Synthesis, structural and electrochemical properties of V₄O₉ cathode for lithium batteries

Highlights

• Low temperature ammonia gas chemical reduction of V₂O₅ creates V₄O₉, a newly formed corner shared polyhedral VO_x structure

• V₄O₉ demonstrates reversible smooth lithium (de) intercalation profiles

• Capacities as high as 225 mAhg⁻¹ are demonstrated between 3.5 to 1.8 V vs. Li metal

Introduction

Classic electrochemically active crystalline vanadium oxides such as V₂O₅ and the lithium-containing version, LiV₃O₈, have long been evaluated in lithium batteries as cathodes owing to their high theoretical capacities when cycled to rock salt type composition; i.e., Li₃V₂O₅ and Li₅V₃O₈ (Panero et al., 1983; Delmas et al., 1991; Delmas et al., 1994; West et al., 1995; Kawakita et al., 1998; Jouanneau et al., 2005; Leger et al., 2005; Chernova et al., 2009). For example, the theoretical lithium (de) intercalation capacities for V₂O₅ is 440 mAhg⁻¹, and for LiV₃O₈ is 372 mAhg⁻¹. Thus, notable high battery energy densities can result for full cells when coupled with metallic Li anodes above ca. 481 Wh kg⁻¹. These high energy densities are linked to the electrochemical activity origin associated with multiple V(V) to V(III) redox state changes and ready insertion of lithium cations (Cartier et al., 1990). However, phase transitions transpire between multiple oxidation states that drive structural changes in these materials which can lead to cycling fade (Delmas et al., 1994). On the other hand, another layered phase LiVO₃ has exhibited good capacity retention after the transformation into rock salt phase due to restricted electrochemical activity to V⁵⁺/V⁴⁺ redox couple but with lower capacity (Pralong et al., 2012).

The early work of S. Passerini et al. (Appetecchi et al., 2001) has focused on V₂O₅ cathodes versus Li foil coupled with a thin polymer electrolyte membrane/separator, such as polyethylene oxide (PEO) plasticized with LiTFSI salt (large amorphous polymer region). Perhaps the earliest successful work that employed vanadium oxide-type materials was the use of LiV₃O₈ with both liquid ester carbonate solvents and PEO polymer electrolyte systems (Bonino et al., 1988). Recently nanotechnology enhancements have taken place, and nanoflowers of V₂O₅ have also been reported with extremely high capacities of 275 mAh/g after 50 cycles (liquid ester carbonate electrolyte based) (Tang et al., 2013). Other nanosized materials such as nanobelts (Li et al., 2006), and nanowires (Li et al., 2007) of V₂O₅ tested in lithium batteries have been investigated, but these materials are complicated to synthesize and the volumetric energy density is compromised by the nanometer dimensionality and requisite low electrode loadings (Wang and Cao, 2006). Micron-sized hollow spheres of V₂O₅ were also reported (Zhu et al., 2008), thus demonstrating that vanadium oxides can reach an effective capacity of 319 mAh/g between 4.0 and 2.0 V cutoffs. After 50 cycles, however, the capacity diminishes to about 210–220 mAh/g in Li cell. Lastly, VO₂ (B) phase is electrochemically active and has been studied as a novel cathode which converts to a Zn-VO_x material in a aqueous Zn battery (Ding et al., 2021).

In the context of Li metal based liquid electrolyte cells, the promise of 500 Wh/kg pouch-format cells is within reach with high Ni cathodes. (Liu et al., 2019). Certainly Li metal allows such an extreme energy dense system, but also brings in the opportunity to use charged (non-lithium containing) electrodes like V₂O₅ or other vanadium oxide phases. An exploitation of the high capacity from two electrons from V(V) to V(III) in such systems may be possible. The Li cells are discharged first thus inserting lithium cations into the host vanadium oxide. Thus far, however, many issues of constricted Li-ion motion, vanadium atomic movement and resultant coordination distortion causes phase changes in the host material (De Jesus et al., 2018), which adversely effects stability and lowering of capacity. In this present work we focus on the synthesis of a different structure; V₄O₉ with large one-dimensional tunnels, and present the electrochemical cycling, and associated mechanism of structural change for the first time. Cycling leads to a more stable crystalline lithiated rock-salt phase with good reversibility and high capacities in Li cells. We compare the capacity retention and values to battery-grade micron-sized polycrystalline V₂O₅ electrode and lend evidence to why the cycling is superior. We find that the V₄O₉ follows a different reaction pathway than V₂O₅, resulting in a more stable structure amiable to reversible and smooth (de) intercalation processes.

