Formation of a Pt-MgO Solid Solution: Analysis by X-ray Absorption Fine Structure Spectroscopy.

Thermal treatment of Pt nanoparticles or Pt(acac)2 supported on MgO resulted in the formation of a solid solution of Pt-MgO, as evidenced by Pt L3-edge X-ray absorption fine structure spectroscopy. The valence of Pt in the Pt-MgO solid solution was determined to be 4+. A characteristic shrinkage of the Pt-O bond distance was observed in comparison with that of the nearest-neighboring Mg-O bond in MgO, which agreed with the density functional theory (DFT) calculations. The segregation of Pt and MgO proceeded with a further increase in the thermal treatment temperature up to 1273 K. The dispersion of Pt on MgO measured through CO adsorption was much higher than that on Al2O3 or SiO2 owing to the formation of the Pt-MgO solid solution.

Introduction

Supported metal catalysts are widely used in petroleum refining, fine chemical synthesis, and most recently, in environmental remediation, and the demand for efficient supported metal catalysts is increasing progressively. Noble metals supported on oxide materials are useful catalysts that exhibit high activity in many reactions. It is desirable to support them in a highly dispersed manner to enhance the surface area of the supported metal species. However, the highly dispersed catalyst particles often aggregate at high temperatures, resulting in an irreversible deactivation of the catalyst, which is a major drawback of such catalysts.[1] One of the possible methods to overcome this issue is to redisperse the aggregated metal or metal oxides by inducing strong interactions between the metal or metal cation and the oxide support. In fact, strong metal–support interaction has been utilized for obtaining highly dispersed Pt on CeO2 or Pd and Rh loaded on perovskite oxides, which have been commercialized for use in the purification of vehicle exhaust.[2,3] Supported Pt catalysts have been widely employed in versatile reactions, such as the purification of the exhaust gases of motor vehicles, total oxidation of CO, and other organic reactions.[4] A solid solution is a promising precursor for the formation of highly active catalysts, as reported for Pd–CeO2 catalysts used in CO oxidation.[5] The interaction between Au and MgO has been extensively studied.[6,7] However, the mechanism of the interaction between the metal Pt and MgO has not been studied in detail.

Recently, significant attention has been paid to catalysts with well-dispersed metal nanoparticles (NPs), in which the finely dispersed metal clusters are protected with polymers.[8−11] The advantage of using polymer-protected NPs is that their size distribution can be regulated by changing the concentration of the metal species and the polymer-to-metal ratio during the preparation of the NPs.[12,13] Therefore, one of the plausible methods to obtain well-dispersed supported metal catalysts is to use polymer-protected nanoparticles. In this study, we employed commercially available polyvinylpyrrolidone (PVP)-protected Pt NPs as the precursor for preparing supported Pt catalysts. The Pt NPs were loaded on three types of supports, MgO, Al2O3, and SiO2, to obtain insights into the origin of the metal–support interactions. For this purpose, the PVP-protected Pt NPs were loaded onto MgO, followed by thermal treatment in air at different temperatures. Platinum(II)bis(acetylacetonate) (Pt(acac)2) was employed as another Pt precursor to assess the influence of the precursor type and preparation conditions on the formation process of Pt–MgO solid solutions. The MgO support loaded with Pt NPs and Pt(acac)2 are denoted as Pt NP/MgO and Pt(acac)2/MgO, respectively. The electronic state and local structure of Pt in the two supported catalysts were analyzed by X-ray absorption fine structure (XAFS) spectroscopy. This is because XAFS is an effective tool for monitoring the oxidation states and local structures of a given element.[14,15]

Experimental Section

Sample Preparation

A 20% ethanol/water suspension of PVP-protected Pt NPs (Wako Chemical Co.) mixed with MgO (JRC-MGO-4, 500A) was evaporated using an evaporator. The obtained solid was crushed with a mortar, followed by thermal treatment in air at 673–1273 K for 3 h in an electric furnace. The PVP-protected Pt NPs were also loaded on Al2O3 (JRC-ALO-7) and SiO2 (Fuji Silysia Co., Q-10) using similar procedures. MgO and Al2O3 were obtained from the Catalysis Society of Japan. The loading of Pt was 1.0 wt % in all cases. Pt(acac)2 (Wako Chemical Co.) impregnated on MgO was thermally treated in air in the same way as that used for preparing Pt NP/MgO. In this case, the Pt loading was fixed at 1.0 wt %, unless otherwise stated.

