Flame Synthesis of Cu/ZnO-CeO2 Catalysts: Synergistic Metal-Support Interactions Promote CH3OH Selectivity in CO2 Hydrogenation.

The hydrogenation of CO2 to CH3OH is an important reaction for future renewable energy scenarios. Herein, we compare Cu/ZnO, Cu/CeO2, and Cu/ZnO-CeO2 catalysts prepared by flame spray pyrolysis. The Cu loading and support composition were varied to understand the role of Cu-ZnO and Cu-CeO2 interactions. CeO2 addition improves Cu dispersion with respect to ZnO, owing to stronger Cu-CeO2 interactions. The ternary Cu/ZnO-CeO2 catalysts displayed a substantially higher CH3OH selectivity than binary Cu/CeO2 and Cu/ZnO catalysts. The high CH3OH selectivity in comparison with a commercial Cu-ZnO catalyst is also confirmed for Cu/ZnO-CeO2 catalyst prepared with high Cu loading (∼40 wt %). In situ IR spectroscopy was used to probe metal-support interactions in the reduced catalysts and to gain insight into CO2 hydrogenation over the Cu-Zn-Ce oxide catalysts. The higher CH3OH selectivity can be explained by synergistic Cu-CeO2 and Cu-ZnO interactions. Cu-ZnO interactions promote CO2 hydrogenation to CH3OH by Zn-decorated Cu active sites. Cu-CeO2 interactions inhibit the reverse water-gas shift reaction due to a high formate coverage of Cu and a high rate of hydrogenation of the CO intermediate to CH3OH. These insights emphasize the potential of fine-tuning metal-support interactions to develop improved Cu-based catalysts for CO2 hydrogenation to CH3OH.

Introduction

Methanol is an important commodity in the current chemical industry and expected to play a central role in the transition toward a sustainable economy as a platform chemical for the production of energy carriers and chemicals.[1] Modern industrial methanol production is based on the conversion of synthesis gas using Cu/ZnO/Al2O3 catalysts operated at elevated temperature (200–300 °C) and pressure (50–100 bar).[2] Catalytic conversion of CO2 to methanol (CO2 + 3H2 → CH3OH + H2O) using green hydrogen generated from sustainable energy sources has recently attracted significant attention.[3,4] This approach allows the reuse of the greenhouse gas CO2 for the production of fuels and chemicals using methanol as a platform.[4,5] Although Cu/ZnO/Al2O3 catalysts are active for CO2 hydrogenation to CH3OH, there are several problems with respect to its practical implementation. A major drawback lies in the high activity in the reverse water–gas shift (rWGS, CO2 + H2 → CO + H2O) reaction of Cu/ZnO/Al2O3 catalysts. The WGS reaction (CO + H2O → CO2 + H2) is important for methanol synthesis from synthesis gas over Cu/ZnO/Al2O3 catalysts because the main reaction pathway involves direct hydrogenation of CO2.[6−8] However, a high rWGS activity is undesirable for CH3OH synthesis from CO2, because the production of CO byproduct decreases CH3OH selectivity and H2 utilization efficiency.[8] The formation of a large amount of water byproduct during CO2 hydrogenation in comparison to synthesis gas hydrogenation poses another challenge to catalyst stability, as water can accelerate the deactivation of Cu/ZnO/Al2O3 catalysts.[9−12]

A strategy to improve Cu–ZnO-based catalysts for CO2 hydrogenation to methanol is to replace Al2O3 by other supports. ZrO2, for instance, has been widely investigated in this respect.[13−16] CeO2 is another promising support material to promote CO2 conversion.[17] CeO2 is more basic and less hydrophilic than Al2O3, which can be beneficial for CH3OH productivity and catalyst stability, respectively.[18,19] CeO2 interacts strongly with Cu, which may improve dispersion and resistance against sintering of the Cu particles.[20] While binary Cu/CeO2 catalyst is known for its high activity in CO hydrogenation to methanol,[21,22] a recent study showed that it is also promising for CO2 hydrogenation to methanol.[23] The influence of CeO2 addition to Cu–ZnO-based catalysts has been investigated as well.[18,24−28] Bonura et al. used a reverse co-precipitation method to prepare promoted Cu-based catalysts under ultrasonic conditions and found that CeO2 was a better promoter for CH3OH formation than ZrO2 in CO2 hydrogenation.[24] Gao et al. prepared a series of promoted Cu–ZnO-based catalysts from hydrotalcite-like precursors[25] and showed that the copper surface area and the fraction of strongly basic sites increased after CeO2 addition, resulting in a higher CO2 conversion and CH3OH selectivity than the unpromoted catalyst. In spite of these interesting results, a detailed structure-performance understanding of Cu–ZnO–CeO2 catalysts in CO2 hydrogenation to CH3OH remains largely unaddressed.

