Metal–organic framework-derived LiFePO4/C composites for lithium storage: In situ construction, effective exploitation, and targeted restoration

Abstract

Hitherto, LiFePO₄ (LFP) is bottlenecked by inferior electronic conductivity and sluggish Li⁺ diffusion, which can be resolved by cation doping, morphological engineering, carbon coating, and so forth. Among these methodologies, morphological optimization and carbon modification can warrant a stable operating voltage and prolong the cycling lifespan, which can be accessible by utilizing metal–organic frameworks as self-sacrificing templates. Herein, we conceptualize a strategy to in-situ construct N-doped carbon-coated LFP with Prussian blue analogues as the template, after which electrochemical tests extensively exploit the lithium storage capacity with 153.2 mAh g⁻¹ after 500 cycles at 0.5 C. However, the capacity failure associated with the inevitable Li⁺ loss and destructed carbon layer provides sufficient room for the restoration of LFP after long-term cycling. Motivated by this, the cell performance of LFP/C after targeted restoration using the 3,4-dihydroxybenzonitrile dilithium salt is investigated, revealing a considerable recovered capacity due to the recuperative LFP crystal and uniform carbon layer with homogeneous N-distribution. The computational study also supports the feasibility of N-doped carbon layer in LFP modification. This study envisages a methodology for the performance improvement of LFP from directional fabrication to targeted recovery, providing insights into the manufacturing and reuse of LIB cathodes.

INTRODUCTION

To satisfy the expanding energy demands for renewable energy sources, people are seeking effective and environmentally-friendly energy storage devices, among which lithium-ion batteries (LIBs) have become a major component due to their remarkable rate performance and considerable energy density.^1–3 Olivine-structured LiFePO₄ (LFP) is a potential contender for advanced cathode materials due to its cost-effectiveness, environmental compatibility, high theoretical specific capacity (170 mAh g⁻¹), stable working voltage plateau (~3.45 V), outstanding cycling performance, and thermal safety (strong PO bond).^4,5 However, the intrinsically inferior electronic conductivity (10⁻⁹–10⁻¹⁰ S cm⁻¹) and sluggish Li⁺ diffusion (10⁻¹⁴–10⁻¹⁶ cm² s⁻¹) result in dissatisfactory high-rate performance and hinder its further practical application in electric vehicles (EVs).^6,7 Recently, orchestrated efforts have been devoted to migrating the above deficiencies with different degrees of success. In general, there are three viable strategies to address these problems: (1) optimizing the morphology and size of LFP particles to reduce Li-ion diffusion distance to speed up electrochemical kinetics^8,9; (2) doping with foreign cation to form a solid solution^9–12; and (3) surface modification with conductive carbon matrix. Noticeably, the solid solution normally exhibits more than one operating voltage platform and the instability of large voltage variation will bring about new challenges for commercially practical production. Therefore, the combination of morphology manipulation and carbon coating is an effective and feasible strategy to overcome these obstacles. The carbonaceous LFP were typically synthesized by mechanically mixing the LFP with external carbon source including small molecule organic materials or polymer, followed by an inert atmosphere thermal treatment to transform the organic species into amorphous carbon, which can be a prevalent technique to boost the surface conductivity.^13,14 Compared with general carbon coating strategies, modifying the carbon matrix with doping anionic heteroatom (such as N, F, S, and B) can effectively generate additional defect active sites and thus enlarge the specific surface area and enhance the electronic conductivity.^15,16 It has been claimed that the nitrogen atoms doping to the carbon layers may serve as electron donors, offer electron carriers, and minimize the band gap, thus facilitating superior electron transfer velocity and excellent electrochemical performance.^17,18 However, it is generally difficult to ensure the uniformity and tightness of the carbon layer with external carbon/nitrogen sources.

As a class of three-dimensional topological coordination network materials, metal–organic frameworks (MOFs) consist of metal nodes and organic linkers and are characterized by substantial specific surface area, high porosity, plenty of active sites, and morphology diversities.^19–21 Considering the adjustable morphology and controllable particle size, MOFs have been regarded as promising self-sacrificing templates for directly in-situ construction, which have drawn extensive attention.^22,23 By choosing an appropriate MOF precursor (such as Prussian blue), the N-doped carbon matrix can be directly generated and uniformly coated in the LFP composites to form a carbonaceous LFP material, which is accredited to the Ostwald ripening upon thermal decomposition. Specifically, Ostwald ripening is a structural evolution process originating from the higher to the lower surface energy, during which the unequal surface energy of internal and outside crystallites promotes the dissolution of the inside structure and partial outer layer. During thermolysis in the inert atmosphere, the low surface energy of carbon is conducive to the formation of the exterior carbon shell and the as-synthesized MOF-derived carbonaceous materials commonly retain the primitive morphological features of precursors since the carbon network can be rendered as a morphology protector, warranting a stable structure during annealing and cycling processes.²⁴ However, despite the solid protective layer and intrinsic strong PO bond that can avoid the collapse of construction, capacity degradation is still inevitable due to the irreversible Li⁺ loss during repetitive cycles. Limited by the one-dimension (1D) Li⁺ transport channel, a portion of the extracted Li-ions are unable to return to their initial sites upon high current density, leading to the formation of Li⁺ vacancies (Li_V), which can not only trigger the higher oxidation state of Fe³⁺ but also result in the migration of Fe²⁺ to Li_V, generating the Fe-Li antisite defects that impede the Li⁺ diffusion pathway.²⁵ The capacity-losing mechanism without irreversible morphological and crystallographic damage provides significant room to directly restore the spent LiFePO₄ (S-LFP) to regenerate a new LFP cathode.

