Enhancing the energy density of safer Li-ion batteries by combining high-voltage lithium cobalt fluorophosphate cathodes and nanostructured titania anodes

Recently, Li-ion batteries have been heavily scrutinized because of the apparent incompatibility between safety and high energy density. This work report a high voltage full battery made with TiO2/Li3PO4/Li2CoPO4F. The Li2CoPO4F cathode and TiO2 anode materials are synthesized by a sol–gel and anodization methods, respectively. X-ray diffraction (XRD) analysis confirmed that Li2CoPO4F is well-crystallized in orthorhombic crystal structure with Pnma space group. The Li3PO4-coated anode was successfully deposited as shown by the (011) lattice fringes of anatase TiO2 and (200) of γ-Li3PO4, as detected by HRTEM. The charge profile of Li2CoPO4F versus lithium shows a plateau at 5.0 V, revealing its importance as potentially high-voltage cathode and could perfectly fit with the plateau of anatase anode (1.8–1.9 V). The full cell made with TiO2/Li3PO4/Li2CoPO4F delivered an initial reversible capacity of 150 mA h g−1 at C rate with good cyclic performance at an average potential of 3.1–3.2 V. Thus, the full cell provides an energy density of 472 W h kg−1. This full battery behaves better than TiO2/Li2CoPO4F. The introduction of Li3PO4 as buffer layer is expected to help the cyclability of the electrodes as it allows a rapid Li-ion transport.

as we know, there are no reports in the literature dealing with TiO 2 /Li 2 CoPO 4 F that can reach theoretical energy densities above 450 W h kg −1 , performance close to the demands of modern applications.
In order to improve the cyclability of high voltage LIBs, the effects of a surface treatment of lithium phosphate on a full cell made with Li 2 CoPO 4 F as cathode and TiO 2 as anode were studied. This report shows how the electrochemical performance of this material compared very favourably with Li 3 PO 4 -free electrodes. The introduction of an inactive matrix such as Li 3 PO 4 for use as a buffer layer is expected to help the cyclability of the electrodes by allowing a rapid transportation of Li ions 13,14 . Figure 1A shows a schematic view of the Li 2 CoPO 4 F structure. It is formed by chains of CoO 4 F 2 octahedra sharing their edges and interconnected with PO 4 tetrahedral oxo-anions by corner sharing. The solid possesses an orthorhombic unit cell with Pnma space group, where there are 3 types of Li; Li1 in 8d sites and Li2 and Li3 in two sets of 4c sites. The Co is in 4a and 4b sites, the P in two sets of 4c sites, the F in two sets of 4c sites, and the O in four sets of 4c and two sets of 8d sites 6 . The cross linked structure forms an opened 3D framework, permitting Li ions to be inserted and extracted from multiple directions [15][16][17] . The XRD pattern of the synthesized Li 2 CoPO 4 F sample shown in Fig. 1B shows diffraction peaks indexed in agreement with the literature values, with a = 10.452 Å, b = 6.3911 Å, and c = 10.874 Å 6,17 . A LiCoPO 4 impurity phase was detected (marked with symbol * in Fig. 1B), which could have been due to the relatively low heat treatment temperature 11,15 . The XPS signal of the Co2p, split into the 2p 3/2 and 2p 1/2 multiplet separated by 15.7 eV, is formed by a double peak at 781.7 eV and 786.2 eV (Fig. 1C), assigned to the Co 2+ in Li 2 CoPO 4 F in very good agreement with the observation reported in ref. 15.

