Linear, Hypervalent Se34– Units and Unprecedented Cu4Se9

Article pubs.acs.org/IC

Linear, Hypervalent Se34− Units and Unprecedented Cu4Se9 Building Blocks in the Copper(I) Selenide Ba4Cu8Se13 Stefan Maier, Olivier Perez, Denis Pelloquin, David Berthebaud, Sylvie Hébert, and Franck Gascoin* Laboratoire CRISMAT UMR 6508, CNRS ENSICAEN, 6 boulevard du Maréchal Juin, 14050 Caen Cedex 04, France S Supporting Information *

ABSTRACT: Single-crystal and polycrystalline Ba4Cu8Se13 were synthesized; the average crystal structure was solved by single-crystal X-ray diffraction, and the structural model was confirmed by a detailed electron microscopy study of polycrystalline Ba4Cu8Se13. The title compound can be rationalized as (Ba2+)4(Cu+)8(Se2−)2(Se22−)4(Se34−) and crystallizes in a new structure type (space group C2/c with a = 9.171(8) Å, b = 9.146(8) Å, c = 27.35(3) Å, β = 93.21(3)°, and V = 2291 Å3). It contains unprecedented Cu4Se9 fragments with planar Cu rectangles. These fragments form two-dimensional layers via regular (2c-2e) Se−Se bonds. Two of these layers are then connected in the third dimension via linear, hypervalent Se34− units, resulting in “sandwichlike”, layered building blocks, which are stacked along c and separated by Ba. Ba4Cu8Se13 is the first example where Se22− and Se34− groups coexist. We were able to visualize the crystal structure by recording HAADF images, which clearly reveal the Cu4Se9 fragments and linear Se34− units. The title compound is a charge-balanced semiconductor and possesses a large Seebeck coefficient (380 μV K−1 at 200 K) and a low thermal conductivity (0.77 W m−1 K−1 at 200 K)two requirements for efficient thermoelectric materials.

■

INTRODUCTION Polychalcogenides exhibit a rich structural chemistry, including a variety of complex anionic frameworks with homonuclear bonds between negatively charged chalcogen atoms (Q).1,2 In chalcogen-rich tellurides, selenides, and sulfides helical Qn2− (n = 1−13) chains are typically found.3,4 More uncommon motifs such as linear chains5 and square-planar nets (both distorted and undistorted variants)6,7 especially occur in polytellurides. The size of the anionic fragment depends on the number of electrons transferred to the chalcogen atom. Quasi-molecular units and homonuclear Q−Q bonds can form if the anionic framework is highly electron rich.8 By using a charge transfer from a strong cation such as Ba2+ toward a Cu−Se network, crystal structures with Se22− or linear, hypervalent Se34− units become accessible. The existence of linear Se34− units was first discovered in Ba2Ag4−xCuxSe5,4 which is the only example up to now where the Se−Se angle in the Se34− fragment is 180°, while an almost linear Se34− fragment (Se−Se angle of 164°) has been reported for Rb12Nb6Se35.9 A variety of other copper chalcogenides such as ACu4Q3 (A = K, Rb, Cs, Tl and Q = S, Se),10−14 Na3Cu4S4,15,16 NaCu6Se4,17 and TlCu2Q2 (Q = S. Se)18 have a single negative charge on the chalcogen atom without the presence of chalcogen−chalcogen bonds, which underlines the rich structural chemistry of polar ternary copper chalcogenides. Originally these compounds with the latest example NaCu4Se319 were separated into two categoriesthose which contain only Cu+ and Q2− 20−22 and those with mixed-valent Cu+/2+ and Q−Q bonds. BaCu2Se2, which belongs to the first © 2017 American Chemical Society

category, is the only known compound in the ternary system Ba/Cu/Se, which has received considerable attention for its promising thermoelectric properties.23,24 Here we report on the discovery of the second member of this system, Ba 4 Cu 8 Se 13 , and its transport properties. Ba4Cu8Se13 is the first compound with Cu4Se9 building blocks containing planar Cu rectangles and an anionic framework with coexisting Se22− and linear, hypervalent Se34− units.

■

EXPERIMENTAL SECTION

Single-Crystal Growth. Single crystals (black plates with metallic luster) with a nominal composition of Ba4Cu8Se13 were grown by classical high-temperature solid-state reactions. The starting composition of the reactions was BaCu2Se4. These reactions are part of exploratory research efforts aimed at the discovery of new copper chalcogenides (a promising class of thermoelectric materials) within the Ba/Cu/Se system. Within 7 h, stoichiometric amounts of Cu shot (Alfa Aesar, 99.5%), Ba pieces (Alfa Aesar, 99+%), and selenium shot (Alfa Aesar, 99.999%) were heated to 500 °C in carbon-coated and flame-sealed silica ampules. The reaction temperature was kept constant for 24 h, subsequently raised to 800 °C within 8 h, and kept constant for another 48 h before cooling the samples to 200 °C within 96 h. The plateau at 500 °C was chosen in order to minimize Se losses during the reaction. This is where this synthesis differs from that reported for BaCu2Se2, which was synthesized over 3 days at 1000 °C. Single-Crystal Structure Determination. Ba4Cu8Se13 is airstable, and therefore single crystals were manually selected in air. Preliminary investigations were performed using a Bruker Nonius fourReceived: May 17, 2017 Published: July 13, 2017 9209

DOI: 10.1021/acs.inorgchem.7b01224 Inorg. Chem. 2017, 56, 9209−9218

Article

Inorganic Chemistry

Figure 1. (a) (h0l)* (left) and (h1l)* (right) planes obtained from experimental X-ray diffraction frames collected on a Ba4Cu8Se13 single crystal. The unit cell (red) corresponds to a = 9.171(8) Å, b = 9.146(8) Å, c = 27.35(3) Å, and β = 93.21(3)°; the profile of a segment of the [13l]* row is depicted in green and shows a quasi-continuum along c*. (b) Schematic drawing of the (h0l)* diffraction plane when only two twin domains associated with a 2-fold axis parallel to c* (one twin domain in black, the second one in red) are considered in the structural model. (c) All observed reflections extracted from the experimental frames of the single-crystal X-ray diffraction experiment projected into the average unit cell. A weak enhancement of the reflection density at 0.31c* allows the definition of the wave vector q = 0.31c* (green). (d) Schematic drawing showing the (h0l)* diffraction plane expected for the cell a = 9.171(8) Å, b = 9.146(8) Å, c = 27.35(3) Å, β = 93.21(3)° when a modulation vector of q = 0.31c* and twinning according to (b) is considered in the structural model.

