Black Carbon Emissions in Gasoline Exhaust and a ReductionAlternative with a Gasoline Particulate FilterTak W. Chan,*,† Eric Meloche,† Joseph Kubsh,‡ and Rasto Brezny‡

†Emissions Research and Measurement Section, Air Quality Research Division, Environment Canada, 335 River Road, Ottawa,Ontario K1A 0H3, Canada‡Manufacturers of Emission Controls Association, 2200 Wilson Boulevard, Suite 310, Arlington, Virginia 22201, United States

*S Supporting Information

ABSTRACT: Black carbon (BC) mass and solid particle number emissionswere obtained from two pairs of gasoline direct injection (GDI) vehicles andport fuel injection (PFI) vehicles over the U.S. Federal Test Procedure 75 (FTP-75) and US06 Supplemental Federal Test Procedure (US06) drive cycles ongasoline and 10% by volume blended ethanol (E10). BC solid particles wereemitted mostly during cold-start from all GDI and PFI vehicles. The reduction inambient temperature had significant impacts on BC mass and solid particlenumber emissions, but larger impacts were observed on the PFI vehicles thanthe GDI vehicles. Over the FTP-75 phase 1 (cold-start) drive cycle, the BC massemissions from the two GDI vehicles at 0 °F (−18 °C) varied from 57 to 143mg/mi, which was higher than the emissions at 72 °F (22 °C; 12−29 mg/mi) by a factor of 5. For the two PFI vehicles, the BCmass emissions over the FTP-75 phase 1 drive cycle at 0 °F varied from 111 to 162 mg/mi, higher by a factor of 44−72 whencompared to the BC emissions of 2−4 mg/mi at 72 °F. The use of a gasoline particulate filter (GPF) reduced BC emissions fromthe selected GDI vehicle by 73−88% at various ambient temperatures over the FTP-75 phase 1 drive cycle. The ambienttemperature had less of an impact on particle emissions for a warmed-up engine. Over the US06 drive cycle, the GPF reduced BCmass emissions from the GDI vehicle by 59−80% at various temperatures. E10 had limited impact on BC emissions from theselected GDI and PFI vehicles during hot-starts. E10 was found to reduce BC emissions from the GDI vehicle by 15% at standardtemperature and by 75% at 19 °F (−7 °C).

■ INTRODUCTION

Atmospheric black carbon (BC) particles are generated fromincomplete combustion processes (e.g., internal combustionengines). When these particles are emitted to the atmosphere,they absorb solar radiation, influence cloud processes, and alterthe melting of snow and ice surfaces.1 BC particles are relativelyinert in the atmosphere but typically have atmospheric lifetimesfrom days to a week1−3 and are considered short-lived pollutants.Atmospheric BC particles are removed by dry and wetdeposition. Exposure assessment studies have shown that BCparticle exposure is linked to various human health issues, such aslung function and blood pressure.4,5 Current research suggeststhat atmospheric BC particles have an overall positive climateforcing only second to carbon dioxide. Reducing the emissions ofshort-lived BC particles could lead to an immediate reduction oflocalized heating in environmentally sensitive areas, such as theArctic, and also bring additional health benefit by reducingparticulate matter (PM) exposure by commuters.1

Modern diesel vehicles are getting cleaner due to thedeployment of efficient emission controls such as dieselparticulate filters to comply with the more stringent regulationson PM.40 While the heavy-duty fleet is progressively gettingcleaner with the new diesel trucks replacing older trucks, therelative BC emissions attributed to the light-duty gasolinevehicles, to the overall BC emissions, is expected to gradually

increase. In North America, the current light-duty vehicle fleet isdominated by gasoline port fuel injection (PFI) vehicles.Gasoline direct injection (GDI) vehicles are one of the latestadditions to the light-duty fleet. The concept of GDI is not new,and the first automotive application of this technology can betraced back to the 1950s, but the popularity of the GDI vehiclehas traditionally been limited due to high production costs andtechnological barriers. The first mass-produced, modern GDIvehicles were introduced to the marketplace in the late 1990s. Atfirst, the concept quickly gained popularity among Japanese andEuropean manufacturers, but since then, technological break-throughs and mass production furthered the deployment of thistechnology in the North American and all other major globalmarkets. Compared to the traditional PFI engines, GDI enginesoffer many advantages, such as better fuel injection control, lowerfuel consumption, less fuel pumping loss, higher compressionratio, and charge air cooling. All these advantages allow adownsized GDI engine to deliver equal or better performancethan a larger PFI counterpart.6,7,26,35 All of the abovecharacteristics position GDI engines as a favorable technologyfor reducing CO2 emissions to help gasoline vehicles meet more

Received: December 9, 2013Accepted: April 23, 2014Published: April 23, 2014

Article

pubs.acs.org/est

© 2014 American Chemical Society 6027 dx.doi.org/10.1021/es501791b | Environ. Sci. Technol. 2014, 48, 6027−6034

stringent emission standards that target fuel consumption and/orvehicle greenhouse gases. At the same time, several recent studieshave shown that GDI vehicles could emit more PM thantraditional PFI vehicles and heavy-duty trucks equipped withdiesel particulate filters,6,10,11 challenging current GDI vehicles tocomply with the recently finalized California LEV III and U.S.EPA Tier 3 particulate emissions standards without the use ofnew emission control strategies, such as a gasoline particulatefilter (GPF).The major difference between the PFI and GDI engines lies in

the fuel injection method and mixture preparation. This leads tothe potential for wetting the cylinder wall in the GDI engines. Inaddition, the inhomogeneity of the fuel mixture under stratifiedoperation inGDI engines during the compression stroke tends tocontribute to higher particulate emissions compared to theexhaust emissions from traditional PFI engines.6,8−11,35,38

