1 Introduction

LiFePO4 (LFP) has been receiving much attention as alternative cathode material for lithium-ion batteries (LIBs) due to its excellent thermal safety, environmental friendliness, and abundant sources for low cost [1,2,3]. Although LFP suffers from intrinsically low electronic and ionic conduction, intensively studies and modifications during recent two decades have enabled a satisfactory rate capability at ambient and high temperatures by methods of coating conductive agents, doping alien ions, and tailoring morphology [4,5,6,7,8,9,10]. However, in cold climates and high-altitude drones, LIBs are required to work at subzero temperature. Cold temperature causes sluggish diffusion rate and slow reaction kinetics. Therefore, both discharge capacity and output voltage of LFP are substantially dropped [11,12,13].

Previous researches have shown that inferior low-temperature performance of LFP is attributed to low Li+ diffusion of electrode materials and high charge-transfer resistance of electrolyte/electrode interface [13, 14]. Most of the approaches to solve this issue are optimizing electrolyte by introducing additives to tune electrolyte/electrode interface for reducing charge-transfer resistance [11, 12, 15,16,17]. In contrast with the electrolyte development, insufficient strides have been achieved in electrode materials design for the low-temperature application. Generally, increasing insertion/desertion channels could accelerate the charge/discharge procedure, even at low-temperature surrounding. Previously, the low-temperature performance of LFP was effectively enhanced by coating Li3V2(PO4)3 material that offers three-dimensional (3D) diffusion channels for insertion/desertion of Li+, in contrast to one-dimensional (1D) diffusion channels of Li+ for olivine-structured LFP along the b-axis [18, 19]. Recently, our group has demonstrated the analogous advantages for providing 3D fast Na+ diffusion channels in NASICON-structured Na3V2(PO4)3 (NVP) crystals, as thus, which is used as an excellent cathode material for all-climate sodium ion batteries systems [19, 20]. Considering the ionic radius of Li+ is much smaller than that of Na+ (0.076 nm vs. 0.102 nm) [21, 22], it can be expected that Li+ could also transfer rapidly in the ion-diffusion channels of NVP. Therefore, coating NVP layer as a fast Li+ diffusion intermediary should be an effective strategy to assistant the insertion/desertion of Li+ for LFP at low temperature.

In this work, a multi-core–shell-structured LFP@NVP@C composite was designed, in which LFP particles are the cores, NVP as the inner layer to improve the interface kinetics and carbon layer as the outer layer to enhance electronic conductivity of the whole composite, as illustrated in Fig. 1. The low-temperature performances of as-prepared electrodes were systematically investigated through electrochemical characterizations. At − 10 °C with the rate of 0.5C, LFP@NVP@C delivered a discharge capacity of 96.9 mAh·g−1, which is higher than LFP@C (80.2 mAh·g−1) without the NVP inner layer and mixed LFP@C/NVP@C composite (80.6 mAh·g−1). This enhancement is due to the NVP layer with 3D diffusion channels on the surface of LFP, enabling more insertion/desertion of Li+ into the crystals and reducing charge-transfer resistance at low temperatures. We do believe this work will open an important window for the research of multifunctional materials for application in low-temperature battery systems.

Fig. 1
figure 1

Schematic illustration used to synthesize LFP@NVP@C composite

Full size image

2 Experimental

2.1 Material synthesis

LFP@C and NVP@C composites were synthesized, respectively, according to our previous report [10, 20]. LFP@C was prepared after experiencing the process of rheological phase, high-energy ball milling and a subsequent carbothermal reduction. While NVP@C was prepared through solid-state strategy, combined with two-step ball-milling and a subsequent carbothermal reduction. In order to synthesize LFP@NVP@C composite, mixture of LFP@C powder and the precursor of NVP@C (95:5 by weight) was ball milled at 300 r·min−1 for 3 h and then calcined at 700 °C for 8 h under Ar atmosphere. For comparison, LFP@C/NVP@C powder was obtained by mixing LFP@C and NVP@C composites (95:5 by weight).

2.2 Materials characterizations

The materials phases investigated by X-ray diffractometer (XRD, Rigaku MiniFlex 600, Japan) was used to analyze the materials phases, the diffraction angle was scanned from 10° to 60° using the Cu Kα radiation (λ = 0.154178 nm). The high-resolution transmission electron microscopy (HRTEM, F30) was used to investigate the morphology and microstructure of as-prepared samples. The elemental compositions of the prepared samples were examined by inductively coupled plasma atomic emission spectrometry (ICP-AES, Perkin Elmer, 300DV) and X-ray photoelectron spectrometer (XPS, Axis Ultra, Kratos Analytical Ltd.).

