Diss ETH Nr. 12235
Laboratory Kinetic and Atmospheric Modelling Studies of the Role of Peroxyacyl Nitrates in Tropospheric Photo-Oxidant Formation
A dissertation submitted to the SWISS FEDERAL INSTITUTE OF TECHNOLOGY ZURICH
for the degree of DOCTOR OF NATURAL SCIENCES
presented by STEPHAN SEEFELD Dip!. Chem. ETH born March 3, 1967 Citizen of Zurich (CH)
Accepted on the recommendation of Prof. Dr. Dieter Imboden Prof. Dr. J. Alistair Kerr Dr. William R. Stockwell
1997 Ce milieu qui nous est echu en partage etant toujours distant des extremes, qu'importe qu'un autre ait un peu plus d'intelligence des choses? S'il en a, ii /es prend un peu de plus haut. N'est-i/ pas toujours infiniment eloigne du bout? Pensees, Blaise Pascal (1623-1662) For Cecile Acknowledgements
With many thanks to Prof. J. A. Kerr and the Swiss Federal Institute of Environmental Science and Technology (EA WAG/ETH) for continuous support and for providing the necessary facilities which made the present study possible. I would like to thank Dieter Imboden for acting as examiner. I would also like to thank Dr. William R. Stockwell for the time I could work in his group and for the many hours he spent introducing me into the secrets of modelling. The following people improved this study in many helpful discussions: Dr. David Stocker, Dr. David Kinnison, Dr. Garrett Locke, Dr. Michael Kuhn, Konrad Stemmler. And last but not least thanks to everybody who supported my in any way during this time. Table of Contents Kurzfassung Abstract
1. Introduction 1 1.1 Synopsis 1 1.2 Air pollution 2 1.3PANs 3 1.3.1 History 3 1.3.2 The "PAN-Family" 4 1.3.3 The Physics of PAN 4 1.3.4 The Chemistry of PAN 5 1.3.5 Toxicology 7 1.3.6 Atmospheric Measurements 8 1.4 Ozone and PAN chemistry 9 1.5 Computer Model of the Troposphere 11
2. Relative Rate Studies, Aldehyde + Cl 13 2.1 Introduction 13 2.1.1 Relative Rate Studies 13 2.1.2 The Flow System 14 2.1.3 Aldehyde + Cl 14 2.2 Experimental 16 2.2.1 On-line Preparation of Gas Mixtures 16 2.2.2 The Reactor and Light Source 17 2.2.3 Analytical System 17 2.2.4 Experimental Procedure 17 2.3Results 18 2.3.1 The Competitive Technique 18 2A Discussion 20 2.4.1 The Flow System 20 2.4.2 Propionaldehyde + Cl vs. Acetaldehyde + Cl 20 2.4.3 Trifluoroacetaldehyde + Cl vs. Acetone + Cl 20
3. PAN Laboratory Study 23 3.1 Introduction 23 3.2 Experimental 24 3.2.1 The Flow System 24 3.2.2 Analysis 26 3.2.3 Measurement of the Rate Data 27 3.2.4 PAN Synthesis 28 3.2.5 Materials 29 3.2.6 Kinetic Model of the Reaction Scheme 29 3.3 Results 29 3.4 Discussion 31 3.4.1 Checks on the Proposed Mechanism of PAN Fonnation 31 3.4.2 Comparison with Literature Data 33 3.4.3 Extension of the Techniques 35 3.4.4 Atmospheric Implications 35
4. PPN Laboratory Study 37 4.1 Introduction 37 4.2 Experimental Section 38 4.2.1 Flow System 38 4.2.2 Analysis 39 4.2.3 Measurement of the Rate Data 39 4.2.4 Combined PAN and PPN measurements 40 4.2.5 Materials 40 4.2.6 PPN Synthesis 41
4.2.7 Peroxytrifluroacetyl Nitrate (PF3AN) and Peroxybenzoyl Nitrate (PBzN) 41 4.3 Results 41
4.3. I k/k2 for PPN 41 4.3.2 Relative Measurements k/k2 for PPN vs. k/k2 for PAN 43 4.4 Discussion 45 4.4.1 Comparison with Literature Data 45 4.4.2 Tropospheric Lifetimes of PPN and PAN 46 4.4.3 Implications for Computer Modelling 47 4.4.4 Measuring the Ratio for Other PANs 47
5. Box Model and Chemical Mechanism 49 5.1 Introduction 49 5.2 The RACM Mechanism 49 5.2. l Inorganic Chemistry 50 5.2.2 Organic Species and Chemistry 50 5.2.3 Radical Chemistry 51 5.2.4 The Future of the RACM Mechanism 52 5.3 Photolysis Rate Coefficients 52 5.4 The SBOX-Model-System 53 5.4.1 The Chemical Compiler 53 5.4.2 The Solver, VODE 55 5.4.3 The Parameter File 56 5.4.4 Output Files 56 5.4.5 Future Development 57 5.5 The PLUME-Case 57 6. Modelling 59 6.1 Nitrogen Chemistry 59 6.1.l Introduction 59 6.1.2 Method 59 6.1.3 Results and Discussion 60
6.2 Chemistry of the Acylperoxy Radical (AC03) 68 6.2.1 Introduction 68 6.2.2 Method 68 6.2.3 Results and Discussion 69
6.3 PAN/03 Ratio 73 6.3.1 Introduction 73 6.3.2 Method 73 6.3.3 Results and Discussion 73 6.4 What Happens when the Light Goes Off? 74 6.4.1 Introduction 74 6.4.2 Method 74 6.4.3 Results and Discussion 75 6.5 Conclusions 76
7. Sensitivity Study 79 7.1 Introduction 79 7.2Method 80 7.2.1 Sensitivity Coefficients 80 7.2.2 The Decoupled Direct Method (DDM) 81 7 .2.3 Time Dependant Rate Coefficients 82 7 .2.4 Sensitivity of Ratios 82 7.3 Results and Discussion 83 7.3.1 General Checks of the DDM 83 7.3.2 Diurnal Varying Photolysis Coefficients 84 7.3.3 Temperature Dependence of Sensitivity Coefficients 87 7.3.4 Influence of Total Emissions and Season 88 7.3.5 Sensitivity of Ratios 89 7.4 Conclusions 91
8. Literature 93
Appendix Appendix A. FACSIMILE PROGRAM Appendix B. RACM Model Species List Appendix C. The RACM Mechanism Appendix D. Curriculum Vitae Kurzfassung
Lange wurde Luft als unerschopflicher Rohstoff angesehen und Luftverschmutzung war nur ein Problem schlechter Durchmischung. Doch der vermehrte Ausstoss reaktiver Gase aus lndustrie, Verkehr, Heizung und anderen menschlichen Aktivitiiten, fiihrte zu zunehmenden Problemen auf lokaler, regionaler und globaler Ebene. Lange Zeit beriichtigt warder Londoner ,,Smog", cine Kombination aus Rauch (smoke) und Nebel (fog), der typischerweise grosse Mengen Staub und Schwefeldioxyd enthielt und vor allem durch die Verfeuerung von Kohle verursacht wurde. Um 1950 wurde von Wissen- schaftlern in der Gegend von Los Angeles, Kalifomien, eine neue Art Smog festgestellt, der im Gegensatz zum ,,Londoner Smog" oxidierend wirkte. Dieser sogenannte photochemische' Smog, der in Zusammenhang gebracht wurde mit Autoabgasen, verursachte Pflanzenschaden und Emteverluste. Bald erkannte man, dass viele weitere Gebiete durch photochemischen Smog belastet waren, was zu intensiver Forschungstiitigkeit auf der ganzen Welt fiihrte. Das Besondere am photochemischen Smog ist, dass viele Schadstoffe erst in der Atmo- sphiire gebildet werden. Sie entstehen durch chemische und photochemische Umformung von Stoffen, die in die Atmosphare emittiert wurden. Um wirkungsvolle Kontrollstrategien zu erarbeiten, ist es daher notwendig, die Prozesse zu verstehen, die zum photochemischen Smog fiihren. Der bekannteste Luftschadstoff, der sich im photochemischen Smog bildet, ist das Ozon". Neben Ozon entstehen jedoch noch eine Reihe weiterer schadlicher Verbindungen, deren Bedeutung je liinger je mehr erkannt wird.
Eine wichtige Klasse von Verbindungen sind die Peroxyacylnitrate (PANs, RC(O)OON02). Sie bilden sich iihnlich wie Ozon in der Atmosphiire beim oxidativen Abbau vieler organischer Verbindungen sowohl menschlichen als auch natiirlichen Ursprungs. PANs sind massgebend an der oxidativen Wirkung des photochemischen Smogs beteiligt. Sie reizen die Augen und werden fiir Pflanzenschaden verantwortlich gemacht. Da reaktive Zwischenprodukte der tropospharischen Chemie voriibergehend in P ANs umgewandelt werden konnen, spielen Peroxyacylnitrate zudem eine wichtige Rolle im Transport von reaktiven Stickstoff- verbindungen in der Troposphiire. Im Rahmen dieser Dissertation wurde eine Reihe von Labor- und Computermodellstudien durchgefiihrt, um mehr zu erfahren iiber die Rolle der Peroxyacylnitrate im photochemischen Smog. Diese Studien und ihre Resultate werden im Folgenden kurz vorgestellt. Es wurden Laborexperimente unter simulierten atmosphiirischen Bedingungen durchgefiihrt. Dazu wurden Gasmischungen durch einen thermostatisierten Durchflussreaktor geleitet und mittels Gaschromatographie analysiert.
• Der Ausdruck ,,photochemisch" weisst darauf hin, dass dieser Smog auf chemischem Weg in der Atmosphlire entsteht und d;iss dazu Sonnenlicht notwendig ist. Da photochemischer Smog im Sommer besonders haufig isl, wird er manchmal auch Sommersmog genannt. •• Gemeint ist hier das troposphiirische Ozon, das im Ubermass gebildet flir die Biosphlire schiidlich ist. Es hat nichts mit dem ,,Ozonloch" zu tun, wie der schleichende Abbau der Ozonschichl in der Stratosphlire oft genannt wird. Aldehyde sind wichtige Vorliiufer fiir PANs. Daher wurden die folgenden relativen Ge- schwindigkeitskoeffizienten bei Raumtemperatur bestimmt: k(Propionaldehyd + Cl-) I k(Acetaldehyd + Cl-) = 1.45 ± 0.11; k(Trifluoracetaldehyd + Cl-) I k(Aceton + Cl-) = 1.09 ± 0.06.
In zwei Studien wurde das Reaktionsystem RC(0)02• + N02 + M -7 RC(0)02N02 + M (I) und RC(0)02• +NO -7 R· + C02 + N02 (2) untersucht, indem die relativen Ausbeuten von
RC(0)02N02 bestimmt wurden, als Funktion des Verhiiltnisses [NO]/[N02]. Im Temperatur- bereich (-30 to +30 °C) wurden die folgenden Verhiiltnisse bestimmt: Fiir Peroxyacetylnitrat
(PAN, R = CH3) kifk2 = 0.41 ± 0.03 und fiir Peroxypropionylnitrat (PPN, R = CH3CH2) k1/k2 = 0.43 ± 0.06, wobei die Fehlergrenzen jeweils 2o entsprechen. Mit Hilfe von RACM (Regional Atmospheric Chemistry Mechanism) und einem neu ent- wickelten ,,Computer-Box-Modell" (SBOX) wurde die Chemie der atmosphiirischen Stick- stoffspezies und speziell von PAN untersucht. Fiir ein Szenario mit mittelmiissiger Luft- verschmutzung, wurde eine Stickstoffbilanz erstellt und die Stofffliisse zwischen den Stick- stoffspezies wurden analysiert. Zudem wurden die wichtigen Bildungs- und Abbaureaktionen des Acetylperoxyradikals identifiziert. Schliesslich wurde die Temperaturempfindlichkeit des
Verhiiltnisses PAN/03 untersucht, sowie die Auswirkungen von pliitzlichen Temperatur- und
Lichtiinderungen auf die errechneten Konzentrationen von PAN und 0 3 analysiert. Mit Hilfe der ,,Decoupled Direct Method" (DDM) wurde die Sensitivitiit der vorhergesagten Ozon- und PAN-Konzentration bestimmt. Eine Methode wurde vorgestellt, mit der man die Sensitivitiit in Bezug auf Verhiiltnisse von Geschwindigkeitskonstanten bestimmen kann. Betreffend den Reaktionen (1) und (2), konnte gezeigt werden, dass die Ozon- und PAN-
Konzentration vie! empfindlicher auf Anderungen des Verhiiltnisses k 1/k2 reagiert, als auf
Anderungen der absoluten Werte von k1 und k2•
Um die Schiiden, verursacht durch Ozon, P ANs und andere Luftschadstoffe, miiglichst effizient zu bekiimpfen, verwendet man heute Computermodelle. Mit Hilfe dieser Modelle, die hauptsiichlich auf Labordaten basieren, liisst sich zum Beispiel ermitteln, welche freigesetzten Schadstoffe am meisten sekundiire Schadstoffe produzieren. Die Studien, die im Rahmen der vorliegenden Dissertation gemacht wurden, kiinnen zur Verbesserung solcher Computer- modelle dienen und sind daher indirekt ein Beitrag zur Verbesserung der Luftqualitiit. Abstract
Peroxyacyl nitrates (PANs, RC(O)OON02) are formed in the atmosphere in the oxidative degradation of many organic compounds of both anthropogenic and biogenic origin. They are important oxidant components of photochemical smog and can cause eye irritation and plant damage. Moreover peroxyacyl nitrates act as temporary reservoirs for reactive intermediates involved in photochemical smog formation and they play an important role in the transport of reactive nitrogen in the troposphere. To investigate the role of peroxyacyl nitrates in tropospheric photo-oxidant formation, a series of laboratory kinetic and computer modelling studies have been carried out. For the laboratory experiments, an atmospheric flow system was employed including on-line preparation of gas mixtures, a thermostated flow reactor and analysis by gas-chromatography. Because aldehydes are important precursors for PANs, the following relative rate coefficients were determined at ambient temperature using the competitive technique: k(propionaldehyde + Cl-)/k(acetaldehyde + Cl.) = 1.45 ± 0.11; k(trifluoroacetaldehyde + Cl·)/k(acetone +Cl·) = 1.09 ± 0.06.
Two relative rate studies have been performed on the reactions RC(0)02 • + N02 + M ~
RC(0)02N02 + M (1) and RC(O)O,- + NO ~R· + C02 + N02 (2), by measuring the yields of
RC(0)02N02, on a relative scale, as a function of the ratio of reactants [NO]/[N02]. Over a range of temperatures (-30 to +30 °C) the following ratios were determined: for peroxyacetyl nitrate (PAN, R CH3) k/k, = 0.41 ± 0.03 and for peroxypropionyl nitrate (PPN, R =
CH3CH,) k/k, = 0.43 ± 0.06, where the error limits are 2cr The Regional Atmospheric Chemistry Mechanism (RACM), with a newly developed photo- chemically box model system (SBOX), was applied to investigate atmospheric nitrogen chemistry in general and PAN chemistry in particular. For a moderately polluted scenario of the planetary boundary layer, a nitrogen budget was calculated and the nitrogen flows were analysed. Additionally the important production and loss channels of acylperoxy radicals, the precursors of PANs, were identified. Finally the temperature dependence of the P AN/03 ratio was investigated and the response of the 0 3 and PAN concentrations to sudden changes in light intensity and to temperature were studied. The sensitivity of the predicted ozone and PAN concentrations to rate coefficients was calculated using the decoupled direct method (DDM). A method was developed to calculate the sensitivity to the ratios of rate coefficients. With respect to reactions (1) and (2), it was shown that ozone and PAN are much more sensitive to the ratio of the rate coefficients, k,/k,, than to the absolute values of k, and k,. 1. Introduction
1.1 Synopsis
Peroxyacyl nitrates, PANs, are a group of chemical species, formed in the atmosphere through the oxidation of many organic compounds of both anthropogenic and biogenic origin. They are strong oxidants, important components of smog and can cause eye irritation and plant damage. This thesis describes several aspects of the chemistry of PANs, investigated by laboratory experiments and computer models. The chapters, which are essentially independent, cover the following subjects: After a brief introduction into air pollution and atmospheric chemistry, in the first chapter, the physical, chemical and toxicological properties of peroxyacyl nitrate are reviewed. Their role in photochemical smog formation and their connection to the ozone chemistry is presented. In the second chapter preliminary kinetic experiments using a flow reactor are described. The rate coefficients of the reactions of chlorine atoms with two aldehydes were measured using the relative rate teehnique. The third chapter describes the measurement of key reactions in the formation of peroxyacetyl nitrate (PAN), using the same flow reactor as in chapter 2. The results are discussed and atmospheric implications are briefly considered. This chapter is based on work already published as "Relative Rate Study of the Reactions of Acetylperoxy Radicals with NO
and N02 : Peroxyacetyl Nitrate Formation under Laboratory Conditions Related to the Troposphere" (Seefeld et al., 1997). The fourth chapter describes a similar experiment for peroxypropionyl nitrate (PPN). It is based on work accepted for publication, as "Kinetics of the Reactions of Propionylperoxy
Radicals with NO and N02: Peroxypropionyl Nitrate Formation under Laboratory Conditions Related to the Troposphere" (Seefeld and Kerr, 1997). A newly developed photochemical box model system is presented in the fifth chapter, and the chemical mechanism involved, which is published by Stockwell et al. (1997), is briefly explained. Chapter six presents results from box model calculations, relating to the atmospheric nitro- gen chemistry in general and the PAN chemistry in particular. And in the seventh chapter, the sensitivity of the model predictions to uncertainties in the input parameters is examined. Here the main focus lies again on PAN chemistry and on the rate coefficients measured in chapters three and four. 2 Introduction
1.2 Air pollution
Air pollution was originally regarded as a sign of industrial activity and economic progress. The air resources of the earth were considered to be limitless. However increased emissions of reactive trace gases from humanities' energy related and industrial activities have led to a wide variety of air pollution problems on urban, regional, and global scales. The term "smog" has been coined to describe a combination of smoke and fog that occurred in London and other coal burning regions. This London smog was associated with high levels of particulate matter and sulphur dioxide. In the late 1940s and early 1950s a new type of smog, so called "photochemical smog", was investigated by a number of researchers in the Los Angeles, California area. In contrast to the smog, which was observed before in London, the new type of smog was found to be chemi- cally oxidising and caused eye-irritation. Photochemical smog was believed to be associated with motor vehicle emissions. Downwind of Los Angeles, many agricultural areas were affected by the photochemically derived pollutants, producing damage to the foliage and subsequent crop-loss (Gaffney et al., 1989). This led to intensive research, first in the US, and later, following similar observations in other countries, all over the world. In Switzerland, at least since the eighties, it has been known that the ozone concentrations are relatively too high in summer (BUW AL, 1989; Neininger and Dommen, 1996). While in 1968 Scorer still concluded that most air pollution problems could be solved by building very tall chimneys, today it is generally accepted that clean air is a limited resource and that the reduction of emissions is required. National air quality standards for oxidants were first established in the US in 1971 and later international guidelines for ozone were proposed by a working group of the World Health Organization (1979). In order to plan efficient control strategies for photochemical smog it is important to under- stand the chemistry of the atmosphere because the emitted primary pollutants are transformed to secondary pollutants by chemical and photochemical reactions. Although at the beginning the main scientific interest in atmospheric chemistry was focused on ozone, it has become more and more obvious, that other chemical species play also an important role in photo- chemical smog formation. An example is the peroxyacyl nitrates, which are discussed in this thesis. Today, there are three main areas of research in atmo- spheric chemistry: field studies, laboratory experiments and computer modelling. As shown in FIGURE 1-1, they are strongly connected. For the interpretation of field studies, laboratory experiments and computer models are necessary; the important species and reactions for laboratory experi- FIGURE 1-1. The three main men ts are often identified in field studies and by computer fields of atmospheric chemistry. Introduction 3 models; and to develop and run a computer model, infonnation from the laboratory and the field are necessary.
1.3 PANs
Peroxyacyl nitrates (PANs, RC(O)OON02) are well known to be important components of
photochemical smog. They are formed in the NOx (NO and N02) polluted atmosphere as a product of the oxidative degradation of many organic compounds. They can cause eye irrita- tion and plant damage. Moreover, PANs act as a temporary reservoir for reactive intennediates involved in smog fonnation and they play an important role in the transport of NO, in the troposphere. This chapter will give a short overview of the current state of knowledge of peroxyacyl nitrates. The reader wishing further details, is referred to one of several excellent reviews. Stephens (1969) provided fundamental infonnation on the chemical and physical properties of peroxyacetyl nitrate, PAN. Gaffney et al. (1989) present a good overview of all areas of PAN chemistry and measurement. A comprehensive review of the chemistry and atmospheric measurements of organic nitrates, including PANs, has been carried out by Roberts (1990). Finally, a recent review of the chemistry and biological effects of PANs has been presented by Kleindienst (1994).
1.3.1 History In 1956 Stephens et al. detected in irradiated mixtures of nitrogen oxides and hydrocarbons in the laboratory a new compound, initially called "compound X", which was soon after also found in Los Angeles air. The precise structural formula of the substance was initially unknown until Stephens et al. (1961) established it to be peroxyacetyl nitrate (PAN): 0 II ,,c, ,,0, H3C 0 N02 The identification and subsequent synthesis of PAN led to numerous studies of the chemical and physical properties of PAN, as well as of its physiological effects on plants and animals. Subsequently PAN was found to cause eye irritation as well as to be the cause of observed unusual bronzing and chlorosis of leaves of crops (Stephens et al., 1961; Gaffney et al., 1989; Kleindienst, 1994).
The name "peroxyacetyl nitrate" assigned by Stephens to the compound CH3C(O)OON02 does not follow official chemical nomenclature. Because it is a mixed acid anhydride, it should be called "ethaneperoxoic nitric anhydride". Unfortunately this leads to a different acronym, EPNA, a fact that may partially explain why the proper name has not been widely adopted. A proposed compromise name that recognises the anhydride structure of the compound, but retains the acronym "PAN", is "peroxyacetie nitric anhydride" (Roberts, 1990). However a 4 Introduction look through the recent literature shows that "peroxyacetyl nitrate" is still most widely used and it was therefore also used in this thesis.
1.32 The "PAN-Family" Peroxyacetyl nitrate, PAN, is the first known member of a homologous series of compounds with similar chemical and physical properties. There have been many failed attempts to
• synthesise the first possible compound in the series, HC(O)OON02 Therefore HC(O)OON02 is believed to be too unstable to exist in the atmosphere. TABLE 1-1 shows the names and structures of the most important members of this series. Most of these species have been found in polluted urban air masses as well as at rural sites. PAN and peroxypropionyl nitrate, PPN, were also found in the free troposphere by aircraft measurements. While originally thought to be only of importance in highly polluted atmospheres, it has become increasingly apparent that PANs may play an important role in regional and global scale chemistry.
TABLE 1-1. Chemical formulas and names of some of the more important peroxyacyl nitrates.
Name Acronym Structural formula
Peroxyacetyl nitrate PAN CH3C(O)OON02
Peroxypropionyl nitrate PPN CH3CH2C(O)OON02
Peroxy-n-butyryl nitrate PnBN CH3CH2CH2C(O)OON02
Peroxybenzoyl nitrate PBzN C6H5C(O)OON02
Peroxytrifluoroacetyl nitrate PF3AN CF3C(O)OON02
Peroxymethacryl nitrate MPAN" CH2=C(CH3)C(O)OON02 " MPAN is a product of the atmospheric photooxidation of isoprene.
1.3.3 The Physics of PAN The physical properties of PAN at room temperature have been difficult to study by classical chemical approaches because of the explosive nature of the pure compound in its liquid state. At standard ambient temperature and pressure, PAN exists as a colourless liquid with the physical properties as listed in TABLE 1-2 (Gaffney et al., 1989).
TABLE 1-2. Physical properties ofPAN
Property Value Comment Melting point -48.5 °C Boiling point 104.5 °C At 1013 hPa, extrapolated from vapour pressure Density 1.2 g/mL Estimated 5 Vapour pressure e·• B6!T(KJ • 19·06 hPa Temperature range 223-330 K p(298K) = 39.3 hPa Introduction 5
Spectroscopy The ultraviolet absorption spectrum of PAN is generally featureless in the region between 200 and 300 nm with a monotonic decrease in absorption over this region (Stephens, 1969). Talukdar et al. (1995) report an absorption cross section range from 4.2.10·18 (196 nm) to l. 7.10·23 cm2 molecu1e·1 (350 nm). The absorption in the range of tropospheric actinic UV (i.e., 290-400 nm) is relatively small and the lifetime of PAN with respect to photolysis is about a month. Since PAN was originally discovered from long-path infrared analysis, considerable effort has been made to describe its infrared spectrum (see for example Bruckmann and Willner, 1983). An infrared spectrum of PAN is shown in FIGURE 3-6 on page 28 of this thesis. For calibration, Kleindienst (1994) recommends the cross section for the band at 1163 cm·1 (i.e. 19 2 1 6.12.10· cm molecule" ).
Solubility PAN was initially thought to be rather soluble in water, since it was observed to undergo rapid base hydrolysis (Stephens, 1969). However, further studies have found that PAN is not very soluble in water and that under more normal atmospheric pHs (i.e. 4-5), it does not undergo rapid hydrolysis. The Henry's law coefficient has been found to be:
H(M/atm) - 9. J .10-rn e65131T(Kl (1-I) From this expression PAN has a Henry's law coefficient of 2.82 M/atm at 25 °C (Kleindienst, 1994 and references therein).
Dry Deposition
1 The dry deposition of PAN over water is reported to be smaller than 0.02 cm s· , while over 1 alfalfa and forested area values of 0.75 and 0.54 cm s· , respectively, are reported (Kleindienst, 1994 and references therein). Schrimpf et al. ( 1996) recently reported the deposition velocity above a field at night to be highly variable with an average of 0.54 cm s·1 and a standard 1 deviation of 0.94 cm s· •
1.3.4 The Chemistry of PAN
Reactions of PAN In the atmosphere, PAN undergoes relatively few reactions, and then with only the most reactive radicals. For the reaction of PAN with the HO radical, IUPAC (1996) recommends a value of 9.5·10·13exp(-650/T(K)) cm3 molecule·• s·•, which yields k = l.1 · 10·13 cm3 molecu1e·1 s·• for 6 3 298 K. Assuming an average 24-h HO concentration of 1·I0 molecules cm· , a lifetime of more than three months is found at 298 K. For higher peroxyacyl nitrates, a higher rate with HO is expected, due to the additional abstractable hydrogen atoms. 6 Introduction
TABLE 1-3. Rate coefficients and lifetimes for the thermal decomposition of PAN at various temperatures
Temp.f'C TempJK k/s-1 a Lifetime t b -30 243 1.84-10"8 1.7 yr -20 253 1.66-10"7 70 d -10 263 1.27-10 6 9.1 d 0 273 8.37-10"6 33 h 10 283 4.81.10·5 5.8 h 20 293 2.46·10·4 1.1 h 25 298 5.32.10·4 31 min 30 303 1.12.10·3 15 min 40 313 4.67-10" 3 3.6min 50 323 1.77·10" 0.94min 3 1 a Calculated using k0 = 4.9-10· exp(-12100/T(K)HMJ s and 16 k00 = 4.0·10 exp(-13600/T(K)) s from Atkinson (1992) at a pressure of 100 kPa. b ~ = l/k, assuming no back reaction
With a rate coefficient <2 x 10·14 cm' molecu1e·1 s·1 at 298 K (IUPAC, 1996), the reaction of PAN with Cl atoms is negligible, and also the reaction of PPN with Cl, with a rate coefficient of 1.1-10·12 cm3 molecule" 1 s·1 is of minor importance (Wallington et al., 1990). The most important atmospheric reaction for PANs is thermal dissociation, reaction(-!), which is highly temperature dependent. TABLE 1-3 shows rate coefficients and lifetimes with respect to thermal decomposition. In the absence of NO, if the N02 concentration is high enough, the products of PANs dissociation, acylperoxy radicals and N02, recombine to form PANs, and reactions (1) and (-1) would be in thermal equilibrium: kl RC(O)OO· + N02 + M RC(O)OON0 + M, (1, -1) k_I 2 where M stands for a third body that absorbs the energy of the reaction. Thermal decomposition is the dominant loss process up to - 7 km, above which photolysis takes over as the major loss process. The loss due to reaction with HO is very small throughout the entire troposphere (Talukdar et al., 1995).
Formation of Peroxyacyl Nitrates In the troposphere, PANs are formed during the photochemical oxidation of many organic molecules that contain more than one carbon atom. The immediate precursors of PANs are the
acylperoxy radicals, RC(0)02 • They are formed by oxidation of an acyl radical, RC(O).
RC(O)· + 0 2 + M RC(O)OO· +M (2) Introduction 7
The acyl radicals, RC(O), come, for example, from the hydroxyl radical attack on an alde- hyde (reaction 3), but also from photolysis of a ketone or from the thennal decomposition of an 13-ketoalkoxy radical.
RCHO+HO RC(O)-+H20 (3)
In the NO, polluted atmosphere acylperoxy radicals react either with N02 to fonn PAN according to reaction (J), or with NO to fonn a acyloxy radical that rapidly decomposes to give an alkyl radical and C02:
RC(O)OO· + NO RC(O)O· + N02 (4)
RC(O)O· R· +C02 (5)
Under low NO, conditions, the acylperoxy radical may react with H02 , other peroxy or peroxyacyl radicals. At night the reaction with NO, may become important. The relative importance of these channels is discussed in more detail in chapter 6 of this thesis.
1.3.5 Toxicology The effect of PANs on plants was mentioned before. The injury to plants has generally been described as a glazing or bronzing of the lower surfaces of leaves and in some cases involves reduced molecular oxygen evolution, tissue collapse, chlorosis, and induced leaf drop (Taylor, 1969). Since relatively small concentrations of PAN create profound effects in plants, it was recognised that PAN was influencing biochemical components (e.g., enzymes) present in relatively small concentrations (Mudd, 1975; Hippeli and Elstner, 1996). This may explain the observation that the sensitivity of different plant species to PANs may be very different. Some of the more sensitive plants are important agricultural crops such as bean, tomato and tobacco (Taylor, 1969). Small stinging nettle, which is very sensitive to PAN, has been proposed as a possible indicator plant for PAN (Posthumus and Tonneijck, 1992). The most apparent physiological effect on animals is the ability of PAN to serve as a strong lacrimator. Along with aldehydes, peroxyacyl nitrates are generally considered to be the main components of photochemical air pollution leading to eye irritation (Kleindienst, 1994). It is generally found that peroxy-n-butyryl nitrate (PnBN) and peroxypropionyl nitrate (PPN) are stronger lacrimators than PAN (Similar findings have been found for plant damage). Acute toxicity effects of PAN on laboratory animals have been reported both for mice and rats. For mice, the median lethal concentration (i.e. LC50) of PAN for a 2-h exposure was 106 ppm. Moreover in subacute exposures of rats to 12 ppm of PAN, a number of abnonnalities were found. In contrast to plants, animals do not show acute toxic effects (i.e. those measured) at levels typically found in urban atmospheres. However there have been suggestions that PANs may cause mutations in animal systems (Kleindienst, 1994). With regard to metabolic and pulmonary functions of human adults, a combination of ozone and PAN typically had a measurably greater effect than exposure to ozone alone. 8 Introduction
1.3.S Atmospheric Measurements Although PAN was first detected in ambient air using long-path infrared spectroscopy, the sensitivity of the method does not make it practical for routine ambient measurements. However one advantage of the infrared detection is the easy calibration due to the well-known infrared band strengths of PAN. With the invention of the electron capture detector (ECD) and it's inherent sensitivity and selectivity for PAN measurement (Darley et al., 1963), gas chromatography (GC) with ECD detection became the dominant technique for ambient air pollution studies. The ECD detector uses an ionising source of radiation to produce a standing current of electrons. By applying a pulsed voltage across two cell electrodes in the detector, the ion current can be monitored. When compounds with high electron affinities such as PAN are in the carrier gas, they can "capture" electrons and thus affect the current. For measurements of PAN in "clean air", cryogenic trapping and preconcentration can be used. The most difficult part of the PAN measurement by GC-ECD is the calibration of the detector. For field measurements typically errors of 10%-15% can be expected, but differences of up to a factor of 2 between two independent systems are possible (Neininger and Dommen, 1996, p. 99). Considerable effort has gone into performing ambient measurements of PAN in urban, non- urban, and remote locations. The highest average and maximum concentrations reported have been found in Los Angeles and the surrounding areas, where daily averages ranged from 10 to 30 ppb with maxima up to 210 ppb. Other cities show usually lower values. FIGURE 1-2 shows an overview over PAN measurements from all over the world. Diurnal profiles of urban PAN measurements have consistently shown daytime, usually mid-afternoon, maxima for PAN concentrations, and higher PAN concentrations in the summertime relative to the winter. See chapter 6 of this thesis for a discussion. Measurements at non-urban sites show in general average PAN concentrations that are substantially lower than those observed in urban air masses. Peak concentrations over I ppb are observed usually at times when heavy anthropogenic impact occurs (Roberts, 1990).
