Chapter 15

Particle Emissions from Vehicles

Jonathan Symonds

Cambustion, UK

15.1 Introduction

Aerosols from vehicle engines originate from five sources: fuel, fuel additives, inlet air, lubrication oil and the mechanical breakdown of preexisting materials. The latter can also form from other sources in the vehicle (e.g., brake dust). Of those formed in the engine, there are four main types of aerosol: carbonaceous, organic, sulfate and ash. These usually appear in combination. Fuel and oil contribute to all four fractions; fuel additives, air and mechanical breakdown contribute to the ash fraction. A typical aerosol from a heavy-duty diesel engine is 41% carbon, 13% ash, 14% sulfate/water, 25% unburnt oil and 7% unburnt fuel (Kittelson, 1998).

The number-based size spectrum (Figure 15.1) of an engine aerosol normally includes some of at least three distinct, lognormal modes (Kittelson, 1998). Homogenous nucleation of volatile materials in the exhaust (or ash particles) can form the so-called nucleation mode, which is usually smaller than 30 nm in size, with a narrow geometric standard deviation (<1.5). After this comes the accumulation mode, which is normally between 60 and 200 nm, with a general standard deviation (GSD) between 1.5 and 2.0. This is where the carbonaceous (‘soot’) agglomerate particles are usually found. Particles larger than this are referred to as the coarse mode and consist mostly of material produced outside the engine, such as brake dust and reentrained soot from the walls of the exhaust system. On a particle number basis, the nucleation- and accumulation-mode particles vastly dominate, but when weighted by mass, the coarse mode can contribute around 20% of the total particulate mass.

15.2 Engine Concepts and Technologies

15.2.1 Air–Fuel Mixture

As we shall see, one of the most important parameters affecting particle formation during combustion is the air–fuel ratio (AFR; e.g. Heywood, 1988, Section 3.3). This is expressed in terms of the mass fraction of air to fuel. If exactly enough air is present to burn all the fuel present then the mixture is called stoichiometric. If too much fuel is present compared with the air, it is called a rich mixture, whereas if too little fuel is present, it is known as a lean mixture. The ratio of the overall AFR to the stoichiometric AFR is represented by λ. Hence, for rich mixtures. λ <1, and for lean mixtures, λ >1.

The ‘ideal’ combustion of a stoichiometric or lean mixture of oxygen and a simple hydrocarbon fuel may be represented as:

If a rich mixture is burnt then the lack of oxygen leads, in the first instance, to the production of carbon monoxide, and then to the production of carbonaceous soot. In reality, all engine combustion processes generate small but finite quantities of CO, NO, NO₂, unburnt fuel (uHC) and particulate matter (PM).

Due to the short time scales involved in combustion compared with the vaporisation of fuel and the mixing of fuel vapour and air, the localised AFR in parts of the cylinder can be quite different from the overall AFR. Depending on the method of introduction of fuel and air into the cylinder, locally rich regions can lead to soot formation (Greeves and Wang, 1982).

If fully vaporised fuel and air are well mixed, the combustion occurs by way of a premixed flame (e.g. typical gasoline engine combustion). If, however, the combustion process is dominated by the rate at which fuel and air mix, then combustion will proceed by way of a diffusion flame (e.g. typical diesel engine combustion).

15.2.2 Spark-Ignition Engines

A spark-ignition (SI) engine relies upon an electrical discharge to ignite a fuel–air mixture. SI engines are usually fuelled by gasoline, though increasingly liquefied petroleum gas (LPG), compressed natural gas (CNG), ethanol, bioethanol or another biofuel, often mixed with gasoline, is used instead. The fuel and air may be premixed in the intake system of the engine, as in carburetted or port fuel-injected (PFI) engines, or the fuel may mix with the air in-cylinder, after being injected directly into it. SI engines use either a four-stroke (intake, compression, power, exhaust) or two-stroke (intake–compression, power–exhaust) working cycle.

The engine management system in a PFI engine will attempt (under closed-loop control) to keep the mixture near stoichiometric, which for gasoline means an AFR of approximately 14.5. Though significantly better fuel economy is achieved with slightly lean mixtures, the simultaneous oxidation of CO and uHCs and reduction of NO_x requires the use of both a stoichiometric mixture and a ‘three-way’ catalyst. Closed-loop control of the fuelling to achieve stoichiometry is achieved by the use of oxygen sensors in the exhaust gas. The stoichiometric and well-mixed fuel/air conditions of PFI gasoline engines also mean that the carbonaceous particle emissions from the fuel in these engines are small.

In recent years, gasoline direct injection (GDI) engines have increased in popularity. In these, gasoline is directly injected into the cylinder, rather than into the intake port. This method of fuelling has a number of advantages. In a PFI engine, the flow through the inlet valve consists of air and fuel vapour. If fully vaporised, about 5% by volume of the flow is fuel vapour, which represents a 5% reduction of the air flow into the engine – and a 5% reduction in the maximum power. In addition, when the fuel evaporates in the cylinder, the cylinder charge is cooler than it is when the evaporation takes place in the port. This allows the engine to be run with a higher compression ratio and hence a higher efficiency. Finally, there are benefits associated with better AFR control, since in a PFI engine some of the fuel entering the cylinder is from previous injections, in which it was deposited on the manifold wall.

In a stoichiometric GDI engine, fuel is normally injected immediately after the exhaust valve closes, which maximises the time available for mixing. Some GDI engines operate, for part of the operating envelope, in stratified mode, and fuel is injected in the latter stages of compression. In principle, this allows a portion of the cylinder to contain a flammable mixture in the region of the spark plug, surrounded by air. If such a mode were ‘perfect’, the engine could be run without throttling, as the power output would be modulated by the relative volumes of the region containing the flammable mixture and the air. Sadly, the high turbulence levels in the cylinder lead to significant mixing, and stratified combustion has not been widely adopted.

In general, GDI engines suffer from higher PM emissions that do PFI engines (e.g. Price et al., 2006). This is due to their reduced air/fuel homogeneity and to impingement of fuel on the piston and cylinder surfaces. These higher emissions cause liquid fuel pools to be formed, which eventually burn by diffusion, again causing higher carbonaceous soot emission (Witze and Green, 2005).

GDI engines are becoming much more common due to their increased efficiency and reduced CO₂ emission, particularly in the USA and Japan, where light-duty diesel vehicles offering similar advantages have not been as widely adopted as they have in Europe. However, the increased particulate emissions are a drawback, and the regulation and reduction of these emissions is a current area of intense research.

15.2.3 Compression-Ignition Engines

Compression-ignition (CI) engines are usually referred to as ‘diesel engines’, after the inventor Rudolf Diesel. Diesel engines have a high thermal efficiency, due to their high compression ratio (of around 15 : 1 to 22 : 1). The compression ratio of gasoline engines is limited by the onset of ‘knocking’; that is, autoignition of part of the mixture late in the burning process. In most diesel engines, the fuel is injected directly into a bowl formed in the piston, near the end of the compression stroke. Ignition then occurs spontaneously in the mixture formed by evaporating fuel and air, due to the temperature generated during the compression stroke (diesel fuel is formulated to have a low autoignition temperature compared to gasoline). Although the use of ever-higher injection pressures leads to enhanced fuel–air mixing, rich combustion inevitably occurs in the fuel vapour plume as it mixes with air. Depending on the crank angle and the location within the cylinder, λ values between 0 and infinity exist, unlike in a stoichiometric PFI gasoline engine, where λ is unity, or nearly so, everywhere. Inevitably, significant PM is formed, and though much of it is oxidised later in the combustion process, significant quantities leave through the exhaust valves. The levels of carbonaceous particle emissions from individual diesel engines are also much higher than those from individual GDI engines.