2.3 Electrochemical measurements

The electrodes slurry was prepared by mixing the as-prepared samples with carbon black and PVDF (polyvinylidene fluoride) binder in a certain weight ratio of 80:10:10, and N-methylpyrrolidone (NMP) was the slurry solvent. After pasting on the Al foil and drying overnight in vacuum at 100 °C, the working electrode could obtain by using the mold (14 mm in diameter) to cut. The cathode loading was about 2–3 mg·cm−2. The CR2025 coin cells were assemble in Ar-filled glove box, which Li foil served as the counter electrode, LiPF6 in carbonate as electrolyte and Celgard 2400 as separator. The LAND CT-2001A (Wuhan, China) battery testing system was used to investigate the galvanostatic charge–discharge behaviors. Cyclic voltammetry (CV) measurements were performed on CHI 660E electrochemical workstation (CHI Instrument, Shanghai, China). The testing voltage window range was 2.5–4.2 V.

3 Results and discussion

Generally, LFP is modified with carbon layer for core–shell structure to enhance electronic conductivity of the composite [6, 8, 23, 24]. Here, we intend to further modify the surface of LFP with NVP inner layer for enhancing insertion/desertion of Li+ through constructing multi-core–shell structure. Microstructure of LFP@C and LFP@NVP@C samples were characterized by TEM in Fig. 2. The LFP@C composite is composed of LFP with a high crystallinity and amorphous carbon coating of 2–3 nm (Fig. 2a). Although core–shell-structured LFP@C composite allows a sufficient diffusion of Li+ in the crystals and electron transfer via outer carbon layer, 1D diffusion channels of LFP along the b-axis only offer a limited window for insertion/desertion of Li+. The LFP@NVP@C displayed dual coatings consisted of NVP inner layer and carbon outer layer (Fig. 2b). It can be observed the interplanar spacing of 0.30 nm corresponds to (121) face of LFP and interplanar spacing of 0.31 nm corresponds to (031) face of NVP, respectively [10, 20]. In addition, there are evident boundaries between LFP, NVP and carbon. The thicknesses of NVP and carbon layers are 3–4 nm and 1–2 nm, respectively. High-angle annular dark-field scanning TEM (HAADF-STEM) images characterized elemental distribution of Fe, V and Na in LFP@NVP@C, as shown in Fig. 2c–f. Uniform distributions of elemental V and Na in the periphery of elemental Fe were witnessed, indicating the existence of a thin NVP coating layer. Owing to 3D Li+ diffusion channels of NASICON, the NASICON-structured NVP is supposed to be an ionic-conductive intermediary with a highly open framework to introduce Li+ into the composite.

Fig. 2
figure 2

TEM images of a LFP@C and b LFP@NVP@C; HAADF-STEM image for LFP@NVP@C: c position and elemental distribution of d Fe, e V and f Na

Full size image

Figure 3 shows XRD patterns of LFP@NVP@C and LFP@C. It can be observed that all characteristic peaks of LFP@NVP@C and LFP@C are indexed as ordered orthorhombic olivine LFP phase (JCPDS No. 81-1173), illustrating that low amount of addition of NVP won’t influence LFP crystal structure. No traces of carbon in LFP@NVP@C and LFP@C were detected due to that residual carbon is amorphous [10]. The carbon contents of LFP@NVP@C and LFP@C are 1.85 wt% and 1.35 wt%, respectively. No diffraction peaks of LFP@NVP@C corresponding to NVP phase were collected due to low amount of NVP component. ICP was further employed to characterize the elemental compositions of LFP@NVP@C and demonstrated that the content of V element was ca. 1.08 wt%, which is consistent well with the V weight ratio of raw materials (1.11 wt%). Further structure information about LFP@NVP@C was characterized by XPS, as shown in Fig. 4. V was confirmed in LFP@NVP@C. High resolution spectrum of C 1s of LFP@NVP@C shows that there is most of sp2-type carbon, indicating a highly graphitized carbon layer that can facilitate electronic migration [25]. In the high resolution XPS spectrum of V 1s, the peak at binding energy of 515.75 eV belongs to the V(III) [20].

Fig. 3
figure 3

XRD patterns of LFP@NVP@C, LFP@C, NVP [18], and LFP (JCPDS No. 81–1173)

Full size image
Fig. 4
figure 4

XPS of LFP@NVP@C: a XPS survey, b high resolution of C 1s curve, and c high resolution of V 2p curve

Full size image

The galvanostatic charge/discharge tests at various temperatures of LFP@NVP@C and LFP@C (shown in Fig. 5) were first investigated. At a rate of 0.5C, LFP@NVP@C electrode shows reversible capacities of 160.9, 111.8, 96.9 and 81.3 mAh·g−1 at 23, 0, − 10 and − 25 °C, respectively, which are 2.2%, 3.6%, 15.5% and 18.6% higher than that of LFP@C. Compared with LFP@C, LFP@NVP@C delivers lower charge plateaus as well as higher discharge plateaus, indicating a smaller polarization for interfacial reaction between electrolyte and electrode. Even at − 25 °C, LFP@NVP@C delivers discharge voltage of ca. 3.2 V, respectively, which potentially ensures sufficient power density of LIBs at low temperatures. Coulombic efficiencies of LFP@NVP@C are near 100% at various temperatures. Therefore, LFP@NVP@C exhibits a better electrochemical performance than LFP@C due to the contribution of NVP layer. The NASICON-structured NVP has open framework to offer fast insertion/desertion of Li+ for LFP@NVP@C.

