New Insulating Antiferromagnetic Quaternary Iridates MLa₁₀Ir₄O₂₄ (M = Sr, Ba)

Abstract

Recently, oxides of Ir⁴⁺ have received renewed attention in the condensed matter physics community, as it has been reported that certain iridates have a strongly spin-orbital coupled (SOC) electronic state, J_eff = ½, that defines the electronic and magnetic properties. The canonical example is the Ruddlesden-Popper compound Sr₂IrO₄, which has been suggested as a potential route to a new class of high temperature superconductor due to the formal analogy between J_eff = ½ and the S = ½ state of the cuprate superconductors. The quest for other iridium oxides that present tests of the underlying SOC physics is underway. In this spirit, here we report the synthesis and physical properties of two new quaternary tetravalent iridates, MLa₁₀Ir₄O₂₄ (M = Sr, Ba). The crystal structure of both compounds features isolated IrO₆ octahedra in which the electronic configuration of Ir is d⁵. Both compounds order antiferromagnetically despite the lack of obvious superexchange pathways and resistivity measurement shows that SrLa₁₀Ir₄O₂₄ is an insulator.

Introduction

Electrons in 3d transition metal oxides exhibit correlated behavior because the bandwidth, W, is relatively narrow, while the electron-electron repulsion, parametrized by a Hubbard U, is significant. The result is a set of collective phenomena including high temperature superconductivity¹ and colossal magnetoresistance^2,3. On the other hand, 5d transition metals are characterized by more extended orbitals and in general, are expected to be uncorrelated metals, as for example IrO₂⁴ and Bi₂Ir₂O₇⁵. Recently, however, Kim et al. found that strong spin-orbital coupling in the layered compound Sr₂IrO₄ leads to a relatively narrow J_eff = ½ band whose width is of the same scale as electron correlation, yielding a Mott insulating state^6,7. With a square IrO₂ network, a U/W ~ 1 and a magnon dispersion qualitatively the same as cuprates⁸, Sr₂IrO₄ has been studied intensely as a potential route to a new class of high temperature superconductor^8,9,10. Some known iridates, including Ba₂IrO₄¹¹ and Ca₄IrO₆¹², have already shown behavior consistent with a J_eff = ½ description by resonant inelastic X-ray scattering, indicating that the phase space of J_eff = ½ materials extends beyond Sr₂IrO₄. These discoveries underscore the importance of identifying new iridates with J_eff = ½ states to better understand the phenomenology of these unusual correlated oxides.

The scope of the present work lies firmly in the regime of discovery synthesis of new compounds in a relatively unexplored regime of crystal chemistry as a first step on the way to classifying and understanding the breadth of spin-orbit driven physics in iridates. Toward this end, we have synthesized two new isostructural tetravalent iridates, MLa₁₀Ir₄O₂₄ (M = Sr, Ba) and characterized their crystal structures and magnetic, transport and thermodynamic signatures. Each is comprised of isolated IrO₆ octahedra in which the nominal electronic configuration of Ir is d⁵. They both order antiferromagnetically and resistivity measurement shows that SrLa₁₀Ir₄O₂₄ exhibits insulating behavior.

The extensively studied iridates are mostly ternary oxides, which is likely due to the lack of approaches for crystallizing iridates that are compositionally diverse. To expand the horizon in studying iridates as J_eff = 1/2 candidates, quaternary and even higher order iridates are desired. Here it is demonstrated that high order iridates can be made with a facile flux crystal growth approach.