Table 1. General Crystallographic Information Obtained from Single-Crystal Structure Solution and Refinement formula wt (amu) space group a (Å) b (Å) c (Å) β (deg) V (Å3) Z F(000) ρcalcd (g cm−3) T (K) cryst dimens (mm) radiation μ(Mo Kα) (mm−1) 2θ limits (deg) data collected

521.1 C2/c (No. 15) 9.171(8) 9.146(8) 27.35(3) 93.21(3) 2290.791 16 3592 6.0432 293 0.096 × 0.101 × 0.249 Mo Kα 34.655 2.98−67.28 −9 ≤ h ≤ 14, −14 ≤ k ≤ 9, −35 ≤ l ≤ 41

no. of measd rflns no. of unique rflns no. of unique rflns with I > 3σ(I) no. of params R(F) for Fo2 > 2σ(Fo2) Rw(Fo2)/Rint goodness of fit Δρmax, Δρmin (e Å−3)

14316 3991 3152 136 0.1250 0.2087/0.0468 1.74 11.07, −5.75

circle diffractometer equipped with an Apex II CCD detector and a Mo Kα Incoatec Microfocus Source (λ = 0.71073 Å). Analyzing the experimental frames revealed overlapping Bragg reflections, diffuse

intensities, and the absence of clear periodicities along a specific direction of the reciprocal space. Hence, not all observed reflections can be indexed in a straightforward fashion. Several crystals exhibit the 9210

DOI: 10.1021/acs.inorgchem.7b01224 Inorg. Chem. 2017, 56, 9209−9218

Article

Inorganic Chemistry

Figure 1 a,d). A detailed interpretation of these findings can be found in the section Nanotwinning, Structural Modulations, and Real Structure Effects along c*. Synthesis of Polycrystalline Ba4Cu8Se13. Small amounts (500 mg) of polycrystalline Ba4Cu8Se13 were synthesized by classical hightemperature solid-state reactions under the same conditions as for the Ba4Cu8Se13 single crystals. The synthesis was upscaled to 5 g by ball milling (15 cycles of 2 min with a milling speed of 700 rpm) starting from a stoichiometric mixture of the elements, i.e., Cu shot (Alfa Aesar, 99.5%), Ba pieces (Alfa Aesar, 99+%), and selenium shot (Alfa Aesar, 99.999%), and using a Pulverisette7 planetary micro mill (Fritsch, Germany), tungsten carbide reaction containers, and balls with a diameter of 10 mm. Spark Plasma Sintering. A 5 g portion of ball-milled powder was sintered (HP D 25/1, FCT, Germany) using high-density graphite dies (Carbon-Lorraine, France), which resulted in cylindrical samples with a diameter and thickness of 15 and 5 mm, respectively. The densification was conducted for 30 min at 400 °C with a heating rate of 23 °C min−1 up to 250 and 3 °C min−1 between 250 and 400 °C, a cooling rate of 13 °C min−1 and an applied mechanical pressure of 51 MPa. The experimental density was determined using the Archimedes method and was found to be 95% of the theoretical density. Powder X-ray Diffraction and Le Bail Refinement. The average crystal structure of polycrystalline Ba4Cu8Se13 synthesized by ball milling was confirmed before and after spark plasma sintering using powder X-ray diffraction. Data were collected on a Bruker D8 Advance Vario1 diffractometer, equipped with a Cu Kα radiation source (λ = 1.54060 Å) and a Johansson monochromator. A Le Bail refinement was carried out using Jana2006, and the average structural model was obtained by single-crystal X-ray diffraction. Instrument parameters were refined using the fundamental approach, the peak shape was modeled using a pseudo-Voigt function, and Legendre polynomials were used to fit the background. Electron Microscopy and Energy Dispersive X-ray Spectroscopy (EDS). A compositional mapping was performed on a Ba4Cu8Se13 single crystal using scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDS). The analysis was carried out on a Supra 55 scanning electron microscope (Zeiss, Germany) equipped with an EDAX EDS detector (Ametek, USA). Transmission electron microscopy (TEM) experiments were performed on single-crystal and polycrystalline Ba4Cu8Se13, synthesized under the same conditions, in order to confirm the structural model obtained from single-crystal X-ray diffraction and to explain the presence of real structure effects observed during the single-crystal structure analysis. A JEM-ARM200F microscope from JEOL was used for all the analyses, and HAADF images were simulated using the software JEMS. Differential Scanning Calorimetry (DSC). A DSC scan with a heating rate of 10 K min−1 and constant Ar flux was performed on polycrystalline Ba4Cu8Se13 synthesized by classical high-temperature solid-state reactions using a NETZSCH STA 449 calorimeter. Thermoelectric Characterization. The TTO option in a Physical Properties Measurements System (Quantum Design, US) was used to measure simultaneously the low-temperature Seebeck coefficient, electrical resistivity, and thermal conductivity on a bar-shaped (2.78 × 2.61 × 11.17 mm), densified bulk sample synthesized by ball milling and subsequent spark plasma sintering.

