Transition-Metal Oxide (111) bilayers∗

Satoshi Okamoto1, † and Di Xiao2

1Materials Science and Technology Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, USA2Department of Physics, Carnegie Mellon University, Pittsburgh, Pennsylvania 15213, USA

Correlated electron systems on a honeycomb lattice have emerged as a fertile playground to ex-plore exotic electronic phenomena. Theoretical and experimental work has appeared to realize novelbehavior, including quantum Hall effects and valleytronics, mainly focusing on van der Waals com-pounds, such as graphene, chalcogenides, and halides. In this article, we review our theoreticalstudy on perovskite transition-metal oxides (TMOs) as an alternative system to realize such exoticphenomena. We demonstrate that novel quantum Hall effects and related phenomena associatedwith the honeycomb structure could be artificially designed by such TMOs by growing their het-erostructures along the [111] crystallographic axis. One of the important predictions is that suchTMO heterostructures could support two-dimensional topological insulating states. The strongcorrelation effects inherent to TM d electrons further enrich the behavior.

I. INTRODUCTION

Two dimensional electron systems on a honeycomblattice have been providing fascinating phenomena. Aunique aspect of the honeycomb lattice is the appear-ance of Dirac points in the band structure, and its topo-logical property is controlled by the gap opening at theDirac points. Haldane first realized that introducingcomplex electron hopping amplitudes on a honeycomblattice could realize the quantum Hall effect (QHE) inthe absence of Landau levels [1]. This originates fromthe nontrivial band topology induced by the time re-versal symmetry (TRS) breaking, which opens the gapat the Dirac point. Later, Kane and Mele realized thatsuch nontrivial band topology could be induced withoutbreaking the TRS when the relativistic spin orbit cou-pling (SOC) exists [2], opening the new field of topolog-ical insulators (TIs).

Graphene, a monolayer of graphite, offers an ideal sys-tem to support such novel phenomena [3, 4]. While theSOC in graphene turned out to be too small to realizeTI states at accessible temperatures [5], early theoreticalwork has predicted a novel integer QHE [6], which hasbeen experimentally confirmed [3, 4]. More recently, afractional QHE was also realized experimentally [7]. Fur-ther interesting phenomena may be realized if the correla-tion effects are strong. It has been argued that graphenehas a potential to realize novel “chiral” superconductiv-

∗ Copyright notice: This manuscript has been authored by UT-Battelle, LLC under Contract No. DE-AC05-00OR22725 with theU.S. Department of Energy. The United States Government re-tains and the publisher, by accepting the article for publication,acknowledges that the United States Government retains a non-exclusive, paid-up, irrevocable, world-wide license to publish or re-produce the published form of this manuscript, or allow others todo so, for United States Government purposes. The Department ofEnergy will provide public access to these results of federally spon-sored research in accordance with the DOE Public Access Plan(http://energy.gov/downloads/doe-public-access-plan)†okapon@ornl.gov

ity by carrier doping [8–11]. In addition to graphene,other van der Waals compounds have also been attract-ing attention [12]. These van der Waals compounds couldcontain transition-metal elements and, therefore, mightbe suitable for exploring phenomena associated with cor-relation effects, such as magnetism.

In light of correlation effects, transition-metal ox-ides (TMOs) have a long history, covering magnetism,high-critical temperature (Tc) superconductivity, colos-sal magnetoresistance effects, and correlation-inducedmetal-insulator transitions.[13] Iridium-based oxides areone of the central focuses of recent studies because theenergy scales of the Coulomb repulsive interaction andthe SOC are compatible [14, 15]. In their seminal work,Shitade and coworkers have performed density functionaltheory (DFT) calculations and proposed that Na2IrO3,having a honeycomb lattice formed by Ir atoms, wouldbecome a two-dimensional TI [16]. On the other hand,Chaloupka and coworkers took an alternative approachfrom a Mott insulating side [17] and proposed that thelow-energy behavior of Na2IrO3 is governed by the so-called Kitaev-Heisenberg model,[18] which is a candidatefor realizing Z2 quantum spin liquid (SL) states. How-ever, later experimental measurements confirmed a mag-netic long-range order in Na2IrO3 [19–21]. To accountfor the experimental results, refined theoretical modelswere developed.[22–24]

While Na2IrO3 turned out to be a trivial insulator withcomplex magnetic ordering, TMOs in general have greatpotential to serve as a major playground to explore novelphenomena originating from the strong SOC and corre-lation effects. Here, we would like to focus on artificialheterostructures of TMOs rather than bulk compounds.This is motivated by the recent development in thin-film growth techniques of TMOs. A variety of TMOheterostructures with atomic precision have been syn-thesized and analyzed [25]. TMO heterostructures havegreat tunability over fundamental physical parameters,including the local Coulomb repulsion, SOC, and carrierconcentration. However, the effect of correlations to pos-sible novel phenomena near Mott insulating states witha strong SOC remains to be explored.

arX

iv:1

705.

0568

3v2

[co

nd-m

at.m

trl-

sci]

26

Jan

2018

2

FIG. 1: (Color online) Crystal structure. Left: cubic per-ovskite ABO3. When a bilayer of ABO3 is grown along thecubic [111] direction, it forms a buckled honeycomb latticewith two B sites, B1 on the top layer and B2 on the bottomlayer (right). B1 and B2 could become inequivalent whensubstrate and capping materials are different or when a gatevoltage is applied along the [111] direction.

In this article, we review our theoretical work on thenovel electronic property of TMO heterostructures. Incontrast to the standard heterostructures grown alongthe [001] crystallographic axis, we consider those grownalong an unconventional axis, i.e., [111] direction. Thisallows us to modify the underlying lattice geometry fromtetragonal to trigonal. By using multiple theoreticaltechniques, we demonstrate the correlated graphene-likebehavior arising from such TMO heterostructures.

II. MAIN IDEA

Throughout this work, we consider perovskite TMOs,which have the chemical formula ABO3. A is normallya divalent alkaline earth element or a trivalent rare earthelement, B is a TM element, and O is oxygen. As shownin Fig. 1, TM B sites form a cubic lattice in the idealbulk perovskite. A crucial observation is that TM B sitesform a triangular lattice in a (111) plane and, then, twoneighboring (111) planes of B sites form a “buckled” hon-eycomb lattice [26]. Thus, one could anticipate “Diracpoints” in their dispersion relations, just like graphene.In reality, it would be extremely difficult to create free-standing (111) bilayers. As in conventional (001) het-eostructures, (111) bilayers must be grown on a substrateand capped properly. To maintain the inversion symme-try, the same material must be used for the substrateand the capping layer. The inversion symmetry can beintentionally broken by using different materials for thesubstrate and the capping layer or by applying a gatevoltage along the [111] direction.