Pt L3-Edge XAFS Measurements and Analyses

The Pt L3-edge XAFS data were collected using synchrotron radiation. The spectra were recorded at the BL01B1 station with the approval of the Japan Synchrotron Radiation Research Institute (SPring-8, JASRI, proposal nos. 2021A1429, 2021B1291, and 2022A1242). The obtained data were collected in the quick mode within 5 min using a Si (111) monochromator. The beam size at the sample position was 5 mm (horizontal) × 0.8 mm (vertical). For the Pt L3-edge extended X-ray fine structure (EXAFS) analysis, the oscillations were extracted using a spline smoothing method. The Fourier transform (FT) of the k3-weighted EXAFS oscillations and k3χ(k) from k-space to r-space was conducted in the range of 3–13 Å–1 for curve fitting analysis. The EXAFS data were analyzed using the REX software (Rigaku Co.) with curve fitting employing the EXAFS data of PtO2 and Pt foil to analyze the Pt–O and Pt–Pt bonds, respectively. For the analysis of the Pt–Mg bond, parameters were obtained using the FEFF8.0 code.[16]

Physicochemical Characterization

Thermogravimetry–differential thermal analysis (TG-DTA) data were collected under air flow using a DTG-60 analyzer (Shimadzu Co.) at a temperature ramp rate of 10 K/min. N2 adsorption isotherms were recorded on a BELSORP-mini-X (MicrotracBEL Co.) instrument. The samples were dehydrated in vacuum at 573 K before the N2 adsorption measurements to obtain isotherm data. The transmission electron microscopy (TEM) images of the Pt/MgO samples were obtained using a JEOL-JEM-2100 microscope. Briefly, an alcohol suspension of the sample was dropped onto Cu grids coated with a C-coated porous thin membrane (NEM, Japan) and dried. Thereafter, TEM observations were performed at an operating voltage of 200 kV. The X-ray diffraction (XRD) patterns of the powders were obtained under ambient conditions using a MiniFlex X-ray diffractometer (Rigaku Co.) with Cu Kα radiation in the 2θ range of 20–90°. The scanning speed was 1°/min. The dispersion of Pt on the supports was evaluated using BELCAT II equipment (MicrotracBEL Co.). The samples were treated with H2 at 773 K for 1 h before the measurements. The dispersion value of Pt was measured via CO adsorption at 323 K using a thermocouple detector (TCD), assuming a CO/Pt ratio of 1 for calculation. Further, temperature-programmed reduction (TPR) of the samples with H2 was conducted with the same equipment for the analysis of dispersion (BELCAT II). The experiments were performed using 5% H2/Ar (50 mL/min flow rate) without pretreatment. The samples were heated at a temperature ramp rate of 10 K/min from room temperature to 923 K. A TCD detector was used to monitor the concentration of H2 in the flowing gas.

Computer Simulation

Density functional theory (DFT) calculations using the DMol3 program package[17] was carried out to calculate the structure of Pt–MgO crystal under three-dimensional periodic condition. The physical wave functions were expanded in terms of accurate numerical basis sets. A double numerical plus d-functions was used for all calculations, employing the generalized gradient approximation functional developed by Perdew–Burke–Ernzerhof.[18] Core electrons were treated by effective core potentials. The k-point was set to 2 × 2 × 2. The convergence tolerances for energy, maximum force, and maximum displacement were less than 2.0 × 10–5 Ha, 0.004 and 0.005 Ha/Å, respectively. A unit cell consists of Pt1Mg30O32, where two nearest Mg atoms were deleted from the perfect crystal of Mg32O32 and the deleted one Mg atom was substituted with a Pt atom to maintain a neutral charge. The cell parameter of the cubic unit cell was a = 8.42240 Å, which is the same as that for a perfect crystal MgO. We found that the cell parameter is slightly increased after the optimization of cell parameters using DFT; however, the change of a cell parameter would be smaller if we consider random substitution with a larger unit cell. Therefore, in the following DFT calculations, the unit cell parameter was fixed to those for the original MgO crystal parameters.

Results and Discussion

Pt L3-Edge EXAFS Studies

Figure a shows the radial distribution functions (i.e., Fourier transform, FTs) of the Pt-L3 edge EXAFS data of Pt NPs supported on different types of supports and thermally treated at 1073 K. The k3χ(k) data are provided in Figure S1a. The data obtained by the curve fitting analysis for Pt/MgO are provided in Table .

Figure 1

Pt L3-edge EXAFS distribution functions of (a) Pt foil, PtO2, and Pt NP loaded on MgO, Al2O3, and SiO2, and heat-treated at 1073 K, and (b) Pt NP/MgO and (c) Pt(acac)2/MgO treated at different temperatures. Fourier transform range: 30–150 nm–1.

