Computational Prediction of an Antimony-Based n-Type Transparent Conducting Oxide: F-Doped Sb2O5

Transparent conducting oxides (TCOs) possess a unique combination of optical transparency and electrical conductivity, making them indispensable in optoelectronic applications. However, their heavy dependence on a small number of established materials limits the range of devices that they can support. The discovery and development of additional wide bandgap oxides that can be doped to exhibit metallic-like conductivity are therefore necessary. In this work, we use hybrid density functional theory to identify a binary Sb(V) system, Sb2O5, as a promising TCO with high conductivity and transparency when doped with fluorine. We conducted a full point defect analysis, finding F-doped Sb2O5 to exhibit degenerate n-type transparent conducting behavior. The inherently large electron affinity found in antimony oxides also widens their application in organic solar cells. Following our previous work on zinc antimonate, this work provides additional support for designing Sb(V)-based oxides as cost-effective TCOs for a broader range of applications.


■ INTRODUCTION
In the field of modern (opto-)electronics, transparent conducting oxides (TCOs) act as pivotal materials, bridging the gap between conductivity and optical transparency. 1,2rom touchscreen devices that can interact with users to solar cells that harvest sunlight energy, the pursuit of highperformance TCOs continues to contribute to a variety of technological applications, yielding more efficient devices and reduced energy consumption.
The developmental history of TCOs spans several decades.Cadmium oxide (CdO), the first recorded TCO, was introduced by Badeker in 1907. 3It was found that a low resistivity of 1.20 × 10 −3 Ωcm and a large mobility of over 100 cm 2 V −1 s −1 at a high carrier concentration of 1 × 10 21 cm −3 can be achieved in CdO. 4 Meanwhile, the Moss−Burstein shift leads to a wide optical bandgap, enabling its high level of transparency. 5Despite the toxic nature of CdO, this discovery laid the foundation and opened up a research area for improved TCOs.−9 Among these metal oxides, In 2 O 3 is the most popular transparent conductor, widely used in cutting-edge technology due to its high conductivity and over 90% transparency. 1owever, the high cost due to the scarcity of indium raises concerns about its long-term sustainability in large-scale applications.The development of new promising TCOs is therefore driven by the demand for cost efficiency, a broader application scope, and enhanced performance. 2 recurring characteristic observed among these successful n-type TCOs is the presence of post-transition metal cations with an electronic configuration of (n − 1)d 10 ns 0 np 0 .A wide bandgap with highly dispersed conduction bands often exists in these oxides due to the strong antibonding hybridization of the valence cation s and the O 2p orbitals.This electronic configuration is primarily found in elements from groups 12, 13, and 14 in the periodic table.−14 Sb is a group 15 element, but in its highest oxidation state [Sb(V)], it adopts the same (n − 1)d 10 ns 0 np 0 electron configuration as conventional post-transition metal TCOs.Furthermore, the increased abundance of Sb can contribute to a reduction in manufacturing costs. 15The discovery of this material not only brings valuable diversity to the field but also opens the possibility of Sb(V)-based oxides as lower-cost alternative TCOs.Following the success of ZnSb 2 O 6 , we consider the underexplored binary system Sb 2 O 5 as a potential candidate due to its wide bandgap and comparable electronic structure to established metal oxide TCOs.The crystal structure of monoclinic Sb 2 O 5 was first studied by Jansen in 1978, where Sb 2 O 5 powder was synthesized by heating Sb 2 O 3 in a steel autoclave that was charged with liquid oxygen. 16The electronic structure of Sb 2 O 5 was briefly investigated along with its competing phases by Allen et al., while its properties as a potential TCO have been overlooked. 17n this work, we initially compared the formation energies of all reported polymorphs of Sb 2 O 5 (Table S1), where monoclinic C2/c Sb 2 O 5 was found to be the most stable crystal structure.We examined the crystal and electronic structures of monoclinic Sb 2 O 5 with hybrid density functional theory (DFT).We find that Sb 2 O 5 possesses a wide optical bandgap over 3.6 eV, suggesting visible light transparency, and high dispersion in the conduction band, suggesting good electron mobility.Charge transport properties were also evaluated, in which intrinsic Sb 2 O 5 was found to exhibit relatively high electron mobility and thus conductivity�if large carrier concentrations can be achieved through extrinsic doping.To investigate the dopability of Sb 2 O 5 , we conducted a full intrinsic point defect analysis, finding the self-consistent Fermi level to reside deep within the bandgap and thus insulating behavior for undoped Sb 2 O 5 .To enhance its electrical conductivity, we considered introducing an extrinsic dopant to the oxygen site, namely, fluorine (F).Our investigation demonstrated that the fluorine-on-oxygen substitution F O serves as a good electron donor in Sb 2 O 5 , shifting the self-consistent Fermi level into the conduction band and achieving equilibrium charge carrier concentrations on the order of 10 19 cm −3 �surpassing the Mott criterion for degenerate conductivity.Thus, we predict F-doped Sb 2 O 5 to be a degenerate n-type TCO.The earth-abundant nature of antimony also renders F-doped Sb 2 O 5 a promising candidate for achieving cost efficiency in optoelectronics applications.The large electron affinity (EA) observed in the band alignment of Sb 2 O 5 extends the range of potential applications for TCOs.This study further demonstrates the feasibility and promising potential of designing Sb(V)-oxides as TCOs for lower cost and broader applications.

