3.2. Intrinsic Defects in ZrO
First, we calculate the defect formation energies of intrinsic defects in ZrO, including vacancy defects and interstitial defects. For vacancies, there are three types of atomic positions (as shown in
Figure 1), resulting in three kinds of vacancy defects labeled as V
Zr1, V
Zr2, and V
O. As for interstitial defects, due to the lower symmetry of the hexagonal structure, we identify two types of interstitial defects, as shown in
Figure 1, which are named X
i-tet and X
i-mid (X = O/Zr). The former is located in the tetrahedral interstitial position while the latter is situated between two Zr atoms.
Table 3 lists all the defect formation energies of intrinsic defects in ZrO. We observe that the formation energy of V
O is 5.31 eV. The vacancy formation energy of V
Zr1 is 3.94 eV, while that for V
Zr2 is 1.99 eV, indicating that V
Zr2 is easier to form than V
Zr1. However, all vacancies exhibit positive formation energies, suggesting that vacancies are unfavorable to form in ZrO. The lowest formation energy corresponds to the oxygen interstitial defect O
i-tet, with a formation energy of −4.04 eV. This indicates that oxygen interstitials can form spontaneously in ZrO. The other interstitial defect of O
i-mid exhibits a slightly positive formation energy of 0.03 eV. This suggests that interstitial oxygen could exist in high concentrations in ZrO under equilibrium. The ZrO phase incorporates a significant amount of dissolved oxygen (O). Once oxygen is dissolved within the solid, it can no longer diffuse further into the underlying metal layer. When the concentration of dissolved oxygen reaches a critical threshold, ZrO undergoes oxidation, leading to the disappearance of the ZrO phase. Without this phase to accommodate additional oxygen, a large amount of oxygen accumulates at the metal/oxide interface, thereby triggering the corrosion transition stage. This phenomenon appears to explain why the ZrO phase emerges prior to the corrosion transition stage but disappears once the transition occurs. The formation energies of Zr interstitial defects are significantly higher than those of O interstitial defects, likely due to atomic size. The larger diameter of Zr atoms makes it challenging for Zr interstitials to stably occupy tetrahedral interstitial positions. In addition, interstitial sites only coordinate with Zr atoms at lattice sites. The coulombic repulsion between metallic atoms in the oxide also significantly increases the formation energies of interstitial Zr atoms in ZrO.
Similarly, the formation energies of interstitial Zr are excessively positive (both exceeding 5 eV), leading us to exclude the migration of Zr interstitial defects from further discussion. The elevated formation energies of Zri are consistent with the observation that ZrOx (0 < x < 1) is a metastable phase. Overall, the most stable defect in pure ZrO is Oi in the tetrahedral interstitial position.
Based on the density of states (DOS) diagram, we conclude that ZrO behaves as a metallic conductor; thus, the properties of charged defects are not considered in the subsequent studies. As shown in
Table 4, previous calculations for t-ZrO
2 by Youssef et al. [
25] indicate that the formation energies of intrinsic defects are higher in t-ZrO
2. Compared with the calculation results in this paper, it can be found that intrinsic defects form more readily in ZrO rather than in the tetragonal phase. Zheng et al. [
26] provided a detailed analysis of intrinsic defects in m-ZrO
2. According to their findings, the formation energy of oxygen vacancy exceeds 6 eV, while the formation energy of Zr vacancy also approaches nearly 6 eV. The formation energies of other interstitial defects were higher than those observed in our study, suggesting that point defects are more likely to form in ZrO.
In general, ZrO exhibits the lowest formation energies of defects, with minimal differences in the formation energies of oxygen vacancies across the three phases of zirconium oxide. Notably, there is a significant discrepancy in the formation energies of interstitial oxygen among the three zirconium oxides: the formation energy of Oi in ZrO is negative, indicating that it can form spontaneously under appropriate conditions, whereas the formation energy in the other two types of ZrO2 is positive, suggesting that it would not be formed spontaneously.
