Effect of Nb-doping on electrochemical stability of Li₄Ti₅O₁₂ discharged to 0 V

Abstract

Li₄Ti_4.95Nb_0.05O₁₂ is synthesized by a citric acid-assistant sol–gel method. X-ray diffraction (XRD) reveals that highly crystalline Li₄Ti_4.95Nb_0.05O₁₂ without any impurity is obtained. The electrochemical performances of the Li₄Ti_4.95Nb_0.05O₁₂ and the Li₄Ti₅O₁₂ in the range from 0 to 2.5 V are investigated. The Li₄Ti_4.95Nb_0.05O₁₂ presents a higher specific capacity and better cycling stability than the Li₄Ti₅O₁₂ due to the improved conductivity. The Li₄Ti_4.95Nb_0.05O₁₂ exhibits a capacity as high as 231.2 mAh g⁻¹ after 100 cycles, which is much higher than the Li₄Ti₅O₁₂ (111.1 mAh g⁻¹). The effect of Nb-doping on electrochemical performance of Li₄Ti₅O₁₂ discharged to 0 V has also been discussed.

Introduction

Spinel Li₄Ti₅O₁₂ has been attracted more attention as a promising anode material for lithium-ion batteries [1–6]. It has many advantages compared to the currently used graphite, such as an excellent lithium-ion insertion/extraction reversibility and very flat voltage plateau at around 1.55 V vs. Li⁺/Li. However, Li₄Ti₅O₁₂ exhibits poor electronic and lithium ionic conductivities, which result in its poor electrochemical performance and prevent Li₄Ti₅O₁₂ from being implemented commercially. In order to improve the conductivity, three methods were proposed, including synthesis of nano-sized Li₄Ti₅O₁₂ particles [7–14], incorporation of good conductive phase of metal powder or carbon [15–20], substituting Li or Ti with metal ions. So far, a lot of metal ions such as Cr³⁺, V⁵⁺, Mn⁴⁺, Fe³⁺, Al³⁺, Ga³⁺, Co³⁺, Cu²⁺, and Ta⁵⁺ were doped into Li₄Ti₅O₁₂ and improved electrochemical performances were observed [21–27].

Recently, the electrochemical behaviors of Li₄Ti₅O₁₂ discharged to 0 V have been widely investigated [28–32]. Ge et al. [28] reported the theoretical capacity of Li₄Ti₅O₁₂ in the voltage range from 2.5 to 0.01 V was 293 mAh g⁻¹, which was calculated on basis of full transition from Ti⁴⁺ to Ti³⁺. Xiang and his coworkers [31] disclosed that the additional capacity of Li₄Ti₅O₁₂ below 1 V could be used to enhance the safety characteristic of the 3 V LiNi_0.5Mn_1.5O₄/Li₄Ti₅O₁₂ cell when the capacity was limited by the Li₄Ti₅O₁₂ anode. So, it is very important to investigate the electrochemical properties and cell performances of Li₄Ti₅O₁₂ discharged to 0 V in order to improve the energy density and safety characteristic of the corresponding batteries. However, from Shu’s [29] and Borghols’ [31] reports, it can be easily found that Li₄Ti₅O₁₂ shows a much rapid capacity loss when discharged to 0 V with increasing the cycling number, especially at a high current rate. Borghols and his coworkers [31] reported the size effect in the Li_4+x Ti₅O₁₂ spinel and concluded that the reversible intercalation and extraction of Li⁺ between Li₄Ti₅O₁₂ and Li_8.5Ti₅O₁₂ were affected by the size of the material. In our previous work, we developed a novel Nb-doping compound, i.e., Li₄Ti_4.95Nb_0.05O₁₂, which possessed better electrochemical performances than Li₄Ti₅O₁₂ due to the improved electronic conductivity when discharged to 1 V [32]. In this paper, the effect of Nb-doping on the electrochemical performance of Li₄Ti₅O₁₂ discharged to 0 V was investigated.

