Metal-Support Synergy of Supported Gold Nanoclusters in Selective Oxidation of Alcohols

: Gold catalysts have been reported to exhibit good performance in aerobic oxidation of alcohols, but the intrinsic origin of the catalytic reactivity is still illusive. The catalyst preparation method, the morphology of the gold particles, and even the support might be key factors that determine the activity. Here, we prepared a series of gold catalysts with di ﬀ erent supports, i.e., the hydrotalcite (HT), ZnO, MgO, Al 2 O 3 , and SiO 2 , by using the atomically controlled Au 25 nanoclusters (NCs) as the gold precursor. The characterization results of the X-ray di ﬀ raction (XRD), UV-vis and transmission electron microscopy (TEM) show that the gold particles were mostly uniformly distributed on the supports, with a mean particle size within 3 nm. In aerobic oxidation of benzyl alcohol, the MgAl-HT-and Al 2 O 3 -supported Au 25 NCs display good performances, with turnover frequency (TOF) values of ~2927 and 2892 h − 1 , respectively, whereas the SiO 2 -, MgO-, and ZnO-supported analogues show much inferior activity. The high resolution TEM and X-ray photoelectron spectra (XPS) results suggest that the interactions between gold and the supports in di ﬀ erent samples are di ﬀ ering, which inﬂuences the morphology and the nature of gold. Our results further point to the importance of acid-base property of the support and the metal-support synergy rather than the gold particle size alone in achieving high-performance in selective alcohol oxidation. Moreover, this work provided a good way to design gold catalysts with controllable sizes that is crucial for understanding the reaction process in aerobic oxidation of alcohol. oxidation reaction of benzyl without the addition of any inorganic base. The N 2 physical adsorption, X-ray di ﬀ raction (XRD), UV-vis, transmission electron microscopy (TEM), high resolution TEM, X-ray photoelectron spectra (XPS), and NH 3 / CO 2 / O 2 -temperature programmed desorption (TPD) property measured regia to 50 mL for analysis. Power X-ray di ﬀ raction were determined on a PW3040 / 60 X’Pert PRO (PANalytical, Almelo, Netherlands) di ﬀ ract meter equipped with a Cu K α radiation source ( λ = 0.15432 nm) operating at 40 kV and 40 mA. The continuous mode was used at a scanning speed of 10 ◦ / min in the scanning angle (2 θ ) range from 10 ◦ to 90 ◦ . UV-vis spectra were recorded on a Cintra (Austin, USA) (GBC) apparatus with BaSO 4 as the reference at room temperature. A continuous mode at a scanning speed of 100 nm / min was used from 200 to 800 nm. The texture properties of the catalysts were measured by N 2 -physical adsorption-desorption test at 77 K using an AutoSorb-1 instrument (Florida, USA). Prior to the measurements, the catalysts were treated at vacuum for 2 h at 120 ◦ C. The pore volume and pore size distribution were determined from the adsorption branches using the NLDFT (the nonlocal density functional theory) method. The transmission electron microscopy (TEM) was recorded with a JEM-2100F ﬁeld emission (Tokyo, TEM. The microstructure of the catalysts was further observed by high resolution transmission electron microscopy (HRTEM) at 200 kV. The temperature-programmed desorption of carbon dioxide (CO 2 -TPD), ammonia (NH 3 -TPD), and oxygen (O 2 -TPD) experiments were performed on AutochemII Instruments (Norcross, Georgia, USA) with a thermal conductivity detector. Prior to the analysis, about 50 the catalysts were added into a U-type quartz tube reactor, and then the sample was a of helium at 300 ◦ C for 1 h. After the temperature decreased to 80 ◦ C, the catalysts saturated CO 2 , NH 3 and O 2 80 ◦


