Structural Changes Induced by Heating in Sputtered NiO and Cr 2 O 3 Thin Films as p-Type Transparent Conductive Electrodes

: NiO and Cr 2 O 3 are transition metal oxides with a partially ﬁlled d electron band that supports p-type conduction. Both are transparent to the visible light due to optical absorption beginning at wavelengths below 0.4 µ m and the creation of holes by metal vacancy defects. The defect and strain effects on the electronic characteristics of these materials need to be established. For this purpose, NiO and Cr 2 O 3 thin ﬁlms were deposited on unheated glass substrates by reactive DC sputtering from metallic targets. Their structural, morphological, optical and electrical properties were analyzed comparatively in the as-grown conditions (25 ◦ C) and after heating in air at 300 ◦ C or 500 ◦ C. The cubic NiO structure was identiﬁed with some tensile strain in the as-grown conditions and compressive strain after heating. Otherwise, the chromium oxide layers were amorphous as grown at 25 ◦ C and crystallized into hexagonal Cr 2 O 3 at 300 ◦ C or above also with compressive strain after heating. Both materials achieved the highest visible transmittance (72%) and analogous electrical conductivity (~10 − 4 S/cm) by annealing at 500 ◦ C. The as-grown NiO ﬁlms showed a higher conductivity (2.5 × 10 − 2 S/cm) but lower transmittance (34%), which were related to more defects causing tensile strain in these samples.


Introduction
Transparent conducting oxides (TCOs) are critical to numerous technological applications, ranging from flat panel displays or light emitting diodes to smart windows and photovoltaic cells [1,2]. The performances of p-type TCOs are behind the n-type counterparts, which limit potential applications in the field of transparent electronics [3]. This discrepancy stems from the localized nature of the O 2p-derived valence band (VB) in contrast with a spatially spread conduction band (CB) composed of metal s orbitals, which results in a much higher effective mass for holes than electrons [4] in addition to easy n-type dopability by oxygen vacancy defects [5]. A design strategy for p-type TCOs is based on the concept of chemical modulation of the VB through the hybridization of O 2p orbitals with metal d or s orbitals, which is advantageous for the mobility of holes [6]. Another approximation is to utilize the electron correlation to promote VB modifications that favor p-type conduction [7]. In this sense, many transition metal oxides with a partially filled d electron band are described by extended Hubbard models [8,9] where the Coulomb interaction between the electrons (U) splits the d band into an upper Hubbard band (UHB) and a lower Hubbard band (LHB) with a separation of U [10]. Consequently, these compounds can support p-type conduction when the VB is composed of O 2p orbitals and metal d orbitals (LHB) driven by the electron correlation.
Transition metal monoxides crystallize in the rock-salt structure and sesquioxides in the corundum structure, showing a similar trend of electronic properties: the values of U gradually decrease with the decrease of the 3d occupation number of the transition metal ion [11]. The electrical conductivity can be improved by decreasing U while maintaining the consequence of the VB modification due to the electron correlation for p-type conduction. Although with differences in their electronic structure, NiO (Ni 2+ 3d 8 ) and Cr 2 O 3 (Cr 3+ 3d 3 ) are equally suitable for acting as p-TCOs [6]. Both are transparent to the visible light due to optical absorption beginning at wavelengths below 0.4 µm and they allow the creation of holes by metal vacancy defects (V Ni or V Cr ) [12,13].
Due to its lower 3d occupation number, a lower U and higher electrical conductivity are expected for Cr 2 O 3 than NiO. Nevertheless, the respective electronic characteristics can be altered as previous studies have revealed that compressive strain can increase the band gap energy and U values in both Cr 2 O 3 [13] and NiO [14]. In addition, the metal vacancy defects play an important role in changing the lattice arrangement and introducing charge transition levels that modify the electronic bands [15,16]. The defect and strain effects on the electronic characteristics of these p-type TCOs need to be established in relation to the specific preparation procedure. For this purpose, NiO and Cr 2 O 3 thin films were deposited on unheated glass substrates by reactive DC sputtering from metallic targets. This is a good technique for the preparation of metal oxide layers with a low cost due to simple equipment and easy extrapolation to large areas [17] and it is used here for the synthesis of compounds that are practically nontoxic [18] although source metals and potential byproducts (such as Cr 6+ species) are problematic. The structural, morphological, optical and electrical properties of the sputtered films were comparatively analyzed in the as-grown conditions (25 • C) and after heating in air at 300 • C or 500 • C, thus studying the evolution of their characteristics with temperature. The main objective was to elucidate the structural effects on the transparency and conductivity of NiO and Cr 2 O 3 thin films prepared under analogous conditions by a low cost and easily scalable deposition technique.