Summarizing available literature provides some general trendswith respect to particulate emissions. For example, cold-start atlow ambient temperature is found to significantly increaseparticle number emissions from a PFI vehicle compared to a GDIvehicle primarily due to the difference in fuel injection andmixture preparation strategies between GDI and PFI engines.This difference typically causes PFI engines to inject significantlymore fuel during cold-start at low ambient temperatures tocompensate for the poor volatility of the fuel.6,9,32 Laboratoryand on-road measurements have also shown that nanoparticleemissions from gasoline spark ignition vehicles can be stronglydependent on vehicle speed and engine load.21−23 The impact offuel ethanol content on BC formation is also not straightforward.A detailed chemical kinetic model suggested that oxygenatedcompounds, such as ethanol, could reduce BC formation becausethe oxygenated carbon is more strongly bonded than non-oxygenated carbon atoms, making the oxygenated carbonunavailable to form BC precursors during combustion, leadingto overall less BC formation.24 However, observations fromflame studies suggest that ethanol can both increase or decreaseBC formation depending on whether a nonpremixed orpremixed flame is involved during the combustion process.25−28

This situation is further complicated when splash blendingethanol with gasoline as this changes several important fuelproperties, such as the aromatic content, vapor pressure, anddistillation profile of the resulting fuel.36,37 These changes in fuelproperties affect particle formation during combustion as well asthe cold and hot weather driveability of the vehicle. The result ismixed observations of particle emissions on ethanol blendedgasoline depending on the type of fuel injection system and theoperating conditions of the vehicle.8,9,11,29−31,38,39 A brief reviewof some recent observations is summarized in Chan et al.8

With the anticipation of more GDI vehicles on the road in thenear future,12 it is important to get a better understanding of theemission characteristics, such as BC mass and solid particlenumber emissions, from modern GDI engines compared totraditional PFI engines because the relative contribution toparticle emissions from GDI vehicles is expected to increasegradually over time. Also, understanding how various factors,such as ambient temperature, driving pattern, and fuel ethanolcontent, influence the BC mass emissions from different enginetechnologies also provides useful information for better assessingfuture emissions in various scenarios. As of September 1, 2010,the Federal Government of Canada has regulated that motorgasoline sold in Canada must contain an annual pool average of5% ethanol while different mandates also exist among differentprovinces. As a result, gasoline sold in Canada may contain up to

10% ethanol. This could have an impact on affecting the BCemissions from on-road GDI and PFI vehicles across differentregions of Canada. By using concurrent solid particle numberand BCmass measurements, this study provides a systematic wayof evaluating emissions from two pairs of GDI and PFI vehiclesover two different drive cycles and at different ambienttemperatures to understand how solid particle number and BCmass relationships vary under the influence of different factors. Inthis study, an aethalometer was used to measure real-time BCemissions while the solid particles were measured using aEuropean Union Particle Measurement Programme compliantsystem. In anticipation that GPFs could become a strategy toreduce BC emissions from future GDI vehicles, this study alsopresented the performance of a prototype GPF on removing BCparticles from a selected GDI vehicle at various conditions.

■ EXPERIMENTAL SECTIONVehicles, Test Details, and Ambient Conditions. Two

midsize sedans and two compact vehicles were used in this study.The midsize sedans were a 2011 2.4 L Hyundai Sonata wall-guided, stoichiometric GDI vehicle (GDI#1) and a 2010 2.4 LVolvo S40 PFI vehicle (PFI#1). The GDI#1 vehicle representsone of the early versions of available GDI vehicles that wereintroduced into the North American market during the start ofthis test program in 2011 while the PFI#1 vehicle is a comparablePFI vehicle in terms of vehicle weight, engine size, and categoryof emission compliance. The two compact vehicles were a 20122.0 L Ford Focus wall-guided, stoichiometric GDI vehicle(GDI#2) and a 2013 2.0 L Ford Transit Connect PFI vehicle(PFI#2). The two compact vehicles are examples of some morerecent vehicles with more advanced improvements but withsmaller engine displacements compared to the GDI#1 and PFI#1vehicles. All vehicles were mileage accumulated on-road prior toconducting all the emission measurements.8 All vehicles wereequipped with three-way catalytic converters (TWCs). Asummary of the specifications of all tested vehicles is given inTable S1 (Supporting Information). The two fuels tested were aTier 2 certification gasoline (E0) and a splash blended 10% byvolume blend of ethanol with certification gasoline (E10). Thefuel specifications and distillation profiles for the test fuels aresummarized in TableS S2 and S3 (Supporting Information).Most of the E0 measurements were conducted at three ambienttemperatures (72 °F/22 °C, 19 °F/−7 °C, 0 °F/−18 °C). Atemperature of 19 °F was chosen to be consistent with thetemperature used for cold CO compliant emission testing while 0°F was chosen to represent one commonly encountered wintertemperature in many parts of Canada. Due to availability of thetest cell, some of the US06 and E10 measurements were onlyconducted at selected temperatures. Table S4 (SupportingInformation) summarizes the test cycle and different testconditions for all four vehicles used in this study. Typicallythree or four repeats were conducted for all test conditions(unless noted otherwise).The drive cycles used in this study were the U.S. Federal Test

Procedure (FTP-75) and US06 Supplemental Federal TestProcedure (US06). The FTP-75 is a city driving cycle, designedto broadly represent the driving conditions in the U.S. The cycleconsists of three phases (i.e., cold-start, urban, and hot-startphases). Phases 1 and 3 of the FTP-75 drive cycle are identicalwith phase 1 being a cold-start. Phase 2 of the FTP-75 drive cyclerepresents a typical urban driving pattern that includes moderateaccelerations and decelerations. The US06 drive cycle is anaggressive driving cycle aimed to simulate aggressive driving

Environmental Science & Technology Article

dx.doi.org/10.1021/es501791b | Environ. Sci. Technol. 2014, 48, 6027−60346028

conditions.20,35 Additional details regarding the drive cycles andtest procedure are given in the Supporting Information. A figuresummarizing the speed trace for the test sequence of a typical testday is given in Figure S2 (Supporting Information).Gasoline Particulate Filter. The GPF used on the GDI#1

vehicle was a custom design device provided by theManufacturers of Emission Controls Association (MECA).The noncatalyzed, wall-flow GPF was made using a cordieriteceramic material and was 5.66 in. (14.4 cm) in diameter and 6 in.(15.2 cm) in length. The GPF was optimized to lower the effectson back pressure and did not increase the fuel consumption overthe test cycles used in this study.9 The cell density of the GPF is200 cells per square inch (cpsi) with 12 mil (0.3 mm) walls andapproximately 50% wall porosity. When in use, the GPF wasinstalled in the underfloor position about 18 in. (46 cm)downstream of the original underfloor TWC converter, replacingthe original resonator.The particle filtration principle of the test GPF is similar to a

