Unravelling the Role of Electrochemically Active FePO4 Coating by Atomic Layer Deposition for Increased High‐Voltage Stability of LiNi0.5Mn1.5O4 Cathode Material

Ultrathin amorphous FePO4 coating derived by atomic layer deposition (ALD) is used to coat the 5 V LiNi0.5Mn1.5O4 cathode material powders, which dramatically increases the capacity retention of LiNi0.5Mn1.5O4. It is believed that the amorphous FePO4 layer could act as a lithium‐ions reservoir and electrochemically active buffer layer during the charge/discharge cycling, helping achieve high capacities in LiNi0.5Mn1.5O4, especially at high current densities.

LiNi 0.5 Mn 1.5 O 4 (LNMO) is a derivative of the commercialized spinel LiMn 2 O 4 with Ni 2+ and Mn 4+ occupying two octahedral sites of 4a and 16d, respectively aimed at suppressing the intrinsic defi ciencies such as the Jahn-Teller distortion of Mn 3+ with the same theoretical capacity as LiMn 2 O 4 (148 mAh g -1 ). [ 1 ] In addition, since the working mechanism of LiNi 0.5 Mn 1.5 O 4 is mainly the redox couple of Ni 2+ /Ni 4+ , the theoretical operating voltage reaches 4.7 V (vs Li + /Li) compared with 4.0 V for LiMn 2 O 4 (vs Li + /Li). [ 1,2 ] Such a high voltage inevitably involves the aggressive oxidation of the electrolyte and the dissolution of transitional metals, which cause the capacity fading. [ 3 ] In order to overcome these drawbacks, various strategies such as surface modifi cations using metal oxides and phosphates like Al 2 O 3 , [ 4 ] ZnO, [ 5 ] MgO, [ 6 ] ZrO 2 , [ 7 ] Li 3 PO 4 , [ 8 ] and AlPO 4 [ 9 ] have been studied. Most of these coatings are, however, still restricted to the poor conductivity and/or nonuniformity. The former defi ciency results in poor kinetics during charging/ discharging, while the latter does not provide full protection of electrode from HF attacking. [ 10 ] Atomic layer deposition (ALD) is a novel coating technique capable of depositing highly conformal and uniform layers with well controlled thickness onto substrates. [ 11 ] ALD-derived ultrathin Al 2 O 3 and LiAlO 2 coatings have been used as protection layers on LNMO recently, [ 12 ] it was found that the coating layer containing lithium favors faster lithium-ion diffusion. Most of the nonlithium-containing coating materials, however, increase the cycling stability at the expense of sacrifi cing capacity. [ 13 ] For example, in their attempt to protect the surface of LNMO by ALD-derived Al 2 O 3 , Jung et al. [ 12b ] used only two ALD cycles of Al 2 O 3 growth on LNMO powders, the capacity dropped by 10 mAh g −1 immediately, when the ALD cycle number was increased to 10, induced by the surface coating, which was also observed in TiO 2 coating. [ 22 ] In order to observe the P content evolution with ALD cycles, the P 2 p spectra were collected by synchrotron XPS technique. It can be observed that these P atoms on the LNMO surface show increasing concentration when the ALD cycle increases, indicating that the amount of surface coating layer correlates to the number of ALD cycles. It is worthwhile to note that the P 2 p XPS spectra of LNMO-40 did not show linear intensity increase, this is probably due to the surface saturation in synchrotron XPS.
The morphologies of the LNMOn samples were characterized by FESEM ( Figure 2 a,b shows LNMO-0 and LNMO-20, the rest are shown in Figure S3, Supporting Information) and HRTEM (Figure 2 c). It can be seen that the pristine LNMO shows sharp crystallized edges, the surface becomes rougher when the ALD cycle number increases, Figure S2 (Supporting Information) shows the EDX mapping of Fe, P, Mn, Ni, and O, it can be seen that the Fe and P are uniformly coated onto the surface of LNMO. The HRTEM images in Figure 2 c reveal that the ultrathin surface coating is about 2 nm in thickness, the growth rate is consistent with our previous fi ndings when depositing FePO 4 onto Si wafer. [ 18 ] The lattice fringe with basal distance of 0.24 nm is consistent with the (222) spacing of cubic phase LNMO. The inset electron diffraction pattern of LNMO-20 indexes a typical spinel lattice structure. Figure 3 a shows the fi rst charge/discharge curves of LNMOn samples, the plateaus at around 4.7 V correspond to the reduction of Ni 4+ to Ni 3+ and Ni 2+ , another small plateau at around 4.0 V corresponds to the reduction of Mn 4+ to Mn 3+ . The bare LNMO delivers highest fi rst discharge capacity of 113 mAh g −1 among all the samples. Nevertheless, the LNMO-0 sample decays rapidly during cycling, and the capacity retention of LNMO-0 is only 79.89% after 100 cycles, as shown in Figure S5 Table 1 . In contrast, the ALD FePO 4coated samples display increasing capacity retention with more ALD cycle numbers, indicating the protective nature of the FePO 4 layer. [ 23 ] It is worthwhile to mention that despite the LNMO-40 sample shows greatly enhanced stability, the capacity is lower, possibly due to the relatively lower electrical conductivity of FePO 4 . Rate capability test (Figure 3 c) also reveals that LNMO-10 presents the highest capacity under high current densities, e.g., more than 80 mAh g −1 at 5 C, while the LNMO-0 sample drops to approximately 0 mAh g −1 . The Coulombic effi ciencies of the samples are shown in Figure S6 (Supporting Information), it can be seen that the Coulombic effi ciency increases with the ALD cycle number, revealing that the presence of FePO 4 has helped to suppress the electrolyte decomposition.
