Identification of chemistry-dependent artifacts on gravimetric PM fine readings at the T1 site during the MILAGRO field campaign

As part of the MIRAGE (MILAGRO) study conducted 7-30 March 2006 in Mexico City and its Metropolitan Area (MCMA), fine particulate matter (PM _2.5) was collected using two Tapered Element Oscillating Microbalance (TEOM) systems, and a Partisol instrument at the T1 super-site (Tecamac, State of Mexico). Inorganic analysis was performed on filter-based (PM ₁, PM _2.5-URG) measurements also collected at this site. Measurements from the gravimetric (TEOMs, Partisol) and URG systems were inter-compared with chemical speciation measurements using a Particle Into Liquid Sampler (PILS) and Thermal Optical methods. Mass and chemical balances applied over the first part of the study (11-22 March) showed that a TEOM using a device (SES) which reduces particle-bound water and retains a fraction of semi-volatile compounds (SVM) gives readings ∼30% larger than a conventional TEOM. In the second part of the study (26-30 March), the loss of SVM during TEOM-heated filter collection (both systems) represented a significant fraction of PM _2.5 mass due to changes in particle composition. Overall, when nonvolatile nitrate dominated (i.e., when associated with crustal species and not NH ₄⁺) and/or sulfate dominates (SO ₄^2-/NO ₃^- molar ratio is >1), PM _2.5 mass readings are in agreement with those reported for the T1 site if TEOM is using a SES device. However, when volatile nitrate dominates (i.e., NH ₄NO ₃) or SO ₄^2-/NO ₃^- molar ratio is <1, a larger fraction is lost from both TEOMs (with or without the SES device). Under the latter regime, uncertainties are large and gravimetric losses may reach 30%-50%. The gravimetric PARTISOL instrument recorded lower readings under all of the aforementioned conditions with differences versus TEOMs decreasing with increasing RH. These findings call for a careful characterization of such volatilization biases to improve current PM (PM ₁₀, PM _2.5) measurements/networks, especially in alkaline-rich environments that can favor such biases. With regards to PM ₁ and PM _2.5 filter-based measurements, findings are: 1) crustal-related elements are important features in the PM _2.5-1 size fraction; 2) a factor of ∼2 overestimation of SO ₄^2- concentrations is recorded on substrates during PM collection and 3) main elements of a typical urban aerosol size distribution are concentrated in the 1 μm (versus 2.5 μm) size fraction.