- Email: [email protected]
Investigation of organic sulphur forms in coals by high pressure temperature-programmed reduction
Investigation of organic coals by high pressure temperature-programmed Christopher J. Lafferty, Colin E. Snape
Stuart
sulphur
forms
in
reduction
C. Mitchell,
Roberto
Garcia
and
Department of Pure and Applied Chemistry, University of Strathclyde, 1 homas Graham Building, 295 Cathedral Street, Glasgow Gl IXL, UK (Received 16 April 7992)
reduction procedure used thus far to investigate the distributions of organic sulphur forms in coals suffers from only a small proportion of thiophenic sulphur actually being observed and from the likelihood of secondary reactions which convert sulphides into thiophenes. These shortcomings have been largely overcome by using a well-swept fixed-bed reactor at relatively high hydrogen pressures (up to 15 MPa) in conjunction with an effective hydrodesulphurization catalyst, sulphided molybdenum. The technique has been applied to three high sulphur lignites, namely Mequinenza (Spain), Rasa (Yugoslavia) and Cayirhan (Turkey) and, for the first two samples, the results have been compared with those from X-ray and other pyrolysis techniques. For all three lignites, thiophenic sulphur is the dominant form with sulphides accounting for 2&35% of the total organic sulphur; the proportions of the latter are, in general, significantly lower than those recently obtained by X-ray techniques and fluidized-bed pyrolysis for the
The temperature-programmed
Mequinenza
and Rasa samples.
(Keywords: organic sulphur; coal; derivatives)
Organic sulphur occurs in coals and petroleum source rocks in both thiophenic and non-thiophenic forms; the latter are thought to comprise aliphatic and aromatic thioethers, thiols and disulphides. Information on the distribution of these forms in coals is needed to improve our current understanding of devolatilization and liquefaction behaviour lp3 . Similarly, organic sulphur plays a key role in oil generation4 but the mechanisms are not fully understood because information is lacking on the forms present. As yet, there is no single established procedure for the speciation of organic sulphur forms in coals’ and petroleum source rocks but X-ray7-” and reductive’ l-l6 techniques have recently received considerable attention. Both X-ray photoelectron spectroscopy (XPS)‘,” and X-ray absorption near-edge structure (XANES)6-8 have been used to distinguish between thiophenic and non-thiophenic sulphur forms in coals but, with both these techniques, the energy differences between these two sulphur forms are relatively small (0.3-0.5 eV in XPS and 0.5 eV for the sulphur K-edge energies in XANES) meaning that there is virtually no visible resolution. Therefore, both these X-ray techniques are heavily reliant on curve-fitting procedures which obviously reduces the accuracy of the results. Further, the number of model sulphur compounds used for calibration thus far is relatively small and possible variations in peak widths in XPS have not been considered. Nonetheless, the use Presented at ‘Environmental Aspects of Coal Utilization and Carbon Science Conference’, 31 March-2 April 1992. University of Newcastle upon Tyne. UK
001&2361/93/03036745 (‘ 1993 Butterworth-Heinemann
Ltd.
of selective oxidation in which non-thiophenic sulphur forms only are converted into sulphoxides and sulphones (S” and Sv’ species) vastly improves the resolution in both XPS and XANES. Results to date from both techniques suggest that low-rank coals contain between 30 and 70% and this decreases with increasing sulphidic sulphur6-lo rank. There is some confusion in the nomenclature used for non-thiophenic (sulfidic) sulphur forms which have been referred to as solely aliphatic sulphides in more than one instance’,“. It is our interpretation of the X-ray data on model compounds that both aromatic and aliphatic sulphides count as non-thiophenic sulphur forms. Temperature-programmed reduction (TPR) is based on the principle that different organic sulphur forms present in solid fuels have different characteristic reduction temperatures at which hydrogen sulphide (H,S) evolves. Calibration with model compounds has indicated that the ease of reduction is in the order of thiols > aliphatic sulphides > aromatic sulphides > thiophenes. The method for coals was pioneered by Attar’l.12 and has been used by others13,‘4 with no modifications to the original design of the reactor in which coal is refluxed in a mixture of low-boiling solvents and H,S is swept from the reactor by a stream of inert carrier gas; a condenser prevents the escape of tar. Attar originally used lead acetate paper to detect H,S but later workers have used potentiometry13 and flame photometric detection14. However, only limited success has been achieved thus far, primarily because only the labile non-thiophenic forms have actually been observed; sulphur balances have been poor with virtually all the thiophenic sulphur remaining in the char due to the low pressures and inappropriate use of
FUEL, 1993, Vol 72, March
367
Organic sulphur forms in coals: C. J. Lafferty et al. Table 1
Elemental
analyses
of the lignites
Mequinenza C (% daf) H (% daf) N (% daf) S (% db) Total Pyritic Sulfatic Organic
Rasa
Cayirhan
66.4 5.8 1.6
80.2 5.2 1.2
64.3 5.1 1.8
9.0 0.5 0.5 8.0
11.8 0.4
4.8 1.0 0.9 2.9
