This principle has been applied in the synthesis of 3NBAO, as described ...... Nielsen, T. Risø-M-2420; Karakterisering Af Polycyclisk Organisk Materiale. (POM) i ...
Risø-R-1183(EN)
Atmospheric Chemistry of Polycyclic Aromatic Compounds with Special Emphasis on Nitro Derivatives Anders Feilberg
Risø National Laboratory, Roskilde April 2000
Abstract. Field measurements of nitro-polycyclic aromatic hydrocarbons (nitro-PAHs) and other polycyclic aromatic compounds (PAC) have been carried out at a semi-rural site and at an urban site. A combination of correlation analyses, PAC indicators, and PAC ratios has been used to evaluate the importance of various sources of nitro-PAHs in ambient air. A major source of nitro-PAHs in urban, heavily traffic-influenced air as well as semi-rural air is atmospheric transformation of PAHs initiated by OH radicals. Especially during long-range transport (LRT) of air pollution from Central Europe, the nitro-PAH composition in Denmark is dominated by nitro-PAHs formed in the atmosphere. Locally emitted nitro-PAHs are primarily from diesel vehicles. Levels of unsubstituted PAHs can also be strongly elevated in connection with LRT episodes. Particularly for non-urban areas, this has a profound effect on the air quality. The ratio of 2-nitrofluoranthene relative to 1-nitropyrene is proposed as a measure of the relative photochemical age of particulate matter. Using this ratio, the relative mutagenicity of particle extracts appears to increase with increasing photochemical age. In connection with the field measurements, a sensitive and selective method for measuring nitro-PAHs in particle extracts based on MS-MS mass spectrometric detection has been developed. The atmospheric chemistry of nitronaphthalenes has been investigated with a smog chamber system combined with simulation with photochemical kinetics software. A methodology to implement gas-particle partitioning in a model based on chemical kinetics is described. Equilibrium constants (KP) for partitioning of 1- and 2nitronaphthalene between the gas phase and diesel exhaust particles have been determined. The gas-particle partitioning equilibrium of nitronaphthalenes, representing relatively volatile nitro-PAHs, is demonstrated to be maintained even if the compounds decay fast in the gas-phase. Mass transfer between the two phases (gas and particulate) appears to occur on a very short timescale. The rate of gas phase direct photolysis of the nitronaphthalenes depends upon the orientation of the nitro-substituent relative to the aromatic plane. Consequently, significantly faster photolysis of 1-nitronaphthalene than of 2-nitronaphthalene is observed. Finally, an approach to model the nitronaphthalene chemistry in ambient air is proposed. The photochemistry of nitro-PAHs, and to some extent other PAC, associated with organic aerosols, such as combustion aerosols, has been studied with chemical model systems simulating the organic fraction of ambient aerosols. Direct photolysis of particle-associated nitro-PAHs can not explain the degradation on combustion-generated aerosols. A number of aerosol constituents, including substituted phenols, benzaldehydes, and oxy-PAHs, are demonstrated to accelerate the photodegradation. A mechanism involving radical chain reactions initiated by electronically excited carbonylcontaining compounds is most consistent with smog chamber observations. The photodegradation of nitro-PAHs and other PAC associated with organic aerosols are strongly dependent on the physical state of the organic medium. Viscosity, temperature, and water content are of importance for the degradation rates. Nitro-PAHs are photochemically reduced to amino-PAHs under conditions of low O2 concentrations and high polarity as demonstrated by using glycerine as a surrogate for the organic fraction of ambient aerosols. Under several experimental conditions it is observed that the PAH decay is much slower than the nitro-PAH decay. Pulse radiolysis experiments have shown that reactions with OH radicals in atmospheric water droplets represent an important sink for water soluble PAC. Rate constants for reactions of N-PAC with OH radicals and for the subsequent reactions of the OHadducts with O2 have been determined.
ISBN 87-550-2700-8 ISBN 87-550-2701-6(Internet) ISSN 0106-2840 Information Service Department, Risø, 2000
Contents Preface 5 Dansk resumé. 6 Motivation and Purpose. 8 List of Abbreviations. 9 1. Background and Introduction. 10 1.1 1.2 1.3 1.4 1.5
Polycyclic aromatic compounds. 10 Primary sources of PAC in air. 11 Occurrence of airborne PAC. 12 Atmospheric transformation and fate. 17 Health effects of airborne polycyclic aromatic compounds. 22
2. Experimental 26 2.1 2.2 2.3 2.4 2.5
Field measurements. 26 Smog chamber experiments. 33 Sampling Artifacts 34 Photodegradation experiments. 36 Pulse radiolysis experiments. 38
3. Field measurements. Results and discussion. 40 3.1 Levels of PAC in urban and semi-rural air. 40 3.2 Separation of 2NF and 3NF. 43 3.3 Correlations with other PAC, inorganic components and meteorological parameters. 43 3.4 Sources and sinks for PAHs, nitro-PAHs and other PAC. 46 3.5 Impact of long-range transport from the continent. 52 3.6 Relation between PAC and mutagenicity levels. 54 3.7 Other significant observations. 56 3.8 Conclusions of Chapter 3. 57 4. Smog Chamber Experiments. Results and Discussion. 59 4.1 4.2 4.3 4.4 4.5 4.6 4.7
Sample system evaluation. 59 Gas-particle partitioning of 1NN and 2NN. 59 Emission of nitronaphthalenes from a diesel vehicle. 63 Direct photolysis of nitronaphthalenes. 63 Formation of nitronaphthalenes. 66 Sources of 1-nitronaphthalene. 69 Conclusions of Chapter 4. 70
Risø-R-1183(EN)
3
5. Effect of Aerosol Chemical composition on the Photodegradation of Nitro-PAHs. Results and Discussion. 71 5.1 5.2 5.3 5.4 5.5 5.6 5.7
Relevance of the applied chemical model systems. 71 Photochemistry in the absence of aerosol constituents. 73 Effect of carbonyl-containing compounds. 75 Influence of photosensitizers. 76 Effect of substituted phenols. 77 Comparison with smog chamber experiments. 80 Conclusions of Chapter 5. 81
6 Physical Factors Influencing the Photostability of Selected PAC. Results and Discussion. 82 6.1 Influence of viscosity and temperature on PAC photodegradation. 83 6.2 Photoreduction of nitro-PAHs - Formation of polycyclic aromatic amines. 89 6.3 Influence of water content in the organic phase. 91 6.4 Influence of polarity on the photodegradation. 93 6.5 Conclusions of Chapter 6. 95 7 Aqueous Phase Reactions of N-PAC. Results and Discussion. 96 7.1 Kinetics of aqueous phase reactions of 5-nitroquinoline and quinoline. 96 7.2 Atmospheric implications. 99 7.3 Conclusions of Chapter 7. 101 8 Summary and Major Conclusions. 102 References 105 Appendix 1 115
4
Risø-R-1183(EN)
Preface The current thesis presents results obtained during my Ph.D.-project carried out at Risø National Laboratory. I have been enrolled as a Ph.D.-student at the University of Southern Denmark – Odense University. My Ph.D.-project has been financed by the Danish Research Academy and Risø National Laboratory. The experimental work has been carried out at Risø National Laboratory and at the University of North Carolina at Chapel Hill (USA). Many people have supported me in various ways during my Ph.D.-project: I thank Jerzy Holcman and Knud Sehested (Risø) for a collaboration on the pulse radiolysis experiments. Jan Tønnesen (Miljøkontrollen, Copenhagen) is acknowledged for sampling particulate matter at H. C. Andersens Boulevard and for providing measurements of various air pollution components. Gunnar Jensen (Risø) is acknowledged for providing meteorological data. Henrik Skov is acknowledged for providing measurements of gaseous pollutants at Risø. I thank Mona-Lise Binderup (Institute of Food Safety and Toxicology) for contributing with mutagenicity tests of ambient air samples. The measurements of nitro-PAHs at Risø was carried out as a joint project with Morten Poulsen (University of Copenhagen), whom I wish to thank for a fruitful collaboration. I am grateful to laboratory apprentices, employees, and other Ph.D.-students at the former Section for Chemical Reactivity at Risø for kindness and assistance during my project. I am indebted to Professor Richard M. Kamens (University of North Carolina) for giving me the opportunity to carry out investigations in the UNC smog chamber facility, for assisting me with these, and for general support and inspiration. Richard Kamens is also acknowledged for lending me a photochemical turntable setup used for some of the experiments described in this thesis. I wish to thank Keri Leach, Myoseon Jang, Kristin Fletcher, Bharad Chandramouli, and especially Michael R. Strommen (all UNC) for assistance with the smog chamber experiments. Christian Lohse (SDU-OU) has been my University supervisor and I wish to thank him for support throughout the Ph.D.-project. Most of all I sincerely wish to thank my supervisor at Risø, Torben Nielsen, for guidance, support, encouragement, and inspiration throughout the project. Finally, I wish to thank my family, Rie and Gunnar, for their love and support at all times. Four manuscripts based on the work presented in this thesis has been published in peerreviewed journals (1-4). A manuscript dealing with the majority of the results presented in Chapter 3 has been accepted by Atmospheric Environment and a manuscript based on the results presented in Chapter 6 has been submitted to Environmental Science and Technology. Reprints are available on request. This thesis has been slightly modified from the thesis that was defended at the University of Southern Denmark – Odense, November 26th 1999, and accepted as partial fulfilment of the requirements to obtain a Ph.D.-degree. The opponents for the defense were: Dr. Henrik Skov (National Environmental Research Institute, Denmark), Dr. Eva Brorström-Lundén (Institut for Vatten- och Luftvårdsforskning, Gothenburg, Sweden), and Dr. Torben Stroyer Hansen (University of Southern Denmark – Odense).
