betaFeedback

WorksPhysics and Astronomy

Inter-defect interactions, oxygen-vacancy distribution, and oxidation in acceptor-doped ABO3 perovskites

Abstract

The effects of inter-defect interaction and impurity disorder on defect thermodynamics, local ion coordination, and oxidation in acceptor-doped wide-gap ABO3 perovskites are explored using the developed statistical theory and Monte Carlo simulations. The results demonstrate that under realistic energy parameters the interaction between oxygen vacancies and impurities generally has a greater impact on the studied properties than inter-vacancy correlations. The influence of inter-vacancy interaction significantly depends on dopant content x: inter-site vacancy repulsion becomes noticeable at sufficiently high x, whereas on-site correlations can be pronounced within a narrow doping range at moderate x values. It is found that a non-uniform impurity allocation, which can result from a sample preparation procedure, considerably affects oxygen-vacancy distribution, and has a weaker effect on short-range order and oxidation. It is also shown that inter-defect interaction reduces the hole concentration, increases the oxidation enthalpy, and can result in their non-trivial dependence on x. The obtained results agree with experimental data on local ion coordination and help to explain the behavior of hole conductivity in the considered perovskites. The findings of this study contribute to understanding the fundamental properties of acceptor-doped oxides, facilitating the development of new materials for clean energy applications.

Introduction

Acceptor-doped perovskites AB1−xRxO3−δAB_{1-x}R_{x}O_{3-\delta} have significant potential as promising materials for clean energy applications [1, 2, 3, 4]. Acceptor doping of ABO3ABO_{3} oxides leads to the formation of oxygen vacancies, which enables such materials to be employed as oxygen, hydrogen or ionic-electronic conductors in solid oxide electrochemical cells [5]. At the same time, acceptor impurities alter various material properties, in particular those important for applications. As has been established experimentally [6, 7, 8, 9, 10, 11] and theoretically [12, 13, 14, 15, 16, 17, 18, 19, 20], interaction between charge carriers and acceptor impurities is one of the most important effects influencing the fundamental properties of the studied oxides. Numerous studies of these effects in proton-conducting perovskites have demonstrated that the interaction of protons and oxygen vacancies with impurity ions strongly affects hydration and transport phenomena (e.g., Refs. [7, 17, 18, 19, 20, 21, 22]). In our previous works, we showed that these interactions impact not only hydration, oxidation and transfer processes [17, 19, 23], but also the performance of electrochemical cells based on proton-conducting membranes [18]. The vacancy–impurity interaction itself has important implications for defect and transport phenomena, including oxidation, oxide-ion and hole conduction [23, 24, 25, 26, 27]. Previously, we addressed the effect of oxygen vacancy trapping on the oxidation of AB1−xRxO3−δAB_{1-x}R_xO_{3-\delta} perovskites assuming only one type of acceptor-bound vacancy states and uniform dopant distribution [18, 23]. In particular, it was shown [18] that vacancy trapping noticeably affects the Fermi level position and hole concentration, which is essential for the transport properties of oxides and their applications in electrochemical devices.

The distribution of oxygen vacancies over lattice sites with different impurity surroundings should also be largely determined by the interactions between vacancies and impurities. The corresponding short-range order related to the local coordination of vacancies, in turn, affects the transport and thermodynamic properties of oxides. The effects of interactions and non-uniform dopant distribution on the local environments of ions in acceptor-doped perovskites were studied experimentally using nuclear magnetic resonance (NMR) technique [6, 11, 28, 29]. Theoretically, the effect of frozen dopant configuration, corresponding to equilibrium at typical sintering temperatures (∼1900\sim1900 K), on the local structure of Y- and Sc-doped BaZrO3_3 was explored using density functional theory (DFT) and Monte-Carlo simulations [30, 31].

Despite numerous studies of inter-defect interactions in acceptor-doped perovskites, a number of fundamental issues related to the implications of these interactions, especially for heavily doped oxides, remain unresolved. Note that in applications, where the studied materials can serve as, e.g., proton-conducting electrolytes in electrochemical cells, oxides with a dopant content of up to ∼20\sim20 % are typically employed. Recent studies reported perovskites that retain a single-phase structure at significantly higher impurity concentrations [10, 32, 33, 34]. In such heavily doped oxides, the effect of non-ideality should be essential and can result in new consequences compared to lightly doped compounds. In particular, the interaction of vacancies with impurities and among each other should significantly affect their distribution over oxygen sites, local structure and oxidation. Furthermore, the effects of defect interactions depend on the distribution of acceptor impurities in oxides. However, the impact of this distribution on the properties under study, as far as we know, has not yet been investigated.

In this work, we explore the effects of inter-defect interaction and quenched impurity disorder on oxygen-vacancy distribution, local ion coordination, and oxidation in acceptor-doped wide-gap perovskites under dry oxidizing conditions. Using the developed statistical theory and Monte Carlo simulations, we investigate the following effects related to non-ideality of defect system. (1) The formation of two- and three-particle complexes of acceptor-bound oxygen vacancies. (2) The impact of on-site Fermi-type correlations arising at high vacancy occupancy of oxygen sites surrounded by impurities. (3) The effects of inter-vacancy repulsion in nearest-neighbor positions. (4) The influence of non-uniform impurity distribution, which can result from a sample preparation procedure.

Our analysis of the above effects highlights the importance of inter-defect interaction for the defect and local structure of acceptor-doped ABO3\mathrm{ABO}_3 oxides. We find that under realistic energy parameters the main effect on the studied properties is generally due to vacancy-impurity interactions. At the same time, short-range inter-vacancy correlations can provide a noticeable contribution to defect thermodynamics within a specific range of dopant concentrations. We show that inter-defect interaction considerably alters the oxide-gas equilibrium constant and oxidation enthalpy, and results in their strong dependence on dopant content. We also demonstrate that non-uniform impurity allocation can significantly influence the vacancy distribution, but has little effect on local coordination and oxidation behavior.

The results of the presented statistical thermodynamic analysis are in agreement with our Monte Carlo simulations. We also verify our findings on the defect and local structure of AB1−xRxO3−δ\mathrm{AB}_{1-x}\mathrm{R}_x\mathrm{O}_{3-\delta} perovskites using the NMR data for acceptor-doped BaZrO3\mathrm{BaZrO}_3 [28, 29]. In addition, our findings on oxidation help to explain the behavior of hole conductivity in Y- and In-doped BaSnO3\mathrm{BaSnO}_3 [35, 36].

Theory

The model

In this study, we consider wide-band-gap acceptor-doped perovskites AB1−xRxO3−δ\mathrm{AB}_{1-x}\mathrm{R}_x\mathrm{O}_{3-\delta} under dry oxidizing conditions. The concentration of holes in the oxide is assumed to be negligibly small compared to that of oxygen vacancies cVc_{\mathrm{V}} (see § 3.5) and therefore cV≈x/2=const.c_{\mathrm{V}} \approx x/2 = \mathrm{const.} In the temperature range under consideration, the mobility of host and dopant cations is low and, accordingly, their distribution is supposed to be frozen.

Let us first consider a model that neglects inter-site interaction between vacancies and accounts only for on-site (Fermi-type) correlations, which are due to the prohibition of site occupation by more than one particle. The Hamiltonian for this model is

H0=∑i=13NEVinVi,(1)H_0 = \sum_{i=1}^{3N} E_{\mathrm{V}i} n_{\mathrm{V}i}, \tag*{(1)}

where nVin_{\mathrm{V}i} is the fermion occupation number of an oxygen site ii by a vacancy (nVi=1n_{\mathrm{V}i}=1 if site ii is vacant and nVi=0n_{\mathrm{V}i}=0 otherwise); EViE_{\mathrm{V}i} is the vacancy energy at site ii; 3N3N is the total number of oxygen sites in a crystal with NN unit cells.

Due to various possible dopant configurations in the region surrounding an oxygen vacancy V, the energy spectrum {EVi}\{E_{\mathrm{V}i}\} in AB1−xRxO3−δ\mathrm{AB}_{1-x}\mathrm{R}_x\mathrm{O}_{3-\delta} perovskites can be quite complex. In this work, however, to reveal the main effects of the considered interactions, we adopt a model with a reduced, three-level, energy spectrum. These levels – EVRE_{\mathrm{VR}}, EV2RE_{\mathrm{V2R}} and EVfE_{\mathrm{Vf}} – correspond to the energies of vacancies with different sets of the B-sublattice cations in the nearest environment: R-V-B, R-V-R, and BV-B, respectively (see Figure 1). The vacancies in these environments are denoted as VRV_{\mathrm{R}}, V2RV_{2\mathrm{R}}, and VfV_{\mathrm{f}}.

Figure 1. Possible nearest environments of oxygen vacancies in acceptor-doped AB1−xRxO3−δ\mathrm{AB}_{1-x}\mathrm{R}_{x}\mathrm{O}_{3-\delta} perovskites: (a) B-V-B (b) B-V-R and (c) R-V-R. The inter-site vacancy interaction characterized by the repulsion energy λ\lambda (see Eq. (3)) is schematically shown in (a).

In what follows, we explore the properties of oxides as a function of the trapping energies of vacancies near one (ΔEVR\Delta E_{V_{\mathrm{R}}}) and two (ΔEV2R\Delta E_{V_{2\mathrm{R}}}) impurities:

ΔEVR=EVf−EVR, ΔEV2R=EVf−EV2R.(2)\Delta E_{V_{\mathrm{R}}}=E_{V_{\mathrm{f}}}-E_{V_{\mathrm{R}}},\ \Delta E_{V_{2\mathrm{R}}}=E_{V_{\mathrm{f}}}-E_{V_{2\mathrm{R}}}. \tag*{(2)}

It is assumed that the effects of crystal and electronic structure reorganization due to the vacancy formation in various configurations are accounted for in ΔEVR\Delta E_{V_{\mathrm{R}}} and ΔEV2R\Delta E_{V_{2\mathrm{R}}} [12, 13, 14, 15].

To analyze the effects of inter-vacancy interaction, we consider a model with infinite repulsion between vacancies in nearest-neighboring sites (the simultaneous occupation of such sites by vacancies is prohibited). The corresponding Hamiltonian HH including the interaction term HintH_{\mathrm{int}} can be specified as

H=H0+Hint=H0+λ∑⟨i,j⟩nVinVj,(3)H=H_{0}+H_{\mathrm{int}}=H_{0}+\lambda\sum_{\langle i,j\rangle}n_{V_i}n_{V_j}, \tag*{(3)}

where H0H_{0} is given by Eq. (1); λ\lambda is the inter-vacancy repulsion energy, and the notation ⟨i,j⟩\langle i,j\rangle implies a sum over all distinct pairs of the nearest-neighboring oxygen sites. The limit λ→∞\lambda\to\infty corresponds to the model with infinite repulsion, and λ=0\lambda=0 is suitable for the model without inter-site vacancy interaction.

In what follows, for brevity, we refer to on-site Fermi-type correlations as weak, and correlations for the model with both on-site and inter-site vacancy repulsion as strong.

Since we consider a system with a quenched impurity disorder, finding the expectation value L‾\overline{L} of any observable LL implies double averaging [37, 38]:

L‾=⟨⟨L⟩⟩ω(4)\overline{L}=\langle\langle L\rangle\rangle_{\omega} \tag*{(4)}

where ⟨⟩\langle\rangle is the usual Gibbs thermodynamic average and ⟨⟩ω\langle\rangle_{\omega} means averaging over the distribution ω\omega of random variables characterizing disorder. In the case of a single-particle energy spectrum {EVi}\{E_{V_i}\}, the value L‾\overline{L} is given by

⟨⟨L⟩⟩ω=∫∏i[ω(EVi) dEVi]⟨L({EVi})⟩,(5)\langle\langle L\rangle\rangle_{\omega}=\int\prod_i[\omega(E_{V_i})\,\mathrm{d}E_{V_i}]\langle L(\{E_{V_i}\})\rangle, \tag*{(5)}

where ω(EVi)\omega(E_{V_i}) is the distribution of energies corresponding to the impurity disorder, and ⟨L({EVi})⟩\langle L(\{E_{V_i}\})\rangle is the Gibbs average for a given set of energies {EVi}\{E_{V_i}\}.

