Maturity-related variations in the bitumen compositions of coals from Tara-1 and Toko-1 wells

Transcription

1 New Zealand Journal of Geology & Geophvsics, 2001, Vol. 44: /01/ $7.00/0 The Royal Society of New Zealand Maturity-related variations in the bitumen compositions of coals from Tara-1 and Toko-1 wells STEVE KLLOPS nstitute of Geological & Nuclear Sciences P.O. Box Lower Hutt, New Zealand PAUL WALKER DAVE WAVREK Humble Geochemical Services P.O. Box 789 Humble, TX 77347, USA Abstract The composition of bitumen extracted from two suites of vitrinite-rich New Zealand coals spanning a rank range associated with the generation and expulsion of oil is reported, together with implications for the factors controlling both the expulsion of oil from coals and the various molecular ratios routinely used as maturity parameters. Solvent-extracted bitumen accounts for a small proportion of the free hydrocarbons present in the coals, but the distribution of both compound classes and components within each compound class in solvent extracts appears representative of the bulk of free hydrocarbons, upon comparison with thermal extracts. Although no obvious increase was observed in solvent-extracted bitumen or its constituent saturates and n-alkanes with increasing maturity, there were changes in the carbon-number distributions of n-alkanes indicative of the generation and expulsion of n- alkanes at Rank(S r ) values >12. The amount of aromatics increased sharply at Rank(S r ) c. 13 in the Tara coals, and bulk compositional changes suggest that paraffinic components were expelled at that rank. The decline in carbon preference index (CP) over the Rank(S r ) range in the Tara coals could be explained by the generation of modest additional amounts of n-alkanes, if expulsion occurred at the same rate as generation leading to the observed approximately constant n-alkane concentrations of c. 1 mg/ g C org recovered by solvent extraction. An alternative mechanism is that the generated n-alkanes may remain largely within closed pores, inaccessible to solvent extraction, although some leakage into open pores would be required to account for the observed changes in n-alkane distributions. Generation of aromatic compounds, which exhibit a greater adsorption affinity for coal than saturates, may aid expulsion, which could account for the association of the onset of significant paraffinic oil expulsion with increase in aromatic hydrocarbon concentrations in the Tara coals at Rank(S r ) c. 13. Only after significant amounts of thermally generated aromatics have accumulated and/or C00025 Received 11 April 2000; accepted 17 October 2000 displaced pre-existing aromatics do maturity-related trends become established. These trends are masked at Rank(S r ) values < 13 by the free aromatic hydrocarbons inherited from diagenesis. Keywords aromatic hydrocarbons; bitumen; coal; maturity; n-alkanes; rank NTRODUCTON A recent study of oil-prone, vitrinite-rich coals from New Zealand suggested that expulsion of the paraffinic oils that are characteristic of coaly source beds (Hvoslef et al. 1988; Radke et al. 1990; Powell et al. 1991) can be modelled by consideration of the polymethylene (PM) content and adsorption capacity of the coals (Killops et al. 1998). The term PM is used here to represent the n-alkyl chains in kerogen components such as cutan, suberan, and cutin (Nip etal. 1986a, b; Tegelaar et al. 1989, 1995). A minimum PM content appeared necessary to overcome adsorption and permit expulsion of oil, allowing threshold parameters to be estimated (Killops et al. 1998). Coals from Toko-1 well in Taranaki Basin did not seem to expel oil but those from Tara-1 well in Great South Basin did, by Rank(S) c. 13 (Rank(S r ) c. 12.3). The onset of expulsion in the Tara coals was suggested by Killops et al. (1998) to be indicated by various compositional changes in bulk coals (a relatively rapid decrease in atomic H/C ratio, a corresponding increase in aromaticity from 13 C n.m.r. spectrometry of bulk coals, CL Q 3 " 0.7 Atomic H/C (mmf) i 13 CNMR f a ' (% aromatic C) " =: w 1 <r h Fig. 1 Bulk-coal compositional data for Tara coals. Horizontal rules indicate maturity range (broken line) and mean (solid line) for the onset of significant paraffinic-oil expulsion (after Killops et al. 1998). (Elemental analysis data after Killops et al. 1998; 13 C n.m.r data after Dickinson etal Rank(S r ) scales derived from Fig. 1; mmf = mineral matter free.)

