Mg2+ hybrid-ion batteries

Journal of Power Sources 346 (2017) 134e142

Contents lists available at ScienceDirect

Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour

Competition between insertion of Liþ and Mg2þ: An example of TiO2-B nanowires for Mg rechargeable batteries and Liþ/Mg2þ hybrid-ion batteries Yuan Meng a, Dashuai Wang a, Yingjin Wei a, *, Kai Zhu a, Yingying Zhao a, Xiaofei Bian a, Fei Du a, Bingbing Liu b, Yu Gao a, **, Gang Chen a, b a b

Key Laboratory of Physics and Technology for Advanced Batteries (Ministry of Education), College of Physics, Jilin University, Changchun 130012, PR China State Key Laboratory of Superhard Materials, Jilin University, Changchun 130012, PR China

h i g h l i g h t s

g r a p h i c a l a b s t r a c t

 TiO2-B nanowires were prepared by the hydrothermal method.  TiO2-B nanowires showed Mg2þ double-layer capacitive in Mg battery. nanowires showed Liþ  TiO2-B pseudo-capacitive in Liþ/Mg2þ hybrid-ion battery.  The hybrid battery showed longer cycle life and larger capacity than Mg battery.

a r t i c l e i n f o

a b s t r a c t

Article history: Received 27 November 2016 Received in revised form 18 January 2017 Accepted 10 February 2017

Titanium dioxide bronze (TiO2-B) nanowires were prepared by the hydrothermal method and used as the positive electrode for Mg rechargeable batteries and Liþ/Mg2þ hybrid-ion batteries. First-principles calculations showed that the diffusion barrier for Mg2þ (0.6 eV) in the TiO2-B lattice was more than twice of that for Liþ (0.3 eV). Electrochemical impedance spectroscopy showed that the charge transfer resistance of TiO2-B in the Mg2þ ion electrolyte was much larger than that in the Liþ/Mg2þ hybrid electrolyte. For these reasons, the Mg rechargeable battery showed a small discharge capacity of 35 mAh g1 resulting from an electrochemical double-layer capacitive process. In comparison, the TiO2-B nanowires exhibited a large capacity (242 mAh g1 at the 20 mA g1 current density), high rate capability (114 mAh g1 at 1 A g1), and excellent cycle stability in the Liþ/Mg2þ hybrid-ion battery. The dominant reaction occurred in the TiO2-B electrode was intercalation of Liþ ions, of which about 74% of the total capacity was attributed to Liþ pseudo-capacitance. © 2017 Elsevier B.V. All rights reserved.

Keywords: Titanium dioxide bronze Positive electrode Magnesium ion battery Hybrid-ion battery First-principles calculations Electrochemical properties

1. Introduction Lithium-metal batteries have been regarded as one of the most

* Corresponding author. ** Corresponding author. E-mail addresses: [email protected] (Y. Wei), [email protected] (Y. Gao). http://dx.doi.org/10.1016/j.jpowsour.2017.02.033 0378-7753/© 2017 Elsevier B.V. All rights reserved.

promising alternatives to traditional lithium ion batteries (LIBs) that using graphite as the negative electrode. Currently, the development of lithium-metal batteries has been seriously impeded by the safety concern posed by the Li electrode due to the progressive formation of lithium dendrite. This problem can be avoided in magnesium rechargeable batteries (MRBs) because the Mg negative electrode in these batteries is free of dendrite