low-boiling compounds, such as tetralin. Furthermore, no account has been taken of the reduction of pyrite to pyrrhotite and retrogressive reactions including the conversion of sulphides into thiophenes, which are extremely likely due to the long residence time of tar in the reactor. The extent of such reactions can be reduced considerably by using well-swept reactors, such as a fluidized bed as used by Calkins and co-workers10.15. on-thiophenic sulphur fo&s were detected and estimated rom the sulphur gases that evolved at given temperatures but thiophenic sulphur was again determined by difference as that remaining in the char. The extent of desulphurization in coal pyrolysis generally increases with hydrogen pressure”j and we recently reported that, in a fixed-bed reactor at a pressure of 15 MPa with a dispersed sulphided molybdenum (MO) catalyst, over 95% of the organic sulphur in both lignite and bituminous coal was released16. Moreover, only about 20% was released as thiophenic compounds in the tars with the remaining 75-80% appearing in the gas phase. This study clearly demonstrated in principle that the use of high hydrogen pressures in TPR could overcome the drawbacks discussed above. This paper describes the first use of high-pressure TPR where this novel approach is illustrated by application to three high sulphur low-rank coals. EXPERIMENTAL The ultimate analyses and the sulphur distributions for the three lignites investigated are listed in Table 1. The samples were chosen because of their high organic sulphur contents and, in the case of Rasa and Mequinenza lignites, their low pyrite contents and the fact that these curiosity samples have already been investigated by XPS nd fluidized-bed pyrolysis”. It was reported previously16 that the concentrations of elemental sulphur in Mequinenza and Cayirhan lignites were below 0.05% w/w. The samples were stored in n-hexane. Pyrite was removed from Cayirhan lignite by washing with dilute nitric acid at ambient temperatures, as described in BS 1016 and ASTM D2492. For tests with catalyst, the lignites were impregnated with ammonium dioxydithiomolybdate to give a nominal MO loading of 1% daf coal as previously described17.18. Prior to each test, the coals were dried in uacuo at 40°C. The high-pressure TPR apparatus is shown schematically in Figure 1. The reactor tube was smaller but otherwise identical to that used previously in fixed-bed hydropyrolysis studies (45.7cm compared to 106.7 cm, both with 14.3 mm o.d., Incoloy) 17,18. Between 1 and 3 g of coal (75-250 pm) mixed with sand (1:3 mass ratio) was loaded into the reactor and held in place with a steel-wool plug.
368
FUEL, 1993, Vol 72, March
The reactor tube was heated resistively from 100 to 600°C at 2°C min- I. Hydrogen pressures of 3 and 15 MPa and a nitrogen pressure of 3 MPa were used with superficial gas velocities in the range 0.2-0.5 m s-l. Tar was condensed in a high pressure trap cooled with ice/water. After pressure let-down through either a metering valve or a mass-flow controller, the gas stream was sampled through a 1.8 m length of heated capillary tubing at a rate of 25 cm3 min-’ into a quadrupole mass spectrometer (VG Monitor, &lo0 a.m.u.). The signal from the control thermocouple situated in the coal bed was also fed to the mass spectrometer facilitating the construction of direct plots of evolved gas concentrations against temperatures. As well as m/z of 34 and 36 for H,S (32S and 34S, respectively), characteristic m/z values for CH,SH, SO, and thiophene, together with those for a number of hydrocarbons were monitored, where appropriate. At the end of each run, tar was recovered from the trap in either dichloromethane (DCM) or toluene, and char was emptied from the reactor. RESULTS
AND DISCUSSION
General operation
Hydrogen sulphide adsorbs strongly on steel and is also likely to dissolve to a considerable extent in the condensed tar. These factors were undoubtedly responsible for the failure to return to the initial baseline concentration of H,S after the peak temperature of 600°C was reached when either a heating rate greater than 2”Cmin-’ or too much sample was used in commissioning tests. Furthermore, it is important to trap the tar using ice/water rather than dry ice where there is a high possibility that much of the H,S will condense in the trap. As a compromise for satisfactory operation at 15 MPa in terms of achieving adequate sensitivity with the quadrupole mass spectrometer and avoiding problems with H,S lag in the system, the mass of coal used was chosen to give an organic sulphur mass of 0.05-0.1 g. Obviously at lower pressures, where the vapour residence times in the system are much shorter
Figure
1
Schematic
diagram
of high-pressure
TPR apparatus
Organic
a
400
200 Temperature
600
(‘C)
sulphur
15 MPa with catalyst. The profiles comprise a dominant high-temperature peak centred at between 450 and 520°C and a leading low-temperature shoulder between 280 and 400°C. From the calibration studies at atmospheric pressure by Attar”“’ and Majchrowicz et al.‘j and our arlier pyrolytic desulphurization study16, we are confident hat the dominant peak is due to thiophenes and the low temperature shoulder is attributable to sulphides. However, there are subtle differences between the profiles. In going to the higher pressure of 15 MPa, the area of the thiophenic peak relative to the sulphidic peak increases due to more hydrogen sulfide being evolved. Also, as with the addition of catalyst at 15 MPa, the thiophenic peak shifts to a lower temperature but the position of the sulphidic peak remains unaffected. The H,S evolution profiles have been fitted to two components using the Redhead equation” and, at 15 MPa, sulphidic sulphur accounts for about 30% of that evolved as H,S. Correcting this value for the 20% thiophenic sulphur in the tars16, it is deduced that sulfidic sulphur accounts for about 25% of the total organic sulphur. There is the possibility that some sulphidic sulphur in aliphatic thioethers will be evolved as mercaptans. Indeed, for Mequinenza lignite, small amounts of CH,SH were detected but these were extremely small compared to the amounts of H,S contributing to the low-temperature peaks in the evolution profiles (Figure 2). Comparison Figure
200 Temperature
Temperature
400
600
(“C)
forms in coals: C. J. Lafferty et al.
