Synthesis and Structural Characterization of Layered Ni^+1/+2 Oxides Obtained by Topotactic Oxygen Release on Nd_2−xSr_xNiO_4−δ Single Crystals

Abstract

Layered nickelate oxides containing Ni¹⁺/Ni²⁺ are isoelectronic to Cu²⁺/Cu³⁺ compounds and of present interest with respect to recent findings of superconductivity in a series of different compositions. It is thereby questionable why superconductivity is still rare to find in nickelates, compared to the much larger amount of superconducting cuprates. Anisotropic d_z² vs. d_x²−y² orbital occupation as well as interface-induced superconductivity are two of the main advanced arguments. We are here interested in investigating the feasibility of synthesizing layered nickelate-type oxides, where the Ni¹⁺/Ni²⁺ ratio can be tuned by oxygen and/or cation doping. Our strategy is to synthesize Sr-doped n = 1 Ruddlesden–Popper type Nd_2−xSr_xNiO_4+δ single crystals, which are then reduced by H₂ gas, forming Nd_2−xSr_xNiO_4−δ via a topotactic oxygen release at moderate temperatures. We report here on structural studies carried out on single crystals by laboratory and synchrotron diffraction using pixel detectors. We evidence the general possibility to obtain reduced single crystals despite their increased orthorhombicity. This must be regarded as a milestone to obtain single crystalline nickelate oxides, which further on contain charge-ordering of Ni¹⁺/Ni²⁺, opening the access towards anisotropic properties.

1. Introduction

Recent findings of superconductivity in Nd_0.8Sr_0.2NiO₂ thin films at Tc~15 K renewed the interest in nickelates for high-T_c superconductivity research which is still essentially dominated by copper oxides [1]. The underlying concept is related to the isoelectronic configuration of Ni¹⁺/Ni²⁺, compared to Cu²⁺/Cu³⁺, i.e., a seemingly critical valence state of the transition metal is 3d⁹ − 3d⁸ [2,3,4]. A stimulating debate resulted in understanding the mechanism of the unconventional superconductivity in nickelates in comparison to the iso-structural and/or iso-electronic copper-based analogues. In particular, the pairing symmetry of the superconducting order parameter in nickelate superconductors is still a controversial and open question [5,6,7]. Theoretical calculations have supported the idea that nickelates have a similar gap symmetry to cuprates, with a dominant (d_x²−y²)-wave pairing, despite its multiorbital nature [1,5,6,7,8,9]. Alongside this theory, other propositions have been put forward, such as a d_x−y-wave gap or a two-gap model d_z² − wave + d_x²−y² − wave [10,11,12] or a transition between a gapped (d + is) − wave to a nodal d-wave pairing state [13] for neodymium-based nickelates. First-principles density functional theory calculations also showed the role of the 4f magnetism [14]. In addition, experimental spectroscopic studies on nickelate thin films have shown a mixture of d-wave and s-wave signals at different locations of the film surface, suggesting the formation of nonstoichiometric or secondary phases, thus complicating the determination of the true coupling symmetry of the superconducting gap [15]. Nickel is right next to copper in the periodic table, and it is long known to form compounds with similar structural configurations as high Tc cuprates. Hole doped superconducting cuprates show a Cu^2+/3+ mixed valence state, i.e., a 3d⁹ − 3d⁸ configuration, which is considered to be a crucial parameter for high-T_C superconductivity [16]. Its nickel analogue containing Ni^1+/2+, and thus with an identical electronic configuration, is therefore also supposed to be a promising candidate for superconductivity.

