1026
Energy & Fuels 1997, 11, 1026-1032
Release of Chlorine from Biomass at Pyrolysis and
Gasification Conditions1
E. Björkman* and B. Strömberg
TPS Termiska Processer AB, Studsvik, S-611 82 Nyköping, Sweden
Received February 24, 1997X
The working hypothesis for the study was that the main part of the chlorine in biomass is in
an inorganic form and therefore should not vaporize appreciably below the melting point of the
corresponding salt (around 700 °C) because the vapor pressure over solid salt is negligible. In
the study, biomass fuels (sugarcane trash, switch grass, lucerne, straw rape) were subjected to
pyrolysis in a flow of nitrogen, and the weight of the residue and its chlorine content were
measured and compared to the original fuel. Contrary to the hypothesis, the results showed
that during pyrolysis of biomass 20-50% of the total chlorine evaporated already at 400 °C,
although the majority of the chlorine was water soluble (in grass 93%) and therefore most probably
ionic species. At 900 °C, 30-60% of the chlorine was still left in the char. At 200 °C less than
10% of chlorine had evaporated from the fuel, indicating that the chlorine is not associated with
water. Another result was that there was no significant difference in the chlorine release between
biomass and synthetic waste, i.e., a mixture of organic and inorganic chlorides. These results
are contradictory with the starting hypothesis and can therefore have new implications for the
use of these fuels in combustion and gasification processes.
Introduction
Emission of hydrogen chloride is the third most
important contribution to the global acidification from
human activities. The two first are SO2 and NOx. The
HCl is a local pollutant, contrary to the other two, since
it is easily dissolved in rain droplets and, therefore,
usually falls down near the emission source. The
hydrogen chloride emission from combustion and gasification processes has been calculated to 3.5 Mt/year.
The major part of the estimated global contribution of
HCl to the atmosphere is evaporation from seas. Even
with a redeposition of 90% HCl to the seas, the estimated emission will reach approximately 120 Mt/year.
Another HCl source in the atmosphere is methyl
chloride (CH3Cl), which reacts with the OH radical and
forms HCl. The main emission sources for CH3Cl are
microorganisms, marine plants on the sea, and the
burning of vegetation on land. The total amount of HCl
produced from CH3Cl in the atmosphere has been
estimated to 4 Mt/year.2
Chlorine, together with alkali metals, has been shown
to play an important role in deposit formation during
biomass combustion. The chlorine facilitates the mobility of many inorganic compounds, especially potassium,
and can stabilize high-temperature gas phase alkalicontaining species.3
X Abstract published in Advance ACS Abstracts, September 1, 1997.
(1) This work was in part presented at the Conference on Development in Thermochemical Biomass Conversion held in Banff, 1996. Due
to misunderstandings, the article was not printed in the proceedings,
and the text presented here is based both on that contribution and on
additional results.
(2) Sloss, L. L. IEA Coal Res. London 1992, IEACR/45.
(3) Baxter, L. L.; Miles, T. R.; Miles Jr., T. R.; Jenkins, B. M.; Milne,
T. A.; Dayton, D.; Bryers, R. W.; Oden, L. L. Developments in
Thermochemical Biomass Conversion; Blackie Academic and Professional: London, 1997; pp 1424-1444.
S0887-0624(97)00031-5 CCC: $14.00
The emitted chloro compounds from combustion processes are formed from chlorine in the fuel, and the
amount is dependent on the concentration and the use
of secondary gas treatment. Both the amount and the
origin of the chlorine vary between different types of
fuels. In coal, the concentration varies normally between 50 and 2000 mg/kg, and the origin is mainly
groundwater which has been incorporated into the coal
after its formation. A smaller amount is from prehistoric vegetation. In the coal, the main part of the
chlorine can be found as chloride ions in the moisture
within the pores of coal particles, which upon drying
forms NaCl. Other forms of chlorine in coal are inorganic and organic chlorides.2 The concentration of
chlorine in peat is 200-500 mg/kg2 mainly from marine
sources, minerals in the soil, and depositions from
human activities. In biomass the chlorine content can
vary between <100 and 7000 mg/kg,4 and the amount
is dependent on closeness of the sea, fertilizers and
leaching of the soil by the rain.
The majority of the emitted chlorine from a combustion process of coal will leave as HCl in the gas phase.