Risø-R-1183(EN)
5
Dansk resumé. Feltmålinger af nitro-polycykliske aromatiske kulbrinter (nitro-PAH) og andre polycykliske aromatiske forbindelser (PAC) er blevet udført på to forskellige lokaliteter, en storby-lokalitet og en lokalitet i et landområde. En kombination af korrelationsanalyser, anvendelse af PAC-indikatorer og PAC koncentrationsforhold er blevet anvendt for at kunne vurdere betydningen af forskellige kilder til nitro-PAH i luften. En vigtig kilde på begge lokaliteter er atmosfærisk omdannelse af PAH initieret af OH radikaler. Især under langtransport af forurenet luft er nitro-PAH sammensætningen i Danmark domineret af nitro-PAH dannet i atmosfæren. Lokalt emitterede nitro-PAH stammer fortrinsvis fra dieselkøretøjer. Niveauerne af de usubstituerede PAH kan også være kraftigt forøgede i forbindelse med langtransport-episoder, og særligt for landområder har dette stor betydning for luftkvaliteten. Forholdet mellem 2-nitrofluoranthene og 1-nitropyrene er foreslået som et relativt mål for den fotokemiske alder af atmosfæriske partikler. Anvendelse af dette forhold tyder på, at den relative mutagenicitet af partikel-ekstrakter stiger med stigende fotokemisk alder. I forbindelse med feltmålingerne er en følsom og selektiv metode til bestemmelse af nitro-PAH i partikel ekstrakter, baseret på MS-MS masse-spektrometrisk detektion, blevet udviklet. Atmosfærekemien af nitro-naphthalener er blevet undersøgt i et smog-kammer system under anvendelse af modellerings-software. En metode til at implementere gas-partikel ligevægte i en fotokemisk model baseret på kinetik-udtryk er beskrevet. Ligevægtskonstanter (KP) for fordelingen af 1- og 2-nitronaphthalen mellem gasfase og partikler fra dieseludstødning er blevet bestemt. Det er vist, at gas-partikel fordelingen er i ligevægt selv under omstændigheder hvor gasfase-komponenten nedbrydes hurtigt. Det fremgår, at stofudvekslingen mellem de to faser foregår på en kort tidsskala. Fotolysehastigheden i gasfasen afhænger af orienteringen af nitro-substituenten i forhold til det aromatiske plan. Dette medfører, at 1-nitronaphthalen fotolyserer væsentligt hurtigere end 2nitronaphthalen. Resultaterne tyder på, at atmosfærekemien af nitronaphthalenerne i luft kan modelleres ved hjælp af de foreslåede typer af mekanismer. Fotokemien af nitro-PAH, samt i mindre grad andre PAC, bundet til organiske aerosoler (f. eks. forbrændingspartikler) er blevet undersøgt ved hjælp af kemiske modelsystemer, der simulerer den organiske fraktion af atmosfæriske partikler. Direkte fotolyse af partikel-bundne nitro-PAH kan ikke forklare nedbrydningen på partikler fra diesel udstødning og brændeovne. Forskellige kendte aerosol-bestanddele, såsom substituerede phenoler, benzaldehyder og oxy-PAH, forstærker nedbrydningshastigheden. En mekanisme, der involverer radikal-kædereaktioner initieret af exciterede carbonyl-forbindelser er mest i overensstemmelse med resultater af smogkammer-forsøg med nitro-PAH (5). Den fotolytiske nedbrydning af nitro-PAH og andre PAC sorberet til organiske partikler i atmosfæren forudsiges at afhænge kraftigt af fysiske forhold vedrørende det sorberende medium. Viskositet, temperatur og vandindhold er faktorer af betydning for nedbrydningshastigheden. Nitro-PAH reduceres fotokemisk til amino-PAH under betingel-
6
Risø-R-1183(EN)
ser med lav iltkoncentration og høj polaritet, som vist ved at anvende glycerin som surrogat for organiske aerosoler. Under flere forskellige eksperimentelle betingelser er nedbrydningen af PAH meget langsommere end nedbrydningen af nitro-PAH. Forsøg i vandig opløsning under anvendelse af pulseret radiolyse har vist, at reaktionen med OH-radikaler i vanddråber udgør en vigtig nedbrydningsvej for vandopløselige PAC. Hastighedskonstanter for reaktioner af N-PAC med OH radikaler den efterfølgende reaktion med O2 er blevet bestemt.
Risø-R-1183(EN)
7
Motivation and Purpose. This Ph.D.-thesis describes experimental studies of the atmospheric chemistry of polycyclic aromatic compounds (PAC) with special emphasis on nitrated polycyclic aromatic hydrocarbons (nitro-PAHs). Many PAC are classified as possible or probable carcinogens by the International Agency for Research on Cancer (6,7). Knowledge about the occurrence and composition of PAC as well as degradation pathways is therefore important. Nitro-PAHs constitute a group of potent direct-acting mutagens in the widely applied Ames test and several of the compounds are also carcinogenic (see Chapter 1 for further information). In order to estimate the risk of exposure to nitro-PAHs relative to the risk of other (more abundant) PAC, especially unsubstituted polycyclic aromatic hydrocarbons (PAHs), experimental data on the atmospheric chemistry of nitro-PAHs (and other PAC) are therefore crucial. However, for nitro-PAHs there is a lack of knowledge about the source contributions under varying conditions, the occurrence in ambient air, and factors and processes which influence the environmental fate of these compounds. Especially the fate of nitro-PAHs associated with aerosol particles and factors that influence this fate are subjects that need investigations. From a chemical viewpoint nitro-PAHs as a group also posses some interesting properties. For example, some of the compounds are emitted directly whereas others are formed in the atmosphere giving rise to source specific isomers. Thus, the occurrence and composition of nitro-PAHs could be indicative of the importance of various atmospheric processes. The main purposes of my Ph.D.-project are as follows. •
•
To provide experimental data regarding the occurrence and composition of particleassociated nitro-PAHs and other PAC, including their relationships with other air pollution components and the importance of various sources and sinks. To provide knowledge about the influence of physical and chemical factors on the degradation of nitro-PAHs and other PAC including knowledge about the distribution of nitro-PAHs between gas and aerosol phases in the atmosphere.
These subjects are studied by a combination of laboratory experiments, field measurements and smog chamber experiments.
8
Risø-R-1183(EN)
List of Abbreviations. Some common abbreviations used in the thesis are listed below in alphabetical order. For abbreviations of PAC the reader is referred to Table 1.1. CB4 CHX CID DCM DFD DL DMF DOP EC ECD FFD GLY G-P HCAB LRT NOx gas-NOy NOz N-PAC O-PAC PAC PAH PAN PEF PKSS PUF SOC S-PAC SPI TIGF TSP UNC
Carbon Bond 4 Cyclohexane (C6H12) Collision induced dissociation Dichloromethane (CH2Cl2) Denuder-filter-denuder Detection limit N,N-Dimethylformamide Di-iso-octylphthalate (bis(2-ethyl-hexyl)phthalate, DEHP) Elemental carbon Electron capture detection Filter-filter-denuder Glycerine Gas-particle H. C. Andersens Boulevard (Copenhagen, DK) Long-range transport NO + NO2 NOx + PAN + PPN + gas-HNO3 + HONO + NO3 + 2×N2O5 + gas-RONO2 = NOy - NOx Endocyclic nitrogen-substituted PAH (aza-arenes) Endocyclic oxygen-substituted PAH (oxa-arenes) Polycyclic aromatic compounds Polycyclic aromatic hydrocarbons Peroxyacylnitrates (also: peroxyacetylnitrate) Potency equivalency factors Photochemical kinetics simulation software Polyurethane foam Semivolatile organic compounds Endocyclic sulphur-substituted PAH (thia-arenes) Septum-purged injection Teflon-impregnated glass fiber Total suspended particles University of North Carolina
Risø-R-1183(EN)
9
1. Background and Introduction. This chapter provides background material concerning the atmospheric chemistry of polycyclic aromatic compounds (PAC). It shall not be regarded as a complete review of the subject, but rather as examples of important investigations and results. Most attention has been given to subjects relevant for understanding the subsequent chapters. In order to understand the influence of airborne PAC on human health issues, the last section of this chapter describes the genotoxicological properties of selected PAC and PAC-containing mixtures, and the risk associated with PAC air pollution.
1.1 Polycyclic aromatic compounds. Polycyclic aromatic compounds (PAC) comprise organic compounds consisting of two or more fused 5- and 6-membered aromatic rings. The main subgroups of PAC are the unsubstituted polycyclic aromatic hydrocarbons (PAHs), the sulphur analogues (SPAC/thiaarenes), the nitrogen analogues (N-PAC/azaarenes), the oxygen analogues (OPAC/oxaarenes) and substituted PAHs, especially nitro-PAHs, oxy-PAHs (including polycyclic aromatic quinones), hydroxy-PAHs, polycyclic aromatic carboxaldehydes and carboxylic acids. Some examples of PAC are shown in Figure 1.1.
N
S Dibenzothiophene
Anthracene
Acridine
NO2
O
O Benzanthrone
O Anthraquinone
1-Nitropyrene
NO2
Benzo(a)pyrene
2-Nitrofluoranthene
Figure 1.1. Examples of important polycyclic aromatic compounds (PAC). It is beyond the scope of this thesis to discuss the physical and chemical properties of PAC in general, as well as the nomenclature. Interested readers are referred to comprehensive textbooks, for example Bjørseth et al. (8), Harvey (9) and - for nomenclature Loening et al. (10).
10
Risø-R-1183(EN)
Table 1.1. The most important PAC in this thesis with abbreviations. Naphthalene
Nap
Perylene
Per
Phenanthrene
Phen
Indeno(1,2,3-cd)pyrene
IP
Anthracene
Ant
Benzo(ghi)perylene
BghiPer
1-Methylphenanthrene
1MePhen
Anthanthrene
AA
2-methylphenanthrene
2MePhen
Coronene
Cor
3-methylphenathrene
3MePhen
Dibenzothiophene
DBT
4,5-methylenephenanthrene
4,5 MePhen
Benzo(b)naphtho(2,1)thiophene
BNTP
9-methylphenanthrene
9MePhen
Benzanthrone
BAO
Sum of methylphenanthrenes
ΣMePhen
Anthraquinone
AQ
Fluoranthene
Fla
1-Nitronaphthalene
1NN
Pyrene
Pyr
2-Nitronaphthalene
2NN
Benzo(ghi)fluoranthene
BghiFla
9-Nitroanthracene
9NA
Benzo(c)phenanthrene
BcPhen
2-Nitrofluoranthene
2NF
Cyclopenteno(cd)pyrene
CP
3-Nitrofluoranthene
3NF
Benz(a)anthracene
BaA
1-Nitropyrene
1NP
Chrysene
Chr
2-Nitropyrene
2NP
Triphenylene
Trp
6Nitrobenzo(a)pyrene
6NBaP
Benzo(b+j+k)fluoranthene
BbjkFla
3-Nitrobenzanthrone
3NBAO
Benzo(a)fluoranthene Benzo(e)pyrene
BaFla
5-Nitroquinoline
5NQ
BeP
Quinoline
Q
Benzo(a)pyrene
BaP
Acridine
Acr
Physical and chemical properties of PAC relevant for their atmospheric behaviour are discussed in subsequent sections. Abbreviations are generally used for the PAC studied in the current work and are listed in Table 1.1.