Regarding methanol synthesis catalyst preparation, academic studies often used impregnation methods to load Cu on a support.[29,30] However, this method is limited by the relatively low Cu loading that can be achieved.[31] Co-precipitation is another often used method to prepare methanol synthesis catalysts, and it is also used to prepare commercial Cu/ZnO/Al2O3 catalysts.[32−34] There are, however, disadvantages associated with co-precipitation as well.[35] First, co-precipitation is time-consuming in comparison with impregnation. A washing step is required to remove the precipitation agent, which involves the use of a large amount of solvent. A promising alternative method to prepare methanol synthesis catalysts is flame spray pyrolysis (FSP), which offers advantages with respect to flexibility, speed, and scalability.[36−41] In an earlier report, Jensen et al. demonstrated that Cu/ZnO/Al2O3 catalysts with a relatively high specific surface area (>100 m2/gcat) can be obtained using a flame combustion method.[41] Using a two-nozzle spray pyrolysis approach, Copéret and co-workers recently prepared a series of Cu/ZrO2 catalysts such that only the Cu particle size was varied.[37] They found that the catalysts with a smaller Cu particle size had a higher CH3OH activity and selectivity. Tada et al. demonstrated that efficient Cu/ZrO2 catalysts with high Cu loadings (up to 80 wt %) can be conveniently prepared by flame spray pyrolysis to convert CO2 into methanol.[36]

In this study, we prepared three series of Cu–Zn–Ce oxide catalysts with varying support composition and Cu loading by a one-step flame spray pyrolysis (FSP) method. This method potentially allows for homogeneous mixing of different components at the nanoscale,[42,43] which is highly advantageous for elucidating the role of Cu–ZnO and Cu–CeO2 interactions in CO2 hydrogenation. The as-prepared Cu–Zn–Ce oxide catalysts were characterized by ICP, N2 physisorption, TEM, STEM-EDX, XRD, TPR, and N2O titration, confirming that well-defined catalysts with controlled Cu dispersion and Cu-support interactions can be achieved via the FSP method. The Cu–Zn–Ce oxide catalysts were evaluated for their CO2 hydrogenation performance under the conditions of 250 °C and 30 bar. The catalytic results showed that (i) combining ZnO and CeO2 led to a higher CH3OH selectivity irrespective of Cu loading as compared to binary Cu/ZnO and Cu/CeO2 catalysts and (ii) the optimized Cu–Zn–Ce oxide catalyst with high Cu loading (∼40 wt %) also displayed an improved CH3OH selectivity as compared to a commercial Cu–ZnO-based catalyst. In situ IR spectroscopy was used to gain insight into the improved CH3OH selectivity over the ternary Cu–Zn–Ce oxide catalysts. Building upon the catalyst characterization and mechanistic studies, the improved CH3OH selectivity of Cu–Zn–Ce oxide catalysts in CO2 hydrogenation is discussed in the context of synergistic interactions between Cu, ZnO and CeO2 components.

Experimental Section

Catalyst Preparation

Cu–Zn–Ce oxide catalysts were prepared by a single-step flame spray pyrolysis (FSP) method using a Tethis NPS10 apparatus. The catalyst precursor solution was prepared by dissolving appropriate amounts of Cu(NO3)2·3H2O (99%, Sigma-Aldrich), Zn(NO3)2·6H2O (99%, Sigma-Aldrich), and Ce(NO3)3·6H2O (99%, Sigma-Aldrich) in a 1:1 (vol %) solvent mixture of ethanol (HPLC, Sigma-Aldrich) and 2-ethylhexanoic acid (99%, Sigma-Aldrich) at room temperature. The total metal (Cu, Zn, and Ce) concentration was 0.15 M. The precursor solution was then injected into the nozzle of the Tethis setup at a flow rate of 5 mL/min. The flame was fed with a 1.5 L/min methane flow and a 3.0 L/min oxygen flow with an additional 5.0 L/min oxygen dispersion flow around it. The resulting catalyst powder was collected from the quartz filter placed after the combustion zone. The as-prepared Cu–Zn–Ce oxide catalysts are denoted as Cu(x)/CeO2, Cu(x)/ZnO, or Cu(x)/ZnO–CeO2(y) where x and y, respectively, stand for Cu loading (wt %) and Zn atomic ratio in the supports (Zn/(Zn + Ce)). Additionally, a commercial Cu–ZnO-based methanol synthesis catalyst (MSC) was purchased from Alfa Aesar (no. 45776).

Catalyst Characterization

Inductively Coupled Plasma Optical Emission Spectroscopy (ICP-OES)

The metal composition (Cu, Zn, and Ce) of the as-prepared catalysts was measured by ICP-OES analysis (Spectro CIROS CCD Spectrometer). To prepare samples, the Cu–Zn–Ce oxide catalysts were dissolved in 5 mL of concentrated sulfuric acid (H2SO4) at 150 °C under stirring for about 1 h.

N2 Physisorption

The textural properties of the as-prepared catalysts were determined by measuring N2 physisorption isotherms at −196 °C on a Micromeritics TriStar II 3020 instrument. For this purpose, about 100 mg of catalyst sample was placed into a glass sample tube and pretreated at 120 °C under a N2 flow overnight. The Brunauer–Emmett–Teller (BET) method was used to calculate the specific surface area of the catalysts.

X-ray Diffraction (XRD)

The crystal structure of the as-prepared catalysts was determined with a Bruker D2 Phaser diffractometer using Cu Kα radiation with a wavelength of 1.5406 Å. The XRD patters were recorded between 20–85° with a step size of 0.02° and a scan rate of 0.5 s/step.

Electron Microscopy (EM)

The morphology of the as-prepared catalysts was studied by transmission electron microscopy (TEM) using a FEI Tecnai 20 (type Sphera) instrument operating at an acceleration voltage of 200 kV. For this purpose, appropriate amount of catalyst samples were dispersed in ethanol under ultrasonication and then deposited on holey Cu grids. The elemental distribution for reduced catalysts was determined by scanning TEM combined with energy dispersive X-ray analysis (STEM-EDX) using a FEI-cubed Cs-corrected Titan instrument operating at an acceleration voltage of 300 kV. To prepare samples, Cu–Zn–Ce oxide catalysts were reduced at 250 °C for 1 h in a flow of 10 vol % H2 in He, followed by passivation at room temperature in a flow of 2 vol % O2 in He. These reduced catalyst samples were then dispersed in ethanol under ultrasonication and deposited on holey Au grids.