2 158 New Zealand Journal of Geology and Geophysics, 2001, Vol. 44 and a levelling out of the Sl/TOC ratio from Rock-Eval) and extractable bitumen (an increase in the saturates: aromatics and C 3 o:c 31 17a(H)-hopane ratios, and a rapid decrease in carbon preference index (CP) values for C25- C31 fl-alkanes) (Killops et al. 1998). The sharp decline in atomic H content and increasing aromaticity, reproduced in Fig. 1, apparently reflect the preferential expulsion of paraffinic components. Significant changes in the amount and composition of hydrocarbons in the bitumen remaining in the coals after the onset of oil expulsion are to be expected. n previous studies on Carboniferous coals from Germany, the amount of extractable bitumen and vitrinite fluorescence reached maxima at a vitrinite reflectance (VR) of c % (equivalent to Rank(S r ) c ), at which point exsudatinite became visible in pores and CP values declined to c. 1.0 (Radke et al. 1980; Littke & Leythaeuser 1993). n Nesberg-1 borehole, there appeared to be direct evidence for expulsion of oil from the C seam into overlying sandstone, possibly accounting for the stabilisation of C15- C30 n-alkane levels in coal samples at c. 1 mg/g C (Littke & Leythaeuser 1993). The amount of saturates from coals does not always show clear maturity-related trends, although there can be pronounced changes in the composition of the aromatics fraction (Radke et al. 1980). Decreasing levels of pristane but fairly constant levels of phytane in bitumen extracted from German Carboniferous coals have been attributed to hydrocarbon expulsion (Littke & Leythaeuser 1993). This proposition was based on the apparent earlier generation of pristane cf. phytane (Brooks et al. 1969), but is equally valid given the similar generation kinetics but different generation potentials recently suggested for these two isoprenoids (Koopmans et al. 1999). We report the analysis of the extractable bitumen from oil-prone New Zealand coals and the insight it provides into the factors controlling both the expulsion of oil from coals and the various molecular ratios routinely used as maturity parameters. Sample suites from Tara-1 and Toko-1 wells were analysed because they appear to represent the extremes of the ability of coals to expel oil and they encompass an appropriate maturity range (Killops et al. 1998). There are some source-related compositional differences between the two sets of coals. The 16 Toko coals are of Eocene age and are derived primarily from angiosperm remains, in which resin components are dominated by triterpenoids, whereas the 29 Tara coals are of Late Cretaceous to early Paleocene age and are primarily derived from gymnosperm remains, in which diterpenoids dominate resin components (Killops et al. 1995). However, resinite is only a minor component in the coals (Killops et al and references therein) and so the main control on oil potential is likely to be PM content. MATERALS AND METHODS The coals represent new sample sets, as Tara samples used in previous studies (Killops et al. 1998) had been exhausted. Cuttings were sampled from 30 ft (9.14 m) depth intervals in Tara-1 and 3 m intervals in Toko-1 well. After washing, air drying, and picking, material of specific gravity <1.7 was isolated by flotation (sodium polytungstate solution), comprising coal with an ash content of up to c. 40%. Following milling, these coal samples (c. 5 g) were extracted with dichloromethane (c. 50 ml) under ultrasonication (20 min). Saturate and aromatic fractions were isolated from the extracted bitumen by thin-layer chromatography (TLC) on silica gel with hexane eluant, and analysed by GC-FD (HP 5890 GC) using a 25 m x 0.2 mm i.d. HP ultra 2 column (5% phenyl-methylsilicone, 0.33 xm film thickness) with helium carrier gas (15 psi). Splitless injection (1 (ll hexane solution) was used at an injector temperature of 290 C; the detector temperature was 305 C. The temperature programme was 2 min at 40 C then C at 57min, with an isothermal period of 25 min at 300 C. Components were quantified against an internal standard (squalane) at a concentration of 100 (Xg/mg fraction weight. Thermal extraction of finely milled coal samples was performed in the modified injector of a HP 6890 Plus GC. Samples (c. 1 mg) were loaded into an injector liner fitted with a glass-wool plug to hold the sample just above the inlet of a fused-silica pre-column (0.5 m x 0.53 mm i.d., 0.25 j,l film thickness, DB5), where liquid nitrogen trapping took place. The pre-column was connected via a low-deadvolume fitting to a 30 m x 0.32 mm i.d. DB5 column (0.25 (im film thickness). Thermal extraction at 320 C (2 min, helium carrier gas split ratio c. 100:1) with liquid-nitrogen trapping of C$+ hydrocarbons (C1-C4 are detected but not resolved) was followed by a temperature programme of -10 to 320 C at 8 /min, and a final isothermal period of 20 min. Components were quantified by peak-area comparison with an internal standard (adamantane, c. 0.5 ^1 of a 0.1 M solution in CS 2 ) using a HP Chemstation C values of bulk coal samples were obtained using a continuous flow system, with a Carlo Erba elemental analyser coupled to a Geo mass spectrometer (Europa nstruments). Bulk coal pyrolysis was performed on a Rock- Eval 2 instrument. Rank(S r ) values, derived from either atomic compositions or from a combination of calorific values and proximate volatiles content (Suggate 2000), are used throughout this article, but limited reference is also made to Rank(S) values (Suggate 1959) to facilitate comparison with previous publications. Rank(S r ), vitrinite reflectance, and Rock-Eval T max maturity trends derived from the coals in Tara-1 and Toko-1 wells are shown in Fig. 2. RESULTS Bulk coal data The 8 3 C values of all Tara and Toko coals lie in the range to -26.6%o (Fig. 3). Only at Rank(S r ) values >14 is there a possible trend of increasing 8 13 C values for the Tara coals, but there is no clear trend for the Toko coals. Rock-Eval data for Tara and Toko coals are presented in Table 1. The parameter B represents S1/TOC, or the amount of free hydrocarbons normalised to the C content, and H represents S2/TOC, or the hydrocarbon-generating potential remaining in the kerogen after removal of the free hydrocarbons. Solvent extraction removes c. 70% of components contributing to the S signal in the Tara coals. The remaining S1 components are probably predominantly large, relatively polar components (e.g., asphaltene-like material), but they also contain a proportion of the generated hydrocarbons that is not amenable to solvent extraction (either because of strong adsorption or because they are present in pores that are inaccessible to the solvent). Evidence for the latter is that solvent extracts c. 35% of the

