Elementary Decomposition Mechanisms of Lithium Hexafluorophosphate in Battery Electrolytes and Interphases

Electrolyte decomposition constitutes an outstanding challenge to long-life Li-ion batteries (LIBs) as well as emergent energy storage technologies, contributing to protection via solid electrolyte interphase (SEI) formation and irreversible capacity loss over a battery’s life. Major strides have been made to understand the breakdown of

common LIB solvents; however, salt decomposition mechanisms remain elusive.In this work, we use density functional theory to explain the decomposition of lithium hexafluorophosphate (LiPF 6 ) salt under SEI formation conditions.Our results suggest that LiPF 6 forms POF 3 primarily through rapid chemical reactions with Li 2 CO 3 , while hydrolysis should be kinetically limited at moderate temperatures.We further identify selectivity in the proposed autocatalysis of POF 3 , finding that POF 3 preferentially reacts with highly anionic oxygens.These results provide a means of interphase design in LIBs, indicating that LiPF 6 reactivity may be controlled by varying the abundance or distribution of inorganic carbonate species or by limiting the transport of PF 6 through the SEI.
Lithium-ion batteries (LIBs) have in recent years become a cornerstone energy storage technology, 1 powering personal electronics and a growing number of electric vehicles.To continue this trend of electrification in transportation and other sectors, LIBs with higher energy density [2][3][4][5] and longer cycle and calendar life 6 are needed, motivating research into novel battery materials.Battery electrolytes, which are typically the limiting factor in terms of LIB potential window and irreversible capacity loss, [7][8][9] are an especially attractive target for research and development to expand the utility of LIBs.
1][12][13][14] Carbonate/LiPF 6 electrolytes have many desirable properties, including weak ion association and high Li + conductivity, [15][16][17] but they are reactive at low potentials.When paired with graphite negative electrodes, carbonate/LiPF 6 electrolytes decompose to form a relatively stable passivation film known as the solid electrolyte interphase (SEI), [18][19][20][21][22][23] which prevents continual electrolyte degradation while allowing reversible charging and discharging.On the other hand, conventional electrolytes based on EC and LiPF 6 are essentially incompatible with high-energy density negative electrodes (e.g.Li metal, 24,25 Si 26,27 ) and form unstable SEIs, resulting in comparatively poor cycle and calendar life. 28,291][32] Such studies have sought to reveal the fundamental processes involved in the exemplar carbonate/LiPF 6 system and to identify opportunities for improvement through electrolyte engineering.An understanding of the decomposition of carbonate solvents, particularly EC, has been developed via a combination of experiment and theory.A wide range of decomposition products -including gases, 33,34 short-chain organic molecules, oligomers/polymers, and inorganic carbonates (e.g.Li 2 CO 3 ) and oxides (e.g.[43][44] In comparison, there are many open questions concerning the decomposition of LiPF 6 . It is widely accepted that LiPF 6 reacts to form LiF, which precipitates and contributes to the SEI. 30,31,45,46A range of other products, including POF 3 , 47 difluorophosphoric acid (PF 2 OOH), 48 and some organophosphorus compounds 49 have been identified by experimental spectroscopy.Moreover, LiPF 6 demonstrates thermal instability, 50,51 and it has long been suggested that an autocatalytic mechanism involving POF 3 (Equations 1-2) is responsible. 52wever, mechanistic explanations for LiPF 6 reactivity remain lacking.Most commonly, hydrolysis 7,45,46,51,53 In this work, we explore the decomposition mechanisms of LiPF 6 using DFT at a high level of theory (see Supporting Information for details).We find that water is not necessary to explain the formation of LiF or POF 3 , but rather that PF 5 can react rapidly with readily available Li 2 CO 3 during early SEI formation.This mechanism is entirely chemical in nature; it does not depend on electrochemical reduction or oxidation of LiPF 6 and can occur at any depth of the SEI as long as the transport of PF 6 -to inorganic carbonate domains is feasible.
