-
Magnetic Anisotropy in Spin-3/2 with Heavy Ligand in Honeycomb
Mott Insulators:Application to CrI3
P. Peter Stavropoulos,1 Xiaoyu Liu,1 and Hae-Young Kee1, 2,
∗
1Department of Physics and Center for Quantum
Materials,University of Toronto, 60 St. George St., Toronto,
Ontario, M5S 1A7, Canada
2Canadian Institute for Advanced Research, Toronto, Ontario, M5G
1Z8, Canada
Ferromagnetism in the two-dimensional CrI3 has generated a lot
of excitement, and it was re-cently proposed that the spin-orbit
coupling (SOC) in Iodine may generate bond-dependent
spininteractions leading to magnetic anisotropy. Here we derive a
microscopic spin model of S=3/2 ontransition metals surrounded by
heavy ligands in honeycomb Mott insulators using a
strong-couplingperturbation theory. For ideal octahedra we find
Heisenberg and Kitaev interactions, which favor themagnetic moment
along the cubic axis via quantum fluctuations. When a slight
trigonal distortion ofthe octahedra is present together with the
SOC, three additional interactions arise, comprised of
theoff-diagonal symmetric Γ and Γ′, and single-ion anisotropy. The
resulting magnetic anisotropy pinsthe moment perpendicular to the
honeycomb plane as observed in a single-layer of CrI3,
suggestingthe significance of SOC and trigonal distortion in
understanding magnetism of two dimensional Mottinsulators.
Comparison to the spin-orbit coupled Jeff= 1/2 and S=1 models is
also presented.
I. INTRODUCTION
Transition metal trihalides (TMT) are layered materi-als
composed of transition metals (M) and halides (X)of the group 9 in
a 1:3 ratio. They have a honeycomblayered structure, and depending
on the filling of the d-orbitals in the transition metals, some are
semiconductorsand some are metals.[1] Among them, RuCl3, VI3 and
CrI3are Mott insulators. Magnetic orderings in these systemsfurther
establish the importance of electronic correlationsand call for a
microscopic understanding of spin models.For example, based on a
strong-coupling perturbation the-ory of the generic spin
model[2–4], it was shown that α-RuCl3 described by the effective
spin Jeff = 1/2 has dom-inant bond-dependent Kitaev and
off-diagonal symmetricΓ interactions.[5, 6] RuCl3 has become an
emergent can-didate of spin-1/2 Kitaev spin liquid[7]. Intense
researchactivities on various properties of RuCl3 have been
carriedout[6, 8–14] and recently a magnetic-field induced spin
liq-uid was suggested.[15–21]
In parallel theoretical interest in the ground state
ofhigher-spin Kitaev models was initiated by classical
modelstudies.[22, 23] The classical Kitaev model has a macro-scopic
degeneracy named a classical spin liquid[22], but thehigher-spin
quantum Kitaev model is not exactly solvable,and the ground state
is currently unknown. Various numer-ical studies such as exact
diagonalization on S=1 suggestedthat the ground state is possibly a
spin liquid with gaplessexcitations.[24] These studies were mainly
of theoretical in-terest, until a microscopic derivation of the S=1
Kitaev-Heisenberg model in multi-orbital systems was
found.[25]Heavy ligand spin-orbit coupling (SOC) and strong
Hund’scoupling in eg orbitals is a way to generate S=1
bond-dependent Kitaev interaction. The magnetic field effectson the
S=1 Kitaev model have also been investigated.[26–28]
The bond-dependent interactions have recently beenadopted into
TMT systems, because the nearest neighbor
∗ [email protected]
(n.n.) Heisenberg J , Kitaev K and Γ interactions are al-lowed
based on the symmetry of the lattice.[4, 29, 30] Inparticular,
ferromagnetism in a single-layer CrI3 has gener-ated excitement in
recent years. [31–45] CrI3 is a ferromag-netic (FM) insulator with
Tc ∼ 61K for bulk samples.[46–48] Single-layer CrI3 was
successfully synthesized, whichshowed an FM ordering with Tc ∼ 45K.
[49] The two-dimensional FM Heisenberg model is insufficient to
explainfinite Tc, i.e., the Mermin-Wagner theorem[50], and
severaltheoretical models were proposed to explain the
magneticanisotropy. They include the XXZ model[51, 52], single-ion
anisotropy and Kitaev[53], and large Kitaev and smallsymmetric
off-diagonal Γ interactions[54].
While the Heisenberg, Kitaev, and Γ interactions are al-lowed by
the symmetry, and found to be significant in theearlier derivations
for lower-spins[3, 4, 25], their strengthsmay not be significant in
S=3/2 systems. Thus, a micro-scopic derivation of S=3/2 model is
necessary to find thesources of the magnetic anisotropy. Here we
derive a n.n.spin model for S=3/2 with three electrons in t2g
orbitalsof transition metal sites and strong SOC in p-orbitals
ofligands. We take into account strong electron-electron
in-teractions in multi-orbital systems including Hund’s cou-pling,
and effects of trigonal distortions present in R3̄rhombohedral
lattice. Contributions from eg orbitals areimportant as shown
below. The minimal n.n. model in-cludes J , K, Γ, another symmetric
off-diagonal Γ′[55], andsingle-ion anisotropy Ac along the ĉ-axis,
denoted as theJ −K − Γ− Γ′ −Ac model.
The rest of the paper is organized as follows. In Sec. II,the
on-site Kanamori interaction and tight binding Hamil-tonian are
presented. In Sec. III, we derive the n.n. spinmodel consisting of
Kitaev and Heisenberg interactions forideal octahedra environment
using standard perturbationtheory. In Sec. IV, we study the spin
model with trigonaldistortion present in R3̄ rhombohedral lattice.
This in-cludes the distortion-induced hopping matrix elements
andthree additional spin interactions generated via combinedeffects
of SOC and distortion. In Sec. V, we apply the the-ory to CrI3 and
present the exchange interaction strengthsusing tight binding
parameter sets obtained by densityfunctional theory. The effects of
the resulting magnetic
arX
iv:2
009.
0447
5v2
[co
nd-m
at.s
tr-e
l] 1
6 M
ar 2
021
mailto:[email protected]
-
2
Figure 1. Edge-shared octahedra honeycomb structure unit cella,
b, in global coordinates xyz. Transition metal sites M ingray and
non-magnetic ligands X in purple. The n.n. bonds X,Y, Z are related
by C3 symmetry. The sites M1, M2, X1, and X2are involved in the
second order strong-coupling expansion onthe Z bond. Indirect
hopping integrals t0, t1 and t2 are shown.
anisotropy on the moment direction are found in Sec. VI. InSec.
VII we discuss the origin of the spin gap, finite Tc, andspin wave
spectrum within the J −K−Γ−Γ′−Ac model.The effects of the the
second n.n. Dzyaloshinskii-Moriya(DM) interaction is also
discussed. Finally in Sec. VIII,we summarize our results and
compare with Jeff=1/2 andS=1 spin models. The detailed calculations
are presentedin the Appendix.
II. KANAMORI INTERACTION AND TIGHTBINDING HAMILTONIAN
The honeycomb network is made of metal (M) d-orbitalsites with
half filled t2g orbitals and octahedra cages of non-magnetic ligand
(X) sites with fully occupied p-orbitals.The full Hamiltonian is
composed of the on-site Kanamoriinteraction and the tight binding
Hamiltonian between twosites.
The on-site Hamiltonian of the M sites is described by
theKanamori interaction [56] as well as crystal field spitting:
Hee = U∑α
nα↑nα↓ +U ′
2
∑α6=β,σ,σ′
nασnβσ′
−JH2
∑α6=β,σ,σ′
c†ασc†βσ′cβσcασ′ + JH
∑α 6=β
c†α↑c†α↓cβ↓cβ↑
+∆c∑α∈eg,σ
c†ασcασ, (1)
where the density operator nασ is given by c†ασcασ, and c
†ασ
is the creation operator with α orbital and spin σ. U and
U ′ are the intra-orbital and inter-orbital Hubbard inter-action
respectively, and JH is the Hund’s coupling for thespin-exchange
and pair-hopping terms. ∆c is a crystal fieldsplitting on the M
sites, originated from the surroundingoctahedra, leading to the
splitting of the d-orbitals into t2gand eg orbitals. In a d
3 system one has half-filled t2g or-bitals, where the Hund’s
coupling JH selects for the S=3/2configuration as the ground state,
and the angular momen-tum is quenched. A table of the excited state
energy spec-trum is show in Appendix 1. The energies of the
exitedstates are larger than the hopping integrals, which allowsus
to treat the tight binding hopping integrals as a
pertur-bation.
In the edge shared octahedra structure, each bond be-tween n.n.
M sites involves two adjacent ligands, as shownin Fig. 1. A tight
binding Hamiltonian between two tran-sition metal sites M1 and M2
including the two adjacentligands X1 and X2 is given below
HTB =
05×5 TM1M2 TM1X1 TM1X2
TM1M2† 05×5 TM2X1 TM2X2
TM1X1† TM2X2
† 03×3 03×3
TM1X2† TM2X2
† 03×3 03×3
, (2)
where 0n×n refers the n × n null matrix. The basisis chosen as
(C†M1,d, C
†M2,d
, C†X1,p, C†X2,p
), where C†Mi,d =(c†i,x2−y2 , c
†i,3z2−r2 , c
†i,yz, c
†i,xz, c
†i,xy,
)are five d-orbitals at
site Mi, and C†Xm,p
=(c†m,px , c
†m,py , c
†m,pz
)are three p-
orbitals at ligand site Xm. Each block of indirect
hoppingbetween Mi and Xm is denoted by TMiXm and the directhopping
between M sites by TM1M2 . The details of eachblock matrix will be
presented latter.
