10
Energy & Fuels 1996, 10, 10-18
Neoformation of Inert Carbon during the Natural Maturation of a Marine Source Rock: Bakken Shale, Williston Basin Gary P. A. Muscio*,† and Brian Horsfield Institute of Petroleum and Organic Geochemistry, Research Centre Ju¨ lich, 52425 Ju¨ lich, Germany Received September 28, 1995. Revised Manuscript Received October 24, 1995X
The Bakken Shale (Mississippian/Devonian) of the Williston Basin was used as a natural laboratory for comparing the disproportionation reactions occurring in nature with those occurring in pyrolysis experiments. A uniform kerogen type and a broad maturity range made this possible. Mass balance calculations, analytical pyrolysis, and MSSV simulation experiments together provided strong evidence that generative yields from open system pyrolysis are not equal to potential petroleum yields in nature for this particular source rock. This is because inert kerogen formation is enhanced under both natural and closed system simulation conditions, ostensibly because of aromatization and condensation reactions involving primary aromatic structures and possibly cross-linked moieties. The outcome is that mass balance models normalized to inert carbon give overestimates of petroleum generated in nature. This is the second case where we have been able to substantiate that the phenomenon occurs in nature. The presence of high concentrations of aromatic moieties in the most immature equivalents is common to both the Alum Shale and Bakken Shale. A maturity zonation based on the residues of simulated and natural maturation allowed the quantitative and qualitative evolution of liquid natural petroleums to be predicted. The occurrence of enhanced concentrations of low molecular weight hydrocarbons in the immature zone nevertheless remains enigmatic.
Introduction The natural maturation of organic matter in petroleum source rocks proceeds in response to progressively increasing temperature and pressure. Similar to the disproportionation process of hydrocarbons in reservoirs,1 the residual organic matter becomes more and more aromatic as hydrogen-rich mobile petroleum is generated. For many years, the quantitative evolution of petroleum generation with increasing maturity was investigated by using the mobile, extractable organic matter fraction to define the “oil generation window”.2-5 However, because of substantial migration-related losses, more emphasis is nowadays placed on the concentration and composition of the residual kerogen. For evaluating the extent of the petroleum generation process itself, the classical approach has been to monitor evolutionary trends in atomic H/C and O/C ratios with due consideration of original kerogen type influence. Directly correlated to changes in elemental composition, the continuous decrease of pyrolysate yields (Rock-Eval † Present address: Integrated Exploration Systems GmbH, Bastionstr. 11-19, 52425 Ju¨lich, Germany. X Abstract published in Advance ACS Abstracts, December 1, 1995. (1) Connan, J.; Le Tran, K.; and van der Weide, B. Proc. 9th World Pet. Congr. 1975, 2, 171-178. (2) Landes, K. K. Bull. Am. Assoc. Petrol. Geol. 1967, 51(6), 828841. (3) Louis, M. C.; Tissot, B. P. Proc. 7th World Pet. Congr. 1967, 2, 47-60. (4) Tissot, B. P.; Califet-Debyser, Y.; Deroo, G.; Oudin, J. L. Bull. Am. Assoc. Petr. Geol. 1971, 55, 2177-2193. (5) Albrecht, P.; Vandenbroucke, M.; Mandengue´ M. Geochim. Cosmochim. Acta 1976, 40, 791-799.
