2.1. Catalytic Runs of Toluene Steam Reforming over CeZrO2 and Ni/CeZrO2
The results of the catalytic runs of steam reforming of toluene (SRT) over CeZrO
2 and Ni/CeZrO
2 conducted at S/C = 2.4 and t
c = 0.36 s are presented in
Table 1, which shows that toluene converts mainly to CO and CO
2, with some insignificant transformation to benzene in the reaction of steam dealkylation (Equation (5)). Some works reported the formation of CH
4 during SRT [
29]; nevertheless, it was not detected in this study. Methane can be produced in the reaction of toluene hydrodealkylation (Equation (6)) and as an important precursor for carbon deposition (Equation (7)), its formation during SRT is unwanted. In the present work, the only by-product detected by GC and GC-MS in the product gas was benzene. From
Table 1, it can be seen that toluene conversion to benzene is higher over CeZrO
2 than over Ni/CeZrO
2, and for both, it decreases with the increase of the reaction temperature. The selectivity to benzene is much higher for CeZrO
2 than for Ni/CeZrO
2. Bion et al. [
42], who studied toluene steam dealkylation over Al
2O
3-supported Rh, Pd, Pt, Ni, Co, Ru, and Ir, found out that contrary to the SRT reaction, toluene steam dealkylation is insensitive to the nature of the metal and the structure of the catalyst (that is also changing due to carbon deposition). However, toluene steam dealkylation was found to be very sensitive to the nature of the support. For example, for metal catalysts supported on alumina, silica, or chromium oxide, toluene dealkylation to benzene (that immediately desorbs to the gas phase) and CH
x species takes place on the metal active sites, while H
2O is activated on the support sites. In the next step, the OH groups diffuse from the support to the CH
x (remaining on the metal sites) and oxidize them to CO and H
2. According to Wang and Gorte [
43], toluene steam dealkylation is not occurring on the CeO
2-supported Pt and Pd catalyst. Nevertheless, in the present work, it is shown that some small amounts of benzene are formed during SRT over CeZrO
2 alone, but production of this hydrocarbon is significantly decreased in the presence of Ni.
In general, the performance of Ni/CeZrO
2 is much better than of CeZrO
2—especially at temperatures up to 700 °C, which is owing to the presence of the Ni active phase. But it must be pointed out that CeZrO
2 alone exhibits relatively good performance in SRT, especially at elevated temperatures (800 and 900 °C). Thus, it is evident that ceria-zirconia provides some catalytic activity to the Ni/CeZrO
2 system. The reduced ceria (Ce
3+) is responsible for H
2O dissociation, while Ce
4+ is the site for toluene dealkylation to CH
x and C
6H
6, followed by the hydrogenation of the latter to a cyclic ring that breaks to form CH
x. The CH
x species are dehydrogenated to H
2 and active carbon species that are oxidized to CO using the oxygen from CeO
2 (supplied by dissociating H
2O molecules). In the Ni/CeZrO
2 catalyst, ceria-zirconia is considered as the oxygen source that is used for the oxidation of carbon species on the Ni, rather than as the active site for toluene activation. Whereas the nickel species Ni
0 and Ni
x+ are responsible for C
7H
8 and H
2O dissociation, respectively.
2.2. The Influence of S/C on the Ni/CeZrO2 Activity in SRT
The results of the catalytic runs conducted at S/C = 1, 1.2, 2.4, and 4 (
Figure 1) prove that the increase in H
2O content in the reaction mixture has a positive influence on C
7H
8 conversion. The most important effect of S/C on hydrocarbon conversion is observed at lower temperatures. For example, at 420 °C, toluene conversion is only 30% for S/C = 1 and 96% for S/C = 4. It should be noticed also that at a high steam content, the conversion of toluene is almost constant in the whole range of applied temperatures (
Figure 1d). A slight increase in S/C (from 1 to 1.2) does not influence toluene conversion; however, it enables to improve the suppression of carbon deposits and protect the catalyst from deactivation. A further increase in H
2O concentration in the feed (
Figure 1c,d) results in a visible increase in hydrocarbon conversion. For example, at a temperature as low as 420 °C, toluene conversion is 66% and 96% for SC = 2.4 and 4, respectively. At an increased steam excess, the Ni/CeZrO
2 catalyst shows very good performance in SRT even at lower temperatures (500–600 °C). For S/C = 2.4, toluene conversion of ca. 90% is observed from 600 °C, while for S/C = 4, over 99% toluene conversion is noticed even from 500 °C.