Can we align the Fermi level and Dirac points? Even ifthis is achieved, is the SOC strong enough to open a gapto induce a TI state? These are fundamental issues torealize TIs and could be overcome by properly choosingA and B elements. For the strong SOC, heavy B ele-ments are certainly favored, such as 4d or 5d TMs rather

2nd order SOC

eg

t2g

jeff = 1/2

jeff = 3/2

10Dqa1g

eg’

λ λ & ∆∆

d

FIG. 2: Single-particle energy level for d electrons. d levelis degenerate in the spherical symmetric environment. Underthe cubic environment, this splits into the triply degeneratet2g states and the doubly degenerate eg states with the levelseparation called 10Dq. With the SOC λ is turned on, the t2glevel splits into the doubly degenerate jeff = 1/2 states andthe four-fold degenerate jeff = 3/2 states, but the eg leveldoes not split while the t2g-eg separation is increased by thesecond-order SOC. The trigonal crystalline field ∆ alone doesnot affect the eg degeneracy either, while t2g level splits intothe doubly degenerate a1g states and the doubly degenerate e′gstates. When both ∆ and λ are turned on, all the degeneracyis lifted except for the Kramers degeneracy.

(a) t2g systems (b) eg systems

σt

δ ′t

δσ ′+ tt 41

43

( )δσ ′− tt43

δσ ′+ tt 43

41

x2−y2

3z2−r2

( )δσ ′−− tt43

0δt

πt

xy

yz

zxπt

πt

δt

δt

πt

00

0

x yz Y

X

FIG. 3: (Color online) Nearest-neighbor transfer integrals fort2g systems (a) and eg systems (b). Different prefactors orig-inate from the Slater-Koster parametrization considering thehybridization between TM d orbitals and oxygen p orbitals.

than 3d TMs. Integer filling required for TIs could berealized relatively easily. If necessary, doping concentra-tion could be tuned by partially replacing A site ions byother ions with different valence. Also, gating could con-trol doping concentration externally. By replacing A siteions between the two (111) triangular lattices of B ionsby other ions with the same valence but different ionicradius, the strain value (distance between B sites andcrystal field) could be controlled. Strain could also bechanged by replacing substrate and capping materials.

One of the benefits of such (111) bilayers is the reducedsymmetry of the crystalline field from octahedral (Oh) totrigonal (C3v). The octahedral crystalline field splits theTM d orbitals into three-fold degenerate t2g(yz, zx, xy)levels and two-fold degenerate eg(3z

2− r2, x2−y2) levelswell separated by the so-called 10Dq on the order of 3eV. The trigonal crystalline field ∆ further splits the t2gmanifold into non-degenerate a1g and two-fold degener-

3

ate e′g, and the SOC λ splits the t2g manifold into two-fold degenerate jeff = 1/2 states and four-fold degen-erate jeff = 3/2 states including Kramer’s degeneracy.On the other hand, each effect alone does not break thedegeneracy in the eg manifold while the SOC shifts theentire eg level upward by ∝ λ2/10Dq, i.e., second-orderin the SOC strength. When both the trigonal field andthe SOC are present, the eg degeneracy is lifted, result-ing in the full degeneracy lifting except for the Kramer’sdegeneracy [27, 28] as shown in Fig. 2. Note that thelevel splitting within the eg manifold is proportional to

λ ∝ λ2∆/(10Dq)2 and vanishes when ∆ vanishes. Thus,TI states could be realized at multiple integer fillings in(111) TMO bilayers.

III. THEORETICAL RESULTS

This section consists of four subsections. In the firsttwo subsections, we consider non-interacting or weaklyinteracting systems to explore possible materials realiza-tion of TIs. In the third subsection, we examine thecorrelation effects by means of the dynamical mean fieldtheory (DMFT).[29] In the fourth subsection, we use astrong coupling approach from a Mott insulating regime.By these multiple approaches, we uncover novel proper-ties of (111) TMO bilayers in a wide parameter space.

A. Tight-binding modeling

With the structural consideration in the previous sec-tion, we first construct tight-binding models for t2g sys-tems and eg systems. By noticing the relation betweenthe cubic coordinate and hexagonal coordinate, it isstraightforward to fix the transfer matrices, which de-pend on the direction and the pair of orbitals. Typi-cal nearest-neighbor transfer integrals are shown in Fig.3. These are d-d hopping through oxygen p orbitals inbetween. Different prefactors originate from the Slater-Koster parametrization[30] considering the hybridizationbetween TM d orbitals and oxygen p orbitals.

Thus, for t2g electron systems, we may have

Ht2g = −∑〈~r~r′〉σoo′∈t2g

{too

′

~r~r′d†~roσd~r′o′σ +H.c.

}+HSOC

t2g +Htrit2g

(1)and, for eg electron systems,

Heg = −∑〈~r~r′〉σoo′∈eg

{too

′

~r~r′d†~roσd~r′o′σ +H.c.

}+HSOC

eg . (2)

The SOC and the trigonal crystal field for t2g systemsare given by

HSOCt2g = λ

∑~r

~l~r · ~s~r, (3)

and

Htrit2g = ∆

∑~rσo6=o′

d†~roσd~ro′σ, (4)

respectively. Here, ~l is the l = 2 angular momentumoperator projected on the t2g multiplet, and ~s is the spinoperator. The second-order SOC for eg electron systemsunder the C3v symmetry is given in a compact form by

HSOCeg = − λ

2

∑~rσ

oo′∈eg

d†~roστyoo′σ

zσσd~ro′σ. (5)

Here, τy is the Pauli matrix acting in the orbital space,and the uniform shift proportional to λ2/10Dq is ab-sorbed in the chemical potential. The spin quantizationaxis is taken along the [111] crystallographic axis or per-pendicular to the plane. Thus, in contrast to the t2gmodel, spin along this quantization axis is conserved andserves as a good quantum number as in the Kane-Melemodel [2].

Using these TB models, we first examine their topolog-ical properties to determine the possible candidate ma-terials for 2D TIs. Because of the multi-orbital natureof TMOs, our model gives rise to a very rich behav-ior of the topological band structure in the parameterspace. For t2g systems, there is a competition betweenthe SOC and the trigonal field, and depending on the rel-ative strength between the two, the system falls into twodifferent phases. We found that the highest two bandshave robust topological properties irrespective to theseparameters. Figures 4 (a) shows the dispersion relationswith ∆/t = 0.5 and λ/t = 1.5 (red) and with ∆/t = 1.5and λ/t = 0 (green). We determined the Z2 topologi-cal invariant by evaluating directly from the bulk bandstructure [31, 32] and by counting the number of edgestates [Fig. 4 (b)]. By inspection, we find that t22g, t

42g

and t52g systems are possible candidates for TIs, while

t12g would become a topological semimetal. For eg sys-

tems, we found that e1g, e2g and e3g systems are possible

candidates for TIs [see Fig. 4 (c) and (d) ].

It is worth mentioning that the lowest or highest bandin the eg model is fairly flat. This originates from the factthat x2− y2 orbitals have very small transfer matrix ele-ments along the z direction tδ′ ; in Fig. 4 (c), tδ′ is takento be zero, and non-zero dispersion comes from the SOC.It turned out that, when the TRS is broken by a magneticfield or spontaneous uniform magnetic ordering, i.e. FM,these bands have nonzero Chern number. Thus, in addi-tion to TIs, the eg model could realize quantum anoma-lous Hall effects at e1g or e3g filling. Furthermore, whenthese bands are partially filled and long-ranged Coulombinteractions are present, fractional QHE could be realizedas pointed out in Refs. [33–37].

