Soot Oxidation in Periphery of Diesel Spray Flame via High-Speed Sampling and HR-TEM Observation.
Diesel engines are expected to play an important role as an effective and practical measure against the global warming problems because of their high thermal efficiency. On the other hand, diesel engines suffer from the tradeoff between soot and NOx emissions due to the high-temperature diffusion combustion nature. Recent diesel combustion models are able to predict NOx emission at an acceptable accuracy, while the prediction of soot emission, including the particle mass (PM) and the particle number (PN), remains to be improved because of the lack of the fundamental understanding of in-cylinder or in-flame soot formation and oxidation processes. Most of the soot particles formed in the cylinder disappear due to the rapid oxidation, and only 0.1% order of the peak soot mass remain unburned through the oxidation phase and are finally emitted as engine-out soot. This means even small fraction of error in soot oxidation prediction might results in the significant error in the soot emission prediction, as discussed by Kamimoto et al. . For this reason, detailed understanding of diesel in-flame soot oxidation process is essential for accurate prediction of soot emission.
Unfortunately, there are many fundamental unknowns about diesel in-flame soot oxidation processes such as the applicability of currently available diesel soot oxidation models based on classical low pressure burner experiments by NSC  and Neoh , to the high pressure and high temperature in-flame conditions equivalent to modern diesel engines. As reviewed by Stanmore et al. in 2001 , the high-temperature soot oxidation in laminar burners and shock tubes have been studied and somewhat reflected to engine equivalent models. However, those classical models are still widely used in recent simulation studies [e. g. 5, 6 and 7] with bold assumptions like soot oxidation rates are proportional to the surface area of sphere-represented soot particles with arbitrarily adjustable coefficients. Additionally, our current understanding is severely lacking even on whether the oxidation process is homogeneous or heterogeneous within the scale of particle or aggregate sizes, which part of the soot particles or aggregates the oxidation starts from and how the oxidation affects separation or collapse of soot aggregates. Such fundamental unknowns on soot oxidation models and processes should be discussed based not on coarse comparisons between predicted and measured engine emissions, but should starts from phenomenological and microscopic understanding of the diesel inflame soot oxidation process itself under engine equivalent conditions.
Soot nanostructure, which refers to physical structure of oriented carbons graphene layers i.e. polycyclic aromatic hydrocarbons (PAHs) within the soot particles, is known to be related to oxidative reactivity [8, 9,10,11, 12]. Therefore, several researchers have used electron microscopy to observe the inner structure of particles and reveal the details of soot oxidation processes [13, 14, 15, 16, 17, 18]. Ishiguro et al. investigated nano-structural changes within the engine-exhaust soot during its oxidation in an electric furnace at 500[degrees]C air atmosphere via TEM observation and FT-IR spectroscopy . It has been observed that, in the early stage of oxidation, void structure is formed by release of soluble organic fraction (SOF), and then, carbon layer planes within the soot particles are re-arranged towards more graphitic. Finally, the outermost shell within the soot is stripped and their primary particle diameter is decreased. Similar oxidation behavior has also been observed in different oxidation systems such as thermogravimetric analyzer (TGA) [11, 12] and laboratory flow reactor . However, these preceding studies are mainly focusing on the soot oxidation by [O.sub.2] at relatively low temperature corresponding to the regeneration of diesel particulate filters (DPF). The diesel in-flame soot oxidation, which is expected to be rapid, heterogeneous and known to be predominated by OH radicals [4, 19, 20, 21] at relatively high temperature around the diesel flame periphery [22, 23, 24, 25], can be significantly different from the ones observed in the preceding studies. Therefore, in order to understand the details of the diesel in-flame soot oxidation process, nano-structural TEM observation of soot particles directly sampled from the oxidation-dominant diesel flame periphery is important.
The authors have been conducting direct sampling of diesel inflame soot particles in a constant-volume vessel and TEM analyses of their morphology and nanostructure variation along different spray axial locations [26, 27, 28] with several different fuels [29, 30, 31, 32, 33]. These studies revealed that the soot aggregate and primary particle sizes notably vary with in-flame locations, fuels and oxygen concentrations. However, the nanostructure of the in-flame soot exhibited only small variation with these parameters. These results seem contradictory to the ones of engine in-cylinder soot sampling and TEM analyses by Song et al. which reports significant changes in nanostructure from the in-cylinder to the exhaust soot [15, 34]. One possible reason for this discrepancy is that the in-flame sampling locations in the authors' previous studies had been limited to the spray flame core regions. In these spray flame core regions, temperature  and concentration of oxidants such as OH [22, 23, 24, 25] are expected to be too low for the soot oxidation to proceed and exhibit notable changes in soot nanostructure. For better understanding of in-flame soot oxidation, sampling of soot particles from diesel spray flame periphery and TEM analyses of their nanostructure are necessary.
