Journal of Analytical and Applied Pyrolysis, 16 (1989) 117-126 Elsevier Science Publishers B.V., Amsterdam - Printed in The Netherlands INFLUENCE OF METAL IONS ON VOLATILE OF PYROLYSIS OF WOOD WEI-PING PAN * and GEOFFREY N. RICHARDS 117 PRODUCTS * Wood Chemistry Laboratory University of Montana, Missoula, MT 59812 (U.S.A.) (Received October 2&h, 1988; accepted February 7th, 1989) ABSTRACT Volatile products from the pyrolysis of wood have been determined by gas-phase Fourier transform infrared spectrometry, coupled to a thermal balance. Progressive yields of water, carbon dioxide, carbon monoxide, acetic and formic acids and methanol have been determined isothermally at 523 K and at 373-823 K. The influence of the removal of inorganic salts and of ion-exchanged cations on these yields has been studied, as also has the effect of adding back potassium ions and calcium ions by ion exchange. Potassium, but not calcium, acts as a catalyst in pyrolytic reactions resulting in formation of carbon dioxide and carbon monoxide (especially from polysaccharides), acetic acid (from hemicelluloses), formic acid (from polysaccharides) and methanol (from lignin). Metal ions; pyrolysis; wood. INTRODUCTION Recently [l] we described an investigation of the primary chemical events in the pyrolysis of wood. The approach was to determine by Fourier transform infrared spectrometry (FTIR) the first volatile products from cottonwood held at 523 K in nitrogen in a thermal balance. These results were related to the progressive changes in the chemical content of the wood to conclude, inter alia, that uranic acid components and lignin begin to degrade first, followed by acetyl and xylose groupings in the hemicelluloses. The cellulose was relatively stable. It was already known that exchangeable metal ions in wood have a major influence on pyrolytic behavior [2] and on char formation. We have now extended this study to determine the influence * Present address: KY 42101, U.S.A. 01652370/89/$03.50 Department of Chemistry, 0 1989 Elsevier Western Kentucky Science Publishers B.V. University, Bowling Green, 118 of specific individual ion exchange bound metal ions on the of formation of the major evolved gases from the first stages of wood. The experimental approach has been to relate weight loss in the wood to the evolved gases as determined by FTIR. nature and rate in the pyrolysis the progressive simultaneously EXPERIMENTAL The thermal and FTIR analytical equipment was exactly as described earlier [l]. All pyrolyses were carried out in flowing nitrogen at 30 ml/mm. The wood samples were prepared as described earlier [l]. For acid washing, the wood (20 g) was immersed in 0.05 M hydrochloric acid, degassed in vacuum and eluted in a column overnight with more acid (2 1) before washing with water (2 1) in the same column and then air-drying in a dust-free environment. All water was deionized to a resistivity greater than 18 MQ cm-‘. To prepare wood samples containing specific metal ions the acid-washed wood was treated exactly as above with a solution of the appropriate metal acetate (0.05 M) replacing the acid. All samples were analyzed by inductively coupled argon plasma spectrometry (ICAPS) and values for the dominant metal ion contents are shown in Table 1. RESULTS AND DISCUSSION Table 1 shows that the indigenous metals ions are almost completely removed from wood by washing with 0.05 M hydrochloric acid at room temperature. The subsequent ion-exchange treatments with calcium and potassium acetate solutions then yield wood samples in which calcium and potassium, respectively, are the only significant metal ions, the metal ions being bound to uranic acid units in the hemicellulose [3]. The effects of the metal ion exchange procedures on isothermal weight loss of wood in nitrogen at 523 K are shown in Fig. 1. The concurrent analyses of volatile products by FTIR are shown in Figs. 2-6. The effects of TABLE 1 Metal ion contents Original wood Acid-washed Calcium-exchanged Potassium-exchanged (ppm, dry weight basis) Ca Mg P K Na Zn 1500 12 1399 12 200 1 4 2 140 8 11 8 1000 5 45 2122 30 7 43 0 9 0 11 0 119 105 I-Y and j ~~~~, 0 30 60 90 120 150 TIME 180 210 240 270 (min) Fig. 1. Weight loss of ion-exchanged woods at 523 K (303-523 K at 10 K/min). CT 60 - 0 5 50- % 8 40- P C E 2 30- 5 lo- 20- 0:. I 20 . I 30 . I 40 I * 50 ’ 60 70 Fig. 2. Formation of water from ion-exchanged + = potassium-exchanged; o = calcium-exchanged; 0: 20 . I 30 . I . 