Nanocomposite thin films containing Pt nanoparticles dispersed in an α-Fe₂O₃ matrix by RF sputtering

Because of efforts toward a sustainable society, iron oxides, which are relatively low-cost and abundant resources, have attracted a lot of attention because they exhibit ferrimagnetic and semiconducting characteristics depending on the degree of oxidation. In particular, α-Fe₂O₃ has an energy bandgap of 2.2 eV,¹ which allows it to absorb near the maximum intensity of sunlight, making it a promising candidate for photoelectrochemical water splitting. Specifically, electrons generated in α-Fe₂O₃ are transported from the conduction band to the electrode (e.g., Pt) for reduction reaction with water, while holes in the valence band undergo oxidation reaction. However, since the water splitting efficiency needs improvement due to the low mobility of minority carriers and the short lifetime of carriers generated in α-Fe₂O₃,² a lot of research has focused on improving its transport properties through impurity doping^3–11 and heterostructuring to introduce band offsets.^12–16 Considering the device of the water splitting system, the Pt electrode and α-Fe₂O₃ should be structurally integrated. In fact, in the nanoparticle form, a structure in which Pt is supported on α-Fe₂O₃ has been configured,^17–20 and in the thin film form, a structure in which a Pt layer is stacked on an α-Fe₂O₃ thin film has been demonstrated.²¹ Since the latter is a one-dimensional electron capture in the stacking direction, it is more efficient to place Pt electrodes three-dimensionally in the thin film, which corresponds to a nanocomposite structure with Pt nanoparticles dispersed in α-Fe₂O₃. In addition, the oxidation and reduction reactions of photoelectrochemical water splitting occur on opposite sides in one-dimensional stacked structures, whereas nanocomposite structures should cause both reduction and oxidation reactions at the photo-irradiated surface. Therefore, nanocomposites are also effective from a practical standpoint because they can be fabricated in large areas on low-cost substrates such as glass. According to the phase diagram of the Pt–Fe–O system,²² α-Fe₂O₃ and the Pt phase separate in thermal equilibrium, allowing the formation of nanocomposite structures. However, in chemically synthesized thin films, Pt is solidly dissolved in α-Fe₂O₃, which ionizes to a tetravalent and behaves as an n-type dopant.^23–26 

In this study, preparation of nanocomposite thin films of Pt nanoparticles dispersed in α-Fe₂O₃ is investigated by employing radio frequency (RF) sputtering, which is a physical vapor deposition method. When α-Fe₂O₃ is used as a sputtering target, oxygen defects are often introduced into the thin film, which are compensated for by subsequent heat treatment in the air. In selecting the heat treatment temperature, the oxidation of Pt nanoparticles and the phase transition of iron oxide to α-Fe₂O₃ should be considered. Specifically, Pt nanoparticles with sizes below 15 nm prepared by vacuum deposition turn into PtO_x upon thermal oxidation exceeding 673 K,²⁷ whereas iron oxide undergoes a phase transition to α-Fe₂O₃ at 673 K.²⁸ Hence, a heat treatment temperature of 673 K was employed in this study. Therefore, the relationship between heat treatment time, improvement in oxygen loss, and oxidation of Pt was investigated in detail over a relatively long period of time, in excess of 100 days.

Thin films were prepared by RF sputtering using a target with a Pt metal chip on a 4-in. diameter α-Fe₂O₃ disk. The area of the metal chip was 5 × 5 mm². The chamber was evacuated to 1.5 × 10⁻⁷ Torr before introducing 6N pure Ar, and the film was deposited on a Corning glass substrate cooled by water circulation at 293 K. The distance between the target and substrate was kept constant at 73 mm. The total Ar pressure was fixed at 2.0 × 10⁻³ Torr. The RF power was maintained at 200 W, the deposition time was kept constant at 60 min, and no RF bias was applied to the substrate. The thickness of the thin films varied from 0.54 to 1.1 µm with an increase with increasing Pt concentration.

The film was structurally characterized using x-ray diffraction (XRD) with Cu Kα radiation (Rigaku, MiniFlex600, Japan). The magnetization of the film was measured at room temperature using a vibrating sample magnetometer (Tamagawa, TM-VSM-2430, Japan). The composition of the film in the as-deposited state was analyzed by energy dispersive spectroscopy (EDAX, Phoenix, USA) operating at 10 kV, using α-Fe₂O₃ and Pt as standard samples to calibrate the results of Pt, Fe, and O analyses. The spectral intensities of the O–K, Fe–L, and Pt–M lines generated by electron beam irradiation on the thin film surface were used to estimate the composition. After calibration with the standard samples and ZAF correction (Z: atomic number correction, A: absorption correction, F: x-ray fluorescence correction), the units were converted from wt. % to at. %. The analysis point observed at a magnification of 2000×, corresponding to the electron beam irradiation area, was near the center of the thin film sample (1 cm square). The microstructures of the samples heat-treated in air were observed by transmission electron microscopy (TEM) (JEOL, JEM-ARM200F, Japan) at an acceleration voltage of 200 kV.

