2.1.1. Textural Properties
Figure 1a,b show the N
2 adsorption–desorption isotherms and the corresponding pore size distributions for the evaporated samples, together with the parent calcined red mud. Data from N
2 physisorption analyses are given in
Table 2.
RM-500 and RME2-500 samples exhibit similar types of isotherm shape and hysteresis loop. The shape is close to type II isotherm with a slow increase in the first half of the N
2 adsorption curve, followed by a sharp increase in the second half of the isotherm. This type is shown by materials which are either non-porous or macro-porous. Additionally, the hysteresis loop matches the H3 hysteresis type, indicating the presence of small amounts of mesopores originating probably from aggregation of particles. The evaporated sample RME2-500 shows a minor decrease in specific surface area (SSA) and increase in total pore volume in comparison to RM-500 (
Table 2). The pore size distribution curves exhibit a narrow peak centered at around 4 nm (
Figure 1b), indicating the good homogeneous distribution of the mesopores in both samples. The leaching with oxalic acid and l-ascorbic acid, followed by evaporation and calcination steps, do not lead to a profound modification in textural properties. Interestingly the evaporated sample RME1-500 shows a marginal increase in SSA and total pore volume compared to that of RM-500. The isotherm of the RME1-500 material is type IV and when the relative pressure exceeds a certain value, the adsorption isotherm tends to level off. The pore size distribution of the RME1-500 material is broad in comparison with that of the RM-500 and RME2-500 samples. Then the leaching with hydrochloric acid following by evaporation and calcination steps seems to open the porosity of the material. This can be explained by a rearrangement of the aggregates forming the particles.
2.1.2. Bulk and Surface Compositions
The X-ray diffraction patterns of the red mud and the evaporated samples are shown in
Figure 2. The XRD pattern of the RM-500 sample (
Figure 2a) matches the main combination of two standard patterns of JCPDS 89-8104 (Hematite α-Fe
2O
3) and JCPDS 79-1906 (Quartz SiO
2). It also has peaks of JCPDS 86-2340 (Calcite, CaCO
3). The XRD pattern of the RME1-500 sample (
Figure 2b) matches the similar combination of the two previous standard patterns, hematite α-Fe
2O
3 and quartz SiO
2. However, in comparison with α-Fe
2O
3, the proportion of SiO
2 phase is significant lower, which is expected from the leaching. In addition, the sample also appears to have sodium chloride (JCPDS 5-628), which is expected from the leaching of red mud in HCl. The identified major mineral phase in the RME2-500 sample (
Figure 2c) is maghemite (γ-Fe
2O
3) (JCPDS 83-0112). Magnetite also has similar peaks but is not likely to be present under the conditions of leaching of the red mud with oxalic acid and l-ascorbic acid, followed by evaporation and calcination at 500 °C. The presence of l-ascorbic acid reduces Fe (III) to Fe (II) and improves leaching in oxalic acid, leading to the formation of ferrous oxalate [
23]. The ferrous oxalate depending on the temperature, atmosphere, and time of calcination, can form magnetite, maghemite, or haematite [
24,
25]; at 500 °C, magnetite forms only at low oxygen partial pressures [
26]. Under the conditions of calcination carried out, haematite is expected. However, due to the presence of reducing conditions from CO arising from oxalate decomposition, maghemite could have formed. No peaks matching with aluminium compounds were observed, which could be due to their small size or amorphous nature.
The values of the binding energies (BE) of the electrons coming from different orbitals from XPS analysis are shown in
Table 3.
In all the samples, the Al2p and Ca2p BE values are consistent with the presence of Al
3+ in Al
2O
3 and Ca
2+ in CaCO
3, respectively.
Figure 3 shows the high resolution of Fe2p spectra with two prominent photopeaks associated with Fe2p
3/2 (711.2 eV) and Fe2p
1/2 (720 eV). Broad peaks to higher BE of the main photopeaks (718.8 eV and 733 eV) are the shake-up satellite peaks of Fe2p
3/2 and Fe2p
1/2, respectively, and indicates the presence of Fe
2O
3 (maghemite or hematite). While Al2p, Ca2p, and Fe2p BE do not significantly vary from one sample to another, Na1s BE is lowered by 0.8–1 eV for the RME2 sample. The lower Na1s BE is ascribed to the presence of sodium as sodium carbonate from the decomposition of sodium oxalate [
27].
Figure 4 represents the O1s spectra of RM, RME1, and RME2 samples. The O1s signal is well described with three components. The O
I at 530.0–530.2 eV is due to the presence of oxygen in transition metal oxides (titanium and iron oxides) and contributed to 45%, 48%, and 47% of total oxygen for RM-500, RME1-500, and RME2-500, respectively. The O
II at 531.6–531.8 eV corresponds to oxygen in alumina and metal carbonates contributed to 33%, 36% and 30% of total oxygen for RM-500, RME1-500 and RME2-500, respectively. The O
III at 533 eV can be attributed to hydroxyl groups and its contribution was the lowest at 18%, 13%, and 18% of total oxygen for RM-500, RME1-500, and RME2-500, respectively. The low intensity photopeak observed at 536.7 eV for RM, RME1, and 536 eV for RM2 is ascribed to the Na KLL Auger peak. Owing to the Na1s BE shift previously observed, the Na KLL Auger peak is also shifted for RME2.
Quantitative information is shown in
Table 4, along with Fe/Al atomic ratios. The surface iron content increases marginally for RME2-500 compared to RM-500, while in the case of RME1-500, there is a reduction. This is similar to the trend in bulk composition (
Table 1). Correspondingly, the Fe/Al atomic ratio is increased for RME2-500 and decreased for RME1-500, compared to RM-500 for both surface composition and bulk composition. The RME1-500 sample shows high Cl content, which corresponds with the presence of NaCl detected in XRD analysis (
Figure 2b).
2.1.3. Redox Properties
H
2-TPR experiments have been performed in order to characterize the redox properties of the different samples (
Figure 5). Since the oxidation of VOC over transition metal oxides takes place through a redox mechanism, a correlation between catalytic activity and H
2-TPR is expected. The unmodified calcined red mud sample (RM-500) contains hematite (α-Fe
2O
3), which mainly contributes to the reduction profile. However, the reduction of α-Fe
2O
3 shows different profiles based on heating rate, moisture content, sample size etc. The red mud sample (RM-500) exhibits three overlapping reduction peaks. The first reduction peak at around 480 °C is usually ascribed to the reduction Fe
2O
3 to Fe
3O
4 while the second reduction peak at a higher temperature range (650–900 °C) is ascribed to the reduction of Fe
3O
4 to Fe. The number of steps involved in the reduction process of pure iron oxide has not been fully clarified. Indeed, this number is highly affected by the quantity of water in the feed [
28]. Furthermore, the metastable nature of FeO and its possible disproportion to Fe
3O
4 and metallic iron according to the reaction 4FeO = Fe
3O
4 + Fe (T < 619 °C) make the reduction of FeO to metallic iron not readily observable [
29].
The evaporated samples also showed reduction peaks in the same temperature range without significant profile modification; however, the onset temperature is higher compared to RM-500.