Reactive Sputtering Process Study for Vanadium Oxynitride Films

Abstract

In this study, vanadium oxynitride thin films were deposited by reactive magnetron sputtering using pure vanadium targets, Ar as a plasma carrier, and a mix of N₂ and O₂ as reactive gases. Various ratios of mass flow rates between two reactive gases were maintained as a constant during the process. To obtain crystalline phases of oxynitrides, rapid thermal annealing in Ar atmosphere at 600 °C and 700 °C for 5 min was conducted after the deposition. This study aims to define the range of the process parameters of magnetron sputtering to deposit vanadium oxynitride thin films. The assessment for the characterization of films utilizes the surface profiler, scanning electron microscope, X-ray diffraction, X-ray photoelectron spectroscopy, four-point probe, Hall analyzer, and UV-visible-NIR spectrometer. Experimental results reveal that the annealed films can be oxynitrides when the oxygen flow rate is below 0.25 sccm, and the ratio of oxygen/nitrogen is no more than ~1/10. The annealed vanadium oxynitride films, in terms of their properties, are closer to vanadium nitrides than to oxides, due to the intended low supply of oxygen during deposition. For instance, the film is more metallic than semi-conductive with dark appurtenance and high optical absorbance across the spectrum between 200 and 900 nm. For practical purposes, the deposition conditions of O₂:N₂ = 1/20, O₂ < 0.25 sccm, and 600 °C annealing are recommended to obtain vanadium oxynitride films with relatively lower resistivity (10⁻² Ω cm) and optical transmittance (<15%) through films.

1. Introduction

Vanadium oxynitride (VN_xO_y) is a metallic compound between vanadium oxides (VO, cubic, V₂O₃, trigonal or V₂O₅, orthorhombic) and vanadium nitride (VN, cubic). The stoichiometric composition of vanadium oxynitride is difficult to achieve in manufacturing processes. This is because the direct replacement of nitrogen by oxygen, or vice versa, is conspicuously hampered by the differences in valences and chemical activities between the two elements. Overall, vanadium oxide is nearly an insulator, while vanadium nitride is semi-conductive, or even closer to metallic compounds. The compound vanadium oxynitride, in its composition between nitrides and oxides, exhibits some interesting properties and functions that may be useful in various optoelectronic applications.

Some known applications of vanadium oxynitride are electro- and photo-catalysts [1], electrochemical capacitors [2] and supercapacitors [3] (both symmetric and asymmetric [4,5]), photovoltaics [6], and solar control coatings [7].

Vanadium is distinguished from other transition metals by its amphoteric oxidation states (−3, −1, 0, +1, +2, +3, +4, +5), due to which it can form a wide range of compounds with specific properties. The study of vanadium oxynitrides thin films thus originates from this unique feature. Due to its non-stoichiometric combination of two distinct crystal structures, its optoelectronic properties are uncertain, even though vanadium oxynitrides should be bonded by oxides and nitrides in principle. Our goal in this study is to explore how different oxygen supplies during deposition can affect the composition and crystal structures of films, and how electrical resistivity and optical transparency are affected. Here, we particularly mention one point: the change in optoelectronic properties in response to different oxygen contents is not well understood [7]. Qualitative answers, such as the semi-conductive-to-insulated transition, can be found in the literature for either vanadium or similar transition metals, although quantitative descriptions, particularly for thin films by sputtering, are usually either difficult to find or overlooked [8,9].

Fabricating oxynitride films involves two opposite approaches. The first approach is to nitrify oxides at elevated temperatures (usually > 800 °C) under a high nitrogen supply [10,11,12]. The reason for such requirements in fabrication is obvious because of the challenge to substitute chemically more active oxygen in oxides with less active nitrogen. In general, metal oxides are chemically more stable than nitrides in their natural states. In contrast, the oxidation of metal nitrides may be easier to achieve, due to the fact that oxygen is more likely to bond with metals than nitrogen [13,14,15,16]. Other advantages of oxidation include a much lower temperature, much fewer supplies of oxygen, and less demanding process controls. Yet, there are also some hidden negatives to oxidation processes. Of the first, partial oxidation is commonly found, as the process temperature for oxidation is usually low. In other words, only part of the nitrides is replaced by oxides. However, this circumstance perfectly fulfills our objective of fabricating oxynitride films. Another drawback is that oxidation is more contagious (spreadable), so delicate metallic parts need to be properly masked to avoid unwanted oxidation. However, this problem can be undertaken with carefully designed masks. Furthermore, it was shown that oxygen dissolution in the lattices of Group 5B metal nitrides is accompanied by the extrication of nitrogen [14,17,18]. This phenomenon is beneficial for the synthesis of vanadium oxynitride, except for forming pure high-oxide phases [19,20].