noncatalyzed wall-flow diesel particulate filter. Starting from aclean filter, filtration efficiency is typically low, and exhaustparticles are expected to deposit inside the porous wall of theGPF, resulting in a change in wall microstructure (i.e., deep-bedfiltration). When the pore space becomes filled, the additionalparticles begin to deposit on the channel walls (inside thechannels) forming a thin soot layer (i.e., soot-cake filtration).Gradually, the soot layer further develops, and this significantlyimproves the filtration efficiency, particularly for ultrafineparticles, as the soot layer also acts as an additional filtrationmedium.13,14,41 Typically, filtration efficiency is the lowest forparticles with diameters of about 100−200 nm because theseparticles are too large to be effectively removed by diffusion andtoo small to be removed by impaction.8,9 As solid particlescontinue to be accumulated in the GPF, the filter graduallybecomes overloaded and eventually leads to filter clogging,reduces the collection capacity, and increases back pressure. Thisproblem is resolved by thermal regeneration of the filter, which inthis case, is triggered by the hot exhaust gas temperature(typically >1022 °F/550 °C). During regeneration, theaccumulated PM will be oxidized (by NO2 and O2) andaccompanied by the release of a large number of ultrafineparticles.8,9,41 After regeneration, the filtration efficiency drops,and the filtration mechanism restarts all over again. The vehicletesting and prepping procedure ensured the emission tests tobegin with a clean GPF at the beginning of the test day. Previousmeasurements showed that regeneration did not occur over theFTP-75 drive cycle except during 0 °F (−18 °C) testing while thehigh exhaust gas temperature generated during the US06 drivecycle was able to caused periodic regeneration during the courseof the cycle.8,9

Sampling Setup. The complete sampling setup is given inFigure S3 (Supporting Information). Descriptions of theindividual components and all gaseous emissions results weredocumented previously8,9 and will not be discussed here. For thepurpose of this paper, descriptions here are only limited to thesolid particle and BC analytical methods. During the testing ofthe GDI#1 and PFI#1 vehicles, diluted exhaust was extracteddirectly from the constant volume sampling (CVS) dilutiontunnel and introduced into a European Union ParticleMeasurement Programme (PMP) compliant system for solidparticle detection (Supporting Information). A Magee ScientificAE51 microaethalometer was also included downstream of thePMP volatile particle remover system to measure BC massconcentration in parallel with the solid particle measurements.

For the testing of GDI#2 and PFI#2 vehicles, the micro-aethalometer was set up to extract diluted exhaust directly fromthe CVS dilution tunnel with secondary dilution. BC measure-ments from the GDI#1 and PFI#1 vehicles obtained using bothconfigurations were compared to verify the consistency of the BCmeasurements between the two methods.

Black Carbon Detection with the Microaethalometer.The AE51 microaethalometer has the same operating principleas a conventional aethalometer.15 Detailed descriptions andderivations of the BC mass concentration from the aethalometermeasurements are documented elsewhere16,17 (SupportingInformation). Briefly, in the microaethalometer, incomingparticles are deposited continuously onto a Teflon-coatedborosilicate glass fiber filter ticket where the light intensities,derived from a LED source at 880 nm, transmitted through thesampled spot and a reference blank spot are comparedcontinuously to determine the light attenuation. The change inlight attenuation at any given time interval is related to the lightabsorption coefficient that can be used to infer BC massconcentration using the predetermined mass attenuation crosssection from the manufacturer.Several studies have shown the need to apply postmeasure-

ment artifact corrections to aethalometer measurements due tothe multiple scattering corrections (caused by the filter fibers)and the filter loading corrections16−18 (Supporting Information).All the microaethalometer measurements reported in this studywere sampled in 1 Hz time resolution. A noise reductionprocedure19 was first applied to the measurements beforeapplying the filter loading and multiple scattering corrections17

to the data. The artifact corrected microaethalometer BCmeasurements were shown to be linearly correlated with thethermally determined elemental carbon (R = 0.83) and therefractory carbonmass measured by laser-induced incandescence(R = 0.91) over the mass concentration range from 0.1 to 10 mg/mi (Supporting Information).

■ RESULTS AND DISCUSSIONImpact of Cold-/Hot-Start and Ambient Temperature

on Black Carbon Emissions. In order to compare the impactof cold- and hot-starts on emissions, results from phase 1 andphase 3 portions of the FTP-75 drive cycle (both share the samevehicle speed trace with phase 1 being a cold-start) are compared.Figure 1 shows how the average BC mass (mg/mi; typicalnumber of repeats = 3−4) (left panel) and the solid particlenumber (particles/mi) (right panel) emissions associated withthe FTP-75, phase 1 drive cycle (cold-start) vary with ambienttemperature. Solid markers correspond to measurements fromtheGDI#1 and PFI#1 vehicles while openmarkers correspond tothe GDI#2 and PFI#2 vehicles. Corresponding results for theFTP-75 phase 3 drive cycle (hot-start) are given in Figure 2. Allthe emission rates are summarized in Tables S6 and S7(Supporting Information). In both Figures 1 and 2, the trendsfor the BC mass and solid particle number emissions wereconsistent suggesting that the solid particles emitted from boththe GDI and PFI vehicles are light absorbing, consistent with anumber of studies that reported a linear relationship betweensolid particle number and BC or PM mass.33−35,42 In all cases,cold-start emissions increased with decreasing ambient temper-atures. At standard temperature, cold-start BC mass emissionsfrom the two GDI vehicles represented 65−84% of the total BCmass emitted over the entire FTP-75 drive cycle. For the two PFIvehicles, cold-start BC mass attributed to 86−100% of the totalBC mass emissions over the entire FTP-75 drive cycle. These

Environmental Science & Technology Article

dx.doi.org/10.1021/es501791b | Environ. Sci. Technol. 2014, 48, 6027−60346029

observations suggest that GDI BC emissions are related to themethod of fuel injection caused by inhomogeneous mixingbetween air and fuel.35 For PFI vehicles, BC emissions are mostlydue to cold-starting and are related to the incompletevaporization of the fuel and wall wetting because PFI enginestend to overfuel during cold start to compensate for the poorvolatility of the fuel.6,32

