3.1. Catalyst Characterization
Results of N
2 adsorption experiments are summarized in
Table 1, where a progressive decrease in the specific surface area can be observed from 64.4 to 46.2 m
2/g as the gallium oxide loading increased from 0 to 40 wt.%, most possibly due to a partial blocking of alumina pores induced by Ga
2O
3 addition [
2,
4,
5,
27]. The BET surface area measured for the bare Ga
2O
3 powder was significantly lower (4 m
2/g). The pore volume and the mean pore diameter were measured over bare Al
2O
3, 10%Ga
2O
3-Al
2O
3 and 30%Ga
2O
3-Al
2O
3 and found to progressively decrease with increasing Ga
2O
3 content, taking values of 0.187, 0.166 and 0.145 cm
3 g
−1, and 8.3, 7.9 and 7.5 nm, respectively.
The X-ray diffractograms obtained from the investigated catalysts are illustrated in
Figure 1A. In the case of Al
2O
3 and Ga
2O
3-Al
2O
3 catalysts, the typical diffraction peaks of the hexagonal and cubic Al
2O
3 structure were detected. Specifically, XRD peaks located at 2θ values equal to 33.07°, 36.69°, 38.51°, 39.52°, 45.87°, 46.62° and 67.4° correspond to (006), (212), (205), (300), (304), (221) and (414) planes of hexagonal Al
2O
3 (JCPDS Card No. 21-10), respectively, whereas diffraction peaks located at 37.8°, 39.67°, 45.90°, 60.53° and 67.34° diffraction angles correspond to (311), (222), (400), (511) and (440) Miller indices of cubic Al
2O
3 (JCPDS Card No. 4-880), respectively. An extra peak centered at 21.63° was detected only for the 30%Ga
2O
3-Al
2O
3 and 40%Ga
2O
3-Al
2O
3 samples and can be attributed to the (004) phase of hexagonal alumina (JCPDS Card No. 21-10). It is of interest to note that the position of the diffraction angle assigned to the (440) plane of cubic alumina was slightly shifted towards lower angles as the Ga
2O
3 content was becoming higher (
Figure 1B). A similar shift was previously attributed to the formation of a solid solution induced by the incorporation of Ga
3+ ions of a higher ionic radius than that of Al
3+ ions into the Al
2O
3 structure [
2,
3,
14,
28,
29].
No diffraction peaks of gallium oxide were discerned for the samples containing 10 and 20 wt.% Ga
2O
3, suggesting that either Ga
2O
3 particles were highly dispersed on the Al
2O
3 surface or Ga
2O
3 was amorphous or a single-phase oxide in Al
2O
3. However, an additional peak located at 2θ = 31.6° assigned to the (002) plane of
β-Ga
2O
3 can be discerned in the XRD pattern of the 30% and 40%Ga
2O
3-Al
2O
3 catalysts [
30]. More peaks corresponding to gallium oxide may also coexist in the diffractograms but cannot be discerned due to overlapping with alumina peaks. Regarding the XRD pattern obtained from the bare Ga
2O
3, it was found to consist of peaks attributed to
β-Ga
2O
3 (JCPDS Card No. 41-1103).
The morphology and the elements distribution of a selected catalyst, specifically, the 10%Ga
2O
3-Al
2O
3, was investigated with SEM and EDS analysis. A representative image, along with the element mapping of Ga and the EDS profile obtained, is presented in
Figure S1. The EDS analysis confirmed the presence of Ga, O, and Al elements, while the elemental mapping demonstrated that Ga was homogeneously distributed on the surface of Al
2O
3. The weight percentage of Ga, Al, and O estimated by the EDS analysis was found to be equal to ca. 5.2 wt.%, 42.1 wt.%, and 52.7 wt.%, respectively. Representative TEM images and the selected area electron diffraction (SAED) patterns were obtained from bare Al
2O
3 and a selected Ga
2O
3-containing catalyst, the 10%Ga
2O
3-Al
2O
3 (
Figure S2). It was found that both Al
2O
3 and 10%Ga
2O
3-Al
2O
3 catalysts consist of spherical Al
2O
3 nanoparticles with a diameter of about 6–8 nm. In the SAED spectra, the observed diffraction rings noted by spots 1, 2, 3, 4, 5, and 6 correspond to d-spacing values equal to 3, 2.4, 1.98, 1.63, 1.43, and 1.39 Å, respectively, of an unknown Al
2O
3 structure (JCPDS Card No. 2-1422). Taking into account that a cubic and hexagonal Al
2O
3 structure was identified in XRD measurements for both bare Al
2O
3 and 10%Ga
2O
3-Al
2O
3 samples, a polycrystalline structure of the Al
2O
3 used as support can be suggested [
8]. No reflections attributed to Ga
2O
3 structure were detected over the 10%Ga
2O
3-Al
2O
3 sample, indicating that Ga
2O
3 particles were either well dispersed or amorphous. Results indicated that the morphology of alumina does not change with the addition of 10 wt.% Ga
2O
3.
Results of CO
2-TPD experiments obtained over the investigated catalysts employing the MS technique are presented in
Figure 2, where the concentration of CO
2 (in ppm) was plotted as a function of temperature for all catalysts examined. A low temperature (LT) desorption peak was observed over bare Al
2O
3 and x%Ga
2O
3-Al
2O
3 catalysts, which is related to CO
2 desorption from weak basic sites [
4,
5,
8,
31,
32]. The position of this peak was progressively shifted from 102 °C for bare Al
2O
3 to 111 °C for the samples containing 20, 30, and 40 wt.% Ga
2O
3 indicating that the strength of CO
2 adsorption was enhanced in the presence of Ga
2O
3 in agreement with previous study [
31]. A high temperature (HT) broad peak can hardly be discerned between 500 and 700 °C for all catalysts examined, which is associated with the desorption of CO
2 from moderate/strong basic sites [
4,
5,
8,
32]. The intensity of both LT and HT peaks was too low for the bare Ga
2O
3 sample, most probably due to the low basicity in combination with the low specific surface area of this sample. The area below the LT and HT peaks was integrated to estimate the amount of CO
2 (in μmol g
−1) desorbed from the weak and moderate/strong basic sites, respectively (
Table S1), which was found to be optimized for the 20%Ga
2O
3-Al
2O
3 catalyst. Although the CO
2 adsorption was expected to decrease as the specific surface area decreases, the observed trend of CO
2 adsorption presented in
Table S1 should not only be related to the variation in the specific surface area but also to the interactions between the Ga
2O
3 and the Al
2O
3 support induced by the increase in Ga
2O
3 content. Since the specific surface area of the investigated catalysts was significantly varied from 4 to 64.4 m
2 g
−1, the results of
Table S1 were normalized by the specific surface area in order not to contain contributions from the variation in this parameter. It was found that the amount of CO
2 desorbed (in μmol m
−2) from both the weak and moderate/strong basic sites was maximized for the sample containing 20 wt.% Ga
2O
3 (
Table 2). This can be clearly seen in
Figure 3, where the total amount of CO
2 desorbed during TPD was plotted as a function of the Ga
2O
3 content. Specifically, the amount of CO
2 was found to increase from 0.54 m
2 g
−1 for the bare Al
2O
3 to 1.54 m
2 g
−1 for the 20%Ga
2O
3-Al
2O
3 catalyst and subsequently decreased to 1.08 m
2 g
−1 with the progressive increase in Ga
2O
3 content to 40 wt.%, while it was further decreased to 0.75 m
2 g
−1 for the bare Ga
2O
3.
