Al³⁺-doped FeNb₁₁O₂₉ anode materials with enhanced lithium-storage performance

Abstract

FeNb₁₁O₂₉ has a high theoretical capacity as a potential anode material for lithium-ion batteries; however, the practical capacity of FeNb₁₁O₂₉ reported was unsatisfactory. In this study, a simple and efficient Al³⁺-doping technique was demonstrated to improve the electrochemical performance of FeNb₁₁O₂₉ successfully. The influences of the Al³⁺ doping amount were investigated. It was found that the crystal structure of FeNb₁₁O₂₉ could be preserved when a suitable amount of Al³⁺ was added, and that Al_0.2Fe_0.8Nb₁₁O₂₉ demonstrated better electrochemical performance than FeNb₁₁O₂₉ because the structure of Al_0.2Fe_0.8Nb₁₁O₂₉ is more stable. At 0.1C, Al_0.2Fe_0.8Nb₁₁O₂₉ possessed a high reversible capacity of 318 mAh g⁻¹ with an initial-cycle Coulombic efficiency of 95.0%. Al_0.2Fe_0.8Nb₁₁O₂₉ exhibited outstanding cycling stability with capacity retention of 92.9% at 10C over after 1000 cycles. Moreover, A LiFePO₄/Al_0.2Fe_0.8Nb₁₁O₂₉ full cell was prepared successfully with a discharge capacity of 206 mAh g⁻¹. The full cell exhibited good cycling stability showing the capacity retention of 84.2% over after 200 cycles at 1C and 89.8% over after 1000 cycles at 5C, respectively. This work suggests that Al_0.2Fe_0.8Nb₁₁O₂₉ has great application prospects in lithium-ion batteries.

Graphical abstract

A suitable amount of Al³⁺ doping in FeNb₁₁O₂₉ significantly improves the electrochemical performance (especially the reversible capacity) of FeNb₁₁O₂₉.

1 Introduction

Lithium-ion batteries (LIBs) are regarded as a kind of green power with good cycling performance and high energy density [1,2,3,4,5], which have become one of the hottest research topics these years. At present, the LIB anodes are normally composed of graphite-like materials because of their inexpensive and high theoretical capacity (372 mAh g⁻¹) [6]. However, due to the relatively low rate capability, it is easy to cause the formation of thick solid electrolyte interface (SEI) layers and the growth of lithium dendrites which are hindering the application of graphite-anode materials [7]. Li₄Ti₅O₁₂ has been explored to overcome these disadvantages [8,9,10], which could avoid the formation of thick SEI layers and display good safety performance. However, the low theoretical capacity of Li₄Ti₅O₁₂ (175 mAh g⁻¹) limits its practical application. Therefore, it is necessary to explore novel anode material substitutes with similar advantages to Li₄Ti₅O₁₂ but much higher theoretical capacities.

Recently, niobium oxide-based materials have attracted great attention due to their high theoretical capacities (374–403 mAh g⁻¹) which are benefitted from the Nb⁴⁺/Nb⁵⁺ and Nb³⁺/Nb⁴⁺ multi-electron redox couples. To date, a series of niobium oxide-based anode materials have been explored by our group, such as Nb₂₅O₆₂ [11], MoNb₁₂O₃₃ [12], AlNb₁₁O₂₉ [13], Al_0.5Nb_24.5O₆₂ [14], CrNb₁₁O₂₉ [15], Mg₂Nb₃₄O₈₇ [16], GaNb₁₁O₂₉ [17], TiNb₂₄O₆₂ [18], and ZrNb₂₄O₆₂ [19]. FeNb₁₁O₂₉ [20], W₃Nb₁₄O₄₄ [21], and K₂Nb₈O₂₁ [22] were also found by other researchers to be good anode material candidates. In particularly, FeNb₁₁O₂₉ has a theoretical capacity of 400 mAh g⁻¹ owing to the existence of three Fe²⁺/Fe³⁺, Nb⁴⁺/Nb⁵⁺, and Nb³⁺/Nb⁴⁺ redox couples (each formula unit of FeNb₁₁O₂₉ could transfer 23 electrons) [20]. However, the practical capacities of FeNb₁₁O₂₉ reported were only 168–273 mAh g⁻¹ at 0.1C [20, 23,24,25]. Therefore, it is of great significance to enhance the specific capacity of FeNb₁₁O₂₉ for practical applications.

