Decomposition of LiPF 6 in High Energy Lithium-Ion Batteries Studied with Online Electrochemical Mass Spectrometry

,

Rechargeable Li-ion batteries are nowadays extensively used to power electronics and are entering the transportation sector by powering electric vehicles (EV).A wide range of both negative (e.g.graphite) and positive electrode materials (e.g.layered cobalt oxides, spinel-type manganese oxides, and olivine-type iron phosphates) have been thoroughly investigated and are now in widespread use in commercial batteries.The specific energy of Li-ion batteries is limited mainly by the positive electrode materials, having typical practical specific charges of ∼150 mAh/g and average operating potentials of ∼3.8 V vs. Li + /Li, which significantly inhibits the introduction of Li-ion batteries as power source in new applications.2][3][4] The origin of such a high specific charge is not yet completely understood, as the exact structure of the HE-NCM materials is highly dependent on the synthesis conditions and models coming from structural characterization are still under debate.Several reports have shown the presence of so-called Li 2 MnO 3 domains in the compound 5,6 whereas other groups demonstrated the monophasic character of their materials. 7However, during the first charge, a long potential plateau at ∼4.5 V vs. Li + /Li, not observed for conventional layered oxides, results from the delithiation process of the Li 2 MnO 3 domains accompanied by oxygen extraction.The extracted oxygen species are believed to be very reactive and partly evolve as O 2 gas, but can also further react with the electrolyte and the products present at the electrode/electrolyte interface. 8For example, Freunberger et al. proposed a mechanism for reactions between oxygen radicals and the carbonate solvents resulting in formation of lithium dicarbonate, formate, acetate, Li 2 CO 3 , CO 2 and H 2 O. 9 Several electrochemically initiated side reactions implicating the solvents (notably oxidation of carbonates HRCO 3 ) and the LiPF 6 salt have been proposed: [10][11][12][13][14][15][16][17][18][19] HRCO 3 → RO • + CO 2 + H + + e − [1]   LiPF 6 ↔ LiF + PF 5 [2]   PF 5 + ROH → POF 3 + HF + RF [3]   z E-mail: aurelie.gueguen@psi.chPOF 3 + ROH → POF 2 (OR) + HF [4]   POF 3 + RCO 3 R → POF 2 OR + CO 2 + RF [5]   For instance, when the cyclic ethylene carbonate (R = C 2 H 3 in Equation 1) is oxidized, 10 it may (via Equation 1) lead to the formation of RO .radicals, which subsequently may (via Equation 3) hydrolyze the salt and cause the formation of various organofluorine compounds, such as flouroethylene C 2 H 3 F. LiPF 6 is known to form an equilibrium (Equation 2) with LiF and PF 5 . 11,124][15][16][17][18][19] Thermally activated decomposition of carbonate and LiPF 6 based electrolytes has however been extensively studied (c.f.Nowak et al. 20 and citations therein).For instance, the presence of several phosphate species was evidenced by ex situ GC-MS and NMR when studying the stability and chemical decomposition of carbonate electrolytes at elevated temperatures (60-100 • C), which strongly accelerated such decomposition reactions. 16,21In situ spectroscopic evidence of electrochemically initiated electrolyte decomposition involving the LiPF 6 salt and high voltage cathodes is however more scarce.
In previous reports, we used online electrochemical mass spectrometry (OEMS) to follow the evolution of gases, such as O 2 and CO 2 , from the interface of the HE-NCM positive electrodes in Li-ion cells containing typical carbonate electrolytes. 22The results, combined with X-ray photoelectron spectroscopy (XPS), allowed us to propose a mechanism for reactions taking place in different potential windows of the two first charge and discharge cycles.Further improvements on the OEMS experimental setup allow us now to monitor the evolution of other gases, such as H 2 and POF 3 , originating from the decomposition of the carbonate electrolyte.The aim of the present work is not only to investigate the formation of O 2 and CO 2 , but also the H 2 and POF 3 gas evolution from HE-NCM during cycling to further disclose the underlying electrolyte decomposition processes.

