Unraveling the Structure and Properties of Layered and Mixed ReO3-WO3 Thin Films Deposited by Reactive DC Magnetron Sputtering.

Tungsten trioxide (WO3) is a well-known electrochromic material with a wide band gap, while rhenium trioxide (ReO3) is a "covalent metal" with an electrical conductivity comparable to that of pure metals. Since both WO3 and ReO3 oxides have perovskite-type structures, the formation of their solid solutions (ReO3-WO3 or Re x W1-x O3) can be expected, which may be of significant academic and industrial interest. In this study, layered WO3/ReO3, ReO3/WO3, and mixed ReO3-WO3 thin films were produced by reactive DC magnetron sputtering and subsequent annealing in air at 450 °C. The structure and properties of the films were characterized by X-ray diffraction, optical spectroscopy, Hall conductivity measurements, conductive atomic force microscopy, scanning and transmission electron microscopy, energy-dispersive X-ray spectroscopy, and X-ray photoemission spectroscopy. First-principles density functional theory calculations were performed for selected compositions of Re x W1-x O3 solid solutions to model their crystallographic structure and electronic properties. The calculations predict metallic conductivity and tetragonal distortion of solid solutions in agreement with the experimental results. In contrast to previously reported methods, our approach allows us to produce the WO3-ReO3 alloy with a high Re content (>50%) at moderate temperatures and without the use of high pressures.

Introduction

Rhenium trioxide (ReO3) is a solid material with a perovskite-type cubic crystalline structure that is formed by a network of regular ReO6 octahedra sharing common vertices in three dimensions with the Re–O–Re angles of 180°.[1] Bulk ReO3 displays metallic behavior with a specific resistivity in the same range as metallic copper,[2] whereas the resistivity of polycrystalline films ReO3 is higher (ρ = 4.0 × 10–3 Ωm).[3] ReO3 can be used as a catalyst in organic synthesis, for example, for amide reduction.[4]

Tungsten trioxide (WO3) has a perovskite-type monoclinic crystal structure that can be described as a slightly distorted cubic ReO3-structure, with the corner-shared and tilted WO6 octahedra.[5] Despite the similarities in the structure, WO3 is an insulator with electrochromic properties that are successfully utilized in the production of smart windows.[6] Applications in gas sensing were also reported.[7]

Considering that both materials have a perovskite-type structure and close values of ionic radii of Re6+ and W6+ in ReO3 and WO3,[8] it should be possible to form solid solutions of mixed ReO3–WO3 (or ReW1–O3) phases with electronic properties different from those of pure WO3 and ReO3. Finding a well-controllable method for production of such materials would open the route for tweaking and fine-tuning the characteristics of the ReO3–WO3 composite.

WO3/ReO3 heterostructures and WO3–ReO3 solid solutions were studied theoretically in several works.[9,10] Ling et al. studied the reactivity of hydrogen and methanol on the (001) surfaces of WO3, ReO3, WO3/ReO3, and ReO3/WO3. Additionally, they studied the hypothetical WReO6 structures and found that (001) layered heterostructures are more stable than the mixed ones.[9] Jiang et al. studied coupling between octahedral rotations and local polar displacements in WO3/ReO3 superlattices.[10] They found that superlattices with sufficiently thick ReO3 layers, the absolute number being three or more layers and the rhenium fraction >50%, tend to be more stable than the separated material phases and also show enhanced octahedral rotations in the WO3 layers.

However, so far, only two experimental works are available on the synthesis of solid solutions of ReO3–WO3.[11,12] The paper by Sleight and Gillson[11] reports the synthesis of WO3-ReO3 solid solutions by the method that requires a high pressure (65 kbar) and high temperatures (800–1100 °C). The authors obtained ReO3–3WO3 crystals with an electrical resistivity of 3.4 × 10–5 Ωm at 25 °C, while the resistivity of pure ReO3 crystals was 3.6 × 10–7 Ωm. Interestingly, the authors also mention the formation of oxygen-deficient ReO3– crystalline phases, having the symmetry reduced from cubic to orthorhombic. The second paper by Helbig et al. describes a new organometallic approach to the synthesis of ReW1–O3 phases with up to x = 0.15.[12]

In this work, we developed a novel method to produce layered ReO3/WO3 and mixed ReO3–WO3 thin films by reactive DC magnetron sputtering and subsequent film annealing in air. In contrast to previously reported methods, our method does not require high pressures and allows high ReO3 content (more than 50%). The crystalline structure of obtained films and their optical and electrical properties were studied by different methods, and the formation of ReO3–WO3 solid solutions was proposed. Additionally, first-principles density functional theory (DFT) calculations were performed for selected compositions of ReW1–O3 solid solutions to model their structural and electronic properties. Our theoretical predictions qualitatively agree with the experimental results.

