Comparative study of the thermal decomposition of pure cellulose and pulp paper

Polymer Degradationand Stability 49 (1995) 275-283 ELSEVIER

0141-3910(95)00077-1

© 1995 Elsevier Science Limited Printed in Northern Ireland, All rights reserved 0141-3910/95/$09.50

Comparative study of the thermal decomposition of pure cellulose and pulp paper S. Soares,* G. Caminot & S. Levchik~t Dipartimento di Chimica lnorganica, Chimica Fisica e Chimica dei Materiali, Universitä di Torino, Via P. Giuria, 7-10125 Torino, Italy ( R e c e i v e d 22 F e b r u a r y 1995; a c c e p t e d 8 M a r c h 1995)

The thermal decomposition of pure cellulose, either powder or paper, and of pulp Kraft paper was studied by thermogravimetry, differential scanning calorimetry, thermal volatilisation analysis and characterisation of the degradation products. Depolymerisation of the pure cellulose with production of levoglucosan takes place at high heating rates. At low heating rates autocatalytic dehydration of the cellulose with char production predominates, as shown by kinetic treatment of thermogravimetric data. Both mechanisms simultaneously contribute to the thermal decomposition of the Kraft paper independently of the heating rate, with the largest production of char. Pure cellulose produces more levoglucosan than Kraft paper on thermal decomposition, which gives furanic compounds as secondary degradation products. A larger amount of volatile carbonylic compounds is formed on thermal decomposition of the Kraft paper.

1 INTRODUCTION

Two competitive pathways have been proposed in the literature 4'9"11'13"1a for the process, as shown in Scheme 1. If the dehydration of cellulose prevails, evolution of CO2, H 2 0 and CO with formation of solid char is mainly observed. If depolymerisation occurs m o r e extensively than dehydration, volatilisation of tar (high boiling products, H B P ) mostly composed of levoglucosan is observed. The H B P might further decompose, with the production of light flammable gases, if it is not promptly r e m o v e d from the heating zone. The pathway of thermal decomposition and the composition of the products are influenced by m a n y physical and chemical factors such as temperature, "~ type of atmosphere, 9 size and texture of the cellulose sample, 15 crystallinity, ~6 presence of impurities such as metals, ~7 and other parameters. 7 In the present work we have c o m p a r e d the cellulose in a p o w d e r sample with pulp paper, either from pure cellulose or the Kraft process, in o r d e r to understand the effect of sample form and impurities. These aspects are of great

Cellulose is the most extensively used natural polymer. The large n u m b e r of studies published on its thermal decomposition is explained by important practical interests regarding fire safety, ~'2 production of chemicals, 3"4 high temperature applications as insulation material in electrical transformers, 5'6 etc. Reviews have been compiled concerning, for example, the mechanism 4-78 and kinetics 9 of cellulose degradation and pyrolysis of cellulose biomass. ~° A decrease in the degree of polymerisation (DP) with negligible production of volatiles occurs at m o d e r a t e t e m p e r a t u r e s ( < 2 5 0 ° C ) , whereas extensive degradation of cellulose takes place at higher temperatures, ~l with evolution of various gases and high boiling products. ~2 * Permanent address: Departamento de Quimica Organica e Inorganica da Universidade Federal do Ceara, Campus do Pici, 60000 Fortaleza, Ceara, Brazil. t To whom all correspondence should be addressed. :~ Permanent address: Research Institute for Physical Chemical Problems, Byelorussian University, Leningradskaya, t4-220080 Minsk, Belarus. 275

276

S. Soares et al. Dehydration ] "Anhydrocellulose"

Cellulose

D

gases (CO2, CO, H20)

Char +

Cellulose D (lower DP) I

Tar + (mainly levoglucosan)

gases

Depolymerisation Scheine

relevance, for example, in the prediction of the lifetime of paper insulation in electrical transformers. 6

1

continued until the evolution of gases was complete or the rate of gas evolution stabilised. The solid residue left and the HBP that condensed on the cool finger were weighed and the amount of gases calculated by difference.

