Probing active sites for carbon oxides hydrogenation on Cu/TiO₂ using infrared spectroscopy

Abstract

The valorization of carbon oxides on metal/metal oxide catalysts has been extensively investigated because of its ecological and economical relevance. However, the ambiguity surrounding the active sites in such catalysts hampers their rational development. Here, in situ infrared spectroscopy in combination with isotope labeling revealed that CO molecules adsorbed on Ti³⁺ and Cu⁺ interfacial sites in Cu/TiO₂ gave two disparate carbonyl peaks. Monitoring each of these peaks under various conditions enabled tracking the adsorption of CO, CO₂, H_2, and H₂O molecules on the surface. At room temperature, CO was initially adsorbed on the oxygen vacancies to produce a high frequency CO peak, Ti³⁺−CO. Competitive adsorption of water molecules on the oxygen vacancies eventually promoted CO migration to copper sites to produce a low-frequency CO peak. In comparison, the presence of gaseous CO₂ inhibits such migration by competitive adsorption on the copper sites. At temperatures necessary to drive CO₂ and CO hydrogenation reactions, oxygen vacancies can still bind CO molecules, and H₂ spilled-over from copper also competed for adsorption on such sites. Our spectroscopic observations demonstrate the existence of bifunctional active sites in which the metal sites catalyze CO₂ dissociation whereas oxygen vacancies bind and activate CO molecules.

The conversion of carbon oxides to products such as fuels is of high industrial relevance, but uncertainties regarding the catalytic mechanisms remain. Here, the authors use in situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) to follow CO, CO₂, H₂ and H₂O molecules as they bind to a Cu/TiO₂ surface, finding metallic copper sites serve as CO₂ dissociation sites, whereas Cu⁺ and oxygen vacancies bind CO molecules for further reductions.

Introduction

The hydrogenation of carbon oxides (CO_x, including CO₂ and CO) to various useful products, such as fuels, has been extensively studied on the surface of various heterogeneous catalysts, to solve both environmental and energy problems^1,2. Copper-based catalysts, for instance, demonstrated the ability to efficiently catalyze such hydrogenation reactions. Some of these catalysts have been already implemented in industry, as in the reverse water-gas shift reaction and methanol synthesis^2,3. Despite the extensive research, there are substantial uncertainties on the mechanism and the role of active sites in CO_x hydrogenation that hamper the rational development of such catalysts. Proposed mechanisms for these reactions are either of dissociative nature, in which CO_x species partially or totally lose oxygen then hydrogenate to products, or of associative nature, in which hydrogen atoms bind to CO_x to form various intermediate species^4–11. Heated debate arises regarding (i) active sites that mediate these reaction steps and (ii) possible reasons behind the synergetic effects between the metal and the support. To account for the synergetic effects, for instance, formation of more active sites at metal/metal oxide interface, such as metals alloys^4,5, oxygen vacancies^12,13, and interfacial Lewis acidic sites^14–16, were discussed. Other groups, however, proposed bifunctional mechanisms¹⁷, in which metal and support coordinate tasks in the reaction.

As a “flagship” of the reducible supports since the strong metal-support interaction (SMSI) was reported, TiO₂ attracted extensive research to explore its role in catalysis^18,19. A common feature in a part of this research is that TiO₂ support significantly enhances the activity and the selectivity of metal catalysts in reactions that involve carbon monoxide, either as a reactant (in CO hydrogenation)^16,20,21 or as an intermediate (in CO₂ hydrogenation)^22–24. Mechanistic investigations pioneered by Somorjai and co-workers highlighted the role of Lewis acidic interfacial sites, generated at the metal–titania interface, in facilitating C−O bond dissociation during carbon oxides transformation^15,16. These interfacial sites are produced from oxygen transfer from the metal oxides to the metal sites^25–30 and were reported in different catalytic systems such as in Pt/CeO₂ (Pt⁺ and Ce³⁺)²⁸, Pt/TiO₂ (Pt⁺ and Ti³⁺)²⁹, Cu/CeO₂ (Cu⁺ and Ce³⁺)³⁰, and Cu/ TiO₂ (Cu⁺ and Ti³⁺)³¹.

