Methanol Synthesis from CO₂ over ZnO-Pd/TiO₂ Catalysts: Effect of Pd Precursors on the Formation of ZnPd-ZnO Active Sites

Abstract

This study examines the factors influencing the formation and structure of ZnPd-ZnO active sites supported on TiO₂ by comparing their genesis from Pd/TiO₂ base catalysts prepared by impregnation using different Pd precursors (Pd(NH₃)₄(NO₃)₂ and Pd(acac)₂). The experimental results demonstrated that, in contrast to the production of CO over Pd/TiO₂ base catalysts, the selectivity of methanol over ZnPd-ZnO/TiO₂ catalysts was significantly affected by the dispersion of ZnPd intermetallic particles and the development of ZnPd-ZnO interfaces. These are determined by the characteristics of Pd particles supported on TiO₂ and their contact with the ZnO particles deposited on them. The Pd/TiO₂ base catalyst prepared by impregnation with neutral Pd precursor (Pd(acac)₂) produces a higher concentration and more effective dispersion of the ZnPd intermetallic phase as well as a wider ZnO-ZnPd interface region in comparison to the Pd/TiO₂ counterpart synthetized using the cationic Pd precursor (Pd(NH₃)₄(NO₃)₂). These differences in the ZnPd-ZnO active sites resulted in notable variations in the methanol yield, achieving the catalysts prepared with the neutral precursor about twice higher methanol selectivity from CO₂ hydrogenation at low temperatures.

1. Introduction

The catalytic conversion of CO₂ with hydrogen represents a promising strategy for reducing CO₂ emissions and vectorizing the H₂ produced by renewable methods. The hydrogenation processes can be used to create e-fuels and high-value chemicals (such as CH₃OH, CO, CH₄, olefins and dimethyl ether), which could have a significant impact on the production of these chemicals of industrial interest [1,2]. The type of product obtained by hydrogenation of CO₂ is largely dependent on the catalyst and the reaction conditions employed.

One of these valuable products is methanol, which has a wide range of potential applications. It can be used as a feedstock for the synthesis of other chemicals, as an e-fuel, or as a storage medium for green hydrogen [3,4,5]. However, the catalytic hydrogenation of CO₂ to methanol (CO₂ + 3H₂ ↔ CH₃OH + H₂O (ΔH_298K = −49.5 kJ·mol⁻¹), is a challenging process due to the simultaneous occurrence of the reverse water gas shift (RWGS) reaction (CO₂ + H₂ ↔ CO + H₂O (ΔH_298K = 41.2 kJ·mol⁻¹) and inherent kinetics limitations [6]. Furthermore, the synthesis of methanol from the very stable CO₂ molecule requires rigorous reaction conditions, such as pressures above 50 bar and temperatures between 250 and 280 °C [7,8]. In these circumstances, Cu-ZnO-Al₂O₃ catalysts, which are traditionally employed for methanol production from synthesis gas (30–60% H₂, 25–30% CO, 5–16% CO₂), are prone to rapid deactivation, mainly due to Cu sintering and oxidation, ZnO growth, and the loss of Cu-ZnO interactions [9]. In this scenario, the aim of the many research studies is to develop alternative catalysts that are highly active and selective to methanol under moderate reaction conditions (low temperature and pressure) and resistant to water deactivation [1,2].

Alternative approaches to the development of catalysts for the synthesis of methanol from carbon dioxide include formulations that facilitate CO₂ adsorption (typically on oxides), metallic particles for the activation and dissociation of hydrogen, and the control of the interfaces between hydrogen activation and CO₂ adsorption sites [10]. The objective of the development of new catalysts is to optimize both CO₂ adsorption and hydrogenation functions to achieve stabilization of intermediates, which selectively produce methanol. In this regard, Pd-based bimetallic catalysts have been demonstrated to be effective for methanol production, exhibiting good long-term stability and resistance to sintering [6]. In fact, in contrast to Cu-ZnO-Al₂O₃, Pd-based catalysts are known to be stable against H₂O-induced sintering [11], and their high intrinsic hydrogenation properties allow efficient hydrogen activation to hydrogenate the adsorbed CO₂ and their intermediates [2,12]. As far as the methanol production is concerned, the surface-catalyzed CH₃OH* formation on Pd is difficult because CO₂ hydrogenation on Pd occurs via RWGS through carboxylic intermediates (COOH*) that decomposes to CO [13]. In contrast to monometallic Pd catalysts, which demonstrate a preference for CO production via RWGS, the catalysts based on intermetallic Pd compounds, such as ZnPd, GaPd₂, or InPd, exhibit high activity and selectivity for methanol formation from CO₂ [6,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29]. Especially interesting for the research community have been catalysts formed by Pd and Zn-based intermetallics, in which the high selectivity towards CH₃OH has been attributed to electron transfer from Zn to Pd [14,20,21,22,23,24,25,26,27,28,29,30]. This electronic modification leads to a stronger adsorption of oxygen-bound species on the ZnPd surface, favoring the formation of the formate intermediate (HCOO*) for methanol synthesis [7]. The activity and selectivity of ZnPd intermetallic catalysts depend not only on the characteristics of ZnPd intermetallics but also on their contacts with ZnO [20,21,29], since ZnPd particles in the absence of ZnO mainly produce CO in CO₂ hydrogenation. The beneficial effect of a large excess of ZnO on methanol formation from CO₂ hydrogenation on ZnPd catalysts has been demonstrated [14,20]. The ZnO loading must be higher than required for the stoichiometric formation of the 1:1 β-ZnPd intermetallic phase [14] and to provide a large interface between ZnPd and ZnO [20]. The geometric structure of the active ZnPd-ZnO contacts is controversial. The catalytic activity of these contacts has been attributed to ZnO_x islands located on the surface of Pd [23,24,25], or to the ZnO overlayer located on top of the PdZn intermetallic nanoparticles, as it has been proposed for methanol steam reforming over ZnPd/ZnO [26,27]. To increase the dispersion and contacts between ZnPd and ZnO particles, they are usually supported on oxides with high surface areas (CeO₂, Al₂O₃, TiO₂…) [20,22,28,29]. Among the different supports studied, those based on TiO₂ appear to be the most effective for the production of highly dispersed ZnPd in close contact with ZnO, leading to superior activity and selectivity in the methanol synthesis from CO₂ hydrogenation [22]. In line with this, our previous work [20] has confirmed that a high surface area TiO₂ anatase is very effective as a support for dispersing and stabilizing highly active and selective ZnPd-ZnO interfaces. However, achieving the precise control of the dispersion and contacts in TiO₂-supported ZnPd-ZnO catalysts has proven to be challenging. In addition, because the segregation propensity of the ZnPd intermetallics depends on the location of the alloy on either the Pd or ZnO [31], the Pd and ZnO loadings on the support should be optimized.

