Functional forms of nitrogen in coals and the release of coal nitrogen as NOx precursors (HCN and NH3)

Twenty-Fourth Symposium(International)on Combustion/TheCombustionInstitute, 1992/pp. 1259-1267

FUNCTIONAL FORMS OF OF COAL NITROGEN

NITROGEN IN COALS AND THE RELEASE A S NO~ P R E C U R S O R S ( H C N A N D NH3)

PETER E. NELSON, ALAN N. BUCKLEY AND MARTIN D. KELLY CSIRO Division of Coal and Energy Technology PO Box 136, North Ryde, Australia 2113

The influence of coal nitrogen on the pyrolytic release of NO~ precursors, (HCN and NH3), has been investigated for a range of Australian coals. Functional forms of nitrogen in the coals were determined by X-ray photoelectron spectroscopy. Pyrrolic-type nitrogen predominated (50-60% coal nitrogen), but pyridinic and quaternary forms were also detected. The proportion of pyridinic increased and that of quaternary decreased with increasing carbon contents of the coals. Four coals, ranging in rank from brown to bituminous and in nitrogen content from 0.6-2.0%, dry ash free basis, were pyrolyzed in a fluidized bed reactor at temperatures ranging from 500 to 1100~ C, gas residence times of 0.3-0.5 s and particle heating rates of 104 K s k Formation of HCN and NH3 occurred at higher temperatures than that required for maximum tar yields; cracking reactions of the tars are a probable source of HCN and NH3. However, the results show that release of nitrogen from structures which are not volatilised as tar also occurs. The major nitrogen-containing components of the tars were identified and quantified by gas chromatography. Nitrogen contained in pyridinic groups was more stable than that in pyrrolic groups; thus, a higher pyrolysis temperature was necessary to release nitrogen as HCN and NH3 from the pyridinic groups. For the range of coals and pyrolysis conditions studied here, however, nitrogen release (as a proportion of total coal nitrogen), in the form of NO~ precursors, did not depend on coal type or coal nitrogen content.

Introduetion Emissions of oxides of nitrogen (NOx) from the combustion of coals are a major environmental problem since they have been shown to contribute to the formation of both acid rain and photochemical smog. Operating conditions used in modern coalfired power plants effectively control the formation of NO~ by fixation of atmospheric nitrogen (thermal NOx);1 hence the major source (~85%) is due to conversion of nitrogen in the fuel (coal nitrogen levels usually fall in the range 0.5-2.0%, dry ash free (dat) basis). However, not all coal nitrogen is converted to NOx and there is no simple relationship between coal nitrogen content and NO~ emissions, although tests performed on a wide range of coals burnt under the same conditions have shown that the NO formed could be correlated with coal nitrogen, pyrolysis HCN yield and non-volatile nitrogen content. In order to improve understanding of the influence of coal nitrogen on NOx formation, we have investigated the functional forms of nitrogen in Australian coals by X-ray photoelectron spectroscopy (XPS), and in the volatiles produced by rapid heating of these coals. The pyrolytic release of nitrogen-containingvolatiles is the first step in the conversion of coal nitrogen to NOx.

The distribution of nitrogen in functional groups in coals is imperfectly understood, but nitrogen is believed to occur principally in heterocyclic moieties. z Recent studies 3-7 using XPS suggest that the majority occurs in pyrrolic- and pyridinic- type structures. However, the use of XPS to determine nitrogen functionality in coals suffers from some limitations and can only be justified because a nondisruptive bulk analytical approach has yet to be devised. In general, the surface composition of a material is different from the bulk composition because of interaction of a freshly exposed surface with the local environment. Consequently in order to deduce meaningful bulk chemical information from XPS results, it is essential for the surface characterised to be as representative as possible of the bulk. This requirement is usually met by the creation of a fracture surface under an inert atmosphere or in ultra high vacuum, or, for those materials which tend to fracture along a weakened interface, by abrasion. The presence of different nitrogen functional groups is revealed in the N(ls) photoelectron spectrum by peaks whose widths are comparable with the peak separation. It is essential, then, that N(is) spectra are determined in such a way as to obtain adequate resolution. In some of the previously