Materials and Methods
Nickel oxide and chromium oxide thin films were prepared on unheated soda-lime glass substrates (2 mm thick) by reactive DC magnetron sputtering from Ni and Cr targets, respectively. After the evacuation of the chamber to a base pressure of 4 × 10 −4 Pa, the reactive and working gases (O 2 and Ar) were introduced by independent mass flow controllers. The O 2 to Ar partial pressure ratio was selected to obtain near stoichiometric NiO and Cr 2 O 3 layers according to previous studies [12,19,20] while the power density was adjusted to get a deposition rate of 20 nm/min for the various films, as detailed in Table 1. For the analogous nickel oxide layers, a previous work showed the evolution of the cation coordination number (N c ) with the oxygen partial pressure being the stoichiometric value (N c = 6) achieved at P(O 2 )/P(O 2 + Ar) = 0.2 [12]. For the chromium oxide samples, several oxygen pressures were also tested and the set value corresponded with the intermediate range between the metallic mode (opaque films) at P(O 2 )/P(O 2 + Ar) ≤ 0.3 and the oxidative mode (with the poisoning of the target [19]) at P(O 2 )/P(O 2 + Ar) ≥ 0.5. The thickness of each layer was measured after deposition with a Dektak 3030 profilometer (Veeco Instruments GmbH, Mannheim, Germany). This value and the corresponding deposition time were used to calculate the growth rate and adjust the power density accordingly. Once the rate of 20 nm/min was reached, the deposition time was set at 5 min to obtain layers with a same thickness of 100 nm. In order to improve the crystallinity of the sputtered samples, post-deposition heat treatments were performed in air at temperatures ranging from 300 • C to 500 • C for 30 min.
The crystallographic properties were examined by X-ray diffraction (XRD) with radiation Cu K α1 (λ = 1.54056 Å) in an X'Pert instrument (PANalytical, Malvern, UK) with a Bragg-Brentano θ-2θ configuration. The crystalline phases were identified by the files given by the Joint Committee of Powder Diffraction Standards (JCPDS) and the mean crystallite size was calculated from the full width at half the maximum of the main diffraction peak according to the Scherrer formula. The microstructure of the films was also analyzed with a system composed of a BAC151B microscope and an i-Raman spectrometer (B&W Tek, Newark, NJ, USA) using a green laser of 532 nm as the excitation source. The topography was examined by atomic force microscopy (AFM) with a Park XE-100 (Park Systems, Suwon, Korea) in contact mode, acquiring digital images to quantify the surface roughness. For the optical characterization, transmittance (T) and reflectance (R) measurements were done with a double beam spectrophotometer Lambda 9 (PerkinElmer Inc., Waltham, MA, USA) in the wavelength range λ = 0.3-1.9 µm taking the air as the reference. The transmittance was then corrected for reflection losses as T c (%) = 100 T(%)/(100 − R(%)). This was directly related to the optical absorption coefficient α = −(1/t) ln(T c /100) [21] being T c ≈ 100% below the band gap energy E g where α ≈ 0. Thus, the E g value could be determined as the maximum of the differential dα/dE or dT c /dE versus the light energy E [22]. The electrical conductivity together with carrier concentration and mobility were determined with an HMS3000 Hall Measurement System (ECOPIA, Gyeonggi-do, Korea) using the Van der Pauw configuration and a magnetic field of 0.5 T. Silver paste contacts with a 1 mm diameter size were placed at the four corners of the sample (1 cm × 1 cm) and four measurements (with the bias applied in different diagonal directions and reversing the field) were taken of each layer. The sheet resistance (R s ) was also obtained with a four-point probe VEECO FPP5000 (Veeco Instruments Inc., Plainview, NY, USA) showing a complete agreement with the equation R s = (σ·t) −1 where σ was the Hall conductivity and t was the film thickness.

