Solid-State Dispersions of Platinum in the SnO2 and Fe2O3 Nanomaterials

The dispersion of platinum (Pt) on metal oxide supports is important for catalytic and gas sensing applications. In this work, we used mechanochemical dispersion and compatible Fe(II) acetate, Sn(II) acetate and Pt(II) acetylacetonate powders to better disperse Pt in Fe2O3 and SnO2. The dispersion of platinum in SnO2 is significantly different from the dispersion of Pt over Fe2O3. Electron microscopy has shown that the elements Sn, O and Pt are homogeneously dispersed in α-SnO2 (cassiterite), indicating the formation of a (Pt,Sn)O2 solid solution. In contrast, platinum is dispersed in α-Fe2O3 (hematite) mainly in the form of isolated Pt nanoparticles despite the oxidative conditions during annealing. The size of the dispersed Pt nanoparticles over α-Fe2O3 can be controlled by changing the experimental conditions and is set to 2.2, 1.2 and 0.8 nm. The rather different Pt dispersion in α-SnO2 and α-Fe2O3 is due to the fact that Pt4+ can be stabilized in the α-SnO2 structure by replacing Sn4+ with Pt4+ in the crystal lattice, while the substitution of Fe3+ with Pt4+ is unfavorable and Pt4+ is mainly expelled from the lattice at the surface of α-Fe2O3 to form isolated platinum nanoparticles.


Introduction
The nanomaterials based on noble metal nanoparticles and semiconducting transition metal oxides have found applications in electrochemistry, photochemistry, biosensing, catalysis and gas sensing. In particular, the Pt nanoparticles dispersed on the non-expensive SnO 2 or Fe 2 O 3 supports have been used in various catalytic reactions or in gas sensing applications. For example, Chen et al. [1] hydrothermally synthesized a Pt/Fe 2 O 3 nanocomposite catalyst with facet and defect structure for the catalytic oxidation of formaldehyde under ambient conditions. Lang et al. [2] reported the synthesis of a thermally stable Pt single-atom catalyst dispersed on an Fe 2 O 3 support. Experimental and computational modeling studies showed that the reducibility of the iron oxide is crucial for the anchoring of the isolated Pt atoms. Ren et al. [3] controlled the dispersion of platinum on Fe 2 O 3 and CeO 2 by using an ethanediamine chelate ligand and rapid thermal treatment. Liu et al. [4] used the co-precipitation method to disperse Pt on FeO x support, while Li et al. [5] used the colloidal deposition method to prepare Pt/Fe 2 O 3 catalysts for low-temperature oxidation CO to CO 2 at room temperature. Li et al. [6] investigated the redispersion behavior of Pt on the surface of Fe 2 O 3 . They found that Pt can be effectively redispersed on the surface of Fe 2 O 3 when it is alternately treated with oxidative and reductive atmosphere. The results of CO oxidation showed that the catalytic activities of Pt/Fe 2 O 3 increased with the decrease of Pt particle size. D'Arienzo et al. [7] reported the one-step preparation of SnO 2 and Pt-doped SnO 2 in the form of inverted opal films obtained by sol-gel synthesis and dip coating. The electrical sensitivity of the inverted opal-based films was enhanced by Pt doping, and it was suggested that the increased porosity and electronic sensitization by the Pt dopant have a synergistic effect in improving the electrical response of the opal films. Dong et al. [8] reported a combustion method to prepare porous Pt-functionalized SnO 2 sheets using urea as fuel. The synthesized Pt-SnO 2 sheets exhibited better response and lower operating temperature than pristine SnO 2 for gas detection of isopropanol gas.
Usually, the Pt dispersions on Fe 2 O 3 and SnO 2 metal oxide supports are prepared by wet impregnation or by sol-gel method using metal chloride precursors such as FeCl 3 , SnCl 4 and H 2 PtCl 6 . In contrast, in this work, we use solid-state dispersion technique (mechanochemical) and compatible Fe(II)-acetate, Sn(II)-acetate and Pt(II)-acetylacetonate powders as organic precursors to better disperse Pt on Fe 2 O 3 and SnO 2 supports and avoid cation (sodium, potassium) [9] or anion (chloride) [10,11] contamination of the synthesized nanomaterial. It is expected that the exclusive use of the divalent metal cations (Fe(II), Sn(II) and Pt(II)) and strongly chelating organic ligands, such as acetate and acetylacetonate, will allow homogeneous mixing of the various M(II) precursors, limit crystal growth and lead to the uniform dispersion of the Pt catalyst on the metal oxide supports.
