Arxiv:2012.07423V1 [Cond-Mat.Str-El] 14 Dec 2020

Twisted superﬂuid and supersolid phases of triplons in bilayer honeycomb magnets

Dhiman Bhowmick1 , Abhinava Chatterjee1,2,∗, Prasanta K. Panigrahi2, and Pinaki Sengupta1 1School of Physical and Mathematical Sciences, Nanyang Technological University, 21 Nanyang Link, Singapore 637371, Singapore and 2Department of Physical Sciences, Indian Institute of Science Education and Research Kolkata, Mohanpur - 741246, India ∗ (Dated: September 2, 2021) We demonstrate that low-lying triplon excitations in a bilayer Heisenberg antiferromagnet provide a promising avenue to realise magnetic analogs of twisted superfluid and supersolid phases that were recently reported for two-component ultracold atomic condensate in an optical lattice. Using a cluster Gutzwiller mean field theory, we establish that Dzyaloshinskii-Moriya interactions (DMI), that are common in many quantum magnets, stabilize these phases in magnetic system, in contrast to pair hopping process that is necessary for ultracold atoms. The critical value of DMI for transition to the twisted superfluid and twisted supersolid phases depends on the strength of the (frustrated) interlayer interactions that can be tuned by applying external pressure on and / or shearing force between the layers. Furthermore, we show that the strength of DMI can be controllably varied by coupling to tailored circularly polarized light. Our results provide crucial guidance for the experimental search of twisted superfluid and supersolid phases of triplons in real quantum magnets.

I. INTRODUCTION atoms [13, 14]. In this work, we show that twisted superfluid and twisted supersolid phases of magnons are The observation of twisted, multi-orbital superfluid in realized in a bilayer honeycomb Heisenberg model. In- binary mixtures of ultracold 87Rb atoms in two different terestingly, pair hopping process of magnons is not es- hyperfine states on a honeycomb optical lattice has at- sential for stabilizing these phases [7], in contrast to ul- tracted heightened interest in this novel quantum state of tracold atomic systems. Instead, a next nearest neighbor matter [1]. The twisted superfluid (or twisted supersolid) (NNN) Dzyaloshinskii-Moriya interaction (DMI) – which state is characterised by a complex order parameter – the is present in many quantum magnets – is sufficient to phase of the local superfluid order parameter at each site yield field induced twisted superfluid (TSF) and twisted changes continuously forming a“twisting pattern”, thus supersolid (TSS) phases over wide ranges of parameters. breaking time reversal symmetry spontaneously. Inter- estingly, complex order parameters have experimentally been shown to be associated with other novel strongly correlated phases such as the superconducting states of Sr2RuO4 [2,3] and UPt 3 [4] and the pseudo-gap state in the cuprate high-Tc superconductor, B-2212 [5,6]. A de- tailed understanding of the twisted superfluid state can help gain insight into these states as well. Subsequent Our paper is structured as follows. In Sec.II, we intro- theoretical studies have shown that the extended Bose duce the microscopic spin Hamiltonian and describe the Hubbard model with an additional pair hopping term can physics in the non-interacting limit by deriving the tight- stabilize a twisted superfluid (TSF) ground state over a binding triplon Hamiltonian and triplon band structure. finite range of parameters [7]. The cluster Gutzwiller mean field theory (CGMFT) is Quantum magnets have long served as a versatile plat- introduced in Sec.III which is used to calculate order- form for realizing magnonic analogs of complex bosonic parameters in interacting limit. In Sec.IV we present arXiv:2012.07423v2 [cond-mat.str-el] 1 Sep 2021 phases, often under less extreme conditions. For exam- the results of our study in the form of the order pa- ple, temperature needed for Bose Einstein condensation rameters and phase diagram as a function of magnetic (BEC) of magnons varies from a few Kelvins in many field, DMI and Heisenberg interactions. In Sec.V, we quantum magnets [8–10] to room temperature in Yttrium propose the possible materials in search of TSS and TSF Iron Garnet (YIG) thin films [11, 12], in contrast to nano- phases. Finally, in Sec.V, we show that circularly po- Kelvin temperature scales required for BEC in ultracold larized light can be used to induce DMI greater than the cutoff DMI required to realize TSS (TSF), whereas presence of frustration among interlayer and intralayer interactions is helpful to lower the cutoff value of DMI ∗ Present address: Department of Physics, Yale University, New (see also Appendix.C). The principal findings are sum- Haven, Connecticut 06520, USA marized in section Sec.VI. II THE BILAYER HONEYCOMB MAGNET AND EFFECTIVE TRIPLON MODEL

z Finally BzSi,m describes a Zeeman coupling between a local spin-moment and an external longitudinal magnetic field. It is noticeable that the change in sign of J and D do not alter the magnetic ground state as well as excitations above the ground state, whereas Jz and J⊥ are strictly positive in this study. For J⊥ |J|, the ground state of the system is a product of the singlet dimers on each interlayer NN bond. In this limit, the lowest excitations of the system are triplons, which are localized S = 1 quasipar- tices. An out-of-plane magnetic field lowers the energy of the Sz = +1 triplons and at a critical magnetic field, it crosses the energy of the singlet state, populating the ground state with a finite density of triplons. The other triplon branches (Sz = 0 and Sz = −1) are separated by a large energy gap. At low temperatures, one can restrict FIG. 1: (Color online) (a) The spins on each lattice site the local Hilbert space of the dimers to the singlet and interact via a strong interlayer anti-ferromagnetic cou- Sz = +1 triplon. By treating the triplons as bosonic pling, resulting in a ground state of a honeycomb lat- quasiparticles, one can formulate a description of the low tice of dimers on each interlayer nearest-neighbor bonds energy physics of the system in terms of hard core bosons (shown by dashed-black line). (b) The ferromagnetic which is known as bond-operator formalism (see also Ap- honeycomb lattice. (c) Triplon band structure at D = pendix.A). In this formalism, the zero field ground state 0.1J, (d) Triplon band structure at D = 0.8J. The made of singlets on each interlayer bonds is considered an other parameters for the band structure are J⊥ = 10J, empty lattice with number of triplons ni = 0 ∀i. At the Bz = 0.0, Jz = 0.0. critical field, a finite density of triplons is generated which increases with increasing field. The inter-dimer exchange interactions induce an effective hopping of the triplons. This delocalization induces a BEC of Sz = +1 triplons II. THE BILAYER HONEYCOMB MAGNET in the ground state. In the spin language, this corre- AND EFFECTIVE TRIPLON MODEL sponds to an canted antiferromagnetic order with a spontaneously broken U(1) symmetry. Considering singlets We start with a S = 1/2 Heisenberg antiferromagnet as a vacuum state in the system and triplon (Sz = +1) on a bilayer honeycomb lattice with out of plane exchange as a hard-core bosonic quasi-particle excitations in vac- anisotropy and Dzyaloshinkii-Moriya interaction (DMI), uum of singlets, we can use the bond operator formalism schematically shown in Fig.1(a) and described by the to express the bi-layer spin Hamiltonian Eq.1 as an ef- Hamiltonian, fective triplon Hamiltonian on a single-layer honeycomb lattice [15–17] (see Appendix.A), X X z H =J⊥ Si,m · Si,n − Bz Si,m i, m∈A i,m J X h i iD X h i H = tˆ†tˆ + H.c. + ν tˆ†tˆ − H.c. n∈B 2 i j 2 ij i j X hiji hhijii + J(Sx Sx + Sy Sy ) + J Sz Sz i,m j,m i,m j,m z i,m j,m J⊥ X Jz X hi,ji,m + − B tˆ†tˆ + nˆ nˆ (2) 4 z i i 2 i j X i + D νijzˆ · (Si,m × Sj,m). (1) hiji hhi,jii,m where, tî is the triplon annihilation operator,n î is the Si,m denotes the spin operator at i-th interlayer bond triplon number operator and index-i represents site in- at layer m (m ∈ {A, B}). h...i and hh...ii denote the dex of effective single-layer honeycomb lattice (which is nearest-neighbor (NN) and next-nearest-neighbor (NNN) equivalent to interlayer bond-index of bi-layer honeycomb inter-dimer bonds respectively. J⊥(> 0) is the strength lattice). The first two terms represent hopping of triplons of (isotropic) interlayer Heisenberg interaction, while between NN and NNN neighbor dimers respectively, the Jz (> 0) and J (> 0) denote the Ising and XX type third term is an on-site potential (effectively a chemical intralayer NN Heisenberg exchange interactions respec- potential) and the last term describes the effects of NN tively. D is the Dzyaloshinskii–Moriya interaction (DMI) interaction between triplons. It is noticeable that the which is constrained by the symmetry of the lattice to NNN hopping has a complex weight which renders the intralayer NNN bonds; νij = +1, if i → j forms part of a Hamiltonian unsuitable for quantum Monte Carlo simu- counterclockwise closed loop connecting the NNN sites in lations. When Jz ≈ 0, Eq.2 reduces to a tight binding a hexagonal plaquette in each layer (see counterclockwise model of non-interacting triplons. The Bloch Hamilto- circular arrows in Fig.1(a)-(b)) and νij = −1 otherwise. nian in the momentum basis, in terms of the momentum

2 II THE BILAYER HONEYCOMB MAGNET AND EFFECTIVE TRIPLON MODEL space triplon operators, is determined via Fourier trans- breaking of valley symmetry transforms the superfluid formation as, order-parameter from a real to a complex value and the resultant BEC is known as twisted superfluid [7]. Based X † H = Ψk [g(k)σ0 + h · σ]Ψk, (3) on this, the ground state on the honeycomb lattice is k no longer a bipartite lattice in twisted superfluid phase, but become a lattice with six-sublattices as shown in the ˆ T ˆ where, Ψk = (âk, bk) .a ˆk (bk) denotes the k-space figure Fig.2. Results based on CGMFT also supports triplon annihilation operator on sublattice-a (b) as shown this scenario as shown in section Sec.III. in Fig.1(a)-(b) and σ is the pseudo-vector of Pauli matrices and σ0 is the two-dimensional identity matrix. The coefficients of the σ-matrices in the Bloch Hamiltonian P are g(k) = (J⊥/4) − Bz, hx(k) = (J/2) cos(k · αi), P Pi hy(k) = (J/2) i sin(k · αi), hz(k) = D i sin(k · βi), where αi and βi are the NN and NNN vectors for each layer respectively. βi’s are chosen such that they form a counter-clockwise triangular loop for sites in sublattice- a in a hexagonal plaquette and clockwise triangular loop for sites in sublattice-b. The energy eigenvalues are given In absence of interaction Jz, Eq.2 transforms into by, well-known bosonic topological Haldane model. At finite temperatures due to non-zero magnetic excitations, E±(k) = g(k) ± |h(k)|. (4) this model is known to exhibit a finite thermal Hall effect due to presence of non-zero Berry-curvature of the The band dispersion is shown in Fig.1(c) and Fig.1(d) bands. On this basis it is expected that Bose-Einstein for two values of DMI. For D = 0, the energy spectrum is condensate at Γ-point or K (K0)-point would provide a identical to that of graphene, with a linear band crossing non-zero Hall conductance even at zero temperature [18]. of the upper and lower bands at the Dirac points K and However, we find that the Berry-curvature at the√ con- 0 K . A finite DMI√ breaks time reversal symmetry and densation momenta – Γ-point when√ |D| < |J|/ 3 and opens a band-gap 6 3D at these points. The energy of K (K0)-point when |D| > |J|/ 3 – is zero. Since the the lower band at Γ and K (K0)-points are respectively density of magnons is concentrated around the conden- given by (at Bz = 0), sation momentum, and is vanishingly small away from it, √ the thermal Hall response of the Bose-Einstein conden- J⊥ 3|J| J⊥ 3 3 sate is vanishingly small in our model. E = − ,E = − |D|. (5) Γ 4 2 K 4 2 In the absence of DMI, the energy minimum is located at the center of the Brillouin zone, the Γ point. For a finite but small DMI, the band minimum remains√ at Γ- point (Fig.1(c)). Increasing DMI to |D| > |J|/ 3 shifts the band minimum from Γ-point to two degenerate minima at the K and K0 (Fig.1(d)). Thus with changing DMI, the ground state changes from a one component BEC (condensation momentum k = 0) to a two compo- An analogous bosonic model can also be observed in 0 nent BEC (condensation momenta at√k = K and K ). terms of Matsubara-Matsuda bosons in magnetically or- The transition happens at |D| = |J|/ 3 independent of dered honeycomb ferromagnets (see Fig.1(b) ) or antifer- J⊥ and Bz for small Bz. romagnets. Thus the non-trivial phases like twisted- In presence of repulsive interaction between superfluid or twisted-supersolid are also expected to triplons ( the last term in Eq.2 ) co-existence of triplons emerge in magnetically ordered systems. However, in at K and K0 points costs no additional energy classi- practice, the presence of anisotropies in magnetically or- cally ( see Appendix.B ) , but the quantum-fluctuations dered systems break the U(1)-symmetries destroying con- around K and K0 points introduces an energy cost [18]. servation of numbers of particles which in-turn preclude Thus spontaneous breaking of valley-symmetry is en- a long time superfluidity response in this systems [15, 21]. ergetically favoured and the quantum fluctuation will Moreover due to presence of strong exchange interac- lead to a ground state with only single valley conden- tions (J, Jz in Eq.2) in general, the phase transitions are sation which is known as “quantum order by disorder” also difficult to study varying the external parameters effect [18–20]. Thus, in the limit of weak interaction, in these systems [21]. That is why the dimerized para- the super-fluid order parameter at lattice site r is either magnets are well suited to study superfluid phases and 0 |b|eiK·r+φ or |b|eiK ·r+φ depending on the valley for the superfluid-Mott transitions. Hence we focus on bilayer Bose-Einstein condensate, where φ is a global phase dimerized honeycomb paramagnetic system to study the independent of position of lattice site. This spontaneous twisted superfluid and twisted supersolid phases.

3 III CLUSTER-GUTZWILLER MEAN FIELD THEORY (CGMFT) AND OBSERVABLES

III. CLUSTER-GUTZWILLER MEAN FIELD sites of the cluster and the mean-ﬁeld region. The form THEORY (CGMFT) AND OBSERVABLES of the boundary Hamiltonian is given by,

J X0 h i iD X0 h i H = tˆ† tˆ + H.c. + tˆ† tˆ − H.c. δC 2 i j 2 i j hi,ji hhi,jii

Jz X0 + nˆ hnˆ i , (7) 2 i j hi,ji

where the primed summations are over the boundary site-

i connected to the mean-field site-j. tˆj and hnˆji are two mean-field parameters denoting the superfluid order parameter and occupation number of triplons at site-j respectively. We choose six inequivalent sites in each cluster (denoted by different patterns in figure Fig.2) to give a total of 12 mean-field parameters. The ground state in FIG. 2: (Color online) The cluster construction for the different parameter regimes are obtained by evaluat- CGMFT. There are 18 sites in the cluster which is lo- ing these mean field parameters self-consistently in the cated within the dashed black box. A periodic boundary following manner, condition is applied along the horizontal direction and (i) Choose an initial set of mean field parameters the mean-field boundary condition is applied along the tˆj , hnˆji , j = 1,..., 6 and then exactly diago- vertical direction. The background of the cluster sites eff is denoted by a pink shade and the background of the nalize the effective Hamiltonian of the cluster HC . mean-field sites are denoted by blue shade. ˆ0 0 (ii) Calculate new mean field parameters tj and nˆj from the sites within blue-dashed rectangle in the The cluster Gutzwiller mean field theory or figure Fig.2 which reside within the cluster. Peri- CGMFT [7, 22] – equivalently, cluster mean field odic boundary condition is chosen along horizontal theory [23–29], self-consistent cluster mean field the- direction to eliminate any boundary effect on the ory [30], multi-site mean field theory [31], hierarchical sites within the blue-dashed rectangle, so that the mean field approach [32, 33], composite boson mean mean field parameters obtained from those sites are field theory [34] – is a powerful technique to study free from boundary effects. superfluid phases in bosonic many body systems with complex hopping terms. CGMFT improves over the (iii) The initial and final set of mean field parameters conventional single-site mean field approach by taking are compared using the tolerance into account the short range correlations present within X ˆ0 ˆ X 0 a small lattice-cluster using exact-diagonalization. = tj − tj + nˆj − hnˆji . (8) Furthermore it is an alternative numerical method j j to study the quantum systems like we described in section Sec.II, where sign problems arises in quantum If the tolerance is less than a certain cutoff then Monte-Carlo methods due to complex hopping terms the obtained mean-field parameters correspond to or geometric frustration [35, 36]. Whereas conventional the ground state of the system. Otherwise the step- mean-field theories [37] and exact-diagonalization of (i) is repeated with new values of mean-field param- ˆ ˆ0 0 small systems [38] fail to show existence of TSF and eters tj = tj and hnˆji = nˆj . TSS phases, Density Matrix Renormalization Group for −10 one dimensional systems [39] and CGMFT for higher We set the cutoff as 10 and start the simulation with dimensional systems [7] are better alternatives for search different initial mean-field parameter sets for a fixed set of these non-trivial phases. of parameters J, D, J⊥, Bz, and Jz. In general, the We explore the ground state phases of the effective simulations with different initial mean field parameter triplon Hamiltonian with CGMFT by decomposing the sets give different ground-states at the boundary of two system into clusters (pink shaded region) and mean-field phases and we selected the phase with minimum energy region (blue shaded region) as shown in figure Fig.2. The as the ground state. effective mean-field Hamiltonian of the cluster is given as, After obtaining the ground state via self-consistent de- termination of the mean field parameters, four order- eff HC = HC + HδC , (6) parameters are calculated to identify the nature of the ground state phase of the system. The magnitude of su- where, HC is the Hamiltonian as in equation Eq.(2) perfluid order parameter is given by, within the cluster and HδC is the Hamiltonian which takes into account the interactions among the boundary |b| = max | tˆj | , (9)

4 IV NUMERICAL RESULTS

where max denotes maximum value of the parameter ob- (c)). The zero field (Bz = 0) ground state corresponds tained out of six-sites within the dashed blue-border in to a singlet phase, or equivalently an empty lattice in Fig.2. Additioally, the average number of particles per the bosonic language. All the order parameters vanish site hniav and difference in number of particles between in this limit. This remains true at small values of the NN sites ∆n are also enumerated. The superfluid order- applied field reflecting a finite gap to lowest excitations parameter is a complex quantity and for twisted super- due to the singlet-triplet gap of the local dimers. When fluid phases in our study the phase difference of super- the applied field exceeds a critical value, the gap is closed fluid order parameter b among NNN sites is obtained to and the ground state acquires a finite density of triplons. be θ = 120o and otherwise θ = 0o. These field induced triplons form a superfluid (SF) driven by the NN triplon hopping and is characterised by a finite SF order parameter, |b|. The mismatch between the IV. NUMERICAL RESULTS occupancy of the two sublattices (∆n) remains zero, reflecting the uniform nature of the SF phase. A vanishing twist angle (θ = 0) completes the characterization of the phase as a normal superfluid. In the weak interaction limit (Jz < 2J) with increasing magnetic field, the density of triplons increases monotonically till full saturation is reached at an upper critical field when each dimer is occupied by a triplon. At saturation, all the order parameters (except average density, hniav, of triplons) vanish denoting a lattice fully occupied by triplons. In this weak interaction limit, the physics is similar to the non- interacting limit as described in sectionII and so the qual- itative feature can be well described using band structure as in figure Fig.1(c).

For strong interactions (Jz > 2J), an intervening charge density wave (CDW) phase, driven by the strong NN-interaction between triplons, appears in addition to the phases discussed above (see Fig.3(c)). With increasing magnetic field, when the density of triplons reaches FIG. 3: (Color online) The order parameters are plot- hniav = 1/2, the triplons form a staggered CDW pattern ted for parameter values (a) D = 0.2J, Jz = J, (b) where one of the sublattices is fully occupied, while the D = 0.9J, Jz = J, (c) D = 0.2J, Jz = 4J, (d) other remains empty. The potential energy-cost due to D = 0.9J, Jz = 4.5J. J⊥ is fixed at value 10J. Order- nearest neighbor interaction is minimized as there are no parameters |b|, hniav and ∆n are plotted as function of nearest neighbor pairs. This is accompanied by a com- magnetic field Bz and denoted by red, black and blue plete quenching of superfluidity, since any hopping of dotted lines respectively. The dots on the lines denote triplons will necessarily involve configurations with en- the points in parameter space where the CGMFT is per- ergetically costly multiple nearest neighbor pairs. The formed and the lines just connect the points. More- CDW phase has a finite gap to the addition of any more over the phase difference of super-fluid order parameter triplons and the density remains constant at hniav = 1/2 θ is shown in the right-side vertical-axis and denoted in over a finite rage of applied field. This phase is charac- green colour. Different phases are indicated by differ- terized by a vanishing superfluid order, and a non-zero ent coloured shades. “E” and “O” denote empty and density mismatch between the two sublattices (∆n), re- fully-occupied phase of the system respectively. All other flecting the staggered order. When the increasing field phases are described in the main text. strength reaches a critical value where the Zeeman energy gain due to increasing magnetization (equivalently, Using CGMFT, we determine the order parameters |b|, adding more triplons) exceeds the potential energy cost of nearest neighbor repulsion, the density of triplons starts hbi , ∆n and θ as a function of magnetic field Bz for av to increase again, resulting in another normal SF phase. different sets of the parameters (D,Jz). The evolution of the order parameters and the resulting field driven phases Finally, as the field is increased above a saturation value, are shown in Figure fig.3 for four illustrative points of Bsat, the ground state enters the fully polarized phase. the (D,Jz) parameter space. In fig.3, the DMI increases The above argument for the appearance of interaction from the left-column of figures to the right-column of driven CDW phase at half-filling does not apply for weak figures, whereas the interaction Jz increases from upper- to moderate interaction strengths (Jz < 2J), as the ki- row of figures towards the lower-row of figures. netic energy gain due to the delocalization of triplons For weak DMI (D = 0.2J), the field driven phase di- exceeds the potential energy cost of NN-interactions. agram resembles that of the canonical extended Bose The sequence of field-driven phase changes markedly Hubbard model for hard core bosons [40] (see Fig.3(a), for strong DMI. As shown earlier in Sec.II, in the non-

5 V MATERIAL REALIZATION interacting limit, the triplon band minimum shifts from the Γ point to the K and K0 (Fig.1(d)) and the BEC of triplons occur at finite momentum. A local minimum persists at the center of the Brillouin zone, and the energy gap between the triplon-sector and the singlet dimer sector, EK in Eq.5 decreases with increasing DMI. This behavior persists in the presence of weak to moderate interaction (Jz < 2J) and is reflected in Fig.3(b). For the present choice of parameters, the energy of the lowest triplon excitation is vanishingly small. The triplon FIG. 4: (Color online) Phase diagram at two different in- density acquires a finite value for an infinitesimally small teraction values (a) J = J and (b) J = 4.5J. Each dot B , and increases monotonically with the strength of the z z z or circle denotes the parameter point where the CGMFT applied field. In this regime, the triplons form a super- is performed. Different color denotes different phases in fluid (|b| > 0). More interestingly, the complex NNN hop- parameter space. The empty and dotted black circles ping process imparts a complex phase to the superfluid represent the empty and fully occupied lattice phases, order parameter, as seen by a finite expectation value of respectively. the twist angle (θ 6= 0). In other words, the ground state in this parameter range is a twisted superfluid (TSF). The triplon density increases monotonically, with the The phase diagram in Bz −D parameter space is shown ground state remaining a TSF, till the fully polarized in the figure Fig.4 for two different values of interaction phase is reached at a saturation field, Bsat. Jz = J and Jz = 4.5J. A comparison of the phase di- agrams at moderate (Jz < 2J) and strong (Jz > 2J) interactions reveal, (i) Appearance of CDW phase at half-filling in the strong interaction limit. The CDW phase appears at half filling dividing the SF-region which appear at weak to moderate Jz into two SF-regions. (ii) In the strong interaction limit the TSF phase is replaced by a TSS phase.

V. MATERIAL REALIZATION

There are two main ingredients to realize TSF or TSS phases; firstly, a bilayer honeycomb valence bond state is required; secondly, a DMI greater than a critical value is required. Although, to the best of our knowledge, there Finally in the strong DMI and strong interaction are no materials described in the literature which satisfies limit (see Fig.3(d)),the twisted superfluid is replaced by our model, a rough idea is sketched here to obtain TSS or a twisted supersolid phase, in addition to the appear- TSF phases based on the real materials Bi3Mn4O12(NO3) ance of an interaction-driven CDW phase at hniav = 1/2 and CrI3. 4+ over a finite range of applied field. In the twisted su- The material Bi3Mn4O12(NO3) consists of Mn ions 3 persolid phase (TSS), the ground state is characterized carrying spin S = 2 arranged in a bilayer honeycomb lat- by a finite ∆n (density mismatch between the two sub- tice with A-A stacking. No magnetic ordering is observed lattices), in addition to a complex superfluid order pa- down to the lowest temperatures. Theoretical studies in- rameter (|b|= 6 0, θ 6= 0,). The finite density difference dicate that the disordered magnetic state may be an in- between two sublattices provide the diagonal order con- terlayer dimer phase which is adiabatically connected to currently with the finite (twisted) superfluid ordering. the direct products of singlets [41–45], although a spin liq- It is surprising that the ground state exhibits TSS or- uid phase cannot be completely ruled out [46–57]. If the der even at low triplon densities. This is understood ground state phase is a spin liquid, an interlayer dimer by recalling that the primary delocalization process in phase can be induced by applying a pressure along the this parameter regime involves the DMI-induced intra- perpendicular direction [44, 51]. sublattice complex next-nearest-neighbor hopping. The On the other hand, experimentally well studied van strong NN-repulsion between the triplons further sup- der Waals material CrI3 is a honeycomb ferromagnet 2+ 1 presses inter-sublattice hopping processes, resulting in a where Cr -ions carry spin- 2 momentum forming a hon- preferential occupation of one of the two sublattices at eycomb lattice. This material has been well studied by small densities. tuning the number of layers [58] as well as varying the

6 A BOND-OPERATOR FORMALISM pressure [59]. It is shown that the interlayer antiferro- TSF phase is induced by DMI greater than a criti- magnetic Heisenberg exchange interaction changes lin- cal value. For Ising-like anisotropy of the intra-plane early with the distance between two layers by applica- Heisenberg interactions, the TSF phase is replaced by tion of pressure on the material and can be achieved a a twisted supersolid (TSS) phase. While the strength interlayer coupling twice as compared with the initial in- of DMI required for the stabilization of TSF and TSS terlayer coupling [59]. Although a valence bond state is phases (D/J & 0.5) is not observed natively in most not detected in CrI3 (and there are indications of struc- quantum magnets, recent experiments have shown that tural phase transition under pressure) [59], the pressure a strong DMI can be induced in thin films of insulating induced high interlayer coupling in CrI3 motivates for magnets, by forming heterostructures with heavy met- searching of pressure induced valance bond states in a als (with strong spin-orbit coupling) [69–72]. Our results family of Van der Walls honeycomb magnets CrBr3 [60], show that circularly polarized light can also induce large CrGeTe3 [61], CrSiTe3 [62], FePS3 [63], NiPS3 [64]. DMI by varying the frequency of light [67]. Finally, the Another challenge for achieving the√ TSF and TSS presence of frustration among interlayer and intralayer phases is to obtain a high DMI (|D| > J/ 3). The mate- Heisenberg exchange interactions in a material can lower rials Bi3Mn4O12(NO3) and CrI3 are experimentally pre- the value of the critical DMI required to realize TSF dicted to possess DMI [65, 66]. However the magnitude of and TSS phases (Appendix.C), thus facilitating their ex- DMI is small for realizations TSS and TSF states. This perimental observation. We propose that the material problem can be overcome by application of circularly po- Bi3Mn4O12(NO3)[41–45] and family of honeycomb mag- larized light. A circularly polarized light couples either to nets CrBr3 [60], CrGeTe3 [61], CrSiTe3 [62], FePS3 [63], the charge [67] or to the magnetic degrees of freedom [68] NiPS3 [64] are promising candidate materials to realize and it has been shown theoretically that both kinds of TSF and TSS phases. mechanisms give rise to scalar spin chiral interaction in a honeycomb magnetic insulator. The synthetic scalar spin chiral interaction as in reference Ref. [67] given by, ACKNOWLEDGEMENT X Hχ = χ νjkSî,m · Sˆj,m × Sˆk,m It is a pleasure to thank Oleg Sushkov for useful dis- hhj,kii,m cussions. Financial support from the Ministry of Educa- 3iχ X h i tion, Singapore, in the form of grant MOE2018-T1-001- ≈ ν hnˆ i tˆ†tˆ − H.c. , (10) 4 jk i j k 021 is gratefully acknowledged. A. Chatterjee acknowl- hhj,kii edges the financial support received from INSPIRE, De- partment of Science and Technology, Govt. of India and where the i-th site is the NNN neighbor of both j- the NTU-India Connect Program. A. Chatterjee would th and k-th sites, χ is synthetic scalar spin chirality. like to thank the School of Physical and Mathematical Comparing the equations Eq.2 and Eq. 10, the effective Sciences at NTU for their kind hospitality. DMI for bond hhjkii due to circularly polarized light is 3χhnii D = 2 . The scalar spin chirality χ is shown to have a 1 U resonance for the frequency of light near ωn = n , where Appendix A: Bond-operator formalism U is onsite interaction representing electron-electron~ repulsion and n is a positive integer. Thus the effective The spin-operators in terms of bond-operators are DMI can be tuned as high as possible tuning the fre- given by [17], quency of the light nearby ωn. However the results in Ref. [67] is based on single-band extended Hubbard † † † † tˆ tˆ + tˆ tˆ¯ sˆ tˆ¯ − tˆ sˆ ˆ+ 1,j 0,j 0,j 1,j j 1j 1,j j model; for a magnetic material which cannot be described Sj,l = √ ± √ by a one band Hubbard model, one needs a more careful 2 2 theoretical treatment. On the other hand, application of tˆ† tˆ + tˆ† tˆ sˆ†tˆ − tˆ† sˆ − 1¯,j 0,j 0,j 1,j j 1,j 1¯,j j light may not be required for a real material if the cutoff Sˆ = √ ∓ √ j,l 2 2 of DMI to realize TSF or TSS states is reduced due to ˆ† ˆ ˆ† ˆ † † presence of frustration among interlayer and intralayer t1,jt1,j − t¯ t1¯,j sˆ tˆ0,j + tˆ sˆj Sˆz = 1,j ± j 0,j , (A1) interactions (see Appendix.C). j,l 2 2 where the upper-sign is for layer l =A and the lower † † † VI. CONCLUSION sign is for the layer l =B. tˆ , tˆ and tˆ create states √1,j 1¯,j 0,j |↑↑i, |↓↓i and (|↑↓i + |↓↑i)/ 2 on j-th bond respectively. To summarize, we have shown that a magnetic ana- After the bond operator transformation, the bilayer hon- log of the novel twisted superfluid (TSF) state reported eycomb spin system transforms into an effective single in recent experiments with ultracold atoms in an opti- layer honeycomb lattice system with triplon and singlet cal lattice [1] can be realized in a bilayer quantum anti- operators on each site. At low temperature, in the limit ferromagnet with realistic interactions. We show that |J⊥| |J|, it can be assumed that the ground state is

7 B INTERACTION TERM IN RECIPROCAL SPACE

ˆ† product of singlets on the interlayer NN-bonds. Assum- where the subscript-1 is omitted from the operator t1,j. ing singlets form the empty or vacuum state and taking advantage of the hardcore nature of the bosons, we can simply transform the following quadratic operators into Appendix B: Interaction term in reciprocal space a single triplon operators in equation Eq. A1, The last term in equation Eq.2 represents the NN re- †ˆ ˆ ˆ† ˆ† sˆjtαj → tαj, tαjsˆj → tαj. pulsive interaction and in reciprocal space it is given by,

Moreover application of magnetic ﬁeld in the z-direction Jz X X i(k−k0)·α † † H = e tˆ tˆ 0 tˆ 00 tˆ 00 0 . (B1) lowers the energy of triplon correspond to the operator V 2 k k k k+k −k ˆ† α k,k0,k00 t1,j and so neglecting all other triplon operators in equation Eq. A1, we get the following bond operator transfor- Considering triplons are only present at K and K0 points mation [15], as well as neglecting quantum ﬂuctuations around these points, the interaction Hamiltonian can be explicitly + 1 † − 1 z 1 † Sˆ = ∓√ tˆ , Sˆ = ∓√ tˆj, Sˆ = tˆ tˆj, written as, j,l 2 j j,l 2 j,l 2 j

3Jz h † † † † † † † † i H = tˆ tˆ tˆ tˆ + tˆ 0 tˆ 0 tˆ 0 tˆ 0 + tˆ tˆ tˆ 0 tˆ 0 + tˆ 0 tˆ 0 tˆ tˆ V 2 K K K K K K K K K K K K K K K K Jz X h i(K−K0)·α † † i(K0−K)·α † † i + e tˆ tˆ 0 tˆ 0 tˆ + e tˆ 0 tˆ tˆ tˆ 0 2 K K K K K K K K α 3Jz 2 2 = nˆ +n ˆ 0 + 2ˆn nˆ 0 2 K K K K 3Jz 2 = (ˆn +n ˆ 0 ) , (B2) 2 K K

thus coexistence of particles at K and K0 points does Appendix C: Introducing more terms in the spin not seem to increase the energy of the system. However, Hamiltonian quantum ﬂuctuations around the points K and K0 increases the energy due to presence of triplons at both K In this appendix, we have taken into account additional and K0 points as shown in the reference Ref. [18]. interlayer bonds (see Fig.5(a), (b)). We introduce intralayer NNN Heisenberg-exchange interaction J1 as well as interlayer NN Heisenberg exchange interaction J2⊥ in the spin Hamiltonian in equation Eq.1,

X X z X x x y y z z H =J⊥ Si,m · Si,n − Bz Si,m + J(Si,mSj,m + Si,mSj,m) + JzSi,mSj,m i, m∈A i,m hi,ji,m n∈B X X + D νijzˆ · (Si,m × Sj,m) + D⊥ νijzˆ · (Si,m × Sj,n) hhi,jii,m hhi,jii m∈A,B, n6=m X X X + J1 Si,m · Si,m + J2,⊥ Si,m · Si,n + J3,⊥ Si,m · Si,n. (C1) hhi,jii,m hi,ji hhi,jii m∈A,B, n6=m m∈A,B, n6=m

Moreover we have added symmetry allowed interlayer NNN DMI D⊥ and Heisenberg interaction J3⊥. The DMI on dimer-bond and NN interlayer bonds are zero due to presence of inversion center at the middle of the bonds. The

8 B INTERACTION TERM IN RECIPROCAL SPACE corresponding real space triplon Hamiltonian is given as,

J − J2⊥ X h i J1 − J3⊥ X h i i(D + D⊥) X h i H = tˆ†tˆ + H.c. + tˆ†tˆ + H.c. + ν tˆ†tˆ − H.c. 2 i j 4 i j 2 ij i j hiji hhijii hhijii J⊥ X J2⊥ + Jz X J1 + J3⊥ X + − B tˆ†tˆ + nˆ nˆ + nˆ nˆ (C2) 4 z i i 2 i j 4 i j i hiji hhijii

It is noticeable that the Heisenberg interaction J2⊥ renor- interactions. The conditions J ≈ J2⊥ and J1 ≈ J3⊥ malizes the interactions J and Jz, whereas J1 and J3⊥ lowers the critical value of the DMI to realize TSF or add additional NNN hopping term and NNN interaction TSS phases. Thus we can conclude that frustration term. Moreover D⊥ renormalizes D by simply adding up among interlayer and intralayer Heisenberg interactions and so D⊥ is absorbed into D in the rest of this section. can lower the critical DMI required to realize TSF and In this appendix, we investigate the triplon Hamiltonian, TSS phases. neglecting the interaction terms. The non-interacting Hamiltonian in k-space is similar to equation Eq.3 and given by,

X † 0 0 H0 = Ψk [g (k)σ0 + h · σ]Ψk, (C3) k where,

0 X g (k) = (J⊥/4) − Bz + ((J1 − J3⊥)/2) cos(k · βi), i 0 X hx(k) = ((J − J2⊥)/2) cos(k · αi), i 0 X hy(k) = ((J − J2⊥)/2) sin(k · αi), i 0 X hz(k) = D sin(k · βi) i FIG. 5: (Color online) (a) Intradimer and interlayer NN , where αi and βi are the NN and NNN vectors for each bonds are indicated by red solid and red dotted lines, re- layer respectively. The band structure for two different spectively. (b) The intralayer and interlayer NNN bonds D-values are plotted in the figure Fig.5(c) and (d). The are indicated by green solid and green dotted lines, re- 0 minima of the bands are at the Γ-point and K (or K )- spectively. (c) Triplon band structure at D = 0.1J, (b) point for D = 0.1J and D = 0.8J respectively. The Triplon band structure at D = 0.8J. The other param- results are same when J2⊥ = 0 and J1 = 0 as in the eters for the band structures are J⊥ = 10J, J1 = 0.8J, figures Fig.1(c) and (d). The condition for the band J2⊥ = 3J, Bz = 0.0, Jz = 0.0, J3⊥ = 0. minima at K or K0-point is given as,

2|J − J | − 3(J − J ) |D| > 2⊥ √ 1 3⊥ . (C4) 2 3 We note that the critical DMI required to realise TSF or TSS phases depends on various Heisenberg exchange

[1] P. Soltan-Panahi, D.-S. L¨uhmann,J. Struck, P. Wind- [3] K. Wysokinski, Condensed Matter 4, 47 (2019). passinger, and K. Sengstock, Nature Physics 8, 71–75 [4] G. M. Luke, A. Keren, L. P. Le, W. D. Wu, Y. J. Uemura, (2011). D. A. Bonn, L. Taillefer, and J. D. Garrett, Phys. Rev. [2] G. M. Luke, Y. Fudamoto, K. M. Kojima, M. I. Larkin, Lett. 71, 1466 (1993). J. Merrin, B. Nachumi, Y. J. Uemura, Y. Maeno, Z. Q. [5] R. P. Kaur and D. F. Agterberg, Phys. Rev. B 68, Mao, Y. Mori, and et al., Nature 394, 558–561 (1998). 100506(R) (2003).

9 B INTERACTION TERM IN RECIPROCAL SPACE

[6] A. Kaminski, S. Rosenkranz, H. Fretwell, J. C. Cam- [36] D. Huerga, J. Dukelsky, N. Laflorencie, and G. Ortiz, puzano, Z. Li, H. Raffy, W. Cullen, H. You, C. Olson, Phys. Rev. B 89, 094401 (2014). C. Varma, and H. Höchst, Nature 416, 610 (2002). [37] S. Choudhury and E. J. Mueller, Phys. Rev. A 87, 033621 [7] O. Jürgensen,K. Sengstock, and D.-S. Lühmann, Scien- (2013). tific Reports 5, 12912 (2015). [38] L. Cao, S. Krönke, J. Stockhofe, J. Simonet, K. Seng- [8] H. Tanaka, A. Oosawa, T. Kato, H. Uekusa, Y. Ohashi, stock, D.-S. Lühmann, and P. Schmelcher, Phys. Rev. A K. Kakurai, and A. Hoser, Journal of the Physical Soci- 91, 043639 (2015). ety of Japan 70, 939 (2001). [39] D.-S. Lühmann, Phys. Rev. A 94, 011603(R) (2016). [9] R. Coldea, D. A. Tennant, and Z. Tylczynski, Phys. Rev. [40] J. Y. Gan, Y. C. Wen, J. Ye, T. Li, S.-J. Yang, and B 68, 134424 (2003). Y. Yu, Phys. Rev. B 75, 214509 (2007). [10] Y. Singh and D. C. Johnston, Phys. Rev. B 76, 012407 [41] H. Zhang, C. A. Lamas, M. Arlego, and W. Brenig, (2007). Physical Review B 93, 235150 (2016). [11] S. Demokritov, V. Demidov, O. Dzyapko, G. Melkov, [42] H. C. Kandpal and J. van den Brink, Phys. Rev. B 83, A. Serga, B. Hillebrands, and A. Slavin, Nature 443, 140412(R) (2011). 430 (2006). [43] H. Zhang, M. Arlego, and C. A. Lamas, Phys. Rev. B [12] O. Dzyapko, V. Demidov, S. Demokritov, G. Melkov, 89, 024403 (2014). and A. Slavin, New Journal of Physics 9, 64 (2007). [44] J. Oitmaa and R. R. P. Singh, Phys. Rev. B 85, 014428 [13] C. C. Bradley, C. A. Sackett, J. J. Tollett, and R. G. (2012). Hulet, Phys. Rev. Lett. 75, 1687 (1995). [45] R. Ganesh, S. V. Isakov, and A. Paramekanti, Phys. [14] K. B. Davis, M. O. Mewes, M. R. Andrews, N. J. van Rev. B 84, 214412 (2011). Druten, D. S. Durfee, D. M. Kurn, and W. Ketterle, [46] A. Mulder, R. Ganesh, L. Capriotti, and Phys. Rev. Lett. 75, 3969 (1995). A. Paramekanti, Phys. Rev. B 81, 214419 (2010). [15] V. Zapf, M. Jaime, and C. D. Batista, Rev. Mod. Phys. [47] F. Wang, Phys. Rev. B 82, 024419 (2010). 86, 563 (2014). [48] S. Okumura, H. Kawamura, T. Okubo, and Y. Mo- [16] S. Sachdev and R. N. Bhatt, Phys. Rev. B 41, 9323 tome, Journal of the Physical Society of Japan 79, 114705 (1990). (2010). [17] J. Romhányi, K. Totsuka, and K. Penc, Phys. Rev. B [49] B. K. Clark, D. A. Abanin, and S. L. Sondhi, Phys. Rev. 83, 024413 (2011). Lett. 107, 087204 (2011). [18] X. Li, S. S. Natu, A. Paramekanti, and S. D. Sarma, [50] D. C. Cabra, C. A. Lamas, and H. D. Rosales, Phys. Nature Communications 5, 5174 (2014). Rev. B 83, 094506 (2011). [19] S. K. Adhikari, B. A. Malomed, L. Salasnich, and [51] R. Ganesh, D. N. Sheng, Y.-J. Kim, and A. Paramekanti, F. Toigo, Phys. Rev. A 81, 053630 (2010). Phys. Rev. B 83, 144414 (2011). [20] C. L. Henley, Phys. Rev. Lett. 62, 2056 (1989). [52] A. F. Albuquerque, D. Schwandt, B. Hetényi, S. Cap- [21] T. Giamarchi, C. Rüegg, and O. Tchernyshyov, Nature poni, M. Mambrini, and A. M. Läuchli, Phys. Rev. B Physics 4, 198–204 (2008). 84, 024406 (2011). [22] D.-S. Lühmann, Phys. Rev. A 87, 043619 (2013). [53] D. C. CABRA, C. A. LAMAS, and H. D. ROSALES, [23] Y.-C. Chen and M.-F. Yang, Journal of Physics Commu- Modern Physics Letters B 25, 891 (2011). nications 1, 035009 (2017). [54] F. Mezzacapo and M. Boninsegni, Phys. Rev. B 85, [24] M. P. Gelfand, R. R. P. Singh, and D. A. Huse, Phys. 060402(R) (2012). Rev. B 40, 10801 (1989). [55] P. H. Y. Li, R. F. Bishop, D. J. J. Farnell, and C. E. [25] D. Yamamoto, G. Marmorini, and I. Danshita, Phys. Campbell, Phys. Rev. B 86, 144404 (2012). Rev. Lett. 112, 127203 (2014). [56] R. Ganesh, J. van den Brink, and S. Nishimoto, Phys. [26] M. Moreno-Cardoner, H. Perrin, S. Paganelli, Rev. Lett. 110, 127203 (2013). G. De Chiara, and A. Sanpera, Phys. Rev. B 90, [57] S.-S. Gong, D. N. Sheng, O. I. Motrunich, and M. P. A. 144409 (2014). Fisher, Phys. Rev. B 88, 165138 (2013). [27] D. Yamamoto, I. Danshita, and C. A. R. Sáde Melo, [58] B. Huang, G. Clark, E. Navarro-Moratalla, D. R. Klein, Phys. Rev. A 85, 021601(R) (2012). R. Cheng, K. L. Seyler, D. Zhong, E. Schmidgall, M. A. [28] D. Yamamoto, A. Masaki, and I. Danshita, Phys. Rev. McGuire, D. H. Cobden, and et al., Nature 546, 270–273 B 86, 054516 (2012). (2017). [29] M. Singh, T. Mishra, R. V. Pai, and B. P. Das, Phys. [59] T. Li, S. Jiang, N. Sivadas, Z. Wang, Y. Xu, D. Weber, Rev. A 90, 013625 (2014). J. E. Goldberger, K. Watanabe, T. Taniguchi, C. J. Fen- [30] S. R. Hassan, L. de Medici, and A. M. S. Tremblay, Phys. nie, and et al., Nature Materials 18, 1303–1308 (2019). Rev. B 76, 144420 (2007). [60] E. J. Samuelsen, R. Silberglitt, G. Shirane, and J. P. [31] T. McIntosh, P. Pisarski, R. J. Gooding, and Remeika, Phys. Rev. B 3, 157 (1971). E. Zaremba, Phys. Rev. A 86, 013623 (2012). [61] C. Gong, L. Li, Z. Li, H. Ji, A. Stern, Y. Xia, T. Cao, [32] L. Isaev, G. Ortiz, and J. Dukelsky, Phys. Rev. B 79, W. Bao, C. Wang, Y. Wang, and et al., Nature 546, 024409 (2009). 265–269 (2017). [33] L. Isaev, G. Ortiz, and J. Dukelsky, Phys. Rev. Lett. [62] T. J. Williams, A. A. Aczel, M. D. Lumsden, S. E. Nagler, 103, 177201 (2009). M. B. Stone, J.-Q. Yan, and D. Mandrus, Phys. Rev. B [34] D. Huerga, J. Dukelsky, and G. E. Scuseria, Phys. Rev. 92, 144404 (2015). Lett. 111, 045701 (2013). [63] J.-U. Lee, S. Lee, J. H. Ryoo, S. Kang, T. Y. Kim, [35] D. Yamamoto, H. Ueda, I. Danshita, G. Marmorini, P. Kim, C.-H. Park, J.-G. Park, and H. Cheong, Nano T. Momoi, and T. Shimokawa, Phys. Rev. B 96, 014431 Letters 16, 7433 (2016). (2017). [64] C.-T. Kuo, M. Neumann, B. Karuppannan, H. Park,

10 B INTERACTION TERM IN RECIPROCAL SPACE

S. Kang, H. Shiu, J. Kang, B. Hong, M. Han, T. Noh, [69] S. Tacchi, R. E. Troncoso, M. Ahlberg, G. Gubbiotti, and J.-G. Park, Scientific Reports 6, 20904 (2016). M. Madami, J. Akerman,˚ and P. Landeros, Phys. Rev. [65] S. Okubo, T. Ueda, H. Ohta, W. Zhang, T. Sakurai, Lett. 118, 147201 (2017). N. Onishi, M. Azuma, Y. Shimakawa, H. Nakano, and [70] M. Heide, G. Bihlmayer, and S. Blügel, Phys. Rev. B T. Sakai, Phys. Rev. B 86, 140401(R) (2012). 78, 140403(R) (2008). [66] L. Chen, J.-H. Chung, B. Gao, T. Chen, M. B. Stone, [71] K. Di, V. L. Zhang, H. S. Lim, S. C. Ng, M. H. Kuok, A. I. Kolesnikov, Q. Huang, and P. Dai, Phys. Rev. X J. Yu, J. Yoon, X. Qiu, and H. Yang, Phys. Rev. Lett. 8, 041028 (2018). 114, 047201 (2015). [67] E. Vinas Boström,M. Claassen, J. McIver, G. Jotzu, [72] D. Gastaldo, N. Strelkov, L. D. Buda-Prejbeanu, B. Di- A. Rubio, and M. Sentef, SciPost Physics 9, 061 (2020). eny, O. Boulle, P. Allia, and P. Tiberto, Journal of Ap- [68] S. A. Owerre, Journal of Physics Communications 1, plied Physics 126, 103905 (2019). 021002 (2017).