Methoxyphenols from burning of Scandinavian forest plant materials

Article abstract of DOI:10.1016/S0045-6535(99)00427-0

Semivolatile compounds in smoke from gram-scale incomplete burning of plant materials were assessed by gas chromatography and mass spectrometry. Gas syringe sampling was shown to be adequate by comparison with adsorbent sampling. Methoxyphenols as well as 1,6-anhydroglucose were released in amounts as large as 10 mg kg^-1 of dry biomass at 90% combustion efficiency. Wood, twigs, bark and needles from the conifers Norway spruce and Scots pine emitted 12 reported 2-methoxyphenols in similar proportions. Grass, heather and birchwood released the same 2-methoxyphenols but also the corresponding 2,6-dimethoxyphenols which are characteristic of angiosperms. The methoxyphenols are formed from lignin and differ in structure by the group in para position relative to the phenolic OH group. Prominent phenols were those with trans-l-propenyl and ethenyl groups in that position. Vanillin, 4- hydroxy-3-methoxybenzaldehyde, was a prominent carbonyl compound from the conifer materials. (C) 2000 Elsevier Science Ltd.

Full text of DOI:10.1016/S0045-6535(99)00427-0

Chemosphere 41 (2000) 735±741
Methoxyphenols from burning of Scandinavian forest plant
materials
*

Jennica Kjallstrand, Olle Ramnas, Goran Petersson


Department of Chemical Environmental Science, Chalmers University of Technology, 412 96 Goteborg, Sweden
Received 19 July 1999; accepted 4 September 1999
Abstract
Semivolatile compounds in smoke from gram-scale incomplete burning of plant materials were assessed by gas
chromatography and mass spectrometry. Gas syringe sampling was shown to be adequate by comparison with ad-
1
sorbent sampling. Methoxyphenols as well as 1,6-anhydroglucose were released in amounts as large as 10 mg kg of
dry biomass at 90% combustion eciency. Wood, twigs, bark and needles from the conifers Norway spruce and Scots
pine emitted 12 reported 2-methoxyphenols in similar proportions. Grass, heather and birchwood released the same 2-
methoxyphenols but also the corresponding 2,6-dimethoxyphenols which are characteristic of angiosperms. The
methoxyphenols are formed from lignin and di€er in structure by the group in para position relative to the phenolic OH
group. Prominent phenols were those with trans-1-propenyl and ethenyl groups in that position. Vanillin, 4-hydroxy-3-
methoxybenzaldehyde, was a prominent carbonyl compound from the conifer materials. Ó 2000 Elsevier Science Ltd.
All rights reserved.
Keywords: Phenols; Wood; Lignin; Combustion; Fire; Smoke
1. Introduction
Gellerstedt, 1991). At ambient temperatures, the meth-
oxyphenols tend to condense, and they are present in
extracts of smoke aerosols together with non-volatile
compounds (Simoneit et al., 1993). Small amounts of the
most volatile methoxyphenols remain gaseous in smoke
from di€erent forest biomass materials (McKenzie et al.,
1995).
Assessment of the chemical components in smoke
from biomass burning is important for both accidental
and planned ®res and with respect to both environ-
mental impact and health hazards.
The lignin of vascular plants decomposes to fuel-
speci®c methoxyphenols from biomass burning. Quali-
tative gas chromatographic and mass spectrometric data
for 36 methoxyphenols from wood burning were re-
The purpose of this study was to determine the
amount and proportions of methoxyphenols released
from natural forest materials. Di€erent biomass mate-
rials, characteristic for Scandinavian boreal forests, were
compared. Fire behaviour in these forests has been de-

cently presented (Kjallstrand et al., 1998), with refer-
ences to earlier qualitative studies. The semivolatile
monocyclic methoxyphenols are emitted as gases, and
quantitative proportions have been reported from con-
trolled analytical pyrolysis of wood and pulp (Kleen and

scribed in some detail (Schimmel and Granstrom, 1997).
2. Experimental
* Corresponding author. Tel.: +46-31-7722998; fax: +46-31-
7722999.
E-mail address: goranp@kmv.chalmers.se (G. Petersson).
The plant materials were collected in interior south
Sweden. The following data apply to biomass materials
and experimental methods.
0045-6535/00/$ - see front matter Ó 2000 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 4 5 - 6 5 3 5 ( 9 9 ) 0 0 4 2 7 - 0

