Porous Nb₄W₇O₃₁ microspheres with a mixed crystal structure for high-performance Li⁺ storage

INTRODUCTION

Lithium-ion batteries (LIBs), as typical representatives of secondary batteries, have dominated power sources for consumable electronics^[1-5]. To satisfy the booming development need for electric vehicles, the exploration of high-performance Li⁺-storage materials with high reversible capacities, safe operation, superior rate capabilities, excellent cyclic stability, and high first-cycle Coulombic efficiencies is highly necessary. As for the anode materials, the intercalation type shows much better cyclic stability than the alloying and conversion types, being more practical for commercial LIBs^[6]. At present, the most popular anode material is based on intercalation-type graphite owing to its large specific capacity (theoretically 372 mAh g^-1) and low cost. However, graphite faces a safety issue of lithium-dendrite formation when fast discharged/charged at its extremely low potential plateau (< 0.1 V vs. Li/Li⁺)^[7]. Intercalation-type Li₄Ti₅O₁₂, which is the second most popular anode material, exhibits a safe and flat potential plateau (~1.55 V), avoiding the above safety issue. Nevertheless, its theoretical capacity is insufficient (175 mAh g^-1 within 1-3 V), hindering its appliance to some extent^[8]. Hence, it is desirable to explore new, practical, and intercalation-type anode materials that simultaneously possess large reversible capacities and high safety performance.

Recently, niobium-tungsten oxide anode materials Nb₁₆W₅O₅₅ and Nb₁₈W₁₆O₉₃, with both large reversible capacities and high safety operation, were reported by Griffith et al.^[5]. Owing to the redox couples Nb⁵⁺/Nb⁴⁺, Nb⁴⁺/Nb³⁺, W⁶⁺/W⁵⁺, and W⁵⁺/W⁴⁺ with two-electron transfer per niobium/tungsten, these niobium-tungsten oxides show significantly larger theoretical/practical capacities than Li₄Ti₅O₁₂. Meanwhile, these four redox couples show safe operating potentials within 0.5-2.0 V, avoiding the lithium-dendrite formation and thus guaranteeing the safe operation. The niobium-tungsten oxides display open crystal structures^[9-15], such as confined ReO₃ structures in Nb₁₆W₅O₅₅ and tetragonal tungsten bronze (TTB) structures in Nb₁₈W₁₆O₉₃, enabling fast Li⁺ diffusivity for high rate performance. The 4 × 4 ReO₃-type blocks constructed by MO₆ octahedra (M = Nb/W) in Nb₁₆W₅O₅₅ are confined through edge sharing of some MO₆ octahedra and corner sharing of WO₄ tetrahedra, which guarantees good structural stability. The TTB structure is composed of pentagonal columns containing MO₇ pentagonal bipyramids by sharing equatorial edges with five MO₆ octahedra per bipyramid, linked in such a way that pentagonal, quadrangular, and triangular tunnels are formed. Compared to the confined ReO₃ structure, the TTB structure owns a larger interlayer spacing and numerous larger tunnels^[16,17], which benefit Li⁺ transport but harm structural stability. Therefore, it can be expected that niobium-tungsten oxide anode materials combining the above two structures can simultaneously have the merits of excellent structural robustness and superior Li⁺ transport. Such niobium-tungsten oxide anode materials with mixed structures, however, have not been reported so far.

Here, porous Nb₄W₇O₃₁ microspheres with a mixed TTB and confined ReO₃ structure are synthesized based on a solvothermal process and applied as a new anode material for LIBs. The synthesis, crystal structures, electrochemical properties, and working mechanisms are systematically studied through various characterizations. As expected, this material exhibits superior Li⁺ diffusion coefficients (D_Li) (1.24 × 10^-11 cm² s^-1 during lithiation and 1.09 × 10^-10 cm² s^-1 during delithiation), outstanding rate capability (10C vs. 0.1C capacity retention percentage of 47.6%), and excellent cyclic stability (95.2% capacity retention after 1,500 cycles at 10C). Furthermore, it shows a large reversible capacity (252 mAh g^-1 at 0.1C), high first-cycle Coulombic efficiency (88.4% at 0.1C), and safe operating potential (~1.66 V). During lithiation, external Li⁺ ions are confirmed to be intercalated into the pentagonal and quadrangular tunnels of the mixed structure, homogeneously and slightly expanding the Nb₄W₇O₃₁ lattice in three dimensions, and this process is highly reversible.

