Coking-resistant dry reforming of methane over Ni/γ-Al2O3 catalysts by rationally steering metal-support interaction.

The coking issue is the main challenge for dry reforming of methane (DRM) over Ni-based catalysts. Herein, we excavate the reasons for the enhanced coking resistance of the bounded Ni over the free state Ni in Ni/γ-Al2O3 catalysts for DRM. Rational metal-support interaction of the bounded Ni would facilitate desorption of CO, thus suppressing CO disproportionation and decreasing carbon deposition. The higher activity of the bounded Ni is ascribed to better methane cracking ability, stronger adsorption, and activation of CO2 by forming polydentate carbonate. The better activation of CO2 over the bounded Ni would also contribute to the gasification of formed coke. We gain an insight into the anti-coking mechanism of DRM determined by metal-support interaction in Ni/γ-Al2O3 catalysts through mechanistic studies. It is believed that our findings would enlighten the design of more efficient catalysts for DRM.

Introduction

Dry reforming of methane (DRM) was attractive for industrial applications since it would simultaneously consume two kinds of greenhouse gases (CH4 and CO2) and produce the syngas (H2 and CO) (Fletcher and Schaefer, 2019). The produced syngas with H2/CO = 1 was conducive to downstream Fischer-Tropsch synthesis for high-value-added products. Noble metals were usually regarded as efficient reactive catalysts, but their inaccessibility and expensiveness would limit the practical use (Shi et al., 2013). As a replacement, the nickel-based catalyst was widely studied, thanks to its high activity and much cheaper price (Akri et al., 2019; Kim et al., 2017; Song et al., 2020). However, problems like sintering and coking of the catalysts have been hampering their commercial application (Li and Gong, 2014; Pakhare and Spivey, 2014; Zhang et al., 2021a).

γ-Al2O3 was extensively researched as support for nickel-based catalysts to catalyze DRM reaction because of its high surface area and good thermostability (Stroud et al., 2018). Actually, the reactivity of original Ni nanoparticles could be modulated by the oxide supports which would determine the DRM performance (Bu et al., 2019). For example, the acidic properties of γ-Al2O3 would contribute to the decomposition of CH4, giving rise to good DRM activity (Osaki et al., 1995; Shehu et al., 2019). However, Al2O3-supported Ni catalysts still severely suffered from carbon accumulation due to the CH4 decomposition and CO disproportionation (Dong et al., 2020; Joo et al., 2020). To enhance its ability of anti-coking, researchers tried to control the size of Ni particles via using specific precursors (Dama et al., 2018; Song et al., 2018) or increase the basicity of the support by incorporating alkali or alkaline earth metals into the support (Mette et al., 2016; Wang et al., 2013). Meanwhile, understanding the intrinsic activity and coke resistance behaviors of specific active sites in Ni/γ-Al2O3 was also very important for the design of better catalysts.

Ni would dissolve into γ-Al2O3 giving rise to nonstoichiometric spinel structure during calcination process, and partial Ni would exsolute from the spinel structure in the reduction process (Xu et al., 2001). Hence, three types of Ni, namely Ni2+ in spinel, the bounded Ni exsoluted from spinel, and the free state Ni on the surface of γ-Al2O3, possibly existed in the reduced Ni/γ-Al2O3 catalysts. These Ni species featured different metal-support interactions (MSIs), which would affect the performance of the catalyst in DRM. As proposed by some literatures, the tetra-coordinated Ni2+ in NiAl2O4 spinel structure was the active site for DRM (Rogers et al., 2016), and metallic Ni particles were less active but responsible for limited coking of the catalysts. However, some researchers concluded that metallic Ni on the surface rather than the Ni-Al2O3 interface was the active site in DRM, but the coke resistance behavior of specific active sites was still incompatible (Foppa et al., 2017). Normally, the Ni loading and experimental temperature (calcination (Zhou et al., 2015), reductions (Srifa et al., 2018), and reaction temperature (Cao et al., 2018)) would influence the properties of Ni which determined the DRM performance of Ni/γ-Al2O3. When the nickel aluminate (mass content of Ni ≥ 33.3 wt%) was directly used as the catalysts, their performance might be overstated (Rogers ) because CH4 might firstly react with NiO to form Ni which would catalyze the DRM reaction subsequently. Additionally, it might be imperfect to explore the roles of different active sites without controlling the temperature or mass loading (Foppa ).