0887-0624/96/2510-0010$12.00/0
Hydrogen and Oxygen Indices6) is the most common reference for mass balance calculations.7-9 Organic carbon in the kerogen can be visualized as being either reactive or inert. Reactive carbon is that which evolves in the form of volatile products during laboratory pyrolysis (most common upper temperature of 550-600 °C) whereas inert carbon is that which is nonvolatile. The proportion of reactive carbon decreases relative to inert carbon in nature, so that Hydrogen Indices for immature kerogens are relatively high and values for mature and overmature equivalents are substantially lower. This could theoretically be caused by one of two probable mechanisms. In one, the absolute quantity of inert carbon can be considered to remain fixed and the absolute quantity of volatilizable components to decrease, whereas in the other, the inert carbon components increase in absolute abundance concomitantly with reactive carbon loss (Figure 1). In the Cooles7 model, inert carbon is always assumed to remain constant throughout natural maturation and is used as the means for normalizing the mass balance calculation. For this approach to be valid, the mechanism of petroleum generation in nature has to be the same as that of laboratory pyrolysis, irrespective of the type and (6) Espitalie´, J.; Deroo, G.; Marquis, F. Rev. Inst. Fr. Pe´ t. 1985, 40, 563-579, 755-784. (7) Cooles, G. P.; Mackenzie, A. S.; Quigley, T. M. In Advances in Organic Geochemistry 1985; Leythaeuser, D., Rullko¨tter, J., Eds.; Org. Geochemistry 10; Pergamon Press: Oxford, U.K., 1986; pp 235-245. (8) Rullko¨tter, J.; Leythaeuser, D.; Horsfield, B.; Littke, R.; Mann, U.; Mu¨ller, P. J.; Radke, M.; Schaefer, R. G.; Schenk, H.-J.; Schwochau, K.; Witte, E.-G.; Welte, D. H. In Advances in Organic Geochemistry 1987; Mattavelli, L., Novelli, L. Eds.; Pergamon Journals: Oxford, U.K.; 1988. (9) Larter, S. R. In Pet. Geochem. Exploration Norwegian Shelf 1985, 269-286.
© 1996 American Chemical Society
Formation of Inert Carbon during the Natural Maturation
Figure 1. Schematic evolution of organic matter (expressed as Rock-Eval terms) as a function of increasing thermal stress for two different scenarios: In sketch A, the inert kerogen portion remains constant during maturation, whereas in B inert carbon is being progressively formed.
structure of its macromolecular (principally kerogen) components, in other words, to be dominated by thermal cracking reactions. Objections to this assumption have been raised on the grounds that coke formation from heteroatomic pyrolysis products in oil shale retorts is dependent on heating rate and pressure.10,11 Additionally, because expelled “oil” yields from hydrous pyrolysis were lower than Rock-Eval yields, and because this oil was considered more similar to natural petroleum, closed system pyrolysis was thought to give a better indication of yields.12 Having said this, it is difficult to completely dismiss open system pyrolysis on the basis of laboratory considerations alone. All pyrolysates are actually dissimilar to petroleum, irrespective of whether open or closed system, in that they contain far higher proportions of polar and high molecular weight compounds.13,14 While it is true that the composition and yield of individual hydrocarbons is system dependent, and progress has recently been made in deciphering the roles played by water and pressure,15 it is still unclear which system is best at simulating natural maturation. Evaluating natural maturation series is the key to addressing this problem. Earlier, we concluded from a combination of field observations and laboratory experiments (on isolated kerogens) that inert carbon had formed during maturation of the Alum Shale (Cambrian, Scandinavia), purportedly via aromatization/condensation reactions. The presence of cross-linking from either radiation damage or inherited from a biological precursor may have made the kerogen follow this reaction pathway during maturation.16 Recently, Lewan et al.17 have visualized petroleum formation reactions in all source rocks as (10) Burnham, A. K.; Braun, R. L. Org. Geochem. 1990, 16, 27-39. (11) Burnham, A. K. Energy Fuels 1991, 5, 205-214. (12) Lewan, M. D. Philos. Trans. Soc. A 1985, 315, 123-134. (13) Horsfield, B. In Petroleum and Basin Evolution; Welte, D. H., Baker, D. R., Horsfield, B. Springer-Verlag: Heidelberg, in press. (14) Larter, S. R.; Horsfield, B. In Organic Geochemistry; Engel, M., Macko, S., Eds.; Plenum: New York, 1993; pp 271-287. (15) Michels, R.; Landais, P.; Philp, R. P.; Torkelson, B. E. Energy Fuels 1995, 9, 204-215. (16) Horsfield, B.; Bharati, S.; Larter, S. R.; Leistner, F., Littke, R., Schenk, H. J.; Dypvik, H. In Early Organic Evolution: Implications for Mineral and Energy Resources; Schidlowwski, M., et al., Eds.; Springer-Verlag: Berlin. (17) Lewan, M. D.; Comer, J. B.; Hamilton-Smith, T.; Hasenmueller, N. R.; Guthrie, J. M.; Hatch, J. R.; Gautier, D. L.; Frankie, W. T. U.S. Geol. Surv. Bull. 1995, 2137.
