Electrochemical Redox Processes Involved in Carbon-Coated KVPO₄F for High Voltage K-Ion Batteries Revealed by XPS Analysis

KVPO₄F and carbon coated KVPO₄F (KVPO₄F-C) were prepared by a two-step reaction. First, VPO₄ was obtained by mixing V₂O₅ (from Sigma-Aldrich, 99.6%) and NH₄H₂PO₄ (from Sigma-Aldrich, 99.99%) by high energy ball-milling (SPEX 8000 M mixer/mill at 1425 rpm) for 1 h followed by a thermal treatment at 800 °C for 7 h under Ar/H₂ (5%). Second, stoichiometric amounts of KF.2H₂O (from Alfa Aesar, 98.5%, dried overnight under vacuum at 250 °C and stored in a glovebox under Ar) and VPO₄ were mixed without (KVPF) or with (KVPF-C) the addition of 10 wt.% sucrose (from Sigma-Aldrich, ACS reagent). The mixture was obtained by SPEX ball-milling for 1 h, pelletized and thermal treated at 650 °C for 8 h under Ar. Rietveld refinement of the X-ray diffraction patterns was performed using the FullProf software and considering an orthorhombic unit cell described in the Pna2₁ space group.²⁴ It confirmed the nature of the grey (KVPF) and black (KVPF-C) powders obtained (Fig. S1 is available online at stacks.iop.org/JES/167/130527/mmedia). The smaller cell parameters and especially volume cell obtained for KVPO₄F-C vs the KVPO₄F phase reported and characterized in-depth by Fedotov et al.¹⁷ suggest a partial O²⁻ substitution for F⁻ and thus the formation of KVPO₄F_1−yO_y. y was estimated to 0.1 from the study of the solid solution KVPO₄F - KVPO₄O (Table SI, and from ³¹P MAS NMR data not reported here). Note that the stoichiometry in K/V/P was determined by chemical analyses (Fig. S1). In addition, thermogravimetric analyses and scanning electron microscopy images confirmed the formation of a carbon coating (2.5 wt.% of C) for KVPF-C compared to KVPF (Fig. S2). Electrodes were prepared by mixing KVPF or KVPF-C, carbon black (Super P, BET = 62 m² g⁻¹, Alfa Aesar) and poly(vinylidene fluoride) according to a 70:25:5 weight ratio, in N-methyl-2-pyrrolidone (32:68 solvent:material weight ratio), by ball-milling for 1 h. The obtained slurry was cast on an aluminium current collector, dried for 24 h under Ar, and finally electrodes of 1.5 mg_KVPO4F/cm² were punch out and dried under vacuum at 80 °C for 12 h. 2032 coin-type cells (316L stainless steel) were assembled under Ar using a Whatman (grade GF/D) and a polypropylene membrane (Celgard) as separators, K metal (99.95%, Alfa Aesar) as negative electrode and 100 μl of 0.8 M KPF₆ in a mixture of EC:DEC by volume as electrolyte (KPF₆, Sigma-Aldrich, ≥99%; EC, Sigma-Aldrich, anhydrous, 99%; DEC, Sigma-Aldrich, anhydrous, ≥99%). All potentials reported in the text are referred to K⁺/K. Note that K metal polarization in symmetrical K//K cell was found negligible during open circuit voltage (<6 mV, Fig. S3) compared to the literature.²⁵ During plating/stripping at 0.04 mA cm⁻² (corresponding to the current used to cycle K//KVPF cells), however, a polarization of 0.1 V was measured. Cycling tests were performed at C/5 (26 mA.g⁻¹, i.e. the exchange of 1 K⁺ in 5 h) and 20 °C using a VMP3 (Biologic, SAS France). Two formation cycles were first performed between 3.5–5 V vs K⁺/K to minimize electrolyte degradation impact (Figs. S4 and 3), as discussed later. KVPF-C cells were then stopped at different potentials during the 3rd cycle and electrodes were recovered from the cells and washed twice with 1 ml of DEC during 30 s. XPS analysis was performed using an Escalab 250 Xi spectrometer, with a monochromatized Al Kα radiation (hν = 1486.6 eV). Electrodes were placed on a sample holder using uPVC insulation tape (3 M part number 655), and transferred to Ar filled glove box connected to the spectrometer. Using the standard charge compensation mode, core spectra were recorded with a 0.15 eV step and a constant 20 eV pass energy. Using CasaXPS software, the binding energy scale was calibrated from the KVPF oxygen peak at 530.0 eV. It is reminded that XPS probes about 5 to 10 nm depth of the sample surface, and here between 6–7 nm for the vanadium, i.e. not only the extreme surface atomic layers (0.5–1 nm) but also the deeper ones, which means a representative part of the bulk is also probed (Fig. S5). Moreover, at the extreme surface, the presence of defects may lead to vanadium with an oxidation state higher or lower than in the bulk. V 2p_3/2 core spectra of KVPF electrodes were thus fitted using peak positions and line shapes constraints corresponding to different oxidation states as determined from reference samples with Vⁿ⁺ in a PO₄ environment (Fig. S5 and Table SII) in agreement with the literature.^26–30 This thorough procedure allows reliably following the evolution of the mean oxidation state upon cycling, especially considering that the mean oxidation state obtained by XPS (3.1) for the pristine KVPF-C material is in good agreement with the XRD analysis (KVPO₄F_1−yO_y with y = 0.1).

