Ni-Sn-Supported ZrO2 Catalysts Modified by Indium for Selective CO2 Hydrogenation to Methanol.

Ni and NiSn supported on zirconia (ZrO2) and on indium (In)-incorporated zirconia (InZrO2) catalysts were prepared by a wet chemical reduction route and tested for hydrogenation of CO2 to methanol in a fixed-bed isothermal flow reactor at 250 °C. The mono-metallic Ni (5%Ni/ZrO2) catalysts showed a very high selectivity for methane (99%) during CO2 hydrogenation. Introduction of Sn to this material with the following formulation 5Ni5Sn/ZrO2 (5% Ni-5% Sn/ZrO2) showed the rate of methanol formation to be 0.0417 μmol/(gcat·s) with 54% selectivity. Furthermore, the combination NiSn supported on InZrO2 (5Ni5Sn/10InZrO2) exhibited a rate of methanol formation 10 times higher than that on 5Ni/ZrO2 (0.1043 μmol/(gcat·s)) with 99% selectivity for methanol. All of these catalysts were characterized by X-ray diffraction, high-resolution transmission electron microscopy (HRTEM), scanning transmission electron microscopy (STEM), X-ray photoelectron spectroscopy, CO2-temperature-programmed desorption, and density functional theory (DFT) studies. Addition of Sn to Ni catalysts resulted in the formation of a NiSn alloy. The NiSn alloy particle size was kept in the range of 10-15 nm, which was evidenced by HRTEM study. DFT analysis was carried out to identify the surface composition as well as the structural location of each element on the surface in three compositions investigated, namely, Ni28Sn27, Ni18Sn37, and Ni37Sn18 bimetallic nanoclusters, and results were in agreement with the STEM and electron energy-loss spectroscopy results. Also, the introduction of "Sn" and "In" helped improve the reducibility of Ni oxide and the basic strength of catalysts. Considerable details of the catalytic and structural properties of the Ni, NiSn, and NiSnIn catalyst systems were elucidated. These observations were decisive for achieving a highly efficient formation rate of methanol via CO2 by the H2 reduction process with high methanol selectivity.

Introduction

The increasing demand for sustainable and environmentally benign alternative energy resources can effectively replace the current natural fossil fuel hydrocarbon sources in the near future.[1] Carbon Capture technologies can be considered as an efficient source of energy because it is easy to store, harvest, and transport. According to several researchers, the catalytic conversion of abundant, inexpensive, renewable, and nontoxic carbon dioxide (CO2) into valuable fuels and chemicals is one of the most practical routes for reducing CO2 emission, and from an economic point of view also, it becomes an attractive C1 feedstock.[2−4] Hence, the production of methanol via CO2 hydrogenation could substitute global dependence on fossil fuels.[5] Methanol has wide range of applications in the chemical industry as a starting component for the production of various upstream and downstream value-added products (e.g., methanol to olefins process). Also, methanol itself can be used as an alternative transportation fuel for both gasoline- and diesel-powered engines because of its efficient combustion, ease of transportation, and wide availability around the globe.[6−8] Industrial production of methanol is usually made from syngas over tricomponent Cu/ZnO/Al2O3 catalysts under mild temperatures (200–300 °C) and high pressures (5–10 MPa).[9,10]

Furthermore, various promoters (Ga, Mg, In, and La) in combination with various metals (such as Cu, Pt, and Pd) have been used for CO2 hydrogenation to methanol, and they showed efficient activity and stability.[11,12] In comparison with that of noble metal catalysts, the use of Ni with promoters dispersed over the support for the hydrogenation reaction is one of the most efficient ways because of its wide availability and low cost.[13] However, a supported monometallic Ni-based catalyst shows predominant selectivity to methane via CO2 hydrogenation[14] and also demonstrated poor stability during the time-on-stream because it has a high tendency to form coke and to sinter.[14]

Apart from traditional catalysts, scientists have been trying to develop catalysts different from the classical copper-based catalysts for the production of methanol via CO2 hydrogenation. The nickel–gallium intermetallic catalyst was shown to exhibit improved activity for CO2 hydrogenation to methanol under mild conditions.[15] Similarly, mesoporous spinel cobalt oxide supported on manganese oxide was found to form methanol in high yield via CO2 hydrogenation but with concomitant production of hydrocarbons.[16] Recently, a trimetallic combination Ni–In–Al supported on phyllosilicate showed efficient activity for CO2 hydrogenation to methanol at low pressure; however, the selectivity for CO is higher than that for methanol.[17] Still, there is an important aspect of studying CO2 hydrogenation in favor of methanol production at atmospheric pressure by designing catalysts that effectively suppress the reverse water gas shift reaction (RWGS) along with the formation of other by-products.

Supported Ni catalysts have been used in combination with Sn as a promoter for various reactions such as ethylene adsorption and ethane and propane dehydrogenation, they could be considered as efficient bimetallic catalysts for various important transformations.[18−23] Dumesic[24] reported that the stability of Ni was improved by involving Sn as a promoter and also that the combination of Ni and Sn catalysts exhibited a remarkable increase in the selectivity toward the formation of hydrogen from biomass-derived hydrocarbons as compared to that of monometallic Ni, which predominantly showed enhanced selectivity toward the production of alkanes but further led to the formation of coke.[24] Although copper is usually claimed to be the main component of the active site for the synthesis of methanol from CO2 hydrogenation, indium, as In2O3, has recently been shown to play an efficient role for activation of CO2. This promoting effect of In2O3 has subsequently been proven by several researchers with the help of density functional theory (DFT).[25−27]

The tentative intuitive explanation behind the above-mentioned observations stems from the fact that hydrogenation of CO2 over the Ni catalyst is mainly responsible for methane production, whereas CO2 hydrogenation catalyzed by NiSn results in methanol production by suppressing methane formation. Moreover, the use of indium as a promoter would improve CO2 hydrogenation activity with effective formation rate of methanol.

