betaFeedback

WorksPhysics and Astronomy

Lattice fluctuations, not excitonic correlations, mediated electronic localization in TiSe2_2

Abstract

TiSe2_2 is thought to be an insulator with a bandgap of ~0.1eV. It has attracted a much interest because, among of a rich array of unique properties, many have thought TiSe2_2 is a rare realisation of an excitonic insulator. Below 200 K, TiSe2_2 undergoes a transition from a high-symmetry ({P-3m1}) phase to a low-symmetry ({P-3c1}) phase. Here we establish that TiSe2_2 is indeed an insulator in both {P-3m1} and {P-3c1} phases. However, the insulating state is driven not by excitonic effects but by symmetry-breaking of the {P-3m1} phase. In the CDW phase the symmetry breaking is static. At high temperature, thermally driven instantaneous deviations from {P-3m1} break the symmetry on the characteristic time scale of a phonon. Even while the time-averaged lattice structure assumes {P-3m1} symmetry, the time-averaged energy band structure is closer to the CDW phase – a rare instance of a metal-insulator transition induced by dynamical symmetry breaking. We establish these conclusions from a high-fidelity, self-consistent form of many body perturbation theory, in combination with molecular dynamics simulations to capture the effects of thermal disorder. The many-body theory includes explicitly ladder diagrams in the polarizability, which incorporates excitonic effects in an ab initio manner. The excitonic modification to the potential is slight, ruling out the possibility that TiSe2_2 is an excitonic insulator. Charge self-consistency is essential distinguish the metallic from insulating state.

Introduction

TiSe2_2 is a layered diselenide compound with space group P3ˉm1P\bar{3}m1. Below 200K, it undergoes a phase transition to a charge density wave (CDW), forming a commensurate 2×2×22\times2\times2 superlattice (P3ˉc1P\bar{3}c1) of the original structure. At the transition there is a softening of the zone boundary phonon, and changes are seen in the transport properties[1, 2]. It is also been observed that if the CDW is suppressed by pressure[3] or intercalation of Cu atoms[4], (unconventional) superconductivity appears. A quantum critical point has also been observed when pressure is applied[5]. The interplay of the CDW and superconductivity has been the subject of many studies, but even at the one-particle level whether the pristine (undoped) system is a semimetal or insulator is not well understood. A proper understanding of the origin of the one-particle spectrum is a prerequisite for understanding the superconductivity, and forms the subject of this work.

Bianco et al.[6] discuss a controversy as to whether the instability originates either from the electrons or by the lattice. Such a distinction is not entirely well posed, because because nuclear fluctuations are correlative to inverse of the static charge susceptibility[7]. A less ambiguous way to frame the question is one of a competition between a one-body effect, e.g. some analog of a Jahn-Teller mechanism[8, 9, 10], or a many-body effect, notably an excitonic insulator[11, 12, 13]. In their paper, Bianco et al.[6] present a detailed review of theoretical and experimental understanding of the nature of the gap in the high temperature, P3ˉm1P\bar{3}m1 phase. It was established early on[14] that the valence band maximum is of Se-4p character and falls at the Γ\Gamma point, while the conduction band minimum has Ti-3d character and falls at L. Experimental reports are voluminous and somewhat contradictory, in part because the distinction between semimetal and a small-gap insulator is obscured by doping. A consensus has emerged, starting perhaps with the work of Ref.[15] that the bandgap is small and positive. The gap is variously reported, in the range 0-0.2 eV, and especially in earlier studies was reported as a semimetallic phase with an inverted gap[14, 16, 17] in the high temperature phase.

The argument for TiSe2_2 being an excitonic insulator largely stems from rather unusual shapes of the spectral function observed by ARPES. In the study of Ref.[13], a large spectral weight from backfolded bands was observed, which they interpreted as a signature of an excitonic insulator. Ref.[18] reported a band of “Mexican hat” shape from a GWGW calculation, also indicative of an excitonic insulator. In this work we indeed find backfolded bands, but they originate from Umklapp scattering processes, static in the CDW phase at 0K and ephemeral at higher temperatures. This will be discussed in more detail below.

Early calculations of the energy band structure found it to be a semimetal with the conduction band at L falling below the valence band at Γ\Gamma [14], within the local-density approximation to density-functional theory (DFT). It predicts a negative gap in both the P3ˉm1P\bar{3}m1 and CDW phases. As the LDA systematically underestimates bandgaps, it is not a reliable predictor of the gap. Indeed, adding UU to the LDA, Bianco et al. [6] found a small opening of the gap in both P3ˉm1P\bar{3}m1 and CDW phases. This is not determinative however, because LDA+UU is ‘second-principles’ method whose answer depends UU, which is not known. Prior to this work, Cazzaniga et al. [18] considered a GWGW calculation as a perturbation to the LDA, and found a gap of ∼200\sim200 meV in the high-temperature P3ˉm1P\bar{3}m1 structure. However, this result turns out to be an artefact of the GLDAWLDAG_{\mathrm{LDA}}W_{\mathrm{LDA}} approximation, as we have reported previously [19]. Changes to the charge density (which GWGW as a first order perturbation to the LDA does not take into account) completely changes the picture. A Quasiparticle Self-Consistent form of GWGW (QSGW) [20] predicts the P3ˉm1P\bar{3}m1 structure to be semimetallic, with a gap approximately similar to the LDA though for different reasons. In the LDA, not only does the conduction band at L fall below the VBM at Γ\Gamma, but the Ti- and Se-derived levels at Γ\Gamma are also inverted, as can be seen from Figure 2 of Ref. 19.

Image not extracted

Figure 2. Energy bands of the Pˉ3c1\bar{P}3c1 superlattice in the QSGW approximation (left) and QSGWc approximation (right), respectively, in the Brillouin zone corresponding to the Pˉ3m1\bar{P}3m1 phase. Red indicates a projection onto Ti-3d character, green onto Se character.

