42
Energy & Fuels 1990, 4, 42-54
activity. The treatment does have an adverse affect on hydrodenitrogenation activity. Alkali-metal and alkaline-earth promotion may therefore be beneficial in applications such as coal liquefaction where the primary function of the catalyst is to hydrogenate and the reaction environment is cqnducive to coking. The performance of these catalysts appears to be simply related to catalyst acidity as evaluated by the TPD of tert-butylamine.
Acknowledgment. This work is jointly sponsored by the US. Department of Energy (Grant DE-FG2288PC88942) and the Amoco Oil Company. We are grateful for both the financial support and consultations provided by these organizations. Registry No. Co, 7440-484; Mo, 7439-907; Li, 7439-93-2;Na, 7440-23-5; K, 7440-09-7; Be, 7440-41-7; Mg, 7439-95-4; Ca, 7440-70-2;Ba, 7440-39-3.
Cross-Linking Reactions during Coal Conversion Peter R. Solomon,* Michael A. Serio, Girish V. Despande,? and Erik Kroo Advanced Fuel Research, Inc., 87 Church Street, East Hartford, Connecticut 06108 Received August 10, 1989. Revised Manuscript Received October 26, 1989
During coal conversion, the breakup of the coal macromolecular network and resulting product formation are controlled by the relative rates of bond breaking, cross-linking, and mass transport. The objective of this work was to systematically study the variations in cross-linking with several parameters (rank, temperature, heating rate, pretreatment, etc.), to identify the factors that control cross-linking, to try to identify the reactions responsible for cross-linking, and to determine the cross-linking rates. This paper describes a study of cross-linking behavior in which chars of a number of coals (including the Argonne premium coal samples) have been pyrolyzed under a variety of temperature histories and analyzed at intermediate extents of pyrolysis for solvent swelling behavior and functional group compositions. The variations in these properties were correlated with the tar molecular weight distribution measured by field ionization mass spectrometry and with the gas evolution. The study of cross-linking as a function of coal rank and pyrolysis temperature shows that there are at least two distinct cross-linking events: one occurs at low temperature prior to tar evolution (in low-rank coals only) and the second occurs at moderate temperatures slightly above that for tar evolution. The low-temperature cross-linking process results in low tar yields, low fluidity (e.g., measured by Geissler plastometer), low extract yields, and low molecular weight tar. Lowtemperature cross-linking is increased by oxidation of the coal and reduced by methylation. Studies that compare char solvent swelling behavior to gas evolution have shown that low-temperature cross-linking occurs simultaneouslywith COPand H 2 0 evolution. Moderate-temperature cross-linking appears to correlate best with methane formation. Studies that compare char swelling behavior to changes in char functional group concentrations have shown that cross-linking reactions occur with the loss of carboxyl groups present in the coal. A clear role for hydroxyl groups in low-temperature cross-linking could not be established, nor could it be ruled out.
Introduction During coal conversion, the breakup of the coal macromolecular network and resulting product formation are controlled by the relative rates of bond breaking, crosslinking, and mass transport. Cross-linking reactions are important in pyrolysis, gasification, and combustion of coal because they control the ultimate tar yield, the tar molecular weight distribution, and the char's fluidity, molecular order, surface area, and reactivity. In liquefaction, cross-linking reactions influence the short contact time yields and the distribution of oils, asphaltenes, and preasphaltenes. The most popular method to follow cross-linking reactions in coal and char has been the measurement of the solvent swelling ratio as described by Green et al.'v2 This method has been employed to determine cross-link density 'Present address: Amoco Performance Products Inc., 12900 Snow Road, Parma, OH 44130.
0887-0624/90/ 2504-0042$02.50/0
changes during pyrolysis*7 and liquefaction.8 The results from pyrolysis studies show that cross-linking is rank dependent, with lignites cross-linking at lower temperatures than bituminous coals.3,41697Cross-linking in lignites occurs prior to tar evolution and the rapid loss of aliphatic hydrogen. Cross-linking in high-volatile bituminous coals occurs at temperatures slightly higher than the tempera(1) Green, T. K.; Kovac, J.; Larsen, J. W. Fuel 1984,63,935. (2)Green, T.K.; Kovac, J.; Larsen, J. W. In Coal Structure; Meyers, R. A., Ed.; Academic Press: New York, 1982;p 199. ( 3 ) Suuberg, E. M.; Lee, D.; Larsen, J. W. Fuel 1985,64, 1668. (4)Suuberg, E. M.; Unger, P. E.; Larsen, J. W. Energy Fuels 1987,1, 305. (5)Suuberg, E. M.; W e , Y.; Deevi, S. Prepr. Pap.-Am. Chem. SOC., Diu.Fuel Chem. 1988,33(1),387. (6)Deshpande, G. V.;Solomon, P. R.; Serio, M. A. Prepr. Pap.-Am. Chem. SOC.,Diu.Fuel Chem. 1988,33(2),310. (7)Solomon, P. R.; Hamblen, D. G.; Carangelo, R. M.; Serio, M. A.; Deshpande, G.V. Energy Fuels 1988,2,405. (8)Bockrath, B. C.; Illig, E. G.; Wassell-Bridger,W. D. Energy Fuels 1987,I , 227.
@ 1990 American Chemical Society
Cross-Linking Reactions during Coal Conversion
ture for maximum tar evolution or alphatic hydrogen loss. Recent research has also demonstrated that substantial reductions in the low-temperature cross-linking reactions can be achieved by ultrarapid heating.g In the case of lignites, the high heating rate conditions produce melting and swelling of char, higher yields of soluble products, and tar molecular weight distributions previously seen only for softening coals. Cross-linking can also be reduced chemically by methylation,6J0 by removal of minerals,6J1 by catalytic reactions,12 by the presence of a solvent! or by the presence of hydrogen? Cross-linking can be increased by oxidation.6 Other properties used to infer cross-linking are the extract yield$f'J3 the fluidity,"16 the fraction of mobile phase determined from measuring NMR relaxation times,17J8the ring cluster size determined by NMR,lg and the tar molecular weight distribution.'JO These measurements also show the rank dependence of cross-linking. Both the extract yield and fluidity of chars are maximum for coals with 83438% carbon and decrease substantially for lower rank coals, presumably due to ~ross-linking.'"~~The mobile phase of a brown coal was shown to decrease at a substantially lower temperature (below that for tar evolution and aliphatic hydrogen loss) than a bituminous coal (slightly above the peak for tar evolution and aliphatic hydrogen loss).18 The ring cluster size determined by NMR is observed to increase at temperatures prior to tar evolution for a lignite, but not for a bituminous ~ 0 a l . l ~ An important aspect in the study of cross-linking reactions is the identification of chemical reactions responsible for cross-linking. In a recent study of cross-linking, Suuberg et al.3*4observed that the low-temperature crosslinking associated with low-rank coals appeared to correlate with the evolution of COP This correlation was confirmed by Solomon and c ~ - w o r k e r s . ~They , ~ * ~also ~ showed that the moderate-temperature cross-linking event seen in bituminous coals appeared to be correlated with CH, evol ~ t i o n . ~It, appears ~.~ that low-temperature cross-linking may be related to the decomposition of carboxyl groups (whose concentration is rank dependent) to form C02, while moderate-temperature cross-linking is related to the release of methyl groups by substitution reactions. The importance of cross-linking is apparent in the model of coal pyrolysis and fluidity being developed in our labo r a t o r y . 6 ~ ~ In J ~ this * ~ model, pyrolysis is described as the decomposition of a macromolecular network. The coal is represented as aromatic ring clusters linked together by bridges. When the coal is heated, the bridges can break (9) Solomon, P. R.; Serio, M. A,; Carangelo, R. M.; Markham, J. R.
Fuel 1986, 65, 182.
(IO) Solomon, P. R.; Squire, K. R. Prepr. Pap.-Am. Chem. SOC.,Diu. Fuel Chem. 1985, 30(4), 346. (11) Tyler, R. J.; Schafer, H. N. S. Fuel 1983,59, 487. (12) Solomon. P. R.: Serio. M. A.: DeshDande. G. V.: Kroo.' E. PreDr. Pap.-Am. Chem. Soc:, Diu.'Fuel Chem. i989,34(3), 803. (13) Fong, W. S.; Peters, W. A.; Howard, J. B. Fuel 1986, 65, 251. (14)Whitehurst, D. D.; Mitchell, T. 0.; Farcasiu, M. Coal Liquefaction, The Chemistry and Technology of Thermal Processes; Academic
Press: New York, 1980. (15) Whitehurst, D. D. In Coal Liquefaction Fundamentals; Whitehurst, D. D., Ed.; ACS Symposium Series 139; American Chemical Society: Washington, DC, 1980; p 133. (16) Solomon, P. R.; Best, P. E.; Yu, Z. Z.; Deshpande, G . V. Prepr. Pap.-Am. Chem. SOC.,Diu. Fuel Chem. 1989,34(3),895. (17) Sakurovs, R.; Lynch, L.J.; Maher, P.; Banerjee, R. N. Energy Fuels 1987, I , 167. (18) Lynch, L. J.; Webster, D. S.; Sakurovs, R.; Barton, W. A,; Maher, T. P. Fuel 1988,67,579. (19) Solum, M. S.; Pugmire, R. J.; Grant, D. M.; Fletcher, T. H.; Solomon, P. R. Studies of Coal Char Structure Euolution: Solid State I3C NMR, in press. (20) Solomon,P. R.; Hamblen, D. G.; Carangelo, R. M.; Serio, M. A.; Deshpande, G . V. Combust. Flame 1988, 71, 137.
