betaFeedback

WorksPhysics and Astronomy

Electronic band structure in pristine and Sulfur-doped Ta2_2NiSe5_5

Abstract

We present an angle-resolved photoemission study of the electronic band structure of the excitonic insulator Ta2_2NiSe5_5, as well as its evolution upon Sulfur doping. Our experimental data show that while the excitonic insulating phase is still preserved at a Sulfur-doping level of 25%\%, such phase is heavily suppressed when there is a substantial amount, ∼\sim 50%\%, of S-doping at liquid nitrogen temperatures. Moreover, our photon energy-dependent measurements reveal a clear three dimensionality of the electronic structure, both in Ta2_2NiSe5_5 and Ta2_2Ni(Se1−x_{1-x}Sx_x)5_5 (x=0.25,0.50x=0.25, 0.50) compounds. This suggests a reduction of electrical and thermal conductivities, which might make these compounds less suitable for electronic transport applications.

Introduction

The dimensionality and degree of anisotropy in electronic transport and optical properties of layered materials, such as, MoS2_2 1,2, ReS2_23 and other transition-metal dichalcogenides4, make them potentially suitable for the fabrication of new electronic and optoelectronic devices. A recent addition to the family of such materials are the ternary chalcogenides, Ta2_2NiX5_5 (X →\to Se/S), which have attracted attention due to their layered crystalline structure and in-plane anisotropic properties5–7.

In both materials, the transition metal atoms and the chalcogen atoms are arranged forming a chain pattern in each layer and the layers are stacked by van der Waals forces8. While the ground state of Ta2_2NiSe5_5 is an excitonic insulator (EI)9–11, Ta2_2NiS5_5 does not show any evidence of hosting an excitonic insulating phase and is a normal semiconductor in its ground state12. An excitonic insulator state is realised in materials where the valence and conduction bands are separated by a very small energy gap (Eg>0E_g > 0) or there is a small overlap in energies of the band edges (Eg<0E_g < 0). The attractive Coulomb force between electrons and holes in a narrow gap semiconductor or in a semimetal lead to the formation of bound electron-hole pairs, known as excitons13–16. The formation of excitons in a system gives rise to an intriguing ground state, namely, the excitonic insulator17, which is characterised by a larger band gap that mirrors the exciton binding energy EbE_b.

The pre-requisite for the spontaneous formation of excitons is ∣Eb∣>∣Eg∣|E_b| > |E_g| and the excitonic state will be most stable around a band gap material (Eg=0E_g = 0). Since EgE_g is usually significantly larger than EbE_b in semiconductors and insulators, materials with nearly zero energy gap or nearly zero energy overlap (Eg≈0E_g \approx0) such that a small EbE_b can establish the EI ground state are chosen as potential candidates in the search of excitonic insulators. The spontaneous formation of excitons is suppressed with increasing values of ∣Eg∣|E_g|, i.e., for more positive values of EgE_g when Eg>0E_g > 0 (in a semiconductor) and for more negative values of EgE_g when Eg<0E_g < 0 (in a semimetal), leading to a decrease in the values of the EI transition temperature TcT_c. As ∣Eg∣|E_g| is increased such that it is comparable to or larger than the exciton binding energy, i.e., when ∣Eg∣≳Eb|E_g| \gtrsim E_b, the excitonic phase becomes increasingly unstable against the semiconducting/semimetallic ground state14,16. The energy gap EgE_g can be controlled through chemical substitution or by applying physical pressure18,19.

The EI phase has been identified in a number of materials, such as TiSe2_220,21, Ta2_2NiSe5_519, TmSe1−x_{1-x}Tex_x22, InAs/GaSb quantum wells23, etc. However, the phase transition in some of these materials is strongly influenced by lattice and spin degrees of freedom24,25, for example, a charge density wave (CDW) or a band Jahn–Teller effect put forward for 1T-TiSe2_226,27. In this respect, Ta2_2NiSe5_5 has emerged as a promising candidate for hosting a canonical excitonic insulator phase, excluding any alternative scenarios such as lattice distortion-mediated CDW mechanism leading to the phase transformation. However, Ta2_2NiSe5_5 also undergoes an orthorhombic to monoclinic structural phase transition8,28 in addition to EI transition which raises a question on the dominant origin of the order parameter in this material. This triggered numerous experimental and theoretical29–31 studies that used either equilibrium[34] [10, 32, 33] or non-equilibrium[35, 36] approaches trying to resolve the issue. The question on the dominant origin of the order parameter in Ta2_2NiSe5_5 has also been addressed in one of our recent works[37]. The scope of this paper is not to identify the origin of the phase transition but to investigate the effect of increasing EgE_g on the electronic band structure of Ta2_2NiSe5_5 and the dimensionality in pure and doped compounds. The increase of EgE_g is obtained by introducing Sulfur atoms at Selenium sites. Beyond a certain doping limit, xx, the excitonic insulating phase in Ta2_2Ni(Se1−x_{1-x}Sx_x)5_5 is completely suppressed19, a signature of which is the absence of a second-order phase transition in resistivity measurements. As xx increases, changes in hybridization between Ni and Se/S valence band orbitals lead to a monotonic increase of the bandgap EgE_g and, as consequence to a reduction of TcT_c. Application of physical pressure also results in the reduction of TcT_c in pristine Ta2_2NiSe5_519.