Herein, we propose a sintering strategy to in-situ construct the MOF-derived LFP/C cathode materials including PBA-LFP/C (PBA means Prussian blue analog), BDC-LFP/C (BDC means 1,4-benzene dicarboxylate), BTC-LFP/C (BTC means 1,3,5-benzene dicarboxylate), and PBA-LFP (without N-doped carbon layer). We systematically study the impact of different self-sacrificing precursors on the morphological features, particle size, and electrochemical behaviors of these various MOF-derived carbonaceous LFP cathodes. As expected, the PBA-LFP/C outperforms its competitors with superior long-term cyclability, better rate performance, and enhanced Li⁺ diffusion kinetics, which is endowed by the higher specific surface area related to the unique nanocube-shape covered with enormous ultrafine particles, abundant actives site and boosted electronic conductivity granted by the N-doped carbon layer. After a series of electrochemical tests, we directly regenerate the degraded PBA-LFP/C using the multifunctional organic 3,4-dihydroxybenzonitrile dilithium salt (Li₂DHBN) to realize the targeted restoration for the Fe-Li antisites in impaired LFP crystal and the flawed N-doping carbon layer to obtain a new efficient cathode for LIBs. Moreover, the density functional theory (DFT) calculation results confirm the favorable electronic structure, lower intercalation energy, and strengthened Fe-O of heterostructured LFP/CN, contributing to accelerated reaction kinetics and stabilized structure and well supporting the experimental results.

RESULTS AND DISCUSSION Structural and morphological characterization

The in-situ construction processes of MOF-derived LFP/C materials are illustrated in Figure 1A. The precursors of Fe-PBA, Fe-BDC, and Fe-BTC were obtained through coprecipitation, water bath, and solvothermal methods, respectively. Subsequently, the Fe-MOFs were intensively mixed with the Li and P sources by grinding them in a mortar. Based on the thermogravimetric analysis results in Figure S1A, the above mixtures were calcinating at 700°C for 8 h in N₂ to gain the heterostructure PBA-LFP/C, BDC-LFP/C, and BTC-LFP/C. In contrast, the PBA-LFP composite is synthesized by adding an extra step, calcinating in air to form iron oxide without the carbon matrix (Figure S1B). As displayed in Figures S2A–C and S3, the characteristic crystalline diffraction peaks in the X-ray powder diffractometer (XRD) curves demonstrate the successful synthesis of Fe-PBA, Fe-BDC, Fe-BTC, and the PBA-Fe₂O₃. The Fourier transform infrared (FT-IR) characterization was performed to further analyze the elemental configuration and functional groups. The band located at 2077 cm⁻¹ in Figure S2D is accredited to the stretching vibration of Fe-CN-Fe, which further identify the presence of Fe₄[Fe(CN)₆]₃.²⁶ As for Fe-BDC in Figure S2E, the bands observed at 1663, 1595, 1438, and 1392 cm⁻¹ are attributed to the acid CO, asymmetric CO, CC, and symmetric CO stretching vibrations, respectively.²⁷ Multiple bands observed in the regions between 1200–1000 and 900–700 cm⁻¹ were associated with the in-plane and out-of-plane CH aromatic linkers.²⁸ The band around 594 cm⁻¹ could be assigned to the Fe-O stretching vibrations.²⁹ For Fe-BTC in Figure S2F, the Fe-O stretching mode located at around 600 cm⁻¹ and the traits at 1631, 1576, 1447, and 1382 cm⁻¹ are also assigned to the acid CO, asymmetric CO, CC, symmetric CO stretching vibrations, respectively. These results prove the successful synthesis and phase purity of Fe-MOF precursors.

FIGURE 1. Structural and compositional characterization. (A) Scheme illustration of the construction of PBA-LFP/C, BDC-LFP/C, and BTC-LFP/C. (B) XRD curves of PBA-LFP/C, PBA-LFP, BDC-LFP/C, and BTC-LFP/C. (C) FT-IR spectra, (D) crystal structure of LiFePO4, XPS spectra of (E) survey, (F) Fe 2p, (G) O 1s, (H) C 1s, (I) P 2p, and (J) Raman spectra of PBA-LFP/C, BDC-LFP/C, and BTC-LFP/C.

