A Low-Cost and Li-Rich Organic Coating on Li 4 Ti 5 O 12 Anode Material Enabling Li-Ion Battery Cycling at Subzero Temperatures

In this paper, we report the surface modification of Li 4 Ti 5 O 12 (LTO) anode material with a freshly prepared Li-rich PTCLi 4 organic molecule by a spray-dryer technique. In addition, burning the resulting powder yielded an electrode material with a few-nanometer-thick carbon coating. For comparison, carbon-coated LTO powder was prepared with graphene oxide (GO) using the same protocol. Organic molecules were first characterized using FTIR, XPS, TGA, XRD, and SEM methods. PTCLi 4 -coated LTO powders were observed by SEM and the corresponding EDX mapping as well as micro-Raman and XPS spectroscopic analyses confirmed the efficient surface coverage of the anode material. After the burning, a graphitic-like carbon coating with an I D /I G of approximately 0.76 and a thickness of a few nanometers was confirmed by TEM observations. Thermogravimetric analyses revealed that the content of carbon varied from 0.3 to 1.5 wt.%, depending on the reaction conditions and material used (i.e., PTCLi 4 or GO). Interestingly, electrochemical cycling at 25 °C of PTCLi 4 -coated LTO electrodes gave superior performance compared to the pristine electrode, especially at high C-rates, and carbon-coated electrodes showed intermediate performance. Most importantly, good cyclability of PTCLi 4 -coated LTO electrodes was observed with a specific capacity of 145 mAh·g −1 after 100 cycles at a C/2 rate with an average coulombic efficiency of 100%. XPS analyses performed on aged electrodes revealed less degradation of the electrolyte with a lower concentration of LiF on the surface of PTCLi 4 -coated LTO electrodes. Finally, cycling of LTO electrodes demonstrated the potential of using PTCLi 4 coating to increase the Li-ion transfer at the electrode–electrolyte interface at subzero temperatures. In fact, the PTCLi 4 -coated LTO electrode delivered almost the same specific capacity at a C/2 rate when cycled at –20 °C as the pristine electrode cycled at 25 °C. Scheme 2. Schematic of the effect of a) PTCLi 4 coating and carbon coatings generated by thermal decomposition b) PTCLi 4 , and c) GO layers on the electron conduction and Li-ion diffusion for the corresponding LTO composites. Red crosses indicate that the Li + or e − transfers are slowed down.


Introduction
Since the introduction of a Li-ion battery to the market by Sony in 1991 [1], battery use cases have proliferated; now, owing to radical changes in the environment, electrical vehicle (EV) and energy storage (ES) are becoming increasingly popular for limiting greenhouse gas emissions and climate change. Therefore, many materials have been investigated to enhance the performance of cells, irrespective of the operation conditions or use cases.
It is believed that lithium titanium oxide (LTO) is an ideal candidate material to meet all the requirements for a battery system, including safety, long cycle life, and rapid charging/discharging [2][3][4][5][6][7]. However, one aspect remains as an obstacle preventing the realization of its full potential, namely, the intrinsic limitations of LTO, such as its low electrical conductivity and poor electrochemical performance temperatures below 0 °C. LTO is an insulator material and several strategies have been explored to overcome its low ionic and electronic conductivities, including carbon coating [8][9][10][11][12][13], miniaturization of particles [14][15][16][17][18][19], and doping [20,21]. However, despite these efforts, the low-temperature performance is still inadequate and can be a source of danger [22], especially when low-temperature electrolytes are used [6,22].
Many groups have proposed alternative strategies to enhance ionic conductivity and lithium diffusion. Controlling the nature of the binder was one of the economical ways to enhance the electrochemical performance of electrodes at high C-rates and low temperatures. For example, poly(acrylic acid) neutralized with Li can effectively reduce the resistance of an electrode, enhancing its adhesion and the cycle life [23][24][25]. Another recurrent problem is the incapability to effectively disperse carbon and active particles and create a uniform solid electrolyte interface 4 (SEI) on active particles [26][27][28]. The perfect covering of all single LTO particles reported by Daigle et al. [13] dramatically enhanced the cell performance, leading to efficient access to all active particles by the electrolyte that maximized deliverable capacity of the LTO anode. Covering LTO particles with polymers [3,29] was effective to prevent gas evolution during cell operation; however, the polymer acts as a barrier and impedes lithium diffusion, thereby adversely affecting fast charge/discharge performance. In addition, the polymer is often not electrochemically active and does not contribute to the cell performance. Organic molecules demonstrated faster lithium ion transfer, as observed in organic electrodes [30][31][32]. Therefore, the incorporation of electrochemically active organic molecules at the surface of LTO should be an interesting trend of research. Based on these observations, the use of single and active organic molecules, which can be well dispersed on active electrode particles by physical absorption (effective covering of the surface) must be suitable for achieving high-performance anode materials at high C-rates and low temperatures.
In this work, we realized a Li-rich organic coating on the surface of LTO anode material to increase its electrochemical performance at subzero temperatures by enhancing the Li-ion transfer at the electrode-electrolyte interface. A semiconducting perylene-based PTCLi 4 molecule was synthesized and coated on the surface of LTO by a spray-dryer technique. First, the successful synthesis of an organic molecule was verified using FTIR, XPS, TGA, XRD, and SEM analyses.
In parallel, the heat-treatment under an argon atmosphere of the resulting powder as well as graphene oxide (GO)-coated LTO yielded carbon-coated LTO materials. TEM, XPS, and micro-Raman investigations revealed that the LTO particles were surrounded by a nanometer-thick carbon layer in both cases, except that large GO sheets were observed for burned GO-coated LTO powder, thereby providing enhanced electronic conductivity for this composite. Carbon contents 5 ranging from 0.3 to 1.5 wt.%, depending on the reaction conditions and material used (i.e., PTCLi 4 or GO), were quantified by TGA. The electrochemical performance of the PTCLi 4 -coated LTO electrodes at 25 °C was better than that of the electrodes made with the same materials after burning, whereas the GO-coated LTO gave the best results. Interestingly, good cyclability of the PTCLi 4 -coated LTO electrodes was observed, as they delivered 145 mAh·g −1 after 100 cycles at a C/2 rate with an average coulombic efficiency of 100%. XPS analyses performed on aged electrodes revealed less electrolyte degradation with a lower concentration of LiF on the surface of the PTCLi 4 -coated LTO electrodes. Finally, the potential of using PTCLi 4 coating to increase the Li-ion transfer at the electrode-electrolyte interface at subzero temperatures was confirmed, as the PTCLi 4 -coated LTO electrode delivered almost the same specific capacity, at a C/2 rate when cycled at −20 °C, as the pristine electrode cycled at 25 °C.

