74
Energy & Fuels 1990,4, 74-77
Solvent Swelling Studies of Two Low-Rank Coals John W. Larsen* and Susan Shawver Department of Chemistry, Lehigh University, Bethlehem, Pennsylvania 18015 Received March 24, 1989. Revised Manuscript Received October 12, 1989
A Wyodak subbituminous coal, a Texas lignite, and their 0-methylated and 0-acetylated derivatives were swollen in both polar and nonpolar solvents (16 total). The lignite swollen with nonpolar solvents does not follow regular solution theory while the subbituminous coal does in spite of similar oxygen contents. The cross-link density of the subbituminous coal was calculated and is slightly greater than that of Illinois No. 6 coal, consistent with the coalification process being a net depolymerization.
Introduction Coals are complex macromolecular solids. Of the techniques used to define the macromolecular structure of various coals, solvent swelling is the one most often used.l It has provided useful insight into the structure of a number of bituminous coals and has also been used to provide valuable information about the course of both pyrolysis and conversion reaction^.^^^ For reasons we will outline later, it is not clear that the solvent swelling approach will be useful with subbituminous coals and lignites. In this work, we explore the utility of solvent swelling to elucidate the macromolecular structure of a subbituminous coal and a lignite. In an equilibrium solvent swelling measurement, the macromolecular network and the solvent are brought into contact and sufficient time is allowed for the solvent to completely dissolve in and swell the coal. At equilibrium, the free energy for dissolution of the solvent in the coal is exactly balanced by the elastic restoring free energy of the coal n e t ~ o r k . To ~ a useful first approximation, this restoring free energy is entirely entropic and is governed by the length of the macromolecular chains between branch points. There are two terms involved in the free energy of dissolution of the solvent in the coal. The first is entropic; the concentration difference between the bulk solvent and the interior of the coal. The second is any interaction between the solvent and the coal, usually given by the Flory x parameter. Except for one now being de~ e l o p e dthe , ~ statistical treatments that have been used to calculate the number-average chain length between branch points (MC) in coals assume that mixing between the solvent and the coal is It is explicitly assumed that there are no specific one-to-one interactions such as hydrogen bonds or charge-transfer complexes formed between the solvents and structural units in the coal. Our previous approach has been to avoid the difficulties caused by one-to-one specific interactions by using (1) Quinga, E. M. Y.; Larsen, J. W. In New Trends in Coal Science; Yurum, Y., Ed., NATO AS1 Series 244; Kluwer Academic Publishers: Boston, 1988. (2) Suuberg, E. M.; Unger, P.E.; Larsen, J. W. Energy Fuels 1987, I , 305-308. (3) Bockrath, B. C.; Illig, E. G.; Wassel-Bridger, W. D. Energy Fuels 1987, 1, 226-227. (4) This discussion follows that in: Flory, P. J. Principles of Polymer Chemistry; Cornel1 University Press: Ithaca, NY, 1963. ( 5 ) Painter, P. C.; Park, Y.; Coleman, M. M. Energy Fuels 1988, 2, 693-702. (6) Kovac, J. Macromolecules 1978, I I , 362-365. (7) Lucht, L. M.; Peppas, N. A. Fuel 1987,66,803-809; J.Appl. Polym. Scr. 1987, 33, 2177-2785.
0887-0624/90/2504-0074$02.50/0
Table I. Elemental Analyses of Coals Useda,* coal % C % H % N % S %Oc %MM Rawhide Big Brown lignite
69.3 69.7
4.42 5.11
0.96 1.35
0.56 1.40
24.3 22.4
8.4 11.0
CHN analysis performed on a Perkin-Elmer 240C elemental analyzer. Sulfur analyses performed by Galbraith Laboratories, Knoxville, TN. bDMMF basis; % MM = 1.08(ASH). % 0 by difference.