The position and contact of the ZnO particles greatly influences the structure and catalytic behavior of the supported ZnPd intermetallic particles [17,19]. For example, the Ga₂O₃ coating of the Pd₂Ga intermetallic phase was shown to have a detrimental effect on methanol production [19]. In fact, the formation of core–shell structures (Pd in core and TiO_2-x in shell) seriously restricted the catalyst activity because of a lower exposition of Pd atoms to hydrogen [17]. This was confirmed by Liu et al., demonstrating that the activity of bulk intermetallic ZnPd was higher than that of ZnPd with a core–shell structure [30]. A variety of methods can be employed to prepare supported ZnPd intermetallics, with the most effective approach being the impregnation of the support using different Pd and Zn salts, including chlorides, nitrates, acetates, and others [7]. The majority of the published research on the formation and structure of ZnPd intermetallics is based on the sequential impregnation of Pd onto preformed ZnO particles. It is, therefore, of interest to study the inverse incorporation of ZnO into supported Pd on TiO₂. This approach allows us to determine the influence of the Pd particle characteristics on the formation, dispersion, and geometric structure of ZnPd intermetallics, as well as their interaction with ZnO particles, which in turn determines the active ZnPd-ZnO interfaces. In this work, different Pd particles supported on TiO₂ (anatase) have been prepared by impregnation using two different Pd precursors: Pd(NH₃)₄(NO₃)₂ and Pd(acac)₂. In the Pd(NH₃)₄(NO₃)₂ precursor, palladium (II) ions are coordinated to four NH₃; on the contrary, the Pd(acac)₂ complex presents two bidentate acetylacetonate ligands [30]. Acetylacetonate complexes are stable in the presence of H⁺ and OH⁻ groups [32,33] and react with coordinatively unsaturated (c.u.s) Ti sites. In contrast, the impregnation of TiO₂ with Pd(NH₃)₄(NO₃)₂ resulted in electrostatic adsorption of Pd(NH₃)₄²⁺ on Ti-O⁻ groups. This article presents a combination of catalytic tests and characterization, which demonstrate the significant influence of the properties of Pd particles deposited on TiO₂ support on the formation of ZnPd intermetallics from ZnO deposited on the Pd/TiO₂. These findings validate the concept of utilizing inverse ZnO-supported Pd configurations as a means of developing efficient catalysts for the selective hydrogenation of CO₂ to CH₃OH.

2. Results

2.1. Characterization of Calcined Catalysts

The supported Pd/TiO₂ base catalysts were prepared using two different Pd precursors, Pd(NH₃)₄(NO₃)₂ and Pd(acac)₂, which were labeled as Pd-CI and Pd-NI, respectively. This nomenclature takes into account the nature of the precursors used in the impregnation, which were cationic and neutral, respectively. The Pd and ZnO loadings of the calcined catalysts, as determined by ICP-OES, are presented in

Table 1. Chemical composition (from ICP-OES) and textural properties (from N₂ physisorption) of TiO₂ support and calcined Pd/TiO₂ and ZnPd/TiO₂ catalysts.

Sample	Pd b (wt.%)	ZnO b (wt.%)	SBET a (m2/gTiO2)	Vpore c (cm3/gTiO2)	dd (nm)
TiO2 a	-	-	76	0.30	13.9
Pd-CI	3.6	-	76	0.29	13.9
Pd-NI	3.9	-	75	0.28	13.5
ZnPd-CI	4.2	17.8	88	0.31	12.1
ZnPd-NI	4.3	19.2	86	0.29	11.9

^a S_BET of pure support calcined at 500 °C for 4 h; ^b normalized S_BET; ^c total pore volume; ^d average pore diameter.

. As can be observed in this table, the Pd and Zn content in the calcined samples is comparable to the nominal values (4.0 wt% for Pd and 23 wt% for ZnO).

The textural properties of the TiO₂ support and calcined catalysts were evaluated from the N₂ adsorption–desorption isotherms at −196 °C (Figure 1). In accordance with the IUPAC classification, all catalysts exhibit a type IV adsorption–desorption isotherm, which is characteristic of mesoporous materials [34]. The H1-type hysteresis loop indicates the presence of a narrow range of uniform pores, while the high slope of the isotherms suggests the existence of slit-like pores in the TiO₂ support that is maintained after the incorporation of Pd. The pore size distribution of the Pd/TiO₂ base catalysts was not affected by the type of Pd precursor (Figure 2). As expected, the incorporation of ZnO into the Pd/TiO₂ base catalysts resulted in a reduction in both the pore volume and size distribution, indicating a dilution effect rather than some occlusion of the TiO₂ pores by the deposited ZnO particles (Figure 2).

The specific BET surface area (S_BET), pore size, and pore volume of calcined samples normalized with respect to the mass of TiO₂ are listed in Table 1. As can be seen, the incorporation of Pd on the TiO₂ support did not substantially change its porous structure. However, the normalized specific surface area (BET) of ZnPd/TiO₂ catalysts increased in comparison to the values of their corresponding Pd/TiO₂ counterparts, indicating that the majority of ZnO particles were incorporated onto the external surface of Pd/TiO₂, and these particles developed additional surface area. Irrespective of the Pd/TiO₂ base catalysts, the S_BET of ZnPd/TiO₂ catalysts was very similar (88 and 86 m²·g_TiO2⁻¹). This suggests that the ZnO particles are homogenously distributed in both catalysts.

The crystallographic properties of the TiO₂ support and calcined catalysts were evaluated by X-ray diffraction. Figure 3 shows the XRD patterns together with the crystalline phase and plane to which each diffraction peak corresponds. Pd/TiO₂ and ZnPd/TiO₂ catalysts show the main diffraction peaks of tetragonal anatase phase of the TiO₂ support (JCPDS 00-021-1272). The intensity of the diffraction peaks associated with the TiO₂ support in the calcined catalysts does not change after Pd incorporation, indicating that the degree of crystallinity of the support has been preserved. Consistent with this, the crystallite size of TiO₂, calculated using the Debye-Scherrer equation in calcined Pd/TiO₂ and ZnPd/TiO₂ catalysts was close to that of the bare TiO₂ support (

Table 2. Crystallite size of TiO₂, PdO, and ZnO and relative PdO (002)/ZnO (100) intensities ^a inTiO₂ support and calcined Pd/TiO₂ and ZnPd/TiO₂ catalysts (from XRD).

Sample	TiO2 (101) (nm)	PdO (002) (nm)	ZnO (100) (nm)	PdO (002)/ ZnO (100)
TiO2	17.4	-	-	-
Pd-CI	17.6	9.2	-	-
Pd-NI	17.6	7.2	-	-
ZnPd-CI	17.6	9.9	16.8	0.09/0.13
ZnPd-NI	17.6	9.0	16.9	0.08/0.10

^a relative peak intensity with respect to TiO₂ (101).

). The calcined Pd/TiO₂ and ZnPd/TiO₂ catalysts also show small diffraction peaks associated with PdO particles of low crystallinity (P42/mmc space group, tetragonal, JCPDS 00-006-0515). The crystallite size of the PdO (002) particles on Pd/TiO₂ catalysts shows differences depending on the precursor used for impregnation, with the sample prepared from the neutral precursor having the smallest size (7.2 nm for Pd-NI vs. 9.2 nm for Pd-CI). The size of the PdO particles on ZnPd/TiO₂ catalysts does not change significantly (taking into account the difficulty of the deconvolutions of the PdO peak (002), which overlaps with the ZnO peak (002)) with respect to the values measured on the Pd/TiO₂ counterparts. Finally, the ZnPd/TiO₂ catalysts show the main diffraction peaks associated with ZnO particles with moderate crystallinity and hexagonal phase (P63mc space group, JCPDS 01-080-0075). The values for the crystalline size of the ZnO particles in the ZnPd/TiO₂ catalysts were similar (~16.8 nm, Table 2) and were not influenced by the characteristics of the Pd/TiO₂ base catalysts. The relative intensities of the (002) peaks of PdO and (100) ZnO with the (101) TiO₂ peak were similar in both ZnPd/TiO₂ catalysts (Table 2), indicating similar crystallinity of PdO and ZnO in the catalysts.