Angle-resolved photoemission spectroscopy is an experimental technique for probing the occupied energy bands in different solid state materials and can be directly compared to band structure calculations. In the present work, we utilize polarization-dependent ARPES to investigate the effect of increasing Sulfur doping levels on the electronic band structure of Ta2_2Ni(Se1−x_{1-x}Sx_x)5_5, notably the changes in EE vs. kk dispersion of the valence band, along different symmetry directions in kk-space. While extensive equilibrium and non-equilibrium studies have been done on pristine Ta2_2NiSe5_5, the TcT_c vs. xx phase diagram19 has not been explored through ARPES so far.

Although the parent compounds are known to possess strongly two-dimensional properties of the electronic structures, which is consistent with the layered structure of Ta2_2NiSe5_5 and Ta2_2NiS5_5, our photon-energy dependent ARPES data reveal a possible three dimensionality of the electronic structure in Ta2_2Ni(Se1−x_{1-x}Sx_x)5_5 compounds.

Experimental Details

Single crystals of Ta2_2Ni(Se1−x_{1-x}Sx_x)5_5 (with x=0.0x = 0.0, x=0.25x = 0.25 and x=0.50x = 0.50) were purchased from HQ Graphene (http://www.hqgraphene.com). The samples were synthesised by chemical vapour transport (CVT)[40]. The ARPES experiments were carried out at the BaDElPh beamline[41] of the Elettra synchrotron in Trieste, Italy. It offers both horizontal (p-pol) and vertical (s-pol) light polarizations, a maximum angular resolution of 0.1∘0.1^\circ and energy resolution of 5.4 meV. A photon energy range from hν=16h\nu= 16 eV to hν=34h\nu= 34 eV was used for this study and all the data presented in the paper were acquired at sample temperature T=80T = 80 K. Prior to ARPES measurements, clean sample surfaces were obtained via cleaving in the direction perpendicular to the atomic planes. The samples were cleaved under UHV pressure better than 5×10−105 \times10^{-10} mbar and the measurements were performed at a base pressure <1×10−10< 1 \times10^{-10} mbar.

Results and Discussion

Ta2_2NiSe5_5, first reported in Ref.[8], has a layered crystalline structure, where each layer consists of double chains of tantalum (Ta) atoms with single chains of nickel (Ni) atoms in-between. The Ta-Ni-Ta trichains are parallel to the crystallographic aa-axis and repeat themselves along the cc-axis, exhibiting a quasi-one dimensional structure. Selenium atoms are tetrahedrally and octahedrally coordinated around the Ni atoms and Ta atoms, respectively. The layers are stacked along bb-axis and are held together through van der Waals forces. The crystal structure of Ta2_2NiSe5_5 in its low temperature monoclinic phase is shown in Figure 1(a). Band structure calculations have shown that the topmost valence band comprises Ni 3dd and Se 4pp orbitals while the conduction band bottom is composed of Ta 5dd orbitals11. The high temperature orthorhombic phase of Ta2_2NiSe5_5 is characterised by a very small direct band gap while its low temperature monoclinic phase is an excitonic insulator, with a second-order phase transition occuring at Tc≈326T_c \approx326 K28. Excitons are formed by Ni 3dd - Se 4pp holes and Ta 5dd electrons with a flattened dispersion of the valence band top characterising the excitonic insulating state from an ARPES point of view. A schematic of the bulk Brillouin zone of Ta2_2NiSe5_5 for the monoclinic phase, along with its projection on the kxkyk_xk_y plane is shown in Figure 1(b).

Crystal structure, Brillouin zone, and experimental geometry of Ta₂NiSe₅

Figure 1. (a) Crystal structure of Ta2_2NiSe5_5 in the monoclinic phase where aa, bb, cc define the crystal axis system and tantalum, nickel and selenium atoms are represented by blue, red and yellow spheres, respectively. (b) (left) Bulk Brillouin zone (BZ) of a monoclinic lattice where Γ\Gamma, X, Y, Z are the high symmetry kk-points, (right) Projection of the BZ on kxkyk_xk_y plane. (c) Schematic of the experimental geometry showing the polarization of incident photons with respect to the sample.