The appearance of superconductivity in Nd_0.8Sr_0.2NiO₂ thin films was therefore an important landmark intensifying research activity to synthesize new layered Ni^1+/2+ containing oxides. However, the stabilization of Ni¹⁺ is not that common but favored in higher-order Ruddlesden–Popper (RP)-type oxides with the general formula of A_n+1B_nO_3n+1, which, e.g., for n = 3 in case of Pr₄Ni₃O₁₀, only contains Ni²⁺. This phase can, however, be reduced in a topotactic reaction under mild conditions (under H₂ at 360 °C) towards Pr₄Ni₃O₈, with an average valence state of Ni^1.33+. The Jahn–Teller distortion in nickelates is generally much less pronounced and a better understanding of their respective electronic structure thus becomes crucial in defining the extent of hybridization of nickel d-orbitals.

Similar to the cuprates, the Ni^1+/2+ mixed valence state appears to be critical for superconductivity in infinite layered compounds [17]. This is further reinforced by the experimentally evidenced superconductivity in undoped Nd₆Ni₅O₁₂ thin films with an average valence state of Ni^1.2+ [18]. Nd₆Ni₅O₁₂ has been obtained by a low-T topotactic reduction via the CaH₂ method, leading to a mixed valence state of the nickel atoms of Ni¹⁺/Ni²⁺. The resulting layered structure is considered to be a T’ structural type similar to the electron-doped Nd_2−xCe_xCuO₄ superconductor [19]. However, unlike cuprates, superconductivity in nickelates was primarily seen only in thin-film samples, until the recent findings on La₃Ni₂O₇ single crystals showing superconductivity at Tc ≈ 80 K under high pressures exceeding 14 GPa [20]. This bulk-type superconductivity puts into question the idea that superconductivity in nickelates is induced by substrate/interface-induced effects, thereby emphasizing the importance of having single crystalline nickelate samples for understanding the origins of superconductivity.

The non-stoichiometric RP phases with n = 1, e.g., La₂CuO₄ or (Pr/Nd)₂NiO₄, present a special case, as they can easily uptake oxygen to form R₂BO_4+δ [21,22,23], and for which its ability to release oxygen is much less studied. Their structure consists of alternating perovskite and rock-salt layers along the stacking axis, as shown in Figure 1a.

Some of these RP oxides have been shown to release oxygen, either by being reduced in H₂ atmosphere or by the CaH₂ method. Cuprates like La₂CuO₄ and Nd₂CuO₄ have been topotactically reduced to form an oxygen vacancy-ordered framework R₂CuO_3.5, containing CuO₂ dumbbells with all copper atoms being formally monovalent Cu¹⁺ only [24,25]. We were therefore interested in exploring to what extent this type of reduction reaction can be transferred to RP nickelates and how oxygen vacancy ordering together with the nickel valence state can be tuned by partially substituting the trivalent rare-earth atoms by bivalent alkaline earth atoms, such as Sr. The feasibility to reduce RP frameworks with H₂ gas has been shown for polycrystalline Nd_2−xSr_xNiO_4−δ, where Ni¹⁺/Ni²⁺ valence states can indeed be formed [26,27].

The appearance or not of superconductivity in nickelates is presently discussed in terms of the d_z² vs. d_x²−y² orbital occupation as well as their degree of hybridization with oxygen orbitals. As an example, the double occupation of the d_z² orbital in Jahn–Teller distorted La₂CuO₄, with an increased apical Cu—O_ap bond length, coupled to a large p-d hybridization (d_x²−y²), is seen as an essential trademark to mediate high-temperature superconductivity upon La substitution by Sr or O doping [16]. The Jahn–Teller distortion in nickelates is much less pronounced, and a better understanding of their respective electronic structures thus becomes crucial in defining the extent of hybridization of nickel d-orbitals.

