Hydrocarbon Nanoparticles Formed in Flames and Diesel Engines

Abstract

The formation of nanoparticles in laboratory hydrocarbon flames is reviewed in terms of particle morphology, chemical composition, and health hazards. The nascent nanoparticles in the nucleation mode have been widely reported in diverse laboratory flames, and are distinguished by their occurrence as singlet particles that form translucent images in transmission electron microscopy (TEM). Their sizes range from about 10 nm or more down to 2 to 3 nm, the limit of resolution of the TEM, and they possess a liquid-like quality. These particles are widely considered to be the precursor stage to the more readily observed carbonaceous aggregates consisting of chained primary particles that are opaque to the electron beam of the TEM. Nanoparticles sampled from the inverse diffusion flame and the particle effluent from diesel engines show a strong resemblance by GCMS analysis, and they contain many of the stabilomer PAHs and their isomers in the 200 to 302 atomic mass range. Many of these chemical species have high relative mutagenicities. Distinctive bimodal particle size distributions can be observed in both flame and engine samples. Recent TEM micrographs of diesel particulates show images of precursor-like nanoparticles, of as yet unknown chemical composition, that are formed in a diesel engine at many operating conditions.

1. INTRODUCTION

Interest in the very smallest particles produced in diesel engines as a potential source of particulate pollution results from health concerns. These particles, referred to as nanoparticles or ultrafine soot particles, are of very small sizes that impede their detection and analysis by the early instrumentation used in monitoring particulate air pollution. Laboratory studies of hydrocarbon flames where diagnostic methods are more readily applied have made marked progress in understanding the origin and fates of the smallest hydrocarbon particles. The results of these studies are pertinent to particle formation in internal combustion engines and in diesel engines in particular. The diesel engine in automotive transport is a potential intermediate term solution for the reduction of both fuel consumption and carbon emissions. A deterrent to this development relates in part to the concern of the health impact of diesel particulate emissions.

Micrographs of very small singlet particles in flames were first reported in the early 1990s, and they have been extensively studied since that time using a variety of experimental methods. These young particles are transparent to the electron beam of the transmission electron microscope (TEM) as shown in Figure 1a, possess a liquid-like consistency, and they apparently grow by coalescent collisions as well as by molecular impingement of growth species. The sizes of the singlet nanoparticles in Figure 1a are about 10 nm down to the limit of resolution of the TEM, near 2 nm. Many terms are used to describe these young particles (e.g., tar, ultrafine particles, nanoparticles, precursor particles; see discussion by Kittelson 1998).

FIG. 1 Particles sampled From Axial Heights of (a) Z = 20 mm and (b) Z = 50 mm of the Laminar Ethene Diffusion Flame. The isolated translucent precursor particles dominate the lower axial region of the flame while opaque aggregates occupy the upper region of the flame.

Exposure of the nanoparticles to elevated temperatures promotes dehydrogenation reactions and the formation of solid particles having higher carbon content and the aggregates that are shown in Figure 1b. These aggregates which are formed by non-coalescent collisions have been more widely sampled and photographed by electron microscopy. The occurrence of the two distinctive morphologies and chemical constituencies of soot particles has complicated their sampling, measurement and characterization both in laboratory flames and in engine tests as well.

2. LABORATORY STUDIES OF PARTICLE FORMATION IN HYDROCARBON FUELS

Several decades of research on particle formation in laboratory flames are summarized in this section with emphasis on the results that may have relevance to studies of engine combustion.

2.1. Precursor Nanoparticles

The existence of two types of particles in laboratory flames is now well documented. Precursor nanoparticles are defined by their physical appearance in TEM as singlet particles of spheroidal shape. They normally range in size from about 10 nm down to the resolution limit of electron microscopy, viz., a few nanometers. They contain large fractions polycyclic aromatic hydrocarbon (PAH) species, and have a C/H ratio of about two. They have been widely depicted in micrographs obtained by ex-situ sampling from flames where they appear as solitary, sometimes faint particles that are semi-transparent to the electron beam but they may display dark edges or dark internal regions. These nanoparticles have been shown in micrographs of samples from premixed flames (Wersborg et al. 1973; Onichuk et al. 2003; Sgro et al. 2003; Öktem et al. 2005), in numerous laminar diffusion flames (Megaridis et al. 1989; Dobbins et al. 1994, 1996, 1998; Dobbins, 1997; Vander Wal, 1996, 1997; Köylü et al. 1997; Lee et al. 2000); in turbulent flames (Hu et al. 2003, 2004; Yang et al. 2005); in unsteady, flickering flames (Zhang and Megaridis 1998); in bituminous coal flames (Ma et al. 1995); in microgravity flames (Konsur et al. 1999); in inverse diffusion flames (Blevins et al. 2002; Lee et al. 2005; Oh et al. 2005); in a two meter pool fire (Jensen et al. 2005); in a well stirred reactor (Blevins et al. 2003); and in an opposed flow flame (Merchan-Merchan et al. 2003). Additional details concerning these studies are summarized in Table A1 of Appendix A.