RESULTS AND DISCUSSION

The reflections in the X-ray diffraction (XRD) pattern [Supplementary Figure 1] correspond to the characteristic peaks of Nb₄W₇O₃₁ (JCPDS 20-1320) except for three weak reflections belonging to tiny WO₃ impurity (JCPDS 43-1035). However, the crystal structure of Nb₄W₇O₃₁ has been in debate. The structure from JCPDS 20-1320, as proposed by N.C. Stephenson^[18], is a TTB structure with the P4 space group (Structure A, Figure 1A). It comprises NbO₆ and WO₆ octahedra in the Nb₄W₇O₃₁ framework with the characteristic of Nb/W mixed arrangement. The numbers of the pentagonal, quadrangular, and triangular prisms are 16, 9, and 16 per unit cell, respectively. A quarter of the pentagonal prisms are completely filled by the -O-Nb/W-O- strings, and 12 pentagonal tunnels, nine quadrangular tunnels, and 16 triangular tunnels [Supplementary Figure 2] in the overall cubic close-packed array are completely vacant. From the energy point of view, Structure B [Figure 1B], which is another TTB structure with filled pentagonal prisms at different positions, is also reasonable since the distance between the nearest filled pentagonal prisms (0.490a) is slightly larger than that in Structure A (0.486a). Additionally, Iijima proposed another reasonable structure for Nb₄W₇O₃₁ (Structure C, Figure 1C)^[19]. A 4 × 4 ReO₃-type block in the middle of the unit cell is confined by a TTB-type matrix, resulting in abundant ion tunnels along the c direction. The numbers of the pentagonal, quadrangular, and triangular prisms are 12, 16, and 12, respectively, and a third of the pentagonal prisms are entirely filled by the -O-Nb/W-O- strings. It is known that the TTB structures tend to lose their stability when less than one-third of pentagonal prisms are filled^[17]. Therefore, compared with Structure A and Structure B, Structure C is more stable. In addition, the total number of the large-sized pentagonal and quadrangular tunnels is larger (24 for Structure C vs. 21 for Structure A and Structure B), enabling faster Li⁺ transport within the Nb₄W₇O₃₁ lattice. The XRD pattern is reasonably Rietveld-refined using a multiple-phase method (weighted profile residual: 9.08%)^[20,21], showing that the mass percentages of Structure A, Structure B, and Structure C are 21.7(3), 22.1(3), and 56.2(2)%, respectively [Figure 1D]. The refined lattice constants are tabulated in Supplementary Table 1. It is noteworthy that the interlayer spacing of Nb₄W₇O₃₁ (equal to the c value) surpasses those of the reported niobate anode materials with confined ReO₃ and tungsten bronze structures [Supplementary Table 1], which undoubtedly benefits the Li⁺ transport within the Nb₄W₇O₃₁ lattice.

Figure 1. Crystal-structure units of Nb₄W₇O₃₁: (A) Structure A with a TTB type, (B) Structure B with another TTB type, and (C) Structure C with a confined ReO₃ type. (D) Rietveld-refined XRD pattern of Nb₄W₇O₃₁. The peaks of WO₃ at 23.8, 24.5, and 33.8° are removed during the refinement. For clarity, not all peaks of Nb₄W₇O₃₁ are indexed.

The synthesized Nb₄W₇O₃₁ sample exhibits a porous-microspherical morphology with sphere diameters ranging from 2 μm to 7 μm [Figure 2A and B]. It is seen that nanorods with diameters between 20 nm and 100 nm and lengths between 60 nm and 1.5 μm are the primary building blocks, which are densely packed to form the porous microspheres [Supplementary Figure 3]. The heterogeneous distribution of nanorods with different sizes can enable a large tap density of the porous microspheres^[22]. Each nanorod is a single crystal [Figure 2C] with its [001] crystal orientation along its longitudinal direction [Figure 2D]. The diffraction-spot deformation in the SAED pattern indicates the existence of abundant defects in the crystal, which is very reasonable since mixed crystal structures always contain defects. The energy-dispersive X-ray spectroscopy (EDX) mapping images in Figure 2E show that the Nb, W, and O elements are well dispersed throughout the tested porous microsphere. Figure 2F presents the N₂ adsorption-desorption isotherm of Nb₄W₇O₃₁, revealing its hierarchically porous characteristic with abundant pores centered at 3.8 and 71 nm [Figure 2F], which respectively correspond to the inter-rod and inter-sphere pores. The Branauer-Emmett-Teller (BET) specific surface area is calculated to be a large value of 11.4 m² g^-1. As a whole, the obtained porous microspheres constructed by nanorods can offer fast electron transport along the nanorods, provide short Li⁺ transport distances within the primary nanorods, give a large electrochemical-reaction area, enable easy electrolyte penetration, and guarantee a large tap density^[23-27], being an ideal morphology for high-performance electrode materials.

