Efficient Stabilization of Na Storage Reversibility by Ti Integration into O′3-Type NaMnO2

Department of Chemistry and Life Science, Yokohama National University, 79-5 Tokiwadai, Hodogaya-ku, Yokohama, Kanagawa 240-8501, Japan Department of Applied Chemistry, Tokyo Denki University, Adachi, Tokyo 120-8551, Japan Elements Strategy Initiative for Catalysts and Batteries, Kyoto University, f1-30 Goryo-Ohara, Nishikyo-ku, Kyoto 615-8245, Japan Institute of Materials Structure Science, High Energy Accelerator Research Organization, Tsukuba, Ibaraki 305-0801, Japan Department of Materials Structure Science, Sokendai (The Graduate University for Advanced Studies), 203-1 Shirakata, Tokai, Ibaraki 319-1106, Japan Advanced Chemical Energy Research Center, Yokohama National University, 79-5 Tokiwadai, Hodogaya-ku, Yokohama, Kanagawa 240-8501, Japan


Introduction
Research interest of rechargeable sodium batteries is continuously growing due to the increasing demand for low-cost energy storage devices for stationary applications. During the past decade, considerable research efforts have been devoted to develop electrode materials for sodium batteries [1,2]. Particularly, research has been focused on the searching of advanced positive electrode materials. Among the various types of potential positive electrode materials, layered oxides and polyanionic compounds have drawn intensive attention [3][4][5][6][7][8]. As the most examined branches of positive electrode materials, layered-type transition metal oxides, Na x MO 2 (M = 3d transition metals), with different layered stacking manners have been intensively examined due to the wider selection for tailoring layered structures and electrode performance [7][8][9][10][11][12][13]. From a practical perspective, the use of electrode materials made from Earth-abundant transition metal elements is preferable, and the oxides with Fe and Mn ions have been studied as strong candidates for sodium battery applications [14].
Among Mn-based oxides studied for sodium storage applications, Na 0.44 MnO 2 (Na 4 Mn 9 O 18 ) with a tunnel-type host structure [15] and Na 0.7 MnO 2 with a P2-type layered structure [16] have been extensively studied. Herein, "P" denotes environment for Na ions, and Na ions are found at prismatic sites [17]. The number "2" corresponds to numbers of MeO 2 slabs in a unit cell. Both oxides show promising electrode performance with excellent reversibility in Na cells. Nevertheless, both oxides are Na-deficient phases, and therefore, a small reversible capacity is obtained for a full cell configuration. To solve this problem, the use of a sacrificial salt, such as Na 2 CO 3 , has been proposed in literature [18]. However, available volumetric energy density is inevitably lowered by addition of sacrificial salts. Therefore, a stoichiometric layered material, NaMnO 2 (or Narich compounds [19]), is more preferable for the full cell configuration. NaMnO 2 has been studied as an electrode material in early 1980s [20]. NaMnO 2 has two polymorphs, a low-temperature alpha phase with a monoclinic distorted O′3-type layered structure, and a high-temperature beta phase with an orthorhombic zigzag-type layered structure. Here, "O" denotes an octahedral environment for Na ions, and the prime symbol indicates the presence of in-plane distortion for MeO 2 slabs [17]. Both phases consist of a cubic close-packed oxygen array, and therefore, intergrowth of both phases is experimentally observed [21]. Electrode performance of alpha-NaMnO 2 was reexamined in 2011 [5], but reversibility is not acceptable for battery applications. In contrast, the high-temperature phase shows much improved electrode performance [21] when compared with the low-temperature phase. However, beta-NaMnO 2 synthesized at high temperatures contains a large fraction of alpha-NaMnO 2 domains [21,22], and this fact probably originates from structural compatibility and similarity of formation energy in both phases. Therefore, to the best of our knowledge, successful synthesis of "phase-pure" beta-NaMnO 2 without alpha domains is not found in the literature. Recently, stabilization of beta-NaMnO 2 by Cu-substitution has been reported and NaCu 0.1 Mn 0.9 O 2 crystallizes into the beta phase with less alpha domains [23,24].
In this article, significant improvement of electrode performance of O3 NaMnO 2 is described through a different and simple methodology without less abundant elements, i.e., Ti substitution for Mn ions. A binary system of x NaMnO 2 -(1x) TiO 2 (1 ≤ x ≤ 0:5) is systematically examined and tested as electrode materials in Na cells. Our previous work has demonstrated that Na-deficient layered phases are successfully obtained in a binary system of NaCrO 2 -TiO 2 , and highly improved electrode performance is realized for solid solution samples in this binary system [25]. A solubility limit of TiO 2 in NaCrO 2 without the destruction of layered structures reaches approximately 40%. In the case of NaMnO 2 , a much narrower solubility limit of TiO 2 is found because of higher phase stability of Na 4 Mn 4 Ti 5 O 18 with a tunnel-type structure [26]. Nevertheless, Ti-substituted and Na-deficient O′3-type layered phases are successfully obtained, and the samples show promising electrode performance as positive electrode materials for rechargeable Na batteries. From these results, the possibility of the development of high-performance batteries made from only Earth-abundant elements is discussed in detail.

Experimental
2.1. Synthesis of Materials. The binary system of x NaMnO 2 -(1x) TiO 2 (0:5 ≤ x ≤ 1) with different chemical compositions was synthesized through a solid-state calcination method. Starting materials consisting of Na 2 CO 3 (99.5%; Wako Pure Chemical Industries), Mn 2 O 3 , and anatase-type TiO 2 (98.5%, Wako Pure Chemical Industries) were thoroughly mixed by ball milling. Mn 2 O 3 was obtained from MnCO 3 (Kishida Chemical) by heating at 850°C for 12 h in air. Mixtures after milling were pressed into pellets and calcined at 800 (0:9 ≤ x ≤ 1) or 1000°C (0:5 ≤ x ≤ 0:8) in Ar for 12 h in a tube furnace. To enhance the electrode performance, the sample was thoroughly mixed with acetylene black (sample : acetylene black = 90 : 10 wt%) by using a planetary ball mill (Pulverisette 7, Fritsch) at 300 rpm for 6 h with a zirconia pot and balls. Thus, obtained carbon composited samples were handled in an Ar-filled glove box to avoid contact to moist air.

Characterization of Materials.
Crystal structures of the samples were examined by using an X-ray diffractometer (XRD) equipped with a high-speed position-sensitive detector (D2 Phaser, Bruker). Nonmonochromatized Cu Kα radiation was utilized as an X-ray source with a nickel filter. The morphological features of the samples were observed using a scanning electron microscope (JEOL, JCM-6000). Synchrotron XRD data were collected at the beamline BL16B2 in SPring-8 synchrotron facility in Japan [27]. The measurement was conducted using an automatic powder diffraction system for Debye-Scherrer geometry using a capillary sample. The wavelength of X-ray was calibrated to be 0.5003 Å with CeO 2 reference. Rietveld refinement was carried out using RIETAN-FP software [28]. A neutron diffraction pattern was collected at BL09 (SPICA) in the Material and Life Science Facility (MLF) at the Japan Proton Accelerator Research Complex (J-PARC) [29]. Structural analysis of the neutron diffraction pattern was carried out using Z-Rietveld [30,31].
The electrode performance of the samples was examined in Na cells. The composite positive electrodes, consisting of 76.5 wt% the carbon composited sample, 13.5 wt% acetylene black, and 10 wt% poly(vinylidene fluoride), pasted on aluminum foil as a current collector were tested. The electrodes were dried at 120°C in a vacuum. Metallic sodium foil was used as a negative electrode. The electrolyte solution used was 1.0 mol dm -3 NaPF 6 dissolved in propylene carbonate (Battery grade, Kishida Chemical). A glass filter (GB-100R, Advantec) was used as separator. Two-electrode electrochemical cells (TJ-AC, Tomcell Japan) were assembled in the Ar-filled glove box and cycled at a rate of 10 mA g -1 at room temperature. The rate capability of the samples was 2 Energy Material Advances also evaluated in Na cells. The cells were charged at the same current density of 20 mA g -1 to 4.5 V vs. Na and then discharged at various rates (20-1280 mA g -1 ) to 1.2 V. Impedance measurement was conducted by using a potentiostat with a frequency response analyzer (SP-200, Bio-Logic) at 10 mV amplitude. Hard X-ray absorption spectroscopy (XAS) at the Mn K-edge was performed at beamline BL-9C of the Photon Factory Synchrotron Source in Japan. Hard XAS spectra were collected with a silicon monochromator in transmission mode. The intensity of incident and transmitted X-rays was measured using an ionization chamber at room temperature. Composite electrode samples were prepared using the twoelectrode cells at a rate of 10 mA g -1 . The composite electrodes were rinsed with dimethyl carbonate and sealed in a water-resistant polymer film in the Ar-filled glove box. Normalization of the XAS spectra was carried out using the program code IFEFFIT [32]. The postedge background was determined using a cubic spline procedure.
In situ XRD patterns were obtained using an electrochemical cell equipped with a Be window and XRD diffractometer (Bruker, D8 Advance). Slurry was pasted on thin Al foil (5 μm) and used as a composite electrode for in situ measurement. The cell was charged at a rate of 5 mA g -1 , and the XRD data were collected for a second charge process.

Results and Discussion
3.1. Synthesis and Structural Characterization of Ti-Substituted NaMnO 2 . Figure 1 shows XRD patterns of a binary system for x NaMnO 2 -(1x) TiO 2 (0:5 ≤ x ≤ 1) prepared by a solidstate calcination method. Observed diffraction lines for NaMnO 2 (x = 1:0) are successfully indexed to an O ′ 3-type layered structure with a space group of C2/m. The refined lat-tice parameters for NaMnO 2 are a = 5:668 Å, b = 2:859 Å, c = 5:800 Å, and β = 113:18°, which is consistent with the reported data in literature [5]. Generally, in-plane distortion of NaMnO 2 originates from the Jahn-Teller effect of Mn 3+ ion (t 2g 3 e g 1 ), and a ratio of a/b values (1.98) is indicative of the degree of in-plane distortion for layered materials. As Ti fractions increase, significant changes on XRD patterns are noted. A nearly single phase of Na 4 Mn 4 Ti 5 O 18 is also obtained at x = 0:5. This phase is essentially isostructural with Na 0.44 MnO 2 [15], and a framework structure consists of Mn(Ti)O 6 octahedra and MnO 5 square pyramids, forming a tunnel-type structure as shown in Figure 1 [26]. Na ions are found at three distinct crystallographic sites in the tunneltype framework structure. Similar phases for Na-Mn-Ti ternary oxides with the tunnel-type structure have been also reported in literature [33,34]. Note that the presence of this phase is evidenced even for the sample of x = 0:9 from the XRD patterns, and the Mn-rich layered oxide coexists with the Tirich tunnel-structured oxide for the samples of 0:5 < x ≤ 0:9. The presence of the Ti-rich oxide and phase segregation is also clearly evidenced from elemental distribution maps shown in Supporting Figure S1. Nevertheless, a small amount of Ti ions is also found in the Mn-rich region as a major phase from the elemental maps. This fact suggests that Ti ions are partially incorporated in the layered phase, forming Na 1-x Ti x Mn 1-x O 2 with the layered structure. Moreover, a detailed structural analysis by the Rietveld analysis also supports this hypothesis and is further discussed in the later section.
Particle morphology of these samples observed by SEM is also shown in Figure 1. NaMnO 2 without Ti substitution consists of agglomerated particles with nonuniform sizes (0.3-1.0 μm). The Ti-substituted sample, x = 0:8, has similar particle morphology with the nonsubstituted sample, but a larger particle size with smooth surface is noted. The 3 Energy Material Advances difference probably originates from the fact that higher synthesis temperature was used for the Ti-substituted sample. Moreover, for Na 0.5 Mn 0.5 Ti 0.5 O 2 , densification of the pellet was observed after calcination. In addition to relatively large micrometer-sized particles, tiny particles are also found, which are formed when the pellet was mechanically crushed into powder form.
Figures 2(a) and 2(b) show a result of Rietveld refinement on neutron diffraction (ND) and synchrotron XRD patterns of the sample, x = 0:8. Refined structural parameters are listed in Supporting Tables S1 and S2. Because similar lattice parameters are noted for the second phase for the all samples of 0:5 ≤ x ≤ 0:9 in Figure 1, the formation of Na 4 Mn 4 Ti 5 O 18 with the tunnel-type structure as the second phase is hypothesized. Structural analysis reveals that nearly 20% (mass percent) of the second phase is present, from which a Ti content of the main phase is estimated and calculated (also see Figure 2(c)). On the basis of this hypothesis, a chemical composition of the major phase is estimated to be Na 0.88 Mn 0.88 Ti 0.12 O 2 with the O′3-type layered structure. The difference in fractions of Ti contents for the layered and tunnel structures is also consistent with the finding of elemental mapping for Ti distributions (Supporting Figure S1). Moreover, the consistent results are also noted for both the SXRD and ND patterns with this two-phase model. Such nonstoichiometric phase segregation has also been discussed in other Fe/Mn-based layered materials [35]. Successful substitution by Ti ions is also supported from a change in lattice parameters for the major phase. Lattice parameters of the major layered phase are calculated to be a = 5:599 Å, b = 2:860 Å, c = 5:821 Å, and β = 112:38°as shown in Supporting Tables S1 and S2. The a/b ratio (1.958) is decreased compared with the nonsubstituted sample mentioned above and close to the ideal value of √3 without in-plane distortion. This fact indicates that Jahn-Teller active Mn 3+ ions are partly and successfully replaced by Jahn-Teller inactive Ti 4+ ions without valence electrons. From the refined result for the neutron diffraction pattern, Bragg diffraction lines with small d-values (~0.5 Å) are nicely refined with the two-phase model. Additionally, large isotropic displacement parameters are found for the sodium sites by Rietveld analysis on both neutron and X-ray diffraction data, which indicates that vacant sites are formed at sodium sites, leading to facilitating self-diffusion of Na ions [36].

Electrode
Performance of Ti-Substituted NaMnO 2 . The influence of Ti substitution on the electrochemical properties of the binary system of x NaMnO 2 -(1x) TiO 2 (0:5 ≤ x ≤ 1) was examined in Na cells, and galvanostatic charge/discharge curves of the carbon composited samples are shown in Figure 3. Note that the electrode performance on this binary system is significantly increased by the preparation of carbon composited samples. For comparison, the electrode performance of "as-prepared" NaMnO 2 is shown in Supporting Figure S2. Large irreversible capacity for the initial cycle (approximately 90 mA h g -1 ) and large polarization are observed. Large irreversible capacity is drastically reduced (approximately 30 mA h g -1 ) for the carbon composite sample, and initial discharge capacity reaches nearly 200 mA h g -1 . For the carbon composite sample, less agglomeration for secondary particles is noted from SEM images in Supporting Figure S2. Reduction of secondary particle sizes and uniform integration with nanosized carbon effectively improves the electrode performance of NaMnO 2 . Nevertheless, a reversible capacity of 35 mA h g -1 is lost during initial 10 cycles. It is noted that a browncolored separator is found after 10 cycles in a Na cell as shown in Figure 3(b), indicating that reduction of the reversible capacity partly originates from dissolution of Mn   Figure 2: Results of Rietveld analysis on (a) synchrotron X-ray and (b) neutron diffraction patterns of the sample of x = 0:8. A major phase crystallizes into the in-plane distorted layered structure, and the presence of a minor phase with the tunnel-type structure is also found. Refined structural parameters obtained by the Rietveld method are summarized in Supporting Tables S1 and S2. A scheme for the process of phase segregation and two-phase formation is also shown in (c).

Energy Material Advances
ions into the electrolyte solution. This observation is also consistent with a recent theoretical study, and it is proposed that the presence of proton accelerates Mn migration in the crystal lattice, leading to Mn dissolution into electrolyte [37]. For NaMnO 2 , stepwise charge/discharge profiles are observed, which suggests that complicated and multiple phase transitions with a two-phase reaction are involved during cycling, probably associated with charge, spin, and orbital ordering on desodiation [38]. The nature of biphasic reaction is also clearly observed in dQ/dE differential curves in Figure 3(c). In the case of Ti-substituted samples, obvious changes in voltage profiles are observed. As Ti contents increase, stepwise profiles are lost, and instead, S-shaped profiles are more pronounced, indicating that single-phase reac-tion proceeds in the entire range. For the Ti-rich sample, x = 0:5, NaMnTiO 4 , a sloping voltage profile without voltage plateaus is observed, and no peak is found in dQ/dE curves. Indeed, Na 0.44 Mn 0.44 Ti 0.56 O 2 , which has a similar chemical composition with x = 0:5 shows a single-phase reaction revealed by in situ XRD study on electrochemical cycles [39]. NaMnTiO 4 shows excellent capacity retention but an available reversible capacity is reduced to 120 mA h g -1 .
Moreover, the initial charge capacity is quite small compared with the initial discharge capacity associated with the formation of the Na-deficient phase.
Among the tested sample in this binary system, the sample of x = 0:8, Na 0.8 Mn 0.8 Ti 0.2 O 2 , shows superior electrode performance. A large initial discharge capacity reaches around 175 mA h g -1 , which is comparable to that of  Energy Material Advances NaMnO 2 . Estimated energy density reaches 460 Wh kg -1 vs. metallic Na. Moreover, capacity retention as electrode materials is significantly improved by Ti substitution. Indeed, no change in color of the separator is observed after 10 cycles as shown in Figure 3(b), and this observation clearly indicates that suppression of Mn dissolution is responsible for the improvement of cyclability. A voltage profile is also influenced by Ti substitution, and less pronounced voltage plateaus are observed in dQ/dE curves (Figure 3(c)). These facts, including the results of structural analysis, conclude the successful synthesis of Ti-substituted O′3type manganese-based layered oxide. However, an initial charge capacity is slightly smaller than that of the initial discharge capacity associated with the formation of the Nadeficient phase. Nevertheless, the concentration of Nadeficient sites is relatively small compared with P2-type Na 0.7 MnO 2 and NaMnTiO 4 , and therefore, for the sample of x = 0:8, a sacrifice in energy density for the full cell configuration would be expected to be limited and small.
The impact of Ti substitution for Mn ions on electrode performance is further examined in Na cells. Quasi-opencircuit voltages (QOCV) for the samples of x = 1:0 and 0.8 were measured by galvanostatic intermittent titration technique (GITT) as shown in Figure 4(a). Although polarization upon charging below 3.5 V is relatively small (∼20 mV) for the sample of x = 1:0, NaMnO 2 , the polarization is increased to >100 mV when charged above 3.5 V. On the other hand, the sample of x = 0:8, Na 0.8 Mn 0.8 Ti 0.2 O 2 , shows much smaller polarization than that of x = 1:0, especially at the high-voltage region (>3.5 V). The reduction of polarization is further supported by impedance measurement as shown in Supporting Figure S3. The rate capability of both samples is also compared in Figure 4(b). Larger discharge capacities are observed for the nonsubstituted sample at slower rates. However, a clear difference in rate capability is evidenced for both samples as increase in current density. For the Ti-substituted sample, a relatively large discharge capacity of approximately 140 mA h g -1 is obtained even at 1280 mA g -1 , corresponding to 80% of the discharge capacity at 20 mA h g -1 . The reduction of polarization and improvement of rate capability are realized by Ti substitution. In general, the substitution of Ti 4+ ions, which have no valence electron, for Mn 3+ ions, is expected to result in reduction of electronic conductivity, but chemical stability is effectively improved by Ti substitution, and thus, dissolution of Mn ions into electrolyte solution is suppressed [40]. From these results, it is hypothesized that dissolution of Mn ions for the nonsubstituted sample results in gradual loss of electrical contact on electrochemical cycles, leading to the increase in the polarization as electrode materials.
The feasibility of the Ti-substituted sample as positive electrode materials is further examined in Na cells. Although the Ti-substituted sample shows much improved capacity retention compared with the non-substituted sample (Figure 3(d)), further improvement is required to use this material for practical applications. Therefore, the cut-off voltage for charge was reduced from 4.5 V to 4.2 V to avoid possible side reactions during high-voltage exposure, and voltage holding at 4.2 V after constant current charge was applied. When cut-off voltage is reduced to 4.2 V, a much improved capacity retention is achieved as shown in Figure 4(c) even though an initial discharge capacity is inevitably reduced to 170 mA h g -1 . In addition, capacity retention is further improved by voltage holding at 4.2 V. An initial charge capacity reaches 185 mA h g -1 , which is almost the same capacity with an initial discharge capacity. The problem of the Na-deficient phase, i.e., a small initial charge capacity, is less pronounced by using voltage holding at 4.2 V. Charge/discharge curves of the Ti-substituted sample without voltage holding is shown in Supporting Figure S4. Approximately 90% of the reversible capacity is retained after 30 cycles at a rate of 20 mA g -1 . A result of an accelerated cycle test at a rate of 50 mA g -1 is also shown in Figure 4(d), and voltage holding at 4.2 V on charge was used in the cycle test. The Ti-substituted sample shows good capacity retention, and 85% of the reversible capacity is retained even after the 100-cycle test.

Sodiation/Desodiation Mechanisms of Na y Mn 0.8 Ti 0.2 O 2 .
Reaction mechanisms of the sample of x = 0:8 has been further examined by ex situ X-ray absorption (XAS) spectroscopy and in situ XRD study. Figure 5(a) shows that Mn K-edge XAS spectra continuously shift towards a higher energy region after charge to 4.2 V, and an opposite trend is noted on discharge. The energy position shifts back to the original position after being discharged to 1.2 V. From the energy position changes observed in XAS spectra, it is concluded that one electron redox of Mn ions, Mn 3+ /Mn 4+ , is effectively utilized in this system and this process is highly reversible. High reversibility of cationic redox reaction is achieved for Na 0.8 Mn 0.8 Ti 0.2 O 2 . It is also anticipated that the use of anionic redox reaction [41], coupled with these cationic redox, would further increase the reversible capacity and energy density of positive electrode materials by designing Na-excess Mn-based oxides [42,43].
To further examine the origin of good electrochemical reversibility for Na 0.8 Mn 0.8 Ti 0.2 O 2 , in situ XRD patterns of the sample on the electrochemical cycle have been collected as shown in Figure 5(b). For a fully discharged phase, a major phase is assigned into the O ′ 3-type layered phase (denoted as O ′ 3 I phase in Figures 5(b) and 6), and the presence of the oxide with the tunnel-type structure is also noted. On charge, the phase transition is clearly observed. The 20-2 diffraction line for the O ′ 3 I phase at 37.9 degrees in 2 theta is weakened, and a new peak appears at 39.0 degrees. Similarly, new peaks are observed at 16.1 and 32.5 degrees, and these peaks are assigned as 001 and 002 diffraction lines for the monoclinic layered structure. This phase is observed in 0:62 ≤ y ≤ 0:82 in Na y Mn 0.8 Ti 0.2 O 2 and denoted as the O ′ 3 II phase in Figure 5(b). A two-phase coexistence of the O ′ 3 I and II phases is noted in the region of 0:86 < y ≤ 0:72. The interlayer distance increases from 5.38 to 5.51 Å for the O ′ 3 II phase. On further charge, the peak intensity of the O ′ 3 II phase is less intensified, and instead, the peak at 41.7 degrees appears. This phase is also assigned to be the O ′ 3-type phase (denoted as the O ′ 3 III phase), but the distortion of the crystal lattice (a/b ratio and β angle for the monoclinic lattice) associated 6 Energy Material Advances with the presence of a Jahn-Teller active Mn 3+ ion with a high-spin configuration is clearly different from the O′3 I and II phases as illustrated in Figure 6. In addition, the interlayer distance is expanded to 5.66 Å. Another biphasic reaction proceeds for both O′3 II and III phases in the range of 0:68 ≤ y < 0:58. Crystallographic parameters of the O′3 III phase are close to those of Na 5/8 MnO 2 reported in literature [44]. Less in-plane distortion of the crystal lattice for the O ′ 3 III phase originates from the fact that the Jahn-Teller active ion in the structure is diluted on charge. On further oxidation, a single-phase reaction proceeds and the in-plane local distortion is relieved. The ratio of a/b values is calculated to be 1.732 (=√3), indicating that its symmetry has been changed from the monoclinic to rhombohedral symmetry. The 20-2 and 111 diffraction lines for the O ′ 3 III phase are merged into a single diffraction line, which is assigned to 7 Energy Material Advances the 104 diffraction line for an O3 phase without in-plane distortion. The single-phase O3 layered structure is found for the range of 0:54 ≤ y ≤ 0:22 in Na y Mn 0.8 Ti 0.2 O 2 . Note that the formation of the O1 phase (a sequence of ABAB for oxygen stacking) is found after being fully charged (y < 0:22), which is clearly evidenced from the disappearance of the 104 line for the O3 phase and appearance of a new peak at 42 degrees. This peak is assigned as the 011 line for the O1 phase. The interlayer distance is also reduced from 5.66 Å, which is the maximum value on phase transition processes, to 5.13 Å for the fully desodiated sample. A similar interlayer distance is observed for the O1 phase (5.08 Å) derived from stoichiometric NaMnO 2 without Ti substitution [45]. Broad diffraction lines observed for the O1 phase suggest the presence of stacking faults along the c-axis direction. For other O3 phases, e.g., Na y Ni 1/3-Fe 1/3 Mn 1/3 O 2 [46] and Na y Ti 0.25 Fe 0.25 Co 0.25 Ni 0.25 O 2 [47], the formation of an OP2 phase (a sequence of ABCA), which contains alternating octahedral and prismatic sites for sodium ion layers, is evidenced. This phase is regarded as the intergrown layered material of P3-and O1-type structures. The interlayer distance of y =~0 in Na y Ni 1/3-Fe 1/3 Mn 1/3 O 2 is observed to be 5.05 Å [46], and this value is also close to that of Na 0.05 Mn 0. 8 Figure 5: (a) Changes in XANES spectra of the sample of x = 0:8 on charge/discharge processes and (b) in situ XRD patterns of the sample x = 0:8 for the second charge process at 5 mA g -1 . The cell was discharged to 1.2 V at 5 mA g -1 and then charged to 4.4 V. Each XRD data has been collected for 1 h on charge. 8 Energy Material Advances Note that phase transition processes simultaneously occur for the minor phase, Na 4 Mn 4 Ti 5 O 18 , with the tunneltype structure. The continuous contraction of the crystal lattice and shift of diffraction lines to the higher diffraction angle are observed on charge, which is consistent with the trend reported in literature [39]. Relatively clear phase transition is noted in the range of 0:38 < y < 0:28 in Na y Mn 0.8-Ti 0.2 O 2 . For instance, the 140 diffraction line at 16.0 degrees shifts to 16.5 degrees as shown in Figure 5(b) on charge. A peak intensity of 140 and 200 is increased on charge, and a similar observation is also noted for Na y Mn 4 Ti 5 O 18 [39]. Less lattice distortion is expected for the charged sample with diluted Jahn-Teller active Mn 3+ ions. These phase transition competes with the formation of the major O3 phase. Continuous shift of diffraction lines is observed in the entire region, but the unit cell volume change is less pronounced for the tunnel phase when compared with the layered phase. Because the tunnel and layered phases are uniformly dispersed as shown in Supporting Figure S1, the strain induced by large volume change of the layered phase would be partly relieved by the presence of the tunnel phase with less volume change around the layered phase. Structural changes of the samples with/without Ti ions after 10 cycles were also examined. Crystallinity of the sample is clearly reduced for the nonsubstituted sample (Supporting Figure S5), indicating that strains are induced after electrochemical cycles. The dissolution of Mn ions from the crystal lattice of NaMnO 2 may result in the broadening of diffraction lines. Although the phase-pure sample is not obtained in the binary system of NaMnO 2 -TiO 2 , both layered and tunnel phases are electrochemically active as clearly shown in Figure 5(b). Therefore, no sacrifice of reversible capacity is observed for these nonpure phases, which are used as highcapacity positive electrode materials for high-performance and cost-effective sodium storage applications combined with carbonaceous negative electrode materials [48] in the future.

Conclusions
The influence of the Ti substitution on the structural and electrochemical properties of NaMnO 2 has been examined   9 Energy Material Advances in this study. A combined study of neutron and synchrotron X-ray diffraction reveals partial substitution of Ti ions and the formation of layered Na 1-x Mn 1-x Ti x O 2 with tunnel-type Na 4 Mn 4 Ti 5 O 18 as the minor phase. The Ti-substituted sample has higher chemical stability, and thus, dissolution of Mn ions into electrolyte on electrochemical cycles is suppressed, leading to good cyclability and rate capability as the electrode material. An in situ XRD study on Na y Mn 0.8-Ti 0.2 O 2 reveals that the partial suppression of unit cell volume change is also evidenced compared with stoichiometric and nonsubstituted NaMnO 2 . Moreover, both layered and tunnel-type oxides are electrochemically active, and simultaneous phase transitions are observed on sodiation/desodiation processes. High reversibility as the electrode material also originates from Mn 3+ /Mn 4+ cationic redox as proofed by XAS measurement. The optimization of synthesis conditions, chemical compositions, and particle morphology would result in further improvement of electrode performance, contributing to the development of cost-effective Na-based battery system without less abundant elements.

Data Availability
All data presented in the paper and the supporting information are available from the corresponding author upon reasonable request.

Ethical Approval
The synchrotron radiation experiments were performed at the BL19B2 of SPring-8 with the approval of the Japan Synchrotron Radiation Research Institute (JASRI) (Proposal 2019B1685). The neutron scattering experiment was approved by the Neutron Scattering Program Advisory Committee of IMSS, KEK (Proposal No. 2019S10). The synchrotron X-ray absorption work was done under the approval of the Photon Factory Program Advisory Committee (Proposal No. 2019G033).

Conflicts of Interest
The authors declare that they have no conflicts of interest.