2 EXPERIMENTAL 2.3 Analysis of degradation products 2.1 Materials Solid residues, gases and HBP collected in TVA experiments were analysed by infrared (FTIR, Perkin Eimer 2000). Gases were also examined by gas chromatography-mass spectrometry ( G C MS) using a Hewlett-Packard 5890/5970 machine. The relative content of the gases was estimated on the basis of the integrated area of the corresponding gas chromatographic peaks

Pure cellulose as powder (Aldrich, ca 20/zm), or ashless Whatman filter paper no. 542 (ash yield 0.008%) and a pulp Kraft paper (Tervakoski Mills, Findland: a cellulose, 87-89%; pentosans, 8-11%; lignin, K number 25-33; metals, mainly Ca 2+ (1350 ppm), Na + (113 ppm) and K + (113 ppm); ash yield < 0.5%) were used. Crystallinity, estimated by infrared spectroscopy 18 decreases in the order powder cellulose > Whatman filter > Kraft paper.

3 RESULTS

2.2 Thermal analysis

3.1 Thermal decomposition behaviour

Thermal decomposition was carried out on 10 mg samples in an aluminium holder either under nitrogen or air flow (60 cm 3 min -]) by thermogravimetry (TG) at heating rates in the range 1.25-40°C min -1 and differential scanning calorimetry (DSC) at 10°C min -1 using a Du Pont 2100 thermal analyser provided with a T G A 2950 module. Thermal degradation under dynamic vacuum ( 1 0 - 3 - 1 0 -4 m m Hg) was carried out by thermal volatilisation analysis ( T V A ) 19 at 10°C min -]. Gases condensable at -196°C were collected in a trap at liquid nitrogen temperature. A watercooled glass finger was introduced in the TVA degradation vessel to condense and collect high boiling products. 2° To estimate the production of HBP during the course of the degradation a TVA step experiment was carried out by successively heating the sample at 10°C min -] to 250, 275, 325 and 420°C. At each temperature isothermal heating was

3.1.1 Inert atmosphere All cellulose samples under study show one main step of weight loss in nitrogen on heating in TG at 10°C min -] (Fig. 1). Apart from a small weight loss at 50-70°C as a result of moisture, the onset of weight loss of the main step of degradation lies at about 275°C for the powder cellulose (curves a and a') and Whatman paper (curves b and b') and at a lower temperature (250°C) for the Kraft paper (curves c and c'). The powder cellulose and the Whatman paper show a maximum rate of weight loss at the same temperature (331-332°C, curves a' and b'), whereas it occurs in Kraft paper about 20°C higher (350°C, curve c). The powder cellulose produces 2.7% of char at 600°C, whereas the Whatman and Kraft papers produce about 2 and 5 times more char, respectively (5.7 and 12.4%). Thus, it seems that apart from different char yields, the powder cellulose and the Whatman paper show similar thermal decomposition behaviour in thermogravimetry

Thermal decomposition of pure cellulose and pulp paper 100 80;~ 6 0 . ~ 40 i

331"(

20-

"~~

(¢)-kmfl " ...........

,h,.what,~,~;~

,

......

0~

332"¢

(a'l-~d«

(b')-whatman . (c')-kraft •

0

F

."

'

100

I

"

!

I

200

'

300

I

I

400

500

'

I

600

T e m p e r a t u r e , *C

Fig. 1. Thermogravimetry (a, b, c) and derivative (a', b', c') curves of: (a, a') powder cellulose, (b, b') Whatman paper and (c, c') Kraft paper. Heating rate, 10°C min 1; nitrogen flow, 60 cm3 min-i. under nitrogen. On the other hand, the Kraft paper begins to decompose at a lower temperature but produces more residue, possibly owing to its metal ions 2~,22 and lignin 23 content, with a shifi of the m a x i m u m rate of weight loss to a higher temperature. The ranking of the thermal stability of the cellulose samples is somewhat different at low heating rate (e.g. 1.25°C min ~, Fig. 2). The weight loss curves of the Whatman and Kraft papers are coincident in the main weight loss step (curves b and c) with rate maxima at 312 and 320°C (curves b' and c', respectively). The powder cellulose seems less stable at low heating rate than both papers (curves a and a'). These effects of heating rate are likely to be due to the occurrence of two overlapping processes in the

