Ind. Eng. Chem. Prod. Res. Dev. 1084, 23, 134-140
134
Effect of Coal Rank on Structure of Tars from Low-Temperature Pyrolysis of Canadian Coals E. Furlmsky;
L. Vancea, and R. Belanger
Energy Research Laboratories, Canada Centre for Mineral and Energy Technoiogy, Department of Energy, Mines and Resources, Onawa, Canada, K1A OG1
Twelve Canadian coals of different rank were pyrolyzed at 535 O C in a Fischer assay retort. The coals and tars were characterized by NMR techniques. Yields of tars decreased linearly with Increasing aromaticity of coals. A linear correlation between the yields of tar and H/Cratio of coals was established for bituminous coals. Lower rank coals exhibited markedly different trends.
Introduction Declining reserves of conventional crudes have resulted in a growing need to produce fuels and petrochemicals from unconventional sources. In this respect, coal has been identified as a potential feedstock. It is known that certain portions of volatile matter, if isolated from coal, may represent valuable feedstock. Various experimental techniques for hydrocarbon liquids and gas production, based on pyrolysis performed in an inert atmosphere, have been developed. The scale of these processes ranges from laboratory, bench, and pilot plant, to commercial size (Edwards et al., 1980; Cortez and La Delfa, 1981). The process exists as both batch and continuous. The need for different pyrolyzing techniques results from the wide range of coals being considered as the feedstock. In other words, it may not be possible to pyrolyze different coals effeciently by using the same process. The composition and yield of liquids and gases depends on the origin of the coal in addition to the technological parameters applied during pyrolysis. Among the latter, temperature, residence time, rate of heating, and size of coal particles are perhaps the most important. The efficiency of coal pyrolysis is then determined by an optimal combination of these parameters. These aspects of coal devolatillisation have been reviewed in detail by Anthony and Howard (1975). The viability of pyrolitic processes for fuels and petrochemical production depends on a complete utilization of the residual char. Thus the feasibility studies are based on a combination of coal pyrolysis with either coal-fired power stations (Herring and Tollefson, 1980) or gasification units (Schulz 1979). Here, hydrocarbon liquids and gases are "skimmed" off the coal prior to its combustion or gasification. The liquid product is a concentrate of aromatic compounds and phenols. Industrial procedures for isolation of these components in pure form were reviewed by Collin and Zander (1980). When production of commercial fuels is preferred, the liquid product is subjected to catalytic hydrorefining; the gas byproduct is hydrocarbon rich and can be a useful petrochemical feedstock or fuel. Little attention has been paid to coal conversion by pyrolysis in Canada. This may be attributed to the fact that the main coal reserves are in the same region as the largest reserves of conventional crude oil and natural gas. However, one feasibility study indicated on a high potential of this process when combined with thermal electric power generation (Herring and Tollefson, 1980). A similar conclusion was reached by Scott (1979) in his work on the short residence time pyrolysis of subbituminous coal from 0196-432118411223-0134$01 .SO10
the Forestburg seam in Alberta. Both these studies emphasized the lack of information regarding the suitability of Canadian coal as feedstocks for production of fuels and petrochemicals by pyrolysis. This work is an attempt to fill this gap. In the present study the assessment of a series of Canadian coals is based on the determination of yields of gaseous and liquid hydrocarbons as well as reaction water. These yields were correlated with properties of coals such as aromaticity and H/C ratios. The aim was to find out whether correlations such as those reported for Australian coals (Tyler, 1980) can be established for Canadian coals. The reliability of the correlations as indications of the suitability of coals as pyrolysis feedstocks is assessed. The correlations, if valid, may be used to predict yields of tar during pyrolysis in a series of coals of different ranks (Saxby, 1980). The tars were subjected to structural evaluation by NMR techniques. Some structural parameters of tars were correlated with their yields. Attempts were also made to discuss the structure of coal feedstocks in terms of various coal models. It was observed that in the range of coal ranks used in this work, at least three different models of coal can be used to explain the structure of coals. Experimental Section Pyrolysis experiments were performed in a modified Fischer assay retort which was connected directly to a volume calibrated trap, immersed in ice. The retort and its 70-g coal charge were weighed before and after each experiment. After charging, the retort was heated (12 "C/min) to 535 "Cand