Abstract. Reference mass spectra are routinely used to facilitate source apportionment of ambient organic aerosol (OA) measured by aerosol mass
spectrometers. However, source apportionment of solid-fuel-burning emissions can be complicated by the use of different fuels, stoves, and burning
conditions. In this study, the organic aerosol mass spectra produced from burning a range of solid fuels in several heating stoves have been
compared using an aerosol chemical speciation monitor (ACSM). The same samples of biomass briquettes and smokeless coal were burnt in a conventional stove
and Ecodesign stove (Ecodesign refers to a stove conforming to EU Directive 2009/125/EC), while different batches of wood, peat, and smoky coal were also burnt in the conventional stove, and the OA mass spectra
were compared to those previously obtained using a boiler stove. The results show that although certain ions (e.g., m/z 60) remain important markers
for solid-fuel burning, the peak intensities obtained at specific m/z values in the normalized mass spectra were not constant with variations
ranging from < 5 % to > 100 %. Using the OA mass spectra of peat, wood, and coal as anchoring profiles and the variation of individual
m/z values for the upper/lower limits (the limits approach) in the positive matrix factorization (PMF) analysis with the Multilinear Engine
algorithm (ME-2), the respective contributions of these fuels to ambient submicron aerosols during a winter period in Dublin, Ireland, were evaluated and
compared with the conventional a-value approach. The ME-2 solution was stable for the limits approach with uncertainties in the range of
2 %–7 %, while relatively large uncertainties (8 %–29 %) were found for the a-value approach. Nevertheless, both approaches
showed good agreement overall, with the burning of peat (39 % vs. 41 %) and wood (14 % vs. 11 %) accounting for the majority of
ambient organic aerosol during polluted evenings, despite their small uses compared to electricity and gas. This study, thus, accounts for the
source variability in ME-2 modelling and provides better constraints on the primary factor contributions to the ambient organic aerosol
estimations. The finding from this study has significant implications for public health and policymakers considering that it is often the case that
different batches of solid fuels are often burnt in different stoves in real-world applications.
SoFi, an Igor based interface for the efficient use of the generalized multilinear engine (ME-2) for source apportionment: application to aerosol mass spectrometer data