Effects of Varying Combustion Conditions

on PCDD/F Formation

JOHANNA AURELL

Akademisk avhandling

som med vederbörligt tillstånd av rektorsämbetet vid Umeå Universitet för avläggande av Teknologie Doktorsexamen i Kemi med inriktning mot miljökemi vid Tekniska-Naturvetenskapliga fakulteten i Umeå, framläggs till offentlig granskning vid Kemiska institutionen, hörsal KB3A9, plan 3, i huset för Kemiskt Biologiskt Centrum (KBC-huset) måndagen den 22 september 2008, klockan 13.00.

Fakultetsopponent: Professor Sukh Sidhu, University of Dayton, Dayton, Ohio, USA

Kemiska institutionen UMEÅ UNIVERSITET

2008

Title: Effects of varying combustion conditions on PCDD/F formation

Author: Johanna Aurell

Organization: Chemistry Department Umeå University SE-901 87 Umeå Sweden

Abstract

Polychlorinated dibenzo-p-dioxins (PCDDs) and polychlorinated dibenzofurans (PCDFs) are by-products emitted from combustion sources such as municipal solid waste (MSW) incineration plants. These organic compounds are recognized as toxic, bioaccumulative and persistent in the environment. PCDD/Fs are removed from flue gases before released from MSW incineration. However, the PCDD/Fs are not destroyed but retained in the residues, thus in the environment. Understanding the pathways that lead to their formation is important in order to develop ways to suppress their formation and prevent their release into the environment. Suppressing the formation can also allow less expensive air pollution control system to be used, and/or the costs of thermally treating the residues to be reduced. The main objective of the studies underlying this thesis was to elucidate process, combustion and fuel parameters that substantially affect the emission levels and formation of PCDD/Fs in flue gases from MSW incineration. The experiments were conducted under controllable, realistic combustion conditions using a laboratory-scale reactor combusting artificial MSW.

The parameter found to most strongly reduce the PCDD/F emissions, was prolonging the flue gas residence time at a relatively high temperature (460°C). Increasing the sulfur dioxide (SO2) to hydrogen chloride (HCl) ratio to 1.6 in the flue gas was also found to reduce the PCDF levels, but not the PCDD levels. Fluctuations in the combustion process (carbon monoxide peaks), high chlorine levels in the waste (1.7%) and low temperatures in the secondary combustion zone (660 °C) all tended to increase the emission levels. The PCDD/PCDF ratio in the flue gas was found to depend on the chlorine level in the waste, fluctuations in the combustion process and the SO2:HCl ratio in the flue gas. The formation pathways were found to be affected by the quench time profiles in the post-combustion zone, fluctuations in the combustion process and addition of sulfur. In addition, increased levels of chlorine in the waste increased the chlorination degrees of both PCDDs and PCDFs. A tendency for increased SO2 levels in the flue gas to increase levels of polychlorinated dibenzothiophenes (sulfur analogues of PCDFs) was also detected, however the increases were much less significant than the reduction in PCDF levels.

Key words: dioxin, formation, PCDD, PCDF, MSW incineration, transient combustion, sulfur, memory effects, quench time, chlorine, combustion, polychlorinated dibenzothiophenes, NO, water

ISBN 978-91-7264-617-9 Language: English Document name: Doctoral Thesis Number of pages: 74 + 4 papers

•

Effects of Varying Combustion Conditions on PCDD/F Formation

JOHANNA AURELL

DOCTORAL THESIS

Department of Chemistry UMEÅ UNIVERSITY

UMEÅ, SWEDEN 2008

Department of Chemistry

UMEÅ UNIVERSITY

SE-901 97 Umeå

SWEDEN

Copyright © 2008 Johanna Aurell ISBN 978-91-7264-617-9 Printed in Sweden by Print & Media, Umeå 2008

•

ABSTRACT

Polychlorinated dibenzo-p-dioxins (PCDDs) and polychlorinated dibenzofurans (PCDFs) are by-products emitted from combustion sources such as municipal solid waste (MSW) incineration plants. These organic compounds are recognized as toxic, bioaccumulative and persistent in the environment. PCDD/Fs are removed from flue gases before released from MSW incineration. However, the PCDD/Fs are not destroyed but retained in the residues, thus in the environment. Understanding the pathways that lead to their formation is important in order to develop ways to suppress their formation and prevent their release into the environment. Suppressing the formation can also allow less expensive air pollution control system to be used, and/or the costs of thermally treating the residues to be reduced. The main objective of the studies underlying this thesis was to elucidate process, combustion and fuel parameters that substantially affect the emission levels and formation of PCDD/Fs in flue gases from MSW incineration. The experiments were conducted under controllable, realistic combustion conditions using a laboratory-scale reactor combusting artificial MSW.

The parameter found to most strongly reduce the PCDD/F emissions, was prolonging the flue gas residence time at a relatively high temperature (460 °C). Increasing the sulfur dioxide (SO2) to hydrogen chloride (HCl) ratio to 1.6 in the flue gas was also found to reduce the PCDF levels, but not the PCDD levels. Fluctuations in the combustion process (carbon monoxide peaks), high chlorine levels in the waste (1.7%) and low temperatures in the secondary combustion zone (660 °C) all tended to increase the emission levels. The PCDD/PCDF ratio in the flue gas was found to depend on the chlorine level in the waste, fluctuations in the combustion process and the SO2:HCl ratio in the flue gas. The formation pathways were found to be affected by the quench time profiles in the post-combustion zone, fluctuations in the combustion process and addition of sulfur. In addition, increased levels of chlorine in the waste increased the chlorination degrees of both PCDDs and PCDFs. A tendency for increased SO2 levels in the flue gas to increase levels of polychlorinated dibenzothiophenes (sulfur analogues of PCDFs) was also detected, however the increases were much less significant than the reduction in PCDF levels.

SAMMANFATTNING (SUMMARY IN SWEDISH)

Polyklorerade dibenso-p-dioxiner (PCDD) och polyklorerade dibensofuraner (PCDF), allmänt känt under samlingsnamnet ”Dioxiner”, är biprodukter som oavsiktligt uppkommer i förbränningsprocesser så som sopförbränning. Dessa organiska föroreningar är toxiska, långlivade i miljön och kan dessutom lagras i vävnader. Utsläpp av PCDD/F från sopförbränning förhindras genom effektiv rökgasreningsutrusning. PCDD/F förstörs dock inte utan är kvar i restprodukterna således även i miljön. PCDD/F kan brytas ned genom termisk behandling. En reducerad bildning av PCDD/F kan leda till lägre rökgasreningskostnader och/eller minskade kostnader för termisk behandling av restprodukterna. Det är därför viktigt att förstå hur PCDD/F bildas för att utveckla tekniker som har avsikt att förhindra utsläppen till miljön. Syftet med denna studie var att klargöra vilka förbränningsparametrar som väsentligt påverkar bildningen och emissionerna av PCDD/F från sopförbränning. Experimenten genomfördes i en fluidiserande bäddreaktor i pilotskala med en artificiell sopa som bränsle.

En förlängd uppehållstid vid 450/460°C var den parameter som effektivast reducerade PCDD/F emissionerna. En ökning av svaveldioxid halt in rökgaserna gav också en reduktion av PCDF koncentrationerna, dock ej för PCDD. Variationer in förbränningsprocessen (ökad CO koncentration i rökgasen), hög halt klor i bränslet (1.7%) och en låg temperatur i den sekundära förbränningszonen (660°C) var parametrar som påtagligt ökade PCDD/F koncentrationen i rökgasen. Kvoten PCDD/PCDF i rökgasen visade sig bero på klorhalten i bränslet, SO2:HCl kvoten i rökgasen och variationer i förbränningsprocessen. Bildningen av PCDD/F påverkades av temperatur profilen i avkylningszonen, variationer i förbränningsprocessen och SO2 halten i rökgasen. En ökad SO2 halt i rökgasen visade tendenser på en ökning i koncentration av polyklorerade dibensothiophener (svavel analoger till PCDF), men halterna var mycket lägre än reduktionen av PCDF koncentrationen.

•

LIST OF PAPERS

This thesis is based on the following papers, which are referred to in the text by their respective roman numerals.

I. Johanna Aurell and Stellan Marklund. Effects of varying combustion conditions on PCDD/F emissions and formation during MSW incineration Manuscript submitted for publication.

II. Johanna Aurell, Stina Jansson and Stellan Marklund. (2008). Effects of quench time profiles on PCDD/F formation in the post-combustion zone during MSW incineration. Accepted for publication in Environmental Engineering Science.

III. Johanna Aurell, Jerker Fick and Stellan Marklund. (2008). Effects of transient combustion conditions on the formation of PCDD/F, PCBz and PAH during MSW incineration. Accepted for publication in Environmental Engineering Science.

IV. Johanna Aurell, Jerker Fick, Peter Haglund and Stellan Marklund. Effects of sulfur on PCDD/F and PCDT formation under stable and transient combustion conditions during MSW incineration.

Manuscript. Papers II and III are presented with the kind permission of Mary Ann Liebert Publishers.

Contribution of the author of this thesis to the papers: The author contributed substantially to the planning of the experiments described in Papers II-IV, and moderately to those described in Paper I. The experimental work (including performance of the experiments, sampling, sample-clean up and analyses) was all done by the author, except Paper II where fifteen of eighteen runs were performed by the author. In addition, the results were evaluated and Papers I-IV were written by the author.

LIST OF ABBREVIATIONS AND DEFINITIONS

APC Air Pollution Control CEM Continuous Emission Monitoring CV Coefficient of Variance dg dry gas dw dry weight EPA Environmental Protection Agency FTIR Fourier Transform Infra-red GC-HRMS Gas Chromatography-High Resolution Mass Spectrometry GC-LRMS Gas Chromatography-Low Resolution Mass Spectrometry I-TEQ International TeCDD-equivalents MSD Mass Selective Detector MSW Municipal Solid Waste PAH Polyaromatic hydrocarbon PCBz Polychlorinated benzens PCDD Polychlorinated dibenzo-p-dioxin PCDF Polychlorinated dibenzofuran PCDT Polychlorinated dibenzothiophenes PCPh Polychlorinated phenol PIC Products of Incomplete Combustion POPs Persistent Organic Pollutants PUFP Polyurethane foam plug PVC Polyvinyl chloride TeCDD Tetra-chlorinated dibenzo-p-dioxin TEF Toxic Equivalent Factor Air factor Actual air supply/stoichiometric (theoretical) air needed

for complete combustion. Congeners Compounds whose structure, function or origin is similar. Homologue A group that is related to each other, i.e. PCDD has 8

homologues and PCDF has 8 homologues. Isomers Compounds with identical molecular formula but differing

in structural formula.

•

TABLE OF CONTENTS

1. Introduction ..................................................................................... 1

2. Background...................................................................................... 3

2.1 Chemical structures and properties ............................................. 3

2.2 Incineration systems and combustion processes.......................... 8

2.3 Formation of PCDD/Fs ............................................................. 11

2.4 Formation of PAH and PCBz..................................................... 18

3. Experimental section..................................................................... 19

3.1 Description of the experimental laboratory-scale fluidized-bed reactor ....................................................................................... 19

3.2 Fuel ............................................................................................ 20

3.3 Experiments................................................................................ 22

3.4 Flue gas sampling ...................................................................... 27

3.4 Extraction, sample clean up and analyses ................................. 29

4. Results and Discussion .................................................................. 33

4.1 Effects of changes in fuel composition and process parameters on combustion conditions .......................................................... 33

4.2 Effects of combustion conditions on PCDD/F emissions and formation pathways ................................................................... 35

5. Conclusions and Future work ...................................................... 61

6. Acknowledgment ........................................................................... 63

7. References ...................................................................................... 65

Appendix: I-TEQ values derived for samples obtained during the experimental runs in Papers I-IV.

•

1. INTRODUCTION

Persistent organic pollutants (POPs) such as polychlorinated dibenzo-p-dioxins (PCDDs), polychlorinated dibenzofurans (PCDFs), polychlorinated benzenes (PCBz) and polyaromatic hydrocarbons (PAHs) are by-products emitted from combustion sources such as municipal solid waste (MSW) incineration plants. There are substantial environmental concerns regarding these compounds due to their high toxicity and persistence in the environment. PCDDs and PCDFs (commonly known as dioxins) were detected in MSW ash in 1977 (1). Since then the formation of PCDD/Fs has been researched, but although great progress has been made, the formation pathways are still not fully understood. However, PCDD/Fs are known to form in the lower temperature region (400-250 °C) of the post-combustion zone in MSW incineration plants.

The aim for incineration of MSW is to reduce the amount of waste, at the same time it generates energy. Regulations (most importantly in the European Union, EU Directives) on the incineration of waste limit the concentration of PCDD/F in flue gases and the residues (2,3). To comply with these regulations, flue gases and residues with PCDD/F concentrations exceeding the limits have to be cleaned before they are released to the air or treated before being deposited in landfills, respectively. Flue gases from modern MSW incineration plants are cleaned of pollutants by efficient air pollution control (APC) systems. The PCDD/F concentrations in the flue gas are usually reduced by using activated carbon as an adsorbent in the APC system, which includes a highly efficient dust removal system. Thus, the PCDD/Fs are not destroyed, but are retained in the residues.

The PCDD/F concentrations in residues can be reduced by thermal treatment. An alternative is to prevent the PCDD/Fs from being formed by optimizing the performance of combustion plants (by adjusting process, combustion and/or fuel parameters), thus reducing the PCDD/F concentrations in the residues and flue gases. Hence, there is a need to understand the pathways that lead to their formation in order to develop ways to suppress their formation, and thus allow less expensive air

1

pollution control systems to be used, and/or the costs of thermally treating the residues to be reduced.

The main objective of the work outlined in this thesis was to investigate the effects of varying combustion parameters on the PCDD/F emissions and formation, in order to elucidate parameters that substantially contribute to the reduction or increase of the release of PCDD/Fs from MSW incineration. The study was primarily focused on the PCDD/Fs but the polychlorinated dibenzothiophenes (PCDTs), PCBzs and PAHs were also analyzed. A 5 kW solid fuel fed laboratory-scale fluidized-bed reactor was used and operated under stable but realistic combustion conditions.

The results of the work have been presented in four papers, which address the following specific issues: • The effects of various combustion parameters, such as chlorine,

sulfur, copper, water levels, and freeboard temperature on PCDD/F emissions and formation pathways (Paper I).

• The effects on PCDD/F emissions and their formation pathways of four different quench time profiles (temperature profiles) in the post-combustion zone (Paper II).

• The effects of transient combustion conditions on PCDD/F, PCBz and PAH emissions and PCDD/F formation pathways (Paper III).

• The effects of sulfur on PCDD/F and PCDT emissions and formation pathways under stable and transient combustion conditions (Paper IV).

2

•

2. BACKGROUND

2.1 Chemical structures and properties

PCDD/Fs, PCDTs and PCBz are chlorinated aromatic compounds, while PAHs are aromatic compounds consisting only of carbon and hydrogen. These compounds have high chemical stability, thus they are persistent in the environment. They can also accumulate in the human body due to their lipophilicity (high affinity for fat and low solubility in water). Polychlorinated phenols (PCPh) are also chlorinated aromatic compounds, although they have higher water solubility than the other mentioned compounds.

PCDD/Fs

PCDDs and PCDFs include 75 and 135 different congeners, respectively, ranging from mono- to octa-chlorinated compounds (Figure 1).

