betaFeedback

WorksPhysics and Astronomy

Zoology of spin and orbital fluctuations in ultrathin oxide films

Abstract

Many metallic transition-metal oxides turn insulating when grown as films that are only a few unit-cells thick. The microscopic origins of these thickness induced metal-to-insulator transitions however remain under dispute. Here, we simulate the extreme case of a monolayer of an inconspicuous correlated metal – the strontium vanadate SrVO3_3 – deposited on a SrTiO3_3 substrate. Crucially, our system can have a termination to vacuum consisting of either a SrO or a VO2_2 top layer. While we find that both lead to Mott insulating behavior at nominal stoichiometry, the phase diagram emerging upon doping – chemically or through an applied gate voltage – is qualitatively different. Indeed, our many-body calculations reveal a cornucopia of nonlocal fluctuations associated with (in)commensurate antiferromagnetic, ferromagnetic, as well as stripe and checkerboard orbital ordering instabilities. Identifying that the two geometries yield crystal-field splittings of opposite signs, we elucidate the ensuing phases through the lens of the orbital degrees of freedom. Quite generally, our work highlights that interface and surface reconstruction and the deformation or severing of coordination polyhedra in ultra-thin films drive rich properties that are radically different from the material's bulk physics.

Introduction

In strongly correlated materials[1, 2] various energy scales compete in defining the ground state. Perturbing their balance, e.g., through pressure or doping, may induce a multitude of different long-range orders or trigger metal-insulator transitions.[3] Particularly rich are phenomena involving the orbital degrees of freedom.[4, 5] Their behavior is extremely sensitive to the local atomic environment that controls hybridizations,[6] crystal-fields,[7] and their degeneracy. Therefore, the advent of epitaxial growth of ultra-thin films and hetero-structures has unlocked vast possibilities[8, 9] to explore orbital physics and electronic anisotropies[10] in general. Indeed, substrate strain, interfacial or surface reconstruction, and varying surface terminations cause distortions, rotations, or even the severing of coordination polyhedra. These structural changes invariably affect the electronic, magnetic, and orbital state, often leading to properties absent in bulk samples of the same material.

Here, we study the extreme case of a monolayer of the perovskite transition-metal oxide SrVO3\mathrm{SrVO}_{3} deposited on a SrTiO3\mathrm{SrTiO}_{3} substrate. A moderately correlated paramagnetic metal in the bulk,[11, 12] SrVO3\mathrm{SrVO}_{3} is known to undergo a metal-to-insulator transition (MIT) in ultra-thin films on SrTiO3\mathrm{SrTiO}_{3} below a critical thickness of 2-3 unit-cells.[13, 14] This Mott insulator[13, 15] has further been suggested as active material in a Mott transistor,[16] where a gate voltage is used to switch between the insulator (OFF) and a metallic (ON) state. Besides the characterization of the charge state, however, little is known about ordered phases or long-range fluctuations in ultra-thin films. Bulk Mott insulators typically order antiferromagnetically (AF) at low enough temperatures, while doping them may lead to various types of fluctuations and symmetry-broken phases, e.g., superconductivity or charge-order. Here, we find that the SrVO3\mathrm{SrVO}_{3} monolayer realizes more than five distinct phases, see Figure 1, including (in)commensurate antiferromagnetism (AF), ferromagnetism (FM), as well as stripe (s) and checkerboard (c) orbital-order (OO). Some of these regimes can be explored by chemical doping or an applied gate voltage. Additionally, we evidence a qualitatively different phase diagram for the two possible choices of surface termination, in which the top is formed by either a VO2\mathrm{VO}_{2}, or a SrO layer (Figure 1a, Figure 1b, respectively). Ultimately, we link the character of dominant fluctuations to the orbital degrees of freedom, that are tuned through the (total) filling nn and the crystal-field splitting Δcfs\Delta_{\mathrm{cfs}} between the two-fold degenerate dxzd_{xz}, dyzd_{yz} orbitals and the dxyd_{xy} orbital residing in the film’s plane. With these key ingredients being common to a multitude of correlated oxides, our study of the SrVO3\mathrm{SrVO}_{3} monolayer anticipates rich phase diagrams in transition-metal oxide ultra-thin films.

Figure 1. Phase diagrams. (a) VO2_2 and (b) SrO terminated monolayers of SrVO3_3 on a SrTiO3_3 substrate (see insets) realize various phases as a function of the number of electrons per site in the low-energy t2gt_{2g} (nn; lower xx-axis) or gate voltage (VGV_G; upper xx-axis): antiferromagnetism (AF: red), ferromagnetism (FM: blue), incommensurate magnetism (iM: blueish), checkerboard orbital order (cOO: green), stripe orbital order (sOO: turquoise). The thus colored domes indicate the formation of long-range order within dynamical mean-field theory (DMFT). The “+”–signs indicate points for which many-body calculations were performed (black signs represent non-divergent, white signs divergent DMFT susceptibilities). Based on the dominant susceptibilities, the color domes have been drawn as a guide to the eye.

The outline of the paper is as follows: Section II briefly discusses the methods employed: density functional theory (DFT) and dynamical mean-field theory (DMFT). Section III A presents the quantitative differences between the two different surface terminations on the one-particle level. Section III B presents a many-body discussion of the Mott insulating state found in the stoichiometric systems, while Section III C presents the electronic structure trends upon doping the SrVO3\mathrm{SrVO}_{3} monolayers. Section III D characterizes magnetic and orbital fluctuations and associated ordering instabilities on the basis of DMFT susceptibilities. Section IV puts our findings into perspective. Finally, Section V summarizes the results and provides an outlook.

Method

DFT calculations are performed with the Wien2k package,[17, 18] using the PBE[19] exchange-correlation potential. We construct systems with one half of a unit cell of SrVO3\mathrm{SrVO}_{3} (VO2\mathrm{VO}_{2} termination) and one unit cell of

SrVO3_3 (SrO termination) on top of a substrate of six unit cells of SrTiO3_3 surrounded by sufficient vacuum of around 10 Å in zz-direction (see insets in Figure 1). In both setups the transition between SrTiO3_3 and SrVO3_3 consists of a TiO2_2 - SrO - VO2_2 interface, consistent with experiment[20, 21], while at the bottom the SrTiO3_3 substrate is terminated via SrO to vacuum. Since experimentally SrVO3_3 is locked to the SrTiO3_3 substrate[22] we initialize the in-plane lattice constant with the PBE-optimized value for bulk SrTiO3_3 aSrTiO3=3.95a_{\mathrm{SrTiO_3}} = 3.95 Å. To treat the surface properly the two unit cells of SrTiO3_3 furthest away from SrVO3_3 are then constrained to aSrTiO3a_{\mathrm{SrTiO_3}}, simulating the transition to the SrTiO3_3 bulk, while all other internal atomic positions are fully relaxed.