Experimental

Synthesis of V₄O₉ was carried out by reducing V₂O₅ precursor under a flow of 3 mol% ammonia (NH₃) in argon (Ar) gas stream (AGA gas; 99.9%). The ammonia/Ar gas flow was 20 cc/min. A gas outlet trap from the tube furnace was filled with dilute acetic acid to neutralize the effluent ammonia gas. Typically 500 mg of V₂O₅ (Sigma, 99.6%) was placed in an alumina crucible inside a tubular furnace and heat treatment was carried out at different temperatures for 12 h. Vanadium oxide (V₂O₅), was purchased (Sigma-Aldrich 99%) and used without further purification. It also was ground and sieved prior to electrode making and electrochemical testing. The purity grade for the electrochemical sample was slightly lower, 99%, versus 99.6% for the synthesis precursor sample. The impact on capacity value is expected to be less than 1%, as the impurities are less than 1%.

Structural characterization was carried out using a Rigaku lab diffractometer (Cu Kα radiation) at a scan rate of 2.0°/min in between 10° and 60° 2θ range. Additional synchrotron-based powder SXRD patterns were collected from 11-BM (Sector 11 bending magnet line) at the Argonne National Laboratory’s Advanced Photon Source. The SXRD patterns were fitted and indexed using Le Bail whole-powder pattern decomposition method.

Electrochemical tests were carried out in coin cell configuration (2032 size; Hohsen) assembled in an argon-filled glovebox (H₂O and O₂ < 0.1 ppm). Cathodes were prepared from a slurry mixture (solvent: N-methylpyrrolidinone; Sigma-Aldrich, 99 + %) of 80wt% V₄O₉ powder, 12wt% carbon black (Super P-Li, Cabot Co.,) and 8wt% poly (vinylidene fluoride) (PVDF, Kynar) which was then coated on an aluminum foil current collector. The electrode laminate was dried at 70°C in a vacuum oven, calendared and punched out as a disc (9/16 in. diameter) and contained an active mass of 3 mg cm⁻². The anode was a lithium metal disc (Hohsen), and the electrolyte (Tomiyama chemical) consisted of 1.2 M LiPF₆ in a mixture of ethylene carbonate and ethyl methyl carbonate (3:7 by weight). These coin cells were set to cycle between 1.8 and 3.5 V vs. Li⁺/Li⁰ at C/10 rate. Likewise the V₂O₅ cell had the electrode configuration 84wt% active, 8wt% PVDF binder and 8wt% Super-P carbon black; it was cycled between 3.5 and 1.8 V at C/10.

⁷Li MAS-NMR experiments were performed at 7.02 T (300 MHz) on a Bruker Avance III HD spectrometer operating at a Larmor frequency of 116.64 MHz, using a 1.3 mm MAS probe. All spectra were acquired at 67 and 60 kHz with a rotor synchronized echo pulse sequence (90°-τ-180°-τ-acq), where τ = 1/ν_r. A π/2 pulse width of 1.6 μs was used with sufficiently long pulse recycle delays of 1 s. Spectra were collected immediately after drying (1–2 h) with 3,072 scans at a constant temperature of 283 ± 0.1 K. Chemical shifts were referenced to 1 M LiCl at 0 ppm. The spectra were normalized by the total number of scans which was the same for all runs and the weight of active materials packed in the rotors for the best possible quantitative analyses of intensity changes of each lithium resonance. Deconvolutions on the spectra were attempted in order to further gain insights into specific Li-coordination(s) via introducing least number of resonances to account for the lineshapes and asymmetries observed (best overlap > 96.5%).