Temperature-Programmed Reduction (H2-TPR)

H2-TPR measurements were carried out using a Micromeritics AutoChem II instrument. Typically, about 50 mg of catalyst sample was held between two quartz wool layers in a quartz U-tube. The sample was pretreated at 200 °C for 1 h in a He flow of 50 mL/min prior to H2-TPR measurement. The H2-TPR profile was recorded by heating the sample from 40 to 700 °C at a ramp rate of 10 °C/min in a 4 vol % H2 in He flow of 50 mL/min. The H2 consumption during temperature ramp was monitored by a thermal conductivity detector (TCD) and the recorded signal was calibrated against a Cu/SiO2 reference sample.

N2O Titration

The surface metallic Cu sites and oxygen vacancies in the reduced Cu–Zn–Ce oxide catalysts were determined by a combined CO2–N2O pulsing titration method using a plug flow setup coupled with a mass spectrometer (Balzers TPG 251). This titration method allows accurate measurement of metallic Cu surface without support interference because CO2 is used to block oxygen vacancies in the support of the reduced catalysts.[44] The detailed experimental procedures and calculations were described in a previous report.[45] In essence, one complete experiment consists of two subexperiments with sample reduction for 1 h in a 10 vol % H2 in He flow at the beginning and sample titration using a diluted 2 vol % N2O in a He gas mixture. The reduction temperature for catalysts with low Cu loading (x = 5) is 250 °C and for catalysts with high Cu loading (x = 45) and commercial MSC is 300 °C. Notably, the first N2O pulse titration was carried out at 50 °C directly after the sample reduction and an extra CO2 pulsing step at 50 °C was added between the sample reduction and the second N2O pulse titration at 50 °C in order to block oxygen vacancies in the reduced support. The total N2O consumption during the first N2O titration was S1 and during the second N2O titration was S2. The surface metallic Cu sites and oxygen vacancies in the reduced catalysts were calculated based on S2 and S1–S2, respectively.

Infrared Spectroscopy

Infrared (IR) spectra were recorded on a Bruker Vertex 70v FTIR spectrometer equipped with a DTGS detector. The experiments were performed in situ by using a home-built environmental transmission IR cell. Self-supporting pellets were made by pressing approximately 12 mg of a sample in a disk with a diameter of 13 mm. Each spectrum was collected by averaging 64 scans with a resolution of 2 cm–1 in the 4000–1000 cm–1 range. The samples were reduced in 10% H2 in He mixture at 250 °C for 1 h after heating at a rate of 5 °C/min. For CO adsorption experiments, the samples were outgassed at 250 °C prior cooling in vacuum to 40 °C. IR spectra were recorded as a function of CO partial pressure in the 0–10 mbar range. For temperature-programmed reaction (CO2 + H2) measurements, the samples were cooled in 10% H2 in He reduction mixture to 50 °C following reduction and prior to exposure to the CO2/H2/He mixture (5:15:80, total flow 200 mL/min). The samples were heated in this reaction mixture to 250 °C at a rate of 5 °C/min, while IR spectra were recorded at intervals of 10 °C. All samples are background subtracted, and the intensity was normalized to the weight of the pellet.

Catalytic Activity Measurements

CO2 Hydrogenation

The performance of Cu–Zn–Ce oxide and commercial MSC catalysts in CO2 hydrogenation was evaluated at 250 °C and 30 bar in a down-flow stainless-steel reactor with an internal diameter of 4 mm. The as-prepared catalysts were pressed, crushed, and sieved to a fraction of 125–250 μm. Typically, 50 mg (for Cu(5)/ZnO-CeO2(y)) or 25 mg (for Cu(45)/ZnO-CeO2(y) and commercial MSC) of catalyst diluted with 200 mg of SiC was loaded into the reactor. Prior to the reaction, the catalyst was reduced in a 10 vol % H2 in He flow of 50 mL/min while heating to 250 °C (for Cu(5)/ZnO-CeO2(y)) or 300 °C (for Cu(45)/ZnO-CeO2(y) and commercial MSC) at a rate of 5 °C/min followed by a dwell time of 1 h. The reaction was started by switching the pretreatment feed to the reaction feed at 250 °C and increasing the pressure in the reactor to 30 bar using a back-pressure regulator. The reaction feed is a gas mixture of H2/CO2/N2 at a volumetric ratio of 3:1:1, and the total flow was 50 mL/min. The effluent gas was continuously sampled and analyzed by an online gas chromatograph (Interscience, CompactGC) equipped with Rtx-1 (FID), Rt-QBond and Molsieve 5A (TCD), and Rt-QBond (TCD) analysis sections. The measurements were taken after ca. 3 h time-on-stream, and CO2 conversion, product selectivity, and product formation rates were calculated as followswhere F stands for the volumetric flow rate calculated based on N2 internal standard using calibrated response factors and Vm for the molar volume of ideal gas at standard temperature and pressure. The CH4 selectivity in all the measurements was very low (<1%). The turnover frequencies (TOFs) were calculated as the product (CH3OH or CO) formation rates normalized by the amounts of surface metallic Cu sites determined by N2O titration.