Introduction
Supported gold catalysts have attracted a lot of attention in the past few decades because of Hutchings and Haruta's pioneer work [1,2]. Highly dispersed gold nanocatalysts displayed good performances in various reactions, including the low temperature CO oxidation [2][3][4][5][6], chemoselective a The actual loadings of gold were evaluated by ICP-AES measurement; b The specific surface area and pore structure were evaluated by N 2 -physical adsorption-desorption test; c The mean particle sizes were calculated from TEM measurement.
The pore structure parameters such as the specific area (S BET ), the pore volume (V mesopore ), and the pore diameter (D pore ) were evaluated by an N 2 -physical adsorption-desorption test. The results were showed in Figure 1 and Table 1. From the results, the Au 25 /MgO-300 and Au 25 /ZnO-300 catalysts showed the adsorption/desorption isotherms almost overlapping, which suggests that the specific area and the pore volume are too small. In addition, the isotherms of the Au 25 /MgAl-300 and Au 25 /ZnAl-300 catalysts were horizontal and parallel to each other at lower P/P 0 ratios, and they arise steeply when the relative pressures approach 1. Thus, the Au 25 /MgAl-300 and Au 25 /ZnAl-300 catalysts might possess slit-shaped pores resulted from the layered structure [40]. Moreover, the Au 25 /Al 2 O 3 -300 and Au 25 /SiO 2 -300 catalysts displayed a typical shape of type IV isotherm according to the International Union of Pure and Applied Chemistry (IUPAC) classification [41,42], with a sharp step at intermediate relative pressures. This type of isotherm indicated the presence of textural mesopores [42].  Table 1 displayed the values of specific area, the pore volume, and the pore diameter. Over the different supported gold catalysts, the SBET decreased as the following order: Au25/SiO2-300 > Au25/Al2O3-300 > Au25/MgAl-300 > Au25/ZnAl-300 > Au25/MgO-300 > Au25/ZnO-300. To be noted, the SiO2 supported gold catalyst possessed a comparable high specific area of 367.1 m 2 /g due to the functionalization of 3-aminopropyltriethoxysilane (APTES) to the surface. Otherwise, the Au25 NCs was hard to infuse into the support and the loadings of gold were below 0.1%. The Au25/Al2O3-300 catalyst had the second largest specific area of 235.4 m 2 /g, with similar pore diameter of Au25/SiO2-300. This might be beneficial for dispersion of the gold and inhibiting the growth and agglomeration of gold species on the surface [43]. MgAl-hydrotalcite (HT) and ZnAl-HT supported gold NCs had a moderate SBET (131.2 and 112.6 m 2 /g) and wide distribution of pore diameter due to their layer structures. In addition, the Au25/MgO-300 and Au25/ZnO-300 catalysts had very low specific area of 10.1 and 4.0 m 2 /g, respectively, with the small pore volume and wide distribution of pore diameter. Notably, the pore structures of the bare oxide supports were almost the same as the corresponding gold catalysts ( Figure S1 and Table S1), which display the intact structure of the oxides after the loading of gold. The mean particle sizes of gold were then calculated from the TEM measurement, which would be discussed in the next paragraph.
The structural properties of the supported Au25 NCs catalysts and the corresponding supports were evaluated from X-ray diffraction (XRD) measurement (Figures 2 and S1). From the XRD patterns, all the catalysts and supports after calcination displayed the specific structure of the precursor or metal oxides. This means that the preparation of six different supported gold catalysts was successful. As showed in Figure 2a, the diffraction peaks could be identified from the order of the ZnO phase, which was similar with the structure of ZnO in Figure 2b. It implied that the ZnAlhydrotalcite had transformed into the atomic mixed oxide after calcination. In addition, the same phase between the supports and the catalysts in Figures 2 and S2 suggested the adsorption of gold into the supports had not changed the structure. Figure 2c showed the diffraction peaks of MgAlhydrotalcite and MgO. Based on the previous study [15], the calcination temperature of 300 °C might not be enough to transform the hydrotalcite into the complete magnesium oxides, while, at this temperature, the MgAl-hydrotalcite supported gold catalysts could display good performances in different reactions [15,38]. In addition, the gold catalysts calcined at the same temperature was imperative for comparison of the different gold catalysts, since the size distribution largely depends  Table 1 displayed the values of specific area, the pore volume, and the pore diameter. Over the different supported gold catalysts, the S BET decreased as the following order: Au 25 /SiO 2 -300 > Au 25 /Al 2 O 3 -300 > Au 25 /MgAl-300 > Au 25 /ZnAl-300 > Au 25 /MgO-300 > Au 25 /ZnO-300. To be noted, the SiO 2 supported gold catalyst possessed a comparable high specific area of 367.1 m 2 /g due to the functionalization of 3-aminopropyltriethoxysilane (APTES) to the surface. Otherwise, the Au 25 NCs was hard to infuse into the support and the loadings of gold were below 0.1%. The Au 25 /Al 2 O 3 -300 catalyst had the second largest specific area of 235.4 m 2 /g, with similar pore diameter of Au 25 /SiO 2 -300. This might be beneficial for dispersion of the gold and inhibiting the growth and agglomeration of gold species on the surface [43]. MgAl-hydrotalcite (HT) and ZnAl-HT supported gold NCs had a moderate S BET (131.2 and 112.6 m 2 /g) and wide distribution of pore diameter due to their layer structures. In addition, the Au 25 /MgO-300 and Au 25 /ZnO-300 catalysts had very low specific area of 10.1 and 4.0 m 2 /g, respectively, with the small pore volume and wide distribution of pore diameter. Notably, the pore structures of the bare oxide supports were almost the same as the corresponding gold catalysts ( Figure S1 and Table S1), which display the intact structure of the oxides after the loading of gold. The mean particle sizes of gold were then calculated from the TEM measurement, which would be discussed in the next paragraph.
The structural properties of the supported Au 25 NCs catalysts and the corresponding supports were evaluated from X-ray diffraction (XRD) measurement ( Figure 2 and Figure S1). From the XRD patterns, all the catalysts and supports after calcination displayed the specific structure of the precursor or metal oxides. This means that the preparation of six different supported gold catalysts was successful. As showed in Figure 2a, the diffraction peaks could be identified from the order of the ZnO phase, which was similar with the structure of ZnO in Figure 2b. It implied that the ZnAl-hydrotalcite had transformed into the atomic mixed oxide after calcination. In addition, the same phase between the supports and the catalysts in Figure 2 and Figure S2 suggested the adsorption of gold into the supports had not changed the structure. Figure 2c showed the diffraction peaks of MgAl-hydrotalcite and MgO. Based on the previous study [15], the calcination temperature of 300 • C might not be enough to transform the hydrotalcite into the complete magnesium oxides, while, at this temperature, the MgAl-hydrotalcite supported gold catalysts could display good performances in different reactions [15,38]. In addition, the gold catalysts calcined at the same temperature was imperative for comparison of the different gold catalysts, since the size distribution largely depends on the heat treatment temperature [32,37]. Subsequently, Figure 2d shows the intact phase structure of MgO, which maintained the originate state of the supports. The diffraction peaks in Figure 2e could be ascribed to the structure of γ-Al 2 O 3 [21]. Due to the low temperature of thermal decomposition, the precursor of Al(OH) 3 dehydrated and transformed to the γ-Al 2 O 3 instead of the α-Al 2 O 3 [21], the latter was generally utilized for the corundum. Figure 2f displayed a broad peak at 22.5 • , which was assigned to the amorphous SiO 2 [44]. In addition, no diffraction peaks of gold appeared on all the catalysts. This was caused by the low loadings of gold.
Catalysts 2020, 10, x FOR PEER REVIEW 5 of 20 on the heat treatment temperature [32,37]. Subsequently, Figure 2d shows the intact phase structure of MgO, which maintained the originate state of the supports. The diffraction peaks in Figure 2e could be ascribed to the structure of γ-Al2O3 [21]. Due to the low temperature of thermal decomposition, the precursor of Al(OH)3 dehydrated and transformed to the γ-Al2O3 instead of the α-Al2O3 [21], the latter was generally utilized for the corundum. Figure 2f displayed a broad peak at 22.5°, which was assigned to the amorphous SiO2 [44]. In addition, no diffraction peaks of gold appeared on all the catalysts. This was caused by the low loadings of gold. UV-vis spectra could also display some information about size distribution of the Au catalyst [37,45]. For the six supported gold catalysts in this work the UV-visible diffuse reflectance spectra were recorded from 200-800 nm at solid mode with the BaSO4 as the reference ( Figure 3). Obviously, there were broad absorption bands located at 520-550 nm on all the catalysts, which is characteristic of the plasmon resonance band of gold nanoparticles [45]. As seen from the spectra, all of the bands had a large half peak width and no sharp peaks appeared in the spectra, which means that the gold particles were not aggregating [37]. UV-vis spectra could also display some information about size distribution of the Au catalyst [37,45]. For the six supported gold catalysts in this work the UV-visible diffuse reflectance spectra were recorded from 200-800 nm at solid mode with the BaSO 4 as the reference ( Figure 3). Obviously, there were broad absorption bands located at 520-550 nm on all the catalysts, which is characteristic of the plasmon resonance band of gold nanoparticles [45]. As seen from the spectra, all of the bands had a large half peak width and no sharp peaks appeared in the spectra, which means that the gold particles were not aggregating [37].  