thermal formation and chlorination of dioxins and dioxin …142298/fulltext01.pdf · thermal...

Thermal Formation and Chlorination of Dioxins and Dioxin-Like Compounds

STINA JANSSON

Akademisk avhandling som med tillstånd av rektorsämbetet vid Umeå universitet för erhållande av Filosofie Doktorsexamen i Kemi med inriktning mot miljökemi vid Teknisk-naturvetenskapliga fakulteten i Umeå, framlägges till offentlig granskning vid Kemiska institutionen, hörsal KB3B1 i KBC-huset, fredagen den 7/11, 2008, klockan 10.00. Fakultetsopponent: Professor Jochen F. Müller, National Research Centre for Environmental Toxicology, The University of Queensland, Australia.

UMEÅ UNIVERSITY DOCTORAL DISSERTATION Department of Chemistry November 2008 SE-901 87 Umeå, Sweden

Stina Jansson


ABSTRACT This thesis contributes to an increased understanding of the formation of dioxins and dioxin-like compounds in combustion processes. Although emissions to air from waste incineration facilities have been greatly reduced by the use of efficient air pollution control measures, the resulting residues (ashes and filters) are highly toxic and are classified as hazardous waste. The main objective of the work underlying this thesis was to elucidate the formation and chlorination pathways of dioxins and dioxin-like compounds in waste combustion flue gases in the temperature range 640-200°C in a representative, well-controlled laboratory-scale reactor using artificial municipal solid waste. This could contribute to the reduction of harmful emissions to air and also reduce the toxicity of waste incineration residues, thus reducing or even eliminating the need for costly and potentially hazardous after-treatment. A comparison of four different quenching profiles showed that the formation of polychlorinated dibenzo-p-dioxins (PCDD) and dibenzofurans (PCDF) was rapid and mainly occurred in the 640-400°C temperature region, with high dependency on sufficient residence time within a specific temperature region. Prolonged residence time at high temperatures (450/460°C) reduced the PCDD yields, even at lower temperatures along the post-combustion zone. PCDD, PCDF and PCN (polychlorinated naphthalene) isomer distribution patterns indicated contributions from chlorophenol condensation as well as chlorination reactions for all three classes of compounds. The formation of PCDDs was largely influenced by chlorophenol condensation and to some extent by chlorination reactions. For the PCDFs, chlorine substitution adjacent to the oxygen bridges was unfavoured, as demonstrated by the notably lower abundance of 1,9-substituted congeners. This was supported by bidirectional orthogonal partial least squares (O2PLS) modelling. The variable with the greatest influence on the distribution of PCDD congeners was the relative free energy (R∆Gf). The O2PLS models displayed distinct clusters, dividing most of the homologues into two or three sub-groups of congeners which seemed to correspond to the probability of origination from chlorophenol condensation. The effects of injection of aromatic structures into the flue gas differed for each class of compounds. Injection of naphthalene increased the formation of monochlorinated naphthalene but the remaining homologues appeared to be unaffected. This was probably due to insufficient residence time at temperatures necessary for further chlorination. Injected dibenzo-p-dioxin was decomposed, chlorinated and re-condensated into PCDDs and PCDFs, whereas injection of dibenzofuran and fluorene reduced the PCDD levels in the flue gas.

Key words: dioxin, dioxin-like compounds, PCDD, PCDF, PCN, PCBz, PCPh, MSW combustion, formation, chlorination, flue gas, quench profiles, isomer distribution pattern, homologue profile, injection, O2PLS.

Language: English Number of pages: 70 + 3 appendices + 4 papers ISBN 978-91-7264-609-4


STINA JANSSON

UMEÅ UNIVERSITY Department of Chemistry

Umeå 2008

© 2008 Stina Jansson

Front cover image: Henrik Jansson

UMEÅ UNIVERSITY Department of Chemistry

SE-901 87 Umeå SWEDEN

ISBN 978-91-7264-609-4

To Elsa and Nils

TABLE OF CONTENTS ABSTRACT............................................................................................................. I SAMMANFATTNING (SUMMARY IN SWEDISH) ....................................... II LIST OF PAPERS .................................................................................................III

Contribution of the author of this thesis to published research.............................iv ABBREVIATIONS AND DEFINITIONS...........................................................V 1. INTRODUCTION..............................................................................................1

Main objective and aim ................................................................................................ 2 2. DIOXINS AND DIOXIN-LIKE COMPOUNDS.............................................3

General description....................................................................................................... 3 PCDD.......................................................................................................................... 3 PCDF .......................................................................................................................... 4 PCN............................................................................................................................. 4 PCBz............................................................................................................................ 4 PCPh ........................................................................................................................... 5

Toxicity of dioxins and dioxin-like compounds ....................................................... 5 Related compounds not addressed in this thesis...................................................... 6

3. THE COMBUSTION PROCESS.......................................................................9 Implications for formation of PCDD and PCDF ....................................................... 9

4. DIOXIN FORMATION AND DEGRADATION.........................................11 Formation pathways................................................................................................... 12

Formation from precursors ........................................................................................ 12 De novo synthesis...................................................................................................... 14 Chlorine substitution................................................................................................. 15

Degradation ................................................................................................................. 16 Reduction strategies.................................................................................................... 17

5. EXPERIMENTAL SECTION ..........................................................................19 General overview........................................................................................................ 19 Combustion experiments ........................................................................................... 20

The laboratory-scale fluidized bed reactor.................................................................. 20 The standardized artificial MSW .............................................................................. 21 Combustion conditions.............................................................................................. 22 Flue gas sampling...................................................................................................... 22 Injection procedure .................................................................................................... 23

Extraction and clean-up ............................................................................................. 25 Analytical procedure .................................................................................................. 26

6. POST-COMBUSTION FORMATION...........................................................27 The influence of quench profiles............................................................................... 27 Formation characteristics during the high-initial temperature profile ................ 31

Homologue profiles .................................................................................................... 31 Isomer distribution patterns ...................................................................................... 33

The effect of poor combustion on PCNs .................................................................. 37 7. INJECTION OF BACKBONE STRUCTURES..............................................39

The effect of dibenzo-p-dioxin injection on PCDD and PCDF .............................. 40 The effect of naphthalene injection on PCN ............................................................ 46

8. INFLUENCE OF MOLECULAR DESCRIPTORS .......................................47 Physicochemical properties ....................................................................................... 47 Chlorine substitution descriptors ............................................................................. 49 O2PLS – bidirectional orthogonal projections to latent structures ....................... 49 Modelling of PCDDs................................................................................................... 49 Modelling of PCDFs ................................................................................................... 52 Orthogonal variation .................................................................................................. 54

9. CONCLUDING REMARKS...........................................................................57 ACKNOWLEDGEMENTS .................................................................................59 REFERENCES ......................................................................................................61 APPENDICES 1-3

PAPERS I-IV

ABSTRACT This thesis contributes to an increased understanding of the formation of dioxins and dioxin-like compounds in combustion processes. Although emissions to air from waste incineration facilities have been greatly reduced by the use of efficient air pollution control measures, the resulting residues (ashes and filters) are highly toxic and are classified as hazardous waste. The main objective of the work underlying this thesis was to elucidate the formation and chlorination pathways of dioxins and dioxin-like compounds in waste combustion flue gases in the temperature range 640-200°C in a representative, well-controlled laboratory-scale reactor using artificial municipal solid waste. This could contribute to the reduction of harmful emissions to air and also reduce the toxicity of waste incineration residues, thus reducing or even eliminating the need for costly and potentially hazardous after-treatment.

A comparison of four different quenching profiles showed that the formation of polychlorinated dibenzo-p-dioxins (PCDD) and dibenzofurans (PCDF) was rapid and mainly occurred in the 640-400°C temperature region, with high dependency on sufficient residence time within a specific temperature region. Prolonged residence time at high temperatures (450/460°C) reduced the PCDD yields, even at lower temperatures along the post-combustion zone. PCDD, PCDF and PCN (polychlorinated naphthalene) isomer distribution patterns indicated contributions from chlorophenol condensation as well as chlorination reactions for all three classes of compounds. The formation of PCDDs was largely influenced by chlorophenol condensation and to some extent by chlorination reactions. For the PCDFs, chlorine substitution adjacent to the oxygen bridges was unfavoured, as demonstrated by the notably lower abundance of 1,9-substituted congeners. This was supported by bidirectional orthogonal partial least squares (O2PLS) modelling. The variable with the greatest influence on the distribution of PCDD congeners was the relative free energy (R∆Gf). The O2PLS models displayed distinct clusters, dividing most of the homologues into two or three sub-groups of congeners which seemed to correspond to the probability of origination from chlorophenol condensation.

The effects of injection of aromatic structures into the flue gas differed for each class of compounds. Injection of naphthalene increased the formation of monochlorinated naphthalene but the remaining homologues appeared to be unaffected. This was probably due to insufficient residence time at temperatures necessary for further chlorination. Injected dibenzo-p-dioxin was decomposed, chlorinated and re-condensated into PCDDs and PCDFs, whereas injection of dibenzofuran and fluorene reduced the PCDD levels in the flue gas.

i

SAMMANFATTNING (SUMMARY IN SWEDISH) Denna avhandling fokuserar på olika aspekter som kan bidra till en ökad förståelse av bildning av dioxiner och dioxin-lika föreningar i förbränningsprocesser. Även om utsläppen till luft från sopförbränningsanläggningar har minskat kraftigt tack vare effektiva rökgasreningsmetoder, så återstår problemet med mycket giftiga rökgasreningsprodukter (askor och filter), vilka klassificeras som farligt avfall. Det huvudsakliga syftet med arbetet bakom denna avhandling var att klarlägga bildnings- och kloreringsvägarna för dioxiner och dioxin-lika föreningar i temperaturintervallet 640-200°C i rökgaser från sopförbränning. Detta kan möjliggöra lösningar för ytterligare emissionsminskningar och en avgiftning av biprodukterna från avfallsförbränning, vilket minskar eller till och med eliminerar behovet av kostsam och riskfylld efterbehandling.

Realistiska och välkontrollerade försök har utförts i en lab-skalereaktor där en artificiell hushållssopa har förbränts. En jämförelse av fyra olika temperatur- och uppehållstidsprofiler visade att bildning av polyklorerade dibenso-p-dioxiner (PCDD) och dibensofuraner (PCDF) sker snabbt och huvudsakligen inom temperaturintervallet 640-400°C. Bildningen var starkt beroende av en tillräckligt lång uppehållstid inom ett visst temperaturområde. En förlängd uppehållstid vid höga temperaturer (>450°C) resulterade i minskade halter av PCDD, vilka förhöll sig låga även senare i efterförbränningszonen. Isomermönstren av PCDD, PCDF och PCN (polyklorerade naftalener) visade alla tecken på att härröra från både klorfenolkondensation och kloreringsreaktioner. PCDD-mönstret visade tydliga indikationer på bildning från klorfenoler, och till mindre grad bildning via klorering. För PCDF var klorsubstitution i positioner angränsande till syrebryggan missgynnad, vilket bekräftades av multivariat modellering (O2PLS). Den variabel som starkast påverkade bildningen av PCDD var relativa fria energin (R∆Gf). Modellerna visade på en distinkt gruppering av PCDD- och PCDF-kongenerna i två eller tre grupper för varje kloreringsgrad, och föreslås vara relaterad till sannolikheten för respektive kongen att bildas via klorfenolkondensation.

Injektion av aromatiska kolstrukturer i rökgaskanalen gav upphov till skilda effekter. Injektion av naftalen ökade bildningen av monoklorerad naftalen medan resterande homologer inte verkade påverkas, sannolikt på grund av för kort uppehållstid för ytterligare klorering. Dibenso-p-dioxin spjälkades sannolikt till fenoliska fragment som klorerades och sedan återkondenserades till PCDD och PCDF, medan dibensofuran och fluoren kraftigt reducerade PCDD-koncentrationerna.

ii

LIST OF PAPERS This thesis is based on the following papers, which are referred to by their respective Roman numerals.

I. Johanna Aurell, Stina Jansson, Stellan Marklund Effects of quench time profiles on PCDD/F formation in the post-combustion zone during MSW incineration

Environmental Engineering Science, in press

II. Stina Jansson, Jerker Fick, Mats Tysklind, Stellan Marklund Post-combustion formation of PCDD, PCDF, PCBz, and PCPh in a laboratory-scale reactor: Influence of dibenzo-p-dioxin injection Submitted to Chemosphere

III. Stina Jansson, Jerker Fick, Stellan Marklund Formation and chlorination of polychlorinated naphthalenes (PCNs) in the post-combustion zone during MSW combustion Chemosphere (2008) 72: 1138–1144

IV. Stina Jansson, Henrik Antti, Stellan Marklund, Mats Tysklind Correlations of PCDD/F isomer distribution patterns from MSW combustion with physicochemical variables and chlorine substitution descriptors Manuscript

Published papers are reproduced with kind permission from Mary Ann Liebert publishers (Paper I), and Elsevier Science (Paper III).

iii

LIST OF PAPERS

Contribution of the author of this thesis to published research

Paper I Stina Jansson planned parts of the study in co-operation with her co-authors, performed five of the experimental runs, and was responsible for sample clean-up and GC-MS analysis for the samples collected during runs 10-12. Stina Jansson also contributed with evaluation and interpretation of the data, and in writing the paper. Paper II Stina Jansson planned the study in co-operation with her co-authors, performed all the experiments and collection of samples, and was responsible for all laboratory work, GC-MS analysis, data evaluation and for writing the paper. Paper III Stina Jansson planned the study in co-operation with her co-authors, performed all the experiments and collection of samples, and was responsible for all laboratory work, GC-MS analysis, data evaluation and for writing the paper. Paper IV Stina Jansson planned the study in co-operation with her co-authors, performed all the experiments and collection of samples, and was responsible for all laboratory work, GC-MS analysis, as well as collection and compilation of the physicochemical property data set. Furthermore, Stina Jansson performed the multivariate data analysis and a major part of the evaluation and writing of the paper.

iv

ABBREVIATIONS AND DEFINITIONS APC air pollution control Cl-PAH chlorinated polycyclic aromatic hydrocarbon CV crossvalidation DD dibenzo-p-dioxin DF dibenzofuran dg dry gas dw dry weight FTIR Fourier Transform infrared GC-LRMS gas chromatography – low resolution mass spectrometry GC-HRMS gas chromatography – high resolution mass spectrometry IS internal standard I-TEQ international toxic equivalents IUPAC International Union of Pure and Applied Chemistry MSW municipal solid waste OPLS orthogonal partial least squares O2PLS bidirectional OPLS PAH polycyclic aromatic hydrocarbon PBT persistence – bioaccumulation - toxicity PCA principal component analysis PCB polychlorinated biphenyl PCBz polychlorinated benzene PCDD polychlorinated dibenzo-p-dioxin PCDD/F polychlorinated dibenzo-p-dioxin and dibenzofuran PCDF polychlorinated dibenzofuran PCN polychlorinated naphthalene PCPh polychlorinated phenol PICs products of incomplete combustion PLS partial least squares POPs persistent organic pollutants PUFP polyurethane foam plug PVC polyvinyl chloride

v

ABBREVIATIONS AND DEFINITIONS

RS recovery standard S/N ratio signal-to-noise ratio TEF toxic equivalency factor TEQ toxic equivalents v:v volume to volume w:w weight to weight Definitions

Congener An individual species within a compound group. Isomer Species with identical molecular formula, but different structural

formula. Homologue A group of species (isomers) with an identical number of chlorine

substituents, but which are substituted in different positions. Profile The levels and/or relative abundance of the respective homologue

groups. Pattern The distribution and/or relative abundance of individual isomers

within a homologue group. Homologue Homologue sum normalized to total sum fraction (ΣHOMOLOGUE/ΣTOTAL) Isomer Isomer concentration normalized to homologue sum fraction (cISOMER/ΣHOMOLOGUE) Substitution prefix

Mo- (Mono-) one Di- (Di-) two Tri- (Tri-) three Te- (Tetra-) four Pe- (Penta-) five Hx- (Hexa-) six Hp- (Hepta-) seven O- (Octa-) eight

vi

1. INTRODUCTION Dioxins are among the most toxic substances known to man. The most toxic dioxin is commonly referred to as TCDD. This is mainly known as the pollutant released in the accident at Seveso, northern Italy in 1976 and as contaminant in the defoliant Agent Orange used during the Vietnam War. In both cases the health effects from exposure were severe, for civilians as well as military personnel. A recent example of the effects of dioxin exposure was the seemingly intentional TCDD-poisoning of the Ukrainian president Viktor Yushchenko during his presidential campaign in 2004 (1).

In addition to their demonstrated toxic potential, dioxins are also highly stable and lipophilic. These properties make them resistant to degradation and prone to accumulation in organisms. The combination of persistence, bioaccumulation and toxicity (often referred to as ‘PBT’) identifies a substance to be an environmental pollutant. Consequently further knowledge of emission sources, formation, environmental distribution and biological effects of dioxins and related compounds is required. The environmental significance of these compounds is emphasized by their inclusion in the Stockholm convention of POPs (persistent organic pollutants) (2), which is an agreement signed by more than 150 nations to ban the production, use and trade of 10 commercially manufactured substances, and to minimize, or if possible eliminate, emissions of unintentionally formed by-products, i.e. dioxins.

Dioxins have never been deliberately manufactured for purposes other than scientific research, and environmental levels are thereby a result of natural formation processes (3) and unintentional anthropogenic release. The latter largely comprises emissions from industrial processes, in which dioxins are present as by-products and the manufacture of chemicals that contain dioxins as contaminants. In 1977, one year after the Seveso accident, Olie and co-workers (4) identified dioxins in municipal solid waste (MSW) incineration effluents. At present, combustion is an important method for waste management because it reduces the volume and recycles the energy in the waste efficiently. Although the use of MSW incineration has increased significantly since the late 1970’s, emissions of POPs to air have been greatly reduced. This was accomplished by the development of efficient air pollution control (APC) measures to meet strict regulations. However the APC residues (ashes and filters) are still highly toxic and are classified as hazardous waste which must be safely disposed and/or destructed. Thus, the dioxins are retained in an ash-bound form, which does not solve the problem, just hands it over to future generations due to the uncertainty about the long-term behaviour of dioxins in deposited MSW ash. Leaching of POPs from

1

INTRODUCTION

deposited residues is incompletely characterised and, although thermal removal of POPs from APC residues shows positive results (5,6), this process is energy consuming and costly.

As industrial emissions to air have been reduced, emissions of POPs from diffuse sources such as domestic heaters and open burning have become correspondingly more significant (7). The most practical, economic and environmentally sound strategy to reduce emissions of dioxins and dioxin-like compounds is to minimize their formation rather than to collect and destroy them after formation. Suitable measures include sorting and/or pre-treatment of waste, optimization of the combustion process, addition of inhibitors and catalytic destruction.

Main objective and aim The reduction of dioxin emissions from combustion sources has received considerable attention in recent decades and the mechanisms of dioxin formation have been studied extensively.

The main objectives of the research upon which this thesis is based were to elucidate the formation and chlorination pathways of dioxins and dioxin-like compounds in waste combustion flue gases between 640 and 200°C in a representative, well-controlled MSW combustion system, and to develop strategies to minimize the formation of dioxins. This could contribute to a further reduction of emissions to air, but would mainly enable the generation of less toxic waste incineration residues and thus reduce, or perhaps eliminate, the need for costly and potentially hazardous after-treatment.

The papers included in this thesis examine the influence of post-combustion zone temperatures and residence times, the effect of injection of unsubstituted aromatic compounds into the flue gas duct, and a multivariate correlation of isomer distribution patterns with physicochemical properties and chlorine substitution descriptors. Combined these studies provide an assessment of the variables that influence the formation of dioxins and dioxin-like compounds in combustion processes.

2

2. DIOXINS AND DIOXIN-LIKE COMPOUNDS The term ‘dioxin’ is often used not only in reference to polychlorinated dibenzo-p-dioxins (PCDDs), but also to polychlorinated dibenzofurans (PCDFs). Structurally related classes of compounds which have similar physicochemical properties, biochemical effects and toxicology to PCDDs and PCDFs, are often referred to as ‘dioxin-like’. The most dioxin-like is generally considered to be the polychlorinated biphenyls (PCB), which were however not included in this work. Figure 1 illustrates the general structures of the compounds addressed in this thesis. In addition to PCDDs and PCDFs, polychlorinated napthalenes (PCN), polychlorinated benzenes (PCBz) and polychlorinated phenols (PCPh) were included.

