Size and composition distributions of particulate matter emissions: Part 1--light-duty gasoline vehicles.

ABSTRACT

Size-resolved particulate matter (PM) emitted from light-duty gasoline vehicles (LDGVs) was characterized using filter-based samplers, cascade impactors, and scanning mobility particle size measurements in the summer 2002. Thirty LDGVs, with different engine and emissions control technologies (model years 1965-2003; odometer readings 1264-207,104 mi), were tested on a chassis dynamometer using the federal test procedure (FTP), the unified cycle (UC), and the correction cycle (CC). LDGV PM emissions were strongly correlated with vehicle age and emissions control technology. The oldest models had average ultrafine [PM.sub.0.1] (0.056- to 0.1-[micro]m aerodynamic diameter) and fine [PM.sub.1.8] ([less than or equal to]1.8-[micro]m aerodynamic diameter) emission rates of 9.6 mg/km and 213 mg/km, respectively. The newest vehicles had [PM.sub.0.1] and [PM.sub.1.8] emissions of 51 [micro]g/km and 371 [micro]g/km, respectively. Light duty trucks and sport utility vehicles had [PM.sub.0.1] and [PM.sub.1.8] emissions nearly double the corresponding emission rates from passenger cars. Higher PM emissions were associated with cold starts and hard accelerations. The FTP driving cycle produced the lowest emissions, followed by the UC and the CC. PM mass distributions peaked between 0.1-and 0.18-[micro]m particle diameter for all vehicles except those emitting visible smoke, which peaked between 0.18 and 0.32 [micro]m. The majority of the PM was composed of carbonaceous material, with only trace amounts of water-soluble ions. Elemental carbon (EC) and organic matter (OM) had similar size distributions, but the EC/OM ratio in LDGV exhaust particles was a strong function of the adopted emissions control technology and of vehicle maintenance. Exhaust from LDGV classes with lower PM emissions generally had higher EC/OM ratios. LDGVs adopting newer technologies were characterized by the highest EC/OM ratios, whereas OM dominated PM emissions from older vehicles. Driving cycles with cold starts and hard accelerations produced higher EC/OM ratios in ultrafine particles.

INTRODUCTION

Recent epidemiological studies have found positive associations between exposure to atmospheric particulate matter (PM) and increased human mortality and morbidity. (1) These associations are particularly strong for the fine and ultrafine fractions of PM (aerodynamic diameter less than 2.5 and 0.1 [micro]m, respectively). (1-3) It has been postulated that ultrafine particles (aerodynamic diameter [less than or equal to]0.1 [micro]m) may be responsible for some of the observed health effects. (4-9) There is some evidence to support the hypothesis that ultrafine PM can localize in the mitochondria of epithelial cells where they induce major structural damage. (10) A better characterization of the composition and source origins of ultrafine particles is necessary to fully investigate their relationships with human health.

Previous studies indicated that combustion processes are the dominant source of ultrafine particles in the urban atmosphere, (11) and that transportation is the leading source of particulate combustion emissions. (11,12) Because of the introduction of new vehicle technologies such as low emission vehicles (LEVs) in the 1990s, the mix of the on-road light-duty gasoline vehicle (LDGV) fleet has changed significantly in the United States over the past decade. A need exists to measure the size and composition distribution of PM emitted from present-day on-road vehicles to help quantify the contribution of these sources to ambient fine and ultrafine PM concentrations.

The purpose of this study is to report the size and composition distributions of PM released from contemporary LDGVs measured using a chassis dynamometer/dilution sampling system that employs filter-based samplers, cascade impactors, and scanning mobility particle size (SMPS) measurements. In the current study ultrafine PM is defined as particles with aerodynamic diameter between 0.056-0.1 [micro]m ([PM.sub.0.1]; as collected by stage 10 of a micro-orifice uniform deposit impactor) and fine PM as particles with aerodynamic diameters less than 1.8 [micro]m ([PM.sub.1.8]; as collected by a reference ambient air sampler filter sample). These are useful functional definitions because very little of the PM mass collected in this study had aerodynamic particle diameters below 0.056 [micro]m or above 1.8 [micro]m ([PM.sub.1.8] is functionally equivalent to [PM.sub.2.5] in the current study). The dataset includes different vehicle types and emission control technologies operated under multiple driving cycles. Particle size and composition distributions in six different size fractions between 0.056 and 1.8 [micro]m are also reported in addition to bulk [PM.sub.1.8] data. Vehicle emissions characteristics as a function of time for different technology types and driving cycles are also presented.

EXPERIMENTAL METHODS

PM emissions from LDGVs were collected at the California Air Resources Board (CARB) Haagen-Smit Laboratory (HSL) in El Monte, CA, during August and September of 2002. LDGVs were tested using a chassis dynamometer and the diluted exhaust was characterized using numerous instruments operated by multiple research groups. LDGV emissions of gas-phase hydrocarbons (HC), carbon monoxide (CO), carbon dioxide (C[O.sub.2]), methane (C[H.sub.4]), nitric oxide (NO), and nitrogen oxides (N[O.sub.x]), as well as alcohols, aldehydes, sulfates, non-methane organic gases (NMOG), and non-methane hydrocarbons (NMHCs) were measured by researchers at CARB. (13) Sodeman et al. (14) characterized PM emissions using aerosol time of flight mass spectrometers (ATOFMS). University of California-Davis (UCD) researchers collected gas-phase emissions and bulk [PM.sub.1.8] samples using annular denuders upstream of filter substrates and polyurethane foam (PUF) plugs. UCD researchers also collected [PM.sub.1.8] and size-resolved samples using filter-based samplers, micro-orifice uniform deposit impactors (MOUDIs), and SMPSs. In the present paper, PM data from the bulk, size-resolved, and time-resolved PM emission samples are reported.

Vehicle Test Fleet

A total of 30 LDGVs were selected to represent the typical on-road fleet in California, taking into account aspects such as weight, emissions control technology, mileage, and manufacturer (Table 1). On the basis of the adopted emissions reduction technology, vehicles were grouped into five categories: LEVs, three-way catalysts (TWCs), oxidation catalysts (OCATs), noncatalysts (NCATs), and vehicles observed to emit blue smoke (SMOKERs). Within these classes, they could be further classified as passenger cars (PCs) or light-duty Trucks (LDTs)/sport utility vehicles (SUVs). Separate PM samples were collected for each LDGV category to obtain enough mass for chemical analyses.

The gasoline used to power the LDGVs was purchased at retail stations throughout Los Angeles, CA. Only two vehicles required additional fuel to complete their chassis dynamometer cycles and were re-fueled at two different retail gasoline stations near the testing facility. Fuel was assumed to be California reformulated gasoline, containing less than or equal to 35 ppm sulfur. Maricq et al. (15) found no appreciable difference in fine PM emissions from a PC on a chassis dynamometer using the Federal Test Procedure (FTP) cycle between this base fuel (35 ppm sulfur) and laboratory variants containing 350 and 600 ppm sulfur. However, because the fuel sulfur content could influence sulfuric acid nucleation, data below 0.056 [micro]m is not reported in this study. Fuel and oil samples were collected from each vehicle and their composition will be reported in future studies.

LDGVs were tested using three different driving cycles: the FTP cycle, the Unified Cycle (UC), and the Correction Cycle (CC). (14) The FTP is characterized by moderate transient sections and the lowest top speed of all three cycles. The UC employs the highest acceleration and greatest speed. The CC has a higher average speed section than the FTP, with some transient driving. (13) Table 2 presents details on each of these driving cycles, and their individual driving traces are provided in each panel of Figure 6.

Table 3 summarizes the complete LDGV sample set, which consists of two background samples and 10 vehicle/class samples. Emissions from as few as two and as many as 40 total driving cycles were composited onto each LDGV sample to ensure that enough PM mass was collected to support chemical analyses. All composited sample sets are integer multiples of the number of vehicles in that test set (i.e., 40 LEV cycles represent four independent cycles from 10 separate vehicles). Total LDGV test sample times ranged from a minimum of 58 min for the vehicles with the highest emission rates (SMOKERs) to a maximum of 1253 min for the vehicles with the lowest emission rates (LEVs).

The breadth and depth of the cumulative sample set allowed several inter-comparisons to be investigated. Emissions released during full FTP cycles were compared between LEVs, TWC PCs, OCAT PCs, NCAT PCs, and SMOKERs. Emissions released from partial FTP cycles were also compared between these classes and TWC LDT/SUVs. Emissions released from partial FTP, UC, and CC cycles were compared across the same set of TWC LDT/SUV vehicles. Using this approach, comparisons between vehicle types using a constant driving cycle, and vice versa, were possible.

Sampling Methodology