Petroleum Expulsion Part 2. Organic Matter Type and Maturity Effects

Jan 5, 2006 - Energy Fuels , 2006, 20 (1), pp 301–308 ... parameters (solubility parameter, cross-link density, and the volume fraction of the kerog...
5 downloads 10 Views 530KB Size
Energy & Fuels 2006, 20, 301-308

301

Petroleum Expulsion Part 2. Organic Matter Type and Maturity Effects on Kerogen Swelling by Solvents and Thermodynamic Parameters for Kerogen from Regular Solution Theory S. R. Kelemen,* C. C. Walters, D. Ertas, and L. M. Kwiatek ExxonMobil Research and Engineering Company, Annandale, New Jersey 08801

D. J. Curry ExxonMobil Exploration Company, Houston, Texas 77252 ReceiVed June 7, 2005. ReVised Manuscript ReceiVed October 26, 2005

The swelling of kerogen in several well-defined solvents was measured to derive thermodynamic parameters specified by an extended Flory-Rehner and Regular Solution Theory framework. These parameters are needed to assess the importance of the selective solubility of individual petroleum compounds into kerogen as a mechanism for the chemical fractionation that occurs during expulsion. The swelling of kerogen by a set of organic solvents was quantified following exposure to 30, 90, and 150 °C temperatures for 24 h in a closed system. The organic solvents were selected to reflect the prevalent hydrocarbon and heteroatom structures in petroleum. A total of 13 different, well-characterized Type II and Type IIIC kerogens were studied that included a maturity suite for each kerogen type. The swelling results were pooled and analyzed using an analysis of variance (ANOVA) statistical method to determine if swelling differences were related to the temperature, solvent, kerogen type, and level of kerogen maturity. There was no statistical difference in the mean volumetric swelling ratio (Qv) at 30 and 90 °C (2.1%) and only a slight increase in the mean Qv at 150 °C (3.0%) after exposing kerogen for 24 h to the solvents used in this study. The weak temperature dependence indicates that these kerogen-solvent systems have nearly equilibrated after 24 h at 30 °C. Statistically significant differences in the mean Qv were found among different kerogens and nonpolar/slightly polar solvents with varying solubility parameter (δ) values. Average swelling data for Type II and Type IIIC kerogens were interpreted to derive thermodynamic parameters (solubility parameter, cross-link density, and the volume fraction of the kerogen network Veq that minimizes its elastic strain energy, termed “native swelling”) by optimizing the match between the experimental data and theoretical predictions. These parameters enable general predictions of swelling behavior of these kerogen types in other solvents and complex solvent mixtures. To verify the model, the swelling of kerogen by 50:50 n-hexadecane/solvent mixtures was calculated and compared to experimental data. Good agreement was observed between the Qv and compositionally based swelling ratio (Qc) for all kerogens.

1. Introduction Expulsion is the initial, poorly understood stage of petroleum migration whereby hydrocarbons are released preferentially from generative organic-rich source rocks into more permeable strata. Chemical fractionation occurs during expulsion such that the composition of expelled petroleum is highly enriched in volatile and saturated hydrocarbons while the retained soluble bitumen is enriched in polar heteroatomic species. The degree and manner of this fractionation appears to vary with organic matter type, source rock richness, and the degree of thermal maturation that influences the chemical nature and quantity of generated petroleum and solid residue.1-3 Numerous theories have been proposed to account for chemical fractionation during expulsion.2-4 These concepts tend to emphasize either movement of hydro* Corresponding author. Phone: (908) 730-2389. Fax: (908) 730-3232. (1) Tissot, B. P.; Welte, D. H. Petroleum Formation and Occurrence, 2nd ed.; Springer-Verlag: Berlin, 1984. (2) Pepper, A. S. In Petroleum Migration; England, W. A., Fleet, A. J., Eds.; Geological Society Special Publication 59; Geological Society: London, 1991; p 9. (3) Lafargue, E.; Espitalie, J.; Broks, T. M.; Nyland, B. Org. Geochem. 1994, 22, 575.

carbons within fine-grained rocks or the release of hydrocarbons from the kerogen matrix. Other mechanisms further contribute to fractionation as petroleum moves from the source to reservoir rocks (secondary migration). Multicomponent equilibrium likely exists between solid and liquid organic matter during generation and expulsion. If the chemical fractionation during primary migration is governed by this equilibrium, a theory may be developed that could accurately model the composition of expelled petroleum and retained bitumen. Sandvik et al.5 emphasized that the compositional fractionation observed for expulsion in natural systems was directionally consistent with documented interactions between solvents and either coals or synthetic polymers.6 Since diffusion effects are modeled to expel fluids preferentially with the opposite compositional fractionation seen in nature (i.e., aromatics > naphthenes > alkanes), (4) Mann, U.; Hantschel, T.; Schaefer, R. G.; Krooss, B.; Leythaeuser, D.; Littke, R.; Sachsenhofer, R. F. In Petroleum and Basin EVolution; Welte, D. H., Horsfeld, B., Baker, D. R., Eds.; Springer: Heidelberg, 1997; pp 397-515. (5) Sandvik, E. I.; Young, W. A.; Curry, D. J. Org. Geochem. 1992, 19, 77. (6) Sandvik, E. I.; Mercer, J. N. Org. Geochem. 1990, 16, 83.

