1. Introduction
Transition metal oxides (TMOs) constitute a class of materials that possesses unique electronic, magnetic, and optical properties, making them attractive for a wide range of applications, including catalysis, energy storage, and electronic devices [
1,
2,
3,
4]. Among the different TMOs, niobium oxides are attractive candidates for these applications due to their hardness and resistance to corrosion, chemical stability, and non-toxicity. However, niobium oxides may present a huge diversity of compositions and crystal structures, which hinders the development of a complete description relating the phases of these oxides with their physico-chemical properties. Despite this, it is worthwhile to increase our comprehension of niobium oxides, as they are materials that withstand aggressive environments well and could therefore increase the useful life of different devices, such as sensors or energy storage devices.
Among the different oxides that niobium can form, niobium pentoxide is the most stable one, in which Nb has an oxidation state of +5 (Nb
2O
5). Nb
2O
5 may have different crystal structures depending on the treatment applied to the material [
5]. The most frequently observed ones are the pseudo-hexagonal phase (TT-Nb
2O
5), which appears when Nb metal, NbO, NbO
2, or amorphous oxides are heated between 320 and 350 °C in air; the orthorhombic phase (T-Nb
2O
5), which is obtained when Nb metal, NbO, NbO
2, amorphous oxide, or TT-Nb
2O
5 are heated between 600 and 800 °C; and the monoclinic phase (H-Nb
2O
5). H-Nb
2O
5 is the most stable phase once formed, and it always appears when the oxide is treated at high temperatures in air (>900 °C).
To obtain Nb
2O
5 nanostructures, several methods have been employed [
6]. The most common ones are hydrothermal synthesis [
7,
8] and sol-gel synthesis [
9,
10]. With these methods, a wide range of structures with different shapes and dimensionality can be obtained. In general, low temperatures are used (in the range of 100–200 °C, although in some cases a post-treatment at higher temperatures is required [
6]) and synthesis times of hours to days. On the other hand, thermal heating of Nb metal foils has also been used to obtain a high density of nanowires [
11,
12,
13]. The growth conditions of this last method would be more comparable with the experiments performed in this work. In general, the thermal treatment of Nb
2O
5 oxide powders only produces structures in the micrometric range [
14]. For all these growth strategies, the time required is in the range of hours. For example, in the case of thermal heating of Nb metal foils, times range from 1 h up to 6 h, temperatures are above 800 °C (generally 900 °C), and the treatment must be performed in a low-oxidizing atmosphere (low vacuum or Ar flux). Rapid growth of Nb
2O
5 nanowires by cold oxygen plasma treatment of Nb foils has been reported [
15], with a growing time of 90 s in vacuum.
In this work, Nb
2O
5 nanowires and micrometric oxide layers have been obtained by Joule resistive heating of Nb metal wires in air. This strategy has been successfully applied for the rapid growth of oxide nanowires of Zn, Cu, W, Mo, or V [
16,
17,
18,
19,
20], and core-shell metal-oxide structures [
21,
22]. However, to the best of our knowledge, this rapid growth has not yet been studied in detail in Nb metal wires. The rapid growth observed in these studies is associated with a diffusion process of the oxygen inside the metal that is accelerated by electromigration due to the high electrical currents employed [
23]. Nb
2O
5 material obtained with this technique can be applied where having the oxide on a metallic substrate is useful, such as in energy storage devices [
24,
25], sensing devices [
12], or field emitters [
11].
2. Materials and Methods
The starting material is a niobium metal wire purchased from Thermo Scientific (Waltham, MA, USA) (purity 99.99% metal basis excluding Ta) with a diameter of 0.25 mm. Segments of 7 cm in length are cut and mounted on the homemade platform used to grow the oxide nanowires (
Figure 1a). As can be seen in the figure, the Nb metal wire is contacted on two steel electrodes, leaving a wire distance of 5 cm suspended in air. The electrodes are contacted to the current source (EA-PS 3016-40 B), which can supply currents up to 40 A (max. power of 640 W). To study the growth conditions, current value and growth time are varied. The voltage variation observed during the treatment is also measured. The final selected growth parameters are shown in
Table 1.
As a consequence of Joule heating, a temperature gradient is established along the metallic wire [
20,
22]. The increase in temperature promotes the formation of oxide nanowires and a micrometric oxide shell covering the original metallic wire. The maximum temperature (
Table 1) is achieved at the center of the length of the Nb wire, while it decreases as we approach the electrodes (
Figure 1b), i.e., electrodes are acting as heat drains. The maximum temperature reached is measured using an optical pyrometer Infratherm IGA 12-S calibrated with the emissivity of Nb. The current value is fixed in the current source before starting the growing process, so that value is applied to the Nb wire instantly, giving a rate of temperature increase of about 500 °C/s.
Electrical characterization of the Nb metal wire is performed using a Keithley 2400 Series SourceMeter (Cleveland, OH, USA). Meanwhile, the oxide formed is characterized by means of X-ray diffraction (XRD), scanning electron microscopy (SEM), energy dispersive X-ray microanalysis (EDX) and micro-Raman spectroscopy (µ-RS), in order to assess their composition and crystal structure. All the measurements are done at room temperature. Cross-section of the Nb metal wires is carried out by embedding the wires in a commercial resin and cutting them with a diamond saw.
XRD measurements are done in a PANalytical X’Pert Powder diffractometer (X-ray diffraction CAI Centers of UCM) in the Bragg–Brentano geometry using Cu–Kα radiation, with a step in 2θ of 0.05°. The morphology of the obtained oxide is studied using a FEI Inspect-S SEM (FEI Company, Eindhoven, Netherlands) working at 20 kV. Additional optical microscope images are recorded with a Leica DFC295 digital camera mounted on a Leica MSV266 microscope (Leica Microsystem, Wetzlar, Germany). For EDX measurements, a Bruker QUANTAX 70 detector (Bruker, Berlin, Germany) attached to a Hitachi TM3000 (Hitachi High Technologies Corporation, Tokyo, Japan) working at 15 kV is employed. The EDX data are analyzed using Bruker ESPRIT QUANTAX 70 EDS software.
µ-RS measurements are carried out with a confocal microscope Horiba JobinYvon LABRAM-HR (Horiba JobinYvon, Villeneuve d’Ascq, France), using the 632.8 nm line of a He–Ne laser. The laser is focused onto the sample using a 100× Olympus objective (0.9 NA), and the scattered light is collected using the same objective (backscattering configuration). The grating employed to analyze the signal has 600 L/mm, and afterwards the signal is recorded with an air-cooled charge-coupled device camera (CCD). µ-RS spectra are analyzed using Labspec 5.0 software.
3. Results and Discussion
Prior to inducing the oxidation of the metallic wires, characterization of the electrical resistivity has been performed. The variation of the resistance (R) with length (L) of the wire is determined using a 4-point probe configuration, and the data are plotted in
Figure 2a.
Then, a linear fit of the data is applied to calculate the resistivity (
) of the Nb metal wire by using Ohm’s law:
where
S is the section of the wire. The obtained resistivity value is
, which matches the expected value for the resistivity of pure niobium at room temperature [
26].
Once the wire is mounted on the growing setup (
Figure 1a) and a high-density current is made to go through it, this value of resistivity is increased. This is related to several factors: temperature increase due to Joule heating, linear expansion of the material due to temperature, and section reduction due to oxide formation on the surface. So, the current source needs to increase the voltage in order to maintain the fixed current, as observed in the measurements of
Table 1. As the section of the metallic wire decreases, higher probability of failure and breaking of the wire is expected, which affects how long the treatment can last. This explains why the time can be expanded for 60 s and more for the lower values of current in
Table 1, whereas times longer than 30 s cannot be maintained for the highest current (5.5 A).
The selected growing parameters allow us to study the evolution of the obtained oxide with current value and time. We observe the formation of niobium oxide in two different scales: nanowires on the surface and the micrometric oxide layer covering the metallic wire (
Figure 2b). We start with the description of the nanowires.
The evolution of the nanowires with current value, keeping a fixed growing time of 60 s, is shown in
Figure 3. For 3.5 A, there is no formation of any nanostructure on the Nb metal wire surface. The formation of nanowires is observed for currents above 4.5 A, whereas for current values between 4.1 and 4.5 A, nucleation points can be identified (
Figure 3a), as well as some isolated nanowires evenly formed, but with low reproducibility of the final growth. As the current increases, the density, length, and diameter of the nanowires increase (
Figure 3b–d). Nanowires are thicker at the base (hundreds of nm) than at the tip (below 100 nm), and they have lengths of a few microns. The best current condition for obtaining a high density of nanowires with high aspect ratio (diameter vs. length) is 4.8 A. At higher currents, nanowires start to evolve to microrods (
Figure 3d). The temperatures reached by the Nb metal wire are comparable to those reported for the growth of niobium oxide nanowires on metal foils by thermal treatment [
11,
12,
13], but the times are clearly shorter. Thus, the rapid growth of the nanowires could be related to an increase in the diffusion rate of oxygen and niobium mobility due to electromigration [
27].
Then, we study the evolution of the nanowires with time for a fixed current value of 4.8 A (
Figure 4). The features of the evolution are similar to those observed for the evolution with current. For a growing time of 5 s (
Figure 4a), nucleation points and tiny nanowires start to appear, but evenly distributed. A high density of nanowires with good aspect ratio is obtained for times between 15 s and 60 s (
Figure 4b–d). Above 60 s (
Figure 4e), the behavior is comparable with the case of 5.0 A in
Figure 3d, i.e., nanowires evolve into microrods. It is also interesting to note that the length of the nanowires or rods does not exceed 5 microns, and no appreciable change is observed when increasing from 60 s to 120 s, as in
Figure 4, or from 4.8 to 5.0 A, as in
Figure 3. This may be an indication of the steps of growth: first nanowires are formed, and their growth is then slowed down due to the formation of the micrometric oxide layer.
The formation of microrods is clearly achieved for current values of 5.5 A (
Figure 5a). These microrods have diameters and lengths of a few microns. On the other hand, for this level of current it is possible to visualize the formation of a micrometric oxide layer by studying the cross-section of the wire (
Figure 5b). In
Figure 5b, the measured oxide layer thickness is around 60 μm, whereas the metallic core diameter has been reduced to 220 μm. It is expected that the thickness of these layers varies with current and time, as observed for the nanowires. We present the results of these experiments in
Figure 6 (current evolution) and
Figure 7 (time evolution).
SEM images in
Figure 6 and
Figure 7 are presented along with the EDX maps of oxygen (cyan) and niobium (red). In both figures, two regions in the Nb wire can be identified, in agreement with the sketch in
Figure 3: a metallic core, where only a niobium signal is detected, and an oxide layer, where both an oxygen and niobium signal are observed. For both current and time evolutions, an increase in the thickness of the oxide layer is clear, associated with a decrease in the diameter of the metallic core. This behavior results from a diffusion phenomenon of oxygen towards the metallic core. It is well known that diffusion rate increases with temperature (higher current means higher temperature) and with time, as derived from Fick’s laws and the diffusion coefficient with temperature.
The formation of the oxide layer is almost symmetric around the metallic core, as seen in
Figure 6 and
Figure 7. To measure the value of the size of the core and the layer, we have measured the height and width, considering the elliptical shape, and the mean values of both measurements are presented in
Figure 8. An increase in time of treatment produces an increase in the thickness of the oxide layer (
Figure 8a). This effect is also observed for an increase in current (
Figure 8b). On the other hand, the formation of the oxide produces a reduction of the metallic core (
Figure 8c–d). The oxidation of the metal produces an increase in the total volume of material [
28], as Nb
2O
5 has close to half the density (4.6 g/cm
3) of Nb metal (8.6 g/cm
3). This ratio agrees with the measured values shown in
Figure 8, mainly for the largest oxide layers. For example, for the fixed current of 4.8 A, the metallic core diameter is reduced from 250 μm to 217 μm (i.e., a reduction of 33 μm) and the oxide layer has a thickness of around 62 μm. The evolution observed with current (which is directly related to the temperature reached) and time is reasonable considering that the growth process is driven by oxygen diffusion from the atmosphere through the oxide layer into the Nb metal [
29].
The diffusion coefficients
D for oxygen in Nb metal and in Nb
2O
5 vary with temperature, following the equations [
30,
31]:
where the pre-exponential coefficient is expressed in m
2/s and the activation energy in J/mol. From these equations, it is clear that the diffusion rate of oxygen is higher in Nb metal than in the oxide. If we calculate the penetration depth (
) of oxygen for different temperature, we can estimate the depth to which diffusion is significant. For the case of oxygen in Nb, the penetration depths are calculated in
Table 2 for different times and temperatures. If we compare these values with the values of reduction of the metallic core, we can see that, if only oxygen diffusion in Nb metal is taken into account, the amount of Nb converted into oxide should be higher. We propose that, in a first stage, an oxide layer is produced through oxygen diffusion from air towards the metallic material. Once this oxide layer is formed, the growth rate would be restricted by the oxygen transference through the oxide, which is much slower. This could explain the almost linear behavior observed for the oxide thickness with time and current (i.e., temperature). Similar behavior was observed in previous studies on the oxidation of niobium by heat treatments [
29,
32,
33]. For example, Kofstad and Kjollesdal [
32] associated the linear kinetics with a rate-determining transport of oxygen ions through an oxide barrier next to the metal. In general, a complex evolution of the oxidation of Nb is observed depending on temperature or oxygen partial pressure, as many variables are involved, e.g., the competition of oxygen diffusion though different Nb
2O
5 polymorphism or the appearance of short-circuit paths for diffusion through the oxide layer [
29].
