Complex metal oxides have been the subject of extensive study due to the wide variety properties they exhibit. These range from electronic and magnetic behaviors such as ferroelectricity, superconductivity and magnetoresistance to an extensive array of catalytic and electrochemical phenomena. As the chemical and physical behaviors exhibited by metal oxides tend to depend strongly on the electric configurations of the metal cations they contain (defined by a combination of oxidation states and coordination environments), there has been an enduring interest in establishing composition-structure-property relations in extended oxide systems to explore these features. These studies have revealed that a number of elements exhibit ‘disfavored’ oxidation states in oxide environments, i.e. oxidation states that appear to be thermodynamically accessible (sufficient lattice energy to overcome the required ionization energy) but are unstable with respect to disproportionation, when the metal is located in an extended oxide framework.

The instability of some of these disfavored states, such as the disproportionation of Pb³⁺ and Bi⁴⁺ in Pb₂O₃ and BiO₂ respectively,^{1, 2} can be accounted for by universal chemical concepts—in this instance the global instability of ns¹ electron configurations in main group metals leads to Pb₂O₃ and BiO₂ being better described as Pb^IIPb^IVO₃ and Bi^IIIBi^VO₄ respectively.

However, similar disproportionations are observed in transition-metal systems, where the instability of the metal oxidation state cannot be easily attributed to a global instability of a particular electron count but appears to arise from the favorability of particular combinations of d-electron count and local coordination environment. For example, AgO is better described as Ag^IAg^IIIO₂,³ with the disproportionation of Ag^II attributed to the favorability of locating d¹⁰ Ag^I in a linear coordination and d⁸ Ag^III in square-planar coordination sites within the oxide framework. Likewise, analogous disproportionations of d⁷ cations, such as Pd^III or Pt^III are observed, driven by the favorability of locating d⁶ Pd^IV/Pt^IV cations in octahedral environments and d⁸ Pd^II/Pt^II cations in square-planar coordinations, in phases such as K₂Pd^II₃Pd^IVO₆ and CdPt^IIPt^IV₂O₆.^{4, 5}

Recently we observed the disproportionation of d⁷ Rh^II during the topochemical reduction of the Ruddlesden-Popper LaSrM_0.5Rh_0.5O₄ (M=Co, Ni) and perovskite LaM_0.5Rh_0.5O₃ oxides, with the reduced phases (LaSrM_0.5Rh_0.5O_3.25 and LaM_0.5Rh_0.5O_2.25 respectively) hosting d⁸ Rh^I in square-planar coordination sites, and d⁶ Rh^III in 5-coordinate, square-based pyramidal sites.^{6, 7} Here we describe the first observation of the disproportionation of d⁷ Co^II in an extended oxide, which occurs during the topochemical reduction of the double perovskite oxide LaSrCoRuO₆ to LaSrCoRuO₅.

Previous work revealed that rapidly quenching the double perovskite oxide LaSrNiRuO₆ through a R-3 to P2₁/n phase transition (T≈400 °C)⁸ increased the reactivity of this oxide phase with CaH₂, allowing the preparation of the infinite layer phase, LaSrNiRuO₄, by topochemical anion deintercalation.^{9, 10} The corresponding cobalt phase, LaSrCoRuO₆,^{11, 12} exhibits an analogous phase transition at T≈450 °C. Rapidly quenching LaSrCoRuO₆ through its R-3 to P2₁/n phase transition also enhances its reactivity enabling the preparation of the infinite layer phase LaSrCoRuO₄ via reaction with binary metal hydrides, as will be described in detail elsewhere. However, in contrast to the LaSrNiRuO_6-x system, quenched samples of LaSrCoRuO₆ can be reduced to a phase of intermediate oxygen content (shown to be LaSrCoRuO₅ by oxidative thermogravimetric analysis) by reaction with a Zr getter at 450 °C.

Synchrotron X-ray powder diffraction (SXRD) data collected from LaSrCoRuO₅ could be indexed using a body-centered monoclinic unit cell (a=5.40 Å, b=5.41 Å, c=8.16 Å, γ=90.5 °) consistent with the retention of the perovskite framework from the LaSrCoRuO₆ parent phase. However, close inspection revealed a series of weak additional reflections in the SXRD data that could not be indexed by this cell. Electron diffraction (ED) data collected from LaSrCoRuO₅, shown in Figure 1, are consistent with a 2×2×1 cell expansion compared to the LaSrCoRuO₆ parent phase (2√2×2√2×2 compared to a simple ABO₃ perovskite unit cell).^{13, 14} This expanded cell accounts for all the additional weak peaks observed in the SXRD data and can also index neutron powder diffraction (NPD) data collected at room temperature from LaSrCoRuO₅.

