Energy & Fuels 1988,2, 815-818 Excessive hydrotreating severity in stage 2 will increase the abundance of lower boiling components in the stage-2 effluent as shown in Figure 8. These components will eventually be removed as light oil in the process (e.g. in the WFD still) thereby further reducing the total solvent inventory. The above two relationships between solvent quality and quantity emphasize the need to hydrotreat with the minimal possible severity sufficient to reach the target SHDI. Concluding Remarks. The SHDI parameter, intended as an index of solvent hydrogen donor capacity, is simple in conception. I t is also simply measured, by recording a single 'H NMR spectrum. The data presented herein and earliers show that coal conversions in noncatalytic hydroliquefaction reactions correlate well with SHDI. Increasing SHDI gives increasing oil yields up to a plateau value. This basic conversion-SHDI relationship has now
815
been shown for three coals in batch trials and for two of these coals (herein) also in continuous processing. Thus far, therefore, the SHDI parameter has been found to serve its intended purpose well. It is a useful guide for the preparation of start-up solvents and the rehydrogenation of recycle solvents in such a way that coal-derived oil yields are maximized.
Acknowledgment. We thank D. J. Cookson for providing all NMR data, A. A. Awadalla for providing all batch autoclave data, and C. P. Lloyd for assistance with numerical analysis. Support for this work was provided under the National Energy Research Development and Demonstration Program, administered by the Australian Commonwealth Department of Primary Industries and Energy. We thank the Broken Hill Proprietary Co., LM., for permission to publish this work.
Coal Solubilization. Factors Governing Successful Solubilization through C-Alkylation Mikio Miyake and Leon M. Stock* Department of Chemistry, The University of Chicago, Chicago, Illinois 60637 Received April 18, 1988. Revised Manuscript Received August 4, 1988
A systematic investigation of the base-catalyzed C-alkylation of a low-volatile bituminous coal from the lower Kittanning Seam was carried out to assess the factors important for the conversion of this 89.3% C (daf) coal into products that could be extracted into pyridine. Although base strength is significant, the chain length of the alkyl group introduced during the reaction is the single most important factor for the achievement of a high, 90%, extractability. Molecular modeling considerations suggest that the larger alkyl groups disrupt the stacking of aromatic compounds and render the material soluble. Introduction The solubilization of coal in conventional organic solvents, such as pyridine, would enable coal chemists to obtain more reliable information about its constitution. Although many chemical reactions have been proposed to transform coal to soluble products,' most of these reactions disrupt strong covalent linkages. Consequently, no unambiguously decisive advances have been made.'S2 More recent interest has been focused on alkylation since the introduction of an alkyl group can be selectively accomplished and since the alkyl group can disrupt hydrogenbonding interactions and intermolecular polarization forces that may bind the solid.2 0-Alkylation does not solubilize most coals to an adequate degree,lp2 but C-alkylation, in which the proton of a carbon acid is abstracted by a strong base to produce an anion that can be alkylated with a suitable alkyl halide, is an attractive alternati~e.~ Ignasiak and co-workers demonstrated that after the reaction was repeated three times, up to 14 ethyl groups per 100 carbon atoms could be introduced into an 83.2% C coal and that up to 64% of the ethylation product of an 87.3% C coal could be solubilized in ~hloroform.~Several research (1) Davison, R. M. Coal Sci. 1982,1,84-160. (2)Stock, L. M.Coal Sci. 1982,1, 161-282. (3)Ignasiak, B.;Carson, D.; Gawlak, M. Fuel 1979,58,833. (4)Gawlak, M.; Carson, B.; Ignasiak, T.; Strausz, 0. P. Preprints of Papers, International Conference on Coal Science, Maastricht, The Netherlands, 1987; pp 57-60.
0887-0624/88/2502-0815$01.50/0
Table I. Analytical Data for PSOC 1197O Proximate Analysis (wt %) moisture 0.72 ash 10.28 volatile matter 16.38 fixed carbon 72.62 Ultimate Analysis (wt %, daf) 89.58 (89.3) H 4.84 (4.7) N 1.78 (2.2) S 1.17 0 (diff) 2.63
C
Maceral Analysis (wt % ) vitrinite 94.4 inertinite 5.6 liptinite 0.0 MAF, Btu
15 924
The observations in parentheses were obtained for the sample used in this study by Commercial Testing and Engineering Co. The other data were provided by The Pennsylvania State University Sample Program.
groups have been intensively investigating C-alkylation reactionsk8 and have obtained disparate results. For ex( 5 ) Mallya, N.; Stock, L. M. Fuel 1986,65,736-738.