The impact of the carbon coating was evaluated regarding its ability to optimize the electrochemical performance of the material in KIBs and to mitigate the KVPF active material surface reactivity against the electrolyte. Overall, both KVPF and KVPF-C showed a similar charge/discharge profile at C/5 with 4 potential domains (Fig. S4), in very good agreement with the literature.¹⁹ However KVPF-C showed a lower average polarization of ∼230 mV compared to ∼300 mV for KVPF (Fig. S4) and a better capacity retention after 70 cycles compared to KVPF (76% vs 62%, (Fig. S6). For both systems, a significant high-potential hysteresis is observed, suggesting that the main capacity loss would be related to the high-potential domain. Since KVPF and KVPF-C have close particles size (Fig. S2), the carbon coating was shown—as expected—to promote better transport properties within the positive electrode, lower polarization of the cell, smaller electrolyte degradation at high voltage and thus optimized performances. XPS analysis confirmed a significantly lower solid electrolyte interphase coverage at the KVPF-C surface compared to KVFP after the first cycle (Table SIII). Note that the lower electrolyte degradation is explained by a passivation effect of the carbon coating at the KVPF surface. Based on these results, KVPF-C was therefore selected to study the electrochemical redox processes involved in KVPO₄F upon cycling.

Figure 1a shows the typical 3rd charge/discharge cycle obtained for K//KVPF-C cells and indicates at which potentials the prepared cells were stopped to perform the analysis of the redox processes involved in KVPF-C. Note that charge/discharge profiles obtained for all the cells showed very close polarization and capacities (Fig. S7), highlighting the great reproducibility of the electrochemical results so that reliable XPS analysis is expected on materials recovered ex situ from this series of cells. Figure 1b shows the V 2p_3/2 XPS core spectra of K_xVⁿ⁺PO₄F-C electrodes as recovered from K//KVPF-C cells during the third cycle at C/5, as function of the K content. For more information, the full V 2p core spectra can be found in Fig. S8. XPS analysis revealed that the pristine KVPF-C showed the presence of V²⁺, V³⁺ and V⁴⁺ (Fig. 1b) although only V³⁺ and V⁴⁺ were expected based on the KVPO₄F_0.9O_0.1-C formula. Note that V²⁺ was also observed for the VPO₄ precursor (Fig. S5). However, considering the very low amount (<1wt.%) of VPO₄ impurity in the KVPF-C material (Fig. S1), the presence of V²⁺ is likely to originate from the KVPF-C extreme surface. This was supported by the decrease of the V²⁺ contribution during oxidation as discussed thereafter.

Zoom In Zoom Out Reset image size

After charge of the cell to 4.10 V, i.e. during the 1st oxidative potential domain, the peak associated with V⁴⁺ increased while the V³⁺ and V²⁺ ones decreased (respectively +7%, −4% and −3%). At almost the end of the 2nd potential domain at 4.30 V (i.e. at the composition of about K_0.625VPO₄F), the peak associated with V⁴⁺ increased while the peaks associated with V³⁺ and V²⁺ decreased (respectively +15%, −12% and −3%). At the end of the 3rd potential domain at 4.46 V (i.e. for approx. K_0.5VPO₄F), a new peak, attributed to V⁵⁺ appeared (+13%) while the V⁴⁺, V³⁺ and V²⁺ peaks decreased (−11%, −1% and −1%, respectively). This phenomenon further continued after the 4th potential domain at 5.00 V, so that the V⁵⁺, V⁴⁺, V³⁺ and V²⁺ peaks represent 20%, 50%, 23% and 7%, respectively. Note that despite the remaining V³⁺ and V²⁺ even at the end of charge, the corresponding mean oxidation states were in good agreement with the theoretical/electrochemical ones, as discussed thereafter. Note that remaining of V³⁺ and V²⁺ at the end of charge should arise from specific vanadium environment at the KVPF surface. During discharge, opposite phenomena were observed (Fig. 1b), indicating a high reversibility of the redox processes involved in KVPF-C upon cycling in the potential window 3.5–5 V vs K⁺/K.

The evolution of the vanadium mean oxidation state as the K content changed during the 3rd cycle for KVPO₄F_0.9-O_0.1C and determined from both the charge/discharge capacity (DO_elec) and XPS analysis (DO_xps) is reported in Fig. 2.

Zoom In Zoom Out Reset image size

Until the end of the 3rd potential domain, i.e. at 4.46 V, a very good agreement was observed between the vanadium mean oxidation states determined from the electrochemistry and from XPS analysis. On the contrary, at the end of the 4th potential domain, i.e. at 5.00 V, DO_elec was +3.9 while DO_xps was only +3.8. This discrepancy indicates that about 25% of the electrons exchanged in the last potential domain are not related to K⁺ deintercalation from KVPF-C but to electrolyte degradation. This is in agreement with the high irreversible capacity observed for the 4th potential domain between the charge and the next discharge (Fig. 1a). Moreover, the evolution of the capacities associated with the different potential domains for the first 20 cycles showed that electrolyte degradation is mainly associated with the 4th potential domain as shown in Fig. 3 by the large difference in height between the full and hatched black bars. Nevertheless, during the first two cycles, irreversibility was observed throughout the reaction, the latter being much more pronounced at high voltage (about 60%, Fig. 3). During the next cycles, a stabilization was observed with an excellent efficiency below 4.6 V (same height for the full and hatched blue and green bars), and thus a capacity loss almost exclusively associated with the high voltage domain (about 90%) and the composition range K_0.5VPO₄F-C—VPO₄F-C. This highlights the relevance of using the 3rd cycle to perform the redox processes study instead of the 1st cycle. During discharge, DO_xps decreased back to +3.1 (Fig. 2), indicating a complete reversibility of the redox processes. This result fully supports the limitation of the participation of the electrolyte degradation to high voltage, only after the 2nd cycle as shown in Fig. 3.

Zoom In Zoom Out Reset image size

Importantly, only 0.7 K⁺ were extracted from KVPO₄F_0.9-O_0.1-C (Fig. 2) instead of 0.9 theoretically. Especially, the capacity associated with the 1st, 2nd and 3rd potential domains nearly reached the theoretical ones and remained almost constant from cycles 3 to 20 (Fig. 3). On the contrary, the 4th potential domain capacity continuously decreased to about half the theoretical one from cycle 3 to 20 (Fig. 3). Interestingly, no polarization increase (due to the accumulation of electrolyte degradation products) was observed (Fig. S9). Thus, the 4th potential domain capacity decrease is not explained by the cut-off potential being reached before the end of the domain. Instead, these results likely originate from structural constraints hindering the extraction (and insertion) of K⁺ from x = 0.5 to x = 0 as proposed previously.^19,31 To evaluate the possibility to further extract the remaining K⁺, a cell was hold at 5 V for 18 h during the 3rd cycle and the corresponding DO_xps was +3.9 (Fig. S10), in agreement with the evolution of the cell parameters determined by XRD (not shown). This result highlights that despite the electrolyte degradation process, it is still possible to extract the potassium but the process is much slower, more likely due to a structural constraint such as cell volume reduction. This structural constraint could increase the K⁺ migration barrier, inducing a decrease in the K⁺ diffusion coefficient,³¹ slowing the global extraction process.

In conclusion, the carbon coating of the KVPF material was found beneficial to the overall electrochemical performance while mitigating electrolyte degradation at high voltage. Then, by comparing the average oxidation state derived from electrochemical and XPS analysis, the electrochemical redox processes of K_xVPO_4.1F_0.9-C were revealed. Interestingly, a high reversibility was found although the K⁺ extraction (and insertion) from x = 0.5 to x = 0 was hindered, possibly by structural constraints. Moreover, the severe electrolyte degradation issue, observed mostly above 4.5 V, is currently under further investigation, as well as possible KVPF dissolution in order to successfully mitigate these irreversible degradation reactions using electrolyte additives and enable highly efficient KVPF based full K-ion cells.

The authors thank the financial support of E2S-UPPA, Région Nouvelle Aquitaine and the French National Research Agency (STORE-EX Labex Project ANR-10-LABX-76-01 and TROPIC project ANR-19-CE05-0026-01). The authors thank Cathy DENAGE, Etienne DURAND and Eric LEBRAUD (ICMCB) for their help with SEM, TGA-FTIR and XRD analyses, respectively. Authors also thank Joël GAUBICHER (IMN-Nantes) for fruitful discussions about redox processes in KVPF.