Table 1

Curve Fitting Analysis of Pt L3-Edge EXAFS Data Measured at Room Temperature for Pt NPs and Pt(acac)2 Loaded on MgO Treated at 1073 and 873 K in Air, Respectively

sample	scatter	CNa	R (Å)b	ΔE0 (eV)c	DW (Å)d	Rf (%)e
Pt NP/MgO treated at 1073 K	O	5.5 ± 0.5	2.04 ± 0.01	1	0.054	1.8
	Mg	9.2 ± 1.3	3.01 ± 0.01	5	0.079
	O	9.8 ± 2.9	3.68 ± 0.01	6	0.061
Pt(acac)2/MgO treated at 873 K	O	5.6 ± 0.5	2.03 ± 0.01	2	0.060	1.2
	Mg	11.1 ± 1.3	3.01 ± 0.01	6	0.080
	O	10.0 ± 2.8	3.69 ± 0.01	8	0.059
MgOf	O	(6)	(2.11)
	Mg	(12)	(3.01)
	O	(8)	(3.69)

Coordination number.

Bond distance.

Difference in the origin of photoelectron energy between the reference and the sample.

Debye–Waller factor.

Residual factor.

Data of X-ray crystallography. Fourier transform range: 3–13 Å–1. Fourier filtering range: 1.3–3.5 Å.

The EXAFS-FTs of Pt/Al2O3 and Pt/SiO2 agreed well with that of the Pt foil, indicating that the agglomeration of loaded Pt nanoparticles occurred on these supports, which was further confirmed by the appearance of the corresponding Pt diffractions in the XRD patterns, as discussed later. The EXAFS-FT of Pt/MgO was however considerably different from those of Pt/Al2O3 and Pt/SiO2; it exhibited two peaks at 1.7 and 2.7 Å (phase shift uncorrected). The first peak could be readily assigned to the Pt–O bond through comparison with the spectrum of the PtO2 reference. The second peak at 2.7 Å was assigned to the Pt–Mg bond based on the curve fitting analysis, and the bond distance corresponding to the peak was calculated to be 3.01 Å, which agreed well with that of the nearest-neighboring Mg–Mg (3.01 Å) bond in MgO (Table ). The possibility of the formation of a Mg–Pt composite oxide (Mg(PtO2)3)[19] may be excluded because the bond distance of the nearest-neighboring Pt–Mg (3.48 Å) bond in Mg(PtO2)3 was much longer than that in the MgO crystal (3.01 Å). The formation of the spinel-type oxide of Mg2PtO4 may also be excluded because the nearest-neighboring Pt–Mg (3.05 Å) is longer in Mg2PtO4 than that in the MgO crystal (3.01 Å), contrary to the previous report.[20] In addition, in the case of Mg2PtO4, the Pt–Pt bond signal would be located at 3.10 Å, which should overlap with the second shell of the Pt L3-edge EXAFS of Pt/MgO found here. However, the second shell of the EXAFS of Pt/MgO that appeared at 2.7 Å could be fitted with the Pt–Mg bond without the contribution of Pt–Pt. Furthermore, no other diffraction assignable to the MgO crystal was found in the XRD pattern of Pt/MgO, as discussed later. The EXAFS oscillation and the FT of Mg2PtO4 are simulated with FEFF8.0 code and compared with those of Pt NP/MgO treated at 1073 K (Figure S2). The enhanced envelope of the oscillation was found up to k = 15 (Å–1) in the EXAFS k3χ(k) of Mg2PtO4 compared with that of the Pt NP/MgO due to the presence of heavy-metal atoms (Pt) around Pt center in the former. In the EXAFS-FT of Mg2PtO4, a clear split of the peak at 2.1–3.5 Å was observed unlike that of Pt NP/MgO. The difference in the EXAFS oscillation and FT supports that the Mg2PtO4 spinel is not included in the Pt NP/MgO. The coordination number (CN) of the Pt–Mg bond was calculated to be 9.2 ± 1.3, which is much smaller than that of the nearest-neighboring Mg–Mg bond (CN = 12). This is probably because the Pt ions in the solid solution were located close to the surface of MgO. The distance of the nearest-neighboring Pt–O bond was calculated to be 2.04 Å, which is shorter than that of the Mg–O bond (2.11 Å) in MgO by 0.07 Å. The CN of the nearest-neighboring Pt–O bond was calculated to be 5.5 ± 0.5, which is almost consistent with that of the nearest-neighboring Mg–O bond of MgO (CN = 6). The curve fitting of the Pt L3-edge EXAFS of Pt NP/MgO heat-treated at 1073 K could be improved by including the second neighboring Pt–O bond, as shown in Figure a and Table . The bond distance of the second neighboring Pt–O bond was calculated to be 3.68 Å, which is close to that of the second neighboring Mg–O bond in MgO (3.69 Å), as listed in Table . The formation of the Pt–MgO solid solution is supported by the agreement of the bond distances.

Figure 2

k3χ(k) (black lines) and simulated (blue lines) Pt L3-edge EXAFS oscillations of (a) Pt NP/MgO treated at 1073 K and (b) Pt(acac)2/MgO treated at 873 K.