Important conclusions to be noted from the above body of research are listed in the following section.

Interparticle collisions at the low heights in flames where the nanoparticle concentrations are highest show no evidence of aggregate formation. Hence, these collisions are coalescent in nature and preserve the isolated singlet morphology of the nanoparticles, which possess a liquid-like consistency.

Micrographs showing nanoparticle captured on the substrate borders of lacey carbon grids (Figure 2, Z = 20, 30 mm flame heights) display surface tension effects that are also characteristic of liquid-like droplets. The opaque aggregates captured from the upper flame Z = 40 mm do not display surface tension effects and appear as solid particles.

FIG. 2 Particles in Transition from Precursor Nanoparticles to Carbonaceous Aggregates Captured on Lacey Carbon Grids at Heights Z = 20, 30, and 40 mm above the Burner With C₂H₄ Fuel. (Reprinted by permission of Elsevier. From Dobbins et al. (1998) p. 286, Vol. 115, Combustion and Flame, Combustion Institute).

Mass spectroscopic analyses (Dobbins et al. 1998) of nanoparticles sampled from an ethene diffusion flame showed the prominence of the benzenoid polycyclic aromatic hydrocarbons (PAHs) in the mass range of 202 u to 300 u. Many of these masses correspond to stabilomer species or their isomers that were previously found by thermodynamic analysis to be the most stable PAHs at temperatures typical of hydrocarbon flames (Stein and Fahr 1985). The discovery of these stabilomer species accounts for their presence on the particulate product of a wide range of hydrocarbon fuels (Dobbins et al. 1998).

The GCMS data (Blevins et al. 2002) was made possible by the use of probe sampling of an inverse diffusion flame and provided isomer identification of the numerous PAHs in the 152 to 302 u mass range. These results both independently confirm and significantly extend the earlier mass spectroscopy data. Furthermore, the data of Blevins et al. (2002) identifies the presence of certain PAH compounds that have been found to be genotoxic in the literature discussed below.

In Figure 3 the nanoparticles are shown to coexist with aggregates in the upper flame regions of buoyancy dominated flames burning CH₄ or C₂H₂ as fuels. The more recent studies of turbulent flames by Koylu and colleagues (Hu et al. 2003, 2004; Yang et al. 2005) show the coexistence of the singlet spheroids and the carbonaceous aggregates and suggest that distributed particle nucleation occurs in these fuel-rich flames even in the presence of the larger soot aggregates. In this event, the particle size distribution (PSD) is clearly bimodal as has been found by the direct observations of laminar premixed flames (Zhao et al. 2003; Öktem et al. 2005).

FIG. 3 Particles Sampled from the Upper Buoyancy Dominated Flames for (a) Methane and (b) Acetylene Fuels. The coexistence of precursor particles and soot aggregates is apparent. (Reprinted by permission of Taylor and Francis Books. From R. A. Dobbins (1997) p. 112 of Physical and Chemical Aspects of Combustion.)

Comparison of the measured PSD and the micrographs shown by (Öktem et al. 2005) indicates the nanoparticles in their Figure 6(a) are associated with the nucleation mode (d_p < 10 nm) while the young agglutinated aggregates in Figure 6b are associated with the accumulation mode (d_p∼ 30 nm). Thus a rational linkage between the particle morphologies and the bimodal PSD is apparent.

The growth of the soot aggregates by the scavenging action of the latter particle mode with the PAH laden young nucleation mode particles can contribute to the detection of PAHs on samples of the aggregates in the accumulation mode. The interaction of fine and coarse modes in bimodal PSD has been described in the literature (Friedlander et al. 1991).

The ubiquity of the precursor nanoparticles in diverse flame geometries, combustion modes, and for various gaseous, liquid, and solid hydrocarbon fuels implicates them to be a universal early stage particle that precedes the formation of the more widely recognized soot aggregate clusters. Precursor nanoparticles are difficult to detect because of their small sizes and their near transparency to the electron beam. In laminar flames precursor nanoparticles are consumed in the formation and growth of aggregates.

Optical methods have provided early detection and quantitative data on the youngest nanoparticles formed in flames. Non absorbing particles that displayed fluorescence to uv radiation were found low in a C₂H₄/O₂ flame where their number densities were order of 10^{+ 13} cm^{− 3} and their sizes to be 3–4 nm (D'Alessio et al. 1992; D'Anna et al. 1994). Two to four ring PAHs were considered to be the source of the fluorescence displayed by the first soot nuclei whose material densities were estimated to be 1.2 gm/ml. Optical observations and sampling methods followed by atomic force microscopy and differential mobility analysis have been used to identify 2 to 4 nm nanoparticles in flames and from practical combustion systems as well (Sgro et al. 2003). This study also showed isolated nanoparticles at Z = 5 mm and a bimodal PSD at Z = 10 mm in the premixed C₂H₄ flame.

