Electron Microscopy Investigation of Particulate Matter from a Dual Fuel Engine

Abstract

The particulate matter (PM) of a dual fuel engine was characterized in size, morphology and fractal geometry by using scanning electron microscopy (SEM) and transmission electron microscopy (TEM). Particulate samples were collected from the diluted exhaust of the engine operated on diesel fuel, natural gas (NG) and synthetic biogas. The engine operating condition was kept the same to compare the results between diesel and dual fuel PM. SEM images yielded agglomerate number, size distributions and a shape descriptor. TEM was used to investigate the primary particle size distribution in agglomerates and the fractal dimensions of the sampled PM. Long chainlike PM agglomerates appeared for the diesel high load condition, whereas PM agglomerates for dual fueling were found to be smaller in size and with more spherical shapes. All of the measured PM appeared to have a bi-modal number size distribution irrespective of engine fueling condition. The average primary particle diameter increased for dual fuel PM (ranging from 26.9 to 29.5 nm) compared to diesel PM (26.4 nm). The average diameter tended to increase with the introduction of CO₂ in the gaseous fuel. PM fractal dimensions were in the range from 1.69 to 1.88 for different PM samples. Higher fractal dimensions (from 1.73 to 1.88) were obtained for dual fuel PM compared to diesel (high load) PM (1.69). This finding also implies that diesel PM are more chainlike and elongated compared to the PM measured for dual fueling conditions. The very different engine fueling conditions used here give valuable understanding of the formation processes of PM.

INTRODUCTION

Diesel engines are widely used in both stationary and mobile applications. They provide high thermal efficiency, and consume relatively cheap fuel. However, diesel engines emit harmful particulate matter (PM). Exposure to diesel PM occurs within many different occupational groups and diesel PM can make an important contribution to ambient PM (Vouitsis et al. 2003). Due to their alleged adverse health and environmental effects, diesel PM have been of great concern in recent times.

The use of gaseous fuels in diesel engines can be a strategy for the simultaneous reduction of PM and NO_x emissions (Mustafi et al. 2007; Mustafi and Raine 2008). Gaseous fuels in diesel engines operate in dual fuel mode where the main energy comes from the gaseous fuel and a minimal amount of diesel fuel acts as the ignition source. A diesel engine can easily be modified to dual fuel operation permitting the engine to run on a variety of gaseous fuels. Since diesel engines operate at high compression ratios, they permit the use of low energy content renewable gaseous fuels such as biogas. The benefits of the use of alternative and renewable gaseous fuels are two fold: substitution for diesel fuel and reduction of harmful pollutants. The use of natural gas in diesel engines is already in practice because of its availability in many parts of the world. Biogas, on the other hand, is a renewable fuel that can be produced from organic wastes.

Current regulations concern PM gravimetric concentrations only, i.e., the mass of all particulates that can be collected from the exhaust. However, in addition to gravimetric measurements, PM characterization in terms of properties such as number, size, shape, or fractal dimension is also important to provide better understanding of particle formation and removal processes. Also the size and the structure of the PM influence their atmospheric transport properties, optical properties, deposition behavior, depth of penetration into the lung, etc. (Park et al. 2004; Chakrabarty et al. 2006).

In diesel engine combustion soot particles are formed as a result of incomplete combustion in fuel rich zones and hydrocarbons are adsorbed or condensed onto their surfaces afterwards (Figler et al. 1996; Kittelson 1998). Although many studies are available in the literature relative to diesel engine PM emissions and their characterization, research gaps remain, for example with respect to measurements of PM generated by dual fuel engines.

A light duty diesel engine is used in this research, which can be a representative of the existing older diesel engines in use. These older engines will remain in use, in the future decades, especially in the developing countries, without aftertreatment facility. Thus this present research provides improved understanding of the effects of dual fueling on PM emissions and many may apply the dual fueling technique in attempts to improve PM emissions of such older engines.