Figure b shows the EXAFS-FTs of Pt NP/MgO treated in the range of 673–1273 K and pristine Pt NP. The corresponding k3χ(k) data are provided in Figure S1b. In the EXAFS-FT of pristine NP, two peaks assignable to the Pt–N and Pt–Pt bonds are observed at 2.0 and 2.6 Å (phase shift uncorrected), respectively. The appearance of the Pt–N peak indicated that the Pt NPs were coordinated with the PVP polymer. For the sample heated at 673 K, a new peak assignable to the Pt–O bond appeared at 1.7 Å owing to the partial oxidation of Pt NPs during thermal treatment in air. The intensity of the Pt–O bond increased after heating 873 K, while that of the Pt–Pt bond decreased, suggesting that the oxidation and dispersion of Pt progressed. The EXAFS-FTs of the samples heated at 973 and 1073 K were similar; they showed two peaks at 1.7 and 2.7 Å, which were assigned to Pt–O and Pt–Mg bonds, as already discussed above. With a further increase in the thermal treatment temperature to 1173 and 1273 K, the peak intensity of Pt–O decreased progressively, while a peak corresponding to the second coordination sphere emerged at 2.6 Å (phase shift uncorrected), probably due to the reappearance of metallic Pt–Pt bonds.

Figure a shows the CNs of the Pt–O, Pt–Mg, and Pt–Pt (metal) bonds of Pt NP/MgO plotted as a function of the thermal treatment temperature. The changes in the CNs of Pt–O and Pt–Mg bonds were similar; the curves showed a bell shape centered at 973–1073 K. At these temperatures, the Pt–Pt bond was not observed, indicating that the formation of the Pt–MgO solid solution was completed at temperatures exceeding 973 K. On the other hand, the change in CN(Pt–Pt) indicated an opposite tendency to those of the Pt–O and Pt–Mg bonds. That is, the CN(Pt–Pt) decreased between 673 and 873 K. The disappearance of the Pt–Pt bond at 973 K indicated that the formation of the solid solution progressed in this temperature range. When the thermal treatment temperature was increased further to 1073 and 1273 K, the CN(Pt–Pt) increased to 9.8. The reappearance of the metal Pt–Pt bond indicated that the Pt–MgO solid solution was no longer stable at temperatures higher than 1173 K and that Pt0 and MgO started to segregate from Pt–MgO solid solution.

Figure 3

Coordination numbers of the nearest-neighboring Pt–O, Pt–Mg, and Pt–Pt bonds determined by Pt L3-edge EXAFS as a function of the thermal treatment temperature for (a) Pt NP/MgO and (b) Pt(acac)2/MgO.

The distances of the Pt–O, Pt–Mg, and Pt–Pt (metal) bonds of Pt NP/MgO are plotted as a function of the thermal treatment temperature in Figure a. The distances were not dependent on the treatment temperature.

Figure 4

Distances of nearest-neighboring Pt–O, Pt–Mg, and Pt–Pt bonds determined by Pt L3-edge EXAFS as a function of the thermal treatment temperature for (a) Pt NP/MgO and (b) Pt(acac)2/MgO.

Figure c shows the EXAFS-FTs of pristine Pt(acac)2 diluted with BN and Pt(acac)2/MgO treated at different temperatures. The corresponding k3χ(k) data are provided in Figure S1c. In the EXAFS of pristine Pt(acac)2, a single peak assignable to tetracoordinated Pt–O bonds appeared at 1.7 Å (phase shift uncorrected). The spectra of Pt(acac)2/MgO treated at 873 and 973 K were similar to those of Pt NP/MgO treated at 973 and 1073 K. This similarity indicated that a Pt–MgO solid solution formed in the case of Pt(acac)2/MgO as well. The curve fitting data of Pt(acac)2/MgO treated at 873 K are included in Figure b and Table . The bond distance in the Pt–MgO solid solution prepared using Pt(acac)2 as the Pt precursor was close to that of Pt NP/MgO, for which the distances of short and long Pt–O bonds and the Pt–Mg bond were observed to be 2.03, 3.69, and 3.01 Å, respectively. The temperature-dependent changes in the CNs are shown in Figure b. The CN(Pt–Pt) became null at 873 and 973 K; at the same time, the maximum CN(Pt–Mg) value (11.5) was reached. This change indicated that a Pt–MgO solid solution was formed at 873 and 973 K for Pt(acac)2/MgO, at a temperature lower than that for Pt NP/MgO by 100 K; the Pt–MgO solid solution was formed between 973 and 1073 K for the latter. The maximum CN(Pt–Mg) value of Pt(acac)2/MgO (CN = 11.5) was larger than that of Pt NP/MgO (CN(Pt–Mg) = 9.2). This difference suggested that the location of Pt ions was different; in the case of Pt(acac)2/MgO, Pt ions were located deep inside the crystals of MgO, while they were located close to the surface of MgO in Pt NP/MgO. The distances of the Pt–O, Pt–Mg, and Pt–Pt (metal) bonds of the solid solution of Pt(acac)2/MgO (Figure b) were consistent with those of Pt NP/MgO (Figure a).