We then perform Wannier projections onto maximally localized V-t2gt_{2g} orbitals, using the WIEN2Wannier[23] interface to Wannier90[24], see Figure 2. These Wannier Hamiltonians are supplemented with an effective SU(2)-symmetric Kanamori interaction of U=5U = 5 eV, J=0.75J = 0.75 eV, U′=3.5U' = 3.5 eV (similar to Ref. 15), for which we perform dynamical mean-field theory (DMFT)[25, 26] calculations at various temperatures. For the undoped bulk, this setup yields the correct mass enhancement and spectra qualitatively congruent with photoemission spectroscopy[27, 28]. These values can be thought of as a lower boundary since in ultrathin films, interaction parameters increase slightly[15] with respect to their bulk values. The Hamiltonians are kept constant under doping. The analytic continuations to real frequencies are performed with the maximum entropy method[29] used in the ana_cont library[30, 31].

Figure 2. Band structures and density of states. (a) VO₂ and (b) SrO terminated monolayer. Left: DFT band structure along a momentum path through the Brillouin zone (black dots) overlain with the t2gt_{2g}-orbital projected Wannier dispersion (red lines). Right: the resulting density of states (DOS) clearly shows the (quasi–) 2D character of the xyxy-orbital (blue) and the 1D character of the locally degenerate xz/yzxz/yz-orbitals (green). The local orbital energy levels are marked as dashed horizontal lines in the DOS. We find a crystal field splitting of Δcfs=−0.252\Delta_{\mathrm{cfs}} = -0.252eV for the VO₂ terminated layer and Δcfs=+0.126\Delta_{\mathrm{cfs}} = +0.126eV for the SrO terminated monolayer.

The DMFT self-consistency cycle as well as the sampling of the two-particle Green’s function is done by continuous-time quantum Monte Carlo simulations in the hybridization expansion[32, 33], using w2dynamics[34] with worm sampling[35]. Momentum-dependent DMFT susceptibilities are calculated from the local vertex, using the AbinitioDΓA[36, 37] program package.

Results

Surface termination, band structure and density of states

Both, SrO and VO2_2-terminated films result in similarly looking DFT band structures whose relevant orbitals around the Fermi level are of vanadium t2gt_{2g} character, see Figure 2. The densities of states (DOS) of these orbitals showcase the abrupt surface termination of the sample: While the xyxy-projection (blue line) keeps its two-dimensional character (as in bulk SrVO3_3) the (locally degenerate) xzxz- and yzyz-projections (green lines) now become one-dimensional and, concomitantly, display a strongly reduced bandwidth. Consequently a van-Hove singularity emerges, which, at zero doping, is in close proximity to the Fermi level. Indeed, in the VO2_2 (SrO) terminated system this singularity is situated slightly below (above) the Fermi level. On top of this dimensionality reduction we find the crystal-field splitting (cfs)

Δcfs=Exz/yz−Exy,(1)\Delta_{\mathrm{cfs}} = E_{xz/yz} - E_{xy}, \tag*{(1)}

to have opposite signs for the two different setups: The positive cfs of the SrO terminated monolayer, Δcfs=\Delta_{\mathrm{cfs}} = +0.13+0.13eV, is a direct result of the tensile strain caused by the (in-plane) lattice mismatch between SrVO₃ and SrTiO₃ (aSrVO3<aSrTiO3a_{\mathrm{SrVO3}} < a_{\mathrm{SrTiO3}}). The in-plane expansion triggers a structural compression in the perpendicular direction, so as to keep the volume approximately constant. This structural anisotropy translates into an electronic anisotropy[10]: The evident breaking of the cubic symmetry of SrVO₃ lifts the three-fold t2gt_{2g} degeneracy, making the xyxy-orbital energetically favorable. The same effects take place in the VO₂ terminated monolayer as well. There, however, the geometric distortion gets overcompensated by the missing SrO layer: severing the apical oxygen of the transition metal coordination octahedron results in a reversed, negative Δcfs=−0.25\Delta_{\mathrm{cfs}} = -0.25eV. The xz/yzxz/yz orbitals have their lobes pointing in the zz-direction, towards the lobes of the oxygen px/pyp_x/p_y orbitals. The missing overlap to the absent apical oxygen leads to less electrostatic repulsion, thus lowering the energy required to occupy these states. Another contributing factor is the abrupt termination to vacuum, removing any restriction in the positive zz-direction for the structural relaxation: The VO₂ (SrO) terminated system results in a concave (convex) final termination, i.e., the last VO₂ (SrO) layer bends inwards (outwards). For both systems we find almost identical xyxy orbitals with a nearest-neighbor hopping txy∼−230t_{xy} \sim-230meV, next-nearest-neighbor hopping txy′∼−70t'_{xy} \sim-70meV and bandwidth Wxy∼2.1W_{xy} \sim2.1eV. The xzxz and yzyz orbitals of both systems, on the other hand, can be described purely by nearest-neighbor hopping along the xx or yy direction, respectively: The VO₂-terminated monolayer allows for a large hopping amplitude (txz/yz∼−300t_{xz/yz} \sim-300meV), resulting in a slightly larger bandwidth Wxz/yz=1.2W_{xz/yz} = 1.2eV in Figure 2a, compared to only Wxz/yz∼−200W_{xz/yz} \sim-200meV for the SrO-terminated monolayer in Figure 2b.

Recent experiments[38, 39] suggest ultra-thin SrVO₃ films to be VO₂-terminated. However, for reasons of stability (surface oxidization, surface protection, etc.), a SrO-boundary could be preferable. This might be achievable either implicitly via a SrTiO₃ capping-layer (see § 2.3) or explicitly via deposition of SrO on top of the VO₂ surface. The latter has in fact been achieved for LaNiO₃ films on LaAlO₃ substrate, where both LaO and NiO terminations are possible through an ablation of La₂O₃ and NiO, respectively[40].