Hence, the porosity, morphology, and transport properties of the SEI also become relevant factors.We then study POF 3 autocatalysis, using PF We begin by considering the formation of PF 5 , which is a key intermediate in essentially all LiPF 6 reaction routes considered in the literature and in this work.We find that the Even once PF 5 is formed, Figure 1 confirms that, at our chosen level of theory, the direct hydrolysis of PF 5 by H 2 O is unfavorable.Each of the three hydrolysis steps -the addition of , the elimination of HF to form PF 4 OH (M 1 −−→ M 2 ), and the elimination of another HF from PF 4 OH to form POF 3 (M 2 −−→ M 3 ) -is predicted to be endergonic.Further, the latter two steps both have energy barriers ∆G ‡ > 1.00 eV, agreeing with the experimental observation that hydrolysis is slow at room temperature.Significant thermal activation beyond temperatures reached in normal LIB cycling conditions would be required to enable LiPF 6 hydrolysis.
An alternate mechanism involves the reaction of PF 5 with Li 2 CO 3 (Figure 2).1][62] The reaction of PF 5 and Li 2 CO 3 is also entirely chemical in nature; none of the reactions in Figure 2 depend on electrochemical oxidation or reduction.As a result, the decomposition should not depend explicitly on applied potential, the proximity to the anode surface, or the availability of electrons.We therefore predict that the decomposition of PF 5 can occur anywhere in the SEI, so long as inorganic carbonates like Li 2 CO 3 are present.This being said, because Li 2 CO 3 is formed in the SEI as a result of electrochemical reduction of EC, 38,44 the overall rate of POF 3 formation via the reaction of PF 5 with Li 2 CO 3 will implicitly have a potential dependence.
While our focus in this work is on LiPF 6 decomposition during SEI formation, it is worth noting that Li 2 CO 3 is an impurity formed during the synthesis of common transition metal oxide positive electrodes. 59Accordingly, the mechanisms described in Figure 2 could occur at the positive electrode as well as at the negative electrode or the SEI.
Figure 2 indicates that POF 3 emerges rapidly by reaction with Li 2 CO 3 during SEI formation.This hints that the proposed autocatalytic mechanisms for POF 3 (re)formation (Equations 1-2), which rely on POF 3 and carbonate species, are chemically plausible.
To confirm the mechanism of POF 3 autocatalysis at elevated temperature, we first consider the formation of PF 2 O 2 R species (Figure 3).Specifically, we explore the formation of PF 2 OOH from H 2 CO 3 (Figure 3a) and LiHCO 3 (Figure 3b) and the formation of LiPF 2 O 2 by Li 2 CO 3 (Figure 3c).In addition to their relevance for POF 3 formation and LiPF 6 decomposition, PF 2 O 2 R species and in particular PF 2 OOH have been blamed as major contributors to the decomposition of SEI species and the loss of battery capacity.oxidation. 63If PF 2 OOH and related species could form at the negative electrode without high potentials, it could have significant implications for the stability of the SEI.Our mechanism confirms the previously reported autocatalytic formation of POF 3 .We find, in agreement with earlier experimental studies, 50,52 that this cycle requires significant thermal activation (T ∼ 150°C).This is primarily due to a sluggish intramolecular fluorine transfer and, specifically for the mechanism requiring PF 2 OOH as an intermediate, the high barrier for HF elimination to reform POF 3 .While we have found a mechanism for POF 3 autocatalysis that does not require any water, the significantly lower barrier for the pathway involving PF 2 OOH indicates that LiPF 6 thermal decomposition could be initiated and accelerated by LiPF 6 hydrolysis, 47 which is accessible at elevated temperature.
To conclude, LiPF 6 is an exceptional salt that is likely to play a major role in the LIB market for years to come.While some decomposition of LiPF 6 is desirable to form a functional SEI, continued breakdown can severely limit the life of LiBs.In this work, we identified a novel and facile elementary decomposition mechanism of LiPF