Results and Discussion

Since QSGW is a much higher fidelity theory [21] than GLDAWLDAG_{\mathrm{LDA}}W_{\mathrm{LDA}}, and moreover it usually tends to slightly overestimate insulating bandgaps, it is unlikely that ideal P3ˉm1P\bar{3}m1 TiSe2_2 is an insulator. It is highly unusual that GLDAWLDAG_{\mathrm{LDA}}W_{\mathrm{LDA}} predicts a gap while QSGW does not, but it can be explained in terms of renormalisation of the quasiparticle band structure owing to corrections to the LDA density [19]. Moreover it offers a hint that excitons are indeed responsible for forming a gap, since excitonic effects are absent from GWGW. However, as we show here, this is not the case. The role of excitonic effects can be assessed in an ab initio framework by extending GWGW to include ladder diagrams in WW (W→WcW\rightarrow W_{\mathrm{c}}) [22]. We find that electron-hole contributions negligibly affect the self-energy. Nevertheless TiSe2_2 is a band insulator in both P3ˉm1P\bar{3}m1 and CDW phases, but the gap forms only when nuclear fluctuations are properly taken into account. We establish it in the next section by combining ab initio molecular dynamics (AIMD) simulations with QSGW calculations of the band structure on configurations sampled from the AIMD trajectories.

The Charge-Density Wave

Below 200 K, Di Salvo et al. [1] found from neutron diffraction experiments that TiSe2_2 reconstructs into a “3qLq_{\mathrm{L}}” charge-density wave (CDW). In the CDW phase (P3ˉc1P\bar{3}c1 symmetry), the three-atom P3ˉm1P\bar{3}m1 unit cell undergoes a 2×2×22\times2\times2 reconstruction into a 24-atom superlattice. The 3qL3q_{\mathrm{L}} superstructure is realised by superimposing on P3ˉm1P\bar{3}m1 displacements from a symmetric linear combination of three L point phonon modes. L points fall at a zone boundary: they are projected onto the Γ\Gamma point under a 2×2×22\times2\times2 reconstruction. Thus the CDW is commensurate with a 2×2×22\times2\times2 superlattice of the P3ˉm1P\bar{3}m1 structure, with P3ˉc1P\bar{3}c1 symmetry.

Bianco et al. [6], used the GGA to estimate in detail the nuclear positions of this 24-atom structure. They tried several kinds of constraints (van der Waals interactions play an important role, which the GGA is known not to treat well), in particular consider a scenario in which the lattice vectors are fixed at experimental values and nuclear positions relaxed under that constraint. The resulting equilibrium geometry is very near the experimentally observed 3qL3q_{\mathrm{L}} distortion of the P3ˉm1P\bar{3}m1 phase, in line with Di Salvo’s experiments. Initially we will adopt the experimental structure, as it is the most reliable.

T1\mathrm{T}_1T2\mathrm{T}_2T3\mathrm{T}_3G1\mathrm{G}_1G2\mathrm{G}_2G3\mathrm{G}_3
Γ\Gamma000000
A001001/2
M1\mathrm{M}_110001/20
M2\mathrm{M}_20101/200
M3\mathrm{M}_31101/21/20
L1\mathrm{L}_110101/21/2
L2\mathrm{L}_20111/201/2
L3\mathrm{L}_31111/21/21/2

Table 1. Linear combinations of primitive lattice vectors T in the P3ˉm1P\bar{3}m1 cell that comprise the P3ˉc1P\bar{3}c1 supercell, and combinations of the corresponding primitive reciprocal lattice vectors G which make up the k points that fold into the Γ\Gamma point.

Ref. [6] explained in some detail how to construct displacements of individual normal modes that form a 3q3q CDWs and we recapitulate their description here. TiSe2_2 is a layered compound with Ti lying in one layer in the basal plane, sandwiched between a pair of Se layers above and below. The P3ˉm1P\bar{3}m1 structure has three independent parameters: lattice parameters aa and cc, and the height hh of the Se plane relative to Ti. The triplet of (Se1_1, Ti, Se2_2) layers repeats periodically along the cc axis. P3ˉc1P\bar{3}c1 is comprised of 8 unit cells of P3ˉm1P\bar{3}m1 (doubling the lattice along each axis); thus 8 k-points fold into the Γ\Gamma point of the larger cell. These points are depicted in Table I, expressed as multiples of primitive lattice vectors of P3ˉm1P\bar{3}m1. A phonon mode of L1_1 symmetry will have equivalent modes at L2_2 and L3_3, related by a three-fold rotation about the cc axis. More details concerning 3qL3q_{\mathrm{L}} structure can be found in the Appendix.

A great deal can be learned by considering the evolution of the CDW along a generalised coordinate xx where

ui=ui0+x(ui3qL−ui0)(1)\mathbf{u}_i=\mathbf{u}_i^0+x(\mathbf{u}_i^{3q_L}-\mathbf{u}_i^0) \tag*{(1)}

{ui0}\{\mathbf{u}_i^0\} and {ui3qL}\{\mathbf{u}_i^{3q_L}\} are coordinates respectively of the 2×2×22\times2\times2 superlattice of the P3‾m1P\overline{3}m1 and the 3qL3q_L CDW. Coordinates are taken from the supplemental material of Ref. [6]; we use the experimental lattice parameters and displacements (ν=3\nu=3), and choose the overall amplitude so that the lattice reaches experimental equilibrium geometry at x=x∗=1x=x^*=1, i.e. when the L1_1 component of ∣u∣Se=0.014|\mathbf{u}|_{\mathrm{Se}}=0.014 Å (see Eq. 10 of Ref. [6]). The displacements are quite small: at x=x∗x=x^* a typical change in bond length is only δd/d≈0.02\delta d/d\approx0.02.

