Hierarchical Ni- and Co-based oxynitride nanoarrays with superior lithiophilicity for high-performance lithium metal anodes

Preparation of NiCoON/NF current collector

NiCoON/NF was prepared by a hydrothermal method and subsequent ammoniation. The surface oxides of the Ni foam were removed by immersion in hydrochloric acid, followed by ultrasonic cleaning with acetone, water and ethanol sequentially. 2.62 mmol Co(NO₃)₂·6H₂O, 1.31 mmol Ni(NO₃)₂·6H₂O, 26.25 mmol urea and 10.5 mmol NH₄F were dissolved in 70 mL of deionized water by stirring and mixing well at room temperature and then transferred to a 100 mL Teflon-lined stainless-steel autoclave. The pretreated Ni foam (2 cm × 4 cm) was vertically immersed into the reaction solution and annealed at 100 °C for 8 h. The sample was then removed and the residual impurities on the surface washed away with deionized water and ethanol successively. The prepared Ni-Co precursor was directly calcined in a NH₃ atmosphere with a constant temperature of 350 °C for 3 h, with a heating rate of 2 °C min^-1 to obtain an oxynitride nanoarray on the Ni foam (NiCoON/NF). For the fabrication of NiCoO, the Ni-Co precursor was heated and calcined at 350 °C in an air atmosphere for 1 h to obtain the NiCoO sample on the Ni foam (NiCoO/NF).

Materials characterization

The surface and cross-sectional morphologies were observed using a field-emission S4800 instrument (Zeiss, Germany). The elemental mappings were captured using energy-dispersive X-ray spectroscopy. The high-resolution transmission electron microscopy (HRTEM) image was taken using a JEOL JEM-2100F. X-ray photoelectron spectroscopy (XPS) was carried out to obtain the surface chemical elements using a K-Alpha ESCALAB 250Xi system. All peaks were calibrated based on the C 1s peak (284.8 eV). X-ray diffraction (XRD) patterns were collected to analyze the phase composition on an XRD-6100 diffractometer coupled with Cu-Kα radiation (Shimadzu, Japan). The pore structure distributions and surface areas from the Brunauer-Emmett-Teller (BET) method were detected using an absorption analyzer (Micromeritics ASAP 2020) and N₂ adsorption-desorption isotherm at 77 K.

Electrochemical measurements

CR2032-type coin cells were used to assemble lithium metal batteries with Li as the reference and NiCoON/NF, NiCoO/NF or NF as the working electrode, where the water and oxygen contents were required to be less than 0.5 ppm. 1 M LiTFSI dissolved in 1,3-dioxolane/1,2-dimethoxyethane (v:v = 1:1) with 2 wt.% LiNO₃ and polypropylene (Celgard 2400) were selected as the electrolyte and separator, respectively. Before testing, all assembled batteries were activated for five cycles under a voltage window of 0.01-1.0 V to remove the contaminants at 50 μA and stabilize the SEI. To investigate the CE, a certain amount of lithium (1 mAh cm^-2) was plated onto different working electrodes, followed by charging up to 1 V to facilitate the release of Li at several current densities (1, 2 and 5 mA cm^-2). The CE was also tested under a higher areal capacity of 3 mAh cm^-2 with 1 mA cm^-2 and 2 mAh cm^-2 with 2 mA cm^-2. For the symmetrical cell, 3 mAh cm^-2 of Li was pre-electroplated on NF, NiCoO/NF and NiCoON/NF, followed by galvanostatic charge and discharge at 1 and 5 mA cm^-2 for 1 mAh cm^-2. With a higher areal capacity of 3 mAh cm^-2 at 1 mA cm^-2, the voltage hysteresis was also measured. All electrochemical properties were measured on multichannel LAND-CT2001A battery test equipment.

