Enhancing surface‐to‐bulk stability of layered Co‐free Ni‐rich cathodes for long‐life Li‐ion batteries

Layered Co‐free Ni‐rich cathodes are the most cost‐effective for high‐energy‐density Li‐ion batteries (LIBs), yet the structural instability and interfacial side reactions seriously hamper their commercial applications versus the Co‐contained counterpart. Herein, a synchronous Ge‐doping and Li4GeO4‐coating of the Co‐free Ni‐rich cathodes have been realized to tackle these limitations during high‐temperature lithiation of the corresponding hydroxide precursors. The nonmagnetic Ge4+ doping effectively relieves the magnetic frustration and the lattice oxygen loss with reduced cation mixing and gas emission. The Li‐ion conductive and anti‐erosive Li4GeO4 coatings contribute to enhance the surface chemistry stability and Li‐ion migration interface kinetics. Consequently, the resulting Co‐free Ni‐rich cathodes delivers a high reversable capacity of 223.3 mAh g−1 at 0.1 C and maintains 127.5 mAh g−1 even at 10C within 2.7–4.4 V. More impressively, it displays high‐capacity retention of 90.5% in coin‐type half‐cell after 150 cycles and 80.5% in pouch‐type full‐cell after 500 cycles at 3C, demonstrating a long‐term cycle life.

alleviated with the absence of Co 3+ , resulting in serious Li/Ni mixing and sluggish Li + transportation. [6][7][8] The ultrahigh Ni content in NM increases the covalency of the Ni-O bond, which would damage the stability of the lattice oxygen framework during repeated cycling. 9,10 Additionally, the abundant highly oxidative Ni 4+ on the NM surface at the deeply charged state would ineluctably induce serious side reactions at the cathode-electrolyte interface, deteriorating the electrochemical performances of the cathodes. 11,12 Surface engineering and lattice doping are proven to be effective tactics to solve the above problems. 13,14 The coating layer on material surface could boost the interfacial stability by preventing the cathodes from directly contacting with the corrosive hydrogen fluoride. 15,16 Several materials (graphene, La 2 O 3 , etc.) have been employed to modify the surface of Co-free Ni-rich cathodes and effectively enhanced their electrochemical performances. 17,18 However, such postprocessing strategies are unable to reduce the severe Li/Ni antisite defects in the Co-free layered oxides. Research indicate that introducing doping elements (e.g., Sr, Al, and Ti) with stable metaloxygen bonds into the material lattice can strengthen its oxygen framework and decrease the cation mixing. [19][20][21] For example, Kang and colleagues 22 compared the energy barriers of the Ni 2+ migration from transition metal (TM) slabs to Li interslabs in LiNiO 2 before and after Ge doping by density function theory (DFT) calculation. The results manifest that doped Ge in TM slabs raises the migration barrier by 2.08 eV, which could reduce the probability of cation mixing in the material and is hopeful to improve the performances. However, the research on Ge-doped Co-free layered cathodes only stays at the level of theoretical calculation. 22 Besides, the heterogeneous element doping cannot resolve the issue of solid-liquid interfacial instability. Therefore, simultaneously stabilizing the crystal structure and material surface through a simple method is of great significance for preparing high-performance Co-free Ni-rich cathodes.
Herein, we demonstrate the surface-to-bulk modified LiNi 0.9 Mn 0.1 O 2 cathode through directly calcining the GeO 2 -coated precursors, during which the hightemperature could induce the in situ formation of Li 4 GeO 4 -coating and Ge-doping. The Li + conductive Li 4 GeO 4 surface skin performs as a barrier to reduce the parasitic reactions, whereas the lattice doping of nonmagnetic Ge 4+ effectively stabilizes the ordered layered structure and the lattice oxygen. The collaborative effect of the Ge 4+ and Li 4 GeO 4 modification enables the material with excellent lithium storage properties.