Energy & Fuels, Vol. 10, No. 1, 1996 11
resulting from a combination of cross-linking and cracking reactions and concluded that open system pyrolysis strongly overemphasizes the relative importance of the cracking pathway. One outcome was that expelled petroleum masses calculated for the New Albany Shale (Illinois Basin) appear to have been grossly overestimated when constant inert carbon was assumed. These are cases where the natural neoformation of inert carbon can be reasonably deduced. There are, however, numerous other examples where inert carbon remains essentially constant. For instance, the Cooles model was calibrated using many well-known source rocks, the list including the Kimmeridge Clay, the Kingak Shale of Alaska, and the Lias δ of the Lower Saxony Basin.7 We may conclude from all of this that some kerogens, such as those rich in naphthenoaromatic or cross-linked structures, are inherently prone to aromatization and condensation reactions whereas others such as highly aliphatic kerogens are not (see average structural models of Be´har and Vandenbroucke18). It was with these considerations in mind that a study of the Bakken Shale (Mississippian/Devonian), Williston Basin, was undertaken. The study was performed using a complete maturity sequence. The broad spectrum of maturity (0.3 to >1.0% Ro19) ensured that the progressive and successive effects of all phases of petroleum generation on this Type II kerogen had been imparted on recoverable samples. Secondly, depositional conditions (input of organic matter) away from the margins could be considered uniform on a basinwide scale.19-23 This simplicity made the Bakken Shale well suited to the specific task at hand, which was to investigate the fates of geomacromolecular components during maturation. Experimental and Analytical Section Twenty-eight wells from the North Dakota portion of the Williston Basin provided a total of 181 core samples which were submitted to analyses as described below. Screening Analyses. All rock samples were analyzed for total organic carbon (using a LECO IR-112 analyzer) and by the Rock-Eval pyrolysis method6 following established procedures. Fifty-five selected whole rock pieces representing all wells had also been examined petrographically using a combination of reflected light and fluorescence techniques.23 Pyrolysis Techniques. Py-GC was performed on 41 whole rock samples from 14 wells. Principally, the high organic richness (10-22% TOC) as well as a relatively low clay content (ca. 30%24) of the Bakken Shale precluded severe mineral matrix effects.25,26 Each sample was placed into the central part of a glass tube (26 mm long; inner sleeve diameter 3 mm). The remaining volume was filled with cleaned quartz wool (630 °C in air, 1 h). After flushing at 300 °C (5 min) to remove (18) Be´har, F.; Vandenbroucke, M. Org. Geochem. 1987, 11, 1524. (19) Webster, R. L. Hydrocarbon Source Rocks of the Greater Rocky Mountain Region. (20) Meissner, F. F. Williston Basin Symp. 1978, 207-227. (21) Krystinik, K. B.; Charpentier, R. R. Bull. Am. Assoc. Pet. Geol. 1987, 71, 95-102. (22) Martiniuk, C. D. Manitoba Energy and Mines, Petroleum Open File 1988, Report POF 8-88. (23) Muscio, G. P. A. Ph.D. Thesis, RWTH Aachen, Germany, 1995. (24) Cramer, D. D. Montana Geological Society 1991 Guidebook to Geology and Horizontal Drilling of the Bakken Formation 1991, 117140. (25) Espitalie´, J.; Madec, J. M.; Tissot, B. Bull. Am. Assoc. Pet. 1980, 64, 59-66. (26) Horsfield, B.; Douglas, A. G. Geochim. Cosmochim. Acta 1980, 44, 1110-1131.