Energy & Fuels, Vol. 4, No. 1, 1990 43 Table I. Ultimate Analyses of Coals Used (% daf) Zap North Dakota lignite Big Brown Texas lignite Montana Rosebud subbituminous Wellmore Kentucky No. 8 bituminous Pittsburgh Seam No. 8 bituminous Pittsburgh Seam (high ash) Pittsburgh Seam (high ash; 239 h oxidized) Pittsburgh Seam (Argonne) Pittsburgh Seam (Argonne sample; 328.5 h oxidized)
C
H
66.5 72.4 72.1 86.0 82.1 77.7 73.5
4.8 5.5 4.7 5.4 5.6 5.5 5.0
S
O ash"
1.1 26.5 7.1 1.2 19.2 12.9 1.2 20.3 10.0 1.2 5.9 7.0
2.4
8.2 4.9 12.6 12.0 19.8 12.0
83.2 5.3 0.9 8.8 9.3 78.5 4.9 15.3 12.3
Dry basis.
and new cross-links can form. Statistical methods are employed to predict the concentration of single aromatic clusters (monomers) and linked clusters (oligomers of n clusters, "n-mers") up to a totally linked network. These concentrations depend primarily on the number of bridges and cross-links per cluster. By assigning molecular weights to the monomers, the fractions of tar,extractables, liquids, or char can then be defined from the distribution of oligomer sizes. These fractions are thus very sensitive to the number of cross-links added during pyrolysis. More cross-links lead to lower yields of tar, extractables, and liquids and a lower average tar molecular weight. The assumption that one cross-link is formed per C02 or CH, evolved agrees reasonably well with the pyrolysis data. If one cross-link is formed for each C02evolved from a lignite (which has one carboxyl group for every two monomers), enough low-temperature cross-links are added to completely join all the monomers, resulting in zero tar,extract, or liquid fractions. The fluidity of a char is related to its total liquid fraction, and hence, it is also reduced with increased cross-linking.16 The objective of this work was to systematically study the variations in cross-linking with various parameters (rank, temperature, heating rate, pretreatment, etc.), to try to identify the reactions responsible for cross-linking, and to determine the cross-linking rates. This paper describes a study of cross-linking behavior in which chars of a number of coals have been pyrolyzed under a variety of temperature histories and analyzed at intermediate extents of pyrolysis for solvent swelling behavior, functional group compositions, and gas evolution. The observed results support the previous o b s e r ~ a t i o n s ~ ~that ~ J ~ Jlow-temper" ature cross-linking is related to the decomposition of carboxyl groups to form C02 (but also, possibly, to reactions involving hydroxyl groups) and moderate-temperature cross-linking appears to be related to the evolution of methyl groups (and, possibly, other peripheral groups).
Experimental Section Coals Examined. Experiments were done with North Dakota (Zap) lignite, Big Brown Texas lignite, Wellmore Kentucky No. 8 bituminous, and Pittsburgh Seam (high-ash) bituminous coals. The ultimate analysis of each coal is given in Table I. Experiments were also done with the Argonne premium coal samples. Analyses for these samples can be found elsewhere.21 Coal Modification. Coals were demineralized according to the technique of Bishop and Ward.22 Coals were also ion exchanged according to the technique outlined by Hengel and Walker23to exchange cations with ammonium ions or load the coal with a certain cation (e.g., calcium) in different amounts. The reaction times were varied from those in ref 23. Experiments were (21) Vorres, K. S. Prepr. Pap.-Am. Chem. SOC.,Diu. Fuel Chem. 1987, 32(4), 221. (22) Bishop, M.; Ward, D. L. Fuel 1958,37, 191. (23) Hengel, T. D.; Walker, P. L.Fuel 1984, 63, 1214.
Solomon et al.
44 Energy & Fuels, Vol. 4, No. 1, 1990 also performed with a perdeuteriomethylated Big Brown Texas lignite provided by Ron Liotta of Exxon, a Pittsburgh Seam (high-ash) bituminous coal that had been oxidized in air a t 110 "C, and a Pittsburgh Seam (Argonne) bituminous coal that had been oxidized in air at 100 "C. Char Preparation. The chars were formed by rapid heating in an inert gas with either an entrained-flow reactor (EFR) or a heated-tube reactor (HTR) for a variety of residence times and maximum temperatures. Samples were studied under slow heating conditions in a thermogravimetric analyzer (TGA) where the volatiles are analyzed on-line by Fourier transform infrared (FT-IR) spectroscopy (TG-FTIR) and in a field ionization mass spectrometer (FIMS) where the pyrolysis tars are analyzed during pyrolysis. HTR. The HTR has been described previously?*" It consists of a 5.08-mm4.d. Inconel 702 tube, which is heated electrically. Coal entrained in cold carrier gas is injected into the heated section of the tube. After a variable residence time, the reacting stream is quenched and the products are separated and analyzed. Heating rates of up to 20000 OC/s, temperatures up to 950 "C, and residence times between 10 and 150 ms have been achieved in this apparatus. EFR. The EFR has also been discussed previ0usly.2~~~~ It consists of a heat exchanger and test section contained in a fumace. An inert gas stream (N2or He) is preheated during transit through the heat exchanger, turns through a U-tube, and enters a 5-cm-diameter test section. Coal is introduced into the test section a t variable positions through a movable water-cooled injector. The apparatus provides temperatures up to 1600 "C, heating rates up to 5000 "C/s, and residence times up to 1 s. TG-FTIR. TG-FTIR analysis was performed with the TG/ Plus manufactured by Bomem, Inc. The instrument components are DUPONT 951 TGA; a hardware interface (including the furnace power supply); an Infrared Analysis 16 pass gas cell with transfer optics; and MICHELSON 100 FT-IR(resolution 4 cm-I). The instrument analyzes a sample suspended on a balance in a heated helium gas stream within a furnace. As the sample is heated, the evolving tars and gases are carried out of the furnace directly into a gas cell for analysis by FT-IR spectroscopy. The sample temperature is determined by a thermocouple located next to the sample. In the current study, most of the samples were heated at 30 "C/min (0.5 "C/s) to a fiial temperature and rapidly cooled. A more detailed description of the apparatus can be found el~ewhere.~~~*' FIMS. Molecular weight distributions of tars produced in pyrolysis were obtained at SRI International using the field ionization mass spectrometry (FIMS) apparatus described by St. John et a1.% The coal samples were pyrolyzed directly in the FIMS apparatus a t a heating rate of 3 *C/min to final temperatures between 450 and 500 O C . The FIMS technique produces little fragmentation of the evolved tars and so provides a good determination of the tar molecular weight distribution. Coal and Char Characterization. The samples were analyzed according to the solvent swelling techniques developed by Green et al.'Z to quantify and monitor the densities of covalent cross-links inside the macromolecular network. Dried coals and chars were swelled in pyridine to determine Q, the ratio of the volume of swollen coal in equilibrium in the solvent (1-5 days) to the volume of the dry coal. As pyrolysis proceeds, values of Q vary, from more than 2 for the dry, raw coal to values approaching 1 (i.e., no swelling) in a high-temperature char. The coals and chars were also analyzed by quantitative FT-IR spectroscopy according to the KBr pellet (24) Serio, M. A.; Hamblen, D. G.; Markham, J. R.; Solomon, P. R. Energy Fuels 1987, I , 138. (25) Solomon, P. R.; Hamblen, D. G.; Carangelo, R. M.; Krause, J. L. 19th Symposium (International)on Combustion; The Combustion In-
stitute: Pittsburgh, PA, 1982; p 1139. (26) Carangelo, R. M.; Solomon, P. R.; Cerson, D. J. Fuel 1987,66960. (27) Whelan, J. K.; Solomon, P. R.; Deshpande, G. V.; Carangelo, R. M. Energy Fuels 1988, 2, 65. (28) St. John, G. A.; Buttrill, S. E., Jr.; Anbar, M. In Organic Chemistry of Coal; Larsen, J., Ed.; ACS Symposium Series 71; American Chemical Society: Washington, DC, 1978; p 223. (29) Solomon, P. R.; Carangelo, R. M. Fuel 1982, 61, 663. (30) Solomon, P. R.; Carangelo, R. M. Fuel 1988, 67, 949.