15.2.4 Two-Stroke Engines

A two-stroke engine completes the combustion cycle in one revolution of the crankshaft. Two-stroke engines are commonly used in motorcycles and horticultural equipment when a high power density is required. In their simplest form, an air, fuel and lubrication oil mixture enters the combustion chamber via the crankcase and a port in the cylinder wall linking the under and top sides of the piston. As the piston rises, the port is covered; when it reaches the top, ignition via a spark plug occurs. As the piston descends, a second port in the wall of the cylinder is uncovered and the exhaust gases exit. The particulate emission levels from two-stroke engines can be high, due to the lubrication oil present in the mixture and to short-circuiting of the fuel–air mixture from the intake port directly to the exhaust port; indeed, visible smoke is often produced. Fuel short-circuiting can be reduced by modern direct-injection technology. The particle size spectrum from two-stroke engines can contain some quite large particles (between 500 and 1000 nm), thought to be condensed lubrication oil, in addition to the nucleation and accumulation modes (Hands et al., 2010). Two-stroke scooters are particularly popular in emerging economies, and particulate emissions from these are currently unregulated.

15.2.5 Gas-Turbine Engines

Particles emitted from jet engines are of particular concern, as they are emitted directly into the upper atmosphere and lower troposphere, where they can directly affect climate forcing (Fortuin et al., 1995). The soot particles produced by these engines tend to be smaller than those produced by diesel engines, usually sub-60 nm (Petzold et al., 2011), and primary spherical soot particles are often directly detected in their unagglomerated form (Schmid et al., 2011). In addition to carbonaceous soot, organic matter and hydrated sulfate are also seen, which form nucleation aerosols in the engine wake. Ground-based gas turbines (for power generation, propulsion, etc.) also produce PM. Unlike piston engines, particle filtration on these devices is not practicable.

15.3 Particle Formation

15.3.1 In-Cylinder Formation

The combustion process converts hydrocarbon fuel molecules, each of which has just a few carbon atoms, into soot particles, which contain many thousands of carbon atoms (Figure 15.2). The initial complex and varied collection of reactions that occurs during fuel combustion with limited oxygen is known as pyrolysis. The first stage sees the formation of poly-aromatic hydrocarbon (PAH) molecules from the fuel hydrocarbon molecules (e.g. Eastwood, 2008, Chapter 3). These PAH molecules, initially in the form of vapour, then undergo a nucleation process to form nuclei particles of less than 3 nm in size. These nuclei particles then grow through the addition of carbon until primary soot particles emerge at between 20 and 50 nm. Electron microscopy shows that these primary particles are spherical. The primary particles then agglomerate together, forming fractal soot agglomerate particles. During all of these processes, a competing oxidation process removes carbon, and final levels of carbon are as much a function of this removal process as they are of the creative processes (Pipho, Ambs and Kittelson, 1986).

During the initial premixed burn stage, the jet of liquid fuel starts to vaporise at its tip as a result of air entrainment (Dec, 1997). Ignition then takes place within the fuel–air mix at the end of the jet and the pyrolysis processes start the conversion of fuel into PAH. The hot nascent soot first appears in pockets and then spreads to the edge of the plume. The soot particles surrounded by the flame envelope at the edge of the plume grow rapidly compared with those within the plume. The process then enters the mixing controlled-combustion stage, until the end of the injection event. Fuel is partially burnt at the tip of the fuel jet, and the products of partial combustion proceed to the tip of the plume, where they enter the diffusion flame. The process of soot formation starts in the premixed flame, and the soot precursors grow as they head towards the plume head, where some oxidation then occurs in the diffusion flame. After fuel injection terminates, the plume collapses. The quality of the jet as the fuel pressure drops may become poor, thereby momentarily increasing soot output before the end of combustion.

The ash fraction arises from inorganic molecules present in the fuel, oil or fuel additives, along with a small amount entrained from mechanical wear of the engine. A certain amount of airborne dust may be present, though the air intake filter should limit this. Common constituents of the ash fraction are metal oxides, particularly oxides of calcium, magnesium and zinc, which arise from the lubrication oil, and iron oxides, which arise from corrosion of the engine materials (Jung, Kittelson and Zachariah, 2003). As oil enters the periphery of the combustion chamber via the piston rings or valve stems, it is not subjected to such intense combustion conditions as the fuel, and therefore the aerosol constituents arising from oil may be chemically quite close to the substances in the original oil.

Additives in fuel designed to improve engine operation (e.g. the anti-knock agent ferrocene) or to catalyse diesel particulate filter (DPF) regeneration (e.g. cerium; Section 15.6.3) can form ash particles in the exhaust (but in the latter case, since a DPF is fitted, these ash particles would be removed before release into the environment). The ash particles produced from fuel additives tend to be in the same size region as the ‘nucleation’ mode; that is, <50 nm (Gidney et al., 2010). Unlike the primary particle precursors of soot agglomerates (which are of a similar size), ash particles do not tend to form larger aggregates. Ash components are of low volatility; therefore, their conversion from the gas to the particulate phase occurs in the heat of the early formation processes, and certainly within the combustion chamber rather than in the exhaust system. Metal fuel additives can in fact have a soot-suppressing effect (Howard and Kausch, 1980). For example, ions formed during chemiionisation of additives in hydrocarbon flames can suppress soot agglomeration by means of mutual electrostatic repulsion.

The organic fraction usually results from unburnt fuel molecules, caused by locally rich regions in the cylinder (e.g., late vaporisation of fuel pools on the cylinder walls; Kato et al., 1997) or an over-mixed charge that is too weak to support combustion. The compounds present in the aerosol phase may be essentially chemically unchanged from their form in the fuel or else may undergo some chemical synthesis in the combustion chamber. Some may be the PAH soot precursor molecules, i.e. molecules that have not evolved as far as soot formation (Fujiwara, Tosaka and Murayama, 1993). Sufficient quantities of unburnt fuel can lead to white smoke formation at the tailpipe. These particles, unlike the nanoparticles produced by combustion, are large enough to be optically detectable. Less volatile hydrocarbons from lubrication oil that pass by the piston ring set are another source of organic aerosol; this is particularly true in two-stroke engines, where the intake air and oil are mixed.

In general, soot formation increases with engine load. As noted earlier, in diesel engines a higher load means increased rich combustion. Also, high load leads to higher combustion temperatures, which means that the processes of soot formation are fully completed. At lower loads, the processes may not complete, and the engine aerosol will consist mainly of nucleation-mode materials created from the organic compounds intermediate in the soot formation process. Furthermore, when soot is produced, material that might otherwise form a nucleation mode (including sulfate) is often absorbed on to the soot, thereby suppressing nucleation mode formation (Abdul-Khalek, Kittelson and Brear, 2000). This results in a characteristic (at times mutually exclusive) interplay of nucleation and accumulation modes as a function of load over a legislated drive cycle or real-world driving when examined with a fast-response particle sizer (e.g. Campbell et al., 2006).

Organic and sulfurous vapour particle precursors often do not enter the particulate phase until they reach the exhaust systems or even the air, where condensation nucleation occurs due to cooling and dilution.