4

-3

-2

-1

0

1

2

3

4

1.8

2.0

0.0 0.2 0.4 0.6 0.8 1.0

-3

-2

-1

0

1

2

3

4

-0.4 -0.2 0.0 0.2 0.4

-1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

-1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

Z2=1

Z2=0

Z2=1

Z2=1

Z2=0

Z2=1

Z2=1

Z2=0

Z2=0

Z2=1

ΓΓ K M

ΓΓ K Mkzigzag [ ]a~32π

(a) t2g systems

(c) eg systems

Ene

rgy/

t πE

nerg

y/t σ

(b) t2g zigzag slab

(d) eg zigzag slab

ΓM

K

kzigzag [ ]a~32π

FIG. 4: (Color online) Tight-binding results of the (111) bilayers. (a,b) t2g electron systems and (c,d) eg electron systems.(a) and (c) are dispersions for bulk (111) bilayers, while (b) and (d) are dispersions for finite-thickness zigzag slabs. Smallhopping parameters tδ and tδ′ are neglected for simplicity. Parameters are taken as ∆/tπ = 0.5 with λ/tπ = 1.5 (thick lines)

and ∆/tπ = 1.5 with λ/tπ = 0 (thin lines) in (a), and λ/tσ = 0.2 (thick lines) and λ/tσ = 0 (thin lines) in (b). Finite SOCgives rise to a nontrivial band topology as shown by band-dependent Z2 indexes and gapless edge modes supporting the spincurrent (thick lines in the right figures). In (b), there are four edge channels between the third and the fourth bands as shownas dash lines. These edge channels do not support the spin current, consistent with Z2 = 0 when the Fermi level is inside thisgap. a is the nearest-neighbor bond length projected on the (111) plane, i.e., a =

√2/3a with a being the lattice constant

of cubic perovskite. The figures ares adopted and modified from Ref. [26]. The inset of (c) is the first Brillouin zone of thehexagonal lattice.

B. Density functional theory analyses

Based on the TB results, we now look for candidatematerials. While t2,42g and e1,3g are also possible candidate

systems, here we focus on t52g and e2g systems because

topological properties are robust. For example, t2,42g sys-tems would tend to become topological semimetals due tothe band overlap, and e1,3g systems could become unsta-

ble against the Jahn-Teller effect or 3x2−r2/3y2−r2-typeorbital ordering. For oxide thin-film growth, SrTiO3 andLaAlO3 are often used as insulating substrates. Basedon this, we consider a TM B ion to have the formal va-lence +4 in the form of SrBO3 or +3 in LaBO3. Thislimitation greatly reduces the number of materials com-bination.

For t52g systems, we come up with LaRu3+O3,

LaOs3+O3, SrRh4+O3 and SrIr4+O3, and for e2g elec-

tron systems LaAg3+O3 and LaAu3+O3. While thesematerials have not been synthesized with the perovskitestructure except for SrIrO3 [38] (for example, accordingto Ref. [39], LaAuO3 has CaF2 structure rather than theperovskite), it is worth examining all these systems toprove our theoretical idea. We hope that advanced crys-tal synthesis techniques will allow fabrication of our pre-dicted structure or that other systems with similar crys-tal and electronic structures will be discovered in the nearfuture.

We examined these (111) bilayers using DFT meth-ods. DFT calculations were performed using the pro-jector augmented wave method[40] with the general-ized gradient approximation in the parametrization ofPerdew, Burke and Enzerhof[41] for exchange correla-tion as implemented in the Vienna Ab Initio Simulation

5

(d) LaAuO3

(b) SrIrO3

Gap: 20 meV

0.0 0.2 0.4 0.6 0.8 1.0

(e) SrIrO3 zigzag slab

-1

0

-2

-0.4 -0.2 0.0 0.2 0.4

(f) LaAuO3 zigzag slab

0

-1

1

0

-1

Gap: 146 meV

ΓΓ K M

(a) LaOsO3

(c) LaAgO3

-2

0

-1

1

0

-1

Ene

rgy

(eV

)E

nerg

y (e

V)

Gap: 55 meV

Gap: 39 meV

1

0

-1

kzigzag [ ]a~32π

kzigzag [ ]a~32π

ΓΓ K MΓΓ K M

ΓΓ K M

FIG. 5: (Color online) Density functional theory results of the dispersion relations of the (111) bilayer of TMOs. (a) LaOsO3,(b) SrIrO3, (c) LaAgO3, and (d) LaAuO3 are for bulk (111) bilayers, and (e) SrIrO3 and (f) LaAuO3 are for finite-thicknesszigzag slabs. Bilayers shown in (a), (c), and (d) are grown between LaAlO3, while that in (b) is grown between SrTiO3. Thesesystems are topological insulators with the gap amplitude indicated. The gappless edge modes at zero energy in (e) and (f)confirm the non-trivial band topology. The Fermi level is taken to be 0 of the vertical axis. Figures (a-d) are adopted andmodified from Ref. [26]. Figures (e) and (f) are taken from Ref. [44].

Package[42]. The detail of our DFT calculations can befound in Ref. [26].

Figure 5 summarizes our DFT results. LaRuO3

and SrRhO3 (not shown) turned out to be topologi-cal semimetals, rather than TIs, because of the overlapbetween the conduction bands and the valence bands.Other candidate systems are found to become TIs withthe Fermi level inside the nontrivial gap. To demonstratetheir nontrivial band topology, we have derived Wannierfunctions [43] to construct finite-thickness zigzag slabsof SrIrO3 (111) bilayer and LaAuO3 (111) bilayer. Asshown in Fig. 5 (e) and (f), these (111) bilayers havegapless edge modes crossing the Fermi level.

C. Dynamical mean field theory analyses

In the previous subsection, we did not consider cor-relation effects which could modify the band structureresults. For example, correlation effects could inducesome symmetry breaking such as ferromagnetic or an-tiferromagnetic (AFM) ordering. Furthermore, a Motttransition (correlation-induced metal to insulator tran-sition) could take place. Here, we examine these twopossibilities using DFT supplemented DMFT [29], i.e.,DFT+DMFT.[44] We focus on SrIrO3 (111) bilayer and

LaAuO3 (111) bilayer because the former is alreadyknown to exist and the latter has the largest nontrivialgap.

The DMFT calculations were carried out using asingle-particle Hamiltonian Hsingle and a many-bodypart HU . Hsingle is parametrized in the Wannierbasis.[43] We express HU in terms of real orbitals, eithert2g or eg, as[45]

HU =U∑o

d†o↑do↑d†o↓do↓ + U ′

∑o 6=o′

d†o↑do↑d†o′↓do′↓

+(U ′ − J)∑o>o′

(d†o↑do↑d

†o′↑do′↑ + d†o↓do↓d

†o′↓do′↓

)+J

∑o 6=o′

(d†o↑do′↑d

†o′↓do↓ + d†o↑do′↑d

†o↓do′↓

), (6)

where o and o′ stand for (yz, zx, xy) for SrIrO3 and(3z2−r2, x2−y2) for LaAuO3. U and U ′ are the intraor-bital Coulomb interaction and the interorbital Coulombinteraction, respectively, and J represents the interor-bital exchange interaction (fourth term) and the interor-bital pair hopping (fifth term). For orbitals with the t2gor eg symmetry, U ′ = U − 2J .

