Real-time methods for estimating organic component mass concentrations from aerosol mass spectrometer data

We use results from positive matrix factorization (PMF) analysis of 15 urban aerosol mass spectrometer (AMS) data sets to derive simple methods for estimating major organic aerosol (OA) component concentrations in real time. PMF analysis extracts mass spectral (MS) profiles and mass concentrations for key OA components such as hydrocarbon-like OA (HOA), oxygenated OA (OOA), low-volatility OOA (LV-OOA), semivolatile OOA (SV-OOA), and biomass burning OA (BBOA). The variability in the component MS across all sites is characterized and used to derive standard profiles for real-time estimation of component concentrations. Two methods for obtaining first-order estimates of the HOA and OOA mass concentrations are evaluated. The first approach is the tracer m/z method, in which the HOA and OOA concentrations are estimated from m/z 57 and m/z 44 as follows: HOA ～ 13.4 × (C(57) - 0.1 × C(44)) and OOA ～ 6.6 × C(44), where C(i) is the equivalent mass concentration of tracer ion m/z i. The second approach uses a chemical mass balance (CMB) method in which standard HOA and OOA profiles are used as a priori information for calculating their mass concentrations. The HOA and OOA mass concentrations obtained from the first-order estimates are evaluated by comparing with the corresponding PMF results for each site. Both methods reproduce the HOA and OOA concentrations to within ～30% of the results from detailed PMF analysis at most sites, with the CMB method being slightly better. For hybrid CMB methods, we find that fixing the LV-OOA spectrum and not constraining the other spectra produces the best results.