When comparing the different pairs of GDI and PFI vehicles,some interesting trends emerge. Over the cold-start, the ratio ofthe stock GDI#1 BC mass to PFI#1 BC mass emissions variedfrom 18 at standard temperature to 1.3 at 0 °F (−18 °C). Thecorresponding ratio of stock GDI#1 solid particle number toPFI#1 solid particle number emissions varied from 7.8 atstandard temperature to 1.4 at 0 °F. When switching to thesmaller displacement GDI#2 and PFI#2 vehicles, increasing BCemissions at low ambient temperatures from both vehicles werestill observed. Interestingly, the BC mass emission trends fromthe GDI#2 vehicle at cold ambient temperatures weresignificantly lower than from the PFI#2 vehicle at the sameconditions, although the solid particle emission trends remained

similar compared to the GDI#1 vehicle and were higher than thePFI#2 vehicle. The difference in BC emission trends between thetwo pairs of GDI and PFI vehicles could be related to differencesin how engines with differing displacements respond to the sameengine load during the two drive cycles. It could also be related tothe different optimized emission control strategies adopted bythe different vehicle manufacturers during cold-start conditions.Future studies are planned to determine if the morphology of theBC particles may have a role in altering the light absorbingproperty of the solid particles that are emitted between the twodifferent GDI vehicles.After the GPF was installed on the GDI#1 vehicle, BC mass

emissions were reduced significantly. At standard temperatureand 19 °F (−7 °C), the GDI#1 post-GPF BC mass emissionswere lower compared to the stock GDI configuration by 73% and88%, respectively. Despite the occurrence of soot regeneration at0 °F (−18 °C),9 the BC mass emission reduction was stillmaintained at 85%. These reductions were consistent with thesolid particle number emission reductions of 75%, 85%, and 68%at standard temperature, 19 °F, and 0 °F, respectively. The lowersolid particle number filtration efficiency at 0 °F compared to BCmass emission reduction was caused by the presence of a largenumber of ultrafine particles emitted during the sootregeneration process, which generated a large number ofultrafine particles, ranging from 6 nm to almost 50 nm indiameter, downstream of the GPF when sampling from the CVStunnel (Supporting Information). Some of these ultrafineparticles contributed to the solid particle number measurementsleading to reduced solid particle filtration efficiency at 0 °F.However, these particles did not significantly contribute to thetotal light absorption detected by the microaethalometer, likelydue to their small particle mass and possibly nonabsorbingnature, and therefore, the BCmass filtration efficiency at 0 °Fwasstill high.Compared to the cold-start results (Figure 1), the hot-start

emission results (Figure 2) were lower by orders ofmagnitude. Ingeneral, both the stock GDI#1 and GDI#2 BC mass emissionswere comparable, varying from about 1.6−2.5 mg/mi at standardtemperature to 4.3−4.5 mg/mi at 0 °F. Although much lowerthan the cold-start results, these values were still highercompared to both the PFI#1 and PFI#2 BC mass emissions,which varied from nondetectable to 0.6 mg/mi at standardtemperature to 0.01−0.5 mg/mi at cold ambient temperatures.The GDI#1 post-GPF BC mass emissions exhibited a slightly

different trend: they varied from about 0.1 mg/mi at standardtemperature to nondetectable at 0 °F, consistent with the solidparticle number emissions trend. The different post-GPF BCemission trends over various ambient temperatures for phases 1and 3 can be explained by the different GPF filtration stages. Aclean GPF (prior to the start of phase 1) has low particle filtrationefficiency. Thus, the postclean GPF BC emissions at variousambient temperatures followed the stock GDI BC emissiontrend. It takes a finite amount of time for the soot layer to developin the GPF before effective filtration begins, and the higher BCemissions at low ambient temperatures reduced the transitiontime from deep-bed filtration to soot-cake filtration, resulting inthe apparent improvement in the particle filtration efficiency inthe GPF (over phase 3) at colder ambient temperatures.9 Theimportance of a well-developed soot layer in improving filtrationefficiency was also demonstrated by the continued reduction inpost-GPF particle number emissions over subsequent cold-startLos Angeles Route Four drive cycles (i.e., phases 1 and 2 of FTP-

Figure 1. BC mass and solid particle number emissions as a function ofambient temperature over the FTP-75 phase 1 drive cycle. Solid markersrepresent measurements from the GDI#1 (circle and square) and PFI#1(triangle) vehicles while open markers represent data from the GDI#2(circle) and PFI#2 (triangle) vehicles.

Figure 2. BC mass and solid particle number emissions as a function ofambient temperature over the FTP-75 phase 3 drive cycle. Solid markersrepresent measurements from the GDI#1 (circle and square) and PFI#1(triangle) vehicles while open markers represent data from the GDI#2(circle) and PFI#2 (triangle) vehicles.

Environmental Science & Technology Article

dx.doi.org/10.1021/es501791b | Environ. Sci. Technol. 2014, 48, 6027−60346030

75) that did not destroy or regenerate the deposited materials inthe GPF between these test cycles.8

Impact of Drive Conditions on Black Carbon Emissions.Observations have shown that nonsteady state and aggressivedriving conditions could have large influences on particleemissions.21−23 To understand how the different drive patternsmay influence particle emissions, hot-start emission results fromthe FTP-75 phases 2 and 3 as well as US06 drive cycles werecompared.As illustrated in Figure S2 (Supporting Information), the

phases 2 and 3 portions of the FTP-75 and the US06 drive cyclesrepresent three very different drive patterns in the order ofincreasing driving speed and aggressiveness (SupportingInformation). Figure 3 shows the comparison of the (a) solid

particle number and (b) BC mass emissions (typical number ofrepeats = 3−4) from the different GDI and PFI vehicles over thethree different driving patterns. In all cases, only standardtemperature and 0 °F (−18 °C) results are shown for clarity. Ingeneral, the solid particle number and BC mass emission trendsare similar and suggest that emissions from the PFI vehiclesdecreased with decreasing drive speed and aggressiveness. Thetwo GDI vehicles show much different patterns. Both GDIvehicles had either similar or much higher emissions during theless aggressive FTP-75 phases 2 and 3 drive cycles than for theUS06 drive cycle, compared to their corresponding PFIcounterparts.

