High Rate Lithium Ion Battery with Niobium Tungsten Oxide Anode

NWO powder was synthesized as previously reported in the literature. ¹⁷ NMC-622 (Targray) and a commercial source of carbon-coated LFP with 0.5 um particle size were used as cathodes (Fig. S1 available online at stacks.iop.org/JES/168/010525/mmedia). The XRD analysis of all active materials confirm that they are phase pure as shown in Fig. S2. High loadings electrodes were prepared mixing Super P carbon (TIMCAL) and polyvinylidene fluoride (PVdF; Kynar) dispersed in N-methyl-2-pyrrolidone. All slurries were composed of 90% active material, 5% super P and 5% PVdF binder, and the mixing was conducted with a Thinky Mixer 250. The NMC and LFP electrodes were dried in an oven at 80 °C for 2 h in a dry room, and the NWO electrodes were dried in an oven at 60 °C overnight under ambient atmosphere. All electrodes were calendared at room temperature and the resulting electrode porosities were in the range of 35%–45%. The electrodes loadings were 8.0–8.3 mg cm⁻² (NMC), 8.4–8.7 mg cm⁻² (LFP) and 8.8–9.4 mg cm⁻² (NWO) with 1 cm² electrode size. Low loading electrodes were prepared in a similar fashion but with a different composition namely having 10% Li250 carbon (Denka), 10% PVdF (Kynar) dispersed in N-methyl-2-pyrrolidone, and 80% of active material. The resulting active material loading were 4.0 mg cm⁻² for NMC, 3.8 mg cm⁻² for LFP and 4.4 mg cm⁻² for NWO.

All electrochemical measurements were evaluated with 2032-type stainless steel coin cells. Prepared cathode and anode electrodes were dried at 100 °C for 3 h under vacuum, then transferred into an argon-filled glove box (MBraun) without exposure to air. Full cells were assembled in the glove box with LP30 electrolyte (Sigma-Aldrich, battery grade), which is composed of 1.0 M lithium hexafluorophosphate (LiPF₆) in ethylene carbonate (EC): dimethyl carbonate (DMC) (1:1 v/v). A polyethylene separator (Toray) was used after drying at 40 °C for 2 h under vacuum. For electrolyte analyses, glass fiber filter paper (Whatman, GE) was used as the separator. The filter was also dried at 150 °C under vacuum in a drying oven (Buchi). For half cells, Li metal chip (250 μm, PI-KEM), LP30 electrolyte and glass fiber filter were used as the counter electrode, electrolyte and separator, respectively. Galvanostatic electrochemical tests were conducted at various current densities by using a galvanostat/potentiostat (BioLogic) in a temperature-controlled oven at 10, 25 and 60 °C. All cells had a negative-to-positive capacity ratio of 1.1–1.2, which is calculated based on the practical capacities of the active materials, i.e. 171.3 mAh g⁻¹ for NWO, 175 mAh g⁻¹ for NMC and 165 mAh g⁻¹ for LFP. Full cell capacities in this study are calculated by active material mass of the cathode. In full cell tests, charge was performed under constant current and constant voltage mode (CC/CV, with the CV step limited by specific current to > C/10) and discharge was operated under constant current (CC) mode. For symmetrical cell tests, two full cells having the same loading were operated at 0.2 C, and impedance was measured at 2.0 V during the charging step. Frequencies from 1 MHz to 100 mHz were scanned with an applied amplitude of 10 mV. Afterwards, cells were disassembled in the glove box, and two symmetric cells were assembled with fresh LP30 electrolyte. Electrochemical impedance was measured again on the symmetric cells under the same conditions. For half cells, working voltages were 1–3 V vs Li/Li⁺(NWO), 3.0–4.2 V vs Li/Li⁺ (NMC), and 2.8–4.0 V vs Li/Li⁺ (LFP).

For characterization, ex situ samples were prepared as follow: cells were disassembled and rinsed with DMC, then dried in the glovebox pre-chamber under vacuum. Electrode microstructures were analyzed by scanning electron microscopy (MIRA3, TESCAN) at 5.0 kV with secondary electron detection. Electron microscopy images of the pristine electrodes of NWO, NMC and LFP are shown Fig. S1. X-ray diffraction patterns of pristine and cycled electrodes were obtained in Bragg–Brentano geometry (Empyrean, Panalytical) at ambient temperature with a Cu-Kα source. The operando XRD measurement at high temperature was performed in the same instrument using a home-made cell with a Be window and heating bars. The scan time was 10 min. Lattice parameters, phase and purity of the material were determined by Rietveld refinement (or Lebail on the NWO material due to its large unit cell) using the FullProf software. ²² X-ray photoelectron spectroscopy (XPS) was carried out with a Thermo Fisher Scientific instrument with a monochromated Al-Kα X-ray source (hν of 1468.7 eV). Samples were adhered to conductive carbon film taped onto an airtight transfer chamber equipped with an X-ray-transparent window.