3 Killops et al. Maturity-related variations, Tara-1 & Toko Fig. 2 Rank(S r ), vitrinite reflectance, and Rock-Eval T max depth trends for Tara and Toko coals. (Rank(S r ) derived from atomic H/ C and O/C data for Tara coals and CV and VM data for Toko coals. Tara data after Sykes et al. 1998; Toko data after Lowery 1988.) Rank(S r) n-alkanes that can be recovered by thermal extraction (Table 2). Bitumen composition Quantities of solvent-extracted saturates, aromatics, and n- alkanes are shown in Fig. 4. The amounts are shown Fig. 3 coals. -27 h-8-14 Carbon isotopic compositions for bulk, unextracted Tara normalised to C org content and also to S2; the latter data exhibit a smaller fluctuation range. Quantities of all the hydrocarbon fractions from the Tara coals are almost twice as great as those from the Toko coals. Total solventextractable bitumen yields average 30 mg/g C org for the Tara coals and 18 mg/g C org for the Toko coals, which compare with a global average coal value of c. 20 mg/g C org (Durand et al. 1977; Radke et al. 1980). Total extracted C ]5 -C 2 9 n- alkanes amount to only c. 1 mg/g C org in the Tara coals and even less in the Toko coals (Fig. 4). n contrast, thermal extraction of four Tara coals yielded «-alkanes in amounts approaching three times greater than solvent extraction (Table 2), as similarly noted for some Japanese Eocene coals that plot within the New Zealand Coal Band on a Van Krevelen diagram (Radke et al. 1990). The efficiency of solvent extraction appears to increase with increasing maturity, from c. 20 to 50% cf. thermal extraction (Table 2). Amounts of solvent-extracted bitumen, total hydrocarbons, and total paraffins fluctuate but do not rise significantly before the previously proposed onset of expulsion in Tara coals (Killops et al. 1998), which is contrary to expectations if the expulsion threshold is simply controlled by the adsorption capacity for paraffins. The Toko coals also do not exhibit any significant increase in these parameters discernible above the general fluctuations. Other studies involving simple solvent extraction have yielded similarly inconclusive trends in saturates abundance with increasing maturity (Radke et al. 1980, 1990). The «-alkane yields from thermal extraction similarly do not indicate a significant change in n-alkane concentrations with increasing rank. n contrast, there is a distinct increase in aromatic abundance near Rank(S r ) 13 in the Tara coals. Saturates composition There are changes in solvent-extracted paraffin distributions with increasing depth/rank in the Tara and Toko coals. Longchain n-alkanes diminish while short-chains increase, as shown in Fig. 5. These changes become most obvious at

4 160 New Zealand Journal of Geology and Geophysics, 2001, Vol. 44 Rank(S r ) values >14 in Tara coals and >13 in Toko coals. The maturity-related trend in Tara-1 is disturbed by the presence of migrated oil in the uppermost coals, which contributes the predominance of short-chain n-alkanes. Figure 6 shows the n-alkane distributions for the four Tara coals subjected to both solvent and thermal extraction. Despite the difference in recovered amounts, the distributions are similar in each solvent/thermal extract pair, and exhibit the expected shift in dominance from long-chain to short-chain components. The carbon preference index (CP2, based on C25-C31 M-alkanes; Marzi et al. 1993) varies little for Toko coals, Table 1 Rock-Eval data and amounts of solvent-extracted bitumen for Tara and Toko coals. ('Mean sample depth below Kelly bushing; 2 H = hydrogen index (S2/TOC); ^B = bitumen index (S/TOC), data in parentheses for coals after solvent extraction; ^Solvent-extracted bitumen normalised to Rock-Eval TOC - Tara coals only.) Depth (m) Tara coals Toko TOC H 2 B 3 (%) (mg/g C org ),mg/g C org ) coals (3) 11 (2) 21 (7) (10) 8 (2) (4) (5) (5) (6) 19 (4) 21 (3) (12) (9) 22 (6) (16) (12) (7) (9) T m Bitumen 4 (mg/gc org ) but there are two trends of significant decline with increasing depth in the Tara coals (Fig. 5). Decreases in CP with increasing rank have been noted in other coal studies (Radke et al. 1980; Littke & Leythaeuser 1993). The absolute abundances of hopanes in solvent extracts from Tara coals decline rapidly above Rank(S r ) c. 12.5, particularly that of the 17a(H)-homohopanes (C31 in Fig. 7), for both C org and S2 normalised data. The relative increase in C30 at Rank(S r ) values >12.5 (shown in the first panel of Fig. 7) appears to be attributable to its slower rate of decline in abundance compared to the C29 and C31 homologues. These data are consistent with previously reported hopane distributions in Tara coals and the proposed onset of oil expulsion at Rank(S r ) c (Rank(S) c. 13; Killops et al. 1998). Hopane distributions in the Toko coals show similar but less-pronounced trends. Aromatics composition The major aromatic components of the solvent-extracted bitumens from both Tara and Toko coals are C1-C3 substituted naphthalenes and C0-C2 phenanthrenes, and the total abundances in each set are similar to the total amount of C15-C29 n-alkanes in solvent extracts. Quantities of naphthalenes in the Tara and Toko samples are similar, although the depth trends of C2 and C3 components are not as closely correlated in Toko as in Tara coals (Fig. 8). n contrast, the abundance versus depth trends of Co, Ci, and C2 phenanthrenes correlate closely within each sample set, but the total amount of phenanthrenes in Toko samples is only about half that of the Tara samples whether normalised to C org or S2 (Fig. 9). n the Tara coals, there appears to be a general increase in the abundance of naphthalenes and phenanthrenes with rank, similar to that noted in studies on some other Late Cretaceous and Tertiary coals (Radke et al. 1990). Such abundance changes are much less obvious in the Toko coals. The notable difference between the Tara and Toko coals is the rapid increase in abundance of naphthalenes and phenanthrenes at Rank(S r ) c. 13 in solvent extracts from the Tara but not the Toko coals. This abrupt abundance maximum is more pronounced in the S2-normalised plots in Fig. 9, and is reflected in the total aromatics abundance, but there do not appear to be corresponding maxima in the saturates and n-alkane abundances (Fig. 4). The methylated naphthalene and phenanthrene series are less readily quantified in thermal extracts due to the presence of saturates. However, dimethylnaphthalenes (DMNs) were resolved from other components during TE-GC analysis of four Tara samples (Table 2). There is a c. 50% increase in Table 2 Compositional data for solvent-extracted (SE) and thermally extracted (TE) bitumen from four Tara coals (all amounts mg/g C org ). Depth (m) SE bitumen SE n-alkanes (C l5 -C 32 ) SECP2(C 25 -C 3 i) SEDMN TE n-alkanes (C 6 -C 32 ) TE«-alkanes(C 5 -C 32 ) TECP2(C 25 -C 3,) TEDMN SE/TE («C i5 -/7C 33 )