only swelling data obtained with nonpolar solvents to calculate Mc values. Even with the relatively nonpolar bituminous coals, the degree of swelling achieved with these nonpolar solvents is not large. The low-rank coals are highly oxygenated and are much more polar than are bituminous coals. Because of this, their interaction with nonpolar solvents is expected to be much less than that with bituminous coals. From a simple but essentially correct viewpoint, the character of the nonpolar solvents is much more like the chemical character of the bituminous coals and so these will interact better (like dissolves like). The greater the oxygen content and the greater the polarity of the low-rank coals, the less the interaction with nonpolar solvents compared to coal-coal interactions and therefore the lower the swelling values. The presence of a variety of highly polar oxygen functional groups also raises the possibilities for more specific interactions as well as more self-association of the coal through internal hydrogen bonding and ionic interactions. All of these will complicate and make more difficult the determination of M cby solvent swelling.* We anticipate that as coal rank decreases and the coals become more polar, the coal-solvent interactions would diverge more from regular solution behavior. At some point, the divergence will become sufficiently great so that solvent swelling approaches which are based on regular solution behavior will fail. Our purpose in this work was to determine the low-rank limit to the use of solvent swelling to gain information about coal macromolecular structure. For the reasons enumerated above, we were pessimistic about the utility of the solvent swelling approach for low-rank coals. Experimental Section Coal Preparation. Both Wyodak and Big Brown lignite were supplied by Dr. Ron Liotta of Exxon Research a n d Engineering ~
(8) A reviewer has raised the issue of mineral contributions to coal
swelling. Control experiments demonstrated that clays do not swell in nonpolar solvents and swell less than coals in pyridine. Their effect on overall swelling is to reduce it slightly, not significantly.
0 1990 American Chemical Society
Energy & Fuels, Vol. 4 , No. 1, 1990 75
Solvent Swelling Studies of Low-Rank Coals Table 11. Solvents Used To Swell Coals solvent solvent no. (~al/cm~)'/~ 1 7.2 n-pentane 2 8.2 cyclohexane 3 8.8 o-xylene 4 8.9 toluene 5 9.2 benzene 6 9.5 chlorobenzene 7 9.5 tetralin 8 1,2-dichloroethane 9.8 9 10.0 carbon disulfide 10 10.0 nitrobenzene 11 biphenyl 10.6 12 tetrahydrofuran 9.1 13 10.7 pyridine 14 11.0 N-methyl-2-pyrrolidone 15 11.9 acetonitrile 12.7 16 ethanol Co. The Wyodak coal is from the Smith Seam (Rawhide) in Gillette, WY, and was received stored under water in 1-in.chunks. These were reduced in size under water in a nitrogen-filled glovebag by use of a mortar and pestle. The ground coal was then transferred to a round-bottom flask and the water removed with a rotary evaporator at about 30 OC. The coal was then ground to -100 Tyler mesh under nitrogen by use of a ball mill and separated into individually stored fractions by use of a riffler, all operations being carried out in a glovebox under nitrogen. The Big Brown lignite was received as a -70 Tyler mesh sample and was ground to -100 Tyler mesh by use of a coffee mill in a glovebox. Before use, coals were dried under vacuum a t approximately 100 "C. The elemental analyses are given in Table I. Solvent Purification. All solvents were purified by standard procedures: except carbon disulfide, cyclohexane, and tetrahydrofuran which were purchased in high purity (Aldrich, Gold Label) and did not require further purification. The T H F was stabilized with butylhydroxytoluene. The solvents employed in the study, their solubility parameters, and their assigned solvent numbers are listed in Table 11. Volumetric Swelling Measurements. Approximately 0.4-0.8 g of coal (-100 Tyler mesh) was placed in an 6-mm-0.d. Pyrex tube (sealed on one end) under nitrogen atmosphere. The tube was placed in a Fisher centrifuge (1725 or 3400 rpm) and repeatedly centrifuged and tapped by hand on a hard surface until the coal bed reached a minimum height, Hl, which was measured. The packed coal bed was broken up by tapping the tube against a hard surface. This coal was then distributed along the tube to prevent clumping due to rapid swelling when solvent was added. A sample of 2-3 mL of solvent was added to the tube and mixed thoroughly with the coal. The tube was allowed to stand vertically for a minimum of 24 h. The tube was again centrifuged and tapped by hand until the measured height of the coal bed reached a minimum height, H2 After thorough shaking and the passage of another 24 h, this process was repeated until a constant value of H2was reached. The volumetric swelling ratio on a dry basis, Qv,dry, is defined as
t
I
1.7 1.6 1.5
I
I
I
,
I I
6 b
Q"
1.1 1.0
p
lL=.-e&
7
a
..
