Enhancement of Li intercalation kinetics of LiFePO₄ nanoparticles with mesoporous carbon

Preparation of MC

MC material was prepared via a fluidized bed reactor with a diameter of 100 mm using CO₂ activation methods. Commercial activated carbon (YP50-F, Kuraray Co., Ltd., Japan) was used as the raw material. The volume flow rate of CO₂ was 7 L min^-1 under 1 atm and the reaction temperature was set to 930 °C. After reacting for 9 h, the pore structure of pristine activated carbon was etched adequately by CO₂ to prepare MC material with abundant mesopores.

Electrode preparation and cell assembly

Nano-sized LFP with carbon-coating was purchased from Pulead Technology Industry Co., Ltd., China, without further purification treatment. LFP and MC were firstly mixed by grinding sufficiently to prepare the composite cathodes (abbreviated as LMC, and LMC@3_1 meant the mass ratio of LFP to MC was 3:1). Then, the active material (LFP or LMC), conductive black (Super P), multi-walled CNTs, and polyvinylidene fluoride (PVDF) were mixed in N-methyl-2-pyrrolidone (NMP) with a mass ratio of 85:10:1:4. After adequately stirring in a homogenizer, the slurry was coated onto three-dimensional Al foam current collector instead of conventional two-dimensional Al foil, which had been proved to be much more conducive to improve the electrochemical performance of batteries and supercapacitors in previous work^[24,25]. The electrodes were dried at 70 °C for 2 h to remove the NMP solvent and then rolled to 150 μm. For assembling the coin cell, all electrodes were punched into 13 mm disks and further dried at 60 °C overnight in a vacuum oven. The average mass loading of active materials for pure LFP and LMC@3_1 electrodes was 6.94 and 5.56 mg cm^-2, where the difference mainly resulted from the distinct tap density between LFP and MC materials. Then, the compaction density of the two electrodes could be calculated by their mass loading and thickness, which were 0.46 and 0.37 g cm^-3.

Coin cells (CR2032) were then assembled in the glove box to investigate the electrochemical performance of pure LFP and LMC@3_1 cathodes, where both battery materials (LFP) and capacitive materials (MC) were considered as active materials. Metallic lithium disks (diameter = 16 mm, thickness = 0.6 mm) were used as the anode to assemble the coin cell, and the two electrodes were separated by a polypropylene membrane (Celgard 2400). The space inside the cell was filled with 200 μL organic electrolyte containing 1 mol L^-1 LiPF₆, which was purchased from Zhangjiagang Guotai Huarong New Chemical Materials Co., Ltd, China (LB-4927BY).

Electrochemical tests

Galvanostatic discharging tests were performed by a battery testing system (LAND CT2001A, China) at 25 °C. A voltage range of 2.4-4.2 V was set for all electrochemical measurements with the cells being charged at 1C (1C = 0.17 A g^-1) and discharged at different rates from 1C to 50C. Galvanostatic charging tests where the cells were charged at different rates from 1C to 50C and discharged at 1C were also performed to obtain the complete Tafel plot. Cyclic tests were carried out by 1C charge and 10C discharge for 1,000 cycles. It should be noted that both current density and specific capacity were based on the total mass of active materials (including LFP and MC) in the cathodes. Energy (E, Wh kg^-1) and power densities (P, W kg^-1) were calculated using:

Here, I was the discharge current (A), U was the cell voltage (V), t was the discharge time (s), and m (kg) was the total active materials mass in the cathode.

Cyclic voltammetry (CV) and EIS tests were performed by an electrochemical workstation (CHI660E). The scan rates of CV measurement were 1.0 mV s^-1 between 2.4 to 4.2 V. EIS measurement was conducted at different states of charge (SOC) with a frequency range from 100 kHz to 10 mHz and an applied amplitude of 5 mV. An equivalent circuit model was applied to analyze the EIS results and the parameters were calculated by ZView3.2 software.