Currently, rare works were reported for improving the electrochemical performance of FeNb₁₁O₂₉. Zheng et al. increased the specific capacity of FeNb₁₁O₂₉ from 226 to 273 mAh g⁻¹ at 0.1C by fabricating FeNb₁₁O₂₉ nanotubes, and the initial-cycle Coulombic efficiency was improved from 75.9 to 90.1% [20]. Lou et al. prepared Cr³⁺-doped FeNb₁₁O₂₉ (Cr_0.2Fe_0.8Nb₁₁O₂₉) with a specific capacity of 254 mAh g⁻¹ [24]. Lou et al. further enhanced the specific capacity of FeNb₁₁O₂₉ to 270 mAh g⁻¹ through generating oxygen vacancies [25]. In electrochemical experiments, there are common methods used to improve the electrochemical performance of electrode materials, such as carbon composite [26,27,28], plasmonization [29, 30], surfactant-assisted process [31], and nanosizing [32, 33]. It is believed that doping is also a common and efficient method to improve the cycling performance and rate capability of electrode materials because of its easy operation, low-consumption, and large-scale production [34,35,36,37,38,39,40,41,42,43,44,45]. Aluminum (Al) has been paid more and more attention as a substitute for various electrode materials owning to its abundant in earth, nontoxic and light characteristics [46, 47]. It has been found that Al doping could enhance the electrochemical performance of anode materials significantly because the strong Al–O bonds favor the structural stability [48,49,50]. Therefore, in this study, FeNb₁₁O₂₉ materials were also modified by doping with Al. The experimental results showed that Al³⁺ successfully and partically replaced Fe³⁺ and that the obtained Al_0.2Fe_0.8Nb₁₁O₂₉ materials preserved the crystal structure, grain size and morphology of FeNb₁₁O₂₉. Better structural stability was achieved due to the strong Al–O bonds, which greatly enhanced the electrochemical performance of FeNb₁₁O₂₉.

2 Experimental

2.1 Material synthesis

The Al_xFe_1-xNb₁₁O₂₉ (x = 0.1, 0.2, and 0.3) materials were synthesized via a simple solid-state reaction method by using Nb₂O₅ (Sinopharm, 99.9%), Al₂O₃ (Aladdin, 99.0%), and Fe₂O₃ (Aladdin, 99.0%) as the raw materials. Nb₂O₅, Fe₂O₃, and different contents of Al₂O₃ were mixed in ethanol, which were then ground in a high-energy ball-milling machine (SPEX 8000 M, USA) for 4 h. After drying, the ball-milled mixtures were sintered in a muffle furnace at 1300 °C in air for 4 h, forming Al_xFe_1-xNb₁₁O₂₉ materials (x = 0.1, 0.2, and 0.3). For comparison, undoped FeNb₁₁O₂₉ counterpart was also fabricated with a similar procedure and without the use of Al₂O₃.

2.2 Material characterization

The crystalline structures of the obtained samples were characterized by X-ray diffraction (XRD) on an X-ray diffractometer (German Bruker D8). The phase purity and lattice parameters were determined by Rietveld refinements, which were conducted by using the GSAS program with the EXPGUI interface [51, 52]. Morphologies, particle sizes, and microstructures of synthesized samples were recorded by scanning electron microscopy (SEM, S-4800), high-resolution transmission electron microscopy (HRTEM, Tecnai G2 F20 S-TWIN, FEI), and energy dispersive X-ray spectrometer (EDX) mapping. The specific surface areas of the samples were determined by nitrogen adsorption/desorption device (ASAP 2460).

