The influence of level and chlorine source on the formation of mono- to octa-chlorinated dibenzo-p-dioxins, dibenzofurans and coplanar polychlorinated biphenyls during combustion of an artificial municipal waste

Article abstract of DOI:10.1016/S0045-6535(00)00155-7

The formation of polychlorinated dibenzo-p-dioxins (PCDDs), polychlorinated dibenzofurans (PCDFs) and the three coplanar polychlorinated biphenyls (pPCBs) was studied during labscale fludized bed combustion of eight artificial municipal solid waste (MSW) fuel mixtures. The level of chlorine as well as the chlorine source varies within the different fuel mixtures. Four different chlorine sources were studied, viz, an inorgnaic (NaCl) and three organic sources, pure PVC plastic and two products (floor and cable) and the total chlorine level varies between 0.28% and 1.1%. The experiments were performed in a 5 kW laboratory scale fluidized bed reactor. A correlation between the total chlorine in the fuel and the formation of the hepta- and octa-chlorinated PCDD/F homologues was found. However, the most important variable for changes in the PCDDs/Fs and pPCBs formation was disturbance in the combustion condition and not the variation in chlorine content of the fuel. Furthermore, no differences in formation between the chlorine sources could be seen.

Full text of DOI:10.1016/S0045-6535(00)00155-7

Chemosphere 43 -2001) 227±234
The in¯uence of level and chlorine source on the formation of
mono- to octa-chlorinated dibenzo-p-dioxins, dibenzofurans
and coplanar polychlorinated biphenyls during combustion of
an arti®cial municipal waste
*

E. Wikstrom , S. Marklund
Department of Chemistry, Environmental Chemistry Umeꢀa University, SE-901 87 Umeꢀa, Sweden
Received 27 September 1999; accepted 5 April 2000
Abstract
The formation of polychlorinated dibenzo-p-dioxins -PCDDs), polychlorinated dibenzofurans -PCDFs) and the
three coplanar polychlorinated biphenyls -pPCBs) was studied during labscale ¯udized bed combustion of eight arti-
®cial municipal solid waste -MSW) fuel mixtures. The level of chlorine as well as the chlorine source varies within the
di€erent fuel mixtures. Four di€erent chlorine sources were studied, viz, an inorgnaic -NaCl) and three organic sources,
pure PVC plastic and two products -¯oor and cable) and the total chlorine level varies between 0.28% and 1.1%. The
experiments were performed in a 5 kW laboratory scale ¯uidized bed reactor. A correlation between the total chlorine in
the fuel and the formation of the hepta- and octa-chlorinated PCDD/F homologues was found. However, the most
important variable for changes in the PCDDs/Fs and pPCBs formation was disturbance in the combustion condition
and not the variation in chlorine content of the fuel. Furthermore, no di€erences in formation between the chlorine
sources could be seen. Ó 2001 Published by Elsevier Science Ltd.
Keywords: PVC; NaCl; PCDD/F; PCB; Labscale ¯uidized bed reactor; MSW
1. Introduction
fuel for the formation of chlorinated organic pollutants
have been performed, but with contradictory conclu-

The in¯uence of the polyvinylchloride -PVC) content
on the formation of polychlorinated dibenzo-p-dioxins
-PCDDs) and polychlorinated dibenzofurans -PCDFs)
during municipal solid waste -MSW) combustion has
been discussed during the last several years. Studies on
the importance of the total amount of chlorine in the
sions. Fangmarket al. -1993) investigated how varia-
tions within di€erent combustion parameters, such as
bed temperature, O2-, HCl-, H2O-concentrations, tem-
perature and residence time of the ¯ue gas in the post-
combustion zone e€ected the PCDD/F formation. The
study showed a weakcorrelation between the amount of
HCl in the ¯ue gas and the formation of PCDDs/Fs;
other parameters, such as the residence time in the post-
combustion zone, were more important. The amount of
HCl added to the reactor in that study corresponded to
combustion of a waste containing approximately 2.2%
of PVC. Further, combustion experiments was per-
formed by Kanters et al. -1996) in a smaller laboratory
scale reactor -1 g fuel/experiment). The reactor was fed
* Corresponding author. Tel.: +1-919-541-5548; fax: +1-919-
541-0554. Present address: U.S. EPA -MD-65), National Risk
Management Research Laboratory, Research Triangle Park,
NC 27711, USA.
E-mail address: wikstrom.evalena@epa.gov -E. Wi-

kstrom).
0045-6535/01/$ - see front matter Ó 2001 Published by Elsevier Science Ltd.
PII: S 0 0 4 5 - 6 5 3 5 - 0 0 ) 0 0 1 5 5 - 7