Air mass Urban ... ••I I • I Ill I I
Non-Urban I •I I I I I I
Remote 0 2 4 6 8 10 12 PAN (ppb}
FIGURE 1-2. Ambient concentration of PAN found in urban, non-urban and remote air masses. The diamonds show averages of measurement series, while the lines show maximum values. The numbers were taken from the compilation ofRoberts (1990). Introduction 9
Concentrations at remote sites are often quite low (<0.02 ppb). Wintertime concentrations are higher than summertime. In remote air masses, PAN is a major fraction of NOY (odd nitrogen, sum of NO, plus all oxidised nitrogen species that represent sources or sinks of NO.). Almost no site is not affeced by the transport of PAN (Singh and Salas, 1983).
OtherPANs Other PANs can be observed with a similar analytical technique as for PAN. PPN/PAN ratios are usually in the range 0.1-0.2. The lower concentrations of PPN relative to PAN are a reflection of the fact that sources of the parent compound, acety Jperoxy, are larger than those of propionylperoxy (Kleindienst, 1994). Similar behaviour is found for other PANs. An average PnBN/PAN ratio of 0.077 was reported by Grosjean et al. (l993c) for a southern California mountain forest location impacted by urban photochemical smog. PBzN/PAN ratios around 0.1 were found in the Los Angeles atmosphere and Williams II et al. (1993) reported an average MPAN/PAN ratio of 0.15, while measuring close to their detection limit of 0.15 ppb.
1.4 Ozone and PAN chemistry
Ozone is present as a natural component in the troposphere (Finlayson-Pitts and Pitts, 1986). One source is injections into the troposphere from the ozone layer in the stratosphere, where ozone is produced in relatively high concentrations by short wave solar radiation. This radiation is absorbed in the stratosphere and does not reach the troposphere. But ozone can also be generated chemically in the troposphere by the reaction of an oxygen atom (in the ground state) with molecular oxygen:
0(3P)+02 +M (6) The oxygen atoms are formed mainly from the photolysis of nitrogen dioxide: N0+0(3P). (7) Because ozone reacts rapidly with nitric oxide,
N0+03 (8) a photostationary steady state including reactions (6) to (8) is reached. This already shows the strong connection between the nitrogen and ozone chemistry but does not explain a net ozone production. However in the HO initiated oxidation of volatile organic compounds (VOCs, with
the simplified chemical structure R'H), NO can be oxidised to N02 without destroying an ozone molecule (see FIGURE 1-3). The HO radical, the key species in the atmospheric oxidation of voes is produced by the photolysis of 03 at wave length< 320 nm producing an
excited O('D) atom, which can react with water, H20, to form two HO radicals.
0 3 + hv(<320 nm) 0 2 +O('D) (9) O('D)+ HzO 2HO (IO) l 0 Introduction
A detailed introduction into atmospheric chemistry in general and ozone chemistry in particular can be found in the book of Finlayson-Pitts and Pitts (1986). A general introduction to the role of nitrogen in the chemistry of the troposphere and stratosphere is presented in the review of Crutzen (1979).
HO· :R'H+-R'· : H20 i02 . R'OO·
R'O·
FIGURE 1-3. Simplified scheme of the connection between VOC and azone chemistry.
In the NO, rich atmosphere, a large fraction of the organic peroxy radicals, R'OO·, react with NO:
R'OO·+NO R'O· +N02 (11)
However, if R' is an acyl group, that is R' = RC(O), the corresponding peroxy radical
RC(O)OO·, can also react with N02 through addition, as shown in reaction (1) on page 6. In this case no NO is oxidised and therefore no ozone produced. On the contrary, an N02 is removed, which potentially decreases the ozone concentration (see FIGURE 1-4). In general, it can be said, that the formation of peroxyacyl nitrate decreases the ozone concentration.
PAN Chemistry Ozone Chemistry HO· RC(O)H-+- RC(O)· H2o io2
+N02 RC(O)OO· RC(O)OONO~2 -N02
RC(O)O· i -- -R·+C02
FIGURE 1-4. Simplified scheme of the connection between PAN and ozane chemistry Introduction 11
Unlike many of the inorganic NOY species (e.g. NO, N02 , N03, N20 5), the PANs can have an appreciable atmospheric lifetime. This is due to their physical and chemical properties which were described earlier. PANs are photolysed slowly, react slowly with HO radicals, and have low dry and wet deposition velocities. This means that they may be transported into remote regions. At higher atmospheric temperatures, PANs may thermally decompose to reform an acylperoxy radical and a N02, according to reaction (-1 ). The N02 produced may photolyse and consequently form ozone. The acyl peroxy radical may oxidise an NO to form another N02 according to reaction (4). Depending on the structure of the resulting radical, R·, additional nitric oxide may be oxidised.
Therefore when PAN decomposes, it is a source of N02 and consequently ozone. Although the reactions shown in FIGURE 1-4 make up the main connection between PAN and ozone chemistry, there are a significant number of additional interactions, many of which are still not fully understood. In the chapters 6 and 7 of this thesis some of these reactions are investigated and discussed in more detail.
1.5 Computer Model of the Troposphere
The chemistry of the atmosphere is a highly non-linear system including hundreds of species and thousands of reactions. In this regard mathematical models have proven to be valuable tools, which can help to understand this system, to interpret field measurements and to develop control strategies for air pollution. Control strategies for urban ozone traditionally have been based on mass reductions in the release of volatile organic compounds (VOCs). Studies show, however, that some organic gas species form an order of magnitude less ozone than equal mass emissions of others. Using chemically detailed photochemical models the photo-chemical ozone (and PAN) creating potential for hydrocarbons can be calculated. An example see Carter and Atkinson (1989) or Derwent and Jenkin (1991), and for an overview and discussion of the uncertainties see Russell et al. (1995). Two classes of photochemical models are most extensively used: (1) Lagrangian models that typically include more detailed and explicit chemistry and (2) three dimensional Eulerian models that incorporate a more comprehensive treatment of meteorological processes but usually involve simplified chemistry. To understand field measurements or to make predictions, three dimensional models are often used, which describe the complicated connections between emissions, meteorology and chemical processes. Because the complexities of the problem exceed the capacities of the fastest computers, it is necessary to simplify all parts of the system. The challenge of model development is to carry out the simplification without omitting the contribution of any significant process. The simplification of a chemical mechanism is described in chapter 5 of this thesis. Whereas an Eulerian model calculates pollutant concentrations at a grid of fixed geographical location, the Lagrangian approach treats "air parcels" individually and follows the trajectory of an air parcel as predicted by the prevailing meteorology. As this parcel, 12 Introduction containing initial pollutant concentrations, moves, it is subject to fresh pollutant emissions, dilution, and chemical reactions. Lagrangian models are very useful for examining the impacts of new pollutant sources and can be less expensive to run than Eulerian models (Finlayson- Pitts and Pitts, 1986). Although the main reactions and species in atmospheric chemistry are known, we are still far from a full understanding of atmospheric chemistry. Computer modelling can help to understand the chemical mechanism. By changing parameters such as pressure, temperature, emissions or reaction rate parameters, it is possible to investigate the role and importance of reactions and species. One topic that warrants detailed examination is for example the time dependent evolution of ratios of various chemical species. Computer models may show if such ratios provide information on the history of air masses, as is suggested by simple theoretical considerations. Additionally, by identifying key species and reactions, computer models can lead to helpful suggestions for the planing of new laboratory and field studies. 13
2. Relative Rate Studies, Aldehyde + Cl
2.1 Introduction
2.1.1 Relative Rate Studies There are two main ways to determine a reaction rate coefficient. In absolute measurements the coefficient, k, is determined independently on an absolute base, whereas in relative rate studies, a rate coefficient, k, is measured relative to the rate coefficient, kREP of an other reaction. An absolute technique often used to investigate gas reactions relevant to the troposphere is the fast flow discharge system (Finlayson-Pitts and Pitts, 1986). It consists of an flow tube in which the reactants are mixed in the presence of a large amount of an inert gas such as He or Ar. One reactant, usually a radical, is generated with a discharge (e.g. microwave discharges). An other popular absolute technique uses flash photolysis to generate the reactive species in a cell (Finlayson-Pitts and Pitts, 1986). The decay of the reactive species is then monitored as a function of time, for example by using resonance or induced fluorescence. The advantage of absolute measurements is, that the rate coefficients can be used directly for applications such as computer modelling studies. Unfortunately absolute rate measurements require sophisticated equipment and often it is impossible to study a reaction under conditions close to the real atmosphere. Fast flow discharge systems for example operate at a pressure of few hPa and use He or Ar as a carrier gas. And in flash photolysis systems the choice of reactants is limited (Finlayson-Pitts and Pitts, 1986). Therefore many rate coefficients of interest are determined by relative rate studies. Thus if the absolute value for one rate coefficient, kREF• has been determined, the second one, k, can then be calculated from the experimentally determined ratio k/kREF' The kinetical treatment is relatively simple, if two competing reactions are compared:
A+X ~ Products k (2-1)
REF + X ~ Products (2-II) where A and REF are the reactants that compete for X. In this case it is sufficient to monitor the decrease of the species A and REF relative to their initial value, to determine the ratio kfkREF. Relative measurements can often be made with greater precision than absolute rate coefficient measurements, because only relative, not absolute, concentrations of A and REF need be measured. In addition, the species X, which is often a highly reactive free radical such 14 Relative Rate Studies, Aldehyde + Cl as HO or Cl need not be monitored in such experiments, Considering tropospheric conditions such as the presence of 0 2 and pressures around one atmosphere, the competitive technique can be applied under conditions more closely resembling those found in the troposphere than most direct techniques. For some applications the knowledge of the relative rates are even more important than the absolute one: for example for comparing lifetimes of organic compounds in the atmosphere or for some mechanistic studies.
2.1.2 The Flow System Relative rate measurements as well as absolute measurements can be performed either in a static system or in a flow system. In the first case the reactants are mixed in a cell, the reaction is started, and the change in the concentrations of reactants or products are monitored versus time. In the second case the species are usually monitored as a function of the path travelled in a flow tube. Although the competitive technique is more commonly applied in static systems such as smog chambers, in the present study a flow reactor has been used, nevertheless the concentrations of reactants were not monitored as a function of the reaction path, but as a function of time. Therefore it was somewhat a mixture between a static and a flow system. Semadeni (1993; 1994) used the same flow reactor to investigate relative rate-coefficients of organic compounds with HO under similar analytical techniques. However the gas-mixing was totally redesigned for the present experiments by using electronic mass flow controllers instead of a premixed bag mixture. One purpose of this experiments was therefore to get used to the new system and to test it
2.1.3 Aldehyde+ Cl While chlorine atoms, Cl., play a key role in the stratospheric ozone depletion, their role in the tropospheric chemistry is generally minor. Their concentration in the troposphere is usually orders of magnitude smaller than that of HO radicals, which show a similar reactivity toward organic compounds. But there has been recurring speculation over the years as to the possible role of atomic chlorine in the chemistry of the marine boundary layer. If the concentration of Cl· in the marine boundary layer is as high as Pszenny et al. (1993) suggested (i.e. 104 -105 3 molecule cm- ), then reactions of atomic chlorine should be taken into account when looking at the atmospheric chemistry of such sites. The comparison of the reactivity of Cl· and HO· toward organic species may be a source of mechanistic information. For example, Wallington et al. (l 988b) observed a linear relationship for the rate coefficients of Cl and HO with alkanes, but no correlation for oxygenated species. This finding may support the proposed complex between the HO· radical and the carbonyl group, prior to H-atom abstraction (Wallington et al., l 988a; Stemmler et al., 1997). Concerns regarding the environmental impact of chlorofluorocarbons (CFC) released into the atmosphere have led to an international effort to remove CFCs from industrial processes Relative Rate Studies, Aldehyde + Cl l 5
GC FID
FIGURE 2-1: Schematic diagram of the atmospheric flow reactor system as used for the propionaldehyde experiment. and consumer products. Hydrofluorocarbons (HFCs) are a class of potential CFC substitutes. In the light of the future widespread use of such compounds, detailed information on the atmospheric chemistry, and hence environmental impact, of these compounds is required.
CF3CHO is a likely oxidation product ofCF3CH3 (HFC-143) and CF3CH2CI (HCFC-133). The reactions
CH3CHO+CI· ~ Products, (I)
C2H5CHO +Cl· ~ Products (2) and
CF3CHO+Cl· ~ Products (3) are also of interest, since they have been used in the laboratory as a source of the acyl radical,
RCO·, and in the presence of 0 2 as a source of the acylperoxy radical, RC(0)02 ., which is a precursor of peroxyacyl nitrates. In fact, the aldehyde +Cl· system was mainly chosen for the present study, because it was originally planed to use it in the PAN experiments described in the chapters 3 and 4.
In the present study the rate coefficient of the reaction of propionaldehyde, C2H5CHO, and atomic chlorine, CI., was measured relative to the rate of acetaldehyde, CH3CHO, with atomic chlorine. To our knowledge, the only other measurement of the rate coefficient of propionaldehyde +Cl· was presented by Wallington et al. (l 988b).
Additionally, the rate coefficient of the reaction of Cl· with trifluoroacetaldehyde, CF3CHO, was measured relative to the reaction of chlorine with acetone:
CH3(CO)CH3 +Cl· Products (4) 16 Relative Rate Studies, Aldehyde + Cl 2.2 Experimental
For the experiment, a gas mixture containing the aldehyde, the reference substance and chlorine in air was drawn through an illuminated reactor. The mixing ratios of the aldehyde and the reference substance were monitored in the output stream as a function of time with the actinic light off and on. A schematic diagram of the apparatus used for the propionaldehyde experiment is shown in FIGURE 2-1. In the trifluoroaldehyde experiment the source of the aldehyde was modifies as explained below.
22.1 On-line Preparation of Gas Mixtures Mass flow controllers (MFC - Brooks, Model 5850E) where used to set up a steady flow of synthetic air containing about 20 ppm of acetaldehyde, 10 ppm of propionaldehyde and 100 ppm of chlorine for one set of experiments and 9 ppm CF3CHO, 8 ppm acetone and 70 ppm chlorine for the other set of experiments. The chlorine was taken from a cylinder containing
1000 ppm Cl2 in air. In the propionaldehyde experiment, to get the required small mixing ratios of aldehydes, a stream of 2.4 ml min·' of synthetic air was split using a set of fused silica tubes with diameters of0.52 and 0.12 mm. The smaller part of the flow (ca. 0.1 ml/min) was lead over a mixture of liquid acetaldehyde (Fluka,:?: 99.5%) and propionaldehyde (Merck, >98%), which was kept at 273 K using an ice bath. The combined streams were then mixed with the main stream as shown in FIGURE 2-1. The mixture of liquid aldehydes was chosen such, that the gas-phase concentrations of both aldehydes were similar despite the different boiling point. While this method was useful to produce a stable flow of aldehydes during the time-scale of a relative rate experiment, the long-time stability of the mixing ratios was not satisfactory. A possible reason were small changes in the flow through the capillaries due to diurnal changes in the room temperature. Additionally this method can be used only for substances whose boiling points are relatively close together and within a certain range. For this reason a different approach for adding the organics was used for the trifluoroacetaldehyde experiments as well as for the PAN and PPN experiments described later. A 25 drn3 glass bulb was filled with a mixture of the organics using a vacuum line and then synthetic air was added until a pressure of about 1250 hPa was reached. During an experiment the gas mixture was passed through an MFC into the main air stream. For more details please see the description of the PAN experiment in section 3.2.1 on page 24. For the trifluoroaldehyde study, the 25 dm 3 bulb was filled with a mixture of 1900 ppm trifluoroacetaldehyde and 1700 ppm acetone (Merck, >99% ). The trifluoroacetaldehyde was synthesised by the dropwise addition of trifluoroacetaldehyde ethyl hemiacetal (FLUKA) to a
H2SOJP20 5 slurry and trapping the aldehyde in a cool trap. The purity was tested using GC- MS analysis and by comparing the FT-IR spectrum of the sample to a literature spectrum (Berney, 1969). Both methods revealed a high purity of the trifluoroacetaldehyde. Relative Rate Studies. Aldehyde + Cl 17
2.2.2 The Reactor and Light Source The combined gas streams from the MFCs were passed through a gas-mixing device prior to entering the flow reactor. All connections were made with Teflon or PFA tubing (3 mm i.d.) which was covered with aluminium foil to prevent prephotolysis of the reactants. The reaction vessel, which is described in more detail in the section 3.2.1, is described here briefly. The Pyrex reaction tube had a diameter of 2.9 cm and a length of 56 cm. The vessel was illuminated through evacuated glass stoppers with quartz windows by a 1 kW Xe arc lamp. The light passed to the flow reactor through a set of collimating lenses, a water filter, an air mass filter (AM2, cut off at 330 nm).
2.2.3 Analytical Sys.tern Gas Samples were drawn from the exit of the reactor through a 2 m fused silica capillary (i.d. 0.56 mm) at a flow rate of 20 cm 3 min·' and transferred via an automatic gas sampling valve using a 500 µI stainless steel loop into a chromatograph which was equipped with an flame ionisation detector (GC-FID). The compounds were separated on a 16 m capillary column (4% PS-255), a gum-type methyl polysiloxane with 1-3% vinyl groups (Grob and 3 1 Grob, 1983), using H 2 as the carrier gas at a flow rate of 3 cm min· • For the propionaldehyde experiment, the GC was operated at 100 °C and the FID at 250 °C. The retention times under this conditions were 0.7 min and 0.9 min for the acetaldehyde and propionaldeyde respectively. For the trifluoroacetaldehyde experiment, the GC was operated at 45 °C and the FID at 150 °C. The retention times under this conditions were 0.55 min and 1.4 min for trifluoroacetaldehyde and acetone respectively. Calibration was carried out, by filling a Teflon bag with a known mixture of the species using a precision pressure gauge and a calibrated volume.
2.2.4 Experimental Procedure The gas mixture was drawn through the reactor at total flow rate of 15 to 40 cm3 min·', which gave a residence time between 9 and 24 min. The reactor was at atmospheric pressure (965 ± 5 hPa). Once reproducible concentration of the test and reference substrates were established, the flow reactor was illuminated by opening a shutter in the optical path. The resulting gas mixture was analysed in short time-steps (1-2 min). After an equilibrium was reached, the shutter was closed again as can be seen from FIGURE 2-2. In this way each analysed sample had a different photolysis time. This procedure confirmed that the system was stable over the time scale of the experiments. To test for photolysis of the substrates, a mixture without chlorine was illuminated in the same way. No photolysis was observed. For one experiment the photolysis time for all illuminated samples was kept constant, and the concentration of chlorine was varied instead. 18 Relative Rate Studies, Aldehyde + Cl
60 ... • 'ii' 50 • • > • • .§. 40 • • QI'. • 0 00 Cl> cP 0 • 0 0 0 • <(.. 30 0 0 : ...... 0 ~ 0 0 IV 0 Cl> 20 0 0 0 "- 0 c 0000000 cPo ii: 10 0 40 60 80 100 120 140 160 180 Time (Min)
FIGURE 2-2. Raw data of the propionaldehyde experiment. The filled circles show acetaldehyde, the empty circles the propionaldehyde.
2.3 Results
2.3.1 The Competitive Technique Assuming that the test (A) and reference (REF) substrates, as defined in eqs. 2-I and 2-11, are consumed solely by Cl· attack, the rate expressions of A and REF are as follows: :
d[A] k [A][Cl·] (2-III) dt d[REF] = kREF[REF][Cl·] (2-IV) dt Combining eqs. 2-111 and 2-IV to eliminate [Cl·], yields
_l_d[A] = k 1 (2-V) [A] dt kREF [REF] dt
Integration from time t = 0 when the initial concentrations are [A]0 and [REF]0, respectively, to time t when the concentrations are [A], and [REF],, gives eq. 2-VI:
In [Alo = _k_ln [REFJo (2-VI) [A], kREF [REF],
Provided that the reaction with Cl· is the only significant removal process for the substrate, plots of the data according to equation 2-VI verify the kinetic model if the data points lie on a straight line passing through the origin. FIGURE 2-3 shows an example which is based on the Relative Rate Studies, Aldehyde + Cl 19 raw data shown in FIGURE 2-2. The initial values were taken from the average of all unilluminated samples. The fitted line represent a least-mean-squares fit.
1.2 ...,.. "C ii c 0 ,. 'ii. 0.8 ...0 0.6 •.. e:,o :!:! ••• m 0.4 c 0 'ii. 0.2 .•.. ...0 a. • c 0 -..J -0.2 ·' -0.2 0 0.2 0.4 0.6 0.8 Ln(Aceta ldehydeo' Acetaldehyde ) 1
FIGURE 2-3. Plot of ln([propionaldehydelof [propionaldehyde],) versus ln([acetaldehyde]of[acetaldehyde],) for the photolysis of a mixture containing Cl,, air, ace/aldehyde and propionaldehyde at a temperature of295 K.
TABLE 2-1. Relative rate coefficients for the reactions of chlorine with a test substrate (A) and a reference.
A REF. k A+Cl·JkREF+CI· Temp./K
CH3CH2CHO CH3CHO 1.48 ±0.05 295
CH3CH2CHO CH3CHO 1.49 ±0.06 295 1.39 ±0.05' 295
CF3CHO CH3(CO)CH3 1.05 ±0.03 291 1.32±0.20 297
CF3CHO CH 3 C~CHO 0.042 ± 0.002 295 ' In this e"periment the chlorine concentration was varied and the illumination time was constant.
The rate coefficient ratios derived from the application of eq 2-VI to the substrate pairs studied are summarised in TABLE 2-1 where the errors quoted are 2o. 20 Relative Rate Studies. Aldehyde + Cl 2.4 Discussion
2.4.1 The Flow System The results of the present study show that the system works satisfactorily, including the newly designed on-line preparation of the gas mixture. This new technology greatly improved the flexibility of the system. The relatively fast response of the substrate concentration to changes in the light intensity (see FIGURE 2-2) indicates that there are plug flow conditions in the reactor and that diffusion is minor.
2.4.2 Proplonaldehyde + Cl vs. Acetaldehyde + Cl As calculated from the results presented in TABLE 2-1, the average ratio of rate coefficients
of the reactions of chlorine with propionaldehyde and acetaldehyde is k/k1 - 1.45 ± 0.11. The
higher value of k2 compared to k 1 can be explained by the additional two hydrogen atoms of the propionaldehyde which can be attacked by the chlorine. The relatively high difference between this two rates, which is almost proportional to the number of additional hydrogen atoms in propionaldehyde, suggests that the reactivity of the hydrogen in the aldehyde group to the chlorine attack can not be much higher than the reactivity of the other hydrogen atoms.
The ratio k 2/k1 presented here is within the error limits of the ratio Jc/k, - 1.34 ± 0.15 where both rate coefficients were measured by Wallington et al. (1988b) relative to the rate of the
reaction of ethane with chlorine at a temperature of 295 K. Taking k1 and k2 from the IUPAC evaluation (1996) gives a ratio k/k, - 1.67, which is somewhat higher than the ratio presented here. However kz in the IUP AC evaluation is based solely on the measurements of Wallington
et al. (I 988b ), while the value of k1 is the average of several different measurements. The ratio of the rate coefficients of the reactions of acetaldehyde and propionaldehyde with chlorine is higher than the corresponding ratio for the reactions of the aldehydes with HO, kcioctt20f0+ttdkctt3CH0+tto - 1.25, based the rate coefficients recommended by IUPAC (1996).
Using the absolute value k, = 7.2.10·11 cm3 molecu1e·1 s·1 (IUPAC, 1996), an absolute rate 3 1 coefficient, k2 = l.O±<:l.1·10.rn cm molecule·• s· , for the reaction of propionaldehyde +Cl· was derived. The quoted error represent 2o, and does not include any errors due to uncertainties in 10 3 the rate coefficient k1• This value is somewhat smaller than the value k2 "" 1.2.10· cm molecule·' s·1 recommended by IUPAC (1996), which is based on the measurements of Wallington et al. (1988b).
2.4.3 Trllluoroacetaldehyde + Cl vs. Acetone + Cl
Because the relative error of the two measurements of kifk4 is very different, weighted
average was calculated to yield k3'k4 - 1.09 ± 0.06. This value is within error limits of the
value k/k4 = 1.14 ± 0.04 reported by Scollard et al. (1993). Relative Rate Studies, Aldehyde + Cl 21
12 3 1 1 Using an absolute value k4 = 3.5·10· cm molecule s· as recommended by IUPAC (1996), 12 3 an absolute value k3 = 3.8±0.2· 10· cm molecule·' s· 1 was derived for the reaction of trifluoroacetaldehyde + CJ.. The quoted error represent 20", and does not include any errors due 12 3 1 to uncertainties in the rate coefficient k4• The value k3 = 2.7±0.1·10· cm molecu1e· s·' 12 3 determined by Scollard et al. (1993), which is based on a absolute value k4 = 2.37·10· cm 1 molecule·' s· is about 30% smaller than measured in the present work while the value k3 = 1.8±0.4·10·12 cm3 molecu1e·1 s·1 reported by Wallington and Hurley (1993), which is based on 13 1 12 the absolute values k(Cl+CH3CI) = 4.9.10· crri' molecule·' s· and k(Cl+C2H5Cl) = 8.0.10· 3 1 1 cm molecu1e· s , is more than 50% smaller than the present value. The discrepancy with the value of Scollard et al. can be explained by the different values of the reference rate coefficients used, while the discrepancy with the value of Wallington and Hurley may be based on the rate coefficient of the reference reaction or on systematic errors in one or both of the systems. However the calculated ratio of absolute rate coefficients k/k, = 0.038 ± 0.004 presented here is in good agreement with the relative rate coefficient ratio k/k, = 0.042 ± 0.002 measured directly in this study. 23
3. PAN Laboratory Study
3.1 Introduction
The precursor of peroxy acetyl nitrate, CH3C(O)OON02 (PAN), in the atmosphere is the acetyl peroxy radical (CH3C(O)OO-). They are formed for example from the hydroxyl radical attack on an aldehyde or from photolysis of a ketone. The initial reaction produces an acetyl radical which is rapidly oxidised to the corresponding acetylperoxy radical under atmospheric conditions:
HO·+CH3CHO Hp+CH,CO·
CH3CO· + 0 2 + M CH3C(O)OO· + M In the NOx polluted atmosphere, there are two main competing reactions for the acetylperoxy radical:
(1,-1)
(2) where reaction (2) is followed by the fast reaction
CH3C(O)O· (3) Note that since PAN is relatively unstable its decomposition via reaction (-1) has to be included at higher temperature.
The ratio of the rate coefficients, k/ki, and the thermal decomposition rate coefficient, k.1• are key values in the atmospheric chemistry of PAN. Computer modelling studies of the troposphere have shown that the uncertainties in the rate coefficients of the PAN chemistry are important when calculating the ozone formation (Yang et al., 1995). The kinetics of the
thermal decomposition of PAN have been extensively studied and the rate coefficients (k.1) are well established (Atkinson et al., 1992). Absolute rate coefficients have been reported (Addison et al., 1980; Basco and Parmar, 1987; Bridier et al., 1991) for reaction (1) and there have been two recent direct measurements (Maricq and Szente, 1996; Villalta and Howard,
1996) of ti over a range of temperatures. The rate coefficient ratio, k/k2, has been measured in several relative rate studies for PAN (Cox et al., 1976; Cox and Roffey, 1977; Hendry and Kenley, 1977; Kirchner et al., 1990; Tuazon et al., 1991) and a few analogous peroxyacyl nitrates (Kenley and Hendry, 1982; Kerr and Stocker, 1985; Kirchner et al., 1992; Zabel et al., 1994).
Prior to the present study, however, only indirect determinations of the ratio k/k2 for PAN
have been carried out. Cox et al. (1976) and Cox and Roffey (1977) measured the NO-to-N02 24 PAN Laboratory Study conversion and absolute yields of PAN and derived k/k.i from a kinetic analysis involving complex reaction schemes. Hendry and Kenley (1977), Kirchner et al. (1990) and Tuazon et al. (1991) investigated the rate of the thermal decomposition of PAN as a function of the
[NO]/[N02] ratio and obtained values ofk/k.i at temperatures above 283 K.
In the present work, the ratio, k/k2, was measured by a method involving the production of : acetylperoxy radicals from the photolysis of biacetyl in the presence of 0 2
(CH3CO)i + hv 2CH3CO· (4)
CH3CO· + 0 2 + M CH3C(0)00· + M (5)
By measuring the resulting relative PAN concentrations as a function of the [NO]/[N02] ratio it was possible to measure k1/k2 at atmospheric pressure and at lower temperatures than hitherto reported.
3.2 Experimental
3.2.1 The Flow System A schematic diagram of the apparatus is shown in FIGURE 3-1. The experiments involved flowing an air mixture through an irradiated reactor with a residence time of about 50 s, and analysing the reactants and PAN product either with an NO,-chemiluminescence analyser or by gas chromatography (GC-ECD-FID). The experiments were carried out at atmospheric pressure (965 ± 5 hPa) and over the temperature range 247-343 K.
GC ECO/ FID
FIGURE 3-1. Schematic diagram of the atmospheric flow reactor system
A steady flow of synthetic air containing biacetyl, NO and N02 in the ppm range was established through the reaction vessel. Mass flow controllers (MFC-Brooks, Model 5850E) consisting of a flow sensor, a control valve and an electronic control system allowed the separate control of all the concentrations without affecting the total flow, which was constant 3 1 over the time scale of the experiments and usually at the rate of 400 cm min· • The mass flow
controllers were calibrated for air (80% N2 and 20% 0 2) and were unaffected by the low PAN Laboratory Study 25 concentrations of reactants in the mixtures. These calibrations were regularly checked using soap-bubble flow meters.
NO in nitrogen and N02 in air were taken from commercial cylinders. For biacetyl, an 8000 ppm mixture in synthetic air was prepared and stored in a 25 dmJ glass bulb at a pressure of about 1250 hPa. To prepare such a mixture, the evacuated bulb was first filled with about 10 hPa of the biacetyl using a precision pressure gauge and then synthetic air was added. Preliminary attempts to generate the air-biacetyl mixture by multiple dilution of a slow flow of air bubbled through a liquid sample of biacetyl failed, owing to the lack of long-term stability. The combined gas streams from the MFCs were passed through a gas-mixing device consisting of a glass tube containing a series of Teflon baffles (Sulzer Chemtech) prior to entering the flow reactor. All connections were made with Teflon or PFA tubing (3 mm i.d.) which was covered with aluminium foil to prevent prephotolysis of the reactants.
FIGURE 3-2. Detailed drawing of the flow reactor.
FIGURE 3-3. Picture of the flow reactor with a length of56 cm.
The flow reactor (see FIGURE 3-2 and FIGURE 3-3) has akeady been used to investigate gas-phase reactions under conditions related to the troposphere (Harris and Kerr, 1988; Semadeni et al., 1993). The Pyrex reaction vessel consisted of three concentric chambers. The inner tube was the actual reaction vessel with an internal diameter of 2.9 cm and a length of about 56 cm. The vessel was irradiated through evacuated glass stoppers with quartz end windows. A thermostated fluid consisting of a 1: l mixture of propylene glycol and water was circulated (thermocirculator, Lauda RKS 20) through the middle chamber to maintain a uniform temperature in the reaction vessel. Before entering the reaction vessel, the gas mixture was equilibrated at the reactor temperature by passage through a Pyrex coil surrounded by the 26 PAN Laboratory Study thennostated fluid in the middle chamber. A metal shield was wrapped around the outer surface of the inner glass tube to prevent the irradiating light beam from entering the middle chamber. The outennost chamber was a vacuum jacket, to reduce the heat flux between the reactor and its surroundings. The reaction vessel was illuminated by a solar simulator (Polytec XS 1000) consisting of a 1 kW Xe arc lamp. The light passed to the flow reactor through a set of collimating lenses, a water filter, an air mass filter (AM2, cut off at 330 nm) and an N02-filter. The latter was used to minimise photolysis of N02 in the reactor and consisted of a glass cell with quartz windows filled with air containing about 1.5% N02 at atmospheric pressure and had a transmission of 10% at 400 nm and 20% at 460 nm. Under these conditions biacetyl was photolysed over the spectral region 350-450 nm but photolysis of N02 was negligible.