Experimental Section

Sample Preparation

Rhenium oxide and tungsten oxide thin films were prepared by reactive DC magnetron sputtering and annealed in air. The targets were a W metal disc (99.9% purity) and a Re metal disc (99.9% purity), which were 50 mm in diameter and 0.3 mm in thickness (GoodFellow). The substrates were 25 × 25 mm fused quartz glasses (SPI Supplies). Magnetron sputtering was performed in the vacuum system SAF25/50 (Sidrabe) at 50–150 W DC magnetron power in a mixed atmosphere of sputtering Ar gas (20 sccm) and reactive O2 gas (10 sccm) at a total pressure of 20 mTorr. The substrate was mounted 15 cm above the magnetrons and maintained at room temperature during the deposition process.

Eight samples were prepared in total. Sample #1 was prepared by the deposition of the amorphous WO film, followed by annealing for 1 h at 450 °C. Samples #2, #3, #4, and #5 had two layers: samples #2 and #3 have WO as the first layer, while samples #4 and #5 have ReO as the first layer. Sample #2 was prepared by the deposition of the amorphous WO film as the first layer and the amorphous ReO film as the second layer, following by annealing for 30 min at 300 °C. During annealing, all these and subsequent samples were capped by quartz glass to prevent ReO3 sublimation. Sample #3 was prepared by the deposition of the amorphous WO film as the first layer and the amorphous ReO film as the second layer, followed by annealing for 30 min at 450 °C. Sample #4 was prepared by the deposition of the amorphous ReO film as the first layer, followed by annealing for 30 min at 300 °C. After deposition of the second layer of amorphous WO, the sample was annealed for 30 min at 450 °C. Sample #5 was prepared by the deposition of the amorphous ReO film as the first layer and amorphous WO of the second layer, followed by annealing for 30 min at 450 °C. Samples #6, #7, and #8 were prepared by simultaneous deposition of WO and ReO with different Re/W ratios. The fabrication of the samples is summarized in Table .

Table 1

Description of Sample Preparation Sequence, Film Thickness, Composition, and Macroscopic Resistivity

sample no.	description of preparation sequence	thickness, nm	resistivity, Ω/sq
1	single layer: amorphous WOx thin film; annealing: 1 h at 450 °C	WO3: 120	a
2	first layer: amorphous WOx (100 nm); second layer: amorphous ReOx (100 nm); annealing: heated for 30 min at 300 °C	WO3: 125	1.2 × 103
		ReO3: 100
3	first layer: amorphous WOx (100 nm); second layer: amorphous ReOx (100 nm); annealing: heated for 30 min at 450 °C	WO3: 125	1.4 × 105
		ReO3: 100
4	first layer: amorphous ReOx (100 nm); annealing 1: heated for 30 min at 300 °C	ReO3: 135	5.3 × 104
		WO3: 135
	second layer: amorphous WOx (100 nm); annealing 2: heated for 30 min at 450 °C
5	first layer: amorphous ReOx (100 nm); second layer: amorphous WOx (100 nm); annealing: heated for 30 min at 450 °C	ReO3: 135	2.3 × 104
		WO3: 135
6	simultaneous deposition of ReOx and WOx (Re/W ratio 1:7); annealing: heated for 30 min at 450 °C	ReO3–WO3: 365	a
7	simultaneous deposition of ReOx and WOx (Re/W ratio 1:2); annealing: heated for 30 min at 450 °C	ReO3–WO3: 500	1.4 × 106
8	simultaneous deposition of ReOx and WOx (Re/W ratio 3:1); annealing: heated for 30 min at 450 °C	ReO3–WO3: 355	2 × 105

Resistivity is too high to be measured by our system.

