2.1. The Nature of Soots
shows the results of elemental analysis (C, N, H, S) performed on a Costech ECS-4010 elemental analyzer. The results reveal that the majority of constituents in diesel particulates and model soot were carbon, oxygen, and hydrogen, with low concentrations of nitrogen and sulphur. In comparison, the contents of N, H, and O (contributed from hydrocarbon and adsorbents) in engine soot are significantly higher than those in the model soot. The change of elemental compositions for flame soot collected at different flame heights is negligible.
shows TEM images of Printex-U carbon black (CB), Tiki©
flame soot FS100, and engine diesel soot (DS). It can be determined that there is a range of particles sizes and the average particle sizes for CB, FS100, and DS are 35 nm, 45 nm, 30 nm respectively. The FS100 particles exhibit slightly larger aggregate and particle size than CB and DS. Figure 2
a–c depict the high-resolution transmission electron microscopy (HRTEM) images of carbon black (CB), flame soot FS100, and real engine soot (DS) respectively. In these images, the graphene layer segments comprising the crystallite stacks are observed as the dark lines blocking/scattering the incident electron beam [10
]. The flame derived soot (Figure 2
b) exhibits short, individual graphene segments that appear to have no orientation relative to each other. Therefore, the flame soot possesses little long-range order and no discernable crystallite structure. This lack of structure is characteristic of disordered carbon. In contrast, the engine soot shown in Figure 2
c has extended carbon lamellae, many of which are oriented parallel to each other and approximately parallel to the particle surface, characteristic of a crystalline structure. For comparison, the HRTEM image of Printex-U carbon black shown in Figure 2
a presents a typical carbon soot nanostructure. Some particles clearly show a recognizable central nucleus. The lamellae can distinctly be seen and show less stacking order while being shorter than in the engine soot particles.
a shows the XRD patterns of the Tiki flame soot FS100, carbon black and diesel soot. All samples show diffraction maxima associated with the (002) and (100) planes. The broad peaks and diffuse nature of the XRD profiles indicate the presence of amorphous carbon. The first peak of the profile corresponds to the (002) peak of the ordered hexagonal graphite structure. The second peak corresponds to the (100) and (101) peaks. Before analysis the diffraction profiles were corrected for absorption and normalization [23
]. The corrected intensities of the samples were then used for further quantitative analysis, including the average crystalline diameter (via the Scherrer equation) and the fraction of crystallinity.
The crystallinity of carbon in the soot was obtained from the integrated intensities of the (002) reflection peak and the amorphous hump background in the range of 2θ
= 18–32°. Both the flame soot FS100 and Printex-U carbon black are mostly amorphous phase with obvious diffraction peaks being absent, while the real engine soot presents a sharp diffraction peak at 2θ
= 26.5°. Figure 3
b shows the final fit of the XRD pattern of real engine soot in the angular range 2θ
= 18–32°. Prior to ascribing the peak to graphitic component, we carefully examined XRD patterns of impurities from engine exhaust (e.g., ash powders) with Si powders as the standard reference. The crystallinity of the real soot sample can be calculated from the integrated intensities of the apparent diffraction peak (A
) and the amorphous hump (B
), and the calculated crystallinity of engine soot
was approximately 10%.
represent the crystalline heights of carbon grain along the c-axis and the a-axis respectively. Table 2
compares the calculated crystal parameters for various soots, where K
and denotes the average number of layers per crystal. The values of d002
for carbon black and flame soot are all substantially greater than that reported for crystalline layered hexagonal graphite (3.354 nm) [PDF#751621]. For the real diesel soot, both an amorphous hump and a broadened peak are present at the (002) position, and the corresponding d002
are 3.593 and 3.364 respectively. This implies that the real diesel soot is partially crystallized, being a mixture of amorphous carbon and nanoscale graphite-like carbon. The change of d002
for flame soot collected at different heights is negligible because all soot samples were collected above the inception zone where the soot formation is thought to be complete [23
]. Meanwhile, the d002
for both flame soot samples and commercial Printex-U are comparable with the d002
of the amorphous phase in real engine soot, at a value of 0.358~0.360 nm.