Some of it condenses on small particles of the fly ashes.
During the combustion of coal in a fluidized bed, at a
temperature of 830 °C, the evaporation of chlorine
started immediately as the particle entered the furnace.
There was no evaporation of NaCl; most of the sodium
could be found in the ashes but not associated with
chlorine in NaCl.2 Furthermore, work on combustion
of waste has shown that HCl together with different
carbon sources are the precursors for dioxin formation5
and that the formation in the postcombustion zone is a
significant contributor.6-8
(4) Gärdenäs S. Vattenfall FUD-Rep. 1991, No SV-UB--91/40, in
Swedish.
(5) Eklund, G.; Pedersen, J.; Strömberg, B. Nature 1986, 320, 155156.
© 1997 American Chemical Society
Release of Chlorine from Biomass
Energy & Fuels, Vol. 11, No. 5, 1997 1027
Table 1. Composition (wt %) on Dry Basis of the Fuels
Used in the Study
fuel
C
H
N
O
Cl
ash
sugarcane trash
switch grass
switch grass leached
lucerne
straw (rape)
synth. waste
44.6
43.8
44.3
44.3
41.1
39.3
6.2
5.8
5.7
5.8
5.8
6.4
0.8
1.2
0.9
2.9
1.3
1.4
43.5
40.4
43.6
38.4
47.2
39.3
0.44
0.79
0.06
0.29
0.18
0.69
5.0
8.8
5.5
8.6
4.6
14.4
During pyrolysis of coal most of the chlorine leaves
at relatively low temperatures. A British coal emitted
71% of the total amount of chlorine at 258 °C during
21-23 h.9 Another coal, this one from Illinois, emitted
more than 90% of its chlorine content between 300 and
600 °C. The same coal had similar profiles for the
release of ammonia and chlorine, which was interpreted
as the chlorine was associated with the nitrogen in the
coal.10 Still another study with coal from Illinois showed
that if NaCl solution had been added to the coal, an
additional peak for the release of chlorine could be
detected. This extra peak appeared at higher temperatures (starting at 770 °C) compared to the original peak
at 300-550 °C. The results showed that, according to
the authors, only a small part of the chlorine in the
original coal was present as NaCl.
The distribution of chlorine between gas, liquid, and
solid phase is dependent on the pressure and the
temperature. Thermodynamic calculation11 of pressurized gasification (10 bar) of peat shows that at 700 °C
most of chlorine was in the liquid phase, but at 800 °C
the gas phase was dominating. Furthermore, the
composition of the gas phase is also dependent of
temperature: the dominating chloro-containing compound below 600 °C was HCl(g) but above 800 °C KCl(g) predominates, followed by NaCl(g).11
Major chlorine-containing compounds released during
gasification of biomass (switch grass) at 800 °C are HCl
and KCl, which are formed both in the devolatilization
and in the char combustion phase.12 The release of
chlorine during pyrolysis of biomass has not, to our
knowledge, been reported in the literature.
Experimental Section
The ultimate analyses of fuels, including the original
chlorine content, that were used in the study are given in Table
1. The table needs some explanation. The sugarcane trash
used consisted mainly of the green leaves and the top of
sugarcane plant (Saccharum officinarum), as the stem itself
is removed for processing in the sugar mill. The switch grass
(Phaláris arundinácea) used was summer harvested and had
a higher chlorine content than the ones that are harvested in
the spring.4 Two other biomass fuels were used in the study,
lucerne (Medicago sativa) and straw of rape (Brassica napus).
The synthetic waste was produced with the aim to have a
homogenous waste fuel for dioxin experiments.13 The chlorine
(6) Fängmark, I.; Strömberg, B.; Berge, N.; Rappe, C. Environ. Sci.
Technol. 1994, 28, 624-629.
(7) Fängmark, I.; Strömberg, B.; Berge, N.; Rappe, C. Chemosphere
1994, 29, 1903-1909.
(8) Fängmark, I.; Strömberg, B.; Berge, N.; Rappe, C. Waste
Management Res. 1995, 13, 259-273.
(9) Gibb, W. H. International Conference Corrosion Resistant Materials for Coal Conversion; Applied Science Publishers: Barking, England,
1983; pp 25-45.
(10) Shao, D.; Hutchinson, E. J.; Cao, H.; Pan, W. P. Energy Fuels
1994, 8, 399-401.