With these intuitive hypotheses, we decided to prepare bimetallic catalysts based on Ni and Sn and also trimetallic catalysts Ni/Sn/In. The choice of ZrO2 as a support was based on its high thermal stability and unique acidic and basic properties.[28−30] Also, the strong anchoring effects and partial activation of CO2 by Zr4+ may efficiently enhance CO2 reduction.[31]

We disclose here that our new catalyst combination associating Ni, Sn, and In over ZrO2, e.g., 5Ni5Sn/10InZrO2, employed in the present investigation exhibits high rate of formation of methanol (0.1043 μmol/(gcat·s)) with very high selectivity for methanol and shows stable performance with time-on-stream with a stability of up to 50 h.

Results and Discussion

Characterization

The textural properties of Ni and NiSn supported on ZrO2 and on InZrO2 catalysts are summarized in Table T1, Supporting Information. The surface area of catalyst samples is constant, which is in the range of 98–99 m2/g. As can be seen from literature study, there was no change observed in the Brunauer–Emmett–Teller (BET) surface area with the addition “Sn” to Ni-supported catalysts.[24] Also, there is no certain change in other textual properties such as pore volume (cm3/g) and pore diameter (Å). The X-ray fluorescence (XRF) results given in Table T2 (Supporting Information) show that the values of real contents of the samples are similar to the nominal values discussed.

XRD patterns for Ni and NiSn supported on ZrO2 and InZrO2 are shown in Figure ; the presence of peaks at 2θ = 30.55, 35.40, 50.70, and 60.35° can be attributed to the (111), (200), (220), and (311) diffraction planes of the tetragonal zirconia. This implies that no obvious phases of Ni and NiSn were observed in the diffraction patterns, which could be due to uniform distribution of metals and metal oxides within the matrix of support and/or low loading of both metals, thereby forming a homogeneous phase of the composite catalyst.[32] The diffraction patterns of the spent 5Ni/ZrO2 and 5Ni/10InZrO2 catalyst samples are shown in Figure , which exhibited a peak at 2θ = 43.5° corresponding to the (111) metallic phase of Ni along with the tetragonal phase of ZrO2. This clear diffraction peak corresponds to the metallic Ni, suggesting the transformation of nickel oxide to metallic Ni. However, the addition of Sn to 5Ni/ZrO2 and 5Ni/10InZrO2 showed new peaks; these diffraction peaks correspond to the NiSn alloy, without any other index peak for metallic Ni.[33] These new phases of NiSn alloy formation quite resemble with those of the Ni3Sn2 alloy, but the presence of NiSn and Ni3Sn4 alloys cannot be ruled out.[34] This study indicates that the introduction of Sn would help inhibit the formation of metallic Ni phases, which is mainly responsible for the formation of coke and methane.

Figure 1

XRD patterns of as-prepared catalyst samples.

Figure 2

XRD patterns of spent catalyst samples (reaction conditions: T = 250 °C, P = 25 bar, gas flow (CO2/Ar/H2) = 5/5/15 mL/min, weight hourly space velocity (WHSV) 30 000 h–1, CO2/H2 = 1:3, time = 24 h).

The 5Ni/ZrO2 catalyst morphology in Figure S1a shows no distinguishable particles of Ni and Ni oxide with a very poor dispersion in the matrix of zirconia. Then, involvement of Sn in Ni-supported catalysts showed that NiSn particles are embedded in the layered structure with a chain bead morphology in the matrix of the ZrO2 support (Figure S1b). Furthermore, this is more evidenced by NiSn supported on InZrO2; this morphology possesses similar features as discussed above and also kept particle size in the range of 10–20 nm with a very high dispersion of the NiSn alloy (Figure S1c,d).

To get more characterization details, these catalyst samples were analyzed using the scanning transmission electron microscopy (STEM) technique. It can be seen that the monometallic Ni supported on ZrO2 showed similar features as discussed above under high magnification and that there is likely agglomeration of Ni particles located on the zirconia support (Figure a,b). However, both 5Ni5Sn/ZrO2 and 5Ni5Sn/10InZrO2 catalysts consisting of Sn species were mixed uniformly and coexisted closely with Ni species, which is further confirmed by the formation of NiSn alloy. These results are in good agreement with the above XRD studies (Figure c–f). Furthermore, elemental mapping along with energy-dispersive system spectrometry (EDS) was also conducted to get a clear vision of the elemental distribution of Ni and NiSn supported on ZrO2 and InZrO2. As can be seen from this study, there is distinct layer formation of Ni, presented in green, which is associated with the ZrO2 support (Figure a,b). On the other hand, the Ni associated with Sn reveals the formation of NiSn alloy (Figure c,d), which is presented in green and blue, anchored to the zirconia support. We consider that both particles of Ni and Sn are most likely of the NiSn alloy located on the surface of zirconia, whereas metallic Ni constitutes the core of the nanocluster and Sn is located as a shell with wide distribution to form a core–shell structure, which resembles with high-magnification and elemental mapping studies (Figure S2). Moreover, the NiSnIn system is shown in tricolors, as green, blue, and magenta (Figure e,f), and this elemental mapping study indicates that the formation of the NiSn alloy is higher than that of both supported catalysts. This may arise due to homogeneous NiSn alloy particles. Aforementioned observations suggest that the addition of Sn to Ni-supported catalysts will help prevent aggregation of metallic Ni species and favor the formation of NiSn alloy and thus avoid undesired hydrogenation reactions, which is clearer with 5Ni5Sn/10InZrO2 catalysts showing efficient formation rate of methanol with 99% selectivity.

Figure 3

STEM images with EDX and selected-area electron diffraction (SAED) patterns of as-prepared catalysts: (a, b) 5Ni/ZrO2, (c, d) 5Ni5Sn/ZrO2, and (e, f) 5Ni5Sn/10InZrO2.

Figure 4

Dark-field STEM images of (EDS) (a) 5Ni/ZrO2, (c) 5Ni5Sn/ZrO2, and (e) 5Ni5Sn/10InZrO2 as-prepared samples. Electron energy-loss spectroscopy (EELS) mapping of the selected regions showing elemental distribution of “Zr” in red, “Ni” in green, Sn in blue, and “In” in magenta: (b) 5Ni-ZrO2, (d) 5Ni5Sn/ZrO2, and (f) 5Ni5Sn/10InZrO2.