Y. Meng et al. / Journal of Power Sources 346 (2017) 134e142

formation [1e3]. Additionally, magnesium presents other advantages such as its abundant reserves in the earth crust, low cost and high volumetric capacity, making MRBs a promising electrochemical energy storage device. However, in such batteries, diffusion of Mg2þ in the positive electrode is kinetically sluggish due to the strong polarity of divalent Mg2þ. Moreover, the lack of suitable electrolytes with a sufficiently wide electrochemical window also has limited the exploration of high-voltage positive electrode. To date, the development of Mg2þ ion electrolytes and Mg2þ storage materials remains one of the biggest bottlenecks for the development of MRBs. Recently, the Liþ/Mg2þ hybrid-ion batteries (LMIBs) have attracted worldwide interest as a new electrochemical energy storage device. This type of battery uses metallic Mg as the negative electrode and a Liþ storage material as the positive electrode, in combination with a Liþ/Mg2þ hybrid electrolyte. During electrochemical processes, only Mg2þ ions deposit on the negative electrode ensuring the high safety of these batteries. Meanwhile, the reaction of the positive electrode is dominated by Liþ intercalation [4], with a small percent of contribution from Mg2þ storage in some cases [5e7]. Some intrinsic shortcomings of conventional MRBs such as large electrode polarization, limited rate capability, and poor cycle stability could be effectively overcome by use of the LMIBs system. Despite these advantages, it should be pointed out that the anodic stability of currently used Liþ/Mg2þ hybrid electrolytes such as (PhMgCl)2eAlCl3þLiBH4/tetrahydrofuran (THF) [8] and Mg(BH4)2þLiBH4/tetraglyme (TG) [9] is lower than 2.5 V vs. Mg2þ/Mg0. So far, studies of LMIBs have been limited to only a few choices of positive electrode. For example, the Chevrel phase Mo6S8 which has been recognized as the most successful Mg2þ intercalation material, has been used as the positive electrode for LMIBs. The material delivered 126 mAh g1 at the 0.1 C rate, and showed excellent stability up to 3000 cycles [10]. Some transition metal disulfides such as MoS2 [7] and TiS2 [4] have also been studied as the positive electrodes for LMIBs. These LMIBs were attractive for their outstanding rate capability because the weak van der Waals forces between adjacent layers allow migration of Liþ ions in the material with a low diffusion barrier. Transitional metal dioxides with electrochemical redox potential compatible with the Liþ/Mg2þ hybrid electrolyte could also be used as the positive electrode for LMIBs. Recently Nuli et al. showed that the anatase TiO2 in LMIB could deliver a discharge capacity of 150 mAh g1 at the 0.1 C rate [6], similar to its behaviors in LIB. Titanium dioxides polymorphs including anatase, rutile, and bronze (TiO2-B) have attracted intensive attention in energy and environmental technologies. Anatase and rutile have been studied for solar cells and photocatalysis. But in the LIBs field, much evidence indicates that TiO2-B could be a better choice than the other two polymorphs. The unique three-dimensional framework of TiO2-B can accommodate more than 250 mAh g1 of Liþ ions [11,12]. For comparison, the practical Liþ capacities for anatase and rutile are limited to 168 mAh g1. Additionally, the unique pseudocapacitive Liþ ion storage mechanism of TiO2-B is more favorable for high rate charge-discharge comparing to the solid-state diffusion processes for anatase and rutile [13,14]. Considering the outstanding Liþ ion storage properties of TiO2B, it is of great interest to explore its availability in LMIB. One could expect that the LMIB cell using TiO2-B as the positive electrode could be a promising electrochemical energy storage device especially for large-scale energy storage where safety, life span, and price are more important concerns than energy density. Herein, we prepared TiO2-B nanowires by the hydrothermal method. Firstprinciples calculations and experimental studies showed that TiO2-B was not a suitable Mg2þ intercalation electrode for MRB. However, when used as the positive electrode for LMIB, TiO2-B

135

showed promising Liþ intercalation properties resulting in large discharge capacity, long cycle stability, and high rate capability. The different electrochemical properties of TiO2-B in MRB and LMIB are consistent with the very different energy storage mechanisms of Liþ and Mg2þ in the same host structure. 2. Computational and experimental details 2.1. Computational method First-principles calculations were performed by the density functional theory (DFT) using the Vienna ab initio Simulation Package (VASP) [15,16]. The ion-electron interaction was described with the projector augmented wave (PAW) method. The exchange correlation energy was described by the generalized gradient approximation (GGA) which has demonstrated its rationality and veracity on TiO2eB without the use of an on-site U parameter [17]. The geometry optimizations were performed by the conjugated gradient method with a convergence threshold of 105 eV in energy and 0.005 eV/Å in force. The Brillouin zone was represented by a Monkhorst-Pack special k-point mesh of 4  8  6 for geometry optimizations and a plane wave energy cut-off of 500 eV was applied. For analysing the diffusion of lithium and magnesium in the TiO2-B lattice, the Climbing Image-Nudged Elastic Band (CINEB) method was used for a 1  3  2 supercell [18,19], utilizing five images constructed via the linear interpolation method. The crystal structures were visualized by the VESTA software. 2.2. Preparation of the TiO2-B nanowires 0.4 g commercial TiO2 (P25, Degussa) was dispersed in 40 ml 10 M NaOH under magnetic stirring and ultrasonic treatment. The resulting suspension was transferred into a 50 ml Teflon autoclave and subjected to hydrothermal treatment at 170  C for 60 h. After cooling to room temperature, a white precipitate was collected by centrifugation and washed with de-ionized water and dilute hydrochloric acid. This intermediate product was constantly stirred in 0.1 M HCl for 8 h to complete the Hþ/Naþ exchange reaction. Then the material was frozen-dried at 30  C for 20 h. Finally, a white powder was collected and heat treated at 400  C for 4 h in air to get the TiO2-B nanowires. 2.3. Characterizations of the material The crystal structure of the TiO2-B nanowires was determined by X-ray diffraction (XRD) on a Bruker AXS D8 X-ray diffractometer with Cu Ka radiation. The lattice parameters of the material were calculated using the Celref Program. The morphology of the material was studied by scanning electron microscope (SEM, JEOL JSM6700F). The microstructure of the nanowires was characterized by high resolution transmission electron microscope (HRTEM, FEI Tecnai G2, 200 kV). Nitrogen adsorption-desorption isotherms were collected on a Micromeritics ASAP 2010 instrument. Specific surface area of the material was determined by the BrunauerEmmett-Teller (BET) method. The ionic conductivity of the electrolyte was measured with a hand-held conductivity meter (AZ Instrument Corp.) 2.4. Electrochemical experiments Electrochemical properties of the TiO2-B nanowires in MRB and LMIB were studied with 2032-type coin cells using Mg foil as the counter electrode. The working electrode was composted of 70 wt% TiO2-B nanowires, 15 wt% Super P conductive additive, and 15 wt% poly-vinylidene fluoride (PVDF) binder which was pasted on a