1259

COAL COMBUSTION

1260

reported studies of coals and coal-derived products, 3-7 powdered coal was mounted on double-sided adhesive tape. While this method is convenient, it is less than satisfactory because of the increased likelihood of differential charging, which leads to peak broadening and shifting, and because of the difficulty of preventing surface contamination and oxidation. In this study XPS analyses were performed on whole coal samples from which fresh surfaces were prepared by fracture or abrasion. Release of nitrogen from coals during pyrolysis has been studied in the pasts-l~ but inadequate attention has been given to the influence of nitrogen functionality on this process. Decomposition of pyrrole-type compounds to form HCN occurs at considerably lower temperatures, and with a fundamentally different mechanism to that of pyridinetype compounds. H-14 It is, therefore, conceivable that the relative amounts of these groups control the rate of release of nitrogen from the volatiles. In this study, results are presented for HCN and NH3 yields as a function of pyrolysis temperature under rapid heating conditions. In addition, the major nitrogen-containing species in the condensable volatiles (tars) have been identified, and the effects of pyrolysis temperature on their yields determined.

Experimental The coals studied were brown, subbituminous and bituminous coals from Australia. Table I gives the complete chemical analyses of these coals. XPS analyses were performed for all coals in Table I(b). The high levels of water in the brown and subbituminous coals made them unsuitable for XPS analysis. For XPS measurements, coal specimens consisting of single pieces shaped from a relatively large lump to an approximate size 8 x 8 x 2 mm were prepared as close as possible to the time of insertion into the glove-box attached to the preparation chamber of the spectrometer. Abrasion was carried out dry, but in such a way as to minimise local temperature rise, under argon flowing through the glove-box from the preparation chamber. Charging, and the extent of oxidation of the surface, was monitored by means of the C(ls) spectrum determined contemporaneously with the N(ls) spectrum. N(ls) spectra were fitted iteratively with the minimum number of Gaussian shaped peaks of unconstrained binding energy and width to produce a fit which was acceptable both visually and on the basis of the chi-square value. Coals chosen for the pyrolysis experiments were crushed and sized to 75-106 tzm. Previous work 15 has shown that tar yields obtained in fluidized bed reactors were constant for coal particle sizes of 53185/zm and 75-106/zm. The coals were pyrolyzed in argon or nitrogen at atmospheric pressure in a

fluidized bed reactor system, which has been described in detail previously. 16-17 Previous work 16 has shown that this system achieves heating rates of at least 104 K s-l; temperature could be varied in the range 500-1050~ C and gas residence times were in the range 0.3-0.5 s, depending on the total gas flowrates. Particle residence times varied, and depended on the mixing patterns in the reactor and on the tendency for the coal to soften during pyrolysis and to adhere to the bed material (zircon sand). Gaseous and condensed products from pyrolysis were characterized by elemental analyses of tar and char; by Fourier transform infrared (FI'IR) spectroscopy, for determination of yields of HCN and NH3; and gas chromatography (GC) with a nitrogen-phosphorous specific detector (NPD), for determination of yields of nitrogen-containing components of the tars. The details of these analyses have been reported previously, 18 apart from the determination of NH3. Yields of this species were determined by long path length (8 m) FTIR spectroscopy. Spectra were recorded at 0.25 cm -1 resolution by the coaddition of 256 scans, and the NH3 line at 1103 cm -1 was used. Calibrations were performed with an NH3 in He gas mixture diluted using mass flow controllers with N2 to the range of concentrations observed in the present work. The NH3 calibration was in excellent agreement with that reported by Freihaut et al. 19 Recovery and analyses of the condensed materials, or tars, were performed in the same way as previously described. 18 Nitrogen balances obtained for combined yields of nitrogen in gas, char and tar were usually in the range 100 ___ 10%. Results and Discussion