12
Energy & Fuels, Vol. 10, No. 1, 1996
Figure 2. Evolution of hydrocarbon generation potential of Bakken Shale kerogens (Rock-Eval HI) as a function of maturity. volatile material, the sample was pyrolyzed using programmed heating from 300 °C (50 °C/min.) to 600 °C (3 min isothermal). The products were cryogenically trapped and separated by online gas chromatography. An HP-1 column (1.65 µm film thickness, 25 m × 0.31 mm i.d.) connected to a flame ionization detector was used, employing helium as carrier gas. The GC oven was programmed from -10 °C (2 min isothermal) to 320 °C at 8 °C/min. Prominent hydrocarbon peaks (especially aromatics) were identified with reference to published relative retention times.27,28 Quantification was performed using nbutane as an external standard. Utilizing thermally extracted rock meant that macromolecular components from the bitumen were pyrolyzed in the above procedure along with the major component, kerogen. Artificial maturation experiments were performed on aliquots of one kerogen concentrate (TOC ) 62.1%; Hydrogen Index 540 mg/g TOC) that had been isolated from an immature Bakken Shale sample (0.49% Ro). Using a convection oven, the following isothermal heating conditions were employed: 300 °C (2 and 5 days), 330 °C (1, 2, and 5 days), and 350 °C (1, 2, and 5 days) using the microscale sealed vessel (MSSV) approach of Horsfield et al.29 Both the artificially generated products as well as the residual kerogen from each artificial maturation level were analyzed via thermovaporization- and pyrolysis-gas chromatography, respectively. GC conditions for analysis of products as well as for residual kerogen were identical as for PY-GC. Total yields were also determined rapidly using a pseudo-Rock-Eval-type analytical setup in which the GC column was replaced by a 1/8 in. stainless steel transfer line directly connected to the FID. S1 and S2 yields were calculated.
Maturation Characteristics of the Natural System Bulk Composition. Figure 2 outlines the evolution of macromolecular organic matter using the Rock-Eval Hydrogen Index. Because kerogen quality (type of organic matter) can be considered uniform, the trend provides a clear measure of how petroleum generation of the Bakken Shale progresses. The trend can be discriminated into three zones: Up to ca. 0.6% Ro, there is only a little decrease in values. Between 0.6 and 0.8% Ro, however, the Hydrogen Index drops rapidly from ca. 500 to 200 mg/g TOC. Beyond 0.8% Ro, the mean HI value remains uniformly low (ca. 200 mg/g TOC) despite (27) Horsfield, B.; Du¨ppenbecker, S. J. J. Anal. Appl. Pyrol. 1991, 20, 107-123. (28) Requejo, A. G.; Allan, J.; Creaney, S.; Gray, N. R.; Cole, K. S. Org. Geochem. 1992, 19 (1-3), 245-264. (29) Horsfield, B.; Disko, U.; Leistner, F. Geol. Rund. 1989, 78, 361374.
Muscio and Horsfield
Figure 3. Evolution of Bakken Shale organic richness (TOC content) as a function of maturity for 129 samples from 12 wells. Trend of average TOC is indicated by dashed line.
Figure 4. Distribution of TOC vs depth for selected Bakken cores from four different stages of thermal evolution.
the fact that Ro suggests maturation is still ongoing. The trend mainly reflects changes in kerogen composition, though, because analyses were conducted using whole rock samples. Hydrogen Indices are to some degree influenced by the presence of macromolecular bitumen components. This amounts to 10-15% for the Ro range 0.9%, according to extract yields and hydrocarbon/non-hydrocarbon ratios23 and using known responses.30 For the sake of brevity, the term kerogen will be used hereafter though the influences of polar components cannot be neglected. Figure 3 shows the fall in TOC associated with increasing maturity. Although the datapoints outline a relatively broad scatter especially at low levels of maturation (0.9% Ro, PGI remains essentially uniform (between 0.8 and 0.9). The trend is similar to those of other marine Type II source rocks, as shown. The datapoints of Figure 5 were approximated by a spline fit, which was then differentiated to give a rate curve. Both curves are given in Figure 6. The spline curve clearly elucidates that ca. 50% of the reactive kerogen of the initial organic matter (at 0.31% Ro) is converted within a relatively narrow interval of maturity (from 0.6 to 0.8% Ro). This is in good agree-
Energy & Fuels, Vol. 10, No. 1, 1996 13
Figure 7. Calculated initial (immature) TOC content for Bakken Shale samples from 10 wells which are in a mature stage at present time. Calculations (average values) are based on the approach by Cooles et al. (1986) and refer to well A (0.31% Ro).