In the CO_x hydrogenation reactions, CO is not only a reactant or intermediate but also a probe^32,33 for distinguishing binding sites when combined with in situ infrared spectroscopy. This provides a unique opportunity to deduce information on the reaction mechanism and active sites. The ability of CO to identify surface sites and to assess their activity was widely implemented in CO oxidation, to develop catalysts for automotive emissions control^34–36, however, it is less explored in CO_x hydrogenation. In the current study, in situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) was employed to follow CO molecules as they bind to two disparate surface sites on Cu/TiO₂ at room temperature. The behavior of both sites was tracked, via monitoring carbonyl peak intensity and position, as isotopically labeled ¹³CO₂ molecules dissociate and H₂O is introduced, and as conditions are changed to in situ hydrogenation conditions. Examining the disparities in their behavior under these conditions revealed novel information that dictated an assignment different from what was proposed previously in the literature. Such observations and discussion provide insights on the bi-functional role played by the metal sites and the interface during CO_x hydrogenation reaction.

Result and discussion

Characterization of Cu/TiO₂ catalysts

Highly dispersed copper sites were prepared on a TiO₂ surface following our published procedure³⁷, as described in the methods section. The presence of copper sites was confirmed with X-ray photoelectron spectroscopy (XPS), CO adsorption, and UV–vis spectroscopy.

The chemical states of copper sites on the Cu/TiO₂ sample were investigated by examining the Cu 2p and Cu LMM regions in the X-ray photoelectron and X-ray-excited Auger electron spectra, respectively (Fig. 1). The catalyst was examined after hydrogen pre-treatment at 300 °C for 1 h (sample denoted as Cu/TiO₂–H₂), and after the reduced sample was annealed under Ar flow at 300 °C for 1 h (sample denoted as Cu/TiO₂–H₂–Ar). The XPS spectrum in the Cu 2p region indicates that Cu⁰ and Cu⁺ are the main species in the hydrogen-treated sample (Fig. 1a). This is evident from the presence of Cu⁰ and/or Cu⁺ peaks at 932.2 eV (Cu 2p_3/2) and 952.0 eV (Cu 2p_1/2), and the absence of Cu²⁺ satellite peaks that typically emerge in between those two peaks. It is difficult to distinguish between Cu⁰ and Cu⁺ species based only on the XPS spectrum in the Cu 2p region³⁸. Nonetheless, the Cu LMM region showed the characteristic peak for Cu⁰ at 918.8 eV (Fig. 1c; in kinetic energy)^39,40, and the amount of Cu⁰ was estimated to be around 37% from spectral fitting (Fig. S1).

Fig. 1. Copper X-ray photoelectron and X-ray-excited Auger electron spectra.

Cu 2p X-ray photoelectron spectra (a, b) and Cu LMM X-ray-excited Auger electron spectra (c, d) of: a, c Cu/TiO₂–H₂, after the catalyst was pretreated at 300 °C under hydrogen for 1 h; b, d Cu/TiO₂–H₂–Ar, hydrogen reduced sample was further treated at 300 °C under Ar.

Annealing the sample at 300 °C under Ar caused the copper speciation to shift more toward the cationic species, as the Cu²⁺ and Cu⁺satellite peaks^38,40 became more discernable (Fig. 1b). Cu²⁺ was also detected in the Cu LMM region at 917.8 eV³⁹, along with Cu⁺ and Cu⁰ (Fig. 1d). Peak deconvolutions indicate that Cu²⁺ is a minor component with around 10% peak area (Fig. S1), which is consistent with the small Cu²⁺ satellite peak in the XPS spectra. Note that in pure CuO, the intensity of the satellite peak is typically ∼0.5 of the main Cu 2p_3/2 peak⁴⁰. The increase in the amount of oxidized species after annealing under Ar flow indicates that the TiO₂ support is likely reduced with copper metal to produce oxygen vacancies, as will be discussed in detail later.