(6)Lazarov, L.; Marinov, S. P.; Stefanova, M.; Angelova, G. Preprints Conference on Coal Science, Maastricht, The Netherlands, 1987;pp 745-748.
of Papers, International
0 1988 American Chemical Society
Miyake and Stock
816 Energy & Fuels, Vol. 2, No. 6,1988
ample, LazarovS reported a clear correlation between coal rank and solubility, while Ignasiak4 found no such correlation. It seemed to us that the differences in the results might arise because of differences in procedures and reagents. Indeed, no report defining the most important experimental factors in coal solubilization has appeared for the C-alkylation reaction. Accordingly, we have systematically studied the influences of different reagents, i.e. the base and the solvent, and different conditions, i.e. the base concentration and the reaction temperature, on the conversion of a high-ranking bituminous coal with a low oxygen content into soluble products. This choice was mandated by our desire to avoid 0-alkylation as a factor in the reaction.
Experimental Section Materials. The low-volatile, bituminous, Lower Kittanning Seam coal was obtained from The Pennsylvania State University Sample Bank. The analytical information is summarized in Table I. The coal sample was dried at 110 OC under vacuum for 48 h prior to use. Commercial anhydrous liquid ammonia (Matheson) was dried by passage through potassium hydroxide and barium oxide. Tetrahydrofuran (Aldrich)was refluxed over potassium metal and distilled prior to use. Alkyl halides (Aldrich),such as iodomethane, 1-iodobutane,and 1-iodooctane,and tetramethylethylenediamine (Aldrich), were dried by 5A molecular sieves. Pyridine and chloroform (Aldrich) were purified by distillation. The other chemicals, including butyllithium (1.6 M solution in hexane), sodium hexamethyldisilazide, lithium diisopropylamide-tetrahydrofuran, ammonium chloride, and methanol were used as received. Reaction Procedure. Reactions of the coal with sodium amide were conducted as described previouslyaS Typically, liquid ammonia (200 mL) was condensed in a flame-dried flask and sodium amide (45 mmol) was added. After 15 min, the coal sample (1 g) was added, and the mixture was agitated vigorously for 6 h at -75 OC. 1-Iodobutane (60 mmol) in tetrahydrofuran (100 mL) was added dropwise to the mixture over a period of 30 min. The supply of coolant was stopped, and the temperature was allowed to rise gradually to room temperature as the ammonia evaporated. Then, additional tetrahydrofuran (100 mL) was added. After 48 h, any remaining base was destroyed by the addition of solid ammonium chloride, and the mixture was diluted with methanol. The solvent and the exceas 1-iodobutanewere evaporated by using a rotary vacuum evaporator. Subsaquently,the product was dried at 90 "C under vacuum overnight. The reaction was conducted under deoxygenated and dried nitrogen. The product was fist washed with acidified aqueous methanol (methanol/water = 1/3 v/v) and then repeatedly with aqueous methanol (more than 20 L). The washing procedure was carried out in a nitrogen atmosphere. The product was collected and dried to constant weight at 110 OC under vacuum for 48 h. The alkylated product (about l g) was exhaustively extracted with pyridine in a Soxhlet apparatus. The other reactions were conducted by suitable modifications of this procedure. It should be noted that the same base concentration (45mmol/g of coal) was used in most experiments and that the reaction with n-butyllithium was carried out with an equimolar amount of tetramethylethylenediamine. Analyses of Products. The relationship between extractability in a Soxhlet apparatus and solubility at room temperature was assessed in the following way. The product of the basecatalyzed octylation reaction of PSOC 1197 coal was extracted in a Soxhlet apparatus with pyridine. Approximately 92% of the octylated coal was extracted. The pyridine-soluble portion of the product was isolated by removal of the pyridine with a rotary (7)Lazarov, L.; Marinov, S. P. Fuel 1987,66, 185-188. (8) Chambers, R. R., Jr.; Hagaman, E. W.; Woody, M. C. Polynuclear Aromatic Hydrocarbons; Ebert,L., Ed.; Advances in Chemistry 217;1987; Chapter 15, pp 255-268. (9) Mallya, N.; Stock,L. M. Prepr. Pap.-Am. Chem. SOC.,Diu.Fuel Chem. 1985, 30(2), 291-297.