X A
"
0
200
400
600
800
Temperature ("C)
Figure 1. Comparison of solvent swelling ratios for coals of various ranks at a series of fiial pyrolysis temperatures at a heating rate of 30 OC/min to final temperature. The uncertainty in using measurements of Q to infer cross-link density changes is that there are two factors which could change Q, variation in the coal-solvent interaction parameter x due to changes in the functional group composition and variations in the cross-link density. Previous researchers have also assumed that it is the latter which is the strongest influence for the change in Q in pyrolysis and have, therefore, assumed that the variation in Q can be used as a qualitative index of the extent of crosslinkig.3p4v8The following observations support this contention. (1)While x is difficult to measure or estimate, various researchers have reported (i) no variation for bituminous coals>1 (ii) an inand (iii) crease of 20% between 65% carbon and 88% an increase of 30% between 69% and 82% carbon.% (2) Among the Argonne coals in this rank range, these small variations in x occur for large differences in the functional group composition. In particular, the polar functional groups that might influence the solvent-coal interaction parameter for pyridine vary by substantial amounts for these coals. (3) The functional group changes that accompany reduced Q in pyrolysis are, in many cases, smaller than the variations among coals. This is true even for the polar functional groups. (4) If it is assumed that the variations in x in chars are in the same range as for the coals &e., less than 30%),then the observed variations in Q must be strongly influenced by variations in the cross-link density since 30% changes in x cannot produce the large observed changes in Q using any of the current solvent swelling theories."~~~
Results and Discussion Observations of Cross-Linking in Coal Pyrolysis. T h i s section considers how cross-linking varies with rank, oxidation, mineral content, chemical modifications, a n d the presence of solvents. T h e properties examined include t h e solvent swelling properties of char, t h e t a r yield a n d molecular weight distribution, a n d t h e char fluidity, a s these properties are all affected by cross-linking. Rank Dependence of Cross-Linking Reactions. T o s t u d y t h e variations in cross-linking with rank, samples were heated at 0.5 O C / s t o temperatures between 200 a n d 900 "C, a n d t h e char samples were analyzed. Figure 1 shows the volumetric swelling ratio, Q, for Zap lignite, Big Brown Texas lignites, Wyodak subbituminous, Wellmore Kentucky No. 8 bituminous, a n d Pittsburgh S e a m (Argonne sample) bituminous coals. T h e swelling ratio profiles are presented as 1- X,where X is t h e change in t h e solvent swelling ratio between the coal and char normalized by t h e maximum change. T h a t is
(31) Green, T.; Kovac, J.; Brenner, D.; Larsen, J. In Coal Structure; Myers, R. A., Ed.; Academic Press: New York, 1982; Chapter 6, p 199. (32) Sanada, Y.; Honda, H. Fuel 1966,45, 295. (33) Lucht, L. M.; Peppas, N. A. Fuel 1987,66,803. (34) Larsen, J. W.; Kovac, J. ACS Symp. Ser. 1978, 71, 36. ' (35) Lucht, L. M.; Peppas, N. A. ACS Symp. Ser. 1981,169,43.
Cross-Linking Reactions during Coal Conversion
Energy & Fuels, Vol. 4, No. 1, 1990 45
Table 11. Summary of Swelling Experiments for Coals and Chars heated at 30 "C/min to 400 "C (Argonne dmmf, Others daf) oxygen max fluidity coal samdes Qm/Qcoal DDPM (dao Q C d D Q40Ob 2 1.12 (1.12) Pocahontas (Argonne) 1.23 1.10 44 Upper Freeport (Argonne) 1.32 (1.32) 5 2.13 1.61 >30 000 Kentucky No. 8 6 1.80 2.10 1.17 7 Pittsburgh Seam (Argonne) 2.30 (2.33, 2.26) 2.37 1.03 29 000 7 Lewiston-Stockton (Argonne) 2.18 (2.35, 2.18) 1.94 0.89 Illinois (Argonne) 2.53 (2.53, 2.53) 10 2.03 0.80 2.74 (2.76, 2.66, 2.74) Blind Canyon (Argonne) 11 0.69 3 1.90 2.70 (2.73) Wyodak (Argonne) 17 2.00 0.74 86 Zap Lignite (Argonne) 2.71 i2.7ij 1.47 19 0.54 Big Brown (Exxon) 20 2.23 0.77 1.71 Zap Lignite (UNDERC) 0.61 1.45c 27 2.38 Zap lignite (UNDERC) ion exchanged with ammonium acetate demineralized Zap lignite (UNDERC) demineralized Upper Freeport perdeuteriom ethylated Big Brown a Volumetric
Altered Coals 2.13 2.52 1.80 1.87
1.28
0.60
1.68 2.16 1.60
0.67 1.20 0.86
swelling ratio for coal. bVolumetricswelling ratio for char at 400 OC. Interpolated (duplicate runs).
Thus, 1 - X = 1 corresponds to the coal and 1 - X = 0 corresponds to fully cross-linked char. The curves are quite different among the five samples. For lignites, cross-linking begins at 200 "C, a much lower temperature than for the bituminous coals. The bituminous coals, on the other hand, first show an increase in 1 - X and do not exhibit a decrease in 1- X (an increase in cross-linking) until 400 "C. The increase in 1 - X is believed to be due to covalent bond breaking reactions, as this is the temperature range where tar evolution, fluidity, and rapid liquefaction are observed. Suuberg et a L 3 s 4 observed similar results but at a higher temperature, consistent with a higher heating rate of 1000 "C/s. The Wyodak subbituminous coal shows intermediate behavior, where 1- X first decreases at low temperature and then increases slightly near 400 "C followed by a decrease at higher temperatures. The results suggest that there are at least two distinct cross-linking events, one at low temperature prior to bridge-breaking reactions (depolymerization) exhibited by low-rank coals, and one at moderate temperatures subsequent to the initial bridgebreaking reactions for higher rank coals. To survey this rank dependence,the remaining Argonne coals were subjected to pyrolysis at 0.5 "C/s to 400 "C and the chars were analyzed. The temperature 400 "C was chosen since, as shown in Figure 1,the largest spread in behavior is exhibited in this range. The results are presented in Table 11together with data for the coals in Figure 1. The coals are listed in order of decreasing rank. T h e variation in Q for duplicate runs with fresh coal samples is typically less t h a n f0.1. The four highest rank coals all show an increase in Q upon heating to 400 "C, while the lowest seven all show a decrease. Clearly, the lowtemperature cross-linking varies systematically with rank. Table I1 also includes fluidity data for the Argonne coals. These data for the maximum fluidity at a heating rate of 3 "C/min were supplied by Commercial Testing and Engineering (Lombard, IL). Ignoring the Pocahontas, the high-rank coals have high fluidities and the maximum fluidity drops with decreasing rank. A recently developed fluidity model attributes this drop almost entirely to the low-temperature cross-linking process.16 As mentioned in the Introduction, cross-linking also affects the molecular weight distribution of tar in pyrolysis.'J0 FIMS analyses for the eight Argonne coals pyrolyzed at 3 "C/min in the FIMS apparatus to 450 "C are shown in Figure 2. The spectra show a distinct progression from low to high rank euen though t h e average aro-
0.60
Utah Blind Canyon e
o.40100 200 300 400 500 600 700 800
I
UpperFreeport b1 o*60[
I
IllinoisNo.6 f
I
Figure 2. Tar molecular weight distributions from pyrolysisFIMS of the Argonne premium coals. The samples were heated a t 3 OC/min into the inlet of the mass spectrometer. The final temperature was 450 "C, except for Pocahontas, Illinois No. 6, and Zap, where the final temperature was 500 "C.
matic ring cluster size measured by NMR remains relatively constant (277 a m u for Zap, 299 a m u for Pocahontas).% The highest rank coals, Pocahontas (Figure 2a) and Upper Freeport (Figure 2b) both show low intensities at low molecular weights (100-200 amu) and high intensity in the 200-600 amu range. The intermediate-rank coals, Pittsburgh (Figure 2c), Lewiston-Stockton (Figure 2d), Utah (Figure 2e), and 11linois (Figure 2 0 all have similar molecular weight distributions showing substantial intensities in the 100-200 amu region as well as in the 200-600 amu range. These spectra extend up to high molecular weights, limited only by the vapor pressure of the oligomers formed in pyrolyThat is, oligomers of all sizes are formed in the (36)Solum, M. S.; Pugmire, R. J.; Grant, D. M. Energy Fuels 1989, 3, 181. (37) Solomon, P. R.; King, H.H.Fuel 1984, 63, 1302.