To further probe the different copper sites, CO adsorption was conducted since the IR signals of surface carbonyls strongly depend on the oxidation states of the metal. The hydrogen-treated Cu/TiO₂ sample was loaded in the in situ diffuse reflectance cell in air and purged with CO for 15 min at room temperature. Subsequently, gaseous CO was purged by flowing Ar prior to spectrum collection. Two carbonyl peaks at 2106 and 2058 cm⁻¹ are present in the spectrum (Fig. S2) for CO adsorbed on surface Cu⁺ and Cu⁰ sites^37,41, respectively, whereas on pure TiO₂ no strong carbonyl bands were observed under the same conditions. Diffuse reflectance UV–Vis spectra were also collected for the Cu/TiO₂ sample and pure TiO₂ (Fig. S3). Unlike pure TiO₂ that shows adsorption only in the UV region, the Cu/TiO₂ sample possesses absorption in the visible region between 400 and 500 nm, due to Ti^IV−O−Cu^I metal-to-metal charge-transfer⁴², and ∼600–800 nm for the d-d transition of Cu^2+,37.

Binding sites and source of the unprompted CO

In order to probe surface sites responsible for CO_x binding, the Cu/TiO₂ sample was pretreated in a Harrick Praying Mantis IR cell at various temperatures (100–400 °C) under constant Ar flow. After the sample was cooled down to room temperature under Ar, the IR cell was closed and the Cu/TiO₂ surface was monitored with in situ DRIFTS. Despite the repeated washing for TiO₂ with H₂O₂ (see the methods section), there was always a slow and spontaneous CO formation on the surface of the pretreated Cu/TiO₂ sample at room temperature, as indicated by the carbonyl peaks associated with surface-adsorbed CO (Fig. 2). The CO molecules could be produced from adventitious carbon on the surface of Cu/TiO₂⁴³. However, in our study, it is likely produced from the recombination of surface oxygen and carbon species that have been formed during the pretreatment step. Heating the sample in this step should trigger decomposition of carbonate-like species on TiO₂^44,45, to produce carbon oxides which in turn dissociate^46–52 on copper to form surface adsorbed oxygen and carbon, as suggested by the change in initial carbonate regions when the pretreatment temperature was increased, Fig. 2.

Fig. 2. DRIFTS spectra of Cu/TiO₂ after activation at different temperatures.

Carbonyl (a–e) and the corresponding carbonate (f–j) regions of Cu/TiO₂ as a function of time (10 min between spectra) after pre-treatment under Ar flow at (a, f) 100 °C, (b, g) 150 °C, (c, h) 200 °C, (d, i) 300 °C, and (e, j) 400 °C. Spectra were collected after the samples are cooled to room temperature. The carbonyl region demonstrates the change in intensity and position of two distinct CO binding sites, low frequency (LF) and high frequency (HF), on Cu⁺ and Ti³⁺, respectively. The corresponding carbonate region indicates the surface adsorbed water at 1620 cm⁻¹.

At low temperatures pretreatments, 100 and 150 °C, a small fraction of adsorbed water was removed and only a low frequency (LF) CO peak was observed on Cu/TiO₂, Fig. 2a, b. On Cu/TiO₂ pretreated at 200 °C and higher, however, more water was removed from the surface and two distinct CO peaks were observed, the LF peak and another CO peak located at a higher frequency (HF), (Fig. 2c–e). Interestingly, the initial peak positions for both the HF and LF peaks showed a strong dependence on pretreatment temperatures. The onset peak position for both peaks is blue-shifted with the increase in the pretreatment temperature. Furthermore, with time after a given pretreatment, the HF peak initially increased in intensity and then decayed at the same wavenumber, whereas the LF peak appeared later and underwent similar changes in peak intensity but red-shifted until it fully decayed. For instance, in the sample pretreated at 300 °C, the HF peak appeared at ∼2130 cm⁻¹ (ν_CO). The intensity of this peak increased gradually and then decreased while the LF peak started to emerge at ∼2118 cm⁻¹ (Fig. 2d). The LF peak slowly shifted to 2111 cm⁻¹ before its disappearance.

