A Highly Active Au/In₂O₃-ZrO₂ Catalyst for Selective Hydrogenation of CO₂ to Methanol

Abstract

A novel gold catalyst supported by In₂O₃-ZrO₂ with a solid solution structure shows a methanol selectivity of 70.1% and a methanol space–time yield (STY) of 0.59 g_MeOH h⁻¹ g_cat⁻¹ for CO₂ hydrogenation to methanol at 573 K and 5 MPa. The ZrO₂ stabilizes the structure of In₂O₃, increases oxygen vacancies, and enhances CO₂ adsorption, causing the improved activity.

1. Introduction

With the development of renewable hydrogen, hydrogenation of CO₂ to methanol (Equation (1)) has received increasing attention worldwide. Significant efforts have been made towards the preparation of catalysts with high CO₂ conversion and high methanol selectivity. The most investigated catalysts are promoted copper catalysts based on the industrial Cu/ZnO catalysts for methanol synthesis from syngas [1,2,3]. Recently, ZnO-ZrO₂ [4], GaNi alloy [5], bimetallic PdZn [6], MOF [7,8], and others [1] were reported to have nice activity for CO₂ hydrogenation to methanol.

CO₂ + 3 H₂ → CH₃OH + H₂O ΔH = −49.7 kJ mol⁻¹

(1)

In 2012, our group predicted via density functional theoretical (DFT) studies that In₂O₃ possesses high activity for CO₂ activation [9]. The further DFT [10] and experimental [11] studies confirmed that In₂O₃ with oxygen vacancies exhibited high activity for CO₂ hydrogenation to methanol, with unusually high methanol selectivity. Since then, In₂O₃-based catalysts have received significant attention, with increasing publications. The reported works include preparation and characterization of In₂O₃ [12]; combination of In₂O₃ with other oxides [13,14], especially ZrO₂ that forms the solid solution with In₂O₃ and improves CO₂ adsorption with optimum oxygen vacancy [15,16,17] and generates strong electronic interaction between oxides [18]; and the addition of metals, including gold [19,20], copper [21,22], cobalt [23], palladium [24,25,26,27], rhodium [28], nickel [29], and platinum [30,31]. The metal catalysts applied significantly improve the hydrogenation ability of In₂O₃. It has been found that In₂O₃ has a strong interaction with metals. This strong interaction causes a total change in catalytic performance. It makes the Ni, Au, Rh, and Pt catalysts become highly selective towards CO₂ hydrogenation to methanol with In₂O₃ support. Especially for the Au catalyst, most of the reported Au catalysts are highly active for oxidation, with almost a thousand papers per year recently. However, only a few papers were published for hydrogenation. The In₂O₃-supported Au catalyst shows the highest selectivity and best activity ever reported for CO₂ hydrogenation to methanol. The methanol selectivity reaches 100% at temperatures below 498 K and is more than 70% at 548 K [19]. It is even 67.8% at 573 K, with a space–time yield (STY) of methanol of 0.47 g_MeOH h⁻¹ g_cat⁻¹ at 573 K, 5 MPa, and 21,000 cm³ h⁻¹ g_cat⁻¹. In this work, we confirm that the In₂O₃-ZrO₂ supported Au catalyst presents even higher activity for CO₂ hydrogenation to methanol.

2. Results

In₂O₃-ZrO₂ oxides were prepared via a co-precipitation method, and then Au nanoparticles (NPs) were loaded on the carrier through the deposition–precipitation method. The loading amount of Au was 1%. The samples after reaction at 573 K and 5 MPa were named with “-AR” at the end. As shown in Supplementary Materials Figure S1, the sample with In/Zr molar ratio of 4:1 shows the best catalytic performance. The sample with this In/Zr ratio (labeled as Au/In₂O₃-ZrO₂) was thus selected for the activity test and further characterization. Based on the N₂ adsorption–desorption test on 77 K, the In₂O₃-ZrO₂ and Au/In₂O₃-ZrO₂ samples exhibit a similar total specific surface area of 111.5 and 121.6 m² g⁻¹, respectively. As shown in Figure 1a, the results of X-ray diffraction (XRD) show that the characterized peaks for all samples containing indium belong to cubic-In₂O₃ (PDF file number 06-0416). Meanwhile, the peaks corresponding to Au cannot be observed, which results from the extremely low loading and ultra-high dispersion of Au NPs. This also means that Au NPs remain stable, preventing deactivation by the agglomeration. The absence of metallic indium after reaction also shows that the structure of the Au/In₂O₃-ZrO₂ catalyst is stable during the reaction. Moreover, the diffraction peak of the In₂O₃-ZrO₂ sample largely differentiates from that of ZrO₂, which can be deduced that Zr may be penetrated into In₂O₃ [32]. In Figure 1b, the high-resolution transmission electron microscopy (HRTEM) image of the Au/In₂O₃-ZrO₂ sample shows the crystalline lattice distances of 0.180, 0.398, 0.275, and 0.291 nm, which are assigned to cubic-In₂O₃ (440), (211), (321), and (222) facets, respectively. No lattice fringes of ZrO₂ or Au can be observed. As shown in Figure 1c, the result of element distribution analysis indicates that the supported Au NPs and Zr are highly dispersed before and after reaction (Supplementary Materials Figure S2). These results further confirm the excellent dispersion of Au NPs and the conclusion that In₂O₃-ZrO₂ is in a solid solution state.