Although the authors had been aware of this importance of the peripheral soot sampling and TEM analyses, the authors' experience difficulties to directly sample soot from diesel flame downstream and periphery. The in-flame soot sampling technique in the authors' previous studies employed a "skim" type soot sampler as shown in Figure 1a, by which a TEM grid is directly exposed to a high temperature diesel spray flame. The primary mechanism of this soot sampling process is the thermophoresis caused by the temperature gradient within the thermal boundary layer formed on the relatively low temperature TEM grid surface. This sampling mechanism works well in the spray flame core where relatively high and constant spray axial velocity helps to maintain the steep temperature gradient within a thin boundary layer. However, in the downstream and periphery of diesel spray flame, lower and fluctuating spray velocity results in a thicker and unsteady boundary layer, making the soot sampling poorly repeatable or impossible. The present study firstly aims to enable repeatable soot sampling from diesel flame periphery while keeping the advantage of the simple and effective thermophoretic soot sampling.
In the present study, a novel soot sampler referred to as "suck" type sampler employing a high-speed solenoid valve was developed for the robust soot sampling in the periphery of the diesel spray flame. Firstly, in order to evaluate the usefulness and the reliability of the suck type sampler, the morphology of soot particles in the flame core sampled by the suck type and conventional skim type samplers was compared. Secondly, the suck type soot sampler was applied to the sampling of soot particles in the oxidation-dominant flame periphery and their morphology and nanostructure were compared with the ones of flame core soot.
Constant Volume Combustion Vessel
The soot sampling from a single-shot diesel spray flames was conducted in an optically-accessible constant-volume combustion vessel using two different types of soot samplers as shown in Figure 1. The constant-volume vessel employs three quartz windows, which have effective view field of [??]35 mm in diameter. The in-flame soot behavior during the sampling process was visualized via a high-speed visible laser shadowgraphy. A common-rail fuel injector with a single-hole nozzle was mounted on the head center and the single-shot fuel spray is injected in a horizontal direction in the combustion vessel. The soot samplers for collecting in-flame soot particles were placed in the vessel with a TEM grid held in a cavity at the sampler tip. The detailed explanation for the samplers is described in the next section, "Soot Samplers". In order to investigate the in-flame sooting tendency, the distance z from the injector nozzle orifice to the soot sampling locations was adjusted by selecting different thickness spacers between the combustion vessel and the head. The displacement of the constant volume vessel varies depending on the thickness of the spacer and is 560 cc for the standard case without the spacers. The combustion vessel, the TEM grid and the soot sampler were electrically pre-heated to 373 K to prevent water condensation.
This section describes two different soot samplers. One is the "skim" type soot sampler as shown in Figure 1a, which has been conventionally used in the previous studies by the authors. The skim sampler was designed to fix a [??]3 mm-diameter TEM grid parallel to the spray axis. The TEM grid is exposed directly to the diesel spray flame, which forms a temperature gradient between the flame and the grid surface. The soot particles trapped in the velocity boundary layer are conveyed and deposited onto the grid surface by the thermophoretic force generated from the temperature gradient. The sufficiently low temperature of grid (373 K) enables to instantly freeze reactions on the sampled soot particles and keep the in-flame soot morphology after the deposition. As explained above, this sampling mechanism works well in the spray flame core, but not in the downstream and periphery of diesel spray flame. For the robust thermophoretic soot sampling regardless of the in-flame sampling locations, it is desirable to hold the TEM grid as close to the sampling location as possible while inducing a flow of soot-laded gas parallel to the grid surface to maintain a relatively thin boundary layer.
In order to achieve this, the "suck" type soot sampler (Figure 1b) was newly designed to rapidly draw out a small portion of soot laden gas from the flame and induces the flow along the surface of a TEM grid hold inside the flow passage close to its entrance. The novel suck sampler is mounted to the combustion vessel by replacing the top quartz window with the sampler assembly. As shown in the cross-section view Figure 2, the suck sampler consists of a high-speed solenoid valve connected to a vacuum pump, a TEM grid holder and an inlet port of the soot laden gases. The solenoid valve (Parker Hannifin, Series 9 Pulse Valve) can operates under high ambient pressure (up to 8.5 MPa) and the shortest valve opening duration is 160 [mu]s by a driver for the exclusive use (Parker Hannifin, IOTA-ONE). This sampler is able to suck out a very small portion of the soot laden gases from the flame instantly. The grids were located close to its entrance as indicated by yellow circle and then exposed to the sucked soot laden gas flow. Potential drawbacks of this sampler are continued soot growth in the suction flow path and soot intrusion to the grid through the open flow path upstream of the solenoid valve at undesired timings. To balance these concerns, the TEM grid was located at 7.5 mm upstream from the inlet port tip. The preliminary study confirmed that the effects of the continued soot growth and soot intrusion are negligible at this location under the condition of the present study . The diameter of sucking flow passage was set to [??]1.5 mm for the sufficient spatial resolution of the soot sampling locations. For morphology and nanostructure observation of the inflame sampled soot particles , two different types of TEM grid, carbon coated copper and molybdenum micro grids, were used respectively and simultaneously exposed to the sucked gas flow. The radial distance r from the spray axis to the inlet port tip was adjusted by changing a spacer between the solenoid valve and the TEM grid holder.
