For ASSLB, the use of solid electrolytes requires them to be stable at high potentials vs Li
+/Li and compatible with high voltage cathode materials (NMC, NCA) operating up to 4.2/4.3 V
[96][119][120] while remaining compatible with the lithium metal anode. Surprisingly, despite high oxidation limits claimed in most publications (largely above 4.0 V vs. Li
+/Li), the electrochemical characterization of these materials in a battery set up is usually performed using LFP cathodes, which have a redox potential of 3.45 V vs. Li
+/Li. The fact that the well-known cathode material NMC, which has a potential window of 3.0–4.3 V vs. Li
+/Li, is not used in these publications to further qualify the stability of their electrolyte leads to questions regarding the validity of the ESW measurement.
The purpose of this section is to highlight articles that are specifically dedicated to the evaluation of the ESW of electrolytes. This chapter is focused on describing “advanced” methods to accurately evaluate the ESW of SPEs and SCEs. Some papers on Inorganic Solid Electrolytes (ISE) or liquid electrolytes are also addressed in this part when an interesting approach to the ESW evaluation is conducted and that can be adapted for use in systems employing SPEs and SCEs.
Even if it is barely addressed, some authors try to improve the evaluation of the electrochemical stability of the electrolytes. Two main characteristics arise regarding a suitable evaluation of the ESW: the cell construction and the optimization of the experimental setup generally based on the techniques previously described (LSV and CV).
The architecture and the configuration of the electrochemical cell for ESW determination is indeed an important parameter. Although standard cells consist of three-electrodes (working electrode, counter electrode and reference electrode), most of the measurements, for convenience, (specifically for battery applications) are made in a two-electrode cell configuration (with one electrode playing the role of both reference and counter electrode). Most of the studies use metallic lithium as both reference and counter electrode assuming that the SPE or SCE materials are stable vs. lithium or indium-lithium alloys and stainless steel as a working electrode. Sometimes these experimental considerations vary (different electrode materials) and strongly impact the value of the ESW. These considerations will be discussed in detail in the next section.
3.1. Cell Configuration
As was mentioned previously, the standard cell configuration for the evaluation of the ESW is a Lithium/electrolyte/stainless steel (SS) cell (
Figure 3). Sometimes, variations, such as the use of other inert planar electrodes, can be observed. For instance, Piana et al.
[62] used Cu metal foil and carbon black coated Al foil as working electrodes for cathodic and anodic scans, respectively, with Li metal as both the counter and the reference electrodes. However, some authors who have focused on characterizing suitable ESW evaluation insist on the importance of the cell configuration and especially on the suppression of the inert planar electrode. Indeed, their geometric surface area and chemical composition is negligible compared to that of the composite electrodes that are used in practical battery devices and, thus, disregard the real electrochemical environment of the evaluated electrolytes. This often results to an overestimation and inappropriate ESW values.
In their work (1999) Xu et al.
[121] recommended that to achieve a real electrochemical stability window (for capacitor and battery applications), the electrode material used for the ESW evaluation should simulate the electrodes used in a real system. The electrolyte stability data generated by conventional approaches (cf.
Section 1) could be inaccurate when applied to electrolytes in real devices. Although this work dealt with liquid electrolytes, it is reasonable to assume that these ideas could be extended to solid electrolytes (i.e., with SPEs and SCEs). Some experimental details of their work are presented here. First, the electrolytes, electrode materials, and subsequent measurements were all handled under vacuum in an argon-filled glove box where both H
2O and O
2 content was below 5 ppm. Solvents of electrolytes were redistilled and well dried until the moisture content decreased below 100 ppm (around 50 ppm as determined according to the Karl Fischer titration method).
Linear or cyclic voltammetry were used for measuring the current-potential (i-V) polarization curves of the tested electrolytes. The three-electrode cell configuration was employed with either Pt wire, glassy carbon (GC), activated carbon (AC) film or lithium battery cathode composite LixMn2O4 as working electrodes, and with Li+/Li as the reference electrode. Lithium foil was used as the counter electrode for lithium battery tests. A scan rate of 5.0 mV/s was typically used, except for cathode composite where the measurement was performed at 0.1 mV/s, owing to the slow lithium-ion diffusion process inside the spinel materials.