Experimental

Li₄Ti_4.95Nb_0.05O₁₂ was synthesized by a sol–gel method with citric acid (AR, Sinopharm Chemical Reagent Co., Ltd., China) as a chelating agent. The stoichiometric amounts of lithium acetate (CH₃COOLi, 99%, Shanghai China Lithium Industrial Co., Ltd. China), tetrabutyl titanate (Ti(OC₄H₉)₄, ≥98%, Shanghai Lingfeng Chemical Reagent Co., Ltd., China) and niobium hydroxide (Nb(OH)₅, ≥99.5%, Guangzhou Litop Non-ferrous Metals Co., Ltd. China) were added as starting materials in ethanol to form a sol. A gel was obtained after the sol was dried at 80 °C to evaporate ethanol. The obtained gel was dried at 120 °C over 10 h to evaporate excess ethanol and yield organic precursor. Furthermore, the precursor was calcined in an oven at 500 °C for 2 h in air and then calcined in the oven at 850 °C for 10 h in N₂ atmosphere to obtain the final powder. The color of the powder is light blue. For comparison, the Li₄Ti₅O₁₂ powder was also prepared using the similar method, while the product is white. The crystal structures of the powders were characterized by X-ray diffraction (XRD, Rigaku DMax-RB) using Cu-Kα radiation (10° ≤ 2θ ≤ 80°).

The electrochemical properties of Li₄Ti_4.95Nb_0.05O₁₂ and Li₄Ti₅O₁₂ were tested in CR2025 cells with lithium as counter electrode. A mixed solvent of ethylene carbonate (EC) and dimethyl carbonate (1:1 w/w) containing 1.0 mol dm⁻³ LiPF₆ was used as electrolyte and Celgard 2325 microporous membrane as separator. The electrodes contained active material, carbon black and polyvinylidene fluoride (82 : 10 : 8, by weight). The cell was assembled in an argon filled glove box (Mikrouna, super 1220), where the oxygen and moisture contents were less than 1 ppm. All the electrochemical tests were carried out at room temperature. Charge–discharge cycling tests were performed at indicated current densities using a NEWARE Battery Testing System.

Results and discussion

Figure 1 shows the XRD patterns of Li₄Ti₅O₁₂ and Li₄Ti_4.95Nb_0.05O₁₂ powders. The diffraction peaks of the powders conform to JCPDS card No. 49-0207, indicating that the synthesized powders are in accordance with the spinel Li₄Ti₅O₁₂ standard without any impurity detected. In our previous work, we have validated that the Nb ions have been successfully incorporated into the lattice structure of Li₄Ti₅O₁₂ [32]. As a result, the lattice constant is slightly enlarged, which is beneficial for fast lithium-ion transfer without lattice stability damaged.

Fig. 1

XRD patterns of Li₄Ti₅O₁₂ and Li₄Ti_4.95Nb_0.05O₁₂

The first, second, and third discharge–charge curves of Li₄Ti₅O₁₂ discharged to 0 V at the current density of 350, 875, and 1,750 mA g⁻¹, and discharged to 1 V at the current density of 1,750 mA g⁻¹ are shown in Fig. 2. The initial discharge capacity of Li₄Ti₅O₁₂ decreases with the current densities increasing. When the current density is increased from 350 to 1,750 mA g⁻¹, the second and third discharge plateaus become slowly inconspicuous. However, Li₄Ti₅O₁₂ discharged to 1 V at the current density of 1,750 mA g⁻¹ shows an obvious discharge plateau. In order to explain the different discharge plateau characteristic for Li₄Ti₅O₁₂ discharged to 1 and 0 V at the current density of 1,750 mA g⁻¹, a sketch map of lithium-ion insertion and extraction for Li₄Ti₅O₁₂ discharged to 1 V and 0 V is proposed in Fig. 3. For discharging down to 1 V at the high current density, the center of Li₄Ti₅O₁₂ particle has not totally transformed to Li₇Ti₅O₁₂ (Fig. 3a). But when being charged to 2.5 V, the original Li₄Ti₅O₁₂ can be resumed, because lithium extraction is easier kinetically than lithium insertion in Li₄Ti₅O₁₂ due to the repulsive interactions between neighboring lithium ions during lithium insertion [33]. However, when discharged to 0 V, Li₄Ti₅O₁₂ firstly transforms to Li₇Ti₅O₁₂ and further to Li_8.5Ti₅O₁₂ [34]. Usually the former process was described by a core-shell model, corresponding to a two-phase transition process, while in the latter process a single-phase transition of spinel/rock-salt occurred [35–38]. When fast lithium insertion and extraction are enforced at a high current rate, as shown in Fig. 3b, the two-phase conversion from Li₄Ti₅O₁₂ to Li₇Ti₅O₁₂ cannot be completely carried out in a limited time due to the intrinsical poor conductivity of this material. Probably, a Li₄Ti₅O₁₂ core exists in the Li₇Ti₅O₁₂, and in the process of the single-phase conversion (Li₇Ti₅O₁₂ → Li_8.5Ti₅O₁₂), a transition phase of Li_4+δTi₅O₁₂ could form around the core. During the lithium extraction, the original Li₄Ti₅O₁₂ cannot be retrieved completely, and Li_4+δ Ti₅O₁₂ could exist inevitably. The higher current rate for lithium insertion/extraction can result in the more Li_4+δ Ti₅O₁₂. In the sequent cycles, the existence of Li_4+δTi₅O₁₂ is also the possible reason for the deformation of a flat discharge plateau. Based on the same mechanism (Fig. 3), Li_4+δTi₅O₁₂ becomes Li_4+δ′Ti₅O₁₂ (δ′ > δ), so the capacity declines. Moreover, solid electrolyte interphase (SEI) film is formed with high impedance as the reduction of electrolyte at the low potential. So Li₄Ti₅O₁₂ exhibits an obvious capacity loss when discharged to 0 V, especially at the high current density. This conclusion also can be deduced from Shu’s work [29].