46
Solomon et al.
Energy & Fuels, Vol. 4, No. 1, 1990 o n
1.5
X
pitts Seam (Argonne,oxidized) Ca Exchanged pins (Argonne,oxidized)
A pins Seam (Argonne)
I
0.5
il
l 0.0 0 200S 400 600 800 . . 1000 0
400
Temperature ("(3)
Figure 4. Effect of oxidation and minerals on cross-linking behavior of Pittsburgh Seam bituminous coal (Argonne). Heating rate was 0.5 OC/s to final temperature.
I!I L
I
200
I
I
I
I
I
I
,
800 1000 1200 1400 Mass 0
400 600
Figure 3. Comparison of FIMS spectra for (a) non-cross-linking and (b and c) cross-linking polymers.
pyrolyzing coal and those with sufficiently high vapor pressure escape as tar. In contrast, the low-rank coals, Wyodak (Figure 2g) and Zap (Figure 2h), show high intensity between 100 and 200 amu, but substantially lower intensity above 200 amu. This has been explained by extensive cross-linking in low-rank c o a l ~ . ~This J ~ cross-linking limits the presence of higher molecular weight volatile oligomers, thus reducing the intensity above 200 amu. A network model for coal pyrolysis which includes weak bridge breaking, low-temperature cross-linking, moderate-temperature cross-linking, and mass transport exhibits these trends with the variation in the amount of low-temperature cross-linking, even though the average aromatic ring cluster size remains c o n ~ t a n t . These ~ ? ~ results show that increased low-temperature cross-linking is sufficient to cause the observed variation in the molecular weight distribution. It is not, however, the only possible explanation. The change could be caused by the lack of plasticity in low-rank coals which limits the mobility of large molecules, but the loss in plasticity appears to be due to the low-temperature cross-linking in any case.16 Model polymers have been used in our laboratory to study the depolymerization and cross-linking reactions believed to occur in coal.10,37~38 For softening coals (little low-temperature cross-linking), polymers with ethylene bridges between unsubstituted aromatic ring clusters were used. The use of this type of polymer allowed the combined bridge-breaking process and evaporation of light fragments (to form tar) to be examined in a well-characterized material. The tar formation from these polymers (38) Squire, K. R.; Carangelo, R. M.; DiTaranto, M. B.; Solomon, P. R. Fuel 1986,65, 833.
has a number of similarities with pyrolysis of bituminous coals. The molecular weight distribution for a naphthalene polymer in Figure 3a shows the same trend of relatively constant intensity for all oligomers up to the vaporization limit. The spectrum, however, extends to higher masses than for melting coals, as expected, due to the higher vapor pressure of the pure hydrocarbon oligomers and slightly higher pyrolysis temperature. A polymer with methoxyl groups on the ring clusters was used to represent low-rank coals. The product molecular weight distribution in Figure 3c for this polymer does show the sharp drop-off in molecular weight due to cross-linking exhibited by low-rank coals (Figure 2g,h). It is interesting to note that this polymer showed a molecular weight distribution (Figure 3b) characteristic of a non-crosslinking material (e.g., Figure 3a) at a lower temperature. The shape changed at higher temperatures, presumably as the cross-linking reactions started. Figure 3 clearly demonstrates that the molecular weight distribution is not related to cluster size, but is in fact a distribution of oligomers whose relative amounts can change with crosslinking. Effect of Oxidation on Cross-Linking Reactions. To further study the role of oxygen in low-temperature cross-linking, a study was made using an oxidized Pittsburgh Seam bituminous coal. It is well-known that weathering of coals (exposure to air even a t room temperature for prolonged periods of time) can cause complete disappearance of both caking and dilation properties.3w1 Several investigator^^^^^^ have carried out mild oxidation of coal under controlled conditions and found a decrease of liquid hydrocarbon yield. This is generally attributed to the formation of oxygen-containing functional groups that on heating convert or participate in the formation of cross-links within the cod macromolecule. An increase in the amount of CO and COz evolution is also usually observed with an increase in the extent of oxidation. Chars of an Argonne Pittsburgh Seam oxidized coal (329 h in air a t 100 "C) were formed a t temperatures between 200 and 900 "C. The solvent swelling data are compared to the data for an unoxidized Argonne Pittsburgh Seam coal in Figure 4. The oxidized coal shows earlier crosslinking, as seen by the decrease in the value of 1 - X at (39) Ignasiak, B. S.; Clugston, D. M.; Montgomery, D. S. Fuel 1976, 51, 76. (40) Mahajan, 0. P.; Komatsu, M.; Walker, P. L., Jr. Fuel 1980, 59, 3.
(41) Maloney, D. J.; Jenkins, R. G. Fuel 1985, 64, 1415. (42) Ignasiak, B. S.; Nandi, B. N.; Montgomery,D. S. Fuel 1970,49, 214. (43) Furmisky, E.; MacPhee, J. A.; Vancea, L.; Ciavaglia, L. A.; Nandi, B. N. Fuel 1983,62, 395.
Cross-Linking Reactions during Coal Conversion 0.4
Energy & Fuels, Vol. 4, No. 1, 1990 47
a
IO0 200 300 400 500 600 700 SO0 Mass (WZ) 0.6
2x10'~sec
I
I
100 200 300
400 500 600 700 800 Mass (WZ)
Figure 5. Molecular weight distribution of tar from Pittaburgh Seam bituminous coal (high ash) determined by FIMS. (a)Fresh coal, (b) 13-h oxidation, and (c) 504-h oxidation.
temperatures above 300 "C. The behavior is similar to what is seen in the Wyodak coal in Figure 1. A comparison of fresh and oxidized Pittsburgh Seam (high-ash) coal was made in an earlier Samples of the coal were oxidized at 110 "C for 13 and 504 h resulting in changes in the oxygen concentration from 10 to 14 and 187'0, respectively. Increases were seen in the hydroxyl, carboxyl, and C-0 regions of the infrared spectra. During pyrolysis, the oxidized coals showed lower fluidity as evidenced by the swelling in the char and lower tar yields. In addition, the tar molecular weight distribution changed with oxidation. Figure 5 shows a shift to smaller species with increased oxidation time as can be seen in the ratio of amplitudes of the 366 amu peak to the 106 amu peak (1.05 for the fresh coal, 0.75 for the 13-h coal, and 0.30 for the 504-h coal). The FIMS spectra for the oxidized bituminous coals are more characteristic of lower rank, less fluid coals. Heating Rate Dependence of Cross-Linking Reactions. In a previous study of coal pyrolysis'O it was noted that the FIMS spectra of tar from a Zap lignite heated very rapidly (20000 OC/s) in the HTR resembled the spectrum of a bituminous coal. Parts a and b of Figure 6 are spectra of tars collected at heating rates of 3 "C/min in the FIMS and 600 "C/s in a heated grid reactor, respectively. Figure 6c is a FIMS spectrum of a tar collected in the HTR at a heating rate of 20000 "C/S. The effect of higher heating rates is to produce tars from lignites that have higher average molecular weights, like the FIMS spectra from higher rank coals (see Figure 2). It was also noted that the chars produced at the high heating rate melted and were ~wollen.~ High heating rates also increased the tar yield. A maximum tar plus hydrocarbon yield of -20% was obtained for a North Dakota lignite at a coal particle temperature of 800 "C and nominal (44) Solomon, P. R.; Hamblen, D. G.; Markham, J. R.; Best, P. E. 'An Investigationof Vaporization and Devolatilizationof Coal/ Water Fuels"; final report for DOE/PETC Contract No. DE-AC22-82PC50254;1984.
Mass W)
Figure 6. Comparison of tar molecular weight distributions for a Zap lignite (UNDERC)determined by FIMS for tar collected
at three different heating rates.
3G 10s
E
B
5-
0
Lil
I A L-f! T I I 100 200 300 400 5&
I
600
700 800 900
Temperature ("0
Figure 7. Effect of temperature on tar yield and pyridine swelling ratio of char for rapid pyrolysis of Zap lignite (UNDERC) in carbon dioxide (solid line) and slow pyrolysis of Zap lignite in helium (dashed line).
gas (helium) residence time of 12 ms in the HTR. A yield of 6% has been obtained for the same coal24at a heating rate of 0.5 OC/s in the TG-FTIR reactor. It, therefore, appeared that rapid heating was reducing the effect of low-temperature cross-linking. To test this hypothesis, the effect of heating rate on the volumetric swelling ratio is compared in Figure 7 for rapid pyrolysis (20000 "C/s) and slow pyrolysis (0.5 "C/s). The tar yield is also presented for reference. It can be clearly seen that the lignite starts to cross-link prior to the tar evolution in slow pyrolysis,whereas cross-linkingoccurs simultaneously with tar evolution in rapid pyrolysis. High heating rate, therefore, reduces the relative rate of cross-linking compared to that for bridge breaking, presumably because of the higher activation energy of the latter. Simulations using the network model discussed previously' are in quantitative agreement with the observed behavior when
"1
48 Energy & Fuels, Vol. 4, No. 1, 1990
x
1.2
Solomon et al.