Fig. 5
figure 5

Electrochemical results of LFP@NVP@C (solid line) and LFP@C (dash line) at various temperatures with 0.5C

Full size image

Rate performances of as-prepared electrodes containing LFP@C/NVP@C were further evaluated at various temperatures, as shown in Fig. 6. At 23 °C, LFP@NVP@C, LFP@C, and LFP@C/NVP@C electrodes exhibit similar electrochemical performances at most rates. Room temperature ensures enough reaction kinetics for the release of capacity of electrode materials. Only one difference for three as-prepared electrodes occurs at the rate of 10.0C. LFP@NVP@C and LFP@C/NVP@C deliver discharge capacities of ca. 100.1 and 93.0 mAh·g−1, respectively, which both exceed LFP@C (82.3 mAh·g−1), indicating that the addition of NVP into LFP is effective for rate performance. In addition, the discharge capacity of LFP@NVP@C is higher than that of LFP@C/NVP@C, illustrating that NVP layer based on multi-core–shell structure plays a beneficial role on facilitating more insertion/desertion of Li+ into LFP crystal. As temperature decreased, effectiveness of NVP coating on improvement of low-temperature performance becomes more distinct. In contrast to LFP@C suffering from inferior low-temperature performance, LFP@NVP@C exhibits enhanced electrochemical properties of Li+ storage. At 0 °C, discharge capacities of LFP@NVP@C are 111.8, 97.6, 84.4 and 61.8 mAh·g−1 at 0.5C, 1.0C, 2.0C and 5.0 C, respectively, which are 8.3%, 10.4%, 16.1% and 33.7% higher than that of LFP@C. Interestingly, discharge capacities of LFP@C/NVP@C are side by side with that of LFP@C at less than 5.0C, while are much close to that of LFP@NVP@C at 10.0C. The NVP@C has been reported to have an excellent low-temperature performance, and therefore NVP@C makes major contribution for discharge capacity of mixed LFP@C/NVP@C electrode [13, 20]. At − 10 °C, LFP@NVP@C still delivered discharge capacities of 96.9, 82.2, 66.7 and 41.6 mAh·g−1 at 0.5C, 1.0C, 2.0C and 5.0C, respectively. Even at − 25 °C, LFP@NVP@C exhibited the best rate performance among three as-prepared electrodes. The above results have proved that the NVP with 3D framework and multi-core–shell structure could significantly enhance the performance of LFP@NVP@C at low-temperature. Moreover, the rate performances of LFP@NVP@C below − 20 °C is compared with that reported in previous work, as shown in Table S1.

Fig. 6
figure 6

Rate performance of LFP@NVP@C, LFP@C and LFP@C/NVP@C at different temperatures: a 23 °C; b 0 °C; c − 10 °C; d − 25 °C

Full size image

In order to understand the mechanisms of LFP@NVP@C with enhanced low-temperature performances, the electrochemical kinetics of LFP@NVP@C and LFP@C were compared by investigating the CV curves as shown in Fig. 7. A pair of redox peaks of LFP@C were evidently collected at 23 °C, corresponding to Li+ insertion/de-insertion in LFP crystals that are in conjunction with Fe2+/Fe3+ redox couple [6, 10, 26]. As operating temperature decreases down to − 25 °C, the potential interval between anodic/cathodic peaks increases from 333 to 787 mV and peak current drops to a quarter of value at room temperature, which indicates a slowdown of electrochemical reactions of LFP@C. As for LFP@NVP@C, similar to LFP@C, increased polarization and reduced peak current were also observed with operating temperature decreasing. Potential intervals of LFP@NVP@C are 298, 442, 579 and 758 mV at 25, 0, − 10 and − 25 °C, respectively. All the values are lower than that of LFP@C, implying NVP coating layer could relieve the polarization of the LFP@NVP@C electrode.

Fig. 7
figure 7

CV curves of LFP@NVP@C (solid line) and LFP@C (dash line) electrode with scanning rate of 0.1 mV·s−1

Full size image

Then the electrochemical impendence spectra (EIS) at various temperatures were analyzed. As displayed in Fig. 8a, b, as operating temperature decreases, the diameters of semicircles in the profiles of LFP@C//Li and LFP@NVP@C//Li half-cell become significantly larger, respectively. Similarly, the EIS curves of symmetric Li//Li cell also demonstrate a rapidly increased semicircle diameter and thus increased impedance (Fig. 8c). Such increases are well consistent with our recent previous works that charge-transfer resistances of half-cell increasing at low temperatures are mainly attributed to the metallic anode [19, 20, 27], which has demonstrated that using half-cell was not best mode to assess charge-transfer resistance of targeted materials.

