2.1. Soot Characterization
Figure 1a shows TGA profiles obtained when heating the different types of diesel soot under air flow, i.e., the Printex U soot model, the one obtained from burnt diesel fuel (LabSoot) and that collected from the exhausts of a bench diesel engine (BenchSoot). LabSoot loses about 93 wt.% mass between 420 and 550 °C, as seen in the corresponding TGA profile, accompanied by an exothermic heat evolution shown in the SDTA profile (
Figure 1b), as indicated by the peak with a maximum at 525 °C. This last temperature is associated with the non-catalytic combustion of soot (carbonaceous fraction), remaining 4.5 wt.% ash after the heating up to 600 °C (
Figure 1a). By XRF analysis, the ash chemical composition was found to be around 50 wt.% Fe, the other 50 wt.% being composed of Ca, S, K, Zr, Zn and Cu.
For BenchSoot, two zones of weight loss can be distinguished in the TGA profile (
Figure 1a), characterized by small differences in their slopes: the first one between 185 and 450 °C and the second one between 500 and 600 °C. The zone at the lower temperature involves two exothermic evolutions that have maxima at 330 °C and 380 °C, whereas the other one at a high temperature shows a peak at 540 °C (exothermic evolution), shown in the respective SDTA curve (
Figure 1b). The low-temperature peaks probably originated from the combustion of unburned hydrocarbons (HC adsorbed and/or condensed on soot), abundant in engine exhausts, which were also observed in a previous work [
16]. The high-temperature peak is associated with the non-catalytic combustion of soot (carbonaceous fraction). However, after heating up to 600 °C, 94.7 wt.% weight remains. This percentage involves both the ash content and the ceramic paper weight, as it was not possible to separate soot particles from the filtering media (ceramic paper) for BenchSoot. By XRF analysis, the ash chemical composition was found to be at around 50 wt.% Fe, and the other 50 wt.% corresponded to Ca, S, K, Zr, Zn, Cu and Mn (values calculated without considering the amounts of SiO
2 and Al
2O
3 that compose the ceramic paper disc).
The composition of the ash varies for different types of soot owing to the presence of several classes of metallic compounds (containing P, S, Ca, Zn, Mg, and possible Na [
33]) coming from additives added to lubricants, trace elements present in the fuel, and products generated by engine wear and corrosion. Nevertheless, as seen by XRF, no significant differences can be observed between the ash composition of LabSoot and BenchSoot, unless the presence of Mn in BenchSoot probably comes from lubricant oils.
On the other hand, the model soot sample (Printex U) is completely gasified between 450 and 670 °C, as observed in the corresponding TGA profile (
Figure 1a), so that no ash was detected after combustion. The mass loss observed correlates to an exothermic evolution at 630 °C, indicated in the SDTA profile of Printex U (
Figure 1b). In addition, the absence of peaks at lower temperatures would indicate the absence of adsorbed or condensed hydrocarbons.
To deepen this study, the soot samples were also characterized by infrared spectroscopy (
Figure 2). The assignment of signals according to the type of groups adsorbed on LabSoot, BenchSoot, and Printex U samples is shown in the respective spectra. The spectrum corresponding to the BenchSoot sample shows signals between 2960 and 2850 cm
−1 ascribed to the asymmetric and symmetric stretching vibrations of the C-H bond of both methyl and methylene species (CH
3, CH
2). Correspondingly, signals at 1384 and 1458 cm
−1 associated to the C-H bond deformation vibrations of these groups are observed. These signals probably come from aliphatic hydrocarbons [
34,
35]. A broad signal appears at 1620 cm
−1, indicating the presence of an aromatic C=C bond [
36]. The signals at 1040 and 1110 cm
−1 are due to the stretching of C-O bonds of C-OH, C-O-C, and C-O groups, corresponding to functional groups of alcohols, ethers, and esters, respectively [
36,
37]. These signals are in line with the peaks observed at low temperatures for BenchSoot (330 and 380 °C) in the SDTA experiment (
Figure 1b) and they are probably associated with unburnt hydrocarbons adsorbed and condensed on this sample.
For LabSoot, similarly to that observed for BenchSoot, the signal of a C=C aromatic at 1620 cm
−1 is clearly observed, whereas a signal observed at ca. 1740 cm
−1 indicates the presence of carbonyl groups [
34]. The lower intensities observed for aliphatic HC in the spectrum of LabSoot are associated to the higher temperature treatment at which the diesel fuel was exposed to produce the artificial soot. In fact, the FTIR technique allowed the identification of surface species, not detected during thermogravimetric analysis (
Figure 1b).
The FTIR spectrum for the Printex U sample indicates the absence of condensed hydrocarbons, also evidenced by the absence of mass loss observed at low temperatures in TGA experiments (
Figure 1a). Only low-intensity signals coming from atmospheric CO
2 (ca. 2200 cm
−1) and rotovibrational signals of water (ca. 1600 cm
−1) are seen in the spectrum.
The presence of surface oxygenated groups provides a fast soot-oxidation pathway. Song et al. [
38], based on a previous contribution of Ishiguro et al. [
39], used the ratio between intensity peaks at 1740 cm
−1 (attributed to the stretching of the C=O groups from carboxylic acids, lactones and ketones) to that at 1620 cm
−1 (corresponding to the stretching of the C=C bond of C=O conjugated aromatic systems), I
1740/I
1620, to estimate the relative abundance of superficial oxygen. This ratio could be estimated for LabSoot (I
1740/I
1620 = 0.91) and for BenchSoot (I
1740/I
1620 = 0.44), whereas in the case of Printex U, these peaks at 1740 and1620 cm
−1 are not detected in the IR spectrum. This suggests the following order according to the content of surface oxygenated species: LabSoot > BenchSoot > Printex U.