Electron microscopy techniques have been used to characterize PM physically and to investigate their morphology (Hinds 1999; Park et al. 2004; Nord et al. 2004). Digital images generated from electron microscopy are analyzed to get the projected two dimensional (2D) properties as well as the three dimensional (3D) structural properties of the PM agglomerates, such as fractal dimensions, which also provide insights into the agglomeration mechanism (Park et al. 2004). In this study, scanning electron microscopy (SEM) is used to determine the number size distributions, and the shape of the measured PM emitted from the diesel engine while operated in diesel and dual fuel modes. The shape of the PM is characterized using the shape factor (SF) as a shape descriptor (Nord et al. 2004). Transmission electron microscopy (TEM) is used to investigate the primary particle diameters of the measured PM. Computer aided analysis of the TEM images also yielded the 3D fractal dimensions of the PM using the 2D properties. Results are compared between diesel and dual fuel conditions. Based on the above observations, possible growth mechanisms for the PM agglomerates are identified.

EXPERIMENTAL

Engine Parameters and Fuels

The investigation was carried out on a Lister Petter, direct injection (DI) diesel engine modified to run in either diesel or dual fuel modes (Table 1). The modification is simple and is described in details in Mustafi (2008) and in Mustafi and Raine (2008). Measurements were taken at two different engine loads: low load (3 nm; equivalent to 10% of the rated output) and high load (28 nm) for diesel fueling and only at high load (28 nm) for dual fueling. For the dual fueling, the amount of diesel fuel was kept the same as for the diesel low load condition and the desired output torque was obtained by increasing the amount of gas flow into the cylinder. About 62% diesel fuel was replaced during dual fueling.

TABLE 1 Engine specifications

Engine type	Lister Petter single- cylinder, DI, water cooled
Bore/Stroke	87.3/110 (mm)
Swept volume	659 (cm^3)
Connecting rod length	231.9 (mm)
Compression ratio	16.5
Injection timing by spill	28°bTDC
Engine speed	1750 rpm

New Zealand low sulfur diesel fuel (detail properties are presented in Supplemental Information), limiting sulfur to a maximum of 50 ppm, was used for the experiments. Natural gas (NG) was obtained from the pipeline supply and the detailed composition is provided in (Mustafi and Raine 2008). Biogas was prepared by mixing NG with CO₂ (and H₂S in one sample) in order to obtain different types of biogas: biogas-1 (80% CH₄ and 20% CO₂); biogas-2 (67% CH₄ and 33% CO₂); biogas-3A (59% CH₄, 41% CO₂ and about 820 ppm H₂S), and biogas-3B (58% CH₄ and 42% CO₂).

PM Sampling

PM Samples were collected for electron microscopy analysis using a partial flow dilution system with a dilution ratio of ∼10:1; further details of the system are provided in (Mustafi and Raine 2008). Collected sample filters were preserved in Petri dishes and sealed carefully until microscopy analysis. The sampling time was kept short to avoid overlapping of PM on filters, which restricts proper image analysis.

SEM Analysis

For SEM analysis, PM samples were collected on 70 mm Isopore^TM polycarbonate membrane filters with 0.4 μm pore sizes (Mustafi et al. 2007). These filters have low contrast and very smooth surfaces, which make them suitable for SEM analysis (Hinds 1999). Samples were cut into small squares (approximately 5 mm × 5 mm) from different parts of the loaded filters and coated and examined using a Phillips XL-30S Field Emission Gun. The SEM parameters maintained during the examination were a small accelerating voltage of the electron beam: 5 kV (Chakrabarty et al. 2006). At least 30 SEM images were recorded for each PM sample in this study.

TEM Analysis

Impaction sampling was used in the study where the TEM grids holder (a thin perforated strip of “post-it”) is attached onto a gravimetric filter through which the diluted exhaust gas passes in the existing partial flow dilution tunnel system (Mathis 2005). A CM12 TEM (Philips, FEI Company, Netherlands) was used to examine the PM samples and was operated at an accelerating voltage of 120 kV. TEM images were observed and digitized with the associated image acquisition system equipped with a Model 792 Bioscan digital camera (Gatan Inc., USA) and stored as 1024 × 1024 pixel computer images. Higher magnifications (88,000× and 110,000×) were used for primary particle observation in agglomerates and lower magnifications (19,000× and 25,000×) were used for PM agglomerates observation collected on TEM grids. Sufficient images were recorded at both lower and higher magnifications to count about 200 PM agglomerates and at least 300 primary particles respectively for any type of PM sample.