Results and Discussion
An XRD analysis conducted on the various samples is illustrated in Figure 1. All of the nickel oxide layers showed diffraction peaks corresponding with the cubic rock-salt structure of NiO (JCPDS No. 4-835). A mean crystallite size of 40 nm was obtained for the as-grown films with a lattice constant a = 0.420 nm that was higher than expected from the standard powder file (0.418 nm) [12]. Such enlargement in the lattice was related to the presence of nickel vacancies (V Ni ) 2− and the concurrent formation of some Ni 3+ to preserve electrical neutrality. The coexistence of Ni 2+ and Ni 3+ produced a local spinel arrangement (as in a mixed valence Ni 3 O 4 [23]) dispersed in the NiO matrix, which altered the cubic rock-salt structure and increased the lattice parameter [24]. During heating in air at 300 • C or 500 • C, NiO crystallites enlarged slightly to 45 nm while the lattice constant decreased to a = 0.416 nm. Otherwise, no diffraction peaks were observed for the as-grown chromium oxide samples, indicating that they were amorphous as reported for analogous layers prepared at room temperature [25,26]. After heating at 300 • C or above, diffraction peaks evidenced hexagonal Cr 2 O 3 (JCPDS 72-3533) with a mean crystallite size of 55 nm and lattice parameters a = 0.492 nm and c = 1.359 nm. These gave a unit cell volume lower than expected from the standard powder file (0.495 nm and 1.358 nm cell parameters) [27] without significant changes with the annealing temperature. At 300 • C, some Cr 2 O 5 (JCPDS 28-0370) also appeared that was related to the coexistence of Cr 3+ and Cr 6+ by the combination of CrO 8 octahedra and CrO 4 tetrahedra [28]. This is a metastable compound favored by the presence of chromium (III) vacancies (V Cr ) 3− together with Cr 6+ species but Cr 2 O 5 decomposes into Cr 2 O 3 around 400 • C [29,30]. In fact, pure Cr 2 O 3 is typically synthetized at temperatures above 450 • C [19,29]. The presence of Cr 2 O 5 in the 300 • C heated sample produced some screening of the (012) Cr 2 O 3 diffraction, which was located between two peaks corresponding with Cr 2 O 5 . It should be noted that a similar evolution has been observed by increasing the heating temperature during sputter deposition [31,32]. However, the film growth rate tends to decrease as the substrate temperature increases [31]. Sputtering on unheated substrates allows the minimisation of the power required for the stated deposition rate and the film thickness remains unchanged after post-deposition heating.
