betaFeedback

WorksPhysics and Astronomy

Finite-Size Corrections to Defect Energetics along One-Dimensional Configuration Coordinate

Abstract

Recently, effective one-dimensional configuration coordinate diagrams have been utilized to calculate the line shapes of luminescence spectra and non-radiative carrier capture coefficients via point defects. Their calculations necessitate accurate total energies as a function of configuration coordinates. Although supercells under periodic boundary conditions are commonly employed, the spurious cell size effects have not been previously considered. In this study, we have proposed a correction energy formalism and verified its effectiveness by applying it to a nitrogen vacancy in GaN.

Introduction

Point defects are ubiquitous in solids and play important roles in many fundamental physics (e.g., superconductors [1] and topological materials [2]), and practical applications (e.g., solar cell absorbers [3] and light-emitting diodes [4]). To better understand and even predict the point defect properties, first-principles calculations have been routinely used over the last few decades [5]. Recently, the line shapes of luminescence spectra [6, 7] and carrier capture coefficients [8, 9, 10, 11] via point defects have been calculated from first principles. The key approximation in these calculations is to simplify the defect-related phonon modes to one effective mode [8] that is parallel to the distortion of the defect geometry when gaining or losing an electron. Total energies as a function of such a one-dimensional (1D) configuration are schematically described with a configuration coordinate (CC) diagram as shown in Figure 1.

Configuration coordinate diagram between charge q and q + Δq

Figure 1. Schematics of the configuration coordinate diagram between charge qq and q+Δqq + \Delta q, where Δq=±1\Delta q = \pm1. The black circles depict energies of structures relaxed at each charge, whereas the red circles represent those relaxed at different charges. The blue circles denote the energy in virtual state (q+αΔq,Rq+Δqq(α))(q + \alpha\Delta q, R^{q}_{q+\Delta q}(\alpha)) (see text for details). The correction energy for (q,Rq+Δqq(α))(q, R^{q}_{q+\Delta q}(\alpha)) is equal to the sum of correction energies for route (i) and (ii).

To accurately determine the formation energies of charged defects using supercells under periodic boundary conditions (PBCs), it is necessary to account for finite cell size errors that

result from spurious electrostatic interactions. Formalisms have been developed for making such corrections, and the Freysoldt-Neugebauer-Van de Walle (FNV) method [12] is currently recognized as the most advanced approach. We have expanded the FNV method to accommodate anisotropic materials with atomic relaxation (eFNV) [13]. Optical transition levels via point defects, also known as vertical transition (VT) energies, can be computed assuming that the atomic configuration remains unchanged during the electron transitions, in accordance with the Franck-Condon principle [14, 15]. Their calculations require a distinct correction methodology, for which we have proposed a method termed the GKFO method [16]. Falletta et al. have also reported a comparable correction energy from a different viewpoint [17].

So far, when calculating the CC diagram, all the energies along the CC were constantly shifted by the correction energies obtained at the ground state using the FNV method. Unfortunately, this approach is incorrect. To understand why, let us begin by defining some notation.

The atomic configuration RR at the displaced configuration from the relaxed structure at charge qq to the relaxed structure at charge q+Δqq+\Delta q (Δq=±1\Delta q = \pm1) in a ratio of α\alpha is denoted as Rq+Δqq(α)R^{q}_{q+\Delta q}(\alpha). Note that α\alpha may take a value other than 0 to 1. The relaxed atomic configuration at charge qq is denoted as Rq+Δqq(0)R^{q}_{q+\Delta q}(0), and Rq+Δqq+Δq(α)=Rq+Δqq+Δq(1−α)R^{q+\Delta q}_{q+\Delta q}(\alpha) = R^{q+\Delta q}_{q+\Delta q}(1-\alpha) is always satisfied. The defect state at charge qq and configuration Rq+Δqq(α)R^{q}_{q+\Delta q}(\alpha) and its energy are denoted as (q,Rq+Δqq(α))(q, R^{q}_{q+\Delta q}(\alpha)), and E(q,Rq+Δqq(α))E(q, R^{q}_{q+\Delta q}(\alpha)), respectively. We also defined its relative energy with respect to E(q,Rq+Δqq(0))E(q, R^{q}_{q+\Delta q}(0)) as ΔE(q,Rq+Δqq(α))\Delta E(q, R^{q}_{q+\Delta q}(\alpha)) (see Figure 1).