The activity test of CO₂ hydrogenation to methanol was conducted in a vertical fixed-bed reactor. The products were analyzed with an online gas chromatograph (Agilent 7890A). For all catalysts, methanol and CO are the main products. Trace methane can be detected at reaction temperatures over 548 K. The carbon balance is better than 99%. As shown in Figure 2a, the Au/In₂O₃-ZrO₂ catalyst exhibits the highest catalytic performance among these three catalysts. The CO₂ conversion of the Au/In₂O₃-ZrO₂ catalyst is 14.8%, with a methanol selectivity of 70.1% at 573 K and 5 MPa. The methanol STY reaches 0.59 g_MeOH h⁻¹ g_cat⁻¹, which is much higher than that over the Au/In₂O₃ catalyst (0.47 g_MeOH h⁻¹ g_cat⁻¹) [19] and other Au-based catalysts (Supplementary Materials Table S1). For all catalysts, CO₂ conversion increases with the temperature increases, while the methanol selectivity decreases due to the competitive reverse water–gas shift (RWGS) reaction (Equation (2)). Although CO can be detected at 498 K for the Au/In₂O₃-ZrO₂ catalyst, the higher methanol selectivity at 523 K and 548 K is obvious, as compared to the Au/In₂O₃ catalyst. This result reveals that the Au/In₂O₃-ZrO₂ catalyst maintains high selectivity for methanol synthesis from CO₂ hydrogenation. On the other hand, the presence of Zr enhances the CO₂ conversion of the Au/In₂O₃ catalysts. As shown in Figure 2c, the methanol STY on the Au/In₂O₃-ZrO₂ catalyst only decreases slightly and is stabilized at 92% of the initial value after a 6 h reaction at 573 K and 5 MPa. Furthermore, the Au/In₂O₃-ZrO₂ catalyst shows the lowest apparent activation energy (Ea) of 74.1 kJ mol⁻¹ (Figure 2d). These results prove the excellent performance for CO₂ hydrogenation to methanol over the Au/In₂O₃-ZrO₂ catalyst.

CO₂ + H₂ → CO + H₂O ΔH = 41.2 kJ mol⁻¹

(2)