ctron. Mater. 2021, 2, FOR PEER REVIEW ature increases [31]. Sputtering on unheated substrates allows the m power required for the stated deposition rate and the film thickness after post-deposition heating. Raman spectroscopy is known to be very sensitive to chem bonding and it was used to complement the XRD data. The results o ious samples are shown in Figure 2. Nickel oxide layers evidenced m corresponding with pure NiO in the first-order transverse optical (1T optical (1LO) phonon modes, the second-order 2TO mode and the TO The films obtained at a relatively low temperature (as-grown or 300 ° a hybridized band of 1TO and 1LO phonons, which is typical of a d [35]. A decrease of the 1LO phonon energy was related to VNi defec with the tensile lattice distortion detected by XRD in the as-grown chromium oxide layers, Raman shifts were assigned to the A1g and [37]. The signal increment observed at ~650 cm −1 for the as-grown s tributed to the forbidden Raman modes of Cr2O3 activated in am [26,38], which was in accordance with the absence of diffraction sponding XRD pattern ( Figure 1). Otherwise, the Raman band that ap after heating was related to the Cr(VI) states in CrO3 [39] or Cr2O5 [40 °C heated sample, the small signal increment around 850 cm −1 was phase detected by XRD, which decomposed into Cr2O3 and CrO3 dur °C. Raman spectroscopy is known to be very sensitive to chemical structures and bonding and it was used to complement the XRD data. The results obtained for the various samples are shown in Figure 2. Nickel oxide layers evidenced molecular vibrations corresponding with pure NiO in the first-order transverse optical (1TO) and longitudinal optical (1LO) phonon modes, the second-order 2TO mode and the TO+LO mode [33,34]. The films obtained at a relatively low temperature (as-grown or 300 • C heated) exhibited a hybridized band of 1TO and 1LO phonons, which is typical of a distorted NiO lattice [35]. A decrease of the 1LO phonon energy was related to V Ni defects [36] in agreement with the tensile lattice distortion detected by XRD in the as-grown NiO films. For the chromium oxide layers, Raman shifts were assigned to the A 1g and E g modes of Cr 2 O 3 [37]. The signal increment observed at 650 cm −1 for the as-grown sample could be attributed to the forbidden Raman modes of Cr 2 O 3 activated in amorphous materials [26,38], which was in accordance with the absence of diffraction peaks in the corresponding XRD pattern ( Figure 1). Otherwise, the Raman band that appeared at~850 cm −1 after heating was related to the Cr(VI) states in CrO 3 [39] or Cr 2 O 5 [40] forms. For the 300 • C heated sample, the small signal increment around 850 cm −1 was related to the Cr 2 O 5 phase detected by XRD, which decomposed into Cr 2 O 3 and CrO 3 during annealing at 500 • C. The morphology of the various NiO and Cr 2 O 3 layers is illustrated in Figure 3, which includes representative AFM images taken on 2 µm × 2 µm areas together with the respective root-mean-square roughness (r). The images showed smooth and homogeneous films being the surface roughness minimum for the as-grown samples (r < 1 nm) and increased gradually after heating at 300 • C (r ≈ 2 nm) and 500 • C (r ≈ 3 nm). These values were in the same order than those reported for analogous NiO [41][42][43] and Cr 2 O 3 [32,38,44] thin films. The increment of the roughness with the annealing temperature was related to the recrystallization process [32,43] and the consequent enlargement of the mean crystallite size, as confirmed by the XRD data. Such crystalline enhancement goes together with the agglomeration of grains and the formation of clusters that increase in size as the annealing temperature increases. In addition, undulations appeared on the surface of the layers heated at the highest temperature, as observed in other works [41]. The roughness of the film was related to its crystallinity but could also affect its optical properties because greater roughness could lead to more scattering at the surface and degrade its transparency by increasing the absorption [32]. In any case, all samples had a root-mean-square roughness well below their mean crystallite size indicating that they were flat and compact layers. Therefore, these low values of film roughness are desirable for optoelectronic applications.
ctron. Mater. 2021, 2, FOR PEER REVIEW Figure 2. Raman spectra corresponding with the nickel oxide and chromium grown at room temperature and after heating in air at 300 °C or 500 °C.
