Laboratory Studies of Organic Peroxy Radical Chemistry: An Overview with Emphasis on Recent Issues of Atmospheric Significance

Transcription

1 Laboratory Studies of Organic Peroxy Radical Chemistry: An Overview with Emphasis on Recent Issues of Atmospheric Significance John J. Orlando and Geoffrey S. Tyndall National Center for Atmospheric Research, Earth System Laboratory, Atmospheric Chemistry Division, Boulder USA Part of the atmospheric chemistry themed issue. ABSTRACT: Organic peroxy radicals (often abbreviated RO 2 ) play a central role in the chemistry of the Earth s lower atmosphere. Formed in the atmospheric oxidation of essentially every organic species emitted, their chemistry is part of the radical cycles that control the oxidative capacity of the atmosphere and lead to the formation of ozone, organic nitrates, organic acids, particulate matter and other so-called secondary pollutants. In this review, laboratory studies of this peroxy radical chemistry are detailed, as they pertain to the chemistry of the atmosphere. First, a brief discussion of methods used to detect the peroxy radicals in the laboratory is presented. Then, the basic reaction pathways involving RO 2 unimolecular reactions and bimolecular reactions with atmospheric constituents such as NO, NO 2, NO 3, O 3, halogen oxides, HO 2, and other RO 2 species - are discussed. For each of these reaction pathways, basic reaction rates are presented, along with trends in reactivity with radical structure. Focus is placed on recent advances in detection methods and on recent advances in our understanding of radical cycling processes, particularly pertaining to the complex chemistry associated with the atmospheric oxidation of biogenic hydrocarbons. 1

2 SECTION 1: INTRODUCTION The chemistry of organic peroxy radicals plays a central role in the chemistry of the Earth s troposphere. Formed as part of the process that oxidizes hydrocarbon species, their subsequent chemistry exerts a controlling influence on the cycling of reactive radicals, and the formation of tropospheric ozone and other important secondary pollutants. e.g.,1-9 In essence, the troposphere can be thought of as behaving like a low-temperature combustion system, with hydrocarbons providing the fuel and radicals of the HO x (OH, HO 2 and organic peroxy radicals, RO 2 ) and NO x (NO and NO 2 ) families acting as catalysts in the oxidative process. 10 The hydrocarbons, including methane and thousands of so-called nonmethane hydrocarbons (NMHCs), are emitted into the atmosphere mostly from natural vegetative sources 11 but also from anthropogenic activity. While the details associated with the removal of these myriad species from the atmosphere are obviously extremely complex, the general oxidative process follows a now well-established pattern. 1-9 The initial step involves reaction of the parent hydrocarbon with a reactive radical - this reactive species is typically OH, although oxidation may also be initiated by NO 3, O 3 and Cl- or Br-atoms. Generally, reaction of OH (and also NO 3 and Cl) with a hydrocarbon occurs via addition to an unsaturated site or abstraction of an H-atom, resulting in both cases in the formation of a carbon-centered (alkyl) radical, e.g., OH + CH 3 CH 3 CH 3 CH 2 + H 2 O OH + CH 2 =CH 2 + M HOCH 2 CH 2 + M These carbon-centered radicals then rapidly add O 2 to produce organic peroxy radicals, collectively designated as RO 2 : CH 3 CH 2 + O 2 + M CH 3 CH 2 OO + M 2

3 HOCH 2 CH 2 + O 2 + M HOCH 2 CH 2 OO + M The peroxy species enjoy a rich chemistry, as reactions with NO, NO 2, HO 2, and other RO 2 species, as well as unimolecular reactions may all contribute, depending on location, time of day, and atmospheric composition. Reactions of RO 2 with NO 3 (nighttime) and with BrO or ClO (maritime regions) may also play a role in certain circumstances. As discussed in more detail in subsequent paragraphs, the RO 2 reaction partner represents a critical branching in the atmospheric oxidative process. Reaction with mostly anthropogenically-emitted NO x leads to net ozone production, and in general propagates radical chains, while reaction with HO 2 /RO 2 (which happens more frequently in more pristine areas) hinders ozone production and leads to dampening of radical cycling. The reaction of RO 2 with NO usually dominates, particularly in continental regions influenced by anthropogenic (NO x ) emissions. The major channel in this class of reactions leads to the formation of an alkoxy radical, RO, along with NO 2, e.g., CH 3 CH 2 OO + NO CH 3 CH 2 O + NO 2 serving to propagate HO x and NO x radical chains and (through subsequent photolysis of the NO 2 ) to generate ozone: RO 2 + NO RO + NO 2 NO 2 + hν NO + O( 3 P) O( 3 P) + O 2 + M O 3 + M Subsequent reactions of the alkoxy species, 12,13 not detailed here, usually lead to the formation of HO 2, and eventually OH (via reaction of HO 2 with NO), thus completing the HO x cycle. Simultaneously, it can be seen that interconversion of NO and NO 2 is occurring, thus completing the NO x cycle. In addition to the radical propagating channel in the RO 2 / NO reaction, there is 3

4 also a minor competing that leads to the formation of organic nitrates; this reaction serves to terminate both HO x and NO x chains, thus dampening ozone production: RO 2 + NO RONO 2 Reactions of RO 2 radicals with NO 2 lead to the formation of peroxynitrates, species that are often thermally unstable: RO 2 + NO 2 + M RO 2 NO 2 + M However, in certain circumstances (e.g., where R is an acyl radical, RC(O)OO ) and/or in colder regions of the atmosphere (such as near the poles, or in the upper troposphere), the lifetimes of the peroxynitrates can be greatly extended, providing an additional sink for both HO x and NO x species. In more pristine regions of the atmosphere, reactions of RO 2 radicals with other RO 2 species and with HO 2 play a larger role and make a contribution to radical removal / chain termination, e.g., RO 2 + HO 2 ROOH + O 2 RO 2 + CH 3 O 2 RO + CH 3 O + O 2 RO 2 + CH 3 O 2 ROH + CH 2 O + O 2 RO 2 + CH 3 O 2 R C=O + CH 3 OH + O 2 There is also an increasing recognition that in these pristine regions (e.g., in the marine boundary layer, or in remote tropical forests), the lifetime of RO 2 species against bimolecular reaction can be increased significantly (to many minutes in extreme cases), allowing for the occurrence of slow, previously unrecognized peroxy radical unimolecular chemistry (e.g., ref. 14). In this review, the various classes of atmospherically-relevant reactions of organic peroxy radicals will be summarized. The review will include a description of each reaction class (e.g., 4