Stoichiometric Mott insulator

Next, we analyze the electronic structure for the stoichiometric samples (n=1n = 1) on the many-body level, using DMFT at room temperature T=290T = 290K: While in the VO₂-terminated monolayer (Figure 3a) the out-of-plane xz/yzxz/yz-orbitals realize a quarter-filled Mott insulator with a gap of 0.50.5eV, in the SrO terminated monolayer (Figure 3c) the in-plane xyxy-orbital hosts an essentially half-filled canonical Mott insulator with a gap of 1.21.2eV. This difference can be traced back to the bare crystal-fields. Indeed, DMFT amplifies the effect of the DFT cfs for both terminations, leading to the depletion of the energetically higher lying orbital(s), i.e., the xyxy and the xz/yzxz/yz orbital(s) for the VO₂ and SrO termination, respectively. This correlation-enhanced orbital polarization[41] leads to an effectively reduced orbital-degeneracy. As a consequence, charge (inter-orbital) fluctuations are suppressed and the critical interaction for reaching the Mott state diminishes[27, 42, 43]. The Coulomb interaction is large enough to open a Mott gap in the SrO (VO₂) terminated monolayer with a single (two-fold) degenerate lowest orbital, while three-fold orbitally degenerate bulk SrVO₃ is a stable metal. Let us note that the evidenced orbital polarization persists when including charge self-consistency, which only yields minor corrections because charge is only redistributed between orbitals, not between sites[44].

Figure 3. Mott insulating ground state. DMFT spectral function A(ω)A(\omega) for (a) the VO2_2 terminated and (c) the SrO terminated structure at U=5U = 5eV and room temperature (T=290T = 290K). In both cases a wide Mott gap forms which is accompanied by a strong orbital polarization. Removing the crystal-field splitting, the reduced bandwidth alone results in a Mott insulator (b,d) with a slightly smaller band gap.

However, for both terminations the insulating behavior is actually driven by a combination of the crystal-field splitting[15] and the reduced band-widths[13]. Whereas the crystal-field splitting is essential for the bilayer system[15], we find that the bandwidth reduction alone is sufficient to drive the monolayers insulating. We illustrate this in Figure 3b and Figure 3d where we take the original Hamiltonians and set the cfs artificially to zero by shifting the local orbital energies. Both systems remain insulating in DMFT. Unswayed by the cfs, however, the Mott gaps turn out smaller and orbital occupations (in both cases: nxy,σ>nxz/yz,σn_{xy,\sigma} > n_{xz/yz,\sigma}) only reflect the asymmetry of the orbitals’ DOS. Let us note here that if we instead keep the cfs unchanged and adjust the xz/yzxz/yz-bandwidths such that Wxz/yz=WxyW_{xz/yz} = W_{xy} both systems remain firmly metallic.

To investigate the stability of the insulating state further, we perform DMFT calculations for various intraorbital interaction strengths UU. While keeping the Hund’s coupling JJ fixed to 0.75eV,[45] we adjust the inter-orbital interaction strength U′U' according to spherical symmetry (U′=U−2JU' = U - 2J)[46]. Figure 4 shows the orbital occupations depending on the interaction UU: Starting from our standard value U=5U = 5eV (vertical dashed line), going to larger interaction strengths simply stabilizes the insulating solution further, while also increasing the orbital polarization slightly. Smaller interaction strengths on the other hand, reduce the orbital polarization until, eventually, the insulating solution can no longer be stabilized. This metal-to-insulator transition is, as expected within DMFT, of first order (hysteresis or coexistence regime marked in gray in Figure 4) and manifests itself by a sudden drop of the orbital polarization. The Mott insulating state is stable down to U=4.5U = 4.5eV (U=4.1U = 4.1eV) for the VO2_2 (SrO) terminated monolayer.

Figure 4. Stability of Mott insulating state. Spin-dependent orbital occupation niσn_{i\sigma} vs. intra-band interaction strength UU at room temperature T=290T = 290K; Hund’s coupling J=0.75J = 0.75eV; inter-band interaction U′=U−2JU' = U - 2J. (a) The VO2_2 terminated monolayer is effectively a two-orbital quarter-filled system (nxz/yz,σ∼0.25n_{xz/yz,\sigma} \sim0.25) while (b) the SrO terminated monolayer behaves effectively as a half-filled one-orbital system (nxy,σ∼0.5n_{xy,\sigma} \sim0.5) at large enough interaction strengths. Both lead to a Mott localization of carriers which can be upheld even if we reduce the interaction. The transition to the metallic solution is accompanied by a tight hysteresis after which the orbital polarization drops rapidly. The calculations under doping in the next figure are performed for U=5U = 5eV (vertical, dashed black line).

The stability of the stoichiometric Mott insulating solution is in particular important when doping away from it, see § 2.1 C. As long as the stoichiometric sample is insulating, we expect that any variation of the interaction will have no qualitative impact on the DMFT phase diagram. A smaller on-site repulsion will merely lead to weaker orbital polarizations and shifted boundaries in the phase diagrams, Figure 1a,Figure 1b.

On top of the Mott physics discussed here, weak localization through disorder may play an additional role in the insulating behavior of transport properties[21]. However, the suppression of the one-particle spectra for thin films[13], magneto-transport results for SrVO3_3 thin films on an LSAT substrate[47] as well as SrVO3_3/SrTiO3_3 superlattices[48] argue against a dominant weak localization scenario for the insulator. Similar observations have been made for CaVO3_3 thin films on SrTiO3_3 substrate[49].

Behavior under doping {#sec:III C}

We now discuss the electronic structure of the doped monolayers in their non-symmetry broken phases. From the information of orbital occupations and degeneracies,

we motivate possible ordering instabilities (that will then be quantitatively assessed in § 2.1 D). First, the stoichiometric insulating state in Figure 3 and the various orbital occupations in Figure 5, indicate that our systems are asymmetrical with respect to doping with electrons (n>1n > 1) or holes (n<1n < 1). The VO₂ terminated monolayer (Figure 5a) somewhat upholds its orbital polarization when holes are introduced to the system. Such orbital occupations of the bipartite lattice system make the system prone to a staggered, checkerboard orbital ordering (cOO)[50, 51], as two orbitals may now be occupied alternately on neighboring lattice sites. This is energetically favorable, since a nearest-neighbor hopping then results in a state where different orbitals are occupied so that a (virtual) hopping process costs only U′=U−2JU' = U - 2J instead of UU.

Figure 5. System trends with doping. Spin-dependent orbital occupation at T=290KT = 290\mathrm{K}, resolved for the i=xyi = xy and xz/yzxz/yz-orbitals. (a) The VO₂ terminated system around nominal filling (n=1n = 1) can be effectively described by two quarter-filled (xz/yzxz/yz) orbitals. (b) The SrO terminated system at and below nominal filling can effectively be described by a single half-filled (xyxy) orbital. The former explains the tendency for checkerboard orbital ordering, while the latter promotes antiferromagnetism in Figure 1. The shaded areas around undoped SrVO₃ represent the doping levels at which we find these checkerboard and antiferromagnetic orderings in DMFT at this temperature. Inset: Semi-empirical condition where stripe orbital ordering emerges: If we frustrate the local site enough, we find a transition from checkerboard to stripe orbital-order (indicated by a change in background color)

Doping with more holes, see Figure 5a, the VO₂-terminated monolayer quickly moves away from quarter-filling by redistributing electrons from the xz/yzxz/yz-orbitals into the xyxy-orbital. As we shall see in § 2.1 D, before OO is fully suppressed upon hole doping the ordering vector changes to stripe orbital ordering (sOO) above a particular filling of the so far auxiliary xyxy-orbital. Electron doping on the other hand maintains the quarter-filled state for longer, where for fillings up to n∼1.2n \sim1.2 the additional electrons solely occupy the xyxy-orbital. Only above n∼1.3n \sim1.3 we see a coincidental increase of all orbital occupations, again disfavoring orbital order.