ment with kinetic predictions of Burnham.31 The differentiated curve more clearly shows that the maximum rate of hydrocarbon generation occurs at ca. 0.62% Ro. Thereafter, the petroleum formation rate drops to minimum amounts at ca. 0.85% Ro. The slight bulge of the curve at 1.1% Ro can be considered an artifact of the spline approximation and is therefore negligible. Figure 7 depicts the initial average TOC content for 10 wells of successively increasing maturity level (% Ro) as calculated using the Cooles model. For the samples which are at present at a mature stage of catagenesis, the values represent the samples’ organic richness at an immature, initial stage of petroleum generation. The actual measured mean value (well A, 0.31% Ro) is shown for comparison. All calculated values range between 19 and 22 wt % TOC and are proportionally higher than the mean value of the reference sample (17 wt %) by between 7 and 22%. There are three possible explanations for these observations: 1. Variations in Depositional Conditions. It has been stated that the depositional environment was uniform in the Williston Basin during Bakken Shale sedimentation.19,20,32,33 Nevertheless, the distribution of TOC in Figure 7 could signify that depositional conditions for all mature Bakken Shale samples were different from the conditions for the reference sample, with higher concentrations of organic carbon being deposited. Microscopical examination of the Bakken Shale revealed that the two predominant macerals are bituminite and alginite (estimated proportions are 25-30% and 10-15%, respectively). While these macerals were found for both shale units and from all depths and locations in the basin, they occur in a horizontally orientated kerogen network at shallow depth and with a more homogeneous texture with no preferred orientation in deeper parts of the basin.23 It cannot, therefore, be unequivocally ruled out that there are indeed at least two types of organofacies represented by the present sample set and hence two different levels of organic richness. 2. Unrepresentative Sampling. In view of the variability in TOC within all wells (Figure 4), the question (31) Burnham, A. K 1992 LLNL Report UCRL. ID-109622. (32) Price, L. C.; Ging, T.; Daws, T., Love, A.; Pawlewicz, M.; Anders, D. Hydrocarbon Source Rocks of the Greater Rocky Mountain Region 1984, 83-133. (33) Hayes, M. D.; Holland, F. D. Bull. Am. Assoc. Pet. Geol. 1983, 67, 1341-1342.
14
Energy & Fuels, Vol. 10, No. 1, 1996
Figure 8. Fingerprints as derived from open-system PY-GC of two whole rock Bakken Shale samples as representative for low-mature and high-mature stages of thermal alteration. Numbers indicate total numbers of carbon atoms. BZ, T, MPX, and TMB refer to benzene, toluene, m-/p-xylene, and 1,2,3,4tetramethylbenzene, respectively.
arises as to whether the mean TOC value determined for the reference sample is actually representative of the entire core. Nonetheless, available data (Figure 3) indicates that the choice was realistic. 3. Progressive Formation of Inert Carbon. Importantly, the discrepancy between calculated and actual TOC becomes more acute with increasing maturity level, signifying a causal relationship. If inert carbon formation were enhanced, the volatile products resulting from such a disproportionation process could be expected to be enriched in compounds with high H/C ratios. Indeed, Muscio et al.34 have elucidated that the thermally mobilized organic components are dominated by low molecular weight hydrocarbons. This is also indicated by an overall high API gravity (40-45°) for Bakken crude oils.35 The difference between predicted and measured TOC values is best explained by the third alternative above because predicted initial TOC values increase as a function of maturity, but this answer is not unequivocal. Pyrolysis-gas chromatography and, later, MSSV pyrolysis experiments were therefore performed to see if molecular considerations could provide an insight into inert carbon formation. Molecular Considerations. The pyrograms of Figure 8 document the changes in bulk kerogen composition on a molecular level with increasing maturity. Based on a previous comparison of 13C NMR and PYGC, it could be assumed that, although absolute concentrations of individual components are low, relative abundances could be considered proportionally representative of the kerogen as a whole, or of its reactive component.36 At early levels of maturation (350 °C/1-2 days
tween the artificial maturation levels 350 °C/1 day and 350 °C/2 days. Clearly, the significance of this parameter is not unambiguous, as its assessment in the natural system involves data from the products which may have been perturbed by expulsion and migration processes. Nonetheless, it may be justified to relate the artificial maturity zone ranging from 350 °C/1 day to 350 °C/2 days to a stage of natural thermal evolution of ca. 0.8% Ro. Hence, by using the aromaticity parameter for low levels of maturation (deflection in trend for both systems at 0.6% Ro and 300 °C/2-5 days, respectively) and the product/residue parameter for higher stages of maturation (cross-over in both systems at 0.8% Ro and 350 °C/1-2 days), it was possible to define the approximate calibration scheme shown in Table 1 (maturity zonation 1, 2, and 3). Before using this framework for quantitative prediction of products, its validity is tested against the evolution of the natural residual kerogen (PY-GC). Figure 15 displays the corresponding diagrams for total yield of pyrolysate and its subfractions (C1-5, and C6-14, and C15+). The range of equivalent values of artificially heated kerogens (MSSV) are superimposed for the maturity zones 1, 2, and 3. As regards the total yield of pyrolysate, there is indeed a rather good match between absolute yields for the natural and artificial series suggesting that the inferred boundaries for the maturity zones are correct and the approach is valid. It must be stated that this good agreement is less pronounced for the gaseous hydrocarbons (C1-5) whereby the naturally matured kerogens consist of more gasyielding moieties than the artificially heated ones,