of the lignites
3 shows the H,S evolution
profile for Rasa lignite obtained at 15 MPa pressure without catalyst. The profile is very similar to that obtained for Mequinenza lignite under identical conditions (Figure 2b) apart from the thiophenic sulphur peaking at a slightly higher temperature and the lowest temperature feature at 25&3Oo”C being slightly more prominent, although this could be partly attributable to the small amounts of pyrite in both lignites (see following). Fitting the trace again to two components gives a similar overall result as for Mequinenza lignite, with sulphides accounting for about 25-30% of the total organic sulphur. In general, these estimates are significantly lower than those determined from fluidized-bed pyrolysis, XPS and XANES by Calkins et al. lo,15 (Table 2), although both the X-ray techniques gave about 25% sulphidic sulphur for Rasa lignite. As mentioned earlier, no visible
(OC)
Figure 2 High-pressure TPR H,S traces for Mequinenza lignite showing fits into sulphidic and thiophenic components: (a) 3 MPa hydrogen without catalyst; (b) 15 MPa hydrogen without catalyst; (c) 15 MPa hydrogen with catalyst
and the gas phase concentrations of H,S and other volatiles are considerably higher, the lag problem is much less acute. Pyrolysis conditions Figure 2 compares
the evolution profiles for H,S in high-pressure TPR at 3 and 15 MPa without catalyst and
Temperature
(“C)
Figure 3 TPR H,S trace for Rasa lignite obtained pressure without catalyst
at 15 MPa hydrogen
FUEL, 1993, Vol 72, March
369
Organic sulphur forms in coals: C. J. Lafferty et al. Table 2 techniques
Comparison of sulphidic sulphur contents (as percentage of total organic sulphur)
High-pressure Fluidized-bed XPS” XANES”
TPR pyrolysis”
by different
Mequinenza
Rasa
Cayirhan
30 67 66 48
35 47 30 26
40 n.d.” n.d.’ n.d.b
“From Ref. 10, where these concentrations were reported as being aliphatic rather than total (aromatic plus aliphatic) sulphides b n.d. = not determined
resolution is observed between the two organic sulphur forms measured and the precision of the XPS and XANES measurement#-lo is probably at best ) 10% of the organic sulphur. The proportion of organic sulphur evolved in high-temperature fluidized-bed pyrolysis of Mequinenza lignite” (-60%) is somewhat more than the 40% at the lower temperature of 520°C in the well-swept fixed-bed reactor used hereI when nitrogen is used as the sweep gas. On the basis of model compounds, Calkins assigned all the evolved sulphur gases to aliphatic sulphides. However, it is probable that the desulphurization behaviour of solids is different to that of volatile compounds in fluidized beds because their residence times are much longer, implying that some thiophenic sulphur may be evolved as H,S. Cayirhan lignite contains a significant concentration of pyrite and this will clearly interfere with TPR much more than for Mequinenza and Rasa lignites. Our recent investigation of pyrolytic desulphurization’6 indicated that the reduction of pyrite to pyrrhotite would appear to be a function of the morphology of the pyrite and cannot therefore be taken as a uniform phenomenon for different coals. Figure 4 compares the H,S evolution profiles for Cayirhan lignite before and after removal of the pyrite. The resolution below 400°C improves markedly due to the removal of the feature associated with the transformation of pyrite to pyrrhotite. In addition, much of the unresolved hump above 400°C also disappears, and this could be associated with both the formation of pyrrhotite and its subsequent reduction to iron under high hydrogen pressures. After pyrite removal, the appearance of the TPR trace is broadly similar to those for Mequinenza and Rasa lignites (Figures 2 and 3) with thiophenes accounting for about 60% of the evolved H,S, but the lowest temperature feature at 300°C is much more prominent. This feature occurs at a higher temperature than the initial pyrite reduction peak at 280°C observed for the parent coal and is tentatively assigned to extremely labile aliphatic sulphides. Indeed, its presence is consistent with our recent study comparing the desulphurization behaviour of Mequinenza and Cayirhan lignites16 in which it was observed that, below 300°C in the volatile matter test, much more of the organic sulphur was lost from Cayirhan. Evidence for secondary reactions at low pressures In the traces for Mequinenza lignite (Figure 2), the
low-temperature sulphidic peak accounts for much less of the observed sulphur at 3 MPa than at 15 MPa (20% compared with 35%), suggesting that some sulphides are being lost via secondary reactions to thiophenes in low-pressure hydrogen. Figure 5 shows the TPR traces
370
FUEL, 1993, Vol 72, March
obtained with 3 MPa nitrogen for Mequinenza and Rasa lignites. Remarkably, the only feature remaining in both cases is a low-temperature peak at 300°C with no thiophenic sulphur being observed above 400°C. Comparison of Figures 2, 3 and 5 reveals that the peak observed in nitrogen occurs at a lower temperature than the sulphidic peaks centred at about 350°C in profiles obtained with hydrogen, and is probably attributable only to the more labile, possibly aliphatic, sulphides. Clearly, calibration measurements are needed to test this hypothesis, but this striking comparison illustrates qualitatively the importance of using high-pressure hydrogen to avoid secondary reactions. Calibration and future work
The results obtained thus far have demonstrated that high-pressure TPR gives considerably better resolution between sulphidic and thiophenic sulphur forms than XPS and XANES, and the results are considerably more quantitative than those derived from earlier attempts at atmospheric pressure TPR”-14. The technique will now be applied to bituminous coals, coal derivatives and petroleum source rocks. However, calibration poses a more serious problem for the well-swept high-pressure reactor than for the reactor pot system used initially 0.25 a
0.20E E if : 2
0.15-
ii m .r :
O.lO-
I
1
0.051 I
0
I
100
I
I
I
I
Temperature 0.25
0.05
,
I
300
200
400
I
I
I
500
600
(“Cl
b
0
1, 100
I 200
I
I 300
Temperature
II
400
I 500
I
(OC)
Figure 4 TPR H,S traces obtained for Cayirhan lignite at 15 MPa hydrogen pressure: (a) initial sample; (b) after washing with dilute nitric acid to remove pyrite
Organic sulphur forms in coals: C. J. Lafferty et al.
