Energy & Fuels 1987,1, 305-308
305
Relation between Tar and Extractables Formation and Cross-Linking during Coal Pyrolysis E. M. Suuberg* and P. E. Ungert Division of Engineering, Brown University, Providence, Rhode Island 02912
J. W . Larsen Department of Chemistry, Lehigh University, Bethlehem, Pennsylvania 18015 Received August 15, 1986. Revised Manuscript Received February 2, 1987
This paper presents a combined study of tar and extractables formation and cross-linking processes during rapid pyrolysis of coal. Tar and extractables were characterized by gel permeation chromatography, and cross-linking was characterized by a solvent swelling technique. The behaviors of coals ranging in rank from low-volatile bituminous to a lignite were examined. I t was noted that the low-rank coal cross-links a t a much lower temperature than the high-volatile bituminous coals and that the low-volatile bituminous coal is already highly cross-linked to begin with. These observations may help explain the widely differing yields and nature of tars and extractables produced by pyrolysis of these different ranks of coal.
Introduction It has recently been shown that the technique of solvent swelling, as has been applied to the analysis of the macromolecular structure of coals,' can also be applied to the analysis of chars produced by pyrolysis.2 This technique is utilized in this paper to help shed further light on the complex processes of "depolymerization" and "repolymerization" (leading to char formation) that occur during pyrolysis of coals. In particular, the differences manifested by coals of different ranks are considered. In this work, data on the molecular weight distributions of pyrolysis tars are combined with data on cross-linking of different rank chars to draw further conclusions about the nature of these complex processes.
Experimental Section The analyses of the four coals examined in this study are provided in Table I. All samples used in&t study had particles in the size range 53-88 pm. Pyrolysis was performed in an electrically heated wire mesh, which assures uniform and rapid heating of all particles. All experimentswere performed with a heating rate of roughly lo00 K/s, up to the indicated peak temperature, followed by cooling at a rate between 200 and 400 K/s. All pyrolyses were performed under atmospheric pressure helium. As used in this paper, the term "tars* refers to room-temperature condensable products that have been expelled from the particles during pyrolysis. These materials generally have a molecular mass greater than 100 and are more than 97% soluble in tetrahydrofuran (THF). Extractables are materials left behind in the char that are THF soluble. The extraction procedure is very mild, involving a 1-h ultrasonicated soak in THF (during which time the THF reached reflux temperatures). The sum of tar plus extractables is referred to as metaplast-the softened, transportablefraction of the coal. It is important to bear in mind that the estimate of metaplast obtained by summing tar plus extractablea is a minimum value for that present in the coal during actual pyrolysis, since some recombination reactions may occur between extractables and unextractable char during cooling. The analysis of the molecular mass distributions of the tars and extractables is performed by gel permeation chromatography (GPC), as has been described else~here.~ It should be noted that calibration of the GPC columns was performed by vapor-phase osmometric measurement of the molecular masses of actual fractionated samples of Bruceton coal tars. 'Present address: Shell Development Corp.,Houston, TX 77002.
Table I. Ultimate Analyses of Coals Examined'
Bruceton Pitts, No. 8 bituminous Hillsboro IL, No. 6 bituminous WV, Pocahontas
C H 80.4 5.3
anal., ?6 0 N S ash moisture 6.7 1.6 1.0 4.6 1.7
67.2 4.6 12.3 1.2 3.4 11.7 84.4 4.2
3.7 0.3 0.5
8.6
6.8
0.2
66.7 3.7 19.5 0.9 0.8 9.3
32.4
Low-Volatile bituminous
Beulah, ND, Lignite
a All results on a dry basis except that for moisture, which is reported on an as-received basis. All analyses by Huffman Laboratories, Inc.