15.3.2 Evolution in the Exhaust and Aftertreatment Systems

Both solid and volatile fractions of engine exhaust aerosol undergo further processes in the exhaust system. At high concentrations, the accumulation-mode soot particles undergo agglomeration, increasing in mean size and decreasing in number but maintaining the same overall mass. As the aerosol cools, the soot particles may adsorb volatile materials, forming a coating. The organic fraction can continue to undergo chemical reactions even in the exhaust systems; for example, PAH can continue to react down to 250 °C (Williams, Perez and Griffing, 1985).

The sulfate aerosol fraction consists primarily of sulfuric acid, as a condensation nucleation aerosol (Abdul-Khalek, Kittelson and Brear, 2000). Sulfur exists in both fuel and lubrication oil, although in many markets its content in fuel is significantly reduced. During combustion, the sulfur oxidises to form mostly sulfur dioxide, with some sulfur trioxide. The sulfuric acid forms from the hydrolysis of the SO₃, but as there is little of this present as a result of combustion itself, the biggest source of the acid is from the preconversion of SO₂ to SO₃, which usually occurs in the aftertreatment system:

Modern diesel vehicles are fitted with a diesel oxidation catalyst (DOC) in the exhaust system in order to ameliorate CO, hydrocarbon and soluble organic fraction (SOF) emissions. However, the DOC is also the main source of conversion of SO₂ to SO₃. The use of low-sulfur diesel mitigates the problem of sulfate formation.

15.3.3 Noncombustion Particle Sources

Sources of coarse-mode particles include: dust from brake linings, tyre wear, road-surface wear, engine wear and rust, dust and scale from the exhaust and catalyst system. Lubrication oil is another source of noncombustion aerosol, as well as of combustion aerosol. The fumes from an engine's crankcase contain fine particulate matter (including ‘blow-by’, which escapes from the combustion chamber via the piston ring set) and larger oil drops on the scale of many microns, whether the engine is being fired or even just rotated by means of the starter motor (Johnson, Hargrave and Reid, 2011). In the USA and Europe, crankcase fumes must either be eliminated by means of a closed crankcase ventilation system (CCV) or be vented to the exhaust system (in which case the total exhaust is still subject to relevant PM emission limits).

15.3.4 Evolution in the Environment

Nucleation of the organic and sulfate fractions into the aerosol phase usually takes place as the aerosol cools in the exhaust system or upon dilution at the end of the tailpipe. When the exhaust finally reaches the environment at the end of the tailpipe, two effects dominate: the exhaust is rapidly cooled and the partial pressure of chemical constituents in the gas phase is reduced. The cooling effect increases the saturation ratio, driving particle nucleation and growth, and dilution decreases the vapour pressure of the constituents, suppressing nucleation and growth and, indeed, eventually shrinking and evaporating volatile particles (Abdul-Khalek, Kittelson and Brear, 2000). Upon addition of a small amount of dilution, the cooling effect dominates (and hence so do nucleation and growth), but the exhaust temperature quickly reaches that of the air, so adding further dilution beyond this point just serves to reduce the concentration of the gas-phase constituents, leading particle shrinkage and evaporation to dominate.

In the near wake of a moving vehicle, the flow becomes complex, with turbulent mixing dominating (Carpentieri, Kumar and Robins, 2011). On longer length scales, further from the vehicle, the wakes from vehicles interact and the dispersion is dominated by the geometry of the locality. For example, street canyons, formed by high buildings on either side of a road, tend to allow levels of particulate to build up. While nucleation-mode particles may continue to mutate, the accumulation-mode particles can remain for up to a month at the concentrations present in the atmosphere, in the absence of rain (Kumar et al., 2010).

15.4 Impact of Vehicle Particle Emissions

15.4.1 Health and Environmental Effects

Acute exposure to diesel-engine aerosol causes irritation (in the eyes, throat and bronchial tubes) and respiratory problems (coughing, exacerbation of asthma) (US EPA, 2002). Chronic effects include likely carcinogenic activity (US EPA, 2002) and an increased risk of heart disease (Miller et al., 2007). The fact that pedestrians in urban environments are placed so close to vehicles and their exhausts, and that the slow speed of urban vehicles creates less dilution in the wake than is found at high speed, means that levels of exposure for pedestrians can easily exceed concentrations that can cause health effects (Buzzard, Clark and Guffey, 2009). Larger particles are deposited in the nose and upper airways, while the smallest nanoparticles can reach the alveolar region and even enter into the bloodstream.

Black carbon, including engine soot, absorbs light and thus acts as a positive radiative forcer and contributes to global warming (Ramanathan and Carmichael, 2008).

15.4.2 Legislation

In order to counter the health and environmental impacts of engine-sourced nanoparticles, most territories have introduced some sort of limit on the levels produced. Limits on particulate mass have been in force in the USA^a since the 1990s (for diesel vehicles) and in Europe since 2000 (Table 1). Limits differ for light- and heavy-duty vehicles, and there are numerous special categories such as low emission vehicle (LEV) standards. The method used to measure particulate mass is almost universally filter paper; see Section 15.5.2.

Table 1 European light-duty particulate emission standards

Vehicles being tested for emissions are set up for monitoring on a chassis dynamometer (a ‘rolling road’) and driven on a standard drive cycle. These are prescribed by legislation, and involve accelerations, decelerations, gear changes and steady-state cruising. In the USA, the FTP-75 (Federal Test Procedure) is common; as it was based on a real-world drive, it has an irregular speed profile. In Europe, the NEDC (New European Drive Cycle) is used for light-duty testing. Its speed profile is much more uniform, consisting of three identical ‘urban’ speed-profile patterns, followed by an ‘extra-urban’ phase, which includes cruises at up to 120 kph. Emissions are usually expressed in terms of amount of particulate per kilometre or mile. Testing is normally conducted from a cold start – a condition that can lead to increased particulate emission at the start of the test. A number of world-harmonised drive cycles are currently under development for use in future legislation.

As emissions standards have become increasingly tight, the filter paper method has become somewhat difficult to use in practice. The mass collected during a test has become very small and avoiding the effect of artefacts has become difficult, or at least very expensive to avoid. In addition, some concern has been raised by those studying health effects that a mass measurement does not adequately relate to health risks. In response to this concern, the United Nations Economic Commission for Europe (UNECE) commissioned a study into alternative methods of particulate-level measurement for use in future European legislation. The Particle Measurement Programme (PMP) undertook experimental investigations into the efficacy of the current methods and, after suggesting a solid-particle-number-based standard (see Section 15.5.3), undertook correlation exercises in order to demonstrate the method's practicality, repeatability, reproducibility and robustness (Andersson et al. 2010). The light-duty report found that the particle number emission from DPF-equipped diesel engines was less than 2.0 × 10¹¹ #/km, with a typical repeatability of 30%. Conventional diesel vehicles produced emissions of around 10¹³–10¹⁴ #/km; that is, 2 orders of magnitude higher than the DPF-equipped vehicles. In additional, particle number emissions from those direct-injection gasoline vehicles tested were around 10¹³ #/km. In 2006, the European Parliament endorsed the suggested solid-particle-number-based measurement system as part of the Euro 5/6 phases of emissions testing legislation in Europe. The limit for diesel light-duty vehicles was set as 6.0 × 10¹¹ #/km, and in 2011 it was proposed that this would also apply to direct-injection gasoline vehicles from 2017, with an interim limit 1 order of magnitude higher than that being introduced from 2014 (Table 1).