However, it is difficult to experimentally access the orbital occupations on polycrystalline samples by spectroscopic methods. This problem can be overcome by resonant inelastic X-ray scattering (RIXS) studied on oriented single crystals (bulk/thin-film), i.e., parallel and perpendicular to the c-axis. Beside the synthetic aspects, our motivation concerns further on the feasibility of Ni¹⁺/Ni²⁺ single crystal growth. Given the straightforwardness to easily reduce polycrystalline R_2−xSr_xNiO₄ at the higher Sr content (x > 0.3), we consequently explored the feasibility to reduce Nd_1.3Sr_0.7NiO_4+δ single crystals. The strategy here is to first synthesize Nd_1.3Sr_0.7NiO_4+δ single crystals, which are then reduced in H₂/N₂ gas mixtures at moderate temperatures towards Nd_2−xSr_xNiO_4−δ. Given that the final stoichiometry is similar to the cuprates reported above, the availability of Nd_2−xSr_xNiO_3.5 type crystals would open up a panoply of studies, i.e., detailed structural features as oxygen vacancy ordering, as well as their anisotropic electronic properties.

We report below on the synthesis conditions and structural characterization of the H₂-reduced Nd_1.3Sr_0.7NiO_4−δ single crystals and polycrystalline powder, with mixed Ni^1+/2+ valence state, using laboratory and synchrotron single-crystal and powder X-ray diffraction techniques.

2. Materials and Methods

Synthesis and single crystal growth. Polycrystalline Nd_1.3Sr_0.7NiO_4+d was synthesized by classical solid-state reactions at high temperature, starting from the precursor oxides. Stoichiometric amounts of Nd₂O₃, SrCO₃ and NiO powders (99.99% purity, Alfa Aesar, Kandel, Germany) were thoroughly ground and calcinated for 12 h at 900 °C in air. The obtained powders were then ground again and heated at 1250 °C for 24 h.

Nd_1.3Sr_0.7NiO_4+δ single-crystal growth was performed by using a two-mirror optical floating zone furnace as described elsewhere [28]. Seed and feed rods for crystal growth were obtained by hydrostatic pressing of the as synthesized Nd_1.3Sr_0.7NiO_4+δ powders at 10 bars in a cylindrical latex tube of 8 mm in diameter and 150 mm in length. To compensate for steady evaporation of NiO during the crystal growth, a 2% molar NiO excess was added to the stoichiometric starting material. Very dense polycrystalline rods for single crystal were obtained by subsequent sintering at 1250 °C for 18 h. The crystal growth was carried out by the travelling floating zone (TFZ) method, using a two-mirror optical floating zone furnace (NEC SC2, Japan) equipped with two 500 W halogen lamps. The growth was carried out in oxygen flux with a speed rate of 3.3 mm/h with the feed and seed rods counter rotating at 30 rpm.

Micron-sized Nd_1.3Sr_0.7NiO_4+δ single crystals and polycrystalline samples were reduced in a 5% H₂/N₂ gas flow at 560 °C, as decomposition into the binary oxides started to set in above 650 °C.

X-ray diffraction. Phase purity of the grown crystals was checked by laboratory X-ray powder diffraction on a D8 Discover diffractometer (Bruker, Karlsruhe, Germany) equipped with a Johansson monochromator for CuK_α1 radiation and a fast detector with high energy resolution (LynxEye XE-T). Powder diffraction patterns were analyzed using the FullProf suite [29] and the crystal structure is visualized using the VESTA freeware (ver. 3.5.8) [30]. Single crystal X-ray diffraction (SXRD) studies were carried out using a STADIVARI (MoK_α source, Xenocs Microfocus tube) diffractometer (STOE, Darmstadt, Germany) equipped with a DECTRIS 200 K Pilatus pixel-detector. In addition, synchrotron X-ray single crystal diffraction was carried out on the beamline ID28 at the European Synchrotron Facility (ESRF, Grenoble, France) (λ = 0.6968 Å), equipped with a 1 M Pilatus detector. Single-crystal data analysis was performed using the SNBL Tool Box [31], the CrysAlis blue (Rigaku-Oxford Diffraction) and an in-house developed software, with ALBULA (DECTRIS) for the visualization, as well as the X-Area software package (version 1.86, STOE, Darmstadt).

Electron microscopy. The crystal quality and elemental composition were checked by scanning electron microscopy (SEM) analysis using a JEOL JSM 6400 microscope (JEOL Ltd., Tokyo, Japan), equipped with an OXFORD INCA EDS instrument for atomic recognition via X-ray fluorescence spectroscopy. SEM/EDS analyses were performed on a cross section (6 mm in diameter) of the grown crystal after an accurate surface polishing and cleaning.

Thermogravimetric studies. To determine the overall oxygen stoichiometry, thermogravimetric (TGA) measurements were carried out using a NETZSCH thermo-balance (NETZSCH, Germany) applying a 5% H₂/N₂ gas mixture for reduction with a heating rate of 5 °C/min. Measurements were performed on ground single crystals.

3. Results and Discussion

3.1. Crystal Growth and Characterization

Figure 2 shows the obtained Nd_1.3Sr_0.7NiO_4+δ single crystal, 6 mm in diameter and 110 mm in length. Phase purity was checked by X-ray powder diffraction on a small section of the as-grown single crystal that was crushed into fine powder. From the corresponding XRD pattern, as shown in Figure 3a, the as-grown single crystal is single-phase without any impurities.

The pattern profile refinement carried out in the profile matching mode through the Fullprof software (5.10, version January 2023) against the I4/mmm space group gives the following lattice parameters: a = b = 3.7755(3) Å and c = 12.4548(1) Å. The chemical composition and homogeneity of the crystals were verified by SEM coupled with the EDX analysis on a section cut from the as-grown crystal. It was found to be free of cracks or inclusions of secondary phases (Figure 3b). The atomic percentages of nickel, strontium and neodymium are constant over the whole cross section, in agreement with the nominal Nd:Sr:Ni stoichiometry 1.3:0.7:1 (inset of Figure 3b).

3.2. Reduction

Polycrystalline Nd_1.3Sr_0.7NiO_4+δ obtained from a crushed single crystal was reduced in a 5% H₂/N₂ gas flow. The reduction curve is depicted in Figure 4, where the weight change is reported as a function of the temperature. The total weight loss amounts to almost 6%: a first oxygen loss is observed above 400 °C, while a second and larger loss starts at around 720 °C, corresponding to the decomposition into (1 − x/2)Nd₂O₃, (x)SrO and Ni, as checked by XRD. It thus results that the oxygen stoichiometry of the starting as-grown single crystal is 4.00(2), and there is the formation of the oxygen deficient phase Nd_1.3Sr_0.7NiO_3.54(2) at 560 °C, with an average Ni oxidation state corresponding to Ni^1.7+.

The important oxygen loss is accompanied by a change in the symmetry towards orthorhombic. After the reduction, the compound is single phase and with the profile matching refinement of laboratory X ray powder diffraction, data could be indexed in the orthorhombic Immm space group with lattice parameters a = 3.6163(4) Å, b = 3.8205(6) Å and c = 12.6505(6) Å (see Figure 5). The initial fit of the X-ray data is rather poor (χ² ≅ 12) as some Bragg reflections (especially those where the a-axis is involved) show significant asymmetries and anisotropic broadening which are difficult to model. The refinement improves by the introduction of anisotropic strains using the spherical harmonics model in the fit (χ² ≅ 5), as implemented in the FullProf suite. Indeed, lattice microstrains could be present due to fluctuations in the cell parameters if the reduction reaction is not homogeneous or to antiphase/twin boundaries originating from the tetragonal-to-orthorhombic transition, as will be discussed in detail in the following section.

The pretty different lattices parameters after reduction imply important changes in the a/b-axes, yielding an elevated orthorhombicity (orthorhombic strain r = (a − b)/(a + b) = 2.7%). From the increase in the c-axis, from 12.4548(1) Å of the starting phase to 12.6505(6) Å of the reduced one, it becomes evident that the O-vacancies are essentially introduced in the equatorial sites. Releasing half of the equatorial oxygen atoms in an ordered way would result in the formation of the so-called S-phase (see Figure 1b).

3.3. Single Crystal X-ray Diffraction

Figure 6a,b show the reconstructed (hk0) layers obtained on the single-crystal STOE STADIVARI diffractometer for the as-grown and reduced phase, respectively. The non-split single-crystal reflections of the starting crystal confirm the tetragonal symmetry, while the reflection splitting for the reduced orthorhombic phase is related to the formation of twin domains.

The symmetry reduction from tetragonal to orthorhombic leads to the loss of the diagonal mirror planes along [110]/[−110], becoming the common planes of the formed twin individuals (see Figure 6c). It leads to the typical splitting of (h00) reflections Δ, which is quantified by Δ = tan⁻¹ $\left(\frac{a}{b}\right)$ − tan⁻¹ ($\frac{b}{a}$). Δ becomes twice this value for (hh0) reflections along [110]. It implies that a maximum of four orthorhombic twin individuals can be achieved. The respective volume fractions can be quantified by analyzing the outer “satellites” of the respective (hh0) and (−hh0) reflections. Typical values found for the reflection splitting Δ in RP phases, related to the tetragonal/orthorhombic phase transitions, are generally below Δ ≈ 1°; as an example, the stoichiometric Nd₂NiO₄ shows a domain separation with Δ = 0.89° [16]. Given the important difference of the a/b lattice parameters in Nd_1.3Sr_0.7NiO_3.54, the domain separation is expected to achieve almost four times this value, i.e., a Δ corresponding to 3.72°. Maintaining a faultless single-crystal characteristic during the reduction reaction thus presents an important obstacle, as the related strain increase is expected to destroy its pristine crystallinity.

The inspection of the reconstructed diffraction planes given in Figure 6 clearly indicates that, despite the elevated orthorhombicity, all diffraction peaks still remain sharp. The systematic presence of four contributions for (h00) reflections (Figure 6e), thereby indicates the formation of four twin individuals, presenting an equal volume fraction of (1/4). This is in agreement with a micro-twinning of the orthorhombic domains, as any unequal volume fraction distribution would rather indicate the formation of large, i.e., individual, domains. A micro twinning is also in favor of minimizing the stress between individual domains. A complete transformation towards the reduced orthorhombic phase is indicated by the entire absence of the former, tetragonal (h00)-reflections, which are supposed to appear in the center of the four orthorhombic (h00/0h0) reflections, thus indicating a complete reduction reaction.

The angular separation of the twin domains in Nd_1.3Sr_0.7NiO_3.54 is determined to correspond to Δ = 3.2° (Figure 7b). The reflection splitting measured is 0.5° lower compared to what is expected from the orthorhombic lattice parameters obtained for polycrystalline Nd_1.3Sr_0.7NiO_3.54. This might be related to reduced reaction kinetics when compared to the powder samples having a much lower grain size. Another reason might be a different microstructure for sintered ceramics and crystallites obtained by recrystallization from the melt.

A more complex domain formation is also indicated by the diffuse scattering as shown for the (−301) reflection in Figure 7d. This is similar to what has been observed during the electrochemical reduction of Pr₂NiO_4+δ performed at room temperature [32].

Investigations to explore possible oxygen ordering have been carried out by single-crystal synchrotron diffraction at ID28@ESRF, as shown in Figure 7 and Figure 8.

While no extra typical superstructure reflections related to oxygen ordering could be uncovered [33], several broader reflections and powder lines are observed, indicating a partial decomposition into the respective binary oxides, i.e., NiO, Nd₂O₃ and SrO (see Figure A1). We underline that the quantity of the decomposed fraction remains very low and essentially requires synchrotron radiation for its detection, as we were unable to confirm these phases with laboratory diffraction methods (less than 0.2%).

The high brilliance of the synchrotron radiation thus enabled us to highlight slightly broadened reflections, appearing for all main reflections. They can be indexed by a modulation vector Q_n = ±0.40 b*, in line with the twin scheme shown in Figure 6c, suggesting a unidirectional Ni¹⁺/Ni²⁺ charge ordering along [010] (see Figure 8b). This value is in between the 1/3 stripe and 1/2 chequerboard like charge ordering as have been reported for Ni²⁺/Ni³⁺ type nickelates, and is theorized in [34]. The charge ordering is thus along the Ni-O vacancy direction, and consequently, it induces a kind of structural disorder, due to the random distribution of the vacancies, related to the non-stoichiometry. The charge ordering is established as an “average” ordering, rather than a discrete ordering scheme, the latter allowing us to establish rational multiples of the unit cell. A rather arbitrary arrangement as encountered here is also supposed to impact the slightly different Ni¹⁺-O and Ni²⁺-O bond length distribution on a local scale. It presents thus a valence state ordering, realized on a shorter length scale, which appears by diffraction in an averaged Ni¹⁺-O and Ni²⁺-O bond length distribution. Despite the random distribution of oxygen vacancies, the charge ordering is of 3D character, as indicated by the presence of respective satellite reflections in higher-order planes, e.g., the (hk1)-plane shown in Figure 7d.

4. Conclusions

High-quality single crystals of Nd_1.3Sr_0.7NiO_4+δ were synthesized by the floating zone method and reduced towards Nd_1.3Sr_0.7NiO_3.54 by annealing in a H₂ atmosphere. Despite their high orthorhombicity, they maintained their crystalline integrity throughout the reduction reaction.

The high brilliance of the synchrotron radiation thus enabled us to highlight slightly broadened reflections, appearing for all main reflections. They can be indexed by a modulation vector Q_n = ±0.4 b*, suggesting unidirectional Ni¹⁺/Ni²⁺ charge ordering along [010] (see Figure 8). This value is in between the 1/3 stripe and 1/2 chequerboard-like charge ordering as has been reported for Ni²⁺/Ni³⁺ type nickelates [34]. It thereby appears unusual to observe Ni¹⁺/Ni²⁺ charge ordering already at room temperature, as this kind of electronic ordering is supposed to appear at much lower temperatures, e.g., as found below 100 K for (Pr_0.3La_0.7)₄Ni₃O₈ [34]. This might be a consequence of the fixed arrangements of the O-vacancies, favoring to establish charge ordering on a local level. The charge ordering is, however, established as an “average” ordering, rather than a discrete ordering scheme, which would involve rational multiples of the unit cell. It presents thus a locally defined valence state ordering, which, from a structural point of view, appears as a random arrangement of Ni¹⁺-O and Ni²⁺-O bond length distribution.

A rather arbitrary arrangement as encountered here is also supposed to impact the slightly different Ni¹⁺-O and Ni²⁺-O bond length distribution on a local scale without long-range ordering. It is, however, unclear whether such an averaged charge ordering in nickelates shows a similar competition towards superconductivity as discussed for d-wave superconductors along the Cu-O-Cu bonds in cuprates [35].

In this regard, Tc gets almost suppressed for striped phased La_2−xBa_xCuO₄, alongside the coexisting charge and spin ordering. To judge this influence, a more systematic study on the evolution of charge ordering with the Sr-doping, i.e., changes in the Ni¹⁺/Ni²⁺ ratio, is required.

The general possibility to obtain reduced single crystals despite their increased orthorhombicity must be regarded as a milestone to obtain single crystalline nickelate oxides containing Ni¹⁺/Ni²⁺, opening to access towards anisotropic properties. The Ni¹⁺/Ni²⁺ valence distribution was evidenced as charge ordering, behaving as in the Sr-doped oxides on an average scale, i.e., not established as long-range translational ordering. An average valence distribution might readily allow charge fluctuations on a shallow energy range, providing an attractive way to identify superconductivity descriptors.