(11) Mojtahedi, W.; Backman, R. J. Inst. Energy 1989, 189-196.
(12) Dayton, D. C.; French, R. J.; Milne, T. A. Energy Fuels, 1995,
9, 855-865.
Figure 1. Percentage of released chlorine as a function of the
pyrolytic temperature for different biomass.
in the synthetic waste was added as PVC (poly(vinyl chloride)),
AlCl3, and CuCl2, i.e., a mixture of organic and inorganic
chlorides.
The pyrolysis experiments were performed by heating a
vessel containing approximately 2 g of the desired fuel in a
quartz tube inside an electric furnace in a stream of nitrogen
(1.3 L/min). The furnace was heated at 50 °C/min to the
desired temperature. This temperature was held for 30 min,
and thereafter the sample was cooled under nitrogen and
weighed. The fraction of solid residue left in the vessel was
calculated and analyzed on its chlorine content.14
The chlorine release was not dependent on the pyrolysis
time at 600 °C, since at that temperature changing the reaction
time from 30 to 10 min gave a negligible change in the results
(less than 0.3%).
The reproducibility of the results was checked by duplicating
some experiments. The largest difference found in the chlorine
release between two identical experiments was 0.9%.
Leaching of the fuel was performed by mixing approximately
10 g of the original fuel with 130 mL of distilled water. The
mixture was stirred for 1 h at room temperature. Thereafter,
the water and fuel were separated by filtration and the
residual fuel was washed three times with 50 mL of H2O.
The influence of elevated pressure on the chlorine release
was studied in thermobalance, which is described elsewhere.15
Sampling of the pyrolysis gas was done in an impinger
bottle, cooled with an ice/salt mixture. The exit of the reactor
was washed with a small amount of acetone (approximately
three times 10 mL) after the finish of each experiment. The
washing solution was added to the acetone solution from the
impinger bottle. The acetone fractions were analysed by GCMS in order to identify any chloroorganic compounds.
The combustion experiments were carried out in a fluidized
bed, which previously has been used for investigating dioxin
formation from combustion of waste6-8 and with a cooling
section with ports for flue gas sampling. The fuel feeding rate
was approximately 1 kg/h.
Results
The released chlorine during pyrolysis of biomass at
different temperatures is given in Figure 1 as percentage of the original content.
The different biomass fuels presented in Figure 1 had
a similar profile for the chlorine release during pyrolytic
(13) Fängmark, I.; Strömberg, B.; Berge, N.; Marklund, S.; Rappe,
C. Studsvik Rep. 1990, No. EP-90/25, in Swedish.
(14) Ion chromatography was used for the chlorine analysis, after
combustion in a small bomb and sampling of the flue gases.
(15) Blackadder, W.; Rensfelt, E. Proceedings of International
Conference on Fundamental of Thermochemical Biomass Conversion;
Elsevier Applied Science: London, 1985; p 747.
1028 Energy & Fuels, Vol. 11, No. 5, 1997
Bjorkman and Stromberg
Figure 2. Amount of alkali metal and chloro atoms (mmol/
100 g fuel) in the biomass used in the study.
Table 2. Amount of Na, K, and Mg in the Fuels
(wt % on Dry Basis)
fuel
Na
K
Mg
sugarcane trash
switch grass
switch grass leached
lucerne
straw
0.027
0.021
0.027
0.061
0.15
1.2
1.8
0.17
1.9
0.84
0.14
0.12
0.037
0.15
0.057
conditions (N2) between 200 and 900 °C, with the
exception of straw (rape). All the fuels had lost between
20 and 50% of all the chlorine at 400 °C. At 200 °C
nearly all of the original chlorine was still in the solid
residue, indicating that the chlorine do not leave with
water when the fuel was dried. The negative value for
the chlorine release from lucerne at 200 °C is probably
due to inhomogeneity in the fuel or to an experimental
error. An experimental error is quite probable since
there are very small changes in both weight loss and
chlorine concentrations at such low temperatures.
Between 400 and 700 °C only small changes in the
chlorine release could be observed. Above 700 °C the
amount of released chlorine did increase with the
temperature for sugarcane trash, grass, and lucerne, but
not for straw.