same features, which seem to be typical for crystals obtained by the previously described synthesis route. This led us to the conclusion that these effects are related to intrinsic real structure effects in the crystal structure. A full data collection was performed at room temperature, and the diffraction frames were analyzed using the Apex II suite, CrysAlisPro and Jana2006.25 The a and b cell parameters could be determined without problems, while the periodicity along the c* direction could not be obtained in a straightforward fashion (cf. Figure 1a), since a quasi-continuum of intensities can be observed for some reciprocal rows of reflections running along c* (see profile of the [31l]* row in Figure 1b). However, on the basis of the strongest reflections, an average periodicity was chosen, leading to the following cell parameters: a = 9.171(8) Å, b = 9.146(8) Å, c = 27.35(3) Å, and β = 93.21(3)°. Analyzing the oriented diffraction planes assembled from the experimental frames (cf. Figure 1a) revealed that the limiting diffraction condition hkl: h + k = 2n is fulfilled, which is consistent with a C centering of the lattice. In the (h0l)* planes the l = 2n condition seems to be fulfilled, which is characteristic for the existence of a c glide plane perpendicular to the b direction. Hence, C2/c was chosen as the space group to proceed with, since the analysis of the reciprocal space is consistent with the space group determined by CellNow.26 At this step, a new data collection was performed at 100 K in order to verify whether or not the observed real structure effects along c* has its origin in temperature-induced disorder in the structure and to obtain diffraction patterns with reduced diffusion along c* to simplify the structure solution. The diffraction patterns show no difference from those collected at room temperature: i.e., the real structure effects causing the overlapping Bragg reflections, diffusion, and the absence of clear periodicities along c* are also present at 100 K. We now discuss the problem of coexistence of periodicities along the c* direction. Pseudomerohedral twinning was included, where a 2-fold axis parallel to a was considered as the twin element. Including this type of twinning allowed a description of one part of the unindexed reflections (cf. Figure 1b). The integration of the data set collected at room temperature was performed considering the cell parameters a = 9.171(8) Å, b = 9.146(8) Å, c = 27.35(3) Å, and β = 93.21(3)° and twinning. The integration and subsequent absorption correction was performed using the Apex II suite and TWINABS.27 The structure was solved using the charge-flipping algorithm implemented in SUPERFLIP28 and refined with Jana2006. Anisotropic atomic displacement parameters were introduced for all the atoms, and refining the site occupancies resulted in a full occupancy of all atomic positions. The final refinement led to a relatively high value for the reliability factor (RF = 12.5%) and the presence of a fairly large residual electron density mainly in the vicinity of the Ba atoms. These features will be discussed in Nanotwinning, Structural Modulations, and Real Structure Effects along c*. However, we obtained a first, chemically reasonable structural model of the average crystal structure by single-crystal X-ray diffraction, which is discussed in Description of the Average Crystal Structure. Detailed information on the data collection and refinement are summarized in Table 1. Equivalent isotropic displacement factors and positional parameters as well as selected interatomic distances and bond angles are summarized in Tables S1 and S2 in the Supporting Information. However, the combination of the chosen unit cell and of two twin domains is not sufficient to describe the reflection sequences observed along the reciprocal rows of reflections along c*. For further investigations, the position of all reflections was gathered from the experimental frames, excluding any information concerning cell parameters. This list of reflections was then introduced into the indexing procedure of Jana2006. Once the previously determined unit cell was introduced, all reflections were projected into the origin of the reciprocal cell (cf. Figure 1c). They show, as expected for a regular crystal, a large density of reflections located at the corners of the parallelepiped. In addition, we observe lines of reflections along the c* direction. An accurate analysis of these lines shows a slight enhancement in reflection density at the positions ±0.31c* (cf. Figure 1c). Hence, including these reflections and describing Ba4Cu8Se13 as a modulated structure are necessary in order to obtain a better description of the observed real structure effects along c* (compare

■

RESULTS AND DISCUSSION Description of the Average Crystal Structure. The crystal structure of Ba4Cu8Se13 (cf. Figure 2) is built up by unprecedented Cu4Se9 fragments containing four slightly distorted, edge-sharing CuSe4 tetrahedra (cf. Figure 3a). In the ab plane the Cu4Se9 fragments are connected by selenium bridges formed by homonuclear Se−Se bonds, resulting in twodimensional layers (cf. Figure 3b). Two of these layers are connected along c via linear, hypervalent Se34− groups, resulting in “sandwichlike”, layered building blocks stacked along c. In 9211

DOI: 10.1021/acs.inorgchem.7b01224 Inorg. Chem. 2017, 56, 9209−9218

Article

Inorganic Chemistry

Se3 (cf. Figure 3c). In Figure 3d it is illustrated how four electrons are delocalized over three selenium atoms, which is typical for hypervalent 3c-4e bonds, resulting in a formal charge of 4− on the Se3 units. Se34− is isoelectronic with the wellknown I3− anion and the Sb37− unit found in Yb14MnSb11.29 The interatomic Se−Se distances of 2.708(3) Å (cf. Table S2 in the Supporting Information) in the linear, hypervalent Se34− units are comparable to those found in Ba2Ag4‑xCuxSe5 (2.7706(9)−2.783(1) Å)4 and Rb12Nb6Se35 (2.68 Å).9 The thermal ellipsoid of Se1 in Ba4Cu8Se13 is slightly elongated along the bond axis (cf. Figure 3c and Table S1 in the Supporting Information), which suggests that Se1 is slightly disordered along this direction, and the disorder can be rationalized using the resonant bonding picture illustrated in Figure 3d. The interatomic distances in the Se22− units (2.401(4) and 2.418(4) Å) are slightly longer but comparable to those found in elemental selenium (2.34 Å in α-Se and 2.37 Å in trigonal Se)30,31 and other chalcogenides such as USe3 (2.36 Å).32 The Cu−Cu distances range from 2.593(5) to 2.632(5) Å, and they are comparable to those found in elemental copper (2.5527(35) Å).33 Comparable values are also reported for other copper selenides such as Rb3Cu8Se6 (2.49 Å), Cs3Cu8Se6 (2.46 Å), and K3Cu8Se6 (2.50 Å).34,35 These short Cu−Cu distances in Ba4Cu8Se13 raise the question of whether Cu(I)−Cu(I) bonding interactions have to be considered. In this case Ba4Cu8Se13 is better described as a cluster compound containing Cu4Se9 clusters with direct Cu− Cu bonds (cf. the Supporting Information). Mehrotra et al. studied such potential Cu(I)−Cu(I) bonding interactions in organocopper(I) compounds, including clusters with Cu44+ squares and Cu−Cu distances of 2.42 Å.36 The authors computed binding energies and overlap populations in these

