Catalytic oxidation of carbon monoxide (CO) has been widely applied in air purification, automotive exhaust emission control, and CO gas sensors [1]. Environmental concerns need to be considered in the design of a catalyst for CO oxidation, such as high temperature (200-600 ℃) for industrial and vehicle exhaust emission control and ambient temperature and high humidity for air purification of semi-enclosed space (underground tunnel or parking lot). Slow speed and frequency/cold start of vehicle results in the accumulation of CO that is usually beyond the normal emission level. It is a challenge to design a CO oxidation catalyst with high activity and stability at ambient temperature and high humidity.

Metal oxides, Co₃O₄ or Hopcalite catalyst with outstanding CO activity, are sensitive to moisture and quickly deactivated in the presence of trace water [2-4]. The weak stability of Hopcalite catalyst against vapor is caused by the strong adsorption of water on the catalyst surface relative to oxygen and CO adsorption [5, 6]. The deactivation of Co₃O₄ is ascribed to the hydrogenation of lattice oxygen and the formation of stable bicarbonates [7]. On the contrary, water facilitates CO oxidation on supported Pt/Pd catalysts [8-12], because OH is more active than lattice oxygen in reacting with CO to yield COOH on account of a smaller energy barrier [7].

Supported Wacker catalyst revealed excellent performance in CO oxidation, especially in the presence of water, and the CO oxidation rate increased 2-3 times compared to that of the case without water [13]. CO could be completely oxidized even at a temperature as low as -40 ℃ over the supported Wacker catalyst prepared by using the NH₃-coordination-impregnation method [14]; however, unfortunately, a continuous loss in activity was still observed at 0 ℃, 0.6% H₂O after 5 h [15]. A further study confirmed that the low temperature and high humidity rendered the whole reaction system unstable [16]. The presence of water facilitates Cu transfer from the surface to the pores, which impaired the contact between the Pd and Cu species [17] and inhibited the re-oxidization of Pd⁰ species [18].

To suppress the performance deterioration of catalysts by water, many methods have been proposed and attempted. One such technique involves the use of an hydrophobic coating for the catalysts for stable operation. Amorphous carbon-coated Mn/Cu-based Hopcalite nanoparticle catalysts showed a certain level of water vapor resistance for CO oxidation [5]. A proper coating of gas-permeable vapor-resistant polymers with a layer of nanofilm for the vapor-sensitive Co-Fe-based nanoparticle catalyst played a major role in its high stability observed during CO oxidation [19].

Another approach is to take advantage of organosilane to form oxane bonds of Si-O-metal and alkyl chain attached onto the surface of metal oxide. Polydimethylsiloxane (PDMS)- coated nanostructured manganese oxide materials exhibited a high selective adsorption of organic molecules in the presence of water [20]. Besides preventing the excess water from adsorbing on the surface of the catalyst [21], the pretreatment of Al₂O₃ support enhanced the metal-support interactions and benefited support reduction, which led to exceptional activity and stability for CH₄ combustion [22].

The above results trigged our interest in the study of how to enhance the stability of supported Wacker catalyst, therefore, a silane coupling agent (diethoxy dimethyl silane) was used to modify the support, and the stability of CO oxidation on Pd-Cu-Cl_x/Al₂O₃ catalysts was investigated for 0.6% H₂O at 0 ℃.

After dissolving diethoxy dimethyl silane in 40 mL ethanol, Al₂O₃ was added and refluxed at 80 ℃ under stirring for 3 h; it was then washed with ethanol and dried at 120 ℃ for 2 h to obtain modified Al₂O₃, which was labeled as Al₂O₃-M (M = modified). Supported Wacker catalysts (Pd-Cu-Cl_x/Al₂O₃) were prepared by the NH₃ coordination-impregnation method using PdCl₂·2H₂O and CuCl₂·2H₂O as precursors [14]. Cu was introduced first and calcined at 350 ℃ for 4 h before the impregnation of PdCl₂·2H₂O. The catalysts prepared from pristine Al₂O₃ and Al₂O₃-M were respectively denoted as R and M.