First, they applied the conventional approach (i.e., linear or cyclic voltammetry on nonporous electrode is used for the determination of the ESW). As they explained, the limiting redox potentials are ascribed where the de-composition current achieves a predefined level. However, the choice of these cutoff criteria is not supported by any theoretical considerations and is therefore arbitrary. Depending on the authors, cutoff currents of 10 µA/cm2, 50 µA/cm2, 0.5 or 1 mA/cm2 (most popular) can be chosen. At the end, the difference in stability data imposed by these arbitrary cutoff criteria would be too important to ignore and could lead to conflicting conclusions on the electrolyte stability evaluation.
Moreover, authors claimed that the difference between cutoff currents is not the only reason for the inaccuracy of conventional ESW evaluation methods. To satisfy this conventional method, they identified two prerequisites. First of all, the capacitance of the working electrode must be negligible. Then, the faradic component of the current (If) must only be dictated by a decomposition mechanism. These settled conditions imply that no other faradic process than electrolyte decomposition takes place and can be sharply approached when nonporous electrodes such as GC or Pt are used in a potentiodynamic experiment due to their stability. This allows the absence of other faradaic processes on the working electrode, except for impurities which can be easily lowered beneath 0.1 mA/cm2 by rigorous drying of the electrolyte solvent and high purification of the electrolyte solution. However, these ideal conditions are no longer valid for battery applications, where electrode materials are not GC or Pt but rather various composite materials with high capacity which results in additional faradic processes. Furthermore, as active material are often moderate electronic conductors, conductive additives (usually carbon black) are added to the electrode formulation. Thus, the non-faradic part of the current density is no longer negligible as in the ideal case mentioned above. Also, the electrolyte decomposition is affected by the surface of the composite electrode. In conclusion, a narrower ESW should result for real systems and results obtained by conventional analyses with GC or Pt working electrodes cannot describe the anodic stability of electrolytes in real battery devices. Thus, according to the authors, using the same electrode material as that employed with the electrolyte in the real battery cell is the only way to obtain a reliable evaluation of the electrochemical stability window.
About their experimental results, they concluded on the inaccuracy of the stability window measurement when employing nonporous electrodes on the battery electrolyte. They determined that using the electrode material approaching the surface state and surface area of the electrodes used in real devices was the best solution. Authors claim that this solution should be universal, instead of being considered confined to the determination of anodic stability of battery electrolytes only. They also concluded about the fact that few ESW values had been obtained in this way at the time of their article. “That time” was in 1999 and, considering that the first part of this review was only based on recent papers (2019–2020), it seems that twenty years did not make a significant difference.
Other works insist on the use of working electrodes that are as close to the real battery electrodes as possible for ESW investigations. Kasnatscheew et al. performed a study
[122] in which they argue that even though a classical measurement (LSV) with an inert working electrode (Pt, glassy carbon) is well adapted for qualitative comparison, it becomes inappropriate in the prediction of a precise electrochemical stability value. This is mainly attributed to the electrode surface area, which is different from that of high surface composite electrodes used in practical devices. This affects the current density and, consequently, the overpotential. Additionally, the electrode composition and surface area are presumed to have an impact on the catalytic activity. To overcome these drawbacks, ESW measurements with LMO electrodes are proposed. This condition reasonably reflects a more realistic battery application. However, concerns and issues linked to this method are still pointed out by authors. There are doubts concerning the validity of this method, as it is not clear whether the obtained stability data can actually be transferred to a real LIB device. First, the chemical composition of the different active materials used in LIBs can, for instance, affect the stability limit via catalytic effects. Then, the determination of the ESW is based on a potentiodynamic principle, while LIB cycling is based on a constant current principle. These differences could have an influence on the determination of the electrolyte stability limit.
3.2. Other Methods: Improved Setups for the ESW Evaluation
In addition to the cell construction and the importance of using composite working electrodes, some authors focus on improved ways to use LSV or CV by adjusting parameters (sometimes articles still mention the use of inert working electrodes but with improved LSV or CV set up).
