Limits of carrier mobility in Sb-doped SnO 2 conducting ﬁlms deposited by reactive sputtering

Electron transport in Sb-doped SnO 2 (ATO) ﬁlms is studied to unveil the limited carrier mobility observed in sputtered ﬁlms as compared to other deposition methods. Transparent and conductive ATO layers are deposited from metallic tin targets alloyed with antimony in oxygen atmosphere optimized for reactive sputtering. The carrier mobility decreases from 24 cm 2 V − 1 s − 1 to 6 cm 2 V − 1 s − 1 when increasing the doping level from 0 to 7 at. %, and the lowest resistivity of 1 . 8 × 10 − 3 Ω cm corresponding to the mobility of 12 cm 2 V − 1 s − 1 which is obtained for the 3 at. % Sb-doped ATO. Temperature-dependent Hall e ﬀ ect measurements and near-infrared reﬂectance measurements reveal that the carrier mobility in sputtered ATO is limited by ingrain scattering. In contrast, the mobility of unintentionally doped SnO 2 ﬁlms is determined mostly by the grain boundary scattering. Both limitations should arise from the sputtering process itself, which su ﬀ ers from the high-energy-ion bombard-ment and yields polycrystalline ﬁlms with small grain size. C 2015 Author(s). All article content, except where otherwise noted, is licensed under a Creative Commons Attribution 3.0 License. [http: dx.doi.org

Electron transport in Sb-doped SnO 2 (ATO) films is studied to unveil the limited carrier mobility observed in sputtered films as compared to other deposition methods. Transparent and conductive ATO layers are deposited from metallic tin targets alloyed with antimony in oxygen atmosphere optimized for reactive sputtering. The carrier mobility decreases from 24 cm 2 V −1 s −1 to 6 cm 2 V −1 s −1 when increasing the doping level from 0 to 7 at. %, and the lowest resistivity of 1.8 × 10 −3 Ω cm corresponding to the mobility of 12 cm 2 V −1 s −1 which is obtained for the 3 at. % Sb-doped ATO. Temperature-dependent Hall effect measurements and near-infrared reflectance measurements reveal that the carrier mobility in sputtered ATO is limited by ingrain scattering. In contrast, the mobility of unintentionally doped SnO 2 films is determined mostly by the grain boundary scattering. Both limitations should arise from the sputtering process itself, which suffers from the high-energy-ion bombardment and yields polycrystalline films with small grain size. C 2015 Author(s). All article content, except where otherwise noted, is licensed under a Creative Commons Attribution 3.0 Unported License. [http://dx.doi.org/10.1063/1.4916586] Doped SnO 2 is an important transparent conductive oxide (TCO) that combines high chemical resistance and excellent thermal stability and has a high work function of 4.9 eV and a hardness of 6.5 Mohs. 1,2 In contrast to other TCOs, such as In 2 O 3 :Sn or doped ZnO that are deposited at large scale by magnetron sputtering, high quality SnO 2 , and specifically F-doped SnO 2 (FTO) for low-emissivity (low-e) window coatings, is deposited by spray pyrolysis directly onto the hot glass at temperatures of 400-600 • C. The need of high temperature restricts the deposition of FTO as top electrodes onto other functional layers for various applications such as photovoltaics or multilayer low-e coatings. Sputtering of FTO 3,4 and other alternatives such as Sb-doped SnO 2 (ATO) 3,5-9 or Ta-doped SnO 2 (TTO) 10 have been attempted at lower temperatures but could not reach resistivity values equal to those for spray pyrolysis or chemical vapor deposition (CVD).