Both CO peaks showed different sensitivity to the residual adsorbed water that remained on the surface after different temperature pretreatments. Comparing the IR regions of surface adsorbed water, either molecularly adsorbed at ∼1620 cm⁻¹ (Fig. 2f–j) or dissociatively adsorbed as hydroxyls at 3700–3000 cm⁻¹ (Fig. S4)^53,54 indicates that the HF CO peak emerges at lower water coverage compared to the LF peak, as can be seen from the initial spectra in h–j in comparison to f and g in Fig. 2. This difference in sensitivity toward the residual water as a Lewis base indicates that the HF site is more Lewis acidic than the LF site. Furthermore, water eventually re-accumulated on the sample surface with the extended time of data collection (up to a few hours in some experiments). The source of this water is the trace amount of adsorbed water that usually exists on the cold inner surfaces of the sample cell walls, a common issue in DRIFTS and other surface studies^54,55. In each experiment, the re-adsorption of water on the sample (Fig. 2h–j) was accompanied by the movement of the CO from the HF to LF sites (Fig. 2c–e). This indicates that the re-adsorbed water will eventually replace CO, since H₂O is a stronger Lewis base, on the HF site prompting CO migration to the less acidic LF site. More discussion is presented in “Water adsorption” section.

The above observations indicate the presence of two disparate CO adsorption sites (HF and LF sites) on the Cu/TiO₂ surface. For metal catalysts supported on reducible metal oxides, it’s well documented that thermal treatment under an oxygen-deficient atmosphere triggers oxygen transfer from the metal oxides to the metal sites at the interfacial region^25–30. Such interactions generate acidic interfacial sites as demonstrated in Pt/CeO₂ (Pt⁺ and Ce³⁺)²⁸, Pt/TiO₂ (Pt⁺ and Ti³⁺)²⁹, and Cu/CeO₂ (Cu⁺ and Ce³⁺)³⁰. Similarly, thermal treatment of Cu/TiO₂ samples under inert gas flow leads to the formation of Cu⁺ and Ti³⁺ sites at the Cu/TiO₂ interface³¹. Such Lewis acidic sites at the interfaces can bind to Lewis basic molecules such as H₂O and CO. In the case of CO, the metal cations that possess partially filled d-shell can back-bond to CO, resulting in IR stretch bands lower than 2143 cm⁻¹, where the IR band of gaseous CO locates⁵⁶. In accordance with the above discussion, the LF CO peak (2118–2111 cm⁻¹) observed in the spectra shown in Fig. 2 can be attributed to the carbonyl stretching mode of CO adsorbed on the surface Cu⁺ sites^31,43. The adsorption site corresponding to the HF CO band (e.g., 2131 cm⁻¹ on Cu/TiO₂ pretreated at 400 °C), however, is more acidic since it required higher pre-treatment temperatures and was more sensitive to water adsorption. More importantly, no progressive redshift was observed in the HF CO peak over time, suggesting that it is associated with a single adsorption site. For these reasons and based on previous theoretical⁵⁷ and experimental^58–62 studies, the observed HF peak can be assigned to CO adsorbed on oxygen vacancies (Ti³⁺). Rigorous pre-treatment conditions (e.g., prolonged treatment at 450 °C under ultra-high vacuum) were often needed to create such CO adsorption sites on pure TiO₂ surfaces^58–62. In the present study, thermal treatment at 200 °C under Ar was sufficient to create oxygen vacancies on TiO₂, due to the presence of surface Cu sites which facilitates the formation of such sites. It is worth mentioning that multiple studies demonstrated that metals supported on reducible metal oxides facilitates the formation of oxygen vacancies which get stabilized via metal/metal oxide Schottky junction^27,63,64.