It can be seen from
Figure 1 that the yields of CO and CO
2 also depend on the applied S/C ratio. The production of CO
2 during catalytic runs of SRT is ascribed to the occurrence of the WGS reaction (Equation (3)), which is thermodynamically favored at lower temperatures. The CO
2 formation during catalytic runs increases with increasing S/C. The higher the H
2O concentration in the feed, the more CO
2 is detected at the reactor’s outlet. For S/C = 4, the CO
2 formation predominates in the whole range of applied temperatures.
The GC-MS analyses at the reactor’s outlet during tests carried out at S/C = 1 and 2.4 (
Table 2) revealed that in the 420–600 °C range, the SRT over Ni/CeZrO
2 produces small amounts of benzene. It is observed that the formation of benzene is inversely proportional to the temperature and steam excess. No benzene is produced at 700–900 °C for S/C = 1 and 2.4. Moreover, benzene was not present in the product gas (in the whole range of temperatures) when S/C = 4. Doumani [
44] reported that during toluene dealkylation at ca. 560 and 620 °C, the conversion to benzene decreased with an increasing H
2O/C
7H
8 molar ratio. The effect of H
2O/C
7H
8 was also found to be more important at higher temperatures.
The results presented in this paper revealed that the effect of the S/C ratio on the performance of Ni/CeZrO
2 in toluene steam reforming is more important at temperatures below 600 °C. At low temperatures, high H
2O excess inhibits catalyst deactivation caused by carbon deposition and enriches the product gas in hydrogen owing to the occurrence of WGS and H
2O dissociation [
43]. However, a high content of H
2O may lead to catalyst sintering, finally causing a decrease of its activity.
According to the results of the catalytic tests presented in this paper, Ni/CeZrO2 shows high activity in SRT. If the step of dissociative adsorption of C7H4 molecules to form surface carbon species is fast and the oxidation of those carbon species is slow, the catalyst may suffer deactivation due to carbon accumulation. The excess of H2O in the reaction feed helps to suppress carbon species from the catalyst surface and inhibits carbon build-up. However, the application of a high S/C ratio may negatively influence the morphology of the catalyst. Specifically, it may cause the sintering of the Ni particles, leading to the decrease of the catalytically active surface. Moreover, when the catalyst is exposed to elevated temperatures and a high steam concentration, the CeZrO2 phase can also agglomerate and cover the Ni phase, reducing its accessibility to the regents. The impact of the S/C ratio on the morphology of Ni/CeZrO2 was examined by using X-ray diffraction (XRD) and scanning electron microscopy (SEM).
The results of the XRD analyses of Ni/CeZrO
2 before SRT and after tests carried out at S/C = 1 and 4 are presented in
Figure 2. The typical reflexes of the CeZrO
2 phase are observed, e.g., at 2θ = 28.8, 33.8, 48.5, 58.6°, while the small-intensity reflexes of the Ni
0 are detected at 44.5 and 52.6°. The mean size of Ni and CeZrO
2 crystallites calculated from the Scherrer equation increases with the S/C ratio applied during the catalytic tests. For the Ni crystallites oriented in the (111) direction, the rise of the mean size from 12.0 nm (fresh catalyst) to 15.0 nm (after tests at S/C = 1) and 20.4 nm (after tests at S/C = 4) was noticed. The mean size of CeZrO
2 crystallites oriented in the (220) direction increased from 8.3 nm to 9.6 and 15.2 nm after tests conducted at S/C = 1 and 4, respectively. Hence, high steam content in the reaction mixture causes the sintering of the support and the active phase, which results in the loss of the catalytically active surface.
The experiments of N
2 sorption (
Table 3) revealed that the specific surface area (S
BET) of the fresh and spent Ni/CeZrO
2 is very similar. Nevertheless, some increase in the mean pore size can be observed for the samples after tests conducted at high steam excess (especially at S/C = 4). As was observed by the XRD, the extended exposition to H
2O at increased temperatures led to the sintering of the catalyst particles, and during this process some small pores could have experienced coarsening to form bigger ones. Nevertheless, the SEM observations (
Figure 3) did not show significant changes in Ni/CeZrO
2 morphology after SRT tests conducted under different S/C ratios.