32.2 Analysis Analysis of NO,. A commercial chemiluminescent NO-NOz-NO, analyser (Thermo
Environmental Instruments Model 42) was used to measure NO and N02 as well as for carrying out absolute PAN calibrations. The instrument monitors the fluorescence from the reaction of nitric oxide with ozone:
For the N02 measurements the sample gas passes through a stainless steal-converter operating at a temperature of 625 ·c, before entering the analysis chamber. The converter reduces nitrogen dioxide (as well as PANs) to nitric oxide. The instrument was calibrated with standard mixtures of NO and N02• Additionally, the conversion rate (99%) was checked by gas-phase titration of the NO standard with ozone.
0.33 ~ 0.32 (ij c Cl 0.31 (i) 0 0.3 () w 0.29
FIGURE 3-4. Gas chromatogram showing about 2 ppb of PAN
PAN Measurement. The PAN was measured by a gas chromatograph (Carlo Erba HRGC 5300) operated at 35 "C and fitted with an electron capture detector (ECD) at 100 °C. Gas Samples were drawn from the exit of the reactor through a 2 m fused silica capillary (i.d. 0.56 mm) at a flow rate of 20 cm3 min ·1 and transferred via an automatic gas sampling valve using a PAN Laboratory study 27
100 µI stainless steel loop into a cryotrap (Carlo Erba). The compounds were separated on a 16 m capillary column (4% PS-255), a gum-type methyl polysiloxane with 1-3% vinyl groups 3 1 (Grob and Grob, 1983), with H2 as the carrier gas at a flow rate of 10 cm min · • The PAN retention time was about 2 min under these conditions. FIGURE 3-4 shows a sample chromatogram collected and processed using a personal computer and a chromatographic software package (Maxima 820, Carlo Erba Instr.). Linearity and calibration checks of the ECD were carried out by measuring ppb levels of PAN simultaneously with the GC-ECD and the NO,-analyser. The PAN samples for the calibration were prepared by passing air through a solution of PAN in n-tridecane and subsequently diluting this gas stream with pure air using mass flow controllers. The ECD response for PAN was linear up to a mixing ratio of 500 ppb and the detection limit (3 o) was 0.5 ppb. Additionally a calibration was done by measuring a PAN sample also with an FT-IR spectrometer (BIORAD FTS45).
3.2.3 Measurement of the Rate Data
During each experiment only the concentrations of NO and N02 were varied. All other parameters such as temperature, total flow, reaction time, light intensity and the biacetyl concentration were kept constant. A typical set of experimental results under the fixed conditions described above is shown in FIGURE 3-5. Starting with no NO in the mixture, the
[NO)/[N02] ratio was increased to a value of about 4 and subsequently decreased stepwise back to O and at each selected value of the [NO)/[N02) ratio the relative amount of PAN formation was measured. This procedure confirmed that the system was stable over the time
6 1.610 4 6 ------ti) 1.410 ···-- Y'f' ...... 3.5 ~ 1.2106 3 cu 'f' f 6 < 110 • 2.5 ..:.: 'f' cu 5 [NO] Q) 810 • 2 [N02 ) Q. 5 0 610 "'f' • 1.5 0 5 • w 410 z •'f' < 5 •• 'f' Q. 210 •••• ...... 0.5 0 0 0 20 40 60 00 100 Time(min)
FIGURE 3-5. Plot of the measured relative mixing ratios of PAN(•) as a function 1 1 0 of [NOJl[NO,] (")for T 25 °C. The dashed line shows the PAN N° " value used in the calculations. 28 PAN Laboratory Study
scale of the experiments. Typical levels of reactants and products, in units of ppm, were: [NO],
0--4; [N02], 1-5; [biacetyl],25; and [PAN],0.01--0.l. As a check on the proposed mechanism for PAN formation in this study, a range of additional experiments was carried out as follows (i) as described above but with different 3 1 flow rates, 153 and 916 cm min· , (ii) with the photolysis of CH3COC1 as the radical source,
(iii) with the photolysis of Cl2 in the presence of CH3CHO as the radical source and (iv) with the reaction vessel lined with a thin Teflon sleeve. In all cases the kinetic data were consistent with the results of the main studies (see Results and Discussion).
32A PAN Synthesis PAN and PPN were synthesised in the liquid phase using a method described by Gaffney et al. (1984) and Nielsen et al. (1982). This involved the nitration of peroxycarboxylic acid with nitric acid:
RC(O)OOH + HN03 H2 + RC(O)OON02 For the synthesis of PAN a 50 ml round bottom flask was filled with 25 ml n-tridecane (>97%, F1uka Chemie) and cooled to 0 °C in a ice bath. Then 640 µl of peracetic acid (32 % solution in dilute acetic acid, Aldrich) and 2 ml of sulphuric acid (95-97%, Merck) were added to the solvent. After stirring and cooling for five minutes, 500 µm of nitric acid (65%, Merck)
0.02
'N 0.015 0 ·~··· Q) ..-~w (.) (]' c 0 ' 0 2000 1500 1000 1 Wavenumber [cm' ] FIGURE 3-6. FT-IR gas spectrum of PAN (optical path length 10 cm; resolution 1 2 cm· ); ""48 ppm PAN in synthetic air. The type of the modes were assigned using the information provides by Bruckmann and Willner ( 1983 ). PAN Laboratory Study 29 was carefully added in 50 µI aliquots. The resulting mixture was stirred for 15 more minutes and then poured onto 30 ml of ice water in a separator funnel. The water phase was discarded and another 30 ml of ice water was poured into the funnel. The tridecane layer containing the peroxyacyl nitrate was separated, dried with approximately 2 g of magnesium sulphate, filtered and stored at -20 °C. FIGURE3-6 shows a infrared spectrum of the synthesised PAN in the gas phase. 3.2.5 Materials The synthetic air (PanGas) was a mixture of 20% 0 2 and 80% N2 • Mixtures of nitric oxide (1004 ± 2 ppm) in N2 and nitrogen dioxide (960 ± 2 ppm) in synthetic air were supplied by Carbagas. The following chemicals were used without further purification other than bulb-to- bulb distillation: biacetyl (2,3-butadione) (Fluka, 3.2.6 Kinetic Model of the Reaction Scheme For a better understanding of the effect of the decomposition reaction (-1 ), a simple model study was carried out using FACSIMILE'" v. 3.05 (AEA, 1994). An experiment as described above was simulated using the program listed in Appendix A. The rate coefficients for the reaction scheme (reactions (1) to (5)) were taken from IUPAC (1992), except the rate coefficient of reaction (2), which was calculated from the rate coefficient of reaction (1) and the ratio determined in the experimental study. The photolysis rate of biacetyl was chosen so that the resulting PAN concentrations were similar to the experiment. To determine k/k2 for a given temperature, a set of runs was made for various ratios NO/N02 and the resulting PAN concentrations were treated the same way as those measured in the experiment. 3.3 Results In the presence of relatively high concentrations of NO and NO,, the formation of PAN and C02 is governed by reactions (1) and (2) above. Thus we can write the rate equations: d[PAN), = k [CH CO ·) [NO ] (3--I) dt t 3 3 t 2 t d[C02 ]' = k [CH CO ·] [NO] (3-II) dt 2 3 3 t t on the assumption that reaction (-1) is negligible, which holds at low temperatures. Taking the ] ratio of eqs 3-I and 3-II and integrating over time assuming that [N0]0 and [N02 0 are constant gives the following equation =~[N0 ) [PAN], 2 0 (3-Ill) [C02 ], k 2 [N0]0 where the subscripts 0 and t indicate concentrations at the beginning of the experiment and at time t respectively. 30 PAN Laboratory Study Initially it was planed to measure the C02 formation as well as the PAN formation. This, however, was not possible as that the PAN decomposed to yield C02 in the heated infrared instrument which was used to measure the C02 from the reaction. There were also additional problems in dealing with the ambient levels of C02 in the atmosphere of the laboratory which interfered with the product C02 levels in the ppb range. It was decided to resort to a kinetic analysis based solely on the measurement of the PAN formation. If the production of the acetyl peroxy radical is independent of the [NO]/[N02] ratio and there are no reactions of the acetyl peroxy radical which are dependent on the [NO]/(N02] ratio other than (1) and (2) then the sum of ([PAN].+ [C02J.) is independent of the [NO]/[N02] ratio. This sum can be determined by measuring [PAN]~NOJ=O, which is [PAN], when [NO]/(N02] - 0 and therefore [C02], = 0. Thus, independent of the [NO]/[N02] ratio, [C02], can be expressed as ] [PAN)~Noi~o [C02 1 = -[PAN],. (3-IV) Combination of eqs 3-III and 3-IV yields the expression [PAN]!,NOJ=o [NO] k __o _l_+J. (3-V) [PAN], [N02 ) 0 k, 0 0.5 1.5 2 2.5 3 3.5 4 [NO]/[NO) FIGURE 3-7. Kinetic plot of the ratio of PAN measured as a function of [NO/l[N02/ according to equation 3-V for T = 25 "C, based on measurements shown in FIGURE 3-5. On the basis of eq 3-V the ratio of the rate coefficients k/ki was derived by a linear least- squares fit from the plots of [P ANJ~NOJ=o /[PAN], as a function of [NO]/[N02]. A typical plot of the present data is shown in FIGURE 3-7. which is derived from the raw data of FIGURE 3-5. PAN Laboratory Study 31 TABLE 3-1. Rate Coefficient Ratios, k11kv for PAN as a Function of Temperature at Atmospheric Pressure (965 hPa) TempJK k.Jk2" TempJK k/k • ···---···---·---···------2 247 0.417±0.019 290 0.430 ± 0.025 253 0.391±0.015 290 0.444 ± 0.046'.g 263 0.401 ± 0.015 290 0.468 ± 0.019[,g 273 0.436 ± 0.02° 298 0.405 ±0.013 282 0.428 ± O.oI8 310 0.324 ± 0.015' 283 0.417 ± 0.02' 343 0.044 ± 0.005 8 ' Errors are the sum of 2a from linear least-squares fit and 2o of PANINOl-tl determination. • Acetyl chloride as the radical source. 'Acetaldehyde + Cl, as the radical source. 'Teflon sleeve inserted into the reactor.' Total flow 153 cm' min·'. 'Total flow 916 cm' min·'. 'Not used lo calculate the average (see text). The rate coefficient ratios found in this study are listed in TABLE 3-1, where the errors quoted are 2u. No significant temperature dependence is found in the temperature range 247- 298 Kand the average of these measurements is 0.41±0.03, where the error is two times the standard deviation. Above 298 K the decomposition reaction ( -1) can not be neglected and these measurements yield a ratio which is underestimated as will be discussed later. 3.4 Discussion 3A.1 Checks on the Proposed Mechanism of PAN Formation As noted in section 3.3 Results on page 29 and in TABLE 3-1, the measurements ofki/ki at 310 and 343 K show considerably lower values of the ratio than over the temperature range 247-298 K. This effect can be explained by the onset of the decomposition of PAN, reaction (-1 ), in relation to the time scale of the flow experiments. Thus at 343 K we calculate the half- life of PAN at high [NO]/[N02] ratios to be 3.5 s compared to the residence time in the flow 3 1 system of -50 s for the standard flow rate of 400 cm min· • As further evidence of this situation the experimental results were compared to those from a kinetic model of the reaction scheme as described in section 3.2.6 on page 29. The results are shown in FIGURE3-8. where the agreement is reasonably satisfactory. Further checks on the mechanism were carried out with alternative sources of acetyl radicals, namely the photolysis of CH,COCl (with the air mass filter removed) and the photolysis ofCl2 in the presence ofCH3CHO. In both cases the results for k/k2 were consistent with the data derived from the biacetyl photolysis. It should be noted, however, that when 32 PAN Laboratory Study relatively high ratios of [CH3CHO]/[Clz] were used, the values ofk/k2 increased to -0.6. This was probably due to the presence of substantial concentrations of HO radicals in the system, generated from the oxidative chain reaction (Cox and Roffey, 1977) involving CH3• radicals produced as result of reaction (2). These HO radicals add to the generation of PAN via HO- initiated attack on CH 3CHO which affects the assumption that the rate of formation of the acetyl radical is independent of the [NO]/[N02] ratio. Such additional HO-initiated PAN formation is not possible with the photolytic radical sources, CH3COCOCH3 and CH3COC1, since HO attack on both of these molecules is unlikely to yield acetyl radicals and hence acetylperoxy radicals and PAN under the conditions of the present experiments. Additional experiments were performed with the main radical source from the photolysis of biacetyl but with higher and lower flow rates through the reaction vessel, as recorded in TABLE 3-1. The results show no systematic trend with increasing flow rate, although in view of the higher errors involved at the lower flow rate and the relatively low mixing ratio of NO and N02 at the higher flow rate, these data were not included in calculating the average value of k/kz over the temperature range 247-298 K. Schurath and Wipprecht (1980) found that PAN may decompose on a surface. Langer et al. ( 1992) have reported that heterogeneous reactions of PAN are too slow to have an influence on PAN decomposition under typical atmospheric conditions, but may affect laboratory measurements of PAN, through the decomposition of the acetylperoxy radical or PAN itself at a surface. They found that the decomposition rate of PAN in air due to glass surfaces follows the relation 0.5 0.4 'ii '8 ::IE k 0.3 .. 1 J!!:I CL ~ 02 E 0 0 0.1 0 -40 -20 0 al 40 Temperature (°C) FIGURE 3-8. Comparison of present data on k,lk2 with a kinetic model of the reaction scheme:(•) measured ratios, (solid line) modelled ratios. PAN Laboratory Study 33 d[PAN] =-~([PAN]· 7·107 +[CH C(O)OOJ· 5·1012 )·e-9382 ncm s·'. (3-VI) dt v 3 In the present system, with a surface to volume ratio, S/V, of about 1.4 cm·• assuming a temperature of 298 K, a PAN concentration of 0.1 ppm, a N02 concentration of 5 ppm and calculating the acetyl peroxy concentration using the equilibrium constant of Bridier et al. (1991 ), the lifetime of PAN in respect to heterogeneous reactions is about 130 h. This is far in excess of the short residence time of PAN (-50 s) in the flow reactor. As further evidence that heterogeneous reactions are not important in the present system, an experiment was carried out with the Pyrex reaction vessel lined with a Teflon sleeve. The kinetic data measured with the Teflon surface were consistent with the results of the main studies carried out with a Pyrex surface (see TABLE 3-1). The above kinetic treatment assumes that [NO] and [N02] do not change significantly during the course of the reaction. Of course, reactions (I) and (2) affect the mixing ratios of these two compounds and every methyl radical produced leads to further NO-to-N02 conversion (Cox et al., 1976). However, in the present system the changes in the mixing ratios of N02 and NO brought about by these reactions are relatively small since the sum of the mixing ratios of NO and N02 was at least an order of magnitude higher than the mixing ratios of the products, PAN and C02• In addition, the photolysis of N02 was minimised by the presence of the N02-filter in the light train. Direct experimental confirmation of the essential stability of the ratio [NO]/[N02] during the course of a reaction was obtained by switching on and off the light or the flow of biacetyl while monitoring NO and N02 • In addition the reaction of peroxy acetyl radicals with other peroxy radicals can be important in the atmosphere under conditions of low NO,, but were not important in the system owing to the high NO, mixing ratio. The photolysis of the PAN product during the course of the experiments can be neglected for a number of reasons. Firstly there is very little overlap of the absorption spectrum (Talukdar et al., 1995) of PAN with the photolytic light (A. > 330 nm). In addition, during the course of a run < 0.5 % of the biacetyl was photolysed and the absorption crosssections of biacetyl are orders of magnitude higher than those of PAN under the experimental conditions. Finally, if the PAN photolysis was significant it would be expected that the values of k/k, would show a dependence on contact time i.e. flow rate through the reactor, which is not the case (see data of TABLE 3-1). 3.4.2 Comparison with Literature Data FIGURE 3-9 shows a plot of the present results in relation to previous determinations of k/k, as a function of temperature and at pressures close to I atm. The temperature independent value of k/k2 - 0.41 ± 0.03 from this study is in agreement with the value k/k, - 0.48 ± 0.07 recommended in the IUPAC evaluation (1992) and by Wallington et al. (1992) on the basis of the data of Cox and Roffey, (1977) Kirchner et al. (1990) and Tuazon et al. (1991). Note that in the present study it was possible to extend the relative rate measurements down to a 34 PAN Laboratory Study temperature of 247 K and at the same time have considerably improved the precision of the measurements. Also shown in FIGURE 3-9 are values of k/k2 calculated from absolute measurements of k 1 and k2 • The values of k1 are from Bridier et al. (1991) which form the basis of the recommendations for k 1 in the IUPAC evaluation (1992) and the NASA panel evaluation (DeMore et al., 1994). These two groups differ in their representation of the pressure effect on k 1 according to the Troe formulation. Here k 1 was calculated at 1 atm pressure as a function of 28 1 3 1 1 temperature from the values ko= 2.7 x 10· (T/298r [M] cm molecu1e· s· , k~ = 1.21 x 11 0 9 3 1 10· (T/298)' • cm molecu1e· s·1, Fe= 0.30, and Ne= 1.41 derived by Bridier et al. (1991) over the temperature range 248-393 K. These k 1 values have been combined with the two recent determinations of k2 to derive the lines plotted in FIGURE 3-9. Villalta and Howard (1996) 12 3 1 1 have reported k2 = 8.1 x 10· exp (270/T) cm molecu1e· s· over the temperature range 200- 402 K, from a direct study of the acetylperoxy radicals using mass spectrometry. Maricq and 12 3 1 1 Szente (1996) on the other hand have reported k2 = 2.1 x 10· exp (570/T) cm molecu1e· s· over the temperature range 228-353 K, from transient IR absorption of N02 and time resolved UV spectroscopy of the acetylperoxy radicals. It is clear from the two resulting lines shown in FIGURE 3-9 that there is a significant difference, about 30% at 298 K, in the calculated values 0.8 k 0.6 0 6. 0 \1 1 oo<> ¢'? \1 • • I e 02 0 0 -40 -20 0 2) 40 ro ro Temperature (°C) FIGURE 3-9. Comparison of present data on k,lk, with literature values: (9) this study, (0) this study, at temperature where reaction (-I) is significant; (l::i.) Cox et al. (1976); fi/) Cox and Roffey (1977); (D) Kirchner et al. (1990); {<>)Tuazon et al. (1991); (------) Hendry and Kenley (1977); ( -- ) ratio of the absolute rate coefficients of Bridier et al. ( 1991) and Villalta and Howard (1996); ( - - - ) ratio of the absolute rate coefficients ofBridier et al. (Bridier et al., 1991) and Maricq and Szente. ( 1996) PAN Laboratory Study 35 of k/Js at atmospheric pressure based on the studies of Villata and Howard (1996) and of Maricq and Szente (1996). Until this discrepancy is resolved it is difficult to compare the calculated values of k/Js with those measured in the relative rate studies, although at face value the calculated ratios k 1/Js derived from the study of Villata and Howard (1996) are in better agreement with the present data than are those of Maricq and Szente (1996). 3.4.3 Extension of the Techniques It is likely that the measurement techniques and the kinetic treatment developed for the present study can be extended to measure other reaction systems involving semistable products of competing reactions. Note that it is sufficient that the species is stable over the time-scale of the reaction and analysis (- l min), not over the time-scale of a experiment (- 2 hours). 0 The techniques still works if it is not possible to measure PAN1N°1" , if for any reason it is impossible to get an NO free system. Equation 3-V can be modified to yield: -[PAN], x [NO lo ~ + [PANJ;Noi~o [PAN], (3-VII) ] [N02 0 k 1 On the basis of eq 3-VII the ratio of the rate coefficients k/k2 can be derived by a linear least-squares fit from the plots of [PAN], as a function of -[PANJi x [NOJi[N02] 0 • The value of [PAN];NOJ•o is not necessary for this treatment, but it can be obtained from the intercept. Some of the PAN data were re-evaluated using eq 3-VII instead of eq 3-V. The ratio of the rate coefficients was the same for both cases, but the error limits were somewhat smaller for most measurements with eq 3-VII. 3.4A Atmospheric Implications While at altitudes above 7 km photolysis is the major loss process of PAN (Kleindienst, 1994; Talukdar et al., 1995), the lifetime, i:, of PAN in the warmer parts of the troposphere is controlled by the rate of its thermal dissociation, reaction (-1 ). In addition, however 't is dependant upon reactions (1) and (2) and may be expressed according to eq 3-VIII, based on a steady-state assumption for the acetylperoxy radical concentration (Cox and Roffey, 1977): 't (l + .!i_ [NO 2 ] )(k )-' k [NO] -i 2 (3-VIII) For example, at a temperature of 263 K in the free troposphere, taking a typical [NO]/[N02] ratio of 3 (Carroll et al., 1990), the lifetime is 20 days, based on the value of k/Js determined here and the value of k_ 1 recommended by IUPAC (1992). This lifetime is 2.2 times lower when calculated simply on the basis of reaction (-1 ). Thus the ratio of rate coefficients, k/Js, plays a significant role in the lifetime of PAN in the troposphere and consequently affects the transport of PAN and therefore the ozone generation. 37 4. PPN Laboratory Study 4.1 Introduction Peroxyacetyl nitrate (PAN, CH3C(O)OONO,), is the most abundant of the peroxyacyl nitrates found in the troposphere. It has been extensively monitored and its chemistry, toxicology and role in NOY transport have been investigated in many studies (Stephens, 1969; Gaffney et al., 1989; Roberts, 1990; Atkinson, 1994; Kleindienst, 1994). Recent studies show that higher peroxyacyl nitrates, especially peroxypropionyl nitrate (PPN, CH 3CH2C(O)OON02), are also often present in ambient air. [PPN]/[PAN] ratios in the range 0.1-0.2 have commonly been reported (Shepson et al., 1992; Grosjean et al., 1993c; Kourtidis et al., 1993; Williams II et al., 1993; Grosjean et al., 1996), but ratios as high as 0.28 have been observed (Grosjean et al., 1993c). There have been indications that PPN could be several times more phytotoxic than PAN (Taylor, 1969) and that PPN may also have biogenic precursors (Grosjean et al., 1993b). In addition tropospheric measurements of PAN and PPN and of the [PPN]/[PAN] ratio have been interpreted to provide information about the history of air masses (Grosjean et al., 1996). In computer modelling studies of the troposphere, to save memory and processor time, all peroxyacyl nitrates are frequently grouped into one model species. Owing to the limited kinetic dataset for higher peroxyacyl nitrates and the large scatter of the data, the kinetic parameters for PAN are often taken to represent those of all the relevant peroxyacyl nitrates. This assumption needs to be tested by further kinetic studies of the higher peroxyacyl nitrates. The precursors of peroxyacyl nitrates in the atmosphere are acyl radicals (RCO·), which are formed from the hydroxyl radical attack on an aldehyde, but also from photolysis of a ketone or from the thermal decomposition of an 13-ketoalkoxy radical. The initial acyl radical is rapidly oxidised to the corresponding acylperoxy radical under atmospheric conditions: RC0·+02 +M RC(O)OO·+M In the NO, polluted atmosphere there are two main competing channels for the acylperoxy radicals: RC(O)OO· + N02 + M RC(O)OON02 + M (1,-1) RC(O)OO· + NO RC(O)O· + N02 (2) where reaction (2) is followed by the fast reaction RC(O)O· (3) 38 PPN Laboratory Study Note that since peroxyacyl nitrates are thermally unstable their decomposition via reaction (-1) becomes significant at higher ambient temperatures. In addition to the decomposition rate coefficient (k.1) the ratio a - k/ls plays an important role in formation and decomposition of peroxyacyl nitrate in the atmosphere. Computer modelling studies of the troposphere have shown a high sensitivity of the calculated ozone formation to the rate coefficients of the PAN chemistry (Yang et al., 1995). For PAN the ratio a'PAN has been studied in detail (Cox et al., 1976; Cox and Roffey, 1977; Hendry and Kenley, 1977; Kirchner et al., 1990; Tuazon et al., 199 l; Seefeld et al., 1997) and a limited number of analogous studies of higher peroxyacyl nitrates have been reported (Kenley and Hendry, 1982; Kerr and Stocker, 1985; Kirchner et al., 1992; Becker and Kirchner, 1994; Zabel et al., 1994). For PPN Kerr and Stocker (1985) measured k/ls= aPPN at a single temperature (302 K) from a kinetic analysis of the rates of formation of NO, N02 and PPN in a flow system while Becker and Kirchner (1994) reported a value [NO]/[N02] ratio. Here we report measurements of the ratio, k/k2 = a.PPN• over a range of tropospherically relevant temperatures by a method involving the production of propionylperoxy radicals from the photolysis of propionyl chloride in the presence of Oz: CH3CH2COC1 + hv CH,CHzCO· + Cl· (4) CH3CHzCO· +Oz + M CH3CH2C(O)OO· + M (5) By measuring the resulting relative PPN concentrations as a function of the [NO]/[N02] ratio it was possible to measure a.PPN at atmospheric pressure and at lower temperatures than hitherto reported. In a second set of experiments a.PPN was measured relative to a.PAN• with the same flow system involving a reactant mixture of acetyl and propionyl chlorides. The relative measurement reduces the possibility of systematic error when comparing a.PPN to a.PAN from the literature data. 4.2 Experimental Section 4.2.1 Flow System The experimental system has been described in detail in section 3.2. l on page 24 of this thesis and will only be discussed briefly here mainly to show the differences. The experiments involved flowing an air-reactant mixture through an irradiated reactor with a residence time of about 50 s, and analysing the reactants and PPN product either with an NO, - chemiluminescence analyser or by gas chromatography (GC-ECD). The experiments were carried out at atmospheric pressure (965 ± 5 hPa) and over the temperature range 249-346 K. PPN Laboratory Study 39 A steady flow of synthetic air containing propionyl chloride, NO and N02 in the ppm range was established through the reaction vessel. Mass flow controllers allowed the separate control of all the concentrations without affecting the total flow, which was constant over the time 1 scale of the experiments and usually at the rate of 400 cm' min- • For the reactant propionyl chloride a mixture in synthetic air (6500 ppm) was prepared and stored in a 25 dm 3 glass bulb at a pressure of about 1250 hPa, by first evacuating the bulb and filling with about 10 hPa of the propionyl chloride using a precision pressure gauge and then adding synthetic air. The combined gas streams from the MFCs were passed through a gas-mixing device prior to entering the flow reactor. The reaction vessel consisted of a Pyrex tube with an internal diameter of 2.9 cm and a length of about 56 cm, which allowed precise control of temperature and reaction time. The reaction vessel was illuminated by a 1 kW Xe arc lamp. The light passed to the flow reactor through a set of collimating lenses, a water filter, and an N02-filter to minimise photolysis of N02 in the reactor. The resulting spectrum has a main peak from 400 nm to 1000 nm with a maximum at 700 nm and a small tail down to 250 nm. 42.2 Analysis The PPN was measured by a gas chromatograph (GC) operated at 35 °C and fitted with an electron capture detector (ECD) at 100 °C. Gas Samples were drawn from the exit of the reactor and transferred via an automatic gas sampling valve using a 100 µl stainless steel loop into a cryotrap. The compounds were separated on a 16 m capillary column (4% PS-255) 3 1 (Grob and Grob, 1983) with H2 as the carrier gas at a flow rate of 10 cm min- • The retention times of PPN and PAN were 4.0 and 2.0 min under these conditions. Linearity checks of the ECD were carried out by measuring ppb levels of PPN simultaneously with the GC-ECD and the NO,-analyser. NO and N02 were measured with a commercial chemiluminescence NO- N02-NOx analyser fitted with a stainless steel N02 to NO converter operating at a temperature of625 °C. 42.3 Measurement of the Rate Data During each experiment only the concentrations of NO and N02 were varied. All other parameters such as temperature, total flow, reaction time, light intensity and the propionyl ] chloride concentration were kept constant. Starting with no NO in the mixture, the [NO]/[N02 ratio was increased to a value of about 4 and subsequently decreased stepwise back to 0 and at each selected value of the [NO]/[N02] ratio the relative amount of PPN formation was measured. This procedure confirmed that the system was stable over the time scale of the experiments. Typical levels of reactants and products, in units of ppm, were: [NO], 0-4; [N02], 1-5; [propionyl chloride], 24; and [PPN], 0.01-0.1. A typical set of results under the conditions described above is shown in FIGURE 4-1. An attempt to produce propionyl radicals from the photolysis of 3,4-hexadione resulted in an unstable PPN signal. The photolysis of Cl 2 in the presence of CH3CH2CHO was not used, 40 PPN Laboratory Study because this system is sensitive to HO production as explained in section 3.4. l on page 31 of this thesis. 2 3 __ ... --...... - --.. --·-· -. 