Characterization

Crystal structure of LFP was studied by XRD pattern from 10° to 60° 2θ with a scanning rate of 2° min^-1 using Cu Kα radiation (λ = 0.15406 nm). The numbers of standard XRD patterns used for comparison were PDF#01-078-7911 for LFP and PDF#01-082-9760 for FePO₄. Morphology of active materials and electrodes was characterized using a scanning electron microscope (SEM, JEOL JSM-7401) and HRTEM (JEOL JEM-2010), and the lattice fringe spacings of LFP particles in HRTEM images were measured by calculating the average spacing among ten lattice fringes. Element analysis of composite cathode was completed by energy dispersive spectrometer (EDS, Oxford Instruments) mapping. Argon sorption isotherms were measured using Autosorb-iQ2-C at 87 K to analyze the pore structure of LFP and MC. In addition, quenched solid density functional theory (QSDFT) method was applied to the study of pore size distribution for MC, in which a slit pore shape was assumed. Particle size distribution of MC powders was measured by a Mastersizer 3000 (Malvern, Britain). Electrical conductivity of active material powders was measured by a ST2722 powder resistivity tester (Suzhou Jingge Electronic Co., Ltd., China) under the pressure of 4 MPa, while the electrical conductivity of electrode was measured by a 4-Point Probes Resistivity Measurement System (RTS-9, FOUR PROBES TECH Co., Ltd., China). Electrolyte uptake tests were conducted by immersing the electrode into electrolyte solution, followed by a complete rest until no obvious liquid existed on the electrode surface. The electrolyte uptake was then calculated based on the mass of active materials in the electrode and the mass difference of the electrode before and after immersion.

For the in-situ XRD experiments, an in-situ electrochemical cell with a high-pure beryllium window was assembled and then charged and discharged at 0.5C and 2C while simultaneously collecting XRD patterns. The 2θ range was set for 24°-38° with 0.02°/step and 0.4 s/step to shorten the collection time per curve and acquire as much original data as possible. Whole pattern fitting (WPF) and Rietveld refinement (RF) were completed using MDI (Materials Data, Inc.) Jade software. The background of the XRD curve was fitted with polynomials (n = 2-5), and the profile shape function (PSF) for all phases was the Pearson-VII function. All parameters that could be refined in the atomic control windows were selected for both LFP and FePO₄ phases. The refinement iteration process continued until the values of the agreement R-factor (7%-9%) and R/E (1.01-1.17) were minimal.

Establishment of local Li⁺-rich electrolyte region around MC during discharge

In this work, nano-sized LFP and micron-sized MC were used as the battery materials and capacitive materials to fabricate the composite cathode of LIBCs with an Al foam-based current collector. Supplementary Figure 1 showed that the LFP spherical particles (monocrystalline with Pnma space group) exhibited a uniform size distribution with a mean diameter of around 100 nm. Supplementary Figure 2 showed the particle size of MC was 1-10 μm with a specific surface area of 2,415 m² g^-1, and the pore volume and mesoporous ratio were 1.209 cm³ g^-1 and 20.4% according to the QSDFT fitting result. Owing to the large size of MC materials, the smaller LFP particles surrounded and contacted MC closely in the LMC electrode [Figure 1A] where LFP and MC both acted as active materials. Such structure prompted the establishment of a local Li⁺-rich electrolyte region around MC in LMC composite cathode, which allowed the fast Li-ion migration between LFP and electrolyte during discharge. EDS mapping revealed the very uniform distribution of element Al, proving that Al wires existed everywhere and connected each other to greatly reduce the contact resistance between active materials and current collectors owing to the application of 3D Al foam^[24,25]. In addition, a significant difference could be seen in the MC particle that element C was present obviously while elements Fe, P, and O were not, proving that LMC was a composite containing LFP and MC and no element was doped on MC. To determine the best composition of LMC, rate performance of composite cathodes with different MC contents was first compared [Supplementary Figure 3A], indicating that the highest specific capacity at 20C-50C could be achieved when the mass ratio of LFP to MC was 3:1 in the composite cathode (i.e., LMC@3_1), which was used to compare with pure LFP cathode in subsequent study.

Figure 1. (A) SEM image and EDS mapping results (elements Al, Fe, P, C, O) of LMC@3_1 electrode with LFP and MC as active materials. (B) Schematic illustration of the establishment of local Li⁺-rich region around MC, contributing to the fast Li-intercalation kinetics of LFP nanoparticles. (C) Voltage curves of LMC@3_1 cathode under 1C charge/discharge, the corresponding capacitive behavior ($$ \mathrm{PF}_{6}^{-} / \mathrm{Li}^{+} $$ adsorption/desorption) and Faraday behavior (Li⁺ extraction/intercalation) in different regions were marked. (D) Specific capacity of MC and the Li⁺-rich concentration range by MC addition at various current densities.