The TEM images and size distribution histograms of supported Au25 NCs catalysts after heat treatment were shown in Figure 4. The dark spots presented in images were defined as the gold particles. From the electronic photos, the gold particles were highly dispersed on various supports. The average particles sizes of gold were within 3 nm with narrow distributions in most of the catalysts, which are 3.0 ± 1.2, 2.6 ± 0.9, 2.7 ± 0.7, 2.7 ± 1.2, 2.7 ± 1.1 nm for the Au25/MgAl-300 (Figure  The TEM images and size distribution histograms of supported Au 25 NCs catalysts after heat treatment were shown in Figure 4. The dark spots presented in images were defined as the gold particles. From the electronic photos, the gold particles were highly dispersed on various supports. The average particles sizes of gold were within 3 nm with narrow distributions in most of the catalysts, which are 3.0 ± 1.2, 2.6 ± 0.9, 2.7 ± 0.7, 2.7 ± 1.2, 2.7 ± 1.1 nm for the Au 25 /MgAl-300 (Figure 4a), Au 25 /MgO-300 (Figure 4b), Au 25 /ZnAl-300 (Figure 4c), Au 25 /ZnO-300 ( Figure 4d) and Au 25 /Al 2 O 3 -300 ( Figure 4e) catalysts, respectively. It meant the Au 25 NCs could be stabilized on MgAl-HT, ZnAl-HT, MgO, ZnO, and Al 2 O 3 supports, whereas the mean particle size of the Au 25 /SiO 2 -300 catalyst was about two times larger than that on others, on which the particle size of gold was 5.9 ± 1.7 nm (Figure 4f). The reason might be caused by the weak interaction between gold and the SiO 2 . In this situation, the gold NCs might grow to be the bigger ones on the surface without the anchor of the carrier. This led to an increased particle size of gold in Au 25 /SiO 2 -300. The supposition could be verified by high resolution TEM (HRTEM). The representative electronic images were displayed in Figure S3. From the photos, the gold particles on the Au 25 /SiO 2 -300 catalyst showed the round shape, yet, over the other catalysts, the gold particles were hexagonal or polygonal. The phenomenon revealed the fact that the gold particles weakly interacted with the supports would lead to a smooth interface, whereas the gold strongly interacted with the supports would result in an irregular shape [33,46]. Thus, the gold and the supports could share more terraces and steps, which facilitated the anchoring of gold on the surface [46,47]. Furthermore, by analyzing the direction of lattice fringes, there was obvious tropism between gold and the supports ( Figure S3a-e), which further suggests the possibility of the strong interaction between gold and the support [37,38].
It is worth noting that the low resolution TEM on the low contrast applied in this study could not reveal the small gold particles on the surface. Thus, some small gold particles were not included in calculation of the mean particle size. It might be the reason why the mean particle sizes in this work were larger than that in our previous study [37,38], even though the loadings of gold were somewhat different. Further study about high resolution TEM or HAADF-STEM was needed to give more precise value. Still, we demonstrated that the preparation method applied in this work was suitable for the synthesis of gold catalysts with controllable sizes. The gold particles were uniformly distributed on different supports such as MgAl-HT, ZnAl-HT, MgO, ZnO, and Al 2 O 3 . In addition, it can exclude the influence of the preparation method on the study of the gold catalysts.
The XPS analyses were a good characterization to test the electronic state of the catalysts. To shed light on the electronic effect, the valent state of gold and the other elements were analyzed. Thus, one can identify whether the interaction between gold and the support existed in the catalysts. In general, the interaction between gold and the support was reported to make an essential difference in catalytic reactions [33,34]. Thus, the supports could modify the electronic properties of gold by this method, and their performances could be adjusted.
In this study, the XPS spectra of Au 4f were presented in Figure 5. All catalysts showed the characteristic bands of Au 4f 5/2 and Au 4f 7/5, and their binding energies of Au 4f 7/2 lines were observed at 83.1, 83.2, 83.2, 83.7, 83.6, and 83.9 eV for Au 25 /MgAl-300, Au 25 /MgO-300, Au 25 /ZnAl-300, Au 25 /ZnO-300, Au 25 /Al 2 O 3 -300, and Au 25 /SiO 2 -300 catalysts, respectively. Since the metallic gold of Au(0) 4f 7/2 line was generally reported at 84.0 eV [15,33], and the valent state of gold in Au 25 /SiO 2 -300 catalyst was 83.9 eV; thus, accordingly, the gold on the Au 25 /SiO 2 -300 catalyst was close to Au(0). Namely, the interaction between gold with the SiO 2 was week. This result was very consistent with the previous HRTEM results-that the weak interaction led to the round shape of gold on the SiO 2 support ( Figure S3f).  It is worth noting that the low resolution TEM on the low contrast applied in this study could not reveal the small gold particles on the surface. Thus, some small gold particles were not included Au(0) 4f 7/2 line was generally reported at 84.0 eV [15,33], and the valent state of gold in Au25/SiO2-300 catalyst was 83.9 eV; thus, accordingly, the gold on the Au25/SiO2-300 catalyst was close to Au(0). Namely, the interaction between gold with the SiO2 was week. This result was very consistent with the previous HRTEM results-that the weak interaction led to the round shape of gold on the SiO2 support ( Figure S3f). However, the gold in Au25/Al2O3-300 and Au25/MgO-300 catalysts showed negative valent state in comparison with the Au25/SiO2-300, of which the binding energies were 83.7 and 83.6 eV, respectively. In particular, the valent states were even lower in Au25/MgAl-300, Au25/ZnAl-300, and Au25/ZnO-300 catalysts. The binding energies were 83.2, 83.1, and 83.2 eV for Au 4f 7/2 lines, respectively. This strongly suggests a stronger interaction between gold and the supports existed in these systems, where the electron-transfer could occur between the two entities. The XPS spectra of O 1s were also provided in Figure S4. For example, two peaks located at 530.0 and 531.4 eV in However, the gold in Au 25 /Al 2 O 3 -300 and Au 25 /MgO-300 catalysts showed negative valent state in comparison with the Au 25 /SiO 2 -300, of which the binding energies were 83.7 and 83.6 eV, respectively. In particular, the valent states were even lower in Au 25 /MgAl-300, Au 25 /ZnAl-300, and Au 25 /ZnO-300 catalysts. The binding energies were 83.2, 83.1, and 83.2 eV for Au 4f 7/2 lines, respectively. This strongly suggests a stronger interaction between gold and the supports existed in these systems, where the electron-transfer could occur between the two entities. The XPS spectra of O 1s were also provided in Figure S4. For example, two peaks located at 530.0 and 531.4 eV in Au 25 /ZnAl-300 and Au 25 /ZnO-300 catalysts could be ascribed to O 2ions on the wurtzite structure of hexagonal Zn 2+ ion array and the O 2− ions in oxygen-deficient regions within the ZnO matrix, respectively [48,49]. It indicates the different valent states and structural features of the supports that appeared in these catalysts.
The basic and acid properties were evaluated by CO 2 and NH 3 pulse adsorption followed by TPD measurements. The results were listed in Figures 6 and 7 and Table S2. The total basicity and acidity could be calculated by the amounts of pulse adsorption (Table S2) [50,51]. From the aspect of basicity, the Au 25 /MgAl-300 catalyst exhibited the maximum basic sites, with the adsorbed CO 2 of 1130 µmol/g. The following order was: Au 25 /Al 2 O 3 -300 (460 µmol/g) > Au 25 /ZnAl-300 (310 µmol/g) > Au 25 /MgO-300 (90 µmol/g) > Au 25 /SiO 2 -300 (20 µmol/g) > Au 25 /ZnO-300 (10 µmol/g). In addition, the profiles of CO 2 -TPD experiment about the strength of base sites over the different supported gold catalysts were showed in Figure 6. From the results, the Au 25 /MgAl-300 catalyst presented one obvious peak at 200-600 • C, which was assigned to the medium base resulted from the M 2+ -O 2− pairs [20,25]. The Au 25 /Al 2 O 3 -300 catalyst showed a broad band at 200-800 • C, which indicated the weak sites, medium sites, and strong sites of the base might be coexisting in the catalyst. The Au 25 /ZnAl-300 and Au 25 /MgO-300 catalysts displayed a peak at 200-400 • C, while the Au 25 /SiO 2 -300 and Au 25 /ZnO-300 catalysts showed almost no basicity. To be noted, the ZnO was generally considered to be a basic oxide and had been used in several base-catalyzed processes [52,53]. However, some others presented no or low basicity of ZnO in different systems [54][55][56], which agree well with our work. We speculated that the low basicity of ZnO in this study might be result from the heat pretreatment at 500 • C before use, since the basic sites of zinc oxide might be decreased at elevated temperatures [55,56].
Among them, the Au 25 /Al 2 O 3 -300 catalyst possessed the three kinds of acid sites. The Au 25 /MgAl-300 catalyst displayed a strong desorption peak of NH 3 at 300-500 • C, which was assigned to the medium acid sites. The Au 25 /SiO 2 -300 catalyst showed a broad band at 350-800 • C, which could be ascribed to the medium and strong acid sites. The Au 25 /ZnO-300, Au 25 /MgO-300 and Au 25 /ZnAl-300 catalysts possessed weak and/or medium acid sites.