1.5 oD D 2 N U) • D 0 -~ 1 z z • 0 0. D -z 0. • 1 0.5 0 • D •••• D 0 D 0 -20 0 20 40 60 80 100 120 TIME (Min) FIGURE 4-1. Plot of the measured peak area of PPN (•)as a function of [NO]l[NO,] 1 1 (0) for an experiment at T = 302 K. The dashed line shows the PPN N° "" value used in the calculations. The x-axis shows the time scale of the experiment, which is not required in the kinetic treatment of the data. 4.2.4 Combined PAN and PPN measurements In a second set of experiments, the propionyl chloride was replaced by a mixture of propionyl and acetyl chlorides obtained by filling the 25 dm3 glass bulb with about 10 hPa of each compound and then adding synthetic air until a total pressure of about 1250 hPa was reached. The larger number of components in the reactant mixture, with a subsequently more complex analyses, made the accurate measurement of PAN and PPN more difficult. This was alleviated by removing the N02 filter, which lead to larger PAN and PPN peaks, but also to an increased photolysis of N02• Additionally the changes in NO and N02 arising from reactions (1) and (2) were increased and hence the NO-to-N02 ratio was no longer as constant over the reaction time (see discussion below). 4.2.5 Materials Most of the chemicals have been described in section 3.2.5. Propionyl chloride (Fluka, >98%) and 3,4-hexadione (Merck, >99%) were used without further purification other than bulb-to-bulb distillation. PPN Laboratory Study 41 42.6 PPN Synthesis PPN synthesised by the liquid phase nitration of perpropionic acid in n-tridecane solution was used for peak identification of the chromatogram and for absolute calibration. PPN was synthesised using the same method described for PAN in section 0 which is based on a description of Gaffney et al. (1984). The perpropionic acid was made by mixing 75 ml propionic acid anhydride (>98%, Merck) with 0.5 ml sulphuric acid (95-97%, Merck) and carefully adding 10 ml Hp2 (Perhydrol® 30% pro analysis, Merck) using a dropping funnel. An ice bath made sure the temperature in the solution was always below IO °C. The mixture was stirred for a further hour and stored at room temperature over night. The peroxide contents of the solution (12%) was tested using iodometry (Becker and Co-workers, 1976). 4.2.7 Peroxytrifluroacetyl Nitrate (PF 3AN) and Peroxybenzoyl Nitrate (PBzN) An attempt was made to synthesise peroxytrifluroacetyl nitrate (PF3AN) in the same way as PAN and PPN. However it was not possible to identify the corresponding peak in the chromatogram because of the many side products of the synthesis containing fluorine atoms, which also give signals on the ECD. Additionally the retention time of PF3 AN is probably shorter than those of PAN on the column used, which leads to interference with the oxygen peak. The same problems arose when PF,AN was synthesised in the gas phase by illuminating a mixture of trif1uroacetaldehyde, N02 and Cl2 • Injecting samples of PF3AN synthesised in the gas or liquid phase into a gas chromatogram coupled with an mass spectrometer (GC-MS) did not allow an identification of the PF3AN peak either, because a) the sensitivity of the MS to PF3AN is orders of magnitude smaller than the sensitivity of an ECD and b) most observed peaks showed a similar mass spectrum. When synthesising PBzN by reaction ofbenzoylperoxide with BF4N02 (Kenley and Hendry, 1982), a violent explosion destroyed the fume cupboard and no further experiments with PBzN were carried out. 4.3 Results 4.3.1 k1/k2 for PPN In the presence of relatively high concentrations of NO and N02 , the formation of PPN and (CH3CH2 • + CO;i) are governed by reactions (1) and (2) above. Assuming that reaction (-1) is negligible, which holds at low temperatures, leads to the rate expressions d[PPN], ---=[CH,CH 2 C(O)O,.J[N02 ]k1 and (4-I) dt d[CH,CHd, (4-II) dt Combining eqs 4-1 and 4-II and integrating the result over time assuming that [NO] and [N02) are constant leads to the following equation 42 PPN Laboratory Study [PPN], _!s_ [ N 0 ] 2 0 (4-III) [C02 ]1 k 2 [N0]0 where the subscripts 0 and t indicate concentrations at the beginning of the reaction and at time t respectively. The ratio k/ki is derived from a kinetic analysis involving solely the measurement of the PPN fonnation as a function of the [NO]/[N02] ratio. This approach is based on the fact that the sum ([PPN], + [C02] 1) is independent of the [NO]/[N02] ratio, which holds assuming that the production of the propionylperoxy radical is independent of the [NO]/[N02) ratio and that there are no reactions of the propionylperoxy radical which are dependent on the [NO)/(N02] 1 0 ratio other than (1) and (2). This sum can be detennined by measuring [PPNJ;N° • , which is [PPN), when [NO]/[N02] = 0 at which point [C02), - 0. Thus, independent of the [NO]/[N02) ratio, [C02) 1 can be expressed as ] [C02 1 [PPNJ;NOJ•O -[PPN]" (4-IV) Combining eqs 4-III and 4-IV and replacing k,lki by a.PPN yields [PPNJlNOJ=O (4-V) [PPN], On the basis of eq 4-V the ratio of the rate coefficients a.PPN = k/kz was derived by a linear least-squares fit from the plots of [PPN];NoJ=o /[PPNJ. as a function of [NO]/[N02], as shown in FIGURE4-2. 8 6 z D...... /1 D.. i" 4 ~ •• z •• D.. D.. 2 .•. • •• 0 0 0.5 1 1.5 2 2.5 NO/NO 2 FIGURE 4-2. Kinetic plot of the ratio of PPN measured as a function of {NO]/ {NO,] according to 4-V for T = 302 K, based on measurements shown in FIGURE 4-1. The rate coefficient ratios found in this study are listed in TABLE 4-1, where the errors quoted are 2o. No significant temperature dependence is found over the temperature range PPN Laboratory Study 43 249-302 K and the average of these measurements is k,lk2 = 0.43 ± 0.07, where the error limits are twice the standard deviation. Above 302 K the decomposition reaction (-1) can not be neglected and these measurements yield a ratio which is underestimated. This effect can be explained by the onset of the decomposition of PPN, reaction (-1), in relation to the time scale of the flow experiments, as discussed in more detail in the seetions 3.2.6 and 3.4.1 of this thesis. TABLE 4-1. Rate Coefficient Ratios, k1/k2, for PPN as a Function of Temperature at Atmospheric Pressure (965 hPa) Temp./K k,/k,' Tem12./K k/k2' 249 0.454 ± O.Q38 290 0.421 ± 0.053 258 0.438 ± 0.028 302 0.398 ± O.Q18 268 0.415 ± O.Dl8 310 0.345 ± 0.035b 273 0.494 ± 0.016 322 0.284 ± 0.040b 279 0.400±0.016 346 0.026 ± 0.002b "Errors limits are the sum of 2o from linear least-squares fit and 2o of PPN[NOJ-O determination. 0Not used to calculate the average because thennal decomposition of PPN can not be neglected at these temperatures. 4.3.2 Relative Measurements kik:i for PPN vs. klkz for PAN The peroxyacetyl radical shows the same reactivity towards NO and N02 as the peroxypropionyl radical and therefore the ratio, a = k,!Js, is important in both cases. Although the simultaneous measurement of aPPN and aPAN is theoretically possible with the present experimental system, the experiments were not carried in this way owing to the change in the NO and N02 concentrations which are described in the experimental section. However the system is suitable for measuring the ratio aPPN/aPAN' By combining eq 4-V and the corresponding equation for PAN, [PANJ;NOJ~o (4-VI) [PANJi we get 1 0 [PPNJ\N° " /[PPNJ. (4-VII) [PANJ\NOJ~o /(PAN), -1 UPPN The same equation can be reached by combining the rate equation for PAN d[PANJ =[CH C(O)O ·][NO ]kPAN and (4-VIII) dt 3 2 2 I (4-IX) 44 PPN Laboratory Study with eq 4-1 and 4-11 and integration as a function of time. In this case NO and N02 cancel out before the integration. It follows that NO and N02 do not have to be constant over the reaction time for equation 4-VII to be valid. On the basis of eq 4-VII, the ratio of the a.PPN/a.m, was derived by a linear least-squares fit from plots of ( [PPN];NOJ=o /[PPN], -1) as a function of ( [PAN];No1=o /[PAN]1 -1) as shown in FIGURE 4-3. The ratios found in this study at three temperatures are listed in TABLE 4-2, where the errors quoted are 2o. No systematic trend with temperature is found and average of these measurements is a.PPN/a.PAN = 0.89 ± 0.13, where the error limits are twice the standard deviation. TABLE 4-2. Relative Rate Coefficient Ratios, a.,,;a,,AN' as a Function of Temperature at Atmospheric Pressure (965 hPa) TempJK (l.PPN/(J.PAN • 262 0.92±0.09 284 0.85 ±0.16 295 0.91 ±0.13 •Errors are the sum of 2o from linear least-squares fits, 2o for PAN[NOJ-O and 2o for pp~NOJ-0 determination. 5 4 z a. a. 3 en 6'- ~ 2 za. a. 0 0 2 3 4 5 PAN[NOJ::O/PAN FIGURE 4-3. Plot of the ratio of PPN measurements as a function of the ratio of PAN measurements with varying [NO]/[NO,] values, according to eq 4-V/l at T 295 K. PPN Laboratory Study 45 4.4 Discussion 4.4.1 Comparison with Literature Data FIGURE 4-4 shows a plot of the present results in relation to previous determinations of a.PPN as a function of temperature and at pressures close to I atm. Our temperature independent value of aPPN = 0.43 ± 0.07 is in agreement with the value aPPN = 0.53 ± 0.11 recommended by Atkinson (1994), which is based on the measurements of Kerr and Stocker (1985) but is significantly smaller than the value aPPN = 0.70 ± 0.06 of Becker and Kirchner (1994). Unfortunately there appear to be no absolute determinations of k 1 or k 2 in the literature to compare with our relative rate measurements of k/k, for the propionylperoxy radical. The present value of ratio (1994) and a:PAN measured by the same group and by the same technique (Kirchner et al., 1990) yields the ratio, aPPN/a.PAN = 1.67 ± 0.31 which is significantly higher. 0.8 0.7 0.6 0.5 k it 0.4 2 0.3 0.2 0.1 0 0 -40 -20 0 2 0 4 0 60 80 Temperature (°C) FIGURE 4--4. Comparison of present data on k,lk, for PPN with literature values: (e) this study; (0) this study, at temperature where reaction ("l) is significant; (0) Kerr and Stocker (1985 ); (0) Becker and Kirchner (1994). The reason for the smaller value of the direct determination of a:PPN/a.PAN compared with the independent measurements of aPPN and 4.4.2 Tropospheric Lifetimes of PPN and PAN The lifetime, 1:, of peroxyacyl nitrates can be expressed in tenns of reactions (1), (2) and (-1) according to the equation (Cox and Roffey, 1977) •=(l+S,_[N02])(k )-' (i+a.[N02]J(k )-1. (4-X) k, [NO] _, [NO] -i FIGURE 4-5 shows the expected lifetimes of PAN and PPN calculate according to eq 4-X using a a.PPN - 0.43 and a.PAN= 0.41. k";N was taken from Mineshos and Glavas (1991) and k~fN was taken from IUPAC (1996) for a pressure of 1013 hPa. While the temperature is the dominating parameter for the lifetime, it can be seen that the NO/NO ratio is much more important than the difference between the kinetics of PAN and PPN. 1 08 HO+ PAN 1 07 e PAN Photolysis ... 106 (!) E ;:: (!) 105 -:i 104 103 -20 -10 0 10 20 30 40 Temperature (°C) FIGURE 4-5. Lifetime of PPN (straight line) and PAN (dashed line) with respect to the thermal decomposition as a function of temperature for a series of NO,INO ratios calculated according to equation 4-X. For comparison the figure also shows the lifetime of PAN with respect to the HO reaction (IUPAC, 1996) calculated with a HO concentration of 10·• molecule cm·' and the lifetime of PAN with respect to the ground level photolysis rate coefficient (averaged for July 4) as taken from figure 6 ofTalukdar et al. ( 1995) without consideration of the small temperature dependence. PPN Laboratory Study 47 4A.3 Implications for Computer Modelling «PAN and a.PPN are close enough to justify using a.PAN for the ratio of the combined reactions in computer models which lump species and reactions. However in computer models which have a species representing PPN or higher PANs, the usage of the somewhat higher ratio, aPPN• for PPN is recommended. The ratio, a., of the reactions of acylperoxy with N02 and NO presented in this study is somewhat smaller than the one used in many computer models today. It can be expected that the new value leads to somewhat smaller predicted levels of PAN in the troposphere. A more detailed analysis of this conclusion is presented later in chapter 7 of this thesis, which discusses the sensitivity of a tropospheric box model to a. 4.4A Measuring the Ratio for Other PANs With the technique presented in this work it should be possible to measure the ratio of the coefficients of the reactions of the acylperoxy radical with NO and N02 for other peroxyacyl nitrates. As described in section 4.2.7 an attempt was made to measure the ratio for peroxytrifluroacetyl nitrate (PF3AN). PF3AN is expected to be a product of the photochemical degradation of CF3CH3 (HFC 143), a substitute for fully halogenated chlorofluorocarbons (Wallington et al., 1994b; Zabel et al., 1994). The CF3CO, CF3, CF30 2 and CFp radicals shows quite different reactivities to those of the corresponding CH3CO, CH3, CH30 2 and CH30 radicals (Wallington et al., 1994a; Dibble et al., 1995; Maricq et al., 1995). Consequently the chemistry of PF3AN may be somewhat different from that of PAN. This may explain the difficulties in identifying the PF3AN peak in the chromatogram as described in section 4.2.7. Possibly a different detection techniques such as FT-IR is necessary for this purpose. For the synthesis in the liquid phase it may be more appropriate to use the reaction of trifluoroaeetyldehyde with N20 5 (Solomon, l 973a; l 973b ). An other problem may be to find a suitable source for the CF3C(0)02 radical. CF3C(O)Cl is not suitable for this purpose because its major photochemical dissociation pathway is the formation of CF3 , CO and Cl· (Maricq and Szente, 1995). Zabel et al. ( 1994) reported a ratio kCF3c(oJ02+No/kcFJqo)m+No = 0.64 at a temperature of 315 K from the thermal decomposition of CF3C(0)02N02 as a function of the NO/N02 ratio. The CF3C(0)02N02 was synthesised in the gas phase by illumination of an mixture of CF3CHO, Cl2 and N02• This ratio is 50% larger than the corresponding ratio for PAN as reported in this 12 thesis. On the other hand the ratio of the absolute determination of kc.3qoJo2+Noi = 6.6 · 10· 3 1 1 11 3 1 cm molecu1e· s· from Wallington et al. (1994b) and kcFiC(O)oz+No = 2.7 · 10· cm molecu1e· 1 s· from Maricq et al. (1996) is kcF3C(O)oz+Nozlkc.3c PF3AN chemistry by measuring kCF3C(oJ02+N02/kcFJC(0)02+No at different temperatures and possibly by different techniques. Peroxybenzoyl nitrate (PBzN) is a product of the photoehemical degradation of aromatic compounds such as toluene in the presence of N02 • PBzN concentrations between 0.1 and 2 48 PPN Laboratory Study ppb are often found in polluted air but concentrations up to 5 ppb have been reported (Roberts, 1990 and references therein). PBzN was reported by Reuss and Glason ( 1968) to be 200 times more irritating to the human eye than formaldehyde. This would mean than PBzN is approximately 100 times more effective as an eye irritant than PAN (Taylor, 1969). The literature contains reports of two measurements of the ratio, ka;H5qoJo2+No!ka;HS3C(o)02+No• which are in agreement. Kenley and Hendry (1982) presented a ratio of 0.66 at 304 K, while Kirchner et al. (1992) reported a ratio of 0.63 in the temperature range 310-322 K. However no measurement at lower temperatures have been made. By using the measurement technique reported in this study it should be possible to extend the temperature range. This would require the setting up of an analysis method for PBzN. A first step could be to synthesise PBzN in the liquid phase according to the description of Grosjean et al. (1993a), which is probably safer than the method tried in the present study. An other important member of the PAN family is peroxymethacrylic nitrate (CH2=C(CH3)C(0)0iN02, MPAN), which is a product of the secondary photooxidation of isoprene (Roberts and Bertman, 1992). MPAN concentration up to 1.2 ppb have been observed (Grosjean et al., 1993c; Williams II et al., 1993). No measurement of the ratio of the corresponding acylperoxy radical with NO and N02 appear to have been reported in the literature. In applying the present technique to measure this ratio, it is important to keep in mind that MPAN may react with ozone or other reactive species in the mixture. 49 5. Box Model and Chemical Mechanism 5.1 Introduction The chemistry of the atmosphere is a very complex system. It includes thousands of chemical reactions in the gas-phase, in the aqueous-phase and on particle surfaces. It is influenced by the meteorology and the actinic flux, as well as by emissions and deposition. A complete mathematical model of the troposphere would have to include a detailed description of all of these components. Because of the complexity of any single component, most models of the troposphere omit some of the components, or use very simplified representations of them. Nevertheless many simplified models are able to make quite accurate predictions of ozone concentrations. Additionally, simplified models allow investigations of cause and effect relationships which are not possible when using a more complete model. This chapter describes the mathematical model of the troposphere used for the studies presented in chapter 6 and 7 of this thesis. The model includes a detailed gas-phase chemistry, but no reactions in the aqueous phase or on aerosols. Because it is a zero-dimensional box model, many meteorological factors such as transport are omitted. A radiative transfer model is used to calculate the actinic flux. Although deposition was omitted, emissions into the box were included. 5.2 The RACM Mechanism The gas-phase chemical mechanism is one of the most important components of an atmospheric chemistry model. A chemical mechanism is a simplified representation of the known and assumed gas-phase chemistry. There are several important mechanisms which are widely used for modelling the chemistry of the troposphere including: the mechanism of Lurmann et al. (1986), the Carbon bond IV mechanism of Gery et al. (1989) and the mechanism for the Regional Acid Deposition Model (RADM2) (Stockwell et al., 1990). For example, the RADM2 mechanism is used in many photochemical transport I transformation atmospheric chemistry models to predict concentrations of oxidants and other air pollutants. The "Regional Atmospheric Chemistry Mechanism" (RACM) used in this study is a completely revised version of the RADM2 mechanism (Stockwell et al., 1990). It is described in detail by Stockwell et al. (1997). The following overview is presented here and details important for the studies described in chapter 6 and 7 are discussed. 50 Box Model and Chemical Mechanism The mechanism includes 17 stable inorganic species, 4 inorganic intennediates, 32 stable organic species ( 4 of these are primarily of biogenic origin) and 24 organic intermediates (Appendix B). The RACM mechanism includes 237 reactions (Appendix C). 5.2.1 Inorganic Chemistry Tropospheric inorganic chemistry is relatively well known. The RACM mechanism has a reasonably complete set of inorganic reactions. The inorganic rate coefficients were set to the values recommended by DeMore et al. (1994). Care was taken to include a complete set of night-time reactions, for example by including the N03 self reaction. The reaction was excluded from the RACM gas-phase mechanism, because the homogeneous gas-phase reaction rate is very slow and experimental measurements are uncertain due to a strong heterogeneous component. 5.2.2 Organic Species and Chemistry Emissions' inventories include hundreds of emitted volatile organic compounds (VOC) (Middleton et al., 1990). Given the huge number of all possible reactions and intennediates resulting from atmospheric oxidation of these emissions, it is necessary to group organic compounds together to fonn a manageable set of model classes. For the same reason many multiple pathways are fonnulated as one reaction in the RACM mechanism and not all organic intermediates (for example alkyl radicals) are explicitly described. Aggregation of Organic Compounds For the RACM mechanism the hundreds of VOCs emitted in the real atmosphere are aggregated into 16 anthropogenic and 3 biogenic model species. The grouping of organic chemical species into the RACM model species is based on the magnitudes of the emission rates, similarities in functional groups and reactivity of the compounds towards HO (Middleton et al., 1990; Stockwell et al., 1990; Stockwell et al., 1997). The rate coefficients, for the reactions of the model species, were calculated as the weighted mean of the rate coefficients of all chemical species aggregated together into a single model species The method of reactivity weighting was used to account for the differences in reactivities between chemical species and model species. The reactivity weighting was based on the assumption that the effect of an emitted chemical species on a simulation is approximately proportional to the amount of the compound that reacts with HO on a daily basis. Under this assumption an emitted compound can be represented by a model species that reacts at a different rate provided that an aggregation factor is applied to the emission rate. Box Model and Chemical Mechanism 51 Alkanes Methane and ethane are treated explicitly in the RACM mechanism as CH4 and ETH. All other alkanes along with alkynes, alcohols, esters and epox.ides are aggregated according to their HO rate coefficients into three additional model species, HC3, HCS and HC8. Alkanes react with HO by the abstraction of hydrogen atoms. The product of this reaction is an alkyl radical, which under atmospheric conditions react immediately with 0 2 to form an alkyl perox.y radical. In RACM the intermediate alkyl radical is omitted and the resulting perox.y radicals are represented as MO,. ETHP, HC3P, HCSP and HC8P. Carbonyls The carbonyl species in the mechanism include formaldehyde (HCHO), acetaldehyde and higher saturated aldehydes (ALD), acetone and higher saturated ketones (KET), unsaturated dicarbonyls (DCB), glyox.al (GL Y) and methylglyox.al and other species of the form RC(O)CHO (MGL Y). The reactions of HCHO are represented by an explicit set of reactions. The chemistry of ALD is treated as acetaldehyde. One important product of the photochemical oxidation of carbonyls, are acylperoxy radicals, which are represented by AC03• Acylperoxy radicals may react with N02 to form acylperox.y nitrate (PAN). Alkenes Four model species were used to represent the emitted anthropogenic alkenes. Ethene is explicitly represented by ETE. Propene and other terminal alkenes are represented by OLT. Internal alkenes are represented by the model species OLI. Dienes are aggregated into DIEN. As well as reacting with HO, alkenes can also react with ozone and NO,, which leads to a large number of reactions and products. Possible secondary products are for ex.ample unsaturated perox.yacyl nitrates, TPAN. Aromatic Compounds The aromatic species are grouped into three model species. TOL for toluene and less reactive aromatics, XYL for xylene and more reactive aromatics and CLS for cresol and other hydroxy substituted aromatics. Aromatic compounds are oxidised in the atmosphere through reaction with HO which can either abstract a hydrogen or add to the aromatic ring. 52.3 Radical Chemistry Kirchner and Stockwell (1996) showed that many radical-radical reactions, for ex.ample the reactions of acyl perox.y radicals with organic perox.y radicals and the acylperox.y self-reaction, may have an important influence on PAN and ozone concentrations. Special care was therefore taken to include the reactions of organic perox.y radical with NO,, HO; and with other organic perox.y radicals in the RACM mechanism. The ex.tended perox.y radical chemistry, together with the limited complexity of the hydrocarbon chemistry, make the RACM mechanism very useful for the study of the chemistry of PAN in the troposphere. 52 Box Model and Chemical Mechanism 5.2.4 The Future of the RACM Mechanism The RACM gas-phase mechanism is still developing and improving. New laboratory data will be added to the mechanism in a manner that improves its chemistry but without greatly increasing its complexity. A main goal will be to develop a more objective method for the aggregation of chemical species and to incorporate the organic chemistry using a template reaction approach so that the mechanism can be more easily adapted to represent the compounds in different emission inventories. 5.3 Photolysis Rate Coefficients Sunlight is the main driving force behind the chemistry of the atmosphere. Molecules are dissociated by sunlight into fragments which are often highly reactive. Models of the chemistry of the atmosphere must include an accurate description of photodissociation processes. For the present study, the photolysis rate coefficients for the 23 photochemical reactions were calculated according to the methode of Madronich (1987). For a species i, the photolysis rate coefficient, Ji, is the integral of the product of the actinic flux jA(/c,t), the absorption cross section o~A,), and the quantum yield FIGURE 5-1 shows the photolysis rate coefficients used for N02 and 0 3 for winter and summer scenarios. 35 30 O> -7 25 1i &. a: - 20 .!! c Ill Cl> 15 ~:2 0 Cl> 10 &.-= 0 0.. (.) 5 6 12 18 24 Time of the day (h) 0.12 0.10 Cl>';"- 1i &. 0.08 a: - .llJ c 0.06 ~·~ 0.04 s=0 Cl> &. 0 0.. (.) 0.02 6 12 18 24 Time of the day (h) FIGURE 5-1. Photolysis rate coefficients of NO, and 0 3• calculated/or a longitude o/0° and a latitude of 40°; straight line: summer day (21 June); dashed line: winter day (21 December). 5.4 The SBOX-Model-System A new box model system was developed. The new model, called "SBOX", is based upon a "chemical compiler" and a FORTRAN library to solve the ordinary differential equations (ODE). The new box model is much easier to use and has much more logical input and control files than the former box model, which was based on the old CHEMK model (Whitten and Meyer, 1975). FIGURES-2 shows an overview of the box model. 5.4.1 The Chemical Compiler One characteristic of SBOX is, that the mechanism is compiled and linked into the program and not read in from a parameter file at execution time. This means that a new box model program needs to be created every time the mechanism is changed. Advantages of this approach are: a fast code, a high flexibility, and a high compatibility with 30 models. 54 Box Model and Chemical Mechanism Photolysis Source Code Mechanism DATA Box Model: SBOX File Solver: VODE (or SENDDM) Photolysis Program Chemical Compiler FORTRAN77 Mechanism Photolysis Input COMPILER FORTRAN SOURCE Parameterflle SBOX Outputfiles FIGURE 5-2. Schematic diagram of the SBOX box model system. In the diagram the square boxes signify ASCII files, and the rounded boxes signify executable binary files. For a description see text. The creation of a new box model program with a new or modified mechanism was simplified by writing a chemical compiler. The chemical compiler reads an input file in which the mechanism's chemical reactions and their rate coefficients are written in a form that is very natural to chemists. Rate coefficients may be given in various formats including, "Arrhenius", "Troe" or coded in FORTRAN. The chemical compiler translates the mechanism directly into a set of FORTRAN routines, which can be compiled and linked to the new box model or alternatively it may become possible to directly incorporate them into an 3D- Eulerian chemical transport model. The following subroutines were provided by the mechanism code: RATE: Calculates the rate coefficient as a function of pressure and temperature for all reactions except the photolysis reactions. No parameterisations are made and all expressions including Troe behaviour are fully coded. This slows down the calculation, but makes sure, that the rate coefficients are valid for large temperature and pressure ranges. F: Calculates the rate expressions (derivatives of the concentrations) which define the ODE system. If C is the vector of all species concentrations and p is a vector of all system parameters (rate coefficients, product yields, emissions, etc.), then the routine F calculates the first-order derivatives: Box Model and Chemical Mechanism 55 dC - = F(C,p,t), (5-11) dt where the function F and the vector p are defined by the mechanism. A derivative dC/dt of a species C, is calculated as dC ~ ~ -' = ""-P.-""-L.+E-D, (5-III) dt j IJ j IJ I I where P,i and L,i stand for the production and loss, respectively, of species i by reaction j, E, are the emissions into the system and D, is a dilution term. If dilution is included in a calculation, then D, is the same for all species except for those, which are not diluted like 0 2 and N2 , for which D, is zero. JAC: Computes the Jacobian matrix, J, which is defined as the partial derivative of the rate expressions to the species concentration (McRae et al., 1982a): J= ()F ac (5-IV) and has the elements: J = _a(~dC_k_/d~t) k,1 ac1 (5-V) Although it is not necessary to provide the solver with an explicit Jacobian matrix, for stiff problems, the solution is highly improved if it is available. FFORC: Computes the forcing functions of all reactions. The forcing functions are the products of the reactants with the rate coefficients. I. e. for a reaction "A + B ~ Products" with the rate coefficient k, the forcing function is FORC = [A]·[B]·k. (5-VI) MECHINFO: Provides information about the mechanism such as the number of species and reactions. SPECIES: Connects the species number to its name. DFDP: Optional, the derivatives 8F/8p are calculated. They can be used to build a set auxiliary equations to calculate sensitivity coefficients as described in more detail in chapter 7. 5.4.2 The Solver, VODE Finding the solution of ordinary differential equations (ODE) is a basic problem in computer modelling. In the early 1970s, one of the more heavily used ODE initial value solvers was the GEAR package (Gear, 1971). The box model presented here uses an improved GEAR routine, VODE (Brown et al., 1989) to integrate the chemical reactions. VODE uses variable coefficient Adams-Moulton and Backward Differentiation Formula (BDF) methods in Nordsieck form, as taken from the older solvers EPISODE and EPISODEB. VODE has a flexible interface that is nearly identical to that of the ODEPACK solver LSODE. 