Electrical and Optical Properties of Amorphous SnO 2 :Ta Films, Prepared by DC and RF Magnetron Sputtering: A Systematic Study of the Influence of the Type of the Reactive Gas

: By reactive magnetron sputtering from a ceramic SnO 2 :Ta target onto unheated substrates, X-ray amorphous SnO : Ta films were prepared in gas mixtures of Ar/O 2 (N 2 O, H 2 O). The process windows, where the films exhibit the lowest resistivity values, were investigated as a function of the partial pressure of the reactive gases O 2 , N 2 O and H 2 O. We found that all three gases lead to the same minimum resistivity, while the width of the process window is broadest for the reactive gas H 2 O. While the amorphous films were remarkably conductive ( ρ ≈ 5 × 10 − 3 Ω cm), the films crystallized by annealing at 500 °C exhibit higher resistivities due to grain boundary limited conduction. For larger film thicknesses (d ≳ 150 nm), crystallization occurs already during the deposition, caused by the substrate temperature increase due to the energy influx from the condensing film species and from the plasma (ions, electrons), leading to higher resistivities of these films. The best amorphous SnO 2 :Ta films had a resistivity of lower than 4 × 10 − 3 Ω cm, with a carrier concentration of 1.1 × 10 20 cm − 3 , and a Hall mobility of 16 cm 2 /Vs. The sheet resistance was about 400 Ω /  for 100 nm films and 80 Ω /  for 500 nm thick films. The average optical transmittance from 500 to 1000 nm is greater than 76 % for 100 nm films, where the films, deposited with H 2 O as reactive gas, exhibit even a slightly higher transmittance of 80 %. These X-ray amorpous SnO 2 :Ta films can be used as low-temperature prepared transparent and conductive protection layers, for instance, to protect semiconducting photoelectrodes for water splitting, and also, where appropriate, in combination with more conductive TCO films (ITO or ZnO).


Introduction
Tin dioxide (SnO2) belongs to the class of wide bandgap, oxidic semiconductors that can be, in order to achieve high conductivities, doped up to high carrier concentrations (> 10 20 cm −3 ). SnO2, like indium oxide and zinc oxide, also belonging to this material class, exhibits isotropic metal 5s orbitals which form the conduction band of these compound semiconductors. The isotropy of their conduction bands, different from other semiconductors, like silicon or GaAs, is advantageous for the good transport properties (high carrier mobility) of these transparent conductive oxides even in the amorphous state [1,2].
SnO2 is much cheaper than indium oxide, but more expensive than ZnO, the other two widely used transparent conductive oxides (TCO) [3]. An inherent advantage of SnO2 is its high chemical [1]. Such amorphous oxide films in the In-Ga-Zn-O system with low carrier concentrations have been used successfully for the preparation of transparent and flexible field effect transistors, where the absence of grain boundaries in the channel regions increased the field effect mobility significantly [37]. Recently, indium oxide films were investigated in detail with respect to the relation between structural and electronic transport properties [38]. Buchholz et al. analyzed the transition from amorphous to crystalline In2O3 films both experimentally and theoretically.
Another beneficial effect of the absence of grain boundaries in amorphous films is that amorphous films typically exhibit lower diffusion coefficients. This property of amorphous films can be exploited for use in protective films [39].
In this article, the deposition of X-ray amorphous, Ta-doped SnO2 films prepared by reactive magnetron sputtering from a ceramic target is investigated systematically. The depositions were performed onto unheated glass substrates in order to explore the possibility of depositing these SnO2 films onto temperature-sensitive substrates (plastics) or sensitive underlying films or devices (thin film solar cells or photoelectrodes for water splitting). We investigated in detail the effects of the plasma excitation (DC or RF, 13.56 MHz) and the type of the reactive gas (O2, H2O, N2O) on the electrical properties of the SnO2:Ta films. Due to its superior chemical stability, such SnO2:Ta films are intended for applications as transparent and conductive protection layers [39].