+ Zap Lignite (UNDERC) ignite (UNDER)
X Zap
Ion-Exchanged
0 Demineralized
Zap
Lignite (UNDERC)
Figure 8. Effect of removal of Ca by ion exchange or demineralization on cross-linkingbehavior of Zap lignite (UNDERC). Heating rate was 0.5 O C / s to final temperature.
the rate of bridge breaking has a higher activation energy than the rate of cross-linking. Effects of Minerals on Cross-Linking Reactions. Tyler and Schafer" found that the removal of cations from Gelliondale coals increased the yields of both tar and total volatile matter. The converse was also observed in that the addition of Ca2+ions to an acid-form coal reduced the yields. The reversibility of the phenomena shows that the increase in yields on cation removal is not the result of any permanent chemical modification or degradation of the organic structure of the coal during the acid treatment. Since tar yield is related to the cross-linking behavior, a study of the effect of minerals on cross-linking reactions was undertaken. As calcium has been the cation most extensivelystudied and low-rank coals have a large amount of calcium cations attached to the carboxylate anions, calcium was the focus of the study. Indian Head Zap lignite which contains 1.66% calcium was used. Samples of raw, ion-exchanged, and demineralized Zap lignites (UNDERC) were pyrolyzed in the TG-FTIR apparatus. The swelling ratio profiles are compared in Figure 8. The cross-linking behavior for ion-exchanged Zap lignite is similar to the raw lignite and shows early crosslinking. It should be noted here that the calcium has not totally been removed by the ion-exchange technique in this study. The calcium content after ion exchange was reduced from 1.66 to 0.2%. The calcium content after demineralization was decreased to 0.01%. The cross-linking behavior of demineralized Zap lignite is different from that of raw Zap lignite. It shows a shift to a higher temperature for the same loss in swelling due to cross-linkingreactions. The tar evolved was higher (daf basis) and the CO and COz evolutions were lower for the demineralized case. An increase in tar evolution is expected due to reduced cross-linking reactions. When pyrolysis of the two lignites was carried out at high heating rate (5000 K/s) in the entrained-flow reactor, the tar yield was increased from 10% (for 0.5 "C/s) to 14% (for raw lignite) and from 12% (for 0.5 "C/s) to 22% for the demineralized lignite. It should be noted that the increase in tar yield due to the removal of calcium found by Tyler and Schafer'l was for coals whose cations were removed by acid treatment. The possibility exists that this treatment results in some depolymerization reactions. To test this possibility, both the lignite and the acid-demineralized lignites were exhaustively extracted by pyridine in a Soxhlet apparatus to determine whether any depolymerization of lignite had occurred. The amount of pyridine-extractable material was 0% for the raw lignite and 2.6% for the demineralized lignite. The volumetric swelling ratio in pyridine was 2.4 for the raw lignite and 2.5 for the demineralized lignite. Both results show that a very small degree of depolym-
erization had occurred, if any, due to the demineralization procedure employed to remove the cations. Hence the increase in tar amount and shift of the cross-linking curve to a higher temperature for the demineralized Zap lignite appears to be due to the reduced mineral content. To study further the role of calcium cations on crosslinking behavior, calcium was added to a Pittsburgh Seam coal. Pittsburgh Seam (Argonne) coal was first oxidized in air at 100 "C for 329 h. A weight gain of 4.6% was observed. The chemical analyses of the raw coals and the oxidized coals are compared in Table I. As expected, the carbon and hydrogen contents were decreased and the oxygen content was increased due to oxidation. A slightly modified procedure of Hengel and Walkerz3was then employed to ion exchange calcium onto the coal. The elemental composition of the minerals in the Pittsburgh Seam coal and the calcium-exchanged oxidized Pittsburgh Seam coal were characterized by a scanning electron microscope (SEM) with dispersive energy X-ray analysis. The added calcium was no more than 0.04%. The swelling ratio profiles are compared in Figure 4 for Pittsburgh Seam (Argonne), oxidized Pittsburgh Seam (Argonne), and calcium-exchanged oxidized Pittsburgh Seam (Argonne) coals. The cross-linking behavior for the calcium-exchanged oxidized Pittsburgh Seam bituminous coal shows early cross-linking similar to the raw lignite. It is surprising to see this much change in the cross-linking behavior with so little calcium exchange. Even if it is assumed that one cross-link is introduced per calcium atom, the low amount of calcium added is not enough to introduce significant numbers of additional cross-links for the calcium-exchanged coal. It is possible that the calcium may be acting catalytically for the cross-linking reaction. We recently became aware of some work by Ignasiak et al.42where a high-volatile bituminous coal was oxidized. It was found that exchange with barium cations restored the loss in fluidity that was observed in pyrolysis experiments with the oxidized coal. This is a surprising result since it would suggest that barium inhibits cross-linking, the opposite of what is believed for calcium. Consequently, it would appear that the mineral effects observed with calcium are not generalizable. Chemical Modification of Cross-LinkingReactions. To study the relationship of low-temperature cross-linking in lignites to oxygen functional groups, measurements were made on methylated coals. The methylation modifies the hydroxyl and carboxyl groups. We have studied the pyrolysis of Big Brown Texas and perdeuteriomethylated Big Brown Texas lignites carried out in the TG-FTIR. Under pyrolysis conditions, the methylated coal shows higher fluidity and tar yield (14% vs 10%) compared to the parent coal. The effect of the methylation is also on the molecular weight distribution of the tar. Figure 9 compares the FIMS spectra for raw and perdeuteriomethylated Big Brown Texas lignites. The average molecular weight is substantially increased from that of the methylated coal and now resembles that of a bituminous coal. Clearly, modification of the hydroxyl and carboxyl groups affects low-temperature cross-linking, implicating these groups in the cross-linking chemistry. Effect of Solvents on Cross-Linking. In order to assess the influence of the liquefaction solvent on the cross-linking of Zap lignite, chars were made in the presence of tetralin or helium at different temperatures. The coal was heated in a sealed electrically heated tube reactor to the desired temperature over a 120-s interval and the temperature was held for 3 min. The amount of solvent in the tube and the experimental geometry ensured that
Cross- Linking Reactions during Coal Conversion
Energy & Fuels, Vol. 4, No. 1, 1990 49
-
w o n ofLow
Region of Low Temp. Crosslinking
Temp. Crosslinking
*
-
g
-
Time (minutes)
Time (minutes)
2.5
12.0
8
9.0
f
1.0 1.6 0.0
.oo 0
8
16
24
32
0.0
8
40
24
32
40
Time (minutes)
Time (minutes)
3
16
U .08 0.3
Mass (M/Z)
o.o!, . 0
Figure 9. Comparison of FIMS spectra for raw and perdeuteriomethylated Big Brown Texas lignite.
1
,,
8
Timr (minutes)
~{d.. ' . 32' . 40L!0.0 16 24 7
'
Time (minutes)
Figure 11. TG-FTIR analysis of Zap lignite (Argonne) for py-
- ,x Solvent Present - - - - ,A SolventAbsent
rolysis a t 0.5 "C/s. The region where low-temperature crosslinking occurs is shaded.
100
OS5
1
800
800 400
0.0 0
200
400
Tempera-
600
800
200
1000
("C)
0
0
8
16
32
24
Time (minuted
40
00
Figure 10. Effect of solvent on cross-linking behavior of Zap
15.0
lignite (Argonne). Heating period was 120 s; isothermal period a t maximum temperature was 180 s.