6 , (cal/cma)'h Figure 1. Swelling ratio of pyridine-extracted Rawhide coal ( 0 ) and its 0-methylated (m) and 0-acetylated (A)derivatives as a function of swelling solvent solubility parameter. (See Table I1 for solvent identification.) Pyrex tube, obscuring the coal bed. A relationship between the height of the coal column (HI)and the weight of the coal in the 6-mm tube was established as follows: The internal height of the tube is measured when the tube is weighed (to 0.1 cm). After addition of the coal and repeated centrifugation and tapping (assuming that when the other samples have reached the minimum H1the tube in question has also reached the minimum Hl), a ruler cut to fit inside the 6-mm tube is gently inserted into the tube until it rests lightly on the top of the coal bed. The height from the top of the coal bed to the top of the tube is measured. The difference in the internal height of the empty tube and height measured from the top of the coal bed is taken as H1. The weight of the 0-acetylated coal sample was then used to calculate HI in subsequent experiments. 0-Acetylation. Following the procedure of Blom et al.," approximately 10 g of dry coal was weighed into a 250-mL round-bottom (rb) flask. A mixture of 200 mL of acetic anhydride and pyridine (1:2 by volume) was added. The mixture was refluxed with stirring under nitrogen for 24 h. The mixture was then cooled and transferred to a 1000-mL Erlenmeyer flask containing 500 mL of distilled water. This mixture was filtered through a medium-porosity Whatman filter using a Buchner funnel under nitrogen. The 0-acetylated coal was washed with warm distilled water until free of acid (the wash water was tested with litmus paper) and placed in a vacuum oven at 110 "C to dry overnight. The complete derivatization of the hydroxyl groups was confirmed by diffuse-reflectance FT-IR. 0-Methylation. The basic procedure developed by Liotta et a1.12 was used with a few modifications. Approximately 10 g of coal was weighed into a 500-mL round-bottom flask. Samples of 34.5 mL of 1M aqueous tetrabutylammonium hydroxide, 100 mL of THF, and 17.5 mL of methyl iodide were added. The mixture was allowed to stir overnight under nitrogen. The mixture was tested with litmus paper and not found to be basic. The mixture was then filtered through a dried Soxhlet extraction thimble (under nitrogen).and washed with an additional 1.5-2.0 L of hot distilled water. The thimble was placed in a Soxhlet extractor charged with approximately 250 mL of distilled water and washed for 5-6 days, under nitrogen. The thimble containing the coal was dried in a vacuum oven a t 105-110 "C overnight. Complete 0-methylation was confirmed by diffuse-reflectance FT-IR.
where v, is the volume of solvent absorbed by a unit volume of dry coal, uC,* The reproducibility of the swellings was good. For example, the average of six measurements using Illinois No. 6 coal and toluene gave a swelling ratio of 1.5 f 0.04 rsd. Thii procedure is very similar to that published and has been shown to give results identical with those of gravimetric swelling measurements.'O Normally, in a solvent swelling experiment, the initial height of the coal bed, HI, is measured visually. In the case of the 0-acetylated coals, the direct observance of H1 was difficult, if not impossible. The 0-acetylated coal clings to the side of the
Figure 1shows the dependence of the swelling ratio of pyridine-extracted Rawhide coal and i t s acetylated and 0 - m e t h y l a t e d derivatives vs the solubility parameters of eight nonpolar solvents. The methylation and acetylation procedures are known to remove internal hydrogen bonding i n the coal b y derivatizing the hydroxyls.12 Removing the internal hydrogen bonds should increase the
(9) Perrin, D. D.;Amarego, W. L. F. Purification of Laboratory Chemicals, 2nd ed.; Pergamon Press: Elmsford, N.Y., 1980. (10) Green, T . K.; Kovac, J.; Larsen, J. W. Fuel 1984,63,935-938.
(11) Blom, L.; Edelhauser, L.; van Krevelen, D. W. Fuel 1959, 18, 531-538. (12) Liotta, R.; Rose, K.; Hippo, E. J. Org. Chem. 1981,46, 277-283.
Results
76 Energy & Fuels, Vol. 4, No. 1, 1990 1.7 1.6
Larsen and Shawuer
k
j
c
J
3 4 5 7
i Qv
1.2 1.o7 1.1
8
0
.
10
9
11
12
5 6
0 .
4.:
6, (cal/cm3)"2 Figure 2. Swelling ratio of pyridine-extractedBig Brown lignite ( 0 )and its 0-methylated (w) and 0-acetylated (A)derivatives as a function of swelling solvent solubility Parameter. (See Table I1 for identification.)
: :I
2.21
Qv
I
I
I
1
1.0 I 7
?