The results of
Figure 3 provide evidence that the surface basicity of Ga
2O
3-Al
2O
3 catalysts depends strongly on the Ga
2O
3 concentration, which is in accordance with previous studies. For example, Li et al. [
33], who investigated the surface basicity of x%Ga
2O
3-ZrO
2 (x: 0, 5, 10, 15, 20 wt.%) catalysts by CO
2-TPD, found that a maximum number of basic sites appeared for Ga
2O
3 content of 15 wt.%. Moreover, Michorczyk et al. [
34] reported that the density of basic sites on the surface of Ga
2O
3-Al
2O
3 catalysts increased with increasing Ga
2O
3 loading from 0 to 20 wt.%, in excellent agreement with the results of the present study. The adsorption of CO
2 was also found to be facilitated by increasing the concentration of Ga
2O
3 over Ni/Ga
2O
3-Al
2O
3 catalysts [
31]. Similarly, Orlyk et al. [
32] demonstrated that the total surface basicity of GaxSiBEA (x: 1, 2, 4 wt.%) zeolites increased almost proportionally to the content of Ga. Furthermore, the addition of Ga
2O
3 on Ce
0.
6Zr
0.
4O
2 with loadings varying between 0 and 15 wt.% was found to enhance the surface basicity, which was optimized over the sample containing 5 wt.% Ga
2O
3 [
35].
The CO
2 adsorption/desorption characteristics were also investigated by in situ FTIR spectroscopy, and the results obtained are presented in
Figure S3. The DRIFT spectrum collected at 25 °C in He flow for bare Al
2O
3 (
Figure S3a) following its interaction with 5%CO
2 (in He) was consisted of various bands in the 1700–1200 cm
−1 region previously attributed to bicarbonate species (1658, 1433 and 1229 cm
−1), as well as to unidentate and bidentate carbonates (1627, 1558 and 1373 cm
−1) [
36,
37,
38,
39,
40,
41,
42]. According to previous studies, the formation of carbonate-like species on the catalyst surface occurs via CO
2 interaction with the basic sites of the metal oxide, i.e., the surface hydroxyl groups and/or the low-coordination oxygen anions [
43,
44]. In particular, it was suggested that CO
2 interaction with the surface OH groups is responsible for bicarbonate formation, while unidentate, bidentate, and bridged carbonate species are mainly generated by CO
2 interaction with oxygen anions [
36,
39,
40,
43,
45,
46]. Increase in temperature under He flow led to a decrease in the intensity of all bands, which almost disappeared above 250 °C, implying that the corresponding species were desorbed from the alumina surface (
Figure S3a).
Similar bands were detected in the spectra obtained from the Ga
2O
3-modified Al
2O
3 catalysts following CO
2 adsorption, implying that the same surface species were formed independently of the Ga
2O
3 content (
Figure S3b–e). It should be noted that CO
2 adsorption on Ga
2O
3 surface was previously found to result in the formation of bicarbonates and bidentate carbonates, giving rise to the development of bands located at wavenumbers close to those discussed above for Al
2O
3 [
41,
42,
47]. Therefore, part of the detected bands in the FTIR spectra may be related to carbonate-like species associated with Ga
2O
3 particles. The shoulder appeared at 1690 cm
−1 for the sample containing 10 wt.% Ga
2O
3 was previously attributed to bidentate carbonates adsorbed on Al
2O
3 surface or bridged carbonates adsorbed on Ga
2O
3 surface [
42,
47]. This band may also be present in the spectra obtained from the rest of the catalysts examined but cannot be distinguished due to the coexistence of more than one band in the corresponding wavenumber region. It is of interest to note that the relative population of surface species seems to be maximized for the 20%Ga
2O
3-Al
2O
3 catalyst and eliminated above 250 °C for all composite metal oxides (
Figure S3), in excellent agreement with the results of CO
2-TPD experiments discussed above (
Figure 2,
Table 2).
Concerning the spectra obtained from the bare Ga
2O
3, only two weak peaks were detected at 1621 and 1333 cm
−1 due to bicarbonate and bidentate carbonate species, respectively, which desorbed from the catalyst surface below 200 °C [
41,
42,
47]. The low CO
2 adsorption capacity of this sample may be correlated with its low specific surface area and agrees well with the results of
Figure 2.
The surface acidity of the investigated metal oxides was examined by the potentiometric titration method described above, and the results obtained are presented in
Figure S4, where the potentiometric titration curves of the dried catalysts and the control sample, fitted by the Boltzmann function, are presented along with the corresponding Gran’s function plots and their linearization. Based on the Gran’s method described above, two equivalence points were determined for each catalyst,
Va and
Vb, which represent the equivalence volumes obtained from the acidic and basic slopes of the Gran’s function, respectively. Two similar equivalence points were also extrapolated,
VaN and
VbN, in the case of the control sample (0.05 M KNO
3).
The concentration of the different types of acid sites was estimated by the intersections of the straight lines with the Volume axis using the Equations (6)–(8) and results obtained (
Table S2, in μmol·g
−1) were normalized with respect to the specific surface area of each catalyst and presented in
Table 3 and
Figure S5 (in μmol·m
−2). It was found that bare Al
2O
3 and x%Ga
2O
3-Al
2O
3 catalysts consisted of three different types of acid sites: very weak, weak, and strong acid sites. The density of very weak (A
vw) and strong acid sites (A
s) was, generally, maximized for the sample containing 30 wt.% Ga
2O
3, while that of weak acid sites (A
w) was progressively increased with increasing the Ga
2O
3 content from 10 to 40 wt.% and found to be lower than that of bare Al
2O
3 in the case of the 10, 20 and 30 wt.% Ga
2O
3-Al
2O
3. Only weak and strong acid sites were determined over bare Ga
2O
3, which, given its low specific surface area, exhibited significantly higher A
w and A
s values (in μmol·m
−2) than the rest of the catalysts examined. The total surface acidity of the investigated catalysts, defined as the sum of A
vw, A
w, and A
s values, was found to gradually increase from 6.64 to 85.00 μmol·m
−2 with increasing Ga
2O
3 content from 0 to 100 wt.% (
Table 3,
Figure 3).