Synthesis of graphene oxide
Graphene oxide powder (GO) was prepared by a modified Hummers method [33]. Synthetic globular graphite microspheres of MAG brand (~20 µm particle size, manufactured by Hitachi Kasei Coke & Chemical Co.) were utilized in our experiments. Approximately 5 g of this carbon was placed in a 250-mL round-bottom flask, followed by the addition of 5 g of NaNO 3 (Aldrich) and 200 mL of concentrated H 2 SO 4 (Fisher Chemical). The mixture was intensively mixed with a magnetic bar for at least 2 h and cooled to 0 °C with an ice-filled container. Subsequently, 30 g of KMnO 4 (Aldrich) was slowly added over a few minutes and the mixture was stirred at 40 °C for 2 h. Then, the mixture was transferred in a 2-L round-bottom flask, approximately 400 mL of deionized water (from a Milli-Q ® ultrapure lab water system) was poured into the flask, and the 6 temperature of the bath was increased to 95 °C. This mixture was magnetically stirred for 15 h.
After cooling the mixture to 25 °C, 800 mL of deionized water was added. After 1 h, 80 mL of H 2 O 2 (50 wt. %, Aldrich) was poured into the flask and the mixture was intensively mixed for 1 h before being vacuum-filtered using a Büchner assembly and a 0.22-µm-pore nylon filter. The synthesized gel-like product was washed several times with a 5% HCl solution, followed by washing with deionized water until a neutral pH was obtained. Subsequently, the brown product was dispersed in 250 mL of deionized water and placed in an ultrasonic bath for at least 2 h. Next, the volume of deionized water was adjusted to 3 L and it was left standing for 1 week for dialysis to remove residual metallic ions and acids. The resulting GO gel was isolated and concentrated by repeated centrifugation steps (5000 RPM, 30 min in 500-mL centrifuge tubes). The thick gel obtained was allowed to dry slowly for one week at 45 °C under constant vacuum to yield a brown GO powder (see Scheme 1 for its hypothetical chemical structure).

Synthesis of PTCLi 4
The one-pot synthesis of PTCLi 4 was inspired by the procedure reported by Fédèle et al. [B]. The rapid hydrolysis and lithiation of PTCDA (Aldrich) were realized in deionized water at 100 °C using LiOH⋅H 2 O (Aldrich) as a lithium source. PTCLi 4 powder was recovered by centrifugation and washed successively with deionized water and ethanol until a neutral pH was obtained. The yellow powder was dried in an oven at 60 °C prior to characterization. The chemical structures and colors of PTCDA and PTCLi 4 molecules are depicted in Scheme 1. 7 Scheme 1. Schematic representation of experiments: PTCLi 4 and GO powders were first synthesized before being utilized to make organic coatings on LTO surface by a spray-dryer method. After burning, carbon-coated LTO powders were obtained.