277

major weight loss stage, which become partially separated in Fig. 2 (shoulders on derivative curves a', b', c'). The DSC curves for all cellulose samples heated at 10°C min -1 in nitrogen show a strong e n d o t h e r m followed by an exothermic peak (Fig. 3). The e n d o t h e r m for both the powder cellulose and the W h a t m a n paper occurs with a m a x i m u m at a comparable temperature, 340-343°C, whereas the exotherm for the powder occurs at a lower t e m p e r a t u r e (371°C, curve a) than for the W h a t m a n paper (389°C, curve b). For the Kraft paper the e n d o t h e r m is observed at a higher temperature (362°C, curve c) indicating an apparently greater thermal stability, which is in agreement with thermogravimetry (Fig. 1, curve c). The exotherm occurs in the Kraft paper at a temperature (384°C) comparable to that for the W h a t m a n paper. It is generally accepted 16,22,24 that the end o t h e r m is mainly due to depolymerisation of cellulose with formation of levoglucosan and its evaporation, whereas the exotherm is due to char formation. 25 Both the powder cellulose and the W h a t m a n paper show endotherms of comparable intensity, whereas the exotherm for the W h a t m a n paper is more pronounced (curve b) than the exotherm for the powder, which is in correspondence with thermogravimery, where a slightly larger char yield is observed for the paper (Fig. 1, curve b). In comparison with the W h a t m a n paper, the Kraft paper shows a smaller e n d o t h e r m and exotherm (Fig. 3, curve c), which apparently contradicts the higher char yield of the Kraft paper. However, since the e n d o t h e r m

k•_•_der°

100 -

10.1WIg

80ùe 6 0 -

~

340"C 389"C

40 -

O

(c)-kraft ~..

(b)-whatman

(b)-whatman

tu

~--Z/2"22'-'--_

202ù.c

+

0-

(a')-powder

5

(s)-powder

?_')~~~_m_,ù . . . .

343"C 384°C

((;)-kraft

312"C,~ . ............

ij

320"C :. +ù:

(c')~raft

, .................... '

0

i

100

'

i

• '

200

•

362°C

i,. I

300

' '

I

'

400

i

500

'

i

600

T e m p e r a t u r e , *C

Fig. 2. Thermogravimetry, samples and curves as in Fig. 1. Heating rate, 1-25°C min 1; nitrogen flow, 60 cm~ min ~.

0

I

100

'

t

200

'

i

300

•

I

'

400

Temperature,

I

500

'

I

600

*C

Fig. 3. DSC curves of (a) powder cellulose, (b) Whatman paper and (c) Kraft paper. Heating rate, 10°C min-~nitrogen flow, 60 c m 3 min '.

S. Soares et al,

278

for the Kraft paper is shifted to high temperature, it is likely partially to overlap the exotherm, with partial compensation. Figure 4 shows the dependence of the apparent activation energy, calculated by the Ozawa method,26 on the degree of decomposition. Pure cellulose (powder and Whatman paper) shows different activation energy (Ea) dependencies at different heating rates. At low heating rates (1-25-5°C min -t) Ea for the powder cellulose and the Whatman paper decreases with increasing degree of decomposition (Fig. 4, curves a and b), which may indicate the occurrence of autocatalytic processes. A steep increase in Ea is observed at a high degree of decomposition (85% for the powder and 75% for the Whatman paper), where charring starts to occur. In the interval 10-75% of decomposition, the Whatman paper has a higher activation energy than the powder cellulose. A reverse situation is found at high heating rates (5-40°C min-'), with the powder cellulose showing a higher activation energy (210-220 kJ/mol) than the Whatman paper (165-180 kJ/mol). In both cases at decomposition above 70-80% the energy of activation increases steeply because of charring processes, as. for low heating rates. Two activation energies, approximately corresponding to those found here for the powder cellulose, have been reported in the literature for the thermal decomposition of pure cellulose. 14'23 The low E~ was attributed to the thermal decomposition of cellulose controlled by dehydration, whereas the high En should be responsible for depolymerisation of cellulose with production of levoglucosan (see scheme above)

Furthermore, it was shown that the pathway controlled by dehydration is autocatalytic, z7 which is in agreement with the decreasing of Ea at low heating rates in (Fig. 4). The Ea for Kraft paper (curve c) is independent of heating rate and is similar to that for depolymerisation of the Whatman paper (high heating rate, curve b'); thus, it seems that, independently of heating rate, the thermal decomposition of the Kraft paper mostly proceeds through depolymerisation. However, the increment of Ea observed at the beginning of thermal decomposition ( < 2 0 % degree of decomposition) might be due to partial contribution of low energy dehydration, which decreases as degradation progresses. Furthermore, the slight decrease of Ea at 20-70% degree of decomposition might be the manifestation of autocatalysis found for the dehydration process. 27 Thus, a mixed depolymerisation and dehydration mechanism might be considered for the Kraft paper independently of heating rate. 3.1.2 Under air

The three samples of cellulose show similar thermal stability in air on heating at 10°C min-', with a main step of weight loss at 270-340°C (Fig. 5, curves a-c) characterised by a maximum rate at 317-329°C (curves a'-c'). Thus the oxygen of air seems to accelerate the thermal degradation of the Kraft paper, whereas the effect on pure cellulose samples is somewhat reduced. In the presence of air the char yield at 350°C for powder cellulose and Whatman paper

100 -

250"

~

80"

(a~-powder high ~ t l n g

o

"~ ................... .