Table 1 summarizes the experimental conditions. A diesel-like high-pressure and high-temperature ambient gas condition (pa=9.5 kg/[m.sup.3], [P.sub.a]=2.5 MPa, [T.sub.a]=1,070 K, 21%[O.sub.2]) was simulated by a spark-ignited combustion of a mixture composed of acetylene ([C.sub.2][H.sub.2]), oxygen ([O.sub.2]) and nitrogen ([N.sub.2]). The ambient gas density and pressure were set below a typical modern diesel in-cylinder condition, mainly due to the safety limitations of the experimental devices used in the present study. However, the TEM based morphology analysis of inflame soot particles sampled under a higher density and pressure condition ([[rho].sub.a]=22.8 kg/[m.sup.3], [[rho].sub.a]=6.7 MPa, [T.sub.a]=1000 K) [32, 33] did not show notable differences in an absolute value of the primary soot diameter and the general trend of the soot formation and oxidation processes on the spray axis, compared with the previous studies conducted under the identical conditions. Therefore, the obtained results in the present study are considered to be useful for discussing the in-flame diesel soot oxidation processes. JIS#2 Japanese conventional diesel fuel was used as the test fuel. Figure 3 shows the examples of time-sequence and time-integrated high-speed laser shadowgraphs of this spray flame without the sampler under the identical ambient conditions with the sampling cases. The liquid core length of the spray has been measured to be around 20mm at the present condition and the shadow in Figure 3 is mainly attributed to soot cloud. The horizontal arrows in the figure indicates the TEM grid exposing duration of the conventional skim type sampler and the solenoid valve opening duration of the novel suck type sampler. The dots plotted on the time-integrated shadowgraphs indicates the soot sampling locations in this spray flame by each types of the sampler. Note that the sampler intrusion to the spray flame indeed has a significant influence on the spray and flame propagation in the upstream of the spray flame beyond the sampler intrusion location. However, for the downstream from the nozzle to the sampler intrusion locations, notable differences of the spray and flame propagation have not been experimentally observed between the cases with and without the sampler. Therefore, the sampled soot is reasonably considered equivalent to the ones from the original spray flame. In addition, the soot sampling experiments had been carried out only at the mid- to downstream locations where the sampler intrusion does not affect the ignition. Regarding the suck sampling, the solenoid valve was kept open for 2.5 ms in all sampling locations, which corresponds to the duration from the spray tip arrival to the tail passage.
Firstly, the morphology of soot particles sampled by the suck and skim type samplers was compared and the usefulness and the reliability of the novel suck sampler was evaluated. The soot particles were sampled from in-flame core regions at 4 different spray axis locations, z=60, 80, 100 and 120 mm away from the injector nozzle tip, which correspond to midstream peak soot to the downstream oxidation-dominant regions in this spray flame. Secondly, soot particles in the peripheral region of spray flame were sampled by the novel suck type sampler and compared with the ones sampled at in-flame core regions. The peripheral soot sampling was conducted at three different locations, z=60 mm: r=6mm, z=80 mm: r=12 mm, z=100 mm: r=12 mm for axial and radial distances respectively, which were determined based on the high-speed laser shadowgraphs.
HR-TEM Observation of Sampled Soot Particles
The sampled soot particles on the TEM grid surface were observed using an HR-TEM (JEOL, JEM-2100F, an acceleration voltage of 200 keV, a point resolution of 0.19 nm). All images were digitally recorded with a CCD camera (Gatan, Ultrascan 1000/FIRST LIGHT, image resolution of 2048x2048 pixels). The grids were observed without any thermal or chemical pre-treatments to maintain the original properties of sampled soot particles produced in the diesel spray flames. Figure 4 shows two different types of the TEM grids used for the soot sampling experiments as explained in the previous section and example TEM images of in-flame soot particles. TEM observation was conducted at 5 different on-grid locations with more than 5 images from each location to compensate their relatively large morphology fluctuation observed in the authors' previous study . TEM observations were conducted at three different magnifications with x6,000, x20,000 and x500,000 . The x6,000 magnified images were binarized and the projection area ratio [A.sub.r] was calculated to compare sampled on-grid soot concentration between the skim and suck type sampler cases. The x20,000 images were used to observe soot morphology qualitatively. The highest magnification x500,000 is useful for observation of soot nanostructure. Since the view field size is indeed inversely proportional to the magnification, the view field of x500,000 nanostructure image often covers only one primary particle as seen in Figure 4c. Therefore, in order to grasp a whole image of the aggregate with detailed nanostructure, several high-mag images of primary particles constituting one aggregate were taken as shown in Figure 5 and stitched image will be discussed in the later section, "Soot Nanostructure at Oxidation-dominant Flame Periphery". Note that this high-mag observation needs to be carried out in short period of time to avoid notable damages on the soot nanostructure by intense electron beam irradiation.
RESULTS AND DISCUSSION
The Results and Discussion is divided into two different sections. Firstly, the morphology of soot particles sampled by two different samplers are compared qualitatively via TEM observation to evaluate the usefulness and reliability of the newly designed suck sampler. Secondly, the nano-structural changes of soot particles sampled at the oxidation-dominant flame periphery observed by the HR-TEM image stitching are presented.