The experimental geometry of the sample plane with respect to the analyser slit and direction of light polarization is schematically represented in Figure 1(c). Figure 2 shows an overview of the band structure in Ta2_2NiSe5_5 for the low temperature phase, revealed by our ARPES measurements along Γ\Gamma–XX direction, using light with different linear polarizations. The ARPES spectra in Figure 2(a)–(Figure 2b) show that the top part of the valence band centered at Γ\Gamma (kx=0k_x = 0) is characterised by a flat band dispersion9 which is the signature of an excitonic insulating phase. The stacked energy distribution curves (EDCs) along kxk_x extracted from the (a)–(b) ARPES intensity plots are shown in Figure 2(c)–(Figure 2d), respectively, to give a better perspective of the band dispersions, especially the valence band below Fermi level, EFE_F (referenced to 0). To give a clear picture of the characteristic band flatness close to EFE_F, the peak energy positions of the topmost valence band determined from the EDCs in (c) and (d) are plotted as a function of kxk_x in Figure 2(e) and Figure 2(f), respectively.

ARPES intensity maps and extracted energy distribution curves of Ta₂NiSe₅

Figure 2. ARPES intensity maps of Ta2_2NiSe5_5 along Γ\Gamma−X direction acquired at T=80T = 80 K using (a) p-polarized (horizontal) and (b) s-polarized (vertical) light of photon energy hν=19h\nu= 19 eV. The extracted energy distribution curves (EDCs) stacked along kxk_x-axis for p-pol (c) and s-pol (d). The blue curve corresponds to the EDC at Γ\Gamma (kx=0k_x = 0) in each case. A zoomed view of the valence band dispersion showing the band flatness near Fermi level, EFE_F, where the peak positions of the band obtained from the EDCs in (c) and (d) are plotted as a function of kxk_x for both polarisations, (e) p-pol and (f) s-pol. The dashed line denotes the Fermi level, EFE_F.

We observe that the top valence band lies at ∼0.2\sim0.2 eV below EFE_F and its dispersion around Γ\Gamma is hole-like with a flattened top, giving an overall shallow M-shaped dispersion. The parts away from Γ\Gamma and more towards XX are strongly dispersive. A characteristic M-shaped dispersion which becomes sharper on lowering the temperature has been previously reported from ARPES studies done using different light polarization geometries34.

Moving to the polarization dependence of the ARPES spectra, we note from Figure 2(a) and Figure 2(b) that the photoemission spectra acquired using two different polarization geometries, namely, p-polarization (E\mathbf{E} lying in xzxz plane in Figure 1(c)) and s-polarization (E\mathbf{E} along yy-direction in Figure 1(c)), exhibit strong differences. The spectral weight of the flat top part of the valence band centered at Γ\Gamma is relatively strong for s-polarization. For the dispersive parts of the valence band which are away from Γ\Gamma and along the Γ−X\Gamma-X line, the spectral weight is stronger when probed with p-polarized light. This polarizationdependent contrast continues to exist at different photon energies explored and hence is attributed to the symmetry selection rules of the photoemission process[42, 43]. Similar symmetry contrasts are also observed in 25% and 50% S-doped Ta₂NiSe₅ (Supplementary Material, Section A). Such drastic changes in the spectral intensity of polarization-dependent ARPES data can help us identify the mirror symmetries associated with different atomic orbitals. From now on, we will only show the data acquired using s-polarized light due to the better visibility of the top part of the valence band, which is relevant for the excitonic insulator phase.

Although Ta₂NiSe₅ is believed to form a quasi-one-dimensional structure, the hopping of mobile carriers along cc-axis gives rise to a two-dimensional (2D) character of the electronic band structure. There exists an anisotropy in the band dispersions of Ta₂NiSe₅[34, 44, 45] and Ta₂NiS₅12 along Γ − X and Γ − Y directions. Our ARPES data reveal similar anisotropic properties of the electronic structure in Sulfur-doped Ta₂NiSe₅ as well. The ARPES intensity maps along Γ − X and Γ − Y (refer to Figure 1(b) for the k-space symmetry points) of Ta₂Ni(Se₁−ₓSₓ)₅ for xx = 0.25 are shown in Figure 3(a) and Figure 3(b), respectively. We observe that the dispersion around Γ point along kxk_x-axis (Γ−X direction) is characterised by a flat feature of the valence band top while the dispersion of the valence band at Γ\Gamma along kyk_y-axis (Γ−Y\Gamma-Y direction) does not exhibit the hybridized flattened band dispersion. The periodicity of the band dispersion along kyk_y in 3(b) could be easily captured with the photon energy range used in this study, due to its small periodicity along Γ−Y\Gamma-Y direction. For better clarity of the band dispersions along both directions, the corresponding stacked EDC plots are displayed in Figure 3(c) and Figure 3(d). A zoomed view of the dispersion for the top part of the valence band centered at Γ\Gamma along kxk_x and kyk_y directions is presented in Figure 3(e) and Figure 3(f), respectively. The dispersions have been obtained from the peak energy positions of the corresponding EDCs (3(c) and 3(d)) for the valence band top below EFE_F. The flat band feature around Γ\Gamma along kxk_x is clearly seen while the dispersion of the valence band top along kyk_y seems to be parabolic.