To account for the indirect d to p hoppings we integrateout the
p-orbitals through a perturbative procedure trun-cated at second
order, leading to an effective d to d hoppingmodel:
TeffMiMj =∑
(a,m)
TMiXm |a〉〈a|TXmMj∆Ea
, (3)
where (a,m) represent a sum over all single hole statesa of all
sites Xm. The hole states are SOC states, thuscreating two energy
costs ∆Ea = ∆−λp/2 or ∆+λp, where∆ = �d− �p is the atomic energy
difference between M andX sites, and λp is the SOC in p-orbitals.
The SOC willintroduce explicit spin dependence in the effective d
to dhopping. The total effective hopping between the two Msites now
reads
HeffTB =
010x10 TM1M2 + TeffM1M2TM1M2
† + TeffM1M2†
010x10
,(4)
where we still retain the bare direct hoppings TM1M2 . Be-low we
focus on the ideal honeycomb structure and firstexamine the effects
of TeffM1M2 before adding the direct hop-pings TM1M2 and
summarizing the resulting spin model.
-
3
III. IDEAL HONEYCOMB STRUCTURE
To understand the microscopic origin of the spin model,we start
with the ideal honeycomb network surrounded byperfect edge shared
octahedra. It was shown that the sym-metry of the edge-shared
octahedra Z bond allows Heisen-berg J , Kitaev K, and symmetric
off-diagonal Γ interac-tions [4, 29, 30]. However, since their
strengths depend onvarious exchange processes, we perform the
strong couplingperturbation theory to determine the exchange
terms.
Truncating at second order in perturbation theory, we ar-rive at
the following Heisenberg-Kitaev (J−K) spin modelfor the ideal
honeycomb octahedra.
H =∑〈ij〉∈γ
J0Si · Sj +K0Sγi Sγj , (5)
where γ = x, y, z bond, and J0 and K0 refer to Heisenbergand
Kitaev interactions for the ideal octahedra.
Below we present the details of the derivation of Heisen-berg
and Kitaev interactions. An explanation of the ab-sence of the Γ
interaction within the second order pertur-bation theory is also
discussed. The exchange processes in-clude the contributions from
both indirect and direct hop-pings. We focus on the Z bond of the
honeycomb Fig. 1,as the other two bonds are related by C3
symmetry.
A. Superexchange path: indirect hopping
We first consider indirect hoppings between the M and Xsites,
which are the largest hopping integrals. They enterthrough the
effective hoppings of Eq. (4). The non-zero in-direct hopping
between M1 and X1 sites t0, t1, t2 are shownin Fig. 1. These
hoppings are incorporated in TeffM1M2 ,which can be simplified
using the Slater-Koster decompo-sition and symmetry related M −X
bonds as shown in theAppendix. There are two contributions to both
Heisenbergand Kitaev interactions, i.e., each interaction is
composedof two exchange terms; one is from t2g − t2g hoppings
andthe other is from eg − t2g hoppings
J0 = Jt2g0 + J
eg0 , K0 = K
t2g0 +K
eg0 , (6)
where the superscript t2g and eg refer to its
correspondinghopping processes. Below we present each exchange
path
leading to Jt2g0 , K
t2g0 , J
eg0 , and K
eg0 .
1. t2g − t2g contributions
Introducing the effective hopping integral teff =t203
(2
∆− λp/2+
1
∆ + λp
)between M1 and M2 via p-
orbitals, and the ratio r =2λp
2∆ + λpbetween SOC λp and
the atomic energy difference ∆, the hopping matrix involv-ing
only t2g orbitals, denoted by t0 in Fig. 1, can be sim-
plified in block form to
TeffM1M2(t2g ⊗ t2g) = teff
02×2 σo i
r
2σx
σo 02×2 −ir
2σy
−i r2σx i
r
2σy 02×2
,(7)
where t2g = {dyz, dxz, dxy} and σi with i = x, y, z are thePauli
matrices carrying the spin degrees of freedom and σois the 2 × 2
identity matrix. The holes in the intermedi-ate sates at the X site
and their indirect hopping TMiXmdetermine the type of σ matrices in
the effective hoppingFor example in the limit λp → 0, only the dyz
− dxz termteffσ0 → t20/∆ is present which contributes to the
directhopping channel. The new terms present for non-zero λpare the
spin-flip (SF) terms between dyz/dxz and dxy. Suchterms would
normally not appear in the second order per-turbation process, as
they involve a dyz/dxz − pz hoppingfollowed by a px/py−dxy hopping,
which will only occur ifpz is entangled with px/py. The SOC among
the px, py, pzgenerates such entanglement. Furthermore, when SOC
isthe dominant energy scale of the hole states, the wavefunc-tions
are inevitably mixtures of p-orbitals and their spin,leading to σi
dependence proportional to the r ratio of theSOC and atomic energy
difference.
The superexchange process involving only t2g orbitalsEq. (7)
result in
Jt2g0 =
8t2eff9 (U + 2JH)
, Kt2g0 = −
4(rteff)2
9 (U + 2JH). (8)
The spin-dependent hoppings have generated a Szi Szj Ki-
taev interaction. This can be rudimentarily understood bythe
following steps. To simplify the steps, we focus onone spin 1/2
electron hopping along the Z bond, throughonly a SF hopping.
Imagining two sites starting in (↑, ↑)state, the SF hopping can
lower the energy by the pro-cess: (↑, ↑) − SF − (0, ↓↑) − SF − (↑,
↑), at a energy costof −(rteff)2/U . On the other hand, if the two
sites startin (↑, ↓) the SF process is forbidden from Pauli
exclusionprinciple. Thus the exchange path starting from (↑, ↑)
andending in (↑, ↑) lowers the energy, which generates the FMSzi
S
zj interaction. Carrying out the details for the S=3/2
leads to the expression of the spin exchange terms shownin Eq.
(8).
Among the symmetry allowed terms, the symmet-ric off-diagonal
term with operator Sxi S
yj + S
yi S
xj =
i
2
(S−i S
−j − S
+i S
+j
)does not occur in the second order per-
turbation results. This operator would connect (↑, ↑) to(↓, ↓),
which is definitely possible from a SF process, how-ever, there is
a subtle cancellation. Having a single Paulimatrix in the SF term
creates such cancellation among thetwo paths within the spin block,
resulting in a null Γ term.Thus it becomes finite only when higher
order perturbationterms are included, or when the octahedra are no
longerideal, as we will show in the later Sec. IV.
-
4
2. eg − t2g contributions
Given that the p-orbital’s hybridization with eg is largerthan
with t2g, denoted by t1 and t2 in Fig. 1,this con-tribution is
essential. The final form of the effectiveTeffM1M2(eg⊗t2g) hopping
includes SF terms in the dyz, dxzto eg blocks, as well as
spin-independent −2teff(t2/t0)σ0hopping between dxy and d3z2−r2 .
The effective hoppingmatrix involving eg − t2g hoppings, in block
matrix form,reads
TeffM1M2(eg⊗t2g) = teff
ir
2
t1t0σy i
r
2
t1t0σx 02×2
−i r2
t2t0σy i
r
2
t2t0σx −2
t2t0σo
,(9)
where eg = {dx2−y2 , d3z2−r2} and t2g = {dyz, dxz, dxy}.Carrying
out the strong coupling expansion, these hoppingslead to the
additional contribution to the Heisenberg andKitaev
interactions
Jeg0 = −
16 JH t2eff
3 (∆c + U ′ − JH) (∆c + U ′ + 3JH)t22t20,
Keg0 =
2 JH(r teff)2
3 (∆c + U ′ − JH) (∆c + U ′ + 3JH)t21 + t
22
t20.
(10)
Similar to the t2g − t2g case, the SF terms contribute tothe
Kitaev interaction, while the spin-independent hop-ping,
−2teff(t2/t0)σ0, generates a FM Heisenberg term.The FM Heisenberg
interaction originates from the com-petition of two exited states
separated by Hund’s coupling,i.e, eg paths consistent with the
earlier findings obtainedby the first principle calculations.[45,
57]
B. Direct hopping
The direct hopping, denoted by td1, td2, td3 and t̃d0 inTM1M2
between M1 and M2 is given by
TM1M2 (t2g ⊗ t2g) =
td1 td2 0td2 td1 00 0 td3
⊗ σo,TM1M2 (eg ⊗ t2g) =
(0 0 00 0 t̃d0
)⊗ σo.
(11)
The Slater-Koster decomposition is explained in the Ap-pendix.
Since there is no SF hopping terms, this does notgenerate the
Kitaev interaction, but changes the Heisen-berg interaction.
C. Summary and Comments
Combining both indirect and direct hopping contribu-tions, the
two exchange interactions for the ideal octahedra
environment Eq. (5) are found to be
J0 =4(2t2d1 + 2(teff + td2)
2 + t2d3)
9 (U + 2JH)
−4JH
(2teff(t2/t0)− t̃d0
)23 (∆c + U ′ − JH) (∆c + U ′ + 3JH)
,
K0 = −4 (r teff)
2
9 (U + 2JH)
+2 JH(r teff)
2
3 (∆c + U ′ − JH) (∆c + U ′ + 3JH)t21 + t
22
t20. (12)
Thus for the ideal honeycomb structure, the J −K modelis derived
within the second order perturbation theory.
Some comments are useful, which will also motivate fur-ther
investigation on the effects of trigonal distortion pre-sented in
the next section. Kitaev and Heisenberg inter-actions have been
generated while the Γ interaction hasnot appeared at second order
due to a subtle cancella-tion. The Kitaev has been generated purely
from SF hop-pings and has a prefactor of r2 = (2λp/(2∆ + λp))
2, whilethe Heisenberg includes spin-independent hopping
contri-butions. Nevertheless both K and J have two
contributionswhich come with opposite signs: one from t2g only
pathsand the other from the paths involving eg. This leads to
asmaller Kitaev compared to Heisenberg interaction, unlessthe
contribution from eg paths reduces the overall strengthof the
Heisenberg interaction. So long as the crystal fieldspitting is not
excessively large, the naturally larger eg hop-pings will drive the
system to a FM Heisenberg interaction.
