The impact of fuel properties within combustion modes, as well as the effect of combustion strategies, is discussed in the following section.
Criteria Pollutant Emissions
The criteria pollutant emissions in the engine-out exhaust are shown in Table 3. The NOx and CO emissions were measured with an FTIR, and THC emissions were measured with a FID. The HCCI and PFS points had very low NOx emissions, with concurrent high CO and THC, as expected for kinetically controlled combustion. These combustion modes can be classified as low-temperature combustion as they resulted in extremely low engine-out NOx and low soot as measured by the MSS . Emission trends for the HFS points were less clear. The 87 RON fuel saw a decrease in the NOx and CO emissions as stratification increased between the two HFS points. For the 70 RON fuel, the NOx increased for the EGR sweep as EGR dropped from 25% to 0%. Dropping the amount of EGR increased stratification demonstrating the need for high EGR rates to achieve conventional LTC operation.
The speciated aldehydes were measured with DNPH cartridges from the dilute exhaust and are shown in Fig. 5. The aldehydes produced at these combustion points were dominated by formaldehyde, for both fuels, and reached emissions rates from ~25% to 60% of the THC emission rates measured by the FID, see Fig. 5 and Table 4. It is important to note that formaldehyde is detected by the FID significantly less sensitively, so that fraction of the aldehyde emissions will be underrepresented in Fig. 5b THC rate and must be considered along with the THC rate as part of the calculation when determining non-methane organic gases (NMOG). The longer-chain aldehydes can be measured by an FID but at varying response levels. High levels of formaldehyde in kinetically controlled ACI combustion studies have been previously reported [19, 20].
The higher aldehyde and THC emission rates seen for the HCCI and PFS modes, compared to the HFS modes, may be related to the slightly lower combustion temperatures common for kinetically controlled combustion. These temperatures might explain the increase in both unburned, partially oxidized, and cracking product HCs, likely fuel HCs, evident in the high THC and aldehyde emissions measured. The 87 RON fuel had lower aldehyde formation, especially formaldehyde, but higher THC in the HCCI mode than the 70 RON fuel. Since formaldehyde formation is more likely to proceed through oxidation of the partial cracking products of saturated HCs, rather than via hydroxyl radical attack of unburned or partially oxidized aromatic or olefinic HC fuel constituents, the difference in HCCI emissions between the two fuels may be related to the fuel composition. To that point, the 70 RON fuel had a higher saturated HC fraction than the 87 RON fuel, which instead had a higher fraction of aromatics (Table 2). Correspondingly, the higher THC emissions of the less reactive, 87 RON fuel, relates to the fuel’s lower saturated HC and increased aromatic fuel HCs. This suggests that the fuel HC composition contributed to the aldehyde-to-THC emissions trade-off. This fuel effect is further supported by the 87 RON fuel’s production, especially in the HCCI mode, of higher benzaldehyde emissions (C7 aldehyde, Fig. 5a), which is a partial oxidation product of toluene, typically the largest aromatic in gasoline.
While a significant drop in total aldehydes and THC was seen for the HFS modes compared to the kinetically controlled modes, regardless of fuel, a larger drop in THC than aldehyde emissions was typically seen (Table 4). Like the kinetically control modes, formaldehyde was still the dominant aldehyde for all of the HFS modes. Comparison of the 87 RON-HFS 1 point to the 70 RON-HFS 1 point showed a similar fuel effect on formaldehyde production as that discussed for the kinetically controlled modes. However, an increase in formaldehyde production combined with a drop in the THC emissions rates was seen between the two 87 RON-HFS points, suggesting other variables may play a more prominent role in impacting these emissions in the HFS modes.
PM Mass and Organic Carbon/Elemental Carbon Emissions
Filter samples of exhaust PM were collected at each condition and analyzed for mass and carbon compositions, Table 5. Figure 6 shows the total PM mass collected and mass quantified from EC/OC data analyses using the NIOSH method for both the premixed and stratified modes. The NIOSH method is a thermo-optical method to identify PM mass as either organic carbon (OC) or elemental carbon (EC) . The PM2.5 PTFE membrane filter used for EPA regulated total PM mass measurements was also collected  and compared to the total carbon (TC) measured by the NIOSH method. Because of the varying EGR rates, and thus exhaust flow rates, the PM concentrations were converted to mass emission rates for comparison.