Results and Discussion In an earlier communication, it was shown that the Bruceton bituminous coal apparently cross-links a t a somewhat higher temperature than does the North Dakota lignite.2 This is consistent with the viewpoint that early cross-linking prevents softening of low-rank coals during pyrolysis. This observation is also consistent with the low tar formation tendency of low-rank coals, in the sense that tar precursors are quickly cross-linked into the char of low-rank coals, prior to their escape. The solvent-swelling tendency of the coal chars in pyridine has been used as a qualitative index of the extent of cross-linking; the higher the swelling ratio (i.e., the volume of the solid when swollen divided by its original volume), the lower the degree of cross-linking. In this light, comparative solvent swelling data are shown for chars of all four ranks of coal in Figure 1. -The low-volatile bituminous coal swells to an extent of less than 5% in both the unpyrolyzed state and as a char. Consequently, it is difficult to say much about changes in the network structure of this coal during pyrolysis, on the basis of solvent swelling information. It is, however, likely that this coal is very highly cross-linked even in its unreacted state, given its small tendency to swell. (1) Green, T. K.; Kovac, J.; Larsen, J. W. Fuel 1984, 63, 935 and references therein. (2) Suuberg, E. M.; Lee, D.; Larsen, J. W. Fuel 1985, 64, 1668. (3) Unger, P. E.; Suuberg, E. M. Fuel 1984, 63, 606.
0887-0624/87/2501-0305$01.50/00 1987 American Chemical Society
Suuberg et al.
306 Energy & Fuels, Vol. 1, No. 3, 1987
300
500 700 900 TEMPERATURE ( K )
1100
Figure 1. Pyridine swelling ratio as a function of peak pyrolysis temperature for four coals: ( 0 )Illinois No. 6; (A)Bruceton; ( X ) Beulah Lignite; (0) Pocahontas. Table 11. Tar Yields and Number Average Molecular Masses of the Tars sample Pocahontas
ND Lignite Illinois No. 6 Bruceton
temp, K
yielda
737 1083 737 1133 807 1001 1190 727 770 819 897-1315'
3.0 9.1 3.0 4.2 7.7 17.4 19.9 8.1 14.8 16.9 23.2-25.9
l
Mn 244 199 323 279 330 333 332 307 384 347 325-357
All on a weight percent, as-received basis. Many experiments in this temperature range, all leading to the same general tar yield and number average molecular mass.
The Illinois No. 6 high-volatile bituminous coal has swelling behavior very similar to that observed for the Bruceton high-volatile bituminous coal. Both these coals soften markedly, and both yield copious amounts of tar during pyrolysis. As noted previously,2both of these coals show little further tar formation a t temperatures above about 900 K, a temperature a t which new cross-link formation is quite measurable by this technique. It may be noted that the tars of these coals also show remarkably similar molecular mass distributions.' The essentially identical pyridine swelling of Bruceton and Illinois No. 6 coals has another implication for this work. These two coals differ significantly in hydroxyl content4but not in pyridine swelling. This demonstrates that swelling is not dominated by the strong pyridinehydroxyl interaction, although it undoubtedly plays an important role. Since hydroxyl loss occurs during pyrolysis, it is important that pyridine swelling does not follow this parameter. In fact, pyridine swelling shows no dependence on oxygen content between 8% and 30% oxygen. Below 8% oxygen, the effects of changes in oxygen content and in the macromolecular network on pyridine swelling cannot be ~ e p a r a t e d . ~ In contrast, little tar formation is seen a t temperatures much in excess of 700 K in the case of the North Dakota lignite.2 On the basis of the high-volatile bituminous and lignite data, it might be tempting to ascribe significance to a swelling ratio of about 2 as the lowest value a t which tar formation is possible. This, of course, is inappropriate since the low-volatile Pocahontas coal yields significantly more tar than the lignite, despite its highly cross-linked nature. Table I1 shows the yields of tar from the Pocahontas sample and all other samples studied in this work. The number average molecular mass of the Pocahontas tar is seen to significantly decline with increasing tem(4) Larsen, J. W.; Green, T. K.; Kovac, J. J. Org. Chen. 1985,50,4729. ( 5 ) Larsen, J. W.; Quinga, E. M. Y., unpublished observations.
.. . . . . ,
~
~
1000
0
I.
~
2000
*
,I
l
4000
3000
MOLECULAR WEIGHT
Figure 2. Molecular mass distribution of tars evolved during pyrolysis of Pocahontas low-volatilebituminous coal. Curves show range of pyrolysis temperatures over which the tar was evolved. T h e curves are partially integrated, as described in the text. L",
$ [r
w
a
I!, 0
~,
,
,\;, \
'.-- - -2000
IO00
,
3 000
,
,
,I 4000
MOLECULAR WEIGHT
Figure 3. Molecular mass distributions of tars evolved during pyrolysis of Illinois No. 6 high-volatile bituminous coal. Curves show range of pyrolysis temperatures over which the tar was evolved and are partially integrated.
