Fabrication and Characterization of Nb 4 N 5 -Nb 2 O 5 Heterostructure. The desirable properties of electrocatalysts for fast conversion of LiPSs are dominated by three main factors: (a) high adsorption ability to anchor LiPSs, (b) rich catalytic active sites to enhance the conversion kinetics, and (c) good electrical conductivity for effectively electron transfer24, 25. Based on that, 2D Nb4N5-Nb2O5 heterostructure integrates those properties with synergetic effect of Nb2O5 and Nb4N5. To be more specific, the bare Nb2O5 possesses desirable chemical adsorption, which is necessary for the LiPSs. However, electrical conductivity is not high for electrons transfer26. In contrast, polar Nb4N5 with superior electrical conductivity facilitates efficient conversion of LiPSs, but poor affinity of the LiPSs on its surface27. Therefore, the novel design of Nb4N5-Nb2O5 heterostructure by coupling the merits of highly adsorptive Nb2O5 and conductive Nb4N5 can enable a fast nucleation and conversion of the LiPSs. Consequently, the LiPSs shuttling is effectively confined and the utilization of sulfur is greatly promoted (Fig. 1a).
The 2D Nb4N5-Nb2O5 heterostructures were prepared by a facile hydrothermal and subsequent ammonia annealing treatment. The scanning electron microscope (SEM) image showed uniform flower morphology, composed of holey nanosheets (Fig. 1b and Supplementary Fig. 1). Significantly, the composition of Nb4N5-Nb2O5 heterostructure could be adjusted by nitriding temperature and time. As a result, Nb2O5-Nb4N5 heterostructures with the ratio of around 1:1 (Supplementary Table 1) were successfully obtained (Fig. 1c). The high-resolution transmission electron microscope (TEM) images illustrated that the lattice spacings were 0.25 nm and 0.39 nm, matching the (211) planes of Nb4N5 (Supplementary Fig. 2) and (001) planes of Nb2O5 (Supplementary Fig. 3), respectively. More importantly, a heterostructured interface between them was clearly observed (Fig. 1d). Such an interface was not only conducive for fast electron transportation, but also improved the adsorption and conversion of LiPSs24. In addition, the porous structure was beneficial to offer easy accessibility to LiPSs with rich active sites for the consequent adsorption and catalytic conversion (Fig. 1e and Supplementary Fig. 4). X-ray photoelectron spectroscopy (XPS) further confirmed the existence of Nb-O, Nb-N bonding configuration (Fig. 1f)28. The intact connection and uniform distribution of Nb2O5 and Nb4N5 were verified by the energy-dispersive spectrometry element mapping results (Fig. 1g).
Electrocatalytic and Adsorption Effects of Nb 4 N 5 -Nb 2 O 5 Heterostructure. First-principle calculations were performed to disclose the chemical interaction of LiPSs (Li2S4 was chosen as the protype for modelling) with Nb2O5 and Nb4N5 configurations. The adsorption energies of Li2S4 on Nb4N5 (211) and Nb2O5 (001) surfaces were − 4.8 eV and − 6.2 eV respectively, indicating that Nb2O5 had higher adsorption affinity for LiPSs29. Further, the superior conductivity of Nb4N5 guaranteed the fast diffusion of LiPSs from Nb2O5 surface to Nb4N5 surface across the interfacial migration between them (Fig. 2a, b). As a result, Nb4N5-Nb2O5 could efficiently enhance the electrochemical reactions and accelerate the oxidation conversion of dissolved LiPSs to solid Li2S in the discharge process, which ensured a discharging/charging loop process with excellent reversibility (Fig. 2c).