To further understand the factors responsible for the enhanced catalytic performance, Raman analyses were performed. As shown in Figure 3a, the band at 132 cm⁻¹ is assigned to the In-O vibration (V_In-O) [25]. The scattering features at 306 cm⁻¹ (I₁) and 493 cm⁻¹ belong to the bending vibration and stretching vibrations of InO₆ octahedrons, respectively [33]. The stretching vibration of In-O-In, occurring at 364 cm⁻¹ (I₂), was confirmed to reflect the degree of oxygen defects [19]. For fresh samples, the band of V_In-O visibly shifts towards lower frequencies due to the addition of Zr. Moreover, the characteristic peaks of the In₂O₃-ZrO₂ and Au/In₂O₃-ZrO₂ samples are completely the same as those of In₂O₃ without the peaks attributed to ZrO₂ (Supplementary Materials Figure S3). These results affirm the penetration of Zr atoms into the indium oxides [32]. The intensities of I₂ increase with the addition of Zr, suggesting that more oxygen vacancies are generated. However, the loading of Au results in lower I₂ intensities, likely owing to the coverage of Au NPs on oxygen vacancies [30]. Subsequently, the variation of related peaks over the samples after H₂ reduction for different times under 473 K (labeled with “-HR-1.5h” or “-HR-1h”at the end) was further investigated. As shown in Figure 3b,c, H₂ reduction results in the increase of oxygen vacancies for all samples, according to the much larger I₂ band [30]. Moreover, prolonging reduction time enables oxygen vacancies to increase sequentially. The area ratio of I₂ and I₁ (labeled as I₂/I₁) was calculated to characterize the amount of oxygen vacancies (Figure 3d). The oxygen vacancies of the Au/In₂O₃ sample are 40% of fresh In₂O₃, while those of the Au/In₂O₃-HR-1.5h sample are 46% of the In₂O₃-HR-1.5h sample. Comparing Au/In₂O₃-ZrO₂ with In₂O₃-ZrO₂ under the same treatment condition, we see that the ratio of their I₂/I₁ value increases by 21% from fresh samples to the samples after 1.5 h H₂ reduction. This result indicates that the well-dispersed Au NPs benefit the formation of oxygen vacancies under the atmosphere including H₂. Furthermore, normalized by the I₂/I₁ value of fresh In₂O₃, the oxygen vacancies of samples after 1.5 h H₂ reduction increase by 2.04, 0.92, 1.00, and 0.90 times in In₂O₃, In₂O₃-ZrO₂, Au/In₂O₃, and Au/In₂O₃-ZrO₂, respectively. During the H₂ reduction process, the oxygen vacancies of In₂O₃ increase much more drastically than the other samples. Noticeably, some bulk information of the sample can be obtained under the excitation laser with the wavelength of 532 nm [34]. The rapid increment of oxygen vacancies can be attributed to the over-reduction of pure In₂O₃ [30]. It results in metallic indium without activity for H₂ activation [13]. Thus, it can be deduced that the strong metal-support interaction (SMSI) between Au and carriers could be conducive to maintain the structure of In₂O₃ [19], which can also be performed through the incorporation of Zr into In₂O₃. Furthermore, compared to Au/In₂O₃, the Au/In₂O₃-ZrO₂ sample exhibits the higher I₂/I₁ values for not only fresh samples but also the samples after reduction, implying the increase of oxygen vacancy density. These results coincide with the better catalytic activity of the Au/In₂O₃-ZrO₂ catalyst.

H₂ temperature-programmed reduction (H₂-TPR) was employed to investigate the reduction behaviors of the catalysts. The H₂-TPR profiles are shown in Figure 4a. For all samples, the reduction peaks located below 500 K can be assigned to the surface reduction [30]. The onset of reduction peak at 530 K belongs to the reduction of bulk In₂O₃. Due to the fact that In₂O₃ can be reduced by hydrogen more easily than ZrO₂ [17], the enhanced stability of Au/In₂O₃-ZrO₂ under the hydrogen atmosphere is thereby achieved. The XRD patterns of the spent samples were further investigated. As shown in Supplementary Materials Figure S4, there is a noticeable shift towards a lower angle for the In₂O₃-AR sample compared to fresh In₂O₃, while the Au/In₂O₃ and Au/In₂O₃-ZrO₂ samples can remain at the same angle after the reaction at 573 K and 5 MPa. Our previous work reported that the lattice expansion of In₂O₃ caused by thermal effect and oxygen vacancies could be performed on the shift of XRD diffraction peak [19]. Moreover, it is similar for lattice parameters of In₂O₃ and Au/In₂O₃, heated under argon and reactant mixture at 573 K, respectively [19]. As the thermal effect can be excluded, it can be conducted that this shift may be attributed to the increasing oxygen vacancies of bulk In₂O₃. Thus, it can be deduced that the addition of Zr and Au could effectively hinder the over-reduction of In₂O₃ under a reactant mixture including H₂. In the temperature range from 275 to 530 K, the surface reduction peak shifts towards the lower temperature (below 360 K) with the addition of Au NPs. This result confirms that Au NPs promote the formation of surface oxygen vacancies. In addition, a broader and intense H₂ consumption peak at 350 K for the Au/In₂O₃-ZrO₂ sample is obvious. The incorporation of Zr into In₂O₃ promotes the formation of surface oxygen vacancies on the Au/In₂O₃-ZrO₂ sample. The higher onset temperature of surface reduction and the disappeared peak at 456 K indicate the stabilization function of Zr on In₂O₃ structure. These results are in line with the results of Raman spectra above. The H₂-TPR profiles demonstrate the increasing surface oxygen vacancies over the Au/In₂O₃-ZrO₂ catalyst, as a reason for its prominent catalytic activity.