Experimental
The SnO2:Ta films were prepared at different substrate temperatures by reactive magnetron sputtering from a ceramic target in a Leybold Z400 sputtering system, equipped with a fast load-lock. The base pressure of the system is about 5 × 10 −4 Pa. While the reactive gases O2 and N2O were fed into the sputtering chamber in the conventional way with mass flow controllers (MFC) from gas bottles, the water vapour (pH2O = 23 hPa at 20 °C) was fed into the chamber by a special low-pressure MFC from the liquid H2O reservoir, a quartz container held at room temperature.
The ceramic Sn0.98Ta0.02O2 target (purity 4N, supplier: EVOCHEM Advanced Materials GmbH) with 75 mm diameter was bonded on a Cu cooling well. It was sintered under reducing conditions, thus also allowing DC plasma excitation. The depositions were done at a target-to-substrate distance of 55 mm onto stationary substrates. Before the deposition, a substrate pretreatment with an RF argon plasma was performed for 1 min (pAr = 1.8 Pa, PRF = 100 W, VDC ≈ 1000 V). The substrate holder could be heated by a resistance heater up to glass substrate temperatures of about 400 °C, measured with a thermocouple.
The film thickness was measured using a DEKTAK profilometer. The optical parameters were obtained by spectroscopic ellipsometry in the photon energy range from 1.5 to 4.2 eV (SE 850, Sentech GmbH, Berlin, Germany). These spectra were fitted and analysed with the commercial optical analysis program 'SpectraRay3' from Sentech. The optical transmittance (T) and reflectance (R) spectra were measured with an UV/Vis double-beam spectrometer (Cary 05E, Varian) in the spectral range from 200 to 3200 nm (6.2 to 0.38 eV).
The structural properties of the SnO2:Ta films were analyzed by X-ray diffraction (XRD) with a silicon-stripe detector (D2 Phaser with a Cu-anode and a Lynxeye detector, Bruker AXS) in Bragg-Brentano geometry. The diffraction peaks were fitted by double peaks with Lorentzian peak-shape, thus yielding the positions and widths of the two CuKα peaks.
The thickness-averaged elemental composition was measured using Rutherford backscattering spectrometry (RBS) on samples deposited onto glassy carbon substrates. He ions with an energy of 1.7 MeV at normal incidence were used for the analysis. Typically, a 4 He + charge of 10 µC was used to collect RBS spectra with good counting statistics. The RBS spectra were simulated with the software SIMNRA [40]. In order to determine the electron concentration and the Hall mobility, Hall and conductivity measurements were performed with a home-built setup at room temperature at a magnetic flux of 0.74 T. The 10 × 10 mm 2 samples were contacted in the corners in the van der Pauw geometry. The radial profiles of the electrical parameters were measured in order to investigate the role of negative oxygen ion bombardment which is radially inhomogeneous [41].
The cross-section of the SnO2:Ta films was prepared with Argon ion beam polishing and investigated with transmission electron microscopy (TEM Zeiss LIBRA 200FE) in the bright-field mode including diffraction contrast and zero-energy-loss filtering.
Post-annealing treatments were carried out with an IR-lamp heater (maximum intensity at a wavelength of about 1.5 µm) in vacuum (p = 1 × 10 −2 Pa) or in hydrogen atmosphere (pH2 = 200 hPa) at a temperature of T = 500 °C. Prior to the annealing, the vacuum chamber was evacuated to a vacuum pressure of about 1 × 10 −4 Pa. The heating rate was about 2 K/s and the time at the set temperature was about 2 min. Figure 1a shows the deposition rate as a function of the discharge power for DC and RF (13.56 MHz) plasma excitation. For both discharge modes, the rate increases linearly with the discharge power, as expected [30], and is independent of the type of the reactive gas (O2, N2O, H2O). While the linear fit curves start at the origin of the x axis for DC excitation, the RF power where the deposition rate is larger than zero is shifted to a value of about 20 WRF. This is caused by the fact that the discharge voltages (ie., the target voltages) are much lower for RF plasma excitation compared to the DC discharge [30]. For RF powers lower than 20 W, the target voltage is lower than the sputtering threshold voltage [42]; thus, no deposition can take place. The deposition rates for RF sputtering are much lower compared to the DC deposition rates; the slopes of the linear rate(power) curves in Figure  1a are 0.24 and 0.66 nm/(min . W) for RF and DC excitation, respectively. This is caused by two effects:

Deposition Characteristics
• The lower discharge voltages leading to lower sputtering yields; • The fact that, in the RF case, the target is only at a sufficiently high negative potential every second half-wave, allowing an acceleration of positive Ar ions to the target surface [43]. Figure 1b displays the partial pressures of the reactive gases O2, N2O, H2O, where the deposited films exhibit the resistivity minimum as a function of the discharge power. It can be seen that more reactive gas is needed with increasing discharge power, i.e., increasing deposition rate. This can be explained by the fact that a certain amount of (reactive) oxygen per deposited amount of metal atoms is needed to prepare highly conductive SnO2:Ta films. This behaviour is also known for other TCO materials, for instance, for ZnO [44]. The amount of reactive gas needed to prepare highly conductive films depends strongly on the type of reactive gas, of the mode of plasma excitation and the deposition temperature. This is due to the different amounts of oxygen, supplied by the different reactive gases. Oxygen gas itself contains the highest relative amount of oxygen compared to N2O and H2O; thus, the lowest partial pressure is needed for pure O2. Although the molecules N2O and H2O contain the same relative amount of oxygen (one third), there is still a difference for both reactive gases which can be explained by the different chemical action of nitrogen (N2O) and hydrogen (H2O): while nitrogen is almost nonreactive, hydrogen is reducing the growing film. Therefore, in case of H2O as reactive gas, a higher amount of oxygen is needed in order to have the same oxidizing effect as N2O. One also has to take into account that the dissociation energies of the stable molecules O2 and H2O (≈ 5.1 eV [45]) are much higher in comparison to the metastable N2O molecule (Edis ≈ 1.6 eV [46]) which shifts the resistivity minimum for N2O nearer to that of O2. This is clearly visible if the necessary reactive gas partial pressure is plotted versus the deposition rate, as shown in Figure 1c. Besides the effects of the reactivity and composition of the reactive gas, the effect of the different degrees of gas dissociation in a DC and an RF discharge can also be derived from Figure 1c. For the case of DC plasma excitation, the films were deposited both at room temperature as well as at a substate temperature of 400 °C. Obviously, the reactivity of the gas (O2 in this case) is significantly improved at a higher substrate temperature, thus a lower oxygen partial pressure is required at a higher deposition temperature.

Phase Composition
The XRD patterns of the as-deposited films exhibit no diffraction peaks; this means that these films are X-ray amorphous at film thicknesses below about 200 nm. While thicker films (≈ 500 nm) deposited by DC magnetron sputtering are still X-ray amorphous, the thicker films, prepared by RF magntron sputtering, exhibit X-ray peaks, corresponding to the tetragonal (rutile-like) structure of cassiterite (SnO2) (powder diffraction file PDF ® _01_070_4175, The International Centre for Diffraction Data, Newton Square, USA).   While the X-ray amorphous DC-sputtered films had a constant resistivity of about 5 × 10 −3 Ωcm, the films which started to crystallize at larger thicknesses showed an increased resistivity for a thickness of about 500 nm (see below). The Raman spectra (not displayed here) of as-deposited SnO2:Ta films show only weak, broad features ascribed to the glass substrates. Since Raman spectroscopy is able to detect structural features/phases for crystallite sizes in the few nm range, i.e., smaller than is possible by XRD, we conclude that the SnO2:Ta films are really amorphous.