In the case of q=0q = 0, E(0,R0,Δq0(0))E(0, R^{0}_{0,\Delta q}(0)) does not require any electrostatic correction. However, when e.g. α=1\alpha= 1, the neighboring ions move slightly as if charge Δq\Delta q is located at the defect site. Such ion distortion occurs linearly with respect to α\alpha and is effectively regarded as a polarization charge [17], which interacts with its images under PBCs and requires energy corrections as a function of the CC. Our objective is to develop a formalism for correcting these energies and verify its effectiveness.

To obtain the correction energy for ΔE(q,Rq+Δqq(α))\Delta E(q, R^{q}_{q+\Delta q}(\alpha)) (ΔEcor\Delta E_{\mathrm{cor}}), we introduce a virtual state (q+αΔq,Rq+Δqq(α))(q + \alpha\Delta q, R^{q}_{q+\Delta q}(\alpha)), in which the ionic dipoles is exactly balanced with the defect charge q+αΔqq + \alpha\Delta q. While computing such a state explicitly is computationally demanding, it is not necessary in the final formulation. ΔEcor\Delta E_{\mathrm{cor}} is the sum of the (i) eFNV and (ii) GKFO correction energies, as illustrated in Figure 1. Initially, we consider the cubic system and later generalize it.

The correction energies for the relaxed structures with charge qq and q+αΔqq+\alpha\Delta q are expressed in the eFNV approach as EPCq(ε0)−qCqE_{\mathrm{PC}}^{q}(\varepsilon_{0})-qC^{q} and EPCq+αΔq(ε0)−(q+αΔq)Cq+αΔqE_{\mathrm{PC}}^{q+\alpha\Delta q}(\varepsilon_{0})-(q+\alpha\Delta q)C^{q+\alpha\Delta q}, respectively. Here, EPCq(ε)E_{\mathrm{PC}}^{q}(\varepsilon) is the point-charge (PC) correction energy for charge qq screened by a dielectric constant ε\varepsilon and CqC^{q} and Cq+αΔqC^{q+\alpha\Delta q} are alignment constants chosen such that the short-range potential decays to zero far from the defect (see [12, 13]). ε0\varepsilon_{0} is the static dielectric constant, which is the sum of the ion-clamped (ε∞\varepsilon_{\infty}) and ionic (εion\varepsilon_{\mathrm{ion}}) dielectric constants.

The correction energy for route (i) is then described as

ΔEcor(i)=EPCq+αΔq(ε0)−EPCq(ε0)−[(q+αΔq)Cq+αΔq−qCq].(1)\Delta E_{\mathrm{cor}}^{(\mathrm{i})}=E_{\mathrm{PC}}^{q+\alpha\Delta q}(\varepsilon_{0})-E_{\mathrm{PC}}^{q}(\varepsilon_{0})-\left[(q+\alpha\Delta q)C^{q+\alpha\Delta q}-qC^{q}\right]. \tag*{(1)}

Route (ii) corresponds to the VT, where the charge transitions from q+αΔqq+\alpha\Delta q to qq, and it is corrected with the GKFO method as follows:

ΔEcor(ii)=−2αΔqq+αΔqEPCq+αΔq(ε0)+EPC−αΔq(ε∞)−(−αΔqC−αΔq−αΔqCq+αΔq+ε∞ε0(q+αΔq)C−αΔq).(2)\Delta E_{\mathrm{cor}}^{(\mathrm{ii})}=-\frac{2\alpha\Delta q}{q+\alpha\Delta q}E_{\mathrm{PC}}^{q+\alpha\Delta q}(\varepsilon_{0})+E_{\mathrm{PC}}^{-\alpha\Delta q}(\varepsilon_{\infty})-\left(-\alpha\Delta qC^{-\alpha\Delta q}-\alpha\Delta qC^{q+\alpha\Delta q}+\frac{\varepsilon_{\infty}}{\varepsilon_{0}}(q+\alpha\Delta q)C^{-\alpha\Delta q}\right). \tag*{(2)}

Here, C−αΔqC^{-\alpha\Delta q} is an alignment constant caused by introducing additional charge −αΔq-\alpha\Delta q without altering the atomic configuration Rq+Δqq(α)R_{q+\Delta q}^{q}(\alpha) [16]. As derived by Komsa et al. [18], it can be written as:

C−αΔq=2π∫d3rΔρ−αΔqr2ε∞L3,(3)C^{-\alpha\Delta q}=\frac{2\pi\int d^{3}r\Delta\rho^{-\alpha\Delta q}r^{2}}{\varepsilon_{\infty}L^{3}}, \tag*{(3)}

where LL is the side length of the supercell, and Δρ−αΔq\Delta\rho^{-\alpha\Delta q} is the defect charge distribution modified by adding charge −αΔq-\alpha\Delta q.