Antimony is used as an effective dopant because the similar radii of Sb 5+ and Sn 4+ allow an efficient incorporation of Sb 5+ as n-type dopant with the doping efficiency as high as 100% in the case of molecular beam epitaxy (MBE) deposition. 11 The main drawback of sputtered ATO is the low mobility on the order of 10 cm 2 V −1 s −1 as compared to other deposition techniques which can yield films with mobilities above 20 cm 2 V −1 s −1 (Refs. 5 and 7) and even up to 100 cm 2 V −1 s −1 for epitaxial films. 11 Table I compares Hall mobility µ and carrier density N for selected highquality ATO films obtained by various deposition techniques. Only optically transparent (transmittance > 80%) polycrystalline films with the lowest reported electrical resistivity ρ are considered. The film thickness is comparable for all methods except MBE, which allows us to conduct a TABLE I. Comparison of the electronic parameters ρ, N , and µ of selected ATO films prepared using different deposition techniques. The Sb dopant concentration c (in cm −3 ) is calculated from the Sb content Sb/(Sb + Sn) (in at. %) assuming the SnO 2 density of 6.99 g cm −3 . 12 All values refer to reported optically transparent samples with the lowest electrical resistivity.

Method
Temperature comparative analysis. The thicker MBE-grown ATO film can also be considered since it was stated that "electron concentration, mobility, and resistivity were independent of thickness." 11 One can see that for similar N, sputtered ATO features lower µ values as compared to films grown by MBE, spray pyrolysis, or CVD. Here, we attempt to maximize µ in optically transparent ATO films while using industrially relevant reactive sputtering. With a combination of temperature-dependent Hall effect measurements, X-ray diffraction (XRD) and near-infrared (NIR) spectroscopy, we determine the mechanism responsible for mobility limitations in sputtered ATO and discuss possible origins of them. Antimony doped tin targets (2 ′′ , purity 99.999%, Plasmaterials) with varying dopant content (0, 1, 3, 5, and 7 at. % Sb relative to Sn) were reactively sputtered onto 1 mm thick soda-lime glass substrates in an ultra-high vacuum system (AJA Intl.) equipped with unbalanced magnetrons. The substrate temperature of 450 • C was chosen to ensure a high adatom mobility comparable to other high-temperature methods from Table I. The films were deposited with a DC power density of 3.45 W cm −2 , a discharge voltage of 390 V, and a target to substrate distance of 12 cm in a reactive Ar + O 2 atmosphere (gas purity 99.995%) at pressure p tot = 5 × 10 −3 mbar, which resulted in a growth rate of ∼12 nm min −1 . In order to obtain transparent and conductive layers, reactive sputtering of Sn demands an accurate control of the oxygen flow. 4 Therefore, for each sample in this study, a prior optimization was performed by depositing a series of thin films at various oxygen flows from 0 to 25 sccm (pp O2 ∼ 1 × 10 −3 mbar) at 450 • C in order to identify the resistivity minimum. All films had a comparable thickness of 230 ± 20 nm measured by stylus profilometry. The Sb content in the films was assumed to be the same as in the sputter target. Grazing incidence XRD (GI-XRD) measurements at θ = 1.5 • were performed on a Bruker D5000 diffractometer operating with CuK α radiation source. All films were polycrystalline with the cassiterite crystal structure. An Ecopia HMS 3000 system was used for ambient Hall measurements, while temperature-dependent Hall investigations (100 K-300 K) were carried out in a Physical Property Measurement System (PPMS) from Quantum Design. In both cases, the samples were contacted in the Van der Pauw geometry. A Shimadzu UV-3600 spectrometer with integrating sphere was used to measure transmittance and reflectance in the wavelength range from 300 to 1400 nm. NIR reflectance measurements were performed on a Nicolet 8700 Fourier transform infrared (FTIR) spectrometer (thermo-scientific) in the wavelength range from 1000 to 20 000 nm. All transmittance and reflectance data shown refer to the entire ATO/glass stack, which was taken into account when fitting the NIR spectra. Fig. 1 compiles the effects of the Sb dopant concentration on optical and electrical parameters. The electrical resistivity reaches its minimum at 3 at. % Sb. The carrier density N increases monotonously within the 10 20 cm −3 range with increasing the Sb content, although the dopant activation efficiency N/c can be calculated to decrease from 60% to 20% indicating a saturation for heavy doping levels. In the case of unintentionally doped (UID) SnO 2 , the carrier density below 10 19 cm −3 is reported to stem from the oxygen vacancies V O formed in the oxygen-deficient sputtering environment 1 or hydrogen doping. 13 Importantly, the room temperature Hall mobility µ decreases from ∼24 cm 2 V −1 s −1 to 6 cm 2 V −1 s −1 when increasing doping.  The absorptance increases with the doping level, which can be explained by a free carrier absorption α ∝ N/µ and increased plasma frequency ω p as described by the Drude model. 14,15 The UID layer does not show any significant free carrier absorption in the given range because of the low carrier density N, and its UV cutoff is shifted to longer wavelength because the Burstein-Moss effect is less pronounced. 12 A conclusion can be drawn from Fig. 1 that it is difficult to achieve simultaneously a high optical transparency and a low electrical resistivity ρ because the low carrier mobility µ in sputtered ATO necessitates high doping levels for achieving a sufficiently low ρ. 15 In order to understand how extrinsic doping influences the electron mobility in sputtered ATO, the mechanism of electron scattering was studied for different doping levels. In-grain and grain boundary scattering are two main processes contributing to the total resistance of a polycrystalline film, 16 is usually dominated by one of these mechanisms. Whereas µ g b is determined by the rate of thermionic emission over grain boundary potential barriers 17 or tunneling through them, 18 µ ig is limited by the scattering at ionized impurities, phonons, or neutral impurities. 16,19 To access the nature of the dominant scattering mechanism in sputtered ATO films, the electron mobility was probed by complimentary optical and electrical approaches. First, the optical mobility µ opt was determined by fitting NIR reflectance measurements with the classical Drude free carrier model. 20,21 The dielectric function was fitted in proximity to the reflectance edge with Γ and ω p as free parameters. The measurements and fits are shown in the inset of Fig. 2. From the damping frequency Γ, the optical mobility can be calculated as µ opt = e/Γm * assuming an effective electron mass of m * = 0.39m e and a relative permittivity ε ∞ = 3.5. 3 Since there is a wide spread of reported m * values, 3,22,23 error bars are added to the calculated optical mobility µ opt in order to represent possible deviations for the range of electron masses m * = 0.39 ± 0.15 m e . Note that the relative permittivity does not influence the calculation of µ opt . The calculated µ opt reflects the in-grain scattering because the NIR excitation is assumed to deflect free electrons only over a few nm, i.e., shorter than the grain size and without allowing them to cross grain boundaries. 24 In contrary, the mobility µ Hall obtained from Hall effect measurements accounts for both intra-grain and grain boundary scattering since electrons are transported through the layer across a large number of grains. Fig. 2 displays room-temperature µ opt and µ Hall for different doping levels. For a wide range of m * , the similar mobility values for all doped samples suggest that in-grain scattering processes limit the carrier mobility. For the UID case, µ opt is higher than µ Hall indicating that the grain boundary scattering prevails.