Once the growth behavior is determined, we perform XRD and µ-RS measurements in order to assess the type of oxide that is formed. The typical XRD pattern recorded is shown in
Figure 9. XRD measurements are performed on the material forming the oxide layer by separating this oxide from the metallic core, giving us information about the main phase of the oxide formed. For all the selected growth conditions where oxide is formed, the diffraction maxima can be associated with the H-Nb
2O
5 monoclinic phase (ICDD card no. 00-037-1468). This observation is coherent with the measured temperatures during the growth (
Table 1). Oxide layer and nanowires appear for currents above 4.1 A, i.e., for temperatures well above 900 °C, which is the condition for forming the H-Nb
2O
5 phase [
5].
To obtain more information about the formed oxide at a micrometric level, µ-RS measurements are carried out on the surface as well as on the cross-section of the Nb metal wires. The spectra measured on the surface where nanowires are formed for different current values are presented in
Figure 9b. All of them correlated well with the expected results for H-Nb
2O
5 phase due to the growing temperature [
14,
34,
35]. Three frequency regions can be identified. First, below 200 cm
−1, modes are related to external lattice vibrations (for which the octahedron is considered as a rigid unit) [
36]. Between 200 cm
−1 and 500 cm
−1, modes related to bending of the Nb-O bodings in the [NbO
6] octahedra are found [
34,
37,
38]. Between 500 and 800 cm
−1, the bands are associated with stretching vibrations in the [NbO
6] octahedra [
34]. In general, several modes are detected as the octahedra in the Nb
2O
5 structure are slightly distorted. Finally, the band located at 992 cm
−1 is generally ascribed to the symmetric stretching modes of the terminal bonds of Nb=O in octahedra with a high degree of distortion [
34,
37], which are related to the existence of shear planes (edge-shared octahedra) in H-Nb
2O
5 [
38,
39]. The main difference observed with the current variation is an increase in the definition of some of the Raman bands, mainly in the low- and medium-frequency region. This phenomenon is observed for an increased crystal quality or a reduction of oxygen deficiency [
40], as well as for an increase in the crystalline size [
35], which is in agreement with our SEM observations of the samples. Separated nanowires show similar Raman spectra associated with the H-Nb
2O
5 phase (
Figure 10a).
For the wires oxidized with the higher values of current (i.e., 5.0 A and 5.5 A), a variation of the RS signal is observed along the length of the wire (
Figure 10). Considering that the temperature reached at the nanowire has a profile that decreases towards the electrodes (see
Figure 1b), we record the spectra at the center of the wire (2.5 cm from the electrode, ‘high’ temperature in
Figure 10b,c); at 1.25 cm from the electrode (‘medium’ temperature); and at 0.7 cm from the electrode (‘low’ temperature). In the Raman spectra, we observe less defined bands as we move away from the center of the Nb metallic wire. This is in agreement with a reduction in the crystal quality or the increase of structures in the nanometric scale. An inspection of the structures obtained along the metallic wire in the selected positions indicate that, as the temperature profile reduces, the structures evolve towards nanowires, in an evolution similar to the one observed for the current variation (
Figure 4 and
Figure 9b).
Finally, we study the Raman signal recorded on the cross-section of a Nb wire, to check if there is any modification along the thickness of the oxide layer. The optical section of the metallic wire (
Figure 11a) allows us to identify the oxide layer (white in the optical micrograph) and the metallic core (dark blue in the micrograph). The line profile of the evolution of the Raman spectra is shown in
Figure 11b. In the graph, the position at which the spectrum is recorded (Y-axis) is presented in terms of Raman shift (X-axis) and intensity of the Raman signal (Z-axis). Similar RS bands are recorded along the thickness of the layer, confirming that the oxide layer is uniform in crystal structure. No signal related to sub-oxides (NbO, NbO
2) is detected. Spectra recorded at selected points are presented in
Figure 11c. At the center of the metallic core, no RS signal is obtained, as it is a metal without oxidation. On the other hand, the RS signal recorded along the oxide layer shows the same features as on the surface (see
Figure 9b). It is at the boundary between the oxide and the metallic core that some changes in the spectra are seen, mainly on the spectrum recorded at the boundary on the side of the metal (‘boundary in’). Even on a region that appears metallic in the optical micrograph, an oxide signal is still detectable, confirming that oxygen is diffusing inside the metal to form new oxide layers.