Pt L3-Edge XANES

X-ray absorption near-edge structure (XANES) is sensitive to long-range order up to 10 nm because of the large mean free path of the photoelectron in the energy range of 1–100 eV above the absorption edge.[21]Figure a shows the Pt-L3 edge XANES profiles of the Pt NP/MgO treated at 1073 K, Pt(acac)2/MgO treated at 973 K, and reference samples. The inflection point of the white line tended to shift to a higher energy with an increase in the valence of Pt in the reference compounds (Pt foil, PtCl2, PtO2, and H2Pt(OH)6). This tendency is consistent with that reported previously for these samples.[22] The energy of the peak maximum of the white line in the XANES profile of Pt NP/MgO and Pt(acac)2/MgO matched with those of PtO2 and H2Pt(OH)6 (11 568 eV), indicating that the valence of Pt in Pt/MgO was 4+. The edge energy at the inflection point was plotted as a function of the thermal treatment temperature (Figure b). The dependence curve was bell-like in the cases of both Pt NP/MgO and Pt(acac)2/MgO, for which the highest energy was attained at ∼1000 K. This relationship suggested that the formation of the Pt–MgO solid solution progressed up to 1000 K as the treatment temperature was increased. With a further increase in the temperature, the edge energy decreased, probably due to the segregation of the Pt–MgO solid solution to form metallic Pt species, as indicated by the Pt L3-edge EXAFS results presented earlier (Figures and 3) and XRD patterns discussed later. Apart from the white line, another characteristic peak appeared in the Pt L3-edge XANES profile at 11 579 eV in both Pt NP/MgO and Pt(acac)2/MgO (Figure a). A similar peak was observed in the Ni K-edge XANES of the NiO–MgO solid solution[23] (Figure S3). A similar peak was found in the Mg K-edge XANES of MgO simulated with FEFF8.0 (Figure S4). However, it should be noted that the intensity of the white line of Pt NP/MgO and Pt(acac)2/MgO was higher than that of PtO2. The white line of the Pt-L3 edge XANES results from the electron transition from the 2p to 5d band of Pt.[24] It has been reported that the intensity of the white line is higher for hexacoordinated Pt4+ cations than for tetracoordinated Pt4+.[25] The intensity and energy of the white line of the Pt–MgO solid solution matched with those of hexacoordinated H2Pt(OH)6,[26] indicating that Pt4+ in the Pt–MgO solid solution was hexacoordinated with O2– ions. This inference was supported by the EXAFS analysis, namely, CN(Pt–O) in the solid solution was ca. 6. The ionic radius of Pt4+ has been reported to be 0.625 Å, which is comparable to that of hexacoordinated Mg2+ (0.720 Å).[27] The similarity of the ionic radii resulted in the formation of a Pt–MgO solid solution. The difference in the ionic radii is 13%, which is within the general criteria for the formation of a solid solution (<15%, Hume–Rothery’s rules[28]). The Pt L3-edge XANES of Pt/MgO prepared by impregnation with Pt(acac)2 thermally treated at 973 K agreed with that of the Pt/MgO prepared using Pt NPs at 1073 K (Figure a). The formation of the Pt–MgO solid solution at these temperatures was already confirmed by EXAFS analyses. Therefore, we infer that the formation of Pt–MgO occurred commonly with the use of both Pt NP and Pt(acac)2 as Pt precursors. The intensity of the white line is plotted as a function of the thermal treatment temperature in Figure c. The curves of both Pt NP/MgO and Pt(acac)2/MgO show a bell-like shape, as was found for the change in the CN(Pt–Mg) obtained by the curve fitting analysis (Figure a,b).

Figure 5

(a) Pt L3-edge XANES profiles of Pt NP/MgO treated at 1073 K, Pt(acac)2/MgO treated at 973 K, and reference samples. (b) Edge energy and (c) intensity of white lines of Pt L3-edge XANES plotted as a function of the thermal treatment temperature for Pt NP/MgO and Pt(acac)2/MgO. (d) Dependence of the mean valence of Pt on the thermal treatment temperature of Pt NPs and Pt(acac)2 loaded on MgO determined by Pt L3-edge XANES; the valence of 1.0 wt % Pt NP/MgO treated at 1073 K was assumed to be 4+.