XPS Studies of Nitrogen Functionality in Coals: Representative N(ls) spectra are shown in Fig. la (Norwich Park) and Fig. lb (Bayswater). These spectra, and those from most other coals studied, could be fitted adequately with three components at 398.6 • 0.1 eV, 400.3 • 0.1 eV and 401.5 • 0.1 eV. Fitted relative intensities for several coals are listed in Table II. In previous XPS investigations of (northern hemisphere) coals, it was claimed that the N(ls) spectra were well fitted by 2 components, one at 400.3 • 0.3 eV due to pyrrolic nitrogen and one at 398.7 • 0.3 eV due to pyridinic nitrogen. Burchill and Welch 6 specifically noted that no components corresponding to primary aromatic amines (2399.7 eV) or quaternary nitrogen (2401.4 eV) were required for an acceptable fit. On th~ other hand, in a similar study of coal-derived liquids, 7 a small, high binding energy component from an un-

PYROLYTIC RELEASE OF C O A L NITROGEN

1261

TABLE I Chemical Analyses of coals (wt % daf basis) (a) low-rank coals Property

Coal

Volatile matter Fixed carbon Carbon Hydrogen Nitrogen Sulphur Oxygen (diff)

Yallourn

Millmerran

53.3 46.7 67.4 4.6 0.6 0.3 27.1

50.7 49.3 77.0 6.2 1.2 0.7 14.9

(b) bituminous coals Coal Property Volatile matter Fixed carbon Carbon Hydrogen Nitrogen Sulphur Oxygen (diff)

Liddell

Bayswater

Blair Athol

Peak Downs

Norwich Park

42.8 57.2 80.4 5.7 2.0 0.6 11.3

39.0 61.0 82.1 5.3 1.8 1.3 9.5

33.4 66.6 79.8 4.6 2.0 0.3 13.3

25.1 74.9 86.1 5.0 2.0 0.7 6.2

18.8 81.2 89.4 4.6 2.0 0.8 3.2

identified species was evident in some cases. More recently, Patience et al, 2~ found that a component near 402.3 eV (of intensity between 10 and 25%), tentatively assigned to quaternary nitrogen, and a peak near 399.4 eV assigned to amino nitrogen were required in addition to the peaks at 398.7 eV and 400.3 eV in order to fit N(ls) spectra from surface sediments. For the coals studied in the present work, a component at a binding energy near 401.5 eV and of intensity less than 20% of the total N(ls) intensity was usually required in addition to the pyrrolic and pyridinic nitrogen peaks to obtain an acceptable fit. There are several possible reasons for a component near 401.5 eV being required to obtain a satisfactory fit to the N(ls) spectrum from coal. One possibility is that such a high binding energy component arises from an excited final state satellite. However, if these satellites were the origin of the component near 401.5 eV, then its intensity would be expected to vary with the intensity of the pyridinic nitrogen peak near 398.6 eV. No such correlation was discerned. There are cogent arguments for also dismissing electron energy loss processes, or physically adsorbed nitrogen as possibilities. It is unlikely that the component is due to residual swarf from abrasion or shards from fracture

in poor electrical contact with, and hence more positively charged than, the bulk of the specimen, as in many cases the intensity of this component was greatest for the wet polished surface. In fact the latter observation suggested that the origin of the peak at 401.5 eV was oxidation of the pyridinic nitrogen to the N-oxide. The N-oxide of pyridine itself is known to be easily formed with oxidants such as hydrogen peroxide, and is quite stable. Oxidation of the pyridinic nitrogen in coal may be possible under some seam conditions or post-extraction conditions, especially aqueous environments. In particular, preparation of a sample surface by conventional wet polishing has been observed to produce an appreciable increase in the intensity of the peak near 401.5 eV and a corresponding decrease in the intensity of the pyridinic nitrogen peak at 398.6 eV (see Table II). The C(ls) spectrum from the wet polished surface was broadened on the high binding energy side due to contributions from surface carbon-oxygen species, whereas the C(ls) spectrum from an abraded or fracture surface indicated considerably less intensity from carbon-oxygen species. Thus the postulated oxidation of nitrogen at the surface is supported by the presence of oxidised carbon. The question remains as to whether all of the intensity near 401.5