10.1021/ef0580220 CCC: $33.50 © 2006 American Chemical Society Published on Web 01/05/2006

302 Energy & Fuels, Vol. 20, No. 1, 2006

they concluded that kerogen-fluid phase partitioning is more important that diffusivity.7 More recent work in this area concluded that this mechanism may account for the directionality but not the magnitude of the chemical fractionation observed in natural systems.8 These conclusions, however, were based on limited published data. Additional experiments and modeling are needed to quantitatively assess the magnitude of selective solubility of liquids in kerogen to access fully its importance in chemical fractionation during expulsion. Solvents interact with kerogen in a manner comparable to cross-linked organic polymers.5,9-12 The swelling of polymers by solvents of varying molar volume and solubility parameter is a well-known technique for characterizing the physical structure of the polymer network and defining its chemical nature.13-15 The behavior of solvents that have no specific interactions with the chemical structure of polymer can be treated within the Regular Solution Theory framework for which the theoretical basis has been outlined in detail.13-20 An extended Flory-Rehner and Regular Solution Theory has been developed for modeling the swelling of kerogen by solvents and solvent mixtures.21 Quantitative information concerning the swelling of kerogen and coal in pure solvents and solvent mixtures is limited.5,9-12,21,22 Observations indicate that solvent swelling of amorphous kerogens occurs via nonspecific interactions that can be modeled within the context of a Regular Solution Theory framework. In contrast, the interaction of some solvents with coal and chars does not follow Regular Solution Theory due to a predominance of specific interactions.11 Accurate modeling of kerogen interactions in laboratory experiments using well-defined solvents and conditions is needed prior to extending the theory to the more complex natural systems that occur during primary migration. By analyzing these results within a theoretical framework,21 thermodynamic parameters (solubility parameter, cross-link density, and native swelling) can be determined for average Type II and Type IIIC kerogens by optimizing the match between predicted and experimental data. The solubility parameter (δ) is a numerical value that indicates the relative solvency behavior of a specific solvent. The cross-link density (νe) of the network of organic matter of kerogen reflects the sum of all bondbreaking and bond-making reactions that have taken place during maturation. Native swelling is the volume fraction of the kerogen network Veq that minimizes its elastic strain energy. These parameters are needed to model the multicomponent equilibrium for kerogen and complex mixtures representative of those (7) Thomas, M. M.; Clouse, J. A. Geochim. Cosmochim. Acta 1990, 54, 2793. (8) Ritter, U. J. Org. Geochem. 2003, 34, 319. (9) Larsen J. W.; Li, S. Energy Fuels 1994, 8, 932. (10) Larsen, J. W.; Li, S. Energy Fuels 1997, 11, 897. (11) Larsen, J. W.; Li, S. Org. Geochem. 1997, 26, 305. (12) Larsen, J. W.; Parikh, H.; Michels, R. Org. Geochem. 2002, 33, 1143. (13) Flory, P. J. Principles of Polymer Chemistry; Cornell University Press: Ithaca, NY, 1953. (14) Hilderbrand, J. H.; Scott, R. L. The Solubility of Nonelectrolytes; Dover: New York, 1964. (15) Hilderbrand, J. H.; Praudnitz, J. M.; Scott, D. L. Regular and Related Solutions; Van Nostrand Reinhold: Princeton, NJ, 1970. (16) Barton A. F. M. CRC Handbook of Solubility Parameters and Other Cohesion Parameters; CRC Press: Boca Raton, FL, 1983; p 48. (17) Brandrup, J.; Immergut, E. H. Polymer Handbook, 3rd ed.; Wiley & Sons: New York, 1989. (18) Flory, P. J.; Rehner, J., Jr. J. Chem. Phys. 1943, 11, 521. (19) Hoy, K. L. J. Paint Technol. 1970, 42, 76. (20) Wiehe, I. A. Fuel Sci. Technol. Int. 1996, 14, 289. (21) Ertas, D.; Kelemen, S. R.; Halsey, T. C. Energy Fuels 2006, 20, 295. (22) Ballice, L.; Larsen, J. W. Fuel 2003, 82, 1317.

Kelemen et al. Table 1. Solubility Parameter and Molar Volumes of the Solvents Used in Kerogen Swelling Studies

solvent

solubility parameter (J/cm3)1/2

n-decane n-hexadecane cyclohexane decalin toluene tetralin 1-methylnaphthalene 2-5 dimethylpyrrole pyridine benzofuran benzothiophene