As follows, we first analyzed the energy storage mechanism of MC in LIBCs. Shellikeri et al. had reported the different mechanisms of ion adsorption in the cathode (only activated carbon as active materials) of Li-ion capacitors (LICs) depending on the open circuit potential (OCP)^[26]. This may lead to the initial coulomb efficiency (ICE) higher than 100% owing to the capacity contribution arising from the adsorption of Li-ions in MC in the later stage of the first discharge [Supplementary Figure 3B]. In addition, Guo et al. reported the “recharging phenomenon” in LIBCs, i.e., an inner current flowing from battery materials to capacitive materials could be formed during the rest period after discharge due to their difference in polarization resistance^[22]. Stepien et al. proved the adsorption of $$\mathrm{PF}_{6}^{-}$$ in porous carbon was dominant with respect to Li⁺; i.e., the capacitance of adsorbing $$\mathrm{PF}_{6}^{-}$$ was larger than that of adsorbing Li^+[15]. Based on these mechanisms, we proposed that MC would also be recharged by LFP upon the rest period after the first discharge, and thus, some adsorptive Li-ions in MC would be desorbed, followed by the adsorption of $$\mathrm{PF}_{6}^{-}$$ in MC during subsequent charge. It may be hard to affirm if all adsorptive Li-ions were desorbed during such recharging processes. However, it was definite that the adsorption and desorption of $$\mathrm{PF}_{6}^{-}$$ could occur both in the whole range higher than OCP and in some range lower than OCP, contributing to the capacitive effect of MC in LIBCs.

Considering the adsorption and desorption of $$\mathrm{PF}_{6}^{-}$$ in MC, Figure 1B illustrated the establishment of a local Li⁺-rich region around MC and Figure 1C showed the corresponding voltage curves of LMC@3_1 cathode under 1C charge/discharge. Upon charge, $$\mathrm{PF}_{6}^{-}$$ anions were adsorbed rapidly on the pore surface of MC due to the EDLC effect (two red oblique lines in Figure 1C), and Li-ions would also be extracted from LFP crystals when the voltage reached a specific value (blue horizontal line in Figure 1C), thus generating amounts of free Li-ions in the electrolyte around MC. Meanwhile, Li-ions in the bulk electrolyte phase would migrate from the cathode to the anode side under the electric field effect. On account of the slow Li-ion diffusion rate in electrolytes, a local Li⁺-rich electrolyte region formed around the MC particles. When discharging, partial desorption of $$\mathrm{PF}_{6}^{-}$$ anions would happen firstly, corresponding to the yellow oblique line I in Figure 1C. As the voltage decreased to a specific value (defined as “inflexion point” meaning the shift from capacitive behavior to Faraday behavior during discharge), Li-ions were intercalated to FePO₄ crystals to generate LFP (green horizontal line in Figure 1C). It was worth noting that some $$\mathrm{PF}_{6}^{-}$$ anions did not undergo desorption until the Li-intercalation of FePO₄ (i.e., the redox reaction inside LFP particles) completed, which corresponded to the yellow oblique line II in Figure 1C^[27-29]. Therefore, the local Li⁺-rich electrolyte region around MC was still present to enhance the Li-intercalation kinetics of LFP during discharge. Based on two extreme cases (one was adsorption/desorption of $$\mathrm{PF}_{6}^{-}$$ occurred in the entire range and another was it only occurred in that higher than OCP), we derived the Li⁺-rich concentration range by MC addition according to the specific capacity of pure MC cathode at various current densities (Figure 1D, Supplementary Figure 4, Supplementary Table 1, detailed calculation process in Supplementary Calculation 1). It could be seen that the Li⁺-rich concentration showed the same trend as the specific capacity, i.e., decreasing with the increase of current density. The Li⁺-rich concentration in LMC@3_1 cathode could reach 0.139 mol L^-1 at a low current density of 0.17 A g^-1 owing to the EDLC effect of MC, which was comparable to the original Li-ion concentration in bulk electrolyte (1 mol L^-1). Even at a current density as high as 8.50 A g^-1, the Li⁺-rich concentration could remain 0.049 mol L^-1, proving the superiority of adding MC for increasing the concentration gradient of Li-ions. In summary, the addition of sufficient amounts of MC in the LFP electrode did render a local Li⁺-rich electrolyte region to improve the Li-ion diffusion performance from electrolyte to LFP and then accelerate the electrochemical reaction.