To determine the oxidation state of nickel and tin, high-resolution X-ray photoelectron spectroscopy (XPS) spectra of Ni 2p, Sn 3d, and In 3d core levels were obtained for the 5Ni5Sn/10InZrO2 catalyst before and after the reaction, as indicated in Figures –7. The Ni 2p core-level spectrum (Figure a) shows two intense peaks at around ∼855.6 and 871.1 eV binding energies, attributed to Ni 2p3/2 and Ni 2p1/2, respectively, accompanied by those of their corresponding satellites at around ∼861.2 and 879.6 eV.[35−37] These peaks are attributed to Ni in the form of oxide and hydroxide phases.[38] Additionally, a peak located at 852.1 eV in Figure b indicates that the metallic nickel (Ni0) phase is present on the catalyst surface. The Sn 3d core-level spectra (Figure a) show two intense peaks at around ∼486.2 and 494.6 eV binding energies, attributed to Sn 3d5/2 and Sn 3d3/2, respectively. The Sn 3d5/2 peak was fitted using three components located at 484.2, 486.2, and 487.1 eV, assigned to Sn0 and Sn2+ in SnO and Sn4+ in SnO2, respectively (Figure b).[39−41] Moreover, the XPS results show that after the reaction the concentration of both metallic nickel and tin increases, indicated by increasing both Ni and Sn metallic peak intensities. The In 3d core-level spectra (Figure a,b) show two intense peaks at around ∼444.09 and 451.67 eV binding energies corresponding to the In 3d5/2 and In 3d3/2 states, respectively. These obtained binding energy values are associated with the In+3 oxidation state in the form of In oxide (In2O3).

Figure 5

Ni 2p XPS spectra of 5Ni5Sn/10InZrO2 catalysts: (a) as-prepared sample, (b) spent sample (reaction conditions: T = 250 °C, P = 25 bar, gas flow (CO2/Ar/H2) = 5/5/15 mL/min, WHSV 30 000 h–1, CO2/H2 = 1:3, time = 24 h).

Figure 7

In 3d5/2 spectra of 5Ni5Sn/10InZrO2 catalysts: (a) sample pretreated at 120 °C, (b) spent sample (reaction conditions: T = 250 °C, P = 25 bar, gas flow (CO2/Ar/H2) = 5/5/15 mL/min, WHSV 30 000 h–1, CO2/H2 = 1:3, time = 24 h).

Figure 6

Sn 3d5/2 spectra of 5Ni5Sn/10InZrO2 catalysts: (a) sample pretreated at 120 °C, (b) spent sample (reaction conditions: T = 250 °C, P = 25 bar, gas flow (CO2/Ar/H2) = 5/5/15 mL/min, WHSV 30 000 h–1, CO2/H2 = 1:3, time = 24 h).

This reduction did not occur for monometallic 5Ni/ZrO2 because we did not observe any peak corresponding to the metallic Ni (Figure S3a). This observation suggests that the combination of both metallic Ni and Sn (Figure S3b,c) may consist of different forms of NiSn alloys, which is in good accordance with XRD and TEM results.

We investigate the reducibility of supported Ni catalysts with Sn and In species (Figure ). The monometallic 5Ni/ZrO2 catalyst temperature-programmed reduction (TPR) profile exhibits sharp shoulder peaks with a tail in the temperature range from 325 to 600 °C, and these observations suggest that the reduction of nickel oxide to metallic Ni requires high reduction temperature.[42−44] Furthermore, for Ni supported on the InZrO2 catalyst, the TPR profile shows a very broad peak in the range of 300–700 °C, and the nature of peak confirms that there are several types of Ni and In oxide species with different ease-of-reducibility patterns. However, both NiSn supported on ZrO2 and InZrO2 catalysts possess two reduction peaks at 570 and 798 °C, which are indexed to the Ni and Sn oxide reduction profiles. The incorporation of Sn to supported Ni catalysts helps ease NiO reducibility, as evidenced by the shift of the peak maxima to lower reduction temperatures at 350 and 357 °C, respectively. Also, use of In further results in an increase in Ni and Sn oxide reducibility, as demonstrated by a shift of peaks to a lower temperature and a decrease in intensity. Moreover, the addition of Sn favors NiSn alloy formation, which is further confirmed by the low reduction temperature required in the presence of NiO because of the small crystallite size, which is closely related to XRD and high-resolution transmission electron microscopy (HRTEM) results discussed above.

Figure 8

H2-TPR profiles of as-prepared supported Ni catalysts.

CO2 temperature-programmed desorption (TPD) study (Figure ) shows that the monometallic 5Ni/ZrO2 catalyst possesses three distinct CO2 desorption peaks mainly in the low (200–300 °C)- and medium (300–500 °C)-temperature regions, and this confirmed that CO2 weakly adsorbed on the catalyst surface. As reported in the literature, the weak basic sites may ascribe to the linear adsorption of CO2, whereas the strong basic sites may result from the bridge adsorption of CO2.[45,46] In contrast, the addition of Sn to the 5Ni/ZrO2 catalyst resulted in two CO2 desorption peaks in the regions of 250–500 °C, medium strong site region, and 450–650 °C, strong site region, which indicate the formation of new CO2 absorption sites on the catalyst surface. Indeed, the involvement of Sn helps modify basic sites of 5Ni/ZrO2 catalysts, but the use of In acts as a promoting agent to improve the basic strength by CO2 absorption sites in both medium and strong regions. This study implies that the incorporation of Sn into the supported 5Ni/ZrO2 catalyst may create new basic sites for CO2 adsorption.

Figure 9

CO2-TPD study of as-prepared supported Ni catalysts.