Impurity problems for DMFT are solved by using a fi-nite temperature exact diagonalization technique [46–48].

6

One of the advantages of the exact diagonalization im-purity solver is the direct access to the spectral functionwithout employing a maximum-entropy analytic contin-uation. The detail of our DFT+DMFT calculations canbe found in Ref. [44].

Are SrIrO3 and LaAuO3 in AF trivial insulating phasesor TI phases? If they are in AF trivial phases, can weturn them into TI phases by suppressing magnetic or-dering? To answer these questions, we estimate realisticCoulomb interactions by using the constrained randomphase approximation (cRPA)[49, 50]. For SrIrO3 bilayer,we took the Slater parameters F0,2,4 for Sr2IrO4 fromRef. [51] and deduced U and J as 2.232 eV and 0.202eV, respectively, for the {xy, yz, zx} basis. For LaAuO3

bilayer, we directly computed these parameters for the{3z2 − r2, x2 − y2} basis, and the resultant U and J areU = 1.80 eV and J = 0.225 eV, respectively.

Using these parameters, we found that SrIrO3 (111) bi-layer is unstable against AFM ordering. When the mag-netic ordering is suppressed, it recovers a TI state, butthe gap amplitude is undetectably small because the nearFermi energy band dominated by jeff = 1/2 state has ex-tremely small quasiparticle weight Z ∼ 0.07. The othertwo bands dominated by jeff = 3/2 states show onlymoderately renormalized quasiparticle weight Z ∼ 0.88and 0.79. This behavior can be seen clearly from angle-resolved spectral function in Fig. 6 (a), where nearlyflat bands (indicating small Z) appear at the Fermi levelwhile far below the Fermi level spectral intensity nicelyfollows DFT dispersions. Such a behavior reminds usof an orbital-selective Mott transition [52]. When themagnetic ordering is allowed, a fairly large gap opens[Fig. 6 (b)]. On the other hand, the topological natureof LaAuO3 (111) bilayer is found to be robust againstcorrelation effects; magnetic ordering is unstable withthe realistic parameters and the quasiparticle weight israther large Z ∼ 0.81 for two Kramers states. As shownin Fig. 6 (c), the peak position of the spectral functionand band dispersion differ only slightly.

Noticing J/U ≈ 0.1 in both SrIrO3 and LaAuO3,we also studied their systematic behavior as a func-tion of U with the fixed ratio J/U = 0.1. It is foundthat SrIrO3 (111) bilayer is indeed in the vicinity of anorbital-selective Mott transition with the critical interac-tion Uc ∼ 2.5 eV (when J/U = 0.1). Since this transitionaccompanies the change in the band topology from non-rivial to trivial, this might be called an orbital-selectivetopological Mott transition. This orbital-selective Motttransition does not only come from the different bandwidth of three states dominated by Jeff = 1/2 statesor Jeff = 3/2 states but also from different fillings; oneband dominated by Jeff = 1/2 states are nearly halffilled while the other bands dominated by Jeff = 3/2states are nearly fully filled. The critical interactionfor the onset of AFM ordering is found to be rathersmall, UAFM,c ∼ 0.5 eV (when J/U = 0.1), reflectingthe small effective band width of jeff = 1/2-dominatedband, W ≈ 0.5 eV. On the other hand, LaAuO3 shows

-3.0

-2.5

-2.0

-1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

ω (

eV)

-3.0

-2.5

-2.0

-1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

ω (

eV)

-1.2

-0.8

-0.4

0.0

0.4

0.8

1.2

1.6

ω (

eV)

(c) LaAuO3 PM

(b) SrIrO3 AFM

Gap: 110meV

Gap: 500meV

ΓΓ K M

(a) SrIrO3 PM

ΓΓ K M

ΓΓ K M

FIG. 6: Color online) DFT+DMFT result of the ARPESspectra of bulk SrIrO3 (111) bilayer in (a) PM and (b) AFMphases and bulk LaAuO3 (111) bilayer (c). Thin dashed linesare DFT (Wannier) dispersion curves. Four lowest dispersionsdominated by jeff = 3/2 states are downshifted by 0.5 eV in(a). Parameters used are U = 2.232 eV and J = 0.202 eV for(a) and (b), and U = 1.80 eV and J = 0.225 eV for (c). Notethat these bulk spectra are very similar to the ones presentedin Ref. [44] with U = 2.0 eV and J = 0.2 eV.

a standard Mott transition where two bands undergo amass divergence simultaneously at Uc ∼ 4.3 eV. AFMordering is also stabilized but it requires larger interac-tion (UAFM,c ∼ 2.1 eV) than SrIrO3. It is worth notic-ing that, for LaAuO3, there is a small window 2.1 eV<∼ U <∼ 2.15 eV for an AFM TI where the gapless edgestates and bulk magnetic ordering coexist. This comes

7

0.0 0.2 0.4 0.6 0.8 1.0-3.0

-2.5

-2.0

-1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

-0.4 -0.2 0.0 0.2 0.4

-0.8

-0.4

0.0

0.4

0.8

1.2

1.6

(a) SrIrO3

(b) LaAuO3

No gapless edge mode

kzigzag [ ]a~32π

ω(e

V)

kzigzag [ ]a~32π

ω(e

V)

FIG. 7: (Color online) DFT+DMFT results of the ARPESspectra of the (111) bilayer of TMOs on a finite-thicknesszigzag slab. (a) SrIrO3 with the SrTiO3 substrate, (b)LaAuO3 with the LaAlO3 substrate. Parameter values arethe same as in Fig. 6. Because of the AFM ordering, SrIrO3

(111) bilayer is in a trivial phase and there is no gapless mode,while LaAuO3 (111) bilayer is in a non-trivial phase with gap-less modes. Figures are taken from Ref. [44].

from the fact that the spin component perpendicular tothe [111] plane is conserved in our eg electron model,and therefore the LaAuO3 bilayer consists of two copiesof Chern insulators with up and down electrons havingthe opposite Chern number, leading to zero Hall responsebut quantized spin Hall response.

The trivial nature of SrIrO3 and non-trivial natureof LaAuO3 can be seen in the edge spectrum of finite-thickness slab. As shown in Fig. 7 (a), SrIrO3 does nothave gapless edge modes. On the other hand, LaAuO3

has gapless edge modes [Fig. 7 (b)].

