1 Introduction

Lithium-ion batteries are important energy storage devices for portable electronics, power tools, and electrical vehicles [1, 2]. In the past few decades, lithium-ion batteries (LIBs) have been power sources of choice for popular mobile electronic devices such as cellular phones, notebooks and MP3 players [3, 4]. However, a profitable use of LIBs is still limited in the case of large scaled energy storage applications requiring fast charge–discharging power rates [4], e.g., hybrid and electric vehicles (HEVs), renewable energy (wind and solar) plants. Accordingly, it is advisable to explore new nanostructured electrode materials capable of improving the rate performance of LIBs by assuring enhanced kinetics of the solid-state diffusion of the Li+ intercalation process as well as a high value of electronic conductivity [5]. A promising safe anode alternative to the commercial carbon/graphite is the spinel-type Li4Ti5O12 (LTO). This material has attracted attention because of the low cost, satisfactory safety and easy preparation. Spinel Li4Ti5O12 material has an excellent reversibility of Li-ion intercalation and deintercalation with a theoretical capacity of 175 mAh g−1 and it also exhibits no volume changes during charge and discharge cycles combined with an excellent safety performance [6–8]. It also has a very flat voltage plateau at around 1.55 V (vs. Li/Li+), which is higher than the reduction potential of most organic electrolytes [9–11]. However, the power performance of Li4Ti5O12 is greatly limited by its low electronic conductivity (ca. 10−13 S cm−1) and moderate Li diffusion coefficient (10−8–10−11 cm2 s−1) [12, 13]. In order to improve the electrochemical performance of the Li4Ti5O12 anode, extensive works concentrated on forming nanoparticles [14–17], doping [18–23] with metal cations and composing with carbon or metal powders [21, 24–29]. The formation of nanoparticles can reduce the Li-ion diffusion path as well as provide a large contact area between the nanoparticles. Doping by other metal cations and adding carbon or metal powder into the Li4Ti5O12 can increase electronic conductivity. Conventional organic solvents [ethylene carbonate, propylene carbonate (PC), dimethyl carbonate, etc.] used in Li-ion cells are flammable due to their relatively high vapour pressure. Instead of a salt solution in volatile solvents, non-volatile molten salts may be applied as electrolytes. Salts liquid at room temperature, usually called ionic liquids (ILs), are non-volatile solvents and show high thermal, chemical and electrochemical stability. They exhibit a very low vapour pressure and high non-flammability properties that are of importance in their use as safe electrolytes in Li-ion batteries [30–32]. Another approach is to test low vapour pressure molecular solvents, e.g., tetramethylene sulfone (sulfolane, TMS, T b = 280 °C) [33] or gamma-butyrolactone (GBL, T b = 205 °C) [34].

The general aim of the present paper was to study the Li4Ti5O12 anode in two non-flammable electrolytes (1 M LiPF6 in TMS + 5 wt% VC and 0.7 M LiNTf2 in MePrPyrNTf2 + 10 wt% GBL) and compare the results with those for the system Li4Ti5O12/Li with a classical electrolyte: LiPF6 in PC + DMC (1:1).