2.3 Half-cell measurement

The electrochemical performance of Al_xFe_1-xNb₁₁O₂₉/Li (x = 0, 0.1, and 0.2) was evaluated by employing CR2016-type coin cells. Firstly, the working electrodes were prepared by mixing 70 wt% of Al_xFe_1-xNb₁₁O₂₉ (active material) powder, 20 wt% of conductive carbon (Super P, battery level), and 10 wt% of polyvinylidene fluoride (PVDF) binder, thoroughly mixed in N-methylpyrrolidone (NMP) to form a homogeneous slurry. After stirring the slurry for 8 h, the slurry was coated onto a copper foil evenly. The mass loading of Al_xFe_1-xNb₁₁O₂₉ was ~ 1.0 mg cm⁻². Then, the coated foils were dried in a vacuum oven at 120 °C for 10 h and were cut into a circular plate with a diameter of 10 mm. The CR2016 coin cells were assembled in a glove box filled with argon gas (O₂/H₂O < 0.1 ppm) and consisted of an above as-prepared electrode, a Li foil that acted as counter and reference electrodes, a microporous polypropylene film (Celgard 2325) and electrolyte containing 1 M LiPF₆ (DAN VEC) in a mixed solvent of ethylene carbonate (EC), dimethyl carbonate (DMC), and diethylene carbonate (DEC) with a volume ratio of 1:1:1. Galvanostatic discharge–charge measurements and galvanostatic intermittent titration technique (GITT) tests were performed on a Neware battery tester. Cyclic voltammetry (CV) measurements were conducted on a CHI660E electrochemical workstation. The electrochemical performance was examined within 3.0–0.8 V. To prepare the Al_xFe_1-xNb₁₁O₂₉ electrodes for ex situ XRD, the half cells at different states of discharge/charge were disassembled in the glove box, followed by washing the obtained Al_xFe_1-xNb₁₁O₂₉ electrodes by dimethyl carbonate.

2.4 Full-cell measurement

Full cells (CR2032-type coin cells) were assembled with the as-prepared Al_0.2Fe_0.8Nb₁₁O₂₉ as the anode and commercial LiFePO₄ as the cathode. Similarly, LiFePO₄ powder, conductive carbon (Super P), and PVDF with a weight ratio of 7:2:1 were thoroughly mixed in N-methylpyrrolidone (NMP) to form a homogeneous slurry. Then, the slurry was spread evenly on an Al foil and made into a circular plate with a diameter of 10 mm. The LiFePO₄:Al_0.2Fe_0.8Nb₁₁O₂₉ weight ratio was fixed to be 2:1. Galvanostatic discharge–charge measurements were performed within 1.0–2.5 V.

3 Results and discussion

Figure 1 shows the XRD patterns of FeNb₁₁O₂₉ and Al_xFe_1-xNb₁₁O₂₉ (x = 0.1, 0.2, and 0.3), and the corresponding Rietveld-refined data are demonstrated in Table S1. It can be seen that the XRD patterns of Al_xFe_1-xNb₁₁O₂₉ (x = 0.1 and 0.2) completely match those of FeNb₁₁O₂₉ with an orthorhombic shear ReO₃ crystal structure and Amma space group (JCPDS#22–0352; Fig. S1). No other phases (such as Nb_xO_y or Fe₂O₃) are observed. Table S1 shows that the weighted profile residual (R_wp) values of the samples are small (about 10%), revealing that the Rietveld refinement results are credible. The fractional atomic parameters of FeNb₁₁O₂₉ and Al_0.2Fe_0.8Nb₁₁O₂₉ are shown in Tables S2 and S3, respectively. Since the ionic radius of Al³⁺ is smaller than Fe³⁺, it is reasonable that the lattice parameters get smaller [45, 53]. It is noteworthy that the b values of Al_xFe_1-xNb₁₁O₂₉ (x = 0.1 and 0.2) are larger than that of the previous Ti₂Nb₁₀O₂₉ (b = 3.805 Å) reported [54], suggesting the existence of larger gap sites in Al_xFe_1-xNb₁₁O₂₉ that are more conducive to Li⁺ diffusion. However, when the Al³⁺ doping amount further increases (x = 0.3), the crystal structure of orthorhombic shear ReO₃ changes, indicating that the allowable Al³⁺ doping amount is limited (i.e., x ≤ 0.2; Fig. 1) [53, 55].

Fig. 1

XRD patterns and Rietveld refinement results of FeNb₁₁O₂₉ and Al_xFe_1-xNb₁₁O₂₉ (x = 0.1, 0.2, and 0.3) materials

Figure 2 illustrates the SEM images of FeNb₁₁O₂₉ and Al_xFe_1-xNb₁₁O₂₉ (x = 0.1, 0.2, and 0.3). It can be observed that all the samples consist of grains which possess a platelet morphology. The BET-specific surface area of Al_0.2Fe_0.8Nb₁₁O₂₉ is 0.27 m² g⁻¹ (Fig. S2b), which is slightly smaller than FeNb₁₁O₂₉ (0.24 m² g⁻¹; Fig. S2a), verifying that Al_0.2Fe_0.8Nb₁₁O₂₉ and FeNb₁₁O₂₉ have similar particle sizes. Thus, there are almost little influences on the grain morphology and size of FeNb₁₁O₂₉ through the Al³⁺ doping.