Isotope studies using ¹³CO₂

To probe the roles of oxygen vacancies and surface Cu⁺ sites in CO₂ dissociation, we carried out isotope labeling experiments where different amounts of ¹³CO₂ were introduced into the IR cell after Cu/TiO₂ was thermally treated at 300 °C and cooled down to room temperature under Ar. Formation of both CO isotopes, ¹²CO from surface carbon residues and ¹³CO from gaseous ¹³CO₂, on the HF/LF sites was monitored as a function of time and as the amount of ¹³CO₂ admitted was increased (Fig. 3).

Fig. 3. ¹³CO₂ isotope labeling experiment.

DRIFTS spectra of surface-adsorbed CO as a function of time on Cu/TiO₂ samples pretreated at 300 °C under Ar flow. Different amounts of ¹³CO₂ were present in the IR cell: a 0 bar, b 0.1 bar, and c 1 bar. To the right, the corresponding integrated CO peak areas are plotted as a function of wavenumber and time.

The presence of a relatively small amount of ¹³CO₂ (0.1 bar) led to the formation of ¹³CO on Cu/TiO₂, as indicated by the appearance of the HF peak at 2082 cm⁻¹ (Fig. 3b). The evolution of this peak follows the same pattern as the ¹²CO HF peak at 2130 cm⁻¹, which gradually decayed while the LF peak started to develop. Both CO isotopes followed almost identical behavior in evolution, in terms of preferential adsorption on the HF sites and their migration to the LF sites. Increasing the pressure of ¹³CO₂ to 1 bar resulted in more surface-adsorbed ¹³CO (Fig. 3c). In the spectra collected immediately after the introduction of ¹³CO₂, the intensity of the ¹³CO HF peak at 2082 cm⁻¹ is significantly greater than that of the ¹²CO HF peak. This is likely because the amount of ¹³CO produced from gaseous ¹³CO₂ is much larger than that of ¹²CO produced from surface carbon. The increase in the onset relative amount of ¹³CO as the amount of injected ¹³CO₂ increased confirms the occurring of dissociation, whereas the relentless formation of ¹²CO and its displacement to ¹³CO from their binding sites supports that each of the CO isotopes originates from two opposing reactions on the catalyst as discussed earlier.

The presence of gaseous ¹³CO₂ inhibited the migration of CO adsorbed on oxygen vacancies (the HF peak) to surface Cu⁺ sites (the LF peak), as shown by the comparison in Fig. 3. In the absence of gaseous ¹³CO₂ (Fig. 3a), the LF peak shifted from 2118 to 2111 cm⁻¹, and its maximum integrated area was slightly greater than that of the HF peak. Introducing a small amount of ¹³CO₂ (0.1 bar, Fig. 3b) significantly reduced the amount of CO adsorbed on the surface Cu⁺ sites, as shown by the relatively small integrated areas of the LF peaks (both ¹²CO and ¹³CO). Further increasing the amount of gaseous ¹³CO₂ led to a nearly complete absence of the LF peak and slow decay of the HF peak (Fig. 3c). These results suggest that the gaseous ¹³CO₂ competes with both CO isotopes during adsorption on the Cu⁺ sites but not on the Ti³⁺ sites. This is further supported by previous studies demonstrating that CO₂ became strongly adsorbed as carbonates on oxidized copper sites^46,65,66.

Water adsorption

As an omnipresent Lewis base, water molecules bind to the interfacial acidic sites on the surface of the Cu/TiO₂ catalyst. Since Ti³⁺ is a stronger Lewis acid than Cu⁺ and H₂O is a stronger Lewis base than CO, the activation temperature required to desorb water from the Ti³⁺ was higher and CO was preferably adsorbed on such site compared to Cu⁺. On Cu/TiO₂ samples with adsorbed CO, readsorption of water will replace any adsorbed CO on the Ti³⁺ sites then Cu⁺ sites, as discussed in “Binding sites and source of the unprompted CO” section.

The role of water was further confirmed by purposely introducing water vapor in the middle of the rise of the HF CO peak. Spectra were collected for a Cu/TiO₂ sample at room temperature under continuous Ar flow after pretreatment at 300 °C. During the rise of the HF, Ar flow was bypassed to flow above degassed water for 3 s then the Ar was switched back to the dry flow. As can be seen in Fig. 4, the decay of the HF peak and the rise/decay of the LF peak occurred immediately after water introduction, in less than two minutes as compared to a few hours in the absence of dosed water. The decay in the CO peaks was accompanied also by a rapid rise in the molecularly adsorbed water at 1620 cm⁻¹ and hydroxyl peak at 3695 cm⁻¹ (Fig. 4a, c). The latter peak is attributed to water healing oxygen vacancies in the vicinity of copper sites. Multiple studies^54,67,68 have shown that this hydroxyl peak (peak at 3695 cm⁻¹) is for Ti⁴⁺-OH peaks that was previously reactive defect site (Ti³⁺) but was filled with dissociatively adsorbed water.

Fig. 4. DRIFTS spectra upon introduction of water vapor.

Changes in the a hydroxyl, b carbonyl, and c water regions in the DRIFTS spectra of Cu/TiO₂ before (blue, collected every 10 min) and after (red, collected every 33 s) the introduction of water vapor into the IR cell.

H₂ admission

In our study, introducing hydrogen at room temperature to activated Cu/TiO₂ either before or during the rise of CO on the HF site did not interrupt the behavior of surface adsorbed CO, either during the rise of CO on the HF site or its transition to the LF site. Such behavior could be attributed to the inactivity of copper toward hydrogen at room temperature^69,70.

After pretreatment at 300 °C under H₂ flow, only the LF CO peak was observed in the DRIFTS spectra upon CO injection. However, upon a second pretreatment under Ar, both HF and LF CO peaks appeared (spectra not shown here for brevity). Likely, hydrogenic species formed on the HF binding site upon hydrogen pretreatment is responsible for the absence of the HF CO peak. In line with these observations, CO adsorption on Cu/TiO₂ studies showed that the appearance of the HF peak required the application of delicate pretreatment conditions^71,72, however, in such studies the HF site was assigned to different binding sites. Thermal pretreatment with hydrogen precluded the formation of the HF peak, and re-oxidation with N₂O could not retrieve it⁷¹. When the H₂ pretreatment was followed with an evacuation step at the same reduction temperature, the HF peak was observed with even an enhanced intensity^71,72. This intriguing behavior could be attributed to the competitive adsorption of CO and H₂ on surface oxygen vacancies. This behavior was demonstrated previously, with different characterization tools, over Pt/TiO₂²⁹ and Rh/TiO₂^73–76 catalysts. In such studies, it was concluded that the metal site facilitates the reduction of the titania support to produce hydrogenic species, which inhibited CO adsorption on titania.

To further investigate the role of the HF and LF CO binding sites in carbon oxides hydrogenation, those two peaks were monitored as the H₂ gas is co-adsorbed and as the temperature is increased to the reaction temperature. When the sample was purged with a mixture of Ar and CO at room temperature, only the LF CO peak was observed (Fig. 5a, b). As the temperature increased, the CO peak shifted gradually from the LF sites to the HF sites. Moreover, when Ar flow in the gas mixture was replaced with an equal amount of hydrogen, the HF CO adsorption peak was not affected until the temperature reached 200 °C. At such temperature (and higher), a red-shift and a rapid decrease in the peak intensity were observed as H₂ was admitted (Fig. 5c), for H₂ and CO coadsorption at 275 °C. Starting at 275 °C and above, the production of methane gas was observed. The CO peak intensity and position were partially retrieved again when H₂ and CO flow was switched back to Ar and CO flow. Such cyclic changes in HF peak occurred whenever the flow was switched back and forth between Ar + CO and H₂ + CO. The observed ability of the HF site (Ti³⁺) to maintain CO binding at high temperatures and to interact with both H₂ and CO at different temperatures strongly suggests that such sites are the active centers in CO hydrogenation over Cu/TiO₂.

