Diesel soot aerosol particles are usually agglomerated to varying degrees, depending on whether they have freshly emerged from the tailpipe, or whether they are aged (i.e. have undergone collisional growth) as in the case of prolonged residence times in a street tunnel. The mechanical stability of soot agglomerate particles is of considerable interest. When used as dye, easy disintegration of "lamp-black" in liquid suspension is usually a desired property; as a reinforcement in automobile tires it is generally not. Shear flow induced soot reentrainment in diesel particulate filters has a strong influence on the over-all system performance [1]. In conjunction with the assessment of health risks posed by inhaled diesel soot in contact with tissue and body fluids, the potential for breaking up larger agglomerate structures into smaller units may have various implications, as well.
Soot, as well as many other anthropogenic or "engineered" nanoparticulate aerosols, is formed in a series of chemical and physical growth processes involving nucleation from supersaturated vapor and coagulation. These initial steps can be accompanied by secondary structural transformations such as condensation of unburned fuel and/or lubricant oil vapors onto the freshly formed, fractal agglomerates [2], some of which may actually be removed again during thermal after-treatment of the soot, for example by oxidation catalysts. After its release into the atmosphere, the diesel aerosol will generally undergo further changes. For example it may undergo collision growth by Brownian coagulation due to slow dilution in confined spaces such as tunnels.
Each of these processes has implications on the stability of a soot agglomerate, and hence on the likelihood of its breaking up later on, under mechanical stress or electrochemical actions in fluid suspension. While pure Brownian coagulation of dry particles will result mostly in van der Waals bonding, the various secondary processes are expected to cause a significant increase in adhesion strength due to the formation of solid or liquid bridges. Hence a soot agglomerate with a complex history may actually consist of substructures with different degrees of mechanical stability. Almost no information is available however on how the various stages of soot formation might affect the strength of individual diesel soot particles. The vast majority of studies concern downstream issues such the effects of carbon black fillers on the tensile strength of rubber. Indeed Friedlander [3] suggested links between agglomerate structure and polymer behavior, and some measurements in this direction have been made lately with AFM techniques [4].
Recently [5, 6] a method was described for measuring the probability of fragmentation of agglomerated aerosol nanoparticles by direct impact from an aerosol stream onto a solid substrate (usually a TEM grid). The method is based on a modified single-stage low pressure impactor (SS-LPI). Particles impact at variable but defined velocities and the degree of fragmentation is determined by image analysis of individual particles in the deposit. The paper also presented first data on the fragmentation probability of nano-agglomerates of silver, nickel, titanium as a function of impact energy. While it does not give a direct measure of interparticle forces, the technique has proven useful to quantify changes and relative magnitudes of effects. For example, one can show the influence of bond strength manipulations on polystyrene particle adhesion [7], or of gradual sintering on the degree of fragmentation of silica nano-aerosols [Seipenbusch, M, Rothenbacher S, Kirchhoff M, Schmid HJ, Weber AP and Kasper G: Interparticle forces in silica nanoparticle agglomerates (submitted)]. These results will be discussed later, in the context of experimental results. We also know that thermally induced rearrangement of primary particles within an agglomerate requires less energy than break-up [8]; experiments have furthermore shown that this rearrangement can be slowed by modification of the nanoagglomerate surface [9]. Generally however, quantitative data on the strength of agglomerate aerosol particles (other than soot) are either scarce or inexistent, especially in the nano size range.
In this paper, we present exploratory results obtained with the above mentioned technique on the fragmentability of soot aerosol particles produced by a passenger vehicle diesel engine, and on the effect of an exhaust after-treatment on the agglomerate strength. The after-treatment was simulated by a thermal denuder, which removes the majority of the adsorbed hydrocarbons at a temperature of 280°C [10]. Since it became clear very quickly during the experiments, that the primary diesel aerosol from the engine exhaust cannot be fragmented at all – with or without removal of adsorbates – the initial question was modified to investigate the effect of adsorbates on an aged diesel aerosol. The aging was done by giving the primary aerosol time to coagulate.
Experimental Setup, Techniques and Aerosol Characteristics
Exp. set-up and aerosol
The diesel soot aerosols were taken from the exhaust gas of an idling Volkswagen 1.4 l TDI (55 kW) engine operated with diesel fuel of a sulfur content below 10 ppm. We know from a detailed analysis of soot and exhaust gas composition for a comparable Volkswagen TDI engine operated under comparable idling conditions that the organic carbon (OC) mass fraction of the soot ranged between 10 and 25% [11].