Comparison between "Suck" and "Skim" Type Soot Samplers
As mentioned in the previous section, the "suck" type sampler was developed to achieve the robust soot sampling at the oxidation-dominant flame periphery. The novel sampler needs to collect sufficient amount of soot particles with high-repeatability without affecting the in-flame soot morphology and nanostructure during the sampling process especially at the flame downstream and periphery where the sampling become challenging by the conventional sampler. In-flame soot sampling experiments were conducted by both types of the samplers at several different axial locations (z=60, 80, 100 and 120 mm) in the spray flame core region. Then TEM observation was conducted to investigate whether the novel sampler affects the sampled soot morphology or not. Furthermore, the sampling experiments were performed several times at each location and the repeatability of the sampled soot amount was examined.
Figure 6 shows example TEM images of soot particles sampled by the conventional skim (top) and novel suck (bottom) type samplers at 4 different axial locations in the flame core (r=0 mm). In order to observe the sampled soot amount and morphology, two different magnifications, (a) x6,000 and (b) x20,000, were used and the sampling locations are indicated at the top of the figure. As will be explained in the later section, the soot amount sampled by the conventional skim sampler exhibited a significant fluctuation especially in the downstream locations thus the TEM images presented here are the representative on-grid soot concentration at each sampling location. Note that this sampling fluctuation is expected to become more significant in flame periphery than in flame downstream due to steeper velocity gradient of diesel spray in radial direction than in axial direction.
In Figure 6, the x6,000 magnification images show that the sampled on-grid soot concentration decreases from the midstream (z=60 mm) to the downstream (z=120 mm) regardless of the sampler type. This trend, decrease towards downstream, is often observed in the previous studies [26, 27, 28]. As seen in the x20,000 images, the sampled in-flame soot are a mixture of separately existing single primary particles and aggregates consisting of several primaries especially around the z=60 and 80 mm. From the qualitative observation of the soot morphology, the aggregate and primary particle size is quite similar to each other at all sampling locations. Additionally, the fractal structure of aggregates does not show notable difference between the samplers. The slight decrease of the aggregate size towards the downstream due to oxidation had also been found in previous studies [26, 27, 28]. From the qualitative observation, the soot particles sampled by the novel suck sampler exhibit quite similar morphology to that of the conventional skim sampler case. This represents that the sampling process of the newly developed sampler does not affect the morphology of sampled soot significantly at among axial locations but not among the heavily oxidized radial locations.
The above explained soot sampling and TEM observation were repeated more than twice at each location to facilitate the repeatability analysis based on a larger number of TEM images. The obtained fluctuation of sampled on-grid soot concentration is shown in Figure 7a. The histograms shown in the figure were created by plotting the normalized frequency of the projection area ratio or the relative amount of soot on each individual TEM image out of 50 to 100 TEM images for different sampling locations and samplers. The histograms for the conventional skim and novel suck sampler cases are plotted as blue and red lines, respectively. The number of grid samples and processed images are indicated on the right of each histogram. The example images for the most downstream, z=120 mm, were also displayed in Figure 7b. As qualitatively seen in the x6,000 magnification images as in Figure 6, the on-grid concentrations by both samplers are gradually decreasing from the midstream to the downstream probably due to the oxidation.
The sampled on-grid soot concentration shows large fluctuations regardless of the sampler type at z=60 and 80 mm, possibly due to the combined effects of the in-flame soot inhomogeneity, shot-by-shot fluctuation and the fluctuation of the sampling process. Author's previous investigations of time-resolved soot concentration fluctuation in the flame core using a novel "scope-rod window" technique  indicates that the in-flame soot inhomogeneity is significant enough to explain the fluctuation. As a general trend, the concentration fluctuation reduces towards the downstream for both sampler cases that could be reflecting the reduced soot concentration itself. However, at the downstream locations at z=100 and 120 mm, the histograms for the novel suck sampler exhibit notably narrower distribution or smaller fluctuation compared to the conventional skim sampler case. As a result of the larger fluctuation, the example TEM images for the skim sampler case also frequently appeared as empty, while the "suck" sampler constantly collected adequate amount of soot particles. This poorly repeatable soot sampling by the conventional skim sampler is considered due to lower and fluctuating spray velocity in the flame downstream resulting in a thicker and unsteady boundary layer on the grid surface. The suck sampler, on the other hand, can effectively improve the sampling repeatability by inducing a flow of soot-laded gas parallel to the grid surface to maintain a relatively thin boundary layer with a steep temperature gradient. These results suggest that the newly developed suck sampler improves the repeatability of soot sampling without affecting the in-flame soot morphology and thus ensures sampling of sufficient amount of on-grid soot required for statistically reliable TEM morphology analysis. In the next section, this newly developed technique is applied for the first time to sample soot from the oxidation-dominant flame periphery where the lower and fluctuating spray velocity nature, is similarly observed.