136

Y. Meng et al. / Journal of Power Sources 346 (2017) 134e142

graphite foil (Gif) current collector. The electrode was cut to 8  8 mm2 after drying at 120  C for 12 h in a vacuum oven. Each electrode contained about 0.8 mg TiO2-B active material. A 0.4 M APC/THF electrolyte was used as the Mg2þ ion electrolyte. The Liþ/ Mg2þ hybrid electrolyte was prepared by adding 0.4 M LiCl into the Mg2þ ion electrolyte. For comparison, the electrochemical properties of the TiO2-B nanowires in LIB were studied using a 1.0 M lithium hexafluorophosphate (LiPF6) electrolyte dissolved in ethylene carbonate (EC) and dimethyl carbonate (DMC) (EC: DMC ¼ 3: 7). The counter electrode was used as Li foil and the electrode slurry was pasted on a Cu current collector. Galvonostatic charge-discharge was performed on a LAND-2010 automatic battery tester. Cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) were performed on a Bio-Logic VSP multichannel electrochemical workstation. The impedance data were collected in a frequency range from 1 MHz to 1 mHz by using an ac voltage of 5 mV. 3. Results and discussion 3.1. Structural and morphological properties of the TiO2-B nanowires The crystallographic properties of the hydrothermal-prepared TiO2-B were studied by powder X-ray diffraction as shown in Fig. 1a. The XRD pattern exhibits all the diffraction peaks of monoclinic TiO2-B. The broad peaks indicate that the crystallite size of the material was in the nanometer scale. The lattice parameters of the material were calculated as a ¼ 12.2508 Å, b ¼ 3.7544 Å, c ¼ 6.5249 Å, a ¼ g ¼ 90 , and b ¼ 105.89 , which are consistent with those recorded in JCPDS 97-004-1056. Fig. 1b shows a SEM photograph of the TiO2-B material. It is composed of well-dispersed nanowires with 50e200 nm in width and several micrometers in length, resulting in a large aspect ratio. The nanowires were further characterized by HRTEM as shown in Fig. 1c. The clear lattice fringes indicate high degree of crystallinity of the material. Fast Fourier transformation (FFT) analysis showed the (200) and (110)

diffractions of monoclinic TiO2-B, confirming the formation of the target material. But, distinct defects were observed on the surface of the nanowires, an issue that should be resolved in future studies. The nitrogen adsorption-desorption measurement of the nanowires shows typical III-type isotherms of a porous material (Fig. 1d). However, significant meso- or micro-pores were not observed on the TiO2-B nanowires, therefore, the porosity should be mainly due to aggregation of the nanowires. Additionally, the BET surface area of the material was determined as 39.6 m2g-1. 3.2. First principle calculations of Liþ and Mg2þ migrations in TiO2B As an essential step of theoretical studies, the structural optimization of TiO2-B was performed according to first-principles calculations. The crystal structure of TiO2-B is monoclinic with the space group of C2/m, and all atoms occupy the wyckoff 4i site. In the TiO2-B structure, every titanium atom is coordinated 6-fold to oxygen atoms, forming the octahedral TiO6 unit (Fig. S1, supporting information). The optimized structural parameters of TiO2-B were a ¼ 12.2754 Å, b ¼ 3.7656 Å, c ¼ 6.6146 Å, a ¼ g ¼ 90.0 , and b ¼ 106.855 as listed in Table 1. These values are in good agreement with the previously reported theoretical results [17] and the experimental results of our hydrothermally-prepared TiO2-B. A small systematic overestimation of cell volume always occurs when employing the GGA method, resulting in a reasonable difference of 2.63% of cell volume between the experimental and theoretical Table 1 Experiment and calculated lattice parameters of TiO2-B. Lattice parameters (Å,  )

Experimentala Calculatedb Calculateda a b

V(Å3)

a

B

c

a

b

g

12.2508 12.2875 12.2754

3.7544 3.7746 3.7656

6.4452 6.5832 6.6146

90.0 90.0 90.0

105.89 107.054 106.855

90.0 90.0 90.0

This work. Ref. [17].

