New Environment for a Two-Dimensional Topological Insulator with

Article pubs.acs.org/cm

New Environment for a Two-Dimensional Topological Insulator with Hexagonal Channels Hosting Diiodido-bismuthate(I) Anions in a Singlet State Bertold Rasche,† Wouter Van den Broek,§ and Michael Ruck*,†,‡ †

Department of Chemistry and Food Chemistry, TU Dresden, 01062 Dresden, Germany Max Planck Institute for Chemical Physics of Solids, 01187 Dresden, Germany § Institute of Experimental Physics, University of Ulm, 89069 Ulm, Germany ‡

S Supporting Information *

ABSTRACT: A honeycomb layer of transition metal-centered bismuth cubes has attracted attention as part of the weak topological insulator (TI) Bi14Rh3I9. The intermetallic layer itself is a two-dimensional TI and has been found in Bi13Pt3I7 and Bi12Pt3I5, as well as in the new material Bi38Pt9I14, yet with different stacking sequences and types of insulating spacer layers in all cases. The arrangements strongly influence the electronic situation, especially the coupling between the TI layers. X-ray diffraction and electron microscopy show that in Bi38Pt9I14 the intermetallic layers are exclusively separated by single layers of iodide ions. The eclipsed stacking of the honeycomb layers results in hexagonal channels. These host hitherto unknown linear [BiI2]− anions with bismuth atoms in the uncommon +I oxidation state. Quantum chemical calculations as well as magnetic susceptibility data indicate a singlet state stabilized by strong spin−orbit coupling. Bi38Pt9I14 is a semiconductor with a very narrow energy gap, as resistivity measurements show. The nontrivial topological invariants of the electronic bands at the Fermi energy and the coupling of TI layers throughout the crystal suggest even a strong TI.

■

INTRODUCTION Topological insulators (TIs)1,2 are in the focus of materials research, as they possess certain surface or edge states in which the spin and momentum of an electron are locked.3,4 This feature strongly suppresses backscattering and lays the foundation for novel types of information processing such as spintronics5 by providing pure spin currents,6 or fault tolerant quantum computation by using the Majorana Fermions at interfaces of these topological states with superconductors.7 In general, bismuth-rich materials have been proven to be a good source in the search of topological nontrivial phases.8−10 The reason is the strong spin−orbit coupling of bismuth, which can open a nontrivial band gap as soon as bismuth states are involved in electronic states close to the Fermi level. Besides, its structural and chemical flexibility, its almost nontoxic properties compared to its neighbors lead, thallium, and mercury, and its availability and price compared to those of gold and platinum provide further advantages. One group of bismuth-rich materials with a manifold of structures are the bismuth subhalides.11,12 This class of materials distinguishes itself by low-dimensional intermetallic structural fragments being incorporated into ionic matrices. That has been proven to lead to very unusual physical properties, such as the coexistence of ferromagnetism and superconductivity.13 © 2015 American Chemical Society

Within this group of materials, the weak TI Bi14Rh3I9 has been discovered and lately been established by scanning tunneling microscopy (STM).14−16 The essential structural motive is a two-dimensional (2D) honeycomb net of edgesharing, transition metal-centered bismuth cubes. Such a layer is itself a 2D TI.16,24 In Bi14Rh3I9, the TI layers are effectively isolated from each other by saltlike iodido-bismuthate(III) spacer layers. Consequently, the TI layers couple only weakly, and their topologically protected one-dimensional edge states can be revealed by STM on the step edges of the (001) surface, which is the cleaving surface.16 Herein, we present the new bismuth-rich compound Bi38Pt9I14, in which an analogous intermetallic honeycomb layer is found in a, so far, unknown environment. Here, neighboring TI layers are separated by only a thin layer of iodide ions, which should strongly influence the coupling between them. Bi38Pt9I14 expands the class of bismuth-rich halides and is the fifth layered structure in the Bi−Pt−I system. While Bi8Pt5I3 and Bi16Pt11I617 are essentially different in their layered structure, Bi13Pt3I718 and Bi12Pt3I519 as well as Bi38Pt9I14 share the Bi−Pt honeycomb motive. Together with the Received: November 18, 2015 Revised: December 30, 2015 Published: December 31, 2015 665

DOI: 10.1021/acs.chemmater.5b04496 Chem. Mater. 2016, 28, 665−672

Article

Chemistry of Materials rhodium compound Bi14Rh3I9, they provide the seed for a family of materials to explore the physics of topological nontrivial materials.

■

SYNTHESIS The existence of five bismuth-rich compounds in the Bi−Pt−I system made a thorough investigation of the conditions under which they form indispensable. We explored the relevant area of the Bi−Pt−I phase system via differential scanning calorimetry (DSC), powder X-ray diffraction (PXRD), and energy dispersive X-ray spectroscopy (EDX). Starting from a mixture of bismuth, platinum, and BiI3, we recorded for the nominal composition “Bi3PtI” an exothermic signal superimposed with endothermic signals that indicate formation reactions (Figure 1). Another sample of the mentioned Figure 2. Heating (blue) and subsequent cooling (red) of Bi38Pt9I14, with bismuth contaminations in a DSC experiment. A and B indicate the decomposition of Bi38Pt9I14 and Bi13Pt3I7, respectively. A′/B′ indicates the simultaneous crystallization of the two compounds.

signal in the relevant range is observed, but both ternary compounds can be identified in the final mixture via PXRD (Figure S2). A repetition of the experiment with a cooling rate of −1 K/h and holding at 290 °C for 5 days yielded mainly crystals of Bi38Pt9I14. All further experiments were performed on these. The composition was confirmed via EDX spectroscopy on several crystals (Figure S3 and Table S2). For a scanning electron microscopy (SEM) picture of such a crystal, see Figure S4.