Results and Discussion
The Li 3 PO 4 -coated TiO 2 anode was prepared at room temperature, with some material also annealed at 500 °C in air for 2 h. The XRD patterns of the materials exhibit different features (Fig. 2). While for the as-prepared specimen, only Ti reflections (JCPDS file 05-0682 and space group P6 3 /mmc) can be observed, the annealed sample shows very intense peaks of anatase (JCPDS 21-1272 and space group I4 1 /amd). In both cases the presence of Li 3 PO 4 could not be detected by XRD, due to the low crystallinity and small amounts of solid controlled by the conditions of the electrolytic deposition. Since the Li 3 PO 4 is on the surface of TiO 2 nanotubes further experiments allowed us to detected the (110), (101), (210) and (002) reflections of β -Li 3 PO 4 (JCPDS 25-1030) phase for large time (20 min) and high current density (75 mA cm −2 ) during electrolytic deposition 13 . Previous experiments conducted to synthesize thick layers of lithium phosphate covering the complete surface of titania nanotubes 13 . So, this over 20 μ m thick layer would passivate the active electrode and cannot be beneficial for such cycling purposes as electrodes in batteries. For this reason, we scaled down the fabrication of Li 3 PO 4 layer to the nanometric size, observing that the best ratio of Li 3 PO 4 to TiO 2 is 9.03·10 −3 g Li 3 PO 4 /g TiO 2 . By using optimal current densities and deposition times of ca. -3.75 mA cm −2 and 1 min a finely dispersed layer of Li 3 PO 4 can fill the titania nanotubes (see Supplementary Fig. S1 online). Under these conditions no diffraction peaks either of β -Li 3 PO 4 or γ -Li 3 PO 4 were detected in the X-ray diffraction patterns which were recorded from 10-80° (°2θ ) with 0.02° of step size each 2 seconds as discussed above. Having a detailed inspection with HRTEM and SAED, the formation of β -Li 3 PO 4 (at room temperature) or γ -Li 3 PO 4 (when annealing) on titania nanotubes was unveiled as discussed below.
The lithium phosphate seems to play an important role enhancing the electrochemical response, both in Li half cells and in full Li-ion batteries, and deserves to be studied in more detail 13,[17][18][19] . The β -Li 3 PO 4 has a basic wurtzite structure where one position of the tetrahedral sites, T+ or T-, is fully occupied, along with cation ordering (Fig. 2C). It has twice the value of the unit cell along the axis when the phase transition from β -to γ -Li 3 PO 4 ( Fig. 2D) occurs. The γ -phase also consists of hexagonal close-packed oxide layers, but these are more distorted in comparison with the β -structure. Moreover, the cations are distributed over both sets of T+ and T-sites, leading LiO 4 tetrahedra to share some of their edges, while only corner-sharing is present in the β -structure [20][21][22] .
In order to examine the formation of phosphate phases on titania nanotubes and discover whether or not an additional phase is formed at the interface, HRTEM and SAED measurements were performed. The HRTEM image of the "AD" sample (which was not calcined) is shown in Fig. 3A, which shows areas having visible lattice fringes measured and labelled according to their particular crystal structures. It can be seen that β -Li 3 PO 4 was successfully formed, with the image containing a small region of ordered crystalline structure with a (110) interplanar spacing of 0.399 nm 23 . Here, amorphous TiO 2 was not detected, but fringes corresponding to the (222) reflection of Li 4 Ti 5 O 12 with an interplanar spacing of 0.245 nm were found 24 . This image was taken at the tip of the nanotube and we can observe that lithium titanate appeared at both sides of the region of lithium phosphate, which could explain the lack of TiO 2 detection. The SAED data in Fig. 3B matches these d-spacings and confirms the presence of these materials. Figure 3C shows HRTEM imagery of the ADC sample (which was calcined), with regions of visible lattice fringes measured and labeled to identify their respective crystal structures. Here, a small region showing an ordered crystalline structure with a (011) interplanar spacing of 0.352 nm was detected, corresponding to anatase TiO 2 25 . However, the observed form of lithium phosphate was γ -Li 3 PO 4 as deduced from the (200) reflection (d 200 = 0.247 nm), as expected after thermal annealing at 500 °C 26 . In addition, fringes corresponding to lithium titanate are also visible (d 113 = 0.255 nm) 24 . The anatase TiO 2 , γ -Li 3 PO 4 and Li 4 Ti 5 O 12 regions are labeled "1", "2" and "3" respectively to avoid confusion. The SAED data in Fig. 3D shows fine diffraction points corresponding to these three phases. The presence of small amounts of the Li 4 Ti 5 O 12 phase at the interface between Li 3 PO 4 and TiO 2 gives an additional understanding of the structure of these composites.