For the performance test of the full battery, NiCoON/NF, NiCoO/NF and NF were used as substrates for pre-lithiation to obtain Li@NiCoON/NF, Li@NiCoO/NF and Li@NF as anodes, respectively. Commercial LiFePO₄ was chosen as the stable cathode material. LiFePO₄ powder, Super P and polyvinylidene difluoride at a mass ratio of 80:10:10 were mixed in a N-methyl-2-pyrolidone solvent, followed by continuous stirring overnight. The homogeneous slurry was coated onto clean aluminum foil and the solvent was removed in a vacuum environment at 80 °C for 12 h with an area loading of 3.5 mg cm^-2. The corresponding full cells were charged and discharged between 2.4 and 4.2 V.

Preparation and structure of NiCoON/NF

NiCoON/NF was obtained by a typical hydrothermal method and subsequent ammoniation [Figure 1]. The Ni-Co precursor was first in situ grown on Ni foam though a hydrothermal process. After the low-temperature ammoniating treatment, the hydroxide was converted into a NiCoO₂/CoO/Ni₃N nanocomposite (abbreviated as NiCoON). As shown in Supplementary Figure 1, the bare Ni foam shows a smooth and flat surface with an open porous structure. The XRD pattern in Supplementary Figure 2 displays characteristic peaks. An interesting architecture with nanorod-on-nanosheet arrays was uniformly deposited onto a 3D Ni skeleton [Supplementary Figure 3]. After ammoniation, the morphologies of the Ni-Co precursor were well inherited. As shown in Figure 2A, the NiCoON/NF exhibits a rough surface. The high-resolution scanning electron microscopy (SEM) images in Figure 2B and C show a well-maintained nanorod-on-nanosheet structure with a unique hyperbranched architecture, which can afford sufficient sites for lithium nucleation and a large electroactive surface area for Li plating/stripping. The corresponding EDS elemental mapping of NiCoON/NF confirms the homogeneous distribution of elemental Ni, Co, O and N throughout the nanoarray matrix, indicating the successful implantation of nitrogen species [Supplementary Figure 4].

Figure 1. Schematic of preparation procedure of NiCoON/NF.

Figure 2. (A-C) SEM images of NiCoON/NF host. (D) Low-magnification and (E) high-resolution TEM images of NiCoON/NF. (F) XRD pattern of NiCoON/NF. (G) Ni 2p, (H) O 1s and (I) N 1s XPS spectra of NiCoON/NF. SEM: Scanning electron microscopy; TEM: transmission electron microscopy; XRD: X-ray diffraction; XPS: X-ray photoelectron spectroscopy.

The TEM image in Figure 2D further reveals a typical piece of the nanorod-on-nanosheet structure. From the HRTEM [Figure 2E], three well-resolved lattice intervals of 0.244, 0.214 and 0.246 nm can be ascribed to the (111) plane of rock-salt NiCoO₂, the (002) plane of hexagonal Ni₃N and the (111) plane of cubic CoO, respectively, proving the co-existence of NiCoO₂, CoO and Ni₃N. The formation of the NiCoON nanocomposite was further confirmed by the XRD patterns in Figure 2F. The typical diffraction peaks of NiCoO₂ (JCPDS 10-0188), CoO (JCPDS 48-1719) and Ni₃N (JCPDS 10-0280) are also consistent with the TEM findings. For comparison, the SEM images of NiCoO/NF prepared by calcination under air at 350 °C reveal a similar hierarchical nanostructure and uniform elemental distribution [Supplementary Figure 5]. However, no nitride peak in the XRD pattern of NiCoO/NF is found, corresponding with the TEM findings [Supplementary Figure 6]. In addition, the measured BET surface area of the NiCoON nanoarrays is 107.8 m² g^-1[Supplementary Figure 7], which is larger than that of NiCoO (101.2 m² g^-1). Furthermore, NiCoON presents a wide range of pore size distributions from 2 to 30 nm. These results show that NiCoON/NF with a large specific surface area can effectively reduce the local current density and guide the uniform distribution of lithium ions.