Results

Lattice constants and space group (I4₁/a) extracted from single-crystal diffraction measurements were similar to that of the known compound Sr₉La₂Mo₄O₂₄¹³ and indeed the crystal structures (see Fig. 1) of the iridates could be solved from this starting model. For SrLa₁₀Ir₄O₂₄, no discernible site mixing between Sr and La was found in the refinement. For BaLa₁₀Ir₄O₂₄, the Ba and La sites were assigned based on the apparent differences in bond lengths of Ba-O and La-O. The structure features isolated IrO₆ octahedra with M (M = Sr, Ba) and La atoms located interstitially. There are two independent Ir sites, both occupying special positions, with the Ir atoms in a distorted octahedral coordination environment. The lattice parameters of SrLa₁₀Ir₄O₂₄ are a = 11.58 Å, c = 16.24 Å, while those of BaLa₁₀Ir₄O₂₄ are a = 11.66 Å, c = 16.17 Å. Compared to SrLa₁₀Ir₄O₂₄, the Ba analogue has a larger a lattice parameter and a smaller c lattice parameter. In other words, the lattice is “squashed” rather than simply enlarged. The cell volume of BaLa₁₀Ir₄O₂₄ (2196.9 ų) is only slightly larger than that of SrLa₁₀Ir₄O₂₄ (2183.4 ų), a difference of ~0.05%. This weak volume perturbation is not surprising, because although the ionic radius of Ba²⁺ (1.42 Å) is significantly larger than that of Sr²⁺ (1.26 Å), the alkaline earth metal is only a small fraction of the unit cell contents. For SrLa₁₀Ir₄O₂₄, the bond length of Ir1-O ranges from 1.967(15) Å to 2.030(16) Å, while the bond length of Ir2-O ranges from 2.001(16) Å to 2.073(16) Å. For BaLa₁₀Ir₄O₂₄, the bond length of Ir1-O ranges from 1.983(12) Å to 2.046(12) Å, while the bond length of Ir2-O ranges from 2.006(12) Å to 2.086(13) Å. These bond lengths are typical of Ir(IV) in oxides^14,15.

Figure 1

The crystal structure of MLa₁₀Ir₄O₂₄ (M = Sr, Ba).

The purple color denotes IrO₆ octahedra and the green and blue spheres denote La and M atoms (M = Sr, Ba), respectively (a) viewed from approximately (001) direction (b) viewed approximately from (010) direction.

The temperature-dependent magnetic susceptibility of SrLa₁₀Ir₄O₂₄ shows a cusp characteristic of antiferromagnetic order at T_N = 12 K (Fig. 2, top), indicating a non-negligible magnetic exchange among the isolated octahedra. Above this cusp, the magnetic behavior is well approximated as that of a Curie-Weiss paramagnet. A fit of the data to χ = C/(T – θ) + χ₀ (where χ₀ phenomenologically accounts for all diamagnetic contributions) from 20 K to 300 K (the results are largely insensitive to the choice of fitting range) gives a Weiss constant of −8.5 K, in good agreement with T_N. The effective moment (μ_eff) of 1.11 μ_B/Ir is considerably reduced from the 1.73 μ_B/Ir expected for low-spin d⁵ Ir(IV) in a rigorously J_eff = ½ configuration (cubic crystal field, k_BT/λ → 0)¹⁶. We note that such reduced effective moments are not uncommonly reported among iridates, including Sr₂IrO₄ (0.50 μ_B /Ir)¹⁷, Sr₃Ir₂O₇ (0.69 μ_B /Ir)¹⁸, 9 M BaIrO₃ (0.13 μ_B /Ir)¹⁷ and 6 M BaIrO₃ (0.276 μ_B /Ir)¹⁹. However, other iridates, notably Na₂IrO₃²⁰, with a reported μ_eff = 1.79 μ_B /Ir, follow more closely the expected behavior. Discrepancies such as these suggest an ‘effective g-factor’ significantly reduced from the free electron value, deriving potentially from non-cubic symmetry, an admixture of configurations other than t_2g⁵, or the effect of hybridization with the O sublattice network.

Figure 2

Temperature dependent DC magnetic susceptibility of SrLa₁₀Ir₄O₂₄ (top) and Ba La₁₀Ir₄O₂₄ (bottom).

Open red circles are the data and the solid black line is a Curie-Weiss fit including a temperature independent parameter, χ₀, which takes the values 2.2 × 10⁻⁴ emu/f.u.•Oe and 3.7 × 10⁻⁴ emu/f.u•Oe, respectively. The blue line is the inverse T-dependent component of the susceptibility. Measuring field is 5000 Oe. The insets show the low temperature ranges (a) SrLa₁₀Ir₄O₂₄ (b) BaLa₁₀Ir₄O₂₄.