The experiments of N
2 sorption also revealed some minor increases in the catalyst’s surface area after SRT was carried out at S/C = 1 (
Table 3). It could be related to the presence of carbon deposits detected by SEM (
Figure 4), especially to the filamentous carbons. According to TEM (
Figure 5), the structural carbon deposits which have formed on Ni/CeZrO
2 are tubular and curly (
Figure 5a). It can be noticed that the rhombohedral nanoparticles of Ni/CeZrO
2 are attached individually or as agglomerates to the walls of CNTs (
Figure 5a–c). Concerning the morphology of the obtained structural carbon deposits, one can state that both the bamboo-like carbon nanofibers (CNFs) and the carbon nanotubes (CNTs) may form during SRT carried out at S/C = 1. The outer diameters of the structures pictured in
Figure 5b–d range from ca. 20 to 80 nm. The detection of those structures in the Ni/CeZrO
2 sample with TEM was difficult. Firstly, the deposits were located only in some areas of the sample, and secondly, the catalyst was found to be very resistant to carbon deposition, so the concentration of the filamentous carbon species was very low. The structural carbon deposits were not observed in the Ni/CeZrO
2 samples after SRT carried out in the excess of H
2O (i.e., at S/C = 1.2, 2.4, and 4). According to the TGA, the weight loss of the Ni/CeZrO
2 sample spent in SRT at S/C = 1 (
Table 4) was only 5.4%, which corresponds to 0.45 mg of carbon deposited over 1g of catalyst per 1 h of the test run. For example, Zhou et al. [
24] reported that the amount of carbon deposited over Pt/C
e1−xZr
xO
2 was 1.7 mg/g
cat/h; 3.8 times more than for the Ni/CeZrO
2 reported here. From
Table 4, it is evident that even a small increase in S/C, i.e., from 1 to 1.2, considerably enhances the suppression of the carbon species left on the catalyst surface during hydrocarbon dehydrogenation, reducing carbon deposition from 0.45 to 0.26 mg/g
cat/h. The impressive reduction of carbon accumulation during SRT over Ni/CeZrO
2 can be observed at S/C = 4. Zhu et al. [
16], who studied the Ni/Al
2O
3 catalyst in SRT, also noticed the decrease in coke formation when the S/C increased. According to their work, 0.29, 0.22 and 0.17 g of carbon deposited per 1 g of the catalyst after 5 h time-on-stream at S/C = 1, 2, and 3, respectively. Converted to mg/g
cat/h, it gives the following values: 58, 44, and 34, which are drastically higher than for the Ni/CeZrO
2 reported here. However, such a comparison is only sketchy. There are few factors that play a role in increased C deposition over Ni/Al
2O
3 [
16] and must be underlined here: (i) the Ni loading, which was 20 wt.%, i.e., twice as high as in the case of Ni/CeZrO
2 (this is very important since Ni is the site for C deposition), (ii) the support, which is not reducible, and unlike CeZrO
2, does not provide the active oxygen species for suppressing carbon species, and (iii) the contact time as low as 1.6 × 10
−5 s (shorter contact times results in increased coke formation; t
c in the present work is 0.36 s).
Based on the obtained results, the reaction pathways for the formation of filamentous carbon during SRT over Ni/CeZrO
2 can be proposed (
Figure 6).
The toluene molecule adsorbs on the Ni active site, which leads to its demethylation. The first route of filamentous carbon formation (I) assumes the formation of a biphenyl from two adsorbed phenyl groups. The combination of a few biphenyls leads first to the graphene layer, which can stack and roll up to form CNT [
41]. The biphenyl formation via dissociation of the C-I bond in iodobenzene adsorbed over Cu (111) was studied by Xi and Bent [
45]. The authors reported that phenyl groups adsorb on Cu (111) with their π-rings approximately parallel to the surface. These phenyl groups were found to be thermally stable to above 300 K, while above that temperature they can couple to form a biphenyl. The decomposition of hydrocarbon can lead to graphite layers and to the formation of CNFs [
46,
47]. The radical mechanism of the formation of bipyridyl molecule catalyzed by Ni (111) using a pyridine precursor was studied by the DFT (density functional theory) calculations by Feng et al. [
48]. In that work, the bipyridyl was assumed as the initial process of the growth of N-doped graphene. In the second route (II), CNTs or CNFs are obtained from methane decomposition. The molecules of CH
4 are released to the gas phase after the hydrogenation of the methyl groups adsorbed on the metal active site after toluene demethylation. The filamentous carbon can be also formed from CO (route III), which is a product of the steam reforming reaction. According to [
49], the potential of aromatic molecules towards carbon formation is reported to be several times higher than CO or methane.