Figure 2. (A) FESEM image, (B) TEM image, (C) SAED pattern, (D) HRTEM image, (E) EDX mapping images, and (F) N₂ adsorption-desorption isotherm (inset: BJH pore-size distribution curve from desorption branch) of porous Nb₄W₇O₃₁ microspheres.

To study the formation mechanism of the porous microspheres, three solvothermal experiments using different precursors are carried out [Supplementary Figure 4], and their final products are compared. First, when a precursor solution containing 0.86 mmol of H₂₈N₆O₄₁W₁₂ and 5.3 mmol of NbCl₅ dissolved in 60 mL isopropyl alcohol (IPA) is used, only nanoparticles can be found in the final product. Second, when 2 g tetrabutylammonium bromide (TBA) is added to the above precursor solution, only nanorods form. The proper amount of TBA can selectively bind to the (001) oriented surfaces of the Nb₄W₇O₃₁ crystals, thus lowing the energy density of these surfaces and eventually guiding the growth of Nb₄W₇O₃₁ nanorods along the [001] crystal orientation^[28]. Third, when 8 g hexadecyl trimethyl ammonium bromide (CTMAB) and 3 mL hydrochloric acid are further added into the precursor solution, the nanorods are self-assembled to yield hierarchical porous microspheres. This self-assembly process is accomplished with the aid of the CTMAB surfactant with a long, flexible, and non-branched hydrophobic chain^[29]. CTMAB may modify the surface interaction between the Nb₄W₇O₃₁ nanorods and the acid solution, increasing the surface free energy of the nanorods as the liquid-solid phase separation proceeds, and thus favoring the self-assembly of nanorods into porous microspheres based on the lowest energy principle^[30-32]. After this self-assembly, Ostwald ripening proceeds, during which large nanorods become larger while small nanorods become smaller in the porous microspheres. Clearly, the addition of TBA, CTMAB, and hydrochloric acid in proper amounts is the key to achieving the novel porous microspheres assembled by nanorods.

Figure 3A shows the galvanostatic discharge-charge curves of the Nb₄W₇O₃₁/Li half cell at current rates of 0.1C-10C (1C = 274 mA g^-1) in the potential window of 0.8-3.0 V. Nb₄W₇O₃₁ delivers a large first-cycle discharge/charge capacity of 285/252 mAh g^-1 with a high Coulombic efficiency of 88.4% at 0.1C. The average operating potential is ~1.66 V during lithiation/delithiation, which is similar to those of the popular Li₄Ti₅O₁₂ and TiNb₂O₇ and suggests the high safety performance of Nb₄W₇O₃₁^[8,33]. Each discharge/charge curve of Nb₄W₇O₃₁ presents two sloping regions and no plateaus, reflecting two solid-solution reactions throughout lithiation/delithiation, which is different from the two-phase-reaction involved behavior in Li₄Ti₅O₁₂ and TiNb₂O₇. The absence of two-phase reactions in Nb₄W₇O₃₁ is beneficial for its electrochemical kinetics since it is known that the Li⁺ diffusion during two-phase reactions is relatively slow^[34]. Additionally, the rate capability of Nb₄W₇O₃₁ is remarkable. With the current rate increasing, Nb₄W₇O₃₁ retains large reversible capacities of 235, 196, 171, 144, and 120 mAh g^-1, respectively at 0.5C, 1C, 2C, 5C, and 10C [Figure 3B]. As the current rate turns back to 0.5C, the capacity has no significant decay. Long-term cycling tests are further performed, revealing 97.7% capacity retention at 1C after 100 cycles [Figure 3C and Supplementary Figure 5] and 95.2% retention at 10C after 1,500 cycles [Figure 3D], and the porous-microspherical morphology can mainly be retained during the cycling [Supplementary Figure 6]. This superior cyclic stability of Nb₄W₇O₃₁ is among the best results obtained from the reported anode materials with the intercalation characteristic [Supplementary Table 2].