■ COMPUTATIONAL METHODS
−23 Density functional perturbation theory and phonon dispersion calculations were performed using the Perdew−Burke−Ernzerhof for solids (PBEsol) exchange− correlation functional, which adopts the generalized gradient approximation. 24The rest of the calculations were performed using the PBE0 hybrid functional, which has been shown to accurately predict the electronic structure of Sb(V)-based oxides. 10,25Planewave cutoff and k-point sampling were tested and converged using vaspup2.0, 26with a cutoff energy of 500 eV and a k-point sampling of 6 × 6 × 5 for the 14-atom primitive cell of Sb 2 O 5 found to be sufficient for total energies converged to 1 meV/atom.The crystal structure of Sb 2 O 5 was visualized using VESTA. 27The electronic band structure and density of states (DOS) were plotted using the SUMO package. 28Charge transport properties were obtained and plotted using AMSET and ThermoParser. 29,30Phonon dispersion curves were calculated using PHONOPY and plotted using ThermoParser. 30,31he band alignment of bulk Sb 2 O 5 was calculated using the core and vacuum energies from a 30 Å thick (001) surface slab with 20 Å of vacuum using the SURFAXE package. 32or defect calculations, an 112-atom supercell from a 1 × 2 × 2 expansion of the Sb 2 O 5 conventional cell was used to minimize the interactions between periodic defects.To sample the defect configuration landscape and search for ground and metastable defect structures, the ShakeNBreak approach was used to initially generate Γ-point-only relaxations for each defect with 10 different local distortions. 33,34The ground-state structure found in these initial relaxations was then selected for structural optimization with a converged Γ-centered 2 × 2 × 2 k-point mesh to obtain the total energy of the ground-state defective supercell.The formation energy of a defect with the charge state q can be calculated using the equation below 35

E
−39 An additional correction term (E corr ), specifically the image-charge correction, is also needed to account for spurious finite-size supercell effects.Each term in eq 1 along with the formation energies of all intrinsic point defects and extrinsic defects are summarized in Tables S8 and S9.The Doped Python package was used to manage all the defect calculations and analysis. 40Data produced during this work are freely available at doi.org/10.5281/zenodo.10075593.
■ RESULTS AND DISCUSSION Crystal Structure.Sb 2 O 5 crystallizes in a monoclinic structure with the C2/c space group (Figure 1).Each Sb atom is surrounded by six oxygen atoms, forming a mixture of distorted SbO 6 corner-and edge-sharing octahedra, yielding an infinite three-dimensional framework.Three crystallographically distinct oxygen sites exist, with one edge-sharing and two corner-sharing environments.The edge-sharing oxygen has C 1 point symmetry and forms three different O−Sb bonds with an average bond length of 2.07 Å. Between the corner-sharing sites, one exhibits two identical O−Sb bonds with a length of 1.91 Å, yielding C 2 symmetry, while the other has two nonequivalent O−Sb bond lengths of 1.92 and 1.89 Å, exhibiting C 1 symmetry.As shown in Table 1, the calculated PBE0 lattice parameters are in excellent agreement with the experimental values. 16PBEsol-and HSE06-relaxed crystal structures were also compared (Table S2), where the PBEsol functional slightly overestimates the lattice parameters by around 2%, while the HSE06 lattice parameters are basically the same as those of PBE0.The phonon dispersion was also calculated (as shown in Figure S1 in the Supporting Information) and no imaginary modes were witnessed, demonstrating the dynamic stability of this compound.
