Past and Future Issues in Exhaust Particle Measurement and Control David B. Kittelson Center for Diesel Research University of Minnesota Aristotle University Thessaloniki, Greece 23 July 2003

Acknowledgements We have had help from many collaborators –In the Center for Diesel Research »Feng Cao, Marcus Drayton, Jason Johnson, Hee Jung Jung, Duane Paulsen, Winthrop Watts, Robert Waytulonis, Qiang Wei, Darrick Zarling –In the Particle Technology Lab »Peter McMurry, Kihong Park, Hiromu Sakurai –At UC Riverside »Herbert Tobias, Paul Ziemann –At Paul Scherrer Institute »Nick Bukowiecki, Urs Baltensperger And many sponsors –Coordinating Research Council, U.S. Office of Heavy Vehicle Technologies, Engine Manufacturers Association, Southcoast Air Quality Management District, California Air Resources Board, Cummins, Caterpillar, Perkins, Volkswagen, and Volvo.

In recent years many concerns have been raised about extremely tiny particles, smaller than 50 to 100 nm in diameter, that are emitted by engines and other combustion sources. This presentation considers issues associated with the formation and measurement of these particles from current and future low emission Diesel engines

Outline Introduction On-road measurements Structure and Composition How they form Formation downstream of aftertreatment devices Alternatives to mass measurements Conclusions

Outline Introduction On-road measurements Structure and Composition How they form Formation downstream of aftertreatment devices Alternatives to mass measurements Conclusions

Typical Diesel Particle Size Distributions, Number, Surface Area, and Mass Weightings Are Shown

Heavy-Duty Engine Emission Standards It was thought that the 1994 standards would lead to the use of exhaust filters - but high performance electronically controlled engines met standards without them The combination of 90% PM reduction and 95% NOx reduction in 2007 makes aftertreatment virtually certain Recent concern about global warming by elemental carbon particles makes particle control by filtration even more attractive

Emissions of Ultrafine and Nanoparticles from Engines Current emission standards are based on filter mass. Recently interest in other measures, i.e, size, number, surface, has increased. Concerns about particle size –New ambient standards on fine particles –Special concerns about ultrafine and nanoparticles – number or surface area may be a better measure of biological response than mass –Indications that reductions in mass emissions sometimes increase number emissions Difficulties associated with measurement of ultrafine and nanoparticles –Typically, with current engines more than 90% of particle number and more than 35% of particle mass are formed from volatile precursors during exhaust dilution. –These fractions will be even higher with future ultra low emission engines –Particle formation during sampling and dilution is highly nonlinear and extremely sensitive to conditions

Background - Historical Perspective A HEI study published in 1996 showed dramatic increase in nanoparticle emissions for a prototype 1991 Diesel engine compared to 1988 engine –This raised concerns that low mass emission engines might increase number emissions –It was suggested that these emissions were tiny carbon particles that might constitute a new health threat Further investigation has alleviated some of those concerns –The recent CRC E-43 study (2002) showed for typical modern Diesel engines »Nanoparticles mostly volatile rather than solid »Both reduced mass and number emissions are reduced In any case, high nanoparticle emissions are not a new development! –Roadside measurements made starting in the 60’s nearly always show a large nuclei (nanoparticle) mode –Both Diesel and spark ignition sources

Outline Introduction On-road measurements Structure and Composition How they form Formation downstream of aftertreatment devices Alternatives to mass measurements Conclusions

U of M Mobile Laboratory built to study formation of nanoparticles in the atmosphere for the CRC E-43 project Instruments (primary instruments highlighted in blue) –SMPS to size particles in 9 to 300 nm size range –ELPI to size particles in 30 to 2500 nm size range –CPC to count all particles larger than 3 nm –Diffusion Charger to measure total submicron particle surface area –Epiphaniometer to measure total submicron particle surface area –PAS to measure total submicron surface bound PAH equivalent –CO 2, CO, and NO analyzers for gas and dilution ratio determinations

During the DOE / CRC E-43 program particles from modern Diesel engines were measured on road and in the laboratory Here we have a 1999 highway tractor with its plume being characterized by the University of Minnesota Mobile Emission Laboratory

Comparison with previous studies: Nuclei mode particles from newer engines are at lower concentrations and somewhat smaller in diameter Corrected for dilution ratio Not corrected for particle losses All data except HEI are for standard on-highway EPA/Federal fuels. HEI fuel lower S ~ 100 ppm

