New Route for "Cold-Passivation" of Defects in Tin-Based Oxides.

Transparent conductive oxides (TCOs) are essential in technologies coupling light and electricity. For Sn-based TCOs, oxygen deficiencies and undercoordinated Sn atoms result in an extended density of states below the conduction band edge. Although shallow states provide free carriers necessary for electrical conductivity, deeper states inside the band gap are detrimental to transparency. In zinc tin oxide (ZTO), the overall optoelectronic properties can be improved by defect passivation via annealing at high temperatures. Yet, the high thermal budget associated with such treatment is incompatible with many applications. Here, we demonstrate an alternative, low-temperature passivation method, which relies on cosputtering Sn-based TCOs with silicon dioxide (SiO2). Using amorphous ZTO and amorphous/polycrystalline tin dioxide (SnO2) as representative cases, we demonstrate through optoelectronic characterization and density functional theory simulations that the SiO2 contribution is twofold. First, oxygen from SiO2 passivates the oxygen deficiencies that form deep defects in SnO2 and ZTO. Second, the ionization energy of the remaining deep defect centers is lowered by the presence of silicon atoms. Remarkably, we find that these ionized states do not contribute to sub-gap absorptance. This simple passivation scheme significantly improves the optical properties without affecting the electrical conductivity, hence overcoming the known transparency-conductivity trade-off in Sn-based TCOs.

Introduction

Sn-based oxides are wide band gap semiconductors of high technological importance, with applications ranging from smart windows to batteries and solar cells. The optoelectronic properties of tin dioxide (SnO2) can be tuned over a wide range of conductivity and transparency and, hence, adapted to the requirements of each of these technologies. For example, reported electrical conductivity values span from[1] ∼10–6 to[2] 104 S cm–1. This is achieved by tuning the density of oxygen deficiencies (VO) or by adding dopants such as fluorine, antimony, barium, or molybdenum.[3−7] Other elements may be added to transparent conductive oxides (TCOs) to modify, for example, their microstructure and thermal stability.[8] In this regard, the addition of zinc to SnO2 [zinc tin oxide (ZTO)] yields thin films with attractive properties such as a total transmittance higher than 75% in the visible and near-infrared spectral range[9−11] and an amorphous microstructure when deposited at room temperature, which is preferable for applications in flexible and organic devices.[12] Furthermore, this microstructure remains stable up to temperatures as high as 550 °C.[11,13] Because of these properties, ZTO has already been applied as a transparent contact in organic light-emitting diodes,[12,14] as a channel in thin film transistors,[15,16] and as a recombination layer in silicon-perovskite tandem solar cells.[17] Theoretical and experimental evidence suggests that the presence of oxygen deficiencies in ZTO creates both shallow and deep sub-gap states, with the latter acting as absorption centers in the visible part of the spectrum.[11,18,19] Even though these defects can be passivated by postdeposition treatments in air at temperatures >400 °C,[11] annealing in these conditions is thermally costly and/or not convenient for devices with low thermal budgets, such as solar cells based on thin hydrogenated amorphous silicon layers or hybrid organic–inorganic perovskite materials.[20−22]

Alternatively, previous investigations have shown that the codeposition of silicon dioxide (SiO2) with different TCOs, mainly with zinc oxide, may decrease the density of VO defects[23−26] but also lower the refractive index,[23,27] decrease the resistivity,[27−29] and amorphize the TCO.[28,30,31] For the case of Si in Sn-based TCOs, Kang and co-workers[26] used first-principle calculations to suggest that silicon atoms alter the coordination number of Sn, leading to an increase in the formation energy of VO deficiencies. Yet, this passivation mechanism leads to a strong decrease in electrical conductivity as these deficiencies are the source of free carriers. Furthermore, it was recently proposed that Si modifies the band gap of ZTO, resulting in improved TFT performance.[32] However, the role of Si in the sub-gap structure of ZTO was not fully clarified at the atomistic level in this study. In contrast to previous reports, here we combine experimental and computational techniques to explain the effect of Si on the optoelectronic properties of SnO2-based materials. We demonstrate that adding SiO2 during deposition of Sn-based TCOs (using ZTO and SnO2, as case examples) results in a decrease in the sub-gap absorption while keeping electrical properties unchanged. By combining these experimental results with density functional theory (DFT) calculations, we find that, while the oxygen from SiO2 passivates deep sub-gap defects, the addition of Si decreases the ionization energy of VO and shifts the corresponding sub-gap defect states close to the conduction band minimum (CBM). Thanks to this effect, the defect no longer contributes to the formation of detrimental sub-gap absorption centers and provides free carriers.