Fig. 2

SEM images of a FeNb₁₁O₂₉, b Al_0.1Fe_0.9Nb₁₁O₂₉, c Al_0.2Fe_0.8Nb₁₁O₂₉, and d Al_0.3Fe_0.7Nb₁₁O₂₉

Figure 3 depicts the HRTEM images of FeNb₁₁O₂₉ (Fig. 3a) and Al_0.2Fe_0.8Nb₁₁O₂₉ (Fig. 3b), revealing their lattice fringes of 0.351 and 0.349 nm, respectively, which correspond to (311) planes. The interplanar spacing contraction in Al_0.2Fe_0.8Nb₁₁O₂₉ may be due to the smaller ionic radius of Al³⁺ in comparison with Fe³⁺. The results further verify that the crystal structure of FeNb₁₁O₂₉ does not change after doping with a suitable amount of Al³⁺, which are also consistent with the XRD characterization. The EDS elemental mapping images shown in Fig. 3c, d indicate that Fe, Nb, and O elements, and Al, Fe, Nb, and O elements, are uniformly distributed in FeNb₁₁O₂₉ and Al_0.2Fe_0.8Nb₁₁O₂₉ materials, respectively, further confirming the successful doping of Al³⁺ into FeNb₁₁O₂₉.

Fig. 3

HRTEM images of a FeNb₁₁O₂₉ and b Al_0.2Fe_0.8Nb₁₁O₂₉. EDX elemental mapping images of c FeNb₁₁O₂₉ and d Al_0.2Fe_0.8Nb₁₁O₂₉

In order to understand the electrochemical mechanism of FeNb₁₁O₂₉ and Al_0.2Fe_0.8Nb₁₁O₂₉, CV tests with different scanning speeds were carried out on the FeNb₁₁O₂₉ and Al_0.2Fe_0.8Nb₁₁O₂₉-made electrodes under a potential window of 3.0–0.8 V. Figure 4a, b show the first four-cycle CV curves of these two electrodes tested at 0.2 mV s⁻¹. The redox peak of the FeNb₁₁O₂₉ electrode can be attributed to Nb³⁺/Nb⁴⁺, Nb⁴⁺/Nb⁵⁺, and Fe²⁺/Fe³⁺ redox pairs (Fig. 4a). In particular, the Nb³⁺/Nb⁴⁺ peak is at ~ 1.10/1.30 V, the Nb⁴⁺/Nb⁵⁺ peak is at ~ 1.53/1.72 V, and the Fe²⁺/Fe³⁺ peak is at ~ 2.36 V. The average working potential of FeNb₁₁O₂₉ is calculated to be ~ 1.61 V. For the Al_0.2Fe_0.8Nb₁₁O₂₉ electrode (Fig. 4b), the Nb³⁺/Nb⁴⁺ peak is at ~ 1.10/1.30 V, the Nb⁴⁺/Nb⁵⁺ peak is at ~ 1.59/1.72 V, and the Fe²⁺/Fe³⁺ peak is at ~ 2.35 V. The average working potential of Al_0.2Fe_0.8Nb₁₁O₂₉ is also ~ 1.61 V. Clearly, Al_0.2Fe_0.8Nb₁₁O₂₉ and FeNb₁₁O₂₉ have a very similar CV characteristic, hinting the similar reaction kinetics.

Fig. 4

The first four CV cycles of a FeNb₁₁O₂₉ electrode and b Al_0.2Fe_0.8Nb₁₁O₂₉ electrode at 0.2 mV s⁻¹. The CV curves of c FeNb₁₁O₂₉ electrode and d Al_0.2Fe_0.8Nb₁₁O₂₉ electrode at different scan rates

Figure 4c, d illustrate the CV curves of FeNb₁₁O₂₉ and Al_0.2Fe_0.8Nb₁₁O₂₉ electrodes at different scanning speeds, respectively. It can be seen that with the increase of the scanning speeds, the current values of the redox front increase in both FeNb₁₁O₂₉ and Al_0.2Fe_0.8Nb₁₁O₂₉ electrodes. However, compared with FeNb₁₁O₂₉, Al_0.2Fe_0.8Nb₁₁O₂₉ has a stronger redox front, revealing that Al³⁺ doping enables better electrochemical-kinetics behavior.