The cathode half-cell reaction can be written as: The theoretical capacity of Li 2 CoPO 4 F is 287 mA h g −1 for x = 2. However, recent studies have indicated that Li 2 CoPO 4 F can reach a maximum reversible capacity of 150 mA h g −1 , with an outstanding high-voltage operation of ~5 V vs. Li + /Li 11,15,[27][28][29][30] . Because of the high inefficiency from the first to the second cycle observable in Li 2 CoPO 4 F, these electrodes were subjected to activation cycles before being used in the complete lithium-ion cell (see Supplementary Fig. S2 online). The first-cycle irreversible capacity due to electrolyte decomposition was then avoided in the full cells 11,15 . Anatase is well known in the literature to exhibit a high reversibility in the first cycle and to operate at a safe average potential of 1.8-1.9 V vs. Li + /Li. The theoretical capacity delivered by anatase is around 167 mA h g −1 according to the following reaction: -nt TiO 05Li 05e nt Li TiO 2 2 0 5 2 Taking into account their individual voltages, the combination of nt-TiO 2 with Li 2 CoPO 4 F could give rise to a battery operating in the 3.1-3.2 V range. While considering the expected capacity of each electrode, the main overall reaction that may take place in the full cells can be summarized as follows: Capacity balance was carried out by assuming 140 mA h g −1 reversible capacity of the cathode after activation and 160 mA h g −1 of the anode (see Fig. 4), the resulting cathode mass to anode mass was: m+/m-= 1.14. Figure 4A,B compare the reversible voltage profile versus Li of the TiO 2 with and without Li 3 PO 4 (bottom) and of the Li 2 CoPO 4 F cathode (middle). The anode operates reversibly with continuous, plateaued charge-discharge curves with a reversible capacity of 150 mA h g −1 at an average voltage value of about 1.8-1.9 V, while the Li 2 CoPO 4 F cathode cycles with a reversible capacity of 148 mA h g −1 at a voltage value of 5 V vs. Li with a flat plateau, typical of the two phase reaction of lithium-cobalt fluorophosphates 19 .
The upper plots of Fig. 4A,B show the trend of the full-cell voltage profile, demonstrating very stable behaviour. The cell operates with an average voltage of around 3.1-3.2 V, while the voltage profile is the combination of the flat voltage of the Li 2 CoPO 4 F cathode (Fig. 4 middle plots) and the flat voltage of the TiO 2 (Fig. 4B bottom plot) or TiO 2 /γ -Li 3 PO 4 anodes (Fig. 4A bottom plot). The reversible capacity of the full battery measured at a state of discharge is about 150 mA h g −1 , reaching about 99% of the maximum reversible capacity. The achieved energy density is 472 W h kg −1 , an enhanced value as compared to the majority of published batteries [31][32][33][34][35][36] .
Li 4 Ti 5 O 12 is a well-known material for LIBs and typically shows a stable plateau at 1.54 V (vs. Li + /Li) 7,37-39 . However, such a plateau is not visible in the charge/discharge curves (Fig. 4A,B bottom plots). Instead, the typical plateau of anatase TiO 2 can be seen. The lithium titanate phase is formed in a minute fraction at the interphase between TiO 2 and Li 3 PO 4 , as detected by SAED and HRTEM measurements (Fig. 3). However, this phase was not detected by XRD (Fig. 2), due to its particularly low proportion. Then, its contribution to battery functionality is expected to be negligible.