The morphology of the various NiO and Cr2O3 layers is illustrate includes representative AFM images taken on 2 μm × 2 μm areas to spective root-mean-square roughness (r). The images showed smooth films being the surface roughness minimum for the as-grown sample creased gradually after heating at 300 °C (r ≈ 2 nm) and 500 °C (r ≈ 3 were in the same order than those reported for analogous NiO [32,38,44] thin films. The increment of the roughness with the annealin related to the recrystallization process [32,43] and the consequent mean crystallite size, as confirmed by the XRD data. Such crystalline together with the agglomeration of grains and the formation of clus size as the annealing temperature increases. In addition, undulatio surface of the layers heated at the highest temperature, as observed The roughness of the film was related to its crystallinity but could a properties because greater roughness could lead to more scattering degrade its transparency by increasing the absorption [32]. In any cas root-mean-square roughness well below their mean crystallite size i The optical transmission spectra of the various samples are given in Figure 4. This included the standard photopic vision V(λ), which described the spectral sensitivity of the human eye centered at λ V = 0.55 µm to better asses the transparency of the films. The visible transmittance (that is, the value T V taken at λ V ) showed the same evolution for the NiO and Cr 2 O 3 layers, increasing from T V = 36 ± 2% for the as-grown films to 54% after heating at 300 • C and higher, to 72%, when the temperature increased to 500 • C. An increment in the visible transmittance with an increasing annealing temperature has been reported for analogous coatings [45][46][47] with a maximum value around 70% at a 0.55 µm wavelength [46,47]. In general, the optical transmission is expected to depend on various film properties such as impurity centers, surface roughness and level of crystallinity. For these smooth layers, structural defects and interstitial O atoms existing in the as-deposited samples acted as impurities and led to scattering and/or the absorption of incident light. As a result of annealing, the mean crystallite size increased and the interstitial O atoms could diffuse out [45] leading to a decrease in the impurity level and the subsequent increment of the film transmittance with the heating temperature. Furthermore, the coexistence of metal ions in different valence states (Ni 2+ /Ni 3+ or Cr 3+ /Cr 6+ ), which share oxygen anion ligands, allows facile charge transfer processes involving optical absorption in the visible range [23]. The structural data indicated that such coexistence was favored in the as-grown samples but subsequent heating produced an evolution to the most stable valence state.
Both the crystallite size increase and the interstitial O out-diffusion led to the removal of impurity levels and charge transfer processes (that absorbed visible light) after heating, which had a greater effect than roughness on the film transmittance. The optical transmission spectra of the various samples are given in Figure 4. This included the standard photopic vision V(λ), which described the spectral sensitivity of the human eye centered at λV = 0.55 μm to better asses the transparency of the films. The visible transmittance (that is, the value TV taken at λV) showed the same evolution for the NiO and Cr2O3 layers, increasing from TV = 36 ± 2% for the as-grown films to 54% after heating at 300 °C and higher, to 72%, when the temperature increased to 500 °C. An increment in the visible transmittance with an increasing annealing temperature has been reported for analogous coatings [45][46][47] with a maximum value around 70% at a 0.55 μm wavelength [46,47]. In general, the optical transmission is expected to depend on various film properties such as impurity centers, surface roughness and level of crystallinity. For these smooth layers, structural defects and interstitial O atoms existing in the as-deposited samples acted as impurities and led to scattering and/or the absorption of incident light. As a result of annealing, the mean crystallite size increased and the interstitial O atoms could diffuse out [45] leading to a decrease in the impurity level and the subsequent increment of the film transmittance with the heating temperature. Furthermore, the coexistence of metal ions in different valence states (Ni 2+ /Ni 3+ or Cr 3+ /Cr 6+ ), which share oxygen anion ligands, allows facile charge transfer processes involving optical absorption in the visible range [23]. The structural data indicated that such coexist-  Figure 5 represents the first derivative of the optical transmittance as a function of the radiation energy for the same spectra shown in Figure 4. The optical gap energy given by the first derivative maximum [22] was clearly identified by prominent peaks in the heated samples, which were in agreement with those expected for the respective material; that is, E g = 3.70 eV for the NiO films [22,24,46] and E g = 3.20 eV for the Cr 2 O 3 layers [27,47]. The chromium oxide layer heated at 500 • C showed an additional maximum at 2.50 eV, which corresponded with the band gap energy of CrO 3 [48] according to the CrO 3 signal also detected in the Raman spectrum of this sample (in Figure 2). Otherwise, the first derivative maximum appeared less pronounced and shifted towards lower energies in the as-grown layers. The same behavior was observed for other amorphous or poorly crystalline films, which showed a dependence of the band gap energy with the crystallite size and/or with the structural strain. A decrease in the optical gap was noted when the NiO lattice parameter increased and the film turned into tensile strain [14] taking into account that the mean crystallite size decreases with tensile strain as observed here for the as-deposited NiO. Theoretical calculations indicated that tensile strain would also reduce the gap of Cr 2 O 3 [13]. Furthermore, E g < 3.2 eV has been reported for NiO coatings with mean crystallite sizes below 20 nm [49] and E g < 2.7 eV for Cr 2 O 3 crystallites below 28 nm [50]. Thus, it can be assumed that the fundamental gap of the crystalline phase and the amorphous phase would be different [31]. Indeed, the near-edge spectrum of amorphous Cr 2 O 3 was a slightly broadened version of the single crystal spectrum with a shift of the peaks of~0.4 eV to lower energies [51]. The origin of the band gap narrowing was then related to structural imperfections (bond angle and distance distortions in relation to defects such as vacancies or interstitials) in the NiO and Cr 2 O 3 films obtained at room temperature. For these samples, the decrease in the optical gap energy led to a decrease in visible transmittance.