The pristine hematite (α-Fe 2 O 3 ) sample FE-0 was synthesized by grinding Fe(II) acetate in a planetary mill "FRITSCH PULVERISETTE 7 premium line", Idar-Oberstein, Germany, with 40 balls of zirconia size 5 mm at a speed of 1000 rpm for two hours. After grinding, the orange-red powder was stored in a tube furnace for 30 min in an argon stream at 200 • C, 30 min in an argon stream at 400 • C, 1 h in an argon stream at 600 • C and 2 h in an air stream at 600 • C; a dark red powder was obtained, which was assigned as a sample FE-0.
The pristine tin oxide (cassiterite) sample SN-0 was synthesized in the same manner as hematite sample FE-0, except that the starting chemical was tin(II) acetate instead of iron(II) acetate. A white powder was obtained and assigned as sample SN-0.
The platinum was dispersed in the SnO 2 or Fe 2 O 3 in three ways: (i) In one step, by mixing powdered Fe (II) acetate or Sn (II) acetate precursor with 1 mol% Pt(acac) 2 and activation in a planetary mill, followed by annealing in a tube furnace for 30 min in an argon stream at 200 • C, 30 min in an argon stream at 400 • C, 1 h in an argon stream at 600 • C and 2 h in an air stream at 600 • C (samples SN-1 and FE-1). (ii) In two steps, where first hematite and cassiterite were synthesized at 600 • C, and then in the second step, these pristine powder samples were mixed with the 1 wt% Pt(acac) 2 powder precursor and homogenized by grinding in a planetary mill at a speed of 400 rpm for 2 h using zirconia 5 mm balls. Subsequently, the homogenized powders were annealed in an argon stream for one hour and in an air stream for two hours at 400 • C (samples SN-2 and FE-2), (iii) In two steps, in which the hematite and the cassiterite were first synthesized at 600 • C, and then in the second step, the pristine powder samples were mixed with 1 wt% Pt(acac) 2 previously dissolved in a certain volume of toluene, and then the paste obtained was homogenized in a planetary mill at a speed of 400 rpm for 2 h using zirconia 5 mm balls and annealed in an argon stream for one hour and in an air stream at 400 • C for two hours (samples SN-3 and FE-3).
Mössbauer spectra were collected in a transmission mode using standard instrumental configuration (WissEl GmbH, Starnberg, Germany). 57 Co in rhodium matrix was used as a source of radiation. The spectrometer was calibrated at room temperature using a standard spectrum of α-Fe foil.
Atomic resolution scanning transmission electron microscope (AR STEM), model Jeol ARM 200 CF, with voltage emission of 200 kV coupled with Gatan Quantum ER system and with electron energy loss spectroscopy and energy dispersive X-ray spectrometry (Jeol Centurio 100), Tokyo, Japan was used.
Nitrogen adsorption measurements at 77 K for Brunauer-Emmett-Teller (BET) analysis and necessary degassing pre-treatment were conducted on Quantachrome Autosorb iQ3 system, Anton Paar QuantaTec 1900 Corporate Drive Boynton Beach, FL, USA.
Temperature-programmed reduction (TPR) experiments were performed using a Quantachrome Autosorb iQ3-AG-C instrument. In a typical experiment, 100 mg of sample was placed in a quartz tube, and the temperature was raised from room temperature to 120 • C in a stream of helium at a constant rate of 20 • C/min. The sample temperature was maintained at 120 • C for 40 min and then cooled to 40 • C. The sample was then heated to 800 • C under a reducing gas mixture (10% H 2 /90% Ar) at a rate of 10 • C/min. A thermal conductivity detector (TCD) was used to measure the changes in thermal conductivity of the effluent gas. At the end of the measurements, the samples were cooled in a stream of nitrogen.