Cyclic voltammetry (CV) measurements were carried out on the LNMOn samples with normalized active material loading and electrolyte amount (Figure 3 d). Three redox couples can be observed in the CV profi les. The weak and broad pair at around 4.0 V corresponds to the Mn 3+ /Mn 4+ , indicating that the LNMO is mostly in the phase of P4 3 32, [ 21 ] in accordance with the Raman spectra. Two pairs of intense redox couples at 4.6-4.9 V are related to the Ni 2+ /Ni 3+ /Ni 4+ , which are the main sources of capacity. CV curves enlarged at 4.9-5.0 V ( Figure S7   higher resident current value at the cutting voltage of 5.0 V than other samples, implying that the electrolyte oxidation in bare sample is more severe than coated samples. The lower area of the LNMO-40 sample is also in accordance with the lower capacity. Table 1 summarizes the potential positions of the redox peaks. The redox peak potentials varied from 0.140 and 0.143 V for LNMO-0 to 0.082 and 0.084 V for LNMO-40 FePO 4 , suggesting that FePO 4 coatings alleviates the polarization of the LNMO materials. In the effort to understand the formation of solid electrolyte interface (SEI) on the surface of the electrodes, AC electrochemical impedance spectra (EIS) were conducted on each LNMOn sample after cycling for 100 times and subsequently charged to 5.0 V as shown in Figure 4 a. It can be seen that the LNMO-0 sample shows only one semicircle, whereas those with FePO 4 coatings show two semicircles in the range of high and medium frequencies. A possible equivalent circuit is proposed to illustrate the impedance behaviors on the surface as shown in Figure 4 b. R Ω stands for the Ohmic electrolyte resistance. The semicircle at high frequency is suggested to be a resistor R s and a constant phase element (CPE), which are related to the migration of Li + through the surface fi lm, in this case, it refl ects the resistance of SEI. Another semicircle at medium frequency is related to the charge transfer reaction composed of R ct and another CPE, together with the fi nite length Warburg impedance. [ 24 ] The values of the R s are presented in Table 1 , it can be found that without any FePO 4 coating, the R s for LNMO-0 is 173.1 Ω, however, the existence of FePO 4 coating layer helped to decrease the R s values dramatically, which vary monotonically with the number of ALD cycles, to only 57.9 Ω for the LNMO-40 sample. The drop of R s clearly reveals the less formation of insulating SEI, which is a result of electrolyte decomposition, therefore FePO 4 is effective towards suppressing the electrolyte decomposition.   Table 1. Potentials of the oxidation/reduction peaks of the fi rst CV scan, the capacity retentions and R s after 100 charge/discharge cycles.
