Sulfate residuals on Ru catalysts switch CO2 reduction from methanation to reverse water-gas shift reaction Authors: Min Chen, Longgang Liu, Xueyan Chen, Xiaoxiao Qin, Jianghao Zhang, Shaohua Xie, Fudong Liu, Hong He, Changbin Zhang Published: 2024 ABSTRACT Efficient heterogeneous catalyst design primarily focuses on engineering the active sites or supports, often neglecting the impact of trace impurities on catalytic performance. Herein, we demonstrate that even trace amounts of sulfate (SO42−) residuals on Ru/TiO2 can totally change the CO2 reduction from methanation to reverse-water gas shift (RWGS) reaction under atmospheric pressure. We reveal that air annealing causes the trace amount of SO42− to migrate from TiO2 to Ru/TiO2 interface, leading to the significant changes in product selectivity from CH4 to CO. Detailed characterizations and DFT calculations show that the sulfate at Ru/TiO2 interface notably enhances the H transfer from Ru particles to the TiO2 support, weakening the CO intermediate activation on Ru particles and inhibiting the further hydrogenation of CO to CH4. This discovery highlights the vital role of trace impurities in CO2 hydrogenation reaction, and also provides broad implications for the design and development of more efficient and selective heterogeneous catalysts. Introduction At present, the atmospheric CO2 level has surged to a historically high of approximately 416 ppm, further reinforcing the existing concerns about its significant contribution to global climate change1–4. Long treated as waste, CO2 is now considered as a potentially useful carbon source for producing fuels and chemicals through photocatalytic, electrocatalytic, and thermal catalytic reduction in the presence of H2, which can be obtained from water splitting using solar, wind, or other renewable energy sources5–8. Thermal catalytic reduction typically offers high reaction efficiency and is extensively used in practical applications. By designing supported metal catalysts and controlling the reaction conditions, a wide range of products, such as methane (CH4)9–11, carbon monoxide (CO)12,13, methanol (CH3OH)14,15, and even long-chain hydrocarbons can be obtained16. Catalytic hydrogenation of CO2 at atmospheric pressure typically involves either methanation reaction, yielding CH4, or reverse water-gas shift (RWGS) reaction, leading to CO production1,2,16–19. Both methanation and RWGS reactions play crucial roles in industrial processes related to hydrogen utilization and synthesis gas production. CO2 hydrogenation to CH4 or CO with high selectivity is desired according to specific application requirements but remains challenging. Previous research has established that the supported Ru/TiO2 catalysts are one of the most active and stable catalysts in CO2 hydrogenation reaction20–22. Very recently, some reports have highlighted that the crystal structure of TiO2 support significantly impacted the selectivity of CO2 hydrogenation on Ru/TiO2 catalysts18,23–25. Qiao at al. observed that the selectivity of CO2 hydrogenation could be completely reversed when Ru particles were supported on anatase-TiO2 (high CO selectivity) versus on rutile-TiO2 (high CH4 selectivity)23. This phenomenon was attributed to the different electron transfer processes from Ru to the TiO2 supports as a result of varying extents of hydrogen spillover related to crystal structure23. Wang et al. reported that annealing Ru/rutile-TiO2 in air enhanced the CO2 conversion to CH4, while annealing Ru/anatase-TiO2 in air decreased the CO2 conversion and led to CO production. They ascribed these differences to the contrasting metal-support interaction between Ru and anatase or rutile18. In contrast, Debecker et al. found that Ru supported on anatase, rutile, or a mixture of the two exhibited a variety of CO2 conversions, but high CH4 selectivity was observed on all catalysts24. The disparity in the observed CO or CH4 selectivity among researchers indicates that the crystal structure of TiO2 supports may not be the sole determining factor for the catalytic performance of CO2 hydrogenation. Upon careful examination of the literatures mentioned above, we noticed that the TiO2 supports used in these studies were usually obtained from commercial sources, with some samples possibly containing trace amounts of residual impurities. Some impurities might remarkably affect the catalytic performance of Ru catalysts in CO2 hydrogenation; however, their distinct significance was often overlooked in the course of research, potentially leading to flawed conclusions. In this work, we observed that the typical Ru/TiO2 catalysts using both anatase and rutile supports displayed excellent performance in CO2 methanation reaction. Surprisingly, the Ru/TiO2 catalysts containing trace amount of