formed by reduction from their precursor salts at temperatures close to ambient and clearly may be more effective than sulphided MO, especially for the sulfidic sulphur forms. However, even more so than for sulfided MO, it will be important to keep the active metal concentration low as the corresponding metal sulfide will inevitably be formed from H,S released. To achieve high activities with extremely low metal loadings, the use of hydrous titanium oxides coated onto coal surfaces will be particularly advantageous*’ as the active metals are introduced by ion-exchange rather than impregnation.
ACKNOWLEDGEMENTS II
0
I
I
I
I
I
200
300
400
500
600
I
100
Temperature
(“C)
The authors thank the Science and Engineering Research Council for financial support (grant no. GR/G/26600). The sample of Rasa lignite was supplied courtesy of Dr C. White (Pittsburgh Energy Technology Centre). R. Garcia thanks FICYT for funding the work at the University of Strathclyde.
REFERENCES
1 0
100
I
200
I
300
Temperature
Figure 5 TPR H,S traces obtained at 3 MPa nitrogen
I
400
I 1
500
600
8
("Cl
9 for (a) Mequinenza pressure
and (b) Rasa
lignite
by Attar and others”-i4, where both model sulphur compounds and polymers with low melting points could be used. Clearly, only non-melting solids are suitable for calibrating the fixed-bed reactor used here. Sulfur compounds immobilized on silica offer a particularly attractive option for calibration. The use of these substrates to probe the effects of restricted motion in the solid state on the free radical chemistry of coal liquefaction has been pioneered by Buchanan and co-workers20.21 who have investigated the reactivity of a number of silica-immobilized diphenylalkanes. Indeed, we are currently collaborating with the Oak Ridge group to synthesize a number of silicaimmobilized sulphur compounds including diphenylsulphide and dibenzothiophene from the corresponding monohydric phenols, in order to calibrate the highpressure TPR apparatus. The sulphided MO catalyst used is formed at about 250°C and clearly has no catalytic effect at lower temperatures. Group VIII transition metals, such as Pd and Rh, which are effective hydrogenation catalysts, are
10
11 12 13
14
15 16 17 18
19 20 21 22
Snape, C. E. Fuel Process. Technol. 1987. 15, 257 Snape. C. E. Fuel 1991. 70, 285 and references therein Khan, M. R. Fuel 1989, 68, 1439 Kohnen, M. E. L., Sinninghe Damaste, J. S., Kock-van Dalen, A. C., ten Haven, H. L.. Rullkotter, J. and de Leeuw. J. W. Geochim. Cosmochim. Acta 1990, 54. 3053 Attar, A. Fuel 1978, 57, 201 Huffman, G. P., Mitra, S., Huggins. F. E., Shah, N., Vaidya. S. and Lu, F. Energy Fuels 1991. 5. 574 Huffman. G. P.. Shah, N.. Taghiei, M. M.. Lu, F. and Huggins, F. E. Prepr. Am. Chem. Sot. Div. Fuel Chem. 1991, 36 (3). 1204 Kelemen, S. R.. George, G. N. and Gorbaty, M. L. Fuel 1990, 69, 939 Gorbaty. M. L.. George. G. N. and Kelemen, S. R. Fuel. 1990. 69, 945. 1065 Calkins, W. H., Torres-Ordofiez, R. J., Jung, B., Gorbaty, M. L.. George, G. N. and Kelemen, S. R. Proceedings of the 1991 International Conference on Coal Science, ButterworthHeinemann, Oxford, 1992, p. 985 Attar, A. in ‘Analytical Methods for Coal and Coal Products’, Vol. III. Academic Press, New York, 1979, Ch. 56 Altar, A. DOE/PC/30145TI Technical Report, US DOE, 1980 Majchrowicz, B. B., France, D. V.. Yperman, J., Reggers. G., Gelen, J.. Martens, J.. Mullens, J. and van Poucke. L. C. Fuel 1991.70.434 Dunstan, B. T. and Walker, L. V. Final report to Australian National Enernv Research Develonment and Demonstration Council, Canberra, 1988 Calkins, W. H. Energy Fuels 1987, 1, 59 Garcia. R.. Moinelo, S. R.. Lafferty, C. J. and Snape. C. E. Energy Fuels 1991, 5. 582 and references therein Snape, C. E. and Lafferty, C. J. Prepr. Am. Chem. Sot. Div. Fuel Chem. 1990.35 (l), 1 Snape. C. E., Lafferty. C. J.. Mitchell, S., Donald, F.. McArthur, C. A., Eglinton, G., Robinson, N. and Collier, R. Final Report, EC Project EN3V-0048-UK(H). Brussels, 1991 Redhead, P. A. Vacuum 1962, 12. 203 Buchanan, A. C. III, Britt, P. F. and Poutsma, M. L. Prepr. Am. Chem. Sot. Div. Fuel Chem. 1990. 35 (I), 217 Britt, P. F. and Buchanan, A. C. III J. Org. Chem. 1991,56,6132 Klavetter, E., Sylvester, A., Wilcoxon, J.. Lafferty, C. J., Mitchell, S. and Snape, C. E. Proceedings of the 1991 International Conference on Coal Science. ButterworthHeinemann. Oxford, p. 699
FUEL, 1993, Vol 72, March
371
Stuart
sulphur
forms
in
reduction
C. Mitchell,
Roberto
Garcia
and
Department of Pure and Applied Chemistry, University of Strathclyde, 1 homas Graham Building, 295 Cathedral Street, Glasgow Gl IXL, UK (Received 16 April 7992)
reduction procedure used thus far to investigate the distributions of organic sulphur forms in coals suffers from only a small proportion of thiophenic sulphur actually being observed and from the likelihood of secondary reactions which convert sulphides into thiophenes. These shortcomings have been largely overcome by using a well-swept fixed-bed reactor at relatively high hydrogen pressures (up to 15 MPa) in conjunction with an effective hydrodesulphurization catalyst, sulphided molybdenum. The technique has been applied to three high sulphur lignites, namely Mequinenza (Spain), Rasa (Yugoslavia) and Cayirhan (Turkey) and, for the first two samples, the results have been compared with those from X-ray and other pyrolysis techniques. For all three lignites, thiophenic sulphur is the dominant form with sulphides accounting for 2&35% of the total organic sulphur; the proportions of the latter are, in general, significantly lower than those recently obtained by X-ray techniques and fluidized-bed pyrolysis for the
The temperature-programmed
Mequinenza
and Rasa samples.
