A new mechanism for regional atmospheric chemistry modeling

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 02, NO. D22, PAGES 25,84-25,89, NOVEMBER 2, 99 A new mechanism for regional atmospheric chemistry modeling William R. Stockwell, Frank Kirchner, and Michael Kuhn Fraunhofer Institute for Atmospheric Environmental Research (IFU), Garmisch-Partenkirchen, Germany Stephan Seefeld Swiss Federal Institute for Environmental Science and Technology (EAWAG), ETH Ztirich, Switzerland Abstract. A new gas-phase chemical mechanism for the modeling of regional atmospheri chemistry, the "Regional Atmospheric Chemistry Mechanism" (RACM) is presented. The mechanism is intended to be valid for remote to polluted conditions and from the Earth's surface through the upper troposphere. The RACM mechanism is based upon the earlier Regional Acid Deposition Model, version 2 (RADM2) mechanism [Stockwell et al., 990] and the more detailed Euro-RADM mechanism [Stockwell and Kley, 994]. The RACM mechanism includes rate constants and product yields from the most recent laboratory measurements, and it has been tested against environmental chamber data. A new condensed reaction mechanism is included for biogenic compounds: isoprene, a-pinene, and d-limonene. The branching ratios for alkane decay were reevaluated, and in the revised mechanism the aldehyde to ketone ratios were significantly reduced. The relatively large amounts of nitrates resulting from the reactions of unbranched alkenes with NO 3 are now included, and the production of HO from the ozonolysis of alkenes has a much greater yield. The aromatic chemistry has been revised through the use of new laboratory data. The yield of cresol production from aromatics was reduced, while the reactions of HO, NO3, and 03 with unsaturated dicarbonyl species and unsaturated peroxynitrate are now included in the RACM mechanism. The peroxyacetyl nitrate chemistry and the organic peroxy radical-peroxy radical reactions were revised, and organic peroxy radical + NO 3 reactions were added.. Introduction The gas-phase chemical mechanism is one of the most important components of an atmospheric chemistry model. These models require high-quality gas-phase chemical mechanisms to calculate the concentrations of atmospheric chemical species. The concentrations of ozone and other air pollutants are determined by the emissions of nitrogen oxides and reactive organic species, gas- and aqueous-phase chemical reaction rates, deposition, and meteorological conditions. There are several important mechanisms which are widely used for modeling the chemistry of the troposphere including the mechanism of Lurmann et al. [986], the carbon bond IV mechanism of Gery et al. [989], and the mechanism for the Regional Acid Deposition Model (RADM2) [Stockwell et al., 990]. For example, the RADM2 mechanism is used in many photochemical transport/transformation atmospheric chemistry models to predict concentrations of oxidants and other air pollutants [Chang et al., 99; Hass et al., 995; Vogel et al., 995]. Since these mechanisms have been published, much new laboratory work has become available [Le Bras, 99; DeMote et al., 994;Atkinson et al., 992b;Atkinson, 994]. The purpose of this paper is to present a completely revised version of the mechanism of Stockwell et al. [990] and to show the effect of the revisions on calculated chemical concentrations. Since the purpose of the new mechanism is to describe atmospheric Copyright 99 by the American Geophysical Union. Paper number 9JD /9/9JD chemistry on a regional scale, we have named it the "Regional Atmospheric Chemistry Mechanism" (RACM). The RACM mechanism was created to be capable of simulating the troposphere from the Earth's surface through the upper troposphere and to be valid for simulating remote to polluted urban con- ditions. The mechanism includes stable inorganic species, 4 inorganic intermediates, 32 stable organic species (4 of these are primarily of biogenic origin), and 24 organic intermediates (Table ). The RACM mechanism includes 23 reactions (Table 2). The mechanism and its use are described in the subsequent text. The text is divided into section 2 on the inorganic chemistry and section 3 on the organic chemistry. The organic chemistry section includes a description of the aggregation procedures for emissions and the development of the chemistry for alkanes, carbonyls, alkenes, aromatics; the decomposition products of aromatics; and the primarily biogenic species, isoprene, a-pinene, and d-limonene. The oxidation mechanism for isoprene, a-pinene, and d-limonene is much more detailed and realistic than that in the RADM2 mechanism [Stockwell et al., 990]. The aggregation factors for the organic emissions are given in Table 3. Sections 4 and 5 present comparisons of the mechanism with environmental chamber data and with the RADM2 mechanism, respectively. The treatment of peroxy radical reactions were described by Kirchner and Stockwell [996a]. 2. Inorganic Chemistry Tropospheric inorganic chemistry is relatively well known. During the day the chemistry is driven by the photolysis of 25,84

2 25,848 STOCKWELL ET AL.: REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM Table. RACM Mechanism Species List Carbon Molecular No. Species Definition Number Weight Oxidants 03 2 H202 Nitrogenous compounds 3 NO 4 NO 2 5 NO 3 6 N20 5 HONO 8 HNO 3 9 HNO 4 Sulfur compounds 0 SO2 SULF Carbon oxides 2 CO 3 CO 2 4 N H20 H O3p OlD Odd hydrogen HO HO 2 Alkanes CH 4 ETH HC3 HC5 HC8 Alkenes ETE OLT OLI DIEN Stable biogenic alkenes ISO API LIM Aromatics TOL XYL CSL Carbonyls HCHO ALD KET GLY MGLY DCB MACR UDD HKET ozone hydrogen peroxide nitric oxide nitrogen dioxide nitrogen trioxide dinitrogen pentoxide nitrous acid nitric acid pernitric acid sulfur dioxide sulfuric acid Stable Inorganic Compounds carbon monoxide carbon dioxide nitrogen oxygen water hydrogen Abundant Stable Species Inorganic Short-Lived Intermediates ground state oxygen atom, O(3p) excited state oxygen atom, O(D) hydroxy radical hydroperoxy radical Stable Organic Compounds methane.0 ethane 2.0 alkanes, alcohols, esters, and alkynes with HO rate 2.9 constant (298 K, atm) less than 3.4 x 0-2 cm 3 S - alkanes, alcohols, esters, and alkynes with HO rate 4.8 constant (298 K, atm) between 3.4 x 0-2 and 6.8 x 0-2 cm 3 S - alkanes, alcohols, esters, and alkynes with HO rate.9 constant (298 K, atm) greater than 6.8 x 0-2 cm 3 s- ethene 2.0 terminal alkenes 3.8 internal alkenes 5.0 butadiene and other anthropogenic dienes 4.0 isoprene 5.0 a-pinene and other cyclic terpenes with one double 0.0 bond d-limonene and other cyclic diene-terpenes 0.0 toluene and less reactive aromatics xylene and more reactive aromatics cresol and other hydroxy substituted aromatics formaldehyde acetaldehyde and higher aldehydes ketones glyoxal methylglyoxal and other a-carbonyl aldehydes unsaturated dicarbonyls methacrolein and other unsaturated monoaldehydes unsaturate dihydroxy dicarbonyl hydroxy ketone

3 STOCKWELL ET AL.' REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM 25,849 Table. (continued) Carbon Molecular No. Species Definition Number Weight Organic nitrogen 46 ONIT 4 PAN 48 TPAN Organic peroxides 49 OP 50 OP2 5 PAA Organic acids 52 ORAl 53 ORA2 Peroxy radicals from alkanes 54 MO 2 55 ETHP 56 HC3P 5 HC5P 58 HC8P Peroxy radicals from alkenes 59 ETEP 60 OLTP 6 OLIP Peroxy radicals from biogenic alkenes 62 ISOP 63 APIP 64 LIMP Radicals produced from aromatics 65 PHO 66 ADDT 6 ADDX 68 ADDC organic nitrate 4.0 peroxyacetyl nitrate and higher saturated PANs 2.0 unsaturated PANs 4.0 methyl hydrogen peroxide higher organic peroxides peroxyacetic acid and higher analogs formic acid acetic acid and higher acids Organic Short-Lived Intermediates methyl peroxy radical peroxy radical formed from ETH peroxy radical formed from HC3 peroxy radical formed from HC5 peroxy radical formed from HC8 peroxy radicals formed from ETE peroxy radicals formed from OLT peroxy radicals formed from OLI peroxy radicals formed from ISO and DIEN peroxy radicals formed from API peroxy radicals formed from LIM phenoxy radical and similar radicals aromatic-ho adduct from TOL aromatic-ho adduct from XYL aromatic-ho adduct from CSL TOLP 0 XYLP CSLP Peroxy radicals with carbonyl groups 2 ACO 3 3 TCO 3 4 KETP Other peroxy radicals 5 OLNN 6 OLND XO2 peroxy radicals formed from TOL peroxy radicals formed from XYL peroxy radicals formed from CSL acetyl peroxy and higher saturated acyl peroxy radicals unsatured acyl peroxy radicals peroxy radicals formed from KET NO3-alkene adduct reacting to form carbonitrates + HO NO3-akene adduct reacting via decomposition accounts for additional NO to NO 2 conversions... ozone and nitrogen dioxide [Warneck, 988]. The photolysis of ozone produces excited oxygen atoms, O( D), and a fraction of these react with H20 to produce the hydroxyl radical (HO). HO reacts with inorganic and organic species to oxidize them. Many of these reactions produce HO 2 or organic peroxy radicals which either react with NO to convert it to NO 2 or (under low NOx conditions) react to produce hydroperoxides. The conversion of NO to NO 2 and the subsequent photolysis of NO 2 produces more ozone. Ozone and NO 2 react to form nitrate radical (NO3) which is a very important reactive species during the nighttime [Morris and Niki, 94; Japar and Niki, 95; Platt et al., 980]. The RACM mechanism has a reasonably complete set of inorganic reactions. The inorganic rate constants were set to the values recommended by DeMore et al. [994], and most of these changes from RADM2 were relatively small. The cross sections and quantum yields for the photolysis of the inorganic species are taken from DeMore et al. [994]. For NO3, DeMore et al. [994] recommended the data from the review of Wayne et al. [99]. Comparing the revised photolysis frequencies to the previous RADM2 values, there is an increase of 5-20% in the photolysis frequencies for HONO and a similar percent increase for both NO3 photolysis channels. Larger relative changes occur for the relatively low photolysis frequencies of HNO3 and HO2NO 2. The photolysis frequencies for NO 2 and 03 remained nearly unchanged after the updates. Several inorganic reactions were added to the RACM mech-

4 25,850 STOCKWELL ET AL.: REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM Table 2a. The RACM Mechanism: Photolysis Reactions Reaction No. Photolysis Frequency, a Reaction s- Cross Section Quantum Yield (R) (R2) (R3) (R4) (RS) (R6) (R) (R8) (R9) (Ri0) (Rll) (R2) (R3) (R4) (R5) (R6) (R) (R8) (R9) (R20) (R2) (R22) (R23) NO 2 --> O3p + NO 03 --> O D > O3p + 02 HONO --> HO + NO HNO3 --> HO + NO 2 HNO4 ---> 0.65 HO 2 q NO 2 q HO NO 3 NO 3 --> NO NO 3 --> NO 2 + O3p H202 HO + HO HCHO --> H 2 + CO HCHO --> 2 HO 2 + CO ALD- MO 2 + HO 2 + CO OP -- HCHO + HO 2 + HO OP2--> ALD + HO 2 + HO PAA --> MO 2 + HO KET --> ETHP + ACO 3 GLY --> 0.3 HCHO +.8 CO H 2 GLY HCHO +.55 CO HO 2 q- 0.5 H 2 MGLY --> CO + HO 2 q- ACO 3 DCB--> TCO 3 q- HO 2 ONIT --> 0.20 ALD KET + HO 2 q- NO2 MACR --> CO + HCHO + HO 2 q- ACO 3 HKET --> HCHO + HO 2 + ACO 3.50 x 0-3 DeMore et al. [994].62 x 0 -s DeMor et al. [994] 4. x 0-4 DeMor et al. [994].63 x 0-3 DeMor et al. [994] 4.50 x 0 - DeMore et al. [994] 3. x 0-6 DeMore et al. [994] 2.33 x 0-2 Waynet al. [99].8 x 0 - Waynet al. [99] 6.00 X 0-6 DeMore et al. [994] 3.50 x 0 -s wavelengths <300 nm: Moongat et al. [980] wavelengths >300 nm: Cantrell et al. [990] 2. x 0 -s same references as (Ri0) 3.6 x 0-6 Martinez et al. [992] 4. x 0-6 DeMore et al. [994] 4. x 0-6 same as (R3).5 x 0-6 H20 2 cros sections scaled by 0.28; Giguere and Olmos [956] 6.6 x 0 - Martinez et al. [992] 5.83 x 0 -s Atkinson et al. [992b] 2.00 x 0 -s Atkinson et al. [992b] DeMore et al. [994] DeMore et al. [994] total yield of O D and O3p assumed to be unity DeMore et al. [994] assumed to be unity assumed to be unity Wayne et al. [99] Wayne et al. [99] assumed to be unity wavelengths <300 nm: Atkinson et al. [994] b wavelengths >300 nm: DeMor et al. [994] b same references as (R0) b Atkinson [994] b DeMore et al. [994] same as (R3) assumed to be unity Atkinson [994] wavelengths <340:0.0 wavelengths >340:0.029 Atkinson et al. [992b] wavelengths <340:0.4 wavelengths >340:0.0 Atkinson et al. [992b] 9.33 x 0 -s Atkinson [994] and estimated from Koch and Staffelbach et al. [995] 4.33 x 0-4 Stockwell et al. [990] Moongat [996] Stockwell et al. [990] 2. x 0-6 assumed to be mixture of Atkinson [994] 20% n-propyl nitrate and 80% i-propyl nitrate; Atkinson [994].33 x 0-6 Gardner et al. [98] assumed equal to acrolein 6.6 x 0 - same as (R6) Gardner et al. [98] same as (R6) atypical photolysis frequencies are given for solar zenith angle 40 ø, June 2, summer surface 40 ø northern latitude. bpressure dependences given by >(X, P) = >(X, atm)/{ >(X, atm) + [ - >(X, atm)] x P}, where P is in atmospheres. anism. The reaction of O(3p) with nitrogen oxides, especially The rate constant for (5) at 298 K is 6. x 0 - s cm -3 S - the reaction of O(3p) with NO2, can be important in the upper [DeMore et al., 994] which is about equal to the rate constant atmosphere [Warneck, 988]: O(3p) + NO2 --> NO + 02 () for the HO + CH 4 reaction, 6.9 x 0 - s cm -3 s - [Atkinson, 994]. If the surface concentrations for H 2 and CH 4 are assumed to be 500 and 00 ppbv, respectively, then the rate of O(3p) + NO2(+ M) - NO3(+ M) (2) the reaction of HO with H 2 is about 30% that of the HO + O(3p) + NO(+ M) - NO2(+ M) (3) CH 4 reaction. For completeness, the reaction of HO with nitrous acid (6) In the original RADM2 mechanism the reaction of O(3p) with was added. The rate constant for this reaction is 4.86 x 0-2 NO 2 produced only NO; however, O(3p) can also add to NO 2 cm -3 s - at 298 K. The rate of the HONO + HO is about to produce NO 3 [DeMote et al., 994]. Both reactions are now included although rapid photolysis of NO 3 would be expected to limit the importance of (2). 2.4% of the HONO photolysis reaction if an HO concentration of I x 0 cm -3 and a HONO photolysis frequency of 2 x 0-3 s are assumed: The reaction of O(3p) with ozone was added because it is important in the upper atmosphere [Warneck, 988]: HO + HONO- NO2 4- H20 (6) O(3p) (4) The reaction of HO with H 2 is a significant loss reaction for HO in the remote atmosphere: HO + H 2 + (02)--> HO2 + H20 (s) The reactions of NO 3 are very important for nighttime atmospheric chemistry. The NO 3 self-reaction () was added for atmospheric conditions with very high NOx and 03 concentrations during the nighttime. For example, for NO 3 concentration of 300 parts per trillion (ppt) [Platt et al., 980] and a NO

5 STOCKWELL ET AL.: REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM 25,85 Table 2b. The RACM Mechanism Reaction No. Reaction A, cm 3 s-! E/R, K a Note (R24) (R25) (R26) (R2) (R28) (R29) (R30) (R3) (R32) (R33) (R34) (R35) (R36) (R3) (R) (R39) (R40) (R4) (R42) (R43) (R44) (R45) (R46) (R4) (R48) (R49) (R50) (R5) (R52) (R53) (R54) (R55) (R56) (R5) (R58) Inorganic Reactions O3p > 0 3 O3p + 03 '--> 2 02 OlD + N 2 --> O3p + N 2 OlD > O3p OlD + H20--> HO + HO 03+HO-->HO HO2-->HO HO + HO2 --> H H202 + HO---> HO2 + H20 HO2 + HO2-* H HO 2 + HO2 + H20---> H H20 O3p + NO ---> NO 2 O3p + NO 2 ---> NO + 02 O3p + NO 2 --> NO 3 HO + NO ---> HONO HO + NO 2 --> HNO 3 HO + NO 3 ---> NO 2 + HO2 HO 2 + NO--> NO 2 + HO HO 2 + NO 2 ---> HNO 4 HNO4 --> HO2 + NO2 HO2 + NO3 --> 0.3 HNO NO HO + 02 HO + HONO---> NO2 + H20 HO + HNO3 '-> NO3 + H20 HO + HNO 4 ---> NO q- H NO--> NO NO2 ---> NO NO + NO > NO2 + NO2 NO 3 + NO-, NO 2 + NO 2 NO3 + NO2 --> NO + NO NO 3 + NO 2 --> N205 N205 --> NO 2 + NO 3 NO 3 + NO 3'->NO 2 + NO HO + H 2 ---> H20 + HO 2 HO + SO2 --> SULF + HO 2 CO + HO ---> HO 2 + CO 2 Table 2f 8.00 x x x x x x X X 0-2 Table 2f Table 2f Table 2d 6.50 x 0-2 Table 2d Table 2d Table 2d 2.20 x x 0-2 Table 2d Table 2e 3.50 x x 0 - Table 2f.30 x x x x 0-39!.50 x X 0-4 Table 2d Table 2e 8.50 X X 0-2 Table 2d Table 2f x x x X X 0 -lø 6.83 x x 0-5. X 0 -ø.0 X X x x x x x X x x! x x x x x x 0-!2.82 x x x 0-3s 2.65 x x x x x x x X 0-3,2,3 4 (R59) (R60) O P + Organic Compounds ISO + O3p --> 0.86 OLT HCHO HO CO DCB HO XO2 MACR + O3p ---> ALD 6.00 x x x x 0 - (R6) (R62) (R63) (R64) (R65) (R66) (R6) (R68) (R69) (R0) (R) (R2) (R3) (R4) (R5) (R6) (R) (R8) (R9) (R80) (R8) (R82) (R83) (R84) (R85) HO + Organic Compounds CH 4 + HO---> MO 2 + H2 ETH + HO---> ETHP + H20 HC3 + HO ---> HC3P + 0. HO ALD ORAl CO GLY HO HCHO + H20 HC5 + HO--> 0.5 HCSP KET HO 2 + H20 HC8 + HO --> 0.95 HC8P ALD HKET HO 2 + H20 ETE + HO --> ETEP OLT + HO--> OLTP OLI + HO--> OLIP DIEN + HO ---> ISOP ISO + HO --> ISOP API + HO--> APIP LIM + HO ---> LIMP TOL + HO ---> 0.90 ADDT XO HO 2 XYL + HO --> 0.90 ADDX XO HO 2 CSL + HO --> 0.85 ADDC PHO HO XO 2 HCHO + HO--> HO 2 + CO + H20 ALD + HO--> ACO 3 + H20 KET + HO--> KETP + H20 HKET + HO--> HO 2 + MGLY + H2 GLY + HO--> HO E + 2 CO + H20 MGLY + HO--> ACO 3 + CO + H20 MACR + HO --> 0.5 TCO HKET MGLY CO HCHO HO XO2 DCB + HO --> 0.50 TCO 3 -I HO 2 -I XO 2 -I UDD GLY MGLY UDD + HO ---> 0.88 ALD KET + HO2 OP + HO --> 0.65 MO HCHO HO Table 2c Table 2c 5.26 x x 0-2!.64 x x x x x x x x 0 -ø.8 x x x x X 0-2 Table 2c 3.00 x X X X x 0-2,0 x 0-ø 2.93 x O x x x x x x x X x x 0 -ø 5.3 x 0 -. x 0 -ø 5.96 x x x x x x x x x x X x 0- lo 5.54 X ,0, 2,3 4 5