X-ray photoelectron spectra (XPS) were recorded using a SPECS instrument (Berlin, Germany) under ultra-high vacuum (UHV) conditions (the typical pressure was in the range of 10 −7 Pa) with an Al Kα X-ray of energy 1486.74 eV and the Phoibos100 electron energy analyzer. The transmission energy was set to 50 eV for the measurements around the Pt 4f core levels. Numerical fitting of the experimental spectra was performed using the mixed Gaussian-Lorentzian functions after Shirley background subtraction. All XPS spectra were corrected for their binding energy shifts using C 1s peaks at 284.2 eV as reference.       Table 1. All three samples are characterized with a sextet. The isomer shift (δ) of 0.35 mm s −1 , the quadrupole shift of −0.17 mm s −1 and the hyperfine magnetic field (B hf ) of~46.89 mm s −1 can be attributed to hematite in agreement with the XRD results. The hyperfine magnetic field of wellcrystallized hematite has a value of~51.0 mm s −1 [12][13][14]. The B hf and line width (Γ) are very sensitive to the crystallinity, particle size and surface properties of hematite [15][16][17][18]. In the present case, the B hf decrease slightly, and Γ broadened from sample FE-1 to sample FE-3 (Table 1).   Table 1. All three samples are characterized with a sextet. The isomer shift (δ) of 0.35 mm s −1 , the quadrupole shift of −0.17 mm s −1 and the hyperfine magnetic field (Bhf) of ~46.89 mm s −1 can be attributed to hematite in agreement with the XRD results. The hyperfine magnetic field of well-crystallized hematite has a value of ~51.0 mm s −1 [12][13][14]. The Bhf and line width (Γ) are very sensitive to the crystallinity, particle size and surface properties of hematite [15][16][17][18]. In the present case, the Bhf decrease slightly, and Γ broadened from sample FE-1 to sample FE-3 (Table 1).          Figure S6 (Supplementary Materials) shows STEM images of samples FE-1 (a), FE-2 (b) and FE-3 (c). The mean particle sizes of 30.9, 20.9 and 8.6 nm were measured for samples FE-1, FE-2 and FE-3, respectively. Figure 6 shows STEM bright-field images with particle size distributions (inset) of small NPs in samples FE-1 (a, b) and FE-2 (c, d) and dark-field images with particle size distribution (inset) of very small NPs in sample FE-3 (e, f). One can see the well-dispersed small distinct NPs of 2.2, 1.2 and 0.8 nm in samples FE-1, FE-2 and FE-3, respectively. The 0.8 nm NPs in the FE-3 sample are so small that they are virtually invisible on the STEM bright-field image, so they are shown in dark-field mode.  and FE-3 (c). The mean particle sizes of 30.9, 20.9 and 8.6 nm were measured for samples FE-1, FE-2 and FE-3, respectively. Figure 6 shows STEM bright-field images with particle size distributions (inset) of small NPs in samples FE-1 (a, b) and FE-2 (c, d) and dark-field images with particle size distribution (inset) of very small NPs in sample FE-3 (e, f). One can see the well-dispersed small distinct NPs of 2.2, 1.2 and 0.8 nm in samples FE-1, FE-2 and FE-3, respectively. The 0.8 nm NPs in the FE-3 sample are so small that they are virtually invisible on the STEM bright-field image, so they are shown in dark-field mode.                   [19,20]. The H2-TPR curve of the pristine α-SnO2 (sample SN-0) showed that the reduction of SnO2 started at about 180 °C and ended with an "incomplete" maximum at 643 °C. The incomplete reduction of SnO2 to Sn can be attributed to the formation of the SnO and/or Sn metal shell around the α-SnO2 particles. The melting point of tin is very low at 232 °C, so molten tin on the surface of α-SnO2 particles can protect Sn 4+ from reduction. The addition of Pt promotes the reduction of α-SnO2, and the reduction started at a lower temperature (~100 °C) and ended with a well-defined maximum at 652 or 733 °C for samples SN-2 and SN-1, respectively.   [19,20]. The H 2 -TPR curve of the pristine α-SnO 2 (sample SN-0) showed that the reduction of SnO 2 started at about 180 • C and ended with an "incomplete" maximum at 643 • C. The incomplete reduction of SnO 2 to Sn can be attributed to the formation of the SnO and/or Sn metal shell around the α-SnO 2 particles. The melting point of tin is very low at 232 • C, so molten tin on the surface of α-SnO 2 particles can protect Sn 4+ from reduction. The addition of Pt promotes the reduction of α-SnO 2 , and the reduction started at a lower temperature (~100 • C) and ended with a well-defined maximum at 652 or 733 • C for samples SN-2 and SN-1, respectively.

