Particle Emissions Volatility And Toxicity Biology Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Particle emissions, volatility and the concentration of reactive oxygen species (ROS) were investigated for a pre-Euro I compression ignition engine to study the potential health impacts of employing ethanol fumigation technology. Engine testing was performed in two separate experimental campaigns, with most testing performed at intermediate speed with four different load settings and various ethanol substitutions. A Scanning Mobility Particle Sizer (SMPS) was used to determine particle size distributions, a Volatilisation Tandem Differential Mobility Analyser (V-TDMA) was used to explore particle volatility and a new profluorescent nitroxide probe, BPEAnit, was used to investigate the potential toxicity of particles. The greatest particulate mass reduction was achieved with ethanol fumigation at full load, which contributed to the formation of a nucleation mode. Ethanol fumigation increased the volatility of particles by coating the particles with organic material or by making extra organic material available as an external mixture. In addition, the particle related ROS concentrations increased with ethanol fumigation and was associated with the formation of a nucleation mode. The smaller particles, increased volatility and the increase in potential particle toxicity with ethanol fumigation may provide a substantial barrier for the uptake of fumigation technology using ethanol as a supplementary fuel.


The transportation sector is in urgent need of alternative fuels due to the peak-oil scenario and the growing global demand for transport, and the accompanying increase in greenhouse gas emissions (1). Biofuels are being pursued as a replacement for diesel in the transportation sector to facilitate global warming mitigation, to reduce exhaust emissions and also for energy security reasons (2, 3). Ethanol is one example of an oxygenated biofuel that is being explored as a potential replacement for diesel in heavy-duty compression ignition (CI) engines (4, 5).

Several ethanol substitution technologies are available for use in CI engines and include the use of ethanol blends, ethanol emulsions, a spark ignition approach, ignition assisting additives, dual injection of diesel and ethanol, and ethanol fumigation (6, 7). The ethanol fumigation approach involves delivering ethanol vapour to the intake manifold of an engine (7) and complements the existing literature on methanol fumigation (8, 9). Up to 50% of the total fuel energy at full load can be provided through ethanol fumigation, which lies between the energy substitutions achievable by blends (~25%) and dual-injection (~90%) (6). The European Union is committed to a 10% substitution (by energy) of transportation fuel by renewable sources by 2020 (10), so it is possible that fumigation technology may play a vital role in achieving this outcome.

A well documented advantage of ethanol usage in CI engines is the significant reduction in particulate mass emissions, especially at full load operation (6, 11). Despite this, a reduction in the mass of particulates emitted by an engine may not be the most appropriate metric for assessing the potential health effects of diesel particulate matter. For example, a study by Peters et al (12) showed that respiratory health effects in asthma sufferers were related more strongly to the number of ambient ultrafine particles, rather than to the mass (measured as PM10) of ambient particulates. In light of these observations, the measurement of particle number distributions from ethanol combustion in CI engines is an emerging area of research interest, with several recent papers having been published on this topic (11, 13-15). This study represents the first attempt to address the issue of particle number distributions from CI engines that employ ethanol fumigation.

Previous research has addressed the issue of regulated emissions from ethanol fumigation (6, 7). Little work, however, has focused on the health-related properties of these emissions. Particle related health effects are still not understood entirely, but a widely accepted hypothesis for the many adverse health effects induced by particles is that the particles contain and/or are able to generate reactive oxygen species (ROS) and, thus, induce oxidative stress at the sites of deposition (16, 17). In addition to the particle-induced generation of ROS, several studies have shown that particles may also contain ROS (18, 19). As a result, knowledge of the amount of particulate matter (PM) related ROS would assist in assessing the potential toxicological impact of particle emissions from engines that employ ethanol fumigation technology.