5 Killops et al. Maturity-related variations, Tara-1 & Toko , LO 2- Toko saturates aromatics n-alkanes h 14 Toko saturates aromatics n-alkanes mg/g c org %os2 60 Fig. 4 Composition of bitumen solvent extracted from Tara and Toko coals (rc-alkanes = EC15-C33; absolute abundances normalised to Rock-Eval C org and S2 measurements). f-8 % total n-alkanes CP2 ( n C 2 5-nC 3 i) Fig. 5 Variations in n-alkane distributions with depth in bitumen solvent extracted from Tara and Toko coals. 2.0

6 162 New Zealand Journal of Geology and Geophysics, 2001, Vol. 44 o O c/g O) a> o c re c 3.a re solvent extraction ~* A! \ A V \ / / y> n ^/ ^ < \\\ Fig. 6 Comparison of the abundance of individual «-alkanes recovered from four Tara coals by solvent and thermal extraction (absolute abundances normalised to Rock-Eval C org measurement) n-alkane carbon number n-alkane carbon number DMNs thermally extracted between the samples from 2934 and 3235 m depth, which is less than the c. 80% increase observed in the corresponding solvent extracts. Changes in the distribution of phenanthrenes in coals have previously been correlated with maturity, although it is known that the type of organic matter (Radke et al. 1986; Radke 1988) and clay-catalysed methylation (Alexander et al. 1995) can also significantly affect distributions. Maturity parameters have been based on the differing thermodynamic stabilities of methylphenanthrenes (the a-methylated MP and 9MP are less stable than their P counterparts 2MP and 3MP) and the apparent consumption of phenanthrene by methylation up to around the middle of the oil window followed by its apparent production thereafter by demethylation reactions. The methylphenanthrene index (MP 1) is based on these thermodynamic stabilities and is defined as 1.5(2MP+3MP)/(P+1MP+9MP) (Radke et al. 1982a, b; Radke & Welte 1983). MP 1 values for Tara coals exhibit scatter up to Rank(S r ) c. 14 but thereafter show a tightly constrained increase (Fig. 10). The VRs (R c ) calculated from MP1 values are significantly lower than measured VRs (R m ) at depths >3 km (Rank(S r ) > 12; Fig. 10). n contrast, the R c and R m trends for the Toko coals correlate quite well (Fig. 10). The main control on MP1 seems to be the relative abundance of MP (Fig. 9). n addition, at Rank(S r ) values > 14, the abundance of 2MP gradually increases relative to the other MPs, but overall the levels of MPs decline slightly relative to phenanthrene (Fig. 9). The abundance of methyldibenzothiophenes (MDBTs) in all Tara and Toko samples was so low that any contribution of co-eluting 1MDBT to the 3MP peak in gas chromatograms was negligible and so was not corrected for during quantification. t has been suggested from factor analysis that phenanthrene has only a minor maturity-related influence on MP1 values for coals and that it may be best to use methylphenanthrenes alone (Kvalheim et al. 1987). Kvalheim et al. (1987) further suggested that, rather than relating negatively correlated components (i.e., 2MP+3MP relative to 1MP+9MP), it is probably best to use the parameter (2MP+3MP)/(MPs) (termed F1), or 2MP/ MPs (F2) if methyldibenzothiophene abundance is likely to be a problem (Kvalheim et al. 1987). The parameter F2 is plotted in Fig. 10, together with corresponding calculated VR values from the formula %R F2 = 3.739(F2) (Kvalheim et al. 1987). R F 2 values are consistently higher than R<. and approach R m values at depth for Tara coals, but deviate considerably for the least mature samples and exhibit significant scatter. For Toko coals, R F 2 slightly overestimates R m. Maturity-related differences in dimethyl- and trimethylnaphthalene distributions can be attributed to varying thermodynamic stabilities of a and (3 substituents, as for MPs (Radke et al. 1982b; Alexander et al. 1985). One maturity parameter based on resolved components during routine GC analysis is DNR1, defined as (2,6DMN + 2,7DMN)/(1,5DMN) (Radke et al. 1982b). There is good correlation between DNR1 trends for the two sets of coals over comparable Rank(S r ) ranges (Fig. 10) but, as for MP 1, a uniform maturity-related increase for the Tara coals only becomes clear at Rank(S r ) >13. Superimposing the %R m calibration lines from Fig. 2 provides a reasonably linear fit between DNR1 and VR for both sets of coals of %RDNR ~ 0.15(DNR1) t has been noted previously that DNR1 can correlate with VR (Radke 1987). DSCUSSON Paraffin generation and expulsion The trends observed in the compositions of bulk coals are consistent with the generation of hydrocarbons and, for the Tara suite, the expulsion of significant amounts of paraffmic hydrocarbons at Rank(S r ) c The large changes observed in atomic H/C ratio and aromatic-c content at that maturity (Fig. 1) are too great to be the result of minor changes in organofacies (there appears to be no significant change in maceral composition; Sykes et al. 1998). Similarly, the dramatic decrease in hopane abundance in general and homohopanes in particular (Fig. 7) is also consistent with a significant expulsion event. The C-isotopic data show a general trend of increasing 8 3 C, consistent with the expulsion of hydrocarbons exhibiting the expected kinetic isotopic fractionation from kerogen cracking that leaves the residue enriched in 13 C. However, the data scatter at Rank(S r ) values <14 obscures any major expulsion event. Although there is no obvious increase in total amounts of extractable bitumen (Table 1), or its constituent saturates

7 o T3 2 < o - 14 % total C 29 -C 31 17oc(H)-hopanes mg/g C org %os2 0.2 Fig. 7 Variations in relative and absolute abundances of C29-C31 17<x(H)-hopanes in bitumen solvent extracted from Tara and Toko coals (absolute abundances normalised to Rock-Eval C org and S2 measurements).