Figure 16. Yield of total volatilisable products (C1+ and subfractions C1-5, C6-14, and C15+) of natural Bakken Shale maturity sequence (thermovaporization-GC) as a function of thermal evolution (% Ro). Shaded areas denote range of equivalent data derived from analysis of artificially matured Bakken kerogen. Approximate boundaries for maturity zones are determined using maturity sensitive parameters of Figure 14. Thermovaporization data was taken from Muscio et al. (1994).
especially for the first two maturity zones. This is thought to be because the sample chosen for the experiments turned out to have a lower than typical content of gas precursors. Interestingly, the general maturity trend of the natural system is nonetheless traced by the laboratory system as far as the C1-C5 components are concerned; it is just the absolute yields which differ. The medium and heavy pyrolysates reveal a rather good to moderate congruency for the natural and laboratory systems. We conclude that the MSSV method provides an adequate simulation of the natural system for the bulk quantities shown. This is important because further MSSV experiments described later in the article were used to quantify the formation of inert carbon as a function of transformation.
Formation of Inert Carbon during the Natural Maturation
In contrast to the evolutionary yields of the residues, there are considerable differences in the concentration of products between the natural and the lab system, especially in the high mature zones (Figure 16): As maturation proceeds in both systems, evolutionary trends follow different pathways: In the closed system, yields drastically rise, whereas in nature the average concentration remains virtually constant with a high variability of data. Interestingly, the “end member” of the simulated maturity sequence exhibits a yield which is almost 3 times as high as the highest concentration of products in the natural sequence (0.99% Ro). The pattern is similar for the subfractions: Both trends diverge significantly from 0.6% Ro on. The latter feature is even more pronounced for the gases (C1-5) as their yield is submitted to a sharp drop in the second maturity zone (0.6-0.8% Ro). The distribution pattern is almost identical for the medium (C6-14) and high molecular weight (C15+) material. Here, natural Bakken Shale products are characterized by uniformly high yields throughout maturation, while yield in the lab system increases continuously. The maturity zone during which natural and artificial maturity trends diverge is of course coincident with the main phase of hydrocarbon generation in the natural system. This serves as a further indicator that the zonation approach is valid. Hence, it may be justified to calculate the amount of generated products which actually have been removed from the source rock system by subtracting the concentration of naturally present hydrocarbons from the amount of artificially formed ones. Clearly, such a relative mass balancing approach can only be carried out for each maturity zone. Furthermore, for each zone one has to select that sample of the natural system which revealed the highest yield of products. This sample is considered as being the least depleted equivalent of the respective natural maturity zone. The correspondence value is balanced against the lowest value of the corresponding artificial maturity sequence. Choosing the lowest value for the closed system sample is of particular importance for high levels of simulated maturation: There is some evidence that the products derived from the most severe stages of artificial maturation have been modified by secondary cracking reactions in that oillike components are cracked to gaseous compounds (Figure 17). The drastic increase in gas yield (30 times as much than at the starting level) paralleled by a more smooth increase of “oil” (5 times as much C6+ material than at the starting level) may also illustrate that effect. In that way, one obtains the minimum difference between the natural and artificial system for each maturity zone. Analogous to the approach of Mackenzie et al.,43 the following equation can be used in order to determine the minimum extent of removal (abbreviated in the following as “EXR”):
(closed system)min - (natural system)max (closed system)min
× 100
The results are listed in Table 2. For the total yield of products, the immature zone (zone 1) of the natural system is enriched in products relative to the closed system. The other two zones which are associated with (43) Mackenzie, A. S.; Leythaeuser, D.; Schaefer, R. G.; BjorØy, M. Nature 1995, 301, 506-509.