Organic coating on LTO particles
LTO anode material (POSCO Company) was covered with 2, 3, 4, or 5 wt. % of PTCLi 4 or GO powder by spray-drying, as represented in Scheme 1. To this end, 20 g of LTO powder was suspended in 100 mL of deionized water and intensively dispersed using a sonotrode for 30 min.
In parallel, a mass of GO or PTCLi 4 corresponding to 2, 3, 4, or 5 wt. % of that of LTO was dispersed in 100 mL of deionized water according to the same method. The two solutions were then combined and completed to 250 mL before being pumped and spray-dried. Spray-drying was performed on a Buchi B-290 mini spray-dryer. The inlet and outlet temperatures were set to 220 8 and 117 °C, respectively. During the entire process, the aspirator was functioning at 100% capacity. The sample solution was pumped at 10 mL/min and the spray gas flow in the nozzle was set to 742 L/h. The surface-modified powders are named LTO@X% PTCLi 4 or LTO@X% GO, depending on the nature and quantity (X wt. %) of organic molecules used in the preparation.

Carbon coating on LTO particles
After applying organic coatings on LTO material, a heat-treatment was carried out to generate a carbon coating on the anode material (see Scheme 1). The powder was heat-treated in a tubular furnace for 4 h at 800 °C (+10 °C/min from ambient temperature to 800 °C) under an argon atmosphere. The carbon-coated powders are named LTO@X% PTCLi 4 burned or LTO@X% GO burned, depending on the nature and quantity (X wt. %) of organic molecules used in the preparation.

Preparation of reduced graphene oxide
Reduced graphene oxide (r-GO) was prepared by heat-treatment of GO powder in a tubular furnace under a nitrogen atmosphere at 800 °C for 4 h. A highly volatile dark powder was obtained.

Characterizations
TGA curves of the synthesized organic molecules and coated LTO powders were recorded using a TGA 550 model (TA instruments) with a heating rate of 10 °C·min −1 and an air flow rate of 90 mL·min −1 from 30 to 700 °C. FTIR measurements were performed on a Bruker Vertex 70 spectrometer equipped with a smart ATR accessory.
Micro-Raman surface analyses were performed with a HORIBA LabSpec 5 apparatus with a 532 nm laser excitation wavelength. 9 XRD analyses were performed using a SmartLab X-ray diffractometer (Rigaku) with Co Kα 1 radiation (λ 1 = 1.78892 Å). Data were collected between 10° and 80°, with a step size of 0.02° and a scan speed of 3.04°/min, using a D/tex Ultra 250 detector.
The chemical compositions (5 nm deep) of the different LTO and organic powders as well as the electrode surface before and after long-cycling coin-cell experiments, were investigated by XPS, using a PHI 5600-ci spectrometer (Physical Electronics). The main XPS chamber was maintained at a base pressure of < 8.10 −9 Torr. A standard aluminum X-ray source (Al Kα = 1486.6 eV) was used to record the survey spectra (1400-0 eV, 10 min) and magnesium was used to obtain highresolution spectra, both without charge neutralization. The detection angle was set at 45º with respect to the normal of the surface and the analyzed area was 0.5 mm 2 . Curve fitting of the highresolution spectra of C 1s, O 1s, F 1s, Li 1s, Ti 2p, and P 2p was performed by means of a least- Tescan Mira SEM was used to evaluate the structural morphology of bare and PTCLi 4 -covered LTO powders. EDX (Oxford X80) analyses were performed to quantify the major carbon concentration on the samples, as well as to produce colored atomic distributions of Ti, O, and C on sample secondary particles. 10 TEM images of carbon-coated LTO materials were obtained using an HF3300 microscope (Hitachi High-Technologies Corporation) operating at 300 kV. The chemical composition was assessed with an electron energy loss spectroscopy (EELS) GIF Quantum system from Gatan, Inc. Samples were dispersed in ethanol and drop-casted on QUANTIFOIL ® holey carbon films prior to the analysis.
Adsorption isotherms were measured using a QuadraSorb Station 3 instrument (version 5.04, Quantachrome Instruments). Pristine and PTCLi 4 -coated LTO powders were characterized using nitrogen as an adsorbent at 77.3 K. The volume of gas adsorbed was recorded for relative pressures (P/P 0 ) ranging from 4.10 −2 to 1. The N 2 adsorption data were used to calculate the Brunauer-Emmett-Teller (BET) specific surface area (S BET ) as well as the total pore volume. The pore size distribution was calculated by simulating the isotherms using density functional theory (DFT) and Monte-Carlo calculations. with a small additional amount of NMP solvent and ten stainless steel beads were added to the mixture and mixed for 10 min. NMP was added in small quantities until a homogenous and viscous solution was obtained after mixing for 10 min. The ink was spread on an aluminum current collector (15 μm), which was then placed in an oven at 75 °C for at least 12 h. A cathode disk 11 (diameter ~19 mm) was punched for each electrode and dried under vacuum at 120 °C for 24 h before testing in a coin cell.

Cell assembly
The coin cells were assembled with an LTO cathode, Li metal counter and reference electrodes, a electrolyte in an argon-filled glove-box (O 2 < 10 ppm).