A B

O

98

7

6 4

3

21

O

O

4

3

2

19

8

7

6

6

5

4

3

2

112

3

456

7

85

4

3

2

6 OHC D E

ClxClx

Clx ClxClx

ClyCly

Cly Figure 1. Generalized structural formula and substitution positions of PCDD (A), PCDF (B), PCN (C), PCBz (D), and PCPh (E).

General description

PCDD The dibenzo-p-dioxin molecule is composed of two benzene rings linked by two oxygen bridges. The three-ring heterocyclic backbone structure is nearly planar and may be substituted by one to eight chlorine atoms in eight different positions, denoted 1-4 and 6-9 in Figure 1A, resulting in a total of 75 possible chlorinated congeners (Table 1). Congeners with the same number of chlorine substituents constitute a homologue, and the distribution of congeners within each homologue is referred to as an isomer distribution pattern. The substitution sites adjacent to the oxygen bridges (1-, 4-, 6-, and 9-) are referred to as α-positions, and the lateral (2-, 3-, 7-, and 8-sites) as β-positions.

3

DIOXINS AND DIOXIN-LIKE COMPOUNDS

Table 1. Number of possible congeners within each homologue for the groups of compounds addressed in this thesis.

No. of Cl PCDD PCDF PCN PCBz PCPh 1 2 4 2 1 3 2 10 16 10 3 6 3 14 28 14 3 6 4 22 38 22 3 3 5 14 28 14 1 1 6 10 16 10 1 7 2 4 2 8 1 1 1 Total 75 135 75 12 19

PCDF The dibenzofuran molecule differs from dibenzo-p-dioxin by the amount of oxygen incorporated into the molecule. The aromatic rings are connected by one oxygen bridge and one carbon-carbon bond (Figure 1B). This difference may seem insignificant, but the single oxygen bridge induces an angular configuration in the dibenzofuran molecule which alters the steric properties of the molecule and thus its substitution pattern. This configuration also results in the loss of a plane of symmetry relative to dibenzo-p-dioxin, which increases the number of possible PCDF congeners to 135 (Table 1). Furthermore, the dibenzofuran molecule is truly aromatic because it has a closed shell of delocalized electrons, making it more stable than dibenzo-p-dioxin, which is not an aromatic system.

PCN The chloronaphthalenes consist of two fused aromatic rings (Figure 1C) with one to eight chlorine substituents, resulting in 75 possible congeners (Table 1). PCNs are widespread environmental pollutants and were previously used in industrial applications (8), but these activities have ceased and PCNs are today mainly unintentionally formed and released through processes such as MSW incineration (9,10). PCNs have been demonstrated to possess dioxin-like biological activity (11,12) and their presence in many environmental matrices (13,14) emphasizes the need to investigate PCN formation and degradation characteristics.

PCBz Polychlorinated benzenes are the least complex of the chloroaromatic compound groups, and are composed of a benzene ring with one to six chlorine substituents (Figure 1D), resulting in a total of 12 possible congeners (Table 1). The fully chlorinated species hexachlorobenzene (HxCBz, or commonly called HCB) was formerly used as an agricultural fungicide and as a synthetic feedstock in industrial chemical processes. HxCBz is one of the ten prioritized substances in the Stockholm convention of POPs (2). Today, the occurrence of HxCBz in the environment is largely due to its formation as a by-product of industrial processes (15).

4


PCPh Polychlorinated phenols consist of a hydroxylated benzene ring substituted by one to five chlorine atoms (Figure 1E). The hydroxyl group greatly increases the polarity of the PCPh molecule relative to that of the corresponding PCBz species (15). Consequently, the PCPh also possesses higher reactivity and water solubility (16), however, the polarity decreases with increasing chlorination (15). Polychlorinated phenols, principally fully-chlorinated pentachlorophenol, have been extensively used as fungicides and for wood conservation purposes. There are 19 PCPh congeners (Table 1), of which 14 are substituted in at least one ortho-position i.e. the carbon site adjacent to the hydroxyl group. The ortho-chlorinated PCPhs are of particular interest in PCDD/F formation studies, since their structure largely determines possible condensation products. This will be further discussed in a later section of this thesis. Ortho-PCPhs are also more stable than other congeners due to intramolecular hydrogen bonding, and hence generally more abundant (16).

Toxicity of dioxins and dioxin-like compounds The toxicity of PCDD and PCDF congeners depends upon the number and position of chlorine substituents. The most toxic congener is, as previously mentioned, ‘TCDD’, or more correctly 2,3,7,8-tetrachlorodibenzo-p-dioxin (2,3,7,8-TeCDD, Figure 2), which is substituted in the lateral (or β-) positions. The mechanism of action involves binding to the aryl hydrocarbon (Ah-) receptor, and the Ah-receptor mediated response was used to derive toxic equivalency factors (TEFs). The TEF value may thus be viewed as a measure of the relative toxicity of a specific congener. The 2,3,7,8-TeCDD TEF value has been set to 1, and by comparing the toxic response of other laterally chlorinated PCDD, PCDF and PCB congeners with that of 2,3,7,8-TeCDD, TEF values have been derived and were recently re-evaluated by the World Health Organization, WHO (Table 2) (17).

O

O

Cl

Cl

Cl

Cl

Figure 2. Molecular structure of 2,3,7,8-TeCDD, the most toxic PCDD congener.

The total dioxin-related toxicity is derived by multiplying the concentrations of each of the 17 PCDD/F and 12 PCB congeners for which TEFs have been assigned by the corresponding TEF value. Summing the products provides the toxic equivalent (TEQ). Other species with confirmed toxic responses, e.g. HxCBz and a number of PCNs, have been proposed for possible inclusion in the TEF scheme but have not been assigned TEF values (17). The association between chlorine substitution pattern and toxicity demonstrates the need to undertake formation studies on the relative abundance of isomers within each

5


homologue as well as total and homologue levels to understand the factors affecting the dioxin-related toxic impacts of MSW incineration.

The TEQ measure has been used in a number of studies attempting to find one or more indicator compounds suitable for online PCDD/F monitoring purposes without time-consuming and expensive dioxin analyses. Several compounds or groups of compounds have been suggested as possible TEQ indicator species, e.g. mono- to tri-chlorinated CDD/F congeners (18,19), 2,3,4,7,8-PeCDF (20), and chlorobenzenes and chlorophenols (20-22).

Table 2. World Health Organization toxic equivalency factors (TEFs) for compounds possessing dioxin-related toxicity (17).

Compound TEF Compound TEF 2,3,7,8-TeCDD 1 3,3',4,4'-TeCB (#77b) 0.0001

1,2,3,7,8-PeCDD 1 3,4,4',5-TeCB (#81) 0.0003a

1,2,3,4,7,8-HxCDD 0.1 3,3',4,4',5-PeCB (#126) 0.1 1,2,3,6,7,8-HxCDD 0.1 3,3',4,4',5,5'-HxCB (#169) 0.03a

1,2,3,7,8,9-HxCDD 0.1 1,2,3,4,6,7,8-HpCDD 0.01

PCD

D

OCDD 0.0003a

Non-o

rtho

PCB

2,3,7,8-TeCDF 0.1 2,3,3',4,4'-PeCB (#105) 0.00003a

1,2,3,7,8-PeCDF 0.03a 2,3,4,4',5-PeCB (#114) 0.00003a

2,3,4,7,8-PeCDF 0.3a 2,3',4,4',5-PeCB (#118) 0.00003a

1,2,3,4,7,8-HxCDF 0.1 2',3,4,4',5-PeCB (#123) 0.00003a

1,2,3,6,7,8-HxCDF 0.1 2,3,3',4,4',5-HxCB (#156) 0.00003a

1,2,3,7,8,9-HxCDF 0.1 2,3,3',4,4',5'-HxCB (#157) 0.00003a

2,3,4,6,7,8-HxCDF 0.1 2,3',4,4',5,5'-HxCB (#167) 0.00003a

1,2,3,4,6,7,8-HpCDF 0.01 Mono-o

rtho P

CB

2,3,3',4,4',5,5'-HpCB (#189) 0.00003a

1,2,3,4,7,8,9-HpCDF 0.01

PCD

F

OCDF 0.0003a a Revised in the 2005 TEF re-evaluation. b #-numbers represent the IUPAC numbering of the 209 PCB congeners.

Related compounds not addressed in this thesis A wide variety of organic and inorganic combustion by-products can be found in MSW incineration flue gases (23). Some of these species are not discussed in depth in this thesis but should be mentioned. Polychlorinated biphenyls (PCB), notably the planar and laterally chlorinated congeners, possess dioxin-like properties. Cl-PAH (chlorinated polycyclic aromatic hydrocarbons) and chlorinated paraffins have recently attracted scientific opinion (24-26). The structural similarities between Cl-PAH and PCDD/Fs suggest that Cl-PAHs may have dioxin-like properties, and recently a study using a yeast assay system reported on eighteen Cl-PAHs with Ah-receptor mediated activity (27). The sulfur analogues of PCDDs and PCDFs, polychlorinated dibenzothianthrenes and dibenzothiophenes, and brominated, alkylated and other substituted aromatics (such as polybrominated diphenyl ethers) may also be of interest in formation studies due to their structural similarity to PCDD/Fs.

6


Some of these compounds may, like PCBs, possess dioxin-like properties and should therefore be studied further to evaluate their potential harm. Some may possibly be added to the list of dioxin-like compounds. The tri- to deca-chlorinated PCBs were analysed in some of the studies that comprise this thesis, however due to temporary analytical interferences the data were not reliable and were thus excluded.

7

3. THE COMBUSTION PROCESS Combustion is a complex sequence of exothermic chemical reactions involving a fuel and an oxidizing agent, i.e. an oxidation process in which chemical energy is transformed to heat. The combustion process is the result of a large number of reactions, many of them involving radical species. Currently the reaction mechanisms are only partially understood, mainly due to the large number of reactions and their complex nature. Ideally, the combustion of an organic fuel results in the formation of only carbon dioxide and water, as is exemplified by the combustion reaction of methane (R1) as follows:

[R1] OHCOOCH 2224 22 +→+

Depending on the combusted fuel composition, different by-products are formed, such as sulfur dioxide and hydrogen chloride. If the oxygen supply is insufficient then complete combustion cannot be obtained. Incomplete combustion conditions may also be induced by poor mixing of the flue gas or quenching of the combustion process by a heat sink such as a solid surface.

The three main parameters affecting combustion efficiency are often referred to as ‘the three T:s’, namely (i) Temperatures high enough to sustain the combustion reaction, (ii) Time sufficient for the reactions to occur, and (iii) Turbulence of air and combustion gases. Under realistic combustion conditions, 100% combustion efficiency is never obtained and soot, uncombusted fuel fractions and products of incomplete combustion (PICs) are always generated to some extent. This is also the case in MSW combustion systems, in which incomplete combustion conditions are partly related to the heterogeneous nature of MSW (28). The fluctuating nature of full-scale combustion processes entails that experimental equipment used to simulate combustion conditions must also allow fluctuations, but within strict limits, to minimise incomplete combustion.

Implications for formation of PCDD and PCDF In the presence of chlorine, chlorinated combustion by-products are formed, such as dioxins and dioxin-like compounds. Several studies have successfully correlated combustion process variables to PCDD and PCDF emissions (29-32). Incomplete combustion conditions have been found to strongly promote the formation of chloroaromatic compounds (33-35) and many studies have shown a positive correlation between PCDD/F formation and CO levels (36,37).

9

THE COMBUSTION PROCESS

Incomplete combustion restricts the oxidation process and thus some carbon does not fully oxidize to carbon dioxide but forms carbon monoxide and organic carbon species of different types. Aliphatic compounds are believed to undergo ring closure to form aromatic species such as PAHs (polycyclic aromatic hydrocarbons). These may then agglomerate and participate in the build-up of residual carbon structures, or soot (38,39). Fly ash is an extremely complex matrix composed of particles ranging from 1-100 µm in size. It consists of the incombustible fraction of the fuel (such as silicates, metals and metal oxides) and, if any, residual carbonaceous material. The composition of the fly ash particles is thus highly dependent on the fuel composition, but also depends on the type of incineration facility and the combustion conditions.

When assessing the impact of CO upon the formation of dioxin-like compounds, it should be noted that the average CO concentration seems to be of less significance than the occurrence and frequency of CO-peaks (36). Previous studies (32,34,35) have demonstrated that one or a few CO-peaks, although limited in height and duration, exert a substantial influence on PCDD/F levels and create favourable formation conditions for an extended time period. This phenomenon is referred to as a ‘memory effect’, and is probably related to soot deposits formed during the incomplete combustion event (33). Combustion experiments conducted during normal, transient and post-transient conditions showed that the elevated PCDD/F concentrations induced by the transient combustion decreased with time (35), which supports a correlation with deposited soot.

Carbonaceous deposits may affect PCDD and PCDF levels in two ways, firstly by supplying essential carbon for PCDD/F formation (40) and secondly by adsorption to particulate deposits followed by desorption, affecting the partitioning between gaseous and solid phase (41). Thus the presence of dioxins in flue gases is not solely a question of formation and/or destruction, but also adsorption and desorption processes.

10

4. DIOXIN FORMATION AND DEGRADATION Dioxins and dioxin-like compounds detected in MSW effluents may originate from the waste and pass through the combustion process intact, or they may be formed from precursors in the combustion and/or post-combustion zone. The first alternative is highly unlikely because combustion zone temperatures exceed temperatures shown to be sufficient for destruction of PCDDs and PCDFs (42). The levels of PCDDs and PCDFs at temperatures above 600°C are low (43) but they increase as the temperature decreases along the post-combustion zone (44,45), indicating that they are post-combustion zone products. This has also been confirmed by laboratory-scale reactor studies that form part of this thesis (Paper I), in which PCDD/Fs were not detected at 650°C.

The concentrations of PCDDs and PCDFs in the flue gas are the result of formation as well as degradation reactions (Figure 3). At high post-combustion zone temperatures these two processes compete, whereas at lower temperatures formation is favoured (46,47) resulting in the observed concentration increase with decreasing temperatures.

Temperature

Rat

e

Decomposition

Formation

Concentration of PCDD and PCDF

300°C

Figure 3. The competing behaviour of formation and decomposition reactions. Adapted from Wehrmeier et al. (47).

11

DIOXIN FORMATION AND DEGRADATION

Formation pathways Thermal formation of dioxins and similar compounds takes place in the gas-phase (homogeneous reactions) at high temperatures, or over the gas- and solid-phase interface (heterogeneous reactions) at temperatures below 500°C. However, the contribution from homogeneous PCDD/F formation is small, not to say negligible (48), and is primarily located to the combustion zone.

Heterogeneous formation of dioxins proceeds via surface-catalyzed reactions. Although the mechanisms of formation have not been fully elucidated, they are generally divided into two main routes; de novo synthesis and formation from precursors, such as PCPh or PCBz (49,50). De novo synthesis originates from residual carbon present in the fly ash matrix (51,52), or carbonaceous intermediates such as PAHs (53). The contribution to PCDD and PCDF levels from each pathway has been intensely discussed in the scientific literature, with the addition of theories linking the two pathways (54,55). The principal issue discussed has been the origin of carbon and specifically the extent to which the carbon atoms incorporated in PCDDs and PCDFs originate from chemically similar precursors. Secondly, if the suggested definition is adopted of de novo synthesis as breakdown reactions that occur in the carbon matrix of fly ash (56), how would formation of PCDD/Fs via aromatic structures in fly ash particulates be classified?

Whatever the exact definitions, these two pathways should not be regarded as mutually exclusive. Since the formation of dioxins and other chloroaromatic molecules proceeds through numerous reaction steps, some of which can be regarded as de novo synthesis, whilst others fit the definition of precursor-mediated formation, the mechanistic classification itself seems insignificant, and even may be superfluous. Since these concepts have been used previously, their continued application seems appropriate, but with less rigid definitions. The de novo synthesis and precursor routes are discussed more extensively below.

Since the PCNs are chemically closely related to PCDDs and PCDFs, improved understanding of PCN formation may enhance our knowledge of the formation pathways of dioxins and furans as well. However, studies of PCNs are sparse when compared to PCDDs and PCDFs. A possible link between PCN and PCDF formation was recently suggested by Kim et al. (57). PCNs were observed to be formed from dichlorophenols, possibly involving an intermediate structure similar to that of PCDFs. PCNs were also reported to be formed from PAHs in a de novo-like pathway (53,56).

Formation from precursors A wide range of combustion by-products may act as precursors for the formation of dioxins and dioxin-like compounds. These include aliphatic compounds, monocyclic aromatics with or without functional groups and bicyclic or polycyclic aromatic compounds. Precursors may be present in the fuel and may be formed during combustion or in the post-combustion zone via

12


multi-step reactions, including ring-closure and chlorination of aliphatic compounds (58,59). Coupling reactions, in which the formation of the tricyclic ring structure of PCDD/Fs is one of the key reaction steps (60) are believed to proceed on active sites on fly ash and to be catalyzed by metals and/or metal oxides (58,59,61).

The most plausible precursors in formation of PCDDs and PCDFs are chlorophenols and chlorobenzenes. These compounds are commonly found together with PCDD/Fs in incinerator effluents and have therefore been considered as indicator compounds for the formation of dioxins (21). In 1987, Karasek & Dickson (62) first demonstrated the formation of PCDDs from pentachlorophenol in fly ash. Many studies have subsequently reported the generation of PCDDs and PCDFs from chlorophenol condensation (63-66).

OH

Cl

Cl

Cl

Cl

O

Cl

Cl

+O

Cl

ClOH

Cl

Cl

Cl

+ ClO

O

Cl

Cl

ClCl

ClRing closure intermediate

Phenoxy ether intermediate

O

OCl

ClCl

Cl

O

OCl

ClCl

Cl

1,3,6,8-TeCDD 1,3,7,9-TeCDD Figure 4. The chlorophenol condensation pathway exemplified by condensation of two 2,4,6-TriCPh entities (radical/molecule). Adapted from Sidhu et al. (63).

Formation of PCDD from PCPhs is generally believed to proceed through dimerization directly to PCDD (67,68). This pathway is exemplified by the suggested pathway in Figure 4, which presents the direct condensation of two 2,4,6-TriCPh entities to form mainly 1,3,6,8- and 1,3,7,9-TeCDD (63). The phenoxy ether intermediate could form 1,3,6,8-TeCDD via simple ring closure or by formation of the second oxygen bridge, but 2,4,6-TriCPh yields equal amounts of 1,3,6,8- and 1,3,7,9-TeCDD, which indicates that the condensation mechanism proceeds via the ring closure intermediate (Figure 4) (63). This intermediate is also referred to as a dioxaspiro-intermediate (69) and is subject to the Smiles rearrangement in which the rings rotate and thus enable formation of both tetra-isomers (63,69). The rearrangement effect increases the number of products of the chlorophenol condensation. Reactions involving dioxaspiro-intermediates do not apply to PCDFs (60).

Phenolic species are relatively easily deprotonated and are therefore largely present in flue gases in their radical form, which is highly reactive. Both radical/radical and radical/molecule reactions have been implicated in the formation of PCDDs as well as PCDFs (63,70). The formation of dibenzo-p-

13


dioxin requires the presence of at least one chlorine substituent in ortho-position on the CPh precursor (71), i.e. the 2- or 6- position (Figure 1E). The formation of dibenzofuran, on the other hand, requires at least one unsubstituted ortho-carbon in the chlorophenol precursor. Thus, unsubstituted phenol gives dibenzofuran, and chlorophenols without chlorine substituents in ortho-positions behave analogously and form chlorinated dibenzofurans. The presence of one ortho-Cl leads to formation of chlorinated dibenzo-p-dioxins and chlorinated dibenzofurans, whereas chlorophenols containing chlorine in both ortho- positions give rise almost exclusively to PCDDs. Therefore, the mixtures of chlorophenol congeners typically present in MSW incineration flue gases generate a combination of PCDDs and PCDFs with varying degree of chlorination (65).

De novo synthesis De novo synthesis involves formation of dioxins and dioxin-like compounds from carbon structures in the fly ash particles entrained in the flue gas and deposited on surfaces in the post-combustion zone. Even though the distinction between precursor-mediated formation and de novo synthesis is imprecise, carbon in a particulate form is considered essential for the de novo synthesis of dioxins, and formation does not proceed via gas phase intermediates.

According to Stieglitz et al. (72) two basic reactions are involved in the de novo synthesis of dioxins from carbon:

• transfer of inorganic chlorine to the macromolecular carbonaceous structure and formation of carbon-chloride bonds

• oxidative degradation of the structure Thus de novo synthesis may be defined as the chlorination and oxidation, although partial, of particulate carbon structures.