■

CRYSTAL STRUCTURE Single-crystal X-ray diffraction (SCXRD) revealed that Bi38Pt9I14 crystallizes in hexagonal space group P6/mmm (No. 191) with a = 922.50(2) pm and c = 2671.09(7) pm at room temperature (RT) and a = 919.15(3) pm and c = 2661.79(9) pm at 120 K (Tables 1−5). The dominating structural motive in the layered structure of Bi38Pt9I14 (Figure 3) is the intermetallic layer 2∞[Bi8/2Pt]+4/3. It consists of platinum-centered bismuth cubes (Pt−Bi, 288.7−289.1 pm) that share edges (Figure 4). Thereby, they form a honeycomb (or with respect to the Pt atoms a Kagome) network with a height of 332.8 pm (defined by the Bi−Bi distance along [001]). Within the trigonal- and hexagonal-prismatic voids of this network, Bi−Bi distances are 320.0 and 347.9 pm, respectively. The intermetallic layers stack primitively, i.e., with the same position in the (001) plane, and are separated by single layers of iodide ions (I−Bi, 363.6−367.2 pm). As a consequence of the eclipsed arrangement of the intermetallic layers, their hexagonal voids form channels along stacking direction [001]. These are filled with a linear sequence of iodide and double the number of [BiI2]− anions (Bi−I, 293.1 and 333.7 pm), creating a three-fold superstructure in [001]. A refinement of the occupancies of the channel atoms resulted in almost the same figures of merit [R1(all) = 3.12; wR2(all) = 3.23; goodness of fit on F2 = 1.129] and only marginal deviations from the nominal occupancies, with atom I2 showing the largest deviation (occupancy of 0.98). The chemical compositions obtained from single-crystal structure refinement as well as from EDX measurements strongly support the formula Bi38Pt9I14. In the diffraction pattern, the superstructure reflections are weak because only a small fraction of the total electron density

Figure 1. Heating of a mixture of Bi, BiI3, and Pt in a DSC experiment. The green area indicates the region where formation of the new compound takes place. A and B indicate the decomposition of Bi38Pt9I14 and Bi13Pt3I7, respectively.

composition annealed at 290 °C, i.e., above the temperature of the formation reaction, gave unindexed reflections in the powder X-ray diffraction (PXRD) experiment, which could later be assigned to the new ternary phase Bi38Pt9I14. Although only few, very small crystals could be extracted manually, we were able to determine the approximate composition by energy dispersive X-ray analysis (EDX). Annealing a mixture of bismuth, platinum, and BiI3 with this composition for 1 week at 290 °C yielded Bi38Pt9I14 as the majority phase and bismuth together with Bi2Pt(hP9) as minority phases (Figure S1 of the Supporting Information). Using this sample, we determined Bi38Pt9I14 to be stable up to ∼430 °C in a closed system (Figure 2). It decomposes peritectically into Bi13Pt3I7, Bi2Pt(hP9), and liquid bismuth following the equation Bi38Pt 9I14 → 2Bi13Pt3I 7 + 3Bi 2Pt + 6Bi

This might explain the minority phases in the product. In general, we observed a strong competition between Bi38Pt9I14 and Bi13Pt3I7. Bi13Pt3I7 itself decomposes at 452 °C into Bi2Pt(hP9) and a melt of bismuth and BiI3. Cooling (−2 K/min) the respective homogeneous melt from 650 °C initially led to the crystallization of Bi2Pt(hP9) at approximately 590 °C (depending on the precise Bi:Pt ratio). The crystallization of Bi38Pt9I14 followed instantaneously the crystallization of Bi13Pt3I7 at 454 °C, as only one exothermic 666