Figure 1. The general structures of PCDDs and PCDFs, and specific structures of 1,3,6,8-TeCDD and 2,3,4,6-TeCDF; x and y represent 0-4 chlorine atoms.

3

Of the 210 PCDD/F congeners, 17 have 2,3,7,8-chlorination patterns, which confer high toxicity (Table 1). The 2,3,7,8-TeCDD congener is the most toxic congener of the PCDD/Fs and by definition is assigned a toxic equivalent factor (TEF) of 1 (Table 2). The TeCDD-equivalent (TEQ) is obtained by multiplying the concentration of a dioxin congener by its TEF-value and summing the result for all congeners. In this way the toxicity of the dioxins present in a sample that contains many congeners can be expressed as a single value (I-TEQ). PCDD/Fs are reportedly carcinogenic, have adverse effects on the reproductive system and induce various other toxic responses including chloracne (4). The I-TEQ values derived for samples obtained during each of the experimental runs described in Papers I-IV are presented in Appendix 1.

Table 1. Toxic equivalent factors (TEFs) of PCDD/Fs for humans and mammals (5).

PCDDs TEF PCDFs TEF

2,3,7,8-TeCDD 1 2,3,7,8-TeCDF 0.1 1,2,3,7,8-PeCDD 0.5 1,2,3,7,8-PeCDF 0.05 1,2,3,4,7,8-HxCDD 0.1 2,3,4,7,8-PeCDF 0.5 1,2,3,7,8,9-HxCDD 0.1 1,2,3,4,7,8-HxCDF 0.1 1,2,3,6,7,8-HxCDD 0.1 1,2,3,7,8,9-HxCDF 0.1 1,2,3,4,6,7,8-HpCDD 0.01 1,2,3,6,7,8-HxCDF 0.1 OCDD 0.001 2,3,4,6,7,8-HxCDF 0.1 1,2,3,4,6,7,8-HpCDF 0.01 1,2,3,4,7,8,9-HpCDF 0.01 OCDF 0.001

4

•

PCDTs

PCDTs are the sulfur analogues of PCDFs, i.e. compound in which the oxygen atom is replaced by a sulfur atom, as shown in Figure 2. Like PCDFs, the PCDTs consist of mono- to octa- chlorinated compounds and include in total 135 congeners. The 2,3,7,8-TeCDT congener has been shown to have potency to induce AHH/EROD, but to a lesser extent than the 2,3,7,8-PCDDs, i.e. the 2,3,7,8-TeCDT congener has similar, but weaker, toxicity properties to 2,3,7,8-TeCDD (6). Data on PCDT emissions are presented in Paper IV.

Figure 2. The general structures of PCDTs, x and y represent 0-4 chlorine atoms.

PCBz

PCBz consist of a benzene ring with 1-6 chlorine atoms as shown in Figure 3, and include 12 congeners. Dichlorobenzenes are suspected to cause liver and kidney damage (7). Data on tri- to hexa-chlorinated benzenes are reported in Paper III.

Figure 3. General molecular structure of PCBz (left) and structure of hexachlorobenzene (HCB) (right), x represents 1-6 chlorine atoms.

5

PAH

PAHs are a large group of compounds, 16 of which are prioritized by the United States Environmental Protection Agency (US EPA) and are therefore frequently analyzed and referred to as the 16 EPA PAHs (shown in Figure 4). PAHs are prioritized compounds because some of them have been reported to have mutagenic and carcinogenic properties (8). Data on 15 of the 16 EPA PAHs (all except naphthalene) are reported in Paper III and IV.

Figure 4. Molecular structures of the 16 EPA PAHs.

6

•

PCPh

As shown in Figure 5, the PCPhs are structurally similar to PCBz, but with a hydroxyl group attached to the molecule. This makes the PCPhs more soluble in water than e.g. PCDD. The PCPh include 19 congeners, none of which were analyzed in the studies reported in Papers I-IV, although they are discussed in the papers.

Figure 5. The general structure of PCPh (left) and 2,4,6-trichlorophenol (right); x represent 1-5 chlorine atoms.

7

2.2 Incineration systems and combustion processes

Incineration systems

Several types of incineration systems are used for combusting MSW, but the two most common are grate and fluidized bed systems. Fluidized bed systems operate at lower combustion temperatures (800-850 °C) than grate systems (1000-1100 °C). Furthermore, the fuel is combusted with sand that is fluidized by air supplied from below, while in grates the fuel is fed on movable grates. Fluidized beds provide good mixing of the flue gases but are sensitive to changes in the fuel feed. Systems based on grates are the most common systems used for incinerating MSW.

Combustion processes

Combustion processes are complex and involve substantial amounts of chemical compounds and reactions. Complete combustion of MSW should in principal yield water (H2O), oxygen (O2), nitrogen (N2) and carbon dioxide (CO2). However, other combustion products are also formed, such as nitrogen oxides (NO, NO2 and N2O), sulfur oxides (SO2 and SO3), hydrogen chloride (HCl), and organic compounds such as PCDD/Fs. For complete combustion it is necessary to add sufficient air to the system, and to ensure that there are sufficiently high temperatures, residence times and mixing of the flue gases in the freeboard (secondary-combustion zone). Usually a temperature of 800 °C to 900 °C in the secondary-combustion zone, residence time of 2 seconds and an air factor* of 1.2 is necessary for complete combustion (9). Incomplete combustion i.e. fuel rich or oxygen deficient conditions, is indicated by elevated carbon monoxide (CO) emissions in the flue gas (since the carbon is not fully oxidized to CO2).

The fuel composition will affect the flue gas composition. For instance, increased sulphur levels in the fuel can result in increased HCl levels in the flue gas, since alkali metal salts (notably potassium chloride, KCl and

* Air factor (λ) = actual air supply/stoichiometric (theoretical) air needed for complete combustion.

8

•

sodium chloride, NaCl) are converted to metal sulphates (K2SO4, Na2SO4) according to R1.

HClSOMOSOOHMCl 2212 42222 +↔+++ (M = K or Na) [R1]

The effects of combustion parameters on NOx (NO and NO2) and N2O levels in the flue gas have been investigated by numerous research groups using fluidized bed incinerators. Due to the difference in combustion temperatures between fluidized bed and grate incineration systems the NOx and N2O levels vary between these systems. In fluidized bed systems most of the NO emissions originate from the fuel via oxidation of NH3. The main formation pathway of N2O occurs via hydrogen cyanide (HCN) as shown in R2. Thus, increasing the O2 level in the flue gas will increase both the NOx and N2O levels (10).

COONNONCOHNCOOHCN+→+

+→+

2 [R2]

In addition, SO2 in the flue gas can cause changes in the distributions of NOx and N2O emissions, which have been suggested to be due to reductions in the rate of HCN oxidation (11). Furthermore, calcium (Ca) has opposite effects on NOx and N2O distributions (10), possibly due to Ca forming the stable compound CaSO4 in the bed. CO has also been shown to reduce N2O emissions, which has been suggested to be due to CO and N2O reacting and forming CO2 and N2, however only in the presence of char (carbonaceous solid product of fuel devolatilization) (12). NO has also been found to have similar effects and participate in similar reactions, but in the absence of oxygen (13). In addition, increasing the secondary combustion zone temperatures while keeping bed temperatures constant results in reduced N2O levels, which has been suggested to be due to the decomposition of N2O to N2 (12,14). The main combustion parameters shown to influence the NOx and N2O levels in the flue gas when fluidized bed systems are used are listed in Table 2.

9

Table 2. Effects of combustion parameters in fluidized bed incinerators on NOx (NO and NO2) and N2O levels.

Increased parameter NOx N2O References

O2 + + (10,12,15)

SO2 - + (15)

Ca + - (10)

CO -a - (12,13)

Freeboard temperatureb No effect - (12,14) a. Without presence of oxygen b. Constant bed temperature

Sulfur dioxide has been shown to inhibit the conversion not only of HCN but also of combustible gases such as CH4 and CO (15). Even if complete combustion occurs the fly ash will always contain residual carbon or macromolecular carbon structures (Figure 6). The residual carbon may contain aliphatic and aromatic fractions and functional groups such as OH, O, COOH, S (16). The fly ash will also contain small organic compounds and metal ions.

Figure 6. Schematic illustration of different combustion stages, PIC = products of incomplete combustion. Modified illustration from Boman (17) and Steiglitz (16).

10

•

2.3 Formation of PCDD/Fs

The formation of PCDD/Fs can occur either in the gas-phase (homogenous reactions) or via gas phase-solid surfaces interaction (heterogeneous reactions). Gas-phase formation occurs at high temperatures, 500-700 °C. However, according to values calculated by Shaub and Tsang (18), gas-phase reactions only produce very low levels of PCDD/Fs. Furthermore, Vogg and Stieglitz (19) found very low concentrations of PCDD/Fs in the gas-phase at 600 °C compared to 300 °C when a stream of air was passed over a MSW ash. It is generally believed that the formation of PCDD/Fs occurs via heterogeneous reactions at temperatures between 450 °C and 250 °C, temperatures typically found in the post-combustion zone of MSW incineration plants. At the same time that PCDD/Fs are formed they are also degraded, since the degradation of PCDD/Fs in fly ash has been shown to occur in the same temperature window as their formation. However, the rate of their degradation increases with increasing temperature (20). Thus, the emission levels found in the flue gases are the net results of both formation and degradation reactions (21).

PCDD/F reaction pathways

The formation mechanisms of PCDD/Fs in combustion processes are very complex (especially in the combustion of highly complex matrices such as MSW) and still not completely understood. However, two major pathways of PCDD/F formation have been shown:

• De novo synthesis; formation from carbonaceous materials/residual carbon found in the ash matrix involving chlorination, with subsequent oxidation and release of PCDD/Fs (22-24), and/or formation from carbon structures such as PAHs (25).

• Formation from precursors such as PCPh and PCBz via condensation reactions (26-32). Such precursors are present in the gas phase and may be adsorbed on the fly ash surface. Formation of PCDDs via CPhs can occur in the gas phase (32) or in reactions catalyzed by fly ash/copper chloride (CuCl2) (30,31,33,34).

11

Both of these pathways occur mainly at temperatures between 400 °C and 250 °C (22,35). However, studies have indicated that PCDFs are formed at higher rates than PCDDs at temperatures >400 °C (36-38), and the major pathways of PCDD and PCDF formation have been suggested to differ. The PCDDs have been associated with condensation reactions and PCDF with de novo synthesis, although PCDDs have been shown to form from de novo created PCPhs (36). Thus, the PCDD/PCDF ratio can provide indications regarding whether de novo or precursor formation pathways predominate in a specific case. A PCDD/PCDF ratio <1 is associated with domination of de novo synthesis and >1 with the dominating formation from precursors (39). Generally, in MSW incineration the PCDF levels are higher than the PCDD levels (40).

PCDD/F homologue profiles and congener patterns

PCDD/F homologue profiles and congener patterns within each homologue group may be used to identify the major formation pathway in a particular case. PCDD/F homologue profiles show the relative proportions of mono- to octa-chlorinated compounds (homologue groups), and may be either described as a list of proportions or simply by a single number indicating the average degree of chlorination. The degree of chlorination ranges from 1 to 8, 8 indicating that the PCDD or PCDF homologue profiles consists solely of OCDD or OCDF. Differences in the PCDD/F homologue profiles have been linked to variations in both fuel composition and combustion conditions (37,41,42).

Congener patterns have been suggested to be a result of formation and degradation reactions (21). The PCDD/F congener patterns found during MSW incineration are similar to those observed during micro-scale laboratory de novo synthesis experiments, in which a stream of air is passed over fly ash. These patterns have been found to be relatively consistent, irrespective of variations in combustion parameters (23,37,43). However, PCDD/F congener patterns in flue gases originating from incineration using fluidized beds reportedly differ from those obtained using from a stoker furnace (44). Furthermore, recent full-scale investigations have found differences between PCDD/F congener patterns obtained during transient combustion conditions (incomplete combustion) and normal, stable combustion conditions (45,46).

12

•

Formation of PCDDs from the most thermodynamically stable PCPh precursors (2,4,6-TeCPh, 2,3,4,6-TeCPh and PeCPh) has been found to generate 1,3,6,8- and 1,3,7,9-TeCDD, 1,2,3,6,8-, 1,2,4,6,8-, 1,2,4,7,9- and 1,2,3,7,9-PeCDD, and 1,2,3,4,6,8-HxCDD congeners (26,30-32,44,47), i.e. these PCPhs give rise to a characteristic congener pattern. Several studies have found that PCDFs tend to be formed at considerably lower levels than PCDDs from PCPh precursors (30,35,47). Although, Ruy et al. (48) found that PCDFs formed twice as abundantly than PCDDs in experiments with 20 phenols. In addition, the 2,4,6,7/2,3,4,9/1,2,3,4/1,4,6,9-TeCDF congeners (which co-eluted from the column they used) strongly appeared to be formed from PCPhs. PCDD/Fs have also been found to form in de novo synthesis experiments with PAHs. The congeners 2,4,6,7-, 2,3,4,6- and 3,4,6,7-TeCDF and 2,3,4,6,7- and 2,3,4,6,8-PeCDF were reportedly the congeners most abundantly formed in de novo synthesis experiments from the PAH perylene (25,49).

Parameters affecting the PCDD/F formation

Chlorine

The chlorine level in the fuel has been found to have conflicting effects on PCDD/F formation: some studies have found correlations between the chlorine content of the fuel and PCDD/F levels in the flue gas (50,51), while others have not (41,52). Wikström et al. (41) investigated the effects of varying total chlorine (Cl) levels in the fuel in experiments with seven fuel mixes (from 0.12 to 1.72 % Cl in the fuels). They only observed a significant increase in PCDD/F formation rate was when the chlorine level in the fuel was 1.72%, and suggested there was a threshold content of ca. 1.7%, above which the PCDD/F formation rate significantly increases. The chlorine source, organic or inorganic such as PVC and NaCl/CaCl2, respectively, have showed no differences on the PCDD/F emissions (41,51,52).

The most common chlorine source in the flue gas is HCl, but it does not act directly as the chlorinating agent in PCDD/F formation. Instead, chlorine gas (Cl2) has been proposed and shown to be a chlorinating agent in PCDD/F formation (53,54). The Cl2 is produced through the Deacon

13

process, a two-step mechanism in which HCl is converted to Cl2 in reactions catalysed by CuCl2/cupper oxide (CuO) (R3-R5). The optimum temperature have shown to be ca. 400 °C down to no conversion of HCl to Cl2 at 300 °C (55).

222 21 ClCuOOCuCl +→+ [R3]

OHCuClHClCuO 222 +→+ [R4]

2222

212 ClOHOHCl CuCl +⎯⎯⎯ →←+ [R5]

Although the metal shown to have the strongest catalytic capacity in PCDD/F formation is copper (Cu), iron (Fe) also has catalytic effects, albeit weaker (55-57). Copper contents of just 0.006% and 0.007% in the fuel have been shown to be sufficient for catalyzing PCDD/F formation (58,59).

More recently, ash-bound chlorine has been suggested to be the most important chlorinating source in the de novo synthesis during full-scale combustion (60). Luijk et al. (30) suggested that CuCl2 could act both as a catalyst and chlorinating agent. In addition, Nonhebel (61) proposed that CuCl2 can act as a chlorinating agent by a ligand transfer oxidation. Furthermore, Taylor et al. (62) suggested that PCDD/F formation may occur from aliphatic hydrocarbons and with CuCl2 as chlorinating agent. This pathway was found to be active down to 150 °C, with maximum rates of formation at 300 °C. Taylor et al. (62) also suggested that the optimum PCDD/F formation temperature window (400-300 °C) was due to the reaction of CuO with HCl generating copper chlorinating agents and not the temperature required for the chlorination step.