Experimental
Electrode preparation.-Thepositive electrodes were prepared by coating thin glass fiber sheets (Whatman, GF/C) with a slurry of 93 wt% HE-NCM or stoichiometric NCM111 (LiNi 1/3 Co 1/3 Mn 1/3 O 2 ) (BASF SE), 3 wt% polyvinylidene fluoride (PVDF Kynar HSV 900, Arkema), 2.64 wt% Super C65 carbon (Imerys) and 1.36 wt% graphite SFG6 (Imerys), dispersed in N-methylpyrrolidone (NMP, Sigma-Aldrich).The porous substrate is required for the OEMS cell configuration. 10The "wet" thickness of the coating was ∼200 μm and the NMP was evaporated under vacuum at 80 • C for 8 hours.Circular electrodes were subsequently punched out (18 mm diameter) and dried overnight at 120 • C before being introduced into an Ar filled glove box.The average loading for the final electrodes was in the range 4.2 to 6.6 mg/cm 2 of active material.For comparison, a similar electrode coating (same thickness) was prepared on a Celgard 2400 (Celgard) monolayer polypropylene (PP) sheet with a loading of ∼12 mg/cm 2 .Self-standing LiFePO 4 counter electrodes were prepared by mixing 80 wt% of LiFePO 4 (BASF SE) and 10 wt% Super C65 carbon (Imerys) with 10 wt% polyetrafluoroethylene (PTFE, Sigma-Aldrich) binder in a solution of isopropanol and water (1:1) to form a viscous slurry.The slurry was sonicated and kept under mechanical stirring at 100 • C to evaporate the solvents and obtain a "dough-like" paste.The electrode sheets were obtained by working the paste with a spatula and mechanical rolling (thickness 200 μm), and then dried at room temperature.Electrodes were subsequently punched out (20 mm diameter) and dried (120 • C under dynamic vacuum) before introduction into the glove box.Delithiated LiFePO 4 electrodes were obtained by electrochemical delithiation of these LiFePO 4 electrodes vs. Li 4 Ti 5 O 12 .Carbon electrodes containing 79 wt% carbon Super C65 and 21 wt% PVDF were similarly prepared by coating a NMP-based slurry on the porous glass fiber or polypropylene substrates.
Electrochemical measurements.-Ahomemade cell was developed for OEMS experiments as described elsewhere. 23All cells were assembled in a glove box filled with argon.The counter electrode was metallic lithium or delithiated LiFePO 4 (see above).Commercial electrolytes composed of 1 M LiPF 6 or 1 M LiClO 4 in a 3:7 (w / w) mixture of ethylene carbonate (EC) and diethyl carbonate (DEC) were used (BASF SE).The amount of trace water in the electrolytes was determined using Karl-Fischer titration and did not exceed 17 ppm.Two types of 28 mm diameter separator were used: thin glass fiber (Sigma-Aldrich) and Celgard 2400 monolayer PP sheets.Before being introduced into the glove box, the PP and glass fiber separators were dried under vacuum at 80 and 120 • C for at least 12 h, respectively.The assembled cells were equilibrated at open circuit potential for 2.5 hours prior to electrochemical cycling.Galvanostatic cycling was carried out between 2.0 and 4.7 V vs. Li + /Li at C/15 rate during the first charge and at C/10 rate for the first discharge and following cycles using a computer-controlled battery cycling device (CCCC, Astrol electronic AG).The specific charges (not reported on the figures) were found similar to those reported previously (∼300 mAh/g and 250 mAh/g at the end of the first charge and discharge, respectively). 24The Super C65 carbon electrodes were subjected to cyclic voltammetry (CV) at a scan rate of 0.075 mV/s, starting from open circuit potential (OCP, ∼ 3.2 V) and cycling between 4.7 and 2.0 V vs. Li + /Li.All measurements were carried out at room temperature.