Article

Chemistry of Materials

intensities vary monotonously with atomic number, the difference in the occupation within the iodide layers due to the superstructure can be visualized. Every third (dark) line along the stacking direction belonging to an iodide layer becomes even darker, as it has only three instead of four iodide ions per unit cell. The ellipsoids representing the anisotropic displacement parameters of the channel atoms are elongated along [001] with the exception of I5. I5 is at the level of the iodide layer separating the intermetallic networks and thus also has an increased number of lateral vibrations. The linear extrapolation of all principal mean square atomic displacements to 0 K shows no remarkable differences between channel atoms and the rest of the structure (Table S1). Hence, the elongation along [001] seems to be no artifact of slight disorder but is mainly caused by thermal vibrations. The thermal expansion coefficients are Δa/ (aΔT) = 2.1 × 10−5 K−1 and Δc/(cΔT) = 2.0 × 10−5 K−1 and therefore almost the same, which is unusual for a layered material. The intermetallic networks of Bi38Pt9I14, Bi14Rh3I9, Bi13Pt3I7, and Bi12Pt3I5 differ only in details (overview in Figure S5). All M−Bi distances (M = Rh or Pt) are almost equal, which can also be taken as a sign of similar electronic states of M in these compounds. Via comparison of the Bi−Bi distances, it is necessary to differentiate among the height of the layer, the edge of the triangle, and the edge of the hexagon. All these Bi− Bi distances range between the lengths of the primary bonds (307 pm) and the secondary bonds (353 pm), i.e., the intraand interlayer distances, in elemental bismuth. In all intermetallic networks, the Bi−Bi distances are ∼320 pm in the triangles and ∼345 pm in the hexagons. Only the layer height differs significantly between platinum-based (334 pm) and rhodium-based (318 pm) compounds, the latter being even the shortest of all Bi−Bi distances. This feature is primarily associated with the occupation of Bi−Bi antibonding states (electron count) and secondarily with the layer charge. On one hand, a platinum atom contributes one valence electron more to the electronic system of the intermetallic layer than a rhodium atom, yet without influencing the layer charge. On the other hand, an iodido-bismuthate [BiI4]− spacer removes fewer electrons from the intermetallic network than a layer of isolated iodide ions, which directly correlates with the charge of the intermetallic layer. Among the four compounds, Bi14Rh3I9 has the lowest electron count per M atom, and Bi38Pt9I14 and Bi13Pt3I7 have the highest. The distances between the intermetallic layers range between 533 and 563 pm for an iodide layer spacer (which appears only in the platinum compounds) and are around 930 pm in the case of an iodido-bismuthate(III) spacer. A hitherto unknown and striking structural fragment is the linear [BiI2]− ion. The Bi−I distances are 334 and 293 pm. The latter is shorter than in BiI3 [305.4 and 312.5 pm, respectively, distorted octahedral coordination of bismuth(III)]23 and among the shortest reported.20−22 This unusual diiodidobismuthate(I) anion seems to profit from its environment in the hexagonal channels. The bismuth(I) atom and the more remote iodide ion are close to the centers of the bismuth hexagons of the intermetallic layer (Bi···Bi, 345.6 pm; I···Bi, 351.3 pm), while the closer iodide ion of the dumbbell is surrounded by iodide ions separating the intermetallic layers (I···I, 407.0 pm) and therefore not very much supported by its environment.

Table 1. Crystal Data and Structural Refinement for Bi38Pt9I14 at 293 and 120 K empirical formula formula weight temperature (K) wavelength (pm) crystal system space group unit cell dimensions (pm) volume (nm3) no. of formula units per cell density (calculated) (Mg m−3) absorption coefficient (mm−1) crystal size range of data collection index ranges

no. of reflections collected no. of independent reflections data merging absorption correction maximal and minimal transmission refinement method data/restraints/parameters goodness of fit on F2 final R indices [Fo > 4σ(Fo)] R indices (all data) extinction coefficient largest difference peak and hole (e Å−3)

Bi38I14Pt9 11473.65 296(1) 120(2) 71.073 hexagonal P6/mmm a = 922.50(2) a = 919.15(3) c = 2671.09(7) c = 2661.79(9) 1.96857(10) 1.94750(14) Z=1 9.678 9.783 105.956 107.103 0.136 mm × 0.094 mm × 0.012 mm 2θmax = 61° 2θmax = 60° −13 ≤ h ≤ 13 −12 ≤ h ≤ 10, −13 ≤ k ≤ 13 −10 ≤ k ≤ 12, −38 ≤ l ≤ 38 −36 ≤ l ≤ 35 42599 26603 1244 1180 Rint = 0.0492 Rint = 0.0342 Rσ = 0.0113 Rσ = 0.0093 numerical 0.3658 and 0.0072 0.3782 and 0.0081 full-matrix least-squares on F2, anisotropic displacement parameters 1244/0/45 1180/0/45 1.129 1.143 R1 = 0.0219, R1 = 0.0273, wR2 = wR2 = 0.0306 0.0674 R1 = 0.0313, R1 = 0.0330, wR2 = wR2 = 0.0324 0.0695 3.9(2) × 10−5 6.3(7) × 10−5 2.18 and −2.15 2.46 and −2.89

Table 2. Atomic Coordinates and Equivalent Isotropic Displacement Parameters (×10−1 pm2) for Bi38Pt9I14 at 293 Ka atom

Wyckoff position

x

y

z

Ueq

Pt1 Pt2 Bi1 Bi2 Bi3 Bi4 I1 I2 I3 I4 I5

6i 3f 12o 12o 12o 2e 1a 2e 3g 6i 2e

0.5000 0.5000 0.43904(4) 0.43546(4) 0.43175(4) 0.0000 0.0000 0.0000 0.5000 0.5000 0.0000

0.0000 0.0000 0.21952(2) 0.21773(2) 0.21587(2) 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

0.33212(2) 0.0000 0.39459(2) 0.06216(2) 0.26988(2) 0.26204(5) 0.0000 0.38696(9) 0.5000 0.16665(4) 0.15233(9)

12.4(1) 12.0(1) 14.8(1) 18.0(1) 15.5(1) 27.7(2) 29.7(5) 34.8(5) 33.1(3) 27.9(2) 51.5(6)

a

Ueq is defined as one-third of the trace of the orthogonalized Uij tensor.

contributes to them. They also fade with increasing diffraction angles, indicating some long-range disorder, probably between the channels. High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) confirms the regular tripling of the unit cell and does not indicate any stacking faults (Figure 5). As in an HAADF-STEM image the 667


Article

Chemistry of Materials Table 3. Anisotropic Displacement Parameters (×10−1 pm2) for Bi38Pt9I14 at 293 Ka

a

atom

U11

U22

U33

U23

U13

U12


12.8(2) 12.1(2) 14.0(2) 20.5(2) 15.8(2) 17.8(3) 20.9(7) 16.1(5) 48.4(7) 36.5(4) 53.6(9)