held at this temperature for 15 min; the heating was then discontinued. Within this time the production of tars was complete in all cases. The tar product was collected and measured in the trap (0 "C) while the sample of gas was taken after passing the water-cooled condenser. Prior to the analyses, the tars were dried with anhydrous sodium sulfate and kept in sealed vials to avoid contact with air. The difference in total weight of the system before and after the experiment was assumed to be equal to the yield of gaseous products. Proximate analysis of coals was performed on a Fischer coal analyzer Model 490. The CHN 240 Perkin-Elmer analyzer and Leco sulfur analyzer were used for the ultimate analysis of all solid and liquid materials. The oxygen was determined by the difference. Gas component analysis was performed with a Sigma Perkin-Elmer gas chromatograph with a three-column system and two detectors (FID and hot wire). The Hewlett-Packard 5710A chromagraph was used to determine @ 1904 American Chemical Society
Ind. Eng. Chem. Prod. Res. Dev., Vol. 23, No. 1, 1984 135 Table I. Properties of Coal Feedstocks proximate
ultimate
coal a
ash
moist
FC
VM
VMdaf
1. Canmore 2. McIntyre 3. ByronCreek 4. Sukunka 5. Balmer 6. Coalspur 7. Shaughnessey 8. Devco 9. Prince 10. E. Blackfoot 11. Bienfait 12. Onakawanab
9.2 8.2 17.3 3.2 10.3 8.5 9.7 2.7 14.7 21.2 7.0 26.2
2.7 2.1 5.1 1.9 0.7 10.5 12.6 1.9 6.4 15.4 31.2 2.3
75.7 71.0 54.0 70.7 68.8 47.7 43.8 58.6 45.1 30.1 32.0 32.5
12.4 18.6 23.5 24.0 20.2 33.1 33.9 36.8 33.9 33.3 30.0 39.0
14.0 20.7 30.3 25.3 22.7 41.0 43.8 38.6 42.9 52.0 48.4 54.5
C 80.5 81.2 66.5 84.0 80.2 64.4 59.1 83.3 63.6 44.9 45.0 48.5
H
S
N
0
H/C
fa
MR
3.6 4.2 3.8 4.7 4.4 4.1 4.2 5.8 4.4 3.7 3.0 3.3
0.8 0.6 0.3 0.5 0.2 0.2 0.7 1.7 4.9 0.3 0.5 5.3
1.6 1.2 1.0 1.3 1.5 1.0 1.5 1.8 1.3 1.0 0.8 0.7
1.6 2.5 6.0 4.4 2.7 11.3 12.1 5.3 4.7 13.5 12.5 13.7
0.55 0.62 0.69 0.67 0.66 0.76 0.85 0.83 0.83 0.99 0.80 0.82
0.86 0.77 0.74 0.79 0.80 0.67 0.65 0.69 0.71 0.58 0.57 0.57
2.12 1.49 0.93 1.41 1.38 0.57 0.50 0.93 0.67 0.49 0.40 0.25
a Bituminous coals no. 2 to 7 are from Western Canada while no. 8 and 9 from eastern Canada. Samples no. 11 and 1 2 are lignites. Dried sample.
the boiling range of liquid products and to estimate the yield of neutral oil and phenol fractions. Sample preparation for reflectance analysis was carried out in accordance with ASTM procedures. Carbon-13 NMR spectra of solid coals and chars were obtained a t 45.28 MHz on a Bruker CXP-180 NMR spectrometer by means of cross-polarization dipolar decoupling (Pines et al., 1972) and magic angle sample spinning techniques (Schaefer et al., 1977). A single matched 1-ms cross-polarization contact was used with radiofrequency field amplitudes of 45 kHz. The delay time between successive contacta was 2 s. Spin-temperature alternation was used as well as flip-back of the proton magnetization after data acquizition (Tagenfeldt and Haeberlen, 1979). The adjustment of the Hartmann-Hahn matching condition was checked during the course of the experiments by inserting a standard hexamethylbenzene sample in the probe. A sweep width setting of 20 kHz was employed to acquire 500 pt free induction decays, which were zero filled to 4K before Fourier transformation. Magic angle spinning was performed at 3.4 kHz with Kel-F rotors of the Andrew type (Andrew et al., 1969). Intensities of the aliphatic and aromatic bands were corrected for spinning side bands, assuming the first-order high- and low-field sides and intensities to be the same. Integrations of the corrected bands yielded content of aromatic carbon (Cm,,) and aliphatic carbon (C&). The C, and total carbon content of corresponding coals were used to calculate the aromaticity (fa = C,O/C). The carbon and hydrogen distributions of liquid products were obtained on a Varian CFT-20 NMR spectrometer according to a technique described elsewhere (Ozubko et al., 1981). Fractions of methyl, methylene, and benzylic hydrogens were obtained by integrating the following regions: 0 to 1.0,l.O to 2.0, and 2.0 to 3.5 ppm, respectively. The solvent used was CDC1, dried over 4A molecular sieves. The tars were dried over anhydrous Na2S04prior to NMR analysis. Contents of phenols in tars were estimated by integration of the phenolic hydrogen peak. To eliminate with certainty the interference of olefinic hydrogen and water, the NMFt analysis was performed twice, i.e., with a diluted and with a concentrated sample. In the latter case no absorption was observed in the region of olefinic and phenolic hydrogen. This confirmed the complete deshielding of phenolic protons under the aromatic envelope as well as the absence of H 2 0 and olefins. Results and Discussions Rank and Chemical Properties of Coals. Proximate and ultimate analyses as well as H/C ratios, aromaticities, and mean reflectance (MR) values of the coal feedstocks
20 W V
z U
+ V
W J
la E
z a W
I O
10
20
3 0 4 0
x)
VOLATILE MATTER d o f
60
,
Figure 1. Mean reflectance vs. volatile matter of coal.