The Catalytic Activities of Different Supported Gold Catalysts
The aerobic oxidation of benzyl alcohol was evaluated to test the catalytic performance of the supported Au 25 NCs catalysts. The general reaction route was illustrated in Scheme 1. Firstly, the benzyl alcohol (BA) was oxidized to benzaldehyde (BZ). Then, the benzaldehyde could be oxidized to benzoic acid (BC). At the same time, BA could further react with BZ or BC under the certain conditions to form the benzyl benzoate (BO) [20,46]. Notably, the benzene and toluene were also probably the products from the reaction, which has been demonstrated in various previous works [57,58]. In this system, different supported gold catalysts for aerobic oxidation of BA were employed. The results were listed in Table 2. Notably, no any benzoic acid could be detected in the reaction mixture, and the benzyl benzoate appeared to be the only by-product, which might be formed by condensation of BA with BZ [57,59]. Obviously, the supported gold catalysts with different supports showed distinctive activities. Among them, the Au25/MgAl-300 and Au25/Al2O3-300 catalysts exhibited the best performances (Table 2, entry 1, 5), with the conversion of BA were 89.5% and 89.2%, respectively, and the selectivities of BZ were 95.0% and 94.9%, respectively. In addition, the Turnover Frequencies (TOF) of Au25/MgAl-300 and Au25/Al2O3-300 catalysts were 993 h −1 and 1033 h −1 , when the whole gold was considered in the reaction. However, according to the general acknowledgement, the gold atoms located at the surface were regarded as the active species [3,24]. Thus, it was reasonable to calculate the TOF values in consideration of the dispersion of gold. In order to precisely concatenate the intrinsic activity with the active sites, two groups of TOF values were provided in Table 2 to give a better correlation. However, the dispersion of gold was not easy to get from the experiment. Hence, it was estimated by the empirical equation of 1/d as the literature reported [60], in which the parameter d refers to the diameter of gold particles. Based on the above consideration, the TOF of Au25/MgAl-300 and Au25/Al2O3-300 increased to 3078 h −1 and 2892 h −1 , respectively. In this system, different supported gold catalysts for aerobic oxidation of BA were employed. The results were listed in Table 2. Notably, no any benzoic acid could be detected in the reaction mixture, and the benzyl benzoate appeared to be the only by-product, which might be formed by condensation of BA with BZ [57,59]. Obviously, the supported gold catalysts with different supports showed distinctive activities. Among them, the Au 25 /MgAl-300 and Au 25 /Al 2 O 3 -300 catalysts exhibited the best performances (Table 2, entry 1, 5), with the conversion of BA were 89.5% and 89.2%, respectively, and the selectivities of BZ were 95.0% and 94.9%, respectively. In addition, the Turnover Frequencies (TOF) of Au 25 /MgAl-300 and Au 25 /Al 2 O 3 -300 catalysts were 993 h −1 and 1033 h −1 , when the whole gold was considered in the reaction. However, according to the general acknowledgement, the gold atoms located at the surface were regarded as the active species [3,24]. Thus, it was reasonable to calculate the TOF values in consideration of the dispersion of gold. In order to precisely concatenate the intrinsic activity with the active sites, two groups of TOF values were provided in Table 2 to give a better correlation. However, the dispersion of gold was not easy to get from the experiment. Hence, it was estimated by the empirical equation of 1/d as the literature reported [60], in which the parameter d refers to the diameter of gold particles. Based on the above consideration, the TOF of Au 25 /MgAl-300 and Au 25 /Al 2 O 3 -300 increased to 3078 h −1 and 2892 h −1 , respectively. reasonable to calculate the TOF values in consideration of the dispersion of gold. In order to precisely concatenate the intrinsic activity with the active sites, two groups of TOF values were provided in Table 2 to give a better correlation. However, the dispersion of gold was not easy to get from the experiment. Hence, it was estimated by the empirical equation of 1/d as the literature reported [60], in which the parameter d refers to the diameter of gold particles. Based on the above consideration, the TOF of Au25/MgAl-300 and Au25/Al2O3-300 increased to 3078 h −1 and 2892 h −1 , respectively.  (Table 2, entry 3, 4, 6), with the TOF values far below that of the Au 25 /MgAl-300 and Au 25 /Al 2 O 3 -300 catalysts. Furthermore, the yields of products did not equilibrate with the transformed reactant over the Au 25 /SiO 2 -300 and Au 25 /ZnO-300 catalysts, which might be caused by the absorption of intermediates on both catalysts. The above results indicated that the support effect might play an essential role in aerobic oxidation of benzyl alcohol.
To clarify the size effect, the gold catalysts with different loadings of gold or different methods were also prepared in this reaction for comparison. The results were listed in Table 2, entry 7-9. It indicated the catalytic performances were not largely affected by the loadings of gold, in which the TOF values were 2792 h −1 for 1.1% Au 25 /MgAl-300 catalyst and 3000 h −1 for 0.5% Au 25 /MgAl-300 catalyst ( Table 2, entries 1 and 7), with the mean particle sizes of 3.0 nm ( Figure 4) and 3.3 nm (Figure S5), respectively. However, the activities of Au/MgAl-300-DP and Au/MgAl-300-IMP catalysts prepared by traditional deposition precipitation (DP) method and incipient-wetness impregnation (IMP) method were far below that of the Au 25 /MgAl-300 catalyst, with TOF values of 1051 and 628 h −1 , respectively ( Table 2, entries 8 and 9). In addition, when the dispersion of gold was not taken into account, their gaps of TOF values with the supported Au 25 NCs catalyst were even larger. This phenomenon suggested that the size effect could influence the catalytic performances at the specific conditions.