15.5 Sampling and Measurement Techniques

15.5.1 Sample Handling

The method by which an engine aerosol sample is collected and transported to an analyser is a most important factor in determining the quality of a subsequent measurement. It is usually necessary to dilute and/or cool an engine aerosol sample before measurement, due to limits upon the concentration and temperature set by the instrumentation used. However, dilution and cooling will inevitably change the nature of the aerosol (Lyyränen et al., 2004). As with the real-world dilution that occurs inside and outside a vehicle's exhaust system, the dilution and cooling necessary for measurement will affect the volatile particle fraction. Cooling with little or no dilution will lead to a large degree of nucleation and condensation – material in the gas phase will enter the aerosol phase and then be measured. If sufficient dilution is used, the partial pressure of the volatile material in the gas phase will be reduced, thereby preventing supersaturation and condensation into the particle phase. This process can be further accelerated by using hot dilution (Kawai, Goto and Odaka, 2004). It comes down to a question of what needs to be measured: a representative sample of aerosol, as would be found in the real environment, or (accepting that this depends on so many variables and conditions as to be almost subjective) a measurement of the much less mutable solid fraction.

As an alternative to extra dilution, a thermodenuder can be used to remove volatile species, if desired. This consists of a hot tube, which ensures volatile material is present only in the gas phase, and a region filled with activated charcoal, which then absorbs the volatile material, preventing subsequent renucleation. One advantage of a thermodenuder is that it does not require additional dilution, which enables the use of less sensitive instrumentation. The use of a catalytic stripper (Kittelson et al., 2004) is another alternative way of removing volatiles. Both thermodenuders and catalytic strippers exhibit some solid particle losses due to diffusion.

The most common on-line method of sampling from an exhaust pipe is the constant volume sampler (CVS; e.g., Burtscher, 2005). In its simplest form, this is a large-diameter tube, with the exhaust pipe fitted into one end and a fan causing a constant volumetric flow to be drawn at the other. The overall flow is much larger than the maximum exhaust flow and the additional flow drawn causes dilution of the exhaust, on the order of 10 : 1. The level of dilution varies with the changes in exhaust flow that occur across a test cycle, but the key to the CVS system is that it is simple to calculate instantaneous rates of emission, whether for particles or for gas-phase emissions. Regardless of the actual concentrations that emerge from the exhaust, the flux of a species in the CVS system is the concentration as measured in the CVS system (for particles, say, in N/cc) multiplied by the volumetric flow (cc/s, giving N/s). It is then simple to integrate the flux over time to give a total particle number (or mass) and then to divide by the distance travelled in the test to give the legislated quantity in N/km (or mg/km). By contrast, if one were to sample directly from the exhaust, it would be necessary to know the flow in the exhaust pipe. Given the pernicious nature of the species present in the exhaust flow, it is not usually possible to measure this flow directly using a mass-flow meter. It can be inferred from the air intake flow to the engine, with an addition to take account of the fuel burnt, but this is considerably more complex and more prone to error than the use of a CVS system. It is usually desirable to filter the clean air intake to the CVS system, especially if it is to be used to sample a DPF-equipped vehicle with relatively low emissions.

A less bulky and less expensive alternative to a CVS system is a partial-flow dilution tunnel. Such tunnels sample a defined proportion of the exhaust flow, but in order to maintain this, measurement of the exhaust flow and a rapid control system are required.

Given the high temperatures of exhaust gases and the fact that most instruments require a cool sample, it is important to avoid thermophoretic loss, which can occur when a sample is gradually cooled. It is therefore usually preferential to cool as quickly as possible by adding cold air, rather than allowing a sample to gradually cool in a length of pipe. Loss of particles by diffusion can be a significant source of error if sufficiently narrow and long tubing is used. It is also vital to use electrically conductive tubing, in order to avoid loss of particles by electrophoresis. This usually means stainless steel where possible, or special electrically conductive silicone or PTFE tubing where flexibility is required, though those nonmetallic materials can deteriorate and emit particles at even moderately high temperatures.

Particles deposited on the walls of the sampling and transport system may at times re-entrain back into the aerosol phase. After re-entrainment, these particles tend to be large, and their effect can be reduced by using a cyclone before the measurement system in order to remove particles above the size of interest. It has been reported that certain types of tubing can absorb hydrocarbons from the exhaust gases, subsequently releasing them in an unpredictable manner, creating additional nucleation-mode particles (Maricq et al., 1999).

15.5.2 Mass Measurement

Most legislation concerning particle emissions from vehicles is still expressed in terms of particle mass collected on filter paper. After dilution (e.g., with a CVS system), a known mass flow of sample is drawn through one (or more) filter papers (often a backup filter is used in series with the main filter in order to catch any material that passes through the latter). The filters are weighed on a sensitive balance both before and after the test and the total particulate mass is taken as the difference – usually on the order of a few milligrams. As the flows in the CVS system and through the filter paper are known, scaling the filter paper mass to represent the total mass over the test is simply a matter of using the ratio of those flows as a scaling factor.

The species adsorbed on the paper include not only the solid carbonaceous and ash fractions but also any volatile material that has condensed. It is almost impossible to know if the volatile material adsorbed would have been in the aerosol phase under real-world conditions, and the levels present are very dependent upon the dilution and sampling conditions. As vehicles become ever cleaner, and in particular with the widespread introduction of DPFs, the level of mass which has to be measured on the filter paper is approaching the practical detection limit for the technique (Liu et al., 2009). Various techniques have been used to improve detectability using filter paper. For example, United States Environmental Protection Agency (EPA) regulation 40 CFR 1066 specifies measures to deal with buoyancy and electrostatic charge.

Various on-line measurement techniques used in the automotive industry give either a direct or an indirect measurement of particle mass. Opacity (light extinction) has been used for many years as a surrogate for particle mass. These instruments either work directly by shining light through the aerosol or indirectly by capturing the particulate matter on a filter paper and measuring the ‘blackness’ of the paper. The former technique is cross-sensitive to NO₂, which is usually present in diesel exhaust. The latter technique is more sensitive but is pseudo-continuous. Two other techniques, the tapered element oscillating microbalance (TEOM) and the quartz crystal microbalance (QCM), both acquire particles on a collection device and use the resonant frequency to measure the mass present. Another device used is the photo acoustic soot sensor (PASS), which uses a light source to irradiate the soot and a sensitive microphone to pick up the resulting sound, which is then used to infer particle mass concentration.

Instruments are available that use opposing electrical and centrifugal forces between concentric spinning cylinders and apply a voltage to classify charged aerosol particles by their mass–charge ratio, such as the aerosol particle mass analyser (APM; Ehara, Hagwood and Coakley, 1996) and the centrifugal particle mass analyser (CPMA; Olfert and Collings, 2005). These have been used as primary standards to measure engine particle mass concentrations, and a comparison with filter-paper measurements shows that the artefact resulting from absorbed volatiles on the filter paper can account for 50% of the total filter-paper mass (Park, Kittelson and McMurry, 2003). Thermogravimetric analysis (TGA) is often used to determine the percentage weight of a sample attributed to carbonaceous soot.

It is also possible to calculate particle mass concentration by appropriately weighting the data from a particle-sizing instrument, assuming a density and a fractal dimension – see Section 15.5.5.

