Measuring and modeling the primary organic aerosol volatility from a modern non-road diesel engine

Primary organic aerosol (POA) in diesel exhaust is semi-volatile and partitions mass between the gas and particle phases. POA volatility is not well understood for alternative fuels, varying engine loads, and for engines that feature modern emissions controls. In this study, we performed filter-based measurements of diesel exhaust from a modern-day non-road diesel engine for two different fuels (conventional diesel and soy-based biodiesel), two different engine loads (idle and 50% load), and with and without an emissions control device. Filters were analyzed offline to determine the POA volatility in two different ways: positive artifact on quartz filters at varying dilution ratios and speciation of alkanes. The POA volatility determined from our data suggests that POA mass emissions from diesel exhaust may be reduced by a factor of five with dilution to atmospherically relevant concentrations. These results are generally consistent with previous literature on POA volatility from non-road diesel engines but not with that from on-road diesel vehicles. POA volatility may hence need to be treated separately for non- and on-road sources in atmospheric models. Surprisingly, the POA volatility did not appear to vary under different combinations of fuel, engine load, and emissions control experiments performed, suggesting that POA might be dominated by unburned lubricating oil and its oxidation products. The POA volatility estimated from the speciation of alkanes was found to agree well with that determined from the dilution experiments. A kinetic model was used to calculate the gas/particle partitioning of POA in the dilution system. The modeling suggests that residence times in the dilution tunnel need to be on the order of minutes to allow the POA in the diluted exhaust to achieve gas/particle equilibrium. The use of short residence times (less than tens of seconds), similar to those used in conventional dilution systems, may bias the measurement of POA mass emissions in such systems and is of particular concern for emissions from cleaner, more modern combustion sources. The precise magnitude and direction of the bias depends on the exhaust temperature before dilution, tailpipe seed concentrations, dilution ratio, and residence times in the dilution tunnel. We recommend that kinetic models such as those used in this work be used, instead of using equilibrium assumptions, to inform the design and operation of the dilution tunnels as well to interpret the POA volatility from measurements made with those dilution tunnels.