The FM J − K model pins the moment along the cu-bic axis when
the quantum fluctuations are taken intoaccount[58, 59], not along
the observed [111] direction inCrI3. We thus investigate if other
interaction terms maybe generated. The Γ interaction allowed by the
symme-try can be finite if higher order perturbation terms
areincluded. However, one may ask if there are other interac-tions
allowed by a slight distortion of the lattice within thesecond
order perturbation theory without invoking higherorder terms.
Indeed TMT materials do not have ideal oc-tahedra, but have either
rhombohedral R3̄ or monoclinicC2/m structures, and their magnetism
strongly depends onstructural differences and number of
layers.[60–63] Belowwe study the effects of distorted octahedra,
which induceadditional hopping integrals which were forbidden
withoutthe distortion.
IV. EFFECTS OF DISTORTION: DISTORTEDOCTAHDERA
CrI3 goes through a structural transition from C2/m toR3̄
structure at low temperature.[48] In the rhombohedralstructure,
there are two types of X ligand position devi-ations from the ideal
octahedra structure. As shown inFig. 2(a), a single octahedron can
be viewed as two shadedtriangles. One distortion is the staggered
rotations of thetwo triangles denoted by δx blue arrows, with
displace-ments of X sites perpendicular to the ĉ = [1, 1, 1]/
√3 di-
rection. The other distortion is the compression of the
dis-tance between these two triangles along the ĉ-axis denoted
-
5
a)
b)
Figure 2. a) The distorted octahedra in R3̄ is shown. The
octa-hedron made of X ligand can be viewed as two yellow
trianglesnormal to the ĉ = [1, 1, 1]/
√3 direction. The blue arrows rep-
resent new positions of X due to the staggered rotations of
thetwo yellow triangular faces. The change of position X due to
thestaggered rotations is parameterized by δx. In addition, thereis
a compression of the two yellow triangles, squeezed as shownby the
orange arrows parallel to the ĉ-axis. The change of posi-tion due
to the compression is parameterized by δx′. b) A topview of
distortions in a unit cell is shown. The dotted circles �indicate
the new position of X moving out of the page along +ĉwhile the
circled cross ⊗ indicate a new position of X movinginto the page
toward −ĉ. The exact positions of X(1,2,3,4,5,6) asa function of
δx, δx′ are found in the Appendix.
by δx′ orange arrows, with displacements along the ĉ
di-rection. Here the dimensionless parameters δx and δx′ arein
units of the distance between the n.n. M sites dM . An-alytic
formulas of the new positions of X sites under stag-gered rotations
and compression are found in Appendix 3.In the R3̄ space group
there are other types of distortions,namely the M1−X2,4,6 bond
length can be different fromthe M1−X1,3,5 bond length. This type of
distortion is gen-erally exceedingly small compared to δx and δx′
and weneglect it in the following analysis. Fig. 2(b) shows a
topview of the the honeycomb unit cell with two such
distortedoctahedra forming the Z bond, and the staggered rotationof
the right octahedron is the mirror image of the left
oc-tahedron.
The distortion-induced hopping matrices have all ele-ments
non-zero in TMiXm as a result of lowering the lo-cal symmetry of
the octahedron from Oh to D3. We de-note the new distortion allowed
hopping integrals as δti.Starting with the distortion-induced
hoppings, we followthe procedure described in Sec. III, namely we
use thedistortion-induced TMiXm matrices to derive the
effectiveTeffM1M2(t2g ⊗ t2g) and T
effM1M2
(eg ⊗ t2g). Details of theirform is deferred to Appendix 4.
Treating the distortion-induced effective hoppings as a
perturbation against theon-site interactions Eq. (1) the minimal
n.n. spin model isfinally given by
H =∑
〈ij〉∈αβ(γ)
[JSi · Sj +KSγi S
γj + Γ(S
αi S
βj + S
βi S
αj )
+ Γ′(Sαi Sγj + S
βi S
γj + S
γi S
αj + S
γi S
βj )]
+∑i
Ac(Si · ĉ)2,
(13)where α, β, (γ) refers to the γ bond taking α and β spin
components [4, 55], and ĉ = [1, 1, 1]/√
3. In addition tothe J −K terms, two symmetric off-diagonal
terms Γ andΓ′ have been generated, as well as the single-ion term
Acallowed by the C3 symmetry present on every site. Γ, Γ
′
and Ac are proportional to distortion induced hoppings δtas well
as r, thus both distortion-induced hoppings andSOC is needed to
bring rise to these terms. The form ofthe spin model J −K −Γ−Γ′ −Ac
to leading order in δtias well as eg contributions are listed in
Table VI in the Ap-pendix. Note that both Heisenberg and Kitaev
interactionsare renormalized by the distortion, but the Heisenberg
in-teraction has a linear term in δt, while Kitaev interactiondoes
not.
V. APPLICATION TO CRI3
To apply the above model to understand the magnetismin CrI3, we
use ab initio calculations to obtain the mi-croscopic parameters.
The calculation is performed withVienna ab initio simulation
package (VASP)[64]. We usethe Perdew-Burke-Ernzerhof (PBE)
functional[65] in ourcalculations. The experimental bulk
structure[48] is usedfor the calculations. We use WANNIER90[66] to
extractthe hopping integrals.
From the density functional theory results, we estimatethe
crystal field splitting ∆c = 1253 meV from on-site Crd-orbitals as
well as a crystal field splitting from the on-site I p-orbitals of
∆p = 528 meV. From Cr d- and I p-orbitals we extract the atomic
energy difference betweenCr and I ∆ = 2070 meV. Finally we find the
dominantindirect p−d hopping integrals t0, t1 and t2 and the
directd − d hoppings parameters, listed in Table I, as well asthe
distortion induced p − d hoppings, listed in Table II.We verified
that t0, t1 and t2 obtained by the ab initiocalculation match well
with the Slater-Koster expectationsof Eq. (A.2).
In the ab initio calculations, we find a sizable crystalfield
spitting ∆p on the ligand I sites, which we have nottaken into
account in the earlier analysis. It is about
-
6
t0 t1 t2 td1 td2 td3 t̃d0590.1 -992.13 -558.3 44.67 -41.26
-147.44 -20.75
TABLE I. Indirect and direct hoppings in units of meV obtainedby
ab initio calculations. The hopping integrals are defined inEq.
(A.1) and Eq. (11).
δt1 δt2 δt3 δt4 δt5 δt′1 δt
′2 δt
′3 δt
′4
2.34 21.76 -22.81 -61.74 65.86 70.97 -43.50 -49.41 29.65
TABLE II. Distortion-induced indirect hoppings in units ofmeV
obtained by ab initio calculations. The hopping integralsare
defined in Eq. (A.5).
528 meV, which is comparable to an estimated SOC pa-rameter of
λp = 630 meV. To capture its effects we re-visit the effective
hopping derivation, and add the crystalfield spitting ∆p on the X
sites. Then the energy level
in Eq. (3) split into three levels as ∆Ea = ∆ − λ2 ,∆ +14
(2∆p + λ±
√(2∆p − λ)2 + 8λ2
). While the holes in I
sites are no longer pure total angular momentum states,the SOC
is sizable enough to carry the spin entanglementthrough to the spin
model. We found that second orderperturbation theory, including the
spitting ∆p in the effec-tive hoppings, does not change the form of
the spin modelEq. (13). The analytic formulas would be vastly more
com-plex in this case, so we proceed to a direct numerical
eval-uation of J −K − Γ− Γ′ −Ac exchange terms using effec-tive
hoppings TeffM1M2 obtained by the ab-inito calculationslisted in
the Tables.
Assuming the spherical symmetry, i.e., U ′ = U − 2JH ,we are
left with two unknown parameters U and JH . Weplot J , K, Γ, Γ′,
and Ac as a function of JH/U for severalU values, with results
shown in Fig. 3. The dominant termis Heisenberg except near the
range 0.11 . JH/U . 0.13where J is almost zero before it changes
the sign. Thesign of J(K) are sensitive to the ratio JH/U and an
ad-equately large Hund’s coupling allows the FM Heisenberg(AFM
Kitaev) interaction to persevere. This reflects thecompetition of
the t2g vs eg terms seen in Eq. (12), witheg eventually wining over
t2g leading to the FM Heisen-berg interaction. The distortion
induced Γ− Γ′ −Ac comein with FM sign, and the single-ion
anisotropy Ac is thesizable term. Adopting the ratio of JH/U ∼ 0.24
ob-tained by cRPA clculations [67], CrI3 sits in the shadedarea in
Fig. 3. CrI3 is a ferromagnet with non-negligibleK − Γ − Γ′ − Ac,
and we now examine the implication ofthese exchange terms on the
observed moment.
VI. MAGNETIC ANISOTROPY
We have shown how the ideal honeycomb structure leadsto J −K
model up to second order in perturbation, whiledistortions
additionally generate Γ−Γ′−Ac. The magneticmoment of the
ferromagnetic state obtained with the FMJ−K model is pinned along
the cubic axis such as [100] andC3 equivalent directions via
quantum fluctuations[58, 59].
Figure 3. Spin model parameters J , K, Γ, Γ′, and Ac
estimatedfrom ab initio parameters and plotted against JH/U .
Solid,dashed, and dotted lines correspond to U = 3eV, 4eV and
5eVrespectively. Shaded region corresponds to JH/U ∼ 0.24 whichis
relevant to CrI3 microscopics.
We take a closer look at the effects of Γ − Γ′ − Ac on themoment
direction. We show three examples in Fig. 4 toillustrate the moment
pinning direction.
Following the method in Ref. [59], we perform exact
diag-onalization calculations on an eight site honeycomb
cluster,with the cluster setup shown in the Appendix. Once
theground state wavefunction |GS〉 is obtained, the
probabilitydistribution P = |〈ΨFM (θ, φ)| GS〉|2 is computed,
where|ΨFM (θ, φ)〉 is a ferromagnetic ansatz with moment direc-tion
pointing at (θ, φ) on the sphere. Results are show inFig. 4 for
different values of J , K, and Γ. Independent of ra-tio of J and K,
the moment is along the cubic axis as shownin the panel (a) and
(b). On the other hand, when a smallFM Γ is introduced, the moment
is along the ĉ direction asshown in panel (c). This effect can be
anticipated from the
a) b) c)
Figure 4. Moment pinning calculation for difrent values
of(J,K,Γ) in the ferromagnetic phase: a) (−1.00,−0.20, 0.00),b)
(−0.20,−1.00, 0.00), c) (−0.20,−1.00,−0.02). Kitaev inter-action
always prefers the cubic axis in panel a) and b). A
smallinteraction like Γ is necessary to pin along the ĉ direction
asshow in panel c).