Through XRD measurements and FTIR analysis, the chemical composition and the structural features of the crystal for the as-prepared MOF-derived LFP products were also determined. The main characteristic crystalline diffraction peaks (Figure 1B) of the as-prepared LFP materials are identified with the pure orthorhombic LFP (JCPDS Card No. 81-1173). The Rietveld refinements results are also used to analyze the crystal structure (Figure S4A–C) and Table S1 recorded the detailed refinement data. In accordance with the FT-IR test in Figure 1C, the shoulders at 947 and 969 cm⁻¹ are linked to the symmetric PO stretching vibration in the PO₄³⁻, while the bands at 1066 and 1158 cm⁻¹ are assigned to the asymmetric PO stretching vibrations.³⁰ The crystal model of LFP depicted in Figure 1D demonstrates the olivine structure, where the LiO₆ octahedrons share corners with four FeO₆ octahedrons, along with a parallel tetrahedron PO₄ chain bridging the LiO₆ and FeO₆.³¹ Attributed to the strong PO covalent linkages in PO₄ tetrahedrons, Li-ion can be extracted in a 1D diffusion across the b-axis without altering the crystalline frameworks.³²

XPS spectra were executed to analyze the elemental oxidation state and surface composition of the MOF-derived LFP/C materials. The survey spectrum shown in Figure 1E reveals the coexistence of Li 1s, P 2p, C 1s, O 1s, and Fe 2p located at around 51.2, 132.4, 283.4, 531.1, and 713.8 eV, respectively. Besides, the weak peak observed at around 400 eV in PBA-LFP/C is assigned to the N element derived from the PBA precursors. The N 1 s spectrum shown in Figure S5 is split into three peaks at 400.4, 398.9, and 397.3 eV, according to the graphitic-N, pyrrolic-N, and pyridinic-N, respectively, confirming the successful N-doping. Noticeably, the introduction of nitrogen can generate additional defect active sites, facilitating enlarged specific surface area, and broadened diffusion pathway, resulting in superior electronic conductivity and accelerated Li-ion diffusion kinetics.³³ As shown in Figure 1F, a doublet Fe 2p_3/2, and Fe 2p_1/2 are observed at around 711.9 and 726.4 eV with an energy difference (ΔE_Fe) of 14.5 eV, identifying the pre of Fe²⁺ oxide state in LFP/C materials, while peaks at 714.3 and 729.2 eV represent the satellite peak.⁹ The fitting O 1s in Figure 1G with the binding energies at 534.3, 533.2, 532.9, and 531.2 eV are related to the CO, CO, the lattice oxygen in PO₄³⁻, and the FeO band, respectively.¹⁰ Additionally, the fitting C 1s in Figure 1H is indexed into three peaks, verifying the coexistence of CO band at 284.7 eV, CO at 286.2 eV, and the CC at 284.8 eV.³⁴ The P 2p spectra in Figure 1I exhibit a wide peak positioned at 134.4 eV, correlative with the PO bond of the PO₄³⁻. As shown in Figure S6A, adsorbed isotherms of LFP/C samples bear upward curves in hypo-P/P₀ with hysteresis loops, indicating type-IV isotherms.³⁵ The Brunauer–Emmett–Teller (BET) specific surface areas of PBA-LFP/C, BDC-LFP/C, and BTC-LFP/C materials are estimated to be 95.8, 25.2, 22.3 m² g⁻¹, respectively, and the corresponding pore size distribution is displayed in Figure S6B. It should be noted that micropores and their derived profound specific surface area can offer affluent active sites, broaden diffusion paths, and accelerate ion/electron transfer, facilitating enhanced electronic conductivity and faster Li-ion diffusion.³⁶ As displayed in Figure 1J, the Raman spectrum is carried out to identify structural characteristics and determine the level of carbon graphitization in the LFP/C composites. The bands located at around 970 cm⁻¹ are associated with the stretching modes of the PO of the PO₄³⁻, while the D-band located at around 1340 cm⁻¹ and the G-band at 1570 cm⁻¹ represent the disordered sp³ carbon and the ordered graphitic sp² carbon, respectively. The I_D/I_G ratios of PBA-LFP/C, BDC-LFP/C, and BTC-LFP/C are calculated to be 0.91, 0.89, and 0.79, respectively, indicating a higher graphitization degree, which is favorable for exalted electronic conductivity.³⁷

Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) were utilized to decipher the surface morphology and internal structure of as-fabricated products, respectively. According to SEM images displayed in Figure 2A, the Fe-PBA particles feature regular nano cubic architecture with a smooth outer layer. The particle size investigation for Fe-PBA exhibited in Figure S7 unravels that the distribution of the nanoparticle diameter ranges from 250 to 700 nm, with an average of 403.87 nm. Meanwhile, the rhombic dodecahedron Fe-BDC and microsphere Fe-BTC are also synthesized successfully (Figure 2B,C). After annealing, all the MOF-derived LFP/C undergoes a shrinkage due to the high-temperature sintering in the nitrogen atmosphere.³⁵ The PBA-LFP/C in Figure 2D transforms from a cube into a smooth rounded hexahedron with plenty of ultrafine particles covered on the surface and the mean diameter is reduced to 254.29 nm (Figure S7). However, the PBA-Fe₂O₃ and PBA-LFP collapse into irregular sphere morphology due to the initial calcination in air (Figure S8A,B). The BDC-LFP/C roughly maintained the rhombic shape of the BDC precursor but with all edges shrinking inward to form an irregular rhombic dodecahedron (Figure 2E). The BTC-LFP/C tends to develop into a hollow sphere, maintaining the morphology of the precursor but containing numerous ultrafine particles scattered on the surface (Figure 2F). This phenomenon could be attributed to the Kirkendall effect, which stands for vacancy generation caused by the directional diffusion of atoms in a diffusion couple.³⁸ Besides, the interior structural property of LFP/C materials was further studied with the help of TEM. Figure 2G,H demonstrate that the LFP particles are enclosed in a carbonate framework, with a uniform thickness of around 2 nm, facilitating stabilized structure, abundant electrochemical active sites, enhanced electron conductivity, and accelerated electrochemical reaction kinetics. As observed in Figure 2I, the ringlike selected area electron diffraction pattern shows notable diffraction rings, which belong to the (200), (011), and (301) planes of LFP of the PBA-LFP/C. The high-resolution TEM images further investigate the crystal structure. The marked region and the corresponding line profiles illustrated in Figure 2J–L reveal the significant lattice fringes with the spacing of 0.369 and 0.278 nm, which are attributed to the (011) and (301) planes of LFP, respectively. Meanwhile, Figure S9A–F presents the TEM images of BDC-LFP/C and BTC-LFP/C as well. High-angle annular dark-field (HAADF) image and elemental mappings displayed in Figure 2M and Figure S10 indicate the homogenous distribution of Fe, P, N, O, and C in the hexahedron PBA-LFP/C, illustrating the successful fabrication of LFP crystals and effective N-doping in the sample. According to the above analysis, using different organic ligands can lead to various MOF particles with diverse morphologies and crystallinity. As for Fe-PBA and Fe-BDC, such MOF nanoparticles with high crystallinity tend to form anisotropy materials, which typically develop into non-spherical geometries with distinct crystal planes and surface metal–ligand coordination. In contrast, the Fe-BTC with relatively low crystallinity exhibit isotropic spherical shapes.^39,40 Consequently, the MOF-derived LFP particles inherit the unique morphologies from the MOF precursors with a uniform carbon layer. Notably, the introduction of the N element into the carbon matrix can create more defect-active sites, increase the specific surface area, and widen the diffusion pathway, allowing for enhanced conductivity and expedited electrochemical kinetics.

FIGURE 2. Structural and morphology characterizations. SEM images of (A) precursor Fe-PBA, (B) Fe-BDC, (C) Fe-BTC, (D) PBA-LFP/C, (E) BDC-LFP/C, (F) BTC-LFP/C, (G) TEM images, (H) high-resolution TEM image, (I) SAED pattern, (J–L) lattice stripes, and (M) HAADF image and elemental mapping images of Fe, P, N, and O elements for PBA-LFP/C.

Electrochemical performances

To systematically investigate the lithium storage behavior of the MOF-derived LFP/C materials, a series of electrochemical measurements were performed. The cyclic voltammetry (CV) curves for the MOF-derived LFP materials at the scanning rate of 0.1 mV s⁻¹ among 2.3–4.8 V are displayed in Figure 3A. All of the CV curves possess a comparable pair of oxidation/reduction peaks, which correlate to the reversible redox couple of Fe²⁺/Fe³⁺. As displayed in Figure 3B, PBA-LFP/C exhibits the smallest voltage separation (224 mV) between oxidation and reduction peaks, superior to BDC-LFP/C (254 mV), BTC-LFP/C (278 mV), and PBA-LFP (383 mV). Noticeably, the sharper and more symmetric redox peak profiles reflect the reduced electrode polarization, boosted electronic conductivity, and accelerated redox kinetics. The rate performances in Figure 3C were carried out at a variety of rates between 0.25–10 C to further verify the electrochemical superiority of PBA-LFP/C. As expected, the PBA-LFP/C possess notable rate capacities of 168.3, 162.7, 152.3, 142.7, 130.3, and 114.9 mAh g⁻¹ at the current density of 0.25, 0.5, 2, 3, 4, and 5 C, respectively. More impressively, it still retained a stable capacity of 100.9 mAh g⁻¹ even at a high rate of 10 C, outperforming its counterparts (BDC-LFP/C for 77.8 mAh g⁻¹), (BTC-LFP/C for 48.3 mAh g⁻¹), and (PBA-LFP for 10.83 mAh g⁻¹). As the current density returned to 0.5 C, the specific capacity of PBA-LFP/C recovered to 164.5 mAh g⁻¹, confirming the distinguished rate performance. Besides, Figure 3D recorded the galvanostatic charge–discharge (GCD) curves at various rates, characterizing the typical flat voltage plateaus at around 3.4 V accredited to the pair of Fe²⁺/Fe³⁺ redox couples, which is in accord with the CV profiles. In addition, Figure S11A,B compare the GCD profiles and the calculated potential interval values of the LFP materials at 10 C, among which the potential interval values of PBA-LFP/C upon (de)lithiation process is the smallest and the voltage plateau is the most stable, indicating that the lowest degree of polarization. Furthermore, the interfacial reaction resistance was explored with electrochemical impedance spectroscopy (EIS). As shown in Figure 3E, the Nyquist plots of all samples are composed of a semicircle at the intermediate frequency area linked to the charge-transfer resistance (R_ct) and a sloping line in the low-frequency area associated with the Warburg impedance (W_s) for lithium ions diffusion.⁴¹ Furthermore, the Li⁺ diffusion coefficient (D_Li⁺) is assessed by the following formula:[Image Omitted. See PDF]where the R, T, A, n, F, c, and σ refer to the gas constant (8.314 J K⁻¹ mol⁻¹), temperature, the area of the disk-like electrode, electron number, the Li⁺ molar concentration, and Warburg coefficient, respectively.⁴² According to the equation, the σ value can be estimated by the slope of the Z′-ω^−1/2 linear relationship at the low-frequency area (Figure 3F). According to the equivalent circuit in Figure S11C, the fitting EIS data were recorded in Figure S11D with the aid of the Z-view software. As expected, the EIS data confirmed the lowest R_ct (179.2 Ω), σ value (17.9 Ω S^−1/2), and highest D_Li⁺ (2.6 × 10⁻¹² cm² S⁻¹) of PBA-LFP/C, indicating enhanced charge transmission and superior Li⁺ diffusion kinetics which could be benefitted from the compositional optimization and structural modification. In contrast, the PBA-LFP shows unsatisfactory results with large R_ct (721.1 Ω), σ value (101.7 Ω S^−1/2), and lowest D_Li⁺ (8.9 × 10⁻¹⁴ cm² S⁻¹), suffering from the inferior electronic conductivity and the crowed Li⁺ diffusion aisles caused by the collapsed sphere-shape agglomerates. Besides, the long-term cycling examination was performed to investigate the electrochemical stability. According to Figure 3G, the PBA-LFP/C delivered a reversible capacity and retained 153.2 mAh g⁻¹ after 500 rounds at 0.5 C, outperforming BDC-LFP/C (76.3 mAh g⁻¹), BTC-LFP/C (86.6 mAh g⁻¹ for the 335th cycle), and PBA-LFP (65.1 mAh g⁻¹). Meanwhile, the comparison of cycling performance and the corresponding synthesis route and material morphology with recently reported LFP materials is displayed in Figure S12 and Table S2, further confirming the effective improvement of PBA-LFP/C.

FIGURE 3. (A) CV curves at the scanning rate of 0.1 mV s−1, (B) enlarged CV pattern, (C) rate performance at different rates in the range of 0.25–10 C of PBA-LFP/C, BDC-LFP/C, BTC-LFP/C, and PBA-LFP. (D) Galvanostatic charge/discharge profiles at various current densities of PBA-LFP/C. (E) Nyquist impedance spectra, (F) the relationship of Z′ and ω−1/2, (G) long-term cycling performance at 0.5 C for 500 cycles, and (H) galvanostatic charge/discharge profiles at long-term cycling performance at 0.5 C for the 200th cycle for LFP materials.

The GCD curves recorded in Figure S13 of PBA-LFP/C show no obvious variation at different cycles, maintaining a small polarization voltage of 65.2 mV, which demonstrates excellent reversibility. The corresponding GCD profiles and the enlarged pattern at long-term cycling performance for 200th are also displayed in Figures 3H and S14A. And Figure S14B recorded the calculated voltage interval values of as-prepared products, revealing the smallest potential interval of PBA-LFP/C, signifying the lowest degree of polarization and excellent reversibility during the cycling process, due to the effective electron transport and superior conductivity granted by the ultrafine particles distributed on the surface of the cubic architecture and the smooth and homogeneous Li-ion insertion/extraction guaranteed by the ultrathin amorphous carbon shells.⁵ In contrast, the PBA-LFP exhibits inferior cycling performance, which should be associated with unfavorable particle growth and accumulation, resulting in poor electronic conductivity and slow diffusion kinetics.³⁵

To further investigate the electrochemical behaviors, the CV profiles at various scan rates upon 0.2–1.0 mV s⁻¹ were conducted to analyze the pseudocapacitive effect and reaction kinetics for LFP products. Figure 4A and Figure S15A,B reveal that the current intensity rises as the scanning rates increase. Herein, the pseudocapacitive contribution can be estimated by the following formulas:[Image Omitted. See PDF] [Image Omitted. See PDF]where i and v stand for the peak current and scanning rate, respectively, while a and b are variable parameters.⁴³ The b value can be quantified by the linear relationship between log(i) and log(v). When the obtained b value is near 1, the electrochemical process tends to be pseudocapacitive-controlled, while the b approaching 0.5 illustrates a diffusion-controlled reaction. In Figure 4B, the b valves of PBA-LFP/C are calculated to be 0.66 for the oxidation peak and 0.82 for the reduction peak, revealing a combination of both the diffusion process and the pseudocapacitive behavior. By contrast, the b valves of BDC-LFP/C (Figure S15C) are quantified to be 0.67 and 0.70, while the b values of BTC-LFP/C (Figure S15D) are 0.49 and 0.48. The b valves of PBA-LFP/C are significantly higher than those of its counterparts, demonstrating a higher pseudocapacitive contribution. More precisely, the ratio of two different processes at different scan rates can be estimated by the formula (3), in which the k₁v and k₂v represent the pseudocapacitive and diffusion behavior, respectively.⁴⁴ The calculated contribution of pseudocapacitive behavior accounts for 73.8% of PBA-LFP/C (Figure 4C), and the proportion continuously rises with the increasing scan rate (Figure 4D). The ratio of BDC-LFP/C and BTC-LFP/C are also calculated and recorded in Figures S15E,F and S16. As expected, the pseudocapacitive contribution of PBA-LFP/C is the highest at all scan rates, due to the boosted electronic conductivity granted by N-doped carbon network and abundant active sites offered by large surface area, facilitating faster reaction kinetics and superior rate performance. It should be noted that the profound pseudocapacitive ratio for the intercalation-type LIBs is associated with the surface-induced capacitive process, such as defect active sites introducing strategy and surface modification method.⁴⁵

