Research ArticlePHYSICS

Spontaneous phase segregation of Sr2NiO3 and SrNi2O3 during SrNiO3 heteroepitaxy

See allHide authors and affiliations

Science Advances  05 Mar 2021:
Vol. 7, no. 10, eabe2866
DOI: 10.1126/sciadv.abe2866


Recent discovery of superconductivity in Nd0.8Sr0.2NiO2 motivates the synthesis of other nickelates for providing insights into the origin of high-temperature superconductivity. However, the synthesis of stoichiometric R1−xSrxNiO3 thin films over a range of x has proven challenging. Moreover, little is known about the structures and properties of the end member SrNiO3. Here, we show that spontaneous phase segregation occurs while depositing SrNiO3 thin films on perovskite oxide substrates by molecular beam epitaxy. Two coexisting oxygen-deficient Ruddlesden-Popper phases, Sr2NiO3 and SrNi2O3, are formed to balance the stoichiometry and stabilize the energetically preferred Ni2+ cation. Our study sheds light on an unusual oxide thin-film nucleation process driven by the instability in perovskite structured SrNiO3 and the tendency of transition metal cations to form their most stable valence (i.e., Ni2+ in this case). The resulting metastable reduced Ruddlesden-Popper structures offer a testbed for further studying emerging phenomena in nickel-based oxides.


Rare-earth nickelates (RNiO3, where R denotes the lanthanide) exhibit a rich array of structural and physical properties (13). Although bulk LaNiO3 (LNO) is a rhombohedral structured paramagnetic metal at all temperatures, other RNiO3 compounds display metal-to-insulator transitions (MITs) from a high-temperature orthorhombic metallic phase to a low-temperature monoclinic insulating phase. The orthorhombic structure (Fig. 1A) evolves from a distortion of the cubic structure due to a small Goldschmidt tolerance factor (t). The drop in t with decreasing R cation radius tends to drive the distortion and reduce the Ni─O─Ni bond angle, resulting in a reduction in Ni 3d–O 2p orbital overlap. Consequently, the bandwidth drops and the transition temperature goes up.

Fig. 1 Epitaxial synthesis of SrNiO3−δ (SNO) films with a 1:1 Sr/Ni ratio.

(A) Schematic crystal structures for bulk NdNi3+O3 and SrNi4+O3. Tolerance factor t=rA+rO2(rB+rO), where rA, rB, and rO are the ionic radius of A and B cations and oxygen, respectively, in a set of ABO3 crystal structures. Sr doping in NdNiO3 (Nd1−xSrxNiO3) would increase t due to the larger ionic radius of Sr2+ ion. (B) XRD θ-2θ scan of SNO films. The left inset shows the RHEED pattern for an as-grown SNO film on LSAT(001) viewed along the [100] zone axis. The right inset shows the magnified 004 peak with clear thickness fringes. # indicates that the peak originated from our XRD setup. a.u., arbitrary units.

Doping and strain engineering are two common ways to modify the electronic states of RNiO3 for exploring emergent physical properties and potential applications (410). The recent discovery of superconductivity in infinite-layered Nd0.8Sr0.2NiO2 on SrTiO3 substrates motivates the synthesis of other related nickelates in searching for high Tc materials and exploring the origin of the observed superconductivity (11). However, insulating behavior without the presence of superconductivity was observed in bulk Nd0.8Sr0.2NiO2 samples (12, 13), suggesting that superconductivity in this system may arise from interface or strain-related effects. Hence, the synthesis of high-quality stoichiometric Nd1−xSrxNiO3 thin films with controlled x becomes a prerequisite for further harnessing superconductivity in this system. However, the synthesis of stoichiometric R1−xSrxNiO3 thin films over a wide range of x has proven to be difficult (10, 11, 14) because of the instability of Ni oxidation states greater than +3. It is thus of considerable interest to explore how the structure and properties of nickelate thin films evolve at higher x, why structural decay occurs as x increases, and whether the end member SrNiO3 can be stabilized as a (pseudo)cubic perovskite by epitaxial growth. In addition, a Ruddlesden-Popper (RP)–type secondary phase is often observed during the growth of nickelate thin films (14). However, it is not clear why the RP secondary phase forms so readily, and further exploration is needed.

In light of the fact that the t for SrNiO3 is ~1.08, bulk SrNiO3 is expected to form a hexagonal structure (see Fig. 1A) with face-shared NiO6 octahedra (15). This structure is not symmetry-matched with traditional perovskite substrates such as SrTiO3 or LaAlO3. First-principles calculations for hexagonal SrNiO3 predict this material to be a half-metallic ferromagnet with promising applications in spintronics and magneto-optical devices (16). In addition, Ni4+ has been proposed to be responsible for the high OER performance found in modified nickelate electrocatalysts (17, 18). SrNiO3 with Ni4+ is unstable due to its high Gibbs formation energy (19), and as a result, there has been limited success in the synthesis of bulk, hexagonal-structured SrNiO3 (15, 20). In addition, successful epitaxial growth of SrNiO3−δ (where 0 ≤ δ < 1) thin films has not yet been reported.

Here, we present a study of the synthesis of SrNiO3−δ thin films by oxygen plasma–assisted molecular beam epitaxy (OPA-MBE). By keeping the Sr/Ni ratio at 1:1 during the deposition, we find that instead of forming phase pure SrNiO3, spontaneous phase segregation occurs. Our combined diffraction, spectroscopy, and microscopy dataset reveals that two complementary oxygen-deficient RP phases Sr2NiO3 and SrNi2O3 evolve together, and that these effectively balance the overall film stoichiometry, stabilize the preferred Ni2+ oxidation state, and satisfy the epitaxial relationships for both in-plane (IP) film/substrate and out-of-plane (OOP) Sr2NiO3/SrNi2O3 grain interfaces. In addition, a combination of x-ray linear dichroism (XLD) measurements and density functional theory (DFT) calculations reveals the presence of a planar structure in the Sr2NiO3 phase with a large IP orbital occupancy, which may open new avenues for investigation of emergent phenomena in this system.


Epitaxial growth of strontium nickel oxide thin films

Twenty-five-nanometer-thick SrNiO3−δ (SNO) thin films with a fixed 1:1 Sr/Ni flux ratio (as calibrated by a quartz crystal microbalance at the growth position before growth) were deposited on (001)-oriented (LaAlO3)0.3(Sr2AlTaO6)0.7 (LSAT) substrates by OPA-MBE. A typical reflection high-energy electron diffraction (RHEED) pattern taken along the [100] zone axis (Fig. 1B) reveals bright, sharp streaks, indicative of a planar surface and an epitaxial orientation with the substrate. The high-resolution x-ray diffraction (XRD) θ to 2θ scan in Fig. 1B shows clear thickness fringes (right inset in Fig. 1B) and strong Bragg peaks, indicating a high extent of crystallographic ordering. A wide-range reciprocal space map (RSM) (fig. S1) confirms the epitaxial relationship. All SNO film peaks in Fig. 1B can be assigned to (00l) reflections with an OOP (c) lattice parameter that is a multiple of 6.3 Å, a value much larger than that of either bulk hexagonal SrNiO3 (a = 5.355 Å and c = 4.86 Å) or the expected cubic perovskite counterpart (a = 3.81 Å) (15, 21). This expanded lattice parameter matches that of Sr2NiO3 (21), for which the Sr/Ni ratio is clearly not 1:1. Scanning transmission electron microscopy (STEM) images in Fig. 2 reveal that the film consists of layered RP phases. The body-centered symmetry of these compounds results in the absence of diffraction peaks for an odd value of h + k + l. Consequently, the XRD indicates an epitaxial RP phase with a c lattice parameter of 12.6 Å aligned perpendicular to the LSAT substrate surface.

Fig. 2 Phase segregation occurs in SNO films.

(A) Cross-sectional STEM image of SNO/LSAT along the [110] LSAT direction. Dashed circles clearly show phase segregation. (B) Magnified images of the Sr2NiO3 phase and the SrNi2O3 phase marked by green and blue rectangles in (A). (C and D) Cross-sectional STEM images of two different phases in SNO/LSAT along the [100] LSAT direction. EELS maps (right) from the area marked with a yellow rectangle in the STEM images (left). Scale bar, 2 nm. (E) APT tip reconstruction for SNO/LSAT. (F) A 3D volume rendering showing regions with greater than 13.5% Ni2+ (upper, gray scale). 2D Ni2+ (middle) and Sr2+ (lower) concentration maps highlighting where spatially localized Ni and Sr enrichment occurs.

Phase segregation in epitaxial SNO films

The high-angle annular dark field (HAADF) STEM image taken along [110] (Fig. 2A) reveals an abrupt interface between the first unit cell (u.c.) and the substrate. Phase segregation is apparent in the thin films consisting of more than 1 u.c. and two different phases can be clearly observed in the STEM image (see the dashed-circle regions). At the onset of SNO film growth, the cubic perovskite substrate can provide a lattice template for nucleation of a NiO2 monolayer followed by a SrO monolayer. This nucleation sequence does not preclude the formation of isolated point defects. As a result, interface engineering and limiting SNO deposition to 1 u.c. in the superlattices could generate novel ground states (such as Ni4+) unobtainable in bulk crystals (22). However, once the SNO film thickness exceeds 1 u.c., the large ionic radius ratio R(Sr2+)/R(Ni4+) and the instability of Ni4+ could promote phase segregation. Figure 2B shows magnified images of these two different phases that are marked by green and blue rectangles in Fig. 2A. Such spatially distributed, well-separated phases are similar to nanocomposite films reported earlier (2327). These phases are tentatively assigned to Sr2NiO3 and SrNi2O3 based on the HAADF STEM contrast, the 1:1 Sr/Ni ratio during deposition, and the presence of Ni2+ as determined by x-ray photoemission spectroscopy (XPS) and x-ray absorption spectroscopy (XAS) measurements (see Fig. 3). As seen in Fig. 2A, Sr2NiO3 and SrNi2O3 show atomically matched, epitaxial interfaces along the film growth direction. Moreover, both phases show the same c-axis lattice constant (~12.6 Å), consistent with the XRD analysis. The HAADF-STEM images and atomically resolved electron energy loss spectroscopy (EELS) mappings (Fig. 2, C and D) for the two domains viewed along [100] further confirm the phase assignment. Within Sr2NiO3 regions, repeating SrO bilayers and NiO monolayers are clearly resolved along the vertical direction (c axis), as shown in Fig. 2C. Within SrNi2O3 regions, SrO single layers and NiO bilayers are connected along the c axis (Fig. 2D), where a weak signal/interference of Sr in the second Ni layer is also observed, indicating that the phase purity of SrNi2O3 is not as high as that of the Sr2NiO3, or the layer may be more accurately defined as Sr1+xNi2−xO3. In addition, localized NiO clusters are observed in the Sr1+xNi2−xO3 regions by EELS mappings at lower magnification (fig. S2), which may account for the overall stoichiometry of SrNi2O3.

Fig. 3 Spectroscopy results of SNO films.

Ni 2p XPS (A) and Ni L edge XAS (B) of SNO films. The reference spectra for Ni2+ and Ni3+ were measured using a vacuum-annealed NdNiO3−δ film (VA-NNO) and a plasma-annealed NdNiO3 film (PA-NNO), respectively, both grown on SrTiO3(001). The spectrum for a NiO film grown on MgO (001) is included for comparison. The blue dashed lines in (A) denote the Ni 2p1/2 and Ni 2p3/2 peak positions of SNO films. The purple and blue dashed lines in (B) are guidelines for peak positions.

Although similar structured SrCu2O3 and CaCu2O3 have been reported in previous studies (2830), the structure of SrNi2O3 has not been predicted or characterized. SrCu2O3 is the first member (m = 3) of the homologous series of compounds with a stoichiometry of Srm−1Cum+1O2m (m = 3, 5, ...). The Cu-O sheet in SrCu2O3 is made of only Cu-O double chains without any bridging network (29). The CaCu2O3 structure is similar to that of SrCu2O3, but the Cu-O sheets in CaCu2O3 are strongly puckered. On the contrary, Sr2NiO3 films have previously been grown on Sr2TiO4-buffered LaSrAlO4 (100) by pulsed laser deposition (21), indicating that Sr2NiO3 phase is more stable and easier to form than SrNi2O3 phase. Therefore, when the SNO film thickness exceeds 1 u.c., the large ionic radius ratio R(Sr2+)/R(Ni4+) and highly unstable Ni4+ could promote the formation of Sr2NiO3 phase first. At the same time, a 1:1 Sr/Ni flux ratio could induce the formation of impure SrNi2O3, leading to the observation of Sr1+xNi2−xO3 phase and NiO clusters. Sr2NiO3 structure can be considered to be a derivative of the layered RP phases (An+1BnO3n+1, where n is the number of the layers of octahedra in the perovskite-like stack) and can be constructed by removing alternate oxygen ions in the two-dimensional (2D) planes along the IP direction of the A2BO4 (n = 1) structure. For simplicity, we continue to refer to the mixture of Sr1+xNi2−xO3 and NiO clusters as SrNi2O3 and refer to both Sr2NiO3 and SrNi2O3 as reduced RP phases because they have the same c-axis lattice parameter (~12.6 Å). The difference between the epitaxial relationships shown in the two white dotted circles in Fig. 2A may be induced by interfacial defects. In the left circle, Sr planes align well to form a coherent interface between the two different phases. In contrast, the grain boundary defects that are commonly found in metal oxide nanocomposites (31) result in poor alignment of Sr planes, as seen in the right circle.

To verify the apparent stabilities of Sr2NiO3 and SrNi2O3, we also deposited and characterized nominally stoichiometric Sr2NiO3 and SrNi2O3 films on LSAT, respectively, by setting the Sr and Ni fluxes to the desired values (either 2:1 or 1:2). By normalizing the Ni 3p XPS spectra in the survey scans, the resulting Sr 3d XPS spectra show distinctly different intensities (fig. S3A), quantitatively confirming the different Sr/Ni ratios in these three samples. Moreover, as shown in fig. S3B, the Sr2NiO3 film displays a RHEED pattern similar to that of the SNO film with a 1:1 Sr/Ni flux ratio. However, the Sr2NiO3 film becomes amorphous once exposed to air. By capping the sample with a 10-nm LaFeO3, the XRD OOP scan of the Sr2NiO3 film shows that the diffraction peak is weak and no thickness fringes are observed. In contrast, the RHEED pattern of the SrNi2O3 film indicates a much rougher surface and a mix of phases. The diffraction peak for the SrNi2O3 film is also weak, but the peak position is the same as those of the Sr2NiO3 film and the SrNiO3−δ film, indicating that all three films have the same c-axis lattice parameter (~12.6 Å). Therefore, the epitaxial interfaces between Sr2NiO3 and SrNi2O3 columns appear to be responsible for stabilizing these metastable phases to form the vertical “nanocomposite” structures when a 1:1 Sr/Ni flux ratio is used.

The phase-separated reduced RP film was further analyzed by 3D atom probe tomography (3D APT) (Fig. 2E). The 3D elemental reconstruction was accomplished through assignment of ionic species from the mass spectra to best reproduce the SNO stoichiometry (fig. S4). In general, the major Ni and Sr species were Ni2+ at 29 to 32 Da and Sr2+ at 42 to 44 Da. Here, 2+ is not the valence of Ni in the solid, but rather the mass-to-charge ratio of the ion detected in APT. The minor isotope 64Ni interferes with 16O21+ at 32 Da and is excluded from analysis of the Ni2+ species. Additional polyatomic/isobaric interferences for the Ni2+ and Sr2+ species were not observed. In turn, the Ni2+ and Sr2+ ionic species were used in evaluating the Ni and Sr distribution for simplicity. Potential elemental partitioning within a small subvolume of the SNO layer was visualized in 3D using iso-concentration surfaces of the major ionic species Ni2+ and Sr2+ species. Ni-rich regions are visualized using iso-concentration surfaces exceeding the average Ni2+ concentration (e.g., 13.5 ionic % Ni2+) in turn enabling observation of more Ni-rich phases (Fig. 2F). Ni-rich regions within the SNO matrix are generally circuitous and randomly distributed in space. 2D concentration maps (middle panel and lower panel of Fig. 2F) further highlight Ni-rich regions correlated to Sr-poor regions, strongly supporting our proposed Sr2NiO3 and SrNi2O3 model.

In situ x-ray photoemission and ex situ absorption spectroscopy

Figure 3A shows Ni 2p XPS spectra of a representative SNO film and three reference samples. Surface charging was consistently observed, and all spectra were shifted so that the O 1s peaks fall at 530 eV. Oxygen plasma annealing has been shown to be effective at healing oxygen vacancies in oxide thin films (32). Accordingly, a plasma-nnealed NdNiO3 film (PA-NNO) was used as a Ni3+ standard and a vacuum-annealed NdNiO3−δ film (VA-NNO) was used as a Ni2+ reference (33). The Ni 2p XPS spectrum for PA-NNO consists of two dominant spin-orbit split features at ~856.3 and 873.9 eV, respectively. These peaks shift to lower binding energy in VA-NNO, consistent with the previous reports (34). These two peaks are significantly narrower in SNO than in PA-NNO, and the binding energies are nearly the same as those for VA-NNO, suggesting the same Ni2+ valence in the two films. In addition, the Ni L edge XAS spectrum for SNO (Fig. 3B) shows the same shape and peak positions as the spectrum for VA-NNO. However, these differ significantly from those of PA-NNO. There are two distinct peaks α and β located at 853.3 and 854.3 eV, respectively, at the Ni L3 edge in the PA-NNO spectrum and a single feature at ~871.6 eV at the Ni L2 edge. In contrast, SNO shows one asymmetric feature located at ~853 eV at the Ni L3 edge and the Ni L2 edge is clearly shifted to lower photon energy at ~870.6 eV. Previous studies show that increasing oxygen nonstoichiometry results in a gradual change in valence from Ni3+ to Ni2+ and the Ni L3 edge is marked by an increase (decrease) in the intensity of peak α (β) as well as a shift to lower (higher) photon energy (3537). This change is clearly seen at the Ni L3 edge in Fig. 3B. Together, the Ni XPS and XAS spectra support the presence of divalent Ni in the SNO films. However, it should be noted that the spectra of VA-NNO and SNO are different from that of the Ni2+-containing standard, a NiO film grown on MgO(001). This difference is most likely due to the significantly different local atomic (RP versus rock salt) structures and bonding configurations.

As with the Cu L edge spectra in cuprates (38), peak β in the Ni L edge (Fig. 3B) is indicative of ligand holes formed as a result of Ni 3d–O 2p hybridization. The diminution of peak β in SNO and VA-NNO implies that the Ni 3d–O 2p hybridization strength decreases relative to PA-NNO. Owing to the strong hybridization between Ni 3d and O 2p bands at the Fermi level, O K edge XAS becomes another useful way to probe the valence of Ni with adjacent ligand holes. Figure 4A shows the O K edge XAS for the SNO film and reference samples. The spectra are normalized to zero and unity at the O K pre-edge and post-edge, respectively, and the spectra are shifted vertically for clarity. The spectra for PA-NNO and VA-NNO agree well with those reported in the literature (5, 37, 39, 40). For PA-NNO, a sharp pre-edge peak (marked by a dashed purple line) centered at ~528 eV, a broad feature near 535 eV, and a doublet at 540 to 545 eV reveal hybridization of the O 2p band with Ni 3d, Nd 4f, and Ni 4sp bands, respectively. The sharp pre-edge peak at ~528 eV in PA-NNO originates from the 3d8→ c̲3d8 (Ni3+) transition, where denotes a ligand hole and c̲ denotes an O 1s core hole (41). The intensity of the pre-edge peak is directly related to the ligand hole density, which in turn strongly affects transport behavior in nickelates films (8, 36, 42, 43). As seen in Fig. 4B and fig. S5, PA-NNO shows metallic bulk-like ρ-T behavior with low resistivity of ~3.2 × 10−4 ohm·cm at 300 K. In sharp contrast, the pre-edge peak is strongly suppressed in SNO, revealing the insulating behavior at 300 K. Moreover, the O K edge spectrum for VA-NNO exhibits notable changes compared with PA-NNO, especially in the Ni 3d band region. For VA-NNO, the pre-edge peak located at ~528 eV disappears and two small features appear at ~528.8 and ~ 530.1 eV (marked by two yellow arrows in Fig. 4A), which match well with those observed in LaNi2+O2.5 (44, 45). Even though Ni ions in both SNO with its mixed phases and VA-NNO are both Ni2+, their O K edge XAS spectra are quite distinct due to differences in band structure that also greatly affect conductivity (see Fig. 4B). Thus, the local coordination environment and associated electronic structure of Ni atoms in SNO with its mix of Sr2NiO3 and SrNi2O3 are different from that found in VA-NNO and LaNiO2.5, in which Ni atoms are in a square-planar structure with chains of NiO6 octahedra (Oh) and NiO4 square-planar (D4h) configurations (44, 46, 47).

Fig. 4 O K edge XAS and electrical transport of SNO.

(A) O K edge XAS for PA-NNO, SNO, and VA-NNO films measured at room temperature. (B) Temperature-dependent resistivity of these three samples. The purple arrow denotes the metal-insulator transition point (TMI).

Orbital polarization in SNO films

We also investigated the orbital polarization in the SNO films using XLD. The dx2 y2 and the d3z2 r2 orbitals can be probed independently by using IP and OOP x-ray polarization, respectively. The absorption spectra for these two polarizations are shown in Fig. 5A. The XLD (defined here as IEIPIEOOP) indicates the predominant occupation of the dx2 y2 IP orbital, suggesting the atomic orbital scheme shown in the inset in Fig. 5A. To quantify the orbital polarization (P), we first calculate the ratio of unoccupied states, r (48)r=h3z2r2hx2y2=3IEOOP4IEIPIEOOP(1)

Fig. 5 Orbital polarization in SNO films.

(A) Polarization-dependent XAS across the Ni L edge for SNO/LSAT(001). The inset shows a sketch of the engineered orbital and spin states, which forms the electronic configuration (Ni2+, 3d8) and orbital polarization (ndx2 y2> nd3z2 r2). (B) Schematic of the crystalline structures of Sr2NiO4 (SNO_214), pyramid Sr2NiO3 (SNO_213 pyramid), planar Sr2NiO3 (SNO_213 planar), and SrNi2O3 (SNO_123). (C) Partial density of states (PDOS) of these four different systems. The Fermi level (the black dashed line) is placed at 0 eV. The black arrows denote the spin state.

Here, hi is the number of holes in orbital i and Ij is the integrated L edge spectral intensity with polarization j. We obtain r = 1.12 by integrating the spectra over the range 850 to 875 eV. This hole ratio value is comparable to previously reported values of r = 1.19 for nickel oxide superlattices in tension and r = 1.05 resulting from spatial confinement in LaAlO3/LNO superlattices (48, 49). The r value can be used to evaluate the orbital polarization, P (50)P=(4neg1)1r1+r(2)where neg is the total number of eg electrons. We obtain a value of P = 5.7% for our SNO films if neg = 2, as expected for Ni2+ (3d8).

To better understand the XLD results in light of the coexistence of Sr2NiO3 and SrNi2O3 in SNO/LSAT, we conducted first-principles calculations (see Materials and Methods for details). We calculated the density of states (DOS) of four different structures: Sr2NiO4 (SNO_214), pyramidal Sr2NiO3 (SNO_213 pyramid), planar Sr2NiO3 (SNO_213 planar), and SrNi2O3 (SNO_123). The calculated minimum-energy structures are shown in Fig. 5B. The calculated c-axis lattice parameters of these four different structures are 12.24 Å for SNO_214, 12.86 Å for SNO_213 pyramid, 12.57 Å for SNO_213 planar, and 10.39 Å for SNO_123, respectively. However, experimental results (XRD and STEM) show that both phases have the same c-axis lattice constant (~12.6 Å). The Sr2NiO3 phase has a calculated lattice parameter closest to our measured value, whereas the SrNi2O3 phase has a much-reduced lattice parameter compared to the experimental value. Partial Sr intermixing within the Ni planes (Sr1+xNi2−xO3) can be invoked to maintain a 12.6-Å c lattice parameter and an epitaxial relationship between the two phases. Figure 5C shows the partial DOS (PDOS) results from which the orbital polarization in each structure can be obtained. For SNO_214 (Ni4+), there is no bandgap and the IP orbital has a larger bandwidth and a slightly lower average energy than the OOP orbital. For SNO_213 pyramid (Ni2+), the IP orbital has a higher energy than the OOP orbital and a large ∆eg (~0.7 eV) is observed. For SNO_213 planar (Ni2+), ∆eg = 0, and it is almost fully polarized with the OOP orbital unoccupied, in qualitative agreement with our XLD measurements. SNO_123 (Ni2+) shows a clear orbital polarization and the IP orbital is mainly unoccupied, which is the opposite of SNO_213 planar. Because of the phase segregation that occurs in our SNO films, the orbital polarization in SNO_123 will partially cancel out that in SNO_213. In addition, the phase purity of SrNi2O3 is not as high as that of the Sr2NiO3. Although it will not affect the net orbital polarization of our SNO films, which is dominated by the contribution from the SNO_213 phase, the orbital polarization P for our SNO film (P ≈ 5.7%) is smaller than that found in the square-planar phase R4Ni3O8 (R = La, Pr) obtained by hydrogen reduction (51), in which ~70% of the holes occupy the dx2 y2 IP orbital. This value is the highest published in the literature for Ni-based complex oxides (4850, 52). Although a large IP orbital occupation has been observed in our SNO film, it should be mentioned here that the infinite NiO2 plane in the SNO_213 planar structure is along the OOP direction, different from the IP infinite NiO2 planes shown in superconducting Nd0.8Sr0.2NiO2 films (11). Moreover, spectroscopy data demonstrate the 3d8 configuration in our insulating SNO films, while a predicted 3d9 configuration should exist in the superconducting Nd0.8Sr0.2NiO2 films (11).

In summary, our study demonstrates that the end member of phase pure SrNi4+O3 thin films in Nd1−xSrxNiO3 where x = 1 cannot be stabilized as a pseudo-cubic structure by epitaxial growth. However, suitable perovskite substrates can be effective templates for the nucleation and growth of structurally compatible oxygen-deficient RP phases, Sr2NiO3 and SrNi2O3. Their coexistence effectively satisfies an overall Sr/Ni ratio of 1:1, leading to an epitaxial c-axis–oriented nanocomposite. The dynamic phase segregation process that occurs during MBE growth of this material requires high cation diffusion rates and extensive vertical redistribution. Although the exact growth mechanism remains to be elucidated, our observations indicate that this system presents a fascinating testbed for an unusual kind of film nucleation that appears to be driven by the high stability of a particular cation valence, Ni2+. We envision that this kind of spontaneous phase segregation and RP faults may occur during the synthesis of Sr-doped RNiO3 where the Sr doping level is high. The presence of randomly distributed, mixed RP or reduced RP phases adds additional synthesis challenges in studying superconductivity but may open other avenues to explore the emergent phenomena in nickelates.


Sample growth

SNO thin films (~25 nm) were grown on (001)-oriented LSAT substrates by OPA-MBE. Sr and Ni were evaporated from the effusion cells, and the evaporation rates were calibrated using a quartz crystal oscillator. The Sr (Ni) fluxes were calibrated by growing SrTiO3 (LaNiO3) films on (001)-oriented SrTiO3 substrate, respectively. Once we obtain stable RHEED oscillations, the exact flux was extracted from the period (deposition time per unit cell) and used for SNO film growth. The substrate temperature was set to 700°C, and the activated oxygen partial pressure was kept at ~3 × 10−6 torr during growth. In situ RHEED was used to monitor the overall growth rate, and surface crystallography and structure. After growth, the sample was cooled down to room temperature at 30°C/min under activated oxygen. Two 12-nm-thick NdNiO3 films grown on (001)-oriented Nb-doped SrTiO3 substrate were annealed in oxygen plasma and vacuum, respectively, to use as reference samples for comparing the Ni XPS and XAS spectra with SNO. One 20-u.c.-thick NiO film grown on MgO (001) substrate was also used as reference sample.

XPS and XAS measurements

In situ XPS measurements were carried out on the as-grown samples. A low-energy electron flood gun was used to neutralize the charging effect. The TEY XAS measurements of the Ni L edge and O K edge were performed at the Singapore Synchrotron Light Source and the photoemission station at Beijing Synchrotron Radiation Facility of Institute of High Energy Physics, Chinese Academy of Sciences. All XPS and XAS were measured with the samples at 300 K. The x-ray incident angles 90° and 30° correspond to IP (Exy) and OOP (Ez) polarizations, respectively.

Sample characterization

A high-resolution x-ray diffractometer (Rigaku SmartLab) was used to measure the lattice structure. HAADF images were obtained by using a probe-corrected ARM200F (JEOL, Tokyo, Japan) STEM operated at 200 kV using a ~1-Å probe with a 27.5-mrad convergence semiangle and an 82-mrad collection inner angle. EELS mapping was performed using a probe-corrected GrandARM-300F (JEOL, Tokyo, Japan) STEM operating at 300 kV, using a ~1-Å probe with a 29-mrad convergence semiangle and a ~113-mrad spectrometer collection inner angle.

Conventional focused-ion beam milling techniques were applied to fabricate APT tips using a dual-beam scanning electron microscope (Thermo Fisher Scientific Helios Nanolab 600i) (53). A description of the preparation techniques, including protective coatings used, is provided in the Supplementary Materials. APT was performed using a CAMECA local electrode atom probe (LEAP) 4000 X-HR at a base temperature and pressure of 40 K and < 3 × 10−11 torr, respectively. The detection efficiency is nominally 36%. Field evaporation of ions from the specimen was induced by a 355-nm-wavelength picosecond laser at a pulse repetition rate of 125 kHz. A detection rate of 0.003 to 0.004 ions per pulse was maintained by varying the applied specimen voltage. The SNO film was captured in three of the four tips, using laser pulse energies between 60 and 160 pJ. One tip is described in detail in the text, representative of observations found across all samples. Data were reconstructed in 3D by using the integrated visualization and analysis software (IVAS 3.8.2) developed by CAMECA (Madison, WI). An average atomic volume of 0.06 nm3 atom−1 was specified in the 3D reconstruction, representing the atomic density of a SrNiO3 lattice [i.e., using the theoretically calculated density of 5.50 g cm−3 determined by Jain et al. (54) (SrNiO3 ID # mp-762506)]. The reconstructions were further optimized to reproduce flat SNO-LSAT interfaces, primarily by adjusting the image compression factor. A voxel size of 1.0 nm by 1.0 nm by 1.0 nm and delocalization of 3.0 nm by 3.0 nm by 1.5 nm were applied. Chemical visualization and analysis of the SNO film were enabled through assignment of Sr-, Ni-, and O-ionic species to the mass spectra (fig. S4), which yielded an average elemental composition of the SNO matrix of ~20.9% Ni, ~36.9% Sr, and ~42.2% O. IP transport measurements were performed in a van der Pauw geometry using a Hall measurement system in the temperature range 85 to 310 K.

DFT calculations

The DFT calculations are carried out using the Vienna ab initio simulation package (55). The atomic potentials are approximated by the projected augmented wave method (56), with Sr 4s24p65s2, Ni 3d94s1, and O 2s22p6 reference configurations. The GGA-PBEsol functional was used as an approximation to the exchange correlation (57). The DFT + U method with U = 6 eV was used to correct the underestimation of on-site Coulomb interaction between the Ni 3d electrons (58). The Kohn-Sham wave functions are represented by a plane wave basis set with an energy cutoff of 550 eV. The Brillouin zone is sampled with an 888 Monkhorst-Pack grid. The forces are reduced to less than 10−3 eV/Å when relaxing the structures.


Supplementary material for this article is available at

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.


Acknowledgments: We acknowledge facility support from W. R. Wiley Environmental Molecular Sciences Laboratory, a DOE User Facility sponsored by the Office of Biological and Environmental Research. We thank the Singapore Synchrotron Light Source (SSLS) and Beijing Synchrotron Radiation Facility for providing the facilities support. The SSLS is a National Research Infrastructure under the National Research Foundation, Singapore. Funding: The film growth and in situ XPS measurements are supported by U.S. Department of Energy (DOE), Office of Science, Office of Basic Energy Sciences, the Division of Materials Sciences and Engineering under award #10122. The XRD, STEM/EELS, and APT work is supported by the U.S. DOE, Office of Science, Early Career Research Program under award #68278. The first-principles calculations are supported by the EU H2020-NMBP-TO-IND-2018 project “INTERSECT” (grant no. 814487). Author contributions: L. Wang and Y.D. conceived this project. S.A.C. and Y.D. directed the project. L. Wang performed the epitaxial thin films synthesis with help from J.L. and L. Wangoh. L. Wang performed the in-plane transport measurements and the in situ XPS measurements. Z.Y. collected and analyzed the data of STEM measurements. S.D.T. and D.E.P. collected and analyzed the APT data. M.E.B. performed XRD measurements and analysis. X.Y., C.S.T., and A.T.S.W. collected and analyzed the XAS data. J.Z. and J.W. collected the Ni L edge XAS of the NiO thin film. X.H. performed the theoretical calculations. L. Wang, S.A.C., and Y.D. coauthored the manuscript. H.Z. contributed to this project by participating in the discussion of the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.

Stay Connected to Science Advances

Navigate This Article