Quantitative analysis of the synergistic effect between LFP and MC and the Li intercalation kinetics of single LFP particle

Figure 2A showed the CV results of pure LFP and pure MC cathodes at a scan rate of 1.0 mV s^-1, which exhibited a couple of typica l redox peaks for LFP and an approximate rectangle shape for MC. In addition, the current density in pure LFP was far larger than that in pure MC on account of the lower capacity of MC compared with LFP. To clearly observe the synergistic effect between battery materials and capacitive materials, a CV curve derived from the linear weighted sum of pure LFP and pure MC cathodes according to their mass ratio (3:1) was compared with that of the LMC@3_1 cathode. It could be seen from Figure 2B that the intensity of redox peaks in LMC@3_1 was much stronger than that of the linear weighted sum of pure LFP and MC, proving the faster and more complete redox reaction occurred inside LFP particles in the presence of MC, i.e., a favorable synergistic effect existed between LFP and MC materials. Besides, the CV curve of LMC@3_1 also showed the approximate rectangle shape in the low (2.4-2.8 V) and the high voltage regions (3.9-4.2 V)^[29,30], demonstrating that the capacitive process (adsorption/desorption of $$\mathrm{PF}_{6}^{-}$$ anions) happened both before and after the redox process of LFP. This result confirmed the viewpoint in Figure 1B that only a part of $$\mathrm{PF}_{6}^{-}$$ anions had desorbed from the pore surface of MC when Li-intercalation reaction proceeded. On the other hand, a smaller potential difference between oxidation and reduction peaks could be observed in the LMC@3_1 cathode as well, indicating the slighter electrode polarization and the better reversibility of LFP particles with the enhancement effect of MC, which was attributed to the higher electrical conductivity [Supplementary Figure 5], the better Li-ion diffusion performance and the faster electrochemical reaction rate that would be further verified in subsequent EIS study.

Figure 2. (A) CV results of pure LFP and pure MC cathodes at scan rate of 1.0 mV s^-1. (B) CV results of LMC@3_1 cathode at scan rate of 1.0 mV s^-1 and the comparison with that of linear weighted sum of pure LFP and pure MC cathodes. (C) Simplified model of single LFP spherical particle considering one-dimensional diffusion channel of Li-ion inside LFP, the blue ring and yellow circle meant the reacted part and unreacted part. (D) Capacity of LFP particles in pure LFP (C_LFP) and in LMC@3_1 (C_e,LFP) cathodes, and the capacity ratio of LMC@3_1 comparing with that of pure LFP (CR = 1) under various discharge rates. (E) Reaction depth of LFP particles in pure LFP and in LMC@3_1 cathodes under various discharge rates, along with the illustration of Li intercalation extent in LFP particle. (F) Linear relationship between intragranular diffusion velocity and the mass fraction of MC in LMC composite cathodes, along with the illustration of Li intercalation extent in LFP particles.

Redox performance of pure LFP and capacitive performance of pure MC could be easily measured in advance, so it was possible to quantificationally evaluate the synergistic effect between battery materials (LFP) and capacitive materials (MC). One important thing to note was that the additional capacity owing to the synergistic effect was supposed to originate from the more capacity release of LFP particles, as the capacity of MC mainly depended on its specific surface and conductivity should remain nearly unchanged in the composite cathode. Thus, different from the calculation of additional capacity “ΔC” based on the mass of both battery materials and capacitive materials in previous work^[31,32], we proposed a novel method to evaluate the synergistic effect more precisely. Moreover, the present LFP nanoparticles were close to spherical shape and had a narrow size distribution [Supplementary Figure 1], which allowed us to obtain the Li intercalation kinetics of single LFP particles.

As shown in Figure 2C, the LFP particle was simplified to a perfect sphere of radius R (equaled to 50 nm in this work). Taking the “shrinking core” model in previous work as a reference^[33-36], we assumed that the Li-intercalation reaction always occurred evenly from the outside to the inside, considering the one-dimensional diffusion channel of Li-ion in the matrix of LFP crystals. Accordingly, some indexes could be defined and calculated by:

Here, C_e and C_t,LFP were the capacity of various cathodes obtained experimentally and the theoretical capacity of LFP (170 mAh g^-1), while C_MC and C_LFP were the specific capacities delivered by pure MC and LFP cathodes. The f_LFP and f_MC were the mass fraction of LFP and MC; d was the intragranular diffusion length, and t was the discharge time. Specially, C_e,LFP obtained from Equation (3) was the actual capacity delivered by LFP in composite cathodes, which was more suitable to evaluate the synergistic effect and analyze the Li intercalation kinetics. The values of calculated C_e,LFP and capacity contribution of MC (CC_MC) in various LMC composite cathodes were listed in Supplementary Table 2. Further, the capacity ratio (CR) was used to compare the capacity of LFP with or without MC addition according to Equation (4), which was directly associated with the synergistic effect. Reaction depth (RD) with a value range from 0 to 1 represented the Li intercalation extent in LFP particles during discharge, and the intragranular diffusion length d could be then obtained according to Equation (5). More importantly, intragranular diffusion velocity (IDV) could also be calculated according to Equation (6) to compare the Li intercalation kinetics of LFP in the presence of MC or not. The values of CR, RD and IDV of LMC composite cathodes with different mass ratios of MC under various discharge rates were shown in Supplementary Figure 6 in detail. Herein, LMC@3_1 was chosen to compare with pure LFP in subsequent analysis because of its excellent rate performance.

Figure 2D showed that the specific capacity of LFP particles in LMC@3_1 cathode (C_e,LFP) was always higher than that in pure LFP cathode (C_LFP) under discharge rates of 1C-50C, and thus the CR values of LMC@3_1 were always larger than 1 (represented pure LFP), indicating the more capacity release of LFP particles as adding MC. In addition, the CR value also increased drastically when the discharge rate exceeded 10C. Quantitatively, this value was 1.12 at 10C and increased to 1.44 and 2.39 at 20C and 50C, demonstrating that the addition of MC in large amounts helped build a much stronger synergistic effect with LFP upon higher current density. These results convincingly validated the existence of a synergistic effect between battery materials (LFP) and capacitive materials (MC), which could be quantificationally evaluated by the calculation of CR. Further, Li intercalation kinetics of LFP could also be analyzed based on the single-particle model. As shown in Figure 2E, RD decreased with the increase of discharge rates for both cathodes, and the LMC@3_1 cathode always exhibited larger RD values than the pure LFP cathode at all rates. Strikingly, the higher the current density was, the larger the gap existed for the RD values between that with and without adding MC. This proved that the Li intercalation extent of LFP particles reduced enormously under high rates owing to the severe electron polarization and ion polarization. It, however, could be alleviated effectively with the addition of MC. Moreover, the value of IDV was enhanced enormously with the discharge rate [Figure 2F]. It reached 0.24 nm s^-1 at 50C for pure LFP (corresponding to the mass fraction of MC with 0), nearly 30 times that at 1C (8.5 × 10^-3 nm s^-1). That meant that the current density increased by 50 times, but the Li diffusion velocity just increased by 30 times, which well explained the limitation of ion diffusion under high current density. In comparison, the value of IDV at 50C for LMC@3_1 (corresponding to the mass fraction of MC with 0.25) was 0.29 nm s^-1 with 1.2 times larger than that for pure LFP, demonstrating that the limitation of ion diffusion did alleviate, to some extent, by adding MC. Specifically, IDV also exhibited a linear relationship with the mass fraction of MC, and the positive slope was improved with increasing discharge rate (from 0.019 nm s^-1 at 1C to 0.18 nm s^-1 at 50C). This validated that the rapid EDLC response of MC offering a Li⁺-rich environment for LFP greatly enhanced the Li intercalation kinetics during discharge, especially at a high rate of 10C-50C. The above results verified a sound synergistic effect between LFP and MC, not merely a linear addition of redox effect and EDLC effect, where MC could also enhance the Li intercalation kinetics of LFP, not merely acted as conductive additives as in most of previous work^[14,37,38].

Analysis of Li intercalation extent and phase transition by HRTEM observation and in-situ XRD characterization

The Li intercalation extent of LFP nanoparticles in pure LFP and LMC@3_1 cathodes after high-rate discharging was directly observed by HRTEM imaging. After discharging at 5C, some LFP particles in pure LFP cathode were intercalated with Li-ions completely [Figure 3A], while others were intercalated partially to generate the intermediate Li_xFePO₄ phase [Figure 3B and C]. Previous work had proved the presence of an intermediate Li_xFePO₄ phase during discharge of LFP particles by ex-situ and in-situ XRD study^[39-41]. Here, the lattice fringe spacings of 3.83-3.85 Å in Figure 3B and C, between 3.74 Å (corresponding to (210) plane of FePO₄) and 3.91 Å (corresponding to (210) plane of LFP), demonstrated the incomplete Li intercalation process after 5C discharge and thus led to the formation of intermediate Li_xFePO₄ phase. A similar phenomenon, but with more LFP phase present, could be found in HRTEM images of LFP particles in LMC@3_1 cathode after 5C discharge [Figure 3D-F], and the phase distribution in Figure 3E confirmed the reasonability of assumption in Figure 2C that the Li intercalation proceeded preferentially on the outside of LFP particles. Considering the illustration of Li intercalation extent in Figure 2E, where both LFP and FePO₄ phases were abundant, the observed intermediate Li_xFePO₄ phase might be the mixed crystal of LFP and FePO₄, which lay in the transition region or interface between these two phases^[33,42,43].

Figure 3. HRTEM images of LFP nanoparticles in pure LFP and LMC@3_1 cathodes after discharging at 5C and 20C, involving the lattice fringe spacings and corresponding phase composition and lattice plane. (A-C) LFP particles in pure LFP cathode after 5C discharge. (D-F) LFP particles in LMC@3_1 cathode after 5C discharge. (G-I) LFP particles in pure LFP cathode after 20C discharge. (J-L) LFP particles in LMC@3_1 cathode after 20C discharge.

After discharging at a high rate of 20C, the Li intercalation extent of LFP particles in pure LFP and LMC@3_1 cathodes exhibited a huge difference. The lattice fringe spacings of 4.94 and 4.43 Å observed in pure LFP cathode were slightly larger than that of 4.91 and 4.30 Å (corresponding to standard FePO₄ phase), which should be attributed to the intercalation of bits of Li-ions into the lattice of FePO₄ [Figure 3G-I]. However, owing to the extremely low Li intercalation extent of LFP particles at such a high discharge rate, the intercalated-free FePO₄ phase was dominant, and there were few LFP phases observed. While for the LMC@3_1 cathode, a much more abundant LFP phase could be observed in Figure 3J-L on account of the enhancement effect of MC to the Li intercalation kinetics of LFP particles compared with that in pure LFP cathode. In addition, the ordered arrangement of LFP and FePO₄ phases in Figure 3K again verified the direction of Li intercalation from the outside to the inside, leading to the formation of the LFP phase at the particle edges and the extended lattice fringe spacings in FePO₄ phase due to the incomplete intercalation process. These direct observations in HRTEM images revealed the intercalation behavior of Li-ions inside LFP nanoparticles, demonstrating that the addition of MC enormously increased the Li intercalation extent and, thus, effectively improved the capacity release of LFP-based cathode during high-rate discharge.

To deeply understand the difference of changes in phase composition and crystal structure of LFP with or without MC during the electrochemical reaction, we further conducted the in-situ XRD experiments under 0.5C and 2C charge/discharge using the in-situ cell of reflection type. Supplementary Figure 7 showed a part of original XRD patterns and the match of peak position to standard cards. Upon charge at 0.5C, the major peaks changed from 25.59°, 29.74°, 32.24°, 35.62°, 36.57° to 25.79°, 30.29°, 30.89°, 36.68°, 37.43° accordingly, which indicated the Li extraction of LFP and the formation of FePO₄ phase during charge process [Figure 4A and B]. Additionally, the opposite changes from FePO₄ to LFP happened when discharging. WPF and RF were performed for acquiring the quantitative results of phase composition and crystal structure. Figure 4C presented the mass fraction evolution of LFP and FePO₄ during 0.5C charge (from 0% to 100% SOC) and 0.5C discharge (from 100% to 0% SOC). In detail, the addition of MC resulted in the timely appearance of the cross-point (50% LFP phase and 50% FePO₄ phase) at 55% SOC during charge, which increased to 62% SOC for pure LFP. Besides, complete Li extraction occurred at 86% SOC as adding MC, while it was associated at 100% SOC for pure LFP. These all indicated the slower Li extraction rate of LFP without MC. Moreover, it was also observed that the FePO₄ phase kept for a much wider SOC range in LMC@3_1, which could be ascribed to the capacitive behavior of MC and, thus, no phase transition around 100% SOC (i.e., at the end of charge and the beginning of discharge). During the discharge process, the complete phase transition from FePO₄ to LFP was achieved from 73% to 7% SOC (a practical range of 66% SOC) for LMC@3_1, while it was from 86% to 0% SOC (a practical range of 86% SOC) for pure LFP. This comparison directly verified the Li intercalation kinetics of LFP particles could be greatly enhanced by large amounts of MC. In addition, variation of lattice parameter b of LFP and FePO₄ reflected the changes of crystal volume in the diffusion orientation of Li-ions. As shown in Figure 4D, the lattice parameter b in the LMC@3_1 cathode changed much more slightly compared with that in pure LFP, which revealed a more uniform reaction inside LFP particles in the presence of MC.

Figure 4. Heat map of original in-situ XRD patterns under 0.5C charge/discharge: (A) pure LFP cathode, (B) LMC@3_1 cathode. Whole pattern fitting and Rietveld refinement results of original in-situ XRD patterns at different SOC under 0.5C charge/discharge: (C) mass fraction of LiFePO₄ and FePO₄, (D) lattice parameter b of LiFePO₄ and FePO₄. Whole pattern fitting and Rietveld refinement results of original in-situ XRD patterns at the end of 2C charge and discharge: (E) end of 2C charge for pure LFP, (F) end of 2C discharge for pure LFP, (G) end of 2C charge for LMC@3_1, (H) end of 2C discharge for LMC@3_1.

Besides, the in-situ XRD patterns were collected under a high charge/discharge rate of 2C, and the original XRD patterns at the end of charge/discharge were further processed by WPF and RF to obtain the quantitative results of phase composition [Figure 4E-H]. For a pure LFP cathode, there was still 8.6 wt% LFP residual at the end of 2C charge, and 32.2 wt% FePO₄ remained unable to finish the Li intercalation process when 2C discharge completed. In sharp contrast, there was only 5.6 wt% LFP residual in the LMC@3_1 cathode at the end of 2C charge, and all FePO₄ was completely intercalated with Li-ions during 2C discharge. The above results in Figures 2-4 all validated that the rapid capacitive effect of MC enhanced the Li intercalation kinetics of LFP nanoparticles in LMC@3_1 cathode by comparing with pure LFP cathode.

Electrochemical characterization of pure LFP and LMC@3_1 cathodes

As follows, EIS results at various SOC were analyzed using equivalent circuit models to understand the enhancement of Li-ion diffusion performance and electrochemical reaction rate of LFP by adding MC. The parameters of R_S, C_SEI//R_SEI, Q_dl//R_ct, Q_neg//R_neg and Q_w in the equivalent circuit model of Figure 5A were explained in Supplementary Figure 8, and the key fitting parameters at various SOC from 10% to 90% were listed in Supplementary Table 3. Specially, charge transfer resistance (R_ct) represented the difficulty for the electrochemical reaction to occur, and the linear part in the low-frequency region was related to the Li-ion diffusive behavior inside LFP particles and expressed by a constant phase element Q_w in series. As shown in Figure 5A, a smaller diameter of the arc was observed in the mid-high-frequency region for the LMC@3_1 cathode, indicating the lower R_ct for combining Li-ions and electrons in the interface of LFP/electrolyte by adding MC. Quantitatively, the average R_ct value from 10% to 90% SOC was 6.03 Ohm for LMC@3_1, much smaller than that (10.64 Ohm) for pure LFP [Figure 5B]. This explained well that the local Li⁺-rich region generated by EDLC effect of MC was favorable to the charge transfer process in LFP/electrolyte interface and, thus, accelerated the subsequent electrochemical reaction inside LFP particles. On the other hand, the solid-phase Li-ion diffusion coefficient ($$ D_{\mathrm{Li}}{ }^{+} $$, cm² s^-1) at various SOC could also be obtained according to the linear fitting of Z_Re to ω^-0.5 in the low-frequency region^[44] (Figure 5C; 0.01-0.1 Hz in this work as the inset of Figure 5A; detailed calculation process in Supplementary Calculation 2). The average $$ D_{\mathrm{Li}}{ }^{+} $$ value was 5.739 × 10^-17 cm² s^-1 for LMC@3_1, approximately six times larger than that (9.184 × 10^-18 cm² s^-1) for pure LFP. This result further validated that the local Li⁺-rich region generated by EDLC effect of MC indeed increased the Li-ion diffusion rate inside LFP particles, a key factor to improve the electrochemical performance of LFP under high current density.

Figure 5. (A) EIS results at 50% SOC and the corresponding equivalent circuit model (inset: linear fitting of Z_Re to ω^-0.5 in the frequency range of 0.01-0.1 Hz). (B) Charge transfer resistance (R_ct) at various SOC. (C) Solid-phase Li-ion diffusion coefficient ($$ D_{\mathrm{Li}}{ }^{+} $$) at various SOC. (D) Voltage-capacity curves at 15C charge/discharge and the schematic illustration for calculating overpotential during charge (η_ch) and discharge (η_dis). (E) Tafel plot and the linear fitting in large overpotential regions. (F) Calculated exchange current density (j₀) during charge and discharge.

In addition, the voltage-capacity curves at 15C charge/discharge (voltage curves at other rates were shown in Supplementary Figure 9) were plotted and used to calculate the overpotential during charge (η_ch) and discharge (η_dis), as shown in Figure 5D. Overpotential was defined as the difference between thermodynamic potential (3.43 V vs. Li/Li⁺ for LFP) and experiment data at half charge/discharge in this work^[45,46]. LMC@3_1 exhibited lower charge voltage, higher discharge voltage and higher specific capacity than pure LFP upon high current density (2.55 A g^-1). Therefore, the electrode polarization extent of LMC@3_1 was far weaker than that of pure LFP. That meant high energy density under high power density was easier to achieve by adding MC for LFP-based devices. Moreover, the Tafel plot in Figure 5E was determined according to the various overpotentials and the logarithm of corresponding areal current densities (j, based on the product of the specific surface area of LFP and its mass in the cathode). Tafel regions at large overpotentials for both charge and discharge processes were then linearly fitted and the intercept at η = 0 represented the magnitude of the exchange current density (j₀), which was summarized in Figure 5F. The larger j₀ for LMC@3_1 compared with pure LFP implied the electrochemical reaction inside LFP particles was easier to occur with smaller electrochemical polarization in the presence of MC, consistent with the results in Figure 5B. Since the redox process did not occur in MC during charge/discharge, the above enhancement effect in LMC@3_1 cathode, i.e., the better Li-ion diffusion performance and the faster electrochemical reaction rate, could be ascribed to the establishment of local Li⁺-rich region around MC due to its rapid EDLC effect, contributing to the fast Li intercalation/extraction kinetics.

According to the total mass of active materials in the cathode, the present LMC@3_1 cathode exhibited a maximum discharge energy density of 437 Wh kg^-1 with a power density of 568 W kg^-1 and maintained 104 Wh kg^-1 with a maximum discharge power density of up to 23.4 kW kg^-1, proving the outstanding high-power characteristic of LFP/MC composite cathode owing to the enhanced Li intercalation kinetics of LFP with MC. Cycle performance of pure LFP and LMC@3_1 was also tested by 1C charge and 10C discharge and the electrochemical performance (specific capacity, energy density, power density, cycle life, etc.) of various reported LIBCs was summarized in Supplementary Table 4. Figure 6A showed the capacity of LMC@3_1 increased slightly in the previous dozens of cycles and nearly 100% capacity retention could be achieved after 1,000 cycles, far more excellent than that for pure LFP (31.5% retention). In addition, the coulombic efficiency of LMC@3_1 was stable at about 100%, while it fluctuated greatly in the late cycles for pure LFP. In order to investigate the reasons for this cyclic difference, we disassembled the coin cells after cycling and characterized the electrode. An SEM image in Figure 6B revealed that severe fragmentation of LFP particles in pure LFP electrodes occurred after 1,000 cycles. The pristine single-crystal particles of around 100 nm [Supplementary Figure 1] were broken down into small fragments from a few to several tens of nanometers, with numerous cracks present. In contrast, although the fragmentation of LFP particles was also observed in the LMC@3_1 electrode [Figure 6C], the majority maintained their intact single-crystal morphology. We believed that the presence of MC favored better Li-ion diffusion performance, which facilitated more uniform Li-ion intercalation and extraction reactions and reduced volume changes of LFP particles during charge and discharge processes (as concluded in Figure 4D). Furthermore, the excellent conductivity of MC helped mitigate the impact of large currents on LFP particles as well^[32]. Therefore, LFP particles in the LMC@3_1 electrode could maintain a more intact morphology after cycling, resulting in superior cycle performance. The in-situ XRD experiments of the cycled electrodes under 2C charge/discharge were also conducted to explore the phase transition process. Figure 6D showed that there were still noticeable intensity peaks at 29.4° and 35.3° at the end of charge, indicating some LFP particles in cycled pure LFP electrode had not completed the Li extraction process (7.8 wt% LFP residual according to the Rietveld refinement result). This might be due to the severe fragmentation of LFP particles and the damage to the crystal structure after prolonged cycling, resulting in the deactivation of some LFP particles and, thus, a sharp decrease in capacity of the pure LFP electrode after cycling. In contrast, Figure 6E showed no intensity peak at 29.4° and 35.3° at the end of charge and no appearance of the FePO₄ phase at the end of discharge. This indicated that even after 1,000 cycles, the LMC@3_1 electrode could still undergo complete charge and discharge processes, thereby maintaining nearly 100% of the initial capacity. These results proved the rapid and sufficient Li-ion supply, the slight ion polarization and electron polarization for LMC@3_1 were favorable to a stable cycle life. Apparently, it was a very simple but robust method to increase the capacity, power density and cycle life of LFP-based devices by adding MC in large amounts.

Figure 6. (A) Cyclic test of pure LFP and LMC@3_1 cathodes under 1C charge and 10C discharge. (B) SEM images of pure LFP electrode after cyclic test. (C) SEM images of LMC@3_1 electrode after cyclic test. Heat map of original in-situ XRD patterns under 2C charge/discharge for (D) pure LFP and (E) LMC@3_1 electrodes after cyclic test.