736
Plant materials
J. Kjallstrand et al. / Chemosphere 41 (2000) 735±741
Softwood (Picea)
Twigs (Picea)
Softwood (Pinus)
Bark (Pinus)
Needles (Pinus)
Fern
Pieces of fresh stem sapwood, cut from Norway spruce (Picea abies)
Needle-free thin twigs with bark, taken three months after felling
Pieces of fresh stem sapwood, cut from Scots pine (Pinus sylvestris)
Thin stem and litter sheets from Scots pine
Season's upper litter needles from a 20±30 yr old planted pine forest
Stem and leaf pieces of dead winter-standing Pteridium aquilinum
Pieces of dead straws of the common forest grass Deschampsia ¯exuosa
Twigs of Calluna vulgaris, with manually defoliated living parts
Pieces of fresh stem sapwood, cut from white birch
Grass
Heather
Hardwood
Moss and lichen
The moss Pleurosium schreberi and the lichen Cladonia rangiferina were collected from
pine forests
Fuels and burning
Fuel preparation
Fuel drying
Wood cut into thin chips and match-size pieces, other fuels broken to pieces
Openly indoors for several days, bark and heather also for 1 h at 100°C
In the range 90±95%, determined by heating at 105°C to constant weight
In the range 1 g (fern, grass) to 10 g (softwood, hardwood)
Fuel dry weight
Amount of fuel
Combustion pot
Covering pot
Ceramic, 1.5 l, with the pile of plant material on the bottom
Clay, 2 l, with bottom hole, placed upside-down for ®re choking and sampling
Flaming burning of half of the material, followed by choking and sampling
Burning sequence
Determination of CO and CO2
Smoke sampling
Gas chromatograph
Separation of CO
Separation of CO2
Detector
30±40 ml, gas-tight syringe (50 ml, SGE)
Perkin Elmer 3920 B, with gas sampling valves, 190 ll (CO) and 290 ll (CO2)
Molecular sieve 13 X, 1:7 m  3:2 mm o.d. column, 80°C, isothermal
HayeSep Q (60±80 mesh) porous polymer, 1:9 m  3:2 mm o.d. column, 80°C
1
Thermal conductivity, 200°C, 150 mA, carrier gas helium, 25 ml min
Calibrations
By injecting pure CO and CO2 from gas tubes
Determination of methoxyphenols
Smoke sampling
Gas chromatograph
Carrier gas
400±500 ll, gas-tight syringe (500 ll, SGE), pre-¯ushed with smoke 3±5 times
Varian 3800 (MS-linked) with splitless injection (220°C)
Helium, 1.1 ml min
1
Column
30 m  0:25 mm i.d. FSOT (fused silica open tubular)
Stationary phase
Oven temperature
Mass spectrometer
Mass spectra
Cyanopropylphenylsilicone, Rtx 1701, 0.25 lm stationary phase layer
1
Increased 5°C min from 50 to 200°C and then kept isothermal at 200°C
Varian Saturn 2000, ion trap (150°C), EI (70 eV)
Mass range m/z 35±400, 1 spectrum s recorded and data-stored
External calibration using guaiacol and vanillin
1
Quantitative data
Adsorbent sampling
Cartridge
Adsorbent
Injector glass liner with a narrow restriction at the lower end
Tenax TA, 60±80 mesh, 0.1 g
Samples
Withdrawn using a 10 ml syringe connected with a te¯on tube
1±2 ml for pot burning and 5±10 ml for open burning
Needle-free thin spruce twigs with bark (Picea abies)
On plate in fume-cupboard, ¯aming and glowing combustion
Thermal, in the injector, temperature raised from 100°C to 220°C (60°C min
Sampling volumes
Biomass fuel
Open burning
Desorption
1
)
The methoxyphenols were identi®ed using recently
published chromatographic and mass spectrometric data
A few minor methoxyphenols overlapped with other
compounds in the chromatograms. Selected ion moni-
toring for speci®c prominent ions was then used to de-
termine peak areas. The mass spectra of the
methoxyphenols are all characterised by an abundant
molecular ion and a few prominent ions from speci®c

(Kjallstrand et al., 1998). Retention times are given in
Table 1 for the speci®c polar gas chromatographic col-
umn used in this work. Automatic integration of peak
areas, manually adjusted when required due to incorrect
baseline drawing, was used for quantitative assessments.

fragmentations (Kjallstrand et al., 1998). The di€erence

J. Kjallstrand et al. / Chemosphere 41 (2000) 735±741
737

738
J. Kjallstrand et al. / Chemosphere 41 (2000) 735±741
in total ion response between the assessed methoxy-
phenols should be less than 20%, because they all have
similar proportions between C, H and O (Fitch and
Sauter, 1983). The response was therefore set equal for
all compounds to facilitate comparisons. The response
of vanillin was determined using a reference solution in
acetonitrile. The magnitude of the response was con-
®rmed for guaiacol using gas injection. A reference
amount of guaiacol was then evaporated in a glass
container (50 l) from which the gas sample was taken.
Carbon monoxide was not resolved from methane on
the molecular sieve column used. Methane was option-
ally assessed speci®cally by changing from thermal
conductivity to ¯ame ionisation detection. The propor-
tion of methane was shown to be less than 10% for a
wood smoke sample, as expected from other biomass
studies (McKenzie et al., 1995). No corrections for
methane were therefore made for any of the carbon
monoxide determinations.
Fig 1. Chromatograms from GC±MS analyses of 4-hydroxy-3-
methoxyphenyl (guaiacyl) compounds in smoke from pinewood
burning (total ion current and ion currents corresponding to the
molecular ions m/z 150 + 164 and 152 + 166 + 180).
which should re¯ect burning in forests better than pyr-
olysis and uniform burning conditions.
3. Results and discussion
Guaiacol and its methyl and ethyl analogues (GuH,
GuCH3, GuCH2CH3) were signi®cant products from all
nine materials with GuCH3 as the most prominent one.
The corresponding Sy compounds were even more
prominent relative to other Sy compounds.
The data given in Table 1 summarise the results of
the study and are arranged to permit comparisons be-
tween di€erent smoke components as well as di€erent
biomass materials.
The methoxyphenols with alkenyl side chains ac-
count for as much as 40±50% of total methoxyphenols
for all materials. The species with trans-1-propenyl side
chains tended to be predominant, but the ethenyl phe-
nols were also very prominent. The well-known eugenol,
GuCH2CH¸CH2, was formed in smaller amounts. In
Fig. 1, the four alkenyl phenols from sapwood of Scots
pine are recorded by molecular ion monitoring below
the total ion chromatogram.
3.1. Methoxyphenol proportions
The conifer samples and the fern sample gave rise to
2-methoxyphenols only, whereas the angiosperm sam-
ples of grass, heather and hardwood gave rise to both 2-
methoxyphenols and 2,6-dimethoxyphenols. The pro-
portion of 2,6-dimethoxyphenols was very high (80%)
for birchwood and quite low (30%) for the grass.
The formation and total proportion of 2,6-dimeth-
oxyphenols re¯ect the long known characteristic lignin
structures of angiosperm plants (Sarkanen and Hergert,
1971). The lignins of conifers almost lack 2,6-dimeth-
oxysubstituted units.
The widely used food ¯avour additive vanillin, Gu-
CHO, constitutes 10±20% of the recorded 2-methoxy-
phenols and was the major carbonyl-containing phenol.
Vanillin is probably partly formed by oxidation of un-
saturated side chains, and its percent proportion tended
to increase with increased air supply. Other prominent 2-
methoxyphenols with carbonyl groups were GuCH2-
CHO, GuCOCH3 and GuCH2COCH3. In Fig. 1, the
presence and separation of these phenols in smoke from
pine are illustrated by molecular ion monitoring. The
prominent trans-coniferaldehyde, GuCH¸CHCHO
(E), was the latest eluting 2-methoxyphenol. The Sy
compounds with low volatility tended to be recorded in
lower proportions than the corresponding Gu com-
pounds.
The observed absence from Pteridium of 2,6-di-
methoxyphenols emphasises the fundamental evolu-
tionary di€erence between ferns and angiosperms.
Lignin is characteristic of vascular plants, and no
methoxyphenols were obtained from burning of two
common Scandinavian moss and lichen species.
Percentage proportions of total methoxyphenols are
given in Table 1 for 12 speci®c 2-methoxyphenols and
the corresponding twelve 2,6-dimethoxyphenols. As ex-
plained in Table 1, these are referred to as guaiacyl (Gu)
and syringyl (Sy) compounds with varying side chains. It
is emphasised that the materials burn in di€erent ways
with respect to parameters such as temperature, ¯aming
and glowing, which in¯uence the proportions of phe-
nols. Each material was allowed to burn uncontrolled
3.2. Sampling methods and combustion conditions
The recorded methoxyphenols di€er considerably
with respect to polarity and reactivity of the side chain.

J. Kjallstrand et al. / Chemosphere 41 (2000) 735±741
739
Table 2
Composition (%) of methoxyphenols from spruce twigsa
Choked pot burning
Glowing pot burning
Open burning
Cartridge
Syringe
Cartridge
Syringe
Cartridge
GuH
GuCH3
5
6
9
11
4
5
6
10
9
7
10
3
GuCH2CH3
GuCH¸CH2
GuCH2CH¸CH2
GuCH¸CHCH3ꢀZ†
GuCH¸CHCH3ꢀE†
GuCHO
3
2
3
11
4
13
3
11
2
9
12
2
3
3
2
2
2
2
17
11
8
15
11
6
12
13
5
10
13
4
12
13
5
GuCOCH3
GuCH2COCH3
GuCH2CH2CH2OH
GuCH¸CHCH2OHꢀE†
GuCH¸CHCHO ꢀE†
Other methoxyphenolsb
8
6
7
5
6
9
8
12
6
11
5
9
5
5
4
7
4
13
4
12
4
9
3
3
6
a Comparison of di€ering burning conditions and sampling methods.
b Mainly GuCH2CH2CH3, GuCH2CHO, GuCOCH2CH3, GuCH2CH2CHO and GuCH¸CHCH2OH ꢀZ†.
To con®rm the results, an optional method for sampling
was developed. Samples were then taken on Tenax ad-
sorption cartridges and the methoxyphenols were anal-
ysed after thermal desorption. Comparative results are
given in Table 2 for the two very di€erent methods. The
results demonstrate similar methoxyphenol proportions
for the two methods, without signi®cant losses of the
most polar and reactive carbonyl compounds, alkenes
and alcohols. Somewhat lower proportions for syringe
sampling are indicated only for the most volatile and
chemically stable phenols, guaiacol and methylguaiacol.
The comparisons in Table 2 also demonstrate sig-
ni®cantly di€erent methoxyphenol proportions for dif-
fering burning conditions. Choked ¯aming burning
causes elevated proportions of the alkenyl phenols. For
glowing and smouldering combustion, an increased
proportion of coniferaldehyde was observed. Adsorbent
cartridges permit larger sampling volumes than gas in-
jection and can therefore be used for smoke from open
burning. The results in Table 2 for open burning com-
pare well with the results from pot burning. It is con-
cluded that the results in Table 1 approximately re¯ect
the methoxyphenol composition in smoke from open
burning.
under nitrogen of wood from Salix, birch and spruce
species (Ingemarsson et al., 1999). A pyrolysis study of
ꢀ
wheat straw lignin (Rodrõguez et al., 1997) reported low
proportions of carbonyl compounds like vanillin. The
major phenol was GuCH¸CH2, which was also the
most prominent methoxyphenol from grass in our study.
Studies of methoxyphenols in smoke from biomass
burning have been made mainly for particles and con-
densates. Smoke from residential wood stoves was re-
ported to contain guaiacol and syringol as the most
prominent methoxyphenols from softwood and hard-
wood (Hawthorne et al., 1989). Several alkenyl and
carbonyl-containing methoxyphenols were observed in
proportions similar to those given here, although the
burning conditions were very di€erent. Guaiacol and
methylguaiacol were the most prominent among few
methoxyphenols determined in condensates from burn-
ing of biomass related to ponderosa pine (McKenzie et
al., 1995). In contrast, carbonyl-containing methoxy-
phenols were observed to be the most prominent in ®ne
particles from biomass burning (Simoneit et al., 1993;
Rogge et al., 1998). For toxicity testing, semivolatile
compounds from wood burning have been collected on
Tenax cartridges with subsequent solvent extraction
(Hauk et al., 1994).
Comprehensive qualitative studies of compounds
from wood pyrolysis (Pouwels et al., 1987; Faix et al.,
1990) include the reported methoxyphenols. Our quali-
tative results compare well with those reported from
analytical pyrolysis of wood of the same spruce and
birch species (Kleen and Gellerstedt, 1991). Apparently,
oxidative pyrolysis conditions were applied in that
study. Vanillin and other carbonyl compounds were
even more prominent than in our results. In contrast,
high proportions of alkenyl phenols but no carbonyl-
containing phenols were reported from slow pyrolysis
3.3. Smoke constituents
From Table 1 it is seen that the CO2 concentration
was typically 1±2% for the nine burning experiments.
The CO concentration was typically 0.1±0.2% and the
combustion eciency consequently typically around
90%.
The concentrations of methoxyphenols and other
smoke components varied for the di€erent plant mate-

740
J. Kjallstrand et al. / Chemosphere 41 (2000) 735±741
rials, which were also burning in di€erent ways. The
highest 2-methoxyphenol concentrations were observed
from pine bark and spruce twigs. The average amount of
methoxyphenols relative to CO was as high as 10% for
the softwood samples and even higher for the birchwood
sample. These amounts of methoxyphenols relative to
CO are at least 10 times higher than the amounts re-
ported for each of the major volatile organic compounds
methane, ethane, methanol, and acetic acid from
smouldering combustion of forest biomass (McKenzie
et al., 1995). They are also 10 times higher than the re-
corded methoxyphenol levels, evidently due to a low
recovery of these semivolatile compounds.
tions of gaseous methoxyphenols in ambient air are
biased towards the most volatile species.
4. Conclusions
The results demonstrate that methoxyphenols are
formed in high concentrations and speci®c proportions
from inecient biomass burning. Syringe sampling im-
mediately followed by GC±MS analysis permits
adequate determination of the methoxyphenols in lab-
oratory experiments. Adsorption on Tenax cartridges,
followed by thermal desorption, is an optional method
suitable also for ®eld sampling.
Approximately 1 kg of CO2 is formed from natural
burning of 1 kg of forest biomass (McKenzie et al.,
1995). From the results in Table 1 it is concluded that
the amount of released methoxyphenols was of the order
The reported smoke components are indicative of
emissions from accidental and planned forest ®res and
other inecient biomass burning. A major environ-
mental concern is the formation of ozone and other
photo-oxidants from gaseous smoke constituents. The
condensation of the methoxyphenols on particles de-
creases their contribution to ground-level ozone forma-
tion. Particles covered by adsorbed methoxyphenols and
anhydrosugars should also have very di€erent health
e€ects compared with particles from trac and other
sources.
1
10 g kg of burnt biomass. Previously, the complex
mixture of volatile hydrocarbons has been studied for
biomass burning under similar conditions (Barrefors
and Petersson, 1995). The ratio of total non-methane
1
volatile hydrocarbons varied from 10
g
kg
for
1
smouldering, to 0.1 g kg for ¯aming combustion of
wood. Similarly increased emissions with decreased
combustion eciency have been reported for other
volatile compounds emitted from forest biomass burn-
ing (McKenzie et al., 1995). It is concluded that meth-
1
oxyphenol emissions markedly exceeding 10 g kg are
References
to be expected from the frequently occurring smoul-
dering burning of lignin-containing forest biomass.
The predominant primary decomposition product
from polysaccharides, 1,6-anhydroglucose, was formed
in similar amounts as the total 2-methoxyphenols. In
spite of its very polar nature with three alcoholic OH
groups, satisfactory chromatographic performance was
observed as illustrated in Fig. 1 by the total ion chro-
matogram. Methoxyphenols and 1,6-anhydroglucose
gave rise to all prominent recorded peaks from semi-
volatile compounds for most of the samples. This is
illustrated in Fig. 1 for the pinewood sample. Levog-
lucosan has been proposed as a tracer for smoke from
biomass burning (Simoneit et al., 1999).
Barrefors, G., Petersson, G., 1995. Volatile hydrocarbons from
domestic wood burning. Chemosphere 30, 1551±1556.
Faix, O., Meier, D., Fortmann, I., 1990. Thermal degradation
products of wood±Gas chromatographic separation and
mass spectrometric characterization of monomeric lignin
derived products. Holz Roh-Werksto€ 48, 281±285.
Fitch, W.L., Sauter, A.D., 1983. Calculation of relative electron
impact total ionization cross sections for organic molecules.
Anal. Chem. 55, 832±835.
Hauk, A., Sklorz, M., Bergmann, G., Hutzinger, O., 1994.
Analysis and toxicity testing of combustion gases I. A new
sampling unit for collection of combustion products.
J. Anal. Appl. Pyrolysis. 28, 1±12.
Hawthorne, S.B., Krieger, M.S., Miller, D.J., Mathiason, M.B.,
1989. Collection and quantitation of methoxylated phenol
tracers for atmospheric pollution from residential wood
stoves. Environ. Sci. Technol. 23, 470±475.
Condensation in the sampling syringe of 1,6-anhy-
droglucose and the methoxyphenols of low volatility
was minimised by ¯ushing the syringe with the hot
smoke before sampling, and by injecting the sample
within 30 s after sampling. Partial or complete losses of
compounds of low volatility are likely to be a common
source of errors in studies of methoxyphenols.
ꢁ
Ingemarsson, A., Nilsson, M., Pedersen, J.R., Olsson, J.O.,
1999. Slow pyrolysis of willow (Salix) studied with GC/MS
and GC/FTIR/FID. Chemosphere 39, 103±112.


Kjallstrand, J., Ramnas, O., Petersson, G., 1998. Gas chro-
matographic and mass spectrometric analysis of 36 lignin-
related methoxyphenols from uncontrolled combustion of
wood. J. Chromatogr. A 824, 205±210.
Many types of phenolic and other compounds of
lower volatility than the methoxyphenols are released on
burning, with immediate formation of smoke particles
(Simoneit et al., 1993). The semivolatile methoxyphenols
are released as gases, but condense on the surface of
particles as the temperature of the smoke decreases
(Hawthorne et al., 1989). This means that the propor-
Kleen, M., Gellerstedt, G., 1991. Characterization of chemical
and mechanical pulps by pyrolysis-gas chromatography/
mass spectrometry. J. Anal. Appl. Pyrolysis 19, 139±152.
McKenzie, L.M., Hao, W.M., Richards, G.N., Ward, D.E.,
1995. Measurement and modeling of air toxins from

J. Kjallstrand et al. / Chemosphere 41 (2000) 735±741
741
smoldering combustion of biomass. Environ. Sci. Technol.
29, 2047±2054.
Sarkanen, K.V., Hergert, H.L., 1971. Classi®cation and distri-
bution. In: Sarkanen, K.V., Ludwig, C.H. (Eds.), Lignins
Occurrence, Formation, Structure and Reactions. Wiley/
Interscience, New York (Chapter 3).
Pouwels, A.D., Tom, A., Eijkel, G.B., Boon, J.J., 1987.
Characterisation of beech wood and its holocellulose and
xylan fractions by pyrolysis-gas chromatography±mass
spectrometry. J. Anal. Appl. Pyrolysis 11, 417±436.

Schimmel, J., Granstrom, A., 1997. Fuel succession and ®re
behavior in the Swedish boreal forest. Can. J. For. Res. 27,
1207±1216.
ꢀ
ꢀ
ꢀ
Rodrõguez, J., Hernandez-Coronado, M.J., Hernandez, M.,
Bocchini, P., Galletti, G.C., Arias, M.E., 1997. Chemical
characterization by pyrolysis/gas chromatography/mass
spectrometry of acid-precipitable polymeric lignin (APPL)
from wheat straw transformed by selected streptomyces
strains. Anal. Chim. Acta 345, 121±129.
Simoneit, B.R.T., Rogge, W.F., Mazurek, M.A., Standley, L.J.,
Hildemann, L.M., Cass, G.R., 1993. Lignin pyrolysis
products, lignans, and resin acids as speci®c tracers of plant
classes in emissions from biomass combustion. Environ. Sci.
Technol. 27, 2533±2541.
Rogge, W.F., Hildemann, L.M., Mazurek, M.A., Cass, G.R.,
Simoneit, B.R.T., 1998. Sources of ®ne organic aerosol. 9.
Pine, oak, and synthetic log combustion in residential
®replaces. Environ. Sci. Technol. 32, 13±22.
Simoneit, B.R.T., Schauer, J.J., Nolte, C.G., Oros, D.R., Elias,
V.O., Fraser, M.P., Rogge, W.F, Cass, G.R., 1999. Levog-
lucosan, a tracer for cellulose in biomass burning and
atmospheric particles. Atmos. Environ. 33, 173±182.