Fig. 2

The first, second, and third discharge–charge curves of the Li₄Ti₅O₁₂ discharged to 0 V at the current density of a 350 mA g⁻¹, b 875 mA g⁻¹, and c 1,750 mA g⁻¹; The first, second, and third discharge–charge curves of the Li₄Ti₅O₁₂ discharged to 1 V at the current density of 1,750 mA g⁻¹ (d)

Fig. 3

Sketch map of lithium-ion intercalation and extraction for Li₄Ti₅O₁₂ discharge to a 1 V and b 0 V at the current density of 1,750 mA g⁻¹

The obvious capacity loss of Li₄Ti₅O₁₂ when discharged to 0 V mentioned above can be improved by doping Nb. The initial discharge–charge curves of Li₄Ti₅O₁₂ and Li₄Ti_4.95Nb_0.05O₁₂ electrodes down to 0 V and 1 V are shown in Fig. 4a and b. When cycled between 2.5 and 0 V, Li₄Ti_4.95Nb_0.05O₁₂ presents an initial discharge capacity of 351.9 mAh g⁻¹ and a reversible charge capacity of 272.8 mAh g⁻¹, while Li₄Ti₅O₁₂ only has a discharge capacity of 271.5 mAh g⁻¹ and a reversible capacity of 209.2 mAh g⁻¹. If all Ti⁴⁺ ions are reduced to Ti³⁺ ions, the theoretical capacity of Li₄Ti₅O₁₂ will be 293 mAh g⁻¹ [29]. In fact, on the basis of Zhong’s [34] report, Li₄Ti₅O₁₂ is transferred into Li_8.5Ti₅O₁₂ (not Li₉Ti₅O₁₂, corresponding to full reduction from Ti⁴⁺ to Ti³⁺) when discharged to 0 V, which is corresponding to a reversible capacity of about 262 mAh g⁻¹. During the initial discharge (lithium insertion), the high discharge capacity is related to the deep lithium insertion (the highest value: 262 mAh g⁻¹) into the active material (Li₄Ti₅O₁₂ or Nb-doped Li₄Ti₅O₁₂), some side reactions for SEI formation and lithium adsorption in the carbon black [39]. The higher reversible charge capacity of Nb-doped Li₄Ti₅O₁₂ is mainly attributed to the lithium extraction from the active material (close to the theoretical value: 262 mAh g⁻¹) and lithium desorption from the carbon black (about 10∼20 mAh g⁻¹). Figure 4c and d show the cycling performances and columbic efficiencies of Li₄Ti₅O₁₂ and Li₄Ti_4.95Nb_0.05O₁₂ at the current of 35 mA g⁻¹ down to 0 V and 1 V. When the cells are discharged to 1 V at the same current density, the 100th discharge capacity of Li₄Ti_4.95Nb_0.05O₁₂ is 157.4 mAh g⁻¹, but the corresponding value of Li₄Ti₅O₁₂ is 135.7 mAh g⁻¹. The improvement in capacity can be attributed to Nb⁵⁺ doped into the lattice of the Li₄Ti₅O₁₂, as we discussed in our previous paper [32]. The 100th discharge capacity of Li₄Ti_4.95Nb_0.05O₁₂ down to 0 V is 231.2 mAh g⁻¹ while the capacity of Li₄Ti₅O₁₂ is only 111.1 mAh g⁻¹. The specific capacity retentions of Li₄Ti_4.95Nb_0.05O₁₂ and Li₄Ti₅O₁₂ are 81.6% and 51.2% except the initial discharge. Obviously, when discharged to 0 V, the two spinel materials have a big gap on the capacity and cycling performance. The improvement in capacity can also be attributed to Nb⁵⁺ doped into the lattice of the Li₄Ti₅O₁₂. Compared with Li₄Ti₅O_12, the lattice constant of the Li₄Ti_4.95Nb_0.05O₁₂ is enlarged, which results in the broader path for lithium insertion and extraction [40, 41]. For the enhanced cycling stability of Nb-doped Li₄Ti₅O₁₂, the main reasons are the enlarged lattice constant and improvement on the conductivity due to the introduction of Nb⁵⁺ in the lattice, which are helpful for suppressing the formation of Li_4+δ Ti₅O₁₂.

Fig. 4

Initial discharge–charge curves of Li₄Ti₅O₁₂ and Li₄Ti_4.95Nb_0.05O₁₂ discharge to a 0 V and b 1 V at the current density of 35 mA g⁻¹; Cycling performance and columbic efficiency of Li₄Ti₅O₁₂ and Li₄Ti_4.95Nb_0.05O₁₂ discharged to c 0 V and d 1 V at the current density of 35 mA g⁻¹

The 1st, 2nd, 3rd, 50th, and 100th discharge–charge curves of Li₄Ti₅O₁₂ and Li₄Ti_4.95Nb_0.05O₁₂ discharged to 0 V at the current density of 35 mA g⁻¹ are plotted in Fig. 5. It can be seen that both Li₄Ti₅O₁₂ and Li₄Ti_4.95Nb_0.05O₁₂ show a loss of the 1.55 V plateau with increasing the cycling number. However, Li₄Ti_4.95Nb_0.05O₁₂ shows a much slower plateau capacity loss than Li₄Ti₅O₁₂ with increasing the cycling number, which indicates that Nb-doping is beneficial to the electrochemical stability of Li₄Ti₅O₁₂ during the electrochemical process for discharging to 0 V. However, there is an obvious irreversible capacity between initial discharge and charge processes. It mainly occurs in the voltage range of 0.75∼0 V for the formation of SEI film with high impedance [39].

Fig. 5

The 1st, 2nd, 3rd, 50th, and 100th discharge–charge curves of a Li₄Ti₅O₁₂ and b Li₄Ti_4.95Nb_0.05O₁₂ discharged to 0 V at the current density of 35 mA g⁻¹

The first, second, and third discharge–charge curves, cycling performance and columbic efficiencies of Li₄Ti_4.95Nb_0.05O₁₂ discharge to 0 V at the current density of 350 mA g⁻¹, 875 mA g⁻¹ and 1,750 mA g⁻¹ are shown in Fig. 6. The discharge capacity of Li₄Ti_4.95Nb_0.05O₁₂ is 219.1 mAh g⁻¹ at the current density of 350 mA g⁻¹ after 100 cycles. When the higher current densities are adopted, Li₄Ti_4.95Nb_0.05O₁₂ still keeps excellent cycling performance and high columbic efficiency (∼100%). And, it retains a discharge capacity of 120.2 mAh g⁻¹ after 100 cycles even at the current density of 1,750 mA g⁻¹. The excellent cycling performance of the Li₄Ti_4.95Nb_0.05O₁₂ at the high current densities is mainly related to the enlargement of the lattice constant and the improvement of electronic conductivity [32].

Fig. 6

The first, second, and third discharge–charge curves (a), (b), and (c); cycling performance and columbic efficiency (d) of Li₄Ti_4.95Nb_0.05O₁₂ discharged to 0 V at the current density of 350, 875, and 1,750 mA g⁻¹

Conclusion

Li₄Ti_4.95Nb_0.05O₁₂ powder has been successfully synthesized by a sol–gel method with citric acid as a chelating agent. XRD patterns show that Li₄Ti_4.95Nb_0.05O₁₂ has good crystallinity. The electrochemical performance of Li₄Ti_4.95Nb_0.05O₁₂ discharged to 0 V has been investigated. It is found that Li₄Ti_4.95Nb_0.05O₁₂ has a higher specific capacity and better cycling stability than Li₄Ti₅O₁₂. Li₄Ti_4.95Nb_0.05O₁₂ exhibits a capacity of 231.2 mAh g⁻¹ at the current of 35 mA g⁻¹ even after 100 cycles. All lines of evidence demonstrate that Nb-doping is beneficial to the electrochemical stability of Li₄Ti₅O₁₂ discharged to 0 V.