| Materials preparation
First, the Ni 0.9 Mn 0.1 (OH) 2 precursors were prepared by a co-precipitation method in a batch reactor (5 L) under an argon atmosphere. Typically, 3.2 mol L −1 NH 4 OH (aq.) and the solution of NiSO 4 · 6H 2 O and MnSO 4 · H 2 O (2 mol L −1 , Ni:Mn = 9:1 in molar ratio) were added dropwise into the reactor, which already contains an aqueous solution of NH 4 OH (0.8 mol L −1 ). Meanwhile, the NaOH (aq.) was added into the reactor simultaneously to make the pH constant (pH 11.0). After stirring for 30 h, the Ni 0.9 Mn 0.1 (OH) 2 precursors were acquired via washing, filtering, and vacuum drying at 100°C for 12 h. The GeO 2 -treated Ni 0.9 Mn 0.1 (OH) 2 precursors were then prepared through a wet-chemical method. Specifically, a certain amount of GeO 2 ((Ni+Mn):Ge = 100:n in molar ratio, n = 1.5, 1.0, 0.5, 0) were dissolved in 30 ml of deionized water, followed by the addition of Ni 0.9 Mn 0.1 (OH) 2 (2.0 g) into the solution. The mixed solution was stirred at 100°C until the water evaporated. The powder was collected and dried in a vacuum overnight. Subsequently, the powder was homogeneously blended with LiOH·H 2 O (Li:TM = 1.05:1 in molar ratio) and preheated at 450°C for 6 h, then sintered at 770°C for 13 h under a flowing O 2 to gain the Co-free Ni-rich cathodes. The cathodes with 0%, 0.5%, 1%, and 1.5% Ge content were denoted as NM, NMGe-0.5, NMGe-1, and NMGe-1.5, respectively.

| Characterization
The crystal structure of Co-free layered oxides was studied by the X-ray diffraction (XRD, Bruker D8 Advance). The corresponding structural parameters were acquired via Rietveld refinement (GSAS software). The element concentration of the prepared samples and the dissolved Ni mass on Li anode were measured by the inductively coupled plasma (Agilent 725). The morphologies of the cathodes were investigated by the scanning electron microscopy (SEM, TESCAN GALA 3). The surface element compositions of the material were researched through the X-ray photoelectron spectra (XPS) using Al Kα radiation. The XPS data were then calibrated by the C 1s peak (284.8 eV) before analyzation. The laser scattering particle size and shape analyzer (Microtrac S3500SI) was used to investigate the particle size distribution. The microstructure and element distribution of the material were studied via the scanning transmission electron microscopy (STEM, FEI Talos F200X) with an accessory of energy dispersive spectrometer.

| Electrochemical measurements
The 2016 coin-type half cells and pouch full cells were fabricated to research the electrochemical performances of the Co-free Ni-rich cathodes. For the half cells, cathodes, poly(vinylidene fluoride) (PVDF) and Super-P (8:1:1 in mass ratio) were first mixed homogeneously in N-methyl-2-pyrrolidone for 3 h to obtain the cathode slurry. Whereafter, the slurry was coated on the Al foil, dried in a vacuum under 120°C for 12 h, and cut into disks (Φ12 mm). The loading mass of the cathodes is around 2.0 mg cm −2 . The 2016-type coin cells were then assembled in the glove box under an Ar atmosphere and the cells were composed of cathode disks, polypropylene separator (Celgard 2400), Li metal, and electrolyte. For the pouch cells, the cathode electrodes with loading mass of~15 mg cm −2 were matched with commercial graphite anodes (N/P ratio of~1.1). The sizes of the positive and negative electrodes used in the pouch-type full cells were 41 × 62 mm 2 and 43 × 65 mm 2 , respectively. After welding the pole ears, the cathode electrodes, separator, and anodes were stacked in sequence, then encapsulated by the Al-plastic film. Subsequently, the electrolyte injection and sealing were conducted under an Ar atmosphere, followed by the formation and degassing process. The electrolyte was fabricated through dissolving the LiPF 6 (1.2 mol L −1 ) and vinylene carbonates (2 wt%) in the mixed solution of ethyl methyl carbonate and ethylene carbonate (7:3 in volume ratio). The pouch cells were charged at 1C and discharged at 3C using the Neware battery testing machine (CT-3008-5V3A-S1) to estimate the cycling performance (25°C, 2.75-4.3 V). The electrochemical impedance spectroscopy (EIS, 10 5 -10 −2 Hz) and the cyclic voltammetry (CV) curves (scan rates of 0.2-2.0 mV s −1 ) were tested at the electrochemical workstation (Autolab, PGSTAT302N). The galvanostatic charge-discharge measurements (25°C, 2.7-4.4 V) under different current density (1C = 185 mA g −1 ) were carried out on battery test machine (LANDCT2001A).