1262

COAL COMBUSTION N(ls)

9 meu

9 - v

(b)

. .,~.,.~,- ~ I 403

I 402

_..y--..~ I 401

I 400

~ I 399

.~:,,~.I 398

I 397

trum was the same within experimental error before and after air exposure of the coal. Moreover, the N(is) spectra from fracture surfaces of Several coals were found to be essentially the same as the corresponding spectra from surfaces carefully abraded dry under argon. Therefore, it can be concluded that the oxidised nitrogen observed at surfaces prepared under dry, inert conditions was present in the bulk of the coal lumps. Such oxidation may have occurred post-extraction, nevertheless it would be relevant to any industrial combustion situation. Even if the intensity of the quaternary nitrogen peak is included with that of the component near 398.6 eV to provide an estimate of the total content of nitrogen in a 6-membered ring, the spectra indicate that the pyrrolic nitrogen content exceeds the pyridinic for each of the coals studied. This finding is in accordance with the conclusions from the previous studies of other coals, Burchill and Welch 6 found that for a series of UK bituminous coals, not only did the pyrrolic form predominate throughout, but also the proportion of pyridinic nitrogen increased with coal rank. We also find that the proportion of pyridinic nitrogen increases with the carbon contents of the coals, mainly at the expense of quaternary nitrogen, although as discussed above, this may be as a consequence of oxidation after extraction.

B I N D I N G ENERGY ( e V )

FIG. 1. N(ls) photoelectron spectrum from an abraded surface of (a) Norwich Park and (b) Bayswater coal fitted with 3 Gaussian-shaped peaks of unconstrained position, width and intensity. eV is an artefact arising from surface oxidation of nitrogen during sample preparation. In order to determine how readily the nitrogen at the surface of a particular coal became oxidised under ambient conditions, freshly exposed surfaces of Bayswater coal were examined spectroscopically then exposed to air for 24 h before re-examination. The N(ls) spec-

Pyrolytic Release of Nitrogen in Tars: Yields of tar and nitrogen-containing species were determined as a function of pyrolysis temperature in the range 500-1100~ C for the Yallourn brown coal, Millmerran subbituminous and Bayswater and Blair Athol bituminous coals. Figure 2 presents yields of tar and HCN for Bayswater coal. Similar results were obtained for the other coals. Previous work 17 has demonstrated that the source of CO and hydrocarbon gases observed at high temperatures during coal pyrolysis is secondary gas phase decomposition of the tars, since maximum yields of tars precede any significant yields of CO and C1-C3 hy-

TABLE II Relative intensities (%) of components fitted to N(ls) spectra from some Australian coals. Surfaces prepared by fracture or dry abrasion unless otherwise indicated. 401.5 +-- 0.1 eV (quarternary)

400.3 --- 0.1 eV (pyrrolie)

398.6 -4- 0.1 eV (pyridinic)

Blair Athol Bayswater Liddell Peak Downs Norwich Park

16 17 17 9 9

59 54 54 56 51

25 29 29 35 40

Bayswater (wet abraded)

