Atmospheric Chemical Mechanisms, Davis, December 2018 Structure-activity relationships for the development of MCM/GECKOA mechanisms Acknowledgement: Bernard Aumont Richard Valorso, Marie Camredon LISA Mike Jenkin Atmospheric Chemical Services Andrew Rickard, Peter Brauer, M. Newland, Mat Evans University of York Camille Mouchel-Vallon, Sasha Madronich, John Orlando NCAR
Why explicit organic chemical mechanisms? Parent hydrocarbon 1 st generation species 2 nd generation species CO 2
Why explicit organic chemical mechanisms? Fundamental understanding of the HOx/NOx/Ox chemistry, SOA formation Reference scheme to examine the reliability of simplified (reduced) schemes. Objective tools to explore our ignorance through comparisons with observations
How big are explicit chemical schemes? Two problems : Explicit schemes are too large to be reasonably written by hand The quantity of physical and chemical data needed to develop explicit schemes is far in excess of available experimental data. Aumont et al., ACP, 2005 Data processing tools are required to: Assimilate all the data provided by laboratory studies Codify the various estimation methods Generate consistent and comprehensive multiphase oxidation schemes on a systematic basis Development of GECKO-A Generator for Explicit Chemistry and Kinetics of Organics in the Atmosphere
The chemical scheme generator precursors Experimental data & estimation methods (SAR) GECKO-A Computing code: generate detailed mechanism (reactions, mass transfer ) Protocol: select chemical pathways provide kinetic & thermodynamic data Explicit chemical schemes + properties (Psat, H,...) parent hydrocarbon oxidation 1 st oxidation 2 nd oxidation generation generation organics organics CO 2 Particle phase 1 st generation SVOC 2 nd generation SVOC Condensed phase assumed to behave as an ideal well mixed liquid homogeneous phase. P vap estimated for each intermediate using structure/properties relationship.
Flow diagram of the generator New reaction products added to the stack VOCn VOC2 VOC1 OH reaction RH + OH RO 2 >C=C< + OH >C(OH)C(O 2 )< O 3 reaction >C=C< + O 3 >C=O + Criegee Criegee products + OH NO 3 reaction RCHO + NO 3 R(O)OO >C=C< + NO 3 >C(ONO 2 )C(O 2 )< Photolysis RCHO + hn R + HCO RONO 2 + hn RO + NO 2 ROOH + hn RO + OH Thermal decomposition RCO(OONO 2 ) RCO(OO)+NO 2 C-H? >C=C<? -CHO? >C=O? -ONO 2? -OOH? -C(O)OONO 2? no Radical? yes RO2? Peroxy alkyl chemistry RO 2 +NO RO+NO 2 RO 2 +NO RONO 2 RO 2 +NO 3 RO+NO 2 +O 2 RO 2 +HO 2 ROOH + O 2 RO 2 +RO 2 products RCO3? Peroxy acyl chemistry RCO 3 +NO R+CO2+NO 2 RCO 3 +NO 2 RCO(OONO 2 ) RCO 3 +NO 3 R+CO 2 +NO 2 +O 2 RCO 3 +HO 2 RCO(OOH) + O 2 RCO 3 +HO 2 RCO(OH) + O 3 RCO 3 +RO 2 products RO? Alcoxy chemistry >CH(O )+O 2 >C=O + HO 2 >CH(O )C 2 C(O.)< >C(O ) C 2 C(OH)< >C(O )-R >C=O + R Treat next species in the stack
The Magnify Project Mechanisms for Atmospheric chemistry: GeneratioN, Interpretation and Fidelity Objective: Provide state-of-the science atmospheric chemical mechanisms based on a sustainable approach project ANR-14-CE01-001 grant NE/M013448/1