2.2. Carbonaceous Aggregates

Soot aggregates consist of filose chains of primary particles whose properties as mass fractals have been widely explored. Micrographs of these particles present a more opaque, black material, and they are shown by several investigations to undergo a serial transformation from precursor particles to aggregates (e.g., Dobbins 1997; Köylü et al. 1997; Vander Wal 1997). Aggregated particles have high elemental carbon content and may have molecular species absorbed on external or interstitial surfaces. Primary particle sizes generally range from 20 to 50 nm, and aggregates consist of chains of ten to hundreds of such primary particles. Smoke and carbon blacks consist of aggregates that are closely related to those released from flames. The densities and optical properties are reasonably well established, and they are therefore well suited to optical and other methods of measurement. Soot aggregates are the seeming universal products of combustion of a wide variety of flammable hydrocarbon containing materials such as biomass, coal, and gaseous or liquid petroleum products.

2.2.1. Fractal Properties of Soot Aggregates

Soot aggregates are fractal-like and comply with the relationship

where N is the number of particles per aggregate, R_g is the aggregate radius of gyration, d_p is the primary particle diameter, D_f is the fractal dimension, and k_f is the prefactor. The properties of the aggregate morphology as mass fractals have been investigated (e.g., Köylü et al. 1995; Sorensen and Feke 1996). The fractal dimensions of aggregates produced in a wide variety of flames have been more recently summarized (Hu and Koylu 2004) who find a near universal result that D_f = 1.82 ± 0.06 for their turbulent flames as well as for earlier studies of laminar flames. The broader variation of the reported values of the prefactor k_f is frequently noted. The experimentally observed value of D_f is in agreement of with theoretical predictions of aggregate formation by diffusion-limited cluster aggregation. The acetylene flame was found (Sorensen and Feke 1996) to produce aggregates with the number of primary particles per aggregate ranging over a seven orders of magnitude yet displaying the same fractal dimension D_f of 1.8. The narrow range of D_f and a relative uniformity of d_p in aggregates from a common point of origin within a given flame facilitate the use of light scattering methods that have been reviewed (Sorensen 2001).

Suction probe sampling has been used to detect soot and condensed hydrocarbon species for subsequent analysis. In a methane diffusion flame, two to four ring PAHs were identified by GCMS and five to ten ring species were found by liquid chromatography (Feitelberg et al. 1993). Table 1 summarizes the current estimates of the properties of the two contrasting particle morphologies found in hydrocarbon flames.

TABLE 1 Estimated physical properties of combustion particles from Hydrocarbon fuels

	Precursor nanoparticles	Carbonaceous soot
Morphology	Singlet spherules, Fig. 1a etc.	Chained fractal aggregates, Fig. 1b etc.
C/H Ratio	1.8 to 2.2 (2)	5 to 10 (2)
Composition	PAHs (1, 2), Aliphatics (4)	C Crystallites, Residual H
Mobility diameter	2 to 10 nm (1, 4)	0.10 to 10 μ m (5)
Material density (gm/ml)	∼ 1.2 (1)	1.77 to 1.8 (3)
Refractive index (visible light)	∼1.6 + i0.05 (1)	Various, e.g., 1.57 + i0.56, 1.9 + i0.55, etc. (6)
References: (1) D'Alessio et al. 1992; (2) Blevins et al. 2002; (3) Park et al. 2004; (4) Öktem et al. 2005; (5) Sorensen et al. 1996; (6) Sorensen, 2001.

2.3. Transformation of Precursor Particles to Aggregates

The experience of the laboratory studies with a variety of combustion modes and devices suggests that the precursor particles are subject to several possible fates. In the laminar flame they are formed early in the combustion process and are transformed to carbonaceous aggregates or may contribute to the further surface growth of the aggregates. Precursor nanoparticles in laminar flames often appear spatially separated from the larger aggregates by a group of ill-defined, agglutinated particles (see Figure 6b, Figure 6c in Dobbins, 1997; Figure 9b in Öktem et al. 2005) suggesting particles in a transition stage between the two morphologies described in Table 1. However, micrographs from turbulent diffusion flames (Hu et al. 2003, 2004; Yang et al. 2005) clearly show precursor nanoparticles in coexistence with mature aggregates. These images represent concrete evidence of a bimodal PSD that has been observed (Zhao et al. 2003) in the premixed laminar flame using a scanning mobility particle sizer.

A striking feature of the micrograph Figure 9 of Öktem et al. is that the singlet nanoparticles found at the 6 mm height are sampled from the flame region where the size distribution given in their Figure 2 was found to be a power law, monomodal PSD descriptive of the nucleation mode. Further, the young aggregate clusters found at the 12 mm flame height correspond to the region where the PSD has developed a clearly bimodal character. This is an important link between particle morphologies and the PSD.