are listed in Table I. Among these feedstocks the Devco and Prince coals as well as Onacawana lignite are from eastern Canada while all the others are from western Canada. The coal rank can be derived from the correlation between MR and volatile matter (VM) contents shown in Figure 1,where the number at each point corresponds to the coal in Table I. According to this classification, coal no. 1is a semianthracite while the coals with MR ranging between 0.5 and 1.5 are bituminous coals with volatilities decreasing as the MR increases. Coals with MR lower than 0.5 are either subbituminous or lignites, e.g., no. 10 is subbituminous while no. 11 and 12 are lignites. It is apparent that the MR values of coals, with the exception of eastern coals and Onakawana lignite, deviate very little from the curve (Figure 1). This confirms a difference in structure between eastern and western coals (Furimsky et al., 1983). For example, the VM content of the Devco and Prince coals are respectively about 8 and 5% higher than for the western coals of similar MR values. The difference is complemented by a relatively high content of exinites in the eastern coals (Furimsky and Ripmeester, 1983). The rank of coals can be determined from their fa values (Saxby, 1980). However, in this case a clear-cut differentiation between lignites and subbituminous coals is not easy to make suggesting that the chemical structure of these materials may be similar. Moisture and ash contents are decisive in determining the classification of low-rank coals based on the heating value of coals. The values of VM determined by standard methods iclude all volatile components of coal (reaction water, gases,
136
Ind. Eng. Chem. Prod. Res. Dev., Vol. 23, No. 1, 1984
Table 11. Yields of Pyrolysis Products (Fischer Assay) products, yield, wt % coal no. 1. Canmore 2. McIntyre 3. Byron Creek 4. Sukunka 5. Balmer 6. Coalspur 7. Shaughnessey 8. Devco 9. Prince
10. E. Blackfoot 11. Bienfait 12. Onakawana a
RW
= water yield
-
tar
gas
water
char
RWa
TYb
TY dafb
tar dafb
0.7 3.3 7.0 4.9 5.7 8.4 9.4 13.5 9.9 3.2 2.9 3.0
2.6 4.8 5.0 6.0 3.9 6.1 6.9 7.4 7.3 10.3 11.4 14.2
3.2 2.9 6.0 2.9 1.9 15.7 17.1 5.9 10.0 21.8 38.3 10.9
93.5 89.0 82.0 86.2 88.5 69.8 66.6 73.2 72.8 64.7 47.4 71.9
0.5 0.8 1.0 1.0 1.2 5.2 4.5 4.0 4.6 6.4 6.8 8.6
3.8 8.9 13.0 11.9 10.8 19.7 20.8 24.9 21.8 19.9 21.5 25.8
4.3 10.0 16.8 12.6 12.1 24.4 26.9 26.1 27.6 31.4 34.7 36.1
0.8 3.7 9.0 5.2 6.4 10.4 12.1 14.2 12.5 5.0 4.7 4.2
moisture.
TY = tar
+ gas + RW.
Table 111. Properties of Liquid Products
1.Canmore 2. McIntyre 3. Byron Creek 4. Sukunka 5. Balmer 6. Coalspur 7. Shaughnessey 8. Devco 9. Prince 10. E. Blackfoot 11. Bienfait 12. Onakawana
84.5 88.4 83.4 85.2 85.9 76.4 78.0 83.4 77.8 81.7 78.4 79.0
5.4 8.0 8.3 8.1 7.3 8.4 8.8 9.3 8.1 9.2 9.1 9.1
1.8 0.8 0.7 0.6 0.7 0.5 1.0 1.0 0.7 0.7 0.5 0.3
0.4 0.4 0.4 0.3 0.2 0.7 0.5 2.6 0.4 0.5 2.3
1.08 1.20 1.14 1.02 1.30 1.35 1.34 1.24 1.35 1.40 1.40
75.1 66.5 72.5 70.0 59.4 60.2 60.0 60.5 59.1 58.5 58.0
24.9 34.5 27.5 30.0 40.6 39.8 40.0 39.5 40.9 41.5 42.0
36.2 32.3 34.4 34.6 28.9 25.8 27.3 25.9 28.4 21.5 27.9
63.8 67.8 65.6 65.4 71.1 74.2 72.7 74.1 71.6 78.5 72.1
0.54 0.56 0.54 0.50 0.48 0.58 0.50 0.54 0.53 0.43 0.53
2.95 2.40 2.69 2.20 2.29 2.50 2.23 2.36 2.34 2.60 2.34
tr 26.7 tr tr 70.8 69.0 30.0 tr 56.0 33.6 58.2
4
25P
20
2.4 7.2 4.7 4.8 14.5 11.5 5.8 10.8 7.3 11.5 9.3
I
\
0 0.4
0.5
0.6
0.7
H/C
OF COALS
0.8
0.9
1.0
Figure 2. Yields of tars and gases vs. the H/C ratios of coals.