9.5(2) 9.4(3) 12.6(1) 15.2(1) 13.1(1) 17.8(3) 20.9(7) 16.1(5) 31.0(8) 30.2(5) 53.6(9)

13.8(2) 13.4(3) 18.3(2) 20.1(2) 18.5(2) 47.4(7) 47.2(16) 72.2(15) 14.2(6) 14.8(4) 47.3(13)

0 0 1.7(1) 3.2(1) −1.5(1) 0 0 0 0 0 0

0 0 3.5(1) 6.5(1) −2.9(1) 0 0 0 0 0 0

4.7(1) 4.7(1) 7.0(1) 10.3(1) 7.9(1) 8.9(1) 10.5(3) 8.0(2) 15.5(4) 15.1(2) 26.8(4)

The exponent of the anisotropic displacement factor takes the form −2π2[h2a × 2U11 + ... + 2hka × b × U12].

Table 4. Atomic Coordinates and Equivalent Isotropic Displacement Parameters (×10−1 pm2) for Bi38Pt9I14 at 120 Ka atom

Wyckoff position

x

y

z

Ueq


6i 3f 12o 12o 12o 2e 1a 2e 3g 6i 2e

0.5000 0.5000 0.43895(5) 0.43549(5) 0.43194(5) 0.0000 0.0000 0.0000 0.5000 0.5000 0.0000

0.0000 0.0000 0.21947(3) 0.21774(3) 0.21597(3) 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

0.33219(2) 0.0000 0.39480(2) 0.06232(2) 0.26978(2) 0.26133(5) 0.0000 0.38592(9) 0.5000 0.16673(4) 0.15146(9)

5.3(2) 5.1(2) 6.2(1) 7.6(1) 6.5(1) 11.1(2) 12.2(5) 13.0(4) 13.2(3) 11.5(2) 20.3(5)

a

Ueq is defined as one-third of the trace of the orthogonalized Uij tensor. Figure 3. Crystal structure of Bi38Pt9I14. The platinum-centered bismuth cubes are colored red.

■

Evidence of the charge of −1 for this three-atom anion, and consequently the bismuth +I oxidation state, was provided by density functional theory (DFT)-based structure optimizations of isolated [BiI2] molecules with different charges of +1, 0, −1, and −2. In all cases, symmetric molecules were obtained and the calculated Bi−I distances were 280.0, 287.9, 301.4, and 322.5 pm, respectively. The experimental mean value of 313.1 pm falls between the distances calculated for charges −1 and −2, excluding the oxidation state of +III for bismuth.

ELECTRONIC STRUCTURE We intended to further corroborate the unusual +I oxidation state of the bismuth atom in the [BiI2]− ion via an analysis of the electronic structure. As absolute charges from quantum chemical calculations, whether from Mulliken population analysis or according to Bader’s quantum theory of atoms in molecules (QTAIM), cannot be transcribed into oxidation states as assigned in chemistry, comparable systems with

Table 5. Anisotropic Displacement Parameters (×10−1 pm2) for Bi38Pt9I14 at 120 Ka

a

atom

U11

U22

U33

U23

U13

U12


5.2(2) 5.2(3) 6.0(2) 8.8(2) 6.3(2) 7.8(3) 9.8(8) 6.4(5) 18.9(6) 14.6(4) 21.1(7)

4.4(3) 4.5(4) 5.5(2) 6.7(2) 5.3(2) 7.8(3) 9.8(8) 6.4(5) 12.5(7) 12.2(5) 21.1(7)

6.1(3) 5.6(4) 7.3(2) 8.1(2) 8.1(2) 17.7(6) 17.0(14) 26.3(11) 6.1(6) 6.8(5) 18.7(10)

0 0 0.6(1) 1.3(1) −0.5(1) 0 0 0 0 0 0

0 0 1.3(1) 2.5(2) −1.1(1) 0 0 0 0 0 0

2.2(1) 2.3(2) 3.0(1) 4.4(1) 3.2(1) 3.9(2) 4.9(4) 3.2(3) 6.2(4) 6.1(3) 10.5(3)

The exponent of the anisotropic displacement factor takes the form −2π2[h2a × 2U11 + ... + 2hka × b × U12]. 668


Article


Figure 4. Two details of the crystal structure of Bi38Pt9I14: (left) hexagonal channels hosting the [BiI2]− anion and (right) [BiI2]− anion with its surroundings.

Figure 6. Resistivity (black) and molar magnetic susceptibility (red) vs temperature for Bi38Pt9I14. A χmol of −0.002 emu/mol indicates the sum of the diamagnetic increments.23 The inset is an Arrhenius plot for the resistivity behavior with linear fits for the extrinsically (green) and intrinsically (orange) dominated regimes. The feature in magnetization near 50 K is very likely due to traces of molecular oxygen in the magnetometer.

