Effects of Phase Change and Cu Doping on the Li Storage Properties of Rutile TiO₂

Abstract

The crystal structure and Li storage properties of Cu-doped rutile TiO₂ after a phase change caused by lithiation were investigated for the first time. Structural analysis results confirmed that undoped rutile TiO₂ was transformed to a distorted layered rock-salt Li_xTiO₂ structure with a small volume expansion of only 1 % when cycled in a potential range of 1.0–3.0 V vs. Li⁺/Li. A substitutional solid solution of Cu²⁺ was formed in layered Li_xTiO₂. The Cu doping increased both the interlayer distance and electronic conductivity of the layered Li_xTiO₂. As an Li-ion battery anode, a Cu-doped TiO₂ electrode exhibited a long cycle life, maintaining a reversible capacity of 120 mAh g⁻¹ over 10000 cycles at 5C and an excellent rate capability of 108 mAh g⁻¹ at 50C. Furthermore, this electrode could also be potentially used as a Na storage material. These attractive properties demonstrate high applicability of Cu-doped rutile TiO₂ as a novel anode material.

1. Introduction

Among various transition-metal oxides, titanium dioxide (TiO₂) has attracted much attention from researchers as a host material for the anode of Li-ion batteries (LIBs) operating at low potentials. Owing to the low cost, abundance, and safe Li insertion potential range (1.3–1.7 V vs. Li⁺/Li) of TiO₂ anodes, they have been extensively studied by many researchers.^1,2 There are eight polymorphs of TiO₂, including rutile (space group: P4₂/mnm), anatase (I4₁/amd), brookite (Pbca), bronze (TiO₂(B), C2/m), ramsdellite (Pbnm), hollandite (I4/m), columbite (TiO₂ II, α-PbO₂ structure, Pbcn), and baddeleyite (TiO₂ III, P2₁/c).³ Among these phases, anatase has been most actively investigated in the literature^4,5 because it demonstrated high activity during Li insertion/extraction reactions. Bronze^6,7 and brookite⁸ were also studied as promising anode materials. Finally, the anode properties of the high-pressure (>12 GPa) columbite phase were examined as well,⁹ while the use of the rutile phase as an anode material for LIBs resulted in many complications.

Typically, micrometer-sized rutile TiO₂ exhibits very small reversible capacities that are lower than 10 % of its theoretical capacity of 335 mAh g⁻¹.^1,9 Maier,³ Wagemaker,¹⁰ and Tarascon¹¹ have discovered an interesting size effect: when the TiO₂ size is reduced to several tens of nanometers, the reversible capacity of rutile TiO₂ considerably increases. Furthermore, rutile TiO₂ has unique Li⁺ diffusion characteristics: its diffusion coefficient along the c-axis direction is as high as 10⁻⁶ cm² s⁻¹, while its magnitude equals only 10⁻¹⁴ cm² s⁻¹ in the ab-plane direction. It was concluded that the size effect of rutile was much stronger than those of other TiO₂ polymorphs¹ because the nanometer-sized rutile particles facilitated Li⁺ transport. These findings have stimulated further research studies of nanostructured rutile TiO₂.^13–17 Meng et al. synthesized TiO₂ nanoflakes with a thickness of 40 nm to improve the TiO₂ anode properties. Lou et al. reported porous rutile TiO₂ submicroboxes with a porous structure, which demonstrated reversible capacities of 150 mAh g⁻¹ at a C-rate of 0.5C (170 mA g⁻¹) during 500 cycles and high rate performance of 80 mAh g⁻¹ at 10C.¹⁶ The authors of this work focused on the impurity doping and high crystallization degree of rutile TiO₂.^18–20 We performed Nb doping of rutile TiO₂¹⁸ and demonstrated that a Nb-doped TiO₂ electrode exhibited a very high rate capability of 120 mAh g⁻¹ even at a high rate of 50C.¹⁹ In addition, we achieved an outstanding cyclability with 170 mAh g⁻¹ during 5500 cycles at 1C by using single-crystalline rutile TiO₂.²⁰ After reporting the remarkable rutile doping effect, the number of research studies on impurity-doped rutile TiO₂ significantly increased.^21,22

An irreversible phase change of rutile TiO₂ during the initial lithiation process has been recently observed by in-situ X-ray diffraction (XRD) and high-resolution transmission electron microscopy (TEM).²³ The first lithiation process leads to a monoclinic distortion of the Li_xTiO₂ rutile structure. At x > 0.8, this distortion causes an irreversible phase change of the rutile structure to a α-NaFeO₂-type layered Li_xTiO₂ phase.²³ More precisely, the irreversible phase change produces a nanostructure consisting of layered Li_xTiO₂ with a domain size of 5 nm and columbite-type Li_xTiO₂ with a size of 1 nm. The α-NaFeO₂-type layered Li_xTiO₂ material undergoes charge–discharge reactions via a solid solution mechanism with a very small volume expansion ratio of 1 %.²³ The excellent cyclability of rutile TiO₂ electrodes is likely due to a such small volume change.