For electrolyte analyses, the glass fiber separator was extracted from cell and soaked in 0.75 ml of DMSO-d₆ (Sigma-Aldrich, 99.9 atom% D, 99% CP) for 5 min, after which the solution was transferred into an air-tight J-Young tube. ¹H, ¹⁹F{¹H} and ³¹P{¹H} NMR spectra were recorded on a Bruker Avance III HD 500 MHz Smart Probe spectrometer using a BBO probe. ¹H decoupling was achieved using the Waltz-16 decoupling sequence. ¹H NMR spectra were internally referenced to DMSO-d₆ at 2.50 ppm and the ¹⁹F and ³¹P NMR spectra were internally reference to LiPF₆ at −74.5 ppm and −145 ppm, respectively.

To assess the stability of NWO anodes upon cycling, LFP cathodes, which are known to show stable cycling without significant degradation, were initially used in a full cell (1.54 mAh capacity). Figure 1a shows the voltage profiles of the charge and discharge reaction, and the operating window was determined to be between 0.5 to 2.6 V, with >90.8% of the capacity being obtained between 1 and 2 V on discharge at C/5 (34.2 mA g⁻¹). Figure 1b shows the discharge capacity as a function of C-rate at 25 °C, the cells showing discharge capacities of 140 mAh g⁻¹ at 0.2 C and 130, 125, 112, 92 and 39 mAh g⁻¹ at 1 C, 2 C, 5 C, 10 C and 20 C, respectively. The cell recovers a capacity of 128 mAh/g when the C-rate is returned to 1 C. The discharge capacity of this NWO/LFP cell at 20 C is significantly reduced, but could be improved by optimizing the electrode structure as suggested by the C-rate performance of lower electrode loadings which delivers up to 89 mAh g⁻¹ at 20 C (Fig. 1b).

Zoom In Zoom Out Reset image size

Based on this result, the long-term cycling performance of NWO/LFP cells was investigated for 1000 cycles at a rate of 5 C and 10 C over the full state-of-charge range. At 5 C, 92.1% of capacity retention is observed at 1000 cycles, 3.9% of the original capacity being lost after 100 cycles and 7.9% at 500 cycles (Fig. 1c). However, less than 0.01% capacity loss was observed during the final 500 cycles. In other words, after the initial 500 cycles, only 0.00002% of the capacity decays per cycle. The capacity retention was further checked by cycling the cell at 1 C after the initial 1000 cycles, the cell recovered 96.5% of the original capacity. Full cell operation at 10 C demonstrated capacity losses of 6.8%, 9.2% and 11.1% for 100, 500 and 1000 cycles, respectively; the corresponding coulombic efficiency remained constant at 99.99% up to 1000 cycles. Note that XRD patterns of the NWO electrode after 200 and 500 cycles at 5 C show that the crystal structure is retained with no peak broadening and signs of amorphization (Fig. 1d). These results clearly illustrate the excellent long-term cyclability of NWO/LFP full cells up to 1000 cycles, which directly demonstrates the applicability of NWO materials as anodes in Li-ion full cells.

At this stage we compare these cell performances with LTO/LFP full cells since LTO has a similar working voltage to NWO but is already commercialized in battery technology. Pohjalainen et al. have reported LTO/LFP full cells with a discharge capacity fade of 5% under 1 C for 300 cycles. ²³ Wang et al. have reported that a carbon-modified LFP and unmodified LTO anode exhibited capacity loss of 11% at 0.5 C over more than 1200 cycles. ²⁴ However, in another example, cells constructed with submicron spherical secondary LTO particles and commercial LFP achieved excellent performances (2% capacity loss for 8000 cycles at 30 C). ²⁵ The present high loading NWO/LFP cells, without optimization of particle size, morphology, binder chemistry or electrolyte composition, shows comparable or better cycle life at higher C-rates than reported for many of these LTO/LFP cells. These results are extremely promising and call for further optimization to investigate the full potential of this chemistry.

To further increase the energy density of the full cell, the NWO anode was tested against an NMC-622 cathode, which has emerged as a leading cathode for high energy density cells against a standard graphite anode. ^26,27 The voltage windows of both electrodes were determined by using a three-electrode cell with a lithium metal reference electrode (Fig. S3a). Figure 2a shows the voltage profile of the NWO/NMC cell (1.78 mAh cell capacity). At a 0.2 C rate, it shows a discharge capacity of 175 mAh g⁻¹ in the working voltage range of 1–3 V, with > 91.6% of the capacity being accessed above 1.5 V and an initial coulombic efficiency of this NWO/NMC cell of 87.6% (Fig. S3b).