228
E.Wikstr om, S.Marklund / Chemosphere 43 (2001) 227±234
with an arti®cial MSW fuel that contained eight times
higher PVC loads than normal. Only a slight increase of
the chlorophenol concentration in the ¯ue gas could be
discerned in this experiment. The same laboratory re-
actor was fed with an chlorine-free arti®cial MSW, and
additional HCl was added to the wet carrier air. Already
at such low level as 0.1 g HCl/-norm) m3 added, the
chlorophenol level in the ¯ue gas was doubled. Further
increasing in the HCl level -up to 2 g HCl/-norm) m3)
did not cause any signi®cantly higher chlorophenol
formation. As a comparison, the average HCl level
during MSW combustion in full scale plants varies be-
tween 0.35 and 0.90 g/-norm) m3, i.e., a level lower than
2 g HCl/-norm) m3 which indicates that higher chlorine
loads in the MSW do not results in increased outputs of
chlorophenols. The only experiment with a low level of
chlorophenols was the one without HCl added to the
¯ue gas.
fractions are then well mixed and pelletized into fuel
pellets. The properties, composition and elementary
content of the arti®cial MSW are further described in

Wikstrom and Marklund -1998a). Table 1 lists the total
chlorine levels, the chlorine origin from the four di€erent
sources, the heat capacity and the dry weight of the
fuels.
In total, 13 experiments were performed in the 5 kW
laboratory scale ¯uidized bed reactor -Fig. 1); a detailed

description of the reactor is given in Wikstrom et al.
-1998b). The reactor was carefully cleaned between each
experiment and the sand bed was replaced with 450 ml
unused, washed sifted sand -250±125 lm), to minimize
the memory e€ects. The reactor is equipped with elec-
trical pre-heating systems; three at the freeboard and
four at the convector section, to prevent unwanted
temperature variations and to achieve stable tempera-
ture conditions faster. To minimize the e€ect of cold
walls and reactions with sooty wall surface, the reactor
was pre-heated with propane for about 3 h before the
pellets fuel feeding was started. Table 2 lists the settings
of the air¯ows, the e€ect of the fuel feeder and electrical
pre-heating systems, pre-heated secondary air was used
in this study. The listed measured temperatures in the
reactor and convector section are shown as average
temperatures Æ the standard deviation among the 13
experiments.
The aim of this study was to investigate how the
amount of inorganic chlorine -introduced as NaCl) and
organic chlorine, bound in PVC plastic, in¯uenced the
emission of PCDD/Fs and coplanar polychlorinated
biphenyls -pPCBs) during MSW combustion. One pure
virgin PVC material and two frequently used PVC
products; ¯oor -product name Norvinyl) and cable
-product name Alcatel) were chosen to study the be-
havior of di€erent additives in PVC products and what
in¯uence these had on the emissions of chlorinated or-
ganic pollutants.
The inorganic gases CO2, CO and O2 were continu-
ously measured at the end of Section 3 in the convector.
One IR-instrument, Siemens Ultramat 22 P for CO2 and
CO -0±1000 ppm) and one, ECOM SG-PLUS instru-
ment that uses electrochemical cells to determine the
amount of O2, CO -100±10,000 ppm) in the ¯ue gases
were used.
2. Materials and methods
Eight fuels with di€erent chlorine content and chlo-
rine sources were prepared by adding di€erent amounts
of chlorine to an arti®cial MSW fuel mixture. The
chlorine content of pure PVC plastic is about 56 wt%
and due to the amount of added additives in the PVC
¯oor and PVC cable the chlorine content is lower in
these product -25 wt%). The arti®cial MSW is a mixture
of eight material fractions, paper, plastic, laminate,
textile, glass, kitchen and garden sweepings, remaining
combustible and remaining incombustible -metals). The
When the temperature conditions in the reactor were
stable, the MSW fuel feeding was started. After one
hour of MSW combustion a particulate matter sample
was taken at the end of convector part ®ve. The sam-
pling period was 60 min, which corresponds to a ¯ue gas
sample of 1.6 m3. Parallel with this sample another two
samples were taken at convector part four to determine
the HCl level in the ¯ue gas. The samples were collected
in an H2CO3/HCO3 bu€er; the volume of the ¯ue gas in
Table 1
Chlorine level and heat capacity of the fuel pellets
Element
Unit
1
2
3
4
5
6
7
8
DW
%
MJ/kg
95.7
18.9
)
95.4
18.8
)
)
0.17
95.5
18.9
0.25
)
)
0.15
95.4
18.7
)
)
)
0.70
95.3
18.6
0.16
0.08
)
95.6
18.9
0.13
)
0.07
0.50
95.9
18.9
)
95.8
18.8
)
)
0.27
Heating value
Clorganic pure PVC
Clorganic PVC ¯oor
Clorganic PVC cable
Clinorganic -NaCl)
Ã
%
0.10
0.35
)
0.18
)
0.65
Ã
%
0.18
0.42
0.83
Ã
Total Cl
* % of dry weight.
%
0.28
0.35
0.40
0.62
0.66
0.70
1.0
1.1