The Influence Factors for Aerobic Oxidation of Benzyl Alcohol over the Supported Gold Catalysts
Based on the above characterizations and catalytic performances in this work, the influence factors for aerobic oxidation of benzyl alcohol over the supported gold catalysts were then discussed. According to an N 2 -physical adsorption-desorption test, the specific area and the pore volume of Au 25 /MgO-300 and Au 25 /ZnO-300 catalysts were obviously smaller than the others (Figure 1 and Table 1), which might lead to their poor activities in oxidation of benzyl alcohol, since the low specific area and small pore volume might hamper the adsorption of substrates onto the catalysts [43].
However, the Au 25 /SiO 2 -300 catalyst which holds a comparably high specific area also exhibited poor activity, which implied that the catalytic performances were not solely influenced by the pore structure. Actually, it is well known that the particle sizes of gold played a vital role in gold catalysts for a large amount of reactions [3,24,37,61]. Because the small gold particles had a higher fraction of low-coordinated atoms, which was proposed to account for the specific catalytic activities [24]. In this study, as demonstrated by TEM results, the mean particle size of gold in Au 25 /SiO 2 -300, Au/MgAl-300-DP and Au/MgAl-300-IMP catalysts were greater than that of the other supported gold catalysts (Figure 4f and Figure S6). Thus, the different catalytic performances could be affected by the particle sizes. Nevertheless, for Au 25 /MgO-300 and Au 25 /ZnO-300 catalysts, the mean particle sizes of gold were almost the same as that in Au 25 /MgAl-300, Au 25 /ZnAl-300 and Au 25 /Al 2 O 3 -300 catalysts, which bear a particle size within 3 nm, while their performances were totally different. This suggests that the catalytic performances were not influenced by the size effect alone.
We summarized the trend about size effect in Figure 8. The star-like icons with different colors displayed the various supported gold catalysts. The horizontal coordinates represented the diameter of gold, and the vertical coordinates revealed the TOF values based on the surface gold. From the data, the catalytic performances of gold catalysts with sizes below 5 nm seem to be independent from the particle size, while the dimensions above 5 nm displayed the obvious size-dependent property. Thus, in a small scale, the activities might be influenced by the other factor instead of the size. of gold, and the vertical coordinates revealed the TOF values based on the surface gold. From the data, the catalytic performances of gold catalysts with sizes below 5 nm seem to be independent from the particle size, while the dimensions above 5 nm displayed the obvious size-dependent property. Thus, in a small scale, the activities might be influenced by the other factor instead of the size. The interaction between gold and the support was often reported to make a big difference in catalytic reactions by modifying the electronic properties of gold [33,34]. In this study, we found the interaction between gold and the SiO2 was too weak to influence the nature of gold. However, it really worked in some other supports, since the valent state of gold was negatively shifted ( Figure 5). Although the interaction between gold and the supports were strong in Au25/MgO-300 and Au25/ZnO-300 catalysts, their performances were not as good as Au25/MgAl-300, and even inferior to Au25/SiO2-300. Thus, it can be concluded that the interaction between gold and the support might not be the key influence factor for catalytic performances of benzyl alcohol oxidation in this system.
The aerobic oxidation of alcohols was usually performed in the presence of a base, which was viewed as essential to form an alkoxide intermediate by the activation of the O-H bond of the alcohol [20,45,59]. The base additive such as KOH and K2CO3 were often deemed as an efficient promoter especially when the support has no basicity [16,20]. However, these additives had big problems of contamination and separation that would greatly prevent its further application in industry. Thus, in this work, the experiments were conducted in the absence of any inorganic base. Thus, the basicity originated from the support was probably important for the performance. Moreover, it was said that the acid sites on the support could facilitate for the metal hydride with oxygen to produce water The interaction between gold and the support was often reported to make a big difference in catalytic reactions by modifying the electronic properties of gold [33,34]. In this study, we found the interaction between gold and the SiO 2 was too weak to influence the nature of gold. However, it really worked in some other supports, since the valent state of gold was negatively shifted ( Figure 5). Although the interaction between gold and the supports were strong in Au 25 /MgO-300 and Au 25 /ZnO-300 catalysts, their performances were not as good as Au 25 /MgAl-300, and even inferior to Au 25 /SiO 2 -300. Thus, it can be concluded that the interaction between gold and the support might not be the key influence factor for catalytic performances of benzyl alcohol oxidation in this system.
The aerobic oxidation of alcohols was usually performed in the presence of a base, which was viewed as essential to form an alkoxide intermediate by the activation of the O-H bond of the alcohol [20,45,59]. The base additive such as KOH and K 2 CO 3 were often deemed as an efficient promoter especially when the support has no basicity [16,20]. However, these additives had big problems of contamination and separation that would greatly prevent its further application in industry. Thus, in this work, the experiments were conducted in the absence of any inorganic base. Thus, the basicity originated from the support was probably important for the performance. Moreover, it was said that the acid sites on the support could facilitate for the metal hydride with oxygen to produce water [25,45]. Hence, the promotion of acidity was also needed to give a reasonable explanation.
Linking the reaction performance with the acid-base properties, it was interesting to find that the order of catalytic performances agreed well with the sequence of acid-base quantities. This meant that the catalysts with more amounts of basic and acidic sites showed better activities than the ones with fewer acid-base sites. It implies that the content of base and acid plays a vital role in aerobic oxidation of benzyl alcohol. In addition, the catalytic activity might be affected simultaneously by both the acidic and basic sites.
The total adsorption amounts of CO 2 , NH 3 as well as O 2 over different supported gold catalysts were illustrated in Figure 9a. Notably, the adsorption of O 2 was not much different in various supported gold catalysts. This implied that the adsorption and dissociation of O 2 might not be the rate-determining step of this reaction. In addition, the relationship between TOF values with the total basicity and acidity was observed in Figure 9b. It clearly suggests that the catalysts with both basic and acid sites displayed good performances in aerobic oxidation of benzyl alcohol, whereas the strength of basicity and acidity displayed no obvious effect in catalytic performances. To be noted, based on the results in Table 2, the catalysts with more basic sites afford higher selectivity towards benzaldehyde, while the catalysts with more acid sites were more active for the conversion of benzyl alcohol.  Table 2, the catalysts with more basic sites afford higher selectivity towards benzaldehyde, while the catalysts with more acid sites were more active for the conversion of benzyl alcohol.