and tars). The VM content is, therefore, not a reliable measure of the tar-producing potential of a coal. It is essential that yield of tars is determined directly. This can be achieved by pyrolyzing coals by Fischer assay (Table 11). The correlation between the H/C ratio of coals and yield of tars and the sums of tar and gas is shown in Figure 2, where the latter were obtained from yields of tars and gases (Table 11) recalculated to a dry ash free basis. A linear correlation is obtained for all bituminous coals and the semianthracite. However, for low rank coals the yields are markedly lower than one would expect on the basis of the H/C values. This contradicts observations made on Australian coals, where a linear correlation was obtained for a whole range of H/C ratios under conditions of low
I
0.6
I
0.7
1
0.8
11
0.9
1
Figure 3. Yields of tars vs. aromaticity of coals.
heating rate, as applied in the present work (Tyler, 1980). The structure of low rank Australian coals must, then, be markedly different from that of Canadian coals. For example, the content of reactive exinites in the Australian coals is usually high (= 15%). Also, chemical structure of the exinites must be different. Thus, despite the high content of exinites in Onakawana lignite (= 15%), the yield of tar was very low compared to that of CO, COz,COP,and reaction HzO. This suggests, that reactions in which such products are formed are important in the overall conversion of the Onakawana exinites during pyrolysis.
Ind. Eng. Chem. Prod. Res. Dev., Vol. 23,No. 1, 1984 137 I
15 0
* 0.7' a
U
+
LL
0
0.5
0.6
0.7
0.s
f a OF COALS Figure 4. Aromaticity of coals vs. aromaticity of tars.
Table IV. Distribution of Aliphatic Hydrogen in Liquid Products (Fractions of Total Aliphatic Hydrogen) hydrogen type coal feed McIntyre Byron Creek Sukunka Balmer Coalspur Shaughnessey Devco Prince E. Blackfoot Bienf ait Onakawana
benzylic methylene methyl
0.57 0.43 0.59 0.55 0.41 0.41 0.39 0.46 0.40 0.32 0.36
0.29 0.40 0.28 0.31 0.43 0.45 0.43 0.39 0.44 0.50 0.47
0.14 0.17 0.13 0.14 0.16 0.14 0.18 0.15 0.16 0.18 0.16
The yields of tar depend also on the aromaticity of coal feedstocks. In fact, a linear correlation is obtained for the group of coals which includes the semianthracite and bituminous coals (Figure 3). Tar yields from the low rank materials are markedly lower in agreement with results shown in Figure 2. Chemical Composition of Tars. Results of elemental analyses and NMR distribution of carbon and hydrogen of the tars are shown in Table 111. The effect of coal structure on the chemical composition of tar is evident from the correlation of f a values of coals and tars shown in Figure 4. The nonlinearity of the correlation is maintained even when eastern coals are excluded. Materials on the low fa end of the curve (Onakawana, Bienfait, and Blackfoot) may exhibit some similarities in their structures. The same should apply for the materials on the high fa end of the curve (Sukunka, Balmer, and McIntyre). In the middle of the fa range four tars have almost identical aromaticities despite different origin of the corresponding coals, i.e., two eastern and two western (Coalspur and Shaugnessey) coals. Distribution of aliphatic hydrogen in tars are shown in Table IV. The results are expressed as fractions of total aliphatic hydrogen. The fraction of benzylic hydrogen also includes methyl groups attached to aromatic rings, while methylene hydrogen includes methylene and hydroaromatic groups in /3 or positions more distant to the aromatic ring. The last column shows the fractions of hydrogen in methyl groups not attached to aromatic rings. It is evident that the presence of C& in coal is essential for the production of tars from bituminous coals. This is confirmed by the decreasing yield of tars with increasing fa of coals (Figure 3). Among different types of H& and
oh 0.2
1
I
0.3
0.4
0.5
FRACTION OF METHYLENE HYDROGEN
Figure 5. Yields of tars vs. the amount of methylene hydrogen in tars.