6 25,852 STOCKWELL ET AL.: REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM Table 2b. (continued) Reaction No. A, E/R, Reaction cm 3 s - K a Note (R86) (R8) (R88) (R89) (R90) OP2 + HO --> 0.44 HC3P ALD KET HO XO2 PAA + HO HCHO ACO 3 q HO 2 q XO2 PAN + HO - HCHO + XO2 + H20 q- NO3 TPAN + HO 0.60 HKET HCHO HO 2 + XO PAN NO 3 ONIT + HO HC3P + NO2 + H x x x x X x x x x x 0-2 6,8 9,9, 20 2 (R9) (R92) (R93) (R94) (R95) (R96) (R9) (R98) (R99) (R00) (R0) (R02) (R03) (R04) (R05) NO s + Organic Compounds HCHO + NO 3 ---> HO 2 q- HNO 3 q- CO ALD + NO 3 - ACO 3 q- HNO 3 GLY + NO 3 --> HNO 3 q- HO2 + 2 CO MGLY + NO 3 --> HNO 3 q- ACO 3 q- CO MACR + NO 3 --> 0.20 TCO 3 q HNO 3 q OLNN CO DCB + NO 3 --> 0.50 TCO 3 q HO 2 q XO 2 q GLY ALD KET MGLY HNO NO 2 CSL + NO3 --> HNO3 + PHO ETE + NO 3 --> 0.80 OLNN OLND OLT + NO 3 ---> 0.43 OLNN OLND OLI + NO 3 --> 0. OLNN OLND DIEN + NO 3 ---> 0.90 OLNN OLND MACR ISO q- NO 3 --> 0.90 OLNN OLND MACR API + NO 3 --> 0.0 OLNN OLND LIM + NO 3 --> 0.3 OLNN OLND TPAN + NO 3 ---> 0.60 ONIT NO 3 q PAN HCHO NO2 + XO x x l x x x x x 0 - Table 2c.9 x x x x! x x x x X X X X x x x x 0-4. x 0-5, 22 22, , 2 8, ,9, 30 (R06) (R0) (R08) (R09) (Rl0) (Rl) (Rl2) (Rl3) (Rl4) (Rl5) 03 + Organic Compounds ETE HCHO CO ORAl HO H HO OLT HCHO ALD CO ORAl ORA HO2 q HO KET KETP CH 4 q H 2 q H202 q ETH MO 2 q- 0.0 ETHP OLI HCHO ALD KET CO H202 q- 0.4 ORA CH 4 q HO 2 q HO MO KETP ETH ETHP DIEN HCHO MACR CO ORAl O3p q HO 2 q OLT HO H ACO 3 q MO 2 q KETP XO H20 2 ISO HCHO MACR CO ORAl O3p q HO 2 q OLT HO H 2 q- 0.5 ACO 3 q MO 2 q KETP XO H20 2 API ALD KET CO ETHP KETP HO HO H202 LIM HCHO OLT CO ETHP KETP HO HO 2 q H202 q- 0.9 MACR ORAl ORA2 MACR > 0.40 HCHO MGLY ORA CO H 2 q ORAl HO 2 q- 0.0 HO q- 0.3 OP ACO 3 DCB > 0.2 HO HO 2 q CO GLY ACO 3 q- 0.6 ALD MGLY + 0. PAA + 0. ORAl ORA2 TPAN > 0.0 HCHO PAN NO 2 q- 0.3 CO H 2 q- 0. ORAl HO 2 q HO ACO x x x x x x x x x l X x x 0-8 8,3 8, ,9, 33 (Rl6) (Rl) (Rl8) (Rl9) (R20) (R2) (R22) (R23) (R24) (R25) (R26) PHO + NO2 --> 0.0 CSL + ONIT PHO + HO2 --> CSL ADDT + NO 2 --> CSL + HONO ADDT TOLP CSL HO 2 ADDT CSL + HO ADDX+ NO 2 --> CSL + HONO ADDX XYLP CSL HO 2 ADDX > CSL + HO ADDC + NO 2 --> CSL + HONO ADDC CSLP CSL HO 2 ADDC CSL + HO Reactions of Intermediates Produced by Aromatic Oxidation 2.00 x x x x x x x x x x x x x X x x x x x x (R2) (R28) (R29) (R30) ACO 3 q- NO 2 --> PAN PAN ---> ACO 3 q- NO2 TCO 3 q- NO 2 ---> TPAN TPAN---> TCO 3 q- NO 2 Peroxyacylnitrate Formation and Decomposition Table 2d Table 2e Table 2d Table 2e x 0-4

7 STOCKWELL ET AL.: REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM 25,853 Table 2b. (continued) Reaction No. A, E/R, Reaction cm 3 s- K a Note (R3) (R!32) (R33) (R34) (R35) (R36) (R3) (R) (R39) (R40) (R4) (R42) (R43) (R44) (R45) (R46) (R4) (R48) (R49) (R50) (R5) (R52) (R53) (R54) (R55) (R56) (R5) (R58) (R59) (R60) (R6) (R62) (R63) (R64) (R65) (R66) (R6) (R68) (R69) (R0) (R) (R2) (R3) (R4) (R5) (R6) (R) (R8) (R9) (R80) (R82) (R83) (R84) NO + Organic Peroxy Radicals MO 2 + NO--* HCHO + HO 2 + NO 2 ETHP + NO--* ALD + HO 2 + NO 2 HC3P + NO --* 0.04 HCHO ALD KET GLY HO MO ETHP XO ONIT NO 2 HC5P + NO HCHO + 0.2! ALD KET HO MO ETHP XO ONIT NO 2 HC8P + NO ALD KET ETHP ONIT NO HO XO2 ETEP + NO-.6 HCHO + HO 2 + NO ALD OLTP + NO ALD + HCHO + HO 2 + NO KET OLIP + NO --> HO 2 +. ALD KET + NO 2 ISOP + NO MACR OLT HO HCHO ONIT NO 2 APIP + NO --* 0.80 HO ALD KET ONIT NO 2 LIMP + NO --> 0.65 HO 2 -I MACR OLI HCHO ONIT NO 2 TOLP + NO--> 0.95 NO 2 -I-0.95 HO 2 -I-0.65 MGLY +.20 GLY DCB ONIT XYLP + NO--> 0.95 NO 2 -I-0.95 HO 2 -I-0.60 MGLY GLY DCB ONIT CSLP + NO--> GLY + MGLY + HO 2 -I- NO 2 ACO 3 + NO--* MO 2 + NO 2 TCO 3 + NO--> ACO 3 + HCHO + NO 2 KETP + NO --> 0.54 MGLY ALD ACO HO XO2 + NO2 OLNN + NO--> HO 2 + ONIT + NO 2 OLND + NO --> 0.28 HCHO +.24 ALD KET + 2 NO 2 MO 2 + HO 2 --> OP ETHP + HO 2 --> OP2 HC3P + HO 2 --> OP2 HC5P + HO 2 --> OP2 HC8P + HO 2 --> OP2 ETEP + HO 2 --> OP2 OLTP + HO 2 --> OP2 OLIP + HO 2 --> OP2 ISOP + HO 2 --> OP2 APIP + HO 2 --> OP2 LIMP + HO 2 --> OP2 TOLP + HO 2 --> OP2 XYLP + HO 2 --> OP2 CSLP + HO 2 --> OP2 ACO 3 + HO 2 --> PAA ACO 3 + HO 2 --> ORA TCO 3 + HO 2 --> OP2 TCO 3 + HO 2 --> ORA KETP + HO 2 --> OP2 OLNN + HO 2 --> ONIT OLND + HO 2 --> ONIT HO 2 + Organic Peroxy Radicals Methyl Peroxy Radical + Organic Peroxy Radicals MO 2 -I- MO 2 -->.33 HCHO HO 2 ETHP + MO 2 --> 0.5 HCHO + HO 2 -I- 0.5 ALD HC3P + MO 2 --> 0.8 HCHO HO 2 -I ALD KET -I MO 2 -I MGLY XO GLY HCSP + MO 2 --> HCHO HO 2 -I ALD KET ETHP MO 2 -I XO2 HC8P + MO 2 --> 0.53 HCHO HO 2 -I- 0.4 ALD + 0.4!9 KET XO ETHP ETEP + MO 2 --->.55 HCHO + HO ALD OLTP + MO 2 --->.25 HCHO + HO ALD KET OLIP + MO 2 --> 0.55 HCHO + HO ALD KET ISOP + MO. 2 --> MACR OLT + HO OLI +.09 HCHO APIP + MO 2 --> HCHO + ALD + KET + 2 HO 2 LIMP + MO 2 -->.4 HCHO MACR OLI + 2 HO 2 TOLP + MO 2 --> HCHO + HO MGLY GLY + DCB XYLP + MO 2 --> HCHO + HO MGLY GLY + DCB CSLP + MO 2 --> GLY + MGLY + HCHO + 2 HO x x x x x x x x x x x x x x x 0 -u 2.0 x 0 -u 4.0 x x x x x x x x x l x x x x l x 0 -u 3.5 x x x x x x x x x x x x x x x 0-4. x x x x 0-! x x x x x OO x x x x x x x x x x x x x 0 -m 4.00 x 0 -m 2.00 x x x x x x x x x x x x x 0 -u.00 x x x x x 0 -u.0 x 0 -u.28 x x l x x x x 0 -u.30 x x x x x x x x x x x l x x x x 0-3,39,39, , 40, 40, , 40 39,

8 25,854 STOCKWELL ET AL.: REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM Table 2b. (continued) Reaction No. A, E/R, Reaction cm 3 s- K a Note (R85) (R86) (R8) (R88) (R89) (R90) (R9) ACO 3 + MO 2 --> HCHO + HO 2 + MO 2 ACO 3 + MO 2 --> HCHO + ORA2 TCO 3 + go 2 --> 2 HCHO + HO 2 + ACO 3 TCO 3 + go 2 --> HCHO + ORA2 KETP + go 2 --> 0.5 HCHO HO MGLY ALD HKET ACO XO 2 OLNN + MO 2 --> 0.5 HCHO + HO 2 + ONIT OLND + go 2 --> 0.96 HCHO HO ALD KET NO NIT 3.2 X X x x X X X X X X X X X x 0-2 (R92) (R93) (R94) (R95) (R96) (R9) (R98) (R99) (R200) (R20) (R202) (R203) (R204) (R205) (R206) (R20) (R208) (R209) Acetyl Radical + Organic Peroxy Radicals ETHP + ACO > ALD HO 2 q- 0.5 MO RA2.03 x 0-2 HC3P + ACO3 ---> 0.24 ALD KET HO MO X ETHP XO HCHO GLY ORA MGLY HC5P + ACO 3 --> 0.6 ALD KET HO MO X ORA ETHP XO HCHO HC8P + ACO 3 --> 0.49 ALD KET HO MO X ORA ETHP XO2 ETEP + ACO3 --> 0.8 HCHO ALD HO MO x RA2 OLTP + ACO3 --> ALD HCHO HO x 0-3 MO ORA KET OLIP + ACO 3 --> 0.94 ALD KET HO MO X ORA2 ISOP + ACO3 ---> 0. MACR OLT HO x 0-3 ORA HCHO MO 2 APIP + ACO3 --> ALD + KET + HO 2-3- MO2.40 x 0-3 LIMP + ACO 3 --> 0.60 MACR OLI HCHO + HO x 0-3 MO 2 TOLP + ACO3 --> MO 2-3- HO MGLY GLY + DCB.40 X 0-3 XYLP + ACO3 --> MO 2-3- HO MGLY GLY + DCB.40 x 0-3 CSLP + ACO3 --> GLY + MGLY + MO 2-3- HO 2.40 X 0-3 ACO 3 + ACO 3 --> 2 MO X 0-2 TCO 3 + ACO 3 --> go 2 + ACO 3 + HCHO 2.80 X 0-2 KETP + ACO 3 --> 0.54 MGLY ALD + 0. KET ACO 3.5 X HO XO 2 '- 0.5 go RA2 OLNN + ACO 3 ---> ONIT RA MO HO X 0-3 OLND + ACO3 --> 0.20 HCHO ALD KET x 0-3 ORA ONIT NO MO X X X X X X X X X X X X X X X X X x (R20) (R2) (R22) NO s-alkene-peroxyradical + NO s-alkene-peroxyradical Reactions OLNN + OLNN --> 20NIT + HO 2.0 X 0-4 OLNN + OLND --> HCHO ALD KET X 0-4 HO ONIT NO 2 OLND + OLND --> HCHO +.2 ALD KET X 0-4 ONIT + NO X x X 0-3 (R23) (R24) (R25) (R26) (R2) (R28) (R29) (R220) (R22) (R222) (R223) (R224) (R225) (R226) (R22) (R228) (R229) (R230) (R23) NO s + Organic Peroxy Radicals MO 2 + NO 3 --> HCHO + HO 2 + NO 2 ETHP + NO 3 --> ALD + HO 2 + NO 2 HC3P + NO3 --> HCHO ALD KET GLY HO MO ETHP XO2 + NO2 HC5P + NO3 --> 0.02 HCHO ALD KET HO MO ETHP XO2 + NO2 HC8P + NO3 --> 0.8 ALD KET HO ETHP XO 2 + NO 2 ETEP + NO3 -->.6 HCHO ALD + HO 2 + NO 2 OLTP + NO3 --> HCHO ALD KET + HO 2 + NO 2 OLIP + NO 3 -->. ALD KET + HO 2 + NO 2 ISOP + NO3 ---> 0.60 MACR OLT HCHO + HO 2 + NO 2 APIP + NO3 --> ALD + KET + HO 2 + NO 2 LIMP + NO3 --> 0.60 MACR OLI HCHO + HO 2 + NO 2 TOLP + NO3 --> 0.0 MGLY +.30 GLY DCB + HO 2 + NO2 XYLP + NO3 --->.26 MGLY GLY + DCB + HO 2 + NO 2 CSLP + NO3 --> GLY + MGLY + HO 2 + NO 2 ACO 3 + NO 3 --> MO 2 + NO 2 TCO3 + NO3 --> HCHO + ACO 3 + NO 2 KETP + NO3 --> 0.54 MGLY ALD + 0. HO ACO XO2 + NO2 OLNN + NO3 --> ONIT + HO 2 + NO 2 OLND + NO3 --> 0.28 HCHO +.24 ALD KET + 2 NO 2.2 X X X X X X x X X X X x X X X X X X X X x X X X X x X X X X X X X X X X X X