40 , 50 . I woods at 523 K. 0 = Original A = acid-washed. wood; . 60 rme(M) Fig. 3. Formation of carbon dioxide from ion-exchanged 2. woods at 523 K. Symbols as in Fig. cations on the behavior of the wood at 523 K may be summarized as follows: 1. The removal of all metal ions (acid-washed wood) resulted in a reduced rate of weight loss and reduced yields of all volatile products compared with the original wood. This effect was particularly dramatic with regard to carbon dioxide and formic acid yields in nitrogen, which were reduced to 120 c1 Q 6- 5 G E 6- 4- 2- = 0 20 30 40 50 60 70 Tlme(min) Fig. 4. Formation of acetic acid from ion-exchanged woods at 523 K. Symbols as in Fig. 2. woods at 523 K. Symbols as in Fig. 2. 5- 4- 3- 2i.p 1 20 30 50 40 60 7 Iime(min) Fig. 5. Formation 20 30 Fig. 6. Formation of formic acid from ion-exchanged 40 of methanol 50 60 70 from ion-exchanged woods at 523 K. Symbols as in Fig. 2. one-third at peak value. In comparison, the effects on yields of other volatiles were much smaller. It is therefore concluded that, rather surprisingly, the pyrolytic decarboxylation of uranic acid salts to produce carbon dioxide is facilitated compared with the free acid form. It is not possible at 121 this stage however, to speculate on the effects influencing the formic acid yields. 2. The behavior of the calcium-exchanged wood was in general very similar to the acid-washed wood. The rates of weight loss at 523 K were almost identical (Fig. 1) and the profiles of yields of individual volatiles versus time were very similar for these two wood samples. 3. The profiles of water yields versus time were remarkably insensitive to cations (Fig. 2) at 523 K. Since the acid-washed and calcium-exchanged samples lose weight more slowly, however, this means that in these woods other degradation mechanisms are inhibited compared with dehydration (i.e. char-forming) reactions. Accordingly, it has previously been observed [2] that potassium-exchanged wood gives a much higher char yield than calcium-exchanged wood. 4. In general, the potassium-exchanged wood behaved similarly to the original wood and rates of weight loss were almost identical. The thermal behavior of the wood therefore may be dominated by the indigenous potassium ions in the wood. Several types of degradation occur even more rapidly in the potassium-exchanged wood in approximate proportion to the potassium present, viz. 1000 ppm in the original wood and 1990 ppm in the potassium-exchanged wood. Thus, the formation of carbon dioxide is increased by 50% at peak and methanol by 70% at peak compared with original wood. This is evidence of the specific catalysis by potassium of pyrolysis of lignin to produce methanol and of pyrolytic decarboxylation of uronates to produce carbon dioxide. 5. The effects of cations observed in this study are essentially the same as those reported previously [2] for pyrolysis of similar samples based on differential thermogravimetric curves, but the addition of FTIR analysis of the product gases provides additional information which may subsequently lead to better understanding of the mechanisms responsible for catalysis of pyrolysis reactions. The study was next extended to progressive heating of the same wood samples in nitrogen at 3 K/mm from 373 to 823 K. The results are shown in Figs. 7-13. Fig. 7 shows the volatile products determined during the pyrolysis. From the earlier isothermal results [l] at 523 K, we assume that the first products (e.g. up to 623 K) are derived mainly from initial and partial decomposition of lignin and hemicelluloses while the later products are derived mainly from cellulose and the more resistant portions of lignin and hemicelluloses. Thus, in Fig. 7, acetic acid from the hemicelluloses reaches a peak at about 570 K and carbon dioxide (at least partly from decarboxylation of uranic acids) shows a marked inflection at a similar temperature, as do water and formic acid, which are both believed to arise predominantly from hemicellulose at Methanol (predominantly from lignin) shows three lower temperatures. peaks at about 510, 570 and 630 K which appear to represent three distinct 122 60 60 40 30 20 IO 3 573 673 773 Temperature(K) Fig. 7. Volatile products from wood heated in nitrogen C = acetic acid; D = CO; E = formic acid; F = methanol. 