8 164 New Zealand Journal of Geology and Geophysics, 2001, Vol. 44 Tara Tara CO. Q. CD Q 4-^ 16 DC < Q. Toko -12 T T 1 ' mg/kg C org %os2 DMN --- STMN - 14 Fig. 8 Variations in abundances of total dimethyl- and trimethylnaphthalenes in bitumen solvent extracted from Tara and Toko coals (absolute abundances normalised to Rock-Eval C org and S2 measurements). 3.0 and M-alkanes (Fig. 4) up to the previously proposed onset of paraffinic oil expulsion in Tara coals, there are clear changes in the carbon-number distribution of n-alkanes (Fig. 5). Solvent extraction does not recover all the free hydrocarbons present in coal, but comparison with the more efficient thermal extraction suggests that compound distributions in the solvent extracts are representative of the bulk of the bitumen. The observed changes indicate that the extractable bitumen experiences loss of «-alkanes at a similar rate to the addition of new w-alkanes derived from PM in the kerogen, so that extracted C15+ «-alkane levels remain fairly constant for the Tara coals. Nevertheless, bulk-coal compositional data (see also Killops et al. 1998) and changes in the abundance of individual components such as 17a(H)- homohopanes (Fig. 7) suggest that a significant amount of saturates must be lost at Rank(S r ) c A number of processes may account for the above observations: (1) The onset of paraffin generation and expulsion occur in rapid succession, together with aromatic generation, and aromatics are preferentially adsorbed by the coal matrix, so there is no obvious increase in paraffin concentrations in the bitumen of the coals. (2) The greater proportion of generated paraffins is not amenable to simple solvent extraction and remains in what can be thought of as closed pores (Sjago et al. 1983; Radke et al. 1990; Littke & Leythaeuser 1993), so that the effects of paraffin generation are only partially reflected in the composition of bitumen extracted from predominantly open pores. There must be some communication between open and closed pores, because of the changes that occur in n-alkane and hopane distributions. Consequently, the lack of paraffin buildup before the main phase of oil expulsion would require leakage from open pores at a rate similar to the rate of transfer of thermally generated paraffins between closed and open pores. The closed pores may be represented, at least to a partial extent during the early stages of n- alkane generation, by the PM in the kerogen, in which the early-generated paraffins remain dissolved (similar to analytes in a GC stationary phase), escaping only relatively slowly by diffusion. (3) The driving force behind large-scale expulsion could be displacement by aromatics, which preferentially occupy adsorption sites (e.g., Sandvik et al. 1992). Expulsion would then require sufficient aromatics to be generated, possibly coinciding with the peak in aromatics abundance at Rank(S r ) c. 13 in Tara coals. f this were the sole factor controlling expulsion, the generation of paraffins would have to coincide with the peak in aromatics generation if no build-up in paraffins is to occur. t is likely that more than one of the above processes applies. n particular, process (2) is likely, otherwise special conditions are required to account for the lack of paraffin build-up. Analyses involving thermal extraction-gc followed by pyrolysis-gc of the Tara and Toko coal suites are planned, which may provide additional information on the composition of generated bitumen, its amenability

9 Killops et al. Maturity-related variations, Tara-1 & Toko Tara h8 h8 / ' : ' Toko \ mg/kg C org %os2 i 1 i i i i Toko - P --- SMP EDMP - 14 i ' i Tara i Q. Toko - Toko - ' ' mg/kg C org 3MP 2MP 9MP 1MP MP --- 2MP 9MP 1MP Fig. 9 Variations in abundances of phenanthrene, total and individual methylphenanthrenes, and total dimethylphenanthrenes in bitumen solvent extracted from Tara and Toko coals (absolute abundances normalised to Rock-Eval C org and S2 measurements).

10 166 New Zealand Journal of Geology and Geophysics, 2001, Vol MP1 o o 1.0 Tara Toko %VR 1.2 DNR1 Fig. 10 Correlation of maturity parameters derived from solvent-extracted methylphenanthrenes (MPl, %Rc, and %RF2) and dimethylnaphthalenes (DNR1) with measured VR values. MPl =.5(2MP+3MP)/(P+1MP+9MP) and %R C = O.6(MP1) (after Radke et al. 1982a, b; Radke & Welte 1983); F2 = 2MP/SMPs, and %R F2 = 3.739(F2) (after Kvalheim et al. 1987); DNR1 = (2,6DMN+2,7DMN)/1,5DMN (after Radke et al. 1982b). towards solvent extraction, and its quantitative relationship towards the remaining paraffin potential. The lower degree of fluctuation of abundance normalised to S2 cf. C org for compound classes among the saturates and aromatics extracted from the Tara coals suggests that retention within the hydrocarbon-generating component of the coal dominates over the general adsorptive capacity of the total C content of the coals, as suggested above for the n-alkanes and PM. n contrast, there is no clear evidence for similar behaviour among the Toko coals. Unless the main phase of oil expulsion occurred from the Toko coals before they reached Rank(S r ) 11, it would seem that the Toko coals have limited paraffinic oil potential despite similar H values to the Tara coals of similar rank. Amount of expelled paraffins The changes in atomic H/C and 3 C n.m.r. saturates/ aromatics ratios near Rank(S r ) 12.5 (Fig. 1) can potentially be used to estimate the amount of paraffinic oil expelled, which is mainly in the form of n-alkanes for typical oil-prone coals. Atomic H/C ratios are affected by the loss of H2O, CO2, and CO as well as by hydrocarbon loss. f the change in atomic H/C ratio at Rank(S r ) c is solely attributed to paraffins with an empirical formula of CH9, some 60 mg/ g C must be expelled (see Appendix 1). Associated loss of CH4 (and, to a lesser extent, other hydrocarbon gases) would lower the amount of paraffinic oil lost. Aromaticity from 13 C n.m.r. data provides an upper limit on amounts of expelled paraffins, because loss of all carbon gases would reduce the estimate based on paraffins alone of c. 120 mg CH 2 /g C at Rank(S r ) c. 13 (see Appendix 1). There are no clear trends in bulk coal 5' 3 C values at Rank(S r ) c (Fig. 3), and estimates of the amounts of expelled paraffins (5 13 C = c %o; after Killops et al. 1994; Killops 1996) would, in any event, be severely affected by losses of relatively small amounts of isotopically light methane (with a 8 3 C Of C. -46%o, based on 8 13 C (me thane-source) «-20%o (Clayton 1991) and a typical initial coal 8' 3 C value of -26.0%o). t is likely that the amount of paraffins expelled near Rank(S r ) 12.5 is less than 60 mg/g C. CP values are generally high in the bitumens from immature higher plant derived material because of the dominant biosynthetic and diagenetic pathways that lead to C25-C33 n-alkanes with odd-numbered C chains. With increasing maturity, thermal cracking of kerogen produces n-alkanes with random chain-length distributions (i.e., CP = 1), so decreasing CP can be used as a maturity measurement (e.g., Killops & Killops 1993 and references therein). Of the two trends of decreasing CP with depth in the Tara coals, the deeper trend may be accounted for by an increasing contribution of n-alkanes with a random chainlength distribution (i.e., CP = 1) derived from PM in the