The reducibility of calcined Pd/TiO₂ and ZnPd/TiO₂ catalysts was investigated by temperature-programmed H₂ reduction followed by mass spectrometry (MS) and thermal conductivity detector (TCD). The TPR profiles corresponding to hydrogen consumption followed by MS (ionic current of m/z = 2) in the temperature range between −33 °C and 100 °C of Pd/TiO₂ and ZnPd/TiO₂ catalysts are shown in Figure 4A and Figure 4B, respectively, while the H₂-TPR consumption of ZnPd/TiO₂ in the range 200–400 °C is illustrated in Figure 4C.

The peaks observed at sub-ambient temperatures in the TPR profiles of Figure 4A are attributed to the consumption of hydrogen for the reduction in PdO particles to metallic Pd [35,36]. It is established that the reducibility of PdO particles depends on their size and interactions with the TiO₂ support. Figure 4A illustrates that the reduction in the PdO particles on the Pd-NI sample occurs more homogeneously and at a higher temperature than on the Pd-CI counterpart (0 °C and −10 °C, respectively). This is in accordance with the sequence of crystallite size of PdO particles determined by XRD (Figure 3) and is consistent with previous results in the literature, which indicates that the reduction in PdO particles occurs at lower temperature as their size increases [37,38]. The Pd/TiO₂ samples do not exhibit any evidence of Pd hydride formation. Given that the formation of Pd hydrides is dependent on the size of the Pd metal particles and their interaction with the support, the absence of hydride formation in the Pd/TiO₂ samples may be attributed to the small size of the Pd particles and their interaction with the TiO₂ support.

The reduction in PdO particles in ZnPd/TiO₂ catalysts (Figure 4B) follows the same differences in temperature observed in the Pd/TiO₂ base catalysts. However, the reduction occurs at a higher temperature as a consequence of the covering and interaction of ZnO in the ZnPd/TiO₂ catalysts. The ZnPd/TiO₂ catalysts show a broad reduction peak (Figure 4C) in the temperature range 200–400 °C (Figure 4C), which is related to the partial reduction in the ZnO by hydrogen spillover from Pd to form the ZnPd intermetallic particles [32]. These ZnPd intermetallic particles are formed by the diffusion of Zn atoms into Pd particles. Moving from ZnPd-NI to ZnPd-CI, the reduction peaks associated with the reduction in ZnO with the formation of ZnPd intermetallic particles are shifted from 304 to 353 °C (Figure 4C), indicating closer Pd-ZnO interactions and a greater ability to form ZnPd intermetallics in the former.

2.2. Characterization of Reduced Catalysts

2.2.1. XRD and HRTEM

XRD patterns of Pd/TiO₂ and ZnPd/TiO₂ catalysts reduced at 450 °C are shown in Figure 5. The reduced Pd/TiO₂ catalysts show diffraction peaks corresponding to the anatase TiO₂ support, and no significant changes were noted with respect to the calcined samples following reduction. Additionally, the (111) diffraction peak, of low intensity, associated with Pd metal particles (cubic JCPDS-00-201-1272) was also observed in the reduced Pd/TiO₂ samples. The crystallite size of Pd metal particles in reduced Pd/TiO₂ samples, calculated by Scherrer equation (9.1 nm for Pd-CI, 7.7 nm for Pd-NI), was found to be similar to the values observed in the calcined counterparts (Table 2). This suggests that the palladium particles did not sinter during the catalyst reduction at 450 °C. The diffraction patterns of reduced ZnPd/TiO₂ catalysts do not show any diffraction peaks associated with metallic Pd. Instead, the diffraction peaks are characteristic of the (111) and (200) planes of crystalline intermetallic ZnPd particles (JCPDS 00-006-0620). The crystallite size of the ZnPd intermetallic particles was calculated using the Scherrer equation and found to be similar to that of the metallic Pd particles, with ZnPd-CI exhibiting a size of 10.7 nm and ZnPd-NI showing a size of 8.5 nm. This confirms that the initial size of the Pd particles in Pd/TiO₂ base catalysts defines the final size of the intermetallic particles. Following reduction, the ZnPd/TiO₂ catalysts exhibited diffraction peaks characteristic of hexagonal ZnO particles, but with a slight reduction in their crystallite size (15.5 nm for ZnPd-CI, 14.1 nm for ZnPd-NI) in comparison to the values observed in the calcined Pd/TiO₂ counterparts (Table 2). This reduction in crystallite size may be attributed to the partial reduction in ZnO particles and the migration of Zn to Pd particles, which leads to the formation of the intermetallic ZnPd phase.

In order to gain insight into the nanomorphology and contacts between ZnO and ZnPd intermetallic particles, the ZnPd-CI and ZnPd-NI catalysts were studied by TEM (Figure 6A and Figure 6B, respectively). For both catalysts, fast Fourier transform (FFT) and inverse FFT (IFFT) analyses indicate the presence of TiO₂ particles with d-spacing of 0.35 nm ((101) planes of the anatase phase), single ZnPd nanocrystals with lattice spacing of 0.219 nm ((111) planes of tetragonal ZnPd) and ZnO particles d-spacing of 0.28 nm ((002) planes of hexagonal ZnO). The HRTEM image of the ZnPd-CI-reduced catalysts (Figure 6A) shows agglomerates comprising ZnPd particles exceeding 10 nm in size, situated in close proximity to ZnO particles. In comparison, the HRTEM image of the ZnPd-NI catalyst (Figure 6B) shows ZnPd nanoparticles located between TiO₂ and ZnO particles. The distribution histograms of the particle size with the standard deviations are shown in Figure 6. In good agreement with XRD, the statistical analysis of the particle size suggests the higher particle size for ZnPd in ZnPd-CI than in ZnPd-NI (10.8 ± 3.9 nm vs. 6.5 ± 3.3 nm). The particle size distribution is significantly narrower in the latter sample. The statistical analysis of the particle size of ZnPd in reduced catalysts, determined by TEM, in accordance with the X-ray diffraction (XRD) findings, indicates that the ZnPd particles in ZnPd-CI are of a greater diameter than those in ZnPd-NI (10.8 ± 3.9 nm vs. 6.5 ± 3.3 nm). Furthermore, the particle size distribution in the latter sample is markedly narrower than that observed in the ZnPd-CI counterpart.

2.2.2. CO-DRIFTS

The transformation of metallic Pd to ZnPd alloy was investigated by DRIFTS study of CO chemisorbed on the surface of Pd/TiO₂ and ZnPd/TiO₂ catalysts reduced at different temperatures (Figure 7). The Pd/TiO₂ catalysts that were reduced at 70 °C (Figure 7A) exhibited three distinct adsorption bands that correspond to linearly bound CO on metallic Pd sites at 2080 cm⁻¹ and a double- and triple-bridged CO adsorption on Pd° sites at 1970 cm⁻¹ and 1850–1890 cm⁻¹, respectively. It is evident that the Pd-NI displays a greater intensity of linearly and bridged-bonded CO bands in comparison to the Pd/NI counterpart. This observation is in accordance with the preceding XRD and TEM results, which indicated a higher Pd surface concentration in the Pd-NI sample as consequence of its lower Pd particle size. A comparison of the DRIFT-CO spectra of ZnPd/TiO₂ reduced at 70 °C with their corresponding Pd/TiO₂ catalysts (Figure 7A) reveals a notable reduction in the intensity of double- and triple-bridged CO adsorption on Pd° in the case of the ZnPd-NI sample, while the decrease is more moderate in the case of the ZnPd-CI sample. This suggests that the contact between ZnO and Pd particles may depend on the dispersion of the Pd particles on the TiO₂; the contact is facilitated when the Pd particles are smaller and more dispersed in the TiO₂ support, as observed in the Pd-NI sample.