Electron. Mater. 2021, 2, FOR PEER REVIEW ence was favored in the as-grown samples but subsequent heating produced an evol tion to the most stable valence state. Both the crystallite size increase and the interstitial out-diffusion led to the removal of impurity levels and charge transfer processes (th absorbed visible light) after heating, which had a greater effect than roughness on th film transmittance. . Optical transmittance spectra for the NiO and Cr2O3 films as grown at room temperatur and after heating in air at 300 °C or 500 °C. The standard photopic vision V(λ), which describes th sensitivity of the human eye, is included for comparison. Figure 5 represents the first derivative of the optical transmittance as a function the radiation energy for the same spectra shown in Figure 4. The optical gap energy give by the first derivative maximum [22] was clearly identified by prominent peaks in th heated samples, which were in agreement with those expected for the respective mater al; that is, Eg = 3.70 eV for the NiO films [22,24,46] and Eg = 3.20 eV for the Cr2O3 laye [27,47]. The chromium oxide layer heated at 500 °C showed an additional maximum 2.50 eV, which corresponded with the band gap energy of CrO3 [48] according to th CrO3 signal also detected in the Raman spectrum of this sample (in Figure 2). Otherwis the first derivative maximum appeared less pronounced and shifted towards lower e ergies in the as-grown layers. The same behavior was observed for other amorphous poorly crystalline films, which showed a dependence of the band gap energy with th crystallite size and/or with the structural strain. A decrease in the optical gap was note when the NiO lattice parameter increased and the film turned into tensile strain [1 taking into account that the mean crystallite size decreases with tensile strain as observe here for the as-deposited NiO. Theoretical calculations indicated that tensile strain wou also reduce the gap of Cr2O3 [13]. Furthermore, Eg < 3.2 eV has been reported for Ni coatings with mean crystallite sizes below 20 nm [49] and Eg < 2.7 eV for Cr2O3 crystallit below 28 nm [50]. Thus, it can be assumed that the fundamental gap of the crystallin phase and the amorphous phase would be different [31]. Indeed, the near-edge spectru of amorphous Cr2O3 was a slightly broadened version of the single crystal spectrum wi a shift of the peaks of ~0.4 eV to lower energies [51]. The origin of the band gap narrow ing was then related to structural imperfections (bond angle and distance distortions relation to defects such as vacancies or interstitials) in the NiO and Cr2O3 films obtaine The electrical data of the sputtered coatings are plotted in Figure 6 as a function the heating temperature. In all cases, the p-type conductivity was dominated by t concentration of free holes, both being maximum (σ = 2.5 × 10 −2 S/cm and N = 2.7 × 1 cm −3 ) for the unheated NiO film grown at 25 °C. The NiO conductivity decreased to 3.1 10 −3 S/cm at 300 °C and 8.9 × 10 −4 S/cm at 500 °C due to the decrease in the carrier co The electrical data of the sputtered coatings are plotted in Figure 6 as a function of the heating temperature. In all cases, the p-type conductivity was dominated by the concentration of free holes, both being maximum (σ = 2.5 × 10 −2 S/cm and N = 2.7 × 10 16 cm −3 ) for the unheated NiO film grown at 25 • C. The NiO conductivity decreased to 3.1 × 10 −3 S/cm at 300 • C and 8.9 × 10 −4 S/cm at 500 • C due to the decrease in the carrier concentration to the 10 15 -10 14 cm −3 range. The same behavior was observed for other sputtered NiO layers [45,52] when the O-rich plasma induced the incorporation of Ni 3+ in the lattice and correlatively the formation of nickel vacancies as the previous structural analysis evidenced for the present samples. These nickel vacancies acted as acceptors (V Ni ) 2− that generated holes to fulfill the