Results

Trade-Off in Optoelectronic Properties of ZTO

Before optimizing ZTO by cosputtering with SiO2, the properties of ZTO films were studied as a function of O2 flow during deposition. As seen in Figure , ZTO films sputtered with a low O2 flow during deposition (1.5–2.5 sccm of Ar–O2) present low conductivity and high absorption in the measured spectral range. Initially, increasing the O2 content improves the film transparency, and its conductivity reaches a maximum of 456 S cm–1. Increasing the Ar–O2 flow above 3.0 sccm reduces the optical absorptance but at the expense of conductivity, which drops by 62%. A trade-off often observed in TCOs is reached: improving the optical properties worsens the electrical ones and vice-versa. As observed in Figure , optimizing the oxygen flow during deposition does not yield a film that combines a conductivity above 400 S cm–1 and a low absorptance at 500 nm (≤5%). Alternative approaches are, hence, required to control the amount of oxygen in the films and to ensure both high conductivity and transparency.

Figure 1

Absorptance of ZTO films as a function of the oxygen flow during deposition. The inset shows the change in conductivity for the same samples. All films were annealed at 200 °C for 30 min in air prior to the measurements.

Combinatorial Deposition of SiO2 and Sn-Based TCOs

To introduce oxygen into Sn-based TCOs in a precise manner, while avoiding high temperature steps,[11] ZTO or SnO2 was cosputtered with SiO2. In the following subsections, we describe in detail the optimization and characterization of ZTO with SiO2 (referred to as SiZTO). The ZTO film with highest conductivity (3.0 sccm of Ar–O2, a composition reported in refs (11) and (14)) will be used as a reference to assess the effectiveness of cosputtering deposition with SiO2. Experimental details of the optimization of SnO2 with SiO2 (SiSnO2) are described in Section I of the Supporting Information.

Reducing Sub-gap Absorption in ZTO Thin Films

We determined the optimal deposition conditions (regarding SiO2 content and Ar–O2 flow), by comparing a simplified figure of merit (FOM) of films sputtered under different conditions. The FOM was calculated as follows: , where σ is the electrical conductivity and A400–800 is the average absorptance from 400 to 800 nm. Therefore, a high FOM is indicative of films with high electrical conductivity and/or low absorptance in the visible spectral range. The SiZTO films with the highest FOM were deposited using 10 W (0.13 W cm–2) in the SiO2 target and 2.5 sccm of Ar–O2 (marked with dashed lines in Figure a). More information about the deposition details can be found in the Methods section. The evolution of the electrical properties of SiZTO with SiO2 content is shown in Figure b (all films deposited with an Ar–O2 flow of 2.5 sccm). The electron mobility increases from 22.2 cm2 V–1 s–1 up to a maximum of 26.8 cm2 V–1 s–1 when the power applied to the SiO2 target is increased from 0 to 10 W. For these powers, the free carrier density remains constant at 1 × 1020 cm–3. Further, increasing the SiO2 content makes the films less absorbing, but it also results in a decrease of free carrier density and mobility.

Figure 2

(a) Plot of the FOM as a function of deposition parameters. The FOM was calculated as the ratio of conductivity and average absorptance in the range of 400–800 nm for SiZTO films deposited with different SiO2 and O2 content; (b) Hall mobility and free carrier density of SiZTO as a function of power applied on the SiO2 target; these films had a constant Ar–O2 flow of 2.5 sccm.

To highlight the effect of adding SiO2 to ZTO, the conductivity and absorptance of the optimized SiZTO and the ZTO reference are compared in Figure . It is worth noting that only a slight difference in conductivity between ZTO and SiZTO is observed (208 and 192 S cm–1 for as-deposited films and 454 vs 429 S cm–1 after a mild annealing at 200 °C), with the clear advantage of SiZTO presenting less absorptance than ZTO in this wavelength range. Indeed, at a wavelength of 500 nm, ZTO has a 5.5% absorptance, whereas SiZTO has an absorptance of only 2.5%. At wavelengths above 1000 nm, SiZTO exhibits an absorptance below 5% (Figure SI2).