Sulfur

Sulfur has been shown to reduce PCDD/F levels in the flue gas (63-68). For instance, injecting SO2 into the flue gas can substantially reduce their levels (68). Co-firing refuse-derived fuel (RDF) with 5% (w/w) high-sulfur coal have also shown to reduce the PCDD/F levels in the flue gas by more than 70% (64). In addition, Ogawa et al. (68) concluded that addition of coal as a sulfur source had greater effects than SO2 on PCDD/F

14

•

formation, even though injecting SO2 to a SO2:HCl ratio of 0.3 (w/w) reduced PCDD/F emissions by 70% in their experiments. However, there are also indications that excessive additions of sulfur to the fuel can diminish reductions in PCDD/F emission levels (63). Furthermore, a time delay of several hours has been observed in full scale investigations before additions of sulfur have resulted in detectable reductions in PCDD/F emissions (69). This was attributed to deposits of ash on the tube walls remaining from incineration prior to sulfur addition, implying that the metal chlorides in the fly ash promote the formation of PCDD/Fs. Sulfur has also been indicated to reduce PCDF formation more than PCDD formation (70).

Griffin (53) proposed that PCDD/F formation could be inhibited by SO2 transforming Cl2 to HCl in the presence of water (R5). Transformation of the Cu catalyst to copper sulphate (CuSO4), via R6 or R7 and R8, has also been suggested to inhibit PCDD/F formation (70,71).

HClSOOHClSO 23222 +→++ [R5]

422 21 CuSOOSOCuO →++ [R6]

)(2)()(2 2 sCuOClgClsCuO ↔+ [R7]

)(2)(2)()(221)(2 4222 gHClsCuSOgOHgSOOsCuOCl +→+++ [R8]

Reaction 5 was shown by Gullett et al. (71) to shift the optimum conversion temperature of the Deacon process (R3 and R4) from around 400 °C to 500 °C. Thus R5 did not decrease the conversion rate of R3 and R4, i.e. the production of Cl2 was not reduced. The CuSO4 has been suggested to reduce biaryl synthesis (71). The inhibitory effect of sulphur on PCDD/F formation has also been suggested to be due to formation of PCDTs and polychlorinated thianthrenes (PCTAs); sulfated analogues of PCDFs and PCDDs, respectively (71). PCDTs have been identified in fly ash from MSW incineration (72), but little is known about their formation, although they are suspected to have similar formation pathways to the PCDD/Fs (72).

15

Oxygen and water

Oxygen is essential in the formation of PCDD/Fs, since O2 is involved both in the Deacon process (R3) and the transformation of CuCl2 to CuSO4 (R6 and R8). Increases in oxygen (O2) levels (from 1% to 4% and 10%) have been found to increase PCDD/F formation (73). However, reductions in O2 levels during combustion could lead to increased CO concentrations (CO-peaks), due to incomplete combustion.

The effects of varying water levels on PCDD/F formation have been investigated with several different laboratory scales. Lenoir et al. (52) found that varying the water level (at 5%, 20% or 25%) had no significant effect on total rates of PCDD/F formation. But increasing the water levels increased the PCDD/PCDF ratio. Increases in the PCDD/PCDF ratio with increases in water vapour levels were also found in a study by Stieglitz et al. (74). However, Ruokojärvi et al. (75) observed that injecting water into the combustion zone had an inhibitory effect on PCDD/F formation during MSW incineration.

Temperatures

Increases in secondary-combustion temperatures (from 700 °C to 900 °C) have been shown to decrease PCDD/F emission levels, most strongly between 700 °C and 800 °C (76).

Temperatures and residence times in the post-combustion zone have shown to influence the PCDD/F formation. Fängmark et al. (77) found that (among the cases they tested) PCDD/F levels were highest with a temperature of 340 °C and residence time of 2.9 seconds, and lowest with a temperature of 260 °C and residence time of 0.9 seconds.

Transient combustion conditions

Numerous researchers have reported that incomplete combustion or transient combustion conditions can lead to increased PCDD/F emission levels (42,46,78-81). However, no correlations between CO levels and PCDD/F levels in the flu gas have been shown (81). Transient combustion conditions occur not only during malfunctions in the system, but also during start-ups and shut-downs of MSW incinerator plants. Recent studies of transient combustion conditions in full-scale plants have shown

16

•

that changes occur in PCDD/PCDF ratios and PCDD/F congener patterns during transient combustion conditions compared to normal combustion conditions (45,82,83). In addition, the PCDD/F homologue profiles have been shown to shift, towards higher proportions of lightly chlorinated DFs, under transient combustion conditions (42,45).

Transient conditions can also cause elevated PCDD/F levels for a time following disturbances, referred to in the literature as ‘memory effects’. Two different memory effects are discussed: de novo based memory effects and adsorptive memory effects (81). The former are due to the formation of deposits of carbon structures by de novo synthesis, while the latter are due to the slow migration of previously formed PCDD/Fs adsorbed to surfaces of tube walls. However, the increased PCDD/PCDF ratios and changes in PCDD/F congener patterns observed during and after transient combustion conditions, compared to normal combustion conditions, have recently prompted researchers to suggest that memory effects may increase rates of precursor-based formation (45).

In addition, contamination of walls by metals from fly ash has been shown to increase PCDD/F levels, and the levels have been shown to decline with time (84-86). Wall deposits have also shown to increase the PCDD/PCDF ratio (65).

17

2.4 Formation of PAH and PCBz

PAHs are formed due to incomplete combustion and occur at high temperature in the combustion zone (> 500 °C). Their formation reactions are complex and still not fully understood. However, formation of PAH can be divided into two steps: formation of the first aromatic ring (often benzene) and further growth. The major pathway leading to the formation of benzene has not yet been fully resolved. Furthermore, the benzene ring does not necessarily work as precursor, compounds as phenylacetylene and styrene can work as precursors as well (87). PAHs are also believed to be precursors to soot. PAHs with five or more benzene rings condense on particles, while PAHs with fewer than four benzene rings generally reside in the gas phase.

The PAH profile has been shown to depend on combustion conditions. PAH profiles during both normal and transient combustion conditions have been investigated in two studies, in both of which the most abundant of the 15 US EPA priority PAHs (excluding naphthalene) were found to be phenanthrene, pyrene, benzoa(a)anthracene and fluoranthene during normal combustion, and domination of phenanthrene, pyrene, fluorine and fluoranthene, during malfunctions/transient combustion conditions (78,88). The largest increases associated with transient combustion conditions were found for anthracene and chrysene. In another investigation the PAH profiles in flue gases from two incinerator plants under transient combustion conditions were found to differ somewhat, the four most abundant PAHs were phenanthrene, fluoranthene, pyrene and acenaphthylene in both cases, but the order of their abundance differed (49).

PCBz have been shown to be formed via de novo synthesis (36), and both from PCPh (89) and PAH (49). However, the congener patterns of the TriCBz and TeCBz were found to differ from those of PCPh and of de novo from carbon than PAHs. The most abundant TriCBz and TeCBz congeners formed de novo from carbon (36) and PCPh (90), were 1,2,4-TriCBz and 1,2,3,5/1,2,4,5-TeCBz, while 1,2,3-TriCBz and 1,2,3,4-TeCBz were formed most abundantly de novo from the PAHs pyrene, perylene and benzo[g,h,i]perylene (49).

18

•

3. EXPERIMENTAL SECTION

3.1 Description of the experimental laboratory-scale fluidized-bed reactor

The experiments reported in Papers I-IV were performed using the 5 kW laboratory-scale fluidized-bed reactor schematically illustrated in Figure 7. Briefly, the reactor consists of a bed section (primary-combustion zone), freeboard (secondary-combustion zone), a convector section (post-combustion zone) and an air pollution control system (APC). The bed and freeboard sections are constructed of a FeCrAl tube (operating temperatures up to 1400 °C). The duct connecting the freeboard to the convector is made of 253MA grade stainless steel (operating temperatures up to 1150 °C), and the tubes in the convector section are made of titanium-stabilized stainless steel suitable for operating temperatures <850 °C.

Figure 7. Schematic illustration of the laboratory-scale fluidized-bed reactor used in studies reported in Papers I-IV, not to scale.

19

This version of the laboratory-scale reactor was originally constructed in 1994 to simulate the behaviour of a full-scale waste incinerator, and was described in detail by Wikström et al. (91). Modifications were made to it during 2005, which were described in Paper III. Briefly: the fuel feeder was updated with a longer tube (containing a v-shaped chute) connecting the fuel box with the bed inlet to minimize the fluctuations of the fuel feeding; 12 N-type sheathed thermocouples were mounted directly into the flue gas duct; and the monitoring system was changed to the Windows-based program LabVIEW 7.1.

For a small scale reactor, such as this laboratory-scale reactor, external heating is required to reach freeboard and post-combustion zone temperatures similar to those of full-scale plants. For this purpose, the laboratory-scale reactor was constructed with seven external heaters (H1-H7, Figure 7), which enable both the freeboard and convector temperatures to be easily varied. Thus, flue gas samples could be collected at different temperatures and flue gas residence times in the convector section. Drops in temperature occurred at each of the 180° bends connecting the convector units since there was no external heating at these points.

3.2 Fuel

The fuel used (Papers I-IV) was an artificial MSW, with a composition based on typical, or ‘average’ Swedish MSW (92), shown in Table 3. The ingredients were mixed and the resulting mixture was compressed into pellets, with a diameter of 6 mm and length of 10-30 mm. The dry weight (dw), ash content and heating value of the fuel were 95%, 18% and 19 MJ/kg dw, respectively.

20

•

Table 3. Ingredients and composition of the artificial MSW.

Material Amount (%) Element Concentration

(% dw)

Paper confetti 36.0 Copper (Cu) 0.007 Wheat starch 15.5 Iron (Fe) 1.0 Sawdust 10.5 Sulfur (S) 0.07 Polyethylene 10.5 Chlorine (Cl) 0.7 Gelatine 7.5 Carbon (C) 45 Peat 5.0 Hydrogen (H) 6.3 Rapeseed oil 5.0 Nitrogen (N) 1.4 Silica gel (SiO2, ∅0.2-0.6 mm)

3.0 Oxygen (O) 31

Sand (∅150-250 μm) 2.5 Silicone (Si) 3.1 Kaolin (Al4OH8SiO10) 1.7 Aluminium (Al) 0.2 Iron powder (Fe) 1.2 Sodium (Na) 0.03 Polyvinyl chloride (PVC) 0.8 Potassium (K) 0.02

Calcium chloride (CaCl2×6H2O) 0.7 Calcium (Ca) 2.7

Copper acetate (CuO4C4H6)

0.02

Advantages of using an artificial MSW include its homogeneity, accurate knowledge of its composition and the ease of varying its composition. Two additional, smaller batches of fuel were made to evaluate the effects of changing its chlorine and copper levels (Paper I). The chlorine level was increased to 1.7% (compared to 0.7% in the starting fuel), by increasing both the PVC and calcium chloride content 2.5-fold. The copper level was increased two-fold relative to the starting fuel, by adding twice as much (0.04%) copper acetate to the fuel mix. However, subsequent analysis of the copper level in the fuel showed it to be 0.011% compared to 0.007% in the starting fuel (not twice the amount).

21

3.3 Experiments

Experimental runs

In the experiments one parameter (process or fuel composition variable) was varied at a time, thus resulting in varying combustion conditions. The reproducibility of the experimental set-up was evaluated by performing five experimental runs using the same target combustion parameters. These runs were referred to either as baseline runs (Papers I and II) or normal combustion conditions (Papers III and IV). The baseline runs were intended to establish similar operating conditions with respect to the following parameters: flue gas residence time and temperatures in the freeboard and convector sections; O2, H2O and SO2 levels in the flue gas; combustion efficiency with respect to the carbon monoxide (CO) level in the flue gas, and fuel load (Table 4). Samples were taken simultaneously at three sampling ports P3 (400 °C), P4 (300 °C) and P7 (200 °C) in three of the five baseline runs and at two sampling ports P4 (300 °C) and P7 (200 °C) in the other two.

Table 4. Target values for the baseline runs described in Paper I-IV.

Combustion parameter Target value Residence/quench time [seconds]

Bed [°C] 800-850 Freeboarda [°C] 800 5.1 P2b [°C] 640 0.1c

P3b [°C] 400 1.4c

P4b [°C] 300 2.3c

P7b [°C] 200 4.4c

H2O [vol%] 8-9 O2 [vol% dg] 10-11 SO2 [ppm dg] <5 CO [ppm dg] <10 Fuel load [kg/hour] 0.9 Air factor (λ) 1.9 a Freeboard temperature at T4 and T5. b Temperature at the sampling port in the convector section. c The residence/quench time in the post-combustion zone was defined as the time the flue gas took to travel from the beginning of the post-combustion zone to the sampling port, calculated from the elemental content of the fuel, air supply, local temperatures and volume of the laboratory-scale reactor.

22

•

In the study reported in Paper I selected process and fuel parameters were varied, as shown in Table 5. Flue gases were simultaneously sampled for PCDD/F at three P3 (400 °C), P4 (300 °C) and P7 (200 °C) or two P4 (300 °C) and P7 (200 °C) in the post-combustion zone. Results from samples collected in P4 (300 °C) were presented in Paper I. Table 5. Experimental runsa described in Paper I.

Description Fuel load [kg/h]

Free- board [°C]

Quench time [sec.]

Cl [%]

Cu [%]

H2O [v-%]

SO2:HCl

CO [ppm]

O2[v-%]

Baseline 0.9 800 2.3 0.7 0.007 7.9 0 <5 10.8

SO2:HCl 0.4 0.9 800 2.3 0.7 0.007 7.7 0.4 <5 10.0

Cu 0.011% 0.9 800 2.3 0.7 0.011 8.0 0 <5 11.1

H2O 21% 0.9 800 2.3 0.7 0.007 21 0 <5 10.9

Cl 1.7% 0.9 800 2.3 1.7 0.007 7.6 0 <5 11.2

Freeboard

950 °C

0.9 950 2.3 0.7 0.007 7.5 0 <5 10.9

Freeboard

660 °C

0.9 660 2.3 0.7 0.007 8.5 0 40 10.5

Fuel load

0.6 kg/h

0.6 800 2.3 0.7 0.007 7.0 0 <5 11.0

Quench time

1.6 sec.

0.9 800 1.6 0.7 0.007 7.5 0 <5 11.6

Transient,

2.5% O2

0.9 800 2.3 0.7 0.007 7.7 0 1300 2.5

Transient,

8.8% O2

0.9 800 2.3 0.7 0.007 9.5 0 1600 8.8

a. Quench time calculated from the start of the post-combustion zone (640 °C) to the sampling point (300 °C); Cl and Cu contents in the fuel; H2O, CO and O2 measured in the flue gas; SO2:HCl ratio calculated by mass.

In the study reported in Paper II the effects of varying quench time/temperature profiles in the post-combustion zone on PCDD/F emissions were investigated, by establishing the four profiles illustrated in Figure 8. Samples were taken simultaneously at three sampling ports P3 (400 °C), P4 (300 °C) and P7 (200 °C). In addition, PCDD/Fs in flue gases

23

at 640 °C were sampled in three separate experimental runs in order to estimate of the amount of PCDD/Fs entering the post-combustion zone.

