3d orbital electron tunning and crystal engineering enables high-capacity vanadium oxide for aqueous ammonium ion batteries

Li ion batteries have dominated the market for energy storage devices due to their high energy densities and superior power densities. Nonetheless, their broad application for grid-scale fields is hindered by the internal high cost and risk of explosion, which drives the research for alternative energy storage devices. Aqueous rechargeable batteries (ARBs) own the unique advantages of non-flammable electrolytes, which avoid the danger of self-burning. In addition, its economical and environmentally friendly characteristics have shown potential for large-scale energy systems.¹ To date, various ARBs based on other metal ion charge carriers, such as Na⁺, K⁺, Mg²⁺, Ca²⁺, Zn²⁺, and Al³⁺, have been widely developed.^2–9 Compared with metal ions, aqueous ammonium ion batteries (AAIBs) with non-metallic ions (NH₄⁺) as charge carriers have garnered particular attention due to their particular energy storage mechanism. Specifically, the light molar mass (18 g mol⁻¹) and small hydrated ionic size (3.31 Å) of NH₄⁺ enable fast ion diffusion in the electrolyte.¹⁰ In addition, in contrast with other non-metallic cations like proton (H⁺) and hydronium (H₃O⁺), NH₄⁺ can avoid the corrosion of the electrode active substance.¹¹ 

Currently, a series of materials have been developed for NH₄⁺ storage, including Prussian blue analog (K_0.72Cu[Fe(CN)₆]_0.78, (NH₄)_1.47Ni[Fe(CN)₆)₂]_0.88, Fe[Fe(CN)₆]_0.88, and Na_0.73Ni[Fe(CN)₆]_0.88),^12–15 organic compound (PTCDI, alloxazine, and PNTCDA),^13,16,17 transition metal oxide (V₂O₅, MnOx, and h-MoO₃),^18–20 and other compounds [K_0.38(H₂O)_0.82MoS₂, QA-COF, and TCNQ-rGO].^21–23 Among them, vanadium pentoxide (V₂O₅) has been recognized as an excellent NH₄⁺ host material beneficial for its cost economical (∼US $7.85/1b) and high theoretical capacity (589.0 mA h g⁻¹ based on two-electron transport).²⁴ For instance, Ji’s group reported a bi-layer hydrated V₂O₅ with a capacity of 103 mA h g⁻¹ for the AAIBs.²⁰ Mai’s group demonstrated a strategy that intercalated polyaniline into V₂O₅ (PVO), which exhibits a good capacity of 192 mA h g⁻¹ at 1 A g⁻¹ due to the increasing inter-layer spacing.²⁵ Xing et al. prepared a Co-doped V₂O₅ electrode with a capacity of 172.0 mA h g⁻¹ at 1 A g⁻¹.²⁶ However, these strategies still have some limitations. For example, the mismatch between doped molecules and crystal structure hinders the further study of vanadium pentoxide.²⁷ While as for the pre-intercalated method, the electrochemical properties of vanadium pentoxide can be improved only when the molecular is intercalated at specific insertion sites.²⁸ Furthermore, these strategies mainly focus on modifying the interlayer distance or the lattice space of V₂O₅, and the capacity, which is lower than 200 mA h g⁻¹, cannot satisfy the requirement for high performance AAIBs.²⁹ 

Here, we report a high capacity V₂O₅ (denoted as N-VO) electrode for NH₄⁺ storage through the 3d-orbital state tunning and crystal engineering strategy. The N-VO with the tailored 3d orbit is synthesized by annealing the NH₄V₄O₁₀ precursor in the N₂ atmosphere. Compared to the sample prepared in the air atmosphere (denoted as A-VO), N-VO has a different crystal structure and a higher content of V⁴⁺. The d electron configuration of V⁴⁺ is $t2g1$⁠, which contained one unpaired t_2g electron (Scheme 1), which can not only enhance the electrostatic force between NH₄⁺ and the vanadium center but also facilitate the electron transport, thus leading to an enhanced capacity of N-VO. As expected, the N-VO electrode exhibits a high specific capacity of 249.3 mA h g⁻¹ at 1.0 A g⁻¹ and 69.5 mA h g⁻¹ at 10 A g⁻¹. Additionally, pairing with a Prussian blue analog cathode, the resultant quasi-solid-state ammonium-ion full cell shows a good cycling stability with capacity retention of 87.9% after 10 000 cycles (at 1 A g⁻¹) and exhibits good mechanical flexibility with no capacity decay after cycling at different bending angles.