Fig. 1. (a) XRD pattern, (b) SEM image, (c) TEM image, (d) nitrogen adsorption-desorption isotherms of the TiO2-B nanowires.

285.12 291.91 292.62

Y. Meng et al. / Journal of Power Sources 346 (2017) 134e142

results. The rate capability of rechargeable batteries is determined by the kinetics of both electron transport and ion diffusion, and is one of the most important indexes of battery performance. Next, we compared the diffusion of monovalent Liþ and divalent Mg2þ ions in the TiO2-B lattice to get insight into their energy storage mechanisms at the atomic level. Based on the described ion insertion sites and the diffusion pathway of TiO2-B [17,20], we selected the most stable adjacent insertion site C as the initial site and the final site for both Liþ and Mg2þ ions, as shown in Fig. 2a. Here, the site C is located at the centre of the quasi-cube constructed by Ti and O atoms, and these face-shared hexahedrons form a channel along the b axis, providing a perfect diffusion pathway for Liþ and Mg2þ ions. With the CI-NEB method, we demonstrated a specific pathway by which the inserted ions migrate along a straight line between two adjacent C sites. The corresponding diffusion barrier for Liþ and Mg2þ ions are plotted in Fig. 2b. The single saddle point indicates that there are no other low energy sites between the two adjacent C sites. The calculated diffusion barriers for the Liþ and Mg2þ ions are about 0.3 eV and 0.6 eV, respectively. This indicates that the diffusion of Mg2þ is much more difficult than the diffusion of Liþ in the same TiO2-B structure. According to the estimation of Ceder et al, the maximum diffusion barrier for guest cations (including monovalent Liþ and multivalent Mg2þ and Al3þ etc.) in nano-sized intercalation materials is about 0.650 eV [21]. The small Liþ diffusion barrier calculated here would allow highly efficient Liþ intercalation in TiO2-B. In contrast, the large diffusion barrier of Mg2þ is just below the 0.65 eV threshold, suggesting that intercalation of Mg2þ in the TiO2-B nanowires may not be favorable. 3.3. Electrochemical properties of the TiO2-B nanowires in MRBs The electrochemical properties of the TiO2-B nanowires in LIB were studied with the TiO2-Bk1 M LiPF6/EC þ DMCkLi cells. The obtained rate capability (Fig. S2, supporting information) and cycling performance (Fig. S3, supporting information) are consistent with those reported in literature [11,22]. The material showed a stable discharge capacity of 269 mAh g1 at the 20 mA g1 current density, and a discharge capacity of 182 mAh g1 was obtained at 1000 mA g1. This remarkable rate capability could be attributed to the low Liþ diffusion barrier of TiO2-B as revealed by the firstprinciples calculations. Moreover, the material showed excellent cycle stability as a large capacity of 243 mAh g1 could be obtained after 200 cycles, which was 90% of the second cycle. To study the

137

electrochemical properties of the TiO2-B nanowires in MRB, we assembled TiO2-Bk0.4 M APC/THFkMg cells and used them for electrochemical measurements. Fig. 3a shows the voltage curves of the MRB cell in the voltage window of 0.01e2.0 V, acquired at a current density of 20 mA g1. The cell showed a first discharge capacity of 34 mAh g1 followed by a much larger charge capacity of 94 mAh g1, resulting in a Coulomb efficiency of 276%. The excessive capacity could be attributed to the decomposition of the electrolyte. Moreover, the stainless steel cell case might be corroded by the APC electrolyte that contains highly concentrated chloride anions [23]. The MRB cell showed slope voltage curves from the second cycle as typical electrochemical double-layer capacitors. The big differences between the second-cycle curves and those of the first cycle indicate that the MRB cell underwent an activation process during the first cycle. In order to study the reaction mechanism of the TiO2-B MRB cell, we collected cyclic voltammgrams at the 0.05 mV s1 scan rate as shown in Fig. 3b. The observed rectangle-like curves indicate that the MRB cell mainly functioned according to the electrochemical double-layer capacitive process. However, one cathodic peak (at 0.44 V) and two anodic peaks (at 0.67/1.04 V) with small current responses were observed. These current peaks were consistent with those of the Gifk0.4 M APC/THFkMg cell (Fig. S4, supporting information), indicating that they did not result from the TiO2-B nanowires. Instead, these additional current peaks were attributed to the redox reactions of the APC/THF electrolyte and/or Mg2þ storage on the Gif current collector as suggested by Wang et al. [24] However, the capacity from these side reactions (~2 mAh g1) could be neglected (Fig. S5, supporting information). Overall, the energy storage of the MRB cell was attributed to the double-layer capacitance of Mg2þ on the TiO2B electrode. Fig. 3c shows the rate capability of the TiO2-B nanowires in MRB. Obviously, the obtained discharge capacities were not sufficient for use as an energy storage device. However, considering the doublelayer capacitive mechanism of this MRB cell, the Mg2þ capacities could be substantially improved by using much thinner nanowires and taking advantage of a large specific surface area. This approach is currently being investigated in our group. Despite the small specific capacities, the TiO2-B MRB cell showed good rate capability. For example, a discharge capacity of 25 mAh g1 was obtained at the 1000 mA g1 current density, 56% of that at 20 mA g1. Fig. 3d shows the cyclic performance of the TiO2-B MRB cell at the 200 mA g1 current density. The cell showed an initial discharge capacity of 19 mAh g1. The discharge capacity increased in the