As described in Sect. 2.3, the customary practice in this laboratory and others has been to take two QFs for each EC/OC measurement . The primary QF collects PM and the secondary QF, in series downstream of the PTFE membrane filter, collects HC that absorbs to the quartz fiber matrix and not part of the PM. The OC measured on the secondary QF is subtracted from the primary QF to correct for this gas phase adsorption artifact. The assumption made in this correction is that the time it takes for the two QFs to come to equilibrium with the gas phase HCs in the exhaust stream is much faster relative to the sampling time. The stacked bar data in Fig. 6a was corrected for QF absorption artifacts as described. For comparison, the correction was not applied in Fig. 6b. There appears to be better agreement between the gravimetric PM mass rates, Fig. 6 (circles), and the TC mass rates calculated when no artifact correction was applied. A possible explanation for the discrepancy seen in Fig. 6 between the TC and gravimetric PM, when the artifact subtraction is done, is that the high level of HC has resulted in there being some hydrocarbon adsorption to the PM already deposited on the filter or on the PTFE membrane filter itself. This phenomenon has been observed by Maricq et al. in PM measurements of gasoline vehicles that have very low solid PM . In previous research on more conventional spark-ignited combustion [23, 24], the correction has been applied with better agreement between TC and gravimetric PM. The discrepancy also points to potential problems with PM measurements during ACI operation. Just as ACI PM cannot be defined as soot carbon, the large amount of semi-volatile hydrocarbons present in ACI exhaust can lead to sampling artifacts such as filter adsorption. Future ACI PM research is needed to address these artifacts and how to avoid them.
Table 5 shows the OC, EC, TC, and gravimetric PM emissions for the different fuel-mode combinations. The results for the kinetically controlled modes (HCCI and PFS, Fig. 6) indicate similar PM mass emission rates for both 70 RON points. The 87 RON-HCCI point had about 50% more gravimetric PM than the 70 RON-HCCI point. This apparent fuel difference may be related to the higher THC emissions produced at the 87 RON-HCCI point compared to the 70 RON-HCCI test condition (Table 3), in which the higher THC emissions may have led to “collection” of more hydrocarbon mass on the PM filter.
For the highly stratified modes (HFS), which are closely approaching stratification levels indicative of CDC, the 87 RON-HFS fuel points exhibited lower mass emissions than the premixed 87 RON-HCCI point. The PM produced at the 87 RON-HFS 1 and -HFS two points were still predominately organic carbon with 97% and 71% OC, respectively (Table 5). This difference correlates with the lower THC emissions for the HFS points (Fig. 5b). Similarly, comparing the 70 RON kinetically controlled and HFS modes, the 70 RON-HFS 1 point had more EC (24%, Table 5), but still lower overall PM emissions. The THC for the 70 RON-HFS 1 point was also much lower than the THC for the kinetically controlled modes (Fig. 5b).
For fuel-to-fuel comparison, the 70 RON-HFS point that correlates most closely, from a combustion standpoint, with the 87 RON-HFS points is the 70 RON-HFS 1, and it produced a lower PM mass rate. The same fuel differences discussed for the aldehyde emissions, fuel composition and reactivity, may also contribute to the higher PM mass rate measured for the 87 RON-HFS points. In gasoline, spark-ignited combustion, higher aromatic content in the fuel, like the 87 RON, are associated with higher PM  emissions. Additionally, the higher reactive of the 70 RON fuel may lead to more complete combustion resulting in less PM with these ACI modes.
In order to achieve equivalent combustion phasing to the 87 RON-HFS points for the 70 RON-HFS 1 point, an increase in EGR (25%) was required, compared to the two 87 RON-HFS points at 15% EGR. Therefore, the effect of EGR on 70 RON-HFS PM was also investigated with an EGR sweep of 25%, 15%, and 0%. As a result of the change in EGR, the centroid of injection shifted slightly, Table 5. This shift in injection shows, as expected, that decreasing levels of EGR increased the level of air-fuel stratification and the total PM mass for the 70 RON EGR sweep. Figure 6 shows that the increase in PM emissions during the EGR sweep was due primarily to the increase in the EC, which reaches 76% of the total PM mass at the 0% EGR point.
MSS and EEPS Results
As stated previously, the MSS is a photoacoustic measurement of the soot PM and thus can be compared to the EC portion of the PM compositional analysis. The plot in Fig. 7 shows a general agreement between the soot PM measured by the MSS and the EC measured by the NIOSH method, in terms of comparable magnitude between test points. However, the MSS consistently measures more mass than the EC fraction measured by the NIOSH method. One potential explanation for the elevated MSS reading is the presence of very small particles that respond to the photoacoustic detector but are too small to have significant mass. Additionally, if some of the organic carbon PM was incorporated into the particulate during the combustion stage, the OC may be located or incorporated into the core of a particulate rather than just on its surface as a result of adsorption or condensation. The MSS, ideal for measuring diesel particulate, which typically has a well carbonized structure and < 20% OC, measures the photoacoustic impact of the EC PM without being influenced by any OC that may have adsorbed or condensed on the diesel PM surface. In this instance, the photoacoustic signal may be skewed by OC with high degrees of conjugated bonds, like soot, or OC within a disordered EC or otherwise carbonized structure. While these results suggest that some of the OC fraction of the ACI PM is being measured by the MSS, it is still only a small fraction of the overall PM OC mass, as can be seen in Fig. 8, which compares the TC PM mass to that measured by the MSS.