diamagnetic increments of −0.002 emu/mol.23 At very low temperatures, impurities lead to a slightly increased magnetic susceptibility (Figure 6 and more detailed Figure S6). On the basis of this evaluation of the [BiI2]− ion, we can assign a charge to the intermetallic layer. The structured formula of Bi38Pt9I14 is then [(Bi8/2Pt4/3+)3]3[(I−)3]3[BiI2−]− 2[I ]. With a charge of +4/3 per Bi8/2Pt cube, the electronic configuration in the intermetallic layer is the same as in Bi13Pt3I7. Moreover, this configuration means that a single intermetallic layer is a 2D TI.16,24 In Bi13Pt3I7, this 2D TI state is canceled out by the so-called “pairing” of TI layers via iodide layers. Within the structure, alternating thick iodido-bismuthate and thin iodide spacers create pairs of intermetallic layers that interact across the short distance and thereby become topologically trivial.16,25−27 Moreover, this interaction turns the compound into a halfmetal. Bi38Pt9I14, however, is a stack of intermetallic layers separated by only iodide spacers. The short interlayer distances provoke interactions between the TI layers. From the viewpoint of theory, different options are possible. It could mean that depending on the number of intermetallic layers the stack could be a TI (odd number of layers) or a trivial insulator (even number of layers).27 Others proposed that the number of layers within the unit cell accounts for the topological character,28 with an odd number yielding a TI. Therefore, the three-fold superstructure of Bi38Pt9I14 is promising. The coupling strength should mainly influence the anisotropy and “shape” of the nontrivial surface state.25,29 Before evaluating the question of topological behavior, we first had to determine whether Bi38Pt9I14 is an insulator at all. The calculated density of states (DOS) suggests a metal, although with a significant drop of the DOS at the Fermi energy. This result is the same for a scalar-relativistic or fullrelativistic approach with a local density approximation (LDA) or a generalized gradient approximation (GGA) exchange correlation potential and for calculations with one spin only or with spin polarization (Figure S7). In contrast, resistivity measurements on powders indicate a semiconductor (Figure

Figure 5. HAADF-STEM image of the [010]* zone of Bi38Pt9I14 in the [010] orientation. Image intensities vary monotonously with atomic number; therefore, the brighter lines belong to the intermetallic layer and the darker lines to the iodide layer. Of the latter, every third is even darker (indicated by arrows), belonging to the less occupied iodide layer due to the superstructure induced by the channel occupation. The inset shows improved contrast in an overlay with the structure model.

confirmed oxidation states are needed. Fortunately, with Bi14Rh3I9 and Bi13Pt3I7, such systems are known, with bismuth atoms in the iodido-bismuthate spacers having the common oxidation state of +III. The Mulliken analysis yields charges of +0.6 and +0.5 for such bismuth atoms in Bi14Rh3I9 and Bi13Pt3I7, respectively. The Mulliken charge of +0.2 for the bismuth atom in the [BiI2]− ion is much smaller, supporting the +I oxidation state. The QTAIM analysis gives similar results, as the bismuth(III) atoms in Bi13Pt3I7 and Bi14Rh3I9 yield a charge of +1, while the bismuth(I) atom of the [BiI2]− ion has a QTAIM charge of +0.2. In combination with the structure optimization argument mentioned above, we see strong evidence for bismuth in the +I oxidation state. One might expect a magnetic moment for a bismuth(I) atom, as two degenerate 6p orbitals should be occupied with unpaired electrons (triplet state). However, strong spin−orbit coupling lifts the degeneracy of the 6p orbitals, resulting in an energetically lower 6p1/2 and two degenerate energetically higher 6p3/2 orbitals, i.e., a singlet state. Accordingly, the calculated magnetic moments of an isolated [BiI2]− ion are 2 μB from a scalar-relativistic approach and 0 μB from a fullrelativistic approach. In accord with the predicted singlet state, measurements of the magnetic susceptibility of Bi38Pt9I14 indicate diamagnetic behavior with χmol matching the sum of 669


Article


Scanning Transmission Electron Microscopy. The single crystal was embedded in the epoxy resin Epon (Fluka) that was then polymerized at 60 °C. Subsequently, thin lamellae were cut perpendicular to the largest facet, i.e., along the stacking direction, with an ultramicrotome (Ultracut, Leica Microsystems) equipped with a diamond knife. The nominal lamella thickness was approximately 50 nm, while locally, crystalline flakes with a thickness of ∼10 nm were observed. High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) studies were performed at the University of Ulm on a FEI Titan F20 microscope with image correction, operating at 300 kV. Thermal Analysis. Differential scanning calorimetry (DSC) was performed on the heat-flux DSC Setaram Labsys ATD-DSC instrument with a type K sample holder (Ni-Cr/Ni-Al) against an Al2O3 reference. Starting materials were fused in silica ampules of 0.1 mL. Powder X-ray Diffraction. Data collection was performed in Bragg−Brentano geometry at 293(1) K on an X’Pert Pro MPD diffractometer (PANalytical) equipped with a curved Ge(111) monochromator using Cu Kα1 radiation (λ = 154.056 pm). X-ray Crystal Structure Determination. X-ray diffraction data were collected at 293(1) and 120 K on an APEX II Kappa CCD single-crystal diffractometer (Bruker) using graphite-monochromatized Mo Kα radiation [λ = 71.073 pm (Table 1)]. Integration, polarization, and Lorentz factor corrections of the data were conducted within the APEX2 suite.34 Optimization of the observed crystal shape based on symmetry-equivalent reflections in the corresponding Laue class and the numerical absorption corrections were performed with X-Red and X-Shape.35,36 For the structure solution with charge flipping methods, Jana2006 was employed.37,38 The refinement was conducted in the SHELX2014 program suite.39 Atomic parameters are listed in Tables 2−5. Graphical representations of the structure were developed with Diamond.40 Detailed crystallographic data can be obtained from the Fachinformationszentrum Karlsruhe, 76344 Eggenstein-Leopoldshafen, Germany [Fax, +(49) 7247-808-666; E-mail, crysdata@fiz-karlsruhe.de]. Inquiries should mention the depository numbers CSD-430467 (Bi38Pt9I14 at 296 K) and CSD-430468 (Bi38Pt9I14 at 120 K). Electrical Resistivity. Powders of the two compounds were coldpressed to cylindrical pellets in a sapphire die cell. The electrical resistivity was measured between 4 and 320 K using four platinum contacts in the Van der Pauw setup. Magnetization Measurements. The samples were filled in precalibrated silica tubes. The magnetization was measured in fields of μ0H = 2 mT and 7 T in a SQUID magnetometer (MPMS-XL7, Quantum Design). A Honda-Owen type correction to the susceptibility for ferromagnetic traces was applied. Quantum Chemical Calculations. All full-relativistic calculations were performed with the Full-Potential Local-Orbital (FPLO) code,41 version 14.00, within the local density approximation (LDA) using the parametrization PW9242 and the generalized gradient approximation (GGA) using the parametrization PBE.43 The Blöchl corrected linear tetrahedron method with a 12 × 12 × 4 k mesh for Bi38Pt9I14 was employed, after checking for convergence with respect to the number of k points. Spin−orbit coupling was implemented on the level of the four-component Dirac equation. Topological invariants were calculated according to the supplement of ref 15. The δi values for the considered bands, calculated from the parity eigenvalues for all timereversal invariant momenta (TRIM), are listed in Table S3. The following basis states are treated as valence states: Bi, 5s, 5p, 5d, 6s, 7s, 6p, 7p, 6d; Pt, 5s, 5p, 5d, 6s, 6p, 6d, 7s; I, 4s, 4p, 4d, 5s, 6s, 5p, 6p, 5d. Mulliken population analysis44 was directly provided by FPLO. Charges according to the quantum theory of atoms in molecules (QTAIM) developed by Bader45 were calculated with the DGrid program package46 via a topological analysis of the electron density, which was provided by an FPLO module.47