The drive pattern for the phase 2 portion of the FTP-75 drivecycle is more transient in nature compared to the phase 3 portionof the FTP-75 drive cycle and for the US06 drive cycle. Inaddition, the average and maximum vehicle speeds for the phase2 portion are also the lowest (Supporting Information). Theobservations on the particle number and BC mass emissionsfrom the different drive patterns for both GDI vehicles suggestthat the additional transient driving nature, even though atrelatively low speeds, led to increased particle emissions from thewarmed GDI vehicles compared to their PFI counterparts. Thiscould be related to the different modes of operation of the GDIengine. At low and medium engine load and speed conditions,the GDI engine typically operates under stratified-charge orhomogeneous lean operationmode, when a compact spray with areduced penetration rate is injected during the compressionstroke, to improve fuel economy.6 BC particle formation ispossible at localized fuel-rich zones during stratified-charge modeoperation. At high engine load and speed, particularly with theneed for fast acceleration, a GDI engine typically operates in ahomogeneous-charge mode using early fuel injection. The bettermixing associated with this type of operation helps to suppressthe formation of BC.6 The exact transition between the differentfuel injection modes varies depending on the engine design thusresulting in different emission trends for different GDI vehicles.In the PFI vehicle case, it would appear that the need to injectmore fuel during high engine load and speed conditions (e.g.,US06) may lead to possible incomplete vaporization of the fuelthat then contributed to the increased BC particle emissionsfrom the two PFI vehicles.The BC mass emissions from the GDI#1 post-GPF

configuration were higher during the US06 drive cycle thanthose from the phases 2 and 3 portions of the FTP-75 drive cycle(Table S6, Supporting Information). However, post-GPF BCmeasurements from the GDI#1 vehicle over the US06 drive cyclewere still much lower than the BC emissions from the stock GDIconfiguration at all ambient temperatures by 59−80%. This isconsistent with the 60−83% reduction in solid particle numberemission over the same test conditions (Table S7, SupportingInformation). These observations are consistent with themultiple spontaneous soot regenerations during the US06drive cycle that impacted the integrity of the soot layer, leadingto decreased particle filtration efficiency, compared to thefiltration efficiency observed over the FTP-75 drive cycle.8,9

Impact of Fuel Ethanol Content on Black CarbonEmissions. Tables S8 and S9 (Supporting Information)summarize the BC mass and solid particle number emissionsfor all four GDI and PFI vehicles (typical number of repeats = 3−4) when operated on E10 fuel. Measurements from the GDI#1vehicle with the GPF installed are also included. Thesemeasurements are presented graphically in Figure 4. Theuncertainties in the figure are the standard deviation andrepresent approximately the margin of error in a 95% confidenceinterval (Supporting Information). Comparing the E10 data withthe E0 results (Tables S6 and S7, Supporting Information) showsthat E10 has minimum impact on both BC mass and solidparticle number emissions from the PFI vehicles in general. ThePFI#1 vehicle appeared to have lower BC mass emissions whenoperated on E10 at 0 °F (−18 °C) although these changes are notstatistically significant.For the GDI#1 vehicle, E10 has a mixed impact on BC

emissions over the US06 and the phases 2 and 3 portions of theFTP-75 drive cycles, suggesting that ambient temperature, driveconditions, and the physical fuel properties of E10 all have

Figure 3. (a) Solid particle number and (b) BC mass emissions for allfour GDI and PFI vehicles over the three different driving patterns. Solidbars represent low ambient temperature measurements whereas open,dashed bars represent standard temperature measurements.

Environmental Science & Technology Article

dx.doi.org/10.1021/es501791b | Environ. Sci. Technol. 2014, 48, 6027−60346031

different degrees of impacts on the BC mass emissions from thisGDI vehicle at different ambient temperatures. In comparison,E10 was observed to have a much larger impact on BC massemissions from both GDI vehicles over certain conditions of thephase 1 portion of the FTP-75 drive cycle (i.e., during cold-start;Figure 4). For example, BC emissions from the GDI#1 vehicle onE10 at standard temperature and 19 °F (−7 °C) were lower thanthe corresponding emissions on E0 by 52% and 75%, respectively(Figure 4). At 0 °F (−18 °C), E10 was found to increase BCemissions from the GDI#1 vehicle by 21% although results werenot statistical significant within a 95% confidence interval. Forthe GDI#2 vehicle, E10 BC emissions over the phase 1 portion ofthe FTP-75 drive cycle at standard and 0 °F were lower than theE0 emissions by 66% and 30%, respectively (no measurements at19 °F for E10 for comparison).The observations from this study suggest that E10 may have

some benefit for GDI engines. Comparing between E0 and E10,E10 has a lower heating value and a larger volume of E10 fuel isneeded to provide the same amount of energy as for E0.29,43

However, the distillation curve also shows that splash blendingethanol will raise the Reid Vapor Pressure (RVP) and increasethe percent fuel evaporated in the 140−200 °F (60−93 °C)range44 (Supporting Information). Compounds in the midrangeof the distillation curve (i.e., the percentage of fuel evaporated upto about 212 °F (100 °C)) are known to have an influence on thewarm-up and the cold and hot weather driveability of thevehicle.37,43 As measurements between the PFI E0 and E10measurements in this study were not statistically different, thissuggests that all the PFI test vehicles adapted well to the E10 fuel,and the increased E10 fuel usage,29 even at low ambienttemperature with cold-starting, is largely compensated for by theincreased RVP of the fuel and did not lead to significantdegradation of the vehicle driveability. Injecting fuel directly intothe combustion chamber helps to reduce the over fueling issueduring cold-start and limits the opportunity of excessive liquidfuel impingement on the piston and cylinder wall and reduces BCemissions.11,38 As GDI engines typically use late injection duringpart load operation, the leaning effect of the E10 fuel couldfurther help to limit BC emissions.31

