J.-F. Müller, K. Ceulemans, S. CompernolleBelgian Institute for Space Aeronomy, Brussels, Belgium

AGU Fall Symposium, Dec. 2007

Factors influencing SOA yields in the simulation of α-pinene photooxidation

experiments

L. Vereecken, J. PeetersKatholieke Universiteit Leuven, Belgium

COOH

COOH

O

CHO

O

COOH

O3, OH, NO3etc.

volatile compounds

semi-volatile compounds

etc.

COOHCOOH

O

CHO

O

COOH

Secondary Organic Aerosol (SOA)

Kp,i (partitioning coefficients)

etc.

Experiments indicate that heterogeneous or particle-phase chemistry generates high-molecular weight compounds, enhancing SOA formation

Previous model studies found necesary to increase the partitioning coefficients by several orders of magnitude in order to match SOA yields from laboratory experiments

Oxidation by OH• mechanism based

on advanced theoretical calculations

• Peeters et al. 2001; Vereecken and Peeters, 2004; Fantechi et al., 2002

• important updates from Vereecken et al., PCCP, 2007

Ozonolysis• mechanism still

preliminary, but theoretical work is in progress

• pathways proposed so far to explain some key observed products (pinic acid, hydroxy pinonic acid) have been demonstrated to be negligible

• As a consequence, the yield of organic acids is too low in this mechanism

• Capouet et al., JGR, in press

Secondary chemistry Explicit part of mechanism : degradation down to primary products +

degradation of pinonaldehyde But the degradation down to CO2 would require maybe billions of

reactions (Aumont et al.)

• Combination of semi-generic chemistry (for high-yield compounds) and generic chemistry (for the rest)• Semi-generic compounds are lumped compounds for which the carbon number and all functionalities are defined, not their precise structure• Generic compounds are lumped compounds for which one functionality (RO2, ROOH, RONO2 etc.) is defined, and further subdivided into 4 volatility classes

Aerosol formation: Partitioning theory (Pankow, 1994)

12

,,

1 4

3 3om

on iAvo g i i i

MW a ak

N D

Aerosol radius

Accomodation coefficient (assumed > 0.1)

Adsorption rate:

12

,, 6 0

, 0 ,

10760

10p i on Avo

p ig i om i i L i off om

C k NRTK

C M MW p k MW

Gas- and particulate phase concentrations at equilibrium

Organic aerosol concentration

Adsorption and desorption rates

Saturation vapour pressure

Partioning coefficient:

obtained from a group contribution method (Capouet and Müller, 2006)

Overall alpha-pinene mechanism

5000 reactions, 1300 compounds (including the gas-aerosol partitioning reactions)

Capouet et al., J. Geophys. Res., in press complete mechanism can be explored at

http://www.oma.be/TROPO/boream/boreammodel KPP/Rosenbrock as chemical solver

http://www.oma.be/TROPO/boream

Ozone formation: simulation of experiments

Ozone (ppm)

Kamens and Jaoui 2001

SAPRC, Carter et al. 2000

D(O3 -NO) (ppm)

O3 (ppm)

SOA formation: Photooxidation experimentsΔVOC

(ppb)ΔVOC/ NOx

OH:O3:NO3

(% )T (K)

J(NO2) (104 /s)

Nozière et al., 1999

(4 experiments)

300 -

1500

0.09–0.52 100:0:0 298 3.5 (lamp)

Kamens et al., 2001(2 experiments)

~ 980 ~ 2 42:44:13

295 - 315 12-35 (sun)

Hoffman et al., 1997(7 experiments)

19 - 95 0.17–0.78 44:31:20

309 - 321 83 (sun)

Takekawa et al., 2003(6 experiments)

55 – 196

1.5-1.9 53:42:4

283 - 303 40 (lamp)

Ng et al., 2006(1 experiment)

108 1.1 62:22:16

293 10 (lamp)

Presto et al., 2005(8 experiments)

11 – 205

0.3 - 30 6:82:11 295 300 (lamp)

Results: SOA yields

Stark contrast with previous modeling studies which required orders of magnitude enhancements of the partitioning coefficients in order to match observed SOA yields

•Simulations with additional acid formation channels in ozonolysis mechanism lead to better agreement in some (not all) low-VOC experiments

• Simulations with additional particle-phase association reactions (ROOH+R’CHO) has little impact except in high-VOC ozonolysis experiments

Time series

Alpha-pinene decay is well reproduced in all cases

SOA formation occurs too late in the simulation of a few Nozière et al. experiments

SOA composition

Particulate compounds are multifunctional

Generic compounds are significant but not dominant in modeled SOA at sampling time

Hydroperoxides make up 25% of compounds in a high-NOx Nozière et al. experiment

Acids dominant (>50%) in ozonolysis experiments (Presto et al.)

Conclusions Low-volatility hydroperoxides from OH-initiated oxidation contribute

significantly to SOA even in presence of NO Theoretical and laboratory work needed to elucidate origin of acids in

ozonolysis Particle-phase association reactions (ROOH+R’CHO) have only a

small impact, except in some conditions (SOA yields enhanced by about 1/3 in high-VOC Presto et al. experiments); more laboratory work needed to investigate the elementary steps

Huge uncertainties in vapor pressures and activity coefficients, chemistry

Near future: calculate activity coefficients, fine-tune the mechanism and the vapor pressures in order to improve agreement

In case of satisfactory results, develop a reduced mechanism and SOA parameterization for implementation in a global model