5 reaction with NO, NO 2, etc.) and some general trends in reaction rate with peroxy radical structure will be presented. Representative rate coefficients for the various classes of reactions will be tabulated, to illustrate the trends described in the text. Special attention will be placed on areas of current research, including the formation of OH in reactions of RO 2 with HO 15 2 and on unimolecular reactions of peroxy species. 14,16 Much of this recent work is focused on understanding puzzling discrepancies in the HO x budget in forested regions of the atmosphere, where issues related to unexpectedly high HO x radical cycling rates, high abundances of HO x species, or to unexplained HOx partitioning have recently been encountered [e.g., refs ]. Another important atmospheric puzzle is with respect to understanding the formation of secondary organic aerosol (SOA), [e.g., refs ] the process whereby forestal emissions, as well as some anthropogenic species such as aromatic compounds and large-chain alkanes, are oxidized in the gas-phase to produce functionalized low-volatility species that subsequently contribute to particle formation or growth. The need to understand the budgets of atmospheric SOA demands a detailed understanding of the gas-phase oxidative processes, as even minor (and often multi-generation) by-products of the chemistry can be major SOA sources. SECTION 2: ADVANCES IN THE DETECTION OF PEROXY RADICALS A number of advances have been made in the detection of HO 2 and organic peroxy radicals. Much of the early work on the kinetics of peroxy radicals was done using ultraviolet (UV) absorption to measure the radical concentrations. Tyndall et al. 4 have critically reviewed the spectra of smaller peroxy radicals. Many organic peroxy radicals have similar UV absorption spectra, however, making it difficult to distinguish individual radicals when mixture are present, or when one radical is formed as a reaction product from another. A method of detecting radicals specifically is clearly desirable. 5

6 Direct spectroscopic techniques have been developed for the detection of HO 2 which improve upon earlier methods using tunable diode lasers in the fundamental region. The newer techniques work in the near-ir, where HO 2 has both an overtone vibrational band and a lowenergy electronic transition. Although these two absorption bands are less intense than the fundamentals, the lasers and detectors available in the near-ir are quieter than those in the mid- IR, and by using suitable averaging techniques good signal-to-noise can be obtained. Taatjes and coworkers 23,24 have detected HO 2 in a number of systems relevant to the atmosphere or low-temperature combustion systems. They have predominantly used frequency modulation with second harmonic detection to improve the detection sensitivity. An analogous technique has been used by Sander and co-workers The method of 2-tone frequency modulation, first proposed for the detection of HO 2 by Johnson et al. in 1991, 29 has recently been adapted by Kanno et al. 30,31 to measure the equilibrium constant for the HO 2 -H 2 O complex and kinetics of the HO 2 self-reaction. Finally, Fittschen and co-workers 32,33 have used a variation on the cavity ring-down technique to detect HO 2 radicals. In these experiments a CW infrared laser is coupled into the absorption cell using an acousto-optic modulator, and by recording a ring-down decay over a range of individual wavelengths both spectroscopic and kinetic information can be acquired. The near-ir electronic band present in HO 2 is common to other peroxy radicals as well. Spectra have been reported for a number of peroxy radicals by Miller and co-workers 34 and used for the detection of methyl- and ethylperoxy radical in a time-resolved cavity ringdown system. 35 However, this spectral region has not seen widespread use in kinetics experiments. In a new variation on the cavity ring-down approach, Okumura and coworkers 36 have detected hydroxyl-peroxy radicals in the OH fundamental stretching region. The radicals were 6

7 formed following the isomerization of C4 and C5 alkoxy radicals, followed by addition of O 2 to the resultant hydroxyalkyl radicals. The appeal of this technique is that any hydroxyalkyl peroxy radicals should have a similar absorption, of comparable intensity, to those studied here. Clearly, however, the technique cannot be extended to unsubstituted peroxy radicals. Lee and coworkers have demonstrated that a number of peroxy radicals can be detected by step-scan Fourier Transform Infrared Spectroscopy. Their studies have included CH 3 O 2, 37 CH 3 C(O)O 2, 38 and the related sulphur-containing radical, CH 3 SOO. 39 While the technique inherently measures the time dependence of the absorption, it has so far not been used to produce any kinetic data. Chemical ionization mass spectrometry has also been used extensively to detect organic peroxy radicals in flow tubes For example, Elrod and coworkers have produced some interesting kinetic data in a turbulent flow tube By the use of proton transfer mass spectrometry they were able to detect a variety of alkyl peroxy radicals, and to measure the kinetics of their reactions with NO. They have also shown that it is possible to detect more complex radicals, such as the bicyclic peroxy radicals formed in the oxidation of aromatics The technique is fairly general, and has opened up the possibility of detecting a large number of - radicals, subject to their ability to be formed. In addition to positive ion detection, the use of O 2 and SF - 6 as ionizing agents also permits the detection of various peroxy radicals , However, while Eberhard et al. were able to detect peroxy radicals as their parent ion using O 2 as reagent ion, HO 2 and CH 3 O 2 radicals were detected at masses 140 (SF 4 O - 2 ) and 52 (FO - 2 ) respectively in the study of their cross reaction by Raventos-Duran et al. 57 The use of I - as reagent ion for detection of acylperoxy radicals has also proven useful in laboratory studies, and is now commonly used in studies of ambient air as well (for detection of PAN species, 7