To further investigate the adsorption behavior of CO₂ on the catalyst surface, temperature-programmed desorption of CO₂ (CO₂-TPD) was conducted. Figure 4b presents the profiles of CO₂-TPD, involving several signals in the temperature ranges of 325–425, 485–570, 570–755, and 730–865 K. The single peak around 360 K for all samples belongs to the desorption of physically absorbed CO₂ [35]. Other peaks from low to high temperatures can be attributed to chemically absorbed CO₂ on the different sites, ascribed to weak, medium, and strong basic sites, respectively. Comparing the Au/In₂O₃-ZrO₂ sample with other samples, we see that the desorption peaks shift to the much higher temperatures of 707 and 798 K, respectively. This result implies that the addition of Zr enhances the strength of CO₂ adsorption on these sites, probably owing to the enhanced basicity [36]. It can be proposed that the enhanced CO₂ adsorption on the Au/In₂O₃-ZrO₂ sample benefits methanol synthesis. Time-resolved in situ diffuse reflectance infrared Fourier transform spectroscopy (in situ DRIFTS) were also conducted to gain the insights into CO₂ adsorption. As shown in Supplementary Materials Figure S5, the CO₂ interaction with Au/In₂O₃-ZrO₂ prefers to produce carbonate species rather than bicarbonate species due to the change of intensities [37]. The bicarbonate species disappears much faster after the Ar desorption, especially over the Au/In₂O₃ sample. Moreover, compared to the Au/In₂O₃ sample, the blue shift over the Au/In₂O₃-ZrO₂ sample for the bands at 900–1100 cm⁻¹, attributed to CO₂ absorbed on vacancies, suggests the stronger adsorption of CO₂ [17,38]. These results reveal that the addition of Zr strengthens the interaction between CO₂ and the catalyst surface. Based on the discussion above, the SMSI between Au and In₂O₃-ZrO₂ solid solution facilitates H₂ dissociation. Meanwhile, the incorporation of Zr into In₂O₃ leads to the increasing oxygen vacancies, as well as the enhanced CO₂ adsorption. The H adatoms, dissociated by the Au NPs strongly interacting with the carrier, can spill over to the oxygen vacancy sites and react with the CO₂ molecule activated on the oxygen vacancy sites. This significantly promotes the CO₂ conversion into methanol.

In conclusion, we have shown that the integration of Au, In₂O₃, and ZrO₂ enables the high selectivity for methanol synthesis from CO₂ hydrogenation. The Au/In₂O₃-ZrO₂ catalyst shows a CO₂ conversion of 14.8% with a methanol STY of 0.59 g_MeOH h⁻¹ g_cat⁻¹ at 573 K and 5 MPa. Such an outstanding activity indicates that the addition of Zr remarkably improves the activity of the Au/In₂O₃ catalyst. The catalyst characterizations demonstrate that Au NPs are well dispersed on the surface of the In₂O₃-ZrO₂ solid solution. Further investigation through Raman spectra, H₂-TPR, and CO₂-TPD indicates that the enhanced H₂ activation is achieved by the Au NPs strongly interacting with catalysts surface due to the SMSI effect. The presence of Zr mainly exhibits four effects: (1) reducing the apparent activation energy, (2) stabilizing the structure of In₂O₃, (3) increasing the amount of oxygen vacancies, and (4) enhancing the CO₂ adsorption. This work not only extends the utilization of gold catalysts for CO₂ hydrogenation but also promotes the fundamental studies for CO₂ conversion.

3. Materials and Methods

3.1. Catalyst Preparation

Indium-zirconium oxide carriers were prepared by a co-precipitation approach. Appropriate amounts of In(NO₃)₃·× H₂O (M = 363.3 g mol⁻¹, Beijing HWRK Chem Co., Ltd., Beijing, China, 99.99%) and ZrO(NO₃)₂·× H₂O (M = 257.9 g mol⁻¹, Shanghai Macklin Biochemicals Co., Ltd., Shanghai, China, 99%) with a fixed molar ratio were firstly dissolved in deionized water (0.145 M of In³⁺). A 0.2 M aqueous solution of Na₂CO₃ (Tianjin Kemiou Chemical Reagent Co., Ltd., Tianjin, China, 99%) was added dropwise under continuous stirring to reach a pH of 10. The mixture was aged at 353 K for 3 h. Thereafter, the precipitate was filtered and then washed with deionized water several times. Prior to the calcination at 723 K (10 K min⁻¹, 3 h), the sample was dried at 353 K overnight. The obtained carrier was labeled as In₂O₃-ZrO₂. The In₂O₃ and ZrO₂ samples were prepared with the same protocol. The Au/In₂O₃-ZrO₂ sample was prepared via a deposition–precipitation method, which is described in detail elsewhere [19]. The only difference was that the sample after freeze-drying overnight was directly used for the catalytic activity test, without a further reduction step. The samples with other molar ratios of In/Zr were denoted as Au/In₂O₃-ZrO₂ (m:n), where m:n is the molar ratio of In/Zr.