Catalytic Activity

The results of the catalytic hydrogenation of CO2 using supported nickel catalysts at 250 °C, 25 bar, and a weight hourly space velocity (WHSV) of 30 000 h–1 are summarized in Figures , 11, and 12. The major products were methanol (MeOH), methane (CH4), and carbon monoxide (CO). All catalytic data were taken when the catalyst reached a steady state. The effect of Sn with 5Ni/ZrO2 and 5Ni/10InZrO2 on the catalytic performance was assessed in terms of the formation rate of methanol per gram of catalysts. Initially, Sn (5Sn/10InZrO2)- and In (10InZrO2)-supported zirconia catalysts showed a very low formation rate of methanol via CO2 hydrogenation. On other hand, the use of Ni without Sn as a catalyst mainly favored the formation of methane, whereas all other catalyst combinations showed selectivity toward methanol formation (Figure S4).

Figure 10

Formation rate of methanol (reaction conditions: T = 250 °C, P = 25 bar, gas flow (CO2/Ar/H2) = 5/5/15 mL/min, WHSV 30 000 h–1, CO2/H2 = 1:3, time = 24 h).

Figure 11

Formation rate of methanol (reaction conditions: T = 250 °C, P = 25 bar, gas flow (CO2/Ar/H2) = 5/5/15 mL/min, WHSV 30 000 h–1, CO2/H2 = 1:3, time = 24 h; catalysts: 1 = 5Ni/ZrO2, 2 = 5Ni/10InZrO2, 3 = 5Ni5Sn/ZrO2, 4 = 5Ni5Sn/10InZrO2, 5 = 10Ni5Sn/10InZrO2, 6 = 5Ni10Sn/10InZrO2).

Figure 12

Selectivity patterns for catalysts 1 = 5Ni/ZrO2, 2 = 5Ni/10InZrO2, 3 = 5Ni5Sn/ZrO2, 4 = 5Ni5Sn/10InZrO2, 5 = 10Ni5Sn/10InZrO2, 6 = 5Ni10Sn/10InZrO2 (reaction conditions: T = 250 °C, P = 25 bar, gas flow (CO2/Ar/H2) = 5/5/15 mL/min, WHSV 30 000 h–1, CO2/H2 = 1:3, time = 24 h).

Also, monometallic nickel catalyst 5Ni/ZrO2 showed nil activity toward the formation rate of methanol; this is proven by complete selectivity for the production of methane (99%). The addition of In to the 5Ni/ZrO2 catalyst resulted in a similar observation; there is no observable methanol formation, which is also due to the high selectivity to methane (99%). As a consequence, monometallic Ni and NiIn species supported on ZrO2 are poor for methanol formation. Interestingly, Sn incorporation in 5Ni/ZrO2 initiates the formation of methanol, which is about 0.0417 (μmol/(gcat·s)) with an increase in selectivity to methanol about 54%. Moreover, the combination of Ni and Sn supported on InZrO2 predominantly increases the formation rate of methanol (1.043 μmol/(gcat·s)) along with an increase in selectivity to methanol (99%) and a concomitant decrease in selectivity to methane. In order to improve formation rate of methanol, we varied the Ni/Sn ratio, and with an increase in the concentration of Ni (10Ni5Sn/10InZrO2), a marginal decrease in the methanol formation rate (0.0798 μmol/(gcat·s)) was observed, which could be due to the increase in selectivity for methane. However, the increase in the concentration of Sn (5Ni10Sn/10InZrO2) did not affect the formation rate of methanol (0.093 μmol/(gcat·s)), but it is more favorable for CO selectivity (22%). Ideally, the use of Sn as a co-metal would help alter the selectivity toward methanol than toward methane and also the NiSnIn catalyst combination helps improve the catalytic activity toward methanol selectivity as well as the production rate via CO2 hydrogenation. Recently, Fan et al. revealed that the trimetallic Ni–In–Al combination shows an effective formation rate of methanol but suffers from poor selectivity to methanol, which is in the range of only 1–2%.[17] However, our catalyst combination possesses not only high selectivity to methanol but also comparable formation rate of methanol. The excellent catalytic performance of our NiSn catalyst supported on InZrO2 can be explained as follows: (i) Sn incorporation into Ni helps in the formation of the NiSn alloy, (ii) advantage of Sn for ease of reducibility of Ni while using predominant oxide supports such as In oxide and ZrO2, (iii) Sn also plays an important role in increasing the basic strength of catalysts toward facile CO2 hydrogenation activity, and (iv) the tricomponent (NiSnIn) showed consistent activity for CO2 hydrogenation to methanol during the time-on-stream study.

The effect of In loading (1, 5, and 10%), by keeping the NiSn (5Ni5Sn) concentration constant, on the formation rate of methanol was studied (Figure ). It was observed that the formation rate of methanol increases from 0.043 to 0.1043 μmol/(gcat·s) with the increasing In loading from 1 to 10% at a constant temperature of 250 °C. The highest methanol formation rate was achieved by 5Ni5Sn/10InZrO2, which is approximately 10 times higher than the formation rate of monometallic 5Ni/ZrO2 catalyst, as against there is no such noticeable effect of In loadings on the selectivity of methanol up to 10%. This clearly confirms that In acts as a promoter in the NiSn combination for selective CO2 conversion with efficient formation rate of methanol.

Figure 13

Effect of In loading on the formation rate of methanol (reaction conditions: T = 225–275 °C, P = 25 bar, gas flow (CO2/Ar/H2) = 5/5/15 mL/min, WHSV 30 000 h–1, CO2/H2 = 1:3, time = 24 h).

The effect of reaction pressure on CO2 hydrogenation was studied for 1, 10, and 25 bar over two different catalysts (5Ni5Sn/ZrO2 and 5Ni5Sn/10InZrO2), by keeping all other experimental parameters same, is shown in Figure . As stated in the literature, high pressure is necessary for selective hydrogenation of CO2 to methanol, which is controlled by the kinetics and thermodynamics of CO2 hydrogenation.[47−50] It can be seen that the results of both catalysts resulted in a linear increase in the methanol formation rate with an increase in the reaction pressure. It can be found that the formation rate of methanol for 5Ni5Sn/10InZrO2 catalysts is in the range 0.039–0.1043 μmol/(gcat·s) from 1 to 25 bar total pressure. The measurable difference observed with the involvement of In as a promoter to NiSn catalysts is that it results in a 2-fold high methanol formation rate as compared to that with NiSn-ZrO2 catalysts within the pressure range studied. This is mainly because NiSn supported on InZrO2 results in more NiSn alloy formation along with high basic strength of catalysts, which is confirmed by high selectivity to methanol discussed earlier in Figure .