6). From an Arrhenius plot (inset of Figure 6), an extrinsically dominated regime and an intrinsically dominated regime can be identified, with energy gaps of ∼0.2 and ∼8 meV, respectively. These very narrow gaps could explain why the DFT-based calculations did not yield a gap at the Fermi energy, as this is a known issue of the method.30 With Bi38Pt9I14 established as a (narrow-gap) semiconductor, the question of topological nontrivial behavior remains. As the calculations yield no gap, the actual character of the measured gap can potentially be identified only via angle-resolved photoelectron spectroscopy (ARPES), via scanning tunneling microscopy (STM), or maybe via 209Bi nuclear magnetic resonance (NMR), as a correlation between topological nature and chemical shielding has been reported recently.31 However, the very narrow gap will also hamper these experiments. At least the topological invariants for the highest fully occupied and all partially occupied bands at the Fermi energy are nontrivial, and in three of four cases, the strong topological invariant is 1, meaning that Bi38Pt9I14 could even be a strong three-dimensional (3D) TI (Table S2).

■

CONCLUSIONS The presented structure of Bi38Pt9I14 is a new member of the class of bismuth subhalides, and in particular in a family of layered structures with a 2D TI as the dominating structural motive. As the environment for the 2D TI within this structure is unique and complements the known environments in Bi14Rh3I9, Bi13Pt3I7, and Bi12Pt3I5, it depicts a potentially fascinating system for studying TI physics. Because the compound has been established as an insulator (semiconductor), an advanced theoretical study is necessary to gain insight into the electronic structure, and to investigate whether Bi38Pt9I14 is a strong 3D TI. Moreover, hexagonal channels in Bi38Pt9I14 host the novel linear anion [BiI2]−, with a bismuth atom in the uncommon +I oxidation state. Apart from its existence, it is a very interesting case in which the expected triplet state is less stable than the singlet state created by the strong spin−orbit coupling of the bismuth atom.

■

EXPERIMENTAL SECTION

Starting Materials. Bismuth (Merck) was treated with streaming H2 at 220 °C before being used. Platinum (ChemPur, >99.95% metal base) was used without further purification. BiI3 was synthesized from the elements and sublimed twice.32 Synthesis. Bismuth, platinum, and BiI3 in a 38:9:14 Bi:Pt:I molar ratio were ground in an argon-filled glovebox [MBraun UNIlab; p(O2)/p0 < 1 ppm; p(H2O)/p0 < 1 ppm] and sealed in a silica ampule (0.1 mL for DSC measurements and 3.5 mL for further synthesis). Powders were obtained by annealing the mixture at 290 °C for 5 days, followed by quenching to room temperature (RT). Single crystals were grown by heating (2 K/min) Bi38Pt9I14 powder to 430 °C and subsequent slow cooling (−1 K/min) to 290 °C. After being held at 290 °C for 5 days, the ampule was quenched to RT. Scanning Electron Microscopy. Scanning electron microscopy (SEM) was performed using a SU8020 instrument (Hitachi) with a triple detector system for secondary and low-energy backscattered electrons (Ua = 2.5 kV). Energy Dispersive X-ray Analysis. The compositions of selected crystals were determined by semiquantitative energy dispersive X-ray analysis (Ua = 30 kV) using a SU8020 SEM instrument (Hitachi) equipped with a Silicon Drift Detector X-MaxN (Oxford Instruments). The data were processed (integration, pulse pile-up correction) within the AZtec software package.33 670


