1283
N- and P- type Doping in Al-rich AlGaN and AlN

Tuesday, 2 October 2018: 14:40
Universal 5 (Expo Center)
B. Sarkar, S. Washiyama, M. H. Breckenridge, A. Klump, J. N. Baker (North Carolina State University), P. Reddy, J. Tweedie, S. Mita, R. Kirste (Adroit Materials Inc.), D. L. Irving, R. Collazo, and Z. Sitar (North Carolina State University)
Significant challenges in point defect control in AlGaN epitaxy has precluded commercialization of AlGaN based devices. Si is typically employed as n-type dopant in AlGaN and exhibits a low activation energy (<50 meV) in AlxGa1-xN with x<0.8. However, Si doped AlGaN exhibits a “knee behavior” resulting in a conductivity and carrier concentration maxima at a specific Si concentration. Hence a high doping limit exists for Si in AlGaN that lowers the maximum achievable carrier concentrations that are necessary for AlGaN based power electronics. Similarly, a low doping limit (a minimum achievable carrier concentration with a corresponding maximum mobility) exists in AlGaN similar to that in GaN which precludes implementation of AlGaN power electronics that require low doped drift regions. Along the same lines, p-type doping is achieved with Mg, but high activation energies preclude its technological implementation. Following results in GaN, compensating point defects are also expected to play a major role in determining the doping efficiency of Mg, even if methods to “reduce” its activation energy are found. Hence a major “point defect problem” exists in AlGaN that needs to be solved for implementation of AlGaN technology.

In this work, we update on three different achievements: (1) identifying and controlling point defects for efficient n-type doping in AlGaN, (2) implantation as a method for achieving high n-type conductivity in AlN and (3) the possibility of high conductivity p-type AlGaN. All these achievements are built upon a point defect control (PDC) framework based on a systematic chemical potential control (CPC) and defect quasi Fermi level control (DQFL). In CPC, the growth environment variables are related to the defect formation energy by determining and controlling the Al,Ga,N and impurity chemical potentials. This helps optimize the growth environment accordingly for minimal point defect incorporation or generation. In DQFL control, the defect quasi Fermi level, the Fermi level describing the probability of the defect level being occupied/unoccupied, is modified by introducing excess minority carriers (by above bandgap illumination). This leads to an increase in the compensating point defect formation energy, reducing their concentration. Both control schemes provide for a theoretical framework that provides a quantitative relationship between point defect formation energies and growth process parameters.

CPC based targeted reduction requires knowledge of the defects responsible and accordingly, we identify the point defects responsible for knee behavior in n-type AlGaN as most likely vacancy-silicon complexes and low doping limit as CN. Note that threading dislocations and vacancy-oxygen complexes are also expected to be compensating and contribute to the low doping limit. Utilizing the vacancy-silicon complex and CN in Al0.7Ga0.3N as a case study, we demonstrate control over the knee behavior (improving the peak carrier concentration by impeding the formation VIII-SiIII complexes) and low doping limit (achieving lower carrier concentrations by reducing the compensating impurity (CN) density) in Al0.7Ga0.3N by controlling the chemical potentials of III/N and growth temperature with excellent agreement with quantitative theoretical predictions. These concepts are further used in Si implanted AlN where a reduction in the formation of vacancy-silicon complexes during implantation annealing is necessary. This was achieved by implementing DQFL control during the implantation annealing stage. Such process led to an increase of nearly 3 orders of magnitude in the conductivity of n-type AlN at room temperature. Finally, after recognizing the importance of point defect reduction during doping, significant achievements in p-type AlGaN were realized. Lower Mg activation energies than those previously reported were observed leading to significant p-type conductivity in AlGaN.

All these results support our PDC framework and the utility of formation energies from equilibrium formalism typically employed in DFT in predicting defect incorporation in non-equilibrium and high temperature MOCVD growth.