Figure 14

Effect of reaction pressure on the methanol formation rate (reaction conditions: T = 250 °C, P = 5–15–25 bar, gas flow (CO2/Ar/H2) = 5/5/15 mL/min, WHSV 30 000 h–1, CO2/H2 = 1:3, time = 24 h).

The hydrogenation of CO2 over 5Ni5Sn/10InZrO2 was studied by variation of total flow rate (space velocity) with different reaction temperatures from 225 to 275 °C and keeping all other parameters constant (Figure ). As expected, the formation rate of methanol is linearly increased with the increasing reaction temperature from 225 to 275 °C and flow rate from 12.5 to 37.5 mL/min. Also, at high reaction temperature, 275 °C, and total flow rate, 37.5 mL/min, the formation rate of methanol is very high (0.176 μmol/(gcat·s)). Because this goes through an increase in the total flow rate, it showed a decrease in the contact time with the catalyst bed; hence, the formation rate of methanol is limited to the feed total flow. Although the formation rate of methanol increases linearly with the total flow rate (12.5, 20, and 37.5 mL/min) and temperature variation from 225 to 275 °C, it suffers from selectivity, which is discussed more clearly with a contact time study further. Also, the effect of temperature on 5Sn/10InZrO2 for CO2 hydrogenation to methanol was estimated (Figure S5). Upon increasing the reaction temperature from 225 to 275 °C, the formation rate of methanol was enhanced up to 0.0275 μmol/(gcat·s), but the selectivity to methanol was decreased with a concomitant increase in the CO selectivity.

Figure 15

Effect of feed flow rate on the methanol formation rate with different reaction temperatures over 5Ni5Sn/10InZrO2 (reaction conditions: T = 225–275 °C, P = 25 bar, feed flow ration (CO2/Ar/H2) = 1:1:3, WHSV 15 000–45 000 h–1, CO2/H2 = 1:3, time = 24 h).

The hydrogenation of CO2 over 5Ni5Sn/10InZrO2 was studied with respect to contact time, and the methanol selectivity results are presented in Figure . To check in a more rational way, different catalyst weights were used for CO2 reduction to obtain different contact times at a constant total feed flow and studied in the temperature range from 225 to 275 °C. The selectivity to methanol decreases linearly at all reaction temperatures from 225 to 275 °C. Therefore, we can consider that the reactor is operating in a differential mode. At low contact time and at low temperature, the selectivity to methanol is high (around >99%). A further increase in both the residence time (0.078–0.24 L/(g·s)) and the reaction temperature (225–275 °C) leads to a decrease in the selectivity of MeOH from 99 to 54% at the cost of selectivity to CO formation. In contrast, the catalyst showed high rate for RWGS to produce CO instead of methanol. It was clearly observed that low catalyst amount and contact time retain high selectivity to methanol. As against, the optimized reaction condition 250 °C with contact time (0.12 L/(g·s)) showed very high selectivity to methanol. Hence, the NiSn combination supported on InZrO2 catalysts shows more favorable selectivity to methanol than CO; thus, these results indicate that the catalyst composition would be more facile to revert methane and CO selectivity to methanol via CO2 hydrogenation.

Figure 16

Methanol (CH3OH) selectivity versus contact time over 5Ni5Sn/10InZrO2 (reaction conditions: T = 250 °C, P = 25 bar, gas flow (CO2/Ar/H2) = 5/5/15 mL/min, WHSV 15 000–45 000 h–1, CO2/H2 = 1:3, time 24 h).

Furthermore, we check the long-term stability of optimized catalysts 5Ni5Sn/10InZrO2 using the time-on-stream stability test by keeping all other parameters constant, as mentioned in Figure . As shown in Figure , the catalyst sustained very high productivity with 0.1043 μmol/(gcat·s) formation rate of methanol over 50 h. Also, this catalyst composition contains NiSn alloy formation, easy reducibility of NiO due to the addition of Sn. Moreover, incorporation of Sn and In is helpful in improving the basic strength, which act as a promoter and stabilizer, and these components are main active components and support the high formation rate of methanol through CO2 hydrogenation. The high productivity of methanol with long-term stability shows that this catalyst is active and stable for CO2 hydrogenation to methanol.

Figure 17

Time-on-stream stability over 5Ni5Sn/10InZrO2 (reaction conditions: T = 250 °C, P = 25 bar, gas flow (CO2/Ar/H2) = 5/5/15 mL/min, WHSV 30 000 h–1, CO2/H2 = 1:3, time = 24 h).

DFT Calculations

Figure shows the various explored DFT-based atomic configurations of the 55-atom Ni28Sn27 bimetallic nanocluster (Ni/Sn ratio of around 1, which is very close to that of the obtained samples in our experiment) generated from a substitution of 27 Ni atoms by 27 Sn atoms at different occupation sites including core–shell and random alloy dispositions of the 55-atom icosahedron Ni55 nanocluster. The relative formation energy per total number of atoms was computed for each structural configuration (given in brackets) to provide insight into the possible structural evolution during heating treatment and to help identify the surface composition of Ni–Sn bimetallic samples under catalytic conditions. The lowest-energy structure (0 meV/atom) as well as the less stable one by 44 meV/atom revealed similar core–shell disposition, where the core is entirely occupied by Ni atoms, whereas all Sn atoms are somehow dispersed on the overall surface, leading to the formation of surface Sn–Ni and Sn–Sn bonds (Figure a,b). This result indicates the probable presence of the Ni (core)–Sn (shell) structure even beyond room temperature until around 510 K, which is very close to the treatment temperature used in our catalytic tests. The configuration showing almost full Sn segregation was found to be 60 meV/atom less stable than the first case (Figure c) and thus this disposition was expected to appear around 696 K. However, other configurations displaying random alloylike structures, with minority of Sn atoms incorporated into the core of the nanocluster and majority remaining on the surface, were obtained at 76 and 88 meV/atom higher energy than the lowest-energy one (Figure d,e). As a typical estimate, these atomic dispositions could become significant at higher temperatures in the 890–1020 K range. In all cases, the obtained structures upon full geometry optimization were found to be much distorted without any specific morphology as compared with the initial icosahedral shape adopted for pure Ni55.