Fig. 2. (a) Diffusion pathway and (b) calculated diffusion barriers for Liþ and Mg2þ ions in TiO2-B.

138

Y. Meng et al. / Journal of Power Sources 346 (2017) 134e142

Fig. 3. (a) Charge-discharge curves, (b) cyclic voltammegrams, (c) rate capability and (d) cycling performance at 200 mA g1 current density of the TiO2-B MRB cell.

early cycles due to activation of the electrode. Afterwards, the cell showed outstanding cycle stability and a relatively stable discharge capacity of 35 mAh g1 was obtained in 200 cycles. However, the Coulomb efficiencies were always larger than 100% (maximum: 110%) throughout the whole charge-discharge experiment, suggesting continuous electrolyte decomposition and corrosion of the cell case. 3.4. Electrochemical properties of the TiO2-B nanowries in LMIBs The above first-principles calculations and experimental results showed that TiO2-B is not a suitable Mg2þ intercalation material. This could be partly attributed to the high Mg2þ diffusion barrier in TiO2-B, which is more than twice of that for Liþ ions. To determine how the TiO2-B nanowires behave in the Liþ/Mg2þ hybrid electrolyte, a series of experiments were carried out on the TiO2-Bk0.4 M APCþ0.4 M LiCl/THFkMg cells. Fig. 4 shows the charge-discharge curves of the LMIB cell at the current density of 20 mA g1, together with those of a conventional TiO2-B LIB cell for comparison. The LMIB cell showed a voltage plateau at ~0.65 V and a specific capacity of 198 mAh g1 during the first discharge, which increased to ~0.75 V and 236 mAh g1 in the second cycle due to activation of the electrode. The LMIB cell showed very similar voltage curves as those of the LIB cell except there was a constant voltage difference resulting from the different redox potentials of Mg2þ/Mg0 (2.37 V vs. SHE) and Liþ/Li0 (3.04 V vs. SHE). This

Fig. 4. Charge-discharge profiles of the TiO2-B nanowires in LMIB (red) and LIB (green) cells in the first three cycles. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

indicates that the electrode reactions of the LMIB cell were dominated by Liþ intercalation. To investigate this possibility, X-ray

Y. Meng et al. / Journal of Power Sources 346 (2017) 134e142

energy dispersive spectroscopy (EDX) affiliated to scanning transmission electron microscopy (STEM) was performed on several single TiO2-B nanowires after the LMIB cell was discharged to 0.01 V (Fig. S6, Supporting Information). Before EDX analysis, the electrode was throughout washed with THF for 7 days to remove any residual absorbance. Analysis showed that only 0.01e0.03 Mg2þ per formula unit of TiO2-B was present in the electrode, providing a negligible Mg2þ ion capacity. This confirms that the energy storage of the TiO2-B electrode was dominated by Liþ ions. The small Mg2þ capacity may be attributed to the double-layer capacitance of the TiO2-B electrode. The experimental capacity (~250 mAh g1) of the EDX sample corresponds to a composition of Li0.75TiO2-B which is close to the Li0.8TiO2-B phase reported by Armstrong et al. [13]. Based on the structure data provided by Armstrong et al. [13], we calculated the theoretical d-spacing of the (200) and (110) diffractions of TiO2-B and Li0.8TiO2-B as shown in Table 2. The theoretical results are very close to the experimental results obtained by ex-situ HRTEM analysis (Fig. 5). This further demonstrates that the capacity of the TiO2-B LMIB cell was dominated by Liþ intercalation. Fig. 6a shows the rate capability of the TiO2-B LMIB cell. A discharge capacity of 242 mAh g1 was obtained at 20 mA g1 and then decreased to 219, 199, 182 141, and 114 mAh g1 as the current density increased to 50, 100, 200, 500, and 1000 mAg1, respectively. When the current density returned back to 50 mA g1, the discharge capacity also recovered to 215 mAh g1. The specific energy of the LMIB cell was calculated based on the net weight of TiO2-B, resulting in a maximum specific energy of 175 Wh kg1. This result is much larger than that reported previously for conventional MRBs, such as 110 Wh kg1 for Mo6S8 [25]. If we calculate the specific energy based on the total mass of the electrode stack, a specific energy of 38 Wh kg1 could be obtained for the current LMIB cell. The cycling performance of the LMIB cell at the 200 mA g1 current density is displayed in Fig. 6b. The discharge capacity increased from 110 to 212 mAh g1 in the initial 20 cycles due to the activation process. The cell then showed perfect cycle stability with almost no capacity fading after 200 cycles. Moreover, the Coulomb efficiency was maintained at 100%, demonstrating excellent electrochemical reversibility of the LMIB cell. We found that the amount of electrolyte had big impact on the electrochemical properties of the LMIB cell. The above charge-discharge performance was obtained with 50 mL electrolyte. By comparison, the electrochemical performance was greatly degraded using less electrolyte (Fig. S7, Supporting Information). This indicates that sufficient amount of electrolyte is necessary a LMIB cell to ensure efficient soaking of the whole battery pack. In addition to the above used APC þ LiCl electrolyte, Mg(BH4)2þLiBH4 is another family of Liþ/Mg2þ hybrid electrolyte [9]. For comparison, we characterized the electrochemical performance of the TiO2-B nanowires in a 0.5 M Mg(BH4)2þ1.5 M LiBH4/TG electrolyte in the voltage window of 0.01e2.0 V. The charge-discharge profiles were identical to those obtained with the APC þ LiCl electrolyte (Fig. S8, Supporting Information), indicating the same electrochemical reactions occurred in both hybrid electrolytes. The LMIB cell showed an initial