Electronic Structure.The PBE0-calculated electronic band structure of Sb 2 O 5 in Figure 2a and S2), where the direct band-to-band absorption spectrum was calculated by using the PBE0 hybrid DFT functional.The absorption coefficient only reaches 10 4 cm −1 at an energy of 3.6 eV.Indeed, an effective optical bandgap of ∼3.6 eV is predicted by determining the point at which the linear fit intersects the x-axis in the direct-gap Tauc plot (Figure S2).We note that this aligns closely with the band gap of ∼3.55 eV from optical transmittance measurements by Mindil et al. on cubic Sb 2 O 5 nanorod films; 41 however, no experimental measurement for C2/c monoclinic Sb 2 O 5 is known.Sb 2 O 5 has a centrosymmetric (C2/c) crystal structure, and so, it is likely that some of the low-energy optical transitions are symmetry-forbidden as in In 2 O 3 , 42 resulting in weak absorption just above the direct bandgap. 43harge Transport Properties.For an ideal TCO, high carrier mobility is required to achieve optimum conductivity.The Mott criterion (n Mott ) describes the critical carrier concentration at which a material is expected to exhibit metallic-like conductivity, given by 10,44−46  The charge transport properties are calculated using AMSET, with the results shown in Figure 3. 29 Figure 3a illustrates how the conductivity changes with temperature at different carrier concentrations, from 1 order of magnitude smaller to 3 orders larger than the Mott criterion value.The conductivity is more sensitive to temperature at lower carrier concentrations, with conductivity decreasing as temperature increases�as expected for band-like carrier transport.At carrier concentrations of 10 19 cm −3 , the conductivity can reach over 100 S cm −1 at room temperature, which can be increased  to just under 10 4 S cm −1 at carrier concentrations of 10 21 cm −3 .Therefore, to achieve sufficient conductivity for TCO applications, high doping concentrations are needed in this system.The change in mobility with carrier concentrations at room temperature is shown in Figure 3b.A maximum mobility of around 106 cm 2 V −1 s −1 is obtained at low carrier concentrations, while it declines with higher carrier concentrations.The typical mobility for n-type TCOs is from 50 to 70 cm 2 V −1 s −1 , and so, Sb 2 O 5 has a competitive mobility compared with that of established TCOs. 51he analysis of the limiting scattering type for mobility is shown in Figure 3c.Polar-optical phonon (POP) is the dominant limiting factor for carrier concentrations below 10 20 cm −3 across all temperatures above 300 K.For carrier concentrations over 10 20 cm −3 , the limiting scattering type becomes an ionized impurity (IMP) at temperatures lower than 800 K.In both regimes, the overall mobility decreases as temperature increases due to increased carrier-phonon (POP) scattering.
Intrinsic Defect Chemistry.Defect chemistry is critical for tailoring the electrical conductivity and optical properties of the TCOs.As shown above, to achieve sufficiently high conductivities for TCO applications, high doping concentrations are required for this material.Thus, to investigate the dopability of this compound, a thorough study of the intrinsic defect chemistry was performed.Chemical potential limits need to be analyzed first as they provide a thermodynamic reference for the formation energy and stability of defects in materials.Table S5 lists the calculated chemical potential limits using the total energy per atom for each competing phase shown in Tables S3 and S4. Figure S4 shows the thermostability region of Sb 2 O 5 where it exhibits a relatively wide Sb chemical potential range of around 2 eV and a smaller oxygen chemical potential range of 0.76 eV.