CO Hydrogenation

The CO hydrogenation activity of Cu–Zn–Ce oxide catalysts with low Cu loading, i.e., Cu(5)/ZnO-CeO2(y), was tested in the same catalytic setup as used for CO2 hydrogenation. The catalyst pretreatment and reaction conditions were kept the same as done for CO2 hydrogenation, except the CO2 in the reaction feed being replaced by the same amount of CO. The measurements were taken after ca. 3 h time-on-stream, and CO conversion, product selectivity, and product formation rates were calculated as followswhere F stands for the volumetric flow rate calculated based on N2 internal standard using calibrated response factors and Vm for the molar volume of ideal gas at standard temperature and pressure. The CH3OH selectivity in all the measurements was very high (>97%).

Results and Discussion

Three series of Cu–Zn–Ce oxide catalysts were prepared by FSP in this study, which are catalysts with a low Cu loading and varying support composition (Cu(5)/ZnO−CeO2(y)), catalysts with a high Cu loading and varying support composition (Cu(45)/ZnO−CeO2(y)), and catalysts with an intermediate Cu loading and constant support composition (Cu(x)/ZnO–CeO2(0.25) 5 < x < 45). In the following sections, the series of Cu(5)/ZnO–CeO2(y) and Cu(45)/Zn–CeO2(y) catalysts are discussed to understand how Cu–ZnO and Cu–CeO2 interactions affect Cu–Zn–Ce oxide catalysts in CO2 hydrogenation. The results of the catalysts with intermediate Cu loadings are provided in the Supporting Information.

Characterization

The physicochemical properties of the Cu(x)/ZnO–CeO2(y) catalysts (x = 5 and 45) prepared by FSP as well as the commercial MSC are listed in Table . The Cu loading and support composition determined by ICP-OES are close to the targeted values for all of the Cu–Zn–Ce oxide catalysts, demonstrating the reliability of FSP method in controlling catalyst composition. N2 physisorption was used to determine the specific surface areas of the as-prepared catalysts. For the catalysts with low Cu loading (x = 5), the values range from 82 to 130 m2/gcat and the Cu/CeO2 and Cu/ZnO samples exhibit, respectively, the largest and smallest surface areas. The catalysts with a high Cu loading of 45 wt % display slightly lower specific surface areas, ranging from 68 to 95 m2/gcat, compared to the catalysts with low Cu loading and the same support composition. The morphology of the Cu(5)/ZnO–CeO2(y) catalysts was studied by TEM. Figure shows that (i) all the samples consist of homogeneously distributed nanoparticles and (ii) the Cu/CeO2 sample has a polyhedral-like morphology, whereas the Zn-containing samples are composed of rounder particles. It is also shown that the average particle size of Cu/ZnO sample (9.1 nm) was substantially larger than the size of the Ce-containing samples (6.3 to 7.3 nm), in line with the specific surface area data.

Table 1

Physicochemical Properties of Cu(x)/ZnO–CeO2(y) Catalysts

catalyst	Cu loadinga (wt %)	Zn/(Zn+Ce) atomic ratioa	Zn/Cu atomic ratioa	SBETb (m2/gcat)	Dcuc (%)	SCu(0)c (m2/gcat)	NCu(0)c (μmol/gcat)	NOvc (μmol/gcat)	RH2/Cu ratiod
Cu(5)/CeO2	4.8	0.00	0.00	130	62.4	19.4	472	174	1.7
Cu(5)/ZnO–CeO2(0.01)	4.5	0.01	0.08	116	64.6	18.8	458	143	1.6
Cu(5)/ZnO–CeO2(0.05)	4.7	0.05	0.37	128	61.1	18.5	452	206	1.6
Cu(5)/ZnO–CeO2(0.10)	4.8	0.10	0.74	115	55.6	17.2	420	208	1.4
Cu(5)/ZnO–CeO2(0.25)	4.7	0.26	2.17	104	52.5	15.9	388	203	1.6
Cu(5)/ZnO–CeO2 (0.50)	4.7	0.51	4.97	112	46.8	14.2	347	132	1.5
Cu(5)/ZnO–CeO2(0.75)	4.8	0.76	9.29	110	30.2	9.4	228	111	1.2
Cu(5)/ZnO	5.0	1.00	15.11	82	10.0	3.2	79	7	0.9
Cu(45)/CeO2	38.9	0.00	0.00	91	6.8	17.2	419	138	1.1
Cu(45)/ZnO–CeO2(0.25)	39.8	0.24	0.10	94	7.4	18.9	461	209	1.1
Cu(45)/ZnO–CeO2(0.50)	40.1	0.51	0.25	93	7.4	19.1	466	171	1.1
Cu(45)/ZnO–CeO2(0.75)	39.6	0.76	0.46	95	6.6	17.0	414	155	1.1
Cu(45)/ZnO	39.9	1.00	0.79	68	3.8	9.8	240	51	1.0
commercial MSC	45.6	1.00	0.37	89	6.5	19.0	464	67	1.0

Determined from ICP-OES.

Determined from N2 physisorption.

Determined from N2O titration.

Determined from H2-TPR.

Figure 1

TEM images of FSP-prepared Cu(5)/ZnO–CeO2(y) catalysts with corresponding average particle size estimations.