Figure 1 shows the results of compositional analysis of Pt-added α-Fe₂O₃ thin films in as-deposited states. As shown in the figure, the Pt concentration tends to increase as the number of Pt chips increases. The composition change in the thin film is closer to the line assuming phase separation of Pt and α-Fe₂O₃ than to the line assuming a solid solution of Pt and α-Fe₂O₃. In addition, the oxygen concentration in the thin film tends to decrease with increasing Pt concentration, approaching the line where Pt and Fe₃O₄ are assumed to phase separate, despite the use of α-Fe₂O₃ as the target. This result suggests that oxygen defects are introduced into the thin film as the Pt concentration increases.

Figure 2(a) shows the optical transmittance spectra of α-Fe₂O₃ thin film without Pt addition heat-treated at 673 K in air. The optical transmittance in the as-deposited state is relatively low, even though the composition shows an excess of oxygen compared to α-Fe₂O₃ at 0 at. % Pt, as shown in Fig. 1. The difference between the two results may be attributed to the relatively thin thickness (0.54 µm), which incorporates oxygen from the glass substrate in the oxygen analysis. The optical transmittance increases with increasing annealing time, reaching almost saturation at 210 m, and the optical absorption edge of α-Fe₂O₃ is clearly observed. This result suggests that the oxygen defects in the α-Fe₂O₃ thin film are compensated for through heat treatment in the air. Figure 2(b) shows the optical transmittance spectra of the α-Fe₂O₃ thin film added with 6.3 at. % Pt heat-treated at 673 K in the air. As in the case without Pt, optical transmittance increases with increasing heat treatment time and almost reaches saturation after 53 days of heat treatment. In other words, the time required to compensate for oxygen defects in the case of Pt addition is longer than that in the case of no Pt addition. This would be closely related to the reductive phase transition from α-Fe₂O₃ to Fe₃O₄ as the oxygen concentration decreases with increasing Pt addition concentration in the as-deposited state (Fig. 1). Since the optical absorption edge is not clear, Raman spectra are shown in Fig. 2(c) to clarify how the iron oxide is formed by heat treatment in the air. In the as-deposited state, a broad peak is observed at 665 cm⁻¹, assigned to the A_1g mode of Fe₃O₄,²⁹ whereas peaks appear at 225, 245, 292, 410, 500, and 612 cm⁻¹, assigned to the A_1g and E_g modes of α-Fe₂O₃²⁹ in the sample annealed at 673 K for 113 days. In addition, the broad peak at 660 cm⁻¹ in the heat-treated sample is attributed to lattice distortions introduced by the many defects in α-Fe₂O₃.³⁰ 

Figure 3(a) shows the XRD patterns focusing on the Pt peak in the α-Fe₂O₃ thin film added with 6.3 at. % Pt with respect to the annealing time at 673 K in the air. The diffraction peak at (111) (closed square) is broad in the as-deposited state, whereas it becomes more distinct with an increase in annealing time. In the peak of iron oxide [Fig. 3(b)], it can be seen that while Fe₃O₄ (closed circle) is present in the as-deposited state, α-Fe₂O₃ (open circle) appears after 82 h of heat treatment, and after 53 days, Fe₃O₄ disappears and finally arrives at a single phase of α-Fe₂O₃. The magnetization at 10 kOe is shown in Fig. 3(c). In the as-deposited state [Fig. 3(b)], the Fe₃O₄ is ferrimagnetic (inset) with 321 emu cm⁻³, whereas the magnetization decreases with annealing time and disappears (inset) after annealing for 113 days. This result is in good agreement with that of the XRD results [Fig. 3(b)]. Therefore, it is clear that heat treatment over a relatively long period of time is needed to compensate for the oxygen deficiency in α-Fe₂O₃.

Figure 4(a) shows the optical transmittance spectra of α-Fe₂O₃ thin films added with Pt after heat treatment in the air at 673 K for 113 days. All spectra also show relatively high optical transmittance regardless of Pt concentration. The optical absorption edge becomes more gradual with an increase in Pt concentration. Figure 4(b) shows the XRD pattern of α-Fe₂O₃ thin film added with Pt after annealing at 673 K in the air for 113 days. Since the XRD patterns at all Pt concentrations consist of α-Fe₂O₃ (open circles) and Pt (closed squares), the oxygen defects that occur in the as-deposited state are ameliorated by heat treatment in the air. The Pt grain size estimated from the Pt (111) diffraction peak by employing Scherrer's formula is shown in Fig. 4(c). While the overall Pt concentration is less than 2 nm in the as-deposited state, it varies from 2 to 15 nm after heat treatment, indicating that the size tends to increase with increasing Pt concentration.