The sum of the PC correction terms can be expressed as

ΔEcorPC=EPC−αΔq(ε∞)−EPC−αΔq(ε0)(4)\Delta E_{\mathrm{cor}}^{\mathrm{PC}}=E_{\mathrm{PC}}^{-\alpha\Delta q}(\varepsilon_{\infty})-E_{\mathrm{PC}}^{-\alpha\Delta q}(\varepsilon_{0}) \tag*{(4)}

(see Supplemental Materials for the detailed derivation). Because Δq=±1\Delta q=\pm1 and the sign of Δq\Delta q is irrelevant for EPC−αΔqE_{\mathrm{PC}}^{-\alpha\Delta q}, ΔEcorPC\Delta E_{\mathrm{cor}}^{\mathrm{PC}} depends on neither initial charge qq nor Δq\Delta q but on α\alpha. Note, however, that ΔEcor\Delta E_{\mathrm{cor}} represents the additional correction to the relative energy with respect to the relaxed structure, and the total correction energy is the sum of ΔEcor\Delta E_{\mathrm{cor}} and the FNV correction energy to E(q,Rq+Δqq(0))E(q,R_{q+\Delta q}^{q}(0)), where the PC term depends on q2q^{2}.

The summation of the alignment terms is written as

ΔEcoralign=−(q+αΔq)Cq+αΔq+qCq+(αΔqC−αΔq+αΔqCq+αΔq−ε∞ε0(q+αΔq)C−αΔq).(5)\Delta E_{\mathrm{cor}}^{\mathrm{align}}=-(q+\alpha\Delta q)C^{q+\alpha\Delta q}+qC^{q}+\left(\alpha\Delta qC^{-\alpha\Delta q}+\alpha\Delta qC^{q+\alpha\Delta q}-\frac{\varepsilon_{\infty}}{\varepsilon_{0}}(q+\alpha\Delta q)C^{-\alpha\Delta q}\right). \tag*{(5)}

We define CalignC^{\mathrm{align}} as the alignment term estimated between (q,Rq+Δqq(α))(q,R_{q+\Delta q}^{q}(\alpha)) and (q,Rq+Δqq(0))(q,R_{q+\Delta q}^{q}(0)) by removing the long-range potential of charge −αΔq-\alpha\Delta q screened by εeff\varepsilon_{\mathrm{eff}}, where 1/εeff=1/ε∞−1/ε01/\varepsilon_{\mathrm{eff}}=1/\varepsilon_{\infty}-1/\varepsilon_{0}, and can be written as

Calign=2π∫d3rΔρ−αΔqr2εeffL3(6)C^{\mathrm{align}}=\frac{2\pi\int d^{3}r\Delta\rho^{-\alpha\Delta q}r^{2}}{\varepsilon_{\mathrm{eff}}L^{3}} \tag*{(6)}

as in Eq. (3). Using CalignC^{\mathrm{align}}, we obtain ΔEcoralign=αΔqCalign\Delta E_{\mathrm{cor}}^{\mathrm{align}}=\alpha\Delta qC^{\mathrm{align}} (see Supplemental Materials for details of the derivation). Thus, the ΔEcor\Delta E_{\mathrm{cor}} is expressed as

ΔEcor=EPC−αΔq(εeff)−(−αΔq)Calign.(7)\Delta E_{\mathrm{cor}}=E_{\mathrm{PC}}^{-\alpha\Delta q}(\varepsilon_{\mathrm{eff}})-(-\alpha\Delta q)C^{\mathrm{align}}. \tag*{(7)}

The obtained ΔEcor\Delta E_{\mathrm{cor}} is simply estimated by considering the effective charge −αΔq-\alpha\Delta q screened by an effective dielectric constant εeff\varepsilon_{\mathrm{eff}}. This can be easily extended to the anisotropic systems by replacing εeff\varepsilon_{\mathrm{eff}} with the tensor form, εˉeff\bar{\varepsilon}_{\mathrm{eff}}. Furthermore, it is applicable to the anharmonic potential energy surface [9] because the ionic displacement distance for the screening linearly depends on α\alpha.

To verify the effectiveness of the correction formalism in Eq. (7), we have calculated the total energies of the nitrogen vacancy (VNV_{\mathrm{N}}) in GaN. To demonstrate that the initial charge state does not affect ΔEcor\Delta E_{\mathrm{cor}}, we calculated the energies at q=+2q=+2 along the CC to the structure at q=+3q=+3. First, we confirm that the conventional eFNV method works well for calculating the formation energy of VN+2V_{\mathrm{N}}^{+2} (E(VN+2)E(V_{\mathrm{N}}^{+2})) in GaN as shown in Figure 2. Specifically, E(VN+2)E(V_{\mathrm{N}}^{+2}) is well corrected when using the 300-atom supercell.

Relative formation energies, atomic site potentials, supercell-size dependence, energy correction, and configuration coordinate

Figure 2. (a) Relative formation energies of VN+2V_{\mathrm{N}}^{+2} in GaN without any corrections, with the PC corrections, and with the eFNV corrections as a function of the supercell size. The zero energy value is set to the energy corrected with the eFNV method at the largest supercell. (b, c) Atomic site potentials caused by VN+2V_{\mathrm{N}}^{+2} at (a) (2,R32(0.5))(2, R_{3}^{2}(0.5)) and (b) (2,R32(1.5))(2, R_{3}^{2}(1.5)) in the 576-atom supercell that are calculated by subtracting the potentials in the pristine supercell. Blue marks are the electrostatic potentials caused by periodic point charges −αΔq-\alpha\Delta q and background charge screened by an effective dielectric tensor, while red crosses are their differences from the atomic site potentials. The vertical dashed lines indicate the radius of the sphere in contact with the supercell. The averaged potential difference far from the defect (CalignC_{\mathrm{align}}) is evaluated outside of the sphere (see Ref. [13] for details). (d) The supercell size dependence of ΔE(2,R32(1.5))\Delta E(2, R_{3}^{2}(1.5)) without any corrections, with PC corrections, and with corrections in Eq. (7). (e) E(2,R32(α))E(2, R_{3}^{2}(\alpha)) as a function of α\alpha without and with corrections in Eq. (7). (f) The configuration coordinate QQ of VN+2V_{\mathrm{N}}^{+2} in GaN at R32(1)R_{3}^{2}(1). In (d) and (e), the zero energy values are set to the energies corrected with Eq. (4) at the largest supercells.

In Figure 2(b) and (Figure 2c), we present the atomic site potentials caused by VN+2V_{\mathrm{N}}^{+2} at two different configuration coordinates, namely (2,R32(0.5))(2,R_{3}^{2}(0.5)) and (2,R32(1.5))(2,R_{3}^{2}(1.5)), in the 576-atom supercell. Despite having identical charge q=+2q=+2, a significant difference in the potential variation is observed between the two coordinates. The atomic site potentials calculated from a charge −αΔq-\alpha\Delta q at the defect site screened by εˉeff\bar{\varepsilon}_{\mathrm{eff}} are also displayed. These potentials adequately reproduce tendencies of the atomic site potentials in the region far from the defect, leading to an estimation of CalignC^{\mathrm{align}}.

Figure 2(d) presents ΔE(2,R32(1.5))\Delta E(2,R_{3}^{2}(1.5)) of VN+2V_{\mathrm{N}}^{+2} as a function of number of atoms in the pristine supercell (NatomsN_{\mathrm{atoms}}). When applying the constant energy correction irrelevant to the CC that corresponds to “Uncorrected” in this Letter, there are noticeable finite-cell size errors. These errors are effectively eliminated when ΔEcor\Delta E_{\mathrm{cor}} in Eq. (7) are applied, thereby validating our formalism. As shown in Figure 2(d), the alignment terms do not exert a significant effect, meaning that the ionic dipoles that screen the additional defect charge Δq\Delta q do not appreciably alter the defect charge distribution.

Figure 2(e) depicts ΔE(2,R32(α))\Delta E(2,R_{3}^{2}(\alpha)) with and without corrections, as a function of the displacement fraction α\alpha. As Eq. (7) suggests, the finite size errors are predominantly dependent on α2\alpha^{2}. When computing the non-radiative carrier capture coefficients, the position at which the potential energy surfaces of two states intersect has a significant effect on the capture rate. Such intersections can occur at values of α\alpha greater than 2, as exemplified by the carbon impurity at the nitrogen site (CN)(C_{\mathrm{N}}) in GaN [8]. Hence, it is likely that the intersection coordinates and energies are significantly modified by applying ΔEcor\Delta E_{\mathrm{cor}}. It is also worth noting that a miscalculation of 0.1 eV in the energies along the CC can result in a tenfold error in the non-radiative carrier capture coefficients [8].

As shown in Figure 2(a) and (Figure 2d), E(VN+2)E(V_{\mathrm{N}}^{+2}) and ΔE(q,R3q(α)q+Δq)\Delta E(q, R_{3}^{q}(\alpha)_{q+\Delta q}) are effectively corrected even using the 72-atom supercell. However, the difference of the configuration coordinate QQ (see Method) between charge qq and q+Δqq+\Delta q (ΔQ\Delta Q) is not accurately estimated using the 72-atom supercell as shown in Figure 2(f). This might also be related to the fact that the 72-atom supercell calculation of E(VN+2)E(V_{\mathrm{N}}^{+2}) shows a different tendency in Figure 2(a). Because errors in QQ have a significant impact on the calculations of the line shapes of luminescence spectra and carrier capture coefficients, we generally recommend using larger supercells, particularly to improve QQ.

The effective dielectric constant εeff\varepsilon_{\mathrm{eff}} quantifies the polarization resulting from ion displacement. It is related to the long-range electron-phonon coupling, and is commonly used when discussing self-trapped polarons [19]. In general, decreasing εeff\varepsilon_{\mathrm{eff}} leads to an increase in EcorE_{\mathrm{cor}}. Because εeff=ε∞+ε∞2/εion\varepsilon_{\mathrm{eff}} = \varepsilon_{\infty} + \varepsilon_{\infty}^{2}/\varepsilon_{\mathrm{ion}}, εeff\varepsilon_{\mathrm{eff}} decreases as ε∞\varepsilon_{\infty} decreases and εion\varepsilon_{\mathrm{ion}} increases. To assess the potential magnitude of the error in realistic materials resulting from neglecting ΔEcor\Delta E_{\mathrm{cor}} in Eq. (7), we calculated εeff\varepsilon_{\mathrm{eff}} for 931 oxide materials for which we calculated the oxygen vacancies in our previous work [20], using the spherical averages of ε∞\varepsilon_{\infty} and εion\varepsilon_{\mathrm{ion}}. The smallest εeff\varepsilon_{\mathrm{eff}}, which is 2.5, is found in CsLi5_{5}(BO3_{3})2_{2}, where the averaged ε∞\varepsilon_{\infty} and εion\varepsilon_{\mathrm{ion}} are 2.1 and 10.8, respectively. Then, ΔEcor\Delta E_{\mathrm{cor}} for the oxygen vacancy at state (1,R21(α=2))(1, R_{2}^{1}(\alpha= 2)) is estimated to be 1.2 eV even using the very large 448-atom supercell.

We can estimate how much the energy correction modifies the effective phonon frequency, which is given by the equation Ω2=∂2E∂Q2\Omega^{2} = \frac{\partial^{2}E}{\partial Q^{2}} in the harmonic approximation [6, 8]. Replacing Ω\Omega with Ω+Ωcor\Omega+\Omega_{\mathrm{cor}}, where Ωcor\Omega_{\mathrm{cor}} is the corrected frequency, and EE with E+ΔEcorE+\Delta E_{\mathrm{cor}}, and assuming that Ωcor≪Ω\Omega_{\mathrm{cor}} \ll\Omega and the PC term is dominant in Eq. (7), we obtain:

Ωcor=EPCΔq(εeff)Ω(ΔQ)2,(8)\Omega_{\mathrm{cor}} = \frac{E_{\mathrm{PC}}^{\Delta q}(\varepsilon_{\mathrm{eff}})}{\Omega(\Delta Q)^{2}}, \tag*{(8)}

(see Supplemental Materials for details of the derivation). For the case of (2,R32(α))(2, R_{3}^{2}(\alpha)) for VNV_{\mathrm{N}}, we find ℏΩcor=3.28\hbar\Omega_{\mathrm{cor}} = 3.28 meV when using the 72-atom supercell, which is 14.5 % of the uncorrected ℏΩcor\hbar\Omega_{\mathrm{cor}}. Note that, because ℏΩcor\hbar\Omega_{\mathrm{cor}} is approximately proportional to EPCΔq(εeff)E_{\mathrm{PC}}^{\Delta q}(\varepsilon_{\mathrm{eff}}), its convergence as a function of NatomN_{\mathrm{atom}} is slow if ΔEcor\Delta E_{\mathrm{cor}} is not applied.