15.5.3 Solid-Particle-Number Measurement

Mass measurement by filter paper has several notable limitations (Liu et al., 2009):

• As engines become progressively cleaner, it becomes harder to accurately measure the low masses of soot produced.
• Filter paper can be affected by artefacts caused by condensed volatile species that would perhaps ‘normally’ be in the gas phase rather than the particulate phase. In many cases, the mass of such volatile material can outweigh the soot produced by a modern engine.
• Mass measurement is inherently biased towards the presence of larger particles. There is much evidence that smaller nanoparticles have greater detrimental health effects than larger nanoparticles; thus, a limitation on particle mass may not correlate all that well with improved health outcomes.

In order to address these concerns in the European Union, a solid-particle-number measurement method has been introduced, in time for the Euro 6 stage of emissions standards (Andersson et al., 2007; UNECE R83). The actual particle number counting is performed with a condensation particle counter (CPC). These are more than sufficiently sensitive to make this measurement, usually having a concentration range (in single-particle counting mode) of between <1 and 10 000 particles per cubic centimetre. Above this range, CPCs often operate in photometric mode, where an estimate of particle number is made from bulk optical scattering by particles, rather than by counting individual particles; they therefore require empirical calibration. The upper limit of a CPC in count mode is usually several orders of magnitude less than the concentration of particulate from a typical diesel vehicle, and thus some form of dilution is required upstream of the CPC if count mode is to be used (as mandated).

The reason the methodology chooses to measure only solid particles, and thus exclude volatile particles, is that the latter are difficult to measure with any repeatability. Also, little medical research has been conducted to confirm that there are health effects associated with volatile particles – unlike the solid particles, which can persist in the lungs. Nucleation-mode particles can readily be created from gas-phase species and destroyed—processes that are highly sensitive to the exact ambient environmental, sampling and dilution conditions used. In order to make any sort of meaningful comparison of vehicles with standards, or to be able to spot trends caused by engineering changes, it is very important for such measurements to be repeatable.

The key to the removal of volatile particles in the system is the dilution systems upstream of the CPC (Figure 15.3). Two stages of dilution are used. The first heats the aerosol to 150 °C and serves to both evaporate existing volatile particles and reduce the partial pressure of gas-phase species to prevent their formation into particles. This stage is followed by an evaporation tube, where the sample is heated to 300 °C for ∼0.2 seconds to evaporate any further semivolatile material. After this tube, the second stage reduces the gas partial pressures further in order to prevent renucleation, and also cools the sample rapidly (to prevent thermophoretic loss) to allow measurement by the CPC.

The type of diluter used for the first stage is often a rotating disc diluter, which consists of a disc with a series of blind- or through-holes parallel to the axis of rotation, into which concentrated aerosol is deposited by a flow from the sample source. As the disc rotates, the holes transfer the sample to a flow of clean air, which then scavenges the particles and dilutes them. The dilution ratio achieved is controlled by the speed of rotation – the lower the speed, the higher the dilution. Alternative systems use careful measurement of the aerosol flow and controlled metered dilution with clean air. It is usually not possible to measure particle-laden flow using conventional mass-flow controllers, so often the pressure drop across an orifice plate is used.

The CPC mandated for use in this type of system is designed, by varying the internal saturator temperature, to have a d₅₀ size cut-off of 23 nm. This is to ensure that nucleation-mode material which is not removed by the dilution stages or the evaporation tube is simply not counted by the CPC. However, this is controversial as not all particles <23 nm can be said to be volatile. Solid ash particles can be smaller than this size, and while not volatile, they would be removed by the CPC cut-off point (Gidney, Twigg and Kittelson, 2010).

15.5.4 Sizing Techniques

The scanning mobility particle sizer (SMPS) has historically been the standard instrument used to measure nanoparticle size distributions. However, with a normal scan time of over 1 minute, it is not suitable for rapidly changing aerosol sources such as engines under non-steady-state conditions. Indeed, legislated drive cycles are by their very nature transient.

One of the first instruments to allow a real-time measurement of the particle size distribution from engines was the electrical low-pressure impactor (ELPI; Ahlvik et al., 1998). This consists of a charger and a series of impactor plates. When a particle lands on a plate, a current is detected and the aerodynamic diameter is inferred. While not having as high a spectral or temporal resolution as the electrical mobility devices described later, the range of particle size measurable extends from a few nanometres up to 10 µm, and the use of impactor plates allows the size-segregated collection of sample for off-line testing, for example by chemical analysis.

In recent years, fast particle-mobility spectrometers have become available, which despite compromising on the sensitivity and spectral resolution of an SMPS, offer much faster data rates (up to 10 Hz) and, more importantly, short response times (down to 200 ms). These differential mobility spectrometers (Reavell, Hands and Collings, 2002; Johnson et al., 2004) first charge the aerosol with a unipolar diffusion charger, which places a higher level of charge than a bipolar charger. As the name suggests, the charge is (usually) net positive rather than a net neutral distribution. The particles then pass into a classification column, which is similar to a differential mobility analyser (DMA) in that there is a central high-voltage electrode (which here repels the positively charged particles), and a sheath air flow, which carries the particles towards the other end of the column. The particles move in trajectories that depend upon their charge–drag ratio, eventually landing on a series of metal detection rings placed along the inside of the outer wall of the classification column. Each detector is connected to a sensitive electrometer circuit, and when a particle lands, a small current is registered (on the order of a few femtoamperes). A data-inversion algorithm uses a charging model and a model of the classifier to generate size spectral density versus diameter from the measured currents. The data in Figure 15.1 were obtained using such an instrument.

15.5.5 Morphology Determination

Transmission electron microscopy (TEM) is commonly used to study the structure of soot aggregates (Figure 15.4). The primary particles are usually clearly distinguishable and their size can be estimated; the number (N) in an agglomerate scales with the radius of gyration (R_g) as:

where D_f is the fractal dimension, d_p is the primary particle diameter and k_g is a constant (Mandelbrot, 1982). The radius of gyration is the root-mean-squared distance of the primary particles from the centre of the aggregate. For soot aggregates, D_f is usually just less than 2.0 (e.g. Park, Kittelson and McMurry, 2004).

It is possible to express the mass of a particle in terms of a power law dependent upon its diameter:

where D_fm is the mass-mobility exponent (also known as the mobility-diameter-based fractal dimension, which is different to the fractal dimension based on radius of gyration). For spherical particles, D_fm = 3, and for agglomerate particles, such as those in the accumulation mode of engine particle emissions, 2 < D_fm < 3. In order to determine D_fm for an aerosol source, it is necessary to measure both the diameter and the mass of the particles. Size selection is usually performed with a DMA and in-series mass selection by an electric-centrifugal particle mass analyser, such as an APM (Park et al., 2003) or a CPMA (Olfert, Symonds and Collings, 2007). Using a CPC as a detector, the peak mass can be determined for a given mobility diameter and particle effective density can be calculated as a function of diameter (Figure 15.5). With this technique, Park et al. (2003) found the mass-mobility exponent of heavy-duty soot for various engines to be between 2.33 and 2.41, and Olfert, Symonds and Collings (2007) found that for one Peugeot diesel engine, the mass-mobility exponent ranged from 2.22 to 2.48, and as high as 2.76 at higher load. A high mass-mobility exponent for soot particles usually indicates absorbed volatile material infilling the agglomerate structure; the particle thus becomes ‘more spherical’. In the case of the Peugeot engine, the volatile material was thought to be sulfate. Soot from gasoline engines tends to have a higher mass-mobility exponent that that from diesel engines, symptomatic of a higher level of semivolatile material in the exhaust stream.