Figure 2. Crystal structure of Ba4Cu8Se13 obtained from single-crystal X-ray diffraction and viewed along [110]. The basic Cu4Se9 building blocks consisting of four edge-sharing CuSe4 tetrahedra (gray) are connected in two dimensions via Se atoms and Se−Se bonds (black), resulting in two-dimensional Cu−Se layers. Two of these layers are connected by hypervalent Se34− units, resulting in covalently bondend, “sandwichlike” building blocks, which are stacked along c, and the Ba atoms are intercalated between each Cu−Se layer. Color scheme: Ba, red; Se, green; Cu, blue.

terms of charge balance Ba4Cu8Se13 can be rationalized using the Zintl formalism as (Ba2+)4(Cu+)8(Se2−)2(Se22−)4(Se34−), assuming a complete charge transfer from the barium atoms and a transfer of one electron from the Cu atoms to the anionic framework. This notation takes into account the Se22− groups formed between Se5 and Se7 as well as Se2 and Se6 and the linear, hypervalent Se34− groups, which are formed by Se1 and

Figure 3. (a) One Cu4Se9 unit in polyhedral representation. (b) Thermal ellipsoids of the linear, hypervalent Se34− unit, which is slightly elongated along the bond axis. (c) Schematic representation of the hypervalent bonding in Se34− which explains the slight elongation of the thermal ellipsoids along the bond axis. (d) Two-dimensional connection of the Cu4Se9 building blocks via Se3 and the Se2−Se6 and Se5−Se7 bonds. (e) SEM image of a Ba4Cu8Se13 single crystal and the corresponding compositional maps. 9212

DOI: 10.1021/acs.inorgchem.7b01224 Inorg. Chem. 2017, 56, 9209−9218

Article

Inorganic Chemistry

Figure 4. (a) HAADF image taken of a polycrystalline sample in the zone axis [110]. (b) Corresponding Fourier transform of area 1 showing no real structure effects. (c) Corresponding Fourier transform of area 2 showing real structure effects along the c* axis. (d) (0kl) diffraction pattern obtained from single-crystal X-ray diffraction. Real structure effects can be observed in both single-crystal and polycrystalline Ba4Cu8Se13.

Cu44+ units as a function of the Cu−Cu distance, including considerations of steric effects of the ligands. For a Cu−Cu distance of 2.57 Å in Cu44+ squares with an ideal D4h symmetry, which comes closest to the situation in Ba4Cu8Se13, a binding energy of −0.984 eV and an overlap population of 0.080 were found. The authors concluded that both direct Cu(I)−Cu(I) bonding interactions and ligand stereochemistry are responsible for the short Cu−Cu distances when considering a mixing of the 3d, 4s, and 4p orbitals of Cu. From these findings, a description of Ba4Cu8Se13 as a cluster compound which contains Cu4Se9 clusters with Cu(I)−Cu(I) bonding interactions also seems reasonable. The dihedral angle between the Cu2−Cu1−Cu3 and Cu4−Cu3−Cu1 planes is 179.75°: i.e., the copper atoms in the Cu4Se9 clusters can be considered as being in a plane resulting in an almost ideal D4d symmetry, which is however broken by the slightly unequal Cu−Cu distances. It is known that isolated, square-planar transition-metal clusters are difficult to stabilize if only d orbitals are considered, since orbital symmetry restrictions allow no more than two in-plane metal−metal bonds.37 This has been demonstrated for the Os4(CO)16 cluster, which shows a strong distortion away from an ideal D4h geometry. It is intriguing that the Cu4Se9 “clusters” in Ba4Cu8Se13 come very close to an ideal D4d symmetry. Possible explanations are s and p orbital contributions allowing four in-plane Cu−Cu bonds and steric effects coming from the ligands and the neighboring Cu4Se9 building blocks. In Figure 3d an SEM image of a Ba4Cu8Se13 single crystal and the corresponding compositional mapping are depicted, and the latter shows a homogeneous distribution of all three elements within the crystal. Nanotwinning, Structural Modulations, and Real Structure Effects along c*. The single-crystal structure solution was complicated by the presence of overlapping Bragg reflections, diffusion, and the absence of a clear periodicity visible in the single-crystal X-ray diffraction patterns along the

c* axis (cf. Figure 1a). These real structure effects are discussed in detail in Single-Crystal Structure Determination. The relatively high reliability factor and residual electron density peaks mainly observed 0.78 Å away from Ba2 at the final stage of the refinement can be considered as directly related to these features. The refined structural model represents an average model describing the main characteristics of the sample, including twinning. In order to obtain additional information concerning the real structure and local order in Ba4Cu8Se13, a detailed transmission electron microscopy analysis was performed on polycrystalline Ba4Cu8Se13. The HAADF image shown in Figure 4a corresponds to a part of a crystallite of polycrystalline Ba4Cu8Se13 synthesized under the same conditions as for the single crystals. Our analysis revealed areas without (cf. Figure 4b) and with real structure effects (cf. Figure 4c) analogous to those observed by singlecrystal X-ray diffraction (cf. Figure 4d). Analyzing area 2 (cf. Figure 4a) of the HAADF image reveals the presence of nanotwins, confirming our findings from singlecrystal X-ray diffraction. Some of the building blocks are rotated by 180° around c. Figure 5 represents an enlargement of parts of the HAADF image shown in Figure 4a, including area 2. Four parts of the image are superimposed by the average structural model obtained by single-crystal X-ray diffraction, and these parts are labeled 1−4. A simple translation of parts of the structural model from layer 2 to layer 3 leads to a perfect overlap of the average structural model with the HAADF image. The same translation with respect to layer 1 and layer 4, on the other hand, results in a misfit of some of the atoms. This misfit is illustrated in Figure 6a, which corresponds to an enlargement of part 1 with parts of the average structural model translated from part 2 and superimposed on part 1. If the whole building block is rotated by 180° around c, the atoms overlap perfectly with the HAADF image again (cf. Figure 6b). Hence, area 2 of 9213