Nuclei Mode Particles Exist in Large Concentrations over Roadways Measurements made with University of Minnesota Mobile Laboratory Large concentrations observed in presence of both Diesel and spark ignition traffic The nuclei mode forms rapidly in the diluting plumes of vehicles – time scale seconds Nuclei mode decays rapidly downwind of roadway – time scale minutes The focus here is on nucleation in diluting exhaust plumes, nucleation also occurs in the atmosphere as a result formation of low vapor species by atmospheric chemistry but usually on a longer time scale

Typical Roadway Number Concentrations from Several Instruments and Road Speed Dp > 3 nm Dp > 29 nm Dp > 9 nm

On-highway measurements made on urban freeways in Minnesota show a large nuclei mode even in the absence of significant Diesel traffic Traffic speed has at least as much influence on the size of the nuclei mode as the presence of Diesel traffic Particle number increases and size decreases as traffic speed increases Particle volume (mass) is higher under low speed congested conditions It appears that slow moving congested traffic leads to storage of volatile materials in the exhaust system As vehicles speed up the exhaust system heats leading to the release of the materials which subsequently form nanoparticles Diesel No Diesel

Nuclei Mode Decays Rapidly Downwind of Roadways Modeling (Capaldo and Pandis, 2001) indicates –For typical urban conditions, characteristic times and transit distances for 90 % reduction of ultrafine concentrations are on the order of a few minutes and m, respectively. –For a given wind speed, ultrafine particles are expected to survive and travel a factor of ten greater distances in a rural flat area as compared to an urban downtown location. Mobile particle sources will influence the aerosol particle number concentrations mainly near roadways.

Outline Introduction On-road measurements Structure and Composition How they form Formation downstream of aftertreatment devices Alternatives to mass measurements Conclusions

Some of the instruments used to determine composition and structure of Diesel particles Particle Technology Laboratory Nano MOUDI

TDPBMS measures the volatility and mass spectra of the volatile fraction of all the particles in selected size ranges between 15 and 300 nm - Summary Results Engines –Deere 4045T medium-duty –Caterpillar C12 heavy-duty –Cummins ISM Fuels –Federal pump fuel, 360 ppm S –California pump fuels, 50 and 96 ppm S –Fischer-Tropsch, < 1 ppm S Test conditions –Light and medium load Composition of volatile fraction –Organic component of total diesel particles and nanoparticles appears to be mainly unburned lubricating oil –Major organic compound classes are alkanes, cycloalkanes, and aromatics –Low-volatility oxidation products and PAHs have been found in previous GC-MS analyses, but are only a minor component of the organic mass –Nanoparticles formed with higher S Federal pump fuel contain small amounts of sulfuric acid but those formed with the lower S fuels show no evidence for sulfuric acid

OC/EC analysis of nano-MOUDI Samples, gives results generally consistent with TDPBMS Analysis performed by Barbara Zielinska, Desert Research Institute There is nearly no EC in nuclei mode The nuclei mode is smaller with lower S fuel, but it is still nearly all OC

Typical Composition and Structure of Diesel Particulate Matter Solid particles are typically carbonaceous chain agglomerates and ash and usually comprise most of the particle mass Volatile or semi-volatile matter (sulfur compounds and organics (SOF)) typically constitutes 35% (5-90%) of the particle mass, 90% (30-99%) of the particle number Carbon and sulfur compounds derive mainly from fuel SOF and ash derive mainly from oil Most of the volatile and semi- volatile materials undergo gas-to- particle conversion as exhaust cools and dilutes

Outline Introduction On-road measurements Structure and Composition How they form Formation downstream of aftertreatment devices Alternatives to mass measurements Conclusions

Nature of particles emitted by Diesel engines Diesel engines produce a bimodal size distribution in the submicron range With current engines –The nuclei mode is in the 3-30 nm diameter range and contains most of the particle number –The accumulation mode is in the nm range and contains most of the particle mass These modes form by different mechanisms The nuclei mode may contain the majority of both the mass and the number emitted by future technology engines

A dilution ratio of 1000 may be reached in s Atmospheric dilution leads to nucleation, absorption, and adsorption - in excess of 90 % of the particle number may form as the exhaust dilutes

Particle formation history – most volatile nanoparticles form during dilution University of Minnesota This is where most of the volatile nanoparticles emitted by engines usually form. There is potential to form solid nanoparticles here if the ratio of ash to carbon is high. Particles formed by Diesel combustion carry a strong bipolar charge

Nuclei mode Nuclei mode particles form mainly from volatile precursors –The nuclei mode typically consists mainly of heavy hydrocarbons, mainly from lubricating oil, and sulfates –Although although the mode is mainly hydrocarbon, its formation is facilitated by sulfur in the fuel –Its formation is very dependent on dilution conditions, especially dilution rate and dilution air temperature –Its formation is favored by low solid carbon and high precursor concentration –It may be all that is left when carbon is removed by exhaust filtration Solid nuclei mode particles may form from metals in the lube oil or fuel –Formed from oil under engine conditions that lead to little solid carbon formation. –Formed from fuel when metallic additives or high metal fuels are used.