Figure 3

Absorptance and conductivity (inset) of optimized SiZTO (10 W in the SiO2 target and 2.5 sccm of Ar–O2) and ZTO as-deposited and after annealing at 200 °C. Although both films show virtually equal conductivities, SiZTO presents a lower absorptance below the band gap when compared to the reference ZTO.

SiZTO Microstructure and Composition

Nanobeam diffraction patterns of the optimized SiZTO films (10 W in the SiO2 target and 2.5 sccm of Ar–O2) indicate an amorphous microstructure (Figure a), analogous to that of ZTO.[11] A scanning transmission electron microscopy (STEM) high-angle annular dark-field (HAADF) image and an energy-dispersive X-ray spectroscopy (EDX) analysis of the cross-section of the optimized SiZTO film (deposited on sapphire) are shown in Figure b,c. The HAADF image of the cross section of the sample indicates a dense and homogeneous microstructure (Figure c), whereas the EDX line profiles demonstrate that the distribution of elements is uniform within the amorphous film. A slight Si accumulation is measured at the top of the film because the SiO2 target shutter was closed slightly after one of the ZTO.

Figure 4

(a) Nanobeam electron diffraction patterns taken along the growth direction of SiZTO thin films. The asymmetric speckles indicate an amorphous structure, unchanged with SiO2 addition[11] and along the growth axis; (b) STEM HAADF image of the cross section of the film (left panel) corresponding to Si K edge EDX map (right panel); (c) at. % line profiles (left to right) of the Si K, O K, Sn L, and Zn K edges quantified using the FEI Velox software (assuming a sample thickness of 100 nm and a density of 6.5 g/cm3 for the absorption correction); (d) nanobeam electron diffraction taken along the growth direction of SiSnO2, showing an increased crystallinity toward the end of the film (arrowheads); (e) STEM HAADF image of the cross section of the film and Si EDX map; (f) quantified line profiles of the elements of interest.

The composition of the optimized SiZTO film determined by Rutherford backscattering (RBS) is Si0.02Zn0.04Sn0.27O0.67, indicating an absolute increase in oxygen concentration of 2 at. % when compared to the reference ZTO (Zn0.05Sn0.30O0.65). EDX and RBS yield a similar composition. In addition, thermal desorption spectroscopy demonstrates that, when heating the films up to 700 °C, the total oxygen desorption of SiZTO is higher than that for ZTO (area under the curve in Figure SI4 of the Supporting Information). Furthermore, the onset of effusion of Zn is postponed to higher temperatures when Si atoms are present in the film. These results suggest that the film decomposition may be postponed to higher temperatures with the addition of Si. Similar to the presence of Zn within the amorphous SnO2, the smaller Si atoms could induce local strain in the amorphous network postponing decomposition.[13]

X-ray photoelectron spectroscopy (XPS) was performed on the optimized ZTO and SiZTO films (before and after annealing) to evaluate the possible changes in the oxidation state of the elements present when adding SiO2 and/or after annealing, which could explain the observed changes in optoelectronic properties. For O 1s, Sn 3d, and Zn 2p bands, a pseudo-Voigt function was fitted to the data to calculate the binding energies and the full width at half maximum. No important differences in the fitted values are seen between the measured samples (Supporting Information Figure SI5). In addition, no signal above the background was observed for the Si 2p peak at the corresponding binding energies, which indicates that the Si content is below the detection limit in these experiments.[33,34]