Similarly, the temperature dependent magnetic measurement of BaLa₁₀Ir₄O₂₄ shows antiferromagnetic ordering at a somewhat lower temperature, T_N = 6 K (Fig. 2, bottom) and also follows Curie-Weiss behavior above this temperature, with μ_eff = 1.35 μ_B, slightly larger than that of SrLa₁₀Ir₄O₂₄. The Weiss constant obtained is −5.1 K, close to the measured T_N. For both SrLa₁₀Ir₄O₂₄ and BaLa₁₀Ir₄O₂₄, at temperatures below the T_N, the susceptibility has a small upturn, which may arise from a small amount of paramagnetic impurity spins.

Heat capacity measurements were carried out on both compounds. Above the magnetic transition, the data can be described by the expression C = C_electron + C_phonon = γT+ βT³ (Fig. 3a). The values of γ and β extracted from the fits for SrLa₁₀Ir₄O₂₄ are 0.405 Jmol⁻¹K⁻² and 0.00205 Jmol⁻¹K⁻⁴ and for BaLa₁₀Ir₄O₂₄ γ and β value are 0.47 Jmol⁻¹K⁻² and 0.00224 Jmol⁻¹K⁻⁴. Corresponding Debye temperatures were calculated to be 333 K for SrLa₁₀Ir₄O₂₄ and 324 K for BaLa₁₀Ir₄O₂₄. The magnetic entropy in the low-T regime can then be calculated as S_mag(T) =  C_mag/T dT (Fig. 3b), where C_mag = C_tot – C_electron - C_phonon, yielding S_mag = 6.31 J mol⁻¹ K⁻¹ for SrLa₁₀Ir₄O₂₄ and 6.17 J mol⁻¹ K⁻¹ for BaLa₁₀Ir₄O₂₄ (Fig. 3b). The expected entropy of the magnetic transition (S_mag) equals to Rln(2 J+1), where R is the gas constant and J is the total angular momentum. For J_eff = ½, the expected value is 5.76 J mol⁻¹ K⁻¹, in fair agreement with that measured here.

Figure 3

Heat capacity of SrLa₁₀Ir₄O₂₄ and BaLa₁₀Ir₄O₂₄.

Red points are for SrLa₁₀Ir₄O₂₄ and blue points are for BaLa₁₀Ir₄O₂₄. Fits are shown in black (see text). (a) Total heat capacity (b) magnetic contribution to the heat capacity. Inset to (b) shows the entropy of the magnetic transition.

Resistivity measurement shows that SrLa₁₀Ir₄O₂₄ exhibits insulating behavior (Fig. 4), which is expected given that the crystal structure features isolated IrO₆ octahedra. With the caveat that the behavior is evaluated in a narrow temperature range of 260 K to 350 K, it was found that the resistivity is best modeled by simple thermally activated hopping, with Ea ~0.26 eV, while three-dimensional and two-dimensional variable range hopping and small polaron models yield poorer agreement with the measured data. Unfortunately, BaLa₁₀Ir₄O₂₄ crystal specimens are too small for a conductivity measurement at this time. Due to the similar crystal structure, one may expect similar electronic transport behavior to that of the Sr analogue.

Figure 4

Resistivity vs. temperature for SrLa₁₀Ir₄O₂₄.

Discussion

Flux crystal growth is an important approach to grow single crystals of new materials^{21,22,23,24,25,26}. For exploratory crystal growth of new iridates, KOH and K₂CO₃ fluxes have typically been used^27,28. Remarkably, for the synthesis of SrLa₁₀Ir₄O₂₄, Ir metal was used as the source of Ir and was oxidized to Ir(IV) in the SrCl₂ flux. It is known that some fluxes like KOH can dissolve O₂ from the atmosphere to provide an oxidizing environment; mostly Ir(V) compounds have been synthesized from KOH flux, but some Ir(VI) and Ir(V) oxides are also reported^29,30. Evidently SrCl₂ also dissolves sufficient O₂ from the atmosphere to oxidize Ir metal to Ir(IV). However, under the conditions of our synthesis, SrCl₂ apparently provides a less oxidizing environment compared to KOH, as we found no higher oxidation state products. EDS shows no evidence of chlorine incorporation in the crystals.