To address the lack of data on the emissions of ROS from CI engines, a novel profluorescent nitroxide probe, BPEAnit, was used to detect and quantify the amount of ROS and free radicals generated from neat diesel and ethanol fumigated particle emissions. BPEAnit is a weakly fluorescent compound, but it exhibits strong fluorescence upon radical trapping or redox activity (20). This makes it a powerful optical sensor for radicals and redox active compounds. The collection of other data involved using a V-TDMA system to explore the volatile properties of particles, along with a Dust-Trak to measure PM2.5 emissions.


Engine, fuel and testing specifications

Emissions testing was performed on a pre-Euro I, 4 cylinder, Ford 2701C engine. Studying pre-Euro I engines (from an Australian perspective) continues to have relevance due to the large percentage of the truck fleet (~40 %) that belongs to this emissions class (21). Detailed specifications for the test engine are documented in the supporting information for this paper. The engine was coupled to a Froude hydraulic dynamometer to provide a brake load to the engine. The major components of the dual-fuel system fitted to this engine include an electronically controlled ethanol injector, a pump and pressure regulator, a heat exchanger for vapourising ethanol, and a separate fuel tank and fuel lines. A 1 kW heater positioned downstream of the ethanol injector was required to fully vapourise ethanol for higher ethanol substitutions.

Testing was performed with commercially available 10 ppm sulphur diesel. The ethanol used in testing had a moisture content of 0.55% (by mass) and was denatured with 1% unleaded petrol (by volume) in accordance with the fuel supplier's legal requirements. Two experimental campaigns were conducted. The first was conducted at 2000 rpm, full load, and the second at intermediate speed (1700 rpm) using four different load settings. Table 1 documents the speed, load and fuel settings used in both experimental campaigns. Note that "EX" denotes that X% of the total fuel energy was provided by ethanol. Consequently, E0 indicates a test that was conducted with neat diesel.

Speed, load and fuel settings used for both experimental campaigns.

Campaign number

Speed (rpm)

Load (%)

Fuels used




E0, E10.6, E16.3, E22.9




E0, E40




E0, E10, E20, E40




E0, E20




E0, E10

Apart from the different speed settings used, the biggest difference between the two experimental campaigns involved the ability to control ethanol fumigation percentages. Full percentage ethanol substitutions (such as 20%) could not be achieved in the first experimental campaign due to using an oversized injector that could not provide the required flow rate. This problem was rectified in time for the second experimental campaign.

For each load setting, all tests were conducted at the brake load associated with neat diesel operation. Tests were designed this way so that any change in the emissions was due to the change in fuel and not due to the different power output of the engine. Data collection did not commence until the exhaust, cooling water and lubricant temperatures and the gaseous emissions had stabilised. In order to prevent the results from being affected, another test procedure involved flushing the fuel lines of ethanol, and leaving the engine to stabilise for approximately half an hour before further tests were conducted.

Particle measurement methodology

A two-stage, unheated dilution system was used to condition exhaust gas before particulate sampling. The first stage of dilution was performed with a dilution tunnel and the second stage with a Dekati ejector diluter (Dekati, Tampere, Finland). Dilution air was passed through a large HEPA filter to provide particle free air for the primary dilution. Filtered compressed air at 2 bar gauge pressure was fed to the ejector diluter for the second stage of dilution. Particulate mass emissions were measured with a Dust-Trak using a specially designed isokinetic sampling port on the dilution tunnel. CO2 was used as a tracer gas to calculate dilution ratios.

After the ejector diluter, the aerosol stream was split into three flows for particle size, volatility and ROS measurements. Figure 1 displays a schematic of the experimental set-up used in this study. The methodology for each type of measurement is described below.

Particle number distributions were measured with a SMPS consisting of a TSI 3071A Classifier (EC) and a TSI 3782 Condensation Particle Counter (CPC). Particles within a 10-400 nm size range were measured. For the neat diesel tests, 15 SMPS scans were taken and at least 5 scans were taken for tests involving ethanol.

Schematic representation of the experimental configuration used in this study.