We attempted to estimate the mean valence of Pt loaded on MgO, assuming that the valence of Pt in Pt NP/MgO treated at 1073 K was 4+ because the XANES profile of the sample agreed with that of H2Pt(OH)6. For this purpose, the Pt L3-edge XANES profiles of Pt NP/MgO and Pt(acac)2/MgO treated at different temperatures were fitted with those of Pt foil and Pt NP/MgO treated at 1073 K after multiplying with appropriate weights. An example of fitting the XANES of Pt(acac)2/MgO treated at 1073 K is shown in Figure S5. Figure d shows the dependence of the change in the average valences of Pt NP/MgO with 1.0 wt % Pt and Pt(acac)2/MgO with 0.7–3.0 wt % Pt on the treatment temperature. In the case of Pt NP/MgO, an increase in the mean Pt valence up to 4+ was observed between 673 and 973 K, followed by a decrease in the mean Pt valence to 1.1 in the temperature range of 1073–1273 K. A similar change in the mean valence of Pt was observed for 0.7 wt % Pt(acac)2/MgO, in which Pt maintained a valence of 4+ between 873 and 1073 K. The temperature range for the formation of Pt4+ narrowed (873–973 K) for 1.0 wt % Pt(acac)2/MgO. Although 3.0 wt % Pt(acac)2/MgO also yielded a bell-shaped curve of Pt valence vs temperature, its maximum mean valence declined to 2.9 at 873 K. The change in the mean Pt valence of Pt(acac)2/MgO depending on the Pt loading suggested that the temperature range for the formation of Pt–MgO became narrower with an increase in the Pt loading, probably due to the partial formation of the Pt–MgO solid solution.

XRD Patterns

Figure a shows the enlarged images of the XRD patterns for Pt NP/MgO treated at different temperatures. The unenlarged images are shown in Figure S6a. The XRD pattern of the as-prepared Pt/MgO agreed with that of pristine (as-received) MgO (Figure S6b) containing a small amount of Mg(OH)2. The XRD patterns of Pt NP/MgO treated between 673 and 1073 K agreed with that of MgO, and no new peaks assignable to the Mg2PtO4 nor metal Pt were found.[29] The feature was different from Pt/Al2O3 and Pt/SiO2 treated at 1073 K, in which intense diffraction assignable to the metal Pt appeared (Figure S7). Figure a shows the intensity of the diffraction peak assignable to the MgO (200) facet that appeared at 42.8° in Pt NP/MgO and MgO as a function of the treatment temperature. An enhancement in the intensity of the diffraction peak of MgO (200) was observed for Pt NP/MgO compared with that of MgO when the comparison was made at the same treatment temperature, probably due to the replacement of Mg2+ in MgO with Pt4+, which promoted the X-ray diffraction efficiency of the MgO lattice. The enhancement of the diffraction intensity was observed not only for the MgO (200) facet but also for diffractions assignable to other facets (Figure S6a,b). In the XRD patterns of Pt NP/MgO treated at 1173 and 1273 K, new peaks assignable to metal Pt appeared (Figure a). The appearance of these peaks agreed well with the EXAFS data, in which the signal of the Pt–Pt bond appeared at the corresponding temperatures.

Figure 6

XRD patterns of (a) Pt NP/MgO and (b) Pt(acac)2/MgO treated at different temperatures.

Figure 7

Intensity of the (200) plane measured from the XRD patterns of (a) Pt NP/MgO and MgO, and (b) Pt(acac)2/MgO and MgO treated with boiling water.

The enlarged images of the XRD patterns for Pt(acac)2/MgO treated at different temperatures are shown in Figure b. The corresponding unenlarged images of Pt(acac)2/MgO and boiling-water-treated MgO are shown in Figure S8a,b, respectively. In the XRD pattern of the as-prepared Pt(acac)2/MgO, peaks assignable to a mixture of Mg(OH)2 and MgO were observed (Figure b). This is likely because of the formation of Mg(OH)2 during the impregnation of MgO with Pt(acac)2, considering that aqueous impregnation was carried out in a water bath using boiling water. The diffractions assignable to Mg(OH)2 diminished, while those of MgO remained unchanged in the XRD pattern of the sample treated at 673 K, probably due to the dehydration of Mg(OH)2 to MgO. An increase in the intensity of the (200) facet in Pt(acac)2/MgO compared with that in the MgO treated with boiling water was observed when the comparison was made at the same treatment temperature, similar to the case of Pt NP/MgO, as shown in Figure b. In the case of Pt(acac)2/MgO heated at 1273 K, peaks assignable to metal Pt appeared (Figure b) as in the case of Pt NP/MgO (Figure a). The similar tendencies of change suggest that the segregation of metallic Pt and MgO from Pt–MgO solid solution occurred commonly in both Pt NP/MgO and Pt(acac)2/MgO as a result of the thermal treatment at 1273 K.