2 Experimental

2.1 Materials

Lithium titanate Li4Ti5O12 powder (Aldrich, BET surface area 32.6 m2 g−1, mean diameter 940 nm), carbon black (CB, Fluka), poly(vinylidene fluoride) (PVdF, Fluka), sulfolane (TMS, Fluka), gamma-butyrolactone (GBL, Aldrich), vinylene carbonate (VC, Aldrich), lithium foil (Aldrich, 0.75 mm thick), N-methyl-2-pyrrolidinone (NMP, Fluka), propylene carbonate (PC, Aldrich), dimethyl carbonate (DMC, Aldrich), lithium hexafluorophosphate (LiPF6, Fluka) and lithium bis(trifluoromethanesulphonyl)imide (LiNTf2, Fluka) were used as-purchased. N-methyl-N-propylpyrrolidinium bis(trifluoromethanesulphonyl)imide (MePrPyrNTf2) was prepared according to the literature [35] by reacting N-methylpyrrolidinium (Aldrich) with bromopropane (Aldrich) followed by metathesis with LiNTf2. Solid LiNTf2 and LiPF6 salts were dissolved in the liquid salt MePrPyrNTf2 (0.7 M LiNTf2 in MePrPyrNTf2) and liquid TMS heated to ca. 35 °C (1 M LiPF6 in TMS), respectively. Electrolytes containing GBL (0.7 M LiNTf2 in MePrPyrNTf2 + 10 wt% GBL) and VC (1 M LiPF6 in TMS + 5 wt% VC) and a classical electrolyte (1 M LiPF6 in PC + DMC, 1:1) were prepared in a dry argon atmosphere in a glove box. Tested anodes were prepared on a copper foil (Hohsen, Japan) by a casting technique, from a slurry of Li4Ti5O12, CB and PVdF in NMP. The ratio of components was Li4Ti5O12:CB:PVdF = 85:5:10 (by weight). After solvent (NMP) vacuum evaporation at 120 °C, a layer of the anode was formed, containing the active material (Li4Ti5O12), an electronic conductor (CB) and the binder (PVdF).

2.2 Procedures and measurements

Particle size of Li4Ti5O12 was determined with Zetasizer Nano ZS (Malvern Instruments Ltd., UK). Specific surface areas of Li4Ti5O12 powders were determined by N2 adsorption (BJH method) using an ASAP 2020 instrument (Micromeritics Instrument Co.). The performance of the cells was characterized using galvanostatic charge–discharge tests, cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS). Cycling efficiency of Li4Ti5O12|Li systems was measured in two compartment cells. Electrodes (metallic-lithium foil and Li4Ti5O12) were separated by the glass microfiber GF/A separator (Whatman, 0.4 mm thick), placed in an adopted Swagelok® connecting tube. Typically, the mass of electrodes was as follows: Li: ca. 45 mg (0.785 cm2) and Li4Ti5O12: 2.5–3.5 mg. Cells were assembled in a glove box in the dry argon atmosphere. Cycling measurements were taken with the use of the ATLAS 0461 MBI multichannel electrochemical system (Atlas-Sollich, Poland) at different current rates (C/20–C/3). CV and ac impedance measurements were performed using the μAutolab FRA2 type III electrochemical system (Ecochemie, the Netherlands). Flash point of electrolytes was measured with an open cup home-made apparatus, based on the Cleveland open cup instrument, with a 1.5 ml cup. The cup was heated electrically through a sand bath and temperature was measured with the M-3850 digital thermometer (Metex, Korea). The apparatus was scaled with a number of compounds of known flash points. The thermal behaviour of electrolytes in the temperature range of 20–300 °C was studied by DSC. A differential scanning calorimeter model DSC XP-10 (Thass GmbH) was used. After electrochemical measurements, cells were disassembled in a glove box, the electrodes were washed with DMC and dried in vacuum at room temperature. The morphology of the Li4Ti5O12 electrode (pristine and after electrochemical cycling) was observed under a scanning electron microscope (SEM, Tescan Vega 5153).

3 Results and discussion

3.1 Cycle voltammetry

Figure 1 presents cyclic voltammograms (CVs) of: Li4Ti5O12/1 M LiPF6 in TMS + 5 wt% VC/Li, Li4Ti5O12/0.7 M LiNTf2 in MePrPyrNTf2 + 10 wt% GBL/Li and Li4Ti5O12/1 M LiPF6 in PC + DMC (1:1)/Li cells obtained in the potential window of 1.0–2.5 V at a scan rate of 0.1 mV s−1. Each curve clearly demonstrates that there is one pair of redox peaks in the range of 1.0–2.5 V, which is in accordance with the typical CV characteristics of spinel Li4Ti5O12 [36]. Oxidation peaks at about 1.65 V and the reduction peak at 1.5 V may be attributed to the oxidation/reduction reactions of the Ti3+/Ti4+ couple in the spinel structure, compensated by lithium extraction–insertion (Fig. 1c). However, compared with Li4Ti5O12 in classical LiPF6 solutions in cyclic carbonates, the peak currents and separation of the peak area of Li4Ti5O12 in sulfolane and IL electrolytes are lower and much larger, respectively (Fig. 1a, b). The difference in currents between anodic and cathodic peaks can be mainly attributed to an increase in viscosity of electrolytes (TMS and IL) [37]. In turn, the slow lithium ion diffusivity in the solid-state body of bulk spinel Li4Ti5O12 may be caused by low lithium salt dissociation and stronger lithium salvation.