NORTH DAKOTA L I G N I T E
a $201
:.. .[
737- I133 K
30 MOLECULAR WEIGHT
Figure 4. Molecular mass distribution of tars evolved during pyrolysis of Beulah lignite. Curves show range of pyrolysis temperatures over which the tar was evolved and are partially integrated.
perature. The actual molecular mass distributions are seen in Figure 2. The molecular mass distributions shown in Figure 2 (and later in Figures 3 and 4) are partially integrated molecular mass distributions, in which the oridinate represents the weight percent of tar at any molecular mass A100 mass units of the abscissa value. The sharp peak of the Pocahontas tar molecular mass distribution in the range 200-400 is quite different from that of the tars of lower rank bituminous coals; as shown in Figure 3, these coals have a peak in the range from 500
~
Coal Pyrolysis
to 700.3 The implication appears to be that the Pocahontas coal is so highly cross-linked to start, that it is unlikely that enough bonds can break so as to yield large tar fragments. Or, viewed another way, the larger the fragment of coal to be released by bond breakage processes and the higher the cross-link density, the lower the probability that enough bonds can be broken during pyrolysis in order to release the fragment. There is another quite speculative explanation for the behavior of the Pocahontas coal. Evidence for an important role for London dispersion interactions in highrank coals has recently appeared.e In both cases, the extractability of coals was enhanced by using selective chemical derivatization to disrupt the London interactions (“stacking interactions”). The Pocahontas coal may be holding tightly to the pyrolysis products by using London forces to that only the lowest molecular mass, most volatile materials, escape. This would truncate the molecular mass distribution a t a low value, as observed. Figure 4 shows the variation of molecular mass distribution with temperature for the North Dakota lignite. The number average molecular masses are given in Table 11. The decrease in number average molecular mass of the tar species with increasing temperature in the case of the lignite (and the low-volatile bituminous coal) is contrary to what might be expected on the basis of vapor pressure arguments (higher temperatures should promote vaporization of even larger fragments of the coal). This trend toward lower number average molecular masses of fragments with increasing severity of thermal treatment is, however, consistent with the data that imply a higher degree of cross-linking with a more severe thermal treatment. It has been shown that in a polycondensation process, the molecular mass of extractables decreases with extent of polycondensation.’ This behavior is not as striking in the case of the high-volatile bituminous coals, which show smaller variations of either number average molecular mass or molecular mass distributions with increasing temperature (or tar yield, see Table II).3 Presumably, this reflects the fact that relatively little crosslinking occurs during the active tar formation period.2 It is important to realize that random bond breaking during pyrolysis will lead to greater molecular mass products with more severe treatment. A kinetic analysis of the delignification of wood fits the present situation well and leads inexorably to the conclusion that increasing processing severity will lead to increasing molecular mass products from a macromolecular network if bonds are broken at random.8 A small fragment will most likely be bound to the insoluble network by only a few bonds and will probably be released early in the bond-breaking process. A high molecular mass fragment will most probably be bound to the network at many points, because its size creates the opportunity for this. Its release requires breaking many bonds, and this will probably occur only after substantial reaction has taken place. The observed product molecular mass distributions are opposite to this prediction in the case of the lignite and low-volatile bituminous coal and are most consistent with the occurrence of facile retrograde reactions (i.e., those reactions forming char). Of course, it could be argued that smaller molecular mass fragments might be expected a t higher temperatures because of a natural distribution of bond strengthsstronger bonds can be broken at higher temperatures, so (6) Stock, L. M.; Mallya, N. Fuel 1986,65,736; Larsen, J. W.; Quinga, E. M. Y. Energy Fuels, preceding paper in this issue. (7) Van Krevelen, D. W. Fuel 1965, 44, 229. (8) Yan, J. F.; Johnson, D. C. J. Appl. Polym. Sci. 1981, 26, 1623.
Energy &Fuels, Vol. 1, No. 3, 1987 307 Table 111. Molecular Mass Distributions of Extractable8 and Metaplast from Bruceton Bituminous Coal” amount, wt % molecular mass range extractables metaplast 100-1100 17 46 1100-2100 31 31 2100-3100 24 13 3100-4100 13 5 >4100 15 5
M.