The stability of chosen electrode materials is another key factor for battery cycling. Figure 5 compares the cycling stability of TiO 2 /Li 3 PO 4 /Li 2 CoPO 4 F and TiO 2 /Li 2 CoPO 4 F measured at 1 C, 2 C and 5 C rates. The battery that utilizes Li 3 PO 4 shows very good cycling behaviour as compared to that of the full cell without Li 3 PO 4 , operating at 1 C, 2 C and 5 C rates for more than 240 charge-discharge cycles with high coulombic efficiencies of 79, 62 and 73%, respectively. These differences were found significant from a statistical analysis of cycling experiments of five different cells for each composition that can be found as Supplementary Table S1 online. As expected, capacity decay from 150 mA h g −1 at 1 C to 120 mA h g −1 is recorded at 2 C, and to 90 mA h g −1 at 5 C. This battery, based on TiO 2 /Li 3 PO 4 /Li 2 CoPO 4 F, exhibits much better performance in terms of cyclability and coulombic efficiency than TiO 2 /Li 2 CoPO 4 F and that previously reported and based on Li 4 Ti 5 O 12 as an anode material 11 . The excellent performance of this battery observed in terms of specific capacity, cycling life and rate capability is to the best of our knowledge only seldom reported, and confirms the great potential of TiO 2 /Li 3 PO 4 as an innovative electrode material that can aid the progress of lithium-based energy storage systems.

Conclusions
The high working voltage and excellent rate capability observed of the TiO 2 /Li 3 PO 4 /Li 2 CoPO 4 F full cell makes it a promising high energy density LIB with acceptable rate performance (472 W h kg −1 at a 1 C rate, and 284 W h kg −1 at a 5 C rate), preventing the emergence of safety issues caused by the highly reactive lithiated graphite present in most LIB systems. The existence of Li 3 PO 4 and the minute fraction of Li 4 Ti 5 O 12 present between the TiO 2 / Li 3 PO 4 interfaces can explain the good cyclability of the full cell as this inactive matrix allows rapid transportation of the lithium ions.

Methods
The Li 2 CoPO 4 F/C nanocomposite was synthesized by the sol-gel (SG) method as previously reported 11,15 . The self-organized titania nanotube (nt-TiO 2 ) layer was fabricated by an anodization process using Ti foils at 60 V for 2 h, with a freshly prepared mixture of EG/water (92:8 vol.) containing 0.3 wt. % NH 4 F as an electrolyte solution. The deposition of electrolytic Li 3 PO 4 was performed on nt-TiO 2 as either amorphous material (labeled as AD) or, after calcination, as a anatase material (labeled as ADC), using a current density of -3.75 mA cm −2 for 1 min. Electrolytic Li 3 PO 4 films were deposited by an electrochemical procedure consisting of proton reduction with a subsequent local increase of pH in the vicinity of the substrate surface, hydrogen phosphate dissociation and Li 3 PO 4 deposition on the surface of the cathode 13 . Optional thermal annealing at 500 °C was performed. The thickness and active mass of the anode was 8 μ m and 0.935 mg cm −2 respectively 13 .
HRTEM and SAED images were collected with a Tecnai F-20 device operating at 200 kV. The X-ray diffraction (XRD) patterns were recorded with a Siemens D5000 instrument utilizing Cu Kα radiation. The X-ray photoelectron spectroscopy (XPS) measurements were performed with a SPECS Phobios 150MCD instrument using a Mg Kα source (1253.6 eV) and a chamber pressure of 4 × 10 −9 mbar.
Electrochemical characterization and cycling properties (discharge−charge) were performed using a three electrode configuration with a Biologic-VMP instrument. The full cells were assembled in a glovebox under an Ar atmosphere. A 9 mm diameter lithium disk was used as reference electrode, with Li 3 PO 4 −ntTiO 2 -based films and Li 2 CoPO 4 F used as counter and working electrodes. The electrolyte solution was 1 M LiPF 6 (EC:DEC) embedded in Whatman glass fiber disks. The full cell was cycled at 1C, 2 C and 5 C rates (C = 0.3 mA cm −2 ). The activation of the positive electrode (Li 2 CoPO 4 F) offers the possibility of achieving a remarkable reversible capacity for the full cell. In the present study, the activation step of the Li 2 CoPO 4 F consisted of two successive cycles of galvanostatic charging to 5.4 V, followed by discharging to 3.0 V, at a 100 mA g -1 current density and using metallic Li as a counter electrode. When designing the full battery, it is quite important to obtain an optimal balance of cathode and anode both in terms of weight and electrochemical properties. The calculation of the energy density of the battery only considered the specific capacity and the working potential (E cathode -E anode ) of the full battery, without further consideration of the mass of the active materials, electrolyte and packing materials.