(Keywords: organic sulphur; coal; derivatives)
Organic sulphur occurs in coals and petroleum source rocks in both thiophenic and non-thiophenic forms; the latter are thought to comprise aliphatic and aromatic thioethers, thiols and disulphides. Information on the distribution of these forms in coals is needed to improve our current understanding of devolatilization and liquefaction behaviour lp3 . Similarly, organic sulphur plays a key role in oil generation4 but the mechanisms are not fully understood because information is lacking on the forms present. As yet, there is no single established procedure for the speciation of organic sulphur forms in coals’ and petroleum source rocks but X-ray7-” and reductive’ l-l6 techniques have recently received considerable attention. Both X-ray photoelectron spectroscopy (XPS)‘,” and X-ray absorption near-edge structure (XANES)6-8 have been used to distinguish between thiophenic and non-thiophenic sulphur forms in coals but, with both these techniques, the energy differences between these two sulphur forms are relatively small (0.3-0.5 eV in XPS and 0.5 eV for the sulphur K-edge energies in XANES) meaning that there is virtually no visible resolution. Therefore, both these X-ray techniques are heavily reliant on curve-fitting procedures which obviously reduces the accuracy of the results. Further, the number of model sulphur compounds used for calibration thus far is relatively small and possible variations in peak widths in XPS have not been considered. Nonetheless, the use Presented at ‘Environmental Aspects of Coal Utilization and Carbon Science Conference’, 31 March-2 April 1992. University of Newcastle upon Tyne. UK
001&2361/93/03036745 (‘ 1993 Butterworth-Heinemann
Ltd.
of selective oxidation in which non-thiophenic sulphur forms only are converted into sulphoxides and sulphones (S” and Sv’ species) vastly improves the resolution in both XPS and XANES. Results to date from both techniques suggest that low-rank coals contain between 30 and 70% and this decreases with increasing sulphidic sulphur6-lo rank. There is some confusion in the nomenclature used for non-thiophenic (sulfidic) sulphur forms which have been referred to as solely aliphatic sulphides in more than one instance’,“. It is our interpretation of the X-ray data on model compounds that both aromatic and aliphatic sulphides count as non-thiophenic sulphur forms. Temperature-programmed reduction (TPR) is based on the principle that different organic sulphur forms present in solid fuels have different characteristic reduction temperatures at which hydrogen sulphide (H,S) evolves. Calibration with model compounds has indicated that the ease of reduction is in the order of thiols > aliphatic sulphides > aromatic sulphides > thiophenes. The method for coals was pioneered by Attar’l.12 and has been used by others13,‘4 with no modifications to the original design of the reactor in which coal is refluxed in a mixture of low-boiling solvents and H,S is swept from the reactor by a stream of inert carrier gas; a condenser prevents the escape of tar. Attar originally used lead acetate paper to detect H,S but later workers have used potentiometry13 and flame photometric detection14. However, only limited success has been achieved thus far, primarily because only the labile non-thiophenic forms have actually been observed; sulphur balances have been poor with virtually all the thiophenic sulphur remaining in the char due to the low pressures and inappropriate use of
FUEL, 1993, Vol 72, March
367
Organic sulphur forms in coals: C. J. Lafferty et al. Table 1
Elemental
analyses
of the lignites
Mequinenza C (% daf) H (% daf) N (% daf) S (% db) Total Pyritic Sulfatic Organic
Rasa
Cayirhan
66.4 5.8 1.6
80.2 5.2 1.2
64.3 5.1 1.8
9.0 0.5 0.5 8.0
11.8 0.4
4.8 1.0 0.9 2.9
low-boiling compounds, such as tetralin. Furthermore, no account has been taken of the reduction of pyrite to pyrrhotite and retrogressive reactions including the conversion of sulphides into thiophenes, which are extremely likely due to the long residence time of tar in the reactor. The extent of such reactions can be reduced considerably by using well-swept reactors, such as a fluidized bed as used by Calkins and co-workers10.15. on-thiophenic sulphur fo&s were detected and estimated rom the sulphur gases that evolved at given temperatures but thiophenic sulphur was again determined by difference as that remaining in the char. The extent of desulphurization in coal pyrolysis generally increases with hydrogen pressure”j and we recently reported that, in a fixed-bed reactor at a pressure of 15 MPa with a dispersed sulphided molybdenum (MO) catalyst, over 95% of the organic sulphur in both lignite and bituminous coal was released16. Moreover, only about 20% was released as thiophenic compounds in the tars with the remaining 75-80% appearing in the gas phase. This study clearly demonstrated in principle that the use of high hydrogen pressures in TPR could overcome the drawbacks discussed above. This paper describes the first use of high-pressure TPR where this