| RESULTS AND DISCUSSION
The Li/Ni mixing and oxygen depletion are reported to be the serious issues faced by the Co-free Ni-rich cathodes, which would lead to the deterioration of their electrochemical performances ( Figure 1A). 23 In-depth analysis reveals that the lone-pair electrons in the e g orbital of Ni 3+ would cause severe magnetic frustration in the material, resulting in an unstable state ( Figure 1B). Thus, the nonmagnetic Li + will exchange with partial Ni ions to relieve the magnetic frustration, but this inevitably disorders the material structure. 24 The Ge 4+ is also nonmagnetic, which could act as a stabilizer in the TM layers to alleviate the magnetic frustration and Li/Ni mixing. 25 Besides, the presence of oxygen vacancy in the material will also aggravate the cation mixing. Theoretical calculations indicate that Ge-doping could effectively inhibit the O 2 release, creating a robust oxygen framework and reducing the cation mixing ( Figure 1C). 26 The sluggish Li + migration is another factor restricting the development of Co-free Ni-rich layered oxides. Based on the DFT calculation, the doped Ge could effectively reduce the Li + diffusion barrier and accelerate the reaction kinetics. 22,27 The electrolyte erosion would also deteriorate the electrochemical performances of the cathodes during cycling, which means surface coating is necessary. To solve the aforementioned problems of Co-free Ni-rich cathodes, we coated the precursors with GeO 2 first and the simultaneously modified LiNi 0.9 Mn 0.1 O 2 with Ge-doping and Li 4 GeO 4coating can be obtained through further one-step calcination (Supporting Information: Figure S1). During the calcination process, the GeO 2 on the hydroxide precursors could react with the lithium source to form Li 4 GeO 4 coating, whereas partial Ge 4+ would diffuse into crystal lattice to achieve the dual modification. 28 Four samples with various Ge contents were prepared and then investigated by the XRD and Rietveld refinement to study the influence of Ge on crystal lattice (Supporting Information: Figure S2 and Table S1). As shown in Figure 1D and Supporting Information: Figure S3, except for the diffraction peaks corresponding to the Li 4 GeO 4 , the other peaks of obtained samples match well with the hexagonal α-NaFeO 2 -type structure (R-3m space group). Additionally, the clear splitting of (006)/(012) and (018)/(110) peaks indicates their well-ordered layered structure (Supporting Information: Figure S4). The lattice constants, a and c, increase in succession with the addition of the Ge content, indicating the successful Ge doping ( Figure 1E and Supporting Information: Table S2). Due to the relieved magnetic frustration and strengthened lattice oxygen, the cation mixing can be significantly reduced ( Figure 1F). According to the Rietveld refinement, the NMGe-0.5 with optimal Ge amount possesses the lowest Li/Ni antisite defects of 1.66%, superior than those of the NM (2.56%), NMGe-1 (2.27%), and NMGe-1.5 (3.05%). It is widely accepted that the large peak intensity ratio of (003) and (104) implies the decreased cation mixing in Ni-rich cathodes and the results also indicate NMGe-0.5 has the most ordered layered structure. The surface element compositions of the prepared cathodes were further researched by the XPS. Figure 1G displays the Ge 3d spectra and the peak intensity of Ge 4+ fortifies with increased Ge concentration. 29 The O 1s peaks at around 531.5 and 528.9 eV are related to the surface oxygen and the lattice oxygen. Obviously, NMGe-1.5 shows the highest lattice oxygen content (12.9%), indicating that the robust Ge-O bond (657.5 kJ mol −1 vs. 366.0 kJ mol −1 of Ni-O and 362.0 kJ mol −1 of Mn-O bonds) is beneficial to stabilizing the oxygen framework ( Figure 1H). 30 For the Ni 2p3/2 spectra shown in Figure 1I, the peaks located at 855.9 and 854.6 eV are identified as Ni 3+ and Ni 2+ , respectively. The introduction of Ge 4+ increases the Ni 2+ content due to charge conservation and overmuch Ni 2+ would aggravate the cation mixing, which results in the highest Li/Ni mixing of NMGe-1.5 among four samples. However, the alleviated magnetic frustration and stabilized oxygen framework originated from modification are helpful to suppress the cation mixing. Therefore, the NMGe-0.5 with suitable Ge content displays the lowest degree of cation mixing under the synergistic influence of above factors.