In contrast, Díaz-Carrasco et al. claimed that the first lithiation of rutile TiO₂ generated not α-NaFeO₂-type layered Li_xTiO₂ but a disordered cubic rock-salt type Li_xTiO₂ structure, which was confirmed by an ex-situ XRD analysis.²⁴ Cyclic voltammetry studies demonstrated that the eversible capacity and high rate performance of rutile TiO₂ most likely originated from capacitive processes. Moreover, Yang et al. also reported a rock-salt Li_xTiO₂ electrode with high rate capability.²⁵

The mechanism of the crystal phase change of rutile TiO₂ remains controversial. Moreover, the effects of impurity doping on the crystal structure and anode properties of rutil TiO₂ electrodes after the phase change have not been examined yet. The main effects of doping rutile TiO₂ electrodes are the increase in electronic conductivity and the expansion of a Li⁺ diffusion pathway, which has been reported by the authors.^18–20 Thus, it is interesting to know how doping affects the TiO₂ electrode after the phase change. In this study, we investigated the crystal structure of TiO₂ after the phase change and the influence of impurity doping on TiO₂ anode properties. Cu²⁺ and Al³⁺ were selected as impurity ions because oxygen vacancies enhancing the anode properties of TiO₂ electrodes were expected to form due to charge compensation.^26,27

2. Experimental

2.1 Active material synthesis

Impurity-doped TiO₂ was synthesized by a hydrothermal method.^28,29 A titanium isopropoxide solution was prepared by mixing 5 mL of Ti[OCH(CH₃)₂]₄ (titanium tetraisopropoxide, 95 %, FUJIFILM Wako Pure Chemical), 5 mL of CH₃CH(OH)CH₃ (isopropanol, 99.5 %, Sigma-Aldrich), and 50 mL of an aqueous solution of glycolic acid (HOCH₂CO₂H, 99.0 %, Sigma-Aldrich). The concentration of titanium tetraisopropoxide was 1.6 mol L⁻¹. When TiO₂ was doped with Cu²⁺, Al³⁺, and Nb⁵⁺ ions, copper ethoxide, aluminum isopropoxide, and niobium ethoxide, respectively, were added to the solution. After stirring for 90 min at a rotation speed of 500 rpm and temperature of 80 °C, a hydrothermal reaction was performed in a Teflon-lined stainless-steel autoclave (HU-50, SAN-AI Kagaku) at 200 °C for 9 h. The resulting precipitate was dried in air at 100 °C for 12 h.

2.2 Material characterization

The formation of single-phase rutile-type TiO₂ in all cases of impurity addition was confirmed by XRD (Fig. S1). The doping amounts of 1 at% Cu, 3 at% Al, and 4 at% Nb were determined by X-ray fluorescence (XRF, EDX-720 Shimadzu Co., Ltd.) (Table 1). The rod-like morphology of impurity-doped rutile TiO₂ particles (Fig. S2) was observed by field-emission scanning electron microscopy (FE-SEM, JSM-6701F, JEOL Ltd.). The oxygen vacancy amount (y) in the impurity (M)-doped titanium oxides (M_xTiO_2−y) was determined using an oxygen analyzer (ONH836, LECO Corp.).²⁸ In the case of Cu-doped TiO₂, the composition can be described as [Cu²⁺]_0.01[Ti⁴⁺]_(1−t)[Ti³⁺]_t[O²⁻]_(2−0.26). Due to the charge compensation, the Ti³⁺ amount t should be 0.58. Thus, the composition is [Cu²⁺]_0.01[Ti⁴⁺]_0.42[Ti³⁺]_0.58[O²⁻]_1.74. Similarly to this, the compositions of other TiO₂ can be estimated as [Al³⁺]_0.03[Ti⁴⁺]_0.29[Ti³⁺]_0.71[O²⁻]^1.69, [Nb⁵⁺]_0.04[Ti⁴⁺]_0.42[Ti³⁺]_0.58[O²⁻]_1.89, and [Ti⁴⁺]_0.8[Ti³⁺]_0.2[O²⁻]_1.90. The electrical resistivity of TiO₂ particles was measured at a pressure of 55 MPa.²⁸ The two-probe resistivity measurements were conducted using a direct current at a room temperature. The nanostructure of TiO₂ after lithiation and delithiation was investigated by TEM (JEMARM200F, JEOL, Co., Ltd.). The cell was disassembled in an Ar atmosphere after charge–discharge cycles.²⁹ TEM observations of the active TiO₂ material were performed without air exposure. The TEM samples were lifted out and thinned by using a focused ion beam of Ga.

Table 1. Characteristics of impurity-doped TiO₂: doping amount, lattice parameters (a and c), and oxygen vacancy amounts. The standard lattice parameters are also provided for rutile-type TiO₂ [Inorganic Crystal Structure Database (ICSD) No. 00-021-1276].

Dopant (M)	Doping amount	Ionic diameter /pm	Lattice parameter		Oxygen vacancy amount (y in MxTiO2−y)
			a /nm	c /nm
Cu2+	1 at%	146	0.4609	0.2955	0.26
Al3+	3 at%	107	0.4601	0.2954	0.31
Nb5+	4 at%	128	0.4609	0.2962	0.11
Undoped (Ti4+)	0 at%	121	0.4603	0.2958	0.10
Standard data of TiO2 (ICSD No. 00-021-1276)			0.4593	0.2959	0