1002
533
“All results for vacuum pyrolysis to a peak temperature of roughly 740 K.
Table IV. Typical Values of i@,
X 0.3 0.4
Mc 1105 1464
for Bruceton Coal
X
M c
0.5 0.6
2172 4204
the coal can break into progressively smaller pieces as the temperature is increased. This does not appear to be bome out by the data on the high-volatile bituminous coals, however, and thus this hypothesis would necessarily have to go hand in hand with a hypothesis of scissile bonds that are similar in very low-rank and very high-rank coals, but quite different in medium-rank coals. This seems unlikely. Some molecular mass distributions of the extracts of the Bruceton high-volatile bituminous coal have been shown previously? These data are summarized in Table 111, along with data on the metaplast (sum of tar plus extractables) molecular mass distribution for this same coal. The difference between the extractables and metaplast arises from the fact that the latter contains a substantial contribution of species that were sufficiently light so as to have evaporated and been measured as tar, outside the particle. A comparison of the number average molecular mass of the extractable pyrolysis fragments and the number average molecular mass between cross-links is useful in characterizing the manner in which the cross-linked structure breaks down. In order to calculate the molecular mass between cross-links from solvent-swelling data, the Flory-Rehner equation is often employedg but is subject to criticism when applied to highly cross-linked, rigid networks.l0J Nevertheless, the Flory-Rehner equation is often used as an approximation because of the lack of information concerning the repeat unit size, needed for more sophisticated m ~ d e l s . ~It~ also J ~ underestimates ATc: so its use in this case is conservative. The Flory-Rehner equation may be expressed as PCMS 41i3
Mc = --
pS
In (1 - 4)
+ 4 + Xd2
where p c is the density of the original coal, p s is the density of the solvent, M, is the molecular mass of the solvent, X is the solvent-network interaction parameter, and 4 is the inverse of the swelling ratio. The evaluation of X is also difficult, particularly since pyridine is a specifically interacting solvent. Table IV shows a range of X values and their effect on for the Bruceton coal (or char up to about 800 K). (9) Flory, P. J.; Rehner, J. J. Chem. Phys. 1943, 11, 521. (10) Larsen, J. W.; Kovac, J. ACS Symp. Ser. 1978, No. 71, 36. (11) Lucht, L. M.; Peppas, N. A. ACS Symp. Ser. 1981, No. 169,43. (12) Kovac, J. Macromolecules 1978, 11, 362. (13) Peppas, N.; Lucht, L.; Hill-Lievense, M.; Hooker, A. DOE Final Technical Report DE-FG22-80PC30222; U.S. Department of Energy: Washington, DC, 1983.
Energy & Fuels 1987,1, 308-309
308
The range of Mcobtained by this method compares well with estimates based on a rigid-chain model applied to comparable ranks of coal,13in the range X = 0.3-0.5. It is seen that the extractable and transportable fragments of the coal have a number average molecular mass that is comparable to or smaller than the number average molecular mass between cross-links, for any reasonable value of
2. The downward shift in molecular mass distributions of pyrolysis tars with increasing temperature from a lignite and a low-volatile bituminous coal is consistent with occurrence of polycondensation during pyrolysis. 3. Pyrolysis fragments appear to be, on the average, comparable in size to or smaller than the number average molecular mass between cross-links in a Bruceton bituminous coal.
Conclusion
Acknowledgment. E.M.S. gratefully acknowledges the support of the US. DOE (Grants DE-FG22-81PC40803 and DE-AC18084FC1061),and J.W.L. thanks the Exxon Education Foundation and the Gas Research Institute for support of this work.
x.
1. A lignite is seen to cross-link at much lower pyrolysis temperatures than high-volatile bituminous coals. A lowvolatile bituminous coal is already too highly cross-linked to use solvent swelling to track pyrolysis behavior.