-
7
classical analysis. The classical J−K−Γ model in the FMstate
with moment S0 = (S
x0 , S
y0 , S
z0 ) has an energy density
per unit cell uc = (3J+K)+2Γ(Sx0Sy0 +S
y0S
z0+S
z0S
x0 ), which
demands min [uc] ⇒ min [sgn(Γ)(Sx0Sy0 + S
y0S
z0 + S
z0S
x0 )].
When Γ < 0 we have min [−(Sx0Sy0 + S
y0S
z0 + S
z0S
x0 )] ⇒
S0 = ĉ leading to a moment pined on the [1,1,1]. We chooseΓ/K �
1 to show that a tiny FM Γ anisotropy results inthe ĉ-axis moment
shown in Fig. 4 (c).
FM Γ′ has the identical effect as FM Γ with respect tothe moment
direction of the ferromagnet. Clearly FM Acalso promotes the
ĉ-axis moment direction. All three inter-actions individually
prefer the moment on the ĉ-axis. Thiseffect is bigger than the
quantum fluctuation effect of cubicaxis pinning from the Kitaev
term resulting in the observedĉ-axis moment direction.
VII. SPIN GAP AND FINITE TRANSITIONTEMPERATURE
The preceding results suggest that it is likely that CrI3has
dominant FM Heisenberg and smaller Kitaev interac-tion, and the
magnetic anistoropy originates from the dis-tortion of the
octahedra together with the SOC of heavy lig-ands. The magnetic
ordering pattern changes from bulk tofilms [49, 63, 67–70],
suggesting a strong coupling betweenmagnetism and crystal
structures, which further impliesthe importance of the distorted
octahedra in the presenceof SOC. Our microscopic model has an
interesting connec-tion to the previous studies. The Kitaev and
single-ionanistoropy in addition to the Heisenberg interaction
werefound in Ref.[53] using density functional theory. If Γ =
Γ′,the J−Γ−Γ′ model maps to the XXZ model[51, 52] in thea− b− c
crystallographic coordinate
JSi · Sj+Γ(Sxi Syj + S
yi S
xj + S
xi S
zj + S
zi S
xj + S
yi S
zj + S
zi S
yj )
=⇒ JabSi · Sj + JcSciScj , (14)
where Jab = J − Γ and Jc = 3Γ.Since the exchange parameters
strongly depend on
JH/U , ∆c/U , and tight binding parameters, it is use-ful to
obtain experimental inputs to determine some pa-rameters. The
inelastic neutron scattering experimentsand magneto-Raman
spectroscopy have reported a spingap of approximately 0.36 meV at
the Brillouin zone(BZ) center Γ-point.[71, 72] This is also
consistent witha small anistoropy found in the ferromagnetic
resonanceexperiment[54], which is about 0.07 meV leading to thespin
gap of 0.3 meV.
Based on our spin wave analysis using the J −K − Γ−Γ′ − Ac model
including the second n.n. DM interaction,the spin wave dispersion
ωk is expressed as ω0+ρ k
2 aroundthe Γ-point. Here ω0 and ρ are the spin gap and
stiffness,respectively, and they are given by
ωo = −S(3Γ + 6Γ′ + 2Ac),
ρ =S
12
∣∣∣3J +K − Γ− 2Γ′ − (K+2Γ−2Γ′)22(2Γ+4Γ′+2Ac+3J+K) ∣∣∣ (15)The
details of the linear spin wave theory (LSWT) is pre-sented in
Appendix 7.
Kitaev and DM interactions do not generate a gap atthe Γ-point
within the LSWT. The classical ferromag-netic ground state under
the Kitaev and DM terms have acontinuous degeneracy, and as a
result expanding aroundthis ground state in the LSWT will not
result in a spingap from these terms. The spin gap is rather small,
asexpected because it is originated from a combination ofslightly
distorted octahedra and SOC, i.e., Γ, Γ′, andAc. While small, it is
essential for a finite Tc in a sin-gle layer CrI3. In the FM
ordered phase, at low tem-peratures the magnons are excited and
their number isgiven by Ns(T ) =
∫d2k 1
eβωk−1 =πβρ
∫βω0
dxex−1 . Without
the spin gap ω0, Ns diverges in two-dimension at any
tem-perature except T = 0, i.e., the celebrated
Mermin-Wagnertheorem[50]. Thus one can understand the essential
role ofω0 which cuts the divergence, and allows the FM orderingat
finite temperatures, as long as ω0(T ) remains finite forT ≤ Tc.
While quantifying the transition temperature re-quires further
analysis [73] , the temperature dependenceof ρ(T ) and ω0(T ) from
the inelastic neutron scatteringmeasurement[71] indicates the
crucial role of ω0(T ) whichvanishes at Tc.
Another important parameter is the Kitaev interac-tion K, which
leads to a gap at the the BZ corner K-point known as the Dirac
gap[54], reported in the neutronscattering.[71, 74] However, the
second n.n. DM term alsogenerates the Dirac gap.[71, 74] We would
like to point outthat Γ and Γ′ also play a part in the Dirac gap as
shownin Eq. (A.14). Estimating the Kitaev interaction by an
in-dependent experimental measurement and further analysison the
individual role of the Kitaev and DM interactionsremain to be
resolved in future studies.
VIII. SUMMARY AND DISCUSSION
We have shown a microscopic derivation of the n.n. spinmodel for
honeycomb Mott insulators with three electronsin t2g orbitals at M
sites surrounded by octahedral cagesof ligands X with strong SOC.
Using the standard strong-coupling perturbation theory, we found
that there are onlyHeisenberg and Kitaev interactions for the ideal
honey-comb lattice among the three symmetry allowed interac-tions
(J,K,Γ), because Γ is zero up to the second order per-turbation.
The exchange paths between t2g and t2g vs. t2gand eg via ligands
generate opposite signs for both Heisen-berg and Kitaev
interactions. The Heisenberg interactionis of order t2eff/U , while
the Kitaev is smaller by a factor ofr2 ∼ (λp/∆)2. The FM Heisenberg
interaction originatesfrom the eg paths, with the hopping integral
between egand p-orbitals being larger compared to t2g and
p-orbitals.
The FM Heisenberg and Kitaev interaction leads to FMordering,
but the moment direction is pinned along thecubic x, y, or z axis,
e.g., [100] (and C3 equivalent direc-tions) via quantum
fluctuations. The ĉ-axis, [111], momentpinning found in CrI3
should thus originate from other in-teractions, which are also
responsible for the spin gap atthe Γ-point in the neutron
scattering measurement[71]. In-cluding the distorted octahedra
present in the rhombohe-dral structure, three additional spin
interactions are found.Inspecting the linear order in the
distortion-induced hop-
-
8
ping paths within the second order perturbation theory, Γ,Γ′ and
single-ion anisotropy Ac contain terms linear in
thedistortion-induced hopping integrals. The Heisenberg
in-teraction also contains such additional linear term, but
theKitaev does not, implying that it is possible to fine-tune
asystem closer to the Kitaev dominant regime via
trigonaldistortions.
In this work we have focused on the nature of the mono-layer,
however, some comments are in place when consider-ing the physics
of the bulk. We showed that the t2g−t2g vs.eg − t2g contributions
to the intra-layer Heisenberg term Jare opposite in sign. This
property should hold for inter-layer Heisenberg interaction J⊥ as
well, because it is deter-mined from superexchange processes. Thus
the magneticordering pattern between layers depends on the details
oforbital compositions. From the ab initio calculation, wefound
that the inter-layer hopping ranges from 10meV to30meV. This leads
to J⊥ of order 0.1 meV. While it doesnot affect the intra-layer
magnetism presented here, it isimportant for the ordering pattern
in the bulk[49, 63, 67–70].
Comparison to Jeff = 1/2, S=1, and 3/2 spin systemswould be
useful. The SOC is necessary to generate thebond-dependent
interaction, as spin and orbitals should beentangled to get such a
directional dependent spin interac-tion. However, the presence of
SOC is not enough to findan exotic phase like a spin liquid,
because the dominant in-teraction is often the Heisenberg
interaction. To comparedifferent spin cases, a summary of the ideal
honeycomb ex-change interactions, J0, K0, and Γ0 including the
effectiveindirect hopping (teff) and direct hopping (td)
integrals,only up to the second order perturbation, is shown in
thefollowing Table III.
Spin Heisenberg J0 Kitaev K0symm.
off-diagonal Γ0
Jeff = 1/2 (d5) O(
t2dt2eff
) O(JHU
) O( tdteff
JHU
)
S = 1 (d8) O(r2) O(r2) 0
S = 3/2 (d3) O(1) O(r2) 0
TABLE III. The leading term of the exchange interactions forthe
ideal honeycomb structure in unit of t2eff/U for different spinS
including only up to the second order perturbation terms.See the
Appendix for the full expression of JKΓΓ′Ac for S=3/2including the
trigonal distortion-induced hopping contributions.
Focusing on the ideal octahedra and n.n. model via sec-ond order
superexchange processes, Jeff = 1/2 is uniquebecause the Heisenberg
term is absent. On the other hand,for the S=1 model from d2 in
eg-orbitals, the heavy ligandSOC generates the Kitaev interaction,
which has the sameorder of magnitude with J . In fact, K = −2J , if
only egpaths are considered.[25] For S=3/2 case, we found that Jis
order 1 in units of roughly t2eff/U , while K is smaller by
r2. Thus it is hard to compete with the Heisenberg inter-action.
We speculate that this is valid for spins equal orhigher than 3/2.
Unlike S=1/2 and S=1 cases, the Heisen-berg interaction is dominant
in S=3/2, but a delicate can-cellation among different
contributions to the Heisenberginteraction, may let the Kitaev
interaction overtake a ma-jor place. In particular, the Heisenberg
interaction is moresensitive to the distortion-induced hopping
integrals thanthe Kitaev term as shown in Table VI in the
Appendix,manipulating ocathedra may be a way to tune the systemto a
desired Kitaev dominant regime.
In summary, in the ideal octahedra environment, we findthat
there are only two spin interactions, Heisenberg andKitaev
interactions. Kitaev interaction is generally weakercompared to the
Heisenberg interaction in contrast to thelower spin models.
Indirect hoppings among t2g and t2gvs. eg and t2g have opposite
contributions. A detailedbalance between the two indirect and
direct hopping con-tributions highly depends on the hopping
integrals, Hund’scoupling strength, and crystal field spitting. Γ
interactionis absent up to the second order due to a subtle
cancella-tion. We further show that trigonal distortion allows
threeadditional interactions, two symmetric off-diagonal
inter-actions Γ and Γ′, and single-ion anisotropy Ac along
theĉ-axis. They are all linearly proportional to a
distortion-induced hopping integral. While they are much
smallerthan the Heisenberg interaction, they are essential for
aspin gap in the FM phase of CrI3 leading to a finite Tc.Our study
offers a microscopic route to the n.n. spin mod-els, J − K − Γ − Γ′
− Ac interactions. Given that thereare five exchange terms within
the n.n. model, and secondn.n. interactions including DM may be
comparable to Γ,Γ′ and Ac, further theoretical and experimental
studies arerequired to determine the microscopic parameters of
CrI3beyond the n.n. model.
-¿
ACKNOWLEDGMENTS
We acknowledge I. Lee, C. Hammel, N. Trivedi, P. Dai,and R.
Valenti for useful discussion. H.Y.K. acknowl-edges the funding
from the Canada Research Chair Pro-gram. This work was supported by
the Natural Sciencesand Engineering Research Council of Canada, the
Centerfor Quantum Materials at the University of Toronto, andthe
Canadian Institute for Advanced Research. Compu-tations were
performed on the Niagara supercomputer atthe SciNet HPC Consortium.
SciNet is funded by: theCanada Foundation for Innovation under the
auspices ofCompute Canada; the Government of Ontario; Ontario
Re-search Fund - Research Excellence; and the University
ofToronto.
[1] M. A. McGuire, Crystals 7, 121 (2017).[2] G. Khaliullin,
Prog. Theor. Phys. Supp. 160, 155 (2005).[3] G. Jackeli and G.
Khaliullin, Phys. Rev. Lett. 102, 017205
(2009).[4] J. G. Rau, E. K.-H. Lee, and H.-Y. Kee, Phys. Rev.
Lett.
112, 077204 (2014).
http://dx.doi.org/10.3390/cryst7050121http://dx.doi.org/10.1143/PTPS.160.155http://dx.doi.org/10.1103/PhysRevLett.102.017205http://dx.doi.org/10.1103/PhysRevLett.102.017205http://dx.doi.org/10.1103/PhysRevLett.112.077204http://dx.doi.org/10.1103/PhysRevLett.112.077204
-
9
[5] K. W. Plumb, J. P. Clancy, L. J. Sandilands, V. V.
Shankar,Y. F. Hu, K. S. Burch, H.-Y. Kee, and Y.-J. Kim, Phys.Rev.
B 90, 041112(R) (2014).
[6] H.-S. Kim, V. Shankar V., A. Catuneanu, and H.-Y. Kee,Phys.
Rev. B 91, 241110(R) (2015).
[7] A. Kitaev, Ann. Phys. (N. Y.) 321, 2 (2006), JanuarySpecial
Issue.
[8] L. J. Sandilands, Y. Tian, K. W. Plumb, Y.-J. Kim, andK. S.
Burch, Phys. Rev. Lett. 114, 147201 (2015).
[9] J. A. Sears, M. Songvilay, K. W. Plumb, J. P. Clancy,Y. Qiu,
Y. Zhao, D. Parshall, and Y.-J. Kim, Phys. Rev.B 91, 144420
(2015).
[10] R. D. Johnson, S. C. Williams, A. A. Haghighirad, J.
Sin-gleton, V. Zapf, P. Manuel, I. I. Mazin, Y. Li, H. O.
Jeschke,R. Valent́ı, and R. Coldea, Phys. Rev. B 92, 235119
(2015).
[11] A. Banerjee, C. A. Bridges, J.-Q. Yan, A. A. Aczel, L.
Li,M. B. Stone, G. E. Granroth, M. D. Lumsden, Y. Yiu,J. Knolle, S.
Bhattacharjee, D. L. Kovrizhin, R. Moessner,D. A. Tennant, D. G.
Mandrus, and S. E. Nagler, Nat.Materials 15, 733 (2016),
article.
[12] H. B. Cao, A. Banerjee, J.-Q. Yan, C. A. Bridges, M.
D.Lumsden, D. G. Mandrus, D. A. Tennant, B. C. Chak-oumakos, and S.
E. Nagler, Phys. Rev. B 93, 134423(2016).
[13] H.-S. Kim and H.-Y. Kee, Phys. Rev. B 93, 155143
(2016).[14] L. Janssen, E. C. Andrade, and M. Vojta, Phys. Rev.
B
96, 064430 (2017).[15] R. Yadav, N. A. Bogdanov, V. M. Katukuri,
S. Nishimoto,
J. van den Brink, and L. Hozoi, Sci. Rep. 6, 37925 (2016).[16]
S.-H. Baek, S.-H. Do, K.-Y. Choi, Y. S. Kwon, A. U. B.
Wolter, S. Nishimoto, J. van den Brink, and B. Büchner,Phys.
Rev. Lett. 119, 037201 (2017).
[17] A. U. B. Wolter, L. T. Corredor, L. Janssen, K. Nenkov,S.
Schönecker, S.-H. Do, K.-Y. Choi, R. Albrecht,J. Hunger, T. Doert,
M. Vojta, and B. Büchner, Phys.Rev. B 96, 041405(R) (2017).
[18] J. Zheng, K. Ran, T. Li, J. Wang, P. Wang, B. Liu,
Z.-X.Liu, B. Normand, J. Wen, and W. Yu, Phys. Rev. Lett.119,
227208 (2017).
[19] N. Janša, A. Zorko, M. Gomiľsek, M. Pregelj, K.
W.Krämer, D. Biner, A. Biffin, C. Rüegg, and M. Klanǰsek,Nat.
Phys. 14, 786 (2018).
[20] Y. Kasahara, T. Ohnishi, Y. Mizukami, O. Tanaka, S. Ma,K.
Sugii, N. Kurita, H. Tanaka, J. Nasu, Y. Motome,T. Shibauchi, and
Y. Matsuda, Nature 559, 227 (2018).
[21] M. Yamashita, J. Gouchi, Y. Uwatoko, N. Kurita, andH.
Tanaka, Phys. Rev. B 102, 220404(R) (2020).
[22] G. Baskaran, D. Sen, and R. Shankar, Phys. Rev. B 78,115116
(2008).
[23] J. Oitmaa, A. Koga, and R. R. P. Singh, Phys. Rev. B
98,214404 (2018).
[24] A. Koga, H. Tomishige, and J. Nasu, J. Phys. Soc. Jpn.87,
063703 (2018).
[25] P. P. Stavropoulos, D. Pereira, and H.-Y. Kee, Phys.
Rev.Lett. 123, 037203 (2019).
[26] C. Hickey, C. Berke, P. P. Stavropoulos, H.-Y. Kee, andS.
Trebst, Phys. Rev. Research 2, 023361 (2020).
[27] Z. Zhu, Z.-Y. Weng, and D. N. Sheng, Phys. Rev. Research2,
022047(R) (2020).
[28] I. Khait, P. P. Stavropoulos, H.-Y. Kee, and Y. B.
Kim,arXiv:2001.06000 [cond-mat.str-el].
[29] Y. Yamaji, Y. Nomura, M. Kurita, R. Arita, andM. Imada,
Phys. Rev. Lett. 113, 107201 (2014).
[30] V. M. Katukuri, S. Nishimoto, V. Yushankhai, A. Stoy-anova,
H. Kandpal, S. Choi, R. Coldea, I. Rousochatzakis,L. Hozoi, and J.
van den Brink, New J. Phys. 16, 013056(2014).
[31] W.-B. Zhang, Q. Qu, P. Zhu, and C.-H. Lam, J. Mater.Chem. C
3, 12457 (2015).
[32] D. Torelli and T. Olsen, 2D Mater. 6, 015028 (2018).[33] S.
S. Pershoguba, S. Banerjee, J. C. Lashley, J. Park,
H. Ågren, G. Aeppli, and A. V. Balatsky, Phys. Rev. X 8,011010
(2018).
[34] F. Zheng, J. Zhao, Z. Liu, M. Li, M. Zhou, S. Zhang, andP.
Zhang, Nanoscale 10, 14298 (2018).
[35] L. Webster and J.-A. Yan, Phys. Rev. B 98, 144411
(2018).[36] J. Liu, M. Shi, P. Mo, and J. Lu, AIP Adv. 8,
055316
(2018).[37] M. Wu, Z. Li, T. Cao, and S. G. Louie, Nat. Commun.
10,
2371 (2019).[38] T. Olsen, MRS Commun. 9, 1142 (2019).[39] O.
Besbes, S. Nikolaev, N. Meskini, and I. Solovyev, Phys.
Rev. B 99, 104432 (2019).[40] V. K. Gudelli and G.-Y. Guo, New
J. Phys. 21, 053012
(2019).[41] C. Xu, J. Feng, M. Kawamura, Y. Yamaji, Y.
Nahas,
S. Prokhorenko, Y. Qi, H. Xiang, and L. Bellaiche, Phys.Rev.
Lett. 124, 087205 (2020).
[42] M. Pizzochero and O. V. Yazyev, J. Phys. Chem. 124,
7585(2020).
[43] M. Pizzochero, R. Yadav, and O. V. Yazyev, 2D Mater.
7,035005 (2020).
[44] E. Aguilera, R. Jaeschke-Ubiergo, N. Vidal-Silva, L. E.
F.Foa Torres, and A. S. Nunez, Phys. Rev. B 102, 024409(2020).
[45] D. Soriano, M. I. Katsnelson, and J.
Fernández-Rossier,Nano Lett. 20, 6225 (2020).
[46] L. L. Handy and N. W. Gregory, J. Am. Chem. Soc. 74,891
(1952).
[47] J. F. Dillon and C. E. Olson, J. Appl. Phys. 36, 1259
(1965).[48] M. A. McGuire, H. Dixit, V. R. Cooper, and B. C.
Sales,
Chem. Mater. 27, 612 (2015).[49] B. Huang, G. Clark, E.
Navarro-Moratalla, D. R. Klein,
R. Cheng, K. L. Seyler, D. Zhong, E. Schmidgall, M. A.McGuire,
D. H. Cobden, W. Yao, D. Xiao, P. Jarillo-Herrero, and X. Xu,
Nature 546, 270 (2017).
[50] N. D. Mermin and H. Wagner, Phys. Rev. Lett. 17,
1133(1966).
[51] J. L. Lado and J. Fernández-Rossier, 2D Mater. 4,
035002(2017).
[52] D.-H. Kim, K. Kim, K.-T. Ko, J. Seo, J. S. Kim, T.-H.Jang,
Y. Kim, J.-Y. Kim, S.-W. Cheong, and J.-H. Park,Phys. Rev. Lett.
122, 207201 (2019).
[53] C. Xu, J. Feng, H. Xiang, and L. Bellaiche, npj
Comput.Materials 4, 57 (2018).
[54] I. Lee, F. G. Utermohlen, D. Weber, K. Hwang, C. Zhang,J.
van Tol, J. E. Goldberger, N. Trivedi, and P. C. Hammel,Phys. Rev.
Lett. 124, 017201 (2020).
[55] J. G. Rau and H.-Y. Kee, arXiv:1408.4811
[cond-mat.str-el].
[56] J. Kanamori, Prog. Theor. Phys. 30, 275 (1963).[57] I. V.
Kashin, V. V. Mazurenko, M. I. Katsnelson, and
A. N. Rudenko, 2D Mater. 7, 025036 (2020).[58] G. Jackeli and A.
Avella, Phys. Rev. B 92, 184416 (2015).[59] J. Chaloupka and G.
Khaliullin, Phys. Rev. B 94, 064435
(2016).[60] D. R. Klein, D. MacNeill, J. L. Lado, D. Sori-
ano, E. Navarro-Moratalla, K. Watanabe, T. Taniguchi,S. Manni,
P. Canfield, J. Fernández-Rossier, and P. Jarillo-Herrero, Science
360, 1218 (2018).
[61] T. Song, X. Cai, M. W.-Y. Tu, X. Zhang, B. Huang, N.
P.Wilson, K. L. Seyler, L. Zhu, T. Taniguchi, K. Watanabe,M. A.
McGuire, D. H. Cobden, D. Xiao, W. Yao, andX. Xu, Science 360, 1214
(2018).
http://dx.doi.org/ 10.1103/PhysRevB.90.041112http://dx.doi.org/
10.1103/PhysRevB.90.041112http://dx.doi.org/10.1103/PhysRevB.91.241110http://dx.doi.org/https://doi.org/10.1016/j.aop.2005.10.005http://dx.doi.org/
10.1103/PhysRevLett.114.147201http://dx.doi.org/10.1103/PhysRevB.91.144420http://dx.doi.org/10.1103/PhysRevB.91.144420http://dx.doi.org/10.1103/PhysRevB.92.235119http://dx.doi.org/10.1038/nmat4604http://dx.doi.org/10.1038/nmat4604http://dx.doi.org/
10.1103/PhysRevB.93.134423http://dx.doi.org/
10.1103/PhysRevB.93.134423http://dx.doi.org/10.1103/PhysRevB.93.155143http://dx.doi.org/10.1103/PhysRevB.96.064430http://dx.doi.org/10.1103/PhysRevB.96.064430http://dx.doi.org/
https://doi.org/10.1038/srep37925http://dx.doi.org/
10.1103/PhysRevLett.119.037201http://dx.doi.org/10.1103/PhysRevB.96.041405http://dx.doi.org/10.1103/PhysRevB.96.041405http://dx.doi.org/
10.1103/PhysRevLett.119.227208http://dx.doi.org/
10.1103/PhysRevLett.119.227208http://dx.doi.org/
10.1038/s41567-018-0129-5http://dx.doi.org/10.1038/s41586-018-0274-0http://dx.doi.org/
10.1103/PhysRevB.102.220404http://dx.doi.org/10.1103/PhysRevB.78.115116http://dx.doi.org/10.1103/PhysRevB.78.115116http://dx.doi.org/10.1103/PhysRevB.98.214404http://dx.doi.org/10.1103/PhysRevB.98.214404http://dx.doi.org/10.7566/JPSJ.87.063703http://dx.doi.org/10.7566/JPSJ.87.063703http://dx.doi.org/10.1103/PhysRevLett.123.037203http://dx.doi.org/10.1103/PhysRevLett.123.037203http://dx.doi.org/10.1103/PhysRevResearch.2.023361http://dx.doi.org/10.1103/PhysRevResearch.2.022047http://dx.doi.org/10.1103/PhysRevResearch.2.022047http://arxiv.org/abs/2001.06000http://dx.doi.org/
10.1103/PhysRevLett.113.107201http://dx.doi.org/10.1088/1367-2630/16/1/013056http://dx.doi.org/10.1088/1367-2630/16/1/013056http://dx.doi.org/
10.1039/C5TC02840Jhttp://dx.doi.org/
10.1039/C5TC02840Jhttp://dx.doi.org/10.1088/2053-1583/aaf06dhttp://dx.doi.org/
10.1103/PhysRevX.8.011010http://dx.doi.org/
10.1103/PhysRevX.8.011010http://dx.doi.org/10.1039/C8NR03230Khttp://dx.doi.org/10.1103/PhysRevB.98.144411http://dx.doi.org/10.1063/1.5030441http://dx.doi.org/10.1063/1.5030441http://dx.doi.org/
10.1038/s41467-019-10325-7http://dx.doi.org/
10.1038/s41467-019-10325-7http://dx.doi.org/10.1557/mrc.2019.117http://dx.doi.org/10.1103/PhysRevB.99.104432http://dx.doi.org/10.1103/PhysRevB.99.104432http://dx.doi.org/10.1088/1367-2630/ab1ae9http://dx.doi.org/10.1088/1367-2630/ab1ae9http://dx.doi.org/10.1103/PhysRevLett.124.087205http://dx.doi.org/10.1103/PhysRevLett.124.087205http://dx.doi.org/10.1021/acs.jpcc.0c01873http://dx.doi.org/10.1021/acs.jpcc.0c01873http://dx.doi.org/10.1088/2053-1583/ab7cabhttp://dx.doi.org/10.1088/2053-1583/ab7cabhttp://dx.doi.org/10.1103/PhysRevB.102.024409http://dx.doi.org/10.1103/PhysRevB.102.024409http://dx.doi.org/10.1021/acs.nanolett.0c02381http://dx.doi.org/10.1021/ja01124a009http://dx.doi.org/10.1021/ja01124a009http://dx.doi.org/10.1063/1.1714194http://dx.doi.org/10.1021/cm504242thttp://dx.doi.org/
10.1038/nature22391http://dx.doi.org/10.1103/PhysRevLett.17.1133http://dx.doi.org/10.1103/PhysRevLett.17.1133http://dx.doi.org/10.1088/2053-1583/aa75edhttp://dx.doi.org/10.1088/2053-1583/aa75edhttp://dx.doi.org/10.1103/PhysRevLett.122.207201http://dx.doi.org/
10.1038/s41524-018-0115-6http://dx.doi.org/
10.1038/s41524-018-0115-6http://dx.doi.org/
10.1103/PhysRevLett.124.017201http://arxiv.org/abs/1408.4811http://arxiv.org/abs/1408.4811http://dx.doi.org/10.1143/PTP.30.275http://dx.doi.org/10.1088/2053-1583/ab72d8http://dx.doi.org/10.1103/PhysRevB.92.184416http://dx.doi.org/10.1103/PhysRevB.94.064435http://dx.doi.org/10.1103/PhysRevB.94.064435http://dx.doi.org/
10.1126/science.aar3617http://dx.doi.org/10.1126/science.aar4851
-
10
[62] D. R. Klein, D. MacNeill, Q. Song, D. T. Larson, S. Fang,M.
Xu, R. A. Ribeiro, P. C. Canfield, E. Kaxiras, R. Comin,and P.
Jarillo-Herrero, Nat. Phys. 15, 1255 (2019).
[63] H. H. Kim, B. Yang, S. Li, S. Jiang, C. Jin, Z. Tao,G.
Nichols, F. Sfigakis, S. Zhong, C. Li, S. Tian, D. G.Cory, G.-X.
Miao, J. Shan, K. F. Mak, H. Lei, K. Sun,L. Zhao, and A. W. Tsen,
Proc. Natl. Acad. Sci. U. S. A.116, 11131 (2019).
[64] G. Kresse and J. Hafner, Phy. Rev. B 47, 558(R) (1993).[65]
J. P. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev.
Lett. 77, 3865 (1997).[66] A. A. Mostofi, J. R. Yates, Y.-S.
Lee, I. Souza, D. Vander-
bilt, and N. Marzari, Comput. Phys. Commun. 178, 685(2008).
[67] S. W. Jang, M. Y. Jeong, H. Yoon, S. Ryee, and M. J.Han,
Phys. Rev. Mater. 3, 031001(R) (2019).
[68] N. Sivadas, S. Okamoto, X. Xu, C. J. Fennie, and D.
Xiao,Nano Lett. 18, 7658 (2018).
[69] P. Jiang, C. Wang, D. Chen, Z. Zhong, Z. Yuan, Z.-Y. Lu,and
W. Ji, Phys. Rev. B 99, 144401 (2019).
[70] D. Soriano, C. Cardoso, and J. Fernández-Rossier,
SolidState Commun. 299, 113662 (2019).
[71] L. Chen, J.-H. Chung, B. Gao, T. Chen, M. B. Stone, A.
I.Kolesnikov, Q. Huang, and P. Dai, Phys. Rev. X 8,
041028(2018).
[72] J. Cenker, B. Huang, N. Suri, P. Thijssen, A. Miller,T.
Song, T. Taniguchi, K. Watanabe, M. A. McGuire,D. Xiao, and X. Xu,
Nat. Phys. 17, 20 (2021).
[73] P. A. Maksimov and A. L. Chernyshev, Phys. Rev. Research2,
033011 (2020).
[74] L. Chen, J.-H. Chung, T. Chen, C. Duan, A. Schneidewind,I.
Radelytskyi, D. J. Voneshen, R. A. Ewings, M. B. Stone,A. I.
Kolesnikov, B. Winn, S. Chi, R. A. Mole, D. H. Yu,B. Gao, and P.
Dai, Phys. Rev. B 101, 134418 (2020).
[75] J. C. Slater and G. F. Koster, Phys. Rev. 94, 1498
(1954).[76] T. Holstein and H. Primakoff, Phys. Rev. 58, 1098
(1940).[77] J. Colpa, Physica A 93, 327 (1978).
APPENDIX
1. Energy levels of on-site Kanamori Hamiltonian
To obtain the n.n. spin interaction, we used the sec-ond order
perturbation theory, where the dominant inter-action is Eq. (1).
First we note that in the lowest energystate there are three
electrons at each metal site M. Theexchange processes then involve
one electron hopping be-tween M sites. Thus the intermediate states
have two elec-trons in t2g-orbitals on one M site and four
electrons ineither t2g- or eg-orbitals on the other M site. On the
otherhand the single-ion anisotropy would follow from a singleM
site, where the three electrons in the ground state inter-act with
an exited three electron state. The energy levelsof all states
involved in these exchange processes are givenin Table IV.
2. Ideal hoppings
In the ideal octahedron environment the allowed indirecthoppings
are show in Fig. 5. On the M1−X1 bond we have
Degen. Energy Microscopics
2 electrons (only in t2g)
1 EA,2,1 = U + 2JH + 2�d
2 EA,2,2 = U − JH + 2�d3 EA,2,3 = U
′ + JH + 2�d
9 EA,2,4 = U′ − JH + 2�d
3 electrons (GS)
4 EA,3,gs = 3U′ − 3JH + 3�d
3 electrons (only in t2g)
4 EA,3,1 = 3U′ + 3�d
6 EA,3,2 = U + 2U′ − 2JH + 3�d
6 EA,3,3 = U + 2U′ + 3�d
3 electrons (2-t2g, 1-eg)
8 EA,3,e1 = U + 2U′ − 2JH + ∆c + 3�d
4 EA,3,e2 = U + 2U′ + JH + ∆c + 3�d
24 EA,3,e3 = 3U′ − 3JH + ∆c + 3�d
24 EA,3,e4 = 3U′ + ∆c + 3�d
4 electrons (only in t2g)
1 EA,4,1 = 2U + 4U′ + 4�d
2 EA,4,2 = 2U + 4U′ − 3JH + 4�d
3 EA,4,3 = U + 5U′ − JH + 4�d
9 EA,4,4 = U + 5U′ − 3JH + 4�d
4 electrons (3-t2g, 1-eg)
4 EA,4,e1 = 6U′ + ∆c + 4�d
6 EA,4,e2 = U + 5U′ + ∆c + 4�d
10 EA,4,e3 = 6U′ − 6JH + ∆c + 4�d
18 EA,4,e4 = U + 5U′ − 4JH + ∆c + 4�d
18 EA,4,e5 = 6U′ − 2JH + ∆c + 4�d
24 EA,4,e6 = U + 5U′ − 2JH + ∆c + 4�d
TABLE IV. Spectrum of Hamiltonian Eq. (1). Listed states
cancontribute to the second order strong coupling perturbation.
the hopping matrix
px py pz
TM1X1 =
t1 0 0−t2 0 00 0 00 0 t00 t0 0
dx2−y2d3z2−r2dyzdxzdxy ,
(A.1)
The other indirect bonds TM2X2 , TM2X1 , TM1X2 are ob-tained
from TM1X2 by consecutively applying the sym-metry operation of the
octahedron C4(0, 0, 1). Use ofEq. (A.1) and symmetry related bonds
are then used inEq. (4) to arrive at effective hoppings Eq. (7) and
Eq. (9).Slater-Koster analysis[75] allows us to decompose t0, t1,
t2and t3 into σ- and π-bonding integrals between d- and p-
http://dx.doi.org/10.1038/s41567-019-0651-0http://dx.doi.org/10.1073/pnas.1902100116http://dx.doi.org/10.1073/pnas.1902100116http://dx.doi.org/10.1103/PhysRevB.47.558http://dx.doi.org/10.1103/PhysRevLett.77.3865http://dx.doi.org/10.1103/PhysRevLett.77.3865http://dx.doi.org/
10.1016/j.cpc.2007.11.016http://dx.doi.org/
10.1016/j.cpc.2007.11.016http://dx.doi.org/
10.1103/PhysRevMaterials.3.031001http://dx.doi.org/
10.1021/acs.nanolett.8b03321http://dx.doi.org/10.1103/PhysRevB.99.144401http://dx.doi.org/
https://doi.org/10.1016/j.ssc.2019.113662http://dx.doi.org/
https://doi.org/10.1016/j.ssc.2019.113662http://dx.doi.org/10.1103/PhysRevX.8.041028http://dx.doi.org/10.1103/PhysRevX.8.041028http://dx.doi.org/
10.1038/s41567-020-0999-1http://dx.doi.org/10.1103/PhysRevResearch.2.033011http://dx.doi.org/10.1103/PhysRevResearch.2.033011http://dx.doi.org/10.1103/PhysRevB.101.134418http://dx.doi.org/10.1103/PhysRev.94.1498http://dx.doi.org/10.1103/PhysRev.58.1098http://dx.doi.org/https://doi.org/10.1016/0378-4371(78)90160-7
-
11
Figure 5. Indirect hoppings from X site to M site, in the
globalcoordinates xyz.
orbitals
t0 = tpdπ, t1 =
√3tpdσ2
, t2 =tpdσ
2, t3 = tpdσ,
tpdσ < 0, tpdπ > 0, |tpdπ| < |tpdσ| .(A.2)
The direct hoppings Eq. (11) include δ-bonding in additionto σ-
and π-bondings
td1 =tddπ + tddδ
2, td2 =
−tddπ + tddδ2
td3 =3tddσ + tddδ
4, t̃d0 =
√3 (tddσ − tddδ)
4,
tddσ < 0, tddπ > 0, tddδ < 0,
|tddδ| < |tddπ| < |tddσ| ,
(A.3)
with the Anderse prediction tddσ : tddπ : tddδ = −6 : 4 :
−1.
3. Positions of X sites under distortions
In the R3̄ space group, octahedra made of two trianglesas shown
in Fig. 2 are rotated around ĉ-axis denoted by δxblue arrows.
Furthermore there is a slight change in thedistance between these
two triangles denoted by δx′ yellowarrows. The new positions of
ligands are parameterizedby dimensionless δx and δx′, in units of
the distance dMbetween M1 and M2. One can track the positions rm
ofXm, as it is related to the undistorted position ro,m, δxand
δx′:
rm = ro,m + δx dM (ξam â + ξ
bm b̂) + δx
′ dMζm ĉ
m ro,m(ξam, ξ
bm
)ζm
X1 (−dM√
20 0 ) ( −1 , −1 ) 1
X2 ( 0dM√
20 ) ( 1 , 1 ) −1
X3 ( 0−dM√
20 ) ( 1 , 0 ) 1
X4 ( 0 0dM√
2) ( −1 , 0 ) −1
X5 ( 0 0−dM√
2) ( 0 , 1 ) 1
X6 (dM√
20 0 ) ( 0 , −1 ) −1
,
(A.4)
where Xm lable sites as in Fig. 2(a) and â, b̂ and ĉ are
unitvectors along the lattice vectors of Fig. 2(b).
4. Distortion-induced hoppings
To get the distortion-induced hoppings one needs to findthe
directional cosines of the indirect d−p bonds using theligand
positions Eq. (A.4). With the metal site located atrM1 and ligand
Xm located at rm, the required directionalcosines follow from rm −
rM1 . They are the inputs in theSlater-Koster formula for rotated
bonds Ref. [75]. Here weonly show the leading term in each matrix
element gen-erated from the above procedure. Note that all
elementsbecome finite, and the distortion-induced hopping
betweent2g and p-oribtials are denoted by δti and eg and
p-orbitalsby δt′i:
TM1X1 =
t1 δt
′1 δt
′2
−t2 δt′3 δt′4δt1 δt2 δt3δt4 δt1 t0δt5 t0 δt1
, (A.5)
where X1 labels the site as in Fig. 2(a). The dis-torted
octahedron realizes the D3 point group, which con-tains C3(1, 1, 1)
and C
′2(−1, 1, 0) rotations. The hoppings
TM1X3 and TM1X5 are recovered by applying C3(1, 1, 1)to TM1X1 .
Further TM1X1 relates to TM1X2 by aC ′2(−1, 1, 0). Finally TM1X2
relates to TM1X4 and TM1X6by C3(1, 1, 1). The direct hopping
integrals denoted byTM1M2 is same the as Eq. (11). Making use of
Eq. (A.5)we derive effective hoppings TeffM1M2 following the
methodin the main text Eq. (4). The distortion induced
effectivehoppings to leading order in δt and listed in Table V.
5. Single-ion anisotropy term
Due to the distortion and SOC, the single-ion anisotropicterm is
also generated via the hopping to anions which canhop back to
create spin-dependent on-site terms denotedby TeffM1M1 = T
effM2M2
. Without the trigonal distortion,the effective hopping
integrals between t2g at M1 are given
-
12
TABLE V. Effective hoppings to leading order in
distortion-induced hoppings, where δt are defined in Eq. (A.5).
Distortion inducedhoppings are grouped in: δta = δt1 − δt4, δtb =
2δt1 + δt4, δtc = δt3 + δt5, δtd = δt1 − δt2, δte = δt1 + δt4, δtf
= δt1 + δt2.
TeffM1M2(t2g ⊗ t2g) = teff
TA TC TDT †C TA UTDU†T †D U
†T †DU TB
TeffM1M2(eg ⊗ t2g) = teff ( T̃A −UT̃AU† T̃CT̃B UT̃BU† T̃D)
U = eiπ 1√
2(σx2 −
σy2 ) =
i√2
(σx − σy)
TA =
(2δt3t0
+td1teff
)σo
TB =
(4δt5t0
+td3teff
)σo
TC =
(1 +
td2teff
)σo + i
r
2
δtat0
(σx + σy)
TD =δtbt0σo + i
r
2
(σx −
δtct0σy −
δtdt0σz
)
T̃A =
(δt′2t0
+δtdt1t20
)σo + i
r
2
((δt3t1t20− δt
′1
t0
)σx +
t1t0σy −
δtet1t20
σz
)T̃B =
(δt′4t0− δtf t2
t20
)σo + i
r
2
((δt3t2t20− δt
′3
t0
)σx −
t2t0σy −
δtat2t20
σz
)T̃C = i
r
2
((δt1t1t20− δt
′2
t0
)(σx + σy)− 2
(δt5t1t20− δt
′1
t0
)σz
)T̃D =
(−2 t2
t0+t̃d0teff
)σo + i
r
2
(δt1t2t20
+δt′4t0
)(σx − σy) .
TABLE VI. Spin model terms, under distortions, to leading order
in δt. Distortion-induced δt defined in Eq. (A.5), and
charactersubscripted δt defined in caption of Table V and Eq.
(A.8)
t2g ⊗ t2g eg ⊗ t2g
J =4(2t2d1 + 2(teff + td2)
2 + t2d3)
9 (U + 2JH)−
4JH(2teff(t2/t0)− t̃d0
)23 (∆c + U ′ − JH) (∆c + U ′ + 3JH)
+32teff
9(U + 2JH)
(td1
δt3t0
+ td3δt5t0
)K = − 4 (r teff)
2
9 (U + 2JH)
2 JH(r teff)2
3 (∆c + U ′ − JH) (∆c + U ′ + 3JH)t21 + t
22
t20
Γ = − 8(rteff)2
9 (U + 2JH)
(δtct0
)− 4JH(rteff)
2
3 (∆c + U ′ − JH) (∆c + U ′ + 3JH)
(t1(t1δt3 − t0δt′1)
t30− t2(t2δt3 − t0δt
′3)
t30
)Γ′ = − 4(rteff)
2
9 (U + 2JH)
(δtdt0
)− 2JH(rteff)
2
3 (∆c + U ′ − JH) (∆c + U ′ + 3JH)
(− t
21δtet30
+t22δtat30
)Ac = −
4(rteff)2
(U + 3JH − U ′)
(δtαt0
)− 4JH(rteff)
2
∆c (∆c + 3JH)
(t1t0
(δtβt1t20
+δt′αt0
)− t2 + t3
t0
(δtβt2 + 2δt1t3
t20+δt′βt0
))
by
TeffM1M1(t2g ⊗ t2g) =
teff
4σo irσz −irσy−irσz 4σo irσxirσy −irσx 4σo
. (A.6)Similarly, effective hopping between t2g and eg is found
as
TeffM1M1(eg ⊗ t2g) =
teff
irt1t0σx ir
t1t0σy −i2r
t1t0σz
irt2 + t3t0
σx −irt2 + t3t0
σy 02×2
.(A.7)
They lead to an Si ·Si term, which is just equal to S(S+1),an
irrelevant constant to the spin model. Thus there is no
single-ion anisotropy without the trigonal distortion. Withthe
trigonal distortion, the single-ion anisotropy along ĉ-axis is
generated:
Ac = −4(rteff)
2
(U + 3JH − U ′)
(δtαt0
)− 4JH(rteff)
2
∆c (∆c + 3JH)
(t1t0
(δtβt1t20
+δt′αt0
)− t2 + t3
t0
(δtβt2 + 2δt1t3
t20+δt′βt0
)),
(A.8)
where δtα = −2δt1 + δt2 + δt3 + δt4 + δt5, δtβ = δt2 + δt3,
δt′α = −δt′1 +√
3δt′4 − δt′22
, and δt′β = −δt′3 +√
3δt′2 + δt′4
2.
-
13
6. Spin interaction with trigonal distortion
Putting them all together, we have Heisenberg, Kitaev,Γ, Γ′, and
single-ion anisotropy Ac term. Their form toleading order in
distortion induced hoppings δt are summa-rized in Table VI bellow.
Note that the Kitaev interactiondoes not have any linear term of
distortion, and Γ, Γ′ andsingle-ion anisotropy Ac along ĉ-axis are
finite due to bothSOC and distortion.
7. Spin Wave Theory
Figure 6. X Y and Z bonds of the J −K − Γ− Γ′ − Ac shownin red,
green and blue respectively. Second n.n are shown inorange. The
orange arrows indicate when sgn(ij) = +1 in theDM term.
We consider the J−K−Γ−Γ′−Ac as well as the secondn.n. DM term
(Dc)
H =∑
〈i,j〉∈αβ(γ)
[JSi · Sj +KSγi S
γj + Γ
(Sαi S
βj + S
βi S
αj
)+Γ′
(Sαi S
γj + S
βi S
γj + S
γi S
αj + S
γi S
βj
) ]+∑〈〈i,j〉〉
Dc · (Si × Sj) +∑i
Ac (Si · ĉ)2 ,
(A.9)where Dc = Dc sgn(ij)ĉ and sgn(ij) = +1 when i to jpoints
along the orange arrows in Fig. 6. The standardHolstein-Primakoff
transformation [76] expanded to linearorder in S read
S+ =√
2S
(1− a
†a
2S
) 12
a '√
2Sa,
S− =√
2Sa†(
1− a†a
2S
) 12
'√
2Sa†,
Sz = S − a†a.
(A.10)
Using the above and Fourier transforming Eq. (A.9) leadsto
H = ECL +∑
k∈BZx†khkxk,
x†k = (a†k, b
†k, a−k, b−k),
hk =
ho−(k) h1(k) 0 h2(k)h1(k)∗ ho+(k) h2(−k) 00 h2(−k)∗ ho+(k)
h1(k)h2(k)
∗ 0 h1(k)∗ ho−(k)
,(A.11)
where the two species of bosons ak and bk correspond tothe two
sublattices of the unit cell. The h(k) terms are
ho±(k) = ho ± hDM (k),
ho = −S (2Ac + 2Γ + 4Γ′ + 3J +K) ,
hDM (k) = 2SDc (sin(a · k) + sin(b · k)− sin((a + b) · k)) ,
h1(k) = −S (Γ + 2Γ′ − 3J −K)
3(1 + e−ia·k + eib·k),
h2(k) =S (2Γ− 2Γ′ +K)
6
((1− i
√3)e−ia·k
+(1 + i√
3)eib·k − 2),(A.12)
where a and b are the lattice vectors in Fig. 6.
Followingstandard methods of diagonalizing BdG Hamiltonians[77]we
find the lowest eigenvalue around the Γ point in the BZ,and upon
expanding to orders of k, we get the spin gap ωoand spin stiffness
ρ:
ωk = ωo + ρk2
ωo = S |3Γ + 6Γ′ + 2Ac|
ρ =S
12
∣∣∣∣∣3J +K − Γ− 2Γ′− (K + 2Γ− 2Γ
′)2
2 (2Γ + 4Γ′ + 2Ac + 3J +K)
∣∣∣∣∣.(A.13)
At the K point in the BZ the Dirac gap is ωK+ − ωK− where
ωK+ = S{
(6Γ′ + 2Ac + 3J) (4Γ + 2Γ′ + 2Ac + 3J + 2K)
+6√
3Dc (2Γ + 4Γ′ + 2Ac + 3J +K) + 27D
2c
}1/2,
ωK− = S∣∣2Γ + 4Γ′ + 2Ac + 3J +K − 3√3Dc∣∣ .
(A.14)
8. Exact diagonalization calculations
The moment pinning calculations are carried out by find-ing the
ground state from exact diagonalization on an 8 site(2x2) honeycomb
cluster with periodic conditions, shownin Fig. 7.
-
14
Figure 7. X,Y and Z bonds show in red, green and blue
respec-tively. Dashed bonds represent the periodic boundary
condi-tions.
Magnetic Anisotropy in Spin-3/2 with Heavy Ligand in Honeycomb
Mott Insulators: Application to CrI3AbstractI IntroductionII
Kanamori interaction and tight binding HamiltonianIII Ideal
honeycomb structureA Superexchange path: indirect hopping1 t2g-t2g
contributions2 eg-t2g contributions
B Direct hoppingC Summary and Comments
IV Effects of distortion: distorted octahderaV Application to
CrI3VI Magnetic AnisotropyVII Spin gap and finite transition
temperatureVIII Summary and Discussion Acknowledgments References
APPENDIX1 Energy levels of on-site Kanamori Hamiltonian2 Ideal
hoppings3 Positions of X sites under distortions4
Distortion-induced hoppings5 Single-ion anisotropy term6 Spin
interaction with trigonal distortion7 Spin Wave Theory8 Exact
diagonalization calculations