D. Strong coupling approach

In the previous subsection, we studied SrIrO3 (111)bilayer and LaAuO3 (111) bilayer using DFT+DMFTtechnique. We found that the SrIrO3 (111) bilayer is inthe vicinity of an orbital-selective Mott transition and

unstable against AFM ordering. Once such magneticordering or correlation is established, a weak-couplingmagnetic insulating regime and a strong-coupling mag-netic insulating regime are adiabatically connected, thusthe two approaches, weak coupling and strong coupling,would be equally valid. Moreover, a strong-coupling ap-proach from a Mott insulating side could provide phys-ically more transparent picture. Here, we focus on theSrIrO3 (111) bilayer using strong coupling approach anddiscuss the origin of magnetic ordering and potentialnovel phenomena.[53]

We start from constructing the low-energy effectiveHamiltonian from a multiband Hubbard model with t2gorbitals. According to Wannier parameterization, typi-cal band parameters are tπ ∼ 0.31 eV and tδ ∼ 0.02 eVfor the NN hoppings, and ∆ ∼ 0.006 eV for the trigonalcrystalline field. Thus, for simplicity, we consider onlytπ in the following discussion. The effective Hamiltonianis derived from the second-order perturbation processeswith respect to the transfer terms and by projecting thesuperexchange-type interactions onto the isospin statesfor Jzeff = ±1/2:

|Jzeff = σ〉 =1√3{i|yz,−σ〉 − σ|xz,−σ〉+ iσ|xy, σ〉}. (7)

The resulting effective Hamiltonian reads

HKH =∑〈~r~r′〉

(JKS

γ~r S

γ~r′ + JH ~S~r · ~S~r′

), (8)

where, ~S is an isospin operator for Jeff = 1/2, not areal spin. JK and JH are given by JK = 4

9 (r1 − r2)

and JH = 827 (3r1 + r2 + 2r3), respectively, where r1 =

t2π/(U − 3J), r2 = t2π/(U − J), r3 = t2π/(U + 2J) withU being the intraorbital Coulomb interaction and J theinterorbital exchange interaction [see Eq. (6)]. γ dependson the direction of ~r-~r′ pair; i.e., γ = x (y, z) when the~r-~r′ bond is along the cubic x (y, z) axis. Thus, thiseffective Hamiltonian has the same form as the celebratedKitaev-Heisenberg model, which is originally proposedfor Na2IrO3. Here, however, both Kitaev and Heisenbergterms have a positive sign, i.e., AFM [54].

First we solved this model on a finite-size cluster us-ing a Lanczos exact-diagonalization method to obtain aground-state phase diagram. Following Ref. [17], weparametrize JK and JH as JK = 2α and JH = 1 − αand vary α from 0 (Heisenberg limit) to 1 (Kitaevlimit). In contrast to a FM-Kitaev/AFM-Heisenbergmodel, a model proposed to describe Na2IrO3 in Ref.[17], there appear two phases in AFM-Kitaev/AFM-Heisenbeg model HKH : an AFM Neel ordered phase atα < αc and a Kitaev spin-liquid phase at α > αc withαc ∼ 0.96.

For a general Kitaev-Heisenberg model, where JK andJH could be either FM or AFM, a more complex phasediagram is obtained, including a zigzag AFM phase anda FM phase [55, 56].

8

Now, a question arises, where would be the realisticrange for SrIrO3 (111) bilayer in this phase diagram?Note that U > 3J is physical because, otherwise, thecharge excitation energy from t52gt

52g to t62gt

42g becomes

negative. From the analytic expressions of JK and JH ,the JK/JH ratio is an increasing function of J ; α variesfrom 0 to 1/5 when J is varied from 0 to U/3. So, thephysical parameter range would be α < 1/5, where a NeelAFM ordering is stabilized. This result is consistent withour DFT+DMFT predictions.

While the novel Kitaev spin liquid phase exists in oureffective model, realizing it using SrIrO3 (111) bilayerseems to be very difficult, even though it may not beentirely impossible. However, carrier doping may be fea-sible. This might induce novel phenomena as carrier dop-ing into a Mott insulator on a square lattice induces d-wave superconductivity [57, 58].

In addition to the effective spin model Eq. (8), weconsider hopping matrices projected into neighboring orJzeff = ±1/2 states. In this representation, the hop-ping matrices are diagonal in the isospin index σ and

given by Ht = −t∑〈~r~r′〉σ(d†~rσd~r′σ+H.c.) with the renor-

malized hopping amplitude t = 23 tπ. d~rσ excludes the

double occupancy because of the strong Coulomb repul-sive interaction. Thus, the effective model for the dopedKitaev-Heisenberg model is given by Heff = Ht +HKH .We analyze this model using a SU(2) slave boson meanfield (SBMF) method developed for high-Tc cuprates [58].Because of the Kitaev interaction, there appears a largernumber of mean field order parameters than for the t-Jmodel. To make the problem tractable, we focus hereon several ansatze which respect the sixfold rotationalsymmetry of the underlying lattice. The detail of theseansatze and the mean field calculations can be found inRef. [53].

The mean field phase diagram as a function of α anddoping concentration δ is shown in Fig. 8. The shadedarea is the realistic parameter range. Since the Heisen-berg interaction dominates in the undoped limit, the ef-fect of carrier doping is also similar to the t-J model ona honeycomb lattice, and a large area is covered by a su-perconducting (SC) state with the d+ id (dx2−y2 + idxy)pairing [9]. Similar to this result, carrier doping to asquare-lattice iridium oxide Sr2IrO4 is theoretically sug-gested to induce the dx2−y2 SC.[59–61] In addition, wefound a s-wave SC state at a large δ regime and a p-waveSC state in the vicinity of the Kitaev spin liquid phase.The relative stability between the d+ id and s wave SCstates is determined by the competition between the pair-ing formation and the coherent motion of carriers. Thep-wave SC state is unique for Kitaev interactions. Thisbreaks the TRS and is adiabatically connected to the un-doped Kitaev spin liquid state. A similar p-wave SC stateis also found in a doped FM Kitaev model in Ref. [62],while Ref. [63] found another p-wave SC state whichdoes not break the TRS. These two p-wave SC statesare found to be separated by a first order transition bychanging δ [55]. Ref. [55] also studied a general Kitaev-

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.00.0

0.1

0.2

0.3

0.4

α

δ

SL

Néel AFM

d + i d

p SC

s

FIG. 8: Schematic phase diagrams for the doped AFM-Kitaev/AFM-Heisenberg model as a function of δ and α. Pa-rameters are taken as JK + JH = tπ. α is the measure ofthe relative strength between JK and JH as JK = 2α andJH = 1 − α. Phase boundaries at finite δ are the resultsof the SU(2) SBMF, while those at δ = 0 are results of theexact diagonalization. The shaded area is the realistic pa-rameter range for 0 < 3J < U . The figure is taken from Ref.[53].

Heisenberge model with finite δ and reported a complexphase diagram.

While the (111) bilayer of SrIrO3 may not be sufficientfor studying the novel physics associated with Kitaev in-teractions, carrier doping or replacing Ir with some otherelements may allow the access to rich behavior as dis-cussed in this subsection.