Online electrochemical mass spectrometry (OEMS).-The
OEMS setup was described elsewhere 23 and operates with a quadrupole mass spectrometer (QMS 200, Pfeiffer) for partial pressure measurements, a pressure transducer (PAA-33X, Keller Druck AG) for total cell pressure, temperature, and internal volume determination, stainless steel gas pipes and Swagelok fittings (3 mm compression tube fittings, Swagelok, OH, US) to connect the OEMS cell, a set of solenoid valves (2-way magnetic valve, Series 99, silver-plated nickel seal, Parker) and a scroll pump (nXDS15i, EDWARDS GmbH) for efficient flushing.The magnetic valves are electronically controlled with a Solid State Relay Module (NI 9485 measurement System, National Instruments) connected to a computer with a LabView Software (NI Labview 2013, National Instruments).For partial pressure and gas evolution rate analysis 0.7 mL of gas are extracted from the headspace (∼3.2 mL) of the cell and replaced by pure Ar (quality 5.0).Calibration gas bottles were utilized to relate the MS ion-current signals at 32 and 44 m/z to known concentrations of O 2 and CO 2 (1000 ppm of O 2 and 200 ppm CO 2 in Ar), before and after the measurement.The ion currents for fragments m/z = 2, 85, and 104 were recorded without calibration and converted into approximate gas evolution rates that allow direct semi-quantitative comparison between the different measurements.

Results and Discussion
Gas evolution during electrochemical cycling.-Figure 1 shows the OEMS data collected for HE-NCM electrodes in 3:7 EC:DEC during the two first galvanostatic cycles, comparing two different electrolyte salts, two different counter electrodes and two different separators.The gas compositions of the cells were continuously probed by analyzing the mass spectra (m/z = 0-110).The gases H 2 , O 2 , CO 2 , and POF 3 were the only identifiable species (appearing at m/z = 2, 32, 44, and 85, respectively) evolving in situ during cycling in our OEMS setup.Under all conditions, the overall evolution of CO 2 and O 2 is similar to previously reported results. 8Several distinct CO 2 formation contributions are apparent during the first charge.The first contribution (CO 2 (1) ) is anticipated to extend from ∼4.2 V until the end of charge, and is observed also during charging in later cycles.In the literature there exists a range of independent reports demonstrating that the corresponding CO 2 formation processes are mainly related to the oxidation of EC. 10,[25][26][27][28][29][30][31] There is a presumably superimposed second contribution (CO 2 (2) ) characterized by a CO 2 evolution maximum at ∼ 4.4 V.As this maximum is observed irrespective of electrolyte salt and counter electrode during the first charge, but not in later cycles and accompanied by the onset of O 2 evolution, it is obvious to relate it to the activation of the Li 2 MnO 3 domains in HE-NCM. 32,33Apart from O 2 , considerable amounts of reactive oxygen species are expected to evolve from the HE-NCM material during Li 2 MnO 3 domain activation, as suggested by the work of Castel et al. and Sathiya et al. 8,34,35 The reactive oxygen, or some oxygen-rich species formed thereof via follow-up reactions, probably decompose at higher potential ∼4.7 V, resulting in the rapid release of CO 2 (denoted CO 2 (3) ) and O 2 observed at the end of the first charge. 8,9The total amount of O 2 gas detected by OEMS varies between 27 and 89 μmol/g active material , depending on the cell composition.These quantities correspond to only a small fraction (<9%) of the theoretical O 2 content of the Li 2 MnO 3 domains in HE-NCM, which is ∼1 mmol/g active material .Thus the majority of oxygen extracted from the Li 2 MnO 3 domains would be expected to be involved in parasitic reactions.Most of the side-products formed are stable at high potential or further react to form products that do not decompose into O 2 gas.
Despite the similarities just described, several important differences depending on the choice of electrolyte salt, counter electrode and separator are apparent from Figure 1.There is an increase in both O 2 and CO 2 evolution when the LiClO 4 electrolyte salt (Figure 1b) or the delithiated LiFePO 4 counter electrode (Figure 1c) are employed.The reactive O-and C-O-rich species may form in larger quantities and/or be converted into non-reactive products either more slowly or to a smaller extent in absence of LiPF 6 and Li metal, respectively.Both LiPF 6 and Li metal appear to be involved in follow-up reactions with reactive oxygen species generated during Li 2 MnO 3 domain activation in such a way that CO 2 (2) is substantially enhanced in the absence of LiPF 6 or Li metal.The fact that CO 2 (1) is decreasing less or not at all from the first to the second cycle with the delithiated LiFePO 4 counter electrode or LiClO 4 , respectively, suggests that both electrolyte salt and counter electrode in addition play important roles in passivating the cell against CO 2 evolution and that LiPF 6 and Li metal are superior to LiClO 4 and delithiated LiFePO 4 in terms of suppressing gas evolution.In this context it should be mentioned that delithiated LiFePO 4 counter electrodes do not have a sufficiently low potential for reducing CO 2 .Therefore, an enhancement in CO 2 evolution is expected as this kind of counter electrode cannot act as a sink for CO 2 .Obviously, there is a significant interaction between the separator and the LiPF 6 salt as the amount of O 2 increased and CO 2 is reduced when the GF separator is replaced by the PP separator (compare Figures 1a and  1d).Further support of this conclusion is found while comparing the formation and consumption of gaseous species other than CO 2 and O 2 .H 2 and POF 3 are particularly important examples of such species.Since H 2 arises from the reduction of protic species ROH, such as H 2 O or alcohols, at potentials close to 2 V vs. Li/Li + , e.g. 30