Sample Characterization

The structure and crystallinity of the films were determined by the X-ray diffraction (XRD) technique. The XRD patterns were recorded using a benchtop Rigaku MiniFlex 600 powder diffractometer. Rietveld refinement was performed with BGMN[13] software using the Profex code.[14] The film morphology was studied with a scanning electron microscope SEM/FIB Lyra XM (Tescan). The crystalline structure of the films was imaged using a transmission electron microscope (Tecnai GF20, FEI) operated at a 200 kV accelerating voltage. The electrical parameters were measured in the van der Pauw configuration using a Hall effect system, HMS5000 (Ecopia). Topography and related electric current maps were obtained in the contact mode by an atomic force microscope (AFM) Park NX10 (Park Systems). Optical reflection and transmission spectra were measured using a spectrophotometer Cary 7000 (Agilent). X-ray photoemission spectroscopy (XPS) measurements were performed using an X-ray photoelectron spectrometer ESCALAB Xi (ThermoFisher), and XPSPEAK41 software was used for peak fitting. Raman scattering spectra were recorded in back-scattering geometry using a TriVista777 confocal micro-Raman system (Princeton Instruments) equipped with an upright Olympus microscope, diode-pumped solid-state laser Cobolt Samba 150 (λ = 532 nm, CW power 150 mW), and an Andor iDus DV420A-OE CCD camera. The laser power at the sample was in the range of 0.46–14.6 mW and controlled by two sets of neutral filters.

First-Principles DFT Calculations

First-principles DFT calculations were performed by the linear combination of atomic orbitals method using the CRYSTAL17 code[15−19] to understand how the cubic structure of pure ReO3 adopts upon substitution of rhenium atoms with tungsten ones. The calculation details are described in the Supporting Information. The computational scheme was validated using cubic ReO3 (Pm3̅m)[20,21] and several known WO3 phases.[22−24] A good agreement was found with the available experimental data[20,22−24] and previous first-principles calculations.[25−28] The calculations of mixed ReO3–WO3 solid solutions were performed using 2 × 2 × 2 supercells containing eight chemical formulas with removed symmetry and a fully relaxed structure.

Results and Discussion

X-ray Diffraction Analysis

The X-ray diffraction patterns of WO3–ReO3 thin film samples are shown in Figure a,b. Their analysis by Rietveld refinement is reported in the Supporting Information, Table S1. The XRD pattern of sample #1 (pure WO3 thin film) corresponds to the monoclinic (P21/n) WO3 phase. Sample #2 has good crystallinity and is composed of a cubic (Pm3̅m) ReO3 phase. A mixture of cubic ReO3 and monoclinic (P21/n) WO3 phases with a reduced crystallite size was found in samples #3 and #4. Samples #5, #6, #7, and #8 correspond to the WO3–ReO3 solid solutions, and their XRD patterns were refined using the tetragonal (P4/nmm) phase. The reduction of lattice symmetry down to tetragonal is responsible for the appearance of two small peaks at 2θ = 29 and 45°, which are absent in the cubic phase. The smallest crystallite size of about 17–27 nm was found in samples #5, #6, and #7. At the highest rhenium content in sample #8, the crystallite size increases to about 95 nm, and the pattern looks close to that of a cubic (Pm3̅m) ReO3 phase. However, also, in this case, the tetragonal (P4/nmm) phase gives a slightly better agreement. The lattice constants of the tetragonal solid solutions with different compositions do not deviate too much and are equal to a = 5.274–5.313 Å and c = 3.748–3.753 Å.

Figure 1

(a) Full spectra of XRD patterns of WO3 (#1), layered WO3/ReO3 (#2, #3), ReO3/WO3 (#4, #5), and mixed WO3–ReO3 (#6, #7, #8) thin film samples deposited on fused quartz substrates. (b) Zoom-in view of the same spectra in regions 23–25°.

It is interesting to note that samples #2 and #3 have an identical composition after film deposition (WO3/ReO3), with the only difference in the annealing temperature (300 °C for sample #2 and 450 °C for sample #3). Only the ReO3 peaks were visible on the XRD spectrum of sample #2 because the annealing temperature was not sufficiently high to crystallize the amorphous WO3 layer (Figure a,b). While both ReO3 and WO3 phases are visible on the XRD spectrum of sample #3. Samples #4 and #5 had a very similar composition (ReO3/WO3) but different XRD patterns (Figure a,b). In sample #4, the rhenium oxide layer was crystallized and converted into ReO3 before deposition of the upper WO3 layer; while in sample #5, the rhenium oxide layer was amorphous before deposition of the upper WO3 layer. Rietveld analysis of XRD spectra of sample #4 reveals two separate cubic ReO3 and monoclinic WO3 phases, while the spectra of sample #5 show a single tetragonal phase attributed to the ReW1–O3 solid solution (Table ). We explain the different behavior of these samples during the annealing process by higher volatility of amorphous rhenium oxide and the ability to interact with amorphous WO3 in comparison to crystalline ReO3.

Table 2

Rietveld Analysis of the XRD Patterns of WO3 and Mixed ReO3–WO3 Thin Filmsa

sample no.	phase 1	phase 2
1		100% WO3
		P21/n monoclinic
		a = 7.322 ± 0.001 Å
		b = 7.544 ± 0.002 Å
		c = 7.682 ± 0.002 Å
		d = 66 ± 3 nm
2	100% ReO3
	Pm3̅m cubic
	a = 3.7508 ± 0.0001 Å
	d = 464 ± 30 nm
3	25% ReO3	75% WO3
	Pm3̅m cubic	P21/n monoclinic
	a = 3.7506 ± 0.0002 Å	a = 7.319 ± 0.001 Å
	d = 129 ± 10 nm	b = 7.500 ± 0.002 Å
		c = 7.680 ± 0.002 Å
		d = 42 ± 4 nm
4	43% ReO3	57% WO3
	Pm3̅m cubic	P21/n monoclinic
	a = 3.7515 ± 0.0002 Å	a = 7.346 ± 0.001 Å
	d = 118 ± 6 nm	b = 7.615 ± 0.002 Å
		c = 7.666 ± 0.002 Å
		d = 42 ± 3 nm
5	100% RexW1–xO3
	P4/nmm tetragonal
	a = 5.305 ± 0.002 Å
	c = 3.753 ± 0.004 Å
	d = 20 ± 2 nm
6	100% RexW1–xO3
	P4/nmm tetragonal
	a = 5.274 ± 0.001 Å
	c = 3.794 ± 0.001 Å
	d = 27 ± 2 nm
7	100% RexW1–xO3
	P4/nmm tetragonal
	a = 5.302 ± 0.001 Å
	c = 3.748 ± 0.002 Å
	d = 17 ± 2 nm
8	100% RexW1–xO3
	P4/nmm tetragonal
	a = 5.313 ± 0.003 Å
	c = 3.756 ± 0.002 Å
	d = 95 ± 2 nm

Space group, lattice parameters (a,b,c), and the crystallite size (d) are reported.

Transmission Electron Microscopy

For investigation of the thin-film morphology in cross sections, transmission electron microscopy (TEM) lamellas were prepared using focused ion beam scanning electron microscopy (SEM-FIB) and are shown in Figure S3 (Supporting Information). TEM was used to determine the thickness of thin film layers (Table ) and to investigate the degree of crystallinity of obtained samples. At the cross section of samples #2 and #3, one can see the bottom WO3 layer and upper ReO3 layer (Figure S3b,c). For samples #4 and #5, ReO3 is the bottom layer and empty lacunas can be seen on both samples in Figure S3d,e, while the upper WO3 layer remains compact. Lacunas are formed due to sublimation and diffusion of rhenium oxide. For mixed WO3–ReO3 samples #6, #7, and #8, the film material is homogeneous (Figure S3f,g,h). The most important information about mixed samples #6, #7, and #8 is the high degree of crystallinity of these films. The crystallographic phase of these mixed ReO3–WO3 samples was identified as tetragonal P4/nmm and interpreted as ReW1–O3 solid solution (Table ). At high magnification TEM images, one can see crystallites and atomic structure of samples #6, #7, and #8 (Figure ). The biggest crystallite size among mixed ReO3–WO3 samples demonstrate sample #8, which well corresponds with Rietveld analysis (Table ).

Figure 2

High-magnification TEM images of samples #6 (a), #7 (b), and #8 (c) with a visible atomic structure.

Optical Properties

Sample #1 (pure WO3 thin film) was highly transparent similar to other works (Figure a).[29,30] All other samples containing rhenium (samples #2–8) were significantly less transparent in comparison to pure WO3. Their transparency is reversely correlated with the Re content, with a maximal transmittance window in the region around 500 nm (Figure b). Visually, all samples containing Re were green-blue in transmitted light (Supporting Information, Figure S1). Maximum transparency of Re-containing samples is close to the transparency window of pure ReO3, which is in the spectral range of 475–525 nm (Figure S2 in the Supporting Information).[3] In spite of the same composition, sample #3 is more transparent in comparison to sample #2 due to partial sublimation of rhenium oxide. The spectrum of sample #5 is shifted a bit to shorter wavelength in comparison to the spectrum of sample #4. The transmittance window for the spectra of samples #6, #7, and #8 is shifted to shorter wavelength; transmittance is reversely correlated with the Re content (Re/W ratio for sample #6, #7, #8 is 1:7, 1:2, 3:1).

Figure 3

(a) Optical transmission spectra of the thin-film samples and (b) zoom-in view of the same spectra in regions 300–750 nm. Note that all samples have different thicknesses.

Raman Spectroscopy

Raman scattering spectra of WO3–ReO3 thin film samples are reported in Figure . The appearance of the crystalline WO3 phase in the as-prepared films depends on the heat treatment process used during their preparation (Figure ). The metallic layer of cubic ReO3 located at the top of samples #1 and #2 screens the Raman signal from the bottom WO3 phase, which is crystalline in sample #1 but amorphous in sample #2. The strong Raman signal in sample #3 is due to the crystallized WO3 phase (a set of sharp bands at about 132, 270, 328, 607, 714, and 806 cm–1 are attributed to the triclinic WO3 δ-phase[31,32]), while rhenium oxide was evaporated from the sample surface upon heating at 450 °C. In samples #4 and #5, the top layer of tungsten trioxide protects the rhenium oxide sublayer from evaporation. Their Raman spectra are similar and include three main broad bands at 400–500, 500–850, and 850–1050 cm–1 additionally to a weak band from the WO3 phase at about 800 cm–1. The latter disappears in the mixed samples #6, #7, and #8, while the three broad bands remain clearly visible. Note that the intensity of the three broad bands is reduced in samples #7 and #8.

Figure 4

Raman scattering spectra of WO3–ReO3 thin-film samples excited by a laser with the wavelength of 532 nm at the power of 0.46 mW.

Since cubic ReO3 shows no or weak Raman activity,[33] the observed sharp peaks in all Raman spectra are related to the WO3 phase. However, the three broad bands at 400–500, 500–850, and about 950 cm–1 in samples #4–#8 are attributed to the WO3–ReO3 solid solutions with a tetragonal lattice, which are formed in the films according to our XRD data (Figure ). Note that the broad band at 950 cm–1 is well observed in the as-prepared samples #4, #5, and #6 but is weaker in samples #7 and #8 with a lower tungsten content. In the literature on different tungsten oxide phases, the origin of this band is usually attributed to the presence of the W=O double bonds. In crystalline tungsten trioxide hydrates, the band at about 950 cm–1 appears as a relatively sharp peak and is ascribed to the stretching mode of the terminal W=O double bond.[34] In amorphous tungsten oxide thin films, the band at about 950 cm–1 has been studied in the past by many authors and is also attributed to the W=O stretching mode.[35−38] At the same time, a weak band located at close wavenumbers was found in ground WO3 powders and WO3–ReO2 mixtures,[39] where it was attributed to reduced tungsten ions W(6– making tungsten–oxygen bonds at the oxide surface or the interface between WO3 and ReO2 phases.

We do not expect to have a significant amount (if any) of W=O bonds in our samples of WO3–ReO3 solid solutions. Note that the Raman spectrum of sample #6 is similar to that of the tungsten oxide sample studied in ref (40) and attributed to the cubic c-WO3 phase produced by the low-temperature (around 250 °C) decomposition of WO3·2H2O. However, it is possible that the cubic c-WO3 phase in ref (40) is indeed tetragonal and was misinterpreted due to the low quality of its XRD pattern. Moreover, the XRD patterns of samples produced using the same technology inref (41) were qualitatively interpreted to be due to cubic or orthorhombic tungsten oxide phases, but again not all peaks in the XRD pattern of the cubic c-WO3 phase were assigned. At the same time, the cubic c-WO3 sample synthesized at high pressure and temperature inref (42) demonstrates good quality XRD pattern refined in space group Pm3̅m but its Raman spectrum is different from that inref (40) and our spectra in Figure a. Thus, the origin of the band at 950 cm–1 in our samples is still obscure.

The correlation between the 950 cm–1 band intensity and the amount of tungsten in WO3–ReO3 solid solutions allows us to assume that the band is related to the tungsten sublattice. Note that the W–O phonon modes with large (>900 cm–1) wavenumber values exist in all pure WO3 phases but they are not Raman active.[28] Therefore, we propose that an increase of the unit cell and the presence of rhenium ions in the solid solutions may lead to the activation of new Raman modes.

X-ray Photoelectron Spectroscopy

An XPS study was performed to qualitatively determine the composition of the surface of the thin film and analyze the oxidation states of the Re and W ions. High-resolution spectra of the Re 4f and W 4f peaks were acquired for each sample (Figure ). The XPS spectrum of sample #1 shows the presence of the W signal: a doublet of 4f7/2 and 4f5/2 at the binding energies of 35.8 and 37.6 eV due to the spin–orbit splitting indicates the presence of W6+ in the film. The spectrum of sample #2 (quartz/WO/ReO annealed at 300 °C) exhibits only the Re signal (Figure b), which is present in two oxidation states: Re doublet 4f 7/2 and 4f 5/2 peak energies at binding energies of 43.5 and 46.0 eV correspond to the 6+ and 7+ oxidation states, respectively, indicating the presence of ReO3 and Re2O7 compounds on the surface of the film.[3,43,44] Sample #3 (quartz/WO/ReO annealed at 450 °C) has signals from both Re and W (Figure c): the spectrum of rhenium is similar to the previous sample. The spectrum of W indicates the presence of W6+ in the film.[45,46] The spectra of the samples #4 and #5 (quartz/c-ReO/WO and quartz/a-ReO/WO annealed at 450 °C) are almost identical (Figure d,e): a strong signal of W 4f doublet peaks characteristic for W6+ and a weak signal of Re 4f doublet peaks for Re7+ are visible. Sample #6 (mixed WO–ReO annealed at 450 °C, Re/W ratio 1:7) also demonstrates a strong signal of W 4f doublet for W6+ and a weak signal of Re 4f doublet for Re7+ and Re6+ (Figure f). The spectra of the samples #7 and #8 (mixed WO–ReO annealed at 450 °C, Re/W ratio 1:2 and 3:1, respectively) also demonstrate strong signals of Re 4f doublets for Re7+ and Re6+ and W 4f doublet for W6+ and W5+ (Figure g,h).[45,47] The presence of the Re7+ signal on the surface of samples #4 and #5 confirms that Re atoms sublime from the bottom Re layer and diffuse through the W layer to the sample surface.

Figure 5

XPS spectra for sample #1 (a), sample #2 (b), sample #3 (c), sample #4 (d), sample #5 (e), sample #6 (f), sample #7 (g), and sample #7 (h).

Electrical Properties

Macroscopic electrical resistivity of the WO3–ReO3 films was measured at room temperature in the van der Pauw configuration using a Hall measurement system. Current–voltage characteristics were linear for all samples, indicating good ohmic contacts. The resistivity of samples #1 and #6 was too high to be measured by our Hall measurement system. The obtained data are summarized in Table . Note, that samples #2, #3, #4, #5 consist of two layers (ReO3 and WO3), and conductivity is due to mainly the ReO3 layer. The minimal resistivity was demonstrated by sample #2 (WO3/ReO3 sample heated at 300 °C), where ReO3 is the upper layer, and the annealing temperature is optimal for the crystallization of ReO3. However, this temperature is not high enough for the crystallization of WO3 (see XRD spectrum in Figure ). The resistivity of sample #3 (WO3/ReO3 sample heated at 450 °C) is significantly higher in comparison to sample #2 due to partial sublimation of ReO3. Samples #4 and #5 (ReO3/WO3 heated at 450 °C) have comparable resistivity; however, sample #5 is two times more conductive than #4. Samples #6, #7, and #8 were prepared by simultaneous deposition of rhenium and tungsten oxides and annealed at 450 °C; resistivity of these samples is reversely dependent on rhenium content and is equal to 0.7 Ωm for sample #7 and 7 × 10–2 Ωm for sample #8 (resistivity of sample #6 was too high to be measured by our system). The only available data on ReO3–WO3 solid solutions resistivity was published by Sleight and Gillson,[11] who reported the electrical resistivity of ReO3–3WO3 bulk crystals equal to 3.4 × 10–5 Ωm, while the electrical resistivity of pure ReO3 bulk crystals 3.6 × 10–7 to 8 × 10–8 Ωm depending on the synthesis method.[11] In our previous work,[3] we found minimal electrical resistivity of pure ReO3 thin films prepared at 250 °C equal to 4.0 × 10–7 Ωm with the tendency to increase at a higher annealing temperature to 1.3 × 10–5 Ωm at 350 °C. In this work, we use an annealing temperature of 450° C to achieve ReO3–WO3 film crystallization, which was impossible at 250 and 350 °C.

Conductive AFM

All samples were investigated using conductive AFM to study the electrical properties at a microscopic scale. Sample #1 or pure WO3 thin film has a very high electrical resistance of 3.8 × 1011 Ω (Figure b). Sample #3 having the WO3 bottom layer and the ReO3 upper layer demonstrates a lower resistance of 4 × 106 Ω (Figure d). Samples #4 and #5 have the ReO3 bottom layer and the WO3 upper layer and show high resistance values of 1.6 × 109 Ω and 2.5 × 109 Ω, respectively (Figure f,h). The resistance of samples #4 and #5 is significantly lower (at least 2 orders of magnitude) in comparison to sample #1. We suggest rhenium oxide diffusion from the bottom of the ReO3 layer through the upper WO3 layer. XPS spectroscopy and tomography data confirm this hypothesis.

Figure 6

Conductive AFM measurements of sample #1 (a—topography, b—current map), sample #3 (c—topography, d—current map), sample #4 (e—topography, f—current map), and sample #5 (g—topography, h—current map). All samples were annealed for 0.5 h at 450 °C. Note different bias voltages and current ranges for all samples.

Mixed ReO3–WO3 samples #6, #7, and #8 show resistance higher than that of sample #3 (having an upper ReO3 layer) and lower in comparison to samples #4 and #5 (having an upper WO3 layer): 1.5 × 108, 1.1 × 107, and 1 × 107 Ω, respectively (Figure ). In other words, the resistivity of mixed ReO3-WO3 samples is in between WO3 and ReO3 film resistivity. The conductivity map of mixed ReO3–WO3 samples demonstrates good homogeneity with the inclusion of small non-conductive regions (discussed later).

Figure 7

Conductive AFM measurements of sample #6 (a—topography, b—current map), sample #7 (c—topography, d—current map), and sample #8 (e—topography, f—current map). All samples were annealed for 0.5 h at 450 °C. Note different bias voltages and current ranges for all samples.

SEM Analysis

The surface morphology was studied using SEM. Randomly oriented closely packed ReO3 crystallites of submicrometer size are visible on the surface of sample #2 (Figure a). ReO3 crystallites are not closely packed on the surface of sample #3, probably due to the partial sublimation of the ReO3 material (Figure b). The surface of samples #4 and #5 looks very close, showing a nanocrystalline surface with some arbitrarily shaped crystallites of submicrometer size.

Figure 8

SEM images of sample #2 (a), sample #3 (b), sample #4 (c), and sample #5 (d).

SEM images of mixed WO3 and ReO3 thin film samples #6, #7, and #8 are shown in Figure . The surface of sample #6 is smooth and featureless, while samples #7 and #8 demonstrate similar small rectangular-shaped crystals.

Figure 9

SEM images of sample #6 (a), sample #7 (b), and sample #8 (c).

There are three types of microcrystals visible on AFM and SEM images. First one is pure ReO3 microcrystals with an arbitrary shape, as shown in Figure a,b (can be seen on sample #2 and #3). On the surface of pure ReO3 thin films annealed at 250–300 °C, these microcrystals have a cubic shape,[3] but at a higher annealing temperature, the shape changes due to sublimation. The second type of crystals are nonconductive and can be seen in AFM images: Figures c–h and 7a–d. In SEM images, these microcrystals appear as black crystals, Figure c,d. We believe that these are HReO4 crystals, which appear at the ReO3 containing surface upon exposure to humid air.[48] HReO4 crystals are insulating and not very reactive with atmospheric gases.[48] The third type—hypothetical ReW1–O3 microcrystals—which appears on the surface of mixed samples #6, #7, and #8 is rectangular parallelepiped. All these microcrystals are shown in Figures S5.4, S5.5, and S5.6 of the Supporting Information.

The results of the first-principles calculations performed within a 2 × 2 × 2 supercell model with removed symmetry and full relaxation of the lattice constants and atomic fractional coordinates are shown in Figure for selected ReW1–O3 solid solutions, which cover the full range of compositions studied experimentally. While our model is rather simple, it predicts the behavior of the cubic ReO3 structure upon substitution of rhenium atoms with tungsten ones. For all studied compositions, the Fermi level is located in the conduction band, suggesting their metallic conductivity. The bottom of the conduction band is mainly due to the 5d(t2g)-states of rhenium and tungsten ions, and the conduction band becomes broadened for the low tungsten content.