C& the methylene groups in positions /3 or more distant to aromatic rings as well as those in hydroaromatic rings are the most important. This follows from the linear correlation of the methylene hydrogen content vs. yield of tars shown in Figure 5. The six points in the correlation, showing little deviation from linearity, are those for tars obtained from western coals. On the other hand, Devco and Prince coals give larger yields than one would expect from the correlation established for western coals. Also in this case yields of tars from pyrolysis of lignites deviate markedly from linearity. It appears that tars derived from Sukunka, Balmer, and McIntyre coals are of similar chemical composition. For example, the contents of benzylic and methylene hydrogen (Table IV)are similar. However, low H&/C& ratio of the Balmer tar (2.2) compared to that of Sukunka and McIntyre (2.7 and 2.9, respectively) indicate that short methylene links joining aromatic structures and/or hydroaromatic carbon are important in the former. On the other hand, methyl or short chains ending with methyl may account for a large portion of Cg in the other two tars. Further, phenols were detected in these tars only in trace quantities despite the presence of oxygen. The fate of 0 is, then, unknown though the presence of furanic, aryl ether, and quinone structures is assumed. Although aromaticities of four tars in the middle of the fa range are almost identical, the structure of major components differ markedly. Thus large quantities of phenols in tars from the Coalspur and Shaughnessey coals account for about half of the 0 in the tars. On the other hand, phenols were detected only in trace quantities in tar from Prince coal despite relatively high 0 content. Furanic rings, aryl ethers, and quinones may account for most of the 0 in this coal. A simple calculation confirms that phenols account for most of the 0 in tar from the Devco coal. However, the content of phenol in this coal is rather low compared to that in the two western coals. Properties of tars obtained from low rank materials such as Blackfoot and Onakawana appear to be similar to those of tars with fa of about 0.60. The tar derived from Bienfait lignite differs with respect to the content of Hdi and the values of H,,/C,, as well as Hdi/Cdi ratios. Lower H,/C, and higher H&/C&ratio of the tar compared to the other tars indicate more extensive ring substitution,
138
Ind. Eng. Chem. Prod. Res. Dev., Vol. 23,No. 1, 1984
Table V. Coinposition of Gaseous Products (vol % )
1. Canmore 2. McIntyre 3. Byron Creek 4. Sukunka 5. Balmer 6. Coalspur 7. Shaughnessey 8. Devco 9. Prince 10. E. Blackfoot 11. Bienfait 12. Onakawana
2.9 1.5 3.5 2.1 1.9 14.0 12.6 3.0 3.3 14.4 6.5 10.2
5.3 1.9 7.6 2.5 2.4 12.1 9.0 2.5 8.8 23.4 29.3 46.5
possibly by methyl or short aliphatic groups. The total amount of heteroatoms is an important criterion for commercial utilization of tars. A high heteroatom content would require extensive hydrotreatment to produce commercial fuels. Large quantities of phenols in some tars may justify their isolation on a commercial scale. The tar from the Devco coal is the most suitable for production of fuels while the tars obtained from Coalspur and Shaughnessey coals might be potential feedstocks for the production of phenols. Properties of Gaseous Products. The composition of gaseous products, expressed on an air-free basis for samples taken at 530 "C are listed in Table V. Air dilution was not significant and in no case exceeded 10%. The composition of samples taken at lower temperatures was rather different. For example, samples taken at 500 "C contained more C3to C5hydrocarbons, H2S, and C 0 2 and less H2and CHI than the gases formed at 530 "C with the exception of Balmer coal, where CHI content in gases formed at 500 "C was higher than that in gases formed at 530 "C. This suggests that the temperature maximum for C H I production, generally observed during pyrolysis of coal (Makino and Toda, 1979), is lower for Balmer coal than for the other coals. The CO contents were also lower in gases formed at 530 "C with the exception of Coalspur and Shaughnessey coals. The amount of C2 hydrocarbons decreased with temperature with the exception of McIntyre and Sukunka coals. Alkenes were formed in small quantities compared to alkanes confirming that primary products did not undergo significant changes (Furimsky et al., 1983). The quality of gaseous products depends on the rank of coal feed. Thus, inert C02 is formed in large quantities during pyrolysis of low rank coals. With increasing rank the content of C 0 2in gas mixtures tend to decrease. Gas products from several coals investigated contained up to 95% of H2 + hydrocarbons. This constitutes a valuable mixture which can be either used as a petrochemical feedstock or efficiently burnt as a fuel contributing to the overall economy of coal pyrolysis. Coal Models and Mechanisms of Pyrolysis. Structural models of coal combined with detailed evaluation of volatile products, may provide some information on pyrolytic reactions. This assumes that molecules contained in pyrolysis products retain some of their original configurations as part of coal molecules. This approach is especially valid for the low rank coals investigated in the present work, mainly because of the almost identical aromaticities of the tars and corresponding feeds. Similar trends were also observed for three low volatile bituminous coals. The structure of tars can be related directly to the structure of coals. An extrapolation of the structure of tars obtained from Sukunka, McIntyre, and Balmer coals to the coals suggests