Figure 8. Quench time profiles investigated in Paper II. Sampling was performed at sampling ports P3, P4 and P7. The steep lines after sampling points indicate the sharp heat losses at the ends of the convector tubes (no external heating).

Paper III describes an investigation of the effects of transient combustion conditions on PCDD/F emissions and formation. Flue gases were simultaneously sampled for PCDD/F, PCBz and PAH were performed at three sampling ports P3 (400 °C), P4 (300 °C) and P7 (200 °C) in the post-combustion zone. Two transient experimental runs were performed, in each case with the baseline quench time profile.

Paper IV reports an investigation of the effects of sulfurs on PCDD/F emissions and formation during normal combustion conditions, and both during and after transient combustion conditions. Flue gases were simultaneously sampled for PCDD/F, PCDT and PAH at three sampling ports P3 (400 °C), P4 (300 °C) and P7 (200 °C) in the post-combustion zone with the baseline quench time profile. The effects of varying the

24

•

sulfur level and SO2:HCl ratio in the flue gas were examined by performing experimental runs with mass ratios of 0, 0.1, 0.4 and 1.6.

Experimental procedure

The experimental runs were performed on separate days, i.e. only one experimental run per day. The sand (600 grams) used in the fluidized bed was a washed sea sand (Barskarpsand 15), which was sieved to a size range of 125-250 μm before use. Fresh sand was used for each experimental run. Before starting the solid fuel combustion the furnace had to be heated to a temperature of about 700 °C. This was accomplished by combusting propane (3.7 l/min) for approximately two hours. As shown in Figures 9 and 10, MSW was then combusted for four hours before starting to sample organic compounds in the flue gas. Samples were collected during 45-minute periods on most sampling occasions (Papers I and II). However, as shown in Figure 10, organic compounds were sampled for 20-30 minutes during and after transient combustion conditions (Papers III and IV). In one experimental run propane was combusted alone for 7 hours to determine background PCDD/F levels, as reported in Paper II.

Figure 9. Timelines for the experimental runs described in Papers I and II.

25

The air supply settings were 110 l/min of primary air and 30 l/min of secondary air in most of the experimental runs. The secondary air was preheated, using a heating-jacket, to about 250 °C before being injected to the bed. After each run the ashes in the bed, convector and cyclone were collected. Furthermore, after each experimental run the reactor was cleaned as thoroughly as possible, by cleaning the freeboard and convector with compressed air and vacuuming, and sweeping the freeboard. The reactor was checked for leaks before each experimental campaign (three campaigns), by adding nitrogen (N2) via injection port 1 (I1) and measuring the O2 level under negative pressure.

Sulfur was added to the combustion zone by injecting SO2 gas (pre-heated to 250 °C by a small heating jacket) using a flow meter (Brooks SHO-RATE model R-2-15-C) through injection port I1 (Figure 7). Water (H2O liquid) was added at a rate of 0.9 l/hour into the combustion zone (through I1) using a peristaltic pump (Alitea Speed Control). A connecting tube between the pump and the combustion zone was heated by conduction from the combustion zone and worked similarly to a carburettor in vaporising the liquid H2O.

In addition, in the experiment with a reduced fuel load (0.6 kg/h rather than 0.9 kg/h) the excess air ratio was kept the same as in the baseline runs by reducing the rate of air supply from 140 l/min to 96 l/min. The effects of transient combustion conditions were also examined in two experimental runs in the study described in Paper I: one in which transient conditions were produced by reducing the O2 level in the flue gas by exchanging the secondary air with N2 gas and reducing the rate of primary air supply from 110 l/min to 90 l/min; and one in which transient combustion conditions were produced by manually varying the fuel feed. The latter approach was also applied in experimental runs described in Papers III and IV, in which organic compounds were also sampled during stable combustion conditions before and after the transient combustion conditions (Figure 10), both with and without SO2 addition.

26

•

Figure 10. Timelines for the experimental runs described in Papers III and IV, M1, M2 = memory effects 1 or 2.

3.4 Flue gas sampling

Continuous emission measuring of H2O, CO2, SO2, NO2, CO, NO, HCl, NH3, N2O and CH4 was performed once every 30 seconds, on average, by a Gas Analyzer (ABB Bomem 9100) based on a Fourier-Transform Infra-red (FTIR) spectrometer. In addition, O2 was measured every second with a ScanTronic OC 2010 oxygen measurement probe with a zirconium dioxide (ZrO2) cell. These gases were measured in the first convector section, as shown in Figure 7.

The organic compounds in the flue gases were sampled isokinetically (i.e. at the same velocity as the flue gases in the duct) to ensure that the particle size distributions in the samples were the same as those in the flue gases. Furthermore, the organic compounds were sampled using the cooled probe polyurethane foam plug (PUFP) sampling technique, according to European standard method EN-1948:1 (93). As shown in Figure 11, the sample train consisted of two glass flasks; the first containing water to collect the condensate and most of the particles and the other containing ethylene glycol to collect the gases. Two PUFPs were attached to the top of the ethylene glycol flask, with a glass microfibre filter between them, to collect the remaining particles, organic compounds and aerosols. Before sampling, 13C12-labeled sampling standards were added to the water flask

27

in the sample train to check for losses of PCDD/Fs, PCBz and PAH during sampling, transport and storage.

To ensure that there were no leaks in the sample train, the filter holder was checked before sampling. In addition, the O2 level in the exhaust gas from the pump was measured during sampling and compared to the O2 level in the flue gas. A higher O2 level in the pump exhaust than in the flue gas would have indicated a leak. The pumps used to withdraw the flue gas were calibrated with a gas meter before and after each experimental campaign. The gas meter had itself been calibrated by the SP Technical Research Institute of Sweden.

Adsorbent (PUFP)

Ice

Convector

Cooled probe

Glass/titanium probe probe

Flue gas Water Water Ethylene glycol

ImpingerPump Filter

00629207

Figure 11. Schematic illustration of the sample train with the cooled probe attached to the convector (flue gas duct), not to scale.

The glass probe inside the cooled probe was made of pyrex glass, quartz glass or titanium. For sampling at temperatures of 100-400 °C (Papers I-IV), pyrex glass was used, quartz glass probes were used for sampling at 450/460 °C (Paper I) and a titanium probe was used for sampling at 640 °C (Paper I). When sampling in a full-scale plant, part of the cooled probe is placed inside the flue gas duct. For the small flue gas duct (convector) used in this study, only the nozzle of the probe was placed inside the duct (Figure 11), in the middle of the flue gas stream. In addition, the cooled probe was bolted to the convector, i.e. no leaks of

28

•

ambient air into the flue gas occurred during sampling. For rapid cooling of the hot flue gases sampled at 640 °C and 460 °C another cooled probe was placed outside the water-cooled probe (Figure 12). This outer probe was filled with a mixture of ice and salt (NaCl) to maintain a temperature of about -12 °C. The temperature outside the glass/titanium probes after cooling was around 20 °C during flue gas sampling. A new glass probe was used for each sampling event. The titanium probe was heated at 550 °C for 390 minutes and rinsed with toluene before reuse.

A

B

Figure 12. The original cooled probe (A) and the modified cooled probe (B) used for sampling at ≤400 °C and ≥460 °C, respectively.

3.4 Extraction, sample clean up and analyses

The extraction and clean up was performed according to standard methods described by Liljelind et al. (94), with modifications described in Paper III. The methods are briefly described below and a flow chart of the sample treatment steps in the laboratory is shown in Figure 13. Before extracting the samples, new 13C12-labeled internal standards (IS) were added to the water flask to compensate for losses during extraction and

29

clean-up. The ethylene glycol and water were combined and filtered twice. In the first filtration a glass fibre pre-filter and a membrane (0.45 μm) filter were used to adsorb particles. In the second filtration organic substances were collected using a solid-phase extraction disk (Envi-disk). The filters from the filtration steps together with the PUFPs and glass microfibre filter from the sample train were extracted overnight under reflux using Soxhlet-Dean-Stark equipment. Each sample extract was divided into three portions by volume: 50% for PCDD/F and PCDT clean-up, 25% for PCBz and PAH clean-up and 25% was retained.

Figure 13. Flow chart of the clean-up procedure in the laboratory.

30

•

For cleaning up PCDD/Fs and PCDTs, extracts were applied to a multisilica-column, eluted with hexane onto an alumina column and eluted with dichloromethane:cyclohexane. The samples were then applied to a AX21 carbon:celite-column, eluted first with n-hexane and then with a mixture of dichloromethane:n-hexane. Finally, the AX21 carbon:celite-column was eluted with toluene to collect the PCDD/Fs and PCDTs. A 13C12-labeled recovery standard (RS) and tetradecane (keeper) were added. The RS was used to calculate the losses during extraction and clean-up, and the keeper to ensure that PCDD/Fs were not lost during final evaporation. For PCBz and PAH clean-up the samples were applied to a deactivated silica column and eluted with cyclopentane. RS and toluene (keeper) were added to the sample. A field blank (consisting of sample equipment but no sample), a laboratory blank (subjected to all of the above laboratory steps, as shown in Figure 13) and a clean-up blank (subjected solely to the clean-up column steps) were also processed to ensure the quality of the sampling and laboratory procedures.

The PCDD/F extracts were analysed using an Agilent 6890 gas chromatograph (GC) equipped with a 60 m×0.25mm×0.32 μm J&W Scientific DB-5MS column and a SUPELCO SP 2331 60m×0.25mm×0.2 μm column and a Waters AutoSpec ULTIMA NT 2000D high resolution mass spectrometer (HRMS), while the PCDTs were analysed using the same GC-MS system, but with a 60m×0.25mm×0.25 μm J&W Scientific DB-5MS column. The PCBz and PAH extracts were analysed using a GC-low resolution mass spectrometer (LRMS) system (Agilent 5975 inert XL Mass Selective Detector) equipped with a 30m×0.25mm×0.25 μm J&W Scientific DB-5MS column. All derived concentrations were normalised (norm m3): to 1013 h Pa, 0 °C, dry gas (dg) and 11vol-% O2. The PCDD/F congeners were identified by published relative retention times for the similar SP 2331 column used in this study (95,96).

31

32

•

4. RESULTS AND DISCUSSION

4.1 Effects of changes in fuel composition and process parameters on combustion conditions

The combustion conditions were stable during the experiments, as demonstrated by the low CO concentrations (< 5 ppm, Papers I-IV). No elevated CO events (CO peaks) were observed within the experimental runs, except during the simulated transient combustion condition runs. The reproducibility/variability was also excellent, as shown by the low coefficients of variance (<14%) for the concentrations of each of the continuously sampled compounds (Table 6).

Table 6. Combustion conditions in baseline runs during flue gas sampling of organic compoundsa.

O2 H2O CO2 CO CH4 SO2 HCl NO NO2 N2O NH3

vol% vol% vol% ppm ppm ppm ppm ppm ppm ppm ppm

Average 11 7.9 9.2 <5 nd nd 210 470 32 52 nd

CV [%] 5 6 9 11 - - 14 9 5 11 -

a CV = coefficient of variance, nd = Not detected, detection limit 1 ppm.

The changes in fuel composition and process parameters altered the combustion conditions. Notably, the addition of SO2 to the combustion zone increased the HCl levels, probably due to the conversion of alkali metals (K, Na) to alkali sulfates according to R1. Addition of SO2 to 570 ppm in the flue gas also induced detectable levels (<5 ppm) of CH4, which were otherwise only observed during transient combustion conditions. The effects of the varied parameters on NO and N2O levels in the flue gas are summarized in Table 7. These results are in accordance with earlier studies, as discussed in Chapter 2.2 and Paper I.

33

Table 7. Effects of different parameters on the NOx, N2O and HCl levels in the flue gas, compared to baseline valuesa.

Increased/decreased parameter NOX N2O HCl

+ Sulfur - + +

+ Freeboard temperature - - 0

+ H2O 0 + 0

+ Cl + 0 +

+ Cu + - 0

- air ratio - - 0 a + = increased levels, - = decreased levels, 0 = levels in the flue gas unaffected, freeboard = secondary combustion zone

The pulsed fuel feed used during the transient combustion conditions or reduced O2 level in the flue gas, resulted in increased CO concentrations (CO peaks) every second to third minute. The CO-peaks varied in concentration (from ca 2000 to 8000 ppm) and with the combustion time amongst the runs. The average CO concentrations during sampling in each of the runs varied between 1300 ppm and 2700 ppm.

34

•

4.2 Effects of combustion conditions on PCDD/F emissions and formation pathways

Baseline profile/normal combustion conditions

The PCDD/F levels in the flue gas samples collected at 640 °C showed no detectable levels*. In addition, as shown in Figure 14, the baseline profile resulted in similar PCDD/F concentrations at each of the three sampling points/temperatures in the post-combustion zone. Thus, the decrease in post-combustion temperature and increases in flue gas residence time did not significantly affect the PCDD/F net formation. Furthermore, the lack of detectable PCDD/Fs at the 640 °C sampling point indicate that most of the net PCDD/F formation occurred between P2 (640 °C) and P3 (400 °C in this quench time profile).

Figure 14. PCDD and PCDF levels (cumulative concentrations) in the flue gas at the three sampling points: P3 (400 °C), P4 (300 °C) and P7 (200 °C). Error bars denote 1 standard deviation.

* PCDD/Fs were detected using the regular cooled probe, but these appeared to arise from excessively slow cooling, and rapid cooling of the hot flue gases seems to be important to avoid such artefacts.

35

The low CVs (< 22%) for the PCDD and PCDF concentrations at each of the sampling points indicate that the reproducibility of the conditions provided by the experimental set-up was excellent. In addition, reducing the fuel load (from 0.9 to 0.6 kg) resulted in a substantial reduction (60%) in PCDD/F yields, indicating that the experimental set-up responded sensitively to changes. The higher concentrations of PCDF compared to PCDD resulted in a PC4-8DD/PC4-8DF ratio (henceforth PCDD/PCDF ratio) of 0.11 ±0.02 (2SD) at all sampling points, indicating that de novo synthesis was the predominant formation pathway. As shown in Figure 15 the PCDD/F homologue profiles were dominated by the tetra- and hexa-chlorinated DDs and the mono- to tetra-chlorinated DFs. These homologue profiles are consistent with profiles observed in full-scale incinerators (97).

Figure 15. PCDD (A) and PCDF (B) homologue profiles for the baseline profile (cumulative concentrations) at the three sampling points: P3 (400 °C), P4 (300 °C) and P7 (200 °C). Error bars denote 1 standard deviation.

36

•

The TeCDD/F-HxCDD/F isomer pattern corresponded well to patterns observed in investigations of de novo formation of PCDD/F from carbon sources on fly ash (43). The congeners that were formed most abundantly in the baseline profile runs are shown in Table 8. Some congeners could not be separated with the column used in this study, and co-eluting isomers are indicated by backslashes (e.g. 2,3/2,8-DiCDD).

Table 8. The most abundant PCDD/F isomers in each homologue group in the baseline profile runs during normal combustion conditionsa.

Homologue PCDDs PCDFs

Mono 2 2, 3

Di [23/28] 24

Tri [237/146], 138 [234/238]

Tetra 1368, 1379, 1378, [1369/1247/1248]

[2468/1238/1467/1236], [1378/1379]

Penta [12468/12479] [12368/13478], [12348/12378], 23468

Hexa [123679/123689] [134678/134679]

Hepta 1234678, 1234679 1234678 a [/]: these isomers co-eluted from the column used. Equal amounts of two-four isomers occurred in some homologue groups.