novel approach is illustrated by application to three high sulphur low-rank coals. EXPERIMENTAL The ultimate analyses and the sulphur distributions for the three lignites investigated are listed in Table 1. The samples were chosen because of their high organic sulphur contents and, in the case of Rasa and Mequinenza lignites, their low pyrite contents and the fact that these curiosity samples have already been investigated by XPS nd fluidized-bed pyrolysis”. It was reported previously16 that the concentrations of elemental sulphur in Mequinenza and Cayirhan lignites were below 0.05% w/w. The samples were stored in n-hexane. Pyrite was removed from Cayirhan lignite by washing with dilute nitric acid at ambient temperatures, as described in BS 1016 and ASTM D2492. For tests with catalyst, the lignites were impregnated with ammonium dioxydithiomolybdate to give a nominal MO loading of 1% daf coal as previously described17.18. Prior to each test, the coals were dried in uacuo at 40°C. The high-pressure TPR apparatus is shown schematically in Figure 1. The reactor tube was smaller but otherwise identical to that used previously in fixed-bed hydropyrolysis studies (45.7cm compared to 106.7 cm, both with 14.3 mm o.d., Incoloy) 17,18. Between 1 and 3 g of coal (75-250 pm) mixed with sand (1:3 mass ratio) was loaded into the reactor and held in place with a steel-wool plug.
368
FUEL, 1993, Vol 72, March
The reactor tube was heated resistively from 100 to 600°C at 2°C min- I. Hydrogen pressures of 3 and 15 MPa and a nitrogen pressure of 3 MPa were used with superficial gas velocities in the range 0.2-0.5 m s-l. Tar was condensed in a high pressure trap cooled with ice/water. After pressure let-down through either a metering valve or a mass-flow controller, the gas stream was sampled through a 1.8 m length of heated capillary tubing at a rate of 25 cm3 min-’ into a quadrupole mass spectrometer (VG Monitor, &lo0 a.m.u.). The signal from the control thermocouple situated in the coal bed was also fed to the mass spectrometer facilitating the construction of direct plots of evolved gas concentrations against temperatures. As well as m/z of 34 and 36 for H,S (32S and 34S, respectively), characteristic m/z values for CH,SH, SO, and thiophene, together with those for a number of hydrocarbons were monitored, where appropriate. At the end of each run, tar was recovered from the trap in either dichloromethane (DCM) or toluene, and char was emptied from the reactor. RESULTS
AND DISCUSSION
General operation
Hydrogen sulphide adsorbs strongly on steel and is also likely to dissolve to a considerable extent in the condensed tar. These factors were undoubtedly responsible for the failure to return to the initial baseline concentration of H,S after the peak temperature of 600°C was reached when either a heating rate greater than 2”Cmin-’ or too much sample was used in commissioning tests. Furthermore, it is important to trap the tar using ice/water rather than dry ice where there is a high possibility that much of the H,S will condense in the trap. As a compromise for satisfactory operation at 15 MPa in terms of achieving adequate sensitivity with the quadrupole mass spectrometer and avoiding problems with H,S lag in the system, the mass of coal used was chosen to give an organic sulphur mass of 0.05-0.1 g. Obviously at lower pressures, where the vapour residence times in the system are much shorter
Figure
1
Schematic
diagram
of high-pressure
TPR apparatus
Organic
a
400
200 Temperature
600
(‘C)
sulphur
15 MPa with catalyst. The profiles comprise a dominant high-temperature peak centred at between 450 and 520°C and a leading low-temperature shoulder between 280 and 400°C. From the calibration studies at atmospheric pressure by Attar”“’ and Majchrowicz et al.‘j and our arlier pyrolytic desulphurization study16, we are confident hat the dominant peak is due to thiophenes and the low temperature shoulder is attributable to sulphides. However, there are subtle differences between the profiles. In going to the higher pressure of 15 MPa, the area of the thiophenic peak relative to the sulphidic peak increases due to more hydrogen sulfide being evolved. Also, as with the addition of catalyst at 15 MPa, the thiophenic peak shifts to a lower temperature but the position of the sulphidic peak remains unaffected. The H,S evolution profiles have been fitted to two components using the Redhead equation” and, at 15 MPa, sulphidic sulphur accounts for about 30% of that evolved as H,S. Correcting this value for the 20% thiophenic sulphur in the tars16, it is deduced that sulfidic sulphur accounts for about 25% of the total organic sulphur. There is the possibility that some sulphidic sulphur in aliphatic thioethers will be evolved as mercaptans. Indeed, for Mequinenza lignite, small amounts of CH,SH were detected but these were extremely small compared to the amounts of H,S contributing to the low-temperature peaks in the evolution profiles (Figure 2). Comparison Figure
200 Temperature
Temperature
400
600
(“C)
forms in coals: C. J. Lafferty et al.