N60~ -

rote

40

"~~200"

{c)-k~n

I~ (c)-kraft ~" " " . ~ * (b)-whatman "- .'~. r

317"C

20 m

(a')-powder

"""""

- -. "" «..

(a)-powder

(b)«hatman ~w h¢at)ng rote - - -. _.

,'

• ( b")-_w_hat_m_a_n. . . .

,

,

100 20 Degree

40

60

of decornposRion,

aO

ù

% ~¢')-kraft

Fig. 4. Activation energy as a function of degree of decomposition for powder cellulose at (a) low heating rates 1-25-5°C min ~ and (a') high heating rates 5-40°C min -j, for Whatman paper at (b) low heating rates 1.25-5°C min-' and (b') high heating rates 5-40°C min ~ and for Kraft paper at (c) heating rates 1.25-40°C min ~.

480°C

324"C .". 403"C

100

'

i

100

•

." ! i

200

,

i

300

Temperature,

j\ '

xt/2

i

'

400

i

500

'

i

600

*C

Fig. 5. Thermogravimetry (a, b, c) and derivative curves (a', b', c') of (a, a') powder cellulose, (b, b') Whatman paper and (c, c') Kraft paper. Heating rate, 10°C min-'; air flow, 60 cm3 min-~.

Thermal decomposition

increases from 6 and 9% obtained in nitrogen (Fig. l), to 13 and 17%, respectively. Oxygen does not significantly affect the char yield of the Kraft paper (ca 20-23%). The char from pure cellulose (powder and Whatman paper) oxidises slowly to volatile products at 350-520°C (curves a and b, maximum rate 480-5OO”C), whereas the char of the Kraft paper rapidly oxidises in a narrower temperature interval of 380-420°C (curve c, maximum rate 403°C). The pattern of DSC curves obtained in air (Fig. 6) is different from that observed in nitrogen (Fig. 3). The weak endotherms observed at 322-328°C for the powder cellulose and the Whatman paper, which are attributed to depolymerisation and evaporation of degradation products as in nitrogen, are followed in air by a strong exotherm at 343-345°C (curves a and b) attributed to charring and oxidation of the products of thermal decomposition of cellulose.2” A second exotherm observed at 489-495°C is likely to be due to oxidation of the char.25 The Kraft paper does not show the endotherm that might be expected at 300-320°C possibly by the large because it is overwhelmed overlapping exotherm at 343°C (Fig. 6, curve c). The second exotherm of char oxidation (43O”C), corresponding to the sharp weight loss in thermogravimetry (Fig. 5, curve c), is very intense in the case of the Kraft paper, which might indicate self-ignition of the char. 3.1.3 Under vacuum Three steps of gas evolution are characteristic for the three samples of cellulose (Fig. 7, solid lines):

0

100

200

300

Temperature,

400

500

600

‘C

Fig. 6. DSC curves of (a) powder cellulose,

279

of pure cellulose and pulp paper

(b) Whatman paper and (c) Kraft paper. Heating rate, 10°C min-‘: air flow, 60 cm7 min-‘.

76

60

150

250

350

Temperature,

450

650

‘C

Fig. 7. TVA curves of (a) powder cellulose,

(b) Whatman paper and (c) Kraft paper. Solid lines, total gases; dashed lines, gases non-condensable at -196°C; heating rate, 10°C min-‘; dynamic vacuum, 10~3-10~4 mm Hg.