An alternative to a tandem DMA-CPMA experiment or equivalent is to use a measure of aerodynamic diameter in tandem with a DMA such as an ELPI. This technique has been used by Maricq and Xu (2004), for example. As aerodynamic diameter depends upon mass, an effective density can be calculated as a function of mobility diameter by comparing the mobility and aerodynamic diameters, and from this the mass-mobility exponent can be obtained. Maricq and Xu (2004) obtained a mobility-diameter-based fractal dimension for diesel vehicle soot of 2.3.

Once D_fm has been determined, it is possible to use spectral instruments to estimate particle mass from a given size spectrum (Kittelson et al., 2004). If a discrete particle size spectrum is used for this calculation then it is prone to excessive noise from the channels representing the largest particles; these have most significance in the calculation as they are the heaviest. One solution to this is to fit a lognormal function to the data (which suppresses spectral noise in the tail of the function) then use Hatch–Choate equations to calculate the diameter of average mass and weight this to calculate the total particle mass (Symonds et al., 2007).

15.6 Amelioration Techniques

15.6.1 Fuel Composition

We have already seen that reducing the sulfur impurities in fuel can reduce the sulfate fraction of engine aerosol dramatically and that organometallic fuel additives can have a soot-suppressing effect (though they can lead to an increase in ash). The combustion quality of diesel fuel is determined by its cetane number. This is a measure of the fuel's ignition delay: the delay between ignition and a detectable rise in pressure. On one hand, decreasing the cetane number of the fuel, and thus increasing the ignition delay, allows more time for air entrainment, favouring premixed combustion, and can therefore reduce carbonaceous soot emission (Li, Chippior and Gülder, 1996); on the other, it can lead to increased organic fraction emission due to an increased chance of wall impingement.

In recent years, biofuels such as biodiesel, ethers (such as dimethyl ether, DME) and ethanol have been increasingly used in fuel blends. Biodiesel consists of long-chain ester molecules. Apart from their other environmental and sustainability advantages (though they are not without controversy), the presence of oxygen in these substances leads to a decrease in soot formation during combustion (Yage, Cheung and Huang, 2009), although an increase in volatile particle emission can sometimes be seen (Northrop et al., 2011). It is not currently known exactly why the presence of oxygenates leads to decreased soot production, although it may be as simple as providing more oxygen for combustion (Rakopoulos, Antanopolous and Rakopoulos, 2006).

15.6.2 Control by Engine Design and Calibration

Reducing particulate emissions by engine design and calibration (the set-up of the engine operating conditions in the electronic engine management system) is an attractive option for engine manufacturers, as it can often make a considerable difference without incurring the extra cost and complication of aftertreatment systems (e.g. Klindt, 2010).

Fuel pressure and timing are among the most influential calibration parameters that affect particulate emissions. By increasing the fuel pressure, the distance from the injector nozzle to the combustion zone is increased, which increases the amount of air entrained in the spray, making the local air–fuel mixture leaner and reducing the amount of unburnt and partially unburnt fuel, which in turn lowers particulate emissions (Picket and Siebers, 2004). The resulting higher combustion temperatures also lead to a higher rate of in-cylinder soot oxidation. Of course, if the injection pressure is raised too much, fuel can impinge on the cylinder walls or piston, with a resulting increase in emissions, including of particulate. The resultant increased cylinder pressure during combustion can also lead to engine damage.

The injection of additional fuel after the main injection event, known as post-injection, can be an effective means of reducing PM emission (Desantes et al., 2011). Splitting the injection means that each injection leads to a less locally rich mixture. However, care must be taken not to post-inject too late and cool the charge, leading to reduced in-cylinder soot oxidation. The relative position and angle of the injector and spark plug in a GDI engine, and whether the fuel is guided by the surface of the piston or localised in the region of the spark plug, can also make a large difference to the quantity and nature of particulate emissions (Price et al., 2006).

Exhaust gas recirculation (EGR) has been introduced to reduce NO_x emissions. A portion of the exhaust gas is recirculated back to the intake manifold (essentially inert gas, which cools the combustion process by lowering the heat capacity and diluting oxygen), thereby reducing the reactions of nitrogen and oxygen favoured at high temperatures which produce NO_x. A side effect of EGR, however, is to increase particulate emissions, due to the decreased availability of oxygen. The trade-off between NO_x and particulate emissions resulting from varying the amount of EGR used presents a difficult challenge to engine calibrators.

15.6.3 Particulate Filters

In recent years, the DPF has provided a very effective solution to the emission of nanoparticles from diesel engines. DPFs consist of a (usually) ceramic substrate placed in a metal container in the exhaust system of the vehicle. The most common substrate construction is the wall flow type (Howitt and Montierth, 1981), in which a grid of channels is formed along the length of the substrate, with alternate channels being blocked off at alternate ends. Exhaust enters via any of the 50% of channels open at the engine end of the DPF, particles are removed as the gas stream passes through the interchannel wall and the particle-reduced gas stream passes into the channels that are open at the tailpipe end. This configuration serves to maximise the surface area available to filtration for a given volume of filter. Common substrate materials include cordierite and silicon carbide. SiC has a higher melting point than cordierite, but it does have a high thermal expansion coefficient, which warrants SiC filters being built in sections (which are cemented to form the overall substrate) in order to allow for thermal expansion.

On a microscale, the walls of a DPF contain many irregular pores. Initially, when a DPF is clean, filtration occurs by deposition of the pore walls (Figure 15.6). After some time, soot gets deposited across the necks of the pores. As these pores fill up, the backpressure across the DPF caused by gas being pumped through it by the engine rapidly increases. Eventually, most of the pores are filled and filtration occurs through a layer of soot on the channel walls – the so-called cake-filtration stage. As more soot arrives, the cake layer increases in thickness and the backpressure continues to rise, although at a slower rate than during the pore-filling stage. The performance of a DPF is normally expressed in terms of its filtration efficiency. When a DPF is clean, at the start of the pore-filling stage, its filtration efficiency can be as low as 50%, but as the DPF begins to fill, its efficiency rapidly increases. At the cake-filtration stage, the efficiency can reach over 99%. This is factored into emissions-testing protocols.

Eventually, the DPF fills with soot to the extent that the backpressure becomes detrimental to the performance of the engine (or more likely, damages the DPF, as discussed later). At this point, it is necessary to regenerate the DPF. There are two forms of regeneration, which apply to different DPF systems to varying degrees. In passive regeneration, soot is continuously oxidised at a slow rate by the thermal action of the exhaust stream. Catalytic coatings on the DPF substrate or catalytic fuel additives can be used to make this reaction more favourable at lower temperatures. In active regeneration, an automated process causes the DPF to be regenerated when the engine management system detects a sufficiently large pressure drop across the DPF. Methods of active regeneration commonly use a change in engine control strategy (e.g. changing injection timing) to cause the exhaust gases to increase in temperature. Other methods include injecting fuel into the exhaust stream and electric heating. There is usually an efficiency and fuel economy penalty to pay during regeneration.

Soot oxidation can occur through reaction with either oxygen or NO₂ (which is usually present in diesel exhaust). One variant of the DPF, the Continuously Regenerating Trap (CRT; Cooper and Thoss, 1989), uses an oxidation catalyst before the filter to oxidise NO to NO₂. Combustion with NO₂ can occur at normal exhaust operating temperatures (∼ 250 °C) and this variant thus improves passive regeneration performance.