37.3 21.1 23.9 27.0 23.5 15.3 19.9 27.8 10.8 13.6 27.1 13.8
0.1 0.3 0.5 0.4 0.3 0.3 1.7 1.8 3.5 0.6 0.2 0.3
46.4 53.1 49.3 48.0 52.8 43.4 48.0 52.8 53.2 36.7 32.9 23.0
6.4 14.1 10.0 12.9 14.1 9.3 7.3 9.7 11.6 7.1 2.3 3.3
1.7 7.6 4.2 7.0 4.8 5.5 3.3 2.4 7.9 4.3 1.6 2.5
that benzylic hydrogen and carbon account for a large portion of total C& and H& Among known coal models, the one proposed by Given (1960) depicts such situation best. Single CH, bridges joining aromatic structures represent an important part of the model molecule. The cleaving of Cmo-C&bonds may be an essential step in the overall mechanism of pyrolysis. Volatile species released in this step will, most likely, retain their original aromatic structure. Phenols were detected in these tars only in trace quantities despite the presence of 0 in these coals. Thus, phenolic OH groups, if present, were converted during pyrolysis either to furanic ring containing structures or aryl ether links (Furimsky et al., 1983). The positions of the OH groups depicted in Given's model make such transformation favorable. Tars derived from Coalspur, Shaughnessey, Prince, and Devco coals exhibit almost the same aromaticities (==0.6) though the fa values of the coals vary between 0.65 and 0.71. This difference in the fa values of tars and corresponding coals can be explained by using Wiser's model of coal (Davidson, 1980). This model includes various cyclic fragments containing up to seven fused rings joined either by methylene bridges or by 0 and S links. It is believed that fragments containing a small number of rings (either aromatic or naphthenic) together with aliphatic and olefinic species which originate from long methylene chains and long ring substituents will, on pyrolysis, become a part of the tar. Fragments containing a large number of fused rings will remain in the pyrolysis char. Similar fa values for tars derived from the four high volatile bituminous coals do not ensure similar tar molecule structures. Namely, tars from Coalspur and Shaughnessey coal differ markedly from tars derived from Devco and Prince coals. The presence of phenols in high concentrations in tars suggests that phenolic 0 accounts for a large portion of total 0 in Coalspur and Shaughnessey coals. Surprisingly, only traces of phenols were detected in tar from Prince coal although the content of 0 is similar to that of Devco coal. A large portion of 0 in the former must, then, be in a nonphenolic form. On the basis of these differences Wiser's model may be modified. Increasing the number of phenolic OH groups will give a better reflection of the structure of Coalspur and Shaughnessey coals and keep the number of such groups at a minimum in the case of Prince coal. Similar contents of 0 in these two eastern coals and a marked difference in the content of phenols in corresponding tars suggests that the main difference in the structure of these coals is the different form of 0 groups. The large content of mineral matter in Prince coal may have had a catalytic effect on conversion of phenolic groups. Little difference was observed between fa values for tars and corresponding lignites. This can be explained by using
Ind. Eng. Chem. Prod. Res. Dev., Vol. 23, No. 1, 1984 139 OH
\ '\CH
/
NCH D
O
c
H
3
/
COOH
I H2
COOH%
I
/
I
0
H
C H20 H
Figure 6. Approximate structure of lignite molecule.
a model proposed by Ludwig et al. (1964). This model includes numerous phenylpropane entities held together by etheric links with the aliphatic part being highly substituted by hydroxyl and carbonyl groups. In its original form this model cannot explain large yields of COz formed during pyrolysis. To account for this, a modified version of the model is shown in Figure 6. A cleavage of weak nonaromatic bonds will release an entity retaining most of its original form. The small size of the entities suggests that conversion to tar should be high. Low yields of tars are attributed to extensive polymerization of radical intermediates to char. This is caused by a limited amount of available hydrogen to stabilize the radicals. Also, a portion of this hydrogen is consumed in reactions in which OH groups are eliminated as HzO. The above discussion indicates that coal structure has a decisive effect on yields of tars. It has been established that the presence of methylene and hydroaromatic structures is essential for high yields of tars. These structures provide cracking sites as well as labile hydrogen needed to stabilize radical intermediates. A large content of 0 in low rank coals, particularly unstable hydroxyl, etheric, carboxyl groups attached to aliphatic carbon, will decompose readily during pyrolysis. This will result in high concentrations of intermediate species. To prevent polymerization to char, high concentrations of stabilizing agents (hydrogen or hydrogen donor) are needed. A large amount of 0 in coal