Figure 18

DFT-based atomic configurations of the 55-atom Ni28Sn27 bimetallic nanocluster along with the computed relative formation energies (given in brackets) per total number of atoms (in meV/atom) (Ni atoms are shown in light blue and Sn atoms in red): (a) lowest-energy and (b–e) metastable structures.

By replacing 37 Ni atoms with 37 Sn atoms at various occupation sites including core–shell and random alloy dispositions into the 55-atom icosahedron Ni55 nanocluster, we have then explored the Sn-rich 55-atom Ni18Sn37 bimetallic nanocluster structures (Ni/Sn ratio of around 0.5, which is very close to that obtained for the samples in our experiment). The obtained DFT-based optimized atomic configurations are displayed in Figure . The most stable structural configuration at room temperature was obtained when all Sn species are located on the surface and the core is fully occupied by Ni atoms, leading to Ni (core)–Sn (shell) disposition (Figure a). The configuration revealing a random alloylike structure, with a limited number of Sn atoms incorporated into the core of the nanocluster and majority remaining on the surface, was found less stable by 42 meV/atom than the previous one (Figure b). This result indicates the probable presence of an alloy structure with the majority of Sn–Sn bonds on the surface at around 490 K, which is very close to the treatment temperature used in our catalytic tests. However, the configuration showing a kind of Ni segregation was found to be 68 meV/atom less stable than the first case (Figure c) and thus this disposition was expected to appear at a relatively high temperature (around 800 K). Similarly as obtained in the cases of Ni28Sn27, all of the obtained structures revealed a strong distortion with respect to the initial icosahedral shape adopted for pure Ni55. This clearly highlighted the substantial impact of substituted Sn on the surface structures of the Sn-rich Ni–Sn bimetallic catalysts.

Figure 19

DFT-based optimized atomic structures of the Sn-rich 55-atom Ni18Sn37 bimetallic nanocluster together with the relative formation energies (given in brackets) per total number of atoms (in meV/atom) (Sn atoms are shown in red and Ni atoms in light blue): (a) most stable and (b, c) metastable structures.

Finally, we have explored Ni-rich 55-atom Ni37Sn18 bimetallic nanocluster structures (Ni/Sn ratio of around 2, which is very close to that obtained for the samples in our experiment) by substituting 18 Ni atoms by 18 Sn atoms at different occupation sites including core–shell and random alloy dispositions of the 55-atom icosahedron Ni55 nanocluster. Figure illustrates the key explored DFT-based optimized atomic configurations. The lowest-energy structure revealed a core–shell disposition, where the core is entirely occupied by Ni atoms, whereas all Sn atoms are located on the surface (Figure a), highlighting this atomic disposition as the most stable one at room temperature. However, the configuration displaying a random alloylike structure, with a minority of Sn atoms incorporated into the core of the nanocluster and the majority of Sn atoms well dispersed on the surface, was found less stable by 47 meV/atom than the lowest-energy one (Figure b). This result indicated the probable presence of a random alloylike structure with the formation of surface Ni–Ni and Sn–Ni bonds at around 545 K, which is very close to the treatment temperature used in our catalytic tests. In contrast, the configuration showing almost full Sn segregation was found to be 112 meV/atom less stable than the first one (Figure c) and thus this disposition becomes significant at around 1300 K. Different from the two previous Ni28Sn27 and Ni18Sn37 cases, the optimized atomic structures obtained for the Ni-rich 55-atom Ni37Sn18 bimetallic nanocluster revealed a much less distortion, keeping roughly the initial icosahedral shape of the pure Ni55. In summary, DFT calculations showed differences between the morphology, surface composition, and structural location of each element on the surface of the three investigated Ni28Sn27, Ni18Sn37, and Ni37Sn18 bimetallic nanoclusters. This discrepancy is relevant with different catalytic activity trends for CO2 hydrogenation to methanol, and this aspect was discussed in the activity results above in Figure and12.

Figure 20

DFT-based optimized atomic configurations of the Ni-rich 55-atom Ni37Sn18 bimetallic nanocluster along with the computed relative formation energies (given in brackets) per total number of atoms (in meV/atom) (Ni atoms are shown in light blue and Sn atoms in red): (a) lowest-energy and (b, c) metastable structures.

Conclusions

In summary, the addition of Sn to Ni/ZrO2 and Ni/InZrO2 resulted in an enhancement in the formation rate of methanol with an order of magnitude via CO2 hydrogenation. The selectivity of the undesired methane and reverse water gas shift reaction was substantially affected by Sn addition, the involvement of Sn showing the change in the selectivity to methanol in the CO2 hydrogenation reaction. The XRD, HRTEM, EELS, and XPS characterizations confirm that the NiSn alloy as well as the Ni core and the Sn shell was formed. Also, the addition of Sn helps in reducing the particle size of agglomerated Ni particles and increases the reducibility of Ni oxide, which was considered as an active site for hydrogenation reactions. This was observed clearly with inhibition of the CO2 methanation reaction, and selectivity to methanol was improved successfully to 99%. Furthermore, with an InZrO2 support, a dramatic increase in the intrinsic activity for CO2 hydrogenation was obtained; these results point out the bifunctional active sites where both NiSn and In oxides were associated in methanol synthesis. These above-mentioned findings are likely due to ease of reducibility of Ni oxide and increase in basic strength of catalysts toward facile CO2 hydrogenation.