Addition of SiO2 to SnO2: General Procedure for Sn-Based TCOs

To test the universality of adding SiO2 to improve the optoelectronic properties of Sn-based TCOs, SiO2 was cosputtered this time with pure SnO2. Details about the combinatorial sputtering of SiSnO2 and microstructure of SnO2 are described in Section I of the Supporting Information. A detailed overview of the microstructure of SiSnO2, described by transmission electron microscopy (TEM), is shown in Figure d–f. The section of the SiSnO2 film in contact with the substrate is amorphous, however, as the material thickens, it crystallizes into rutile c-SnO2 structure. Nanocrystallites are formed halfway through the 150 nm thick film. A composition of Sn0.38O0.62 is obtained by EDX before SiO2 addition. For SiZTO, EDX indicates that Si is homogenously distributed at an average value of 3 at. % within the films, whereas the oxygen content increases slightly to 63 at. %. Furthermore, Si atoms do not accumulate at grain boundaries or inside the bulk (amorphous or crystalline) of SiSnO2 (see Si map in Figure e). No Si-rich clusters are observed, particularly toward the top of the film, where the film is composed of small crystallites (Figure d). As seen in Figure , the conductivity of the as-deposited and annealed SnO2 drops slightly when adding 3 at. % of Si, whereas the absorptance in the visible and near-infrared regions decreases simultaneously (from 6 to 3% at 500 nm). Hall effect measurements indicate a free carrier density of 1.75 × 1020 cm–3 for SnO2 and 1.26 × 1020 cm–3 for SiSnO2, and mobilities of 28.2 and 25.5 cm2 V–1 s–1 for SnO2 and SiSnO2, respectively. Notably, the SnO2 film contains both amorphous and polycrystalline regions (Supporting Information Section I) demonstrating that the addition of SiO2 passivates the sub-gap defects in amorphous and mixed-phase amorphous/polycrystalline thin films. In addition, the presence of Si-atoms in SnO2 retards the onset of crystallization of the films: grains start to appear closer to the top surface in SiSnO2 when compared to SnO2. A similar effect has been previously reported for Zn-modification of SnO2.[13] Finally, the presence/lack of Zn does not appear to modify the passivation mechanism.

Figure 5

Absorptance and conductivity (inset) of as-deposited and annealed SiO2–SnO2 (SiSnO2) and SnO2 films. The films were deposited at the optimal conditions. As for ZTO, both films show similar conductivities. The main advantage of SiSnO2 films is their lower absorption in the visible range.

Computational Assessment of Si Modification to SnO2

The addition of a small amount of SiO2 does not modify the microstructure of ZTO, which remains amorphous, yet improves the optical properties of the film. The gain in optical properties occurs irrespective of whether the microstructure is fully amorphous (SiZTO) or an amorphous/polycrystalline mixture (SiSnO2). Moreover, both Si and O are found by EDX to be homogeneously distributed within the thin films and show no segregation (e.g., Si does not accumulate at the grain boundaries of the polycrystalline SnO2 structure, as shown in Figure d–f). These observations indicate that the addition of SiO2 is modifying the nature of point defects present within the films; point defects must be present in both amorphous and crystalline structures. To understand in detail the nature of these defects and their passivation mechanism by Si addition, DFT calculations were performed. For these calculations, the rutile crystal structure of SnO2 was used as a starting point because (i) the same effect was observed for amorphous and polycrystalline structures, (ii) ZTO crystallizes into rutile SnO2 and has first coordination shells very close to this atomic structure,[13] (iii) Zn does not appear to modify the Si-passivation mechanism (see previous paragraph), and (iv) in a crystalline structure, the effects induced by point defects can be isolated and only a limited number of defect sites needs to be considered compared to an amorphous environment, thus preventing the convolution of different effects (i.e., induced by the aperiodic structure and/or locally missing atoms) that may blur the contribution of individual point defects in an amorphous material.

The stoichiometric phase of crystalline SnO2 has a defect-free band gap of 3.6 eV with no parasitic absorption in the visible range.[35] One possible cause for the optical absorption feature shown in Figure is deep-defect states arising from charge-neutral oxygen vacancies predicted by theoretical models.[18,19] A similar role of oxygen-deficiency-related defects in the sub-gap absorptance was demonstrated for the amorphous ZTO films in ref (11). The link between VO-related defects and the absorptance features at 600 nm observed in Figure is further supported by the observation that increasing oxygen partial pressure during deposition suppresses the absorption (Figure ). The central role of oxygen deficiencies in the sub-gap absorption and its reduction in the presence of silicon suggest an indirect or direct passivation mechanism of the vacancies upon SiO2 addition. In this section, one such possible mechanism is discussed by considering a direct interaction between Si and oxygen vacancies. First, the contribution of oxygen vacancies to the parasitic absorption in SnO2 is described in detail and then the impact of Si addition is elucidated.