The constant current discharge–charge curves of the FeNb₁₁O₂₉ and Al_0.2Fe_0.8Nb₁₁O₂₉ electrodes are demonstrated in Fig. 5a, b, respectively. Each discharge–charge curve was divided into three regions: a short slope at 3.0–1.7 V, a short platform at 1.7–1.6 V, and a long slope at 1.6–0.8 V. Solid solution reactions occur in the first and third regions, and a two-phase reaction occurs in the second region. It can be found that the FeNb₁₁O₂₉ electrode has a reversible capacity of 262 mAh g⁻¹ with an initial-cycle Coulombic efficiency of 91.1%, while the Al_0.2Fe_0.8Nb₁₁O₂₉ electrode has an enhanced reversible capacity of 318 mAh g⁻¹ with an initial-cycle Coulombic efficiency of 95.0%. Table 1 lists the electrochemical performance of the Al_0.2Fe_0.8Nb₁₁O₂₉ and M-Nb–O materials reported in literatures. It is clear that the capacity of Al_0.2Fe_0.8Nb₁₁O₂₉ is higher than those of most reported M-Nb–O materials (except for MoNb₁₂O₃₃ and Mg₂Nb₃₄O₈₇), and the initial-cycle Coulombic efficiency is only lower than GaNb₁₁O₂₉. It is believed that the relatively high working potential of Al_0.2Fe_0.8Nb₁₁O₂₉ (~ 1.61 V) results in the less formation of SEI layer during the electrochemical reaction process, which reduces the consumption of lithium ions and leads to the high initial-cycle Coulombic efficiency [7].

Fig. 5

Electrochemical performance of FeNb₁₁O₂₉ and Al_0.2Fe_0.8Nb₁₁O₂₉ electrodes: a, b discharge–charge curves at 0.1C, c, d discharge–charge curves at 0.1–10C, e rate capabilities, and f cycling stability at 10C after 1000 cycles and Coulombic efficiency of Al_0.2Fe_0.8Nb₁₁O₂₉

Table 1 Electrochemical performance of Al_0.2Fe_0.8Nb₁₁O₂₉ compared with M-Nb–O materials reported

Figure 5c, d show the discharge–charge curves of the FeNb₁₁O₂₉ and Al_0.2Fe_0.8Nb₁₁O₂₉ electrodes at different current density, respectively. When the current density increases from 0.1 to 10C, the specific capacity of FeNb₁₁O₂₉ decreases from 262 to 104 mAh g⁻¹, while that of Al_0.2Fe_0.8Nb₁₁O₂₉ electrode decreases from 318 to 134 mAh g⁻¹. Figure 5e demonstrates the rate capabilities of the FeNb₁₁O₂₉ and Al_0.2Fe_0.8Nb₁₁O₂₉ electrodes at various current densities. Both the specific capacities of the FeNb₁₁O₂₉ and Al_0.2Fe_0.8Nb₁₁O₂₉ electrodes reduce with increasing the current density. It can also be found that the specific capacity of the Al_0.2Fe_0.8Nb₁₁O₂₉ electrode at 10C is even higher than the FeNb₁₁O₂₉ electrode at 5C. Therefore, the Al_0.2Fe_0.8Nb₁₁O₂₉ electrode has achieved a better rate capability. This high rate capability is due to the fact that Al_0.2Fe_0.8Nb₁₁O₂₉ has a stronger redox front. Moreover, Fig. S3a, b illustrate the discharge–charge curves of the Al_0.1Fe_0.9Nb₁₁O₂₉ electrode. The Al_0.1Fe_0.9Nb₁₁O₂₉ electrode also exhibits an improved specific capacity (272 mAh g⁻¹) in comparison with the undoped FeNb₁₁O₂₉ electrode, but lower than that of the Al_0.2Fe_0.8Nb₁₁O₂₉ electrode. This result indicates that more Al³⁺ doping has a more significant effect on improving the reversible capacity of FeNb₁₁O₂₉.