The Magnify project Feedbacks from external users Feedbacks from external users Development of the chemical protocol Development of the reduction protocol sensitivity tests Chemical rules Error quantifications Simplification rules Automatic generation of a MCM/GECKOA mechanism Target : ~ 10 000 species, ~ 100 primary VOC Optimized mechanisms Development of website facility to : - disseminate protocols, mechanisms and tools - receive review from the scientific community on protocols
SAR in the MCM/GECKOA chemical protocol New reaction products added to the stack VOCn VOC2 VOC1 OH reaction RH + OH RO Jenkin M. et al., 2018a, 2 2018b >C=C< + OH >C(OH)C(O 2 )< O 3 reaction Newland >C=C< + M., O 3 Mouchel-Vallon >C=O + CriegeeC. et Criegee al., in prep products + OH NO 3 reaction NO 3 reaction RCHO + NO 3 R(O)OO Kerdouci J., et al., 2014 >C=C< + NO 3 >C(ONO 2 )C(O 2 )< Photolysis Photolysis RCHO + hn R + HCO Brauer RONO 2 P. + hn et al., RO in prep + NO 2 ROOH + hn RO + OH Thermal decomposition Jenkin M. et al., in prep Thermal decomposition RCO(OONO 2 ) RCO(OO)+NO 2 C-H? >C=C<? -CHO? >C=O? -ONO 2? -OOH? -C(O)OONO 2? no Radical? yes RO2? Peroxy alkyl chemistry Peroxy alkyl chemistry RO 2 +NO RO+NO 2 RO 2 +NO RONO Jenkin M. et al., in prep 2 RO 2 +NO 3 RO+NO 2 +O 2 RO 2 +HO 2 ROOH + O 2 RO 2 +RO 2 products RCO3? Peroxy acyl chemistry Peroxy acyl chemistry RCO 3 +NO R+CO2+NO 2 RCO 3 +NO 2 RCO(OONO 2 ) Jenkin RCO 3 +NO M. et 3 al., R+CO in prep 2 +NO 2 +O 2 RCO 3 +HO 2 RCO(OOH) + O 2 RCO 3 +HO 2 RCO(OH) + O 3 RCO 3 +RO 2 products RO? Alkoxy chemistry Vereecken L. and Peeters J., 2009 Vereecken L. and Peeters J., 2010 Alkoxy chemistry >CH(O )+O 2 >C=O + HO 2 >CH(O )C 2 C(O.)< >C(O ) C 2 C(OH)< >C(O )-R >C=O + R Treat next species in the stack
SAR in the MCM/GECKOA chemical protocol The current version GECKOA generates New reaction products mechanism added to for the alkane, stack alkene and (monocyclic) aromatic compounds and their oxidation products (alcohol, ketone, ester, ether, nitrate, nitro, hydroperoxide, VOCn ). VOC2 OH reaction RH + OH RO Jenkin M. et al., 2018a, 2 2018b >C=C< + OH >C(OH)C(O 2 )< NO 3 reaction RCHO + NO 3 R(O)OO Kerdouci J., et al., 2014 >C=C< + NO 3 >C(ONO 2 )C(O 2 )< Photolysis RCHO + hn R + HCO Brauer RONO 2 P. + hn et al., RO in prep + NO 2 ROOH + hn RO + OH Thermal decomposition Jenkin M. et al., in prep Thermal decomposition RCO(OONO 2 ) RCO(OO)+NO 2 C-H? >C=C<? -CHO? >C=O? -ONO 2? -OOH? -C(O)OONO 2? no VOC1 See poster 4 by Andrew Rickard et al. for additional Radical? information about the Magnify project and chemical protocol in MCM/GECKOA O 3 reaction Newland >C=C< + M., O 3 Mouchel-Vallon >C=O + CriegeeC. et Criegee al., in prep products + OH NO 3 reaction Photolysis yes RO2? RCO3? RO? Treat next species in the stack Peroxy alkyl chemistry Peroxy alkyl chemistry RO 2 +NO RO+NO 2 RO 2 +NO RONO Jenkin M. et al., in prep 2 RO 2 +NO 3 RO+NO 2 +O 2 RO 2 +HO 2 ROOH + O 2 RO 2 +RO 2 products Peroxy acyl chemistry Peroxy acyl chemistry RCO 3 +NO R+CO2+NO 2 RCO 3 +NO 2 RCO(OONO 2 ) Jenkin RCO 3 +NO M. et 3 al., R+CO in prep 2 +NO 2 +O 2 RCO 3 +HO 2 RCO(OOH) + O 2 RCO 3 +HO 2 RCO(OH) + O 3 RCO 3 +RO 2 products Alkoxy chemistry Vereecken L. and Peeters J., 2009 Vereecken L. and Peeters J., 2010 Alkoxy chemistry >CH(O )+O 2 >C=O + HO 2 >CH(O )C 2 C(O.)< >C(O ) C 2 C(OH)< >C(O )-R >C=O + R