Figure 1 shows how the band structure evolves with the initiation of the CDW. Panel 1(a) merely reproduces the band structure of the high-symmetry P3‾m1P\overline{3}m1 structure, in the 2×2×22\times2\times2 superstructure which folds L and M into Γ\Gamma. Thus the lines Γ→\Gamma\toL and L→\toM connect equivalent points along different directions. There is no coupling between L-derived states near EFE_F (mostly Ti character, red) and Γ\Gamma-derived states (mostly Se character, green) owing to translational symmetry of the smaller P3‾m1P\overline{3}m1 structure.

Image not extracted

Figure 1. QSGW Energy bands of the P3‾c1P\overline{3}c1 superlattice for: (a) the undistorted structure, x=0x=0; (b) the initial stages of the CDW (x=0.1x=0.1); (c) formation of gap on Γ\Gamma-K and Γ\Gamma-L lines (x=0.2x=0.2); (d) fully formed metal-insulator transition (x=0.5x=0.5); (e) band structure at theoretical equilibrium point (x=1x=1); (f) band structure at x=1x=1 with spin-orbit coupling included. Red and green indicate a projections onto Ti-3d3d and Se characters. Note that L, M, and K are half the size of L, M, and K in the P3‾m1P\overline{3}m1 phase. The latter points fold into Γ\Gamma in the CDW.

The Ti-like dd states at Γ\Gamma (EF−0.04E_F-0.04 eV) are triply degenerate; the Se-like pp states at +0.2+0.2 eV are doubly degenerate. Note in particular a pair of doubly degenerate Se bands at EF−0.05E_F-0.05 eV and a corresponding pair at EF+0.18E_F+0.18 eV. Midway along the Γ\Gamma-M line (which we denote Q for convenience) they cross and form a fourfold degeneracy at the midpoint. Something similar occurs with pairs of threefold degenerate Ti-derived bands at EF−0.04E_F-0.04 eV and EF+0.18E_F+0.18 eV. For a gap to appear these 10 states must split into a group of four occupied and six unoccupied states.

With the initiation of the CDW (see Figure 1, x=0.2x=0.2), two pairs of L-derived (Ti-like) states near Γ\Gamma begin to admix with Γ\Gamma-derived (Se-like) states. They repel and form a gap near Γ\Gamma, with the band maximum offset from Γ\Gamma. The third Ti-like state at EF−0.04E_F-0.04 eV begins to separate from other two. Eventually it rises enough to completely separate from the valence bands and form the conduction band minimum. However the splitting increases quadratically with xx, so for x<0.15x<0.15 it overlaps with the two aforementioned bands and no gap is formed. Around x=0.2x=0.2 there is a pseudogap: a gap on the Γ\Gamma-K and Γ\Gamma-L lines is fully formed, but the system remains metallic on the L-M line. For a gap to form around Q, the aforementioned pair of doubly degenerate Se-like states (−0.05-0.05 and +0.2+0.2 eV at the endpoints) that cross midway between L and M also couple to form bond-antibond pairs. Each of the three kinds of splittings (Se-Se splitting at Q, Ti-Se coupling near Γ\Gamma, Ti-Ti splitting at Γ\Gamma) conspire together to eventually form a gap for sufficiently large xx. The gap is fully formed when x=0.5x=0.5; however, it is indirect with the valence band maximum at Q and conduction band minimum at Γ\Gamma.

As xx increases further to the equilibrium point at x=1x=1 the topology remains similar, and the gap continues to widen. Spin-orbit coupling changes the topology somewhat and reduces the gap to about 0.140.14 eV. The topology of the valence bands is remarkable with several distinct points all within kTkT of each other. The highest band along L-M (nominally the valence band maximum) almost completely dehybridizes and appears like a flat, atomic-like state. The colours also show that a state at Γ\Gamma below the VBM has Ti-3d3d character. This is an echo of the L-derived conduction band minimum in the high-symmetry P3‾m1P\overline{3}m1 phase. The evolution of the gap with respect to xx makes it clear that the gap opens as a consequence of level repulsions in the CDW superlattice that appear because states of unique wave number in the erstwhile P3‾m1P\overline{3}m1 Brillouin zone become coupled. This is in keeping with the observation of Stoffel et al. [15], and of Cercellier et al. [13] who directly observed zone-folding effects in ARPES measurements.

Excitonic effects

We can incorporate electron-hole effects in the GWGW approximation by adding ladder diagrams to the polarizability via a Bethe-Salpeter equation. We use QSGW^\hat{W} to indicate an enhancement of QSGW when ladder diagrams are added to WW. Such a capability was recently developed within Questaal, and Ref. 22 examines in detail the change in spectral properties when WW is augmented to W^\widehat{W} for a wide range of materials systems. It was shown to significantly improve the fidelity of the QSGW theory, particularly when correlations are weak or moderate, as is the case with TiSe2_2. Almost universally, both one-particle spectral functions and dielectric functions are described with remarkably high fidelity.

Thus QSGWc provides a rigorous, ab initio framework to establish the role excitons play in TiSe2_2, and whether it might be an excitonic insulator. Figure 2 compares the QSGW and QSGWc energy band structures in the CDW phase. Differences are very slight: the ladder diagrams reduce the fundamental gap by 0.02 eV. The weak influence of excitons on spectral functions is typical of narrow-gap systems: the nearly metallic electronic structure causes the RPA part of the polarizability to be large, and ladders act to make a small correction to it.

The twin observations that QSGW without excitonic effects yields a small gap (which agrees with experiments as well as they can be measured), and that excitonic additions have almost negligible effect, lead us to conclude with some confidence that TiSe2_2 is not an excitonic insulator in the CDW phase.

