It is commonly recognized that Li vacancy defects (Liv) and Fe occupation of Li site (FeLi) are one of the major issues that are responsible for the performance degradation of LFP cathode. Due to the formation of Liv defects, iron at the iron-rich grain boundaries of degraded LFP is partly oxidized into Fe3+, meanwhile, the partial cations (e.g., Fe2+) start to migrate into lithium layer and occupy lithium site (Fig. 1), forming the as-defined ‘‘anti-site’’ defects, which ultimately obstructs the Li+ diffusions [1, 20, 21]. As it turns out, there are two different phases: olivine-type FePO4 (FP) and LiFePO4. Inspired by the Liv and FeLi defects in cycling LFP material, we should precisely resolve the Liv and anti-site defects, thereby resume their electrochemical activity of the C-LFP. It requires high activation energy to complete the migration process of Fe ions returning to the original position (H0). In our design, a reaction environment which involves ambient-pressure and low-temperature provides relithiation-reduction system, which can be beneficial to re-dose lithium ions and the migration and reduction of Fe ions. The eutectic Li-molten salt solution is considered to be a naturally homogeneous ionic conductor, facilitating the relithiation-related Li+ diffusion into the Li-defificient particles during the reaction process. The citric acid, an electrical conductor, donates electrons in the reaction system, reducing the Fe3+ and subsequently dropping the migration barrier to shift Fe2+ from the H1 to H0 site (Fig. 1). With that in mind, the relithiation mechanism reactions are presented below [22]:
$$x{Li}^{+}+{Li}_{1-x}{FePO}_{4}+x{e}^{-}\to {LiFePO}_{4}$$
1
$${2OH}^{-}+2{e}^{-}\to 2{H}_{2}O+{2O}^{2-}$$
2
$${N{O}_{3}}^{-}+{e}^{-}\to {NO}_{2}+{O}^{2-}$$
3
$${C}_{6}{H}_{8}{O}_{7}+{O}^{2-}-2{e}^{-}\to 4C+{2CO}_{2}+{4H}_{2}O$$
4
To determine the appropriate reaction condition and explain the regeneration process, thermal analysis of C-LFP powders and the mixture (C-LFP and eutectic Li salts) were conducted (Fig. 2a). Two stages of the mass loss were observed in the TG curve of the mixture: the first mass loss of 19% above 100 ℃ corresponds to the loss of crystal water from LiNO3 and LiOH, the second predominant mass loss of about 15% in the range of 150–350 ℃ is associated with the melting of the eutectic Li salts and the gas evolution via Eqs. 2 and 3 because the O2, water (vapor) and NO2 are generated through the lithiation [3, 23]. A mass loss slightly of C-LFP above 550 ℃ is observed, which is attributed to the decomposition of salts. Therefore, a series of temperature ranges of 350–550 ℃ were used for the relithiation experiment in this work.
To promote the relithiation and crystallization of the spent particles, the C-LFP powers were treated by the molten salt thermochemical process under at different annealed temperatures. The chemical compositions of the C-LFP and R-LFP materials were monitored and compared (Fig. 2b). The Li/Fe molar ratio of degraded LFP was only 0.45, which should be largely to blame for the capacity loss in spent LIBs after a long time. After regeneration in molten salt medium, the Li compositions of the C-LFP materials are fully recovered to the ideal stoichiometry (about 1.0 of Li).
The crystalline structures of the spent LFP and R-LFP samples with different reaction temperature were analyzed by XRD, as shown in Fig. 3a. The peak annotated with diamonds were typically attributed to the olivine FP (JCPDS 37–0478), while the other diffraction peaks are consistence with the olivine LFP crystal system (JCPDS 40-1499) [24, 25]. The existence of the FePO4 phase was result of lithium loss after long-term cycles. After relithiation, the XRD patterns of all R-LFP powers show good accordance to the olivine LFP (JCPDS 40-1499) [24, 25]. This directly demonstrated that the Li+ form eutectic Li-molten salt lithium was effectively supplemented into the Li-poor materials through homogeneous thermochemical process, and the delithiated crystal structure of FP was successfully converted to LFP phase. From the quality point of view, it provides a high-purity LFP phase without the Fe2O3 phase in the regenerated product.
Figure 3b shows XPS spectra of Fe 2p for the C-LFP, R-LFP and pristine LFP materials. The binding energies at 711.96 (Fe 2p3/2) and 725.62 eV (Fe 2p1/2) of C- LFP samples indicates the Fe3+ valence state, further confirming the presence of the FePO4 in the spent materials. XPS spectra of R-LFP and pristine LFP displays two obvious shifts of the main peaks: one (Fe 2p3/2) from 711.96 eV (Fe3+) to 710.68 eV (Fe2+) and another (Fe 2p1/2) from 725.62 eV (Fe3+) to723.92eV (Fe2+). Interestingly, a similar peak shift was reported in the literature [25, 26]. The XRD result indicates the conversion of the FePO4 to the LiFePO4 phase through thermochemical healing [27, 28].
The detailed structural evolution of C-LFP, R-LFP samples were carefully examined by high-resolution TEM (HR-TEM). The TEM images (Fig. 4a and Fig. 4c) show that spent LiFePO4 and R-LFP particles are irregular shaped with an average size of nearly 200 nm. HR-TEM of the C-LFP sample has a d-spacing of 0.344 nm, confirming the formation of a FePO4 phase after repeated long-term cycling of the battery. The appearance of FePO4 in C-LFP sample proves that the loss of lithium is the chief factor affecting behind the failure of LFP batteries. Difffferent interplanar spacings of 0.392 nm (Fig. 4d), 0.246 nm (Fig. 4f) and 0.347 nm (Fig. 4g) were observed from lattice fringes of the R-LFP sample, which can be indexed to the (210), (121) and (111) planes of LFP respectively.
To investigate the efficiency of regeneration, the electrochemical properties of spent LFP and all R-LFP were evaluated under a voltage range of 2.5-4.0 V. Figure 5 depicts the initial charge-discharge profiles of all LFP cathodes at 0.2 C. The spent LFP cathode exhibits a low specific capacity of 69.14 mAh g− 1, with the electrochemical platform almost disappear, indicating the spent LFP lose the electrochemical activity after multiple charge-discharge cycle. The specific capacity of R-LFP-350, R-LFP-450 and R-LFP-550 achieves 151.2 mAh g− 1, as large as that of spent LFP cathode. The electrochemical platforms are gradually emerged with the increase of the annealing temperature, verifying that the composition, structure and electrochemical activities of spent LiFePO4 are fully recovered by defect-targeted healing. Most notably, the first charge capacities of R-LFP are much better than the values obtained by other regeneration strategy (Table 1).
Table 1
The capacities of R-LFP and other regeneration strategy.
Regeneration strategy | Discharge capacity (1st)/mAh g− 1 | Discharge capacity (Cycles number)/mAh g− 1 | Ref. |
Solid-state reaction | 147.3 (0.2 C) | 95.31% (100st, 0.2 C) | 29 |
Solid-state reaction | 144 (0.1 C) | 93.75% (100st, 0.1 C) | 30 |
Carbon-thermal reduction | 130–140 (0.2 C) | - | 31 |
Hydrothermal method | 163 (0.1 C) | - | 20 |
Thermal reduction | 151.9 (0.1 C) | - | 2 |
Hydrothermal method | 146.2 (0.2 C) | - | 11 |
Thermochemical Healing | 151.2 (0.2 C) | 97.73% (150st, 1 C) | Our work |
The rate performance of spent LFP and all R-LFP electrode was also evaluated at various current densities (Fig. 6b and Fig. 6c). The spent LFP show specific capacities of 71.69, 62.4, 49.1, 35.5 and 18.9 mAh g− 1 at 0.2, 0.5, 1, 2 and 5 C, respectively. The R-LFP-550-no CA electrode shows inferior rate capability, the capacities are about 140.6, 89.5, and 65.1 mAh g− 1 at 0.2, 2 and 5 C. Moreover, the R-LFP-550 delivers a capacity of 150.2, 120.4, and 101.5 mAh g− 1 at 0.2, 2 and 5 C, superior to that of R-LFP-no-CA, R-LFP-450 and R-LFP-350, especially at high rates. The specific capacities of R-LFP are quickly retained after cycling at 0.2 C. The carbon layer in R-LFP materials is conducive to electron transfer and used as a buffer layer for volume effect of Li+ in the process of charge and discharge. As for the cycling performance, the R-LFP-550 cathode displayed a specific capacity of 128.0 mAh g− 1 in the first cycle at 1 C and 125.1 mAh g− 1 after 150 cycles with a capacity retention of 97.73%. The capacity retention of R-LFP-no-CA only reach 74.59% after 150 cycles. All R-LFP with CA have excellent cycling performances compare to that of R-LFP-no-CA.