0 I 500 600 700 at 3 K/min. A = H,O; B = CO,; I 800 Fig. 8. Formation of water from ion-exchanged woods heated in nitrogen from 373 K at 3 = Original . . . . . = calciumwood; = potassium-exchanged, K/ min. _._._ exchanged; - - - = acid-washed. Ol 500 I 600 700 Fig. 9. Formation of carbon dioxide from ion-exchanged K at 3 K/min. Symbols as in Fig. 8. woods heated in nitrogen from 373 123 -I 500 550 600 650 700 750 600 Temperature(K) Fig. 10. Formation 373 K at 3 K/min. of carbon monoxide from ion-exchanged Symbols as in Fig. 8. 0 ! 500 woods heated in nitrogen from I 700 600 Temperature(K) Fig. 11. Formation of acetic acid from ion-exchanged K/min. Symbols as in Fig. 8. 500 600 700 woods heated in nitrogen from 373 at 3 800 Temperature(K) Fig. 12. Formation of formic acid from ion-exchanged at 3 K/min. Symbols as in Fig. 8. woods heated events in the pyrolysis of lignin. Carbon monoxide, only trace amounts on pyrolysis at 523 K, shows a 550 K and a major peak at 620 K and hence may and/or hemicelluloses as well as from cellulose. The in nitrogen from 373 K which was produced in significant inflection at be derived from lignin major volatile product 124 0 I 500 600 700 800 Temperature(K) Fig. 13. Formation of methanol from ion-exchanged 3 K/min. Symbols as in Fig. 8. woods heated in nitrogen from 373 K at of the pyrolysis detected in these experiments, peaking at about 630 K, is water and this must be produced from both hemicelluloses and cellulose by elimination reactions, e.g. from glycol groups, which yield unsaturated systems and ultimately chars. Formic (but probably not acetic) acid is evidently also a significant product of pyrolysis of cellulose. The mechanism of formic acid production is not known. There is an analogy in the formation of high yields of the same acid in anaerobic alkaline degradation of cellulose, also by unknown mechanisms [4]. The influences of cations on yields of carbon dioxide, carbon monoxide (formed above 523 K), methanol and formic acid during pyrolysis at temperatures up to 700 K are shown in Figs. 8-13. The yields of water (Fig. 8) and acetic acid (Fig. 11) were relatively insensitive to cation variation except in the zone of hemicellulose pyrolysis, where potassium is evidently a more effective catalyst than calcium. Fig. 9 shows that the potassium ions also favor increased formation of carbon dioxide and lower the temperature of peak production compared with the acid-washed or calcium forms. The total yield of this catalyzed formation of carbon dioxide is much greater than could be explained by decarboxylation of uranic acids alone. The mechanism of formation of this carbon dioxide is not known. Since the bulk of the carbon dioxide is formed above 570 K it must be derived at least partly from cellulose. Carbon monoxide was not formed in significant quantities at 523 K, but was produced in similar molar amount to carbon dioxide from original and from potassium-exchanged wood at higher temperatures, peaking at about 620 K (Fig. 10). It seems probable that this product also is produced largely from cellulose. The yield of CO from acid-washed and from calciumexchanged wood was only about one-tenth of the other wood samples. We consider it unlikely therefore that the CO is formed by secondary gasification reaction of CO, with char, because we have previously shown [5] that calcium is a particularly effective catalyst of this gasification reaction. 