11 Killops et al. Maturity-related variations, Tara-1 & Toko kerogen. The shallower trend is associated with the zone of migrated hydrocarbons in the uppermost coals, the effects of which are not straightforward. The low CP value immediately above the deeper maximum is to be expected where migrated hydrocarbons contain a significant proportion of C25-C31 n-alkanes with a CP of c. 1, but the higher CP values at shallower depth suggest that the associated migrated hydrocarbons contain substantially lower proportions of C25-C31 w-alkanes, such that the immature bitumen contribution to CP dominates. The decline in CP from c to 1.15 over the Rank(S r ) range (Fig. 6) would require an increase in abundance of n- alkanes in the C25-C31 range of some 450%, if solely due to the addition of n-alkanes with a CP of 1 (Appendix 1). f the model is adjusted to allow n-alkanes to escape at the same rate they are generated from PM, to accommodate the lack of increase in n-alkane abundance in Fig. 4, the required amount of n-alkane generation in the C25-C31 range with a CP of 1 would not need to be more than c. 180% of the original quantity of free n-alkanes (which is no more than c. 3 mg/g C org ). So, although the change in CP appears dramatic, it can be accounted for by relatively modest levels of n-alkane generation. Aromatic maturity indicators t has been proposed that variations in MP distributions are more closely related to Q5+ hydrocarbon generation in type- ll kerogens than are changes in DMN distributions, although both can be used to assess the maturity of these kerogens (Radke et al. 1986). However, if the solvent-extract concentrations accurately reflect changes in the levels of aromatic hydrocarbons, there appears to be no distinction between methylated phenanthrenes and naphthalenes in terms of maturity-related changes in their abundance in the Tara and Toko coals (Fig. 7). The main influence on aromatic-hydrocarbon distributions at low maturities appears to be the contribution from free aromatic hydrocarbons inherited from diagenesis. t appears that the amounts and distributions of MPs and DMNs from this inherited freehydrocarbon component can vary significantly and obscure maturity-related trends at Rank(S r ) values <13, as seen in the plots ofmpl,f2, and DNR1 infig. 10. At higher ranks, a tightly constrained increase in aromatic maturity parameters is observed (Fig. 10), suggesting that only after expulsion commences do variations in the distribution of generated aromatics exhibit a uniform maturity-related trend. Even over rank ranges in which maturity-related trends in aromatic distributions are clearer, caution should still be exercised in extrapolating equivalent VR values from universal calibration equations, because Fig. 10 shows that the fits to measured reflectance trends are not precise. Closer fits in predicted and measured VR trends may be possible by calibration of individual coal families, which would compensate for variations in the relative abundances of precursors of individual MPs and DMNs associated with variations in plant assemblages. For example, the same gymnosperm-derived isopimaroid precursors that are responsible for high concentrations of isopimarane and pimanthrene in Late Cretaceous coals like those from Tara- 1 well could also generate MP. A similar observation has been made for some German coals (Radke et al. 1990). n contrast, Eocene coals like those from Toko-1 contain much lower levels of isopimaroid precursors. This may account for the higher abundance of MP relative to other MPs in the Rank(S r ) range c in the Tara coals cf. the Toko coals (Fig. 9). As suggested by Kvalheim et al. (1987), the influence of phenanthrene abundance on MP1 is minor; phenanthrene abundance broadly mirrors total MP abundance (Fig. 9). The parameter F2, based upon only MP distributions (Kvalheim et al. 1987), fits a linear depth-related trend more closely than does MP1 (Fig. 10), although the equivalent VR values derived from F2 may not provide a better fit to measured values than those derived from MP1 for all coal types. CONCLUSONS Compositional analysis of bitumens extracted from vitriniterich coals can provide information on the generation and expulsion of paraffinic oil, based on the data obtained from vitrinite-rich coals of predominantly angiosperm origin from Toko-1 well (Taranaki Basin) and of predominantly gymnosperm origin from Tara-1 well (Great South Basin). Solvent-extracted bitumen may not represent all the free hydrocarbons present in coal, but the distribution of compound classes and components within each compound class in solvent extracts appears representative upon comparison with thermal extracts. Although no obvious increase was observed in solvent-extracted bitumen or its constituent saturates and n-alkanes with increasing maturity, there were changes in the carbon-number distributions of n-alkanes that suggest there is generation and expulsion of new n-alkanes from the coals at Rank(S r ) values >11, or possibly even at lower maturity. Bulk compositional changes in the Tara coals, such as variations in atomic H/C and 13 C n.m.r. values, suggest losses of a few tens of mg paraffins/g C org near Rank(S r ) f the approximately constant n-alkane concentration of c. 1 mg/g C org obtained from solvent extraction of bitumen reflects accurately the total n-alkane concentrations in the Tara coals, one possible explanation is that n-alkanes are expelled at approximately the same rate as they are generated. On this basis, the decline in CP for n-alkanes in the range C25-C31 in the Tara coal bitumens over the Rank(S r ) range suggests that only relatively modest n-alkane generation is required (amounting to only c. 180% of initial free n-alkanes). An alternative explanation of approximately constant n- alkane concentrations is that the generated n-alkanes may remain largely within closed pores, where they are not readily amenable to solvent extraction. However, a degree of leakage into open pores would be required, together with displacement of an equivalent amount of n-alkanes from the open pores, to account for the observed changes in n-alkane distributions in solvent extracts. t can be envisaged that generation of aromatic compounds facilitates expulsion because of the greater adsorption affinity of coals for aromatics. The proposed major onset of paraffinic oil expulsion in the Tara coals at Rank(S r ) c may be related to the apparently contemporaneous pronounced increase in aromatic hydrocarbon concentrations. The Toko coals exhibit signs that some generation and loss of hydrocarbons must have occurred, if only on a gradual basis, in order to account for changes in the distributions of extractable n-alkanes. However, there is no sign of a major expulsion event.