Galvanostatic cycling
LTO half-cells were cycled between 1.2 and 2.5 V vs. Li/Li + for five cycles at different cycling rates ranging from C/10 to 5C, where 1 C corresponds to a constant current of 165 mA·g −1 . The experiments were performed at ambient temperature (i.e., 25 °C), 0, and −20 °C.
Long-term cycling for 100 cycles was performed at a C/2 rate between 1.2 and 2.5 V vs. Li/Li + with the unmodified and PTCLi 4 -modified LTO electrodes. Every 20 cycles, the cycling rate is increased to C/10 for one cycle.

Electrochemical impedance spectroscopy
Electrochemical impedance spectroscopy measurements were performed at an open-circuit voltage (OCV) after fully charging the cell to 2.5 V vs. Li/Li + at a C/2 rate for the unmodified and PTCLi 4 -modified LTO electrodes. Measurements were performed with an amplitude of 10 mV and a frequency range of 1 MHz to 10 mHz after the 1 st and the 100 th cycles at C/2. Experimental data were fitted using the Z-fit tool within the EC-Lab software package (V11.12). Additional measurements were recorded at 1.55 V vs. Li/Li + after one charge/discharge cycle at a C/10 rate.

PTCLi 4
PTCLi 4 and PTCDA powders were first analyzed using FTIR spectroscopy. The corresponding FTIR spectra are shown in Figure 1a. The red PTCDA powder presented an intense band at approximately 1750 cm −1 that corresponds to the asymmetric C=O carboxylate stretching mode.
After reaction with strong LiOH base, the band shifted to ≈1595 cm −1 as usually observed for PTCLi 4 molecules [34][35][36]. Other characteristic IR bands such as the aromatic C=C stretching of the perylene structure at 1551 cm −1 , aromatic C-H stretching around 3050-3100 cm −1 , and C=O stretching at 1435 cm −1 are also observed, in good agreement with a recent study on PTCLi 4 anode material [37]. The well-ordered structures of PTCDA and PTCLi 4 were confirmed by XRD analyses. The corresponding XRD patterns, presented in Figure 1b, are similar to those reported for these two organic molecules and confirm the successful lithiation of the PTCDA precursor [37,38]. A SEM image of the synthesized molecule, shown in Figure

Graphene oxide
GO powder is first analyzed by FTIR and its spectrum is shown in Figure 2a. The highly hydrophilic material was effectively characterized by a broad peak centered at approximately 3500 cm −1 that is associated with -OH stretching vibrations [39,40]. In addition, several bands , and C-OH (~1375 cm −1 ) groups were observed. Carboxylic acid and carbonyl moieties (C=O stretching vibrations) were demonstrated by a well-separated single peak at 1725 cm −1 . Finally, a peak at 1600 cm −1 related to unoxidized sp 2 C=C bonds was also present in the IR spectrum. The XRD pattern of the graphite powder used for the synthesis was compared with that of the GO obtained after chemical oxidation ( Figure 2b).
A very intense peak (002) around 2Ɵ = 31° was obtained for the graphite powder, in addition to several smaller peaks corresponding to different crystal planes that are generally observed for this carbon [41]. The XRD pattern of the GO powder was characterized by a strong peak (001) around 2Ɵ = 12° with a smaller peak at 2Ɵ = 49°, which are both usually obtained for GO powder synthesized by Hummers' method [42]. After thermal reduction of GO under flowing N 2 , a dark r-GO powder was obtained and its corresponding XRD pattern showed a large and flat peak at around 2Ɵ = 29°, which proves the formation of graphene nanosheets [43]. TGA data, shown in oxygen functional groups on the GO sheets [44]. The decomposition of more stable oxygen functionalities was associated with the gradual loss of mass observed between 240 and 525 °C.
Above this temperature, the combustion of the carbon skeleton occurred. The r-GO powder was completely burned between 500 and 600 °C, confirming that graphene sheets were successfully synthetized [45].