The total dielectric constant, used to obtain the finite-size correction term E corr in eq 1, was calculated to be 13.99, 12.16, and 11.61 ε 0 along the a, b, and c directions, respectively, using AMSET. 29Herein, the Kumagai−Oba (eFNV) finite-size charge correction scheme was applied due to the anisotropic character of Sb 2 O 5 . 52s a binary semiconductor, Sb 2 O 5 has six possible intrinsic point defects, namely, antimony interstitials (Sb i ), oxygen interstitials (O i ), antimony vacancies (V Sb ), oxygen vacancies (V O ), antimony-on-oxygen antisites (Sb O ), and oxygen-onantimony antisites (O Sb ).However, the analyses of O Sb and Sb O are excluded as they are expected to be energetically unfavorable due to their large size and charge mismatch.where the lowest energy charge state changes directly from +2 to neutral as the Fermi level moves toward the CBM. 54This is because the formation of a neutral oxygen vacancy creates two electrons that are mainly localized on the neighboring Sb s orbital, where it favors full or zero occupation over partial occupation.In addition, no major symmetry-breaking is identified for V O +1 , with similar charge localization to the neutral vacancy state, as shown in Figure 4b, and thus, a negative-U behavior is preferred.The other two oxygen vacancies 1 ] have a narrow stability window of 15 and 10 meV for the +1 charge state, respectively.While oxygen vacancies are the lowest energy intrinsic defect, they still have significant formation energies, with E f (V C O(c, ) 0 1 ) = 1.68 eV�suggesting that they will form in appreciable but not large concentrations unless very high processing temperatures are used.The charge density of the lowest formation energy 1 is shown in Figure 4b.Upon formation of the oxygen vacancy in its neutral state, two extra electrons are generated, which prefer to occupy the lowest energy unoccupied states� the Sb s orbitals of the CBM.This is seen in Figure 4b, where the electron densities associated with V O 0 and V O +1 are mainly localized on the Sb atom neighboring the vacancy site.V O +2 is a fully ionized state where no additional electron/charge is present.
O i is the next lowest energy defect, which has a formation energy of 3.89 eV under O-poor conditions (Figure 4).It is stable as a neutral dimer across all Fermi levels, with an O−O bond length of 1.40 Å.−57 Sb i is a donor defect in Sb 2 O 5 , with (+5/+4), (+4/+3), and (+3/+2) transition levels located within the bandgap, where the transition level is defined as the position of the Fermi level when two different charge states of a defect have the same energy.Sb i prefers to occupy octahedral sites, where it is surrounded by six O atoms, resembling the typical Sb−O coordination environment found in the bulk structure.For Sb i +5 , the average Sb i −O bond length is 1.96 Å, similar to the bulk Sb−O bond length of 1.99 Å. V Sb has much higher formation energy at 6.31 eV for V Sb 0 under O-poor conditions, where V Sb −4 cuts the CBM at around 5.0 eV, and so is much less likely to form.High formation energies for V Sb were also observed in ZnSb 2 O 6 10 and can be attributed to the high oxidation state (+5) of Sb in this compound.However, under Sb-poor conditions, the formation energy of V Sb at the CBM drops by ∼2 to ∼3 eV, becoming one of the lowest energy native defects under these conditions.Notably, V Sb here is found to be a rare four-electron negative-U center, where only the (0/−4) charge transition level occurs within the bandgap�indicating that the equilibrium charge state of antimony vacancies changes directly from neutral to −4 as the Fermi level approaches the CBM.Negative-U behavior is typically related to structural reconstruction and lattice distortion, and so the structures of V Sb 0 and V Sb −4 were investigated in detail. 54As shown in Figure 4c  for V Sb in Sb 2 Se 3 , which also results in 4-electron negative-U behavior. 54−60 This trimer geometry was identified using ShakeNBreak where the energy was found to be around 4 eV lower than that of the structure obtained with a standard (unperturbed) defect geometry relaxation (Figure 4c).On the other hand, for V Sb in +3, +4, and +5 charge states, we find a split-vacancy configuration to be the preferred arrangement (Figure S6).A split-vacancy configuration can be thought of as a divacancy and interstitial cluster as a neighboring host Sb atom displaces toward the vacancy position (creating an additional vacancy and interstitial in the process).This type of vacancy structure is often formed in cation vacancies with high charge states, having also been reported in α-Al 2 O 3 and β-Ga 2 O 3 . 61,62he thermodynamic analysis of intrinsic point defects in Sb 2 O 5 above demonstrates that the oxygen vacancies are the lowest formation energy defects.However, they are still moderately high in energy, and the donor levels are deep in the bandgap (0.9−0.6 eV below the CBM).Therefore, there will not be any significant intrinsic n-type doping in this system.However, Sb 2 O 5 has a large n-type doping window of around 5.0 eV.The doping window is defined as the formation energy of the lowest energy-compensating defect species at the corresponding band edge as this sets an upper limit to the formation energy of dopants, which could push the Fermi level close to the band edge without being negated by ionic charge compensation from intrinsic defects.O i does not chargecompensate as it stays neutral across the bandgap.In this case, the donor doping is charge-compensated by the native acceptor V Sb −4 , cutting the CBM at around 5.0 eV under Opoor conditions.Compared with ZnSb 2 O 6 , 10 Sb 2 O 5 has a wider n-type doping window since the dominant intrinsic acceptor (cation vacancy V Sb ) here has higher energy than that of V Zn because of the higher oxidation state of Sb.