MLa₁₀Ir₄O₂₄ (M = Sr, Ba) compounds have similar lattice parameters as the Mo(VI) oxide, Sr₉La₂Mo₄O₂₄¹³. Although the atomic coordinates of Sr₉La₂Mo₄O₂₄ were not reported¹³, the stoichiometry and similar lattice parameters leads one to expect that the structures of MLa₁₀Ir₄O₂₄ (M = Sr, Ba) and Sr₉La₂Mo₄O₂₄ are closely related. Apparently this structure type can adjust its M:La ratio to accommodate either the tetravalent Ir or the hexavalent Mo. When only alkaline earth metal is involved and rare earth metal is excluded, the structure type can host transition metal with mixed oxidation states, which has been shown by the synthesis of Ca₁₁Re₄O₂₄³¹ Sr₁₁Re₄O₂₄³² and Ba₁₁Os₄O₂₄³³.

Magnetization data show that the effective moment of MLa₁₀Ir₄O₂₄ (M = Sr, Ba) is significantly reduced vis-à-vis that expected for a J = ½ Kramer’s ion. In the case of 6 M-BaIrO₃ it has been suggested that such a reduced effective moment may arise from the d electron hybridization with oxygen p states¹⁸. Putting any such argument on a stronger, more quantitative footing calls for a broader materials search and theoretical input beyond the scope of this report.

As mentioned earlier, the J_eff = ½ state has been implicated as foundational to the understanding of iridate physics, although this description rigorously applies only in the case of an isolated and ideal octahedral crystal field^6,7. The former criterion eliminates band structure effects and super-exchange, while the latter guarantees the symmetry of the J_eff = ½ wavefunction (assuming that the e_g states lie at sufficiently high energy that the contribution from excited configurations such as t_2 g⁴e_g¹ are negligible. This latter assumption has been questioned recently by Katakuri et al. from quantum chemical calculations³⁴. By isolating the octahedra and thus eliminating bandwidth and super-exchange as a competing influence on the electronic structure³⁵, compounds such as MLa₁₀Ir₄O₂₄ offer a platform for testing the intrinsic nature of the J_eff = ½ description.

Conclusion

In summary, we report the discovery and characterization of two new quaternary iridates, SrLa₁₀Ir₄O₂₄ and BaLa₁₀Ir₄O₂₄, with the crystal structure similar to Ca₁₁Re₄O₂₄³¹, Sr₁₁Re₄O₂₄³² and Ba₁₁Os₄O₂₄³³ and Sr₉La₂Mo₄O₂₄¹³. By isolating the octahedra and thus eliminating bandwidth and super-exchange as a competing influence on the electronic structure, compounds such as these can provide a platform for testing the detailed nature and range of applicability of the J_eff = ½ description using, for example, resonant inelastic x-ray scattering. More generally, the synthetic approaches reported here provide valuable insights that can stimulate the efforts in crystal growth of new iridates, particularly quaternary iridates, that will be essential to achieving a broader understanding of correlated electron physics in the presence of strong spin-orbit coupling.

During the proofing process of this manuscript, we became aware of a paper reporting the synthesis and magnetic properties of SrxLa11−xIr4O24 (B.F. Phelan et al. Phys. Rev. B 91, 155117 (2015)). The crystallographic data presented by Phelan et al. for single crystal Sr4.25La6.75Ir4O24 are qualitatively the same as ours with a slightly smaller average Ir-O bond length that can be attributed to a slightly more Ir(V) concentration in the specimen of Phelan et al. Magnetic properties of the x = 1 member of this series (polycrystalline specimens in the Phelan et al. report) are comparable to those here, showing Curie-Weiss behavior at a cusp at ~12 K., Curiously, the effective moment for SrLa10Ir4O24 reported by Phelan et al differs considerably from that we find and is closer to that expected by an isolated Ir(IV). We do not have an explanation for this discrepancy, but note that the synthetic processes are different.

Methods

Syntheses

Crystals of the MLa₁₀Ir₄O₂₄ (M = Sr, Ba) were grown by a flux method. For the synthesis of SrLa₁₀Ir₄O₂₄, La₂O₃ (Alfa Aesar, 99.9%, 0.51 mmol), Ir metal (0.5 mmol) and anhydrous SrCl₂ (12.6 mmol) were loaded into a platinum crucible. The crucible was placed into a box furnace, heated to 1200 °C at 300 °C/hour, held at that temperature for 12 h, cooled to 900 °C at 12 °C/hour and finally cooled to room temperature by turning off the furnace. For the synthesis of BaLa₁₀Ir₄O₂₄, La₂O₃ (Alfa Aesar, 99.9%, 0.82 mmol) IrO₂ (0.82 mmol), anhydrous BaCl₂ (30 mmol) were loaded into a platinum crucible. The crucible was placed into a box furnace, heated to 900 °C at 300 °C/hour, then heated to 1200 °C at 12 °C/hour, held at 1200 °C for 12 h, cooled to 950 °C at 10 °C/hour and finally cooled to room temperature by turning off the furnace. For both compounds, the crystals were separated from the flux by dissolving the flux in water aided by sonication and then isolated with vacuum filtration and rinsing with acetone. The crystals are stable in air and water. They are black in color with an irregular polyhedral shape and the crystal sizes are about 100 microns from one face to the face across.

Single Crystal X-ray Diffraction and EDS

Single crystals with irregular polyhedral shape were selected and mounted on tips of glass fibers for X-ray diffraction. Intensity data were collected at room temperature on a STOE imaging plate diffraction system (IPDS-II) using graphite-monochromatized Mo–Kα radiation (λ = 0.71073 Å) operating at 50 kV and 40 mA with a 34 cm diameter imaging plate. For SrLa₁₀Ir₄O₂₄, individual frames were collected with a 15 min exposure time and a 1° ω rotation at a φ angle of 98°, while for BaLa₁₀Ir₄O₂₄, individual frames were collected with a 5 min exposure time and a 1° ω rotation at a φ angle of 78°. Data reduction and integration absorption correction were performed using X-Area software provided by STOE and the crystal structures were solved with SHELXL 97 software package³⁶. The parameters for data collection and the details of the structure refinement are given in Table 1. Atomic coordinates, isotropic thermal displacement parameters (U_eq) and occupancies of all atoms are given in Table 2 and selected bond lengths are given in Tables 3 for both compounds. Anisotropic displacement parameters are given in the supplemental material. The isotropic thermal parameter for Sr1 in SrLa₁₀Ir₄O₂₄ is relatively large and the anisotropic thermal parameters for Sr1 have an elongated ellipsoid shape. This may be a sign of disordering for this site. The Ba in BaLa₁₀Ir₄O₂₄, on the other hand, is well behaved. Electron dispersive X-ray spectroscopy data were collected on Oxford INCA Model 6498 and no discernible chlorine peaks were detected.

Table 1 Crystal data and structure refinement for MLa₁₀Ir₄O₂₄ (M = Sr, Ba) at 293(2) K.

Table 2 Atomic coordinates and equivalent isotropic displacement parameters (Ų × 10³) for MLa₁₀Ir₄O₂₄ (M = Sr, Ba) at 293(2) K with estimated standard deviations in parentheses.

Table 3 Selected bond lengths [Å] for MLa₁₀Ir₄O₂₄ (M = Sr, Ba) at 293(2) K with estimated standard deviations in parentheses.

Magnetism

The DC magnetic susceptibilities of the ground samples were measured using a Quantum Design MPMS XL SQUID magnetometer. Samples were measured under zero-field-cooled (ZFC) and field-cooled (FC) conditions in an applied field of 5000 G. For SrLa₁₀Ir₄O₂₄, the magnetization was measured upon warming the samples from 1.8 to 300 K. For BaLa₁₀Ir₄O₂₄, the magnetization was measured upon warming the samples from 2 to 300 K. The very small diamagnetic contribution of the gelatin capsule had a negligible contribution to the overall magnetization and was not subtracted.

Electrical Conductivity and Heat Capacity

Electrical conductivity of a single crystal of SrLa₁₀Ir₄O₂₄ was measured on a Quantum Design PPMS with a four-probe method. It was found that below 260 K the resistance is too large to be measured, thus data between 260 K and 350 K were measured. Heat capacity for both compounds was measured on the PPMS from 2 K to 30 K.