DOI: 10.1021/acs.inorgchem.7b01224 Inorg. Chem. 2017, 56, 9209−9218

Article

Inorganic Chemistry

However, they cannot explain the entire reflection sequences observed along this direction. This is why the previously discussed 0.31c* periodicity was included (cf. Single-Crystal Structure Determination) in our structural model, since these additional reflections contain information about structural modulations: i.e., deviations of the atomic positions from their equilibrium positions. To account for these additional reflections, a superspace formalism developed for the description of aperiodic crystal structures was used38 and a modulation wave vector of q = 0.31c* was thus introduced. A schematic representation of the (h0l)* plane, including twinning and satellite reflections, is provided in Figure 1d, which is meant to be compared to Figure 1a. The data, collected at room temperature, were integrated using the unit cell a = 9.171(8) Å, b = 9.146(8) Å, c = 27.35(3) Å, and β = 93.21(3)° and the q vector. Starting from the average structural model and the superspace group C2/c (σ10σ3)0s, harmonic waves were developed up to the first order and introduced for all atoms in order to describe possible atomic displacements. The final model led to a reasonable reliability factor for the satellite reflections (ca. 19%) considering the partial overlap of the reflections (cf. Figure 1a,d). We provide three movies in the Supporting Information, which we created using the software MoleCoolQt39 in order to show the effect of the modulation on the average crystal structure. The modulation can be understood as a cooperative atomic displacement observed within a single layer. As a consequence, each motif keeps its general characteristics, described in Description of the Average Crystal Structure. Thus, the Cu4Se9 units inside the layers are globally identical in each of the “sandwichlike”, layered building blocks. However, considering the stacking direction, each of the different motifs within an individual building block can exhibit small shifts along a and b. These shifts are one possible explanation for the diffuse intensities along c*. The Ba1 atoms within the covalently bonded, “sandwichlike” building blocks exhibit the largest atomic displacements mainly along c. Such a Ba modulation seems to be essential to avoid a mismatch between adjacent Cu−Se layers. The more pronounced modulation of the Ba1 atoms in comparison to the Ba2 atoms might have its origin in the different Cu−Se environments shown in the Supporting Information. The Ba1 atoms are located between Cu−Se layers, which are connected through Se34− groups; these are missing in the environment of the Ba2 atoms. The Se34− groups seem to restrict the movement of the Ba1 atoms in the ab plane, which might result in a large displacement along c. Our study of the real structure effects in Ba4Cu8Se13 provides a basic understanding of the origin of the diffuse intensities and absence of a clear periodicity along c*, and it explains why a relatively large reliability factor and residual electron density were obtained after solving and refining the average crystal structure of Ba4Cu8Se13. Confirmation of the Average Crystal Structure by Transmission Electron Microscopy. The average structural model obtained by single-crystal X-ray diffraction can be confirmed by our TEM analyses of polycrystalline Ba4Cu8Se13 synthesized under the same conditions as the single crystals. This is illustrated in Figure 8, which shows that the structural model ([110] projection) obtained by single-crystal X-ray diffraction perfectly overlaps with the HAADF image of polycrystalline Ba4Cu8Se13 taken in the [110] zone axis and its corresponding simulation (cf. Figure 8b). The HAADF

Figure 5. Enlargement of a part of the HAADF image shown in Figure 3, which contains real structure effects. Four “sandwichlike” building blocks of the crystal structure and four layers (labeled 1−4) of the HAADF image are superimposed by simple translation of the building block, resulting in a misfit of the atoms in layers 1 and 4. The building blocks with the misfit are rotated 180° around c (right part of the image). The rotation leads to a perfect overlap of the atoms. This type of nanotwinning found in layer 1 and layer 4 can explain some of the additional reflections along c*. Color scheme: Ba, red; Cu, blue; Se, green.

Figure 6. Enlargement of layer 1 shown in Figure 4: (a) the misfit of the atoms after translating the building block from layer 2 to layer 1 without a rotation of the building block; (b) perfect overlap of the atoms after rotating the building block 180° around c. Color scheme: Ba, red; Cu, blue; Se, green. The nanotwins are especially visible by looking at Ba2 and Se4.

the HAADF image clearly contains nanotwins corresponding to a 180° rotation of some of the covalently bonded, “sandwichlike” building blocks around the c axis, which is consistent with the twin domains introduced in our single-crystal structural model describing the average crystal structure of Ba4Cu8Se13. These nanotwins are especially visible by looking at the Ba2 and Se4 atoms, as is illustrated in Figure 7. The image shows that a 180° rotation of the covalently bonded, “sandwichlike” building blocks around c is possible without breaking any covalent bonds, since these building blocks are only separated by Ba atoms. These nanotwins lead to a slight shift of the atoms in the ab plane, which alters the periodicity along c, and they can explain some of the additional reflections along c*. 9214

DOI: 10.1021/acs.inorgchem.7b01224 Inorg. Chem. 2017, 56, 9209−9218

Article

Inorganic Chemistry

Figure 7. Illustration of the nanotwins with special focus on the interface of two “sandwichlike” building blocks and on the Ba2 and Se4 atoms: (a) [110] projection of the structural model; b) 180° rotation of the [110] projection around the c axis. This illustration shows that a rotation of entire building blocks is possible without breaking any covalent bonds within the two covalently bonded, “sandwichlike” building blocks.