ROH
it is an indirect marker for the formation of ROH species inside the battery.Similarly, POF 3 is an indicator of LiPF 6 decomposition, which is supposed to be mediated by ROH species (Equation 3) and/or other oxygen containing species such as Li 2 CO 3 and further, similar reactions. 13,36,37Fragments m/z = 85 and 104 evolve in a strongly correlated manner (Figure 1a) and are not observed with LiClO 4 electrolyte salt (Figure 1b) proving that the corresponding signals originate from POF 3 38 and thus that LiPF 6 decomposition indeed is enhanced by galvanostatic cycling of HE-NCM.Substantial quantities of POF 3 evolve during the first discharge regardless of whether Li metal (Figure 1a) or delithiated LiFePO 4 (Figure 1c) is used as counter electrode.Interestingly, POF 3 formation persists at a roughly constant rate throughout further galvanostatic cycling with delithiated LiFePO 4 , whereas it ceases almost completely after the first cycle in presence of a Li metal counter electrode.7][18] Although phosphate species are commonly observed, no evolution of fragments at m/z = 79 (PO 3 groups) and 83 (one of the fragments expected for POF 2 (OH)) were registered.Possibly, a different range of species are formed during room temperature electrochemical cycling compared to, e.g., high temperature storage of Li-ion battery cells.Alternatively, their volatility and/or concentration are insufficient for in situ detection with the employed OEMS setup.
In the following we will discuss in which ways the observations and differences regarding POF 3 evolution relate to ROH, Super C65, separator, cell potential and counter electrode and how we imagine LiPF 6 decomposition to trigger self-sustained decomposition cycles involving further highly reactive species such as HF. ), and chemical follow-up reactions involving primary decomposition products and/or other cell components (ROH (3) ). 13 ROH (1) is expected to be negligible since leak rates were found to be marginal with the employed type of OEMS cell and all electrolyte and cell components were thoroughly dried under vacuum at elevated temperatures for an extended time.Also, the deliberate addition of contaminants, such as water, is known to primarily lead to increased fractions of CO 2 and H 2 during the very early stages of cycling, which is not observed in any of the experiments.In contrast, the gasses in the experimental data in Figure 1 show a substantial potential dependence, suggesting that ROH formation relies mainly on electrochemical decomposition (ROH (2) ) or its chemical follow-up reactions (ROH (3) ).The correlation observed with LiClO 4 between H 2 evolution rate and cell potential (Figure 1b) supports the assumption that ROH (2) is formed at the HE-NCM electrode by electrolyte decomposition with potential dependent kinetics and subsequently diffuses to the negative electrode where it is reduced, e.g., according to Equation 6.A further comparison of the H 2 traces in Figure 1 provides interesting insights into the interdependence between ROH and LiPF 6 .No potential dependent H 2 evolution, as described above for the cell employing LiClO 4 , is observed in presence of LiPF 6 during the first charge and major parts of the first discharge.Obviously, most ROH (2) is scavenged by reactions, such as Equation 3, under these conditions and the formation of H 2 only occurs upon further reduction of some of the remaining ROH, HF or other residual decomposition electrolyte products at the very end of the first discharge when the positive electrode reaches potentials of ∼2 V.