The crystallite heights (Lc
) of Printex-U and flame soot were found to be smaller than that of engine soot. Additionally, the Lc
of the flame soot was found to be constant at 1.41~1.42 nm as the flame height increased. According to Hurt et al. [13
], face-to-face association of the graphene sheets is favored at low temperatures over edge-to-edge coalescence. Hence, the conditions during soot formation might be the main factor causing the differences in the structural parameter of different soot. Conversely, the change of crystallite width (La
) is negligible for all soot.
While crystalline parameters do not change with respect to flame height, it appears that agglomerate size does. Figure 4
shows the size distribution of agglomerates for the soot collected from different flame heights. It was found that the soot from the lowest flame height has smaller agglomerates with a mean size of 195 nm. The agglomerates grow to 460 nm when the flame height is 15 cm from the burner.
2.2. Temperature-Programmed-Oxidation (TPO)
a shows the non-catalytic oxidation curves for Printex-U carbon black, FS100 and engine soot. Samples were heated at 20 °C/min in a 10% O2
gas flow. For real engine soot, two regions of mass loss were observed. In the lower temperature region from 300–500 °C, around 20~25% weight loss resulted from the oxidation of adsorbed hydrocarbons (HCs), while in the higher temperature region from 500–650 °C, carbon soot was burned off. Both flame soot FS100 and Printex-U carbon black have relatively low amounts of HCs, and only one region of weight loss was observed from 550–650 °C. The Tig
for FS100 and Printex-U carbon black are 575 °C and 580 °C respectively. It was also found that the engine soot showed a higher combustion completion temperature than flame soot FS100 and Printex-U carbon black.
Typical uncatalyzed isothermal oxidation (at 525 °C) results for Printex-U carbon black, flame soot FS100 and real soot are shown in Figure 5
b. For all combustion data, the reaction rates reach a maximum at approximately 5% conversion, most likely due to the lag after switching from inert gas to the oxidizing reaction gas. The reaction rates usually remain relatively high during the initial 10% conversion, and this can be attributed to oxidation of more reactive species in the soot (e.g., absorbed HCs) during this stage of the experiment. The curves of reaction rate versus conversion for Printex-U carbon black and flame soot FS100 show an exponential-like behavior, while the reaction rate of diesel soot shows a plateau in the conversion range of 0.25–0.55. This plateau is also visible in other combustion curves of diesel soot at different temperatures.
shows the oxidation curves for flame soot samples that were collected from different flame heights. The initial reaction rates increased with decreasing distance from the burner. The maximum reaction rate for FS20 is more than double that of FS150, whereas for FS50 and FS100 the reaction rates are very similar. When the conversions were larger than 0.4, the oxidation curves for all samples dropped to a comparable range. A correlation between the oxidation behavior and distance from the burner is discussed below.
2.3. Kinetics Analysis Based on Nonisothermal Experiments
The conversion–temperature TGA curves for flame soot FS100 recorded at various constant heating rates are shown in Figure 7
a. When the model-free kinetic analysis method was applied, the apparent activation energy (Ea
) was determined from the slope of
at a constant X
is the extent of conversion). The corresponding results are shown in Figure 7
The conversion vs. temperature TGA curves showed a noticeable dependence on the heating rate, such that the conversion curves were shifted to higher temperatures at higher heating rates. The activation energies and pre-exponential factors at different degrees of conversion were calculated, and it was found that the activation energy varied with the extent of conversion from 137.4 kJ/mol to 154.3 kJ/mol with a mean value of 147 kJ/mol.
The noticeable dependence of the activation energy on the extent of conversion (X) indicated that the assumption of a single-step reaction is not accurate enough in the case of flame soot oxidation. A similar variation of Ea at different conversion levels was also observed for Printex-U carbon black and real engine soot, and the mean values of Ea for these soot samples were 145 kJ/mol and 137 kJ/mol respectively.
2.4. Kinetic Analysis from Isothermal Studies (nth Order Model)
A comparative kinetics study was performed under isothermal conditions for non-catalytic oxidation of FS100. The reaction order of carbon can be determined from a series of isothermal oxidation curves at different temperatures with fixed oxygen partial pressure (10 vol % oxygen in nitrogen). The influence of oxygen partial pressure was studied in the range of 10–40% O2.