Active regeneration occurs at temperatures in excess of 600 °C and the oxidation reaction is itself exothermic. The temperatures reached can be in excess of 1000 °C, which can place considerable stress on the ceramic substrate. The maximum backpressure reached before active regeneration is induced may in practice be much less than that which in itself would be tolerated by the engine, in order to prevent the absolute mass of soot accumulated on the filter from reaching a level such that the heat released during regeneration permanently damages the substrate.

It should be remembered that DPFs are most effective at removing solid particles; that is to say, the carbonaceous and ash fractions. Nucleation-mode particles may not normally even be formed until the exhaust system or real-world dilution, so they may pass through a DPF as their gas-phase precursors. It is common for volatile materials to adsorb on to solid particles, meaning that if the solid particles have been removed by the DPF, homogenous nucleation of volatiles passing through the DPF into a particle mode may be promoted. The formation of sulfuric acid in a DOC and its desorption from the DOC and DPF can be made worse under the high-temperature conditions of DPF regeneration, and large quantities of nucleation aerosol have been observed during regeneration (Campbell et al., 2006).

The DPF also captures ash particles (including any formed from fuelborne additives intended to catalyse regeneration), but these are not usually removed by regeneration. Although the proportion of ash in the engine exhaust is small, over time this can lead to a gradual reduction in the capacity of the DPF.

The widespread adoption of DPFs has been driven by ever more stringent legislation, such as the PMP project and the Euro 5 and 6 standards for light- and heavy-duty engines in Europe and the US2007 standard for heavy-duty engines in the USA (light-duty diesel vehicles are not yet common there). A number of retrofit programmes are in place globally, particularly for off-road applications (especially vehicles used indoors or in confined spaces). A number of bus fleets have been retrofitted and low-emission zones are becoming widespread in Europe, which mandate the retrofitting of DPFs.

With the advent of the Euro 6 emission standards, which will introduce a particle-number standard for gasoline-engine particle emissions, some manufacturers are considering the use of gasoline particulate filters (GPFs, e.g. Mikulic et al., 2010). Due to the increased temperatures involved in gasoline combustion, continuous thermal passive regeneration tends to be a feature of GPFs. This means that the level of soot filling for optimal filtration efficiency may never be reached. However, as the level of particulate emissions from even direct-injection gasoline engines is much less than that from diesel engines, the amount of attenuation required from a GPF is usually much less than that required by a DPF.

Acknowledgements

The author would like to thank Professor Nick Collings of Cambridge University Engineering Department and Chris Nickolaus, Mark Peckham and Kingsley Reavell of Cambustion for their useful suggestions.

References

Abdul-Khalek, I., Kittelson, D.B. and Brear, F. (2000) Nanoparticle Growth During Dilution and Cooling of Diesel Exhaust: Experimental Investigation and Theoretical Assessment. SAE Technical Paper 2000-01-0515.

Ahlvik, P., Ntziachristos, L., Keskinen, J. and Virtanen, A. (1998) Real Time Measurements of Diesel Particle Size Distribution with an Electrical Low Pressure Impactor. SAE Technical Paper 980410.

Andersson, J., Giechaskiel, B., Muñoz-Bueno, R. et al., (2007) Particle Measurement Programme (PMP): Light-duty Inter-laboratory Correlation Exercise (ILCE_LD) Final Report, Joint Research Centre, European Commission, http://www.unece.org/trans/main/wp29/wp29wgs/wp29grpe/grpeinf54.html (last accessed 8 August 2013).

Andersson, J., Mamakos, A., Giechaskiel, B., Carriero, M. and Martini, G., (2010) Particle Measurement Programme (PMP): Heavy-duty Inter-laboratory Correlation Exercise (ILCE_HD) Final Report, Joint Research Centre, Ispra http://www.unece.org.unecedev.colo.iway.ch/fileadmin/DAM/trans/doc/2009/wp29grpe/PMP-24-02e.pdf (last accessed 15 August 2013).

Burtscher, H. (2005) Physical characterization of particulate emissions from diesel engines: a review. Journal of Aerosol Science, 36, 896–932.

Buzzard, N.A., Clark, N.N. and Guffey, S.E. (2009) Investigation into pedestrian exposure to near-vehicle exhaust emissions. Environmental Health, 8(13). doi: 10.1186/1476-069X-8-13.

Campbell, B., Peckham, M., Symonds, J. et al., (2006) Transient Gaseous and Particulate Emissions Measurements on a Diesel Passenger Car Including a DPF Regeneration Event. SAE Technical Paper 2006-01-1079.

Carpentieri, M., Kumar, P. and Robins, A. (2011) An overview of experimental results and dispersion modelling of nanoparticles in the wake of moving vehicles. Environmental Science and Technology, 159, 685–693.

Cooper, B.J. and Thoss, J.E. (1989) Role of NO in Diesel Particulate Emission Control. SAE Technical Paper 890404.

Dec, J.E. (1997) A Conceptual Model of Di Diesel Combustion Based on Laser-Sheet Imaging. SAE Technical Paper 970873.

Desantes, J.M., Bermúdez, V., García, A. and Linares, W.G. (2011) A comprehensive study of particle size distributions with the use of postinjection strategies in DI diesel engines. Aerosol Science and Technology, 45(10), 1161–1175.

Eastwood, P. (2008) Particulate Emissions from Vehicles, SAE International/John Wiley & Sons, Ltd, Chichester.

Ehara, K., Hagwood, C. and Coakley, K.J. (1996) Novel method to classify aerosol particles according to their mass-to-charge ratio—aerosol particle mass analyser. Journal of Aerosol Science, 27, 217–234.

Fortuin, J.P.F., Van Dorland, R., Wauben, W.M.F. and Kelder, H. (1995) Greenhouse effects of aircraft emissions as calculated by a radiative transfer model. Annales Geophysicae, 13(4), 413–418.

Fujiwara, Y., Tosaka, S. and Murayama, T. (1993) Formation Process of SOF in the Combustion Chamber of IDI Diesel Engines. SAE Technical Paper 932799.

Gidney, J.T., Twigg, M.V. and Kittelson, D.B. (2010) Effect of organometallic fuel additives on nanoparticle emissions from a gasoline passenger car. Environmental Science and Technology, 44(7), 2562–2569.

Greeves, G. and Wang, C.H.T. (1982) Origins of Diesel Particulate Mass Emission. SAE Technical Paper 810260.

Hands, T., Nickolaus, C., Symonds, J. and Finch, A. (2010) Real-Time Particle Emissions from 2-Stroke Motorbikes with and without PMP Sampling System. SAE Technical Paper 2010-32-0047.

Heywood, J.B. (1988) Internal Combustion Engine Fundamentals, McGraw-Hill International Editions.

Howard, J.B. and Kausch, W.J. (1980) Soot control by fuel additives. Progress in Energy and Combustion Science, 6, 263–276.

Howitt, J.S. and Montierth, M.R. (1981) Cellular Ceramic Diesel Particulate Filter. SAE Technical Paper 810114.

Johnson, T., Caldow, R., Pocher, A. et al., (2004) An Engine Exhaust Particle Sizer Spectrometer for Transient Emission Particle Measurements. SAE Technical Paper 2004-01-1341.

Johnson, B.T., Hargrave, G.K. and Reid, B. (2011) Crankcase Sampling of PM from a Fired and Motored Compression Ignition Engine. SAE Technical Paper 2011-24-0209.

Jung, H., Kittelson, D.B. and Zachariah, M.R. (2003) The Influence of Engine Lubricating Oil on Nanoparticle Emissions and Kinetics of Oxidation. SAE Technical Paper 2003-01-3179.