15.8 16.3 16.8 17.7 18.2 19.4 20.2 20.3 21.9 21.1 21.8

ref

molar volume (cm3)

ref

16 16 16 23 16 17 17 23 17 23 23

195.9 294.1 108.7 154.2 106.9 136.3 139.4 101.7 80.6 108.3 124.7

16 16 16 24 16 24 24 24 24 24 24

thermally generated by kerogen. In this way, the importance of this chemical fractionation mechanism in primary migration from source rocks under geologic conditions can be assessed. 2. Experimental Section 2.1. Solvents. Organic solvents were selected to cover a wide range of solubility parameter values and reflect the prevalent hydrocarbon and heteroatom structures in petroleum. Table 1 lists the solubility parameters and molar volumes of these solvents at temperatures near 25 °C that were obtained from reference tables16,17 or calculated from estimates of group molar cohesive energy23 and molar volume.24 Several alkyl, naphthenic, and aromatic solvents represent the petroleum hydrocarbon fraction. Almost all nitrogen in kerogen and its thermal decomposition products exist in pyrrolic and pyridinic nitrogen forms. These nitrogen species are represented by the solvents pyridine and 2,5 dimethylpyrrole. Single-bond carbon-oxygen species are among the most abundant oxygen forms in mature organic matter and are represented by benzofuran. Thiophenic sulfur, the most abundant aromatic form of sulfur in kerogen and its thermal decomposition products, is represented by benzothiophene. 2.2. Kerogen. A suite of kerogens was selected for the swelling experiments from marine shales (Type II) and hydrogen-rich coaly shales. These hydrogen-rich coals and equivalent disseminated organic matter are chemically distinct from gas-prone vitrinitic coals as these kerogens are the source of major accumulations of hydrocarbon-rich oil.25 The kerogens contain a variety of macerals (sporinite, resinite, and liptinite, including alginite, cutinite, and suberinite) that contain a significant proportion of aliphatic moieties. We have designated such kerogens as “Type IIIC”.26 Rock-Eval Tmax and hydrogen index (HI) data for the kerogens are listed in Table 2 along with the weight percent carbon and ash. Note that the Type II (D) and Type IIIC (H) kerogens represent maturation suites that span the range from early to late catagenesis as indicated by the Tmax values. Type IIIC samples were not demineralized. Type II kerogens were isolated from the rock matrix by mineral dissolution with HCl/HF according to the standard procedure described by Durand and Nicaise.27 It is recognized that this procedure does not remove pyrite and other minerals protected by encapsulation by organic matter. However, Table 2 shows that carbon content of the Type II samples has been increased from a relatively low level (several weight percent) to levels found in the Type IIIC samples. Table 3 contains compositional analysis. The results are reported relative to carbon. The organic nitrogen and sulfur data were obtained from X-ray photoelectron spectroscopy (XPS) based on the area of the characteristic organic photoelectron (23) Fedors, R. F. J. Polym. Sci. C 1974, 14, 147 and 472. (24) CRC Handbook of Chemistry and Physics, 51st ed.; CRC Press: Boca Raton, FL, 1970. (25) Wilkins, R, W. T.; George, S. C. Int. J. Coal Geol. 2002, 50, 317. (26) Curry, D. J.; Emmett, J. K.; Hunt, J. W. In Coal and coal-bearing strata as oil-prone source rock?; Scott, A. C., Fleet, A. J., Eds.; Geological Society Special Publication 77; Geological Society: London, 1994; p 149. (27) Durand, B.; Nicaise, G. In Kerogen; Durand, B., Ed.; Editions Technip: Paris, 1980; pp 35-53.

Organic Matter and Maturity Effects on Kerogen Swelling

Energy & Fuels, Vol. 20, No. 1, 2006 303

Table 2. Rock-Eval Data and Weight % Carbon and Ash for the Kerogen Used in the Swelling Studies kerogen type (name)

Rock-Eval Tmax (°C)

hydrogen index (mg/g OC)

Wt % carbon

Wt % ash

Type II (A-1) Type II (B-1) Type II (B-2) Type II (C-1) Type II (D-1) Type II (D-2) Type II (D-3) Type II (D-4) Type IIIC (F-1) Type IIIC (G-2) Type IIIC (H-1) Type IIIC (H-2) Type IIIC (H-3)

438 424 nd 413 414 438 443 479 424 411 427 453 479

401 581 nd 577 532 439 242 22 237 251 295 235 120

57.0 49.9 50.6 59.2 63.4 56.1 55.6 42.1 61.3 67.4 61.3 82.8 73.2

30.8 38.8 37.8 17.5 17.7 31.2 34.3 51.2 24.5 8.4 24.0 3.2 18.0

Table 3. Compositional Data for the Kerogen Used in the Swelling Studies per 100 carbon elemental

XPS

kerogen type (name)

hydrogen

organic nitrogen

organic sulfur

organic oxygen

Type II (A-1) Type II (B-1) Type II (B-2) Type II (C-1) Type II (D-1) Type II (D-2) Type II (D-3) Type II (D-4) Type IIIC (F-1) Type IIIC (G-2) Type IIIC (H-1) Type IIIC (H-2) Type IIIC (H-3)

