal desorption unit (TDU) coupled to a time of flight (TOF) mass spectrometer (MS) (py-GC/MS) in order to understand the effect of source vegetation and origin pathways on coal pyrolysis products. As liptinite normally accounts for a minor proportion of the coals of the MKB (Falcon and Snyman 1986; Kruszewska 2003; Van Niekerk et al. 2008), it is omitted from the present discussion. A medium rank C bituminous Witbank coal from the No. 4 Seam Upper was density-fractionated to create the required samples, enhancing compositional differences between the samples without introducing rank-dependent variations. Along with the widely used carbon-13 cross-polarization magic-angle-spinning solid-state nuclear magnetic resonance analysis (¹³C CP-MAS SS NMR), py-GC/MS is used to investigate the organic molecular chemistry of the samples. The ¹³C CP-MAS SS NMR results of these samples (including spectra) were reported by Moroeng et al. (2018a). However, aside from the aromatic cluster sizes, other ¹³C CP-MAS SS NMR chemical-structural parameters were not explored by Moroeng et al. (2018a).

2.1 Samples and preparation

A run-of-mine coal sample (weighing approximately 50 kg) was obtained from the No. 4 Seam Upper, Witbank Coalfield. Following crushing, a quarter of the original sample was reserved to represent the “parent” coal sample. The parent sample was density-fractionated at an RD of 1.3 and 1.8 to create the required vitrinite-rich and inertinite-rich samples, respectively, following South African National Standard (SANS) 7936 (2010).

2.2 Proximate, elemental, and petrographic analysis

The proximate (moisture, ash, volatile matter, and fixed carbon) and elemental analyses (C, H, N, O, and S) were undertaken at a commercial laboratory (Bureau Veritas, Centurion, South Africa) following prescribed SANS standards (SANS 334 1992; SANS 17247 2006; SANS 17246 2011).

The samples were prepared for petrographic analysis as polished blocks following SANS 7404-2 (2015). The maceral analysis was undertaken following SANS 7404-3 (2016) by means of a 500 point-count, using a Zeiss AxioImager m2M reflected light microscope retrofitted with Hilgers Diskus Fossil components and software, at a total magnification of 500× using immersion oil. The mean random vitrinite reflectance (%RoVmr) of the coal was determined on the parent sample by a 100 point-count on collotelinite, following SANS 7404-5 (2016).

2.3 Nuclear magnetic resonance analysis

The ¹³C CP-MAS SS NMR methodology is described extensively by Moroeng et al. (2018a), and will only be briefly described in the present study.

¹³C CP-MAS SS NMR coupled with dipolar dephasing (CP-MAS/DD) was used to compare the coal samples, following the method described by Solum et al. (1989). These two experiments were performed on each sample to determine the structural differences between the parent coal, vitrinite-rich, and inertinite-rich samples. The structural parameters were calculated based on Solum et al. (1989) and Suggate and Dickinson (2004). The spectra were acquired using an Agilent VNMRS 500 MHz two-channel NMR spectrometer using 4 mm zirconia rotors and a 4 mm Chemagnetics TM T3 HXY MAS probe. All CP spectra were recorded with the latest VnmrJ 4.2 instrument software at ambient temperature with optimized high-power proton decoupling; a relaxation delay of 3 s and 10 000 scans were collected for obtaining sufficient signal-to-noise ratio. The power parameters were optimized for the Hartmann-Hahn match; the radio frequency fields were γCB1C = γHB1H ≈ 56 kHz. The contact time selected for CP was 2 ms. The free induction decay recorded 2000 points and was Fourier transformed with a 125 Hz line broadening. Magic-angle-spinning (MAS) rate of 12 kHz and Admantane was used as an external chemical shift standard with the upfield methyl peak-referenced to 38.4 ppm.

The DD experiments were carried out under the same CP parameters with the interrupted decoupling constant of the tancpxidref sequence set to 40 µs, after first evaluating an incremental array of 10–100 µs time constants.

Regarding spectra processing, the integration reset points were adapted from Solum et al. (1989) and Mestrenova 11.02 “preset integral regions table” was made to ensure that the exact same regions were integrated reproducibly. The individual spectra were each phased and baseline corrected manually before integration. All spectra were processed in the same manner and by the same person to minimize operator bias. The absolute integral values were transferred to a Microsoft Excel spreadsheet for calculation of the coal structural parameters as per Solum et al. (1989). No variable contact time was performed, as suggested by Solum et al. (1989), because of the long acquisition times. A contact time of 2 ms was used for all experiments and a correction factor of 1.3, derived from Solum et al. (1989), was used to correct the intensity of the bridgehead aromatic carbons (peak 135–90 ppm for DD) for the loss of magnetization during the interrupted decoupling.