SO42− residuals showed no activity in CO2 methanation, but excellent activity in RWGS reaction. This unique phenomenon suggested that the SO42− residuals on these Ru catalysts, rather than the crystal structure of the TiO2 supports, plays the key role in determining the catalytic performance of CO2 hydrogenation. Further investigation revealed that annealing the sulfate-containing Ru/TiO2 in air induced the sulfate migration from the TiO2 support to the Ru/TiO2 interface. At the interface, the sulfate could strongly promote the transfer of hydrogen from Ru particles to the TiO2 support. The enhanced hydrogen spillover weakened the activation of CO intermediates on Ru particles, leading to significantly higher selectivity for CO production. This work not only introduces a novel viewpoint for elucidating the variation of observed CO or CH4 selectivity of CO2 hydrogenation on Ru/TiO2 catalysts, but also leads to a fundamental guideline for new catalyst design including the careful control of impurity levels and exploiting their positive impacts. Results The effects of sulfate on the catalytic performance of Ru/TiO2 for CO2 hydrogenation A set of Ru/TiO2 catalysts (with anatase TiO2 purchased from Aldrich or Aladdin) were prepared by a wet impregnation method, with the Ru loading of 5 wt.%. The catalytic reduction of CO2 was conducted at atmospheric pressure within the temperature range from 200 to 450 °C, in a fixed-bed flow reactor with a gas mixture composed of CO2 (10 vol.%), H2 (40 vol.%), and N2 balance. The gas weight hourly space velocity (WHSV) was approximately 48,000 mL·g−1·h−1. The CO2 hydrogenation under these conditions typically yields CH4 via CO2 methanation reaction and CO via RWGS reaction. Moreover, CO2 methanation is more thermodynamically favorable compared to the RWGS reaction when the reaction temperature is below 500 °C26. Our calculation results about the thermodynamic equilibrium of CO2 hydrogenation also showed that CH4 was the favored product at lower temperature (<500 °C), while CO was the favored product at higher temperature (>500 °C) (Supplementary Figs. 1, 2). Thus, it remains an ongoing challenge to tune the high CH4 selectivity to high CO selectivity at lower temperatures. Figure 1a illustrates the products comparison on different sets of Ru/TiO2 catalysts (the detailed comparison is provided in Supplementary Fig. 3). CH4 was the main product on some Ru/TiO2 catalysts, while CO was the main product on other Ru/TiO2 catalysts. Considering that the purchased anatase TiO2 supports may contain trace amount of sulfate species as impurity, we conducted the element analysis by inductively coupled plasma mass spectrometry (ICP-MS, see Supplementary Table 1). The results revealed that these TiO2 supports could be divided into two groups, with one group showing nearly no presence of SO42− (i.e., very low S content of 0.01–0.03 wt.%) and the other group showing the presence of trace amount of SO42− with relatively higher S content of 0.1–0.7 wt.%. Surprisingly, it was observed that the Ru/TiO2 catalysts with no SO42− displayed high CH4 selectivity, while the Ru/TiO2 catalysts with trace amount of SO42− displayed high CO selectivity. These results strongly indicate that the presence of trace amount of SO42− on Ru/TiO2 may play a crucial role in impacting the selectivity of CO2 hydrogenation reaction.Fig. 1Catalytic performance of the Ru/TiO2 catalysts.a The products and sulfur content comparison on the different sets of Ru/TiO2 catalysts for CO2 hydrogenation. b Temperature-dependent CO2 conversion and CH4 selectivity of Ru/TiO2 catalysts with or without SO42− species. c Comparison with commercial Ru/Al2O3 catalyst for CH4 productivity at 350 °C and commercial CuO/ZnO/Al2O3 catalyst for CO productivity at 410 °C. d The product selectivity on Ru/Ti-S catalyst with air and/or H2 pretreatment at 350 °C. Catalytic performance of the Ru/TiO2 catalysts. a The products and sulfur content comparison on the different sets of Ru/TiO2 catalysts for CO2 hydrogenation. b Temperature-dependent CO2 conversion and CH4 selectivity of Ru/TiO2 catalysts with or without SO42− species. c Comparison with commercial Ru/Al2O3 catalyst for CH4 productivity at 350 °C and commercial CuO/ZnO/Al2O3 catalyst for CO productivity at 410 °C. d The product selectivity on Ru/Ti-S catalyst with air and/or H2 pretreatment at 350 °C. To further investigate the influence of SO42− on the catalytic performance of CO2 hydrogenation, we prepared the sulfate-free Ru/TiO2 catalysts, in which the sulfate-free TiO2 were synthesized by hydrolyzing tetrabutyl titanate, and also prepared the Ru/TiO2 catalysts containing sulfate by purposely adding ammonium sulfate during the preparation process (with mole ratio of S/Ru set as 0, 0.03, 0.05, and 0.1). Before