In order to obtain additional insight related to the type and strength of acid sites on the surface of the investigated catalysts, pyridine adsorption/desorption experiments using FTIR spectroscopy were applied. Results obtained are presented in
Figure 4. Pyridine adsorption on bare Al
2O
3 (
Figure 4a) resulted in the development of two bands at 1635 and 1453 cm
−1 in the spectrum recorded at 25 °C, which can be attributed to pyridine species interacting with Brønsted and strong Lewis acid sites, respectively [
3,
27,
28,
48,
49,
50,
51,
52,
53,
54,
55]. The former band disappeared at temperatures higher than 150 °C, while the latter one was present in all spectra collected up to 500 °C, indicating that the corresponding species were adsorbed strongly on the alumina surface. It should be noted, however, that the band at 1453 cm
−1 may also contain contributions from physisorbed pyridine at least at low desorption temperatures [
48,
51,
56]. A new band was discerned at ca. 1613 cm
−1 in the spectrum obtained at 150 °C, which was accompanied by the parallel development of a broad shoulder at ca. 1589 cm
−1 at temperatures higher than 400 °C. Both bands were detectable up to 500 °C and can be assigned to pyridine adsorption on strong (1613 cm
−1) and weak/moderate (1589 cm
−1) Lewis acid sites [
28,
48,
50,
52,
53,
55,
56].
No significant variations were observed in the spectra obtained from the 10%Ga
2O
3-Al
2O
3 catalyst (
Figure 4b), besides (a) the appearance of a band at 1596 cm
−1 in the spectrum collected at 25 °C, which was diminished at higher temperatures and was previously attributed to physiosorbed or H-bonded pyridine [
48,
51,
56], (b) the detection of a weak band at 1558 cm
−1 at 100 °C, which was present in all spectra collected up to 450 °C and was due to pyridine protonated by strong Brønsted acid sites [
28,
48,
50] and (c) the absence of the band discussed above at 1589 cm
−1 corresponding to weak/moderate Lewis acid sites. The bands assigned to pyridine species adsorbed on Lewis acid sites exhibited significantly higher intensity than those assigned to pyridine species adsorbed on Brønsted acid sites and were present up to 500 °C, implying that both the number and strength of Lewis acid sites were higher. Further increase in Ga
2O
3 content up to 40 wt.% led to a progressive increase in features owing to pyridine adsorbed on Lewis acid sites (1613–1615, 1585–1590 and 1451–1453 cm
−1) while no characteristic peaks associated with Brønsted acidity (1640 or 1540–1560 cm
−1) was discerned in any of the samples containing 20, 30 and 40 wt.% Ga
2O
3 (
Figure 4c–e). A new band can be discerned at 1492 cm
−1 in the spectra obtained from the 30%Ga
2O
3-Al
2O
3 (
Figure 4d) and 40%Ga
2O
3-Al
2O
3 (
Figure 4e) catalysts, which was previously reported to contain overlapping bands due to pyridine adsorption on both Lewis and Brønsted acid sites [
48,
55]. Results of
Figure 4 indicate that even if the Ga
2O
3-Al
2O
3 catalysts contained Brønsted acid sites, both their number and strength were notably lower compared to those of Lewis acid sites and were eliminated with increasing gallium oxide loading. It is generally accepted that Lewis acid sites in Ga
2O
3-based catalysts are related to coordinatively unsaturated Ga
3+ ions in the tetrahedral position, while Brønsted acid sites are related to Ga-OH groups on the catalyst surface [
28,
55,
56,
57]. It has also been proposed that the Lewis acid sites over Ga
2O
3-SiO
2 catalysts originate from Ga
2O
3 particles that have not been incorporated into the SiO
2 framework, while Ga
2O
3 particles incorporated into the SiO
2 framework are responsible for the creation of Brønsted acid sites [
57]. Therefore, taking into account that the population of Lewis acid sites was higher than Brønsted acid sites for the Ga
2O
3-Al
2O
3 catalysts investigated in the present study, it can be assumed that gallium oxide particles mainly remained on the surface of alumina rather than incorporated into its framework. Although part of Ga
2O
3 may be incorporated into the Al
2O
3 structure, as evidenced by the small shift of the diffraction peak located at 2θ = 67.4° observed in X-ray diffractograms (
Figure 1B), the fraction of Ga
2O
3 particles that remained on the catalyst surface seems to be higher.
It should also be noted that, in the case of catalysts containing 20, 30, and 40 wt.% Ga2O3, two overlapping features can be discerned in the 1445–1453 cm−1 region. The one located at ca. 1452 cm−1 was associated with strong Lewis acid sites, while that detected at ca. 1445 cm−1 was associated with physisorbed pyridine, which in all cases disappeared upon heating the catalyst at 100 °C. This was also the case for the 1594–1598 cm−1 band, which, for all gallium oxide containing catalysts, was only present at 25 °C, further supporting the above suggestion that it was related to physisorbed pyridine. Concerning the bare Ga2O3 sample, although a similar pyridine adsorption/desorption experiment was conducted, no clear peaks could be discerned, most likely due to the significantly lower specific surface area (4 m2 g−1) of this sample.
Results of
Figure 4 clearly indicate that the population of pyridine adsorbed surface species, and therefore, the number of acid sites on the catalyst surface, increased significantly with increasing Ga
2O
3 content from 0 to 40 wt.% and are in excellent agreement with the results of
Table 3 obtained from the potentiometric titration experiments. The increase in the acid site density with the addition of Ga
2O
3 on Al
2O
3, TiO
2, and SiO
2 supports was also reported in our previous studies [
4,
5,
8]. Moreover, Zhou et al. [
50] demonstrated that the total surface acidity increased with the addition of Ga
2O
3 on Al
2O
3, with the distribution of acid sites, however, remaining unchanged, whereas Chen et al. [
28] found a greater Lewis acid site density over spinel-type gallia–alumina solid solution Ga
xAl
10−xO
15 (
x: 0–10) oxides compared to bare alumina. In addition, Ga
2O
3-Al
2O
3, prepared by the atomic layer deposition method, was found to be characterized by higher gallium oxide dispersion and stronger interaction with the alumina support, which was able to form more Ga-O-Al linkages and lead to higher surface acidity [
2]. An increase in the number of weak acid sites was also observed by Ga
2O
3 deposition on the SiO
2 surface [
19]. Moreover, Castro-Fernandez et al. [
58] studied the coordination geometry and Lewis acidity of surface sites over gallia–alumina oxides and proposed that the optimization of the Ga/Al atomic ratio is able to adjust the relative abundance and strength of Ga-related Lewis surface acid sites. They found that Ga-rich samples exhibited a higher fraction of six-coordinated Ga sites, as well as a higher Ga related strong Lewis acidity, in agreement with the results of the present study.