evolution of absorbed water at 50-140°C; a main step of gas evolution at 160-400°C; and a high temperature volatilisation step with constant rate at 400-600°C. Absorbed water is larger in the papers than in the powder. Apart from the first peak, powder cellulose and the Whatman paper show very similar TVA curves for total gas evolution (Fig. 7(a) and (b), solid lines). Apparently, the peak of the main step of gas evolution is wider for powder cellulose, whereas it is higher for the Whatman paper. Nevertheless, the total volume of gas evolved in this step is approximately the same for both samples. The onset of the main step of gas evolution lies at a lower temperature (160°C) for the Kraft paper (Fig. 7(c), solid line) than for the two other samples (1SO’C). The apparently much lower stability of all the samples under vacuum as compared to nitrogen might be due to more efficient removal of a light volatile product of degradation (e.g. H,O), which tends to be retained by the polar matrix at normal pressure, giving negligible weight loss. The weight loss becomes detectable when non-condensable products are evolved (e.g. CO) by thermal degradation of the dehydrated cellulose (see Fig. 1). Besides carbon monoxide, which is one of the main gaseous products of cellulose decompositioqx hydrogen, which is noncondensable at -196°C is expected to be evolved on charring and carbonisation of cellulose.28 It is likely that CO is evolved mostly in the second step of gas evolution at 200-400°C whereas dehydrogenation of the charring residue is a

280

S. Soares et al.

high-temperature process that should mostly occur above 400°C. However, the signal owing to non-condensable gases is relatively small compared to total gas evolution above 400°C showing that a more complex chemical reorganisation of the char occurs, not simply dehydrogenation. Production of non-condensable gases (powder cellulose < Whatman paper < Kraft paper) correlates with the char yield as measured in thermogravimetry (Figs 1 and 2) in agreement with literature data on C0.8~2s~29~30

Hydmy-

pmpanone

Butenal

Buts

B+Io-

“O”l2

alone

oplem camonyls

Fig. 9. Relative content of carbonylic compounds evolved under vacuum in TVA.

in gases

acetone, butenal and butanone tend to decrease from the powder cellulose to the Whatman paper and to the Kraft paper, which gives hydroxypropanone as the largest carbonyl product by far.

3.2 Products of thermal decomposition 3.2.1 Gases

Water and carbon dioxide are the main gaseous products, condensable at -196°C evolved on thermal decomposition of all cellulose samples. Furans (Fig. 8) and carbonyl derivatives (Fig. 9) are the second major gaseous degradation products, whereas alcohols, acids, and aromatic and aliphatic hydrocarbons are minor products. The powder cellulose produces a larger amount of furanic compounds than the Whatman paper and a much larger amount than the Kraft paper (Fig. 8). Furan is the most abundant of these products for powder cellulose and Whatman paper. Methylfurans and furancarboxyaldehyde are also produced in relatively large amounts by powder cellulose, whereas furan, methyl furans and furancarboxyaldehyde are produced in comparably low amounts from the Kraft paper. Acetone, hydroxypropanone, butenal, butanone and butanedione (Fig. 9) are the major carbonyl products detected among the condensable gases in TVA. The amounts of evolved

aldehyde

Fig. 8. Relative content of furanic compounds evolved under vacuum in TVA.

!.I Acetone

in gases

3.2.2 High boiling products Table 1 shows the amount of HBP produced at different temperatures throughout the main step of thermal decomposition of the celluloses. They give comparable amounts of HBP (47-59%) at the end of this step (420°C). However, the major fraction of HBP (40-47%, Table 1) is evolved from the powder cellulose and the Whatman paper at the beginning of the step ( I 275”C), whereas the Kraft paper produces HBP mostly at the end of the step (2 275°C). Infrared spectra of HBP from the powder cellulose (Fig. 10, spectrum a) and the Whatman paper (spectrum b) show an absorption pattern similar to that of levoglucosan (1,6-anhydro-g-Dglucopyranose; spectrum d), with characteristic absorptions at 3350-3270 (OH stretching), 2915-2900 (CH stretching), 1395-1405 (CH deformation), 1130-1140 (C-OC stretching), 1040-1050 (C-OH stretching) and 855-865 cm-’ (CH bending).“‘-36 On the other hand, HBP for the Kraft paper, apart from levoglucosan absorptions at 1136 and 1048 cm-‘, show strong absorptions at 3390 (OH stretching), 2929 (CH stretching), 1713 (C=O stretching) and 1627 cm-’ (C=C stretching)” These absorptions may be caused, for example, by the presence of levoglucosenone, which is reported to be one of the products of cellulose degradation,‘” or possibly by volatilisation of chain fragments consisting of partially decomposed oligosaccharides.’ An approximate estimation”’ of the levoglucosan content in the HBP, based on the intensity of the band at 1130-1140 cm-’ (C-OC, stretching) normalised to CH stretching at