We selected the Nb4N5-Nb2O5 heterostructure (with the ratio of around 1:1) as the model catalyst representation for deducing the reaction mechanism of electrocatalysis in corresponding Li-S batteries. Nb4N5-Nb2O5 mixture (Nb4N5-Nb2O5 mix with the ratio of 1:1), bare Nb4N5 and bare Nb2O5 were also selected for comparison. In order to confirm the improved redox-reaction kinetics in the liquid-liquid transformation process, cyclic voltammetry (CV) measurement using symmetric batteries based on Li2S6 catholyte were employed. The redox current response increased in the order of Nb2O5 < Nb4N5 < Nb4N5-Nb2O5 mix < Nb4N5-Nb2O5 heterostructure (Fig. 2d), reflecting that Nb4N5-Nb2O5 heterostructure possessed better interfacial kinetics30. Apart from the liquid-liquid transformation of LiPSs, the polar and conductive heterostructure also played a key role at liquid-solid boundary. A potentiostatic nucleation experiment was conducted to understand the electrochemical deposition from LiPSs to solid Li2S. Specifically, the capacity 168 mAh g− 1 of the precipitated Li2S on Nb4N5-Nb2O5 heterostructure electrode was much higher than those of Nb4N5-Nb2O5 mix (132 mAh g− 1), Nb4N5 (124 mAh g− 1), and Nb2O5 (88 mAh g− 1). Besides, the battery with Nb4N5-Nb2O5 heterostructure exhibited the highest current density of 0.074 mA, suggesting the role of heterostructure in fast LiPSs trapping and nucleation of Li2S (Fig. 2e-h)31. Moreover, the sluggish oxidation kinetics of solid Li2S at the charging cycle is the dominant factor for the reduced reversibility of Li2S-to-LiPSs interconversion, thus leaving behind unusable electrochemical phases (also called “dead sulfur”). Similarly, kinetic evaluation of Li2S decomposition was conducted by a potentiostatic charging process after full discharge into solid Li2S. Nb4N5-Nb2O5 heterostructure showed an obvious oxidation current peak at 577 s, which was much earlier than those of Nb4N5-Nb2O5 mix (763 s), Nb4N5 (1021 s) and Nb2O5 (2620 s). In addition, the improved Li2S dissolution could reduce the deactivation of the catalyst surface and increase the utilization of sulfur32. Furthermore, the linear sweep voltammetry (LSV) indicated that Nb4N5-Nb2O5 heterostructure electrode exhibited highest reaction peak current (0.43 mA cm− 2) with lowest Tafel slope (89.4 mV dec− 1), further demonstrative of the improved kinetics of LiPSs redox reactions (Fig. 2j and Supplementary Fig. 5)33.
To study the chemical adsorption behavior of the Nb4N5-Nb2O5 heterostructure for LiPSs, a visual adsorption test was conducted by adding the same amount (~ 5 mg) of materials into the Li2S6 solution. As shown in Fig. 2k, the solution with Nb4N5-Nb2O5 heterostructure additive rapidly decolored. The sample with Nb2O5 showed slightly inferior decoloration phenomenon compared to the heterostructure due to the limited specific area (33.9 m2 g− 1) than that of heterostructure (40.5 m2 g− 1), while the control sample with bare Nb4N5 showed minor difference (Fig. 1e and Supplementary Fig. 4). This was consistent with the ultraviolet-visible (UV-vis) absorption and density functional theory (DFT) calculation results. Moreover, the nature of the interaction between LiPSs and Nb4N5-Nb2O5 heterostructure was probed by XPS measurement. In the deconvoluted Li 1s spectrum, two obvious peaks at 60.9 eV and 59.7 eV were observed, corresponding to Li-O and Li-N bond (Fig. 2l). Moreover, additional Nb-S peaks (203.7 eV, 206.3 eV) were detected in Nb 3d XPS spectrum34. These results indicated that the Nb4N5-Nb2O5 heterostructure strongly interacted with LiPSs (Fig. 2m), offering a potential functional material for LiPSs electrocatalysis.