Table 11. Influence of the Concentration of Sodium Amide on the Solubilization of PSOC 1197 alkylation," solubility, wt % mmol of wt % of Bu groups/ NaNHz product 100 C inpy in CHC& 0 15
45 100 200
103 109 138d 145d
0.8 2.4
5 20 50 60 61
26b 28'
OThe degree of alkylation was estimated from the weight gain. The number-average molecular weight was determined to be 6600. 'The number-average molecular weight was determined to be 4500. These high weight gains were realized in early experiments before the product workup procedures had been adequately developed. evaporator with a warm oil bath used to purge the residual pyridine. This solid (0.804g) was then redissolved in pyridine and the solution was passed through a 0.8 pm glass frit. Approximately 6% of the solid (0.047 g) was retained on the frit. In view of the fact that the pyridine-soluble materials were heated for a prolonged interval and exposed to air during their collection and redissolution, the result implies that there is a very close relationship between the measured extractability and the solubility of the alkylated products. Solid-state'V-NMR spectra of the coal samples were measured at Argonne National Laboratory by using a Bruker Model CXP-100 as described previously.1° The molecular weight measurementswere obtained by the VPO method using a Hitachi Perkin-Elmer 115instrument. The observationsfor three different experiments using chloroform were extrapolated to infinite dilution in the usual way.
Results and Discussion The Lower Kittanning Seam coal with 89.3% carbon (daf) was selected for study because of its low oxygen content and because of its high aromatic character. The fa value estimated from 13C NMR spectroscopic measurements is 0.85. Further, we anticipated that this coal as many other coals with 88.5-89.5% carbon (daf) would have novel properties. This feature was anticipated by prior work in this laboratory5 and by early reports of significant changes in the physical and chemical properties of other coals with about 89% carbon, for example, the average crystallite stack height1' and the high solubility in pyridine achieved in exhaustive reduction reactions.12 Since the original coal is only 5 % soluble in pyridine, changes in its constitution brought about by the C-alkylation reaction can be readily discerned. The Lower Kittanning Seam coal was treated with sodium amide in liquid ammonia at -75 OC. Preliminary experiments indicated that the reaction with the base could be carried out successfully in 6 h but that the alkylation reaction required a longer time. The influence of the concentration of sodium amide on the two-step butylation reaction is shown in Table 11. The C-butylation reactions add between one and two alkyl groups per 100 mol of carbon to this coal. Further, the degree of the solubilization is essentially independent of the base concentration. The results for 45,100, and 200 mmol of base/g of the coal gave essentially the same solubility in pyridine and in chloroform. In addition, we found that the number average molecular weight of the (10)Botto; R. E.;Choi, C.; Muntean, J. V.; Stock, L. M. Energy Fuels
1987,1, 270-273.
(11)Cartz, L.; Hirsch, P. B. Philos. Trans. R. SOC.London, A 1960, 2.52. 55'7-599. ---I
(12)Reggel, L.; Raymond, W. A.; Steiner, R. A.; Friedel; Wender, I. Fuel 1961, 40, 339.
Energy & Fuels, Vol. 2, No. 6,1988 817
Coal Solubilization Table 111. Influence of the Electrophilic Agent on the Solubilization of PSOC 1197 mmol of NaNHz 0
100 45 45 45
reagent none NHdCl Me1 1-BuI 1-OctI
wt % of product
alkylation, groups/ 100 C
102 109 109 121
2.4 2.9
solubilization, wt % in py 5 9 10 50
Table IV. Influence of the Base on the Solubiization of PSOC 1197 base NaN[(CH,),si12 NaNH2 LiN(i-P&
n-BuLil
90
wt %
alkylsolubility, of ation, wt % ' product Bu/100 C in py 105 1.5 16
conditions: solvent; pKA T,O C 25" NH,; -75 NHS; -75 NH,; -75 THF; rtc
35' 36" 42b
109 132 104
2.4 1.1
50 53 10
TMED
material in the chloroform extracts decreased from 6600 with 45 mmol of the base to 4500 with 200 mmol of the base. This observation suggests that high concentrations of the base favor bimolecular anionization reactions that lead to fragmentation and, successively, to retrogressive adduction reactions. Thus, we adopted a rather low concentration of the base, 45 mmol, in all the subsequent experiments. A large change in the reaction temperature had only a small influence on the course of the reaction. In sharp contrast, the length of the alkyl group has a major effect on the outcome of the reaction. The observations for protonation and alkylation of PSOC 1197 are presented in Table 111. When the reaction was terminated with ammonium chloride, the product was only 9% soluble in pyridine. This finding implies that the reaction of PSOC 1197 with modest concentrations of sodium amide does not significantly alter the molecular framework through hydrocarbon elimination reactions, eq 1, or other base-catalyzed fragAr2CHCHzCH2Ar+ B
Ar2C-CH2CH2Ar
+
Ar2C-CH2CH2Ar BH+ (la)
-
F?
