A series of Mg(OH)₂-CeO₂ with various Mg(OH)₂ contents were prepared by co-precipitation method [1] - [10] using magnesium nitrate and cerium nitrate as the starting materials. The catalyst was dried overnight in air at 80˚C for 8 h, and then calcined at 550˚C for 4 h in air. Deposition-precipitation method (DP) was used in this study to deposit gold on the support. The detailed preparation method has been described in previous paper [18] . An aqueous solution of HAuCl₄ (2.55 × 10^–3 M) was slowly added into the solution containing support under vigorous stirring at a temperature 65˚C. NH₄OH was used to adjust the pH value of the solution to 7. After aging for 2 h, the sample was filtered and the filtration cake was washed until no chloride ions were detected. Finally, the sample was dried overnight at 80˚C, and was calcined at 180˚C for 4 h. This temperature was high enough to reduce Au³ = cation and low enough to prevent Au from sintering. The nominal loading of Au was 1 wt%. The sample was denoted as Au/Mg(OH)₂-CeO₂(x) where x is the atomic ratio of Mg/(Mg + Ce).

The catalysts were characterized with ICP-MS, N₂-sorption, transmission electron micros copy (TEM), high resolution transmission electron microscopy (HR-TEM), X-ray diffraction (XRD), and X-ray Spectroscopy (XPS). The characterization technique has been described in the previous paper [18] . The real Au contents of the catalysts were measured by ICP-MS.

The surface area of the samples was measured using a Micromeritics ASAP 2010 by Brunauer-Emmett-Teller (BET) method in the range P/P₀ = 0.05 - 0.3. Prior to the experiments, the sample was treated at 120˚C for 8 h at the vacuum pressure below 1 × 10^–5 torr.

XRD analysis was performed using a Siemens D500 powder diffractometer using Cu K_α1 radiation (0.15405 nm) at a voltage and current of 40 kV and 40 mA, respectively. The sample was scanned over the range of 2θ = 20˚ - 70˚ at a rate of 0.05˚/min.

The catalysts were characterized by TEM on a JEM-2000 EX II operated at 120 kV and HRTEM on a JEOL JEM-2010 operated at 160 kV. A small amount of sample was placed into the sample tube filled with a 95% ethanol solution. After agitating under ultrasonic environment for 3 h, one drop of the sample was dipped on a carbon-coated copper mesh (300#) (Ted Pella Inc., CA, USA), then dried in vacuum overnight. Images were recorded digitally with a Gatan slow scan camera (GIF).

XPS analysis was carried out with a Thermo VG Scientific Sigma Prob spectrometer. The XPS spectra were collected using Al K_α radiation at a voltage and current of 20 kV and 30 mA, respectively. The spectrometer was operated at 23.5 eV pass energy and the binding energy was corrected by contaminant carbon (C_1s = 285.0 eV). Peak fitting was done using XPSPEAK 4.1 with Shirley background and 30:70 Lorentzian/Gaussian convolution product shapes.

The PROX reaction was carried out in a fixed-bed glass reactor. 0.1 g catalyst was loaded in the reactor. The reactant gas containing 1.33% CO, 1.33% O₂, 65.33% H₂ and He for balance (vol. %) was fed into the reactor with a total flow rate of 50 ml/min. The ratio [O₂]/[CO] was fixed at 1 in this study for a better comparison of the activity and selectivity among catalysts. The reactor was heated in a furnace with a heating rate of 2˚C/min. After reaching the reaction temperature for 5 min, the product was analyzed by a gas chromatograph equipped with a thermal conductivity detector using MS-5A packed column. CO conversion and selectivity of oxygen reacting with CO were calculated by the following equations:

CO conversion (X_CO%) = [ CO ] in − [ CO ] out [ O ] in ∗ 100 % (3)

O₂ conversion (X_O2%) = [ O 2 ] in − [ O 2 ] out [ O 2 ] in ∗ 100 % (4)

Selectivity (S%) = mol of O₂ reacting with CO/mol of O₂ consumed

= 0.5 ∗ X CO X O 2 ∗ 100 % (5)

In all experiments, the carbon imbalance did not exceed ±1%.