The possible structural evolution of Ni–Sn bimetallic samples at the atomic level during the heating treatment has also been investigated by DFT calculations to help identify their surface compositions under catalytic conditions. The DFT study showed differences between the morphology, surface composition, and structural location of each element on the surface of the three investigated Ni28Sn27, Ni18Sn37, and Ni37Sn18 bimetallic nanoclusters. This was explicitly confirmed by the different trends obtained in the catalytic activity and selectivity for CO2 hydrogenation to methanol. Also, DFT calculations suggested the formation of Ni (core)–Sn (shell) and random alloylike atomic structures, which could apparently match with the STEM, EDS, and elemental mapping studies. The excellent correlation between model surfaces and supported catalysts demonstrated the feasibility of employing this combined approach to guide the design of bimetallic catalysts for CO2 hydrogenation to methanol.

Experimental Section

Materials

Nickel nitrate, indium nitrate, zirconium nitrate, tin acetate, and sodium borohydride were purchased from Sigma-Aldrich and used directly. High-purity gases (H2 99.9995%; CO2, Ar 99.9999%) were purchased from the Specialty Gases Center of Abdullah Hashim Industrial Gases & Equipment Co. Ltd. (Saudi Arabia).

Catalyst Preparation

Support Preparation

For this synthesis, a Chemspeed I-synth platform was used. It allowed volumetric dispensing of solutions and gravimetric dispensing of solids.

ZrO2 Support Preparation

First, 20 mL of 0.5 M K2CO3 was added dropwise (dosing rate: 0.5 mL/min) to already prepared 20 mL of 0.5 M ZrO(NO3)2·H2O solution while stirring at 500 rpm, allowing a well-controlled precipitation. After that, the pH was adjusted by the pH control tool to 9.5. The resulting precipitate was separated using a centrifuge at 8000 rpm for 10 min and washed with deionized water three times and two times with ethanol.

10% InZrO2 Support Preparation

In, Zr, and K2CO3 starting solutions were prepared off-line and put on the platform at defined positions. In and Zr precursor solutions were added together to result in a 20 mL mixed solution having a total concentration of 0.25 M (In(NO3)2·3H2O) and 0.5 M ZrO(NO3)2·H2O with the desired ratio of In and Zr. Then, 20 mL of 0.5 M K2CO3 was added dropwise (dosing rate: 0.5 mL/min) to this solution while stirring at 500 rpm, allowing a well-controlled precipitation. After that, the pH was adjusted to 9.5. Then, the resulting precipitate was separated by using a centrifuge at 8000 rpm for 10 min and washed with deionized water for three times and two times with ethanol.

Wet Chemical Reduction Method

Ni and NiSn catalysts with various loadings on ZrO2 and InZrO2 supports were prepared by a wet chemical reduction method. For the synthesis of 5Ni/ZrO2, a calculated amount of nickel(II) nitrate hexahydrate was dissolved in deionized water with 1 g of ZrO2 support (preparation procedure as given above) and stirred for 30 min at a constant temperature of 90 °C. NaBH4 (99.999%) (metal/NaBH4 = 1:7) was dissolved in 20 mL of deionized water and added dropwise to the solution. The reaction was allowed to proceed for 3 h at 90 °C temperature for complete reduction of the metal salt. After the reaction, the resulting sample was centrifuged at 7000 rpm for 5 min thrice to remove unwanted ions and washed with excess deionized water. The catalyst was dried overnight at 120 °C and used as prepared without any further heat treatment.

The 5Ni5Sn/ZrO2, 5Ni/10InZrO2, and 5Ni5Sn/10InZrO2 catalysts were prepared similarly as follows: calculated amounts of nickel(II) nitrate hexahydrate and tin acetate were dissolved in deionized water with 1 g of ZrO2 and InZrO2 (prepared as above) supports and stirred for 30 min at a constant temperature of 90 °C. NaBH4 (99.999%) (metal/NaBH4 = 1:7) was dissolved in 20 mL of deionized water and added dropwise to the solution. The reaction was allowed to proceed for 3 h at 90 °C, the temperature necessary for a complete reduction of the metal salt. After the reaction, the sample was centrifuged at 7000 rpm for 5 min thrice to remove unwanted ions and washed with excess deionized water. The catalyst was then dried overnight at 120 °C in the open air and used as such without any further heat treatment.

Hydrogenation Experiments and Analysis

All catalytic tests were conducted in the gas phase using a fixed-bed isothermal flow reactor purchased from Process Integral Development Eng&Tech (PID Eng&Tech, Spain). In a typical experiment, 0.050–0.250 g of the as-prepared catalysts diluted with 0.950–0.750 g of inert carborundum (vide supra) was loaded into a stainless steel (SS-316) reactor (length 30 cm and 0.9 cm i.d.). The stainless steel tube was equipped with a 20 μm (mesh size) porous plate made with Hastelloy C, which was used to hold the catalyst inside the reactor. The reactor was then heated to 250 °C (50 °C/min) under CO2, H2, and Ar (1:3:1 respective molar ratio) gas mixture (CO2, 5 mL/min; H2, 15 mL/min; and Ar, 5 mL/min, all of the flow values at NTP) with total feed flow (25 mL/min at NTP). The pressure (1–25 bar) inside the reactor was controlled using an automated micrometric valve.

Product gas analysis was performed on-line using a Varian 450 GC gas chromatograph. A sample from the reactor outlet stream was automatically injected on three parallel channels referred to here as channel A, channel B, and channel C. In channel A, the sample (1 mL at STP) was injected on a set of three packed columns, “Hayesep” Q (CP81073), “Hayesep” T (CP81072), and “Molsieve” 13× (CP81073) connected in series. A set of 10-way and 6-way Valco valves was used to allow automatic injection of the sample, back-flushing of Hayesep T, and bypassing of “Molsieve” 13× columns. This channel was equipped with a TCD detector (He as a reference gas) and used to monitor the amounts of CO and CO2. Channel B used a set of two capillary columns, CP-Wax 52CB (CP8553) as a precolumn and Al2O3 MAPD (CP7432), connected in series. This channel was used to monitor methane, ethane, and ethylene. Meanwhile, channel C was equipped with a CP-wax 52CB column (CP7668) to separate oxygenates. Channels B and C were equipped with a flame ionization detector. A 10-way Valco valve was used to allow simultaneous injection of the gas samples on both channels A (500 μL at STP was injected) and B (250 μL at STP was injected). Estimation of CO2 conversion was made directly from the carbon balance based on the product concentrations for which the sensitivity is higher. The formation rate of methanol is defined in micromoles produced per gram of the catalyst per second.