3.2. Characterization

The phase compositions of the samples were characterized by a Bruker D8 Focus diffractometer equipped with Cu Kα radiation source (λ = 1.54056 Å). Diffraction data were collected at a scanning speed of 8° min⁻¹ over the 2θ range of 10°–70°. The N₂ adsorption/desorption isotherms, measuring on an AUTOSORB-1-C instrument (Quantachrome) at 77 K, were used to determine the total specific surface area (S_BET).

The Raman spectra were recorded by an INVIA Raman analyzer (Renishaw PLC. Ltd., Gloucestershire, UK) with laser irradiation at 532 nm. The SEM images and EDX spectra were used to characterize microstructure and element distribution. All samples were measured on a Hitachi S-4800 SEM at an operating voltage of 25 kV without any conductive coating process. The transmission electron microscopy (TEM) and HRTEM images were obtained using a JEOL JEM-F200 microscope at the accelerating voltage of 200 kV.

H₂-TPR was carried out using the Micromeritics Autochem II 2920 chemisorption analyzer. About 100 mg of the as-prepared sample was pretreated under flowing helium for 1 h at 473 K to remove the absorbed impurities. Thereafter, the sample was cooled down to 273 K by liquid nitrogen under flowing helium. Then the sample was heated to 973 K (10 K min⁻¹) after switching to a mixture gas of 10% H₂ in N₂. The effluent gas was analyzed using a thermal conductivity detector (TCD). The signals were normalized to the weight of the sample for the following analysis.

CO₂-TPD was conducted on the same apparatus as H₂-TPR. After the same pretreating process as H₂-TPR, the sample (about 100 mg) was cooled down to 323 K under flowing helium, followed by CO₂ adsorption for 1 h. After being purged with flowing helium for 1 h to remove physically absorbed CO₂ at the same temperature, the sample was heated to 1073 K under flowing helium at a rate of 10 K min⁻¹. The effluent gas was analyzed by the TCD. The signals were normalized to the weight of the sample for the following analysis.

Time-resolved in situ DRIFTS was performed using a Tensor 27 spectrometer (Bruker) equipped with a MCT detector by recording 32 scans at a resolution of 2 cm⁻¹. After the pretreatment under flowing argon for 2 h at 323 K, the background spectra were firstly collected and subtracted from the sample spectra. For CO₂ adsorption, a 1.0% CO₂ in Ar gas with the flow rate of 20 mL min⁻¹ was fed into the system at 323 K. After the collecting of time-resolved spectra for 30 min, the gas was switched to pure Ar. The desorption time-resolved spectra were collected after flushing for 10 min.

3.3. Catalyst Activity Test

The activity test of CO₂ hydrogenation to methanol was conducted in a vertical fixed-bed reactor setup. The reactor was loaded with 0.2 g of the catalysts and 1.0 g of SiC, which was purged with N₂, at room temperature. After 0.5 h, the gas flow was switched to the reactant mixture (H₂/CO₂/N₂ = 76/19/5, molar ratio) until the pressure reached 5 MPa. The reaction was conducted with a temperature range from 473 to 573 K. The gaseous hourly space velocity (GHSV) was 21,000 cm³ h⁻¹ g_cat⁻¹. The reaction conditions were determined by the reported studies [17,39] and our previous study on Au/In₂O₃ [19] in order to get better catalytic performance. Products were analyzed with an online gas chromatograph (Agilent 7890A) equipped with a flame ionized detector (FID) and a TCD. All the valves and lines between the reactor outlet and GC inlet were heated to 413 K to prevent the condensation of methanol. CO₂ conversion (X_CO2) and selectivity (S_i) and space–time yield (STY, g_i g_cat⁻¹ h⁻¹) of product i (i = methanol or CO) were calculated as follows:

X_CO2 = (F_{CO2, in} − F_{CO2, out})/F_{CO2, in} × 100%

S_i = F_{i, out}/(F_{CO2, in} − F_{CO2, out}) × 100%

STY_i = (F_{CO2, in} × S_i × X_CO2)/m_cat × M_i

where F and M are the molar flow rate and the molar mass of product i, respectively, and m_cat is the catalyst sample weight.