29

56

15

Coal

1263

PYROLYTIC RELEASE OF COAL NITROGEN 30

,

[

~

30

25

~

I

600

800

"0

r

o o

20

9

20

,,~ 15

_=

g

g

10

lO

N

0

5 0 400

o 600

800

1000

1200

0

T e m p e r a t u r e (~ 40

FIG. 2. Yields of tar (0), HCN (V); and NH3 (V) from the pyrolysis of Bayswater coal as a function of pyrolysis temperature. droearbons. Figure 2 demonstrates that this is also the case for HCN and NH3. Thus a major source of HCN and NH3 is secondary, probably gas-phase, decomposition of nitrogen-containing aromatic species in the tars. Under combustion conditions, at higher temperatures, liberation of the nitrogen which remains in the char under the conditions of the present experiment, may also lead to NO~ formation. The processes by which nitrogen is released from the tars can be further investigated by a detailed analysis of the nitrogen-containing species in the volatiles. These analyses were achieved by high resolution GC with a nitrogen-specific detector. At low temperatures (--<600~ C, i.e., before significant tar cracking had occurred), the chromatograms obtained by this technique showed that the nitrogencontaining components of the tars were complex, high molecular weight, highly polar compounds. Extractions of the pyrolysis products formed at 600~ sequentially with diehloromethane, and then with methanol, a more polar solvent, showed that the methanol fraction was significantly enriched in nitrogen compared to both the original coal, and the diehloromethane fraction. This was true for coals ranging in rank from brown to bituminous, and confirms the highly polar nature of the structures in which nitrogen is initially released from the coals. At higher temperatures (>600 ~ C) it was possible to identify the following major components: pyridine and alkyl substituted pyridines, pyrrole, quinolines and substituted quinolines, indoles, acridines, earbazoles and cyano aromatics. Representative chromatograms from 3 of these coals have been published previously, is The compounds quinoline, indole and 1-cyanonaphthalene may be taken

"O

2O

10

0 1000

Temperature (~ FIG. 3. Yields of 2 ring nitrogen-containing aromatic compounds: quinoline ((3); indole (0); 1-cyanonaphthalene (V); as a function of pyrolyis temperature. (a) Blair Athol coal; (b) Bayswater coal. as representative of pyridine-type, pyrrole-type and eyano-type structures in the volatiles. In Fig. 3 yields of these species as a fimetion of pyrolysis temperature are presented for the 2 bituminous coals. Similar results for the brown coal have been previously reported, as For all three species yields increase with temperature in the range 600-800~ C; thus the initial decomposition reactions of the tars release simpler structures, possibly as a result of molecular weight reduction, or a loss of hydroxyl functionality. 17 At temperatures greater than 800~ differences in the behaviour of these species are apparent. Indole is the least stable, and its yield decreases rapidly at temperatures greater than 900~ C; quinoline is more stable, but does begin to decompose at temperatures above 900~ C; 1-eyanonaphthalene is the most stable of these species, and had not begun to decompose at 1000~ C. Cyano type

1264

COAL COMBUSTION

functional groups were not detected by XPS; thus the presence of these species in the tars must be a result of ring opening reactions or of reactions of liberated CN radicals with the aromatic products of coal decomposition. The thermal stability of these species is in accord with that determined previously in model compound studies, tl-14 The results confirm that compounds containing nitrogen in pyridinic type structures are more stable than those of the pyrrole type at temperatures above 800 ~ C. Thus a coal with a higher proportion of pyridine type structures should release nitrogen as HCN at higher temperatures. Clearly further work on a wider range of coals is necessary to establish the importance of nitrogen functionality for the kinetics of nitrogen liberation from coals.

20

15

10

0 0

v

0 Z v

-o

20

.0_

Formation of HCN and NH3: The major gas phase nitrogen-containing products of pyrolysis were HCN and NH3. More complex organic nitriles, such as cyanoacetylene, HCCCN, observed in shock tube pyrolysis studies of pyridine13 and pyrrolela comprise, at most, 2% of the HCN yield and are not considered further here. It was not possible to accurately measure yields of N~ in the fluidized bed reactor, as air leaks into the gas sampling system could not adequately be eliminated. Yields of HCN and NH3 for all four coals responded in a similar way to increases in temperature; Fig. 4 gives results for Yallourn and Blair Athol coals; these may be compared to those for Bayswater coal presented in Fig. 2. Results for the subbituminous coal were similar. HCN and NH3 were not produced in significant quantities at temperatures below 700~ C. Yields of both HCN and NH3 increased with temperature to 800~ C, but at higher temperatures yields of NH3 decreased, while those of HCN continued to increase. The proportion of coal nitrogen released as NH3 was similar for all coals (~5-7% N, daf basis), including the brown coal. Yields of HCN from the 4 coals at 1000~ C represent 15--20% of the coal nitrogen. For all four coals the dependence of yields on pyrolysis temperature is similar and is shown in Fig, 5 any differences are within the experimental uncertainties. These results are consistent with the functional group analyses, determined by XPS, of the nitrogen content of the 2 bituminous coals (Table II), which showed similar compositions. However the observed HCN yields present a problem for the proposition that secondary cracking of the tars is a major pathway to release of coal nitrogen as HCN. Elemental analyses showed that the ratio of the nitrogen content of the tars produced at 600~ to that of the coals was similar for all the coals; however, maximum tar yields, observed at a pyrolysis

>.