The SrO-terminated monolayer (Figure 5b) is even more asymmetrical: Introducing holes does not affect the effective one-band description of the system and the sparsely filled xz/yzxz/yz orbitals remain almost depleted. More interesting is the electron-doped side, where the multi-orbital character is promoted. In this electron-doped regime, the Hund’s coupling JJ will promote a parallel spin alignment of the electrons in the three orbitals. It is natural to expect that the hopping transfers this local spin alignment into an FM order on the lattice, but other orders such as OO may emerge here as well[52]. For strong Coulomb interactions and in an insulating state, these competing phases can be understood by superexchange as in the classical Kugel-Khomskii spin-orbital models[53]. These phases have also been found in early DMFT calculations for a two-band model[50] and an oversimplified Stoner criterion predicts FM order of the mm-fold degenerate Hubbard model for A(0)(U+(m−1)J)≥1A(0) (U + (m - 1)J) \ge1[54]. At extremely large dopings around n∼1.5n \sim1.5, the physics changes once again: The system now consists of three quite equally filled orbitals where the xz/yzxz/yz orbitals approach quarter-filling. Similarly to stoichiometric filling in the VO₂ terminated monolayer, such degenerate quarter-filled orbitals may lead again to orbital ordering.

DMFT susceptibilities

We now put the above analysis of potential ordered phases on firm footing: For the prevailing magnetic and orbital orders, Figure 6a and Figure 6b displays the relevant DMFT susceptibilities at temperatures above the respective instabilities. Maxima in the shown susceptibilities indicate type and Q\mathbf{Q}-vector of the dominant fluctuations. Figure 6c illustrates, for selected examples, the critical behavior of the (inverse) susceptibilities and (inverse) correlation lengths emerging when said maxima turn into instabilities. Magnetic instabilities occur where the static susceptibility in the magnetic channel

Figure 6. DMFT susceptibilities and criticality. Momentum-dependence of the susceptibility χr(q)\chi_r(\mathbf{q}) for (a) the VO2_2 terminated and (b) the SrO terminated SrVO3_3 monolayer in the vicinity of the respective phase transitions. The dominating component can be either found directly in the magnetic channel (r=mr=m, Eq. (2)) or be obtained via a linear orbital combination of the density channel (r=d′r=d', Eq. (3)). (c) Temperature-dependence of the inverse DMFT susceptibility χr−1(Q)\chi_r^{-1}(\mathbf{Q}) (first diverging rr, Q\mathbf{Q} at selected dopings; circles, left axis) for selected points from (a) and (b); lines are linear fits. Intersections with the TT-axis denote the transition temperature for the respective order. On the secondary (right) axis the corresponding inverse correlation lengths ξr−1\xi_r^{-1} are shown (squares, right axis); dashed lines are fits to mean-field behavior.

χm(Q)=g2∑i,j,l,l′eiQ(Ri−Rj)∫dτ⟨TτSilz(τ)Sjl′z(0)⟩(2)\chi_{\mathrm{m}}(\mathbf{Q}) = g^2 \sum_{i,j,l,l'} e^{i\mathbf{Q}(\mathbf{R}_i-\mathbf{R}_j)} \int d\tau\langle T_\tau S^z_{il}(\tau) S^z_{jl'}(0) \rangle \tag*{(2)}

diverges at a critical temperature, indicated by the intercept of χ−1(Q)\chi^{-1}(\mathbf{Q}) with the temperature axis in Figure 6c. In Eq. (2), gg is the Landé factor; i,ji,j are indices for the lattice sites Ri,Rj\mathbf{R}_i,\mathbf{R}_j; l,l′l,l' are orbital indices. The zz-component of the spin operator Silz=(nil↑−nil↓)/2S^z_{il} = (n_{il\uparrow} - n_{il\downarrow})/2 is expressed in terms of the number operator nilσn_{il\sigma} for an electron on site ii in orbital ll with spin σ\sigma. Ferromagnetism (FM) and antiferromagnetism (AF) correspond to the usual ordering vectors Q=(0,0)\mathbf{Q} = (0,0) and Q=(π,π)\mathbf{Q} = (\pi,\pi), respectively. The incommensurate magnetism (iM), that we find for the VO₂-termination, corresponds to an ordering vector Q\mathbf{Q} with fixed length ∣Q∣=δ>0|\mathbf{Q}| = \delta> 0, see Figure 6a and Figure 7.

Figure 7. Incommensurate magnetism (iM). Left: magnetic susceptibility χm(q)\chi_m(\mathbf{q}) along the high-symmetry path Γ→M\Gamma\to\mathrm{M} for the VO₂-terminated monolayer at n=1.3n = 1.3 at various temperatures. Right: magnetic susceptibility χm(q)\chi_m(\mathbf{q}) for T=170 KT = 170\,\mathrm{K} in the planar Brillouin zone. The susceptibility is maximal roughly on a circle centered around q=(0,0)\mathbf{q} = (0,0). Upon lowering temperature the maximum of the susceptibility moves to larger q\mathbf{q}-vectors. Close to the transition temperature we find anisotropy on this circle where the susceptibility is clearly maximal (purple) for q=(±δ,±δ)\mathbf{q} = (\pm\delta,\pm\delta).

We now assess the instabilities resulting in the phase diagrams, Figure 1a,Figure 1b. Antiferromagnetic (AF) order from super-exchange is facilitated by effectively half-filled orbitals. For the d1d^1 configuration of SrVO₃, only the SrO-terminated monolayer provides this favorable condition. Indeed, there, the positive crystal-field realizes a half-filled, Mott insulating xyxy-orbital that then hosts AF order, see the diverging susceptibility in Figure 6(c). Note that AF order was also predicted for a SrO-terminated SrRuO3_3 monolayer on SrTiO3_3 around nominal stoichiometry. There, the d4d^4 configuration results in an essentially fully occupied xyxy-orbital, and the staggered moment is instead carried by half-filled xz/yzxz/yz orbitals. In both cases, doping with either electrons or holes suppresses the AF state.