To demonstrate the favorable LiPSs adsorption and electrocatalytic reactivity of Nb4N5-Nb2O5 heterostructure on the sulfur reactions, the half-cell configurations were fabricated with a Li metal foil as anode and Nb4N5-Nb2O5 heterostructure, Nb4N5-Nb2O5 mix, Nb4N5 and Nb2O5 loaded with sulfur as cathodes respectively to evaluate their electrochemical performance. The CV curves of all the assembled Li-S batteries showed the typical pair of redox peaks, corresponding to the formation of soluble LiPSs (2.2–2.4 V) and solid Li2S (2.1-2.0 V). Obviously, the Nb4N5-Nb2O5 heterostructure exhibited a considerably mitigated electrochemical polarization with the highest current intensity and a good overlap of CV profiles (Fig. 3a and Supplementary Fig. 6), indictive of enhanced redox kinetics and sufficient utilization of the LiPSs along Nb4N5-Nb2O5 heterostructure35. Moreover, the substantial improvement of the charge transfer from the Nb4N5-Nb2O5 heterostructure was further verified by electrochemical impedance spectroscopy (EIS). Obviously, the charge-transfer resistance (Rct) of Nb4N5-Nb2O5 electrode (18.8 Ω) was lower compared with the Nb4N5-Nb2O5 mix (24.5 Ω), Nb4N5 (64.0 Ω) and Nb2O5 (111.8 Ω), implying better interfacial kinetics of Nb4N5-Nb2O5 heterostructure (Fig. 3b)36.
Subsequently, the cyclability was tested at a constant current density of 0.5 C to evaluate the catalytic ability of the cathodes. The initial capacity of the cathode with Nb4N5-Nb2O5 heterostructure was 1108 mAh g− 1, significantly higher than those of Nb4N5-Nb2O5 mix (1015 mAh g− 1), Nb4N5 (975 mAh g− 1) and Nb2O5 (903 mAh g− 1) (Fig. 3c and Supplementary Fig. 7). In addition, the cathode with Nb4N5-Nb2O5 heterostructure showed smaller polarization (150 mV) than those of Nb4N5-Nb2O5 mix (213 mV), Nb4N5 (210mV) and Nb2O5 (259 mV) (Supplementary Fig. 8), suggesting the improved redox reaction37. Apart from that, the Nb4N5-Nb2O5 heterostructure based sulfur cathode retained a high capacity of 1021 mAh g− 1 and high Coulombic efficiency ~ 100% over 120 cycles, with an ultralow capacity decay rate of 0.07%. Impressively, a high capacity of 942 mAh g− 1 was sustained more than 500 cycles, (Supplementary Fig. 9), which suggested that the shutting effect of LiPSs was significantly suppressed as the strong chemisorption. In contrast, only 760 mAh g− 1 for Nb4N5-Nb2O5 mix (capacity degradation rate of 0.18%), 681 mAh g− 1 for Nb4N5 (capacity degradation rate of 0.21%), and 443 mAh g− 1 for Nb2O5 (capacity degradation rate of 0.42%) were remained at 120 cycles (Fig. 3d). The rate performances from 0.5 C to 10 C were further tested to evaluate the superior kinetics of LiPSs conversion. As shown in Fig. 3e, the Nb4N5-Nb2O5 heterostructure based sulfur cathode presented superior rate response and exceptional reversibility compared with those of the reference electrodes. Specifically, when cycled at step current rates (0.5, 1, 2, 3, 5 and 8 C), such a heterostructure electrode could deliver discharge capacities of 1159, 1095, 1063, 1028, 968 and 899 mAh g− 1 with nearly 100% Coulombic efficiency, respectively. Even increasing high current rate to 10 C, an excellent reversible capacity (844 mAh g− 1) was remained. In contrast, the batteries employing Nb4N5-Nb2O5 mix, bare Nb4N5 and Nb2O5 showed much inferior capacities of 675, 225 and 141 mAh g− 1 under the rate of 10 C (Supplementary Fig. 10). Remarkably, the ultrahigh rate capability coupled with ultralow capacity decay rate, has been raraly reported in the former heterostructure materials-based Li-S batteries (Supplementary Table S2), such as MoN-VN (636 mAh g− 1 at 2 C with 0.06% decay rate)38, VO2-VN (587 mAh g− 1 at 5 C with 0.06% decay rate)24 and TiO2-Ni3S2 (534 mAh g− 1 at 5 C with 0.04% decay rate)32. In addition, the maximum power density (12010 W kg− 1) and energy density (1628 Wh kg− 1) calculated by the whole mass of cathode are competitive to majority recent reports (Supplementary Fig. 11), such as VN-S (3058W kg− 1, 1014 Wh kg− 1)39, TiS2-S (3058 W kg− 1, 1014 Wh kg− 1)40 and ZnS-S (3863 W kg− 1, 1396 Wh kg− 1)41.