125 Furthermore, the gasification reaction is extremely slow at the temperature of maximum CO formation. We conclude therefore that the CO is a primary pyrolysis product derived from wood by unknown mechanism(s) catalyzed by potassium. Formic acid (Fig. 12) evidently forms at higher temperatures by potassium-catalyzed pyrolysis reactions from cellulose. The same acid is a major product of alkaline degradation of cellulose in absence of air [4] and it is probable that the greater effectiveness of potassium compared with calcium in catalyzing formation of formic acid (and perhaps methanol) is associated with the greater basicity of the former. The formation of methanol (Fig. 13) shows a distinct second peak at about 630 K for original and for potassium-exchanged wood. Presumably the evolution of methanol above 590 K is catalyzed by potassium and derived by pyrolysis reactions from the more resistant lignin which survives pyrolysis at lower temperatures. Varhegyi and coworkers [6,7] have recently studied the effect on pyrolysis of physical sorption of several inorganic salts on cellulose and on sugar cane bagasse. In general, they used salts which would be stable at the pyrolysis temperature, whereas in the present work the ion exchanged cations (like the ‘native’ cations in wood) are converted into oxides, carbonates or free metals during pyrolysis [8]. Their determination of volatile products was by connection of a mass spectrometer to the thermal balance, so that it was not possible to relate actual yields of any volatile product to weight loss of the solid, but some relative influences of the salts can be related to the present work. Thus it was clear that pyrolysis of cellulose to produce water, CO and CO, is accelerated by addition of sodium chloride (but not by three other salts) [6]. However the opposite effect appears to operate with bagasse [7]. It is difficult to reach conclusions on formation of other primary pyrolysis products in this work because generally there are several types of products which fragment to give the ions detected. These results may be compared with the present study in which the presence of potassium has little influence on extent or rate of water formation from wood, but greatly increases the extent and rate of formation of both CO and CO,. This present study and many earlier ones, show that traces of inorganic salts in lignocellulosics can produce dramatic effects on pyrolysis mechanisms. In the ultimate case, we have shown that 0.01% of sodium chloride added to pure cellulose can almost halve the pyrolytic yield of levoglucosan [8]. This represents one formula unit of salt per 3640 anhydroglucose unit (AGUs), or about one per 700 AGUs if we consider only the amorphous regions of the cellulose where pyrolysis commences. It seems most unlikely that this situation could be achieved by swelling of the solid matrix by the salt as has been suggested [7]. However, we remain unable at present to explain such effects beyond the vague suggestion that they may operate via changes in relative rates of competing pyrolysis mechanisms which are influenced by conductivity or dielectric constant in the solid. It is tempting 126 to couple such a hypothesis with the guess that both free-radical and heterolytic mechanisms may be operative in the pyrolysis, but despite all of the studies on this subject, we remain in virtual ignorance of the chemical mechanisms of pyrolysis of cellulose and of lignocellulosics. ACKNOWLEDGEMENTS The authors are grateful to Bill DeGroot for frequent helpful discussions and experimental assistance. The work was funded by a grant from the Center for Fire Research, National Bureau of Standards. REFERENCES 1 W.F. DeGroot, W.-P. Pan, M.D. Rahman and G.N. Richards, J. Anal. Appl. Pyrolysis, 13 (1988) 221-231. 2 W.F. DeGroot and F. Shafizadeh, J. Anal. Appl. Pyrolysis, 6 (1984) 217-232 and references therein. 3 W.F. DeGroot, Carbohydr. Res., 142 (1985) 172-178. 4 W.M. Corbett and G.N. Richards, Svensk Papperstidning, 60 (1957) 791-794. 5 W.F. DeGroot and G.N. Richards, Fuel, 67 (1988) 352-360. 6 G. Varhegyi, M.J. Antal, T. Szekely, F. Till and E. Jakab, J. Energy Fuels, 2 (1988) 267-272. 7 G. Varhegyi, M.J. Antal, T. Szekely, F. Till, E. Jakab and P. Szabo, J. Energy Fuels, 2 (1988) 273-277. 8 W.F. DeGroot and G.N. Richards, Fuel, 67 (1988) 345-351. 9 M.G. Essig, G.N. Richards and E.M. Schenck, Appl. Polym. Symp., in press.