Ar2C=CH;
+ ArCH2-
(lb)
mentation reactions.2 Although our observations are not encouraging, this feature of the chemistry certainly deserves further study. As anticipated from results of 0-alkylation and reductive alkylation reactions, the chain length of the alkyl group can have an enormous influence on the conversion of the coal to pyridine-soluble p r o d u ~ t s . ~ JThis ~ influence is especially dramatic for the octylation reaction, which provided a product that was 90% soluble in pyridine. This finding implies that the structural bulk of the added alkyl group is a key factor in determining the extractability of the alkylation product. Presumably, the steric requirementa for the alkyl groups are especially important for the high-ranking coals that have a more regular laminar structure. Inasmuch as sodium amide is not sufficiently basic to abstract protons from aromatic compounds, we presume that this amide abstracts protons from aliphatic structural elements with pKA < 35. Alkylation presumably takes place preferentially at acidic benzylic and fluorenylic positi~ns.'~ To examine this feature, we next turned our attention to the use of other basic reagents for the C-butylation of PSOC-1197 coal. Four different reagents with pKA values ranging from 26 to 42 were studied, with results given in Table IV. The results for the three amide bases that could be examined under the same conditions can be discussed with some confidence. The weakest base, sodium hexamethyldisilazide, PKA = 26,15 provides only 16% soluble (13)Wachowska, H.Fuel 1979,58,99-104. (14)Gawlak, M.; Cyr, N.; Carson, D.; Ignasiak, B. Prepr. Pap.-Am. Chem. Soc., Diu. Fuel Chem. 1980,25(4),111. (15)Fraser, R.R.;Mansour, T. S. J.Org. Chem. 1984,49,3443-3444.
"Reference 15. *Reference 16. r t = room temperature.
Scheme I coal^ + Base
=- Coal'
+ BascHt
irreversible sequence in THF
coal- +
0
CodH
+
0
0. -
GH4 + CH,=CHO
Reversible sequence in NH,
Coal- + NH,
CoalH
+ N-I2
products, but the stronger bases, sodium amide, pKA = 35,16 and lithium diisopropylamide, pKA = 36,16 provide much greater yields of soluble products. These observations imply that base strength is very significant in the solubilization reaction. Chambers and his co-workers have also examined the C-alkylation of PSOC-1197.8 They found that no more than two methyl groups were added per 100 mol of carbon in an 0-methylated sample of PSOC 1197 even after the reaction was repeated three times with bases such as trityllithium, pKA = 31, fluorenyllithium, pKA = 22, and 9-phenylfluorenyllithium, pKA = 19. The reaction with trityllithium provided 12% pyridine-soluble material. The yield of soluble products could be increased to only 30% by three repetitive reacti0ns.l' To examine this problem in another way, we attempted to compel anionization through the use of an irreversible, strongly basic reagent, butyllithium/ tetramethylethylenediamine, pKA = 42, in tetrahydrofuran. Only about 10% of the butylation product was soluble, Table IV. The reason for the failure of the reaction with butyllithium has not yet been established. It may simply be that tetrahydrofuran fails to permeate this high-rank coal. It seems more likely, however, that the more reactive coal anions formed in kinetically controlled reactions with the strongly basic organolithium are consumed via the formation of the tetrahydrofuran anion, which decomposes irreversibly to ethene and the enolate of ethanal, Scheme I.2~8J8No side-reactions of this kind can occur for the amides.
Conclusion Amide-catalyzed alkylation reactions can successfully convert higher ranking PSOC 1197 coal into soluble products. The stronger amide bases with pKA values near (16)Cram, D. J. In Fundamentals of Carbanion Chemistry; Academic: New York, 1965;pp 8-20,41. (17) Chambers, R. R., Jr., private communication and ref 8. (18)Bates, R.B.;Kroposki, L. M.; Porter, D. E. J. Org. Chem. 1972, 37,560-562. (19)Streitwieser, A., Jr.; Juaristi, E.; Nebenzahl, L. L. In Comprehensive Carbon Chemistry,Part A; Buncel, E., Durst, T., Eds.; Elsevier: New York, 1980; pp 323-381.