15 10

o 400

- ~~vj_.,.,/'~ 600

i 800

E 1000

1200

Temperature (~ FIG. 4. Yields of HCN and NHa as a function of pyrolysis temperature (a) Yallourn coal; (b) Blair Athol coal.

temperature of 600 ~ C, were 18.0. 31.3, 25.3 and 12.0% w/w (daf basis) for the Yallourn, MiUmerran, Bayswater and Blair Athol coals respectively. Thus the proportion of coal nitrogen released as HCN was invariant in spite of the fact that the tar yields varied by more than a factor of 2. In the case of Blair Athol coal (12.0% tar yield) the proportion of nitrogen released as tar at 600~ C (8.0%) is significantly less than that observed as HCN and NH3 at higher temperatures, whereas for Millmerran coal (31.3% tar yield) the corresponding proportion (21.5% of coal nitrogen as tar at 600~ C) is close to the observed yields of HCN and NH3. This implies that, for some coals, conversion of nitrogen to HCN and NH3 occurs from groups not released as tar at 600~ C, and, in addition, that there is a similar probability of conversion of coal nitrogen from these coals to HCN and NH3 irrespective of whether it occurs in groups volatile enough to be released as tar. Hence the tar yield is not a reliable guide to the conversion of coal nitrogen to NOx precursors.

PYROLYTIC RELEASE OF COAL NITROGEN 25

'~

Conclusions

/

20

"o r

S is Z

0 400

600

800

1000

1265

1200

Temperature (~ Fro. 5. Yields of HCN as a function of pyrolysis temperature (0) Yallouru; (~7) Millmerran; (V) Bayswater, ([]) Blair Athol.

The origin of NH3 during the rapid pyrolysis of coals is difficult to explain. Model compound studies H-14 of the pyrolysis of heterocyclic aromatics show HCN as the major product with no NH3 production reported. It has often been claimed that brown coals and lignites are more likely to contain a higher proportion of coal nitrogen in amine groups than are higher rank coals. These groups would be expected to decompose at low temperatures to produce NH3. However, the yields of NH3 and the effects of temperature on the yields of NH3 from Yallourn coal (Fig. 4) are not noticeably different from those of the higher rank coals, and do not provide evidence for increased NH3 formation from amine group decomposition in brown coals. The XPS analyses do not provide evidence for amine groups in the higher rank coals. Decomposition of clay minerals has been suggested as a possible source of NH3 from organic source rocks. However the brown coal studied here had a very low clay content, and this mechanism for NH3 formation does not seem to be important for these coals. Up to 800~ C yields of HCN and NH3 show a similar response to increasing temperature, suggesting that a common source may be responsible for their formation. Experiments in which NH3 was fed to the reactor with C1-C3 hydrocarbon gases showed that the high temperature decrease in NH3 yields were not due to reactions of the NH3 with these species, or with the reactor surfaces. Further work is required to determine the mechanism of the NH3 yield decrease.