The role of ROH in
While these results prove the importance of electrochemical ROH (2) generation for LiPF 6 decomposition, the question remains to which extent thermally activated chemical ROH (3) formation processes contribute to LiPF 6 instability and POF 3 formation. 13Diverse sources of ROH (3) formed in chemical reactions can be conceived and have been proposed in the literature. 10,39They range from decomposition reactions involving either auxiliary cell components (e.g.conductive carbon additives) or other secondary decomposition products (e.g.HF or OH − ) or combinations thereof (ROH (3a) ) to reactions associated with oxygen species formed during Li 2 MnO 3 domain activation (ROH (3b) ). 9,40The broad range of conceivable decomposition routes of the carbonate solvent implies that several factors may be influential.This led us to consider the roles of the conductive Super C65 carbon additive, the glass fiber separator, as well as the release of reactive oxygen species during Li 2 MnO 3 domain activation in more detail.

The role of carbon, separator, and Li 2 MnO 3 domain activation
in LiPF 6 decomposition.-Based on the BET specific surface areas of 62 and 10 m 2 /g for Super C65 and HE-NCM, respectively, the 2.64 wt% of Super C65 contained in our HE-NCM electrodes are expected to make up about 15% of the overall electrode surface area.The results in Figure 2a show that substantial POF 3 formation occurs with a HE-NCM free Super C65 electrode (solid line, filled squares), verifying that the presence of Super C65 itself contributes to LiPF 6 decomposition.From the concomitant CO 2 release we conclude that oxidation processes, such as a combination of Equation 1 and hydrolysis of the LiPF 6 salt via Equations 2-3 could be involved.
Comparison with a PP separator based cell (dashed line, open squares in Figure 2a) reveals a striking interconnection between the Super C65 dependent decomposition reactions and the separator, supporting the earlier results (compare Figures 1a and 1d) as well as previous reports about the influence of the separator on parasitic reactions inside batteries. 17,41Even though POF 3 and CO 2 also evolve with a PP separator, marked differences exist in the onset potentials and evolution rates compared to the glass fiber separator based cell.The effective contact area between separator and conductive carbon can be assumed to be substantially higher for the glass fiber compared to the PP separator.Consequently, the glass fiber separator not only leads to an overall ∼5-fold increase in evolution rate but more importantly also to substantial, additional POF 3 evolution features in the beginning of the first charge and during major parts of the first discharge.
Moreover, very high specific current peaks are observed with the glass fiber separator early on during the first charge and at the end of the first discharge.These are not observed with the PP separator and dwarf all other specific current features.In line with previous reports 17,41 we are convinced that impurities or reactive surface groups in the glass fiber account for these differences and that reactive electrolyte decomposition products, including ROH (3) , are involved in sustaining or promoting further electrolyte decomposition.For instance, SiO 2 is known to react with HF to form H 2 O and SiF 4 . 42OH forming reactions, such as the ones proposed earlier between carbonates and O 2 •− radicals, 9 suggest that oxygen removal from the HE-NCM active material during Li 2 MnO 3 domain activation might also play an important role in ROH (3) mediated LiPF 6 decomposition.In order to test this hypothesis, the gas evolution characteristics of a HE-NCM cell (Figure 1d) was compared with an NCM111 cell (Figure 2b).NCM111 does not contain any Li 2 MnO 3 domains and therefore lacks the 4.5 V plateau with its characteristic CO 2 (2,3) and O 2 evolution.However, the evolution of CO 2 (1) starts at the same potential (∼4.2 V) as with HE-NCM and the release of a small amount of O 2 is also observed at potentials around 4.55 V at which over-oxidation of the NCM111 material occurs, resulting in extraction of oxygen from the lattice and ensuing Mn dissolution. 43or HE-NCM, a small but non-negligible extent of POF 3 evolution is already observed at potentials between 4.3 and 4.5 V, but not with NCM111 suggesting that there is a contribution of Li 2 MnO 3 domain activation to LiPF 6 decomposition.Thus, contrary to what has been reported in the literature, 44 reactive oxygen species generated during Li 2 MnO 3 domain activation might indeed affect the cell performance as they are not only involved in the evolution of O 2 but also in parasitic reactions such as POF 3 formation.However, mere electrode potential appears to be more important in controlling POF 3 formation, because POF 3 is found to evolve much more strongly at the end of charge whenever the potential exceeds ∼4.5 V, regardless of electrode material and cycle number.