103.5 108.0 109.8 123.2 116.8 111.1 88.9 56.2 89.0 88.3 87.0 76.0 61.0

3.1 2.3 2.3 2.0 2.9 2.0 2.1 2.1 1.7 0.8 1.3 1.1 1.1

1.5 2.1 2.1 3.2 1.4 1.2 0.6 1.0 0.4 2.7 0.2 0.2 0.1

3.9 4.1 4.1 13.7 9.7 5.9 5.0 4.7 9.3 15.4 11.1 6.6 4.0

peaks after correcting for atomic sensitivity.28,29 The amount of organic oxygen was derived from the total XPS oxygen (1s) signal by taking into account inorganic contributions.30 For the Type II samples, the inorganic contributions to the XPS are particularly low owing to the surface sensitivity of XPS and the fact that residual mineral matter including pyrite is mostly encapsulated by organic matter. This situation favors XPS in direct organic compositional analysis for these kinds of samples. The effect of demineralization on the composition of the kerogen organic matter is found to be small. No incorporation of chlorine or fluorine was detected using XPS. Hydrolysis is expected to impact the most immature samples and have negligible impact on the more mature samples.27 All samples were subsequently Soxhlet extracted with toluene and then processed in a Wig-L-Bug to ensure small uniform particle size. 2.3. Volumetric Swelling Ratio. Samples were weighed into ∼3 cm-long NMR tubes (5 mm) and then centrifuged at 1000 rpm for ∼20 min before recording the initial dry sample height. Solvent was added to each individual tube, stirred, topped with a plug of glass wool, and placed in a upright position within a 100-mL Parr high-pressure reactor vessel, which held up to 28 tubes at one time. The tubes were covered with excess solvent, and the reactor was sealed, evacuated, and then pressurized with helium (100 kPa). The sequence of evacuation and pressurization with helium was repeated four times to ensure that the reactor was free of air, which was confirmed by GC analysis of the reactor headspace. The reactor was heated to 30, 90, or 150 °C for 24 h. After cooling, the reactor was opened and each tube again was centrifuged before recording the final height for each tube. Experiments using the three temperatures were run in succession. In general, two to three tubes (28) Kelemen, S. R.; Gorbaty, M. L.; Kwiatek, P. J. Energy Fuels 1994, 8, 896. (29) Kelemen, S. R.; George, G. N.; Gorbaty, M. L. Fuel 1990, 69, 939. (30) Kelemen, S. R.; Afeworki, M.; Gorbaty, M. L.; Cohen, A. D. Energy Fuels 2002, 16, 1450.

of the same kerogen were present in a given experiment and the replicate volumetric swelling data were averaged for each kerogen and for each set of experimental conditions. The ratio of the final volume of kerogen to the initial volume of kerogen is defined as the volumetric swelling ratio Qv. 2.4. Compositional-Based Swelling Ratio. Samples were weighed into ∼3 cm-long NMR tubes (5 mm) and then centrifuged at 1000 rpm for ∼20 min before we recorded the initial dry sample height. Binary 50:50 n-hexadecane/solvent mixtures were prepared and then calibrated for quantitative GC analysis. A binary mixture was added to the kerogen-containing NMR tube that then was sealed within a stainless steel tube using two Swagelok end caps. Two of these closed system tubes could be held in an upright position within the secondary containment vessel, a standard 100 mL Parr reactor vessel. The reactor was heated to 150 °C for 24 h, cooled, and then opened to remove the closed system tubes. The Swagelock fittings were opened, and the NMR tubes containing kerogen and solvent mixture were removed and centrifuged before recording the height of the kerogen. The remaining solvent mixture was sampled immediately and analyzed five times, with the results averaged to yield a composition. Care was taken to minimize the time that tubes were open. The preferential uptake of the solvent relative to n-hexadecane was calculated based on the compositional changes of the solvent mixture and mass balance considerations that were based on the amounts of kerogen, the density of the kerogen, and the amount of solvents used in the experiment. The volumetric swelling predicted from the compositional changes to the solvent mixture and mass balances is defined as the compositionally base swelling ratio (Qc). 2.5. Extended Regular Solution Theory Framework of Kerogen Swelling by Solvents. An extended Flory-Rehner and Regular Solution Theory framework for swelling of kerogen21 was used to interpret the swelling data of kerogen by different solvents and solvent mixtures. Briefly, the framework incorporates concepts from the Flory-Rehner theory of rubber elasticity18 and the Regular Solution Theory of Hildebrand.15 The main hypotheses underlying the applicability of this model to kerogen absorption, in order of importance, are: (1) the system remains in thermodynamic phase equilibrium, (2) the (swollen) kerogen behaves as an elastomer network (i.e., rubber, not glassy), and (3) solvent-solvent and solvent-kerogen interactions are largely nonspecific, so that mixing rules of Regular Solution Theory are applicable. In this treatment, the characteristic chemistry of the kerogen network is simplified to a single effective solubility parameter δ0, and the physical structure of the network is characterized by the cross-linking density (νe) and the volume fraction of the kerogen network Veq that minimizes its elastic strain energy, termed “native swelling”.21 These parameters are determined for a given kerogen by obtaining a best fit between observed and computed values of swelling.21 The quality of the fit is measured by the correlation index R2, defined in terms of measured (exp) and predicted (th) swelling ratios of solvents {i} as

∑(Q R2 ≡ 1 -

(exp) vi

2 - Q(th) vi )

i

∑(Q i

(exp) 2 vi )

1 - ( n

∑Q

(exp) 2 vi )

i

3. Results 3.1. Swelling Temperature. The swelling of Type II and Type IIIC kerogens was quantified following exposure to various solvents at 30, 90, and 150 °C in a closed system. The results were pooled and analyzed using the analysis of variance (ANOVA) method coded in Statview statistical software (Table 4). No statistical differences are observed in the mean volumetric swelling ratio (Qv) at 30 and 90 °C following 24-h exposure to the same set of solvents (Figure 1). The mean Qv at 150 °C was only slightly greater than that at 30 and 90 °C (3.0 and

304 Energy & Fuels, Vol. 20, No. 1, 2006

Kelemen et al.