8 RC(O)OONO 2, which are first thermally dissociated in the instrument inlet to the acylperoxy radical). 59 SECTION 3: LABORATORY STUDIES OF THE ATMOSPHERIC CHEMISTRY OF ORGANIC PEROXY RADICALS Section 3.1 REACTIONS OF ORGANIC PEROXY RADICALS WITH NO Reaction with NO represents the dominant loss process for organic peroxy radicals in many parts of the troposphere, including many continental areas (where NO x levels are elevated due to anthropogenic activity) and in the upper troposphere (where NO x is elevated due to lightning). As presented above, the reaction occurs via two channels, the main channel leading to NO 2 and an alkoxy species, RO, and the minor channel leading to organic nitrate formation. RO 2 + NO RO + NO 2 RO 2 + NO RONO 2 (1a) (1b) Rate coefficient data for reactions of NO with representative organic peroxy species are summarized in Table 1; data are generally taken from the reviews of Calvert et al. 7 and from the IUPAC review panel [ 60 for smaller systems, while data for larger systems are taken from the primary literature. Note that for this, and all subsequent tables, these data represent a subset of the measured rate coefficients to illustrate the basic patterns in reactivity. For unsubstituted (alkane-derived) alkylperoxy species, rate coefficients fall in the range (7-10) cm 3 molecule -1 s -1 near 298 K. The rate coefficients display weak negative temperature dependences, and there is no discernible variation in the rate coefficient with size or structure of the R-group. Early reports of lower rate coefficients for larger radicals, obtained via 8

9 time-resolved NO 2 detection following radical production via pulsed radiolysis 61, are likely in error due to the complications from radical cycling and delayed NO 2 production. 7 For example, reaction of (CH 3 ) 3 CCH 2 O 2 with NO leads to the formation of a cascade of smaller peroxy radicals, each of which reacts further with NO: (CH 3 ) 3 CCH 2 O 2 + NO (CH 3 ) 3 CCH 2 O + NO 2 (CH 3 ) 3 CCH 2 O (+ O 2 ) (CH 3 ) 3 CO 2 + CH 2 O (CH 3 ) 3 CO 2 + NO (CH 3 ) 3 CO + NO 2 Rate coefficients for reaction of halogenated alkylperoxy species (generally obtained via abstraction of an H-atom from a fluorocarbon) with NO tend to be larger than is the case for their unsubstituted counterparts, 7 with values ranging from about (9.5-16) cm 3 molecule -1 s -1. The rate coefficients again display a weak negative temperature dependence, see Table 1 for representative data. As a well-studied example [ref. 60, and references therein], consider the CF 3 O 2 radical reaction with NO; the recommended expression, 60 k = (T/298) -1.2 cm 3 molecule -1 s -1, provides rate coefficients that are roughly times faster than those for CH 3 O 2 over the K temperature range. Note also that Patchen et al. 50 studied a series of mono-chlorinated peroxy species, obtained from Cl-initiated oxidation of alkenes, e.g., Cl + CH 2 =CH 2 + M ClCH 2 CH 2 + M ClCH 2 CH 2 + O 2 + M ClCH 2 CH 2 OO + M The radicals were produced in a turbulent flow system, and detected using a proton-transfer chemical ionization mass spectrometer as discussed earlier. Data for the monoalkenes studied were found to be independent of the nature of the alkene (C 2 -C 4 ), with an average rate coefficient of cm 3 molecule -1 s -1. 9

10 Rate coefficient data have also been obtained for reaction of NO with various oxygensubstituted peroxy species (i.e., those containing alcohol, ether, carbonyl and ester functionality). Again, some representative data are collected in Table 1. For the most part, systematic studies have not been carried out, so patterns in reactivity are not always evident. A notable exception is the acylperoxy radicals, RC(O)OO, whose reactions with NO are rather rapid, cm 3 molecule -1 s -1. Miller et al. 49 conducted a systematic study of a number of C 2 -C 5 hydroxy-peroxy species obtained from the OH-initiated oxidation of alkenes, e.g., OH + CH 2 =CH 2 + M HOCH 2 CH 2 + M HOCH 2 CH 2 + O 2 + M HOCH 2 CH 2 OO + M They found the rate coefficient to be independent of size/structure of the peroxy radical, with k cm 3 molecule -1 s -1 in all cases. Among the systems studied was isoprene, which has been the subject of a number of additional investigations [ref. 60; and references therein]. Note that in the addition of OH to isoprene (with two conjugated double bonds), a total of eight distinct hydroxyl-peroxy species are obtained as shown in Figure 1 below: Figure 1: Structures of the eight distinct hydroxy-peroxy radicals generated in the OH-initiated oxidation of isoprene. 10

11 While most measurements conducted to date have been for the atmospherically-relevant distribution of radicals generated in the OH/isoprene/O 2, Ghosh et al. 62 studied a subset of these radicals (the second through fourth radical from the left in the scheme above) via photolysis of an appropriate iodo-substituted isoprene analog. The IUPAC panel 60 recommends a rate coefficient k = cm 3 molecule -1 s -1 for the entire suite of radicals generated from OH addition to isoprene, based largely on the direct study of Miller et al. 49 The data from Ghosh et al. 62 for the subset of radicals generated in their experiments are essentially identical to this value ( cm 3 molecule -1 s -1 ), implying little variation in rate constant among the various hydroxyperoxy isomers. Data are more limited for other functional groups, but (as examples) values near (8-9) cm 3 molecule -1 s -1 have been obtained for the reaction of NO with the peroxy radicals derived from acetone [CH 3 C(O)CH 2 OO] 63 and dimethyl ether [CH 3 OCH 2 OO]. 64 A faster rate coefficient has been reported for the reaction of the peroxy species derived from the cyclic diether, 1,4-dioxane, k = cm 3 molecule -1 s Recently, Elrod 55 reported a rate coefficient for reaction of NO with bicyclic peroxy radicals derived from the oxidation of 1,3,5- trimethylbenzene, that was similar to those observed for simple systems, k = cm 3 molecule -1 s