Quench time profiles in the post-combustion zone

As shown in Figure 8, the differences in residence times between the quench time profiles in the post-combustion zone were quite slight. The PCDD/F yields with the quench time profile in which residence times of the flue gases to the temperatures 300 °C and 200 °C were approximately a second shorter (the “low-temperature profile”) than in the baseline profile were in the same range as those found in the baseline runs. The PCDD/F homologue profiles and congener patterns were also similar to those found in the baseline runs, indicating that the PCDD/Fs were rapidly

37

formed. However, the two temperature profiles with prolonged residence times at 450/460 °C (the “high-initial temperature” and “high-temperature” profiles) resulted in significantly lower PCDD/F yields, even at lower temperatures downstream of the post-combustion zone, compared to the baseline yields (Figure 16).

Figure 16. PCDD (A) and PCDF (B) levels in the flue gas in the four investigated quench time profiles (cumulative concentrations at the respective sampling points). Error bars denote 1 standard deviation.

38

•

The lower PCDD/F levels and PCDD/PCDF ratios (0.07) found at the 450/460 °C sampling point in the high temperature profiles than those found in the baseline runs (0.11±0.02 (2SD)) are consistent with the previously reported formation maxima of 400-500 °C for PCDFs and 300-400 °C for PCDDs (37). According to recent thermochemical equilibrium calculations (98), the temperature interval 500-450 °C is also a key transition zone for copper speciation, in which gaseous copper species solidify. Thus, the prolonged flue gas residence time at 450/460 °C may prevent or delay the conversion of gas-phase copper to CuCl2 (s), which can act both as a chlorinating agent and as a catalyst in the Deacon process (R3 and R4), reducing the amount of the chlorinating agent Cl2.

As shown in Figure 17, the PCDD and PCDF homologue profiles at 450/460 °C were dominated by the mono- to penta-chlorinated DDs and mono- to di-chlorinated DFs. The higher proportions of lightly chlorinated DD/DFs in the homologue profiles compared to those observed in baseline profiles may have been due either to weaker promotion of chlorination or enhanced promotion of dechlorination. However, since no notable differences in the congener patterns were detected between high-initial/high-temperature and baseline profiles, substantial degradation of PCDD/Fs is unlikely to have occurred. The differences in quench time from 450/460 °C to 200/260 °C resulted in different PCDD/F homologue profiles and congener patterns in these two profiles. The PCDD homologue profiles in the high-initial quench time profile at 300 °C and 200 °C were shifted to profiles similar to those found in the baseline runs, while no changes were found in the PCDF homologue profiles. This was also the case at 360 °C in the high-temperature profile.

39

Figure 17. PCDD and PCDF homologue profiles in the quench time profiles: high-initial temperature (A) and high-temperature (B) (cumulative concentrations at the respective sampling points). Error bars denote 1 standard deviation.

The PCDD yields increased between the sampling points P4 (360 °C) and P7 (260 °C) in the high-temperature profile. However, the PCDD yields were found to decline in three successive replicate runs: from 4.0, to 2.7 and finally to 1.6 ng/m3. This implied that the observed increases between P4 (360 °C) and P7 (260 °C) may have been related to deposits on the convector walls formed during previous experimental runs with disturbances in the combustion process resulting in soot formation. Even though the post-combustion zone was vacuumed and cleaned with compressed air after each experimental runs, a thin covering of fine residue particles always remained on the convector walls. Thus, the decline in PCDD/F levels between successive runs may have been due to consumption of carbon-based structures and/or metals and metal chlorides with time. This hypothesis is in agreement with a study by Lee et al. (85), who found that concentrations of PCDD/Fs, formed due to soot and copper deposits in a quartz reactor, declined with time. Moreover, elevated

40

•

PCDD/F concentrations in quartz reactors have been found to be due to wall contamination by metals from fly ash (84,86). The increased PCDD/PCDF ratio at P7 (260 °C) – 0.20±0.03 compared to 0.11±0.02 (2SD) in the baseline runs – is consistent with previous findings indicating that wall deposits influence PCDD formation more than PCDF formation (65).

In addition, the PCDD/F homologue profiles at 260 °C shifted towards higher proportions of the more highly chlorinated homologue groups (Figure 17). Furthermore, there were changes in congener patterns, presumably due to changes in their routes of formation, conceivably including subsequent chlorination of PCDFs. Taylor et al. (62) suggested that chlorination may occur at lower temperatures if CuCl2 is available, which may explain the chlorination of PCDFs in this temperature window. The PCDD congener patterns in the high-temperature profile changed at 360 °C and 260 °C to patterns characteristic of PCDD formation from PCPhs, implying that chlorophenol condensation reactions were promoted by these temperatures and residence times. The temperatures in the last sections of the post-combustion zone are not normally as high as those during the high-temperature runs, suggesting that when temperatures were raised (as in the high-temperature profile) catalytic sites required for chlorophenol condensation may be activated. The observed differences in PCDF homologue profiles between the high-temperature profile and the high initial-temperature profile may be that the Deacon process (R4) is capable of producing CuCl2 at 360 °C, but not at 300 °C (55). In addition, in de novo synthesis experiments Luijk et al. (30) found higher PCDD/PCDF ratios and a PCDD congener pattern characteristic of formation from PCPhs when levels of CuCl2 were relatively low, which is a possible explanation of the findings in this study.

Secondary combustion zone temperatures

Raising the temperature of the secondary combustion zone from 800 °C to 950 °C did not cause substantial changes in PCDD/F yields compared to those observed in the baseline runs. However, decreasing the temperature in the freeboard to 660 °C resulted in a constant CO concentration (40 ppm) and a significant increase in PCDD/F emission levels (three-fold) compared to baseline runs. However, the only changes in PCDD/F

41

homologue profiles and congener patterns compared to the baseline runs were found in the PCDF homologue profile, due to the increased secondary combustion zone temperature (950 °C instead of 800 °C). As shown in Figure 18, the PCDF homologue profile showed an increase in the degree of chlorination, possibly induced by reductions in the amount of backbone structures, due to enhanced oxidation of the hydrocarbons. These findings are indicative of increased chlorination of PCDFs. No differences were found in the PCDD/F levels, homologue profiles or congener patterns in the flue gas samples collected at sampling points P4 (300 °C) and P7 (200 °C). The similarity of the PCDD/F levels in the flue gas with temperatures of 800 °C and 950 °C in the secondary combustion zone may have been due to the relatively long residence time at 800 °C (ca. 5 seconds compared to the 2 seconds generally recommended).

Figure 18. PCDD (A) and PCDF (B) homologue profiles at sampling point P4 (300 °C) in the post-combustion zone observed with three different secondary combustion zone temperatures: 660 °C (A), 800 °C (B) and 950 °C (C). Error bars denote 1 standard deviation.

42

•

Chlorine

Increasing the chlorine content of the fuel from 0.7% to 1.7% resulted in a relatively large increase in PCDD/F emissions (20-fold) compared to baseline levels. Furthermore, it induced changes in the PCDD/F homologue profiles, increasing the proportions of highly chlorinated PCDD/Fs (Figure 19). The PCDF homologue profile indicated that the increase in chlorine levels promoted the chlorination of lightly or non-chlorinated precursors or carbon structures (de novo synthesis). In addition, no changes were found in the congener patterns compared to the baseline profiles. Although, the increased chlorination activity had a stronger effect on the PCDD emission levels than the PCDFs, as shown by the PCDD/PCDF ratio increasing to 0.31 from 0.11 ±0.02 (2 SD) in the baseline runs. The increased chlorination activity may have been due to increased levels of ash-bound chlorine (metal chlorides) and increased Cl2 concentrations in the flue gas, which have been shown to increase the PCDD/F emissions and degree of chlorination (60,70).

Figure 19. PCDD (A) and PCDF (B) homologue profiles at the sampling points P3 (400 °C), P4 (300 °C) and P7 (200 °C) with 1.7% Cl in the fuel.

43

Copper and water

Increasing the Cu content of the fuel from 0.007% to 0.011% or the H2O level in the flue gas from 8% to 21% had no significant effects on the PCDD/F emission levels compared to baseline levels. Furthermore, no changes in the PCDD/F homologue profiles or congener patterns compared to those observed in baseline runs were found. It is possible that greater differences in Cu contents of the fuel than those used in this study would be needed to affect the PCDD/F formation/degradation pathways and emission levels significantly. Increasing the H2O content resulted in a slight increase in the PCDD/PCDF ratio, from 0.11 ±0.02 (2 SD) in baseline runs to 0.16, which is consistent with results by Lenoir et al. (52). Note, since the bed temperature can be adjusted with external heating, the bed temperatures in runs with addition of water were similar to those in the baseline runs.

Sulfur

Reductions (up to 65%) in both PCDD and PCDF concentrations at the first sampling point, P3 (400 °C), with SO2:HCl ratios of 0.4 and 1.6 were found compared to those in baseline runs with no added SO2 (Figure 20). However, at the last sampling point (P7, 200 °C) the only significant reductions observed were in PCDF concentrations with an SO2:HCl ratio of 1.6 in the flue gas. This resulted in a increase in the PCDD/PCDF ratio with decreased temperature in the post-combustion zone and increases in the SO2:HCl ratio in the flue gas. The PCDD/F homologue profiles and the congener patterns were found to differ between the experiments with SO2:HCl ratios of ≤ 0.1 and SO2:HCl ratios ≥ 0.4 in the flue gas, implying that adding sulfur to the flue gas causes changes in the formation reactions, particularly for the PCDDs. The increased HCl levels induced by the addition of SO2 gas to the combustion zone (320 ppm and 370 ppm with the highest and second highest additions, respectively, compared to 210 ppm in the baseline runs) should theoretically increase Cl2 levels, via the Deacon reactions (R3 and R4) since CuSO4 does not affect Cl2 production (71). Thus, these findings suggest that Cl2 is not the major chlorinating agent in PCDD/F formation, probably because the sulfur reduces the chlorinating agent CuCl2 to CuSO4.

44

•

Figure 20. PCDD (A), PCDF (B) and PCDT (C) levels at the three sampling points – P3 (400 °C), P4 (300 °C) and P7 (200 °C) – with SO2:HCl ratios of 0 (baseline), 0.1, 0.4 and 1.6 in the flue gas. Note, data for the sample collected at 200 °C with an SO2:HCl ratio 0.1 are missing due to losses during clean-up. Error bars denote 2 standard deviations.

As shown in Figure 21, the PCDD homologue profiles at 400 °C with SO2:HCl ratios of 0.4 and 1.6 had lower proportions of highly chlorinated DDs compared to those observed in baseline runs. However, the proportions of highly chlorinated DDs were found to increase with decrease in temperature and increases in residence time in the post-combustion zone, and with SO2:HCl ratios of 0.4 and 1.6 in the flue gas. These changes were found to be largely due to increases in levels of the 1,3,6,8- and 1,3,7,9-TeCDD, 1,2,4,6,8/1,2,4,7,9- and 1,2,3,6,8-PeCDD and 1,2,3,4,6,8-HxCDD congeners; changes characteristically associated with PCDD formation from PCPh. The results imply that the conversion of CuCl2 (and/or other metal chlorides) to CuSO4 (and/or other metal sulfates) does not affect the catalysis of condensation reactions. In addition, Luijk et al. (30) found increases in PCDD/PCDF ratios and PCDD congener patterns characteristic of formation from PCPh when using low levels of CuCl2 in de novo experiments. Furthermore, Hell et al. (36) found increases in these PCDD congeners from de novo created

45

PCPhs with reductions in temperature, implying that the changes observed in the PCDD congener patterns in the present study were due to their formation from carbon residuals (de novo synthesis) in the fly ash.

Only minor changes (including slight reductions in the degree of chlorination) were found in the PCDF homologue profiles in the flue gas with SO2:HCl ratios of 0.4 and 1.6 in the flue gas compared to baseline runs (Figure 21). This was possibly due to inhibition of the chlorination of carbon residuals (de novo synthesis) or of lightly or non-chlorinated precursors. The only changes found in the PCDF congener patterns were increases in the isomer fraction of 1,2,3,4/2,3,4,9-TeCDF congeners at the last sampling point (200 °C) in the experiment with the SO2:HCl ratio of 1.6 in the flue gas, to 11% from 4 ±3% (2 SD) in the baseline runs. These congeners have also been shown to be formed from PCPhs (48). In addition, the isomer fractions of the 1,2,3,4/2,3,4,9-TeCDF congeners were larger with low levels of CuCl2 than with high CuCl2 levels in the study by Luijk et al. (30). The increases in 1,2,3,4/2,3,4,9-TeCDF congeners and PCDD levels at 200 °C with the SO2:HCl ratio of 1.6 in the flue gas, may also be partly due to increased rates of de novo formation of PCPhs induced by increased amounts of carbon residuals, as indicated by the inhibited oxidation of CH4. It should be noted that since the increased PCDD/F congeners were not 2,3,7,8-chlorinated the PCDD/F TEQ values showed up to 60% reductions in all samples collected along the post-combustion zone with SO2:HCl ratios of 0.4 and 1.6 compared to baseline runs (SO2:HCl ratio, 0) (see Appendix for TEQ values).

46

•

Figure 21. PCDD and PCDF homologue profiles at three sampling points – P3 (400 °C), P4 (300 °C) and P7 (200 °C) – with SO2:HCl ratios of 0 (A, baseline), 0.1 (B), 0.4 (C) and 1.6 (D) in the flue gas. Note, data for the sample collected at 200 °C with an SO2:HCl ratio 0.1 are missing due to losses during clean-up. Error bars denote 1 standard deviation. As shown in Figure 20, the PCDT levels obtained with the SO2:HCl ratio of 0 in the flue gas were similar at all three sampling points/temperatures, so most of the net formation occurred prior to the first sampling point (400 °C), as for the PCDD/Fs. The PCDT concentrations significantly increased with the SO2:HCl flue gas ratio of 0.1, compared to those observed in the baseline runs (Figure 22). This increase in PCDTs may have been due to increased sulfonation of the carbon residuals or precursors. This possibility is supported by the PCDT homologue profiles, which indicate that the degree of chlorination was higher with SO2:HCl ratios ≥0.1 than in the baseline runs (Figure 22). The PCDTs occurred in ten-fold lower concentrations than the PCDFs, implying that the sulfur analogues are not formed instead of the PCDFs. Furthermore, the PCDT homologue profiles suggest that the formation pathways of the PCDTs are

47

similar to those of the PCDFs, although the increased PCDT levels at 200 °C with the SO2:HCl of ratio of 1.6 in the flue gas suggest that they may also have a similar pathway to the PCDDs.

Figure 22. PCDT homologue profiles at the three sampling points – P3 (400 °C), P4 (300 °C) and P7 (200 °C) – with SO2:HCl ratios of 0 (baseline, A), 0.1 (B), 0.4 (C) and 1.6 (D) in the flue gas. Note, data for the sample collected at 200 °C with an SO2:HCl ratio of 0.1 are missing due to losses during clean-up. Error bars denote 1 standard deviation.