Fig. 1
figure 1

Cyclic voltammograms of the spinel Li4Ti5O12 in (a) 1 M LiPF6 in TMS + 5 wt% VC, (b) 0.7 M LiNTf2 in MePrPyrNTf2 + 10 wt% GBL and (c) 1 M LiPF6 in PC + DMC (1:1). Scan rate 0.1 mV s−1. Counter electrode Li

Full size image

3.2 Galvanostatic charging/discharging

Figure 2 shows charge/discharge curves for 2, 10 and 50 cycles for 1 M LiPF6 in TMS + 5 wt% VC, 0.7 M LiNTf2 in MePrPyrNTf2 + 10 wt% GBL and 1 M LiPF6 in PC + DMC (1:1) electrolytes. When sulfolane was used as an electrolyte solvent (Fig. 2a), a capacity of 128 mAh g−1 was obtained during the second cycle. This value decreased slightly after subsequent cycles, then finally reached a stable capacity of 120 mAh g−1. This indicates that the capacity after 50 cycles of charge/discharge decreases by about 9 %. A similar situation was found in the case of MePrPyrNTf2 (Fig. 2b). The highest capacity for both intercalation and deintercalation was found in the second cycle of the process (174 and 143 mAh g−1, respectively). A 100 % reversibility of the cells was reached after only ten cycles. During at the first three cycles formed of so-called SEI layer. Its formation is always associated with a loss of capacity.

Fig. 2
figure 2

Galvanostatic charging and discharging of (a) Li4Ti5O12/1 M LiPF6 in TMS + 5 wt% VC/Li, (b) Li4Ti5O12/0.7 M LiNTf2 in MePrPyrNTf2 + 10 wt% GBL/Li and (c) Li4Ti5O12/1 M LiPF6 in PC + DMC (1:1)/Li system (second, tenth, fiftieth cycles) at C/20 rate

Full size image

In subsequent cycles of the charge and discharge processes, the capacity decreased slightly to reach (after 50 cycles) the value of 120 and 123 mAh g−1, respectively. The fastest stabilization of the capacity was observed (Fig. 2c) in the case of the classical electrolyte (1 M LiPF6 in PC + DMC). It occurred after ten charge/discharge cycles, maintaining the same value for both the processes (ca. 110 mAh g−1, which is 63 % of the theoretical capacity). Charging/discharging curves for the Li4Ti5O12/1 M LiPF6 in TMS + 5 wt% VC/Li cell are shown in Fig. 3a. The charging and discharging capacity of the Li4Ti5O12 anode was between 160 and 80 mAh g−1 for different rates (C/2, C/5, C/10 and C/20). In the case of a system working on the basis of a TMS solution, the best value was obtained at a C/5 rate and an average of more than 150 mAh g−1 (86 % of theoretical capacity). The highest capacitance was observed for the Li4Ti5O12/0.7 M LiNTf2 in MePrPyrNTf2 + 10 wt% GBL/Li (Fig. 4a) at a C/20 rate (approx. 150 mAh g−1). When the current regime was increased to C/3, that value decreases to 60 mAh g−1. In contrast, for Li4Ti5O12/1 M LiPF6 in PC + DMC (Fig. 5a) the best capacities were obtained at C/10 and C/3, amounting to 145 and 154 mAh g−1, respectively. The efficiency of the process of insertion and deinsertion was 98 %.

Fig. 3
figure 3

Charging/discharging capacity (a) and Coulombic efficiency (b) of the Li4Ti5O12/1 M LiPF6 in TMS + 5 wt% VC/Li cell

Full size image
Fig. 4
figure 4

Charging/discharging capacity (a) and Coulombic efficiency (b) of the Li4Ti5O12/0.7 M LiNTf2 in MePrPyrNTf2 + 10 wt% GBL/Li cell

Full size image
Fig. 5
figure 5

Charging/discharging capacity (a) and Coulombic efficiency (b) of the Li4Ti5O12/1 M LiPF6 in PC + DMC (1:1)/Li cell

Full size image

That is to say, the structure of Li4Ti5O12 is not destroyed during the charge/discharge cycling. Probably the loss of capacity fluctuation on the capacity-cycle profile at this high current rate also suggests that Li+ insertion/extraction only takes place in the Li4Ti5O12 grains at near the outer surface of the highly aggregated particles. The capacity is much low at high current rates, and needs to be improved.