Herein, the types of active Ni were preciously tuned by controlling the Ni mass loading and the annealing temperature, and then, the primary reasons for the differences of DRM performance between the species were studied. The Ni content was 0.9 wt%, 1.5 wt%, and 5.0 wt%, and the corresponding calcination and the reduced samples were named as (L, M, and H)-NiO/γ-Al2O3 and (L, M, and H)-Ni/γ-Al2O3, respectively. Meanwhile, we also gave the reasons why we chose these metal loadings (Figure S1). Two types of active Ni sites (the bounded Ni in L-Ni/γ-Al2O3 and M-Ni/γ-Al2O3 and the free state Ni in H-Ni/γ-Al2O3) with different MSIs were identified in Ni/γ-Al2O3 via ultraviolet-visible spectroscopy (UV-vis), hydrogen temperature-programmed reduction (H2-TPR), and oxygen temperature-programmed oxidation (O2-TPO) tests. Activity tests and thermogravimetric analysis (TGA) proved that, compared to the free state Ni, the bounded Ni possessed better anti-coking ability (0.087 gc·mmolNi−1·h−1 vs 0.167 gc·mmolNi−1·h−1 for coking rate), higher intrinsic activity (11.2 s−1 vs 8.48 s−1 for CH4 conversion and 13.7 s−1 vs 11.9 s−1 for CO2 conversion), and more robust stability for DRM. According to the results of in situ diffuse reflectance infrared Fourier transform spectroscopy (in situ DRIFTS), the better anti-coking ability and higher turnover frequency (TOF) of the bounded Ni were ascribed to the superior CH4 cracking capacity, stronger adsorption, and better activation of CO2. Additionally, the bounded Ni showed weaker interaction with CO than that of the free state Ni, which contributed to desorption of the products and suppressed the CO disproportionation. Our work clarified the differences in the carbon resistance and reaction mechanism of the two active sites in the Ni/γ-Al2O3 catalyst induced by different MSIs. This mechanistic research was believed to pave a way for devising better catalysts for DRM by precisely controlling the type of active Ni species.

Results

Structure of catalysts

UV-vis spectra of calcination samples are given in Figure 1A to illustrate the coordination state of nickel; all spectra were subtracted from the alumina background. It was clear that Ni2+ species occupying tetracoordinate sites appeared at 596 and 633 nm. These bands did not shift as Ni loading increased (Négrier et al., 2005), which indicated that the coordination environment of Ni2+ at this site got no changes. However, the peak at 350 nm, which was attributed to the d-d transition of the octahedral Ni2+ (Margossian et al., 2017), shifted to 381 nm with Ni content increasing. The bathochromic shift could be associated with a weaker ligand field of Ni2+ (Margossian ). The intensity of the two characteristic peaks enhanced simultaneously with increased Ni loading, meaning that Ni occupied both tetracoordinate and hex-coordinated sites at the same time (Elias et al., 2019). For the reduced samples (Figure S2), the peak positions of UV-vis spectra were similar to each other. However, NiO species in H-Ni/γ-Al2O3 were observed at 233 and 346 nm possibly as a consequence of partial oxidation of surface Ni particles (Rahman et al., 2018) with weaker MSI.

Figure 1

Structure characterization of the catalysts

(A) UV-vis spectra of the calcination samples.

(B) H2-TPR profiles of serial catalysts.

(C) O2-TPO profiles of the pre-reduced catalysts.