PTCLi 4 coating on LTO surface
The LTO material was first coated with Li-rich PTCLi 4 powder by spray-drying, as described in the experimental section. PTCLi 4 is known as a low-cost organic anode material, synthesized from commercially available PTCDA colorant, and possesses a working potential (~1.2 V vs. Li/Li + ) [37] close to that of LTO (~1.55 V vs. Li/Li + ). In addition, PTCDA is a semiconducting material [46,47]. For these reasons, the coating of LTO particles with PTCLi 4 was of interest to yield a material with good surface conduction of Li-ions and electrons. In addition, the thin PTCLi 4 layer can provide overcharge protection in a full-cell configuration, as its redox potential is lower than  Figure SI1 in Supporting Information), and the inlet temperature was set to 220 °C. Under these conditions, the drop of water was instantly evaporated and the highly water-soluble PTCLi 4 molecule gently surrounded the LTO particle, which was in suspension in the water. The thermograms in Figure 3b show that LTO remained stable up to 700 °C under air flow, whereas LTO@PTCLi 4 powders presented a mass loss between 375 and 500 °C, which was attributed to the degradation of the PTCLi 4 coating.
According to the TGA results and considering that a weight loss of 66% for the redox molecule alone was recorded under heating to 550 °C (see thermogram in Figure SI1), it can be calculated than LTO powders were obtained with approximately 2, 3, and 4.9 wt. % of PTCLi 4 , as expected.
The method and parameters are well adapted, as the amounts of PTCLi 4 in the composite powders corresponded to the quantities of PTCLi 4 measured to prepare the solutions. Figure 3c shows photographs of pristine and PTCLi 4 -coated LTO powders, as well as the organic molecule.
Although a small concentration of the organic molecule was present in the composite powders,  Figure SI2 in Supporting Information) revealed that carbon was present everywhere on the surface of the anode material and that the voids between the primary LTO particles are effectively rich in carbon (i.e., PTCLi 4 ). The PTCLi 4 -coated LTO powders were not observed by TEM due to the high sensibility of organic molecules to the electron beam irradiation [50]. Thus, micro-Raman and XPS analyses were 18 chosen to confirm the nanometric coverage of the LTO surface. Raman spectroscopy is a spectroscopic technique generally used to obtain information about vibrational modes of molecules or crystals. In particular, it is very sensitive to the electronic structure of carbon, its degree of hybridization, and the crystal disorder [51]. The micro-Raman technique enabled a chemical surface analysis to a depth of a few nanometers and was employed to demonstrate the coverage of LTO particles with PTCLi 4 and then with carbon from the subsequent heat-treatment. Figure 4a shows the spectrum of LTO material (-) that reveals three bands at 240, 430, and 680 cm −1 , which are assigned to F 1g , Eg, and A 1g vibrations of the LTO crystal, respectively [52]. PTCLi 4 (-) exhibited a strong photoluminescence effect [53,54], inducing a background with a mountain-like shape and small peaks at 1315, 1355, and 1570 cm −1 were hardly observed but often reported for perylene-based structures, such as the PTCDA precursor. More precisely, the two peaks at 1315 and 1355 cm −1 are attributed to the C-H bonds of the perylene structure, whereas that at 1570 cm −1 is associated with C-C bonds [55]. When LTO is coated with 2% (-), 3% (-), and 5% by mass (-) of PTCLi 4 , the general signal for PTCLi 4 was conserved, although the peak intensity decreased as the amount of LTO in the composite powder increased. More importantly, the absence of characteristic bands of LTO was consistent with the coverage of its surface by organic material.  XPS analyses are also performed on pristine and PTCLi 4 -coated LTO powders, and their C 1s and O 1s core level spectra are shown in Figure 5. For the bare LTO, carbon was detected on its surface (single peak at 285 eV, Figure 5a), although this was not expected and probably due to adventitious carbon typically found on the surface of most air-exposed samples [55]. The O 1s core level spectrum, presented in Figure 5d, shows an intense peak at approximately 530 eV, which is attributed to Ti-O bonds of the spinel structure. An additional small peak at 532 eV comes from surface pollution. After surface modification with PTCLi 4 , the shape of the C 1s and O 1s spectra was strongly affected, which confirmed the efficient coverage of the LTO surface. The peak at 285 eV for PTCLi 4 -coated LTO materials increased, which is consistent with the presence of aromatic C=C and C-C bonds (see Figure 5b,c). The appearance of a small peak located at 286- and c,f) LTO@5% PTCLi 4 powders.

Graphene oxide coating on LTO surface
The LTO was also coated with previously synthesized GO powder using the same technique and conditions as for PTCLi 4 . This material is widely studied, especially to generate carbon coatings on electrode materials such as LTO [56,57]. Because PTCLi 4 -coated LTO powders can be heattreated in a second step to generate a carbon coating, we chose to compare the effects of the organic molecules (i.e., GO and PTCLi 4 ) on the deposition of carbon on the LTO surface and their implications for the electrochemical performance. Figure 6 shows the thermal decomposition of different composites made with LTO and various amounts of GO ranging from 2 to 4 wt.%. Two clearly discernable mass losses around 150 and 22 400 °C correspond to the degradation of GO carbon, as explained above (see Figure 2c). However, these decomposition temperatures are slightly lower than those recorded for pure GO powder, probably owing to a catalyst effect of the transition metal at the LTO surface or because of a structural change in the GO sheets, which were less agglomerated after spray-drying. At 650 °C, the GO was totally degraded and mass losses of approximately 1.6, 2.3, and 2.7% were calculated.
These values are lower than the expected losses (2, 3, and 4 wt.%, respectively), mainly owing to experimental parameters such as the outlet temperature set to 220 °C. At this temperature, according to the thermogram of GO powder, shown in Figure 2c, its degradation starts and it is well demonstrated in Figure 6 that the first mass loss is quite low in comparison to the second at a higher temperature. These results assume that the release of labile oxygen functional groups already began during spray-drying.  symmetry. It is associated with disordered carbons or defective graphitic structures and its corresponding intensity increases with the number of defects in the carbon structure obtained [59].