12 TABLE 1. Summary of representative rate coefficient data for reaction of NO with organic peroxy radicals. (Values given only at 298 K have not been determined as a function of temperature.) RO 2 k (298 K) / cm 3 molecule -1 s -1 A-Factor / (E a /R) / s -1 K Reference Unsubstituted Species CH 3 O CH 3 CH 2 O n-c 3 H 7 O i-c 3 H 7 O t-c 4 H 9 O C 5 H 11 O cyclo-c 5 H 9 O Halogenated Species CF 3 O a CF 2 ClO b CFCl 2 O c ClCH 2 CH 2 O Oxygenated Species CH 3 C(O)O CH 3 CH 2 C(O)O CH 3 C(O)CH 2 O HOCH 2 CH 2 O OH-isoprene-O CH 3 OCH 2 O a Temp. dependence given by (T/298) -1.2 b Temp. dependence given by (T/298) -1.5 c Temp. dependence given by (T/298) -1.3 d For data from refs. [7,60], uncertainties are estimated to be ±(15-25%). Other data represent individual measurements, and are estimated to have ±35% uncertainties. The yields of organic nitrates from reaction of (unsubstituted) alkylperoxy radicals with NO have been studied in detail, primarily by Atkinson and co-workers [e.g., ref. 66, and references therein]. For these unsubstituted cases, clear patterns in nitrate yield with radical size and structure, pressure and temperature are evident. The nitrate yield is seen to increase with decreasing temperature and with increasing pressure, and to increase monotonically with size of the R-group, approaching a limit of about 30% for species containing ten carbons or more. In addition, nitrate yields are thought to be highest for secondary radicals (e.g., CH 3 CH 2 CH(OO )CH 3 ), with yields reduced by a factor of 2-3 for primary (e.g., 12

13 CH 3 CH 2 CH 2 CH 2 OO ) and tertiary (e.g., (CH 3 ) 3 COO ) species of comparable size. Atkinson and co-workers [see ref. 66 for the most recent update] have provided the following relationship for the calculation of the ratio of the nitrate production to radical propagation channels for reactions of secondary alkylperoxy radicals [R-CH(OO )-R] with NO: Here, k (R1b) / k (R1a) = [A/(1+(A/B))] F z A = ( ) exp(n) [M] (T/300), with n the number of carbons in the peroxy radical; B = 0.43 (T/300) -8 ; F=0.41; and z = (1 +{log 10 [A/B]} 2 ) -1 Parameterized nitrate yield data, k Nitrate /k Total, obtained from this formulation are provided in Figure 2, and serve to demonstrate the nitrate yield variations as a function of radical size, temperature and pressure. FIGURE 2: Parameterized nitrate yields k (R1b) /k (R1) from the reaction of NO with various organic alkylperoxy radicals. Parameterizations for secondary alkylperoxy radicals from ref Unsubstitued 2ry Alkylperoxy Radicals, 298 K, 1 atm. Unsubstituted 2ry Alkylperoxy Radicals, 298 K, 100 Torr Unsubstituted 2ry Alkylperoxy Radicals, 248 K, 1 atm. Fractional Nitrate Yield Number of Carbon Atoms 13

14 However, we note that there are some indications that the reduction in nitrate yields for primary and tertiary radicals may be overestimated. 7 Data from Espada et al. 67 show roughly equal nitrate yields ( 7-8%) from reaction of 1-butylperoxy and 2-butylperoxy radicals with NO. Studies of isomers of pentylperoxy radicals 68 show roughly equal yields from both secondary and tertiary radicals. Also, unpublished data from our laboratory [quoted in ref. 7] indicate that tertiary organic nitrates, e.g., CH 3 CH 2 C(ONO 2 )(CH 3 ) 2, are thermally unstable at GC injection temperatures, and previously determined yields may be systematically underestimated. A study of the yields of these tertiary species is encouraged, using a non-gas chromatography based technique if possible. It should also be pointed out that the Arey et al. 66 parameterization just presented is based on data for hydrocarbons containing three carbons or more, and thus some discussion of nitrate production for the smallest alkanes is appropriate. Fairly recent studies of nitrate production from reaction of NO with 2-propylperoxy, ethylperoxy and methylperoxy radicals have been conducted by Elrod and co-workers 45,46,48 (generally at Torr, over a range of temperatures) and by Butkovskaya and co-workers (at pressures ranging from roughly Torr, and as a function of temperature for methyl- and ethylperoxy) In general, the yield data show the expected trends, with nitrate production increasing with size of the radical, with decreasing temperature and with increasing pressure. For isopropy nitrate formation from reaction of NO with 2-propylperoxy, the available yield data 48,69 agree favorably with the Arey et al. 66 parameterization. For reaction of NO with ethylperoxy radicals, 46,70 reported ethyl nitrate yields are generally higher than those given by extrapolation of the Arey et al. parameterization for example, the Butkovskaya et al. data extrapolated to atmospheric pressure gives an ethyl nitrate yield of about 3% compared to 2.1% obtained from the parameterization. Deviations 14

15 between the measured and parameterized yields are larger at lower temperature. Reported methyl nitrate yields are very small, 1% at 298 K, 1 atm. total pressure. 71 However, it should be pointed out that this small yield is still considerably higher (by about a factor of 100) than what has been derived from measurements of CH 3 ONO 2 in the lower stratosphere, where the only source should be the CH 3 O 2 + NO reaction. 71,72 Clearly, confirmatory studies of these small nitrate yields from methane and ethane oxidation and comparisons with atmospheric measurements are warranted. Yields of organic nitrates from reaction of NO with functionalized alkylperoxy radicals are less well studied, although for systems that have been studied nitrate yields are typically lower than those obtained for the simple alkylperoxy radicals. For example, Espada et al. 67 showed that the nitrate yield from reaction of 1-bromo-2-propylperoxy radical BrCH 2 CH(OO )CH 3 with NO was reduced by about a factor of two relative to the 2- propylperoxy case, while the yield for the 1-bromo-3-propylperoxy system (where the Br-atom is separated from the peroxy radical functionality) was similar to that predicted for 1-propylperoxy. Espada and Shepson 73 also report nitrate yields from reaction of NO with peroxy radicals generated from oxidation of ethers and glycol ethers. In general, the data show reduced nitrate yields (compared to alkane-derived radicals of similar size and structure) in cases where the peroxy function is adjacent to an ether linkage, but an increase in nitrate yield when the ether and peroxy radical functionality are separated from each other. Ziemann and co-workers have conducted a systematic study of nitrate production in the OH-initiated oxidation of linear and 2-methyl-substituted alkenes, e.g. : OH + R CH 2 C(R)=CH 2 (+ O 2 ) R CH 2 CR(OO )CH 2 OH R CH 2 CR(OO )CH 2 OH + NO R CH 2 CR(O )CH 2 OH + NO 2 15