48

•

The effects of varying chlorine levels in the fuel or use of coal (as a sulfur source) on the PCDD/F emission levels and formation were not investigated in this study. However, the observed effects of increases in chlorine levels in the fuel and increases in SO2 levels in the flue gas on HCl levels imply that the correlations between PCDD/F emissions and chlorine levels are dependent on the levels of alkali metals (K and Na), sulfur and kaolin in the fuel. Kaolin (which consists of aluminum silicates, Al4OH8SiO10) is known to remove alkali from gases at high temperatures (bed temperatures), thereby forming HCl. In addition, Dayton et al. (99) suggested that observed reductions of levels of alkali metals in the gas phase when coal and biomass were co-fired was due to the aluminum silicates in the coal. Thus, the presence of clay minerals, such as aluminum silicates, in coal may account for its stronger inhibitory effects than SO2 on the formation of PCCD/Fs.

Transient combustion conditions

The PCDD/F concentrations increased substantially during and after the transient combustion conditions compared to those found during normal combustion conditions (Figure 23). Furthermore, the PCDD/F emission levels found during transient combustion conditions indicated that most of the formation occurred prior to the 400 °C sampling point in the post-combustion zone. However, during the memory effects periods the PCDD/F concentrations (especially the PCDD concentrations) increased with reductions in temperature in the post-combustion zone.

In addition, the SO2:HCl ratio in the flue gas was found to affect the relative proportions of PCDDs and PCDFs formed, as shown by the increases in PCDD/PCDF ratios with increases in SO2:HCl ratios in the flue gas (Table 9). Thus, transient combustion conditions promoted PCDF formation more than PCDD formation. However, while increases in the SO2:HCl ratio in the flue gas inhibited PCDF formation they enhanced PCDD formation, implying that different pathways are involved in PCDD and PCDF formation. In addition, the results suggest that these pathways compete.

49

0

200

400

600

800

1000

SO2:HCl 0 SO2:HCl 0 SO2:HCl 0.1 SO2:HCl 0.4 SO2:HCl 1.6

PCD

D [n

g/m

3 (nor

m)]

A. PCDD

0

4000

8000

12000

16000

20000

SO2:HCl 0 SO2:HCl 0 SO2:HCl 0.1 SO2:HCl 0.4 SO2:HCl 1.6

PCD

F [n

g/m

3 (nor

m)]

Transient combustion P3 (400 °C)Transient combustion P4 (300 °C)Transient combustion P7 (200 °C)M1 P3 (400 °C)M1 P4 (300 °C)M1 P7 (200 °C)M2 P3 (400 °C)M2 P4 (300 °C)M2 P7 (200 °C)

B. PCDF

Figure 23. PCDD (A) and PCDF (B) levels during and after transient combustion conditions in experiments with SO2:HCl ratios of 0, 0.1, 0.4 and 1.6 in the flue gas.

50

•

Table 9. PC4-8DD/PC4-8DF ratios in the flue gas samples collected at 400, 300 and 200 °C during the different combustion periodsa.

Temp. [°C] Normalb Transient Memory

effects 1 Memory effects 2

400 0.12 0.05 0.09 0.11 SO2:HCl 0 300 0.11 0.06d 0.13d 0.14d

200 0.11 0.06d 0.13d 0.14d

400 0.10 0.16 0.28 0.14 SO2:HCl 0.1 300 0.10 0.21 0.53 0.67 200 -c 0.20 0.51 0.72 400 0.11 0.29 0.32 0.14 SO2:HCl 0.4 300 0.14 0.37 0.83 1.00 200 0.20 0.41 0.87 1.12 400 0.17 0.52 0.42 0.21 SO2:HCl 1.6 300 0.24 0.68 0.90 1.13 200 0.33 0.77 0.98 1.18 a PCDD/PCDF in mass ratio. b Mean values: of five runs of SO2:HCl ratio 0 and two runs of SO2:HCl ratio 0.4. c Excluded due to losses during sample clean up. d Mean value of two runs.

Under the transient combustion conditions the PCDD homologue profiles were not significantly different from those observed in normal combustion conditions. In contrast, the PCDF homologue profiles changed, becoming dominated by the MoCDF homologues, levels of which increased with increases in the SO2:HCl ratio in the flue gas (Figure 24). These findings suggest that sulfur inhibits the chlorination of carbon (de novo synthesis) or lightly- or non-chlorinated precursors.

51

Figure 24. PCDD and PCDF homologue profiles during transient combustion conditions in runs with SO2:HCl ratios of 0 (baseline, A), 0.1 (B), 0.4 (C) and 1.6 (D) in the flue gas.

52

•

The transient combustion conditions induced changes in the PCDD/F congener patterns, compared to those observed in normal combustion conditions. The congener patterns were changed to patterns characteristic of PCDD/F formation from PCPhs, i.e. there were increases in the 1,3,6,8- and 1,3,7,9-TeCDD, 1,2,4,6,8/1,2,4,7,9- and 1,2,3,6,8-PeCDD and 1,2,3,4,6,8-HxCDD congeners. Changes in the PCDF congener patterns indicated increased formation via PAHs and PCPhs, i.e. increases in the 1,2,3,4/2,3,4,9-, 2,4,6,7-, 2,3,4,6- and 3,4,6,7-TeCDF, and 2,3,4,6,8- and 2,3,4,6,7-PeCDF congeners. Thus, the PCDD/F congeners found to be most strongly affected by the transient combustion conditions indicated that the PCDFs and PCDDs were formed by different reaction pathways. In addition, changes in the PCDD congener patterns during transient combustion were found to depend on the O2 level in the flue gas, i.e. reducing the O2 level in the flue gas (from 8.8 to 2.5%) did not induced changes in the PCDD congener pattern compared to those observed in normal combustion conditions. A possible explanation for this is that reductions in levels of oxygen molecules in the carbon residuals (macromolecule) may reduce the active sites required for formation of PCPhs.

These changes in PCDD/F congener patterns were independent of the SO2:HCl ratio in the flue gas and temperature in the post-combustion zone during transient combustion conditions. However, during the memory effect periods the isomer fraction of these congeners increased with decrease in the temperature in the post-combustion zone and the SO2:HCl ratio in the flue gas. For instance, the isomer fraction of 1,3,6,8-TeCDD increased from 18% to 48% with the SO2:HCl ratio of 1.6, and from 15% to 28% with the SO2:HCl ratio of 0, at the sampling points with temperatures of 400 °C and 200 °C, respectively (Figure 25). In addition, the isomer fractions of the TeCDF and PeCDF congeners most affected by the transient conditions were 40-50% lower with an addition of SO2 to the flue gas than without addition of SO2 (Figure 26). This again is indicative of PCDD formation from PCPhs created de novo and that the CuCl2 levels (or other metal chlorides) in the fly ash affects the relative levels of PCDD and PCDF emissions.

53

Figure 25. TeCDD isomer patterns with SO2:HCl ratios of 0 and 1.6 in the flue gas, during normal combustion conditions and memory effect periods (i.e. combustion periods following transient combustion conditions) at 400 °C (A) and 200 °C (B) sampling points in the post-combustion zone. Error bars denote 1 standard deviation.

54

•

Figure 26. TeCDF isomer patterns with SO2:HCl ratios of 0 and 1.6 in the flue gas during normal combustion conditions and transient combustion conditions at the 400 °C sampling point in the post-combustion zone. Error bars denote 1 standard deviation.

Summary

The results show that PCDD/F formation is rapid and that net formation occurs primarily at temperatures between 640 °C and 400 °C in the post-combustion zone. Prolonging the residence time (by approximately one second) at 450/460 °C or increasing the SO2:HCl ratio in the flue gas from 0 to 0.4 or 1.6 in the flue gas were the only investigated factors that induced reductions in the PCDD/F emission levels compared to those observed in the baseline runs. In contrast, transient combustion conditions, reducing the secondary combustion temperatures (from 800 °C to 660 °C) and increasing the chlorine content of the fuel (from 0.7% to 1.7%) all increased the PCDD/F levels.

The changes in PCDF homologue profiles with transient combustion conditions (incomplete combustion) and increases in chlorine levels in the fuel, SO2:HCl ratios in the flue gas and freeboard temperatures (from 800 °C to 950 °C) indicate that PCDFs are mainly formed by chlorination from lightly or non-chlorinated precursors or carbon structures (de novo synthesis). Thus, the PCDF homologue profile appears to depend on the relative amounts of backbone structures and chlorinating agents.

55

In addition, shifts in the PCDD homologue profile towards higher degrees of chlorination than baseline profiles were found with increases in both chlorine levels in the fuel (presumably due to increased chlorination activity), and SO2:HCl ratios in the flue gas (which presumably result in low levels of CuCl2). In the latter case the shift in the PCDD homologue profile was found to be due to increases in condensation reaction products.

The findings indicate that sulfur has a greater inhibitory effect on PCDF formation than on PCDD formation, presumably due to reductions in amounts of the chlorinating agent CuCl2. This appeared to induce shifts in the PCDD and PCDF formation reactions, pointing in an enhancement of condensation reactions.

Transient combustion conditions induced changes in the PCDD/PCDF ratios, homologue profiles and congener patterns. Notably, the PCDD/PCDF ratios in the flue gas were found to increase with increases in the SO2:HCl ratio in the flue gas and ash deposits originating from transient combustion conditions. The transient combustion conditions also induced increases in PCDD congeners characteristically formed in condensation reaction and PAH-related PCDF congeners. In the periods following the transient combustion conditions the relative importance of condensation reactions increased with increases in the SO2:HCl ratio in the flue gas. In addition, the results indicate that the relative activities of the pathways largely responsible for the formation of PCDDs and PCDFs are due to competing reactions, which presumably depend on the level of CuCl2 in the fly ash.

It was also found that increased temperatures (360-260 °C) in the post-combustion sections, in which temperatures are not normally high, activated residues on the tube walls, with consequent increases in PCDD levels and promotion of the condensation reactions.

The overall results also indicate that the relative importance of different pathways shifts in the post-combustion zone, condensation products increasing with decreases in temperature and increases in residence time. However, neither the HCl level in the flue gas nor changes in NOx and N2O concentrations in the flue gas due to the variations of the parameters correlated with the PCDD/F emission levels.

56

•

PAH and PCBz None of the analyzed PAHs were detected in the baseline samples, confirming that virtually complete combustion occurred under baseline conditions. As shown in Table 10, the sum PAH concentrations observed during the sampling periods in the transient combustion period did not correlate with the average CO concentrations in the flue gas. However, while the CO concentration was higher (2700 ppm versus 1900 ppm) the PAH level was similar (11 versus 15 ng PAH/m3) in the flue gas with the SO2:HCl ratio of 1.6 than with the SO2:HCl ratio of 0 in the flue gas, implying that large additions of sulfur may reduce PAH levels in the flue gas during transient combustion conditions.

Table 10. CO concentration and PAH concentration measured under transient combustion conditions in experiments with different SO2:HCl ratios.

SO2:HCl ratio CO ppm

Sum PAH ng/m3

0 1900 11 0 1600 10

0.1 1800 18 0.4 2500 25 1.6 2700 15

a Sum PAH = 15 of the 16 US EPA PAHs (naphthalene not included).

The sum PAH concentration in the flue gas decreased rapidly with time after the transient combustion conditions, resulting in a PAH concentration of < 2 ug/m3 during the last combustion period (memory effects 2). As shown in Figure 27, levels of the 15 analyzed PAHs were similar at all three sampling points/temperatures along the post-combustion zone during the transient combustion conditions. In addition, during the transient combustion conditions the most abundant PAHs were acenaphthylene and phenanthrene, followed by fluoranthene, pyrene, anthracene and fluorene, in accordance with the PAHs found to be most abundantly formed during transient combustion conditions in full-scale investigations (49,78,88). However, as shown in Figure 27 the most abundant PAHs changed with addition of sulfur to phenanthrene, fluoranthene and acenaphtylene. Levels

57

of the PAHs during each of the combustion periods following the transient combustion conditions tended to increase along the post-combustion zone, possibly due to adsorption of PAHs on soot or convector wall surfaces, which may migrate more slowly at 200 °C than at 400 °C.

1

10

100

1000

10000

100000

1000000

10000000

acen

apht

ylen

e

acen

apht

ene

fluor

ene

phen

anth

rene

anth

race

ne

fluor

anth

ene

pyre

ne

benz

o(a)

anth

race

ne

chry

sene

benz

o(b)

fluor

anth

ene+

be

nzo(

k)flu

oran

then

e

benz

o(a)

pyre

ne

inde

no(1

,2,3

-cd)

pyre

ne

benz

o(g,

h,i)p

eryl

ene

dibe

nzo(

a,h)

anth

race

ne

ng/m

3 (n

orm

)

Transient P3 (400 °C) Transient P4 (300 °C) Transient P7 (200 °C)Memory 1 P3 (400 °C) Memory 1 P4 (300 °C) Memory 1 P7 (200 °C)Memory 2 P3 (400 °C) Memory 2 P4 (300 °C) Memory 2 P7 (200 °C)

Run 2, S:Cl 0A. SO2:HCl 0

1

10

100

1000

10000

100000

1000000

10000000

acen

apht

ylen

e

acen

apht

ene

fluor

ene

phen

anth

rene

anth

race

ne

fluor

anth

ene

pyre

ne

benz

o(a)

anth

race

ne

chry

sene

benz

o(b)

fluor

anth

ene+

be

nzo(

k)flu

oran

then

e

benz

o(a)

pyre

ne

inde

no(1

,2,3

-cd)

pyre

ne

benz

o(g,

h,i)p

eryl

ene

dibe

nzo(

a,h)

anth

race

ne

ng/m

3 (nor

m)

Transient P3 (400 °C) Transient P4 (300 °C) Transient P7 (200 °C)Memory 1 P3 (400 °C) Memory 1 P4 (300 °C) Memory 1 P7 (200 °C)Memory 2 P3 (400 °C) Memory 2 P4 (300 °C) Memory 2 P7 (200 °C)

Run 6, S:Cl 1.6B. SO2:HCl 1.6

Figure 27. PAH distributions along the post-combustion zone during and after transient combustion conditions without (SO2:HCl ratio 0, A) and with (SO2:HCl ratio 1.6, B) sulfur addition (logarithmic scale on the y-axis).

58

•

The average sum tri- to hexa-chlorinatedBz concentration in the baseline was about 2 μg/m3 at all sampling points, with CVs < 20%. In addition, the PCBz homologue profile declined with chlorination, at all three sampling points/temperatures: P3 (400 °C), P4 (300 °C) and P7 (200 °C). The PCBz were not analyzed in all samples. However, the PCBz levels were reduced by 60% with a SO2:HCl ratio of 1.6 in the flue gas. In addition, the TriCBz and TeCBz isomer patterns differed between normal and transient combustion conditions, the most abundant isomers in the respective groups changed from 1,2,4-TriCBz and 1,2,3,5/1,2,4,5-TeCBz during normal combustion to 1,2,3-TriCBz and 1,2,3,4-TeCBz during transient combustion conditions (Figure 28). The high abundance of 1,2,4-TriCBz and 1,2,3,5/1,2,4,5-TeCBz isomers during normal combustion conditions is consistent with formation from PCPhs and investigations of their de novo synthesis from carbon (36,90). Whereas the abundance of 1,2,3-TriCBz and 1,2,3,4-TeCBz isomers during transient combustion conditions is consistent with de novo formation from PAHs (49).

Figure 28. Isomer distribution of Tri- and TetraCBz during normal and transient combustion conditions, SO2:HCl ratio 0, at sampling points P3 (400 °C), P4 (300 °C) and P7 (200 °C). Error bars denote 1 standard deviation.

