(image) (image) Previous Next

Impact of Charge Transitions at Atomic Defect Sites on Electronic Device Performance

Chapter 6 Optical Properties of Vacancies in Corundum (α-Al2O3)

The content presented in this section is currently prepared for publication [79]. The interpretation of the absorption line shapes is still under discussion.

In this chapter, optical transitions at vacancies in corundum (α-Al2O3) are analyzed within the framework of effective phonon modes as described in section 2.1.3. The cell parameters of the pristine α-Al2O3 bulk were relaxed until the residual stress was below 1 MPa, resulting in a mass density of 4.0 g cm3, which agrees with experimentally determined mass densities of corundum (3.984 g cm3 [261]). All Al atoms are bonded to 6 O atoms, while all O are bonded to 4 Al. Three of the six Al–O bonds have a bond length of 1.97 Å, the other three 1.86 Å. This is in perfect agreement with experimental data from x-ray measurements (1.97 and 1.86 Å [262]). All defect calculations were performed in a 240-atom 2x4x1 supercell built from the conventional cell of α-Al2O3 with space group R3c. The defect supercell containing a single vacancy was created by removing one atom and optimizing the geometry of the structure in different charge states. More details about the computational setup can be found in Appendix E.

6.1 Charge transition levels

An accurate calculation of the CTL is crucial for theoretically predicting optical properties of defects. As can be seen in Fig 2.1, the energy of the emitted or absorbed photon at the vacancy is related to the energetic position of the CTL relative to the band edges. The results for CTLs as calculated according to section 2.1.3 are listed in Table 6.1 and compared with previous results. While other details of the calculations may also play a role, the differences are mainly attributed to the use of different functionals employed in the DFT calculations.

Table 6.1: Calculated CTLs of vacancies in α-Al2O3. Results obtained with PBE0_TC_LRC are from this work; for comparison, results from previous work are listed, labeled with the functional used (in the case of LDA additional corrections were applied). All energy levels (in eV) are referenced to the VBM.
Defect ε(q1/q) PBE0 HSE [95] PBE [99]/[263] LDA [98]
VO ε(0/+1) 4.29 4.1 –/2.9 3.5
VO ε(+1/+2) 3.66 3.2 –/2.8 3.3
VAl ε(0/1) 1.89 1.4 0.4/0.1 0.6
VAl ε(1/2) 2.66 1.7 1.2/0.2 0.7
VAl ε(2/3) 3.26 2.6 1.3/0.4 1.5
VAl,s ε(0/1) 1.82 0.4/– 0.3
VAl,s ε(1/2) 2.39 0.7
VAl,s ε(2/3) 2.79 1.0
Band gap 8.92 9.2 5.7/7.5 6.9
6.1.1 Oxygen vacancy

The results for the formation energy are shown in Fig. 6.1 and the corresponding CTLs are given in Table 6.1 . For the VO, ε(0/+1) and ε(+1/+2) are located near midgap and follow a +2/+1/0 progression with increasing Fermi energy.

6.1.2 Aluminum vacancy

For both the VAl and VAl,s, ε(0/1), ε(1/2) and ε(2/3) are inside the band gap. For Fermi energies above 3.26 eV (2.79 eV), VAl3 (VAl,s3) is the most stable charge state. The formation energy of the VAl0 configuration is 57 meV lower than for the VAl,s0. In contrast, the energy of VAl is 16 meV higher for q=1, 280 meV higher for q=2 and 0.75 eV higher in energy for q=3 compared to VAl,s. These results qualitatively agree with the results of previous calculations, while quantitatively the calculated defect levels are considerably higher than with PBE and LDA due to the improved evaluation of the electronic band gap. As shown in Fig. 6.1(b), the VAl,s can be regarded as a complex consisting of an interstitial Al atom situated between two VAl sites along the c-axis [98, 99, 100, 101].

(image)

Figure 6.1: (a) Formation energy of the investigated vacancy configurations in α-Al2O3 as a function of the Fermi level. (b) Rendering of α-Al2O3 with a VAl,s defect relaxed in charge state q=3. Al are drawn in silver, O in red. In this vacancy configuration, an Al occupies the interstitial site between to vacant Al sites (drawn in green) along the c-axis.

The migration path for the transition of a VAl to a VAl,s configuration was calculated with the CI-NEB method for q=0 and q=3 and the transition states along the migration path further optimized with the dimer method. The Al atom that moves towards the vacancy is originally at an adjacent lattice site along the c-axis. As this Al atom moves towards VAl, it localizes at a new minimum-energy configuration corresponding to the VAl,s site. The energy barrier Ea therefore occurs for a position of the Al atom in between the VAl and VAl,s configurations. Ea in different charge states are given in Table 6.2.

Table 6.2: Migration barriers between VAl and VAl,s configurations in different charge states.
Migration path q=0 q=3
Ea(VAlVAl,s) [eV] 1.97 1.17
Ea(VAl,sVAl) [eV] 1.92 1.92

These results indicate that the VAl,s is the most prevalent cation deficiency configuration in α-Al2O3. The large energy difference of 0.75 eV for q=3 indicates that most vacancies in the system occupy the split state compared to VAl, since the occupation ratios are given by the Boltzmann factor e(dE/kBT). Furthermore, once a VAl,s defect forms, it is less likely to transition to a VAl site compared to the reverse process.