2.3 Electrochemical studies

TiO₂ electrodes were prepared by a gas deposition method.^30,31 During this process, active material particles were accelerated at a high speed ranging from 150 to 500 m s⁻¹ by the carrier gas emitted from a gas nozzle. The high-speed collision between the particles and the current collector substrate induced their plastic deformation. Because of the high impact energy, mutual diffusion of atoms occurred at the particle interface to produce electrodes consisting of strongly adhered particles.^32,33 When electrodes are prepared via this method, we can evaluate a basic electrochemical properties of active materials because the electrodes do not contain binders and conductive additives. The authors have utilized this method to fabricate the electrodes of rechargeable batteries for the first time.³⁰ In this study, gas deposition was performed at a nozzle diameter of 0.8 mm, Ti current collector thickness of 20 µm, He carrier gas differential pressure of 5.0 × 10⁵ Pa, and nozzle–substrate distance of 5 mm. Other fabrication conditions were described elsewhere.^18,19 The deposition weight of TiO₂ was 40 µg, and the TiO₂ film thickness was approximately 4–5 µm. When the TiO₂ phase change was studied by XRD, electrodes were fabricated by a typical slurry coating process because a large amount of the active material was required to generate XRD peaks. First, we prepared 1.0 g of a mixture consisting of TiO₂ (70 wt%), acetylene black (15 wt%), sodium carboxymethylcellulose (10 wt%), and styrene–butadiene rubber (5 wt%). This mixture was added to 4.0 mL of hot water used as a solvent followed by mixing via ball milling to form a slurry. The slurry was coated onto Cu foil with a thickness of 18 µm, and the produced electrode was dried at 120 °C. The TiO₂ coverage was approximately 5.0 mg cm⁻².

A total of 2032 coin-type cells were assembled under a dry argon atmosphere using lithium metal as the counter electrode and a glass microfiber filter as the separator. Lithium bis(trifluoromethanesulfonyl)amide (LiTFSA) (1 M = 1 mol L⁻¹) dissolved in propylene carbonate (PC; C₄H₆O₃, Kishida Chemical Co., Ltd.) was used as the electrolyte. Galvanostatic testing was conducted at 30 °C in the potential range of 1.0–3.0 V or 1.3–3.0 vs. Li⁺/Li. The current density was set to 35 or 335 mA g⁻¹, which corresponded to 0.1C or 1C, respectively. Electrochemical impedance spectroscopy (EIS) measurements were performed using an impedance analyzer (CompactStat.h 20250e, Ivium Technologies). Measurements were conducted after the first charge–discharge cycle at 3.0 V vs. Li⁺/Li with a potential amplitude of 10 mV in the frequency range of 10 mHz–500 kHz. Nyquist plots were constructed using a Randles equivalent circuit including solution resistance (R_sol), charge transfer resistance (R_ct), electric double layer capacitance (C_dl), and Warburg impedance (Z_W).

To evaluate the applicability of a Na-ion battery (NIB) anode, we assembled a coin cell from Cu-doped TiO₂ electrode, Na metal counter electrode, and 1.0 M sodium bis(fluorosulfonyl)amide (NaFSA)-dissolved/PC electrolyte solution. The charge–discharge properties of the produced cell were determined in the potential range of 0.005–3.0 V vs. Na⁺/Na at a temperature of 30 °C and current density of 50 mA g⁻¹ (0.15C).

The crystal phase change was studied by XRD. After performing ten and three charge–discharge cycles of the undoped TiO₂ electrode and Cu-doped TiO₂ electrode, respectively, the coin cells were disassembled to remove the electrodes. To increase the degree of material crystallinity, a post-annealing treatment was conducted at 800 °C for 20 min in a vacuum of 10⁻⁵ Pa using a turbomolecular pumping system (Desktop YTP, ULVAC) and a tube-type electric furnace (Full-tech, FT-02VAC-50). After the annealed TiO₂ active material was mounted on an Al substrate as an internal standard, XRD measurements were performed.

3. Results and Discussion

3.1 TiO₂ phase change

Figure 1 shows the results of galvanostatic charge–discharge measurements of the undoped TiO₂ electrodes cycled at a current density of 33.5 mA g⁻¹ (0.1C). When lithiation was limited to 1.3 V vs. Li⁺/Li, the TiO₂ electrode exhibited a small discharge (Li extraction) capacity of 124 mAh g⁻¹, while the discharge curves were in good agreement in the initial three cycles (Fig. 1a). The differential capacity (dQ/dV) plots contain a broad peak at 1.8 V vs. Li⁺/Li and a sharp peak at 1.5 V vs. Li⁺/Li in the cathodic profile (Fig. 1b). The first peak originated from the solid solution reaction of rutile TiO₂ (Li_xTiO₂, 0 < x < 0.2).^10–16 The second peak can be ascribed to the reversible phase change of rutile TiO₂ (P4₂/mnm) to monoclinic distorted rutile Li_xTiO₂ (P2/m_RUT).²⁴ In the anodic profile, these reverse reactions occurred at 1.4 and 1.8 V vs. Li⁺/Li, respectively. When lithiation was continued up to 1.0 V vs. Li⁺/Li, a large charge (Li insertion) capacity of 506 mAh g⁻¹ was obtained in the first cycle (Fig. 1c), which exceeded the theoretical capacity of TiO₂ (335 mAh g⁻¹). It was found previously that not only the Li insertion into the TiO₂ lattice but also the pseudocapacitive Li-storage on the TiO₂ surface contributed to the charge capacity.²⁴ A large slope in the low potential region obtained during the first lithiation process is a typical feature of rutile TiO₂ electrodes.^{10–16,18–20} The initial discharge capacity was 257 mAh g⁻¹, and the charge–discharge curves of the second and third cycles were fully consistent with each other, indicating high reversibility of the electrode reactions. The electrode capacity was maintained at 220 mAh g⁻¹ until the 10th cycle (Fig. S3), while the dQ/dV plots confirmed the presence of two cathodic peaks at 1.5 and 1.15 V vs. Li⁺/Li in the initial cycle. Moreover, the peak at 1.15 V was attributed to the irreversible phase change from the distorted rutile Li_xTiO₂ to monoclinic distorted α-NaFeO₂-type (layered rock-salt) Li_xTiO₂ (P2/m_HEX)²³ or disordered cubic rock-salt LiTiO₂ ($Fm\bar{3}m$).^24,25 Hence, the exact phase change mechanism remained controversial. Consequently, our objective was to experimentally confirm the phase composition after the rutile TiO₂ transformation via the charge–discharge reactions in the potential range of 1.0–3.0 V.