Doping the SrVO3_3 monolayer with either termination towards their respective van-Hove singularities, i.e., hole (electron) doping for the VO2_2 (SrO) terminated sample (see Figure 2) results in a strongly increased spectral density around the Fermi level within DMFT. Concomitantly doping generates an orbital configuration where all involved orbitals are close to equally filled, promoting energy minimization through Hund’s exchange JJ and therefore a parallel alignment of the involved spins. This situation, leading to ferromagnetism (FM), is found around n=0.7n = 0.7 in the VO2_2-terminated and near n=1.3n = 1.3 in the SrO-terminated sample. In both cases, ferromagnetism is hosted by the degenerate xzxz and yzyz-orbitals. Subleading non-local AF fluctuations are, however, still present in the xyxy-orbital of the SrO-terminated system. Indeed, an antiferromagnetic stripe pattern, Q=(0,π)\mathbf{Q} = (0,\pi), and symmetrically related at Q=(π,0)\mathbf{Q} = (\pi,0) appears, indicated by additional local maxima in the susceptibility, see Figure 6b. Quite notably, in the absence of Hund’s rule coupling FM fluctuations are strongly suppressed and said frustrated AF spin-fluctuations would be on par with them (additional data, not shown). Moreover, we also find incommensurate magnetic order in the VO2_2-terminated system around n=1.3n = 1.3 in the xz/yzxz/yz-orbitals (iM at n=1.3n = 1.3 in Figure 6a). There, instead of a specific ordering vector Q\mathbf{Q} the magnetic susceptibility is maximal for all vectors Q\mathbf{Q} with origin (0,0)(0,0) and a length of δ=∣Q∣≥0\delta= |\mathbf{Q}| \ge0, i.e., roughly a circle in the q\mathbf{q}-plane. Upon lowering temperature, δ\delta increases and in close vicinity to the ordered phase anisotropy develops; the maximum susceptibility within the circle is found at Q=(±δ,±δ)\mathbf{Q} = (\pm\delta,\pm\delta), see Figure 7. These clear maxima suggest that a kind of frustrated ferromagnetism develops where the xyxy-orbital disturbs the alignment of the xz/yzxz/yz-orbitals. Doping beyond n=1.3n = 1.3 further increases δ\delta (data not shown). Let us note here that throughout the phase diagram we did not find any magnetic instabilities supported by Fermi surface nesting.

The other prevalent type of instabilities we find are of orbital-ordering type between the degenerate xz/yzxz/yz orbitals and can be monitored in the density channel

χdxz/yz(Q)=∑ijσσ′eiQ(Ri−Rj)∫dτ×⟨Tτ(nixzσ−niyzσ)(τ)(njxzσ′−njyzσ′)(0)⟩.(3)\chi_{d}^{xz/yz}(\mathbf{Q}) = \sum_{ij\sigma\sigma'} e^{i\mathbf{Q}(\mathbf{R}_{i}-\mathbf{R}_{j})} \int d\tau\times \left\langle T_{\tau}(n_{ixz\sigma}-n_{iyz\sigma})(\tau)(n_{jxz\sigma'}-n_{jyz\sigma'})(0)\right\rangle. \tag*{(3)}

Towards quarter-filling (ni,σ=0.25n_{i,\sigma} = 0.25) of the xz/yzxz/yz-orbitals, i.e., at and around stoichiometric filling in the VO2_2-terminated monolayer and around n∼1.5n \sim1.5 in the SrO-terminated setup, the wave-vector of critical fluctuations is firmly Q=(π,π)\mathbf{Q} = (\pi,\pi). As previously alluded to, this leads to a checker-board orbital-order (cOO), consistent with model expectations[50, 51]. We note that the xyxy-orbital does not participate in the ordering, as signaled by susceptibility enhancements being confined to components of the other two orbitals. The xyxy orbital can also be passive at larger fillings or valencies: With one electron more, a t2g2t_{2g}^{2} OO—with xyxy-orbitals near half-filling and one electron alternatingly in the xzxz and the yzyz-orbital—can occur in YVO3_3 (LaVO3_3) if Y (La) ions are partially replaced by Ca (Sr)[56], cf. Refs. 57 and 58. An OO with all three t2gt_{2g}-orbitals participating on the other hand is highly frustrated for a cubic lattice[59].

Due to the strong asymmetry around nominal filling in our monolayers, we also find a strong asymmetry of the

corresponding cOO-dome in Figure 1a, where the cOO transition temperature even increases upon electron-doping. If we move too far away from ideal quarter-filling, the ordering temperature is suppressed rapidly. Despite this suppression of cOO we find an additional emerging ordering for n∼0.9n \sim0.9 in Figure 1a. The corresponding ordering can again be described via Eq. (3) with, however, a characteristic vector Q=(0,π)\mathbf{Q} = (0,\pi) (and Q=(π,0)\mathbf{Q} = (\pi,0) related via symmetry), describing stripe orbital-ordering (sOO). The cOO-to-sOO transition under hole-doping is not realized by a continuous move of the ordering vector from Q=(π,π)\mathbf{Q} = (\pi,\pi) to Q=(0,π)\mathbf{Q} = (0,\pi). Instead, increased hole-doping suppresses cOO while simultaneously promoting sOO. We conjecture that this transition can be ascribed to the ‘auxiliary’ xyxy-orbital, that does not contribute to the susceptibility enhancements of either fluctuations. Illustrated in the inset of Figure 5 we find a semi-empirical condition that links the preference for stripe over checkerboard orbital order to the filling of the xyxy-orbital: nxy,σ=(0.25−nxz/yz,σ)/2n_{xy,\sigma} = (0.25-n_{xz/yz,\sigma})/2 below which checkerboard ordering and above which stripe ordering is preferred by the system. Effectively, enough dxyd_{xy} occupation frustrates the local site enough for stripe ordering to be energetically favorable. As for the electron-doped side, the orbital-ordering domes in both systems disappear when doping too far away from quarter-filling.

Discussion

Mapping out all these different magnetic and orbital-ordering instabilities and their corresponding critical temperatures yields the phase diagrams in Figure 1. Naturally, the DMFT long-range order exhibits mean-field criticality (Gaussian fluctuations), i.e., the critical exponents are γ=1\gamma= 1 for the susceptibility χ\chi, and ν=0.5=γ/2\nu= 0.5 = \gamma/2 for the correlation length ξ\xi, see Figure 6. We note that in strictly two dimensions, due to the Mermin-Wagner theorem[60], long-range order can only set in at T=0T = 0. As a spatial mean-field theory, DMFT does not verify this constraint, while it is captured in, e.g., DΓA[61, 62]. In an experimental setting, however, perturbations by disorder (oxygen vacancies, etc.), spin-orbit coupling, single-ion anisotropy, surface adatoms, as well as tunneling into the substrate render strict 2Dness obsolete. This allows for phase transitions at finite temperatures. Still, non-local correlations beyond DMFT[63, 64] are expected to attenuate the tendency for long-range order with respect to our DMFT solution. Non-local fluctuations may even drive pseudogaps, as shown for antiferromagnetic fluctuations in the one-band Hubbard model[62, 65, 66, 67].