Zoom In Zoom Out Reset image size

Figure 2b shows the discharge capacity as a function of C-rate obtained from the NWO/NMC cell. At 0.2 C, the cell showed discharge capacity of 175 mAh g⁻¹, with the successively decreasing capacities of 163, 157, 148 and 136 mAh g⁻¹ at 1 C, 2 C, 5 C and 10 C, respectively. A noticeable drop in capacity to 43 mAh g⁻¹ was obtained upon 20 C cycling. Similarly, to the LFP/NWO full cell, we attribute this drop to kinetic limitation in the electrode since i) close to the initial capacity (162 mAh g⁻¹) could be recovered on returning to 1 C, and ii) 152 mAh g⁻¹ at 20 C is obtained for a full cell with a lower loading. Capacity retention and coulombic efficiency at 5 C for 500 cycles were 86.8% and 99.9%, respectively (Fig. 2c). This means a stable cycling performance with a 0.026% of capacity loss per cycle at a 5 C rate operation.

To gain further insights into the structural changes and the possible sources of the degradation of the NWO/NMC full cell, ex situ XRD and SEM analyses were conducted on the extracted electrodes from the cells that were cycled 500 times over the full state-of-charge range. The XRD pattern from the cycled NWO electrode (Fig. 2c) shows no indication of any new phases; however, the monoclinic lattice has expanded by 1.2% in comparison to that of the pristine material, according to Rietveld refinements (Fig. S4 and Table SI), most likely because of the presence of Li⁺ in the structure. ¹⁷ On the NMC side (Fig. 2d), Rietveld refinement shows a shrinkage along the a- and b-direction and an elongation along the c-direction compared to the pristine material, which is consistent with Li deficiency. ^28–31 Based on the volume changes, it can be proposed that, at the end of the 500th discharge, a quantity of lithium equivalent to 25 mAh g⁻¹ has not been removed from NWO while 42 mAh g⁻¹ is missing from the NMC (detailed in Table SI caption). This suggests that the capacity loss is due to (i) active Li loss and (ii) accumulation of the Li in the NWO electrode.

To investigate the source of cell degradation further (NWO anode or NMC cathode), SEM images of the pristine and 500-cycled NWO anode are compared (Figs. 3a, 3c). The NWO particles having an elongated shape, due to a preferential growth of the crystal along the b direction, show several microcracks and a optically clean surface in both the pristine material and after cycling. On the other hand, SEM images (Figs. 3b, 3d) of the pristine and 500-cycled NMC cathodes demonstrate a clear partial destruction of the spherical particles. Particle cracking during cycling is a relatively well-known degradation phenomenon observed from NMC cathode materials; cracks on the particle surfaces can cause, or accelerate, the dissolution of transition metals such as Mn and Ni. Furthermore, the electrical contact between the active material and the conductive carbon can be lost resulting in degradation of the cell performance. ^29,30,32

Zoom In Zoom Out Reset image size

Electrochemical impedance measurements were conducted in full cells and symmetric cells. Cell impedance was measured in symmetric cells which were assembled with electrodes taken from two full cells operated under the same conditions. Two sets of cells were prepared: one was tested after 2 cycles at 0.2 C and the other was tested after 300 cycles at 5 C. Figure 4a indicates the full cell impedance obtained after the 2nd and 300th cycles; Figs. 4b and 4c represent the cell impedance from cathode and anode symmetric cells, respectively. The data were fitted assuming an equivalent circuit ³³ composed of the bulk resistance (R_b), electrode surface resistance (R_s) and charge transfer resistance (R_ct) as shown in Fig. S5; the fitted parameters are summarized in Fig. 4d. In a full cell, the increase in R_ct upon cycling is the dominant source of the increase in overall cell impedance. By cycling the same electrode films in symmetric cells, the R_ct in the NWO symmetric cell only increases from 1.6 to 3.5 Ω cm² while R_ct in the NMC symmetric cell increases from 2.9 to 11.9 Ω cm². This indicates that the increase of the full cell resistance is mainly due to the increase in charge transfer resistance at the NMC cathode rather than at the NWO anode. Note that the increase of the charge transfer resistance at the NMC cathode is consistent with previous literature and explanations for this include a loss of electrical contact by particle fracture and/or the formation of a resistive surface layer. ^21,26,27