1.2 Primary sources of PAC in air. Numerous investigations concerning the primary emissions of PAC have been carried out in the past. Most of these have focused on unsubstituted PAHs. Sources of PAHs include residential heating, industrial production (coke, aluminium, asphalt), incineration, open fires, power generation and mobile sources (11). Table 1.2. Emission rates or particle concentrations of PAC subclasses for different particle sources. From (12-20). PAC subclass PAH Oxy-PAH N+S-PAC PAC subclass
MC kg d-1 0.5-0.8
NCA µg/km 1406 310 69 C µg/cig. 13.5
CA µg/km 52 44 58 AF µg/g 3370
D µg/km 210 207 29 B ng/kJ 0.1-1.4 0.1-0.5
TD µg/g 226.1 0.45
RD µg/g 58.68 3.83
NG PW OW ng/kJ mg/kg mg/kg PAH 1.2-5.5 11.9 6.8 Oxy-PAH 0.8-4.2 2.5 1.8 N+S-PAC 0.2-0.6 3202 MC: Meat cooking. NCA: Non-catalyst automobile. CA: catalyst equipped automobile. D: Diesel vehicle. TD: Tire debris. RD: Road dust. NG: Natural gas home appliances. C: Cigarette smoke. AF: Asphalt roofing tar pot fumes. B: Boilers burning fuel oil. PW: Pine wood combustion. OW: Oak wood combustion.
A series of papers by Rogge and co-workers (12-20) report detailed chemical analysis of the extractable organic fraction of a variety of fine organic aerosols from different
Risø-R-1183(EN)
11
sources. In Table 1.2, PAC subclasses identified in some of these extracts are listed with their emission rates or their concentrations in particle samples. From Table 1.2 however, it is not possible to deduce the relative source strengths to ambient PAHs or other PAC. However, since mobile sources seem to be among the most important contributors to ambient fine particle mass concentrations (21), it is plausible that mobile sources also contribute substantially to ambient PAH concentrations. PAH emission rates for the Los Angeles (LA) area in 1982 have been estimated to be 125.8 kg day-1 for non-catalyst-equipped automobiles, 7.4 kg day-1 for catalyst-equipped automobiles and 3.4 kg day-1 for diesel vehicles (13). Since 1982, the proportion of automobiles with catalyst has increased so the situation is probably different nowadays. From a tunnel study by Miguel et al. (22) it appears that diesel trucks is a major source of the lighter PAHs, whereas gasoline vehicles is the dominant source of the heavier PAHs. Nielsen (23) estimated that the contribution of traffic to PAH levels in a busy street in Copenhagen (DK) amounted to about 90 % on working days and about 60 % on weekends. In an adjacent city park the traffic contribution was estimated to be only ~40 %. About 2/3 of the traffic PAH emissions were estimated to be due to diesel vehicles (23). Li and Kamens (24) applied a chemical mass balance model on Paris PAH data in order to estimate the relative contribution of residential heating, diesel vehicles and gasoline driven vehicles. The relative contributions were 20 % (residential heating), 53 % (gasoline vehicles), and 27 % (diesel vehicles). Pistikopoulos et al. (25) applied a different mass balance model on the same date and obtained relative contributions of 11 % (residential heating), 57 % (gasoline vehicles), and 32 % (diesel vehicles). The PAH source contributions in sub-urban and rural areas are more uncertain, since many different PAH sources are probably important. In addition, large variations are likely to occur depending on the existence of industrial plants and other point sources. Emission sources of other PAC, such as N-PAC and S-PAC, are less certain, but the major PAH sources are also important for these compounds, since analogous to PAHs the formation is caused by incomplete combustion of organic material, in these cases involving N- and Scontaining material (26). Emissions of nitro-PAHs have caught considerable interest due to the direct-acting mutagenicity of these PAC as discussed in Section 1.5. The dominant emission source of nitro-PAHs is believed to be diesel vehicles (7,27,28), but for example 1NP has also been detected in gasoline vehicle exhaust (7). In addition, 1NP has been detected in stack gases from aluminium smelters and wood stoves, in coal fly ash, and in emissions from coal-fired power plants (7). 1NP was previously present in photocopier carbon black in significant amounts (7).
1.3 Occurrence of airborne PAC. PAC are ubiquitous atmospheric pollutants, but are also found in a variety of other environmental samples (soil, sediment, water). With respect to the global distribution and occurrence of PAC, most attention has been focused on PAHs. A large number of PAHs have been identified in ambient air all over the world both in urban environments and in background air (26). Some examples of observed levels are given in the following. Grimmer (29) reviewed the occurrence of PAHs in city air throughout the world using BaP as a marker for PAHs and found the concentrations to range from a few ng m-3 to several hundred ng m-3 close to PAH sources. Thrane and Mikaelsen (30) measured PAHs in long-range transported air arriving at Birkenes in southern Norway. Clean air
12
Risø-R-1183(EN)
from the Atlantic Ocean (at westerly winds) contained ~10 ng ΣPAH m-3, whereas air parcels arriving from the European continent or Great Britain contained ~25 ng ΣPAH m-3. Nielsen (31) determined the concentrations of PAHs at a rural site in Denmark and observed the concentration of five PAHs to depend on the wind direction with average concentrations of 0.8 ng m-3 at winds from unpolluted regions and average concentrations of 12 ng m-3 at winds from more polluted areas. The levels of BaP and BeP in Copenhagen, Denmark, was in 1992-93 determined to be 4.4±1.2 and 4.4±0.7 ng m-3, respectively, in a busy street and 1.4±0.6 and 1.3±0.3 ng m3 , respectively, in an adjacent amusement park (Tivoli) (23). Lee and Jones (32) analysed samples from a semirural maritime-influenced site in England for both vapour phase and particulate PAHs and found the total concentrations of selected PAHs (13 PAH species) to vary by over a factor of 20 (1.4-20 ng m-3). The relatively large variations in PAH levels were ascribed to variation in source strengths at different wind directions and variation in photochemical activity. Nitrated PAHs, although less extensively studied than the parent PAHs, have also been identified in ambient samples throughout the world (see for example: (33-37)). The concentration levels determined are observed to range from < 1 pg m-3 up to ~1 ng m-3 and are usually a factor of 10-1000 below the parent PAH concentrations. The concentrations of oxy-PAHs can be close to the highest PAH concentrations or even higher (38-40). It is possible that PAHs are converted to oxy-PAHs in catalysts applied by newer gasoline engines (13). S-PAC and N-PAC are usually present in ambient air in concentrations about an order of magnitude less than the corresponding PAHs (41-43). There is a strong seasonal variation in particulate PAH levels in ambient air with higher winter concentrations and lower summer concentrations (32,44,45). This can be ascribed to one or more of the following reasons: 1) The gas –particle distribution changes towards the gas phase at higher temperatures; 2) Degradation of PAHs increases with increasing solar flux; 3) Emissions could be higher in the winter due to cold engine conditions. Which of the three factors is dominating depends on compound properties, such as photostability, volatility, formation mechanism (petrogenesis or pyrogenesis) etc. It appears that the concentrations of PAHs has decreased since the early measurements in the 1950’s (42) although few long-term trend studies have been performed. Recently, Matsumoto et al. (44) reported a long term trend of BaP in Japan with a 75-80 % decline from 1974 to 1992, but no decrease in the levels of mutagenicity activity was observed. Nielsen et al. (42) suggested that the proportion of polar PAC relative to PAHs in the atmosphere has increased since the 50’ies. Gas-particle partitioning. Many PAC are able to partition between the gas phase and airborne particles/aerosols in the atmosphere depending on their vapour pressure. Organic compounds that partition between the gas and particle phases in the atmosphere are denoted semivolatile organic compounds (SOCs). The vapour pressures of compounds considered as SOCs range from ~10-4 to ~10-12 atm. Thus PAHs ranging from Nap (POL = ~10-4 atm) to (at least) BeP/BaP (POL = ~10-10 atm) have semivolatile characteristics. The vapour pressures of the heavier PAHs are not well known, but these PAHs are almost entirely associated with particles under all conditions anyway. The gas particle-partitioning of PAHs in ambient air was first investigated in detail by Yamasaki (46) who developed a gasparticle (G-P) equilibrium constant based on Langmuirian adsorption theory, where TSP (total suspended particles) is used as a surrogate for adsorption sites:
Risø-R-1183(EN)
13
C
K Y = C gas part / TSP
(1.1)
where Cgas is the gas phase concentration (ng/m3) and Cpart is the particle phase concentration (ng/m3). Since TSP is measured in units of µg/m3 or mg/m3, KY also has either of these units. Ky was observed to have a temperature dependence that could be fitted to the general formula (46): log K Y = A T1 + B
(1.2)
where A and B are empirical constants. The G-P partitioning was originally treated theoretically as gas-solid partitioning. A theoretical treatment of the partitioning between the gas and aerosol/particle phases is provided by Pankow (47). KY has subsequently been replaced by Kp (defined as KY-1) as this is more consistent with the usual convention for parameterisation of phase distributions (48): C
K p = C part gas ×TSP
(1.3)
The temperature dependence of Kp was proposed to consist of an Arrhenius expression for desorption (koff ∝ e-E/RT) and an expression for sorption based on the kinetic theory of gases (kon ∝ T-½) (47). Thus, most of the temperature dependence arise from the desorption energetics and Kp can be described by an equation analogous to Eq. 1.2. More specifically Kp for gas-solid adsorption can be expressed as (49): Kp =
Ns a tsp Te( Q l −Q v ) / RT 16P0 ( L )
(1.4)
