skip to main content

Search for: All records

Award ID contains: 1654104

Note: When clicking on a Digital Object Identifier (DOI) number, you will be taken to an external site maintained by the publisher. Some full text articles may not yet be available without a charge during the embargo (administrative interval).
What is a DOI Number?

Some links on this page may take you to non-federal websites. Their policies may differ from this site.

  1. Gas-particle partitioning of secondary organic aerosols is impacted by particle phase state and viscosity, which can be inferred from the glass transition temperature ( T g ) of the constituting organic compounds. Several parametrizations were developed to predict T g of organic compounds based on molecular properties and elemental composition, but they are subject to relatively large uncertainties as they do not account for molecular structure and functionality. Here we develop a new T g prediction method powered by machine learning and “molecular embeddings”, which are unique numerical representations of chemical compounds that retain information on their structure, inter atomic connectivity and functionality. We have trained multiple state-of-the-art machine learning models on databases of experimental T g of organic compounds and their corresponding molecular embeddings. The best prediction model is the tgBoost model built with an Extreme Gradient Boosting (XGBoost) regressor trained via a nested cross-validation method, reproducing experimental data very well with a mean absolute error of 18.3 K. It can also quantify the influence of number and location of functional groups on the T g of organic molecules, while accounting for atom connectivity and predicting different T g for compositional isomers. The tgBoost model suggests the following trendmore »for sensitivity of T g to functional group addition: –COOH (carboxylic acid) > –C(O)OR (ester) ≈ –OH (alcohol) > –C(O)R (ketone) ≈ –COR (ether) ≈ –C(O)H (aldehyde). We also developed a model to predict the melting point ( T m ) of organic compounds by training a deep neural network on a large dataset of experimental T m . The model performs reasonably well against the available dataset with a mean absolute error of 31.0 K. These new machine learning powered models can be applied to field and laboratory measurements as well as atmospheric aerosol models to predict the T g and T m of SOA compounds for evaluation of the phase state and viscosity of SOA.« less
    Free, publicly-accessible full text available May 19, 2023
  2. Free, publicly-accessible full text available March 8, 2023
  3. Individual atmospheric particles can contain mixtures of primary organic aerosol (POA), secondary organic aerosol (SOA), and secondary inorganic aerosol (SIA). To predict the role of such complex multicomponent particles in air quality and climate, information on the number and types of phases present in the particles is needed. However, the phase behavior of such particles has not been studied in the laboratory, and as a result, remains poorly constrained. Here, we show that POA+SOA+SIA particles can contain three distinct liquid phases: a low-polarity organic-rich phase, a higher-polarity organic-rich phase, and an aqueous inorganic-rich phase. Based on our results, when the elemental oxygen-to-carbon (O:C) ratio of the SOA is less than 0.8, three liquid phases can coexist within the same particle over a wide relative humidity range. In contrast, when the O:C ratio of the SOA is greater than 0.8, three phases will not form. We also demonstrate, using thermodynamic and kinetic modeling, that the presence of three liquid phases in such particles impacts their equilibration timescale with the surrounding gas phase. Three phases will likely also impact their ability to act as nuclei for liquid cloud droplets, the reactivity of these particles, and the mechanism of SOA formation and growthmore »in the atmosphere. These observations provide fundamental information necessary for improved predictions of air quality and aerosol indirect effects on climate.« less
  4. Abstract. Mass accommodation is an essential process for gas–particle partitioning oforganic compounds in secondary organic aerosols (SOA). The massaccommodation coefficient is commonly described as the probability of a gasmolecule colliding with the surface to enter the particle phase. It is oftenapplied, however, without specifying if and how deep a molecule has topenetrate beneath the surface to be regarded as being incorporated into thecondensed phase (adsorption vs. absorption). While this aspect is usuallynot critical for liquid particles with rapid surface–bulk exchange, it canbe important for viscous semi-solid or glassy solid particles to distinguishand resolve the kinetics of accommodation at the surface, transfer acrossthe gas–particle interface, and further transport into the particle bulk. For this purpose, we introduce a novel parameter: an effective massaccommodation coefficient αeff that depends on penetrationdepth and is a function of surface accommodation coefficient, volatility,bulk diffusivity, and particle-phase reaction rate coefficient. Applicationof αeff in the traditional Fuchs–Sutugin approximation ofmass-transport kinetics at the gas–particle interface yields SOApartitioning results that are consistent with a detailed kinetic multilayermodel (kinetic multilayer model of gas–particle interactions in aerosols and clouds, KM-GAP; Shiraiwa et al., 2012) and two-film model solutions (Modelfor Simulating Aerosol Interactions and Chemistry, MOSAIC;Zaveri