Design of Composites Based on Lithium Titanate and Carbon Nanomaterials for High-Power Lithium-Ion Batteries †

: The composite nanomaterials based on hydrothermally or sol-gel synthesized Li 4 Ti 5 O 12 and both nitrogen-doped and undoped carbons were prepared and investigated by XRD, SEM, low-temperature nitrogen adsorption, dc-measurements, charge-discharge tests, cyclic voltammetry, and electrochemical impedance spectroscopy. Carbon nanotube introduction provides the formation of a highly conductive 3D network resulting in increase in electronic conductivity, lithium diffusion, and thus improvement of rate-capability and cycling stability of the composites.


Introduction
Lithium-ion batteries are widely used in electronic devices, first of all, owing to their high energy density, low self-discharge, and relatively small weight. Li 4 Ti 5 O 12 is considered as a promising anode material for lithium-ion batteries, primarily due to low degradation during cycling and high safety. However, its electronic and ionic conductivities are relatively low that limits its practical application. The design of composites with conductive materials (e.g., with carbon), is one of the approaches to improve the electrochemical properties of materials. In this work, the Li 4 Ti 5 O 12 /carbon nanomaterial composites were prepared by ball-milling. Different synthesis methods (sol-gel and hydrothermal) of lithium titanate as well as both nitrogen-doped and undoped carbons were used to compare their effect on specific electrochemical capacity and rate capability of the obtained composites.

Materials and Methods
Lithium titanate was prepared by a sol-gel method (LTO sg ) in the presence of citric acid (final calcination temperature 800 • C) or hydrothermally (LTO H , final calcination temperature 400 • C), using procedures described elsewhere [1,2]. To prepare the LTO/C composites a mixture of lithium titanate and 5 or 10 wt% carbon nanomaterial was ballmilled at 200 rpm for 1 h and 5 min for LTOsg and LTO H , respectively. Carbon nanotubes (CNTs), agglomerates of graphene particles (carbon nanoflakes, CNFs), N-doped ones (N-CNTs and N-CNFs, respectively) or CNTs (CNFs) together with 5 wt% carbon black (CB)) were used as carbon nanomaterials. Hereinafter, the obtained composites are denoted as LTO H /5CNT, LTOH/10N-CNF, LTOsg/5CB/5N-CNT, etc. indicating the carbon nanomaterial used and its amount.
X-ray diffraction (Rigaku D/MAX 2200 diffractometer with CuK α radiation) was used to examine the crystal structure. Their morphology and particle size were studied by scanning electron microscopy using a Carl Zeiss NVision 40. The specific surface area of the samples was measured by BET (Brunauer-Emmett-Teller) method at -196 • C on a Sorbtometr-M instrument.
Electrochemical characteristics were studied using hermetically sealed three-electrode electrochemical cells with lithium as counter and reference electrodes. The working electrodes were prepared by pasting a mixture of the composite (85 wt%), carbon black (10 wt%), and polyvinylidene fluoride binder (5 wt%) dissolved in N-methyl-2-pyrrolidone on a stainless-steel grid (to be used as a current collector). The amount of the active material on the electrode was 10-15 mg/cm 2 . The electrodes were pressed at 0.1 GPa and dried at 120 • C under vacuum. The Celgard 2300 polypropylene membrane and 1 M LiPF 6 solution in a 1:1:1 mixture of ethylene carbonate, diethyl carbonate, and dimethyl carbonate were used as the separator and the electrolyte, respectively. The cells (lithium metal|electrolyte|active material) were assembled in an argon-filled glove box and cycled using a ZRU 50 mA-10 V charge-discharge system (JSC NTTs Buster) at potentials in the range of 1-3 V and at current densities in the range of 20-6400 mA/g. Cyclic voltammograms were recorded using an Elins P-8NANO electrochemical workstation at different scanning rates (0.1-1.6 mV/s). Electrochemical impedance spectra were obtained using an Elins Z500 PRO analyzer in the frequency range of 10 Hz to 2 MHz at an amplitude of 80 mV.

