High-pressure synthesis, crystal structure, and magnetic properties of KSbO3-type 5d oxides K0.84OsO3 and Bi2.93Os3O11

5d Solid-state oxides K0.84OsO3 (Os5.16+; 5d 2.84) and Bi2.93Os3O11 (Os4.40+; 5d 3.60) were synthesized under high-pressure and high-temperature conditions (6 GPa and 1500–1700 °C). Their crystal structures were determined by synchrotron x-ray diffraction and their 5d electronic properties and tunnel-like structure motifs were investigated. A KSbO3-type structure with a space group of Im-3 and Pn-3 was determined for K0.84OsO3 and Bi2.93Os3O11, respectively. The magnetic and electronic transport properties of the polycrystalline compounds were compared with those obtained theoretically. It was revealed that the 5d tunnel-like structures are paramagnetic with metallic charge conduction at temperatures above 2 K. This was similar to what was observed for structurally relevant 5d oxides, including Bi3Re3O11 (Re4.33+; 5d 2.66) and Ba2Ir3O9 (Ir4.66+; 5d 4.33). The absence of long-range magnetic order seems to be common among 5d KSbO3-like oxides, regardless of the number of 5d electrons (between 2.6 and 4.3 per 5d atom).


Introduction
Crystalline KSbO 3 -type [1] or comparable-type solid-state oxides are attractive for their possible applications in ionically Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. conductive and electrocatalytic materials [2][3][4][5]. This prospect may be related to the presence of tunnel motifs in the crystal structure of these oxides [6]. Furthermore, structurally related La 4 Ru 6 O 19 has received considerable attention because of its role in novel electronic transport in non-Fermi-liquids [7,8]. The KSbO 3 -type family of solid-state oxides is currently an emergent subject in the field of inorganic chemistry. They can be used to develop advanced technologies for energy-related applications and to help understand correlated electron properties.
Our recent studies have focused on the synthesis of solidstate osmium oxides in order to develop 5d electronic properties and 5d materials for possible advancements in the field of spintronics and related scientific devices [28][29][30][31]. During our attempted syntheses of compositionally new osmium oxides under high-pressure and high-temperature conditions, an additional oxide K 0.84 OsO 3 was synthesized (at 6 GPa). The polycrystalline compounds K 0.84 OsO 3 was studied by using synchrotron x-ray diffraction and magnetic and charge transport measurements. The refined crystal structure indicated that the crystalline oxide has a KSbO 3 -type structure and shares a tunnel structural motif with a related Os oxide Bi 2.93 Os 3 O 11 [15,16]. Herein, we report the synthesis, crystal structure, and primary electrical and magnetic properties of the newly synthesized KSbO 3 -type oxide K 0.84 OsO 3 and compare those with the properties of the structurally comparable oxide Bi 2.93 Os 3 O 11 .

Materials and methods
Polycrystalline K 0.84 OsO 3 was synthesized by a solid-state reaction method in a belt-type high-pressure and high-temperature apparatus (Kobe Steel, Japan), in which a pyrophyllite cell was used to produce a quasi-hydrostatic environment at an elevated pressure [32]. The starting materials Os (99.95%, Heraeus Materials Technology) and KO 2 (O 2 -45.6%, yellow powder, Sigma-Aldrich) were mixed at a molar ratio of 1:2 in a glove box under argon. The mixture was sealed in a platinum capsule, followed by heating in a compressed pyrophyllite cell at 1500°C for 1 h. The capsule pressure was maintained at 6 GPa during the heating process. The capsule was then quenched to ambient temperature within a minute by cutting off the electric power supply before releasing the pressure. The final product was a dense pellet of part of it was ground in an agate mortar and pestle. The powder was then rinsed in an ultrasonic water bath multiple times to remove any residue. The high-pressure method is helpful in reducing the risk of human exposure to possible presence of toxic OsO 4 during the synthesis.
Polycrystalline Bi 2.93 Os 3 O 11 was similarly prepared using fine powders of Bi 2 O 3 (99.999%, Kojundo Chem. Lab, Japan) and OsO 2 (Os-83%, Alfa Aesar) in the high-pressure apparatus. A small amount of an oxygen source (KClO 4 , 99.5%, Kishida Chem) was added to a stoichiometric mixture of the starting materials. The elevated pressure was maintained at 6 GPa during the heating process at 1700°C for 1 h. Residues in the final product (including KCl) were removed in a water bath.