FIGURE 4. Quantitative kinetics analysis for PBA-LFP/C cathode. (A) the CV curves at various scanning speeds during 0.2–1.0 mV s−1, (B) the fitting b values for redox peaks, (C) the pseudocapacitive-dominated contribution at 1.0 mV s−1, and (D) the pseudocapacitive-dominated contribution at 0.2, 0.4, 0.6, 0.8, and 1.0 mV s−1. (E) The GITT curves, the calculated Li+ diffusion coefficient during (F) delithiation, and (G) lithiation process. (H) The GCD curve and the ex-situ XRD curves upon the corresponding (de)lithiation state. (I) The illustration of the (de)lithiation process.

To systematically investigate the superiority of the PBA-LFP/C, the galvanostatic intermittent titration technique (GITT) was implemented to find the diffusion kinetics of the Li-ion. The titration process of PBA-LFP/C and BTC-LFP/C for the second cycle was displayed in Figure S17A,B, where the voltage variations in single relaxation steps reflect the smaller polarization degree of PBA-LFP/C.⁴⁶ Furthermore, Figure 4E recorded the GITT curves, in which PBA-LFP/C displays the shortest relaxation spikes, suggesting accelerated reaction kinetics accompanied by lower polarization and quicker equilibration.⁴⁷ To quantificationally determine the diffusion kinetics of Li⁺, the diffusion coefficient (D_Li⁺) can be determined by the following relation:[Image Omitted. See PDF]where τ stands for constant current pulse duration of the GITT measurement, M_B, V_M, m_B, and A represent the molecular weight, molar volume, electroactive mass, and surface area of the electrode, respectively.⁴⁸ While the ΔE_S and ΔE_τ mean the change of steady-state voltage and a cell voltage variation at the corresponding (de)lithiation step, which can be acquired from the single-step GITT titration in Figure S18A,B. The calculated D_Li⁺ values of all samples upon delithiation and lithiation are displayed in Figure 4F,G, suggesting the highest diffusion coefficient of PBA-LFP/C, indicating that the carbon matrix modified with N-doping can facilitate smoother Li⁺ insertion/extraction, easier redox reaction, and superior reversibility. The ex-situ XRD curves upon the corresponding (de)lithiation state for PBA-LFP/C were conducted to excavate the transformation mechanism. As demonstrated in Figure 4H, the PBA-LFP/C follows the traditional nucleation process with both well-resolved Li-rich and Li-poor phases, belonging to the triphylite and heterosite phases, respectively, which is illustrated in Figure 4I.^49–51 During the whole (de)lithiation process, the XRD patterns show high crystallinity without obvious diffraction angle shifts, and then the XRD curve recovers back to the pure single phase of LFP at the completely lithiated state, revealing the remarkable reversibility and structural stability, which helps to explain the excellent cycling performance.

Although the stable structure of LFP ensures that the construction will not collapse during the cycling process, the capacity still experiences fluctuation and attenuation, which is suffered from the 1D Li⁺ transport channel during cycling.⁵² Partial extracted Li⁺ fails to return to their original sites along with the b axis at a high current, resulting in a considerable number of Li⁺ vacancies. In this situation, Fe ions partially transfer and fill the Li⁺ vacancies, forming Fe-Li antisite (Fe_Li) defects that prevent Li⁺ from migrating across the channel.⁵³ Apart from these, the aggregation and irrecoverable electrolyte erosion of LFP particles would also result in electrochemical instability. To restore the structural degradation, we directly regenerate the spent PBA-LFP/C cathode (Figure 5A) utilizing a multifunctional organic lithium salt Li₂DHBN (its synthetic process is shown in Figure S19A). The XRD curves of 3,4-dihydroxybenzonitrile and Li₂DHBN are shown in Figure S19B,C help to illustrate the structure. Furthermore, the corresponding FT-IR analysis is displayed in Figure S19D, in which Li₂DHBN shows a characteristic band at ~500 cm⁻¹, belonging to Li-O vibration, confirming successful synthesis.⁵⁴ Firstly, the inductively coupled plasma-optical emission spectrometry (ICP-OES) results displayed in Figure S20 reveal the lithium deficiency of the spent PBA-LFP/C cathode (S-LFP), when compared with the primary PBA-LFP/C cathode and regenerated LFP cathode (R-LFP) (more details can be found in Table S3). Noticeably, the lithium vacancy can not only result in the capacity decline but also trigger the higher oxidation state of Fe³⁺. The XRD curves in Figure 5B certainly display a clear phase of FePO_4, confirming the formation of Fe³⁺. Besides, the XPS survey in Figure S21 and the fitting XPS spectrum of Fe 2p in Figure 5D also verify the existence of Fe³⁺ and the fitting area ratio of Fe 2p_3/2 (Table S4) is applied to quantify the ratio of Fe³⁺/Fe²⁺, which is calculated to be 2.17, suggesting the predominant phase of FePO₄.⁵⁵ After regeneration, the FePO₄ phase disappears according to the Rietveld refinement result of R-LFP in Figure 5C, and the Fe³⁺ peak is also invisible in R-LFP in Figure 5D, suggesting successful restoration for both composition and structure. Furthermore, the distribution of nitrogen and carbon for S-LFP in Figure 5E looks uneven, existing C and N vacant areas, which could be corroded by the electrolyte and suffered from some irrecoverable mechanical damage during the long-term cycling process. In contrast, the C and N mapping patterns of R-LFP in Figure 5F show excellent homogeneity, indicating uniform carbon coating modified with N-doping, revealing the carbon layer-recoating ability of Li₂DHBN. Therefore, utilizing the Li₂DHBN to directly restore the S-LFP is a promising strategy. The distorted PBA-LFP/C can effectively combine with the functional groups so that the ferric phase can be prevented and lithium vacancies can be filled by Li ions under the reductive atmosphere provided by the cyano group. Meanwhile, the pyrolysis of Li₂DHBN can purposefully heal the conductive carbon layer and supplementary nitrogen, facilitating accelerated Li-ion diffusion and boosting electronic conductivity.⁵³ In contrast, the utilization of Li₂CO₃ can merely repair the losing Li without restoring the N-doped carbon matrix.