56 Box Model and Chemical Mechanism As long as a similar interface is used, it is easy to replace VODE in SBOX by another solver. For example for the calculation of the sensitivity coefficients, the VODE solver was replaced by a solver called SENDDM, which uses the decoupled direct method (DDM) as described in chapter 7. 5.4.3 The Parameter File While the mechanism code is linked directly into the box model code, many parameters of a run are read from an ASCII file. Special care was taken to make the format of the input file as flexible as possible. Each line consists of a keyword (for example "Temperature") and a value including a unit (for example "298 K"). Comments can be added everywhere to make the file more readable. Parameters controlled by the input file include: Pressure, temperature, start- and end-time of the integration, print intervals as well as relative and absolute tolerance. For species it contains initial concentrations and possibly emissions and dilution. The parameter file may contain a title and additional comments, which are copied as header lines into all output files. The parameters may by split into several files, which are then included by the main parameter file. This allows the re-use of parameters, for example emissions or photolysis rate coefficients in many runs. The photolysis rate coefficients, computed by the photolysis program described in section 5.3, are in a format, which can be directly included into a parameter file. Finally the parameter file controls what output files are written and how they are named. 5.4.4 Output Files The SBOX model may produce a number of different output files: • A report file, which lists all important parameters together with information or possible error messages from the solver. • A table with the concentration of species as a function of time. The unit of the time and concentration can be set in the parameter file. • A table with derivatives of species concentrations, dC/dt, as a function of time. The unit of the time and concentration can be set in the parameter file. • A table with the photolysis rate coefficients as a function of time, interpolated for the output time steps. • A table of forcing functions as a function of time. Forcing functions as defined in section 5.4. l on page 53 are a measure on the mass flux of a reaction. If the time steps are small, the integrated forcing functions are equal to the values of counter species (Leone and Seinfeld, 1985). • A set of species concentrations at the end of a run, in a form which can directly be used to initialise starting concentrations in a new parameter file. Box Model and Chemical Mechanism 57 TABLE 5-1. lnirial Concenrrarions ofModel Species in the PLUME-CASE. Model Species Initial Model Species Initial Concentration Concentration H20 1.0 % CH4 1700 ppb 03 50 ppb H1 500 ppb NO 0.2 ppb Hi02 2ppb N02 0.5 ppb HCHO 1 ppb HN03 0.1 ppb 02 20.9% co 200ppb Ni 78.1 % 5.4.5 Future Development The modular structure of SBOX and the flexible syntax of the parameter file make it easy to extend the SBOX model in the future. Planed extensions are the inclusion of time dependent emission, temperature and pressure profiles. This would allow the simulation of some meteorological effects on the air parcel chemistry. The inclusion of deposition would be helpful for the study of the relative importance of chemistry and deposition on species concentrations. This could require the extension of the model to a 1-dimensional model, so that boundary layer effects and turbulence could be simulated. Other possible extensions in the future include aqueous phase reactions and heterogeneous chemistry. 5.5 The PLUME-Case Most simulations in this study were performed for a moderately polluted scenario of the planetary boundary layer (in this study called PLUME-case) which corresponds to the PLUME/2 case in the model intercomparison of Kuhn et al. (Kuhn et al., 1997) which is based TABLE 5-1. Emissions into the system in the PLUME-CASE. Model Species Emission Model Species Emission 1 1 (ppmmin" ) (ppmmin' ) ALD 3.62E-08 KET 3.l2E-07 co 5.65E-06 OLI l.88E-07 ETE 4.56E-07 OLT 2.19E-07 ETH 2.4IE-07 S02 5.18E-07 HC3 2.91E-06 TOL 5.73E-07 HC5 7.69E-07 XYL 5.l9E-07 HC8 4.55E-07 HCHO l.39E-07 NO 2.59E-06 58 Box Model and Chemical Mechanism on the data of Derwent and Jenkin (1991). To get the starting conditions and emissions of the model species used in this study, as listed in TABLE 5-1 and TABLE 5-2, the emissions of the chemical species were multiplied by the aggregation factors of the RACM mechanism. The PLUME-case was chosen, because it describes a typical moderated polluted case in Europe and is well documented in the literature. 59 6. Modelling 6.1 Nitrogen Chemistry The chemistry of peroxyacy 1 nitrate (PAN) is fundamentally coupled to the chemistry of its precursors, N02 and acylperoxy radical (AC03). This chapter presents a number of computer model studies investigating the chemistry of the atmospheric nitrogen species and of the acylperoxy radical. Please note that in this chapter the term "PAN" stands for a model species that includes all saturated peroxyacyl nitrates and not only for peroxyacetyl nitrate. 6.1.1 Introduction While oxides of nitrogen in the stratosphere participate in an important set of catalytic reactions which convert ozone (03) to molecular oxygen, at lower altitudes they act catalytically to produce ozone by "smog" reactions (Crutzen, 1979). For 1990, Friedrich (1996) reported a total NO, emission in Europe of 18 Mio tons, almost half of which comes from road transport. The fate of all this reactive nitrogen is still not totally clear. It is even possible, that yet unknown species play an important role (Nielsen et al., 1996). The reactive nitrogen cycles affect the 0 3 concentration on local, regional and global scales. Although the main reactions of the nitrogen chemistry are believed to be well known, there are still many questions, which are not fully understood. Understanding the nitrogen chemistry helps to predict the 0 3 and PAN concentrations and therefore may improve the control strategies of these species. While ozone is the best known oxidant component of photochemical smog, it is not the best marker for photochemical activity, because there are other sources such as transport from the stratosphere. PAN or organic nitrates have been proposed as better markers (see for example Penkett and Brice, 1986). This would require a detailed understanding of the formation processes of these compounds. Other questions of interest are for example the role of peroxyacyl nitrates in long distant transport of reactive nitrogen, the importance of the night- time chemistry (Stockwell et al., 1995), or the effect of temperature on the nitrogen chemistry. It is not possible to answer all these questions with a single model study. In the present study the basic behaviour of atmospheric nitrogen species was examined by the use of a zero dimensional box model with the Regional Atmospheric Chemistry Mechanism, RACM, (Stockwell et al., 1997). 6.1.2 Method The newly developed zero dimensional box model (SBOX) and a new version of a chemical mechanism (RACM) were used in the present study. Both are explained in detail in chapter 5 60 Modelling of this thesis. A list of all species and reactions of the chemical mechanism can be found in the Appendix. In this thesis "R###", where "#" is a digit, is used to refer to a RACM model reaction. Simulations were performed for a moderately polluted scenario of the planetary boundary layer. The scenario is described in chapter 5. All calculations were done on a SUN Spare station running SunOS 5.5.1. The rate (forcing function) of the individual reaction was used as a measure of the mass flux through the reaction. All nitrogen species used in the RACM mechanism are listed in TABLE 6-1. Please note that all inorganic nitrogen model species correspond to a single chemical species, while the organic nitrogen species represent a group of lumped chemical species. For this study, the fifth day (96 h < t::;; < 120 h) was chosen for the evaluation to make sure that the system was not affected by the initial conditions. TABLE 6-1. Nitrogen species in the RACM mechanism Inorganic nitr~~~---···- Organic nitrogen ~···------NO Nitric oxide ONIT Organic nitrate N02 Nitrogen dioxide PAN Saturated peroxyacyl nitrates (PANs) N03 Nitrate radical TPAN Unsaturated PANs Ni Os Dinitrogen pentoxide HONO Nitrous acid Organic 11itrogen containi11g peroxy radical_s__ HNO, Nitric acid OLND N03-akene adduct reacting via de- composition HN04 Pernitric acid OLNN N03-alkene adduct reacting to form Carbonitrates + 6.1.3 Results and Discussion Nitrogen Budget While the RACM mechanism does not conserve carbon, it does conserve nitrogen. This means that the stoichiometry of nitrogen in all reactions is balanced. Consequently a nitrogen budget can be calculated to determine the relative importance of the various nitrogen species. In the following section a nitrogen budget is presented and the basic processes of the nitrogen chemistry are discussed. FIGURE 6-1 shows the nitrogen budget calculated for the fifth day of the PLUME-case for a temperature of 298 K. The slope of the increasing total nitrogen corresponds to the emissions of NO into the system. Please note that in FIGURE 6-1 the Y-axes does not start at zero. Everything below 10 ppb is nitric acid, HN03, which has built up during the previous four days. HN03 is a sink for reactive nitrogen species. However the HN03 concentration is probably overestimated in comparison with the real atmosphere because neither wet nor dry deposition was included in the model. Modelling 61 20 -----·---- II ONIT NO N02 .a 18 lil!l -a. ~ N03 a. 11111 -0 ;: a:J 16 • a: fS'j•. cal )( !i 14 c G) al .....0 z 12 10 100 104 108 112 116 120 124 Time (h) FIGURE 6-1: Nitrogen budget on the 5th day of the Plume-case at a temperature of298 K. The N02 photolysis coefficient is shown to illustrate the day-night cycle. The HN03 concentration shows a sharp increase starting about two hours after sunrise. While practically no increase can be observed during sunset, HN03 increases again during the night. This implies that there are two different formation processes, one during the day and one during the night. This observation is supported by field measurements of Bottenheim and Sirois (1996), which suggest that the predominant formation mechanisms of HN03 are different in winter and summer. During the night, dinitrogen pentoxide (N 20 5) and the nitrate radical (N03) appear. These two species are in equilibrium through the fast reactions: (R053) (R054) The equilibrium constant for this pair of reactions is defined to be K = kR05 /kR-0s 4 = [Nz05]/([NO,HN02}}. R054 is the only reaction of Nz0 5 included in the mechanism and this species is therefore a temporary sink of N03• The more N03 is converted to N20 5 the less is left to react with organics. The ratio 62 Modelling [NO,] (6--1) kR053 . N02 + kR054 shows the fraction of NO, that is available to react with organics. The ratio [N03]/([N03]+[N,05]) depends not only on the N02 concentration, but is also highly temperature and pressure dependant. Assuming a N02 mixing ratio of 1.45 ppb during the night, as calculated in the plume case, and a pressure of 1013.25 hPa, the ratio [N03]/([N03]+[N,05]) is 50% for T = 298.15 K, 20% for T = 288.15 Kand 6% at T = 278.15 K. Therefore at lower temperatures the N03 night-time chemistry is reduced because most of the N03 is converted to N20 5. On the other hand, at lower pressure the ratio [N03]/([N03]+[N,05]) rises. For example at 298.15 Kand 700 hPa it is 59%. Note that in FIGURE 6-1 the Y-axis shows mixing ratios of nitrogen-equivalents not species. At sunrise N03 photolyses rapidly and only small concentrations of N03 and N,05 are left during the day. Most of NO, and N,05 is converted to N02 and a N02 peak can be observed shortly after sunrise. In the PLUME-case at 298 K, the concentration of NO, and N20 5 at noon are as low as 0.37 ppt and 0.22 ppt, respectively. The only nitrogen source in the plume-case is the emission of NO. One could therefore expect to find a high fraction of NO in the nitrogen budget. However if there is a high ozone concentration, as in the PLUME-case, the emitted NO reacts rapidly with ozone forming N02 according to: N0+03 N02 + 0 2 (R048) As mentioned in chapter 1, N02 is photolysed during the day: N02 +hv NO +O('P) (ROOl) Because the oxygen atom, O('P), reacts mainly with 0 2 to form ozone, one can assume a photo stationary steady state for the O('P) and consequently the NO concentration can be calculated according to: [NO]= [NO,) kR001 ' (6--11) [0,] kR048 where kRooi is a time dependant photolysis rate coefficient. Using eq 6--11, the diurnal variation of the NO concentration can be calculated approximately. Doing so, the NO concentration is overestimated by about 20% during the day and by about 70% a short time after sunrise. On the other hand, during the night eq 6--11 calculates an NO concentration of zero, because kR001 is zero, while the box model predicts an NO concentration of about 2.7 ppt. All of these calculations show that additional processes influence the NO concentration. The concentration of organic nitrates, ONIT, does not show remarkable diurnal variations. This is somewhat surprising, because the formation and lost reaction rates show a strong diurnal variation. The formation of ONIT is discussed later. PAN and to a smaller extent TPAN build up during the day at the cost of N02. It is interesting that the formation is not proportional to the light intensity, as can be seen from Modelling 63 FIGURE 6-2, which shows the derivatives of the PAN concentration with respect to the time. For 298 K, PAN production starts about two hours after sunrise. It reaches its maximum at about 9 o'clock, and subsequently decreases again. In the afternoon there is a net loss of PAN, which reaches its maximum in the late afternoon. During the night we observe only a small PAN loss. Please note that the behaviour of PAN as observed in FIGURE 6-2 is not a product of a diurnal temperature variation, because all model runs were isothermal. For runs at lower temperatures. the PAN production peak in the morning becomes taller and broader but shows a similar behaviour, except that the PAN loss rate in the evening gets lower and no loss is observed at temperatures below 278 K. 0.4 1---.-.----···~· 35 0.3 ,t, \ 30 N 0 1 l 5 .cc. 0.2 _ ~~11 l 202 CoefficientPhotol;sls ..!:!.: 1 - __}/; ' J 15 ( h" 1) ! ~·~L~100 1 05 11 0 ~~~:115 120 125 Time (h) FIGURE 6-2. Derivative ofPAN (solid lines) during the 5th day of the PLUME-case for a set of temperatures: a: 298 K, b: 288 K, c: 278 K. The NO, photolysis rate coefficient (dashed line) shows the day-night cycle. Sinks for Nitrogen 1 In the PLUME-case there is an NO emission of 0.1554 ppb h· • This is the only nitrogen species emitted. Where does the nitrogen go? To answer this question a least square fit was made for the mixing ratio of all nitrogen species versus the time from the noon of the second day to the noon of the 8 day. The slope found represents an estimate of the overall formation of the nitrogen containing species. The slope of each curve was then ratioed to the NO emissions. These ratios are given in TABLE 6-1. As can be seen from TABLE 6-1, most emitted nitrogen is finally converted into PAN or nitric acid. While at high temperature HN03 is the most important sink for the emitted nitrogen, PAN becomes more and more important when the temperature decreases. At 278 K, more than half of the emitted nitrogen finally becommes PAN. As noted before, the RACM mechanism does not include any heterogeneous chemistry. While the reactivity of HN03 in the gas phase is minor, it should be kept in mind, that in the atmosphere the wet and dry deposition of HN03 may be important. This would decrease the HN03 found in the gas phase. PAN has been found almost everywhere (Roberts, 1990). Unfortunately many reported measurements of nitrogen species do not include temperature information, which can make the interpretation of these measurements somewhat difficult. An extended dataset covering 5 years of measurements of 0 3, PAN, HN03 and particle nitrate including meteorological parameters 64 Modelling was presented by Bottenheim and Sirois (1996). The mean temperature was derived from 5 days backward trajectories. Their data shows a negative correlation of PAN with the temperature, while HN03 shows a positive correlation, which is in agreement with the results presented in TABLE 6-1. The PAN formed during lower wintertime temperatures is expected to decompose in spring when the temperature rises. This may well lead to an increased ozone production in spring. However one must also take into account that hydrocarbons of low reactivity can build up during the winter as well, and reactions of these will have a similar effect. More studies are necessary to explain the ozone spring peaks. Nitrogen Flows To get an overview of all reactions involving nitrogen compounds, the flow from any nitrogen species to any other nitrogen species was calculated. This was done by adding up the forcing functions of all reactions, which convert one species to another. For reactions with several nitrogen products, the forcing function was split according to the product yields. FIGURE 6-3 shows the flow at noon and FIGURE 6-4 shows the flow at night (3 AM) both calculated for the PLUME-case at 298 K. The importance of many reactions already discussed in section 6.1.3 was confirmed by this analysis. During the day most nitrogen cycles between NO and N02• The reaction R048 (NO + 0 3 ~ ) N02 + 0 2 accounts for 83% of the NO-to-N02 conversion. Other reactions of importance for this conversion are the reaction of NO with H0 2 , methyl peroxy (M02) and acylperoxy (AC03). Another cycle goes from N02 to pernitric acid (HN04). The only reaction known to form HN04 is the reaction of N02 with HOi- Because the thermal decomposition of HN04 back to reactants is fast at 298 K, no HN04 can build up. The lifetime of HN04 at 298 K with respect to thermal decomposition is only about 12 s, much shorter than that for photolysis or reaction with HO. However, at lower temperatures and pressures, the lifetime of HN04 is significantly longer and these reactions become more important. TABLE 6-1: Importance of species as sink for the emitted nitrogen Species' Proportion of emitted nitrogen b 278 K 288K 298K HN03 40.6±0.5 % 73.8±0.7 % 91.7±0.5 % PAN 54.4±0.7 % 20.6±0.5 % 4.0±0.4 % ONIT 3.2±0.1 % 2.7±0.1 % 2.3±0.1 % N02 1.0±1.0 % 2.0±1.0 % J.5±0.7 % 2 · N20 5 0.8±0.6 % 0.9±0.5 % 0.3±0.3 % • Species not listed had slopes indistinguishable from zero. ' Increase of nitrogen species divided by the total emitted nitrogen (emitted as NO). See text for description. The error is 2 times the standard deviation. Modelling 65 A similar cycle can be observed from N02 to PAN. As was discussed earlier in this thesis, the thermal decomposition of PAN is very temperature dependent. While at 298 K the thermal decomposition is 98.6 % of the formation of PAN, it drops to 91.2 % at 288 K. The reaction of PAN with HO leading to NO 3 is of minor importance, as will be discussed in more detail later. TPAN shows a similar behaviour than PAN except that the HO reaction is relatively more important. The last fast cycle goes from N02 to NO,. In contrast to PAN and HN04, N03 does not thermally decompose, but photolyses rapidly during the day. The photolysis accounts for 76% of the N03 to N02 conversion. Additionally N03 can react with NO to form two N02 molecules, which accounts for 23% of the N03-to-N02 conversion during the day. During the night, the situation looks somewhat different, as can be seen from FIGURE 6-4. Because N03 does not photolyse, it can build up and undergo a large number of reactions with inorganic species, organic radicals as well as stable organic species such as aldehydes or alkenes (see for example Cantrell et al., 1984; Finlayson-Pitts and Pitts, 1986; Canosa-Mas et al., 1996). One important reaction discussed earlier, which can be clearly seen from FIGURE 6-4 is the formation of N20 5 from the combination of N02 and NO,. An important day-time formation reaction of organic nitrates (ONIT) is the reaction of organic peroxides with NO. This reaction has a second reaction path, resulting in N02 and an alkoxy radical: a· (RO·+ N02) +(I-a)· RON02, where a is the branching coefficient. Because one branch of this reaction converts an NO to an N02, which consequently increases the ozone, while the other branch produces an organic nitrate, organic nitrates were proposed as marker for photochemical smog formation. Unfortunally the analysis is more complicated because the reaction of peroxy radicals with NO is not the only formation reaction for organic nitrates. During the 5th day of the PLUME-case, 69% of the organic nitrates are formed by reactions of organic peroxides with NO. Other important formation reactions include: N03 reactions (12% ), the reaction of phenoxy radicals (PHO) with N02 (9%) and the reaction of N03-alkene adducts with H02 (8%). This means that the night-time chemistry plays an important part in the formation of organic nitrates and this has to be taken into account, when using organic nitrates as a marker. 66 Modelling ~::;:::.._---- 0.1 FIGURE 6-3. Nitrogen flow at noon (t = 108 h) for the PLUME-case at 298 K, photolysis rate coefficients for June 21. The numbers show the flow in units of pptlh. The flow was calculated by adding the forcing functions of all reactions that convert one species to the other. For graphical reasons, the thickness of the arrows is proportional to the cubic root of the flow. Modelling 67 0.37 (OLN~ ~ (ONIT) ~(OLND) /"601 2.5 Emissions 155.4 ~ 1.1E-3 FIGURE 6-4. Nitrogen flow by night (3 AM, t = 123 h) for the PLUME-case at 298 K, photolysis rate coefficients for June 21. The numbers show the flow in units of pptlh. Flows were calculated and the arrow thicknesses were determined as noted in the caption for FIGURE 6-3. 68 Modelling 6.2 Chemistry of the Acylperoxy Radical (AC03) 6.2.1 Introduction Organic peroxy radical (R02) reactions have received a great deal of attention during the last decade because they represent an important class of intermediates formed in the oxidation process of hydrocarbons (Lightfoot et al., 1992; Wallington et al., 1992; Lesclaux, 1997). In particular, the acetylperoxy radical, CH3C(0)02, is known to be formed as a consequence of the photooxidation of carbonyl compounds such as acetaldehyde, acetone, methylvinylketone, methylglyoxal, etc. As shown in the introduction of this thesis, CH3C(0)02 is a significant contributor to smog formation as a precursor to peroxyacetyl nitrate. To understand the formation of PAN it is important to know the source of the precursor acylperoxy radicals (AC03). Using model calculations, Kirchner and Stockwell (19%) showed the importance of the peroxy radical chemistry for the calculated concentrations of PAN, higher organic hydroperoxides, and peroxyacetic acid. Furthermore, Kirchner and Stockwell as well as Canosa-Mas et al. (1996) reported that under some conditions the reaction of the acetylperoxy radical with N03 can be important in the night time chemistry. Recently Villenave et al. (1996; submitted) measured new rate coefficients for the reaction of the acetylperoxy radical with other peroxy radicals. Their rate coefficient value is a factor of 2 to 4 larger than those used previously in most models of tropospheric chemistry for acylperoxy (AC03) radical cross reactions. In the present study it was estimated to which extent the results of model calculations are affected when their new value is taken into account. Additionally the relative importance of the possible reactions of AC03 was examined for a set of different conditions. 6.2.2 Method The RACM mechanism was used in a zero-dimensional box model as described earlier. To determine the influence of the VOC and NOx emissions on the chemistry some runs were made where the starting concentrations and emissions of VOC or NOx or of both were reduced compared to the PLUME case. In the LOW NOx case the NOx was reduced to 10%, in the LOW voe case, the VOC was reduced to 10% and in the LOW voe & NOx case both NOx as well as VOC were reduced to 10% and the initial CO concentration was set to I 04 ppb which is close to the atmospheric background value. The initial 0 2, N2, H2 , CH4 and H20 concentration were identical in all four cases. To investigate the effects of the recommendations of Villenave et al. (1996; submitted), the rate coefficients of all reactions AC03 +higher peroxy radicals (reactions R l 92-R204, R207- R209 and R234 of the RACM mechanism) were set to 10·11 cm3 molecule·1 s·1 for two runs corresponding to the recommendations of Villenave et al. Important for the present study is, that RACM includes reactions of the acylperoxy radical with peroxy radicals as well as acylperoxy self reactions, as discussed in detail by Kirchner and Stockwell (1996 ). Modelling 69 6.2.3 Results and Discussion Formation of AC03 It is well known, that the reaction of aldehyde with HO is an important source of acylperoxy radicals, AC03 (Finlayson-Pitts and Pitts, 1986). However, there are other possible source reactions and it is therefore interesting to look at the relative importance of these reactions. To identify the main processes producing AC03, the reactions leading to AC03, were grouped in 9 categories. The AC03 formation rate for each category was then calculated by adding the product of AC03 yield times forcing function for all reactions of a group. 800 HO+ALD :::"" 700 :c HO+ MGLY a. 600 HO + PAAfTPAN -a. Ketone Photolysis -c 500 NO + TC03/KETP 0 0 3 +Alkene ;:: 400 Ill M02 + TC03/KETP E N03 + TC03/KETP/TPAN ..0 300 u.. N03 + ALD/MGLY .., 200 0 (.) < 100 0 100 104 108 112 116 120 124 Time (h) FIGURE 6-5. lmportance of the ACO,formation reactions for the PLUME case at 298 K, photolysis rates for June 21. Noon is at t = 108 h, midnight at t = I20h. FIGURE 6-5 shows the diurnal variation of the AC03 formation calculated for the PLUME- case at a temperature of 298 Kand with photolysis rates of June 21. The AC03 coming from the thermal decomposition of PAN was omitted from the evaluation, because at ambient temperature there is an equilibrium between AC03 and PAN. Most AC03 produced by the thermal decomposition of PAN rapidly reacts back to reform PAN. TABLE 6-2 shows the relative importance of the categories for both summer and winter photolysis rate coefficients. The reaction of aldehyde (ALD) with HO accounts for about 60% of the AC03 formation. If the production from the HO reaction with methylglyoxal (MGL Y) and peroxyacetic acid (PAA) is added, HO reactions account for a total of 72% of the AC03 production in summer and for 53% with the winter photolysis rate coefficients. It is interesting that the relative importance of the photolysis of ketones, on daily average, is almost independent of the photolysis rate coefficients. During the night the reactions of N03 with ALD and MGLY are the dominant process. The relative importance of the night-time chemistry increases with the winter photolysis rate coefficients, because of the reduced actinic flux. 70 Modelling TABLE 6-2. Relative importance of the ACO,formation reactions Summer (June 21) Winter (Dec. 21) ___ ~-- Time• __! N03 + ALD/MGL Y 0.0% 65.0% 4.5% 0.0% 66.7% 17.9% NO,+ TCO,IKETP/TPAN 0.0% 4.6% 0.4% 0.0% 5.8% 1.8% NO + TCO/KETP 8.1% 0.0% 8.5% 16.6% 0.1% 11.4% M02 + TCOiKETP 0.4% 0.2% 0.3% 0.1% 0.3% 0.1% + Alkene 1.1% 2.4% 1.5% 1.8% 5.6% 3.8% Total flow ppt!h 751 57 267 492 68 141 ' Noon: t- 108 h, Midnight t = 120 h. Avg. day: Average 96 h :5 t Sinks of AC03 When examining the loss reactions of AC03, one has to take into account, that at ambient temperature, there is an equilibrium between the acylperoxy radicals (AC03) and the peroxyacyl nitrntes (PAN): AC03 +N02 PAN (Rl27) PAN ACO,+N02 (Rl28) When looking at the importance of the sink reaction for the acylperoxy radicals, the reaction with N02 would in most cases be the dominant process. However this reaction is not a real loss process for AC03 because most of the AC03 radicals are recycled through reaction R128. Therefore the reactions R 127 and R 128 were neglected, but the reaction of PAN with HO was included, which indirectly also destroys an AC03 radical. FIGURE 6-6 shows the diurnal variation of the important AC03 destruction reactions. Remarkable is the change from the daytime to the night-time regime, and vice versa at sunrise and sunset. During the day, the reaction with NO is the dominant loss process for AC03 (77 %), while the reaction with H02 accounts still for 12 % of the loss. The methylperoxy , , CH30 2 and higher peroxy radicals, R02 sum up to almost 7 % of the reactions, followed by the AC03 self reaction. The reaction with N03 is neglectable during the day. However, during the night the reaction with N03 accounts for 85 % of the AC03 mass flux, and only 0.6 % of the ACO, radicals react through NO. The overall mass flux of the AC03 loss processes at midnight is 11 % of the one at noon. Modelling 71 8E-04-.-A-C_0_3_+_R_02----~.."""'"------(A-C-0-3-A-C0--3-)-2-----~-.0 ,,--,, + ~ AC03+M02 II ,/ '\ AC03+N03 D 30 ..... AC03+H02 .::<'6E-04 II/ \ AC03+NO 8 i=. :c I ' ... I I ' \ c I \ " I>. .. = ,' \ N02 Photolysis 2 0 ;. ~ I I ~ 4E-04 I \ ..... I \ r-. I \ ~"' I I ·:;;"' c ,' ,, '.1'.- ... ,, \, \,',. ',, \,',, '.,,'· \ ... .