The Proposed Reaction Mechanism
Based on the previous literature [20,25,45,59] and the above results in this work, we proposed an acid-base promoted mechanism for aerobic oxidation of benzyl alcohol (Figure 10). At this situation, the catalysts should exhibit both basicity and acidity. Among them, the base acts as the sites to cleave the O-H bond of the alcohol, in order to form an alkoxide intermediate [20,25,45], while the acid site might participate in the reaction of transforming the metal hydride and the oxygen to form water [45,46]. To be noted, the blank experiment and the support itself were not active for this reaction ( Table 2, entries 11 and 12), which suggests that the gold and the supports synergistically catalyzed the reaction. Furthermore, the uncalcined Au25/MgAl-HT sample showed extreme low activity and selectivity ( Table 2, entry 10), which indicates that the exposed gold atoms after removal of the protectant were critical for the reaction, and the active sites might be located at the interface of gold with the support [62].

The Proposed Reaction Mechanism
Based on the previous literature [20,25,45,59] and the above results in this work, we proposed an acid-base promoted mechanism for aerobic oxidation of benzyl alcohol (Figure 10). At this situation, the catalysts should exhibit both basicity and acidity. Among them, the base acts as the sites to cleave the O-H bond of the alcohol, in order to form an alkoxide intermediate [20,25,45], while the acid site might participate in the reaction of transforming the metal hydride and the oxygen to form water [45,46]. To be noted, the blank experiment and the support itself were not active for this reaction ( Table 2, entries 11 and 12), which suggests that the gold and the supports synergistically catalyzed the reaction. Furthermore, the uncalcined Au 25 /MgAl-HT sample showed extreme low activity and selectivity ( Table 2, entry 10), which indicates that the exposed gold atoms after removal of the protectant were critical for the reaction, and the active sites might be located at the interface of gold with the support [62]. Catalysts 2020, 10, x FOR PEER REVIEW 15 of 20 Figure 10. The proposed reaction mechanism over the supported gold catalysts with the synergetic acid-base sites. The M refers to the metal cation.
The proposed reaction mechanism could be divided into four steps. In the first step, the benzyl alcohol attacked the basic site of M-OH δ− on the supports. Then, the deprotonation by the hydroxyl group of the support occurred to form an alkoxide intermediate at the interface. Simultaneously, the molecule water was formed in the process. After that, the alkoxide intermediate proceeded with the coordination with the gold to form a metal-hydride species. Synchronously, the β-hydride was eliminated from the interface by the gold to produce the benzaldehyde. According to literature [25,45], this step might play a significant role in determining the selectivity of benzaldehyde. In addition, the process was expected to be the rate-determining step. This was why the small gold particles exhibited higher activity-because the coordinatively unsaturated gold atoms with smaller sizes were more active for the cleavage of the C-H bond. However, when the particle sizes were within a small range, the acid-base property dominated the reaction rate instead of the particle size. In the last step, the reaction underwent a rapid oxidation of the metal-hydride specie by oxygen to form the water, which promoted by the acid site of the support. Then, the catalysts recovered at the initial metallic state, thereby to complete the catalytic cycle. For the Au25/MgAl-300 catalyst, we also examined the recyclability for the oxidant of benzyl alcohol. As shown in Figure S7, the conversion of benzyl alcohol decreased from 87% to 19% after reusing five times. This means that the Au25/MgAl-300 catalyst was not very stable in the reaction. The TEM results of Au25/MgAl-300 catalyst after reusing three times indicated that the particle sizes of gold increased from 3.0 nm (Figure 4a) to 4.5 nm ( Figure S8), and agglomeration appeared in part of the catalyst, which might be deemed as a key point for the reduction of catalytic performance.
In summary, we proposed a reaction mechanism of acid-base synergistically catalyzed the reaction route with the help of small gold particles based on the experiment results and characterizations. The fundamental reasons for the high activity and selectivity were provided to give a deep understanding of the system. This was crucial to make it clear how the different factors influenced the catalytic performances.