Figure 5f illustrates the cycling performance of the FeNb₁₁O₂₉ and Al_0.2Fe_0.8Nb₁₁O₂₉ electrodes at a current density of 10C. After 1000 cycles, the capacity retention rate of Al_0.2Fe_0.8Nb₁₁O₂₉ is 92.9%, which is obviously higher than that of FeNb₁₁O₂₉ (85.8%), proving that Al³⁺ doping is very beneficial to the cycling stability of FeNb₁₁O₂₉. The excellent cycling stability is due to the fact that Al_0.2Fe_0.8Nb₁₁O₂₉ owns excellent structural stability.

To study the Li⁺ diffusivity of the FeNb₁₁O₂₉ and Al_0.2Fe_0.8Nb₁₁O₂₉ electrodes, the GITT tests were employed at room temperature (Fig. 6). Figure 6a, b record the initial GITT curves of the FeNb₁₁O₂₉ and Al_0.2Fe_0.8Nb₁₁O₂₉ electrodes at 0.1C, respectively. The diffusion coefficient of Li⁺$({D}_{L{i}^{+}})$ can be calculated according to Fick’s second law (Eq. (1)):

Fig. 6

GITT initial curves of a FeNb₁₁O₂₉ electrode and b Al_0.2Fe_0.8Nb₁₁O₂₉ electrode tested at 0.1C. Calculated ${D}_{L{i}^{+}}$ from GITT curves of c FeNb₁₁O₂₉ electrode and d Al_0.2Fe_0.8Nb₁₁O₂₉ electrode

$$D_{Li^+}=\frac4\pi\left(\frac{m_BV_m}{M_BS}\right)^2\left(\frac{\Delta E_s}{\tau\left(dE_\tau/d\sqrt\tau\right)}\right)^2(\tau<<\frac{L^2}{D^2})$$

(1)

$$D_{Li^+}=\frac {4}{\pi\tau} \left(\frac{m_{B}V_{m}}{M_{B}S}\right)^2\left(\frac{\Delta E_{s}}{\Delta E_{\tau}}\right)^2 (\tau << \frac{L^2}{D^2})$$

(2)

where M_B and m_B are the molar mass and unit mass of the active material, respectively; V_m stands the molar volume of the active material; S stands the polar area; τ is titration time; L stands for the diffusion length; and ΔE_S and ΔE_τ represent the change of equilibrium potential and the change in potential during a single-step titration (Fig. S4a, b), respectively. Since there is a linear relationship between τ^0.5 and the potential during the single-step titration (Fig. S4c, d), Eq. (1) can be simplified as Eq. (2). Figure 6c, d describe how the calculated ${D}_{\text{L}{\text{i}}^{+}}$ values vary with the potential. During the Li⁺ insertion process, the calculated ${D}_{\text{L}{\text{i}}^{+}}$ values of the FeNb₁₁O₂₉ electrode are 8.03 × 10⁻¹³–3.71 × 10⁻¹¹cm² s⁻¹ with an average value of 1.47 × 10⁻¹¹ cm² s⁻¹, while the Al_0.2Fe_0.8Nb₁₁O₂₉ electrode has calculated ${D}_{\text{L}{\text{i}}^{+}}$ values of 1.67 × 10⁻¹²–3.41 × 10⁻¹¹ cm² s⁻¹ with an average value of 1.76 × 10⁻¹¹ cm² s⁻¹. For the Li⁺ extraction process, the calculated ${D}_{\text{L}{\text{i}}^{+}}$ values of the FeNb₁₁O₂₉ electrode are 3.95 × 10⁻¹²–1.83 × 10⁻¹¹ cm² s⁻¹ with an average value of 1.28 × 10⁻¹¹ cm² s⁻¹, while the Al_0.2Fe_0.8Nb₁₁O₂₉ electrode has calculated ${D}_{\text{L}{\text{i}}^{+}}$ values of 7.24 × 10⁻¹²–2.60 × 10⁻¹¹ cm² s⁻¹ with an average value of 1.64 × 10⁻¹¹ cm² s⁻¹. These similar Li⁺ diffusivity can be explained by the similar lattice parameters and unit-cell volumes of the two materials (Table S1).