9 STOCKWELL ET AL.: REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM 25,855 Table 2b. (continued) Reaction A, E/R, No. Reaction cm 3 s- K k a Note Operator Reactions (R232) XO 2 q- HO 2 --> OP2.66 X x 0 - (R233) XO2 + MO 2 --> HCHO + HO x 0 - s x 0-3 (R234) XO2 q- ACO3 --> MO x x 0-2 (R235) (R236) XO2 + XO2 -- XO 2 + NO --> NO 2.3 X X X X 0-2 (R23) XO2 + NO 3 --> NO2.20 X X 0-2 Notes, DeMore et al. [994]; 2, products estimated (see text); 3, products from Uselman et al. [99]; 4, Atkinson and Lloyd [984]; 5, Paulson et al., [992a]; 6, rate constant taken to be same as for iso-butene + O3p [Cvetanovic, 98];, Atkinson [994]; 8, rate constant and product distribution calculated for average U.S. emissions (see text); 9, Kirchner and Stockwell [996b]; 0, E/R taken to be same as for the toluene + HO reaction;, taken as m-cresol; 2, O2-addition channel [Atkinson, 994] is ignored for simplification (see text); 3, Tuazon and Atkinson [990]; 4, rate constant from Bierbach et al. [994]; 5, rate constant from Wiesen et al. [995]; 6, rate constant estimated as for propyl hydroperoxide;, rate constant as for CH302H q- HO; 8, rate constant is upper limit; 9, TPAN treated as methacrylic peroxy nitrate; 20, rate constant from Grosjean et al. [993a], E/R from Kirchner and Stockwell [996b], and products from Grosjean et al. [993b]; 2, rate constantaken to be as HC3; decomposition is assumed; 22, E/R is treated as ALD + NO3; 23, rate constant is assumed to be equal to 2 x k(ald + NO3) - k(eth + NO3); 24, rate constant is treated as ALD + NO3; 25, rate constant is taken to be as crotonaldehyde [Atkinson, 994] and E/R and products from Kirchner and Stockwell [996b]; 26, rate constant is estimated as the mean of a mixture of 50% methylphenol and 50% dimethylphenol [Atkinson, 994]; 2, rate constant is uncertain because of large variation in rate constant for alkenes grouped into model species; 28, Atkinson [99]; 29, rate constant is the mean of eight measurements, and products are described by Kirchner and Stockwell [996b]; 30, rate constantaken to be 80% of estimated rate constant for methacrolein by Kirchner and Stockwell [996b]; 3, rate constant and products estimated from Atkinson [994]; 32, rate constant from Atkinson [994], and the products were estimated from Kirchner and Stockwell [996b]; 33, rate constant is from constant from Grosjean et al. [993a], the yield of HCHO was from Grosjean et al. [993b], and other products and E/R were estimated by Kirchner and Stockwell [996b]; 34, rate constant taken to be same as CH30 q- NO2; 35, rate constant taken to be same as C6HsCH202 q- HO 2 [Le Bras, 99]; 36, the difference between Bierbach [994] and Atkinson [994] was assumed (see text); 3, rate constant same as corresponding reaction for ADDT;, Kirchner and Stockwell [996a]; 39, Lightfoot et al. [992]; 40, Le Bras [99]; 4, rate constant taken to be same as HO 2 + CH2=CHCH20 [Le Bras, 99]; 42, rate constantaken to be same as HCP + NO 3. athe rate constants are for 298 K and atm. The units for rate constants of first-ordereactions are s- ; of second-order reactions, cm 3 s- ; and for third-ordereactions, cm 6 s -. For second-order reactions the rate constants are given by k =.4 exp [(-E/R)/T] unless indicated otherwise. concentration of 0.2 ppt, the NO3 radical self-reaction occurs at a rate that is 2.6% of the rate of its reaction with NO: NO3 + NO3 2 NO () Some studies have found evidence for the thermal decomposition of NO 3 [Johnston et al., 986; Davidson et al., 990; Hjorth et al., 992]; this reaction could be especially important during warm summer nights: NO3 ---> NO + 02 (8) However, these results are contradicted by the studies of Russell et al. [986] and Davis et al. [993]. DeMore et al. [994] rule out (8) on the basis of the results of Davis et al. [993], who determined that barrier to thermal dissociation is 4.3 kcal mo-. On this basis, (8) was not included in the new mecha- nism. Cantrell et al. [985] showed that the reaction of HO 2 with NO3 was important for nighttime chemistry. Further research by several groups [Hall et al., 988; Becker et al., 992; Mellouki et al., 993] has shown that the rate constant and especially the products for this reaction in the RADM2 mechanism had to be revised. DeMote et al. [994] recommend a weighted average (3.5 x 0-2 cm 3 s -) ofhallet al. [988],Melloukiet al. [988, 993], and Becker et al. [992] for the overall rate constant. This recommendation was adopted. Yields of 0. for the HOproducing channel and 0.3 for the HNO3-producing channel were taken from Le Bras [99]: HO2 q- NO3 '--> 0.3 HNO3 q- 02 q- 0. HO + 0. NO2 (9) These yields are consistent with those of Becker et al. [992] and Mellouki et al. [988, 993] within the experimental uncertainties. Reaction (0) was added for completeness to the RACM mechanism because of nighttime production of HO from (9) and other sources: HO + NO3---> NO2 + HO2 ( 0) In the atmosphere, N20 5 reacts with water to produce HNO 3 primarily through heterogeneous processes: Table 2c. The RACM Chemical Mechanism: Reaction Rate Constants of the Form k = r 2 C exp (-D/T) Reaction C, No. Reaction K -2 cm 3 s - D, K Note (R6) CH 4 + HO --> MO 2 q- H20.44 x (R62) (R8) ETH + HO--> ETHP + H20 KET + HO--> KETP + H20.5 X x (R98) ETE + NO3 --> 0.80 OLNN OLND 4.88 x Note, Atkinson [994].

10 25,856 STOCKWELL ET AL.: REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM Table 2d. The RACM Chemical Mechanism: Troe Reactions Reaction k, k3, No. Reaction cm 6 s- n cm 3 s- rn Note (R35) O3p + NO - NO X X 0 -ll 0.0 (R3) O3p + NO 2 --> NO X X 0 -ll 0.0 (R) HO + NO - HONO.00 X x (R39) HO + NO 2 --> HNO X X (R42) HO 2 + NO 2 --> HNO 4.80 X X (R53) NO 3 + NO 2 --> N X X (R5) HO -- SO 2 ---> SULF + HO X X (R2) ACO 3 + NO 2 --> PAN 9.0 X X (R29) TCO 3 + NO 2 --> TPAN 9.0 X X Here, k (cm 3 molecule - s - ) = {ko(t)[m]/( + ko(t)[m]/k (T)))0.6 t +[løg ø( :ø(t)[m]/ : (T))]2}- where: ko(t ) = k øø(t/300)-n; k (T) = k3 øø(t/300)-m; and [M] is the concentration of air in molecules cm -3 [DeMoret al., 994]. Note, DeMor et al. [994];, Kirchner and Stockwell [996a, 99]. N205 + H20 --> 2 HNO3 ( ) However, the gas-phase reaction of N20 5 with H20 is not a fast reaction due to entropy considerations [Calvert and Stockwell, 983]. Experimental measurements of the gas-phase rate constant are complicated by a strong heterogeneous component to the reaction rate. The rate constant utilized by RADM2, 2.0 x 0-2 cm 3 s -, is the upper limit recommended by DeMote et al. [994]. Other workers report an upper limit of 5 x 0-22 cm 3 s - [Sverdrup et al., 98]. Given the entropy considerations and the strong interference of heterogeneous reactions in the available measurements, this re- action was omitted from the RACM mechanism. 3. Organic Chemistry The recent mechanisms of Andersson-Sk6d [995] and Jenkin et al. [996] include highly detailed descriptions of atmospheric organic oxidation mechanisms. The mechanism of Andersson-Sk6d was developed to describe the gas-phase chemistry of 90 emitted compounds, and this mechanism includes more than 00 species with nearly 2000 reactions. The mechanism of Jenkin et al. is even larger with 20 emitted organic compounds reacting in 2500 chemical species and 000 chemical reactions. However, emissions inventories include hundreds of emitted volatile organic compounds (VOC) [Middleton et al., 990]. Given the need to conserve computational resources for a transport/transformation model and the complexity of explicit detailed mechanisms, it is necessary to group organic compounds together to form a manageable set of model classes. For the same reason, many multiple pathways are formulated as one reaction in the RACM mechanism, and not all organic intermediates (i.e., alkyl radicals) are explicitly described. 3.. Aggregation of Organic Compounds For the RACM mechanism the hundreds of VOCs in the real atmosphere are aggregated into 6 anthropogenic and 3 biogenic model species (Table 3). 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 the compound's reactivity toward HO [Middleton et al., 990; Stockwell et al., 990]. The aggregation procedure was done in two steps to simplify the process. First, hundreds of anthropogenic VOC are grouped into 32 emission categories, and second, these categories are aggregated into 6 model species (Table 3). For further details the reader is referred to Middleton et al. [990] and Stockwell et al. [990]. Here we focus on our revisions to the procedure. In the text the following subscripts are used: "i" for chemical species, "cat" for Middleton et al. emission categories, and "m" for model species. It should be stressed that the aggregated rate constants presented in Table 2b and the aggregation factors presented in Table 3 were calculated on the basis of a specific emissions inventory [Middleton et al., 990]. The procedure is completely general and can be used to adapt the mechanism's rate constants and the aggregation factors to other inventories. The procedure is illustrated in detail for alkanes, and it is applicable to the other classes of organic compounds as well. One of the parameters used for the aggregation is the reactivity of the VOC toward HO. The rate constants km,ho for the reactions of the model species with HO were calculated as the weighted mean of the rate constants kcat,ho of all categories aggregated together into a single model species (for an example, see Table 4). The rate constants kcat,ho of the categories were calculated as the weighted mean of the rate constants ki,ho at 298 K of the chemical species grouped into this Table 2e. The RACM Chemical Mechanism: Troe Equilibrium Reactions Reaction k, k3, No. Reaction A B cm 6 s- n cm 3 s- rn Note (R43) HNO 4 --> HO 2 + NO X X X (R54) N > NO 2 + NO X X X (R28) PAN --> ACO 3 + NO 2.6 x x x (R30) TPAN --> TCO 3 + NO 2.6 x x x Here, k (S-I) =.4 exp (-B/T) x {ko(t)[m]/( + ko(t)[m]/ko (T))}0.6 { +[løg'ø(kø(r)[ ul/k (r))2}-, where ko(t), k (T), and [M] are as defined for Table 2f [DeMore et al., 994]. Notes, DeMore et al. [994];, Kirchner and Stockwell [996a].

11 STOCKWELL ET AL.: REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM 25,85 Table 2œ The RACM Chemical Mechanism: Reactions With Special Rate Expressions Reaction No. Reaction Rate Constant Expression a (R24) (R33) (R34) (R46) (R58) O3p >03 [M] x 6.0 x 0-34 X (T/300 K) -2'3 HO 2 + HO 2 H x 0-3 x exp (600/T) +. x 0-33 x [M] x exp (000/T) HO 2 + HO 2 + H20---> H H x 0-34 x exp (2800/T) + 2. x 0-54 x [M] x exp (3200/T) HO + HNO3 - NO3 + H20 k = ko + k3/( + k3/k2 ) ko =.2 x 0-5 x exp (85/T) k 2 = 4. X 0-6 X exp (440/T) k3 =.9 x 0-33 x exp (25/T) x [M] CO + HO---> HO2 + CO2.5 x 0-3 x ( x 0-20 x [M]) Note, DeMore et al. [994]. athe units for second-order rate constants are cm 3 s- and for third-order the units are cm 6 s-. For all of the above, [M] is the concentration of air in molecules cm -3. Note Table 3. Allocation of Emissions Categories to RACM Species Category a RACM Aggregation Emission Category Description a Species Factor methane CH ethane ETH.00 3 propane HC alkanes with HO rate constant (298 K, atm) between HC3.. x 0-2 and 3.4 x 0-2 cm 3 s- 5 alkanes with HO rate constant (298 K, atm) between HC x 0-2 and 6.8 x 0-2 cm 3 s - 6 alkanes with HO rate constant (298 K, atm) between HC X 0-2 and.36 x 0- cm 3 s- alkanes with HO rate constant (298 K, atm) greater HC8.4 than.36 x 0- cm 3 S- alkane/aromatic mix HC ethene propene alkenes (primary) alkenes (internal) alkenes (primary/internal mix) XYL 0.09 ETE.00 OLT.00 OLT.00 OLI.00 OLI 0.50 OLT 0.50 TOL 0.29 TOL benzene, halobenzenes aromatics with HO rate constant (298 K, atm) less than.36 x 0- cm 3 s- 6 aromatics with HO rate constant (298 K, atm) XYL.00 greater than.36 x 0 - cm 3 s - phenols and cresols CSL.00 8 styrenes OLT.00 TOL.00 9 formaldehyde HCHO higher aldehydes ALD.00 2 acetone KET higher ketones KET.6 23 organic acids b ORA acetylene HC haloalkenes HC unreactive 2 others with HO rate constant (298 K, atm) HC less than. x 0-2 cm 3 s- 28 others with HO rate constant (298 K, atm) HC3.3 between. x 0-2 and 3.4 x 0-2 cm 3 S - 29 others with HO rate constant (298 K, atm) HC5.0 between 3.4 x 0-2 and 6.8 x 0-2 cm 3 s - 30 others with HO rate constant (298 K, atm) HC greater than 6.8 x 0-2 cm 3 s- unidentified c unassigned c aemission inventory and definition of emission categories taken from Middleton et al. [990]. bthere was no separate category for formic acid emissions given by Middleton al. [990]. If formic acid emissions are available, they should be grouped into ORAl with an aggregation factor of.00. Not assigned to any model species.

12 25,858 STOCKWELL ET AL.: REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM Table 4. Allocation of Alkane Emissions to RACM Species Model Species Chemical Species Categories a and Emission Rate, c k i,ha, d Chemical Species b Mmol yr cm -3 s - Category Percent Emission, e k cat,ha, f %Eca t 0-2 cm-3 s- Model Species Aggregation km Ha, Factor g 0-2 C' -3 S- CH4 ETH HC3 HC5 HC8 category methane CH4 aggregated rate constant category 2 ethane ETH aggregated rate constant category propane 2,88.5 category n-butane 8, /-butane 5, ,2-dimethylbutane category acetylene 4, category methanol, ethyl ace tate 6.48 h methylacetate h category ethanol 4, isopropyl acetate h dimethylether, category 25 0 i perchlorethylene 2,29 0. trichloroethylene,02.5 ethylene dichloride HC3 aggregated rate constant category i-pentane 668 j 3.9 hexane 0 k 5.6 n -pentane 3 j me thylp ent ane 9k' methyl pentane 9 k' 5. 2,2,4- trimethylp ent an e ,3-dimethylbutane cyclopentane j 5.08 category isopropanol n-butyl acetate m n-propyl acetate m HC5 aggregated rate constant category heptane methylcyclohexane 322 n 0.4 n-decane 208 ø.6 nonane 52 p 0.2 2,4- dim ethylhexane q methylcyclopentane q octane ,3,3-trimethylpentane q 3-methylhexane 4 r 0. q n -undecane cyclohexane.49 2,4-dimethylpentane ,5-dimethylheptane q 3,4-dimethyloctane q dimethylcyclohexane q 2,2,5-trimethylhexane q 4-methylheptane 8.5 q 3-methylheptane q 2,4,5-trimethylheptane q category n-pentadecane methyldecane 6 s 2.52 q

13 STOCKWELL ET AL.: REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM 25,859 Table 4. (continued) Model Species Categories a and Chemical Species b Chemical Species Category Model Species Emission Rate, c ki,ho, d Percent Emission, e k cat HO, f Aggregation km HO, Mmol yr cm -3 s - %Eca t C'i --3 S- Factor g 0-2 C'i -3 S- diethylcyclohexane q n- do decane n-tridecane 23 6 n-tetradecane 2 9 category 30 glycol ether t propylene glycole q ethylene glycole 9 5. m HC8 aggregated rate constant acategory defined by Middleton et al. [990]. bspecies used to calculate kcat,ho and the aggregation factor. Species with emissions less than 20 Gg/yr were usually neglected. CAs from Middleton et al. [990]. dho rate constant at 298 K, from Atkinson [994] if not mentioned otherwise. epercent category emission into RACM, percent contribution by moles of the category to the RACM class. fho rate constant of the category 298 K, calculated according to equation (4). gcalculated according to equation (6). hwallington et al. [988]. ihaloalkenes were not used to calculate khc3+ho. JThe emissions of "isomers of pentane" and "C5 paraffin" were interpreted as 62% i-pentane, 35% n-pentane, and 3% cyclopentane corresponding to the emissions of these species. kthe emissions of "isomers hexane" were interpreted as 55% hexane, 28% 2-methylpentane, and % 3-methylpentane corresponding to the emissions of these species. IThe emissions of "methylpentane" were interpreted as 62% 2-methylpentane and % 3-methylpentane corresponding to the emissions of these species. mwilliams et al. [993]. nthe emissions of "C-cycloparaffins" were interpreted as methylcyclohexane. øthe emissions of "isomers of decane" were interpreted as n-decane. PThe emissions of "isomers of nonane" were interpreted as nonane. qcalculated using Kwok and Atkinson [995]. rthe emissions of "methylhexane" were interpreted as 3-methylhexane. SThe emissions of "isomers of undecane" were interpreted as 2-methyldecane. tdagaut et al. [989]. category. The weighting was done on the basis of the emissions E in units of moles per year taken from the U.S. emissions inventory [Middleton et al., 990]' with Z (E catkcat,ho) Z (%Ecatkcat,HO) cat m km'ho = E E cat : carom cat m 00 (2) Ecat = EE,, %Ecat = 00 E Ecat Eca t (3) i cat cat Z (Eiki,HO) i cat The rate constant was calculated kcat'ho = EEi (4) igcat for 23 and 298 K and fitted to an Arrhenius equation to determine the activation energy of the rate constant for the 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 [Stockwell et al., 990]. Under this assumption an emitted compound can be represented by a model species which reacts at a different rate provided that an aggregation factor, Aggi, is applied to the compound emissions. The factor is the ratio of the fraction of the chemical species i which reacts with HO during the day to the fraction of the model species m which reacts: Agg,: - exp (--km,ho f [HO] dt) (ls) For.the daily average integrated HO radical concentration, J' [HO] dt, a value of.63 x 0 TM s cm -3 was used [Stockwell et al., 990]. The aggregation factor becomes unity if both the model and emitted VOC are highly reactive. To use this aggregation procedure, it is necessary to know the HO rate constants of every chemical species that is to be aggregated in a model species. However, for cases where this rate constant is not known we calculated aggregation factors for categories, Aggca t. Because species in a category react similarly, the aggregation factors of categories, Aggcat, can be used as an approximation for the aggregation factors of chemical species, Agg i belonging to a category. Table 3 shows the aggregation factors, calculated using