The influence of Ge-doping and Li 4 GeO 4 -coating on material morphology was further studied by the SEM and STEM technologies. As shown in Figure 2A,B and Supporting Information: Figure S5a, the NMGe-0.5 displays uniform spherical morphology with the median diameter (D50) and distribution span value of 12.14 μm and 0.53, which is similar with those of the pristine NM (Supporting Information: Figure S6a,b). In addition, the small distribution span value manifests that the NMGe-0.5 particles are uniform. The samples before and after modification are all consisted by nano-sized primary particles. The surface of NMGe-0.5 becomes rough, whereas NM exhibits a smooth surface, implying the formation of surface Li 4 GeO 4 coating on modified sample (Supporting Information: Figures S5b and S6c). The high-angle annular dark-field imaging-STEM image and relevant elemental mappings are depicted in Figure 2C, which demonstrate the homogeneous distribution of Ni, Mn, and Ge elements. Moreover, there is a continuous coating skin on the surface of NMGe-0.5, whereas that of NM is fresh ( Figure 2D and Supporting Information: Figure S6d). High-resolution STEM image reveals that In addition, the relevant FFT pattern shows that the bulk region of NMGe-0.5 still exhibits a well-ordered structure.
The 2016-type half cells assembled by the prepared samples and Li anode were then tested under 25°C to estimate the electrochemical performances of the material with different Ge concentration. Owing to the stabilized lattice oxygen and solid-liquid interface, the NMGe-0.5 exhibits an outstanding initial Coulombic efficiency of 88.7%, much higher than the pristine NM (84.1%) ( Figure 3A and Supporting Information: Figure S7). Besides, the NMGe-0.5 could deliver a high reversible capacity of 223.3 mAh g −1 at 0.1C ( Figure 3B). When the current density comes to 10C, the NMGe-0.5 possesses the highest capacity of 127.5 mAh g −1 , outperforming the NM (106.9 mAh g −1 ), NMGe-1 (115.3 mAh g −1 ), and NMGe-1.5 (104.9 mAh g −1 ). Impressively, the capacity of NMGe-0.5 at 0.1C could return to 218.0 mAh g −1 after charge-discharge at 10C, whereas that of NM drops to 203.8 mAh g −1 . The cycle performances tested at 2.7-4.4 V reveal that the NMGe-0.5 with proper Ge content displays the supreme capacity retention of 90.1% ( Figure 3C), far better than those of the unmodified NM (81.1%), NMGe-1 (86.1%), and NMGe-1.5 (84.2%). The development of voltage hysteresis during repeat charge-discharge at 1C for NM and NMGe-0.5 is detailly analyzed in Figure 3D. Notably, the polarization of NMGe-0.5 (0.027 V) is less than half that of the NM (0.055 V), reflecting the Ge-doping and Li 4 GeO 4 -coating could effectively accelerate reaction kinetics and enhance structural stability of Co-free Ni-rich cathodes. Even when cycles at 3C for 150 cycles, the NMGe-0.5 could still deliver a capacity retention of 90.5%, outperforming the NM (76.8%) ( Figure 3E). The self-discharge profiles were further measured at 55°C to investigate the surface stability of the materials. In detail, the coin cells were charged to 4.4 V (Supporting Information: Figure S8) and kept at 55°C for 20 h, then discharged to 2.7 V. It can be seen that the open-circuit voltage retention during storge ( Figure 3F) and discharge capacity ( Figure 3G) of NMGe-0.5 are both higher than those of the NM, which indicates that the dual modification could effectively reduce the side reactions and create a stable surface. The pouch cells with the Co-free Ni-rich layered oxides and graphite were assembled and then tested at a 1C-3C charge-discharge rate to assess their long-term cycling performances in full cells ( Figure 3H). When cycles at a high discharge rate of 3C for 500 cycles within 2.75-4.3 V, the NMGe-0.5 still exhibits a discharge capacity retention of 80.5%, whereas that of NM is only 52.9%, reflecting the Ge-doping and Li 4 GeO 4 -skin are beneficial to enhancing the structural stability and cycling performance.