The surface areas of the samples were measured by N₂ adsorption-desorption at a low temperature (-196 ℃) on a NOVA 4200e surface area and porosity analyzer. The catalysts were degassed at 200 ℃ before the measurement.

The powder X-ray diffraction (XRD) patterns of the samples were obtained on a Brook D8 focus diffraction spectrometer with Cu-K_α radiation (40 kV, 40 mA, λ = 0.15406 nm) at the scanning rate of 6°/min. The average crystallite sizes of the catalysts were estimated using the Scherrer equation.

The contents of Pd and Cu were measured by an inductively coupled plasma-atomic emission spectrometry (ICP-AES) instrument (Varian 710-ES). The results showed that the pristine catalyst (R) consisted of 1.5 wt% Pd and 4.3 wt% Cu and the modified catalyst (M) contained 1.5 wt% Pd and 4.5 wt% Cu. ICP-AES analysis showed similar loadings of Pd and Cu on the catalysts.

H₂-temperature programmed reduction (H₂-TPR) was effected using a suitable equipment (Penxiang Co., Tianjin, China). The sample was heated in a flow of 5%H₂-95%N₂ from ambient temperature to 800 ℃ at the rate of 10 C/min. The amount of H₂ consumed was measured by a thermal conductivity detector (TCD) using H₂ consumption with CuO as the reference for quantitative analysis.

X-ray photoelectron spectroscopy (XPS) was performed at 25 ℃ with an AXIS Ultra DLD spectrometer using Mg K_α (hν = 1253.6 eV) radiation. Charged samples were avoided by setting the binding energy of adventitious carbon (C 1s) to 284.8 eV.

The in situ diffuse reflectance infrared Fourier transform spectra (DRIFTS) of the CO adsorbed onto the catalyst were recorded on a Nicolet Nexus 670 spectrometer at 25 ℃ every 2 min. The process followed this sequence: (1) mixture gas (0.15% CO/N₂) flowed into the reactor at the rate of 50 mL/min, (2) the gas passed through a water vapor saturator to the reactor, and (3) O₂ was added to the mixture gas.

The stability test for CO oxidation was performed in a quartz tubular U-tube fixed-bed reactor at 0 ℃, and a 0.2 g sample (20-40 mesh) was used in each test. The mixture 0.15%CO-20%O₂/balanced N₂ at the weight hour space velocity (WHSV) of 15000 mL/(g·h) was directed through a water vapor saturator immersed in a 0 ℃ water bath into the reactor. The water concentration in the feed gas was about 0.6%. The CO concentration was analyzed by an on-line gas chromatograph equipped with a FID detector and a methanator.

The water adsorption experiment was conducted at 10 ℃; 0.2 g of the catalyst was used. N₂ was directed at the rate of 50 mL/min through a water vapor saturator at 0 ℃, and the mass of the catalyst was measured at different time to obtain the amount of water adsorbed.

The stability of the pristine and modified catalysts are shown in Fig. 1. CO conversion on the pristine catalyst was equal to that of the modified catalyst after the first 1 h, which indicated that support silylation did not have an effect on catalyst activity. Although a slightly increased CO conversion was observed for the modified catalyst, the CO conversions finally became similar after 4 h. With a further increase in time, an obvious difference is observed in Fig. 1. The CO conversion of the pristine catalyst displayed a decreasing trend, and the value of the conversion sharply dropped from 86% to 52% even after 23 h. On the contrary, the conversion was rather stable after 150 h in the case of the modified catalyst. These experiments illustrated that support silylation was an efficient way to improve catalyst stability in the presence of water at a low temperature (0 ℃).

Water adsorption experiments were conducted at 100% relative humidity and 0 ℃, and the results are shown in Fig. 2. During the initial time, the water adsorption amount on the modified catalyst was higher than that on the pristine catalyst, but with further exposure to water, the water adsorption rate on the pristine catalyst was four times that on the modified catalyst after the first 5 h. When water adsorption reached the saturation level, the saturated water adsorption amount on the modified catalyst was only one-third that on the pristine catalyst. The water adsorption rate and saturation amount significantly decreased after support silylation, which confirmed that support silylation made the surface hydrophilic.