Furthermore, we investigated the migration behaviors of intrinsic defects in ZrO oxide using the climbing image nudged elastic band (CI-NEB) method [
52]. As shown in
Figure 3, we assumed that the oxygen vacancy could diffuse in two directions: intra-plane and inter-plane. For intra-plane diffusion, we selected two diffusion paths to analyze the migration of the oxygen atom. One path (Path 1) is between the nearest two neighboring vacancies, separated by 2.798 Å, while the other path (Path 2) is between the next-nearest two neighboring vacancies separated by 3.734 Å. For inter-plane diffusion, we considered two vertical oxygen vacancies (Path 3) in the c-direction.
Table 5 presents the migration energies of the oxygen vacancies. According to the results obtained from the climbing image nudged elastic band (CI-NEB) method, it is evident that the oxygen atom is more inclined to diffuse along Path 2, which has a migration barrier of 2.96 eV. The migration barrier for oxygen vacancy diffusion along Path 1 is 3.28 eV. Interestingly the longer diffusion path results in a lower migration energy. The migration barrier for oxygen vacancies diffusing between planes lies between the barriers of the other two paths, suggesting that the diffusion of oxygen vacancies is inherently difficult.
Additionally, Youssef et al. [
53,
54] calculated the migration barriers of O vacancies in t-ZrO
2 and m-ZrO
2, finding ranges of 1.35 eV~3.96 eV and 1.84 eV~2.48 eV, respectively. This indicates that the diffusion of oxygen vacancies in ZrO
2 is more favorable.
Since Zr has two types of inequivalent positions, we consider migration between equivalent sites and non-equivalent sites. As shown in
Figure 4, we selected four migration paths for Zr vacancies. The migration paths are named as follows: Since the formation energy of V
Zr2 is lower than that of V
Zr1, the system energy naturally tends to decrease, indicating that Zr vacancy migrates from V
Zr1 to V
Zr2. This migration is called Path I. The interplanar diffusion of V
Zr1 along the c direction is named Path II. The in-plane diffusion of V
Zr2 is called Path III, which is the shortest diffusion path for the Zr vacancy. The diffusion of V
Zr2 between planes along the c direction is set as Path IV.
Table 6 displays the migration barriers for Zr vacancy defects. From our previous calculations of defect formation energies, it is evident that V
Zr2 is easier to form. However, in the calculation of CI-NEB calculations, we found that the diffusion of V
Zr between two V
Zr1 sites has the lowest diffusion barrier of 2.44 eV. This indicates that the Zr vacancies tend to migrate between vacancies at the V
Zr1 site, while the diffusion barrier is maximized when the Zr vacancy moves between the V
Zr1 site and V
Zr2 sites. Furthermore, it is apparent that Zr vacancies residing at equivalent positions exhibit relatively lower migration barriers when diffusing along shorter paths, which aligns with our previous speculations.
The two interstitial positions of the oxygen atom in ZrO exhibit distinct formation energies. The interstitial oxygen in the tetrahedral interstitial position has a significantly lower defect formation energy compared to that in the mid-position between Zr atoms; thus, we do not consider the case of O
i-mid in subsequent discussions. As shown in
Figure 5, for the interstitial oxygen in the tetrahedral interstitial position, we investigated two diffusion mechanisms, selecting two paths for each mechanism. In the hop mechanism, the interstitial oxygen atom diffuses from one tetrahedral interstice to another. In contrast, the kick-off mechanism involves the diffusion of two oxygen atoms: the interstitial atom moves from the tetrahedral interstice to the nearest lattice site, while an atom from the original lattice site migrates to another tetrahedral interstice.
Table 7 presents the migration energies associated with all aforementioned paths. The results indicate that all migration paths of interstitial oxygen atoms in ZrO exhibit high migration barriers. The path with the lowest migration energy occurs in the intra-plane hop mechanism, corresponding to a migration energy of 2.91 eV. This suggests oxygen atoms in tetrahedral interstices encounter significant difficulty in diffusing, which aligns with the experimental observations. Moreover, the migration of interstitial atoms via the kick-off mechanism typically has a lower migration barrier. Notably, the intra-plane hop mechanism for oxygen interstitials is energetically favored in ZrO. Conversely, migration between planes adheres to general trends, as interstitial atoms typically do not traverse across a plane.