ARPES intensity maps, energy distribution curves, and valence band dispersion of 25% S-doped Ta2NiSe5

Figure 3. ARPES intensity maps of 25% S-doped Ta2_2NiSe5_5 along (a) Γ\Gamma–XX and (b) Γ\Gamma–YY directions acquired at T=80T = 80 K and photon energy hν=20h\nu= 20 eV using s-polarized photons. The extracted energy distribution curves (EDCs) stacked along kxk_x-axis in (c) and kyk_y-axis in (d). The blue curve corresponds to the EDC at Γ\Gamma [kx=0k_x = 0 in (a), ky=0k_y = 0 in (b)]. A zoomed view of the valence band dispersion near Fermi level, EFE_F, where the peak positions of the EDCs in (c) and (d) are plotted as a function of kxk_x and kyk_y in (e) and (f), respectively.

For completeness, the anisotropic dispersions for pristine Ta2_2NiSe5_5 and Ta2_2Ni(Se1−x_{1-x}Sx_x)5_5 (x=0.5x = 0.5) are presented in the Supplementary Material (Section B). Because the bands also show a dispersion in the (in-plane) direction perpendicular to the chains, the quasi-1D structure often assumed for this family of compounds does not fully describe the electronic structure in these materials.

We now study the effect of Sulphur doping on the electronic structure of Ta2_2NiSe5_5, i.e., the band dispersions in Ta2_2Ni(Se1−x_{1-x}Sx_x)5_5, with increasing values of xx. Figure 4 shows the evolution of the band dispersion along Γ−X direction in Ta2_2Ni(Se1−x_{1-x}Sx_x)5_5 with xx changing from 0 to 0.5. We observe that for Sulfur doping levels of x=0x = 0, 0.25 and 0.5, the band structure retains the main dispersive features down to approximately 2 eV below EFE_F, except for the top-part of the valence band at Γ. The relative changes in intensity are due to different photoionization cross sections at different photon energies. What is interesting to note here is that while the band flatness with the shallow M-shape is present for pristine Ta2_2NiSe5_5 (0% S-doping in Figure 4(a), (Figure 4d)) and remains unaffected under 25% Sulfur doping (see Figure 4(b), (Figure 4e)), it is significantly distorted for a substantial S-doping level of 50% (see Figure 4(c), (Figure 4f)). Since the flattened dispersion of the valence band top is a fingerprint of the excitonic insulator phase, the observed deviation is indicative of the suppression of an excitonic insulator phase under heavy S-doping. For better clarity, the valence band dispersions around Γ extracted from their respective EDC plots are displayed in Figure 4(g)–(Figure 4i). The flat band distortion can be understood considering the size of the band gap EgE_g in the high temperature phase of Ta2_2Ni(Se1−x_{1-x}Sx_x)5_5. Indeed, with increasing levels of Sulfur doping, the band gap EgE_g monotonically increases, thereby, suppressing the spontaneous formation of excitons. Hence, the excitonic insulating state becomes unstable against the semiconducting ground state. This emphasizes that although an excitonic insulator phase in Ta2_2NiSe5_5 can be realized with 25% of S-doping at 80 K, the excitonic ground state is destroyed when the doping level is ≈ 50%. Electrical transport studies have revealed that a similar S-doping level (i.e., x≥0.5x \ge0.5) does not allow to observe the excitonic phase transition, down to 2K.

ARPES intensity plots, stacked energy distribution curves, and band dispersions for Ta₂NiSe₅ and sulfur-doped Ta₂NiSe₅

Figure 4. (a)-(c) ARPES intensity plots of Ta₂NiSe₅, Ta₂Ni(Se₀.₇₅S₀.₂₅)₅ and Ta₂Ni(Se₀.₅S₀.₅)₅, respectively, along Γ − X direction acquired at photon energies hνh\nu = 24 eV, 26 eV, 25 eV (s-polarization) respectively, and at TT = 80 K. (d)-(f) Stacked energy distribution curves (EDCs) extracted from their respective ARPES plots in (a)-(c); the blue curve corresponds to the EDC at Γ in each case. (g)-(i) Band dispersion of the top part of the valence band near EFE_{\mathrm{F}} for xx = 0, 0.25, 0.5, respectively, obtained from the peak positions of the corresponding EDCs.

We now turn to the evolution of the band structure along Γ − Y direction as a function of S-doping. Figure 5(a)-(Figure 5c) shows the band structure of Ta2_2Ni(Se1−x_{1-x}Sx_x)5_5 for x=0x = 0, 0.25, 0.5 and the corresponding stacked EDCs are plotted in Figure 5(d)-(Figure 5f) for better visualization of the band dispersions. As already shown in Figure 3(b), (Figure 3d) and (Figure 3f), the valence band near EFE_F is characterised by a hole-like parabolic dispersion along kyk_y centred at Γ-point. From the ARPES intensity maps Figure 5(a)-(Figure 5c), we see that the band dispersions along kyk_y-direction have subtle changes in the binding energies of different states for 25% S-doping [Figure 5(e)] but pronounced changes for 50% S-doping [Figure 5(f)], when compared to the energy states in pristine Ta2_2NiSe5_5 [Figure 5(d)]. The dispersion of the valence band around Γ near EFE_F for the three doping levels are compared in Figure 5(g)-(Figure 5i). We notice that by increasing the amount of S-doping, the parabolic dispersion centered at Γ shifts to higher binding energies resulting in an overall enhancement of the energy gap along Γ − Y direction. As stated before, the valence band is composed of Ni 3d3d and Se 4p4p (S 3p3p) orbitals. Therefore, increasing substitution by Sulfur atoms at the Se atom sites brings the states at the top of the valence band to higher binding energies. Another interesting fact that can be noted from the band dispersions is the change in effective mass ratio, m∗/mem^*/m_e (where, mem_e is the rest mass of an electron), in Ta2_2Ni(Se1−x_{1-x}Sx_x)5_5 at different S-doping levels. The valence band dispersions along Γ − Y have been fitted with quadratic functions, E0−ℏ2k2/2m∗E_0 - \hbar^2 k^2/2m^* (hh is the Planck’s constant), with two fitting parameters E0E_0 and m∗m^*. The experimental data in Figure 5(g)-(Figure 5i) are well reproduced by the fits, an example of which is shown in Figure 6(a). The values of m∗m^* obtained from the fits are plotted in units of mem_e as a function of S-doping level in Figure 6(b). We observe that while there is a small difference in the values of m∗/mem^*/m_e between x=0x = 0 and x=0.25x = 0.25, it is comparatively larger for x=0.5x = 0.5. This is expected due to stronger carrier localization with increasing S-doping levels which arises from hybridised Ni 3d3d and S 3p3p orbitals, leading to higher effective masses. Such a result is consistent with our observations in Figure 5(g)-(Figure 5i), where the energy states move towards higher binding energies as S-doping is increased.

ARPES intensity plots, stacked energy distribution curves, and band dispersions for Ta₂NiSe₅ and sulfur-doped Ta₂NiSe₅

Figure 5. (a)–(c) ARPES intensity plots of Ta2_2NiSe5_5, Ta2_2Ni(Se0.75_{0.75}S0.25_{0.25})5_5 and Ta2_2Ni(Se0.5_{0.5}S0.5_{0.5})5_5, respectively, along Γ−Y\Gamma-Y direction acquired at photon energy hν=22h\nu= 22 eV (s-polarization) and at T=80T = 80 K. (d)–(f) Stacked energy distribution curves (EDCs) extracted from their respective ARPES plots in (a)–(c); the blue curve corresponds to the EDC at Γ\Gamma in each case. The dashed lines in (a)–(f) denote the Fermi level, EFE_F. (g)–(i) Band dispersion of the top part of the valence band near EFE_F for x=0x = 0, 0.250.25, 0.50.5 respectively obtained from the peak position of the corresponding EDCs. The dashed line denotes the peak position at Γ\Gamma for x=0x = 0.

Fit to the E vs. $k$ dispersion and effective mass at different doping levels

Figure 6. (a) Fit (red curve) to the EE vs. kk dispersion (black circles) around Γ in pristine Ta2_2NiSe5_5. (b) Effective mass m∗m^* (in units of electron rest mass, mem_e) at different doping levels, xx.

Next, we investigate whether the band dispersion in Ta2_2Ni(Se1−x_{1-x}Sx_x)5_5 exhibits a three-dimensional character. The three dimensionality of the electronic structure can be determined by probing the out-of-plane Γ − Z direction which corresponds to k⊥k_\perp in our experimental geometry. A photon energy dependent ARPES study is required to obtain such an information. Figure 7(a)-(Figure 7e) shows the EE vs. kk maps of Ta2_2Ni(Se1−x_{1-x}Sx_x)5_5, x=0.25x = 0.25, along Γ − X direction acquired at different photon energies. Due to low photoionization cross section at photon energies below 15 eV and above 35 eV, a photon energy range from 16 eV to 34 eV was chosen. Looking at ARPES intensity maps, one can notice a photon energy-dependent cross sectional contrast, as well as a considerable variation of the binding energies of different bands, especially for bands lying between 0.5 eV and 1 eV below EFE_F and the flat band near EFE_F. We will mainly concentrate on the energy modifications of the ‘flat band’ with variations in photon energy. The EDCs at Γ-point [along the red dotted line in Figure 7(a)] within 0.3 eV below EFE_F, capturing only the spectral intensity of the flat band, are plotted for different photon energies in Figure 7(f). From the ARPES images and the EDC plots, the obvious trend that we observe is the shift of the flat band towards EFE_F on increasing the photon energy. This is clearly seen if one notes the energy separation between 0 eV binding energy (or EFE_F, shown by black dashed line) and the flat band below it in Figure 7(a)-(Figure 7e). The dispersion of the flat band along k⊥k_{\perp} can be visualized from the red dashed line denoting the peak position of the EDCs at different photon energies in Figure 7(f). The peak position of the flat band relative to EFE_{\mathrm{F}} at different photon energies has been obtained by fitting each of the EDCs with a Gaussian-Lorentzian distribution function after Shirley background subtraction. Figure 7(g) shows that the peak position of the flat band at Γ\Gamma moves towards EFE_{\mathrm{F}} with increasing photon energy, with a dip at hν≈20h\nu\approx20 eV. A decrease in the binding energy of the flat band in Ta2NiSe5\mathrm{Ta}_{2}\mathrm{NiSe}_{5} and the distorted flat band in Ta2Ni(Se0.5S0.5)5\mathrm{Ta}_{2}\mathrm{Ni}(\mathrm{Se}_{0.5}\mathrm{S}_{0.5})_{5} at Γ\Gamma-

ARPES intensity plots and valence-band peak positions

Figure 7. (a)-(e) ARPES intensity plots of Ta2Ni(Se0.75S0.25)5\mathrm{Ta}_{2}\mathrm{Ni}(\mathrm{Se}_{0.75}\mathrm{S}_{0.25})_{5} (T=80T = 80 K) along Γ−X\Gamma- X direction for ss-polarized light at photon energies, hν=17h\nu= 17 eV, 21 eV, 26 eV, 29 eV, 32 eV respectively. The Γ\Gamma-point is denoted by the red dotted line in (a) and the dashed lines in (a)-(e) at 0 eV binding energy denote EFE_{\mathrm{F}}. (f) EDCs at Γ\Gamma for different photon energies covering a range from hν=16h\nu= 16 eV to 32 eV with a step size of 1 eV. The red dashed line is a guide to the eye following the peak position of the valence band with increasing photon energy. (g) Peak position of the valence band obtained by fitting the EDCs with a Lorentzian-Gaussian line shape at Γ\Gamma as a function of photon energy. (h), (i) Dependence of the valence band peak position on photon energy in pristine and 50% S-doped Ta2NiSe5\mathrm{Ta}_{2}\mathrm{NiSe}_{5}, respectively.

point (Γ−X\Gamma-X direction) with increasing photon energy are shown in Figure 7(h) and (Figure 7i), respectively. There are no pronounced qualitative differences in the trend of binding energy vs photon energy curves amongst the three compounds (Supplementary Material, Section C). For a perfectly two-dimensional electronic structure often used for Ta2_2NiSe5_5 and Ta2_2NiS5_5, the band dispersions should be independent of kzk_z (k⊥k_\perp in our experimental geometry) but the above observations clearly show a dispersion along kzk_z, i.e., along Γ−Z\Gamma-Z direction in Figure 1(b). This demonstrates the three-dimensionality of electronic band structure in this family of compounds.

For completeness, we also present the photon energy dependence of the band structure along Γ−Y\Gamma-Y direction. Figure 8(a)-(Figure 8d) show the ARPES intensity plots at different photon energies for pristine Ta2_2NiSe5_5. Again, an overall energy shift of the band structure towards EFE_F (denoted by black dashed lines in the 2D-plots) is observed with increasing photon energies. The EDCs containing only the top-part of the valence band at Γ\Gamma are plotted for various photon energies in Figure 8(e). A similar trend in the shift of valence band peak position as

ARPES intensity plots and valence-band peak positions for pristine Ta$_2$NiSe$_5$

Figure 8. (a)-(d) ARPES intensity plots of pristine Ta2_2NiSe5_5 (T=80T = 80 K) along Γ−Y\Gamma- Y direction for s-polarized light at photon energies, hν=18h\nu= 18 eV, 22 eV, 26 eV, 30 eV respectively. The Γ\Gamma-point is denoted by the red dotted line in (a) and the dashed lines in (a)-(d) at 0 eV binding energy denote EFE_F. (e) EDCs at Γ\Gamma for various photon energies. The red dashed line is a guide to the eye following the peak position of the valence band with increasing photon energy. (f) Peak position of the valence band obtained by fitting the EDCs at Γ\Gamma as a function of photon energy.

that in Figure 7(f) can be noted here (marked by the red dashed line). The photon energy-dependent peak positions are plotted in Figure 8(f), where a decrease in binding energy is observed, thereby demonstrating a clear dispersion along Γ−Z\Gamma-Z direction. The above results of the photon energy dependent ARPES measurements in all the three compounds do emphasise that inspite of the well known layered structure characterising the Ta2_2Ni(Se1−x_{1-x}Sx_x)5_5 compounds, the electronic structure is not strictly two-dimensional.

Conclusion

To summarize, we investigated the evolution of band structure in Ta2_2Ni(Se1−x_{1-x}Sx_x)5_5, as a function of Sulfur concentration xx and its three-dimensionality using ARPES. The anisotropy of the in-plane band dispersions along Γ−X\Gamma-X and Γ−Y\Gamma-Y directions in the doped compounds is clearly revealed. A substantial amount of Sulfur doping ∼50%\sim50\% is able to suppress the excitonic ground state in Ta2_2Ni(Se1−x_{1-x}Sx_x)5_5. This can be claimed due to the observed deviation of the dispersion of valence band top from its flattened EE vs. kk feature along Γ−X\Gamma- X at x=0.5x = 0.5 but not at x=0.25x = 0.25. The suppression of the excitonic insulator state is followed by a pronounced increase in the effective mass m∗m^{*} for the highest doping level, indicating stronger localization of the charge carriers, which is in agreement with the weakening of hybridization between Ni 3d and Se 4p (S 3p) orbitals that also leads to larger values of EgE_{g}. Our photon-energy dependent ARPES data show a gradual decrease in the binding energy of states near Fermi level on increasing the photon energy (increasing kzk_{z}). This emphasizes that the two-dimensional picture often used for Ta2_2NiSe5_5 (and Ta2_2NiS5_5, not studied in this work) does not fully represent the entire electronic structure and that three-dimensionality must be taken into account when studying Ta2_2Ni(Se1−x_{1-x}Sx_x)5_5 compounds. Dimensionality of the underlying electronic structure in a material is an important parameter while choosing potential candidates for electronic applications. Low-dimensional systems have surpassed their bulk counterparts in this field due to their high electrical and thermal conductivities. Our results therefore suggest that the three-dimensional nature of Ta2_2Ni(Se1−x_{1-x}Sx_x)5_5 needs to be taken into account when considering these compounds for applications in the field of electronics and optoelectronics.

ACKNOWLEDGMENTS

This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 654360 NFFA-Europe. The authors acknowledge Elettra Sincrotone Trieste for providing access to its synchrotron radiation facilities and to the BaDElPh beamline that contributed to the results presented in this work. We thank L. Sancin for technical assistance during experiments and J. Mravlje and F. Galdenzi for fruitful discussions during the beamtime.

Supplementary Material for: Electronic band structure in pristine and Sulfur-doped Ta2NiSe5\mathrm{Ta}_2\mathrm{NiSe}_5

Tanusree Saha; Luca Petaccia; Barbara Ressel; Primož Rebernik Ribič; Giovanni Di Santo; Wenjuan Zhao; Giovanni De Ninno

Polarization dependent ARPES data in Ta2Ni(Se0.75S0.25)5\mathrm{Ta}_2\mathrm{Ni}(\mathrm{Se}_{0.75}\mathrm{S}_{0.25})_5 and Ta2Ni(Se0.5S0.5)5\mathrm{Ta}_2\mathrm{Ni}(\mathrm{Se}_{0.5}\mathrm{S}_{0.5})_5

Figure S1 shows the ARPES intensity plots of 25% and 50% Sulphur-doped Ta2NiSe5\mathrm{Ta}_2\mathrm{NiSe}_5 acquired using xzxz (p) and yy (s)-polarized light. The intensity of the top part of the valence band around Γ\Gamma is enhanced for s-polarization [see Figure S1 and (d)] while the dispersive parts of the valence band away from Γ\Gamma (and towards XX) display higher intensities for p-polarized light [see Figure S1 and (c)]. This reveals the mirror symmetries associated with different orbitals forming the valence band in these compounds with respect to the xzxz plane (for our experimental setup).

ARPES intensity maps for Ta2Ni(Se0.5S0.5)5 and Ta2Ni(Se0.75S0.25)5 with p- and s-polarized light

Figure S1. ARPES intensity maps of Ta2Ni(Se0.5S0.5)5\mathrm{Ta}_2\mathrm{Ni}(\mathrm{Se}_{0.5}\mathrm{S}_{0.5})_5 acquired along Γ−X\Gamma-X direction at T=80T=80 K and at photon energy 26 eV using (a) p-polarized light and (b) s-polarized light. The same for Ta2Ni(Se0.75S0.25)5\mathrm{Ta}_2\mathrm{Ni}(\mathrm{Se}_{0.75}\mathrm{S}_{0.25})_5 in (c) and (d), respectively at photon energy 20 eV.

Anisotropic in-plane band dispersions in Ta2_2NiSe5_5 and Ta2_2Ni(Se0.5_{0.5}S0.5_{0.5})5_5

The in-plane anisotropy of the energy band dispersions along Γ−X\Gamma-X and Γ−Y\Gamma-Y directions for Ta2_2NiSe5_5 having 0% and 50% of Sulphur-doping is shown in Figure S2(a), (b) and Figure S2(c), (d), respectively. The valence band near the Fermi level, EFE_{\mathrm{F}}, exhibits a hole-like dispersion centered at Γ\Gamma and is characterized by a stronger dispersion along Γ−X\Gamma-X direction (parallel to the Ta-Ni-Ta chains) than along Γ−Y\Gamma-Y direction (perpendicular to the chains). This indicates that the hopping properties of the carriers are different along the aa-axis and cc-axis.

Anisotropic band dispersions along $\Gamma-X$ and $\Gamma-Y$ directions for Ta$_2$NiSe$_5$ and Ta$_2$Ni(Se$_{0.5}$S$_{0.5}$)$_5$

Figure S2. (a) and (b) Anisotropic band dispersions along Γ−X\Gamma-X and Γ−Y\Gamma-Y directions in Ta2_2NiSe5_5. Band dispersions along Γ−X\Gamma-X and Γ−Y\Gamma-Y in Ta2_2Ni(Se0.5_{0.5}S0.5_{0.5})5_5 in (c) and (d), respectively. A photon energy of 26 eV and s-polarized light is used.

Comparison of photon energy dependence of valence band peak position among Ta2_2Ni(Se1−x_{1-x}Sx_x)5_5 compounds

Figure S3 shows a comparison of the valence band peak position at Γ\Gamma along Γ−X\Gamma- X direction as a function of the incident photon energy for the three doped (0%, 25%, 50%) compounds. The binding energy values have been plotted on the (Eb−Eb,min⁡)/(Eb,max⁡−Eb,min⁡)(E_b - E_{b,\min})/(E_{b,\max} - E_{b,\min}) scale for each of the compounds. Apart from the quantitative differences in the binding energy, EbE_b, values between the three curves (red, black, blue in Figure S3), the overall trend, i.e., a decrease in the binding energy with increasing photon energy, does not show any pronounced qualitative differences between the three compounds, at least within the photon energy range used for this study.

Peak position of the valence band at Gamma as a function of incident photon energy for pure and S-doped Ta2NiSe5

Figure S3. Peak position of the valence band at Γ\Gamma [shown in Figure 7(g)-(Figure 7i)] as a function of incident photon energy for pure and S-doped Ta2_2NiSe5_5, plotted on the same relative scale for comparison. EbE_b stands for the binding energy.

All the data analysis has been performed using Igor Pro (wavemetrics.com) and Casa XPS (http://www.casaxps.com) softwares and the crystal structure in Figure 1(a) was generated in VESTA1^1.

References

  1. [34]M. D. Watson, I. Marković, E. A. Morales, P. Le Févre, M. Merz, A. A. Haghighirad and P. D. C. King, Phys. Rev. Research 2, 013236 (2020).
  2. [35]S. Mor, M. Herzog, D. Golež, P. Werner, M. Eckstein, N. Katayama, M. Nohara, H. Takagi, T. Mizokawa, C. Monney, and J. Stähler, Phys. Rev. Lett. 119, 086401 (2017).
  3. [36]E. Baldini, A. Zong, D. Choi, C. Lee, M. H. Michael, L. Windgätter, I. M. Mazin, S. Latini, D. Azoury, B. Lv, A. Kogar, Y. Wang, Y. Lu, T. Takayama, H. Takagi, A. J. Millis, A. Rubio, E. Demler and N. Gedik, arXiv:2007.02909.
  4. [37]T. Saha, P. D. Golež, G. D. Ninno, J. Mravlje, Y. Murakami, B. Ressel, M. Stupar and P. R. Ribič, Phys. Rev. B 103, 144304 (2021).
  5. [40]For a review of transport reactions see: Harald Schäfer, “Chemical Transport Reactions”. Academic Press, New York, (1964).
  6. [41]L. Petaccia, P. Vilmercati, S. Gorovikov, M. Barnaba, A. Bianco, D. Cocca, C. Mascio vecchio and A.Goldoni, Nuclear Instruments and Methods in Physics Research A 606, 780 (2009).
  7. [42]J. Harmanson, Solid State Commun. 22, 9 (1977).
  8. [43]A. Damascelli, Z. Hussain and Z. X. Shen, Rev. Mod. Phys. 75, 473 (2003).
  9. [44]K. Fukutani, R. Stania, C. Il Kwon, J. S. Kim, K. J. Kong, J. Kim and H. W. Yeom, Nature Physics 17, 1024 (2021).
  10. [45]Y. Wakisaka, T. Sudayama, K. Takubo, T. Mizokawa, N. L. Saini, M. Arita, H. Namatame, M. Taniguchi, N. Katayama, M. Nohara and H. Takagi, J. Supercond. Novel Magn. 25, 1231 (2012).

Paper details

vlnc:W_n3w38th3
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