XPS was performed to determine the composition and chemical states of the as-synthesized NiCoON/NF. The corresponding full-scan spectra contain elemental Ni, Co, O and N, consistent with the EDS results [Supplementary Figure 8A].Figure 2G exhibits the high-resolution Ni 2p XPS spectrum containing Ni³⁺ (856.6 eV) and Ni²⁺ (854.8 eV), corresponding to Ni 2p_3/2 and Ni 2p_1/2, respectively^[35]. Importantly, an obvious peak of Ni⁺ (853.1 eV) is found in NiCoON/NF, indicating the partial reduction of Ni²⁺ to Ni⁺ after the nitriding treatment. The N 1s spectrum clarifies the existence of Ni₃N [Figure 2I]^[36]. Furthermore, the fitted O 1s spectrum shows three characteristic peaks and the peak at 530.9 eV corresponds to the oxygen defect/vacancy species [Figure 2H]^[37,38]. The Co 2p peaks of the NiCoON/NF composite show two more obvious central satellite peaks, indicating that some Co³⁺ were reduced to Co²⁺ and formed oxygen vacancies [Supplementary Figure 8B]^[39]. Compared to NiCoON/NF, no N element or Ni⁺ (853.1 eV) peak was found in the full-scan spectra of NiCoO/NF and the high-resolution Ni 2p XPS spectrum [Supplementary Figure 9], respectively. The corresponding oxygen defect/vacancy species at 531.4 eV in the O 1s spectrum of NiCoO/NF is also smaller than that of NiCoON/NF and the Co 2p peak of NiCoO/NF showed no visible satellite peaks. In general, N element is successfully introduced and the concentration of oxygen defect sites increases after the nitriding treatment, which can help to enhance the conductivity and lithiophilicity.

Electrochemical performance of NiCoON/NF

To better evaluate the electrochemical features of NiCoON/NF, the CE was tested via the assembly of half cells using Li as the reference electrode. As depicted in Figure 3A, NiCoON/NF shows the lowest nucleation overpotential of only 22 mV, much lower than that of NiCoO/NF (32 mV) and bare NF (42 mV) at a current density of 1 mA cm^-2. This clearly evidences that the strong lithiophilicity of NiCoON/NF is favorable for guiding the uniform Li nucleation. Figure 3B illustrates the CE comparison among the three samples. The NiCoON/NF host exhibits an inferior value of ~93.7% in the first cycle, which may be due to the conversion reaction between Li and NiCoON. This gradually rises to > 98% upon subsequent cycles and maintains stable cycling for 600 cycles. Compared with the published studies in Supplementary Table 1, such a CE is one of the best for 3D hierarchical hosts. In contrast, the CE of NiCoO/NF suffers from a sharp decline after 300 cycles, while bare NF shows a poor CE with huge fluctuations after 100 cycles due to the large consumption of electrolyte and lithium metal.

Figure 3. (A) Voltage profiles of nucleation overpotentials on different electrodes at 1 mA cm^-2. Comparison of CEs on NF, NiCoO/NF and NiCoON/NF current collectors with a fixed capacity of 1 mAh cm^-2 at (B) 1 mA cm^-2, (C) 2 mA cm^-2 and (F) 5 mA cm^-2 and (G) CE at 2 mA cm^-2 with 2 mAh cm^-2. (D, E) Corresponding discharge-charge curves on NiCoON/NF and NF at 2 mA cm^-2 with 1 mAh cm^-2. (H) Electrochemical impedance spectroscopy (EIS) after 1st cycle at 1 mA cm^-2. CE: Coulombic efficiency.

When the applied current density is increased to 2 mA cm^-2, the NiCoON/NF electrode still shows stable values of > 98% after 400 cycles, while the NiCoO/NF and NF are still in a state of poor CE [Figure 3C]. In addition, the corresponding discharge-charge curves of the NiCoON/NF host fit well with the CE and show a low nucleation overpotential in the first cycle. The voltage hysteresis is smaller than that of bare NF in the subsequent cycles, indicating the excellent stability of NiCoON/NF [Figure 3D and E]. Furthermore, even at a higher current density of 5 mA cm^-2, the NiCoON/NF still exhibits an excellent CE, while the NiCoO/NF and bare NF present severe fluctuations [Figure 3F]. Such excellent cycling stability of the NiCoON/NF electrode is mainly attributed to the formation of a stable SEI and the successful suppression of lithium dendrites.