19.0
the solvent remained in the liquid state and in contact with the coal. The results of these experiments are shown in Figure 10. The solvent swelling data indicate that the loss in swelling is lower when the solvent is present during pyrolysis at all temperatures. It is speculated that the presence of solvent may reduce the extent of the crosslinking reactions by capping some of the cross-link sites with hydrogen from the donor-solventtetralin. The results are consistent with the findings of Bockrath et a1.8 Chemistry of Cross-Linking. The study of crosslinking behavior described above indicates that there are two distinct cross-link processes which occur at different rates and which vary in importance with rank. Since it is likely that the cross-linking reactions will be accompanied by observable chemical changes in the coal, a study was undertaken to identify either gas evolution or functional group changes which can be correlated with crosslinking (i.e., have similar kinetic rates and variations with rank). Correlation of Cross-Linking with Gas Evolution during Pyrolysis. To determine whether cross-linking might correlate with the evolution of some gas species, the solvent swelling data of Figure 1 was compared to gas evolution profiles. The Zap lignite and Pittsburgh Seam coal were picked as representing the cases where low-temperature cross-linking and moderate-temperature crosslinking dominated, respectively. The results for Zap lignite are presented in Figure 11. The weight loss and .temperature are presented as a function of time in Figure lla. The rate of evolution and
_ _ _ - - - 6.0 0.0 0
8
16
24
32
40
Time (minutes)
0
8 16 24 32 Time (minuted
Time (minuted
40
0
8
16
24
32
40
Time (minuted
Figure 12. TG-FTIR analysis of Pittsburgh Seam bituminous
coal (Argonne) for pyrolysis a t 0.5 "C/s. (-) Raw coal and (- - -) oxidized coal (329 h at 100 "C). The region where low-temperature cross-linking occurs is lightly shaded and the region where moderate-temperature cross-linking occurs is darkly shaded (see text).
the integrated yield are presented for tar, CHI, H20, C02, and CO in Figure llb-f. The region where cross-linking occurs is shown shaded. Clearly, it is the oxygen-containing gases H20, C02, and CO which evolve during crosslinking and the start of C 0 2 and H 2 0 evolution coincides most closely with the start of cross-linking. Figure 12 shows results for Pittsburgh Seam coal (Argonne), both fresh (solid lines) and oxidized (329 h, dashed lines). The regions for both low-temperaturecross-linking exhibited by the oxidized sample and the moderate-temperature cross-linking that dominates the fresh sample are shown by light and darker shading, respectively. There is little gas evolution in the low-temperature cross-linking regions for the fresh sample, which does not undergo sig-
50 Energy & Fuels, Vol. 4, No. 1, 1990
Solomon et al.
0.8
0.6
v
8.0
6.4 $
0 MontanaRosebud
0.4
SubbituminousCod, HTR
4.8 8 1
0
8
16 24 3240 Time (minutes)
0
8
16
24
52
40
3.2
B
1.6
[
0.0
8
2 B
F
.oo 0
A~ap~ignite,~~~ - % C02 YieW44
0.2
-
0.0
0
.20
16 21 Sa40 Tim0 (minub.)
8
Figure 13. TG-FTIR analysis of Big Brown Texas lignite for pyrolysis at 0.5 "C/s. (-) Perdeuteriomethylatedcoal and (-- -) raw coal. The region where low-temperaturecross-linking occurs is shaded.
nificant low-temperature cross-linking. The oxidized sample, however, shows significant increases in all three oxygen-containing gases over the low-temperature crosslinking region. The fresh sample undergoes cross-linking in the darker shaded region where CH, and tar evolution are observed. There are also other gases such as ethane, propane, and larger paraffins evolved in this region. Gas evolution profiles were also compared for the raw Big Brown Texas and perdeuteriomethylated Big Brown Texas lignites as shown in Figure 13. The untreated sample data are shown as dashed lines, and the perdeuteriomethylated sample data are shown as solid lines. It can be seen that, during the pyrolysis stage, the initial amounts of C02, CO, and H 2 0 evolved are less for the perdeuteriomethylated coal, while the amount of CH4 evolved is slightly more. For the untreated sample, which exhibits early cross-linking, the largest gas yield is C02. A more comprehensive correlation with gas yields was performed by Solomon et al.7920 The most consistent correspondence was that the moderate-temperature cross-linking correlated with the molar yield of methane for Pittsburgh Seam bituminous coal and the low-temperature cross-linking correlated with the molar yield of carbon dioxide for North Dakota Zap lignite. The amount of cross-linking produced per mole of gas is roughly equivalent for both cases. Since the lignite reaches maximum cross-linking before the start of methane evolution and the Pittsburgh Seam bituminous coal evolves little C02, correlations could be made separately between cross-linking and C02 evolution in the lignite and crosslinking and CHI evolution in the Pittsburgh Seam bituminous coal. The model described in refs 7 and 20 demonstrated that the assumption that one cross-link is formed for each C02or CH, molecule evolved, gives predictions that agree well with the data. In order to see the combined effect of C02 and CHI evolution on cross-linking, a subbituminous coal was examined since, as shown for the Wyodak coal in Figure 1, a subbituminous coal exhibits both low- and moderatetemperature cross-linking. Chars from the pyrolysis of a Montana Rosebud subbituminous coal produced at high heating rates were analyzed for cross-link densities by the
PittsburghSeam Bituminous Cod,EFR
0.1
0.3
02
% CO, Yield
44
_--% CH4 Yieldn6 I
+
d.4
I
0.5
% CH4 Yield
16
Figure 14. Effect of CH4and C02yield during pyrolysis on the volumetric swelling ratio of chars of Montana Rosebud subbituminous coal, Zap lignite (UNDERC), and Pittsburgh Seam bituminous coal (No. 8).
solvent swelling method. The swelling and gas evolution data (swelling data normalized as in eq 1) are shown in Figure 14. The data for Zap lignite and Pittsburgh Seam bituminous coal are also shown in the same figure. The molar yields of C02and CH, in Figure 14 are shown by solid and dashed lines, respectively. It can be seen that, for Zap lignite, the C02yield (solid line) is predominant and C02 evolution correlates with the cross-linking. The Zap lignite has a very high concentration of carboxyl groups, which appear to be responsible for both the C02 evolution and the cross-linking. The CH4 evolution (dashed line) contributes to cross-linking only at a higher severity of pyrolysis (higher values of pyrolysis gas yields). For Pittsburgh Seam bituminous coal, the CH4 yield (dashed line) is predominant and CH, evolution correlates with the cross-linking. The C02 yield (solid line) is low because of very low concentrations of carboxyl groups in the Pittsburgh Seam bituminous coal. In the case of Montana Rosebud subbituminous coal, the C02 yield (solid line) is predominant in the initial stages of pyrolysis and hence, cross-linking initially takes place during evolution of C02. In the later stages of pyrolysis, the C02yield (solid line) is constant while methane yield (dashed line) starts increasing. The cross-linking in this stage now coincides with methane evolution. The relationship between gas yield and cross-linking is further examined in Figure 15. Figure 15a presents 1 X at 300 and 400 "C as a function of oxygen in the coal. These low temperatures were chosen so that only the low-temperature cross-linking process would dominate. The two highest rank coals (Pocahontas and Upper Freeport) have very low Q values in the starting coal (see Table 11), and Q is observed to increase substantially ((1 - X ) > 1) even at low temperatures. Also, there is only a small percentage change in 1 - X between 300 and 400 "C. At 10% oxygen and above, there is a loss in swelling ((1 - X ) < 1) which increases with rank and the loss is more severe at the higher temperature. The Pittsburgh Seam coal behaves like the two high-rank coals while the Lewiston-Stockton coal appears anomalous in behaving more like a low-rank coal. Parts b, c, and d of Figure 15 present the C02,H20,and CO yields up to 300 and 400 "C for the same coals on a molar basis. The water yield is pyrolysis water (the yield above 200 "C only). The C02 and H,O yields are the highest on a molar basis, while there is almost no CO yield at 300 "C. This suggests that if low-temperature cross-
Energy & Fuels, Vol. 4, No. 1, 1990 51
Cross-Linking Reactions during Coal Conversion
li
;ia A
1.1
5
0
10
15
20
% Oxygen (DhfMF)
a
25 O 0.0 a5* 0
10
2 %CO 3 ,Yield
1
4
5
6
0m c
b
m 4wc 2.0
b
03XPC
.4OWC
X
A
5
0
10
15
% Oxygen (DMMF)
20
l.O-" m a
25
. 0
0.5
-
0.