Material Characterization
According to PDF-2 database file no. 72-0426, all of the diffraction peaks in the X-ray diffraction patterns of the composites based on LTO H and LTO sg can be assigned to the cubic Li 4 Ti 5 O 12 (Figure 1). In the case of the LTO H /carbon composites, a weak reflection at 2Θ ≈ 24-26 • results from the (002) crystallographic plane of graphene [3]. In the case of the composites based on LTO sg , no reflections of graphene and carbon nanotubes were detected. Apparently, this is due to the low content of carbon materials in the composites and/or the low intensity of their reflexes against the reflection (111) of LTO sg .
Eng. Proc. 2022, 19, 3 2 of 5 X-ray diffraction (Rigaku D/MAX 2200 diffractometer with CuKα radiation) was used to examine the crystal structure. Their morphology and particle size were studied by scanning electron microscopy using a Carl Zeiss NVision 40. The specific surface area of the samples was measured by BET (Brunauer-Emmett-Teller) method at -196 °С on a Sorbtometr-M instrument.
Electrochemical characteristics were studied using hermetically sealed three-electrode electrochemical cells with lithium as counter and reference electrodes. The working electrodes were prepared by pasting a mixture of the composite (85 wt%), carbon black (10 wt%), and polyvinylidene fluoride binder (5 wt%) dissolved in N-methyl-2-pyrrolidone on a stainless-steel grid (to be used as a current collector). The amount of the active material on the electrode was 10-15 mg/cm 2 . The electrodes were pressed at 0.1 GPa and dried at 120 °С under vacuum. The Celgard 2300 polypropylene membrane and 1 M LiPF6 solution in a 1:1:1 mixture of ethylene carbonate, diethyl carbonate, and dimethyl carbonate were used as the separator and the electrolyte, respectively. The cells (lithium metal|electrolyte|active material) were assembled in an argon-filled glove box and cycled using a ZRU 50 mA-10 V charge-discharge system (JSC NTTs Buster) at potentials in the range of 1-3 V and at current densities in the range of 20-6400 mA/g. Cyclic voltammograms were recorded using an Elins P-8NANO electrochemical workstation at different scanning rates (0.1-1.6 mV/s). Electrochemical impedance spectra were obtained using an Elins Z500 PRO analyzer in the frequency range of 10 Hz to 2 MHz at an amplitude of 80 mV.

Material Characterization
According to PDF-2 database file no. 72-0426, all of the diffraction peaks in the X-ray diffraction patterns of the composites based on LTOH and LTOsg can be assigned to the cubic Li4Ti5O12 ( Figure 1). In the case of the LTOH/carbon composites, a weak reflection at 2Θ ≈ 24-26° results from the (002) crystallographic plane of graphene [3]. In the case of the composites based on LTOsg, no reflections of graphene and carbon nanotubes were detected. Apparently, this is due to the low content of carbon materials in the composites and/or the low intensity of their reflexes against the reflection (111) of LTOsg.  Figure 2 illustrates the SEM images of the composites based on lithium titanate prepared hydrothermally and by sol-gel method. In the composites based on LTOH and carbon nanotubes, the latters are homogeneously distributed over the LTOH surface, indicating the formation of a well-connected conductive network on the lithium titanate particles (Figure 2b). Unlike the composites based on LTOH, carbon nanotubes are une-  The BET surface area is 598, 287, 1198, and 1028 m 2 /g for CNTs, N-CNTs, CNFs, and N-CNFs, respectively. Undoped CNTs and CNFs have larger BETsurface areas than that of nitrogen-doped carbon nanomaterials. The specific surface area of LTOH and LTOsg are 103 and 5-7 m 2 /g, respectively. Among the obtained composites, the composites with CNTs have the highest BET surface area.
Ball-milling of lithium titanate with CNTs or CNFs leads to a pronounced increase in the dc conductivity of the obtained composites. The dc conductivity of the composites based on LTO prepared hydrothermally increases in the series LTOH << LTOH/N-CNF < LTOH/CNF < LTOH/СNT < LTOH/N-СNT. Electron conductivity of the LTOH/СNT com- The BET surface area is 598, 287, 1198, and 1028 m 2 /g for CNTs, N-CNTs, CNFs, and N-CNFs, respectively. Undoped CNTs and CNFs have larger BETsurface areas than that of nitrogen-doped carbon nanomaterials. The specific surface area of LTO H and LTO sg are 103 and 5-7 m 2 /g, respectively. Among the obtained composites, the composites with CNTs have the highest BET surface area.
Ball-milling of lithium titanate with CNTs or CNFs leads to a pronounced increase in the dc conductivity of the obtained composites. The dc conductivity of the composites based on LTO prepared hydrothermally increases in the series LTO H << LTO H /N-CNF < LTO H /CNF < LTO H /CNT < LTO H /N-CNT. Electron conductivity of the LTO H /CNT composite is 7 times lower than that of the LTO H /nitrogen-doped CNT composite (1.33 S/cm). In the case of the composites with carbon nanoflakes, electron conductivity is much lower, due to a non-effective contact between CNFs and lithium titanate particles (9.6 × 10 −5 and 6.7 × 10 −5 S/cm for the composites with CNFs and nitrogen-doped CNFs, respectively). The dc conductivity of the composites based on LTO sg increases from 0.01 to 0.36 S/cm in the series LTO sg /10N-CNF < LTO sg /10CNF < LTO sg /10CNT ≈ LTO sg /5CB/5N-CNT < LTO sg /10N-CNT.
The Nyquist plots for the LTO H -based composites consist of a semicircle (correspond to the charge transfer process at the electrode/electrolyte interface) and a straight line, which represents lithium-ion diffusion (the Warburg impedance) in the high-and lowfrequency regions, respectively. The Li + diffusion coefficients of LTO H , LTO H /5CNT, and LTO H /5CNF calculated using the Warburg impedance are 1.1 × 10 −13 , 1.2 × 10 −12 , and 8.0 × 10 −14 cm 2 /s, respectively. Figure 3a,b show cycling performance of the composites based on Li 4 Ti 5 O 12 prepared hydrothermally and by sol-gel method, respectively. At a low current density (20 mA/g), the discharge capacities of all of the composites are in the range of 140-170 mAh/g. Taking into account the low electrochemical capacity of the carbon nanomaterials in the potential range of 1-3 V and their content in the composites, the discharge capacity of the LTO H /CNT, LTO sg /10CNF and LTO sg /5CB/5N-CNT composites approaches the theoretical value of Li 4 Ti 5 O 12 (175 mAh/g). At high current densities, the best rate capability was found for the composites with CNTs.