For our model with three energy levels, the probability distribution ω\omega takes the form

ω(EVi,x)=pRδ(EVi−EVR)+p2Rδ(EVi−EV2R)+pfδ(EVi−EVf),(6)\omega(E_{V_i},x)=p_{\mathrm{R}}\delta(E_{V_i}-E_{V_{\mathrm{R}}})+p_{2\mathrm{R}}\delta(E_{V_i}-E_{V_{2\mathrm{R}}})+p_{\mathrm{f}}\delta(E_{V_i}-E_{V_{\mathrm{f}}}), \tag*{(6)}

where pRp_{\mathrm{R}}, p2Rp_{2\mathrm{R}} and pfp_{\mathrm{f}} are the probabilities that a randomly chosen oxygen site has the corresponding local environment (R-O-B, R-O-R and B-O-B). Obviously, these probabilities depend on the dopant content xx. In the case of the uniform distribution of impurities over B-sites, the probabilities are given by

pR=2x(1−x),p2R=x2,pf=(1−x)2.(7)p_{\mathrm{R}}=2x(1-x),\quad p_{2\mathrm{R}}=x^2,\quad p_{\mathrm{f}}=(1-x)^2. \tag*{(7)}

We explore the effects of inter-defect interactions on the oxide properties for both uniform and non-uniform dopant distributions. The implications of the non-uniform distribution are considered in § 3.4 and § 3.5.

Defect thermodynamics

To consider the equilibrium properties of the studied oxides, we employ: (1) analytical methods of statistical mechanics and (2) Monte Carlo simulations. The model neglecting inter-site vacancy repulsion is treated using both approaches, and the effect of this repulsion is studied only by Monte Carlo method (analytical treatment of the latter effect requires complex, poorly controlled approximations, and is beyond the scope of this work).

For the Hamiltonian H0H_0 (Eq. (1)) without inter-site vacancy interaction, the grand partition function ZZ and equilibrium occupation probability ⟨nVi⟩\langle n_{\mathrm{V}i}\rangle are well-known (e.g., Ref. [39]):

Z=∏i=13NTr⁡{exp⁡[(μV−EVi)nVikT]}=∏i=13N[1+exp⁡(μV−EVikT)],(8)Z = \prod_{i=1}^{3N} \operatorname{Tr}\left\{\exp\left[\frac{(\mu_{\mathrm{V}}-E_{\mathrm{V}i})n_{\mathrm{V}i}}{kT}\right]\right\} = \prod_{i=1}^{3N}\left[1+\exp\left(\frac{\mu_{\mathrm{V}}-E_{\mathrm{V}i}}{kT}\right)\right], \tag*{(8)}
⟨nVi⟩=[1+exp⁡(EVi−μVkT)]−1,(9)\langle n_{\mathrm{V}i}\rangle= \left[1+\exp\left(\frac{E_{\mathrm{V}i}-\mu_{\mathrm{V}}}{kT}\right)\right]^{-1}, \tag*{(9)}

where μV\mu_{\mathrm{V}} is the chemical potential of oxygen vacancies.

Using Eq. (4) with the distribution ω(EVi,x)\omega(E_{\mathrm{V}i},x) (6), we obtain the averaged grand potential

Ω‾=⟨−kTln⁡Z⟩ω=−3NkT∑m=13pmln⁡[1+exp⁡(μV−EVmkT)],(10)\overline{\Omega}=\langle-kT\ln Z\rangle_{\omega}=-3NkT\sum_{m=1}^{3}p_m\ln\left[1+\exp\left(\frac{\mu_{\mathrm{V}}-E_{\mathrm{V}m}}{kT}\right)\right], \tag*{(10)}

where mm denotes the type of local vacancy environment m=Vfm=\mathrm{V}_{f}, VR\mathrm{V}_{R} or V2R\mathrm{V}_{2R}, corresponding to the configurations B-V-B, R-V-B and R-V-R.

The concentration of vacancies of type mm per formula unit is determined by the occupation probability ⟨nVm⟩\langle n_{\mathrm{V}m}\rangle (9) averaged over ω(EVi,x)\omega(E_{\mathrm{V}i},x) (6):

cVm=3pm⟨nVm⟩=3pm[1+exp⁡(EVm−μVkT)]−1.(11)c_{\mathrm{V}m}=3p_m\langle n_{\mathrm{V}m}\rangle=3p_m\left[1+\exp\left(\frac{E_{\mathrm{V}m}-\mu_{\mathrm{V}}}{kT}\right)\right]^{-1}. \tag*{(11)}

The chemical potential μV\mu_{\mathrm{V}} is determined by the self-consistency condition

cVf+cVR+cV2R=cV=x/2,(12)c_{\mathrm{V}f}+c_{\mathrm{V}R}+c_{\mathrm{V}2R}=c_{\mathrm{V}}=x/2, \tag*{(12)}

which gives a cubic equation for the absolute activity exp⁡(μV/kT)\exp(\mu_{\mathrm{V}}/kT).

To consider the effects of non-ideality in the vacancy system, we also use the activity coefficient γV\gamma_{\mathrm{V}}. We define γV\gamma_{\mathrm{V}} in a standard way using EVfE_{\mathrm{V}f} as the reference energy:

μV=EVf+kTln⁡γVcV3.(13)\mu_{\mathrm{V}}=E_{\mathrm{V}f}+kT\ln\gamma_{\mathrm{V}}\frac{c_{\mathrm{V}}}{3}. \tag*{(13)}

The solution of Eqs. (11) and (12) yields the activity coefficient γV\gamma_{\mathrm{V}} (or μV\mu_{\mathrm{V}}) and the concentrations cVmc_{\mathrm{Vm}} that account for both trapping effects and Fermi-type correlations.

If the concentrations of vacancies of each type are much smaller than the concentrations of corresponding oxygen sites (cVf≪cOfc_{\mathrm{Vf}} \ll c_{\mathrm{Of}}, cVR≪cORc_{\mathrm{VR}} \ll c_{\mathrm{OR}} and cV2R≪cO2Rc_{\mathrm{V2R}} \ll c_{\mathrm{O2R}}), Fermi-type correlations are negligible, and we arrive at Boltzmann statistics. In this case, the expressions for cVmc_{\mathrm{Vm}} (11) take the simple form:

cVf=pfγVcV,(14)c_{\mathrm{Vf}} = p_{\mathrm{f}}\gamma_{\mathrm{V}}c_{\mathrm{V}}, \tag*{(14)}
cVR=pRγVcVeΔEVRkT,(15)c_{\mathrm{VR}} = p_{\mathrm{R}}\gamma_{\mathrm{V}}c_{\mathrm{V}}e^{\frac{\Delta E_{\mathrm{VR}}}{kT}}, \tag*{(15)}
cV2R=p2RγVcVeΔEV2RkT,(16)c_{\mathrm{V2R}} = p_{\mathrm{2R}}\gamma_{\mathrm{V}}c_{\mathrm{V}}e^{\frac{\Delta E_{\mathrm{V2R}}}{kT}}, \tag*{(16)}

where the activity coefficient γV\gamma_{\mathrm{V}} for the Boltzmann statistics is given by

γV=(pf+pReΔEVRkT+p2ReΔEV2RkT)−1.(17)\gamma_{\mathrm{V}} = \left(p_{\mathrm{f}} + p_{\mathrm{R}}e^{\frac{\Delta E_{\mathrm{VR}}}{kT}} + p_{\mathrm{2R}}e^{\frac{\Delta E_{\mathrm{V2R}}}{kT}}\right)^{-1}. \tag*{(17)}

Monte-Carlo simulation technique

The Monte Carlo modeling was conducted using a supercell containing 30×30×3030 \times30 \times30 unit cells with periodic boundary conditions. Dopant concentration and temperature were varied from 0.05 to 0.5 (in steps of 0.05) and from 600 to 1500 K (in steps of 25 K), respectively.

For each combination of dopant content and temperature, the simulation started with the distribution of acceptor impurities over the B-sublattice and oxygen vacancies over the O-sublattice of the perovskite structure. Impurity ions were allocated according to the specified distribution types, and vacancies were distributed taking into account the prohibition rules for site occupancy within the adopted models. The ensemble of vacancies was then equilibrated using a sequence of random Monte Carlo steps with transition probabilities given by the Metropolis algorithm that accounts for inter-vacancy interactions within one of the adopted models (see § 3.1).

Our calculations indicate that 10510^{5} Monte Carlo steps were sufficient to achieve equilibrium for the difference in vacancy energies in neighboring sites up to ∼0.4\sim0.4 eV and the temperature range of 600-1500 K. After equilibration, the simulation continued for an additional 10410^{4} steps, with the concentration of vacancies, cVmc_{\mathrm{Vm}}, with different local surroundings recorded every 500 steps. To minimize the impact of random fluctuations, the resulting concentration values were obtained by averaging these measurements.

To ensure that the attained equilibrium state corresponds to the global minimum of the thermodynamic potential and to exclude possible dependence of the results on specific initial impurity configurations, the calculations were carried out repeatedly. Each simulation was performed about 10 times for every dopant content and temperature, each time starting with a new allocation of vacancies and impurities. The close agreement between repeatedly simulated cVm(x,T)c_{\mathrm{Vm}}(x,T) dependencies indicates a low probability of occurrence of the above effects during modeling.

Similar Monte Carlo simulation schemes are described in more detail in our previous works [20, 40].

Results and discussion

Distribution of oxygen vacancies

First, we discuss the impact of trapping energies, dopant content and temperature on the distribution of vacancies over oxygen sites with different number of neighboring impurities (B-O-B, B-O-R, and R-O-R). Figure 2 shows the concentrations of acceptor-bound vacancies, cVRc_{\mathrm{VR}} and cV2Rc_{\mathrm{V2R}}, calculated using Eqs. (11) and (12) as a function of the trapping energy ΔEVR/kT\Delta E_{\mathrm{VR}}/kT and the ratio ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}}. Hereafter, all vacancy concentrations are normalized to their total concentration cV=x/2c_{\mathrm{V}} = x/2. As seen in Figure 2a, cVRc_{\mathrm{VR}} attains its maximum value at high ΔEVR\Delta E_{\mathrm{VR}} energies and low ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} ratios. An increase in ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} results in a decrease in cVRc_{\mathrm{VR}}. Correspondingly, in the region of high values of ΔEVR\Delta E_{\mathrm{VR}} and ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}}, V2R\mathrm{V2R} vacancies become the dominant defects, as shown in Figure 2b.

Concentrations of oxygen vacancies in bound states near one and two impurities

Figure 2. Concentrations of oxygen vacancies in bound states near one (cVRc_{\mathrm{VR}}) and two (cV2Rc_{\mathrm{V2R}}) impurities as a function of the trapping energy ΔEVR/kT\Delta E_{\mathrm{VR}}/kT and the ratio ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} (x=0.2x = 0.2). Results are presented for the model with weak (Fermi-type) inter-vacancy correlations.

Figure 3 depicts the concentrations of vacancies in free and bound states as a function of dopant content xx for the typical trapping energy ΔEVR=0.4\Delta E_{\mathrm{VR}} = 0.4 eV and various ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} ratios. The Monte Carlo simulation results perfectly match the analytical calculations. With increasing xx, the fraction of free vacancies cVf/cVc_{\mathrm{Vf}}/c_{\mathrm{V}} decreases rapidly, so that bound vacancies become the dominant defects even at low dopant levels. The dependence of cVR/cVc_{\mathrm{VR}}/c_{\mathrm{V}} on xx is non-monotonic, with its maximum shifting towards lower xx as ΔEVR/kT\Delta E_{\mathrm{VR}}/kT and/or ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} increase (for the typical trapping energies used in Figure 3, the maximum occurs at x<10%x < 10\%). In contrast, cV2R/cVc_{\mathrm{V2R}}/c_{\mathrm{V}} increases monotonically with increasing xx. At high ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} ratios and doping levels, most vacancies reside at oxygen sites surrounded by two impurities.

Concentrations of acceptor-bound and free oxygen vacancies as a function of dopant content

Figure 3. Concentrations of acceptor-bound (cVRc_{\mathrm{VR}} and cV2Rc_{\mathrm{V2R}}) and free (cVfc_{\mathrm{Vf}}) oxygen vacancies, normalized to the total vacancy concentration cV=x/2c_{\mathrm{V}} = x/2, as a function of dopant content xx (T=1000 KT = 1000\ \mathrm{K}, ΔEVR=0.4 eV\Delta E_{\mathrm{VR}} = 0.4\ \mathrm{eV}). Results are presented for three ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} ratios: (a) 1, (b) 1.5 and (c) 2. Lines and symbols correspond to the results of analytical theory and Monte Carlo simulations, respectively, for cVRc_{\mathrm{VR}} (red), cV2Rc_{\mathrm{V2R}} (blue) and cVfc_{\mathrm{Vf}} (green).