of the lignites
3 shows the H,S evolution
profile for Rasa lignite obtained at 15 MPa pressure without catalyst. The profile is very similar to that obtained for Mequinenza lignite under identical conditions (Figure 2b) apart from the thiophenic sulphur peaking at a slightly higher temperature and the lowest temperature feature at 25&3Oo”C being slightly more prominent, although this could be partly attributable to the small amounts of pyrite in both lignites (see following). Fitting the trace again to two components gives a similar overall result as for Mequinenza lignite, with sulphides accounting for about 25-30% of the total organic sulphur. In general, these estimates are significantly lower than those determined from fluidized-bed pyrolysis, XPS and XANES by Calkins et al. lo,15 (Table 2), although both the X-ray techniques gave about 25% sulphidic sulphur for Rasa lignite. As mentioned earlier, no visible
(OC)
Figure 2 High-pressure TPR H,S traces for Mequinenza lignite showing fits into sulphidic and thiophenic components: (a) 3 MPa hydrogen without catalyst; (b) 15 MPa hydrogen without catalyst; (c) 15 MPa hydrogen with catalyst
and the gas phase concentrations of H,S and other volatiles are considerably higher, the lag problem is much less acute. Pyrolysis conditions Figure 2 compares
the evolution profiles for H,S in high-pressure TPR at 3 and 15 MPa without catalyst and
Temperature
(“C)
Figure 3 TPR H,S trace for Rasa lignite obtained pressure without catalyst
at 15 MPa hydrogen
FUEL, 1993, Vol 72, March
369
Organic sulphur forms in coals: C. J. Lafferty et al. Table 2 techniques
Comparison of sulphidic sulphur contents (as percentage of total organic sulphur)
High-pressure Fluidized-bed XPS” XANES”
TPR pyrolysis”
by different
Mequinenza
Rasa
Cayirhan
30 67 66 48
35 47 30 26
40 n.d.” n.d.’ n.d.b
“From Ref. 10, where these concentrations were reported as being aliphatic rather than total (aromatic plus aliphatic) sulphides b n.d. = not determined
resolution is observed between the two organic sulphur forms measured and the precision of the XPS and XANES measurement#-lo is probably at best ) 10% of the organic sulphur. The proportion of organic sulphur evolved in high-temperature fluidized-bed pyrolysis of Mequinenza lignite” (-60%) is somewhat more than the 40% at the lower temperature of 520°C in the well-swept fixed-bed reactor used hereI when nitrogen is used as the sweep gas. On the basis of model compounds, Calkins assigned all the evolved sulphur gases to aliphatic sulphides. However, it is probable that the desulphurization behaviour of solids is different to that of volatile compounds in fluidized beds because their residence times are much longer, implying that some thiophenic sulphur may be evolved as H,S. Cayirhan lignite contains a significant concentration of pyrite and this will clearly interfere with TPR much more than for Mequinenza and Rasa lignites. Our recent investigation of pyrolytic desulphurization’6 indicated that the reduction of pyrite to pyrrhotite would appear to be a function of the morphology of the pyrite and cannot therefore be taken as a uniform phenomenon for different coals. Figure 4 compares the H,S evolution profiles for Cayirhan lignite before and after removal of the pyrite. The resolution below 400°C improves markedly due to the removal of the feature associated with the transformation of pyrite to pyrrhotite. In addition, much of the unresolved hump above 400°C also disappears, and this could be associated with both the formation of pyrrhotite and its subsequent reduction to iron under high hydrogen pressures. After pyrite removal, the appearance of the TPR trace is broadly similar to those for Mequinenza and Rasa lignites (Figures 2 and 3) with thiophenes accounting for about 60% of the evolved H,S, but the lowest temperature feature at 300°C is much more prominent. This feature occurs at a higher temperature than the initial pyrite reduction peak at 280°C observed for the parent coal and is tentatively assigned to extremely labile aliphatic sulphides. Indeed, its presence is consistent with our recent study comparing the desulphurization behaviour of Mequinenza and Cayirhan lignites16 in which it was observed that, below 300°C in the volatile matter test, much more of the organic sulphur was lost from Cayirhan. Evidence for secondary reactions at low pressures In the traces for Mequinenza lignite (Figure 2), the
low-temperature sulphidic peak accounts for much less of the observed sulphur at 3 MPa than at 15 MPa (20% compared with 35%), suggesting that some sulphides are being lost via secondary reactions to thiophenes in low-pressure hydrogen. Figure 5 shows the TPR traces
370
FUEL, 1993, Vol 72, March
obtained with 3 MPa nitrogen for Mequinenza and Rasa lignites. Remarkably, the only feature remaining in both cases is a low-temperature peak at 300°C with no thiophenic sulphur being observed above 400°C. Comparison of Figures 2, 3 and 5 reveals that the peak observed in nitrogen occurs at a lower temperature than the sulphidic peaks centred at about 350°C in profiles obtained with hydrogen, and is probably attributable only to the more labile, possibly aliphatic, sulphides. Clearly, calibration measurements are needed to test this hypothesis, but this striking comparison illustrates qualitatively the importance of using high-pressure hydrogen to avoid secondary reactions. Calibration and future work
The results obtained thus far have demonstrated that high-pressure TPR gives considerably better resolution between sulphidic and thiophenic sulphur forms than XPS and XANES, and the results are considerably more quantitative than those derived from earlier attempts at atmospheric pressure TPR”-14. The technique will now be applied to bituminous coals, coal derivatives and petroleum source rocks. However, calibration poses a more serious problem for the well-swept high-pressure reactor than for the reactor pot system used initially 0.25 a
0.20E E if : 2
0.15-
ii m .r :
O.lO-
I
1
0.051 I
0
I
100
I
I
I
I
Temperature 0.25
0.05
,
I
300
200
400
I
I
I
500
600
(“Cl
b
0
1, 100
I 200
I
I 300
Temperature
II
400
I 500
I
(OC)
Figure 4 TPR H,S traces obtained for Cayirhan lignite at 15 MPa hydrogen pressure: (a) initial sample; (b) after washing with dilute nitric acid to remove pyrite
Organic sulphur forms in coals: C. J. Lafferty et al.
formed by reduction from their precursor salts at temperatures close to ambient and clearly may be more effective than sulphided MO, especially for the sulfidic sulphur forms. However, even more so than for sulfided MO, it will be important to keep the active metal concentration low as the corresponding metal sulfide will inevitably be formed from H,S released. To achieve high activities with extremely low metal loadings, the use of hydrous titanium oxides coated onto coal surfaces will be particularly advantageous*’ as the active metals are introduced by ion-exchange rather than impregnation.