E.Wikstr om, S.Marklund / Chemosphere 43 (2001) 227±234
229
Fig. 1. Drawing of the 5-kW laboratory scale ¯uidized bed reactor. The sampling ports used in this study are marked with arrows.
each sample was about 50 dm3. The chlorine ion con-
centration was then determined by ion chromatography
according to method SS-EN ISO 10304-1, and the HCl
level in the ¯ue gas was calculated. Directly after the
particle sampling, two consecutive samples -denoted as
sample A and B) for determination of the PCDD/Fs
were started, i.e., in total 26 ¯ue gas samples were taken
-13Â2). Due to one unsuccessful sampling the real
number of samples were 25 instead of 26. The samples
were taken after convector part three and were per-
formed according to European standard EN 1948:1 with
the isokinetically cooled probe technique. The gas vol-
ume was noramlized -norm) to 1 atm dry gas at 0°C and
10% CO2 -Marklund et al., 1992; European standard
1997). The residence time of the ¯ue gas was 4.6 s in the
freeboard and further 2.2 s in the convector section -port
1±3). The sampling time for each sample was 50 min,
which corresponds to a sample volume of 0.75 m3. The
samples were analyzed for mono- to octa-PCDD/Fs and
the three coplanar PCBs #77, #126 and #169. The ex-
traction and clean-up procedure was performed ac-
cording to standard methods -Marklund, 1990; van
Bavel et al., 1992). The sampling train was spiked with
13C labeled compounds before the sampling was started,
this enables correction of the levels for losses during the
sampling and clean-up procedure. The PCDD/Fs and
the pPCBs were analyzed with a sector magnetic
HRGC/HRMS -VG 70 S, with a Supelco 2330 column).
Table 2
The parameter settings during the experiments and achieved
temperatures in the reactor and convector section. The location
of the thermocouple can be seen in Fig. 1
Parameter
Setting
Fuel and air¯ows
Fuel
940 g/h
-4.9 kW)
80 l/min
Primary air
Pre-heating temperature:
25°C
Pre-heating temperature:
Secondary air
70 l/min
2.1. Principal component analysis
150°C
Temperature settings
Bed section
Measured temperature
Bed: 860 Æ 32
F1: 910 Æ 30
F2: 840 Æ 21
F3:750 Æ 52 F4:
760 Æ 64
Principal component analysis -PCA) was used to
evaluate similarities and dissimilarities within the re-
sulting data. An exhaustive description of the method
used in PCA can be found in Jolli€e -1986) and Jackson
-1991). The primary intention of PCA is to obtain an
overview of the dominant ``pattern'' in a data set. The
data set is divided into objects -here; each PCDD/F ¯ue
gas sample period) and measured variables -here; e.g.,
concentrations of di€erent PCDD/F homologues and
CO levels during the sampling period). The variables is
700°C
Middle section 600°C
Top section
800°C
Convector 1
Convector 2
Convector 3
Convector 4
400°C
350°C
300°C
250°C
C1: 630 Æ 12
C2: 419 Æ 5
C3: 351 Æ 6
C4: 292 Æ 7
C5: 281 Æ 7

230
E.Wikstr om, S.Marklund / Chemosphere 43 (2001) 227±234
used to characterize the objects, e.g. the unique PCDD/F
homologue level in the ¯ue gas for each sampling period.
The results of the PCA are visualized by two ®gures, a
so-called score plot and a loading plot. Samples with
similar PCDD/F emissions and pattern will be located
close to each other in the score plot, while those which
have divergent emission pattern will be located further
apart. The loading plot shows how the measured vari-
ables relate to each other. Variables close to each other
in the loading plot correlate, and variables diagonally
across from each other have a negative correlation. The
most important variables for the PCA model are located
far from the origin and the insigni®cant variables are
positioned close to the origin in the loading plot. Since
the variables characterize the objects, it is possible to
establish which variables -here: a speci®c homologue
concentration) that are responsible for the separation of
the samples in the score plot by examining the loading
plot. Prior to the PCA, the data was scaled to unit
variance and mean-centered. Cross-validation was used
to calculate the number of principal components -PC) to
be included to provide a signi®cant valid model. The
total fraction of the systematic variation in the data
explained by the model is expressed as an R2 value and
the prediction ability of the model is expressed as a Q2
value. The validity of a model is decided by the R2 and
Q2 value -<1.0).
vided into three groups: taken during bad, fair and good
combustion conditions -Table 3). The samples were di-
vided according to their emissions of CO during the
sampling period and number of CO-peaks over 2000
ppm. The data from the samples taken during fair and
good combustion conditions were further evaluated to-
gether as one group. The average PCDD and PCDF
pro®le of the 17 samples taken during good and fair
combustion are shown in Fig. 2. The average pro®le
agrees with an average PCDD/F pro®le found in full-
scale MSW combustors, which shows that the formation
conditions -e.g., catalytical reactions, temperatures) are
similar in the labscale and full-scale systems.
A plot of the total chlorine load in the fuels versus the
emissions of I-TEQs during good and fair combustion
conditions are shown in Fig. 3. A larger variation within
the same chlorine level is discerned than between the
di€erent chlorine levels, consequently no correlation
between the chlorine content and the emissions appears
to exist with in the data set -R2 ˆ 0:07).
Further evaluation of the data was done with PCA.
This model consists of 17 observations -all samples
taken at fair and good combustion conditions) and 25
variables: mono- to octa-PCDD/F homologues, the
three pPCBs the O2, CO2, CO, HCl, particulate matter
and Total Cl in the fuel. The two ®rst components of the
model are shown in Figs. 4 and 5. Component 1 de-
scribes most of the important information relating to
di€erences between the samples -R2 ˆ 0:50), which is
mainly the di€erence in PCDD/F and pPCB emission
levels. The second component -R2 ˆ 0:28) mainly de-
scribes the variation in the homologue pattern between
the di€erent samples. The loading plot -Fig. 4) show as
well a distinct correlation between the formation of the
chlorinated homologues of the PCDDs/Fs and pPCBs.
For example, the PeCDD, PeCDF and PeCB -#126) are
placed close to each other in the loading plot, the same
can be seen for the tetra- and hexa-chlorinated homo-
logues -Fig. 4). This similarity indicates that they are
formed through a similar formation mechanism. A
correlation between the Total Cl in the pellets and for-
mation of the higher chlorinated PCDD/F; namely the
hepta- and octa-chlorinated congeners can be noticed as
well in the loadingplot. A separate PCA evaluation -not
shown) of the ten samples taken during good combus-
tion conditions showed as well a signi®cant positive
correlation between the Total Cl level in the fuel, HCl,
hepta- and octa-chlorinated PCDDs/Fs in the ¯ue gas
-Validity of the model: R2 ˆ 0:91, Q2 ˆ 0:60). A study
performed by Vehlow et al. -1996) in the TAMARA
reactor found a similar correlation between the higher
chlorinated PCDDs and the chlorine load in the fuel.
Our study shows that the amount of chlorine in the
waste is not the limiting parameter for formation of
chlorinated compounds. The amount of chlorine is
presented in percentage levels in the waste, which is
3. Results and discussion
Table 3 lists the measured levels of PCDDs, PCDFs,
pPCBs, O2, CO2, CO, HCl, particulate matter and
number of CO-peaks in the ¯ue gas during each sam-
pling. The O2-, CO2- and CO-levels are all calculated as
average values during the PCDD/F sampling period
-ꢀ50 min). The O2 level in the 13 experiments varies
between 10.3% and 13.3% while the corresponding
variation in CO2 emission is between 8.0% and 11.0% of
CO2. One of the largest di€erences between the experi-
ments is the variation in CO emission -9±727 ppm). The
higher CO emissions was caused by uncontrolled dis-
turbances of the fuel feeding system of the reactor. A
PCA evaluation -not shown) of the complete data set
showed that the eciency of the combustion conditions
is a parameter that in¯uences the formation of PCDDs/
Fs and pPCBs more than the total chlorine content in
the fuel -validity of the model: R2 ˆ 0:86, Q2 ˆ 0:51). A
positive correlation between PCDDs/Fs, pPCBs forma-
tion and CO concentration was noticed. A conclusion
that can be seen in Table 3 as well. Even, a negative
correlation between the Total Cl and formation of
PCDDs/Fs and pPCBs was found within the complete
data set.
To evaluate the data with minimal in¯uence from
di€erent combustion conditions, the samples were di-

Table 3
The levels of PCDD/Fs, pPCBs -as -norm) m3), O2, CO2, CO, HCl and particulate matter in the ¯ue gas in all experiments
Good combustion
Fair combustion
3A
3B
4Aa
4Ba
1Aa
1Ba
7Aa
7Ba
6Aa
2A
6A
6B
4A
7A
7B
8B
5Ba
1±8 CDF -ng/m3)
1±8 CDD -ng/m3)
I-TEQ -ng/m3)
177
18
170
17
57
12
47
9
103
13
134
19
73
16
90
18
648
62
501
53
593
68
1194
126
25.8
5.5
116
16
471
66
219
41
960
170
22.4
1.9
3.7
1.1
12.8
8.3
63
911
101
21.1
NA
NA
NA
10.3
10.5
234
5
4.4
0.8
4.0
0.6
0.4
0.1
10.8
10.5
20
1.3
0.2
0.2
0.1
12.3
8.5
9
1.0
0.1
0.1
0.1
12.1
8.8
10
2.2
0.3
0.3
0.1
12.3
8.7
67
3.5
0.3
0.2
0.1
12.8
8.1
10
1.6
0.2
0.2
0.1
12.9
8.5
20
2.0
0.2
0.2
0.1
13.3
8.0
12
16.0
4.1
3.0
0.7
10.7
10.3
40
11.6
3.7
2.6
0.6
11.9
NA
40
13.8
2.0
4.1
1.8
12.1
8.7
157
7
2.9
0.5
0.5
0.2
11.7
9.4
121
5
11.4
1.2
1.8
0.8
11.9
8.9
146
8
5.4
0.6
0.9
0.4
12.1
8.6
88
PCB #77 -ng/m3)
PCB #126 -ng/m3) 0.5
PCB #169 -ng/m3) 0.1
10.5
3.7
O2
CO2
%
11.1
10.3
20
12.0
8.9
%
CO -ppm)
Number of
CO-peaksb
68
4
0
0
0
0
4
0
0
0
1
0
3
4
HCl -mg/Nm3)
PM -mg/Nm3)c
150
650
520
650
250
2300
750
2300
960
700
580
700
370
1100
720
1100
160
1200
NA
830
230
620
350
620
320
560
470
650
600
650
1050
1200
220
870
Bad combustion
5A
5B
4B
8A
6Ba
5Aa
1A
1B
1±8 CDF -ng/m3)
1±8 CDD -ng/m3)
I-TEQ -ng/m3)
1730
555
39.0
3.6
1360
139
29.0
1.3
1090
157
27.4
2.8
2600
445
87.8
7.8
425
49
547
52
4700
628
106
12.3
35.1
13.1
10.7
10.9
727
43
25 000
3900
670
12.0
2.2
2.6
0.8
9.5
11.0
334
11
16.0
2.1
PCB #77 -ng/m3)
40.5
433.9
15.0
10.9
9.6
PCB #126 -ng/m3) 13.2
PCB #169 -ng/m3) 5.8
6.2
3.1
6.2
2.8
36.7
10.3
12.6
8.5
3.3
0.9
O2
CO2
%
11.2
9.6
515
22
10.8
10.1
535
23
11.2
9.9
10.9
10.6
334
23
%
CO -ppm)
Number of
CO-peaksb
268
10
143
8
441
19
HCl -mg/Nm3)
PM -mg/Nm3)
240
680
730
680
420
560
1100
1200
260
1200
290
870
NA
1000
NA
1000
a These samples were taken during a second -repeated) combustion experimental run.
b Number of CO-peaks: Total number of CO-peaks over 2000 ppm during the sampling period of chlorinated organic compounds.
c PM: Particulate matter.

