Faculty of Science - Department of Chemistry - Division of Quantum Chemistry and Physical Chemistry Katholieke Universiteit Leuven Structure-Activity Relationships - Mechanism development Luc Vereecken Research group on reaction kinetics Department of Chemistry Quantum Chemistry and Physical Chemistry K.U.Leuven, Belgium

Structure activity relationships : WP2 Introduction Task 2.1 : alkoxy decomposition and isomerisation Task 2.2 : Site-specific NO 3 and OH addition on alkenes Task 2.3 : O 3 cycloaddition on alkenes Task 2.4 : H-abstraction by OH from hydrocarbons Mechanism development : WP3 - WP5 Task 3.1 : OH +  -pinene Task 3.2 : O 3 +  -pinene,  -humulene,  -caryophyllene Task 5 : Oxygenates + OH : T,P-dependent mechanism

Chemical mechanisms for modeling Introduction - SARs Large, explicit mechanisms (e.g. MCM) 100s to 1000s of reactions/compounds  But no direct experimental or theoretical data on many of these  Use of SAR’s, predictive correlations Increasing demand for ever-better accuracy Policy-supporting predictions, what-if analyses: - Smog-episodes, chemical weather, climate - Emission control (compounds and quantities)  Need for accurate Structure Activity Relationships

SAR’s and correlations Structure-Activity Relationship or Predictive Correlation: Good predictive accuracy Easy to use Continuous development Working model: Independent, additive site-specific rate coefficients k tot =  k site (even for different types of reaction) Most rate coefficients depend primarily on local effects Inductive, hyperconjugative effects don’t carry very far H-bonds, resonances, … must be treated explicitly Linear models are easy to work with

Addition of OH-radicals on (poly-)alkenes Introduction

OH-addition on (poly-)alkenes Alkenes The rate of addition depends mainly on the substituents of the radical site C b after addition : X 3 X 4 k prim = 0.45  cm 3 s ‑ 1 k sec = 3.0  cm 3 s ‑ 1 k tert = 5.5  cm 3 s ‑ 1 Conjugated Alkenes : some contribution from second radical site k sec/prim = 3.0  cm 3 s ‑ 1 k sec/sec = 3.8  cm 3 s ‑ 1 k sec/tert = 5.1  cm 3 s ‑ 1 k tert/prim = 5.7  cm 3 s ‑ 1 k tert/sec = 8.3  cm 3 s ‑ 1 k tert/tert = 9.9  cm 3 s ‑ 1 CCHCC R R R OH k CCHCC OHR R R. C CH CC R R R OH. resonance sec/tert

OH-addition on (poly-)alkenes Non-cyclic compounds: Average deviation 9% All compounds: Average deviation 13% Max. deviation 54% Can this be improved ?  Yes Residual errors mostly due to H-abstraction contributions Publication submitted to J. Phys. Chem. A

OH-addition on (poly-)alkenes Linear and mono-cyclic compounds

OH-addition on (poly-)alkenes + bicyclic and (near-)conjugated compounds

H-abstraction by OH-radicals Introduction

H-abstraction by OH radicals

Excellent correlation with bond strength Rate coefficient of abstraction determined by D(C  H) Correlation is non-linear (data can be fitted by quadratic eq.) log (k 298K ) =  D  D Resonance stabilization shifts curve: e.g. vinoxy stabilisation log (k 298K ) =  D  D – Dependence similar for all compounds Angle and curvature similar for all resonances: Hyperconjugation, allyl, super-allyl, vinoxy. In 1 st order approximation: use same value for all Different resonance stabilizations have different shift Correlation will break down for oxygenates/H-bonding at low T At room temperature: Carboxylic acids are already different