IV. RELATED WORK

After our first proposal [26], a number of related workshave appeared.[64] Here, we list some of the importantones.SrIrO3: Theoretical work based on the DFT calcu-

lation, similar to ours, has appeared in Ref. [65]. Theexperimental realization was quite challenging because ofthe large lattice mismatch between SrIrO3 and SrTiO3.In Ref. [66], Hirai and coworkers were successful to fab-ricate a bilayer of iridate by replacing a half of Sr byCa, i.e., Ca0.5Sr0.5IrO3, on SrTiO3. They found such aniridate (111) bilayer is strongly insulating and magnetic.While the detail of magnetic ordering is not clarified yet,the trivial insulator nature with magnetism is consistentwith our DFT+DMFT results.LaAuO3 and LaAgO3: Similar to our proposal,

LaAuO3 and LaAgO3 (111) bilayers were also theoreti-cally studied. In Ref. [67], Liang and Hu proposed grow-ing those TMOs between LaCrO3 to realize controllableChern insulators.LaNiO3: LaNiO3-based superlattices have been ac-

tively studied because of its potential to realize the un-conventional superconductivity similar to cuprates.[68](111) bilayers of nickelates were first studied using model

9

Hamiltonians to predict novel anomalous Hall insulatorsdue to the complex orbital ordering.[69, 70] This or-bital ordering accompanies a finite expectation value of〈τy〉. Thus, the SOC-like effect is dynamically generated.Later, realistic DFT calculations were performed.[71–73]Experimental efforts on such nickelate (111) bilayer havealso appeared.[74, 75]

La(Ni,Co)O3: Correlation induced novel phenomenahave been theoretically predicted, including odd-paritysuperconductivity in a (111) bilayer of LaNi7/8Co1/8O3

[76] and quantum anomalous Hall phases in a (111) bi-layer of LaCoO3.[77]

SrTiO3: SrTiO3/LaAlO3 heterostructures are one ofthe most intensively studied systems due to the high-mobility electron gases.[78] In Ref. [79], (111) bilayers ofSrTiO3 grown on LaAlO3 were theoretically studied.LaMnO3: When eg band is fully spin polarized, the

Fermi level for an e1g system could be just inside thenontrivial gap [see Fig. 4 (c)]. Since the spin quantumnumber along the [111] direction is conserved, a result-ing electronic phase could be a Chern insulator. Such astate was indeed theoretically reported in LaMnO3 (111)bilayers.[80, 81]

Other 4d and 5d perovskite TMOs: There have also ap-peared a number of theoretical studies on 4d and 5d per-ovskite TMOs. Among these TMOs, Refs. [82, 83]showedthat LaOsO3 (111) bilayer could become a Chern insula-tor with high Curie temperature. Ref. [83] also showeda similar Chern insulating state in LaRuO3 (111) bi-layer. Further, Ref. [84] showed that SrRuO3 (111) bi-layer could also show a Chern insulating state when elec-trons are doped. Ref. [85] considered BiIrO3 (111) bilay-ers sandwiched by BiAlO3. Since an Ir ion has the t62gconfiguration, this system has a trivial band topology.Nevertheless, novel spin-valley coupled phenomena[86]could result.

Double perovskite: When the top B sites and the bot-tom B sites of ABO3 (111) bilayer are different, it is a(111) monolayer of double perovskite. In Ref. [87], suchheterostructures were studied using Fe on the top layerand Os on the bottom layer to predict novel Chern insu-lating states.

Dice lattice: When a trilayer of ABO3 is grown alongthe (111) direction, B sites form a “dice” lattice. Ref.[88] discussed possible Chern insulators associated withnearly flat bands induced in such a dice lattice.

Pyrochlore oxides: In addition to perovskite oxides, py-rochlore oxides have been studied theoretically. Since the(111) plane is the natural cleavage plane, this class of ma-terials might be grown along the [111] direction more eas-ily. Novel topological phases have been predicted usingmodel Hamiltonians [89–92] and the first principle DFTcalculations.[93] Further, topological magnon states andunconventional superconductivity in pyrochloe iridatesare discussed in Ref. [94].

Corundum and other hexagonal TMOs: As mentionedin Ref. [26], corundum materials are possible candidatesfor realizing novel quantum states associated with honey-comb lattice because the [0001] plane of corundum M2O3

involves a honeycomb lattice formed by M atoms. Ref.[95] reported nontrivial band topology using DFT tech-niques with M = Au and Os grown between sapphireAl2O3.

Similar to the aforementioned Na2IrO3, SrRu2O6 has ahoneycomb lattice formed by Ru ions. However, becauseof the robust G-type AFM ordering with relatively highNeel temperature TN ∼ 565 K,[96–98] its topologicalproperty has not attracted attentions. A recent theoret-ical study showed that the band topology of SrRu2O6 isindeed trivial even when the AFM ordering is suppressedbut this could be turned to nontrivial by a slight modi-fication of lattice parameters [99]. While realizing non-trivial band topology in SrRu2O6 remains challenging,it was suggested that the topological transition could beinduced under accessible pressures by replacing Sr by Caand Ru by Os,[99] resulting in a strong TI.[100] Althoughthis is not a heterostructure, thin film-growth techniquesdeveloped for the TMO heterostructures could make im-portant contributions for growing such materials and in-troducing pressure or strain induced by a substrate.

V. CONCLUSION

In this article, we reviewed our theoretical work on(111) bilayers of perovskite TMOs as a fertile play-ground to explore novel electronic phenomena. Our workstarted from an extremely simple idea; honeycomb-likestructure could be realized using perovskite TMOs whengrown along the [111] direction, which is almost like anegg of Columbus. We found surprisingly rich behaviorfrom (111) bilayers of perovskite TMOs, including 2-dimensional TIs even for eg electron systems in whichthe SOC is quenched in bulk systems, orbital-selectiveMott transition, Kitaev physics, and doping-driven un-conventional superconductivity. Beside our original pro-posal, there has appeared much research on (111) TMOheterostructures. While an experimental realization oftruly novel or topological phenomena as theoretically pre-dicted has yet to appear, we hope that further develop-ments in the thin-film growth technique will eventuallyrealize some of the theoretical predictions and our workwill make important contributions to modern materialsscience.

Acknowledgments

The research by S.O. is supported by the U.S. De-partment of Energy, Office of Science, Basic Energy Sci-ences, Materials Sciences and Engineering Division. D.X.acknowledges support by the Air Force Office of Scien-tific Research under Grants No. FA9550-12-1-0479. Wewould like to thank W. Zhu, Y. Ran, N. Nagaosa, Y.Nomura, and R. Arita for collaborations and fruitful dis-cussions at every stage of this work.

10

[1] F. D. M. Haldane, Phys. Rev. Lett. 61, 2015 (1988).[2] C. L. Kane and E. J. Mele Phys. Rev. Lett. 95, 146802

(2005).[3] K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang,

M. I. Katsnelson, I. V. Grigorieva, S. V. Dubonos, andA. A. Firsov, Nature 438, 197 (2005).

[4] Y. Zhang, Y.-W. Tan, H. L. Stormer, and P. Kim, Na-ture 438, 201(2005).

[5] H. Min, J. E. Hill, N. A. Sinitsyn, B. R. Sahu, LeonardKleinman, and A. H. MacDonald, Phys. Rev. B 74,165310 (2006).

[6] Y. Zheng and T. Ando, Phys. Rev. B 65, 245420 (2002).[7] K. I. Bolotin, F. Ghahari, M. D. Shulman, H. L.

Stormer, P. Kim, Nature 462, 196 (2009).[8] G. Baskaran, Phys. Rev. B 65, 212505 (2002).[9] A. M. Black-Schaffer and S. Doniach, Phys. Rev. B 75,

134512 (2007).[10] S. Pathak, V. B. Shenoy, and G. Baskaran, Phys. Rev.

B 81, 085431 (2010).[11] R. Nandkishore, L. S. Levitov, and A. V. Chubukov,

Nat. Phys. 8, 158 (2012).[12] G. R Bhimanapati, Z. Lin, V. Meunier, Y. Jung, J. Cha,

S. Das, D. Xiao, Y. Son, M. S Strano, V. R Cooper, L.Liang, S. G. Louie, E. Ringe, W. Zhou, S. S. Kim, R.R. Naik, B. G. Sumpter, H. Terrones, F. Xia, Y. Wang,J. Zhu, D. Akinwande, N. Alem, J. A. Schuller, R. E.Schaak, M. Terrones, J. A Robinson, Acs Nano 9, 11509(2015).

[13] M. Imada, A. Fujimori, and Y. Tokura, Rev. Mod. Phys.70, 1039 (1998).

[14] B. J. Kim, H. Jin, S. J. Moon, J.-Y. Kim, B.-G. Park,C. S. Leem, J. Yu, T. W. Noh, C. Kim, S.-J. Oh, J.-H. Park, V. Durairaj, G. Cao, and E. Rotenberg, Phys.Rev. Lett. 101, 076402 (2008).

[15] D. Pesin and L. Balents, Nat. Phys. 6, 376 (2010).[16] A. Shitade, H. Katsura, J. Kunes, X.-L. Qi, S.-C. Zhang,

and N. Nagaosa, Phys. Rev. Lett. 102, 256403 (2009).[17] J. Chaloupka, G. Jackeli, and G. Khaliullin, Phys. Rev.

Lett. 105, 027204 (2010).[18] A. Kitaev, Ann. Phys. (N.Y.) 321, 2 (2006).[19] Y. Singh and P. Gegenwart, Phys. Rev. B 82, 064412

(2010).[20] X. Liu, T. Berlijn, W.-G. Yin, W. Ku, A. Tsvelik, Y.-J.

Kim, H. Gretarsson, Y. Singh, P. Gegenwart, and J. P.Hill, Phys. Rev. B 83, 220403 (2011).

[21] F. Ye, S. Chi, H. Cao, B. C. Chakoumakos, J. A.Fernandez-Baca, R. Custelcean, T. F. Qi, O.B. Korneta,and G. Cao, Phys. Rev. B 85, 180403 (2012).

[22] J. G. Rau, E. K.-H. Lee, and H.-Y. Kee, Phys. Rev.Lett. 112, 077204 (2014).

[23] Y. Yamaji, Y. Nomura, M. Kurita, R. Arita, and M.Imada, Phys. Rev. Lett. 113, 107201 (2014).

[24] Y. Sizyuk, C. Price, P. Wolfle, and N. B. Perkins, Phys.Rev. B 90, 155126 (2014).

[25] H. Y. Hwang, Y. Iwasa, M. Kawasaki, B. Keimer, N.Nagaosa, and Y. Tokura, Nat. Mater. 11, 103 (2012).

[26] D. Xiao, W. Zhu, Y. Ran, N. Nagaosa, and S. Okamoto,Nat. Commun. 2, 596 (2011).

[27] J. T. Vallin, Phys. Rev. B 2, 2390 (1970).[28] G. Chen, L. Balents, and A. P. Schnyder, Phys. Rev.

Lett. 102, 096406 (2009).

[29] A. Georges, G. Kotliar, W. Krauth, and M. J. Rozen-berg, Rev. Mod. Phys. 68, 13 (1996).

[30] J. C. Slater, and G. F. Koster, Phys. Rev. 94, 1498(1954).

[31] L. Fu and C. L. Kane, Phys. Rev. B 74, 195312 (2006).[32] T. Fukui and Y. Hatsugai, J. Phys. Soc. Jpn. 76, 053702

(2007).[33] E. Tang, J.- W. Mei, and X.- G. Wen, Phys. Rev. Lett.

106, 236802 (2011).[34] K. Sun, Z. Gu, H. Katsura, and S. Das Sarma, Phys.

Rev. Lett. 106, 236803 (2011).[35] T. Neupert, L. Santos, C. Chamon, and C. Mudry, Phys.

Rev. Lett. 106, 236804 (2011).[36] J. W. F. Venderbos, M. Daghofer, and J. van den Brink,

Phys. Rev. Lett. 107, 116401 (2011).[37] Y.- F. Wang, Z.-C. Gu, C.-D. Gong, and D. N. Sheng,

Phys. Rev. Lett. 107, 146803 (2011).[38] G. Cao, V. Durairaj, S. Chikara, L. E. DeLong,

S. Parkin, and P. Schlottmann, Phys. Rev. B 76,100402(R) (2007).

[39] M. Ralle and M. Jansen, J. Solid State Chem. 105, 378(1993).

[40] G. Kresse and D. Joubert, Phys. Rev. B 59, 1758 (1999).[41] J. P. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev.

Lett. 77, 3865 (1996).[42] G. Kresse and J. Furthmuller, Phys. Rev. B 54, 11169

(1996).[43] A. A. Mostofi, J. R. Yates, Y.-S. Lee, I. Souza, D. Van-

derbilt, and N. Marzari, Comput. Phys. Commun. 178,685 (2008).

[44] S. Okamoto, W. Zhu, Y. Nomura, R. Arita, D. Xiao,and N. Nagaosa, Phys. Rev. B 89, 195121 (2014).

[45] S. Sugano, Y. Tanabe, and H. Kamimura, Multipletsof Transition-Metal Ions in Crystals (Academic, NewYork, 1970).

[46] M. Caffarel and W. Krauth, Phys. Rev. Lett. 72, 1545(1994).

[47] C. A. Perroni, H. Ishida, and A. Liebsch, Phys. Rev. B75, 045125 (2007).

[48] R. B. Lehoucq, D. C. Sorensen, and C. Yang, ARPACKUsers Guide (SIAM, Philadelphia, 1997).

[49] F. Aryasetiawan, M. Imada, A. Georges, G. Kotliar,S. Biermann, and A. I. Lichtenstein, Phys. Rev. B 70,195104 (2004).

[50] A. Kozhevnikov, A. Eguiluz, and T. Schulthess, SC’10Proceedings of the 2010 ACM/IEEE International Con-ference for High Performance Computing, Networking,Storage, and Analysis (IEEE Computer Society, Wash-ington, DC, 2010), pp. 1-10 (2010).