2.4 Pyrolysis-gas chromatography mass spectrometry analysis

The py-TD-GC-TOFMS system used in this study consisted of an Agilent 6890B (Agilent Technologies, USA) gas chromatograph and a Pegasus III TOF-MS (LECO, USA) equipped with a Gerstel autosampler, thermo desorption unit (TDU, Gerstel), pyrolysis module (Pyro, Gerstel) and cooled injection system (CIS, Gerstel) housed at the University of South Africa (Department of Chemistry, Florida, Johannesburg).

The samples were milled to a particle size of − 212 µm for the py-GC/MS analysis. Samples (± 3 mg) were placed in small quartz tubes and inserted into the pyrolysis unit. The samples were heated to 700 °C at a rate of 5 °C/s and kept at that temperature for 1 min. Thermal desorption was then carried out from 40 to 350 °C (3 min hold time) at 240 °C min⁻¹. The desorbed volatiles were cryogenically focused using a Gerstel cooled injection system (CIS) at − 25 °C using liquid nitrogen. The analytes moved onto the GC column through the transfer line heated at 350 °C and then separated using a Restek Rxi-17Sil MS capillary column (30 m × 0.25 mm × 0.25 µm) with helium as the carrier gas at a constant flow rate of 1.4 mL min⁻¹ throughout the analysis. The initial oven temperature was 50 °C and held for 1 min, then increased to 300 °C at a rate of 9 °C/min and held constant for 2 min. The injection port and transfer line temperatures were maintained at 300 °C and 310 °C, respectively. The coal sample remained in the pyrolysis module during the entire GC run.

Mass acquisition was from 50 to 650 mass units at 10 spectra s⁻¹ data acquisition rate. The electron energy was 70 eV whilst mass spectrometer transfer line and ion source temperatures were 310 °C and 225 °C respectively. Data acquisition and processing was carried out using ChromaTOF software version 4.0 (LECO, USA). Peak identification was made using library matching using the National Institute of Standards & Technology (NIST) mass spectral library. Semi-quantitative data of the compounds was obtained by using the peak area for the compound as a percentage of the total area for all compounds detected. For each sample, analysis was done in triplicate and average results presented.

3.1 Proximate, elemental, and petrographic composition

The proximate, elemental, and petrographic data are presented in Table 1. Based on the proximate analysis, the parent coal, vitrinite-rich, and inertinite-rich samples have similar moisture contents. The vitrinite-rich sample has the highest volatile matter content as well as the lowest ash content. The higher volatile content for the vitrinite-rich sample relative to the inertinite-rich counterpart is in agreement with published literature for bituminous coals (van Krevelen 1993; Van Niekerk et al. 2008, and references therein). The H/C atomic ratio, which is indicative of aromatic fraction, is also in agreement with published literature regarding the greater aromatic fraction of bituminous inertinite-rich coals relative to vitrinite-rich counterparts at similar rank (Davidson 2004; Van Niekerk et al. 2008). The parent sample has the highest total sulphur content, which is significantly reduced in the density-fractionated samples, including the inertinite-rich sample despite the higher ash content. This suggests that the major form of sulphur in the parent coal is present as sulphide minerals (or other sulphur-bearing mineral phases), which reported to the 1.8 sink fraction, with organic sulphur comprising a minor proportion. The total nitrogen content of the samples is comparable, though the vitrinite-sample has the slightly higher proportion of this chemical element.

In terms of maceral group composition, the parent coal sample consists of 44.3 vol% vitrinite and 51.7 vol% inertinite (mineral-matter-free basis). This maceral composition is relatively enriched in vitrinite in comparison to most Witbank coals (e.g., Falcon and Snyman 1986; Roberts et al. 2015), although comparable to a sample used by Okolo et al. (2015). The 0.65%RoVmr (SD = 0.067) indicates that the coal is medium rank C bituminous. As a result, the maceral nomenclature used follows that of the International Committee for Coal and Organic Petrology for bituminous rank coals (ICCP 1998, 2001; Pickel et al. 2017).

The semifusinite and inertodetrinite macerals were further subdivided into reactive and inert sub-macerals, as is the convention for South African coals (Falcon and Snyman 1986; Hagelskamp and Snyman 1988; Snyman 1989; Okolo et al. 2015).