The formation of the intermetallic phase was investigated by the evolution of CO-DRIFTS on the ZnPd/TiO₂ catalysts reduced from 70 to 450 °C (Figure 7B). Reduction at 350 °C produces changes in the CO adsorption profile for both the ZnPd-NI and ZnPd-CI catalysts. The bands of multiple-bound CO, which are characteristic of metallic Pd, exhibit a notable decrease in intensity, while the linear CO band shows an increase. These changes are indicative of the formation of the ZnPd intermetallic phase, as previously observed in the literature reporting that there is a decrease in the concentration of multiple-bound CO species with the increase in the formation of intermetallic ZnPd because of the increase in the Pd-Pd distance [11,20]. The changes are more pronounced in the ZnPd-NI catalyst, which exhibits a very sharp linear CO band, indicating the easier formation of ZnPd intermetallics in this sample in respect to the ZnPd-CI-counterpart. The subsequent increase in reduction temperature at 450 °C produces a decrease in the signals for CO on both ZnPd/TiO₂ catalysts that should be a result of the additional formation of a ZnO_x over-layer at the surface of the ZnPd particles induced by SMSI [39]. The differences in the intensity of CO bands of the ZnPd/TiO₂ catalyst reduced at 450 °C (Figure 7B) indicate a higher surface concentration of ZnPd intermetallic on the ZnPd-NI with respect to the ZnPd-CI counterpart. This corroborates the previous results obtained by XRD and HRTEM, which showed a lower size and close contact of ZnPd intermetallic particles with ZnO in the ZnPd-NI catalyst.

2.3. Activity Tests

The activity and selectivity of the Pd/TiO₂ and ZnPd/TiO₂ catalysts in the CO₂ hydrogenation were evaluated at moderate reaction conditions (250 °C, P = 30 bar) and high GHSV (12,000 mL_CO2/H2·g_cat⁻¹·h⁻¹) to achieve low CO₂ conversions [29]. The activity data for the Pd/TiO₂ and ZnPd/TiO₂ catalysts under stationary conditions are presented in

Table 3. CO₂ conversion, selectivity, and methanol production in the CO₂ hydrogenation over Pd/TiO₂ and PdZn/TiO₂ catalysts (T = 250 °C; P = 30 bar, TOS = 12 h, data corresponding to the average during the entire test after stabilization).

Catalyst	χCO2 (%)	Selectivity (%)			STY (mmol·min−·molPd−1)
		MeOH	CO	CH4	CH3OH a	CO
Pd-CI	5.5 ± 0.1	-	94.8 ± 0.2	5.2 ± 0.2	-	313.2 ± 6.2
Pd-NI	5.9 ± 0.1	-	95.0 ± 0.6	5.0 ± 0.6	-	319.8 ± 6.3
ZnPd-CI	2.3 ± 0.02	35.5 ± 0.01	64.5 ± 0.01	-	41.6 ± 1.0	71.3 ± 1.4
ZnPd-NI	2.2 ± 0.02	64.7 ± 0.2	35.3 ± 0.2	-	67.8 ± 0.6	37.4 ± 0.3

^a Spatial methanol yield (STY) calculated considering Pd wt.% from ICP-OES.

, and Figure 8 illustrates the space–time yields for CO and methanol production. The only products formed in the hydrogenation of CO₂ over the Pd/TiO₂ catalysts were CO and CH₄. In contrast, methanol and CO were formed as main products in the CO₂ hydrogenation over the ZnPd/TiO₂ catalysts (Table 3). Both Pd/TiO₂ catalysts exhibit very high selectivity for CO production (≈95%), which means that they are highly active for the RWGS reaction. The use of supported Pd/oxide systems as active catalysts for RWGS is a relatively unexplored area of research, which makes it challenging to draw comparisons between the catalytic behavior of our Pd/TiO₂ catalysts [14,40]. Similarly to the approach taken with the Pt/TiO₂ catalyst in the RWGS reaction, the activity of our Pd/TiO₂ catalysts may be associated with the Pd-TiO₂ interfaces and the formation of Pd–O_v–Ti³⁺ sites [41]. In accordance with this hypothesis, the slightly higher RWGS activity of the Pd-NI in comparison with the Pd-CI may be attributed to the better dispersion and homogeneity of the Pd particles achieved in the former. In contrast to the results obtained with RWGS over Pd/TiO₂ catalysts, the ZnPd/TiO₂ catalysts exhibited higher methanol selectivity and lower CO₂ conversion (Table 3), which is in good agreement with previous studies [14,20]. Furthermore, the ZnPd/TiO₂ catalysts demonstrated stability over the duration of the reaction. The ZnPd-NI and ZnPd-CI catalysts exhibited different methanol selectivities (Table 3), with the former displaying significantly higher activity than the latter (Figure 8). A comparison of the results obtained for the ZnPd-NI catalysts with those for the catalyst prepared by sequential Pd incorporation onto ZnO/TiO₂ (as previously studied by our research group [20]) revealed that the inverse incorporation of ZnO on Pd/TiO₂ exhibited higher CH₃OH selectivity at the same reaction conditions.

2.4. Characterization of Used Catalysts

In order to evaluate the possible alterations in the crystal structure of the catalysts during the course of the reaction, the Pd/TiO₂ and ZnPd catalysts were analyzed after use by XRD (Figure 9). The Pd/TiO₂ catalysts show diffraction peaks associated with the crystalline Pd⁰ phase (cubic, JCPDS-00-002-1439). In line with XRD of the freshly reduced catalysts (Figure 5), the used Pd-CI shows greater crystal size of Pd⁰ particles than the Pd-NI counterpart (9.8 vs. 7.0 nm). The crystallite size of the metallic Pd phase in both Pd/TiO₂ samples was found to be similar to that of the freshly reduced catalysts (

Table 4. Crystallite sizes of freshly reduced and used Pd/TiO₂ and ZnPd/TiO₂ catalysts (from XRD).

Catalysts	Pd0 (111)	ZnPd (111)	ZnO (100)	IZnO(100)/IZnO(002)
	Freshly red./used	Freshly red./used	Freshly red./used	Freshly red./used
Pd-CI	9.1/9.8	-	-	-
Pd-NI	7.7/7.0	-	-	-
ZnPd-CI	-	10.7/10.1	15.5/15.0	1.44/1.39
ZnPd-NI	-	8.5/9.4	14.1/13.8	1.31/1.41

), which suggests that the Pd particles did not sinter after reaction. This finding contrasts with the Pd particle sintering reported in the literature, which was explained as being due to an increase in the mobility of Pd species in the presence of CO, facilitating the ripening mechanism through lowering the barriers for Pd cluster diffusion [42,43,44]. For example, in the reaction carried out under high hydrogen pressure conditions, the migration of Pd particles on the surface of the TiO₂ carrier was observed by Chen et al. [45]. However, it has recently been demonstrated that the sintering of Pd on TiO₂ is dependent on both Pd-support interactions and the exposed facets of TiO₂ [46]. Consequently, the observed stability of Pd/TiO₂ catalysts may be attributed to the development of strong Pd-TiO₂ support interactions that are formed after reduction via the SMSI effect.