59

60

•

5. CONCLUSIONS AND FUTURE WORK

The main objective of the work outlined in this thesis was to investigate the effects of varying combustion parameters on PCDD/F emissions and formation, in order to elucidate parameters that substantially contribute to reductions or increases in the release of PCDD/Fs from MSW incineration. The major conclusions from the work can be summarized as follows.

The following investigated parameters substantially affected the PCDD/F emission levels:

• Prolonging the residence time (by approximately one second) at 450/460 °C can reduce the PCDD/F emission levels by up to 60%, even downstream of the post-combustion zone.

• Increasing the SO2:HCl ratio from 0 to 0.4 or 1.6 can reduce the PCDD/F (TEQ) emissions levels by up to 60%.

• Transient combustion conditions substantially increased (100-fold) the PCDD/F emission levels.

• Increasing the chlorine content of the fuel from 0.7% to 1.7% resulted in a 20-fold increase in PCDD/F emission levels.

• Reducing the secondary combustion zone temperature from 800 °C to 660 °C resulted in a 3-fold increase in PCDD/F emission levels.

Future work: The potential for prolonging the residence time at 450/460 °C to reduce PCDD/F emission levels from full-scale plants should be confirmed. In addition, it would be interesting to examine the combined effects of prolonging residence times at 450/460 °C and raising SO2:HCl ratios to 0.4 or 1.6 in the flue gas to determine if PDD/F emissions could be further, synergistically reduced.

Effects of the investigated parameters on PCDD/F formation:

• PCDFs appeared to be formed by chlorination of carbon structures (de novo synthesis) or lightly or non-chlorinated precursors. Furthermore, the PCDF homologue profile appeared

61

to depend on the relative amounts of backbone structures and chlorinating agents.

• The findings indicate that sulfur has a greater inhibitory effect on PCDF formation than on PCDD formation, suggesting that the PCDD and PCDF formation pathways differ.

• Addition of sulfur and/or transient combustion conditions induced shifts in the PCDD and PCDF reaction pathways, pointing in an enhancement of condensation reactions, which increased with decreased temperature and increased residence time in the post-combustion zone.

• The findings indicate that PCDD/PCDF ratios found in the flue gas from MSW incineration depend on both the SO2:HCl ratio in the flue gas and ash deposits originating from transient combustion conditions.

• A tendency for increased SO2 levels in the flue gas to increase levels of PCDTs was also detected, but the increases were much less significant than the reductions in PCDF levels. Thus, the reductions in PCDF levels were not compensated by increases in PCDTs, the sulfur analogues of PCDFs.

• Increasing the chlorine content of the fuel increased the degrees of chlorination of both the PCDFs and PCDDs.

Future work: In order to determine the effects of sulfur on PCDD/F formation more comprehensively, the Cl2 levels in the flue gas and the ash composition of Cl-containing compounds, such as CuCl2, at various sampling points/temperatures in the post-combustion zone should be analyzed in detail. Another approach would be to add Cl2 gas at different locations in the post-combustion zone.

62

•

6. ACKNOWLEDGMENT

The research project was in part a collaboration between Umeå University, Rambøll (Denmark) and FORCE Technology (Denmark). The studies were partly financed by PSO (Public Service Obligations) funds from the Danish Ministry for Transport and Energy, Energinet.dk, and the Faculty of Science and Technology at Umeå University, whose support is gratefully acknowledged. Special thanks are due to Jens Kjems Toudal and Tore Hulgaard at Rambøll, and Jonas Dahl at FORCE Technology for all their valuable comments along the way. In addition, The Kempe foundation (JC Kempes minnes akademiska forskningsfond), the Swedish Association of Graduate Engineers (Sveriges Ingenjörer), and Ångpanneföreningen's Foundation for Research and Development (Ångpanneföreningens Forskningsstiftelse) are gratefully acknowledged for enabling the research work to be presented at conferences in Italy, Japan and USA.

Special thanks to my supervisors Stellan Marklund and Barbro Andrsson.

Special thanks are also due to:

The anonymous reviewers for providing valuable comments in order to improve the papers.

Jerker Fick for assisting during experiments, Per Liljelind for help with the GC/MS analyses and discussions, and Mats Tysklind for critical reading of the manuscript.

Marie Calla for assisting during experiments and laboratory work, and for always showing a professional, positive attitude despite the tight schedule.

Ingmar Olofsson and Ulf Nordström of the Energy Technology and Thermal Process Chemistry Group at Umeå University for help and assistance during the renovation of the laboratory-scale fluidized bed reactor.

In addition, Lisa Lundin and Stina Jansson are gratefully acknowledged for collaboration, discussions and for being great traveling companions throughout the six years we have worked together.

Thanks to all people at the former Environmental Chemistry.

63

64

•

7. REFERENCES

(1) Olie, K.; Vermeulen, P. L.; Hutzinger, O. Chlorodibenzo-p-dioxins and chlorodibenzofurans are trace components of fly ash and flue gas of some municipal incinerators in the netherlands. Chemosphere. 1977, 8, 455-459.

(2) European Union Council Directive 2000/76/EG. 2007.

(3) Council regulation (EC) No 1195/2006 amending Annex IV to regulation (EC) No 850/2004 of European Parliament and of the Council on persistent organic pollutants. Official Journal of the European Union. 2006.

(4) Safe, S. Polychlorinated-biphenyls (PCBs), dibenzo-para-dioxins (PCDDs), dibenzofurans (PCDFs), and related compounds - environmental and mechanistic considerations which support the development of toxic equivalency factors (TEFs). Critical Reviews in Toxicology. 1990, 21 (1), 51-88.

(5) van den Berg, M.; Birnbaum, L.; Boseveld, A. T. C.; Brunström, B.; Cook, P.; Feeley, M.; Giesy, J. P.; Hanberg, A.; Hasegawa, R.; Kennedy, S. W.; Kubiak, T.; Larsen, J. C.; van Leeuwen, F. X. R.; Liem, A. K. D.; Nolt, C.; Peterson, R. E.; Poellinger, L.; Safe, S.; Schrenk, D.; Tillitt, D.; Tysklind, M.; Younes, M.; Waern, F.; Zacharewski, T. Toxic equivalency factors (TEFs) for PCBs, PCDDs, PCDFs for humans and wildlife. Environmental Health Perspectives. 1998, 106, 775-792.

(6) Kopponen, P.; Sinkkonen, S.; Poso, A.; Gynther, J.; Karenlampi, S. Sulfur analogs of polychlorinated dibenzo-p-dioxins, dibenzofurans and diphenyl ehters as inducers of CYP1A1 in mouse heptatoma-cell culture and structure-activity-relationships. Environmental Toxicology and Chemistry. 1994, 13 (9), 1543-1548.

(7) Birgerson, B.; Sterner, O.; Zimerson, E. Kemiska hälsorisker Toxikologi i kemiskt perspektiv. Malmö: Liber-Hermods. 1995, ISBN 91-23-01731-7.

(8) Pickering, R. W. A toxicology review of polycyclic aromatic hydrocarbons. Journal of Toxicology-Cutaneous and Ocular Toxicology. 1999, 18 101-135.

(9) Niessen, W. R. Combustion and incineration processes 3rd ed. Marcel Dekker, Inc. :New York. 2002, (ISBN: 0-8247-0629-3).

65

(10) de Diego, L. F.; Londono, C. A.; Wang, X. S.; Gibbs, B. M. Influence of operating parameters on NOx and N2O axial profiles in a circulating fluidized bed combustor. Fuel. 1996, 75 (8), 971-978.

(11) Dam-Johansen, K.; Åmand, L.-E.; Leckner, B. Influence of SO2 on the NO/N2O chemistry in fluidized bed combustion 2. Interpretation of full-scale observations based on laboratory experiments. Fuel. 1993, 72 (4), 565-571.

(12) Bonn, B.; Pelz, G.; Baumann, H. Formation and decomposition of N2O in fluidized bed boilers. Fuel. 1995, 74 (2), 165-171.

(13) Zevenhoven, R.; Hupa, M. The reactivity of chars from coal, peat and wood towards NO, with and without CO. Fuel. 1998, 77 (11), 1169-1176.

(14) Lu, Y.; Hippinen, I.; Jahkola, A. Control of NOx and N2O in pressurized fluidized-bed combustion. Fuel. 1995, 74 (3), 317-322.

(15) Miccio, F.; Löffler, G.; Wargadalam, V. J.; Winter, F. The influence of SO2 level and operating conditions on NOx and N2O emissions during fluidised bed combustion of coals. Fuel. 2001, 80, 1555-1566.

(16) Stieglitz, L. Selected topics on the de novo Synthesis of PCDD/PCDF on fly ash. Environmental Engineering Science. 1998, 15 (1), 5-18.

(17) Boman, C. Particulate and gaseous emissions from residential biomass combustion. Doctoral Thesis, Energy Technology and Thermal Process Chemistry, Umeå University, Umeå. 2005.

(18) Shaub, W. M.; Tsang, W. Dioxin Formation in Incinerators. Environmental Science & Technology. 1983, 17, 721-730.

(19) Vogg, H.; Stieglitz, L. Thermal behavior of PCDD/PCDF in fly ash from municipal incinerators. Chemosphere. 1986, 15 (9-12), 1373-1378.

(20) Weber, R.; Sakurai, T.; Hagenmaier, H. Formation and destruction of PCDD/PCDF during heat treatment of fly ash samples from fluidized bed incinerators. Chemosphere. 1999, 38 (11), 2633-2642.

66

•

(21) Wehrmeier, A.; Lenoir, D.; Schramm, K.-W.; Zimmermann, R.; Hahn, K.; Henkelmann, B.; Kettrup, A. Patterns of isomers of chlorinated dibenzo-p-dioxins as tool for elucidation of thermal formation mechanisms. Chemosphere. 1998, 36 (13), 2775-2801.

(22) Stieglitz, L.; Zwick, G.; Beck, J.; Roth, W.; Vogg, H. On the de-novo synthesis of PCDD/PCDF on fly ash of municipal waste incinerators. Chemosphere. 1989, 18 (1-6), 1219-1226.

(23) Luijk, R.; Dorland, C.; Kapteijn, F.; Govers, H. A. J. The formation of PCDDs and PCDFs in the catalysed combustion of carbon: implications for coal combustion. Fuel. 1993, 72, 343-347.

(24) Stieglitz, L.; Vogg, H.; Zwick, G.; Beck, J.; Bautz, H. On formation conditions of organohalogen compounds from particulate carbon of fly ash. Chemosphere. 1991, 23 (8-10), 1255-1264.

(25) Iino, F.; Imagawa, T.; Takeuchi, M.; Sadakata, M. De novo synthesis mechanism of polychlorinated dibenzofurans from polycyclic aromatic hydrocarbons and the characteristic isomers of polychlorinated naphthalenes. Environmental Science & Technology. 1999, 33, 1038-1043.

(26) Addink, R.; Cnubben, P. A. J. P.; Olie, K. Formation of polychlorinated dibenzo-p-dioxins/dibenzofurans on fly ash from precursors and carbon model compounds. Carbon. 1995, 33 (10), 1463-1471.

(27) Altwicker, E. R.; Konduri, R. K. N. V.; Lin, C.; Milligan, M. S. Rapid formation of polychlorinated dioxins/furans in the post combustion region during heterogeneous combustion. Chemosphere. 1992, 25 (12), 1935-1944.

(28) Ghorishi, S. B.; Altwicker, E. R. Rapid formation of polychlorinated dioxins/furans during the heterogeneous combustion of 1,2-dichlorobenzene and 2,4-dichlorophenol. Chemosphere. 1996, 32 (1), 133-144.

(29) Karasek, F. W.; Dickson, L. C. Model Studies of Polychlorinated Dibenzo-p-Dioxin Formation During Municipal Refuse Incineration. Science. 1987, 237, 754-756.

(30) Luijk, R.; Akkerman, D. M.; Slot, P.; Olie, K.; Kapteijn, F. Mechanism of formation of polychlorinated dibenzo-p-dioxins and dibenzofurans in the catalyzed combustion of carbon. Environmental Science & Technology. 1994, 28, 312-321.

67

(31) Mulholland, J. A.; Ryu, J.-Y. Formation of polychlorinated dibenzo-p-dioxins by CuCl2-catalyzed condensation of 2,6 chlorinated phenols. Combustion Science and Technology. 2001, 169, 107-126.

(32) Sidhu, S. S.; Maqsud, L.; Dellinger, B. The homogeneous, gas-phase formation of chlorinated and brominated dibenzo-p-dioxin from 2,4,6-trichloro- and 2,4,6-tribromophenols. Combustion and Flame. 1995, 100, 11-20.

(33) Born, J. G.; Mulder, P.; Louw, R. Fly ash mediated reactions of phenol and monochlorophenols: oxychlorination, deep oxidation, and condensation. Environmental Science & Technology. 1993, 27, 1849-1863.

(34) Dickson, L. C.; Lenoir, D.; Hutzinger, O. Inhibiton of chlorinated dibenzo-p-dioxin formation on municipal incinerator fly ash by using catalyst inhibitors. Chemosphere. 1989, 19 (8/9), 1435-1445.

(35) Milligan, M. S.; Altwicker, E. R. Chlorophenol reactions on fly ash. 1. Adsorption/desorption equilibria and conversion to polychlorinated dibenzo-p-dioxins. Environmental Science & Technology. 1996, 30, 225-229.

(36) Hell, K.; Stieglitz, L.; Dinjus, E. Mechanistic aspects of the de-novo synthesis of PCDD/PCDF on model mixtures and MSWI fly ashes using amorphous 12C- and 13C-labeled carbon. Environmental Science & Technology. 2001, 35, 3892-3898.

(37) Wikström, E.; Ryan, S.; Touati, A.; Gullett, B. K. Key parameters for de novo formation of polychlorinated dibenzo-p-dioxins and dibenzofurans. Environmental Science & Technology. 2003, 37, 1962-1970.

(38) Altwicker, E. R. Relative rates of formation of polychlorinated dioxins and furans from precursor and de novo reactions. Chemosphere. 1996, 33, 1897-1904.

(39) Huang, H.; Buekens, A. On the mechanisms of dioxin formation in combustion processes. Chemosphere. 1995, 31 (9), 4099-4117.

(40) Everaert, K.; Baeyens, J. The formation and emission of dioxins in large scale thermal processes. Chemosphere. 2002, 46, 439-448.

68

•

(41) Wikström, E.; Löfvenius, G.; Rappe, C.; Marklund, S. Influence of Level and Form of Chlorine on the Formation of Chlorinated Dioxins, Dibenzofurans, and Benzenes during Combustion of an Artificial Fuel in a Laboratory Reactor. Environmental Science & Technology. 1996, 30 (5), 1637-1644.

(42) Wikström, E.; Marklund, S.; Tysklind, M. Influence of Variation in Combustion Conditions on the Primary Formation of Chlorinated Organic Micropollutants during Municipal Waste Combustion. Environmental Science & Technology. 1999, 33, 4263-4269.

(43) Addink, R.; Govers, H. A. J.; Olie, K. Isomer distributions of polychlorinated dibenzo-p-dioxins/dibenzofurans formed during de novo synthesis on incinerator fly ash. Environmental Science & Technology. 1998, 32, 1888-1893.

(44) Weber, R.; Hagenmaier, H. PCDD/PCDF formation in fluidized bed incineration. Chemosphere. 1999, 38 (11), 2643-2654.