Magnetron sputtering is considered an effective method to deposit high-quality functional films, due to its simple and cost-effective process control. Today, magnetron sputtering has been employed to deposit VO₂ [21], VO_x [22], doped-V_xO_y [23], VN [24], vanadium oxynitrides [25], and other films for many industrial applications. Despite several limitations, magnetron sputtering is undeniably capable of producing uniform and compact films. Moreover, the thickness of films can be controlled simply by changing a few process parameters, such as power or gas supply. In addition, magnetron sputtering has excellent repeatability and reproducibility suitable for industrial-scale production. Regarding the synthesis of vanadium oxynitride thin films, magnetron sputtering has not been fully utilized and understood, due to the complexity of its reactive plasma. For example, there is still a lack of data on how sputtering parameters, such as reactive gas flow rates, and the post-annealing temperature, correlate to the trilogy, structure-property-function of vanadium oxynitrides films.

2. Experiment

The vanadium oxynitride thin films are deposited by magnetron sputtering (CS-400, Junsun Tech. Co., New Taipei City, Taiwan) on quartz glass (20 mm × 20 mm × 0.5 mm, for optical measurements) or Si wafer (ϕ = 4”, p-type, 20 mm × 20 mm × 0.5 mm, for material characterizations). For reference purposes, the experimental plan for material characterizations and process parameters of sputtering and rapid thermal annealing are listed in

Table 1. Experimental plan for material characterizations.

Chamber Plasma	Optical Emission Spectrometer (OES)
Crystal Structure	XRD
Surface Morphology	Scanning electron microscope (SEM), Surface profiler
Chemical Composition	X-ray photoelectron spectroscopy (XPS)
Electrical Analysis	Four-point probe, Hall effect analyzer
Optical Analysis	UV-Visible-NIR spectrometer

and

Table 2. Process parameters for sputtering and rapid thermal annealing.

Sputtering
Target power (W)	DC 190
Target composition	V (99.9 wt. %)
Background pressure (Pa)	6.4 × 10−4
Working pressure (Pa)	~0.733–~0.933
Substrate temperature (°C)	Room temperature
Target-substrate distance (cm)	10
Substrate bias (V)	0
Key selected O2/N2 flow rate (sccm)	0.15/3, 0.3/3, 0.25/5, 0.5/5, 0.5/10, 1/10
Ar flow rate (sccm)	20
Annealing
Temperature (°C)	600, 700
Time (min)	5
Ar flow rate (sccm)	2000

, respectively.

2.1. Substrate Preparation

A 4” silicon wafer (p-type, <100>, Ultimate Materials Technology Co., Ltd. Hsin Chu, Taiwan) was cut into 20 mm × 20 mm pieces using a diamond scribe. The cut substrate was etched in buffered oxide etchant (40% NH₄F in water and 49% of HF in water mixed in a 6:1 volumetric ratio) for 20~30 s to remove silicon dioxide (SiO₂) or silicon nitride (Si₃N₄). After etching, the substrate was ultrasonically cleaned for 10 min in various solutions, following the sequence of (1) de-ionized (DI) water, (2) ethanol (95%), (3) DI water, and (4) acetone (99.9%, Echo Chemical Co., Taipei, Taiwan). After cleaning, the substrate was blown to dry in the 99.95% nitrogen air (Cinphong Gas Industrial Co., New Taipei City, Taiwan).