The differences in the size of Pd particles in the used Pd/TiO₂ catalysts, determined by XRD, were verified by TEM (Figure 10). As illustrated in Figure 10A, the Pd-NI catalyst exhibits smaller and more homogeneously dispersed Pd particles on the support surface than the Pd-CI counterpart. Statistical analysis of the main diameter of the Pd particles confirmed a lower average Pd particle size of Pd-NI with respect to its Pd-CI counterpart (10.2 nm vs. 12.4 nm, Figure 10B), as well as a narrower particle size distribution.

The XRD patterns of used ZnPd/TiO₂ catalysts (Figure 9) show diffraction peaks of crystalline ZnPd intermetallic (tetragonal, JCPDS card 00-006-0620) and ZnO phases (hexagonal, JCPDS card 01-080-0075). The comparison of the crystallite sizes of the ZnPd intermetallic phase after the reaction indicates that there is no significant sintering of this phase for both ZnPd/TiO₂ catalysts (Table 4). The used ZnPd/TiO₂ catalysts maintain the differences in crystallite size of the ZnPd phase observed in the freshly reduced state. It is important to consider the crystalline structure and orientation of the ZnO surfaces when evaluating the activity of ZnPd/TiO₂ catalysts. The adsorption of CO₂ is influenced by these factors, with Zn-terminated ZnO polar surfaces demonstrating higher activity than their nonpolar counterparts. The crystallite size of the ZnO phase in both ZnPd/TiO₂ used samples was found to be similar to that of the fresh reduced catalysts (Table 4), which indicates that the ZnO particles did not sinter after the reaction. Changes in the orientation of the ZnO surfaces were also analyzed. Table 4 provides a comparison of the intensity ratio of nonpolar (2θ at 31.7°) and polar (2θ at 4.4°) crystal facets of the ZnO phase of the fresh and used catalysts (I_ZnO(100)/I_ZnO(002)) [47]. As can be seen, the ZnO crystals of both used ZnPd/TiO₂ catalysts have a larger exposure of their nonpolar facets, exhibiting a similar ratio of their nonpolar/polar surfaces. There is not a significant change in the orientation of the ZnO under reaction conditions when going from freshly reduced to used catalysts. Therefore, the observed structural stability of the ZnPd and ZnO crystals in both ZnPd/TiO₂ catalysts under reaction conditions is a key factor in their overall stability throughout the course of the reaction.

The surface analysis of the Pd/TiO₂ and ZnPd used catalysts was evaluated by XPS. Figure 11A and Figure 11B present the Pd 3d spectra of Pd/TiO₂ and ZnPd/TiO₂, respectively, whereas Figure 11C show the Zn LMN Auger spectra of ZnPd/TiO₂ used catalysts. The deconvolution of the Pd 3d_5/2 level spectra in Pd/TiO₂ used catalysts (Figure 11A) reveals two distinctive Pd 3d_5/2 components at binding energies (BEs) of 335.0 ± 0.1 eV and 338.0 eV. The former corresponds to metallic Pd⁰ species, while the latter suggests the presence of a Pd²⁺ ionic species. The presence of a Pd²⁺ ionic species in used catalysts is probably due to air exposure of the samples prior to XPS analysis [48]. For Pd-NI, the stronger intensity of the Pd 3d_5/2 peak suggests a better Pd surface exposure of this catalyst than that of Pd-CI counterpart. Quantification of Pd/Ti atomic surface concentration (

Table 5. Surface atomic ratios for Pd/TiO₂ and ZnPd/TiO₂ used catalysts calculated from XPS data.

Atomic Ratios	Pd-CI	Pd-NI	ZnPd-CI	ZnPd-NI
Pd/Titotal	0.12	0.27	0.20	0.33
Zn/Ti	-	-	15.88	13.7
ZnPd/Pdtotal	-	-	0.48	0.76

) corroborates the better Pd surface concentration on the Pd-NI catalysts respect the Pd-CI counterpart, which is in line with previous XRD and TEM data.

As compared to Pd/TiO₂, the Pd 3d core-level of the ZnPd/TiO₂ used catalysts (Figure 11B) shows Pd 3d_5/2 contributions located around 335.1 eV and 334.7 eV. The BE of the former peak is corresponding to metallic Pd while that of the latter is due to ZnPd intermetallic phase [20]. The chemical state of Zn atoms in used ZnPd/TiO₂ catalysts was investigated through the analysis of Zn LMM auger electron spectra (Figure 11C). The Auger LMM electron spectra of the Zn level of the used ZnPd/TiO₂ catalysts show a primary peak of ZnO at 983 eV and a secondary peak at 992 eV, while the main peak of metallic Zn species are observed at 992 eV with a shoulder at 996 eV. The presence of metallic Zn contributions is related with the formation of the ZnPd intermetallic particles and/or partially reduced ZnO_x particles.

Analysis of the surface composition data in Table 5 shows that the concentrations of Pd and ZnPd in the used ZnPd/TiO₂ catalysts differ significantly. In the case of the ZnPd-NI catalyst, the total surface concentration of palladium is somewhat higher than in the ZnPd-CI counterpart, which maintains the differences already detected in the Pd/TiO₂ catalysts they derived from. The used ZnPd-NI catalyst also shows a higher ZnPd/Pd ratio than the ZnPd-CI sample, indicating that the catalyst maintains the more favorable formation of the ZnPd intermetallic that was previously observed in the freshly reduced ZnPd-CI sample by CO-DRIFTS (Figure 6).

3. Discussion

The characterization of the Pd/TiO₂ based catalysts revealed differences in the interaction and dispersion of the Pd particles depending on the metal precursor used in the impregnation of the TiO₂ support. Pd particles in Pd-NI show smaller particle size and a stronger interaction than Pd-CI, as confirmed by XRD (Figure 3) and TPR-MS (Figure 4). The use of the neutral precursor Pd(acac)₂ was more effective for the Pd dispersion than the use of the cationic Pd precursor Pd(NH₃)₄(NO₃)₂. It is known that, depending on the pH of the solution, the oxygen functional groups of titania behave as amphoteric oxides, dissociating or protonating according to the following equations:

Ti-OH₂⁺ ↔ Ti-OH + H⁺

Ti-OH ↔ Ti-O⁻ + H⁺

In this situation, the acetylacetonate complexes do not interact with the OH⁻ groups [31,32] and prefer to react with coordinatively unsaturated (c.u.s) sites. In contrast, the impregnation of TiO₂ with Pd(NH₃)₄(NO₃)₂ led to electrostatic adsorption of Pd(NH₃)₄²⁺ on Ti-O⁻ groups. The better Pd dispersion on the surface of the reduced Pd-NI catalyst compared to the Pd-CI counterpart could then be attributed to the presence of c.u.s. sites on the TiO₂ support, where planar Pd(acac)₂ adsorption was facilitated. The poorer Pd dispersion of the Pd-CI catalyst may also be related to the uncontrolled growth of PdO particles during calcination of the adsorbed Pd[(NH₃)₄]²⁺ surface complexes [49]. Regardless of the Pd precursor, both Pd-NI and Pd-CI catalysts showed similar high activity and selectivity for CO in the hydrogenation of CO₂.