In order to investigate the reducibility of supported Ga
2O
3 catalysts, H
2-TPR experiments were conducted over the 10%Ga
2O
3-Al
2O
3 and 30% Ga
2O
3-Al
2O
3 samples (
Figure S6). No reduction peaks were observed in the H
2-TPR profile of the 10%Ga
2O
3-Al
2O
3 catalyst, indicating that this catalyst was not able to be reduced by hydrogen. Increase in Ga
2O
3 content to 30 wt.% led to the appearance of a single weak peak in the H
2-TPR profile centered at ~180 °C, which can be attributed to the reduction of well-dispersed Ga particles and/or GaO
+ species interacting with the support [
17,
59]. The total amount of hydrogen consumed during the H
2-TPR experiment was estimated by integrating the area below the H
2 response curve and found to be 27.9 μmol g
−1. Contradicting results have been reported in the literature related to the reducibility of supported Ga
2O
3 catalysts. For example, treatment of Ga
2O
3-SiO
2 catalyst with H
2 led to the appearance of peaks at low binding energies in the XPS spectra previously ascribed to the formation of gallium hydrides, Ga
2+ or Ga
+ species, implying that the reduction of Ga
3+ is feasible [
16]. This was also the case for Ga
2O
3-Al
2O
3 and Ga
2O
3-ZSM-5 catalysts explored by H
2-TPR experiments [
17]. The reduction ability of gallium from Ga
3+ to Ga
+ in H
2 atmosphere were also studied employing X-ray absorption near edge spectroscopy (XANES) and corroborated by an observed shift of the XANES edge energy upon exposure of Ga-H-ZSM5, Ga–H-BEA and Ga–H-ZSM5 to H
2 at 500–550 °C [
60,
61,
62,
63]. Contrarily, Getsoian et al. [
13] demonstrated that Ga
3+ is not reduced to Ga
+ when Ga–SiO
2 and Ga–H-BEA catalysts are exposed to hydrogen at high temperature. Results presented in
Figure S6 clearly indicate that the reducibility of Ga
2O
3-Al
2O
3 catalysts is generally limited, but it can be slightly enhanced with increasing Ga
2O
3 content.
3.2. Catalytic Performance Tests for the CO2-ODP Reaction
Results of catalytic performance experiments carried out over the x%Ga
2O
3-Al
2O
3 catalysts for the CO
2-ODP reaction are presented in
Figure 5. It was observed that both propane conversion (
Figure 5a) and propylene yield (
Figure 5b) of the composite metal oxides were, for all Ga
2O
3 loadings, higher than that of bare Al
2O
3 and Ga
2O
3. At temperatures lower than 600 °C,
and
increased with increasing Ga
2O
3 content from 0 to 30 wt.%, while they were decreased with the addition of 40 wt.%Ga
2O
3, as well as for the bare Ga
2O
3, which exhibited identical performance with that of bare Al
2O
3. The most active 30%Ga
2O
3-Al
2O
3 catalyst was activated above 475 °C and reached maximum
= 58% and
= 39% at 605 °C. It is of interest to note that the samples containing 20, 30, and 40 wt.% Ga
2O
3 presented a decrease in both
and
in the temperature range of ~600–670 °C, which was then increased again with further increase in temperature to 750 °C. This behavior—which was not observed for the 10 wt.% Ga
2O
3-Al
2O
3, Al
2O
3 and Ga
2O
3 samples—was more intense as Ga
2O
3 loading was becoming higher and, as it will be discussed below, can be attributed to side reactions occurring in parallel, hindering propylene generation.
Figure 6 shows the selectivities towards reaction products as a function of temperature for the investigated catalysts. In the case of bare alumina (
Figure 6a), propylene selectivity (
) increased from 27 to 43% with increasing temperature from 600 to 635 °C, respectively, remained almost constant with further increase in temperature to 700 °C, whereas it was subsequently decreased to 19% with gradual increase in temperature to 750 °C. Selectivity towards CO (
) followed the opposite trend, taking, generally, lower values varying between 17 and 34%. In addition to C
3H
6 and CO, C
2H
4 and CH
4 were also detected with their selectivities (
and
) remaining almost stable in the entire temperature range examined at 27–30% and 14–18%, respectively. The addition of Ga
2O
3 (
Figure 6b–e) resulted in a significant increase in
, which reached 90% at ~450 °C for the sample containing 40 wt.% (
Figure 6e). An increase in the reaction temperature led to a progressive decrease in
, which became more intense in the temperature range of 600–670 °C, where, as mentioned above,
and
presented a sharp decrease over the 20, 30, and 40%Ga
2O
3-Al
2O
3 catalysts. In the same temperature range,
, which showed, in general, a mild upward trend with temperature, exhibited an abrupt increase, which was always followed by a decrease at the same levels as those obtained below 600 °C. Interestingly, C
2H
4 and CH
4 formation were limited below 600 °C for all composite metal oxides, while their production was enhanced at higher temperatures, with the corresponding selectivities, at a given temperature, decreasing as Ga
2O
3 content was increased from 10 to 40 wt.%. In contrast, bare Ga
2O
3 exhibited the highest values of
and
in the entire temperature range examined, and the lowest
at elevated temperatures (
Figure 6f).
The addition of Ga
2O
3 also led to the formation of traces of C
2H
6 due to propane hydrogenolysis:
The decrease in
and
between 600 and 670 °C can be attributed to the reaction of propylene decomposition, which may be partially responsible for the observed production of CH
4 and C
2H
4:
Taking into account that
was always higher than that of
, part of these compounds may also be produced via the following reaction:
whereas the reactions of propane decomposition (15) and (16) and/or propane hydrogenolysis (17) cannot be excluded:
Reaction (14) in combination with the reverse WGS (2) and reverse Boudouard (3) reactions may also contribute to the notable increase in
between 600 and 670 °C. High reaction temperatures are also known to favor the reaction of dry propane reforming (18), which also favors the formation of CO:
The effect of Ga
2O
3 content on the propane conversion, propylene yield, and product selectivities can be better seen in
Figure 7, where the corresponding measurements were obtained at 600 °C. Catalytic activity was optimized in the presence of 30 wt.% Ga
2O
3. Specifically,
and
were remarkably increased from 4 to 58% and from 1.5 to 39%, respectively, with increasing Ga
2O
3 content from 0 to 30 wt.%, followed by their gradual decrease to the initial values with further increase in Ga
2O
3 content to 40 and 100 wt.% (
Figure 7a). Interestingly,
at 600 °C increased significantly from 28% for bare Al
2O
3 to ~68% for the samples containing 10%, 20%, and 30%Ga
2O
3 and decreased to 64 and 41% for the 40%Ga
2O
3-Al
2O
3 and bare Ga
2O
3, respectively (
Figure 7b). The opposite trend was observed for
and
measured at 600 °C, which were minimized to the same value of ~4.5% for all composite metal oxides, while higher values were obtained for the bare metal oxides (
= 17% and
= 30% for Al
2O
3,
= 16.5% and
= 7% for Ga
2O
3). The effect of Ga
2O
3 on
was less important (ranging between 21 and 35%), most possibly because, as discussed above, CO may originate from various reactions under CO
2-ODP conditions (reactions (1)–(3), (14) and (18)), which may be affected in a different manner by Ga
2O
3 loading. Therefore, the increase in
induced by some of these reactions may be balanced by the decrease in
caused by others, thus leading to relatively low fluctuations.
Results of
Figure 5,
Figure 6 and
Figure 7 clearly indicate that the catalytic activity is strongly affected by the concentration of Ga
2O
3, which is not only able to increase propane conversion to propylene but also to suppress side product formation at temperatures of practical interest (<600 °C). This can also be seen in
Figure S7a, where the ratio of
/
at 600 °C was plotted as a function of Ga
2O
3 content. As it is observed, the
/
ratio goes through a maximum value of 17 for the 20%Ga
2O
3-Al
2O
3 catalyst, which was 18- and ~7-fold higher than those measured for the corresponding bare Al
2O
3 and Ga
2O
3, respectively. This implies that the C–H bond cleavage against that of the C–C bond can be optimized with the addition of a suitable amount of Ga
2O
3 content.