Thermal treatments of GO-and PTCLi 4 -coated LTO powders
The G band (G for graphitic) observed at ~1598 cm −1 is ascribed to the E 2g phonon of sp 2 -bonded carbon atoms, which is a characteristic feature of graphitic layers [44]. The intensity ratio of the D to G bands (I D /I G ) was approximately equal to 0.76 for the three LTO@PTCLi 4 burned powders.
This ratio is particularly low and demonstrates that the carbon layer is mostly composed of perylene stacking to form a structure with a relatively high degree of graphitic carbon on the surface of LTO. In fact, I D /I G ratios of approximately 1, 0.84, and 0.09 were reported for GO, r-GO, and graphite [60]. The shape and position of the 2D band at approximately 2660 cm −1 indicate the presence of a graphitic structure composed of few layers of carbon atoms [61][62][63] that is consistent with a thin carbon layer, which will be confirmed by TEM observations (results shown below).
In addition, the XPS profiles of burned GO-and PTCLi 4 -coated LTO powders shown in Figure   SI3 confirm the deposition of a thin carbon layer on the surface of LTO. The C 1s and O 1s spectra 24 are similar for the carbon-coated samples made with either PTCLi 4 or GO. Moreover, the small peak observed at 289 eV for PTCLi 4 -coated LTO powders (see Figure 5b,c) and attributed to -COOLi bonds disappeared after burning, which is consistent with the degradation of the molecule and its conversion in carbon.  The TEM images presented in Figure 8 show that the pristine LTO (left side) has a well-defined crystal structure with a succession of (111) planes, confirming that the surface has a typical spinel structure [63,64]. For the LTO@3% PTCLi 4 burned powder, an additional thin layer was observed on the LTO particle surface and attributed to the carbon layer generated by the decomposition of PTCLi4 molecules. As mentioned above with the Micro-Raman analyses of burned LTO@PTCLi 4 samples, the shape of the 2D band suggests that the carbon coating was composed of a few layers of carbon atoms [65]. If we take into account the reported value for the thickness of a PTCDA monolayer, which is approximately 0.32 nm [66,38], then the carbon coating should be composed of at least six to seven layers of perylene-like carbon. A rough calculation, considering the largest 26 lattice parameter of the PTCDA monoclinic crystal (P2 1 /c space group, approximately 2 nm) [38], quantity of carbon estimated by TGA (0.89 wt.%), and BET surface area for the LTO sample (see Figure SI4 and Table SI1 [68]. Such a carbon-coated material will provide good electronic conductivity owing to the improved electron-transfer on the LTO surface, but the absence of large carbon sheets as observed for the LTO@GO burned sample will limit the electron movement between the LTO particles or agglomerates. Thus, slightly lower

Electrochemical performance of LTO electrodes
Spinel-LTO (Li-poor phase) anode material can accommodate up to three lithium ions per formula unit at a potential of 1.55 V vs. Li/Li + [69,70]. During the discharge process, associated with the reduction of three Ti 4+ atoms, the spinel structure is converted to a Li-rich rock-salt phase, which provides a theoretical capacity of 175 mAh·g −1 [71,72]. LTO has many advantages, such as an excellent Li-ion insertion/extraction reversibility, negligible volume expansion during the charge/discharge process, and a flat potential plateau [73]. However, owing to its poor electronic conductivity, in the range of 10 −8 -10 −13 S·cm −1 [74,75], LTO presents low specific capacity and poor capability for high-rate performance. Generally, a practical specific capacity of approximately 150-160 mAh·g −1 is obtained, as will be shown below for our batteries. Finally, the shape and the particle sizes of an electrode material strongly influence the performance at low and high C-rates [76]. For instance, for high power density applications, nanometric LTO particles is preferred in order to facilitate Li + ion transport by dramatically reducing the diffusion distance [77]. In contrast, for energy demanding applications, micrometric particles size is needed with high tap density. Increasing the tap density to gain higher volumetric energy density is a crucial factor for commercialization [78]. We selected a LTO material designed for energy demanding applications and composed of primary particles of nanometer size forming large secondary spherical aggregates of a few micrometers to 20-30 µm in diameter (see SEM images in Figure   SI2). Stable capacities at low C-rates were obtained with this material, as it will be shown below, but it hardly cycled at high C-rates, which is consistent with the cycling of micrometric LTO spheres [78].