Particle volatility methodology

A Volatilisation Tandem Differential Mobility Analyser (V-TDMA) was used to investigate the volatility of particles (22, 23). The system is composed of an electrostatic classifier that pre-selects particles of a set size, followed by a thermodenuder that heats the pre-selected particles (see Figure 1) to a set temperature. Once particles are heated, the change in particle size is measured with an SMPS. The SMPS consists of an identical classifier to the one that pre-selects the particles and also consists of a TSI 3010 CPC. The temperature difference between the saturator and condenser of the CPC was increased to 21 °C to improve the particle detection efficiency down to 8 nm. The thermodenuder temperature was increased in discrete steps and the change in the particle diameter was recorded as the volatile components evaporated.

Accumulation mode particles with a diameter of 80 nm were pre-selected for V-TDMA analysis. Pre-selecting this particle size was based on the mode of the neat diesel particle size distribution as derived by the SMPS system at full load. Scan times of 90 seconds were chosen for the SMPS system downstream of the thermodenuder and measured particles within an 8-109.1 nm size range. The thermodenuder was set up to scan temperatures in an approximate range of 30 to 320 oC, with temperature increments of 25 to 30 oC between scans. All testing with the V-TDMA system was performed at intermediate speed. The thermodenuder temperatures were calibrated in the laboratory after testing.

Particle volatilisation is presented through the volume fraction remaining (VFR),

, where is the particle diameter before heating, and is the particle diameter after heating in the thermodenuder to a temperature . The dependence of the VFR on thermodenuder temperature provides a volatility signature for particles that enables basic hypotheses regarding the chemical composition and formation mechanisms of particles to be tested. For example, using a V-TDMA system, Sakurai et al (24) demonstrated that diesel nanoparticles had a volatility signature consistent with heavy hydrocarbons (C24-C32) that are prominent constituents of lubricating oil.

In diesel particles, it is common to observe particles of the same size but significantly different composition and therefore volatility (24). Particles such as this are known as external mixtures. As a measure of external mixing, we have quantified the percentage of volatile particles (PVP) according to the following equation:



is the total particle concentration measured by the V-TDMA before the thermodenuder temperature is increased, and is the total concentration of the non-volatile peak at the highest temperature applied by the thermodenuder.

ROS concentration measurement - BPEAnit assay

Samples were collected by bubbling aerosol for 15 to 20 minutes through an impinger containing 20 mL of 4 µM BPEAnit solution, after which fluorescence was measured. Impingers were placed after the two-stage dilution system. For each fuel type and test mode, fluorescence measurements from both the test sample and a HEPA filtered control sample were measured.

The new profluorescent nitroxide probe BPEAnit and its methyl adduct (BPEAnit-Me), which was used for the calibration curve, were synthesised in our laboratory. The details of the synthesis are presented in Fairfull-Smith and Bottle (20) and details of the evaluation of the probe for applications in particle bound ROS quantification are presented in Miljevic et al (25). The solvent used in all experiments was AR grade dimethyl sulfoxide (DMSO). Impingers used in this study were custom made and consisted of a Quickfit Dreschel bottle head, they were sintered (porosity grade 1: pore size of 100-160 µm) and were modified to fit a Quickfit 75 mL test tube (Barloworld Scientific, Staffordshire, UK).

Fluorescence spectra were recorded using a USB2000 fibre-optic spectrometer combined with a cuvette holder and a pulsed Xenon lamp (both Ocean Optics, Dunedin FL, USA) which used a narrow bandpass filter at 430 nm (Edmund Optics, Barrington, NJ, USA). In all of the fluorescence measurements a 10 mm quartz cuvette (Starna Pty Ltd, Hainault, UK) was used.

In order to conduct quantitative chemical analysis on the particles collected by the impingers, it was important to know the collection efficiency of the impinger. This was determined as described in Miljevic et al (26).