Regardless of the mechanism of the RWGS reaction, a selective RWGS catalyst needs a mild C–O dissociation ability and a weak CO adsorption ability to avoid the dissociation of CO and the further hydrogenation of CO to CH₄ and methanol. Therefore, the balance between C–O dissociation and hydrogenation is a key factor for highly active and selective RWGS catalysts. The hydrogenation ability can be provided by the Pd particles whereas the C–O breaking ability may be enhanced by the TiO₂ support to improve CO₂ adsorption. The similar and very high production of CO by the RWGS reaction over both Pd/TiO₂ catalysts suggests their comparable number of active sites that react weakly with CO₂ (that are selective for CO) and the low development of the active sites that strongly react with CO₂ (that are selective to the formation of CH₄) [50]. Considering that CO production through the rWGS reaction is highly dependent on particle size, structural defects/distortions and chemical non-stoichiometry [51,52], the differences in the physicochemical properties of Pd/TiO₂ catalysts (confirmed by XRD, HRTEM, and XPS) could explain the CO selectivities (Figure 12). The characterization of the Pd-NI used catalyst by XPS shows a higher Pd surface atomic ratio with respect to the Pd-CI counterpart (Table 5). This fact implies a smaller Pd particle size and a higher number of Pd support interfaces in the Pd-NI catalyst. Nevertheless, the Pd-NI catalyst shows similar activity to its Pd-CI counterpart. Since the XPS data of the catalysts used confirmed that the formation of Ti³⁺ on the surface of the TiO₂ support was very low (max. 2%), the participation of the support in the reaction mechanism through the adsorption of CO₂ on its oxygen vacancies was minimal and the role of TiO₂ was limited to improving the dispersion of the active Pd phase [53,54,55,56]. Considering the study by Wang et al. that suggested that there is no direct dissociation of CO₂ on Pd [50], the similar activity and CO selectivity observed on the Pd-NI could be explained by considering that its better dispersion and higher Pd-TiO₂ interface on this catalyst could be compensated by its stronger Pd-TiO₂ interactions that alter the intrinsic hydrogenation activity of the Pd particles for the RWGS reaction.

The comparison of the selectivity of ZnPd/TiO₂ with Pd/TiO₂ catalysts (Figure 12) showed the change in the nature of the active sites for the selective synthesis of CO by RWGS observed on the Pd/TiO₂ catalysts to methanol synthesis on the ZnPd/TiO₂ catalysts [57].

The characterization of the ZnPd/TiO₂ catalysts confirms the role of the initial characteristics of the supported Pd particles on TiO₂ in the genesis of the ZnPd intermetallics prepared by the reverse incorporation of ZnO on the Pd/TiO₂. The differences in the dispersion and size of Pd on TiO₂ observed in the supported Pd/TiO₂ base catalysts produce differences in the development of contacts with the inverse decoration of ZnO particles deposited on them since the size of the ZnO particles is similar in both cases (around 17 nm). As mentioned above, the characterization of the Pd/TiO₂ based catalysts showed that the PdO particles in Pd-NI have a smaller particle size and a stronger interaction than Pd-CI, as confirmed by XRD and TPR-MS. The better dispersion of PdO particles achieved on Pd-NI base catalysts provides a better contact with ZnO particles as H₂-TPR demonstrates (Figure 4). This better contact between PdO and ZnO particles facilitates the formation of ZnPd intermetallics, since these ZnPd intermetallics are formed by the diffusion of reduced Zn atoms (facilitated by the dispersed Pd particles) into the closer Pd particles [58]. The formation of the ZnPd intermetallic by the reduction of H₂ at 450 °C implies that the ZnPd intermetallic formed is in close contact with the ZnO particles. The easier formation of ZnPd intermetallics on the ZnPd-NI catalysts was confirmed by CO-DRIFTS (Figure 6B) and their close contact between ZnPd and ZnO particles was clearly observed by HR-TEM (Figure 6). HR-TEM and XRD results also confirm that the initial size of the Pd particles in Pd/TiO₂ base catalysts defines the final size of the ZnPd intermetallic particles in ZnPd/TiO₂ reduced catalysts.

The selectivity to methanol on the ZnPd/TiO₂ catalysts varies (64.7% on ZnPd-NI and 35.5% on ZnPd-CI) and its comparison with the characterization results indicated that this is directly related to the formation of the ZnPd intermetallics on the catalysts, as it was demonstrated by the XPS data of the used catalysts. In fact, the methanol/CO ratio on the ZnPd/TiO₂ catalysts directly correlates with the surface PdZn/Pd ratio on the used catalysts, as determined by XPS (Table 5). Therefore, the selectivity to methanol in the CO₂ hydrogenation on ZnPd/TiO₂ catalysts prepared by inverse ZnO incorporation on Pd/TiO₂ depends on the relative formation of ZnPd intermetallics with respect to metallic Pd (Figure 13) under reaction conditions. This is explained by the fact that the ZnPd intermetallics have a stronger electron donor character compared to Pd, allowing weak adsorption of carbon-bonded species and strong adsorption of oxygen-bonded species [8]. This implies that the carboxyl and formate intermediates are preferentially formed on Pd and ZnPd intermetallics, respectively, which are intermediates that selectively evolve to CO and methanol, respectively [8]. Nevertheless, the contact between ZnPd and ZnO is also important for the selective formation of methanol. As mentioned above, the formation of the ZnPd intermetallics on ZnPd/TiO₂ catalysts prepared by inverse ZnO incorporation implies that the formed ZnPd alloy is in close contact with ZnO after reduction. Thus, the higher formation of the ZnPd intermetallics on the surface of the ZnPd-NI catalyst also implies the presence of a large interface between ZnPd and ZnO (Figure 13). In this regard, the previous study by Zabilskiy et al. showed that the ZnPd-ZnO contacts are necessary for hydrogen activation (ZnPd) and CO₂ activation (ZnO) [21]. Thus, in the case of ZnPd-NI, the formation of a higher amount of ZnPd phase implies higher hydrogen dissociation on ZnPd, allowing a higher hydrogenation of formate to methanol over ZnO [21]. In the CO₂-methanol process, the presence of CO₂ and the H₂ produced during the reaction can create an oxidative environment capable of destabilizing the ZnPd intermetallic phase. Taking into account this fact, and despite maximizing the formation of ZnPd in the fresh reduced catalysts, the ZnPd alloy partially decomposes under reaction conditions and is, therefore, hardly able to completely avoid the RWGS reaction on the Pd sites formed by the decomposition of the ZnPd intermetallic under a reaction condition. Our group has recently investigated the effect of the Zn/Pd ratio on the formation and dispersion of ZnPd-ZnO interfaces supported on TiO₂ [20], and we have concluded that the use of a high Zn/Pd ratio controls the formation and stability of the ZnPd intermetallics under reaction conditions and can be maximized, but their partial decomposition under reaction conditions and, therefore, the participation of RWGS associated with a Pd species cannot be completely avoided.

4. Materials and Methods

4.1. Catalysts Preparation

Commercial TiO₂ (anatase 99.6%; pellets; specific surface area 156 m²·g⁻¹) and palladium (II) nitrate hydrate (Pd(NO₃)₂·2H₂O, ≥99 wt.%) were purchased from Alfa Aesar (Ward Hill, MA, USA), whereas NaOH (98%, pellets) was supplied by PanReac AppliChem (Darmstadt, Germany). Tetraammine palladium (II) nitrate ([Pd(NH₃)₄] (NO₃)₂, 10 wt.% aqueous solution), Pd(acac)₂ (acac = acetylacetonate; Pd(C₅H₇O₂)₂; 99%) and Zn(acac)₂ (acac = acetylacetonate, 99%) were supplied by Sigma-Aldrich (Saint Louis, MO, USA).