For material characterizations, quartz glass (99.99%, Ultimate Materials Technology Co., Ltd. Hsin Chu, Taiwan) was cut into 20 mm × 20 mm pieces using a diamond scribe. The cut glass was ultrasonically cleaned by KOH (1 g/100 mL in water, Echo Chemical Co., Taipei, Taiwan), at first, and then followed the same sequence of the above-mentioned solutions for 10 min, respectively. Lastly, the substrate was blown to dry in 99.95% nitrogen air.

2.2. Deposition and Rapid Thermal Annealing

Vanadium targets (99.99 wt.% Ultimate Materials Technology Co., Ltd. Hsin Chu, Taiwan) are used for the deposition of films following the process parameters listed in Table 2. Prior to all deposition processes, the physical vapor deposition (PVD) chamber was vacuumed to 6.4 × 10⁻⁴ Pa under room temperature. The working gases are 20 sccm Ar (99.999%, Cinphong Gas Industrial Co., New Taipei City, Taiwan) as the plasma carrier, 3–10 sccm N₂ (99.999%), and 0–1 sccm O₂ (99.999%, Cinphong Gas Industrial Co., New Taipei City, Taiwan) as reactive gases. All gases were purchased from local suppliers. The ratio between nitrogen and oxygen flow rates is chosen to be O₂/N₂ = 0, 1/20, and 1/10 to deposit films of various compositions and microstructures. During deposition, a direct current (DC) power of 190 W was supplied to the vanadium target. No substrate bias is applied during deposition. The temperature in the chamber during the deposition is controlled at around 21–24 °C.

To clean up volatile contaminants on the target, we always pumped Ar into the chamber at first, and maintained it for 10 min before the deposition. All other gases were supplied later, depending on our intention.

To transform the as-deposited amorphous films into crystalline structures, a rapid thermal annealing process is conducted in a chamber under the Ar atmosphere. The annealing temperatures are deliberately selected at 600 and 700 °C for 5 min after several rounds of trials before this study to make sure successful annealing of films.

2.3. Plasma and Material Characterization

2.3.1. Chamber Plasma

For the detection of excited species in the chamber plasma, an optical emission spectrometer (OES, Emicon HR system, Plasus, German) was utilized for the observation of optical emission from plasma. The optical probe of the spectrometer was attached to the chamber window, which is made of quartz, to minimize its optical absorption. The range of spectrum for scanning was set to the wavelength of 195–950 nm. The spectrometer can automatically repeat the recording and take an average for each round of measurement.

2.3.2. Thickness and Surface Morphology of Films

The thicknesses of deposited films were evaluated by a surface profiler (Surfcorder ET200, Kosaka, Japan). To make a step profile of deposited films for the thickness measurements, half of the substrate is masked by heat-resistant tapes before deposition. The tape was removed to create a step profile after deposition. The measurement was taken along five randomly selected lines on the sample surface across the edge of the step. The average thickness of films is counted by these measurements. The average deposition rate can be calculated as the quotient of the average thickness to the deposition time.

Note that the deposition rate is measured each time before the deposition to warrant an accurate control of the film’s thickness.

The surface morphology of films was examined by scanning electron microscopy (SEM, S-3400N, Hitachi, Tokyo, Japan) with accelerating voltage and chosen magnification of 15 kV and 3000, respectively.

2.3.3. Crystal Structure

The crystal structures of deposited films were examined by an X-ray diffractometer (XRD, PANalytical XPert PRO MPD) in the thin film mode, using monochromatic high-intensity Cu Kα radiation (λ = 1.5425 Å). The scanning angle was from 10°–2θ to 90°–2θ with a step size of 0.02°, and the measuring time is 0.5 s per step with an incident angle of 0.5°.

We can estimate the crystal size (D) using Scherrer’s formula:

$$D=\frac{k\lambda }{\beta cos\theta }$$

(1)

where θ is the scattering angle of the crystal plane, β is the full width at half-maximum (FWHM) of a peak, k is the shape factor (0.9), and λ is the wavelength of X-rays (1.5425 Å). Numerical calculations for the crystal sizes were conducted, mainly using Jade^® 6 (Materials. Data Inc., Livermore, CA, USA).