Table 1 shows the amount of Au loading on the catalyst. The isoelectric point of support would influence the Au loading and the particle size of Au. After adding Mg(OH)₂ in CeO₂, the isoelectric point of support increased. In this study, the pH value during DP process was fixed at 7, therefore the amount of Au loading increased after adding Mg(OH)₂. The results are in accord [1] [10] .

The BET surface areas of Au/Mg(OH)₂-CeO₂ catalysts are listed in Table 2. It shows that the addition of Mg(OH)₂ in Au/CeO₂ increased the BET surface area of the catalyst. Since CeO₂ was crystalline phase, it had low surface area. Mg(OH)₂ was in amorphous phase and had higher surface area. Adding Mg(OH)₂ in CeO₂ increased the BET surface area of the catalyst, as expected.

In the XRD pattern of catalysts with different magnesium contents (Figures 1(b)-(e)), the characteristic peaks of CeO₂ at 2θ = 28.549˚ (111), 33.077˚ (200), 47.483˚ (220) were detected in all catalysts. However, the peaks of Au, which should present at 2θ = 38.184 (111), 44.392 (200) and 77.547 (311), were not detected in the XRD patterns, indicating that the particle size of gold particles were smaller than 5 nm [12] [13] [14] [15] . The results are consistent with the TEM results, which will be reported in the later section.

It should be noted that the pure MgO support calcined at 380˚C existed in phase of both MgO (2θ = 42.917 (200), 62.304 (220)) and Mg(OH)₂ (2θ = 38.016 (101), 50.854 (102), 58.640 (110)), as shown in Figure 1(g). After loading with gold, only Mg(OH)₂ phase was detected. Mg(OH)₂ was formed by reaction with water during depeosition-precipitation (DP) process:

MgO + H 2 O → Mg ( OH ) 2 (6)

Mg(OH)₂ can be converted to MgO as the temperature was higher than 400˚C. Since the calcination temperature of the sample was 180˚C, Mg would be in the form of Mg(OH)₂. There was no significant difference in particle size of cerium oxide with different Mg contents.

The characteristic peaks of Mg(OH)₂ were not observed in the XRD patterns for all of the catalysts, excepted the one with Mg/(Mg + Ce) ratio of 50%. This indicates that Mg(OH)₂ was highly dispersed on CeO₂ surface. MgO phase was not observed in all samples. It was only detected in the Au/MgO sample.

As shown in the TEM photos in Figure 2, the gold particles were observed as small dark spots. The gray bulk mass was ceria support, while Mg(OH)₂ was surrounded on ceria with a lighter color. It shows that Mg(OH)₂ was on the surface of CeO₂ as a thin layer when the content of Mg(OH)₂ was low. Au was deposited on either CeO₂ or Mg(OH)₂ depending on the content of Mg(OH)₂. The average particles diameters of Au in these Au/Mg(OH)₂-CeO₂ with different Mg contents are listed in Table 3. The average size of Au on these catalysts was between 3.5 nm and 5 nm, the Au particle size decreased as the content of Mg increased.

The HR-TEM image of Au/Mg(OH)₂-CeO₂ (20%) catalyst are shown in Figure 3. The d-spacing of 0.3063 nm is corresponded to CeO₂ (111) with standard d-spacing of 0.3124 nm. The d-spacing of 0.2351 nm is corresponded to Mg(OH)₂ (101) with standard d-spacing of 0.2365 nm. Gold with d-spacing 0.2335 nm (the standard d-spacing is 0.2355 nm) attached on the support was seen as well.

In the XPS analysis of Mg 2p, the binding energy of MgO is 50.8 eV and Mg(OH)₂ is 49.5 eV. The XPS Mg 2p spectra of this series of Au/Mg(OH)₂-CeO₂ catalyst are shown in Figure 4. The binding energies of Mg 2p peaks in all samples are tabulated in Table 4. The binding energies of Mg 2p of Au/Mg(OH)₂-CeO₂ catalysts are around 49.3 eV, indicating that the magnesium species were mainly existed as Mg(OH)₂, in agreement with the XRD results. The Mg/(Mg + Ce) ratios on the surface, obtained by XPS analysis, are listed in Table 4. As the nominal content of magnesium reached 37.5%, significant accumulation of Mg on the surface were observed.