0.0
AIL
r"
/ \/
41
/
0
I
/
10
H
1
10
S
10
Argonne coals. (a) With C02 yield and (b) with H2O yield. a
5
4 6 % H 2 0 Yield
Figure 16. Correlation of swelling loss with gas yield for six
Y
0
2
15
% Oxygen (DMMF)
20
25
i.apwte 2. BigBrown Texas Idgdte 3. WYOdak Gubbihrminm 4. RUSbUIlZhSeamb g " e ) 5. WellmoR coal
Bitrrminova
O m C
d
400°C
81
-
--___
01 0
5
10
15
-a
Q Oxygen (DMMF)
20
25
Figure 15. Dependence of swelling and gas evolution on oxygen content. (a) 1- X,(b) C02yield, (c) H20 yield, and (d) CO yield.
linking is associated with a gas, it is probably not CO. Between the two, the C 0 2 appears to match the loss in swelling in that it increases sharply above 10% oxygen and the Lewiston-Stockton coal appears to be higher in COz than the Pittsburgh. This correlation is further examined in Figure 16a, which presents 1 - X as a function of C 0 2 yield f o r the Argonne coals (the Pocahontas and Upper Freeport coals were not considered because of their extremely low values of Q in the coal). The decrease in 1X shows a fair correlation with the increase in CO,. The correlation with H 2 0 yield in Figure 16b is poor, although this could be related to the greater difficulty in quantitative measurement of H20 yields compared to C 0 2 yields due to larger background interferences. From the results of the gas evolution data, presented here and p r e v i ~ u s l ~ - the ~ * "possible gases involved in the low-temperature cross-linking reactions are water and carbon dioxide. The possible gases evolved in moderate temperature cross-linking reactions are CHI and other hydrocarbon By correlating the low-temperature cross-linking with COP(or H20) and the moderatetemperature cross-link with CHI, the rates for the crosslinking events are established as being the same as for these gases.
Figure 17. FT-IR spectra of coals and oxidized coals. (a) Comparison of the dry, mineral matter corrected FT-IR spectra of several coals and (b) comparison of the dry, mineral matter corrected and base-line corrected FT-IR spectra of Pittsburgh Seam bituminous coals and oxidized coals.
The relationship between cross-linking and gas yields were quantitatively tested in the network model where it was assumed that approximately one low-temperature cross-link and one moderate-temperature cross-linking is formed in the network for each COz and CHI evolved, respectively.16.20The predicted results show the proper trends in variations in Q, fluidity, tar yields, tar molecular weight distributions, and extract yields with rank, temperature, time, heating rate, and oxidation.16*20 Relation between Functional Groups and CrossLinking. To aid in the identification of the cross-linking chemistry, consideration was given to the functional group composition of the coals and chars. How the functional groups varied with rank and changed in pyrolysis was correlated with the changes in solvent swelling behavior.
52 Energy & Fuels, Vol. 4, No. 1, 1990
sample UF WY ILL#6 PITT#8 POC#3 UT
wv
ZAP
HOH
3.43 3.03 3.41 3.60 1.97 4.79 3.48 2.02
0.11 0.33 0.23 0.16 0.06 0.16 0.23 0.34
Solomon et al.
Table 111. Data on AmDule SamDles (wt % dmmf)" hydrogen aromatic hydrogen carbon 1 adj 2 adj 3 or more Har Htntd Har/Htntdl Cd 2.08 5.62 0.37 0.66 0.71 0.71 22.87 1.73 5.09 20.20 0.34 0.52 0.78 0.43 22.73 2.07 5.71 0.36 0.69 0.78 0.60 2.07 5.83 0.36 0.67 0.80 0.60 24.00 2.19 4.22 0.52 0.73 0.86 13.93 0.60 31.93 1.90 6.05 0.28 0.51 0.80 0.58 2.12 5.83 0.36 0.79 0.67 0.67 23.20 0.94 0.40 0.37 1.58 0.46 0.74 13.47
oxygen OOH h e r 1.75 0.75 5.25 5.0 3.75 2.25 2.5 1.88 1.0 1.25 2.5 4.0 3.75 1.75 5.5 5.0
carbonylb e2 13 2 9.0 9.8 500 "C char 0.78 0.22 2.28 3.28 0.69 5.20 6 3.50 6.75 22.1 600 "C char 0.32 0.19 3.36 3.87 0.87 2.10 6 3.00 4.38 30.0 0.73 1 1.75 5.25 35.4 3.92 0.94 700 "C char 0.11 0.11 3.70 800 "C char 0.00 0.08 3.01 3.09 0.97 0.00 0 1.25 7.00 43.0
Q 2.38 2.38 1.77 NDb 1.07 1.00 NDb NDb
"Peak height at 1700 cm-' (arbitrary units). bND, not determined.
Table V. Data on Coal and Chars Produced at 3 W m i n Heating Rate, No Hold Time (wt % dmmf) oxygen hydrogen carbon sample Hd HoH H, Htna HU/Hb~ Cd carbonyl" OOH Oher wtloss coal 1.93 0.43 1.50 3.86 0.39 12.89 17 6.83 6.69 200 "C char 1.80 0.42 1.51 3.73 0.40 12.00 16 6.69 7.00 1.6 250 "C char 1.83 0.39 1.48 3.70 0.40 12.20 15 6.25 7.25 3.8 1.53 0.32 1.41 3.26 0.43 10.23 13 5.13 300 "C char 6.75 6.7
Q 2.38 2.22 2.03 1.79
Peak height at 1700 cm-' (arbitrary units).
Figure 17a shows FT-IR spectra for the five coals whose solvent swelling behavior was compared in Figure 1. The coals are arranged so that the Kentucky No. 8, which exhibited the least low-temperature cross-linking, is on top and the Zap lignite, which exhibited the most low-temperature cross-linking, is on the bottom. It is the oxygen functional groups, hydroxyl (3400 cm-I), C = O (1700 cm-'), and C-0 (12W cm-I), whose increase matches the increase in low-temperature cross-linking. Of these, the C=O group variation best matches the low-temperature cross-link variation, being close to zero for Kentucky No. 8 and Pittsburgh, which exhibit little low-temperature crosslinking, and increasing for Wyodak, Big Brown, and Zap lignite, which exhibit significant low-temperature crosslinking. Figure 17b shows the variation in functional group composition with oxidation for oxidized Pittsburgh Seam coal. It is the C=O band at 1700 cm-I that undergoes the largest increase with oxidation. It is the oxidized coals that exhibit low-temperature cross-linking. The rank dependence is further compared in Figure 18, which presents the hydroxyl and carbonyl contents determined from FT-IR functional group analysis of the Argonne coals as a function of oxygen content. The complete functional group composition for these coals is presented in Table 111. I t is the carbonyl that best matches the loss in swelling above 10% oxygen. The functional group concentrations measured by FT-IR for a variety of chars have been correlated with solvent swelling data. Pyrolysis of Zap lignite was carried out by using three heating rates: (1)0.5 "C/s (hold time of 3 min at 150 "C) and final temperatures of 200-800 "C in the TG-FTIR reactor; (2) heating a t 3 "C/min with no hold time; and (3) isothermal pyrolysis for 24 h. The chars and
d
15 -
a
10 -
5-
0,
'
8
'
'
I
'
1
5
0.0
5 0 5 10 15 20 25 lo Oxygen (DMMF)
Figure 18. Oxygen functional group compositions as a function of oxygen concentration. (a) Carbonyl and (b) hydroxyl.
the parent coal were analyzed by quantitative FT-IR spectroscopy using the KBr pellet method and by measuring the volumetric swelling ratio in pyridine to estimate cross-link densities. The functional group compositions, swelling and weight loss data are presented in Tables IV-VI. It can be seen
Cross-Linking Reactions during Coal Conversion
Energy & Fuels, Vol. 4,No. 1, 1990 53
Table VI. Data on Coal and Chars Produced at 24-h Hold Time (wt % dmmf) sample coal 50 "C char 100 "C char 150 "C char 200 " C char 250 "C char 300 "C char 350 "C char 400 " C char 500 "C char
hydrogen H b ~ H,/Hb~
Hd
HOH
H,
1.93 1.77 1.77 1.81 1.83 1.85 1.58 1.62 0.93 0.52
0.43 0.39 0.39 0.39 0.39 0.39 0.31 0.28 0.14 0.06
1.50 1.54 1.54 1.45 1.45 1.61 1.71 2.01 1.23 3.43
3.86 3.70 3.70 3.65 3.67 3.85 3.60 3.91 2.30 4.01
0.39 0.42 0.42 0.40 0.39 0.42 0.47 0.51 0.53 0.86
oxygen Osthe, 6.83 6.69 6.13 6.00 6.13 6.13 6.31 6.75 6.13 6.37 6.13 6.25 4.87 8.37 4.50 7.50 6.13 2.30 0.90 3.75
carbon Cd
carbonyl"
12.89 11.83 11.80 12.10 12.20 12.33 10.53 10.80 6.20 3.47
17 16 16 16 16 13 11 8 6 1
OOH
wtloss
Q
0.1 0.7 0.3 1.2 3.5 9.8 15.9 26.5 34.4
2.38 2.38 2.38 1.93 1.89 1.88 1.46 1.53 1.36 1.10
"Peak height at 1700 cm-' (arbitrary units). that, as the temperature is increased, the first change is in the C=O region at 1700 cm-' followed by the hydroxyl region (broad peak between 3600 and 2200 cm-l) and the aliphatic peak (2900 cm-l). The C=O intensity a t 1700 cm-' is a measure of carboxyl groups in coal. The loss of carboxyl intensity is found to increase with an increase in the char formation temperature, and there is a significant loss between 300 and 400 "C. A large loss of the intensity in the hydroxyl stretching region is also found at reasonably low temperatures where cross-linking reactions take place. The loss in aliphatic intensity occurs a t temperatures above the region for low-temperature cross-linking. The two functional groups that are most closely correlated with low-temperature cross-linkingare, therefore, the hydroxyl and carbonyl groups. These are correlated in Figure 19 with the loss in solvent swelling. As can be seen, there is a reasonable correlation with both the hydroxyl and carbonyl. Sipce, in coal, the carbonyls are most likely on carboxyl groups which also exhibit absorption in the hydroxyl region (3400 cm-'), the hydroxyl and carboxyl should both decrease but the hydroxyls will also have a contribution from phenolic OH which will not change with the carbonyl band, unless the cross-linking reactions involved both carboxyl and phenolic hydroxyl groups. It does appear in Figure 19b that the loss in swelling is complete before all of the hydroxyls are removed. From this study, it can be concluded that cross-linking in Zap lignite occurred with a decrease in the concentrations of the carboxyl and the hydroxyl groups. As the FT-IR is not able to distinguish between hydroxyl of the carboxyl groups and hydroxyl groups by themselves (as in phenol), it is not clear whether the decrease of the hydroxyl band is occurring only because of the decrease in concentration of the carboxyl groups.