16 R CH 2 CR(OO )CH 2 OH + NO R CH 2 CR(ONO 2 )CH 2 OH Note that these hydroxynitrates were detected in the condensed phase, indicating the importance of the formation of these multi-functional species from long-chain alkane oxidation in SOA production. Their studies show that nitrate yields obtained from β-hydroxy-substituted radicals are dependent on the structure of the radical, with higher yields observed for tertiary radicals than for secondary, and for secondary relative to primary. As with the simple alkylperoxy species, yields plateau at high carbon number, with limiting values of 0.25 (tertiary), 0.15 (secondary) and 0.12 (primary). 76 However, it is apparent that the yield for a radical of given size is lower 76,77 than what is seen for the standard alkyl cases. The OH-initiated oxidation of long-chain alkanes leads in part to the formation of 1,4- hydroxyperoxy radicals, following isomerization of an alkoxy species, e.g., OH + RCH 2 CH 2 CH 2 CH 2 CH 3 (+ O 2 ) RCH 2 CH 2 CH 2 CH(OO )CH 3 + H 2 O RCH 2 CH 2 CH 2 CH(OO )CH 3 + NO RCH 2 CH 2 CH 2 CH(O )CH 3 + NO 2 RCH 2 CH 2 CH 2 CH(O )CH 3 (+ O 2 ) RCH(OO )CH 2 CH 2 CH(OH)CH 3 Complementary data from Arey et al. 66 and Lim and Ziemann 78 indicate that yields of nitrates obtained in the reactions of these 1,4-substituted peroxy species with NO are reduced from those obtained from linear alkanes, with a limiting yield of 15% for high carbon number. 78 Data on these hydroxyl-substituted systems are of particular relevance because of the predominant contribution of unsaturated biogenic hydrocarbons (particularly isoprene) to the global VOC budget. For example, it can be estimated that 10% or more of the NO x emitted in the eastern USA is converted into isoprene nitrates (assuming a 8% nitrate yield, see discussion below, and following the calculation of Chen et al.); 79 thus, quantifying the yield of isoprene- 16

17 derived nitrates and their subsequent fate (particularly whether or not they act as permanent NO x sinks) has critical bearing on NO x budgets on regional and even global scales Studies of nitrate formation from isoprene are complicated by the existence of the eight distinct hydroxyperoxy species illustrated in Figure 1; note that these radicals are all β- hydroxyperoxy or 1,4-hydroxyperoxy species, similar to those just described (although there is an additional double bond present in the isoprene-derived species, compared to those derived from simple alkanes and alkenes). There have been numerous measurements of the overall nitrate yield in the reaction of NO with isoprene-derived hydroxyperoxy radicals, using a variety of techniques to quantify the nitrates including FTIR absorption, chemical ionization mass spectrometry, GC-based techniques, and analysis of OH radical time profiles; data vary from about 4-15%. 79,84-89 More recent values fall in the middle of this range, and a consensus value of (8±3)% is now recommended. 60 It should also be noted that Lockwood et al., 89 via the synthesis of three individual isoprene nitrate isomers and additional chromatographic and mechanistic information and deduction, were able to identify and quantify yields of each of the eight isoprene nitrates produced in the OH-initiated oxidation of isoprene. Overall, a hydroxynitrate yield of 7 ( +2.5 / -1.5 ) % was obtained, with the two β-hydroxynitrate species accounting for 80% of the total yield. It is also interesting to note that the Lockwood et al. 89 data imply nitrate yields from individual RO 2 /NO reactions that vary from about 1-15%, although these data likely possess significant uncertainties (especially for minor peroxy species). On the basis of a measured overall isoprene nitrate yield of 12%, the identification of a wide range of products in the OH-initiated oxidation of isoprene using a CIMS technique and mass balance arguments, Paulot et al. 88 were able to draw some conclusions regarding yields of various types of isoprene nitrates from the corresponding RO 2 /NO reaction. Their analysis 17

18 suggested larger yields for the 1,4-hydroxyperoxy / NO reactions (24.7%) compared to those for the β-hydroxyperoxy / NO case (6.7%). Clearly, there exists the likelihood that the nitrate yield is different for each of the eight isoprene-derived hydroxyperoxy radicals 60 and, despite the recent efforts just described, further work to obtain more precise yields of each species and their subsequent fate is of central importance. The study of Paulot et al. 88 also provided estimates for the nitrate yields from reactions of the β-hydroxyperoxy species obtained from the OH-initiated oxidation of methyl vinyl ketone (MVK) and methacrolein (MACR), the two major products of isoprene oxidation: OH + CH 2 =CHC(=O)CH 3 HOCH 2 CH(OO )C(=O)CH 3 OH + CH 2 =C(CH 3 )CH=O HOCH 2 C(CH 3 )(OO )CH=O Their analysis led to nitrate yields of 11% and 15% for these two cases, respectively, significantly higher than yields expected from pentene-derived species. The study of Chuong and Stevens, 90 which is based on observed OH recycling in a discharge flow system, suggests yields of 10±10% for reaction of these two radicals with NO. We conclude this section by noting that very little is known about the nitrate yields from OH-initiated oxidation of more complex biogenic species, such as monoterpenes, which contain not only unsaturated functionality, but also ring structures in some cases. For example, estimates in the literature for production of hydroxynitrates from the OH-initiated oxidation of α-pinene are significantly different. The 18±9% yield reported by Nozière et al. 91 using FT-IR absorption to identify/quantify products may be in error due to interferences from absorption due to other nitrates, while the 1% yield reported by Aschmann et al., 92 obtained using chemical ionization techniques may be an underestimate due to fragmentation of the parent hydroxynitrate upon protonation, or sampling losses. Lee et al. 93 considered the possible formation of organic nitrates 18