Results
In the pristine α-Fe 2 O 3 (sample FE-0), the reduction starts at 300 • C and shows two peaks at 421 and 629 • C. The first maximum at 421 • C corresponds to the reduction of α-Fe 2 O 3 → Fe 3 O 4 , and the second maximum at 629 • C to the reduction of Fe 3 O 4 → FeO → Fe [6]. Doping with platinum in one step (sample FE-1) significantly increases the reduction of α-Fe 2 O 3 , two reduction maxima at 421 and 629 • C integrate a broad maximum at 360 • C, and a new small and very broad reduction maximum at 146 • C can be attributed to Pt 4+ reduction in Pt 0 and partial Fe 3+ reduction ions bound in Fe-O-Pt groups. When Pt is doped in two steps (samples FE-2 and FE-3), two reduction maxima are again present, with the maximum shifted to lower temperatures (281 and 270 • C) due to α-Fe 2 O 3 → Fe 3 O 4 reduction, while Fe 3 O 4 → FeO → Fe reduction is shifted to higher reduction temperatures (678 • C), and the maximum is at a lower temperature at 635 • C for sample FE-3. For the two-step platinum doping, the Pt 4+ → Pt 0 reduction maxima are well defined and distinct at 112 and 128 • C for samples FE-2 and FE-3, respectively. In the pristine α-Fe 2 O 3 (sample FE-0) and α-SnO 2 (sample SN-0) there is no platinum, and consequently, there is no reduction maxima below 200 • C. In the pristine α-Fe2O3 (sample FE-0), the reduction starts at 300 °C and shows two peaks at 421 and 629 °C. The first maximum at 421 °C corresponds to the reduction of α-Fe2O3 → Fe3O4, and the second maximum at 629 °C to the reduction of Fe3O4 → FeO → Fe [6]. Doping with platinum in one step (sample FE-1) significantly increases the reduction of α-Fe2O3, two reduction maxima at 421 and 629 °C integrate a broad maximum at 360 °C, and a new small and very broad reduction maximum at 146 °C can be attributed to Pt 4+ reduction in Pt 0 and partial Fe 3+ reduction ions bound in Fe-O-Pt groups. When Pt is doped in two steps (samples FE-2 and FE-3), two reduction maxima are again present, with the maximum shifted to lower temperatures (281 and 270 °C) due to α-Fe2O3 → Fe3O4 reduction, while Fe3O4 → FeO → Fe reduction is shifted to higher reduction temperatures (678 °C), and the maximum is at a lower temperature at 635 °C for sample FE-3. For the two-step platinum doping, the Pt 4+ → Pt 0 reduction maxima are well defined and distinct at 112 and 128 °C for samples FE-2 and FE-3, respectively. In the pristine α-Fe2O3 (sample FE-0) and α-SnO2 (sample SN-0) there is no platinum, and consequently, there is no reduction maxima below 200 °C. Figure 11 shows the results of XPS analysis for samples SN-2 (a) and sample FE-2 (b). The presence of small Zr impurities in the samples due to the ball milling process (ZrO2 balls) is shown in the inset of the XPS survey scanning spectrum of the samples SN-2 and FE-2. The high-resolution XPS spectra of Sn 3d, O 1s, and Pt 4f of sample SN-2 are shown in panel (a). The binding energies of Sn 3d3/2 and Sn 3d5/2 at 495.4 and 486.9 eV, respectively, are attributed to Sn 4+ , at 493.7 and 485.2 eV to Sn 2+ , and at 490.8 and 482.5 eV to Sn 0 [21,22], which is consistent with the NIST X-ray photoelectron spectroscopy database. The binding energy of O1s at 531 eV (530.9 eV in Table 2) can be assigned to the O 2-of lattice oxygen. The peaks of Pt 4f are rather weak, which can be attributed to the low concentra-  Figure 11 shows the results of XPS analysis for samples SN-2 (a) and sample FE-2 (b). The presence of small Zr impurities in the samples due to the ball milling process (ZrO 2 balls) is shown in the inset of the XPS survey scanning spectrum of the samples SN-2 and FE-2. The high-resolution XPS spectra of Sn 3d, O 1s, and Pt 4f of sample SN-2 are shown in panel (a). The binding energies of Sn 3d 3/2 and Sn 3d 5/2 at 495.4 and 486.9 eV, respectively, are attributed to Sn 4+ , at 493.7 and 485.2 eV to Sn 2+ , and at 490.8 and 482.5 eV to Sn 0 [21,22], which is consistent with the NIST X-ray photoelectron spectroscopy database. The binding energy of O1s at 531 eV (530.9 eV in Table 2) can be assigned to the O 2− of lattice oxygen. The peaks of Pt 4f are rather weak, which can be attributed to the low concentration of Pt in the samples. The Pt 4f peak was fitted without background subtraction. XPS peaks Pt 4f 5/2 and Pt 4f 7/2 at 77.6 and 74.1 eV, respectively, are attributed to Pt 4+ and at 75.8 and 72.3 eV, to Pt 2+ (Table 2) [23][24][25]. Figure 11, panel (b) shows the high-resolution Fe 2p spectrum. The two distinct peaks at binding energies of 710.8 eV for Fe 2p 3/2 and 724.7 eV for 2p 1/2 are consistent with Fe 3+ in α-Fe 2 O 3 [26]. In addition, two satellite peaks characteristic of α-Fe 2 O 3 at 719.8 eV and 732.5 eV are clearly visible [22,25]. The O 1s peak consists of three deconvoluted peaks at 526.8, 530.1, and 533.1 eV ( Table 1)

Discussion
The dispersion of platinum (Pt) on metal oxide supports is important for catalytic and gas sensing applications. Usually, Pt is dispersed by a wet impregnation method starting from H 2 PtCl 6 (chloroplatinic acid) or K 2 PtCl 6 (potassium hexachloroplatinate). In contrast, in this work, we mechanochemically dispersed Pt in nanocrystalline α-Fe 2 O 3 and α-SnO 2 starting from compatible Fe(II) acetate, Sn(II) acetate and Pt(II) acetylacetonate powders. In a one-step procedure, the Pt(acac) 2 powder was mixed with Fe(II) acetate or Sn(II) acetate powder and homogenized in a planetary mill. Subsequently, the samples were annealed at 600 • C. The dispersion of platinum over α-SnO 2 is clearly different from the Pt dispersion over α-Fe 2 O 3 . Figure 4a shows STEM images of the samples SN-1 with the 10.3 nm nanoparticles, while EDXS analysis ( Figure 5) confirms that these nanoparticles belong to SnO 2 and that there are no isolated Pt clusters. The Sn, O and Pt elements are homogeneously dispersed, strongly suggesting that the platinum is incorporated into the α-SnO 2 structure. In contrast, the STEM image and EDXS analysis of the sample FE-1 (Figures 7a,b and 8) show the well-dispersed isolated Pt nanoparticles with a size of 2.2 nm. Moreover, the size of the Pt nanoparticles could be controlled by changing the experimental conditions. In the two-step procedure, mixing the synthesized pure α-Fe 2 O 3 powder (sample FE-0) with Pt(acac) 2 powder in the planetary mill and then annealing at 400 • C resulted in smaller, well-dispersed, discrete Pt nanoparticles with a size of 1.2 nm. When the Pt(acac) 2 powder was previously dissolved in a certain volume of toluene and then mixed with pure α-Fe 2 O 3 in a planetary mill, the obtained PtNPs are 0.8 nm in size. In the case of SnO 2 samples, there were no isolated PtNPs regardless of the experimental conditions. The unexpectedly different Pt dispersion over α-SnO 2 and α-Fe 2 O 3 reducible metal oxide supports could be explained by the different chemistry and oxidation state of Sn 