
1 Introduction
Ultra-wide bandgap (UWBG) semiconductors have
recently emerged as an exciting class of materials
due to their potential applications in power elec-
tronics, optoelectronics, and radio frequency devices.
UWBG materials are defined as materials that have
a bandgap larger than that of GaN (3.4 eV)[1]. The
large bandgap of materials impacts many device per-
formance parameters, such as thinner drift layers
and lower specific on resistance, allowing significant
miniaturization of devices like switches and transis-
tors. Furthermore, UWBG materials are generally
characterized by high bandgap, very large break-
down fields (>106V/cm), high thermal conduc-
tivity, and reduced impact ionization rates and tun-
nelling due their high bandgaps and high mechanical
strengths. AlN and BN are the largest gap group-
III nitrides, and AlN has already been explored for
application in the field of power electronics and in
UV device applications[2–4]. AlGaN alloys have also
been used extensively in power electronics due to the
tunability of their bandgaps between the bandgap of
AlN (6.2 eV) and that of GaN (3.4 eV).
Realization of BxAl1−xN alloys could allow fur-
ther tunability of bandgaps beyond what is avail-
able via AlGaN alloys. Figure 1 shows the crys-
tal structures of the wurtzite AlN and BN, as well
as the ground state phase of BN, the hexagonal
phase. The predicted bandgap of w-BN ranges from
5.44 to 7.70 eV[5], much larger than that of GaN
(3.4 eV). Additionally, BxAl1−xN alloys are expected
to display high dielectric constants[6]. Therefore,
BxAl1−xN may have tunable dielectric constants and
high bandgaps without losing the other excellent me-
chanical and thermal properties. However, there are
several challenges in the realization of BxAl1−xN al-
loys. For example, the lattice mismatch between
AlN and BN is very large, 18%. Moreover, BN has a
preference for existing in the hexagonal phase rather
than the metastable wurtzite phase, which makes
introduction of boron into the wurtzite AlN lattice
challenging. Consequently, it leads to the formation
of polycrystalline BN phases in BxAl1−xN alloys and
a high density of grain boundaries that can be detri-
mental to device design and performance[7].
Despite these challenges, thin-films of w-BxAl1−xN
have been successfully grown on AlN and sapphire
substrates in very recent years with thicknesses of up
to 300 nm and B-fractions up to x= 0.30[2,6,7,11–15].
Figure 1: The crystal structures of (a) wurtzite phase of alu-
minum nitride as well as (b) hexagonal, (c) cubic and (d)
wurtzite phase of boron nitride.
In addition, experimental efforts have been made to-
wards understanding their crystal growth and struc-
tural properties[7,11,14].
Theory and simulations have been used to study
the structure of a broader range of B-fractions of
BxAl1−xN alloys. Mainly two classes of methods
have been used. The first is the method of cation
substitution in the w-AlN host lattice[15–17]. The B-
fractions that can be studied using this method are
dependent on the size of the supercell used to gen-
erate the possible BxAl1−xN structures. Also, as the
BxAl1−xN alloys are highly mismatched alloys, the
restriction of the lattice to the wurtzite phase may
preclude other structures that could be formed in
BxAl1−xN . Another class of methods is the one used
by Ahmed et al that uses density-functional theory
(DFT) based evolutionary structure searches. This
method overcomes the limitations associated with
supercell size and constraints set on the symmetry
of the crystal to predict the structures. However,
due to the high computational resource and time
requirement of this method, only three B-fractions,
x= 0.25,0.5,0.75, were studied by Ahmed et al [18].
In this study, we investigate the structure and
bonding environment of BxAl1−xN over the entire
boron fraction range using the ab-initio cluster ex-
pansion (CE) method. In this method, special quasi-
random alloys structures are predicted based on a
2