(45) Neuer-Etscheidt, K.; Nordsieck, H. O.; Liu, Y.; Kettrup, A.; Zimmermann, R. PCDD/F and other micropollutants in MSWI crude gas and ashes during plant start-up and shut-down processes. Environmental Science & Technology. 2006, 40, 342-349.

(46) Oh, J.-E.; Gullett, B. K.; Ryan, S.; Touati, A. Mechanistic relationships among PCDD/Fs, PCNs, PAHs, CIPhs, and CIBzs in municipal waste incineration. Environmental Science & Technology. 2007, 41, 4705-4710.

(47) Hell, K.; Altwicker, E. R.; Stieglitz, L.; Addink, R. Comparison of 2,4,6-trichlorophenol conversion to PCDD/PCDF on a MSWI-fly ash and a model fly ash. Chemosphere. 2000, 40, 995-1001.

(48) Ryu, J.-Y.; Mulholland, J. A.; Takeuchi, M.; Kim, D. H.; Hatanaka, T. CuCl2-catalyzed PCDD/F formation and congener patterns from phenols. Chemosphere. 2005, 61, 1312-1326.

(49) Weber, R.; Iino, F.; Imagwa, T.; Takeuchi, M.; Sakurai, T.; Sadakata, M. Formation of PCDF, PCDD, PCB, and PCN in de novo synthesis from PAH: Mechanistic aspects and correlation to fluidized bed incinerators. Chemosphere. 2001, 44, 1429-1438.

(50) Preto, F.; McCleave, R.; McLaughlin, D.; Wang, J. Dioxins/furans emissions from fluidized bed combustion of salt-laden hog fuel. Chemosphere. 2005, 58, 935-941.

69

(51) Hatanaka, T.; Imagawa, T.; Takeuchi, M. Formation of PCDD/Fs in Artificial Solid Waste Incineration in a Laboratory-Scale Fluidized-Bed Reactor: Influence of Contents and Forms of Chlorine Sources in High-Temperature Combustion. Environmental Science & Technology. 2000, 34, 3920-3924.

(52) Lenoir, D.; Kaune, A.; Hutzinger, O.; Mutzenich, G.; Horch, K. Influence of operating parameters and fuel type on PCDD/F emissions from a fluidized bed incinerator. Chemosphere. 1991, 23 (8-10), 1491-1500.

(53) Griffin, R. D. A new theory of dioxin formation in municipal solid waste combustion. Chemosphere. 1986, 15 (9-12), 1987-1990.

(54) Gullett, B. K.; Bruce, K. R.; Beach, L. O. Formation of chlorinated organics during solid waste combustion. Waste Management & Research. 1990, 8 (3), 203-214.

(55) Gullett, B. K.; Bruce, K. R.; Beach, L. O. The effect of metal catalysts on the formation of polychlorinated dibenzo-p-dioxin and polychlorinated dibenzofuran precursors. Chemosphere. 1990, 20 (10-12), 1945-1952.

(56) Nestrick, T. J.; Lamparski, L. L.; Crummet, W. B. Thermolytic surface-reaction of benzene and iron (III) chloride to form chlorinated dibenzo-p-dioxins and dibenzofurans. Chemosphere. 1987, 16 (4), 777-790.

(57) Ryan, S. P.; Altwicker, E. R. Understanding the role of iron chlorides in the de novo synthesis of polychlorinated dibenzo-p-dioxins/dibenzofurans. Environmental Science & Technology. 2004, 38, 1708-1717.

(58) Söderström, G.; Marklund, S. Formation of PBCDD and PBCDF during flue gas cooling. Environmental Science & Technology. 2004, 38, 825-830.

(59) Hatanaka, T.; Kitajima, A.; Takeuchi, M. Role of copper chloride in the formation of polychlorinated dibenzo-p-dioxins and dibenzofurans during incineration. Chemosphere. 2004, 57, 73-79.

(60) Wikström, E.; Ryan, S.; Touati, A.; Telfer, M.; Tabor, D.; Gullett, B. K. Importance of chlorine speciation on de novo formation of polychlorinated dibenzo-p-dioxins and polychlorinated dibenzofurans. Environmental Science & Technology. 2004.

70

•

(61) Nonhebel, D. C. Copper-catalysed single-electron oxidations and reducuctions. Chemical Society (London) Special Publications. 1970, 24, 409-437.

(62) Taylor, P. H.; Sidhu, S. S.; Rubey, W. A.; Dellinger, B.; Wehrmeier, A.; Lenoir, D.; Schramm, K.-W. Evidence for a unified pathway of dioxin formation from aliphatic hydrocarbons. Symposium (Internationl) on Combustion. 1998, 27th, 1769-1775.

(63) Chang, M. B.; Cheng, Y. C.; Chi, K. H. Reducing PCDD/F formation by adding sulfur as inhibitor in waste incineration processes. Science of the Total Environment. 2006, 366, 456-465.

(64) Gullett, B. K.; Raghunathan, K.; Dunn, J. E. The effect of cofiring high-sulfur coal with municipal waste on formation of polychlorinated dibenzodioxin and polychlorinated dibenzofuran. Environmental Engineering Science. 1998, 15 (1), 59-70.

(65) Gullett, B. K.; Dunn, J. E.; Raghunathan, K. Effect of cofiring coal on formation of polychlorinated dibenzo-p-dioxins and dibenzofurans during waste combustion. Environmental Science & Technology. 2000, 34 (2), 282-290.

(66) Hunsinger, H.; Seifert, H.; Jay, K. Reduction of PCDD/F formation in MSWI by a process-integrated SO2 cycle. Environmental Engineering Science. 2007, 24 (8), 1145-1159.

(67) Lindbauer, R. L. Combustion dioxin suppression in municipal solid waste incineration with sulfphur additives. Chemosphere. 1992, 25 (7-10), 1409-1414.

(68) Ogawa, H.; Orita, N.; Horaguchi, M.; Suzuki, T.; Okada, M.; Yasuda, S. Dioxin reduction by sulfur component addition. Chemosphere. 1996, 32 (1), 151-157.

(69) Hunsinger, H.; Seifert, H.; Jay, K. Reduction of PCDD/F formation in MSWI by a process-integrated SO2 cycle. IT3-Conference. 2006, May 15-19.

(70) Ryan, P. S.; Li, X.-D.; Gullett, B. K.; Lee, C. W.; Clayton, M.; Touati, A. Experimental study on the effect of SO2 on PCDD/F emissions: Determination of the importance of gas-phase versus solid-phase reactions in PCDD/F formation. Environmental Science & Technology. 2006, 40, 7040-7047.

(71) Gullett, B. K.; Bruce, K. R.; Beach, L. O. Effect of sulfur dioxide on the formation mechanism of polychlorinated dibenzodioxin and dibenzofuran in municipal waste combustors. Environmental Science & Technology. 1992, 26 (10), 1938-1943.

71

(72) Buser, H. R. Identification and sources of dioxin-like compounds: I. polychlorodibenzothiphenes and polychlorothianthrenes, the sulfur-analogues of the polychlorodibenzofurans and polychlorodibenzodioxins. Chemosphere. 1992, 25 (1-2), 45-48.

(73) Vogg, H.; Metzger, M.; Stieglitz, L. Recent findings on the formation and decomposition of PCDD/PCDF in municipal solid waste incineration. Waste Management & Research. 1987, 5, 285-294.

(74) Stieglitz, L.; Zwick, G.; Beck, J.; Bautz, H.; Roth, W. The role of particulate carbon in the de-novo synthesis of polychlorinated dibenzodioxins and -furans in fly-ash. Chemosphere. 1990, 20 (10-12), 1953-1958.

(75) Ruokojärvi, P. H.; Asikainen, A.; Ruuskanen, J.; Tuppurainen, K. A.; Mueller, C.; Kilpinen, P.; Yli-Keturi, N. Urea as a PCDD/F inhibitor in municipal waste incineration. Journal of the Air & Waste Management Association. 2001, 51 (March), 422-431.

(76) Hatanaka, T.; Imagawa, T.; Kitajima, A.; Takeuchi, M. Effects of combustion temperature on PCDD/Fs formation in laboratory-scale fluidized-bed incinerator. Environmental Science & Technology. 2001, 35, 4936-4940.

(77) Fängmark, I.; Strömberg, B.; Berge, N.; Rappe, C. The influence of post combustion temperature profiles on the formation of PCDDs, PCDFs, PCBzs and PCBs in a pilot incinerator. Environmental Science & Technology. 1994, 28, 624-629.

(78) Blumenstock, M.; Zimmermann, R.; Schramm, K.-W.; Kettrup, A. Influence of combustion conditions on the PCDD/F-, PCB-, PCBz- and PAH-concentrations in the post-combustion chamber of a waste incineration pilot plant. Chemosphere. 2000, 40, 987-993.

(79) Hunsinger, H.; Seifert, H.; Jay, K. Formation of PCDD/F during start-up of MSWI. Organohalogen compounds. 2003, 60-65.

(80) Leclerc, D.; Duo, W. L.; Vessey, M. Effects of combustion and operating conditions on PCDD/PCDF emissions from power boilers burning salt-laden wood waste. Chemosphere. 2006, 63, 676-689.

(81) Weber, R.; Sakurai, T.; Ueno, S.; Nishino, J. Correlation of PCDD/PCDF and CO values in a MSW incinerator-indication of memory effects in the high temperature/cooling section. Chemosphere. 2002, 49, 127-134.

72

•

(82) Neuer-Etscheidt, K.; Orasche, J.; Nordsieck, H.; Streibel, T.; Zimmermann, R.; Kettrup, A. Changes in PCDD/PCDF formation processes during instationary phases of combustor operation - exemplified by the use of Cl4DD isomer patterns. Chemosphere. 2007, 67, S205-S216.

(83) Gass, C. H.; Luder, K.; Wilken, M. PCDD/F-emissions during cold start-up and shut-down of a municipal waste incinerator. Organohalogen compounds. 2002, 56, 193-196.

(84) Kanters, J.; Louw, R. Thermal and catalysed halogenation in combustion reactions. Chemosphere. 1996, 32 (1), 89-97.

(85) Lee, C. W.; Kilgroe, J. D.; Raghunathan, K. Effect of Soot and Copper Combustor Deposits on Dioxin Emissions. Environmental Engineering Science. 1998, 15 (1), 71-84.

(86) Wikström, E.; Ryan, S.; Touati, A.; Tabor, D.; Gullett, B. K. Origin of carbon in polychlorinated dioxins and furans formed during sooting combustion. Environmental Science & Technology. 2004, 38, 3778-3784.

(87) Richter, H.; Howard, J. B. Formation of polycyclic aromatic hydrocarbons and their growth to soot-a review of chemical reaction pathways. Progress in Energy and Combustion science. 2000, 26, 565-608.

(88) Zimmermann, R.; Blumenstock, M.; Heger, H. J.; Schramm, K.-W.; Kettrup, A. Emission of Nonchlorinated and Chlorinated Aromatics in the Flue Gas of Incineration Plants during and after Transient Disturbances of Combustion Conditions:Delayed Emission Effects. Environmental Science & Technology. 2001, 35, 1019-1030.

(89) Ryu, J.-Y.; Mulholland, J. A.; Kim, D. H.; Takeuchi, M. Homologue and Isomer Patterns of Polychlorinated Dibenzo-p-dioxins and Dibenzofurans from Phenol Precursors: Comparison with Municipal Waste Incinerator Data. Environmental Science & Technology. 2005, 39, 4398-4406.

(90) Ryu, J.-Y.; Mulholland, J. A. Metal-mediated chlorinated dibenzo-p-dioxin (CDD) and dibenzofuran (CDF) formation from phenols. Chemosphere. 2005, 58, 977-988.

(91) Wikström, E.; Andersson, P.; Marklund, S. Design of a laboratory scale fluidized bed reactor. Review of Scientific Instruments. 1998, 69, 1850-1859.

73

(92) Wikström, E.; Marklund, S. Combustion of an artificial municipal solid waste in a laboratory fludised bed reactor. Waste Management & Research. 1998, 16 (4), 342-350.

(93) European Committee for Standardization. Stationary source emissions - Determination of the mass concentration of PCDDs/PCDFs. 1997, EN 1948:1-3.

(94) Liljelind, P.; Söderström, G.; Hedman, B.; Karlsson, S.; Lundin, L.; Marklund, S. Method for multiresidue determination of halogenated aromatics and PAHs in combustion-related samples. Environmental Science & Technology. 2003, 37 (16), 3680-3686.

(95) Bacher, R.; Swerev, M.; Ballschmiter, K. Profile and pattern of monochloro- through octachlorodibenzodioxins and -dibenzofurans in chimney deposits from wood burning. Environmental Science & Technology. 1992, 26, 1649-1655.

(96) Ryan, J. J.; Conacher, H. B. S.; Panopio, L. G.; Lau, B. P. Y.; Hardy, J. A. Gas chromatographic separations of all 136 tetra- to octa-polychlorinated dibenzo-p-dioxins and polychlorinated dibenzofurans on nine different stationary phases. Journal of Chromatography. 1991, 541, 131-183.

(97) Gullett, B. K.; Dunn, J. E.; Bae, S.-K.; Raghunathan, K. Effects of combustion parameters on polychlorinated dibenzodioxin and dibenzofuran homologue profiles from municipal waste and coal co-combustion. Waste Management. 1998, 18, 473-483.

(98) Lundholm, K. Fate of Cu, Cr, As and some other trace elements during combustion of recovered waste fuels. Doctoral Thesis, Energy Technology and Thermal Process Chemistry, Umeå University, Umeå. 2007.

(99) Dayton, D. C.; Belle-Oudry, D.; Nordin, A. Effect of coal minerals on chlorine and alkali metals released during biomass/coal cofiring. Energy & Fuels. 1999, 13, 1203-1211.

74

•

Appendix I-TEQ values derived from experimental runs in Paper I-IV (ng/m3 (norm)).

Sampling point/temperature Paper Exp. Description P3

(400 °C) P4

(300 °C) P7

(200 °C)

I-IV Baseline/normal combustion conditions

0.60 0.63 0.64

CV [%] 10 24 24 IV 2 SO2:HCl 0.1 0.66 0.57 -

3 SO2:HCl 0.4 0.21 0.25 0.26 4 SO2:HCl 1.6 0.12 0.22 0.34 I 7 Cu 0.011% - 0.30 0.43 8 H2O 21% - 0.46 0.94 9 Cl 1.7% 5.8 7.5 6.11 10 Freeboard 950 °C - 0.74 0.71 11 Freeboard 660 °C - 2.2 1.7 12 Fuel load 0.6 kg/h 0.14 0.12 0.12

III-IV T SO2:HCl 0 - 100 100 M1 - 28 36 M2 - 4.0 6.6 T SO2:HCl 0 63 52 55 M1 44 77 80 M2 4.8 11 12 T SO2:HCl 0.1 15 12 24 M1 15 21 32 M2 2.5 5.6 8.9 T SO2:HCl 0.4 20 26 28 M1 8.8 15 15 M2 1.2 6.4 6.1 T SO2:HCl 1.6 11 11 12 M1 4.7 9.0 8.9 M2 0.77 2.9 3.1 P3

(300 °C) P4

(200 °C) P7

(100 °C)

II 8-9 Low-temperature 0.50 0.36 0.22 P3

(460 °C) P4

(360 °C) P7

(260 °C)

II 13-15 High-temperature 0.17 0.21 0.55

1

2

•