Article


■

(10) Isaeva, A.; Rasche, B.; Ruck, M. Bismuth-Based Candidates for Topological Insulators: Chemistry beyond Bi2Te3. Phys. Status Solidi RRL 2013, 7, 39−49. (11) Ruck, M. From the Metal to the MoleculeTernary Bismuth Subhalides. Angew. Chem., Int. Ed. 2001, 40, 1182−1193. (12) Ruck, M. Between Covalent and Metallic Bonding: From Clusters to Intermetallics of Bismuth. In Reference Module in Chemistry, Molecular Sciences and Chemical Engineering; Elsevier: Amsterdam, 2015. (13) Herrmannsdörfer, T.; Skrotzki, R.; Wosnitza, J.; Köhler, D.; Boldt, R.; Ruck, M. Structure-Induced Coexistence of Ferromagnetic and Superconducting States of Single-Phase Bi3Ni Seen via Magnetization and Resistance Measurements. Phys. Rev. B: Condens. Matter Mater. Phys. 2011, 83, 140501. (14) Rasche, B.; Isaeva, A.; Ruck, M.; Borisenko, S.; Zabolotnyy, V.; Büchner, B.; Koepernik, K.; Ortix, C.; Richter, M.; van den Brink, J. Stacked Topological Insulator Built from Bismuth-Based Graphene Sheet Analogues. Nat. Mater. 2013, 12, 422−425. (15) Rasche, B.; Isaeva, A.; Gerisch, A.; Kaiser, M.; Van den Broek, W.; Koch, C. T.; Kaiser, U.; Ruck, M. Crystal Growth and Real Structure Effects of the First Weak 3D Stacked Topological Insulator Bi14Rh3I9. Chem. Mater. 2013, 25, 2359−2364. (16) Pauly, C.; Rasche, B.; Koepernik, K.; Liebmann, M.; Pratzer, M.; Richter, M.; Kellner, J.; Eschbach, M.; Kaufmann, B.; Plucinski, L.; Schneider, C. M.; Ruck, M.; van den Brink, J.; Morgenstern, M. Subnanometre-Wide Electron Channels Protected by Topology. Nat. Phys. 2015, 11, 338−343. (17) Rasche, B.; Schnelle, W.; Ruck, M. The Bismuth Subiodides Bi8Pt5I3 and Bi16Pt11I6 − Layered Metals with Covalent Platinum Networks. Z. Anorg. Allg. Chem. 2015, 641, 1444−1452. (18) Ruck, M. Bi13Pt3I7: Ein Subiodid mit einer pseudosymmetrischen Schichtstruktur. Z. Anorg. Allg. Chem. 1997, 623, 1535−1541. (19) Kaiser, M.; Rasche, B.; Isaeva, A.; Ruck, M. Low-Temperature Topochemical Transformation of Bi13Pt3I7 into the New Layered Honeycomb Metal Bi12Pt3I5. Chem. - Eur. J. 2014, 20, 17152−17160. (20) Hsueh, H. C.; Chen, R. K.; Vass, H.; Clark, S. J.; Ackland, G. J.; Poon, W. C.-K.; Crain, J. Compression Mechanisms in Quasimolecular XI3 (X = As, Sb, Bi) Solids. Phys. Rev. B: Condens. Matter Mater. Phys. 1998, 58, 14812−14822. (21) Robertson, B. K.; McPherson, W. G.; Meyers, E. A. Crystal Structures of Bismuth Halide Complex Salts. I. 2-Picolinium Tetrabromobismuthate (III) and Tetraiodobismuthate (III). J. Phys. Chem. 1967, 71, 3531−3535. (22) Sidey, V. I.; Voroshilov, Y. V.; Kun, S. V.; Peresh, E. Y. Crystal Growth and X-Ray Structure Determination of Rb3Bi2I9. J. Alloys Compd. 2000, 296, 53−58. (23) Lueken, H. Magnetochemie: Eine Einführung in Theorie und Anwendung; B. G. Teubner: Stuttgart, Germany, 1999. (24) Rasche, B.; Isaeva, A.; Ruck, M.; Koepernik, K.; Richter, M.; van den Brink, J. Correlation between Topological Band Character and Chemical Bonding in a Bi14Rh3I9-Based Family of Insulators. Scientific Reports. Sci., 2016, 6, just accepted. (25) Ringel, Z.; Kraus, Y. E.; Stern, A. Strong Side of Weak Topological Insulators. Phys. Rev. B: Condens. Matter Mater. Phys. 2012, 86.10.1103/PhysRevB.86.045102 (26) Mong, R. S. K.; Bardarson, J. H.; Moore, J. E. Quantum Transport and Two-Parameter Scaling at the Surface of a Weak Topological Insulator. Phys. Rev. Lett. 2012, 108, 076804. (27) Yan, B.; Müchler, L.; Felser, C. Prediction of Weak Topological Insulators in Layered Semiconductors. Phys. Rev. Lett. 2012, 109, 116406. (28) Kim, Y.; Kane, C. L.; Mele, E. J.; Rappe, A. M. Layered Topological Crystalline Insulators. Phys. Rev. Lett. 2015, 115, 086802. (29) Lau, A.; Ortix, C.; van den Brink, J. One-Dimensional Dirac Electrons on the Surface of Weak Topological Insulators. Phys. Rev. B: Condens. Matter Mater. Phys. 2015, 91, 085106. (30) Martin, R. M. Electronic Structure: Basic Theory and Practical Methods; Cambridge University Press: Cambridge, U.K., 2004.