The self-consistent Fermi level under O-poor conditions was predicted using the PY-SC-FERMI package. 63,64An annealing/ growth temperature of 700 °C was assumed as it is a widely reported growth temperature in the literature for Sb 2 O 5 synthesis. 16,41,65To understand the defect behavior at operating (room) temperature, the "frozen defect approach" was applied where the total concentration of each defect is fixed to the equilibrium concentration at the annealing temperature.Upon cooling, the total concentration of each defect remains unchanged, whereas the relative populations of each defect charge state are allowed to re-equilibrate.The selfconsistent Fermi level is then recalculated using the operating temperature (300 K). Figure 4b shows that the self-consistent Fermi level in undoped Sb 2 O 5 sits at around 0.50 eV lower than the CBM at room temperature, corresponding to an essentially negligible electron concentration of around 10 8 cm −3 , and showing undoped Sb 2 O 5 to be highly insulating as expected.
Extrinsic Defect Chemistry.From the charge transport analysis earlier, Sb 2 O 5 was found to have good electron mobility, but to achieve a high conductivity comparable to conventional TCOs, high carrier (and thus doping) concentrations are required.However, at high doping concentrations, the impact of extrinsic dopants on the host lattice and carrier scattering can severely diminish the carrier mobilities.Therefore, to optimize the electrical conductivity, the objective is to achieve electron-doping while minimizing disruption of the CBM states (in order to retain decent carrier mobilities).As shown in Figure 2, Sb 5s states mainly comprise the CBM, followed by the states of O 2p.Considering O is the minor contributor to the CBM, doping on the O site is preferred since it is likely to have a smaller impact on electron mobilities.In addition, the relatively small V O but large V Sb formation energies shown in the intrinsic transition level diagram (Figure 4b) also suggest that substituting on O will be easier as removing O from the lattice requires less energy than removing Sb.Fluorine (F) was selected as a potential dopant due to its similar ionic radius (1.29 Å) to that of oxygen (1.35 Å) and adjacent position to that of oxygen on the periodic table (having similar 2p valence orbital energies).Combined, these similarities are likely to yield low formation energies for F O substitutional donors and thus high achievable electron concentrations.−68 Figure 5a shows the transition level diagram for the lowenergy extrinsic point defects in F-doped Sb 2 O 5 under Sb-rich/ O-deficient conditions, in which the limiting secondary phase is Sb 2 F 7 .F i is found to be a high-energy defect with a formation energy of around 3.7 eV in its negative charge state at the during annealing (gray) and upon cooling to operating temperature (300 K; black).The light blue block shows a range of growth temperatures from 500 to 700 °C reported in the literature. 16,41,65,69(c) Room-temperature carrier and defect concentrations as a function of annealing temperature.Defects with concentrations lower than 1 × 10 10 cm −3 are omitted from the legend and are plotted using DOPED and PY-SC-FERMI. 40,63,64BM, precluding its formation at any significant concentration within the material and thus having a negligible impact on the conductivity.By comparing F i with the O i interstitial, it can be seen that F i has a higher formation energy for most (n-type) Fermi levels.Usually, the smaller ionic radius of F would suggest F i formation to be more favorable, especially in ionic compounds. 70However, here, O i is more stable due to the formation of the O−O dimer, which stabilizes the neutral state.It is also noticeable that F i is a negative-U center with F i +1 stability across a large range of Fermi levels, despite this being a somewhat unusual charge state from simple oxidationstate considerations.Similar negative-U behavior of F i +1 is observed in another n-type oxide, BaBi 2 O 6 . 71This defect species is stabilized by the formation of a short O−F i bond with a corner-sharing oxygen atom, with a bond length (1.40 Å) similar to that in OF 2 (1.41 Å), annihilating the two holes associated with this species. 72luorine-on-oxygen substitutions, on the other hand, have quite low formation energies with the most favorable substitution site 2 have lower formation energies (0.60 and 0.77 eV at the CBM, respectively), reflecting the trend in oxygen vacancy formation energies (Figure 4a).