The amount of BPEAnit that reacted when exposed to engine exhaust was calculated from a standard curve obtained by plotting known concentrations of methyl adduct of BPEAnit (BPEAnit-Me; fluorescent) against fluorescence intensity at 485 nm. Based on the difference in the fluorescence signal between the test and the control sample, the amount of ROS for each test mode was calculated and normalised to the PM mass calculated from the SMPS data. The portion of particles remaining in the impinger upon bubbling was calculated by multiplying the size distribution by the impinger collection efficiency curve.


Particle size distributions

Full load size distribution data from the SMPS is shown for the neat diesel (E0) and E40 test in Figure 2. Full load size distributions are shown since a clear difference was exhibited for the E0 and E40 tests. Size distributions, at all other loads, for the neat diesel and ethanol tests were quite similar, since nucleation occurred in each size distribution.

SMPS derived particle number distributions at intermediate speed (1700 rpm), full load, for neat diesel (E0) and 40% ethanol (E40) engine operation. Error bars denote ± one standard error.

The mode for the diesel size distribution was about 90 nm, which was in very good agreement with that expected for diesel particulate matter (27). A 40% ethanol substitution (on an energy basis) markedly changed the neat diesel size distribution. The ethanol size distribution had a large peak in the nucleation mode, and it also had a reduced particle mode diameter and reduced accumulation mode particle concentrations.

A correlation between particle size and ethanol substitution is shown in Figure 3, for tests conducted in the first experimental campaign at 2000 rpm, full load. The count median diameter (CMD) of the SMPS-derived particle size distribution was used as a metric for the size of particles. Relative to the neat diesel case (E0), the E22.9 test reduced the CMD by approximately 20%, from 81 nm to 63 nm. It can be seen that the CMD is anti-correlated with the ethanol substitution percentage (=-0.939), where is the Pearson correlation co-efficient.

Correlation of particle size (CMD) with the ethanol substitution percentage for tests conducted at 2000 rpm, full load (=-0.939). Error bars denote ± one standard error.

Brake-specific particulate matter (PM2.5) emissions, for tests conducted at intermediate speed (1700 rpm) and various loads settings during the second experimental campaign, are shown in Figure 4. In general, the addition of ethanol significantly reduced PM emissions, especially at full load operation during the E40 test. The results at idle mode were not consistent with the general trend, since E10 led to an increase in PM emissions, relative to E0. No explanation can be provided for this result. Full-load PM reductions from ethanol were significantly greater than those observed at half or quarter load.

Brake-specific PM2.5 emissions at intermediate speed (1700 rpm) with various load settings and ethanol substitutions. Error bars denote ± one standard error.

Particle volatility

The volume fraction remaining (VFR) curves are displayed in Figure 5a and 5b. It can be observed that for neat diesel at full load (Figure 5a), heating the particles results in a very small reduction in particle volume, whilst for the E40 test at the same load there was a significant reduction in particle volume. For loads other than full load, heating of particles introduces a second, far more volatile peak in the size distributions. This more volatile peak separates from the initial distribution of particles pre-selected for V-TDMA analysis, and consequently has to be analysed separately to the less volatile peak for volatilisation information (see Figure 5b). Further, it should be noted that significant volatilisation occurred between 50 and 100 oC, suggesting the presence of fuel or lubricating oil derived organic material (23, 24). Size distribution information for the V-TDMA scans can be found in the Supporting Information of this paper.

Volume fraction remaining (VFR) versus thermodenuder temperature at intermediate speed (1700 rpm). (a) 100% load E0 and E40. (b) 25% load E0 and E20. Note well the linear scale on the ordinate for (a) and the logarithmic scale on the ordinate for (b). Error bars are calculated using the uncertainties in the diameter measurement.