Hallinan et al.
[123] proposed an electrochemical approach based on a series of adjusted LSV measurements from different, large over potentials to open circuit voltage, which the authors name “variable reverse linear sweep voltammetry” for evaluating the ESW of solid polymer electrolytes. By applying relaxation times to the cell between each polarization, the first data points of each voltammogram are not limited by mass transfer. This allows the current vs overpotential data to be analyzed by a kinetic model such as the Butler–Volmer one.
Electrochemical measurements were conducted on PEO and SEO (Polystyrene–b–poly(ethyleneoxide)) containing LiTFSI salt in cells with different working electrode materials in both two and three electrode cell configurations at a scanning rate of 15 mV/s. The electrode materials are either used as current collectors in lithium-ion batteries or are intended as inert electrodes for examining oxidative degradation of the polymer electrolytes. An electrochemical impedance spectroscopy (EIS) measurement was performed for every sample before and after every set of electrochemical tests to ensure that the sample was still exhibiting behavior in the same way as a pristine cell. Among the results obtained by these upgraded LSV measurements, the authors found an electrochemical stability for the SEO electrolyte of around 5 V at 40 °C. Based on their experiments on Cu/SEO/Li cells, they concluded that over-discharge should be avoided to prevent Cu corrosion (as with liquid electrolytes for battery applications). This work demonstrated a different way of using LSV methods to improve precision in the determination of the ESW.
Kasnatscheew et al.
[122] assessed the validity of the potentiodynamic based ESW method by comparing the data with that obtained by galvanostatic method on commercial positive electrodes. They demonstrated the good agreement of the two methods on the determination of the oxidation stabilities of electrolytes. Additionally, they were able to quantify the parasitic reactions by comparing the specific capacity losses obtained in half of the cells during cycling experiments.
Even though this work is devoted to liquid electrolytes, it highlights the use of composite battery electrodes to obtain a more realistic evaluation of the ESW of the electrolyte, and these observations can be transferred to SPEs and SCEs.
A similarly comparative study based on potentiodynamic and galvanostatic results for several active materials was conducted by Homann et al.
[124]. In their work, they consider PEO-based SPE under battery cell operation to evidence cell failure and to clear up reported ambiguities regarding oxidation stability. Then, they conducted electrochemical stability evaluations on various cathode materials that are usually used in Lithium batteries (NMC, LMO, LNMO and LFP) via LSV (potentiodynamic) and galvanostatic measurements. The onset of oxidation can be detected by an exponential (e.g., Butler–Volmer) current increase and a potential plateau, respectively. LSV measurements were conducted with an applied scanning rate of 0.1 mV/s. For the galvanostatic approach, electrodes were charged with a specific current of 15 mA/g. Preliminary evaluations were conducted by LSV on a Pt inert working electrode and revealed an Eox around 4.9 V (the choice of the cut-off current was fixed arbitrary without further explanation). However, as mentioned earlier in this review and by the present authors, the validity of this result is questionable, as the surface area of the Pt foil is small compared to that of composite battery electrodes in practical cells. Then, they decided to implement a galvanostatic approach to improve the accuracy of their measurement. Thus, LSV on a conductive carbon electrode with higher surface area was applied and a lower onset oxidative potential of 4.6 V was obtained. Moreover, to be closer to real evaluating characterizations and battery conditions, galvanostatic experiments were performed on a conductive carbon electrode, NMC, LNMO, LMO and LFP working electrodes to confirm the LSV observations. Like the LSV experiment, the galvanostatic approach revealed that the onset of oxidation occurred at 4.6 V vs. Li
+/Li, as seen by the respective potential plateau. These results are shown in
Figure 7.
Figure 7. LSVs of PEO based SPE with a scan rate of 0.1 mV/s on (
a) Pt and (
b) conductive carbon working electrode resulting in an exponential increase in current density of 4.9 and 4.6 V vs. Li/Li
+, respectively. (
c) Determination of the onset of main oxidation of PEO based SPEs via overcharge of the working electrode with a specific current of 15 mA g
−1 using different positive electrodes. Extract with permission from ref.
[124]. Copyright 2020 Springer Nature.