232
E.Wikstr om, S.Marklund / Chemosphere 43 (2001) 227±234
Fig. 2. The relative abundance Æ the 95% con®dence interval of PCDD and PCDF in the samples taken during good and fair
combustion conditions.
Fig. 3. The total Cl level of the fuel and the I-TEQ level in the 17 samples taken during good and fair combustion condition.
magnitudes higher than the amount of chlorine needed
for formation of chlorinated compounds, such as
PCDDs/Fs. If the chlorine level is less than 1% or more
is not signi®cant for the formation, since the chlorine is
present in excess amount already at very low -< 0:1)
percentage level. In a previous investigation, the for-
mation of PCDD/Fs and chlorobenzenes was studied
during combustion of seven arti®cial MSW fuels with a
chlorine content varying between 0.12 and 1.7 wt%
and chlorobenzenes was noticed when a fuel with 1.7
wt% of chlorine was combusted. A correlation that
cannot be seen in this data set. The major di€erences
between these two studies are the complicity of the waste
mixture, which probably e€ects the catalytic activity of
the fuel. The recently used fuel origin from a much more
complex mixture than the previous one. The lower cat-
alytic activity of the fuel could be the reason why the
high chlorine level showed signi®cant higher emissions
with this fuel and not with the more complex fuel.
Furthermore a single-tailed ANOVA analysis of the
2,3,7,8-substituted congeners in the samples from the
four di€erent chlorine sources was done to evaluate if
the isomer pattern of these isomers changed between the

-Wikstrom et al. 1996). The results showed no correla-
tion between the chlorine source -organic or inorganic
chlorine), chlorine loading -if less than 0.84 wt%) and
the formation of PCDD/Fs and chlorobenzenes. How-
ever, a distinct increase in the formation of PCDD/F

E.Wikstr om, S.Marklund / Chemosphere 43 (2001) 227±234
233
Fig. 4. The loading plot from the PCA evaluation of the data combusted during fair and good combustion conditions. The complete
model consists of four signi®cant components with a total R2 ˆ 0:88 and a total Q2 ˆ 0:57.
Fig. 5. The score plot of the 17 samples taken during fair and good combustion conditions.
sources. The calculation showed no signi®cant di€er-
ences between the chlorine sources and the congener
pattern formed during the combustion process
-Fcrit: ˆ 2:27, Fcal: ˆ 0:0086). The same congener pattern
and the correlation between the Cl-homologues in Fig. 4
indicates that the chlorinated organic compounds are
formed through the same mechanism independently of
the chlorine sources. One explanation could be that they
are formed in two steps, one where the organic skeletons
-e.g., the dibenzofuran) are formed and a consecutive
in order to obtain better insights in formation pathways
and mechanisms.
4. Conclusions
·
·
·
Variation in combustion conditions is a parameter
that in¯uenced the emissions of PCDDs/Fs and
pPCBs more than the chlorine content of the arti®cial
MSW fuel -0.28±1.1% Cl).