Energy & Fuels, Vol. 10, No. 1, 1996 17
Figure 17. Relationship of gaseous (C1-5) to oillike (C6+) compounds of the artificial maturity sequence (MSSV) for products and residues. TVAP refers to unheated sample. Table 2. Results from Mass Balancing of the Natural and Artificial System Using Maturity Zonation Approach for Bulk Compound Groupsa fraction C1+ C1-5 C6-14 C15+
zone
CSmin
NSmax
EXR (%)
zone 1 zone 2 zone 3 zone 1 zone 2 zone 3 zone 1 zone 2 zone 3 zone 1 zone 2 zone 3
49.6 82.9 172.8 2.3 7.5 29.3 8.9 23.5 70.5 38.4 51.9 73.0
83.0 55.0 94.1 6.9 6.0 3.3 47.0 31.3 50.3 30.7 19.8 43.2
-67.3 33.7 45.5 -200.0 20.0 88.7 -428.1 -33.2 28.7 20.1 61.8 40.8
a CS min refers to the minimum yield of the respective compound group of the artificial (closed) systems, NSmax to the maximum yield of the respective compound group of the natural system. EXR is calculated according to the equation as displayed in the text.
comprehensive generation of hydrocarbons in nature indeed reflect a difference in masses between both systems (33.7% for zone 2 and 42.3% for zone 3). These values are significantly lower than Cooles Model predictions. This pattern is similar for the subfractions. Zone 1 is characterized by an enrichment of naturally generated products relative to the artificial maturation sequence. This is most pronounced for the medium boiling range (C6-14) which contains high abundances of cycloalkanes and aromatic hydrocarbons.34 A noteworthy feature can be observed for the gaseous products: Here, the high mature zone (zone 3) exhibits the highest value for EXR, indicating that discrepancy between the two systems is most enhanced for gaseous compounds at high levels of maturity. This phenomenon may be explained by differences in formation mechanisms for gases as a function of maturity level. As illustrated in Figure 17, it is very likely that the yield of gaseous products in the lab system at high artificial maturation stages (>350 °C/1 day including zone 3) has a significant contribution from “oil” (C6+) to gas cracking. It must be pointed out, however, that it is difficult to determine which would be the equivalent natural maturity stage of the end member of the artificial maturity sequence, as there is no suitable parameter for this. Therefore, in the course of closed system
18
Energy & Fuels, Vol. 10, No. 1, 1996
Figure 18. A mass balance of products and residues as derived from closed-system MSSV-type artificial maturation (cf. Horsfield et al., 1992) implies that inert kerogen is progressively being formed during thermal evolution under laboratory conditions. RT ) original sample.
