Passivation of the beryllium acceptor in GaN and a possible route for p-type doping

Abstract

Using theory and experiment, we address a possible p-type doping route of GaN using the Be acceptor. Hybrid functional calculations suggest that straightforward incorporation of Be in GaN will likely lead to compensated high resistivity samples. In this case, beryllium interstitials and nitrogen vacancies are the most likely compensating donors. However, calculations also show that simultaneous incorporation of beryllium and hydrogen in the nitrogen-rich growth regime is likely to lead to the formation of neutral BeGa-Hi complexes. These complexes are energetically favorable and do not exhibit any transition levels in the bandgap. Subsequent annealing in hydrogen-free or nitrogen ambient will dissociate these complexes and activate shallow BeGa acceptors in GaN. Experimental measurements, where an ultraviolet photoluminescence band related to the shallow BeGa acceptor emerges after annealing the Be-doped GaN, confirm these findings.

Direct evidence of Be as an amphoteric dopant in GaN

“…While both light group II atoms can occupy substitutional Ga and interstitial sites, the amphoteric character of Be is considerably more pronounced. Interstitial fractions of 11 Be were found substantially higher than previously for the case of 27 Mg [7,8], which supports theoretical predictions [4,16,19,[21][22][23]25,31] that Be is more prone to self-compensation and thus unsuitable for p-type doping. The exact locations of the interstitial sites of Mg parallel to the c-axis (−0.6±0.14 Å from the O site in p-GaN) and Be (+0.69±0.08 Å) differ, with Be i located closer to the N atoms.…”

“…From the theoretical side, Be is one of the most studied impurities in GaN and considerable efforts have been made in order to understand its doping behavior [4,[16][17][18][19][20][21][22][23][24][25][26][27][28][29][30][31][32]. However, up to now no consensus has been reached regarding theoretical predictions of its acceptor level(s), for which values of <200 meV [16], 60 meV [17], 185-241 meV [18], 183-190 meV [20], 60 meV [22], 120 meV [23], 550 meV [25], 47-93 meV [26], 600 meV [28], 720-800 meV [29], and 205 meV [31] can be found in the literature. One theoretical study [24] and recent experimental results [30] point towards the hypothesis that Be has two acceptor levels, a relatively shallow one (EA≈113 meV), and a second, which is deeper (E A ≈580 meV) and characterized by large lattice relaxations.…”

“…One theoretical study [24] and recent experimental results [30] point towards the hypothesis that Be has two acceptor levels, a relatively shallow one (EA≈113 meV), and a second, which is deeper (E A ≈580 meV) and characterized by large lattice relaxations. Moreover, several theoretical studies [4,16,19,[21][22][23]25,31] have predicted that Be is an amphoteric impurity in GaN, meaning that it can occupy substitutional Ga positions (Be Ga ), where it acts as an acceptor, but also interstitial sites (Be i ), where it acts as a double donor. This implies that the formation of interstitial Be i instead of substitutional Be Ga should become more favorable in p-GaN, which was forecast to be the case once the Fermi level is closer than ≈1.2 eV [16], 0.8 eV [19], 0.5 eV [21], 0.3-0.5 eV [22], 1.4-1.8 eV [25], or 1.2-1.6 eV [31] from the valence band.…”

“…Moreover, several theoretical studies [4,16,19,[21][22][23]25,31] have predicted that Be is an amphoteric impurity in GaN, meaning that it can occupy substitutional Ga positions (Be Ga ), where it acts as an acceptor, but also interstitial sites (Be i ), where it acts as a double donor. This implies that the formation of interstitial Be i instead of substitutional Be Ga should become more favorable in p-GaN, which was forecast to be the case once the Fermi level is closer than ≈1.2 eV [16], 0.8 eV [19], 0.5 eV [21], 0.3-0.5 eV [22], 1.4-1.8 eV [25], or 1.2-1.6 eV [31] from the valence band. The lack of p-type characteristics following Be doping could thus be explained by self-compensation, i.e.…”

Substantial P‐Type Conductivity of AlN Achieved via Beryllium Doping

“…In another study, Demchenko and Reshchikov studied the formation energies of Be‐related energy levels in GaN via Heyd–Scuseria–Ernzerhof (HSE) hybrid functional calculations. [ 27 ] From HSE calculations, they found that the formation energy of Be Ga –H i complex is much lower as compared to the formation energy of Be on a Ga site (Be Ga ) under both N‐rich and Ga‐rich growth conditions, concluding that the presence of hydrogen enhances Be incorporation during growth. In our prior metal modulated epitaxy (MME) work, we did not see any limitation in the incorporation of Be into GaN nor did we see a need to use hydrogen to enhance its incorporation.…”