Prior to the preparation of Pd/TiO₂ based catalysts, the TiO₂ support was ground and sieved to select the fraction with sizes between 212 and 425 μm. The sieved TiO₂ particles were then thermally stabilized by calcination in static air at 500 °C for 4 h with a heating ramp of 4 °C/min. The Pd loading in Pd/TiO₂ catalysts was fixed at 5% wt because, taking into account the surface area of the TiO₂ support, we are far from the theoretical monolayer (5% wt Pd = 0.25 theoretical monolayers) and, therefore, we studied the effect of the Pd precursors on Pd interactions/dispersion on TiO₂ more clearly.

The first Pd/TiO₂ base catalyst was prepared with a nominal palladium loading of 5 wt.% by impregnating TiO₂ with tetraamminepalladium (II) nitrate. For this impregnation, 2.8 mL of Pd(NH₃)₄ (NO₃)₂ (10 wt.% aqueous solution) was dissolved in 3.2 mL of Milli-Q water in a beaker. After stirring for 5 min under Ar flow, this solution was added to a 100 mL round flask containing 1.896 g TiO₂. Since the Pd(NH₃)₄²⁺ ion is stable in neutral and moderately basic environments and the point zero charge (PZC) of TiO₂ is about 5.36 [59], the impregnation was performed at pH~8. At these moderately basic conditions, the surface of TiO₂ was negatively charged, and the positively charged palladium complex ion could be adsorbed [60]. After sonication for 2 min under vacuum, the mixture was kept at room temperature for 2 h in a rotary evaporator (BÜCHI RE-111, 461 Water Bath (Büchi AG, Uster, Switzerland). Subsequently, the excess water was evaporated under vacuum at 90 °C. Finally, the impregnated solids were dried at 80 °C for 16 h and calcined in static air at 350 °C for 2 h (a heating ramp of 2 °C/min) to form PdO species and to remove remaining nitrate ions. Hereinafter, the catalyst synthetized by impregnation with cationic palladium complex ion is referred to as Pd-CI.

The second Pd/TiO₂ base catalyst was prepared with a nominal palladium loading of 5 wt.% by impregnation of TiO₂ with Pd(acac)₂ precursor. In this method, 300 mg of Pd(acac)₂ previously dissolved in 30 mL of chloroform was introduced into a 100 mL flask containing 1.896 g of TiO₂ support. After sonicating for 2 min, the impregnation process was carried out by rotating the flask at room temperature for 2 h. The chloroform was then removed under vacuum (50 °C), and the resulting slurry was dried in an oven at 80 °C for 16 h. Finally, the solid was calcined in static air at 350 °C for 2 h with a heating ramp of 2 °C/min to form PdO particles. Hereinafter, the catalyst prepared with Pd(acac)₂ precursor is referred to as Pd-NI (NI: impregnation with neutral precursor).

The ZnPd-CI and ZnPd-NI catalysts were prepared by impregnation of the corresponding base sample (Pd-CI or Pd-NI) with a Zn(acac)₂ precursor dissolved in methanol (17 mL·g_cat⁻¹). The amount of precursor used was calculated to obtain 23 wt.% ZnO in the final catalyst. After stirring for 5 min, this solution was added to a 100 mL flask containing the Pd/TiO₂ samples. After sonication for 2 min and stirring for 2 h, the solvent was removed at 50 °C under vacuum. Finally, the solid was dried at 80 °C for 16 h and then calcined in static air at 350 °C for 2 h (heating ramp of 2 °C·min⁻¹) to form ZnO particles. The reduced catalysts were obtained after treatment under H₂/Ar gas mixture (10 vol.%, 55.56 mL(N)·min⁻¹) under temperature-programmed conditions at a heating rate of 5 °C·min⁻¹ up to 450 °C and held at 450 °C for 1 h.

4.2. Catalysts Characterization

Inductively coupled plasma optical emission spectroscopy (ICP-OES) was applied to analyze the catalyst composition and the Pd and Zn loadings on the prepared catalysts using Perkin Elmer Optima 3300 DV apparatus. For the analysis, the solids were first digested with a mixture of HF, HCl, and HNO₃ acids in a microwave oven for 2 h, and then aliquots of the solution were diluted in 50 mL of deionized water.

Powder X-ray diffraction measurements were performed to determine the crystalline phases of the calcined, reduced, and spent catalysts by a Polycrystal X’Pert Pro PANalytical computerized diffractometer (Malvern Panalytical, Worcestershire, UK) equipped with a Cu Kα radiation source (λ = 0.15406 nm). The catalysts were scanned from 4° to 90° using 0.0335^o step size and 200 s step acquisition time. Crystalline phases were assigned according to the Joint Committee on Powder Diffraction Standards (JCPDS) diffraction cards, while crystal sizes were calculated using the Debye-Scherrer equation.

The textural properties of the calcined catalysts and the bare TiO₂ support were determined by nitrogen adsorption–desorption isotherms at −196 °C carried out in a Micromeritics TriStar 3000 apparatus (Micromeritics, Norcross, GA, USA). Prior to the experiments, the samples were degassed at 140 °C for 16 h. The specific surface area of samples was calculated by using the Brunner–Emmet–Teller (BET) method, while the cumulative pore volume at P/P₀ ~0.99 and pore size distributions (PSD) were determined using the Barrett–Joyner–Halenda (BJH) model [61]. To avoid the artefact of tensile strength (TSE), the pore size distribution (PSD) curves were calculated by applying the BET method to the adsorption branches of the N₂ isotherms.

X-ray photoelectron spectroscopy (XPS) spectra were collected using a SPECS GmbH (Berlin, Germany) System equipped with a PHOIBOS 150 9MCD energy analyzer. A non-monochromatized Al Kα X-ray source with a power of 200 W and voltage of 12 kV was used to irradiate the samples. The obtained binding energies were calibrated by using the C1 s peak at 284.8 eV. CasaXPS program was used for the peak’s deconvolution. The surface atomic ratios were calculated from the intensities of the XPS signals and normalized to their atomic sensitivity factors [62].

The reducibility of catalysts was analyzed by hydrogen temperature programmed reduction (H₂-TPR) experiments carried out in a quartz tube reactor (U-shaped) connected to a Balzer Prisma™ quadrupole mass spectrometer (QMS 200) (Pfeiffer Vacuum, Hessen, Germany) for online gas analysis, as well as in a Micromeritics TPD/TPR 2900 instrument equipped with a thermal conductivity detector (TCD). The temperature below ambient (−65 °C) was archived by immersing the reactor in a vessel containing a mixture of isopropanol and liquid nitrogen. The temperature was ramped from −33 °C to 450 °C at a heating rate of 5 °C-min⁻¹ (H₂-TPR) or 10 °C (TPR-MS) (50 mL/min) using a 10% H₂/Ar gas mixture. The H₂ signal (m/z = 2) and H₂O signal (m/z = 18) in the mass spectrometer were continuously recorded to follow the reduction processes of the palladium oxides.