See also Figures S2–S9 and Tables S1, S2, and S5.

X-ray diffraction (XRD) was then used to characterize the crystal structure of reduced catalysts (Figure S3). The peak appearing at 51.7° for H-Ni/γ-Al2O3 was indexed to the facet (111) of Ni (PDF#04–0850) (Zhang et al., 2017). The characteristic XRD peaks of Ni were negligible over both L-Ni/γ-Al2O3 and M-Ni/γ-Al2O3, which might attribute to the low mass loading and better dispersion. The results of Scanning Electron Microscope-Energy Dispersive Spectrometer (SEM-EDS) mapping of the catalysts also prove this point (Figure S4), where Ni species were well dispersed in L-Ni/γ-Al2O3 and M-Ni/γ-Al2O3, but obvious agglomeration of Ni species occurred for the H-Ni/γ-Al2O3. The grain size of Ni in H-Ni/γ-Al2O3 calculated by the Scherrer equation was about 11 nm, which also coincided with the results of transmission electron microscopy (TEM). As was seen in Figures S5–S7, the size of Ni particles slightly increased from 8.9 ± 1.3 for L-Ni/γ-Al2O3 to 11.3 ± 2.5 nm for H-Ni/γ-Al2O3. On the other hand, the specific surface area (SAA) (Figure S8 and Table S1) was similar with slight increase for the catalysts. The gradual increase of the SAA might be related to the decomposition of nitrate in the pores of the support during calcination process, where a large amount of gas was generated to expand the pores. The higher the metal loading, the more the SAA increased.

H2-TPR was performed (Figure 1B) to investigate the MSI of the reduced catalysts. The peaks appeared above 750°C derived from Ni exsolution from nonstoichiometric NiAlxOy spinel structure on the surface of Al2O3 support (Mette et al., 2016). The peak gradually shifted to lower temperatures with higher Ni loading, suggesting the weaker MSI (Margossian ). It was worth noting that the peak at 467°C only appeared in H-Ni/γ-Al2O3, which showed the existence of weak-bonded NiO species (Li et al., 2014). O2-TPO of pre-reduced samples at 750°C was performed to further evaluate the catalysts' MSI and the oxidation resistance ability of active metals (Figure 1C). For H-Ni/γ-Al2O3, a strong peak appeared at 320°C which might be related to the oxidation of the weak-bonded Ni. By contrast, there was no obvious peak for L-Ni/γ-Al2O3 and M-Ni/γ-Al2O3 in the test temperature range. This might be attributed to their relatively strong MSI, which would make Ni species more difficult to be re-oxidized (Zhao et al., 2018). The results above showed that the relationship of MSI among the catalysts was L-Ni/γ-Al2O3>M-Ni/γ-Al2O3>H-Ni/γ-Al2O3.

The High Resolution-Transmission Electron Microscope (HR-TEM) images of the catalysts are shown in Figure S9 to determine the specific structure of the active sites in the catalysts. It could be observed that all the catalysts showed the (111) crystal plane of Ni with the lattice spacing of 0.204 nm (Bu et al., 2020), but the structures of the catalysts were still significantly different: a spinel overlayer was found on the Ni particle in L-Ni/γ-Al2O3 (Mette et al., 2016), while the distortion of Ni lattice without spinel layer was found for M-Ni/γ-Al2O3. For H-Ni/γ-Al2O3, the Ni particles exposed (111) crystal plane without obvious distortion and coated layer. The differences of HR-TEM images among these three catalysts suggested that the MSI would influence the properties of the Ni particles derived from exolution, then influencing the electronic state of Ni. This conclusion was further proved by X-ray photoelectron spectroscopy (Table S2) where higher binding energy was found in L-Ni/γ-Al2O3 and M-Ni/γ-Al2O3 compared with H-Ni/γ-Al2O3. Those results proved that two kinds of active Ni would emerge with increasing Ni loading: the Ni particles exsoluted from octahedral sites during reduction exhibited stronger MSI and were regarded as the bounded Ni. With high Ni content, another kind of Ni (the free state Ni) derived from weak-bonded NiO with weaker MSI would appear in H-Ni/γ-Al2O3.