In this study, the authors also concluded that the main source of the sudden battery failure was the Li/SPE interface and, particularly, Li dendrite formation and penetration through the SPE membrane rather than the SPE/NMC interface. Finally, they claim that “it is the cell set-up (PEO thickness, negative electrode), which is crucial for the voltage-noise associated failure, and counterintuitively not the high potential of the positive electrode.”
Another interesting work proposes an improved CV setup to analyze the ESW of a solid electrolyte
[125]. Dewald et al. applied what they call a “stepwise cyclic voltammetry” method to evaluate the practical oxidative stability of various inorganic solid electrolytes (SE) such as Li
10GeP
2S
12, Li
2S-P
2S
5 or Li
6PS
5Cl. For reasons already mentioned in the previous section about the necessity of avoiding the use of planar electrodes, the authors decided to replace the traditional working electrode by a SE-carbon black composite electrode with higher surface area in order to increase the interfacial contacts between the components and, thus, the sensitivity of the measurement. Additionally, as the electrolytes are expected to decompose in contact with Li metal, indium metal is used as a counter electrode to minimize the reaction current arising from the decomposition on the anode side and to ensure that no additional lithium source is present. Their results, presented in
Figure 8, clearly show the impact of the replacement of the inert working electrode by a composite electrode on the ESW determination.
Figure 8. Visual comparison of two types of electrode morphology in CV experiments with the thiophosphate solid electrolyte Li
10GeP
2S
12. Reprinted with permission from ref.
[125]. Copyright 2019 American Chemical Society.
By employing In/InLi as both the reference/counter electrode and planar stainless steel as the working electrode, only small currents are detected (black curve). Using a carbon−solid electrolyte composite electrode to the cell (orange curve) leads to higher currents. Consequently, important oxidative decomposition reactions are now visible, mainly due to the huge difference in surface area between a flat (steel) and 3D conductive electrode (composite carbon electrode). In order to determine more precisely the oxidation onset potential of the electrolyte they established a stepwise CV approach. Each CV was measured twice at a low scan rate (0.1 mV/s) followed by a stepwise increase of the potential range by 0.1 V up to 4.4 V vs. In/InLi (approx. 5 V vs. Li
+/Li). Authors once again pointed out the necessity of avoiding the use of classical working planar electrodes (SS, Pt) for a better evaluation of the ESW. Their stepwise cyclic voltammetry method could be transferred to the evaluation of SPEs and SCEs. Other works propose the use of solid electrolyte composite electrodes for better ESW evaluations with carbon
[126]. It has to be noted that the use of carbon for such measurement could be biased by the presence of residual water within the carbon. Careful drying of carbon is therefore required
[125]. Other authors will then prefer gold to carbon for instance
[127].
Lastly, Amanchukwu et al.
[128] synthesized a new class of fluorinated ether electrolytes that combine the oxidative stability of hydrofluoroethers (HFEs) with the ionic conductivity of ethers in a single compound. Among their main results, they showed that their fluorinated ether electrolytes can achieve an ionic conductivity of 2.7 × 10
−4 S/cm at 30 °C with a higher oxidative stability of up to 5.6 V compared to classical ether electrolytes. They used two methods to determine the oxidative stability of the electrolytes. In the first approach they used a classical LSV measurement in a SS/Li cell with an applied scanning rate of 0.1 mV/s from an open circuit to 6 V. They observed good electrochemical stabilities up to 5 V. They also used potentiostatic holds (
Figure 9), also called Potentiostatic Intermittent Titration Technique (PITT), to accurately probe the oxidative stability of the fluorinated ether electrolytes. They argue that long Potentiostatic hold experiments are less sensitive to the influence of impurities which could lead to an early increase of the current in LSV experiments not corresponding to the actual oxidation of the compound. Finally, their conclusions were made by using a stainless steel, aluminum, or Ni-rich NMC 811 electrodes with different electrolytes and by holding the potential for 3 h at increasingly higher potentials. The recorded current should decrease if no undesired Faradaic reactions occurs. Their best synthesized fluorinated ether electrolyte reached an oxidative stability of 5.6 V. Despite, the fact that this study was conducted on liquid electrolytes, the electrochemical “potentiostatic holds” method could serve as a solid base and could be extended to the ESW evaluation of Solid polymer electrolytes. For instance, Zhang et al.