The EEPS particle size distributions, shown in Figs. 9 and 10, further highlight the differences in ACI PM composition as a result of injection mode and fuel properties. The common delineation of particle size modes for emission particulate has been shaded in the EEPS figures as nuclei mode (< 25 nm) and accumulation mode (25–300 nm) with all particles greater than 300 simply labeled larger agglomerates. The PMP conditioning of the exhaust before EEPS sampling, briefly described in Sect. 2.3 , was developed for diesel PM emission measurements. The injection strategies, temperatures, and fuels of ACI combustion strategies, like those used in this study, are different than CDC, and therefore, may be expected to produce different types of particulates . Therefore, the assumption that all nuclei mode particles, measured after the 350 °C evaporation tube of the PMP, are solid EC particulate for diesel particulate may not hold true for ACI particulate. While detailed speciation of ACI PM was beyond the scope of this project, on-going work in our lab on ACI PM will include further investigation into the chemical difference which result from changes in injection mode and/or fuel properties.
In Fig. 9, the size distributions for the LTC modes (HCCI and PFS) for both fuels are shown as log plots (a) and semi-log plots (b). Since the PM composition for all three points was shown to be > 95% OC, and even 100% OC for the 70 RON-HCCI point, it suggests that the particles measured by the EEPS would have to be largely organic carbon PM even in the nuclei mode region. The bimodal distribution for all three kinetically controlled conditions can clearly be seen in the log plot of Fig. 9a, with particles in both the nuclei and accumulation mode ranges. The Fig. 9b semi-log plot shows a significant increase in nuclei mode particles was produced by the 87 RON fuel at the same, premixed, HCCI combustion mode compared to the 70 RON fuel. As the air-fuel stratification was increased from HCCI to PFS for the 70 RON fuel, an increase in both nuclei and accumulation mode PM particles was seen. The PM mass doubled when switching to the lower reactivity 87 RON fuel in the same HCCI mode (Fig. 6) even though an increase in particle number was only observed in the low mass nuclei mode range. Therefore, the difference in PM mass was likely from the organic portion of the PM mass that was volatilized by the PMP and not measured by the EEPS. In contrast, an increase in both nuclei and accumulation mode range particles resulted in little change in the total PM mass for the 70 RON fuel when the air-fuel stratification mode was changed from HCCI to PFS. Following the same logic, this suggests that even within the kinetically controlled combustion modes, the speciation of the organic portion of the PM may be impacted by changes in the fuel properties.
A comparison of the particle size distributions for the two 87 RON-HFS points and the 70 RON-HFS 1 point, which is most closely related to them (25% EGR), is shown in the log plots (c) and semi-log plots (d) of Fig. 9. The point with the earliest centroid of injection of these three, 87 RON-HFS-1, clearly shows a bimodal distribution in the log plot, Fig. 9c, again indicating the presence of nuclei mode particles. All three curves in (c) and (d) show a significant quantity of accumulation mode particles, consistent with the presence of soot PM indicated by the MSS and EC measurements. While a clear bimodal distribution was not seen for 87 RON-HFS-2 or 70 RON-HFS-1, all three contain more nuclei mode particles (< 25 nm) than were observed for any of the kinetically controlled modes which show bimodal distribution in the Fig. 9 EEPS plots. The smaller amount of PM mass for the 70 RON-HFS-1 point shown in Fig. 6 appears at first to be inconsistent with the size distribution in Fig. 9c, d. The 87 RON-HFS 1 point has the smallest number of particles and smallest size distribution; however, it also has more large agglomerates (i.e., particles > 300 nm) which tend to have more mass. This particular point also has low EC mass (Table 5), which implies that the large agglomerates likely make up a significant portion of OC mass. A future study using size-selective speciation of the PM could determine the make-up of these larger particles.
In Fig. 10, the particle size distributions for the 70 RON EGR sweep (HFS-1, HFS-2, HFS-3) are shown. The increasing size of the mean particle diameter and amounts of particles in the accumulation mode correlate with the increase in soot PM and EC shown in Fig. 8. The mean diameter also increased from 25.5 nm to 39.2 nm, Fig. 10, indicating particle growth in the accumulation mode.
The PM size distributions observed for the kinetically controlled HCCI and PFS points were similar to the ones observed for RCCI combustion with this engine . In that study, the PM was also dominated by OC mass and in some cases EC mass was undetectable. Speciation of that RCCI PM found that the high boiling point components of the high reactivity fuel (ULSD or biodiesel) were the primary components of the OC. Both the 70 RON and 87 RON fuels are gasoline-range and do not have high boiling components like diesel, so the OC for these fuels may be composed of partially oxidized fuel species and other combustion products that have a lower volatility than the fuel HCs. Understanding the composition of the PM OC will be important to identifying PM control techniques for ACI operation, whether it is filter-based, oxidation catalyst based, or a combination of both.