Fig. 5. DRIFTS spectra for high-temperature CO and CO-H₂ adsorption.

a CO adsorption peaks under continuous CO flow over Cu/TiO₂ and as the temperature increases. The spectra were collected 10 min. after reaching the desired temperature. b Peak areas for the HF and LF CO peaks that were deconvoluted from a. c CO adsorption peaks at 275 °C under continuous CO flow over Cu/TiO₂, and as the flow was switched back and forth between Ar + CO and H₂ + CO.

CO₂ dissociation and carbon oxides chemical conversion

The observed disparities between the surface binding sites confirm the existence of bi-functional catalytic sites that can work collaboratively to catalyze CO₂ hydrogenation. Such bifunctionality enables heterogeneous catalysts to efficiently catalyze CO₂ conversion to higher hydrogenated products with one-pot synthesis^77,78. As discussed earlier, the introduction and dissociation of CO₂ affected only the LF site and the produced CO accumulated on the HF sites (Fig. 6). Such observations confirm the role of the metal sites in the dissociative adsorption of CO₂. Multiple studies have demonstrated that CO₂ dissociation, a key step in CO₂ chemical conversion, takes place spontaneously on pure copper metal^46–50. It is worth mentioning that accumulation of the produced oxygen poisons the copper metal surface^50,79, which can be regenerated with hydrogen, based on redox mechanism⁸⁰ in reverse water gas shift reaction. Moreover, studies of CO_x hydrogenation have demonstrated that surface copper sites can protect oxygen vacancies, the proposed active sites, from healing when CO₂ is introduced to a feed mixture of CO and H₂¹².

Fig. 6. Schematic of competitive adsorption of CO with H₂O and CO₂.

H₂O competes with CO on oxygen vacancies while CO₂ adsorbs on copper sites.

The HF site demonstrated a higher affinity to bind CO even at high temperatures. This suggests that the oxygen vacancies will stabilize the CO produced from CO₂ dissociation occurring on the neighboring metal sites. And since hydrogenic species compete with CO on such sites upon H₂ introduction, it can be concluded that oxygen vacancies activate both molecules for the reaction. Calculations and experimental observation in multiple studies have confirmed such role of the oxygen vacancies during the conversion of syngas (or a mixture of syngas and CO₂) to methanol on reducible metal oxides and copper supported on reducible metal oxides^{12,13,63,81,82}.

The observed strong adsorption of water molecules on the acidic interfacial sites, which displace CO from HF binding sites, highlights the detrimental role of water on the interfacial active sites during the hydrogenation reaction. This is in line with the previous observations that water produced during CO₂ hydrogenation deactivates the Cu/metal oxide catalysts, and for this reason, CAMERE (Carbon dioxide hydrogenation to form methanol via a reverse-water gas shift reaction) process was implemented to minimize water percentage in the reaction mixture¹¹. In such process CO₂ was reduced first to CO then the CO was fed to another reactor for further reaction.

Conclusion

We have employed in situ DRIFTS to investigate surface sites responsible for CO_x hydrogenation on Cu/TiO₂. Introducing ¹³CO₂ at room temperature to a thermally activated Cu/TiO₂ catalyst produces a mixture of ¹³CO and ¹²CO that likely originated from ¹³CO₂ spontaneous dissociation and carbon residue oxidation, respectively. The ratio of ¹³CO/¹²CO isotopes increased as the introduced amount of ¹³CO₂ was increased, however, with time ¹²CO eventually displaced ¹³CO on the catalyst surface.