The inertinite-rich sample comprises of 65.7 vol% total inertinite macerals, where fusinite (8.7 vol%), semifusinite (reactive = 7.5 vol% and inert = 22.9 vol%) and inertodetrinite (reactive = 2.2 vol% and inert = 22.8 vol%) are the principal inertinite macerals. In contrast, the vitrinite-rich counterpart is dominated by the vitrinite macerals (81.3 vol% total vitrinite), collotelinite and collodetrinite, with lower proportions of telinite and corpogelinite. Because the dominant macerals of the density-fractionated samples were derived from similar plant organs, i.e. wood (ICCP 1998, 2001; Hower et al. 2013; O’Keefe et al. 2013), the pyrolysis products of the samples may be expected to either be similar or different. This would depend on which is the overriding parameter between source vegetation and origin pathways determining the pyrolysis products of the coal samples.

3.2 Comparison of organic geochemistry of the vitrinite-rich and inertinite-rich samples

Along with py-GC/MS, ¹³C CP-MAS SS NMR was used to compare the molecular chemistry of the coal samples. Moroeng et al. (2018a) concluded that the ¹³C CP-MAS SS NMR results are generally comparable with results obtained by others using advanced, direct polarization SS ¹³C NMR coupled with spectral-editing techniques, for example the results reported by Cao et al. (2011, 2013). The results of the ¹³C CP-MAS SS NMR analysis are given in Table 2. The greater aromatic fraction of inertinite-rich coals relative to vitrinite-rich counterparts at the bituminous rank is well-established (van Krevelen 1993; Davidson 2004; Van Niekerk et al. 2008), and the results of the present study largely concur with this (Table 2). Further, the results indicate that the inertinite-rich sample has the marginally greater fraction of phenolics (f^P_a = 0.13 vs. 0.09 for the vitrinite-rich sample). Although this sample has the greater overall aromatic fraction, the degree of alkyl substitution is greater for the aromatics of the vitrinite-rich sample. Despite lower overall aliphatic fraction (f_al), the inertinite-rich sample has a significantly higher proportion of aliphatic carbons bonded to oxygen (f^o_al =0.05 vs. 0.002 for the vitrinite-rich counterpart), along with a higher fraction of carbonyl/carboxyls (f^CO_a). The proportion of non-protonated carbons and methyl groups in aliphatic compounds is larger for the vitrinite-rich sample (f^N*_a = 0.1) relative to the other samples. The number of attachments per cluster (σ + 1) is larger for the inertinite-rich sample in comparison to the vitrinite-rich counterpart. Therefore, the ¹³C CP-MAS SS NMR chemical-structural parameters for the density-fractionated samples are markedly different, reflecting the greater aromatic fraction and oxygen-bearing functionalities for the inertinite-rich sample as well as the greater degree of alkyl substitution of aromatics for the vitrinite-rich sample. Moreover, aside from the average aromatic cluster size (C) for the inertinite-rich sample, the ¹³C CP-MAS SS NMR results are generally comparable with results obtained by others when evaluating the chemical-structural parameters for vitrinite-rich and inertinite-rich South African bituminous coals (e.g., Van Niekerk et al. 2008; Okolo et al. 2015; Roberts et al. 2015).

The pyrolysis chromatograms for the parent, inertinite-rich, and vitrinite-rich coal samples are presented, respectively, in Figs. 1, 2 and 3. Of the compounds detected during the py-GC/MS analysis, chemical compounds with an abundance (area %) greater than 1% are given in Table 3. These compounds represent 75%, 78% and 71% of the pyrolysis products detected for the parent, inertinite-rich, and vitrinite- rich samples, respectively. The chemical compounds presented were identified with a similarity of more than 70%, and were arranged in order of increasing retention time in Table 3 to aid with peak identification on the chromatographs. For relatively simple compounds common in the pyrolysis products for the three samples such as phenol, benzene, xylene isomers, and fluorene, the py-GC/MS retention times are greatly comparable (Table 3). Differences appear to become more prominent for larger and/or more substituted compounds, thus with higher retention times.

