Contribution of intermediate-volatility organic compounds from on-road transport to secondary organic aerosol levels in Europe

Abstract. Atmospheric organic compounds with an effective saturation concentration (C∗) at 298 K between 103 and 106 µg m−3 are called intermediate-volatility organic compounds (IVOCs), and they have been identified as important secondary organic aerosol (SOA) precursors. In this work, we simulate IVOCs emitted from on-road diesel and gasoline vehicles over Europe with a chemical transport model (CTM), utilizing a new approach in which IVOCs are treated as lumped species that preserve their chemical characteristics. This approach allows us to assess both the overall contribution of IVOCs to SOA formation and the role of specific compounds. For the simulated early-summer period, the highest concentrations of SOA formed from the oxidation of on-road IVOCs (SOA-iv) are predicted for major European cities, like Paris, Athens, and Madrid. In these urban environments, on-road SOA-iv can account for up to a quarter of the predicted total SOA. Over Europe, unspeciated cyclic alkanes in the IVOC range are estimated to account for up to 72 % of the total on-road SOA-iv mass, with compounds with 15 to 20 carbons being the most prominent precursors. The sensitivity of the predicted SOA-iv concentrations to the selected parameters of the new lumping scheme is also investigated. Active multigenerational aging of the secondary aerosol products has the most significant effect as it increases the predicted SOA-iv concentrations by 67 %.


S1 Lumped IVOC gas-phase chemistry and the SOA-iv parametrisation in PMCAMx-iv.
In PMCAMx-iv, all seven lumped IVOC species react individually in the gas-phase with the hydroxyl radical.These seven oxidation reactions lead to the formation of volatile products, which remain in the gas-phase, and lower volatility products, which partition to the aerosol phase and form SOA-iv.For the introduction of these seven reactions to the existing SAPRC gas-phase mechanism, it is assumed that their volatile products and their reaction rate constants are the same as the ones used to simulate OH radical reactions of larger VOCs, which are part of the pre-existing mechanism.
For the four lumped alkane species (ALK6, ALK7, ALK8 and ALK9), the oxidation reactions are based on the reaction of ALK5 with the hydroxyl radical.For example,  +  → 0.653 RO2R + 0.347 RO2N + 0.948 R2O2 + 0.026 HCHO + 0.099 CCHO +0.204 RCHO + 0.072 ACET + 0.089 MEK + 0.417 PROD + ∑ a i OCG i where, RO2R is the organic peroxy radical converting NO to NO2 with HO2 production, RO2N is the organic peroxy radical converting NO to organic nitrate, R2O2 is the organic peroxy radical converting NO to NO2, HCHO is formaldehyde, CCHO is acetaldehyde, RCHO represents the higher aldehydes (based on propionaldehyde), ACET is acetone, MEK is methyl ketone, PROD represents other organic products, ai is its NOx-dependent mass-based yield and OCGi is the ith oxygenated condensable lower volatility product which can partition to the aerosol phase (all seven new lumped species contribute to the formation of the same five lower volatility products).The reaction rate constant for reaction (R1) is assumed to have a value of 1.4 × 10 4 ppm -1 min -1 .
For the PAHs species (PAH1 and PAH2) and the new aromatic species (ARO3), their reactions are based on the reaction of ARO2 with the hydroxyl radical.where, HO2 is the hydroperoxyl radical, GLY is glyoxal, MGLY is methylglyoxal, BACL is biacetyl, CRES is cresol, BALD is benzaldehyde and DCB1-DCB3 represent three different aromatic ring-opening dicarbonyl products.The reaction rate constant for reaction (R2) is assumed to have a value of 3.9 × 10 4 ppm -1 min -1 .
The five lower volatility products have respectively an effective saturation concentration of 0.1, 1, 10, 100 and 10 3 μg m -3 .Each of the lower volatility products follows a partitioning reaction (R3) forming SOA-iv products with the corresponding volatility.
For all seven lumped species, the NOx dependent stoichiometric yields are estimated based on recent experimental data.

Figure S1 :
Figure S1: Average diurnal profiles of the on-road diesel and gasoline vehicle emissions of the seven lumped IVOC species over Paris for May 2008.

Figure S2 :
Figure S2: Average diurnal profiles for the ground-level concentrations of on-road transportation IVOCs over Paris as predicted by PMCAMx (sum of the four surrogate species) and PMCAMx-iv (sum of the seven lumped IVOC species) for May 2008.

Figure S3 :
Figure S3: Domain-averaged ground-level concentrations of the SOA-iv products (a) in the aerosol phase and (b) in the gas-phase predicted by PMCAMx, PMCAMx-iv and by the different sensitivity tests for the May 2008 simulations over Europe.

Figure S4 :
Figure S4: Estimated averaged ground-level PM2.5 concentrations of (a) SOA and (d) OA concentrations and the corresponding contributions from on-road diesel and gasoline vehicles IVOCs over Europe for May 2008 using (b,e) PMCAMx-iv and (c,f) PMCAMx-iv.

Figure S5 :
Figure S5: (a) PMCAMx-iv estimated hourly-averaged ground-level concentrations of PM2.5 SOAtotal, OAtotal and on-road SOA-iv over Moscow and (b) the contributions of on-road IVOCs to the total SOA and OA concentrations for May 2008.

Figure S8 :
Figure S8: (a) Absolute and (b) percentage differences between the average ground-level concentrations of the sum of the seven lumped IVOC species that are predicted in the "base case" and the "IVOC emissions × 2" simulations.

Figure S9 :
Figure S9: PM2.5 hourly-averaged timeseries for the predicted ground-level concentrations of on-road transport SOA-iv by the "VBS scheme", the "base case" and the "lumped IVOC emissions × 2" simulations over the city of Athens for May 2008.

Figure S10 :
Figure S10: The maximum hourly PM2.5 SOA-iv concentrations for May 2008 over Paris as predicted by the "base case" and the six sensitivity tests of PMCAMx-iv.Not all max values are predicted to occur at the same time.

Figure S11 :
Figure S11: (a) Absolute and (b) percentage differences between the average ground-level concentrations of O3 predicted in the "base case" and in the "no gas-phase chemistry" simulations.

Figure S12 :
Figure S12: Predicted ground-level ozone concentrations by PMCAMx ("base case" test), PMCAMx-iv ("new IVOC scheme" test) and the "no gas-phase chemistry" sensitivity test over Milan (Italy) on the 8 th of May 2008.

Figure S13 :
Figure S13: Diurnal profile of the PM2.5 ground-level on-road transport SOA-iv concentrations over Paris predicted by the "base case", the "multigenerational aging" test and the "ΔH effect" test.

Figure S14 :
Figure S14: Diurnal profile of the PM2.5 ground-level on-road transport SOA-iv concentrations over Finokalia predicted by PMCAMx, PMCAMx-iv and the "different yields" sensitivity test.

Table S1 :
Average percent contributions of IVOCs emitted from on-road diesel and gasoline vehicles to the formation of PM2.5 OA and SOA over 44 European capital cities for May 2008.The mean ground-level concentrations of SOA and OA in μg m -3 are included in the parenthesis.

Table S2 :
Prediction skill metrics of PMCAMx, PMCAMx-iv and the six sensitivity tests against AMS hourly ground measurements of PM1 OA concentration from the four measuring stations that were part of the EUCAARI campaign in May 2008.

Table S3 :
Average percent contributions of on-road IVOCs to the formation of PM1 OA concentrations over the four measuring stations were part of the EUCAARI campaign in May 2008.The contributions are calculated for all eight simulation cases.