The mechanism of formation of PCDDs and PCDFs from carbon is unclear and alternative routes have been proposed with wider definitions of de novo synthesis. The alternatives proposed include: (i) direct release from the carbonaceous structure (suggesting that the PCDD/F structures already exist in the carbon matrix) (73); (ii) formation from particulate carbon via surface-bound chlorinated or non-chlorinated aromatic intermediates such as PAHs or phenolic compounds (56,74,75); (iii) formation resulting from precursor molecules that are adsorbed or chemisorbed onto the carbon surface, implying that formation from carbon may be an artefact (76,77). The de novo synthesis of PCDDs and PCDFs from aromatic compounds also requires an oxychlorination step, i.e. the decomposition or cleavage of a polyaromatic structure (53,56). Several publications have identified de novo synthesis as the predominant formation pathway for the generation of the PCDFs, whereas the PCDDs seem to originate largely from precursors such as chlorophenols with a limited contribution from de novo-type formation (74,78). This results in a PCDD/PCDF ratio < 1 which is considered to be characteristic of de novo synthesis and is commonly observed in MSW incineration flue gases (79). Precursor-dominated formation, on the other hand, is accompanied by a PCDD/PCDF ratio > 1.

14


As previously mentioned, there is a significant overlap between the definitions of de novo and precursor formation of dioxins. Luijk et al. (55) linked the two major pathways by suggesting initial de novo formation of chlorophenols from carbon in fly ash followed by chlorophenol condensation reactions. The pathway suggested by Taylor and co-workers (54) on the other hand, suggests the chlorination and molecular growth of aliphatic species adsorbed to a catalytic site as a crucial first step, followed by their release to the gas phase.

Chlorine substitution Regardless of the formation pathway, the resulting PCDD and PCDF structures may be further chlorinated and/or dechlorinated in the post-combustion zone. Since PCDD/F toxicity is related to 2,3,7,8-chlorination, enhanced understanding of the chlorination reactions could enable detoxification measures. More than 90% of the chlorine in the fuel is released as hydrogen chloride, HCl, in the flue gas. HCl cannot act directly as a chlorinating agent in reactions with aromatic species and must first be oxidized to chlorine gas (Cl2) (80), which can subsequently chlorinate aromatic molecules (81).

The conversion of HCl is believed to proceed through the Deacon process (82), which is a two-step mechanism (R2 and R3) catalyzed by copper species, e.g. CuCl2 (82,83):

222 2

1 ClCuOOCuCl +→+ [R2]

OHCuClHClCuO 222 +→+ [R3]

The Deacon process produces an optimum yield of Cl2 at about 400°C but at 300°C the yield is almost zero (84). The Cl2 then participates in chlorination of aromatic molecules (R4):

[R4] HClArClClClArH +→⋅⋅+

Today there is no ambiguity regarding the status of copper chloride species (CuCl and CuCl2) as the most efficient Deacon catalysts. This has been shown repeatedly (83-86), with maximum efficiency in the presence of divalent copper chloride (85,86). Besides the copper chlorides, elemental copper and the oxide species, CuO and Cu2O, have also been found to display catalytic activity for the Deacon process (83,84). Copper compounds have also been identified as catalysts in chlorophenol condensation reactions (64) and chlorination/ dechlorination reactions (87,88), which emphasizes their importance in dioxin formation. Very small amounts of copper in the waste (0.007%) have been shown to be sufficient to catalyze PCDD/F formation (87).

In addition, several other metals and metal oxides have been found to act as catalysts for dioxin formation, including iron (51,58,84,85), aluminium (58), nickel (85,86), lead (51,85), zinc (51,85) and chromium (86), although all with significantly lower catalytic efficiency than copper.

However, the chlorination pathways are still under debate, and some researchers have questioned the importance of the Deacon reaction. An

15


alternative chlorination route was proposed, in which chlorination of aromatic compounds proceeds directly by reaction with CuCl2 (51,54), as in the following sequence (R5 and R6), suggested by Stieglitz et al. (51):

ClICuArHClClIICuArH )()( 2 +⋅→+ [R5]

HClClICuArClClIICuArHCl ++→+⋅ )()( 2 [R6]

Such a direct transfer of chlorine is in accordance with the ligand transfer oxidation mechanism described by Nonhebel (89). The divalent Cu(II) is reduced to monovalent Cu(I), but is readily re-oxidized in the presence of oxygen and thus available to repeat the reaction cycle (51).

Degradation The degradation of dioxins and dioxin-like compounds is the result of two different types of reactions, namely dechlorination/hydrogenation and decomposition, both of which are catalysed by fly ash. The low-temperature (< 300°C) dechlorination/hydrogenation of PCDD and PCDF by waste incineration fly ash (46) and copper powder (88) was described by Hagenmaier and co-workers over two decades ago. The resulting PCDD isomer distribution patterns indicated that dechlorination was directed to the aromatic ring possessing the highest number of chlorine substituents, preferably forming laterally substituted congeners (88). In the same year (1987), Vogg et al. (40) observed the ready decomposition of PCDD/Fs at temperatures exceeding 500°C and, in the absence of oxygen, at temperatures as low as 300°C. Furthermore, it was suggested that PCDD/F degradation is initiated by dechlorination and continues with cleavage of the oxygen bridges.

O

98

7

6 4

3

21

ClxCly

Figure 5. The dibenzofuran backbone structure with the sterically crowded so-called bay-sites (1,9-positions) encircled.

Weber and co-workers have investigated PCDD/F degradation in several studies (90-92), and have identified copper as the most catalytically efficient element with the highest dechlorination potential found for Cu2O (91). The degradation behaviours of PCDDs and PCDFs were found to differ. Eight hours at 310°C and 10% O2 exerted a greater effect on PCDDs than on PCDFs, suggesting that PCDD degradation proceeded faster and was a result of both dechlorination and decomposition, whereas PCDFs seemed to be mainly influenced by dechlorination (90). Degradation was accompanied by a 40% increase of the total TEQ due to the increased abundance of laterally substituted congeners, or merely a result of higher relative abundance of the

16


tetra- and penta-chlorinated dechlorination products (90). Dechlorination of PCDFs seemed to commence at the sterically crowded 1,9-positions located at the so-called bay-sites of the dibenzofuran backbone (Figure 5) (91). The extent of dechlorination of α- vs. β–positions was also evaluated, but no difference was observed between PCDDs and PCDFs treated with fly ash (92).

Such research is highly useful to develop technical applications with environmental benefits. For example, a thermal degradation technique has been developed for detoxification of MSW incineration filter ash (5). This gave a 99% reduction of the total PCDD/F concentration and of the TEQ (6). In addition, chemically related compounds such as PCBs and HxCBz were also removed. The chemical composition of the ash had a major impact on the degradation efficiency, and this factor largely influenced the temperature needed for total degradation (5).

Reduction strategies Most emission control techniques are mainly concerned with ‘end of pipe‘ or secondary measures, such as wet/dry scrubbing, fabric filtration, and activated carbon adsorption. Such approaches address the removal of PCDD/F from flue gases after formation using flue gas-cleaning devices. Primary control measures, on the other hand, include actions to prevent PCDD/F formation, such as optimization of the combustion process (37,93), reduction of HCl levels in the flue gas (94), minimisation of chlorine and metal catalysts in the fuel (95;96), and source sorting and/or pre-treatment of the waste (97). Catalytic destruction (46,98) and the optimisation of combustion technology differ from flue gas cleaning methods as they address the minimization of PCDD/F formation.

An approach to reduce the formation of dioxins and their precursors, and thus avoid the problems associated with the presence of these substances in fly ash, is the addition of inhibitors to the waste or their introduction into the post-combustion zone. Several types of inhibitors have been studied, particularly nitrogen and sulfur compounds such as urea (96), elemental sulfur (99), sulfur dioxide (100-102), and sulfur-containing coal (101,103). However, the chemical species responsible for the inhibitory effect may not be the added compound itself but one of its decomposition products.

The mechanism of inhibition falls outside the scope of this thesis, but demonstrates the importance of fully elucidating the mechanisms of formation and degradation of dioxins. Inhibition of dioxin formation in full-scale systems would be particularly beneficial if the inhibitor could be co-combusted with the waste, e.g. by using sulfur-containing coal, or car tyre-waste. It has not yet been determined whether other potentially harmful compounds are formed instead of the inhibited dioxins, in which case the inhibition of dioxin formation may not be environmentally beneficial.

17

5. EXPERIMENTAL SECTION

General overview A brief summary of the combustion experiments and papers that support this thesis is presented below. A more extensive description of the experiments is given in the respective papers, and an additional overview listing each individual sequence is presented in Appendix 1.

Paper I Four different post-combustion zone quench time profiles were investigated, referred to as the baseline profile (400-200°C), low-temperature profile (300-100°C), high-initial temperature profile (450-200°C), and high-temperature profile (460-260°C). Flue gas samples were collected at three locations in the post-combustion zone and these were analysed for mono- to octa-chlorinated PCDD/F.

Paper II The high-initial temperature profile from Paper I was further investigated (referred to as ‘reference samples’), and compared to an experimental run in which dibenzo-p-dioxin was injected into the post-combustion zone in order to promote chlorination. Samples were collected at three locations in the post-combustion zone (450, 300, and 200°C), and analysed for PC0-8DD, PC0-8DF, PC2-

6Bz, and PC1-5Ph.

Paper III An approach was used similar to that of Paper II with sampling at three locations in the post-combustion zone (450, 300, and 200°C) before and during injection. The injected compound was naphthalene and the purpose was to study formation of unchlorinated to octa-chlorinated PCNs. In addition, samples were collected during an incomplete combustion event.

Paper IV The PC1-7DD and PC1-7DF isomer distribution patterns in the reference samples were correlated with physicochemical properties and chlorine substitution descriptors using principal component analysis, PCA, and the recently developed bidirectional orthogonal partial least squares projection method, O2PLS.

19

EXPERIMENTAL SECTION

Combustion experiments

The laboratory-scale fluidized bed reactor In all combustion experiments a 5kW laboratory-scale fluidized-bed incinerator (Figure 6) was used. The reactor consists of a sand bed (primary-combustion zone), freeboard (secondary-combustion zone), convector (post-combustion zone), and air pollution control system, which included a cyclone, metal particle filter, textile filter, wet scrubber and active carbon filter. The convector tubes were equipped with electrical heaters, allowing convector temperatures to be easily varied, and thus samples to be collected from the fixed sampling ports at different temperatures and residence times.

I4

I3

I2

T1

T2

T6

T7

T8

T9

T10 T11

T12

P1P2

P4

P3

P5

P6

Secondary air

Primary airPropane

O2 measurementFTIR online monitoring

P7

Fuelfeeder

Induced draft fan

C

D

A

H1

H2

H3

H4

H5

H6

H7

H8

H9

Freeboard

BConvector

Bed

P8

I1

T5

T4

T3

Injection

Ice

Water Ethylene glycol

PumpFilter

00629207

Adsorbent

Figure 6. The laboratory-scale fluidized-bed reactor, with sampling ports denoted P1-P8, electrical heaters H1-H9, thermocouples T1-T12, and injection ports I1-I4. The air pollution control system consists of a cyclone (A), a metal and textile filter (B), a wet scrubber (C), and an active carbon filter (D). Flue gas samples were collected at P3, P4, and P7. Injections were made at P2.

The reactor was constructed to simulate the behaviour of a full-scale waste incinerator and has been described by Wikström et al. (104). Prior to the experiments presented in this thesis it was modified as follows:

• the duct connecting the freeboard with the convector was replaced with a high-temperature grade of stainless steel (253 MA, Avesta Sheffield, Sweden), with an operating temperature of up to 1150°C,

• new injection ports were installed (denoted I1-I4, Figure 6), enabling the injection of precursor compounds or additives in liquid or gaseous form directly into the combustion zone or into the flue gas stream,

20


• 12 N-type sheathed thermocouples were mounted directly into the flue gas duct (T1-T12), to improve the precision of temperature monitoring,

• a longer tube was installed to connect the fuel box screw to the bed inlet - this improved the evenness of fuel feeding,

• the operating system was updated and redesigned using the software LabVIEW 7 Express (National Instruments Corporation, Austin, TX, USA).

All operating parameters and reactor temperatures were continuously measured and logged using LabVIEW 7 Express at selected intervals for subsequent evaluation. For on-line monitoring of flue gas concentrations, an ABB Bomem 9100 Gas Analyzer based on a Fourier-Transform Infrared (FTIR) spectrometer was connected to the end of the first convector section. H2O, CO2, SO2, NO2, CO, NO, HCl, NH3, N2O, and CH4 were continuously measured and 30-second average values were logged. The O2 levels were monitored by an on-line ScanTronic OC 2010 oxygen measuring probe with a zirconium dioxide cell taking readings every second.

Table 3. Composition of the artificial MSW used in all experimental runs.

Material Amount [%] Element Concentration [% dw] Paper confetti 36.0 C 45 Wheat starch 15.5 O 31

Polyethylene powder 10.5 H 6.3 Sawdust 10.5 Si 3.1 Gelatine 7.5 Ca 2.7

Peat 5.0 N 1.4 Rapeseed oil 5.0 Fe 1.0

Silica gel (SiO2 Ø 0.06-0.2 mm) 3.0 Cl 0.7 Sand (Ø 150-210 µm) 2.5 Al 0.2 Kaolin (Al2Si2O5(OH)4) 1.7 S 0.07

Iron powder 1.2 Na 0.03 Polyvinylchloride (PVC) powder 0.8 K 0.02 Calcium chloride (CaCl2×6H2O) 0.7 Cu 0.007

Copper acetate (Cu(CH3COO)2×H2O) 0.02

The standardized artificial MSW To maximise the reproducibility of the experiments, a standardized, artificial MSW fuel was prepared using the materials listed in Table 3 (35). The fuel was thoroughly mixed and formed into MSW fuel pellets with a diameter of 6 mm. The dry weight (dw) of the artificial MSW was 95%, with an ash content of 17% dw, and a calorimetric heating value of 19 MJ/kg dw. The material and elemental composition of the artificial MSW (Table 3) was similar to that of typical Swedish MSW. Previous experiments with a similar artificial MSW resulted in similar PCDD/F levels and homologue profiles when compared to those typically observed in full-scale MSW combustion (105).

21


Combustion conditions All experiments commenced by pre-heating the reactor using propane combustion for two hours. MSW combustion was initiated and maintained at an average fuel feeding rate of 0.86±0.02 kg/h. The primary air flow was set to 110 l/min and the secondary air flow was set to 30 l/min. The bed temperature varied between 800-850°C, freeboard temperatures were 800°C, and convector temperatures declined from 650°C to 260 or 100°C, depending on the quenching profile used. The reactor was thoroughly cleaned between experiments and the bed material was replaced with washed and sifted sea sand (125-250µm). The ash deposited in each of the reactor zones (bed, convector and cyclone) was collected and weighed after each run. The flue gas residence times from the inlet of the convector to each of the sampling ports were calculated based on fuel feed, air supply, reactor, post-combustion zone temperatures and the internal volume of the combustion system. Steady state was reached after approximately four hours of MSW combustion.

The measured concentrations of combustion gases (Table 4) were generally stable, and the low CO levels (below 3 ppm on average) indicated that the combustion efficiency was high. This was the case for all runs except the ‘poor combustion event’, which is discussed with regard to PCN formation in Paper III. In addition, the variations in combustion conditions between runs were minor, as shown by the low coefficients of variance. The non-detectable sulfur dioxide concentration in the flue gas (Table 4) were due to the low levels of sulfur in the MSW fuel.

Table 4. Experimental combustion conditions during sampling, average values calculated from the three reference samples.

Parameter Average Coefficient of variance [%]

H2O [%] 7.4 1.8 O2 [%] 10.1 2.0 CO2 [%] 8.1 2.6 CO [ppm] 2.9 7.5 NO2 [ppm] 45.2 8.3 NO [ppm] 485 3.5 N2O [ppm] 50.1 2.0 HCl [ppm] 191 4.7 SO2 [ppm] n.d.a NH3 [ppm] <1 CH4 [ppm] n.d.a a n.d. = not detected

Flue gas sampling Flue gas samples were collected simultaneously at sampling ports P3, P4, and P7 after four hours of MSW combustion (Figure 7). Sampling was performed isokinetically using the cooled probe polyurethane foam plug (PUFP) sampling technique according to the standard method EN 1948:1 (106). Three different types of probes were used depending on the sampling temperature; pyrex glass

22


at 400-100°C, quartz glass at 460°C and 450°C and titanium at 640°C. For sampling at 640°C and 460°C, an additional cooled-probe was used, placed outside the existing water-cooled probe. The outer cooled-probe was filled with a mixture of ice and salt (NaCl) to maintain a temperature of around -12°C. This enabled fast cooling of the hot flue gases. New glass probes were used for each sampling event and the titanium probe was heated at 550°C for 390 minutes and rinsed with toluene before reuse. The sampling train is presented in Figure 6 and consisted of a glass probe, a collector filled with water, an impinger filled with ethylene glycol, and a glass housing for two PUFPs with an aerosol filter placed between, as described elsewhere (107). A leak check of the sampling equipment was performed by comparing the O2 level in the flue gas leaving the pumps with the O2 level in the post-combustion zone. The concentrations of analytes were calculated on molar basis and were normalized to 1 atm, 0°C, dry gas (dg), and 11% O2. Field, laboratory, and cleanup blanks were analyzed.

0 1 2 0 1 2 3 4 5 6 7 8 9

Time (hour) Sampling Sampling

Start dibenzo-p-dioxin injectionStop injection

Propane MSW combustion

Combustion terminated

0 1 2 0 1 2 3 4 5 6 7 8 9

Time (hour) Sampling Sampling

Start naphthalene injectionStop injection

PropaneMSW combustion


Start solvent injectionSamplingSampling

Stop solvent injectionSampling

Paper III

Paper II

Time (hour)0 1 2 0 1 2 3 4 5

Sampling

Propane MSW combustionPapers I and IV


Time (hour)

Figure 7: Timelines for the experimental runs in Papers I-IV.

Injection procedure In order to demonstrate the influence of chlorination reactions on the formation characteristics of dioxins and dioxin-like compounds, and to determine the importance of backbone structures already present in the flue gases, eight

23


compounds were selected for the injection study. The selection was made based on the objective to identify and study important precursors and/or intermediates with specific emphasis on the chlorination process. The compounds in Table 5 were used. The unchlorinated parent compounds dibenzo-p-dioxin (DD) and dibenzofuran (DF), and the fully chlorinated analogues octachlorodibenzo-p-dioxin (OCDD) and octachlorodibenzofuran (OCDF) were selected to represent the two extremes of the degree of chlorination. Unsubstituted biphenyl and three PAHs with dioxin-like backbone structures (naphthalene, phenanthrene and fluorene) were included in the study to enable an estimation of the contribution from other polyaromatics to the formation of dioxins. Separate injection solutions were prepared containing each species in n-heptane (Merck, pro analysi) at concentrations intended to obtain a flue gas concentration approximately 10 times higher than the normal flue gas levels of the respective compounds (Table 5). The purities of dibenzo-p-dioxin, naphthalene, and n-heptane were checked by GC-LRMS (full scan) analysis using an Agilent 5975 MSD low-resolution mass spectrometer (Agilent Technologies, Inc., USA) fitted with a J&W Scientific DB-5MS (30 m, 0.25 mm i.d., 0.25 µm) capillary column. No impurity exceeding a S/N (signal-to-noise) ratio ≥ 3 was detected.

Table 5. Concentrations of the solutions used in the injection experiments.

Injected compound

DD DF OCDD OCDF Biphenyl Naphthalene Phenanthrene Fluorene

cINJ SOL [µg/µl]

0.30 3.0 0.03 0.03 7.5 30 15 3.0

cFLUE GAS

[µg/m3] 0.11 1.1 0.011 0.011 2.8 11 5.6 1.1

FLUEGAS

DUCTA B C D E F

PURIFIEDAIR

Figure 8. Schematic diagram of the injection equipment used in Papers II and III, consisting of a metal holder (A), ceramic liner (B), glass connector (C), nebulizer (D), syringe needle (E), and syringe (F).