In Figure 1 we further indicate that instead of chemical doping, the phase diagram can be perused by applying a gate voltage VGV_{\mathrm{G}}. We note that the estimated values of VGV_{\mathrm{G}} measure the necessary potential directly in the monolayer, not the truly external one applied at a certain distance through a dielectric medium. Realizing the whole phase diagram with a sheet carrier change of 0.50.5 electrons/unit cell (3⋅10143 \cdot10^{14} electrons/cm2\mathrm{cm}^{2}) in a single device might be challenging[68]. Ionic electrolytes as dielectrics, however, allow for such a large amount of induced charges thanks to the large capacitance of polarized ions[69]. Experimentally, the phase diagram Figure 1 can hence be realized either by doping or via gate voltage, or a combination of both.

Finally, let us comment on an important difference between our semi-abab initio calculations and typical model setups. In multi-orbital (Hubbard) models, orbital complexity is often simplified, in the sense that the hopping of each orbital is considered to be isotropic vis-à-vis all neighboring atoms. Many model studies even employ semi-circular densities of states, corresponding to the Bethe lattice with infinite coordination number. Such simplifications allowed distilling essential behaviors of, e.g., correlation enhancements of crystal-fields[70], the influence of Hund’s physics on the Mott transition[71], or correlations in band-insulators[72, 73]. Indeed, qualitative spectral properties of 3D systems without broken symmetries, are mainly controlled by the orbitals’ filling and the kinetic energy they mediate. Near symmetry-broken phases, however, the associated fluctuations in physical susceptibilities are strongly dependent on electronic-structure details. In fact, van-Hove singularities, nesting, or Kohn points may well be the cause of an instability. In our context, besides anisotropies, e.g., induced by the tetragonal distortion[10] and the obvious geometric restriction, a crucial ingredient is the per se 2D nature of the t2gt_{2g}-orbitals: In perfectly 3D cubic perovskites (e.g., bulk SrVO₃) each transition-metal t2gt_{2g} orbital only hybridizes with four of the six oxygen ions of the coordination octahedron. Therefore, both in bulk and ultra thin films, the in-plane dxyd_{xy} DOS is 2D-like. The loss of hopping along the zz-direction in ultra-thin films further reduces the effective dimensionality, leading to the 1D-like DOS of the dxz,yzd_{xz,yz} orbitals, see Figure 2. Our analysis suggests, that even for qualitative phase diagrams of ultra-thin oxide films or heterostructures, both, the crystal geometry and the orbital structure have to be accounted for.

Conclusion and outlook

We have studied a single layer of SrVO3_3 on a SrTiO3_3 substrate, using DFT and DMFT, considering both possible surface terminations, VO2_2 and SrO, to vacuum. We demonstrated that stoichiometric samples (n=1n=1) are bandwidth-controlled Mott insulators: Depending on the surface termination, SrO or VO2_2, the monolayer is an effectively half-filled one-orbital or a quarter-filled two-orbital Mott insulator. We showed this orbital polarization to derive from the crystal-field splitting having opposite signs for the two terminations and to be significantly enhanced by electronic correlations. Electron

or hole-doping reveals multi-orbital effects: For the SrO-termination, AF-fluctuation are dominant around nominal filling. Doping with electrons populates the xz/yzxz/yz-orbitals; they order ferromagnetically (n∼1.3n\sim1.3) or realize checkerboard orbital orbital order (n∼1.5n\sim1.5). For the VO2_2 termination checkerboard xz/yzxz/yz orbital-order already dominates around nominal filling. Doping then instead promotes the xyxy-orbital which acts as a mediator for ferromagnetism and stripe orbital-order on the hole-doped side and incommensurate magnetism on the electron-doped side. While the change in magnetic fluctuations and orders could be observed in neutron experiments, experimentally evidencing the orbital fluctuations is only possible indirectly: the staggered pattern of xzxz and yzyz-orbitals will result in a dynamic (potentially static) alternation of the bond-length in the xx and yy direction, possibly detectable in future x-ray measurements.

In all, the orbital polarization is the essential driver of the phase diagram of the SrVO3_3 monolayer on SrTiO3_3. We therefore summarize our results in Figure 8 in form of an orbital occupation map. The considered surface terminations each realize, under doping, a characteristic trajectory in the nxyn_{xy} vs. nxz+nyzn_{xz}+n_{yz} space. As an outlook, we include a third possibility—a SrVO3_3 monolayer with SrTiO3_3 on both sides: At nominal filling this sandwich is, again, a Mott insulator. However, owing to the symmetric embedding, the crystal-field is minute (but positive). The computation of ordering instabilities of capped ultra-thin films, in which quantum confinement effects could be studied in a more controlled fashion, is left for future work.

Figure 8. Global phase diagram of SrVO3_3 monolayers on SrTiO3_3. DMFT orbital occupations of each setup are mapped into a nxyn_{xy} vs. nxz+nyzn_{xz}+n_{yz} graph. Due to the reduced bandwidth in the monolayer on a SrTiO3_3 substrate, any orbital occupation repartitioning of the nominal n=1n=1 filling (orange line) realizes a Mott insulator. SrO-terminated (black) and VO2_2-terminated (gray) monolayers and their dominant non-local fluctuations (shaded background; colors identical to Figure 1, but lighter) are discussed in this publication. As an outlook we showcase the effect of embedding a SrVO3_3 monolayer in a SrTiO3_3 sandwich (navy blue, dashed): The crystal-field and orbital structure is somewhat intermediate to the two uncapped monolayers. The computation of respective instabilities is left for future work.

A different future avenue are even more realistic setups of the current geometries, e.g., atomic position relaxation with the inclusion of correlation effects and recent advances[74, 75] which allow for calculations of forces and phonons within DFT+DMFT to test the dynamical stability against superstructure formations.

Acknowledgments

We thank R. Claessen, M. Fuchs, J. Gabel, A. Galler, G. Sangiovanni, M. Sing, P. Thunström and Z. Zhong for fruitful discussions. The authors acknowledge support from the Austrian Science Fund (FWF) through grants P 30819, P 30997, P 32044, and P 30213. Calculations were performed in part on the Vienna Scientific Cluster (VSC).