where Ns is the surface concentration of adsorption sites (sites cm-2), atsp is the specific surface area of the particulate matter (m2 g-1), Ql is the enthalpy of desorption from the surface (kJ mol-1), Qv is the enthalpy of evaporation of the liquid (sub-cooled if necessary) compound (kJ mol-1), R is the gas constant, T is the temperature (K), P0(L) is the vapour pressure of the specific compound as a (sub-cooled) liquid. Thus, it can be anticipated that log Kp is correlated to log P0(L):
log K p, ad = m r log P0 ( L ) + b r
(1.5)
A linear relationship within compound classes, such as PAHs or linear alkanes, according to Eq. 1.5 has indeed been observed in a number of field studies reviewed recently by Goss and Schwarzenbach (50). A prerequisite of the successful application of Eq. 1.5 is that the energy difference Ql-Qv remains constant within the compound class studied. This matter has been discussed by Pankow (48) who found that the difference Ql-Qv was fairly constant for a series of PAHs. In that case it is predicted that mr = -1 for adsorption partitioning and hence br = log(NsatspT×exp(Ql-Qv)/RT). An expression analogous to Eq. 1.4 for G-P partitioning involving absorption into a liquid or amorphous phase was later derived by Pankow (49): K p,abs =
14
f om 760 RT Mw om γ om P 0 ( L )10 6
(1.6)
Risø-R-1183(EN)
where fom is the weight fraction of the absorbing phase with “om” denoting organic material although it could in principle also be inorganic, e.g. water. Mwom is the average molecular weight of the absorbing medium, and γom is the activity coefficient of the specific compound in the absorbing phase. This expression can also be rewritten into an equation analogous to Eq. 1.5: log K p, abs = − log P 0 ( L ) + log
f om 760 RT Mw om γ om 10 6
(1.7)
Eq. 1.7 predicts that when absorption partitioning dominates, a plot of log Kp vs. log P0(L) for a given compound class should yield a slope of –1 if the activity coefficients remain constant within the group (49). The same conclusion was drawn for adsorption based on Eq. 1.4 and 1.5 (47). However, in the thorough review by Goss and Schwarzenbach (50) it is demonstrated that even within a homologous series of compounds (e. g. PAHs) the slopes may deviate significantly from -1 both for absorption and adsorption. In the case of absorption, this is because it is quite likely that activity coefficients differ within compound classes (50). For adsorption of neutral organic molecules, the energy difference Ql-Qv is governed by van der Waals forces and – for some compounds – Lewis acid-base interactions between the adsorbed molecule and the surface to which it is attached. These interactions may very well change within a series of compounds (50). Goss and Schwarzenbach (50) concluded that for absorption the slope of log Kp vs. log P0(L) plots may vary between +0.2 and –1.0 whereas for adsorption the slope may vary between –0.6 and –1.1 for systems in equilibrium. They furthermore concluded that 1) the slopes of such plots for ambient data vary significantly even for the same location at different times, indicating the variability in sorbent chemical properties, and 2) based on a combination of slopes of log Kp vs. log P0(L) plots and the absolute sorption constants absorption seems to dominate over adsorption in most cases. Liang et al. (51) also concluded that, at least for urban particulate matter, absorption into an organic layer seems to be the dominating sorption process for PAHs (and alkanes). The importance of activity coefficients for absorption partitioning has been demonstrated by Jang et al. (52) who obtained very reasonable log Kp vs. log P0(L) plots for different types of compounds by adjusting Kp for the activity coefficient of the compounds in the organic fraction of the aerosols studied (diesel soot, wood smoke, secondary organic aerosol). Most of the work on G-P partitioning of PAC has been done on unsubstituted PAHs although recently N-PAC have also been studied (43). By using data obtained by Pankow and Bidleman (53) the expected particle phase fraction of a series of PAHs is calculated at T = 298 K and T = 273 K with TSP = 100 µg/m3. The results are shown in Table 1.3. Table 1.3. Vapour pressures and calculated particle phase fraction for selected PAHs at 298 K and 273 K, respectively, using TSP = 100 µg m-3 (based on Pankow and Bidleman (53)). Compound Phenanthrene Fluoranthene Pyrene Benz(a)anthracene Chrysene Benzo(b)fluoranthene Benzo(a)pyrene
Risø-R-1183(EN)
log P0(L) (atm) -6.38 -7.42 -7.61 -8.90 -8.94 -10.0 -10.25
Cpart/Ctot (298K) 0.002 0.021 0.028 0.359 0.359 0.889 0.953
Cpart/Ctot (273K) 0.040 0.320 0.395 0.944 0.944 0.997 0.999
15
From Table 1.3, it is evident that the phase distribution of PAHs of intermediate volatility (Fla to Chr) is most sensitive to temperature changes. The lighter PAHs are mostly in the gas phase even at low temperature, whereas the heavier PAHs are always predominantly in the particle phase. The N-PAC are less volatile than the analogous PAHs and are correspondingly also sorbed to particles to a higher degree (43). For example, based on the parameterisation by Chen and Preston (43) it can be calculated that 64% of azapyrene and 85% of azachrysene is in the particle phase at T = 298K and TSP = 100µg m-3, which is significantly higher than the percentages for pyrene (3 %) and chrysene (36%) from Table 1.3. The dynamic behaviour of gas-particle partitioning of PAHs and other semivolatile compounds has only been studied in a limited number of cases. Often it has been assumed that partitioning equilibrium exists, but this may not hold for all conditions. Kamens et al. (54) observed that for diesel and wood soot particles under cold conditions (~0°C) G-P partitioning could progressively deviate from equilibrium with time. This effect was ascribed to slow desorption under cold conditions so that particle-to-gas conversion could not keep up with gas phase loss. Under warmer conditions, however, the observed Kp values remained constant over time (54). The difference could be due to changes in the physical state of the particle surface as a function of temperature, i.e. the viscosity of the organic fraction of the particle might decrease at increasing temperatures (55). The dynamics of gas-particle partitioning for diesel soot particles has been modelled by radial diffusion models (56,57) with the best results obtained by a dual-impedance two-layer model (57). Some significant findings from these modelling studies are that 1) surface mass transfer is rapid and not limiting, 2) diffusion in both the outer (liquid) layer and especially the inner (porous, carbon-aggregates) layer is restricted by adsorption and tortuosity. At least for relatively volatile compounds such as fluorene and phenanthrene it seems that the dynamic gas-particle partitioning occurs on a very short time scale (58), whereas for higher molecular weight compounds longer time scales have been suggested (59). Furthermore, SOCs that have been absorbed by a diesel particle will diffuse towards the centre of the particle and therefore with time the desorption rate will be slower (57). Size distribution of particulate PAHs. The size distribution of PAHs is very important both for the fate of particle-associated PAHs and for health effects. Venkataraman and Friedlander (60,61) investigated the size distribution of PAHs at different locations. In traffic tunnels, the size distributions were very close to unimodal with the mode in the 0.05 – 0.12 µm range and thus following closely the size distribution of elemental carbon (EC) (61). In ambient air the size distribution of both PAHs and EC is bimodal with peaks in the 0.05 – 0.12 µm (mode 1) and 0.5 – 1.0 µm (mode 2) ranges (61). The more volatile of the PAHs studied were distributed more into mode 2, whereas the heavier PAHs as well as EC predominate in the smaller particles (mode 1) (61). These observations reflect that the lighter PAHs partition more into the gas phase and therefore evaporate from the primary (mode 1) particles and condense onto both mode 1 and 2 (61). However, the existence of EC and high molecular weight PAHs in mode 2 show that particle growth due to gasparticle conversions of secondary organic reaction products and water vapour contribute to the bimodal distribution (61). This was corroborated by transmission electron microscopy (TEM) photographs showing that mode 1 particles resembled primary soot agglomerates whereas mode 2 particles resembled soot agglomerates covered by a liquid film (61).
16
Risø-R-1183(EN)
Allen et al. (59) investigated the size distributions of PAHs in urban air and observed that larger PAHs (MW ≥ 228 g/mol) were found associated primarily with smaller particles ( ~19 µm. The sampling flow is measured with a top-loading orifice calibrator (model 330, Sierra Instr.) connected to a 16 inch slack tube manometer (model 305-808, Sierra Instr.) to measure the pressure drop across the orifice. The pressure reading is converted to air flow (m3 min-1) with a certified calibration curve provided with the equipment and corrections for temperature and pressure are made. The sampling flows were typically about 1.3 – 1.7 m3 min-1. The filters were weighed prior to sampling and were reweighed immediately after exposure. The filters were stored in a freezer (-18 °C) until extraction and analysis. The gravimetric procedures were done as uniformly as possible to get a rough estimate of TSP. Extraction and fractionation. The filters were cut into pieces and extracted ultrasonically with ~50 ml of dichloromethane (DCM, p.a. Merck) twice for 30 min. each time and with ~50 ml acetone (p.a. Merck) once for 30 min. The samples were protected against light during extraction and in the subsequent steps to avoid photodegradation. Some samples were then divided in two equal parts; one part was for mutagenicity testing and the other was for chemical analysis. 5 µl of solutions of internal standards were added with a microsyringe to the part for chemical analysis. For samples that were not tested for mutagenicity, the internal standards were added directly to the filters. Perdeuterated internal standards applied were dibenzothiophene-d8, fluoranthene-d10, triphenylene-d12, perylene-d12 and coronene-d12 for the analysis of native PAHs and oxy-PAHs and 2-nitrofluoranthene-d9 for the nitro-PAHs.