Electrochemical Performance
Eng. Proc. 2022, 19, 3 4 of 5 posite is 7 times lower than that of the LTOH/nitrogen-doped CNT composite (1.33 S/cm). In the case of the composites with carbon nanoflakes, electron conductivity is much lower, due to a non-effective contact between CNFs and lithium titanate particles (9.6 × 10 −5 and 6.7 × 10 −5 S/cm for the composites with CNFs and nitrogen-doped CNFs, respectively). The dc conductivity of the composites based on LTOsg increases from 0.01 to 0.36 S/cm in the series LTOsg/10N-CNF < LTOsg/10CNF < LTOsg/10CNT ≈ LTOsg/5CB/5N-CNT < LTOsg/10N-CNT. The Nyquist plots for the LTOH-based composites consist of a semicircle (correspond to the charge transfer process at the electrode/electrolyte interface) and a straight line, which represents lithium-ion diffusion (the Warburg impedance) in the high-and low-frequency regions, respectively. The Li + diffusion coefficients of LTOH, LTOH/5CNT, and LTOH/5CNF calculated using the Warburg impedance are 1.1 × 10 −13 , 1.2 × 10 −12 , and 8.0 × 10 −14 cm 2 /s, respectively. Figure 3a,b show cycling performance of the composites based on Li4Ti5O12 prepared hydrothermally and by sol-gel method, respectively. At a low current density (20 mA/g), the discharge capacities of all of the composites are in the range of 140-170 mAh/g. Taking into account the low electrochemical capacity of the carbon nanomaterials in the potential range of 1-3 V and their content in the composites, the discharge capacity of the LTOH/CNT, LTOsg/10CNF and LTOsg/5CB/5N-CNT composites approaches the theoretical value of Li4Ti5O12 (175 mAh/g). At high current densities, the best rate capability was found for the composites with CNTs.

Conclusions
The composites based on Li4Ti5O12 synthesized by hydrothermal and sol-gel methods and carbon nanomaterials were prepared by a simple ball-milling method. The introduction of latters leads to an increase in the electrical conductivity of the composites and the reversible discharge capacity even at high charge/discharge rates. In the composites with carbon nanotubes, as latters can bridge Li4Ti5O12 particles a high-conductive network forms providing fast electron and lithium transport. In the case of lithium titanate prepared by sol-gel method, the highest discharge capacity (97 mAh/g at ~18C rate) is observed for composites with CNTs and carbon black. In the case of Li4Ti5O12 prepared hydrothermally, at ~18C rate, the discharge capacities of lithium titanate with CNTs and N-doped CNTs are 107 and 114 mAh/g, respectively. The use of carbon nanoflakes is less efficient since their large agglomerates are retained in the composite

Conclusions
The composites based on Li 4 Ti 5 O 12 synthesized by hydrothermal and sol-gel methods and carbon nanomaterials were prepared by a simple ball-milling method. The introduction of latters leads to an increase in the electrical conductivity of the composites and the reversible discharge capacity even at high charge/discharge rates. In the composites with carbon nanotubes, as latters can bridge Li 4 Ti 5 O 12 particles a high-conductive network forms providing fast electron and lithium transport. In the case of lithium titanate prepared by sol-gel method, the highest discharge capacity (97 mAh/g at~18C rate) is observed for composites with CNTs and carbon black. In the case of Li 4 Ti 5 O 12 prepared hydrothermally, at~18C rate, the discharge capacities of lithium titanate with CNTs and N-doped CNTs are 107 and 114 mAh/g, respectively. The use of carbon nanoflakes is less efficient since their large agglomerates are retained in the composite even after ball-milling and, hence, the composite is less uniform and there is no a network of highly conductive contacts between the lithium titanate particles.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.