Figure 3. Electrochemical properties of Nb₄W₇O₃₁/Li half cell: (A) discharge/charge curves at various current rates, (B) rate capability, and cyclic stability at (C) 1C and (D) 10C (after rate-capability test). Electrochemical properties of LiFePO₄//Nb₄W₇O₃₁ full cell: (E) charge/discharge curves at various current rates, (F) rate capability, and (G) cyclic stability at 1C (after rate-capability test, inset: LED lit up by full cell). (H) dQ/dV curves of Nb₄W₇O₃₁/Li half cell at 0.1C.

To prove the practicability of Nb₄W₇O₃₁, it is coupled with a LiFePO₄ cathode, and the resultant full cell is studied. Its average working voltage is ~1.75 V [Figure 3E], matching well with the operating potentials of both LiFePO₄ [Supplementary Figure 7] and Nb₄W₇O₃₁. It exhibits a large reversible capacity of 200 mAh g^-1 at 0.1C. The reversible capacity remains 180, 156, 110, and 87 mAh g^-1 at 0.5C, 1C, 2C, and 5C, respectively [Figure 3F]. It remains at 136 mAh g^-1 with high-capacity retention of 88.1% after 200 cycles at 1C [Figure 3G]. In addition, a bright LED bulb can be lit up by the full cell after enduring long-term cycling [Figure 3G]. Although the gravimetric energy density of Nb₄W₇O₃₁-based full cells is lower than that of traditional graphite-based full cells, Nb₄W₇O₃₁ owns several merits compared with graphite, including inherent safety performance, high rate capability, and good cyclic stability. Therefore, Nb₄W₇O₃₁ is a very promising anode material for safe, fast-charging, and long-life LIBs.

In order to know why Nb₄W₇O₃₁ owns comprehensively good electrochemical properties, several in-depth characterizations are conducted. The dQ/dV curves of the Nb₄W₇O₃₁/Li half cell in Figure 3H are deduced from its galvanostatic discharge-charge curves at 0.1C, revealing the redox mechanism of Nb₄W₇O₃₁. During the first cycle, there exist three peak pairs, respectively located at ~2.1/~2.2, ~1.5/~1.7, and ~1.1/~1.2 V. The first pair could be assigned to the redox reaction from the W⁶⁺/W⁵⁺ redox couples, the second pair could be assigned to the Nb⁵⁺/Nb⁴⁺ couple, and the third pair could be rooted in the combination of Nb⁴⁺/Nb³⁺ and W⁵⁺/W⁴⁺ couples^[35-37]. The subsequent sweep, however, is slightly different from the first one, which can be ascribed to the irreversible polarization, incomplete delithiation, and thin SEI-film formation during the first cycle^[38]. The survey X-ray photoelectron spectroscopy (XPS) spectrum in Supplementary Figure 8 clearly shows the existence of Nb and W elements. Before discharge, the Nb-3d spectrum consists of a Nb-3d_5/2 (207.8 eV) and Nb-3d_3/2 (210.5 eV) doublet [Figure 4A]^[39], and the W-4f spectrum comprises a W-4f_7/2 (35.7 eV) and W-4f_5/2 (37.9 eV) doublet [Figure 4B]^[9], indicating the valences states of Nb and W are respectively +5 and +6, as expected. After discharged to 0.8 V, the lower doublet at 205.3 and 208.0 eV can correspond to Nb-3d_5/2 and Nb-3d_3/2 of Nb⁴⁺, and the shoulder peak at 203.9 eV can be attributed to Nb^3+[40]; the W-4f spectrum can be well fitted by the characteristic peaks of W⁵⁺ and W^4+[41]. Therefore, the two electrons transfer in each Nb/W during the electrochemical reaction of Nb₄W₇O₃₁. Excitingly, after being charged to 3 V, the valence states of Nb and W almost fully recover +5 and +6, respectively, suggesting the excellent electrochemical reversibility of the multi-electron transfer in Nb₄W₇O₃₁. In addition, the electrochemical impedance spectroscopy (EIS) results of the Nb₄W₇O₃₁/Li half cell after different cycles show that the charge-transfer resistance gradually decreases with the increase of the cycle number and that the formed solid electrolyte interface (SEI) becomes stable after ten cycles [Supplementary Figure 9A-C].