maturation in the lab the stage of metagenetic gas generation indeed has been entered in contrast to the natural system. The phenomenon that the yield of C1-5 and C6-14 material is higher in zone 1 of the natural system is surprising. In case of the gaseous subfraction, this may be explained by the retention capability of the Bakken Shale for gas. Muscio et al.34 discovered that coarsely crushed samples contained high concentrations of gas while powdered ones yielded lower amounts. For MSSV experiments, powdered samples were utilized. Therefore, the gas yields in the closed system may have been affected by sample treatment in that the amount of gas initially present prior to the heating procedure was less than in the samples which were submitted to natural thermal evolution. Alternatively, this mismatch might indicate that the natural gas encountered in the low mature zone (zone 1) is not a result of thermal degradation of the host source rock but was derived from another maturity zone by lateral intraformational migration processes. This scenario has already been called upon based on compound specific isotope analysis.34 However, such a hypothesis is only valid with the prerequisite that the maturity zonation approach between natural and lab system indeed is applicable and that both systems are comparable in terms of identical reaction pathways. Nonetheless, the results as deduced from quantitative comparison of artificial and natural thermal evolution imply that intraformational migration may have contributed to the unusually enhanced concentrations of gaseous hydrocarbons in the immature zone of the Bakken Shale. Bulk Composition of Artificially Matured Residues and Products. Additional artificial maturation experiments were conducted using the immature Bakken Shale kerogen concentrate described above. In this case, both volatile products (dS1) and pyrolyzable residues (dS2) were measured in bulk-flow mode,44 that is, without chromatographic separation of complex mixtures, using flame ionization detection (Figure 18). The mass balance calculation reveals that with increasing thermal stress S1 is generated in increasing amounts at the expense of S2. Under the pyrolysis conditions applied, transformation ratios (TR ) S1/(S1 (44) Du¨ppenbecker, S. J.; Horsfield B. In Advances in Organic Geochemistry 1989; Durand B., Be´har F., Eds.; Pergamon: Oxford, U.K., 1990.
Muscio and Horsfield
+ S2) from 0.02 to 0.63 were induced. Importantly, the sum of S1 and S2 was noted to decrease. For the most severe level of artificial maturation (350 °C/5 days), up to 22% of the reactive kerogen originally available for conversion to volatile hydrocarbons (according to open system pyrolysis of the original sample) is actually converted to inert products. It should be noted that a similar progressive evolution was also encountered for the MSSV pyrolysis of Alum Shale kerogen, where 33% of the original S2 was converted into inert components at a TR of 0.76 (Horsfield et al., 1992). By way of contrast, isolated kerogen from immature Posidonia Shale (Toarcian, Germany; Ro ) 0.48%; kerogen Type II) generate secondary inert carbon only at very high levels of thermal stress (TR > 0.9) using programmed temperature MSSV pyrolysis (unpublished results). Lewan12 has also documented this phenomenon during hydrous pyrolysis of other organic matter-rich shales. These experimental results corroborate inferences from the natural system that inert carbon formation is quantitatively significant in the case of Bakken Shale kerogen. Conclusions The natural and artificial maturation of Bakken Shale source rock samples provides evidence that generative yields from open system pyrolysis are not equal to potential petroleum yields in nature. This is because inert kerogen formation is enhanced under both natural and closed system simulation conditions. The outcome is that mass balance models normalized to inert carbon give overestimates of petroleum generated in nature. This is the second case where we have been able to document this phenomenon, the first being the Alum Shale of Scandinavia. Immature Alum Shale kerogen is aliphatic but because it is cross-linked/alicyclic, aromatizes, and condenses during closed system pyrolysis and natural maturation alike. Immature Bakken Shale kerogen is hydrogen-rich, yet generates large proportions of aromatics on pyrolysis. In this case, a significant proportion are derived from primary aromatic structures in the kerogen. An abundance of aromatic nuclei or readily aromatizable structures appears to be a common denominator. MSSV pyrolysis was shown to provide a satisfactory simulation of maturation as far as boiling ranges were concerned. Compositional features which are common to both the natural and artificial maturation sequence could be used to construct zonations. While this maturity zonation approach could be used to predict the quantitative and qualitative evolution of natural products and the degree of petroleum expulsion, the occurrence of enhanced concentrations of low molecular weight hydrocarbons in the immature zone remains unexplained. Acknowledgment. This work was performed as part of G.M.’s Ph.D at KFA Ju¨lich with the support of Prof. D. H. Welte. The authors express their gratitude to Conoco Inc. for sponsorship, and to Eric Michael for advice. The North Dakota Geological Survey, and Julie Lefever in particular, are thanked for their invaluable assistance in providing samples and offering advice. Technical assistance provided by Franz Leistner and Elmy Biermanns is gratefully acknowledged. Reviews by Alan Burnham, Raymond Michels, and Eric Michael helped to improve the quality of the manuscriptsthank you. EF950189D