Impressively, when the deposition capacity is substantially increased to 3 mAh cm^-2, NiCoON/NF achieves a stable CE at 1 mA cm^-2 for 160 cycles [Supplementary Figure 10]. Similarly, even at 2 mA cm^-2 for 2 mAh cm^-2 of Li, NiCoON/NF still retains a high CE of 97% over 200 cycles [Figure 3G]. In strong contrast, NiCoO/NF only maintains a stable state for 60 cycles and drops suddenly. Apparently, bare NF also experiences severe disturbances and faster drops, which are mainly attributed to the internal short circuit of the cell caused by Li dendrite formation. Electrochemical impedance spectroscopy measurements were also implemented to study the charge transfer resistance after one cycle. As displayed in Figure 3H, NiCoON/NF shows a significantly smaller charge transfer resistance than NiCoO/NF and bare NF, which confirms the outstanding Li plating and stripping kinetics and stable electrode interface. Therefore, these satisfactory results indicate that the higher electrical conductivity and better lithiophilicity of the NiCoON/NF nanoarrays largely reduce the Li nucleation barrier and facilitate the homogeneous Li plating/stripping.

Next, galvanostatic cycling tests based on symmetrical cells were carried out to explore the long-term plating/stripping stability at different current densities and capacities. Figure 4A displays the voltage-time profiles of NiCoON/NF, NiCoO/NF and NF at 1 mA cm^-2 with a fixed deposition capacity of 1 mAh cm^-2. Although both NiCoON/NF and NiCoO/NF present ultrastable cycling, the voltage hysteresis value of NiCoON/NF (23 mV) is lower than that of NiCoO/NF (26 mV), running over 1000 h without obvious potential oscillation. As a control, bare NF undergoes larger voltage hysteresis (30 mV) and irregular potential fluctuation appears only for 400 h due to the emergence of Li dendrites. At a current density of 5 mA cm^-2, the remarkable electrochemical stability of NiCoON/NF is still evident with a relatively low overpotential of 65 mV for 150 h, while NiCoO/NF and NF illustrate a short cycle life of 120 h with poor variability [Figure 4B], suggesting an unstable SEI layer and internal short circuit induced by aggravated Li dendrite formation. Moreover, even with a high capacity of 3 mAh cm^-2 at 1 mA cm^-2 [Figure 4C], NiCoON/NF still retains a conspicuous cyclability with an ultralong cycling life (2000 h) and ultralow overpotential (~20 mV), implying excellent electrochemical performance among the different 3D current collectors, as presented in Supplementary Table 2. However, the bare NF decays quickly with increased overpotential. The remarkably advanced cycling performance of NiCoON/NF can be ascribed to the nitrogen-implanted nanoarray architecture with enhanced conductivities and superior Li₃N from the lithiation conversion enriched SEI layer, which can promote reduced voltage hysteresis and homogeneous Li deposition/dissolution.

Figure 4. Long-term cycling performance of symmetrical batteries under different current densities and deposition capacities: (A) 1 mA cm^-2 with 1 mAh cm^-2; (B) 5 mA cm^-2 with 1 mAh cm^-2; (C) 1 mA cm^-2 with 3 mAh cm^-2.

Lithium deposition behavior

The Li metal deposition morphology on NF, NiCoO/NF and NiCoON/NF is also systematically evaluated via ex-situ SEM. As shown in Figure 5A, after depositing 1 mAh cm^-2 of Li onto the NF electrode, Li tends to sparsely and locally nucleate in the early stages. As the amount of Li deposition increases from 1 to 5 and 15 mAh cm^-2, visible protuberances and cotton-like lithium dendrites accumulate along the NF skeleton [Figure 5B and C]. There is no doubt that the poor affinity of the NF host for Li aggravates the random nucleation of lithium, thereby facilitating the growth of dendrites. Moreover, the cross-sectional SEM image shows that the lithiophobicity of the Ni foam limits the internal deposition of lithium, resulting in a large amount of dendrites remaining on the top [Figure 5D].

Figure 5. Top-view SEM images of Li deposition with 1, 5 and 15 mAh cm^-2 at 1 mA cm^-2: (A-C) NF; (E-G) NiCoO/NF; (I-K) NiCoON/NF. Cross-sectional SEM images of Li deposition with 15 mAh cm^-2: (D) NF; (H) NiCoO/NF; (L) NiCoON/NF. SEM: Scanning electron microscopy.

