p-type transparent superconductivity in a layered oxide

See allHide authors and affiliations

Science Advances  15 Jul 2020:
Vol. 6, no. 29, eabb8570
DOI: 10.1126/sciadv.abb8570


Development of p-type transparent conducting materials has been a challenging issue. The known p-type transparent conductors unsatisfy both of high transparency and high conductivity nor exhibit superconductivity. Here, we report on epitaxial synthesis, excellent p-type transparent conductivity, and two-dimensional superconductivity of Li1−xNbO2. The LiNbO2 epitaxial films with NbO2 sheets parallel to (111) plane of cubic MgAl2O4 substrates were stabilized by heating amorphous films. The hole doping associated with Li+ ion deintercalation triggered superconductivity below 4.2 kelvin. Optical measurements revealed that the averaged transmittance to the visible light of ~100-nanometer-thick Li1−xNbO2 was ~77%, despite the large number of hole carriers exceeding 1022 per cubic centimeter. These results indicate that Li1−xNbO2 is a previously unknown p-type transparent superconductor, in which strongly correlated electrons at the largely isolated Nb 4dz2 band play an important role for the high transparency.


Search for new transparent conductors (TCs) is one of the most important subjects not only for practical use but also for advances in materials science (1). In case of n-type TCs, a variety of materials have been developed and used in industry. For example, Sn-doped In2O3 (ITO) exhibits high conductivity (~6000 S cm−1) and high transparency (~80%) (2). Furthermore, even a transparent superconductor (TSC), which is regarded as an ultimate TC, was reported for LiTi2O4 epitaxial films (3). In contrast to n-type TCs, however, p-type TCs (p-TCs) are subjects of fundamental research since their performances are still low (4).

Zhang et al. (5) recently proposed new strategy for designing TCs based on transition-metal oxides (TMOs). In contrast to a traditional concept (taking ITO as an example, high-mobility semiconductor In2O3 is doped with a small amount of Sn), they shed light on correlated metals, where large number of carriers resides in low-mobility TMOs. On the basis of this strategy, perovskite-type AMO3 (A = Ca, Sr, and La; M = V, Mo, and Cr) have been found to be good TCs, both n- and p-type (58). However, the studies have been so far limited to the perovskite-type TMOs. Thus, further investigation on other TMOs will pave a way for better p-TC performances.

To realize an excellent p-TC and p-type TSC (p-TSC), one may investigate two-dimensional materials, which have been enthusiastically studied in these days (9). Among various TMOs, we focus on a layered niobate, which is the only isostructural oxide to 2H-type transition-metal dichalcogenide (TMD) such as MoS2 (10). The crystal structure of LiNbO2 consists of NbO6 triangular prisms (Fig. 1A), which are thought to create characteristic d-band splitting for realizing p-type conduction (11, 12). Similar to TMDs, exotic properties can be expected (13, 14). In addition, the wide bandgap nature of TMOs provides high transparency in the visible range (15). However, electronic properties of LiNbO2 are still unclear. Although superconductivity was observed below 5 K in a Li-deficient phase Li1−xNbO2 (x ~ 0.5) (16), its electronic properties have not been investigated in detail (17). Moreover, there is no report about single-crystalline superconducting films, and their transport and optical properties are not yet fully understood.

Fig. 1 Three-step synthesis method and structural analysis for Li1−xNbO2 epitaxial films.

(A) Schematic crystal structure of LiNbO2. The green, blue, and red spheres indicate lithium, niobium, and oxygen atoms, respectively. The magnified unit of NbO6 takes a triangular prism structure. (B) Schematic illustrations of three-step synthesis method. The experimental conditions and produced compounds at each step are shown in each box. (C) Out-of-plane x-ray diffraction (XRD) profiles for the direct film and the films at each step. Filled circles indicate reflections coming from the MgAl2O4 substrates. Filled and open triangles indicate reflections coming from a sample stage and secondary phases, respectively. arb. units., arbitrary units. (D) XRD ϕ scans of reflections for MgAl2O4 440 in the substrate, LiNbO2 103 in the step 2 film, and Li1−xNbO2 103 in the step 3 film.

Here, we fabricated superconducting Li1−xNbO2 epitaxial films using three-step synthesis method as illustrated in Fig. 1B and investigated their electronic properties. p-type doping associated with Li+ ion deintercalation to LiNbO2 triggered superconductivity and enhanced transparency in the visible range, and we created a p-TSC with strongly correlated electrons in two-dimensional NbO2 conduction layers and isolated Nb 4dz2 band derived from the NbO6 triangular prism coordination.


Three-step synthesis and structural properties

Figure 1C shows out-of-plane x-ray diffraction (XRD) profiles of Li-Nb-O films. The films prepared by the high-temperature direct growth (direct) showed 002 reflection of layered LiNbO2 at 2θ = 17.0° and a number of reflections from secondary phases, Li3NbO4 222, NbO2 440, and LiNbO3 006 at 2θ = 36.8°, 37.3°, and 38.9°, respectively (18, 19). The pulsed laser deposition (PLD) growth of Li-containing materials such as LiCoO2 and LiTi2O4 often tends to form Li-deficient phases because Li atoms are easily scattered by gaseous species and/or reevaporated from the growing surface (20, 21). In our case, however, the Li-rich phase (Li3NbO4) indicated much stronger peak than the Li-deficient phase (NbO2) and stoichiometric phases (LiNbO2 and LiNbO3). The inclusion and unusual stability of Li3NbO4 were further verified by its excellent crystallinity irrespective of growth conditions (see section S1).

To avoid the formation of the secondary phases and to create layered LiNbO2, we developed a three-step synthesis, which was initiated from PLD growth of Li-Nb-O films on the unheated substrates (step 1). Out-of-plane XRD profiles of the films at each step are also shown in Fig. 1C. In step 1, no film peak was observed, suggesting the amorphous phase. After in situ reductive annealing (step 2), clear 00l reflections of LiNbO2 phase appeared without Li3NbO4 peaks despite the condition identical to the direct growth. A small peak of the oxidized phase (LiNbO3) is thought to originate from partial oxidation of the LiNbO2 phase. LiNbO3 has wide bandgap and insulating nature and does not affect the transport and optical properties described below. Last, the LiNbO2 films were immersed into an I2/CH3CN solution to deintercalate Li+ ions (step 3). This solution process is known to take place with the following redox reaction (22)Anodic reaction:LiNbO2Li1xNbO2+xLi++xe(1)Cathodic reaction:I2+2e2I(2)

After the Li+ ion deintercalation, only small shift of LiNbO2 00l peaks was observed, suggesting that NbO2 layers in a host structure were preserved. Note that the crystallinity of the films (full width at half maximum of ω-scan profiles for LiNbO2 002 reflections) was unchanged.

We also measured asymmetric reflections (Fig. 1D). For both step 2 and step 3 films, the LiNbO2 103 reflection was clearly observed with a sixfold rotational symmetry, demonstrating solid-phase epitaxy at hexagonal planes (10). The angles of 440 reflections of the substrate coincided with those of the film, indicating an epitaxial relationship of LiNbO2 [100] || MgAl2O4 <11¯0>. The lattice parameters for step 2 (step 3) films were calculated to be a = 2.906 (2.914) Å and c = 10.446 (10.468) Å. Bulk values of Li0.93NbO2 (Li0.45NbO2) were reported to be a = 2.906 (2.923) Å and c = 10.447 (10.455) Å (16), suggesting that the step 2 films had nearly stoichiometric composition, while the step 3 films had Li+ ion deficiency. In addition, the in-plane lattice parameters at each step are considerably larger than that of MgAl2O4 (d22¯0 = 2.858 Å), indicating unstrained film lattices.

Transport properties and strongly correlated electrons

The direct and step 1 films indicated resistivity (ρ) exceeding a measurement limit. Figure 2A shows the temperature dependence of ρ for the films at the following steps. The step 2 films showed ρ less than 10 milliohm·cm at 300 K and metallic conduction (dρ/dT > 0). The step 2 films showed positive Hall coefficient, and hole concentration (n) was estimated to be 2.4 × 1021 cm−3 at 300 K (see section S2). On the other hand, decrease in ρ and change of the film color were found over the course of days, suggesting that oxidation upon exposure to air led to hole doping (23). To avoid any reaction in air, we prepared a capping layer (several-nanometer-thick amorphous alumina) before removing from the chamber (capped step 2). The capped step 2 films showed insulating behavior (dρ/dT < 0) with higher ρ (50 milliohm·cm at 300 K) and lower n (6.4 × 1020 cm−3 at 300 K). Therefore, it is concluded that the metallic conduction in the step 2 films arises from easily oxidized (hole dopable) nature of LiNbO2. It should be emphasized that the electronic ground state of LiNbO2 has been controversial (17). Our results strongly suggest that LiNbO2 is an insulator and supports a theoretical study (24).

Fig. 2 Transport properties and Fermi-liquid analysis.

(A) Temperature dependence of resistivity for the films at each step. The labeled values are hole concentration at 300 K (n) obtained by Hall measurements. Inset: Superconducting properties of the step 3 films. Magnification of that is indicated in the main panel, and its magnetic field (along the c axis) dependence was taken up to 0.9 T. (B) T and (C) T2 dependence of ρ normalized by the values at 300 K for the films at each step. Dashed lines are linear fits to the plots.

The step 3 films showed much lower ρ (3 milliohm·cm at 300 K), higher n (2.5 × 1022 cm−3 at 300 K), and a superconductivity below 4.2 K, demonstrating superconducting transition in Li1−xNbO2 triggered by hole doping associated with the Li+ ion deintercalation. Inset of Fig. 2A shows the magnification in the low-temperature range and modulation of superconductivity under magnetic field applied parallel to the c axis of Li1−xNbO2. The superconductivity was completely suppressed under a magnetic field as low as 0.9 T, which supports two-dimensional hole superconductivity in the NbO2 plane (25). The critical magnetic field is much lower for a polycrystal (26). Moreover, superconductivity remained up to 4.2 T, when magnetic field applied perpendicular to the c axis, details of which will be published elsewhere. Further investigation on the two-dimensional hole superconductivity in our ideal samples is important, especially in connection with two-dimensional superconductors such as TMD and high–critical temperature (TC) cuprate superconductors.

To elucidate how electron correlation works in Li1 − xNbO2, we replot Fig. 2A for Fig. 2 (B and C), where the normalized resistivity (ρ/ρ300K) is plotted as a function of T (Fig. 2B) and T2 (Fig. 2C). A comparison between these plots allows us to elucidate a Fermi-liquid picture in the transport properties. We clearly demonstrate enhanced strength of electron correlation and non-Fermi liquid ground states in the superconducting Li1−xNbO2 films. We observed the linear T2 dependence of ρ for the metallic step 2 and step 3 films, indicating Fermi-liquid ground states. However, their temperature ranges are different from one to another. The step 2 films showed good linearity lasting up to 300 K. In contrast, the step 3 films showed deviation from ~100 K and, instead, followed T linear dependence. This characteristic temperature, referred to as T*, is regarded as an important phase point and/or a boundary in electronic phase diagrams of well-known cuprate, iron-pnictide superconductors, and heavy-fermion compounds (27, 28). We find that, in our Li1−xNbO2, T* showed a monotonic decrease by increasing the hole-doping level similar to well-established quantum critical systems, which we wish to publish elsewhere.

We performed fitting of T2 plots to Fermi-liquid equation, i.e., ρ = ρ0 + AT2. We obtained A = 3.3 × 10−2 and 5.4 × 10−2 microhm·cm K−2 for step 2 and step 3 films, respectively. The A parameters are directly associated with effective mass of carrier m* (Am*2). Conventional metals and metal-oxide conductors exhibit A ~ 10−6 microhm·cm K−2 (29). On the other hand, strongly correlated metal oxides in the vicinity of Mott-Hubbard transitions (30) and heavy-fermion systems (29) show A in a range of 10−3 to 10 microhm·cm K−2. The values obtained for our Li1−xNbO2 films are comparable to or larger than those of the latter, reflecting strong electron correlations. Apart from a class of metal oxides, electron correlations in Li1−xNbO2 are comparable to that in superconducting Cu0.07TiSe2, an intercalated TMDs (1.1 × 10−2 microhm·cm K−2) (31). Furthermore, the roles of electron correlation in electronic structures of Li1−xNbO2 are also expected theoretically (32).

Optical properties and strongly isolated band structure

We have observed marked color changes throughout the three-step synthesis as shown in Fig. 3A. The direct and step 1 films indicated dark color typical of doped TMOs. In contrast, the step 2 films indicated characteristic bloody red, which is consistent with the previous results (27). Furthermore, the step 3 films indicated transparent olive. Figure 3B shows optical transmittance in infrared to visible range for the films at each step and a pristine substrate. The first characteristic coloration in the step 2 films originated from drop of transmittance below 600 nm, which was consistent with calculated bandgap of layered LiNbO2 (~2 eV) (24, 33, 34). The second color bleaching in the step 3 films resulted in increasing transmittance in the visible range. Unexpectedly, the average transmittance (Tave) in the visible range (1.5 to 3.0 eV) reached ~50% despite a heavily hole-doped medium. Note that this value is the external transmittance, where substrate and light reflection at the surface and interface cause intensity loss. The Tave after subtracting these contributions was as high as ~77%.

Fig. 3 Sample photographs and optical properties.

(A) Optical images of the films at each step. Substrate size is about 5 mm by 5 mm. The transparent regions at the diagonal corners are trace of clamps that were used to hold the substrate to a metal backplate. (B) Optical transmittance spectra for the films at each step and MgAl2O4 substrate.

Taking difference between internal absorption spectra of LiNbO2 and Li1−xNbO2 films into account, notable effects of the Li+ ion deintercalation (hole doping) can be clearly captured (see section S3). The near–band edge absorption (2 to 2.5 eV) is greatly suppressed upon hole doping. In addition, the near-infrared absorption is hardly enhanced, although n increases by an order magnitude. The origin of this behavior will be explained later.

Performance as a p-TC

The p-TC performance of Li1−xNbO2 is summarized and compared with other p-TCs in Fig. 4, where their conductivities (σ) at room temperature (RT) are plotted as a function of Tave. In general, σ and Tave tend to indicate trade-off relationship as recently demonstrated in high-performance p-TCs of La1−xSrxCrO3 (LSCO) films. In this system, increase in x (number of holes) leads to increase in σ and decrease in Tave (8). The lightly doped metallic LiNbO2 (step 2) marks almost on the LSCO line. In contrast, the heavily doped superconducting Li1−xNbO2 marks beyond the border, highlighted by gradient background, and even violates the general tendency. Given the record large figure of merit as a p-TC and superconductivity, Li1−xNbO2 can be regarded the first p-TSC.

Fig. 4 Performances as a p-TC.

Graphical representation of conductivity and transmittance for the LiNbO2 (step 2) and Li1−xNbO2 (step 3) epitaxial films and reported p-TCs. Some of p-TCs are merely labeled, but all details including other p-TCs are listed in section S5. n-type ITO was also plotted as the ultimate performance for the purpose of exploring p-TCs. A series of LSCO, which is regarded as the guidance of TMO-based p-TC, was highlighted by a dashed arrow (8).


Let us explain the origin of the anomalous transparency of Li1−xNbO2. The degree of transparency in near-ultraviolet to near-infrared range can generally be divided into lower (near-infrared) and higher (near-ultraviolet) photon energy regime, which must be maintained against (i) free-carrier absorption, also known as a plasma reflection, and (ii) interband optical absorption, respectively (5). In conventional TCs, the higher σ becomes, the lower Tave becomes due to (i). As for TMO-based TCs, Tave can relatively be maintained against (i) due to large effective mass m* of quasiparticle carriers. In contrast, Li1−xNbO2 rather harvests Tave in terms of (ii) as will be explained as follows (schematically illustrated in Fig. 5).

Fig. 5 Origins of p-type transparent superconductivity in Li1−xNbO2.

Transmittance spectra in an extended range of photon energy and energy diagram of p-type transparent superconductivity in the Li1−xNbO2 epitaxial films. Strongly correlated electrons in Li1−xNbO2 decrease plasma energy. On the other hand, the strongly isolated Nb 4dz2 band located at EF increases band absorption energy in each process. As a result, “transparent window” of Li1−xNbO2 becomes wider as almost comparable to visible range.

Regarding (i), strongly correlated electrons decreased plasma reflection. Plasma energy (hωp) is associated with ratio of n to m* (5). Our results allowed us to estimate m* numerically, taking previously reported parameters into account (24). Suppose hωp ~ 1 eV for the step 3 films from decrease of transmittance in near-infrared region (Fig. 5), we obtain m* ~ 100 m0 (see section S4). On the other hand, hωp ~ 0.5 eV was estimated for the step 2 films. The heavy mass is consistent with large A (3.3 × 10−2 and 5.4 × 10−2 microhm·cm K−2 for step 2 and step 3 films, respectively) described above. Li1−xNbO2 turned out to form strongly correlated electronic states, resulting in the lower hωp even when n exceeds 1022 cm−3. Upon hole doping into LiNbO2, hωp did not shift into the visible range. Thus, further carrier doping is acceptable to enhance the performance of p-TSC.

Regarding (ii), strongly isolated Nb 4dz2 band decreased interband optical absorption. According to our results and the reported band calculations, the Fermi level (EF) was thought to locate in the band on Nb 4dz2 orbital, which was strongly isolated from the upper (degenerated Nb dxy and dx2−y2 orbitals) and the lower (O 2p orbitals) bands. Note that highly isolated single band has a potential to create high density of states at EF and strong electron correlation, both of which are favorable to superconductivity (34). The characteristic layered NbO6 triangular prism structure derives this prominent nature (24) and leads to the anomalous transparency. According to Fig. 5, bandgap becomes wider as EF drops with increasing n. This effect known as Burstein-Moss effect extends a short wavelength cutoff to blue range (35). Combining these two effects in Li1−xNbO2 realizes appropriate transparent window ranging from 1.0 to 2.4 eV. One expects to expand this window to violet upon further deintercalation of Li+ ions. Synergetic increase in σ allows Li1−xNbO2 to approach to ITO (Fig. 4). The transmittance can also be improved by reducing film thickness.

In summary, we have first grown two-dimensional Li1−xNbO2 epitaxial films using specialized three-step synthesis method. The hole doping associating the Li+ ion deintercalation substantially enhances visible-light transparency and leads to superconductivity. We conclude that strongly correlated electrons in the isolated Nb 4dz2 band realize the p-type transparent superconductivity in Li1−xNbO2. Our study demonstrates a new approach to creating high-performance TCs.


Sample fabrication

Li-Nb-O films were grown on MgAl2O4 (111) substrates using PLD method with KrF excimer laser (1.0 J cm−2). A Li excess ceramic (Li1.2NbO2+δ) was prepared by conventional solid-state reaction steps, starting from mixing Li2CO3 and Nb2O5 powders with a molar ratio of 1.2:1. The direct growth was conducted at substrate temperature (Ts) of 800°C under a chamber pressure (PAr/H2) of 0.1 mtorr set by continues flow of Ar/H2 (1 volume percent H2) gas. The three-step synthesis method is illustrated in Fig. 1B. In step 1, using Li1.2NbO2+δ target, amorphous Li-Nb-O films were deposited in vacuum (background pressure, 1.0 × 10−7 torr) at RT. In step 2, Ar/H2 gas was fed into the chamber to set PAr/H2 = 0.1 mtorr, and the step 1 films were annealed in situ at Ts = 800°C for 1 hour. In step 3, the step 2 films were exposed to air and immersed in 0.02 M I2/CH3CN solution for 15 min. Some films were capped by several-nanometer-thick alumina films using PLD at RT in vacuum for avoiding reactions with air (capped step 2).


Using a stylus profiler, film thickness was regulated to be ~100 nm. The crystal structures and epitaxial relationship were investigated by a laboratory XRD apparatus with Cu Kα1 radiation. The temperature dependence of resistivity and the Hall voltage were measured by a standard four-probe method using a physical property measurement system (Quantum Design). The optical properties were investigated by ultraviolet-visible near-infrared spectroscopy at RT.


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: Funding: This work was partly supported by MEXT Elements Strategy Initiative to Form Core Research Center (grant no. JPMXP0112101001) and a Grant-in-Aid for Scientific Research (nos. 18H03925, 18 J10171, 19H02588, and 20K15169) from the Japan Society for the Promotion of Science Foundation. T.S. acknowledges the financial support from JSPS Research Fellowship for Young Scientists. Author contributions: T.S. performed the experiments and analyzed the experimental data. All authors discussed the results and wrote the manuscript. A.O. directed the project. 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