Figure 1. Comparison of the volumetric swelling ratio (Qv) after exposure to different solvents for 24 h at 30 and 90 °C. Table 4. Mean Volumetric Swelling Ratios Qv for Neat Solvents at 30, 90, and 150 °C temp (°C)

Qv mean

ANOVA ( 95%

data points

30 90 150

1.267 1.278 1.305

0.019 0.019 0.020

153 153 148

Table 5. Mean Volumetric Swelling Ratio Qv for Type II Kerogena kerogen type (name)

mean swelling ratio (Qv)

ANOVA ( 95%

data points

Type II (A-1) Type II (B-1) Type II (B-2) Type II (C-1) Type II (D-1) Type II (D-2) Type II (D-3) Type II (D-4)

1.386 1.364 1.271 1.421 1.364 1.387 1.298 1.152

0.033 0.034 0.033 0.033 0.024 0.033 0.034 0.033

33 32 33 33 64 33 32 33

a Data from swelling experiments conducted at 30, 90, and 150 °C pooled for each Type II kerogen.

Table 6. Mean Volumetric Swelling Ratio (Qv) for Different Solvents for Type II Kerogena solvent name

mean swelling ratio (Qv)

ANOVA ( 95%

data points

n-decane n-hexadecane cyclohexane decalin toluene tetralin 1-methylnaphthylene 2,5 dimethylpyrrole pyridine benzofuran benzothiophene

1.216 1.204 1.278 1.250 1.358 1.318 1.400 1.439 1.416 1.375 1.371

0.036 0.036 0.036 0.036 0.036 0.036 0.036 0.036 0.040 0.036 0.036

27 27 27 27 27 27 27 27 23 27 27

a

Data from all Type II kerogen pooled for each solvent.

2.1%, respectively). The weak temperature dependence of these data indicates that these kerogen-solvent systems have nearly equilibrated after 24 h at 30 °C. 3.2. Analysis of Type II Kerogen. The Qv values derived from ANOVA analysis for data pooled from experiments conducted at 30, 90, and 150 °C exhibit statistically significant differences among the Type II kerogens (Table 5) and the solvents (Table 6). Most notable are the differences among the Type II (D) kerogens that are related to sample maturity and discussed later. Slightly smaller, but still significant, differences are observed in the mean swelling ratio using different solvents. Higher solubility parameter solvents tend on average to swell kerogen to a greater extent (Figure 2).

Figure 2. Plot of the mean swelling ratio for Type II kerogen as a function of the solubility parameter of the solvent. Error bars represent 95% confidence intervals. Table 7. Thermodynamic Parameters for the Average Type II Kerogen and Type IIIC Coal from Optimum Fits of Experimental Swelling Data to Regular Solution Theory

kerogen type

solubility parameter (J/cm3)1/2

cross-link density (mol/cm3)

native swelling fraction

correlation index R2

average Type II average Type IIIC

22.5 23.3

0.16 0.25

0.76 0.85

0.923 0.962

Both the size (molar volume) and the solubility parameter of the solvent will influence its ability to swell a given kerogen. The extent of kerogen swelling for each solvent can be predicted within a Regular Solution Theory framework by knowing the solvent parameters and the thermodynamic parameters of kerogen. Conversely, it is possible to use results of swelling experiments to predict the most likely kerogen thermodynamic parameters by maximizing the correlation index R2. This procedure was applied to the mean swelling data for Type II kerogen (Table 6) to identify its most likely thermodynamic parameters. Table 7 shows the optimum thermodynamic parameters determined for the average Type II kerogen. Figure 3 shows three orthogonal two-dimensional sections of correlation index contours in the three-dimensional parameter space, centered around the optimum fit. This provides an estimate of the confidence level for these results. The average experimental swellings of Type II kerogen by different solvents were compared to the predicted swelling results using the kerogen parameters listed in Table 7. Figure 4 shows that the swelling behavior of all solvents can be accounted for reasonably well within a Regular Solution Theory model (R2 ) 0.92). 3.3. Type IIIC Kerogen. Statistically significant differences for the mean Qv swelling were found between the Type IIIC kerogens (Table 8) and solvents with varying solubility parameters (Table 9). In particular, the Type IIIC (H-1), (H-2), and (H-3) samples swell less than the other Type IIIC kerogens and will be discussed later in the context of maturity effects. As with the Type II kerogens, higher solubility parameter solvents tend on average to swell kerogen to a greater extent (Figure 5). One solvent, pyridine, results in an extreme increase in the volumetric swelling ratio. Pyridine is known to disrupt noncovalent bonding interactions in coal (i.e., hydrogen bonds, etc.) that act as secondary cross-links.11,31,32 The behavior of pyridine with Type IIIC kerogen indicates that this solvent likely interacts (31) Xiong, J.; Maciel, G. E. Energy Fuels 2002, 16, 497. (32) Wertz, D. L.; Smith, E. R. Energy Fuels 2003, 17, 482.