Compared to the bare NF, Li is preferentially plated on the lithiophilic surface of NiCoO/NF as a result of the efficacious surface area of dense nanorod-on-nanosheet arrays and more Li nucleation sites. With increasing Li from 1 to 5 and 15 mAh cm^-2, the growth of Li dendrites is largely controlled. However, inhomogeneous Li deposition and a small amount of dendrites are still unavoidable, which may be related to the low conductivity of the metal oxide [Figure 5E-G]. The cross-sectional SEM image in Figure 5H is consistent with this discussion. Li can enter the skeleton of NiCoO/NF, while there are still large chunks of Li deposits. Notably, the NiCoON/NF exhibits no visible lithium dendrites under various capacities of 1, 5 and 15 mAh cm^-2 [Figure 5I-K]. After depositing Li at 1 mAh cm^-2, Li uniformly enters the dense skeleton of NiCoON/NF, owing to the sufficient lithiophilic nucleation sites. Under deep plating conditions from 5 to 15 mAh cm^-2, the wide internal space and ample pores of NiCoON/NF are gradually filled due to the cooperative action of enhanced surface lithophilicity and electrical conductivity, resulting in good Li nucleation and deposition. In particular, Li is uniformly deposited inside the 3D porous framework with the original thickness and no obvious lithium dendrites are found [Figure 5L]. These morphological results indicate that NiCoON/NF, as an ideal host, can provide enhanced surface lithiophilicity for inducing uniform nucleation, thereby promoting the formation of dense lithium deposition.

To clarify the underlying mechanism, XRD analysis of Li@NiCoON/NF obtained after depositing 1 mAh cm^-2 of lithium was performed. These peaks have a positive effect on the dense Li deposition due to the form of Li₃N and Li₂O [Supplementary Figure 11]^[40]. XPS was also performed to further confirm the chemical composition change after lithiation in [Supplementary Figure 12]. The three peaks at 55.1, 55.8 and 56.7 eV correspond to Li₂O, Li₃N and LiF, respectively^[41]. In addition, the N 1s spectrum includes three peaks at 399.7, 404.2 and 407.7 eV, attributed to Li₃N, LiNO₂ and LiNO₃, respectively^[42]. This result indicates the formation of a stable SEI film rich in Li₃N and Li₂O. After lithiation, the 3D structure and nanorod-on-nanosheet morphology are maintained, suggesting uniform Li deposition [Supplementary Figure 13]. The corresponding EDS elemental mapping of Li@NiCoON/NF confirms the homogeneous distribution of elemental Ni, Co, O and N, indicating the successful formation of Li₂O and Li₃N. The unique 3D NiCoON/NF nanoarrays combining a large surface area and enhanced lithiophilicity have the ability to strategically weaken the local current density and ensure the successful suppression of Li dendrites. Furthermore, the in-situ formed Li₂O and Li₃N layer with higher Li-ion conductivity further decreases the overpotential and induces Li deposition.

The morphological changes are also evaluated to realize the Li plating/stripping behavior on the NiCoON/NF and bare NF at 1 mA cm^-2 under the areal capacity of 1 mAh cm^-2. After the first electroplating, the surface of NF covers massive scattered dendritic lithium [Figure 6A], while metallic Li can enter the lithiophilic surface of NiCoON/NF [Figure 6D]. After stripping, moss-like Li remains on the NF surface [Figure 6G], while Li can be uniformly removed from NiCoON/NF [Figure 6J]. During the subsequent 50th plating/stripping, a significant amount of thick and loose Li is deposited on the NF [Figure 6B] and needle-like dead lithium is retained upon the 50th stripping [Figure 6H]. In sharp contrast, the NiCoON/NF architecture could spatially direct uniform lithium distribution along the 3D matrix and vertical nanoarrays could be clearly exposed again after Li stripping [Figure 6E and K]. After 100 cycles, the deposited lithium becomes worse [Figure 6C] and a significant quantity of lumped dead lithium accumulates on the bare body of NF after stripping [Figure 6I]. Such poor lithium deposition and the exfoliated dendritic morphology are responsible for the poor CE. Satisfactorily, Li is uniformly deposited on the NiCoON/NF and produces a dense flat Li layer [Figure 6F]. After Li stripping, the NiCoON/NF still maintains a comparatively clean and flat surface [Figure 6L], confirming its outstanding reversibility. These results indicate that NiCoON/NF can effectively hinder the dendrite growth and acquire more compact Li plating/stripping morphologies, which stem from the large specific surface area and promising lithiophilic properties.