Influence of coating nature
All the materials presented above were electrochemically characterized in a coin-cell. Figure 10 presents charge/discharge curves for pristine LTO and LTO@2%PTCLi 4 electrodes cycled at 25 °C, at rates ranging from C/10 to 5C. Both cells showed the predominance of a flat discharge plateau around 1.55 V vs. Li/Li + at a low cycling rate, corresponding to the Ti 3+ /Ti 4+ redox couple.
However, the PTCLi 4 -coated LTO electrode delivered a higher initial discharge capacity of approximately 162 mAh·g −1 , whereas the pristine LTO yielded a mere 154 mAh·g −1 owing to the low electronic conductivity of the uncoated micrometric LTO. In fact, the specific capacity remained quite stable up to a rate of 1C for the modified LTO electrode and the polarization only severely increased at a rate ≥ 2C. For the bare LTO electrode, a rapid loss of specific capacity was observed with increasing C-rate; for example, the same capacity (approximately 140 mAh·g −1 ) was recovered for the pristine and PTCLi 4 -coated LTO electrodes at rates of C/5 and 2C, respectively. In addition, the bare LTO electrode was not able to cycle at a rate of 5C, owing to the narrow potential window selected (1.2-2.5 V) and the excessively high polarization of the electrode. At this rate, the LTO@2%PTCLi 4 electrode delivered an average 85 mAh·g −1 . This improvement is due to the thin layer of PTCLi 4 on the surface of LTO, which is characterized by a strong π-π conjugation between the layer-stacked molecules favoring fast electron transfer [79].
In fact, an electronic conductivity for nanometer-thick PTCDA stacking of approximately 3 × 10 −2 S·cm −1 was reported, which is much higher than that for LTO (i.e., 10 −8 -10 −13 S·cm −1 ) [74,75]. Nevertheless, when the amount of PTCLi 4 was increased in the composite electrode material, the electrochemical performance slightly decreased, especially at high C-rates, as often reported for other types of organic surface treatments with excessive amounts of material [80,81].
The rate capabilities of the pristine and PTCLi 4 -coated LTO electrodes are shown in Figure 11a, which clearly demonstrate the strong beneficial effect of the organic coating on the electrochemical performance. These results indicate that patchy coverage of the LTO surface is obtained for samples with low carbon content, which should be higher than 1.5% by mass to achieve performance near that obtained with the burned-GO coating.
Interestingly, by comparing Figure  and dendrite formation [83]. It is worth noting that at high C-rates (from 2C to 5C), the specific capacities for the LTO@4%PTCLi 4 electrode were inferior to those for the LTO@2%PTCLi 4 anode (see Figure 11a). This behavior is attributed to the higher quantity of PTCLi 4 molecules on the surface of LTO@4%PTCLi 4 material, which are beneficial at low C-rates for improving the electronical conductivity but detrimental at high C-rates [81,84].