~ I ,,',,',,',,',,',,\,',,'.,."",,',,',, \ IO 0 ~ 2E-04 0 ',' ,, >>>>>>>>>>:<<,, ,, ,, "',,,,,,,,,,"',,,,,, \' .:: .. i:>. ~ ,' ...... >~ ... >>>>>>~< ... >>~·'\ \ <<<<<<<<<'>~<<<<\, I ,' "'',. ',,. ',, './'./'./'./",,',, './'.1 './',,'./',, ',,' \ O' OE+OO....,:w<·~:-:~-:·.:-:~-:·.:-:~-:·~· T'·,'~,'-,~',~'-,'~,-',~'-,'~,-',~'-,'~,~',·'~,'·,~',·'~,'·,~',..,,...... ,....,...... ,,_ 0 z 105 llO 120 TIME (h) FIGURE 6-6. Fraction of AC03 reaction rates during the 5th day of the PLUME case at 298 K with JOO% emissions. The dotted line shows the NO, photolysis frequencies and gives a measure of the day-night cycle. When integrated over 24 hours, the Ae03 + N03 reaction accounts still for 8 % of the Ae03 mass flux. TABLE 6-3 shows the relative importance of each channel for the loss of Ae03 radicals during the 5th day. When going from 298 K to 288 K, the picture does not change much as shown in column 3 of TABLE 6-3. The importance of the Ae03 + N03 reaction decreases, which can be explained by the lower N03 concentrations found in this scenario. When both, the NO, and the voe, are reduced, the relative importance of the reaction of Ae03 with NO does not change significantly but the reaction with H02 becomes more important. Interesting is that the contribution of the methylperoxy reaction, eH30 2 + Ae03 , almost doubles, whereas the importance of the reaction with higher peroxy radicals decreases to half the value of the more polluted case. The importance of the night-time chemistry decreases by a factor of 4. In the case with high voe and low NO, emissions, the Ae03 + NO reaction is only responsible for about half of the Ae03 loss. The Ae03 self reaction accounts for 9 % of the loss, while the methylperoxy and higher peroxy radicals account for 16 %, which is about the same as H02 • In the opposite case, with low voe and high NO, emissions, the reaction of Ae03 with peroxy radicals and its self reaction are not important. During the day about 90 % of the Ae03 , reacts with NO and 10 % with H02 while during the night almost I 00% reacts with N03• All the above runs were made with the Ae03 rate coefficient of the RAeM mechanism. The rate coefficients of the reactions (Ae03 + peroxy radicals) in this mechanism are calculated 72 Modelling from the self reactions of these radicals as explained by Kirchner and Stockwell (1996). These rate coefficients were in general 2 to 4 times smaller, than the values suggested by Villenave et 11 3 1 1 al. (1996; submitted). After setting all coefficients to 1.0 x 10· cm molecu1e· s· , the relative importance of the higher peroxy radicals rises from 2.4 % to 4.6 %, and the importance of the AC03 + NO reaction decreases by about 2 % for the case with full emissions. This change has no significant influence on the 0 3 concentration after 5 days, but decreases the PAN concentration by about 2 % after the same period. The absolute mass flow of the reaction AC03 +higher peroxy radicals almost doubles. In the LOW NO, case, the relative importance of the reaction of AC03 with higher peroxy radicals rises from 7.8 to 13 % when the rate coefficients are updated according to the new measurements. Including CHp2, the peroxy radicals account for more than 20 % of the ACO, loss reactions. This has no influence on the 0 3 concentration, but the PAN concentrations decrease more than 4 % compared with the run using the old rate coefficients. The reactions of AC03 with organic nitrate radicals and the reaction of PAN with HO are not important in any of these cases. In order to quantify the contribution of peroxy radicals cross reactions to the total loss of AC03 radicals, Calculations were performed for the PLUME-case and variations thereof. As expected, the contribution is of minor importance or negligible for "high NO," situations, the AC03 +NO reaction remaining the principal fate of AC03 radicals. In contrast, at "low NO," levels, AC03 cross reactions with peroxy radicals become quite significant since they can TABLE 6-3. Fraction ofAC0 3 sinks in the PLUME-Case and Variations Low Updated Rate Temp. Coefficientsa CASE PLUME PLUME LOW LOW LOW PLUME LOW NO,&VOC NO, voe NO, NO Emissions 100% 100% 10% 10% 100% 100% 10% VOC Emissions 100% 100% 10% 100% 10% 100% 10% 298K 288K 298K 298K 298K 298K 298K AC03 +NO 73.1% 75.2% 71.2% 55.8% 80.5% 71.5% 53.0% AC03 +H02 9.4% 9.5% 16.9% 17.2% 7.6% 9.2% 16.3% AC03 +M02 3.4% 3.0% 6.6% 8.3% 1.9% 3.3% 7.8% 2·(AC03+AC03) 3.3% 3.3% 1.8% 9.1% 0.4% 3.2% 8.2% AC03 + Higher 2.4% 2.6% 1.2% 7.8% 0.3% 4.6% 13.0% peroxy radicals PAN+HO 0.2% 0.6% 0.1% 0.0% 0.2% 0.2% 0.0% AC03 +N03 8.1% 5.5% 2.2% 1.2% 9.1% 8.0% 1.1% AC03 +0LN 0.1% 0.2% 0.1% 0.5% 0.0% 0.2% 0.6% 'rate coefficients of all reactions ACO, +higher peroxides set to 10'11 molecules·' cm' s·' Modelling 73 account for more than 20 % of the total AC03 loss. This shows that this class of reactions plays a significant role in the tropospheric chemistry, under "low NO," levels. It is likely that this role is even more important in very clean remote atmospheres such as the marine boundary layer. The direct consequence of increasing the cross reaction rates for AC03 radical reactions is to decrease of PAN production. However, the PAN concentration is moderately reduced: 4% in the "low NO," level situation, but again, this reduction could be larger in very clean atmospheres. It must be pointed out that the ozone formation is not significantly affected by increasing the efficiency of the AC03 cross reactions. 6.3 PAN/03 Ratio 6.3.1 Introduction The PAN/03 ratio of polluted air usually shows an annual and diurnal variation. This ratio has received much attention recently, because it is expected to provide some information on the history of air masses (see for example Shepson et al., 1992; Kourtidis et al., 1993; Hartsell et al., 1994; Bottenheim and Sirois, 1996). It is not clear yet, whether the observed nocturnal decay of PAN is due to the chemistry (Stockwell et al., 1995) or a rapid rate of dry deposition as suggested by Shepson et al. (1992). Deposition would mean that the PAN reacts on plant surfaces, and a high deposition rate for PAN implies lower ozone production rates because of the loss of reactive nitrogen. However one aspect often not fully considered is, that the O,JPAN ratio is temperature sensitive and that the temperature shows a diurnal variation as well. In this study the diurnal variation of the O/PAN ratio was compared to the influence of temperature. 6.3.2 Method The RACM mechanism was used in a zero-dimensional box model as described earlier. The PLUME-case scenario with photolysis frequencies for June 21 was used. 6.3.3 Results and Discussion A number of isothermal runs were made for various temperatures in the range 263 K - 313 K. For every run the PAN/03 ratio was determined for all calculated data points of the fifth day. FIGURE 6-7 shows the lowest and highest ratios for each temperature together with the average value. The average ratio PAN/03 for a number of U.S. cities and polluted air masses in the Netherlands was reported as 0.07 with a wide range of variability (Shepson et al., 1992). This is lower than the model prediction for ambient temperature. 74 Modelling 0.1 0.08 0.04 0.02 0 260 270 280 290 300 310 320 Temperature (K) FIGURE 6-7.Temperature dependence of the [PAN]l[O,] ratio. Shown is the range and the mean value of the ratio for the fifth day (96 h 5 t < 120 h) of the PLUME-case scenario with photolysis rate coefficients for June 21. At ambient temperatures the diurnal change in the P AN/03 ratio is relatively small compared to the temperature sensitivity. Therefore a change of the temperature during the day may have a greater influence on the ratio than the variation of the actinic flux. This suggests that the knowledge of the temperature history of an air parcel is necessary to understand its PAN/03 ratio. 6.4 What Happens when the Light Goes Off? 6A.1 Introduction The day night cycle has a big influence on the atmospheric chemistry of the boundary layer. Not only do the photolysis rate coefficients change, but also the temperature varies. An understanding of these factors may help to understand observed PAN and ozone values and ratios. Not only does the zenith angle of the sun influence actinic flux, but also clouds may cause rapid changes in the photolysis rate coefficients. To investigate the effect of light and temperature change on PAN and 0 3 , a step-functions for the temperature and the actinic flux were used. 6.4.2 Method In the following study the PLUME case scenario was used in a zero dimensional box model. A run was started for a temperature of 298 K and constant photolysis rate coefficients corresponding to an 24 h average of 21 June. At one point in time (t = 48 h) the photolysis rate coefficients were set to zero or the temperature to 288 K or both. Modelling 75 6.4.3 Results and Discussion FIGURE 6-8 shows the step response of the PAN (A) and ozone (B) concentration as well as of the PAN/03 ratio (C). While the light is on, the owne concentration is increasing. When the light goes off, it starts to decay constantly mainly due the reactions of ozone with NO, (58%), H02 (20%) and NO (17%). If at the same time the temperature decreases, the ozone loss is 35 % smaller, which 30 40 50 60 70 80 90 0.26 0.24 i 0.22 o'" 0.2 0.18 ,_ ...... ,,,, ...... 0.16 0.008 ~····~~~~~~~~~-·~~~~~·~~~ 0.007 <-··················"· - 0.006 E ~ 0.005 ,_ ...... , ...... ~ 0.004 >-····· ...... , ...... 0.003 }-······· .. 0.002 i:...... ~cc ...... , ...... , ······.·············• 0.001 0.04 ~~-~~~~~~.,.....,.-,-,~--~~-,-,~~~...., 0.035 0.03 '2M 0.025 :z ~ 0.02 0.015 0.01 0.005 ~ .... ~~~~~~···~-~~ ...... ~~~~~~.. ~~~ 30 40 50 60 70 80 90 TIME (h) FIGURE 6-8. Effect on PAN and O, when switching off photolysis or decreasing the temperature. Starting with 298K and constant light, at t = 48 h, the light was switched off, or the temperature decreased suddenly or both. 76 Modelling can be explained by the temperature dependence of the 0 3 sink reactions. If the temperature decreases to 288 K while the light is still on, the 0 3 formation (slope of the curve plotted) drops within two hours to about 25% compared to the value for 298 K PAN shows an opposite temperature behaviour than ozone. The PAN formation rate, d[PAN]/dt, increases immediately by a factor of 5 as the temperature decreases from 298 K to 288 K (see FIGURE 6-9). As time goes on, the formation rate decreases again to reach a constant value, which is twice as high as the 298 K value, after about 15 hours. The response of the PAN concentration is interesting when photolysis is turned off. Then PAN concentration first increases for about two hours at a high rate and starts then to decrease again. A similar behaviour of the PAN formation can be observed in FIGURE 6-2 during the sunset. In the case of the combined temperature and light change, the PAN concentration shows a fast increase for about 5 hours and then keeps about constant. The sensitivity of PAN towards the temperature and light change is about an order of magnitude larger than the sensitivity of ozone. Therefore, the PAN/03 ratio shows a similar characteristic as the PAN concentration. While the step response of ozone is relatively fast, for PAN it may take several hours to reach an equilibrium (see FIGURE 6-9). Additional studies are necessary to fully explain the behaviour of PAN. 10 8 6 .2 .2 ... 4 ... I!: 2 I!: 0 -2 0 5 10 15 20 25 30 35 TIME (h) FIGURE 6-9. Relative change of the PAN formation, d[PAN]ldt, when switching off photolysis (left hand side) or decreasing the temperature (right hand side). 6.5 Conclusions The newly developed box model, SBOX, combined with the Regional Atmospheric Chemistry Mechanism (RACM) (Stockwell et al., 1997) as used to investigate the PAN and nitrogen chemistry of the troposphere. It has been shown that SBOX is a flexible tool, which can provide much information on a mechanism. By means of a nitrogen budget the diurnal variation of the reactive nitrogen was studied and some basic relationships were found. HN03 and PAN were identified as the main sinks for emitted nitrogen and the temperature dependence of this processes was stressed. An overview over the nitrogen flows in the RACM mechanism was presented and the important cycles in the nitrogen chemistry were determined. Modelling 77 In the PLUME-case at 298 K, the reaction of aldehyde with HO accounts for about 60% of the produced acylperoxy radical. Other reactions of importance are the photolysis of ketones, the reaction of HO with methylglyoxal, and at night the reaction of N03 with aldehydes or methylglyoxal. Although the reaction with NO is the dominant process for the loss of acylperoxy radicals, the reactions of AC03 with organic peroxy radicals can not be neglected. In the case of low NO, and high VOC, the reactions of AC03 with higher peroxy radicals account for 8 % of the loss reactions when using the rate coefficients of RACM. This value increases to 13 % when the rate coefficients used are those recommended in the paper of Villenave et al. (1996; submitted) and to more than 20 % when CH,02 is included. These changes are reflected in the PAN concentrations, which are decreased by 4 %. It was shown that the ratio, O/PAN, is very temperature sensitive. This should be considered when field measurements are interpreted. The high sensitivity of the ratio on the temperature means also, that the observed diurnal changes of the ratio can be a product of the chemistry alone and are not necessary a indicator of deposition The response of the PAN and ozone concentration on abrupt changes in the temperature or actinic flux were studied and an interesting behaviour of the PAN response was found. Additional studies are necessary to fully understand these observations. Most of these studies showed the high temperature sensitivity of the PAN and nitrogen chemistry. The influence of the temperature on PAN may be an order of magnitude higher than on ozone. To understand observed annual and diurnal variations of the nitrogen species, the temperature variation needs to be taken into account. For this purpose, it would be useful, to extend the box model, SBOX, to allow calculations for diurnal temperature profiles. 79 7. Sensitivity Study 7.1 Introduction In many branches of science and engineering, computer models are used to describe the dynamics of complex systems. One direct consequence of using models is that approximations involved in the model and uncertainties in the input parameters lead to inevitable uncertainties in the predictions. To improve a model and to estimate its reliability, it is important to identify those parameters that have the most impact on system performance. The goal of any sensitivity analysis is to determine the change of the system output that results from the variations in the model input parameters. Estimates of the contribution of individual parameters to the overall uncertainty in gas phase chemical mechanisms can provide guidance for revising or developing mechanisms and gives hints which laboratory measurements are important. Additionally a sensitivity analysis can give valuable information which can help to understand the complex interactions of species and reactions in a chemical mechanism. Traditionally, the analysis of the sensitivity of models to small perturbations in parameters is called local sensitivity analysis. Methods for local sensitivity analysis are discussed below. A global sensitivity analysis is a procedure where the sensitivity of the solution to large variations of parameters is determined. Typically global sensitivity is combined in an appropriate manner with a measure of the actual degree of uncertainty in the parameter's values. Commonly used methods include Monte Carlo analysis and the Fourier amplitude sensitivity test (FAST). In Monte Carlo analysis, repeated model runs are made, each drawing a value at random from the probability distribution function of each parameter (Gao, 1995). The probability distribution functions for the output variables are then estimated based on the results of these runs. Stratified sampling techniques such as Latin Hypercube Sampling (LHS) improve the Monte Carlo techniques. The FAST technique associates each uncertain parameter with a specific frequency in the Fourier transform space of the system. The system sensitivities are then determined by solving the system equations for discrete values of the Fourier transform variable and then computing the Fourier coefficients associated with each parameter frequency (Cukier et al., 1978; McRae et al., l 982b ). Several studies considered sensitivity in mechanisms for examining the factors that contribute to discrepancies between mechanisms (see for example Milford et al., 1992). Russell et al. (1995) describe a ozone control strategy based on VOC reactivity weighting whereas Yang et al. (1995; 1996) used sensitivity analysis to estimate uncertainties in calculated incremental reactivities of organic compounds. Gao et al. (1995) applied sensitivity and uncertainty analysis methods to the RADM2 mechanism to evaluate the sensitivity of 5 species towards a large number of rate coefficients and stoichiometric coefficients. Most of 80 Sensitivity Study these studies used constant photolysis rates to facilitated interpretation of the sensitivities. The present study presents an approach to calculate the local sensitivity with varying photolysis frequencies. Using this method, the local sensitivity of 0 3 and PAN in the new RACM mechanism was investigated at various temperatures and conditions. 7.2 Method 7.2.1 Sensitivity Coefficients Chemical rate equations can be written in the compact form d~~t) = F(C(t), t;p(t) ]; C(to) C0, (7-I) where C(I) = [C" .. ., Cnl is a vector of species concentrations at time t, p(t) = [p, (t), ... , Pm(t)J is a vector of system parameters and F[C(t), t; p(t)] is an-component vector [F,(p,C), .. ., F0 (p,C)], which couples the equations together and is generally a non-linear function of C. C0 are the initial values of C at time 1,i. The sensitivity of the model can be defined as the change in system output .6.C that results from a parameter variation ti.p away from some nominal value p i.e. ti.C = C(p + .6.p) - C(p). (7-II) Therefore a matrix s(t) of linear sensitivity coefficients can be defined as s,;(t) (7-III) A [CJ ..: p p FIGURE 7-1. A) Definition of a linear sensitivity coefficient describing the sensitivity of a species concentration [CJ towards a parameter p. The linear sensitivity is the derivative iJ[C]ldp around the nominal value of a model parameter. B) Hypothetical response of a predicted concentration [C} to uncertainties in model parameters p1 and p2 (after Stockwell and Milford, 1995 ). Sensitivity Study 81 The possible effect of varying a single parameter is illustrated schematically in FIGURE 7-1. Please note that local sensitivity coefficients ClC/0pi provide direct information on the effect of a small variation in each parameter about its nominal value, pi' on each species concentrations, they do not indicate the effect of simultaneous, large variations in all parameters on the species concentrations. 72.:1. The Decoupled Direct Method (DOM) Several approaches have been developed to calculate local sensitivity coefficients. The simplest of these is the indirect method, in which the input parameters are varied one by one and the model equations solved a new for each new set of values of the input parameters. The variational method presented by Seigneur et al. (1982) introduces an objective function which permits one to obtain the sensitivity of several species simultaneously in a single analysis. In the direct method the sensitivity coefficients are calculated from an auxiliary set of equations derived from the model equations 7-1 by differentiating with respect to parameter pi (Gao, 1995): ds.(t) _J- = J[C(t),t;p(t) ]· si(t) + FJC(t),t;p(t)] (7-IV) dt _ClC(t 0 ) sj ( to ) - a Pi where: s --·ac with respect to pi" j - Clpj . sensitivity coefficient F=~: accounts for any explicit dependence F on pi. J Clpj J = ClF: ac Jacobian matrix. If pi is an element of the initial concentration vector C(ta), then Fi is zero and the last term in the right-hand side of equation 7-IV disappears. If pi is an element of the rate coefficients, the initial value of si reduces to zero. Solving the coupled system of model equations 7-1 and auxiliary equations 7-IV has been found to be unstable for stiff problems and it is also quite inefficient. The Green's function method with the analytically integrated Magnus modification (GFM/ AIM) was developed to overcome the difficulties connected with the direct method (Kramer et al., 1981). The GFM calculates Green's function for the auxiliary equations used in the direct method and then obtains the sensitivity coefficients from integrals over Green's function. An simpler method which has proved more efficient and stable with equal or better accuracy than GFM is the decoupled direct method (DDM) (Dunker, 1984). The DDM solves the auxiliary equations separately from the model equations but uses exactly the same time steps and numerical approximations. Dunker (1984) showed that this method has advantages in simplicity, 82 Sensitivity Study stability, accuracy, efficiency and storage requirements. The DDM code used for this study, which uses the method of Gear (1971) to solve the equations, was initially developed by Mccroskey and McRae (1987) using the LSODE code and was extended by Gao (1995). For the sensitivity calculations a version of the SBOX model was created where the VODE solver was replaced by the DDM solver. Additionally to the subroutines 'F' and 'JAC' (see section 5.4), an subroutine 'DFDP' had to be supplied, which calculates the vector oF/dpi for each parameter PJ of interest. The chemical compiler was extended to generate the code for this subroutine. 7.2.3 Time Dependant Rate Coefficients In this study the sensitivity of species towards rate coefficients was determined. Rate coefficients may be time dependent, either because of changing temperature, or because of changing light intensity in the case of photolysis reactions. This can make interpretation of eq 7-III difficult if PJ is a rate coefficient. To prevent this problem, the rate coefficients were assumed to be a product of a, possibly time dependent, term without uncertainty and an uncertain term equal unity: ki(t) - k*i(t) ·pi, (7-V) where kJ(t)=k'(t); pi l; o(k')(t)=O; o(k) is the uncertainty of k and k is its mean value. Assuming that the relative uncertainty, a(ki)(t)/ki(t), is not time dependant, it is equal to the uncertainty of pi: o(pi) = o(ki)(t)/ ki(t). (7-Vl) For this reason and for purposes of comparing sensitivities across parameters, normalised coefficients, (7-VII) are used in this study. 7.2.4 Sensitivity of Ratios Many rate coefficients are not measured as absolute values, but relative to some other coefficients (see experimental part of this thesis). In these cases it is often useful to look at the sensitivity to ratios of rate coefficients. System parameters as defined in eq 7-II are not limited to rate coefficients. If we have two reactions with the rate coefficients k1 and k2 respectively, where kz was measured relative to k1, we can replace k1 and kz with the parameters K and a, k 1 = K·Jfi. (7-VIII) k2 K/ Jfi. Sensitivity Study 83 and consequently a is the ratio, k, /k2 , and K: is the geometrical mean .Jk1 x k 2 • When assuming that these coefficients are a product of a term without uncertainty and a uncertain term equal unity, p, one gets (7-IX) which can be simplified to get P, = (7-X) For a small variation, Ap, we get p, + Ap, "'.J{Pi + iip 1)-(p2 + Ap2) and (7-XI) Pa +Ap" = p, +Ap, P2 +AP2 Because all p are unity, if all Ap << I, one gets Ap, "' Ap1 + Ap 2 and Apa "' Ap 1 Ap 2 • (7-XII) For small disturbances, the change in system output, AC, is linear to the disturbances: (7-XIII) wheres* is a normalised sensitivity coefficient. Using eqs 7-XII and 7-XIII one can calculate the new sensitivity coefficients: and s: from the original coefficients; and s;: s; +s; (7-XIV) 2 2 7.3 Results and Discussion 7.3.1 General Checks of the DOM To insure the implementation of the DDM code was working, the calculated species concentrations were compared to the concentrations calculated using the VODE solver for a few runs. The results of both solvers were the same within 4 significant digits. To make sure the sensitivity calculations were working, the sensitivity coefficients from DDM were compared to sensitivity coefficients calculated by the indirect method. For these, two runs were made using the VODE solver. In these runs the rate coefficient of reaction Rl45 (AC03+NO) was increased respectively and decreased by one percent. The sensitivity was then calculated from the ratio of the differences in the 0 3 yield and the differences in the rate coefficient: 84 Sensitivity Study kR145 d(03] ,,,, kRJ45 ([03]+1% -[03]_1%) (7-XV) [03] dkRJ45 (03] (kRl45•l% -kRl45-1%) With constant as well as with varying photolysis coefficients the sensitivity calculated with the indirect method were the same as the sensitivities calculated with DDM within three significant digits. This shows that the DDM code can be used to calculate sensitivities for systems with varying photolysis coefficients. 7.3.2 Diurnal Varying Photolysis Coefficients To examine the effects of varying photolysis coefficients, a sensitivity analysis of the R001: N02 + h v R061: CH4 + HO 7% ,-,-~F'C"'F"""'~,--·-,--·:·--, Q. % 6% 1--··'···k'-····.; ...... ;...... ;::-"""''=·····'- --1 '.!:! 30% ;;' ;:::; 5o/o !-·····'···•···'-····• ...... ;...... ~ 20% 2. 4% "Cl 3% x x 10% 1-·ff:...... i ...... : ...... ;... : 2% ;;' 1% f-...... ·i ...... ·~ QO/o §- 0% 20 40 60 80 100 120 140 100 Time (h) R127: AC03 + N02 R128: PAN decomposition 5%------~ x x ;;' -3% ;;- g. -4% 1-...... ;...... :...... t/··""'+~-i/·\·~•··J..1""'-J-< g. 0% I-··-"+ ...... ;.,, ....: ...... •, ...... ;...... , ...... ;..... -J -1 % '-----'--'----'-~'----'---'----'----' - ~ 1001M1~100 20 40 60 80 100 120 140 160 0 M 00 00 Time (h) Time (h) R091: HCHO + N03 R049: 03 + N02 0.2% ~--~·---·-----~~ % 0% ~-0.2% 2.-0.4% "Cl_0.6% x -0.8%. ~-1~~~: 1-·····•·········'·····--•········•·········•··---··-····•··-·\-cl -1 .4% ,____,__,__ __,_ __,___.___,____,_~ 0 ~ ~ 00 00 1001~1~100 Time (h) FIGURE 7-2. Effect of varying photolysis coefficients on the normalised sensitivity of the ozone concentration towards some rate coefficients. The dashed lines show calculations with constant photolysis coefficients, the straight lines show calculations with a day- night cycle. Case: 298 K, 21 June. Sensitivity Study 85 PLUME case at 298 K was made once for a normal day - night cycle of 21 June and once with constant photolysis coefficients set to the daily average. FIGURE 7-2 shows the sensitivity of the ozone concentration for some sample reactions showing low, medium and high change when going from constant photolysis coefficients to diurnal varying photolysis coefficients. reactions. Photolysis at The N02 photolysis and the CH4 + HO reactions are typical daytime night is neglectable and HO reactions cause little change in 0 3 concentration during the night because of the low HO concentration during the night. Consequently the sensitivity of the 0 3 concentration to these reaction changes little during the night. On the other hand the sensitivity is larger during the day than in the case of average light intensity. This explains the steeper TABLE 7-1. Normalised Sensitivity of 0 3 concentrations to selected reaction rate parameter/ Temperature I K 298 298 288 278 288 288 2 Photolysis : Average Summer Summer Summer Summer Winter Summer Emissions: 100% 100% 100% 100% 10% 100% 3 Ozone / ppb 135 137 130 110 47 75 ROOI: N02 + hv 34.9% 35.9 % 37.8 % 37.5 % 30.1 % 35.2 % R048: 0 3 +NO -31.3 % -31.4 % -32.7 % -32.0 % -23.3 % -33.2 % R039: HO + N02 -26.5 % -25.4 % -24.7 % -21.7 % -24.5 % -20.7 % R041: H02 +NO 21.1 % 19.1 % 16.2 % 12.7% 17.1 % 12.6 % Rl27: ACO, + N02 -2.8 % -4.0% -10.8 % -17.1 % -3.2 % -10.8 % % R030: 0 3 + H02 -13.7 % -10.3 % -9.2 % -8.3 % -8.8 % -2.6 R145: AC03 +NO 3.4% 5.4% 10.6% 15.2 % 3.0% 12.1 % R061: CH4 +HO 6.3 % 6.1 % 5.5% 5.0% 12.6 % 4.2% R002: 0 3 +hv -9.8 % -8.0 % -9.3 % -12.2 % -12.3 % 5.4% R049: 0 3 + N02 -2.9 % -9.5 % -5.1 % -2.0% -1.5 % -7.7 % ROI I: HCHO + hv 0.3% I.I% 0.9% 0.6% 0.6% I !.I% R028: 0 1D + HzO -9.1 % -7.4 % -8.7 % -11.3% -11.4 % 5.1 % Rl28: PAN decomp. 2.8 % 3.9% 9.8% 10.3 % 2.9% 9.3% R026: 0 1D + N2 6.5 % 5.2% 6.1 % 8.0% 8.1 % -3.6 % R058: CO+HO 5.8 % 5.9% 7.2% 9.2% 7.7% 4.2% R074: XYL +HO 0.2 % 0.6% 0.7% 0.5 % 0.2% 7.3% R063: HC3 + HO 3.6% 2.8 % 2.0% 0.0% 0.3 % 2.9% R027: 0 1D + 02 2.7 % 2.2% 2.5 % 3.3 % 3.3 % -1.5 % R073: TOL + HO 0.6% 0.7% 0.7% 0.2% 0.2% 5.2% ROJ9: MGLY + hv 0.0% 0.3 % 0.2% 0.1 % 0.0% 5.0% 1 Shown are the daily average (96 h ~ t < 120 h) of the normalised sensitivity of the 0 3 concentration to the most important reactions. 