9). Detailed information on the refinement is summarized in Table 2.

Figure 9. Le Bail refinement of an experimental PXRD pattern recorded after spark plasma sintering: (red) experimental intensities (Iobs); (black) calculated intensities (Icalc); (blue) Iobs − Icalc; (green) reflection positions of Ba4Cu8Se13.

Figure 8. (a) Projection of the Ba4Cu8Se13 structure in the [110] zone axis without coordination polyhedra. (b) HAADF image (left) taken from a polycrystalline sample in the [110] zone axis and the corresponding simulation (right), both superimposed by the structural model shown in (a); the Cu4Se9 building blocks (c) and Se34− units (d) are clearly visible. Color scheme: Ba, red; Cu, blue; Se, green.

Ba4Cu8Se13 has a melting point of 796 K (cf. Figure 10a) and is an intrinsic p-type semiconductor (cf. Figure 10b), which is indicated by the positive Seebeck coefficient. The semiTable 2. Le Bail Refinement Results

image also clearly shows the presence of the Cu4Se9 and Se34− units. Transport Properties. The crystal structure of Ba4Cu8Se13 is highly complex; the Ba atoms in Ba4Cu8Se13 can act as phonon scattering centers, and the more rigid Cu−Se network facilitates electron conduction. This is ideal for thermoelectric applications according to the phonon-glass electron-crystal (PGEC) concept introduced by Slack as being a key concept toward materials with inherently low thermal conductivities and high thermoelectric performance.40 Since Ba4Cu8Se13 appears to fulfill this concept, the low-temperature transport properties were measured on a densified bulk sample synthesized by ball milling and subsequent spark plasma sintering. In order to confirm the crystal structure and to ensure phase purity of the sample, a Le Bail refinement was performed on an experimental PXRD pattern recorded after spark plasma sintering (cf. Figure

nominal composition space group Z a (Å) b (Å) c (Å) β (deg) F(000) 2θ limits (deg) radiation constraints no. of refined params Rf/Rwp GOF 9215

Ba4Cu8Se13 C2/c (No. 15) 16 9.14556(15) 9.14137(15) 27.356(7) 93.14 3592 5.01−119.99 Cu Kα 0 27 0.0496/0.0735 1.89 DOI: 10.1021/acs.inorgchem.7b01224 Inorg. Chem. 2017, 56, 9209−9218

Article

Inorganic Chemistry

Figure 10. (a) DSC scan showing the melting point of Ba4Cu8Se13 and (b) electrical resistivity, (c) thermal conductivity, and (d) Seebeck coefficient of polycrystalline Ba4Cu8Se13. The abbreviations TTO-1 and TTO-2 refer to data obtained from two subsequent measurements on the same polycrystalline sample.

BaCu2Se2,23 which suggests a similar origin of the previously discussed anomalies in Seebeck coefficient and resistivity in these two compounds.

conducting behavior of the electrical resistivity is consistent with the bonding analysis, which results in the charge-balanced description of the title compound as (Ba2+)4(Cu+)8(Se2−)2(Se22−)4(Se34−). This is in contrast to mixed-valence copper chalcogenides such as NaCu4Se3, which are metallic.19 Ba4Cu8Se13 possesses a very low thermal conductivity κ (cf. Figure 10c) even below room temperature (e.g., 0.77 W m−1 K−1 at 200 K). Comparable values were measured for BaCu2Se2 (1.523/0.6824 W m−1 K−1 at room temperature) and BaCu2Te2 (2.0 W m−1 K−1 at room temperature).23 Ba4Cu8Se13 also has a large Seebeck coefficient (cf. Figure 10d) over a broad temperature range between 100 and 200 K, reaching a value of 380 μV K−1 at 200 K. The electronic contribution to the thermal conductivity is negligible, and the temperature dependence of κ is therefore governed by the phononic thermal conductivity (κph). The shape of the thermal conductivity is typical for crystalline materials, where grain boundary and point defect scattering are dominant at low temperatures, while Umklapp scattering processes dominate at intermediate and high temperatures.41 The Seebeck coefficient shows two anomalies at ∼50 and ∼125 K. Below ∼125 K charges become increasingly localized, which is reflected in the increase in resistivity. The Seebeck coefficient, on the other hand, decreases sharply below ∼125 K and shows a change in slope at ∼50 K, as does the resistivity. The drop in S is rather typical for a transition from a semiconductor to a metal, which stands in contradiction to the increase in resistivity. One possible explanation for these anomalies is a change in the electronic structure associated with order−disorder transitions, as proposed for KCu7‑xS4.42 Providing a detailed explanation of the origin of these transitions is beyond the scope of this publication and will be the subject of further studies. However, the transport properties of Ba4Cu8Se13 are similar to those of

■

CONCLUSION The new copper(I) selenide Ba4Cu8Se13 was discovered, and its average crystal structure was determined by single-crystal X-ray diffraction and transmission electron microscopy. The compound crystallizes in a new structure type and is the first example which simultaneously contains Se22− and linear, hypervalent Se34− units as well as unprecedented Cu4Se9 building blocks with hitherto unknown planar Cu rectangles. When homonuclear bonds between selenium atoms are taken into account, the compound can be rationalized as (Ba2+)4(Cu+)8(Se2−)2(Se22−)4(Se34−): i.e., it is charge balanced, which is consistent with the semiconducting behavior of the electrical resistivity. We were able to visualize the crystal structure by taking HAADF images, which clearly reveal the presence of the Se34− and Cu4Se9 building blocks as well as nanotwins, introduced by a 180° rotation of some of the covalently bonded, “sandwichlike” building blocks that are built up by Se22−, Se34−, and Cu4Se9 fragments. These nanotwins and a structural modulation with a modulation wave vector of q = 0.31c* provide a basic understanding of the real structure effects observed along c*, which made the structure solution unusually complex. The title compound possesses a very low thermal conductivity (e.g., 0.77 W m−1 K−1 at 200 K) and a large Seebeck coefficient (380 μV K−1 at 200 K) over a wide temperature range. Hence, Ba4Cu8Se13 fulfills two requirements for efficient thermoelectric materials. The Seebeck coefficient shows anomalies at ∼50 and ∼125 K, where also the resistivity 9216