Figure 4. Ex-situ XPS spectra of pristine (OCV), lithiated (0.8 V), and delithiated (3.0 V) Nb₄W₇O₃₁ samples: (A) Nb-3d and (B) W-4f.

To study the Li⁺ diffusivity of Nb₄W₇O₃₁, its D_Li at various discharge/charge states are calculated from GITT data [Figure 5A] by using Equation (1)^[42,43].

Figure 5. GITT characterization and intercalation-pseudocapacitive behavior of Nb₄W₇O₃₁: (A) first-cycle GITT curve at 0.1C, (B) E vs. τ curves for a single GITT step, (C) linear relationship of E vs. τ^0.5 during a typical titration, and (D) variations in Li⁺ diffusion coefficients during lithiation and delithiation. (E) CV curves of Nb₄W₇O₃₁/Li half cell at various scan rates. Pseudocapacitive contributions at (F) 0.2, (G) 0.4, (H) 0.7, and (I) 1.1 mV s^-1.

where m_a, M_a, V_m, S, and L present the mass, molar mass, molar volume, electrode surface area, and electrode thickness of Nb₄W₇O₃₁, respectively; τ is the time during which a constant current is applied; ΔE_s and ΔE_τ respectively embody the change in the equilibrium potential and the change in potential during the current pulse, as presented in Figure 5B. This equation is reasonable and valid since the potential during each single titration can deliver a linear relationship with τ^0.5 [Figure 5C]. The average D_Li value of Nb₄W₇O₃₁ during lithiation is 1.24 × 10^-11 cm² s^-1, and that during delithiation is 1.09 × 10^-10 cm² s^-1 [Figure 5D]. The fact that the former value is obviously smaller than the latter one suggests that lithiation is the rate-determining step during the electrochemical reaction in Nb₄W₇O₃₁, which is similar to the cases in Li₄Ti₅O₁₂ and TiNb₂O₇^[8,33]. The large D_Li value of Nb₄W₇O₃₁ is confirmed by the EIS method [Supplementary Figure 9D and E] and surpasses those of the previously reported niobate, vanadate, and titanate anode materials (< 1 × 10^-11 cm² s^-1,Supplementary Table 3). Such fast Li⁺ transport in Nb₄W₇O₃₁ is undoubtedly rooted in its criss-cross Li⁺ transport network, allowing longitudinal transport along the large-sized pentagonal/quadrangular tunnels and transverse transport between the large-spaced MO_x (x = 6 and 7) layers.

For further understanding of the electrochemical kinetics of Nb₄W₇O₃₁, additional analyses are conducted based on the CV curves at various scan rates [Figure 5E], quantitatively determining the detailed contributions of the capacitive process (jv) and the diffusion-controlled process (kv^0.5) according to Equation (2)^[44,45]:

where I represents the detected current at a fixed potential, and j and k are adjustable factors. The resulting percentage of the capacitive contribution increases from 44.6% (0.2 mV s^-1, Figure 5F) to 53.7% (0.4 mV s^-1, Figure 5G), 60.6% (0.7 mV s^-1, Figure 5H), and 65.4% (1.1 mV s^-1, Figure 5I) with the scan-rate increasing. Therefore, this Nb₄W₇O₃₁ material reveals significant intercalation-pseudocapacitance behavior. This desirable behavior, the intrinsic charge-transport characteristic, and the extrinsic porous-microspherical morphology work together to achieve the superior rate capability of the Nb₄W₇O₃₁ material in this study.