Catalyst Characterization

Surface area measurements were carried out using N2 adsorption/desorption isotherms at liquid nitrogen temperature (77 K) on a Micromeretics ASAP2420 instrument. The surface areas of the samples were analyzed using the multipoint Brunauer–Emmett–Teller (BET) model. Prior to these measurements, the catalyst samples were degassed under vacuum at 250 °C for 2 h. X-ray diffraction (XRD) patterns for all samples were collected using a Bruker D8 Advanced A25 diffractometer with Cu K radiation, which is operated at 40 kV and 40 mA. The data sets were acquired in the step scan mode in the 2θ range of 20–80°, using a step interval of 0.05° and a counting time of 10°/min.

Morphology of all samples was characterized by transmission electron microscopy (TEM), using a Titan 80-300 ST microscope from Thermo-Fisher Scientific. This microscope was equipped with a probe-corrector to perform the Cs-corrected scanning STEM analysis. It was also equipped with an X-ray energy dispersive spectrometer (EDS) and a post-column energy filter for the determination of the elemental composition of samples by acquiring EDS and electron energy-loss spectroscopy (EELS) spectra, respectively. The specimens for this analysis were prepared by mixing the powdered samples in a pure ethanol solution. A small amount (<5 μL) of resultant solution from each sample was then placed on Holey-carbon-coated copper grids. These grids were then air-dried at ambient condition for several hours before performing the analysis. STEM imaging of the samples was completed by operating the microscope at an accelerating voltage of 300 kV, and the images were recorded using a high-angle annular dark-field (HAADF) detector and were recorded by setting the microscope in a given range of magnifications. Both EDS and core-loss EELS spectra were acquired by placing the electron beam on samples for several seconds. The selected-area electron diffraction (SAED) patterns were also acquired in the TEM mode to investigate the crystal structure of samples. At the end, spectrum-imaging (SI) data sets in the STEM-EELS configuration were also acquired to investigate the distribution of elements in the samples.

H2 temperature-programmed reduction (H2-TPR) measurements were carried out using an Altamira Instrument (AMi-200Ip) equipped with a TCD detector to measure the H2 uptake. The catalyst sample (about 0.0250 g) was kept in a U-shaped quartz tube and pretreated under Ar flow (50 mL/min) for 2 h at 250 °C to remove moisture and other surface impurities present in the sample. Furthermore, the sample was allowed to cool down at room temperature under the flow of Ar. Finally, the temperature was raised from 50 to 750 °C (10 °C/min) under a flow of 5% H2/Ar (50 mL/min).

High-resolution X-ray photoelectron spectroscopy (XPS) was used to study the chemical composition and oxidation state of the catalyst surfaces. XPS studies were carried out using a Kratos Axis Ultra DLD spectrometer equipped with a monochromatic Al Kα X-ray source (hν = 1486.6 eV) operating at 150 W, a multichannel plate, and a delay line detector under a vacuum of 1 × 10–9 mbar. The survey and high-resolution spectra were recorded at fixed analyzer pass energies of 160 and 20 eV, respectively, and quantified using empirically derived relative sensitivity factors provided by Kratos analytical. Samples were mounted in floating mode to avoid differential charging. Charge neutralization was required for all samples. Binding energies were referenced to the C 1s peak of C–C and C–H bonds, which was set at 284.8 eV. The data were analyzed with commercially available software (CasaXPS).

Before CO2-TPD, the catalyst samples (about 0.0250 g) were first reduced in 5% H2/Ar at 50 mL/min for 2 h at 300 °C. Then, the gas was changed to 1% CO2/He with 50 mL/min at 50 °C and kept for 1 h. TPD was performed under Ar flow at 50 mL/min to study the basic strength of the catalysts. The temperature was increased from 50 to 750 °C at the ramp rate of 10 °C/min. The effluent gases also were analyzed on a mass spectrometer (MS) (Hiden Analytical).

DFT Computations

For structural simulations of bimetallic Ni–Sn nanoclusters, we considered as a starting material the initial 55-atom geometrical model of the nickel Ni55 nanocluster with the commonly observed FCC-derived icosahedral shape containing 42 atoms on the surface and 13 atoms in the core. Several key atomic configurations of bimetallic Ni–Sn nanoclusters were modeled, in which 27, 37, and 18 Ni atoms were substituted by 27, 37, and 18 Sn atoms at different sites including core–shell and random alloy dispositions, to mimic the three samples obtained experimentally with Ni/Sn ratios of around 1, 0.5, and 2, respectively. The relative formation energy was systematically computed for each structural configuration to give a hint about the possible structural evolution during the heating treatment to help identify the surface composition of these nanoparticles under catalytic conditions.

Spin-polarized density functional theory (DFT) together with the plane wave approach as implemented in the Vienna Ab initio Simulation Package[51−54] were applied for total energy calculations of the various explored bimetallic Ni–Sn bimetallic structures. Each nanocluster was incorporated into a relatively large vacuum box to avoid any possible interaction with neighboring nanoclusters introduced by the periodic boundary conditions. The Perdew–Burke–Emzerhof exchange-correlation functional[55] and the projector-augmented plane wave approach[56] were employed to describe the electron–electron and electron–ion interactions, respectively. The convergence criterion for the electronic self-consistent-field cycles was fixed at 10–5 eV. To obtain reliable computed electronic energies of the simulated bimetallic Ni–Sn nanoclusters, the atomic coordinates were fully relaxed until the three components of the Hellmann–Feynman forces on each atom reached values below 10–2 eV/Å.