Accumulation mode The accumulation mode is where most “soot” or “smoke” resides It consists primarily of carbonaceous agglomerates and adsorbed hydrocarbons Particles in this mode are strongly light absorbing Most of the lubricating oil ash is found in this mode The density of accumulation mode particles decreases with increasing size –Fractal like behavior –Low densities cause size to be underestimated by some methods Accumulation mode particles carry a significant bipolar electrical charge that is produced during the basic combustion process –Offers potential for particle sensing –Could influence filtration Accumulation mode particles are not strongly influenced by dilution conditions Accumulation mode particles have been reduced sharply by better engine technology

Studies of Diesel Nanoparticle Formation Using a Variable Residence Time Dilution System

Influence of Residence Time and Temperature on Number Weighted Size Distributions Medium-duty Diesel engine running at medium speed and load Increasing residence time in the primary dilution chamber from 230 ms to 1 s increases the size of the nuclei mode by two orders of magnitude Decreasing the temperature in the primary dilution chamber from 66 to 32 °C increases the size of the nuclei mode by about one and one half orders of magnitude

Influence of residence time and temperature on total number concentrations – nearly all in nuclei mode Heavy-Duty Diesel Medium Load and Speed Primary Dilution Ratio = 12

A 2-stage, porous tube/ejector dilutor could simulate on-road nuclei mode formation under summer highway conditions Results shown are composite of loaded and unloaded highway cruise for a modern heavy-duty Diesel engine with full electronic engine management The relative sizes of the two modes are more significant than the absolute levels – due to uncertainty in on-road dilution ratios In general the CA fuel produced a smaller nuclei mode than EPA fuel These results show that it is possible to simulate carefully defined on-road conditions At present, it is unclear which on-road conditions should be simulated. There are many variables including –Temperature –Previous operating history –Road speed –Exhaust system design –Others …. Lab On-road Lab On-road EPA fuel CA fuel

Outline Introduction On-road measurements Structure and Composition How they form Formation downstream of aftertreatment devices Alternatives to mass measurements Conclusions

Influence of an uncatalyzed exhaust filter on particle size and concentration Filters remove accumulation mode particles very effectively Nuclei mode particles may form downstream of filters, especially at longer residence times These particles form from volatile materials that are in the gas phase in the filter and are thus not collected Peak power speed 25% load Rated Torque

Formation of a volatile nuclei mode downstream of a catalyzed Diesel particulate filter (DPF) The results shown are for tests of a catalyzed DPF on a modern Cummins engine DPF very efficient for solid particles May form a very large nuclei mode downstream of DPF at high load conditions Size of mode increases with sulfur content of fuel, but still observed with near zero sulfur fuel Sulfuric acid likely major component 26 ppm S fuel 1 ppm S fuel

Filter performance is best evaluated under sampling conditions that minimize nuclei mode formation Results shown are for a VW TDI with an uncatalyzed DPF Here we are investigating how trap loading influences filtration performance Concentrations are shown on the left, penetrations on the right Increasing time

On road characterization of aftertreatment devices– we sniff our own exhaust plume Driver side sample point Passenger side sample point Background sample point Stacks Plume sampling points on MEL

On road tests of particle filtration device All tests done with 15 ppm S fuel (post 2006) with standard or low sulfur lubricant Engine out with low S oil

On road tests of particle filtration device Engine out with low S oil Filter with low S oil, August All tests done with 15 ppm S fuel (post 2006) with standard or low sulfur lubricant

On road tests of particle filtration device Engine out with low S oil Filter with standard oil, August Filter with low S oil, August All tests done with 15 ppm S fuel (post 2006) with standard or low sulfur lubricant

On road tests of particle filtration device Engine out with low S oil Filter with standard oil, August Filter with standard oil, October Filter with low S oil, August All tests done with 15 ppm S fuel (post 2006) with standard or low sulfur lubricant

On road tests of particle filtration device – volume (mass) distributions Engine out with low S oil Filter with standard oil, August Filter with standard oil, October Filter with low S oil, August All tests done with 15 ppm S fuel (post 2006) with standard or low sulfur lubricant

