Abstract. Gas-particle equilibrium partitioning is a fundamental concept used to describe the growth and loss of secondary organic aerosol (SOA). However, recent literature has suggested that gas-particle partitioning may be kinetically limited, preventing volatilization from the aerosol phase as a result of the physical state of the aerosol (e.g. glassy, viscous). Experimental measurements of diffusion constants within viscous aerosol are limited and do not represent the complex chemical composition observed in SOA (i.e. multicomponent mixtures). Motivated by the need to address fundamental questions regarding the effect of the physical state and chemical composition of a particle on gas-particle partitioning, we present the design and operation of a newly built 0.3 m3 continuous-flow reactor (CFR), which can be used as a tool to gain considerable insights into the composition and physical state of SOA. The CFR was used to generate SOA from the photo-oxidation of α-pinene, limonene, β-caryophyllene and toluene under different experimental conditions (i.e. relative humidity, VOC and VOC∕NOx ratios). Up to 102 mg of SOA mass was collected per experiment, allowing the use of highly accurate compositional- and single-particle analysis techniques, which are not usually accessible due to the large quantity of organic aerosol mass required for analysis. A suite of offline analytical techniques was used to determine the chemical composition and physical state of the generated SOA, including attenuated total reflectance infrared spectroscopy; carbon, hydrogen, nitrogen, and sulfur (CHNS) elemental analysis; 1H and 1H-13C nuclear magnetic resonance spectroscopy (NMR); ultra-performance liquid chromatography ultra-high-resolution mass spectrometry (UHRMS); high-performance liquid chromatography ion-trap mass spectrometry (HPLC-ITMS); and an electrodynamic balance (EDB). The oxygen-to-carbon (O∕C) and hydrogen-to-carbon (H∕C) ratios of generated SOA samples (determined using a CHNS elemental analyser) displayed good agreement with literature values and were consistent with the characteristic Van Krevelen diagram trajectory, with an observed slope of −0.41. The elemental composition of two SOA samples formed in separate replicate experiments displayed excellent reproducibility, with the O∕C and H∕C ratios of the SOA samples observed to be within error of the analytical instrumentation (instrument accuracy ±0.15 % to a reference standard). The ability to use a highly accurate CHNS elemental analyser to determine the elemental composition of the SOA samples allowed us to evaluate the accuracy of reported SOA elemental compositions using UHRMS (a commonly used technique). In all of the experiments investigated, the SOA O∕C ratios obtained for each SOA sample using UHRMS were lower than the O∕C ratios obtained from the CHNS analyser (the more accurate and non-selective technique). The average difference in the ΔO∕C ratios ranged from 19 % to 45 % depending on the SOA precursor and formation conditions. α-pinene SOA standards were generated from the collected SOA mass using semi-preparative HPLC-ITMS coupled to an automated fraction collector, followed by 1H NMR spectroscopy. Up to 35.8±1.6 % (propagated error of the uncertainty in the slope of the calibrations graphs) of α-pinene SOA was quantified using this method; a considerable improvement from most previous studies. Single aerosol droplets were generated from the collected SOA samples and trapped within an EDB at different temperatures and relative humidities to investigate the dynamic changes in their physiochemical properties. The volatilization of organic components from toluene and β-caryophyllene SOA particles at 0 % relative humidity was found to be kinetically limited, owing to particle viscosity. The unconventional use of a newly built CFR, combined with comprehensive offline chemical characterization and single-particle measurements, offers a unique approach to further our understanding of the relationship between SOA formation conditions, chemical composition and physiochemical properties.
Read full abstract