Theoretical Study of Native Point Defects in AlN and InN
- PDF / 423,248 Bytes
- 6 Pages / 414.72 x 648 pts Page_size
- 67 Downloads / 190 Views
ABSTRACT We have studied native point defects in AIN and InN using density-functional calculations employing both the local-density and generalized gradient approximations for the exchange-correlation functional. For both materials we find that the nitrogen vacancy acts as a compensating center in p-type material. For AIN in the zinc-blende structure, the aluminum interstitial has an equally low formation energy as the nitrogen vacancy. For n-type material the aluminum vacancy is the dominant compensating center in AIN. For n-type InN, all defect formation energies are high. INTRODUCTION Much progress has been made in the growth and fabrication of optoelectonic devices based on III-V nitrides [1-3]. The particular material properties responsible for this interest are the large band gaps and very strong interatomic bonds. Further investigations are still required, however, for example concerning the control of doping levels; the issues include suppression of background n-type conductivity, achieving and understanding n- and p-type doping of AlGaN alloys, and compensation effects by native defects. Valuable insight into such problems can be obtained from first-principles calculations which provide information about the atomic and electronic structure, formation energies, and compensation effects. In the present work we have studied all native point defects in the relevant charge states in AIN and InN. The character and behavior of the defects are in many ways similar to those
which have been reported in GaN [4,5]. We find, however, several important differences which have implications for compensation issues. We also compare results obtained using the local-density approximation (LDA) [6] and the generalized gradient approximation (GGA) [7] for the exchange-correlation functional. Below we give a brief description of the theoretical method and discuss our results for AIN and InN in the two subsequent sections. We then compare LDA and GGA results. The final section contains the conclusions. THEORETICAL METHOD The calculations are performed using the density-functional pseudopotential method [8]. The wave functions are expanded in a plane-wave basis set and the system modeled using the supercell approach. We employ a tight-binding initialization scheme for the electronic wave functions to improve convergence. The soft pseudopotentials are created using the scheme of Troullier and Martins [9], and for the GGA calculations we include the GGA in creation of the pseudopotential as well as in the self-consistent total energy calculations so that the description is consistent [10]. We primarily treat the indium 4d electrons using the non-linear core correction (nlcc) introduced by Louie et al. [11]. For several defects we have also performed calculations including the 4d electrons as valence states, confirming our results. The majority of the calculations are performed using 32-atom zinc-blende 905
Mat. Res. Soc. Symp. Proc. Vol. 482 01998 Materials Research Society
supercells. We have also investigated low-energy defect
Data Loading...