Fig. 1 shows a schematic diagram of the set-up used for the fragmentation experiments. The engine exhaust gas was diluted 12:1 with dry air to avoid condensation of water in the sampling lines. Under idling conditions the condensation of the volatile organic fraction onto the soot particles occurs in the exhaust system and is therefore not significantly influenced by the dilution. For the experiments with thermal after-treatment, the aerosol was led into the thermodenuder (TD) developed at the Institute of Hydrochemistry, TU Munich. In the TD the diluted aerosol could be heated to 280°C and then drawn into a desorber tube consisting of a perforated stainless steel inner tube surrounded by a filling of activated charcoal grains. The tube geometry (i.d. 10 mm, length 0.6 m) gave an aerosol residence time of 0.43 s. In thermodenuders the highly volatile organic carbon (OC) fraction attributed mainly to fuel hydrocarbons and their fragments is removed between 30 and 175°C, low volatility OC mainly stemming from lube oil emissions is removed between 175 and 300°C [12]. Under the operating conditions described above, the TD removes the highly volatile and the majority of the low volatility OC whose fraction is determined by the complex interaction of engine operating conditions and lube oil properties (Dronia, 2003) [10]. For the experiments without thermal treatment, a bypass line with comparable aerosol residence time was used. Downstream of the thermodenuder, the diesel soot aerosol could be aged by Brownian coagulation in a flow tube (not shown in Fig. 1) with a residence time of 12 s, which suffices to produce a modest degree of additional agglomeration.
Fig. 2 (left) shows a cumulative mobility size distribution of the agglomerated diesel soot particles, with and without TD. The total number concentration of the aerosol bypassing the denuder was about 105cm-3. The agglomerates passing the denuder showed a mean mobility diameter of 94 nm and geom. standard deviation of 1.7, the ones without denuder treatment of 96 nm, indicating that the TD did not affect the size distribution noticeably, despite losses on the order of 30% in the denuder.
Fig. 2 (right) shows a representative size distribution of the primary particles without thermodenuder, obtained by image analysis of 2000 primary particles from transmission electron micrographs. The count mean particle size is 22 nm with a standard deviation of 1.48, which is in good agreement with earlier investigations [13, 14]. No significant difference in primary size distribution was detected between the treated and untreated particles, emphasizing again that the primary particle generation process is terminated at the engine outlet.
The technique for fragmenting airborne agglomerates and determining their degree of fragmentation as a function of impact velocity has been described thoroughly in earlier papers [5, 6], so that we can confine ourselves here to summarizing its essential aspects. Initially, a narrow size fraction of the diesel aerosol having a mean mobility diameter of 150 nm is selected by DMA (TSI Model 3071; operated at a flow ratio of 1:10). This fraction is then introduced to the fragmentation device, a modified single-stage low pressure impactor (SS-LPI) of the type described first by de la Mora [15]. A schematic drawing of the device is shown in Fig. 3. The impactor is generally operated at Stokes numbers well above the range typically used for size classification. Hence all particles are collected directly on TEM grids; bounce is insignificant. By varying the nozzle flow velocity via the pressure in the impaction chamber from about 30 to 300 m/s, one can adjust the impact velocity (ca. 85% of mean nozzle exit velocity [16]) and hence the degree of fragmentation.
The degree of fragmentation is determined by image analysis of a representative number of individual agglomerates collected on TEM grids (see Fig. 4 for illustration). Since the observed fragments have to be assigned to the original agglomerates, it is quite important to ensure appropriately low deposit densities. The inlet concentration of the SS-LPI is therefore monitored by a parallel CPC to adjust the optimal sampling time, based on experience. The degree of fragmentation, F, which we use here as a measure for strength, is defined as the fraction of broken bonds in an agglomerate. It is irrelevant whether any of the fragments are deformed, i.e. the primary particles are rearranged upon impact, as long as the components are not dislocated. We do assume however that the number of bonds can still be determined with reasonable accuracy. The degree of fragmentation is based on the total number of bonds per agglomerate particle, which must be determined without impaction. This is done by depositing agglomerates by diffusion onto TEM grids. For this purpose the impactor is operated with a different flow path with the nozzle module removed (see Fig. 4).
A point to be addressed is sampling volatile or semi-volatile materials with a low-pressure impactor. In this context Berner (private communication) has pointed out that the passage times from the entrance nozzle to the impaction surface are so short (microseconds), that no significant evaporation takes place before impact, even for water droplets.
Systematic and statistical errors in the impact velocity are on the order of a few percent at most, since the device is quite accurately controlled. A given value for the fragmentation degree is typically based on the evaluation of several hundred particle images; errors are thus estimated to be on the order of about 10%.