The volatility results require external and internal mixtures to be defined. An external mixture in automotive exhaust entails carbon and other aerosol particles (such as volatile droplets) existing as distinct, or separate, particles. Alternatively, internally mixed particles have the various components incorporated together and could consist of a carbon core coated with other aerosol particles (28). The level of external mixture can be presented through the percentage of volatile particles (PVP). Figure 6 presents the PVP as a function of ethanol substitution for three load conditions, namely: idle, 25% and 50% load. The PVP at full load is not shown as the particles were not externally mixed and consequently the PVP was zero. For all three loads, an increase in the percentage of volatile particles was observed with ethanol. For the half load case, where several ethanol substitutions were measured, a clear increasing trend in PVP can be observed with higher ethanol substitutions. The volume fraction of organic material coating the non-volatile particles increased with increasing ethanol substitutions at all loads, with results appearing in the Supporting Information of this paper (see Figure S2).

Percentage of volatile particles at intermediate speed (1700 rpm) and 50%, 25% load and idle mode for various ethanol substitutions. Error bars have been calculated using the statistical uncertainty in the counts .

ROS concentration results

Figure 7 shows fluorescence emissions for the BPEAnit solution when exposed to diesel exhaust, and also for the HEPA-filtered control samples, at intermediate speed (1700 rpm) and full load for the neat diesel (E0) and 40% ethanol (E40) tests. There was an increase in the fluorescence signal for BPEAnit when exposed to the engines exhaust, with the fluorescence signal for E40 being significantly higher than for E0. An increase in fluorescence for the HEPA-filtered control samples (grey curves in Figure 7) is due to gaseous reactive species, whereas the fluorescence of test samples (black curves in Figure 7) represents the response due to aerosol being bubbled through the impinger. The difference between the black and the grey curves is, therefore, the fluorescence induced by particles. A small increase in fluorescence was observed for PM emissions from neat diesel testing. On the contrary, PM emissions from the E40 test led to a 4.5-fold increase in fluorescence, relative to the neat diesel case.

Fluorescence spectra of BPEAnit control (HEPA filtered) and test samples for neat diesel (E0) and 40% ethanol (E40) at intermediate speed (1700 rpm) and full load.

Normalisation of the amount of BPEAnit being converted to fluorescent product, with respect to the PM mass, represents a measure of ROS concentration. Figure 8 displays ROS concentrations calculated for PM emissions at intermediate speed with various load settings and ethanol substitutions. ROS concentrations for neat diesel emissions at 0% (idle), 25%, 50% and 100% load show a significant increase with decreasing engine load. ROS concentrations for 10%, 20% and 40% ethanol tend to exhibit the same increasing trend as the load is decreased, similar to the neat diesel emissions. At a particular load setting, the ROS concentrations for the E10 and E20 tests, relative to E0, do not differ by any more than 20%. At half load, however, the E40 test resulted in approximately double the ROS concentration relative to E0, and for full load, the E40 test resulted in a ROS concentration almost 40 times higher than for neat diesel.

ROS concentrations at intermediate speed (1700 rpm) with various load settings and ethanol substitutions. Error bars denote ± one standard error.


Several recent papers have addressed the issue of particle number distributions emanating from CI engines using the ethanol blending approach (11, 13-15). Although measurements were taken at different speeds, and different loads and ethanol blend percentages, a common feature of the work of Di et al (11, 13) and Lapuerta et al (15) is that ethanol reduces the peak particle concentration and shifts the CMD of the size distribution to a smaller particle diameter. As a result, ethanol blending technology produces a higher percentage of particles that reside in the ultrafine (<100 nm) size range, but the overall number concentration with ethanol blends is lower relative to the neat diesel case. Using fumigation technology, the result presented in this study (see Figure 2) is different. The reduction in particle concentration in the accumulation mode (>50 nm) is still evident, but the concentration in the nucleation mode (<50 nm) is higher by a factor of approximately 8.