Each compound was injected in a separate combustion experiment, conducted on separate days with thorough cleaning of the reactor between experiments. To minimize carry-over effects between experiments, combustion continued

24


with elevated post-combustion zone temperatures for two hours after sampling was completed. Injection was performed using a gas-tight 10 µl syringe (Hamilton) mounted in a microdialysis pump (CMA/100, CMA microdialysis) as illustrated in Figure 8. The syringe injected at a rate of 3 µl/h into the nebulizer (TR-30-A3, Meinhard) with a 100 ml/min flow of purified air. The combination of a low injection rate and the high flow in the nebulizer ensured complete volatilization, and the gas-phase mixture was directly injected into the post-combustion zone at 650°C (P2, Figure 6) through a ceramic liner. After 30 minutes of continuous injection, three flue gas samples were collected simultaneously at 450, 300, and 200°C (P3, P4 and P7, Figure 6). The performance of the micro-injection pump was tested using n-heptane to validate the injection rate and to ensure that no condensation occurred. A solvent control experiment was also performed in which n-heptane was injected, following the same procedure. Two successive flue gas samples were collected at 300°C (P4, Figure 6) and verified that the injection of n-heptane had no significant influence on the concentrations of the target analytes. In addition, no effect was observed for the n-heptane injection upon the on-line monitoring of combustion gases.

Extraction and clean-up After addition of IS (internal standards), the flue gas samples were filtered, soxhlet extracted and evaporated. All laboratory work was performed according to validated methods which are fully described elsewhere (108,109).

Half of the resulting concentrated extract was prepared for the determination of PC1-8DD, PC1-8DF, and PC3-8N according to the following cleanup scheme. The extract was applied to a multilayer silica column and eluted with n-hexane directly onto an alumina column. The alumina column was eluted with n-hexane (which was discarded), and was then eluted with dichloromethane:cyclohexane 1:1 (v:v). 40 µl of n-tetradecane (Fluka, puriss, olefin free) was added as a keeper solvent prior to evaporation in a TurboVap II concentration workstation (Zymark Corp.). A 1:11.7 (w:w) mixture of AX21 carbon (Anderson Development Company) and Celite 545 (Fluka) was used for fractionation. The carbon:celite column was pre-washed with toluene, dichloromethane:methanol:toluene 15:4:1 (v:v:v), dichloromethane:n-hexane 1:1 (v:v), and finally n-hexane. The samples were applied and eluted with dichloromethane:n-hexane 1:1 (v:v), and then back-eluted with toluene. 40 µl of n-tetradecane was added to the toluene fraction which was then evaporated. Finally recovery standards (RS) were added.

25 vol% of the concentrated extract was liquid-liquid extracted using 3×1 ml of 0.5M NaOH (aq). The resulting organic phase was prepared for the determination of non-chlorinated DD/Fs, PC2-6Bz, naphthalene and PC1-2N by clean-up on a 10% deactivated silica column, and elution with cyclopentane. 1 ml of toluene was added, the cyclopentane was evaporated and RS was added. The aqueous phase was dried on a Na2SO4 column, eluted with cyclopentane, 1

25


ml of toluene was added prior to evaporation and RS addition, and the final extract was analysed for PC1-5Ph.

Analytical procedure Analyses of PC1-8DD, PC1-8DF, PC2-6Bz and PC1-5Ph were performed according to methods used regularly in the laboratory. The analytical procedure for the PCNs was developed specifically for the samples included in Paper III. Furthermore, interferences in the analysis of DD and DF using GC-LRMS required analytical modifications resulting in a GC-HRMS method with a modified GC temperature programme. This approach was successful in separating the target peaks from interfering peaks.

DD, DF, PC1-2N, and PC1-5Ph were analyzed by GC-HRMS using a VG AutoSpec mass spectrometer (Fisons Instruments, VG Analytical, Manchester, UK) and a Hewlett-Packard 5890 gas chromatograph equipped with a J&W Scientific DB-5MS (30 m, 0.25 mm i.d., 0.25 µm) capillary column.

PC1-8DD, PC1-8DF, and PC3-8N were determined by GC-HRMS using a Waters AutoSpec ULTIMA NT 2000D high resolution mass spectrometer (Micromass UK Limited, Manchester, UK) and an Agilent 6890N Network GC System gas chromatograph. PC1-8DD and PC1-8DF were analysed in two injections, first using a Supelco SP 2330 (60 m, 0.25mm i.d., 0.2 µm) capillary column and then with a J&W Scientific DB-5 (60 m, 0.25mm id, 0.25 µm) capillary column. PC3-

8N were analysed using a J&W Scientific DB-5 (60 m, 0.25mm id, 0.25 µm) capillary column. In order to verify the complete elution of PCNs from the carbon:celite column into the second fraction, the first fraction from two of the samples was also analyzed. No peaks corresponding to the target analytes were detected in the first fraction.

PC2-6Bz and naphthalene were analyzed by GC-LRMS using an Agilent 5975 MSD mass spectrometer (Agilent Technologies, Inc., USA) with a J&W Scientific DB-5MS (30 m, 0.25 mm i.d., 0.25 µm) capillary column.

The mass spectrometers were all operated in electron impact ionization/selected ion monitoring mode and analytes were quantified using the isotope dilution technique. Target analytes were identified by comparing (i) their retention times with those of quantification congeners in standard mixtures, and (ii) published PCDD/F (110,111) and PCN (10,112) GC-MS chromatograms. PCN GC-MS chromatograms for one of the reference samples are presented in Appendix 2.

26

6. POST-COMBUSTION FORMATION Dioxins formed during combustion processes display characteristic ‘fingerprints’, typically a PCDD/PCDF ratio < 1 and isomer distribution patterns that contain almost every isomer (113). The maximum formation rates for PCDDs and PCDFs have been observed at 300-400°C and at 400-500°C, respectively (52), which may indicate different formation pathways for PCDDs and PCDFs. The temperature effect on formation maxima is possibly related to low concentrations of chlorophenol precursors at higher temperatures (74) responsible for shifting the promotion of PCDD towards lower temperatures.

Figure 9. Post-combustion quenching profiles for each of the investigated profiles, with temperatures and residence times at sampling points P3, P4 and P7 (Paper I).

The influence of quench profiles In addition to the post-combustion zone temperature, the flue gas residence time has been demonstrated to affect the yields of PCDD/Fs (29). To further investigate the effects of temperature, residence time and the combined influence of these two variables, four different quench/time profiles were investigated. These experiments were reported in Paper I. The profiles comprised the following: falling from 400°C to 200°C (baseline profile), 300°C

27

POST-COMBUSTION FORMATION

to 100°C (low-temperature profile), 450°C to 200°C (high-initial temperature profile) and 460°C to 260°C (high-temperature profile). The temperature profiles are reproduced in Figure 9. The corresponding residence times were 1.4 to 4.4 seconds, 1.4 to 5.4 seconds, 1.3 to 4.3 seconds, and 1.3 to 4.0 seconds, respectively.

Figure 10. PCDD and PCDF levels in the flue gas of the four investigated quench time profiles (Paper I). Error bars represent ±1 standard deviation.

The study focused primarily on the effects of post-combustion zone temperatures and flue gas residence times upon the formation of mono- to octa-chlorinated PCDDs and PCDFs. Flue gas samples were collected simultaneously at three fixed sampling points (P3, P4 and P7, Figure 6), with additional sampling of the flue gas entering the post-combustion zone (P2, 640°C). Flue gas from the four quench time profiles resulted in different total PCDD/F yields (Figure 10) and different homologue profiles (Figure 11) and congener patterns. Thus, the formation pathways seem to be closely linked to the residence time within a specific temperature region. PCDD/Fs were not

28


detected in flue gas samples taken at 640°C. In the baseline profile, the observed PCDD/F concentrations did not increase or decrease significantly below 400°C (Figure 10). It was therefore concluded that the formation of PCDD/Fs was rapid and occurred mainly in the 640-400°C temperature region.

Figure 11. Homologue profiles of mono-octa-CDD/Fs at sampling points P3, P4 and P7 during the four investigated quench time profiles (Paper I): the baseline profile (A), low-temperature profile (B), high-initial-temperature profile (C) and high-temperature profile (D). Error bars represent ±1 standard deviation.

29


The PCDD/F homologue profiles (Figure 11) observed for the baseline and low-temperature profiles were dominated by the Te-HxCDDs and the Mo-TeCDFs. These quench time profiles did not indicate additional chlorination/dechlorination of already-formed PCDD/Fs along the post-combustion zone. The prolonged residence time at 450/460°C obtained in the high-initial temperature and high-temperature profiles resulted in lower PCDD/F yields than were measured in the baseline profile The low yields observed for the high-initial temperature profile were carried all the way through the post-combustion zone.

Recently reported thermochemical equilibrium calculations (114) indicates that the temperature interval 450-500°C is significant for copper speciation, because it is a transition zone between solid and gaseous copper species. A prolonged flue gas residence time at 450°C may prevent or delay the conversion of gas-phase copper to the catalytically potent CuCl2 (s), and may also result in reduced levels of carbon backbone structures available for formation of PCDD/Fs. The PCDD and PCDF homologue profiles at 450/460°C were dominated by the Mo-PeCDDs and the Mo-DiCDFs. These changes in homologue profile compared to baseline may be due to a reduced rate of chlorination or more rapid dechlorination. Since no notable differences in the congener patterns were evident when comparing the high-initial temperature/high-temperature and baseline profiles, a decreased rate of chlorination of PCDD/Fs is the probable reason. The differences in quench time between 450/460°C and 200/260°C resulted in different PCDD/F homologue profiles and congener patterns in the high-initial temperature and high-temperature profiles. At 260°C, the PCDD/F homologue profiles contained a higher proportion of the more highly chlorinated homologues. Changes in congener patterns were also observed. These factors suggest altered routes of formation, in which subsequent chlorination of PCDFs is plausible.

PCDD concentrations significantly increased between P4 (360°C) and P7 (260°C) in the high-temperature profile (Figure 10) and this was also reflected in the homologue profiles (Figure 11). When temperatures were raised to 260°C in this region of the post-combustion zone, one or more formation pathways seemed to be promoted that enhanced the formation of PCDDs. Interestingly, PCDD concentrations at 260°C declined with run order, implying that the observed increase was related to deposits on the convector walls formed during previous experimental runs in which combustion process disturbances formed soot. Even though the post-combustion zone was vacuumed and cleaned with compressed air after each experimental run, a thin covering of fine particles always remained on the convector walls. Thus, the decline of PCDD/F levels with run order may be due to the consumption of carbon-based structures and/or metals and metal oxides with time. This is in agreement with a study by Lee et al. (115), who found that PCDD/F concentrations, formed due to soot and copper deposits in a quartz reactor, declined with time. The effect observed at 260°C can be explained, at least partially, by a shift in the equilibrium between iron chlorides and iron oxides in the wall coatings. At temperatures exceeding 250°C, iron chlorides may be conversed and/or

30


volatilized, in some cases accompanied by release of Cl2 (116,117), which may then participate in chlorination reactions.

The high-temperature profile samples collected at 360°C and 260°C contained a greater abundance of congeners that have been shown to form from, or correlate with, 2,4,6-trichlorophenols (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) (63,64). This implies that chlorophenol condensation reactions were promoted in this temperature/residence time profile.

Formation characteristics during the high-initial temperature profile The quench time-profile that resulted in the lowest PCDD concentrations, i.e. the high-initial temperature profile, was further investigated in Papers II and III. Homologue profiles and isomer distribution patterns were evaluated for PCDDs, PCDFs, PCNs and the two principal precursors, PCBz and PCPh.

The total concentrations of PCDFs, PCNs, PCBzs, and PCPhs did not change significantly between 450 and 200°C, suggesting that their molecular backbone structures are predominantly formed at or above the first sampling temperature (450°C). However, the maximum concentration of PCDDs was observed at 300°C, indicating continued PCDD formation below 450°C. This agrees with the previously mentioned maximum formation temperatures reported by Wikström et al. (52) located between 500-400°C for the PCDFs and 400-300°C for PCDDs. The lower optimum temperature for PCDD formation is possibly related to their lower aromatic stability and the lower abundance of PCDD precursors at high temperatures (52). At 200°C the PCDD concentrations decreased slightly (Figure 10 and 11), possibly due to adsorption effects affecting mainly the least volatile compound class, namely PCDDs (118).

Homologue profiles The PCDF and PCN homologues decreased in concentration with increasing degree of chlorination and were stable between 450 and 200°C (Figures 11C and 12). However, the PCN profile displayed an even more pronounced domination of the mono- to tri-chlorinated homologues, with MoCN constituting approximately 50% of the total concentration of PCNs. The distinctive domination of the least-chlorinated PCN homologues and the successive decline in abundance with each additional degree of chlorination is strongly suggests a PCDF and/or PCN formation route involving successive chlorination of the unsubstituted aromatic backbone and its chlorination products.

The PCDD profile contained higher proportions of tetra- to hepta-chlorinated homologues and displayed a slight shift towards higher chlorinated homologues in flue gas samples collected along the post-combustion zone (Figure 11C). This shift may be interpreted as a larger influence of chlorination reactions in PCDD formation within the investigated temperature region which is contradictory to previous findings (119) claiming that dibenzofurans are

31


more subjected to chlorine substitution. However, the observed shift may alternatively indicate a temperature-related difference in maximum chlorination rate for PCDD and PCDF, e.g. chlorination of the respective species mainly occurring within different temperature regions.

0

1

2

3

4

5

DiCBz TriCBz TeCBz PeCBz HxCBz

PC

Bz

[nm

ole

m-3 ]

0

1

2

MoCPh DiCPh TriCPh TeCPh PeCPh

PC

Ph

[nm

ole

m-3 ]

0

100

200

MoCN DiCN TriCN TeCN PeCN HxCN HpCN OCN

PC

N [p

mol

e m-3

]

450°C

300°C

200°C

Figure 12. PC0-8N, PC2-6Bz, and PC1-5Ph homologue profiles at 450°C, 300°C, and 200°C. Error bars represent ±1 standard deviation. The mechanistic correlation of PCBzs to PCDDs and PCDFs is unclear but the present work showed that the PCBz mimicked the PCDF and PCN homologue profiles. All three displayed a successive decrease in abundance for each additional degree of chlorination (Figure 12). Between 450 and 300°C the PCBz profiles seemed to shift towards the more highly chlorinated homologues, suggesting that additional chlorine substitution reactions were occurring within this temperature interval. The PCPh were largely dominated by the monochlorinated homologue (Figure 12). In addition, an increase in the MoCPh concentration along the post-combustion zone was observed. This may indicate a dependency on temperature and/or residence time for the chlorination of unsubstituted phenol.

32


Isomer distribution patterns The PCDD isomer distribution patterns seemed to be influenced largely by chlorophenol condensation and to some extent by chlorination reactions. The monochlorinated homologue was dominated by 2-MoCDD (Table 6), which is the most thermodynamically stable isomer (120) and thus the MoCDD product most likely to be formed by chlorination (121). Another significant feature was the high abundance of 1,3,6,8-TeCDD and 1,3,7,9-TeCDD, both of which are well-known products of the condensation of 2,4,6-TriCPh (64), which was one of the most abundant chlorophenol species in this study. All PCDD homologues reflected isomer distribution patterns consistent with PCPh condensation pathways, but the patterns were also slightly similar to those associated with chlorination of the dibenzo-p-dioxin backbone and/or lower chlorinated condensation products, i.e., mainly directed to the lateral positions. The order of dibenzo-p-dioxin and dibenzofuran chlorination by electrophilic aromatic substitution has been described previously (121) as 2→8→3→7→1→4→6→9. This was only partly matched by the PCDD isomer distribution patterns in the present study.

Table 6. Most abundant isomers within each PCDD and PCDF homologue.

Homologue PCDD PCDF Mono- 2- 3- > 2-

Di- 1,3- > 2,8- [1,2- 2,4-] > 2,3- ≥ 2,8- Tri- 1,3,8- > [2,3,7- 1,4,6-]a > 1,3,7- [2,3,4- 2,3,8-] > 2,3,6- > 2,4,7- Tetra- 1,3,6,8- > 1,3,7,8- ≥ [1,2,3,4- 1,2,3,8- 1,2,4,6- 1,2,3,9-]

[1,3,7,8- 1,3,7,9-] ≥ [1,4,7,8- 1,3,6,9- 1,2,3,7-] ≥ [1,2,4,9- 2,3,6,8-] ≥ 1,3,6,8-

Penta- [1,2,4,6,8- 1,2,4,7,9-] > 1,2,3,6,8- > 1,2,3,7,9-

[1,2,3,6,8- 1,3,4,7,8-] > 2,3,4,6,8- ≥ 1,2,4,6,8- ≥ 1,3,4,6,8- = 1,2,4,7,8- = [1,2,3,4,8- 1,2,3,7,8-]

Hexa- [1,2,3,6,7,9- 1,2,3,6,8,9-] > 1,2,3,4,6,8-

1,3,4,6,7,8- ≥ 1,2,4,6,7,8- > 2,3,4,6,7,8-

Hepta- 1,2,3,4,6,7,8- 1,2,3,4,6,7,8 > 1,2,3,4,6,7,9- a Brackets denote coeluting congeners.

The PCDF isomer distribution patterns were more complex, partly due to the greater number of congeners (135 PCDF congeners in compared to 75 for the PCDDs) but also due to chromatographic co-elutions under the analytical conditions reported in Paper II. The most abundant MoCDFs were the laterally substituted 2- and 3-isomers (Table 6), which agrees well with stability data which predict that initial DF chlorination adjacent to the oxygen bridge is less favourable (122). Data for the tetra- and penta-CDF homologues were also confounded by co-elutions, but the formation of these homologues appeared to be influenced by chlorination as well as precursor condensation reactions. The most abundant HxCDF isomers were 1,3,4,6,7,8-, 1,2,4,6,7,8-, and 2,3,4,6,7,8-HxCDF, and the most abundant HpCDF isomer was 1,2,3,4,6,7,8-HpCDF,

33


which illustrates the negative influence upon formation of the 1- and 9-positions. For the higher chlorinated homologues especially, the 1- and 9-positions seem to be highly unfavourable for substitution, as exemplified by the minimal formation of 1,3,4,9-, 1,4,6,9-, and 1,2,3,9-TeCDF.

For the PCNs, the mono- to tri-homologues displayed a relatively even distribution of isomers, but a more distinct pattern was evident in the tetra-to hepta-chlorinated homologues. The isomer distribution patterns were partially consistent with chlorination by electrophilic aromatic substitution, i.e. primary chlorination at ortho-positions. The β-positions, i.e. the 2,3,6,7-substitution sites (Figure 1), seemed to be somewhat favoured for chlorination. There is thus an interesting analogy with PCDD and PCDF formation, in which the lateral 2,3,7,8-positions are preferred for chlorine substitution (53,121). The preference for chlorination at the β-positions may be explained by the greater relative stability of congeners that are chlorinated in the β-positions rather than the α-positions (123). By reference to the most abundant isomers in each homologue, two parallel chlorination pathways from mono- to heptachlorinated naphthalene were postulated. These are illustrated in Figure 13. Chlorination of MoCN is directed to the second benzene ring, which confers the highest stability (123). Further chlorination appears to be by ortho-directed, electrophilic, aromatic substitution. A few of the congeners may be attributed to formation via chlorophenol condensation (57) or by PAH breakdown reactions (56). Thus, it seems likely that PCNs are not formed by a single pathway but by a combination of successive chlorination of naphthalene, de novo synthesis from PAHs, and chlorophenol condensation. However, during these conditions stepwise chlorination of naphthalene appears to be the most significant pathway.

The most abundant PCBz and PCPh congeners were probably formed by subsequent chlorination. All PCPh homologues were dominated by single congeners. When arranged from mono- to pentachlorophenol, these major isomers suggest a possible chlorination sequence which is presented in Figure 14. The PCPh chlorine substitution pattern clearly follows the ortho-para directionality mediated by the hydroxyl group, as in the electrophilic, aromatic substitution of phenols. Chlorination of phenol (66) has been shown to form the same chlorophenol species as the major PCPh isomers identified in the present study, which infers a chlorine substitution pathway for PCPh formation. With regard to the formation of dioxins and dioxin-like compounds from PCPh, at least one ortho-substituted chlorine is needed for PCDD formation and an ortho-hydrogen is required for PCDF formation (71). For example, 2-MoCPh was the most abundant mono-isomer, and the presence of an ortho-chlorine and an unsubstituted ortho-carbon site enables formation of both PCDD and PCDF. However, formation of PCDDs is favored over PCDF formation, resulting in DD and 1-MoCDD as major products (124).