testing and characterization, the obtained sulfate-free and sulfated Ru/TiO2 catalysts were annealed in air at 400 °C and then reduced by H2 at 500 °C (denoted as Ru/Ti-AR and Ru/Ti-S-AR, Ru/Ti = Ru supported on anatase TiO2, S = sulfated, AR = air annealing and H2 reduction). Figure 1b shows that the Ru/Ti-AR exhibited excellent activity for CO2 hydrogenation between 150 and 410 °C, and the CO2 conversion reached the highest of ca. 89% at 330 °C and slightly decreased to 83% at 410 °C. The CH4 selectivity maintained above 95% within this temperature range. With increasing the mole ratio of S/Ru from 0 to 0.1 (note that the ratio of S/Ru in Ru/Ti-S-AR was 0.1), the CO2 conversion dropped sharply and the product distribution dramatically changed from CH4 to CO (Supplementary Fig. 4). Besides, we tested the activity of the Ru/Ti-S-AR by altering contact time (Supplementary Fig. 5). The results showed that the CO2 conversion was enhanced by increasing contact time, but high CO selectivity still remained, further confirming the high CO selectivity on Ru/Ti-S-AR. Compared with the commercial Ru/Al2O3 catalyst for CO2 methanation and CuO/ZnO/Al2O3 catalyst for RWGS reaction (Fig. 1c), the CH4 production on Ru/Ti-AR at 350 °C was 348 mmol g−1 h−1, which was 1.4 times higher than that on Ru/Al2O3. The CO production on Ru/Ti-S-AR at 410 °C was 160 mmol g−1 h−1, which was close to that on Cu/Zn/Al (191 mmol g−1 h−1). These results suggest that the Ru/Ti-AR performed as an efficient catalyst for CO2 methanation reaction, while the Ru/Ti-S-AR performed as an efficient catalyst for RWGS reaction. Considering that the crystal structure of TiO2 was frequently discussed in influencing the product selectivity on Ru/TiO2 catalyst in CO2 hydrogenation reaction, we also synthesized a series of Ru/rutile catalysts, including the sulfate-free Ru/rutile-AR and Ru/rutile-R, as well as the sulfate-containing Ru/rutile-S-AR and Ru/rutile-S-R (AR = air annealing and H2 reduction, S = sulfated, R = direct H2 reduction). Upon analyzing the testing results depicted in Supplementary Fig. 6, we observed the similar trends as that on Ru/Ti catalysts. In short summary, both the sulfate-free Ru/TiO2 catalysts (using anatase or rutile as support) showed high CH4 selectivity, while the sulfate-modified Ru/TiO2 showed high CO selectivity. These results emphasized that the presence of sulfate residuals in TiO2 support, rather than the crystal structure of TiO2, was the key factor influencing the catalytic performance of CO2 hydrogenation. Pretreating catalysts under different atmospheres commonly impacted the performance of catalysts in many reactions. Figure 1d shows that the high CO selectivity on Ru/Ti-S-AR was observed at 350 °C by annealing the as-prepared Ru/Ti-S in air followed by H2 reduction. The pretreatment condition was also switched to direct H2 reduction without pre-annealing in air. Surprisingly, the high CH4 selectivity was observed (Fig. 1d). Afterwards, this sample was further treated with air annealing and subsequent H2 reduction, and the product selectivity could achieve high CO (see detailed activity results as shown in Supplementary Fig. 7). In contrast, when the as-prepared sulfate-free Ru/Ti was pretreated under similar conditions, no such switch of product selectivity was observed at all (Supplementary Fig. 8). These findings suggest that when there was trace amount of sulfate species on Ru/TiO2 catalysts, annealing the catalysts in air was highly crucial for regulating the product selectivity in CO2 hydrogenation. Sulfate-induced structural modification of Ru/TiO2 To understand the effect of sulfates on the structure of Ru/TiO2 catalysts, we investigated the geometric states of Ru nanoparticles (NPs) on different samples. The Ru/TiO2 catalysts were prepared using the traditional wet impregnation method, which typically resulted in a wide range of metal dispersion on the support. High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) measurement of the Ru/Ti-S-AR catalyst revealed numerous Ru particles distributed on the TiO2 support, with sizes ranging from 5 to 12 nm (average size of 5.7 nm) (Fig. 2a and Supplementary Fig. 9). The Ru/Ti-AR exhibited a similar size distribution of Ru particles to that of Ru/Ti-S-AR. As shown in Supplementary Fig. 10, the Ru size distribution on Ru/Ti-R was in the range of 1–7 nm, with an average size of 2.9 nm, which was also comparable to that on Ru/Ti-S-R. The sizes of Ru derived from HAADF-STEM were in good agreement with XRD data (Supplementary Fig. 12). A summary about the Ru particles size and dispersion was shown in Supplementary Table 2. These results suggested that the air annealing at high temperature followed by H2 reduction led to a higher degree of Ru particle aggregation compared with