is not necessarily the main reason for low yields of tar. For example, tar yields from Coalspur and Shaughnessey coal were high despite high 0 content in these coals. Large quantities of phenols in the tars indicate high concentrations of phenolic OH in these coals. Such groups have a better chance of surviving pyrolytic conditions than those attached to aliphatic carbon. It is known that mineral matter, particularly calcium, can affect the yield of pyrolytic products. Thus, the addition of CaO to coal prior to its pyrolysis resulted in a decrease in tar yields (Franklin et al., 1982). Contents of CaO (on dry basis) in the low rank materials, i.e., Onakawana (4.0%), Blackfoot (1.4%), and Bienfait (2.3%),are markedly higher than in any of the coals with f a less than about 0.7 used in this work (Tibbetts et al., 1978). Apparently, Ca ions catalyze repolymerization of primary products to char and thus affect tar yields. Significance of the Experimental Correlations. The developed correlations are useful tools to select the best or to eliminate the least suitable coals in a series of coals as feedstocks for production of hydrocarbon liquids by pyrolysis. The reliability of such correlations may increase by including well-tested referee coals. The Fischer assay, in all its simplicity, appears to be a useful method for a
relatively rapid assessment of a series of coal. The yields of tars are generally lower when compared to those obtained in processes based on a rapid pyrolysis technique. A direct relation may be established between yields from rapid pyrolysis and those obtained from the tests conducted under similar conditions as applied in the present work. According to Tyler, tar yields from rapid pyrolysis are about twice as high as those obtained during the slow heating pyrolysis used in the present work (Tyler, 1980). Interesting information may be extrapolated from the high rank side of the correlations. For example, a coal having H/C ratios of about 0.4 or lower and aromaticity of about 0.9 should contain little volatile matter. No volatile products should be formed, under the pyrolysis conditions applied in the present study for coals with H/C ratios lower than 0.5 or coals containing less than 10% volatile matter as determined by standard methods. Conclusions The presence of aliphatic carbon, presumably in a naphthenic form or as methylene bridges, is essential for high tar yields during coal pyrolysis. Such structures provide potential sites for cracking large coal molecules. Methylene and hydroaromatic hydrogen effectively participate in the stabilization of reactive intermediates, thus preventing their conversion to char. The presence of 0 in coals is generally believed to have an effect on tar yields during pyrolysis. Very low yields of tar from low rank coals studied in this work coincide with high contents of 0. A substantial portion of O-containing groups in these feeds is assumed to be attached to aliphatic carbon. During pyrolysis such groups decompose readily producing HzO, CO, and COz rather than contributing to tar yields. In contrast, coals having a large portion of total 0 in phenolic form may still give large tar yields even though the 0 content is comparable to that in low rank coals. This was observed for Coalspur and Shaughnessey coals. The form of 0 groups may be more important than total 0 content in coals with respect to yields of tars from pyrolysis. In the series of coals investigated in this work, a linear correlation between tar yields and coal structures expressed in terms of H/C ratios and aromaticities was found to apply only to bituminous coals. This differs from the observations made with Australian coals. Among the tested coals, the two from eastern Canada (Devco and Prince) are most suitable for production of liquid fuels by pyrolysis. Among the coals from western Canada, Coalspur and Shaughnessey give relatively good yields of tars. The high concentration of phenols, however, means that extensive hydrotreatment is required to produce commercial fuels. The main difference between the two eastern and the two western coals is the higher concentration of phenolic structures in the latter. Literature Cited Andrew, E. R.; Farrell. L. F.; Firth, M.; Gledhlil, T. D.; Robert, I . J . Magn. Reson. 1080, 1 , 27. Anthony, 8. D.; Howard, J. B. A I C h € J . 1075, 22(4), 625. Collln, 0.;Zander, M. €r&l Kohk 1080, 33(12), 557. Cortez, D. H.; La Delfa, C. J. Hydrocarbon Process. 1081, 2 , 11 1. Davldson, R. M. "Molecular structure of coal"; IEA Coal Research Report No. ICTIS/TR 08; 1980. Edwards, J. H.; Smith, 1. W.; Tyler, R. J. Fuel 1080, 59(10), 681. Franklin, H. D.; Peters, W. A.; Howard, J. B. Fuel 1082. 61(2), 155. Furlmsky. E.; MacPhee, J. A.; Vance, L.; Clavagila, L. A.; Nandl, B. N. Fuel 108g960, 4. Furimsky. E.; Ripmeester. J . Fuel Process Techno/. 1083, In press. Glven, P. H. Fuel 1080, 39. 145. Herring, I. W.; Tollefson, E. L. Can. J . Pet. Technnol. 1080, 2 , 93. Ludwig, C. H.; Nist, B. J.; McCarthy, J. L. J . Am. Chem. SOC. 1064, 86, 1196. Maklno, M.; Toda, Y. Fuel 1070, 58, 574. Ozubko. R. S.; Clugston, D. M.; Furlmsky, E. Anal. Chem. 1081, 53, 183. Pines, A.; Glbby. M. G.; Waugh, J. S. J . Chem. Phys. 1073, 59, 569.
140
Ind. Eng. Chem. Prod. Res.