Figure 6. Li metal plating (A-C) and stripping morphologies (G-I) of NF at 1st, 50th and 100th cycles, respectively. Li metal plating (D-F) and stripping morphologies (J-L) of NiCoON/NF at 1st, 50th and 100th cycles, respectively.

Electrochemical properties of full cells

To evaluate the practicality of the NiCoON/NF current collector, full cells were assembled and tested by pre-depositing Li onto NF, NiCoO/NF and NiCoON/NF to obtain Li@NF, Li@NiCoO/NF and Li@NiCoON/NF, respectively, as anodes with LiFePO₄ as the cathode. Figure 7A illustrates the long-term cycling properties of full devices at 1 C (160 mA g^-1). Li@NiCoON/NF-LFP achieves improved cycling behavior with a maximum reversible capacity of 119 mAh g^-1 over 400 cycles, corresponding to a superior capacity retention of 85%. In comparison, Li@NiCoO/NF-LFP delivers a relatively reduced discharge capacity of 88.8 mAh g^-1 over 400 cycles with a poor capacity retention of 66.8%. However, the Li@NF-LFP experiences faster capacity degradation over 100 cycles with a capacity retention of only 34.9%, which is mainly introduced by inhomogeneous Li plating and fast SEI depletion. According to the corresponding charge/discharge profiles of the three full cells at the 100th cycle, Li@NiCoON/NF-LFP exhibits the lowest voltage polarization, indicating its good ability to inhibit Li dendrites [Figure 7B]. Li@NiCoON/NF-LFP also exhibits a better rate performance and a high discharge capacity, especially at a high current density of 5 C [Figure 7C]. However, Li@NiCoO/NF-LFP and Li@NF-LFP present obviously lower capacities at the same rate. In addition, the voltage polarization of the Li@NiCoON/NF-LFP cell is less than that of the Li@NiCoO/NF-LFP and Li@NF-LFP cells at 0.2 C and 5 C, indicating the excellent electrochemical kinetics during the discharge and charge process [Figure 7D and E]. These encouraging results can be partially attributed to the superior lithiophilicity of the NiCoON/NF nanoarrays.

Figure 7. (A) Cycling performance of full cells with different Li anodes at 1 C (160 mA g^-1). (B) Comparison of full cells in voltage profiles at 100th cycle. (C) Rate capability. (D, E). Comparison of full cells in voltage hysteresis at 0.2 C and 5 C.

CONCLUSION

In summary, hierarchical Ni- and Co-based oxynitride nanoarrays with superior lithiophilicity have been constructed as stable hosts for uniform lithium deposition. The well-arranged nanoarrays with oxygen vacancies can offer high electronic conductivity and increased surface area to delocalize the practical current density and homogenize the Li-ion flux. Importantly, the in-situ formed Li₃N by the conversion reaction of NiCoON (i.e., NiCoO₂/CoO/Ni₃N) and Li can enhance the electron/ion conductivity for the kinetic induction of the homogeneous deposition of Li. Consequently, a promising Coulombic efficiency of 98.4% up to 600 cycles at 1 mA cm^-2 and an ultralong cycling life of 2000 h with a reduced voltage hysteresis under a high deposition capacity of 3 mAh cm^-2 have been achieved. Furthermore, Li@NiCoON/NF-LFP full cells also exhibit extraordinary electrochemical performance. This work illustrates the potential advantages of Ni- and Co-based oxynitrides in the preparation of stable and dendrite-free lithium anodes to achieve safe and high-energy batteries.