direct H2 reduction treatment, and the presence of sulfate had negligible influence on the Ru particle size distribution.Fig. 2The geometric states of Ru NPs on TiO2.a HAADF-STEM image of Ru/Ti-S-AR. b EDX mapping images of Ru and S elements on Ru/Ti-S-AR. c HAADF-STEM image of Ru/Ti-S-R. d EDX mapping images of Ru and S elements on Ru/Ti-S-R. e Schematic illustration of the evolution of Ru and S species on TiO2 during air-H2 or direct H2 treatment. The geometric states of Ru NPs on TiO2. a HAADF-STEM image of Ru/Ti-S-AR. b EDX mapping images of Ru and S elements on Ru/Ti-S-AR. c HAADF-STEM image of Ru/Ti-S-R. d EDX mapping images of Ru and S elements on Ru/Ti-S-R. e Schematic illustration of the evolution of Ru and S species on TiO2 during air-H2 or direct H2 treatment. The distribution of sulfate species on Ru/Ti-S-AR and Ru/Ti-S-R was investigated using energy dispersive X-ray (EDX) mapping. As shown in Fig. 2b, d, the S element on Ru/Ti-S-AR tended to accumulate near Ru particles, while the sulfates on Ru/Ti-S-R were randomly distributed on Ru/Ti-S-R. The distribution pattern of Ru and S elements on Ru/Ti-S-R-AR was similar to that on Ru/Ti-S-AR (Supplementary Fig. 11). This distinct relationship between Ru and S element distribution suggested that the air annealing could effectively drive the migration of sulfates on TiO2 to the Ru/TiO2 interface. In general, the sulfate species are bonded with TiO2 through the chemical bonding of Ti-S. However, when the Ru species are loaded on TiO2, a stronger chemical bonding of Ru-S may be present. During the air annealing at high temperatures, the surface sulfates on TiO2 likely migrated to the Ru/TiO2 interface to form a more stable state with the stronger chemical bonding of Ru-S (as illustrated in Fig. 2e). The spatial distribution of sulfates differed noticeably between Ru/Ti-S-R and Ru/Ti-S-AR, implying that the chemical states of sulfate species were also expected to be distinct. To prove this point-of-view, we performed X-ray photoelectron spectroscopy (XPS) measurement. The S 2p1/2 spectra between binding energies of 160–172 eV are shown in Fig. 3a. For both Ru/Ti-S-AR and Ru/Ti-S-R, the doublet peaks of S 2p1/2 were observed at 169.4 and 168.2 eV, corresponding to the presence of sulfate ions (SO42−)27. In addition, for Ru/Ti-S-AR, the peaks at 161.5 and 162.5 eV were detected, indicating the presence of S2−28, while these two peaks were considerably weaker for Ru/Ti-S-R. The presence of SO42− could be attributed to the introduction of (NH4)2SO4 during the preparation process. The SO42− bonded on the surface of TiO2, while the appearance of S2− was indicative of the formation of Ru-S bonds, confirming that the air annealing indeed facilitated the migration of a certain amount of sulfate from the TiO2 support to the Ru/TiO2 interface.Fig. 3Characterization on the chemical states of S and Ru species in different catalysts.a S 2p XPS for Ru/Ti-S-AR, Ru/Ti-S-R, Ru/Ti-AR and Ru/Ti-R. b H2-TPR profiles of Ru/Ti, Ru/Ti-A, Ru/Ti-S and Ru/Ti-S-A. c CO-DRIFTS on different catalysts at 25 °C, probing the surface states of Ru nanoparticles. Characterization on the chemical states of S and Ru species in different catalysts. a S 2p XPS for Ru/Ti-S-AR, Ru/Ti-S-R, Ru/Ti-AR and Ru/Ti-R. b H2-TPR profiles of Ru/Ti, Ru/Ti-A, Ru/Ti-S and Ru/Ti-S-A. c CO-DRIFTS on different catalysts at 25 °C, probing the surface states of Ru nanoparticles. H2-temperature-programmed reduction (H2-TPR) experiments were performed to investigate the reducibility of catalysts and the interactions at the metal-support interfaces. The H2-TPR results (Fig. 3b) showed that all the samples exhibited two main reduction peaks, centered at 95–103 °C and 128–131 °C, corresponding to the reduction of surface RuOx species weakly and strongly interacting with TiO2, respectively25,29,30. The surface RuOx species were typically reduced by H2 easily, displaying peaks at lower temperatures (95–103 °C), whereas the interfacial RuOx required higher temperatures and exhibited H2 reduction peaks at 128–131 °C. Furthermore, the ratio of interfacial RuOx to surface RuOx species in the Ru/Ti-S-R and Ru/Ti-R samples was noticeably higher than in the Ru/Ti-S-AR and Ru/Ti-AR samples. This observation could be attributed to the size difference of the Ru particles. The HAADF-STEM results showed that the Ru particles in the Ru/Ti-S-R and Ru/Ti-R samples were smaller compared to those in the Ru/Ti-S-AR and Ru/Ti-AR samples. Smaller Ru particles typically exhibited more interfacial RuOx species on TiO2, explaining the more abundant interfacial RuOx in the Ru/Ti-S-R and Ru/Ti-R samples. Notably, the Ru/Ti-S-AR sample also displayed two additional peaks at 230 and 322 °C, which were not observed on other samples. These peaks could be attributed to the