Organic Matter and Maturity Effects on Kerogen Swelling

Energy & Fuels, Vol. 20, No. 1, 2006 305

Figure 4. Comparison experimental results and predicted swelling of the average Type II kerogen for different solvents calculated using Regular Solution Theory and kerogen parameters listed in Table 6. Experimental error bars reflect 95% confidence intervals. Table 8. Mean Volumetric Swelling Ratio Qv for Type IIIC Kerogena kerogen type (name)

mean swelling ratio (Qv)

ANOVA ( 95%

data points

Type IIIC (F-1) Type IIIC (G-2) Type IIIC (H-1) Type IIIC (H-2) Type IIIC (H-3)

1.346 1.391 1.204 1.051 1.063

0.047 0.051 0.048 0.047 0.047

33 30 32 33 33

a Data from swelling experiments conducted at 30, 90, and 150 °C pooled for each Type IIIC kerogen.

Table 9. Mean Volumetric Swelling Ratio (Qv) for Different Solvents for Type IIIC Kerogena solvent name

mean swelling ratio (Qv)

ANOVA ( 95%

data points

n-decane n-hexadecane cyclohexane decalin toluene tetralin 1-methylnaphthalene 2,5 dimethylpyrrole pyridine benzofuran benzothiophene

1.081 1.105 1.174 1.131 1.175 1.173 1.222 1.261 1.581 1.213 1.226

0.070 0.070 0.079 0.070 0.070 0.070 0.070 0.070 0.073 0.070 0.070

15 15 12 15 15 15 15 15 14 15 15

a

Figure 3. Correlation index contours from attempts at fitting the average swelling data from Type II kerogen by varying the solubility parameter, cross-link density, and native swelling of kerogen. (A) Native swelling versus cross-link density, (B) native swelling versus solubility parameter, and (C) cross-link density versus solubility parameter.

with specific sites in a manner similar to its well-known behavior with coal. It is not possible to predict the swelling behavior of Type IIIC kerogen for pyridine within a Regular Solution Theory framework due to this specific interaction; therefore, the swelling data from pyridine were excluded in subsequent analyses for Type IIIC kerogen.

Data from all Type IIIC kerogen pooled for each solvent.

The mean swelling data shown in Table 9 were used as described in the previous section to determine the most likely thermodynamic parameters for the Type IIIC kerogen samples. Optimum thermodynamic parameters for the average Type IIIC kerogen are summarized and compared to those found for the average Type II kerogen in Table 7. Figure 6 shows three orthogonal two-dimensional sections of correlation index contours in the three-dimensional parameter space, centered around the optimum fit. As seen in Figure 7, the Regular Solution Theory framework can account for the swelling behavior of Type IIIC kerogens for all of the solvents, excluding pyridine (R2 ) 0.96). 3.4. Kerogen Maturation Effects. The similarity in the mean swelling ratio among seven relatively immature Type II kerogens (Qv ) 1.26 to 1.42) is striking. Three relatively immature Type IIIC kerogens also have similar mean swelling ratios (1.201.39). The maximum swelling ratio for the average Type II and Type IIIC kerogens was 1.45 and 1.25, respectively.

306 Energy & Fuels, Vol. 20, No. 1, 2006

Kelemen et al.

Figure 5. Plot of mean swelling ratio (Qv) for type IIIC kerogen as a function of the solubility parameter of the solvent (δ). Experimental error bars reflect 68% confidence intervals.

Swelling response changes as kerogen becomes highly mature. The influence of thermal maturation was first studied by Larsen and Li,10 who used Type I kerogen from the Uinta Basin and found that the optimum swelling occurred at a solubility parameter of 19.4 (J/cm3)1/2 at all levels of maturity. The principal conclusions from this study are that maturity has a negligible effect on the kerogen solubility parameter and there is little decrease in swelling ratio until the latest stages of maturity. However, the maximum Qv values of ∼3.0 reported by Larsen for all but the most mature Type I kerogens are far larger than that observed for other coals or kerogens where the maximum swelling ratios range from 1.2 to 1.6.5,9,11 These lower Qv values are consistent with our observations. It is possible that the highly aliphatic nature of Type I kerogens causes them to behave differently than Type II and IIIC kerogens with respect to swelling and thermal maturation. The maximum swelling response decreases at higher levels of sample maturity. A plot of the mean Qv versus Rock-Eval Tmax for the Type II (D) and Type IIIC (H) samples illustrates this behavior in two maturity suites of genetically related kerogens (Figure 8). For the Type II (D) samples, the mean swelling ratio appears relatively constant at low Tmax, followed by a small but statistically significant drop in the sample with a Tmax ) 443 °C, and a much greater drop in the mean swelling ratio for the sample with a Tmax ) 479 °C. The Type III (H) samples swell less than the Type II (D) samples at comparable Tmax temperature but qualitatively exhibit the same decrease in Qv with increasing Tmax. The highest Tmax samples, Type IIIC (H-2) and Type IIIC (H-3), have the lowest mean swelling ratios among all samples. 4. Discussion An extended Flory-Rehner and Regular Solution Theory framework21 for swelling of kerogen was used to interpret the average swelling data for Type II and Type IIIC kerogens in different solvents. All swelling data for kerogen, except for experiments involving Type IIIC kerogens and pyridine, could be interpreted within this theoretical framework. Optimizing the match between the experimental data and regular solution predictions enabled the thermodynamic parameters (solubility parameter, cross-link density, and native swelling) to be determined for an average Type II and a Type IIIC kerogen. These parameters enable general predictions of swelling behavior of these kerogens to other solvents and complex mixtures of solvents. For example, Table 10 shows the