The Cu–Zn–Ce oxide catalysts with low Cu loading were stuided by XRD (Figure a). The Cu/CeO2 and Cu/ZnO end members of this series of catalysts exhibited XRD patterns coresponding to the fluorite and wurtzite structures of CeO2 (PDF No. 00-004-0593) and ZnO (PDF No. 00-065-3411), respectively. The broadening of the diffraction peaks points to the nanocrystalline nature of the support material (CeO2 or ZnO). All of the Ce-containing samples display clear diffraction peaks due to CeO2. Zn addition led to a small shift of the CeO2 (111) diffraction peaks to lower angle compared to Cu(5)/CeO2 (Figure S2), pointing to the expansion of the CeO2 lattice. As substitution of Ce4+ (0.97 Å) by the smaller Zn2+ (0.90 Å) would lead to a lattice contraction,[46] the observation of the opposite can be explained by the formation of reduced Ce sites. This does not exclude that a small part of copper is also included in CeO2. Such substitution does not necessarily lead to a lattice parameter shift.[47] Only for catalysts with a high Zn content, i.e., y ≥ 0.50, clear ZnO diffraction peaks can be observed. No diffraction peaks assignable to CuO (PDF No. 00-041-0254) or Cu2O (PDF No. 00-05-0667) are present in all the XRD patterns, indicating that the Cu phases in these catalysys are either well-dispersed or amorphous. The diffraction patterns of the Cu–Zn–Ce oxide catalysts with high Cu loading are shown in Figure b. Similar to the series of Cu(5)/ZnO–CeO2(y), CeO2 diffraction peaks are present in all the Ce-containing samples and ZnO diffraction peaks are only observed in the samples with high Zn content (y ≥ 0.50). Further, all the Cu(45)/ZnO–CeO2(y) catalysts display clear CuO diffraction peaks, which is expected given high Cu loading of these samples. Notably, the intensity of the CuO diffraction peaks are much weaker in the Ce-containing samples as compared to the Cu(45)/ZnO, indicating that the presence of CeO2 can enhance Cu dispersion.

Figure 2

XRD patterns of (a) Cu(5)/ZnO–CeO2(y) and (b) Cu(45)/ZnO–CeO2(y) catalysts.

H2-TPR was carried out to probe Cu-support interactions in the Cu–Zn–Ce oxide catalysts. Figure a shows that two distinct reduction peaks were observed in all the Ce-containing samples for the series of Cu(5)/ZnO–CeO2(y). These two reduction peaks (α and β) can be assigned, respectively, to dispersed Cu species strongly interacting with ceria support and a Cu–CeO2 solid solution.[48,49] The amount of H2 consumed was normalized to the Cu content (Table ). As the resulting values were higher than unity (corresponding to the reduction of CuO to Cu), we can conclude that the ceria support was also reducing, implying the formation of oxygen vacancies. Moreover, similar two-peak reduction profiles were observed for all of the Ce-containing samples, implying that Cu preferably interacts with CeO2 over ZnO in the ternary catalysts. It is also found that the α and β reduction peaks already shifted to higher temperatures after addition of a small amount of ZnO, which is likely related to the modification of CeO2 support (as indicated by the XRD data) or the decoration of Cu particles by Zn species. The H2-TPR profile of Cu(5)/ZnO sample shows an asymmetric reduction peak around 200 °C with a small low-temperature shoulder, which can be attributed to dispersed Cu species interacting with ZnO.[50] The Cu–Zn–Ce oxide catalysts with high Cu loading (x = 45) were also investigated by H2-TPR. Figure b shows that the α and β reduction peaks in the Ce-containing samples and the main reduction peak in the Cu(45)/ZnO sample shifted to higher temperature in comparison to the catalysts with the same support composition in Figure a. Such shifts can be explained by the fact that Cu particles are much larger at higher Cu loading (see below).

Figure 3

H2-TPR profiles of (a) Cu(5)/ZnO–CeO2(y) and (b) Cu(45)/ZnO–CeO2(y) catalysts.

We used N2O titration to quantify the metallic Cu surface sites in the reduced Cu–Zn–Ce oxide catalysts, which is a key descriptor for methanol synthesis for Cu-based catalysts.[51,52] The results are listed in Table . The Cu dispersion as a function of Zn content in the support is also plotted in Figure . For the series of Cu(5)/ZnO−CeO2(y), Figure a shows that the Cu dispersion gradually decreases from ∼60% for Cu(5)/CeO2 to ∼10% for Cu(5)/ZnO. The decreasing trend of Cu dispersion with respect to Zn content indicates that CeO2 interacts stronger with Cu than ZnO.[20,53] Notably, the metallic Cu surface areas remain almost the same (∼19 m2/gcat) after addition of a small amount of Zn (y ≤ 0.05). In addition to metallic Cu sites, the N2O titration data reveal that oxygen vacancies exist in the Ce-containing samples after H2 reduction (Table ). These findings agree with the quantitative H2-TPR quantification analysis. The Cu dispersion data of the Cu(45)/ZnO–CeO2(y) series are plotted in Figure b. Clearly, the higher Cu loading in these catalysts leads to a much lower Cu dispersion. It is also shown that (i) all of the Ce-containing catalysts display similar Cu dispersion and (ii) the Cu(45)/ZnO exhibits the lowest Cu dispersion (3.8%) in this series.

Figure 4

Cu dispersion of (a) Cu(5)/ZnO–CeO2(y) and (b) Cu(45)/ZnO–CeO2(y) catalysts determined from N2O titration.