Total ion chromatogram for the parent coal sample

Total ion chromatogram for the inertinite-rich sample

Total ion chromatogram for the vitrinite-rich sample

The pyrolysis products of the coal samples of the present study generally comprise of similar compounds reported for various coals from around the world (Table 3) (Meuzelaar et al. 1984; Senftle and Larter 1987; Eglinton et al. 1991; Powell et al. 1991; Nip et al. 1988, 1992; van Krevelen 1993, and references therein; Hartgers et al. 1994; Veld et al. 1994; Zieger et al. 2018), and South Africa (Van Niekerk et al. 2008). The aliphatic fraction of the pyrolysis products is comprised of a range of low molecular weight (carbon number, C₄) to relatively high molecular weight (C₃₀) cyclic and branched alkanes and alkenes. With the aid of flash-pyrolysis GCMS, Van Niekerk et al. (2008) reported long-chain alkane and alkene compounds ranging up to C₂₃ for vitrinite-rich Waterberg and inertinite-rich Highveld bituminous coals. Further, the pyrolysis products of the vitrinite-rich Waterberg sample were reported to comprise of a greater proportion of aliphatic compounds relative to the inertinite-rich sample (Van Niekerk et al. 2008), consistent with the ¹³C CP-MAS SS NMR results of this study (Table 2).

The aromatic fraction of the pyrolysis products for the samples of the present study comprise of a significant amount of substituted and unsubstituted 6-aromatic carbon ring functional groups. For the inertinite-rich sample, the abundance of these monocyclic rings is confirmed by the ¹³C CP-MAS SS NMR results (Table 2), specifically the average number of aromatic carbons per cluster (C). The most abundant polycyclic aromatic hydrocarbons (PAHs) detected in the pyrolysis products are naphthalene and flourene, similar to the results obtained by Zieger et al. (2018) for vitrinites of Carboniferous Age from the Ruhr Basin, Northwest Germany. Other larger PAHs such as pyrene, anthracene, and phenanthrene are present in minor amounts, hence they were excluded from Table 3. The PAHs mostly occur as alkyl-substituted derivatives, consistent with other studies (e.g., Van Niekerk et al. 2008; Zieger et al. 2018).

Ethane-tribromo- and trichloromethane (halogenated hydrocarbons) were also detected during the analysis. Of these, the latter is common to all the samples, whereas the former only present in the pyrolysis products of the inertinite-rich sample. Despite the fact that the parent coal and the inertinite-rich sample have comparable ash contents, the inertinite-rich sample has a lower total sulphur content. This may be interpreted to suggest that the organic sulphur content of the coals is lower, assuming that the sulphides reported in the 1.8 sink fraction, as previously suggested. This is also supported by the minor sulphur-bearing organic compounds (e.g., 3, 4-Epoxytetrahydrothiophene-1, 1-dioxide for the parent coal) detected during the py-GC/MS analysis. Likewise, nitrogen-bearing compounds (e.g., 1H-Pyrrole, 2-methyl-) were detected with a low abundance. However, it should be mentioned that Phiri et al. (2018) observed the dependence of pyrrolic nitrogen on pyrolysis temperature.

Through visual inspection, it becomes evident that the pyrolysis chromatogram for the parent coal (Fig. 1) is more comparable to that of the inertinite-rich sample (Fig. 2), likely reflecting similarities in petrographic composition (Table 1). The chromatograph for vitrinite-rich sample (Fig. 3) is markedly different, exhibiting greater chemical diversity in comparison to the other samples, possibly resulting from the greater aliphatic fraction and alkyl substitution as reflected in the ¹³C CP-MAS SS NMR results (Table 2).

Van Niekerk et al. (2008) also observed marked differences in the molecular chemistry for vitrinite-rich Waterberg and inertinite-rich Highveld coal samples, with greater variability observed for the former sample, thus in agreement with the finding of the present study. Further, most compounds detected during the py-GC/MS analysis (Fig. 3) appear to be generally more abundant in the vitrinite-rich sample. Because the coals used by Van Niekerk et al. (2008) were obtained from two different coalfields, an argument could be made that differences between vitrinite-rich Waterberg and inertinite-rich Highveld also reflect differences in source vegetation and/or depositional environments. However, the results of the present study may be taken to contradict such an interpretation, instead suggesting inherent differences in the pyrolysis products for the vitrinite and inertinite macerals.