reduction of RuSx30,31, formed due to the migration of sulfate to the Ru/TiO2 interface during air annealing. This phenomenon resulted in a strong interaction between Ru and sulfate, leading to the presence of RuSx peaks in H2-TPR profile for Ru/Ti-S-AR sample. To further investigate how the sulfate species induced the structural modification of Ru/TiO2 catalyst, the chemical states of Ru were also characterized. The XPS results of Ru 3d for both Ru/Ti-AR and Ru/Ti-S-AR (Supplementary Fig. 14) revealed that the Ru species in the samples with or without sulfate were both in metallic state. Specifically, the presence of sulfate induced a slight shift of the metallic Ru peak from 279.7 to 279.9 eV, which might be due to the formation of Ru-S bonds at the Ru/TiO2 interface32. For the impact of sulfate species on the structure TiO2, no obvious changes were observed in Ti 2p on Ru/Ti-AR and Ru/Ti-S-AR (Supplementary Fig. 15), indicating that the presence of trace amount of sulfate species had negligible influence on the TiO2 support, which was in line with the XRD and Raman results (Supplementary Figs. 12, 13). To gain further insights into the surface states of Ru nanoparticles, we measured the CO adsorption at 25 °C using in situ diffuse-reflectance infrared Fourier transform spectroscopy (in situ DRIFTS). We firstly compared the CO adsorption on the sulfate-free Ru/Ti-R and Ru/Ti-AR samples. As shown in Fig. 3c, three CO vibrational bands at 2128, 2068, and 2005 cm−1 appeared after CO adsorption on the sulfate-free samples, corresponding to the adsorption of CO on different Ru sites. Specifically, the bands at 2128 and 2068 cm−1 could be assigned to the vibrations of CO on adsorbed on interfacial sites of Ru particles that interacted with the TiO220,33, while the band at 2005 cm−1 could be ascribed to the characteristic of CO adsorbed CO on top sites of Ru particles that interacted with all surrounding sites by Ru-Ru bonds18,33. In addition, the bands at 2178 cm−1 could be ascribed to the CO adsorption on cationic Ti sites34. Notably, the intensities of the CO adsorption bands on Ru/Ti-AR were much lower than that on Ru/Ti-R, which could be due to the reduced exposure of Ru sites by air annealing. This observation was consistent with the HAADF-STEM results, which revealed that the Ru particle size in Ru/Ti-AR was apparently larger than that in Ru/Ti-R. Next, the CO adsorptions on Ru/Ti-S-R and Ru/Ti-S-AR catalysts were examined. With the introduction of the sulfate to TiO2, the CO adsorbed on TiO2 (2178 cm−1) nearly disappeared on Ru/Ti-S-R and Ru/Ti-S-AR, which should be due to the covering of cationic Ti sites by sulfate. In addition, the intensities of CO adsorption band associated with top sites of Ru particles (2004 cm−1) and interfacial sites of Ru particles (2127 and 2074 cm−1) on the Ru/Ti-S-R were comparable with the Ru/Ti-R, where only a very slight decrease was observed. This indicated that, during direct H2 treatment, the introduced sulfate mainly stayed on TiO2 and did not migrate to Ru particles. Notably, compared with the Ru/Ti-AR, the intensity of CO adsorption band (2127 and 2074 cm−1) was dramatically decreased on the Ru/Ti-S-AR, implying that most interfacial sites of Ru particles were covered by sulfate. Meanwhile, CO adsorbed on top sites of Ru particles exhibited a shift to higher wavenumbers (from 2004 to 2022 cm−1). This shift suggested an evolution of the top Ru atoms, in which the nearby interfacial Ru atoms occupied by sulfate might impact the Ru-CO bond of CO on top sites of Ru particles. The above results clearly showed that the sulfate on TiO2 tended to migrate to Ru particles after air-H2 treatment (Ru/Ti-S-AR), and did not migrate to Ru particles after direct H2 treatment (Ru/Ti-S-R), which were consistent with the EDX mapping results. Origin of the catalytic performance modification by the introduction of trace sulfates It was usually considered that the activity and selectivity of CO2 hydrogenation on supported Ru catalysts were affected by the size of Ru particles. Single Ru sites or small Ru clusters less than 1 nm were suggested to be selective for CO2 hydrogenation to CO, while larger Ru particles were typically more active for methanation reaction21,22. Our HAADF-STEM results indicated that the presence of surface sulfate had little influence on the size distribution of Ru particles. Moreover, the valence state distribution of Ru species was not obviously affected either by the surface sulfate species (Supplementary Fig. 14). Therefore, the effects of Ru particle size and the Ru valence state related to sulfates on the catalytic performance could be excluded. The CO-DRIFTS data strongly suggested that the sulfate species significantly modified the interfacial Ru sites, which were likely the