In order to confirm these findings, the Hall mobility µ Hall was studied as a function of temperature. Fig. 3 displays N and µ Hall in the 100 K-300 K range for three selected samples: UID, 1, and 7 at.% Sb. In all three cases, the carrier density N is found to be constant as expected from degenerately doped semiconductors. 25 Also, the µ Hall show no significant temperature dependence for the doped samples with 1 at. % and 7 at. % Sb. This behavior is a typical signature of the FIG. 3. Carrier density N and mobility µ Hall for three samples with 0, 1, and 7 at. % Sb, estimated from the temperaturedependent Hall measurements in the range from 100 K to 300 K. The inset shows a re-plotted temperature dependency of mobility for the UID (0 at. % Sb) and the linear trend indicates an agreement with the Petritz model. 17 ingrain scattering at ionized or neutral impurities. 16,19 In contrary, the µ Hall of the UID SnO 2 sample increases from 14 cm 2 V −1 s −1 to ∼21 cm 2 V −1 s −1 with increasing temperature. Such thermally activated mobility can be explained by the Petritz model which describes thermionic emission of electrons across grain boundaries. 17 Here, T denotes temperature, k is the Boltzmann constant, and ϕ b is the energetic height of the intergrain barrier. The inset in Fig. 3 shows the temperature dependance of mobility in UID rescaled according to this model. The linear fit indicates a good agreement to the model over a wide temperature range, and an activation energy of 13 meV can be extracted. Such potential barriers arise from the charged trap states at grain boundaries that induce band bending. 26 With decreasing N, the free carrier shielding decreases, i.e., the potential barrier height ϕ b increases and the mobility is reduced. 16 There could be several origins of the limited electron mobility in sputtered ATO films. In order to rule out the effect of Sb doping on the crystalline quality, grazing incident XRD measurements were performed. Full-width-at-half-maximum (FWHM) values for [101] and [211] reflexes did not reveal any significant influence of Sb concentration on the film crystallinity. An average crystal coherence length of around 15 nm was calculated from Scherrer's formula 28 for all samples regardless the Sb content. Therefore, it is unlikely that the Sb doping reduces the mobility by significant changes in the film morphology, and there must be an another reason why the electron µ in sputtered ATO is a factor of 2-3 lower than for MBE, spray pyrolysis, or CVD at comparable doping concentrations (Table I) limited mobility, but certain ingrain defects inherent to the sputtering process must also be involved. White et al. 11 reported a 100% doping efficiency for the doping range ≤ 1 at. % Sb, whereas in our sputtering case, the doping efficiency was ∼60% for 1 at. % Sb (Table I). The remaining inactive dopants can induce scattering centers that further reduce the electron mobility. A second possible reason for the reduced mobility in sputtered SnO 2 was suggested by Ellmer and Welzel, 29 who observed a higher flux of high-energetic ions during the reactive sputtering of SnO 2 as compared to other metal oxide. It is quite plausible that the high flux of negative oxygen ions emitted from the Sn target impinges on the growing films and causes the ingrain scattering defects. As such ions are inherent to the sputtering process, this can explain the reduced mobility in comparison to other deposition techniques. A third possible reason could be grain boundary potential barriers induced by the chemisorbed oxygen-species from the oxygen plasma environment. 27 In order to desorb these oxygen species, we irradiated sputtered ATO samples with the UV radiation for different times (250 W iron-doped Hg lamp, 50 mW cm −2 UVA). No changes in mobility could be detected, which is an additional indication for the ingrain nature of the mobility limiting defects. Inhomogeneous dopant distribution, e.g., by formation of Sb oxide phases 30 or dopant clustering 31 is another model that could account for reduced mobility and relatively low dopant activation. Despite the fact that GI-XRD measurements did not show such secondary phases, we cannot strictly exclude such effects here.
In conclusion, the mobility of sputtered ATO is found to be a factor of 2-3 lower than for other deposition techniques, despite the high deposition temperature and a careful optimization of the reactive gas flow. A minimum resistivity of 1.8 × 10 −3 Ω cm is obtained for a doping level of 3 at. % Sb with N = 3 × 10 20 cm −3 , µ = 12 cm 2 V −1 s −1 , and an average absorptance of 4% in the visible spectral range. For heavy doped layers (N > 10 20 cm −3 ), the in-grain scattering is dominant, whereas the UID SnO 2 layer exhibits a thermally activated mobility indicating the grain-boundary limited carrier transport. The high degree of inactive dopants and/or the influence of high energetic O − ions impingement are possible origins of the low mobility in sputtered ATO. Rising the deposition temperature (as demonstrated for TTO 10 ) or the post-deposition annealing (as demonstrated for FTO 4 ) could eventually improve mobility but present additional technologic challenges. Innovative concepts such as the magnetic lenses for deflecting energetic negative ions 32 or the chemically activated sputtering to improve the doping efficiency should be explored to improve carrier mobility in sputtered ATO.