26
Risø-R-1183(EN)
Ultrasonic filter extraction
Raw extract in DCM/acetone Rotavap + N2 blowdown
Rotavap + N2 blowdown
Extract for muta-genicity testing (5 ml)
Preconcentrated extract Add. CHX + N2 blowdown
CHX extract (1 ml) Three extractions with 1 ml DMF/H2O (9:1) Aliphatic fraction
3 ml DMF/H2O (9:1) extract Add. 2.4 ml H2O → DMF/H2O = 1:1 Three extractions with 3 ml CHX Polar fraction
Moderately polar fraction Wash w. H2O. Dry w. Na2SO4 N2 blowdown → 200 µl.
HPLC fractionation on Nucleosil CHX/toluene gradient elution.
50-5
with
Rotavap + N2 blowdown
PAH fraction ~100 µl
nitro-PAH fraction ~25 µl
Figure 2.1. Extraction and fractionation procedure for field samples collected in this work.
Risø-R-1183(EN)
27
The extracts were purified by a liquid-liquid extraction procedure (41). The samples collected during 1998 (Risø) were further purified by means of HPLC fractionation (137). The complete extraction and clean-up procedures are outlined on Figure 2.1. First, the extracts were concentrated to ~ 5 ml by rotary evaporation and further concentrated to 1 ml under a stream of N2 at 40 °C (N2 blowdown). Then 2 ml of cyclohexane (CHX) were added and the residues of DCM and acetone were evaporated together with CHX by N2 blowdown to a final volume of 1 ml. This solution was extracted three times with a 1 ml mixture of dimethylformamide (p.a. Merck) and millipore water. The DMF/H2O ratio was 9:1. The nonpolar aliphatic fraction remains in CHX, while the polar and moderately polar fractions are extracted. The combined extracts (3 ml) were added 2.4 ml H2O to reach a DMF/H2O ratio of 1:1. This solution was extracted with 3 ml CHX three times, to remove the moderately polar PAC from the polar fraction. The combined cyclohexane extracts were washed with H2O, dried with anhydrous Na2SO4 and concentrated by N2 blowdown to approximately 1 ml in case of the HCAB samples or 200 µl in case of the Risø samples. The HCAB samples were analysed for selected PAC, as described later. To analyse for nitro-PAHs it was necessary to concentrate the samples further to a volume of ~100 µl. Table 2.1. Quantified compounds and their respective I.S. Compound Dibenzothiophene (DBT) Phenanthrene (Phen) Anthracene (Ant) 1-Methylphenanthrene (1-MePhen) 2-Methylphenanthrene (2- MePhen) 3-Methylphenanthrene (3- MePhen) 9-Methylphenanthrene (9- MePhen) 4,5-Methylene-phenanthrene (4,5- MePhen) Fluoranthene (Fla) Pyrene (Pyr) Anthraquinone (AQ) Benzo(2,1-b)naphthothiophene (BNTP) Benzo(ghi)fluoranthene (BghiFla) Benzo(c)phenanthrene (BcPhen) Cyclopenteno(cd)pyrene (CP) Benz(a)anthracene (BaA) Chrysene/Triphenylene (Chr/Trp) Benzanthrone (BAO) Benzo(b+j+k)fluoranthene (BbjkFla) Benzo(a)fluoranthene (BaFla) Benzo(a)pyrene (BaP) Benzo(e)pyrene (BeP) Perylene (Per) Dibenz(ac+ah)anthracene (DBacahAnt) Dibenz(aj)anthracene (DbajAnt) Indeno(1,2,3-cd)pyrene (IP) Benzo(ghi)perylene (BghiPer) Anthanthrene (AA) Coronene (Cor) Nitro-PAHs analysed by GC-MS-MS: 9-Nitroanthracene (9NA) 2-Nitrofluoranthene (2NF) 3-Nitrofluoranthene (3NF) 1-Nitropyrene (1NP) 2-Nitropyrene (2NP)
Quant. ion 184 178 178 192 192 192 192 190 202 202 208 234 226 228 226 228 228 230 252 252 252 252 252 278 278 276 276 276 300 193 189 191 191 189
I.S.
Quant. ion
DBT-d8
192
Fla-d10
212
Trp-d12
240
Per-d12
264
Cor-d12
312
2NF-d9
198
The Risø samples were further fractionated by high performance liquid chromatography (HPLC). The fractionation procedure was based on the method developed by Nielsen (137). The HPLC system consisted of a low pressure gradient Shimadzu LC10 equipped with a diode array UV-vis detector and thermostated oven. The column (25 cm × 8.0 mm) was packed with Nucleosil Si-50-5 silica gel and the mobile phase program em-
28
Risø-R-1183(EN)
ployed was: cyclohexane/toluene (9:1) for 10 min., then a linear gradient to 100 % toluene over 25 min. and held at 100 % toluene for 30 min. The PAH fraction was collected between 6 and 12 min. and the nitro-PAH fraction was collected between 18.5 and 26 min. Prior to a series of fractionations the column was activated with 50-100 ml each of CH3CN, Ethyl acetate, CH2Cl2 and toluene. The samples were injected manually into a 500 µl loop. The fractions were collected with a Gilson fraction collector (model 201).It is important that the stationary phase is saturated with toluene to avoid retention time drifts (137). Since the more polar components are eliminated in the liquid-liquid fractionation procedure, the column was not eluted with a polar solvent between runs as the oxy-PAHs eluted in 100 % toluene. Thus, the column was kept saturated with toluene and it was not necessary to re-activate the column between runs, which would be very time-consuming. The reason for using toluene instead of the more often applied dichloromethane is that nitro-PAHs and oxy-PAHs are well separated with toluene, whereas with dichloromethane the separation is not as efficient (137). The retention times were tested regularly with a mixture of PAHs (Ant, BghiPer), nitro-PAHs (9NA, 3NF) and oxy-PAHs (AQ, BAO) and were very constant for many consecutive runs. A sample of anthracene was run several times a day to ensure that no retention time drifts occurred. For the N-PAC samples, the isolation procedure described by Nielsen et al. (41) was applied. The N-PAC were extracted with a strong aqueous phosphoric acid solution and subsequently the solution was neutralised and extracted with dichloromethane. Quinoline-d7, Acridine-d9, and 10-azabenzo(a)pyrene were used as internal standards for NPAC. The recovery for basic azaarenes was previously found to be in the range 75 – 95 % (41). Analysis. Compound identification and quantifications were made using a gas chromatography (GC) – mass spectrometry (MS) system consisting of a Varian 3400 Star gas chromatograph and a Varian Saturn III ion trap mass spectrometer. The GC was equipped with a septum purge injection (SPI) temperature-programmable injector (Varian, model 1078) and a 30 m XTI-5 fused silica capillary column (Restek Corp.). For analysis of PAHs, the mass spectrometer was run in the electron impact ionisation mode with single-stage ion trap detection. For the nitro-PAHs with concentration levels 10-1000 times less than the PAHs, a more sensitive ion trap MS-MS method was developed and applied. In this method the molecular ion is isolated from the matrix and subsequently fragmented by collisions with the Helium carrier gas. The collisions are induced by applying a specific voltage resulting in a well-defined fragmentation pattern. The optimum fragmentation voltages were obtained by a series of analyses with increasing fragmentation voltage. The fragmentation patterns of some nitro-PAHs are shown in Figure 2.2a/b. The voltages that gave the highest abundance of the fragment ions were used for the analyses. Fragmentation ions used for quantification were the [M-NO]+ ion, the [MCNO2]+ ion or the [M-C2H2NO]+ ion. The identification of nitro-PAHs was based on both the fragmentation pattern of the compounds and their retention times by comparison with authentic standards. For both PAHs and nitro-PAHs, the quantification was made by calibrating the signal relative to the deuterated internal standards. The compounds analysed for and the respective internal standards are shown in Table 2.1.
Risø-R-1183(EN)
29
1.10 m/z 223
Relativ e intens ity
0.90
A
m/z 193 m/z 176
0.70 0.50 0.30 0.10 -0.10 1.10
m/z 247
Relativ e intens ity
0.90
B
m/z 217 m/z 189
0.70 0.50 0.30 0.10 -0.10 0.00
10.00
20.00
30.00
40.00
50.00
60.00
70.00
80.00
CID v o ltag e (Vo lt)
Figure 2.2. MS-MS fragmentation pattern of 9NA (A) and 2NF (B) as a function of CID voltage.
Syntheses. As a part of this work, 2NF, 2NF-d9, and 3NBAO were synthesised by two different nitration methods. Fla and Fla-d9 were nitrated by the method described in detail by Zielinska et al. (84) The method uses N2O5 as nitrating agent, and this compound was thus prepared according to Kamens et al. (76): NO2 was collected at –18 °C in a salt/ice/acetone bath as N2O4. O3 from an electrical discharge O3 generator using dry O2 as source was bubbled through this solution. The effluent was led through three 1-L flasks and collected at dry ice temperature. N2O5 was collected as a white solid and purified by two consecutive sublimation-condensation steps under a stream of O3. The nitration was performed at 25°C in CCl4 by adding N2O5 dissolved in CCl4 to a solution of Fla or Fla-d10 in CCl4. 2NF was isolated from unreacted fluoranthene and dinitrofluoranthenes by HPLC fractionation on a Nucleosil 50-5 column using CHX/DCM (3:1) as isocratic eluent. The purity of the isolated product was checked on GC-MS and HPLC. 3NBAO was synthesised by the so-called Kyodai-nitration (89,138): NO2 was bubbled through a solution of benzanthrone in DCM at 0 °C. Then O3 was bubbled through the NO2 saturated solution. The following reaction sequence is expected to occur (138): NO2 + O3 → NO3 + O2
30
(2.1)
Risø-R-1183(EN)
NO3 + BAO → NO3- + BAO+
(2.2)
BAO+ + NO2 → [BAO-NO2]+ → 3-Nitro-BAO + H+
(2.3)
3NBAO was isolated from unreacted BAO and other nitro-benzanthrone isomers (formed in much lower yields) by HPLC fractionation as described above. The purity of the isolated product was checked on GC-MS and HPLC.