main active sites controlling the CO2 conversion and product selectivity. It was reported that the activation of H2 and the transfer of H were critical steps in CO2 hydrogenation23. Our H2-TPR results demonstrated that the H2 activation on Ru particles occurred easily at low temperatures (<150 °C), indicating that the H2 activation was not the rate-determining step. Recent reports indicated that the strong hydrogen spillover, associated with enhanced H and electron migration, could lead to the reduced activation of intermediate CO, subsequently resulting in the distinctly low CH4 selectivity18,23. This suggested that the sulfate species modifying the interfacial Ru sites might significantly influence the migration of H and electrons, thus affecting the catalytic performance of Ru/TiO2 catalyst for CO2 hydrogenation. We conducted a comprehensive in situ DRIFTS study of H2 reactions with different samples to investigate the migration of H atoms and electrons. In this process, H2 molecules dissociated into H atoms at metallic sites and spilled over to O sites on the surface of TiO2 forming localized Ti-O(H)-Ti species. Simultaneously, the electrons were donated into the shallow trap states in the band gap of TiO2, leading to a broad IR absorbance in the spectrum23,35–37. As shown in Fig. 4a, the introduction of H2 at 523 K resulted in a very broad absorbance across the range of 4000 to 1000 cm−1 on Ru/Ti-S-AR, indicating the accumulation of electrons on the TiO2 surface due to H2 reduction. This broad IR absorbance was also observed on Au/TiO2 and Rh/TiO2, and it was attributed to the strong hydrogen spillover process38,39. In clear contrast, the Ru/Ti-R, Ru/Ti-AR, and Ru/Ti-S-R samples did not show such an adsorption feature, indicating that the intensity of hydrogen spillover on these samples was considerably lower than that on the Ru/Ti-S-AR. The temperature-dependent hydrogen spillover on Ru/Ti-S-AR was also depicted in Fig. 4b. As the temperature increased, the intensity of the broad absorbance across the range of 4000 to 1000 cm−1 was significantly enhanced, signifying that the hydrogen spillover was facilitated at higher temperatures. These findings demonstrated that the presence of sulfate species, which modified the interfacial Ru sites, greatly enhanced the hydrogen spillover process.Fig. 4Revealing the origin of catalytic performance modification by trace sulfates.a In situ DRIFTS following the exposure to H2 gas for Ru/Ti-S-AR, Ru/Ti-S-R, Ru/Ti-AR and Ru/Ti-R. b In situ DRIFTS following the exposure to H2 gas for Ru/Ti-S-AR at different temperature. c DFT calculations of the H transfer process on sulfate-free and sulfate-containing Ru/TiO2 catalysts (red, O; gray, Ti; cyan, Ru; yellow, S; green, H). d Schematic illustration of the mechanisms of CO2 hydrogenation on sulfate-free and sulfate-containing Ru/TiO2 catalysts. Revealing the origin of catalytic performance modification by trace sulfates. a In situ DRIFTS following the exposure to H2 gas for Ru/Ti-S-AR, Ru/Ti-S-R, Ru/Ti-AR and Ru/Ti-R. b In situ DRIFTS following the exposure to H2 gas for Ru/Ti-S-AR at different temperature. c DFT calculations of the H transfer process on sulfate-free and sulfate-containing Ru/TiO2 catalysts (red, O; gray, Ti; cyan, Ru; yellow, S; green, H). d Schematic illustration of the mechanisms of CO2 hydrogenation on sulfate-free and sulfate-containing Ru/TiO2 catalysts. We next carried out DFT calculations to investigate how the sulfate modification induced changes in the H migration. Based on the experimental results presented earlier, the theoretical calculations primarily focused on H migration reactions occurring at perimeter sites between TiO2 and Ru (Supplementary Figs. 16 and 17). The calculations revealed that the H migration from Ru to the bridge O of TiO2 on the sulfate-free Ru-TiO2 interface had a barrier of 1.91 eV and an endothermicity of 0.09 eV. In contrast, when the Ru/TiO2 interface was modified by sulfate, this barrier decreased to 1.47 eV, and the reaction became exothermic with an energy release of 0.46 eV (Fig. 4c). This result suggested that the sulfate modification greatly enhanced the H transfer process at the Ru/TiO2 interface. During the CO2 hydrogenation on Ru/TiO2 catalysts, the general reaction mechanism involved the initial adsorption of CO2 at the Ru-TiO2 interface, accompanied by H2 activation and dissociation to H on the Ru sites9,11,20,40,41. With the assistance of dissociated H, the adsorbed CO2 could be activated to form intermediate CO11,20. The presence of sufficient H and electrons allowed the intermediate CO to further convert into CH4. We performed DRIFTS studies under steady-state CO2 hydrogenation conditions, and it was observed that CO2 was easily converted to intermediate CO on Ru/Ti-AR and