References

  1. [1]L. H. Greene, J. Thompson, and S. J., Rep. Prog. Phys. 80, 030401 (2017).
  2. [2]F. Giustino, J. H. Lee, F. Trier, M. Bibes, S. M. Winter, R. Valenti, Y.-W. Son, L. Taillefer, C. Heil, A. I. Figueroa, B. Placais, Q. Wu, O. V. Yazyev, E. P. A. M. Bakker, J. Nygård, P. Forn-Díaz, S. D. Franceschi, J. W. McIver, L. E. F. F. Torres, T. Low, A. Kumar, R. Gal- ceran, S. O. Valenzuela, M. V. Costache, A. Manchon, E.-A. Kim, G. R. Schleder, A. Fazzio, and S. Roche, Journal of Physics: Materials 3, 042006 (2021).
  3. [3]M. Imada, A. Fujimori, and Y. Tokura, Rev. Mod. Phys. 70, 1039 (1998).
  4. [4]Y. Tokura and N. Nagaosa, Science 288, 462 (2000).
  5. [5]D. I. Khomskii and S. V. Streltsov, Chemical Reviews 121, 2992 (2021), PMID: 3334912.
  6. [6]J. M. Tomczak, Phys. Rev. B 101, 035116 (2020).
  7. [7]H. Bethe, Annalen der Physik 395, 133 (1929).
  8. [8]Y. Tokura, Physics Today 56, 50 (2003).
  9. [9]J. Ngai, F. Walker, and C. Ahn, Annual Review of Materials Research 44, 1 (2014).
  10. [10]B. Klebel-Knobloch, T. Schäfer, A. Toschi, and J. M. Tomczak, Phys. Rev. B 103, 045121 (2021), 2005.01369.
  11. [11]I. H. Inoue, O. Goto, H. Makino, N. E. Hussey, and M. Ishikawa, Phys. Rev. B 58, 4372 (1998).
  12. [12]A. Sekiyama, H. Fujiwara, S. Imada, S. Suga, H. Eisaki, S. I. Uchida, K. Takegahara, H. Harima, Y. Saitoh, I. A. Nekrasov, G. Keller, D. E. Kondakov, A. V. Kozhevnikov, T. Pruschke, K. Held, D. Vollhardt, and V. I. Anisimov, Phys. Rev. Lett. 93, 156402 (2004).
  13. [13]K. Yoshimatsu, T. Okabe, H. Kumigashira, S. Okamoto, S. Aizaki, A. Fujimori, and M. Oshima, Phys. Rev. Lett. 104, 147601 (2010).
  14. [14]M. Kobayashi, K. Yoshimatsu, T. Mitsuhashi, M. Kitamura, E. Sakai, R. Yukawa, M. Minohara, A. Fujimori, K. Horiba, and H. Kumigashira, Scientific Reports 7, 16621 (2017).
  15. [15]Z. Zhong, M. Wallerberger, J. M. Tomczak, C. Taranto, N. Parragh, A. Toschi, G. Sangiovanni, and K. Held, Phys. Rev. Lett. 114, 246401 (2015).
  16. [16]D. M. Newns, J. A. Misewich, C. C. Tsuei, A. Gupta, B. A. Scott, and A. Schrott, Applied Physics Letters 73, 780 (1998).
  17. [17]P. Blaha, K. Schwarz, G.-K.-H. Madsen, D. Kvasnicka, and J. Luitz, Vienna University of Technology, Austria (2001), ISBN 3-9501031-1-2.
  18. [18]P. Blaha, K. Schwarz, F. Tran, R. Laskowski, G. K. H. Madsen, and L. D. Marks, The Journal of Chemical Physics 152, 074101 (2020).
  19. [19]J. P. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev. Lett. 77, 3865 (1996).
  20. [20]J. Li, D. Yin, Q. Li, R. Sun, S. Huang, and F. Meng, Phys. Chem. Chem. Phys. 19, 6945 (2017).
  21. [21]G. Wang, Z. Wang, M. Meng, M. Saghayezhian, L. Chen, C. Chen, H. Guo, Y. Zhu, E. W. Plummer, and J. Zhang, Phys. Rev. B 100, 155114 (2019).
  22. [22]C. Wang, H. Zhang, K. Deepak, C. Chen, A. Fouchet, J. Duan, D. Hilliard, U. Kentsch, D. Chen, M. Zeng, X. Gao, Y.-J. Zeng, M. Helm, W. Prellier, and S. Zhou, Phys. Rev. Materials 3, 115001 (2019).
  23. [23]J. Kunes, R. Arita, P. Wissgott, A. Toschi, H. Ikeda, and K. Held, Computer Physics Communications 181, 1888 (2010).
  24. [24]A. A. Mostofi, J. R. Yates, Y.-S. Lee, I. Souza, D. Vanderbilt, and N. Marzari, Computer Physics Communications 178, 685 (2008).
  25. [25]A. Georges, G. Kotliar, W. Krauth, and M. J. Rozenberg, Rev. Mod. Phys. 68, 13 (1996).
  26. [26]K. Held, Advances in Physics 56, 829 (2007).
  27. [27]E. Pavarini, S. Biermann, A. Poteryaev, A. I. Lichtenstein, A. Georges, and O. K. Andersen, Phys. Rev. Lett. 92, 176403 (2004).
  28. [28]S.-K. Mo, J. D. Denlinger, H.-D. Kim, J.-H. Park, J. W. Allen, A. Sekiyama, A. Yamasaki, K. Kadono, S. Suga, Y. Saitoh, T. Muro, P. Metcalf, G. Keller, K. Held, V. Eyert, V. I. Anisimov, and D. Vollhardt, Phys. Rev. Lett. 90, 186403 (2003).
  29. [29]M. Jarrell and J. E. Gubernatis, Physics Reports 269, 133 (1996).
  30. [30]D. Geffroy, J. Kaufmann, A. Hariki, P. Gunacker, A. Hausoel, and J. Kuneš, Phys. Rev. Lett. 122, 127601 (2019).
  31. [31]J. Kaufmann and K. Held, (2021), arXiv:2105.11211 [cond-mat.str-el].arxiv.org/abs/2105.11211
  32. [32]P. Werner, A. Comanac, L. de’ Medici, M. Troyer, and A. J. Millis, Phys. Rev. Lett. 97, 076405 (2006).
  33. [33]E. Gull, A. J. Millis, A. I. Lichtenstein, A. N. Rubtsov, M. Troyer, and P. Werner, Rev. Mod. Phys. 83, 349 (2011).
  34. [34]M. Wallerberger, A. Hausoel, P. Gunacker, A. Kowalski, N. Parragh, F. Goth, K. Held, and G. Sangiovanni, Computer Physics Communications 235, 388 (2019).
  35. [35]P. Gunacker, M. Wallerberger, E. Gull, A. Hausoel, G. Sangiovanni, and K. Held, Phys. Rev. B 92, 155102 (2015).