Quality assurance of field measurements. The quality assurance of the field measurements included repeatability, analysing blank samples, routinely performing calibration procedures, and characterising analyte losses during the workup procedures. The use of deuterated internal standards (I.S.) ensure that any sample losses during the extraction and fractionation steps are accounted for under the condition that the selected internal standards are representative of the analytes. A series of deuterated PAHs covering vapour pressures and polarities of the PAH analytes have therefore been applied as internal standards. For the nitro-PAHs, perdeuterated 2NF, 2NF-d9, was used as I.S. The recovery of representative analytes and internal standards were checked several times to make sure that i) the analyte and I.S. losses were the same and ii) the sample loss was too small to affect the sensitivity of the method. The repeatability was determined by running a real sample (from Risø) five times. This gave standard deviations ranging from 4 to 7 % for the nitro-PAHs. Furthermore, all samples were analysed twice and the average relative standard deviation was 7±2 % for the Risø samples. For the HCAB data, the repeatability was poorer because lack of HPLC fractionation of these samples gave rise to a higher noise level. The repeatability based on duplicate analyses was generally within 25%. Before analysing the 1996 samples the recovery of several compounds were determined for the liquid-liquid fractionation. The following representative compounds were tested in triplicates: DBT-d12, TRP-d12, Per-d12, Cor-d12, Phen, AQ, BNTP, 1NP, BbFla, Per, Cor. The recoveries of these compounds were all in the range of 85 – 100 % with no significant difference between analytes and internal standards. Recovery experiments were performed several times throughout the 1998 campaign with a smaller set of compounds. Recovery results for one of these experiments with representative PAC are shown in Table 2.2. Table 2.2. Recoveries of selected PAC during liquid-liquid fractionation. Compound Phenanthrene Fluoranthene Benzo(e)pyrene 3-Nitrofluoranthene 1-Nitropyrene
Extraction efficiency (%) 73.6 76.1 79.0 81.5 81.7
Standard deviation (%) 1.7 1.1 2.5 10.4 6.7
Blank (unexposed) filters were extracted and analysed several times, but no analytes were found in these blank samples. Blanks were also taken from the HPLC fractionation procedure but in general, no analytes were detected in these. In a few HPLC fractionated blank samples very low levels of PAHs were detected, but in much too small amounts to have any significance for the measurements. Furthermore, the solvents were
Risø-R-1183(EN)
31
concentrated similarly to the samples and analysed. Some background material was present giving rise to peaks in the chromatograms, especially in the case of toluene. Some of the peaks were due to plasticizers (phthalates) found in all solvents, while others (in toluene) were due to unidentified alkyl esters. None of the analytes were detected in any of these runs. Table 2.3. Selected regression parameters for second-order calibration functions and estimated detection limits for nitro-PAHs. R2
Vx0 (%)
0.9998 0.9989 0.9999 0.9999 0.9999
1.94 4.80 1.90 2.29 1.42
Compound 9NA 2NF 3NF 1NP 2NP
DL instr. (pg) 35 20 22 24 22
DL method (pg m-3) 0.9 0.5 0.6 0.6 0.5
3.5
A
2
y = 0.139x + 0.9016x
3
2
R = 0.9998
/A
2.5 2
A
1.5 1 0.5 0 0
0.5
1
1.5
2
2.5
3
m9 N A /mI S 4.5 4
B
2
y = 0.206x + 0.5485x
3.5
2
R = 0.9989
/A
3 2.5
A
2 1.5 1 0.5 0 0
0.5
1
1.5
2
2.5
3
3.5
4
m2 N F /mI S
Figure 2.3. Second-order calibration curves for 9NA (A), 2NF (B). Calibration procedures were performed regularly. For the unsubstituted PAHs, linear calibration curves (R>0.99) could always be obtained. It is also possible to obtain linear calibration curves for the nitro-PAHs with a new column and a new injection glass liner. However, prolonged use results in positively curved calibration curves, especially at
32
Risø-R-1183(EN)
low concentrations. This is presumably due to the thermolabile nature of the nitroPAHs, which is enhanced by the development of active sites in the injection liner and column. For most of the data, second-order calibration curves have therefore been used (see Figure 2.3). The process variation coefficients (Vx0) calculated according to Funk et al. (139) from a typical set of calibration curves are presented in Table 2.3. Since the nitro-PAHs are only present in trace levels (compared to unsubstituted PAHs), detection limits of the overall method have been estimated for nitro-PAHs. It is assumed in the following that the extracted and concentrated samples have a volume of 25 µl and that the sampled air volume is 2000 m3 for 24 hour samples. The detection limit has been determined as the mass corresponding to 3 times the average noise level for 6 different real samples by measuring the noise levels immediately before and after the compound peak. This is believed to give a much more realistic measurement of the noise level than by analysing blank samples. For the DL of the overall method, a recovery of 50 % has been used as a conservative estimate for all nitro-PAHs. These detection limits are valid for the Risø samples, which were HPLC fractionated in addition to liquid-liquid fractionation. For the HCAB samples, it was not possible to do HPLC fractionation because this would separate the nitro-PAHs and the I.S. In the HCAB samples a higher background level and a higher noise level was observed. The detection limits of these samples were therefore correspondingly higher. However, there was a lot of variation in the background levels, which makes it difficult to estimate detection limits. Compared with the Risø samples the detection limits were approximately 5 times higher for the HCAB samples.
2.2 Smog chamber experiments. Experiments were performed in a 190 m3 Teflon film chamber at the UNC environmental facility in Pittsboro, North Carolina. Diesel exhaust was generated with a 1980 Mercedes sedan (Model 300SD) either in the idling mode or with ~2500 rpm for several minutes in order to obtain a particle concentration of approximately 1-2 mg m-3. Chemicals were injected with a hot manifold at ~200°C (90). Gas- and particle-phase samples were collected simultaneously with a sampling train consisting of a 47-mm Teflon impregnated glass fibre filter (TIGF) (type T60A20, Pallflex Products Corp., Putnam, CT, USA) followed by another TIGF and a 40-cm 5-channel annular gas-phase denuder (ID = 2.4 cm with each annulus separated by a space of 0.1 cm) coated with fine XAD-4 resin (Supelco, Bellefonte, PA, USA) (140). The denuders were coated according to Fan et al. (72) and were extracted and reused in the field. The second filter in the system was used to quantify adsorption of gas-phase components to the front filter (141). As a part of this study the above described sampling system was also tested against a sampling system consisting of a 40-cm denuder followed by a TIGF followed by a 24cm denuder (72). The second denuder collects compounds originally in the particle phase, which have been blown off the filter. Both denuders and filters were extracted with either a mixture of hexane, acetone, and methylene chloride (5:3:2) or a mixture of hexane and methylene chloride (1:4). The filters were soxhlet extracted for at least 3 hours with a cycle rate of ~8 hour-1, which was sufficient to extract >99.9% of the nitronaphthalenes. Extractions and work-up procedures have been described in detail elsewhere (54,72,142). Samples were collected for 20 min. with a flow rate of ~ 20 L min-1. For experiments addressing the possible formation of nitronaphthalenes on the filters during daytime sampling, annular denuders coated with aqueous solutions of NaHCO3 and glycerine (1% w/w) were positioned in front of the sampling train to remove gaseous HNO3 (143).
Risø-R-1183(EN)
33
2-Nitrobiphenyl was used as internal standard for 1NN and 2NN. Quantitative analysis was carried out on a Hewlett Packard 5890II/5971A gas chromatograph/mass spectrometer (GC/MS) equipped with a cool on-column injector. A 25 m DB-5 fused silica column (J&W) was used for chromatographic separation. Table 2.4. List of smog chamber experiments performed in this study. Date
Type
Temp. Range (K)
TSP mg m-
1NN
2NN
Injected compounds (ppm) Propylene NO
Naphthalene
3
05-20-97 Night 290 - 294 1.48 0.0020 0.0016 07-15-97 Day 314 - 318 1.46 0.92 0.2 0.4 08-25-97 Day 309 - 312 0.90 0.95 0.2 0.4 09-15-97 Night 290 - 294 2.18 0.0043 0.0019 09-30-97 Day 292 - 314 0.80 0.0134 0.0089 Additionally, a daytime experiment was performed on 08-08-97 to investigate the effect of sample ageing. Since there was no observable effect, these data are not presented.
The precision for 1NN and 2NN was generally within ± 15 % (± 2 S.D.) based on several sets of duplicate chamber samples. Repetitive injections of chamber samples gave analytical uncertainties of less than ± 5% (± 2 S.D.). O3, NO and gas NOy were monitored with Bendix Models 8101-B and 8002 chemiluminiscence analysers (Roncerverte, WV 24970). Hydrocarbons were monitored with an online gas chromatograph (Shimadzu Model 8A, column: 1.5 m, 3.2 mm stainless steel packed with Supelco 5% Bentanone 34) equipped with flame ionisation detection (FID). The GC-FID was calibrated with a propylene gas sample of known concentration. The particle size distribution for particles ranging from 0.023 – 1.0 µm was monitored by an Electrical Aerosol Analyser (EAA) (Thermo Systems, Inc., Model 3030, Minneapolis, MN). Total aerosol number concentrations for particles > 0.0016 µm was also measured with a Condensation Nuclei Counter (CNC, Model Rich 100, Environment One Corp., Schenectady, NY). Temperature, humidity and solar radiation were also monitored during the experiments (7577). In all experiments SF6 was injected as an inert tracer to account for chamber leakage. SF6 was monitored with a gas chromatograph with electron capture detection. The initial concentrations and temperatures for the experiments are presented in Table 2.4. NO and NO2 are injected into the chamber along with the diesel exhaust in variable amounts. Typical concentrations were ~0.6 ppm NO and ~0.3 ppm NO2.