Oxygen Vacancies

The structure of the SnO2 crystal containing an oxygen vacancy is shown in Figure a. Local relaxations of the three-neighboring tin atoms following the creation of an oxygen vacancy result in two symmetry inequivalent Sn-sites labeled site (A) and site (B) in the inset of Figure a. An isolated VO is seen to be stable in two charge states in the crystalline SnO2 film (see Figure b): an ionized q = +2 charge state when the Fermi level is below 2.77 eV and in a charge neutral q = 0 state when the Fermi level is approaching the conduction band. In agreement with the previous studies,[36−38] we observe electronic defect states in the mid-gap region for a charge neutral VO (Figure a), which would contribute to parasitic absorption. In contrast, a doubly ionized VO (Figure b) results in electronic states at the edge of the CBM of stoichiometric SnO2, which would not detrimentally affect the optical properties of TCO. This transition of electronic defect states from deep to shallow is a result of local atomic relaxations that follow the ionization of the vacancy. Similar metastable shallow donor state formation via ionization has also been reported for other TCOs, namely ZnO and In2O3.[39,40]

Figure 6

(a) SnO2 surrounding an oxygen vacancy defect. Sn atoms are shown as purple spheres, oxygen—red, Si—blue, and the vacancy is indicated in green. Right panel: A and B number the two substitutional Si sites neighboring the vacancy. Left panel: The distance between a substitutional SiSn far from VO is indicated; (b) formation energies (O-rich) of isolated defects and defect-clusters as a function of the Fermi level. ϵ(2/0) transitions are indicated by light gray lines. Δ marks the distance between ϵ(2/0) transition and the CBM. This distance, important in determining the ratio between different charge states, is shifted toward the CBM in the presence of Si.

Figure 7

Electronic densities of states for oxygen-vacancy-related defects in SnO2. Results for the charge neutral (q = 0) and for doubly ionized (q = 2) supercells are shown. Colored lines correspond to defect geometries described in detail in the main text. Defect-induced states are highlighted by dashed circles. (Left) Cold-passivation of Sn-based TCOs by cosputtering with SiO2: a new method to design materials with enhanced optoelectronic properties. (Right) General band structure schematics for TCOs, wide band gap materials with sub-gap states near the CBM.

Whether an oxygen vacancy contributes to parasitic absorption or not is, therefore, determined by the position of the Fermi level, ϵF. The Fermi energy at which two different charge states of a given defect have the same formation energy (i.e., form in equal concentrations according to Boltzmann statistics) is known as the thermodynamic transition level. The calculated thermodynamic transition levels, ϵ(2/0), are indicated by gray lines in Figure . In the case of an isolated oxygen vacancy, the ϵ(2/0) transition was found to occur at a Fermi level of 2.77 eV above the valence band. However, in an n-type TCO material, ϵF is expected to lie at or above the CBM. The distance, Δ, between the CBM and the thermodynamic transition level is, therefore, the quantity that determines the ratio between the concentrations, Cq, in which the different charge states, q, will form.

In the case of an isolated VO, a value of 0.855 eV for Δ was obtained. As a consequence, in an n-type SnO2, the majority of oxygen vacancies is expected to be charge-neutral and likely to lead to parasitic absorption.

Addition of Silicon

The EDX measurements reveal a uniform distribution of Si atoms in the SnO2 and ZTO atomic networks; hence, Si clustering in the modeling process was not considered. The rutile structure of SnO2 offers two obvious substitutional sites for Si incorporation: the oxygen, SiO, or the tin, SiSn, site. We found that silicon preferentially substitutes Sn with a formation energy of 2.04 eV and remains electrically inactive for Fermi levels across the band gap, as demonstrated in Figure b. O-site substitution, on the other hand, results in a formation energy over 10 eV higher than that of an Sn site (not-shown), which suggests this defect-type is unlikely to occur.

We then consider the formation of SiSn-VO defect clusters, where the Si atom takes one of the two symmetry inequivalent Sn sites neighboring the oxygen vacancy, marked by A and B on the right panel of Figure a. The calculated binding energies of the ionized SiSn-VO clusters were found to be 0.757 eV on site A and 0.927 eV on site B. The positive binding energy suggests that Si substitutionals prefer to incorporate nearby undercoordinated Sn atoms.

As seen in Figure , in all cases, the electronic defect states associated with a VO formation are not strongly affected by the presence of a neighboring Si atom. However, Figure b reveals that when the SiSn-VO pair is formed, the thermodynamic transition energies ϵ(2/0) are shifted closer to the conduction band and values of Δ equal to 0.635 eV (site B) and 0.655 eV (site A) are obtained. The exponential dependence on the value of Δ suggests that a 25% shift observed in the presence of Si could significantly affect the ratio between the different charge states of oxygen vacancies present in the TCO. The presence of silicon is, accordingly, seen to promote the formation of ionized oxygen vacancies, that is, charge states that do not contribute to the parasitic absorption.

Finally, we validate our results by placing the SiSn and the VO defects inside the same cell, but as far away from each other as the cell size allows, this defect geometry is shown in the left panel of Figure a. In the limit of an infinite cell, one should recover the exact sum of the behaviors of the two defects in isolation. Instead, Figure b reveals a small shift of 20 meV in the thermodynamic transition level ϵ(2/0), when compared to isolated VO. Changes of similar magnitudes are seen in the electronic defect states, as shown in Figure . These shifts reflect the size of the error that results from the choice of the supercell and demonstrate the validity of the SiSn-VO cluster calculations.

Discussion

The increase in oxygen content in the Sn-based thin films when cosputtering with SiO2 eliminates some VO-related defects, subsequently improving the transparency of Sn-based TCOs. As the same optical/electrical trade-off cannot be achieved solely by tuning the oxygen partial pressure during deposition (Figure ) or by mild annealing in air,[11] some additional effects linked to the presence of Si atoms are expected. Our DFT calculations show that the incorporation of Si atoms nearby oxygen-deficient sites is energetically favored, at least in the rutile SnO2 lattice. This is due to a positive binding energy between a substitutional Si atom and a VO. The binding of Si is further seen to promote the ionization of the oxygen defect, releasing charge carriers into the host material. Local structural relaxations following the ionization of VO lead to electronic defect states at the edge of the optical band gap range and thus provide a potential explanation for the success of silicon in passivating optically detrimental states in Sn-based TCOs. A combination of the two phenomena, namely a direct passivation of VO by the oxygen atoms of SiO2 and an indirect passivation of VO because of a shift of the electronic defect states to higher energies close to the band gap edge, could explain the experimental results shown in Figures and 5. Interestingly, we report the same effect in both amorphous ZTO and mixed phase amorphous/polycrystalline SnO2 samples, showing the generality of this “cold-passivation” approach.

Conclusions

In this work, we demonstrated an effective defect passivation scheme for Sn-based materials via SiO2 addition. The addition of SiO2 is experimentally seen to be equally effective for amorphous and mixed phase amorphous/polycrystalline microstructures. In addition, we provide a plausible explanation for the mechanisms governing the cold passivation using DFT calculations. The approach simultaneously preserves the electrical conductivity and improves the transparency of the films, opening new perspectives on low-temperature defect-selective passivation. The compatibility of this cosputtering methodology with temperature-sensitive processes and substrates (<200 °C) enables its application in transparent and flexible electronics. Finally, this approach should serve as an inspiration to design and discover oxides that could potentially play a similar role in other TCOs as SiO2 does in SnO2 and ZTO.