2 "Summer" is 21 June, "Winter" is 21 Dec, "Average Summer" is constant photolysis coefficients daily average of 21 June. 3 Average daily concentration 86 Sensitivity Study TABLE 7-2. Normalised Sensitivity of the PAN concentrations to selected reaction rate parameters' Temperature I K 298 298 288 278 288 288 2 Photolysis : Average Summer Summer Summer Summer Winter Summer Emissions: 100% 100% 100% 100% 10% 100% PAN3 IJ2J2b 0.98 0.80 3.23 6.54 0.12 2.90 Rl28: PAN decomp. -91.9 % -107.0 % -69.0% -21.5 % -94.3 % -42.7 % Rl27: AC03 + N02 92.4% 106.7 % 73.7 % 34.5 % 97.6% 47.8% Rl45: AC03 +NO -76.9 % -71.7 % -56.2 % -28.5 % -72.0 % -37.3 % R039: HO + N02 -49.0% -41.1 % -39.4% -30.2 % -48.5 % -34.6 % ROOl: N02 + hv -38.8 % -47.0% -30.0 % -8.3 % -44.4 % -10.0 % R048:03 +NO 35.0% 40.0% 25.3 % 7.0% 34.6% 9.1 % R041: H02 +NO 23.3 % 15.4% 16.5 % 10.8% 18.2 % 15.9 % R049: 0 3 + N02 -2.l % -19.7 % -8.4% -1.4 % -5.8 % -2.0% R063: HC3 +HO 14.5 % 13.8 % 13.0% 10.0% 16.4% 9.5 % R227: AC03 + N03 -0.2% -15.7 % -4.9% -0.7 % -4.4 % -6.9% R078: KET+ HO 14.2% 14.7 % 9.9% 5.1 % 10.9% 3.4% ROI I: HCHO + hv 0.4% 0.3% 1.1% 1.1% -0.5 % 13.0 % R073: TOL + HO 1.2 % 0.7 % 1.8 % 2.6% 1.6 % 12.2 % R074: XYL +HO 0.2% 0.7% 1.4% 2.2% 0.9% 11.2 % R077: ALD + HO 4.5% -1.5 % 3.7% 5.2% 4.6% 7.5% R019: MGLY + hv 0.3% -0.l % 0.5% 0.8% 0.1 % 7.0% R002: 0 3 +hv -0.3 % -3.3 % 2.2% 4.4% -6.9% 6.6% R028: O'D +HP -0.3 % -3.0% 2.1 % 4.1 % -6.4% 6.1 % R092: ALD + N03 0.0% 5.3% 0.7% 0.2% 0.6% 6.4% R014:0P2+hv 3.2% 3.4% 2.8% 2.4% 6.0% 0.4% R016: KET+ hv 5.8% 5.5% 4.4% 3.1 % 5.9% 1.2 % R020: DCB + hv -0.2 % -0.1 % 0.3 % 0.5% -0.l % 5.8% R076: HCHO +HO -0.4% -0.3 % -0.7 % -0.8 % 0.3% -5.6 % R046: HO + HN03 4.7% 5.5 % 4.0% 2.7 % 4.6% -0.4% R164: AC03 + H02 -3.1 % -3.4% -2.5 % -1.2 % -5.4% -0.9 % R165: AC03 + H02 -2.3 % -2.5 % -2.l % -1.2 % -5.1 % -0.4% Rt 47: KETP + NO 4.7% 4.5% 3.1 % 1.7 % 4.1 % 0.3% R185: AC03 + M02 -2.4% -3.3 % -1.7 % -0.5 % -4.6% -0.3 % 1 R026: 0 D + N2 0.2% 2.1 % -1.5 % -2.9% 4.5% -4.3 % R168: KETP + H02 -4.4% -4.5% -3.0% -1.6 % -4.4% -0.3 % 1 Shown are the daily average (96 h S t < 120 h) of the normalised sensitivity of the PAN concentration to the most important reactions. 2 "Summer" is 21 June. "Winter" is 21 Dec, "Average Summer" is constant photolysis coefficients daily average of 21 June. ' Average daily concentration Sensitivity Study 8 7 slope during the day and causes the sensitivity of the 0 3 concentration to these kinds of reactions to look like stairs. However the sensitivity calculated for the varying photolysis coefficients follows the sensitivity for the averaged coefficients quite well. When looking at a daily average, the differences are small (see also TABLE 7-1). When averaged constant photolysis coefficients are used, night-time chemistry is neglected. Species such as N0 3 have no chance to build up. Therefore the sensitivity towards reactions which are involved in the night time chemistry changed most. Reaction R091 (HCHO + N03) is a typical night time reaction. It is not important when using average photolysis coefficients, but has a significant influence if night-time chemistry is available. Reaction R049 (03 + N02 ) even ~ N03 + 0 2 is an important ozone loss reaction during the day. However it becomes more important during the night, when N03 is not immediately photolysed and a significant part of the reactive nitrogen is converted into N03 and N20 5 (see chapter 6). ) Reaction Rl27 (AC03 + N02 and Rl28 (PAN decomposition) are other examples for reactions which are important during the day but are also involved in night-time chemistry. The first two data columns of TABLE 7-1 and TABLE 7-2 show the sensitivity of the 0 3 and PAN analysis respectively, with average photolysis coefficients and with diurnal varying coefficients. For 0 3 the highest difference in the normalised sensitivity coefficients between ) this two cases was found for reaction R049 (03 + N02 changing from -2.9% to -9.5, for ) reaction R030 (03 + H02 changing from -13.7% to -10.3% and for reaction Rl45 (AC03 + NO) changing from 3.4 to 5.4%. For PAN, the highest difference in the normalised sensitivity coefficients was also found for ) reaction R049 (03 + N02 going from -2.1 % to -19.7%. A big change was also found for ) reaction R227 (AC03 + N03 from -0.2% to -15.7%, for reaction Rl28: (PAN decomposition.) ) from -91.9% to -107.0% and for reaction Rl27 (AC03 + N02 changing from 92.4 to 106.7%. The relative change in the sensitivity for night-time reactions is very high for both species. For example the ozone sensitivity to reaction R227 (AC03 + N03) changes by more than two orders of magnitudes from -0.01 % to -1.1 %. and the PAN sensitivity to reaction R23 I (OLND + N03) changes by more than three orders of magnitudes from 0.0004% to 0.7%. 7.3.3 Temperature Dependence of Sensitivity Coefficients To investigate the temperature dependence of sensitivity coefficients a sensitivity analysis was made for the PLUME case with temperature of 278K, 288K and 298K. TABLE 7-1 shows the resulting normalised sensitivity coefficients for ozone. Remarkable is the increase of the sensitivity of the ozone concentration to all PAN reactions (Rl27: AC03 + N02, Rl28: PAN decomposition., Rl45: AC03 + NO), when going to lower temperatures. The increase is a factor of 3 to 4 when going from 298K to 278K. Interesting is, that the sensitivity to the PAN decomposition greatly changes when going from 298K to 288K, but only a small increase is observed when decreasing the temperature to 278K. It can be expected, that the 0 3 sensitivity to the PAN decomposition decreases again when going to temperatures below 278 K because this reaction becomes very slow. 88 Sensitivity Study The sensitivity of PAN (see TABLE 7-2) shows a totally different behaviour to temperature changes. The sensitivity to most reactions decreases significantly when going to lower temperatures. However note that TABLE 7-2 shows normalised sensitivity coefficients and that the PAN concentration increases by a factor of 8 when going from 298 K to 278 K. At lower temperatures the night-time chemistry becomes less important for PAN. For example PAN has a high sensitivity coefficient of -15.7% to reaction R227 (ACOJ + NOJ) at 298K, but at 278 the same coefficient is only -0. 7%. Please note that this behaviour can not be observed when averaged photolysis coefficients were used, because in that case this sensitivity coefficient is only -0.2% at 298 K. 7.3A Influence of Total Emissions and Season Two additional sensitivity calculations have been done to investigate the dependence of sensitivity coefficients on the scale of emissions and on the season, results of which can be found in the last two columns of TABLE7-1 and TABLE 7-2. When the NO, and non-methane hydrocarbon emissions are decreased by a factor of 10 (as in the LOW-VOC-NOx-Case explained in section 6.2.2) the ozone sensitivity to PAN reactions becomes smaller and so does the sensitivity to R048 (OJ +NO). On the other hand the reaction R061 (CH4 +HO) becomes more important. However the sensitivity of ozone to other reactions only changes little despite the big change in the available NO, and VOC. The sensitivity of the PAN concentration is more influenced by the decrease of the emissions. The sensitivity to the first 6 reactions listed in TABLE 7-2 increases by about 30%. It seems that the uncertainty of predicted PAN concentrations increases when looking at cases with cleaner conditions. The influence of the actinic flux on the sensitivity coefficients was examined by doing a sensitivity analysis using photolysis rate coefficients calculated for January 21. Compared with the summer photolysis rates some major changes in ozone and PAN sensitivity were observed. Some coefficients were even changed in sign. The biggest change in the ozone sensitivity coefficients was found for reaction R002: (03 + hv ~ O'D) changing from -9.3% to +5.4%, to 1 reaction R028 (0 0 + H20 ~ 2 HO) changing from -8.7% to +5.1 %, to reaction RO! I (HCHO 1 + hv) changing from +0.9% to + 11.1 % and to reaction R026: (0 0 + N2) changing from +6.1 % to -3.6%. This suggests that the ozone production in the winter scenario is limited by the production of radicals. Reaction R002 destroys ozone but may, together with reaction R028, produce HO radicals. Photolysis of organic species as in reaction RO! I and reaction RO 19 (MGL Y + hv) lead also to higher radical concentrations. It is interesting to note that the sensitivity of ozone to many photolysis reactions increases despite the lower photolysis coefficients in winter. This supports the suggestion that under these conditions ozone production is limited by production of radicals. For PAN the sensitivity to the key reactions (R128: PAN decomposition, Rl27: AC03 + N02, Rl45: AC03 +NO) decreases by about one third in the winter, while on the other hand the sensitivity to the photolysis of organic species and their reaction with HO increases. Sensitivity Study 89 7.35 Sensitivity of Ratios The reactions of acylperoxy with NO and N02 are a good example of competing reactions where the ratio is better known than the absolute values. This reaction system is interesting because the sensitivity of 0 3 to the reaction with NO is positive while the sensitivity to the reaction with N02 is negative. FIGURE 7-3 shows the relative change in the predicted 0 3 concentration resulting from small changes in the rate coefficient of reaction R127 (AC03 + N02) and R 145 (AC03 + NO). These calculations were done for the PLUME case at 288 K using diurnal varying photolysis coefficients for June 21. The resulting surface is almost a plane as can be expected for small changes in the parameter. Interesting is that the resulting ozone does not change significantly when moving from the centre toward the A and C, which corresponds to a change in kR 127 and kR 145 while keeping the ratio kR 12 /kR 145 constant. On the other hand the resulting ozone changes when moving from the centre toward B and D, which corresponds (for small changes) to a change in kR 127 and kR 145 while keeping the geometric mean ~kR 127 • kR 145 constant. FIGURE 7-3. Relative response of the predicted ozone concentration to small changes in rate coefficients of reaction Rl27 (ACO, +NO,) and RJ45 (ACO, +NO). Calculated for the PLUME case at 288 K, June 21. The undisturbed response is marked with a circle. This justifies eqs 7-XIV, which are consequently used to calculate the sensitivity of PAN and ozone to the ratio and the geometric mean of reaction R 127 and R 145 as found in TABLE 7-3. It is obvious that the relative sensitivity of the PAN and ozone concentration to the ratio of the rate coefficients, kRl2/kR 145 , is much higher than to its geometric mean, ~kR 127 • kR 145 • 90 Sensitivity Study TABLE 7-3. Sensitivity to the ratio ofk.,,,tk"'45 and the geometric mean ~kRm X kR145 • Ozone sensitivity PAN sensitivity 1 Case ~kR127 • kR145 ~kR127 • kRl4S 298 K, Summer, 100% -4.7% 0.7% 84.6% 7.7% 288 K, Summer, 100% -10.7% -0.1% 89.2 % 17.5 % 278 K, Summer, 100% -16.1 % -1.0% 65.0% 8.8% 288 K, Summer, 10% -3.1% -0.1% 31.5 % 3.0% 288 K, Winter, 100% -11.5% 0.6% 84.8% 12.8 % 1 Temperature, photolsysis and emissions. This finding justifies relative measurements such as the one presented in the chapters 3 and 4 of this thesis. For bigger changes of a parameter the system response is not necessary linear. In FIGURE 7-4 the relative change of the 0 3, N02 and PAN concentration are plotted as a function of the ratio kR 12/kR 145 • A vertical line marks the ratio, ~ 12.,lkR 145 = 0.41, which is the one presented in this thesis for peroxyacetyl nitrate. When using this value the predicted noon concentration for c 0 60% ::: ...I'll c Cl) 40% -(.) c 0 0 20% .5 0% Cl) cCD ~ -20% 0 Cl) -40 % .? iii 'ii) -60 % a: 0 0.2 0.4 0.6 0.8 1.2 1.4 FIGURE 7-4. Relative change of the 0 3 , NO, and PAN noon concentration as a function of the ratio of the rate coefficients of reaction Rl27 (ACO_, +NO,) and R145 (ACO, + NO). The vertical line indicates a ratio of 0.41 corresponding the ratio presented in chapter 3 of this thesis. The reference ratio used is 0.46 as defined in the RACM mechanism. Parameters used: Emissions: PLUME-case, actinic flux: 21 June, Temp.: 288 K, time: 108 h. Sensitivity Study 91 PAN decreases by 5.5 % and the OJ and N02 concentrations rise by I respectively 2 % compared to the ratio, kRl27/kR 145 - 0.46 at 288 K, defined in the RACM mechanism (Stockwell et al., 1997). 7.4 Conclusions The SBOX model system was modified to allow the calculation of sensitivity coefficients using the decoupled direct method. By splitting the rate coefficients into a time dependant term and a time independent constant equal to unity, it was possible to calculate sensitivity coefficients for a system with time dependant rate coefficients. It was shown, that the sensitivity to reactions involved in night-time chemistry are underestimated when average constant photolysis coefficients were used. The temperature dependence of the sensitivity coefficients of ozone and PAN was investigated. It was found that the sensitivity of 0 3 to PAN reactions increases with decreasing temperature. The normalised sensitivity of PAN to most reactions decreases when going to lower temperatures, but the absolute sensitivity increases because of the increased PAN concentration. The sensitivity of ozone to most reactions changes little when the total emissions in the system are reduced, while the sensitivity of PAN to most reactions increases. With winter-time photolysis coefficients, the sensitivity of ozone to reactions producing radicals is increased. This suggests that the ozone production is limited by the number of radicals in the system. Finally a method was presented to calculate the sensitivity to ratios of rate coefficients. For the reaction system acylperoxy + NO, acylperoxy + N02, it was shown that ozone and PAN are much more sensitive to the ratio of the corresponding rate coefficients than to their geometric mean. 8. Literature Addison, M.C., J.P. Burrows, R.A. Cox, and R. Patrick, Absorption Spectrum and Kinetics of the Acetylperoxy Radical, Chemical Physics Letters, 73 (2), 283-287, 1980. AEA, FACSIMILE 3.0 User Guide, AEA, United Kingdom Atomic Energy Authority, Harwell, Didcot, Oxfordshire, United Kingdom, 1994. Atkinson, R., Gas-Phase Tropospheric Chemistry of Organic Compounds, Journal of Physical and Chemical Reference Data, Monograph 2, 1-216, 1994. Atkinson, R., D.L. Baulch, R.A. Cox, R.F. Hampson Jr., J.A. Kerr, M.J. Rossi, and J. Troe, Evaluated Kinetic and Photochemical Data for Atmospheric Chemistry: Supplement V, Enclosed Diskette, Atmospheric Environment, 30 (22), 3903-3904, 1996. Atkinson, R., D.L. Baulch, R.A. Cox, R.F. Hampson, J.A. Kerr, and J. Troe, Summary of Reactions and Preferred Rate Data, Journal of Physical and Chemical Reference Data, 21 (6), 1125, 1992. Basco, N., and S.S. Pannar, The Reaction of Acetylperoxy Radicals with N02 , International Journal of Chemical Kinetics, 19, 115-128, 1987. Becker, H., and Co-workers, Organikum: organisch-chemisches Grundpraktikum, VEB Deutscher Verlag der Wissenschaften, Berlin, 1976. Becker, K.H., and F. Kirchner, Kinetische Untersuchungen an Peroxynitraten und Peroxy- Radikalen, Bericht Nr. 29, Bergische Universitat Gesamthochschule Wuppertal, Fachbereich 9 Physikalische Chemie, 1994. Berney, C.V., Spectroscopy of CF3COZ compounds - III Vibrational spectrum and barrier to internal rotation of trifluoroacetaldehyde, Spectrochimica Acta, 25A, 793-809, 1969. Bottenheim, J.W., and A. Sirois, Long-Term daily mean mixing ratios of 0 3, PAN, HN03, and particle nitrate at a rural location in eastern Canada: Relationships and implied ozone production efficiency,Journal of Geophysical Research, 101 (D2), 4189-4204, 1996. Bridier, I., F. Caralp, H. Loirat, R. Lesclaux, and B. Veyret, Kinetic and Theoretical Studies of the Reactions CH3C(0)02 + N02 + M <=> CH3C(0)02N02 + M between 248 and 393 K and between 30 and 760 Torr, Journal of Physical Chemistry, 95 (9), 3594-3600, 1991. Brown, P.N., G.D. Byrne, and A.C. Hindmarsh, VODE: A Variable-Coefficient ODE Solver, Journal on Scientific and Statistical Computing, JO (5), 1038-1051, 1989. Bruckmann, P.W., and H. Willner, Infrared Spectroscopic Study of Peroxyacetyl Nitrate (PAN) and Its Decomposition Products, Environmental Science & Technology, 17 (6), 352- 357, 1983. 94 Literature BUWAL, Ozan in der Schweiz, Schriftenreihe Umweltschutz, Nr. 101, Bundesamt filr Umwelt, Wald und Landschaft (BUW AL), Bern, 1989. Canosa-Mas, C.E., M.D. King, R. Lopez, C.J. Percival, R.P. Wayne, D.E. Shallcross, J.A. Pyle, and V. D~le, Is the reaction between CH3C(0)02 and N02 important in the night- time troposphere?, Journal of The Chemical Society, Faraday Transactions I, 92 (12), 2211-2222, 1996. Cantrell, C.A., W.R. Stockwell, L.G. Anderson, K.L. Busarow, D. Pemer, A. Schmeltekope, J.G. Calvert, and H.S. Johnston, Kinetic Study of the N03-CH20 Reaction and its Possible Role in Nighttime Tropospheric Chemistry, Journal of Physical Chemistry, 79, 139-146, 1984. Carroll, M.A., D.R. Hastie, B.A. Ridley, M.O. Rodgers, A.L. Torres, D.D. Davis, J.D. Bradshaw, S.T. Sandholm, H.I. Schiff, D.R. Karecki, G.W. Harris, G.l. Mackay, G.L. Gregory, E.P. Condon, M. Trainer, G. Hubler, D.D. Montzka, S. Madronich, D.L. Albritton, H.B. Singh, S.M. Beck, M.C. Shipham, and A.S. Bachmeier, Aircraft Measurements of NO, Over the Eastern Pacific and Continental United States and Implications for Ozone Production, Journal of Geophysical Research, 95 (D7), l 0205- 10233, 1990. Carter, W.P.L., and R. Atkinson, Computer Modeling Study of Incremental Hydrocarbon Reactivity, Environmental Science & Technology, 23, 7, 1989. Chang, J.S., F.S. Binkowski, N.L. Seaman, J.N. McHenry, P.J. Samson, W.R. Stockwell, C.J. Walcek, S. Madronich, P.B. Middleton, J.E. Pleim, and H.H. Lansford, The Regional Acid Deposition Model and Engineering Model, State-of-Science/Technology, Report 4, National Acid Precipitation Assessment Program. Washington D.C, 1990. Cox, R.A., R.G. Derwent, P.M. Holt, and J.A. Kerr, Photolysis of Nitrous Acid in the Presence of Acetaldehyde, Journal of The Chemical Society, Faraday Transactions I, 72, 2061-2075, 1976. Cox, R.A., and M.J. Roffey, Thermal Decomposition of Peroxyacetylnitrate in the Presence of Nitric Oxide, Environmental Science & Technology, 11 (9), 900-906, 1977. Crutzen, P.J., The Role of NO and N02 in the Chemistry of the Troposphere and Stratosphere, Annual review of earth and planetary sciences, 7, 443-472, 1979. Cukier, R.I., H.B. Levine, and K.E. Shuler, Nonlinear Sensitivity Analysis of Multiparameter Model Systems, Journal of Computational Physics, 26, 1-42, 1978. Darley, E.F., K.A. Kettner, and E.R. Stephens, Analysis of Peroxyacyl Nitrates by Gas Chromatography with Electron Capture Detection, Analytical Chemistry, 35 (4), 588-591, 1963. Demerjian, K.L., K.L. Schere, and J.T. Peterson, Theoretical estimates of actinic (spherically integrated) flux and photolytic rate constants of atmospheric species in the lower troposphere, Advances in Environmental Sciences and Technology, JO, 369-459, 1980. Literature 95 DeMore, W.B., S.P. Sander, D.M. Golden, R.F. Hampson, M.J. Kurylo, C.J. Howard, A.R. Ravishankara, C.E. Kolb, and M.J. Molina, Chemical Kinetics and Photochemical Data for Use in Stratm,pheric Modeling, Evaluation Number 11, 94-26, NASA Panel for Data Evaluation, JPL Publication 94-26, 1994. Derwent, R.G., and M.E. Jenkin, Hydrocarbons and the Long-Range Transport of Ozone and PAN Across Europe, Atmospheric Environment, 25A (8), 1661-1878, 1991. Dibble, T.S., M.M. Maricq, and J.J. Szente, Kinetics of the Reaction of CF30 with NO, Journal of Physical Chemistry, 99 (48), 17394-17402, 1995. Dunker, A.M., The decoupled direct method for calculating sensitivity coefficients in chemical kinetics, Journal of Physical Chemistry, 81 (5), 2385-2393, 1984. Eberhard, J., and C.J. Howard, Rate Coefficients for the Reactions of some C3 to C5 Hydrocarbon Peroxy Radicals with NO, Journal of Physical Chemistry, 101 (18), 3360- 3366, 1997. Finlayson-Pitts, B.J., and J.N. Pitts, Atmospheric Chemistry: Fundamentals and Experimental Techniques, John Wiley & Sons, 1986. Friedrich, R., European Emissions Distribution and Trends, in Proceedings of the EUROTRAC Symposium '96, Transport and Transformation of Pollutants in the Troposphere, Volume 2: Emissions, Deposition, Laboratory Work and Instrumentation, edited by P.M. Borrell, P. Borrell, K. Kelly, T. Cvitas, and W. Seiler, pp. 29-34, Computational Mechanics Publications, Southampton, 1996. Gaffney, J.S., R. Fajer, and G.I. Senum, An improved Procedure for High Purity Gaseous Peroxyacyl Nitrate Production: Use of Heavy Lipid Solvents, Atmospheric Environment, 18 (1), 215-218, 1984. Gaffney, J.S., N.A. Marley, and E.W. Prestbo, Peroxyacyl Nitrates (PANs): Their Physical and Chemical Properties, in The Handbook of Environmental Chemistry, edited by 0. Hutzinger, pp. 1-38, Springer, Berlin (etc.), 1989. Gao, D., Sensitivity and Uncertainty Analysis in Chemical Systems for Air Quality Modeling, Ph.D. Thesis, University of Connecticut, 1995. Gao, D.F., W.R. Stockwell, and J.B. Milford, First-order sensitivity and uncertainty analysis for a regional-scale gas-phase chemical mechanism, Journal of Geophysical Research, 100, 23153-23166, 1995. Gear, C.W., Numerical Initial Value Problems in Ordinary Differential Equations, Prentice- Hall, Englewood Cliffs, N. J., 1971. Gery, M.W., G.Z. Whitten, J.P. Killus, and M.C. Dodge, A photochemical mechanism for urban and regional scale computer modeling, Journal of Geophysical Research, 94, 12925- 12956, 1989. Grob, K., and G. Grob, Capillary Columns with Very Thick Coatings, Journal of High Resolution Chromatography & Chromatography Communications, 6, 133-139, 1983. 96 Literature Grosjean, D., E.L. Williams II, and E. Grosjean, Ambient Levels of Peroxy-n-Butyryl Nitrate at a Southern California Mountain Forest Smog Receptor Location, Environmental Science & Technology, 27 (2), 326-331, 1993a. Grosjean, D., E.L. Williams II, and E. Grosjean, A Biogenic Precursor of Peroxypropionyl Nitrate: Atmospheric Oxidation of cis-3-Hexen-l-ol, Environmental Science & Technology, 27, 979-981, 1993b. Grosjean, D., E.L. Williams II, and E. Grosjean, Peroxyacyl Nitrates at Southern California Mountain Forest Locations, Environmental Science & Technology, 27 ( l ), 110-121, l 993c. Grosjean, E., D. Grosjean, M.P. Fraser, and G.R. Grass, Air Quality Model Evaluation Data for Organics. 3. Peroxyacetyl Nitrate and Peroxypropionyl Nitrate in Los Angeles Air, Environmental Science & Technology, 30 (9), 2704-2714, 1996. Harris, S.J., and J.A. Kerr, Relative Rate Measurements of Some Reactions of Hydroxyl Radicals with Alkanes Studied under Atmospheric Conditions, International Journal of Chemical Kinetics, 20, 939-955, 1988. Hartsell, B.E., V.P. Aneja, and W.A. Lonneman, Relationships between peroxyacetyl nitrate, 0 3, and NOY at the rural Southern Oxidants Study site in central Piedmont, North Carolina, site SONIA,Journal of Geophysical Research, 99 (DJO), 21033-21041, 1994. Hendry, D.G., and R.A. Kenley, Generation of Peroxy Radicals from Peroxy Nitrates (R02N02). Decomposition of Peroxyacyl Nitrates, Journal of the American chemical Society, 99 (9), 3198-3199, 1977. Reuss, J.M., and W.A. Glasson, Hydrocarbon Reactivity and Eye Irritation, Environmental Science & Technology, 2 (12), 1109-1116, 1968. Hippeli, S., and E.F. Elstner, Mechanisms of Oxygen Activation during Plant Stress: Biochemical Effects of Air Pollutants, Journal of Plant Physiology, 148 (3), 249-257, 1996. IUPAC, see Atkinson et al., 1992. IUPAC, see Atkinson et al., 1996. Joseph, J.H., and W.J. Wiscombe, The Delta-Eddington Approximation for Radiative Flux Transfer, Journal of the Atmospheric Sciences, 33, 2452-2459, 1976. Kenley, R.A., and D.G. Hendry, Generation of Peroxy Radicals from Peroxynitrates (ROON02). Decomposition of Peroxybenzoyl Nitrate (PBzN), Journal of the American Chemical Society, 104, 220-224, 1982. Kerr, J.A., and D.W. Stocker, The Kinetics and Mechanism of the Photo-Oxidation of Propionaldehyde under Simulated Atmospheric Conditions, Journal of Photochemistry, 28, 475-489, 1985. Kirchner, F., and W.R. Stockwell, Effect of peroxy radical reactions on the predicted concentrations of ozone, nitrogenous compounds, and radicals, Journal of Geophysical Research, JOI (DIS), 21007-21022, 1996. literature 97 Kirchner, F., F. Zabel, and K.H. Becker, Determination of the Rate Constant Ratio for the Reactions of the Acetylperoxy Radical with NO and N02, Berichte der Bunsengesellschaft fur Physikalische Chemie, 94, 1379-1382, 1990. Kirchner, F., F. Zabel, and K.H. Becker, Kinetic behaviour of benzoylperoxy radicals in the presence of NO and N02, Chemical Physics Letters, 191(1,2),169-174, 1992. Kleindienst, T.E., Recent Developments in the Chemistry and Biology of Peroxyacetyl Nitrate, Research on Chemical Intermediates, 20 (3/4/5), 335-384, 1994. Kourtidis, K.A., P. Fabian, C. Zerefos, and B. Rappengliick, Peroxyacetyl nitrate (PAN), peroxypropionyl nitrate (PPN) and PAN/ozone ratio measurements at three sites in Germany, Tetlus, 45B (5), 442-457, 1993. Kramer, M.A., J.M. Calo, and H. Rabitz, An improved computational method for sensitivity analysis: Green's function method with 'AIM', Applied Mathematical Modelling, 5, 432, 1981. Kuhn, M., H. Builtjes, D. Poppe, D. Simpson, W.R. Stockwell, Y. Andersson-Skoeld, A. Baart, M. Das, F. Fiedler, 0. Hov, F. Kirchner, P.A. Makar, J.B. Milford, M.G.M. Roemer, R. Ruhnke, A. Strand, B. Vogel, and H. Vogel, Intercomparison of the Gas-Phase Chemistry in Several Chemistry and Transport Models, Atmospheric Environment, submitted, 1997. Langer, S., I. Wangberg, and E. Ljungstriim, Heterogeneous Transformation of Peroxyacety !nitrate, Atmospheric Environment, 26A, 3089-3098, 1992. Leone, J.A., and J.H. Seinfeld, Comparative Analysis of Chemical Reaction Mechanisms for Photochemical Smog, Atmospheric Environment, 19 (3), 437-464, 1985. Lesclaux, R., Combination of peroxyl radicals in the gas phase, in Peroxy Radicals, edited by Z.B. Alfassi, John Wiley and Sons, London, 1997. Lightfoot, P.D., R.A. Cox, J.N. Crowley, M. Destriau, G.D. Hayman, M.E. Jenkin, G.K. Moortgat, and F. Zabel, Organic Peroxy Radicals: Kinetics, Spectroscopy and Tropospheric Chemistry, Atmospheric Environment, 26A (10), 1805-1961, 1992. Lurmann, F.W., A.C. Lloyd, and R. Atkinson, A chemical mechanism for use in long-range transport/acid deposition computer modeling, Journal of Geophysical Research, 91, 10905- 10936, 1986. Madronich, S., Photodissociation in the Atmosphere; 1. Actinic Flux and the Effects on Ground Reflections and Clouds, Journal of Geophysical Research, 92 (D8), 9740-9752, 1987. Maricq, M.M., and J.J. Szente, The 193 and 248 nm Photodissociation of CF3C(O)Cl, Journal ofPhysical Chemistry, 99 (13), 4554-4557, 1995. Maricq, M.M., and J.J. Szente, A Temperature-Dependent Study of the CH3C(0)02 + NO Reaction,Journal of Physical Chemistry, 100 (30), 12380-12385, 1996. 98 Literature Maricq, M.M., J.J. Szente, G.A. Khitrov, T.S. Dibble, and J.S. Francisco, CF3CO Dissociation Kinetics, Journal ofPhysical Chemistry, 99 (31), 11875-11822, 1995. Maricq, M.M., J.J. Szente, G.A. Khitrov, and J.S. Francisco, The CF3C(0)02 Radical. Its UV Spectrum, Self-Reaction Kinetics, and Reaction with NO, Journal of Physical Chemistry, 100 (10), 4514-4520, 1996. McCroskey, P.S., and G.J. McRae, Documentation for the Direct Decoupled Sensitivity Analysis Method--DDM, Report, Department of Chemical Engineering, Carnegie Mellon University, Pittsburgh, PA, 1987. McRae, G.J., W.R. Gooden, and J.H. Seinfeld, Mathematical Modeling of Photochemical Air Pollution, EQL Report No. 18, Environmental Quality Laboratory, California Institute of Technology, Pasadena, California, United States, 1982a. McRae, G.J., J.W. Tilden, and J.H. Seinfeld, Global Sensitivity Analysis A Computational Implementation of the Fourier Amplitude Sensitivity Test (FAST), Computers & Chemical Engineering, 6 (I), 15-25, 1982b. Middleton, P., W.R. Stockwell, and W.P.L. Carter, Aggregation and analysis of volatile organic compound emissions for regional modeling, Atmospheric Environment, 24 A (5), 1107-1133, 1990. Milford, J.B., D. Gao, A.G. Russell, and G.J. McRae, Use of sensitivity analysis to compare chemical mechanisms for air-quality modeling, Environmental Science & Technology, 26, 1179-1189, 1992. Mineshos, G., and S. Glavas, Thermal Decomposition of Peroxypropionyl Nitrate; Kinetics of the Formation of Nitrogenous Products, Reaction kinetics and catalysis letters, 45 (2), 305- 312, 1991. Mudd, J.B., in Responses of Plants to Air Pollution, edited by J.B. Mudd, and T.T. Kozlowski, pp. 97-119, Academic Press, New York, 1975. Neininger, B., and J. Dammen, POLLUMET - Luftverschmutzung und Meteorologie in der Schweiz, Umwelt-Materialien Nr. 63, Bundesamt fiir Umwelt, Wald und Landschaft (BUWAL), Bern, 1996. Nielsen, T., A.M. Hansen, and E.L. Thomsen, A convenient method for preparation of pure standards of peroxyacetyl nitrate for atmospheric analyses, Atmospheric Environment, 16 (10), 2447-2450, 1982. Nielsen, T., K. Pilegaard, A.H. Egel0v, K. Granby, P. Hummelsh0j, N.O. Jensen, and H. Skov, Atmospheric Nitrogen Compounds: occurrence, composition and deposition, The Science of the Total Environment, 1891190, 459-465, 1996. Penkett, S.A., and K.A. Brice, The spring maximum in photo-oxidants in the Northern Hemisphere troposphere, Nature, 319, 655-657, 1986. Literature 99 Posthumus, A.C., and A.E.G. Tonneijck, Monitoring of Effects of Photo-Oxidants, in Monitoring of air pollutants by plants, pp. 115-119, Dr W. Junk Publishers, The Hague, 1992. Pszenny, A.A.P., W.C. Keene, D.J. Jacob, S. Fan, J.R. Maben, M.P. Zetwo, M. Springer- Young, and J.N. Galloway, Evidence of Inorganic Chlorine Gases Other than Hydrogen Chloride in Marine Surface Air, Geophysical Research Letter, 20 (8), 699-702, 1993. Roberts, J.M., The Atmospheric Chemistry of Organic Nitrates, Atmospheric Environment, 24A (2), 243-287, 1990. Roberts, J.M., and S.B. Bertman, Peroxymethacrylic Nitric Anhydride (MPAN), International Journal of Chemical Kinetics, 24, 297-307, 1992. Russell, A., J. Milford, M.S. Bergin, S. Mcbride, L. Mcnair, Y. Yang, W.R. Stockwell, and B. Croes, Urban ozone control and atmospheric reactivity of organic gases, Science, 269, 491- 495, 1995. Schrimpf, W., K. Lienaerts, K.P. Mueller, J. Rudolph, R. Neubert, W. Schussler, and I. Levin, Dry Deposition of peroxyacetyl nitrate (PAN): Determination of its Deposition Velocity at Night from Measurements of the Atmospheric PAN and 222Radon Concentration Gradient., Geophysical Research Letter, 23 (24), 3599-3602, 1996. Schurath, U., and V. Wipprecht, Proc. I st European Symp. on the Physico-Chemical Behaviour of Atmospheric Pollutants, edited by B. Versino, and H. Ott, pp. 157-166, Commiss. European Commun., Ispra, 1980. Scollard, D.J., J.J. Treacy, and H.W. Sidebottom, Rate Constants for the Reaction of Hydroxyl Radicals and Chlorine Atoms with Halogenated Aldehydes, Journal of Physical Chemistry, 97 (18), 4683-4688, 1993. Scorer, R., Air Pollution, Pergamon Press, Oxford, Great Britain, 1968. Seefeld, S., and J.A. Kerr, Kinetics of the Reactions of Propionylperoxy Radicals with NO and N02: Peroxypropionyl Nitrate Formation under Laboratory Conditions Related to the Troposphere, Environmental Science & Technology, in press, 1997. Seefeld, S., D.J. Kinnison, and J.A. Kerr, Relative Rate Study of the Reactions of Acetylperoxy Radicals with NO and N02: Peroxyacetyl Nitrate Formation under Laboratory Conditions Related to the Troposphere, J oumal of Physical Chemistry, 101 (I), 55-59, 1997. Seigneur, C., G. Stephanopoulos, and R.W. Carr Jr., Dynamic Sensitivity Analysis of Chemical Reaction Systems, A Variational Method, Chemical Engineering Science, 37 (6), 845-853, 1982. Semadeni, M., Hydroxyl Radical Reactions with Volatile Organic Compounds Under Simulated Tropospheric Conditions: Tropospheric Lifetimes, Ph.D. Thesis, Swiss Federal Institute of Technology Zurich (ETH), 1994. 