Results of
Figure 2,
Figure 3 and
Figure 4 and
Table 2 and
Table 3 showed that the acid site density was progressively increased with increasing Ga
2O
3 content, while a volcano-type behavior was found to exist between this parameter and surface basicity, with the maximum value being observed for the 20%Ga
2O
3-Al
2O
3 catalyst. In an attempt to understand the effect of acid/base properties on catalytic activity,
and
measured at 600 °C were plotted as a function of the total amount of CO
2 desorbed during CO
2-TPD (
Table 2) and the acid site density measured by the potentiometric titration measurements (
Table 3). Results are presented in
Figure 8, where a noteworthy correlation was found to exist. Specifically,
and
exhibited optimum values for intermediate values of surface basicity and acidity, which both correspond to the sample containing 30 wt.% Ga
2O
3. This is in agreement with our previous studies over M
xO
y-TiO
2 (M: Zr, Ce, Ca, Cr, Ga) and M
xO
y-SiO
2 (M: Ca, Sn, Cr, Ga) catalysts [
4,
5], providing evidence that the number of both acid and basic sites determines the CO
2-ODP activity. Comparing the results of
Figure 8 with those presented in
Figure 2 and
Figure S5 shows that catalytic activity was mainly determined by the strong acidic and the weak basic sites of the catalyst surface.
The surface basicity was also found to influence the
ratio, which was significantly higher for all Ga
2O
3-Al
2O
3 catalysts and optimum for the 10% and 20%Ga
2O
3-Al
2O
3 catalysts, (
/
), compared to the bare single metal oxides characterized by lower surface basicity (
Figure S7b). On the other hand, the
ratio exhibited a volcano-type correlation with respect to the acid site density, with the 10% and 20%Ga
2O
3-Al
2O
3 catalysts presenting the maximum values (
) (
Figure S7c). Although the aforementioned optimum
/
ratios were not observed for the most active 30%Ga
2O
3-Al
2O
3 catalyst, results clearly indicate that the C–H bond cleavage was facilitated compared to C–C bond break over samples characterized by moderate surface basicity and acidity. Based on the above, the acid/base properties of catalysts can be considered as the key physicochemical properties for the CO
2-ODP reaction.
An optimum propane dehydrogenation activity for intermediate Ga-related Lewis surface acidity, which was achieved by optimizing the Ga/Al atomic ratio, was also reported by Castro-Fernandez et al. [
58]. Among various Ga/Al atomic ratios examined (1:6, 1:3, 3:1, and 1:0), superior catalytic activity, propylene selectivity, and stability were obtained for Ga/Al = 1:3, which were attributed to the higher abundance of four-coordinated Ga sites and the higher relative number of weak/medium Lewis acid sites. The increase in Ga
2O
3 loading from 1 to 9 wt.% was found to increase the fraction of gallium in the oxidized state over xGa
2O
3/SBA-15 catalysts, with the TOF of propane conversion, however, being maximized for an intermediate Ga
2O
3 loading (5 wt.%) [
17]. An intermediate Ga/Al ratio equal to 3:8 of Ga
2O
3-Al
2O
3 nanofibers was also reported to present superior
and
of 48.4 and 96.8%, respectively, for the CO
2-assisted oxidative dehydrogenation of propane at 500 °C [
3]. Moreover, Tedeeva et al. [
19] pointed out the importance of the acid sites in achieving high catalytic activity for the propane dehydrogenation in the presence of CO
2 over Ga/SiO
2 catalysts. The authors studied catalysts with different Ga content in the range of 3–50 wt.% dispersed on three different SiO
2 powders characterized by different textural properties, and they found that both the nature and texture of the support, as well as the Ga content, influence catalytic activity. The best results (
= 33% and
= 84% at 650 °C) were obtained when Ga
2O
3 oxide with Ga content of 7 wt.% was supported on the SiO
2 powder characterized by the highest specific surface area, which also exhibited a higher number of Brønsted acid sites. Besides the high initial activity of this sample, a decrease in
was observed after 10 h on stream which, was stabilized at 20% from 10 to 20 h of continuous operation. The higher surface total moderate acidity of alumina supported Ga
2O
3 catalysts synthesized by the atomic layer deposition method was also found to facilitate the conversion of C
3H
8 to C
3H
6 by decreasing the energy barrier for the activation of C-H bond [
2]. The effect of Ga loading was investigated in the range of 1–2.9 wt.%, and results showed that the sample containing 2.9 wt.% Ga presented the highest performance, which, although it was drastically deactivated during the first 45 min on stream,
and
were stabilized at 38 and 82%, respectively, at 600 °C for the next ~2 h. It is of interest to note that propane conversion and propylene yield obtained in the present study for the 30%Ga
2O
3-Al
2O
3 catalyst (
= 59% and
= 39% at ~600 °C) was higher than most of those reported thus far in the literature, enabling the operation of the reaction at low temperatures which besides the advantage of inhibiting side reactions, offer the benefit of low energy requirements and thus, low operational cost.
3.3. TOS Stability Tests over 30%Ga2O3-Al2O3
The most effective 30%Ga
2O
3-Al
2O
3 catalyst was subjected to a time on stream stability test for a period of ~12 h at a constant temperature of 550 °C, and results are presented in
Figure 9a in terms of C
3H
8 conversion (
) and product selectivity (
) versus time. It was observed that after an initial period of 4 h on stream, where
progressively decreased from 35 to 23%, it exhibited a rather stable performance, periodically fluctuating between 22 and 32% until the end of the stability test. Interestingly, the selectivity towards reaction products remained constant during the entire time of the experiment, taking values of
= 72%,
= 22%,
= 2.5%,
= 1.2% and
< 1%. In order to investigate the potential carbon deposition on the catalyst surface during this experiment, the temperature was decreased to 25 °C in He flow, and the catalyst was exposed to 1%O
2 (in He) flow, followed by a TPO experiment using a heating rate of 10 °C/min. Results (
Figure 10) showed that CO
2 started to elute above 40 °C, giving rise to a weak peak centered at 110 °C, followed by a major peak above 300 °C with a maximum at ~580 °C. The oxidation of carbon was not completed when the temperature reached 800 °C, and thus, the catalyst remained at this temperature for ~6 min until the CO
2 response returned to the baseline. The amount of coke formed was estimated by integrating the area below the CO
2 response curve versus time and found to be 2950.8 μmol·g
−1. Results indicate that despite the significant amount of the so-formed carbon, the 30%Ga
2O
3-Al
2O
3 catalyst exhibited a sufficiently stable performance for 12 h on stream. When the TPO experiment was completed, the catalyst was again exposed to the reaction mixture at 550 °C for 5 h, and the results (
Figure 9b) showed that the values of both
and
were identical to those presented in
Figure 9a.