Thermal decomposition

281

of pure cellulose and pulp paper

Table 1. Percent ratio of solid residues, high boiling products (HBP) and gases produced in the main step of thermal decomposition (160-420°C) of different celhdoses under vacuum

Whatman

Powder cellulose

T(“C)

250 275 325 420

Kraft paper

paper

Solid residue

HBP

Gases”

Solid residue

HBP

Gases”

Solid residue

HBP

Gases“

69 32 14 4

22 47 53 58

9 21 33 38

80 51 27 6

13 40 57 59

7 9 16 35

83 64 35 12

2 8 32 47

15 28 33 41

“By difference. 2900-2930

cm-‘, indicates -30% less levoglucosan content in HBP for the Kraft paper than for the two pure celluloses. 3.2.3 Solid residue The change of the infrared spectra of the solid residue of all celluloses in TVA experiments is similar. Therefore the infrared characterisation of the solid residues from the powder cellulose only (Fig. 11) is considered in the following discussion. The attributes of the main absorptions of the virgin powder cellulose (Fig. 11, spectrum a) are listed in Table 2. The infrared pattern of the solid residue at 250°C (spectrum b) retains the characteristic features of virgin cellulose, despite 31% weight loss (Table 1). A new band at 1730 of carbonyl cm-’ is due to the appearance functionalities. The characteristic absorptions of the glucosidic structure (1200-900 cm-‘) decrease in intensity on heating to 275°C (spectrum c) where 68% weight loss occurs (Table 1). Simultaneously, there is an increase of double bond (1637-1612 cm-‘) and carbonyl (1730-1707 cm-‘) concentra-

I

!

3500

3000

2500

2000

Wavenumber,

Fig. 10. FTIR

tion. Complete disappearance of the glucosidic structure is observed at 325°C (spectrum d) where only 14% of solid residue remains from the powder cellulose (Table 1). Furthermore, the characteristic infrared fingerprint of aliphatic structure” (3000-2800 and 1440 cm-‘) appears on heating to 325°C. It is likely that at this temperature a crosslinked unsaturated aliphaticcarbonylic structure is formed, which might be the precursor of the char. The first step of charring of this precursor structure is observed at 325-42O”C, since a decrease of carbonyl functionalities (1707-1701 cm-‘) and the appearance of CH stretching of aromatics (3040 cm-‘) and of the characteristic triplet of CH wagging of aromatic structures (900-700 cmp’)37 occur (spectrum e). On further heating to 550°C (spectrum f) the intensity of aromatic absorptions (3041 and 900-700 cm-‘) increases. Furthermore, a new band at 1517 cm-‘,

1500

1000

500

cm-’

spectra of high boiling products from (a) powder cellulose, (b) Whatman paper and (c) Kraft paper: (d) spectrum of levoglucosan. Pellets in KBr.

3000

2500

2000

Wavenumber,

1500 -1

1000

500

cm

Fig. 11. FTIR of (a) the initial powder cellulose and of solid

residues collected under vacuum at: (b) 250°C (c) 275°C (d) 325°C (e) 420°C and (f) 550°C. Pellets in Kbr.

282

S. Snares et

Table 2. Assignments

of the main absorptions der cellulose

of the pow-

Wavenumber (cm-‘)

Assignment

References

3347 2901 1637 1428 1377 1320 1167 1130-1000 894

Y (OH) hydrogen bonded v (CH) 6 (OH) absorbed water S (CH,) scissors 6 (CH) bending S (COH) bending v Glucopyranose ring 6 (OH) S (CH) wag

30,33-36 30,33,35,36 31,33 33,35,36 30,35 31,35 33,36 33-36 7,33,36

owing to aromatic semicircle stretching,“’ appears at this temperature. In contrast, the intensity of aliphatic absorptions (3000-2800 cm-‘) decreases, which is in accordance with carbonisation and formation of a thermostable char at this temperature.