Performance of catalysts

Deduced from the results of H2-TPR and relevant tests (Figures S10–S14), the bounded Ni would gradually exsolute to replenish active sites for DRM during the reaction at high temperature (750°C) (Wong et al., 2019). It would make it difficult to distinguish the differences between the bounded Ni and the free state Ni. To accurately verify their differences of activity and coking resistance, a DRM test at lower temperature 650°C was adopted to exclude the effect of exsolution, with results listed in Figures 2 and S15–S19.

Figure 2

DRM performance of the catalysts

(A) TGA curves of the used M-(blue line) and H-Ni/γ-Al2O3 (red line) samples after 20-hr DRM test at 650°C.

(B) Coking rate of the spent samples after 20-hr DRM test at 650°C based on TGA results.

See also Figures S1 and S10–S20 and Tables S3, S4, and S6.

The results of the apparent reaction rate (CH4) are shown in Figure S15. It was easy to observe that both L-Ni/γ-Al2O3 and M-Ni/γ-Al2O3 show similar reaction rates (530.6 mmol·min−1·gNi−1 and 565.3 mmol·min−1·gNi−1, respectively) derived from the bounded Ni. However, the initial reaction rate of H-Ni/γ-Al2O3 (131.5 mmol·min−1·gNi−1) was much lower compared to the former two samples. Combined with the previous analysis, it could be primarily inferred that the activity of the bounded Ni was higher than that of the free state Ni. Corresponding CO2 apparent conversion rate (Figure S16) was higher than that of CH4 because of the existence of reverse water gas shift reaction and Ni(111) possessed higher activation ability for CO2 than CH4 (Das et al., 2018; Foppa et al., 2016). It should also be noticed that M-Ni/γ-Al2O3 got less activity loss than L-Ni/γ-Al2O3 (52.4 mmol·min−1·gNi−1 to 181.2 mmol·min−1·gNi−1). This might be because when Ni loading was too low, the catalyst was more likely to be deactivated due to the dissolution of Ni into the support because of the too strong MSI during the reaction. Therefore, it could be speculated that a rational MSI was very important for the stability of the catalysts, as in the M-Ni/γ-Al2O3. The H2/CO value of ∼0.8 (Figure S17) was also similar with that of others (Al-Fatesh et al., 2020), and the mass balance and relationship between the actual reactant conversion and the thermodynamic equilibrium conversion were also given here. To further confirm the intrinsic activity between the two kinds of Ni, TOF for CH4 and CO2 was employed at 650°C (Figure S18 and Table S3). It was clear that M-Ni/γ-Al2O3 had higher TOF for both CH4 and CO2 than that of the H-Ni/γ-Al2O3 (11.2 s−1 vs 8.48 s−1 for CH4, 13.7 s−1 vs 11.9 s−1 for CO2), suggesting the higher intrinsic activity of the bounded Ni than the free state Ni. The TOF values of L-Ni/γ-Al2O3 were close to those of M-Ni/γ-Al2O3. It should also be noted that Ni/γ-Al2O3 with the bounded Ni was ahead of other catalysts in terms of the intrinsic activity (Table S4).