According to TGA, SDTA and FTIR results for the BenchSoot sample, the SDTA peaks at 330 and 380 °C correspond to the combustion of different types of organic compounds, such as aliphatic, aromatic and oxygenated hydrocarbons, condensed on the carbonaceous core of soot particulates, whereas the high temperature peak (540 °C) corresponds to the combustion of the carbonaceous nuclei. In the case of LabSoot, the burning of the carbonaceous core appears at lower temperatures (525 °C), which could be associated with the higher content of oxygenated species (
Figure 2). As these oxygenated species are superficial, no detectable exothermic peaks associated to their decomposition are observed in TGA experiments.
Raman spectroscopy is a powerful method to characterize carbon materials, since it is very sensitive to the short range disorder of carbon. The carbon materials are integrated with the linear chain of C-C atoms forming the monomers and polymers and can undergo different structural changes due to the rearrangement of the atoms through different hybridization. The existence of π-states in carbon materials with the sp
2 bonds have very long-range polarizability, leading to a large Raman cross section, making Raman spectroscopy a suitable choice [
40].
Graphite is the stacked sheet of carbon atoms in the basal planes arranged in the sp
2 tetragonal configuration with weak interaction between the layers. It has two-dimensional hexagonal shapes with the fourth electron in the π-orbital and perpendicular to the weakly bonded hexagon sheets [
41]. The graphitic structure has a high degree of disorder due to strong C-C bonding.
Accordingly, the different classes of soot were further analyzed utilizing LRS and
Figure 3 displays the resulting spectra, composed of the G-peak, which relates to the graphitic-like structures and is originated from the doubly degenerated vibrational mode E
2g at about 1580 cm
−1 and the D-peak at around 1360 cm
−1 related to the disorder induced in the graphite and linked to A
1g symmetry. Both the ratio of amorphous to graphitic carbon and the degree of the ordering of the graphene layer were calculated by the deconvolution of the spectra to relate them to the reactivity of soot in the TPO runs. To this end, every Raman spectrum was fitted considering the contributions listed in
Table 1, as proposed by Sadezky et al. [
42].
The graphitic carbon content, expressed as a percentage, was obtained according to Equation (1) [
30]:
in which
is the percentage of graphitic carbon and I
G and I
T are the corresponding areas of G peak and that of the sum of all the areas of the different peaks obtained from fitting the spectra, respectively. The resulting percent graphitic carbon content was 30.6%, 21.4%, and 19.9% for LabSoot, Printex U and BenchSoot, respectively (
Table 2).
On the other hand, the percentage of amorphous carbon (% AC) was calculated according to Equation (2):
where I
D3 is the intensity (peak area) of the D3 peak that correlates with the amount of amorphous carbon.
Table 2 lists amorphous carbon contents, which were similar for Printex U and BenchSoot samples, with values of 17.7% and 16.6%, respectively, whereas a lower value (6.0%) was observed for LabSoot. This type of carbon is easier to burn than graphitic carbon, which will influence TPO assays [
43].
In addition, the I
D1/I
G ratio, where I
D1 is the intensity of the D1 peak related to the disordered graphitic lattice, is indicative of the degree of graphene layers disordering, which increases with the value of I
D1/I
G. In this sense, values shown in
Table 2 indicate higher disordering for BenchSoot and, consequently, the following order can be established according to the degree of graphitization: LabSoot > Printex U > BenchSoot.
On the other hand, the internal structure and morphological aspects of soot samples, such as size distribution, particle shape and the orientation of the graphene layer, were studied by TEM. The images obtained (
Figure 4) show that particle size and agglomeration are different, depending on the method of obtaining soot samples. As
Figure 4 shows, BenchSoot primary particles have an irregular shape with distribution sizes between 8 to 57 nm, and a compact agglomeration. In contrast, Printex U primary particles, with distribution sizes between 24 to 70 nm, have a spherical geometry and a chain-like agglomeration, as LabSoot particles do. However, the latter are the largest particles, between 48 to 154 nm, and oval in shape. From
Figure 4 (bottom), the size order of mean primary particle size is BenchSoot < Printex U < LabSoot, with values of 23.8 nm, 47.5 nm and 102.5 nm, respectively.
In a detailed view (
Figure 5), the orientation of the graphene layers for the different primary soot particles can be observed, consisting of an outer shell, related to graphitic carbon, and an inner core, related to amorphous carbon. The LabSoot sample shows a more defined orientation of the graphene layers with a small inner core, while for the Printex U and Bench Soot samples, the graphene layers acquire a certain degree of disorder and the size of the inner cores increases, which agrees with the results shown in
Table 2 obtained from LR spectra. That is, the ratio of amorphous to graphitic carbon increases in the order LabSoot < Printex U < BenchSoot and, on the contrary, the degree of graphitization follows the order LabSoot > Printex U > BenchSoot.
It is noteworthy that both the soot particle agglomeration and the higher graphene layer disorder observed for the BenchSoot sample are similar to those found in vehicle soot particles [
43], and are consistent with the higher amount of adsorbed hydrocarbons or oxygenated species found by TGA/SDTA and FTIR, with the more disordered soot structure being easier to burn.