Interfacial sites in Cu/TiO₂ catalyst gave rise to two distinct CO binding sites, Cu⁺ and Ti³⁺. The Cu⁺-CO spanned the range 2118–2111 cm⁻¹ whereases the Ti³⁺-CO (CO on oxygen vacancies) gave rise to a carbonyl peak at a single wavenumber in the range 2126–2131 cm⁻¹, depending on the activation temperature employed. Being more acidic, the HF site (Ti³⁺) required a higher temperature (200 °C or higher) to relinquish adsorbed atmospheric water and showed more affinity to CO than the LF site (Cu⁺). Likewise, water re-adsorption on the surface that already contains adsorbed CO prompted the migration of CO from the oxygen vacancies to the neighboring Cu⁺ sites. CO₂ admission, on the other hand, suppressed the LF CO peak area and limited the CO migration from the oxygen vacancies.

Hydrogenic species, formed from H₂ spill-over during pretreatment, prevented the formation of the CO HF peak at room temperature. However, this did not prevent the formation of the LF CO peak. Furthermore, the HF sites demonstrated the ability to interact with both CO and H₂ at high temperatures necessary to form methane.

On the Cu/TiO₂ surface, the adsorption of CO was affected by the presence of other molecules in the hydrogenation reaction mixture, including CO₂, CO, H_2, and H₂O. The observed disparities of carbonyl signals suggest the existence of bifunctional catalytic sites, in which metallic copper sites serve as CO₂ dissociation sites, whereas the Cu⁺ and the oxygen vacancies bind the produced CO molecules for further reductions.

Methods

The Cu/TiO₂ catalyst was prepared via a simple precipitation method using CuCl₂ (Sigma-Aldrich, 99.995%), P25 (TiO₂, obtained from Evonik), and ammonia (aq, BDH, ACS grade). To remove surface organic contaminants, 300 mg TiO₂ was washed with 30 mL of 1:3 by volume solution of 30% H₂O₂ (aq, J.T. Baker, CMOS grade) in Milli-Q water, using sonication to disperse the powder and centrifugation to retrieve it. A 10 mL solution of ammonia was added to 30 mL of the washed TiO₂ suspension before adding 10 mg of CuCl₂ under constant stirring. The resulting Cu/TiO₂ was separated by centrifugation, washed with Milli-Q water, and vacuum dried at room temperature overnight. Prior to infrared studies, the synthesized Cu/TiO₂ was annealed under a flow of H₂ for 3 h at 300 °C in a tube furnace.

Additional pre-treatment of powder samples was done in a Harrick Praying Mantis diffuse reflectance IR cell attached to a Thermo Nicolet 6700 FTIR spectrometer. The catalyst in powder form was compressed in the sample holder and purged with Ar (99.999%) prior to thermal treatment at the desired temperature for 1 h under Ar flow (400 mL/min). The sample was then cooled down to room temperature under continuous Ar flow before the IR cell was closed and spectra were collected. In the isotope experiments, specific amounts of gaseous ¹³CO₂ were introduced into the IR cell by using a syringe or flowing ¹³CO₂ briefly to produce desired pressure. Water vapor was introduced into the IR cell by using a bubbler with Ar flow. Typically, IR spectra were collected every 10 min, unless otherwise mentioned. UV–Visible spectra were obtained on a Cary 50 Bio spectrophotometer. A Barrelino diffuse reflectance probe was used to collect UV–Visible spectra of powder samples using BaSO₄ as a standard. X-ray Photoelectron Spectroscopy (XPS) investigations were performed with a KRATOS Axis Supra. Monochromated Al Kα was used for the excitation of the photo and Auger electrons. The binding energy scale was referenced to the C1s at 285 eV. A glove bag was used to ensure that the hydrogen-treated sample is always kept under a dry atmosphere of Ar during loading in the XPS instrument.

Author contributions

E.S and G.L. conceived the idea and planned the research. E.S. carried out the experiments. E.S. and G.L. analyzed the results and wrote the manuscript.

Peer review

Peer review information

Communications Chemistry thanks the anonymous reviewers for their contribution to the peer review of this work. Peer reviewer reports are available.

Data availability

Any relevant data are available from the authors upon reasonable request.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42004-022-00650-2.