34


Cl

Cl

Cl

Cl Cl

Cl

Cl

Cl Cl

ClCl

Cl

ClCl

Cl

Cl ClCl

Cl

Cl

ClCl

Cl

Cl

ClCl

Cl

Cl

ClCl

Cl

Cl

Cl

ClCl

Cl

ClCl

ClCl

Cl

ClCl Cl

ClCl

Cl

Cl Cl

Cl

ClCl

Cl

Cl Cl

Cl Cl

1-MoCN 2-MoCN

1,7-DiCN 2,6-DiCN

1,2,7-TriCN 1,3,7-TriCN

1,2,4,7-TeCN 1,2,6,7-TeCN 1,2,3,7-TeCN 1,2,5,7-TeCN

1,2,3,5,7-PeCN1,2,4,6,7-PeCN

1,2,3,4,6,7-HxCN 1,2,3,5,6,7-HxCN

1,2,3,4,5,6,7-HpCN

ClCl

Cl

Cl Cl

Cl Cl

Cl

1,2,3,4,5,6,7,8-OCN

Figure 13. Suggested PCN chlorination pathways based on the dominating isomers within each homologue (Paper III).

35


1,4-DiCBz

ClCl

Cl

Cl

Cl

Cl

ClCl

Cl

ClCl

ClCl

ClCl

ClClCl

ClCl

ClClCl

Cl

1,2-DiCBz1,2,4-TriCBz

1,3-DiCBz1,2,3,5-TeCBz

PeCBz HxCBz

> =

OHCl

ClCl

Cl

Cl

2-MoCPh

OHCl

OHCl

Cl2,4-DiCPh 2,4,6-TriCPh

OHCl

Cl

Cl

OHCl

Cl

Cl

Cl

2,3,4,6-TeCPh PeCPh

Figure 14. Suggested chlorination pathways of PCBz and PCPh, based on the dominating congeners.

The DiCBz isomer pattern was quite evenly distributed for the DiCBz. The TriCBz isomer pattern contained one highly abundant isomer (1,2,4-triCBz) which may be formed from chlorination of all three DiCBz isomers. As for the PCPh, the major PCBz congeners can be arranged in a chlorination sequence (Figure 14) which partly displays the ortho-para substitution associated with electrophilic aromatic substitution. However, formation via pathways other than chlorine substitution cannot be disregarded. A comparison with literature data indicated that the PCBz isomer distribution patterns obtained in Paper II were highly similar to patterns obtained by reacting acetylene with HCl over a municipal waste incinerator fly ash (50). This indicates that the coupling of aliphatic entities governs the additional routes suggested above.

Discussing the mechanisms of chlorination raises the question of the chlorine source. At temperatures exceeding 420°C the chlorinating agent may be diatomic chlorine formed from HCl via the Deacon process. However, as was shown by Procaccini and co-workers, some elemental chlorine is present as radical species in the flue gas (125). As was discussed in Paper I, a significant contribution of chlorine substituents is likely originating from divalent copper chloride (CuCl2) via direct chlorination, which further emphasizes the importance of copper speciation and solidification. When considering the chlorine substitution characteristics of all the compounds studied, it is seems likely that more than one chlorination mechanism is active. For example, in some aspects the patterns strongly suggest chlorination via electrophilic aromatic substitution; however this mechanism does not produce the meta-substituted chlorobenzenes observed in the above mentioned chlorination sequence. Furthermore, the observed differences in optimum temperature region for chlorination support chlorination via different pathways.

36


The effect of poor combustion on PCNs During a solvent control run it was confirmed that injection of the pure solvent (n-heptane) did not affect either total levels or homologue profiles of PCNs. However, during the collection of a second sample a short period of incomplete combustion occurred (referred to as poor combustion). This was caused by fuel feeding irregularities, and manifested as a 770 ppm CO-peak. Although the CO-peak was limited in both magnitude and duration, the effect of the poor combustion event was substantial, clearly increasing the formation of naphthalene and PCNs, with the largest increase seen for the MoCN homologue. However, the most striking effect of the poor combustion event was the extensive formation of 1,8-DiCN, 1,2,3-TriCN and 1,2,8-TriCN. The increases were 40-, 5-, and 12-fold respectively (Figure 15). This implies a preference for chlorination of the α-positions under such conditions, in contrast to the patterns observed during normal combustion. Interestingly, these thermodynamically unfavourable (123) congeners were of low abundance in the reference samples and the injection samples, but they were very abundant in the poor combustion sample, probably due to substantial PAH formation. Iino et al. (56) have observed a similar 1,8-substitution pattern and identified the TriCN isomers noted above as characteristic products of de novo synthesis from PAHs, in which 1,2,3-TriCN was formed from dibenzopentacene and 1,2,8-TriCN from benzo[ghi]perylene. Moreover, formation of 1,8-DiCN has been observed during combustion of 2,6-DiCPh (57), and was identified as the most abundant DiCN product. This was possibly as a result of formation via a dibenzo-p-dioxin intermediate, rather than from a dibenzofuran intermediate.

0.0

0.1

0.2

136- 135- 137- 146- 124- 125- 126- 127- 167- 123- 138- 145- 128-

0.0

0.1

0.2

0.3

0.4

13- 14- 16- 15- 27- 26- 17- 12- 23- 18-

450°C 300°C200°C CO Peak

Isom

er F

ract

ion

Isom

er F

ract

ion

Figure 15. The effect of a poor combustion event (770 ppm CO) on Di- and TriCN isomer distribution patterns, compared to the triplicate reference samples collected at 450°C, 300°C and 200°C. Error bars represent ±1 standard deviation.

37

7. INJECTION OF BACKBONE STRUCTURES The compounds selected for injection included unsubstituted and fully substituted DDs and DFs, and also the four aromatic structures shown in Figure 16. These compounds, namely biphenyl, naphthalene, phenanthrene and fluorene, share several structural features with DD and DF. This enabled an estimation of the contribution from dioxin-like polyaromatics to the formation of dioxins.

A C DB

Figure 16. Structural formulas for biphenyl (A), naphthalene (B), fluorene (C) and phenanthrene (D).

The effects of injecting the compounds differed substantially. Injection of the octachlorinated compounds did not result in any significant divergence from the reference samples (Figure 17), not even regarding the concentration of the injected species. The combustion experiment in which OCDF was injected was disturbed by strong winds affecting the fans, and causing fluctuation in the negative pressure. To maintain the required temperatures, the fuel feeding was increased and hence the OCDF injection profiles were incomparable and thus excluded. An undetectable change in the OCDD concentrations during injection of this species may be explained by three different scenarios, namely (i) injection of these high-weight/low-volatility molecules did not successfully transfer the OCDD to the flue gas stream, (ii) the OCDD were introduced into the flue gas stream but were diluted or adsorbed, making them undetectable downstream, or (iii) the injected OCDD were subjected to degradation and/or dechlorination reactions in the post-combustion zone, but these reactions did not result in detectable levels of possible products.

Injection of the unsubstituted DF did not significantly affect the formation of PCDF, as can be seen in the homologue profile (Figure 17). However, a reduction of mainly Tri-PeCDD was observed. It seems plausible that the availability of the dibenzofuran backbone is not a limiting factor in PCDF formation, but the elevated concentrations of DF may compete with reactions that form PCDDs, especially the tri- to penta-homologues.

39

INJECTION OF BACKBONE STRUCTURES

0.0

0.2

0.4

0.6

0.8

1.0

MoCDD DiCDD TriCDD TeCDD PeCDD HxCDD HpCDD OCDD

PCD

D [p

mol

e m

-3]

0

20

40

60

80

MoCDF DiCDF TriCDF TeCDF PeCDF HxCDF HpCDF OCDF

PCD

F [p

mol

e m

-3]

AverageDFBiphenylNaphthaleneFluorenePhenanthreneOCDD

Figure 17. PCDD and PCDF homologue profiles during injection of the respective compounds. Profiles obtained during DD-injection are shown separately (Figure 18) and OCDF was excluded due to disturbances during combustion.

The biphenyl, naphthalene and phenanthrene injections did not result in any observable effects on the PCDD and PCDF homologue profiles, whereas injection of fluorene significantly reduced the concentrations of Tri-HxCDD (Figure 17). This is an interesting finding for two reasons. Firstly, fluorene is rarely mentioned in dioxin formation contexts even though it is structurally similar to DF. Secondly, the effect of fluorene injection seems to replicate the effects of injection of DF, and for some homologues is even more pronounced. Whether this is due to competition for catalytic sites or because chlorine substitution is favoured for fluorene in comparison to dibenzo-p-dioxin cannot be determined from the present data set, but is definitely a subject for future research.

The injection of dibenzo-p-dioxin and naphthalene resulted in the findings reported in Papers II and III and is discussed below.

The effect of dibenzo-p-dioxin injection on PCDD and PCDF The dibenzo-p-dioxin was dissolved in n-heptane at a concentration of 0.30 µg/µl and injected into the post-combustion zone at approximately 650°C. Under normal conditions the non-chlorinated dibenzo-p-dioxin and dibenzofuran were the most abundant species in the PCDD and PCDF homologue profiles. The injection resulted in a 10-fold increase in DD concentrations, from 3.1 pmole/m3 at 200°C in the reference samples, to 40 pmole/m3 during injection (Figure 18). DD concentrations decreased markedly from 450°C to 200°C during injection probably as a result of chlorine substitution reactions and/or decomposition of the dibenzo-p-dioxin backbone. However, at these temperatures the DD structure should not decompose, according to literature data (126).

40


0

1

2

3

4

5

DD MoCDD DiCDD TriCDD TeCDD PeCDD HxCDD HpCDD OCDD

PCD

D [p

mol

e m

-3]

450°C

300°C

200°C

83 67 40

0

100

200

DF MoCDF DiCDF TriCDF TeCDF PeCDF HxCDF HpCDF OCDF

PC

DF

[pm

ole

m-3

]

Figure 18. PCDD (upper) and PCDF (lower) homologue profiles during injection of dibenzo-p-dioxin (bars) and in the reference samples (bullets) at 450°C, 300°C, and 200°C (Paper II). Error bars represent ±1 standard deviation.

The total PCDD concentrations were also elevated during DD injection, resulting in a 73% increase of the mono- to octa-chlorinated PCDDs at 200°C in the injection sample compared to the reference samples. Interestingly, the observed increase was limited to the tri- to hexachlorinated homologues (Figure 18), which is not consistent with chlorination of the injected DD. This suggests promotion of other pathways which favour the formation of the intermediately chlorinated dioxins. The effect of DD injection on the PCDFs on the other hand was not very pronounced - the total PCDF levels were the same as those for the reference samples. However, the concentration of DF decreased at 300°C and 200°C compared to the reference samples and to some extent also MoCDF, which may suggest that the reactions to form the dibenzofuran backbone were restricted by competition induced by the injection of DD.

The injection of dibenzo-p-dioxin did not result in an increase of PCDD congeners originating from chlorine substitution as anticipated, but enhanced the isomer distribution pattern found in the reference samples. The congeners observed at increased levels (Figure 19) were all identified to be products of chlorophenol condensation or additional chlorination of condensation products. Several PCDF congeners were affected (Figure 20). For example, an increase was observed in the concentrations of 1,3-DiCDF and several tetrachlorinated congeners (1,3,6,8-, 1,3,7,8-/1,3,7,9-, 1,2,4,7-, 1,2,4,6-/1,2,6,8- and 1,2,3,8-/1,4,6,7-/1,2,3,6-TeCDF), while 2,3,4-/2,3,8- and 2,3,6-TriCDF decreased. It should be noted that although the dibenzo-p-dioxin injection was performed in a single experimental run, the three samples that were collected simultaneously were highly similar, which adds support to the observed effects.

41


0.0

0.2

0.4

0.6

0.8

2- 1- 13- 23- 28- 14-17-

18- 12- 19-

Mo-

DiC

DD

[pm

ole

m-3

] 450°C

300°C

200°C

0.0

0.2

0.4

137- 138- 136- 124- 237-146-

148- 123- 178- 127- 128-147-

126-

TriC

DD

[pm

ole

m-3

]

0.0

0.2

0.4

0.6

0.8

1368- 1379- 1378- 1247-1248-

1369- 1268- 1478- 2378- 1237- 1234-1238-1246-1239-

1279- 1236- 1278- 1239- 1269- 1267- 1289-

TeC

DD

[p

mo

le m-3

]

0.0

0.2

0.4

0.6

12468-12479-

12368- 12478- 12379- 12469- 12347- 12378- 12369- 12467-12489-

12346- 12367- 12389-

PeC

DD

[pm

ole

m-3

]

0.0

0.2

0.4

124679-124689-

123468- 123679-123689-

123478- 123678- 123469- 123789- 123467- 1234679-

1234678-

Hx-

HpC

DD

[pm

ole

m-3

]

1234679- 1234678-

Figure 19. PCDD isomer distribution patterns during injection of dibenzo-p-dioxin (bars) and in the reference samples (bullets) at 450°C, 300°C, and 200°C. Error bars represent ±1 standard deviation. A reaction pathway that offers an explanation to the changes in PCDF isomer distribution patterns that were observed during DD injection is presented in Figure 21. It is initiated by the decomposition of dibenzo-p-dioxin introduced into the hot flue gases (650°C), probably via a cleavage of the C-O bonds (43) to create phenolic decomposition products. These would be rapidly chlorinated to form the specific PCPh congeners discussed previously and would then re-condense to form chlorinated dibenzo-p-dioxins and dibenzofurans. The

42


0

10

20

1- 3- 2- 4-

MoC

DF

[pm

ole

m-3

]

450°C300°C200°C

0

5

10

15

13- 17- 18- 12-24-

37- 27- 23- 36-28-

26- 19- 46-

DiC

DF

[pm

ole

m-3

]

0

1

2

3

4

5

124-167-

123- 127- 247- 248- 246- 234-238-

236-

TriC

DF

[pm

ole

m-3

]

0

1

2

3

1368

-

1378

-137

9-

1347

-

1468

-

1247

-

1346

-124

8-

1246

-126

8-

1678

-123

4-

1238

-146

7-

1349

-

1278

-

1267

-

1469

-

1249

-236

8-

2467

-

2347

-

2378

-

2348

-

2346

-

2367

-

3467

-

TeC

DF

[pm

ole

m-3

]

0.0

0.5

1.0

1.5

1346

8-

1246

8-

1367

8-

1347

9-

1236

8-

1247

8-

1247

9-

1246

7-

1234

7-

1346

9-

1234

8-

1234

6-

1237

9-

1236

7-

1246

9-

1267

9-

1236

9-

2346

8-

1234

9-

1248

9-

2347

8-

1238

9-

2346

7-

PeC

DF

[pm

ole

m-3

]

0.0

0.5

1.0

123468- 134679-134678-

124678- 124679- 123478-123479-

123678- 124689- 123467- 123679- 123689-123469-

123789- 123489- 234678-

HxC

DF

[pm

ole

m-3

]

0

1

2

1234678- 1234679- 1234689- 1234789-

HpC

DF

[pm

ole

m-3

]

1236

-

1236

8-13

478-

1234

7-14

678-

1234

8-12

378-

1247

9-13

467-

1246

9-12

678-

Figure 20. PCDF isomer distribution patterns during injection of dibenzo-p-dioxin (bars) and in the reference samples (bullets) at 450°C, 300°C, and 200°C (Paper II). Error bars represent ±1 standard deviation.

43


reconstructed PCDD/F structures are at this point chlorinated at substitution sites indicative of chlorophenol condensation reactions. The dibenzo-p-dioxin injected into the high temperature region of the post-combustion zone may generate a number of decomposition/chlorination products. However, the levels of PCBzs and PCPhs are several orders of magnitude greater than the PCDDs and PCDFs, and consequently the formation of phenolic intermediates proposed above cannot be verified based on concentration increase.

+OH

Cly

OH

Clx

O

O

Clx Cly

OH

OHO

O

+

Figure 21. Suggested decomposition – chlorination - condensation pathway during the injection of dibenzo-p-dioxin.

0

100

200

300

MoCN DiCN TriCN TeCN PeCN HxCN HpCN OCN

PCN

[pm

ole

m-3

]

450°C300°C200°C

Figure 22. PCN homologue profiles during injection of naphthalene (bars) and in the reference samples (bullets) at 450°C, 300°C, and 200°C (Paper III). Error bars represent ±1 standard deviation.

44


0

0.2

0.4

0.6

123467-123567-

123457- 123568- 124568-124578-

123578- 123456- 123458- 1234567-

1234568-

Hx-

HpC

N [p

mol

e m

-3]

0

1

2

3

12357-12467-

12457- 12468- 12346- 12356- 12367- 12456- 12478- 12358-12368-

12458- 12345- 12378-

PeC

N [p

mol

e m

-3]

0

50

100

150

1- 2- 13- 14-16-

15-27-

26-17-

12- 23- 18-

Mo-

DiC

N [p

mol

e m

-3]

450°C

300°C

200°C

0

2

4

6

8

136-

135-

137-

146-

124-

125-

126-

127-

167-

123-

138-

145-

128-

TriC

N [p

mol

e m

-3]

0

1

2

3

4

5

1357- 1257-1246-1247-

1367- 1467- 1368-1256-

1235-1358-

1237-1234-1267-

1245- 2367- 1248- 1258-1268-

1458- 1238- 1278-

TeC

N [p

mol

e m

-3]

1234567- 1234568-

124-146- 125- 126-137- 127-135-136- 123-167- 138- 145- 128-

Figure 23. PCN isomer distribution patterns during injection of naphthalene (bars) and in the reference samples (bullets) at 450°C, 300°C, and 200°C (Paper III). Error bars represent ±1 standard deviation.

45


The effect of naphthalene injection on PCN In Paper III naphthalene was dissolved in n-heptane at a concentration of 30 µg/µl and injected at P2 (650°C, Figure 6). The injection of naphthalene increased the total levels of PCNs by approximately 30% at 200°C compared to the reference samples. This was accompanied by a reduction in the degree of chlorination and, in contrast to the reference samples, the degree of chlorination tended to decrease with the temperature. The homologue profiles at 200°C during naphthalene injection (Figure 22) indicated increased formation of monochlorinated naphthalene which can be seen in the distribution of individual congeners (Figure 23). The remaining homologues appeared to be unaffected by the injection of naphthalene, probably due to insufficient residence time in the required temperature interval limiting further chlorination. This further supports the hypothesis that the major PCN formation route involves successive chlorination of naphthalene.

Injection of naphthalene and dibenzo-p-dioxin provide an opportunity to compare the effects on dioxin formation of the two structures, which differ by the oxygen bridges incorporated into the DD backbone. Since the aromatic substitution of chlorine is believed to mainly be directed to the lateral positions (121), chlorination should lead to similar substitution patterns for the two classes of compounds thus produced. The α-positions of the dibenzo-p-dioxin molecule would naturally also exert an influence and behave differently due to the close proximity to the oxygen for the dibenzo-p-dioxin molecule. The seemingly small structural difference of the oxygen bridges of the dibenzo-p-dioxin molecule influence the difference in formation characteristics and create a larger, three-ring that lacks true aromaticity and consequently has reduced stability.

46

8. INFLUENCE OF MOLECULAR DESCRIPTORS PCDD and PCDF isomer distribution patterns from combustion generally contain almost all congeners, but not in equal amounts (113). Since the thermodynamically less stable congeners are generally less abundant (127), a thermodynamic impact on the relative abundance of individual congeners seems probable. Previous attempts to correlate PCDD/F formation in thermal processes with thermodynamic properties (127-130) have only addressed the tetra- to octa-chlorinated PCDD/F congeners, or have focused on the effects of addition of reagents such as sulfur dioxide and chlorine (131), or the correlation of PCPh and PCDD/F (22). Since experimentally determined physicochemical data are available for few of the 210 mono- to octa-chlorinated PCDD/F congeners, models are needed to derive thermodynamic data for the complete set of congeners.

In order to evaluate the impact of thermodynamics and chlorine substitution on the PCDD/F isomer distribution patterns reported in Paper II, a data set was compiled for a total of 128 resolved mono- to hepta-chlorinated PCDD and PCDF, as is described in Paper IV. Physicochemical property estimates collected from the literature were included, as were chlorine substitution descriptors for all investigated congeners (Table 7). The main objective was to examine the data set using the recently developed chemometric projection method O2PLS (132) in order to identify parameters affecting PCDD and PCDF formation in the post-combustion zone of the laboratory-scale fluidized-bed reactor. Separate models were applied to PCDDs and PCDFs since their formation is believed to proceed via different pathways.

Physicochemical properties The physicochemical variables that were investigated comprised total energy (TE), electronic energy (EE), core-core repulsion energy (CCR), enthalpy of formation (∆Hf), entropy (S0), relative free energy (R∆Gf), vapour pressure (-LogP), energy of the highest occupied molecular orbital (EHOMO), energy of the lowest unoccupied molecular orbital (ELUMO), EHOMO – ELUMO gap (ELUMO-EHOMO), largest negative net atomic charge on a carbon atom (q-C), largest positive net atomic charge on a hydrogen atom (q+H), largest positive net atomic charge on a chlorine atom (q+Cl), dipole moment (DM), mean molecular polarizability (MP), and molecular volume (MV) (120,133-137).

47

INFLUENCE OF MOLECULAR DESCRIPTORS

Table 7. The physicochemical properties and chlorine substitution descriptors used in the O2PLS modelling (Paper IV).

Variable General Description Unit Reference TE Total energy eV Chen et al. 2004 (134) EE Electronic energy eV Chen et al. 2004 (134)

CCR Core-core repulsion energy eV Chen et al. 2004 (134) ∆Hf Enthalpy of formation kJ/mol Wang et al. 2004 (120) and

2005(135) S0 Entropy J/mol K Wang et al. 2004 (120) and

2005(135) R∆Gf Relative free energy kJ/mol Wang et al. 2004 (120) and

2005(135) -LogP Vapor pressure Pa Govers and Krop 1998 (137) EHOMO Energy of the highest occupied

molecular orbital eV Yang et al. 2006 (133)

ELUMO Energy of the lowest unoccupied molecular orbital

eV Yang et al. 2006 (133)

ELUMO-EHOMO

EHOMO – ELUMO gap eV Yang et al. 2006 (133)

q-C Largest negative net atomic charge

on a carbon atom acu a Chen et al. 2004 (134)

q+H Largest positive net atomic charge

on a hydrogen atom acu a Chen et al. 2004 (134)

q+Cl Largest positive net atomic charge

on a chlorine atom acu a Chen et al. 2004 (134)

DM Dipole moment Debye Yang et al. 2006 (133) MP Mean molecular polarizability esu b Yang et al. 2006 (133) MV Molecular volume Å3 Yang et al. 2006 (133) Clα Number of chlorine substituents in α-position (1,4,6,9) Clβ Number of chlorine substituents in β-position (2,3,7,8) ClOM Number of ortho/meta chlorine pairs (1-2, 3-4, 6-7, 8-9) Cl1-3 Number of chlorine pairs in position 1 and 3 or analogous Cl1-4 Number of chlorine pairs in position 1 and 4 or analogous ClSEQ3 Number of three-chlorine sequences

Cl-O-Cl Simultaneous chlorine substitution adjacent to the oxygen bridge(s) (4-6 for PCDF and 4-6, 1-9 for PCDD)

ClR1 Number of chlorine substituents of the first aromatic ring (1,2,3,4) ClR2 Number of chlorine substituents of the second aromatic ring (6,7,8,9) ClBAY Number of chlorine substituents located at the bay side of the dibenzofuran

molecule (1,2,8,9) a atomic charge unit b electrostatic unit of charge

The molecular structural variables TE, EE, CCR, MP, and MV are relevant for molecular size and/or polarizability, while the thermodynamic properties ∆Hf, S0, and R∆Gf may be regarded as stability parameters. Since vapour pressure is a measure of the volatility of a compound it indicates the availability of a congener for potential reaction on the fly ash surface. The remaining descriptors (EHOMO, ELUMO, ELUMO-EHOMO, q-C, q+H, q+Cl, and DM) were selected to represent each congeners charge distribution and propensity for intermolecular interactions.

48


Chlorine substitution descriptors The chlorine substitution descriptors were selected to describe position, symmetry and orientation of chlorine substituents at the 1- to 8-positions (Figure 1), with the intention to indicate PCDD/F formation and chlorination pathways. The chlorine substitution descriptors were denoted Clα (number of chlorines in α-positions; 1,4,6,9), Clβ (number of chlorines in β-position; 2,3,7,8), ClOM (number of ortho/meta chlorine pairs; 1-2, 3-4, 6-7, 8-9), Cl1-3 (number of chlorine pairs in position 1 and 3, or analogous), Cl1-4 (number of chlorine pairs in position 1 and 4, or analogous), ClSEQ3 (number of three-chlorine sequences), Cl-O-Cl (simultaneous chlorine substitution adjacent to the oxygen bridge(s); 4-6 for PCDFs and 4-6 or 1-9 for PCDDs), ClR1 (number of chlorines at the first aromatic ring; 1,2,3,4), ClR2 (number of chlorines at the second aromatic ring; 6,7,8,9), and ClBAY (number of chlorines at the bay side of the dibenzofuran molecule; 1,2,8,9). All chlorine substitution descriptors were normalized to the total number of substituents (ClN) to correct for the degree of chlorination. Finally, a set of descriptors indicating the substitution status of each carbon site was added to the data set, where 0 and 1 indicated unsubstituted and substituted, respectively. These descriptors will be referred to as #Cl 1 – #Cl 9.

O2PLS – bidirectional orthogonal projections to latent structures O2PLS is a novel extension of PLS (partial least squares projections to latent structures), which enables bi-directional modelling between multivariate data sets and separation of the systematic variation in the data sets into common variation (correlated, or predictive) and variation unique to each data set respectively (orthogonal). Conventional PLS uses variables assigned as X-block to construct a model of the responses that describe the Y-block. However, in many data sets there is a risk of systematic variation in X not linearly correlated with Y, which may interfere with the interpretation of the model. In O2PLS this uncorrelated systematic variation, also referred to as structured noise, may be described and interpreted since the predictive and orthogonal variation is separated. The division into correlated and uncorrelated components greatly facilitates the interpretation of data and improves the understanding of the data. Furthermore, the bidirectionality of O2PLS means that X can be used to predict Y, and vice versa. In Paper IV, the Simca P+ 12 software (138) was used for PCA and O2PLS modelling. The PCDD and PCDF isomer fractions at 300°C (Paper II) were defined as the Y-block and the physicochemical variables and chlorine substitution descriptors comprised the X-block.

Modelling of PCDDs The O2PLS modelling reported in Paper IV was based on the physicochemical variables and chlorine substitution descriptors (X block) and the PCDD and PCDF isomer fractions (Y-block). The model extracted predictive components (referred to as P-components) representing joint X/Y variation, and Y-orthogonal components (referred to as OY-components) that described the

49


variation in the X-block not correlated to the Y-block. No X-orthogonal components were found to be significant in any of the models according to cross validation (CV). This indicated that all systematic variation in the Y-block was correlated to the X-block, i.e. related to the physicochemical variables and chlorine substitution descriptors. The cross-validation procedure excludes 1/7 of the observations from the model development, then uses the resulting model to predict the values of the excluded observations, and compares the predicted and actual values. This procedure is repeated until every observation has been excluded once, i.e. seven times. The scores plots display the distribution of the observations (here; congeners) projected onto a plane representing the largest variation in the multivariate space, in which congeners located close to each other are similar while congeners far apart from each other are dissimilar. To interpret the scores plots the corresponding loadings plots are used, in which the distribution of the variables is explaining the congener distribution seen in the scores plot, with variables located far from the centre of the plot exerting the highest influence on the projection.

The PCDD model resulted in two P-components and two OY-components. Since the cross-validated explained variance (Q2) did not increase with the second P-component, only the first component was evaluated. Visualisation of the model was performed by plotting X-score vectors (t) versus the Y-score vectors (u), in which the congeners were coloured according to homologue to facilitate interpretation (Figure 24). Plotting the t-vector versus the u-vector provides an illustration of the goodness of fit, and the linearity of the t1/u1 scores plot reflects the degree of correlation between the Y- and X-block variables, i.e. the correlation between the isomer distribution data set and the molecular descriptors (which constitute the X-block). Although the t1/u1 scores plot may seem cluttered, some useful observations were made, including the PCDD congener extremes denoted as 3 and 47 (2-MoCDD and 1,4,6,9-TeCDD) which differ in a number of aspects. Firstly, in terms of relative abundance, 2-MoCDD had an isomer fraction of 0.66 (the highest absolute value) whereas the isomer fraction of 1,4,6,9-TeCDD was 0.002 (the lowest). Secondly, these two congeners represent extreme cases regarding degree of α- and β-oriented substitution and also differ largely in relative free energy, R∆Gf, which was shown to be the single most influential variable for the distribution of PCDD congeners.

A further evaluation of Figure 24 showed that the congeners located to the right (39, 37, 42, 29 and 16, corresponding to 1,2,6,9-, 1,2,6,7-, 1,2,8,9-, 1,2,3,6-TeCDD and 1,2,6-TriCDD, respectively) all exerted low abundance and chlorine substituents in 1,2-positions. On the other side of the plot a group of high-abundance congeners were observed (5, 22, 65 and 74, corresponding to 1,3-DiCDD, 1,3,8-TriCDD, 1,2,3,4,6,8-HxCDD and 1,2,3,4,6,7,8-HpCDD, respectively) of which the first four are substituted at alternating positions (1,3-substitution, or analogously), a trait not applicable for the hepta-congener. Additionally, clustering of PCDD congeners was discerned, as exemplified by the tetra-homologue in which congeners 43, 45 and 46 (1,3,6,8-, 1,3,7,8- and 1,3,7,9-TeCDD) form a sub-group in the left part of the t1/u1 scores plot and the

50


non-laterally substituted TeCDD isomer (no. 47, Figure 24) is separated in the right hand side of the plot. This suggests that 1,3,6,8-, 1,3,7,8-, and 1,3,7,9-TeCDD, which are well-known chlorophenol condensation products (63,64), are not only favoured during formation but that their high abundance also seems to be related to the X-block variables.

-3.0

-2.0

-1.0

0.0

1.0

2.0

3.0

4.0

-2.0 -1.0 0.0 1.0 2.0 3.0

u[1]

tcv[1]

2

3

4

5

9

1015 2640

4

50

52

53

54

6

5760

61

1

13

14

16

1720

2122

25

29

30

3237

38

39

41 42

43

44

5

46

47

48

49

74 75

151

555

64

65

66

67 6871

SIMCA-P+ 12 - 2008-09-08 11:06:52 (UTC+1)

t1 [CV] (R2X: 7%, Q2: 68%)

u 1 (R

2 Y: 8

9%)

Figure 24. The crossvalidated t1/u1 scores plot for the PCDD model (Paper IV), visualizing the correspondence between the X- and Y-block. Colouring according to homologue (mono: yellow; di: turquoise; tri: pink; tetra: blue; penta: red; hexa: green; and hepta: black). For congener annotations, see Appendix 3.

-4.0

-2.0

0.0

2.0

4.0

-4 -2 0 2 4

u[1]

tcv[1]

2

34

5

7

1

44

962

5

88

91

92

9497

98

99 107

08

110

1114

11

1310 1

1213

1617

20

21

2225

4145

46

54

56 56667

68

69

70

75

76

79 81

8286

132

133

134

5

158084

8

87

96

01

103105

1

1

111135

116117 119

120121

122

123

124

125

126

127

129

130

131

SIMCA-P+ 12 - 2008-09-08 11:19:43 (UTC+1)

t1 [CV] (R2X: 12%, Q2: 51%)

u 1 (R

2 Y: 6

5%)

Figure 25. The t1/u1 scores plot for the PCDF model (Paper IV), visualizing the correspondence between the X- and Y-block. Colouring according to homologue (mono: yellow; di: turquoise; tri: pink; tetra: blue; penta: red; hexa: green; and hepta: black). For congener annotations, see Appendix 3.

51


To further elucidate the factors governing these observations, separate models were calculated for the homologues containing the highest numbers of isomers, namely the Te- and PeCDD/Fs. The TeCDD model resulted in one P-component explaining 80% of the variation in Y and 24% of the variation in the X-block, and one OY-component. The first P-component was grouping the isomers in the same way as in the PCDD-model, but in a more pronounced way. The X-block variables that exerted the highest influence on the model were R∆Gf, ∆Hf, and TE, whereas the chlorine substitution descriptors seemed to be less significant. The chlorine substitution descriptor that exerted the strongest influence on projection was Cl1-3, which supports an alternating chlorine substitution pattern resulting from the condensation of 2,4,6-TriCPh. In Paper II, this specific chlorophenol congener was shown to be largely abundant in the flue gas and its main condensation products 1,3,6,8- and 1,3,7,9-TeCDD were identified as the most abundant TeCDD. The PeCDD model extracted two P-components, the first of which induced a division into three clusters, to repeat the tendency seen in Figure 24 with a grouping of 1,2,3,6,8-, 1,2,3,7,8-, 1,2,3,7,9-, and 1,2,4,7,8-PeCDD (53, 55, 56 and 61, respectively), but with 1,2,3,4,6-PeCDD (50) separated.

Modelling of PCDFs The PCDF model resulted in one P-component and two OY- components. The crossvalidated t1/u1 scores plots in Figure 25 display greater scatter of data for PCDFs than for the PCDDs, however linear which indicates a relationship between X and Y. The clustering behaviour was more pronounced than for PCDD and most of the homologues appeared to divide into two or three sub-groups of isomers. A closer investigation of these sub-groups revealed that the cause of separation was the same as for the PCDDs, namely correlation with chlorophenol condensation products. The tetrachlorinated homologue displayed two, or possibly three, clusters, in which the right-most group isomers (54, 76, and 68, i.e. 1,2,3,9-, 1,4,6,9- and 1,3,4,9-TeCDF) were substituted in the highly unfavoured 1- and 9-positions. These positions are sometimes referred to as the bay-sites of the dibenzofuran backbone structure and were shown in Figure 5. The 1,9-positions are more sterically crowded than the α-positions on the opposite side of the molecule (4- and 6-) or the corresponding sites of the dibenzo-p-dioxin molecule. The remaining isomers are grouped in the centre of the plot, whereas 70 and 86 are separated at the left-hand side. These two congeners (1,3,6,8- and 2,4,6,8-TeCDF) are predicted (139) and confirmed (63) products of the condensation of 2,4,6-TriCPh and 2,4-DiCPh. A similar division was seen for the PeCDFs, in which two distinct groups were observed (Figure 25). The right-hand group consisted solely of 1,9-substituted congeners while the left-hand sub-group contained PCPh condensation products and/or congeners formed from the further chlorination of tetrachlorinated PCPh condensation products. An identical distinction was observed for the HxCDFs and for the HpCDFs. The suppression of formation of 1,9-substituted PCDF congeners agrees with predictions of PCDF formation by chlorophenol condensation, in which PCDF congeners originating from

52


meta-chlorinated phenols are favoured over those formed from ortho- or para-substituted PCPhs, resulting in a characteristic 1,3,6,8-/1,3,7,9- pattern (139). The TriCDFs were clearly separated into two groups, one of which contained 2,4,6-, 2,4,7- and 2,4,8-TriCDF (44, 45 and 46, respectively) and the second contained 1,2,3-, 1,2,7- and 2,3,6-TriCDF (22, 25 and 41, respectively). This was possibly caused by the steric effects of adjacent chlorine substituents.

Modeling of the TeCDF data resulted in one P-component which isolated the same three congeners as in the PC1-7DF model, namely 1,2,3,9-, 1,4,6,9- and 1,3,4,9-TeCDF (54, 76 and 68, respectively). The most influential variables were the substitution of #Cl 9, ∆Hf, and R∆Gf, which further illustrates the significance of the sterically crowded 9-position. The PeCDF model did not result in the distinct separation into two groups as in the PCDF model, but rather an outstretched appearance with a reduced tendency towards sub-grouping. The most influential variables were ∆Hf, R∆Gf and CCR, the latter implying the increased impact of crowding effects from the increasing number of substituents.

An interesting feature of the PC1-7DD/F models and the Te-PeCDD/F models was the distribution of congeners observed in the t1/u1 scores plots, which may be considered analogous to a gradient describing the probability of origination from chlorophenol condensation.

-3.0

-2.0-1.0

0.0

1.0

2.0

3.0

-6 -4 -2 0 2 4 6

tocv

[2]P

tocv[1]P

23

45

910

11

13

14

1516

1720

2122

25

26

2930

323738

39

40

450 51

52

53

54

55 56

57

60

61 14243

44

45

4647

48

49

74 756465

66

6768

71

SIMCA-P+ 12 - 2008-09-19 13:00:31 (UTC+1)

t o2C

V (R

2 X: 1

2%)

to1CV (R2X: 42%)

-0.6

-0.4

-0.2

-0.0

0.2

0.4

0.6

-0.30 -0.20 -0.10 -0.00 0.10 0.20 0.30 0.40

po[2

]

po[1]

TEEECCR

Hf

S_0

RGf

- LogP HOMOLUMOLUMO-HOMO

Q-_C

Q+_H

Q+_ClDm

MpMv

Cl_A/Cl_N

Cl_B/Cl_N

Cl_OM/Cl_NCl_1~3/Cl_N

Cl_1~4/Cl_N

Cl_SEQ3/Cl_N

Cl-O-Cl/Cl_NCl_R1/Cl_N

Cl_R2/Cl_N

#1

#2#3

#4#6

#7 #8

#9

SIMCA-P+ 12 - 2008-09-19 13:07:24 (UTC+1)

p o2

po1

Figure 26. Scores and loadings plots showing the first two OY-components of the PCDD model (Paper IV). Colouring according to homologue (mono: yellow; di: turquoise; tri: pink; tetra: blue; penta: red; hexa: green; and hepta: black). For congener annotations, see Appendix 3.

53


Orthogonal variation The non-correlated variation in the O2PLS models was restricted to variation in the X-block orthogonal to Y, i.e. systematic variation unique to the X-block which indicated that the X-block was more informative than the Y-block. Hence it was concluded that the systematic variation in the Y-block isomer distribution patterns were fully described by the predictive components discussed previously.

The orthogonal variation was analysed by retrieving scores and loadings plots for the OY components. The PCDD congeners were distinctly separated according to the degree of chlorination by the first OY-component (Figure 26), which explained 42% of the variation in the X-block. A similar effect was observed for the PCDF (explaining 36% of the variation), but this was less pronounced (Figure 27). The high degree of explanation of the X-block variation for PCDD and PCDF, respectively, which demonstrates that there is a substantial proportion of variation in the X-block which is not related to the Y-block PCDD/F isomer fractions. The successive distribution of homologues was explained mainly by the physicochemical property variables, as was shown by their peripheral location along the first OY component (Figures 26 and 27), and most strongly by TE, EE, CCR, MP, -LogP, S0, EHOMO, ELUMO and EHOMO-ELUMO gap.

-3.0

-2.0

-1.0

0.0

1.0

2.0

3.0

-6 -4 -2 0 2 4 6

tocv

[2]P

tocv[1]P

2

34

5

7

101112

13

15

161720

21 22

2541

44

454654

566667

70

7

7905

88

91

92

94

9697

98103

105

107108

110

113

114

11559

62

68

69576

881

82 84

8

86

87

12

132133

99

101111

116117

119120121122

123

124

5126127

129

130131

134

135

SIMCA-P+ 12 - 2008-09-19 13:22:48 (UTC+1)

-0.4

-0.2

-0.0

0.2

0.4

-0.30 -0.20 -0.10 -0.00 0.10 0.20 0.30 0.40

po[2

]

po[1]

TEEE

CCRHf S_0

RGf- LogP

HOMOLUMOLUMO-HOMO

Q-_C

Q+_H

Q+_ClDm Mp

Mv

Cl_A/Cl_N

Cl_B/Cl_N

Cl_OM/Cl_NCl_1~3/Cl_N

Cl_1~4/Cl_N

Cl_SEQ3/Cl_N

Cl-O-Cl/Cl_NCl_R1/Cl_N

Cl_R2/Cl_N

Cl_BAY/Cl_N

#1

#2#3

#4

#6

#7#8

#9

SIMCA-P+ 12 - 2008-09-19 13:25:23 (UTC+1)

p o2

po1

to1CV (R2X: 36%)

t o2C

V (R

2 X: 8

%)

Figure 27. Scores and loadings plots showing the first two OY-components of the PCDF model (Paper IV). Colouring according to homologue (mono: yellow; di: turquoise; tri: pink; tetra: blue; penta: red; hexa: green; and hepta: black). For congener annotations, see Appendix 3.

54


The second OY-component was highly influenced by the extent of lateral and non-lateral chlorination (Clβ and Clα, respectively), which was illustrated by the peripheral positions of these variables in the loadings plots (Figures 26 and 27). Examining the scores plots confirmed this by the positioning of the β-substituted PCDD congeners denoted 49 and 71 (2,3,7,8-TeCDD and 1,2,3,7,8,9-HxCDD), and PCDF congener no. 84 (2,3,7,8-TeCDF) as well as the largely α-substituted 1,2,4,6,9-PeCDD and 1,2,3,4,6,9-HxCDD (congeners 60 and 66) and 1,2,3,4,6-PeCDF (88). Combined with the chlorination degree-relationship indicated by the first OY-component, the Y-orthogonal variation appeared closely connected with chlorination processes. Thus, the characteristic chlorination-related patterns in the X-block were not related to the Y-block isomer distribution patterns.

The findings in Paper IV has emphasized the complexity of PCDD and PCDF formation and has demonstrated the application of multivariate techniques, in particular the bi-directional O2PLS method, to this multidimensional problem. The physicochemical properties and chlorine substitution descriptors were efficient in describing PCDD and PCDF isomer distribution patterns, as was shown by the absence of X-orthogonal components in the Y-block, i.e. no systematic variation unique to the Y-block was found significant. The impact of other factors such as combustion conditions and waste composition may thus lie mainly in affecting PCDD/F levels, not relative isomer abundance.

55

9. CONCLUDING REMARKS A number of conclusions have been drawn from the work underlying this thesis. First, the formation of PCDDs and PCDFs was found to be rapid, mainly associated with temperatures between 640 and 400°C, and seemed to be highly dependent on sufficient residence time in this temperature region. Prolonged residence time at 450/460°C resulted in lower PCDD/F concentrations than during baseline conditions, even at lower temperatures along the post-combustion zone. This opens the possibility of reducing PCDD concentrations by adjusting the temperature drop to the temperature interval least favourable for PCDD formation.

Secondly, the PCDD and PCDF isomer distribution patterns indicated contribution from chlorophenol condensation as well as from chlorination reactions, for both classes of compounds. The PCDDs were largely influenced by chlorophenol condensation and to some extent chlorination reactions. For the PCDFs, chlorine substitution adjacent to the oxygen bridges was found to be unfavoured. This was highlighted by the notably lower abundance of 1,9-substituted congeners. The adverse influence of the sterically crowded bay-sites of the dibenzofuran backbone structure (1,9-positions) was also identified by O2PLS modeling, in which chlorination of the 9- position was the most influential variable in PCDF formation. The variable with the greatest influence on the distribution of PCDD congeners was the relative free energy (R∆Gf). The O2PLS models also displayed distinct clustering which was most pronounced for the PCDFs and divided most of the homologues into two or three sub-groups of congeners. These sub-groups appeared to correspond to the probability of origination from chlorophenol condensation.

PCN formation also appeared to proceed via more than one pathway, reflecting a combination of successive chlorination of naphthalene, de novo synthesis from PAHs and chlorophenol condensation. However, the stepwise chlorination of naphthalene seemed to be the most influential pathway.

Injection of naphthalene increased the formation of MoCN but the remaining homologues appeared to be unaffected. This was probably due to insufficient residence time at temperatures necessary for further chlorination. Injection of dibenzo-p-dioxin resulted in decomposition followed by chlorination and re-condensation to form chlorinated dibenzo-p-dioxins as well as dibenzofurans. This scenario was supported by the observation that the affected PCDD and PCDF congeners were not characteristic chlorination products, but congeners that are identified products of chlorophenol condensation.

57

CONCLUDING REMARKS

The injection of structurally similar dibenzofuran and fluorene resulted in similar effects, with a significant reduction of Tri-HxCDD. This was possibly caused by competition for catalytic sites or the chlorination of fluorene proceeding more readily than that of dibenzo-p-dioxin.

The present work has illustrated the scientific complexity of PCDD and PCDF formation and has also demonstrated bi-directional O2PLS to be a valuable technique with which to address such a multidimensional problem. The physicochemical properties and chlorine substitution descriptors were observed to exert a substantial influence on the PCDD and PCDF isomer distribution patterns, and although these were also affected by numerous other factors (e.g. combustion conditions and waste composition), thermodynamic considerations cannot be neglected.

Future research interests include the improvement of analytical techniques for full scan analysis, with the intention to improve separation and reduce analysis time. This would provide more comprehensive data for the wide array of organic compounds in combustion-related samples and reduce the risk of excluding valuable information when selecting analytical conditions. Further knowledge regarding formation and degradation of the di- and tri-chlorinated dibenzo-p-dioxin and dibenzofuran isomers requires more comprehensive identification of congeners for homologues that are poorly characterised at present. The observed effect of injection of dibenzofuran and fluorene on Tri-HxCDDs could not be determined from the present data set, but is a promising subject for future research.

Paper I discusses catalytically active copper species on fly ash surfaces and highlights the importance of combining formation studies of chloro-organic compounds in combustion systems with the characterisation of particulates. The importance of copper in dioxin formation is well documented but its speciation and distribution on fly ash particles requires further investigation, particularly at different temperatures and under varying combustion conditions. Finally and most importantly, results obtained using laboratory-scale experimental equipment must be verified in larger scale facilities and applications require continuing development.

58

ACKNOWLEDGEMENTS Many people around me deserve to be thanked. I take the liberty of being untraditional and start with thanking Barbro Andersson. Without you this thesis would not have been written. Naturally, I would also like to thank my supervisors Stellan Marklund and Mats Tysklind; you certainly complement each other and have contributed to the work underlying this thesis in very different ways. I have learned a lot from you both. I am also grateful to my other co-authors Johanna Aurell, Jerker Fick and Henrik Antti for interesting discussions and good collaboration. Thanks also to Dr. Jingwen Chen at Dalian University of Technology, China for generously sharing property data, and to colleagues and anonymous reviewers for valuable comments on data presented at conferences and in submitted papers.

I am naturally deeply grateful for the financial support I have received for research and conference travels: J. Gust. Richert Memorial Fund, Ångpanneföreningen's Foundation for Research and Development, Swedish Chemical Society, The Knut and Alice Wallenberg Foundation, The Kempe Foundation, and the Publishers Association Acta Chemica Scandinavica (FACS).

The importance of a positive and inspiring working environment cannot be overestimated. I would like to thank present and former co-workers at what used to be Environmental Chemistry, it has been a pleasure getting to know and working with you all! Thank you, Jerker, my combustion experiment associate, for practical help, critical reading and lots of inspiration. Thanks also to Per for critical reading and valuable advice, and to all the people that I have harassed during the work with the front cover. And of course, particularly warm thanks to Lisa and Johanna for being excellent room-mates, discussion partners and therapists. Good luck with your research assignments abroad!

My family and friends also deserve a big hug for bearing with me during the last year(s). I would especially like to thank my mother, Tobbe and Eva for being my greatest supporters and always giving me strength, unconditionally. And Eva, you are a truly unique combination of neighbour, mother in law and friend. Thank you for always being there for me and my family. Also, in trying times, a huge latte and a close friend are invaluable. Thanks to Jessica for lunch dates, coffee nights, and shopping streaks; I really appreciate your friendship! And although I can’t complain about the hardships of being a PhD student anymore, I am sure that I can find other issues to disentangle over a cup of coffee.

59

ACKNOWLEDGEMENTS

Finally, my beloved Patrik, what would I do without you? You are the foundation and backbone of our family, thank you for your patience and for always listening. I love you a little bit more today than yesterday, and a little bit less than tomorrow. Elsa and Nils, my wonderful children, you are more important to me than anything else in this world. Actually, you have contributed to this work by giving me time to breathe and recover some strength. Thank you for all your hugs and kisses and for making sure that I don’t get a chance to think about work when I am with you. I love you!

60

REFERENCES 1. Young,A.L., Regens,J.L. (2005). Serum TCDD levels and health effects from

elevated exposure: Medical and scientific evidence. Environmental Science and Pollution Research (12) 1-4.

2. United Nations Environment Programme (UNEP). (2001). Stockholm convention on Persistent Organic Pollutants (POPs).

3. Gaus,C., Brunskill,G.J., Connell,D.W., Prange,J., Müller,J.F., Päpke,O., Weber,R. (2002). Transformation processes, pathways, and possible sources of distinctive polychlorinated dibenzo-p-dioxin signatures in sink environments. Environmental Science & Technology (36) 3542-3549.

4. Olie,K., Vermeulen,P.L., Hutzinger,O. (1977). Chlorodibenzo-p-dioxins and chlorodibenzofurans are trace components of fly ash and flue gas of some municipal incinerators in the Netherlands. Chemosphere (6) 455-459.

5. Lundin,L., Marklund,S. (2005). Thermal degradation of PCDD/F in municipal solid waste ashes in sealed glass ampules. Environmental Science & Technology (39) 3872-3877.

6. Lundin,L., Marklund,S. (2007). Thermal degradation of PCDD/F, PCB and HCB in municipal solid waste ash. Chemosphere (67) 474-481.

7. Quaß,U., Fermann,M., Bröker,G. (2004). The European Dioxin Air Emission Inventory Project - Final results. Chemosphere (54) 1319-1327.

8. Falandysz,J. (1998). Polychlorinated naphthalenes: an environmental update. Environmental Pollution (101) 77-90.

9. Eiceman,G.A., Clement,R.E., Karasek,F.W. (1979). Analysis of fly ash from municipal incinerators for trace organic compounds. Analytical Chemistry (51) 2343-2350.

10. Abad,E., Caixach,J., Rivera,J. (1999). Dioxin like compounds from municipal waste incinerator emissions: assessment of the presence of polychlorinated naphthalenes. Chemosphere (38) 109-120.

11. Hanberg,A., Wærn,F., Asplund,L., Haglund,E., Safe,S. (1990). Swedish Dioxin Survey: Determination of 2,3,7,8-TCDD toxic equivalent factors for some polychlorinated biphenyls and naphthalenes using biological tests. Chemosphere (20) 1161-1164.

12. Villeneuve,D.L., Kannan,K., Khim,J.S., Falandysz,J., Nikiforov,V.A., Blankenship,A.L., Giesy,J.P. (2000). Relative potencies of individual polychlorinated naphthalenes to induce dioxin-like responses in fish and mammalian in vitro bioassays. Archives of Environmental Contamination and Toxicology (39) 273-281.

13. Falandysz,J., Strandberg,L., Bergqvist,P.-A., Kulp,S.E., Strandberg,B., Rappe,C. (1996). Polychlorinated naphthalenes in sediment and biota from the Gdañsk Basin, Baltic Sea. Environmental Science & Technology (30) 3266-3274.

61

REFERENCES

14. Järnberg,U., Asplund,L., de Wit,C., Egebäck,A.-L., Wideqvist,U., Jakobsson,E. (1997). Distribution of polychlorinated naphthalene congeners in environmental and source-related samples. Archives of Environmental Contamination and Toxicology (32) 232-245.

15. Bernes,C. (1998). Persistent Organic Pollutants. Swedish Environmental Protection Agency.

16. Han,J., Deming,R.L., Tao,F.-M. (2004). Theoretical study of molecular structures and properties of the complete series of chlorophenols. Journal of Physical Chemistry A (108) 7736-7743.

17. Van den Berg,M., Birnbaum,L.S., Denison,M., De Vito,M., Farland,W., Feeley,M., Fiedler,H., Hakansson,H., Hanberg,A., Haws,L., Rose,M., Safe,S., Schrenk,D., Tohyama,C., Tritscher,A., Tuomisto,J., Tysklind,M., Walker,N., Peterson,R.E. (2006). The 2005 World Health Organization reevaluation of human and mammalian toxic equivalency factors for dioxins and dioxin-like compounds. Toxicological Sciences (93) 223-241.

18. Gullett,B., Wikström,E. (2000). Mono- to tri-chlorinated dibenzodioxin (CDD) and dibenzofuran (CDF) congeners/homologues as indicators of CDD and CDF emissions from municipal waste and waste/coal combustion. Chemosphere (40) 1015-1019.

19. Oser,H., Thanner,R., Grotheer,H.-H., Gullett,B.K., Natschke,D., Raghunathan,K. (1998). Lowly chlorinated dibenzodioxins as TEQ indicators. A combined approach using spectroscopic measurements with DLR Jet-REMPI and statistical correlations with waste combustor emissions. Combustion Science and Technology (134) 201-220.

20. Kato,M., Urano,K. (2001). Convenient substitute indices to toxic equivalent quantity for controlling and monitoring dioxins in stack gas from waste incineration facilities. Waste Management (21) 55-62.

21. Kaune,A., Lenoir,D., Schramm,K.-W., Zimmermann,R., Kettrup,A., Jaeger,K., Rückel,H.G., Frank,F. (1998). Chlorobenzenes and chlorophenols as indicator parameters for chlorinated dibenzodioxins and dibenzofurans in incineration processes: Influences of various facilities and sampling points. Environmental Engineering Science (15) 85-95.

22. Tuppurainen,K., Ruokojärvi,P., Asikainen,A., Aatamila,M., Ruuskanen,J. (2000). Chlorophenols as precursors of PCDD/Fs in incineration processes: Correlations, PLS modeling, and reaction mechanisms. Environmental Science & Technology (34) 4958-4962.

23. Jay,K., Stieglitz,L. (1995). Identification and quantification of volatile organic components in emissions of waste incineration plants. Chemosphere (30) 1249-1260.

24. Ohura,T., Kitazawa,A., Amagai,T., Makino,M. (2005). Occurrence, profiles, and photostabilities of chlorinated polycyclic aromatic hydrocarbons associated with particulates in urban air. Environmental Science & Technology (39) 85-91.

25. Horii,Y., Ok,G., Ohura,T., Kannan,K. (2008). Occurrence and profiles of chlorinated and brominated polycyclic aromatic hydrocarbons. Environmental Science & Technology (42) 1904-1909.

26. Ohura,T., Fujima,S., Amagai,T., Shinomiya,M. (2008). Chlorinated polycyclic aromatic hydrocarbons in the atmosphere: Seasonal levels, gas-particle partitioning, and origin. Environmental Science & Technology (42) 3296-3302.

62

REFERENCES

27. Ohura,T., Morita,M., Makino,M., Amagai,T., Shimoi,K. (2007). Aryl hydrocarbon receptor-mediated effects of chlorinated polycyclic aromatic hydrocarbons. Chemical Research in Toxicology (20) 1237-1241.

28. Kilgroe,J.D., Lanier,W.S., Von Alten,T.R. (1992). Development of good combustion practice for municipal waste combustors. Proceedings of the National Waste Processing Conference (15) 145-156.

29. Fängmark,I., van Bavel,B., Marklund,S., Strömberg,B., Berge,N., Rappe,C. (1993). Influence of combustion parameters on the formation of polychlorinated dibenzo-p-dioxins, dibenzofurans, benzenes, and biphenyls and polyaromatic hydrocarbons in a pilot incinerator. Environmental Science & Technology (27) 1602-1610.

30. Gullett,B.K., Raghunathan,K. (1997). Observations on the effect of process parameters on dioxin/furan yield in municipal waste and coal systems. Chemosphere (34) 1027-1032.

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

32. Wikström,E., Marklund,S., Tysklind,M. (1999). Influence of variation in combustion conditions on the primary formation of chlorinated organic micropollutants during municipal waste combustion. Environmental Science & Technology (33) 4263-4269.

33. Blumenstock,M., Zimmermann,R., Schramm,K.-W., Kettrup,A. (2000). 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 (40) 987-993.

34. Zimmermann,R., Blumenstock,M., Heger,H.J., Schramm,K.-W., Kettrup,A. (2001). 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 (35) 1019-1030.

35. Aurell,J., Fick,J., Marklund,S. (2008). Effects of transient combustion conditions on the formation of PCDD/F, PCBz and PAH during MSW incineration. Environmental Engineering Science, (in press).

36. Bergström,J.G.T., Warman,K. (1987). Production and characterization of trace organic emissions in Sweden. Waste Management & Research (5) 395-401.

37. Kilgroe,J.D., Nelson,L.P., Schindler,P.J., Lanier,W.S. (1990). Combustion control of organic emissions from municipal waste combustors. Combustion Science and Technology (74) 223-244.

38. Huang,H., Buekens,A. (2000). Chlorinated dioxins and furans as trace products of combustion: some theoretical aspects. Toxicological and Environmental Chemistry (74) 179-193.

39. Violi,A., D'Anna,A., D'Alessio,A. (2001). A modeling evaluation of the effect of chlorine on the formation of particulate matter in combustion. Chemosphere (42) 463-471.

63

REFERENCES

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

41. Mätzing,H., Baumann,W., Becker,B., Jay,K., Paur,H.-R., Seifert,H. (2001). Adsorption of PCDD/F on MWI fly ash. Chemosphere (42) 803-809.

42. Miller,H., Marklund,S., Bjerle,I., Rappe,C. (1989). Correlation of incineration parameters for the destruction of polychlorinated dibenzo-p-dioxins. Chemosphere (18) 1485-1494.

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

44. Wikström,E., Marklund,S. (2000). Secondary formation of chlorinated dibenzo-p-dioxins, dibenzofurans, biphenyls, benzenes and phenols during MSW combustion. Environmental Science & Technology (34) 604-609.

45. Behrooz Ghorishi,S., Altwicker,E.R. (1995). Formation of polychlorinated dioxins, furans, benzenes, and phenols in the post-combustion region of a heterogeneous combustor: Effect of bed material and post-combustion temperature. Environmental Science & Technology (29) 1156-1162.

46. Hagenmaier,H., Kraft,M., Brunner,H., Haag,R. (1987). Catalytic effects of fly ash from waste incineration facilities on the formation and decomposition of polychlorinated dibenzo-p-dioxins and polychlorinated dibenzofurans. Environmental Science & Technology (21) 1080-1084.

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

48. Shaub,W.M., Tsang,W. (1983). Dioxin formation in incinerators. Environmental Science & Technology (17) 721-730.

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

50. Froese,K.L., Hutzinger,O. (1997). Mechanisms of the formation of polychlorinated benzenes and phenols by heterogeneous reactions of C2 aliphatics. Environmental Science & Technology (31) 542-547.

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

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

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

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

64

REFERENCES

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

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

57. Kim,D.-H., Mulholland,J.A., Ryu,J.-Y. (2007). Chlorinated naphthalene formation from the oxidation of dichlorophenols. Chemosphere (67) S135-S143.

58. Froese,K.L., Hutzinger,O. (1996). Polychlorinated benzene, phenol, dibenzo-p-dioxin, and dibenzofuran in heterogeneous combustion reactions of acetylene. Environmental Science & Technology (30) 998-1008.

59. Wehrmeier,A., Lenoir,D., Sidhu,S.S., Taylor,P.H., Rubey,W.A., Kettrup,A., Dellinger,B. (1998). Role of copper species in chlorination and condensation reactions of acetylene. Environmental Science & Technology (32) 2741-2748.

60. Tuppurainen,K., Halonen,I., Ruokojärvi,P., Tarhanen,J., Ruuskanen,J. (1998). Formation of PCDDs and PCDFs in municipal waste incineration and its inhibition mechanisms: a review. Chemosphere (36) 1493-1511.

61. Gullett,B.K., Bruce,K.R., Beach,L.O., Drago,A.M. (1992). Mechanistic steps in the production of PCDD and PCDF during waste combustion. Chemosphere (25) 1387-1392.

62. Karasek,F.W., Dickson,L.C. (1987). Model studies of polychlorinated dibenzo-p-dioxin formation during municipal refuse incineration. Science (237) 754-756.

63. Sidhu,S.S., Maqsud,L., Dellinger,B., Mascolo,G. (1995). 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 (100) 11-20.

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

65. Ryu,J.-Y., Mulholland,J.A., Kim,D.H., Takeuchi,M. (2005). Homologue and isomer patterns of polychlorinated dibenzo-p-dioxins and dibenzofurans from phenol precursors: Comparison with municipal waste incinerator data. Environmental Science & Technology (39) 4398-4406.

66. Ryu,J.-Y. (2008). Formation of chlorinated phenols, dibenzo-p-dioxins, dibenzofurans, benzenes, benzoquinnones and perchloroethylenes from phenols in oxidative and copper (II) chloride-catalyzed thermal process. Chemosphere (71) 1100-1109.

67. Rappe,C., Marklund,S., Buser,H.-R., Bosshardt,H.-P. (1978). Formation of polychlorinated dibenzo-p-dioxins (PCDDs) and dibenzofurans (PCDFs) by burning or heating chlorophenates. Chemosphere (7) 269-281.

68. Jansson,B., Sundström,G., Ahling,B. (1978). Formation of polychlorinated dibenzo-p-dioxins during combustion of chlorophenol formulations. The Science of the Total Environment (10) 209-217.

69. Tuppurainen,K., Halonen,I., Ruuskanen,J. (1996). Dioxaspiro-type structures as intermediates in the formation of polychlorinated dibenzo-p-dioxins. A semi-empirical AM1 molecular orbital study. Chemosphere (32) 1349-1356.

65

REFERENCES

70. Wiater,I. Louw,R. (1999). Reactions of diphenyl ether with chlorine and bromine atoms around 750 K - Relevance for gas-phase "dioxin" formation. European Journal of Organic Chemistry , 261-265.

71. Wiater-Protas,I. Louw,R. (2001). Gas-phase chemistry of chlorinated phenols - Formation of dibenzofurans and dibenzodioxins in slow combustion. European Journal of Organic Chemistry , 3945-3952.

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

73. Jay,K., Stieglitz,L. (1991). On the mechanisms of formation of polychlorinated aromatic compounds with copper (II) chloride. Chemosphere (22) 987-996.

74. Hell,K., Stieglitz,L., Dinjus,E. (2001). 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 (35) 3892-3898.

75. Iino,F., Imagawa,T., Takeuchi,M., Sadakata,M., Weber,R. (1999). Formation rates of polychlorinated dibenzofurans and dibenzo-p-dioxins from polycyclic aromatic hydrocarbons, activated carbon and phenol. Chemosphere (39) 2749-2756.

76. Albrecht,I.D., Naikwadi,K., Karasek,F.W. (1992). Investigation of the De Novo process using 13C amorphous carbon. Organohalogen Compounds (8) 217-220.

77. Albrecht,I.D., Karasek,F.W. (1995). Effects of solvent extractions on the thermal formation of PCDD/F on flyash samples enriched with 13C-labelled amorphous carbon. Organohalogen Compounds (23) 323-327.

78. Wikström,E., Ryan,S., Touati,A., Gullett,B.K. (2004). In situ formed soot deposit as a carbon source for polychlorinated dibenzo-p-dioxins and dibenzofurans. Environmental Science & Technology (38) 2097-2101.

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

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

81. Procaccini,C., Bozzelli,J.W., Longwell,J.P., Sarofim,A.F., Smith,K.A. (2003). Formation of chlorinated aromatics by reactions of Cl•, Cl2, and HCl with benzene in the cool-down zone of a combustor. Environmental Science & Technology (37) 1684-1689.

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

83. Hisham,M.W.M., Benson,S.W. (1995). Thermochemistry of the Deacon process. Journal of Physical Chemistry (99) 6194-6198.

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

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

86. Öberg,T., Bergbäck,B., Öberg,E. (2007). Different catalytic effects by copper and chromium on the formation and degradation of chlorinated aromatic compounds in fly ash. Environmental Science & Technology (41) 3741-3746.

66

REFERENCES

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

88. Hagenmaier,H., Brunner,H., Haag,R., Kraft,M. (1987). Copper-catalyzed dechlorination/hydrogenation of polychlorinated dibenzo-p-dioxins, polychlorinated dibenzofurans, and other chlorinated aromatic compounds. Environmental Science & Technology (21) 1085-1088.

89. Nonhebel,D.C. (1970). Copper-catalysed single-electron oxidations and reductions. In: Hey,D.H., Waters,W.A., Norman,R.O.C. (eds), Essays on free-radical chemistry, Chemical Society, Special Publication no 24, London, pp. 409-437.

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

91. Weber,R., Nagai,K., Nishino,J., Shiraishi,H., Ishida,M., Takasuga,T., Konndo,K., Hiraoka,M. (2002). Effects of selected metal oxides on the dechlorination and destruction of PCDD and PCDF. Chemosphere (46) 1247-1253.

92. Weber,R., Takasuga,T., Nagai,K., Shiraishi,H., Sakurai,T., Matuda,T., Hiraoka,M. (2002). Dechlorination and destruction of PCDD, PCDF and PCB on selected fly ash from municipal waste incineration. Chemosphere (46) 1255-1262.

93. Kaune,A., Lenoir,D., Nikolai,U., Schramm,K.-W., Kettrup,A. (1995). Seven-year decreasing trend of PCDD/F emissions from a waste incinerator showing importance of primary measures. Organohalogen Compounds (23) 477-480.

94. Gullett,B.K., Lemieux,P.M., Dunn,J.E. (1994). Role of combustion and sorbent parameters in prevention of polychlorinated dibenzo-p-dioxin and polychlorinated dibenzofuran formation during waste combustion. Environmental Science & Technology (28) 107-118.

95. Addink,R., Paulus,R.H.W.L., Olie,K. (1996). Prevention of polychlorinated dibenzo-p-dioxins/dibenzofurans formation on municipal waste incinerator fly ash using nitrogen and sulfur compounds. Environmental Science & Technology (30) 2350-2354.

96. Ruokojärvi,P., Halonen,I., Tuppurainen,K., Tarhanen,J., Ruuskanen,J. (1998). Effect of gaseous inhibitors on PCDD/F formation. Environmental Science & Technology (32) 3099-3103.

97. Hedman,B., Burvall,J., Nilsson,C., Marklund,S. (2007). PCDD/F in source-sorted waste fractions and emissions from their co-combustion with reed canary-grass. Waste Management (27) 1580-1592.

98. Liljelind,P., Unsworth,J., Maaskant,O., Marklund,S. (2001). Removal of dioxins and related aromatic hydrocarbons from flue gas streams by absorption and catalytic destruction. Chemosphere (42) 615-623.

99. Bechtler,R., Stieglitz,L., Zwick,G., Will,R. (1996). The influence of elemental sulfur on the de-novo-synthesis and the oxidation of residual carbon on fly ash. Organohalogen Compounds (27) 10-13.

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

101. Raghunathan,K., Gullett,B.K. (1996). Role of sulfur in reducing PCDD and PCDF formation. Environmental Science & Technology (30) 1827-1834.

67

REFERENCES

102. Ryan,S.P., Li,X.-D., Gullett,B.K., Lee,C.W., Clayton,M., Touati,A. (2006). 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 (40) 7040-7047.

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

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

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

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

107. Marklund,S., Söderström,G., Ljung,K., Rappe,C., Kraft,M., Hagenmaier,H. (1992). Parallel sampling for dioxins using various sampling techniques at a Swedish municipal solid waste incinerator. Waste Management & Research (10) 21-36.

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

109. Jansson,S., Fick,J., Marklund,S. (2008). Formation and chlorination of polychlorinated naphthalenes (PCNs) in the post-combustion zone during MSW combustion. Chemosphere (72) 1138-1144.

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

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

112. Schneider,M., Stieglitz,L., Will,R., Zwick,G. (1998). Formation of polychlorinated naphthalenes on fly ash. Chemosphere (37) 2055-2070.

113. Rappe,C. (1992). Sources of PCDDs and PCDFs. Introduction. Reactions, levels, patterns, profiles and trends. Chemosphere (25) 41-44.

114. Lundholm, K. 2007. Fate of Cu, Cr, As and some other trace elements during combustion of recovered waste fuels. Ph.D. thesis, Department of Applied Physics and Electronics, Umeå University, Sweden.

115. Lee,C.W., Kilgroe,J.D., Raghunathan,K. (1998). Effect of soot and copper combustor deposits on dioxin emissions. Environmental Engineering Science (15) 71-84.

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

117. Saeki,Y., Matsuzaki,R., Fujiwara,S. (1978). Reaction process between iron(II)chloride and oxygen. Bulletin of the Chemical Society of Japan (51) 3527-3529.

68

REFERENCES

118. Chang,S.H., Yeh,J.W., Chein,H.M., Hsu,L.Y., Chi,K.H., Chang,M.B. (2008). PCDD/F adsorption and destruction in the flue gas streams of MWI and MSP via Cu and Fe catalysts supported on carbon. Environmental Science & Technology (42) 15-5727.

119. Ryu,J.-Y., Mulholland,J.A., Dunn,J.E., Iino,F., Gullett,B.K. (2004). Potential role of chlorination pathways in PCDD/F formation in a municipal waste incinerator. Environmental Science & Technology (38) 5112-5119.

120. Wang,Z.-Y., Zhai,Z.-C., Wang,L.-S., Chen,J.-L., Kikuchi,O., Watanabe,T. (2004). Prediction of gas phase thermodynamic function of polychlorinated dibenzo-p-dioxins using DFT. Journal of Molecular Structure - Theochem (672) 97-104.

121. Luijk,R., Dorland,C., Smit,P., Jansen,J., Govers,H.A.J. (1994). The role of bromine in the De novo synthesis in a model fly ash system. Chemosphere (28) 1299-1309.

122. Thompson,D., Ewan,B.C.R. (2007). A group additivity algorithm for polychlorinated dibenzofurans derived from selected DFT analyses. Journal of Physical Chemistry A (111) 5043-5047.

123. Zhai,Z., Wang,Z. (2005). Computational study on the relative stability and formation distribution of 76 polychlorinated naphthalene by density functional theory. Journal of Molecular Structure - Theochem (724) 221-227.

124. Zhang,Q., Li,S., Qu,X., Shi,X., Wang,W. (2008). A quantum mechanical study on the formation of PCDD/Fs from 2-chlorophenol as precursor. Environmental Science & Technology (42) 7301-7308.

125. Procaccini,C., Bozzelli,J.W., Longwell,J.P., Smith,K.A., Sarofim,A.F. (2000). Presence of chlorine radicals and formation of molecular chlorine in the post-flame region of chlorocarbon combustion. Environmental Science & Technology (34) 4565-4570.

126. Cieplik,M.K., Epema,O.J., Louw,R. (2002). Thermal hydrogenolysis of dibenzo-para-dioxin and dibenzofuran. European Journal of Organic Chemistry , 2792-2799.

127. Unsworth,J.F., Dorans,H. (1993). Thermodynamic data for dioxins from molecular modelling computations: prediction of equilibrium isomer composition. Chemosphere (27) 351-358.

128. Thompson,D. (1994). Thermodynamic considerations in dibenzodioxin and dibenzofuran formation: Concentrations of chlorinated dioxins and furans in model fuel-rich combustion gases. Chemosphere (29) 2583-2595.

129. Tan,P., Hurtado,I., Neuschütz,D., Eriksson,G. (2001). Thermodynamic modeling of PCDD/Fs formation in thermal processes. Environmental Science & Technology (35) 1867-1874.

130. Tysklind,M., Lundgren,K., Rappe,C., Eriksson,L., Jonsson,J., Sjöström,M., Ahlborg,U.G. (1992). Multivariate characterization and modeling of polychlorinated dibenzo-p-dioxins and dibenzofurans. Environmental Science & Technology (26) 1023-1030.

131. Li,X., Zhang,J., Yan,J., Cen,K., Ryan,S.P., Gullett,B.K., Lee,C. (2007). Experimental and modeling study of de novo formation of PCDD/PCDF on MSW fly ash. Journal of Environmental Sciences (19) 117-122.

132. Trygg,J., Wold,S. (2003). O2-PLS, a two-block (X-Y) latent variable regression (LVR) method with an integral OSC filter. Journal of Chemometrics (17) 53-64.

133. Yang,G., Zhang,X., Wang,Z., Liu,H., Ju,X. (2006). Estimation of the aqueous solubility (-lgSw) of all polychlorinated dibenzo-furans (PCDF) and

69

REFERENCES

polychlorinated dibenzo-p-dioxins (PCDD) congeners by density functional theory. Journal of Molecular Structure - Theochem (766) 25-33.

134. Chen,J., Harner,T., Ding,G., Quan,X., Schramm,K.-W., Kettrup,A. (2004). Universal predictive models on octanol-air partition coefficients at different temperatures for persistent organic pollutants. Environmental Toxicology and Chemistry (23) 2309-2317.

135. Wang,Z.-Y., Zhai,Z.-C., Wang,L.-S. (2005) Prediction of gas phase thermodynamic properties of polychlorinated dibenzo-furans by DFT. Journal of Molecular Structure - Theochem (725) 55-62.

136. Govers,H.A.J., Luijk,R., Evers,E.H.G. (1990). Calculation of heat of vaporization, molar volume and solubility parameter of polychlorodibenzo-p-dioxins. Chemosphere (20) 287-294.

137. Govers,H.A.J., Krop,H.B. (1998). Partition constants of chlorinated dibenzofurans and dibenzo-p-dioxins. Chemosphere (37) 2139-2152.

138. Simca P+ 12. 2008. Umetrics AB, Umeå, Sweden.

139. Ryu,J.-Y., Mulholland,J.A., Oh,J.-E., Nakahata,D.T., Kim,D.-H. (2004): Prediction of polychlorinated dibenzofuran congener distribution from gas-phase phenol condensation pathways. Chemosphere (55) 1447-1455.

70

APPENDIX 1

APPENDIX 1 Overview of the experimental runs underlying this thesis.

Run Description Sampling Ports

Sampling Temp. [°C]

Quench Time [sec]

Paper

P3 400 1.4 P4 300 2.3

1 Background check (propane combustion)

P7 200 4.4

I

2 Extended baseline profile P4 300 2.3 I 3-7a Baseline profile P3 400 1.4 P4 300 2.3 P7 200 4.4

I

8-9a Low-temperature profile P3 300 1.4 P4 200 2.7 P7 100 5.4

I

P3 450 1.3 P4 300 2.2

10-12a High-initial temperature profile

P7 200 4.3

I-IV

10b P3 450 1.3

Dibenzo-p-dioxin injection P4 300 2.2

P7 200 4.3

II

11b Naphthalene injection P3 450 1.3 P4 300 2.2 P7 200 4.3

III

12-17b P3 450 1.3 P4 300 2.2

Injection of dibenzofuran, OCDD, OCDF, biphenyl, fluorene and phenanthrene P7 200 4.3

–

18b,c Heptane injection P4 300 2.2 III High-temperature profile P3 460 1.3

P4 360 2.1 19-21a

P7 260 4.0

I

22-24a Hot flue gas sampling P2 640 0.1 I a Replicate runs. b Injection performed directly after collection of high-initial temperature profile samples. c Solvent control.

1

APPENDIX 2

APPENDIX 2 PCN analysis PC1-2N: determined by GC-HRMS using a VG AutoSpec mass spectrometer (Fisons Instruments, VG Analytical, Manchester, UK) and a Hewlett-Packard 5890 gas chromatograph equipped with a J&W Scientific DB-5MS (30 m, 0.25 mm i.d., 0.25 µm) capillary column. The injector was operated in splitless mode and set to 250°C. The oven temperature was kept at 80°C for 2 min and then increased by 8.0°C/min to 180°C, and by 12.0°C/min to 300°C, held at 1.0 min. PC3-8N: determined by GC-HRMS using a Waters AutoSpec ULTIMA NT 2000D high resolution mass spectrometer (Micromass UK Limited, Manchester, UK) and an Agilent 6890N Network GC System gas chromatograph with a J&W Scientific DB-5 (60m×0.25mm i.d.×0.25 µm film thickness) capillary column. Helium carrier gas was used at a flow rate of 1.2 mL/min. The injector was operated in splitless mode and set to 280°C. The oven temperature was kept at 190°C for 2 min and then increased by 3.0°C/min to 310°C and held for 1.0 min. The mass spectrometers were operated in electron impact ionization/selected ion monitoring mode and analytes were quantified by the isotope dilution technique using 2H-naphthalene, 13C-TriCDD and 13C-PeCDD as internal standards, and 2H-phenanthrene and 13C-TeCDD recovery standards, in the absence of 13C-labelled PCN standards in the laboratory. Native naphthalene and PCNs were used as quantification standards.

Monitored mass Compound Quantification

ion Control ion

Quantification Congeners

Naphthalene 128.00a 126.00a Naphthalene MoCN 162.0236 164.0207 2-MoCN DiCN 195.9847 197.9817 2-MoCN TriCN 229.9457 231.9427 1,2,7-TriCN TeCN 265.9038 263.9067 1,2,3,4-TeCN PeCN 299.8648 301.8618 1,2,3,6,7-PeCN HxCN 333.8258 335.8229 1,2,3,4,6,7-HxCN HpCN 367.7868 369.7839 1,2,3,4,5,6,7-HpCN

Target Analytes

OCN 403.7449 401.7479 1,2,3,4,5,6,7,8-OCN 2H Naphthalene 136.1086

13C 2,3,7-TriCDD 297.9759 299.9729 Internal

Standards 13C 1,2,3,7,8-PeCDD 367.8949 365.8978

2H Phenanthrene 188.1410 Recovery Standards 13C 1,2,3,4-TeCDD 333.9339 331.9368 a Analyzed by GC-LRMS.

1

APPENDIX 2

GC-HRMS chromatograms for Mo-HpCN in a reference sample.

TeCN

DiCN

TriCN

MoCN

PeCN

HxCN

HpCN

2- 1-

14/1

6-

13-

15/2

7-

26/1

7-

12-

23-

18-

136-

146-

135-

124-

137-

125-

126-

127-

167-

123-

138-

128-14

5-

1368

/125

6-

1467

-13

67-

1246

/124

7/12

57-

1357

-

2367

-12

45-

1267

-

1235

/135

8-

1238

-

1458

-

1258

/126

8-

1248

-

1278

-

1235

7/12

467-

1245

7-

1247

8-

1245

6-

1236

7-

1235

6-12

346-

1246

8-

1234

5-

1245

8-

1235

8/12

368-

1237

8-

1234

67/1

2356

7-

1234

57-

1235

78-

1245

68/1

2457

8-

1235

68-

1234

58-

1234

568-

1234

567-

2

APPENDIX 3

APPENDIX 3 Annotations of PCDF and PCDD congeners included in Paper IV.

PCDD PCDF PCDF D2 3- F2 1- F88 1,2,3,4,6- Mo D3 2- F3 2- F91 1,2,3,4,9- D4 1,2- F4 3- F92 1,2,3,6,7- D5 1,3-

Mo

F5 4- F94 1,2,3,6,9- D9 1,8- F7 1,3- F96 1,2,3,7,9- D10 1,9- F10 1,7- F97 1,2,3,8,9- D11 2,3- F11 1,8- F98 1,2,4,6,7-

Di

D13 2,8- F12 1,9- F99 1,2,4,6,8- D14 1,2,3- F13 2,3- F101 1,2,4,7,8- D15 1,2,4- F15 2,6- F103 1,2,4,8,9- D16 1,2,6- F16 2,7- F105 1,2,6,7,9- D17 1,2,7- F17 2,8- F107 1,3,4,6,8- D20 1,3,6- F20 3,7- F108 1,3,4,6,9- D21 1,3,7-

Di

F21 4,6- F110 1,3,4,7,9- D22 1,3,8- F22 1,2,3- F111 1,3,6,7,8- D25 1,4,7- F25 1,2,7- F113 2,3,4,6,7-

Tri

D26 1,7,8- F41 2,3,6- F114 2,3,4,6,8- D29 1,2,3,6- F44 2,4,6-

Pe

F115 2,3,4,7,8- D30 1,2,3,7- F45 2,4,7- F116 1,2,3,4,6,7- D32 1,2,3,9-

Tri

F46 2,4,8- F117 1,2,3,4,6,8- D37 1,2,6,7- F54 1,2,3,9- F119 1,2,3,4,7,8- D38 1,2,6,8- F56 1,2,4,7- F120 1,2,3,4,7,9- D39 1,2,6,9- F59 1,2,6,7- F121 1,2,3,4,8,9- D40 1,2,7,8- F62 1,2,7,8- F122 1,2,3,6,7,8- D41 1,2,7,9- F66 1,3,4,7- F123 1,2,3,6,7,9- D42 1,2,8,9- F67 1,3,4,8- F124 1,2,3,7,8,9- D43 1,3,6,8- F68 1,3,4,9- F125 1,2,4,6,7,8- D44 1,3,6,9- F69 1,3,6,7- F126 1,2,4,6,7,9- D45 1,3,7,8- F70 1,3,6,8- F127 1,2,4,6,8,9- D46 1,3,7,9- F75 1,4,6,8- F129 1,3,4,6,7,8- D47 1,4,6,9- F76 1,4,6,9- F130 1,3,4,6,7,9- D48 1,4,7,8- F79 2,3,4,6-

Hx

F131 2,3,4,6,7,8-

Te

D49 2,3,7,8- F80 2,3,4,7- F132 1,2,3,4,6,7,8- D50 1,2,3,4,6- F81 2,3,4,8- F133 1,2,3,4,6,7,9- D51 1,2,3,4,7- F82 2,3,6,7- F134 1,2,3,4,6,8,9- D52 1,2,3,6,7- F84 2,3,7,8-

Hp

F135 1,2,3,4,7,8,9- D53 1,2,3,6,8- F85 2,4,6,7- D54 1,2,3,6,9- F86 2,4,6,8- D55 1,2,3,7,8-

Te

F87 3,4,6,7- D56 1,2,3,7,9- D57 1,2,3,8,9- D60 1,2,4,6,9-

Pe

D61 1,2,4,7,8- D64 1,2,3,4,6,7- D65 1,2,3,4,6,8- D66 1,2,3,4,6,9- D67 1,2,3,4,7,8- D68 1,2,3,6,7,8-

Hx

D71 1,2,3,7,8,9- D74 1,2,3,4,6,7,8- Hp D75 1,2,3,4,6,7,9-

1

thermal formation and chlorination of dioxins and dioxin …142298/fulltext01.pdf · thermal...

Documents