4 DISCUSSION It is likely that the general mechanism of thermal decomposition is similar for the different celluloses under study, since the changes in the structures of their solid residues is similar, as shown by infrared (Fig. 11). However, important details of the process depend upon the type of cellulose or paper sample. The high boiling products produced from the pure celluloses consist mainly of levoglucosan (Fig. 10, spectra a and b), whereas chain fragments with structures similar to that of the solid residue at 325420°C (Fig. 11, spectra d and e) contribute, together with levoglucosan, to the HBP of the Kraft paper (Fig. 10, spectrum c). The pure celluloses produce much more volatile furanic compounds than the Kraft paper with the (Fig. S), which is in agreement suggestion in the literature”.3x that furans mostly originate from thermal decomposition of levoglucosan. The abnormally high yield of hydroxypropanone from the Kraft paper (Fig. 9) is likely to be due to a contribution from pentosans and lignin.“’ The kinetic approach shows that the thermal decomposition of pure celluloses is controlled by two different processes, depending on the heating rate used in non-isothermal experiments (Fig. 4). At low heating rates, decomposition of cellulose is likely to occur in the less ordered amorphous

al.

regions with char formation.24 At higher heating rates the depolymerisation of cellulose to yield levoglucosan becomes the major decomposition pathway and should start in crystalline regions through an unzipping mechanism.‘6,24 Since the activation energy of depolymerisation is higher than that of thermal decomposition of ‘anhydrocellulose”6 the depolymerisation should contribute mostly to degradation at the higher temperatures, whereas dehydration followed by chain scissions should contribute mainly at low temperatures. It is likely that the dehydration pathway mostly contributes to the charring of cellulose (see Scheme 1); however, the char might also be obtained throughout depolymerisation as a result of secondary reactions, if escape of levoglucosan is restricted. The char yield from celluloses (Figs 1, 2 and 5) depends on many factors: the packing density of the original material, purity of the cellulose, heating rate, the presence of oxygen in the atmosphere, etc. The difference in char production from the powder cellulose and from the Whatman paper might be explained by retention of levoglucosan in the heating zone, which induces its secondary degradation. Furthermore, the Whatman paper has a lower degree of crystallinity than the powder cellulose and a larger fraction is expected to degrade through the dehydration pathway. Thermal decomposition occurs in the lower temperature range if lower heating rates are applied. Thus, the possibility exists of an increasing occurrence of dehydration reactions, which also provokes an increase of charring (Figs 1 and 2). Oxygen shows a similar effect by decreasing the temperature of decomposition (Figs 1, 3, 5 and 6), thus playing a catalytic role that was compared in the literature4” to that of prehydrolysis. On the other hand, oxygen accelerates free radical processes in the decomposition of cellulose,2y which can also contribute to char production. The lower degree of crystallinity of the Kraft paper, which decreases the levoglucosan yield, together with the effect of impurities, pentosans, lignin and metals, can explain the relatively large char production. For example, it was speculated4’ that alkaline cations might retard the unzipping reaction of depolymerisation and therefore favour the dehydration pathway. It was also suggested7 that inorganic salts can catalyse the dehydration by scission of both exocyclic and

Thermal decomposition

of pure cellulose and pulp paper

en&cyclic C-O bonds. In our experiments we can not exclude any of these possibilities. Furthermore, lignin produces more char on thermal decomposition than cellulose,2 although the threshold of weight loss is observed at lower temperatures. Such behaviour can also explain the onset of weight loss of the Kraft paper at a lower temperature than for pure cellulose. Conflicting data are reported in the literature about the contribution of char in the thermal degradation2,2” of pentosans. ACKNOWLEDGEMENTS

The help of Prof. L. Costa in the organization of experiments and of Dr J. Scheirs in discussing the results is gratefully acknowledged. S. Soares is grateful to the Fundaqao Coordenaqao de Aperfeiqoamento de Pessoal de Nivel Superior (CAPES) for a fellowship. REFERENCES 1. Faroq, A. A., Price, D., Milness, G. L & Horrocks, A. R., Polym. Deg. Stab., 33 (1991) 155. 2. Hirata, T., Kawamoto. S. & Nishimoto, T., Fire Mater., 15 (1991) 27. 3. Richards, G. N., J. Anal. A&. Pyrol., 10 (1987) 251. 4. Radlein, D., Piskorz, J. & Scott, D. S., J. Anal. Appl. Pyrol., 19 (1991) 41. 5. Myers, S. D., Kelly, J. J. & Parish, R. H., A Guide to Transformer Maintenance. Division of S. D. Myers Transformer Maintenance Institute, 1981. 6. Emsley, A. M., Polym. Deg. Stab., 44 (1994) 343. 7. Shafizadeh, F., Adv. Carbohydr. Chem., 23 (1968) 419. 8. Shafizadeh, F., Appl. Polym. Symp., 28 (1975) 153. 9. Antal, M.J., Friedman, H. L., Jr & Rogers, F. E., Comb. Sci. Technol., 21 (1980) 141.