Figure 5(a) shows a high-resolution TEM plan-view of a 6.3 at. % Pt-added α-Fe₂O₃ thin film annealed in the air for 113 days at 673 K. It can be seen that spherical particles of ∼5 nm in size with a relatively dark contrast are dispersed in the film. Fast Fourier transform (FFT) analysis results for the area enclosed by the green circle are shown in the inset. The plane widths of the spots labeled A, B, and C were 0.222, 0.117, and 0.136 nm, respectively, and the angles between the spots were 59° at AOB, 91° at AOC, and 32° at BOC. These values are close to (11-1), (3-1-1), and (2-20) of the standard data for Pt (PDF No. 00-004-0802), suggesting that this nanoparticle is Pt. The observed size is in good agreement with the size of 4.8 nm [Fig. 4(c)] estimated from the XRD peaks. The different image [Fig. 5(b)] shows that an extensive region with a relatively bright contrast is crystallized surrounding the nanoparticles. The results of the FFT analysis of the region marked by the green circle are shown in the inset. The plane widths of spots A, B, and C were 0.181, 0.146, and 0.145 nm, respectively, and the lattice plane angles AOB, AOC, and BOC were 67°, 114°, and 47°, respectively. These values are close to those for spots in the lattice planes (2-24), (-1-24), and (-300) of the standard data for α-Fe₂O₃ (PDF No. 00-033-0664) when the electron beam is incident from [021]. Thus, spots A, B, and C are assigned to (2-24), (-1-24), and (300) of α-Fe₂O₃, respectively. These observations indicate that the fabricated thin film forms a nanocomposite with Pt nanoparticles dispersed in α-Fe₂O₃.

Figure 6(a) shows the high-angle annular dark field scanning TEM plan-view of the α-Fe₂O₃ thin film added with 6.3 at. % Pt after heat treatment in the air for 113 days at 673 K. Densely dispersed spherical particles with relatively bright contrasts are observed, which are detected in Pt [Fig. 6(c)] in the energy dispersive x-ray spectroscopy (EDS) elemental mapping analysis, whereas O [Fig. 6(b)] and Fe [Fig. 6(d)] are observed in the other regions. In other words, nanocomposite thin films with Pt nanoparticles dispersed in α-Fe₂O₃ can be fabricated by sputter deposition using an α-Fe₂O₃ target attached with Pt chips, followed by heat treatment in the air.

In a similar study, when a target with Pt wires placed on a TiO₂ disk was employed for sputtering, Pt ionizes in TiO₂ in the as-deposited state, while Pt phase-separates from TiO₂ to form a composite after heat treatment.³¹ Even though the Pt elements are thermodynamically more stable in the TiO₂ matrix because the difference between the heat of formation of the Pt oxide (i.e., PtO₂) and that of TiO₂ is larger than that of α-Fe₂O₃,³² the Pt elements phase-separate in the α-Fe₂O₃ matrix both in the as-deposited state and after heat treatment. In particular, it should be noted that the Pt nanoparticles are nanoscaled and relatively uniform in size [Figs. 5(a) and 6(a)]. This structure should effectively capture the carriers generated in α-Fe₂O₃ by light absorption. Specifically, for pure α-Fe₂O₃ films without Pt, a thickness of 20 nm³³ or 50 nm³⁴ is optimal due to the relatively short carrier lifetime,² whereas in this study, high-resolution TEM images (Fig. 5) show that Pt nanoparticle electrodes can be brought close to those values. The distance between nanoparticles can be technically controlled by varying the number of Pt chips in the sputter deposition process. Considering photoelectrochemical water splitting, it is expected that the electrons captured by the Pt nanoparticle electrode will cause a reduction reaction at the film surface, while the holes generated in α-Fe₂O₃ will cause an oxidation reaction. Hence, Pt nanoparticles embedded inside the film would not contribute directly to the reduction reaction but might be supplied to the Pt electrode at the surface through the electrical pathway associated with the partial segregation of the Pt nanoparticles. Oxygen defects in α-Fe₂O₃ could be compensated for in a shorter time by increasing the heat treatment temperature, although this study took a relatively long time (i.e., 113 days).

We investigated the preparation of nanocomposite thin films in which Pt nanoparticles are dispersed in α-Fe₂O₃. In the as-deposited state, the increase in Pt addition concentration induces oxygen deficiency in the iron oxide to form Fe₃O₄, followed by heat treatment in the air to transform the iron oxide into a single phase α-Fe₂O₃. The added Pt phase separates from the iron oxide both in the as-deposited state and after annealing in the air, with particle size varying from 2 to 15 nm. In particular, TEM images after heat treatment of the thin film with 6.3 at. % Pt in the air for 113 days show that spherical particles of ∼5 nm in size are dispersed. Thus, it is concluded that the thin films form nanocomposites in which Pt nanoparticles are dispersed in α-Fe₂O₃.

I am grateful to N. Makabe [Research Institute for Electromagnetic Materials (DENJIKEN), Sendai, Japan] for assisting in the experiments.

The author has no conflicts to disclose.

Seishi Abe: Conceptualization (lead).

The data that support the findings of this study are available from the corresponding author upon reasonable request.