The surface areas of the samples are provided in Table 1. The surface areas of both pristine and modified catalysts have similar values, suggesting that support pretreatment had no effect on the surface area.

The XRD patterns of the pristine/modified catalysts and support are shown in Fig. 3. No characteristic Pd diffraction peaks were observed, indicating that either the Pd species was highly dispersed on the support surface or Pd loading was below the detection limit of the XRD equipment.

The XRD patterns of the pristine and modified supports reveal that silylation has no effect on the phase structure. The characteristic diffraction peaks at 16.3°, 32.7°, and 40.2° were attributed to the Cu₂(OH)₃Cl species. The crystallite sizes are 26 and 40 nm for the pristine and modified catalysts, according to the Scherrer equation, based on the (101) peak of Cu₂(OH)₃Cl. Compared with the pristine catalyst, support silylation increased the crystallite sizes of Cu₂(OH)₃Cl in the case of the modified catalyst.

The H₂-TPR profiles and H₂ consumption of catalysts are shown in Fig. 4. and Table 2, respectively. Pd/Al₂O₃ had an obvious H₂ reduction peak at 38 ℃, while Cu/Al₂O₃ revealed two such peaks at 239 and 311 ℃. With the introduction of Cu to Pd/Al₂O₃, two distinct reduction peaks at 100-250 ℃ and 250-400 ℃ were observed. Compared with the single metal catalysts, it was known that the presence of Pd promoted the reduction of Cu species. The intense reduction peak with a shoulder in the temperature range 100-250 ℃ can be de-convoluted into two reduction peaks: one (peak A) was ascribed to the reduction of Pd species and the other (peak B) to the reduction of copper species, which was in close contact with the Pd species. The high-temperature reduction peak (peak C) at 250-400 ℃ was attributed to the reduction of isolated copper species [23, 24].

Owing to the same Pd loading amounts, the H₂ consumption during low-temperature reduction was dominated by the amount of Cu species in close contact with the Pd species. The lower reduction temperature and higher H₂ consumption values demonstrated the higher reducibility of the catalyst.

Although the reduction temperature of peak B slightly shifted to a higher value, its H₂ consumption value increased from 0.215 mmol/g (pristine catalyst) to 0.302 mmol/g (modified catalyst). Support pretreatment decreased the amount of isolated Cu species, and the reduction temperature was shifted to a low value. The H₂ consumption ratio of peak B/(B+C) increased from 0.44 to 0.54, indicating that most Cu species were in close contact with Pd species after support pretreatment.

The XPS patterns of Pd 3d, Cu 2p, and Auger Cu LMM spectra for the reference and modified catalysts are shown in Fig. 5, and the surface compositions and chemical states listed in Table 1.

The Pd 3d spectrum exhibited Pd²⁺ and Pd⁺ peaks at 337.5 and 335.7 eV, respectively, while the Cu 2p spectrum shows two peaks at 934.4 and 932.4 eV, respectively. It is easy to infer that the peak at 934.4 eV is attributable to Cu²⁺, however, it is difficult to distinguish the Cu⁰ and Cu⁺ species due to their similar binding energies in the Cu 2p XPS pattern. Auger electron spectroscopy of the Cu species was performed, and the results are shown in Fig. 5. The binding energies of Cu⁺ and Cu⁰ are 570 and 568 eV, respectively, in Auger Cu LMM spectroscopy [25]. Only one peak at 570 eV was detected in all the samples, so the presence of the metal Cu species on the catalyst surface was excluded. The Cu species existed in the form of Cu²⁺ and Cu⁺.

As shown in Table 1, the effects of support modification on Cu ion distribution are more obvious than those on Pd ion distribution. The ratio Cu/Pd even increased after support modification. A higher ratio of Cu/Pd indicated closer contact between the Pd and Cu species. This closer contact facilitated the reduction of the Cu species, as confirmed by H₂-TPR. It is worth mentioning that the surface concentration of Cl decreased from 4.3% to 2.8% after support pretreatment, indicating less residual Cl on the modified catalyst.