To clarify the crystal-structural evolution of Nb₄W₇O₃₁ during lithiation/delithiation, an in-situ XRD experiment is carried out at 0.3C, as exhibited in Figure 6A and B. The weak WO₃ peaks located at 22.8, 23.7, 24.5, 33.8, and 34.5° do not shift during the electrochemical reaction, indicating that the tiny WO₃ impurity is inactive. The diffraction peaks initially located at 22.8, 23.2, 26.5, 28.1, 29.1, 30.3, 31.2, 33.0, and 35.1° correspond to the (001), (111), (270), (650), (800), (280), (660), (840), and (461) planes of Nb₄W₇O₃₁, respectively. During the first lithiation, these diffraction peaks show gradually decreased intensity and complicated shifts. From open-circuit voltage (OCV, ~2.7 V) to 1.8 V, they gradually shift to lower 2θ angles. Then, they suddenly shift to higher 2θ angles in a very short time. Finally, from 1.7 V to 0.8 V, they shift to lower 2θ angles again. During the subsequent delithiation, they monotonously shift to higher 2θ angles and almost recover their original intensities and positions at 3.0 V. This peak-evolution behavior confirms that only solid-solution reactions occur in the electrochemical process and suggests that the short and irreversible process at ~1.7 V during the first lithiation contributes to the small capacity loss [Figure 3A]. During the second lithiation, the peak evolution is almost contrary to that during both the first and second delithiation, demonstrating the intercalation-type nature of Nb₄W₇O₃₁ with superior structural and electrochemical reversibility after the first lithiation.

Figure 6. First two-cycle (A) pristine and (B) two-dimensional in-situ XRD patterns of Nb₄W₇O₃₁/Li in-situ cell with corresponding discharge/charge curves within 3.0-0.8 V at 0.3C (: PE preservative film; : WO₃). (C) Variations in lattice constants of Nb₄W₇O₃₁. Ex-situ HRTEM images of (D) pristine (OCV), (E) lithiated (0.8 V), and (F) delithiated (3.0 V) Nb₄W₇O₃₁ samples.

The lattice constants of Nb₄W₇O₃₁ at various discharge/charge states are calculated by a simple method. For the tetragonal structure, the interlayer spacing of the (280) plane is only associated with the lattice constant a, which can be calculated from the crystal and diffraction data by using Equation (3).

In this equation, (hkl) embodies the plane index (h = 2, k = 8, and l = 0 when calculating the a values), and d represents the interlayer spacing, which can be determined through Bragg’s Law. The resultant a-variation is revealed in Figure 6C and Supplementary Table 4. When Li⁺ ions gradually insert into the Nb₄W₇O₃₁ lattice, the a-variation follows a sequence of slight increase → tiny decrease → small increase. When Li⁺ ions are gradually extracted from the lattice, the a value monotonously decreases, which is highly reversible in the second cycle. This a-variation behavior matches well with the (270), (650), (800), (280), (660), and (840) peak shifts. The maximum a-value change is only +4.1% at 0.8 V. Similarly, the interlayer spacing of the (461) plane is only associated with a and c. Since the a values have been obtained, the c values can be easily achieved also using Equation (3). The c evolution follows a sequence of slight decrease → small increase with the maximum change of only +2.7% during lithiation and is highly reversible during delithiation [Figure 6C and Supplementary Table 4]. This slight and reversible c variation is confirmed by the ex-situ high-resolution transmission electron microscopy (HRTEM) characterization, showing that the (001) interlayer spacing enlarges from ~0.395 nm (OCV, Figure 6D) to ~0.404 nm (0.8 V, Figure 6E) and fully recovers (3.0 V, Figure 6F). The maximum V-value change, which only links to the a and c changes, is +11.2%. This percentage is similar to those of graphite (+13.2%) and Mo₃Nb₁₄O₄₄ (+10.6%) with similar layered structures [Supplementary Table 5]^[46,47]. The volume increase of graphite and Mo₃Nb₁₄O₄₄ is mainly caused by the interlayer expansion (+10.2% for graphite and +9.9% for Mo₃Nb₁₄O₄₄)^[45,46]. In sharp contrast, the interlayer expansion along the c direction in Nb₄W₇O₃₁ is only 2.7%, and the expansion along the a and b directions (4.1%) also contributes to its moderate volume increase. Compared with the one-dimensionally considerable expansion in graphite and Mo₃Nb₁₄O₄₄, the three-dimensionally uniform and small expansion in Nb₄W₇O₃₁ can more effectively alleviate the damage of volume expansion, greatly benefiting the structural and cyclic stability.