To deeply understand the influence of simultaneous optimization on lithiation-delithiation, the first three CV curves were measured at 0.2 mV s −1 under 2.7-4.4 V. As depicted by Figure 4A,B, the NM and NMGe-0.5 undergo similar reaction processes during charging and discharging, as proved by the analogous CV curves, manifesting the working mechanism of the material still maintains after the dual modification. Nevertheless, the potential intervals of CV peaks for the NMGe-0.5 (0.111 and 0.112 V) are much smaller than those of NM (0.147 and 0.205 V), which reflects the reaction kinetics can be greatly accelerated due to the strengthened lattice oxygen and superionic conductive Li 4 GeO 4 skin. It is worth noting that the latter two CV patterns of NMGe-0.5 overlap better than the pristine NM, displaying the enhanced redox reversibility of NMGe-0.5. The CV patterns with various sweep rates (0.2-2.0 mV s −1 ) are shown in Supporting Information: Figure S9. As the scan rate increases, the CV peaks of NM exhibit severe polarization, whereas those of NMGe-0.5 are still recognizable. According to those CV curves, the peak current (i p ) is linearly dependent on the square root of the sweeping rate (v 1/2 ), indicating that the reaction is controlled by the diffusion procedure ( Figure 4C). 31,32 Moreover, the Li + diffusion coefficient of NMGe-0.5 is 1.23 × 10 −9 cm 2 s −1 /2.07 × 10 −10 cm 2 s −1 for the de-/ lithiation process based on the Randles-Sevcik equation, which is superior to the NM (4.31 × 10 −10 cm 2 s −1 / 1.39 × 10 −10 cm 2 s −1 ). To analyze the effect of dual modification on electrochemical process during cycling, the calculated dQ/dV curves of different cycles are shown in Figure 4D. Clearly, both samples undergo multi-step phase transitions during delithiation: originally from hexagonal phase (H1) to monoclinic phase (M), then to hexagonal phase (H2), and finally to another hexagonal phase (H3). In the process of lithiation, the above phase transitions will happen in reverse. 33 The H2-H3 redox peaks of different cycle numbers for NMGe-0.5 represent negligible change in peak potential and intensity ( Figure 4E), whereas those of the pristine NM decay quickly, indicating the robust oxygen framework and reduced side reactions could increase the reaction reversibility. The EIS measurement before and after cycling was then conducted to analyze the impedance growth. As depicted in Supporting Information: Figure S10, the Nyquist plots are consisted of two semicircles and a slope line, corresponding to the surface resistance (R sf ), charge transfer resistance (R ct ), and Warburg impedance, respectively. To quantitatively analyze the EIS data, the Nyquist plots were fitted and the equivalent circuit is presented in Supporting Information: Figure S11. Due to the optimized structure, the R ct of NMGe-0.5 (44.6 Ω) is lower compared with the NM (54.1 Ω) in the first cycle ( Figure 4F and Supporting Information: Table S3). After long-term cycling, accompanied by the lattice oxygen loss and electrolyte erosion, the R ct of the material will inevitably increase. 34 Impressively, the R ct of NM after 100 cycles (447.2 Ω) is about 2 times higher than that of NMGe-0.5 (144.3 Ω), reflecting the co-modification of Co-free Ni-rich cathodes is conducive to suppressing the structural degradation and unfavorable side reactions.