The final products were characterized by synchrotron xray diffraction (SXRD) using a large Debye-Scherrer camera at the beam line BL15XU in the SPring-8 synchrotron radiation facility, Japan [33]. The diffraction profiles were collected at room temperature between 2 θ of 3°and 81°at 0.003°intervals using a monochromatized beam (λ = 0.65 298 Å or 0.40 025 Å). The wavelength was confirmed by measurements of a standard material (CeO 2 ). Each powder was placed into a Lindenmann glass capillary (inner diameter: 0.1 mm) and rotated during the measurements. The SXRD profiles were analyzed by a Rietveld method using the program RIETAN-FP [34].
The dc magnetic susceptibility (χ) of the compound was measured in the Magnetic Property Measurement System (MPMS, Quantum Design) between 2 and 395 K in an applied magnetic field of 10 kOe. Each powdered compound was loosely gathered in a sample holder and cooled to the temperature limit. The magnetic field was then applied to the holder. The holder was gradually warmed to 395 K (zero-field cooling, ZFC), followed by cooling in the field (field cooling, FC). The isothermal magnetization of the compounds was also measured in the apparatus with a magnetic field range between -70 and 70 kOe at 5 K. The specific heat C p of a piece of the physically compressed bulk material was measured in the Physical Property Measurement System (PPMS, Quantum Design) between 2 and 300 K. In the apparatus, the electrical resistivity (ρ) of a pellet piece was measured by a 4terminal method using platinum wires and a silver paste.
The K content of polycrystalline K 0.84 OsO 3 was determined by inductively coupled plasma spectrometry. Waterrinsed fine powder was used in the analysis and the average K content was found to be 0.837 (7) in accordance with the formula unit.
First-principles calculations of the electronic state of the stoichiometric hosts KOsO 3 and Bi 3 Os 3 O 11 were performed by a generalized gradient approximation [35] of the density functional theory. The WIEN2k program [36] was used, which was based on the full-potential augmented plane-wave method. The muffin-tin radii were chosen to be 2.4 atomic unit (au) for K, 2.2 au for Bi, 1.9 au for Os, and 1.6 au for O atoms. The spin-orbit interaction was included as a perturbation to the scalar-relativistic equations. The cut-off wave vector K was fixed at RK = 8, where R is the smallest muffintin radius (i.e. 1.6 au). The Brillouin zone integration was approximated by the tetrahedron method with 294 k points in an irreducible zone for KOsO 3 and 76 k points for Bi 3 Os 3 O 11 . We assumed that K atoms occupy the K1 site for KOsO 3 and Bi atoms occupy Bi1 (8e) and Bi2 (4b) sites for Bi 3 Os 3 O 11 to avoid fractional occupation.

Results and discussion
The crystal structure of K 0.84 OsO 3 was characterized well by a cubic model with a space group of Im-3, similar to KBiO 3 and AgBiO 3 [4,11,12]. Figure 1 shows the SXRD pattern for K 0.84 OsO 3 at room temperature. Rietveld analysis conducted on the pattern with the cubic model resulted in a well-refined profile with R indices below 7%. The refined cubic lattice parameter was a = 9.47 164(1) Å, which is smaller than the corresponding cubic parameters of KBiO 3 (10.0194(6) Å) and AgBiO 3 (9.7852(2) Å). The smaller ionic radius (0.575 Å) of Os(V) in an octahedral environment than that of Bi(V) (0.76 Å) may account for this observation [37]. We concluded that a reasonable fit was established; the final structural parameters, including the refined atomic coordinates, are listed in table 1. Although the true chemical composition was slightly under-stoichiometric (K 0.84 OsO 3 ), we analyzed the pattern without considering the small amount of K deficiencies that we were unable to refine. However, the thermal parameters for all atoms remained within a reasonable level regardless of the K deficiencies. The small amount of deficiencies may have been distributed almost equally over the three crystallographic K sites, minimizing impact on the analysis.