The N-VO was prepared via a two-step process, which includes the hydrothermal synthesis of the NH₄V₄O₁₀ precursor (Figs. S1–S3) and a post-anneal treatment at 400 °C for 5 h in the N₂ atmosphere. For comparison, an A-VO sample was also prepared via a similar heating treatment in an air atmosphere. The crystal structure of N-VO and A-VO was analyzed by x-ray diffraction (XRD) [Fig. 1(a)]. The diffraction peaks of N-VO located at 15.1°, 25.3°, 30.3°, 43.1°, 45.8°, and 49.3° fit well to the (200), (110), (400), (−404), (114), and (−603) planes of monoclinic V₆O₁₃ (PDF#89-0100),³⁰ while the diffraction peaks of A-VO at 20.3°, 21.7°, 31.1°, 32.4°, 45.5°, 47.3°, and 51.2° are index to (001), (101), (301), (011), (411), (600), and (020) of orthorhombic V₂O₅ (PDF#77-2418).³¹ The morphologies of A-VO and N-VO were investigated by the scanning electron microscope (SEM). As revealed in Fig. 1(b), the A-VO sample shows a nano-sheet like morphology with the uneven size. In contrast, the morphology of the optimized N-VO sample is transformed into a smaller nano-sheet [Fig. 1(c)]. Low transmission electron microscopy (TEM) characterization of N-VO revealed the diameter of the nanosheet is around 81.9 nm [Fig. 1(e)], which is thinner than A-VO (139.3 nm) [Fig. 1(d)]. Besides, the high-resolution TEM (HRTEM) of the A-VO sample in Fig. 1(f) displayed that the lattice distance of 0.34 nm corresponds to the (110) phase of orthorhombic V₂O₅ (PDF#77-2418). The selected area electron diffraction (SAED) pattern of A-VO exhibits a bright and well-organized diffraction spot indexed with (512), (321), (422), and (701) of orthorhombic V₂O₅ (PDF#77-2418), reflexing high purity with a single crystal structure. An inter-layer distance of 0.212 nm was measured for N-VO [Fig. 1(g)], which is well indexed to the (−404) plane of monoclinic V₆O₁₃ (PDF#89-0100). The SAED pattern of N-VO displays two diffraction rings corresponding to (002) and (112) phases, demonstrating the polycrystalline nature of N-VO. According to energy dispersive spectra (EDS) elemental mapping images in Figs. 1(h) and 1(i), V and O elements are evenly distributed in the as-prepared N-VO and A-VO samples.

Raman spectra and Fourier transform infrared (FT-IR) spectra were carried out to further analyze the structure information of the samples. As shown in Fig. 2(a), the characteristic peaks of A-VO at 103, 148, 288, 307, 407, 483, 529, 703, and 996 cm⁻¹ can be assigned to the vibration of orthorhombic V₂O₅.³² In contrast, N-VO exhibits Raman vibrational modes located at 148, 259, 289, 419, 512, and 692 cm⁻¹, which correspond to the representative bands of monoclinic V₆O₁₃. The 419 and 512 cm⁻¹ peaks are attributed to the V–O–V stretching mode for N-VO. The maximum intensity peak located at 148 cm⁻¹ corresponds to the characteristic skeleton bent vibration and demonstrates the B_g symmetry.³³ The lower intensity peaks of N-VO prove that the N-VO is less crystalline than A-VO, which is consistent with the SAED result.³⁴ In the FT-IR spectra, the peaks of A-VO located at 616.9, 829.0, and 1020.5 cm⁻¹ correspond to V–O–V, V–O, and V=O stretching vibrations, which are matched well with the structure of V₂O₅.³⁵ As for N-VO, the absorption peaks at 998.9 and 877.5 cm⁻¹ can be attributed to V=O and V–O stretch vibrations of V₆O₁₃. Noticeably, the peaks at 535 and 753 cm⁻¹ are indexed with V–O–V symmetric and asymmetric stretching vibration.³⁶ Additionally, no additional peaks assigning to the N–H bond are detected in the range of 400–4000 cm⁻¹, suggesting the absence of NH₄⁺ in N-VO and A-VO samples.³⁷ X-ray photoelectron spectrum (XPS) was conducted to clarify the chemical composition and valence state of N-VO and A-VO. No N 1s peaks were observed in the XPS spectra of N-VO and A-NO (Fig. S4). The high resolution XPS spectra were carried out to further analyze the valence states of N-VO and A-VO. As displayed in Fig. 2(c), the V 2p spectra can be split into two peaks located at 517.3 and 516.6 eV, corresponding to V⁵⁺ and V⁴⁺.³⁸ It is worth noting that the content of V⁴⁺ for N-VO is significantly higher than that of A-VO, verifying that the crystal transition during the annealing process is accompanied by the reduction from V⁵⁺ to V⁴⁺. The reduction of the vanadium atom could inject the electrons into the d band of V and increase the electron concentrations of t_2g of vanadium’s 3d-orbital state, scilicet higher t_2g orbital electron filling.^39,40 The O 1s spectra are fitted to three peaks at ∼531.8, ∼530.8, and ∼530.1 eV indexed with the absorbent water, V–O–V and V=O, respectively [Fig. 2(d)].⁴¹ Surface area is another essential effect for the NH₄⁺ storage capability. Figure 2(e) compares the surface area of N-VO and A-VO via the dye-absorption method. Obviously, the absorption peak intensity of N-VO decreases significantly compared to that of A-VO, revealing that the N-VO has a higher surface area than A-VO.⁴² 