Analyses of the nitrogen groups in a range of Australian bituminous coals by XPS has shown that the nitrogen is present in three functional forms: pyridinic, pyrrolic and quaternary. The pyrrolic form predominates and accounts for 50-60% of the coal nitrogen. The proportion of pyridinic groups increases and that of quaternary groups decreases with increasing carbon contents of the coals, although this could be due to post extraction oxidation. Pyrolysis experiments on a group of coals ranging in rank from brown to bituminous, and in coal nitrogen content from 0.6-2.0% N enabled the release of nitrogen from the coal to be investigated. Formation of the gas phase NO~ precursors, HCN and NH3, followed the production of tar, and cracking reactions of the tars are a probable source of HCN and NH3. The major nitrogen-containing components of the tars have been identified, and their thermal stability (Pyrrolic type < pyridinic type < cyanoaromatic type) is consistent with model compound studies. However, the results clearly show that release of nitrogen from structures which are not volatilised as tar also occurs. Acknowledgements

This project is supported by the Energy Research and Development Corporation (Project No. 1457). We thank Dr K. D. Bartle for supplying a preprint of reference 20, and K. Riley and W. C. Godbeer for the coal analyses.

REFERENCES 1. CnEN, S. L., HEAP, M. P., PERSHING, D. W. AND MARTIN, G. B.: Nineteenth Symposium (International) on Combustion, p. 1271, The Combustion Institute, 1982. 2. ATTAR, A. AND HENDRICKSON, G. G.: Coal Structure (R. A. Meyers, Ed.), p. 155, 1982. 3. JONES, R. B., McCouRr, C. B. AND SWIFT, P.: Proe. Int. Conf. on Coal Science, p. 657, Gliickauf, 1981. 4. PERRY, D. L. AND GRINT, A.: Fuel 62, 1024 (1983), 5. CLARK, O. Z. AND WILSON, R.: Fuel 62, 1034

(1983). 6. BURCHILL, P. AND WELCH, L. S.: Fuel 68, 100

(1989). 7. WALLACE,S., BARTLE, K. D. AND PERRY, D. L.: Fuel 68, 1450 (1989). 8. BLAre, D. W., WENDT, J. O. L. AND BARTOK, W.: Sixteenth Symposium (International) on Combustion, p. 475, The Combustion Institute, 1977.

1266

COAL COMBUSTION

9. SOLOMON, P. R. AND COLKET, M. B.: Fuel 57, 749 (1978). 10. FREIHAUT, J. D., PROSCIA, W. M. AND SEERY, D. J. : 1987 Symposium on Stationary Combustion Nitrogen Oxide Control, Volume 2, p. 3 6 1, EPRI CS-5361, 1987. 11. AXWORTHY,A. E.: AIChE Symposium Series 71, 43 (1975). 12. BRUINSMA,O. S. L., GEERTSMA, R. S., BANK, P. AND MOULIJN, J. A.: Fuel 67, 334 (1988). 13. MACKIE, J. C., COLKET, M. D. AND NELSON, P. F.: J. Phys. Chem. 94, 4099 (1990). 14. MACKIE, J. C., COLKET, M. D., NELSON, P. F. AND ESLER, M.: Int. J. Chem. Kinetics 23, 733 (1991).

15. EDWARDS, J. H., SMITH, I. W. AND TYLER, R. J.: Fuel 59, 681 (1980). 16. TYLER, R. J.: Fuel 58, 680 (1979). 17. NELSON, P. F., SMITH, I. W., TYLER, R. J. AND MACKIE, J. C.: Energy Fuels 2, 391 (1988). 18. NELSON, P. F., KELLY, M. D. AND WORNAT, M. J.: Fuel 70, 403 (1991). 19. FREmAUT, J. D., PROSCIA, W., KNICHT, B., VRANOS, A., HOLLICK, H. AND WICKS, K.: United Technologies Research Center Report no. 957553-F, 1989. 20. PATIENCE, R. L., BAXBY, M., BARTLE, K. D., PERRY, D. L., REES, A. G. W. AND ROWLAND, S. J.: Org. Geochem., in press.