Particles from future engines Solid particles in both the nuclei and accumulation modes may be nearly completely eliminated by filtration Filters cannot directly remove the gas phase precursors that lead to the formation of nuclei mode particles –Precursors may be removed by adsorption on collected particles followed by.. –Hydrocarbon precursors may be destroyed in catalyzed systems to the extent allowed by kinetics (mainly mass transfer) but sulfuric acid may be formed If filters remove nearly all of the solid particles the only thing left will be volatile particles in the nuclei mode, with all the sampling problems associated with this mode

Outline Introduction On-road measurements Structure and Composition How they form Formation downstream of aftertreatment devices Alternatives to mass measurements Conclusions

Filter mass measurements with low emission engines - issues Very difficult to accurately measure filter mass at low levels For volatile particles, filter mass does not adequately represent suspended mass (what we breathe) –DPF performance measured with filters is very different from that measured with instruments that measure suspended particles (VERT (Switzerland), SWRI,..) –Absolute suspended mass measurements made with a new instrument developed at of U of M (Park, et al., 2003) show that filters may significantly overstate the mass of volatile particles –Sampling and dilution may strongly influence results Other measures may better predict environmental impact –Number –Surface area –Black carbon The European Union is expected to institute a solid particle number or surface area standard to supplement filter mass standard Fast response, low cost instruments needed for engine diagnostics, conventional smokemeters inadequate

An instrument suite that would offer an attractive alternative to filter mass A sampling and dilution system that adequately simulates atmospheric dilution conditions A device to remove volatile particles Condensation particle counter An active surface area instrument A black carbon instrument For more fundamental work one of the new fast response sizing instruments

Outline Introduction On-road measurements Structure and Composition How they form Formation downstream of aftertreatment devices Alternatives to mass measurements Conclusions

The particulate mass emitted by future Diesel engines will be very low and difficult to measure accurately. –This would be true even if all the material were solid –Unfortunately much of the material will consist of volatile nuclei mode particles that form during exhaust dilution –Volatile particles sampled on a filter may be very different from those that we breathe Instruments that directly measure suspended particles rather than collecting them on a filter avoid these difficulties

Future Issues Both the fuel and lubricating oil properties will play an important role in the performance of aftertreament devices –Sulfur and phosphorus –Metallic ash Many low emission engine concepts without traps have potential to for significant nanoparticle emissions mainly associated with the lubricating oil –Spark ignition engines – especially older worn engines –Gaseous fuel engines –Homogeneous charge compression ignition engines (HCCI) Recent measurements on a Boeing 757 gave number emissions in the same range or higher than current Diesel engines Certain roadways, on ramps, airports, etc. are likely nanoparticle “hotspots”

Alternatives to filter mass – dilution and sampling, sample conditioning System used for dilution and sampling of exhaust must adequately simulate nuclei mode formation under representative atmospheric conditions The system must be designed to minimize particle losses that may be very large for sub 10 nm particles A system to differentiate between solid and volatile particles should be included –Thermal denuder, Dekati, Matter –Catalytic Stripper, U of M, SWRI

Alternatives to filter mass - particle sizing of suspended particles Current –Scanning Mobility Particle Sizer, size range 3 to 700 nm, TSI »Commonly used »Not suitable for transient measurements –Electrical Low Pressure Impactor, size range ~10 to 2500 nm, Dekati »Widely used in Europe, Japan »Poor response to nuclei mode New –Electrical Diffusion Battery, size range ~10 to 200 nm, Matter »Fast response »Poor size resolution –Differential Mobility Spectrometer, size range 5 to 500 nm, Cambustion »Fast response, < 1 s »Expensive –Real Time Mobility Particle Sizer, size range 5 to 500 nm, new from TSI »Fast response, < 1 s »Expensive

Alternatives to filter mass - fast response integral measures of suspended particles Total number –Condensation Particle Counter, size range > 3 to 11 nm, TSI »Time resolution, 1 – 13 s »Extremely sensitive Active surface area –Diffusion Charger, size range > 10 nm, (Matter, Dekati) »Inexpensive »Dp 1.4 »Time resolution, 1 s –Electrical Aerosol Detector (TSI) »Inexpensive »Dp 1.1 »Time resolution, 1 s

Alternatives to filter mass - fast response integral measures Black carbon mass –Laser induced incandescence, Sandia »Expensive »Time resolution, < ms –Photo acoustic, DRI »Expensive »Time resolution, ~1 s –Aethalometer, Magee Scientific »Particles collected on filter, not an issue for solid particles »Inexpensive »Time resolution, ~1 s – 1 min Total mass (particles collected subject to artifacts similar to filters) –Tapered Element Oscillating Microbalance, Rupprecht & Patashnick »Time resolution, ~1 s –Quartz Crystal Microbalance, Booker »Time resolution, ~1 s