The refined crystal structure is illustrated in the inset of figure 1. The structural view indicates that an Os atom occupies the center of the octahedron. The view clearly shows a characteristic tunnel motif as was observed for other KSbO 3 -type and related oxides. In the octahedra, each edge or corner is shared by neighboring octahedra, causing the shortest Os-Os distance to be 2.6610(4) Å. The distance is 8.7% longer than the bonded Ru-Ru distance of La 4 Ru 6 O 19 (2.448 Å) and 12.5% shorter than the non-bonded Ru-Ru distance of La 3 Ru 3 O 11 [7,8]. The distances may suggest that Os-Os bonding has formed. It can be noted that the ionic size of Os is slightly larger than that of Ru [37]; however, the ionic size difference is unlikely to affect the observation. Figure 2 shows the SXRD pattern for Bi 2.93 Os 3 O 11 measured at room temperature (as well as the refined pattern). In the refinement, Bi 2.93 Os 3 O 11 was assumed to be  [34]. The proposed crystal structure is presented in the inset, in which Os and O atoms are drawn as large red and small blue balls, respectively. K atoms are not shown for clarity.  (7) Note. The space group is Im-3 (no. 204), a = 9.47164(1) Å, Z = 12, V = 849.718(2) Å 3 , and d cal = 6.50 g cm −3 . R Indices were R wp = 3.55%, R p = 2.47%, R B = 6.58%, and R   [34]. The proposed crystal structure is presented in the inset, in which Os and O atoms are drawn as large red and small blue balls, respectively. K atoms are not shown for clarity. isostructural to Bi 3 Mn 3 O 11 [24]; fractional atomic coordinates for Bi 3 Mn 3 O 11 were tested in early refinements. The Bi1 atom was found to be disordered as in Bi 3 Mn 3 O 11 [24] and Bi 3 GaSb 2 O 11 [39]. Eventually, a refinement with an occupation factor (g) of 1/3 for Bi1 and 1 for Bi2 resulted in a negative thermal parameter (B) for O1; however, the B(O1) was positive when we allowed the refinement of g(Bi1) and g (Bi2). A small amount of vacancies was therefore suggested at these sites; the refined composition was Bi 2.93 Os 3 O 11 . The under-stoichiometric composition may be connected to the presence of a small amount of impurities in the compound. We also note that splitting of Bi2 from the ideal 4b site (0,0,0) to 8e site (x, x, x) resulted in slightly better R factors at x = 0.0029 (4). Note that the shortest Os-Os distance is 2.5653 (3) Å, comparable with the Os-Os distance in K 0.84 OsO 3 . Details of the refinement are summarized in table 2.
The OsO 6 framework of Bi 2.93 Os 3 O 11 is tunnel-like, similar to K 0.84 OsO 3 , as shown in the inset of figure 2. The Bi atoms are not shown in the structural views for clarity. When assuming an ionic picture of the compounds, the formal Os valence should be +5. 16  The temperature dependence of ρ for a piece of polycrystalline K 0.84 OsO 3 was measured (figure 3), and a metallic temperature dependence was found over the studied temperature range. Although ρ (∼0.025 ohm cm) at room temperature is approximately one order of magnitude higher than what is expected for a polycrystalline conducting oxide, the  (7) Note. The space group is Pn-3 (no. 201) at origin choice 2, Z = 4, a = 9.35993(2) Å, and V = 820.007(4) Å 3 . R Indices were R wp = 6.42%, R p = 4.03%, R B = 3.52%, and R F = 1.75%.     Figure 4 shows the temperature dependence of χ for the oxides, revealing weakly temperature-dependent paramagnetic features over the temperature range. Any possible magnetic transition was unlikely over the measurements. Although a broad and small bump is seen at approximately ∼50 K for K 0.84 OsO 3 , any corresponding anomalies were not obvious in the ρ and C p measurements (shown later). The inset shows the isothermal magnetizations of the oxides at low temperature (5 K) and only quasi-linear behaviors with trivial magnetizations were observed. Therefore, we tentatively assumed that the magnetic bump is likely impurity driven. The overall magnetic measurements suggested that the compounds are both paramagnetic at temperatures above 2 K. The temperature dependence of χ at the low temperature limit remains unconnected to any magnetic model and impurities.