Coulombic efficiency of the process was found to be close to 95–99 % (±4 %), depending on the discharge rate and it was identical for all the three electrolytes (Figs. 3b, 4b, 5b). However, it can be seen that in the case of the lowest C/10 rate it was 90 % for the TMS-based electrolyte (Fig. 1a).

3.3 Impedance spectroscopy (SEI formation)

Anodes characterized by a low potential, such as lithium metal or lithiated graphite, react spontaneously with electrolytes. In the case of lithium metal, the growth of dendrite crystals on its surface is observed. It has also been shown that lithiated graphite requires the formation of a protective coating, similarly to metallic lithium. Consequently, the electrolyte for Li-ion batteries is expected to form SEI, protecting the lithium anode [38]. A similar phenomenon has also been found for the Li4Ti5O12 anode. A solid electrolyte interface can be formed on the Li4Ti5O12 anode after several cycles. This is different in the case of a graphite anode where the SEI film forms during the initial cycle [39, 40]. The passivation of electrodes in a Li-ion battery may be observed with the help of impedance spectroscopy. Electrochemical impedance spectra (EIS) of the LTO/electrolyte/Li were measured before discharge and after 20 cycles (at C/10 and C/2 rates). The impedance spectra were recorded in two electrode cells using Li counter. Then, the EIS results also include impedance of lithium interface. Figure 6 shows impedance of three: Li4Ti5O12/0.7 M LiNTf2 in MePrPyrNTf2 + 10 wt% GBL/Li, Li4Ti5O12/1 M LiPF6 in TMS + 5 wt% VC/Li and Li4Ti5O12/1 M LiPF6 in PC + DMC/Li unsymmetrical cells before cycling. As it can be seen, spectra taken immediately after the cell assembling consisted of a semicircle, followed by a long linear part. The semicircle can be attributed to phenomena such as polarization resistance and the formation of the SEI layer. The distinct linear part is due to the diffusion of Li+ in the electrolyte and SEI, or Li in a solid electrode (insertion/extraction). It may be noted that the high lithium diffusion resistance in the anode material may be the rate-determining step. The diffusion coefficient D of the lithium ions diffusing in the Li4Ti5O12 is ca. 1 × 10−10–1 × 10−11 cm2 s−1 [13]. However, after galvanostatic charging/discharging cycles at different rates (Figs. 7, 8, 9) a flat quasi-semicircle or two semicircles can be seen at the high-frequency region and a short straight line at the low-frequency region. The first semicircle reflects the formation of the SEI layer and the second the charge transfer process, while the line at the low-frequency region represents the diffusion process. The equivalent circuit used for impedance spectra deconvolution (Fig. 10) consists of two RC elements describing resistance of SEI (R SEI) and resistance of the charge transfer reaction (R ct), in series with electrolyte resistance R el and diffusion impedance represented by the Warburg element (Z W). It can be seen that impedance for all the three increases with the current rate increase. Charge transfer resistances (R ct) (Fig. 8) were found to be 1,550 and 2,890 Ω for Li4Ti5O12/0.7 M LiNTf2 in MePrPyrNTf2 + 10 wt% GBL/Li cell at C/10 and C/2 rates, respectively, indicating a high kinetic barrier. However, a much smaller increase of impedance was observed in the case of the 1 M LiPF6 in TMS + 5 % VC (just 120 Ω at C/10 rate) system (Fig. 7). Then, the SEI resistance and the charge transfer resistance were 31 and 14 Ω at a C/10 rate, respectively. In the case of the 1 M LiPF6 in PC + DMC (1:1) electrolyte impedance spectrum had a shape of a semicircle followed by a line (Fig. 9) when the system was charged/discharged at a low rate (C/10). However, after charging/discharging at higher rates (C/2 rate) two semicircles were present in the spectrum. The charge transfer resistance and the SEI resistance were 104 and 12 Ω at a C/2 rate, respectively. The particle morphologies of pristine Li4Ti5O12 and Li4Ti5O12 after 50 cycles of galvanostatic charging/discharging at different electrolytes observed by SEM are shown in Figs. 11 and 12, respectively. The pristine LTO powders (Fig. 11) had a spherical shape, dense structure and non-aggregation characteristics. The size of the LTO particles was in the micron range (0.5–2 μm). As shown in Fig. 12c, d, the spherical morphology and microstructures slightly changed after galvanostatic charging/discharging at 0.7 M LiNTf2 in MePrPyrNTf2 + 10 wt% GBL (changes are visible only at higher magnifications). However, SEM images of the LTO particles after galvanostatic charging/discharging at 1 M LiPF6 in TMS + 5 % VC and 1 M LiPF6 in PC + DMC (1:1) in Fig. 12a, b, e, f show that they were coated by a uniform layer. Probably this film can be identified as the SEI layer.