/y
9
9
Communications New Utilization of NaCl as a Catalyst Precursor for Catalytic Gasification of Low-Rank Coal Sir: Among many gasification catalysts, alkali-metal carbonates have been generally accepted to be the most promising ones. Although alkali-metal chlorides are cheaper than carbonates, they have two intrinsic disadvantages: (1)The affinity between cation and anion is so strong that the alkali-metal cation can not favorably interact with char; therefore, the catalytic effectiveness is small. (2) The C1 may muse corrosive problems on various parts of materials once introduced into the gasification system. Several successful attempts have been made to overcome the first demerit,ld3but none of them have been satisfactory with respect to the removal of chlorine. In the present paper, a new method utilizing NaCl as the catalyst raw material in the steam gasification of brown coal is presented. Not only high catalytic activity but also complete removal of C1 was easily achieved. Yallourn coal (C, 67.1; H, 4.8; N, 0.8; S, 0.3; C1,0.09; 0, 26.9 w t % (daf)), an Australian brown coal, was used. It was received in the form of briquettes and was crushed up to 100-200 mesh. Carboxyl and hydroxyl groups in the coal were determined by ion exchange with barium acetate and by acetylation? the amounts of these groups being 1.7 and 6.0 mequiv/g (daf) of coal, respectively. The oxygen-containing functional groups in such a low-rank coal act as cation-exchange sites. The method of catalyst addition essentially consists of a cation-exchange step and a water-washing step. Thus, 10 g of coal powder was soaked in 200 mL of NaCl solution in the concentration range 0.3-5 N, and the pH of the solution was adjusted by adding an adequate amount of NH,. As the ion exchange progressed, the pH of the solution decreased. The ion exchange was allowed to continue until there was no further change in pH. The exchanged coal was separated from the solution by fitration, washed twice with deionized water (100 mL each), and then dried at 380 K in N2. The amount of Na incorporated into the coal was determined (1)Huttinger, K. J.; Minges, R. Fuel 1984, 63, 9-12. (2) Lang, R. J. Fuel 1986,65, 1324-1329. (3) Huttinger, K. J.; Minges, R. Fuel 1985, 64, 486-490. (4) Zhou, P.; Dermer, 0. C.; Crynes, B. L. In Coal Science; Gorbaty, M. L., Larsen, J. W., Wender, I., Eds.; Academic: London, 1984; Vol. 3, pp 253-300.
Table I. Catalyst Content and Reactivity in Steam at 923 K content, w t % (dry) specific Na C1 rate, h-' catalyst addn method final pH 0.09 0.15 none 0.27 0.3O (O.l)b 0.10 ion exchange 2.6 NaCl 1.7 (1.6) 1.6 ion exchange 5.5 NaCl 2.6 (2.4) 0.10 1.8 6.2 ion exchange NaCl 3.5 (3.4) 2.2 ion exchange 9.2 NaCl 4.4 (4.2) 0.09 2.1 10.3 ion exchange NaCl 11.1 5.3 (5.3) 2.6 ion exchange NaCl 2.7 (4.4) 3.7 0.21 impregnation NaCl 3.6 (3.8) 2.0 Na2CO8 impregnation 5.3 (5.6) 2.4 Na2C03 impregnation Determined by the flame spectrochemical analysis. bEstimated from the ash content in the gasification residue.
by back-exchange with a dilute HC1 solution followed by flame spectrochemical analysis. The C1 content in the resulting sample was determined by the Eschka method. The steam gasification was conducted in a thermobalance at 923 K. Gasification was allowed to continue for 2 h, and the remaining char was completely burnt to check the total amount of ash and catalyst. Details of the gasification procedure have been described el~ewhere.~ The reaction consisted of the devolatilization stage and the following char gasification stage. The reactivities of Na2C03-impregnated and NaC1-impregnated coals were determined for comparison. These samples were prepared by evaporating Na2C03or NaCl solutions to dryness in the presence of coal. The results are summarized in Table I. The amount of exchanged Na was not dependent on the NaCl concentration but on the final pH of the solution; the extent of exchange increases with an increase in the pH. At pH 9.2, the Na content was 3.5 wt %, which corresponds to 1.6 mequiv/g (daf) of coal. It was very close to the amount of carboxyl groups in the raw coal, 1.7 mequiv/g (daf). This result suggests that only carboxyl groups are exchangeable sites for Na ion a t pH less than 9. A t a high pH of 11, the Na loading increased to 5.3 wt % (2.4 mequiv/g). This suggests that Na ion was exchanged not only on carboxyl groups but also on phenolic hydroxyl groups ( 5 ) Tomita, A,; Takarada, T.; Tamai, Y. Fuel 1983, 62, 62-68.
0887-062418712501-0308$01.50/0 Q - 1987 American Chemical Societv