The calculated self-consistent Fermi levels and defect/ electron concentrations upon F doping are presented as functions of the annealing temperature in Figure 5b,c.During the annealing process, a significant amount of 1 is formed, with its concentration equal to the overall electron concentrations at all temperatures due to F C O(c, ) 1 being the dominant contributor to the conductivity of F-doped Sb 2 O 5 .Due to the lack of low-energy-compensating acceptor species, the electron concentrations and Fermi levels under annealing and after cooling are mostly similar.The self-consistent Fermi level E F at room temperature is predicted to lie at 0.18 eV above the conduction band, assuming an annealing temperature of 700 °C. 16,41,65The electron concentration at this predicted E F reaches 10 19 cm −3 , which is around 1 order of magnitude higher than the Mott criterion.At this carrier concentration, the predicted mobility and conductivity of Fdoped Sb 2 O 5 are around 103 cm 2 V −1 s −1 and 100 S cm −1 , respectively, at room temperature, where the limiting scattering mechanism is mainly POP (Figure 3).In comparison to that of conventional n-type TCOs, this predicted conductivity for Fdoped Sb 2 O 5 is somewhat modest.If a greater conductivity is desired, a higher annealing temperature can be used, as demonstrated in Figure 5c, or potentially nonequilibrium growth conditions, where more fluorine can be incorporated to yield higher carrier concentrations.Moreover, the neglect of temperature-dependent renormalization of the host band gap may also contribute to a slight underestimation of the equilibrium electron concentrations under F-doping here.
The calculation is also limited by the relatively simple model used in the AMSET package to address impurity scattering as the effect of a relatively significant F O concentration on the lattice is not explicitly accounted for in the carrier mobility model.Overall, the low-energy formation of resonant F O defects is predicted to push the self-consistent Fermi level into the conduction band, yielding a strongly degenerate semiconducting behavior upon F-doping in Sb 2 O 5 .
Band Alignment.The electronic band alignment of Sb 2 O 5 was investigated and compared with that of existing TCOs, as shown in Figure 6.The ionization potential and the EA were calculated to be 9.89 and 6.80 eV, respectively.The EA of Sb 2 O 5 is significantly larger than that of established TCOs.This is because the Sb 5s states mainly contribute to the CBM, as for ZnSb 2 O 6 and similar to other TCOs where the cation ns orbitals comprise the CBM. 10 However, the ns energy levels differ due to the effective nuclear charge and electron shielding effects.Other conventional post-transition metals have higher energy for the unoccupied ns states since they have lower oxidation states (+2/+3/+4) than Sb(V) and thus reduced nuclear attraction. 70Therefore, the Sb 5s orbitals in the +5 oxidation state have a smaller nuclear-electron distance, which has less shielding and stronger attraction, causing the Sb s states to sit at lower energy.Indeed, the fifth ionization energy of Sb is much higher than the fourth and third ionization energies of Sn and In. 76Moreover, this brings the Sb s orbital closer in energy to the deep 2p states, giving rise to a stronger interaction in the conduction band.The high oxidation state and smaller ionic radius of Sb 5+ also result in a reduced Sb−O bond length compared to those of Sn−O and In−O, contributing to the strength of the Sb 5s−O 2p interactions.The resulting dispersed conduction band further extends the energy range of the unoccupied states, yielding a larger EA.This large EA could be particularly beneficial for TCO applications in organic solar cells, aiding the acceptance of photoexcited electrons from the organic active layer and thus facilitating electron extraction, reducing the likelihood of electron−hole recombination, and potentially boosting efficiencies. 77Sb 2 O 5 with a high EA could therefore be a cheaper choice of TCO that can provide more diversity to the optoelectronics market.

■ CONCLUSIONS
An investigation into the potential of Sb 2 O 5 as an n-type TCO was conducted by using first-principles calculations.The analysis of its electronic structure revealed a highly dispersed conduction band, indicative of high electron mobility.A large optical band gap of around 3.6 eV was predicted, demonstrating visible-light transparency.Although undoped Sb 2 O 5 was to be insulating, a wide n-type doping window facilitates the introduction of extrinsic dopants, which can provide high charge carrier concentrations.Fluorine was chosen as a candidate substitutional dopant due to its similar size to oxygen and the low formation energies of oxygen vacancies.The predicted carrier concentration of F-doped Sb 2 O 5 surpassed the Mott criterion, pushing the self-consistent Fermi level into the conduction band.The results show that Fdoped Sb 2 O 5 exhibits degenerate n-type transparent conducting behavior.Furthermore, the electronic band alignment demonstrates an extremely large EA in Sb 2 O 5 , enhancing its suitability in organic solar cells, where the overall efficiency can be improved along with a lower cost.These findings corroborate the potential of Sb(V)-based oxides as alternative earth-abundant n-type TCOs, providing much-needed diversity in the field of transparent conducting materials.

* sı Supporting Information
The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.chemmater.3c03257.Calculated lattice parameters, phonon dispersion, Tauc Plot, COHP analysis, data for competing phase formation energies for calculating the chemical potential limits of both intrinsic and extrinsic defects, thermostability region plot, intrinsic and extrinsic transition level diagrams under p-type (Sb-poor and O-rich) conditions, visualization of V Sb −4 split-vacancy configuration, and all the energy terms in eq 1 used for calculating the defect formation energy (PDF) ■ AUTHOR INFORMATION

Figure 1 .
Figure 1.(a) Primitive unit cell of Sb 2 O 5 , where Sb atoms are in blue and O atoms in yellow.(b,c) show the Sb 2 O 5 crystal structure looking down the c and b axes, respectively, showing the corner-and edge-sharing SbO 6 octahedra.
shows a slightly indirect bandgap of 3.08 eV, from the VBM located between Γ and A [0, 0, 0.5] to the conduction band minimum (CBM) at Γ.It has a direct bandgap of 3.11 eV at Γ.The conduction band is highly dispersed and has electron effective masses of 0.30, 0.30, and 0.34 m e along the Γ → Y, Γ → V, and Γ → A directions, respectively.This results in a mean effective mass of 0.31 m e for Sb 2 O 5 , indicating the potential for high electron mobility.The high dispersion of the conduction band arises due to the strong antibonding interaction of the unoccupied cation s orbitals and O 2p states as in other post-transition metal oxides, which can be seen from the crystal orbital Hamilton population (COHP) analysis shown in Figure S3.A flat valence band is observed, reflecting the localized character of the O 2p bonding states which dominate the VBM.This results in large hole effective masses of 1.48 and 1.84 m e from VBM to A and VBM to Γ, respectively.The optical properties are shown in the optical absorption and Tauc plots (Figures 2b a 0 is the effective Bohr radius (18.9 Å for Sb 2 O 5 ); ε is the total dielectric constant, which is equal to the calculated total relative dielectric constant([13.99,12.16, 11.61]) times the permittivity of free space (ε 0 ); and m* is the reduced effective mass.Using the electron and hole effective masses obtained from the Sb 2 O 5 band structure, a reduced effective mass of 0.263 m 0 was determined, showing a comparable value to that of traditional TCOs (0.13−0.55 m 0 ).47,48The Mott criterion for Sb 2 O 5 predicts a critical carrier concentration of 1.09 × 10 18 cm −3 based on these values, closely aligning with the conventional TCOs such as In 2 O 3 , SnO 2 , and Ga 2 O 3 , which have Mott criterion concentrations ranging from 1 × 10 18 cm −3 to 3 × 10 18 cm −3 .48−50

Figure 2 .
Figure 2. (a) Electronic band structure and orbital-decomposed DOS of Sb 2 O 5 calculated using hybrid DFT (PBE0), where the conduction bands are shown in yellow and the valence bands in blue.(b) Band−band optical absorption for Sb 2 O 5 plotted using SUMO.28 Figure 4b shows the formation energy diagram for all intrinsic point defects under Sb-rich/O-poor conditions (i.e., the conditions most conducive to n-type behavior).Three different V O exist due to the three crystallographically distinct oxygen sites [one edge-sharing, V C O(e, ) 1 , and two corner-sharing, V C O(c, ) 1 and V C O(c, ) 2 ], each acting as deep donors and having the lowest formation energies among all intrinsic point defects�as is often the case in conventional TCOs such as In 2 O 3 , SnO 2 , and ZnSb 2 O 6 . 8,10,53Among the three oxygen vacancies, V C O(c, ) 2 exhibits negative-U behavior

Figure 3 .
Figure 3. (a) Simulated conductivity of Sb 2 O 5 over a range of carrier concentrations chosen with respect to the Mott criterion.(b) Calculated electron mobility of Sb 2 O 5 over a range of carrier concentrations at room temperature.(c) Effect of scattering mechanisms on mobility at low and high carrier concentrations.ADP is acoustic deformation potential scattering, IMP is ionized impurity scattering, and POP is polar-optical phonon scattering.
, an O−O−O trimer is formed in V Sb 0 , similar to the Se−Se−Se trimer found

Figure 4 .
Figure 4. (a) Transition level diagram of intrinsic defects in Sb 2 O 5 under Sb-rich and O-poor (n-type) conditions, with a self-consistent Fermi level at 2.58 eV above the VBM.(b) Charge density isosurfaces of the highest-occupied state of V O(c,Cd 1 ) in 0 and +1 charge states using a wireframe structure, plotted with an isosurface depth of 0.032 e Å −3 and viewed along the c direction.O atoms correspond to the intersection of red lines, Sb to the intersections of purple lines, and the black dotted circle indicates the oxygen vacancy.(c) Visualization of the O−O−O trimer formed in the ground-state structure of V Sb 0 , alongside the ShakeNBreak 33,34 plot of energy versus initial bond distortion factor.Here, the ground-state structure is found with −0.4 (−40%) bond distortion.

Figure 5 .
Figure 5. (a) Transition level diagram of low-energy extrinsic defects in F-doped Sb 2 O 5 under Sb-rich and O-poor (n-type) conditions with the predicted self-consistent Fermi level located at 0.18 eV above the CBM.(b) Calculated self-consistent Fermi level positions in F-doped Sb 2 O 5during annealing (gray) and upon cooling to operating temperature (300 K; black).The light blue block shows a range of growth temperatures from 500 to 700 °C reported in the literature.16,41,65,69(c) Room-temperature carrier and defect concentrations as a function of annealing temperature.Defects with concentrations lower than 1 × 10 10 cm −3 are omitted from the legend and are plotted using DOPED and PY-SC-FERMI.40,63,64

1 +
[F C O(c, ) 1 ] having a formation energy of 0.60 eV in the +1 charge state at the CBM.In F C O(c, ) 1 , the newly formed two Sb−F bonds have longer bond lengths (2.07 and 1.99 Å) than the two Sb−O bond lengths in bulk Sb 2 O 5 (1.92 and 1.89 Å).The low formation energy of F O indicates its ready solubility in Sb 2 O 5 , and is analogous to that observed in ZnSb 2 O 6 . 10All F O substitutions are resonant defects where the (+1/0) transition level sits well within the conduction band, indicating that high carrier concentrations can be achieved through F doping.Among the three substitutional defects, the corner-sharing sites F C O(c, ) 1 and F C O(c, )

Table 1 .
16mparison between the Conventional Crystal Lattice Parameters Calculated Using the PBE0 DFT Functional and the Experimental Values Measured Using Powder X-ray Diffraction.16