Synthetic waste was incorporated in Figure 1 for
reasons of comparison, and it clearly shows that the
chlorine release from waste and biomass are of the same
magnitude within the investigated temperature interval.
The chlorine release reactions did not seem to be
dependent on the heating rate of the fuel, since no
significant changes in the loss of chlorine by increasing
the heating rate could be detected.
To get an estimate of the proportion of inorganic
chlorides in the fuels, the amount of sodium, potassium,
and magnesium were determined (Table 2). Recalculating the values in Table 2 to mmol/100 g dry fuel and
comparing with the corresponding values for chlorine
shows that the amount of potassium was high enough
to bind all the chlorine as KCl (Figure 2).
Changing the reaction conditions from pyrolytic to
gasification did not significantly change the chlorine
release (Figure 3). The only exception was that CO2
seemed to enhance the chlorine release at higher
temperatures. This may be explained by increased
gasifying reactivity toward CO2 at 800 °C and above,
which is also indicated by the fact that 97% of the fuel
was consumed when sugarcane trash was gasified in
CO2 (in N2 approximately 70% had left at 900 °C).
Figure 3. Percentage of released chlorine under different
gasifying atmospheres.
Figure 4. Percentage of released chlorine from sugarcane
trash at elevated pressure. The point at 0.1 MPa was made to
check the reproducibility between different apparatuses.
Figure 5. Percentage of released chlorine from two different
coals during pyrolytic conditions. The curve from switch grass
is added as a reference.
Increased pressure did not significantly affect the
chlorine release reactions (Figure 4). The figure indicates that at elevated pressures the chlorine release
reactions are somewhat slower, but still significant
amounts of chlorine are released at higher pressures.
The chlorine release from two different coals was also
studied (Figure 5). The results are similar to those that
can be found in the literature9,10 and can be summarized
by the observation that the chlorine is easily removed.
In order to establish the amount of water-soluble
Release of Chlorine from Biomass
Energy & Fuels, Vol. 11, No. 5, 1997 1029
Table 3. Chloroorganic Compounds That Are Formed during Combustion of Lucerne and Waste in a Small
Fluidized Beda
short residence time
compound
tetrachloroethylene
1,2,5-trichlorobenzene
1,2,3-trichlorobenzene
1,2,4-trichlorobenzene
1,2,3,4- + 1,2,4,5-tetrachlorobenzene
1,2,3,5-tetrachlorobenzene
pentachlorobenzene
hexachlorobenzene
long residence time
lucerne
(µg/m3)
lucerneb
(µg/m3)
10.6
5.48
2.92
4.26
2.99
1.60
0.98
0.92
1.39
1.67
1.93
0.955
0.808
0.586
lucerne
(µg/m3)
lucerneb
(µg/m3)
synth waste
(µg/m3)
6.66
37.3
104.4
119.7
200.5c
96.6
62.8
71.2
6.68
30.3
70.3
98.9
106.4
84.4
51.8
393.8
5.3
34.6
88.1
93.7
144.1
87.1
53.4
a The sampling was made in the cooling section at 250-300 °C and after two different residence times. b Duplicate sample. c The
quantification number is uncertain due complicated spectra.
Figure 6. Amount of solid residue (on ash-free basis) left after
pyrolysis at different temperatures.
chlorine, a leaching test was performed on the switch
grass (summer harvest). During leaching for 1 h in
water, the chlorine content was reduced from 0.79% to
0.06%. In other words, approximately, 93% of all
chlorine in grass is extractable and therefore most
probably in the form of a salt. The rest of the chlorine
left the fuel before 400 °C at pyrolytic conditions. (In
the calculations above, it was assumed that the only
leachable substance in the original grass was KCl.) As
a comparison, the majority of the chlorine in coal16 and
in biomass17 (rice straw) may be removed as chloride
ions with exhaustive washing.
The amount of char (calculated on ash-free basis) left
after the pyrolysis was similar for all fuels, except for
the leached grass (Figure 6). When comparing the
amount of char left from the leached and the unleached
switch grass it seems that lowering the chlorine content
lowered the amount of char. The tendency that increased chlorine content increases the char yield has
also been observed when mixture of wood and poly(vinyl
chloride) (PVC) was pyrolyzed.18 The explanation of the
authors was that there is some interaction of HCl and
cellulose below 600 °C, which decreases the char reactivity.