2.3.4. X-ray Photoelectron Spectroscopy

The chemical binding energy among vanadium, nitrogen, and oxygen in the deposited films is investigated by X-ray photoelectron spectroscopy (XPS, PHI 5000 VersaProbe III, ULVAC-PHI. Inc. Japan), using monochromatic Al K_α X-rays of power 25.1 W and energy resolution (FWHM) ≤ 0.5 eV at Ag 3d^5/2 peak. The passing energy of the He ultraviolet light source is set to 224 eV, and the incident angle is set at 45.0°. The measurement is performed under a vacuum of 6.7 × 10⁻⁸ Pa. The area for analysis is circular of radius 100 μm.

Numerical fitting for the analysis of XPS spectra was carried out using XPSPEAK 4.1. Our target elements are vanadium, oxygen, and nitrogen.

2.3.5. Electrical Resistivity

The electrical resistivity of the annealed films was measured by a four-point probe (RT-70/RG-7, Naspon, Tokyo, Japan) and a Hall-effect analyzer (AHM-800B, Advanced Design Technology, Hsinchu, Taiwan).

For the four-point probe measurement, the resistivity is calculated following Ohm’s law, using the voltage difference measured between two selected points out of four on the center of the film’s surface. For better accuracy, the resistivity of each sample was repetitively measured 5 times by rotating the probe randomly.

Other relevant information, such as the carrier density, mobility, and Hall coefficient (the ratio of the induced electric field to the product of the current density and the applied magnetic field), was further measured by the Hall-effect analyzer.

To validate measurements, we cross-examined and compared measurements from the four-point probe and the Hall-effect analyzer to avoid erroneous experimental results.

2.3.6. Optical Measurement for Films

Optical transmittance and reflectance of the films were measured by a UV-VIS-NIR double-beam spectrophotometer (JASCO-V670, USA) with wavelengths ranging from 190 to 900 nm. The optical absorption is calculated by subtracting the transmittance (T_film) and reflectance (R_film) of film from 100%, i.e.:

A_film = 100% − T_film − R_film

(2)

Each sample was respectively measured 5 times to have a more accountable average optical absorbance.

Optical band gaps of the deposited films can be estimated from the spectra of the UV-Vis-NIR spectrometer. The calculation procedure combines the Beer-Lambert law and the Tauc formula [26,27]. The Beer-Lambert law presumes that optical transmittance (T) decays exponentially through the thickness of the film through which the light travels, i.e.:

T = exp(−αd)

(3)

where d and α are, respectively, the thickness and optical absorption coefficient of a film. By assuming a form of a power law, the Tauc formula gives the electron transition across parabolic energy bands as:

(αE)^x = B(E − E_g)

(4)

Where E = hc/λ is the incident photon energy, E_g is the Tauc bandgap, and B is the microstructure-related constant called the Tauc parameter, which represents the slope of the linear part of the curve in the plot (αE)^x against E [26,27,28]. Different numerical values of the exponent x characterize different schemes of electron transition at the band edge by optical absorption. The most common values for x are ½ for the indirect transitions in many amorphous structures; and 2 for the direct bandgap transitions with conserved momentum during the transition [29,30,31,32,33,34,35].

3. Results

3.1. Deposition Rate

Among all deposition parameters, we present the set of the nitrogen flow rate of 10 sccm herein to discuss the influence of different oxygen supplies on the variations in average deposition rate (nm/min). The choice of this set for our illustration is because a higher nitrogen flow permits a wider range of oxygen flow to form a whole gamut of nitrides, oxynitrides, and oxides.

The average deposition rates of films deposited under different oxygen flow rates are shown in Figure 1, where the average deposition rate initially increases from 9.39 nm/min under 0 sccm oxygen to a maximum of 9.98 nm/min under 0.35 sccm oxygen flow. Then the average deposition rate gradually declines as the oxygen supply increases. The average deposition rate decreases to 5.2 nm/min under a 3 sccm oxygen supply.

The initial rise of the deposition rate between 0 and 0.35 sccm O₂ supply can be attributed to the formation of oxynitrides, because the density and atomic packing factors of oxynitride are lower than those of nitrides (cubic VN: 6.13 g/cm³ and trigonal V₂O₃: 4.87 g/cm³). In other words, oxynitrides films would be thicker than nitrides if the same quantity of nitrides were deposited.

Usually, introducing oxygen into a chamber would lead to the increase of oxides and thus reduce the electrical conductivity of plasma or elevate the target voltage. However, for a small amount of oxygen being introduced, 0–0.5 sccm of O₂ in our case, the corresponding target voltage remains almost plateau. This can be caused by a higher density of secondary electrons generated in plasma, as the valance electrons of oxygen are easier to be ionized. These ionized free electrons can generate even more free electrons by the collisions among oxygen, nitrogen, and argon.

The subsequent decrease in deposition rate from 0.5 sccm of O₂ onwards is simultaneously influenced by two factors. One is the formation of large amounts of oxynitrides or oxides, which demote the electrical conductivity of chamber plasma and hamper the deposition. Another is the poisoning of vanadium targets by the high supply of oxygen, particularly for cases of O₂ flow rate exceeding 1 sccm. These two factors combined require higher power to sputter the vanadium, manifested by the increasing voltage at the target. Figure 1 shows the monotonical increase in the target voltage from 424 to 492 V when the O₂ flow rate varies from 0.5 to 3 sccm.

Films deposited under other nitrogen flow rates (3 and 5 sccm) present a similar pattern of average deposition rates to those shown in Figure 1. We omit these data for the brevity of the context.

For all subsequent studies, we control the thickness of deposited films around 200 nm to ensure the formation of films rather than islands for better characterization and measurements.

3.2. OES for Chamber Plasma

Optical spectra of the excited atoms/molecules in plasma under different O₂ flow rates are shown in Figure 2, where major peaks from the emission of N, O, Ar, and H are numerically found and listed in Appendix A,

Table A1. Major OES peaks for N, O, Ar, and H.

Specie	Wavelength (Nm)	Molecules/Atoms	Reference
N2	337.078	N2 (C³Πu—B³Πg)	[36,37,38,39,40,41,42,43,44,50]
	357.899	N2	[36,37,38,39,40,41,42,43,44,50]
	391.134	N2+ (B²Σu+—B²Σg+)	[36,37,38,39,40,41,42,43,44,50]
	394.097	N2 (B³Πu+—B³Πg+)	[36,37,38,39,40,41,42,43,44,50]
	427.817	N2+	[36,37,38,39,40,41,42,43,44,50]
O2	458.7	O+: 3p4D → 3s4P	[36,41,42,43,44,45,50]
	525.3	O2+: b4Σg− → a4Σu	[36,41,42,43,44,45,50]
	560.7	O2+: b4Σg− → a4Σu	[36,41,42,43,44,45,50]
	615.9	O: 4d5D → 3p5P	[36,41,42,43,44,45,50]
	678.9	O+	[36,41,42,43,44,45,50]
	777.4	O: 3p5P → 3s5S	[36,41,42,43,44,45,50]
	844.6	O: 3p3P → 3s3S	[36,41,42,43,44,45,50]
Ar	696.5	Ar (4p [1/2] → 4s [3/2])	[46,47,48,49,50]
	738.4	Ar (2p3 → 1s4)	[46,47,48,49,50]
	750.6	Ar (4p [1/2] → 4s [3/2])	[46,47,48,49,50]
	763.5	Ar (4p [3/2] → 4s [3/2])	[46,47,48,49,50]
	772.4	Ar (4p [1/2] → 4s [1/2])	[46,47,48,49,50]
	794.82	Ar (2p3 → 1s5)	[46,47,48,49,50]
	801.5	Ar (1s5–2p8)	[46,47,48,49,50]
	811.1	Ar (2p9 → 1s5)	[46,47,48,49,50]
	826.45	Ar (2p2 → 1s2)	[46,47,48,49,50]
	852.14	Ar (2p4 → 1s2)	[46,47,48,49,50]
	912.3	Ar (4p [1/2] → 4s [3/2])	[46,47,48,49,50]
H2	434.312	Hγ	[50]
	486.132	Hβ	[50]
	656.126	Hα	[50]