The microstructure and particle distribution after the ZnPd/TiO₂ catalyst reduction at 450 °C and after reaction (Pd/TiO₂) were determined by using a JOEL (Peabody, MA, USA) JEM 2100 (LaB6) high-resolution transmission electron microscope (HRTEM) equipped with an ORIUS (Melbourne, Australia) SC1000 CCD digital camera (model 832) and operating at 200 kV and. The solids were ultrasonically dispersed in ethanol and deposited on an amorphous carbon-coated Cu TEM grid. The electron microscopic images were processed using a Gatan Digital Micrograph computer program capable of the Fourier analysis of image marked area (fast Fourier transformation, FFT) and its filtration (inverse FFT or IFFT). The creation of a diffraction image form selected area that was used to determine the lattice spacing of specific crystals, as it was described elsewhere [63]. The average particle diameter was calculated from the mean diameter frequency distribution with the formula: d = Σn_id_i/Σn_i, where n_i is the number of particles with particle diameter d_i in a certain range.

In situ diffuse reflectance infrared (DRIFT) spectra of the reduced samples before and after CO adsorption were recorded on a JASCO (Tokyo, Japan) FT/IR-6300 spectrometer equipped with a liquid nitrogen-cooled MCT detector and connected to a flow-system of purified gasses. First, the catalysts diluted with KBr (20% wt.%) were transferred into the DRIFTS cell and a DRIFT spectrum of dry KBr was recorded as a background. Then, the sample was reduced in situ with a 10% H₂/He gas mixture at 70, 250, 350, or 450 °C for 10 min (flow rate of 4 mL(N)·min⁻¹). After removing H₂ by purging in He for 10 min, the reduced sample was cooled in flowing He and the DRIFT spectra were recorded prior and after CO adsorption. Prior to CO adsorption, the spectra were recorded to evaluate acidity of the samples. Next, the sample was saturated with CO at 25 °C for 20 min using a 5 vol.% CO/He gas mixture, after which the gas phase was removed by He purge. All spectra were recorded at a spectral resolution of 4 cm⁻¹ with 300 accumulation scans.

4.3. Activity Tests

Catalytic tests were performed in a fixed-bed continuous-flow tubular reactor (PID Eng&Tech) placed in an electric furnace, as already described elsewhere [20]. The reaction was carried out at 250 °C and at 30 bar for 16 h using a mixture of reactant gasses CO₂/H₂/N₂ = 21.3%/64%/14.7% passing through the catalyst bed at a rate of 50 mL(N) min⁻¹ and a gas hourly space velocity (GHSV) set at 12,000 mL_CO2/H2·g_cat⁻¹·h⁻¹). The objective of the use of high space-velocity was to achieve low conversions to exclude the effect of equilibrium and to visualize the differences in deactivation. Briefly, the powder catalyst (0.25 g) mixed with 1.44 g of SiC was activated in situ by reduction in a flow of H₂/Ar gas mixture at 450 °C for 1 h (50 mL·min⁻¹; 1 atm). The reaction was carried out for 12 h, and the data in Table 3 represent the average during the entire test after stabilization. The effluent products were analyzed online by a gas chromatograph (Varian 450) (Palo alto, CA, USA) equipped with a thermal conductivity detector and two capillary columns: Msieve-5A and PoraBOND Q. For all catalysts, methanol, CH₄, and CO were the only products observed. GC and TD parameters are summarized in Table S1, while a representative GC analysis is shown in Figure S1. Taking into account that the only products detected were CH₃OH and CO, the CO₂ conversion, selectivity, and space time yield (STY) of products were calculated by Equations (1)–(3), respectively:

$$\mathrm{C}{\mathrm{O}}_2\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{v}\mathrm{e}\mathrm{r}\mathrm{s}\mathrm{i}\mathrm{o}\mathrm{n}={\displaystyle \frac{\mathrm{C}{\mathrm{O}}_2\mathrm{i}\mathrm{n}-\mathrm{C}{\mathrm{O}}_2\mathrm{o}\mathrm{u}\mathrm{t}(\mathrm{m}\mathrm{o}\mathrm{l}/\mathrm{m}\mathrm{i}\mathrm{n})}{\mathrm{C}{\mathrm{O}}_2\mathrm{i}\mathrm{n}(\mathrm{m}\mathrm{o}\mathrm{l}/\mathrm{m}\mathrm{i}\mathrm{n})}}\times 100$$

(1)

$$\mathrm{C}{\mathrm{H}}_3\mathrm{O}\mathrm{H}\left(\mathrm{C}\mathrm{O},{\mathrm{C}\mathrm{H}}_4\right)\left(\mathrm{\%}\right)\mathrm{s}\mathrm{e}\mathrm{l}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{i}\mathrm{t}\mathrm{y}={\displaystyle \frac{\mathrm{C}{\mathrm{H}}_3\mathrm{O}\mathrm{H}\left(\mathrm{C}\mathrm{O},{\mathrm{C}\mathrm{H}}_4\right)\mathrm{o}\mathrm{u}\mathrm{t}\left(\mathrm{m}\mathrm{o}\mathrm{l}\right)}{\mathrm{t}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{p}\mathrm{r}\mathrm{o}\mathrm{d}\mathrm{u}\mathrm{c}\mathrm{t}\mathrm{s}\left(\mathrm{m}\mathrm{o}\mathrm{l}\right)}}\times 100$$

(2)

$$\mathrm{S}\mathrm{T}\mathrm{Y}={\displaystyle \frac{\mathrm{C}{\mathrm{H}}_3\mathrm{O}\mathrm{H}\left(\mathrm{C}\mathrm{O}\right)\mathrm{o}\mathrm{u}\mathrm{t}(\mathrm{m}\mathrm{o}\mathrm{l}/\mathrm{m}\mathrm{i}\mathrm{n})}{\mathrm{P}\mathrm{d}\mathrm{i}\mathrm{n}\mathrm{c}\mathrm{a}\mathrm{t}\mathrm{a}\mathrm{l}\mathrm{y}\mathrm{s}\mathrm{t}\left(\mathrm{m}\mathrm{o}\mathrm{l}\right)}}\times 100$$

(3)

5. Conclusions

This work clearly demonstrated the pivotal importance of the genesis of the ZnPd/TiO₂ inverse catalyst for methanol synthesis via CO₂ hydrogenation. In this regard, the adequate selection of a Pd precursor is a decisive factor influencing the final physicochemical properties of the ZnPd/TiO₂ catalysts. The characterization of the Pd/TiO₂ base catalysts revealed that the use of the neutral precursor Pd(acac)₂ (Pd-NI) was more effective for the Pd dispersion than the use of the cationic Pd precursor Pd(NH₃)₄(NO₃)₂ (Pd-CI). Regardless of the Pd precursor, both Pd-NI and Pd-CI catalysts showed similar high activity and selectivity for CO in the hydrogenation of CO₂. The characterization of the ZnPd/TiO₂ catalysts confirms the role of the initial characteristics of the supported Pd particles on TiO₂ in the genesis of the ZnPd intermetallics prepared by the reverse incorporation of ZnO on the Pd/TiO₂. The better dispersion of PdO particles achieved on Pd-NI base catalysts provides a better contact with ZnO particles that facilitates the formation of ZnPd intermetallics. The comparison of the selectivity of ZnPd/TiO₂ with Pd/TiO₂ catalysts showed the change in the nature of the Pd active sites for the selective synthesis of CO by RWGS to methanol synthesis. CO formation occurs on the metallic Pd phase, whereas methanol production occurs exclusively when a ZnPd intermetallic phase is formed. The selectivity to methanol on the ZnPd/TiO₂ catalysts varies (64.7% on ZnPd-NI and 35.5% on ZnPd-CI) and its comparison with the characterization results indicated that it is directly related to the easier formation of the ZnPd intermetallics on the ZnPd-NI catalyst.