Attempts were made to measure HCl in the product
gas, which unfortunately failed. The sampling of HCl
(16) Fynes, G.; Herod, A. A.; Hodges, J.; Stokes, B. J.; Ladner, W.
R. Fuel 1988, 67, 822-829.
(17) Jenkins, B. M.; Bakker, R. R.; Baxter, L. L.; Gilmer, J. H.; Wei,
J. B. Developments in Thermochemical Biomass Conversion; Blackie
Academic and Professional: London, 1997; pp 1316-1330.
(18) McGhee, B.; Norton, F.; Snape, C. E.; Hall, P. J. Fuel 1995, 74,
28-31.
was made by passing the gas through impinger bottles,
containing water. The water solution was sent for
analysis of its chloride content, but due to high concentrations of acetic acid, the chloride measurements were
uncertain.
We could not find any chloroorganic compounds, such
as aliphatic (one to six carbons) and aromatic (one to
two ring) compounds, in the tar produced during pyrolysis of switch grass at 300, 400, and 500 °C, which
are temperatures known to maximize the dioxin formation during combustion conditions.19 Most of the identified compounds were the decomposition products from
the lignin in the biomass fuel. The main result, besides
the fact that no chloroorganic compounds were found,
was that the amount of tar molecules increased with
the pyrolytic temperature, which can be expected.
Furthermore, we could not see any influence from the
gasifying media on the product composition of the tar.
On the other hand, during combustion of biomass,
chloroorganic compounds were formed in comparable
amounts to that of the combustion of waste (Table 3).
The sampling for the analysis of chloroorganic compounds was made in the cooling section at a temperature of around 300 °C, where dioxins are known to be
formed. We did not analyze dioxins but chlorinated
benzenes, and we have assumed that there is a correlation between the formation of chlorinated benzenes and
dioxins. Oxygen seemed to be a vital component for the
formation of chlorinated benzenes, since the difference
between gasification/pyrolysis reactions and combustion
reactions is the presence of oxygen. The same importance of the presence of oxygen for the formation of
chloro-organic compounds has also be found in the
chemistry of dioxins,19 in which the formation reactions
are catalyzed by fly ash and oxygen is essential.
Since no chloro-containing products from the pyrolysis
have been identified, we have unfortunately not been
able to do any Cl balance.
Discussion
The aim of this project was to determine the type of
chlorine in biomass by measuring the release of chlorine
during pyrolytic conditions. The working hypothesis
was that chlorine from inorganic salts will not leave
below their melting point, approximately 750 °C, while
organic chlorine would leave at significantly lower
temperatures. The background for the hypothesis is
that the vapor pressure of a chloride-containing alkali
(19) Strömberg, B. Chemosphere 1991, 23, No. 8-10, 1515-1525.
1030 Energy & Fuels, Vol. 11, No. 5, 1997
Bjorkman and Stromberg
Figure 7. X-ray photo of pyrolyzed switch grass. The X-ray pictures are based on the same photo (the bottom right photo): the
blue dots represent potassium atoms, the green dots chloro atoms, and the yellow dots silica atoms.
salt increases significantly over the melt compared to
the solid phase.20
Unexpectedly, since the majority of the chlorine in
biomass is believed to be KCl, our result showed that
20-50% of the chlorine left the fuels during pyrolysis
already at temperatures around 400 °C and the behavior of biomass fuels was not significantly different from
that of synthetic waste, even though the synthetic waste
did not contain any alkali.
The location of the chloro atoms on grass seemed to
be correlated with the potassium atom, as can be seen
from the X-ray photo of pyrolyzed switch grass in Figure
7. This indicates that the chloro atoms probably are
bounded to potassium as KCl. The figure may need
some explanation, the X-ray pictures are based on the
same photo (the bottom right photo), the blue dots
(20) Knacke, O., Kubaschewski, O., Hesselmann K. Thermochemical
Properties of Inorganic Substances I; Springer-Verlag: Berlin, 1991;
p 964.
represent potassium atoms, the green dots chloro atoms
and the yellow dots silica atoms.