Image Analysis

Images obtained from SEM and TEM examinations were analyzed using public domain image processing software ImageJ, Version 1.38l (Rasband 1997–2007). ImageJ yielded count, measurement of projected areas (A _a), perimeters (P), projected area equivalent diameters (D _p), fitted best-fit ellipses and measurement of maximum projected length (L _max), and width (W _max) of the PM agglomerates. Figure 1 presents how the primary particles are selected and measured randomly from a TEM image. It also provided the means to determine the primary particle diameter (d _p) from an agglomerate.

FIG. 1 TEM micrograph showing the measurements of primary particle area in an agglomerate and the maximum projected length (L_max) and width (W_max) of the agglomerate.

RESULTS AND DISCUSSION

Distribution of Primary Particle Diameters

From the TEM micrographs, primary particles with distinguishable boundaries in the PM agglomerates were selected randomly and their diameters were measured. Results are presented for different engine operating conditions in Figure 2. The average primary particle diameters measured here for different fueling, range from 25.9 to 29.5 nm, which are in the range of those for light duty diesel particulates as obtained by Zhu et al. (2005). It can be observed that all the primary particles have mean diameters less than 30 nm and the differences in diameters obtained for different engine fueling conditions are not great. However, there is a trend that as the engine load increased during diesel fueling, the mean primary particle diameter ( _p ) decreased from 28 nm (Figure 2a) to 26.4 nm (Figure 2b). A similar trend was reported by Lee et al. (2003) and Zhu et al. (2005) for light duty diesel PM.

FIG. 2 Primary particle size distributions at different engine fueling conditions: (a) diesel low load (3 nm); (b) diesel high load (28 nm); (c) diesel-NG (28 nm); (d) diesel-BG1 (80% CH₄ and 20% CO₂) (28 nm); (e) diesel-BG2 (67% CH₄ and 33% CO₂) (28 nm); (f) diesel-BG3B (58% CH₄ and 42% CO₂) (28 nm); and (g) diesel-BG3A (59% CH₄, 41% CO₂ and about 820 ppm H₂S) (28 nm).

The difference in _p values between diesel low and high load conditions can be attributed to mainly the difference in combustion temperatures in those two cases. The existence of a primary particle and its subsequent growths in the combustion field depends on the competition between nucleation and oxidation. Therefore, combustion temperature can be considered as the most important parameter affecting particulate formation as both particulate nucleation and oxidation rates are a function of temperature (Zhu et al. 2005). At high temperature (for high load), particularly, the oxidation process is much enhanced which would cause the size of the primary particles to decrease (Jung et al. 2004). In other studies (Lee et al. 2002; Zhu et al. 2005), it was found that the decreasing size trend started above the exhaust temperature of 400°C. In this study, the exhaust temperature varied from 445°C to 485°C for diesel (high load) and dual fueling while that for diesel (low load) was about 200°C. The effect of combustion temperature is also noticed when the value of _p is compared between diesel low load (Figure 2a) and diesel-NG fueling (Figure 2c). The amount of diesel fuel injected into the cylinder in both cases is nearly the same but because of high combustion temperature during NG fueling, the oxidation rate is higher than that for diesel low load and thus the _p decreases. The difference in _p values between diesel high load and dual fuel conditions can be attributed to the combined effects of combustion temperature and duration of combustion. Like high combustion temperature, longer combustion duration can also cause a reduction of the primary particle diameter. Both the exhaust temperature and combustion duration for dual fueling were found to be lower and shorter than those for diesel (high load) fueling condition. Thus the _p value for dual fueling (Figure 2c, Figure 2d, Figure 2e, Figure 2f, Figure 2g) falls between those for diesel low and diesel high load conditions.