Fig. 8
figure 8

EIS characterizations at different temperatures: cell structure and resultant EIS curves of a LFP@C, b LFP@NVP@C, and c symmetric Li disks

Full size image

Therefore, the EIS spectra of LFP@C and LFP@NVP@C with symmetric cells were further assessed. As shown in Fig. 9a, b, symmetric LFP@C//LFP@C and LFP@NVP@C//LFP@NVP@C cells offer a single semicircle with similar diameters, illustrating that the Li electrode is responsible for the increased impedance [20]. In EIS curves of Fig. 9a, b, the intersection on the x-axis represents bulk resistance of the cell (Ru) including the electrolyte and electrodes [4], and semicircle diameter represents corresponding charge-transfer resistance (Rct) reflecting electrochemical kinetics of interfacial reaction [19]. All the values of Ru and Rct are listed in Table 1. The Ru values gradually increase with temperature decreasing can be attributed to slow down Li+ diffusion in the electrolyte [28,29,30,31,32]. An increase in the values of Rct indicates that decreased operating temperatures make reaction kinetics of electrode difficult. Compared with LFP@C, LFP@NVP@C exhibits lower Rct values from 43.74 to 54.55 Ω with temperature decreasing. The NVP offers easy access to Li+ for LFP and thus enables LFP@NVP@C more capacity and lower polarizations than LFP@C during the charge/discharge. In addition, the apparent Li+ diffusion coefficient (DLi) is calculated by the straight slope lines at low frequency region [31, 33, 34]. The Warburg coefficient (σ) can be calculated by Eq. (1):

$$\left| {Z^{\prime } } \right| = R_{\rm{s}} + R_{\rm{ct}} + \sigma \omega^{ - 0.5}$$
(1)

where the Y-axis of −Zre′ represents imaginary part of impedance (Z′), while the X-axis of Zim′ represents real part of impedance (Z′), ω is frequency, and σ is Warburg coefficient. The plot of Zim versus the reciprocal square root of the low angular frequency is given in Fig. 9c, d. The slope of the fitted line is the Warburg coefficient (σ). The calculated values of σ for LFP@NVP@C and LFP@C are given in Table 1. Subsequently, because ionic diffusion coefficient (DLi, cm2·s−1) is inversely proportional to σ, the DLi values can be further calculated by the following Eq. (2):

$$D_{\text{Li}} = R^{2} T^{2} /2A^{2} n^{4} F^{4} c^{2} \sigma^{2}$$
(2)

where R is the gas constant, T is the absolute temperature (K), A is the apparent area of electrode (cm2), n is the number of electrons per molecule during the intercalation (n = 1), F is the Faraday constant, c is the Li+ concentration (mol·cm−3) and σ is the Warburg coefficient. Therefore, the calculated DLi values at various temperatures are shown in Table 1.

Fig. 9
figure 9

EIS characterizations of symmetrical electrode at different temperatures: EIS curves of a LFP@NVP@C and b LFP@C; fitting curves of −Zim and ω−1/2 at low frequency region for c LFP@NVP@C and d LFP@C

Full size image
Table 1 Ru, Rct, σ and DLi values of LFP@C and LFP@NVP@C
Full size table

The variation of the values of DLi is higher for LFP@C than LFP@NVP@C, which is benefited from NVP layer with a stable 3D framework offering sufficient interstitial window to allow fast insertion/desertion of Li+ into the crystal [35]. The reduced charge-transfer resistance and improved insertion/desertion of Li+ bring an enhanced low-temperature performance of LFP@NVP@C.

4 Conclusion

A multi-core–shell-structured LFP@NVP@C composite was successfully synthesized to enhance low-temperature performance of LIBs. NVP demonstrates an open framework for easy insertion/desertion of Li+ into/out of the crystal, and the introduction of NVP interlayer is to modify the surface of LFP crystal by offering more diffusion channels of Li+. As the temperature falls, LFP@NVP@C exhibited an enhanced low-temperature performance by ca. 8%–33% in comparison with LFP@C, which is attributed to reduced charge-transfer resistance and enhanced Li+ diffusion. Even at − 10 °C with 0.5 C, LFP@NVP@C delivered a discharge capacity of ca. 96.9 mAh·g−1 and discharge voltage of ca. 3.3 V. Our results suggest that tuning the surface of cathodes or anodes with materials of multi-dimensional diffusion channels is an effective method to enhance low-temperature performance of LIBs.