The actual carbonization process wherein precursor nanoparticles are transformed to carbonaceous aggregates has been studied using dark field TEM (DFTEM) which displays bright domains when Bragg diffraction effects are present (Vander Wal 1996, 1997; Chen and Dobbins 2000). The development of bright domains is attributed to the parallel alignment and stacking of planar PAHs that form the crystallites composing the carbonaceous aggregates. The progressive development of bright domains in the nanoparticles is well documented by the transformations shown in the DFTEM micrographs in the above references. Noteworthy are the faint Bragg diffraction effects from the nanoparticles in the lower flame where the presence of planar PAHs has been detected by chemical analysis. As carbonization develops in the upper flame region the bright domains become progressively more intense. (A seeming fortuitous outcome is the widespread TEM observation that primary particles within one aggregate are of a nearly uniform primary diameter even though the number of primaries per aggregate is highly variable. This outcome greatly facilitates the use of optical methods that observe a large number of particles within a small sample volume. Exceptional samples extracted from a two meter JP-8 pool fire [Jensen et al. 2005] display micrographs in which mixed aggregates possessing distinctly different primary sizes are found at various flame locations. These more complex morphologies are here interpreted to be the result of later stage particle inception as a consequence of more complex turbulent flame structures produced in pool fires).

2.4. Chemical Properties of Young and Carbonaceous Soots

Small samples, about 10^{− 12} gm, of precursor nanoparticles were obtained by thermophoretic sampling for analysis by the sensitive laser microprobe mass spectrometry (LMMS) (Dobbins et al. 1998). It was found that the precursor particles that occupy the central axial region of the 88 mm high laminar ethene diffusion flame mainly consisted of the four to eight ring PAHs ranging in mass from 202 to 350 u. These masses correspond to the PAHs previously identified in the oily matter on carbon blacks (Sweitzer and Heller 1956). Further, the LMMS analysis showed that the detected masses generally fell within the boundaries of the stabilomer grid representing the most stable PAHs at typical fuel-rich flame conditions (Stein and Fahr 1985). The presence of the pericondensed (most compact), benzenoid (consisting solely of six member rings) PAHs at the centerline of the stabilomer grid is noted. The lighter PAHs were not detected in the above study presumably because of their high vapor pressures at the flame temperatures from which they were extracted.

Figure 4 shows a comparison of the mass spectra obtained by LMMS of particles sampled from an ethene diffusion flame and from a truck tunnel. The common mass sequence corresponds to the stabilomer PAH masses that are detected in both the flame burning pure gaseous hydrocarbon and the truck tunnel sample originating from diesel fuels.

FIG. 4 Mass Spectra of Particles Sampled from an Ethene Diffusion Flame and from the Ft. McHenry (truck) Tunnel. Spectra were obtained using laser microprobe mass spectrometry (Fletcher et al. 2004).

The more detailed chemical analysis of the nanoparticles (Blevins et al. 2002) was obtained by using suction probe sampling of the inverse diffusion flame (IDF). These large samples provided isomer specific identification of the various PAHs in this flame by the use of GCMS. The results are shown in Table 2, which illustrates the several PAH groups of isomers at 202 u, 226 u, 252 u, 276 u, and 300 u to be among the more prominent. The presence of these stabilomer PAHs is confirmed in these more detailed analytical results which support the earlier thermodynamic prediction (Stein and Fahr 1985). The benzenoid PAHs of the IDF flame burning a pure gaseous fuel are in the mass range of 152 to 300 u. They originate in the combustion process per se and are therefore referred to as pyrogenic PAHs.

TABLE 2 PAHs, Mutagenicities, and GCMS yields for diesel and IDF particulates

				PAH yields
Mass (u)	Formula	PAH compound	Rel. Mtgnty BaP = 1.0 ^1	Diesel exht ^2	Diesel part. ^3	IDF ^4
152	C12H8	Acenaphthylene	5.6E-4	—	—	2295
154	C12H10	Biphenyl	—	—	—	97
154	C12H10	Acenaphthene	—	—	—	163
166	C13H10	Fluorene	—	—	—	2710
178	C14H10	Anthracene	—	—	1.5	2285
178	C14H10	Phenanthrene	—	—	68.4	5775
190	C15H10	4,5-Methylenephenanthrene	—	—	—	5635
192	C15H12	Methylphenanthrenes	2.5E-3	—	194	1083
202	C16H10	Fluoranthene	—	1426	49.9	7000
202	C16H10	Acephenanthrene	—	—	—	5870
202	C16H10	Aceanthrylene	—	—	—	1880
202	C16H10	Pyrene	—	2220	47.5	8990
226	C18H10	Benzo[ghi]fluoranthene	—	—	12.1	3040
226	C18H10	Cyclopenta[cd]pyrene (CpP)	6.9	—	—	8550
228	C18H12	Benzo[a]anthracene	0.08	3640	6.33	—
228	C18H12	Chrysene & Triphenylene	1.7E-2, 6E-3	4800	26	1102
252	C20H12	Benzo[b]fluoranthene (BbF)	0.25	1396	8.81	1040
252	C20H12	Benzo[k]fluoranthene (BkF)	0.11	500	2.64	660
252	C20H12	Benzo[j]fluoranthene (BjF)	0.26	—	3.52	885
252	C20H12	Benzo[e]pyrene	1.7E-3	—	7.44	790
252	C20H12	Benzo[a]pyrene, (BaP)	1.00	540	1.33	2760
252	C20H12	Perylene	—	—	0.16	235
276	C22H12	Indeno[1,2,3-cd]pyrene(IcdP)	0.31	—	5.62	2300
276	C22H12	Benzo[ghi]perylene (BghiP)	0.19	—	6.50	3150
300	C24H12	Coronene & isomers	—	—	2.0	4731
302	C24H14	17 Isomers	High	—	—	1000