Some SAR assessment used in the MCM/GECKO protocols
VOC+OH Estimation of rate constants based on SAR The experimental database for aliphatic species: 147 hydrocarbons 213 monofunctional species (alcohol, aldehyde, ketone, nitrate, ether, ester, nitro, carboxylic acid, hydroperoxide) 115 di-functional species 8 tri-functional species [Jenkin et al., ACP, 2018]
VOC+OH Estimation of rate constants based on SAR agreement of the calculated k within 40 % no significant bias [Jenkin et al., ACP, 2018]
VOC+OH Estimation of rate constants based on SAR the reliability of the SARs decreases with the number of functional groups on the carbon skeleton [Jenkin et al., ACP, 2018]
VOC+OH Estimation of rate constants based on SAR agreement of the calculated k within 20 % [Jenkin et al., ACP, 2018]
VOC+OH Estimation of rate constants based on SAR the reliability of the SARs decreases with the number of functional groups on the carbon skeleton [Jenkin et al., ACP, 2018]
Assessment of branching ratio estimation based on SAR (VOC+OH reaction) Example: isobutane branching ratio (@298K): k t SAR: Exp: α t = k t k SAR = 0. 739 α p = 0. 74 ± 0. 08 k t = 1. 53 10 12 cm 3 s 1 + OH 3k p SAR: α p = 3k p k SAR = 0. 261 Exp: α p = 0. 26 ± 0. 08 3k p = 5. 41 10 13 cm 3 s 1 k SAR = 3k p + k t = 2. 07 10 12 cm 3 s 1 k Exp. = 2. 10 ± 0. 42 10 12 cm 3 s 1
Assessment of branching ratio estimation based on SAR (VOC+OH reaction) SAR: Exp: α t = k t k SAR = 0. 739 α p = 0. 74 ± 0. 08 SAR: Exp: α p = 3k p k SAR = 0. 261 α p = 0. 26 ± 0. 08 Database for branching ratios (Acknowledgment: John Orlando, NCAR) - 135 branching ratios for 79 species (C 2 without halogen) - Alkane, mono-functional species (alcohol, ether, ester, carbonyls, acid) and multifunctional species - Values inferred (mostly) from product distributions provided in Calvert et al. books
alkane N = 11 RMSE = 0.02 Database for branching ratios (Acknowledgment: John Orlando, NCAR) - 135 branching ratios for 79 species (C 2 without halogen) - Alkane, mono-functional species (alcohol, ether, ester, carbonyls, acid) and multifunctional species - Values inferred (mostly) from product distributions provided in Calvert et al. books
Assessment of branching ratio estimation based on SAR (VOC+OH reaction) alkane alcohol carbonyl + acid N = 11 RMSE = 0.02 N = 24 RMSE = 0.12 N = 15 RMSE = 0.14 ester ether multi functional N = 12 RMSE = 0.20 N = 10 RMSE = 0.14 N = 31 RMSE = 0.19
Assessment of branching ratio estimation based on SAR (VOC+OH reaction) alkane alcohol carbonyl + acid N = 11 RMSE = 0.02 N = 24 RMSE = 0.12 N = 15 RMSE = 0.14 The reliability of the estimated branching ratios decreases with the number of functional groups: RMSE is 2% for alkanes, 15% for monofunctional species and 19 % for di-functional species. Simulation ester of SOA production ether show high sensitivity to multi branching functional ratios. Additional constrains for multifunctional species needed. N = 12 RMSE = 0.20 N = 10 RMSE = 0.14 N = 31 RMSE = 0.19
VOC+NO3 Estimation of rate constants based on SAR Selected SAR for the MCM/GECKO protocol: group contribution method by Kerdouci et al., Atmos. Env., 2014 The experimental database for aliphatic species: 100 hydrocarbons 99 monofunctional species (alcohol, aldehyde, ketone, ether, ester) 17 multifunctional species Acknowledgments: Carter et al., CRC project, 2017
VOC+NO3 Estimation of rate constants based on SAR agreement of the calculated k within a factor 3 no significant bias
VOC+NO3 Estimation of rate constants based on SAR the reliability of the SARs decreases with the number of functional groups on the carbon skeleton
MCM 3.2 vs. GECKOA/MCM
Development of scenarios to compare mechanisms Objectives & Constrains Should represent a set of conditions, from remote to urban environments Oxidation mechanism of each VOC must be tested under identical conditions: - - - Same chemistry for inorganic species The tested mechanism should not interfere with the chemical scenario Chemical conditions must be buffered Initial conditions with only tiny amount of the tested VOC
Development of scenarios to compare mechanisms Constrained concentrations: Remote Remote continental Continental Polluted continental Urban NOx (ppb) O 3 (ppb) CH 4 (ppb) CO (ppb) NMHC k OH (s -1 ) Add. HCHO (ppb) Org. Aero. (µg m -3 ) 0.010 0.025 0.5 2 20 40 40 40 40 40 1750 1750 1750 1750 1750 120 120 150 200 300 0 1 6 9 13 0 0 2 5 10 10 10 10 10 10 Simulation conditions: - Solar zenith angle: 45 (fixed) - Temperature: 298 K - Relative humidity: 70 % - Deposition (avoid accumulation) : HNO 3, H 2 O 2, CH 3 OOH
Butane oxidation GECKOA/MCM scheme generation: - Chemistry ignored if P vap < 10-13 atm - Functional isomers lumped if yield < 5% - Up to 8 generations accounted 2 230 species 18 000 reactions MCM scheme: 171 species 552 reactions Urban scenario GECKOA / MCM MCM v3.2 Continental scenario Remote scenario SOG HC SOG CO
Butane oxidation top contributors to the C budget CONTINENTAL SCENARIO CH3CHO CO2 CO C4H10 CH3COC2H5
Butane oxidation top contributors to the C budget CH3CH2(OOH) PAN CH3(OOH) CH3CHO CO2 HCHO CO C4H10
Butane oxidation top contributors to the C budget CONTINENTAL SCENARIO
Butane oxidation top contributors to the C budget Urban scenario Continental scenario Remote scenario
Dodecane oxidation GECKOA/MCM scheme generation: - Chemistry ignored if P vap < 10-13 atm - Functional isomers lumped if yield < 5% - Up to 8 generations accounted 1 10 5 species 7 10 5 reactions MCM scheme: 400 species 1 200 reactions Urban scenario Continental scenario Remote scenario GECKOA / MCM MCM v3.2 CO SOA SOA CO SOA HC SOG SOA CO
Dodecane oxidation organic carbon budget CONTINENTAL SCENARIO Carbon chain length gas phase only GECKOA / MCM MCM v3.2 C12 C10 C8 C6 C4 C6 C12 C2 C12 C3 MCM favors fragmentation routes compared to MCM/GECKOA
Dodecane oxidation org. functional group distribution R i/c = CONTINENTAL SCENARIO Functional group per carbon ratio gas phase only Number of carbons bearing function i Total number of organic carbons GECKOA / MCM MCM v3.2 25 % 25 % # functional groups # organic C # functional groups # organic C