Methods

Experimental Section

Thin films (150 nm) of SiZTO and SiSnO2 were deposited onto aluminoborosilicate glass in a Leybold Univex RF sputtering system from separate targets of SnO2, Zn0.05Sn0.3O0.65, and SiO2. Depositions were performed using two targets simultaneously, that is, ZTO and SiO2 to deposit SiZTO or SnO2 and SiO2 to deposit SiSnO2. The ZTO composition was optimized as described in ref (14). The power applied to the ZTO and SnO2 targets was fixed to 80 W (1.02 W cm–2), and the power on the SiO2 target was varied between 0, 15, and 20 W (up to 0.25 W cm–2, all targets were 10 cm in diameter). Depositions with 5 W applied to the SiO2 targets did not yield a stable plasma and were hence not performed. Before deposition, the pressure in the working chamber was ∼6 × 10–7 mbar. Substrate temperatures of 100 and 25 °C were used for ZTO and SnO2 respectively, because these conditions were found to yield high-quality films. Depositions were done with a constant flow of 10 sccm of Ar, whereas the O2 partial pressure was changed by increasing or decreasing the flow of an Ar–O2 gas mixture (95 at. % Ar and 5 at. % O2) from 1.0 to 3.5 sccm to optimize the optoelectronic properties. The resulting working pressures were between 4 and 10 × 10–4 mbar. Following depositions, the films were subjected to a thermal treatment at 200 °C for 30 min in air using a hot plate. The free carrier densities and Hall mobilities of the films were obtained with a Hall effect HMS-5000 system in the Van der Pauw configuration. Their optical properties were measured using a PerkinElmer Lambda 900 spectrophotometer equipped with an integrating sphere. The absorptance of the films was calculated using the total transmittance and the total reflectance. To assess the microstructure and composition of the films, TEM was performed in  FEI TITAN Themis (STEM EDX) or a FEI Osiris (nanobeam diffraction) microscope, both operated at 200 kV. Samples were characterized in cross section. Thin lamellae were extracted using the conventional focused-ion beam lift-out method in a Zeiss NVision 40. RBS spectrometry was used to assess the atomic concentration of the different atomic species in SiZTO and ZTO. During RBS measurements, high-energy He2+ ions are directed onto the samples, and the energy distribution and yield of the backscattered He2+ ions at a 160° angle is measured. For the calculation of the atomic concentration, the substrate and the background signals were subtracted. For the RBS measurements, uncertainties from statistical errors are shared for all films because all samples were deposited in the same sputtering system and were exposed to the same atmospheres and possible contaminants from the atmospheric environment. TDS was performed using an ESCO spectrometer equipped with a quadrupole mass spectrometer and a halogen lamp at a base pressure of 10–9 mbar. By comparing the total effusion and desorption rates from TDS, it was possible to compare total oxygen, tin and zinc desorption for ZTO and SiZTO while heating the samples at a constant rate of 20 °C/min up to 700 °C.

Defect Calculations

Thermodynamic transition levels ϵ(q1/q2) between two charged states q1 and q2 of a given defect show the Fermi energy, ϵF, at which the stable charge state changes. They were calculated using eq ,where EDF(q,ϵF = ϵV) is the formation energy of a defect D in a charge state q when the Fermi energy is set equal to the valence band maximum ϵV. Formation energies of the charged defects for each charge state were calculated using eq .where ED is the energy of the supercell containing the defect D in a charge state q. ESnO is the energy of the pure SnO2 crystal in the same-sized supercell, n is the number of atoms of species i added to the supercell to create the defect, and μ is the chemical potential of that species. Chemical potential bounds were imposed by SnO2 and SiO2 formation. More detailed explanations of the methodology and the correction term, Ecor, applied to charged defect calculations can be found in ref (38).

The binding energy between two defects, X and Y, was calculated as the energy difference between the formation energies of the isolated defects and the formation energy of the X–Y defect cluster.

According to the definition in eq , a positive binding energy implies a preference for the two defects to cluster, whereas a negative binding energy suggests a preference for isolated defects. As the formation energy of a given defect (eq ) depends on the Fermi level and the charge state of the defect, so does the binding energy.

Computational Details

All DFT calculations were performed using the PBE0 hybrid functional as implemented in the VASP electronic structure code.[41−44] Si 3s and 3p (4), O 2s and 2p (6), and Sn 5s, 5p, and 4d (14) electrons were included in the valence. All defects were introduced into a 2 × 2 × 3 (72 atom) supercell of rutile SnO2 phase. The atomic positions were relaxed using a 2 × 2 × 2 Monkhorst–Pack k point mesh until the forces were below 0.02 eV/Å. Final densities of states were obtained using a 3 × 3 × 3 Γ-centered k-point mesh. The volume of the supercell was fixed to that of the (expanded) perfect crystal calculated via fitting the Birch–Murnaghan[45] equation of state. A 3 × 3 × 4 (216 atoms) supercell was tested to verify convergence with respect to supercell size, and a good qualitative agreement was found.