The presence of the stabilomer species suggested in the LMMS data, which is not isomer specific, is unambiguously confirmed by the GCMS data from the IDF studied by Blevins et al. (2002). The targeted PAHs found in the IDF show that eight of the compounds, ranging from the three-ring C₁₂H₈ to the seven ring C₂₄H₁₂ have structures that were specifically identified on the stabilomer grid (Stein and Fahr 1985). This prediction of chemical thermodynamics explains the universality of soot chemistry in a wide variety from flame types and hydrocarbon fuel compositions as has been noted in the past.

The conversion of soot precursor particles to mature soot aggregates in flames is accompanied by an increase of the carbon to hydrogen ratio (C/H) from about 2 to 6 or higher. This carbonization process has been studied by subjecting pure hydrocarbons to elevated temperatures under inert atmospheres. Carbonization (e.g., Lewis and Singer 1988) is described in terms of the formation of activated complexes, molecular rearrangement, polymerization, and evolution of hydrogen. The kinetics of the carbonization of ethene tar pitch was found to be first order. Observations of carbonization of precursor particles (Dobbins et al. 1996) lead to Arrhenius rate constants that were used (Dobbins 2002) to explain the reported soot inception temperatures by three teams of investigators. These temperatures range from 1275 to 1690 K depending upon the sensitivity of the experimental method and the temporal derivative of temperature profile in the various flames that were tested. On the other hand, another study (Basile et al. 2002) concluded that the pyrolysis of precursor particles depends on the coagulation process and is second order in the particle number concentration. Further study of the carbonization process is clearly warranted.

2.5. The Coexistence of Precursor Particles and Carbonaceous Soot

Precursor nanoparticles are often observed low in laminar flames in isolation from the chained aggregates that form in the higher flame as shown in Figure 1a and Figure 1b. Higher fuel flow rates produce a near turbulent or fully turbulent environment where these two particle morphologies can coexist (see Figure 3 above and micrographs of particles in a turbulent flame, Figure 2 of Hu and Koylu 2004). In this case the PSD is bimodal showing both the nucleation mode and the accumulation mode. The experimentally derived particle size distributions (PSD) have clearly shown (Zhoa et al. 2003; Öktem et al. 2005) the evolution of a bimodal PSD as a function of time and height in the laminar premixed flames. The coexistence of these two modes is expected to be the source of growth species for the aggregates in the accumulation mode and to contribute to an outer layer PAH species on the aggregates. Actual escape of the precursor nanoparticles to the surroundings constitutes an alternative fate with undesirable consequences that requires careful consideration.

3. DIESEL ENGINE COMBUSTION

Combustion in diesel engines takes place at elevated pressures (40–100 bars), and temperatures (ca. 1800 to 2500 K), in a highly unsteady mode, and in a turbulent environment (Heywood 1988). However, engine studies reveal that many of the features of the laboratory studies have pertinence to the description of diesel engine combustion. The initial burning takes place as premixed combustion when the fuel is injected into the hot, high-pressure gases near the end of the compression stroke. There intense local mixing promotes ignition under premixed flame conditions. Yet most of the fuel is later consumed in a diffusion flame after molecular and turbulent mixing produce favorable fuel/air mixture ratios. The extensive use of optical diagnostic methods on test engines representative of direct-injection heavy duty engines (Dec 1996) confirm a diffusion flame to be an important component of the mixing controlled combustion. Although the combustion ensues in the interior of the cylinder of a reciprocating engine, the reaction rates are much faster than the mechanical oscillations and the combustion processes, are considered to be quasi-steady. The limited numbers of studies of soot formation at elevated pressures conducted to date indicate that increased soot production results from pressure elevation.

A major difference in the diesel combustion relative to the laboratory studies is the complex composition of the diesel fuel which consists of thousands of C₁₀ to C₂₀ hydrocarbons including straight chain and cyclic paraffins, naphthenes, monocyclics, and smaller fractions of polycyclic aromatic species (Song et al. 2000). The actual composition of the fuel depends on its geographic origin and the detailed refining processes. Several analyses indicate that the two and three ring alkylated naphthalenes and alkylated phenanthrenes are the dominant aromatic components comprising about 90% by weight of the polyaromatics which are predominately limited in mass range to 128 u to 206 u (Dobbins et al. 2006). While heavier PAHs are present in diesel fuels in small quantities, the near absence as fuel components of the four ring and larger species is particularly noteworthy.