Structural fluctuations at finite temperature

Above the ordering temperature TSDW∼200T_{\mathrm{SDW}}\sim200 K, the CDW is lost and the system changes to the high-symmetry Pˉ3m1\bar{P}3m1 phase. However, it assumes Pˉ3m1\bar{P}3m1 only when averaged over time: nuclei are not fixed but fluctuate stochastically about the ideal lattice positions. At very low temperature excursions will be distributed as harmonic excursions around one of the several possible 3qL3q_L deformations to the CDW. Above the transition temperature the lattice assumes the Pˉ3m1\bar{P}3m1 structure on average but will undergo fluctuations which will be weighted to a locale resembling one of the 3qL3q_L CDWs. What effect this has on the energy-band structure can only be determined from knowledge of the stochastic motion of the nuclei.

To this end we performed classical ab initio molecular dynamics AIMD simulations of a 96-atom simulation cell at the density-functional (DFT) level, and took selected snapshots separated far enough in time so as to be uncorrelated. This amounts to making an adiabatic approximation for the electronic structure, and follows a strategy similar to that used successfully to determine the dielectric response and temperature dependence of the bandgap renormalisation in Si[23]. For context, Pˉ3c1\bar{P}3c1 has 3 atoms in the unit cell: the 96-atom cell corresponds to a 4×4×24\times4\times2 replication of it, and a 2×2×12\times2\times1 superlattice of the CDW. The 24-atom CDW is the smallest system capable of revealing differences above and below the CDW phase transition; however, to allow for additional sampling of thermal fluctuations on length scales larger than supported with 24 atoms, we focused on the 96 atom cell. More complete details of how we performed the MD simulations are given Section IIIB.

As we show below, the gap is indeed preserved at finite temperature. This is a rare instance when thermal nuclear fluctuations qualitatively change the topology of the Fermi surface, from metal to insulator. These are largely a result of Umklapp processes that couple kk points near Γ\Gamma to those near L. While the effect of nuclear fluctuations on the bandgap has been noted in other systems[24, 25], TiSe2_2 is unique in that the Umklapp processes give rise to a metal-insulator transition. These Umklapp processes arise because of dynamical Peierls-like displacements that result from the “softness” of the system, which follow as a consequence of the internal energy being more favourable for lattices deformed relative to the Pˉ3m1\bar{P}3m1 structure.

Electronic structure from molecular dynamics trajectories

Molecular dynamics simulations

dˉ\bar{d}Δd\Delta dθˉ\bar{\theta}Δθ\Delta\thetaEˉG\bar{E}_{G}ΔEG\Delta E_{G}
Pˉ3m1\bar{P}3m1 (Expt)2.554-90.02.3∼\sim0.1-
CDW(Expt)2.5560.03890.02.50.18-
120 K2.5640.06790.03.70.090.05
300 K2.5680.08989.94.40.140.06

Table 2. Data extracted for the Pˉ3c1\bar{P}3c1 in the experimental geometry[1], and from classical dynamics simulations in the 96 atom cell at 120 K and 300 K. Columns correspond to mean nearest-neighbour Ti-Se bond length dˉ\bar{d} and fluctuations about the mean (Å); mean Se-Ti-Se bond angle θˉ\bar{\theta} (degrees) and fluctuations about the mean, and the mean QSGW bandgap EˉG\bar{E}_{G} (eV) and root-mean-squared fluctuations about the mean.

Details on how the AIMD were carried out are given in a later section IIIB. Table II shows that average bond lengths and angles are similar in the Pˉ3c1\bar{P}3c1 and Pˉ3m1\bar{P}3m1 phases, so it would be difficult to distinguish them based on pair distribution functions. Indeed, the thermally broadened radial distribution functions look similar at both 120 K and 300 K, albeit with slightly greater broadening of peaks at 300 K.

We computed the QSGW band structure from twelve snapshots calculated at nuclear configurations generated from the AIMD, at 120 K and 300 K. Band edge states are quite fluid in both conduction and valence bands: the valence band maximum fluctuates between the Γ\Gamma and A points. At either temperature the gap varies between 0 and 0.25 eV, with RMS fluctuation of ∼\sim0.06 eV (Table II). As for the disagreement between the absolute values of the gap between theory and experiments (∼\sim60 meV), it is worth noting that realistically, experiments are performed with samples that are degenerately nn-doped with ∼\sim1% excess electrons, and are not stoichiometric. In such situations the change in the gap values between the two phases is a more physical quantity to consider. An intriguing observation is the reduction of the gap at lower temperatures. We note that the lower bound of our estimated gap values at 300 K is ∼\sim80 meV which is close to the experimental observation of ∼\sim74 + ±\pm15 meV gap at the same temperature[26]. At 120 K, we estimate a gap of 90 meV ±\pm 60 meV of RMS fluctuations when the QSGW gap values are sampled over all MD snapshots, i.e. is a reduction of about 50 meV in the mean gap value from 300 K. This is in remarkable agreement with the estimated experimental reduction in gap of 60 meV inside the CDW phase (experimentally estimated gap at 10 K is 15 meV)[26]. Note that our calculations are performed at 120 K while the experiments are performed at 10 K, and the difference in temperature could account for ∼\sim10 meV in thermal broadening.