secondary chlorination reaction as shown by Wikstrom
and Marklund -2000).
Positive correlations between the hepta- and octa-
chlorinated PCDDs/Fs and the Total Cl content in
the MSW were noticed in the samples taken during
good and fair combustion conditions.
Further research is needed on how the two chlorine
sources, inorganic and organic, in¯uence ± and con-
tribute to ± the formation of chlorinated organic pollu-
tants during combustion under less ecient conditions
The ANOVA analysis showed that there was no sig-
ni®cant di€erence in the formation of PCDD/Fs be-

234
E.Wikstr om, S.Marklund / Chemosphere 43 (2001) 227±234
Marklund, S., 1990. Dioxin emissions and environmental
tween the two PVC products, pure PVC and NaCl in
this study.
ꢀ
imissions, Environmental Chemistry, Umea University,
Sweden.


Marklund, S., Soderstrom, G., Ljung, K., Rappe, C., Kraft,
M., Hagenmaier, H.-P., 1992. Parallel sampling using
various sampling techniques at a Swedish municipal solid
waste incinerator. Waste Manage. Res. 10, 21±36.
Acknowledgements
The study was supported by TFR -Swedish Research
Council for Engineering Sciences, Teknikvetenskapliga
Forskningsrꢀadet in Swedish) and NorskHydro AB.



van Bavel, B., Fangmark, I.E., Marklund, S., Soderstrom, G.,
Ljung K., Rappe, C., 1992. Quantitative analysis of organic
trace compounds by isotope dilution: Formation of the non-
ortho PCBs 77, 126, 169 during incineration. Organohalo-
gen Compound vol. 8, pp. 225±228.

Special thanks to Per Liljelind, Gunilla Soderstrom, Lisa
Borjeson, Gunilla Lofvenius and Maria Hjelt for all



Vehlow, J., Rittmeyer, C., Prang-Stotz, G., Vogg, H., Mark,
F.E., Kayen, H., Freiesleben, W., Lescuyer, J.-L., 1996. Co-
combustion of waste plastics and municipal solid waste in
the Karlsruhe test incinerator TAMARA, Solid Waste
Manage. Thermal Treat. Waste-to-Energy Technology, pp.
405±414.
their help with experimental and laboratory work.
References
European Standard, 1997. Stationary source emissions-deter-
mination of the mass concentration of PCDDs/PCDFs. EN,
1948, pp. 1±3.

Wikstrom, E., Marklund, S., 1998a. Combustion of an arti®cial
municipal solid waste in a laboratory ¯uidised bed reactor.
Waste Manage. Res. 16, 342±350.


Fangmark, I.E., van Bavel, B., Marklund, S., Stromberg, B.,
Berge, N., Rappe, C., 1993. In¯uence of combustion
parameters on the formation of polychlorinated dibenzo-
p-dioxins, dibenzofurans, benzenes and biphenyls and
polyaromatic hydrocarbons in a pilot incinerator. Environ.
Sci. Technol. 27, 1602±1610.

Wikstrom, E., Marklund, S., 2000. Secondary formation of
chlorinated dipenzo-p-diokins, dibenzofurans, biphenyls,
benzenes and phenols during MSW combustion. Environ.
Sci. Technol. 34, 604±609.

Wikstrom, E., Andersson, P., Marklund, S., 1998b. Design of a
laboratory scale ¯uidized bed reactor. Rev. Sci. Instr. 69,
1850±1859.
Jackson, J.E., 1991. A user guide to principal components.
Wiley, ISBN-0-471-62267-2.


Wikstrom, E., Lofvenius, G., Rappe, C., Marklund, S., 1996.
In¯uence of level and form of chlorine on the formation of
chlorinated dioxins dibenzofurans and benzenes during
combustion of an arti®cial fuel in a laboratory reactor.
Environ. Sci. Technol. 30, 1637±1644.
Jolli€e, I.T., 1986. Principal component analysis. Springer,
ISBN-0-387-96269-7.
Kanters, M.J., VanNispen, R., Louw, R., Mulder, P., 1996.
Chlorine input and chlorophenol emission in the lab-scale
combustion of municipal solid waste. Environ. Sci. Technol.
30, 2121±2126.