The structural stability of Al_0.2Fe_0.8Nb₁₁O₂₉ during Li⁺ insertion-extraction processes was studied by XRD. Figure 7 demonstrates the XRD results of the fresh Al_0.2Fe_0.8Nb₁₁O₂₉ electrode and the Al_0.2Fe_0.8Nb₁₁O₂₉ electrodes after discharged to 0.8 V as well as charged to 3.0 V in 1st cycle and 10th cycles at 0.1C. It can be found that after the discharge–charge process, no impurity phases (such as Al₂O₃, Fe₂O₃, or Nb_xO_y) appear in the XRD patterns, and no obvious changes are observed in the peak intensities. The XRD patterns of the FeNb₁₁O₂₉ electrodes after the similar discharge–charge process are shown in Fig. S5. Very similar results are obtained. Therefore, Al³⁺ doping does not change the crystal structure of FeNb₁₁O₂₉ material indeed, and the Al_0.2Fe_0.8Nb₁₁O₂₉ framework is very stable during the electrochemical reaction.

Fig. 7

Ex situ XRD patterns of fresh Al_0.2Fe_0.8Nb₁₁O₂₉ electrode and Al_0.2Fe_0.8Nb₁₁O₂₉ electrodes after discharged to 0.8 V in 1st cycle, charged to 3.0 V in 1st and 10th cycles at 0.1C

To assess Al_0.2Fe_0.8Nb₁₁O₂₉ practical application value, we prepared a LiFePO₄/Al_0.2Fe_0.8Nb₁₁O₂₉ full cell with LiFePO₄ as the cathode material and Al_0.2Fe_0.8Nb₁₁O₂₉ as the anode material. Figure 8a shows the discharge–charge curves of the full cell at 0.1C. It delivered initial-cycle charge/discharge capacities of 241/206 mAh g⁻¹, respectively. Its initial-cycle Coulombic efficiency was 85%. This result is better than the previously reported LiNi_0.5Mn_1.5O₄/AlNb₁₁O₂₉ full cell with charge/discharge capacities of only 238/195 mAh g⁻¹ and an initial-cycle Coulombic efficiency of 82% [13]. Figure 8b, c show the discharge–charge curves and rate capability of the LiPFeO₄/Al_0.2Fe_0.8Nb₁₁O₂₉ full cell at different current density from 0.5 to 5C. The discharge capacities are 172, 152, 125, and 83 mAh g⁻¹ at 0.5, 1, 2, and 5C, respectively. In particular, compared with the original current density at 0.5C, there is no significant decrease appeared in the rate-capability curve when the current density is from 5C back to 0.5C (Fig. 8c). It indicates that the full cell has good electrochemical reversibility. Figure 8d, e show that the capacity retention of the full cell reaches 84.2% after 200 cycles at 1C, and 89.8% after 1000 cycles at 5C, respectively, demonstrating good cycling stability.

Fig. 8

Electrochemical performance of the LiFePO₄/Al_0.2Fe_0.8Nb₁₁O₂₉ full cell: a discharge–charge curves at 0.1C, b discharge–charge curves at 0.1-5C, c rate capabilities, d cycling stability at 1C over 200 cycles and Coulombic efficiency, and e cycling stability at 10C over 1000 cycles and Coulombic efficiency

4 Conclusions

In summary, the electrochemical performance (especially the specific capacity) of FeNb₁₁O₂₉ is improved significantly by doping with Al³⁺. A reasonable doping amount can preserve the crystal structure, grain size, and morphology of FeNb₁₁O₂₉. Al_0.2Fe_0.8Nb₁₁O₂₉ shows improved electrochemical performance, with a reversible capacity of 318 mAh g⁻¹ at 0.1 C, high initial-cycle Coulombic efficiency of 95.0%, and relatively high and safe working potential of about 1.61 V. Al_0.2Fe_0.8Nb₁₁O₂₉ further possesses excellent cycling stability with capacity retention of 92.9% over after 1000 cycles. In addition, we assembled a LiFePO₄/Al_0.2Fe_0.8Nb₁₁O₂₉ full cell, which also show excellent electrochemical performance. The discharge capacity is up to 206 mAh g⁻¹ at 0.1C. The capacity retention of the full cell reaches 84.2% after 200 cycles at 1C, and 89.8% after 1000 cycles at 5C. Therefore, Al_0.2Fe_0.8Nb₁₁O₂₉ may find practical applications in LIBs.