The electrochemical performance of N-VO and A-VO was evaluated by cyclic voltammetry (CV) and galvanostatic charge–discharge (GCD) tests in the three-electrode system with 1M ammonium acetate as the electrolyte. Figure 3(a) shows the CV curves of the two samples at the scan rate of 1 mV s⁻¹. The N-VO electrode exhibits an obvious pair of redox peaks at −0.75 and −0.26 V, respectively, which could be attributed to the reversible NH₄⁺ insertion/extraction in the N-VO host. Note that the increasing area of the CV curve of N-VO reveals its electrochemical reactivity is significantly improved during the annealing treatment. The GCD profile at 1 A g⁻¹ of the N-VO electrode delivers a high capacity of 249.3 mA h g⁻¹, which is about 23 times higher than that of the A-VO (10.5 mA h g⁻¹) [Fig. 3(b)]. More encouragingly, the N-VO electrode also achieves considerable discharge capacity of 251.9, 248.9, 216.7, 165.4, 126.2, 93.7, and 64.4 mA h g⁻¹ at different current densities from 0.6 to 10 A g⁻¹, outperforming the A-VO electrode [Fig. 3(c)]. In addition, these values of N-VO are incredibly better than previously reported anode material [Fig. 3(d)], such as VO_x@PPy (194.5 mA h g⁻¹, 0.2 A g⁻¹),⁴³ PTCDI (119.0 mA h g⁻¹, 0.24 A g⁻¹),¹³ CF@PANI (63.0 mA h g⁻¹, 0.1 A g⁻¹),⁴⁴ QA@COF(220.4 mA h g⁻¹, 0.5 A g⁻¹),²² h-MoO₃ (115.0 mA h g⁻¹, 0.1 A g⁻¹)¹⁹ and titanic acid (95 mA h g⁻¹, 0.5 A g⁻¹).⁴⁵ The above-mentioned results proved that the NH₄⁺ storage capability of N-VO can be significantly boosted by the modulation of the 3d-orbital state and crystal structure of N-VO via the annealing process.

A series of measurements, including ex-situ XRD, FT-IR, and XPS, were used to probe the electrochemical behaviors of the NH₄⁺ intercalation/deintercalation process. Figure 4(a) shows the ex-situ XRD patterns of N-VO at different charging/discharging states at five points labeled the GCD curve. During the charging/discharging process, there are no new peaks forming, and the diffraction peaks, e.g., (110), (−203), (400), and (310) do not shift, suggesting the intercalation and deintercalation of NH₄⁺ do not cause the phase transition of N-VO [Fig. 4(b)].³¹ For the FT-IR spectra, the absorption peaks at around 1350 cm⁻¹ can be assigned to be the vibration of the N–H bond [Fig. 4(c)]. Upon the charging and discharging process, the intensity of the N–H bond increased first and then decreased, indicating the insertion/desertion of ammonium ions into/from N-VO. Figure 4(d) presents the N 1s spectra of the N-VO at full charging/discharging states. In the charged state (CS), two N 1s groups –NH⁺-(41%) and –NH-(59%) are detected, while the –NH⁺- group increases to 55% and the –NH– group decreases to 45% in the discharged state (DS) due to the inserted NH₄⁺.⁴⁶  Figure 4(e) displays the V 2p spectra of the N-VO electrode at the different states. The V 2p spectra at CS show two peaks at 524.9 and 517.5 eV corresponding to V⁵⁺ and two peaks at 523.2 and 516.1 eV corresponding to V⁴⁺.³⁸ When at the DS, the intensity of V⁵⁺ decreases and the constant of V⁴⁺ increases. It can be seen that the V⁵⁺ is reduced to V⁴⁺ when the NH₄⁺ are intercalated; hence the amount of V⁵⁺ increases and that of V⁴⁺ decreases during the intercalated process.

Finally, aqueous ammonium ion full cells were assembled to further validate the practical application of the N-VO electrode. The full cells were prepared with the N-VO or A-VO as anodes and the synthesized Prussian blue analog NiHCF as the cathode (denoted as N-VO//NiHCF and A-VO//NiHCF). As shown in Fig. 5(a), the N-VO//NiHCF displays a specific capacity of 83.9 mA h g⁻¹ at 0.5 A g⁻¹, much higher than that of the A-VO//NiHCF (36.6 mA h g⁻¹). In addition, the capacity of N-VO//NiHCF at different current densities is also superior to that of A-VO//NiHCF [Figs. 5(b) and S10]. The capacity of N-VO//NiHCF is higher than that of A-VO//NiHCF (94.8 vs 35.9 mA h g⁻¹) at 0.2 A g⁻¹. When the current density returns back to 0.2 A g⁻¹, its capacity can recover to 104.3 mA h g⁻¹, indicating the excellent rate capability of N-VO//NiHCF. Remarkably, the cyclic performance of N-VO//NiHCF could sustain its performance with 61.0 mA h g⁻¹ after 1000 cycles at 1 A g⁻¹, superior to A-VO//NiHCF (23.2 mA h g⁻¹) [Fig. 5(c)]. The improving electrochemical performance can be ascribed to the decreased resistance from 8.2 to 2.3 Ω, which demonstrates that the optimized 3d orbit of the vanadium site allows higher electrical conductivity for AAIBs [Figs. 5(d) and S11]. Finally, we attempt to estimate the performance of a quasi-solid-state ammonium battery with polyacrylamide gel (denoted as N-VO/PAM/NiHCF) on a more practical application in the field of flexible devices. As shown in Fig. 5(e), the capacity of N-VO/PAM/NiHCF increases in the first 400 cycles, which can be ascribed to the electrochemical activation of the electrodes.²⁴ More importantly, N-VO/PAM/NiHCF displays impressive cycling performance with capacity retention of 87.9% after 10 000 cycles at 1 A g⁻¹. Furthermore, the N-VO/PAM/NiHCF remains similar in electrochemical performance with a relevant working potential platform and exhibits good cycling stability with no capacity decay after 400 cycles at 1 A g⁻¹ under different bending angles, which preserved its performance subjected to continuously mechanical manipulations [Figs. 5(f) and S12]. In addition, a 1.5 V LED lamp was successfully driven by 5 series N-VO/PAM/NiHCF batteries [Fig. 5(g)], demonstrating the feasibility for the practical application.

In conclusion, a strategy that tunes the 3d orbit and modifies the crystal structure of vanadium oxide is developed to achieve a high-capacity N-VO electrode for AAIBs. Thanks to the increasing orbital electron content of t_2g of the vanadium center, the active sites of the vanadium site are enhanced. As a result, the N-VO electrode delivers a superior capacity of 249.3 mA h g⁻¹ at 1.0 A g⁻¹ and 69.5 mA h g⁻¹ at 10 A g⁻¹. Moreover, the assembled quasi-solid-state N-VO/PAM/NiHCF device delivers excellent cycling stability with capacity retention of 87.9% after 10 000 cycles at 1 A g⁻¹ and exhibits good mechanism flexibility with no capacity decay after cycling at different bending angles. Such a study offers a simple and efficient way to reach high-capacity electrodes, providing a new strategy for preparing advanced AAIBs.

Synthesis of NH₄V₄O₁₀ precursor: 0.628 g H₂C₂O₄⋅2H₂O was dissolved in 30 ml deionized water with 0.562 g NH₄VO₃. The mixture was stirred at 80 °C for 20 min to obtain a dark green solution. The solution was then transferred to a 50 ml reactor, sealed tightly, and reacted in an oven at 180 °C for 4 h. After the reactor was cooled to room temperature, the dark green powder was separated by centrifugation and washed with deionized water and ethanol three times. Finally, the NH₄V₄O₁₀ powder was dried overnight in an oven at 60 °C to obtain it.