FIGURE 5. (A) The regeneration process of PBA-LFP/C cathode. (B) The XRD curves of the S-LFP and the R-LFP, and (C) the Rietveld refinement result of R-LFP. (D) Fe 2p XPS spectrum, of S-LFP and R-LFP. HAADF image and elemental mapping images of Fe, N, C for (E) S-LFP and (F) R-LFP. (G) The enlarged CV pattern at 0.1 mV s−1, (H) the cycling performance at 0.5 C for 40 cycles, and (I) Nyquist impedance spectra for S- and R-LFP.

To further investigate the lithium storage behaviors of S-LFP and R-LFP electrodes, a series of tests were conducted. Figures S22A and 5G recorded CV profiles at 0.1 mV s⁻¹ and the enlarged patterns, respectively. The potential gap between the Fe³⁺/Fe²⁺ peaks is calculated to be 378 and 275 mV for S-LFP and R-LFP, respectively, revealing higher reversibility of the redox process of R-LFP, which suggests effective regeneration. Furthermore, the cycling performances of the above electrode (Figure 5H) were compared in the potential range of 2.3–4.8 V at 0.5 C, in which the R-LFP exhibits a considerable capacity of 139.7 mAh g⁻¹ at 0.5 C after 40 loops, surpassing the S-LFP (71.6 mAh g⁻¹). Synchronously, the polarization voltage of S-LFP (191 mV) is much higher than R-LFP (99 mV) at the 40 cycles (Figure S22B), indicating a significant improvement in the reaction kinetics of R-LFP. Furthermore, the EIS measurements in Figure 5I and S22C were also used to evaluate the reaction resistance (the equivalent circuit is displayed in Figure S11C. and more details can be found in Figure S23). Compared with the S-LFP, the R-LFP processes smaller R_ct value, Warburg impedance, and higher D_Li⁺, proving the superior electronic conductivity and accelerated Li-ion diffusivity. All the results of characterization and electrochemical tests indicate that directly regenerating the degraded PBA-LFP/C by using the multifunctional organic lithium salt (Li₂DHBN) is an efficient and feasible approach to realize the targeted restoration for Fe_Li antisites and flawed N-doped carbon layer, which provides an opportunity for the spent LFP materials.

Theoretical study

DFT study was performed to unveil the heterostructured interaction and associated effect on enhanced electrochemical performance of the N-doped PBA-LFP/C. According to XPS results, a N-doped carbon network with the atomic ratio (C:N = 9:1) was created. In general, electronic properties and charge distribution are critical factors in electrochemical dynamics.⁵⁶ Therefore, the density of states (DOS) coupled with Fermi levels (E_f) were employed to examine the electronic properties of LFP/CN and bare LFP. As delivered in Figure 6A,B, the LFP presents a band gap of 3.48 eV, showing a characteristic of the semiconductor, while the Fermi levels of LFP/CN across the conduction bands and imply a characteristic of the metal. The difference suggests that the LFP encapsulated by the N-doped carbon matrix contains more available active electrons, facilitating enhanced electronic conductivity.⁵⁷ Besides, the 3D differential charge density distribution displayed in Figure 6C further confirms the optimized electronic structure. Noticeably, the N-doped carbon matrix is surrounded by valence electrons, which strengthens the connection between the carbon layer and the LFP crystal. The LFP/CN and LFP models are also used to investigate the Bader charge to further study the synergistic effect of the N-dope carbon coating on electronic variation and charge distribution. As displayed in Figure 6D, a charge redistribution exists in the LFP/CN with less electronegative O, more electropositive Fe, and slightly more electropositive Li, indicating a weakened attraction between the closed anions (O²⁻) and diffused Li-ions, which may facilitate accelerated Li migration.^58,59 In addition, the average bond length of Fe-O for LFP/CN (2.16 Å) and LFP (2.19 Å) are recorded in Figure S26A. The shorter bond length in LFP/CN represents strengthened Fe-O bonds, which can effectively prevent Fe migration and the generation of the antisite defects during (de)lithiation.⁵⁵

FIGURE 6. Density functional theory. Total and partial density of states (DOS) plots for (A) LFP and (B) LFP/CN. (C) Differential charge density distribution of the LFP/CN heterostructure. (D) Bader charge distribution of bare LFP and LFP/CN. (E) The illustration of intercalation processes for LFP and LFP/CN. (F) Schematic illustration of the PBA-LFP coated by N-doped carbon layer.

To further explore the effect of N-doped carbon layer on the reaction kinetics and intercalation process, the mono-vacancy models of LFP/CN and LFP were designed by eliminating a single Li atom to evaluate the Li-ion intercalation energy in Figure 6E (the top view and side view are displayed in Figure S24 and Figure S25), which can be obtained by following equations^60–62:[Image Omitted. See PDF] [Image Omitted. See PDF]where the ΔE_LFP and ΔE_LFP/CN stand for the energy change upon Li ion intercalation process in LFP and LFP/CN, respectively; while the E_tot represents the total energy of whole system and the E_tot (Li) symbolizes the electrode potential of fcc metallic lithium (−1.997 eV). On the basis of the optimized E_tot valves (displayed in Figure S26B,C), the intercalation energy of LFP and LFP/CN (Figure S26D) can be calculated to be −4.602 and −5.191 eV, respectively. It is noteworthy that the lower intercalation energy of LFP/CN can not only demonstrate the easier occurrence of Li intercalation during electrochemical process, it also implies the smoother Li-ion compensation during restoring procedure.

From the experimental side, the PBA-LFP/C possesses excellent electrochemical properties with considerable cycling stability (153.2 mAh g⁻¹ after 500 cycles at 0.5 C) and remarkable rate performance (100.9 mAh g⁻¹ at 10 C). In contrast, the absence of a carbon layer, the dearth of active sites, and the large particle size caused by the annealing procedure in air or the inferior monomer selection led to structural deterioration, miserable electronic conductivity, and particle aggregation during cycling process. Meanwhile, the DFT theoretical studies well support the experimental results, verifying the boosted electronic conductivity and strengthened crystal stability during de(lithiation) process. Therefore, combining the theory with experimental results, the superiority of PBA-LFP/C heterostructure could be summarized to advantages shown in Figure 6F and can be interpreted as follows: (1) N-doping into carbon network can provide extra defect active sites, allowing for enlarged specific surface area and broaden diffusion pathway. (2) The N-doped carbon layer can rend as morphology protector, avoiding the direct contact and adverse reactions between LFP and electrolyte, facilitating stabilized structure. (3) The Fe-O bonding can be strengthened after establishing a heterogeneous interface between the LFP and N-doped carbon matrix, which can effectively restrain the migration of Fe ions and prevent the formation of Fe-Li antisite defects during (de)lithiation process. (4) Due to the systematic effect, the lower intercalation energy and the higher Li⁺ diffusion coefficient of PBA-LFP/C demonstrate accelerated extraction/insertion kinetics and faster Li-ion diffusion.

CONCLUSION

In conclusion, we elaborately design an in-situ constructing strategy for a series of MOF-derived LFP cathode materials, among which the PBA-LFP/C heterostructures encapsulated in an ultrathin N-dope carbon shell exhibit impressive cycling performance (153.2 mAh g⁻¹ after 500 cycles at 0.5 C) and remarkable rate capability (100.9 mAh g⁻¹ at 10 C), overcoming the intrinsic subpar electronic conductivity and sluggish Li⁺ transportation and thereby successfully exploiting the advanced lithium storage capacity. Considering the structural stability ensured by the two-phase transition mechanism, we directly regenerated the degraded PBA-LFP/C using Li₂DHBN which can provide a reductive atmosphere to prevent the formation of Fe³⁺, force Li⁺ to fill vacancies, and effectively reclaim the impaired N-doped carbon matrix, endowing the regenerated LFP with recovered cycling performance. Meanwhile, the DFT calculation further substantiates the ameliorated electronic conductivity and exalted lithium storage activity imparted by the N-doped carbon matrix. With the aid of experimental and computational studies, this research envisions a conceptual strategy for directional synthesis, modification, and reuse of the LFP cathode, which provides valuable guidelines for LIB cathode production and recycle and well satisfies the “3R” (reduce, reuse, and recycle) principle.

AUTHOR CONTRIBUTIONS

Yilin Li, Ziqiang Fan, and Zhijian Peng contributed equally. Yilin Li, Zhijian Peng and Xinyu Zhang conceived the project and designed the experiments. Ziqiang Fan contributed to the DFT calculations. Yilin Li, Jian-En Zhou and Zhenyu Wu performed the electrochemical studies and characterizations. Zhaohui Xu, Xiaoming Lin, Enyue Zhao, and Ronghua Zeng supervised the project. Yilin Li, Ziqiang Fan, and Zhijian Peng co-write the manuscript. All of the authors discussed the results and commented on the manuscript.

ACKNOWLEDGMENTS

We gratefully acknowledge the financial support from the National Natural Science Foundation of China (no. 21875076), the Science and Technology Planning Project of Guangzhou City (2023B03J1278), Moreover, the authors would like to thank the Shiyanjia lab (https://www.shiyanjia.com) for the XPS test.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflict of interest.