2.3 Sampling Artifacts Both in the smog chamber experiments and in the field measurements, potential sampling artifacts must be considered. The two types of artifacts that may occur are 1) chemical transformation of the compounds during sampling and 2) processes that affect the gas-particle partitioning of the compounds, including volatilisation from and adsorption to filter material. The chemical transformation artifact might be important for nitro-PAHs since they can be formed by heterogeneous nitration of particle associated PAHs (76,144) and since in ambient air nitro-PAHs are only present in trace amounts compared to the PAHs. PAHs associated with filter-retained particles are continuously exposed to nitrating agents such as HNO3, NO2, and N2O5 (at night) during sampling. Furthermore, O3 may react with PAHs on the sampling filters leading to underestimation of the PAH levels. Arey et al. (145) reported that formation of nitro-PAHs during ambient air sampling is insignificant with filter-formed 1NP accounting for 2-3 % of the total 1NP measured in the filter extracts at the most, and even less for other nitro-PAHs. The recoveries of deuterated fluoranthene and pyrene including the gas phase fraction were close to 100
34
Risø-R-1183(EN)
%, whereas the recoveries of deuterated BaP and Per were significantly smaller (~48-87 %). These results were obtained during a high-NOx episode (145). The low recoveries of BaP and Per can be ascribed to reactions of these PAHs either with O3 or with nitrating agents giving rise to non-nitro products. Kamens et al. (75) reported that 10-20 % of BaP added to filter-retained wood smoke particles was degraded upon exposure to 0.5 ppm O3 for 5 min, whereas clean air pulled across the filter did not lead to BaP loss. Pitts Jr. et al. (85) also reported significant loss of PAHs associated with ambient particulate matter exposed to 200 ppb O3 on sampling filters. On the other hand, Grosjean et al. (83) did not observe any significant loss of BaP, Per or 1NP deposited on Teflon filters loaded with either fly ash, diesel exhaust or ambient particulate matter when these filters were exposed to 100 ppb O3 or 100 ppb NO2 for 3 hours under high-volume sampling conditions. This suggests that the BaP degradation observed by Kamens et al. (75) might be specific for wood smoke, but it is also in contradiction with the work Pitts Jr. et al. (85). However, Nielsen et al. (146) previously reported that there was no effect of sampling time (2×24 hours vs. 48 hours) on measured concentrations of PAHs at the Risø site. This indicates that no significant PAH degradation occurred under the specific conditions of the sampling event, including O3 levels of 30±10 ppb and NO2 levels of 5±3 ppb (146). To conclude, it is not certain to what extent PAHs are degraded by O3 during sampling but it may occur in the presence of very high O3 levels. For the field samples collected in the centre of Copenhagen (1996) O3 degradation is likely to be unimportant, since O3 is titrated by engine exhaust NO and does not reach very high ambient levels (7±3 ppb). For the semi-rural samples, the O3 levels were typically in the range 2 – 40 ppb with an average of 25 ppb. It cannot be ruled out that some O3 degradation of the most reactive PAHs occurred but due to the contradictory results reported in the literature this remains an uncertain issue (75,83,85,146). During the smog chamber experiments described in this thesis, a new type of transformation artifact was observed as discussed in detail in (1). This involves gaseous HNO3 which can be partially collected by particle loaded filters and react with gas phase PAHs to form nitro-PAHs, as observed for naphthalene forming nitronaphthalenes. This artifact is probably not important for ambient conditions since it was observed under extremely high HNO3 and naphthalene levels (modelled HNO3max ~0.5 ppm, [Naphthalene]t0 = 0.4 ppm). If the compounds are able to partition between the gas and particle phases, the phase distribution may be affected by the sampling method leading to either over- or underestimations of the concentrations in each phase. For conventional high-volume sampling of particulate matter, two types of artifacts should be considered: First, gas phase material may be physically adsorbed to the sampling filter or to the collected particulate matter and secondly, particle associated material may evaporate due to the pressure drop across the filter, the latter artifact commonly being denoted “blow-off”. Previous studies (141,147) have shown that for organic carbon the adsorption artifact is important whereas “blow-off” is normally insignificant. This is probably because the pressure drop experienced by an individual fine particle is much lower than the total pressure drop across the whole filter (the pressure decreases gradually across the filter, which is much thicker than the particle diameter. For accurately determining gas-particle distributions of semivolatile compounds it is important to quantify the vapour adsorption artifact. Traditionally, gas-particle distributions of SOCs have been determined by using a HiVol sampler equipped with a filter followed by polyurethane foam (PUF) plugs (30,46). Measurements with this system have probably overestimated the particle phase concentrations as no corrections for vapour adsorption have been made. However, for low-
Risø-R-1183(EN)
35
volatility compounds mainly present in the particle phase, adsorption does not contribute significantly to the total measured particle phase concentration, as the gas phase concentration is very low. Furthermore, even if volatilisation (“blow-off”) should take place for these heavier compounds, they would immediately be exposed to a large filter surface area to which they would be adsorbed and thus remain on the filter. A different sampling method has been proposed by Gundel et al. (140) to determine the gas-particle partitioning of SOCs. This system consists of an annular denuder to collect gas phase material followed by a filter to collect particulate matter followed by another annular denuder to collect compounds blown off the filters. In this case, “blow-off” will occur to some extent because the gas phase component has been removed and the particle phase component will respond by evaporating in order to re-establish gas-particle equilibrium. The problem with this sampling system is, that as the gas phase component is removed when the sampled air passes through the denuder, the particle component will evaporate to some extent in order to re-establish equilibrium. Hence, material originally in the particle phase will be quantified as gas phase material. In connection with the smog chamber experiments presented in this work, two sampling trains were tested: One was based on the traditional sampling system but with a backup filter used to quantify vapour adsorption and with an annular denuder in place of PUF plugs as the denuders exhibit less breakthrough (filter-filter-denuder: FFD). The other was the sampling system proposed by Gundel et al. (140) (DFD). The results of this comparison are presented and discussed in Section 4.1. For the ambient measurements, we have only sampled the particle phase and no corrections have been made with respect to adsorption. However, since the measurements have been focused on compounds that exist almost entirely in the particle phase (for a discussion on the gas-particle partitioning of the compounds measured in this study, see Chapter 1), this does not represent a significant error. For the more volatile PAHs measured, ranging from DBT to Pyr, the concentrations may have been overestimated. From the smog chamber studies with nitronaphthalenes it was estimated that adsorption could account for 10-20 % of the particle phase concentrations even though the second filter only removed 0.05-0.1 % of the gas phase component. The volatility of the nitronaphthalenes is comparable to the volatility of Phen and Ant. However, the glass fibre filter material used in the field sampling is probably more adsorptive than Teflon impregnated filter material. Hence, significant filter adsorption of PAHs ranging from Mw 178 to Mw 228 cannot be ruled out.
2.4 Photodegradation experiments. Two series of experiments were performed. The first was aimed at investigating the effect of the chemical composition of organic aerosols on the photodegradation of nitroPAHs associated with such aerosols. This was done by irradiating samples of nitroPAHs in the presence of a series of known constituents of atmospheric aerosols. The other was aimed at studying the effect of various physical factors on the photodegradation of selected PAC. Irradiations were performed with a turntable photoreactor using a 450 W medium pressure mercury lamp as light source. The lamp was placed in a quartz immersion well with cooling water. The immersion well was surrounded by a Pyrex sleeve to filter out high-energy UV bands (λ > ~290 nm). Although this does not resemble the spectral distribution of the actinic flux, it does reproduce the wavelengths normally encountered in the atmosphere. The photoreactor was positioned in a water bath with constant water circulation. The circulation and temperature in the water bath was
36
Risø-R-1183(EN)
maintained with a HETO control unit. The solutions were irradiated in 13 × 100 mm quartz reaction tubes. Effect of chemical composition (co-solutes). The cosolutes used for the chemical composition experiments are listed in Table 5.1. A few test experiments were performed with a monochromatic light source. This setup consists of a Hg lamp (Bausch & Lomb) with the emission focused into a monochromator. A wavelength of 366 nm was used to selectively produce excited nitro-PAHs and avoid photolysis of the cosolutes. Samples were irradiated in pairs in quartz cuvettes. One sample contained only the nitro-PAHs, in the other a cosolute was added. To test that the photon flux into the two cuvettes were the same, two tests were performed. First, the classic ferrioxalate actinometer (148) were irradiated and secondly, identical solutions of 1NP were irradiated. The production of Fe2+ (complexed with 1,10phenanthrolin) and the loss of 1NP were spectrophotometrically determined. The difference between the two cuvettes was less than 5 %. The concentrations of nitro-PAHs were ~10-4 M and the co-solute concentrations were in the range 1×10-3 - 5×10-3 M. In experiments with polar cosolutes small amounts of dichloromethane or ethyl acetate were added to all samples in equal amounts in order to dissolve the cosolutes. The chemical composition samples were analysed by HPLC using a Shimadzu HPLC (LC10-AD) equipped with diode array UV-vis detection (SPD-M10A). The column was packed with Nucleosil 50-5 C18-material. Methanol containing 10 or 20 % Millipore H2O was used as isocratic eluent. The response of the analytical system was a linear function of concentration (r2 > 0.99). Effect of physical factors on the PAC photodegradation. In these experiments, duplicate samples were irradiated with anthraquinone added as a radical sensitizer to one of the samples, whereas the other contained only a PAC mixture. The PAC mixture consisted of ACR, BaA, BAO, 3NF, 1NP, BaP and BeP as examples of different PAC. The conditions for the experiments addressing the influence of physical factors are listed in Table 2.5 The concentrations of PAC were ~8×10-5 M and the concentration of anthraquinone was ~1×10-3 M. 6 ml solutions were irradiated and 400 µl of each sample was removed after 0, 30, 60 and 120 min. and transferred to brown chromatography vials. 10 µl of a solution of deuterated PAC (TRP-d12, Per-d12) and phenanthridine were added as internal standards immediately. GC-MS analysis is more suitable for these PAC mixtures in order to get good separation. However, some of the samples needed pre-treatment prior to GC-MS analysis as described in the following. The MeOH samples (Exp. 1) were added 2 ml H2O (Millipore) and extracted with 0.5 ml CHX. The CHX phase was removed and dried with Na2SO4 (anhydrous). The dioxane/H2O (Exp. 2) samples were added 1 ml CHX and dried with Na2SO4. The glycerine samples (Exp. 3) were added 2 ml H2O and 1 ml CHX and shaked. The CHX phase was removed and dried with Na2SO4. The DOP samples (Exp. 4-7) were added 2 ml CHX and extracted with 1 ml DMF/H2O (9:1). The DMF/H2O extracts were added 1 ml H2O and 1 ml CHX and shaked. The CHX phase was removed and dried with Na2SO4. The samples were analysed by GC-MS as described for PAH analysis of atmospheric samples. The linearity of the GC-MS analysis was tested on 3NF and 1NP, since the nitro-PAHs are most sensitive towards active sites etc. The response was linear (R2 > 0.996) in the concentration range relevant for these experiments.