3
I
I
I
I
I
8
I
9
10
11
12
13
6, (cal/cm3)'/2 Figure 4. Swelling ratio of pyridine-extracted Big Brown lignite as a function of swelling solvent solubility parameter. (See Table I1 for solvent identification.)
Table 111. x Parameters for the Nonpolar Solvents and Pyridine-Extracted, 0-Acetylated, and 0-Methylated Rawhide Coal PEXTd ACT, ME'P
1.6 1.4
2
0 9
6, (cal/cm3)'/* Figure 3. Swelling ratio of pyridine-extractedRawhide coal as a function of swelling solvent solubility parameter. (See Table I1 for solvent identification.)
swelling in nonpolar solvents as observed. The methylation procedure used tetrabutylammonium hydroxide, which is now known to be tenaciously retained by the c0a1s.l~ If the coal/nonpolar solvent systems follow regular solution theory, smooth bell-shaped curves for the plot of swelling ratio vs solubility parameters are expected.14 A skew plot is observed for the pyridine-extracted coal while reasonable regular solution behavior is shown for the 0-methylated and 0-acetylated derivatives. This coal follows regular solution theory much more closely than we initially anticipated. A similar plot is shown for Big Brown lignite in Figure 2. The pyridine-extracted and methylated coals do not show good regular solution behavior. While a bell-shaped curve has been drawn through the 0-acetylated coals, the data do not fit this curve well and we conclude that this system does not follow regular solution theory. Figures 3 and 4 show the plots of the swelling ratio of Rawhide and Big Brown lignite, respectively, vs solubility parameters of both nonpolar and polar solvents which are good hydrogen bond acceptors. As anticipated, hydrogen bond acceptors swell the coal significantly more than do the nonpolar solvents. This is undoubtedly due to a combination of their greater interaction with the coals and the fact that they disrupt the coal-coal hydrogen bonding leading to a system that effectively has a lower cross-link density. With both coals, the best swelling solvents are
solvent n-pentane cyclohexane o-xylene toluene benzene chlorobenzene tetralin carbon disulfide biphenyl
VI: cm3/mol
xH*
xc
xHe
xc
116.2 108.7 121.2 106.8 89.4 102.1 136.0 60.0 154.1
1.042 0.311 0.101 0.065 0.014 0.000
1.342 0.611 0.401 0.365 0.314 0.300 0.300 0.325 0.616
0.788 0.184 0.033 0.016 0.000 0.016 0.021 0.065 0.312
1.088 0.485 0.333 0.316 0.300 0.316 0.321 0.365 0.612
0.000
0.025 0.316
'At 25 "C. *Calculated as described in ref 14. The solubility *parameterof the pyridine-extracted Rawhide coal estimated to be 9.5 (cal/cm3)'~*. cCalculated as described in ref 14. xs = 0.3. dPEXT = Pyridine-extracted. 'ACT = 0-acetylated; MET = 0methylated. Calculated as described in ref 14. The solubility parameter of the 0-acetylated and 0-methylatedRawhide coals estimated to be 9.2 (cal/cm3)'/*.
pyridine and N-methylpyrrolidone, which are both good hydrogen bond acceptors and have similar solubility parameters. We do not understand why the relative swelling abilities of these two solvents should be reversed for the two coals.
Discussion It is clear from the data presented that the Big Brown lignite does not follow regular solution theory. This makes it impossible to calculate x for coal-solvent pairs. This calculation requires knowledge of the solubility parameter (6) of the coal, and this can only be obtained by using this approach if the coal/solvent system follows regular solution theory. Swelling ratios for the coal and its derivatives in nonpolar solvents are all below 1.6. Solvents that are polar and good hydrogen bond acceptors lead to swelling ratios on the order of 2.0, values a bit smaller than those encountered with bituminous coals. We have previously published solvent swelling values for these coals after deminera1i~ation.l~This process increases the solvent swelling with polar solvents to values between 2.5 and 3, probably due to increased solvent-coal interactions. Rawhide coal and its 0-methylated and 0-acetylated derivatives do appear to follow regular solution theory. Looking at the curves in Figure 1, it seems that there is
(13)Ettinger, M.; Nardin, R.; Mahasay, S. R.; Stock, L. M. J. Org.
Chem. 1986,51, 2840-2842.
(14) Larsen, J. W.; Green, T. K.; Kovac, J. J. Org. Chem. 1985, 50, 4729-4735.
(15) Larsen, J. W.; Pan, C.-S.; Shawver, S. Energy Fuels 1989, 3,
557-561.