When comparing results of _p between NG (Figure 2c) and different biogas fueling (Figure 2d, Figure 2e, Figure 2f), an increasing _p is noticed with the increase in CO₂ content in biogas. As the amount of diluent, CO₂, increases in biogas, _p increases from 27.3 nm (Figure 2d) to 29.5 nm (Figure 2f), which agrees with the results obtained by Lee et al. 2002. It is likely that the maximum flame temperature is lowered due to the presence of high CO₂ in the fuel. On the other hand, as the amount of diluent increases in the fuel, the amount of available oxygen in the cylinder also decreases and this would have negative impact on net oxidation rate throughout the combustion cycle as compared to NG fueling. Thus the higher value of _p for higher CO₂ containing biogases can be speculated as the effect of flame temperature due to diluent and oxygen content.

For biogas3A (with H₂S), the value of _p is found to be decreased to 25.9 nm, which is small compared to other biogas samples. In the case of biogas3A, the presence of H₂S could enhance particulate nucleation when the exhaust cools. Sulfur and organic compounds (which are known as the precursors of nuclei) are generally in the vapor phase in the tailpipe, and undergo gas-to-particle conversion during dilution and cooling (Kittelson et al. 2006). These nuclei mode particles then undergo coagulation processes and form PM agglomerates which thus contain relatively lower size primary particles. Thus the primary particles for biogas3A may have a lower average primary particle diameter, _p than other biogas samples due to the contribution of these PM agglomerates.

Number Size Distribution

SEM images of PM agglomerates were analyzed to yield the projected area equivalent diameter (D _P), which is defined as the diameter of a circle with the same area as the projected agglomerate. In order to get a statistical representation of PM appearance on SEM filters, large numbers of particulates (more than 500 from about 15 different images) have been analyzed for every fueling condition. Figure 3 shows the measured PM (agglomerates) number size distribution with fitted bimodal lognormal distributions for different engine operating conditions. It can be observed from Figure 3 that the size of PM agglomerates measured on SEM filters has a bi-modal number size distribution irrespective of type of fueling. All the measured PM on the SEM filters appear to be in the range of fine particles, D _P < 2.5 μm, which are composed of ultrafine particles, D _P < 0.1 μm and nanoparticles D _P < 0.05 μm (Kittelson 1998).

FIG. 3 Number size distribution with fitted bimodal lognormal distribution of the PM agglomerates measured for different engine operating conditions. CMD = count median diameter; GSD = geometric standard deviation.

It is observed from Figure 3 that PM agglomerates size measured on SEM filters has mainly 2 modes of distributions: nuclei (D _P less than 0.1 μm) and accumulation mode (0.1 < D _P < 1.0 μm) except the diesel high load condition where PM size modes are found to be an accumulation and an almost coarse mode (D _P larger than 1.0 μm). Kittelson (1998) reported a higher peak for the nuclei mode or nanoparticles for diesel PM than the peak for accumulation mode. This can be attributed to the completely different technique that has been used in the study and also the pore size (0.4 μm) of the polycarbonate filters used for PM sampling, which has limitations to capture all the nuclei mode particles. However, no distinguishable nuclei mode particles are observed here for diesel (high load) PM. This can be attributed to the fact that a smaller amount of soluble organic fraction (SOF) or volatile fraction (VF) is produced during diesel high load condition as compared to either diesel low load or dual fueling (Mustafi 2008), which contributes to the nuclei mode particles in the exhaust. On the other hand, no distinguishable coarse mode of the measured PM is observed for either diesel low load or dual fueling.

The first peaks (nuclei mode) are found to appear at D _P = 0.08 μm for diesel low load; 0.05 μm for diesel-NG; and for diesel-BG1 and diesel-BG2 and 0.07 μm for diesel-biogas3A fueling. In summary, the median diameter for dual fueling in nuclei mode region is found to be about 0.05 μm. The small shift in the case of biogas3 may be due to the extra adsorption and condensation that could happen with the formation of H₂SO₄ in the exhaust.

The second peak (accumulation mode) of the lognormal fits in Figure 3 are observed at D _P = 0.16 μm to 0.14 μm for diesel low load and dual fuel operation. But for diesel high load condition, the peak of the accumulation mode is observed at D _P = 0.28 μm. This indicates that the PM agglomerates measured on SEM filters have significantly larger median diameter for diesel (high load) fueling compared to other fueling conditions for the same mode of particles. Figler et al. (1996) obtained by SEM analysis the peak mean diameter for diesel particle size distribution at 0.25 ± 0.03 μm, in good agreement with the present observation. Lapuerta et al. (2003) also measured mean diameter of the emitted particulates for a diesel engine at about 0.23 μm at 70% of full load condition by SEM analysis. From the above discussions it is clear that the number of smaller particulates observed on SEM filter images is more in the case of dual fueling compared to that for diesel fueling.