Risø-R-1183(EN)
37
Table 2.5. List of experiments used to study of the effect of physical factors on PAC photo-degradation Experiment 1. 2. 3. 4.-7.
Description Photolysis in 5 solvents: CHX, toluene, ethyl acetate, acetonitrile, MeOH with and without AQ Photolysis in DIOX/H2O mixtures with 0, 3, 6 and 10 % H2O with and without AQ Photolysis in CHX and glycerine (GLY) with and without AQ Photolysis in CHX and bis-(2-ethyl-hexyl)-phthalate (DEHP/DOP) with and without AQ at 12, 20, 30 and 40°C
No. of samples 10 8 4 4×4
Although it was attempted to reduce the DOP concentration in the final samples by liquid-liquid extraction as mentioned above, the amount of DOP was still sufficient to cause peak splitting and broadening for Acr, BaA and BAO. It was therefore not possible to analyse for these compounds in the DOP samples, whereas the chromatography of the compounds eluting later than DOP (1NP, 3NF, BeP and BaP) was not disturbed by DOP.
2.5 Pulse radiolysis experiments. As a part of this PhD-project pulse radiolysis experiments were performed in order to investigate the aqueous phase chemistry of PAC. Several different PAC have been detected in rain samples (100) and especially N-PAC are expected to partition into atmospheric water droplets (cloud, fog or rain) due to the basic nitrogen atom which make them relatively water-soluble compared to other PAC. Results for two N-PAC, quinoline (Q) and 5-nitroquinoline (5NQ), are reported. Both are reasonably water soluble and 5NQ can be regarded as a model compound for some of the extremely genotoxic nitro-N-PAC detected by Sera et al. (128) in environmental samples. A pulse radiolysis setup was used to generate OH radicals in aqueous solution. The setup consists of a 10 MeV HRC Linac accelerator delivering 0.5 - 3.0 µs electron pulse into a quartz cell. An optical detection system consisting of a 150 W Varian xenon lamp, a 5.1 cm light-path cell, a Perkin-Elmer double quartz prism monochromator, a 1P28 photomultiplier and a LeCroy 9400 digital oscilloscope was used to monitor transient species. The experimental setup has been described in detail elsewhere (149). A unique specially designed high-pressure high-temperature cell (150) was used in order to increase the O2 concentration. O2 concentrations were controlled by allowing the solutions to be saturated with O2 at elevated pressures ranging from 2 to 114 atm. Pulse radiolysis of water yields the following species within 1 ns (151): H2O e-(aq)(2.7) + H(0.55) + OH(2.7) + H2(0.45) + H2O2(0.71) + H3O+(2.7)
(2.4)
where numbers in parenthesis are G-values. The G-values represent the number of species formed for each 100 eV energy absorbed. To convert to SI units the following expression is used: 1 molecule per 100 eV = 0.1036 µmol J-1. N2O saturated solutions were used in order to convert e-(aq) into more OH radicals: N2O + e-(aq)(2.7) + H2O → N2 + OH + OH-
(2.5)
The overall yield of OH radicals then becomes 5.2 (2 × 2.7) molecules per 100 eV or 0.539 µmol J-1. This is about a factor of 10 higher than the yield of H atoms, so any interferences of these are within the experimental uncertainty. All solutions were prepared
38
Risø-R-1183(EN)
in triply distilled water. pH was adjusted with HClO4 or NaOH. Typical electron doses were in the range of 10 - 15 Gy (Gray, 1 Gy = 1 J kg-1).
Risø-R-1183(EN)
39
3. Field measurements. Results and discussion. 3.1 Levels of PAC in urban and semi-rural air. Average concentration levels (in ng m-3) of PAC (PAHs, nitro-PAHs, S-PAC) measured in this study are tabulated in Table 3.1 for both of the two sampling sites. In addition, the average levels of selected PAC excluding samples heavily influenced by long-range transport (LRT, see Section 3.5) are tabulated in Table 3.2, since these levels are probably more representative of the typical concentration levels encountered at this site. Some general features can be derived from Table 3.2. Table 3.1. Average levels of PAHs, nitro-PAHs and S-PAC at two sampling sites. HCAB (n=15) Conc. Conc. PAC PAC ng m-3 ng m-3 PAHs: IP 2.0 Phen 1.4 BghiPer 5.1 Ant 0.4 AA 0.6 4.0 Cor 3.8 ΣMePhen S-PAC: Fla 5.2 Pyr 9.0 DBT 0.1 BghiFlaa 3.5 BNTP 0.4 Nitro-PAHs: CP 2.4 BaA 3.1 9NA 0.071±0.032 Chr/Trp 4.0 2NF 0.114±0.067 BbjkFla 6.5 3NF 0.043±0.023 BeP 3.1 1NP 0.144±0.053 BaP 2.2 2NP 0.023±0.009 Per 0.49 a Includes BcPhen
Risø (n=24) PAC
Conc. ng m-3
PAC
Conc. ng m-3 0.19±0.09 2.62±2.47 1.25±1.06 0.41±0.26 1.83±1.65
PAHs: Phen Ant ΣMePhen Fla Pyr BghiFla BcPhen CP BaA Chr/Trp BbjkFla BeP BaP
4.40±5.50 0.77±0.71 7.09±8.82 2.32±2.18 2.16±1.99 1.20±0.99 0.48±0.46 0.69±0.71 0.98±0.92 1.50±1.26 4.51±3.99 1.73±1.33 1.69±1.45
Per IP BghiPer AA Cor S-PAC: DBT 0.050±0.050 BNTP 0.117±0.094 Nitro-PAHs: 9NA 0.038±0.015 2NF 0.097±0.045 3NF 0.031±0.017 1NP 0.040±0.014 2NP 0.022±0.013
Table 3.2. Average levels of PAHs, nitro-PAHs and S-PAC at two sampling sites excluding data from long-range-transport episodes. HCAB (n=14) Conc. Conc. PAC PAC ng m-3 ng m-3 PAHs: IP 2.0 Phen 1.37 BghiPer 5.1 Ant 0.40 AA 0.6 4.0 Cor 3.8 ΣMePhen S-PAC: Fla 5.2 Pyr 9.0 DBT 0.1 BghiFlaa 3.5 BNTP 0.4 Nitro-PAHs: CP 2.5 BaA 3.1 9NA 0.071±0.032 Chr/Trp 3.9 2NF 0.114±0.067 BbjkFla 6.2 3NF 0.043±0.023 BeP 2.9 1NP 0.144±0.053 BaP 2.1 2NP 0.023±0.009 Per 0.46 a Includes BcPhen
PAC PAHs: Phen Ant ΣMePhen Fla Pyr BghiFla BcPhen CP BaA Chr/Trp BbjkFla BeP BaP
RISØ (n=20) Conc. Conc. PAC ng m-3 ng m-3 Per 0.12±0.07 IP 0.37±0.17 0.72±0.54 BghiPer 0.33±0.40 0.43±0.32 AA 0.58±0.24 0.28±0.25 Cor 0.61±0.32 0.58±0.44 S-PAC: 0.64±0.36 DBT 0.43±0.37 0.017±0.006 BNTP 0.13±0.11 0.05±0.02 Nitro-PAHs: 0.17±0.13 9NA 0.28±0.22 0.030±0.015 2NF 0.52±0.36 0.060±0.022 3NF 1.37±0.82 0.032±0.018 1NP 0.67±0.32 0.030±0.015 2NP 0.54±0.26 0.008±0.004
The levels of nitro-PAHs relative to the native PAHs are higher in the semi-rural area than in the urban area, except for 9NA where the Ant/9NA ratios are comparable.
40
Risø-R-1183(EN)
Fla/2NF at Risø is ~13 and at HCAB it is ~57. The corresponding Pyr/1NP ratios are ~75 (HCAB) and ~21 (Risø). In general, the PAH/nitro-PAH ratio is in the order of 10 – 100. The S-PAC levels are about a factor of 10 lower than typical PAH levels. Levels of other PAC (N-PAC, oxy-PAHs) determined at the HCAB site are presented later. The average levels of nitro-PAHs range from ~10 pg m-3 (2NP, Risø) to 118 pg m-3 (1NP, HCAB). However, there is a lot of variability in the data. For example 2NP was below detection limit (~0.5 pg m-3) in certain samples, whereas the highest nitro-PAH levels were reached by 2NF during a LRT episode (430 pg m-3, HCAB, 22/4/1996). The nitro-PAH levels measured in this work are comparable to levels measured in other parts of the industrialised world, such as California (36,152), Italy (37) and previously at Risø (33). Although the PAH emissions in the industrialised world may have decreased in recent years (44) (see also Section 3.7) the levels of nitro-PAHs and PAHs at Risø does not appear to have decreased when the results presented here are compared with the levels of 9NA and 1NP determined by Nielsen et al. in 1982 (33) (see Table 3.3). On the contrary, the 1NP concentration is higher in 1998, but it should be noted that the two sample sets may not be truly comparable. The temperature range for the 1982 data (-10 to +10°C) was lower than for the current data (-3 to +16°C). BeP, BbjkFla, and IP have been used for this comparison since they are stable, relatively indifferent to changes in ambient temperature (with respect to gas-particle partitioning), and relatively indifferent to changes in source composition. In addition, it is the subjective impression of Torben Nielsen (T. Nielsen, personal comm.) that the road transport of goods with heavy duty vehicles have increased at Risø from 1982 to 1998. Table 3.3. Comparison of 1998 data with previous measurements (LRT data not included). 9NA
1NP
BeP
BbjkFla
IP
1998
0.030±0.015
0.030±0.015
0.67±0.32
1.37±0.82
0.72±0.54
Nielsen et al. (33)
0.03±0.01
0.009±0.005
0.6±0.2
1.3±0.5
0.7±0.2
The average nitro-PAH levels at HCAB are only significantly different from the Risø levels in the case of 1NP (t-test, P