Dodecane oxidation org. functional group distribution Functional group per carbon ratio gas phase only R i/c = CONTINENTAL SCENARIO Number of carbons bearing function i Total number of organic carbons GECKOA / MCM MCM v3.2 -ONO 2 -OH -ONO2 -OH -OOH >C=O -OOH >C=O -CHO -PAN -C(O)OH -OH -OOH >C=O -CHO -CHO -C(O)OOH -PAN
Dodecane oxidation oxidation trajectories CONTINENTAL SCENARIO GECKOA / MCM Aqueous phase (particles) time = 0 MCM v3.2 Aqueous phase (particles) Gas phase Gas phase Org. particles Org. particles Phase partitioning (> 99 %) assuming: - C OA = 10 µg m -3, MW aerosol = 200 g mol -1 - LWC = 10 µg m -3 (deliquescent particle)
Dodecane oxidation oxidation trajectories CONTINENTAL SCENARIO GECKOA / MCM Aqueous phase (particles) time = 1 t parent MCM v3.2 Aqueous phase (particles) Gas phase Gas phase Org. particles Org. particles Phase partitioning (> 99 %) assuming: - C OA = 10 µg m -3, MW aerosol = 200 g mol -1 - LWC = 10 µg m -3 (deliquescent particle)
Dodecane oxidation oxidation trajectories CONTINENTAL SCENARIO GECKOA / MCM Aqueous phase (particles) time = 2 t parent MCM v3.2 Aqueous phase (particles) Gas phase Gas phase Org. particles Org. particles Phase partitioning (> 99 %) assuming: - C OA = 10 µg m -3, MW aerosol = 200 g mol -1 - LWC = 10 µg m -3 (deliquescent particle)
Dodecane oxidation oxidation trajectories CONTINENTAL SCENARIO GECKOA / MCM Aqueous phase (particles) time = 5 t parent MCM v3.2 Aqueous phase (particles) Gas phase Gas phase Org. particles Org. particles Phase partitioning (> 99 %) assuming: - C OA = 10 µg m -3, MW aerosol = 200 g mol -1 - LWC = 10 µg m -3 (deliquescent particle)
Dodecane oxidation oxidation trajectories CONTINENTAL SCENARIO GECKOA / MCM Aqueous phase (particles) time = 10 t parent MCM v3.2 Aqueous phase (particles) Gas phase Gas phase Org. particles Org. particles Phase partitioning (> 99 %) assuming: - C OA = 10 µg m -3, MW aerosol = 200 g mol -1 - LWC = 10 µg m -3 (deliquescent particle)
SOG normalized reactivity Ox produced per C Alkane series comparison of integrative variables CONTINENTAL SCENARIO SOG normalized reactivity = k i OH SOG i t OH k parent parent 0 Consistent results provided by MCM/GECKOA and MCM3.2 for short carbon skeletons. Discrepancies in oxidative trajectories increase with chain length. Ox produced per C = Ox produced t C in parent 0
Summary & Conclusion SAR for mechanism generation: Available SARs are sound for simple hydrocarbons and mono-functional VOC but less reliable for multifunctional species (rate constants and branching ratios). Critical need of additional experimental and/or theoretical data for multifunctional species to better constrain SAR GECKOA/MCM versus MCM comparison: Consistent results provided by GECKOA and MCM3.2 for hydrocarbons having small carbon skeleton: The reduction protocol applied in MCM seems appropriate to describe the Ox/NOx/HOx chemistry Discrepancies highlighted between GECKOA and MCM3.2 in simulated oxidative trajectories for hydrocarbons having large carbon skeleton: - MCM favors fragmentation routes - GECKOA favors functionalization routes New protocols are required to reduce the size of the generated mechanism down to typical MCM size (10 4 species). A real challenge