3.1. PAHs Found on Diesel Particulates and IDF Nanoparticles

Presented in Figure 5 is a graph of H atoms vs. C atoms for the PAHs of both the SRM 1650a sample of diesel soot and the particles sampled from the IDF of pure ethene C₂H₄. The staircase curve corresponds to the thermodynamically predicted most stable PAHs under typical flame conditions (Stein and Fahr 1985). This curve explains the universality of the PAHs produced in a wide variety of flames burning diverse hydrocarbon fuels (Dobbins et al. 1988). Both of the above analyses employ GCMS that provides isomer specific identification of species contained in the analyte. Since all the PAHs detected in the pure ethene flame are the products of combustion per se, they are said to be of pyrogenic origin. The important distinction between the pyrogenic PAHs and the unburned PAH species contained in the fuel was made in the analysis of burn residues from oil spills (Wang et al. 2000). This distinction has relevance to the PAHs found in diesel emissions.

FIG. 5 The H—C diagram from GCMS analyses of (a) Diesel Particulate, o, (NIST SRM 1650a) and (b) Precursor Nanoparticles From the IDF Flame, ◊, (Blevins et al. 2002). Points common to both data sets are slightly offset in the y coordinate for clarity.

The relative scarcity in diesel fuels of the four ring, 228 u, and larger PAHs supports the conclusion that they are formed in the combustion process in the engine as in the case of the laboratory flames burning pure gaseous hydrocarbons. The prominence in engine exhaust gases of the dimethyl and trimethyl PAHs, which are not found in flames of pure gaseous fuels, suggests their origin is the incomplete combustion of all fuel components. On the other hand, the presence of the pyrogenic PAHs in the diesel exhaust indicates their formation in the engine cylinder even in the presence of complete combustion of all fuel components. A previous analysis of diesel particulates (Dobbins et al. 2006) obtained from earlier engines (ca. 1990) showed both the petrogenic and pyrogenic PAHs to be present. Recognition of the origin of the PAHs emitted from engines may be pertinent to their control.

3.2. Morphology of Engine Particulates

The chemical complexity of diesel fuels and the mechanical design of the diesel engine complicate the capture and the analysis representative particulate samples. Burscher (2005) has discussed diesel particulate properties, sampling, and instrument issues.

Very fine < 10 nm singlet nanoparticles, seemingly formed by nucleation of sulfuric acid from the high sulfur fuel, were detected from a 1 L commercial engine (Shi et al. 2000). Twenty-three not identified PAHs were found on the particle phase by GCMS. TEM micrographs first show faint images of precursor-like nanoparticles at light engine loads (Lee et al. 2002). The recognizable attributes of precursor nanoparticles are singlet particle morphology, a translucent response to the electron beam of the TEM, and a near absence of crystallinity or black opaqueness. Such particles appear very faintly in the TEM images (Lee et al. 2002). Similar images were present in the micrograph from a commercial four-cylinder engine and but were thought to be artifacts of the sampling process (Park et al. 2004). In contrast, a light-duty engine with fuel containing 110 ppm sulfur produced nucleation mode particles at high speeds and were considered to be an actual emission component of the diesel engine (Giechaskiel et al. 2005). Amorphous particles suspected of containing soluble organic compounds at idling conditions were shown in the micrographs of a light duty engine (Zhu et al. 2005). Tests on a light duty engine using 50 ppm sulfur fuel (Neer and Koylu 2006) produced particle images showing both singlet morphologies and larger fractal aggregates. Their micrographs show many singlet particles in the nucleation mode that provide the background of the larger aggregates in the accumulation size mode. These investigators found small clusters of one or a few particles on the TEM grids to be present under many engine conditions. The chemical composition of these particles remains to be investigated.

The recognition of the nanoparticles, which contribute negligibly to the total mass of the particulate effluent, as an adverse health issue has lead to a redefinition of the particle emission standard to be specified in terms the particle number density rather than the mass density (Kennedy 2006). Test results citing compliance with the proposed particle emission limit of 10^{+ 11} particles/km by the use of a diesel particle filter are described.

3.3. Fine Particle Roadway Emissions

A series of roadway studies involving curbside or chase vehicles have provided information relating to the release nanoparticles from spark ignition (SI) and or diesel powered vehicles. Photoelectric charging of nanoparticles has been developed as an efficient means for the detection of PAHs carried by nanoparticles (Zhiqiang et al. 2000). This method was employed to measure nanoparticle pollution in major cities in Europe, North America, and in Asia. Among other observations in recent citations are the prominence of the nanoparticles, < 10 nm, representing about one third of the curbside particle number concentration in UK urban atmospheres (Shi et al. 2001). Bimodal PSDs were noted to decay with downwind distance apparently caused by coagulation and atmospheric dilution in several studies (Zhu et al. 2002a, 2002b). Nucleation mode particles < 50 nm in concentrations as high as 10⁺⁶ ml⁻¹ were noted to be usually composed of all volatile material on-road sampling (Kittleson et al. 2004). High-speed chase tests revealed bimodal PSD in which the nucleation mode was related to the sulfur content of the fuel and vehicle speed (Giechaskiel et al. 2005).

Chemical analyses where performed provide a more definitive link between the nanoparticles found in flames and those found in engine effluents. In Table 2 the composition of particle-associated samples identified several PAHs (Bagley et al. 1996) in the 228 to 252 u mass range that correspond to the toxic compounds detected in the more detailed GCMS analysis of other diesel particulates (NIST SRM 1650a, 2000) and from the nanoparticles from a pure ethene C₂H₄ IDF flame (Blevins et al. 2002). It is noted that the diesel samples listed in Table 2 were not size segregated with respect to the nucleation and the accumulation modes.

3.4. Toxic Properties of PAHs in Engine Emissions

Health effects of the ultrafine particles of aerodynamic diameter d_p < 100 nm in urban air have been discussed (Oberdörster 2001; Oberdörster and Utell 2002). Mass monitoring instruments may fail to detect even moderate quantities of ultrafine nanoparticles which contribute negligibly to the total particulate mass. These particles were noted to have high specific surface area, have higher retention probability in the alveolar region of the human lung, and can thus carry toxic species deep into the lung where translocation to other organs may occur (Oberdörster 2001).

Recent interest on the particles released by diesel engines has directed attention to the nanoparticles, d_p < 50 nm (Kittleson 1998), and ultrafine particles (UFP), d_p < 100 nm (Kleeman et al. 2000). These particles appear to contribute to adverse health effects owing to their high surface-to-volume ratio and their higher likelihood of lung penetration and retention. (The small particles that may be formed by fragmentation or homogeneous condensation in the engine exhaust system or at its exit are not within the purview of the present work.) Limited chemical analyses of diesel particulate and gaseous effluents leads to the conclusion that many of the chemical species are carcinogenic or mutagenic. Surprisingly, tests of a later engine design not employing emission control device showed strong release of “… small, primary particles” (Bagley et al. 1996).

The extensive study of human cell mutagenicity has been presented for 39 PAHs and nine nitro-PAHs (Durant et al. 1996) is useful for assessing the relative mutagenicities of many of the species discussed herein which are briefly summarized in Table 2. Benzo[a]pyrene (BaP) has long been considered to be a harmful species and is used as the standard for comparison. One lighter PAH, cyclopenta[cd]pyrene (CpP), was found to have minimum relative mutagenicity (RM) that is 6.9 times that of BaP. Also notable are the several six-ring PAHs of mass 302 u have high mutagenicity including dibenzo[a,l]pyrene with a RM of 24. These investigators concluded that the reported concentrations of a subset of four PAHs found in some ambient aerosols could account for a greater mutagenicity than BaP.

The relative mutagenicities (RM) of the PAHs species found in IDF flame tests and in the engine particulate samples (Bagley et al. and NIST Certificate of Analysis, SRM 1650a, NIST 2000) are listed in Table 2 where it is apparent that CpP and BaP with its 252 u isomers BbF and BjF as well as two 276 u isomers appear as the more mutagenic species. Chemical structures of selected pyrogenic PAHs found in flames and in diesel emissions are shown in Figure 6. That the RM rating is sensitive to the species molecular structure as is clearly illustrated in Figure 6 where the RM for the two 252 u isomers are shown to differ by factor of nearly 600.

FIG. 6 Chemical Structures of Selected Pyrogenic PAHs Detected in Hydrocarbon Flames and in Diesel Emissions. The relative mutagenicities (RM) from are given for certain species (Durant et al. 1996).

Of particular interest are the results of a regional study of the human cell mutagenicity of atmospheric samples of respirable particles taken from Rochester, New York; Boston, Massachusetts; and rural Massachusetts (Pedersen et al. 2005). These investigators concluded that the 226 to 302 u unsubstituted PAHs and a single oxy-PAH accounted for a significant fraction of the total mutagenicity of the atmospheric samples taken at all three geographically dispersed locations in northeastern United States. NitroPAHs were measured to be at lower concentrations and did not contribute significantly to the sample mutagenicities in this study.

Carcinogenic properties of PAHs are described by toxic equivalency factors (TEFs) which are based on tumor incidence rates for measured dose levels of individual PAHs in animal tests (Table 4, Nisbet and LaGoy 1992). The uncertainties of the TEFs are noted to range up to a factor of at least three. The estimated TEFs for the PAHs found in diesel effluents are rated relative to benzo[a]pyrene and are highest for the various 252 u, 276 u, and 278 u species.

