Gasoline cars produce more carbonaceous particulate matter than modern filter-equipped diesel cars

Abstract

Carbonaceous particulate matter (PM), comprising black carbon (BC), primary organic aerosol (POA) and secondary organic aerosol (SOA, from atmospheric aging of precursors), is a highly toxic vehicle exhaust component. Therefore, understanding vehicle pollution requires knowledge of both primary emissions, and how these emissions age in the atmosphere. We provide a systematic examination of carbonaceous PM emissions and parameterisation of SOA formation from modern diesel and gasoline cars at different temperatures (22, -7 °C) during controlled laboratory experiments. Carbonaceous PM emission and SOA formation is markedly higher from gasoline than diesel particle filter (DPF) and catalyst-equipped diesel cars, more so at -7 °C, contrasting with nitrogen oxides (NO_X). Higher SOA formation from gasoline cars and primary emission reductions for diesels implies gasoline cars will increasingly dominate vehicular total carbonaceous PM, though older non-DPF-equipped diesels will continue to dominate the primary fraction for some time. Supported by state-of-the-art source apportionment of ambient fossil fuel derived PM, our results show that whether gasoline or diesel cars are more polluting depends on the pollutant in question, i.e. that diesel cars are not necessarily worse polluters than gasoline cars.

Figure 1

Schematic (not to scale) of the experimental set up. The test vehicle and smog chamber were operated inside the temperature controlled test cell with instrumentation outside. During testing, instruments were operated at the tailpipe and from a constant volume sampler (CVS), while a small fraction of emission were sampled into the smog chamber via a heated ejector dilutor (total dilution factor ~150). Tailpipe instrumentation included a Fourier-transformed infrared spectrometer (FTIR), non-dispersive infrared sensor (NDIR), flame ionisation detector (FID), and a chemiluminescence detector (CLD). At the CVS were NDIR, CLD, and FID instruments as well as a heated gravimetric filter sampler. After testing, primary emissions were investigated and then aged in the smog chamber under UV lights. Gas phase instruments were a proton transfer reaction time-of-flight mass spectrometer (PTR-ToF-MS), FID, cavity ring-down spectrometer, nitrogen oxide (NO_X) detectors, an ozone (O₃) monitor and relative humidity and temperature sensors (RH/T). Aerosol instrumentation was a high-resolution time-of-flight aerosol mass spectrometer (HR-ToF-AMS), a scanning mobility particle sizer (SMPS) and an aethalometer. Table S2 lists all chamber instruments in detail while Platt et al. give a fuller description of this set-up.

Figure 2

Carbonaceous emissions/secondary organic aerosol formation from modern diesel and gasoline passenger cars. (a) Aerosol emission factors (g kg⁻¹ fuel) measured in this study. SOA is at OH exposure = 10⁷ molec. cm⁻³ h. Diesel vehicles did not produce measurable OA so gravimetric PM is given. For comparison the average and standard deviation of PM emission factors from non-DPF medium duty diesels from Gordon et al. 2013 ref. , is shown. (b) Average ratio of diesel/gasoline emission factors for OA, BC, PM, methane (CH₄), total hydrocarbon (THC), aromatic hydrocarbons (Ar. HC), nitrogen oxides (NO_X) and carbon monoxide (CO) at 22 °C and −7 °C. Euro 5 values are obtained using the NEDC, while US LEV2 and DPF-equipped vehicles use the US unified driving cycle, UC^, . While a true ratio cannot be calculated for OC and BC, a maximum can be, based on detection limits, highlighted by the red asterisks (*). In contrast to CO and especially NO_X, PM and hydrocarbon emissions from gasoline cars are higher than from diesel cars. (c) Averaged THC-normalised exhaust composition (temperature in parentheses) in the smog chamber with uncharacterised emissions in grey. The single largest fraction of gasoline THC consists of methyl benzenes present in the fuel, while the rest likely consists of surviving linear/branched, saturated/unsaturated hydrocarbons. Meanwhile, diesel emissions mainly comprise pyrolysis products, including small carbonyls (formaldehyde, acetaldehyde) and carboxylic acids (formic, acetic), which are not efficient SOA precursors. Supporting material related to Fig. 2 available in SI Tables S3–7, SI.

Figure 3

Secondary organic aerosol yields from Euro 5 gasoline car emissions. Modelled secondary organic aerosol (SOA) yields as a function of suspended organic aerosol (C_OA) concentration from three Euro 5 gasoline cars at 22 and −7 °C (orange and blue, respectively). Error bars represent one standard deviation. Also shown: SOA yields from light aromatic hydrocarbons (Yield_Ar), and diesel and gasoline vapours (Yield_{diesel vapour}, Yield_{gasoline vapour}), from Jathar et al.. The comparison clearly highlights that SOA yields from gasoline emissions are higher than those based on the fuel aromatic content or expected from gasoline vapour, suggesting the presence of unidentified SOA precursors in the exhaust. Meanwhile, DPF-diesel emissions did not produce any measureable SOA, in contrast to the oxidation of diesel vapours. Taken together both of these observations indicate that SOA yields from gasoline and diesel emissions cannot be predicted based on the yields of fuel vapours. Note that the gasoline car SOA yields are not corrected for vapour phase wall losses (see Methods), for the purposes of intercomparison with previous studies.

Figure 4

Time resolved exhaust total hydrocarbon (THC) concentrations from Euro 5 gasoline (number vehicles, n = 11) and diesel (n = 6) passenger cars. (a) Median time-resolved THC exhaust concentrations in the diesel and gasoline passenger car exhausts during the NEDC driving cycle, at −7 and 22 °C (target speed, grey shaded region) (b) Ratio of the distributions whose medians are given in A, shown as a probability density function (PDF, colour scale), for the 22 °C case. While the ratio of diesel to gasoline exhaust THC is broadly distributed (>factor 10) and variable, particularly during initial operations at higher speed, the difference is still statistically significant. The grey line gives the time-integrated median of the distribution (i.e. from the test start to the denoted time, lower than the time-resolved ratio for much of the cycle due to the high absolute concentrations and low gasoline:diesel THC during the cold start, and below unity, indicating that total gasoline exhaust THC emissions are higher). Note the logarithmic scale in both panels. While THC emissions from diesel cars are up to one order of magnitude higher than gasoline emissions during almost the entire driving period, gasoline emissions are up to two orders of magnitudes higher when the catalyst is cold. Because of the cold start effect, integrated emissions from gasoline cars exceed diesel, even after a journey of several kilometres (~14 km). This effect is more pronounced at −7 °C (SI Fig. S3).

Figure 5

Estimates of the contribution of diesel and gasoline passenger cars to ambient urban PM and carbonaceous aerosol. (a) Contributions of gasoline (blue, shaded) and diesel cars (pink, shaded) to fossil SOA (green, shaded) in the L.A. basin, incorporating ambient OH exposure measurements, chamber SOA yields for LEV1/LEV2 vehicles, and ambient tracer concentrations (see SI for methodology). Maxima-minima for total vehicular SOA given by black dashed lines. Blue markers show an estimate from Euro 5 yields for comparison. Uncertainties in OH exposures are unaccounted for, possibly explaining the time shift between model/observed peak concentrations. For the measured fossil SOA, the range is determined by propagating the best estimate of the errors in the determination of SOA from Hayes et al. and those related in the determination of SOA fossil fraction from ¹⁴C measurements in Zotter et al.. (b) Black carbon (BC), hydrocarbon-like organic aerosol (HOA) and fossil-fuel related (f) SOA and national share of diesel on-road vehicles (all types) according to the GAINS model. The composition of the fossil carbonaceous matter in Europe and the US is clearly different, with the dominance of BC from diesel emissions in Europe and of fossil SOA from gasoline emissions in the US consistent with bottom-up calculations. (c) Fractional contribution of diesel passenger cars to primary vehicular PM (gasoline + diesel) as a function of total passenger car fuel consumption and fraction of DPF diesels, at 22 °C. Dashed lines show the effect of the DPF fraction on diesel vehicles’ share of PM emissions at any given fraction fuel consumption. A projection for the EU is shown with data on the fraction of Euro 5 diesel from the TREMOVE model. For the current fleet, primary PM from gasoline cars would only exceed those from diesel when 97% of diesels are DPF-equipped. SOA would exhibit a similar, but less pronounced trend. Additional data are shown in SI Fig. S5 and SI Table S11.