Materials
The main chemicals were purchased from Aladdin (Georgia, USA). The MgO and ZnO were pretreated at 500 °C for 4 h before use to remove the impurity. The Al2O3 was obtained by calcination of Al(OH)3 at 500 °C for 3 h. The SiO2 was functionalized with 3-aminopropyltriethoxysilane (APTES, Figure 10. The proposed reaction mechanism over the supported gold catalysts with the synergetic acid-base sites. The M refers to the metal cation. The proposed reaction mechanism could be divided into four steps. In the first step, the benzyl alcohol attacked the basic site of M-OH δ− on the supports. Then, the deprotonation by the hydroxyl group of the support occurred to form an alkoxide intermediate at the interface. Simultaneously, the molecule water was formed in the process. After that, the alkoxide intermediate proceeded with the coordination with the gold to form a metal-hydride species. Synchronously, the β-hydride was eliminated from the interface by the gold to produce the benzaldehyde. According to literature [25,45], this step might play a significant role in determining the selectivity of benzaldehyde. In addition, the process was expected to be the rate-determining step. This was why the small gold particles exhibited higher activity-because the coordinatively unsaturated gold atoms with smaller sizes were more active for the cleavage of the C-H bond. However, when the particle sizes were within a small range, the acid-base property dominated the reaction rate instead of the particle size. In the last step, the reaction underwent a rapid oxidation of the metal-hydride specie by oxygen to form the water, which promoted by the acid site of the support. Then, the catalysts recovered at the initial metallic state, thereby to complete the catalytic cycle. For the Au 25 /MgAl-300 catalyst, we also examined the recyclability for the oxidant of benzyl alcohol. As shown in Figure S7, the conversion of benzyl alcohol decreased from 87% to 19% after reusing five times. This means that the Au 25 /MgAl-300 catalyst was not very stable in the reaction. The TEM results of Au 25 /MgAl-300 catalyst after reusing three times indicated that the particle sizes of gold increased from 3.0 nm (Figure 4a) to 4.5 nm (Figure S8), and agglomeration appeared in part of the catalyst, which might be deemed as a key point for the reduction of catalytic performance.
In summary, we proposed a reaction mechanism of acid-base synergistically catalyzed the reaction route with the help of small gold particles based on the experiment results and characterizations. The fundamental reasons for the high activity and selectivity were provided to give a deep understanding of the system. This was crucial to make it clear how the different factors influenced the catalytic performances.

Materials
The main chemicals were purchased from Aladdin (Georgia, USA). The MgO and ZnO were pretreated at 500 • C for 4 h before use to remove the impurity. The Al 2 O 3 was obtained by calcination of Al(OH) 3 at 500 • C for 3 h. The SiO 2 was functionalized with 3-aminopropyltriethoxysilane (APTES, 99%) by heating at 80 • C in alcohol under reflux for 24 h. The MgAl-HT and ZnAl-HT were homemade by solution synthesis. The other chemicals in the experiment were used without further purification.

Preparation of the Au 25 NCs
The hydrosoluble Au 25 NCs were synthesized according to the literature [37]. In a typical synthesis, 0.5 mmol of the HAuCl 4 solution and 0.75 mmol of the cysteine solution were added successively to 150 mL of deionized water under vigorous stirring. After that, 30 mL of NaOH solution (1 M) was introduced into the mixture. Then, five times of the equivalent NaBH 4 solution were freshly prepared and added to the above solution, followed by vigorous stirring for 3 h. Finally, the products were washed repeatedly with the ethanol/water mixture (V/V = 3:1) and collected after lyophilization to obtain the Au 25 nanoclusters.

Preparation of the Hydrotalcite (HT)
The ZnAl-HT and MgAl-HT supports were prepared as follows. Firstly, solution A was obtained by adding 0. 21

Preparation of the Supported Au 25 Nanoclusters (NCs)
The supported Au NCs were synthesized via the impregnation method. Typically, 15 mg of the cysteine-caped Au 25 NCs were dissolved in 20 mL of deionized water. Then, 2.00 g of the supports were added to the above suspension under gentle stirring. After 1 h, the products were washed with water and collected by centrifugation. The residue was then dried at 80 • C and calcined at 300 • C for 2 h to obtain the samples. The obtained samples were denoted as Au 25 /MgAl-300, Au 25 /ZnAl-300, Au 25 /ZnO-300, Au 25 /MgO-300, Au 25 /Al 2 O 3 -300, and Au 25 /SiO 2 -300. For comparison, the supported gold NCs with different loadings of Au were prepared by regulating the amounts of Au 25 .