Energy & Fuels, Vol. 4,No. 1, 1990 77
Solvent Swelling Studies of Low-Rank Coals 3400
I
I
I
I
I
162
Table IV. Organic Oxygen Functional Groups per 100 Carbon Atoms (Reference 17) hvdroxvl carboxvlic nonacidic 4.8 3.2 9.8
coal Rawhide Texas lignite
Figure 5. Relationship between number-average cluster molecular weight (M,)and assumed values for number of clusters between branch points (M for pyridine-extracted Rawhide coal ( 0 )and its 0-methylated (m) and 0-acetylated derivatives (A).
a shift in the solubility parameter of the coal on derivatization from about 9.5 for the pyridine-extracted coal to about 9.2 and perhaps 9.1 for the 0-methylated coal. This decrease generally parallels the change in polarity expected by the derivization. The assignments are somewhat uncertain, and in the calculations that follow the pyridineextracted coal is assigned a solubility parameter of 9.5, while both the 0-acetylated and 0-methylated Rawhide are assigned a solubility parameter of 9.2. Using these values for the solubility parameter of the coal leads to the values for the Rawhide coal and its derivatives shown in Table 111. These x values were used to calculate the cross-link density of the coals using the Kovac equation as previously de~cribed.'~ The average values of the M, (number-average molecular weight between cross-links) calculated from the nonpolar solvent swelling data were used to calculate the curves shown in Figure 5. Here, M ois the number-average molecular weight of the individual clusters in the network and N is the number of rotatable repeating units between cross-links. Thus N = Mc/Mo. The lines in the figure are experimental and are derived from the average of the swelling measurements in the nonpolar solvents. The figure is used in the following way: one's favorite estimate for the number-average molecular weight of the clusters is selected and one moves across the figure horizontally to find the intersection of that weight with the line for the coal or coal derivative in question. From that intersection,
4.8
3.7
7.8
the number of clusters ( N ) can be determined. The product of N times the estimated cluster size gives the number-average molecular weight between cross-links. For example, if the average cluster molecular weight is 300 (dashed line), Mc for both the acetylated and the methylated coal is 600. As was true for the bituminous coals, the treatment indicates that essentially every cluster interacts with another cluster for the pyridine-extracted coals. Acetylation or methylation results in an increase in the chain length between branch points. The molecular weight between cross-links in this coal appears to be significantly lower than that for Illinois No. 6 and Bruceton coals; that is, this coal is more highly cross-linked. This is consistent with the notion that the coalification process is a net depolymerization.16 It appears that solvent swelling with nonpolar solvents may be a useful technique with subbituminous coals. It clearly fails with this lignite. The overall picture which emerges is that of a tightly cross-linked highly polar material. Reliance on a single technique for characterization of a network of this type is unsafe. Mechanical measurements have been carried out on some coals, but rely on a statistical analysis of the network which forms part of the mathematical treatment of solvent swelling." We are currently exploring other ways to characterize the macromolecular structure of low-rank coals and will report on them in due course. Both of these coals contain similar amounts of organic oxygen, similarly distributed among the functional groups (see Table IV), yet one follows regular solution theory when swollen with organic solvents and the other does not. Clearly, factors other than the bulk and composition are important. We do not yet have any knowledge of what these factors are.
Acknowledgment. We are grateful to the U.S. Department of Energy and to the Exxon Education Foundation for financial support of this work. We gratefully acknowledge the aid of Dr. Andy Baskar (dec) in obtaining the IR spectra. Conversations with Tom Green materially aided the progress of this work. We thank Dr. Ron Liotta for his most helpful criticism of an earlier draft of this paper. Registry No. n-Pentane, 109-66-0; cyclohexane, 110-82-7; o-xylene, 95-47-6; toluene, 108-88-3; benzene, 71-43-2; chlorobenzene, 108-90-7;tetralin, 119-64-2;carbon disulfide, 75-15-0; biphenyl, 92-52-4; pyridine, 110-86-1. (16) Larsen, J. W.; Wei, Y.-C. Energy Fuels 1988,2, 344-350. (17) Howell. J. M.: Petmas. N. A. Fuel 1987.66.810-814. Barr-Howell. B. D.;'Howell,'J.M.;Pe&s; N. A. P r e p . Pap-Am. Chem. SOC.,Diu: Fuel Chem. 1985, 30(1),64-66. (18) Liotta, R. Personal communication.