The varied pyrolysis products for the vitrinite-rich sample, as reflected in Fig. 3, is interpreted to suggest that humification and gelification processes result in the formation of chemical compounds able to give off diverse products upon pyrolysis. Similarly, the lack of diversity for inertinite-rich sample (Fig. 2) reflects the origin pathway for the dominant inertinite macerals. Moroeng et al. (2018a, b, c) proposed that the dominant inertinite macerals of the coal samples of the present study were formed through charring of plant matter. Accepting charring as an origin pathway for these macerals, this process would likely have resulted in the volatilization of certain chemical species, which were driven off, and are currently reflected by the lack of chemical diversity in the chromatogram for the inertinite-rich sample (Fig. 2). The inertinite macerals may essentially be interpreted to have been pre-pyrolyzed in their geologic history as a result of the origin pathway, prior to the py-GC/MS analysis. Further, the lack of chemical diversity for the pyrolysis products of the inertinite-rich sample may also be interpreted to suggest the dominant inertinite macerals, fusinite, semifusinite, and inertodetrinite, have similar origin pathways and source vegetation, especially considering that this sample also has a significant proportion of vitrinite (31.3 vol%—mineral-matter free). The lack of diversity may, therefore, be taken to be consistent with the view that inertodetrinite represents fusinitized and semifusinitized plant matter that was subsequently fragmented (Hagelskamp and Snyman 1988; ICCP 2001; O’Keefe et al. 2013).

Various other workers have argued for a fire-origin for certain inertinite macerals, such as fusinite and semifusinite (Austen et al. 1966; Scott 1989, 2002, 2010; Diessel 1992, 2010; ICCP 2001; Glasspool 2003a, b; Hower et al. 2013; Jasper et al. 2013; O’Keefe et al. 2013; Moroeng et al. 2018a, b, c). Along with inertodetrinite, these are the abundant inertinite macerals in the Witbank coals of this study. The greater proportion of phenolics for the inertinite-rich sample (Table 2) may also be consistent with a fire-origin for the major inertinite macerals of this sample, as concluded by Moroeng et al. (2018a). This is based on the fact that pyrolysis experiments using lignin-rich wood at temperatures below 400 °C produces phenolic moieties (Asmadi et al. 2011; Kawamoto 2017). However, the phenolic moieties produced during the pyrolysis experiments are methoxy-substituted. The absence of methoxy-substituted phenols in the pyrolysis products of the inertinite-rich sample, in particular, may suggest that the methoxy functionalities were expelled during the course of coalification (Austen et al. 1966; Senftle and Larter 1987; Nip et al. 1988; van Krevelen 1993, and references therein). It should be noted that methoxy-substituted phenols have been observed in low rank coals (Senftle and Larter 1987, and references therein; van Krevelen 1993, and references therein).

Despite the diversity exhibited by the vitrinite-rich sample, the pyrolysis products identified for three samples of this study are generally similar (Table 2). The abundance of the compounds identified appear to vary between the samples (Figs. 1, 2, 3), possibly depending on the dominant macerals of the each coal sample. This observation is consistent with the findings of Nip et al. (1992) for single maceral fractions (cutinite, resinite, sporinite, vitrinite, pseudovitrinite, semifusinite, and fusinite) of Carboniferous Age from the Brazil Block seam (Roaring Creek Mine, Parke County) prepared using density gradient centrifugation. For the present study, phenolic compounds are a prominent example of a functional group reflected in the py-GC/MS and ¹³C CP-MAS SS NMR results. According to Powell et al. (1991), phenolics are abundant in the pyrolysis products of low-rank coals as opposed to higher rank counterparts. Similar observations were made by Senftle et al. (1986) using vitrinite concentrates from the Lower Kittanning Seam (Pennsylvania), ranging in maturity from medium rank D to A. The fraction of phenolics for all three coal samples of this study comprises of a range unsubstituted and substituted (e.g., Phenol, 3, 5-dimethyl-; Phenol, 2-ethyl-6-methyl-) derivatives.

Nip et al. (1988) found that p-xylene and m-xylene were the major C₂-alkylbenzenes in the pyrolysis products of vitrinite and inertinite macerals from the Lower and Middle Yorkshire Coalfield (UK), whereas o-xylene was observed to be a major product of liptinite pyrolysis. This finding contrasts somewhat with the results of the present study. In the first instance, m-xylene was not detected in the pyrolysis products of the Witbank coal samples. Secondly, p-xylene and o-xylene were detected in the pyrolysis products for the parent, vitrinite-rich, and inertinite-rich samples, though in varying proportions. For the three samples, p-xylene was consistently observed to be more abundant than o-xylene.

Given that the inertinite-rich and vitrinite-rich samples are products of the same coal, differences in the proportions of the xylene isomers present in the pyrolysis products reflect differences between the major maceral groups, and not rank-dependent differences (Nip et al. 1988). This may be taken to suggest that for Witbank coals, p-xylene and o-xylene are the major pyrolysis products for vitrinite macerals in particular. Alternatively, if the xylene isomers are pyrolysis products for the dominant inertinite macerals as well, this would have to be interpreted to suggest that origin pathways for the vitrinite and inertinite macerals produce chemical species that are able to give off the xylene isomers during pyrolysis. Although forming through different pathways, the dominant inertinite and vitrinite macerals of the samples are derived primarily from the woody tissues of plants, and the xylene isomers are thought to be derived from the lignin component of the wood (Nip et al. 1988). As a result, the presence of xylene isomers in the pyrolysis products for the vitrinite-rich and inertinite-rich samples may suggest that these reflect the comparable source vegetation as opposed to the different origin pathways (Senftle and Larter 1987).

Assuming that the xylene isomers reflect source vegetation for macerals present in these Witbank coals, the detection of largely similar pyrolysis products for both the vitrinite-rich and inertinite-rich samples may be interpreted to suggest a continuum in chemical composition between the dominant vitrinite and inertinite macerals (Nip et al. 1992). Because collotelinite, collodetrinite, fusinite, and semifusinite have similar botanical precursors, similarities in the pyrolysis products of these macerals may be expected, as suggested by Powell et al. (1991). For South African coals, this has been established based chiefly on petrographic observations. As an example, the semifusinite component of South African coals is typically subdivided into a reactive and an inert sub-maceral (Falcon and Snyman 1986; Hagelskamp and Snyman 1988; Snyman 1989; Okolo et al. 2015). The subdivision accounts for reactive semifusinite possessing petrographic and chemical properties more comparable to vitrinite, whereas the inert counterpart possesses properties similar to fusinite of the same coal sample.

According to Nip et al. (1992), pyrolysis products can be used as a “fingerprint” to discriminate between different macerals. In this regard, the findings of the present study may be considered somewhat inconclusive and thus warrant further investigation, specifically with the use of advanced maceral separation techniques. This derives from the differences observed in the chromatograms for inertinite-rich and vitrinite-rich samples, suggesting differences in molecular composition which may be obscured by the relative impurity of the coal samples. With purer maceral fractions, it may be possible to assign pyrolysis products to specific macerals with a higher degree of confidence. However, it should be mentioned that even for largely pure maceral fractions prepared using density-gradient centrifugation from the Argonne Premium Coal Sample Program (for example, fractions comprised of 96.5 vol% liptinite, 100 vol% vitrinite, and 99 vol% inertinite, respectively), Hartgers et al. (1994) encountered difficulties with assigning pyrolysis products to specific macerals, instead suggesting assignment on the basis of an assemblage of functional groups detected.

The study evaluated the organic geochemistry of a vitrinite-rich and an inertinite-rich sample derived from a medium rank C bituminous coal (derived from the same coal, i.e., iso-rank) from the No. 4 Seam Upper of the Witbank Coalfield (South Africa) using pyrolysis py-GC/MS and ¹³C CP-MAS SS NMR. The vitrinite-rich sample (81.3 vol% vitrinite, mineral-matter-free basis) had an abundance of collotelinite and collodetrinite. In contrast, the inertinite-rich counterpart (65.7 vol% inertinite) was dominated by fusinite, and variable proportions of both the reactive and inert semifusinite and inertodetrinite sub-macerals. Pyrolysis products for the vitrinite-rich and inertinite-rich samples generally comprised of similar organic molecules, although the former sample was more diverse and had a greater abundance of most compounds. The lack of diversity for the inertinite-rich sample was interpreted to suggest that the dominant macerals were pre-pyrolyzed in their geologic history, thus supporting a fire-origin for the specific macerals. According to the ¹³C CP-MAS SS NMR analysis, the inertinite-rich sample had a higher fraction of phenolics, which comprised of a range of substituted and unsubstituted functionalities along with polycyclic aromatic compounds (PAHs) reflected in the py-GC/MS results. The abundance of phenolics for the inertinite-rich sample was also interpreted to be consistent with a fire-origin for the dominant inertinite macerals. The C₂-alkylbenzene isomer, p-xylene, was observed to be more abundant in the samples than o-xylene, whereas m-xylene was not detected. The xylene isomers are likely derived from the lignin component of woody plant matter, the precursor for the dominant vitrinite and inertinite macerals of the coal samples, thus suggesting that the isomers reflect source vegetation. The relative petrographic impurity of the samples rendered the assignment of the compounds to specific macerals difficult.