with their corresponding transition states. Major oxygen peaks in the current study during sputtering are 458.7, 525.3, 560.7, 615.9, 678.9, 777.4, and 844.6 nm [36,37,38,39,40,41,42,43,44,45,46,47,48,49,50]. Some oxygen peaks are also marked in Figure 2 to illustrate the differences in emission among various oxygen supplies during sputtering. These spectra provide us a confirmation of the presence of oxygen in the plasma, although the intensities are relatively low. This is because the supply of O₂ is low compared to the flow rates of nitrogen and argon, which, nevertheless, is our original intention: to produce oxynitrides without complete oxidation of nitrides.

3.3. XRD

We present the XRD pattern for annealed films deposited under different nitrogen and oxygen flow rate levels and then compare different cases. For readers’ reference, we also schematically present three primary crystal structures of V₂O₃, VO₂, and VN in Appendix B, Figure A1.

Figure 3 demonstrates the XRD pattern of films annealed at 600 and 700 °C under a nitrogen supply of 3 (Figure 3a), 5 (Figure 3b), and 10 (Figure 3c) sccm, respectively. The fitted data is carefully examined according to various PDF cards, for example, [PDF#01-073-0528 for cubic VN, PDF#74-1216 for cubic VN, PDF#32-1413 for hexagonal VN, PDF#96-230-0547 for tetragonal V₃₂-N₂₆]; [PDF#37-1178 for VN_0.52O_0.26, PDF#00-037-1179 for cubic VN_0.95O_0.07, PDF#89-2532 for cubic V₂NO]; [PDF#39-0774 for monoclinic V₂O₃, and PDF#84-0316 for hexagonal V₂O₃].

It is challenging to accurately determine the crystalline phase of vanadium oxynitride due to the simultaneous presence of oxynitrides, nitrides, and oxides. As well, the oxynitrides are most likely entangled with both nitrides and oxides. However, there are several distinctive peaks for oxynitrides from the XRD pattern: 2θ = ~36.5°–36.1° (3,1,1), ~38.1°–38.2° (2,2,2), and ~64.17°–64.85° (4,4,0) [51,52,53,54,55,56,57,58,59,60,61,62].

For each nitrogen flow rate, there exists an optimal oxygen flow to produce vanadium oxynitride films. One optimal flow rate is found to be about ~1/20 of the nitrogen flow rate in the current study. For example, if N₂ = 3 sccm, then O₂ = 0.15 sccm can yield vanadium oxynitrides, almost for sure.

3.4. XPS

To further examine the elemental composition across the surface, we selected 3 sample films subjected to XPS to investigate the binding energy among elements. These samples are films annealed at 600 °C and deposited under O₂/N₂ = 0.15/3 (Figure 4a), 0.25/5 (Figure 4b), and 0.5/10 (Figure 4c), respectively.

Figure 4 shows the full spectra of three samples, where the differences between annealed and as-deposited films can be seen from the intensities of N1s, O1s, V2s, V2p_3/2, and V2p_1/2. The annealed films all have higher intensities of O1s and V2s, whereas as-deposited films all exhibit higher intensities in N1s. V2p_3/2, and V2p_1/2, if the oxygen flow rate exceeds 0.25 sccm. This implies an obvious switch (or replacement) of chemical bonds from vanadium-nitrogen to vanadium-oxygen by annealing under sufficient oxygen flow (O₂ > 0.25 sccm).

The splits of N1s, O1s, V2p_3/2, and V2p_1/2 by numerical fitting for annealed films are shown in smaller insets. For V2p_3/2 and V2p_1/2, the splits are V(0) (~512 eV), V(1, 2) (~513 eV), V(3)-OH (single) (~514.5 eV), and V(3) (~515.3 eV). For O1s, the splits are O-V-N (~528 and ~528.8 eV), and O-V (~530.5 eV). For N1s, the splits are O-V-N (~394, ~395, and ~396 eV) [63,64,65,66,67,68,69,70,71,72,73,74,75,76,77]. These splits strongly suggest that the lower oxide states of vanadium (1, 2, and 3), bonded with oxygen and nitrogen, are commonly present in these films. So, we may lend this result to ascertain the formation of vanadium oxynitrides in annealed films at 600 °C by a low oxygen supply of O₂/N₂ = 0.15/3, 0.25/5, and 0.5/10 sccm.