Fig. 6
figure 6

Impedance spectroscopy of (a) Li4Ti5O12/0.7 M LiNTf2 in MePrPyrNTf2 + 10 wt% GBL/Li, (b) Li4Ti5O12/1 M LiPF6 in TMS + 5 wt% VC/Li and (c) Li4Ti5O12/1 M LiPF6 in PC + DMC (1:1)/Li systems recorded after cell assembling (OCV ~3.1 V). Frequency range 105–10−2 Hz

Full size image
Fig. 7
figure 7

Impedance spectroscopy of the Li4Ti5O12/1 M LiPF6 in TMS + 5 wt% VC/Li cell after 20 cycles of galvanostatic charging/discharging at various rates: (a) C/10 (OCV 1.59 V) and (b) C/2 (OCV 1.56 V). Frequency range 105–10−2 Hz

Full size image
Fig. 8
figure 8

Impedance spectroscopy of the Li4Ti5O12/0.7 M LiNTf2 in MePrPyrNTf2 + 10 wt% GBL/Li cell after 20 cycles of galvanostatic charging/discharging at various rates: C/10 (OCV 1.51 V) and C/2 (OCV 1.59 V). Frequency range 105–10−2 Hz

Full size image
Fig. 9
figure 9

Impedance spectroscopy of the Li4Ti5O12/1 M LiPF6 in PC + DMC (1:1)/Li cell after 20 cycles of galvanostatic charging/discharging at various rates: C/10 (OCV 1.56 V) and C/2 (OCV 1.63 V). Frequency range 105–10−2 Hz

Full size image
Fig. 10
figure 10

An equivalent circuit representing the electrode/electrolyte system

Full size image
Fig. 11
figure 11

SEM images of the pristine Li4Ti5O12 anode. Magnification ×5,000 (a) and ×20,000 (b)

Full size image
Fig. 12
figure 12

SEM images of the Li4Ti5O12 anode after 20 charge/discharge cycles at C/10 rate. Electrolyte: 1 M LiPF6 in TMS + 5 wt% (a, b), VC 0.7 M LiNTf2 in MePrPyrNTf2 + 10 wt% GBL (c, d) and 1 M LiPF6 in PC + DMC (1:1) (e, f). Magnification: ×5,000 (a, c, e) and ×20000 (b, d, f)