ACKNOWLEDGEMENTS II
0
I
I
I
I
I
200
300
400
500
600
I
100
Temperature
(“C)
The authors thank the Science and Engineering Research Council for financial support (grant no. GR/G/26600). The sample of Rasa lignite was supplied courtesy of Dr C. White (Pittsburgh Energy Technology Centre). R. Garcia thanks FICYT for funding the work at the University of Strathclyde.
REFERENCES
1 0
100
I
200
I
300
Temperature
Figure 5 TPR H,S traces obtained at 3 MPa nitrogen
I
400
I 1
500
600
8
("Cl
9 for (a) Mequinenza pressure
and (b) Rasa
lignite
by Attar and others”-i4, where both model sulphur compounds and polymers with low melting points could be used. Clearly, only non-melting solids are suitable for calibrating the fixed-bed reactor used here. Sulfur compounds immobilized on silica offer a particularly attractive option for calibration. The use of these substrates to probe the effects of restricted motion in the solid state on the free radical chemistry of coal liquefaction has been pioneered by Buchanan and co-workers20.21 who have investigated the reactivity of a number of silica-immobilized diphenylalkanes. Indeed, we are currently collaborating with the Oak Ridge group to synthesize a number of silicaimmobilized sulphur compounds including diphenylsulphide and dibenzothiophene from the corresponding monohydric phenols, in order to calibrate the highpressure TPR apparatus. The sulphided MO catalyst used is formed at about 250°C and clearly has no catalytic effect at lower temperatures. Group VIII transition metals, such as Pd and Rh, which are effective hydrogenation catalysts, are
10
11 12 13
14
15 16 17 18
19 20 21 22
Snape, C. E. Fuel Process. Technol. 1987. 15, 257 Snape. C. E. Fuel 1991. 70, 285 and references therein Khan, M. R. Fuel 1989, 68, 1439 Kohnen, M. E. L., Sinninghe Damaste, J. S., Kock-van Dalen, A. C., ten Haven, H. L.. Rullkotter, J. and de Leeuw. J. W. Geochim. Cosmochim. Acta 1990, 54. 3053 Attar, A. Fuel 1978, 57, 201 Huffman, G. P., Mitra, S., Huggins. F. E., Shah, N., Vaidya. S. and Lu, F. Energy Fuels 1991. 5. 574 Huffman. G. P.. Shah, N.. Taghiei, M. M.. Lu, F. and Huggins, F. E. Prepr. Am. Chem. Sot. Div. Fuel Chem. 1991, 36 (3). 1204 Kelemen, S. R.. George, G. N. and Gorbaty, M. L. Fuel 1990, 69, 939 Gorbaty. M. L.. George. G. N. and Kelemen, S. R. Fuel. 1990. 69, 945. 1065 Calkins, W. H., Torres-Ordofiez, R. J., Jung, B., Gorbaty, M. L.. George, G. N. and Kelemen, S. R. Proceedings of the 1991 International Conference on Coal Science, ButterworthHeinemann, Oxford, 1992, p. 985 Attar, A. in ‘Analytical Methods for Coal and Coal Products’, Vol. III. Academic Press, New York, 1979, Ch. 56 Altar, A. DOE/PC/30145TI Technical Report, US DOE, 1980 Majchrowicz, B. B., France, D. V.. Yperman, J., Reggers. G., Gelen, J.. Martens, J.. Mullens, J. and van Poucke. L. C. Fuel 1991.70.434 Dunstan, B. T. and Walker, L. V. Final report to Australian National Enernv Research Develonment and Demonstration Council, Canberra, 1988 Calkins, W. H. Energy Fuels 1987, 1, 59 Garcia. R.. Moinelo, S. R.. Lafferty, C. J. and Snape. C. E. Energy Fuels 1991, 5. 582 and references therein Snape, C. E. and Lafferty, C. J. Prepr. Am. Chem. Sot. Div. Fuel Chem. 1990.35 (l), 1 Snape. C. E., Lafferty. C. J.. Mitchell, S., Donald, F.. McArthur, C. A., Eglinton, G., Robinson, N. and Collier, R. Final Report, EC Project EN3V-0048-UK(H). Brussels, 1991 Redhead, P. A. Vacuum 1962, 12. 203 Buchanan, A. C. III, Britt, P. F. and Poutsma, M. L. Prepr. Am. Chem. Sot. Div. Fuel Chem. 1990. 35 (I), 217 Britt, P. F. and Buchanan, A. C. III J. Org. Chem. 1991,56,6132 Klavetter, E., Sylvester, A., Wilcoxon, J.. Lafferty, C. J., Mitchell, S. and Snape, C. E. Proceedings of the 1991 International Conference on Coal Science. ButterworthHeinemann. Oxford, p. 699
FUEL, 1993, Vol 72, March
371