Resistivity
The reactive gas partial pressure dependence of the resistivity, the carrier concentration and the Hall mobility of the SnO2:Ta films are displayed in Figure 3. While the discharge voltage and the deposition rate are nearly constant in the investigated partial pressure range, the resistivity exhibits a strong decrease with increasing oxygen partial pressure. A resistivity minimum is found at an oxygen partial pressure of about 2 × 10 −2 Pa; further increasing the O2 partial pressure leads to a slight increase in the resistivity. The resistivity minimum comes about by a local maximum of the carrier concentration and a maximum Hall mobility at the oxygen partial pressure where the minimum resistivity is obtained.
The effects of the different plasma excitation modes (DC or RF, 13.56 MHz) on film resistivity are shown in Figure 4a. It can be seen that at the same deposition rate, realized by different discharge powers for DC (25 W) and RF excitation (100 W), the O2-partial pressure where the minimum resistivity is obtained is significantly lower for RF-, compared to DC-excitation. This, we ascribe to the much higher density of reactive oxygen species (O, O*, O + ) in the RF discharge, which was measured already in 1998 by us for the reactive magnetron sputter deposition of ZnO:Al films [47]. The minimum resistivity is about the same for both excitation modes (ρmin ≈ 5 × 10 −3 Ωcm); however, the width of the minimum is significantly wider in case of DC plasma excitation. Figure 4b shows the reactive gas partial pressure dependence of the resistivity for the three reactive (oxygen-containing) gases O2, N2O and H2O. For all three gases, a resistivity minimum exists with nearly the same value. The position and the width of the ρ(preactive) curves are clearly different for the three oxygen-containing reactive gases. This is plausible, taking into account the different molar concentrations of oxygen in the different molecules-see also Section 3.1. Using H2O as a reactive gas, the resistivity minimum occurs at the highest partial pressure, although H2O contains the same relative amount of oxygen as N2O. This can be explained by the reductive effect of hydrogen, thus requiring a higher partial pressure to have the same oxidizing effect, as in the case of N2O. An advantage of H2O as reactive gas is the wider resistivity minimum, making it easier to adjust the reactive gas partial pressure. In order to analyze the carrier transport process, the Hall mobility is displayed as a function of the carrier concentration, see Figure 5a−c. It can be seen that, independently of the used reactive gases, the mobility first increases steeply with increasing reactive gas partial pressure, reaching maximum values of about 15-20 cm 2 /Vs at carrier concentrations of about 1 × 10 20 cm −3 . The strong increase in the electron mobility with increasing reactive gas partial pressure at a nearly constant carrier concentration (see the arrows in Figure 5) can be tentatively explained by a higher defect density (Sn or SnO inclusions) at low reactive gas partial pressure. These defects act as scattering centers, thus reducing the carrier mobility. A theory for the explanation of this steep µ(N) dependence has yet to be developed.
Further increasing preactive leads to a gradually decreasing mobility, less for O2, and stronger for N2O and H2O. The carrier concentration decreases much more, by nearly two orders of magnitude. This behaviour is similar to that of other amorphous semiconducting oxide films, for instance, InGaZnO4 [37,48]. According to Nomura et al., the electronic transport in these amorphous oxides is governed by a percolation conduction over a distribution of potential barriers at the conduction band edge. Other than in single-or polycrystalline semiconductors, the band edges are no longer straight over larger volumes of material, i.e., the amorphous oxides exhibit small potential barriers on the nanoscale, caused by the randomness of the amorphous material in general and by the random distribution of defects (TaSn, Oi, VO etc.) especially.
The maximum carrier concentration is relatively low, compared to other TCO materials (ITO, ZnO). Taking into account the concentration of Ta in the SnO2:Ta films (Ta/Sn ≈ 0.5 at%), the electrical activation of Ta, assuming that Ta on tin lattice sites (TaSn) acts as a donor, is quite low, in the order of 22 %, using the molecular number density of SnO2 of 3.125 × 10 22 molecules/cm 3

Film Composition (RBS)
The film composition was determined by Rutherford back-scattering analysis (RBS). To this purpose, the SnO2:Ta films were deposited onto glassy carbon (Sigradur) substrates which produce, due to the low atom mass of carbon, a very low background for the backscattering peaks of oxygen, tin and tantalum, thus allowing an exact determination of the average film composition [51].
Typical RBS spectra of two SnO2:Ta films on logarithmic scale are shown in Figure 6a. Besides the edge of the carbon substrate, the peaks for O, Sn and the shoulder for the dopant Ta are clearly visible. At channel numbers around 600, a small peak is visible that belongs to argon. It is caused by two effects: Ar is implanted into the glassy carbon during the surface cleaning of the substrates by the Ar-RF plasma treatment (narrow peak at channel 590); (ii) Ar is included (buried) into the growing film during the deposition of the SnO2:Ta film. This is caused both by the continuous coverage of the film surface by neutral argon atoms from the sputtering atmosphere as well as the bombardment of the film surface by energetic Ar ions, atoms and metastable atoms [52,53] which are then covered by the film atoms.
The concentration of argon in the films is about 0.4 at% for films deposited at room temperature. For the SnO2:Ta films deposited at a high temperature (400 °C), the Ar content is much lower (≈ 0.1 at%), caused by the higher desorption of the Ar atoms during the film growth [54,55].
By integrating the peaks for O, Sn and Ta, the average composition of the films was determined, displayed in Figure 6b as a function of the oxygen partial pressure. It is clearly visible that the dopant (Ta) concentration does not depend on the reactive gas partial pressure and the type of gas. It is nearly constant at a concentration of 1.5 at%. This means that the observed variation in the electrical parameters as a function of the reactive gas pressure is not caused by a variation in the chemical dopant amount, but by a varying electrical activation of the dopant Ta and/or by compensation of the electrons by acceptor-like defects-see, for instance, [49,50]. The oxygen-to-metal ratio is almost constant within a measurement accuracy of about ≈ 2 at%. Since we observe a strong variation in the electrical parameters when varying the reactive gas partial pressure, this has to be caused by an oxygen concentration variation which cannot be detected by RBS.