Full size image

3.4 Electrolyte ignition point

The flash point of ethylene carbonate and propylene carbonate is 143 and 132 °C, respectively, while in the case of DMC it is only 16 °C (data from a Sigma-Aldrich catalogue). The flash point of neat sulfolane is much higher 177 °C [41] and the MePrPyrNTf2 is a non-volatile solvent. Only SEI-forming additives (VC, GBL) may increase volatility of the investigated electrolytes. Flash points of the VC and the GBL are 73 and 98 °C, respectively (Sigma-Aldrich catalogue). On the other hand, during SEI formation the volatile additive is converted into a solid polymeric component of the interface. Probably, also GBL molecules in 0.7 M LiNTf2 in MePrPyrNTf2 are involved in ion salvation. Consequently, the amount of the volatile compound decreases to a low value, increasing the flash point of the electrolyte. The flash point determined for the LiNTf2 solution in MePrPyrNTf2, with GBL is 152 °C. In turn, the system under study, i.e., 1 M LiPF6 + TMS + 5 wt% VC, shows a flash point of ca. 160 °C. Figure 13 presents thermal gravimetric analysis (DSC) of the (a) Li4Ti5O12 + 1 M LiPF6 + TMS + 5 wt% VC; (b) Li4Ti5O12 + 0.7 M LiNTf2 in MePrPyrNTf2 + 10 wt% GBL; (c) Li4Ti5O12 + 1 M LiPF6 in PC + DMC (1:1) electrolyte under nitrogen atmosphere. In the case of the electrolyte based on sulfolane (Fig. 13a), two peaks can be observed at 160 °C (2.81 J g−1) and 180 °C (28.1 J g−1), corresponding to the boiling point of VC (162 °C) and flash point of TMS (177 °C), respectively. Energy transformation in the vicinity of 140–150 °C amounting to 19.32 J g−1 can probably answer GBL ignition of the temperature flash point when added to 0.7 M LiNTf2 in MePrPyrNTf2 liquid (Fig. 13b). In the case of the classical electrolyte, again two peaks can be seen (15.32 and 36.6 J g−1) at a temperature of 70–80 and 90–110 °C, which is probably associated with boiling temperatures of solvents used (PC and DMC, Fig. 13c) [41].

Fig. 13
figure 13

DSC profiles related to the heating test of the (a) Li4Ti5O12 + 1 M LiPF6 + TMS + 5 wt% VC, (b) Li4Ti5O12 + 0.7 M LiNTf2 in MePrPyrNTf2 + 10 wt% GBL, (c) Li4Ti5O12 + 1 M LiPF6 in PC + DMC (1:1) electrolyte (nitrogen atmosphere)

Full size image

4 Conclusions

LTO have a working potential at 1.55 V against lithium which is above the voltage where electrochemical decomposition of the electrolyte takes place for most of the common electrolytes. This work is connected with testing the performance of work, a Li4Ti5O12 anode using a non-flammable electrolyte with a low vapour pressure, such as 1 M LiPF6 + TMS + 5 wt% VC and 0.7 M LiNTf2 in MePrPyrNTf2 + 10 wt% GBL. The Li4Ti5O12/Li batteries were investigated by EIS and SEM techniques.

  1. 1.

    SEM images of LTO particles after galvanostatic charging/discharging show that they were covered with relatively uniform layers with conglomerates, which can be identified as a protective SEI coating.

  2. 2.

    Capacity of the LTO anode depends on the current rate (C/20, C/10, C/5 or C/3) and the electrolyte. In the case of a system working on the basis of a TMS solution the best value was obtained at a C/5 current and an average of more than 150 mAh g−1 (86 % of theoretical capacity). For Li4Ti5O12/0.7 M LiNTf2 in MePrPyrNTf2 + 10 wt% GBL/Li the largest capacitance value may be observed at a C/20 current (it is about 150 mAh g−1) and it decreases at higher currents. However, for Li4Ti5O12/1 M LiPF6 in PC + DMC (1:1) the best capacity was obtained at C/10 and C/3, amounting to 145 and 154 mAh g−1 (which represents about 85 % of the theoretical capacity). Coulombic efficiency of the discharging process was found to be close to 95–99 % (±4 %), depending on the discharge rate and was the same for all three electrolytes. All the results obtained indicate that LTO anode may be successfully used in lithium-ion batteries.

  3. 3.

    The 1 M LiPF6 in TMS + 5 wt% VC and 0.7 M LiNTf2 in MePrPyrNTf2 + 10 wt% GBL electrolytes had a greater thermal stability and no DSC peak was observed below 150 °C. The classical electrolyte (1 M LiPF6 in PC + DMC) was stable only to a temperature of about 75 °C.

These results suggest that IL (MePrPyrNTf2) and sulfolane (TMS) electrolytes were a good candidate to be safely used as an electrolyte in LIBs.