818 Energy &Fuels, Vol. 2, No. 6,1988
Miyake and Stock
Table V. Hydrocarbon DK. Data' compd
compd
PKA
& ax2 & & 18*5
PKA
23.0 30.6
19.6
lgV8
(PhhCH
31.5
@CHz*
31'8
(Ph)&H2
33.4
@
m
21.4 lgS9
33.5
6%
@& a Reference
19.
35 are more effective than the weaker amide bases with pKAvalues near 25. The extractability yields, in general, parallel the number of added butyl groups, which, in turn, parallel the base strength. Also, the dimensions of the alkyl group have a very profound influence on the degree of solubilization. Examination of the pKA data for hydrocarbons that may have structural elements representative of the coal molecules in PSOC 1197, Table V, suggest that C-alkylation with the less basic reagents occurs on fluorenylic structures whereas the more effective C-alkylation reactions with strong amide bases occur on hydroaromatic fragments and on benzylic structural elements. This feature of the chemistry was probed by an examination of the 13CNMR spectrum of the methylation product obtained by using m e t h y P C iodide, Figure 1. The spectroscopic observations reveal that there is only weak signal intensity above 50 ppm; hence, 0-alkylation is a minor side reaction. The much more intense resonances of the C-alkylation products span a rather broad region from 10 to 50 ppm that encompasses carbon, nitrogen, and sulfur alkylation. Subtle maxima appear near 20, 23, and 27 ppm. The chemical shift data are compatible with the methylation products of a variety of different hydrocarbons including fluorenyl derivatives such as 9-methylfluorene and 9-methyl-9,9-dimethylfluorene aa well as benzylic alkylation products such as 9-methyl-
9,10-dihydrophenanthrene,9,9-dimethyl-9,10-dihydrophenanthrene, 1,l-diphenylethane, and 2,2-diphenylpropane.
H
CHJ
H
CHJ
1 9 . 6 ppm
27.9 ppm
Chambers and his coworkers pointed out that not less than 20% of the reactions occur on benzylic positions? Previous research in our laboratory has also led to the conclusion that benzylic alkylation products were prominent in reductive alkylation reactions.20 Although it is quite im-
200
159
I99
PPM
59
a
I
18C NMR spectra of the Lower Kittanning Seam coal, PSOC-1197(bottom), and the product obtained in the sodium amide catalyzedmethylation reaction with methyl-18Ciodide (top).
Figure 1.
possible to specify the exact reaction sites, the available data are all consistent with an important role for alkylation a t benzylic methylene bridges. Several features of this study imply, as noted previous1y,2*sthat base-catalyzed hydrocarbon elimination reactions also occur under the conditions of our experiments (eq 1). However, the experiments with ammonium chloride reveal that products of these reactions are not soluble. Moreover, we used low concentrations of the base to suppress this bimolecular, molecular weight reducing sidereaction. Thus, while molecular weight reduction may occur, and may be necessary, it is not sufficient, under these conditions, to solubilize this coal. These observations imply that solubilization is realized because the basecatalyzed reactions introduce alkyl groups on the crosslinking and side-chain groups of the aromatic structural elements. Such reactions could place the alkyl groups in strategic positions to disrupt the stacking patterns of the coal. When the alkyl fragment is sufficiently large, the polarization bonding forces between the structural fragments of the coal can be disrupted and the mixture solubilized. In view of the fact that the methyl group has essentially the same steric requirements as a sheet of aromatic molecules, it has little influence on the solubility of the coal. On the other hand, molecular modeling confirms that the larger butyl and octyl groups have much greater steric requirements that cannot be relieved by simple rotations. In this situation, favorable polarization interactions with the solvent exceed those between the molecules in the solid, and solubilization is realized.
Acknowledgment. We thank Francis Flores for his assistance in the conduct of these experiments and Dr. R. R. Chambers for generously sharing unpublished information concerning the alkylation reactions. This research was supported by the U.S. Department of Energy. (20) Willis,
R. S.;Stock, L. M. J. Org. Chem. 1985, 50, 3566-3573.