Synthesis of N-VO and A-VO samples: 200 mg of NH₄V₄O₁₀ precursor was annealed at 400 °C in N₂ atmosphere for 5 h (the heating rate is set to 5 °C min⁻¹) to obtain the N-VO powder. Other conditions were the same, and the atmosphere was changed to the air to obtain the A-VO powder.

Synthesis of NiHCF: 0.184 g Ni(CH₃COOH)₂⋅4H₂O and 3 g polyvinylpyrrolidone (PVP) were mixed in 100 ml deionized water. A 100 ml aqueous solution containing 0.164 g K₃[Fe(CN)₆] was then slowly added into the above-mentioned solution with vigorous stirring for 30 min at 25 °C. The sediment was then centrifuged and washed with deionized water and ethanol for at least three times. The NiHCF powder was obtained after drying in an oven at 60 °C overnight.

Assemble of N-VO/PAM/NiHCF quasi-solid-state ammonium ion batteries: 7 g acrylamide (AM) was added to 22 ml deionized water, stirring for 10 minutes to dissolve it. Then, 7 mg N, N′-methylene bisacrylamide was added and stirred for another 30 min. Subsequently, 110 mg of ammonium persulfate was added to the solution and stirred at 0 °C with N₂ injecting for 30 min. The mixed solution was transferred to plastic wrap and dried in an oven at 60 °C for 2 h to obtain a dry polyacrylamide (PAM) gel. Then PAM gel was immersed in 16 ml of 1M NH₄Ac electrolyte for 8 h to obtain NH₄Ac gel electrolyte. Finally, it is assembled according to the order of N-VO anode, NH₄Ac gel electrolyte, and NiHCF cathode. With a vacuum seal, the N-VO/PAM/NiHCF flexible quasi-solid-state battery is obtained.

Characterizations: Scanning electron microscope (SEM) images were acquired by using a Thermal Field Quanta 400F microscope. A Transmission Electron Microscope (TEM, FEI Tencai G2 F30) operated at 300 kV was performed to research the microstructure and the element mapping of the material. The crystal structure was analyzed by the x-ray diffraction (XRD) of LIMF smart Lab. X-ray photoelectron spectroscopy (XPS) was carried out on ESCALab250 using the Alα x-ray as the excitation source. Fourier transform infrared spectrum (FT-IR, NICOLET 6700) was conducted for a chemical bond of the electrode material. Thermal gravity analysis was employed by Thermogravimetry (TG209F1 libra).

Electrochemical measurements: The working electrodes were prepared by mixing 70 wt. % as-prepared material, 20 wt. % conductive carbon, and 10 wt. % polyvinylidene difluoride (PVDF) with a moderate addition of N-Methylpyrrolidone (NMP). The mixture was coated on carbon paper and dried under vacuum at 60 °C overnight. The mass loading of the active material is about 2.5 mg cm⁻². The electrochemical performance of N-VO and A-VO was measured in a three-electrode system with a saturated calomel electrode working as the reference electrode and the graphite rod serving as the counter electrode. Galvanostatic charge/discharge curves (GCD), cyclic voltammograms (CVs), and electrochemical impedance spectroscopy (EIS) are carried out using an electrochemical workstation (CHI 760E). The cycling stability measurements were conducted on a Neware battery testing system (CT-3008-5V10mA-164, Neware, Shenzhen, China). The galvanostatic intermittence titration technique (GITT) was conducted by a series of galvanostatic discharge pulses of 10 s at 1 A g⁻¹, followed by a 10 min rest.

See the supplementary material for the morphological and structural characterizations and electrochemical measurements, including Figs. S1–S12 and Table S1.

This work received financial support from the Science and Technology Planning Project of Guangzhou City (Grant No. 2023A04J1942).

The authors have no conflicts to disclose.

Tzu-Hao Lu: Investigation (lead); Methodology (lead); Writing – original draft (lead). Qiyu Liu: Formal analysis (lead); Validation (lead). Ang Yi: Data curation (lead). Hao Liu: Resources (lead). Yanxia Yu: Funding acquisition (lead); Project administration (lead); Writing – review & editing (equal). Xihong Lu: Conceptualization (equal); Supervision (lead); Writing – review & editing (equal).

The data that support the findings of this study are available within the article and its supplementary material.