10. Shafizadeh, F., J. Anal. Appl. Pyrol., 3 (1982) 283. 11. Arseneau, D. F., Can. J. Chem., 49 (1971) 632. 12. Pouwels, A. D., Eijkel, G. B. & Boon, J. J., J. Anal. Appl. Pyrol., 9 (1986) 121.

13. Kishore, K. & Mohandas, K. Fire Mater., 6(2) (1982) 54. 14. Bradbury, A. G. W., Sakai, Y. & Shafizadeh, F., J. Appl. Polym. Sci., 23 (1979) 3271. 15. Helmy, S. A., Polym. Deg. Stab., 33 (1991) 155.

283

16. Bash, A. & Lewin, M., J. Polym. Sci., Polym. Chem. Ed., 11 (1973) 3071. 17. Pan, W. P. & Richards, G. N., J. Anal. Appl. Pyrol., 16 (1989) 117. 18. Bash, A., Wasserman, T. & Lewin. M., J. Polym. Sci., Polym. Chem. Ed., 12 (1974) 1143. I. C., in Developments in Polymer Degradation, Vol. 1, ed. N. Grassie. Applied Science

19. McNeil],

Publishers, London, 1977, p. 43. 20. McNeill, I. C. & Rincon, A., Polym. Deg. Stab., 24 (1989) 59.

21. Richards, G. N. & Hshien, F. Y., in Fire and Polymer. Hazard Identification and Prevention. ACS Symp. Ser. Vol. 425, ed. G. L. Nelson. AC& Washington, DC,

1990, p. 361. 22. Tang, W. K. & Eickner, H. W., Forest service research paper. FPL 82, January 1968. 23. Shafizadeh, F., Pure Appl. Chem., 55 (1983) 705. 24. Halpern, Y. & Patai, S., Isr. J. Chern., 7 (1969) 673. 25. Jain, R. K., Lal, K. & Bhatnagar, H. L., J. Appl. Polym. Sci., 30 (1985) 897.

26. Ozawa, T., Bull. Chem. Sot. Jpn, 38 (1965) 1881. 27. Roberts, A. F., Combust. Flame, 14 (1970) 261. 28. Fitzer, E., Mueller, K. & Schaefer, W., in Chemistry and Physics of Carbon. A Series of Advances. Vol. 7, ed. P. L. Walker, Jr. Marcel Dekker, NY, 1971, p. 237. 29. Shafizadeh, F. & Bradbury, A. G. W., J. Appl. Polym. Sci., 23 (1979) 1431.

30. Jain, R. K., Lal, K. & Bhatnagar, H. L., Makromol. Chem., 183 (1982) 3003. 31. Lin-Vien, D., Colthup, N. B., Fateley, W. G. & Grasselli, J. G., The Handbook of Infrared and Raman Characteristic Frequencies of Organic Molecules. Academic Press, Boston, 1991. 32. Bash, A. & Lewin, M., J. Fire Flam., 4 (1973) 92. 33. Higgins, H. G., J. Polym. Sci., 28 (1958) 645. 34. Sekiguchi, Y., Frye, J. S. & Shafizadeh, F., J. Appl. Polym. Sci., 28 (1983) 3513.

35. Zawadski, J., In Chemistry and Physics of Carbon. A Series of Advances, Vol. 21, ed. P. A. Thower. Marcel Dekker, NY, 1990, p. 147. 36. Gilliland, B. F. & Smith, B. F., J. Appl. Polym. Sci., 16 (1972) 1801. 37. Morterra, C. & Low, M. J. D., Carbon, 21 (1983) 283. 38. Shafizadeh, F. & Lai, Y. Z., J. Org. Chem., 37 (1972) 278.

39. Piskorz, J., Radlein,

D. & Scott, I>. S., J. Anal. Appl.

Pyrol., 9 (1986) 121.

40. Bash, A. & Lewin, M., J. Polym. Sci., Polym. Chem. Ed., 11 (1973) 3095. 41. Piskorz, J., Radlein, D. S. A., Scott, D. S. & Czernik, S., J. Anal. Appl. Pyrol., 16 (1989) 127.