Ru/Ti-S-AR at low temperature. Additionally, the intermediate CO adsorbed at Ru site of Ru/Ti-AR could be converted to CH4 when the reaction temperature was above 473 K, while the intermediate CO adsorbed at Ru site of Ru/Ti-S-AR was stable and no CH4 was obtained (Supplementary Fig. 18). In the case of Ru/Ti-S-AR, where the Ru-TiO2 interface was modified by sulfate, the H transfer process was greatly enhanced. This led to more H and electrons migrating from Ru to TiO2 via the S medium, resulting in fewer H atoms remaining on the Ru sites. Consequently, this could effectively result in the low product selectivity to CH4. In contrast, on the sulfate-free Ru/TiO2, the hydrogen spillover and charge transfer could not proceed effectively. Therefore, the hydrogenation of adsorbed CO proceeded more smoothly, leading to the high CH4 selectivity (Fig. 4d). Catalytic performance can be remarkably affected, both advantageously and detrimentally, by the presence of trace impurities. However, the distinct significance of these trace impurities is often underestimated during the research process, which can potentially lead to erroneous conclusions. For example, sulfate, while capable of acting as a catalyst poison, can induce deactivation or reduced efficiency on some catalysts42. Conversely, on other types of catalysts, sulfate can play a positive role by enhancing metal dispersion or functioning as a promoter such as in photocatalytic water splitting for hydrogen production43. Other impurities such as chlorine, fluorine, or alkali cations can also yield similar positive or negative outcomes on catalytic performance44–46. In this work, the presence of trace amount of sulfate species on Ru/TiO2 could significantly change the product distribution from high CH4 selectivity to high CO selectivity. We also observed the similar product selectivity inversion by sulfate modification on other methanation catalysts such as the Rh/TiO2 and Ni/TiO2 catalysts (Supplementary Figs. 19, 20). These results further suggested the ubiquitous role of residual sulfate in controlling the product selectivity in the CO2 hydrogenation. This groundbreaking discovery serves as a poignant reminder of the paramount importance of comprehending the intricate interplay between impurities and catalyst structure in the endeavor to design catalysts that are not only more efficient but also exhibit heightened selectivity. Researchers must account for the origin of impurities, control the impurity levels, and develop strategies to mitigate their disadvantageous effects while harnessing their positive impacts on catalytic performance. Discussion In summary, we discovered that the presence of residual sulfate species in commercial TiO2 support, rather than the crystal structure of TiO2, played a pivotal role in determining the product selectivity on the Ru/TiO2 catalysts during CO2 hydrogenation. Sulfate-free Ru/TiO2 catalysts exhibited high CH4 selectivity, whereas Ru/TiO2 catalysts containing residual sulfate species displayed high CO selectivity. The annealing process in air at high temperatures induced the migration of sulfate on TiO2 to the Ru/TiO2 interface, where the interfacial sulfate species acted as an intermediate between the Ru sites and TiO2 support, significantly promoting the H transfer from the former to the latter. The strong H spillover on Ru/TiO2 catalysts containing residual sulfate species weakened the further activation of CO intermediates, resulting in low CO2 conversion but very high selectivity to CO. These findings shed light on the role of trace impurities in heterogenous catalysis, and they can inform future research and development into ever more efficient and selective heterogeneous catalysts. Methods Syntheses of Ru/TiO2 catalysts Tetrabutyl titanate (TBOT, 99.5%) and ruthenium(III) nitrosyl nitrate were purchased from Aladdin. Commercial TiO2 supports were purchased from Sigma-Aldrich and Aladdin. Ammonium sulfate was purchased from Beijing Innochem Science & Technology Co., LTD. TiO2 were synthesized by hydrolyzing TBOT in a mixture of anhydrous ethanol and distilled water with a molar ratio of n(TBOT)/n(C2H5OH)/n(H2O) = 1:15:4. Distilled water was dropped into the mixture of TBOT and anhydrous ethanol. The obtained precipitates were next dried and calcinated at 400 °C for 2 h. All Ru/TiO2 catalysts with 5 wt.