In order to further investigate the effect of support pretreatment on CO oxidation stability, CO adsorption and oxidation were further investigated by DRIFT. In in situ DRIFT spectra, the bands at 2300-2400 cm^-1 are ascribed to gaseous CO₂ [26-28], while the bands in the regions 1600-1700 and 1500-1300 cm⁻¹ are assigned to the OH groups and carbonates, respectively [29, 30]. The bands at 2126 and 2162 cm⁻¹ correspond to the adsorbed CO in the Cu⁺ complex (Cu⁺-CO) and the terminal CO groups in the Pd²⁺ complex (Pd²⁺-CO). The bands at 1991, 1934, and 1810 cm⁻¹ originate from the bridge-bonded CO on the Pd metal surface (Pd₂-CO), bridged carbonyl ligands in the Pd⁺ complexes (Pd⁺-CO), and triply bonded CO on metallic Pd (Pd₃-CO) [31-33].

In situ DRIFT spectra of CO adsorbed over catalysts in CO/Ar as functions of time are illustrated in Fig. 6. Weak bands for Pd²⁺-CO (2158 cm⁻¹) and Cu⁺-CO (2126 cm⁻¹) appeared first after 4 min exposure to CO on the pristine catalyst. The appearance of Pd₂-CO (1991 cm⁻¹), lagging behind Pd⁺-CO (1934 cm⁻¹), indicated that the Pd metal originated from the reduction of the Pd⁺ species by CO. Meanwhile, the carbonate species (1300-1500 cm⁻¹) kept intensive with increasing concentration of Pd₂-CO, suggesting that this species originated from the transformation of Pd⁺ to Pd metal.

Unlike CO adsorption on the pristine catalyst, Pd²⁺-CO, Cu⁺-CO, and Pd⁺-CO were clearly observed simultaneously after introducing CO for 4 min on the modified catalyst. Pd₂-CO appeared after exposure in CO for 8 min, and no bands corresponding to the carbonate were observed. However, it is worth mentioning that the band ratio of Cu⁺-CO/ Pd⁺-CO over the pristine catalyst was higher than that over the modified catalyst. The intense CO adsorption on the Cu⁺ species indicated the facile oxidation of the metallic Pd by high-valence Cu species.

Water was introduced after CO adsorption equilibrium, and the results are shown in Fig. 7. Pd²⁺-CO and Cu⁺-CO maintained their peak intensities with the increase in the exposure time to water, while Pd⁺-CO revealed an opposite trend. The disappearance of Pd⁺-CO was accompanied by the intensification of CO adsorption on metallic Pd (Pd₂-CO, Pd₃-CO). As a result of the admission of water, the carbonate species was further enhanced in amount compared to that observed in the absence of water in the feed gas.

Oxygen gas was introduced after exposure to water for 20 min, which stabilized the bands of Pd²⁺-CO and Cu⁺-CO quickly. Because the redox potential of Cu⁺/Cu²⁺ (0.15 V) was much lower than that of Pd⁰/Pd²⁺ (0.92 V), O₂ would oxidize Cu⁺ first rather than Pd⁰. Shen et al. [18] also confirmed the fast oxidation of Cu⁺ to Cu²⁺ by O₂.

Although the absorption bands for the two catalysts displayed similar tendencies, compared with the modified catalyst, a faster disappearance of Pd⁺-CO over the reference catalyst was still observed. Combined with Fig. 6, it further confirmed that the presence of water favored the oxidation of metallic Pd by the Cu²⁺ species. The gradually intensifying Pd²⁺-CO band seemed to contradict the competitive adsorption between CO and water on active sites. Considering the first step in CO oxidation, which is the fast reduction of Pd²⁺ to Pd⁺ by CO, it is suggested that the presence of water strongly inhibited the reduction of Pd²⁺ by CO, which led to the deactivation.