Lithium Dendrite Suppression of Nb 4 N 5 -Nb 2 O 5 Heterostructure. In fact, the performance of Li-S batteries is severely limited by both unacceptable Li dendrite formation and unstable solid electrolyte interphase (SEI)42. In particular, the uneven Li-ion flux during the repeating plating/stripping is regard as the main reason causing the safety risks and short lifetime of Li metal anode43. Our 2D Nb4N5-Nb2O5 heterostructure with dominating hole (1.7–3.5 nm) (Fig. 4a and Supplementary Fig. 12) is expected to benefit the Li-ion redistributor for improving the uniformity of Li-ion flux (Fig. 4b). Besides, the abundant and homogeneously distributed N and O element on the Nb4N5-Nb2O5 surface can be regarded as lithiophilic sites (binding energy with Li atom: -3.5 eV for Nb2O5 and − 5.3 eV for Nb4N5) to guide the uniform Li nucleation (Supplementary Fig. 13 and Supplementary Fig. 14)44. Half-cell configurations paired with Li foil as counter electrode were developed firstly to explore the reversibility of Nb4N5-Nb2O5 electrodes. For comparison, the bare Cu electrode was also assembled. After pre-cycling between 0.01-1 V for 4 cycles to stabilize the SEI film and clean impurity (Supplementary Fig. 15)45, the Nb4N5-Nb2O5 showed significantly improved Coulombic efficiency of 99.9% with a steady voltage hysteresis (~ 53 mV) for 300 cycles than the bare Cu foil electrode (58.6% for 16 cycles) at 0.5 mA cm− 2 for tripping/plating capacity of 0.5 mAh cm− 2 (Fig. 4c and Supplementary Fig. 16). Impressively, ultralong cycling lifespan of 1400 h coupled with high average Coulombic efficiency of 99.7% was enabled at 0.5 mA cm− 2 without dendrite growth (Supplementary Fig. 17). In addition, the voltage dropped sharply to -174 mV (vs. Li+/Li) at the Li nucleation stage on Cu foil electrode, while the voltage curve of Nb4N5-Nb2O5 electrode exhibited much smoother voltage dip with a smaller nucleation overpotential of only 56 mV (Fig. 4d). The results indicated the high lithiophilic property of the Nb4N5-Nb2O5 surface46. Then, the long cycling stability of Nb4N5-Nb2O5 electrode was evaluated by the Li||Li symmetrical configuration. With the pre-stored capacity of 1 mAh cm− 2 at 1 mA cm− 2, excellent cycling stability (1000 h) with highly stable overpotential (~ 10.5 mV) was highlighted for Nb4N5-Nb2O5/Li||Nb4N5-Nb2O5/Li symmetric batteries. However, a limited cycle lifespan with obvious fluctuant overpotential (162 mV for 102 h) was observed for the Cu-Li||Cu-Li symmetric batteries (Fig. 4e, f). Even at high current density of 3 mA cm− 2 and large plating capacity of 3 mAh cm− 2, the Nb4N5-Nb2O5/Li based anode could also exhibit long cycling life more than 900 h with a stable voltage hysteresis of ~ 11.0 mV (Fig. 4g and Supplementary Fig. 18). SEM characterizations were conducted to clearly elucidate the Li deposition morphology after cycling. For the Cu/Li electrode, the cavities and islands were formed on the surface after Li plating and striping at 1 mA cm− 2 (Supplementary Fig. 19). As for the Nb4N5-Nb2O5/Li anode, no obvious “dead Li” and Li dendrite were observed after cycling (Fig. 4h and Supplementary Fig. 20), indicative of the homogeneous Li deposition, due to the uniform distribution of Li ions by the holey and lithiophilic Nb4N5-Nb2O5 nanosheets.