Since elevated potentials promote ROH (2) and ROH (3) formation, 13 the participation of ROH in these potential dependent LiPF 6 decomposition processes needs to be taken into account.In this respect it is important to realize that the contribution of ROH (3b) is likely of minor importance compared to ROH (2) and ROH (3a) regarding overall ROH formation inside the cell, because there is hardly any difference in H 2 evolution between the first and second cycle with LiClO 4 (Figure 1b).

The role of the counter electrode in LiPF 6 decomposition.-
Counter electrode dependent differences in the evolution of CO 2 and O 2 were already reviewed above.The discussion will now be extended by considerations about the observed similarities and differences in POF 3 and H 2 evolution (Figures 1a, 1c).In fact, the evolution of POF 3 is very similar with the two different counter electrodes during the first cycle.Only in the second cycle does the impact of the counter electrode become apparent.While POF 3 evolution is substantially decreased in the second compared to the first cycle with the Li metal counter electrode, it restarts very early during the second charge and persists at a more or less constant, elevated rate throughout the second cycle with the delithiated LiFePO 4 counter electrode.A drop in POF 3 evolution accompanied by substantial H 2 evolution is observed at the end of discharge regardless of the choice of counter electrode but only in presence of LiPF 6 .This supports the notion that LiPF 6 decomposition is intimately connected with elevated potentials and that H 2 is generated via reductive reactions involving hydrated LiPF 6 decomposition ROH intermediates. 13Interestingly, a similar extent of end-of-charge H 2 formation as with LiClO 4 / Li (Figure 1b) is observed with LiPF 6 / Li (Figure 1a) but not with LiPF 6 / delithiated LiFePO 4 (Figure 1c) during the second cycle.Obviously, further LiPF 6 decomposition ceases after the first cycle in presence of the Li metal counter electrode even though more ROH (2,3) is being generated.This suggests that ROH is necessary but not sufficient for LiPF 6 decomposition and that the Li electrode acts as a "sink" for ROH products that are required to sustain LiPF 6 decomposition.Summary.-Asschematically depicted in Figure 3, electrochemical decomposition reactions inside the cell -strongly influenced by auxiliary battery components, such as the separator and/or the conductive carbon additive -give rise to many species such as diverse radicals, H 2 , O 2 , CO 2 , and ROH ("Source").These can enter both chemical and electrochemical follow-up reactions with each other, the electrodes, the solvent, or LiPF 6 and its decomposition intermediates ("Reactant").Our in situ study allowed us to adapt the thermally activated electrolyte decomposition scheme from previous reports 14,42 by showing that similar reactions are also triggered by the potential of the positive electrode.Once the potential reaches the oxidation potential (∼4.2 V vs. Li + /Li) of the carbonate solvents, ROH species may arise that react with PF 5 to form POF 3 .A ubiquitous side-product of the latter reaction is HF, 36,45 which together with POF 3 cause the generation of more ROH 45,46 in an autocatalytic self-sustaining LiPF 6 decomposition cycle as outlined by Campion et al. 13 This autocatalytic decomposition cycle can be stopped by competing deposition / degradation processes that inactivate at least one of the involved species.We observed that a key player in controlling the nature and extent of these inactivation processes is the counter electrode ("Sink").These types of reactions are most relevant in full cell configurations (including graphite negative electrodes), where HF corrodes particularly the Mn rich cathodes 43 and where the accumulation of cathode/electrolyte decomposition products on the anode considerably increases cell impedance and reduces cycle life.