STEM-EDX was used to investigate the elemental distribution (Cu, Zn and Ce) in the reduced Cu–Zn–Ce oxide catalysts (Figure ). Although these data do not warrant a statistical analysis of the distribution of Zn and Ce in the support, inspection of the STEM-EDX mapping results points to the segregation of these two components at the nanometer scale (Figure b). Moreover, Cu is homogeneously and finely dispersed in the reduced Cu(5)/CeO2 and Cu(5)/ZnO–CeO2(0.25). In contrast, larger Cu particles can be distinguished in the reduced Cu(5)/ZnO. Clearly, the presence of CeO2 in the support can increase Cu dispersion, in line with the N2O titration results. To summarize, we demonstrate that FSP is a suitable method to prepare catalysts containing dispersed Cu particles supported on (mixed) oxides of CeO2 and ZnO. The resulting catalysts display similar morphological and textural properties in a broad support compositional range. The Cu dispersion of FSP-prepared Cu–Zn–Ce oxide catalysts can be easily tuned by changing the Cu content and the Zn–Ce ratio in the precursor solutions.

Figure 5

STEM images with corresponding elemental mappings of the reduced (a) Cu(5)/CeO2, (b) Cu(5)/ZnO–CeO2(0.25), and (c) Cu(5)/ZnO. Reduction conditions: 250 °C, 1 h, and 10 vol % H2 in He.

The series of Cu(5)/ZnO−CeO2(y) catalysts were evaluated for CO2 hydrogenation under the conditions of 250 °C and 30 bar. Figure a shows how CO2 conversion and CH3OH selectivity change with Zn content in the catalyst support. The CO2 conversion is lowest for the Cu(5)/CeO2 and increases with Zn content until y = 0.05. At higher Zn content (0.05 ≤ y ≤ 0.75), the ternary Cu–Zn–Ce oxide catalysts display nearly the same CO2 conversion, while the Cu(5)/ZnO shows the highest CO2 conversion. Regarding product distribution, we observe that the ternary Cu–Ze–Ce oxide catalysts show a higher CH3OH selectivity compared to the binary Cu–Zn and Cu–Ce oxide catalysts. Figure b reports how CH3OH and CO formation rates vary as a function of Zn content in the catalyst support. The CH3OH formation rate increases more than twice after adding a small amount of ZnO (y ≤ 0.05). As the Cu dispersion is nearly the same for these catalysts, the improved CH3OH formation rate points to a clear Zn promotion effect on Cu surface for CH3OH synthesis from CO2.[54] At higher Zn content, the CH3OH formation rate remains nearly unchanged, suggesting that a maximum Zn promotion is already achieved at a relatively low Zn content (y = 0.05). Regarding the formation of CO, the activity is similar for all the Ce-containing catalysts, while the corresponding rate of the Cu(5)/ZnO is nearly two times higher. These observations indicate that CeO2 in these ternary catalysts can inhibit the rWGS reaction. As it has been suggested that CO can be a reaction intermediate for CH3OH synthesis from CO2 hydrogenation,[55] we also evaluated the activity of the Cu(5)/ZnO–CeO2(y) catalysts in CO hydrogenation under the same conditions (Figure S4). The results show that the CH3OH activity in CO hydrogenation decreased after adding a small amount of ZnO (y ≤ 0.05) and remained nearly the same at higher ZnO content (0.10 ≤ y ≤ 0.75). Notably, all the Ce-containing catalysts displayed a significantly higher CH3OH formation rate than Cu(5)/ZnO.

Figure 6

(a) CO2 conversion and CH3OH selectivity; (b) CH3OH and CO formation rates as a function of support composition for Cu(5)/ZnO–CeO2(y) catalysts. Reaction conditions: 250 °C, 30 bar, H2/CO2/N2 = 3:1:1, and SV = 60 L/(gcat × h).

The Cu–Zn–Ce oxide catalysts with higher Cu loading (x = 45) were also tested in CO2 hydrogenation. Similar trends of CO2 conversion, CH3OH selectivity and product formation rates as a function of Zn content were observed for this series of catalysts, irrespective of the substantial difference in Cu loading compared to Cu(5)/ZnO–CeO2(y). Specifically, Figure a shows that (i) the Cu(45)/CeO2 and Cu(45)/ZnO, respectively, have the lowest and highest CO2 conversion and (ii) the ternary Cu–Zn–Ce oxide catalysts (y = 0.25, 0.50 and 0.75) have similar CO2 conversion and higher CH3OH selectivity than the corresponding binary catalysts. Figure b shows that addition of a small amount of ZnO (y = 0.25) leads to a significant increase in CH3OH activity and a further increase in Zn content has a minor influence on CH3OH activity. Addition of a large amount of ZnO (y = 0.75 and 1.00) resulted in a significant increase of CO activity. Besides these similarities, it was observed that the maximum Zn promotion for CH3OH formation in Cu(45)/ZnO–CeO2(y) catalysts (2.7 → 19.1 mmol/(gcat × h)) is significantly higher than that in Cu(5)/ZnO–CeO2(y) catalysts (2.0 → 6.6 mmol/(gcat × h)). This difference implies that the Zn promotion for CH3OH synthesis from CO2 is dependent on Cu particle size.

Figure 7

(a) CO2 conversion and CH3OH selectivity; (b) CH3OH and CO formation rates as a function of support composition for Cu(45)/ZnO–CeO2(y) catalysts. Reaction conditions: 250 °C, 30 bar, H2/CO2/N2 = 3:1:1, and SV = 120 L/(gcat × h).