  36. [36]A. Galler, P. Thunström, P. Gunacker, J. M. Tomczak, and K. Held, Phys. Rev. B 95, 115107 (2017).
  37. [37]A. Galler, P. Thunström, J. Kaufmann, M. Pickem, J. M. Tomczak, and K. Held, Computer Physics Communications 245, 106874 (2019).
  38. [38]Y. Okada, S.-Y. Shiau, T.-R. Chang, G. Chang, M. Kobayashi, R. Shimizu, H.-T. Jeng, S. Shiraki, H. Kumigashira, A. Bansil, H. Lin, and T. Hitosugi, Phys. Rev. Lett. 119, 086801 (2017).
  39. [39]J. Gabel et al. (in preparation), (2021).
  40. [40]M. Golakhani, Q. Lei, R. U. Chandrasena, L. Kasaei, H. Park, J. Bai, P. Giorgiani, J. Ciston, G. E. Sterbinsky, D. A. Arena, P. Shafer, E. Arenholz, B. A. Davidson, A. J. Millis, A. X. Gray, and X. X. Xi, Nature Communications 9, 2206 (2018).
  41. [41]A. I. Poteryaev, J. M. Tomczak, S. Biermann, A. Georges, A. I. Lichtenstein, A. N. Rubtsov, T. Saha-Dasgupta, and O. K. Andersen, Phys. Rev. B 76, 085127 (2007).
  42. [42]O. Gunnarsson, E. Koch, and R. M. Martin, Phys. Rev. B 54, R11026 (1996).
  43. [43]G. Keller, K. Held, V. Eyert, D. Vollhardt, and V. I. Anisimov, Phys. Rev. B 70, 205116 (2004).
  44. [44]S. Bhandary, E. Assmann, M. Aichhorn, and K. Held, Phys. Rev. B 94, 155131 (2016).
  45. [45]Note that, in contrast to U, J is hardly screened so that there is much less uncertainty and ambiguity than for U.
  46. [46]T. Ribic, E. Assmann, A. Tóth, and K. Held, Phys. Rev. B 90, 165105 (2014).
  47. [47]M. Gu, S. A. Wolf, and J. Lu, Advanced Materials Interfaces 1, 1300216 (2014).
  48. [48]J. Wang, N. Gauquelin, M. Huijben, J. Verbeeck, G. Rijnders, and G. Koster, Applied Physics Letters 117, 133105 (2020), https://doi.org/10.1063/5.0020615.
  49. [49]M. Gu, J. V. Balachandran, B. Chen, K. E. Smith, S. A. Wolf, and J. Lu, Journal of Applied Physics 113, 133704 (2013).
  50. [50]K. Held and D. Vollhardt, The European Physical Journal B - Condensed Matter and Complex Systems 5, 473 (1998).
  51. [51]T. Momoi and K. Kubo, Phys. Rev. B 58, R567 (1998).
  52. [52]Y. Tokura and N. Nagaosa, Science 288, 462 (2000).
  53. [53]D. Khomskii and K. Kugel, Solid State Communications 13, 763 (1973).
  54. [54]P. Fazekas, Lecture Notes on Electron Correlation and Magnetism. (World Scientific, 1999).
  55. [56]W. Brzezicki, A. M. Oleś, and M. Cuoco, Phys. Rev. X 5, 011037 (2015).
  56. [59]G. Khaliullin and S. Maekawa, Phys. Rev. Lett. 85, 3950 (2000).
  57. [60]N. D. Mermin and H. Wagner, Phys. Rev. Lett. 17, 1307 (1966).
  58. [61]G. Rohringer, A. Toschi, A. Katanin, and K. Held, Phys. Rev. Lett. 107, 256402 (2011).
  59. [62]T. Schäfer, N. Wentzell, F. Simkovic, Y. Y. He, C. Hille, M. Klett, C. J. Eckhardt, B. Arzhang, V. Harkov, F.M. Le Régent, A. Kirsch, Y. Wang, A. J. Kim, E. Kozik, E. A. Stepanov, A. Kauch, S. Andersson, P. Hansmann, D. Rohe, Y. M. Vilk, J. P. F. LeBlanc, S. Zhang, A. M. S. Tremblay, M. Ferrero, O. Parcollet, and A. Georges, Phys. Rev. X 11, 011058 (2021).
  60. [63]G. Rohringer, H. Hafermann, A. Toschi, A. A. Katanin, A. E. Antipov, M. I. Katsnelson, A. I. Lichtenstein, A. N. Rubtsov, and K. Held, Rev. Mod. Phys. 90, 025003 (2018).
  61. [64]J. M. Tomczak, P. Liu, A. Toschi, G. Kresse, and K. Held, The European Physical Journal Special Topics 226, 2565 (2017).
  62. [65]T. Schäfer, F. Geles, D. Rost, G. Rohringer, E. Arrigoni, K. Held, N. Blümer, M. Aichhorn, and A. Toschi, Phys. Rev. B 91, 125109 (2015).
  63. [66]T. Schäfer, A. Toschi, and K. Held, Journal of Magnetism and Magnetic Materials 400, 107 (2016).
  64. [67]J. Gukelberger, E. Kozik, and H. Hafermann, Phys. Rev. B 96, 035152 (2017).
  65. [68]C. H. Ahn, A. Bhattacharya, M. Di Ventra, J. N. Eckstein, C. D. Frisbie, M. E. Gershenson, A. M. Goldman, I. H. Inoue, J. S. Hannahrt, A. J. Millis, A. F. Morpurgo, D. Natelson, and J.-M. Triscone, Rev. Mod. Phys. 78, 1185 (2006).
  66. [69]T. Katase, H. Hiramatsu, T. Kamiya, and H. Hosono, Proceedings of the National Academy of Sciences 111, 3979 (2014).
  67. [70]A. I. Portnoyev, M. Ferrero, A. Georges, and O. Parcollet, Phys. Rev. B 78, 045115 (2008).
  68. [71]L. de’ Medici, J. Mravlje, and A. Georges, Phys. Rev. Lett. 107, 256402 (2011).
  69. [72]M. Senf, J. Kuneš, P. Werner, and A. P. Kampf, Phys. Rev. B 80, 155116 (2009).
  70. [73]J. M. Tomczak, J. Phys.: Condens. Matter (Topical Review) 30, 183001 (2018).
  71. [74]I. Leonov, V. I. Anisimov, and D. Vollhardt, Phys. Rev. Lett. 112, 146401 (2014).
  72. [75]C. P. Koçer, K. H. Rau, G. L. Pascut, and B. Monserrat, Phys. Rev. B 102, 245104 (2020).

Paper details

vlnc:W_aja28n7q
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents