Unconventional Pressure Dependent Interorbital and Interlayer Doping in Superconducting Nickelates

Understanding high temperature (high-$T_{c}$) cuprate superconductivityBednorz and Müller (1986) is one of the most important problems in condensed matter physics. Although the mechanism of superconductivity has remained elusive, research since 1986 has revealed many features of the cuprates and their superconductivity. Much of the understanding is related to phase diagrams, particularly the complex behavior of the electronic structure, transport, and superconductivity, as well as the emergence of competing phases as a function of doping, i.e., the electron counts in the Cu $d$-orbitals, especially the deviation from half-filling of the $d_{x^{2}-y^{2}}$ orbital. The recent discoveries of nickelate superconductivity provided the first example of a Cu-free material showing superconductivity with the key characteristics of the cupratesLi et al. (2019); Gu et al. (2020); Cheng et al. (2024); Zeng et al. (2022). The commonalities and differences from the cuprates hold the promise of providing crucial insights into the seemingly common superconductivity of these materials.

The initial discovery of nickelate superconductivity was in the doped infinite layer $R$NiO₂ ($R$=rare earth) systemLi et al. (2019). The nominal valence of Ni in the stoichiometric material is Ni⁺, with the same $d$-electron count as the Cu²⁺ in the high-$T_{c}$ parent La₂CuO₄. Similar to the cuprates, NdNiO₂ is not superconducting but is an insulator, while superconductivity emerges with doping by Sr. This would correspond to hole doping similar to the (La,Sr)₂CuO₄ system. This was followed by the discovery of superconductivity with $T_{c}=80$ K under pressure in the bilayer compound La₃Ni₂O₇ with nominal valence Ni^+2.5 and without chemical dopantsSun et al. (2023). This was understood in terms of accommodation of some holes in the Ni $d_{z^{2}}$ orbital in addition to the $d_{x^{2}-y^{2}}$ orbitalSun et al. (2023); Yang et al. (2024). This involvement of two distinct $d$ orbitals on the active Ni site is distinct from the self-doped cuprates such as YBa₂Cu₃O₇ Wu et al. (1987) and Bi₂Sr₂CaCu₂O₈ Subramanian et al. (1988), where the CuO₂ planes are doped by other layers, i.e., the chain-Cu layer in YBa₂Cu₃O₇ Pickett et al. (1992) and Bi-O and Tl-O bands in the Bi- and Tl-based cupratesKrakauer and Pickett (1988); Singh and Pickett (1992); Hussey et al. (2003). The consequence is that more complex superconducting states other than the $d$-wave $x^{2}-y^{2}$ order of the cuprates become possibleYang et al. (2023). Finally, superconductivity has also been reported in the three-layer compound La₄Ni₃O₁₀ under pressure without chemical dopingLi et al. (2024a, b); Zhang et al. (a); Zhu et al. (2024); Nagata et al. , although no superconductivity was found in polycrystalline samplesZhang et al. (2024a). Nonetheless, experiments show strong dependence of the superconducting behavior on the precise O stoichiometry indicating the importance of electron countNagata et al. . From a structural point of view, this compound can be considered intermediate between the bilayer La₃Ni₂O₇ and the infinite layer LaNiO₂. The nominal Ni valence in La₄Ni₃O₁₀ is also different, Ni^+2.67. The compound also differs from the other two in that it has two different Ni sheets. This provides an additional degree of freedom that may help unravel the puzzle of doping of the superconductors. These experimental findings have motivated several theoretical proposals for the superconductivity LaBollita et al. ; Zhang et al. (b); Qin et al. ; Huang and Zhou ; Zhang et al. (c, d); Chen et al. ; Oh et al. ; Yang et al. ; Tian et al. (2024); Leonov (2024); Sakakibara et al. (2024).

Here we use an analysis of pressure dependent first principles calculations and find an unconventional pressure dependent doping of this material. This includes an interlayer component with transfer of carriers between the outer and inner layer Ni and an interorbital component involving the interplay of the $d_{x^{2}-y^{2}}$ and $d_{z^{2}}$ orbitals.

La₃Ni₂O₇Wang et al. (2024) maintains $Amam$ symmetry under ambient conditions. Upon cooling, this compound transitions to $Fmmm$ symmetry and, below 40 K, takes $I4/mmm$ symmetry when the pressure exceeds 19 GPa. La₄Ni₃O₁₀ has $P2_{1}/a$ symmetry under normal conditions. However, applying pressure greater than 12.6 to 13.4 GPa, even at ambient temperatures, induces a transition to $I4/mmm$ symmetry. Both La₃Ni₂O₇ and La₄Ni₃O₁₀ have $I4/mmm$ symmetry in the superconducting phase. La₄Ni₃O₁₀ features three Ni-O layers, whereas La₃Ni₂O₇ has two. $I4/mmm$ La₄Ni₃O₁₀ contains two nonequivalent nickel atoms, Ni1 and Ni2, as illustrated in Fig.1(a). In contrast, La₃Ni₂O₇ consists of only one type of nickel atom, as shown in Fig.1(b). Ni1 and Ni2 have distinct properties, with interlayer doping. It is to be noted that the La-O layers between the NiO layers have net positive charge based on the valence.

We study electronic properties of high-pressure structures from 15.3 GPa to 44.3 GPa reported in Ref.Li et al. (2024a). We used the efficient APW+lo method Sjostedt et al. (2000) as implemented in the WIEN2k packageBlaha et al. (2020), and the Perdew-Burke-Ernzerhof (PBE) generalized gradient approximation (GGA). We used $30\times 30\times 30$ k-point meshes were used to sample the Brillouin zone during the self-consistent iterations. The muffin-tin radii in Bohr were Ni:1.82,La:2.2,O:1.56 and the basis set cutoff parameter was $R_{min}K_{max}$=7. Both muffin-tin radii and $R_{min}K_{max}$ were fixed in all calculations in order to facilitate comparisons.

The octahedral crystal field leads to $e_{g}$ orbitals near the Fermi level in these compounds, specifically the $d_{x^{2}-y^{2}}$ and $d_{z^{2}}$ orbitals, which are hybridized with in-plane and apical O $p$ orbitals, respectively. In Fig.2, we show La₄Ni₃O₁₀ at 30.5 GPa as an example. We show orbital resolved bands and Fermi surfaces for Ni1 $d_{z^{2}}$, Ni2 $d_{z^{2}}$, Ni1 $d_{x^{2}-y^{2}}$ and Ni2 $d_{x^{2}-y^{2}}$. The high symmetry points of BZ are labeled in Fig.2(e). The Fermi surfaces are labeled in Fig.2(f) as $\gamma,\gamma^{\prime},\alpha,\beta,\beta^{\prime}$

In Fig.2(b), we applied nonzero Hubbard $U$=6 eV and Hund $J$=1 eV compared to Fig.2(a). Surprisingly, the $d_{z^{2}}$ and $d_{x^{2}-y^{2}}$ bands are little changed in terms of the band width or orbital occupancy with applied $U$ and $J$. The main effect of $U$ and $J$ is pushing $t_{2g}$ bands significantly downwards to higher binding energy. Here we emphasize results that do not depend on the choice of $U$. In Fig.2(c) and (d), we plot bands under different pressure together and focus on the range $E_{F}$-0.5 eV to $E_{F}$+0.5 eV. We can see that pressure has a significant impact on the $\gamma^{\prime}$ pocket; as the pressure increases, the $\gamma^{\prime}$ pocket becomes larger. In addition, $U$ and $J$ mainly affect the $\gamma^{\prime}$ pocket, especially with relatively low pressures of 15.3 GPa and 19.2 GPa, where applying $U$ and $J$ suppresses the $\gamma^{\prime}$ pocket. Under higher pressures, $U$ and $J$ also shrink the $\gamma^{\prime}$ pocket. The other Fermi pockets, however, are insensitive to both pressure and $U$, $J$. Since the influences of $U$, $J$, and pressure on the energy bands near the Fermi level are minimal, our following discussion is based on the bands where $U$ and $J$ are zero, and the pressure is 30.5 GPa.

From Fig.2(a), we can see clear separation of $d_{z^{2}}$ and $d_{x^{2}-y^{2}}$ on $\Gamma-M$ and $\Gamma-X$. $\Gamma-M$ is in $kz$ direction, and $d_{x^{2}-y^{2}}$ is flat along this direction, but $d_{z^{2}}$ has dispersion. The $\gamma$ pocket at $\Gamma$ point is from Ni2 $d_{z^{2}}$ without any Ni1 component as shown in Fig.2(a)(g) on the whole $\Gamma-M$ and $\Gamma-X$ path. As seen from Fig.S1(d), the $p_{z}$ orbitals of both the O2 and O4 contribute to this band. A similar situation is found on $\gamma^{\prime}$ pocket at X point. The $\gamma^{\prime}$ pocket contributed by $d_{z^{2}}$ of both Ni1 and Ni2, with Ni1 contributing slightly more. However, considering that there are actually two Ni2 atoms, overall, Ni2 contributes more. However, as in Fig.S1(d), there is no contribution from O2’s $p_{z}$ orbital at X point. The $\beta^{\prime}$ pocket is contributed by Ni2’s $d_{x^{2}-y^{2}}$ and $d_{z^{2}}$, mainly $d_{x^{2}-y^{2}}$. However, $\beta$ pocket is a mix of Ni2’s $d_{x^{2}-y^{2}}$ and Ni1’s $d_{z^{2}}$ and $d_{x^{2}-y^{2}}$ The $\alpha$ pocket is contributed by all four orbitals, but most weight concentrated on $\Gamma-X$ direction by Ni1’s $d_{x^{2}-y^{2}}$

We present plots of electron occupancy vs pressure for different Fermi surfaces of La₄Ni₃O₁₀ in Fig.3. As pressure increases, the $\gamma^{\prime}$ band exhibits a noticeable decrease in occupancy, indicating electron depletion, due to electrons being transferred to other bands. Conversely, the $\beta^{\prime}$ and $\beta$ bands show substantial increases in occupancy, suggesting that these bands are being significantly electron-doped under pressure. This indicates a pronounced transfer of electrons into these bands from the $\gamma^{\prime}$ band. The $\alpha$ and $\gamma$ bands exhibit slight increases in occupancy with increasing pressure, although these changes are less pronounced compared to the $\beta^{\prime}$ and $\beta$ bands. This suggests a minor self-doping effect in the $\alpha$ and $\gamma$ bands. Overall, the pressure-induced self-doping effect in La₄Ni₃O₁₀ results in a redistribution of electrons among the Fermi surface sheets, with the $\beta^{\prime}$ and $\beta$ bands having increased electron occupancy coming from the $\gamma^{\prime}$ band.

We emphasize the similarity between infinite layer NdNiO₂ doped with strontium and cuprates, for example, their Fermi surfaces are very similarSakakibara et al. (2020). Magnetic modes observed by resonant inelastic x-ray scattering are also similar to those of the typical doped Mott insulator in cupratesLu et al. (2021). However, in La₃Ni₂O₇ Ni has valence Ni^2.5+, quite different from cuprates. Nonetheless, recent experimentsZhang et al. (2024b) show superconductivity similar to cuprates and features such as linear temperature-dependent resistivity. We emphasize with the different Ni the electron count is very different from cuprates but our calculations show the $\beta$ and $\beta^{\prime}$ Fermi surfaces in La₄Ni₃O₁₀ remain similar shape to those in cuprates. However, the differences are also significant. Both cuprates and doped NdNiO₂ have main Fermi surface sheets primarily from the $d_{x^{2}-y^{2}}$ orbitals hybridized with in-plane O_p orbitals. The $\beta$ and $\beta^{\prime}$ Fermi surfaces, also dominantly from $d_{x^{2}-y^{2}}$ orbitals, with hybridization by both in-plane O, and also the $d_{z^{2}}$ and apical O are analogous both in shape and orbital composition to the superconducting sheets in the cuprates and doped NdNiO₂. The $\beta^{\prime}$ sheet has occupancy close to half-filling while $\beta$ sheet is doped away from half-filling by approximately $0.09\sim 0.10$ holes, depending on pressure. Both $\beta$ and $\beta^{\prime}$ have Ni2 $d_{x^{2}-y^{2}}$ character, and in the case of $\beta$ there is additionally Ni1 $d_{x^{2}-y^{2}}$ character almost twice that of Ni2 $d_{x^{2}-y^{2}}$ per atom. The strongest effect of pressure is an increase in electron count towards half-filling with pressure for the $\beta$ sheet.

We now turn to the connections between structure and the doping of the sheets. In Fig.4(d), we plot variations of various Ni-O bonds vary with pressure. In Fig.4(a)(b)(c), we plot 1$s$ core level relative to $E_{F}$ for oxygen and nickel sites and their differences vs pressure. As expected, all Ni-O bond lengths decrease with pressure. The planar bond Ni1-O1 and Ni2-O3 are the shortest. The next longer bonds are Ni2-O2 and Ni1-O2 in which O2 bridges Ni1 and Ni2. Finally, the longest bond is Ni2-O4, which is the outer apical O bond of the nickel trilayer. Ni2 octahedra has Ni2-O2 and Ni2-O4 with different lengths. The 1$s$ core level of Ni2 is lower than Ni1 indicating their different environment. Both the Ni1 and Ni2 core levels decreases with increasing pressure. Interestingly, despite Ni2-O4 being the longest bond, under low pressure, both the 1$s$ core level of O4 and the core level difference between Ni2 and O4 are at intermediate values. As the pressure increases to the maximum, the 1$s$ core level of O4 decreases faster than those of other O atoms and reaches the lowest value, while the core level difference between Ni2 and O4 goes up above all other core level differences under pressure.

Thus, the Ni2-O4 bond, which represents the behavior of the outer layer apical O is distinct. The downward shift indicates a stabilization of the ionic nature of this outer apical O with pressure. It is to be noted that the crystal field split $e_{g}$ orbitals are antibonding states. Therefore, the consequence of the shortening of the softer Ni2-O4 bond relative to the others is expected to push bands associated with the Ni2 $d_{z^{2}}$ orbital to higher energy. Hence, the reduced occupation under pressure of the $\gamma^{\prime}$ Fermi surface.

Therefore, under pressure electrons are transferred from the outer-layer Ni2 $d_{z^{2}}$ of $\gamma^{\prime}$ Fermi surface to the $\beta$ and $\beta^{\prime}$ sheets, which have mixed Ni1 and Ni2 $d_{x^{2}-y^{2}}$ character. This represents an unconventional interlayer doping, with transfer from a Fermi surface originating on the outer layer, to Fermi surfaces from both the inner and outer layer. The pressure dependent occupations are particularly interesting in relation to a scenario where the $\beta$ and $\beta^{\prime}$, $d_{x^{2}-y^{2}}$ sheets are the active sheets for superconductivity.

Firstly, it may be noted that there are only two such sheets, while there are three Ni-O layers. This is a consequence of the band formation, which pushes one of the $d_{x^{2}-y^{2}}$ sheets to higher energy as shown in Fig. 2. This, plus the presence of other sheets of Fermi surface, representing partially occupied bands, accounts for the valence difference from the Ni¹⁺ of NdNiO₂, while still maintaining the cuprate-like structure of the $\beta$ and $\beta^{\prime}$ surfaces. Secondly, the $\beta^{\prime}$ sheet is near half filling, and becomes increasingly close to half filling with pressure. In cuprates, half-filling is associated with a Mott insulating state, incompatible with superconductivity. Doping away from half-filling has at least two effects. The first is destruction of the Mott insulating state in favor of a conducting state, with Fermi surfaces, compatible with superconductivity.

The second is suppression of antiferromagnetism with reduction of the corresponding spin-fluctuations as the hole doping is increased, particularly towards the over-doped region of the phase diagram. In present case, due to the multi-orbital nature of the bands and the multiple bands crossing $E_{F}$ there is no half-filled orbital that would favor a Mott insulating state consistent with the fact that La₄Ni₃O₁₀ is not insulating. However, the cuprate-like near half-filled $\beta^{\prime}$ may lead to nearness to antiferromagnetism, with the doping level of this band being a key parameter. This suggests studies probing the proximity to magnetism of this compound both in terms of the O stoichiometry and pressure, especially in relation to the superconducting properties.

Thus, in spite of the very different electron counts the superconductivity of the different nickelate superconductors can be unified in terms of the Fermi surface structure, which universally shows similar $d_{x^{2}-y^{2}}$ sheets characteristic also of the cuprate high-Tc materials. This provides a framework that enables similar superconductivity in cuprates and nickelates. In addition, by avoiding the Ni¹⁺ valence state and the chemical instability of this state, but maintaining an electronic structure favorable for superconductivity materials like La₃Ni₂O₇ and La₄Ni₃O₁₀ point to the possible existence of many more layered superconducting nickelates that remain to be discovered.

Y. N. Huang thanks W. C. Bao and H. Q. Lin for helpful discussions. Y. N. Huang is supported by the National Natural Science Foundation of China (Grant No. 11904319).

Supplemental Materials: Unconventional Pressure Dependent Interorbital and Interlayer Doping in Superconducting Nickelates