Zoom In Zoom Out Reset image size

Next, in order to evaluate the performance of the cell at different temperatures, NWO/NMC and NWO/LFP cells were tested at 10 °C, 25 °C and 60 °C (Fig. 5). The discharge capacities of the two cells tested at different C-rates at 60 °C and 10 °C show higher and lower relative values, respectively, as compared to the 25 °C cell, as expected from the thermal activation of the kinetics of charge transfer and diffusion. Long-term cycling performance of NWO/NMC was evaluated for 300 cycles under three different conditions: 10 C rate at 60 °C (Figs. 5b) and 5C rate at 10°C and 25 °C (Fig. 5c). Cell cycling at 60 °C (10 C rate) resulted in a loss of 30.8% of discharge capacity after 300 cycles while the cells at 25 °C (5 C rate) and 10 °C (5 C rate) showed 9.2% and 15.5% capacity loss in the same cycling window. Variable-temperature cycling of NWO/LFP was conducted for 1000 cycles at 5 C rate. At temperatures of 10 °C, 25 °C and 60 °C (Figs. 5e), 6.9%, 7.9% and 18.1% capacity losses were observed over 1000 cycles, respectively. This indicates NWO is suitable at different temperatures and that the NWO/LFP combination has better cycling stability and operational temperature range than NWO/NMC with the electrolyte used here.

Zoom In Zoom Out Reset image size

Owing to the excellent cycling stability of the LFP/NWO full cell at 60 °C, we conducted an operando XRD experiment at this temperature during the 1st cycle of a NWO half-cell at C/5 (Fig. 6a), and plotted the results of the Lebail refinements (Figs. 6b, 6c). A solid solution mechanism is observed across the full cycle, with an increase of the b parameter and a more complex variation of the a and c parameters. Such evolution is very similar to the room temperature behaviour from Ref. 17, as shown in Fig. 6c in which the trace of the unit cell volume variation during Li insertion/extraction at 60 °C is nearly superimposed with the one recorded at 25 °C. The similar structural evolution is consistent with the good cycling behaviour at 60 °C and also calls for the investigation of the temperature limit at which this material can cycle without structural decomposition.

Zoom In Zoom Out Reset image size

To understand the degradation of the electrolyte during full cell cycling, NMR and XPS analyses were carried out to track the presence of dissolved and surface species, respectively.

¹H, ¹⁹F and ³¹P NMR spectra of the electrolyte extracted from NWO/NMC cells cycled at 5 C for 300 cycles at 60 °C and 500 cycles at 25 °C are compared to the spectra of a pristine electrolyte (Fig. 7). On the ¹⁹F and ³¹P spectra no extra peaks are observed after cycling. The ¹H spectra of the cycled samples present a small peak at 4.3 ppm originating from trace quantities of lithium ethylene dicarbonate (LEDC) likely due to hydrolysis of EC with trace moisture. Integration of the ¹H peak area gives an estimated content of LEDC:EC as 0.1 mol.% (300 cycles, 60 °C) and 0.05 mol.% (500 cycles, 25 °C).

Zoom In Zoom Out Reset image size

F 1s, C 1s, O 1s, Li 1s and P 2p photoelectron spectra of NWO anode from cycled NWO/NMC cells (Fig. 8) show the presence of surface species after cycling at room temperature (200 cycles) as well as 60 °C (300 cycles). The F 1s spectra show that LiF forms after the 1st cycle but does not change significantly after 200 cycles (Fig. 8a). However, a new peak appears around 686.4 eV due to PO_xF_y species after 200 cycles, presumably arising from the degradation of LiPF₆ via reaction with moisture or oxygen. ^34,35 Formation of PO_xF_y species is also confirmed by P 2p spectra ³⁶ after cycling at room temperature (Fig. 8b). On cycling at 60 °C, the extent of LiPF₆ degradation has increased, as seen by a significant rise in both the F 1s and the P 2p XPS intensity due to PO_xF_y components. In addition, the O 1s spectra for the 60 °C cycled sample shows a large increase in peak intensity at 534.4 eV arising from singly-bonded oxygen moieties, including P–O species ³⁷ (Fig. 8c). The C 1s and O 1s spectra show that the increase in intensity of C–O and C=O species at the end of cycling could be due to the accumulation of alkyl carbonate and lithium carbonate species (Figs. 6c and 6d). C 1s XPS of the 60 °C cycled sample shows an intense peak around 288.9 eV assigned to O–C=O bonds, ³⁶ likely to be originating from lithium ethylene dicarbonate (LEDC) as also observed in the NMR studies. Li 1s shows a broad band from 54–58 eV that includes contributions from LiF and ROCO₂Li surface species (Fig. 8e). ³⁶ A trace amount of Li₂O is detected by O 1s and Li 1s spectra after long-term cycling at room temperature.

Zoom In Zoom Out Reset image size

Overall, it can be concluded that i) a very small amount of EC is being reduced at the NWO surface leading to a trace amount of LEDC in the electrolyte and the accumulation of lithium alkyl carbonate at the surface of electrode, ii) residual water leads to the formation of LiPF₆ decomposition products such as deposited PO_xF_y species and soluble [PO₂F₂]⁻. ^38,39 This suggests only minor electrolyte decomposition even in the absence of additives.