Figure 4 shows the vacancy occupation probabilities ⟨nVR⟩\langle n_{\mathrm{VR}}\rangle and ⟨nV2R⟩\langle n_{\mathrm{V2R}}\rangle (given by Eq. (9)) as a function of dopant content. The results for free vacancies are not shown because of the very low ⟨nVf⟩\langle n_{\mathrm{Vf}}\rangle values. As expected, with decreasing temperature, the occupancy of R-O-R sites increases, and the occupancy of R-O-B sites decreases, providing ΔEV2R/ΔEVR>1\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} > 1. Figure 4 indicates that the occupation probability ⟨nVR⟩\langle n_{\mathrm{VR}}\rangle remains small over the considered range of xx values (⟨nVR⟩≪1\langle n_{\mathrm{VR}}\rangle\ll1), whereas the mean occupancy of sites surrounded by two impurities ⟨nV2R⟩\langle n_{\mathrm{V2R}}\rangle can be significant at low xx and high ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} ratios. Thus, in the latter range of parameters, Fermi-type correlations should substantially contribute to defect thermodynamics (see § 3.2 and § 3.5).

Occupation probabilities for oxygen vacancies bound near one and two impurities as a function of dopant content

Figure 4. Occupation probabilities (Eq. (9)) for oxygen vacancies bound near one (⟨nVR⟩\langle n_{\mathrm{VR}}\rangle) and two (⟨nV2R⟩\langle n_{\mathrm{V2R}}\rangle) impurities as a function of dopant content xx at different ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} ratios (T=1000T = 1000 K, ΔEVR=0.4\Delta E_{\mathrm{VR}} = 0.4 eV).

The unusually high values of ⟨nV2R⟩\langle n_{\mathrm{V2R}}\rangle at low xx are explained by the low concentration of R-O-R sites and their preferential occupancy at ΔEV2R/ΔEVR>1\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} > 1. The maximum on the dependence of ⟨nV2R⟩\langle n_{\mathrm{V2R}}\rangle on xx appears at ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} ratios exceeding a certain value (ΔEV2R/ΔEVR)∗(\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}})^{*}, which can be estimated using Boltzmann statistics as

kTΔEVRln⁡(2eΔEVRkT−1).\frac{kT}{\Delta E_{\mathrm{VR}}}\ln\left(2e^{\frac{\Delta E_{\mathrm{VR}}}{kT}}-1\right).

For ΔEVR=0.4÷0.6 eV\Delta E_{\mathrm{VR}} = 0.4 \div0.6\ \mathrm{eV} and T=600÷1200 KT = 600 \div1200\ \mathrm{K}, we obtain (ΔEV2R/ΔEVR)∗=1.1÷1.2(\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}})^{*} = 1.1 \div1.2. With increasing the ratio ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} above (ΔEV2R/ΔEVR)∗(\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}})^{*}, the occupancy ⟨nV2R⟩\langle n_{\mathrm{V2R}}\rangle increases and approaches unity at ΔEV2R/kT≳2ΔEVR/kT≫1\Delta E_{\mathrm{V2R}}/kT \gtrsim2\Delta E_{\mathrm{VR}}/kT \gg1, when nearly all vacancies are surrounded by two impurities. We can estimate the region of xx values, in which ⟨nV2R⟩\langle n_{\mathrm{V2R}}\rangle can be high, using the condition cV=x/2≳3p2Rc_{\mathrm{V}} = x/2 \gtrsim3p_{\mathrm{2R}}, yielding x≲0.17x \lesssim0.17. The concentration x0x_{0}, at which ⟨nV2R⟩\langle n_{\mathrm{V2R}}\rangle attains its maximum, shifts to higher xx with decreasing ΔEVR/kT\Delta E_{\mathrm{VR}}/kT and/or ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}}. At realistic energy parameters ΔEVR=0.4 eV\Delta E_{\mathrm{VR}} = 0.4\ \mathrm{eV} and ΔEV2R/ΔEVR=1.5\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} = 1.5, the x0x_{0} value falls within the range 0<x0<0.070 < x_{0} < 0.07 at temperatures between 600 and 1200 K.

Correlations

Correlations between oxygen vacancies can play a significant role in their distribution, particularly in heavily doped perovskites. We explore the effects of these correlations using analytical calculations and Monte-Carlo simulations for the proposed model (see § 2.1). The analytical approach accounts only for weak (Fermi-type) correlations, while the Monte Carlo modeling incorporates both weak and strong (inter-site) correlations.

Figure 5 illustrates the impact of correlations on vacancy distribution over oxygen sites with different impurity surroundings. The correlations lead to an increase in the concentration of VRV_{\mathrm{R}} vacancies, cVRc_{\mathrm{VR}}, and a decrease in the concentration of V2RV_{\mathrm{2R}} vacancies, cV2Rc_{\mathrm{V2R}}. This effect becomes more pronounced at higher values of the ratio ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}}. The reason for such an opposite trend lies in the significantly different occupation probabilities of VRV_{\mathrm{R}} and V2RV_{\mathrm{2R}} vacancies (see Figure 4). Since ⟨nV2R⟩\langle n_{\mathrm{V2R}}\rangle is relatively high, inter-vacancy correlations noticeably reduce the number of available oxygen sites, thereby decreasing cV2Rc_{\mathrm{V2R}}.

Concentrations of oxygen vacancies bound near acceptor impurities

Figure 5. Concentrations of oxygen vacancies bound near one (cVRc_{\mathrm{VR}}) and two (cV2Rc_{\mathrm{V2R}}) acceptor impurities as a function of (a) dopant content xx (T=1000T = 1000 K) and (b) temperature TT (x=0.2x = 0.2). Results are presented for ΔEVR=0.4\Delta E_{\mathrm{VR}} = 0.4 eV and ΔEV2R/ΔEVR=1.5\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} = 1.5. Symbols indicate Monte Carlo results for three models: no correlations (open circles), weak correlations (closed squares) and strong correlations (open triangles). Dashed and solid lines represent analytically calculated dependencies for the models without correlations and with weak inter-vacancy correlations, respectively.

The effect of inter-vacancy interactions is most pronounced at low temperatures, as shown in Figure 5b. When inter-site vacancy repulsion is neglected, almost all vacancies occupy the lowest energy states (R-O-R sites). Accounting for this repulsion leads to a decrease in cV2Rc_{\mathrm{V2R}} and an increase in cVRc_{\mathrm{VR}}. In this case, the fraction of VRV_{\mathrm{R}} vacancies is significant even at low temperatures. The effect of strong correlations diminishes with decreasing dopant concentration. The change in temperature mainly results in the redistribution of vacancies between B-O-R and R-O-R sites, as the fraction of free vacancies cVf/cVc_{\mathrm{Vf}}/c_{\mathrm{V}} is small under the considered conditions.

To analyze the effect of inter-vacancy correlations on local structure, we consider the short-range order parameter α\alpha, which characterizes deviations in the local distribution of host ions and/or defects from an uncorrelated arrangement. We define α\alpha following [41] and express it as

α=1−ZZ0,(18)\alpha= 1 - \frac{Z}{Z_0}, \tag*{(18)}

where ZZ and Z0Z_0 are the first-nearest-neighbor (1NN) coordination numbers with and without inter-vacancy correlations.

Let us consider the average coordination numbers of oxygen vacancies around acceptor impurities ZR−VZ_{\mathrm{R-V}} and host B cations ZB−VZ_{\mathrm{B-V}}. For the uniform dopant distribution, these coordination numbers are given by

ZR−V=6[x⟨nV2R⟩+(1−x)⟨nVR⟩],(19)Z_{\mathrm{R-V}} = 6\left[x\langle n_{\mathrm{V2R}}\rangle+ (1-x)\langle n_{\mathrm{VR}}\rangle\right], \tag*{(19)}
ZB−V=6[x⟨nVR⟩+(1−x)⟨nVf⟩].(20)Z_{\mathrm{B-V}} = 6\left[x\langle n_{\mathrm{VR}}\rangle+ (1-x)\langle n_{\mathrm{Vf}}\rangle\right]. \tag*{(20)}

Using the expressions for pmp_m (Eq. (7)) and the relationship between occupation probabilities ⟨nVm⟩\langle n_{\mathrm{Vm}}\rangle and concentrations cVmc_{\mathrm{Vm}} (Eq. (11)), from Eqs. (19) and (20) we obtain

ZR−V=cVR+2cV2Rx,(21)Z_{\mathrm{R-V}} = \frac{c_{\mathrm{VR}}+2c_{\mathrm{V2R}}}{x}, \tag*{(21)}
ZB−V=x−cVR−2cV2R1−x.(22)Z_{\mathrm{B-V}} = \frac{x-c_{\mathrm{VR}}-2c_{\mathrm{V2R}}}{1-x}. \tag*{(22)}

The calculated short-range order parameters αR−V\alpha_{\mathrm{R-V}} and αB−V\alpha_{\mathrm{B-V}} as a function of dopant content are given in Figure 6. The contribution of weak correlations exhibits a non-monotonic dependence on xx, the shape of which is governed by the behavior of the occupation probabilities varying non-monotonically with xx (see Figure 4).

Four-panel plot of short-range order parameters

Figure 6. Short-range order parameters for the distribution of vacancies around (a,b) impurities (αR−V\alpha_{\mathrm{R-V}}) and (c,d) host B cations (αB−V\alpha_{\mathrm{B-V}}) as a function of dopant content xx (T=1000T = 1000 K). Results are presented for ΔEVR=0.4\Delta E_{\mathrm{VR}} = 0.4 eV and two ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} ratios: (a,c) 1.5 and (b,d) 2. Lines and symbols indicate the results of analytical theory and Monte Carlo simulations for weak (red lines and squares) and strong (blue circles) correlations.

Specifically, high values of ⟨nV2R⟩\langle n_{\mathrm{V2R}}\rangle in a narrow doping range lead to large deviations of vacancy concentrations from their values in the absence of correlations, which in turn results in large deviations of the short-range order parameters from zero. At low xx, αR−V\alpha_{\mathrm{R-V}} and αB−V\alpha_{\mathrm{B-V}} can be largely determined by weak correlations, while at high xx, strong correlations play a dominant role.

Coordination numbers

Studying coordination numbers is useful for understanding the effect of inter-defect interactions on the local structure of oxides. Figure 7 depicts the average coordination numbers ZB−OZ_{\mathrm{B-O}} and ZR−OZ_{\mathrm{R-O}} of host B cations and impurities R, respectively, surrounded by 1NN oxygen ions as a function of xx. These coordination numbers are calculated as ZB−O=6−ZB−VZ_{\mathrm{B-O}} = 6 - Z_{\mathrm{B-V}} and ZR−O=6−ZR−VZ_{\mathrm{R-O}} = 6 - Z_{\mathrm{R-V}} using Eqs. (19) and (20). At uniform defect distribution, when inter-defect interactions are absent, ZB−OZ_{\mathrm{B-O}} and ZR−OZ_{\mathrm{R-O}} reduce to 6−x6 - x (black dashed line in Figure 7a).

Coordination numbers and short-range order parameters for B cations and impurities R

Figure 7. (a) Coordination numbers of B cations (ZB−OZ_{\mathrm{B-O}}) and impurities R (ZR−OZ_{\mathrm{R-O}}) surrounded by oxygen ions and (b) the corresponding short-range order parameters αB−O∗\alpha_{\mathrm{B-O}}^{*} and αR−O∗\alpha_{\mathrm{R-O}}^{*} as a function of dopant content xx (T=1000T=1000 K, ΔEVR=0.4\Delta E_{\mathrm{VR}}=0.4 eV). Results are provided for different ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} ratios. Colored lines represent analytically calculated dependencies for the model with weak correlations. Closed and open symbols correspond to Monte Carlo results for the model with weak and strong correlations, respectively. Dashed black lines correspond to the uniform distribution of defects. The fragment of the perovskite structure illustrating the nearest-neighbor coordination of B/R cations is given in (a), where oxygen and B/R ions are indicated by red and violet-green spheres, respectively.

With increasing xx, and accordingly the concentration of vacancies in the oxide, both ZB−OZ_{\mathrm{B-O}} and ZR−OZ_{\mathrm{R-O}} decrease. However, at high values of ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}}, most vacancies are surrounded by two impurities, and ZB−OZ_{\mathrm{B-O}} remains close to 6 (Figure 7a). In the latter case, an impurity on average traps one vacancy and ZR−OZ_{\mathrm{R-O}} approaches 5.

To assess the effect of vacancy-impurity interactions, we consider the short-range order parameter α∗\alpha^{*}, defined analogously to Eq. (18), but with Z0Z_{0} being replaced by Z0∗Z_{0}^{*} – the coordinating number in the absence of inter-defect interactions. This parameter characterizes deviations from the uniform defect distribution. As seen in Figure 7, the average coordination numbers ZB−OZ_{\mathrm{B-O}} and ZR−OZ_{\mathrm{R-O}} lie, respectively, above and below their values at uniform distribution, leading to αB−O∗<0\alpha_{\mathrm{B-O}}^{*}<0 and αR−O∗>0\alpha_{\mathrm{R-O}}^{*}>0. Note that αB−O∗\alpha_{\mathrm{B-O}}^{*} decreases monotonically with increasing xx, while αR−O∗(x)\alpha_{\mathrm{R-O}}^{*}(x) exhibits a maximum at moderate xx. As ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} increases, the defect distribution becomes increasingly non-uniform (Figure 7b).

The Monte Carlo simulation results in Figure 7 are shown for the cases of weak and strong correlations. Although the coordination numbers remain close in both models, accounting for inter-site vacancy interactions slightly increases ZR−OZ_{\mathrm{R-O}} due to the reduced number of vacancies surrounding each impurity.

We now proceed to analyze the behavior of the average coordination number of acceptor impurities R around oxygen vacancies V, given by

ZV−R=2ZR−V=cVR+2cV2RcV(23)Z_{\mathrm{V-R}} = 2Z_{\mathrm{R-V}} = \frac{c_{\mathrm{VR}}+2c_{\mathrm{V2R}}}{c_{\mathrm{V}}} \tag*{(23)}

Figure 8a shows that ZV−RZ_{\mathrm{V-R}} increases with increasing xx and is higher than its value at uniform distribution of defects (2x2x). Correspondingly, the parameter αV−R∗\alpha^{*}_{\mathrm{V-R}} is negative (Figure 8b), indicating the attractive interaction between vacancies and impurities. The effect of strong correlations results in the slightly reduced ZV−RZ_{\mathrm{V-R}} values (Figure 8a). As shown in Figure 8b, the largest deviation from the uniform distribution occurs at low xx. With increasing xx, αV−R∗\alpha^{*}_{\mathrm{V-R}} increases and gradually approaches zero at high doping levels. Increasing the ratio ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} leads to an increase in the concentration of R-V-R complexes and, as a result, ZV−RZ_{\mathrm{V-R}} approaches 2.

Coordination number and short-range order parameter as a function of dopant content

Figure 8. (a) Coordination number ZV−RZ_{\mathrm{V-R}} of oxygen vacancies V surrounded by acceptor impurities R and (b) the corresponding short-range order parameter αV−R∗\alpha^{*}_{\mathrm{V-R}} as a function of dopant content xx (T=1000T = 1000 K, ΔEVR=0.4\Delta E_{\mathrm{VR}} = 0.4 eV). Results are provided for different ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} ratios. Colored lines indicate analytical theory results for the model with weak correlations. Closed and open symbols represent Monte Carlo simulation results for the model with weak and strong correlations, respectively. The dashed black line corresponds to the uniform distribution of vacancies over oxygen sites. The fragment of the perovskite structure illustrating the nearest-neighbor coordination of an oxygen vacancy is given in (a). Oxygen ions, vacancies and B/R cations are indicated by red, white and violet-green spheres, respectively.

A comparison between the short-range order parameters describing the trapping (αR−V∗\alpha^{*}_{\mathrm{R-V}}, αB−V∗\alpha^{*}_{\mathrm{B-V}}, αB−O∗\alpha^{*}_{\mathrm{B-O}}) and correlation (αR−V\alpha_{\mathrm{R-V}}, αB−V\alpha_{\mathrm{B-V}}, αB−O\alpha_{\mathrm{B-O}}) effects shows that the former significantly exceeds the latter for most local coordination environments, indicating the predominant role of vacancy-impurity interactions.

Non-uniform distribution of impurities

Our considerations thus far have focused on the problem at uniform dopant distribution in oxides. The actual distribution, however, can deviate from uniform due to a sample preparation method. In this section, we discuss the implications of a non-uniform distribution of acceptor impurities for the defect and local structure of the studied AB1−xRxO3−δ\mathrm{AB}_{1-x}\mathrm{R}_{x}\mathrm{O}_{3-\delta} perovskites. As an example, we use a frozen high-temperature dopant arrangement corresponding to the equilibrium structure obtained via combined DFT and Monte Carlo simulations [30]. Figure 9a presents the data extracted from Ref. [30] on the average Y-Y coordination at 1NN sites in Y-doped BaZrO3_3 at a typical sintering temperature of 1900 K. It is seen that the probability of finding a Y ion at a B-site nearest to another Y ion is noticeably lower than expected for the uniform dopant distribution.

Conditional probabilities and probability ratios for Y-doped BaZrO3

Figure 9. (a) Conditional probabilities p(R∣R)p(\mathrm{R}|\mathrm{R}) and p(R∣B)p(\mathrm{R}|\mathrm{B}) extracted from the Y-Y coordination data at 1NN sites in Y-doped BaZrO3_3 at ∼\sim 1900 K obtained using DFT and Monte Carlo simulation [30]. The dashed black line corresponds to the probability for the uniform impurity distribution. (b) The ratio of the probabilities for a non-uniform dopant distribution (pmp_{\mathrm{m}}) to those at uniform doping (pmup_{\mathrm{m}}^{\mathrm{u}}), plotted as a function of dopant content xx. Symbols indicate these ratios calculated from the data given in (a) using Eqs. (24).

To incorporate this non-uniform dopant distribution into our theory, we extract the necessary probabilities pRp_{\mathrm{R}}, p2Rp_{2\mathrm{R}} and pfp_{\mathrm{f}} from the Y-Y coordination data [30] (see Figure 9a). This is achieved by determining from the data [30] the conditional probabilities p(j∣i)p(j|i), which describe the likelihood of finding a cation of type jj at a B-site adjacent to a cation of type ii (i,j=B,Ri,j=\mathrm{B},\mathrm{R}). These probabilities are related to pmp_{\mathrm{m}} as follows:

pR=x(1−p(R∣R))+(1−x)p(R∣B),(24)p_{\mathrm{R}}=x\left(1-p(\mathrm{R}|\mathrm{R})\right)+(1-x)p(\mathrm{R}|\mathrm{B}), \tag*{(24)}
p2R=xp(R∣R),(25)p_{2\mathrm{R}}=xp(\mathrm{R}|\mathrm{R}), \tag*{(25)}
pf=(1−x)(1−p(R∣B)).(26)p_{\mathrm{f}}=(1-x)\left(1-p(\mathrm{R}|\mathrm{B})\right). \tag*{(26)}

For the uniform dopant distribution, we have p(R∣R)=p(R∣B)=xp(\mathrm{R}|\mathrm{R})=p(\mathrm{R}|\mathrm{B})=x.

The R-R coordination number at a 1NN site is given by 6p(R∣R)6p(\mathrm{R}|\mathrm{R}). Thus, using the data from Ref. [30], we determine p(R∣R)p(\mathrm{R}|\mathrm{R}) and p(R∣B)p(\mathrm{R}|\mathrm{B}), allowing us to calculate the desired probabilities pmp_{\mathrm{m}} for each dopant content. Figure 9b demonstrates that pRp_{\mathrm{R}} and pfp_{\mathrm{f}}, calculated using Eqs. (24a) and (24c), are close to their values (pRup_{\mathrm{R}}^{\mathrm{u}} and pfup_{\mathrm{f}}^{\mathrm{u}}) at uniform doping, whereas p2Rp_{2\mathrm{R}} (25) is significantly lower than p2Rup_{2\mathrm{R}}^{\mathrm{u}}.

With the obtained probabilities pRp_{\mathrm{R}}, p2Rp_{2\mathrm{R}} and pfp_{\mathrm{f}}, we can compute the defect concentrations using Eq. (11). Fig. 10a illustrates that the employed dopant distribution results in a decrease in the concentration of R-V-R complexes (cV2Rc_{\mathrm{V2R}}) and an increase in the concentration of R-V-B complexes (cVRc_{\mathrm{VR}}) compared to the uniform distribution.

The coordination numbers ZR−VZ_{\mathrm{R-V}} (19) and ZB−VZ_{\mathrm{B-V}} (20) for the considered non-uniform dopant distribution can be found using the determined probabilities p(j∣i)p(j|i) and mean occupancy numbers ⟨nm⟩\langle n_m\rangle:

ZR−V=6[p(R∣R)⟨nV2R⟩+p(B∣R)⟨nVR⟩],(27)Z_{\mathrm{R-V}} = 6[p(\mathrm{R}|\mathrm{R})\langle n_{\mathrm{V2R}}\rangle+ p(\mathrm{B}|\mathrm{R})\langle n_{\mathrm{VR}}\rangle], \tag*{(27)}
ZB−V=6[p(R∣B)⟨nVR⟩+p(B∣B)⟨nVf⟩].(28)Z_{\mathrm{B-V}} = 6[p(\mathrm{R}|\mathrm{B})\langle n_{\mathrm{VR}}\rangle+ p(\mathrm{B}|\mathrm{B})\langle n_{\mathrm{Vf}}\rangle]. \tag*{(28)}

Using Eqs. (24), these expressions reduce to the previously derived formulas (21) and (22).

Fig. 10b shows the ratio of the coordination numbers at non-uniform impurity distribution to their values at uniform doping. The coordination numbers ZR−OZ_{\mathrm{R-O}} and ZB−OZ_{\mathrm{B-O}} computed using Eqs. (27) and (28) are virtually the same as those for the uniform dopant distribution (the results in Fig. 10b are shown only for ZR−OZ_{\mathrm{R-O}}). At the same time, the employed non-uniform dopant distribution slightly reduces the average number of Y ions surrounding V, especially at high ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} ratios. This reduction arises from a decreased probability of the formation of R-R complexes (Fig. 10b).

To obtain a more general assessment of the influence of non-uniform dopant distribution, we calculated the coordination numbers at realistic trapping energies in a wide range of temperatures and dopant concentrations, varying the probability p(R∣R)p(\mathrm{R}|\mathrm{R}) by ±50%\pm50\% relative to its value at uniform distribution. The results indicate that the maximum deviation from the values at uniform doping does not exceed ∼2%\sim2\%, ∼5%\sim5\% and ∼25%\sim25\% for ZB−OZ_{\mathrm{B-O}}, ZR−OZ_{\mathrm{R-O}} and ZV−RZ_{\mathrm{V-R}}, respectively.

Note that the obtained results should be considered as an illustration of the possible impact of non-uniform dopant distribution rather than as general predictions for the studied perovskites. In real oxides, dopant distribution depends on the synthesis method and sample preparation history.

Coordination and defect concentration ratios for non-uniform dopant distributions

Figure 10.

Image not extracted

Figure 10 (PDF p. 16). Ratios of the (a) oxygen vacancy concentrations (cVRc_{\mathrm{VR}} and cV2Rc_{\mathrm{V2R}}) and (b) coordination numbers (ZR−OZ_{\mathrm{R-O}} and ZV−RZ_{\mathrm{V-R}}) at non-uniform impurity distribution to their values at uniform doping (cuc^{\mathrm{u}} and ZuZ^{\mathrm{u}}) as a function of dopant content xx (T=1000T = 1000 K, ΔEVR=0.4\Delta E_{\mathrm{VR}} = 0.4 eV). Open symbols and crosses indicate the results of our analytical theory and Monte Carlo modeling, respectively. The dependencies are presented for the non-uniform dopant distribution corresponding to a frozen high-temperature (∼1900\sim1900 K) cation allocation obtained via DFT and Monte Carlo simulations [30].

Oxidation

In this section, we examine the influence of inter-defect interactions on the oxidation of wide-band-gap AB1−xRxO3−δ\mathrm{AB}_{1-x}\mathrm{R}_{x}\mathrm{O}_{3-\delta} perovskites without mixed-valence cations. In such oxides, the states of electron holes are derived mostly from O2p\mathrm{O}2p orbitals [25, 26, 27, 42]. Typically, small hole polarons in wide-band-gap oxides are energetically more favorable than free valence-band holes [25, 42, 43, 44, 45]. The holes are mainly localized on oxygen ions adjacent to impurities, forming acceptor-bound small hole polarons [23, 25, 42, 45, 46].

As we showed earlier [23, 46], if the acceptor-bound hole states are located deep enough in the band gap, the oxidation reaction can be exothermic, and the concentration of bound holes can be considerable, and even comparable to that of ionic defects. In most cases, however, one can expect endothermic oxidation of the studied oxides and, accordingly, the hole concentration chc_{\mathrm{h}} much lower than cVc_{\mathrm{V}}. Assuming the condition ch≪cVc_{\mathrm{h}} \ll c_{\mathrm{V}}, we can neglect the effect of oxidation on the oxygen content and hence cV=x/2c_{\mathrm{V}} = x/2.

Despite the fact that ch≪cVc_{\mathrm{h}} \ll c_{\mathrm{V}}, the hole conductivity σh\sigma_{\mathrm{h}} can be comparable to or even exceed the oxygen-ion conductivity in perovskites under consideration due to high hole mobility (see, e.g., [18, 23, 46, 47]). As we previously demonstrated, the hole transport data in these oxides can be described by both band-like transfer and small polaron hopping models, owing to the small activation energy of hopping [23, 46]. In this study, for definiteness, we consider holes as small polarons (consideration of band-like holes instead of small polarons does not affect the results on oxidation).

The oxidation reaction and the corresponding oxide-gas equilibrium condition can be written as

0.5O2(g)⇌O2−+2h+,(29)0.5\mathrm{O}_{2(\mathrm{g})} \rightleftharpoons\mathrm{O}^{2-} + 2\mathrm{h}^{+}, \tag*{(29)}
0.5μO2=2μh−μV,(30)0.5\mu_{\mathrm{O}_2} = 2\mu_{\mathrm{h}} - \mu_{\mathrm{V}}, \tag*{(30)}

where μO2\mu_{\mathrm{O}_2} is the chemical potential of oxygen molecules in the gas phase, defined as μO2=μO20+kTln⁡pO2\mu_{\mathrm{O}_2} = \mu_{\mathrm{O}_2}^{0} + kT\ln p_{\mathrm{O}_2} (μO20\mu_{\mathrm{O}_2}^{0} is the standard chemical potential and pO2p_{\mathrm{O}_2} is the oxygen partial pressure); μh\mu_{\mathrm{h}} and μV\mu_{\mathrm{V}} are the chemical potentials of holes and vacancies, respectively.

Given the low concentration of holes, their chemical potential can be written as

μh=μh0+kTln⁡ch,(31)\mu_{\mathrm{h}} = \mu_{\mathrm{h}}^{0} + kT\ln c_{\mathrm{h}}, \tag*{(31)}

where chc_{\mathrm{h}} is the concentration of hole carriers and μh0\mu_{\mathrm{h}}^{0} is their standard chemical potential.

Using Eqs. (13), (30) and (31), we can write the mass action law for the oxidation reaction (27) as follows:

Kox=Kox0Koxint=3ch2cVpO2(32)K_{\mathrm{ox}} = K_{\mathrm{ox}}^{0} K_{\mathrm{ox}}^{\mathrm{int}} = \frac{3c_{\mathrm{h}}^{2}}{c_{\mathrm{V}}\sqrt{p_{\mathrm{O}_{2}}}} \tag*{(32)}

where we introduce the effective oxidation constant KoxK_{\mathrm{ox}}, which includes the equilibrium constant Kox0K_{\mathrm{ox}}^{0} for an ideal defect solution and the non-ideality factor KoxintK_{\mathrm{ox}}^{\mathrm{int}}. The components Kox0K_{\mathrm{ox}}^{0} and KoxintK_{\mathrm{ox}}^{\mathrm{int}} are determined as

Kox0=exp⁡(EVf+2μH0+0.5μO20kT)(33)K_{\mathrm{ox}}^{0} = \exp\left(\frac{E_{\mathrm{Vf}}+2\mu_{\mathrm{H}}^{0}+0.5\mu_{\mathrm{O}_{2}}^{0}}{kT}\right) \tag*{(33)}

and

Koxint=γV,(34)K_{\mathrm{ox}}^{\mathrm{int}} = \gamma_{\mathrm{V}}, \tag*{(34)}

where γV\gamma_{\mathrm{V}} is the activity coefficient of vacancies, defined by Eq. (13) in § 2.2. In the case of Boltzmann statistics, Koxint=γVK_{\mathrm{ox}}^{\mathrm{int}}=\gamma_{\mathrm{V}} can be explicitly expressed by Eq. (17) in terms of the trapping energies (ΔEVR\Delta E_{\mathrm{VR}} and ΔEV2R\Delta E_{\mathrm{V2R}}) and probabilities pmp_m. The attractive interaction between vacancies and impurities (ΔEVR,ΔEV2R>0\Delta E_{\mathrm{VR}},\Delta E_{\mathrm{V2R}}>0) yields γV<1\gamma_{\mathrm{V}}<1 (see Eq. (17)), which means that the trapping effect reduces the oxidation constant KoxK_{\mathrm{ox}}, and, correspondingly, the hole concentration.

The non-ideality factor KoxintK_{\mathrm{ox}}^{\mathrm{int}} as a function of dopant content is plotted in Figure 11a. To assess the impact of Fermi-type correlations, we also plotted the ratio of γV\gamma_{\mathrm{V}} to its value calculated using Boltzmann statistics (Eq. (17)). As shown in Figure 11, these correlations have a small effect on oxidation, with the exception of a narrow range of parameters characterized by low temperatures, large ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} ratios and moderate dopant concentrations (x<0.2x<0.2). This range of parameters corresponds to the high occupancy ⟨nV2R⟩\langle n_{\mathrm{V2R}}\rangle of R-O-R sites (see § 3.1). In contrast, vacancy trapping strongly influences oxidation, with the magnitude of this effect varying significantly with energy parameters, dopant content and temperature. In particular, the contribution of trapping to KoxK_{\mathrm{ox}} increases with increasing xx and decreasing TT, as illustrated in Figure 11a. Larger ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} ratios enhance both trapping and correlation effects.

Figure 11. Thermodynamic parameters of the oxidation reaction related to inter-defect interactions. (a) Component of the equilibrium constant Koxint=γVK_{\mathrm{ox}}^{\mathrm{int}}=\gamma_{\mathrm{V}} and (b) oxidation enthalpy ΔHoxint\Delta H_{\mathrm{ox}}^{\mathrm{int}} as a function of dopant content xx (ΔEVR=0.4\Delta E_{\mathrm{VR}}=0.4 eV, ΔEV2R/ΔEVR=1.5\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}}=1.5 and 2). The ratio of the activity coefficients γV/γVno corr\gamma_{\mathrm{V}}/\gamma_{\mathrm{V}}^{\mathrm{no\ corr}}, shown in (a), illustrates the effect of Fermi-type correlations on the equilibrium constant (γV\gamma_{\mathrm{V}} and γVno corr\gamma_{\mathrm{V}}^{\mathrm{no\ corr}} are calculated with and without correlations). Symbols represent the parameters calculated with the non-uniform dopant distribution, corresponding

To find the contribution of inter-defect interactions to the oxidation enthalpy, we use the standard relation ΔHoxint=kT2dln⁡Koxint/dT\Delta H_{\mathrm{ox}}^{\mathrm{int}}=kT^{2}d\ln K_{\mathrm{ox}}^{\mathrm{int}}/dT. The dependence of ΔHoxint\Delta H_{\mathrm{ox}}^{\mathrm{int}} on xx is shown in Figure 11b. The positive sign of ΔHoxint\Delta H_{\mathrm{ox}}^{\mathrm{int}} is explained by the fact that the trapping effect reduces the formation energy of oxygen vacancies (ΔEVR,ΔEV2R>0\Delta E_{\mathrm{VR}},\Delta E_{\mathrm{V2R}}>0), rendering oxidation less favorable. When the majority of vacancies occupy R-O-B sites, γV∼exp⁡(−ΔEVR/kT)\gamma_{\mathrm{V}}\sim\exp(-\Delta E_{\mathrm{VR}}/kT) and therefore ΔHoxint≈ΔEVR\Delta H_{\mathrm{ox}}^{\mathrm{int}}\approx\Delta E_{\mathrm{VR}}. Similarly, when the majority of vacancies are surrounded by two impurity ions, γV∼exp⁡(−ΔEV2R/kT)\gamma_{\mathrm{V}}\sim\exp(-\Delta E_{\mathrm{V2R}}/kT) and ΔHoxint\Delta H_{\mathrm{ox}}^{\mathrm{int}} approaches ΔEV2R\Delta E_{\mathrm{V2R}} at high xx. These scenarios, illustrated in Figure 11b, occur in the regions of low (x≈0.1x\approx0.1) and high (x>0.2x>0.2) dopant concentrations, respectively, at reasonably large ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} ratios and low temperatures. At moderate ratios (ΔEV2R/ΔEVR∼1.5\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}}\sim1.5) and/or higher temperatures, the first plateau (ΔHoxint≈ΔEVR\Delta H_{\mathrm{ox}}^{\mathrm{int}}\approx\Delta E_{\mathrm{VR}}) on the dependence ΔHoxint(x)\Delta H_{\mathrm{ox}}^{\mathrm{int}}(x) disappears.

For the region of parameters, where the effect of Fermi-type correlations is negligible (see above), we can obtain a good estimate of ΔHoxint\Delta H_{\mathrm{ox}}^{\mathrm{int}} using Eq. (17):

ΔHoxint=ΔEVRpReΔEVRkT+ΔEV2Rp2ReΔEV2RkTpf+pReΔEVRkT+p2ReΔEV2RkT.(35)\Delta H_{\mathrm{ox}}^{\mathrm{int}}=\frac{\Delta E_{\mathrm{VR}}p_{\mathrm{R}}e^{\frac{\Delta E_{\mathrm{VR}}}{kT}}+\Delta E_{\mathrm{V2R}}p_{2\mathrm{R}}e^{\frac{\Delta E_{\mathrm{V2R}}}{kT}}}{p_{\mathrm{f}}+p_{\mathrm{R}}e^{\frac{\Delta E_{\mathrm{VR}}}{kT}}+p_{2\mathrm{R}}e^{\frac{\Delta E_{\mathrm{V2R}}}{kT}}}. \tag*{(35)}

If ΔEVR=ΔEV2R=ΔEV\Delta E_{\mathrm{VR}}=\Delta E_{\mathrm{V2R}}=\Delta E_{\mathrm{V}}, Eq. (35) simplifies to

ΔHoxint=ΔEV1+pfpR+p2Re−ΔEVkT.(36)\Delta H_{\mathrm{ox}}^{\mathrm{int}}=\frac{\Delta E_{\mathrm{V}}}{1+\frac{p_{\mathrm{f}}}{p_{\mathrm{R}}+p_{2\mathrm{R}}}e^{-\frac{\Delta E_{\mathrm{V}}}{kT}}}. \tag*{(36)}

which was previously obtained in Ref. [23] for low dopant concentrations. In this case, the oxidation enthalpy has only one limit, ΔHoxint≈ΔEV\Delta H_{\mathrm{ox}}^{\mathrm{int}}\approx\Delta E_{\mathrm{V}}, attained at high xx and ΔEV/kT\Delta E_{\mathrm{V}}/kT.

Figure 11 also shows the equilibrium constant and oxidation enthalpy calculated for the frozen non-uniform dopant distribution, corresponding to the high-temperature defect configuration obtained in Ref. [30] (see § 3.4). Although the results for the uniform and non-uniform distributions slightly differ, the overall trends in the behavior of the thermodynamic parameters of the oxidation reaction are similar.