M-Ni/γ-Al2O3 and H-Ni/γ-Al2O3 were mainly used for the comparison in the next section while L-Ni/γ-Al2O3 would be omitted because L-Ni/γ-Al2O3 got similar active sites with M-Ni/γ-Al2O3 but was seriously deactivated in the reaction at 650°C. Severe carbon (21 wt%) accumulated on the H-Ni/γ-Al2O3 after 20 hr DRM test at 650°C, while the much lower coke (4 wt%) happened to M-Ni/γ-Al2O3 (Figure 2A). To eliminate the influence of active sites numbers on the amount of carbon deposit, the weight loss value was normalized to the active Ni, and the result is given in Figure 2B. The average coke amount of M-Ni/γ-Al2O3 was much lower than that of H-Ni/γ-Al2O3, which strongly verified that the bounded Ni had more excellent anti-carbon property than the free state Ni. SEM and TEM images of the used catalysts are given in Figure S19 to show morphology of catalysts after reaction. As described in the TGA results, the amount of carbon deposition observed in the used L-Ni/γ-Al2O3 and M-Ni/γ-Al2O3 samples was very small, while a large amount of coke species was observed in the used H-Ni/γ-Al2O3. With further observation, we found that these carbon deposition species should be mainly filamentous carbon and carbon nanotubes. On the other hand, XRD patterns of the used catalysts are listed in Figure S20 to study the reasons for catalysts' deactivation. It was easy to observe that a strong carbon species peak at 26° appeared in the XRD pattern of the spent H-Ni/γ-Al2O3. For used L-Ni/γ-Al2O3 and M-Ni/γ-Al2O3, the intensity of C species was very low, indicating that the amount of carbon deposition was small. At the same time, we found that after DRM reaction, Al2O3 was mainly in the γ phase, but α-Al2O3 (PDF#71–1123) was also observed at the diffraction angles of 35.04, 43.4, and 57.4, which correspond to (104), (113), and (116) crystal plane of α-Al2O3, respectively. This phenomenon showed that after a long-time test, the carrier Al2O3 underwent a partial transformation from γ phase to α phase. Nevertheless, the intrinsic activity, coking resistance, and stability of M-Ni/γ-Al2O3 were all better than those of H-Ni/γ-Al2O3, which might be related to the rational MSI of the bounded Ni in M-Ni/γ-Al2O3.

Temperature-programmed surface reaction mass spectra test

Summarizing from the previous characterization, the bounded and the free state Ni were proved to have different activity and anti-coking abilities for DRM. To explore the mechanism of these differences, a series of temperature-programmed surface reaction mass spectra (TPSR-MS) experiments were subsequently performed.

The results of CH4 TPSR-MS are given in Figure 3A–3C. During the TPSR process, H2 generation was always accompanied by CO2 production which suggested CH4 decomposed into H2 and CHx on the surface of Ni, while the latter would be oxidized on the surface of the catalysts to form CO and H2. It should be noted that there was a signal mutation when H2 starts to appear. This might be a consequence of a reaction between CH4 and highly active O∗ species produced in the reduction process on the surface. The slope of the hydrogen generation signal decreased with the increase of Ni loading, which suggested that the bounded Ni presented better CH4 activation ability. The signal of H2 and CO2 quickly decreased for all samples, possibly because of the limited amount of active O∗ on the surface of these catalysts. Subsequently, H2 was secondarily generated deriving from the direct thermal cracking of CH4 in the high temperature.

Figure 3

TPSR profiles of the catalysts

(A–C) (A) CH4-TPSR MS profiles of L-Ni/γ-Al2O3, (B) M-Ni/γ-Al2O3, and (C) H-Ni/γ-Al2O3.

(D–F) (D) CH4-CO2 TPSR MS profiles of L-Ni/γ-Al2O3, (E) M-Ni/γ-Al2O3, and (F) H-Ni/γ-Al2O3.

See also Figure S21.