Similar experiments were carried out at 600 and 650 °C, aiming to explore the effect of reaction temperature on the catalyst’s stability and tendency towards carbon formation. It was found that the catalyst interaction with the gas stream at 600 °C led to a gradual decrease in the propane conversion from 56 to 13% within the first 7.5 h on stream, which then remained almost constant up to 12.5 h (
Figure S8a). This decrease was accompanied by a decrease in
(from 61 to 37%) and a parallel increase in
(from 28 to ~38%),
(from 4 to 8.3%) and
(from 4.9 to 15%), indicating that the oxidative dehydrogenation of propane was hindered most possibly due to the enhancement in the side reactions discussed above ((12)–(17)), which lead to the undesired CH
4, C
2H
4, and most possibly carbon formation, which eventually results in catalyst deactivation. The increase in
implies that either reaction (14) becomes significant and/or part of the coke formed was gasified via the Reverse Boudouard reaction (3). The deposition of carbon was corroborated by the TPO experiment conducted immediately after the stability test. The profile of the CO
2 thus produced was qualitatively similar with that presented in
Figure 10, with the amount of CO
2 thus produced being significantly higher (6534.5 μmol·g
−1) and accompanied by a parallel, but smaller, production of CO (94.4 μmol·g
−1) when the reaction was taking place at 600 °C (
Figure S9). The simultaneous consumption of CO
2 above 750 °C, where CO was eluted, implies that part of the CO
2 produced during TPO interacted with the accumulated carbon, generating CO most possibly through the reverse Boudouard reaction [
64]. As it is observed in
Figure S9, a significantly longer time (95 min) of stay at 800 °C was required in order for carbon to be completely removed from the catalyst surface compared to that shown in
Figure 10b. Although catalytic activity was partially regained following the complete carbon oxidation during the TPO experiment, a similar loss of catalytic activity was observed after the subsequent catalyst exposure to the gas stream for 4 h (
Figure S8b). The affinity of 30%Ga
2O
3-Al
2O
3 catalyst towards coke formation was most likely related to the large number of acidic sites characterized in this sample (
Table 3,
Figure 4) [
2,
28,
65].
Catalyst was also deactivated when the TOS stability test was conducted at 650 °C, as shown by the substantial decrease in propane conversion from 45 to 5.5% after continuous catalyst operation for ~12 h (
Figure S10a). Contrary to what was observed at 600 °C,
remained stable at 650 °C ranging between 38 and 42%, while
decreased from 35 to 22% and
and
increased from 8 to 14% and from 10 to 26%, respectively. These findings demonstrate that the formation of CH
4 and C
2H
4 was enhanced with time at the expense of CO, implying that the side reactions ((12)–(17)), which produced C
2H
4, CH
4 and C, inhibited the RWGS (2) and reverse Boudouard (3) reactions, which produced CO. It can then be suggested that the rate of carbon deposition was higher than the rate of carbon gasification through the reverse Boudouard (3) reaction when the reaction occurred at 650 °C. This was also confirmed by the TPO experiment (
Figure S11) conducted after the TOS stability test shown in
Figure S10a, where higher amounts of CO
2 (6743.0 μmol·g
−1) and CO (184.3 μmol·g
−1) were produced. It should be noted that the catalyst needs to remain at 800 °C for 120 min in order for the oxidation of carbon to be completed. The subsequent exposure of the catalyst to CO
2-ODP conditions showed that catalytic activity was restored following carbon oxidation, but it was rapidly lost within the first 5 h on stream.
The results discussed above clearly indicate that carbon deposition on the catalyst surface is enhanced as the reaction temperature increases, leading to gradual catalyst deactivation. However, catalytic activity remains stable under conditions (T < 600 °C) where the formation C2H4, CH4 and C2H6 is limited providing evidence that the 30%Ga2O3-Al2O3 is an efficient catalyst for the CO2-assisted hydrogenation of propane provided that the reaction conditions, and especially reaction temperature, are properly selected in order side reactions to be suppressed.
3.4. In Situ DRIFTS Studies for the CO2-ODP Reaction
The CO
2-ODP reaction was also investigated by in situ DRIFTS in an attempt to identify the reaction intermediates formed on the catalyst surface under reaction conditions. Representative DRIFT spectra collected at selected temperatures over bare Al
2O
3 and 30%Ga
2O
3-Al
2O
3 catalysts following their interaction with a feed stream of 1%C
3H
8 + 5%CO
2 (in He) at 25 °C are presented in
Figure 11. In the case of bare Al
2O
3 (
Figure 11a), the spectrum recorded at 25 °C was consisted of two negative bands (3751 and 3680 cm
−1) due to surface OH groups of Al
2O
3 which may serve as active sites for CO
2 adsorption, two bands previously assigned to bicarbonate species (1428 and 1226 cm
−1), three bands due to unidentate and bidentate carbonates (1636, 1570 and 1381 cm
−1) and several bands in the
ν(C-H) region (3000–2850 cm
−1) [
36,
37,
38,
39,
40,
41,
42]. It should be noted that the pair of peaks at 1570 and 1381 cm
−1 is also characteristic of formate species and therefore, their formation on the catalyst surface cannot be excluded [
3,
66,
67,
68]. If this is the case, then the occurrence of the RWGS may be possible since formates have been proposed as active intermediates in this reaction [
69]. Regarding the bands detected in the wavenumber range of 3000–2850 cm
−1, they can be better seen in
Figure 11b, where six bands can be clearly discerned which can be attributed to asymmetric (2980 and 2967 cm
−1) and symmetric (2960 cm
−1) C–H stretching vibrations in methyl groups (CH
3,ad), to asymmetric (2902 cm
−1) and symmetric (2875 cm
−1) vibrations in methylene groups (CH
2,ad) as well as to
νs(CH
2)/ν
as(CH
3) of gaseous propane (2886 cm
−1) [
5,
67,
70,
71].
Stepwise increase in temperature at 350 °C led to a gradual decrease in the relative intensity of all bands as well as to the splitting of the peak at 1570 cm
−1 into two peaks located at 1588 and 1507 cm
−1. As mentioned above, the former one was previously assigned to bidentate or formate species, while the latter one can be attributed to unidentate carbonates adsorbed on the Al
2O
3 surface [
3,
40,
47,
66,
68,
72,
73]. It is of interest to note that a similar band detected at 1507 cm
−1 under propane dehydrogenation conditions, both in the presence and absence of CO
2 over Ga
2O
3-Al
2O
3 catalysts, was also attributed by Han et al. [
3] to adsorbed C
3H
7* species, the dehydrogenation of which was found to be the rate-determining step. As can be seen in
Figure 11a, a further increase in temperature to 500 °C resulted in an increase in the relative intensity of the band at 1507 cm
−1, implying that the formation of the corresponding species was enhanced at elevated temperatures. This further supports the above suggestion that C
3H
8 dehydrogenation to adsorbed C
3H
7* species may take place under the present reaction conditions.
Similar peaks were detected in the DRIFT spectra obtained over the 30%Ga
2O
3-Al
2O
3 catalyst (
Figure 11c,d), indicating that the addition of Ga
2O
3 did not affect the nature of the species formed under reaction conditions. The main difference observed was that the band at 1507 cm
−1 started to be developed at lower temperatures (~250 °C) compared to bare Al
2O
3 and its relative intensity was found to be higher at a given temperature implying that the formation rate of the corresponding species was higher over the most active 30%Ga
2O
3-Al
2O
3 catalyst (
Figure S12). This reinforces the above assumption that the band at 1507 cm
−1 was due to adsorbed C
3H
7* species. Moreover, the relative intensity of the bands due to bicarbonates (1430 and 1228 cm
−1) was higher in the presence of 30%Ga
2O
3 on Al
2O
3, especially below 250 °C (
Figure 11), most possibly due to the higher basicity of this sample which enhanced the CO
2 activation in agreement with the results of
Figure S3. In addition to CO
2 activation, the enhanced surface basicity has also been reported to hide the adsorption of the undesired C
2H
4 on the catalyst surface, thus inhibiting its subsequent deep oxidation to carbon oxides [
74]. It should be noted that a new band at 3734 cm
−1 was developed at ~400 °C, which, according to the literature, was due to surface OH groups created by H
2O adsorption [
3,
5,
75]. This band increased in intensity with increasing temperature up to 500 °C and was accompanied by the appearance of a new weak band at 3588 cm
−1, which was previously suggested to be raised through H
2O interaction with weak basic hydroxyl groups on the metal oxide surface [
75]. Water adsorption may be generated through the RWGS reaction [
3,
5], which seems to be enhanced over the most active 30%Ga
2O
3-Al
2O
3 catalyst, as evidenced by the absence of similar bands from the spectra obtained from the least active Al
2O
3 support at least below 500 °C.
In an attempt to further explore the reactivity of surface species formed under CO
2-ODP conditions, a DRIFTS experiment was conducted under transient conditions at constant temperature over the most active 30%Ga
2O
3-Al
2O
3 catalyst. In this experiment, the catalyst was exposed to 1%C
3H
8 + 5%CO
2/He at 500 °C, followed by spectra recording as a function of time. As it can be seen in
Figure S13, the spectrum collected at 2 min is characterized by bands due to (a) asymmetric (2982 and 2966 cm
−1) C–H stretching vibrations in methyl groups (CH
3,ad), (b) asymmetric (2903 cm
−1) vibrations in methylene groups (CH
2,ad), (c) bicarbonate species (1435 and 1228 cm
−1), (d) bidentate carbonate species (1634 cm
−1) and adsorbed C
3H
7* species (1505 cm
−1). Stepwise increase in reaction time to 20 min resulted in a progressive decrease in the relative population of bicarbonates and bidentate carbonates, accompanied by an increase in that of C
3H
7* species and the gradual development of two new bands at 1588 and 1390 cm
−1 due to formate formation on the catalyst surface (
Figure S13b). Although the bands in the 3100–2750 cm
−1 region remained unaffected with increasing the reaction time up to 20 min, a new band was discerned at 3085 cm
−1 after 30 min on stream, which according to previous studies can be attributed to asymmetric vibrations of the C-H bond of methylene (CH
2,ad) groups of adsorbed propylene on the catalyst surface (
Figure S13a) [
76]. Further increase in reaction time up to 60 min led to an additional increase in the intensity of bands due to formates, C
3H
7* species and adsorbed propylene and the development of two shoulders at 2935 and 2921 cm
−1 which can be assigned to
νas(CH
3) and
νs(CH
3) of adsorbed propylene [
76]. Results of
Figure S13 clearly indicate that both the RWGS and propane dehydrogenation reactions are operable at 500 °C over the 30%Ga
2O
3-Al
2O
3 catalyst.
In order to corroborate the contribution of adsorbed and/or gas phase propylene to the bands detected under CO
2-ODP conditions, an additional DRIFTS experiment was carried out where the interaction of the 30%Ga
2O
3-Al
2O
3 catalyst with a 10% C
3H
6 (in He) mixture was investigated in the temperature range of 25–500 °C. Results (
Figure S14) showed that propylene adsorption led to the appearance of several spectral features in the
ν(C-H) region which were due to asymmetric and symmetric C-H bond vibrations of the methyl (CH
3,ad) and methylene (CH
2,ad) groups of adsorbed or gas phase propylene [
76,
77]. The bands at 3085, 2935, and 2921 cm
−1 observed in
Figure S13a are also discernible in
Figure S14b, confirming the production of propylene when the catalyst interacts with the 1%C
3H
8 + 5%CO
2 (in He) mixture. Six bands were also observed below 1700 cm
−1 (
Figure S14a) which were attributed to the C=C bond stretch (1665 and 1638 cm
−1) as well as to asymmetric and symmetric bending vibrations of the methyl (CH
3,ad) groups (1475, 1442, 1393, 1377 cm
−1) of adsorbed or gas phase propylene [
76,
77]. Taking into account that some of these features were also present in the spectra of both
Figure 11 and
Figure S13, it can be argued that generated propylene may also contribute to their development.
Based on the above, it can be suggested that both C
3H
8 and CO
2 are activated on the catalyst surface as evidenced by the formation of CH
x and carbonate-like species, respectively, at low reaction temperatures. Regarding the CO
2-ODP reaction mechanism, two general mechanistic schemes have been proposed: the one-step oxidative route and the two-step oxidative route, which differ mainly in the role of CO
2 [
3,
5,
18]. According to the former one, the lattice oxygen ions abstract hydrogen atoms from C
3H
8, producing C
3H
6 and H
2O, while CO
2 re-oxidizes the reduced surface following the Mars–Van Krevelen mechanism to complete the redox cycle [
18]. In order to explore if the catalyst surface is able to easily re-oxidized by CO
2, immediately after the H
2-TPR experiment conducted over the 30%Ga
2O
3-Al
2O
3 catalyst (
Figure S6), the flow was switched to a 5%CO
2/He mixture at 500 °C for 30 min followed by a subsequent H
2-TPR under identical conditions with those discussed above. No reduction peak was observed in the H
2-TPR profile (
Figure S15), implying the CO
2 was not able to re-oxidize the catalyst surface, providing additional evidence that the one-step oxidative route was not operable for the Ga
2O
3-Al
2O
3 catalysts of the present study. Results are in agreement with those reported by Getsoian et al. [
13], who demonstrated that Ga
3+ was not reduced to Ga
+ over Ga–SiO
2 and Ga–H-BEA catalysts rendering the redox mechanism unfavorable and was also supported by computational studies over Ga-zeolite catalysts, which demonstrated that non-redox mechanisms of alkane dehydrogenation reactions proceed with much lower energy barriers than those required for the reduction of Ga
3+ to Ga
+ [
78,
79].