Figure 1.

(a) Charge–discharge curves and (b) differential capacity plots (dQ/dV) obtained during the initial three cycles for the undoped TiO₂ electrode in the potential range of 1.3–3.0 V vs. Li⁺/Li using 1.0 M LiTFSA/PC electrolyte at a current density of 33.5 mA g⁻¹ (0.1C). (c) Charge-discharge curves and (d) dQ/dV plots of the TiO₂ electrode cycled in the potential range of 1.0–3.0 V vs. Li⁺/Li.

Figure 2a shows the XRD patterns of the undoped TiO₂ electrodes recorded before and after delithiation in the 10th cycle. No XRD peak appeared without a post-annealing due to a very low crystallinity, whereas XRD peaks appeared and their intensities increased with increasing the post-annealing temperature of the electrode (Fig. S4). By the annealing at 800 °C, the diffraction peaks of the distorted layered rock-salt Li_xTiO₂ structure clearly appeared (Fig. 2b). Borghols et al. reported that lithiated rutile TiO₂ (Li_0.85TiO₂) had a crystal structure with the space group P2/m_HEX,¹⁰ whose atomic arrangement closely resembled that of LiTiO₂ with the hexagonal space group $R\bar{3}m$. Christensen et al. suggested that the distorted layered Li_xTiO₂ underwent a solid solution reaction. The crystal structure of the distorted layered Li_xTiO₂ was fully maintained during the charge–discharge cycles.²³ In this study, the XRD pattern of delithiated TiO₂ matched that of the distorted layered Li_xTiO₂ (P2/m_HEX) phase.^10,23 However, unlike Christensen’s study,²³ no columbite phase (Li_xTiO₂) was formed. These results demonstrate that the crystal phase changed from rutile TiO₂ (Fig. 2c) to layered rock-salt Li_xTiO₂ (Fig. 2d) via the charge–discharge reactions. Note that the analyzed crystal structures were created using the Visualization for Electronic and Structural Analysis (VESTA) package developed by K. Momma and F. Izumi.³⁴ The distorted layered Li_xTiO₂ has basically same crystal structure the lithiated rutile TiO₂ (Li_0.85TiO₂) reported by Borghols.¹⁰ The composition of the lithiated rutile TiO₂ (Li_0.85TiO₂) can be estimated as [Li⁺]_0.85[Ti⁴⁺]_0.15[Ti³⁺]_0.85[O²⁻]₂ on the assumption that there is no oxygen deficiency. However, it should be noted that Ti⁴⁺ amount will be increased with decreasing Li amount (x) in Li_xTiO₂ because Ti is oxidized.

Figure 2.

(a) XRD patterns of the undoped TiO₂ electrodes recorded before charge–discharge cycling and after the delithiation at the 10th cycle conducted at 0.1C in the potential range of 1.0–3.0 V vs. Li⁺/Li. (b) An enlarged view of the XRD peak centered at 18.4°. Crystal structures of (c) rutile TiO₂ and (d) layered rock-salt Li_xTiO₂ (Li_0.85TiO₂).

Figure 3 displays the results of TEM observations of the undoped TiO₂ electrodes. Before the charge–discharge cycle, crystallites with sizes of several nanometers were detected (Fig. 3a), and two types of lattice fringes orthogonal to each other were observed (Fig. 3b). Their interlayer spacings were 0.257 and 0.227 nm, which were consistent with the lattice spacings of the (2 0 2) and (0 2 0) planes in the rutile TiO₂ structure, respectively. After the first lithiation cycle, crystallites with a larger size were formed (Fig. 3c), and the interlayer spacings of the lattice fringes were equal to 0.226 nm, which corresponded to the lattice spacing of the (4 0 0) plane in the distorted layered rock-salt Li_xTiO₂ structure (Fig. 3d). These results suggested that the atoms of the rutile phase rearranged to produce Li_xTiO₂ crystallites upon lithiation. After the first delithiation process, crystallites with similar sizes were detected (Fig. 3e), while the observed lattice fringes had interlayer spacings of 0.226 and 0.203 nm (Fig. 3f), which were consistent with the lattice spacings of the (4 0 0) and (3 1 $\bar{5}$) planes in layered Li_xTiO₂, respectively. Therefore, the TEM observations clearly confirmed the phase change from rutile TiO₂ to distorted rock-salt Li_xTiO₂.

Figure 3.

TEM images of the undoped TiO₂ electrodes obtained (a, b) before the charge-discharge cycles, (c, d) after the first lithiation process, and (e, f) after the first delithiation process. The lattice fringes corresponded to the interlayer spacings of rutile TiO₂ and layered rock-salt Li_xTiO₂.

3.2 Effect of impurity doping on the properties of the Li_xTiO₂ phase

We also investigated the effect of impurity (M) doping on TiO₂ structural properties. Table 1 lists the lattice parameters (a and c) of the rutile phase and oxygen vacancy amounts y in M_xTiO_2−y species. When Cu²⁺ and Nb⁵⁺ were introduced into the TiO₂ structure, the lattice parameter a increased because their sizes were larger than that of the Ti⁴⁺ ion. In contrast, the lattice parameter a decreased after Al³⁺ doping because of the smaller size of the Al³⁺ ion. These results confirmed that the introduced impurity elements formed a substitutional solid solution. By doping impurity ions (Cu²⁺ and Al³⁺) with lower valences than that of Ti⁴⁺, oxygen vacancies were expected to be introduced into the TiO₂ lattice due to charge compensation. The obtained oxygen amounts revealed that a small amount of oxygen vacancies (y = 0.10) was present even in undoped TiO₂. After Nb⁵⁺ doping, the oxygen vacancy amount y remained almost the same. Meanwhile, the value of y apparently increased to 0.26 and 0.31 after doping Cu²⁺ and Al³⁺ ions, respectively. These magnitudes were very close, although the doping amount of Cu²⁺ (1 at%) was much smaller than that (3 at%) of Al³⁺ due to the lower valence of Cu²⁺ ions. To confirm the electrical conductivity improvement caused by oxygen vacancies, electrical resistivity measurements were performed for pressed TiO₂ powders. The electrical conductivity of undoped TiO₂, 1 at% Cu-doped TiO₂, and 4 at% Nb-doped TiO₂ were 4.0 × 10⁻², 4.5 × 10⁻¹, and 9.0 × 10⁻² S cm⁻¹, respectively, whereas the conductivity of 3 at% Al-doped TiO₂ could not been measured due to particle aggregation (Fig. S2). Moreover, Cu-doped TiO₂ exhibited a higher conductivity than that of Nb-doped TiO₂ even at a smaller doping amount. Hence, the larger number of oxygen vacancies increased the electronic conductivity of TiO₂.

Figure 4 compares the charge–discharge curves and dQ/dV plots of the undoped TiO₂ electrode and Cu-doped TiO₂ electrodes cycled in the potential range of 1.0–3.0 V vs. Li⁺/Li at 1.0C. The results obtained for the Al-doped TiO₂ electrode and Nb-doped TiO₂ electrode are presented in Fig. S5. The initial discharge capacity was 72 mAh g⁻¹ (Fig. 4a), which was much lower than the value of 257 mAh g⁻¹ obtained at 0.1C. In the dQ/dV plots, a cathodic peak at 1.1 V vs. Li⁺/Li was observed, which was ascribed to the two-phase reaction of distorted rutile Li_xTiO₂ and distorted layered Li_xTiO₂. Meanwhile, we could not clearly detect a cathodic peak at 1.5 V vs. Li⁺/Li. There is one possible reason for the absence of this peak at 1.5 V. The reason is the ten times higher current density of 1.0C than 0.1C. Due to fast Li⁺ kinetics, the electrode reaction of rutile Li_xTiO₂–distorted rutile Li_xTiO₂ at 1.5 V is restricted. In contrast, the Cu-doped TiO₂ electrode produced a cathodic peak at 1.4 V vs. Li⁺/Li in the dQ/dV plots (Fig. 4b), suggesting that Cu-doped TiO₂ could strongly react with Li⁺ ions even at a higher current rate of 1.0C. This electrode exhibited a high discharge capacity of 280 mAh g⁻¹ during the first cycle. Thus, we investigated the effect of Cu doping on the crystal structure, electrical conductivity, and anode performance of TiO₂ after the phase change.

Figure 4.

Charge–discharge curves and differential capacity plots (dQ/dV) obtained during the initial three cycles for the (a) undoped TiO₂ electrode and (b) Cu-doped TiO₂ electrode. The charge–discharge tests were performed in the potential range of 1.0–3.0 V vs. Li⁺/Li using 1.0 M LiTFSA/PC electrolyte at a current density of 335 mA g⁻¹ (1C).

The crystal structure of Cu-doped TiO₂ was first studied by XRD. Figure 5a displays the XRD pattern of the Cu-doped TiO₂ electrode obtained after three cycles. For comparison, the figure also shows the pattern of the undoped TiO₂ electrode recorded after ten cycles. The peaks of the distorted layered rock-salt Li_xTiO₂ emerged during the third cycle, while those of the remaining rutile phase appeared because the TiO₂ phase change was not complete yet. Figure 5b compares the peak positions of layered rock-salt Li_xTiO₂ at 18.4°, indicating that the peak of Cu-doped TiO₂ shifted to lower angles. From these results, we verified for the first time the presence of a substitutional solid solution of impurity elements in layered rock-salt Li_xTiO₂. The obtained XRD results revealed that the lattice parameters a, b, and c in undoped monoclinic layered Li_xTiO₂ were equal to 0.4803, 0.2980, and 0.5154 nm, respectively, and that Cu-doped TiO₂ had larger lattice parameters (a = 0.4826 nm, b = 0.2986 nm, and c = 0.5161 nm). The interlayer spacing, c sin(β), was 0.5079 nm at β = 79.77° for the monoclinic layered Li_xTiO₂ structure.²³ This value exceeded the interlayer distance of 0.5072 nm obtained for undoped TiO₂ though the difference was very small. As illustrated in Fig. 5c, Li⁺ diffuses in the two-dimensional direction between the layers composed of TiO₆ octahedra. Thus, Cu doping made the crystal structure more suitable for Li⁺ diffusion by expanding the interlayer distance because of its larger size of the Cu²⁺ ion.

Figure 5.

(a) XRD pattern of the Cu-doped TiO₂ electrode recorded at the delithiation state after three cycles conducted at 0.1C in the potential range of 1.0–3.0 V vs. Li⁺/Li. For comparison, the figure also shows the pattern of the undoped TiO₂ electrode obtained after ten cycles. (b) Effect of Cu doping on the XRD peak position of layered rock-salt Li_xTiO₂ at 18.4°. (c) Schematic illustration of the interlayer spacing of the layered rock-salt Li_xTiO₂.

The electrical conductivity of Cu-doped TiO₂ was determined by EIS. Figure 6a shows the Nyquist plots of the Cu-doped and undoped TiO₂ electrodes obtained in the discharged state at 3.0 V vs. Li⁺/Li during the third cycle. In the high-frequency region, one semicircle was observed. On the other hand, we confirmed one semicircle in the charged state at 1.0 V (Fig. S6). The semicircle corresponds to the charge transfer resistance (R_ct) of the delithiation reaction, and the x-axis intercept shows the solution resistance (R_sol) (Fig. 6b). The R_sol of Cu-doped TiO₂ was identical to that of undoped TiO₂. Furthermore, Cu-doped TiO₂ exhibited a lower R_ct value than that of undoped TiO₂ in the discharged state. We have concluded that Cu doping likely increased the electronic conductivity of TiO₂ due to the presence of oxygen vacancies in delithiated layered Li_xTiO₂. In contrast, no increase in R_ct was observed in the charged state at 1.0 V (Fig. S6), indicating that the delithiated layered Li_xTiO₂ had a semiconductive electronic structure, while the lithiated phase possessed a metallic electronic structure.

Figure 6.

(a) Nyquist plots of the Cu-doped TiO₂ electrode and undoped TiO₂ electrode obtained in the discharged (delithiated) state of the third cycle. (b) Equivalent circuit used for fitting analysis, which includes the electrolyte solution resistance (R_sol), charge-transfer resistance (R_ct), Warburg impedance (Z_W), and capacitance of the electric double layer (C_dl). (c) Fitting results.

The anode performance was evaluated by charge–discharge cycling at 1.0C. Figure 7 shows the cycling performance of the Cu-doped TiO₂ electrode. The small capacity decay observed during the initial four cycles is attributed to the incomplete phase change from rutile TiO₂ to layered rock-salt Li_xTiO₂. After the 10th cycle, the Cu-doped TiO₂ electrode exhibited outstanding cyclability (a discharge capacity of 230 mAh g⁻¹ was maintained up to the 300th cycle). The layered rock-salt Li_xTiO₂ structure underwent a small volumetric expansion of only 1 % during lithiation.²³ The outstanding cyclability of the Cu-doped TiO₂ electrode is likely due to the expanded Li⁺ diffusion pathway, higher electronic conductivity, and small volume changes. We consider that the conductivity enhancement is more dominant for improving cyclability than the diffusion pathway expansion because the increase in the interlayer distance of Cu-doped TiO₂ is very small. Meanwhile, the Nb-doped and Al-doped TiO₂ electrodes exhibited inferior performance characteristics. The electronic conductivity of Nb-doped TiO₂ did not increase as much as that of Cu-doped TiO₂ because the amount of oxygen vacancies in the former structure could not be increased. Furthermore, the capacity of Al-doped TiO₂ was lowered due to particle aggregation despite the large number of oxygen vacancies.

Figure 7.

Cycling performance of the Cu-doped TiO₂ electrode evaluated at 1.0C. For comparison, the figure also shows the data obtained for the Al-doped TiO₂ and Nb-doped TiO₂ electrodes.

3.3 Cu-doped TiO₂ as a novel anode material

Figure 8 shows the rate capability of the Cu-doped TiO₂ electrode cycled at current densities ranging from 0.1C (33.5 mA g⁻¹) to 100C (33.5 A g⁻¹). The observed rate performance trend was identical to that of the cycle performance. The Cu-doped TiO₂ electrode exhibited the best rate capability with discharge capacities of 140, 108, and 87 mAh g⁻¹ at 20C, 50C, and 100C, respectively. In terms of the rate capability, the performance of Cu-doped TiO₂ was superior to that of a Li₄Ti₅O₁₂ anode reported by Tarascon.³⁵ Moreover, the Li_xTiO₂ electrode demonstrated excellent rate performance due to the contribution of the capacitive process,²⁴ which could be further boosted by Cu doping. The long-term cyclability of the Cu-doped TiO₂ electrode was subsequently evaluated at 5C (1.675 A g⁻¹) after the rate capability test. Even after a very long period of 10000 cycles, the Cu-doped TiO₂ electrode maintained a capacity of 120 mAh g⁻¹ (Fig. 9), indicating its structural robustness caused by the small volumetric change of only 1 % during the charge–discharge cycling of the layered Li_xTiO₂ phase. The charge–discharge cycling test was carried out for the electrode prepared by the conventional slurry-coating method, whereas XRD analysis was conducted for the electrode prepared by GD method. The authors have confirmed that these two types of Nb-doped TiO₂ electrodes exhibit similar anode properties.³⁶ It is thus considered that the phase change and the performance improvement occur in the same way in Cu-doped TiO₂ regardless of the electrode preparation method.

Figure 8.

Rate capabilities of various impurity-doped TiO₂ electrodes determined at various current rates between 0.1C and 100C.

Figure 9.

Long-term cycling performance of the Cu-doped TiO₂ electrode evaluated at a high current density of 1.675 A g⁻¹ (5C).

Rutile TiO₂ is a promising material not only as an LIB anode but also as an NIB anode because Na insertion and Na extraction are reversible processes.^18,19 In this work, we evaluated the applicability of the Cu-doped TiO₂ electrode as an NIB anode at 0.15C (50 mA g⁻¹). The electrode underwent reversible charge–discharge reactions (Fig. 10a), while Cu doping significantly improved the NIB anode performance; as a result, a high discharge capacity of 300 mAh g⁻¹ was achieved after the second cycle (Fig. 10b), while the undoped TiO₂ electrode demonstrated a capacity of approximately 200 mAh g⁻¹. Although the phase change of TiO₂ in NIBs has not been sufficiently elucidated, the obtained results indicated that Cu doping drastically improved the NIB anode performance. In the next study, we will examine in detail the phase changes occurring in NIBs and the effect of impurity doping on NIB characteristics.

Figure 10.

(a) Charge–discharge curve of the 1 at% Cu-doped TiO₂ electrode utilized as a NIB anode in an electrolyte containing 1.0 M sodium bis(fluorosulfonyl)imide (NaFSA) solution in PC during the second cycle at a current density of 50 mA g⁻¹ (0.15C). (b) Discharge capacity dependence on the cycle number of the Cu-doped TiO₂ electrode.

4. Conclusions

In this study, we investigated the TiO₂ crystal structure after the phase change induced by lithiation and the influence of impurity doping on LIB anode properties. When Cu²⁺ ions were introduced into the TiO₂ lattice the oxygen vacancy amount increased due to charge compensation, which increased the electronic conductivity of the material. The obtained XRD data confirmed that the undoped rutile TiO₂ phase was transformed into a distorted layered rock-salt Li_xTiO₂ structure. In addition, we verified for the first time the formation of a substitutional solid solution of Cu²⁺ ions in the layered rock-salt Li_xTiO₂ lattice. Cu doping made the TiO₂ crystal structure more suitable for Li⁺ diffusion by expanding the interlayer distance because of the relatively large size of Cu²⁺ ions. The results of EIS analysis revealed that Cu-doped TiO₂ possessed a lower charge transfer resistance compared with that of undoped TiO₂ and higher electronic conductivity due to the presence of oxygen vacancies in layered Li_xTiO₂. Furthermore, the Cu-doped TiO₂ electrode exhibited outstanding cyclability because a discharge capacity of 230 mAh g⁻¹ was maintained up to the 300th cycle, which was attributed to the expanded Li⁺ diffusion pathway, higher electronic conductivity, and small volume changes. We also confirmed an excellent rate capability of 108 mAh g⁻¹ at 50C and a very long cycle life with 120 mAh g⁻¹ at 5C for 10000 cycles. In addition, the Cu-doped TiO₂ electrode can be potentially used as a NIB anode due to its high discharge capacity of 300 mAh g⁻¹.

Acknowledgments

This study was partially supported by the Japan Society for the Promotion of Science (JSPS) KAKENHI (grant numbers 19H02817, 19K05649, 20H00399, and JP18K04965) and the National Institute for Materials Science (NIMS) Joint Research Hub Program. It was also supported by the Adaptable and Seamless Technology Transfer Program through Target-driven R&D (A-STEP) from the Japan Science and Technology Agency (JST) (grant number JPMJTR20T2). The TEM measurements were performed at the NIMS Battery Research Platform. The author greatly thanks the two reviewers for their prompt and very educational suggestions.

Data Availability Statement

The data that support the findings of this study are openly available under the terms of the designated Creative Commons License in J-STAGE Data at https://doi.org/10.50892/data.electrochemistry.18843530.

CRediT Authorship Contribution Statement

Hiroyuki Usui: Conceptualization (Lead), Project administration (Lead), Writing – original draft (Lead)

Yasuhiro Domi: Data curation (Lead), Formal analysis (Lead)

Thi Hay Nguyen: Investigation (Lead), Doping experiments (Lead)

Shin-ichiro Izaki: Methodology (Lead)

Kei Nishikawa: TEM observations (Lead)

Toshiyuki Tanaka: Oxygen analysis (Lead)

Hiroki Sakaguchi: Supervision (Lead)

Conflict of Interest

The authors declare no conflict of interest in the manuscript.

Funding

Japan Society for the Promotion of Science: 19H02817

Japan Society for the Promotion of Science: 19K05649

Japan Society for the Promotion of Science: 20H00399

Japan Society for the Promotion of Science: JP18K04965

Adaptable and Seamless Technology Transfer Program through Target-Driven R and D: JPMJTR20T2

Footnotes

H. Usui, Y. Domi, K. Nishikawa, and H. Sakaguchi: ECSJ Active Members

T. H. Nguyen and S.-i. Izaki: ECSJ Student Members

References

• 1)  M. Wagemaker and F. M. Mulderv, Acc. Chem. Res., 46, 1206 (2013).
• 2)  V. Aravindan, Y.-S. Lee, R. Yazami, and S. Madhavi, Mater. Today, 18, 345 (2015).
• 3)  Y.-S. Hu, L. Kienle, Y.-G. Guo, and J. Maier, Adv. Mater., 18, 1421 (2006).
• 4)  M. Wagemaker, W. J. H. Borghols, and F. M. Mulder, J. Am. Chem. Soc., 129, 4323 (2007).
• 5)  J. Ye, W. Liu, J. Cai, S. Chen, X. Zhao, H. Zhou, and L. Qi, J. Am. Chem. Soc., 133, 933 (2011).
• 6)  T. Beuvier, M. Richard-Plouet, M.-L. Granvalet, T. Brousse, O. Crosnier, and L. Brohan, Inorg. Chem., 49, 8457 (2010).
• 7)  A. G. Dylla, G. Henkelman, and K. J. Stevenson, Acc. Chem. Res., 46, 1104 (2013).
• 8)  D. Dambournet, K. W. Chapman, M. V. Koudriachova, P. J. Chupas, I. Belharouak, and K. Amine, Inorg. Chem., 50, 5855 (2011).
• 9)  K. Mukai and I. Yamada, J. Electrochem. Soc., 164, A3590 (2017).
• 10)  W. J. H. Borghols, M. Wagemaker, U. Lafont, E. M. Kelder, and F. M. Mulder, Chem. Mater., 20, 2949 (2008).
• 11)  E. Baudrin, S. Cassaignon, M. Koelsch, J.-P. Jolivet, L. Dupont, and J.-M. Tarascon, Electrochem. Commun., 9, 337 (2007).
• 12)  D. Wang, D. Choi, Z. Yang, V. V. Viswanathan, Z. Nie, C. Wang, Y. Song, J.-G. Zhang, and J. Liu, Chem. Mater., 20, 3435 (2008).
• 13)  M. Vijayakumar, S. Kerisit, C. Wang, Z. Nie, K. M. Rosso, Z. Yang, G. Graff, J. Liu, and J. Hu, J. Phys. Chem. C, 113, 14567 (2009).
• 14)  M.-C. Yang, Y.-Y. Lee, B. Xu, K. Powers, and Y. S. Meng, J. Power Sources, 207, 166 (2012).
• 15)  J. Chen, W. Song, H. Hou, Y. Zhang, M. Jing, X. Jia, and X. Ji, Adv. Funct. Mater., 25, 6793 (2015).
• 16)  X.-Y. Yu, H. B. Wu, L. Yu, F.-X. Ma, and X. W. Lou, Angew. Chem., Int. Ed., 54, 4001 (2015).
• 17)  H. Usui, K. Wasada, M. Shimizu, and H. Sakaguchi, Electrochim. Acta, 111, 575 (2013).
• 18)  H. Usui, S. Yoshioka, K. Wasada, M. Shimizu, and H. Sakaguchi, ACS Appl. Mater. Interfaces, 7, 6567 (2015).
• 19)  H. Usui, Y. Domi, S. Yoshioka, K. Kojima, and H. Sakaguchi, ACS Sustain. Chem. Eng., 4, 6695 (2016).
• 20)  H. Usui, Y. Domi, S. Ohnishi, N. Takamori, S. I. Izaki, N. Morimoto, K. Yamanaka, K. Kobayashi, and H. Sakaguchi, ACS Mater. Lett., 3, 372 (2021).
• 21)  M. Lübke, I. Johnson, N. M. Makwana, D. Brett, P. Shearing, Z. Liu, and J. A. Darr, J. Power Sources, 294, 94 (2015).
• 22)  D. Imazeki, C. C. van Gils, K. Nishio, R. Shimizu, and T. Hitosugi, ACS Appl. Energy Mater., 3, 8338 (2020).
• 23)  C. K. Christensen, M. A. H. Mamakhel, A. R. Balakrishna, B. B. Iversen, Y.-M. Chiang, and D. B. Ravnsbæk, Nanoscale, 11, 12347 (2019).
• 24)  P. Díaz-Carrasco, A. Duarte-Cárdenas, A. Kuhn, and F. García-Alvarado, J. Power Sources, 515, 230632 (2021).
• 25)  H.-D. Yang, Y.-Y. Kang, P.-P. Zhu, Q.-W. Chen, L. Yang, and J.-P. Zhou, J. Alloys Compd., 872, 159759 (2021).
• 26)  H. Usui, Y. Domi, S. Ohnishi, and H. Sakaguchi, ACS Appl. Nano Mater., 2, 5360 (2019).
• 27)  W. Wang, M. Wu, P. Han, Y. Liu, L. He, Q. Huang, J. Wang, W. Yan, L. Fu, and Y. Wu, ACS Appl. Mater. Interfaces, 11, 3061 (2019).
• 28)  H. Usui, S. Toriu, Y. Domi, T. Tanaka, and H. Sakaguchi, J. Energy Storage, 44, 103497 (2021).
• 29)  H. Usui, Y. Domi, S. Ohnishi, and H. Sakaguchi, ACS Appl. Nano Mater., 2, 5360 (2019).
• 30)  H. Sakaguchi, T. Toda, Y. Nagao, and T. Esaka, Electrochem. Solid-State Lett., 10, J146 (2007).
• 31)  Y. Domi, H. Usui, K. Nishikawa, and H. Sakaguchi, ACS Appl. Nano Mater., 4, 8473 (2021).
• 32)  J. Akedo, J. Am. Ceram. Soc., 89, 1834 (2006).
• 33)  J. Akedo, J. Thermal Spray Tech., 17, 181 (2008).
• 34)  K. Momma and F. Izumi, J. Appl. Crystallogr., 44, 1272 (2011).
• 35)  A. S. Prakash, P. Manikandan, K. Ramesha, M. Sathiya, J.-M. Tarascon, and A. K. Shukla, Chem. Mater., 22, 2857 (2010).
• 36)  H. Usui, Y. Domi, M. Shimizu, A. Imoto, K. Yamaguchi, and H. Sakaguchi, J. Power Sources, 329, 428 (2016).