We then compared the performance of a selected Cu–Zn–Ce oxide catalyst, i.e., Cu(45)/ZnO–CeO2(0.25) to a commercial MSC in CO2 hydrogenation to CH3OH. Figure shows that the Cu–Zn–Ce oxide catalyst displays a higher initial CH3OH selectivity than the commercial MSC at similar CO2 conversion (∼6%). Upon increasing time on stream (TOS), the CH3OH selectivity of the Cu–Zn–Ce oxide catalyst slowly increased at the expense of CO2 conversion. A similar trade-off between CH3OH selectivity and CO2 conversion has been earlier reported for Cu-based catalysts and can be attributed either to product inhibition by H2O and/or CH3OH or to CH3OH decomposition.[56,57] After 12 h on stream, the CH3OH selectivity of the Cu–Zn–Ce oxide catalyst was ∼16% higher than the commercial MSC. A catalyst stability test of 100 h (Figure S8) shows that the deactivation of Cu(45)/ZnO–CeO2(0.25) was faster in the beginning (first 30 h) than after prolonged reaction. Notably, the CO formation rate decreased more substantially with time on stream than the CH3OH formation rate.

Figure 8

Time-on-stream CO2 conversion and CH3OH selectivity comparison between Cu(45)/ZnO–CeO2(0.25) and commercial methanol synthesis catalysts. Reaction conditions: 250 °C, 30 bar, H2/CO2/N2 = 30:10:10 (mL/min).

In Situ IR Spectroscopy

CO adsorption was carried out to characterize the Cu speciation in the reduced Cu–Zn–Ce oxide catalysts. For the Cu/CeO2 sample, we observed that the carbonyl band at 2068 cm–1 strongly blue-shifted with increasing CO partial pressure (Figure a). The shift is due to dipole–dipole coupling of CO molecules adsorbed on metallic Cu sites.[58] The presence of a weak broad band around 1960 cm–1 due to bridge-bound carbonyls on metallic Cu confirms this assignment. The low-frequency tailing of the band can be explained by the presence of a range of Cu particle sizes with different extents of dipole–dipole coupling interactions. The spectra also feature a strong band at 2109 cm–1, which can be assigned to interfacial Cu+ sites as reported before for reduced Cu/CeO2.[59−61] The corresponding CO IR spectra of Cu/ZnO in Figure b show a much lower intensity of carbonyl bands, which is in keeping with the lower Cu dispersion. Similar spectra have been reported in literature for Cu/ZnO.[62−64] The most likely interpretation is that the 2087 cm–1 band is due to carbonyl on metallic Cu and the broad band at lower frequency to Zn-decorated Cu sites.[65] The CO IR spectra of Cu(5)/ZnO–CeO2(0.05) resemble the one of Cu/ZnO with a substantially higher intensity (Figure c). Accordingly, we assign these spectra to similar Zn-decorated Cu sites as in Cu/ZnO but with a much higher dispersion. Notably, the carbonyl band due to Cu+ sites is absent in the reduced ternary catalyst. Although this may indicate that the Cu–CeO2 interface is lost, we tentatively assign the difference with Cu/CeO2 to Zn decoration of the Cu surface.

Figure 9

IR spectra of CO adsorbed on reduced (a) Cu(5)/CeO2, (b) Cu(5)/ZnO, and (c) Cu(5)/ZnO–CeO2(0.05) (Tads = 40 °C with increasing CO partial pressure from orange to blue with a maximum CO partial pressure of 10 mbar).

We also investigated the surface of the reduced Cu–Zn–Ce oxide catalysts by IR spectroscopy during temperature-programmed CO2 hydrogenation. The C–H vibration and carbonyl regions are displayed in Figure . For all of the catalysts, formate was present on the Cu surface (HCOO-Cu) with increasing temperature to 250 °C as evidenced by the band at 2849 cm–1.[52,66−68] The band at 2876 cm–1 in the C–H vibration region observed for Cu/ZnO can be attributed to formate adsorbed on ZnO (HCOO-Zn).[68,69] Notably, formates were formed at lower temperature on Cu/ZnO compared to the other two catalysts. A comparison of Cu(5)/ZnO–CeO2(0.05) and Cu(5)/CeO2 shows that ZnO addition led to a carbonyl band at 2120 cm–1 in the spectrum at 50 °C. The Cu/ZnO sample also contains a weaker carbonyl band at 2104 cm–1. This carbonyl band is less stable on Cu/ZnO than on Cu(5)/ZnO–CeO2(0.05) as can be appreciated by comparison of spectra with increasing temperature (Figure b,c). Clearly, the appearance of carbonyl bands for the two Zn-containing catalysts points to facile CO2 activation on Zn-decorated Cu. Furthermore, the carbonyl bands observed during CO2 hydrogenation at 50 °C were substantially blue-shifted in comparison to those recorded upon CO adsorption (Figure ). We also observed that the carbonyl band on Cu(5)/ZnO–CeO2(0.05) was located at a higher wavenumber (2120 cm–1) than on Cu/ZnO (2104 cm–1). Such difference can be explained by electron depletion of the Cu surface, which may be due to the high coverage of oxygen caused by CO2 dissociation at this temperature. A similar effect was also observed by comparing the spectra recorded at 250 °C where formate is the electro-withdrawing adsorbate.[70−72] It was found that the intensity of the HCOO-Cu band on Cu(5)/CeO2 and Cu(5)/ZnO–CeO2(0.05) were much higher than on Cu(5)/ZnO, indicating a higher formate coverage on the Ce-containing catalysts.

Figure 10

IR spectra during CO2 hydrogenation over (a) Cu(5)/CeO2, (b) Cu(5)/ZnO–CeO2(0.05), and (c) Cu(5)/ZnO. IR spectra at reaction temperature of 50 and 250 °C over (d) Cu(5)/CeO2, (e) Cu(5)/ZnO–CeO2(0.05), and (f) Cu(5)/ZnO. Reaction conditions: CO2 5%, H2 15%, He 80%, total flow of 200 mL/min, and 50–250 °C.

Discussion

A key finding from the results of this work is that the combination of Cu with ZnO–CeO2 supports results in a higher CH3OH selectivity in CO2 hydrogenation in comparison with binary Cu/ZnO and Cu/CeO2 catalysts. The role of CeO2 oxygen vacancies in CO2 conversion was examined because such vacancies are known to play an important role in many ceria-based catalysts.[73,74] For this purpose, we plotted CH3OH and CO formation rates against the amount of oxygen vacancy estimated by H2-TPR and N2O titration measurements (Figure S9). The absence of strong correlations indicates that oxygen vacancies in CeO2 support are not the key descriptor for CO2 hydrogenation to CH3OH or CO. The blocking of oxygen vacancies by in situ formed carbonates may explain their limited role in CO2 hydrogenation.[21,45,75]

We next discuss the improved CH3OH selectivity in the context of Cu–ZnO and Cu–CeO2 interactions. Scheme presents three simplified models for the Cu/CeO2, Cu/ZnO–CeO2, and Cu/ZnO catalysts investigated in this work. To facilitate the discussion, turnover frequencies (TOFs) for CH3OH and CO formation are plotted against the Cu particle sizes estimated by N2O titration (Figure S10). On the one hand, the improved CH3OH selectivity is clearly related to Zn addition to Cu/CeO2, which promotes CH3OH reaction rate as shown in the catalytic data. We found that such Zn promotion already reaches its full extent at a relatively low Zn content, e.g., in Cu(5)/ZnO–CeO2(0.05). As diffraction peaks due to bulk ZnO are absent in the XRD pattern of this catalyst, the Zn promotion of CH3OH synthesis can be correlated to highly dispersed Zn species on Cu as also suggested in an earlier study.[76] A comparison of CO IR spectra (Figure ) shows that the ternary catalysts are not a physical mixture of Cu/CeO2 and Cu/ZnO. The addition of ZnO to Cu/CeO2 results in the formation of Zn-decorated Cu sites upon reduction of Cu(5)/ZnO–CeO2(0.05) as also observed for Cu(5)/ZnO. The formation of Zn-decorated metallic Cu in Cu/ZnO as active sites for CH3OH synthesis has been well documented,[2,54,77] with main propositions for the formation of ZnO moieties on Cu surface and a CuZn surface alloy. Comparing the CO IR spectra of our catalysts to those reported for Cu/ZnO by Topsoe et al.,[62,64] it is more likely that Cu is modified by ZnO in our catalysts. This is consistent with the relatively low reduction temperature employed. On the other hand, CeO2 addition inhibits the rWGS reaction as can be inferred from Figure S10b: all the Ce-containing catalysts display significantly lower TOF(CO) compared to Cu/ZnO. In situ IR spectra recorded at 250 °C during CO2 hydrogenation indicate that the formate coverage on Cu is higher on Ce-containing catalysts than on Cu/ZnO. We propose that the ceria-induced inhibition of the rWGS is associated with the higher formate coverage, which blocks metallic Cu sites active in the rWGS as proposed before.[45,78−81] Besides the blocking of Cu surface by formate, in situ CO-to-CH3OH conversion during CO2 hydrogenation can also contribute to the ceria-induced inhibition of the rWGS. It has been reported that Cu–CeO2 interface is highly active for CO hydrogenation to CH3OH.[21,22] Our catalytic data (Figure S4) also show that the Ce-containing catalysts showed a substantially higher CH3OH formation rate in CO hydrogenation than Cu/ZnO catalyst. As such, we propose that the ceria-induced rWGS inhibition can also be due to the fact that the in situ formed CO during CO2 hydrogenation can be faster hydrogenated to CH3OH at Cu–CeO2 interface in Ce-containing catalysts than Cu/ZnO.

Scheme 1

Proposed Models of FSP-Prepared Cu/CeO2, Cu/ZnO–CeO2, and Cu/ZnO Catalysts

Conclusions

Well-defined high-surface-area Cu–Zn–Ce oxide catalysts were prepared by a one-step FSP method. The Cu loading and support composition were systematically varied to investigate how Cu-support interactions affect CO2 hydrogenation. CeO2 interacts more strongly with Cu than ZnO, leading to a better Cu dispersion in the presence of CeO2. CO2 hydrogenation results show that (i) the Cu–Zn–Ce oxide catalysts display a significantly higher CH3OH selectivity in comparison to the binary Cu–Zn and Cu–Ce catalysts and (ii) the Cu–Zn–Ce oxide catalyst with a high Cu loading (∼40 wt %) outperforms a commercial methanol synthesis catalyst in terms of CH3OH selectivity. Evaluation of the relation between the structure and the catalytic performance reveals that the improved CH3OH selectivity of Cu–Zn–Ce oxide catalysts can be explained by Zn promotion of the Cu surface for CO2 hydrogenation to CH3OH and by ceria-induced inhibition of the rWGS reaction. Based on in situ IR studies and other characterization results, we propose that (i) the promotion of CH3OH synthesis from CO2 is due to the formation of Zn-decorated Cu sites and (ii) the ceria-induced inhibition of the rWGS can be explained by blocking of the Cu surface by formate species and a higher rate of hydrogenation of the CO intermediate. This study also demonstrates that FSP is a promising method to prepare multicomponent methanol synthesis catalysts. The use of different metal oxide promoters that affect Cu-support interactions can help to improve the overall performance of methanol synthesis catalysts in CO2 hydrogenation.