. . I
.
O 0.4 a6I
0.2 -
.
X
I
X
i 0.0 0.2 Oa4*
0.0
0.1
0.2
0.3
0.4
0.5
H a Weight Percent (DMMF)
Figure 19. Correlation between loss of swelling and loss of functional groups for three heating profiles. (0)3 "C/min; (u) 30 "C/min after 3 min. Hold at 150 "C;(X) isothermal pyrolysis for 24 h. (a) Carbonyl determined at 1700 cm-' and (b) hydroxyl determined at 3400 cm-*.
low molecular weight tar. This variation is consistent with the results of a recently developed macromolecular network model which considers the fragmentation of the network under the combined influence of bond breaking, low-temperature cross-linking, moderate-temperature cross-linking, Conclusions and mass transport. The model predicts that increased Cross-linking reactions in coal pyrolysis were studied in low-temperature cross-linking will result in low tar yields, a series of chars from the Argonne premium sample library low fluidity (e.g., measured by Geissler plastometer), low and other coals. The solvent swelling ratio in pyridine, Q, extract yields, and low molecular weight tar.l6Y2" was measured in the coal and chars and the reduction in 3. Low-temperature cross-linking is increased by oxiQ during pyrolysis was employed as a qualitative indicator dation of the coal and reduced by methylation. of cross-linking. The molecular weight distribution of tar 4. The effect of cross-linking reactions has been shown measured by FIMS was also used to investigate crosto be reduced in low-rank coal by heating particles at a rate slinking. The following conclusions were reached from this of over 20000 OC/s. Tar yields of up to 20% have been study combined with the results of previous s t ~ d i e s . ~ - ' J ~ * obtained ~ for Zap lignite compared to yields of 6% ob1. An investigation of the cross-linking as a function of tainable in low heating rate pyrolysis (0.5 OC/s). coal rank and pyrolysis temperature shows that there are 5. The role of exchangeable cations in cross-linking at least two distinct cross-linkingevents: one occurs at low reactions was demonstrated by removing these ions from temperature prior to tar evolution in low-rank coals only; Zap lignite by demineralization and introducing calcium the second occurs at moderate temperatures slightly above cations in the oxidized Pittsburgh Seam bituminous coal. that for tar evolution. Demineralized lignite shows a lower rate and extent of 2. Coals that exhibit low-temperature cross-linking cross-linking and higher tar yield due to the reduced process also exhibit low tar yields, low fluidity (e.g., mineral content. Calcium-exchanged oxidized Pittsburgh measured by Geissler plastometer), low extract yields, and Seam bituminous coal shows a higher rate and extent of
54
Energy & Fuels 1990,4, 54-60
pyrolysis. The oxidized coal showed early cross-linkingand cross-linking than the oxidized coal. The results are consistent with increased cross-linking in the presence of orlower tar evolution. ganically bound calcium. 10. From this study and previous s t ~ d i e s ,mod~~~J~~~~ 6. The rate and extent of cross-linking is found to be erate-temperature cross-linking appears to correlate best lower in donor-solvent liquefaction than pyrolysis. It is with methane formation. 11. The assumption that moderate-temperature crossprobably due to the partial capping of newly created cross-link sites by hydrogen from the donor solvent. linking occurs at the rate of CH4 evolution and low-tem7. From this study and previous s t ~ d i e s ~ -which ~ J ~ - ~ ~perature cross-linking occurs at the rate of C02evolution compare char solvent swelling behavior to gas evolution, has been employed in a network mode1.16720The model it is shown that low-temperature cross-linking (prior to tar predicts the observed variation in solvent swelling, tar molecular weight distribution, fluidity, tar yield, and exevolution) occurs simultaneously with C02and H,O evotract yield with rank, heating time, temperature, heating lution. The loss of swelling correlates best with the evorate, and oxidation. lution of COP. 8. Studies that compare char swelling behavior to Acknowledgment. We thank Dr. Ron Liotta of Exxon changes in char functional group concentrations have Corporation for supplying the perdeuteriomethylated shown that cross-linking reactions occur with the loss of sample and Mr. George Engelke of Commercial Testing carboxyl groups present in the coal. A clear role for hyand Engineering for providing the fluidity data. We also droxyl groups in low-temperature cross-linking could not thank Professor Eric Suuberg of Brown University, with be established, nor could it be ruled out. Chemical modwhom we have had several helpful discussions on solvent ification of carboxyl and hydroxyl groups in lignites by swelling measurements and cross-linking mechanisms. methylation results in reduced low-temperature crossThis work was supported under Contracts US DOE/PETC linking. NO. DE-FG22-85PC80910, US DOE/METC NO. DE9. The role of carboxyl groups in cross-linking reactions AC21-86MC23075, and US DOE/PETC SBIR No. DEwas further supported by introducing carboxyl groups in AC01-88ER80560. a Pittsburgh Seam bituminous coal by oxidation in air and showing the oxidized coal to behave more like a lignite in Registry No. COz, 124-38-9;CH4, 74-82-8; water, 7732-18-5.
Chemical Percolation Model for Devolatilization. 2. Temperature and Heating Rate Effects on Product YieMst Thomas H. Fletcher* and Alan R. Kerstein Combustion Research Facility, Sandia National Laboratories, Livermore, California 94551 -0969
Ronald J. Pugmire and David M. Grant Department of Chemistry, University of Utah, Salt Lake City, U t a h 84112 Received September 8, 1989. Revised Manuscript Received October 23, 1989 The chemical percolation devolatilization (CPD) model previously developed to describe the devolatilization behavior of rapidly heated coal was based on the chemical structure of the parent coal. Percolation lattice statistics are employed to describe generation of finite tar clusters as labile bonds are cleaved in the infinite coal lattice. The model is used here to describe effects of heating rate and temperature on tar and gas release from coal. Coefficients for the net rate of competition between char formation and side-chain formation are generated from heated screen data performed at five different heating rates. The model also compares well with heated screen data obtained at 1000 K/s and different hold times at the final temperature as well as with data from entrained-flow reactors obtained at higher heating rates (lo4K/s) where particle temperatures have been measured. Results indicate that the CPD model predictions yield good agreement with pub!ished data for a wide range of coals and particle heating rates. Introduction It is well-known that the yield of volatile matter obtained from a pulverized coal is dependent upon the temperature history of the particle. However, the effect of heating rate Work supported by the US. Department of Energy’s Pittsburgh Energy Technology Center’s Direct Utilization AR&TD Program, the DOE Division of Engineering and Geosciences through the Office of Basic Energy Sciences, and by the National Science Foundation through the Advanced Combustion Engineering Research Center (ACERC) at Brigham Young University and the University of Utah. * Author to whom correspondence should be addressed. 0887-06241901 2504-0054$02.50/0
on the yield of volatiles is difficult to study independently of final temperature. For example, the yields obtained in an entained-flow reactor by Kobayashi et al.’ increase with both temperature and heating rate, but the independent contribution of heating rate could not be assessed; heating rates greater than lo5 K/s could only be obtained for the high-temperature conditions (greater than 1500 K). This is typical of most entrained-flow coal devolatilization (1) Kobayashi, H.; Howard, J. B.; Sarofim, A. F. Sixteenth Symposium (International) on Combustion; The Combustion Institute: Pittsburgh, 1976; p 411.
0 1990 American Chemical Society