DOI: 10.1021/acs.inorgchem.7b01224 Inorg. Chem. 2017, 56, 9209−9218

Article

Inorganic Chemistry

(8) Sturza, M.; Han, F.; Shoemaker, D. P.; Malliakas, C. D.; Chung, D. Y.; Jin, H.; Freeman, A. J.; Kanatzidis, M. G. NaBa2Cu3S5 A Doped p-Type Degenerate Semiconductor. Inorg. Chem. 2013, 52, 7210− 7217. (9) Dürichen, P.; Bolte, M.; Bensch, W. Synthesis, Crystal Structure, and Properties of Polymeric Rb12Nb6Se35, a Novel Ternary Niobium Selenide Consisting of Infinite Anionic Chains, Built Up by Nb2Se11 Units Containing an Uncommon Se34‑-Fragment. J. Solid State Chem. 1998, 140, 97−102. (10) Brown, D. B.; Zubieta, J. A.; Vella, P. A.; Wrobleski, J. T.; Watt, T.; Hatfield, W. E.; Day, P. Solid-state structure and electronic properties of a mixed-valence two-dimensional metal, potassium copper sulfide (KCu4S3). Inorg. Chem. 1980, 19, 1945−1950. (11) Berger, R.; Tergenius, L. E.; Noren, L.; Erisson, L. The crystal structure of room-temperature synthesized orthorhombic TlCu4Se3 from direct methods on X-ray powder data. J. Alloys Compd. 1995, 224, 171−176. (12) Stoll, P.; Näther, C.; Jess, I.; Bensch, W. KCu4Se3. Acta Crystallogr., Sect. C: Cryst. Struct. Commun. 1999, 55, 286−288. (13) Burschka, C. CsCu4S3 und CsCu3S2: Sulfide mit tetraedrisch und linear koordiniertem Kupfer. Z. Anorg. Allg. Chem. 1980, 463, 65− 71. (14) Vajenine, G. V.; Hoffmann, R. Compounds Containing CopperSulfur Layers: Electronic Structure, Conductivity, and Stability. Inorg. Chem. 1996, 35, 451−457. (15) Burschka, C. Na3Cu4S4 − Thiocuprate with isolated infinity [Cu4S4]-Chains. Z. Naturforsch. B: J. Chem. Sci. 1979, 34B, 396−397. (16) Peplinski, Z.; Brown, D. B.; Watt, T.; Hatfield, W. E.; Day, P. Electrical properties of sodium copper sulfide (Na3Cu4S4), a mixedvalence one-dimensional metal. Inorg. Chem. 1982, 21, 1752−1755. (17) Sturza, M.; Malliakas, C. D.; Bugaris, D. E.; Han, F.; Chung, D. Y.; Kanatzidis, M. G. NaCu6Se4: A Layered Compound with Mixed Valency and Metallic Properties. Inorg. Chem. 2014, 53, 12191−12198. (18) Berger, R.; Van Bruggen, C. F. TlCu2Se2: A p-type metal with a layer structure. J. Less-Common Met. 1984, 99, 113−123. (19) Sturza, M.; Bugaris, D. E.; Malliakas, C. D.; Han, F.; Chung, D. Y.; Kanatzidis, M. G. Mixed-Valent NaCu4Se3: A Two-Dimensional Metal. Inorg. Chem. 2016, 55, 4884−4890. (20) Iglesias, J. E.; Pachali, K. E.; Steinfink, H. Structural Chemistry of Ba2CdS3, Ba2CdSe3, BaCdS2, BaCu2S2 and BaCu2Se2. J. Solid State Chem. 1974, 9, 6−14. (21) Maier, S.; Prakash, J.; Berthebaud, D.; Perez, O.; Bobev, S.; Gascoin, F. Crystal Structure of the four new quaternary copper(I)selenides A0.5CuZrSe3 and ACuYSe3 (A = Sr, Ba). J. Solid State Chem. 2016, 242, 14−20. (22) Mansuetto, M. F.; Keane, P. M.; Ibers, J. A. Synthesis, structure, and conductivity of the new group IV chalcogenides, KCuZrQ3 (Q = S, Se, Te). J. Solid State Chem. 1992, 101, 257−264. (23) McGuire, M. A.; May, A. F.; Singh, D. J.; Du, M.-H.; Jellison, G. E. Transport and optical properties of heavily hole-doped semiconductors BaCu2Se2 and BaCu2Te2. J. Solid State Chem. 2011, 184, 2744−2750. (24) Li, J.; Zhao, L.-D.; Sui, J.; Berardan, D.; Cai, W.; Dragoe, N. BaCu2Se2 based compounds as promising thermoelectric materials. Dalton Trans. 2015, 44, 2285−2293. (25) Petricek, V.; Dusek, M.; Palatinus, L. Crystallographic Computing System JANA2006: General features. Z. Kristallogr. Cryst. Mater. 2014, 229, 345−352. (26) Sheldrick, G. M. A short history of SHELX. Acta Crystallogr., Sect. A: Found. Crystallogr. 2008, 64, 112−122. (27) Sheldrick, G. M. TWINABS; Bruker AXS Inc, Madison, WI, USA, 2002. (28) Palatinus, L.; Chapuis, G. Superflip − a computer program for the solution of crystal structures by charge flipping in arbitrary dimensions. J. Appl. Crystallogr. 2007, 40, 786−790. (29) Chan, Y. C.; Olmstead, M. M.; Kauzlarich, S. M. Structure and ferromagnetism of the rare-earth zintl compounds: Yb14MnSb11 and Yb14MnBi11. Chem. Mater. 1998, 10, 3583−3588.