4+ and Fe 3+ cations. The Sn 4+ cation is easily reducible and can be easily reduced to metallic state Sn 0 [27], as shown by the TPR and XPS results (Figures 10 and 11). Moreover, the TPR results ( Figure 10) showed that Pt promotes the reduction of Sn 4+ . It is known that SnPt nanoparticles can be synthesized from the same organic precursors used in this work, i.e., Sn(II) acetate and Pt(II) acetylacetonate by a polyalcohol reduction in the liquid phase [28]. However, every attempt to find Pt or SnPt in SnO 2 samples was unsuccessful. For instance, the HAADF image of sample SN-1 shows a uniform contrast, whereas Pt nanoparticles and clusters are clearly visible in sample FE-1 ( Figure S7 in the Supplementary Materials). Sn and Pt have very different atomic numbers (50 vs. 78), and the intensity of the HAADF image (besides the thickness) is approximately related to Z 1.7 . In the case of Pt-based particles or Pt-rich surface layers, this should be seen as areas of higher contrast. On the other hand, the ionic radii of Fe 3+ (0.645 Å), Sn 4+ (0.690 Å) and Pt 4+ (0.625 Å) in octahedral coordination are similar, and formally, Pt 4+ could be incorporated as the smallest cation in the crystal lattice of α-SnO 2 and α-Fe 2 O 3 . However, Pt 4+ can be stabilized in the α-SnO 2 structure by replacing Sn 4+ with Pt 4+ in the crystal lattice, resulting in a (Pt,Sn)O 2 solid solution, while the substitution of Fe 3+ with Pt 4+ is unfavorable and Pt 4+ is mainly ejected as Pt 0 from α-Fe 2 O 3 during annealing. Anenburg et al. [29] showed that Pt 4+ doping of hematite was only possible at high pressure, which allowed the oxygen fugacity to be maintained at sufficiently high values to stabilize Pt 4+ in hematite. In contrast, Murata et al. [30] have shown that Pt 4+ is located at the Sn 4+ site of the α-SnO 2 lattice, thus forming a (Pt,Sn)O 2 solid solution at Pt loading up to 10 at%. Our solid-state dispersion of platinum is therefore in agreement with the references [29,30] and showed that the cationic platinum (Pt 4+ ) was homogeneously dispersed in the α-SnO 2 structure under oxidative annealing ( Figure 5), whereas in Fe 2 O 3 , despite the oxidative conditions, Pt 4+ was expelled from the α-Fe 2 O 3 crystal lattice and dispersed mainly as isolated platinum nanoparticles over α-Fe 2 O 3 (Figures 7 and 8).
EDXS results ( Figure 5) have shown that Sn, O and Pt elements are homogeneously dispersed in α-SnO 2 , indicating the formation of a (Pt,Sn)O 2 solid solution.
Platinum is dispersed in α-Fe 2 O 3 mainly in the form of isolated Pt nanoparticles (Figures 7 and 8). The size of the dispersed Pt nanoparticles over α-Fe 2 O 3 can be controlled by changing the experimental conditions and is set to 2.2, 1.2 and 0.8 nm (Figure 7).
The rather different Pt dispersion in α-SnO 2 and α-Fe 2 O 3 is due to the fact that Pt 4+ can be stabilized in the α-SnO 2 structure by replacing Sn 4+ with Pt 4+ in the crystal lattice, while the substitution of Fe 3+ with Pt 4+ is unfavorable, and Pt 4+ is mainly expelled from the lattice at the surface of α-Fe 2 O 3 to form isolated platinum nanoparticles despite the oxidative conditions during annealing.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.

Acknowledgments:
The authors thank Robert Peter and Mladen Petravić for X-ray photoelectron spectroscopy (XPS) measurements and discussions on XPS results.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.