Shape Analysis of the Agglomerates

The shape factor, SF, is used to define the external shape of PM measured on SEM filters. This shape descriptor is calculated from the minor and major axes of the best-fitted ellipses around PM agglomerates (Nord et al. 2004):

SF is sensitive to particulate elongation. SF close to 1.0, indicates that the particles are nearly spherical and SF close to 0.1 indicates that the particles are long-chained or elongated (Nord et al. 2004).

Figure 4 presents the shape factor of the measured PM for diesel and dual fueling. Comparing the results between diesel high load and dual fuel PM, a significant difference in the nature of the SF curves is noted. In the case of dual fuel operations, all the peaks are close to 0.7, compared to diesel (high load) at about 0.5. This indicates that the PM measured in the case of dual fuel operations are more nearly spherical compared to diesel (high load) fueling. The peak of diesel-biogas3A is at about 0.65 indicating more irregular shaped particulates compared to the other dual fuel PM samples.

FIG. 4 Shape factor (SF) trends of the PM collected on SEM filters for diesel and dual fueling (1750 rpm, and pilot = 0.6 kg/h for dual fueling).

From the thermogravimetry analysis (Mustafi 2008), it is observed that the PM in the case of diesel (low load) and dual fueling are composed of much higher volatile fractions compared to diesel (high load) PM. When the exhaust is diluted and cooled these materials may cause nucleation and can form a large number of nuclei mode (0.005–0.05 μm). These particles lead to large number concentrations, but with insignificant contribution to either volume or mass concentrations (Kittelson 1998). This has already been observed in the case of PM mass measurements where PM mass emissions for diesel low load and for all the dual fueling were found to be significantly lower compared to those for diesel high load condition (Mustafi and Raine 2008). These nuclei mode particles are then either adsorbed onto the soot surfaces or collide together to form a particulate cluster. As the availability of solid soot particles is less at these engine-operating conditions, large numbers of small, nearly spherical, particulate clusters as well as individual particles composed of soluble organic or volatile organic materials are obtained. On the other hand, at high engine load, the exhaust temperature is high and the SOF in the exhaust decreases as it is more effectively oxidized (Kittelson 1998).

In the coagulation processes, spherical solid soot particles collide with each other and coalesce and form a larger primary spherical particle. When the rate of particle growth slows down, continued collision between the spherical primary particles results in agglomeration to form large clusters of primary particles, which appear to be chain-like. The rate of agglomeration is proportional to the square of the primary particle number density (Smith 1981). The major source of PM formation can be attributed to the diesel fuel injected into the cylinder. The use of gas dual fueling is thus expected to produce little PM compared to the diesel (high load) as diesel injection is minimized. As the number density of the primary particles formed in the case of diesel high load operation will be much higher than that in the case of the diesel low load or dual fueling condition, the corresponding rate of production of agglomerates is also higher in the first case.

During dilution and cooling of the exhaust, organic materials adsorbing and condensing onto the agglomerated particles further increase their physical size and enhance the formation of larger irregular shaped chain-like agglomerates. The rate of adsorption and condensation is also higher in this case as the availability of the soot surfaces is higher. On the other hand, the exhaust temperature for dual fueling remains sufficiently high (∼ 450°C) Therefore, if any large agglomerates form during the combustion cycle, these would be oxidized or broken down to smaller sizes later in the course of the PM growth phases. However, due to high temperatures the emissions of soluble organic or volatile organic materials in the exhaust are lower compared to that in diesel low load condition and thus the peaks of shape factor (SF) for dual fueling lie in between the diesel high and low load conditions.