Based on the DRIFTS results of
Figure 11 and
Figures S12–S14, the reaction seems to proceed via the two-step oxidative route, according to which C
3H
8 is dehydrogenated on the catalyst’s acid sites, leading to the formation of an intermediate adsorbed surface C
3H
7* species. Hydrogen produced from this step is removed with the indirect contribution of CO
2, which is activated on the catalyst’s basic sites and participates in the RWGS reaction, shifting the thermodynamic equilibrium towards C
3H
6 formation. The RWGS, which has been proposed to occur via intermediate formation of formate species (originating by carbonates/bicarbonates interaction with hydrogen atoms [
80,
81,
82]), is operable under the present reaction conditions as evidenced by the detection of bands due to adsorbed formates and steam and seems to be enhanced in the presence of Ga
2O
3. The enhancement in the RWGS may also be responsible for the inhibition of the C–C bond cleavage of the intermediate C
3H
7* species, thus leading to a decrease in the formation rates of the undesired C
2H
x and CH
4 and their corresponding selectivities (
Figure 4d and
Figure S7). The above findings demonstrate that although CO
2 does not directly participate in the dehydrogenation step, its role is decisive in propylene production. Results of the present study clearly show that the aforementioned steps of the C
3H
7* formation and the RWGS reaction are favored over the 30%Ga
2O
3-Al
2O
3 catalyst characterized by a moderate number and strength of both acid and basic sites, confirming the crucial role of acid/base properties on propylene production.
3.5. CO2-ODP Reaction Scheme Under Transient Conditions
The reaction scheme was also investigated by transient-MS technique over bare Al
2O
3 and 30%Ga
2O
3-Al
2O
3 catalysts using a feed composition consisting of 1%C
3H
8 + 5% CO
2 (in He) and a linear temperature ramp of 10 °C/min. The TPSR profile obtained from bare Al
2O
3 is shown in
Figure 12a, where it is observed that the concentrations of reactants, CO
2 and C
3H
8, started to decrease at temperatures higher than 650 °C. This decrease was accompanied by the simultaneous evolution of C
3H
6, CO, H
2, CH
4, and C
2H
4, implying that in addition to the propane oxidative dehydrogenation, the reactions of propane and propylene decomposition and/or propane hydrogenolysis ((12)–(17)) were taking place. The concentrations of CH
4 and C
2H
4 became higher than that of C
3H
6 above 720 °C, implying that the latter undesired reactions were favored with increasing temperature, in excellent agreement with the results of catalytic performance tests (
Figure 6a).
The TPSR pattern obtained from 30%Ga
2O
3-Al
2O
3 catalyst is presented in
Figure 12b where it can be seen that propane dehydrogenation was initiated at significantly lower temperatures compared to bare Al
2O
3, as evidenced by the onset of C
3H
8 consumption and C
3H
6 and H
2 evolution at ~450 °C, i.e., at temperatures where the FTIR band assigned to adsorbed C
3H
7* species (1507 cm
−1) was clearly discerned (
Figure 11b). The concentration of C
3H
6 went through a maximum at around 570 °C and then progressively decreased with further increasing temperature, in excellent agreement with the results of
Figure 5b, where
was found to be optimized at ~600 °C. As noted above, the decrease in C
3H
6 concentration can be attributed to its consumption via the propylene decomposition reaction (13), which may be responsible for the evolution of CH
4 and C
2H
4. Contrary to C
3H
6, the hydrogen concentration was progressively increased with increasing temperature up to 680 °C and was slightly decreased when the temperature reached 750 °C. This indicates that the origin of H
2 generation was not limited to propane dehydrogenation reaction, but as discussed above it may be also produced through the propane decomposition (16) and the dry reforming of propane (18), with the former reaction being more possible taking into account that H
2 concentration is twice that of CH
4 (
Figure 12b). This was further supported by the fact that although C
2H
4 was eliminated above 670 °C, CH
4 concentration increased continuously with increasing temperature to 750 °C.
Propane hydrogenolysis via reactions (12) and (17) may also contribute to the continuous upward trend of CH
4 response, as well as to the observed production of C
2H
6 traces (reaction (12)). The concentration of CO
2 started to decrease at similar temperatures with C
3H
8 (~450 °C), and as discussed above, it can be converted to CO via the RWGS (2), the reverse Boudouard (3), and/or the dry reforming of propane (18). However, CO started to elute above 550 °C. This can be correlated with DRIFTS results of
Figure 11c, where it was shown that CO
2 was initially (at low reaction temperatures) adsorbed on the catalyst surface in the form of carbonate-like species which most possibly interact with hydrogen atoms that were abstracted from propane molecule during its dehydrogenation, yielding formates (RWGS reaction), which are considered, at least in part, as precursors of CO (generated at higher reaction temperatures). Moreover, the formation of carbonate-like structures was previously found to be advantageous for the reaction between CO
2 and coke during the reverse Boudouard reaction pathway [
83].
Immediately after completion of the TPSR experiments presented in
Figure 12a,b, TPO experiments were conducted in order to estimate the amount of carbon deposited on the catalyst surface during CO
2-ODP reaction. The profile of CO
2 (
Figure 12c) thus produced from the bare Al
2O
3 sample exhibited a weak peak at 315 °C as well as a shoulder at around 550 °C followed by a peak of higher intensity centered at around 675 °C, indicating that three distinct carbon species are present on the surface of “spent” Al
2O
3 which can be attributed to the propylene and/or propane decomposition reactions ((13) and (16)). This was also the case for the 30%Ga
2O
3-Al
2O
3 catalyst, the CO
2 response curve of which consisted of two weak peaks at 175 and 670 °C and a major one with a maximum at 545 °C. The amounts of CO
2 produced during TPO experiments, which is equivalent to the amount of carbon deposited during TPSR experiments, were estimated by integrating the area below the CO
2 response curves and found to be significantly higher for the 30%Ga
2O
3-Al
2O
3 catalyst (496.5 μmol g
−1 corresponding to 9.6 μmol m
−2) than bare Al
2O
3 (115.8 μmol g
−1 corresponding to 1.8 μmol m
−2). This may be due to the higher acid site density of the 30%Ga
2O
3-Al
2O
3 catalyst (
Table 3), which has been previously accused of the enhanced tendency of the catalyst towards carbon deposition [
2,
8,
14,
17,
84] and is most possibly correlated with the short lifetime of catalyst at reaction temperatures higher than 600 °C (
Figures S8 and S10). It should be mentioned that despite the higher amount of carbon deposition on the 30%Ga
2O
3-Al
2O
3 surface, the evolution of the CO
2 peaks appeared at lower temperatures for this sample, implying that coke gasification is facilitated compared to bare Al
2O
3.
The overall carbon balance of the TPSR experiments was calculated using the following equation, where the amount of coke formed on the catalyst surface estimated by TPO experiments was taken into account as follows:
For both catalysts examined, the carbon balance was satisfactory, with a deviation of 1% for Al2O3 and 5% for 30%Ga2O3-Al2O3.
Overall, it can be suggested that the dispersion of a suitable amount of gallium oxide on the alumina surface is able to modify the acid/base properties of the alumina support, providing the appropriate number of both acidic and basic sites. This facilitates the activation of reactants and their selective conversion to C3H6 and CO at low temperatures, where undesired reactions are suppressed, ensuring a stable catalyst performance with time on stream.