lattice charge neutrality resulting in the highest conductivity. Nevertheless, the as-grown NiO was thermodynamically unstable and it was easy to release excess oxygen atoms by heat treatment at 300 • C or 500 • C, which produced the out-diffusion of the interstitial oxygen atoms bonded weakly in the sputtered NiO films [52]. Otherwise, the electrical characteristics of Cr 2 O 3 remained stable at σ = (4.5 ± 0.1) × 10 −4 S/cm and N = (1.5 ± 0.2) × 10 14 cm −3 over the entire temperature range. In this case, the incorporation of Cr 6+ led to the formation of Cr 2 O 5 and CrO 3 phases (detected by XRD and Raman spectroscopy) but was less effective in creating chromium vacancy defects that acted as acceptors. The crystallization and disappearance of the disorder-induced Raman mode (at~650 cm -1 ) did not correlate here with a change in conduction. With regard to the carrier mobility, it is represented in the Figure 6 inset as a function of the respective carrier concentration. It was observed as practically constant at µ = 18 ± 4 cm 2 /Vs for the various layers with N ≤ 10 15 cm −3 and decreased to µ = 6 cm 2 /Vs for the as-grown NiO with N = 2.7 × 10 16 cm −3 . These mobility values were within the highest reported for NiO [14,35] and Cr 2 O 3 [53] thin films. Depending on the localization length of the holes (L h ) in relation to the lattice constant (a), the conductivity could proceed by hopping (when L h < a) or by a band-like transport (when L h > a) [54]. In the first case, the mobility was expected to be extremely low (µ < 0.1 cm 2 /Vs [54]) while the high mobilities observed in the present samples indicated a band-like transport. The lower mobility in the as-grown NiO was due to its higher proportion of defects, which scattered the movement of holes [14]. These sputtered NiO samples exhibited better electrical conduction than Cr 2 O 3 for analogous visible transmittance, which is desirable for TCOs. After heating at 500 • C, the NiO films had a sheet resistance R s = 1.1 × 10 5 kΩ comparable with that achieved by other p-type conductors with a visible transmittance above 70% [6].

Conclusions
NiO and Cr2O3 thin films were deposited by reactive DC sputtering on unheated glass substrates and subsequently annealed in air at 300 °C or 500 °C. Beyond their different electronic structure, the optical and electrical characteristics were dominated by

Conclusions
NiO and Cr 2 O 3 thin films were deposited by reactive DC sputtering on unheated glass substrates and subsequently annealed in air at 300 • C or 500 • C. Beyond their different electronic structure, the optical and electrical characteristics were dominated by the respective structural defects, which could be changed by the heating temperature.
All of the nickel oxide samples showed a cubic NiO structure with some tensile strain in the as-grown conditions and compressive strain after heating. The tensile strain was related to the presence of Ni 3+ and nickel vacancy defects, which acted as optical absorption centers and charge acceptors (V Ni ) 2− . This resulted in a low visible transmittance (34%) and high p-type conductivity (2.5 × 10 −2 S/cm) for the as-grown NiO. Nickel vacancy defects, the concentration of holes and conductivity decreased after heating (down to 8.9 × 10 −4 S/cm at 500 • C) while the visible transmittance increased (up to 72% at 500 • C).
Otherwise, the chromium oxide layers were amorphous as grown at 25 • C and crystallized into hexagonal Cr 2 O 3 at 300 • C or above also with compressive strain after heating. The presence of some Cr 6+ led to the formation of Cr 2 O 5 (at 300 • C) and CrO 3 (at 500 • C) but was less effective in creating chromium vacancy defects that acted as acceptors. Thus, the electrical conductivity remained stable at~4.5 × 10 −4 S/cm over the entire temperature range although the visible transmittance increased with heating in the same way as for NiO.