The reaction orders of carbon were determined by fitting combustion curves as shown in Figure 8
a with the nth order model. ln(reaction rate) was plotted as function of ln(1 − x
) in Figure 8
b. The nth order model fits well to the combustion curves for flame soot in the conversion range of 0.2 to 0.9 with a mean value of 0.78 for the reaction order of carbon. The standard deviation of the determined reaction order is less than 0.1.
To determine the influence of the oxygen partial pressure on the conversion rate, isothermal experiments at various partial pressures (10% to 40%) were performed. The isotherm temperature was chosen to be 525 °C (a temperature in kinetics regime). For each partial pressure, the conversion rate was calculated at defined conversions (X
= 0.2 to 0.9). A linear regression between ln(reaction rate) and
was performed (Figure 9
) and the order of reaction of oxygen was calculated. It was found that the order of oxygen concentration is dependent on the conversion level. The order slightly increased from 0.67 to 0.75 with the increase of conversion (from X
= 0.2 to X
= 0.8). This agrees with our previous studies [24
], in which the order of reaction of oxygen was evaluated in a pilot reactor.
To determine the activation energy for each isotherm, the reaction rate was calculated at defined conversion: X
= 0.2–0.9. For each conversion, ln(rs
) was represented as a function of 1/T
and a linear fit was performed to calculate the activation energy. It can be observed from the Arrhenius plot in Figure 10
that the activation energy is independent of the conversion level, resulting in an average activation energy of 167 kJ/mol. This result is consistent with our previous studies and the literature, in which Ea
in the range of 130–170 kJ/mol appears frequently for different soot sources [16
]. In contrast, the kinetics parameters for Printex-U carbon black and engine soot using a step-response technique [17
] and the same fitting procedure are tabulated in Table 3
2.6. Flame Deposit Contact vs. Realistic Contact
In lab-scale studies, a loose contact condition is crucial to evaluate the performance of the soot oxidation catalysts and approximate real conditions. Hence, one would like to know how well Tiki©
lamp flame deposition imitates real soot contact. To address this issue, a sample holder was placed in the exhaust pipe of a diesel engine test stand to produce realistic contact and capture real engine soot for further comparative study by TGA. A blank inert substrate was used to capture pristine engine soot, and a catalyst coated substrate (previously discussed) [25
] was used to produce a realistic contact condition between the soot and catalysts. Figure 11
shows the curves for catalyzed and non-catalyzed real diesel soot oxidation under real contact conditions. Equivalent substrates were also used to capture flame soot 10 cm (FS100) away from the burner.
It was found that the catalyzed soot oxidation Tig with realistic contact is 310 °C, while for flame deposit contact it is ~410 °C. The much lower catalytic soot oxidation Tig of engine soot can be attributed to the high content of HCs, which usually have lower intrinsic ignition temperatures than carbon. In contrast, the catalytic oxidation T50, FC (where 50 percent of soot is burned off) of flame deposited soot was roughly 60 °C higher than engine produced soot (410 °C). The T50, RC dropped by 220 °C for catalyzed oxidation with the realistic contact condition, while the T50, FC dropped by 200 °C with catalyzed oxidation with the flame deposit contact condition. The change in T50 due to the presence of the catalyst is very similar in both cases, implying that the flame deposit method is able to produce comparably realistic, albeit slightly looser, contact conditions to that resulting from the real exhaust system. The higher hydrocarbon content of the real soot gives lower Tig and T50, while the flame deposition technique provides meaningful, conservative contact testing conditions relative to a realistic exhaust system.
It is also worthy to note that the catalyzed soot oxidation T50, SC
using “gently shaking” method (noted as Shaking Contact, catalyst in the powder form) [5
] dropped by ~75 °C from the T50
of non-catalytic model soot oxidation. In comparison with 200 °C drop and 220 °C drop in T50
when using the flame deposit method (Flame Contact) and the engine deposit method (Realistic Contact) respectively, “Shaking Contact” seems to be too loose compared to the real-world established loose contact condition. It implies a risk of missing lead candidate compounds when using “Shaking Contact” for catalyst screening studies due to the measured T50, SC
being too conservative.