To reveal the real crystal structure of Nb₄W₇O₃₁ and its Li⁺-storage sites after lithiation, spherical-aberration-corrected STEM is employed. The high-angle annular dark field (HAADF, Figure 7A) and the annular bright field (ABF, Figure 7B) STEM images of lithiated Nb₄W₇O₃₁ (at 0.8 V) can be viewed along the [001] zone axis, which are acquired under optimal imaging conditions for identifying heavy atoms (Nb and W) and light atoms (Li and O), respectively^[48]. The a value obtained from the STEM images is 25.30 Å, in good agreement with the in-situ XRD results. Since Nb₄W₇O₃₁ is an intercalation-type Li⁺-storage material with small expansion along all three dimensions, the Nb₄W₇O₃₁ framework should be well-kept during repeated lithiation/delithiation. Consequently, the HAADF-STEM image containing the exact positions of Nb and W can reflect the real structure of Nb₄W₇O₃₁. In Figure 7A, the intergrowth of Structure A, Structure B, and Structure C with different orientations and some defects can be observed, confirming the structural mixing in Nb₄W₇O₃₁, which can be the main cause of the three-dimensionally uniform and small expansion/shrinkage. In order to intuitively identify the light-atom positions, inverted image contrast with colors is applied for the magnified ABF-STEM image [Figure 7C]. As a result, the inserted Li⁺ ions can be seen, and some examples are indicated by dotted circles and highlighted by the white arrows. Clearly, the inserted Li⁺ ions occupy the pentagonal and quadrangular tunnels in the three structures, as schematically illustrated in Figure 7D-F.

Figure 7. HAADF and ABF STEM results of lithiated Nb₄W₇O₃₁ (0.8 V) confirming its Li⁺-storage sites: (A) atomic-resolution HAADF-STEM and (B) ABF-STEM images viewed along [001], (C) magnified ABF-STEM color image with inverted image contrast showing the positions of inserted Li⁺ ions (some examples are indicated by dotted circles and highlighted by white arrows). Unit cells corresponding to Structure A, Structure B, and Structure C are respectively highlighted by red, blue, and black squares in (A and C). For clarity, in (C), Nb/W and Li are shown in only one-quarter of each unit cell due to the symmetry of the P4 space group. Crystal structure models of (D) lithiated Structure A, (E) lithiated Structure B, and (F) lithiated Structure C corresponding to HAADF-STEM and ABF-STEM images.

CONCLUSIONS

Porous Nb₄W₇O₃₁ microspheres formed by self-assembly of nanorods are developed as a new intercalation-type anode material with comprehensively good Li⁺-storage properties. The presence of proper surfactants during the solvothermal process is the key to forming this hierarchical morphology. Nb₄W₇O₃₁ owns an interesting crystal structure with mixed TTB and confined ReO₃ units, abundant large-sized tunnels, and a large interlayer spacing (~3.95 Å), resulting in the three-dimensionally uniform expansion/shrinkage (≤ 4.1%) and fast Li⁺ transport (1.24 × 10^-11 cm² s^-1 during lithiation and 1.09 × 10^-10 cm² s^-1 during delithiation) when the external Li⁺ ions insert into/extract from the pentagonal and quadrangular tunnels. These structural merits combined with the morphological merits enable excellent cyclic stability (95.2% capacity retention after 1,500 cycles at 10C) and superior rate capability (10C vs. 0.1C capacity retention percentage of 47.6%). The highly reversible redox reactions of the multiple Nb⁵⁺/Nb⁴⁺, Nb⁴⁺/Nb³⁺, W⁶⁺/W⁵⁺, and W⁵⁺/W⁴⁺ couples in Nb₄W₇O₃₁ are confirmed and enable the large reversible capacity (252 mAh g^-1 at 0.1C), safe operating potential (~1.66 V), and high first-cycle Coulombic efficiency (88.4%). Additionally, the LiFePO₄//Nb₄W₇O₃₁ full cell also exhibits prominent electrochemical properties, demonstrating the practicability of the porous Nb₄W₇O₃₁ microspheres for high-performance LIBs. The structural and morphological designs in this work can find its extensional applications for enhancing the electrochemical properties of other energy-storage materials.