That increasing thermal lattice fluctuations cause the electronic gap to widen is intriguing. How itinerant electrons localize is at the heart of microscopic principles of electronic correlations, as well as devices based on localisation. For a large class of semiconductors, electrons are Bloch states that open up a gap at the Fermi energy due to reflections at the Brillouin zone boundary, a phenomenon that is known as metal to band insulator transition. Most non magnetic semiconductors are trivial band insulators where the emergence of the insulating phase has a one-to-one correspondence with Landau Fermi liquid theory and for all such cases the electronic properties can be described by single Slater determinant. However, for Mott insulators, magnetic fluctuations, in particular, a multi-reference scenario (multiple Slater determinants) are ubiquitous in describing the localization of electrons starting from the metallic phase and formation of a gap, a phenomenon that has no analogue in Landau Fermi liquid theory for electronic excitation. However, once the electrons get localized and the gap at the Fermi energy is formed, phonons, often than not, play a secondary role of reducing the band gaps[24, 25] by enhancing the electronic screening. In solid state systems, it is rare to find an occasion where electronic localization in itself is mediated by phonons, making TiSe2_2 a special case. It is in that sense that the electronic localization in TiSe2_2 is more akin to how localization happens in liquids. In TiSe2_2 the lattice remains significantly soft and squeezy over a large range of temperatures, so much so, that it can be a source of disorder in itself that is sufficient to localize electrons.

To recover an effective band structure in the Pˉ3m1\bar{P}3m1 phase, we use a band unfolding technique along the lines of work presented in Refs.[27, 28]. Zone unfolding allows us to represent kk-resolved spectral information of the 4×4×24\times4\times2 supercell as statistically averaged quantities of the Pˉ3m1\bar{P}3m1 Brillouin zone. The Brillouin zone of the primitive cell is folded into the supercell so that NN kk-points in k1,…,kN\mathbf{k}_1,\ldots,\mathbf{k}_N, become reciprocal lattice vectors G1,…,GN\mathbf{G}_1,\ldots,\mathbf{G}_N of the supercell (N=24N=24 in this case), with one of the G\mathbf{G} vectors being G=0\mathbf{G}=0. While details of our implementation will be presented elsewhere, in brief it is a method to resolve an eigenstate ψnk\psi^{nk} of the supercell into linear combinations of eigenstates ψ~n′k1…ψ~n′kN\tilde{\psi}^{n'k_1}\ldots\tilde{\psi}^{n'k_N} of the primitive cell. Since ψnk\psi^{nk} is normalised, it will be distributed among partial contributions from the ψ~n′k1…ψ~n′kN\tilde{\psi}^{n'k_1}\ldots\tilde{\psi}^{n'k_N}. The partial weights are the scattering amplitudes into these states from Umklapp processes. The G=0\mathbf{G}=0 weight for eigenstate wknw^n_k is less than unity, approaching it when kk is a good quantum number, i.e. if the supercell is merely a periodic replica of the primitive cell. Thus we can express the spectral function as

A(k,ω)=∑iwki δ[ω−(ϵi−EF)](2)A(\mathbf{k},\omega)=\sum_i w^i_{\mathbf{k}}\,\delta[\omega-(\epsilon_i-E_F)] \tag*{(2)}

where ii runs over the eigenstates of the supercell at k\mathbf{k}. Eq. (2) provides a reasonable description for A(k,ω)A(\mathbf{k},\omega) in the Brillouin zone of the primitive cell, for any snapshot, but averaging over snapshots gives a better representation. The result is shown in Figure 3. The poles in Eq. (2) were broadened out by a Lorenzian broadening of 20 meV.

Image not extracted

Figure 3. QSGW spectral functions folded to the Brillouin zone of the P3ˉm1P\bar{3}m1 structure for (a) the ideal P3ˉm1P\bar{3}m1 structure, (b) the CDW phase, (c) the statistically averaged snapshots from the 120 K AIMD simulations and (d) the same for the 300 K simulations. The color scale ranges from no intensity (white) to full intensity (red) with green midway between.

Fig. 3(a) shows the energy bands of the ideal Pˉ3m1\bar{P}3m1 structure. kk is a good quantum number and bands are well defined bands without Umklapp processes. Figure 3(b) shows how Umklapp processes open a gap in the CDW. A replica of the highest valence band on the A–Γ\Gamma line appears on the L–M line. This couples to the Ti-3d3d state on the L–M line, forming one bonding state and two antibonding states, thus opening a gap. Similarly, the Ti-dd state at L picks up a replica at Γ\Gamma; indeed the majority of the QP weight at L is transferred to Γ\Gamma. Turning to the 120 K case, A(k,ω)A(\mathbf{k},\omega) falls between the Pˉ3m1\bar{P}3m1 and CDW limiting cases. The Umklapp processes are weaker but nevertheless present, causing a gap to open very similar to the CDW case. The highest valence band on the A-Γ\Gamma line is almost dispersionless, with significant incoherence, and very likely these factors are the driving force for many-body effects such as superconductivity in doped TiSe2_2. At 300 K, Umklapp processes are weaker but still present and the gap is preserved. This is a remarkable result, since only some kinds of deformations are sufficient to open a gap (see Figure 1). Intriguingly enough, the blob of spectral weight at L (which is missing in the ideal Pˉ3m1\bar{P}3m1 structure but is only there when the lattice fluctuations are taken into account), can be observed in experiments too[29]. At low temperatures (deep inside the CDW phase) one obviously sees the strongly backfolded hole bands. However at temperatures immediately above 200 K (outside the CDW phase), one can still clearly observe this blob of spectral weight with similar intensity as that of the CDW phase that gradually disappears with increasing temperatures and by 300 K it is mostly gone. The fact that this blob of spectral weight at and above 200 K persists and is temperature dependent are strong signatures that these are related to the fluctuations of the 2×2×22 \times2 \times2 CDW order. However, it is only natural that the exact nature of the coherence of these blobs are sensitive to the de-phasing that comes from the finite size effects in theory. The larger the supercell is higher is the de-phasing and more quickly the blobs will be washed out as we go further away and above the CDW temperature.