Black Carbon Mass and Solid Particle NumberRelationship. A particle number standard has been adoptedin Europe as part of the Euro 5/6 light-duty vehicle and Euro VIheavy-duty vehicle emission requirements in order to address the

need for a more sensitive method to detect particle emissionsfrom very low emitting vehicles. A number of studies havereported a positive relationship between particle number and BC(or PM) mass suggesting particle number measurements couldbe used to infer refractory (or particle) mass. Giechaskiel et al.33

summarized results from more than 20 heavy-duty engines andmore than 150 light-duty vehicles and observed that the solidparticle number to total PM mass ratio varied from 1 to 4 × 1012

particles/mg. When using refractory BC mass instead of PMmass, variability of the measurements was improved, resulting inan increased ratio. Their measurements showed that solidparticle number to BC mass ratio approached 6 × 1012 particles/mg for particles with a count median diameter (CMD) of 50 nmand the ratio decreased to 1× 1012 particles/mg for a CMD of 75nm. Khalek et al.42 reported a solid particle number to PM massratio of 2.3 × 1012 particles/mg and 3.2−3.9 × 1012 particles/mgfor solid particle number to BC mass for a GDI vehicle (particlediameter ∼45−60 nm). In two studies by Maricq et al.,34,35 asolid particle number to PM mass ratio of 1.9−3.0 × 1012

particles/mg were observed for various GDI vehicles (particlediameters ∼60−90 nm).Figure 5 shows the correlation between solid particle number

and BC mass measurements at all ambient conditions observedfrom this study. The three straight lines on the figure represent

Figure 4. BCmass emissions from all four GDI and PFI vehicles as a function of ambient temperature over the phase 1 portion of the FTP-75 drive cyclewhen operated on E0 (solid color) and E10 (hashed area).

Figure 5. Variations of the solid particle number emissions versus BCmass emissions over the US06 and individual phases of the FTP-75 drivecycles.

Environmental Science & Technology Article

dx.doi.org/10.1021/es501791b | Environ. Sci. Technol. 2014, 48, 6027−60346032

three different solid particle number to BC mass ratios that varyfrom 0.2 × 1012 to 2 × 1012 particles/mg. Consistent with otherstudies, measurements in Figure 5 show that solid particlenumber and BC mass were generally linearly correlated. Most ofthe scattering occurs at low concentrations and was attributedmostly to the PFI#1 and PFI#2 vehicles measurements over thephases 2 and 3 portions of the FTP-75 drive cycle, when the BCmass measurements were often too low to be accuratelydetermined. Measurements from all four vehicles were typicallybounded between the ratios of 1−2 × 1012 particles/mg. Incomparison, the cold temperature measurements over the FTP-75 drive cycle for various vehicles appeared to approach the solidparticle number to BC mass ratio of about 0.2 × 1012 particles/mg (Supporting Information). Particle number size distributions(Supporting Information) showed that the geometric meandiameter of the particles shifted from 60 nm at standardtemperature to as much as 90 nm at 0 °F (−18 °C). The size shiftcould be caused by particle coagulation due to the increasedparticle number emissions at cold ambient temperatures. A shiftof the particle number size distribution may not have a significantimpact on the solid particle number measurements as mostparticles are larger than 23 nm. Larger BC particles absorbsignificantly more light than smaller BC particles, and thus,increased particle diameter could increase the BC massmeasurements and lower the particle number to BC mass ratio.The reduced particle number to BC mass ratio with increasingparticle diameter trend in this case was consistent with theobservation from Giechaskiel et al.33

Implications. Observations from this study revealed severalimportant implications regarding BC emissions obtained at real-world, on-road ambient conditions. First of all, both BCmass andsolid particle number emissions increased considerably withdecreasing ambient temperature, particularly evident duringcold-starts from both the GDI and PFI vehicles. In addition, thisstudy also showed that driving conditions have different impactson BC emissions between GDI and PFI vehicles. The twowarmed-up PFI vehicles had very low emissions over the phases 2and 3 portions of the FTP-75 drive cycle compared to emissionsobserved during the US06 drive cycle. However, the twowarmed-up GDI vehicles still emitted considerable amounts ofBCmass during phases 2 and 3 portions of the FTP-75 and US06drive cycles. Although the change in BC mass emissions at lessdemanding driving conditions varies from one GDI vehicle toanother, these emission levels were still considerably higher thantheir corresponding PFI counterparts. This implies thatinformation such as the relative fleet mix of GDI and PFIvehicles could be important in projecting ambient BC emissionsin areas with mixed amounts of highways and intracity roadswhere driving conditions vary considerably.Another observed finding was that the 10% by volume ethanol

in gasoline had limited impact on BC mass emissions from bothPFI test vehicles. However, a significant reduction in BC massemissions, up to 52−66%, from the two GDI vehicles wasobserved during the initial engine start up and operation at 22 °C.The reduction in BC mass emissions at low ambient temper-atures associated with the ethanol-containing fuel varied betweenthe different GDI vehicles.Particles that are emitted from GDI engines are primarily solid

in nature; therefore, a GPF with a similar operating principle as adiesel particulate filter was expected to be useful in removingthese emission particles. The assessment of the BC massemissions from the prototype GPF used in this study showed thatan optimum designed GPF can provide one alternative to reduce

BC mass and solid particle number emissions from future GDIvehicles without compromising fuel economy. Over the FTP-75phase 1 drive cycle, the GPF reduced BC emissions by 73−88%at various ambient temperatures. Over the US06 drive cycle, theBC emissions were reduced by 59−80%.

■ ASSOCIATED CONTENT*S Supporting InformationFurther details on background of the current emission testprogram, vehicles, fuels, test cycles, test procedure, samplingsetup, error analysis and margin of error determination,microaethalometer BC measurements, additional figures, andtabulated emissions data. This material is available free of chargevia the Internet at http://pubs.acs.org.

■ AUTHOR INFORMATIONCorresponding Author*Phone: (613) 998-7913; fax: (613) 952-1006; e-mail: tak.chan@ec.gc.ca.

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThe authors would like to acknowledge the contribution of theERMS staff for their assistance in conducting this vehicleemissions research project. Financial support was provided bythe Government of Canada’s Program of Energy Research andDevelopment (PERD) P&E, Project C11.006, AFTER10, andMECA. The prototype GPF for this work was provided byMECA. The authors would also like to thank the anonymousreviewers for their helpful comments.