% Ru were prepared using the impregnation method. TiO2 supports and a certain amount of Ru(NO3)3 were mixed in distilled water with stirring. The solution was evaporated at 60 °C under vacuum until dry. The resulting samples were dried at 100 °C overnight and then were calcined at 400 °C for 2 h in air or directly reduced with H2 to prepare various Ru/TiO2 catalysts. For the samples with sulfate addition, ammonium sulfate was incorporated during the impregnation process (with mole ratio of S/Ru set as 0, 0.03, 0.05, and 0.1). Characterization N2 adsorption-desorption isotherms for the catalysts were measured at 77 K on a Quantachrome instrument. To remove the effects of other adsorbed species, all samples were degassed at 300 °C for 6 h under vacuum before the tests. X-ray diffraction (XRD) analyses were performed on a Bruker D8 Advance diffractometer, using Cu Kα radiation (λ = 0.15406 nm) at 40 mA and 40 kV in the range 5° <2θ < 90° with a step size of 0.02°. The phase compositions of the catalysts were identified by comparison of the patterns with the Power Diffraction Files (PDF). The elemental analysis was performed using an inductively coupled plasma mass spectrometer (ICP-MS, Agilent 7700 s) equipped with a concentric nebulizer and a cyclonic spray chamber. High-angle annular dark field scanning transmission electron microscopy and element mapping images were taken by a JEOL JEM-ARM 200 F, operating at 200 kV. X-ray photoelectron spectra measurements were carried out on an AXIS Supra instrument, using a standard Al Kα X-ray source (150 W) and a pass energy of 40 eV. The binding energies (BE) of spectra were adjusted by carbon calibration (C 1s = 284.8 eV). H2-temperature-programmed reduction (H2-TPR) measurements were conducted using a Micromeritics Chemisorb 2920 analyzer. The samples (~100 mg) were placed into a U-shaped quartz tube and pretreated in an Ar (30 mL·min−1) atmosphere at 300 °C for 0.5 h. Then the samples were heated from 50 to 800 °C at a heating rate of 15 °C·min−1 in a 10% H2/Ar (50 mL·min−1) flow. The effluent gas was passed through a cold trap to remove H2O, and the signal was recorded by a thermal conductivity detector (TCD). For CO pulse chemisorption dispersion measurements, 70 mg of calcined catalyst was loaded into a U-shaped sample tube and reduced at 673 K for 1 h in 10% H2/Ar. The catalyst was then flushed with He for 30 min. After cooling the sample to 323 K, pulse chemisorption measurements were performed with 10% CO/He while monitoring the effluent with a TCD. In situ diffuse reflectance infrared Fourier transform spectroscopy (in situ DRIFTS) was performed on a Thermo Nicolet iS50 spectrometer equipped with a smart collector, and a liquid N2-cooled MCT detector. The flow of the feed gas mixture was controlled using mass flow meters. All the spectra were measured with a resolution of 4 cm−1 and an accumulation of 32 scans. A background spectrum was subtracted from each spectrum. CO adsorption experiments were carried out at 50 °C, and the mixture gas contained 500 ppm of CO and N2 balance at a total flow rate of 100 mL/min. Spin-polarized DFT calculations were carried out using the Vienna Ab initio Software Package (VASP)47. The ion-electron interactions are described using the projector-augmented wave (PAW) method and Perdew–Burke–Ernzerhof (PBE) generalized gradient approximation (GGA) functional48,49. The TiO2 anatase (101) surface was modeled with a slab consisting of three O-Ti-O layers and a 15 Å vacuum gap. The bottom layers were fixed to their bulk structure, while only the top layer was allowed to relaxed. The Ru/TiO2 model was constructed according to previous reported literature50. A planewave with a cut-off energy of 400 eV was employed. Γ-point calculations were performed for geometry optimization, with the convergence criteria for the energy and force were set to 10−5 eV and 0.02 eV/Å, respectively. The transition states for H transfer reaction were identified by relaxing the force below 0.05 eV/Å via the climbing image nudged-elastic band (CINEB) method51. Catalytic tests The evaluation of the CO2 hydrogenation reaction was carried out in a quartz tube fixed-bed reactor under atmospheric pressure. The catalyst (50 mg) was loaded into the quartz tube and reduced with 10 vol.% H2/N2 (40 mL·min−1) at 300 °C for 30 min prior to the catalytic performance evaluation. Then, the CO2 hydrogenation reaction was performed at 250–550 °C under the reaction atmosphere of 4 vol.% CO2, 16 vol.% H2 and N2 balance. The total flow rate was 40 mL·min−1, and the gas hourly space velocity (GHSV) was 48,000 mL·h−1·gcat−1. The outlet stream was analyzed by an online infrared gas analyzer. The experimental data variability for the activity tests was less than ±5%. CO2 conversion in the activity test was defined as ([CO2]inlet − [CO2]outlet)/[CO2]inlet × 100%, and the selectivity for CH4 and CO were calculated as [CH4]/([CH4] + [CO]) × 100% and [CO]/([CH4] + [CO]) × 100%, respectively. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary Information Reporting Summary Transparent Peer Review file Supplementary Information Reporting Summary Transparent Peer Review file Source data Source Data 1 Source Data 2 Source Data 1 Source Data 2