14 25,860 STOCKWELL ET AL.: REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM Agg at = X Mcat (6) --exp(--km,hof [HO] dt) where Mca t is the mixing factor, which is for categories which belong only to one model species and /n for categories which belong to n model species. The rate constants used to calculate k cat,ho were taken from the recommendations of Atkinson [994] or calculated according to the procedure of Kwok and Atkinson [995] by using a structure-reactivity relationship. As an example of the aggregation process, Table 4 shows the complete set of rate constants and aggregation factors for chemical species, classes, and model species for alkanes. Middleton et al. [990] is not followed for anthropogenic emissions of isoprene and terpenes. Although isoprene is a terminal alkene, it is treated as the separate model compound, ISO, because it is primarily emitted from biogenic rather than anthropogenic sources. Any anthropogenic emissions of isoprene and terpenes should be aggregated together into the primarily biogenic model species Alkanes Alkanes are very important chemical species which may be transported over long distances. Methane and ethane are treated explicitly in both the RADM2 and RACM mechanisms. All other alkanes along with alkynes, alcohols, esters, and epoxides are aggregated according to their HO rate constants into three additional model species, HC3, HC5, and HC8. HC3 represents species with relatively low HO reaction rate constants (lower than 3.4 x 0-2 cm 3 s-l); HC5 represents species with moderate HO reaction rate constants (between 3.4 x 0-2 and 6.8 x 0-2 cm 3 s-l); and HC8 represents species with relatively high HO reaction rate constants (>6.8 x 0-2 cm 3 The determination of the reaction products and product yields of the model species HC3, HC5, and HC8 is described in the following. Alkanes react with HO by the abstraction of hydrogen atoms. The product of this reaction is an alkyl radical, which under atmospheric conditions reacts immediately with 02 to form an alkyl peroxy radical: For example, n-butane: RH + H O- R. + H20 () R RO2 (8) CH3CH2CH2CH 3 + HO (+ 02) CH3CH2CH2CH202' where/? ',HO is the product yield fraction of the reaction of the chemical species i with HO giving the product p. The values for/3{',x-io were taken from the recommendations of Atkinson [994] or calculated according to the procedure of Kwok and Atkinson [995]. For example, /3CH3CH2OH,HO is 0.95 because CH3CH2OH + HO - 5% 'O2CH2CH2OH + 95% (CH3CHO + HO2) + H20 (2) 3.3. Peroxy Radical Formed From Alkanes The species CH4, ETH, HC3, HC5, and HC8 react with HO and 02 to form the peroxy radicals MO2, ETHP, HC3P, HC5P, and HC8P, respectively. The composition of the model peroxy radicals HC3P, HC5P, and HC8P is determined by the composition of the model species HC3, HC5, and HC8 and the reactivity of the chemical species which are aggregated in them. In general, organic peroxy radicals RO 2- can react with NO, NO2, NO3, HO2, and R'O2-. RO2' + NO - x RONO2 + ( - x)(ro' + NO2) (22) RO2' + NO2 --> RO2NO2 (23) RO2' + NO3 --> RO' + NO2 q- 02 (24) RO2' + HO2 --> ROOH + 02 (25) RO2' + R'O2' --> products (26) The determination of the rate constants for (22)-(26) are discussed by Kirchner and Stockwell [996a]. The alkoxy radicals RO. formed in (22) and (24) may react with 02 or undergo isomerization or decompose. When an RO. radical reacts with 02, an HO 2 radical and a carbonyl species are produced. An example for n-butane is shown by CH3CH2CH2CH NO- CH3CH2CH2CH20. q- NO2 (2) CH3CH2CH2CH20' > CH3CH2CH2CHO q- HO2 (28) Alkoxy radicals may also decompose to produce a carbonyl species and after addition of 02, a new organic peroxy radical is produced: CH3CH2C(O')HCH3(+O2) CH3CHO + CH3CH2OO' (29) or they can undergo a,5-h-isomerization leading normally to an HO 2 and a carbonyl species with one additional NO to NO 2 conversion: CH3CH2(CHO2')CH3 + H20 (9) CH3CH2CH2CH20' --> HOCH2CH2CH2CH2' (3o) However, alcohols and alkynes, which are grouped together with alkanes in the model species HC3, HC5, and HC8, do not HOCH2CH2CH2CH2' > HOCH2CH2CH2CH202' always form peroxy radicals but may form other species directly. For that reason the reactions of HC3, HC5, and HC8 HOCH2CH2CH2CH202' + NO with HO form products like ALD, GLY, or CO besides HC3P, --> HOCH2CH2CH2CH20' + NO2 (32) HC5P, and HC8P, respectively. The product yie!d fractions, HOCH2CH2CH2CH20' apm,ho, of the reaction of model species m with HO giving + 02 productp were calculated as the weighted mean of the product yields of all chemical species belonging to that model species: --> HOCH2CH2CH2CHO + HO2 (33) E AggiEi]3 Ho In the RACM mechanism, production of the operator radical, XO2, accounts for the additional NO to NO 2 conversation step i Grn due to isomerization. For individual alkoxy radical species the apm'hø = 5 Agg Ei (20) rate constants for isomerization of the alkoxy radicals were i m determined, and these were used to determine the appropriate

15 STOCKWELL ET AL.: REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM 25,86 XO2 yield from the average composition of each lumped alkoxy radical species [Kirchner and Stockwell, 9968]. The calculation of the product yields of the model peroxy following considerations: Acetone reacts to produce CH3COCH20. (reactions(39)-(40)). The CH3COCH20. may either react with 02 to form methylglyoxal and HO 2 or to radical are based on the product yields of the chemical species decompose to produce an acetyl peroxy radical and formaldein these reactions. Given a general reaction of a model peroxy radical RO2(m) with a species X, with X being NO, NO2, NO3, hyde. The branching ratio between these two reaction processes is uncertain [Atkinson and Lloyd, 984; Le Bras, 99]. HO2, or R'O2(m): For the RACM mechanism, decomposition of the RO2(m) + X--> O? O2(m),xP + ø? O2(m),xP' +''' (34) CH3COCH20 2. radical was not considered, and further experimental data are required: the product yield fraction, CgRO2(m p ),X, for a product p from this CH3COCH3 + HO (+ 02) --> CH3COCH202' + H20 reaction is given by the weighted mean of the product yield (39) fractions, /3{', o,x, for a product p from the reaction of the chemical species i with HO giving an peroxy radical and the subsequent reaction of this species peroxy radical with X: CH3COCH20 2. q- NO --> CH3COCH20. q- NO2 CH3COCH20' q > a CH3COCHO q- a HO 2 (40) E AggiEi/3,P,I-IO,X i m cty ø2(m)'x = E AggiEi (35) igm where/?, ',i-io,x is normalized to account only for reaction products which come from the formation of an RO2.. Therefore p /3/,i-io,x V i X (36) + b CH3CO3' q- b HCHO (4) The reactions of methyl ethyl ketone are known to be considerably more complicated. The reaction of methyl ethyl ketone with HO and the subsequent reaction of the organic with 02 radical leads to [Atkinson, 994] CH3CH2COCH3 + HO (+ 02) --> 0.46'O2CH2CH2C(O)CH CH3CH(O2.)C(O)CH3 Here/3 [,i-io,x was calculated using product yield of the reactions R. + HO, RO 2' q- X and RO- decomposition. If no product yields were available from laboratory studies, they were calculated following the procedure proposed by Kwok and Atkinson [995] 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), glyoxal (GLY), and methylglyoxal and other species of the form RC(O)CHO (MGLY). DCB and GLY are produced through the photooxidation of aromatic compounds. Most MGLY is produced through aromatic oxidation and some is produced through ketone reactions. The reactions of HCHO are represented by an explicit set of reactions. The chemistry of ALD is treated as acetaldehyde. The chemistry of KET is treated as a mixture of 50% acetone and 50% methyl ethyl ketone (by mole fraction). Although methyl ethyl ketone and higher ketones are the main photochemical oxidation products of VOC, the emissions of acetone are twice as high as methyl ethyl ketone on a mole basis [Middleton et al., 990]; therefore this split is a reasonable compromise between the composition of the emissions and the photochemical production. Atkinson [994] recommends the following rate constants for the reaction of HO with acetone and methylethyl ketone: Acetone k = 5.34 x 0-8 cm 3 s - X T 2 exp (-230/T) (3) Methyl ethyl ketone k = 3.24 x 0-8 cm 3 s - X T 2 exp (44/T) () The rate constant of HO with KET was derived from the arithmetic means of the activation energies and the rate constants at 298 K (Table 2c). The scheme for the reactions of KET was based upon the q CH3CH2C(O)CH20 2' q- H20 (42) The three peroxy radicals may react with NO or other species to produce oxy radicals. CH3C(O)CH2CH202' + NO--> CH3C(O)CH2CH20' + NO2 CH3CH(O2')C(O)CH3 + NO (43) - CH3CH(O.)C(O)CH3 + NO2 (44) CH3CH2C(O)CH202' + NO --> CH3CH2C(O)CH20. + NO2 (45) According to the rate constants given by Atkinson [994], 58% of the primary oxy radicals, CH3C(O)CH2CH20. and CH3CH2C(O)CH20., react through isomerization and the remaining 42% react through abstraction of H atoms by 02. The isomerization reactions involve the transfer of an H atom to the terminal O atom: CH3C(O)CH2CH20' '-> HOCH2CH2C(O)CH2' (46) CH3CH2C(O)CH20'--> HOCH2C(O)CH2CH2' (4) In the RACM mechanism the effects of the HOCH2CH2C(O)CH 2. radical are represented through the reactions of ALD, XO2, and HO2, and the effects of the HOCH2C(O)CH2CH 2. radical are represented through the reactions of MGLY, XO2, and HO 2. The H abstraction reactions produce aldehydes and an HO2: CH3C(O)CH2CH > CH3C(O)CH2CHO + HO2 CH3CH2C(O)CH20' > CH3CH2C(O)CHO + HO2 (48) (49)

16 25,862 STOCKWELL ET AL.: REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM CH3C(O)CH2CHO is treated as ALD and CH3CH2C(O)CHO is treated as MGLY. The secondary oxy radical CH3CH(O.)C(O)CH 3 rapidly decomposes [Atkinson, 994]: cm3rm(o')r(o)cm3(+ 02) CH3CHO 4- cm3c(o)o 2' (so) These products are characterized as ALD and ACO 3 in the RACM mechanism. For methyl ethyl ketone if (43)-(50) are considered along with the radical yields and given the assumption that KET represents a 50%-50% mixture of acetone and methyl ethyl ketone, (5) for the reaction of the KETP radical with NO is obtained: KETP + NO MGLY ALD ACO HO XO 2 4- NO2 (5) HCOCOH + HO- HCO. + CO + H20 (52) mcorom 4- mo (4-02)----> mc(o)c(o)o 2 4- m20 (53) The relative percent yields of (52) and (53) are 60% and 40%, respectively. The HCO radical reacts to produce HO 2 and CO: HCO HO2 + GO (54) If (52) and (54) were the only reaction pathways for HCO- COH, the overall reaction would be HCOCOH + HO (+ 02)-o HO2 + 2 CO + H20 (55) An explicit consideration of the reaction pathway for HC(O)- C(O)O2' would require two additional species HC(O)- C(O)O2' and HC(O)C(O)O2NO 2 and many additional RO 2' reactions. The subsequent chemistry of HC(O)C(O)O 2. is highly uncertain, but it would be expected that the radical reacts with NO: mc(o)c(o)o2-4- NO (+ 02) ----> mo2 4- GO 4- CO2 4- NO2 (56) Probably, HC(O)C(O)O 2- can react with NO 2 to produce a peroxy nitrate with a thermal stability similar to peroxyacetyl nitrate (PAN). UC(O)C(O)O2-4- NO2 HC(O)C(O)O2NO2 (5) Because of the a-dicarbonyl group and the aldehyde hydrogen atom, HC(O)C(O)O2NO 2 would be expected to have a much higher HO reaction rate constant and to have a much higher photolysis frequency than PAN: HC(O)C(O)O2NO2 + mo CO + GO 2 4- NO3 + m20 ue(o)e(o)o2no 2 4- hv (+ 02) (58) HO2 + GO + GO 2 4- NO3 (59) Given these differences between HC(O)C(O)O2NO 2 and PAN, it would be wrong to group this peroxynitrate with PAN. However, the mechanism can still be simplified if a few assumptions are made. The NO pathway for HC(O)C(O)O 2. involves two peroxy radicals HC(O)C(O)O 2. and HO 2 and therefore two possible NO to NO 2 conversions. The NO 2 pathway produces an overall total of fewer than NO to NO 2 conversions because of HC(O)C(O)O2NO 2 formation and consumption of the HO radical in (58). If it is assumed that the NO and NO 2 pathways for HC(O)C(O)O 2. are of equal importance, then the total number of NO to NO 2 conversions will be close to. This assumption is justified because only the photolysis reaction directly produces additional peroxy radicals (although the rapid photolysis of NO3 does produce some 03). The HC(O)C(O)O 2. reaction pathway will have about the same effect on 03 and NOx concentrations as the pathway given by (55). For this reason and given their high uncertainty, the possible HC(O)C(O)O 2. reactions are ignored. Therefore the RACM model reaction is completely analogous to (55): The dicarbonyl species glyoxal (GLY) and methylglyoxal (MGLY) have very similar chemistry and therefore their prod- GLY 4- mo(4-02)----> mo2 4-2 CO + m20 (60) uct yields were derived in a similar manner. For the reaction of glyoxal with HO, Atkinson [994] recommends two reaction The major difference for MGLY is that its reaction with HO produces an acetyl radical rather than an HO 2 (Table 2b). The channels: reactions of DCB are described in section because unsat- urated dicarbonyls are products of aromatic oxidation. The reactions of the NO3 radical with carbonyl species HCHO, ALD, GLY, MGLY, and DCB are included because these reactions can be significant during the nighttime [Stockwell and Calvert, 983]. The NO3 radical reacts with saturated aldehydes through hydrogen atom abstraction: RCHO + NO3 (+ 02) --'-> RCO3 ø 4- HNO3 (6) The rate constants for the reaction of NO3 with HCHO and acetaldehyde (ALD) were taken from DeMore et al. [994]. The NO3-MGLY rate constant was assumed to be equal to the rate constant of ALD. For glyoxal there are only aldehydic hydrogen atoms; therefore the rate constant for GLY was estimated as 2 x k(acetaldehyde + NO3) - k(ethane + NO3). The activation energies for the reactions of NO 3 radical with all these aldehydes were assumed to be equal to the activation energy for the acetaldehyde + NO 3 reaction given by DeMore et al. [994]. The rate constant for the DCB + NO 3 reaction was taken from the estimates of Bierbach et al. [994], and the activation energy was estimated to be the mean of the activation energy for the reaction of NO 3 with acetaldehyde and the activation energies for the reaction of NO 3 with unbranched internal alkenes. Photolysis of carbonyls represents a significant loss process for these species. The quantum yields, cross sections, along with representative photolysis frequencies used for the carbonyl species are given in Table 2a. The photolysis frequencies for HCHO and ALD remain nearly unchanged from those used by Stockwell et al. [990], while the KET values increased by about 0%. The photolysis frequencies for GLY is significantly changed due to new recommendations from Atkinson et al. [992b]. For MGLY, new values for the cross sections [Atkinson et al., 992b; Staffelbach et al., 995] and the quantum yield measurements [Koch and Moortgat, 996] were adopted Alkenes Alkenes are important constituents of the polluted and rural troposphere. Owing to the double bond, they are very reactive

17 STOCKWELL ET AL.: REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM 25,863 Table 5. Estimated Rate Constants for Alkenes Used for All Alkenes for Which Data Were Not Available RACM HO NO Mechanism Alkene k (298 K), E/R (HO), k (298 K) E/R, k (298 K), E/R, Species Structure cm 3 s- K cm 3 s- K cm 3 s- K OLT OLI RCH=CH x x x 0-00 R2C=CH x x x 0-00 RCH=CHR 6.4 x x x R2C=CHR 8.8 X X X R2C=CR x x x 0 -ls 300 species that have relatively high rate constants for reaction with HO radical and they react with ozone and NO 3 radicals [Finlayson-Pitts and Pitts, 986]. Four model species were used to represent the anthropogenic emitted alkenes to better represent their diverse reactivities. Ethene is explicitly represented by ETE because of its lower rate constants for reactions with HO and 0 3 and its relatively high concentrations. Propene and other terminal alkenes (alkenes with a double bond attached to a carbon atom at an end of the molecule) are represented by OLT. Internal alkenes (alkenes with a double A!kene + NO3 radical. The initial reaction of NO 3 with alkenes is addition to the double bond. The adduct reacts with 02 to form a peroxy radical which was represented by the model species OLN in the original RADM2 mechanism [Stockwell et al., 990]. The NO3-alkene-peroxy radicals may react with NO, NO3, or RO 2- to form oxy radicals. For the oxy radicals produced in this reaction, two fundamentally different reaction pathways are possible: decomposition (pathway D) or formation of organic nitrates (pathway N). Recent experimental studies on the reaction of NO 3 which alkenes have shown bond within the molecule) such trans-2-butene or cycloalkenes that the product yields for the two pathways depend strongly are represented by the model species OLI. Dienes and,3- butadiene are aggregated into a new model species, DIEN. on the structure of the alkene from which the NO3-alkene-oxy radical was formed [Barnes et al., 990; Hjorth et al., 990; Skov DIEN was introduced into the RACM mechanism because the et al., 992; Tuazon et al., 993]. Radicals formed from unrate constant for the reaction of,3-butadiene with HO is very different from the other terminal alkenes and its chemistry is more like isoprene than terminal alkenes [Atkinson, 994]. branched alkenes and dienes produce high yields of carbonyl nitrates (pathway N), while the reactions of NO 3 with branched monoalkenes have higher yields of carbonyls and For the RACM mechanism the rate constants for the reac- NO 2 (pathway D) and relatively low yields of carbonyl nitrates. tions of alkenes with HO, NO3, and 03 were taken from Atkinson [994] when possible. For alkenes not listed by Atkinson [994] the rate constants were estimated by taking the mean of alkenes with known rate constants and similar structure (Table 5). To distinguish between these two reaction pathways, the species OLN in the RADM2 mechanism was replaced by two model species for the NO3-alkene-peroxy radical, OLNN, which primarily produces nitrates (pathway N), and OLND, which primarily decomposes to produce carbonyls and NO Reaction with HO. The reaction of HO with alk- (pathway D). It is assumed that the total yield of OLNN and enes occurs by addition to the double bond, followed by the addition of 02 to form a peroxy radical. The peroxy radicals may react with NO, NO3, or RO 2. to produce oxy radicals which are believed to decompose completely [Atkinson, 994]. The only significant exception is the oxy radical resulting from O LND is unity. Following the general procedure for building the RACM mechanism, the product yields O/i,NO3 Nitrate for the model species OLNN and OLND from the reactions of the species ETE, OLI, and OLT with NO 3 depend on the rate constants ki,no 3 ethene; about 20% of this reacts via H abstraction by 02 which and nitrate product yields, b'i,no /2lNitrate 3, for the reaction of the produce hydroxyacetaldehyde (H2C(OH)CHO) which is chemical species with NO 3 and the composition of alkene treated as ALD in RACM: emissions [Middleton et al., 990]. Our estimated nitrate yields, H2C(OH)CH20' H2C(OH)CHO + HO2 (62) 3 i,no Nitrate 3, for alkenes are compiled in Table 6, but there are relatively few experiments, and the agreement between them is not good [Barnes et ai., 990; Hjorth et al., 990; Tuazon et al., 993]: The rate constants for the reactions of the model species OLT and OLI with HO producing the peroxy radicals OLTP and OLIP were calculated according to (2). In contrast to the alkanes, for which measured data for many rate constants are available, only a few measurements exist for alkenes [Atkinson, 994]. Where no experimental rate constants were available, estimated values from Table 5 were taken as ki,ho in (4). The composition of the model peroxy radicals ETEP, OLTP, and OLIP was calculated by using the same procedure that was applied to the alkanes. For alkenes higher than ethene the OLT + NO3 '--> O OLT,NO OLNN 3 O LNN + (- O OLT, OLNN NO3), OLND (63) OLI + NO3 croli,no _ OLNN 3 O LNN + ( - aoli,no3) OLNN x OLND (64) with E K'/,- J-' itv i,no3 b i,no3 Nitrate igm OLNN _ (65) O/m'NO3- E Eiki,NO3 igm reaction with HO is so fast on a daily basis that on the regional scale an initial amount of alkenes would be almost completely where rn is equal to either OLI or OLT. The reactions of consumed. Therefore according to (6), no reactivity, weighting OLNN and OLND with NO and with other peroxy radicals is required. were described by Kirchner and Stockwell [996a].

18 25,864 STOCKWELL ET AL.: REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM Table 6. Estimated Yields of Nitrate // Nitrate k/, i,no3,/ Carbonyls and NO2 From Nitrate Radical-Alkene Adduct Reactions Alkene Estimated Estimated Yield of Structure Nitrate Carbonyls and NO 2 b i,no 3 ETE CH2=CH HCHO RCH--CH R2C=CH OLT 0.20 HCHO ALD 0.80 HCHO ALD OLI RCH=CHR ALD R2C--CHR ALD KET R2C=CR KET Ozonolysis of alkenes. Ozone reacts with alkenes to form ozonides which immediately decompose to yield excited Criegee intermediates [Finlayson-Pitts and Pitts, 986] and carbonyl species, for example, CH3CH=CH > CH3CHO + [CH2OO]* (66) CH3CH=CH > HCHO + [CH3CHOO] * (6) Table. Products and Yields of Criegee Intermediates Used in This Work Criegee Product Atkinson [994] This Work The dominant atmospheric loss process of the stabilized Criegee intermediates appears to be its reaction with H20 therefore these products are included in the new mechanism. Paulson et al. [992b] proposed that stabilized Criegee intermediates may react with H20 to produce HO: R2COO + H20 --> R2CO + 2 HO (69) This reaction was used to fit the HO yields reported by Atkinson [994]. The HO yields were fit by varying the fraction of Criegee intermediates which react by (69). The fraction of Criegee intermediates that react by (69) was estimated to be 0.0, 0.4 and 0.6 for CH2OO, RCHOO, and R2COO. The resulting HO yields for alkenes of various structures are given in Table 8. The reaction of stabilized Criegee intermediates with water may also produce hydrogen peroxide [Becker et al., 993]: R2COO + H20 --> R2CO + H202 (0) A H20 2 yield of 0.5% was determined by the measurements for CH2OO and for R2COO the estimated yield was about 3%. Formic acid is also produced from the reaction of R2COO with water [Moongat et al., 99]. The results of Moongat et al. [99] suggesthat the HCOOH and higher organic acids are formed through the rapid decomposition of The branching ratio for the formation of Criegee intermedi- organic hydroperoxides. Assuming the yield for RCHOO to be ates from CH3CH=CH 2 was investigated by Horie and Moongat the geometric mean of the CH2OO and R2COO yields and [99] and was found to be [CH2OO]/[R2COO] = 0.:0.62. allowing for the formation of HO and organic acids, the fol- To estimate these branching ratios for other alkenes, this ratio lowing reactions result: of 4:6 was used for all R-CH=CH2 and for R2C=CHR. The R2C=CH 2 branching ratio was calculated to be 0.4 to be consistent with the branching ratios for R-CH=CH2 and R2C=CHR: CH2OO + H20 '-> H HCHO HCOOH () RCHOO + H20 '-> H202 [CH2OO]* [CH2OO]* [RCHOO]* 4 x 4 [R2COO]* [RCHOO]* [R2COO]* 6 x (68) HO RCHO RCOOH (2) The excited Criegee intermediates are either stabilized or they decompose. Table shows the products and yields of excited Criegee intermediates from Atkinson [994] and the yields R2COO + H20 '-> 0. H HO + 0. RC(O)R RC(O)OR (3) assumed in this work. Some HO is directly produced through the decomposition of the excited Criegee intermediates [Atkinson and Aschmann, 993; Atkinson, 994]. However, direct Further reactions of the ester, RC(O)OR, are ignored because it is relatively unreactive and only a relatively small amount is produced through gas-phase reactions. The alcohols which reproduction of HO is not enough to account for the concentra- sult from gas-phase chemistry are treated similarly. tions determined byatkinson and Aschmann [993] even if the upper limits given in Table are used. The chemistry of the unsaturate dicarbonyl species (DCB) and the unsaturated peroxynitrate (TPAN) has been revised by including HO and NO3 addition reactions and ozonolysis. These reactions are described below because unsaturated car- bonyls and peroxynitrates are products of aromatic decay. [CH2OO] = CH CO 2 + H 2 '"' CO H CO + H HCO + HO [CH3CHOO] CH3CHO GO 2 q- CH CO2 + CH3 + H CO + CH3OH CO + CH3 + HO HCO + CH [(CH3)2COO : (CH3)2CO CH3C(O)CH 2 + HO Table 8. Estimated HO Yields Produced From the Ozonolysis of Alkenes Alkene Structure Experimental Fit [Atkinson, 994] (This Work) CH2=CH CH2=CHR CH2=CR CHR=CHR CHR=CR CR2=CR

19 STOCKWELL ET AL.: REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM 25, Aromatic Chemistry Aromatic compounds are oxidized in the atmosphere through reaction with HO. When aromatics react with HO radicals, the radicals either add to the aromatic ring or react with a substituent group. However, the reaction of HO radicals with aromatic substituent groups is usually a much less important channel than HO addition to the aromatic ring. For typical aromatic compounds containing alkyl substituent groups the measured fraction of HO radicals reacting with a substituent group ranges between 0.04 and 0.2 [Atkinson, 994]. It is assumed that an overall fraction of 0.0 of the HO radicals Three different aromatic-ho adducts were added to the mechanism: ADDT (from toluene), ADDX (from xylene), and ADDC (from cresol). The measured rate constants for the reactions of the methylhydroxycyclohexadienyl radical with 02 and NO 2 [Knispel et al., 990] were used for these adducts. The reactions of NOx, H202, and aromatics with the aromatic-ho adduct increase cresol yields. The difference in the cresol yields measured by Bierbach [994] and Atkinson et al. [994] can be further rationalized if the aromatic-ho adduct reacts with ozone through aromatic-ho adduct Cresol + HO (4) react by abstraction with the aromatic compounds treated in the RACM mechanism. Given that the aromatic-ho adduct reacts with 02, NOx, H202, and aromatics, it would be very surprising if it did not The HO radicals that react with alkyl substituent groups react with ozone. In the toluene experiments of Atkinson and abstract hydrogen atoms. This hydrogen atom abstraction Aschmann [994] the concentrations of 02 and 03 were about leads to the formation of peroxy radicals which may react with -5 x 0 8 and-5 x 0 TM cm -3, respectively. Using the con- NO or other species to form oxy radicals. Subsequently, the oxy centrations of 02 and 03 and the difference between the mearadicals react with oxygen to produce aromatics with aldehyde sured cresol yields of Bierbach [994] andatkinson et al. [994], or ketone substituent groups. The NO to NO 2 conversions the rate constant for (4) is calculated to be about 05 times affected by the oxy radicals are accounted for by the XO2 greater than the rate constant for the reaction of aromatic-ho operator radical in the RACM mechanism. The product aroadduct with 02. This implies that the rate constant for the matics with aldehyde or ketone substituent groups have relatoluene-ho adduct with 03 is about 5 x 0 - cm 3 s - which tively small yields relative to the products formed from the is near the rate constant for the reaction of the aromatic-ho addition of HO to the aromatic ring; therefore the subsequent adduct with NO2,-3.6 x 0 - cm 3 s - [Knispel et al., 990]. reactions of these products are not included. Including (4) in the mechanism improves the agreement be Treatment of cresol formation from aromatic-ho tween simulated concentrations and environmental chamber adducts. The aromatic-ho adduct reacts with 02 to either data for the experiments containing toluene. abstract an H atom to form a cresol or to add the 02 to form For o-xylene oxidation experiments that did not contain a peroxy radical. In the original RADM2 mechanism, large NOx, a 2,3-dimethylphenol yield of 2.4 _+ 0.9% was measured amounts of cresols were formed from aromatic photooxida- [Atkinson and Aschmann, 994]. They measured a 2,3- tion: 25% from toluene and % from xylene. There is much dimethylphenol to 3,4-dimethylphenol ratio of 3:2 in systems disagreement in the literature about the relative importance of containing NOx and o-xylene. If this same ratio applies to the cresol formation, but the most recent measurements suggest systems not containing NOx, the overall cresol yield is 4.0 +_ that the cresol yields have been significantly overestimated..5%. This result implies that the reaction of the HO-xylene For example, Atkinson et al. [989, 99, 992a] found in the adduct with 03 must be somewhat slower than for the reaction presence of NOx that the yield of cresol from the oxidation of of the HO-toluene adduct. A value of x 0 - cm 3 S - was toluene was 25%, while for ortho-xylene, meta-xylene, and assigned to the rate constant for the reaction of aromatic-ho para-xylene the yields of cresol were 6, 2, and 9%, respec- adduct with 03. tively. More recent experiments provide yields that are Further parameterization of Aromatic-HO adduct 40% lower. Furthermore, cresol formation decreases with de- reactions. In the RACM mechanism the aromatic-ho adcreasing NO x concentrations [Atkinson and Aschmann, 994]. duct reacts with 02 through addition to produce a peroxy In the absence of NOx they measured a cresol yield of 2% radical as was assumed in several previous mechanisms [Lurfrom toluene. The 2,3-dimethylphenol yield from o-xylene was mann et al., 986; Calvert and Madronich, 98; Stockwell et al., 4 times lower in experiments without NOx than in experiments 990]. The existence of an aromatic peroxy radical and its with NOx, although it is possible that in these experiments subsequent reaction mechanism are both highly uncertain other phenolic compounds may have been formed which were [Barnes et al., 996]. However, the mechanism presented here not measured. Even lower yields of cresols were measured by Bierbach [994]. Bierbach used the photolysis of H202 as the HO radical is consistent with available laboratory and environmental chamber data. Many more experimental data are needed to understand the atmospheric oxidation of aromatic species. If source for the oxidation of aromatics. He determined that the the aromatic-ho adduct reacts with 02 through addition, the yield of cresols increased with increasing initial concentrations resulting peroxy radical could react through three different of NOx, H202, and aromatics. A cresol yield less of than % was measured for toluene if the initial H202 concentration was low. For high H202 initial concentrations, Bierbach measured a cresol yield of % if NOx was absent. Overall, these results suggesthat much cresol is produced through the reactions of the aromatic-ho adduct with NOx, H202, and aromatics. Thus high yields of cresol from aromatic oxidation should not be expected for most typical atmospheric conditions but to allow pathways (Figure ): In pathway I, the peroxy radical may react with NO, NO3, or R'O 2. to produce an RO. radical. The RO. radical would decompose to form a C6-dicarbonyl-diene (A) and HO 2. In pathway II, the oxygen molecule can form a bridge across the ring followed by addition of a second oxygen molecule. The resulting radical reacts to form an RO. that decomposes to produce unsaturated dicarbonyls, glyoxal or methylglyoxal, and for highly polluted conditions with high NOx concentrations HO 2 radicals; these are represented in the RACM mechanism the yield of cresol formation is dependent on the NOx concen- as DCB, GLY or MGLY, and HO2, respectively. Some of the tration in the RACM mechanism. adducts may decay through this channel because C4-

20 25,866 STOCKWELL ET AL.: REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM HO'O2q ooo HO I NO NO HO 2 øxn HO,O 2 IIIa o- ] o IIIb Peroxynitrate [ O Figure. Possible decomposition pathways for the peroxy radical resulting from 02 addition to the aro- matic-ho adduct. HO2 o O dicarbonyls are experimentally measured aromatic compound decay products [Bierbach et al., 994]. guarantee that the environmental chamber simulations lead to a unique determination of their relative importance of the In pathway III, the peroxy radical can react to form a species three reaction pathways. The calculation of the branching rawith two 02 bridges across the ring followed by addition of a third oxygen molecule. This radical reacts to form an RO. tios is dependent in part upon the secondary products formed from the reactions of the unsaturated dicarbonyls. Knowledge radical which decomposes to produce glyoxal or methylglyoxal of the chemistry of the unsaturated dicarbonyl products is very and HO 2 radicals. The relative importance of these three possibilities for the uncertain, and as new data become available, these branching ratios must be reevaluated. reaction of the aromatic-ho-o2 adduct was determined The simulations show that the rate of ozone formation is through the simulation of environmental chamber data including the experiments listed in Tables 9 and 0, but there is no different depending upon the decay channel (Figure 2). Channel I produces ozone at the greatest rate after a short inhibition

21 STOCKWELL ET AL.: REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM 25,86 Table 9. Comparison of RACM and RADM2 With SAPRC Ozone Data Peak Ozone Concentrations Time of Peak Ozone Concentrations SAPRC Principal Experi- RACM Differ- RADM2 Differ- Experi- RACM Differ- RADM2 Differ- Experi- Organic ment, Simulation, ence, Simulation, ence, ment, Simulation, ence, Simulation, ence, ment Compounds a ppm ppm % ppm % min min % min % EC-42 ethene >360 > > EC-43 ethene EC-8 n-butane >495 > > EC-26 propene EC-23 mixture EC-232 mixture >390 >390 --' > EC-233 mixture > EC-23 mixture EC-2 mixture EC-24 mixture >360 > > EC-242 mixture EC-243 mixture EC-245 mixture EC-246 mixture >55 > >50... EC-254 acetaldehyde >30 > > EC-305 n-butane >360 > > EC-33 toluene n -butane EC-340 toluene >330 > > EC-344 m-xylene EC-345 m-xylene amixture consists of n-butane, 2,3-dimethylbutane, ethene, propene, tert-2-butene, toluene, m-xylene, HCHO, and CO. period. Channel II shows no inhibition period, but the rate of ozone production is somewhat less. Channel III shows the slowest rate of ozone formation, and the rate depends on whether the final products are an unsaturated-dihydroxydione, glyoxal and methyglyoxal (IIIa), or only glyoxal and methyglyoxal (IIIb). For xylene the agreement between simu- lations and experiment was best if 00% of the adducts react by channel II, while for toluene the best fit was obtained if it was assumed that 50% of the adducts react by channel II, 50% react by channel IIIb, and none react through channel I. It was furthermore assumed that the adducts have a 5% nitrate yield Treatment of unsaturated dicarbonyl products. The description of the chemistry of unsaturated dicarbonyl products (DCB) is primarily based on the work of Bierbach et al. [994] and Wiesen et al. [995]. For the reaction of HO with unsaturated dicarbonyls, Bierbach et al. found that 50% of the HO radicals react by H abstraction and about 50% add to the double bond. This ratio of abstraction to addition for the Table 0. Comparison of RACM and RADM2 With SAPRC Nitrogen Dioxide Data Peak NO 2 Concentrations Time of Peak NO 2 Concentrations SAPRC Principal Experi- RACM Differ- RADM2 Differ- Experi- RACM Differ- RADM2 Experi- Organic ment, Simulation, ence, Simulation, ence, ment, Simulation, ence, Simulation, ment Compounds a ppm ppm % ppm % min min % min EC-42 ethene EC-43 ethene EC-8 n-butane EC-26 propene EC-23 mixture EC-232 mixture EC-233 mixture EC-23 mixture EC-2 mixture EC-24 mixture EC-242 mixture EC-243 mixture EC-245 mixture EC-246 mixture ! EC-254 acetaldehyde EC-305 n-butane EC-33 toluene n -butane EC-340 toluene EC-344 m-xylene EC-345 m-xylene amixture consists of n-butane, 2,3-dimethylbutane, ethene, propene, tert-2-butene, toluene, m- lene, HCHO, and CO. Differ- ence, %

22 25,868 STOCKWELL ET AL.: REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM :2 I nism was assumed, and the excited Criegee intermediates from the unsaturated dicarbonyls were characterized as [HC(O)CHOO]*, [HC(O)C(CH3)OO]*, and [RC(O)CHOO]. [HC(O)C(R)OO]* is also a product of methacrolein ozonolysis [Kirchner and Stockwell, 996b]. These excited Criegee intermediates would be expected to produce glyoxals, aldehydes, organic acids, and radicals. If it is further assumed that the ozonolysis of DCB produces equal yields of all three Criegee intermediates, and if the carbonyl products are included, the following reaction results: DCB > 0.5 GLY MGLY ALD CO + 0. PAA + 0. ORAl ORA ACO HO HO (5) 0 Unsaturated acyl peroxy radicals (TCO3) are produced from DCB through photolysis and through reactions with HO and NO 3 [Stockwell et al., 990]. The reaction of unsaturated Time, Min acyl peroxy radicals with NO leads to the production of Figure 2. Simulated ozone formation profiles for decompo- NO2, CO2, and compounds of the form R-C(O)-CH=CH.. sition pathways (I, I!, and III) for the peroxy radical resulting from 02 addition to the aromatic-ho adduct aromatic decay R-C(O)-CH=CH. reacts with 02 to produce R-C(O)CHO, CO, and HO2. The dicarbony compounds R-C(O)CHO react (see text). Two different alternatives are presented for pathway with HO to produce acyl radicals and CO and HO2 radicals. III: IIIa, decomposition to 3,4-dihydroxy-3-hexene-2,5-dione, These products are almost the same as the products from the glyoxal and methyglyoxal; and IIIb, glyoxal and methyglyoxal aidehyde-hydrogen atom abstraction from methacrolein only. These simulations are based upon EC-33. [Tuazon and Atkinson, 990]: reaction of HO with unsaturated dicarbonyls is similar to the ratio for the reaction of HO with methacrolein [Atkinson, 994]. The products for the addition of HO with unsaturated dicarbonyls are based upon the several studies [Bierbach et al., 994; Wiesen et al., 995; Tuazon et al., 985]. These researchers have studied the products of the HO addition reaction with 3-hexene-2,5-dione. The initial reaction sequence is the same as for alkenes: the HO radical adds to the double bond, and this is followed by addition of 02. The peroxy radical reacts with NO, NO3, or other peroxy radicals to form an RO. radical. In Contrasto most alkenes the RO. decomposition reaction is much less important. About 30% of the RO. radicals decompose, and 0% react via H abstraction by 02 or through H atom shift isomerization [Bierbach et al., 994]. The main prod- dicarbonyl species (DCB) was based on the estimates of Bierbach et al. [994]. There are no available product studies for the ozonolysis for these compounds. A Criegee mecha- CH2CCH3CHO + HO (+ 02) --> CH2C(CH3)C(O)OO' + H20 (6) CH2C(CH3)C(O)OO' + NO ---> NO2 + CH2C(CH3)C(O)O' () CH2C(CH3)C(O)O' ---> CH2C(CH3)' q- CO2 (8) CH2C(CH3)'(+ 2 02) --=> CH3C(O)O2' q- HCHO (9) HCHO + HO (+ 02)----> CO q- HO2 q- H20 (80) To simplify the mechanism, the model species TCO 3 was also used to represent the unsaturated acyl products. It is probable that the acyl radicals produced from unsaturate dicarbonyls uct for the reaction of HO With 3-hexene-2,5-dione was found react with NO2 to produce peroxynitrates; however, these to be 3,4-dihydroxy-3-hexene-2,5-dione, an unsaturated- products have not been observed. The comparable product, dihydroxy-dicarbonyl (UDD) [Wiesen, 995]. To keep the RACM mechanism reasonably condensed, the peroxy radical is not treated separately rather it is represented by XO2 and its products from the peroxy radical-no reaction. produced in methacrolein degradation, methacrylic peroxynitrate, has been reasonably well investigated. The rate constants and products for the reactions of our model species TPAN with HO and 03 are based upon the experimental measurements The addition reaction of HO with 3,4-dihydroxy-3-heXene- for methacrylic peroxynitrate. These reactions are further dis- 2,5-dione is very fast (k 2. x 0 -ø cm 3 s -) [Wiesen, cussed by Kirchner and Stockwell [996b].!995]. Therefore it was assumed that the H abstraction reac Treatment of cresols. For the reaction of HO with tion is unimportant for UDD. The main products of the 3,4- dihydroxy-3-hexene-2,5-dione + HO reaction were HO 2 and cresol (CSL) it was assumed that an unsaturated dicarbonyl enol was formed which can isomerize via keto-enol-tautomerie 3,3-dihydroxyhexane-2,4,5-trione. The 3,3-dihydroxyhexane- into a saturated tricarbonyl compound which is treated as 2,4,5-trione may dehydi'ate to form hexane-2,3,4,5-tetraone. MGLY. For the reaction of NO 3 with cresol it was assumed Generalizing this to UDD, the reaction products of HO with that the only reaction channel was the abstraction of the UDD are treated as HO 2 and a mixture of ALD and KET. The rate constant for the reaction of ozone with unsaturated hydrogen atom from the HO group. There may be other possible reactions [Atkinson et al., 994], but now no other reactions are known. The resulting phenoxy radical (PHO) can react with NO 2 followed by a hydrogen shift to form ortho-nitrophenol:

23 STOCKWELL ET AL.: REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM 25,869 OH CH3 H 3 + NO 3 + HNO3 (8) O CH 3 stants for the peroxy radical reactions were compared. The calculated concentrations (especially the nighttime concentrations) of PAN, higher organic hydroperoxides, peroxyacetic acid organic peroxy radicals, HO2, HO, and NO3 were strongly affected by the revisions to peroxy radical chemistry. On the basis of that study it appears that RO2. + NO3 reactions are more important in the nighttime atmosphere than RO2- + RO2. reactions. O CH3 +NO 2 NO 2 CH3 OH (82) The rate constant for (82) is not known; therefore the rate constant for the analogous CH30. + NO2 addition reaction was used. For the PHO + HO2 reaction the rate constant for Ph-CH2OO- + HO2 was used. Because orthonitrophenol is less reactive than phenol by order of magnitude, the products are described in the mechanism as 0.CSL + ONIT. There is also experimental evidence for the formation of quinones from cresols [Wiesen et al., 995] through the reactions of phenoxy radicals, but not enough is known about the quinone chemistry to include it in the RACM mechanism. 3.. Biogenic Organic Species The oxidation mechanism for isoprene was completely revised from the mechanism of Stockwell et al. [990]. Isoprene is oxidized in the atmosphere through its reactions with HO, NO3, and 03. Our oxidation scheme for the isoprene + HO reaction follows the measurements of Paulson et al. [992a]. The new isoprene mechanism is a good representation of the atmospheri chemistry of methacrolein and MPAN. It includes treatment of isoprene ozonolysis, hydroperoxide production, and production of carbonitrates from the reaction of isoprene with NO 3. The RACM mechanism also includes a mechanism for the oxidation of a-pinene and d-limonene. The new oxidation mechanism for a-pinene and d-limonene is based upon the limited available experimental data for terpene reactions [Atkinson, 994]. Because many of the reactions for a-pinene and d-limonene oxidation are unknown, our mechanism was completed by using reactions that were analogous to known reactions for lower alkenes. To include the large number of different monoterpenes in the mechanism, it was necessary to group the individual monoterpenes into a few groups. This grouping was done according to similarities in rate constants and structure. Structure and rate constants of sabinene and A3-carene The reactions of organic peroxy radical with NO, NO3, HO2, and with other organic peroxy radicals were revised as described by Kirchner and Stockwell [996a]. The revised rate constants were significantly different from those used by Stockwell et al. [990] in several cases. Simulations made using the original mechanism and a version with the updated rate con- 4. Comparison With Environmental Chamber Data Simulation of environmental chamber experiments allows an atmospheric chemical mechanism to be tested as a complete system. The chamber typically contains a mixture of NOx and organic species that are exposed to either sunlight or artificial light and concentration measurements are made as the compounds react [Carter et al., 995]. Unfortunately, the usefulness of environmental chamber data is limited by several factors including the relatively high concentrations used, wall effects, and photolysis rate uncertainties [Stockwell et al., 990]. Chamber experiments are usually conducted at concentrations greater than those in ambient air to permit the concentrations to be measured. The walls of the chamber affect the concen- trations of 03, NOx, aldehydes, and ketones and are a significant source of radicals [Lurmann et al., 986; Stockwell et al., 990; Carter et al., 995]. There may be differences between sunlight and the spectral distribution in the actinic flux in indoor chambers, while for outdoor chambers the photolysis rates may be uncertain due to clouds, wall adsorption and reflection, and other chamber effects [National Center for Atmospheric Research, (NCAR), 986]. Taken together, all of these uncertainties in environmental chamber data limit the evaluation of predicted ozone concentrations to no better than about _+30%[Stockwell et al., 990]. The environmental chamber runs modeled here are the same experiments used to test both the mechanism of Lurmann et al. [986] and the RADM2 mechanism [Stockwell et al., 990]. Extremely extensive testing of the RADM2 mechanism that used the results from several different chambers showed that the experiments used here are representative for ozone, nitrogen oxides, and VOC [Carter and Lurmann, 989]. The reevaluated data set of Carter et al. [995] was used for these tests. The 20 environmental chamber runs were made using the Evacuable Chamber (EC) of the Statewide Air Pollution Research Center (SAPRC) of the University of California, Riverside [Carter et al., 995]. This indoor chamber has a volume of 5800 L. Typically, the relative humidity was near 50% for the modeled experiments. The RACM and RADM2 mechanisms were compared by running both against these experiments. The RADM2 simulations presented here are are similar to those of a-pinene; therefore those terpenes are treated as a-pinene. The mechanism was successfully tested against environmental chamber runs [Kirchner and Stockwell, slightly different from the simulations of Stockwell et al. [990] 996b]. because the description of the chamber radical sources, other wall effects, actinic flux was updated to the chamber model 3.8. Parameters for Organic Peroxy Radical-Peroxy Radical developed by Carter et al. [995]. It should be noted that the Reactions and the Reactions of Nitrate Radical original chamber runs used by Stockwell et al. included exper- With Organic Peroxy Radicals iment EC-306 but it had to be dropped from our comparisons because the experiment had been removed from the reevaluated data set. Figures 3, 4, and 5 show profiles of NOx, ozone, and the aromatics toluene and xylene, respectively, for experiment EC- 23. which consisted of the photolysis of a mixture of nitrogen oxides and a complex mixture of reactive organic species:

24 25,80 STOCKWELL ET AL.: REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM Time, Min Figure 3. Predicted and observed concentration profiles of NOx for SAPRC environmental chamber experiment EC-23. Triangles represent experimental values of NO; diamonds represent experimental NO2; solid lines represent RACM simulations; dotted line represents RADM2 simulations. ethene, propene, t-2-butene, n-butene, 2,3-dimethylbutene, toluene, and m-xylene [Carter et al., 995]. Experiment EC-23 is a "benchmark" case that has been presented by several mechanism developers [McRae et al., 982; Lurmann et al., 986; Stockwell et al., 990]. The experiment provides a clear illustration of the chemistry of ozone formation. Figure 3 shows that NO is converted to NO 2 which is due to the reaction of peroxy radicals with NO. The NO 2 photolyzes to produce ozone, while the reduction in NO concentrations decreases the rate of the 03 + NO reaction. Therefore the net effect of the NO to NO 2 conversions is to increase ozone concentrations (Figure 4). The profiles of NOx and ozone and the timing of the maxima of NO 2 and ozone are better fit by the new mechanism than by RADM2 when the simulations are made with the revised chamber model of Carter et al. [995]. For experiment EC-23 the NO-to-NO2 conversion rate predicted by RACM is slightly greater than the rate predicted by RADM2. The timing of the maximum NO 2 concentration predicted by RACM is earlier and in better agreement with the chamber data than the RADM2 calculation. Figure 4 shows the comparison of simulated ozone concentrations with the experimental values for EC-23. The RACM profile shows an ozone production rate more consistent with the chamber data than that predicted by RADM2. The peak concentration predicted by the RACM mechanism is slightly higher than predicted by the RADM2 mechanism. However, both predicted ozone profiles are well within +_30% of the experimental ozone concentration. The timing of the ozone peak predicted by RACM is somewhat later than the experimental value but is in better agreement than the RADM2 calculations. The predicted and measured concentrations of toluene and xylene are in good agreement for both mechanisms for these chamber conditions (Figure 5). This agreement indicates that for this experiment both mechanisms accurately predict the HO concentration because the only processes which reduce aromatic concentrations are reaction with HO and dilution in the chamber. Tables 9 and 0 summarize the calculated and observed maximum 03 and NO 2 concentrations and the time at which they occur for all 20 of the modeled EC experiments. In a few cases the experiments were concluded before the ozone reached a maximum or the simulations did not predict a ozone maximum during the time of the experiment. For these cases ß ** 0.4 ** ' ' Time, Min Figure 4. Predicted and observed concentration profiles of ozone for SAPRC environmental chamber experiment EC- 23. Circles represent experimental values of 03; solid line represents RACM simulations; dotted line represents RADM2 simulations. Time, Min Figure 5. Predicted and observed concentration profiles of toluene for SAPRC environmental chamber experiment EC- 23. Circles represent experimental values of toluene; squares represent xylene; solid lines represent RACM simulations; dotted lines represent RADM2 simulations.

25 STOCKWELL ET AL.: REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM 25,8 the timing of the ozone maximum in Table 9 is reported as "greater than experimental time," and the maximum ozone concentration is the concentration at the end of the experiment. These cases are included in Figure 6, showing the peak 03 concentrations observed and predicted by the two mechanisms, but they are not included in Figure in which the timing of the peak 03 concentrations is compared. Overall, the RACM mechanism generally predicts higher peak ozone concentrations than those predicted by RADM2. For the peak ozone concentrations the mean normalized deviations of the RACM and RADM2 calculated values from the EC experimental values are 9 and 3%, respectively. The timing of the ozone maximum is better represented by the RACM mechanism for most cases, except for a few cases with an aromatic hydrocarbon as the principal component (EC-33, EC-344, and EC-345). If these cases are included, the mean of the normalized deviations of the RACM calculated timing of the peak ozone concentrations from the EC experimental values is 30%, while RADM2 gives a value of 2%. For EC-340 the agreement between the final experimental and simulated ozone concentrations was better for the RACM mechanism. If runs EC-33, EC-340, EC-344, and EC-345 are excluded, the mean normalized deviations are and 25% for RACM and RADM2, respectively. For experiment EC-33 the RACM mechanism predicted the toluene profile better than the RADM2. Because toluene concentrations in the chamber were affected only by HO reaction and dilution, this suggests that the RACM mechanism predicted HO concentrations better than the RADM2 mechanism. For EC-344 run where m-xylene was the major aromatic compound, the ozone profile and xylene profiles were both simulated better by the original RADM2. However, simulations of aromatic photo-oxidation experiments made in another SAPRC chamber (the evacuatable Teflon chamber (ETC)) gave better agreement between model I} $oo ' 400 t loo o Experimental Time of Ozone Maximum, Min Figure. Time of the maximum of the ozone concentrations predicted by RACM and RADM2 mechanisms plotted against SAPRC experimental values. Only experiments where an ozone peak was observed are included. Circles refer to RACM mechanism; asterisks refer to RADM2. Middle line is the - line; top and bottom lines represent _+30% deviation from the - line. and experiment for the RACM mechanism, but these experiments were performed under black light illumination and low relative humidity (-5%). Furthermore, the chamber studies showed that the RADM cresol mechanism produced too much ozone and that the timing of the peak ozone was late. The revised RACM mechanism predicts the timing of the ozone peak better, but the amount of ozone production from cresol requires improvement. Although the aromatic chemistry in the RACM mechanism is based on more recent data than the RADM2 mechanism, future versions will require additional research. Both mechanisms predict peak NO 2 concentrations very well (Figure 8). The mean normalized deviations of NO 2 peak concentrations from the EC chamber experimental measurements for RACM and RADM2 are 0 and 9%, respectively. The timing of the NO 2 peak concentration predicted by RACM and RADM2 are very similar (Figure 9). The timing of the simulated NO 2 maxima is later in most cases, but the overall performance of RACM is only slightly improved (Figure 9). The mean normalized deviations of timing of the NO 2 peak concentrations for RACM and RADM2 are both near 3%. The ability of the mechanism to correctly simulate the NO and NO 2 profiles suggests that not only the HO concentrations but also the underlying acyl peroxy radical concentrations and their effect on the chemistry NO 2 are well simulated because under these conditions, significant amounts of PAN and PAN analogs are produced. Experimental Maximum Ozone, ppm Figure 6. Maximum ozone concentrations predicted by RACM and RADM2 mechanisms plotted against SAPRC experimental values. Circles refer to RACM mechanism; asterisks refer to RADM2. Middle line is the - line; top and bottom lines represent _+30% deviation from the - line. 5. Comparison of the RACM and RADM2 Mechanisms The concentrations of 03, H202, H2SO4, HNO3, and PAN predicted by the RACM and the RADM2 mechanisms were compared. The purpose of this comparison was to allow an

26 25,82 STOCKWELL ET AL.- REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM Experimental Maximum NO2, ppm Figure 8. Maximum NO 2 concentrations predicted by RACM and RADM2 mechanisms plotted against SAPRC experimental values. Circles refer to RACM mechanism; asterisks refer to RADM2. Middle line is the - line; top and bottom lines represent _+30% deviation from the - line. assessment of the effect of the revisions on the calculated concentrations of species that are of particular interest to the developers of regional air quality models. No emissions, dilution, or deposition were included in the first 8 cases. The impacts of emissions on the relative performance of the mechanisms were assessed through cases 9-2 discussed below. The first 8 simulations are from an extensive mechanism intercomparison study [Stockwell and Lurmann, 989]. The 200 l simulations were performed for typical rural and urban conditions in a manner similar to a comparison of the first and second version of the RADM mechanisms [Stockwell et al., 990]. Two days were simulated for summer surface conditions starting and stopping at 0600 LT. The simulations employed diurnally varying solar radiation for 40 ø latitude. The fluxes are calculated from a radiative transfer model based on the delta- Eddington technique [Joseph et al., 96; Madronich, 98] as described by Stockwell et al. [990]. The original and updated photolysis rates were used for the RADM2 and the RACM mechanisms, respectively. The composition of the rural and urban mixtures shown in Table were used, and Table 2 shows the initial concentrations for these simulations. The rural simulations have initial NOx concentrations ranging from 0.5 to 5 ppb, total nonmethane organic compound (NMOC) concentrations ranging from 0 to 00 ppb C. The RADM2 aggregation factors were used for both mechanisms. The rural initial NMOC/NOx ratios ranged from 2 to 200. The urban simulations have initial NOx concentrations ranging from 0 to 00 ppb, initial VOC concentrations ranging from 00 to 000 ppb C, and initial NMOC/NOx ratios ranging from to 00. The RACM and RADM2 mechanisms predict similar 03 concentrations for typical atmospheri conditions (Table 3). The average ratio of 03 predicted by RACM to that predicted by RADM2 is.08. The RACM mechanism predicts higher peak 03 concentrations for of the 8 cases. The 03 predictions for the rural cases are all in excellent agreement. The RACM mechanism predicts more 03 than the RADM2 mechanism for all but one of the urban cases with the greatest differences occurring for cases, 2, and 5. The urban cases have a greater proportion of highly reactive alkenes and aromatics than the rural cases, and therefore the urban cases are Table. Nonmethane Organic Composition Used for Cases -8 Rural Summer Surface Urban Summer Surface = 50 o Fraction Fraction of Fraction Fraction of of Nonmethane of Nonmethane Organic Group/ Group, a Organic Group, a Organic Compound % Carbon, % % Carbon, % Z Experimental Time of NO 2 Maximum, Min Figure 9. Time of the maximum of the NO 2 concentrations predicted by RACM and RADM2 mechanisms plotted against SAPRC experimental values. Circles refer to RACM mechanism; asterisks refer to RADM2. Middle line is the - line; top and bottom lines represent _+30% deviation from the - line. Alkanes Ethane Propane n -butane n -pentane n-hexane n-heptane n-octane Alkenes Ethene Propene tert-2-butene Isoprene Aromatics Benzene Toluene m-xylene Mesitylene Aldehydes Formaldehyde Acetaldehyde From Stockwell et al. [990]. All organic fractions are calculated on a carbon basis.

27 STOCKWELL ET AL.' REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM 25,83 more strongly affected by the revisions to the aromatic chemistry and the nighttime NO 3 q- alkene chemistry. The RACM mechanism predicts less H20 2 than the RADM2 mechanism for 5 of the cases due to the revisions to the peroxy radical chemistry [Kirchner and Stockwell, 996a]. The predicted H20 2 concentrations agree well for the cases which produce large amounts H202, but the mechanismshow their greatest differences for the cases 6 and where both predicted little H20 2. The average ratio of H20 2 predicted by RACM to that predicted by RADM2 is 0.90 if these two cases are excluded and 0.99 if they are included. The RACM mechanism predicts higher H2SO 4 concentrations than the RADM2 for all 8 cases; the average predicted peak H2SO 4 concentration ratios of RACM to RADM2 are.20. This also shows that the RACM mechanism predicts higher HO concentrations for these rural and urban conditions because the reaction of HO with SO2 is only gas-phase reaction which produces H2SO 4 in the mechanisms. The differences between the RACM and RADM2 mechanisms are greatest for the urban cases, and these differences are greater for those cases with the higher initial NMOC/NOx ratios. The RACM mechanism predicts slightly higher HNO3 con- centrations than the RADM2 for 4 of the 8 cases. The Table 2. Initial Conditions for Mechanism Intercomparison Cases -8 Meteorological Conditions Value Case Date June 2 Latitude 40øN Elevation, km 0 Temperature, K 298 Pressure, atm Nonmethane Organic Carbon, NOx, Duration, ppb C ppb days Rural Summer Surface Ratio of NMOC to NOx Urban Summer Surface Other Species CH4, ppb CO, ppb S02, ppb 03, ppb NO/NOx Rural Summer Surface Urban Summer Surface

28 25,84 STOCKWELL ET AL.: REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM Table 4. Initial Conditions for Cases 9-2 Case 9 Case 20 Compound Initial Initial Case 2 Initial Concentrations, Emissions, Concentrations, Emissions, Concentrations, ppb ppt/min ppb ppt/min ppb NO 0.2 NO HNO3 0. H20 X H SO2 '" CO 200 Inorganics ß ß ß -. x x 0 ß ß Alkanes CH 4 00 '-' 00 '-' 00 ETH '" 0.24 "' HC3 '" 2.6 "' HC5 '" 0.6 '" HC8 '-' 0.45 '" ETE ". OLI "- OLT '" ISO '-' Alkenes TOL XYL '" '-' Aromatics HCHO.0 ALD.-. KET... PAN... Carbonyls From Kirchner and Stockwell [996a]. average predicted peak HNO 3 concentrations ratios is.08. The greatest differences are seen for the urban cases and the differences are greatest for those cases with the higher initial NMOC/NOx ratios. The concentrations are affected by the differences between the predicted HO concentrations and the differences in the nighttime NO 3 chemistry which affects the predicted nitrogen balance. The differences in predicted PAN concentrations are very significant, and these results are in accord with our previous results [Kirchner and Stockwell, 996a]. The RACM mechanism predicts less PAN at noon on the second day for all cases except case 2. The average predicted noontime PAN concencases -8. Emissions are included for cases 9 and 20. Case 9 represents a somewhat polluted atmosphere with emissions of NOx and organic compounds. Case 20 represents a relatively cleaner atmosphere with the initial NOx and organic compound concentrations and emission rates reduced by a factor of 0 from those of case 9. Case 2 represents the situation of aged air mixing with rural air. The three cases were simulated using the RADM2 mechanism, the RADM2 mechanism with updated peroxy radical chemistry (run D of Kirchner and Stockwell [996a]), and the new RACM mechanism. Table 5 shows the results of this intercomparison for cases 9 to 2. The effect of the revisions to peroxy radical chemistry tration ratios of RACM to RADM2 are The increased was discussed by Kirchner and Stockwell [996a]; therefore we rates of reaction of the AGO 3 radical with NO and the addi- focus the differences between the RACM and RADM2 mechtional reaction of AGO 3 with NO 3 cause the predicted concentrations of PAN predicted by the RACM mechanism to be significantly reduced. This illustrates the importance of the revisions to the NO 3 and peroxy radical chemistry. anisms with revised peroxy radical chemistry (RADM2-D). The 03 concentrations were not strongly affected by the revisions. However, the RACM mechanism predicted nighttime HO and HO 2 concentrations that were much greater than To further determine the cause of the observed differences those of RADM2-D. The ratios of the nighttime HO concenbetween RACM and RADM2, three additional cases were simulated (Table 4). The first two cases are based upon conditions used in an international intercomparison of chemical mechanisms [Poppe et al., 996], and all of these cases were used to determine the effect of revisions to peroxy radical trations predicted by the RACM mechanism to that predicted by the RADM2-D mechanism were 3.34, 2.90, and 2.04, for cases 9, 20, and 2, respectively. The ratios of the nighttime HO 2 concentrations predicted by the RACM mechanism to that predicted by the RADM2-D mechanism were.39,.49, chemistry on predicted concentrations [Kirchner and Stockwell, and.28 for cases 9, 20, and 2, respectively. The most im- 996a]. Cases 9, 20, and 2 were cases, 2, and 3, respectively, portant reason for the greater nighttime HO concentrations of Kirchner and Stockwell [996a]. The same meteorological predicted by the RACM mechanism is the revised product conditions were used for these simulations as were used for composition of the reaction of HO 2 with NO 3. For case 9 the

29 STOCKWELL ET AL.' REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM 25,85

30 25,86 STOCKWELL ET AL.' REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM HO2 + NO3 reaction becomes the most important HO source during the night, and for cases 20 and 2 this reaction occurs at a rate that is about 5 and 50% the rate of the O3 + HO2 reaction. The higher nighttime HO concentrations increase the importance of their reactions during the nighttime. For case 9 a counter species analysis [Leone and Seinfeld, 985] of the RACM simulation showed that the fraction of organic compounds which react with HO slightly exceed the fraction reacting with NO3. The fraction of organic compounds which react with 03 was significantly less than the fraction reacting with either HO or NO 3 during the nighttime hours. On the fifth night of this simulation, for example, the amount of HCHO that reacted with NO3 is the same as the amount that reacted with HO. There are additional slow reactions of NO3 with alkanes which are not included in the RACM mechanism organic nitrate concentrations predicted by the RACM mech- To make the mechanism suitable for three-dimensional anism. Although the organic nitrate photolysis frequency is higher in the RACM mechanism, the lower organic nitrate concentrations are due to lower production from aromatic oxidation. In the RADM2 and RADM2-D mechanisms, large amounts of cresols are produced from the oxidation of aromatic species. During the night the cresol rapidly reacts with NO3 to produce intermediates that react with NO 2 to form organic nitrate. Organic nitrates are produced in the RACM mechanism through the reaction of PHO with NO 2. Phenoxy (3-D) models, the RACM mechanism uses grouped organic species. Aggregation factors for reactive organic species were developed according to the methods of Middleton et al. [990] and Stockwell et al. [990]. The organic chemistry was significantly revised. Revisions to the branching ratios for alkane decay led to significantly reduced aldehyde to ketone ratios. The decomposition and nitrate formation channels for the reactions of NO 3 with alkenes are now more explicitly treated, and the relatively large amounts of nitrates resulting from the radicals are produced from the reaction of cresol with NO3 and reactions of unbranched alkenes are included in the RACM partially from the reaction of cresol with HO. In the RACM mechanism the cresol yields are much lower than in the mechanism. The production of HO from the ozonolysis of alkenes has a much greater yield in the RACM mechanism RADM2 and RADM2-D mechanisms; this leads to much than in the RADM2 mechanism. lower concentrations of phenoxy radicals and organic nitrate. During daytime the lower cresol yield from aromatics in combination with a faster rate of HO 2 production and a reduced organic peroxy radical production rate from the reaction of HO with HC3 lead to higher daytime HO, HO2, and NO x The aromatic chemistry has been revised to be more consistent with recent laboratory data. Many more aromatic reaction products are accounted for in the RACM mechanism. Large amounts of products from the aromatic-ho addition reactions had been ignored in the RADM2 mechanism; this has been concentrations for the RACM mechanism. The revised rate corrected in the RACM mechanism. The chemistry of the parameter for the HO + HC3 reaction is significantly lower in unsaturated dicarbonyl species and the unsaturated peroxyni- RACM. The increase in the HO concentrations leads to in- trate has been revised by including HO and NO3 addition creased aldehyde (ALD) production from the hydrocarbons, reactions and ozonolysis. The amount of cresol production has and consequently, the rate of the ALD + HO reaction is been reduced. For the RADM2 mechanism, environmental increased. This produces more acetyl peroxy radicals (ACO3), chamber studies showed that the timing for the cresol chemperoxy acetic acid (PAA), and PAN for the RACM mechanism istry was poor and that the cresol chemistry produced too relative to the RADM2-D mechanism. The concentrations of much ozone. The new RACM mechanism has improved the PAN and PAA are further increased because the revised rate constants for the reaction of these compounds with HO are lower than those used in RADM2. The increase of methyl hydrogen peroxide (OP) concentrations results from the updated photolysis rate constant which is about 25% lower than the one used in RADM2. This leads to a significant increase in the predicted methyl hydrogen peroxide concentrations. The increase in the higher organic peroxide concentrations is much smaller because it has a much greater rate constant with HO which makes its loss by photolysis relatively less important. 6. Conclusions The Regional Atmospheric Chemistry Mechanism is a substantially revised version of the RADM2 mechanism [Stockwell et al., 990], with rate constants and product yields from the most recent laboratory measurements. The mechanism has been compared with the previous RADM2 mechanism, and it has been tested against environmental chamber data. The mechanism includes stable inorganic species, 4 inorganic intermediates, 32 stable organic species, and 24 organic intermediates in 23 reactions. Several inorganic reactions were added to the RACM mechanism including reactions of O(3p) with nitrogen oxides and ozone, the reaction of HO with H2, HO with nitrous acid, and a few additional NO 3 reactions. It should be stressed that the RACM mechanism considers which may increase the relative fraction of organic compounds that react with NO3 and slightly decrease the relative importance of the HO reaction. However, the comparison of the only the gas-phase chemistry of the troposphere. Therefore the RACM mechanism with the RADM2-D mechanism shows reaction of N20 s with H20 was dropped from the RACM that during the nighttime the reaction of organic species with HO can be important. The revised nighttime production of HO now included the RACM mechanism produce greater HO 2 and organic peroxy radical concentrations. The revisions incorporated into the aromatic oxidation scheme in the RACM mechanism lead to greater NO x concentrations. Comparison of the RACM simulations with the gas-phase mechanism because it now appears that the N2Os + H20 reaction is a heterogeneous reaction. Heterogeneous and aqueous phase reactions have a strong effect on the gas-phase chemistry, and these should be included in photochemical atmospheric chemistry models [Seinfeld, 986]. However, much more data from both laboratory and field measurements will be required before this aspect of atmospherichemistry is well RADM2-D simulations shows that this increase is due to lower understood. timing for the cresol chemistry, but the level of ozone production requires additional improvement. The peroxyacetyl nitrate chemistry and the organic peroxy radical-peroxy radical reactions were revised. Organic peroxy radical + NO3 reactions were added according to Kirchner and Stockwell [996a]. These revisions lead to significantly lower calculated PAN concentrations. The oxidation mechanism for isoprene was completely re-

31 STOCKWELL ET AL.: REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM 25,8 vised from the RADM2 mechanism. The new isoprene mechanism includes a good representation of methacrolein, and it includes treatment of isoprene ozonolysis, hydroperoxide production, and production of carbonitrates from the reaction of isoprene with NO 3. The RACM mechanism includes a mechanism for the oxidation of a-pinene and d-limonene. Tests against environmental chamber experiment show a reasonably good agreement between simulation and experiment [Kirchner and Stockwell, 996b]. Both the RADM and RACM mechanisms reproduce the concentrations of O., NO2, and hydrocarbons and the timing of peak O. and NO 2 concentrations measured in environmental chambers to within about +30%. Intercomparisons for rural and urban conditions show that the RADM and RACM Although many rate constants for the primary reactions of HO, O., and NO3 with organic compounds have been measured, there has been insufficient progress in the understanding of the secondary organic chemistry. Product yields and the reactions of many photooxidation products remain to be investigated. This is especially true for the chemistry of higher molecular weight organic compounds. There are few available data on the secondary chemistry for compounds with carbon numbers greater than 3 or 4. For example, the nature, yield, and fate of many of the carbonyl products produced from the high molecular weight alkanes, alkenes and other compounds requires more laboratory data. The reactions of organic peroxy radicals with NO, HO2, RO2', and NO 3 need to be better characterized because these reactions may strongly affect the concentrations of PAN, organic nitrate, organic peroxides, H202, and ozone. For higher molecular weight alkoxy radicals the relative importance of unimolecular decomposition, reaction with oxygen, and isomerization reactions is unknown, and this may effect simulated ozone production rates. More environmental chamber and field data are required to test the mechanisms. Uncertainties in photolysis rates, wall effects, and initial conditions and the relatively high concentrations that are used in environmental chambers limit the usefulness of the data now available. Cleaner environmental experiments with a wider number of measured species are required. Field measurementsuffer from uncertainties in meteorological conditions, photolysis rates, emissions, initial concentrations and air mass age, effects of heterogeneous and aqueous phase reactions, deposition rates, etc. One possible solution to the problem is comparison of in situ measurements of HO and other radical intermediates with model simulations based upon a reasonably complete dataset of meteorological parameters, photolysis frequencies, NOx, and VOC concentrations [Poppe et al., 994]. These tests and similar ones should be repeated for the new mechanism. Finally, simultaneous field measurements of NOx, PAN, organic nitrates, and HNO3 would be a useful test of the mechanism because the secondary peroxy radical chemistry strongly affects the partitioning between nitrogen species [Stockwell et al., 990, 995; Kirchner and Stockwell, 996a]. Acknowledgments. Support for this research was provided by the predict similar concentrations of O., H202, H2SO4, and German Bundesministerium ffir Bildung, Wissenschaft, Forschung HNO3, but the PAN concentrations are significantly different. und Technologie, projects "Spurenstoffkreisliiufe," contract Because the RACM chemistry is based upon significantly more reliable laboratory measurements and evaluated rate con- 0VSK0/4 and "Der Einfluss biogener Kohlenwasserstoffe auf das troposphiirische Oxidationsverhalten: Abbaumechanismus biogener Alkene fiber Reaktionen mit OH, NO 3 und O3," and by the Commisstants, we strongly recommend this mechanism for use in resion of the European Communities, projects "Biogenic Emissions in gional atmospheric chemistry models. the Mediterranean Area (BEMA)" and "Biogenic VOC Emissions and Research on the development of tropospheric gas-phase re- Photochemistry in the Boreal Regions of Europe (BIPHOREP)." The action mechanisms must be considered to be a continuing process because of significant gaps and uncertainties in the authors thank Karl H. Becker and colleagues at the Bergische Universitfit Wuppertal for helpful discussions. laboratory kinetic data. These important gaps include several topics for research. The reactions of ozone and NO3 with alkenes and their reaction products require better characterization. For these compounds, uncertainties include the nature References Andersson-Sk6d, Y., Updating the chemical scheme for the IVL phoand yield of radicals, hydrogen peroxide, hydroperoxides, and tochemical trajectory model, VL Rep. B 5, Swed. Environ. Res. Inst., G6tegorg, Sweden, 995. organic acids. The oxidation mechanisms of biogenic hydro- Atkinson, R., Gas-phase tropospheric chemistry of organic comcarbons including isoprene and terpenes are very complex and pounds: A review, Atmos. Environ., 24/, -4, 990. poorly understood. The uncertainties in aromatic chemistry Atkinson, R., Kinetics and mechanisms of the gas-phase reactions of are very high; many of the products have not been character- the NO3 radical with organic compounds, J. Phys. Chem. Ref. Data, ized. The initial fate of the aromatic-ho adduct is not known. 20, , 99. Atkinson, R., Gas-phase tropospheric chemistry of organic com- The point during the oxidation cycle where the aromatic ring pounds: A review, J. Phys. Chem. Ref. Data, Monogr. 2, -26, 994. breaks is not known nor is the location of the break known. Atkinson, R., and S. M. Aschmann, OH radical production from the gas-phase reactions of 03 with a series of alkenes under atmospheric conditions, Environ. Sci. Technol., 2, , 993. Atkinson, R., and S. M. Aschmann, Products of the gas-phase reactions of aromatic hydrocarbons: Effect of NO 2 concentration, Int. J. Chem. Kinet., 26, , 994. Atkinson, R., and A. C. Lloyd, Evaluation of kinetic and mechanistic data for modeling of photochemical smog, J. Phys. Chem. Ref. Data, 3, , 984. Atkinson, R., S. M. Aschmann, J. Arey, and W. P. L. Carter, Formation of ring-retaining products from the OH radical-initiated reactions of benzene and toluene, lnt. J. Chem. Kinet., 2,80-82, 989. Atkinson, R., S. M. Aschmann, and J. Arey, Formation of ringretaining products from the OH radical-initiated reactions of o-, m-, and p-xylene, Int. J. Chem. Kinet., 23, -9, 99. Atkinson, R., S. M. Aschmann, and J. Arey, Reactions of OH and NO 3 radicals with phenol, cresols and 2-nitrophenol at 296 _+ 2 K, Environ. Sci. Technol., 26, , 992a. Atkinson, R., D. L. Baulch, R. A. Cox, R. F. Hampson, J. A. Kerr, and J. Troe, Evaluated kinetic and photochemical data for atmospheric chemistry: Supplement IV, IUPAC subcommittee on gas kinetic data evaluation for atmospherichemistry, J. Phys. Chem. Ref. Data, 2, , 992b. Atkinson, R., E. C. Tuazon, I. Bridier, and J. Arey, Reactions of NO3-naphthalene, adducts with 02 and NO2, Environ. Sci. Technol., 28, , 994. Barnes, I., V. Bastian, K. H. Becker, and T. Zhu, Kinetic and products of the reactions of NO 3 with monoalkenes, dialkenes, and monoterpene, J. Phys. Chem., 94, , 990. Barnes, I., B. Klotz, and K. H. Becker, Aromatic hydrocarbon oxidation: New developments, in Proceedings of the/it & Waste Management Association 89th Annual Meeting & Exhibition, June 23-28, 996, Nashville, Tennessee, Air and Waste Manage. Assoc., Pittsburgh, Pa., 996. Becker, E., M. M. Rahman, and R. N. Schindler, Determination of the

32 25,88 STOCKWELL ET AL.: REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM rate constants for the gas phase reactions of NO 3 with H, OH and HO2 radicals at 298 K, Ber. Bunsenges. Phys. Chem., 96, 6-83, 992. Becker, K. H., J. Bechara, and K. J. Brockmann, Studies on the for- mation of H202 on the ozonolysis of alkenes, Atmos. Environ., 2A, 5-6, 993. Hass, H., H. J. Jakobs, and M. Memmesheimer, Analysis of a regional Bierbach, A., Produktuntersuchungen und Kinetik der OH-initiierten model (EURAD) near surface gas concentration predictions using Gasphasenoxidation aromafischer Kohlenwasserstoffe sowie aus- observations from networks, Meteorol. Atmos. Phys., 5, 3-200, gewfihltcr carbonylischer Folgeprodukte, dissertation, Bergische 995. Univ. Gesamthochschule Wuppertal, Wuppertal, Germany, 994. Hjorth, J., C. Lohse, C. J. Nielsen, H. Skov, and G. Restelli, Products Bierbach, A., I. Barnes, K. H. Becker, and E. Wiesen, Atmospheric and mechanisms of the gas phase reactions between NO 3 and a chemistry of unsaturated carbonyls: butendial, 4-oxo-2-pentenal, 3- series of alkenes, J. Phys. Chem., 94, , 990. hexene-2,5-dione, maleic anhydride, 3H-furan-2-one, and 5-methyl- Hjorth, J., J. Notholt, and G. Restelli, A spectroscopic study of the 3H-furan-2-one, Environ. Sci. Technol., 28, 5-29, 994. equilibrium NO 2 + NO 3 + M <--> N205 + M and the kinetics of the Calvert, J. G., and S. Madronich, Theoretical study of the initial prod- O3/N2OsfNO3/NO2/air system, Int. J. Chem. Kinet., 24, 5-65, 992. ucts of the atmospheric oxidation of hydrocarbons, J. Geophys. Res., Horie, O., and G. K. Moortgat, Decomposition pathways of the excited 92, , 98. Criegee intermediates in the ozonolysis of simple alkenes, Atmos. Calvert, J. G., and W. R. Stockwell, Acid Generation in the tropo- Environ., 25.4, , 99. sphere by gas phase chemistry, Environ. Sci. Techno!.,, 428A- Japar, S. M., and H. Niki, Gas-phase reactions of the nitrate radical 443A, 983. with olefins, J. Phys. Chem., 9, , 95. Cantrell, C. A., W. R. Stockwell, L. G. Anderson, K. L. Busarow, Jenkin, M. E., S. M. Saunders, and M. J. Pil!ing, Constructing a de- D. Perner, A. Schmeltekopf, J. G. Calvert, and H. S. Johnston, tailed chemical mechanism for use in tropospheri chemistry mod- Kinetic Study of the NO3-CH20 Reaction and its possible role in els, report, AEA Technol., Natl. Environ. Technol. Cent., Culham, nighttime tropospherichemistry, J. Phys. Chem., 89, 39-46, 985. England, 996. Cantrell, C. A., J. A. Davidson, A. H. McDaniel, R. E. Shetter, and Johnston, H. S., C. A. Cantrell, and J. G. Calvert, Unimolecular de- J. G. Calvert, Temperature-dependent formaldehyde cross sections composition of NO 3 to form NO and 0 2 and a review of N2Os/NO 3 in the near-ultraviolet spectral region, J. Phys. Chem., 94, , kinetics, J. Geophys. Res., 9, , Joseph, J. H., W. J. Wiscombe, and J. A. Weinman, The delta- Carter, W. P. L., and F. W. Lurmann, Evaluation of the RADM Gas- Eddington approximation for radiative flux transfer, J. Atmos. Sci., Phase Chemical Mechanism, U.S. Environmental Protection Agency 33, , 96. Cooperative Agreement CR Statewide Air Pollut. Res. Kirchner, F. and W. R. Stockwell, Effect of peroxy radical reactions on Cent., Univ. of Calif., Riverside, 989. the predicted concentrations of ozone, nitrogenous compounds, and Carter, W. P. L., D. Luo, I. L. Malkina, D. Fitz, The University of radicals, J. Geophys. Res., 0, 2,00-2,022, 996a. (Correction to California, Riverside Environmental Chamber Data Base for Evaluat- "Effect of peroxy radical reactions on the predicted concentrations ing Oxidant Mechanisms, vol. and 2, Univ. of Calif., Riverside, of ozone, nitrogenous compounds and radicals" by F. Kirchner and 995. W. R. Stockwell, J. Geophys. Res., in press, 99.) Chang, J. S., et al., NAPAP report 4, the Regional Acid Deposition Kirchner, F., and W. R. Stockwell, Isoprene and terpene oxidation Model and Engineering Model, Acidic Deposition: State-of-Science/ mechanisms, in Proceedings of the EUROTRAC Symposium '96, Technology, vol., Emissions, Atmospheric Processes, and Deposi- Transport and Transformation of Pollutants in the Troposphere, Voltion, edited by P.M. Irving, 92 pp., Natl. Acid Precip. Assess. ume : Clouds, Aerosols, Modelling and Photooxidants, edited by Program, Washington, D.C., 99. P.M. Borrell et al., pp. 5-9, Comput. Mech. Publ., Southamp- Cvetanovic, R. J., Chemical kinetic studies of atmospheric interest, ton, England, 996b. paper presented at 2th International Symposium on Free Radicals, Knispel, R., R. Koch, M. Siese, and C. Zetzsch, Adduct formation of Laguna Beach, Calif., Jan. 4-9, 96. Cvetanovic, R. J., Evaluated chemical kinetic data for the reactions of OH radicals with benzene, toluene and phenol and consecutive atomic oxygen O(3p) with unsaturated hydrocarbons, J. Phys. Chem. reactions of the adducts with NOx and 02, Ber. Bunsenges. Phys. Chem., 94, 35-39, 990. Ref. Data, 6, , 98. Dagaut, P., R. Liu, T. J. Wallington, and M. J. Kurylo, The gas phase Koch, S., and G. K. Moortgat, Photolysis of methylgloxyal, in.4jr reactivity of aliphatic polyethers towards OH radicals: Measure- Pollution Research Report Series Research Report 5, Homogeneous ments and predictions, Int. J. Chem. Kinet., 2, 3-80, 989. and Heterogeneous Chemical Processes in the Troposphere, Joint Davidson, J. A., C. A. Cantrell, R. E. Shetter, A. H. McDaniel, and Workshop LACTOZ-HALIPP, September 25-26, Strasbourg, France, J. G. Calvert, The NO 3 radical decomposition and NO 3 scavenging pp , Off. for Off. Publ. of the Eur. Commun., Luxembourg, 996. in the troposphere, J. Geophys. Res., 95, 3,963-3,969, 990. Davis, H. F., B. Kim, H. S. Johnston, and Y. T. Lee, Dissociation Kwok, E. S.C., and R. Atkinson, Estimation of hydroxyl radical rate energy and photochemistry of NO3, J. Phys. Chem., 9, , constants for gas phase organic compounds using a structure reactivity relationship: An update, Atmos. Environ., 29, , 995. 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. Le Bras, G. (Ed), Chemical Processes in Atmospheric Oxidation, Spring- Molina, Chemical kinetics and photochemical data for use in stratospheric modeling, Eval., Natl. Aeronaut. and Space Admin., Jet Propul. Lab., Calif. Inst. of Technol., Pasadena, 994. Finlayson-Pitts, B. J., and J. N. Pitts Jr., Atmospheric Chemistry: Fundamentals and Experimental Techniques, John Wiley, New York, 986. Gardner, E. P., P. D. Sperry, and J. G. Calvert, Photodecomposition of acrolein in O2-N 2 mixtures, J. Phys. Chem., 9, , 98. Gery, M. W., G. Z. Whitten, J.P. Killus, and M. C. Dodge, A photochemical mechanism for urban and regional scale computer modeling, J. Geophys. Res., 94, 2,925-2,956, 989. Giguere, P. A., and A. W. Olmos, Sur le spectre ultraviolet de l'acide perac tique et l'hydrolyse des peracetates, Can. J. Chem., 34, , 956. Grosjean, D., E. L. Williams II, and E. Grosjean, Gas phase reaction of the hydroxyl radical with the unsaturated peroxyacyl nitrate CH2=(CH3)C(O)OONO2, Int. J. Chem. Kinet., 25, , 993a. Grosjean, D., E. Grosjean, and E. L. Williams II, The reaction of ozone with MPAN, CH2--C(CH3)C(O)OONO2, Environ. Sci. Technol., 2, , 993b. Hall, I. W., R. P. Wayne, R. A. Cox, M. E. Jenkin, and G. D. Hayman, Kinetics of the reaction of NO 3 with HO2, J. Phys. Chem., 92, O54, 988. er-verlag, New York, 99. Leone, J. A., and J. H. Seinfeld, Comparative analysis of chemical reaction mechanisms for photochemical smog, Atmos. Environ., 9, , 985. 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 radi- cals: Kinetics, spectroscopy and tropospherichemistry, Atmos. Environ., 26A, , 992. Lurmann, F. W., A. C. Lloyd, and R. Atkinson, A chemical mechanism for use in long-range transport/acid deposition computer modeling, J. Geophys. Res., 9, 0,905-0,936, 986. Madronich, S., Photodissociation in the atmosphere,, Actinic flux and the effects of ground reflections and clouds, J. Geophys. Res., 92, , 98. Martinez, R. D., A. A. Buitrago, N. W. Howell, C. H. Hearn, and J. A. Joens, The near U.V. absorption spectra of several aliphatic aldehydes and ketones at 300 K, Atmos. Environ., 26A, 85-92, 992. McRae, G. J., W. R. Gooden, and J. H. Seinfeld, Mathematical Mod-

33 STOCKWELL ET AL.: REGIONAL ATMOSPHERIC CHEMISTRY MECHANISM 25,89 eling of Photochemical Air Pollution, EQL Rep. 8, Environ. Qual. Lab., Calif. Inst. of Technol., Pasadena, 982. Mellouki, A., G. Le Bras, and G. Poulet, Kinetics of the reactions of NO3 with OH and HO2, J. Phys. Chem., 92, , 988. Mellouki, A., R. K. Talukdar, A.M. R. P. Bopegedera, and C. J. Howard, Study of the kinetics of the reactions of NO3 with HO 2 and OH, Int. J. Chem. Kinet., 25, 25-39, 993. Middleton, P., W. R. Stockwell, and W. P. L. Carter, Aggregation and analysis of volatile organic compound emissions for regional modeling, Atmos. Environ., 24A, 0-33, 990. Moortgat, G. K., W. Klippel, K. H. Mobus, W. Seiler, and P. Warneck, Laboratory measurement of photolytic parameters for formaldehyde, Final Rep. FAA-EE-80-4, Off. of Environ. and Energy, Fed. Aviat. Admin., Washington, D.C., 980. Moortgat, G. K., et al., Laboratory studies of peroxy radicals, carbonyl compounds and ozonolysis reactions of tropospheric importance, in Chemical Processes in Atmospheric Oxidation, edited by G. Le Bras, pp , Springer-Verlag, New York, 99. Morris, E. D., Jr., and H. Niki, Reaction of the nitrate radical with acetaldehyde and propylene, J. Phys. Chem., 8, 33-3, 94. National Center for Atmospheric Research (NCAR) Preliminary Evaluation Studies with the Regional Acid Deposition Model (P, ADM), NCAR Tech. Note NCAR/TN-265+STR, Natl. Cent. for Atmos. Res., Boulder, Colo., 986. Paulson, S. E., R. C. Flagan, and J. H. Seinfeld, Atmospheric photooxidation of isoprene, I, The hydroxyl radical and ground state atomic oxygen reactions, Int. J. Chem. Kinet., 24, 9-0, 992a. Paulson, S. E., R. C. Flagan, and J. H. Seinfeld, Atmospheric photooxidation of isoprene, II, The ozone-isoprene reaction, Int. J. Chem. Kinet., 24, 03-25, 992b. Platt, U., D. Perner, A.M. Winer, G. W. Harris, and J. N. Pitts Jr., Detection of NO3 in the polluted troposphere by differential optical absorption, Geophys. Res. Lett.,, 89-92, 980. Poppe, D., et al., Comparison of measured OH concentrations with model calculations, J. Geophys. Res., 99, 6,633-6,642, 994. Poppe, D., et al., Gas-phase reactions in atmospheric chemistry and transport models: A model intercomparison, report, EUROTRAC Int. Sci. Secr., Garmisch-Partenkirchen, Germany, 996. Russell, A. G., G. R. Cass, and J. H. Seinfeld, On some aspects of nighttime atmospheric chemistry, Environ. Sci. Technol., 20, 6-2, 986. Seinfeld, J. H., Atmospheric Chemistry and Physics of Air Pollution, John Wiley, New York, 986. Skov, H., J. Hjorth, C. Lohse, C. J. Jensen, and G. Restelli, Products and mechanisms of the reactions of the nitrate radical (NO3) with isoprene,,3-butadien and 2,3-dimethyl-l,3-butadien in air, Atmos. Environ., 26A, 2-283, 992. Staffelbach, T. A., J. J. Orlando, G. S. Tyndall, and J. G. Calvert, The UV-visible absorption spectrum and photolysis quantum yields of methylglyoxal, J. Geophys. Res., 00, 4,89-4,98, 995. Stockwell, W. R., and J. G. Calvert, The mechanism of NO 3 and HONO formation in the nighttime chemistry of the urban atmosphere, J. Geophys. Res., 88, , 983. Stockwell, W. R., and D. Kley, The Euro-RADM Mechanism: A gasphase chemical mechanism for European air quality studies, Forschungszentrum, Jfilich, Jfilich, Germany, 994. Stockwell, W. R., and F. W. Lurmann, Intercomparison of the ADOM and RADM gas-phase chemical mechanisms, report, Electr. Power Res. Inst., Palo Alto, Calif., 989. Stockwell, W. R., P. Middleton, J. S. Chang, and X. Tang, The second generation regional Acid Deposition Model chemical mechanism for regional air quality modeling, J. Geophys. Res., 95, 6,343-6,36, 990. 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, Atmos. Environ., 29, , 995. Sverdrup, G. M., C. W. Spicer, and G. F. Ward, Investigation of the gas phase reaction of dinitrogen pentoxide with water vapor, Int. J. Chem. Kinet., 9, 9-205, 98. Tuazon, E. C., and R. Atkinson, A product study of the gas phase reaction of methacrolein with the OH radical in the presence of NOx, Int. J. Chem. Kinet., 22, , 990. Tuazon, E. C., R. Atkinson, and W. P. L. Carter, Atmospheric chem- istry of cis- and trans-3-hexene-2,5-dione, Environ. Sci. Technol., 9, , 985. Tuazon, E. C., J. Arey, R. Atkinson, and S. M. Aschmann, Gas-phase reactions of 2-vinylpyridine and styrene with OH and NO 3 radicals and 03, Environ. Sci. Technol., 2, , 993. Uselman, W. M., S. Z. Levine, W. H. Chan, J. G. Calvert, and J. H. Shaw, A kinetic study of the mechanism of peroxynitric acid formation in irradiated mixtures of chlorine, hydrogen, nitrogen dioxide and nitric oxide in air, in Nitrogenous Air Pollutants, Chemical and Biological Implications, edited by D. Grosjean, Ann Arbor Sci., Ann Arbor, Mich., 99. Vogel, B., F. Fiedler, and H. Vogel, Influence of topography and biogenic volatile organic compounds emission in the state of Baden- Wfirttemberg on ozone concentrations during episodes of high air temperatures, J. Geophys. Res., 00, 22,90-22,928, 995. Wallington, T. J., P. Dagaut, R. Liu, and M. J. Kurylo, The gas-phase reactions of hydroxyl radicals with a series of esters over the temperature range K, Int. J. Chem. Kinet., 20, -86, 988. Warneck, P., Chemistry of the Natural Atmosphere, Academic, San Diego, Calif., 988. Wayne, R. P., et al., The nitrate radical: Physics, chemistry, and the atmosphere, Atmos. Environ., 25A, -203, 99. Wiesen, E., OH-initiierte Oxidation von p-xylol und Produktuntersu- chungen an ausgewaehlten Carbonylverbindungen (995), dissertation, Bergische Univ. Gesamthochschule Wuppertal, Wuppertal, Germany, 995. Wiesen, E., I. Barnes and K. H. Becker, Study of the OH-initiated degradation of the aromatic photooxidation product 3,4-dihydroxy- 3-hexene-2,5-dione, Environ. Sci. Technol., 29, 0-6, 995. Williams, D.C., L. N. O'Rji, and D. A. Stone, Kinetics of the reactions of OH radicals with selected acetates and other esters under simulated atmospheri conditions, Int. J. Chem. Kinet., 25, , 993. F. Kirchner, M. Kuhn, and W. R. Stockwell, Fraunhofer Institute for Atmospheric Environmental Research, Kreuzeckbahnstr. 9, D-8246 Garmisch-Partenkirchen, Germany. ( stockwel@ifu.fhg.de) S. Seefeld, EAWAG, Swiss Federal Institute for Environmental Science and Technology, ETH Zfirich, CH-8600 Dfibendorf, Switzerland. (Received December 6, 996; revised March, 99; accepted March, 99.)