Water cannot be avoided in the environment, so it is important to explore catalysts with high water tolerances. A hydrophobic catalyst or an active component inert to water revealed excellence performance in a moist environment. Although small amounts of water could promote CO oxidation over the supported Wacker catalyst, with the increase in the amount of deposited water, the active sites would be covered and the close-knit structure between the Pd and Cu species broken [16]. Such detrimental effects lead to both reversible and irreversible deactivation.

Considering the solubility of CuCl₂ in water, it seems that the decrease in the hydrophilic support will be efficient to control CuCl₂ transfer in the presence of water. Therefore, diethoxy dimethyl silane was used to form oxane bonds to increase the hydrophobicity of the support. Silylanization made the surface of the modified catalyst more hydrophobic compared to that of the pristine catalyst. The catalyst based on the pretreated support revealed no obvious decline in activity even after 150 h at 0 ℃ and 100% relative humidity.

Support silylanization had no effect on specific surface area (Table 1), but it increased the particle size of Cu₂(OH)₃Cl from 26 to 40 nm (Fig. 3 and Table 1). Meanwhile, the amount of the Cu species in close contact with the Pd species enhanced according to H₂-TPR and XPS analyses (Figs. 4 and 5). It should be mentioned that the concentration of chlorine in the case of the modified catalyst was much lower than that of the pristine catalyst (Table 1). With the increase in the amount of Cl species coordinated with the Pd species, the CO substitution energy will increase accordingly, which inhibits CO adsorption [34].

The water adsorption experiment showed that silylanization inhibited water deposition on the modified support (Fig. 2). The CO oxidation stability was correlated with the water concentration, while the role of water could not be clearly demonstrated. Therefore, in situ DRFTIR was used to clarify the relationship between them. The reaction mechanism for CO oxidation on supported Wacker-type catalysts was similar to that involving homogeneous catalysts: Pd species are the active sites for CO oxidation, and CO oxidation is achieved by the dual redox cycle of the Pd and Cu species. Although it is well accepted that Pd ions are the active component in CO oxidation, the nature of Pd itself is ambiguous. Choi et al. [35] thought that the structure of the Pd species was similar to that of PdCl₂, and was probably modified by a carbonyl ligand. DFT modulation suggested that it should exist as PdOH, based on the de-chlorination of PdCl₂ [36].

As shown in Fig. 6, the presence of CO accompanied by the appearance of the Pd⁺ species suggest the easily reduction of Pd²⁺ to Pd⁺ by CO. The faster disappearance of Pd⁺ indicated that this species was more active than Pd²⁺ during CO oxidation, especially in the presence of water (Fig. 7). Meanwhile, most of carbonate species accumulated with the transformation of Pd⁺ to Pd⁰, which had detrimental effects on the stability of CO oxidation. The high concentration of Cu⁺ on hydrophilic catalysts indicated that water favored the reoxidation of Pd⁰ by Cu²⁺. However, it is worth mentioning that the Pd²⁺ species intensified in the presence of water. Although the accumulated carbonate species had negative effects on CO oxidation, considering the important role of Pd⁺ in the oxidation, the deactivation in the presence of water was mainly due to its inhibition to produce Pd⁺.

Silylanization is an efficient way to improve hydrophobicity, so diethoxy dimethyl silane (C₆H₁₆O₂Si) was used to modify Al₂O₃ and prepare the hydrophobic catalysts Pd-Cu-Cl_x/Al₂O₃. This support pretreatment significantly promoted the CO oxidation stability, and CO conversion remained at 78% even after 150 h at saturated humidity and freezing point. Support silylanization improved catalyst hydrophobicity and promoted the reduction of the Cu species. Compared with the pristine catalyst, the concentration of Cl ions decreased significantly, which benefited CO adsorption.

Pd⁺ species was more active than Pd²⁺ species in CO oxidation, and the oxidation was achieved by the dual redox cycle of the Pd and Cu species. The presence of water favored CO oxidation over active Pd⁺ and Pd⁰ reoxidation by Cu²⁺. Meanwhile, water also inhibited the formation of active Pd⁺ species and helped to produce carbonate species. Compared with the form of the carbonate species, the inhibition of the production of the active Pd⁺ species played a main detrimental role in determining the stability.