Considering the A₂BB'O₅ composition and the 2√2×2√2×2 cell expansion of the phase, a number of anion-vacancy ordered and B-site cation ordered perovskite structural models were considered for LaSrCoRuO₅. It was observed that a good fit to the SXRD and NPD data could be achieved using a model based on the anion-vacancy ordered structure of LaNi_0.9Co_0.1O_2.5 which consists of a network of apex-linked 6-coordinate octahedral, 5-coordinate square-based pyramidal and 4-coordinate square planar BO_x units.¹⁵ The model was modified to take account of the rock salt ordering of the Co and Ru cations, so that the Ru centers were exclusively located within 5-coordinate sites, while the Co centers occupied both 6- and 4-coordinate sites within a monoclinic unit cell (a=10.8128(2) Å, b=10.8231(2) Å, c=8.1626(1) Å, γ=90.55(1) °) with P112₁ space group symmetry, as shown in Figure 2. The model was refined against the NPD data to achieve a good fit (wRp=6.33 %) as shown in Figure 1 and described in detail in the Supporting Information.¹⁶

The crystal structure of LaSrCoRuO₅ shown in Figure 2 reveals that the cobalt cations occupy two distinct sites within the oxide framework. A 4-coordinate planar site and a 6-coordniate octahedral site. The location of the cobalt cations in two distinct sites is reminiscent of the local-coordination-driven disproportionation of transition metals with d⁷ electron counts observed for Pd³⁺ and Pt³⁺ and more recently Rh²⁺, and suggests the disproportionation of Co²⁺ into Co¹⁺ (square-planar) and Co³⁺ (octahedral). Analysis of the local coordination environments of the cobalt centers is hampered by the lack of reported Co¹⁺O₄ units for comparison. However, the observed bond lengths of the CoO₄ units in LaSrCoRuO₅ (Co−O=2.032(11) Å×2; 2.119(9) Å×2) are significantly longer than those in the Co²⁺O₄ units reported in Sr₃Co₂O₄Cl₂ (Co−O=2.007(1) Å×4)¹⁷ or Sr₂CoO₂Cu₂S₂ (Co−O=1.995(1) Å×4)¹⁸ consistent with assignment of Co¹⁺O₄ for the units present in LaSrCoRuO₅. Bond valance sums (BVS)¹⁹ calculated using parameters for Co²⁺ yield values of LaSrCoRuO₅ : Co+1.42, Sr₃Co₂O₄Cl₂ : Co+1.70 and Sr₂CoO₂Cu₂S₂ : Co+1.76. The CoO₆ units in LaSrCoRuO₅ have a rather irregular shape but exhibit an average bond length of <Co−O=2.004 Å> (BVS=Co+2.69) compared to <Co−O=2.033 Å> (BVS=Co+2.38) the Co^IIO₆ units in the LaSrCoRuO₆ parent phase.¹² Thus, it can be seen that the BVS values of the square-planar (BVS=Co+1.42) and octahedral sites (BVS=Co+2.69) in LaSrCoRuO₅ differ by 1.27 units. This difference is significantly larger than the difference between the octahedral and tetrahedral sites in the Co²⁺ brownmillerite phase La₂Co₂O₅ (CoO₆ BVS=Co+2.23; CoO₄ BVS=Co+2.07, Δ=0.16)²⁰ or the difference between octahedral and square-planar sites in the Ni²⁺ phase La₂Ni₂O₅ (NiO₆ BVS=2.08; NiO₄ BVS=2.11, Δ=0.03)²¹ and provides strong support for the disproportionation of Co²⁺ in LaSrCoRuO₅.

In an attempt to further confirm the disproportionation of Co²⁺, cobalt EELS data collected from LaSrCoRuO₅. These data show a single set of Co L₂ and L₃ peaks (Figure S14 in the Supporting Information) and thus represent the superposition of signals from both the square-planar and octahedral cobalt sites. In the absence of a Co¹⁺ oxide standard we are unable to know if a Co¹⁺/Co³⁺ oxidation state combination would be expected to lead to a resolvable splitting of the L₂ and L₃ peaks. It should be noted that splitting of Co²⁺/Co³⁺ signals is not resolvable for Co₃O₄.²² The L₃/L₂ intensity ratio (4.83) and L₃-L₂ energy difference (15.06 eV) from the data are broadly consistent with Co²⁺.

Magnetization data collected from LaSrCoRuO₅ indicate that, in common with many other topochemically reduced phases containing cobalt, samples of LaSrCoRuO₅ contain small quantities of ferromagnetic, elemental cobalt not detectable by diffraction. The magnetization of LaSrCoRuO₅ was therefore measured using the ‘ferromagnetic subtraction’ method described in the Supporting Information. A plot of the magnetic susceptibility of LaSrCoRuO₅ against temperature (Figure 3a) can be fit by the Curie–Weiss law in the range 140<T/K<300. However, the extracted Curie constant (C=3.76 cm³ K mol⁻¹; θ=+82.7 K) is much larger than can be accounted for by a combination of Co¹⁺, Co³⁺ and Ru³⁺ cations (even with the cobalt centers in high-spin states), suggesting strong magnetic interactions are present between the metal centers in this temperature range.

On cooling below 120 K there is a large increase in the saturated ferromagnetic moment of the samples (Figure 3b), increasing from 0.03 μB per fu at 150 K (a value that is attributed to the presence of an elemental Co impurity) to ≈1.25 μB per fu at 2 K, indicative of a ferromagnetic state, however NPD data collected at 5 K show no additional features indicative of magnetic order, as described in the Supporting Information.

The bond lengths of the square-planar and octahedral cobalt sites in LaSrCoRuO₅ are consistent with a high spin, S=1 Co¹⁺ center, and a low spin, S=0 Co³⁺ respectively. Thus, the most significant magnetic couplings in the system will be between the square-planar Co¹⁺ centers, which have a (d_xz/yz)⁴(d_xy)²(d_z2)¹(d_x2−y2)¹ electronic configuration, and the Ru³⁺ centers located in square-based pyramidal sites which have a (d_xz/yz)⁴(d_xy)¹(d_z2)⁰(d_x2−y2)⁰ electronic configuration.

As shown in Figure 3c, the Co¹⁺ and Ru³⁺ centers are magnetically coupled by either (Ru4d_x2−y2)−O2p−(Co3d_x2−y2) or (Ru4d_z2)−O2p-(Co3d_x2−y2) σ-type super exchange or (Ru4d_z2)−(Co3d_z2) direct exchange. Given that the Ru 4d_x2−y2 and 4d_z2 orbitals are empty and the corresponding Co3d orbitals are half filled, all of these interactions will be ferromagnetic,²³ consistent with the low-temperature magnetization data.

The disproportionation of Co²⁺ evident in LaSrCoRuO₅ is surprising. As noted above, other transition metal cations with d⁷ electron counts (e.g., Pd³⁺, Pt³⁺, Rh²⁺) are observed to disproportionate in oxide environments, driven by the presence of ‘preferred’ coordination sites. However, to date, this behavior has been restricted to 4d and 5d transition metals where the stronger ligand fields (compared to 3d metals) provide a larger energetic stabilization for the d⁶ octahedral and d⁸ square-planar electron-count/coordination combinations. It is therefore unexpected to see Co²⁺, a common oxidation state with a modest ligand field in oxides, undergo a coordination-site driven disproportionation.

There are limited examples of 3d transition metal cations, such as Fe⁴⁺ and Ni³⁺ disproportionating in extended oxides. However, in these cases the disproportionation of the metal center (e.g. Fe⁴⁺ in CaFeO₃ or BaFeO₃; Ni³⁺ in TlNiO₃)^24-26 is driven by a metal-insulator phase transition driven by the presence of a single electron in the σ-band of these oxides phases, rather than coordination site preference.

The unique observation of coordination-site driven disproportionation of Co²⁺ in LaSrCoRuO₅ suggests that the topochemical reaction which forms LaSrCoRuO₅ may act to ‘select’ this phase, as the disproportionated structure is a local energy minimum in composition-structure space in the reaction path between LaSrCoRuO₆ and LaSrCoRuO₄. Indeed, the same argument can be applied to the topochemical reactions which form the Rh^I/Rh^III disproportionated phases reported previously.^{6, 7} In combination these observations suggest further coordination-site driven disproportionated oxide phases could be accessible by this type of low-temperature reaction, presenting an opportunity to prepare a range of transition metal oxides with a rich variety of novel metal oxidation-state/coordination geometry-combinations.