changes its slope. Further studies are required in order to reach an understanding of the origin of these anomalies.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.inorgchem.7b01224. Equivalent isotropic displacement and positional parameters, selected interatomic distances and angles, cluster description of the Cu4Se9 building blocks, and local environments of the Ba atoms (PDF) Effect of the modulation on the average crystal structure (MP4) Effect of the modulation on the average crystal structure (MP4) Effect of the modulation on the average crystal structure(MP4) Accession Codes

CCDC 1549764 contains the supplementary crystallographic data for this paper. These data can be obtained free of charge via www.ccdc.cam.ac.uk/data_request/cif, or by emailing data_ [email protected], or by contacting The Cambridge Crystallographic Data Centre, 12 Union Road, Cambridge CB2 1EZ, UK; fax: +44 1223 336033.

■

AUTHOR INFORMATION

Corresponding Author

*F.G.: tel, +33-2-31452605; fax, +33-2-31951600; e-mail, [email protected]. ORCID

Sylvie Hébert: 0000-0001-7412-2367 Franck Gascoin: 0000-0002-9791-1358 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors acknowledge financial support from the french “Agence Nationale de la Recherche” (ANR) through the program “Investissements d’Avenir” (ANR-10-LABX-09-01), LabEx EMC3.

■

REFERENCES

(1) Sheldrick, W. S. Polychalcogenide Anions: Structural Diversity and Ligand Versatility. Z. Anorg. Allg. Chem. 2012, 638, 2401−2423. (2) Mayasree, O.; Sankar, C. R.; Kleinke, K. M.; Kleinke, H. Cu clusters and chalcogen-chalcogen bonds in various copper polychalcogenides. Coord. Chem. Rev. 2012, 256, 1377−1383. (3) Böttcher, P. Tellurium-Rich Tellurides. Angew. Chem., Int. Ed. Engl. 1988, 27, 759−772. (4) Assoud, A.; Xu, J.; Kleinke, H. Structures and Physical Properties of New Semiconducting Polyselenides Ba2CuδAg4‑δSe5 with Unprecedented Linear Se34‑ Units. Inorg. Chem. 2007, 46, 9906−9911. (5) Schewe-Miller, I.; Böttcher, P. Ternäre Telluride mit W5Si3-TypStruktur: MxK4Te3 (M = Ca, Sr). J. Alloys Compd. 1992, 183, 98− 108. (6) Patschke, R.; Kanatzidis, M. G. Polytelluride compounds containing distorted nets of tellurium. Phys. Chem. Chem. Phys. 2002, 4, 3266−3281. (7) Malliakas, C. D.; Billinge, S. J. L.; Kim, H. J.; Kanatzidis, M. G. Square nets of tellurium: Rare-earth dependent variation in the chargedensity wave of RETe3 (RE = rare-earth element). J. Am. Chem. Soc. 2005, 127, 6510−6511. 9217

DOI: 10.1021/acs.inorgchem.7b01224 Inorg. Chem. 2017, 56, 9209−9218

Article

Inorganic Chemistry (30) Burbank, R. D. The crystal structure of α-monoclinic selenium. Acta Crystallogr. 1951, 4, 140−148. (31) Cherin, P.; Unger, P. The crystal structure of trigonal selenium. Inorg. Chem. 1967, 6, 1589−1591. (32) Salem, A. B.; Meerschaut, A.; Rouxel, J. Chem. Informationsdienst 1984, 299, 617−619. (33) Bragg, W. L. The crystalline structure of copper. Phylos. Mag 1914, 28, 355−360. (34) Schils, H.; Bronger, W. Ternäre Selenide des Kupfers. Z. Anorg. Allg. Chem. 1979, 456, 187−193. (35) Nather, C.; Rohnert, D.; Bensch, W. Synthesis, crystal structure and low-temperature X-ray investigations of K3Cu8Se6. Eur. J. Solid State Inorg. Chem. 1998, 35, 565−577. (36) Mehrotra, P. K.; Hoffmann, R. Cu(I)-Cu(I) Interactions. Bonding Relationships in d10-d10 Systems. Inorg. Chem. 1978, 17, 2187−2189. (37) Lauher, J. W. What is Wrong with a Square Metal Cluster? Int. J. Quantum Chem. 1988, 34, 309−317. (38) De Wolff, P. M.; Jansen, T.; Janner, A. The superspace groups for incommensurate crystal structures with a one-dimensional modulation. Acta Crystallogr., Sect. A: Cryst. Phys., Diffr., Theor. Gen. Crystallogr. 1981, 37, 625−636. (39) Hübschle, C. B.; Dittrich, B. MoleCoolQt − a molecule viewer for charge-density research. J. Appl. Crystallogr. 2011, 44, 238−240. (40) Slack, G. A. In New Materials and Performance Limits for Thermoelectric Cooling - CRC Handbook of Thermoelectrics; Rowe, D. M., Ed.; CRC Press: Boca Raton, FL, 1995; pp 407−439. (41) Tritt, T. M. Thermal Conductivity − Theory, Properties and Applications; Springer: New York, 2004. (42) Kuo, Y.-K.; Skove, M. J.; Verebelyi, D. T.; Li, H.; Mackay, R.; Hwu, S.-J.; Whangbo, M.-H.; Brill, J. W. Unusual physical properties of KCu7-xS4 at diffusive, one-dimensional ordering transitions. Phys. Rev. B: Condens. Matter Mater. Phys. 1998, 57, 3315−3325.

9218

DOI: 10.1021/acs.inorgchem.7b01224 Inorg. Chem. 2017, 56, 9209−9218