3.5. SEM

The selected SEM images of annealed films at 600 and 700 °C are shown in Figure 5 for cases of deposition under O₂/N₂ = 0/3. 0.15/3, 0/10, and 0.5/10. The annealing has an obvious impact on the increase of surface roughness, which should be caused by the precipitation of crystalline phases. However, the change of surface morphology is not significantly different for films deposited under a low oxygen supply (≤ 0.15 sccm), whether annealed at 600 or 700 °C. In contrast, the 700 °C annealing, along with a higher oxygen flow rate of 0.5 sccm, can provide sufficient thermal energy to make noticeable precipitation of crystalline phases, as seen from the presence of numerous small aggregates on the surface.

The selected SEM images provide pieces of evidence to show that a low oxygen supply with an appropriate O₂/N₂ ratio can interrupt the formation of large crystals, even under 700 °C annealing. Also, from XRD, we recognize that these aggregates on the film’s surface are very likely to be vanadium oxides.

3.6. Electrical Resistivity

The electrical resistivities of as-deposited and annealed films at 600 °C are shown in Figure 6 for various O₂ flow rates. The measurement of resistivities from the four-point probe and Hall analyzer are quite consistent in their trends. It is common among all cases that a higher oxygen supply increases the electrical resistivity caused by the formation of less conductive vanadium oxides. This is also truthful following the lower carrier density. The carrier density is an indicator of the available free-charge carriers in films. More free-charge carriers would imply a lower electrical resistivity in semi-conductive materials, although this correlation is not necessarily linear.

Another finding is that annealing helps to reduce electrical resistivity. This reduction is particularly noteworthy for cases of O₂/N₂ = 0.25/5 and 0.15/3. We conjecture that such a reduction may be caused by some defects (e.g., vacancies) in crystalline phases of oxynitrides, due to the partial oxidation of vanadium nitrides. Around these defects, more free electrons could be available for electrical conduction. Nevertheless, the density of defects should be sufficiently low to avoid hampering electrons’ motion.

Note that pure transition metal nitrides are usually semi-conductive [1,2,3,4,5,6,7,8,9,10], while pure oxides of these metals are almost insulated. Therefore, the electrical resistivity of vanadium nitrides is close to metals (low resistivity), though it is in fact semi-conductive with narrow band gaps. Therefore, the reduction of nitrogen supply during deposition would make films more metal than nitride and thus lower the electrical resistivity of films. On the contrary, either annealing at sufficiently high temperatures or increasing oxygen supply during deposition would help to produce more oxides and hence increase the electrical resistivity.

The electrical resistivity of films in this study is higher than the pure vanadium nitride (10⁻²–10⁻⁴ Ω cm), somewhat in the range of pure V₂O₃ (rhombohedral, 10⁻⁴–10⁻² Ω cm), and lower than pure VO₂ (monoclinic, 10⁻²–10² Ω cm) [78,79,80,81,82,83,84,85,86,87]. This result may reflect the fact that V₂O₃ could be the major oxide within vanadium nitrides.

3.7. UV-Visbile-NIR Spectrum

The optical measurements of selected films (O₂/N₂ = 0.15/3, 0.25/5, and O₂/N₂ = 0.5/10), whether as-deposited or annealed at 600 °C, are shown in Figure 7, where averages of optical transmittance and absorbance (calculated using Equation (2)) are plotted against the wavelength of the incident light in the range of 200–900 nm. We omit the reflectance for the brevity and clarity of the presentation. Both the annealing and oxygen supply assist to form more oxides or oxynitrides and thus make the films more transparent. However, since the oxygen supply is much less than nitrogen, the major constituents of films are still vanadium nitrides, with the intrinsic appearance of opaque and dark brown color. Overall, the optical transmittance is less than 20%, while the absorbance is above 65% in all samples, even after 600 °C annealing. This optical character further illuminates previous results found in XRD and XPS, i.e., lower oxides formed in nitride-rich microstructures. Thus, the optical functions of films should be closer to that of vanadium nitrides rather than oxides.