Thickness Dependence
Most of the deposited SnO2:Ta films had thicknesses of about 100 nm. For applications where low sheet resistances are needed, thicker films have to be deposited. Therefore, the thickness dependence of the electrical film properties was investigated. Figure 7a shows the dependence of deposition rate, resistivity, carrier concentration and Hall mobility on the film thickness up to about 450 nm. As expected, the deposition rate is constant. In contrast to what is known for other TCO materials (see Figure 7b, [56]), the resistivity increases significantly with increasing film thickness, caused mainly by the decrease in the carrier concentration. A similar behaviour was observed by Minami et al. and by Brousseau et al. for undoped SnO2 films, also deposited by RF magnetron sputtering [57,58].
The well-known behaviour of a decreasing resistivity with increasing film thickness can be explained by an increasing grain size and the decreasing effect of surface and interface carrier scattering, described by the Fuchs-Sondheimer theory, see [56,59,60].  [56].
To explain the contrary behaviour of our SnO2:Ta films, one has to take into account that the films with larger film thicknesses (> 200 nm) are no longer X-ray amorphous, i.e., these films are, at least partly, polycrystalline with grain boundaries. Figure 8 shows a TEM cross-sectional picture of a 500 nm thick SnO2:Ta film, deposited by RF magnetron sputtering. It can clearly be seen that the film is composed of an amorphous part, directly grown on the glass substrate. After a certain thickness is reached, varying locally, a spontaneous crystallization sets in and the SnO2:Ta growth is transformed to polycrystalline growth. A similar behaviour was observed by us earlier for the growth of indium-tin oxide (ITO) films by reactive magnetron sputtering from a metallic InSn10wt% target [61]. In this case, the films were deposited onto intentionally unheated substrates and the increase in the substrate temperature occurred during the film growth.
The crystallization of the thicker SnO2:Ta films is due to the increasing substrate temperature during the deposition, caused by the total energy flux to the growing film [62,63]. Based on the measured energy flux, it is estimated that the substrate temperature reaches values > 100 °C. That the increased substrate temperature causes the film crystallization was proven by depositing some films stepwise, making a pause for cooling down every 100 nm. Thick films (500 nm) deposited in this way were X-ray amophous and had resistivity values comparable to thin films.

Radial Profiles
The radial profiles of the electrical film properties were analyzed for DC and RF plasma excitation. Such profiles are an important aspect for magnetron sputtering deposition onto a stationary substrate, since the plasma/ion assistance of the film growth varies radially significantly, caused by the plasma torus which is formed in front of the target [18]. For the deposition of TCO films it is often reported that radial profiles of the resistivity exhibit pronounced maxima or minima opposite the erosion groove of the target which are formed due to the torus-like shape of the magnetron plasma-see, for instance, [18,[64][65][66]. Figure 9 displays the radial distributions of the electrical parameters and the film thickness for DC and RF plasma excitation. While the radial thickness distributions for both excitation modes are almost the same, exhibiting the expected bell-shaped curve, the resistivity distributions in the asdeposited state are clearly different for DC and RF plasma excitation. While the resistivity of the RFdeposited films is only slightly varying over the diameter of the substrate, the resistivity distribution of the DC-deposited films shows a significant increase in ρ towards the border of the substrate by a factor of about 3. The regions, where the resistivity is high, coincide with the radial position of the erosion groove of the target. This points to the effect of a high-energetic negative ion bombardment of the growing film, as discussed recently by us [17,18]. It has become known in recent years that, in sputtering processes with electronegative elements (for instance O, F, Cl, S, Se, but also Au, Pt, Ag, Sn, Bi-see reference [67] for the electron affinity values) negative ions are accelerated away from the negatively biased target surface towards the substrate, i.e., the growing film. Due to the high discharge/target voltages, even in the modern magnetron sputtering systems, the energies of these negative ions are in the range of some 100 eV, which therefore leads to the generation of defects in the growing film. In the beginning of the sputtering technology, before the invention of the magnetron sputtering sources, the discharge voltages were much higher (some kV) which led to a very high-energy bombardment of the growing films, sometimes ending up not in film deposition but in film and substrate etching [68]. Short-term post-anneling of the SnO2:Ta films at 500 °C in the hydrogen atmosphere of the dc-sputtered films reduces their resistivities even in the region opposite the target groove, while the resistivity of rfsputtered films was nearly unaffected.