[129] used this method of potentiostatic holds to determine the anodic stability of different polymer electrolytes.
Figure 9. Potentiostatic holds for 3 h at different voltages with stainless steel as the working electrode and 0.1 M LiFSA in tetraglyme and tetraglyme:TTE. Reprinted with permission from ref.
[128]. Copyright 2020 American Chemical Society.
Recently, Li et al.
[130] determined the absolute anodic stability threshold of polymer electrolytes via a capacity-based electrochemical method. The objective was to address the limitations imposed by traditional approaches such as LSV. The authors claim that “the inconsistency of LSV is intrinsic to the method”. The comparison of electrolytes with distinct conductivities is problematic due to the proportionality of the current density and the electrolyte conductivity. Mass transport limitations cause difficulty in the theoretical analysis and lead to greater error for any method based on current density. Their point is clearly presented in
Figure 10.
Figure 10. Linear sweep voltammogram of PEO:LiTFSI (O/Li = 10) from 3.5 to 5.5 V vs. Li
+/Li at room temperature. Examples of “stability thresholds” determined by (
a) extrapolating the voltammogram and (
b) using an arbitrary onset current density. (
c) Description of the systematic error in estimating the oxidative stability from current-based methods. Reprinted with permission from ref.
[130]. Copyright 2021 American Chemical Society.
The authors explain that close to the true stability potential of the electrolyte (Eonset), the Faradaic current (ionset) and the capacitive current (iC) may be too intricate to be distinguished. Thus, to clearly assign a faradic reaction, a significant increase of the observed current density is needed and the systematic error is the difference between ionset and iobs. Due to the slow kinetics and low diffusion processes in solid state systems, the i-E curve will be flattened, and a larger error is expected compared to liquid electrolytes. Also, the shape of the voltammogramm can influence the value of the systematic error which becomes unique to each case and quasi-impossible to correct between samples or by repeated measurements. Thus, a noticeable disparity is expected to be inherent to the LSV method for the ESW determination of polymer electrolytes.
Thus, in order to overcome these limitations, they developed an alternative method based on capacity measurements which they named the reversibility test. Briefly, a cyclic voltammetry is performed, and the charge capacities of the cathodic and anodic responses are compared. The ratio between the anodic and the cathodic capacity corresponds to the irreversibility of the process. The capacity ratio remains similar when the electrolyte is electrochemically stable but varies when the potentials exceed the stability threshold of the electrolyte. By this method, the anodic stabilities of poly(ethylene oxide) (PEO) and hydrogenated nitrile butadiene rubber (HNBR), both blended with lithium bis(trifluoromethanesulfonyl) imide (LiTFSI), were identified to be 3.6 and 3.7 V vs. Li+/Li, respectively.
Finally, other studies use computational methods (as DFT calculations) to determine the electrochemical stability of liquid and solid electrolytes
[131][132] or a complementary study involving both experimental and computational methods
[126][127][133][134][135]. In their work, Thompson et al.
[134] proposed a complementary computational/experimental work by using alternating current electrochemical impedance spectroscopy, direct current chronoamperometry, and optical absorption band gap measurements combined with first-principles calculations to characterize the electrochemical window of the Li
7La
3Zr
2O
12 (LLZO) solid electrolyte. These first-principles calculations were used to predict the density of states (DOS), band gap, and absolute positions of the band edges for LLZO. Authors employed three different levels of theory: (1) the semi-local generalized gradient approximation (DFT-GGA) of Perdew, Burke, and Ernzerhof (PBE); (2) the hybrid functional of Heyd, Scuseria, and Ernzerhof (HSE06); and (3) quasi-particle (QP) calculations based on many-body perturbation theory (G0W0 method). These methods were already used to predict the ESW of liquid electrolytes at electrode interfaces. This work is one of the good examples of the complementarity between experimental and computational methods that can improve the accuracy of the ESW evaluation of electrolytes including SPEs and SCEs.