4. Discussion

4.1. Combined Conditions of Oxygen Supply and Annealing for Depositing Oxynitrides Films

XRD reveals some information about the annealing process. First, the annealing temperature at 600 °C can assist in the formation of oxynitrides, regardless of the nitrogen/oxygen ratio. However, the higher the nitrogen flow rate, i.e., 10 sccm, the more difficult to form oxynitrides. This phenomenon is due to the significant disparity between Gibb’s free energy of VN and V₂O₃. The standard Gibb’s energy of formation for VN is around −11.318 kJ for VN and −30.444 kJ for V₂O₃ (please refer to Appendix C). If the supply of oxygen increased to a certain level, then the thermodynamic equilibrium is in favor of oxides rather than oxynitrides. Such a tip of balance is even more prominent at high temperatures, because of the multiplication of entropy and temperature in Gibb’s free energy. In our case, the critical oxygen flow for directly forming vanadium oxides is somewhere around O₂/N₂ = 1/20. Above this critical ratio, V₂O₃ can form spontaneously without observable transitions from oxynitrides.

In addition, the annealing temperature of 700 °C seems sufficiently high to create various vanadium oxides, even in the case of a low oxygen flow rate of 0.15 sccm (N₂ = 3 sccm). This result indicates that oxynitrides can be produced if the annealing process is controlled below 700 °C. In the present study, films annealed at 600 °C can be a good choice for the formation of oxynitrides, as seen from the XRD.

4.2. Estimated Optical Bandgap

The Beer-Lambert law and Tauc formula in Equations (3) and (4) allow us to estimate the optical bandgaps of films from the UV-Visible-NIR spectra. The determination of bandgaps E_g is numerically implemented by extrapolating the linear part of (αE)^1/2 to intercept the horizontal axis of E. Examples of such calculations are shown as insets in Figure 8 for films deposited under N₂ = 3, 5, and 10 sccm, and various O₂ flow rates.

Among all samples, the annealing at 600 °C leads to the formation of more oxides and thus widens bandgaps. The widened bandgaps also correspond to the higher transmittance (transparency), as seen in the UV-Visible-NIR spectra. However, because these films are still very much vanadium nitrides with relatively high optical absorbance, the estimated bandgaps are narrow, less than ~2.0 eV (~620 nm), and excitable by the orange-red light.

If compared to pure vanadium nitrides, the introduction of a low oxygen supply (<~0.25 sccm) can modify the conductive nature of nitrides to be more semi-conductive oxynitrides in our current setup.

It is also interesting to notice that the band gap is not only dependent on the O₂ flow rate, but also the supply of N₂, despite the fact that the ratio of O₂/N₂ can be identical. A higher N₂ supply can produce more semi-conductive vanadium nitrides in contrast to films deposited under a lower N₂ flow rate, which is virtually more metallic vanadium, as seen from the XRD. The fewer the vanadium nitrides that are presented in films, the easier for vanadium oxides or oxynitrides to form with sufficient oxygen supply and by annealing at appropriate temperatures.

5. Conclusions

Vanadium oxynitride thin films were fabricated using magnetron sputtering and post-rapid thermal annealing at 600 and 700 °C. Various ratios of reactive gas supplies, i.e., O₂/N₂, were set for the deposition process. The films deposited by lower oxygen supply, i.e., O₂/N₂ = 0.15/3, 0.25/5, and 0.5/10 sccm, and annealed subsequently at 600 °C, can produce vanadium oxynitrides films following the analysis by XRD and XPS. These annealed films have an opaque and dark brown appearance. The average electrical resistivity is measured to be ~10⁻² Ω cm, the optical transmittance is somewhat less than 15% in the visible light range, and estimated bandgaps are less than ~2 eV. These oxynitride films have microstructures and properties closer to nitrides but acquire semi-conductive features by the minor components of oxygen.