Results and discussion

The synthesis of single phase V₄O₉ is complex and challenging as its formation depends on temperature as well as the strength and type of reducing agents (Marezio et al., 1973; Kawashima et al., 1975; Yamazaki et al., 2010). Improper chemical treatment leads to formation of multiple vanadium oxide phases in the sample. Herein, we found that a soft reducing agent, i.e., ammonia/argon or NH₃/Ar gas stream can be used to synthesize single phase black color V₄O₉ simply from V₂O₅ powder precursor (Figure 1A). Figure 1B shows XRD patterns collected on vanadium oxide samples which are heat treated for 12 h at different temperatures. The pattern collected on the sample heat treated at 150°C showed the presence of additional reflections along with V₂O₅ precursor at 2θ = 27.4° and 28°. As the temperature was raised to 250°C, the intensity of new reflections increases at the expense of V₂O₅ precursor and at 300°C, only the former is visible and the corresponding new phase is identified to be V₄O₉. However, increasing the annealing temperature to 400°C resulted in another phase transformation from V₄O₉ to V₄O₇ (Liu et al., 2019). Thus, the product phase formed is critically tied to the reaction temperature.

FIGURE 1. (A) Scheme of reduction reaction of V₂O₅ under the stream of 3 mol% NH₃ in a balance of argon and (B) XRD patterns of vanadium oxide precursor and heat treated samples at different temperatures.

Vanadium oxide V₄O₉, consists of an average oxidation state of + 4.5 thus possibly suggesting oxygen vacancies in the material. In fact, oxygen vacancies have been tied to better performance for oxygen deficient V₂O₅ (or V-V₂O₅); with ∼ 3% more oxygen deficiency than pristine V₂O₅ (based on deconvolution of XPS spectra) (Sun et al., 2018; Chotsawat et al., 2022). Certainly, at only 3% more O vacancies than pristine V₂O₅, the resultant global long-range structure V-V₂O₅ is hardly affected. The unit cell parameters of pristine V₂O₅ are slightly smaller than V-V₂O₅ due to slightly more (larger) V⁴⁺ ionic radii. As compared to V₄O₉ with theoretically 10% O vacancies (in V₂O₅ nomenclature), the comparisons end there. Moreover, the V-V₂O₅ voltage profile, just like pristine V₂O₅ contains steps and kinks associated with phase changes with resultant low voltage output (Sun et al., 2018), unlike V₄O₉ voltage profile which will be discussed later.

Figure 2 shows the SXRD pattern of V₄O₉ collected using a synchrotron X-ray source. The powder pattern could be fitted well with the Cmcm space group where lattice parameters were found; a = 10.381 Å, b = 8.196 Å and c = 16.588 Å, in accordance to the reported values (Marezio et al., 1973; Kawashima et al., 1975; Yamazaki et al., 2010). Our V₄O₉ has a much larger cell volume compared to other materials made by soft chemistry routes (Wang et al., 2021), which may explain why Li ion conduction and smooth voltage profiles are observed. The crystal structure (Figure 2 inset) contains V⁴⁺ cations in both pyramidal VO₅ and octahedral VO₆ coordination, and V⁵⁺ cation in tetrahedral VO₄ coordination (The Materials Project, 2022). In the ab-plane, VO₅ square-based pyramids and VO₆ octahedra share edges to make pairs which are then connected to VO₄ tetrahedra through corners to form layers. These layers are further linked together via corner sharing of VO₄-VO₄ tetrahedra to form double layers along the c-axis. Further, VO₅ pyramids and VO₆ octahedra present in the double layers are sharing corners to make three-dimensional structure with hexagon-shaped cavities or large tunnels along the b-axis.

FIGURE 2. Synchrotron XRD pattern collected on V₄O₉ sample prepared at 300°C. The red dots and black and blue solid lines correspond to observed and calculated patterns and their difference, respectively. (Crystal structure of V₄O₉ is shown in inset).

For determining the V₄O₉ electrochemical lithium ion intercalation properties, cathodes were fabricated and cycled against Li metal at C/10 between 1.8 and 3.5 V vs. Li⁺/Li⁰. Representative data is shown in Figure 3A. During the initial discharge, the voltage gradually drops to the cutoff of 1.8 V with a capacity of 320 mAh/g. On the subsequent charge, the voltage raises smoothly back to 3.5 V vs. Li⁺/Li⁰ with a return capacity of 286 mAh/g. The coulombic efficiency cycle-to-cycle after the first cycle is above 99%. Approximately 4.2 M ratio of Li⁺ has been intercalated into the host on discharge, while only 3.7 M ratio can be extracted; these values are consistent with previous literature (Hammouche and Hammou, 1987). However, the voltage profile shapes are different as the former possesses a multitude of voltage steps whereas the material synthesized and evaluated here (i.e., “Li3.7V4O9”) exhibits, instead a classic S-shaped curve. To better explain this result, the corresponding dQ/dV curves are shown in Figure 3B which includes a set of voltage peaks or features at 2.63, 2.52, 2.38, and 2.04, and 2.21, and 2.57 V vs. Li⁺/Li⁰ during first discharge and charge respectively. These slight voltage features are eliminated or significantly muted on the second cycle.

FIGURE 3. (A) Voltage profiles of Li/V₄O₉ & Li/V₂O₅ cells at C/10 rate in between 1.8 and 3.5 V (B) the dQ/dV (first two cycles) profiles of Li/V₄O₉ cell, and (C) capacity vs. cycle number plots of V₄O₉ and V₂O₅ cathodes.

These electrochemical results are compared to amongst the best and most recent cycling performance reported in the literature on morphologically-optimized high-performance V₂O5 that is made in a nano-plate stacked form, whereby the capacity is stabilized at 258 mAhg⁻¹ (4.0–2.0 V; 50 mAg⁻¹) (Sim et al., 2023). If a nanocomposite of V₂O₅ made by a hydrothermal method is synthesized in the presence of conductive reduced graphene oxide (rGO) scaffold then an even higher capacity of 280 mAhg⁻¹ (4.0–2.0 V; 50 mAg⁻¹) was observed (Alsherari et al., 2023). Both of these recent results used nano-architecture designed materials, unlike the micron-sized V₄O₉ particles and their electrochemical cycling results presented here.

For the electrochemical-structural mechanism, we compare the pathway of phase transformation for the layered V₂O₅ prototype form versus that of V₄O₉ three dimensional vanadium oxide. Both phases proceed to cation disordered rock salt type structures upon different stages of lithiation. For example, V₂O₅ undergoes a series of phase transformations depending upon the amount of lithium inserted, α- (x < 0.01), ε- (0.35 < x < 0.7), δ- (x = 1), γ- (1 < x < 2) and ω- (x = 3) (Delmas et al., 1994). The initial transformations up to two Li per formula unit does not significantly alter the pristine structure and the intercalation occurs by puckering of VO₅ pyramid layers. As the intercalation proceeds beyond this limit, an irreversible phase transformation results in the formation of rock salt type structure with the end composition Li₃V₂O₅. This is clearly shown from the change in the voltage profile from the initial discharge to subsequent cycles as shown in Figure 3A. The puckering is too significant with deeply intercalated lithium and the vanadyl–V(V) = O bonding unit cohesiveness strength is lost. Thereafter the material still cycles, but the capacity fades somewhat significantly after 20 cycles.

As demonstrated, significant changes in the voltage profiles are observed for these comparative vanadium oxide cathodes which could be attributed to successive phase transformations upon cycling (Delmas et al., 1991; Delmas et al., 1994). To compare and contrast the V₄O₉ results, we collected ex situ synchrotron XRD patterns on fully discharged and charged V₄O₉ electrodes during the first cycle (Figure 4). Both patterns could be partially indexed with cubic space group Fd-3m which is characteristic of disordered rock salt structure, as previously observed for other Li-V-O systems, such as V₂O₅ and LiVO₃ (Chieh et al., 1981; Delmas et al., 1991; Pralong et al., 2012). The corresponding cell parameters of discharged and charged phase have found to be a = 8.228 and 8.197 Å, respectively. In addition, both patterns consist of a few un-indexed peaks which could be due to the formation of secondary phase upon electrochemical activation. These peaks do not shift or change in between discharge and charge process, thus confirming the inactivity but entrapment of some lithium as observed in the electrochemical measurements during the first cycle. Peak broadening indicates crystallographic stresses have ensued upon lithium intercalation into the host material, but does not adversely affects the electrochemical activity.

FIGURE 4. Ex situ synchrotron XRD patterns collected on (A) discharged and (B) charged V₄O₉ electrode samples. The red dots and black and blue solid lines correspond to observed and calculated patterns and their difference, respectively. The broad bump at 7° 2-theta is due to the Kapton tape that is protecting the sample from ambient air.

To shed some preliminary insights into the detailed distribution of lithium ions in this disordered rock salt phase, ⁷Li MAS NMR measurements were performed on discharged and charged samples (Figure 5). Deconvolution of the spectra show a major broad resonance peak at −11.9 ppm and two minor peaks at −40 and 45 ppm, thus indicating a wide distribution of various Li cations in the structure with an envelope of associated chemical shifts. The results suggest multiple broad distributions of Li sites are available in the host lattice for lithium occupancy with respect to cycling. The intensity of these resonances diminishes to approximately 25% upon charging indicating Li removal from these sites but without disturbing the lattice. Note that the signal at 0 ppm is due to Li-containing diamagnetic species from the electrolyte/SEI.

FIGURE 5. Ex situ ⁷Li MAS NMR spectra collected on discharged (black) and charged (red) V₄O₉ electrode samples. * indicates spinning sidebands. Insets show corresponding deconvolutions of the spectra.

In the electrochemical cell, V₄O₉ is expected to follow the similar multi-step lithium intercalation process as observed in the case of V₂O₅ before transforming into rock-salt type structure. We have confirmed this mechanism by ex situ XRD measurements. From the electrochemical results, we found that the activated rock salt phase-“Li3.7V4O9” exchanges ∼ 0.93 mol of Li⁺ per mol vanadium in the structure at an average intercalation voltage of 2.5 V vs. Li⁺/Li⁰. This is ascribed to the activity of V⁵⁺/V⁴⁺ redox couple as reported in the literature (Pralong et al., 2012). The voltage profiles of subsequent cycles exhibit S-shaped curves-a characteristic of solid-solution formation upon intercalation and de-intercalation phenomena as confirmed by our ex situ synchrotron XRD studies. Thanks to the low volume changes in the unit cell (ΔV/V ∼ 0.34%) and restricted redox activity confined to V⁵⁺/V⁴⁺couple (capacity value driven), the rock salt phase yields a capacity of 225 mAh/g even after 50 cycles (Figure 3C), which is higher than battery grade micron-sized V₂O₅.

Conclusion

In summary, we report a simple methodology to prepare V₄O₉ cathode by utilizing NH₃/Ar gas stream. The as-prepared cathode was then transformed into disordered rock salt type “Li3.7V4O9” upon electrochemical lithiation. Once it has formed, it reversibly exchanges lithium ions at an average intercalation voltage of 2.5 V vs. Li⁺/Li⁰ with an initial capacity of 260 mAh/g. With restricted redox activity V⁵⁺/V⁴⁺ and low volume changes, the disordered rock-salt type cathode renders a reversible stable capacity of 225 mAh/g after 50 cycles.

Data availability statement

The raw data supporting the conclusion of this article will be made available by the authors, without undue reservation.

Author contributions

CJ and PS conceived the idea. PS wrote the original manuscript. CJ edited the manuscript and provided the efforts to receive funding. PS conducted the synthesis, structural characterization, and electrochemical characterization. EL assisted in structural characterization and synthesis. BK conducted NMR experiments and section writing.

Acknowledgments

This work was supported as part of the Joint Center for Energy Storage Research (JCESR), an Energy Innovation Hub funded by the U.S. Department of Energy, Office of Science, Basic Energy Sciences. This research used resources of the Advanced Photon Source, a U.S. Department of Energy (DOE) Office of Science User Facility operated for the DOE Office of Science by Argonne National Laboratory under Contract No. DE-AC02-06CH11357. The submitted manuscript has been created by UChicago Argonne, LLC, Operator of Argonne National Laboratory (“Argonne”).

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

References