CH4-CO2 TPSR-MS experiment was performed to estimate the activation of CO2, and the results are given in Figures 3D–3F. For L-Ni/γ-Al2O3 sample, peaks of CO appeared at 430°C. A similar peak was also noticed in M-Ni/γ-Al2O3 sample where a new low-temperature peak emerged at around 361°C, which was due to more the bounded Ni of M-Ni/γ-Al2O3. In Figure 3F, nearly no signal of CO was detected in H-Ni/γ-Al2O3 between 200°C and 500°C, and the continuing release of CO happened above 600°C which suggested the weaker CO2 activation ability and stronger interaction between CO and H-Ni/γ-Al2O3. This conclusion could be further proved by the results of CO-Temperature Programmed Desorption (CO-TPD) (Figure S21). The desorption temperature of CO from the surface of M-Ni/γ-Al2O3 was about 368°C, while for H-Ni/γ-Al2O3 it was 449°C. It could be inferred that the adsorption strength of CO on the free state Ni was stronger than that of the bounded Ni. It was also interesting to notice that H2 production temperature was brought forward about 70°C compared to the counterparts in CH4-TPSR, meaning that CO2 activation could further favor H2 production from CH4 decomposition. Based on the results of TPSR-MS, the bounded Ni showed not only better capability toward CH4 cracking than the free state Ni but also better activation ability for CO2.

Operando Raman and in situ DRIFTS

Operando Raman tests were performed at 650°C to explain the differences in carbon deposition resistance of the M-Ni/γ-Al2O3 and H-Ni/γ-Al2O3 (Figure 4A). The DRM reaction was firstly carried out for around 1 hr until peak intensity of both D band (∼1400 cm−1) and G band (∼1580 cm−1) did not change significantly (Zhang et al., 2013); then, CH4 was solely cut off to evaluate the effect of CO2 on carbon species. For M-Ni/γ-Al2O3, carbon species accumulated gradually during the 1 hr DRM test. After cutting the CH4 gas flow, the peak intensity at D and G band reduced immediately and eventually was steady for 15 min. On the contrary, the intensity of the G band peak of the H-Ni/γ-Al2O3 grew rapidly and remained stable for 10 min during the DRM test. After cutting CH4, the intensity of the G band decreased at the 10th minute, but quickly rebounded at the 15th min, then keeping constant finally. The special rebound might be attributed to CO disproportionation to form carbon species since carbon deposition was more likely to accumulate through CO disproportionation when MSI of Ni particles was not strong enough (Foppa ).

Figure 4

Reaction mechanism analysis

(A) Operando Raman spectra of the fresh samples (blue: M-Ni/γ-Al2O3, red: H-Ni/γ-Al2O3), test condition: 650°C, gas flow rate of CH4 and CO2: 15 mL/min, purge gas: 50 mL/min N2.

(B) CO2 signal in CO2-TPD MS profiles of the catalysts.

(C) In situ DRIFTS of M-Ni/γ-Al2O3 at 650°C.

(D) In situ DRIFTS of H-Ni/γ-Al2O3 at 650°C.

See also Figures S22 and S23.

CO2-Temperature Programmed Desorption Mass Spectrometer (CO2-TPD MS) was carried out to explore the differences of CO2 adsorption over the catalysts (Figure 4B). All samples got a sharp peak at about 90°C, which was attributed to desorption of weak-bonded bicarbonate (Keturakis et al., 2014). Smooth peaks at 370°C and 550°C, which originated from desorption of monodentate carbonate (weak basic site) and polydentate carbonate (strong basic site), respectively (Ewald and Hinrichsen, 2019; Grünbacher et al., 2019; Zheng et al., 2019), appeared in L-Ni/γ-Al2O3. The intensity of these two peaks intensified with Ni loading and a new desorption peak at 500°C appeared, which was possibly derived from a kind of bidentate carbonate (Busca and Lorenzelli, 1982). More bidentate carbonate desorbed at around 430°C (moderate basic site) over H-Ni/γ-Al2O3 (Ewald and Hinrichsen, 2019). Carbonates were generally thought to be formed through two ways: CO2 might combine with the pre-adsorbed oxygen on the metal surface to form carbonate or it could directly form carbonate on the support near the metal-support interface (Bitter et al., 1998; Solymosi, 1991). Combined with the results above, it was possible that on the bounded Ni with stronger MSI, the O atom of the support might participate in the formation of polydentate carbonate. The CO signal (m/z = 28) in CO2-TPD MS of the fresh catalysts is also given in Figure S22 to imply the capability for CO2 activation. H-Ni/γ-Al2O3 showed a distinct peak at 425°C, while M-Ni/γ-Al2O3 exhibited a lower temperature peak at 340°C beside the peak at 425°C. This suggested that M-Ni/γ-Al2O3 had better CO2 activation ability to produce CO and active O∗ than H-Ni/γ-Al2O3, and it could be inferred that CO2 was easier to be activated on the bounded Ni than the free state Ni.

Furthermore, CO2-TPSR DRIFTS at different temperature was explored to assess CO2 activation ability and CO adsorption ability at the same time, and relevant results are shown in Figure S23. It was easy to find that CO species at ∼2100 cm−1 began to appear at 400–450°C for H-Ni/γ-Al2O3 (Tang et al., 2017), which was in good agreement with the results of CO2-TPD MS, and CO peaks intensity increased monotonically with the temperature. However, almost no peak of CO was found in M-Ni/γ-Al2O3. This could be weaker CO adsorption ability of the bounded Ni in M-Ni/γ-Al2O3, thanks to rational MSI compared with the free state Ni in H-Ni/γ-Al2O3. As reported in relevant articles, strong MSI would weaken the adsorption strength of small molecules such as CO and H2 on metal particles (Han et al., 2020), and the free energy of C-C coupling would increase significantly due to enhanced MSI (Zhang et al., 2021b). Hence, the CO disproportionation reaction (Boudouard reaction) would be inhibited due to these above reasons on the bounded Ni, thereby reducing the accumulation of carbon depositions.

In situ DRIFTS was performed to explore the differences of the DRM reaction mechanism between the bounded Ni and the free state Ni, and the results are given in Figures 4C and 4D. As shown in Figure 4C for M-Ni/γ-Al2O3, the peaks of polydentate carbonate (a broad peak around 1400–1500 cm−1), bidentate carbonate (1558 cm−1), and bidentate formate (1542 cm−1) were found during the reaction, which meant that these species were the reactive intermediates (Binet et al., 1999; Takano et al., 2016). Different from the results of the M-Ni/γ-Al2O3, the peak of polydentate carbonate species was not shown over H-Ni/γ-Al2O3 (Figure 4D). Instead, peaks assigned to monodentate carbonate at 1506 cm−1 appeared during the reaction (Binet ), which was not observed in M-Ni/γ-Al2O3. At the same time, the peaks of bidentate formate at 1542 cm−1 and bidentate carbonate at 1558 cm−1 were also observed. However, it was interesting to find that the peak intensity of monodentate carbonate and bidentate carbonate strengthened and weakened alternately. This phenomenon showed that the monodentate and bidentate carbonates would transform into each other during the reaction. According to previous literature, monodentate carbonates did tend to be converted into bidentate carbonates (Galenda et al., 2010). Combined with the conclusions of the catalysts' TOF value, it could be proposed that the interconversion would delay the conversion from carbonate to formate, thus hindering the conversion of CO2.

Reaction mechanism studies

The mechanism schematic is given in Figure 5 summarized from the previous analysis. For M-Ni/γ-Al2O3 sample, where only the bounded Ni existed, CH4 was activated more effectively to produce CHx and H2. The activation of CO2 on M-Ni/γ-Al2O3 was divided into two ways: on the one hand, CO2 could directly decompose into active O∗ and CO, the active O∗ then reacted with CHx species (x = 0, 1, 2, 3) to generate CHxO, and finally CHxO would degrade directly or indirectly to produce H2 and CO (Zhu et al., 2009). On the other hand, CO2 was mainly adsorbed and activated on the catalysts in form of the polydentate/bidentate carbonate, as observed in CO2-TPD. It could be inferred that the three ligands of the polydentate carbonate might involve Al and Ni atoms simultaneously considering the strong interaction between Ni and γ-Al2O3 at the metal-support interface. And the polydentate carbonate was believed to guarantee both the high activity and good coke resistance of the catalysts for DRM by forming Ni-Al surface as reported in a recent literature (Chen et al., 2020). The polydentate/bidentate carbonate could later quickly react with H∗ species to form bidentate formate, which would subsequently decompose into CO and H2. This efficient conversion would further accelerate CH4 decomposition on the bounded Ni, proved by the results of CH4-CO2-TPSR-MS. The desorption of the product (H2 and CO) and C-O coupling on the bounded Ni was preferred due to rational MSI (Han ; Zhang et al., 2021b), so the side reactions (Reverse water gas shift (RWGS) and Boudouard reaction) were suppressed, thus reducing the coke deposition. As for the H-Ni/γ-Al2O3, where the bounded Ni and the free state Ni coexisted, CO2 might preferentially adsorb and be activated in the form of the bidentate and monodentate carbonate. The bidentate formate was then formed via a reaction between H∗ and the carbonate intermediates (Bobadilla et al., 2017). However, this process was blocked to a certain extent since interconversion between the carbonates at the same time. This finding could explain why the TOF of CO2 of M-Ni/γ-Al2O3 was higher than that of the H-Ni/γ-Al2O3. Furthermore, the free state Ni with weaker MSI got stronger CO adsorption ability than that of the bounded Ni. The stronger CO adsorption ability meant more CO accumulation on the surface of catalysts during the reaction, which would exacerbate CO disproportionation and thus causing more serious coking. In summary, a rational MSI with γ-Al2O3 made the bounded Ni more capable of activating CO2, which in turn promoted the conversion of CH4 and the elimination of carbon depositions, and also facilitated the desorption of CO, thereby addressing the trade-off issue of catalytic activity and carbon resistance.

Figure 5

Proposed DRM mechanism diagram on the bounded Ni and the free state Ni

Discussion

In this study, we made an insight into the differences between the bounded Ni and the free state Ni in Ni/γ-Al2O3 for DRM reaction in terms of carbon resistance and reaction mechanism. The bounded nickel, which held rational MSI, showed better coking resistance, higher TOFCO2 and TOFCH4 than the free state nickel due to the better CH4 cracking ability, and CO2 adsorption and activation. CO2 adsorbed on the bounded Ni mainly in the form of the polydentate/bidentate carbonate and converted into bidentate formate efficiently, which finally decomposed into CO. By contrast, the interconversion between monodentate carbonate and bidentate carbonate was detected on the free state Ni, which would slow down the conversion efficiency of CO2 and further reduced the TOF of CH4. Moreover, the CO generated by the reaction was easier to desorb from the bounded Ni compared with the free state Ni due to rational MSI, which would further improve coking resistance by inhibiting the CO disproportionation. We believed that our observations would strengthen the understanding of structure-performance relationship in catalysis and could enable the design of more efficient catalysts for DRM in the future.

Limitations of the study

In this study, we explored the differences in the DRM performance and reaction mechanism of two different active Ni species (the bounded Ni and the free state Ni) over Ni/γ-Al2O3, which would be useful for the future design of high-performance DRM catalysts. However, the work also has some shortcomings. For example, it might be difficult to control the size of the Ni particles obtained by the exsolution to reach the cluster level according to relevant researches, so studying the DRM properties of smaller Ni species (such as Ni1, Ni2, and Ni4) would be more interesting, although it also seems more difficult but is worth exploring.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Chemicals, peptides, and recombinant proteins
γ-Al2O3	Titan Scientific Co., Ltd	CAS: 1344-28-1
Nickel(II) nitrate hexahydrateNi(NO3)2·6H2O	Sinopharm Chemical Reagent Co., Ltd.	CAS: 13478-00-7