2.3. The Influence of Contact Time
The influence of contact time (t
c) on toluene conversion and hydrogen yield during the steam reforming reaction carried out at different S/C ratios is presented in
Figure 7. For all experiments, the increase in contact time increases toluene conversion and hydrogen yield. From the presented graph, the impact of t
c is more pronounced at lower temperatures. As the temperature increases (for each S/C ratio), the dissimilarities in C
7H
8 conversions and H
2 yields for different t
c values are less significant. Toluene steam reforming, as an endothermic reaction, favors high temperatures. At elevated temperatures, the rate of reaction is high; hence, the impact of parameters such as S/C or t
c on toluene conversion is minor. At lower temperatures, the rate of SRT decreases, thus the effect of S/C and t
c is more pronounced. From the collected data, one can say that the performance of Ni/CeZrO
2 at lower temperatures (420–600 °C) can be significantly improved by increasing the contact time, while at higher temperatures, the impact of t
c is not that profound. For all experiments, toluene conversion increases with reaction temperature. The increase in hydrogen yield with temperature is observed only for S/C = 1 and 1.2, while for the experiments carried out at higher steam excess (at S/C = 2.4 and 4), the H
2 yield decreases above 500 °C owing to the thermodynamic limitations of the WGS reaction (Equation (3)).
The calculation of the H
2/CO and H
2/(CO + CO
2) ratios can give more insight into the phenomena occurring during the SRT reaction over Ni/CeZrO
2, such as the “overproduction” of hydrogen. The steam reforming reaction, especially when conducted under steam excess, is accompanied by a WGS reaction. Moreover, the presence of reduced cerium species in the CeZrO
2 allows for H
2O dissociation to H
2 and lattice oxygen. The H
2/CO will be more suitable to describe the SRT conducted at stoichiometric steam content (S/C = 1), while the H
2/(CO + CO
2) is the proper one to explain the WGS-accompanied SRT because it also considers the CO which was produced in the SR reaction but was further converted into CO
2 in the WGS reaction. An overall reaction of toluene conversion concerning the participation of three steps, i.e., SRT (Equation (8)), WGS (Equation (3)) and H
2O dissociation (Equation (9)), may be written as Equation (10). The σ is the stoichiometric number of the step. The H
2/CO and H
2/(CO + CO
2) ratios are expressed by Equations (11) and (12). By knowing the experimental values of both ratios (presented in
Figure 8), one can calculate the stoichiometric numbers for the WGS step (x) and the H
2O dissociation step (y).
For the SRT reaction taking place alone, “x” and “y” value zero. In such a case, the H
2/(CO + CO
2) and H
2/CO are equal (because CO
2 is not produced) and amount to 1.57. If H
2/CO is above that value, it indicates the occurrence of the parallel reactions producing H
2 (Equations (3) and (9)) and is typical for the ceria-zirconia supported metal catalyst.
Figure 8 shows the H
2/CO and H
2/(CO + CO
2) ratios as a function of temperature during the SRT catalytic test carried out under changing t
c and at different steam excess.
For all S/C ratios and contact times, the H
2/CO (
Figure 8a–c) exceeds 1.57, indicating the occurrence of said parallel reactions. The highest production of hydrogen out of the steam reforming catalytic cycle is observed for higher S/C ratios and at lower temperatures (especially at 420 °C, and to a lesser extent at 500 °C). This is explained by the fact that at high steam concentrations, the WGS equilibrium is shifted towards H
2 production, not to mention that this reaction favors low temperatures. The H
2/(CO + CO
2) ratios (
Figure 8d–h) also show an overproduction of hydrogen because their values exceed 1.57 (marked with dotted line “A”).
It can be observed that the contact time has no effect or a minor effect on the H
2/(CO + CO
2) ratios. Some influence of contact time is noticed at lower temperatures and for lower t
c values. The increase in S/C ratio increases H
2/(CO + CO
2), owing to increased participation of WGS reaction. If the stoichiometric number of the WGS reaction “x” (Equation (3)) reaches its maximal value of 7 (because one catalytic cycle of SRT gives 7 moles of CO, and each mole of CO can be converted into 1 mole of CO
2 within one catalytic cycle of WGS), the H
2/(CO + CO
2) is equal to 2.57, and that is indicated with line “B”. It can be seen from the graphs in
Figure 8g and h that the application of high steam excess (S/C = 2.4 and 4) results in significant overproduction of H
2 in the whole range of applied temperatures.
One must remember that except for the SRT and WGS, another reaction which produces hydrogen during toluene steam reforming over the studied Ni/CeZrO
2 catalyst is H
2O dissociation (Equation (9)). The adsorptive dissociation of steam occurs on oxygen vacancies and surface oxygen (O
2−) of ceria-zirconia and produces H
2 and Ce
4+ via formation of the surface hydroxyl (OH
−) as described by Equations (13) and (14) [
50]. These reactions are the elementary steps occurring within the catalytic cycles of SRT and WGS reactions; however, owing to the nature of CeZrO
2 support, they can also take place beyond those cycles, increasing the overall H
2 production.
The formation of oxygen vacancies in pure ceria starts from 350 °C [
51], thus allowing the formation of oxygen vacancies on which H
2O can dissociate—also beyond the catalytic cycles of the SR and WGS reactions. It is also widely known that the formation of oxygen vacancies is facilitated when ceria is doped with zirconium, because due to the smaller ionic radius of the Zr compared to Ce, the mobility of the surface oxygen is increased.
The surface hydroxyls that form during H
2O dissociative adsorption are thermally stable up to 130–330 °C and may produce either H
2 or H
2O [
52,
53]. The DFT calculations performed by Hansen and Wolverton [
50] revealed that H
2O adsorbed near the surface vacancy easily dissociates to OH
−. This step is highly exothermic. Whereas the decomposition of hydroxyls to H
2 requires the breakage of the O-H bond and is endothermic. On a fully hydroxylated ceria, the OH
− species can decompose to H
2 at high temperatures. Also, according to Henderson [
54], the rate of H
2 desorption from the hydroxylated surface of ceria increases with temperature. Carrasco et al. [
55] studied Ni/CeO
2 (111) in H
2O dissociation and found that Ni
2+ sites support the dissociation of the O-H bond. Those Ni
2+ sites are generated by the strong metal-support interaction or by the formation of Ce
1−xNi
xO
2−y solid solutions. On the contrary, the Ni atoms which form clusters and are not in an intimate contact with ceria show low activity in H
2O dissociation (which is comparable to the performance of unsupported Ni).
The contribution of H
2 (
x) produced in SRT, WGS, and H
2O dissociation over the Ni/CeZrO
2 catalyst in the product gas from toluene steam reforming was calculated using Equations (15)–(17) and is presented in
Figure 9.
As is displayed in
Figure 9a, the contribution of the H
2 produced in the SRT reaction increases with temperature and is the highest at S/C = 1. Moreover, it decreases with increasing steam excess owing to the growing participation of WGS and H
2O dissociation (
Figure 9b,c). As is presented in
Figure 10, the stoichiometric numbers of both reactions, i.e., “x” for WGS and “y” for H
2O dissociation, increase with raising the S/C ratio because more H
2O is introduced to the reaction feed. In addition, “x” and “y” decrease with temperature owing to the reactions’ thermodynamics. Based on the collected results, one can state that the impact of t
c on the contribution of both reactions occurring in parallel to SRT is minor, especially at higher temperatures. Nevertheless, to resolve this issue, more sets of catalytic runs should be conducted under varying S/C and t
c.
Table 5 shows the carbon formation over Ni/CeZrO
2 catalyst during catalytic tests of toluene steam reforming carried out under S/C = 2.4 at different contact times. The highest carbon deposition determined by TGA is observed for the lowest contact time, 0.003 s, and it decreases with increasing t
c. Hence, the application of higher contact times results in a higher conversion of toluene and H
2 yield (
Figure 7) and decreased carbon formation.