19 in the OH-initiated oxidation of a series of terpenes (including α-pinene) and sesquiterpenes, using proton-transfer mass spectrometry to detect the nitrates. They noted that yields of products with even m/z (i.e., those likely containing an N-atom) were low, less than 1% to 2% in all cases, although the possibility of sampling issues and/or fragmentation of the parent ions was also considered possible. Clearly, further studies of these biogenic hydrocarbon systems, and development of standards and calibrated measurement techniques for the nitrate compounds, are required. Section 3.2 REACTIONS OF ORGANIC PEROXY RADICALS WITH NO 2 The reactions of organic peroxy radicals with NO 2 result in almost all cases in the formation of a peroxynitrate species via a third-body process, RO 2 + NO 2 + M ROONO 2 + M (2) Note that the most reasonable bimolecular reaction channel, to form RO and NO 3, is endothermic for standard alkylperoxy radicals. Rate coefficients for reaction of a few simple alkylperoxy species with NO 2 have been determined and representative data are summarized in Table 2. The rate coefficient data are adequately described by the Troe-type expression for third-body reactions, 60 k = k o [M] / {1+(k o [M] / k } F c z ; where z = {1+[log 10 (k o [M] / k )] 2 } -1 ; k o is the (third-order) limiting rate coefficient at low pressure, and varies with temperature, k o = k o (T/300) -m ; k is the (second-order) limiting rate coefficient at infinite pressure, and also varies with temperature, k = k (T/300) -n ; and F c is the broadening-factor. Typical high pressure limits are near cm 3 molecule -1 s -1 and rate coefficients near 298 K and 1 atm total pressure approach this value (e.g., for the C 2 H 5 O 2 + NO 2 19

20 reaction, k 1 atm., 298 K = cm 3 molecule -1 s -1, k = cm 3 molecule -1 s -1 ). 60 Interestingly, the rate coefficient typically remains fairly constant throughout the troposphere, due to compensating effects of decreasing temperature and pressure. The alkylperoxy nitrates formed in reaction (2) are bound by about kcal/mole, and are thermally unstable in the lower troposphere. For example, the rate coefficient for decomposition of C 2 H 5 OONO 2 is about 4 s -1 at 298 K, 1 atm. pressure: C 2 H 5 OONO 2 + M C 2 H 5 O 2 + NO 2 + M Thus, these species serve only as temporary reservoirs for reactive NO x and HO x radicals under these conditions. However, this picture changes dramatically in the colder conditions of the upper troposphere, 94 or in polar regions. For example, the lifetime of C 2 H 5 OONO 2 against thermal decomposition increases to many hours / many days for typical conditions near 10 / 15 km. Here, additional loss processes (including photolysis and reaction with OH) can become significant. In polar regions, near the Earth s surface, depositional losses may also be of significance. Kinetic data for reactions of NO 2 with functionalized (oxygenated, halogenated) peroxy species are rather limited, but most additional studies report rate coefficients in the range (5-10) cm 3 molecule -1 s -1 under atmospheric conditions, similar to the values obtained for standard alkylperoxy species. A few examples are provided in Table 2 below. The thermal stability of the functionalized alkylperoxy nitrates has been considered in works by Zabel and coworkers [e.g., refs ], who demonstrated that the stability of the ROONO 2 species is controlled by inductive effects associated with the nature of the R-group. Thus, peroxynitrates derived from acetone [CH 3 C(O)CH 2 OONO 2 ] and dimethyl ether [CH 3 OCH 2 OONO 2 ] have similar bond strengths as the standard alkyl peroxynitrates, while halogenated methyl 20

21 peroxynitrates [e.g., CF 2 ClOONO 2 ] are significantly longer lived than their non-functionalized alkylperoxy counterparts. Some representative data are presented in Table 3. The reaction of NO 2 with acylperoxy radicals, RC(O)OO, represents something of an extreme case, as the peroxynitrates (referred to as PANs, peroxyacyl nitrates) formed in these cases are more stable than their alkylperoxy nitrate counterparts: RC(O)O 2 + NO 2 + M RC(O)OONO 2 + M First, formation rate coefficients for this class of reactions are roughly similar to those for the alkylperoxy cases, with k 298K/1 atm cm 3 molecule -1 s -1 for the reaction of NO 2 with CH 3 C(O)OO. 60 More importantly, however, the O-N bond tends to be considerably stronger for these species, about 28 kcal/mole, and lifetimes for thermal decomposition increase accordingly for peroxyacetyl nitrate (CH 3 C(O)OONO 2, PAN), the data suggest lifetimes of 40 min near the surface, increasing to many months near 6 km and many centuries in the upper troposphere. Again, photolysis 98 and depositional processes become important losses at reduced temperatures. Because of their increased lifetime, these PAN species play a well-known role in the redistribution of NO x on regional and global scales. 21

22 TABLE 2. Summary of rate coefficient data for reaction of NO 2 with organic peroxy radicals. (Values given only at 298 K have not been determined as a function of temperature / pressure.) RO k o (cm 6 m k (cm 3 molec -1 s -1 ) n F c k (298 K, 1 atm) Reference molec -2 s -1 ) ALKYL PEROXY SYSTEMS CH 3 O CH 3 CH 2 O CH 3 C(O)CH 2 O cyclo-c 6 H 11 O CF 3 CH 2 OCH(O 2 )CF cyclo-c 3 H 5 O 3 -O CH 3 OCH 2 O CF 3 CH 2 O CCl 3 CH 2 O CF 3 O ACYL PEROXY SYSTEMS CH 3 C(O)O CF 3 C(O)O d Uncertainties for data from ref. [60] are estimated to be ±(20-30%) at 298 K. Other data represent individual measurements, and are estimated to have ±40% uncertainties. TABLE 3. Summary of rate coefficient data for thermal decomposition of RO 2 NO 2 species. (Values given only at 298 K have not been determined as a fn. of temperature.) Species k o (cm 3 molec -1 s -1 ) k (s -1 ) F c k (298 K, 1 atm), s -1 Reference ALKYL PEROXY SYSTEMS CH 3 O exp(-9690/t) exp(-10560/t) CH 3 CH 2 O exp(-9285/t) exp(-10440/t) CH 3 C(O)CH 2 O exp(-10730/t) 3 60 CH 3 OCH 2 O CF 2 ClO exp(-10500/t) exp(-11990/t) ,96 CFCl 2 O exp(-10860/t) exp(-12240/t) ,96 ACYL PEROXY SYSTEMS CH 3 C(O)O exp(-12100/t) exp(-13830/t) C 2 H 5 C(O)O exp(-11280/t) exp(-13940/t) CH 2 =C(CH 3 )C(O)O exp(-13472/t) a a Data obtained at 1 atm. total pressure, and are probably close to the high pressure limit. 60 b Typical uncertainties are estimated to be ±50% to a factor of two for all data given at 298 K, 1 atm. 22