The mechanism responsible for nucleation in this case appears to be consistent with a theory developed by Kittelson et al (29). The accumulation mode is very effective at absorbing organic material, due to the large surface area available. With the high ethanol substitutions achieved with fumigation, the accumulation mode surface area is reduced to a level such that organic material has very little particle surface area upon which to condense. So instead of the accumulation mode acting as a "sponge", absorbing organic material and hence reducing its vapour pressure, the organic material resides in the vapour phase with an increased vapour concentration. Under conditions where the vapour pressure of a nucleating species is high and exhaust gas dilution cools the organic material, thereby decreasing its saturation vapour pressure, the saturation ratio of organic material is significantly increased and nucleation can occur instead (29). In this case, nucleation occurred solely due to the change in fuel and was not related to some artefact of the dilution process, such as having different dilution ratios or tunnel temperatures during each test. This study provides the first experimental evidence that high ethanol substitutions are capable of inducing nucleation in particle size distributions, in addition to decreasing the concentration of particles in the accumulation mode.

Results from the V-TDMA analysis suggest the presence of an organic substance, either derived from fuel or lubricating oil, which coats particles and potentially leaves a sufficient concentration in the vapour phase for nucleation to occur. The amount of volatile material available for nucleation is proportional to the percentage of volatile particles (PVP) and increases with ethanol substitution (see Figure 6). Ethanol fumigation increased the volatility of particulates, either through coating particles with volatile, organic material or through making organic material available for nucleation to occur, producing an external mixture of purely volatile and partially volatile particles.

Ethanol fumigation increased the particle related ROS concentration, especially at full load operation, although the ROS concentration was reduced at idle mode operation with E10. The lowest ROS concentration occurred with the full load E0 test, which was the only size distribution measured which did not exhibit a nucleation mode. For all the other tests (involving a nucleation mode), at least a 30 fold increase in ROS emissions occurred, relative to the E40 full load test. Therefore, significantly higher ROS concentrations are associated with the formation of nucleation mode particles. An explanation for the mechanism governing the formation of ROS due to ethanol fumigation and its relationship to the formation of a nucleation mode is not possible with the data collected in this study; therefore, further investigation is recommended. The significant increase in potential particle toxicity with ethanol fumigation may provide a substantial barrier for the uptake of fumigation technology using ethanol as a supplementary fuel. Other supplementary fuels should be investigated with fumigation technology to explore the potential toxicological impacts.

The reduction in particle size was also strongly anti-correlated with the percentage of ethanol fumigated (see Figure 3). A possible mechanism for this observation is the oxidation of particulate matter by OH radicals. It has been suggested that OH radicals are much more effective at oxidising the soot surface than other oxidants such as O2 (30), which are decreased by ethanol fumigation (see calculation in the Supporting Information for this paper) since the intake air is being replaced by fumigated mixture. At temperatures relevant for combustion (1200-2200K), Daly and Nag (31) showed through kinetic modelling that the peak concentration of OH radicals during combustion were approximately doubled for a 10% ethanol blend. The concentration of OH radicals tends to peak at the onset of soot depletion (32); therefore, ethanol combustion potentially involves more available oxidant to attack the particle surface and hence reduce its size. An increase in OH radicals with increasing ethanol fumigation percentages is also confirmed by an AVL Boost simulation conducted as part of this study. The Boost program solves the one-dimensional Euler equations for in-viscid, compressible flow and is coupled to a zero-dimensional combustion model. A detailed chemistry module is available in the program for performing emissions simulations (33). Results for this simulation appear in the Supporting Information for this paper.

A modelling study conducted by Benvenutti et al (27) demonstrated that the methyl radical (CH3) is an important precursor for the formation of excited species such as OH radicals. That fact that the nitroxide probe traps radicals and that, in general, radical concentrations were greater with ethanol fumigation, suggests that ethanol combustion provides a pathway capable of significant CH3, and subsequently, OH radical production. This could also be one of the reasons for a significant increase in the ROS concentration observed with ethanol substitution.

A detailed characterisation of particle emission properties has been undertaken for a pre-Euro I engine without after-treatment. Newer engine technologies, with after-treatment devices, should be investigated to ascertain if the same qualitative trends are evident.

Supporting Information Available

The supporting information includes one table and four figures that augment the results presented in the manuscript. This information is available free of charge via the Internet at