Kato, S., Takayama, Y., Sato, G.T. et al., (1997) Investigation of Particulate Formation of DI Diesel Engine with Direct Sampling from Combustion Chamber. SAE Technical Paper 972969.

Kawai, T., Goto, Y. and Odaka, M. (2004) Influence of Dilution Process on Engine Exhaust Nano-particles. SAE Technical Paper 2004-01-0963.

Kittelson, D.B. (1998) Engines and nanoparticles: a review. Journal of Aerosol Science, 29(5/6), 575–588.

Kittelson, D., Hands, T., Nickolaus, C. et al., (2004) Mass correlation of engine emissions with spectral instruments. Proceedings of 2004 Japanese SAE Annual Congress. Paper No. 20045462.

Klindt, K. (2010) Reducing the particulate emission numbers in DI gasoline engines. Proceedings of Directions in Engine-Efficiency and Emissions Research (DEER) Conference.

Kumar, P., Robins, A., Vardoulakis, S. and Britter, R. (2010) A review of the characteristics of nanoparticles in the urban atmosphere and the prospects for developing regulatory control. Atmospheric Environment, 44, 5035–5052.

Li, X., Chippior, W.L. and Gülder, Ö. (1996) Effects of Fuel Properties on Exhaust Emissions of a Single Cylinder DI Diesel Engine. SAE Technical Paper 962116.

Liu, Z.G., Vasys, V.N., Dettmann, M.E. et al. (2009) Comparison of strategies for the measurement of mass emissions from diesel engines emitting ultra-low levels of particulate matter. Aerosol Science and Technology, 43(11), 1142–1152.

Lyyränen, J., Jokiniemi, J., Kauppinen, E.I. et al. (2004) Comparison of different dilution methods for measuring diesel particle emissions. Aerosol Science and Technology, 38, 12–23.

Mandelbrot, B. (1982) The Fractal Geometry of nature, Freeman, San Francisco, CA.

Maricq, M.M., Chase, R.E., Podsadlik, D.H. and Vogt, R. (1999) Vehicle Exhaust Particle Size Distributions: A Comparison of Tailpipe and Dilution Tunnel Measurements. SAE Technical Paper 1999-01-1461.

Maricq, M.M. and Xu, N. (2004) The effective density and fractal dimension of soot particles from premixed flames and motor vehicle exhaust. Journal of Aerosol Science, 35, 1251–1274.

Mikulic, I., Koelman, H., Majkowski, S. and Vosejpka, P. (2010) A study about particle filter application on a state-of-the-art homogeneous turbocharged 2l DI gasoline engine. Aachener Kolloquium Fahrzeug- und Motorentechnik, 19, 1–18.

Miller, K.A., Siscovick, D.S., Sheppard, L. et al. (2007) Long-term exposure to air pollution and incidence of cardiovascular events in women. New England Journal of Medicine, 356, 447–458.

Northrop, W.A., Madathil, P.V., Bohac, S.V. and Assanis, D.N. (2011) Condensation growth of particulate matter from partially premixed low temperature combustion of biodiesel in a compression ignition engine. Aerosol Science and Technology, 45(1), 26–36.

Olfert, J. and Collings, N. (2005) New method for particle mass classification — The Couette centrifugal particle mass analyser. Journal of Aerosol Science, 36, 1338–1352.

Olfert, J.S., Symonds, J.P.R. and Collings, N. (2007) The effective density and fractal dimension of particles emitted from a light-duty vehicle with diesel oxidation catalyst. Journal of Aerosol Science, 38, 69–82.

Park, K., Cao, F., Kittelson, D.B. and McMurry, P.H. (2003) Relationship between particle mass and mobility for diesel exhaust particles. Environmental Science and Technology, 37, 577–583.

Park, K., Kittelson, D.B. and McMurry, P.H. (2003) A closure study of aerosol mass concentration measurements: comparison of values obtained with filters and by direct measurements of mass distributions. Atmospheric Environment, 37, 1123–1230.

Park, K., Kittelson, D.B. and McMurry, P.H. (2004) Structural properties of diesel exhaust particles measured by transmission electron microscopy (TEM): relationship s to particle mass and mobility. Aerosol Science and Technology, 38, 881–889.

Payne, S. (2011) Experimental studies of diesel particulate filtration. PhD thesis. University of Cambridge.

Petzold, A., Marsh, R., Johnson, M. et al. (2011) Evaluation of methods for measuring particulate matter emissions from gas turbines. Environmental Science and Technology, 45, 3562–3568.

Pickett, L.M. and Siebers, D.L. (2004) Soot in diesel fuel jets: effects of ambient temperature, ambient density and injection pressure. Combustion and Flame, 138, 114–135.

Pipho, M.J., Ambs, J.L. and Kittelson, D.B. (1986) In-cylinder Measurements of Particulate Formation in an Indirect Injection Diesel Engine. SAE Technical Paper 860024.

Price, P., Stone, R., Collier, T. and Davies, M. (2006) Particulate Matter and Hydrocarbon Emissions Measurements: Comparing First and Second Generation DISI with PFI Engines in Single Cylinder Optical Engines. SAE Technical Paper 2006-01-1263.

Rakopoulos, C.D., Antanopolous, K.A. and Rakopoulos, D.C. (2006) Multi-zone modelling of diesel engine fuel spray development with vegetable oil, bio-diesel or diesel fuels. Energy Conversion and Management, 47, 1550–1573.

Ramanathan, V. and Carmichael, G. (2008) Global and regional climate changes due to black carbon. Nature Geoscience, 1, 221–227.

Reavell, K., Hands, T. and Collings, N.A. (2002) Fast Response Particulate Spectrometer for Combustion Aerosols. SAE Technical Paper 2002-01-2714.

Schmid, O., Hagen, D.E., Whitefield, P.D. et al. (2011) Methodology for particle characterisation in the exhaust flows of gas turbine engines. Aerosol Science and Technology, 38(11), 1108–1122.

Symonds, J.P.R., Reavell, K.S.J., Olfert, J.S. et al. (2007) Diesel soot mass calculation in real-time with a differential mobility spectrometer. Journal of Aerosol Science, 38, 52–68.

UNECE R83 (United Nations Economic Commission for Europe), (2011) World Forum for Harmonization of Vehicle Regulations, Regulation Number 83—Emission of Pollutants According to Engine Fuel Requirements, available from http://www.unece.org/fileadmin/DAM/trans/main/wp29/wp29regs/r083r4e.pdf (last accessed 15 August 2013).

US EPA (United States Environmental Protection Agency) (2002) Health Assessment Document for Diesel Engine Exhaust.

Williams, R.L., Perez, J.M. and Griffing, M.E. (1985) A Review of Sampling Condition Effects Upon Polynuclear Aromatic Hydrocarbons (PNA) from Heavy-duty Diesel Engines. SAE Technical Paper 852081.

Witze, P.O. and Green, R.M. (2005) Comparison of Single and Dual Spray Fuel Injectors during Cold Start of a PFI Spark Ignition Engine Using Visualization of Liquid Fuel Films and Pool Fires. SAE Technical Paper 2005-01-3863.

Yage, D., Cheung, C.S. and Huang, Z. (2009) Comparison of the effect of biodiesel-diesel and ethanol-diesel on the particulate emissions of a direct injection diesel engine. Aerosol Science and Technology, 43(5), 455–465.

^a California tends to have its own set of emissions standards, often leading the rest of the USA.