English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

The formation, properties and lifetime of secondary organic aerosols in theatmosphere are largely determined by gas-particle partitioning coefficients ofthe participating organic vapours. Since these coefficients are often difficultto measure or compute, we developed a machine learning (ML) model to predictthem given molecular structure as input. Our data-driven approach is based onthe dataset by Wang et al. (Atmos. Chem. Phys., 17, 7529 (2017)), who computedthe partitioning coefficients and saturation vapour pressures of 3414atmospheric oxidation products from the master chemical mechanism using theCOSMOtherm program. We train a kernel ridge regression (KRR) ML model on thesaturation vapour pressure ($P_{sat}$), and on two equilibrium partitioningcoefficients: between a water-insoluble organic matter phase and the gas phase($K_{WIOM/G}$), and between an infinitely dilute solution with pure water andthe gas phase ($K_{W/G}$). For the input representation of the atomic structureof each organic molecule to the machine, we test different descriptors. Ourbest ML model predicts $P_{sat}$ and $K_{WIOM/G}$ to within 0.3 and $K_{W/G}$to within 0.4 logarithmic units of the original COSMOtherm calculations. Thisis equal or better than the typical accuracy of COSMOtherm predictions comparedto experimental data. We then apply our ML model to a dataset of 35,383molecules that we generated based on a carbon 10 backbone and functionalizedwith 0 to 6 carboxyl, carbonyl or hydroxyl groups to evaluate its performancefor polyfunctional compounds with potentially low $P_{sat}$. The resulting$P_{sat}$ and partitioning coefficient distributions were physico-chemicallyreasonable, and the volatility predictions for the most highly oxidizedcompounds were in qualitative agreement with experimentally inferredvolatilities of atmospheric oxidation products with similar elementalcomposition.

Predicting Gas-Particle Partitioning Coefficients of Atmospheric Molecules with Machine Learning