Figure 6. Correlation index contours from attempts at fitting the average swelling data from Type IIIC kerogen by varying the solubility parameter, cross-link density, and native swelling of kerogen. (A) Native swelling versus cross-link density, (B) native swelling versus solubility parameter, and (C) cross-link density versus solubility parameter.

calculated swelling by 50:50 mixtures of n-hexadacane/solvent. These predictions are confirmed by experimentation where the swelling behaviors of two Type II and two Type IIIC kerogens were investigated using 50:50 n-hexadecane/solvent mixtures at 150 °C. All solvents listed in Table 3, except for n-decane, were considered. A compositionally based swelling ratio, Qc, was calculated using the compositional changes in the solvent

Organic Matter and Maturity Effects on Kerogen Swelling

Figure 7. Comparison experimental results and predicted swelling of the average Type IIIC kerogen for different solvents calculated using Regular Solution Theory and kerogen parameters listed in Table 6. Experimental error bars reflect 68% confidence intervals.

Figure 8. Plot of mean swelling ratio as a function of Rock-Eval Tmax for Type II and IIIC maturity samples. Table 10. Predicted Swelling in 50:50 Hexadecane/Solvent Mixtures and the Percentage Solvent Contribution to Swelling for an Average Type II Kerogen percent swelling contribution cyclohexane decalin toluene tetralin 1-methylnaphthalene 2,5 dimethylpyrrole pyridine benzofuran benzothiophene

Qv

hexadecane

solvent

1.260 1.267 1.301 1.320 1.341 1.359 1.409 1.374 1.388

9.7 10.5 4.2 3.0 1.9 1.2 0.4 0.9 0.7

90.3 89.5 95.8 97.0 98.1 98.8 99.6 99.1 99.3

mixture as measured by gas chromatography and mass balances for the solvent-kerogen system, assuming that there is no uptake by the reference solvent (n-hexadecane). The densities of the solvent and kerogen components are needed to complete the mass balance calculations. For the solvents, this is a well-defined quantity. For kerogen, the apparent density can be measured for the finely powdered kerogen particles after centrifugation. Table 11 shows the apparent density values for the kerogens used in the compositionally based swelling experiments. The theoretical density of the organic matter of each kerogen was calculated based on a chemical structural model and an established correlation between chemical composition and density (Table 11). The calculated density of kerogen falls in a much narrower range than the measured apparent density. There

Energy & Fuels, Vol. 20, No. 1, 2006 307

Figure 9. Comparison of the volumetric swelling ratio (Qv) and the compositional-based swelling ratio (Qc) for kerogen. Table 11. Apparent Density and Eight Percent of Ash for Different Kerogens Used in the Compositionally Based Swelling Studies with Solvent Mixtures kerogen type (name) Type II (A-1) Type II (C-1) Type IIIC (F-1) Type IIIC (G-2)

Wt % ash

apparent density (g/cm3)

calculated density (g/cm3)

30.1 17.5 24.5 8.4

0.57 0.25 0.79 0.75

1.13 1.13 1.27 1.26

is good directional agreement between the volumetric swelling ratio Qv and compositionally based swelling ratio Qc derived using calculated densities for each kerogen (Figure 9). This agreement confirms the preferential absorption of the solvent over the reference n-hexadecane. Among the requirements for interpreting the swelling response of kerogen within a Regular Solution Theory framework are that the kerogen behaves like an elastomer network upon swelling (i.e., rubbery, not glassy state) and that the system is equilibrated. Almost all studies of coal swelling by solvents have been conducted near room temperature and atmospheric pressure. Recent results indicate that coal does not have a measurable glass transition temperature (Tg) below its degradation point (T > 250 °C).33 However, organic solvent molecules dissolved in coal are believed to act as plasticizers and greatly lower Tg.34 The Tg of coal is expected to increase with increasing pressure as the compression of polymers decreases free volume and increases Tg. Coals are anisotropically deformed,35 which is understandable if the macromolecular structure of coal is developed under the influence of lithostatic pressure and is in a “glassy” state. The predominance of a hydrostatic pressure influence and a rubbery physical state would favor isotropic behavior. Temperature may influence the rate of swelling as the movement of molecules within coal will depend on its physical state and will be much faster in a rubbery state than in a glassy state. The kinetics of swelling of Argonne Premium Coal was studied at temperatures ranging from 10 to 60 °C.34 In general, the ultimate maximum extent of swelling of Argonne Premium Coal is achieved after 300 min at room temperature and 10 min near 60 °C. These results confirm earlier findings conducted at higher temperatures, up to 100 °C, that the ultimate swelling ratio is insensitive to temperature.36-39 The physical state of (33) Opaprakasit, P.; Painter, P. Energy Fuels 2003, 17, 354. (34) Larsen, J. W. Int. J. Coal Geol. 2004, 57, 63. (35) Larsen, J. W.; Flowers, R. A., II; Hall, P. J.; Carlson, G. Energy Fuels 1997, 11, 998. (36) Otake, Y.; Suuberg, E. M. Energy Fuels 1997, 11, 1156.