N2 Adsorption Isotherms

The N2 adsorption isotherms of boiling-water-treated MgO and Pt(acac)2/MgO prepared by the impregnation method are displayed in Figure S9a,b, respectively. MgO was subjected to boiling water treatment for comparison with the Pt(acac)2/MgO prepared in a water bath using boiling water. The former and the latter showed much higher isotherms compared to those of MgO (Figure S10a) and Pt NP/MgO (Figure S10b) when the comparison was made for samples heated at temperatures lower than 773 K. This is probably because the formation of Mg(OH)2 progressed owing to the reaction between MgO and water as confirmed with XRD patterns (Figure b).[30] The specific surface areas calculated from the N2 adsorption isotherms are shown in Figure . A monotonic decrease in the specific surface area was observed for the boiling-water-treated MgO sample. Although the specific surface area of the Pt(acac)2/MgO showed a similar tendency, its surface area was lower than that of the boiling-water-treated MgO, when the comparison was made at the same temperature. This is probably because of the formation of the Pt–MgO solid solution via the inclusion of Pt atoms in the defect sites of MgO,[31] which caused a decrease in the surface area and pore volume of the MgO phase. The specific surface area of the MgO did not change in the temperature range of 673–1273 K (Figure , O). The surface areas of the Pt NP/MgO samples were slightly lower than those of MgO, similar to the case of Pt(acac)2/MgO.

Figure 8

Specific surface areas of Pt NP and Pt(acac)2 loaded on MgO, bare MgO, and boiling-water-treated MgO determined by N2 adsorption as a function of the thermal treatment temperature.

TG-DTA Analysis

Figure a shows the TG-DTA curves of the as-received MgO and Pt NP/MgO. In the TG-DTA curve of the as-received MgO, an endothermic peak in the DTA curve appeared at 625 K; at the same time, its weight decreased by 10%. This weight loss was likely caused by the dehydration of Mg(OH)2, which coexisted with MgO as the major component as confirmed by the XRD pattern (Figure S6b).[32] In the case of Pt NP/MgO, an additional exothermic peak appeared at 565 K apart from endothermic peak at 625 K in the DTA curves (heat flow). The former could be attributed to the combustion and removal of the PVP polymer, considering that it has been reported that removal of the PVP polymer in PVP-protected Pt occurs up to 773 K.[33]

Figure 9

TG-DTA curves of (a) Pt NP/MgO and MgO and (b) Pt(acac)2/MgO and MgO treated with boiling water.

The TG-DTA curves of the MgO sample treated with boiling water and Pt(acac)2/MgO together with Mg(OH)2 are shown in Figure b. The endothermic peak was observed in the DTA curve of boiling-water-treated MgO at 625 K, accompanied by a weight loss of 27%. The extent of weight loss was larger than that of the untreated MgO (Figure a), probably due to the formation of Mg(OH)2 during the preparation step. In agreement with this, the TG-DTA curve of Mg(OH)2 exhibited a similar endothermic peak between 570 and 630 K, with a weight loss of 34%. This weight loss may be attributed to the dehydration of Mg(OH)2 to MgO, considering that the XRD studies indicated the transformation of Mg(OH)2 to MgO occurred at a temperature lower than 673 K (see Figure S8b). The dehydration of Mg(OH)2 to MgO has been reported to occur via the dehydration of the −OH groups in monolayer sheets.[34] A similar change was observed for Pt(acac)2/MgO. However, a small exothermic peak was observed additionally at 450 K, which could be attributed to the combustion of the ligand (i.e., acac ligand).

H2-TPR Analysis

Figure a,b presents the H2-TPR plots of Pt NP/MgO and Pt(acac)2/MgO thermally treated at different temperatures, together with that of pristine PtO2, respectively. Two major peaks appeared at 319 and 450 K in the H2-TPR of PtO2, which could be attributed to the reduction of Pt4+ to Pt2+ and Pt2+ to Pt0, respectively.[35] In the H2-TPR plots of Pt NP/MgO treated at 1073 K (Figure a), a major reduction peak appeared at 620 K, at a temperature much higher than that of PtO2, probably due to the formation of the Pt–MgO solid solution, which retarded the reduction of Pt4+. A similar H2-TPR profile was observed in Pt(acac)2/MgO treated at 973 K (Figure b). Such a shift was previously observed for Pt/Al2O3, in which oxidized species strongly interacted with the Al2O3 support.[36] The H2/Pt ratio of Pt/MgO was estimated based on the H2-TPR curves (Figure c). The highest value of the H2/Pt ratio was observed in the samples treated at 973 K in both Pt NP/MgO and Pt(acac)2/MgO. The temperature almost agreed well with those of CN(Pt–Mg) found in Pt L3-edge EXAFS (Figure a,b).

Figure 10

H2-TPR profiles of (a) Pt NP/MgO and (b) Pt(acac)2/MgO treated at different temperatures, and pristine PtO2. (c) H2/Pt ratio plotted as a function of the thermal treatment temperature measured with H2-TPR.

Dispersion of Pt Measured through CO Adsorption

Figure shows the dispersion values of Pt in Pt NP/MgO, Pt NP/Al2O3, and Pt NP/SiO2, together with that of Pt(acac)2/MgO as a function of the thermal treatment temperature. The dispersion of Pt loaded on Al2O3 and SiO2 was less than 6%; it was lower than those of Pt NP/MgO and Pt(acac)2/MgO when the comparison was made at the same treatment temperature. The low dispersion of Pt NPs loaded on Al2O3 and SiO2 indicated severe aggregation of Pt progressed on these supports, as confirmed by the appearance of peaks assignable to metallic Pt in the corresponding XRD patterns (Figure S7). The aggregation of Pt on Al2O3 and SiO2 indicates that the interaction of Pt with these supports was weak. In the case of Pt(acac)2/MgO, although the dispersion decreased with increasing temperature, the dispersion value increased again at 1173–1273 K, probably because of the segregation of metallic Pt and MgO, as confirmed by the Pt L3-edge EXAFS (Figures c and 3b) and XRD results (Figure b). The dispersion of Pt in Pt NP/MgO decreased continuously up to 1273 K, and a shoulder appeared at 1173 K. The appearance of the shoulder suggested the occurrence of a similar phenomenon (Pt segregation) as that in the Pt(acac)2/MgO (Figure ).

Figure 11

Dependence of the dispersion of Pt in Pt NP/MgO and Pt(acac)2/MgO on the thermal treatment temperature measured with CO adsorption.

TEM Study

The TEM images of untreated and thermally treated Pt NP/MgO samples are shown in Figure . The TEM image of as-prepared Pt NP/MgO exhibited well-dispersed Pt particles of ∼2.0 nm diameter (Figure a), while Pt particles could not be found in the TEM image of Pt NP/MgO thermally treated at 1073 K, probably due to the formation of a Pt–Mg solid solution (Figure b). In addition, the shape of the MgO support turned angular owing to the crystallization and sintering of MgO at 1073 K. Pt crystals were not found in the TEM image of Pt NP/MgO treated at 1273 K as well (Figure c) despite the appearance of small diffraction assignable to metal Pt in XRD patterns (Figure b). Probably, this is because most of the Pt present as the solid solution with MgO even after treatment at 1273 K.

Figure 12

TEM images of Pt NPs loaded on MgO: (a) as-prepared samples, (b) the sample thermally treated at 1073 K, and (c) the sample thermally treated at 1273 K.

The structure of the Pt–MgO solid solution was simulated by calculations. For this, two Mg2+ ions in the MgO lattice were replaced with one Pt4+ ion. The location of Pt4+ in the MgO crystal remained almost unchanged after the replacement of Mg2+ with Pt4+, as shown in Figure . The mean distance of the nearest-neighboring Mg–O bond was calculated to be 2.064 Å (Figure S11, Table S1). The distance was shorter than that of Mg–O in the MgO crystal (2.106 Å), and a characteristic shrinkage of the Pt–O distance in the solid solution was observed compared with that of the Mg–O as observed with the Pt L3-edge EXAFS. This is probably because Pt4+ with a higher valence than that of Mg2+ attracted the O2– anion, leading to the shrinkage of the Pt4+–O2– bond in the solid solution. On the other hand, the nearest-neighboring Pt–Mg distance in the Pt–MgO solid solutions (Figure S12) was close to that of the Mg–Mg bond in MgO (3.01 Å), which was consistent with the EXAFS data (Table ).

Figure 13

Optimum structure of the Pt–MgO solid solution calculated by DFT.

Conclusions

The formation and segregation processes of a Pt–MgO solid solution were analyzed in detail using Pt L3-edge XAFS coupled with other techniques. The temperature for the formation of a stable Pt4+–MgO solid solution was observed to be in the ranges of 973–1073 and 873–973 K when Pt NP and Pt(acac)2 (Pt: 1.0 wt %) were employed as the Pt precursors, respectively. The formation of the Pt4+–MgO solid solution progressed via Pt0, while the thermal treatment of the Pt–MgO solid solution at 1173 or 1273 K resulted in the segregation of Pt and MgO again. The behavior of Pt indicated that the oxidation state as well as the location of Pt changed sensitively depending on the thermal treatment temperature. Such dissolution in MgO and segregation behaviors of Pt suggested the possibility of redispersing the Pt species, using the strong interaction between Pt and MgO.