Figure 10

Total and projected densities of states (DOSs) of the set of atomic orbitals for the four selected models of ReW1–O3 solid solutions and pure ReO3 for comparison. The energy zero is set at the Fermi level EF. The supercells used in the calculations are shown at the bottom, and lattice parameters are shown for each model.

In pure WO3, the conduction band is separated from the valence band by a gap, whose size depends on the oxide phase. Our calculations (Figure S6) suggest that the band gap increases from about 2.3 eV in cubic WO3 (Pm3̅m) to 3.7 eV in orthorhombic (Pcnb) and 4.1 eV in triclinic (P1̅) WO3 phases. These results are in good qualitative agreement with previous theoretical studies.[26−28] In solid solutions with a high tungsten content (Re0.125W0.875O3), the calculations predict a separation (a gap) of about 2.2 eV between the top of the 2p(O) states and the bottom of 5d(Re/W)-states. Upon increasing the rhenium concentration, this gap decreases down to about 0.6 eV in Re0.875W0.125O3 and is equal to 0.4 eV for pure cubic ReO3. Thus, the value of the gap between the 2p(O) and 5d(Re/W)-states in ReW1–O3 solid solutions is between that for pure WO3 and ReO3 phases. The Mulliken population analysis indicates that the charges on tungsten and rhenium ions decrease slightly from Z(Re) = +2.76 and Z(W) = +2.81 in Re0.125W0.875O3 to Z(Re) = +2.70 and Z(W) = +2.77 in Re0.875W0.125O3 as a result of electron delocalization.

The most intriguing result of our calculations is the strong evidence of the supercell tetragonal distortion occurring upon an increase of the tungsten content (Table S1), which is also found by the Rietveld refinement of the experimental diffraction patterns (Figure and Table S2). At a large rhenium concentration (Re0.875W0.125O3), the shape of the supercell is cubic, and both rhenium and tungsten ions are located within the regular octahedra. The W–O bond length is slightly longer (by 0.02 Å) than that of Re–O. However, already in Re0.75W0.25O3, the early evidence of the tetragonal distortion starts to appear as slightly non-equivalent Re–O bond lengths. In Re0.5W0.5O3 solid solution, the difference between a, b, and c lattice parameters and the axial distortion of the WO6 and ReO6 octahedra is well observed and indicates the tetragonal distortion of the lattice. At lower rhenium concentration (Re0.25W0.75O3), the lattice distortion increases further, promoting the rotation of the metal–oxygen octahedra and off-center displacements of metal ions (Figure ). The tetragonality of the lattice is well reflected in the lattice parameters with c being larger than a and b. Thus, our first-principles DFT calculations suggest that the addition of tungsten to ReO3 induces the tetragonal distortion of the lattice in agreement with the XRD results discussed above.

Conclusions

In this work, we produced layered WO3/ReO3, ReO3/WO3 and mixed ReO3–WO3 thin films by reactive DC magnetron sputtering and subsequent annealing in air. In contrast to previously reported methods, our synthesis method benefits from high flexibility in the Re-to-W ratio in the resulting material, allowing high Re content (>50%) at moderate temperatures and without the use of high pressure. The possibility to vary Re content over a wide range opens a route for tweaking and fine tuning the properties of ReO3–WO3 to meet the needs of potential applications.

Separate cubic ReO3 (Pm3̅m) and monoclinic WO3 (P21/n) phases were found by Rietveld refinement of XRD patterns in layered thin films; WO3/ReO3 (sample #3) and ReO3/WO3 (sample #4) were annealed at 450 °C. However, the tetragonal (P4/nmm) phase was revealed in layered ReO3/WO3 (sample #5) and mixed ReO3–WO3 films (samples #6, #7, #8), which possibly indicates the formation of ReW1–O3 solid solutions.

Conductive AFM measurements confirm rhenium oxide diffusion through the WO3 layer during annealing (samples #4 and #5). Macroscopic electrical measurements and conductive AFM show correlation of film conductivity with the increase of Re content in mixed ReO3–WO3 thin films.

First-principles DFT calculations were performed for selected compositions of ReW1–O3 solid solutions to predict their crystallographic phase and electronic properties. The calculations suggest that ReW1–O3 solid solutions possess electrical conductivity, which was confirmed by microscopic and macroscopic electrical measurements. Moreover, our DFT calculations predict that the addition of tungsten to ReO3 induces tetragonal distortion of the lattice in agreement with the XRD results.

The new synthesis method of crystalline layered ReO3/WO3 and mixed ReO3–WO3 thin films opens the door for further studies.