23 While the formation of peroxynitrates in reactions of RO 2 with NO 2 is nearly universal under atmospheric conditions, there are isolated cases in the literature regarding the formation of NO 3 as a product of this chemistry. For example, Orlando and Tyndall 106 provided evidence that this was the case for the HC(O)C(O)OO radical, formed in the oxidation of glyoxal: OH + HC(O)CHO HC(O)CO + H 2 O HC(O)CO + M HCO + CO + M HC(O)CO + O 2 + M HC(O)C(O)OO + M HC(O)C(O)OO + NO 2 HC(O)C(O)O + NO 3 HC(O)C(O)O HCO + CO 2 In addition, there is recent evidence 107 that NO 3 production also occurs from reaction of NO 2 with the peroxy radical formed during the Cl-atom initiated oxidation of methacrolein, Cl + CH 2 =C(CH 3 )CHO (+ O 2 ) ClCH 2 C(CH 3 )(OO )CHO ClCH 2 C(CH 3 )(OO )CHO + NO 2 ClCH 2 C(CH 3 )(OONO 2 )CHO ClCH 2 C(CH 3 )(OO )CHO + NO 2 ClCH 2 C(CH 3 )(O )CHO + NO 3 Although too few examples are available to draw firm conclusions, the formation of the NO 3 in these reactions appears correlated with the formation of a thermally unstable alkoxy radical for example, in the glyoxal case above, the loss of CO 2 from the HC(O)C(O)O radical is likely very favorable. Section 3.3 REACTIONS OF ORGANIC PEROXY RADICALS WITH NO 3 Reactions of organic peroxy radicals with NO 3 can be of significance in the nighttime troposphere. 108 Here, NO 3 radicals build up through reaction of NO 2 with O 3, and often exist in equilibrium with N 2 O 5 [e.g., ref. 109]: 23

24 NO 2 + O 3 NO 3 + O 2 NO 3 + NO 2 + M N 2 O 5 + M Peroxy radicals then can be formed via reaction of the resultant NO 3 radicals with VOCs. These reactions can occur (usually rather slowly) via abstraction, and more rapidly via addition of the NO 3 to the VOC, leading in both cases to RO 2 production, e.g., NO 3 + CH 3 CHO (+ O 2 ) HNO 3 + CH 3 C(O)OO NO 3 + CH 2 =CH 2 (+ O 2 ) O 2 NOCH 2 CH 2 OO The fate of the resultant RO 2 radicals can be varied but may include reaction with NO 3, with an alkoxy radical and NO 2 as the dominant products, e.g.: NO 3 + RO 2 NO 2 + RO + O 2 As summarized by Vaughan et al., 108 rate coefficient data for reaction of NO 3 with RO 2 radicals are only available for select cases; the available data are presented in Table 4. These reactions are very difficult to study. Since they involve two reactive radicals, the possibility of consumption of either species by unwanted side-reactions is high. Earlier studies, were typically conducted with [NO 3 ] > [RO 2 ], allowing for the possibility of complications due to reactions of NO 3 with either RO or HO 2 radicals. 108 The study of Vaughan et al. 108 employed a more sensitive detection method for NO 3 (cavity-enhanced absorption), allowing for experiments to be conducted with [NO 3 ] < [RO 2 ]; however, even in this case, RO 2 concentrations needed to be kept low to avoid consumption via self-reaction and related processes, changes in [NO 3 ] in the flow tube were thus small, and kinetic modeling of the system was required to retrieve the rate coefficient. Despite these limitations, the generally very good agreement for simple systems (CH 3 O 2 and C 2 H 5 O 2 ) is encouraging. 24

25 TABLE 4. Rate coefficients for reaction of NO 3 with organic peroxy radicals at 298 K. Source PEROXY RADICAL k (298 K) / cm 3 molecule -1 s -1 CH 3 O ± 0.6 a C 2 H 5 O ± 0.6 b CH 3 C(O)O ± 1.4 ( 0.15) cyclo-c 6 H 11 O ± cyclo-c 5 H 9 O ( / -0.5 ) 108 CH 2 FO ± CH 2 ClO ( / -2.6 ) 108 CF 3 O ± (CF 3 ) 2 CFO ± a Average of determinations from refs. 108, 110, 111. Quoted uncertainty is based on uncertainties reported by the original investigators. b Average of determinations by 108, 112, 113. Quoted uncertainty is based on uncertainties reported by the original investigators. Vaughan et al. 108 have considered the rate coefficients for the various peroxy radicals studied in terms of frontier orbital theory. That discussion will not be repeated here, but it is clear that the presence of electron-withdrawing groups (e.g., F-atoms, CF 3 ) leads to a significant decrease in the measured rate coefficient. SECTION 3.4: REACTIONS OF ORGANIC PEROXY RADICALS WITH HO 2 : Reactions of organic peroxy radicals with HO 2 are of central importance in the atmosphere, as they serve as sinks for HO x radicals and terminators for ozone-generating chain reactions, e.g., RO 2 + HO 2 ROOH + O 2 In fact, except for high NO x (typically urban) areas, these reactions represent the major loss for atmospheric HO x. These reactions typically possess rate coefficients on the order of cm 3 molecule -1 s -1, and occur with significant negative temperature dependences. As will be 25

