Quantum Monte Carlo Study of Semiconductor Artificial Graphene Nanostructures
G¨okhan ¨
Oztarhan, E. Bulut Kul, Emre Okcu, and A. D. G¨u¸cl¨u
Department of Physics, ˙
Izmir Institute of Technology, 35430 Urla, ˙
Izmir, Turkey
(Dated: October 3, 2023)
Semiconductor artificial graphene nanostructures where Hubbard model parameter U/t can be of
the order of 100, provide a highly controllable platform to study strongly correlated quantum many-
particle phases. We use accurate variational and diffusion Monte Carlo methods to demonstrate a
transition from antiferromagnetic to metallic phases for experimentally accessible lattice constant
a= 50 nm in terms of lattice site radius ρ, for finite sized artificial honeycomb structures nanopat-
terned on GaAs quantum wells containing up to 114 electrons. By analysing spin-spin correlation
functions for hexagonal flakes with armchair edges and triangular flakes with zigzag edges, we show
that edge type, geometry and charge nonuniformity affect the steepness and the crossover ρvalue of
the phase transition. For triangular structures, the metal-insulator transition is accompanied with
a smoother edge polarization transition.
Keywords: artificial graphene, graphene quantum dots, quantum simulators, variational Monte Carlo, diffu-
sion Monte Carlo
In recent years, technological advances in photonic and
condensed matter based artificial superlattices gives us
opportunities to develop practical quantum simulators
[1–14]. These quantum simulators allow us to replicate
complex systems that are hard to fabricate and pro-
vide us with a playground to verify theoretical predic-
tions. In this respect, artificial graphene (AG) nanos-
tructures, designed by imitating the 2D honeycomb pat-
tern of graphene, have been proven to be good candi-
dates for being reliable and controllable sources for both
fabrication and investigation of many physical phenom-
ena related to Dirac fermions [13–19]. In particular, AG
nanostructures can be formed using semiconductor ma-
terials. While earlier reports on nanopatterned artifi-
cial graphene on GaAs quantum well (QW) structures
found no evidence of massless Dirac fermions (MDFs),
presumably because of the relatively large lattice periods
[20–22], in recent experimental works using modulation-
doped AlGaAs/GaAs quantum wells [13, 14] shrinking
down of the lattice constant of the honeycomb array to
approximately 50 nm allowed the observation of the pre-
dicted graphene-like behavior [21, 23–25].
Observation of Dirac fermions in AG also opens up a
fresh way of studying graphene quantum dots [26] where
geometry, size and edge type is expected to give rise
to several physical properties such as bandgap opening
[27], edge magnetization [28–30] and optical control [31].
However, the fabrication and reliability issues such as
edge reconstruction or presence of impurities, make it
harder to observe interesting phenomena predicted to
occur in nanostructured graphene. Semiconductor AG
nanostructures, on the other hand, offer several advan-
tages such as tunability of system parameters including
lattice constant, site radius and potential depth, which,
in turn, allow to control electron-electron interactions
and tunneling strength between sites, in particular Hub-
bard parameter U/t. In experimental structures with
lattice constant a= 50 nm [14], U/t can be as high as
350 (as we will argue below) i.e. two orders of magni-
tude larger than the critical value for antiferromagnetic
Mott transition predicted by calculations based on Hub-
bard model for honeycomb lattice [32–35]. Moreover, un-
like in real graphene, long range electron repulsion does
not cancel the attraction of the artificial confining po-
tential in AG even near charge neutrality. For such large
and long-ranged electron interactions, a non-perturbative
many-body approach is desirable for careful treatment of
correlation effects.
Earlier theoretical work on electron interaction effects
in semiconductor AG nanostructures based on density
functional theory (DFT) showed that Dirac bands were
stable against interactions [23, 24] which was also con-
firmed using path integral Monte Carlo calculations [25].
However, recent calculations using Hartree-Fock and ex-
act diagonalization approaches for a triangular zigzag ge-
ometry with a= 12.5−15 nm show that a transition from
antiferromagnetic (AF) insulator to metallic phases oc-
curs, pointing to the importance of electron interactions
[36].
In this work, we use continuum variational Monte
Carlo (VMC) and diffusion Monte Carlo (DMC) methods
for non-perturbative and accurate treatment of many-
body correlations within the fixed-node approximation
to study GaAs based AG nanostructures. First, we con-
sider an hexagonal armchair geometry which serves as a
bridge between the finite-size samples and bulk graphene
[27], with lattice constant a= 50 nm following recent
experimental work [14]. We show that, a transition from
AF to metallic phase occurs, but is affected by a nonuni-
form charge distribution in the sample due to finite size
effects. This charge nonuniformity which is not present in
real graphene quantum dots, causes the phase transition
to be steeper and to occur at a smaller value of ρ. We
also investigate AG quantum dots with triangular zigzag
geometry and show that edge magnetization survives the
phase transition, in agreement with previous theoretical
arXiv:2210.14696v3 [cond-mat.mes-hall] 1 Oct 2023