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.chemmater.5b04496. Extrapolated principal mean square displacements, powder diffraction patterns for analysis of the synthesis, additional electron microscopy data, comparison of distances for all compounds with the honeycomb TI layer, detailed magnetic susceptibility curves, density of states for different methods, and calculated topological invariants (PDF)

■

AUTHOR INFORMATION

Corresponding Author

*Department of Chemistry and Food Chemistry, TU Dresden, 01062 Dresden, Germany. E-mail: [email protected]. Funding

German Research Foundation (DFG). W.V.d.B. acknowledges financial support by the Carl Zeiss Foundation. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We thank A. Isaeva and A. Baranov for helpful discussions, W. Schnelle and R. Koban for the magnetization and resistivity measurements, and U. Burkhardt for additional energy dispersive X-ray measurements. We are grateful to E. Schmid for the ultramicrotomy and to P. Walther and U. Kaiser for the electron microscopy time. We are indebted to ZIH TU Dresden for computational facilities.

■

ABBREVIATIONS TI, topological insulator; DSC, differential scanning calorimetry; PXRD, powder X-ray diffraction; SCXRD, single-crystal X-ray diffraction; QTAIM, quantum theory of atoms in molecules; STM, scanning tunneling microscopy; STEM, scanning transmission electron microscopy

■

REFERENCES

(1) Kane, C. L.; Mele, E. J. Z2 Topological Order and the Quantum Spin Hall Effect. Phys. Rev. Lett. 2005, 95, 146802. (2) Kane, C. L.; Mele, E. J. Quantum Spin Hall Effect in Graphene. Phys. Rev. Lett. 2005, 95, 226801. (3) Qi, X.-L.; Zhang, S.-C. Topological Insulators and Superconductors. Rev. Mod. Phys. 2011, 83, 1057−1110. (4) Hasan, M. Z.; Kane, C. L. Colloquium: Topological Insulators. Rev. Mod. Phys. 2010, 82, 3045−3067. (5) Moore, J. E. The Birth of Topological Insulators. Nature 2010, 464, 194−198. (6) Brüne, C.; Roth, A.; Buhmann, H.; Hankiewicz, E. M.; Molenkamp, L. W.; Maciejko, J.; Qi, X.-L.; Zhang, S.-C. Spin Polarization of the Quantum Spin Hall Edge States. Nat. Phys. 2012, 8, 486−490. (7) Alicea, J. New Directions in the Pursuit of Majorana Fermions in Solid State Systems. Rep. Prog. Phys. 2012, 75, 076501. (8) Ando, Y. Topological Insulator Materials. J. Phys. Soc. Jpn. 2013, 82, 102001. (9) Cava, R. J.; Ji, H.; Fuccillo, M. K.; Gibson, Q. D.; Hor, Y. S. Crystal Structure and Chemistry of Topological Insulators. J. Mater. Chem. C 2013, 1, 3176. 671


Article

Chemistry of Materials (31) Nowak, B.; Kaczorowski, D. NMR as a Probe of Band Inversion in Topologically Nontrivial Half-Heusler Compounds. J. Phys. Chem. C 2014, 118, 18021−18026. (32) Ruck, M. Darstellung und Kristallstruktur von fehlordnungsfreiem Bismuttriiodid. Z. Kristallogr. - Cryst. Mater. 1995, 210, 650− 655. (33) AZtec; Oxford Instruments Technology Tools Ltd.: Abingdon, U.K., 2013. (34) APEX2-Software; Bruker AXS Inc.: Madison, WI, 2014. (35) X-Red32; STOE & Cie GmbH: Darmstadt, Germany, 2009. (36) X-Shape, Program for Crystal Optimisation for Numerical Absorption Correction; STOE & Cie GmbH: Darmstadt, Germany, 1999. (37) Jana2006, The Crystallographic Computing System; Institute of Physics: Praha, Czech Republic, 2015. (38) Petricek, V.; Dusek, M.; Palatinus, L. Crystallographic Computing System JANA2006: General Features. Z. Kristallogr. Cryst. Mater. 2014, 229, 345−352. (39) Sheldrick, G. M. SHELX2014, Program for Crystal Structure Determination; University of Göttingen: Göttingen, Germany, 2014. (40) Brandenburg, K. DIAMOND 3.2k, Crystal and Molecular Structure Visualization; Crystal Impact GbR: Bonn, Germany, 2010. (41) Koepernik, K.; Eschrig, H. Full-Potential Nonorthogonal LocalOrbital Minimum-Basis Band-Structure Scheme. Phys. Rev. B: Condens. Matter Mater. Phys. 1999, 59, 1743−1757. (42) Perdew, J. P.; Wang, Y. Accurate and Simple Analytic Representation of the Electron-Gas Correlation Energy. Phys. Rev. B: Condens. Matter Mater. Phys. 1992, 45, 13244. (43) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient Approximation Made Simple. Phys. Rev. Lett. 1996, 77, 3865−3868. (44) Mulliken, R. S. Electronic Population Analysis on LCAO−MO Molecular Wave Functions. I. J. Chem. Phys. 1955, 23, 1833−1840. (45) Bader, R. F. W. Atoms in Molecules: A Quantum Theory; Oxford University Press: Oxford, U.K., 1990. (46) Kohout, M. DGrid; Radebeul, Germany, 2013. (47) Ormeci, A.; Rosner, H.; Wagner, F. R.; Kohout, M.; Grin, Y. Electron Localization Function in Full-Potential Representation for Crystalline Materials. J. Phys. Chem. A 2006, 110, 1100−1105.

672


New Environment for a Two-Dimensional Topological Insulator with

Recommend Documents