COMMENTS Prof. lost Wendt, University of Arizona, USA. You may wish to point out that under combustion conditions, pyrolysis occurs under oxidative conditions, or at least in the presence of some oxygen, while your experiments were completed in the total absence of oxygen. This may explain the apparent discrepancy between your data, and those from many studies in which coal was pyrolysed in the presence of oxygen, and which show that as coal rank decreased the NH3 yield (in the bulk gas phase) increased. Do you plan to extend your results to oxidative pyrolysis conditions? Author's Reply. The gaseous environment in which the initial pyrolytic decomposition of coal occurs is the subject of conjecture. It obviously will be very different from the inert gas which we have used in the present experiments; however, it is also unlikely that it will approximate to that of the overall combustion stoichiometry. It is probable that the nature of the pyrolysis/combustion gas will influence the relative amounts of gas-phase nitrogen species; indeed we already have evidence that pyrolysis in 10% Ha in N2 can result in substantial increases in the NH3 yields from these coals. ~ Hence, we agree that pyrolysis gas composition will be an important variable in future studies. Differences between the prolytic release of nitorgen from Australian coals and from coals from other parts of the world also need to be investigated.

REFERENCE 1. Nelson, P. F. AND Kelly, M. D.: unpublished results.

Prof. Dr. Klaus R. G. Hein, University of Technology, Delft, The Netherlands. Based on my own pyrolysis investigations using different methods I wonder if you compared your results from a small fluidised bed with other pyrolysis apparatus. We found with brown coal considerable differences between methods, in particular, HCN vs. NH3-yields, which are important to explain NO and N20-preeursers.

Author's Reply. W e have not performed experiments in other pyrolysis apparatus, but we think it likely that differences in reactor type and configuration could result in changes in the nitrogen distributions. One of the conclusions drawn from the present results is that tar cracking (or secondary reaction) is one mechanism for the release of nitrogen as HCN. Reactor type or configuration will influence the extent of secondary cracking and hence the formation of IICN. Interactions between the other components of the gas and the nitrogen-containing species are also likely to affect the relative amounts of HCN and NH3. Conditions of the present experiments are such that the coal pyrolysis products are very dilute in the inert gas; this may also influence the HCN and NH3 concentrations.

B. Leckner, Chalmers University of Technology, Sweden. It is stated that this paper provides in formation for the estimation of NO~ emissions from pulverized coal combustion. The paper can, however, be even more important for the prediction of NzO emission from fluidized bed combustors, where HCN is known to give NzO whereas the formation of N~O from NH3 is small. Also, the formation of NzO is highly d e p e n d e n t on the rank of the fuel

PYROLYTIC RELEASE OF COAL NITROGEN (or on some property of the fuel connected to the rank). More knowledge is needed of the primary formation of HCN and NH3 linked to the fuel structure to explain the features of N20 formation.

Author's Reply. It is true that, because of the pyrolysis temperatures studied, this work has relevance to NzO formation from the fluidized bed combustion (FBC) of coals. However, some differences should be emphasized. In this study, 75-106 ~m coal particles were injected into a fluidized bed of inert material (zircon sand). These conditions were chosen to give high heating rates (~>104 K s-~), approaching those found in pulverized coal combustors. Conditions of coal particle size and heating rates are significantly different in FBC. Nevertheless, the study provides initial results on the relationship between fuel structure and ttCN and NH~ formation.

Thomas B. Brill, University of Delaware, USA. The formation of a larger amount of HCN relative

1267

to NH3 as the temperature of decomposition increases is interesting. The same result is obtained upon rapid thermolysis of polymers containing only one source of nitrogen, such as a methyl azide side groups (paper 220 of this symposium). Do you think that your findings can be explained by the preference to retain C - - N bonds at higher temperatures oa the basis of entropy? This explanation was put forth recently (Applied Spectroscopy, 46, 900 (1992)).

Author's Reply It is interesting to hear of a single source of nitrogen giving rise to both HCN and NH3, and these species showing similar responses to increased pyrolysis temperatures. The thermodynamic argument you propose seems feasible. However it is possible that interconversion of HCN and NH3 resulting from interactions with the hydrocarbon products of pyrolysis also plays a role. In this context it is interesting to note that there is a commercial process (Shawinigan) for the conversion of NH3 to HCN using propane (or butane) in a fluidized bed of coke, albeit at higher temperatures (>1200 ~ C) than those used in the present experimeats.