Capacity To investigate the change of Mn valence state in the LNMOn samples, XANES was collected on the Mn L 3,2 -edges. Mn L 3,2edges illustrate the electronic transition from Mn 2 p 3/2 and 2 p 1/2 to an unoccupied 3 d state. [ 25 ] Figure 5 a depicts the total electron yields (TEY) of LNMO-0 and LNMO-20, which is surface sensitive with a probing depth of 5-10 nm. The L 2 -edge often appears to be broader due to the core hole lifetime as explained by Coster-Kronig Auger decay. [ 26 ] It can be seen that both the LNMO-0 and the LNMO-20 show predominantly Mn 4+ features that fi t well with standard MnO 2 , the small peak at 646 eV corresponds to Mn 3+ , and this is also consistent with the Raman spectra, the unchanged spectra reveal that the coating process did not generate changes to the surface phase of LNMO. However, after charge/discharge cycling, Mn 4+ at the surface was partially reduced to Mn 2+ , and the LNMO-0 shows much higher intensity ratio of Mn 2+ /Mn 4+ than the coated LNMO sample. The bulk-sensitive fl uorescence yield (FYI) spectra of LNMO-20, LNMO-20 after battery cycling and LNMO-0 after battery cycling are shown in Figure S8 (Supporting Information). It can be seen that the bulk Mn exhibits subtle changes after cycling. The less reduced Mn valence on coated LNMO surface also reveals weaker reduction by the electrolyte, which can be attributed to the inhibitive role of FePO 4 against the electrolyte oxidation. [ 27 ] It is also generally accepted that the presence of Mn 3+ triggers the Jahn-Teller distortion because of its ( t 2g 3e g 1 ) confi guration, resulting in its charge disproportionation into non Jahn-Teller active Mn 2+ and Mn 4+ , described as 2Mn 3+ → Mn 2+ + Mn 4+ . [ 3a , 28 ] In the presence of HF from the LiPF 6 salt, Mn 2+ ions dissolve in the electrolyte and migrate through the separator followed by depositing on the anode as Mn metal, with a secondary phase formed on the surface of cathode materials. [ 29 ] The suppression of Jahn-Teller distortion by FePO 4 coating prevents the formation of Mn 2+ , thereby decreases the chance of Mn 2+ dissolution in HF, and improves the stability of LNMO. [ 30 ] Fe L 3 -edges XANES of standard FePO 4 , LNMO-20, and LNMO-20 after battery cycling were performed to determine the chemical states of the FePO 4 coatings before and after charge/discharge cycling. As shown in Figure 5 b, the spectrum of LNMO-20 fi ts well with the standard FePO 4 spectrum, the intense peak at 713.5 eV (can be ascribed to the dominant spectral feature of Fe 3+ ) and the weaker peak at 712.2 eV are related to the spin-orbit, interplay of crystal-fi led, and electronic interactions. Their intensity ratio reveals the Fe 3+ /Fe 2+ ratio. [ 31 ] Nevertheless, upon battery cycling, there is an obvious drop in the Fe 3+ /Fe 2+ ratio, indicating that part of the Fe 3+ has been reduced, and the position of the right peak is, interestingly, shifting to higher energy value. Such shift was also observed in our previous study on the soft XANES spectroscopies of FePO 4 -related various phases. [ 31 ] In this regard, we believe that the insertion of lithium ions into the matrix of amorphous FePO 4 has resulted in the partially lithiated FePO 4 domains, which acts as a lithium-ion reservoir and exhibited improved performance at high current densities by providing abundant Li + diffusion pathways.   Based on the aforementioned results, the schematic illustration of the protecting role of FePO 4 is presented in Figure 6 . The LNMO-0 exposed to electrolyte suffers from side reactions such as fi erce transitional metal dissolution and continuous electrolyte decomposition. On the contrary, LNMO with FePO 4 coating is resistant to the side reactions. This is because it was found that the noncoated sample displayed Mn at reduced state on the surface after cycling, which is much more prone to dissolution compared with Mn 3+ and/or Mn 4+ . [ 29b ] Additionally, the amorphous FePO 4 layer accommodates lithium ions rapidly during cycling, thus provides fast lithium diffusion. More specifi c role of FePO 4 is shown in Figure 6 b with the electrolyte's highest occupied molecular orbital (HOMO) and work functions of FePO 4 and LiNi 0.5 Mn 1.5 O 4 . The electrolyte gets readily oxidized when the electrochemical potentials of cathode materials are below the HOMO of it. [ 20 ] Unlike other conventional insulating ALD coating materials such as Al 2 O 3 or ZrO 2 , and FePO 4 is electrochemically active with an open-circuit voltage of ≈3 V, [ 15a ] the FePO 4 ultrathin layer on the surface prevents the direct contact of LNMO with the electrolyte, helping to avoid the oxidation of electrolyte that results in the reduction and dissolution of Mn ions.
We have proposed a new FePO 4 coating on high voltage LNMO cathode material enabled by ALD. Different thicknesses of FePO 4 have been deposited onto LNMO powders with 5, 10, 20, and 40 ALD cycles. The LNMO coated with 10 ALD cycles of FePO 4 showed the best performance including the highest capacity and stabilized capacity retention under all the current rates. When the LNMO was coated with 40 ALD cycles of FePO 4 , the capacity retention increased up to 100%. XANES study showed that the ultrathin FePO 4 suppressed the surface Mn 4+ from being heavily reduced to Mn 2+ by the reduction from the electrolyte and the Jahn-Teller distortion, less amount of Mn 2+ helped to retain the surface consistency without severe dissolution into the electrolyte. The FePO 4 coating layer was slightly reduced due to the remaining Li + in the structure after charge/discharge cycling. Compared with the most widely used insulating Al 2 O 3 , amorphous FePO 4 presents many advantages on the electron/ion diffusion on the surface. Our work provides an alternative option of depositing materials onto powders instead of electrode sheets directly using ALD, which expands the deposition temperature, owing to the electrochemically active nature of FePO 4 .