Table 2 Theoretical d-spacing of the (200) and (110) diffractions of TiO2-B and Li0.8TiO2-B, comparing with the corresponding experimental d-spacing obtained from ex-situ HRTEM analysis.

(200) (110)

Calculated d-spacing (nm)

Experimental d-spacing (nm)

TiO2-B

Li0.8TiO2-B

TiO2-B

Li0.75TiO2-B

0.586 0.358

0.605 0.375

0.585 0.356

0.602 0.371

139

discharge capacity of 208 mAh g1 and then decreased to 174 mAh g1 after 200 cycles (Fig. S9, Supporting Information). This indicates that the electrochemical performance obtained with the Mg(BH4)2þLiBH4 electrolyte was inferior to that obtained with the APC þ LiCl electrolyte. Fig. 7a shows the CV curves of the TiO2-B LMIB cell at different scan rates. The material showed a reduction peak at around 0.5 V vs. Mg2þ/Mg0 and an oxidation peak at around 1.0 V. The CV curves are similar to those of the TiO2-B LIB cell (Fig. S10, Supporting Information), except a simultaneous voltage increase of about 0.7 V in the current peaks due to the different standard electrode potentials of Mg2þ/Mg0 and Liþ/Li0. It is well-known that the TiO2-B electrode in LIB has a large percent of capacity from pseudo-capacitance [14]. Considering that the specific capacity of the TiO2-B LMIB cell was dominated by Liþ ions, it is reasonable to conclude that the LMIB cell also had a large contribution from pseudo-capacitance. According to Dunn's work [26], the peak current obeys the power-law relationship with the scan rate:

i ¼ avb

(1)

The b value can be changed between 0.5 and 1.0. The value of 0.5 means a diffusion controlled process and the value of 1.0 means a capacitive process. The b value can be determined by the slope of log(i) versus log(v). As shown in Fig. 7b, the b value of the reduction and oxidation peaks are 0.69 and 0.68, indicating that the capacity of the LMIB cell contained mixed contributions of a diffusion controlled process and a pseudo-capacitance process. And they can be quantitatively discriminated from the total capacity via the following equation:

i ¼ k1 v þ k2 v1=2

(2)

where k1v indicates the capacitive current and k2v1/2 represents the diffusion controlled current. The parameters k1 and k2 can be determined from the liner relationship of iv1/2 versus v1/2, i.e.,

iv1=2 ¼ k1 v1=2 þ k2

(3)

Thus, the corresponding current contribution can be quantificationally calculated as a function of potential. As shown in Fig. 7c, the capacitive currents (k1v) are calculated and distinguished from the total measured currents. Apparently, the diffusion-controlled process is mainly generated at around the reduction and oxidation peaks, indicating that the diffusion process is facile at this region and corresponds to the redox reaction of Ti4þ/Ti3þ. Based on the quantitative calculation, about 74% of the total capacity is pseudo-capacitance at the scan rate of 2 mV s1. This value is similar to that of 70% obtained for the LIB cell using the same TiO2-B electrode (Fig. S10, Supporting Information). 3.5. Electrochemical impedance spectroscopy Even though the calculated Mg2þ diffusion barrier in TiO2-B (0.6 eV) is below the 0.65 eV threshold that was suggested by Ceder et al, no evidence of Mg2þ intercalation was observed for our MRB and LMIB cells. As well known, the electrochemical properties of rechargeable batteries are not only controlled by the ionic diffusion in the electrode bulk but also depend on the kinetics at the electrode/electrolyte interface. To further investigate why Mg2þ ions cannot migrate in the TiO2-B lattice, we performed EIS analysis to study the interfacial kinetics of the MRB and LMIB cells. Fig. 8a shows the Nyquist plots of the symmetrical TiO2-B cell using the APC/THF electrolyte. The electrodes of the symmetrical cell were collected from two identical TiO2-B MRB cells after they were discharged to 0.01 V. The use of symmetrical cells eliminated the

140

Y. Meng et al. / Journal of Power Sources 346 (2017) 134e142

Fig. 5. HRTEM image of the pristine TiO2-B nanowires (a) and after the first discharge (b).

Fig. 6. Rate capability (a) and cycling performance (b) of the TiO2-B LMIB cell at the 200 mA g1 current density.

influence of the Mg anode and thus the EIS data obtained reveal the “real” kinetic properties of the TiO2-B electrode. The semicircle in the high-to-medium frequency region is ascribed to the charge transfer process. The corresponding charge transfer resistance was 1516 U. This value is much larger than that of 73 U obtained in the Liþ/Mg2þ hybrid electrolyte (Fig. 8b), indicating that the interfacial kinetics of the MRB cell were much more sluggish than the LMIB cell. Only a few magnesium species exist in the APC/THF electrolyte, the most important of which are MgCl2, MgClþ, Mg2Clþ 3 , Ph2Mg, and PhMgCl, all coordinated by THF. Due to their high polarity and charge density, the Mg2þ ions couple with Cl and THF forming hexacoordinated complexes as shown previously [27]:

The strong bonding effect between Mg and its coordinated ligands made it very difficult for the Mg2þ ions to dissolve from these large complexes, resulting in the observed large charge transfer resistance. As a consequence, these Mg2þ complexes tended to adsorb on the electrode surface under the electric gradient, resulting in the double-layer capacitance of the MRB cell. For comparison, the TiO2-B electrode showed a much smaller charge transfer resistance in the Liþ/Mg2þ hybrid electrolyte. According to Aurbach et al [27], the LiCl salt in the hybrid electrolyte was a Lewis-base and acted as a strong Cl donor. For example, the  Mg2Clþ 3 species could accept these Cl anions to form MgCl2. The remaining Liþ ions may form Liþ[AlPh4]- and Liþ[AlCl4]- instead of coordinating with the THF ligands. This is supported by the improved ionic conductivity that changes from 2.56 mS cm1 to 2.66 mS cm1 with the addition of 0.4 M LiCl into the APC/THF electrolyte (LiCl is only slightly soluble in THF forming a solution with no practical ionic conductivity [28]). Because of the much lower polarity and charge density, the Liþ ions could easily dissolve from Liþ[AlPh4]- and Liþ[AlCl4]-. Then, the dissolved Liþ ions could penetrate into the TiO2-B nanowires and diffuse in the lattice under a small kinetic barrier, improving the electrochemical performance of the LMIB cell. 4. Conclusions We prepared TiO2-B nanowires by the hydrothermal method

Y. Meng et al. / Journal of Power Sources 346 (2017) 134e142

141

Fig. 7. CV curves of the TiO2-B LMIB cell at different scan rates (a); linear fitting of the log(i) versus log(v) plots (b); fitted pseudo-capacitance contribution (shadow area) at the scan rate of 2 mV s1.

Fig. 8. Nquist plots of the symmetrical TiO2-B cell using the Mg2þ ion electrolyte (a) and the Liþ/Mg2þ hybrid electrolyte (b).

and studied the electrochemical properties of the material in MRB and LMIB. The results showed that TiO2-B was not functional for the intercalation of Mg2þ due to the difficulty of Mg2þ desolvation and a large diffusion barrier. The practical capacity obtained for the MRB cell was attributed to the double-layer capacitance of Mg2þ ions. In this light, preparation of nano-sized TiO2-B with larger specific surface area would improve the Mg2þ capacitance of the material. The TiO2-B nanowires exhibited large capacities, excellent cycle stability, and high rate capability in LMIB. Most of the capacity was attributed to the pseudo-capacitance of Liþ in

the nanowries. This work not only reveals the very different energy storage mechanisms of Liþ and Mg2þ in the same TiO2-B structure, but also demonstrates the use of traditional Liþ storage materials to construct high performance LMIB. If a new Liþ/Mg2þ hybrid electrolyte with wider voltage window can be developed, other Liþ storage materials with larger specific capacity and higher working voltage could be used to construct LMIBs. If so, the specific energy and volumetric energy of LMIBs could be significantly increased. This would make LMIBs more competitive than traditional LIBs for large-scale energy storage where safety,

142

Y. Meng et al. / Journal of Power Sources 346 (2017) 134e142

cycle life, and price are more important concerns than energy density. Acknowledgements This work was supported by the Ministry of Science and Technology of China (No. 2015CB251103), the National Natural Science Foundation of China (No. 51472104, 21473075, 51272088), the Department of Science and Technology of Jilin Province (No. 20140101083JC), and the Graduate Innovation Fund of Jilin University (No. 2016009). Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.jpowsour.2017.02.033. References [1] [2] [3] [4] [5] [6] [7] [8]

M. Matsui, J. Power Sources 196 (2011) 7048e7055. C. Ling, D. Banerjee, M. Matsui, Electrochim. Acta 76 (2012) 270e274. €ckle, A. Groß, J. Chem. Phys. 141 (2014) 174710. M. Ja T. Gao, F. Han, Y. Zhu, L. Suo, C. Luo, K. Xu, C. Wang, Adv. Energy Mater. 5 (2015) 1401507. T. Ichitsubo, S. Okamoto, T. Kawaguchi, Y. Kumagai, F. Oba, S. Yagi, N. Goto, T. Doi, E. Matsubara, J. Mater. Chem. A 3 (2015) 10188e10194. S. Su, Z. Huang, Y. NuLi, F. Tuerxun, J. Yang, J. Wang, Chem. Commun. 51 (2015) 2641e2644. C.-J. Hsu, C.-Y. Chou, C.-H. Yang, T.-C. Lee, J.-K. Chang, Chem. Commun. 52 (2016) 1701e1704. Q. Miao, Y. NuLi, N. Wang, J. Yang, J. Wang, S.-i. Hirano, RSC Adv. 6 (2016) 3231e3234.

[9] F. Tuerxun, Y. Abulizi, Y. NuLi, S. Su, J. Yang, J. Wang, J. Power Sources 276 (2015) 255e261. [10] Y. Cheng, Y. Shao, J.-G. Zhang, V.L. Sprenkle, J. Liu, G. Li, Chem. Commun. 50 (2014) 9644e9646. [11] X. Yan, Y. Zhang, K. Zhu, Y. Gao, D. Zhang, G. Chen, C. Wang, Y. Wei, J. Power Sources 246 (2014) 95e102. [12] A.R. Armstrong, G. Armstrong, J. Canales, R. García, P.G. Bruce, Adv. Mater. 17 (2005) 862e865. [13] A.R. Armstrong, C. Arrouvel, V. Gentili, S.C. Parker, M.S. Islam, P.G. Bruce, Chem. Mater. 22 (2010) 6426e6432. , M. Kalba  [14] M. Zukalova c, L. Kavan, I. Exnar, M. Graetzel, Chem. Mater. 17 (2005) 1248e1255. [15] G. Kresse, J. Furthmüller, Phys. Rev. B 54 (1996) 11169e11186. [16] G. Kresse, J. Furthmüller, Comput. Mater. Sci. 6 (1996) 15e50. [17] C. Arrouvel, S.C. Parker, M.S. Islam, Chem. Mater. 21 (2009) 4778e4783. nsson, J. Chem. Phys. 113 (2000) [18] G. Henkelman, B.P. Uberuaga, H. Jo 9901e9904. [19] K. Tibbetts, C.R. Miranda, Y.S. Meng, G. Ceder, Chem. Mater. 19 (2007) 5302e5308. [20] F. Legrain, O. Malyi, S. Manzhos, J. Power Sources 278 (2015) 197e202. [21] Z. Rong, R. Malik, P. Canepa, G. Sai Gautam, M. Liu, A. Jain, K. Persson, G. Ceder, Chem. Mater. 27 (2015) 6016e6021. [22] Y. Zhang, Q. Fu, Q. Xu, X. Yan, R. Zhang, Z. Guo, F. Du, Y. Wei, D. Zhang, G. Chen, Nanoscale 7 (2015) 12215e12224. [23] J. Muldoon, C.B. Bucur, A.G. Oliver, J. Zajicek, G.D. Allred, W.C. Boggess, Energy Environ. Sci. 6 (2013) 482e487. [24] L. Wang, K. Asheim, P.E. Vullum, A.M. Svensson, F. Vullum-Bruer, Chem. Mater. 28 (2016) 6459e6470. [25] E. Lancry, E. Levi, Y. Gofer, M.D. Levi, D. Aurbach, J. Solid State Electrochem. 9 (2005) 259e266. [26] G.A. Muller, J.B. Cook, H.-S. Kim, S.H. Tolbert, B. Dunn, Nano Lett. 15 (2015) 1911e1917. [27] N. Pour, Y. Gofer, D.T. Major, D. Aurbach, J. Am. Chem. Soc. 133 (2011) 6270e6278. [28] Y. Vestfried, O. Chusid, Y. Goffer, P. Aped, D. Aurbach, Organometallics 26 (2007) 3130e3137.