An intriguing aspect of systems with an ordered phase is the fluctuation of the order parameter that persists at much higher temperatures compared to the critical temperature for the ordered phase. Experimental methods can barely measure the fluctuation itself above the ordered phase and often the information of the fluctuations are inferred from indirect microscopic signatures that reflect in various measurements. A study of these indirect microscopic signatures in various materials is of extreme interest. Knowledge of the parameter space (temperatures and pressure) where these fluctuations survive is key to finding experimental parameters that can help in condensing the fluctuations into ordered phases at different temperatures and pressures. For example, the experimental signatures of preformed Cooper pairs in cuprates[30, 31, 32, 33], iron based superconductors[34], organic superconductors,[35] CDW fluctuations in various chacogenides[36] and cuprates[37] and beginning of formation of Kondo cloud[38] have been matters of intensive studies in the recent years. Often, multiple orders and their associated order-parameter fluctuations are present in these materials at a certain part of the phase diagram and these fluctuations can compete, co-operate or coexist. In such situations it becomes difficult to attribute certain microscopic signatures to the fluctuations associated with a particular order parameter. A similar situation occurs in TiSe2_2 where (since the spin fluctuations are absent) the charge fluctuations (particularly the excitonic correlations) and lattice fluctuations can leave their imprint on various measurements. Our work is a step change in this regard, where we unambiguously determine the signature of the CDW order parameter fluctuations that persist upto 300 K and show that the spectral blobs observed in the ARPES can not be due to the excitonic correlations but fully determined by the lattice fluctuations. The additional presence of the magnetic fluctuations in various cuprates, iron base superconductors make the analysis more involved, but a systematic diagrammatic improvement of our many-body perturbative approach is probably the desired way forward in addressing the outstanding problem in those class of magnetic materials too.

Computational Methods

Band structure and QSGW

The Quasiparticle Self-Consistent GWGW[20, 39] form of GWGW dramatically improves its fidelity compared to the usual (DFT-based) GWGW. Details of the theory and its implementation in Questaal[40] can be found in Refs[39, 41]. QSGW largely eliminates the starting-point dependence (a problem for GWGW implementations) and importantly, errors are highly systematic, which clarifies which diagrams are missing. In the spectrum between Hartree Fock (which overestimates gaps) and DFT (which underestimates them) QSGW has a tendency to err slightly on the HF side. As we have shown recently[22], adding ladder diagrams to the polarizability eliminates the tendency to underscreen greatly reduces this error. As a consequence excited states and optical properties are shown to be predicted with high fidelity, particularly if correlations are modest as they are here. As we have shown, the high fidelity of QSGW and QSGW^\hat{W} is essential to distinguish artefacts of approximations in the theory from physical behaviour of TiSe2_2.

Ab initio molecular dynamics

All AIMD simulations were performed at the density functional theory (DFT) level using the Vienna Ab-Initio Simulation Package (VASP 5.4)[42, 43, 44, 45] with the

VDWΓ\Gamma2×2×22 \times2 \times23×3×33 \times3 \times34×4×44 \times4 \times45×5×55 \times5 \times5
None683.72560.87562.65556.04572.91
TSa^{a}619.11502.25500.13499.04500.49
TS-Hb^{b}629.93507.10507.20503.93505.68
DFT-D2c^{c}619.24527.44526.40524.25523.73
DFT-D3d^{d}620.56519.68518.76517.73518.68

Table 3. 24 atom cell volume with different combinations of van der Waals correction (VDW) and Γ\Gamma-centered k space mesh, where Γ\Gamma denotes that only the gamma point is used. van der Waals dispersion corrections are: a^{a} Tkachenko-Scheffler^{50}; b^{b} TS with Hirschfeld Iteration^{51}; c^{c} Grimme D2^{52}; d^{d} Grimme D3^{49}.

Perdew-Burke-Ernzerhof (PBE) exchange correlations functional^{46} and the associated projector augmented wave (PAW) pseudopotentials^{47,48} with a plane-wave energy cutoff of 400 eV and periodic boundary conditions in all three directions. A 3×3×33 \times3 \times3 Γ\Gamma-centered k-point grid was used, and Grimme’s D3 van der Waals dispersion-energy correction (DFT-D3)^{49} was applied with a timestep of 2 fs. Fermi smearing was applied with a smearing energy of 0.026 eV (consistent with a temperature of 300 K; for consistency this was used at 120 K as well).

We determined the minimum size of k-point mesh and the choice of van der Waals correction for the AIMD by taking the 24 atom CDW cell and allowing both the atomic positions and the lattice dimensions to relax to minimize the energy, for several different combinations of k-space mesh and van der Waals dispersion correction. We chose a combination of mesh and dispersion correction that closely matches the experimentally determined cell volume of 521.47 Å. The results of these relaxations are shown in Table III. Relaxation of the lattice with the Grimme DFT-D2 and DFT-D3 corrections and 2×2×22 \times2 \times2 and 3×3×33 \times3 \times3 meshes gives geometries close to the experimental value. The DFT-D3 van der Waals dispersion correction was selected because it gives volumes slightly smaller than the experimental volume, which is consistent with there being some contraction from relaxation (essentially a zero temperature result). The 3×3×33 \times3 \times3 k-space mesh (bolded) was selected because it is the smallest tested mesh that contains the Γ\Gamma point and gives a volume within 0.5% of the experimental volume.

Configurations for the 96 atom system were obtained below and above the CDW phase transition from AIMD simulations at 120 K and 300 K, respectively. The system was first equilibrated at constant temperature and pressure (the so-called NpT ensemble) using VASP’s implementation of the Parinello-Rahman algorithm,^{53,54} which allows the cell volume and shape to fluctuate, and an applied pressure of 1 atmosphere. The NpT runs were used to determine average lattice parameters which were then used in subsequent runs at constant temperature and volume (NVT ensemble). Configurations were sampled at intervals of 2 ps from the NVT runs and used to compute the thermally-averaged electronic structure of the TiSe2_2

Stateaa (Å)cc (Å)c/ac/a
P3ˉc1P\bar{3}c1 (CDW)7.0812.011.70
P3ˉm1P\bar{3}m17.0612.011.71
120 K7.06 (0.02)12.14 (0.02)1.71 (0.01)
300 K7.08 (0.05)12.07 (0.05)1.70 (0.01)

Table 4. Lattice parameters (a and c) for the different TiSe2_2 structures studied here. Units are Å and root-mean-square fluctuations are given in parentheses.

system above and below the CDW transition temperature.

Determining lattice parameters from NpT runs: At both 120 K and 300 K, we found that the hexagonal lattice structure was preserved, with fluctuations in the 120 degree angle between the two lattice vectors in the xyxy plane (<0.3∘<0.3^\circ), and a smaller (<0.2∘<0.2^\circ) fluctuation between the two in-plane lattice vectors and the out-of-plane lattice vector. At 300 K, we ran NpT dynamics for 32 ps and observed that the lattice parameters fluctuated but did not drift over this timescale, with average lattice vector lengths of approximately 14.19 Å, 14.14 Å, and 12.07 Å, with root-mean-squared fluctuations of 0.03, 0.06 and 0.09 Å, respectively. These lattice dimensions can be related to the dimensions of the minimum sized CDW lattice by halving the first two (∼\sim14 Å) values. The two xyxy lattice vectors can be taken, within the fluctuations, to be the same so we average their values to produce the average 300 K lattice vectors given in Table IV. At 120 K, we ran NpT dynamics for 22 ps and observed a gradual increase in the zz-oriented lattice vector over the first 12 ps. We therefore averaged only over the final 10 ps to find average lattice vector lengths of 14.13 Å, 14.12 Å, and 12.14 Å, all with root-mean-squared fluctuations of 0.02 Å. We again averaged the lengths of the two vectors in the xyxy plane to produce the average 120 K aa lattice constant given in Table IV.

Summary

Using a combination of the QSGW and QSGW^\hat{W} approximations with AIMD simulations, TiSe2_2 is shown to be a band insulator in both the low-temperature and high-temperature phases, because of symmetry lowering deformations of the P3ˉm1P\bar{3}m1 structure. We show explicitly in an ab initio framework, that excitons play a negligible role, ruling out the common assertion that TiSe2_2 is an excitonic insulator. The gap forms in the low-temperature CDW phase because of level repulsions between Ti-3dd states at L coupling to Se-4pp states at Γ\Gamma arising from matrix elements that appear by Umklapp processes from symmetry-breaking. A gap appears in the nominal P3ˉm1P\bar{3}m1 phase for similar reasons: but now the symmetry-breaking is dynamical, from instantaneous deformations about the P3ˉm1P\bar{3}m1 structure. TiSe2_2 offers a rare instance of topological changes to the Fermi surface originating from lattice fluctuations. The literature on TiSe2_2 has been confusing and contradictory because it is difficult to account for all competing phenomena that are somewhat unique to this system. On the scale of the small bandgap, slight perturbations such as small nuclear deformations or approximations to the charge density, play a large role. It is only possible to disentangle the various effects with an exceptionally high fidelity description of the electronic structure, which the QSGW (QSGW^\widehat{\mathrm{QSGW}}) approximations provide.

Acknowledgements

This work was authored by the National Renewable Energy Laboratory, operated by Alliance for Sustainable Energy, LLC, for the U.S. Department of Energy (DOE) under Contract No. DE-AC36-08GO28308, funding from Office of Science, Basic Energy Sciences, Division of Materials. We acknowledge the use of the National Energy Research Scientific Computing Center, under Contract No. DE-AC02-05CH11231 using NERSC award BES-ERCAP0021783 and we also acknowledge that a portion of the research was performed using computational resources sponsored by the Department of Energy’s Office of Energy Efficiency and Renewable Energy and located at the National Renewable Energy Laboratory.

References

  1. [1]F. J. D. Salvo, D. E. Moncton, and J. V. Waszczak, Phys. Rev. B 14, 4321 (1976).
  2. [2]M. Holt, P. Zschack, H. Hong, M.-Y. Chou, and T.-C. Chiang, Phys. Rev. Lett. 86, 3799 (2001).
  3. [3]A. F. Kusmartseva, B. Sipos, H. Berger, L. Forró, and E. Tutiš, Phys. Rev. Lett. 103, 236401 (2009).
  4. [4]E. Morosan, H. Zandbergen, and B. e. a. Dennis, Nature Phys 2, 544 (2006).
  5. [5]Y. Joe, P. Chen, X. Ghaemi, and et al., Nat. Phys. 10, 421 (2014).
  6. [6]R. Bianco, M. Calandra, and F. Mauri, Phys. Rev. B 92, 094107 (2015).
  7. [7]R. M. Pick, M. H. Cohen, and R. M. Martin, Phys. Rev. B 1, 910 (1970).
  8. [8]Y. Yoshida and K. Motizuki, J. Phys. Soc. Jpn. 49, 898 (1980).
  9. [9]N. Suzuki, A. Yamamoto, and K. Motizuki, J. Phys. Soc. Jpn. 54, 4668 (1985).
  10. [10]N. Suzuki, H. Yoshiyama, K. Motizuki, and Y. Takaoka, Synth. Met. 19, 887 (1987).
  11. [11]M. Traum, M. Margaritondo, N. Smith, J. E. Rowe, and F.J.DiSalvo, Phys. Rev. B 17, 1836 (1978).
  12. [12]J. A. Wilson, Solid State Commun. 22, 551 (1977).
  13. [13]H. Cercellier, C. Monney, F. Clerc, C. Battaglia, L. Despont, M. G. Garnier, H. Beck, P. Aebi, L. Patthey, H. Berger, and L. Forro, Phys. Rev. Lett. 99, 146403 (2007).
  14. [14]A. Zunger and A. J. Freeman, Phys. Rev. B 17, 1839 (1978).
  15. [15]N. G. Stoffel, S. D. Kevan, and N. V. Smith, Phys. Rev. B 31, 8049 (1985).
  16. [16]R. Z. Bachrach, M. Skibowski, and F. C. Brown, Phys. Rev. Lett. 37, 40 (1976).
  17. [17]C. H. Chen, W. Fabian, F. C. Brown, K. C. Woo, B. Davies, B. DeLong, and A. H. Thompson, Phys. Rev. B 21, 615 (1980).
  18. [18]M. Cazzaniga, H. Cercellier, M. Holzmann, C. Monney, P. Aebi, G. Onida, and V. Olevano, Phys. Rev. B 85, 195111 (2012).
  19. [19]S. Acharya, D. Pashov, A. N. Rudenko, M. Rößner, M. van Schilfgaarde, and M. I. Katsnelson, npj Comput. Mater 7, 208 (2021).
  20. [20]M. van Schilfgaarde, T. Kotani, and S. Faleev, Phys. Rev. Lett. 96, 224402 (2006).
  21. [21]M. van Schilfgaarde, T. Kotani, and S. V. Faleev, PRB 74, 245125 (2006).
  22. [22]B. Cunningham, M. Grüning, D. Pashov, and M. van Schilfgaarde, Phys. Rev. B 108, 165104 (2023).
  23. [23]M. Zacharias and F. Giustino, Phys. Rev. B 94, 075125 (2016).
  24. [24]T. Etienne, E. Mosconi, and F. D. Angelis, J. Phys. Chem. Lett. 7, 1638 (2016).
  25. [25]S. McKechnie, J. M. Frost, D. Pashov, P. Azarhoosh, A. Walsh, and M. van Schilfgaarde, PRB 98, 085108 (2018).
  26. [26]M. D. Watson, O. J. Clark, F. Mazzola, I. Marković, V. Sunko, T. K. Kim, K. Rossnagel, and P. D. C. King, Phys. Rev. Lett. 122, 076404 (2019).
  27. [27]T. G. Dargam, R. B. Capaz, and B. Koiller, Phys. Rev. B 56, 9625 (1997).
  28. [28]P. V. C. Medeiros, S. Stafström, and J. Björk, Phys. Rev. B, 89, 041407 (2014).
  29. [29]C. Monney, E. F. Schwier, M. G. Garnier, N. Mariotti, C. Didiot, H. Beck, P. Aebi, H. Cercellier, J. Marcus, C. Battaglia, H. Berger, and A. N. Titov, Phys. Rev. B 81, 165104 (2010).
  30. [30]Y. He, S.-D. Chen, Z.-X. Li, D. Zhao, D. Song, Y. Yoshida, H. Eisaki, T. Wu, X.-H. Chen, D.-H. Lu, C. Meingast, T. P. Devereaux, R. J. Birgeneau, M. Hashimoto, D.-H. Lee, and Z.-X. Shen, Phys. Rev. X 11, 031068 (2021).
  31. [31]Y. Seo, W. Choi, S.-i. Kimura, and Y. S. Kwon, Scientific reports 9, 3987 (2019).
  32. [32]I. Božović and J. Levy, Nature Physics 16, 712 (2020).
  33. [33]K. Maki, Progress of Theoretical Physics 40, 193 (1968).
  34. [34]B. L. Kang, M. Z. Shi, S. J. Li, H. H. Wang, Q. Zhang, D. Zhao, J. Li, D. W. Song, L. X. Zheng, L. P. Nie, T. Wu, and X. H. Chen, Phys. Rev. Lett. 125, 097003 (2020).
  35. [35]T. Furukawa, K. Miyagawa, M. Matsumoto, T. Sasaki, and K. Kanoda, Phys. Rev. B 105, 023165 (2023).
  36. [36]Y. Feng, J. Wang, R. Jaramillo, J. Van Wezel, S. Haravifard, G. Srajer, Y. Liu, Z.-A. Xu, P. Littlewood, and T. Rosenbaum, Proceedings of the National Academy of Sciences 109, 7224 (2012).
  37. [37]R. Arpaia, S. Caprara, R. Fumagalli, G. De Vecchi, Y. Peng, E. Andersson, D. Betto, G. De Luca, N. Brookes, F. Lombardi, et al., Science 365, 906 (2019).
  38. [38]I. V. Borzenets, J. Shim, J. C. Chen, A. Ludwig, A. D. Wieck, S. Tarucha, H.-S. Sim, and M. Yamamoto, Nature 579, 210 (2020).
  39. [39]T. Kotani, M. van Schilfgaarde, and S. V. Faleev, PRB 76, 165106 (2007).
  40. [40]https://www.questaal.org, Questaal code website.
  41. [41]D. Pashov, S. Acharya, W. R. L. Lambrecht, J. Jackson, K. D. Belashchenko, A. Chantis, F. Jamet, and M. van Schilfgaarde, Comp. Phys. Comm. 249, 107065 (2020).
  42. [42]G. Kresse and J. Hafner, Physical Review B 47, 558 (1993).
  43. [43]G. Kresse and J. Hafner, Physical Review B 49, 14251 (1994).
  44. [44]G. Kresse and J. Furthmüller, Computational Materials Science 6, 15 (1996).
  45. [45]G. Kresse and J. Furthmüller, Physical Review B 54, 11169 (1996).

Paper details

vlnc:W_qa3k9wz4
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