Fractal Morphologies of the Agglomerates

Although the primary particles are nearly spherical, the sizes and shapes of agglomerates vary significantly. The complex morphology of the agglomerates can be characterized as fractals. PM fractal dimensions are calculated in this study for the different engine operating conditions. The necessary formula and the associated assumptions used in calculation are presented in Supplemental Information. Fractal dimension based on maximum projected length (D _fL) is used in this study instead of fractal dimension based on radius of gyration (D _f). The two fractal dimensions, D _fL and D _f, have identical values and can be used interchangeably (Chakrabarty et al. 2006).

Figure 5 shows an example of the results as a plot of the number of primary particles in an agglomerate, N versus L _max/ _p on logarithmic scales (where fractal dimension D _fL, is represented by the slope). Table 2 presents the fractal dimensions for the PM measured for different engine operating conditions. It can be observed that the fractal dimension values (D _fL) are in the range 1.69 to 1.88 among which 1.72 is for diesel (low load) and 1.69 for diesel (high load) fueling. Lee et al. (2003) and Zhu et al. (2005) measured fractal dimensions of sampled PM in the range of 1.46–1.70 for a light duty diesel engine while the engine was operated at different low to high speeds and loads. At speed 1000 rpm, D _f = 1.52 and 1.48 was obtained for 25% and 75% engine loads, respectively, and at 2500 rpm 1.70 and 1.59 for 10% and 70% loads, respectively. These values are relatively lower than the values obtained in the present study. The main reason is that their calculation did not consider the primary particles overlapping which is an obvious case in any agglomerate. Primary particles overlapping were considered in this study to calculate the fractal dimension (see the Supplemental Information) as suggested by Oh and Sorensen (1997). Besides this, the engine and the operating conditions were quite different compared to the present study. However, a similar trend in values is obtained which are a higher fractal dimension at low load and a relatively lower fractal dimension at high load condition. When comparing the fractal dimension between diesel and dual fueling (Table 2), a higher value (1.73 to 1.88) is always obtained for the latter case. No published literature has been found to compare with our results obtained for dual fuel PM samples. However, several articles are found where morphology was studied on PM obtained from premixed methane/oxygen flames. A summary of the previous results along with the present work is given in Table 3, where it is seen that the fractal dimensions obtained by light scattering (LS) and TEM methods of PM sampled for methane/oxygen premixed flame are in the range of 1.6–1.82. The value obtained in the present work is therefore in the range of the existing values but for methane/oxygen premixed flame instead of a dual fuel engine environment. The highest fractal dimension (1.88) is obtained for biogas3A (with H₂S) fueling.

TABLE 2 Fractal dimensions of the PM sampled for different engine fueling conditions

Engine fueling	Fractal dimension, D fL
Diesel low load	1.72
Diesel high load	1.69
Diesel-NG	1.73
Diesel-BG2 (67% CH4 and 33% CO2)	1.80
Diesel-BG3B (58% CH4 and 42% CO2)	1.84
Diesel-BG3A (59% CH4, 41% CO2 and about 820 ppm H2S)	1.88

TABLE 3 Fractal dimension measured for the PM sampled from CH₄/O₂ premixed flames and diesel-NG dual fueling in diesel engine.

Investigators	PM source	Method	Fractal dimension
Zhang et al. (1988)	CH4/O2 premixed flame	TEM	1.72 ± 0.10
		LS	1.61 ± 0.06
Gangopadhyay et al. (1991)	CH4/O2 premixed flame	LS	1.6–1.8
Sorensen et al. (1992)	CH4/O2 premixed flame	LS	1.70–1.79
Cai et al. (1993)	CH4/O2 premixed flame	TEM	1.70
Cai et al. (1995)	CH4/O2 premixed flame	TEM	1.74
This work (diesel engine)	Diesel-NG dual fueling	TEM	1.73
	Diesel-biogas dual fueling	TEM	1.8–1.88

FIG. 5 Plot of fractal data of PM agglomerates sampled for diesel low load operation (best fit with D _fL = 1.72). The number of primary particles per agglomerate N is correlated with the fractal dimension as, N = k _L(L _max/

_P ) ^{D _fL} ; While the fractal dimension, D _fL represents the slope, k _L (a correlation prefactor) determines the magnitude of the least-squares linear fit to the data in the ln (N) versus ln (L _max/

_p ) plot.

Fractal agglomerates grow through either primary particle-cluster (PC) or cluster-cluster (CC) collisions in three different growth mechanisms: reaction-limited, ballistic, and diffusion-limited (Schaefer and Hurd 1990; Lee et al. 2002). In real combustion processes, particle agglomerates are likely to grow through either ballistic or diffusion-limited mechanisms with both PC and CC type collisions. It can be speculated that at high engine load conditions, the particulate agglomeration growth is dominated by the diffusion-limited mechanism since the mean free path is generally smaller than the particulate sizes. On the other hand, at low load condition, the growth may be dominated by the ballistic mechanism and therefore forms agglomerates, which are more spherical in shape. Typically, smaller fractal dimensions indicate more chainlike and larger fractal dimensions indicate more spherical particulate agglomerates (Lee et al. 2002). In the present work, however, the mean free path of particulates was not calculated and it was assumed that the mean free path for dual fuel PM would be of similar range of diesel (low load) PM as the parameter mainly depends on the number concentrations of the particulates (Bird 1960). It has been observed that the PM mass concentration is almost one-third for dual fueling compared to diesel (high load) fueling (Mustafi and Raine 2008). Thus the particulate growth in the case of dual fuel PM may also be dominated by the ballistic mechanism and results in more spherical shaped agglomerates than diesel high load conditions. The fractal dimensions obtained for dual fueling (1.73–1.88) are higher than the diesel (high load) fueling (1.69) indicating more chainlike agglomerates in the latter case.

CONCLUSIONS

Measurements have been made of the physical and morphological characteristics of PM emitted by a diesel engine operated on diesel fuel and in the dual fuel mode using natural gas and simulated biogas. Based on these measurements the following conclusions can be drawn:

•	The average primary particle diameters measured for different fueling, range from 25.9 to 29.5 nm. As the engine load increased from low to high during diesel fueling, the mean primary particle diameter ( p ) decreased from 28 nm to 26.4 nm. This is mainly due to the higher combustion temperature in the latter case. The p value for dual fueling falls between those for diesel low load and diesel high load conditions. This can be attributed to the combined effects of low combustion temperature and short combustion duration in the case of dual fueling. However, for biogas fueling p increases from 27.0 to 29.5 nm as CO2 content increases in biogas.
•	All the measured PM agglomerates have a bi-modal number size distribution irrespective of type of fueling. The size of PM agglomerates measured on SEM filters has mainly 2 modes of distributions: nuclei (D P less than 0.1 μm) and accumulation mode (0.1 < D P < 1.0 μm) except the diesel high load condition where PM size modes are found to be an accumulation and an almost coarse mode (D P larger than 1.0 μm).
•	According to shape factor analysis, the PM agglomerates measured in the case of dual fuel operations are more nearly spherical compared to diesel (high load) fueling.
•	Fractal dimensions of the PM agglomerates measured by TEM (based on maximum projected length, D fL) are found to be in the range of a light duty diesel engine (1.69 to 1.88). For diesel PM, a higher fractal dimension at low load (1.72) is obtained than at high load condition (1.69). However, D fL is always higher (1.73 to 1.88) for dual fuel PM compared to diesel (high load) PM, indicating more chainlike agglomerates in the latter case. The particulate agglomeration growth in the case of diesel fueling may be dominated by diffusion-limited mechanism while that for dual fueling may be dominated by the ballistic mechanism.
•	These measurements have given improved understanding into the formation, coagulation and oxidation processes of PM under diesel-like combustion conditions.

Acknowledgments

The authors sincerely thank the following from around the University of Auckland: Alan Eaton and Martin Ryder (Thermodynamics Laboratory), Stephen Elder (EFRU), Jacqueline Ross (Biomedical Imaging Research Unit); Dr. Bryony James, and Catherine Hobbis (Research Center for Surface and Materials Science); Dr. Adrian Turner (Microscopy and Graphics Unit). In addition, the financial support provided by the University of Auckland International Doctoral Scholarship is gratefully acknowledged.

[Supplementary materials are available for this article. Go to the publisher's online edition of Aerosol Science and Technology to view the free supplementary file.]