To conclude, we derived the correction formalism for the defect energies along the 1D CC, which has not been considered so far. The correction energy is described with the effective charge −αΔq-\alpha\Delta q and effective dielectric constant εeff\varepsilon_{\mathrm{eff}}. Its effectiveness has been verified with a nitrogen vacancy in GaN. Since our correction method is automatically applied with negligible computational cost compared to first-principles calculations, we believe that it would be routinely used in the future.

Method. All the calculations were performed using the projector augmented-wave (PAW) method [21, 22] implemented in VASP [23]. We adopted the HSE06 hybrid functional [24]. A fraction of Fock exchange parameter was set to 0.31 to reproduce the experimental band gap [8]. See the Supplemental Materials for the computational details. The harmonic phonon frequencies were calculated using Nonrad [25]. All the VASP input settings were generated with the VISE code (version 0.7.0) [26], while the processing related to defects was done with pydefect [27]. The configuration coordinate QQ is defined as ∑αMα⋅ΔRα2\sqrt{\sum_{\alpha} M_{\alpha}\cdot\Delta R_{\alpha}^{2}}, where MαM_{\alpha} and ΔRα\Delta R_{\alpha} are the atomic mass and the displacement vector from the equilibrium position of atom α\alpha, respectively [6, 8]. The ion-clamped dielectric constants (ε∞\varepsilon_{\infty}) for 931 oxides were obtained from the long-range wavelength limit in the real part of dielectric functions calculated using the dielectric dependent hybrid functional [28] (see Ref. [29] for details). The ionic contributions (εion\varepsilon_{\mathrm{ion}}) were calculated using the density functional perturbation theory [30] (see Ref. [20] for details).

Acknowledgement. Discussion with Soungmin Bae is sincerely appreciated. This study was financially supported by PRESTO (JPMJPR16N4) from the Japan Science and Technology Agency, KAKENHI (Grant No. 22H01755), and the E-IMR project at IMR, Tohoku University.