XPS spectra of O 1s were applied to analyze the contents of various oxygen species on the surface (Figure 5). The O 1s spectra were separated into two peaks: lattice oxygen at 529 eV and hydroxide species at 531 eV. The lattice oxygen was attributed to the oxygen existed in the lattice of CeO₂ while the hydroxide species assigned to the Mg(OH)₂ species. Among all of the Au/Mg(OH)₂-CeO₂ catalysts, the one with 6.5% of Mg/(Mg + Ce) ratio had the highest amount of hydroxide species compared to the others (Table 5).

The XPS spectra of Au 4f are shown in Figure 6, and the concentrations of various electronic states of Au on Mg(OH)₂-CeO₂ with different Mg/(Mg + Ce) atomic ratios are listed in Table 6. Gold displayed two peaks of Au 4f_7/2 and Au 4f_5/2 at around 84 eV and 88 eV. The peak of Mg 2s at 89 eV overlapped with Au 4f_5/2. The metallic gold and oxidic gold displayed peaks at 87.57 eV and 88.2 eV in Au 4f_5/2, and at 83.9 eV and 84.7 eV in Au 4f_7/2 [1] [2] [3] [4] [5] . Table 6 shows that the higher concentration of metallic gold species was observed with higher content of Mg(OH)₂ in the catalyst. In the samples with low content of Mg(OH)₂, most of Au was deposited on CeO₂. Au would have strong interaction with CeO₂. Since CeO₂ has more defect sites, it would cause Au in electron-deficient. In the samples with high concentration of Mg(OH)₂, most of Au would deposit on the surface of Mg(OH)₂, resulting in metallic gold species.

*Mg/(Mg + Ce) in catalyst.

Figure 7 and Figure 8 show the effect of Mg content on the performance of PROX over Au/Mg(OH)₂-CeO₂ catalysts. CO conversion increased with increasing temperature until 40˚C or 60˚C, depending on sample, and then level off. At high temperature, H₂ would compete with CO to react with O₂, and resulted in low CO selectivity. At lower reaction temperature, the existence of magnesium did not facilitate the CO conversion; instead, lower CO conversion was observed, except the sample with Mg/(Mg + Ce) ratios of 6.5%. Figure 8 shows that the samples modified with Mg(OH)₂ had high selectivity of CO oxidation at low reaction temperature, indicating that the presence of Mg(OH)₂ could suppress H₂ oxidation. Au/Mg(OH)₂-CeO₂ catalysts showed higher CO oxidation selectivity than the unmodified Au/CeO₂ catalyst at any reaction temperature. All catalysts had high CO conversion at temperature 80˚C - 100˚C. Nevertheless, the Au/Mg(OH)₂-CeO₂ still had a higher CO conversion, although the difference was very small, since Au/CeO₂ already has conversion reaching 99%. The Au/Mg(OH)₂-CeO₂ (20%) catalyst was run at 80˚C for 120 h time on stream, it did not show any decay, indicating this catalyst was very stable.

Au/Mg(OH)₂-CeO₂ (20%) catalyst showed the highest activity among all catalysts, it also possessed the highest ratio of hydroxyl species. This confirms that the OH species between the interface of gold and support is the species to activate CO oxidation reaction. Molina and Hammer [19] reported that the Au-Au coordination determined the local reactivity of the Au atoms and the presence of the MgO support that, besides providing excess electrons to the Au clusters, forms ionic bonds to the peroxo part of the CO-O₂ reaction intermediate. They [19] reported that the type of interface boundary likely to be predominant for medium-sized nanoparticles provides the optimal degree of low-coordinated Au atoms in the neighborhood of the MgO support. Our results are in accord.

The metallic gold was believed to be the active site for PROX reaction [20] - [28] . The amount of Au⁰ in the catalysts increased with the increase of magnesium content as shown in Table 5. However, overdoes of Mg(OH)₂ would retard the CO adsorption and suppress CO oxidation. It also suppressed H₂ oxidation and thus there was enough O₂ to react with CO [13] [21] . The results clearly demonstrate that adding Mg(OH)₂ in Au/CeO₂ could suppress H₂ oxidation and therefore increase CO oxidation activity.