26 discussed in detail toward the end of this section, this can be attributed to the formation of a short-lived complex during the course of the reaction. Rate coefficients are available for reaction of HO 2 with a variety of simple alkyl peroxy radicals, as well as with numerous halogen- and oxygen-functionalized peroxy species. A summary of representative rate coefficients, taken primarily from previous reviews, 4,7,60 is presented in Table 5. As originally pointed out by Boyd et al., 116 rate coefficients near 298 K for standard alkyl peroxy species (and hydroxyl-substituted species as well) increase with the number of carbon atoms; Calvert et al. 7 used the data for unsubstituted alkyl peroxy species from Table 5 to derive the following updated function to describe this rate coefficient increase: k = {21.3 [1-exp( n)]} cm 3 molecule -1 s -1, where n is the number of carbons in the alkyl group. Note that the rate coefficient for reaction of HO 2 with the atmospherically-relevant distribution of isoprene-derived peroxy species was determined as part of the Boyd et al. study; 116 the measured rate coefficient, k = cm 3 molecule -1 s -1, is well described by this function. 26

27 TABLE 5. Rate coefficients for reaction of HO 2 with representative organic peroxy radicals. (Values given only at 298 K have not been determined as a fn. of temperature.) PEROXY RADICAL k (298 K) (cm 3 molec -1 s -1 ) A (cm 3 molec -1 s -1 ) E a /R (K) Source CH 3 O CH 3 CH 2 O CH 3 C(O)CH 2 O HOCH 2 CH 2 O (CH 3 ) 2 C(OH)CH 2 O , 116 CH 2 ClO CF 2 ClO CF 3 CF 2 O CF 2 ClCH 2 O cyclo-c 6 H 11 O C 10 H 21 O ,116 C 14 H 29 O ,116 HO-C 5 H 8 -O HO-α-pinene-O CH 3 C(O)O a Typical uncertainties range from ±30-35% for unsubstituted species to ± (factor of 2) for halogenated and biogenic species. Although the dataset is limited, it appears that the rate coefficients are modified by the presence of F-atoms at the peroxy-containing carbon, as the rate coefficient for reaction of HO 2 with CF 3 CF 2 O 2 is significantly smaller than with ethylperoxy or CX 3 CH 2 O 2 species (X=Cl, F). 60 Note also that reaction of HO 2 with acylperoxy radicals, RC(O)O 2, tend to be faster than reaction with their standard alkylperoxy counterparts. 60 Further studies of rate coefficients for reactions of HO 2 with other functionalized peroxy species (e.g., those derived from ethers, larger carbonyl species, organic acids, esters) would be useful in completing this dataset and assessing patterns in reactivity. Products obtained from reaction of HO 2 with representative peroxy radicals are summarized in Table 6. It is now well established that in the case of simple alkylperoxy radicals, the reaction generates predominantly/exclusively a chain-terminating hydroperoxide species, [e.g., refs. 57,58,117,118] 27

28 C 2 H 5 O 2 + HO 2 C 2 H 5 OOH + O 2 However, current research is heavily focused on studying specialized cases where OH production, and hence radical propagation, is occurring. It has been known for some time that acylperoxy radicals react with HO 2 to from peroxyacids and organic acids (with O 3 as a coproduct of the second channel): CH 3 C(O)OO + HO 2 CH 3 C(O)OOH + O 2 CH 3 C(O)OO + HO 2 CH 3 C(O)OH + O 3 (3a) (3b) However, Wallington and co-workers [e.g., refs. 119,120] provided evidence for significant formation of OH in the reaction of HO 2 with a series of perfluorinated acylperoxy radicals, C x F 2x+1 C(O)O 2, x=1-4. Further, Hasson et al. 15 showed that the yields of these acid/peracid products were insufficient to account for all acetylperoxy radicals consumed in their chamber experiments and, via FTIR- and HPLC-based end-product analysis in the presence of varying ratios of [CH 3 C(O)OO]:[HO 2 ], were able to demonstrate that a third channel was operative in the reaction: CH 3 C(O)OO + HO 2 CH 3 C(O)O + OH + O 2 (3c) Despite an initial report to the contrary, 121 additional studies have now confirmed that the production of OH in the reaction of acetylperoxy (and other straight-chain acylperoxy species) leads to significant OH production, see Table 6. Although certainly not a ubiquitous process, production of OH also appears to occur in reaction of HO 2 with other types of peroxy species - for example, for α-carbonyl peroxy species; 15, CH 3 C(O)CH 2 O 2 + HO 2 CH 3 C(O)CH 2 OOH + O 2 28

29 CH 3 C(O)CH 2 O 2 + HO 2 CH 3 C(O)CH 2 O + OH + O 2 Additional α-carbonyl peroxy species studied, and the reported OH yields, are summarized in Table 6. Although the dataset is limited, there is a suggestion that the degree of substitution at the peroxy radical carbon influences the OH yield, with more complex species generating larger OH yields. 124 Significant OH generation (20% and 19% yield, respectively) in the reaction of both HOCH 2 OO and CH 3 OCH 2 OO with HO 2 has also been reported. 122,126 Theoretical studies of RO 2 / HO 2 reactions have provided insight into the OH production mechanism, and explained why it is only operative in select cases. In essence, formation of the hydroperoxide (or peracid in the case of acylperoxy species) is thought to occur on a triplet surface via a hydrogen-bonded complex, ROO HOO, which then proceeds through a transition state (located lower in energy than the reactants) to products, ROOH and O 2. On the other hand, OH production appears to originate from a singlet tetroxide intermediate, ROO OOH. In the case of simple alkyl systems (e.g., R=CH 3 ) at atmospherically relevant temperatures, there are no readily accessible product channels for the tetroxide and decomposition back to reactants is its dominant fate. However, hydrogen bonding (involving the -OOH hydrogen and the carbonyl groups in the acylperoxy or α-carbonylperoxy radical, via 7- or 8-membered rings) lowers the energy barriers to tetroxide decomposition, allowing for decomposition to RO + O 2 + OH products and, in the case of the acylperoxy radicals the organic acid and O In a series of experiments conducted predominantly in the 1990 s Wallington and coworkers, using FTIR end-product analysis in environmental chamber studies, reported the formation of carbonyl species in some haloalkylperoxy / HO 2 reactions, e.g, CH 2 FO 2 + HO 2 CH 2 FOOH + O 2 29