Soot Nanostructure at Oxidation-Dominant Flame Periphery
Based on the discussion in the previous section, the soot particles at the oxidation-dominant periphery of spray flame were sampled by the newly developed suck type sampler and qualitatively compared with the ones at the core in midstream and downstream locations. Figure 8 shows example HR-TEM images of the whole aggregates typically found at flame core mid- (z=60 mm) and downstream (z=100 mm) and downstream periphery (z=100 mm, r=12 mm). The time integrated laser shadowgraph of the spray flame is shown in the mid-right of the figure to show the overall sooting tendency. The sampling locations, mid- and downstream core and downstream periphery are indicated as blue, purple and red dots, respectively on the shadowgraph. The sampling of flame-peripheral soot was conducted also at two other locations (z=60 mm: r=6 mm, z=80 mm: r=12 mm) as indicated as gray dots. The HR-TEM images shown in Figure 8c are representative examples of these flame-peripheral soot. The dark shadow at around z=60 mm suggests that the midstream core is formation dominant region in this spray, while the slightly lightened shadow at the core of z=100 mm is possibly due to the oxidation. The red and gray dots correspond to the edge of shadow region where the soot oxidation by OH is expected to be dominant.
The HR-TEM images of soot show that for the soot sampled at the midstream core region, the nanostructure exhibits layered and ordered graphitic structure, as previously observed for the in-flame core and engine exhaust soot [8, 29]. On the other hand, the peripheral soot shows clearly different nanostructure. The outer layers of the peripheral soot particles exhibit lumpy surfaces, which is possibly due to stripping and disintegration of the layers by rapid oxidation. Although it is not easy to identify the boundaries of primary particle within these aggregate because of the lumpy surface, an apparent primary particle size of the peripheral soot seems clearly smaller than that of the flame core soot. These peripheral soot nano-structural features were consistently observed for the other peripheral sampling locations mentioned above. These observations suggest that the diesel in-flame soot oxidation starts from the outer shell of the primary particles and proceeds to their inner core.
The black arrows in the Figure 8c indicate example locations of another feature observed on the peripheral soot, blurred outline boundaries of the primary particles. These blurs were frequently observed on the aggregates sampled from the periphery exhibiting wholly collapsed outer layers especially at the perimeter of the aggregates. These blurs were hardly observed on the midstream core soot which exhibited clearly layered structure at the outermost boundary as shown in Figure 8a. The downstream core soot only occasionally or partially exhibited these blurs as indicated by white arrows in Figure 8b. These blurs may be the fragments of soot particles right before the complete collapse by oxidation. Note that these blurs are not the results of the soot particle damaging due to the above mentioned electron beam focusing during high-mag observations.
These observations are consistent with the authors' previous results that soot nanostructure does not show significant variation towards downstream in the flame core where temperature and concentration of OH are expected to be too low for the soot oxidation to rapidly proceed and exhibit notable changes in soot nanostructure . The partially observed nano-structural features such as the smaller primary particle size and the blurs at the downstream core may correspond to the first stage of the in-flame soot oxidation which starts from the edge sites of the aggregates.
In contrast to the above explained diesel in-flame soot oxidation, soot particles oxidized in DPF and furnace frequently exhibit hollow structure with graphitized outer shells as reported by several researchers [11, 13, 14]. The inner core of soot particle is known to be more amorphous and less resistant to oxidation . Therefore, under the relatively low temperature (~800 K) in DPF, the soot oxidation gradually starts from the inner core, while the outer shells become further graphitized and more resistant to oxidation due to annealing effects. Stanmore et al. discuss in their review on experiments and models of soot oxidation that significant oxidant penetration into soot particles occurs under lower temperatures, while increased oxidative reactivity decreases the penetration under higher temperatures . Neoh et al. also suggest that oxygen molecules tend to penetrate into soot particles and collapse their inner core resulting in internal burning, while OH radicals hardly penetrate into soot particles due to their high reactivity . The above explained present observation in diesel in-flame soot oxidation and the discussions in literatures on soot oxidation by different oxidants are consistent each other. In addition, it confirm that in diesel in-flame soot oxidation OH is the important oxidant and rapidly oxidizes the soot particles from the outer surface towards the inner core regardless of the graphitic/amorphous nanostructures.
Another and more important finding of the present study is that the engine exhaust soot which exhibits increasingly graphitic structure during the exhaust process  is actually similar to the flame core soot before oxidation. The engine exhaust soot is significantly different from the flame periphery soot which exhibits mostly amorphous structure, lumpy outer layers, apparently smaller primary particle size and blurs or fragmental particles. In addition to the qualitative comparisons of the peripheral and engine-out soot, authors are now working on the quantitative image analysis of these soot nanostructure  using a MATLAB-based software developed by the authors  based on the algorithm by Yehliu . The analysis results, to be published elsewhere, suggest that the crystal nano-layer separation notably increases from the flame core to periphery. This is possibly due to the disintegration of the tighter graphitic outer layers by rapid oxidation at the flame periphery and survival of the looser amorphous inner core within the primary particles. On the other hand, the variation of the nano-layer separation of the engine soot from in-cylinder TDC to the exhaust exhibits the opposite trend and clearly decreases as reported by Li et al. . It is therefore suspected that the engine exhaust soot is not the remains of incomplete or partial oxidation, but the runaways escaped from the flame core to the exhaust without being attacked by in-flame OH radicals. Enhanced in-flame encounter between soot particles and OH radicals might therefore be an effective approach for soot emission reduction. For further understanding of the diesel in-flame soot oxidation process, authors are preparing for high-speed UV laser shadowgraphy imaging of the time-sequential interaction between soot and fundamental OH radicals (not the chemiluminescent excited OH*) around the flame periphery, coupled with the HRTEM morphology analysis of the soot particles being oxidized. The combination of time-sequential optical diagnostics and sampling/TEM investigation is expected to provide useful information which conventional single-shot diagnostics based on the planar OH-LIF [22, 23,42, 43] could not reveal.
For better understanding of in-flame diesel soot oxidation processes, soot particles at the oxidation-dominant periphery of diesel spray flame were sampled by a newly developed high-speed suck type soot sampler and their morphology and nanostructure were observed via HR-TEM. Firstly, in order to evaluate the usefulness and the reliability of the suck type sampler, the morphology of soot particles in the flame core sampled by the suck type and conventional skim type samplers was compared. Secondly, the suck type soot sampler was applied to the sampling of soot particles in the oxidation-dominant flame periphery and their morphology and nanostructure were compared with the ones of flame core soot. The obtained conclusions are as follows.
1. The morphology of soot particles in the flame core sampled by the suck type and conventional skim type samplers did not show notable differences based on qualitative observation. The on-grid concentration of soot particles sampled by the novel suck sampler at downstream in-flame locations exhibited notably smaller fluctuation compared to the ones by conventional skim sampler. These results suggest that the suck sampler can effectively improve the soot sampling repeatability without affecting the in-flame soot morphology, which enables robust soot sampling from the peripheral and downstream regions in diesel spray flame.
2. The nanostructure of soot particles sampled at the oxidation-dominant in-flame periphery by the suck type sampler showed that the outer layers of the soot particles exhibit lumpy surfaces which is possibly due to stripping and disintegration of the layers by rapid oxidation and makes it difficult to identify boundaries of primary particles within aggregates. The apparent primary particle size was smaller than that of the flame core soot, suggesting that the diesel in-flame soot oxidation starts from the outer shell of the primary particles and proceeds to their inner core.
3. The high-resolution images of the whole soot aggregates produced by the stitching of multiple HR-TEM images showed that blurs or fragments of soot particles, in addition to the above mentioned lumpy surfaces and smaller primary particle size, were partially seen in the aggregates sampled from the downstream core region which may correspond to the first stage of the inflame soot oxidation. On the other hand, the aggregates sampled from the flame periphery mostly exhibited wholly collapsed outer layers especially at the perimeter of the aggregates.
4. The nanostructure of engine exhaust soot found in literature was similar to the flame core soot before oxidation and significantly different from the flame periphery soot exhibiting above mentioned oxidation-induced nano-structural features. It is therefore suspected that the engine exhaust soot is not the remains of incomplete or partial oxidation, but the runaways escaped from the flame core to the exhaust without being attacked by in-flame OH radicals.
[1.] Kamimoto T., Uchida N., Aizawa T., Kondo K. and Kuboyama T., "Diesel Flame Imaging and Quantitative Analysis of In- Cylinder Soot Oxidation," International Journal of Engine Research (2016), pp.1-14.
[2.] Nagle J. and Strickland-Constable R. F.., "Oxidation of Carbon between 1000-2000C," Fifth Carbon Conference, Vol.1, pp.154-164, (1962).
[3.] Neoh K. G., Howard J. B. and Sarofm A. F.., "Soot Oxidation in Flames," Particulate Carbon Formation during Combustion, pp.261-282, (1981).
[4.] Stanmore B, Brilhac J and Gilot P. "The oxidation of soot: a review of experiments, mechanisms and models." Carbon 2001; 39: 2247-2268.
[5.] Liu, Y., Tao, F., Foster, D., and Reitz, R., "Application of A Multiple-Step Phenomenological Soot Model to HSDI Diesel Multiple Injection Modeling," SAE Technical Paper 2005-01-0924, 2005, doi:10.4271/2005-01-0924.
[6.] Tao, F., Foster, D., and Reitz, R., "Soot Structure in a Conventional Non-Premixed Diesel Flame," SAE Technical Paper 2006-01-0196, 2006, doi:10.4271/2006-01-0196.
[7.] Cheng X., Chen L., Hong G., Yan F. and Dong S., "Modeling Study of Soot Formation and Oxidation in DI diesel engine using an improved soot model," Appl. Therm. Emg., Vol. 62 (2014), pp.303-312.
[8.] Ishiguro. T., Takatori. Y., and Akihama K., "Microstructure of Diesel Soot Particles Probed by Electron Microscopy: First Observation of Inner Core and Outer Shell," Combust. Flame, Vol. 108 (1997), pp.231-234, doi: 10.1016/S0010-2180(96)00206-4.
[9.] Vander Wal, R. L. and Tomasek, A. J., "Soot Oxidation: Dependence upon Initial Nanostructure," Combust. Flame, Vol. 134 (2003), pp.1-9.
[10.] Vander Wal, R. L. and Tomasek, A. J., "Soot Nanostructure: Dependence upon Synthesis Conditions," Combust. Flame, Vol. 136 (2004), pp.129-140.
[11.] Song, J., Alam, M., Boehman, A. L. and Kim, U., "Examination of the Oxidation Behavior of Biodiesel Soot," Combust. Flame, Vol. 146 (2006), pp.589-604.
[12.] Vander Wal, R. L., Yezerets, A., Currier, N. W., Kim, D. H. and Wang C. M., "HRTEM Study of Diesel Soot Collected from Diesel Particulate Filters," Carbon, Vol. 45 (2007), pp70-77.
[13.] Strzelec, A., Vander Wal, R. L., Toops, T. J. and Daw, S., "Nanostructure and Burning Mode of Light Duty Diesel Particulate with Conventional Diesel, Biodiesel and Intermediate Blends," International Journal of Engine Research, pp. 1-12, (2016). doi:10.1177/1468087416674414.
[14.] Ishiguro, T., Suzuki N., Fujitani Y. and Morimoto H., "Microstructural Changes of Diesel Soot during Oxidation," Combust. Flame, Vol. 85 (1991), pp.1-6.
[15.] Li, Z., Song, C., Song, J., Lv, G., Dong, S., and Zhao, Z., "Evolution of the Nanostructure, Fractal Dimension and Size of In-Cylinder Soot during Diesel Combustion Process," Combust. Flame, Vol. 158, No.8 (2011), pp.1624-1630.
[16.] Ye Peng, Sun Chenxi, Lapuerta Magin, Agudelo John, Wal Randy Vander, Boehman Andre L, Toops Todd J and Daw Stuart, "Impact of rail pressure and biodiesel fueling on the particulate morphology and soot nanostructure from a common-rail turbocharged direct injection diesel engine," International Journal of Engine Research, (2016).
[17.] Yehilu, K., Armas, O., Vander Wal, R. L. and Boehman, A. L., "Impact of Engine Operating Modes and Combustion Phasing on the Reactivity of Diesel Soot," Combust. Flame, Vol. 160 (2013), pp.682-691.
[18.] Al-Qurashi K and Boehman AL., "Impact of exhaust gas recirculation (EGR) on the oxidative reactivity of diesel engine soot," Combust. Flame; Vol. 155 (2008): pp.675-695.
[19.] Roth P., Brandt O., Von Gersum S., "High temperature oxidation of suspended soot particles verified by carbon monoxide and carbon dioxide measurements," Symposium (International) on Combustion 23 (1990) 1485.
[20.] Fenimore C.P., Jones G.W., "Oxidation of soot by hydroxyl radicals," J. Phys. Chem. A 71 (1967) 593.
[21.] Mulcahy M.F.R., Young B.C., "Reaction of hydroxyl radicals with carbon at 298 K," Carbon, Vol. 13 (1975), pp.115-124.
[22.] Kosaka, H., Aizawa, T. and Kamimoto, T., "Two-Dimensional Imaging of Ignition and Soot Formation Processes in a Diesel Flame," International Journal of Engine Research, Vol. 6, No. 1 (2005), pp.21-42.
[23.] Dec, J., "A Conceptual Model of DI Diesel Combustion Based on Laser-Sheet Imaging*," SAE Technical Paper 970873, 1997, doi:10.4271/970873.
[24.] Ito, T., Hosaka, T., Ueda, M., Senda, J. et al., "Detailed Kinetic Modeling and Laser Diagnostics of Soot Formation Process in Diesel Jet Flame," SAE Technical Paper 2004-01-1398, 2004, doi:10.4271/2004-01-1398.
[25.] Kosaka, H., Nishigaki, T., Kamimoto, T., and Harada, S., "A Study on Soot Formation and Oxidation in an Unsteady Spray Flame via Laser Induced Incandescence and Scattering Techniques," SAE Technical Paper 952451, 1995, doi:10.4271/952451.
[26.] Kondo, K., Yamaguchi, T., Nishigai, H., Takano, S. et al., "High-Resolution Transmission Electron Microscopy of Soot Directly Sampled at Different Axial Locations in Diesel Spray Flame," SAE Technical Paper 2011-24-0068, 2011, doi:10.4271/2011-24-0068.
[27.] Yamaguchi, T., Kondo, K., Nishigai, H., Takano, S. et al., "Direct Sampling, TEM Analysis and Optical Measurement of Soot Particles at Different Axial Locations in a Transient Spray Flame," SAE Int. J. Fuels Lubr. 5(1):316-328, 2012, doi:10.4271/2011-01-2051.
[28.] Kondo, K., Yamaguchi, T., Nishigai, H. and Aizawa, T., "Transmission electron microscopy of soot directly sampled at different axial locations in diesel spray flame," Transactions of the JSME (in Japanese), Vol.81, No.830 (2015).
[29.] Aizawa, T., Takahata, N., Okabe, K., and Mizutani, Y., "Effect of Fuel Aromatics on In-Flame Diesel Soot Nanostructure via HRTEM," SAE Technical Paper 2015-01-1829, 2015, doi:10.4271/2015-01-1829.
[30.] Okabe, K., Sakai, M., Mizutani, Y., and Aizawa, T., "Aromatic Additive Effect on Soot Formation and Oxidation in Fischer-Tropsch Diesel (FTD) Spray Flame -Morphology and Nanostructure Analysis of InFlame Soot Particles via HRTEM-," SAE Int. J. Fuels Lubr. 6(3):807-816, 2013, doi:10.4271/2013-01-2681.
[31.] Kook, S. and Pickett, L.M., "Soot volume fraction and morphology of conventional and surrogate jet fuel sprays at 1000-K and 6.7-MPa ambient conditions", Proc. Combust. Inst., Vol.33, no.2, pp.2911-2918, (2011).
[32.] Aizawa, T., Nishigai, H., Kondo, K., Yamaguchi, T. et al., "Transmission Electron Microscopy of Soot Particles Directly Sampled in Diesel Spray Flame - A Comparison between US#2 and Biodiesel Soot," SAE Int. J. Fuels Lubr. 5(2):665-673, 2012, doi:10.4271/2012-01-0695.
[33.] Nerva, J., Yamaguchi, T., Iguma, H., Nishigai, H. et al., "Transmission Electron Microscopy of Soot Particles sampled directly from a Biodiesel Spray Flame," SAE Technical Paper 2011-01-2046, 2011, doi:10.4271/2011-01-2046.
[34.] Wei, J., Song, C., Lv, G., Song, J., Wang, L. and Pang, H. "A comparative study of the physical properties of in-cylinder soot generated from the combustion of n-heptane and toluene/ n-heptane in a diesel engine," Proceedings of the Combustion Institute 35(2), pp.1939-1946 (2015)
[35.] Aizawa, T. et al., "Thermocouple temperature measurements in diesel spray flame for validation of in-flame soot formation dynamics," International Journal of Engine Research (2017), doi: 10.1177/1468087416657280.
[36.] Takahata, N., Toyama, Y., Kondo, H. and Aizawa, T. "Soot Oxidation in Periphery of Diesel Spray Flame via High-Speed Sampling and HRTEM Observation," The 27th Internal Combustion Engines Symposium (in Japanese), Japan (2016)
[37.] Kondo, K., Aizawa, T., Kook, S., and Pickett, L., "Uncertainty in Sampling and TEM Analysis of Soot Particles in Diesel Spray Flame," SAE Technical Paper 2013-01-0908, 2013, doi:10.4271/2013-01-0908.
[38.] Neoh KG, Howard JB, Sarofim AF., "Effect of oxidation on the physical structure of soot," In: Twentieth international symposium on combustion, Pittsburgh: The Combustion Institute, 1984, pp. 951-7.
[39.] Vander Wal, R., "Soot Nanostructure: Definition, Quantification and Implications," SAE Technical Paper 2005-01-0964, 2005, doi:10.4271/2005-01-0964.
[40.] Sakai, M., Iguma, H., Kondo, K., and Aizawa, T., "Nanostructure Analysis of Primary Soot Particles Directly Sampled in Diesel Spray Flame via HRTEM," SAE Technical Paper 2012-01-1722, 2012, doi:10.4271/2012-01-1722.
[41.] Yehliu, K., Vander Wal, R. L. and Boehman, A. L., "Development of an HRTEM image analysis method to quantify carbon nanostructure," Combust. Flame, Vol. 158, No. 9 (2011), pp. 1837-1851.
[42.] Bruneaux, G., "Combustion Structure of Free and Wall Impinging Diesel Jets by Simultaneous Laser Induced Fluorescence of Formaldehyde, PAH and OH," International Journal of Engine Research, Vol. 9, Issue 3, pp.249-265, 2008.
[43.] Skeen, S., Manin, J., Pickett, L., Cenker, E. et al., "A Progress Review on Soot Experiments and Modeling in the Engine Combustion Network (ECN)," SAE Int. J. Engines 9(2):883-898, 2016, doi:10.4271/2016-01-0734.
The authors are sincerely grateful to Hideyuki Yoshimura of Department of Physics, Meiji University for providing the opportunity to use the HRTEM for soot observation.
Yoshiaki Toyama, Nozomi Takahata, Katsufumi Kondo, and Tetsuya Aizawa
Table 1. Experimental conditions Surrounding conditions Ambient gas density [[rho].sub.a] 9.5 kg/[m.sup.3] Ambient pressure [P.sub.a] 2.5 MPa Ambient temperature [T.sub.a] 1070 K [O.sub.2] concentration 21% Injection conditions Injector type G2S (Solenoid) Nozzle orifice [??]0.14 mmx1 Fuel JIS#2 Injection pressure [DELTA][P.sub.inj] 80 MPa Injection duration [DELTA][t.sub.inj] 2.3 ms Injection amount [Q.sub.inj] 10.3 mg Soot sampling conditions Sampler types Skim and Suck Distance from nozzle tip z 60, 80, 100 and 120 mm Distance from spray axis r 0, 6 and 12 mm TEM grid types Carbon-coated copper grid Molybdenum micro-grid Temperature of sampler [T.sub.sampler] 373 K
|Printer friendly Cite/link Email Feedback|
|Author:||Toyama, Yoshiaki; Takahata, Nozomi; Kondo, Katsufumi; Aizawa, Tetsuya|
|Publication:||SAE International Journal of Engines|
|Date:||Dec 1, 2017|
|Previous Article:||Water Injection: a Technology to Improve Performance and Emissions of Downsized Turbocharged Spark Ignited Engines.|
|Next Article:||Functional Requirements to Exceed the 100 kW/l Milestone for High Power Density Automotive Diesel Engines.|