■ REFERENCES(1) Bond, T. C.; Doherty, S. J.; Fahey, D. W.; Forster, P. M.; Bemtsen,T.; DeAngelo, B. J.; Flanner, M. G.; Ghan, S.; Kar̈cher, B.; Koch, D.;Kinne, S.; Kondo, Y.; Quinn, P. K.; Sarofim, M. C.; Schultz, M. G.;Schulz, M.; Venkataraman, C.; Zhang, H.; Zhang, S.; Bellouin, N.;Guttikunda, S. K.; Hopke, P. K.; Jacobson, M. Z.; Kaiser, J. W.; Klimont,Z.; Lohmann, U.; Schwarz, J. P.; Shindell, D.; Storelvmo, T.; Warren, S.G.; Zender, C. S. Bounding the role of black carbon in the climatesystem: A scientific assessment. J. Geophys. Res.: Atmos. 2013, 118,5380−5552.(2) Ramanathan, V.; Carmichael, G. Global and regional climatechanges due to black carbon. Nat. Geosci. 2008, 1, 221−227.(3) Cape, J. N.; Coyle, M.; Dumitrean, P. The atmospheric lifetime ofblack carbon. Atmos. Environ. 2012, 59, 256−263.(4) Suglia, S. F.; Gryparis, A.; Schwartz, J.; Wright, R. J. Associationbetween traffic-related black carbon exposure and lung function amongurban women. Environ. Health Perspect. 2008, 116 (10), 1333−1337.(5) Wilker, E. H.; Baccarelli, A.; Suh, H.; Vokonas, P.; Wright, R. O.;Schwartz, J. Black carbon exposures, blood pressure, and interactionswith signle nucleotide polymorphisms in MicroRNA processing genes.Environ. Health Perspect. 2010, 118 (7), 943−948.(6) Zhao, F.; Lai, M. C.; Harrington, D. L. Automotive spark-igniteddirect-injection gasoline engines. Prog. Energy Combust. Sci. 1999, 25,437−562.(7) Alkidas, A. C. Combustion advancements in gasoline engines.Energy Convers. Manage. 2007, 48, 2751−2761.(8) Chan, T. W.; Meloche, E.; Kubsh, J.; Rosenblatt, D.; Brezny, R.;Rideout, G. Evaluation of a gasoline particulate filter to reduce particleemissions from a gasoline direct injection vehicle. SAE Int. J. Fuels Lubr.2012, 5 (3), 1277−1290.(9) Chan, T. W.; Meloche, E.; Kubsh, J.; Brezny, R.; Rosenblatt, D.;Rideout, G. Impact of ambient temperature on gaseous and particle

Environmental Science & Technology Article

dx.doi.org/10.1021/es501791b | Environ. Sci. Technol. 2014, 48, 6027−60346033

emissions from a direct injection gasoline vehicle and its implication onparticle filtration. SAE Int. J. Fuel Lubr. 2013, 6 (2), 2013-01-0527.(10) Graskow, B. R.; Kittelson, D. B.; Ahmadi, M. R.; Morris, J. E.Exhaust particulate emissions from a direct injection spark ignitionengine. Soc. Automot. Eng. Tech. Pap. Ser. 1999, 350−371.(11) He, X.; Ratcliff, M. A.; Zigler, B. T. Effect of gasoline directinjection engine operating parameters on particle number emissions.Energy Fuels 2012, 26, 2014−2027.(12) State of California Air Resources Board. Preliminary discussionpaper − Proposed amendments to California’s low-emission vehicleregulations − Particulate matter mass, ultrafine solid particle number,and black carbon emissions, 2010.(13) Zhong, D.; He, S.; Tandon, P.; Moreno, M.; Boger, T.Measurement and prediction of filtration efficiency evolution of sootloaded diesel particulate filters. Soc. Automot. Eng. Tech. Pap. Ser. 2012,2012-01-0363.(14) Tandon, P.; Heibel, A.; Whitmore, J.; Kekre, N.; Chithapragada,K. Measurement and prediction of filtration efficiency evolution of sootloaded diesel particulate filters. Chem. Eng. Sci. 2010, 65, 4751−4760.(15) Hansen, A. D. A.; Rosen, H.; Novakov, T. The aethalometer An instrument for the real-time measurement of optical absorption byaerosol particles. Sci. Total Environ. 1984, 36, 191−196.(16) Ferrero, L.; Mocnik, G.; Ferrini, B. S.; Perrone, M. G.; Sangiorgi,G.; Bolzacchini, E. Vertical profiles of aerosol absorption coefficientfrommicro-aethalometer data and Mie calculation over Milan. Sci. TotalEnviron. 2011, 409, 2824−2837.(17) Weingartner, E.; Saathoff, H.; Schnaiter, M.; Streit, N.; Bitnar, B.;Baltensperger, U. Absorption of light by soot particles: Determination ofthe absorption coefficient by means of aethalometers. J. Aerosol Sci.2003, 34, 1445−1463.(18) Collaud Coen, M.; Weingartner, E.; Apituley, A.; Ceburnis, D.;Fierz-Schmidhauser, R.; Flentje, H.; Henzing, J. S.; Jennings, S. G.;Moerman, M.; Petzold, A.; Schmid, O.; Baltensperger, U. Minimizinglight absorption measurement artifacts of the aethalometer: Evaluationof five correction algorithms. Atmos. Meas. Tech. 2010, 3, 457−474.(19) Hagler, G. S. W.; Yelverton, T. L. B.; Vedantham, R.; Hansen, A.D. A.; Turner, J. R. Post-processing method to reduce noise whilepreserving high time resolution in aethalometer real-time black carbondata. Aerosol Air Qual. Res. 2011, 11, 539−546.(20) Kamboures, M. A.; Hu, S.; Yu, Y.; Sandoval, J.; Rieger, P.; Huang,S. M.; Zhang, S.; Dzhema, I.; Huo, D.; Ayala, A.; Chang, M. C. O. Blackcarbon emissions in gasoline vehicle exhaust: A measurement andinstrument comparison. J. Air Waste Manage. Assoc. 2013, 63 (8), 886−901.(21) Harris, S. J.; Maricq, M. M. Signature size distributions for dieseland gasoline engine exhaust particulate matter. J. Aerosol Sci. 2001, 32,749−764.(22) Maricq, M. M.; Podsiadlik, D. H.; Chase, R. E. Gasoline vehicleparticle size distributions: Comparison of steady state, FTP, and US06measurements. Environ. Sci. Technol. 1999, 33, 2007−2015.(23) Kittelson, D. B.; Watt, W. F.; Johnson, J. P.; Schauer, J. J.; Lawson,D. R. On-road and laboratory evaluation of combustion aerosols-part 2:Summary of spark ignition engine results. J. Aerosol Sci. 2006, 37, 931−949.(24) Westbrook, C. K.; Pitz, W.; Curran, H. J. Chemical kineticmodeling study of the effects of oxygenated hydrocarbons on sootemissions from diesel engines. J. Phys. Chem. A 2006, 110, 6912−6922.(25) Golea, D.; Rezgui, Y.; Guemini, M.; Hamdane, S. Reduction ofPAH and soot precursors in benzene flames by addition of ethanol. J.Phys. Chem. A 2012, 116, 3625−3642.(26) Maricq, M. M. Soot formation in ethanol/gasoline fuel blenddiffusion flames. Combust. Flame 2012, 159, 170−180.(27) McNesby, K. L.; Miziolek, A. W.; Nguyen, T.; Delucia, F. C.;Skaggs, R. R.; Litzinger, T. A. Experimental and computational studies ofoxidizer and fuel side addition of ethanol to opposed flow air/ethyleneflames. Combust. Flame 2005, 142, 413−427.(28)Wu, J.; Song, K. H.; Litzinger, T.; Lee, S. Y.; Santoro, R.; Linevsky,M.; Colket, M.; Liscinsky, D. Reduction of PAH and soot in premixed