100 Literature Semadeni, M., D.W. Stocker, and J.A. Kerr, Further Studies of the Temperature Dependence of the Rate Coefficients for the Reactions of OH with a Series of Ethers under Simulated Atmospheric Conditions, Journal of Atmospheric Chemistry, 16, 79-93, 1993. Shepson, P.B., D.R. Hastie, K.W. So, and H.I. Schiff, Relationships between PAN, PPN and 0 3 at Urban and Rural Sites in Ontario, Atmospheric Environment, 26A (7), 1259-1270, 1992. Shettle, E.P., and J.A. Winman, The Transfer of Solar Irradiance Through Inhomogeneus Turbid Atmospheres Evaluated by Eddington's Approximation, Journal of the Atmospheric Sciences,27, 1048-1055, 1970. Singh, H.B., and L.J. Salas, Methodology for the analysis of peroxyacetyl nitrate (PAN) in the unpolluted atmosphere, 17, 1507-1516, 1983. Solomon, I.J., A Study of PAN-Type Compounds by Oxygen-17 NMR Techniques, Summary Report, C6268-2, IIT Research Institute Technology Center Chicago, Illinois 60616, 1973a. Solomon, I.J., and J.N. Keith, Synthesis of Halogen compounds in High Valence States, Final Technical Report, IITRI-C6140-12, UT Research Institute Technology Center Chicago, Illinois 60616, 1973b. Stemmler, K., W. Mengon, D.J. Kinnison, and J.A. Kerr, OH Radical-Initiated Oxidation of 2- Butoxyethanol under Laboratory Conditions Related to the Troposphere: Product Studies and Proposed Mechanism, Environmental Science & Technology, 31(5),1496-1504, 1997. Stephens, E.R., The Formation, Reactions, and Properties of Peroxyacetyl Nitrates (PANs) in Photochemical Air Pollution, Advances in Environmental Sciences and Technology, 1, 119- 147, 1969. Stephens, E.R., E.F. Darley, O.C. Taylor, and W.E. Scott, Photochemical Reaction Products in Air Pollution, International Journal ofAir and Water Pollution, 4 (l/2), 79-100, 1961. Stephens, E.R., P.L. Hanst, R.C. Doerr, and W.E. Scott, Reactions of Nitrogen Dioxide and Organic Compounds in Air, Industrial and Engineering Chemistry, 48 (9), 1498-1504, 1956. Stockwell, W.R., F. Kirchner, M. Kuhn, and S. Seefeld, A New Mechanism for Regional Atmospheric Chemistry Modeling, Journal of Geophysical Research, in press, 1997. Stockwell, W.R., P. Middleton, X. Tang, and J.S. Chang, The Second Generation Regional Acid Deposition Model Chemical Mechanism for Regional Air Quality Modeling, Journal of Geophysical Research, 95 (DJO), 16,343-16,367, 1990. Stockwell, W.R., and J.B. Milford, Mechanisms for Atmospheric Chemistry, Sensitivity, Uncertainty and Chemical Kinetic Data, in EUROTRAC Newsletter No. 16, pp. 2-7, 1995. Stockwell, W.R., J.B. Milford, D. Gao, and Y.-J. Yang, The Effect of Acetyl Peroxy-Peroxy Radical Reactions on Peroxyacetyl Nitrate and Ozone Concentrations, Atmospheric Environment, 29 (14), 1591-1599, 1995. Literature 101 Talukdar, R.K., J.B. Burkholder, A.-M. Schmoltner, J.M. Roberts, R.R. Wilson, and A.R. Ravishankara, Investigation of the loss processes for peroxyacetyl nitrate in the atmosphere: UV photolysis and reaction with OH, Journal of Geophysical Research, 100 (D7), 14163- 14173, 1995. Taylor, O.C., Importance of Peroxyacetyl Nitrate (PAN) as a Phototoxic Air Pollutant, Journal of the Air Pollution Control Association, 19 (5), 347-351, 1969. Tuazon, E.C., W.P.L. Carter, and R. Atkinson, Thermal Decomposition of Peroxyacetyl Nitrate and Reactions of Acetyl Peroxy Radicals with NO and N02 over the Temperature Range 283 313 K, Journal of Physical Chemistry, 95, 2434-2437, 1991. Villalta, P.W., and C.J. Howard, Direct Kinetics Study of the CH3C(0)02 + NO Reaction Using Chemical Ionization Mass Spectrometry, Journal of Physical Chemistry, 100 (32), 13624-13628, 1996. Villenave, E., and R. Lesclaux, Kinetics of the Reactions of Selected Peroxy Radicals with CH3C(0)02, in Proceedings of the EUROTRAC Symposium '96, Transport and Transformation of Pollutants in the Troposphere, Volume 2, edited by P.M. Borrell, P. Borrell, T. Cvitas, K. Kelly, and W. Seiler, Computational Mechanics Publications, Southampton, 1996. Villenave, E., R. Lesclaux, S. Seefeld, and W.R. Stockwell, Kinetics and Atmospheric Implications of Peroxy Radical Cross Reactions involving the CH 3C(0)02 Radical, Journal of Geophysical Research, submitted. Wallington, T.J., J.M. Andino, J.C. Japar, and M.S. Ball, Fourier Transform Infrared Studies of the Reaction of Cl Atoms with PAN, PPN, CH300H, HCOOH, CH3COCH3 and CH 3COC2H5 at 295±2 K, Journal ofAtmospheric Chemistry, 10, 301-313, 1990. Wallington, T.J., P. Dagaut, and M.J. Kurylo, Ultraviolet Absorption Cross Sections and Reaction Kinetics and Mechanisms for Peroxy Radical in the Gas Phase, Chemical Reviews, 92, 667-710, 1992. Wallington, TJ., P. Dagaut, R. Liu, and M.J. Kurylo, The Gas Phase Reactions of Hydroxyl Radicals with a Series of Esters Over the Temperature Range 240-440 K, International Journal of Chemical Kinetics, 20, 177-186, 1988a. Wallington, T.J., and M.D. Hurley, A Kinetic Study of the Reaction of Chlorine and Fluorine Atoms with CF3CHO at 295 ± 2 K, International Journal of Chemical Kinetics, 25, 819- 824, 1993. Wallington, T.J., M.D. Hurley, O.J. Nielsen, and J. Sehested, Atmospheric Chemistry of CF3CO, Radicals: Fate of CF3CO Radicals, the UV Absorption Spectrum of CF3C(0)02 Radicals, and Kinetics of the Reaction CF3C(0)02 + NO -> CF3C(0)0 + N02, Journal of Physical Chemistry, 98, 5686-5694, 1994a. l 02 Literature Wallington, T.J., J. Sehested, and 0.J. Nielsen, Atmospheric chemistry of CF3C(0)02 radicals. Kinetics of their reaction with N02 and kinetics of the thermal decomposition of the product CF3C(0)02N02, Chemical Physics Letters, 226, 563-569, l 994b. Wallington, T.J., L.M. Skewes, W.O. Sieg!, C.-H. Wu, and S.M. Japar, Gas Phase Reaction of Cl Atoms with a Series of Oxygenated Organic Species at 295 K, International Journal of Chemical Kinetics, 20, 867-875, 1988b. Whitten, G.Z., and J.P. Meyer, CHEMK: A Computer Modeling Scheme for Chemical Kinetics, Report Number CS75-0R2, Systems Applications Incorporated, San Rafeal, California, 1975. Williams II, E.L., E. Grosjean, and D. Grosjean, Ambient Levels of the Peroxyacyl Nitrates PAN, PPN and PMAN in Atlanta, Georgia, Journal of the Air & Waste Management Association, 43, 873-879, 1993. World Health Organization, Photochemical oxidants, Environmental health criteria No. 7, WHO, Geneva, Switzerland, 1979. Yang, Y.-J., W.R. Stockwell, and J.B. Milford, Uncertainties in Incremental Reactivities of Volatile Organic Compounds, Environmental Science & Technology, 29 (5), 1336-1345, 1995. Yang, Y.-J., W.R. Stockwell, and J.B. Milford, Effect of Chemical Product Yield Uncertainties on Reactivities of VOCs and Emissions from Reformulated Gasolines and Methanol Fuels, Environmental Science & Technology, 30 ( 4), 1392-1397, 1996. Zabel, F., F. Kirchner, and K.H. Becker, Thermal Decomposition of CF3C(0)02N02, CCIF2C(0)02N02, CC1 2FC(0)02N02 , and CC1 3C(0)02N02, International Journal of Chemical Kinetics, 26, 827-845, 1994. A-1 Appendix A: FACSIMILE Program * Simulation of the reaction scheme of the PAN-Experiment Neglects NO->N02 conversion etc. Question: What happens at higher temperatures, when the backreaction ( 1) becomes important? Date: 21. 06. 96 Author: St. Seefeld (EAWAG) Parameter: Startingconditions: [(CH3CH0)2] 10 ppm [NO)+[N02] - 7 ppm [02} 21 % Rate constants used: kl 6.lE-12 cm2 molecule-1 s-1 k2 - kl I (ratio as measured) (CH3CH0)2 -> CH3CHO : k-0.00125 sA-1 VARIABLES ppm Conversion ppm -> molecule cmA-3 kl Rate coeff. CH3C03 + N02 k2 Rate coeff. CH3C03 NO k3 Rate cocff, PAN-> CH3C03 + N02 k3_o Low-pressure rate coefficient of k3 k3_oo High-pressure rate coefficient of k3 EXEC OPEN 7 "pan04_T-313.out" NEW ; * Define constants; PARAMETER T 313; * Temp I K; PARAMETER p 925; • Pressure / hPa; PARAMETER R 83.145; * Gas constant / hPa dmA3 molA-1 KA·l; PARAMETER Na 6.02E23; * Avogadros constante; PARAMETER NOX 7; * [NO]+[N02] / ppm; PARAMETER BRANCHINGRATIO . 41; * Startingconditions VARIABLE ppm CH3COCOCH3 CH3CO 02 CH3C03 N02 NO N02_0 NO_O PAN C02 kl k2 k3 k3_o k3_oo k3_f fact NON02RATIO xx xxl PANNOO COMPILE INSTANT; * Rate constants; ppm M (p/(R*T)*Na*lE-9); * conv. factor ppm-> molecule cmA·3; A-2 FACSIMILE Program kl 6.lE-12; k2 kl / BRANCHINGRATIO; k3 o - 4.9E-3*EXP(-12100/T)*(lE6 *ppm); k3_oo - 4.0E16*EXP(-13600/T); k3_ffact-(LOG10(0.3)/(l+LOG10(k3_o/k3_oo)**2)); k3 -k3_o*k3_oo/(k3_o+k3_oo) * 10@k3_ffact; **i COMPILE INITIAL; * Set the initial values of the variables; * This procedure is called by the 'BEGIN' command; CH3CO 0.0; CH3C03 0.0; PAN 0. 0; C02 0. 0; CH3COCOCH3 - 10 * ppm; NO_O (NOX *ppm) * NON02RATIO / (l+ NON02RATIO); N02 0 (NOX *ppm) * 1 / (l+ NON02RATIO); N02 N02_0; NO NO_O; 02 . 21 * ppm * 1E6; * 21 % Oxigen TIME 0.0; **; * Is Executed at the beginning; COMPILE INSTANT; CALL WRHEADER; **; COMPILE EQUATIONS; * Definieren der Reaktionen; % .000125 CH3COCOCH3 - CH3CO + CH3CO; * Photolysis of CH3COCOCH3 % 9.4E-12 CH3CO + 02 CH3C03; % kl CH3C03 + N02 PAN; % k2 CH3C03 + NO C02 ; •no N02 production!; % k3 PAN - CH3C03 + N02; * Back reaction; **; * Control of output; WHENEVER * l) TIME - 0.0 + 2.0•20% CALL OUT2 l) TIME 40. % CALL OUT2; ** ; COMPILE INSTANT; WRITE 1-7, "PARAMETER"; 11 11 WRITE 1=7, "T = T," K ; WRITE 1=7 1 "p -" p," hPa"; WRITE 1-7, "kl/k2-", BRANCHINGRATIO; 11 11 *WRITE 1=7, TIME= 1 40, "s"; WRITE 1-7, "NO(O)+N02(0)-", NOX, " ppm"; WRITE 1-7, "k3 -", (El0,5), k3," l/s"; WRITE 1-7, "l/k3 -", &(l/k3)," s - ",&(l/k3/60), "min - ",&(l/k3/3600)," h"; II ___ -- ______-- ______WRITE 1-7, 11 WRITE 1-7, " NO/N02(0) NO/N02{t)" "NO N02 PAN C02 Alfa(O)" Alfa(t) PANO/PAN"; FACSIMILE Program A-3 **; COMPILE OUT2; • Wenn NO~O wird PANNOO gesetzt; if (NO_O) * * 2; PANNOO ~ PAN; LABEL 2; • Calculate the ratio kl/k2; xx ~ O; xxl ~ 0; IF ( C02) 1 1 •; • Ueberspringen, falls <~O"; xx~ (PAN/ C02) • (NO_O / N02_0); xxl ~ (PAN/ C02) • (NO I N02); LABEL l; WRITE 1~7, (Fll, 4)' NON02RATIO, (( FlO, 7))' &(NO I N02)' &(NO/ppm), &(N02 I ppm)' &(PAN I ppm)' &(C02 I ppm)' xx, xxl, (Fl2, 7), &(PANNOO/PAN) %; **; COMPILE OUTl PSTREAM 4 · ** ; • output of one Run; PSTREAM 3 7 12; TIME CH3COCOCH3 CH3CO CH3C03 N02 NO PAN C02; ** ; • Output of final result; PSTREAM 4 7 12; NON02RATIO NO_O N02 0 NO N02 PAN C02; ** i EXEC NON02RATIO~O; BEGIN; EXEC NON02RATIO~.l; BEGIN; EXEC NON02RATI0~.25; BEGIN; EXEC NON02RATI0~.5; BEGIN; EXEC NON02RATIO~l; BEGIN; EXEC NON02RATIO~l.5; BEGIN; EXEC NON02RATI0~2; BEGIN; EXEC NON02RATI0~2.5; BEGIN; EXEC NON02RATI0~3; BEGIN; EXEC NON02RATI0~3.5; BEGIN; EXEC NON02RATI0~4; BEGIN; A-4 FACSIMILE Program EXEC NON02RATI0~4.5; BEGIN; EXEC NON02RATI0~5; BEGIN; STOP; B-1 Appendix B. RACM Model Species List Stable Inorganic Compounds Stable Organic Compounds Oxidants Alkanes 0 3 Ozone CH 4 Methane H20 2 Hydrogen peroxide ETH Ethane HC3 Alkanes, alcohols, esters and al- Nitrogenous compounds kynes with HO rate constant NO Nitrogen oxide (298K, I atm) less than 3.4 x 10"12 1 cm3 s- • N02 Nitrogen dioxide HCS Alkanes, alcohols, esters and al- N03 Nitrogen trioxide kynes with HO rate constant N205 Dinitrogen pentoxide 12 (298K, latm) between 3.4 x 10- HON Nitrous acid and 6.8 x 10·12 cm3 s·'. 0 alcohols, esters and al- HN0 Nitric acid HC8 Alkanes, 3 kynes with HO rate constant HN0 Pemitric acid 4 (298K, latm) greater than 6.8 x I 0- 12 cm3 s·'. Sulfur compounds S02 Sulfur dioxide Alkenes SULF Sulfuric acid ETE Ethene OLT Terminal alkenes Carbon Oxides OLI Internal alkenes CO Carbon monoxide DIEN Butadiene and other anthropoge- C02 Carbon Dioxide nic dienes Abundant Stable Species Stable biogenic alkenes N2 Nitrogen ISO Isoprene 0 Oxygen 2 API a-pinene and other cyclic H20 Water terpenes with one double bond Hydrogen LIM d-limonene and other cyclic diene-terpenes Inorg. Short-Lived Intermediates Aromatics Atomic Oxigen TOL Toluene and less reactive aroma- 3 0 P Ground state oxygen atom, 0(3P) tics 1 O'D Excited state oxygen atom, 0( 0) XYL Xylene and more reactive aroma- tics Odd hydrogen CSL Cresol and other hydroxy substi- HO Hydroxy radical tuted aromatics H02 Hydroperoxy radical B-2 RACM Model Species List Stable Organic Compounds (cont.) Organic Short-lived Intermediates Peroxy radicals from alkanes Carbonyls M0 Methyl peroxy radical HCHO Formaldehyde 2 ETHP Peroxy radical formed from ETH ALD Acetaldehyde and higher aide- hydes HC3P Peroxy radical formed from HC3 KET Ke tones HC5P Peroxy radical formed from HC5 GLY Glyoxal HC8P Peroxy radical formed from HC8 MGLY Methylglyoxal and other a- Peroxy radicals from alkenes carbonyl aldehydes ETEP Peroxy radicals formed from ETE DCB Unsaturated dicarbonyls OLTP Peroxy radicals formed from OLT MACR Methacrolein and other unsatu- OLIP Peroxy radicals formed from OLI rated monoaldehydes UDD Unsaturated dihydroxy Peroxy rad. from biogenic alkenes dicarbonyl ISOP Peroxy radicals formed from ISO HKET Hydroxy ketone APIP Peroxy radicals formed from API LIMP Peroxy radicals formed from LIM Organic nitrogen ONIT Organic nitrate Radicals produced from aromatics PAN Peroxyacetyl nitrate and higher PHO Phenoxy radical and similar ra- saturated PAN s dicals ADDT Aromatic-HO adduct from TOL TPAN Unsaturated PANs ADDX Aromatic-HO adduct from XYL ADDC Aromatic-HO adduct from CSL Organic peroxides TOLP Peroxy radicals formed from TOL OPl Methyl hydrogen peroxide XYLP Peroxy radicals formed from XYL OP2 Higher organic peroxides CSLP Peroxy radicals formed from CSL PAA Peroxyacetic acid and higher analogs Peroxy rad. with carbonyl groups AC03 Acetyl peroxy and higher saturated Organic acids acyl peroxy radicals ORAl Formic acid TC03 Unsatured acyl peroxy radicals ORA2 Acetic acid and higher acids KETP Peroxy radicals formed from KET Other peroxy radicals OLNN N03-alkene adduct reacting to form Carbonitrates + H02 OLND NO,-akene adduct reacting via decomposition X02 Accounts for additional NO to N02 conversions :Sr Reaction k' :Sr Reaction k' Photolysis Reactions R025 O'P+03 -; 2 o, 7.96.UJ" 3 1 ROOI NO,-; 0 P+NO 7.50·1{)' R026 0 D+N,-; O'P+N, 2.60·10" 1 1 R002 o,_, 0 D+O, l.62·!0' R027 0 D+02 -; O'P+O, 4.05·10" 3 1 R003 0,-; 0 P+O, 4.17-JO"' R02S 0 D+H,O-; HO+HO 2.20·Jff10 3 R004 HONO-; HO+NO 1.63·!0" R029 O,+HO-; H02 +02 6.83·10" ROOS HNO,-; HO+NO, 4.50-HT' R030 0 3 +H02 -; H0+202 2.05·UJ" 10 R006 HN04-; 0.65 H02 + 0.65 NO,+ 0.35 HO+ 0.35 3.17-10" R031 HO+HO,-; H,O+O, Ll 1·l0' +NO, ...., R032 H20 2 +HO-; H02 +H20 l.70.!0"' R007 NO,-; N0+02 2.3HQ·' :r R033 H02 +H02-; H,O,+O, 2.92·!0·12 G ROOS N03 -; N02 +o'P 1.87-10" R034 HO,+ HO, + H20 -; H20 2 + 0 2 + H,O 6.58·!0""' ~ -; R009 H20 2 HO+HO 6.00-10' 3 R035 0 P+NO-; NO, 1.66-IO"' >(') ROJO HCHO-; H,+CO 3.50·!0'' 3 R036 0 P+N02 -> N0+02 9.72-IO"' ROii HCHO-; 2HO,+CO 2.17-10' s;: R037 O'P+NO,-; NO, l.58·!0" R012 ALD-; MO,+HO,+CO 3.67·!0' s;: R038 HO+NO-; HONO 4.87·1012 G R013 OP!-; HCHO + HO,+ HO 4.17·!0' 11 ('") R039 HO+NO,-; HN03 l.15·10" :r R014 OP2-; ALD+HO,+HO 4.lHO' $:>) R040 HO+NO,-; NO,+HO, 2.20·10" ::i ROl5 PAA-; M02 +HO 1.57.10• R041 HO,+NO-; NO,+HO 8.56·10 12 en ROl6 KET-; ETHP+AC03 6.67·!0'' -· -; s R042 H02 +N02 HNO, l.39·10'" R017 GLY-; 0.13 HCHO + 1.87 CO+ 0.87 H, 5.83·!0" 2 R043 HNO,-; H02 +N02 8.62-10 ROIS GLY-; 0.45 HCHO + 1.55 CO+ 0.80 HO,+ 0.15 2.()().lO" H, R044 HO,+NO,-; 0.3 HN03 + 0.7 NO,+ 0.7 HO+ 0 2 3.50·!0·12 ROI9 MGLY-; CO+ HO, + ACO, 9.33-10'' R045 HO+HONO-; N02 +H20 4.86·10"' R020 DCB-; TC03 +H02 4.33·!0" R046 HO+HN03 -; N03 +H,O l.52·10"' 2 R021 ONJT-; 0.20 ALO+ 0.80 KET+ HO,+ NO, 2.IHO' R047 HO+HNO,-; N02 +02 +H20 4.65-10" 1 R022 MACR-; CO+ HCHO +HO,+ ACO, 1.33.10• R048 0 3 + NO-; N0,+02 1.82-10 ' R023 HKET-; HCHO + HO,+ ACO, 6.67-10'' R049 O,+NO,-; NO,+O, 3.23·10"7 Inorganic Reactions R050 NO+NO +02 -;NO,+ N02 l.95·10"' R024 O'P+02 -; o, J.50·10"' R051 NO,+NO-; NO,+NO, 2.65·10'" 0 _,I 0 Nr Reaction k' Nr Reaction k' I I\.) R052 N03 +NO,-> NO+N0,+02 6.56·10"' R075 CSL+HO-; 0.85 ADDC + 0.10 PHO+ 0.05 H02 + 6.00·lff" 0.05 XO, R053 N03 +N02 -> N20 5 1.27·!0"' ..... R076 HCHO+HO-> H02 +CO+H,O l.00· lff" ;;;; R054 N,O, -> N02 +N03 4.36-lff' CJ) R077 ALD+HO AC03 +H20 1.69·!0"' :v R055 N03 +N03 -> N02+N02 +0, 2.29·10"' l> R078 KET+HO-> KETP+H,O 6.87·!0"' 0 R056 HO+H2 -> H,O+HO, 6.69-10-" $: R079 HKET+HO-> H02 + MGL Y + H,O 3.00·10"' R057 HO+SO,-> SULF+H02 8.89·10"' $: (J) ROSO GLY+HO-; HO, +2CO +H20 1.14-Iff" (") R058 CO+HO-> H02 +C02 2.40-!0-" ::r ROSI MGLY+HO-t ACO, + CO + H20 1.72·!0"' Q O'P + Organic Compowuis :J R082 MACR+HO-> 0.51 TCO, + 0.41 HKET + 0.08 MGL Y + 3.35-10"' R059 ISO+ 0 3P-> 0.86 OLT + 0.05 HCHO + 0.02 HO + O.DI 6.00·10"' 0.41 CO +0.08 HCHO + 0.49 H02 + 0.49 3'°" CO+ 0.13 DCB+ 0.28 HO,+ 0.15 XO, xo, 3 R060 MACR + 0 P -t ALD L66·10"' R083 DCB+HO-> 0.50 TC03 + 0.50 HO, + 0.50 XO, + 0.35 5.04-lff" UDD+0.15 GLY +0.15 MGLY HO+ Organic Compowuis R084 UDD+HO-> 0.88 ALD + 0.12 KET+ HO, 2.70·!0"' R061 CH,+HO-t M02 +H20 6.86·!0"' R085 OP! +HO-> 0.65 MO, + 0.35 HCHO + 0.35 HO 5.54·10'" R062 ETH+HO ETHP+H20 2.57·1ff" R086 OP2+HO 0.44 HC3P + 0.08 ALD + 0.41 KET+ 0.49 6.43·lff" R063 HC3+H0-> 0.583 HC3P + 0.381 HO, + 0.335 ALD 2.20-!0"" H0+0.07X0 + 0.036 ORAi + 0.036 CO + 0.036 GL Y 2 R087 PAA+HO-> + 0.036 HO+ 0.0 I 0 HCHO + H 0 0.35 HCHO + 0.65 AC03 + 0.35 HO, + 5.54-IO"' 2 0.35 xo, R064 HC5 +HO-> 0.75 HC5P + 0.25 KET+ 0.25 H01 + H,O 4.77·!0"' R088 PAN +HO -t HCHO + X02 + H20 + NO, 4.00-lff" R065 HC8+HO-> 0.951 HC8P + 0.025 ALD + 0.024 HKET 1.08-IO"' R089 TPAN+HO-> 0.60 HKET + 0.40 HCHO + 0.40 HO, + l.74-IO"' + 0.049 H01 + H,O X02 + 0.40 PAN+ 0.60 NO, R066 ETE+HO-> ETEP 8.52·10"' R090 ONIT+ HO-> HC3P +NO,+ H,O 2.22·10"' R067 OLT+HO-> OLTP 3.06-10"' N0 + Organic Compounds R068 OLJ+HO-> OLJP 7.!2·10"' 3 R091 HCHO + N03 -> H02 + HN03 +CO 5.79·10"' R069 DIEN +HO-> ISOP 6.65·10"' R092 ALD+N03 -> AC03 +HN03 2.38·!0"' R070 ISO+ HO -t !SOP LOl·lO"' R093 GLY +NO,-> HN03 +H02 + 2CO 4.94-!0"' R071 APl+HO-> APIP 5.45·10"' R094 MGL Y + N03 -> HNO, + ACO, +CO 2.38·!0"' R072 LIM+HO-> LIMP l.70·!0"' R095 MACR + N03 -> 0.20 TC03 + 0.20 HN03 + 0.80 OLNN + 5.00·10"' R073 TOL+HO-> 0.90ADDT + 0.10X02 + 0.10 H02 5.96·lff" 0.80CO R074 XYL+HO-> 0.90 ADDX + 0.10 XO,+ 0.10 H02 2.40-10"' Nr Reaction k' Nr Reaction k' R096 DCB+N03 -; 0.50 TC03 + 0.50 H02 + 0.50 X01 + 0.25 l.OO·lOi' RllO IS0+03 -t 0.90 HCHO + 0.39 MACR + 0.36 CO + l.28·10''' 3 GL Y + 0.25 ALD + 0.03 KET + 0.25 0.15 ORAi + 0.09 0 P + 0.30 H02 + 0.35 MGL Y + 0.5 HN03 + 0.5 N02 OLT + 0.28 HO+ 0.05 H2 + 0.15 AC03 + 0.03 M0 + 0.02 KETP + 0.13 XO, + R097 CSL +N03 -t HN03 +PHO 2.20·!0'" 2 0.001 H20 2 R098 ETE+N03 -t 0.80 OLNN + 0.20 OLND 2.05·10"' Rill APT +03 -; 0.65 ALD + 0.53 KET+ 0.14 CO+ 0.20 8.69·!0"' R099 OLT+N03 -; 0.43 OLNN + 0.57 OLND 3.95·!0'" ETHP + 0.42 KETP + 0.85 HO + 0.10 H02 + 0.02 H20 2 RlOO OLI +N03 -; 0.1 I OLNN + 0.89 OLND 3,91.10·' R!l2 LIM+0 -t OLT + 0.14 CO+ 0.16 1 3 0.04 HCHO + 0.46 2.00.10·" RlOI DIEN+N03 -; 0.90 OLNN + O. l 0 OLND + 0.90 MACR LO· lff ~ ETHP + 0.42 KETP + 0.85 HO+ 0.10 HO, R\02 ISO+ N03 -; 0.90 OLNN + 0.10 OLND + 0.90 MACR 8.96·!0"!) + 0.02 H20 2 + 0.79 MACR + 0.01 ORAi + 0.070RA2 Rl03 APT+ N03-; 0.10 OLNN + 0.90 OLND 6.16·10"' Rll3 MACR+03 -t 0.40 HCHO + 0.60 MGL Y + 0.13 ORA2 + 1.14· JO'" Rl04 LIM+N03-; 0.13 OLNN + 0.87 OLND l.22·10"" 0.54 CO + 0.08 H2 + 0.22 ORAi + 0.29 HO, +0.07 HO+ 0.13 OP2 + 0.13 AC0 RIOS TPAN +N03 -; 0.60 ONIT + 0.60 N03 + 0.40 PAN+ 0.40 4.11-10" 3 HCHO + 0.40 NO, + XO, Rll4 DCB+03 -; 0.21 HO + 0.29 HO, + 0.66 CO + 0.50 2.00.10-l' GLY + 0.28 AC03 + 0.16 ALD + 0.62 0 3 + Organic Compounds MGLY + 0.11PAA+0.Jl ORAi +0.21 Rl06 ETE+03 -; HCHO + 0.43 CO + 0.37 ORAi + 0.26 1.59· 10" ORA2 HO, + 0.13 H2 + 0.12 HO RIIS TI'AN + 0 3 -; 0.70 HCHO + 0.30 PAN + 0.70 N02 + 8.19·10"' Rl07 OLT +03 _, 0.64 HCHO +0.44 ALD + 0.37 CO+ 0.14 J.03.10-CI' 0.13 CO + 0.04 H, + 0.11 ORAi + 0.08 ORA! + 0.10 ORA2 + 0.25 H02 + 0.40 HO,+ 0.036 HO+ 0.70 AC03 HO + 0.03 KET + 0.03 KETI' + 0.06 CH4 Reactions Of Intermediates Produced by Aromatic Oxidation + 0.05 H2 + 0.006 H20 2 + 0.03 ETH + 0.19 MO,+ 0.10 ETHP Rll6 PHO+N02 _, 0. IO CSL+ ONIT 2.00·10'" RIOS OLI +0 _, 0.02 HCHO + 0.99 ALD + 0.16 KET+ -1 3 2.58·10'" Rll7 PHO+H02 -l CSL l.00-10"" :; 0.30 CO+ 0.011 H20 2 + 0.14 ORA2 + 0.07 (j) Rll8 ADDT + N02 4 CSL+ HONO 3.60·10"" CH4 + 0.22 HO, + 0.63 HO + 0.23 MO, ;;n )> + 0.12 KETI' + 0.06 ETH+ 0.18 ETIIP Rll9 ADDT+0 4 0.98 TOLP + 0.02 CSL+ 0.02 H0 2 2 5.52·10"' () _, Rl09 DIEN+03 0.90 HCHO + 0.39 MACR + 0.36 CO + 6.33·10'" _, ~ 3 R120 ADDT +03 CSL+HO 5.00·10'" 0.15 ORAi + 0.09 0 P + 0.30 H02 + 0.35 ~ Rl21 ADDX + N0 _, CSL+ HONO OLT + 0.28 HO + 0.05 H, + 0.15 AC03 2 3.60·10"' (l) + 0.03 MO,+ 0.02 KETP + 0.13 X0 + () 2 Rl22 ADDX +02 -l 0.98 XYLP + 0.02 CSL+ 0.02 HO, 5.52·10'" :J 0.001 H20 2 0 RJ23 ADDX+O,_, CSL+HO \.()().JO-" :::l RI24 ADDC + N02 _, CSL + HONO 3.60·!0"" ~r 0 ~ 0 Nr Reaction k' Nr Reaction k' b. R125 ADDC+O,-; 0.98 CSLP + O.OZ CSL + 0.02 H02 5.52 x!O" Rl42 TOLP+NO-; 0.95 NO, + 0.95 H02 + 0.65 MGL Y + 1.20 4.00·10"" GLY + 0.50 DCB+ 0.05 ONIT " -I Rl43 XYLP+NO-; 0.95 NO, + 0.95 HO, + 0.60 MGL Y + 0.35 "::J Rl26 ADDC +03 -; CSL+HO 5.00·10-" 4.00·10-" GL Y + 0.95 DCB + 0.05 ONIT (!) Peroxyacylnitrate Formation and Decomposition Rl44 CSLP+NO-; GLY + MGLY +HO,+ NO, 4.00.10·" "')> Rl27 AC03 +N02 -; PAN 8.66·10·" 0 Rl45 AC03 +NO-; MO,+NO, 2.00-10"" ~ Rl28 PAN-; AC03 +N02 4.63· HJ" R146 TC03 +NO-; AC03 + HCHO +NO, 2.00-!0-" ~ (!) Rl29 TC03 +N02 -; TPAN 8.66-10-" 0 Rl47 KETP+NO-; 0.54 MGL Y + 0.46 ALD + 0.23 AC03 + 4.00-10-" ::r Rl30 TPAN-; TC01 +N01 4.63·!0" 0.77 HO,+ 0.16 xo,+ NO, Cl ::J Rl48 OLNN+NO-; H02 + ONIT +NO, 4.00·!0"' q;- NO + Organic Peroxy Radicals 3 Rl49 OLND+NO-; 0.287 HCHO + 1.24 ALD + 0.464 KET+ 4.00·10"" Rl31 M02 +NO-; HCHO +HO,+ N02 7.68-10'" 2N02 Rl32 ETHP+NO-; ALD + H0 + N0 8.70-10'" 2 2 HO, + Organic Peroxy Radicals Rl33 HC3P+ NO-; 0.047 HCHO + 0.233 ALD + 0.623 KET + 4.00-10'" R150 M02 +HO,-; OP! 5.57·10'" 0.063 GLY +0.742 HO,+ 0.15 M02 + 0.048 ETHP + 0.048 XO,+ 0.059 ONIT + Rl51 ETHP+HO,-; OP2 7.86·10"" 0.94! NO, Rl52 HC3P+H02 -; OP2 l.30·10"' Rl34 HC5P+NO-; 0.021 HCHO + 0.211 ALD + 0.722 KET+ 4.00·!0"' Rl53 HC5P+H02 -; OP2 1.30·10"' 0.599 HO, + 0.031 M02 + 0.245 ETHP + 0.334 X01 + 0.124 ONIT + 0.876 NO, Rl54 HC8P+HO,-; OP2 l.30-10'" HC8P+NO-; Rl35 0.15 ALD + 0.642 KET+ 0.133 ETHP + 4.00-!0"' R155 ETEP+HO,-; OP2 1.50·10"" 0.261 ONIT +0.739 N02 + 0.606 H02 + R156 OLTP+ HO,-; 0.416X02 OP2 l.30·!0"' Rl36 ETEP+NO-; 1.6 HCHO + H02 + NO,+ 0.2 ALD 9.00·10"' R157 OLIP+H02 -; OP2 l.30-10"' R137 OLTP+ NO-; 0.94 ALD + HCHO + HO, + NO, + 0.06 4.00-10"" Rl58 !SOP+ HO,-; OP2 l.00-10"" KET Rl59 APIP+H02 -; OP2 150-10"' Rl38 OLIP+NO-; HO,+ 1.71 ALD +0.29 KET+ N02 4.00·!0"" R160 LIMP+HO,-; OP2 150-10"" Rl39 !SOP+ NO-; 0.446 MACR + 0.354 OLT + 0.847 HO,+ 4.00-10"" TOLP+ HO,-; OP2 0.606 HCHO + 0.153 ONIT + 0.847 N02 Rl61 l.Ol·IO"" Rl40 APIP+NO-; 0.80 HO, + 0.80 ALD + 0.80 KET + 0.20 4.0-10"' Rl62 XYLP+ H02 -; OP2 1.01· 10"" ONIT + 0.80 N02 Rl63 CSLP+H02 -; OP2 1.01-10"" Rl41 LIMP+NO-; 0.65 HO, + 0.40 MACR + 0.25 OLI + 0.25 4.0-10"" HCHO + 0.35 ONIT + 0.65 NO, Rl64 AC03 +H02 -; PAA 7.28· IO"" Rl65 AC03 +HO,-; ORA2+0, 2.72·10"" Nr Reaction k' Nr Reaction k' RI66 TCO,+ H01 _, OP2 7.28·10" Rl86 AC03 +M02 _, HCHO+ORA2 L22·10" RI67 TCO,+H02 _, ORA2+01 2.72·!0"' Rl87 TC03 +M02 _, 2 HCHO + H02 + AC03 133·10" RI68 KETP+HO, _, OP2 9.02·!0"' Rl88 TC03 +M02 _, HCHO+ORA2 L22·10'" Rl69 OLNN + H02 _, ONIT L30·IO"' Rl89 KETP+M02 _, 0.75 HCHO + 0.88 HO, + 0.40 MGL Y + 3.80·10" 0.30 ALO+ 030 HKET + 0.12 AC03 + Rl70 OLND + H02 _, ONIT L30·10" o.osxo, Methyl Peroxy Radical+ Organic Peroxy Radicals Rl90 OLNN +MO,_, 0.75 HCHO + H02 + ONIT L72·!0'" Rl71 M02 +M02 _, 1.33 HCHO + 0.66 H02 3.68·10'" Rl91 OLND + MO, _, 0.96 HCHO + 0.5 HO,+ 0.64 ALD + 0. 149 L04·10" KET+ 0.5 N0 + 0.5 ONIT Rl72 ETHP +MO, _, 0.75 HCHO + H02 + 0.75 ALD 2.01·!0'' 2 Rl73 HC3P+MO, _, 0.81 HCHO + 0.992 HO, + 0.58 ALD + 4.02·!0"" Acetyl Radical+ Organic Peroxy Radicals 0.018 KET+ 0.007 MO,+ 0.005 MGLY + Rl92 ETIIP + ACO, _, ALD + 0.5 H02 + 0.5 M02 + 0.5 ORA2 2.09·10"' 0.085 X02 +0.119 GLY Rl93 HC3P+AC03 ->0.724 ALD + 0.127 KET+ 0.488 H0 + 1 2 3.23·10" Rl74 HC5P+MO, _, 0.829 HCHO + 0.946 H02 + 0.523 ALO+ 4.79-10- ) 0.508 MO,+ 0.006 ETHP + 0.071 XO,+ 0.24 KET+ 0.014 ETHP + 0.049 M02 + 0.091 HCHO + 0. JO GL Y + 0.499 ORA2 + 0.245 xo, 0.004MGLY Rl75 HC8P+M0 _, 0.753 HCHO + 0.993 HO,+ 0.411 ALO+ 2 3.63·!0"" R194 HCSP + AC03 _, 0.677 ALO + 0.33 KET + 0.438 H02 + 3.22·!0"' 0.419 KET+ 0.322 X0 + 0.013 ETHP 2 0.554 M02 + 0.495 ORA2 + 0.018 ETHP + _, Rl76 ETEP+ M02 1.55 HCHO + H02 + 0.35 ALD l.84·lff" 0.237 X02 + 0.076 HCHO R177 OLTP+MO,_, l.25 HCHO + H02 + 0.669 ALD + 0.081 l.57·10"' Rl95 HCSP+Aco, .... o.497 ALO+ 0.581 KET+ 0.489 HO,+ 2.44·10" KET 0.507 M02 + 0.495 ORA2 + O.QJ5 ETIIP + 0.318 XO, Rl78 OLIP+M02 _, 0.755 HCHO + H02 + 0.932 ALO+ 0.313 9.87·10" KET Rl96 ETEP + AC03 _, 0.8 HCHO + 0.6 ALD + 0.5 HO, + 0.5 l.24·1ff" Rl79 ISOP+MO, _, 0.550 MACR + 0.370 OLT +HO,+ 0.08 M02 + 0.5 ORA2 1.46·10"' ::1' OLI + 1.09 HCHO R197 OLTP+AC03 -70.859 ALD + 0.501HCHO+0.50! HO,+ l.06·10" - Rl80 HCHO + ALO + KET + 2 HO, 0.501 MO,+ 0.499 ORA2 + 0.141 KET :0 APIP+M02 -> 3.83·10"' )> Rl98 OLJP + AC03 _, 0.941 ALO + 0.569 KET + 0.51 H02 + 6.63·10" () Rl8l LIMP+ M02 _, L4 HCHO + 0.60 MACR + 0.40 OLI + 2 3.83·10'" 0.51 M0 + 0.49 ORA2 HO, 2 ;;:: R!99 ISOP+AC03 -> 0.771 MACR+0.2290LT+0.506H02 + 9.90·!0"' TOLP+M0 _, ;;:: Rl82 2 HCHO + HO,+ 0.35 MGL Y + 0.65 GLY + 3.83· IO"' 0.494 ORA2 + 0.340 HCHO + 0.506 MO, CD DCB () R200 APJP + AC03 _, ALO+ KET+ H02 +MO, 9.63·10"' ::r Rl83 XYLP + M03 _, HCHO +HO,+ 0.63 MGL Y + 0.37 GLY + 3.83· lff" 0 :l DCB R201 LIMP+ AC03 ~ 0.60 MACR + 0.40 OLI + 0.40 HCHO + 9.63·10"' Rl84 CSLP+MO, _, GLY + MGLY +HCH0+2 HO, 3.83·10"' H02 +M02 ~ R202 TOLP + ACO, ~MO, + H02 + 0.35 MGL Y + 0.65 GLY + 9.63·10"' Rl85 ACO,+M02 -'> HCHO + H02 + MO, 7.33·10"' DCB 0 Q, () Nr Reaction k' Nr Reaction k' 6- R203 XYLP+ AC03 -tMO, + H02 + 0.63 MGLY + 0.37 GLY + 9.63-10" R220 OLIP+ N03 -t 1.71 ALD + 0.29 KEf + H02 + N02 1.20·10"' DCB -< R221 ISOP+N03 -; 0.60 MACR + 0 .40 OLT + 0.686 HCHO + J.2Q.JO"" ::r (J) R204 CSLP + AC03 -; GL Y + MGL Y + MO,+ H02 9.63·10" H0 +N0 2 2 :v R205 AC03 + AC03 -; 2 MO, 1.66-10'" R222 APIP+N03 -; ALO + KET + H02 + N02 1.20-10" )> 0 R206 TC03 + AC03 -; M01 + AC03 + HCHO l.66-10" R223 LIMP+N03 -; 0.60 MACR + 0.40 OLI + 0.40 HCHO + 1.20·10'" cs: H02 +N02 R207 KETP + AC03 -t0.54 MGLY + 0.35 ALD + 0.11 KET + 5JXHO'" cs: R224 TOLP+ N0 -t 0.70 MGLY + 1.30 GLY + 0.50 DCB+ CD 0.12 AC03 + 0.38 H02 + 0.08 X02 + 0.5 3 1.20·10" () M0 + 0.5 ORA2 H02 +N02 ::r 1 0 R225 XYLP+N0 -t l.26 MGLY + 0.74 GLY +DCB+ H0 + :> R208 OLNN + AC03 -; ONIT + 0.5 ORA2 + 0.5 M01 + 0.50 H02 1.15·10" 3 2 1.20·10" ;;;· N02 R209 OLND + AC03 -; 0.207 HCHO + 0.65 ALD + 0.167 KET+ 7.00·10" 3 R226 CSLP +N03 -t GL Y + MGL Y + H01 + N02 1.20·10" 0.484 ORA2 + 0.484 ONIT + 0.516 N0 2 + 0.516 M02 R227 AC03 +N03 -t M02 +N02 4.QO.JO"' NO,-Alkene-Peroxyradical + NO rAlkene-Peroxyradical R228 TC03 +N03 -; HCHO + AC03 +NO, 4.00-10" Reactions R229 KETP+N03 -; 0.54 MGLY + 0.46 ALD + 0.77 HO,+ 1.20·!0" R2!0 OLNN + OLNN -; 2 ONIT +HO, 2.00-!0"' 0.23 AC03 + 0.16 X02 + N02 R211 OLNN + OLND -; 0.202 HCHO + 0.64 ALD + 0.149 KET+ 1.22·10" R230 OLNN + N03 -t ONIT +HO,+ NO, l.20· IO'" 0.50 H02 + 1.50 ONIT + 0.50 N01 R231 OLND + NO, -; 0.28 HCHO + 1.24 ALO + 0.469 KET + 2 1.20.10" R212 OLND + OLND -> 0.504 HCHO + 1.21 ALO+ 0.285 KET+ 8.50-10" N01 ONIT+N01 Operator Reactions N03 + Organic Peroxy Radicals R232 X02 +H02 -t OP2 1.30·10"' R213 M02+ N03 -t HCHO + H02 + N01 1.20-10'" R233 X02 +M02 -t HCHO+H02 9.50· IO"' R214 ETHP+N03 -t ALO + H02 + N02 l.20-10" R234 X02 +AC03 -> M02 6.38·!0" R215 HC3P+N03 -; 0.048 HCHO + 0.243 ALD + 0.67 KET + 1.20.10"' R235 XO,+ X01 --> 1.42-10'" 0.063 GLY +0.792 HO,+ 0.155 M01 + 0.053 ETHP + 0.05 l X02 + N02 R236 XO,+NO-t NO, 4.00·!0" R216 HC5P+N03 -; 0.021 HCHO + 0.239 ALO+ 0.828 KET+ 1.20·10"' R237 X02 +N03 -t N02 1.20-10'" 0.699 HO, + 0.04 MO, + 0.262 ETHP + 0.391 XO,-+ NO, - "The rate constants are for 298 K and I atm. The units for rate 1 constants of first order reactions are s· ; of second order reactionst R2l7 HC8P+N03 -> 0.187 ALO+ 0.88 KET+ 0.845 H02 + 1.20·10" s~•; 0.155 ETHP + 0.587 X02 + N02 cm' and for third order reactions, cm• s·' . Photolysis frequencies are given for solar zenith angle 40°, June 21, summer R218 ETEP+ N03 -t 1.6 HCHO + 0.2 ALO+ H02 + N02 l.20·JO" surface 40° northern latitude. R219 OLTP+N01 -t HCHO + 0.94 ALO + 0.06 KET + HO, + 1.20·10" NO,