The cycled half cells were disassembled in glove box and researched through multiple characterization to fully comprehend the function of Ge-doping and Li 4 GeO 4doping. As depicted by Figure 5A, the well-ordered layered structure of NMGe-0.5 is still preserved after repeated Li + intercalation-deintercalation, whereas the structure of NM experiences serious cation mixing with the emergence of harmful rock-salt phase NiO, impeding the Li + migration. The fragmentation and pulverization of cathode material particles and the resulting electrolyte corrosion along cracks are critical failure mechanisms for Ni-rich cathodes. 35,36 Therefore, the morphology of the cycled cathodes was studied by the SEM (Supporting Information: Figure S12). Clearly, there is no distinct microcracks in the NMGe-0.5 particles owing to the stabilized bulk structure and solid-liquid interface, as evidenced by the numerous cracks in the unmodified NM particles. During the cycling, the TM ions would dissolve in electrolyte and then migrate to the anode side, destroying the crystal lattice of the cathode and the solid electrolyte interphase on anode. 11 Therefore, the digital photos of Li anode and the deposited Ni mass on it were F I G U R E 4 (A, B) Initial three cyclic voltammetry (CV) curves at 0.2 mV s −1 within 2.7-4.4 V, (C) linear relationship between anodic/ cathodic peak current (i p ) and the square root of the scan rate (v 1/2 ), (D, E) the calculated dQ/dV curves during 100 cycles at 1C, and (F) the charge transfer resistance before and after cycling for the NMGe-0.5 and NM. acquired to investigate the TM dissolution ( Figure 5B). It is apparent that the black region associated with the TM ions deposition in NMGe-0.5 is much smaller than that in the NM. The dissolved Ni mass of NMGe-0.5 is 0.85 μg, far lower than the pristine NM (4.60 μg), reflecting that the bulk Ge-doping and surface Li 4 GeO 4coating could suppress the dissolution of active substances and endow the material with enhanced cycling stability. XPS characterization was then used to analyze the surface compositions of the cathode electrodes. In the C 1s spectra (Figure 5C), the peaks located at 284.8, 285.7, 287.0, 288.9, and 290.9 eV are assigned to C-C (super P), C-H (PVDF), C-O, C=O, and C-F (PVDF), in which the C-O and C=O are derived from the electrolyte decomposition products. Notably, the peak area of the byproducts for NMGe-0.5 is much smaller than that of the NM, which proves the synergistic modification is beneficial to inhibiting the electrolyte decomposition. For the F 1s spectra in Figure 5D, the peak area of Li x PO y F z and LiF/NiF 2 for NMGe-0.5 is much smaller compared with the NM, demonstrating the interfacial parasitic reactions are significantly suppressed because of the robust lattice oxygen and surface protective layer. 37,38

| CONCLUSION
In summary, the concurrently modified Co-free LiNi 0.9 Mn 0.1 O 2 with lattice Ge-doping and interface Li 4 GeO 4 -coating (NMGe) was synthesized via a facile one-step sintering the precursors. The magnetic frustration can be effectively alleviated owing to the nonmagnetic Ge 4+ doping at TM slabs, which helps to reduce the cation mixing. The robust Ge-O bond is conducive to suppressing the oxygen loss and stabilizing the crystal structure. Moreover, the superionic conductive Li 4 GeO 4 skin could accelerate the Li + migration and inhibit the interfacial side reactions. Thus, the dual-modified strategy endows the Co-free NMGe with enhanced reaction kinetics and cycling stability. In the coin-type half-cells, the NMGe delivers a high specific capacity of 223.3 mAh g −1 at 0.1C and 127.5 mAh g −1 even at 10C. When charge-discharge at 3C for 150 cycles, the NMGe could still deliver a superior capacity retention of 90.5%. Besides, a pouch-type full-cell assembled by the graphite and NMGe exhibits a capacity retention of 80.5% even after 500 cycles at 3C. This work proposes a novel synergetic modification tactic to promote the large-scale applications of Co-free Ni-rich layered oxides.
F I G U R E 5 (A) X-ray diffraction (XRD) patterns, (B) the dissolved Ni mass and the photos of corresponding Li anode, and (C, D) C 1s and F 1s X-ray photoelectron spectra (XPS) spectra for the NMGe-0.5 and NM.