Conclusions
Further insights into the parasitic reactions involving LiPF 6 can be obtained from in situ OEMS measurements by monitoring POF 3 and H 2 evolution during the electrochemical cycling of HE-NCM based Li-ion batteries and analyzing its dependence on different system components and parameters: 1. Carbonate solvent oxidation (>4.2 V vs. Li + /Li) leads to the formation of reactive species, such as ROH, which in turn hydrolyze the LiPF 6 electrolyte salt to form POF 3 species.In addition, it can be concluded that the evolution of POF 3 with concomitant consumption of ROH is indirect proof for the formation of HF (e.g. according to Equation 3), which is one of the major causes for failure of Li-ion batteries during long-term cycling.HF contributes to the leaching of transition metal ions from positive electrode materials, whereby the metal ions and their complexes cross over to the negative electrode and ruin the operability of the SEI.Our results show that POF 3 can be used as a probe to investigate the HF formation in situ, which in turn may be the key to develop rational strategies for its mitigation.Resolving the issues of interfacial electrolyte decomposition on high energy and potential cathodes is of paramount importance for further development of Li-ion batteries.

Figure 2 .
Figure 2. (a) Plots of the CO 2 evolution rate and approximate evolution rate for POF 3 for the two first CV scans with an electrode composed of 79 wt% Super C65 and 21% PVDF cycled vs. Li with 1 M LiPF 6 in EC:DEC (3:7).Filled squares correspond to glass fiber separator based cell whereas open squares correspond to PP separator based cell.(b) CO 2 and O 2 evolution rate and approximate evolution rate for fragments with m/z = 85 for the two first galvanostatic cycles of a NCM111 electrode vs. Li with PP separator and 1 M LiPF 6 in EC:DEC (3:7).

Figure 3 .
Figure 3. Schematic loop, adapted from Campion et al., 13 including the different reactions involved in the POF 3 evolution detected during cycling of HE-NCM electrodes vs. Li metal in 1M LiPF 6 in EC:DEC (3:7) electrolyte.

2 .
Li 2 MnO 3 domain activation contributes considerably to POF 3 formation between 4.3 and 4.5 V vs. Li + /Li during the first charge.3. The contribution of Super C65 conductive carbon to electrolyte decomposition and POF 3 (and HF) formation is substantial.4. Impurities or surface functions of glass fiber cause dramatically enhanced and even additional POF 3 evolution.5. Electrochemically initiated LiPF 6 hydrolysis triggers an autocatalytic electrolyte decomposition cycle involving POF 3 as a reactive intermediate.6.The autocatalytic electrolyte decomposition cycle, as observed by the maintained POF 3 formation, is impeded by counter-or working-electrode potentials more negative than ∼2.5 V vs. Li + /Li, which leads to scavenging of reactive intermediates.