4. SUMMARY

Precursor nanoparticles have been widely found in laboratory flames burning hydrocarbon fuels in diverse flame geometries. Ex situ sampling and TEM examination reveal these particles to consist of a singlet particle morphology in the 10 nm and smaller size range, and they possess a liquid-like consistency. The ubiquity of these nanoparticles in diverse flames types burning various hydrocarbon fuels leads to the conclusion that they are a likely universal antecedent stage to the more widely observed carbonaceous aggregates.

Particles sampled from diesel engines show similar chemical composition to the samples nanoparticles from IDF flames. Both types of samples in GCMS analysis prominently display the stabilomer PAHs and their isomers in the 200 to 302 u mass range. Many of these species are noted in the literature to be carcinogenic or mutagenic. Nanoparticles (d_p < 10 nm) from engines are also found on and near highways showing high particle number concentrations but may fail to be detected by mass-sensitive air pollution monitors. Nanoparticles are of particular concern to human health because of their small sizes, their singlet morphology which grants access to the alveolar lung region, and their possible toxic chemical content. The recent micrographs (Neer and Koylu 2006) reveal singlet morphology in the nanoparticles sampled from a diesel engine under many operating conditions. The chemical composition of these engine-formed nanoparticles remains to be determined. Particulate control strategies must address the reduction of the pyrogenically formed nanoparticles of the more adverse morphology.

APPENDIX A

TABLE A1 TEM observations of precursor nanoparticles in flames

Citation, fig. no.	Fuel(s)	Flame description	Comment
Wersborg et al. 1973, fig. 3a	C2H2/O2	Laminar premixed	Low pressure (20 torr)
Megaridis et al. 1989, fig. 3 & fig. 4	C2H4	Laminar diffusion
Dobbins et al. 1994, fig. 17.4	C2H4	Buoyancy dominated
Ma et al. 1995, fig. 7	Coal	Early coal particle combustion
Dobbins et al. 1996, fig. 1, fig. 3	C2H4	Laminar diffusion
Vander Wal, 1996, fig. 4	C2H4	Laminar diffusion	Carbonization shown
Dobbins, 1997, fig. 2 to fig. 6	CH4, C2H2, C2H4	Buoyancy dominated	Transition shown
Köylü et al. 1997, fig. 3	C2H4	Laminar diffusion	Transition shown
Vander Wal, 1997, fig. 4	C2H4	Laminar diffusion	Transition shown
Dobbins et al. 1998, fig. 1.	C2H4	Laminar diffusion	MS data shown
Zhang et al. 1998, fig. 4	CH4/Air	Unsteady diffusion	Flickering flame
Konsur et al. 1999, fig. 4	C2H2/N2	Laminar diffusion	In microgravity
Lee et al. 2000, fig. 2	CH4	Laminar premixed	O2 Concentration studied
Blevins et al. 2002, fig. 2	C2H4	Inverse diffusion	GC/MS data shown
Blevins et al. 2003, fig. 2	C2H4	Well stirred reactor	GC/MS data shown
Merchan-Merchan et al. 2003, fig. 2, fig. 4	CH4	Opposed-Flow flame
Onichuk et al. 2003, fig. 5	C3H8	Laminar diffusion	Electric charge studied
Hu et al. 2003, fig. 3	C2H4	Turbulent premixed
Sgro et al. 2003, fig. 2B	C2H4	Laminar premixed	Bimodal PSD shown
Hu et al. 2004, fig. 2	C2H2	Turbulent diffusion
Jensen et al. 2005, fig. 26	JP-8	Pool fire	GC/MS data shown
Oh et al. 2005, fig. 3	C2H4	Inverse diffusion
Lee et al. 2005, fig. 4	C2H4/N2	Inverse diffusion
Yang et al. 2005, fig. 7	C2H4	Turbulent diffusion
Öktem et al. 2005, fig. 9	C2H4/O2/Ar	Premixed laminar	Bimodal PSD & MS data shown

Acknowledgments

This work is based in part on an earlier collaboration with Robert A. Fletcher of the National Institute of Standards and Technology. The support from the Center for Fire Research of the NIST and the U.S. Army Research Office under various grants at an earlier time is also acknowledged with gratitude. Electron microscopy in our laboratory was conducted initially by Dr. C. M. Megaridis during the development of the thermophoretic sampling method. Later TEM work was conducted by Mr. W. Lu and Dr. H.-C. Chang under the supervision of Mr. A. F. Schwartzman. The capable assistance of Mr. Jeffrey S. Brown in the preparation of the manuscript is deeply appreciated.

Notes

¹ Durant et al. 1996.

²Units (μ g/g). Sum of particulate and vapor PAH, 50% engine load, without aftertreatment, Table 35, Bagley et al. (1996).

³Units (μ g/g). Certificate of Analysis SRM1650a, NIST, 2000.

⁴Units (ng/m³). Average of two samples from the IDF, Blevins et al. (2002).