Influence of cycling temperature
The electrolyte plays the crucial role in controlling the mass transport in the cell, including Li + conduction through the bulk electrolyte as well as Li + migration through the electrode/electrolyte interface [88]. Many efforts have been devoted to designing low-temperature electrolytes with the use of low-freezing-point solvents [89][90][91] or mixtures of lithium salts [92][93][94][95]. The electrolyte 35 formulation strongly influences the composition of the SEI, the solvation of lithium ions, and thus the energy barrier associated with the desolvation of lithium at the electrolyte/electrode interface [88,96]. According to recent studies, Li + migration across the SEI layer [97] and the Li + desolvation process [98,99] are limiting steps for the fast movement of lithium ions in the cell.
These are the reasons why we believe the design of an artificial SEI layer (e.g., PTCLi 4 coverage of LTO surface) could strongly improve the Li + ion transfer between the electrolyte and the surface of the active material, especially at low temperatures.
Hence, additional electrochemical tests were performed at 0 and −20 °C to confirm the potential of the PTCLi 4 coating to increase the Li + ion transfer through the electrode/electrolyte interface. Figure 13a presents the rate capabilities for pristine and PTCLi 4 -coated LTO electrodes cycled at 0 °C. First, at this temperature, a lower initial capacity at C/10 was obtained for all the batteries; for instance, 165 and 160 mAh·g −1 were delivered for the LTO@2% PTCLi 4 electrode at 25 and 0 °C, respectively, whereas pristine LTO yielded 155 and 140 mAh·g −1 at the same temperatures.
Second, a progressive loss of capacity was observed for PTCLi 4 -coated LTO electrodes up to a cycling rate of 1C, at which 130 mAh·g −1 was still delivered with the LTO@2% PTCLi 4 electrode.
In contrast, for the same rate, the bare LTO anode was not able to cycle and showed a poor rate capability. Finally, at a rate of 2C, an average value of 70 mAh·g -1 was calculated for the LTO@2% PTCLi 4 electrode. As observed at ambient temperature, the carbon-coated LTO electrodes showed intermediate results at 0 °C (see Figure 13b), but the performance of the LTO@4% GO burned Experiments performed at −20 °C showed the same trend, except that the initial discharge capacities at C/10 and the rate performances were more adversely affected by the low temperature.
However, it was observed that carbon-coated electrodes with excessive content of graphitic carbon (e.g., LTO@3% and 5% PTCLi 4 burned electrodes) delivered very low discharge capacities at C/10 (~100-110 mAh·g −1 ), which are even lower than that obtained for the pristine LTO electrode (125 mAh·g −1 ). Consequently, these batteries showed the poorest electrochemical performance at −20 °C, owing to their high charge-transfer resistance caused by the hindrance by graphene layers of the fast Li + ion transfer between the electrolyte and the host material. In contrast, the LTO@2% PTCLi 4 electrode delivered 90 mAh·g −1 at a rate of C/2 and −20 °C, which is similar to that obtained at 25 °C with the pristine LTO electrode. 38 of PVDF binder at 286.5 and 291.7 eV and conductive carbon at 285 eV [100]. The O 1s spectrum ( Figure 14b) is dominated by the metal oxide peak at ~530 eV and a broad and low-intensity peak centered at 532 eV associated with the presence of oxygen-containing species on the carbon surface. Finally, the F 1s core level spectra, shown in Figure 14c, contain a peak at 688 eV characteristic of PVDF binder [101] and a small shoulder at 685 eV that could be due to adsorbed or entrapped fluorine [102,103]. The atomic concentrations of C, O, Ti, and F determined by XPS are given in Table 1. It is worth noting that a higher concentration of carbon was observed for the LTO@4% PTCLi 4 electrode, owing to the coverage of the LTO surface by the organic molecules, which is also consistent with the lower percentage of Ti (4.5%) detected in comparison to the LTO electrode (5.8%).
After the long cycling experiments, the C 1s region for the surface of LTO@4% PTCLi 4 electrode ( Figure 14d) displayed additional contributions that were assigned to species containing C-O-C bonds (286.5 eV), O-C-O, and/or C=O bonds (288 eV) [104] as well as a possible contribution from carbonates (Li 2 CO 3 or RCO 3 Li) at 290 eV [101]. All these species, coming from electrolyte degradation, are generally observed at the surface of electrodes after cycling. The atomic concentration of carbon at the surface of the pristine LTO electrode was slightly lower and its environment appeared different from that of the LTO@4% PTCLi 4 electrode. For instance, the peak at 292 eV (CF 2 from binder) is no longer visible, and it is probably hidden by the presence of a large amount of LiF on the surface the film. The O 1s core level spectra for both the electrodes are relatively similar (Figure 14e). The metal oxide peak at 530 eV is still visible, but with a low intensity that is consistent with the low concentration of Ti detected for the two samples (see Table   1) due to the presence of the passivation layer. In addition, at least two other contributions are observed around 532 and 534 eV, which are attributed to the formation of degradation products 39 containing Li 2 CO 3 /C=O and O-C=O/C-O-C bonds [105], respectively. In contrast, there is no evidence for the formation of Li x PF y O z generated by the degradation of the lithium salt and generally observed around 535 eV [106]. However, small amount of phosphorous was detected by XPS and could be attributed to this type of degradation products. Finally, the most important difference between the pristine and the PTCLi 4 -coated electrodes is the concentration of fluorine.
In fact, fluorine accounts for 51.1% of the atoms present on the surface of the pristine electrode, compared to 42.5% for the modified electrode (see Table 1). The F 1s core level spectra for the two electrodes after cycling are shown in Figure 14f. It is clear that for the LTO electrode, a higher quantity of LiF (~685.5 eV) is formed because the binder peak at 688 eV has almost disappeared.
This result is in agreement with the increase in the internal resistance observed for the LTO electrode after long-term cycling (Figure 12c). In fact, LiF is known to exhibit poor electronic conductivity and its formation must be avoided [107]. In conclusion, although not expected, the PTCLi 4 coating on the LTO material could prevent or reduce the formation of resistive LiF at the surface of the electrode, thus leading to better cyclability and rate performance.

Proposed mechanism
Based on previous explanations for carbon-coated LTO [108] and in order to explain our results, we proposed a schematic (see Scheme 2) to depict Li-ion and electron transfers at the surface of the different prepared LTO powders. Carbon coating strongly increases the electronic conductivity at the surface of LTO anode material and although some particles are not totally covered [13] (e − transfer is slowed down, see red crosses on Scheme 2b,c), the electrochemical performance for carbon-coated LTO anodes is better than that obtained with the bare material. In particular, as represented in Scheme 2c, the LTO@GO burned sample possesses a thin layer of carbon on the LTO surface (see Figure 8) as well as large r-GO sheets (see Figure 9b) that surround the LTO electrochemical performance to that for pristine LTO at 25 °C but similar performance as that for r-GO-coated LTO, because the electronic conductivity is a limiting factor at this temperature. In contrast, when the temperature is decreased, the ionic conductivity and solvation/desolvation processes become the limiting factors [89,92,93]. Thus, the electrochemical performance for PTCLi 4 -coated LTO was increasingly better than that of the other carbon-coated materials as the temperature decreased ( Figure 13). In addition, the high graphitization degree of carbon layers, demonstrated by Micro-Raman investigations for burned PTCLi 4 (I D /I G = 0.76) and widely reported for r-GO material [60], limits the Li + ion diffusion and more precisely in the perpendicular direction to graphene layers [108,109] (represented with red crosses in Scheme 2b,c). According to the electrochemical results presented in Figure 13d, it is concluded that an excessive amount of graphitic carbon (e.g., LTO@5% PTCLi 4 burned electrode) is detrimental to electrochemical performance at low temperatures and ionically conductive coatings are preferable.

Conclusions and perspectives
We proposed a facile way to make an electronically and Li-ionically conductive coating on the