Preparation of the Supported Au Nanoparticles (NPs)
In comparison to the supported Au 25 NCs catalysts, the gold nanoparticles were synthesized by the traditional deposition precipitation (DP) method and incipient-wetness impregnation (IMP) method. In a typical DP method, the pH of the HAuCl 4 solution was firstly adjusted to nearly 7 with 0.1 M NaOH solution. Then, the support was added to the solution under vigorous stirring. After the pH reached to about 9 with NaOH solution, the solution was filtered, washed with 500 mL of ultrapure water, and dried overnight in an oven at 80 • C. In the IMP method, a special volume of the HAuCl 4 solution was prepared in advance. Then, the solution was slowly added to 2 g of the support under continuous stirring until the dry powder was saturated with the solution. After that, the wet powder dried 80 • C. In addition, before evaluating for the reaction test, they were calcined at 300 • C for 2 h.

Characterization of the Catalysts
The actual content of gold in the catalysts was measured with an inductively coupled plasma atomic emission spectrometer (ICP-AES) on an IRIS Intrepid II XSP instrument (Thermo Electron Corporation, Madison, USA). The measured amounts of catalysts were dissolved in a fresh aqua regia and hydrofluoric acid at boiling condition and diluted to 50 mL for analysis. Power X-ray diffraction (XRD) were determined on a PW3040/60 X'Pert PRO (PANalytical, Almelo, Netherlands) diffract meter equipped with a Cu Kα radiation source (λ = 0.15432 nm) operating at 40 kV and 40 mA. The continuous mode was used at a scanning speed of 10 • /min in the scanning angle (2θ) range from 10 • to 90 • . UV-vis spectra were recorded on a Cintra (Austin, USA) (GBC) apparatus with BaSO 4 as the reference at room temperature. A continuous mode at a scanning speed of 100 nm/min was used from 200 to 800 nm. The texture properties of the catalysts were measured by N 2 -physical adsorption-desorption test at 77 K using an AutoSorb-1 instrument (Florida, USA). Prior to the measurements, the catalysts were treated at vacuum for 2 h at 120 • C. The pore volume and pore size distribution were determined from the adsorption branches using the NLDFT (the nonlocal density functional theory) method. The transmission electron microscopy (TEM) was recorded with a JEM-2100F field emission (Tokyo, Japan) TEM. The microstructure of the catalysts was further observed by high resolution transmission electron microscopy (HRTEM) at 200 kV. The temperature-programmed desorption of carbon dioxide (CO 2 -TPD), ammonia (NH 3 -TPD), and oxygen (O 2 -TPD) experiments were performed on AutochemII 2920 Instruments (Norcross, Georgia, USA) with a thermal conductivity detector. Prior to the analysis, about 50 mg of the catalysts were added into a U-type quartz tube reactor, and then the sample was heated in a flow of helium at 300 • C for 1 h. After the temperature decreased to 80 • C, the catalysts were saturated with CO 2 , NH 3 and O 2 at 80 • C for 20 times. Then, the sample was purged in helium at 50 • C for 1 h. Finally, the CO 2 /NH 3 /O 2 -TPD signal was recorded from 100 • C to 800 • C at a rate of 10 • C/min with a cold trap. X-ray photoelectron spectra (XPS) were conducted on an ESCALAB 250 X-ray photoelectron spectrometer (Madison, USA) equipped with a monochromated Al Kα anode. The binding energies of Au 4f were calibrated for surface charging by referencing them to the energy of the C1s peak at 284.6 eV.

Aerobic Oxidation of Alcohols
The aerobic oxidation of benzyl alcohol was performed in a three-necked flask. In a typical reaction, 100 mg of the catalysts were added in the reactor. Then, 1 mmol substrate dissolved in 10 mL toluene was added without addition of any inorganic bases. The o-xylene was used as the internal reference. The reaction was conducted at 80 • C under stirring. After the temperature reached a fixed value, the O 2 was bubbled through the solution at atmospheric pressure with the gas flow of 30 mL/min. The products were analyzed by gas chromatography (Agilent 7890 GC, California, USA) equipped with a HP-5 capillary column (25 m × 0.32 mm) and flame ionization detector (FID). The initial reaction rate and the turnover frequency (TOF) were measured below 20% conversion.

Conclusions
In this work, the MgAl-HT, ZnAl-HT, ZnO, MgO, Al 2 O 3 , and SiO 2 supported Au 25 NCs catalysts were successfully synthesized. Characterization results of XRD, UV-vis, and TEM implied that the gold particles were uniformly distributed on the supports, with a mean particle size within 3 nm in most of the above catalysts. For aerobic oxidation of benzyl alcohol, the Au 25 /MgAl-300 and Au 25 /Al 2 O 3 -300 catalysts presented the best activities, with the TOF values of~2927 and 2892 h −1 , respectively. The next one was the Au 25 /ZnAl-300, with a TOF of 1536 h −1 . The Au 25 /ZnO-300, Au 25 /MgO-300, and Au 25 /SiO 2 -300 catalysts displayed poor activities. Based on the TEM, HRTEM, and XPS measurements, the morphology and size distributions of gold as well as the interaction between gold with the supports were largely affected by the supports effect. These factors might not be the key points for catalytic performances, but the acid-base property might be the critical one. The synergic catalytic route between the acid site and the basic site was proposed over the supported Au 25 NCs catalysts. We believe that this work provided a good way to design the gold catalysts with controllable sizes, and it was crucial for understanding the reaction process in aerobic oxidation of alcohol.

Supplementary Materials:
The following are available online at http://www.mdpi.com/2073-4344/10/1/107/s1, Figure S1: N 2 adsorption/desorption isotherms and pore size distributions of different supports, Figure S2: XRD patterns of different supports, Figure S3: HRTEM images of different supported gold catalysts, Figure S4: XPS spectra of O 1s on different supported gold catalysts, Figure S5: TEM image and size distribution histogram of 1.1% Au 25 /MgAl-300 catalyst, Figure S6: TEM images and size distribution histograms of gold catalysts with different preparation methods, Figure S7: The reusability of Au 25 /MgAl-300 catalyst in benzyl alcohol oxidation, Figure S8: The TEM images and size distributions of Au 25 /MgAl-300 catalyst after reusing three times, Table S1: Physicochemical properties of different supports, Table S2: The adsorption quantities of CO 2 , NH 3