Raman Spectroscopy of V₄O₇ Films

Abstract

A thin film of vanadium oxide Magnéli phase V₄O₇ was produced using cathodic arc sputtering. X-ray diffraction, Rutherford backscattering spectrometry and Raman investigations confirmed the formation of this phase. The Raman spectrum of V₄O₇ differs considerably from the spectrum of another Magnéli oxide, V₃O₅, showing that Raman spectroscopy is an excellent tool for distinguishing between these two phases. Temperature-dependent Raman measurements revealed a significant change of the spectra near the V₄O₇ metal–insulator phase transition.

1. Introduction

V₄O₇ belongs to a series of vanadium oxides with a general formula of V_nO_2n−1 (n = 3–9) called Magnéli phases [1]. All these phases can be described as modifications of the rutile structure with periodic sheared planes resembling stacking faults into which extra planes of metal atoms are introduced [2]. All Magnéli phases, except V₇O₁₃, undergo a metal–insulator transition with a little discontinuity in lattice constants [3]. For V₄O₇, such transition is observed between 237 and 250 K [3,4]. At room temperature, V₄O₇ is metallic, it has a triclinic structure and may be described in space group A$\overline{1}$ (four vanadium and seven oxygen atoms per unit cell with parameters of a = 5.509 Å, b = 7.008 Å, c = 12.258 Å, α = 95.09°, β = 95.19°, γ = 109.21° at 298 K) [5]. Below 250 K, the resistivity increases discontinuously by a factor of 8–50, the lattice symmetry is unchanged, and there are only slight changes of lattice constants or atomic positions [4,5].

The similarity of structures of Magnéli phases may make it challenging to identify the phase precisely using X-ray diffraction, especially if the sample is nanostructured (for instance, thin film), highly textured and has large lattice stress. Recently, we showed that Raman spectroscopy could be very sensitive to tiny variations of the crystal structure by measuring the spectral features of V₃O₅ around the phase transition at ~420 K [6]. Later, our results were further developed and fully confirmed by the other group [7]. Meanwhile, Raman scattering features for V₄O₇ (as well as for the other Magnéli phases with higher n) remain an open question. The only recent report claiming the synthesis and Raman characterization of V₄O₇ demonstrate the typical spectrum of V₂O₅, probably due to sample degradation during the measurement or the presence of impurities, was compiled in [8].

From the practical point of view, V₄O₇ shows potential as a thin film material for smart windows, optical switching and liquid crystal displays [9]. V₄O₇ nanoflakes demonstrate good performance as a photocatalyst [8], V₄O₇ nanowires may be used as optical sensors [10], and V₄O₇ nanocrosses are considered a promising candidate as a cathode for lithium-ion batteries [11].

Typically, V₄O₇ is produced from a mixture of V₂O₃ and V₂O₅ powders by vacuum heating [1,3] or vapor transport using TeCl₄ [4,5,12]. Alternatively, this Magnéli phase can be obtained by solvothermal methods [8,9,10,11]. Direct sputtering of vanadium in an oxygen atmosphere can also yield V₄O₇, but such reports are scarce [13]. In our work, we used cathodic arc sputtering of vanadium target in reactive oxygen, and under certain conditions, V₄O₇ films were formed. We studied these films using Raman spectroscopy at room and low temperatures and showed that the phase transition significantly influenced the observed spectra.

2. Materials and Methods

Vanadium oxide films were deposited by a cathodic arc sputtering of a vanadium target in an oxygen atmosphere, similarly to our previous work [6]. The substrate was 10 mm × 10 mm (111)-oriented Si crystal with ~200 nm thermally grown oxide layer SiO₂. The substrate temperature during the deposition was 600 °C, and the pressure of reactive oxygen was 0.045 Pa. The arc current was 55 A. The sample was placed about 50 cm away from the cathode, at the point where growth speed was about 17 nm/min. The total deposition time was 15 min, yielding a thickness of about 250 nm.

An X-ray diffraction (XRD) study of the films in θ–2θ geometry was performed using a Bruker AXS D8 DISCOVER setup with a Cu Kα (0.15418 nm) radiation (Bruker, Karlsruhe, Germany). Temperature-dependent measurements were performed using the Anton Paar temperature control unit in the range of −120~+100 °C.

Raman scattering measurements were performed using a Horiba Jobin Yvon micro-Raman spectrometer LabRam HR800 with an ×100 magnification objective lens (numerical aperture of 0.9, Horiba, Villeneuve d’Ascq, France). Measurements were conducted at room and lower temperatures in the argon environment to avoid water condensation on the sample. Peltier element supplied by Kryotherm, Saint-Petersburg, Russia, was used to achieve temperatures of up to −75 °C. A He-Ne laser with a 632.8 nm wavelength was used to excite Raman scattering. Laser power on the sample was 0.5 mW, and a spot diameter was about 10 μm (it was intentionally defocused to avoid sample degradation and reduce overheating). The total acquisition time for each sample was at least 1 h.

Stoichiometry of produced samples was investigated by Rutherford backscattering spectrometry (RBS) using the Van de Graaff accelerator AN2500 (High Voltage Engineering Europa B.V., Amersfoort, The Netherlands) located at Immanuel Kant Baltic Federal University (Kaliningrad, Russia). A sample was exposed to a 1.5 MeV 4He⁺ beam. An angle between the beam and normal to the sample surface was 5°, an exit angle was 21°, and a scattering angle was 160°. A beam current and a total accumulated charge were about 20 nA and 65 μC, respectively. Experimental spectra were processed using SIMNRA software (version: 7.03) [14].

Morphology of the samples was investigated using the atomic force microscope (AFM) SmartSPM Aist-NT in the tapping mode (amplitude 25 nm, AIST-NT Inc., Novato, CA, USA) and the scanning electron microscope (SEM) Zeiss Cross Beam XB 540 (Carl Zeiss Microscopy GmbH, Oberkochen, Germany), which is part of a unique scientific facility «SynchrotronLike», operated at 3 kV.

3. Results and Discussion

XRD pattern for a powder V₄O₇ sample was calculated based on the crystal structure from ref. [5] using Profex software [15] (Figure 1a). Our V₄O₇ film (Figure 1b) is highly textured, and the only strong peak is located at 2θ = 39.04°. The nearest reflection for the bulk sample is ($2\overline{22}$) at 2θ = 38.86°, so the compressive stress in the film is about 0.4%. A ($2\overline{22}$) reflection may be rewritten in rutile VO₂ (R) subcell coordinates as (200)_R [5]. Such texturing is typical for VO₂ films [16,17], including samples produced by arc sputtering [18].

One of the characteristic reflections of V₄O₇ is ($10\overline{2}$) at 2θ = 21.14° (for V₅O₉ the equivalent reflection is ($10\overline{2}$) at 2θ = 22.36° [19] and for V₃O₅ the equivalent reflection is (200) at 2θ = 19.13° [20]). For a sample with a texture of ($2\overline{22}$), this reflection can be observed in θ–2θ scans if a sample tilt of ψ = 43° is applied (this angle is calculated according to ref. [21]). In Figure 1b we indeed see such a reflection at 2θ = 21.13°, confirming the formation of V₄O₇. The experimental configuration at ψ = 43° supports the observation of several other expected reflections: ($1\overline{2}0$) at 2θ = 26.81° (expected at 2θ = 26.76°, ψ = 32°), ($10\overline{4}$) at 2θ = 32.22° (expected at 2θ = 32.3°, ψ = 49°), ($30\overline{2}$) at 2θ = 53.6° (expected at 2θ = 53.45°, ψ = 45°), ($20\overline{6}$) or ($1\overline{26}$) at 2θ = 54.2° (expected at 2θ = 53.96°, ψ = 45° or 2θ = 54.16°, ψ = 43°, respectively), ($1\overline{42}$) at 2θ = 56.25° (expected at 2θ = 56.13°, ψ = 43°) and ($2\overline{4}2$) at 2θ = 56.7° (expected at 2θ = 56.63°, ψ = 43°). Since we observe all the expected peaks (with relatively high intensities) and only the expected reflections, we can conclude that our film is indeed V₄O₇ and contains no other phases.

Temperature dependence of the position of ($2\overline{22}$) reflection is shown in Figure 1c. We can see the abnormal expansion of the lattice during the sample cooling from 0 to −65 °C. Such behavior is associated with the phase transition in V₄O₇. However, the discontinuity on the graph, which could help to locate the exact transition temperature, can hardly be seen even for single crystals [5]. For thin films, the phase transition is usually not abrupt, which is related to the complex internal strain dynamics in the film [7,22]. Thus, we can only conclude that the phase transition temperature in our sample is above −65 °C.

The formation of V₄O₇ is further confirmed by elemental analysis performed by RBS (Figure 2). An experimental spectrum was fitted, suggesting structures with different values of x in the VO_x layer. The number of incident particles was adjusted to minimize the quadratic deviation for the signal from vanadium (channels 245–280). Then, the value of the reduced quadratic deviation ${\chi }_R^2$ was analyzed for the oxygen region (channels 75–115). The film composition of V₄O₇ gives the best agreement between experiment and simulation (${\chi }_R^2$ = 1.4). The accuracy of the method rules out the formation of V₃O₅ or Magnéli phases with n ≥ 6 (${\chi }_R^2$ = 8.6 and 5.3 for V₃O₅ and V₆O₁₁, respectively). However, it would be challenging to distinguish between V₄O₇ and V₅O₉, since V₅O₉ also gives an acceptable agreement between experiment and simulation (${\chi }_R^2$ = 2.3).

Morphology of V₄O₇ film was investigated using SEM and AFM (Figure 3). From the Scherrer equation (FWHM of ($2\overline{22}$) peak is 0.15°), we can estimate the average crystallite size as 60 nm. This value is consistent with visible grain sizes (Figure 3a,b,d). Additionally, using a watershed algorithm, we determined that the distribution of equivalent disk diameter for grains has a maximum of 60 nm and a mean value of 100 nm. From the cross-section image (Figure 3c), we can estimate the film thickness as 240 nm, which is in good agreement with the value obtained by RBS.

Raman spectrum of V₄O₇ film is shown in Figure 4. The spectrum is similar to one observed in our previous work (VOx, Figure 12 in ref. [6]). However, this VO_x phase was not properly identified, and the sample contained impurities. V₄O₇ undergoes a phase transition between −36 and −23 °C, so two different spectra are presented: above (20 °C, Figure 4b) and below (−75 °C, Figure 4d) the phase transition temperature. Cooling the V₄O₇ sample leads to a blueshift of all Raman bands. Moreover, low-temperature modification has four additional peaks at 166, 556, 596 and 714 cm⁻¹. Such spectral changes can be compared with the behavior of another Magnéli phase, V₃O₅ (Figure 4a,c). Similarly to V₄O₇, heating of V₃O₅ above the phase transition leads to the disappearance of strong peaks at higher wavenumbers: 564, 591 and 740 cm⁻¹.

Temperature-dependent Raman measurements are summarized in Figure 5. Only slight blueshifts of the bands are observed up to −50 °C. At −55 °C, we can see the appearance of new peaks at 165, 553, 593 and 709 cm⁻¹. Further cooling does not significantly influence the spectrum. The spectral changes are fully reversible if the sample is heated to room temperature after the cooling. From our measurements, we can conclude that the sample undergoes a phase transition at a temperature between −50 and −55 °C. This temperature is slightly lower than the phase transition temperature of bulk V₄O₇ (the reported values are −36 [4] or −23 °C [5]). The discrepancy can be explained by the effect of external surface and grain-boundary surface energies [23]. This effect shifts the phase transition temperature by up to several tens of degrees to lower or higher values and is typical for thin films. Note that the phase transition in other Magnéli phases occurs at much lower temperatures (the highest one is about −100 °C for V₆O₁₁ [3]).

Temperature dependence of the V₄O₇ film electrical conductivity is given in Figure 6. Similar to XRD, we can see a broad transition region between −25 and −55 °C. The absolute value of conductivity at room temperature agrees well with the value previously reported for the bulk crystal [4]. For the low-temperature region, our curve is shifted by 20–25 K. Such shift is consistent with our observations derived from Raman spectroscopy.

4. Conclusions

Magnéli phases of vanadium oxides were directly produced by a sputtering technique. During the cathodic arc sputtering of the vanadium target, V₄O₇ grows by 17 nm/min at 600 °C and 0.045 Pa of oxygen pressure. Produced V₄O₇ films have a preferred orientation of ($2\overline{22}$) corresponding to (200) in rutile coordinates. The stoichiometry of V₄O₇ was confirmed by Rutherford backscatter spectrometry. The room temperature Raman spectrum of this phase is relatively weak, and its measurement requires careful tuning of the laser power on the sample. The spectrum consists of one strong narrow peak at 203 cm⁻¹ and several weaker and broader bands at 176, 247, 303, 357, 414, 431 and 495 cm⁻¹. Metal–insulator phase transition in V₄O₇ was observed by Raman spectroscopy between −50 and −55 °C. The low-temperature spectrum is characterized by four additional bands at 166, 556, 596 and 714 cm⁻¹ (at −75 °C).