3.3. Effect of Alloying Elements in ZrO
Many researchers have demonstrated that the incorporation of alloying elements significantly affects the corrosion rate of zirconium alloys [
55,
56,
57,
58]. Appropriate additions of these elements can enhance the corrosion resistance of the matrix. In our investigation, we selected four common alloying elements used in zirconium alloys: Cr, Fe, Nb, and Sn. Given that zirconium possesses two inequivalent sites, we first evaluated the favorable substitution sites for each alloying element.
Table 8 presents the substitution energies of all identified defects. Notably, the substitution energies are all positive, indicating that substitution does not occur spontaneously; however, sites with lower substitution energies are still considered stable. The substitution of Cr at the Zr2 site is more favorable, corresponding to a substitution energy of 3.27 eV. Similarly, Fe tends to substitute at the Zr2 site, with a corresponding substitution energy of 3.26 eV. Conversely, Nb and Sn exhibit lower substitution energies at the Zr1 site, corresponding to substitution energies of 2.04 eV and 3.17 eV, respectively. As indicated by the density of states (DOS) diagram, ZrO behaves as a metallic crystal; hence, we do not account for the charged state of the substituting elements.
Once the substitution sites of the alloying elements were determined, we investigated their influence on the behavior of oxygen defects. Regardless of where the substitution defects occur, we considered both the nearest and next-nearest oxygen vacancies.
Table 9 summarizes the calculated formation energies of oxygen vacancies with and without substitution defects.
By comparing the formation energies of oxygen vacancies in pure ZrO and those in ZrO with substitution defects, we found that the presence of alloying elements reduces the formation energies of oxygen vacancies, indicating that it becomes easier to form oxygen vacancies in ZrO when substitution defects are present. The results in
Table 9 further demonstrate that, regardless of the substitution site, the likelihood of losing an oxygen atom from the nearby lattice to form an oxygen vacancy increases. A positive formation energy for the oxygen vacancy indicates that oxygen vacancies do not form spontaneously in ZrO. The formation energy of an oxygen vacancy with Sn
Zr in ZrO is reduced to 4.39 eV, the lowest among the four substitution defects. Additionally, as the distance between the oxygen vacancy and the substitution defect increases, the impact of the substitution defect on reducing vacancy formation energy diminishes. Notably, the formation energy of the oxygen vacancy near the substitution defect Fe
Zr is the highest.
To elucidate the mechanism by which alloying elements enhance corrosion resistance, we investigated the diffusion barriers of oxygen vacancies in ZrO with substitutions using the climbing image nudged elastic band (CI-NEB) method. Similar to the migration paths in pure ZrO, we considered three types of paths: diffusion of the oxygen vacancy between the two nearest positions adjacent to the substitution defect (Path 1), diffusion between the nearest and next-nearest positions (Path 2), and vertical diffusion (Path 3).
Table 9 displays the migration energies for these paths in ZrO with substitution defects.
It is evident that the oxygen vacancy has a lower migration energy with a CrZr substitution defect in ZrO, with calculated migration barriers decreasing by 1.34 eV, 0.16 eV, and 1.13 eV, for Path 1, Path 2, and Path 3, respectively. These results suggest that oxygen vacancies in ZrO with Cr dopants exhibit higher mobility compared to those in pure ZrO. For Path 2, the migration energy reaches 2.80 eV, which may indicate that Cr exerts a certain repulsive effect on the oxygen atom. In the case of the alloying element Fe, a strong chemical interaction appears to exist between Fe and the oxygen atom, resulting in non-convergence of the calculation for Path 3. However, for Paths 1 and 2, the migration barriers for oxygen vacancies both decreased. Interestingly, the Fe atom also migrated alongside the vacancy during the relaxation process. Conversely, in ZrO with Nb dopants, the results are quite different: the migration barriers for Paths 2 and 3 slightly increase, while the barrier for Path 1 decreases by 0.69 eV from 3.28 eV to 2.59 eV. The migration barriers of oxygen vacancies in ZrO with Sn dopants increase to 3.95 eV and 3.76 eV, respectively, for Path 2 and Path 3, while the calculation of the migration barrier for Path 1 does not converge, as the oxygen atom transitions to a tetrahedral interstice position during the calculation.
Overall, the migration barriers of oxygen vacancies provide insight into the motion of oxygen atoms. Based on the calculated formation energies of oxygen vacancies, it is evident that oxygen vacancies are more mobile in ZrO with substitution defects. Thus, migration barriers serve as a standard for evaluating the influence of alloying elements. Consequently, it is clear that Sn acts as a beneficial alloying element that decreases the mobility of oxygen atoms.
Additionally, to further understand the influence of alloying elements on the corrosion resistance of zirconium oxide, we investigated the behavior of interstitial oxygen atoms in ZrO doped with alloying elements. Similar to the interstitial oxygen located in pure ZrO, we chose the tetrahedral interstitial sites, composed of three zirconium atoms and one substitutional alloying atom, as the preferred interstitial positions. Based on previous calculations, we observed that the effect of substitutional atoms on oxygen vacancies depends on the distance between the oxygen atom and the alloying atom. Therefore, we sought to determine whether the same trend applies when interstitial oxygen atoms are introduced into the system.
To explore this, we selected three tetrahedral interstitial sites at varying distances from the substitutional alloying atom, denoted as O
ia, O
ib, and O
ic, with increasing distance from the alloying defect as shown in
Figure 6.
Table 10 summarizes the calculated formation energies for interstitial oxygen atoms in both pure ZrO and ZrO doped with alloying elements.
Table 11 lists the formation energies of all interstitial oxygen atoms. Compared to pure ZrO, we found that the formation energies of interstitial oxygen atoms in ZrO doped with substitutional defects generally increase, although they remain negative. This implies that, while the interstitial oxygen atoms exhibit slightly higher formation energies in doped systems, they are still thermodynamically stable and would persist in ZrO regardless of doping.
Interestingly, the influence of alloying elements on the formation energy of interstitial oxygen varies depending on the specific dopant. For Cr and Fe substitutions, no consistent trend was observed with respect to the distance between the substitutional alloying atom and the interstitial oxygen atom. However, in the case of NbZr and SnZr defects, the formation energies of interstitial oxygen decreased as the distance between the substitutional defect and the interstitial oxygen increased. This suggests that alloying elements such as Nb and Sn have a stabilizing effect on interstitial oxygen at greater distances, while Cr and Fe exhibit more localized or complex interactions.
We schematically depict the migration pathways of interstitial oxygen atoms in
Figure 7. Two migration mechanisms were considered, as in the case of pure ZrO: the hop and kick-off mechanisms. However, an additional factor must be accounted for in the doped system. While we already understand that the effect of alloying elements on the formation energy of interstitial oxygen is not dependent on distance, it remains necessary to confirm whether their influence on the migration behavior of interstitial oxygen is distance-dependent.
Table 12 presents the migration energies of interstitial oxygen in both pure ZrO and ZrO doped with alloying elements. Interestingly, although the formation of interstitial defects becomes more challenging in all doped systems, the migration of oxygen atoms exhibits varying behaviors based on the type of alloying element.
The results can be categorized based on the substitution sites of the alloying elements. When Nb and Sn occupy the Zr1 site, they reduce the migration barrier of oxygen atoms in nearby interstitial sites. As with Cr and Fe, the influence of Nb and Sn diminishes with increasing distance from the interstitial defect. Notably, Sn maximizes the formation probability of interstitial oxygen defects while minimizing the migration barrier of nearby interstitial oxygen atoms.
In contrast, when Cr and Fe occupy the Zr2 site, they increase the migration barrier of the interstitial oxygen atom in nearby sites. As the distance between the substitutional atom and the interstitial oxygen atom increases, the migration barrier decreases, suggesting that the effect of alloying elements on the migration barrier is distance-dependent. The kick-off mechanism consistently exhibits a lower barrier compared to the hop mechanism. Additionally, the impact of Cr and Fe on migration barriers follows a similar trend.