et al., 2014) but deviate substantially frommore »earlier modelingapproaches not considering the influence of penetration depth and relatedparameters. For highly viscous or semi-solid particles, we show that the effective massaccommodation coefficient remains similar to the surface accommodationcoefficient in the case of low-volatility compounds, whereas it can decrease byseveral orders of magnitude in the case of semi-volatile compounds. Such effectscan explain apparent inconsistencies between earlier studies deriving massaccommodation coefficients from experimental data or from molecular dynamicssimulations. Our findings challenge the approach of traditional SOA models using theFuchs–Sutugin approximation of mass transfer kinetics with a fixed massaccommodation coefficient, regardless of particle phase state and penetrationdepth. The effective mass accommodation coefficient introduced in this studyprovides an efficient new way of accounting for the influence of volatility,diffusivity, and particle-phase reactions on SOA partitioning in processmodels as well as in regional and global air quality models. While kineticlimitations may not be critical for partitioning into liquid SOA particlesin the planetary boundary layer (PBL), the effects are likely important foramorphous semi-solid or glassy SOA in the free and upper troposphere (FT–UT)as well as in the PBL at low relative humidity and low temperature.« less
  5. Abstract. Polycyclic aromatic hydrocarbons (PAHs) are carcinogenic air pollutants. The dispersion of PAHs in the atmosphere is influenced by gas–particle partitioning and chemical loss. These processes are closely interlinked and may occur at vastly differing timescales, which complicates their mathematical description in chemical transport models. Here, we use a kinetic model that explicitly resolves mass transport and chemical reactions in the gas and particle phases to describe and explore the dynamic and non-equilibrium interplay of gas–particle partitioning and chemical losses of PAHs on soot particles. We define the equilibration timescale τeq of gas–particle partitioning as the e-folding time for relaxation of the system to the partitioning equilibrium. We find this metric to span from seconds to hours depending on temperature, particle surface area, and the type of PAH. The equilibration time can be approximated using a time-independent equation, τeq≈1kdes+kads, which depends on the desorption rate coefficient kdes and adsorption rate coefficient kads, both of which can be calculated from experimentally accessible parameters. The model reveals two regimes in which different physical processes control the equilibration timescale: a desorption-controlled and an adsorption-controlled regime. In a case study with the PAH pyrene, we illustrate how chemical loss can perturb the equilibrium particulatemore »fraction at typical atmospheric concentrations of O3 and OH. For the surface reaction with O3, the perturbation is significant and increases with the gas-phase concentration of O3. Conversely, perturbations are smaller for reaction with the OH radical, which reacts with pyrene on both the surface of particles and in the gas phase. Global and regional chemical transport models typically approximate gas–particle partitioning with instantaneous-equilibration approaches. We highlight scenarios in which these approximations deviate from the explicitly coupled treatment of gas–particle partitioning and chemistry presented in this study. We find that the discrepancy between solutions depends on the operator-splitting time step and the choice of time step can help to minimize the discrepancy. The findings and techniques presented in this work not only are relevant for PAHs but can also be applied to other semi-volatile substances that undergo chemical reactions and mass transport between the gas and particle phase.« less
  6. Abstract. Secondary organic aerosols (SOA) are major components of atmospheric fineparticulate matter, affecting climate and air quality. Mounting evidenceexists that SOA can adopt glassy and viscous semisolid states, impactingformation and partitioning of SOA. In this study, we apply the GECKO-A(Generator of Explicit Chemistry and Kinetics of Organics in the Atmosphere)model to conduct explicit chemical modeling of isoprene photooxidation andα-pinene ozonolysis and their subsequent SOA formation. The detailedgas-phase chemical schemes from GECKO-A are implemented into a box model andcoupled to our recently developed glass transition temperatureparameterizations, allowing us to predict SOA viscosity. The effects ofchemical composition, relative humidity, mass loadings and mass accommodation on particle viscosity are investigated in comparison withmeasurements of SOA viscosity. The simulated viscosity of isoprene SOAagrees well with viscosity measurements as a function of relative humidity,while the model underestimates viscosity of α-pinene SOA by a feworders of magnitude. This difference may be due to missing processes in themodel, including autoxidation and particle-phase reactions, leading to theformation of high-molar-mass compounds that would increase particleviscosity. Additional simulations imply that kinetic limitations of bulkdiffusion and reduction in mass accommodation coefficient may play a role inenhancing particle viscosity by suppressing condensation of semi-volatilecompounds. The developed model is a useful tool formore »analysis andinvestigation of the interplay among gas-phase reactions, particle chemicalcomposition and SOA phase state.« less