In order to better understand the results in Figure 1,
we made mixtures of pulverised wood chips and KCl
(inorganic chloride) or PVC (organic chloride). Potassium chloride was added to wood fuel as a water
solution. The PVC was pulverised and mixed with the
wood by stirring. Additional to these fuel mixtures,
solid KCl was also mixed with silica sand. The resulting
chlorine concentrations are given in Table 4 and the
release of chlorine during pyrolysis is shown in Figure
8. In the figure the release of chlorine from switch grass
is added for comparison reasons.
Figure 8 clearly shows that if KCl is added to wood
and there is no other chlorine present, between 30 and
50% of the chlorine will leave already at temperatures
between 400 and 600 °C, while if the KCl is mixed with
sand there was no loss of chlorine below 600 °C. The
chlorine releases from the sand-KCl mixtures were
Release of Chlorine from Biomass
Energy & Fuels, Vol. 11, No. 5, 1997 1031
Figure 8. Percentage of released chlorine from wood mixed
with PVC, wood mixed with KCl, and sand mixed with KCl
during pyrolytic conditions. The curve from switch grass is
added as a reference.
Table 4. Chlorine Concentrations in the Model Fuels
Used in the Study
fuel
Cl (wt %)
wood
wood + PVC
wood + KCl
sand + KCl
0.02
0.18
1.82
2.19 (calcd)
calculated with the assumption that the measured
weight loss corresponded to KCl evaporation.
The reason for this unexpected behavior is not known.
It is clear that there was some interaction between wood
and KCl at temperatures between 300 and 600 °C.
Wood, compared to sand, is a reactive media during
pyrolysis, and therefore it may react and stabilize ion
that can be formed during thermal treatment. When
adding an inorganic salt to wood and increasing the
temperature, the chlorine atoms may move on the
surface of the fuel and thereby associate to other atoms
than the alkali metal.
When chlorine release from biomass during combustion/gasification has been studied in the literature, it
has been generally studied coupled with alkali release.
It has been found in these studies that the KCl(g) is
one of the most stable alkali containing species in the
gas phase at high temperatures3,12 (800 °C and above).
In many cases, the amount of alkali vaporized during
biomass combustion is more dependent on the amount
of chlorine available to form stable vapors than by the
amount of alkali in the fuel.3 Unfortunately, the
vaporization of KCl cannot explain our results, since we
got a substantial loss of chlorine already at 300-400
°C and at these temperatures the vapor pressure of KCl
is negligible,20 which has been confirmed by the negligible chlorine release from the sand-KCl mixture in
shown in Figure 8.
One possible explanation for the chlorine release is
that the potassium chloride reacts with steam formed
from the biomass itself to give gaseous HCl and KOH.
Equilibrium calculations21 indicate that this formation
is small. But if an acidic oxide is added (in our
calculations SiO2), the formation of HCl(g) formation is
significant and probably coupled to the formation of
potassium silicates, since both begin to appear at the
(21) The program used for the calculations was HSC Chemistry for
Windows, Outokumpu, Finland.
Figure 9. Percentage of released chlorine from switch grass
and cellulose with added KCl(aq) during pyrolytic conditions.
The concentrations of chlorine in unpyrolyzed fuel were 0.79
and 0.78, respectively. The curve from switch grass is added
as a reference.
same temperature; see the proposed reaction below:
2KCl + nSiO2 + H2O(g) f K2O(SiO2)n + 2HCl(g)
This reaction seems to be important already at 400 °C,
according to the thermodynamic calculations. The
amount of HCl(g) and potassium silicate formed increases with temperature and steam concentration.
The calculations indicate that the vaporization of KCl
starts at higher temperature, around 650 °C.
Summarizing, Figure 8 and the equilibrium calculations clearly indicate that even if the chlorine in the
biomass is present as an inorganic salt, it can leave at
temperatures well below the corresponding melting
point for the salt during pyrolytic conditions.
There was no loss of chlorine from a sand-KCl
mixture when simulated dry or wet fuel gas was used
as the surrounding atmosphere. This result indicates
that the chlorine release reactions are governed by the
solid phase and cannot be induced by gas phase reactions only.
Increasing the alkali metal concentration in a fuel
increases the sintering ability of the fuel. This may be
the explanation why 60% of the chlorine still was left
after heating the sand-KCl mixture to 900 °C (Figure
8). The effect could be seen when the reactor was taken
apart. Sintering can also be one explanation why the
curve for straw in Figure 1 has a different shape than
the curves for the other fuels. The straw contained
significantly higher amounts of sodium than the other,
which is known to lower sintering temperatures.