References

  1. [1]J. G. Bednorz and K. A. Müller, “Possible high Tc superconductivity in the Ba–La–Cu–O system,” Zeitschrift für Physik B Condensed Matter 64, 189 (1986).DOI
  2. [2]Y. Ando, “Topological Insulator Materials,” Journal of the Physical Society of Japan 82, 102001 (2013).DOI
  3. [3]P. K. Nayak, S. Mahesh, H. J. Snaith, and D. Cahen, “Photovoltaic solar cell technologies: analysing the state of the art,” Nature Reviews Materials 4, 269 (2019).DOI
  4. [4]S. Nakamura, N. Iwasa, M. S. M. Senoh, and T. M. T. Mukai, “Hole Compensation Mechanism of P-Type GaN Films,” Japanese Journal of Applied Physics 31, 1258 (1992).DOI
  5. [5]C. Freysoldt, B. Grabowski, T. Hickel, J. Neugebauer, G. Kresse, A. Janotti, and C. G. Van de Walle, “First-principles calculations for point defects in solids,” Rev. Mod. Phys. 86, 253 (2014).DOI
  6. [6]A. Alkauskas, J. L. Lyons, D. Steiauf, and C. G. V. d. Walle, “First-Principles Calculations of Luminescence Spectrum Line Shapes for Defects in Semiconductors: The Example of GaN and ZnO,” Physical review letters 109, 267401 (2012).pubmed.ncbi.nlm.nih.gov/23368616
  7. [7]Y. K. Frodason, K. M. Johansen, T. S. Bjørheim, B. G. Svensson, and A. Alkauskas, “Zn vacancy as a polaronic hole trap in ZnO,” Physical Review B 95, 094105 (2017).DOI
  8. [8]A. Alkauskas, Q. Yan, and C. G. V. d. Walle, “First-principles theory of nonradiative carrier capture via multiphonon emission,” Phys. Rev. B 90, 075202 (2014).DOI
  9. [9]S. Kim, S. N. Hood, and A. Walsh, “Anharmonic lattice relaxation during nonradiative carrier capture,” Physical Review B 100, 041202 (2019).DOI
  10. [10]X. Zhang, M. E. Turiansky, and C. G. V. d. Walle, “Correctly Assessing Defect Tolerance in Halide Perovskites,” The Journal of Physical Chemistry C 124, 6022 (2020).DOI
  11. [11]S. R. Kavanagh, D. O. Scanlon, A. Walsh, and C. Freysoldt, “Impact of metastable defect structures on carrier recombination in solar cells,” Faraday Discussions 239, 339 (2022).
  12. [12]C. Freysoldt, J. Neugebauer, and C. V. d. Walle, “Fully Ab Initio Finite-Size Corrections for Charged-Defect Supercell Calculations,” Physical Review Letters 102, 016402 (2009).pubmed.ncbi.nlm.nih.gov/19257218
  13. [13]Y. Kumagai and F. Oba, “Electrostatics-based finite-size corrections for first-principles point defect calculations,” Phys. Rev. B 89, 195205 (2014).
  14. [14]J. Franck and E. G. Dymond, “Elementary processes of photochemical reactions,” Transactions of the Faraday Society 21, 536 (1926).DOI
  15. [15]E. Condon, “A Theory of Intensity Distribution in Band Systems,” Physical Review 28, 1182 (1926).
  16. [16]T. Gake, Y. Kumagai, C. Freysoldt, and F. Oba, “Finite-size corrections for defect-involving vertical transitions in supercell calculations,” Physical Review B 101, 020102 (2020).
  17. [17]S. Falletta, J. Wiktor, and A. Pasquarello, “Finite-size corrections of defect energy levels involving ionic polarization,” Physical Review B 102, 041115 (2020).DOI
  18. [18]H.-P. Komsa, T. T. Rantala, and A. Pasquarello, “Finite-size supercell correction schemes for charged defect calculations,” Physical Review B 86, 045112 (2012).DOI
  19. [19]D. Emin, Polarons, edited by C. U. Press (2013).
  20. [20]Y. Kumagai, N. Tsunoda, A. Takahashi, and F. Oba, “Insights into oxygen vacancies from high-throughput first-principles calculations,” Physical Review Materials 5, 123803 (2021).
  21. [21]P. E. Blöchl, “Projector augmented-wave method,” Physical Review B 50, 17953 (1994).pubmed.ncbi.nlm.nih.gov/9976227
  22. [22]G. Kresse and D. Joubert, “From ultrasoft pseudopotentials to the projector augmented-wave method,” Physical Review B 59, 1758 (1999).DOI
  23. [23]G. Kresse and J. Furthmüller, “Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set,” Physical Review B 54, 11169 (1996).pubmed.ncbi.nlm.nih.gov/9984901
  24. [24]A. V. Krukau, O. A. Vydrov, A. F. Izmaylov, and G. E. Scuseria, “Influence of the exchange screening parameter on the performance of screened hybrid functionals,” The Journal of Chemical Physics 125, 224106 (2006).pubmed.ncbi.nlm.nih.gov/17176133
  25. [25]M. E. Turiansky, A. Alkauskas, M. Engel, G. Kresse, D. Wickramaratne, J.-X. Shen, C. E. Dreyer, and C. G. V. d. Walle, “Nonrad: Computing nonradiative capture coefficients from first principles,” Computer Physics Communications 267, 108056 (2021).
  26. [26]Y. Kumagai, vise (2021), https://github.com/kumagai-group/vise.
  27. [27]Y. Kumagai, pydefect (2021), https://github.com/kumagai-group/pydefect.
  28. [28]J. H. Skone, M. Govoni, and G. Galli, “Self-consistent hybrid functional for condensed systems,” Physical Review B 89, 195112 (2014).
  29. [29]Y. Kumagai, “Computational Screening of p-Type Transparent Conducting Oxides Using the Optical Absorption Spectra and Oxygen-Vacancy Formation Energies,” Physical Review Applied 19, 034063 (2023).
  30. [30]X. Gonze and C. Lee, “Dynamical matrices, Born effective charges, dielectric permittivity tensors, and interatomic force constants from density-functional perturbation theory,” Physical Review B 55, 10355 (1997).DOI

Paper details

vlnc:W_jdpnjmb2
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 3

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents