Entangled phase diagrams of the 2D Kugel-Khomskii models Wojciech Brzezicki Andrzej M. Oleś M. Smoluchowski Institute of Physics, Kraków, Poland Kugel-Khomskii model • Spin-orbital model; Si - spins ½, τiγ (γ=a,b,c) - eg orbital operators defined in terms of pseudospins ½ - σi. • r1, r2 , r4 depend on Hund coupling JH and Hubbard U ratio η= JH/U: • J is a superexchange constant: J=4t2/U Meaning of the operator • For all {i, γ}: • Eigenstates: -1/2: 1/2: where orbitals’ orientation is given by γ. For all {i}: Origin of the KK model • Effective superexchange model for transition metal oxides (eg. La2CuO4) or fluorides (eg. KCuF3) with a single hole localized at the metal ion in d9 orbital configuration • Only eg are active, no Jahn-Teller distortions • Superexchange Hamiltonian contains: kinetic term Ht, Hund and Hubbard interactions between the hole Hint and crystal-field splitting Hz favorizing one of the eg orbitals: • Effective hopping amplitudes between two sites depend on the shape of occupied orbitals – only cigar-cigar amplitudes are non-zero. Only two orthogonal orbitals for three space directions! • Effective (Kugel-Khomskii) Hamiltonian can be derived from the atomic limit treating Ht as a perturbation (see: A. M. Oleś, L. F. Feiner, and J. Zaanen, Phys. Rev. B 61, 6257 (2000)) Motivation for cluster mean-field calculations for bilayer and single layer 2D systems • Interlayer singlet phase found in a bilayer fluride K3Cu2F7 by magnetic susceptibility measurement using SQUID magnetometer H. Manaka, et. al. J. Phys. Soc. Jpn. 76, 044710 (2007). • Pressure induced phase transition in monolayer K2CuF4: AO order with FM spin configuration changes into FOx with AF order at 9-10 GPa • Lattice structure changes: orthorombic Bbcm → Ammm • Crystal-field splitting energy Ez in KK model can be regarded as uniaxial pressure along c direction • Rough estimation: J=0.225 eV (for CuO2 planes and x orbitals) Ez/P=4.5 meV/GPa (for t2g states of iron) gives: Ez=0.18-0.2 J for P=9-10 GPa M. Ishizuka, I. Yamada, K. Amaya and S. Endo, J. Phys. Soc. Jpn. 65, 1927 (1996) • Rich phase diagram of a three-dimensional KCuF3: L. F. Feiner, A. M. Oleś, and J. Zaanen, Phys. Rev. Lett. 78, 2799 (1997). (cited 155 times) • Role of spin-orbital entanglement in t2g systems; violation of the Goodenough-Canamori rules • In shaded area spins order AF despite Jij<0 for d1 and d2 models • Cluster MF can capture all of these effects! A. M. Oleś, P. Horsch, L. F. Feiner, and G. Khaliullin, Phys. Rev. Lett. 96, 147205 (2006). Single-site mean field for a bilayer • The KK Hamiltonian: • The MF decoupling: • Because of single-site character of this approximation, no spin flutuactions can be taken into account (total Mz is conserved) • This means that spin and orbital degrees of freedom factorize • We assume certain magnetic order to determine orbital order • The order parameters are: with γ=a,b and γ γ where we assume the AO order as the most general in a single site approach • Self consistency equations can be solved exactly (see W. Brzezicki and A. M. Oleś, Phys. Rev. B 83, 214408 (2011)) • Comparing the ground-state energies of different phases we obtain phase diagram Single-site MF phase diagram for a bilayer Cluster MF for a bilayer • Cluster apprach gives the possibility of spin fluctuations inside the cluster – we can get singlet phases and spin-orbital entanglement • Our cluster: • Solid lines = operator – operator bonds • Dashed lines = operator – <operator> bond • We cover whole lattice with such clusters Spin and orbital order • Our order parameters are: with γ=a,b • Configuration of site 1 determine all other sites inside and outside of the cluster depending on assumed spin and orbital order Spin order: • FM phase – for all i: si = s • AF phase – for all i є {1,4,5,8}: si = s and for all i є {2,3,6,7}: si = -s + assumption that neighboring clusters are identical Orbital order: • AO phase – for all i є {1,4,5,8}: tia,b = ta,b and for all i є {2,3,6,7}: tia,b = tb,a + assumption that neighboring clusters are identical • PVB phase – for all i є {1,...,8}: tia,b = ta,b + assumption that neighboring clusters are rotated by π/2 in the ab plane Spin-orbital order: we treat via,b as if via,b = s tia,b • Example of the PVB orbital order: Iterative procedure • We choose some values of {s, ta,b, va,b}, desired order and (Ez, η) • Using Lanczos algorith me find the ground state ane new {s, ta,b, va,b} • We continue until the convergence is reached • For AF phases we work in zero magnetization sector, d = 17920 • For FM phases we work in maximal magnetization sector, d = 256 • To determine the relevance of the non-factorizable SO field we first assume that va,b = s ta,b Factorizable phase diagram for a bilayer • Singlet phases shaded yellow • No perfect orderings • 2nd order phase transition between VBz and VBm • High frustration for nonzero η • No C-AF phase Phase diagram for a bilayer • Phases with SO entanglement are shaded orange: ESO, EPVB • Intermediate phase between PVB and G-AF: continuous transitions • Four-critical point at Order parameters, correlations and entanglement • In what follows we will use the notation for the correlation functions: where γ= direction of the bond <1,i >, i=2,3,7. • The simplest parameters of SO entaglement are: - on-site SO non-factorizibility - bond SO non-factorizibility Passage through the intermediate PVB-AF Entangled phases Passage through the intermediate VBm Bilayer KK model: conclusions • Including quantum fluctuations within the cluster gives new phases comparing to the single-site MF phase diagram: PVB and VBz singlet phases and destroys perfect order • Independent SO order parameter stabilizes new phases in the most frustrated part of the phase diagram • Spin-orbital order can exist independently from magnetic and orbital orders • Spin-orbital entaglement can exist in some phases away from the phase transitions, as a permanent effect Cluster MF for a monolayer • Cluster = four-site square • Calculations in T>0 – thermal averages • We consider phases: AO-FM and PVB-AF • Symmetries used: Mz conservation and invariance under the permutation of sites (1→3, 2→4, 3→1, 4→2) • Diagonal blocks of the cluster Hamiltonian: Mz = 0 d+= 52, d– = 44, Mz = ±1 d+= 32, d– = 32, Mz = ±2 d+= 10, d– = 6. Phase diagram for a monolayer in T=0 • Phases observed before plus.. • Plaquette Liquid Phase (orange) PSL phase • all inter-cluster spin-spin correlations are zero • short-range FM order inside the cluster • orbital and spin-orbital order • each cluster is degenerate between two configurations: • the degeneracy is not lifted by the mean fields • average cluster configuration (not superposition) is characterized by: a) zero magnetization b) no orbital alternation (FO order) c) spin-orbital order such that for all i: and the global pattern for via,b is: Melting PSL • Ez=-0.7J, η=0.17 • thermal fluctuations lift the degeneracy • global FM order appears, entropy S drops, orbitals start to alternate • we get on-site non-factorizibility high but not necessarily entanglement • orbital order survives longer than magnetization • maximal entropy = 0.554 • bond non-factorizibility low – plot shows 10*Ra,b Monolayer KK model: conclusions • Two-fold degeneracy of the cluster’s ground state in certain parameter range for T=0 (and T>0) not lifed by the mean fields • This stabilizes a spin liquid phase with uniform orbital order and SO order doubling the unit cell • PSL phase exhibits SO non-factorizibility which is not an effect of SO entanglement even in T=0 • Thermal decay of the PSL phase is due to the order by disorder mechanism and leads to the FM phase Thank you for your attention! Gracias por su atención!