Experimental Section
Materials Synthesis : LiNi 0.5 Mn 1.5 O 4 was synthesized via a two-step hydrothermal-assisted carbonate precipitation method followed by thermal treatment. Ni(NO 3 ) 2 ·6H 2 O (99%, Aldrich, 0.005 mol) and Mn(NO 3 ) 2 · 4H 2 O (99%, Aldrich, 0.015 mol) were dissolved in deionized water (5 mL). Na 2 CO 3 (99%, Aldrich, 1 mol L -1 , 20 mL) solution was subsequently added to the above mixture of nickel nitrate and manganese nitrate under vigorous stirring at a rate of 0.25 mL min -1 , then the green precipitation was transferred to a 40 mL Tefl on-lined autoclave and kept at 140 °C for 10 h. After cooling down to room temperature (RT), the precipitation was fi ltered and washed with water several times and dried at 80 °C overnight. The carbonate powders were annealed at 450 °C for 4 h in air so as to obtain corresponding oxides. Thereafter, the oxide powders were mixed with Li 2 CO 3 (99%, Sigma-Aldrich, 0.00503 mol) in 1:1 water and ethanol mixture (10 mL) and left to dry under stirring at 60 °C. The mixed precursor was subsequently sintered in O 2 at 800 °C for 6 h and then cooled to 600 °C in 3 h. After keeping at 600 °C for another 6 h, the furnace was cooled to RT at a cooling rate of 1 °C min −1 to obtain the fi nal LNMO.
Characterization Methods : The morphology of LNMOn was characterized by a Hitachi S-4800 fi eld emission scanning electronic microscopy (FESEM) equipped with an energy dispersive X-ray spectroscope (EDS), Hitachi H-7000 transmission electron microscope (TEM), and a high-resolution transmission electron microscope (HRTEM, JEOL 2010F). Raman scattering (RS) spectra was collected from a HORIBA Scientifi c LabRAM HR Raman spectrometer system with a 532.4 nm laser and optical microscope at RT. X-ray diffraction (XRD) patterns were collected on a Bruker D8 Advance Diffractometer using Cu K α radiation at 40 kV and 40 mA. The X-ray absorption near  edge structure (XANES) measurements at total electron yield (TEY) and fl uorescence yield (FYI) modes of Mn L 2,3 -edge and Fe L 3 -edge were performed at the Canadian Light Source (CLS) on the high resolution Spherical Grating Monochrometer (SGM) beamline using a 45 mm planar undulator and three gratings with a photon energy range of 250-2000 eV, LNMO-20 was chosen as the target sample. The P 2 p X-ray photoemission spectroscopy (XPS) was performed at the variable line spacing plane grating monochromator (VLS PGM) beamline at 200 eV photon energy with a total resolution of 100 meV.
Electrochemical Measurements : The LNMOn powders were mixed with poly(vinylidene fl uoride) binder and acetylene black in a ratio of 8:1:1 in N-methyl-pyrrolidione (NMP) solvent to form slurries. The slurries were subsequently casted onto aluminum foils as the current collector and dried at 80 °C under vacuum overnight. The electrode was assembled in an Ar-fi lled glovebox with moisture and oxygen concentrations below 1 ppm. A CR-2032 type coin cell using a lithium metal as the counter electrode and Celgard K2045 as the separator was utilized. The electrolyte was composed of 1 M LiPF 6 salt dissolved in ethylene carbonate (EC) and dimethyl carbonate (DMC) in a 1:1 volume ratio (BASF Corp.). Cyclic voltammetry (CV) was performed on a versatile multichannel potentiostat 3/Z (VMP3), with a scanning rate of 0.1 mV s −1 and a potential range of 3.5−5.0 V (vs Li + /Li) at RT. Electrochemical impedance spectroscopy (EIS) was also performed on the versatile multichannel potentiostat 3/Z (VMP3) by holding the cells at 5.0 V. Galvanostatical charge/discharge was performed on Arbin BT2000 at various current densities between 3.5 and 5.0 V (vs Li + /Li), the stability performance test was done under 0.5 C, which is 73.5 mA g -1 .

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.