Transmittance and Reflectance
The transmittance and reflectance spectra of SnO2:Ta films are shown exemplarily in Figure  10a,b. With increasing N2O partial pressure, the amount of non-oxidized Sn in the films is reduced and the films become more transparent in the visible spectral range. The absorption edge shifts to smaller wavelengths. This means that the optical band gap energy is increased, caused by the Burstein-Moss effect [69,70]. From the transmittance and reflectance curves (see Figure 10a,b), the optical constants n and k have been calculated. The extinction coefficient k of three SnO2:Ta films is shown in Figure 10c as a function of the wavelength. These films were prepared with the three reactive gases O, N2O and H2O under conditions that yielded minimum resistivities of about 4.7 × 10 −3 Ωcm. While the films deposited with O or N2O exhibit a minimum k value of about 7 × 10 −3 in the visible spectral range, the films deposited in Ar+H2O show a much lower extinction coefficient, lower than 1 × 10 −3 . A tentative explanation of this striking effect is the passivation of deep defects in the band gap of the SnO2:Ta films by the hydrogen from the sputtering atmosphere when using H2O as a reactive sputtering gas.

Conclusions
Conductive and transparent SnO2:Ta films were prepared at low substrate temperatures (< 100 °C) by reactive magnetron sputtering from a ceramic target in various gas mixtures: Ar/O2/(N2O, H2O). The width of the process window with respect to the reactive gas partial pressure depends on the type of the reactive gas; it is wider for N2O and H2O, mostly due to the dilution of the oxygen content in the compound gases. The SnO2:Ta films are X-ray amorphous in the as-deposited state. By heating the films to temperatures above 500 °C, the films start to crystallize, accompanied by an increased resistivity. For RF deposition (larger energy flux to the growing film) at larger film thicknesses (d >≈ 150 nm), i.e., longer deposition times, crystallization occurs during the deposition, caused by the substrate temperature increase due to the energy influx from the condensing film species and the plasma. While the amorphous films are remarkably conductive (ρ ≈ 5 × 10 −3 Ωcm), the crystallized films exhibit higher resistivities due to grain-boundary-limited conduction. The best amorphous SnO2:Ta films had a resistivity of better than 4 x 10 −3 Ωcm with a carrier concentration of 1.1 × 10 20 cm −3 , and a Hall mobility of 16 cm 2 /Vs. The sheet resistance was about 400 Ω/ for 100 nm films and 80 Ω/ for 500 nm thick films. The average optical transmittance from 500 to 1000 nm is greater than 76 % for 100 nm films, where the films, deposited with H2O as a reactive gas, exhibit a slightly higher transmittance of 80 %.
The crystallization of the SnO2:Ta films is not only detrimental for the electrical transport properties of the films; grain boundaries are also diffusion paths for atoms and molecules in gaseous or liquid environments. Since these films were deposited at low temperatures, they are amorphous, thus improving the resistivity against degradation and etching.
These X-ray amorpous SnO2:Ta films can be used as low-temperature transparent and conductive protection layers, for instance to protect semiconducting photoelectrodes for water splitting, and also, where appropriate, in combination with more conductive TCO films (ITO or ZnO). Funding: This research received no external funding. Please add: "This research received no external funding" or "This research was funded by NAME OF FUNDER, grant number XXX" and "The APC was funded by XXX". Check carefully that the details given are accurate and use the standard spelling of funding agency names at https://search.crossref.org/funding, any errors may affect your future funding.