The considered inter-defect interactions significantly affect the position of the Fermi level (defined as εF=−μh\varepsilon_{\mathrm{F}}=-\mu_{\mathrm{h}}) and hole concentration. In accordance with Eqs. (13) and (30), the shift in the Fermi level ΔεFint\Delta\varepsilon_{\mathrm{F}}^{\mathrm{int}} is given by

ΔεFint=−0.5ΔμV=−0.5kTln⁡γV.(37)\Delta\varepsilon_{\mathrm{F}}^{\mathrm{int}}=-0.5\Delta\mu_{\mathrm{V}}=-0.5kT\ln\gamma_{\mathrm{V}}. \tag*{(37)}

When correlation effects are small, ΔεFint\Delta\varepsilon_{\mathrm{F}}^{\mathrm{int}} can be approximated by

ΔεFint=0.5kTln⁡(pf+pReΔEVRkT+p2ReΔEV2RkT).(38)\Delta\varepsilon_{\mathrm{F}}^{\mathrm{int}}=0.5kT\ln\left(p_{\mathrm{f}}+p_{\mathrm{R}}e^{\frac{\Delta E_{\mathrm{VR}}}{kT}}+p_{2\mathrm{R}}e^{\frac{\Delta E_{\mathrm{V2R}}}{kT}}\right). \tag*{(38)}

Earlier, we derived this equation for the special case of ΔEVR=ΔEV2R\Delta E_{\mathrm{VR}}=\Delta E_{\mathrm{V2R}} and low xx [18].

Furthermore, the hole concentration can be expressed through ΔεFint\Delta\varepsilon_{\mathrm{F}}^{\mathrm{int}} as follows

ch=ch0exp⁡(−ΔεFintkT)=ch0γV.(39)c_{\mathrm{h}}=c_{\mathrm{h}}^{0}\exp\left(-\frac{\Delta\varepsilon_{\mathrm{F}}^{\mathrm{int}}}{kT}\right)=c_{\mathrm{h}}^{0}\sqrt{\gamma_{\mathrm{V}}}. \tag*{(39)}

where ch0c_{\mathrm{h}}^{0} is the hole concentration in the absence of inter-defect interactions.

Eq. (38) indicates that trapping shifts the Fermi level to higher energies in the band gap (as predicted in Ref. [18]), thereby reducing the hole concentration (see Eq. (39)). The effect of trapping on εF\varepsilon_{\mathrm{F}} and chc_{\mathrm{h}} diminishes with increasing temperature. For example, at x=0.2x=0.2 and realistic energy parameters (ΔEVR=0.4\Delta E_{\mathrm{VR}}=0.4 eV, ΔEV2R/ΔEVR=1.5\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}}=1.5), ΔεFint\Delta\varepsilon_{\mathrm{F}}^{\mathrm{int}} decreases from 0.20 to 0.16 eV and ch/ch0c_{\mathrm{h}}/c_{\mathrm{h}}^{0} increases from 10−1.710^{-1.7} to 10−0.710^{-0.7} as temperature rises from 600 to 1200 K.

It should be noted that the effect of inter-defect interactions and impurity disorder on oxidation is essential for understanding the behavior of hole conduction. In particular, the oxidation enthalpy ΔHox\Delta H_{\mathrm{ox}} provides a direct contribution to the effective activation energy of hole conductivity σh\sigma_{\mathrm{h}}. This contribution, ΔEaoh\Delta E_{\mathrm{a}}^{\mathrm{oh}}, resulting from the ch(T)c_{\mathrm{h}}(T) dependence, is

ΔEaoh=ΔHox0+ΔHoxint2.(40)\Delta E_{\mathrm{a}}^{\mathrm{oh}}=\frac{\Delta H_{\mathrm{ox}}^{0}+\Delta H_{\mathrm{ox}}^{\mathrm{int}}}{2}. \tag*{(40)}

where ΔHox0\Delta H_{\mathrm{ox}}^{0} is the term associated with the equilibrium constant Kox0K_{\mathrm{ox}}^{0}, independent of inter-defect interactions.

As discussed above, at realistic trapping energies, ΔHoxint\Delta H_{\mathrm{ox}}^{\mathrm{int}} can be significant and strongly depend on dopant content (see Figure 11b), resulting in the corresponding behavior of hole conductivity and its activation energy.

Figure 12 shows that the dependence of the hole concentration on xx is strongly affected by the trapping energy ratio ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}}. For ΔEV2R/ΔEVR≈1\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} \approx1, chc_{\mathrm{h}} increases monotonically with increasing xx; for larger ratios, a maximum appears on the dependence ch(x)c_{\mathrm{h}}(x) at low xx. Note that the monotonic behavior of ch(x)c_{\mathrm{h}}(x) at not too high ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} ratios correlates with the observed behavior of the hole conductivity σh(x)\sigma_{\mathrm{h}}(x) in Y- and In-doped BaSnO3_3 [35, 36]. A detailed analysis of σh\sigma_{\mathrm{h}} in these oxides requires elucidating the mechanisms of hole mobility and the effects of impurity disorder on σh\sigma_{\mathrm{h}}, which is beyond the scope of this work. However, at not too large energy barriers for hole polaron hopping, the dependence σh(x)\sigma_{\mathrm{h}}(x) is largely determined by the ch(x)c_{\mathrm{h}}(x) behavior. This can be easily understood by analogy with the behavior of proton hopping conductivity in perovskites under consideration [18, 19]. Using the effective media approximation, we demonstrated that at low hopping barriers, the carrier’ mobility weakly depends on xx outside the region of low xx [18, 19]. Since the estimated barriers for hole polaron hopping in acceptor-doped BaSnO3_3 are comparable to kTkT (∼0.1\sim0.1 eV) [45], we can expect that the dependences σh(x)\sigma_{\mathrm{h}}(x) and ch(x)c_{\mathrm{h}}(x) behave similarly at relevant temperatures.

Dependence of the normalized hole concentration on dopant content and trapping-energy ratio

Figure 12. Dependence of the normalized hole concentration ch/ch0(x0)c_{\mathrm{h}}/c_{\mathrm{h}}^{0}(x_{0}) on dopant content xx and ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} ratio (T=1000T=1000 K, ΔEVR=0.4\Delta E_{\mathrm{VR}}=0.4 eV). ch0(x0)c_{\mathrm{h}}^{0}(x_{0}) is the hole concentration at x0=0.05x_{0}=0.05 in the absence of inter-defect interactions.

Concluding this section, we would like to note that the obtained results on oxidation, predicting the position of εF\varepsilon_{\mathrm{F}} and chc_{\mathrm{h}} behavior, can be important for understanding the performance of electrochemical devices based on the studied oxides (see, e.g., Refs. [1, 18, 48]).

to a frozen high-temperature configuration (see § 3.4). Dashed horizontal lines in (a) and (b) correspond to Koxint=γV=1K_{\mathrm{ox}}^{\mathrm{int}}=\gamma_{\mathrm{V}}=1 and the values of the trapping energies (ΔEVR\Delta E_{\mathrm{VR}} and ΔEV2R\Delta E_{\mathrm{V2R}}), respectively.

Comparison with experiment and other simulations

Our results on the distribution of oxygen vacancies can be compared with the findings of the NMR studies of local structure in acceptor-doped perovskites. Oikawa et al. [28] studied the 45Sc{}^{45}\mathrm{Sc} NMR spectra of vacuum-dried BaZr1−xScxO3−δ\mathrm{BaZr}_{1-x}\mathrm{Sc}_{x}\mathrm{O}_{3-\delta} (x=0.01x=0.01–0.150.15) and determined the integral intensities of the peaks assigned to the five- and six-oxygen-coordinated Sc. Given the relatively low dopant content in the studied samples [28], we will use for comparison our results for the case of weak correlations. For the uniform distribution of impurities, the fractions of the RO6\mathrm{RO}_{6} and RO5\mathrm{RO}_{5} groups are given, respectively, by

[x(1−⟨nV2R⟩)+(1−x)(1−⟨nVR⟩)]6(41)\left[x\left(1-\langle n_{\mathrm{V2R}}\rangle\right)+(1-x)\left(1-\langle n_{\mathrm{VR}}\rangle\right)\right]^{6} \tag*{(41)}
6[x⟨nV2R⟩+(1−x)⟨nVR⟩][x(1−⟨nV2R⟩)+(1−x)(1−⟨nVR⟩)]5(42)6\left[x\langle n_{\mathrm{V2R}}\rangle+(1-x)\langle n_{\mathrm{VR}}\rangle\right]\left[x\left(1-\langle n_{\mathrm{V2R}}\rangle\right)+(1-x)\left(1-\langle n_{\mathrm{VR}}\rangle\right)\right]^{5} \tag*{(42)}

Fitting the experimental data [28] with Eqs. (41) and (42) yields ΔEVR/kT≈3.0\Delta E_{\mathrm{VR}}/kT\approx3.0 for ΔEV2R/ΔEVR≈1.5\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}}\approx1.5 and ΔEVR/kT≈2.6\Delta E_{\mathrm{VR}}/kT\approx2.6 for ΔEV2R/ΔEVR≈2.0\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}}\approx2.0. If we assume that the defect configuration is frozen at 1473 K (at which the samples were vacuum dried for 22 h [28]), then ΔEVR≈0.3\Delta E_{\mathrm{VR}}\approx0.3–0.40.4 eV. It should be noted that variations in the parameter ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}}, in contrast to ΔEVR/kT\Delta E_{\mathrm{VR}}/kT, have little effect on the accuracy of the fitting. The determined trapping energy ΔEVR\Delta E_{\mathrm{VR}} is close to the values obtained by ab initio calculations for Sc-doped BaZrO3\mathrm{BaZrO}_{3} (0.37 eV [12] and 0.29 eV [14]). For Y-doped BaZrO3\mathrm{BaZrO}_{3}, Draber et al. [24] obtained the ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} ratio of ∼1.5\sim1.5.

Figure 13a shows very good agreement between our theoretical results and the experimental data [28] for both values of the ratio ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} (1.5 and 2.0). As Sc substitution increases, the fraction of RO5_5 groups increases due to the preferable location of oxygen vacancies near Sc ions. The ratio of the 5 to 6 oxygen-coordinated Sc increases with increasing xx, significantly exceeding the ratio at uniform distribution of vacancies (Figure 13b). The latter indicates a substantial effect of vacancy trapping on their distribution. A similar qualitative trend in the behavior of the 5:6 coordinated Sc ratio as a function of xx was reported by Buannic et al. [29] for Sc-doped BaZrO3_3.

Figure 13. (a) Fractions of RO5_5 and RO6_6 groups and (b) their ratio as a function of dopant content xx. Lines represent theoretical dependencies for two ΔEV2R/ΔEVR\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} ratios: 1.5 (ΔEVR/kT=3.0\Delta E_{\mathrm{VR}}/kT = 3.0) and 2.0 (ΔEVR/kT=2.6\Delta E_{\mathrm{VR}}/kT = 2.6). The trapping energies (ΔEVR/kT\Delta E_{\mathrm{VR}}/kT) were determined from fitting of the experimental data [28]. Symbols correspond to the experimental values of five- and six-coordinated scandium (ScO5_5 and ScO6_6) in BaZr1−x_{1-x}Scx_xO3−δ_{3-\delta} extracted from the 45^{45}Sc NMR spectra [28]. The dashed black line in (b) indicates the ratio for the uniform distribution of vacancies over oxygen sites (6x/(6−x)6x/(6-x)).

Buannic et al. [29] also analyzed the 17^{17}O NMR spectra of BaZr1−x_{1-x}Scx_xO3−δ_{3-\delta}, assigning relative peak intensities to the Zr-O-Zr, Zr-O-Sc and Sc-O-Sc environments, which roughly correspond to the probabilities pfp_f, pRp_R and p2Rp_{2R}, respectively. For samples with 15 % Sc, the observed probabilities agree with a uniform Sc distribution, but at 30 % Sc, the experimental probability p2Rp_{2R} is lower, indicating a reduced concentration of R-O-R groups. The latter correlates with the high-temperature dopant distribution simulated by Kasamatsu et al. [30] (see § 3.4). It should be noted, however, that the experimental probabilities [29] pfp_f, pRp_R and p2Rp_{2R} do not satisfy the condition 2x=pR+2p2R2x = p_R + 2p_{2R} at x=0.3x = 0.3 and therefore cannot be considered as reliable descriptors of the dopant distribution.

We further compare our findings obtained within our model with the combined DFT and Monte Carlo simulations [31]. Hoshino et al. [31] calculated the equilibrium defect configurations in 22.2% Sc-doped BaZrO3_3 over a wide temperature range. The authors reported an almost uniform impurity distribution at ∼1900\sim1900 K. Figure 14 presents the temperature dependencies of the oxygen vacancy concentrations computed in Ref. [31] for a frozen high-temperature dopant configuration. We calculated the same dependencies within our model analytically taking into account weak inter-vacancy correlations and by Monte-Carlo simulations considering both weak and strong correlations. The results are presented for the energy parameters ΔEVR=0.3\Delta E_{\mathrm{VR}} = 0.3 eV and ΔEV2R/ΔEVR=2.0\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} = 2.0, which were chosen based on our fitting of the experimental data (see Figure 13) as well as the DFT calculations for Sc-doped BaZrO3\mathrm{BaZrO_3} [12, 14]. Figure 14 shows that the vacancy concentrations calculated for the case of strong inter-vacancy correlations almost perfectly match those reported in Ref. [31], while for weak correlations the results are noticeably different. Specifically, strong correlations result in a decrease in the concentration of R-V-R complexes and an increase in the concentration of R-V-B complexes at low TT. This indicates the importance of inter-site inter-vacancy repulsion effects in heavily acceptor-doped oxides, especially at low temperatures (see § 3.2). It should be emphasized, however, that the calculated equilibrium defect distribution is unlikely to be achieved in experiments at low temperatures, when the diffusivity of oxygen vacancies is low.

Concentrations of acceptor-bound and free oxygen vacancies as a function of temperature

Figure 14. Concentrations of acceptor-bound (cVRc_{\mathrm{VR}} and cV2Rc_{\mathrm{V2R}}) and free (cVfc_{\mathrm{Vf}}) oxygen vacancies in 22.2% Sc-doped BaZrO3\mathrm{BaZrO_3} as a function of temperature TT. Closed squares correspond to the results of the combined DFT and Monte Carlo simulations [31]. Solid lines and open symbols represent the results of analytical theory and Monte-Carlo modeling, respectively, obtained within our model at ΔEVR=0.3\Delta E_{\mathrm{VR}} = 0.3 eV and ΔEV2R/ΔEVR=2.0\Delta E_{\mathrm{V2R}}/\Delta E_{\mathrm{VR}} = 2.0. Analytical calculations account for weak inter-vacancy correlations, and Monte-Carlo simulations incorporate both weak and strong correlations.

A brief discussion of the proposed model: limitations and possibilities

Our considerations are based on the model proposed to explore the effects of oxygen vacancies interaction with frozen-in acceptor impurities and with each other (see § 2.1). In this model, the interaction with impurities and defect-induced lattice distortions is described using a three-level energy spectrum. These levels correspond to the three possible sets of the B-sublattice cations in the nearest environment of vacancies. The concentration and distribution of acceptor impurities are incorporated into defect thermodynamics through the probability of finding these sets. At first glance, such an approach seems oversimplified, since the model energy spectrum of vacancies does not depend on dopant content and configuration of impurities beyond the first neighbors.

However, similar models, treating acceptor-bound states of protons and vacancies even in a simpler form, provided a good description of hydration and proton transport in a number of acceptor-doped perovskites [10, 17, 18, 19]. This indicates that the contribution of long-range inter-defect interactions is not critical for the defect thermodynamics of the studied compounds. The pivotal role of the interaction between vacancies and their local environment is largely related to the high static permittivity ε0\varepsilon_0 of the considered wide-gap perovskites, which reduces the energy of long-range Coulomb interactions [17].

Another model limitation is related to long-range deformation interactions. These interactions can substantially contribute to the defects energy at a considerable ionic radii RiR_i mismatch between dopants and substituted host cations. According to the DFT results [24, 49], in perovskites with large-RiR_i dopants, the trapping energies of vacancies on the dopant nearest- and second nearest-neighbor oxygen sites can be comparable. In this case, our model is directly inapplicable, since the trapping zones for vacancies overlap even at moderate dopant content xx, and the energy spectrum of defect states becomes more complex and xx-dependent. Consideration of such a problem would require compound-specific analysis with a low degree of generality, which is not in line with the goals of this work. It should be also noted that dopants with large RiR_i typically exhibit low solubility and tend to substitute A-sublattice cations in the studied perovskites [50, 51, 52], which makes such materials less suitable for clean energy applications.

Obviously, the exploited approach cannot describe long-range effects, such, e.g., as ordering of oxygen vacancies. Note, however, that even in the case of materials with orderable defects, models with several defect energy states can be useful for analyzing bulk properties beyond phase transition regions. For example, a two-level model for orderable oxygen ions in the basal plane of YBa2Cu3O7−δ\mathrm{YBa_2Cu_3O_{7-\delta}} allowed us to describe pO2p_{\mathrm{O}_2}-TT-δ\delta diagrams and oxidation of this high-TcT_c oxide over a wide range of external conditions (with worse than the mean-field and Bethe-Kikuchi approximations, but acceptable accuracy), see [53] and Refs therein.

The inter-vacancy interaction in the proposed model is treated as infinite short-range repulsion: (1) on-site (Fermi-type correlations) and (2) on nearest neighboring sites. The latter reflects the high Coulomb interaction energy and expectable instability of an oxygen octahedron with two such vacancies in a cubic perovskite. Our results show that both types of inter-vacancy interaction can significantly affect defect thermodynamics, local structure, and oxidation. Fermi-type correlations are most pronounced at relatively low dopant content xx within a narrow xx range (see § 3.2, § 3.5). The effect of inter-site inter-vacancy repulsion increases with increasing xx and can become noticeable even at moderate doping levels in the low-temperature region (see § 3.2, § 3.6).

Despite the above-mentioned limitations, the proposed model is useful not only for qualitative, but, in some cases, for quantitative analysis of oxide properties (at moderate dopant content). The considerations based on this model have allowed us to elucidate the influence of vacancy-impurity and inter-vacancy interactions on the local structure and oxidation of AB1−xRxO3−δ\mathrm{AB}_{1-x}\mathrm{R}_{x}\mathrm{O}_{3-\delta} perovskites. The obtained results describe well the NMR experimental data on local structure and correlate with the hole conductivity data for acceptor-doped perovskites (see § 3.6). We also note that our predictions for the oxygen-vacancy distribution in 22.2% Sc-doped BaZrO3\mathrm{BaZrO}_{3} are consistent with the results of computationally intensive combined DFT and Monte-Carlo simulations (§ 3.6). Additionally, we have revealed the influence of the frozen non-uniform dopant distribution on the studied properties.

Furthermore, the model has predicted several interesting physical effects. In particular, we have shown the possibility of significant on-site Fermi-type correlations at relatively low dopant content and elucidate their implications. One of the most important predictions is related to the oxidation of acceptor-doped perovskites: we have revealed the possibility of a strong and unusual dependence of oxidation enthalpy and hole concentration on dopant content.

Finally, it should be noted that the proposed model can also be useful for a fairly wide class of problems in the theory of defect and transport phenomena in acceptor-doped perovskites. In particular, as we mentioned in § 3.5, the results on oxidation are essential for understanding transport processes and functioning of the studied oxides in various electrochemical devices. Another potential application of the model – important for employing these oxides as proton-conducting materials – is the analysis of their properties (related, e.g., to hydration) in hydrogen-containing atmospheres. Hydration theory taking into account the effects of acceptor-bound states and briefly considering the role of 3-particle defect complexes (including R-V-R) was previously developed in our work [17]. Several important implications of 3-particle complexes were revealed [17], and further developments of the hydration theory using the new findings presented here is under way. We would also like to point out that the proposed approach can be instrumental in developing a theory of transport phenomena of oxide-ion charge carriers. Sketches of such a theory for the oxygen-vacancy conductivity, mobility and oxide-ion thermopower can be found in our recent papers [54, 55].

Concluding remarks

In summary, we have investigated the effects of inter-defect interaction on the properties of wide-gap AB1−xRxO3−δ\mathrm{AB}_{1-x}\mathrm{R}_{x}\mathrm{O}_{3-\delta} perovskites under dry oxidizing conditions. Using the developed statistical theory and Monte Carlo simulations, we have elucidated the influence of these interactions, along with quenched impurity disorder, on the oxygen-vacancy distribution, local coordination of host ions and defects, and oxidation.

The effects of vacancy-impurity and vacancy-vacancy interactions manifest themselves differently depending on external conditions and dopant content xx. (a) Interaction between vacancies and impurities generally exerts a greater influence on the studied properties compared to inter-vacancy interaction. (b) The effect of on-site Fermi-type correlations, arising at high vacancy occupancy of oxygen sites surrounded by two impurities, can be considerable within a narrow dopant range at relatively low xx. (c) The impact of short-range inter-site inter-vacancy correlations increases with xx and can become noticeable even at moderate doping levels in the low-temperature region.

The main results regarding the studied properties can be summarized as follows:

  1. The concentrations of vacancies in various cation environments behave differently depending on dopant content. The fraction of vacancies located near a single impurity varies non-monotonically with xx, reaching a maximum at relatively low xx. The fraction of vacancies surrounded by two impurities increases with xx, and, under typical energy parameters, these vacancies become dominant even at moderate xx values;

  1. Inter-defect interactions result in the preferential formation of oxygen vacancies in the vicinity of acceptor impurities. Stronger vacancy-impurity interaction renders the local coordination increasingly non-uniform, while inter-vacancy interaction has the opposite and a significantly less pronounced effect;

  1. Inter-vacancy correlations result in the dependence of short-range order parameters on dopant content. At low doping levels, the effect can be largely due to Fermi-type correlations, whereas at high dopant content, inter-site vacancy repulsion plays a dominant role;

  1. Non-uniform dopant allocation, which can result from a sample preparation method, can noticeably influence the vacancy distribution over different types of oxygen sites, but has little effect on the local coordination and oxidation;

  1. Inter-defect interactions considerably affect the oxidation behavior of acceptor-doped perovskites, reducing the oxidation constant. These interactions can result in a strong and unusual dependence of oxidation enthalpy and hole concentration on dopant content.

The obtained results elucidate the implications of defect interactions and impurity disorder in acceptor-doped oxides, enabling the optimization of their properties required for applications through the selection of dopant type and content.

Conflicts of interest

There are no conflicts of interest to declare.