12 168 New Zealand Journal of Geology and Geophysics, 2001, Vol. 44 Changes in the distributions of components within various aromatic hydrocarbon classes, such as MPs and DMNs, are routinely used as maturity indicators for types and kerogens. However, data from the Tara and Toko coals suggest that maturity-related changes may only become clear at Rank(S r ) >13. At lower maturities, the free aromatic hydrocarbons inherited from diagenesis tend to cause random distributions, and it is only after significant amounts of thermally generated aromatics have accumulated, and have displaced the inherited aromatics in the case of the Tara coals, that maturity-related trends are established. Even then, equivalent VR values obtained from previously proposed universal conversion equations are not always accurate. More accurate absolute maturity indications probably require calibration of individual molecular aromatic maturity indicators against measured VR or Rank(S r ) for different coal types. ACKNOWLEDGMENTS We are grateful to Karyne Rogers (nstitute of Geological & Nuclear Sciences, Lower Hutt, New Zealand) for the carbon isotopic analyses, Lloyd Snowdon (Geological Survey of Canada, Calgary, Canada) for Rock-Eval analyses, Henry McGrath (AgriQuality, Lower Hutt, New Zealand) for GC analyses, and Rod Weston (RL, Lower Hutt, New Zealand) for helpful review comments. We thank the Foundation for Research, Science and Technology, New Zealand (contract number C05806) for supporting this research. GNS publication number REFERENCES Alexander, R.; Kagi, R..; Rowland, S. J.; Sheppard, P. N.; Chirila, T. V. 1985: The effects of thermal maturity on distributions of dimethylnaphthalenes and trimethylnaphthalenes in some ancient sediments and petroleums. Geochimica et Cosmochimica Ada 49: Alexander, R.; Bastow, T. P.; Fisher, S. J.; Kagi, R : Geosynthesis of organic compounds:. Methylation of phenanthrene and alkylphenanthrenes. Geochimica et Cosmochimica Ada 59: Brooks, J. D.; Gould, K.; Smith, J. W. 1969: soprenoid hydrocarbons in coal and petroleum. Nature 222: Clayton, C. 1991: Carbon isotope fractionation during natural gas generation from kerogen. Marine and Petroleum Geology 8: Dickinson, W. W.; Newman, R. H.; Collen, J. D. 1991: Maturation parameters measured from 13C NMR spectra compiled from Upper Cretaceous and Tertiary coals, New Zealand. Geology Board of Studies Publication 9. Wellington, Victoria University of Wellington. Durand, B.; Nicaise, G.; Roucache, J.; Vandenbroucke, M.; Hagemann, H. W. 1977: Etude geochimique d'une serie de charbons. n: Campos, R.; Goni, J. ed. Advances in organic geochemistry Enadimsa. Pp Hvoslef, S.; Larter, S. R.; Leythaeuser, D. 1988: Aspects of generation and migration of hydrocarbons from coalbearing strata of the Hitra formation, Haltenbanken area, offshore Norway. Organic Geochemistry 13: Killops, S. 1996: A geochemical perspective of oil generation in New Zealand basins. n: 1996 New Zealand Petroleum Conference Proceedings, Vol. 1. Wellington, Ministry of Commerce. Pp Killops, S.; Killops, V. J. 1993: An introduction to organic geochemistry. Harlow, Longman. Killops, S. D.; Woolhouse, A. D.; Weston, R. J.; Cook, R. A. 1994: A geochemical appraisal of oil generation in the Taranaki Basin, New Zealand. American Association of Petroleum Geologists Bulletin 78: Killops, S. D.; Raine, J..; Woolhouse, A. D.; Weston, R. J. 1995: Chemostratigraphic evidence of higher-plant evolution in the Taranaki Basin, New Zealand. Organic Geochemistry 23: Killops, S. D.; Funnell, R. H.; Suggate, R. P.; Sykes, R.; Peters, K. E.; Walters, C; Woolhouse, A. D.; Weston, R. J.; Boudou, J.-P. 1998: Predicting generation and expulsion of paraffinic oil from vitrinite-rich coals. Organic Geochemistry 29: Koopmans, M. P.; Rijpstra, W.. C; Klapwijk, M. M.; de Leeuw, J. W.; Lewan, M. D.; Sinninghe Damste, J. S. 1999: A thermal and chemical degradation approach to decipher pristane and phytane precursors in sedimentary organic matter. Organic Geochemistry 30: Kvalheim, O. M.; Christy, A. A.; Telnaes, N.; Bjarseth, A. 1987: Maturity determination of organic matter in coals using the methylphenanthrene distribution. Geochimica et Cosmochimica Ada 51: Littke, R.; Leythaeuser, D. 1993: Migration of oil and gas in coals. n: Law, B. E.; Rice, D. D. ed. Hydrocarbons from coal. AAPG Studies in Geology 38: Lowery, J. H. 1988: Catalogue of vitrinite reflectance measurements and coal analyses from oil prospecting wells in Taranaki Basin, New Zealand. nstitute of Geological & Nuclear Sciences Report Ml68. Lower Hutt, New Zealand. Marzi, R.; Torkelson, B. E.; Olson, R. K. 1993: A revised carbon preference index. Organic Geochemistry 20: Nip, M.; Tegelaar, E. W.; de Leeuw, J. W.; Schenck, P. A.; Holloway, P. J. 1986a: A new non-saponifiable highly aliphatic and resistant biopolymer in plant cuticles: evidence from pyrolysis and ' 3 C n.m.r. analysis of present day and fossil plants. Naturwissenschaften 73: Nip, M.; Tegelaar, E. W.; Brinkhuis, H.; de Leeuw, J. W.; Schenck, P. A.; Holloway, P. J. 1986b: Analysis of modern and fossil plant cuticles by Curie-point Py-GC and Curie-point Py- GC-MS: recognition of a new highly aliphatic and resistant biopolymer. Organic Geochemistry 10: Powell, T. G.; Boreham, C. J.; Smyth, M.; Russell, N.; Cook, A. C. 1991: Petroleum source rock assessment in non-marine sequences: pyrolysis and petrographic analysis of Australian coals and carbonaceous shales. Organic Geochemistry 17: Radke, M. 1987: Organic geochemistry of aromatic hydrocarbons. n: Brook, J.; Welte, D. ed. Advances in petroleum geochemistry, Vol. 2. London, Academic Press. Pp Radke, M. 1988: Application of aromatic compounds as maturity indicators in source rocks and crude oils. Marine and Petroleum Geology 5: HA-liii. Radke, M.; Welte, D H. 1983: The methylphenanthrene index (MP1): a maturity parameter based on aromatic hydrocarbons. n: Bjoray, M. et al. ed. Advances in organic geochemistry Chichester, John Wiley & Sons Ltd. Pp Radke, M.; Schaefer, R. G.; Leythaeuser, D.; Teichmiiller, M. 1980: Composition of soluble organic matter in coals: relation to rank and liptinite fluorescence. Geochimica et Cosmochimica Ada 44: Radke, M.; Welte, D. H.; Willsch, H. 1982a: Geochemical study on a well in the Western Canada Basin: relation of the aromatic distribution pattern to maturity of organic matter. Geochimica et Cosmochimica Ada 46: 1-10.

13 Killops et al. Maturity-related variations, Tara-1 & Toko Radke, ML; Willsch, H.; Leythaeuser, D.; Teichmuller, M. 1982b: Aromatic components of coal: relation of distribution pattern to rank. Geochimica et Cosmochimica Ada 46: Radke, M.; Welte, D. H.; Willsch, H. 1986: Maturity parameters based on aromatic hydrocarbons: influence of the organic matter type. Organic Geochemistry 10: Radke, M.; Willsch, H.; Teichmuller, M. 1990: Generation and distribution of aromatic hydrocarbons in coals of low rank. Organic Geochemistry 15: Sandvik, E. 1.; Young, W. A.; Curry, D. J. 1992: Expulsion from hydrocarbon sources: the role of organic absorption. Organic Geochemistry 19: Sjago, Cs.; Maxwell, J. R.; Mackenzie, A. S. 1983: Evaluation of fractionation effects during the early stages of primary migration. Organic Geochemistry 5: Suggate, R. P. 1959: New Zealand coals: their geological setting and its influence on their properties. New Zealand Department of Scientific and ndustrial Research Bulletin 134. Suggate, R. P. 2000: The Rank (S r ) scale: its basis and its applicability as a maturity index for all coals. New Zealand Journal of Geologv and Geophysics 43: Sykes, R.; Suggate, R. P.; Funnell, R. H.; Boudou, J. P.; Killops, S. D.; Cook, R. A.; Newman, J. 1998: Petroleum generation and expulsion from coals and coaly shales in Tara-1, Great South Basin, New Zealand: facies controls, maturity thresholds and expulsion history. n: Handbook and abstracts, nternational Conference on Coal Seam Gas and Oil, Brisbane. P. 22. Tegelaar, E. W.; de Leeuw, J. W.; Derenne, S.; Largeau, C. 1989: A reappraisal of kerogen formation. Geochimica et Cosmochimica Acta 53: Tegelaar, E. W.; Hollman, G.; van der Vegt, P.; de Leeuw, J. W.; Holloway, P. J. 1995: Chemical characterization of the periderm tissue of some angiosperm species: recognition of an insoluble, non-hydrolyzable aliphatic biomacromolecule (Suberan). Organic Geochemistry 23: 239. APPENDX 1 Estimates of «-alkane expulsion near Rank(S r ) 12.5 based on the changes in atomic H/C ratio, aromatic C content from 13 C n.m.r., and CP values. (1) Atomic H/C ratio This ratio changes from c at Rank(S r ) 11.5 to c at Rank(S r ) 12.5, and can be represented by the equation: CiHo.85 = C(i-x)H(o.85-2x) + C X H2 X, assuming the only C and H loss is in the form of «-alkanes (CH2). This yields x = 0.05, or c. 60 mg CH2/ (2) 3 C n.m.r. aromaticity The proportion of aromatic C atoms changes from c at Rank(S ik(s r ) 13.0 to c at Rank(S r ) 13.5, and can be represented by the equation: C sat + C aro = C sat + C( aro _ X ) + C x, where C x is the C in expelled «-alkanes, and no other C loss occurs. The equation reduces to: 0.67/(l-x) = 0.75, yielding x = 0.11, or c. 120 mg CH 2 /g C. (3) CP The CP value changes from c at Rank(S r ) 11 to c at Rank(S r ) 13. Assuming that the «-alkanes in the C25-C31 range generated from kerogen have a random chain length (i.e., CP = 1), the change in CP by simple addition of such /j-alkanes is given by the equation: 1.15= 1.85/(l+x)+ 1.00(x)/(l+x), which yields x = 4.67, or c. 470% more /j-alkanes than originally present have to be generated. Performing an iterative calculation allowing 5% additions and the immediate escape of an equal amount of the resulting mixture gives x = 1.8.