[51] R. Arita, J. Kunes, A. V. Kozhevnikov, A. G. Eguiluz,and M. Imada, Phys. Rev. Lett. 108, 086403 (2012).

[52] V. I. Anisimov, I. A. Nekrasov, D. E. Kondakov, T. M.Rice, and M. Sigrist, Eur. Phys. J. B 25, 191 (2002).

[53] S. Okamoto, Phys. Rev. Lett. 110, 066403 (2013).[54] G. Jackeli and G. Khaliullin, Phys. Rev. Lett. 102,

017205 (2009).[55] S. Okamoto, Phys. Rev. B 87, 064508 (2013).[56] J. Chaloupka, G. Jackeli, and G. Khaliullin, Phys. Rev.

Lett. 110, 097204 (2013).[57] G. Baskaran, Z. Zou, and P. W. Anderson, Solid State

Commun. 63, 973 (1987).

11

[58] P. A. Lee, N. Nagaosa, and X.-G. Wen, Rev. Mod. Phys.78, 17 (2006).

[59] F. Wang and T. Senthil, Phys. Rev. Lett. 106, 136402(2011).

[60] H. Watanabe, T. Shirakawa, and S. Yunoki, Phys. Rev.Lett. 110, 027002 (2013).

[61] Z. Y. Meng, Y. B. Kim, and H.-Y. Kee, Phys. Rev. Lett.113, 177003 (2014).

[62] Y.-Z. You, I. Kimchi, and A. Vishwanath, Phys. Rev.B, 86, 085145 (2012).

[63] T. Hyart, A. R. Wright, G. Khaliullin, and B. Rosenow,Phys. Rev. B 85, 140510 (2012).

[64] X. Liu, S. Middey, Y. Cao, M. Kareev, J. Chakhalian,MRS Communications 6, 133 (2016).

[65] J. L. Lado, V. Pardo, and D. Baldomir, Phys. Rev. B88, 155119 (2013).

[66] D. Hirai, J. Matsuno, and H. Takagi, APL Materials 3,041508 (2015).

[67] Q.-F. Liang, L.-H. Wu, and X. Hu, NJP 15, 063031(2013).

[68] J. Chaloupka and G. Khaliullin, Phys. Rev. Lett. 100,016404 (2008).

[69] A. Ruegg and G. A. Fiete, Phys. Rev. B 84, 201103(R)(2011).

[70] K.-Y. Yang, W. Zhu, D. Xiao, S. Okamoto, Z. Wang,and Y. Ran, Phys. Rev. B 84, 201104(R) (2011).

[71] A. Ruegg, C. Mitra, A. A. Demkov, and G. A. Fiete,Phys. Rev. B 85, 245131 (2012).

[72] A. Ruegg, C. Mitra, A. A. Demkov, and G. A. Fiete,Phys. Rev. B 88, 115146 (2013).

[73] D. Doennig, W. E. Pickett, and R. Pentcheva, Phys.Rev. B 89, 121110(R) (2014).

[74] S. Middey, D. Meyers, M. Kareev, E. J. Moon, B. A.Gray, X. Liu, J. W. Freeland, and J. Chakhalian, Appl.Phys. Lett. 101, 261602 (2012).

[75] S. Middey, D. Meyers, D. Doennig, M. Kareev, X. Liu,Y. Cao, Zhenzhong Yang, Jinan Shi, Lin Gu, P.J. Ryan,R. Pentcheva, J.W. Freeland, and J. Chakhalian Phys.Rev. Lett. 116, 056801 (2016).

[76] B. Ye, A. Mesaros, and Y. Ran, Phys. Rev. B 89, 201111(2014).

[77] Y. Wang, Z. Wang, Z. Fang, and X. Dai, Phys. Rev. B91, 125139 (2015).

[78] A. Ohtomo and H. Y. Hwang, Nature 427, 423 (2004).[79] D. Doennig, W. E. Pickett, and R. Pentcheva, Phys.

Rev. Lett. 111, 126804 (2013).[80] Y. Weng, X. Huang, Y. Yao, and S. Dong, Phys. Rev.

B 92, 195114 (2015).[81] D. Doennig, S. Baidya, W. E. Pickett, and R.

Pentcheva, Phys. Rev. B 93, 165145 (2016)[82] H. K. Chandra and G.-Y. Guo, Phys. Rev. B 95, 134448

(2017).[83] H. Guo, S. Gangopadhyay, O. Koksal, R. Pentcheva, W.

E. Pickett, npj Quantum Materials 2, 4 (2017).[84] L. Si, O. Janson, G. Li, Z. Zhong, Z. Liao, G. Koster,

and K. Held, Phys. Rev. Lett. 119, 026402 (2017).[85] K. Yamauchi, P. Barone, T. Shishidou, T. Oguchi, and

S. Picozzi, Phys. Rev. Lett. 115, 037602 (2015).[86] D. Xiao, G.-B. Liu, W. Feng, X. Xu, and W. Yao, Phys.

Rev. Lett. 108, 196802 (2012).[87] A. M. Cook and A. Paramekanti, Phys. Rev. Lett. 113,

077203 (2014).[88] F. Wang and Y. Ran, Phys. Rev. B 84, 241103(R)

(2011).[89] X. Hu, A. Ruegg, G. A. Fiete, Phys. Rev. B 86, 235141

(2012).[90] M. Trescher and E. J. Bergholtz, Phys. Rev. B 86,

241111(R) (2012).[91] B.-J. Yang and N. Nagaosa, Phys. Rev. Lett. 112,

246402 (2014).[92] E. J. Bergholtz, Z. Liu, M. Trescher, R. Moessner, and

M. Udagawa, Phys. Rev. Lett. 114, 016806 (2015).[93] X. Hu, Z. Zhong, and G. A. Fiete, Sci. Rep. 5, 11072

(2015).[94] P. Laurell and G. A. Fiete, Phys. Rev. Lett. 118, 177201

(2017).[95] J. F. Afonso and V. Pardo, Phys. Rev. B 92, 235102

(2015).[96] C. I. Hiley, M. R. Lees, J. M. Fisher, D. Thompsett, S.

Agrestini, R. I. Smith, and R. I. Walton, Angew. Chem.,Int. Ed. Engl. 53, 4423 (2014).

[97] W. Tian, C. Svoboda, M. Ochi, M. Matsuda, H. B. Cao,J.-G. Cheng, B. C. Sales, D. G. Mandrus, R. Arita, N.Trivedi, and J.-Q. Yan, Phys. Rev. B 92, 100404(R)(2015).

[98] C. I. Hiley, D. O. Scanlon, A. A. Sokol, S. M. Woodley,A. M. Ganose, S. Sangiao, J. M. De Teresa, P. Manuel,D. D. Khalyavin, M. Walker, M. R. Lees, and R. I.Walton, Phys. Rev. B 92, 104413 (2015).

[99] M. Ochi, R. Arita, N. Trivedi, and S. Okamoto, Phys.Rev. B 93, 195149 (2016).

[100] L. Fu and C. L. Kane, Phys. Rev. B 76, 045302 (2007).