References

  1. [1]C. Duan, J. Huang, N. Sullivan, R. O’Havre, Proton-conducting oxides for energy conversion and storage, Appl. Phys. Rev. 7 (2020) 011314.DOI
  2. [2]A.V. Kasyanova, I.A. Zvonareva, N.A. Tarasova, L. Bi, D.A. Medvedev, Z. Shao, Electrolyte materials for protonic ceramic electrochemical cells: Main limitations and potential solutions, Mater. Rep.: Energy 2 (2022) 100158.DOI
  3. [3]L.A. Dunyushkina, S.A. Belyakov, N.M. Filatov, Proton-conducting alkaline earth hafnates: A review of manufacturing technologies, physicochemical properties and electrochemical performance, J. Eur. Ceram. Soc. 43 (2023) 6681–6698.DOI
  4. [4]S.A. Rasaki, C. Liu, C. Lao, Z. Chen, A review of current performance of rare earth metal-doped barium zirconate perovskite: The promising electrode and electrolyte material for the protonic ceramic fuel cells, Prog. Solid State Chem. 63 (2021) 100325.DOI
  5. [5]K.D. Kreuer, Proton-conducting oxides, Annu. Rev. Mater. Res. 33 (2003) 333–359.DOI
  6. [6]Y. Yamazaki, F. Blanc, Y. Okuyama, L. Buannic, J.C. Lucio-Vega, C.P. Grey, S.M. Haile, Proton trapping in yttrium-doped barium zirconate, Nat. Mater. 12 (2013) 647–651.DOI
  7. [7]E. Gilardi, E. Fabbri, L. Bi, J.L.M. Rupp, T. Lippert, D. Pergolesi, E. Traversa, Effect of dopant–host ionic radii mismatch on acceptor-doped barium zirconate microstructure and proton conductivity, J. Phys. Chem. C 121 (2017) 9739–9747.DOI
  8. [8]D. Han, K. Shinoda, S. Sato, M. Majima, T. Uda, Correlation between electroconductive and structural properties of proton conductive acceptor-doped barium zirconate, J. Mater. Chem. A 3 (2015) 1243–1250.DOI
  9. [9]K.D. Kreuer, S. Adams, W. Münch, A. Fuchs, U. Klock, J. Maier, Proton conducting alkaline earth zirconates and titanates for high drain electrochemical applications, Solid State Ionics 145 (2001) 295–306.DOI
  10. [10]L.P. Putilov, N.A. Shevyrev, A.M. Mineev, A.S. Farlenkov, D.A. Medvedev, V.I. Tsidilkovski, Hydration of acceptor-doped BaSnO₃: Implications of the bound states of ionic defects, Acta Mater. 190 (2020) 70–80.DOI
  11. [11]I. Oikawa, H. Takamura, Correlation among oxygen vacancies, protonic defects, and the acceptor dopant in Sc-doped BaZrO₃ studied by ⁴⁵Sc nuclear magnetic resonance, Chem. Mater. 27 (2015) 6660–6667.DOI
  12. [12]M.E. Björketun, P.G. Sundell, G. Wahnström, Structure and thermodynamic stability of hydrogen interstitials in BaZrO₃ perovskite oxide from density functional calculations, Faraday Discuss. 134 (2007) 247–265.
  13. [13]E. Bevillon, J. Hermet, G. Dezanneau, G. Geneste, How dopant size influences the protonic energy landscape in BaSn₁−ₓMₓO₃−ₓ/₂ (M = Ga, Sc, In, Y, Gd, La), J. Mater. Chem. A 2 (2014) 460–471.DOI
  14. [14]T.S. Bjørheim, E. Kotomin, J. Maier, Hydration entropy of BaZrO₃ from first principles phonon calculations, J. Mater. Chem. A 3 (2015) 7639–7648.DOI
  15. [15]A. Løken, T.S. Bjørheim, R. Haugstrud, The pivotal role of the dopant choice on the thermodynamics of hydration and associations in proton conducting BaCe₀.₉X₀.₁O₃−δ (X = Sc, Ga, Y, In, Gd and Er), J. Mater. Chem. A 3 (2015) 23289–23298.DOI
  16. [16]Y. Yamazaki, A. Kuwabara, J. Hyodo, Y. Okuyama, C.A.J. Fisher, S.M. Haile, Oxygen affinity: The missing link enabling prediction of proton conductivities in doped barium zirconates, Chem. Mater. 32 (2020) 7292–7300.DOI
  17. [17]L.P. Putilov, V.I. Tsidilk ovski, Impact of bound ionic defects on the hydration of acceptor-doped proton-conducting perovskites, Phys. Chem. Chem. Phys. 21 (2019) 6391–6406.DOI
  18. [18]L.P. Putilov, V.I. Tsidilkovski, Improving the performance of protonic ceramic fuel cells and electrolyzers: The role of acceptor impurities in oxide membranes, Energy Convers. Manage. 267 (2022) 115826.DOI
  19. [19]L.P. Putilov, V.I. Tsidilkovski, Proton conduction in acceptor-doped BaSnO₃: The impact of the interaction between ionic defects and acceptor impurities, Materials 15 (2022) 4795.DOI
  20. [20]V.I. Tsidilkovski, M.Z. Uritsky, A.N. Varaksin, A.Y. Fishman, The influence of defects on proton diffusion in perovskites AᴵᴵBᴵⱽ₁−xRᴵᴵᴵₓO₃−δ: Monte Carlo study, Defect Diffus. Forum 258–260 (2006) 124–129.
  21. [21]M.E. Björketun, P.G. Sundell, G. Wahnström, Effect of acceptor dopants on the proton mobility in BaZrO₃: A density functional investigation, Phys. Rev. B 76 (2007) 054307.
  22. [22]C.Y.R. Vera, H. Ding, J. Urban-Klaehn, M. Li, Z. Zhao, F. Stewart, H. Tian, X. Liu, Y. Dong, J. Li, M. Zhou, H. Luo, D. Ding, Improving proton conductivity by navigating proton trapping in high scandium-doped barium zirconate electrolytes, Chem. Mater. 35 (2023) 5341–5352.DOI
  23. [23]L.P. Putilov, V.I. Tsidilkovski, The role of deep acceptor centers in the oxidation of acceptor-doped wide-band-gap perovskites ABO₃, J. Solid State Chem. 247 (2017) 147–155.DOI
  24. [24]F.M. Draber, J.R. Denninger, P.C. Müller, I.K. Sommerfeld, M. Martin, The impact of nanoscale percolation in yttrium-doped BaZrO₃ on the oxygen ion and proton conductivities: A density functional theory and kinetic Monte Carlo study, Adv. Energy Sustainability Res. 3 (2022) 2200007.DOI
  25. [25]A. Lindman, P. Erhart, G. Wahnström, Polaronic contributions to oxidation and hole conductivity in acceptor-doped BaZrO₃, Phys. Rev. B 94 (2016) 075204.DOI
  26. [26]P.G. Sundell, M.E. Bjørketun, G. Wahnstrøm, Thermodynamics of doping and vacancy formation in BaZrO₃ perovskite oxide from density functional calculations, Phys. Rev. B 73 (2006) 104112.
  27. [27]E. Bevillon, G. Dezanneau, G. Geneste, Oxygen incorporation in acceptor-doped perovskites, Phys. Rev. B 83 (2011) 174101.DOI
  28. [28]I. Oikawa, M. Ando, Y. Noda, K. Amezawa, H. Kiyono, T. Shimizu, M. Tansho, H. Maekawa, Defects in scandium doped barium zirconate studied by Sc-45 NMR, Solid State Ionics 192 (2011) 83–87.DOI
  29. [29]L. Buannic, F. Blanc, I. Hung, Z. Gan, C.P. Grey, Probing the local structures and protonic conduction pathways in scandium substituted BaZrO₃ by multinuclear solid-state NMR spectroscopy, J. Mater. Chem. 20 (2010) 6322–6332.DOI
  30. [30]S. Kasamatsu, O. Sugino, T. Ogawa, A. Kuwabara, Dopant arrangements in Y-doped BaZrO₃ under processing conditions and their impact on proton conduction: a large-scale first-principles thermodynamics study, J. Mater. Chem. A 8 (2020) 12674–12686.DOI
  31. [31]K. Hoshino, S. Kasamatsu, J. Hyodo, K. Yamamoto, H. Setoyama, T. Okajima, Y. Yamazaki, Probing local environments of oxygen vacancies responsible for hydration in Sc-doped barium zirconates at elevated temperatures: In situ X-ray absorption spectroscopy, thermogravimetry, and active learning ab initio replica exchange Monte Carlo simulations, Chem. Mater. 35 (2023) 2289–2301.DOI
  32. [32]J. Hyodo, K. Kitabayashi, K. Hoshino, Y. Okuyama, Y. Yamazaki, Fast and stable proton conduction in heavily scandium-doped polycrystalline barium zirconate at intermediate temperatures, Adv. Energy Mater. 10 (2020) 2000213.DOI
  33. [33]Y. Wang, A. Chesnaud, E. Bevillon, J. Huang, J. Yang, Preparation and characterization of In-substituted BaSnO₃ compounds, Funct. Mater. Lett. 6 (2013) 1350041.DOI
  34. [34]I.A. Zvonareva, A.M. Mineev, N.A. Tarasova, X.-Z. Fu, D.A. Medvedev, High-temperature transport properties of BaSn₁−ₓScₓO₃−δ ceramic materials as promising electrolytes for protonic ceramic fuel cells, J. Adv. Ceram. 11 (2022) 1131–1143.
  35. [35]I.A. Zvonareva, G.N. Starostin, M.T. Akopian, G.K. Vdovin, X.-Z. Fu, D.A. Medvedev, Ionic and electronic transport of dense Y-doped barium stannate ceramics for high-temperature applications, J. Power Sources 565 (2023) 232883.
  36. [36]G.N. Starostin, M.T. Akopian, G.K. Vdovin, I.A. Starostina, G. Yang, D.A. Medvedev, Transport properties of highly dense proton-conducting BaSn₁−ₓInₓO₃−δ ceramics, Int. J. Hydrogen Energy 69 (2024) 306–316.DOI
  37. [37]C. De Dominicis, I. Giardina, Random fields and spin glasses. A field theory approach, Cambridge University Press, Cambridge, 2006.
  38. [38]T.C. Lubensky, in Ill-Condensed Matter, ed. R Balian, R Maynard, G. Toulouse, Thermal and geometrical critical phenomena in random systems, 1984, 405–475, DOI: 10.1142/9789814412728_0006.DOI
  39. [39]L.D. Landau, E.M. Lifshitz, Statistical Physics. Vol. 5, Butterworth-Heinemann, Oxford, 1980.
  40. [40]M.Z. Uritsky, V.I. Tsidilkovski, Role of an acceptor impurity in the proton transfer in proton-conducting oxides, Phys. Solid State 56 (2014) 2173–2180.DOI
  41. [41]J.M. Cowley, An approximate theory of order in alloys, Phys. Rev. 77 (1950) 669–675.DOI
  42. [42]J. Hermet, F. Bottin, G. Dezanneau, G. Geneste, Thermodynamics of hydration and oxidation in the proton conductor Gd-doped barium cerate from density functional theory calculations, Phys. Rev. B 85 (2012) 205137.DOI
  43. [43]J.B. Varley, A. Janotti, C. Franchini, C.G. Van de Walle, Role of self-trapping in luminescence and p-type conductivity of wide-band-gap oxides, Phys. Rev. B 86 (2012) 081109(R).
  44. [44]M. Swift, A. Janotti, C.G. Van de Walle, Small polarons and point defects in barium cerate, Phys. Rev. B 92 (2015) 214114.DOI
  45. [45]G. Geneste, B. Amadon, M. Torrent, G. Dezanneau, DFT+U study of self-trapping, trapping, and mobility of oxygen-type hole polarons in barium stannate, Phys. Rev. B 96 (2017) 134123.DOI
  46. [46]V.I. Tsidilkovski, L.P. Putilov, The role of deep acceptor levels in hydration and transport processes in BaZr₁−ₓYₓO₃−δ and related materials, J. Solid State Electrochem. 20 (2016) 629–643.DOI
  47. [47]D. Han, Y. Noda, T. Onishi, N. Hatada, M. Majima, T. Uda, Transport properties of acceptor-doped barium zirconate by electromotive force measurements, Int. J. Hydrogen Energy 41 (2016) 14897-14908.DOI
  48. [48]L.P. Putilov, V.I. Tsidilkovski, A.K. Demin, Revealing the effect of the cell voltage and external conditions on the characteristics of protonic ceramic fuel cells, J. Mater. Chem. A 8 (2020) 12641–12656.DOI
  49. [49]G. Geneste, G. Dezanneau, Competition between elastic and chemical effects in the doping, defect association, and hydration of barium stannate, Solid State Ionics 308 (2017) 121–132.DOI
  50. [50]D. Han, K. Shinoda, T. Uda, Dopant Site Occupancy and Chemical Expansion in Rare Earth-Doped Barium Zirconate, J. Am. Ceram. Soc. 97 (2014) 643–650.DOI
  51. [51]S. Imashuku, T. Uda, Y. Nose, G. Taniguchi, Y. Ito, Y. Awakura, Dependence of Dopant Cations on Microstructure and Proton Conductivity of Barium Zirconate, J. Electrochem. Soc. 156 (2009) B1-B8.DOI
  52. [52]L. Casillas-Trujillo, D.A. Andersson, B. Dorado, M. Nikl, K.E. Sickafus, K.J. McClellan, C.R. Stanek, Intrinsic defects, nonstoichiometry, and aliovalent doping of A²⁺B⁴⁺O₃ perovskite scintillators, Phys. Status Solidi B 251 (2014) 2279–2286.
  53. [53]V.I. Tsidilkovski, I.A. Leonidov, A.A. Lakhtin, V.A. Mezrin, The role of the electron-hole system in the thermodynamics of YBa₂Cu₃O₇−δ–gas equilibrium, Phys. Status Solidi B 168 (1991) 233–244.
  54. [54]L.P. Putilov, M.Z. Uritsky, V.I. Tsidilkovski, Oxygen-vacancy transport in acceptor-doped perovskites: effective medium approximation approach, Electrochem. Mater. Technol. 4 (2025) 20254055.DOI
  55. [55]L.P. Putilov, V.I. Tsidilkovski, Oxide-ion thermopower of acceptor-doped perovskites: The impact of inter-defect interactions, Electrochem. Mater. Technol. 4 (2025) 20254056.DOI

Paper details

vlnc:W_z26tkf8w
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

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