Possible binding-association sites, other than alkali
metal ions, for the negative chloride ions in the biomass
are basic nitrogen sites. It has been suggested in the
literature10 that the chloro atoms were associated with
the nitrogen atoms, since the evaporation curve for HCl
and NH3 had a similar form. To test this hypothesis
we mixed KCl(aq) with pure cellulose (without any
nitrogen at all) and pyrolyzed it. The achieved loss of
chlorine (30%) at 400 °C (Figure 9) was comparable with
that of switch grass. Our result showed clearly that the
presence of nitrogen atoms in the fuel is not necessary
for chlorine release reactions, and thereby the proposed
association between nitrogen and chloro atoms is unlikely.
1032 Energy & Fuels, Vol. 11, No. 5, 1997
Since we have not observed any difference in the
chlorine release between biomass and waste and it is
known that dioxins are formed during combustion of
waste, the formation of dioxins from thermal treatment
of biomass cannot be ignored. However, no precursors,
i.e., simple chlorinated organic compounds, for dioxins
was found in the tars produced when pyrolyzing or
gasifying biomass. A comparison with coal shows that
during a mild pyrolysis (<300 °C) of a coal with a high
chlorine content (most probably as brine derived chlorides) no chlorinated organic compounds could be detected.16 On the other hand, combustion of lucerne
produced similar amounts of chlorinated benzenes as
combustion of synthetic waste (Table 3), indicating that
oxygen is an essential component in the formation of
chlorinated organic compounds. The importance of
oxygen may be explained by the theory that Cl2 is a
much more powerful agent for chlorination than HCl
and that Cl2 is formed in the flue gases via the Deacon
process reaction from HCl and O2, according to22
2HCl + 1/2O2 T Cl2 + H2O
We have not studied the loss of potassium but are
planning to determine this in the near future. Preliminary results have indicated that the release of potassium follows the release of chlorine even at low temperatures, but this does not imply that chlorine leaves
as KCl(g). Indications on the opposite are the low
temperatures for the chlorine release and the similarities between biomass and synthetic waste despite the
difference in the chlorine source (inorganic vs organic
chlorides).
Conclusions
The results in this work can be summarized as
follows:
1. During pyrolysis of biomass, 20-50% of the
chlorine was released at such low temperatures as 300
to 400 °C.
2. At 900 °C, 30-60% of the total chlorine was still
left in the char.
(22) Raghunathan, K.; Gullett, B. K. Environ. Sci. Technol. 1996,
30, 1827.
Bjorkman and Stromberg
3. There was no significant difference in the chlorine
release between biomass and synthetic waste during
pyrolysis.
4. Different gasifying atmospheres did not significantly change the chlorine release, nor did elevated
pressure.
5. 93% of the chlorine in switch grass was water
soluble.
6. No chloroorganic compounds could be found in the
tars produced when pyrolyzing switch grass at 300, 400,
and 500 °C.
7. Increasing the chlorine content seemed to increase
the char yield.
8. Chlorine could evaporate from a mixture of wood
and KCl at 400 °C but not from sand and KCl(s) under
similar conditions.
9. The assumption that the chlorine from inorganic
salt does not evaporate below the appropriate melting
point (i.e., that the vapor pressure over the solid salt is
negligible) is too simplified.
10. The presence of nitrogen atoms in the fuel is not
necessary for the chlorine release reactions.
11. Equilibrium calculations indicated that the chlorine release may be explained by the formation of
gaseous HCl coupled with the formation of potassium
silicates.
12. No chloroorganic compounds could be found in
the tar produced when pyrolyzing switch grass.
13. Combustion of biomass can give the same problem with emission of chloro compounds as the combustion of waste.
14. The partial release of chlorine already at low
temperatures does not make it possible to separate the
bulk of chlorine into either the char residue or to gas
products by pyrolysis. Such a treatment does not give
any benefit compared to direct combustion or gasification.
Acknowledgment. We thank NUTEK and The
Board for Non-Nuclear Research at Studsvik for financing the project. Furthermore, we thank Torbjörn Nilsson
and Frank Zintl for helpful advice, and Ewa Burek,
Patrik Gustafsson, Jan Oskarsson, and Camilla Larsson
for helping with the laboratory work.
EF970031O