Saxby. J. D. Fuel 1980, 59(5), 281. S*efer, J.; Steikal, E. 0.; BucMahl. R. Macromolecuks 1977, 10, 384. Schulz. H. Pure AD@. Ghem. 1979, 51, 2225. Scott, D. S. “Investigation of the Short Residence Time Fyolysis of Coal”, Final Report DSS Canada, File No. 07SE-23440-&9105. Tagenfeldt, T.; Haeberlen. U. J . M a p . Reson. 1979, 36, 453. Tibbetts, T. E.;Montgomery, W. J.; Faurschou. D. K. “Analysis Directory of
Dev. 1984, 23, ~
140-143
Canadian Commercial Coals”; CANMET Report 78-7; 1978 ~ R , J, Fuel i Igso, ~ 59(4), ~ ,218,
Receiued f o r review April 28, 1983 Accepted August 15, 1983
Potentiometric Titration of Sulfate with Lead and Barium Ions with Various Indicating Electrodes Walter S. Sellg Lawrence Llvermore National Laboratoty, Unlverslty of Callfornla, Llvermore, Callfornie 94550
Several types of graphite were used as sensors in the potentiometric titration of 25 to 75 p t o l of sulfate vs. lead(1I) and barlum(I1) and compared with tkratlons obtained with a lead ion-selective electrode (ISE). Pyrolytic graphite and highdenslty graphite, conditioned In neutral potassium permanganate, were found to be good alternatives to the lead ISE. A qualitative study was made of a variety of commercially available ISE’s and other materials as sensors in the titration of 5 pmol of sulfate vs. lead(I1). Every ISE and conducting material tested yielded a usable response. while that of the commonly used lead ISE was largest, some other ISE’s and metal rods also function satisfactorily as sensors in this titration. All titrations were carried out in a partially nonaqueous medium, which is required even for the lead ISE at the low sulfate levels lnvestlgated.
Introduction Ion chromatography is probably the most useful tool for the analysis of sulfate in water and other matrices (Buchholz et al., 1982; Roberts et al., 1981). It is of particular advantage because other anions of interest can be determined simultaneously on a single sample. The equipment is, however, costly and not readily available in the field. We should like to demonstrate in this paper that it is indeed possible to devise simple, inexpensive methods for the analysis of sulfate (and possibly other ions in a similar fashion) which can be used in the field and require only equipment readily available in any laboratory, such as a pH meter and burets. The method described is based on the potentiometric titration of sulfate with barium or lead ions. The advent of ion-selective electrodes (ISEs) has led to a renaissance in analytical potentiometry in recent years. However, no satisfactory electrode is available for the direct determination of sulfate ion. While it is relatively easy to achieve a Nernstian response to sulfate, it is extremely difficult to achieve high selectivity (Rechnitz et al., 1967). The lead ISE has, however, been successfully used to monitor the potentiometric titration of sulfate vs. lead ions ( h s and Frant, 1969; Selig, 1970; Mascini, 1973) and vs. barium ions (Harzdorf, 1972; Selig, 1975). In every instance it has been found that end points were enhanced by using a partially nonaqueous medium to reduce the solubility of the precipitated sulfate species. Mixed oxide electrodes were recently used (Schumacher et al., 1982) to monitor the precipitation titration of sulfate vs. lead(I1) or barium(I1) in partially nonaqueous media. Our recent work (Selig, 1983) has shown that the potentiometric titration of fluoride vs. lanthanum(II1) or thorium(1V) can be monitored with sensors other than the fluoride ISE. Indeed, even platinum as well as a variety of graphite sensors can accomplish this purpose. In this 0196-4321/84/1223-0140$01.50/0
Table I. Response of Various Electrode Couples in Titration of Sulfate vs. Pb(I1) in 80% Methanol reference
results
lead ISE (standard)=
sensor
single-junction
pyrolytic graphite pyrolytic graphite
single-junction vitreous carbon
pyrolytic graphite vitreous carbon high-density graphite e spectroscopic graphite
platinum single-junction single-junction single-junction
symmetrical, sharp breaks sharp breaks small breaks, dip prior t o end point small breaks sharp breaks sharp breaks breaks smaller than above
Orion Research Inc., Model 94-82, cost $350. B. F . Goodrich Supertemp, cost $60.25 (in lots of 12). Beckman Instruments platinum thimble, Model 39271, Sigri Corp. K rod, cost $40 (in lots of 12). cost $40. e Poco eraphite Inc., Decatur, TX, Model AXF-9QBG1, cost $6.31. Ultra Carbon Corp, Bay City, MI, Model UF-4S, cost $1.29.
paper we report the results of our investigation of sensors other than those used heretofore in the titration of sulfate vs. lead and barium ions. Experimental Section The sensing electrodes that we tested are listed in Table I (footnote). The graphite sensors, with the exception of vitreous carbon, were 152.4 mm (6 in.) long and 6.35 mm (1/4 in.) in diameter. The vitreous carbon sensors were of the same length but 3 mm in diameter. The reference electrode was a single-junction Orion No. 90-01 Ag/AgCl electrode with a salt-bridge containing 0.1 N NaN03. The graphite sensors were attached to the potentiometer by means of an alligator clip. Sodium sulfate was Baker Analyzed reagent. The titrants were 0.01 and 0.002 M lead nitrate and 0.01 M 0 1984 American Chemical Society