ethylene-air flames by additional of ethanol. Combust. Flame 2006, 144,675−687.(29) Hsieh, W. D.; Chen, R. H.; Wu, T. L.; Lin, T. H. Engineperformance and pollutant emission of an SI engine using ethanol-gasoline blended fuels. Atmos. Environ. 2002, 36, 403−410.(30) Poulopoulos, S. G.; Samaras, D. P.; Philippopoulos, C. J.Regulated and unregulated emissions from an internal combustionengine operating on ethanol-containing fuels. Atmos. Environ. 2001, 35,4399−4406.(31) Chen, L.; Braisher, M.; Crossley, A.; Stone, R.; Richardson, D.The influence of ethanol blends on particulate matter emissions fromgasoline direct injection engines. Soc. Automot. Eng. Tech. Pap. Ser. 2010,2010-01-0793.(32) Andrews, G. E.; Zhu, G.; Li, H.; Simpson, A.; Wylie, J. A.; Bell, M.;Tate, J. The effect of ambient temperature on cold start urban trafficemissions for a real world SI car. Soc. Automot. Eng. Tech. Pap. Ser. 2004,2004-01-2903.(33) Giechaskiel, B.; Mamakos, A.; Andersson, J.; Dilara, P.; Martini,G.; Schindler, W.; Bergmann, A. Measurement of automotivenonvolatile particle number emissions within the European legislativeframework: A review. Aerosol Sci. Technol. 2012, 46, 719−749.(34) Maricq, M. M.; Szente, J.; Loos, M.; Vogt, R. Motor vehicle PMemissions measurement at LEV III levels. SAE Int. J. Engines 2011, 4,587−609.(35) Maricq, M. M.; Szente, J. J.; Adams, J.; Tennison, P.; Rumpsa, T.Influence of mileage accumulation on the particle mass and numberemissions of two gasoline direct injection vehicles. Environ. Sci. Technol.2013, 47, 11890−11896.(36) United States Environmental Protection Agency. Accessing theEffect of Five Gasoline Properties on Exhaust Emissions from Light-DutyVehicles Certified to Tier 2 Standards: Analysis of Data from EPAct Phase 3(EPAct/V2/E-89), EPA report EPA-420-R-13-002; U.S. EPA: Wash-ington, DC, 2013.(37) Stradling, R.; Antunez Martel, F. J.; Ariztegui, J.; Beeckmann, J.;Bjordal, S. D.; Blosser, P.; Canovas, J.; Clark, A.; Elliott, N.; Farenback-Brateman, J.; Gomez-Acebo, P.; Martinez Sanchez, P. M.; Scorletti, P.;McArragher, J. S.; Zemroch, P. J.; Rose, K. D. Volatility and VehicleDriveability Performance of Ethanol/Gasoline Blends: A Literature Review,CONCAWE Report 8/09; CONCAWE: Brussels, Belgium, 2009;http://www.biofuelstp.eu/viewreport.php?viewid=75.(38) Barone, T. L.; Storey, J. M. E.; Youngquist, A. D.; Szybist, J. P. Ananalysis of direct-injection spark-ignition (DISI) soot morphology.Atmos. Environ. 2012, 49, 268−274.(39) Storey, J. M. E.; Barone, T. L.; Norman, K. M.; Lewis, S. A., Sr.Ethanol blend effects on direct injection spark-ignition gasoline vehicleparticulate matter emissions. Soc. Automot. Eng. Tech. Pap. Ser. 2010,2010-01-2129.(40) Johnson, T. V. Diesel emission control in review. Soc. Automot.Eng. Tech. Pap. Ser. 2009, 2009-01-0121.(41) Barone, T. L.; Storey, J. M. E.; Domingo, N. An analysis of field-aged diesel particulate filter performance: Particle emissions before,during, and after regeneration. J. Air Waste Manage. Assoc. 2010, 60,968−976.(42) Khalek, I. A.; Bougher, T.; Jetter, J. J. Particle emissions from a2009 gasoline direction injection engine using different commerciallyavailable fuels. SAE Int. J. Fuel Lubr. 2010, 3 (2), 623−637.(43) Orbital Engine Company. Environment Australia: A LiteratureReview Based Assessment on the Impacts of a 20% Ethanol Gasoline FuelBlend on the Australian Vehicle Fleet, Environment Australia Report;2002; http://www.environment.gov.au/archive/fuelquality/publications/review-vehicle-fleet/index.html.(44) Adler, J.; Harvey, C. A. Vehicle Driveability with Gasoline/AlcoholBlends, EPA Technical report EPA-AA-TSS-PA-86-02; U.S. EPA:Washington, DC, 2002.

Environmental Science & Technology Article

dx.doi.org/10.1021/es501791b | Environ. Sci. Technol. 2014, 48, 6027−60346034