Specific heat measurements for the oxides were conducted; their C p versus T curves are shown in figures 5(a) and (b). Over the temperature range, C p varies monotonically and any indicative anomaly for a transition is unobvious. The C p versus T curve was analyzed by a linear combination of the Debye and Einstein model, as was conducted for related materials [40]. The analytical formula was The Einstein term added to the Debye term increased the quality of fitting, suggesting a possibility that the phonon density of state (DOS) forms a rather complex structure over the whole temperature range. However, analysis on the C p /T 3 versus T plots (not shown) indicated that anomalous Einstein contributions, which are indicative of lattice rattling, were not obvious (unlike related oxides) [40]. Further analysis of the phonon modes is needed to clarify the phonon DOS structure of both the compounds. The low-temperature measurements of the C p /T versus T 2 plot for each compound were analyzed by an approximated Debye model (figures 5(c) and (d)), which was C p /T = βT 2 + γ, where β and γ are a constant and the Sommerfeld coefficient, respectively. The fit of the plots yielded β = 3.03(9) × 10 −4 J (one mole of osmium atoms (mol-Os)) −1 K −4 and γ = 16.8(2) mJ mol-Os −1 K −2 for K 0.84 OsO 3 ; and β = 7.00(9) × 10 −4 J mol-Os −1 K −4 and γ = 1.6(2) mJ mol-Os −1 K −2 for Bi 2.93 Os 3 O 11 . The T D for K 0.84 OsO 3 and Bi 2.93 Os 3 O 11 were calculated from β of 315(4) K and 250(1) K, respectively. It appeared that the γ for Bi 2.93 Os 3 O 11 is nearly one tenth of the γ for K 0.84 OsO 3 ; we therefore carefully investigated the electronic state of the compounds by a theoretical method. Figures 6(a) and (b) show the theoretically predicted electronic DOS structure for the stoichiometric hosts K 0.84 OsO 3 and Bi 2.93 Os 3 O 11 , respectively. The total DOS was found to consist of mainly Os and O contributions and little from K/Bi. Both compounds have a nontrivial electronic DOS at the Fermi level (E F ). Therefore, the hosts are expected to be metallic. The estimated γ for KOsO 3 from the DOS at E F is 15.6 mJ mol-Os −1 K −2 , which is nearly comparable to the observed γ for K 0.84 OsO 3 (16.8(2) mJ mol-Os −1 K −2 ). However, the estimated γ for Bi 2.93 Os 3 O 11 is 9.7 mJ mol-Os −1 K −2 , which is much larger than the observed γ for Bi 2.93 Os 3 O 11 (1.6(2) mJ mol-Os −1 K −2 ). This disagreement between the expected and observed γ for the compounds is possibly owing to a steep change of DOS in the vicinity of E F with lowering the Bi stoichiometry. This phenomenon has been previously discussed for related Os oxides [41]. Although we need to carefully investigate possible contributions from spin-polarization and spin-orbit interactions over the DOS structure, opening a full gap at E F for Bi 2.93 Os 3 O 11 seems unlikely.

Conclusions
Materials with 5d electrons may show characteristic features owing to radially extended valence orbitals and large spinorbit couplings of 5d electrons over 3d electrons. For example, a perovskite-type oxide, NaOsO 3 , shows a Slater-like transition [30,31] and a LiNbO 3 -type oxide, LiOsO 3 , shows a ferroelectric-like transition in the metallic state [29]. The results lead to the reasonable expectation that KOsO 3 , if synthesized, also shows characteristic features of 5d electrons. Nevertheless, KSbO 3 -type crystalline K 0.84 OsO 3 shows only a weak temperature-dependent paramagnetic feature. In addition, KSbO 3 -type Bi 2.93 Os 3 O 11 was also synthesized under high-pressure and high-temperature conditions. The Os-O network in this system was found to form a similar tunnel motif to that of K 0.84 OsO 3 . Although the formal valence of Os decreased from +5.16 (K) to +4.40 (Bi), the observed magnetic and electronic properties did not change significantly.
In contrast to the remarkable 5d properties of related compounds NaOsO 3 [30,31] and LiOsO 3 [29], K 0.84 OsO 3 and Bi 2.93 Os 3 O 11 seem to have less characteristic 5d properties above 2 K. Both compounds showed rather weak temperature-dependent paramagnetism and metallic transports. Disorders such as K/Bi vacancies and the polycrystalline nature of the compounds (including grain boundaries and impurities) could possibly complicate the observed 5d properties. Nevertheless, the magnetic and electronic properties are quite similar to what was observed for structurally relevant 5d oxides, including Bi 3 Re 3 O 11 (Re 4.33+ ; 5d 2.66 ) [18] and Ba 2 Ir 3 O 9 (Ir 4.66+ ; 5d 4.33 ) [26], regardless of the number of 5d electrons. The absence of a long-range magnetic order seems to be common among the tunnel-like structures of 5d oxides. Further studies on high-quality single crystals of newly synthesized KSbO 3 -type material K 0.84 OsO 3 may reveal the 5d characteristic features and help establish a comprehensive picture of the 5d electronic system.