Addition of NO 3 -radicals on (poly-)alkenes Introduction

NO 3 -addition on (poly-)alkenes Addition of NO 3 radicals: double interaction The rate of addition depends on substitution on both carbons: F prim = 1.28  cm 3/2 s ‑ 1/2 f prim = 1.28  cm 3/2 s ‑ 1/2 F sec = 7.27  cm 3/2 s ‑ 1/2 f sec = 3.30  cm 3/2 s ‑ 1/2 F tert = 3.85  cm 3/2 s ‑ 1/2 f tert = 7.02  cm 3/2 s ‑ 1/2 Radical site: factor F Addition site: factor f k add = F  f k add,site = F  f  k add,tot =  k site Open questions: - Corrections for allyl-resonance stabilization of radical - H-abstraction (e.g. with allyl-resonance stabilization)

NO 3 -addition on (poly-)alkenes Average deviation  1.2

NO 3 -addition on (poly-)alkenes Average deviation  2.2

NO 3 -addition on (poly-)alkenes Systematic underestimation

NO 3 -addition on (poly-)alkenes

Possible influence of H-abstraction: e.g. series of 1-alkenes - Could be sizable for large hydrocarbons - Affected by addition followed by HNO 3 elimination ?

NO 3 -addition on (poly-)alkenes Addition to conjugated alkadienes: Substitution effect different than for OH-addition (partial stabilisation of radical electron by allyl-resonance) Underestimation seems different for linear and cyclic Linear: underestimation by  0.3 Cyclic: underestimation by  0.1  Different addition scheme across  -bonds ? Allyl-resonance Interaction across  -bonds

Decomposition of alkoxy radicals Introduction

Alkoxy radical decomposition Decomposition barrier depends mostly on ,  -substituents A first version of this SAR was published as: J. Peeters, G. Fantechi, L. Vereecken, J. Atmos. Chem. 48, 59 (2004) k(T) =  × 1.8×10 13 exp(-E b /RT) s -1 E b / kcal mol -1 =  n  -alkyl  n  -alkyl  n ,  -hydroxy  n  -oxo  n  -oxo curvature for small E b < 7 kcal mol -1 : E b ' / kcal mol ‑ 1 = E b  (9.0-E b ) 2

Alkoxy radical decomposition

Current developments (in progress) : - More quantum chemical methods 6-31G(d,p), G(2df,2pd), aug-cc-pVTZ MPW1K, BB1K, MPWKCIS1K, (CC, Gx, QCI) - Multi-rotamer TST with (modified) Arrhenius fit  SAR for E a, A, (n) - More substituents (preliminary) / kcal mol -1 :  -OR : -9.1  -OR : -9.0  -OOR : -7.5  -OOR :  =C :  =C : +4.6  -C=C : -5.0  -C=C : -9.6  -ONO 2 : -3.1  -ONO 2 : -2.7  -ONO : -4.2  -ONO : -6.2

Alkoxy radical decomposition Future work: - Use multi-rotamer TST for alkoxy isomerisation (H-shift) L. Vereecken, J. Peeters, J. Chem. Phys. 119, 5159 (2003) - Perform URESAM calculations on these systems:  Pressure dependence SAR for Troe Parameters: F c, k 0, … O 3 cycloaddition No results yet, but see literature

Conclusions - I OH-addition SAR: Very good accuracy Can only be improved by explicitly incorporating H-abstraction H-Abstraction correlation Very good correlation with bond strength Curvature and slope similar, delocalisation shifts curve NO 3 addition SAR Very good accuracy for most compounds (  1.2,  2.2) Conjugated alkenes are underpredicted  delocalisation effects Alkoxy decomposition SAR: Being extended (substituents and methodology) Data serves as basis for alkoxy isomerisation SAR Four site-specific predictive SARs:

Part II: Mechanism development Terpenes and sesquiterpenes Introduction - Mechanism development

OH-initiated oxidation of  -pinene using traditional chemistry: Chemistry of  -pinene + OH Prediction of 60 % acetone formation Experiment: acetone yields 8% (Aschmann et al, 1998) 2% (Orlando et al., 2000) 13% (Wisthaler et al., 2001) ?

Peroxy ringclosure in isoprene / terpenes : Chemistry of unsaturated (per)oxy radicals

Ring closure in  -pinene + OH

 -pinene + OH Nopinone: 25 %

 -pinene + OH Peroxy ring closure path forms dicarbonyl dihydroxy compound

 -pinene + OH About 4 % Chemistry with oxy ring closure finds low acetone yield  comparable to experimental findings Compounds formed are highly oxygenated  cyclic esters, formates

 -pinene + OH 10 ppt 100 ppt 1 ppb 10 ppb 100 ppb 1 ppm [NO] Peroxy chemistry ROO + R’OO/HOO (pre- and post ring closure) Peroxy ring closure di-OH-di-carbonyl Oxy ring closure Degradation mechanism depends on [NO], [HO 2 /RO 2 ]

 -pinene + OH Minor H-abstraction channels (Klara Petrov) Mainly formation of larger (multisubstituted) oxygenates. Larger products should nearly all be reactive to OH, O 3, NO 3

 -pinene + O 3 Other mechanisms Some additional theoretical verification on impact of - ring closure - low-NO x chemistry Mechanism sufficiently mature for modeling (see BIRA) Sesquiterpenes + O 3 No results yet

Oxygenates + OH Introduction

General mechanism: Oxygenates + OH

T,P-dependences: Oxygenates + OH Barriers above reactants: Formation of pre-reactive complexes not too important Positive T-dependence (except at low T: tunneling) No P-dependence Barriers below reactants: Chemical activation effects Negative T-dependence at all T Pressure dependent See: Peeters and Vereecken, Int. Symp. Gas Kin. 2006

Specific issues for theoretical work on oxygenate+OH reactions Oxygenates + OH - Calculation of tunneling contributions Small-curvature corrections most often used e.g. Masgrau et al., J. Phys. Chem. A 106, (2002) tunneling contribution  22 at 202 K for acetone+OH - Variational effects H-abstraction over H-bonds: low and broad TS Variational effects can be important (kinetic bottleneck not at energy maximum) e.g. Masgrau et al (acetone+OH) variational effects up to order of magnitude - Specific reaction pathways (See acids)

Acetone + OH The reaction of acetone + OH shows a curved Arrhenius plot: Wollenhaupt, Carl, Horowitz, Crowley, J. Phys. Chem. A 104, 2695 (2000) Gierczak, Gilles, Bauerle, Ravishankara, J. Phys. Chem. A 107, 5014 (2003); Talukdar et al., J. Phys. Chem. A 107, 5021 (2003)

Acetone + OH Theoretical work shows the general features of the PES: Vandenberk, Vereecken and Peeters, PCCP 4, 461 (2002) Similar PESes by Masgrau et al., J. Phys. Chem. A 106, (2002) Vasvári et al., PCCP 3, 551 (2001)

Hydroxyacetone + OH The reaction of hydroxyacetone + OH : Negative T-dependence Dillon, Horowitz, Hölscher, Crowley, Vereecken, Peeters, PCCP, 8, 236, 2006

Hydroxyacetone + OH Accuracy of barrier heights did not allow for final theoretical kinetic predictions.

Glycolaldehyde + OH The reaction of CH 2 OH  CHO+ OH : No T-dependence Karunanandan, Hölscher, Dillon, Horowitz, Crowley, Vereecken, Peeters, submitted for publication -Slowdown relative to CH 3 CHO: due to charge distribution - Lack of T-dependence: due to specific barrier height: RRKM-ME simulation

Stringent requirements for theoretical methodologies Oxygenates + OH Quantum chemical methods: very high level needed Calculation of energies But also for calculation of geometries and frequencies Mechanism development Unexpected mechanisms can exist Kinetic methodologies: Important effects of Tunneling (SCT or better needed) Variational effects Anharmonicity effects Multi-conformer (multi-well) effects Multiple pathways Internal rotors

Conclusions - II  -pinene + OH Very complex reaction mechanism Depends strongly on [NOx] versus [ROO/HOO] Many fast unimolecular reaction steps  reduction of mechanism possible In progress Terpenoids + O 3 In progress Oxygenates + OH : Very complex kinetics Stringent demands on theoretical methodology T,P-dependence of k(T) or product distribution still difficult Mechanism development