308 Energy & Fuels, Vol. 20, No. 1, 2006

Type II kerogen and its response to solvents are expected to be somewhat different than that of coal as the chemical structure of kerogen is significantly more aliphatic. This chemical difference is expected to produce a “rubbery” physical state at lower temperature. Organic solvent molecules dissolved in Type II and IIIC kerogens also are believed to act as plasticizers and lower Tg similar to their effect on coal. In geologic situations, the ability of Type II kerogen to produce significant quantities of bitumen during maturation also will favor formation of a “rubbery” state. However, the glass to rubber transition and the effects of bitumen, solvents, pressure, and form (e.g., isolated kerogen versus whole rock) have not been quantitatively explored. The ultimate swelling level of kerogen in most solvents does not appear to be very sensitive to temperature. Our experiments found no statistical difference in the Qv values for Type II and IIIC kerogens at 30 and 90 °C, and the mean Qv at 150 °C is only slightly greater than that at lower temperatures (3.0 and 2.1%, respectively). The weak temperature dependence of the swelling data indicates that these kerogen solvent systems have nearly equilibrated after 24 h at 30 °C. It is clear from the response of Type IIIC kerogen toward pyridine that specific interactions exist. Nevertheless, the data for all other solvents used could be interpreted within the basic regular solution model framework. Pyridine is expected to disrupt hydrogen-bonding interactions, and these interactions aid the self-association of the organic network structure and function as noncovalent bonded cross-links. These kinds of interactions are not prominent in Type II kerogens studied. This is a favorable situation for application of Regular Solution Theory for quantitative prediction. More work is needed for a comprehensive understanding of Type IIIC kerogen. Macerals composition is believed to be an important control on oilproneness of coal25 and may be a significant factor in determining the swelling behavior of Type IIIC kerogen toward different solvents. Hydrogen-bonding interactions may be less important in aliphatic-rich oxygen-poor macerals, and these macerals are more likely to participate in the oil formation processes.25 Methylation of Type IIIC kerogen samples and maceral fractions (37) Otake, Y.; Suuberg, E. M. Fuel 1989, 68, 1609. (38) Ndaji, F. E.; Thomas, K. M. Fuel 1993, 72, 1525. (39) Cody, G. D.; Eser, S.; Hatcher, P. G.; Davis, A.; Sobkowiak, M.; Shenoy, S.; Painter, P. C. Energy Fuels 1992, 6, 716.

Kelemen et al.

followed by solvent swelling studies would help to elucidate the significance of hydrogen-bonding interactions in Type IIIC kerogen. 5. Summary The volumetric swelling ratio was determined for a suite of Type II and Type IIIC kerogens in solvents that reflect the major components of petroleum. Comparable swelling ratios were obtained after exposure to solvents at 30, 90, and 150 °C after 24 h, indicating that these systems are nearly equilibrated. Mean Qv differences are found among different solubility parameter solvents. Mean Qv differences were found among different kerogens. There is a significant drop in the mean Qv between immature and overmature Type II and IIIC kerogens. All swelling data for kerogen, except for experiments involving Type IIIC kerogens and pyridine, could be interpreted within an extended Regular Solution Theory modeling framework. Optimizing the match between the experimental data and theory predictions enabled the thermodynamic parameters (solubility parameter, cross-link density, and native swelling) to be determined for an average Type II and a Type IIIC kerogen. These parameters enable general predictions of swelling behavior of these kerogen types to other solvents and complex mixtures of solvents. Good correlations are observed between the predicted and observed swelling of kerogens in binary mixtures of n-hexadecane/solvent mixtures. These results indicate that this theoretical framework can be used to determine the multicomponent equilibrium for kerogen and complex mixtures representative of those thermally generated by kerogen. In this way, it should be possible to determine the composition of bitumen in the swollen kerogen and in the expelled fluids and test the importance of this chemical fractionation mechanism in petroleum formation. Acknowledgment. We thank J. W. Larsen, P. J. Mankiewicz, A. E. Bence, W. A. Symington, Y. Xiao, H. Freund, M. L. Gorbaty, B. G. Silbernagel, T. C Halsey, and M. Siskin for valuable discussions during the course of the work. We also thank M. Vandenbroucke and F. Behar of IFP for supplying some of the demineralized kerogen samples. EF0580220