Figure 1 :
Figure 1: Hydrolysis of PF 5 to form POF 3 and 2 HF.This mechanism is overall thermodynamically unfavorable and involves two reactions with high barriers (∆G ‡ > 1.00 eV).

CO 3 .Figure 2 :
Figure 2: Energy diagrams for the formation of POF 3 from PF 5 and Li 2 CO 3 .a) LiPOF 4 forms via by simultaneous elimination of LiF and CO 2 from a PF 5 -Li 2 CO 3 adduct; LiPOF 4 can then eliminate LiF to form POF 3 .b) Alternate, less favorable mechanisms in which LiF is eliminated from the adduct without simultaneously eliminating CO 2 .

Figure 3 :
Figure 3: Reactions between POF 3 and simple inorganic carbonates (a) H 2 CO 3 , b) LiHCO 3 , and c) Li 2 CO 3 ) to form CO 2 and either PF 2 OOH or LiPF 2 O 2 .A trend between the partial charge of the reacting oxygen(s) and the reaction energies with POF 3 for each carbonate considered is shown in d).A linear fit, ∆G = 4.39q + 4.47, where q = the most negative oxygen partial charge, shows strong correlation (R 2 = 0.96) among the three carbonates.9

Figure 3a shows a
Figure3ashows a mechanism for a chemical reaction between H 2 CO 3 and POF 3 .The initial addition reaction between POF 3 and H 2 CO 3 (H 2 CO 3 +POF 3 −−→ M 14 ) is thermodynamically unfavorable (∆G = 1.62 eV).Subsequent reactions to form HF, CO 2 , and PF 2 OOH do not face significant barriers and should occur rapidly.The reaction between POF 3 and LiHCO 3 (Figure3b) follows a similar mechanism.The addition step (M 17 −−→ M 18 ) is also endergonic (∆G ‡ = 0.48 eV, ∆G = 0.52eV ), though we suggest that it could be accessed at moderate temperatures.Addition by LiHCO 3 is followed by the elimination of LiF (M 19 −−→ M 20 ), which is analogous to the elimination of HF in Figure 3a, (M 14 −−→ M 15 ).Following the complete removal of LiF, M 20 can undergo the same concerted proton transfer and CO 2 elimination shown in Figure 3a (M 15 −−→ M 16 ).In contrast, POF 3 adds easily to Li 2 CO 3 (Figure 3c, M 21 −−→ M 22 ), with ∆G ‡ = 0.15 eV and ∆G = −0.01eV.We explain the difference in the thermodynamics of the reactions between POF 3 and H 2 CO 3 , LiHCO 3 , and Li 2 CO 3 by considering atomic partial charges (Figure 3d).POF 3 is reactive towards the highly anionic oxygens in Li 2 CO 3 , but not towards the less charged oxygens in LiHCO 3 and H 2 CO 3 .A similar trend is found for the reaction between PF 5 and inorganic carbonates (see Supporting Information).The difference in behavior can also be rationalized in terms of acid-base chemistry.POF 3 and PF 5 (both Lewis acids) prefer to react with CO 3 2 -(a Lewis base) over HCO 3 -(depending on context, either a weak acid or a weak base) and H 2 CO 3 (an acid).Though PF 2 OOH formation via LiHCO 3 is possible, the difficulty of addition with protonated carbonates suggests that, barring electrochemical processes, LiPF 2 O 2 should be more abundant at the negative electrode than PF 2 OOH.Nonetheless, the prediction that PF 2 OOH and LiPF 2 O 2 can form at or near the SEI without the need for cross-talk from the positive electrode motivates further efforts to understand the interactions between these species and other SEI components.Mechanisms for the reformation of POF 3 , completing the autocatalytic cycle in Equation 2, are shown in Figure 4. Following a similar trend to that shown in Figure 3d, the attack of PF 5 by the acidic PF 2 OOH (Figure 4a, PF 2 OOH + PF 5 −−→ M 26 ) is thermodynamically

Figure 4 :
Figure 4: Possible routes for the reformation of POF 3 from PF 2 OOH (a) and LiPF 2 O 2 (b).Both mechanisms are kinetically limited due to an extremely unfavorable intramolecular fluorine transfer step (M 26 −−→ M 27 , M 30 −−→ M 31 ), which makes POF 3 autocatalysis unlikely at modest temperatures.Rate coefficients for the fluorine transfer step are provided in c) for the PF 2 OOH pathway and in d) for the LiPF 2 O 2 pathway.
on the basis of the observed evolution of CO 2 and POF 3 upon mixing of LiPF 6 and Li 2 CO 3 , but this route has largely been neglected in favor of hydrolytic mechanisms.Moreover, no elementary mechanism for the reaction between LiPF 6 -like species and Li 2 CO 3 has been reported.We find that PF 5 reacts vigorously with Li 2 63,64Jayawardana et al. have argued that PF 2 OOH should form at the positive electrode as a result of PF 6 - 6 using firstprinciples DFT simulations.Our results imply that under normal battery cycling conditions, the major decomposition mechanism of LiPF 6 does not depend on water or on electrochemical salt reduction.Rather, LiPF 6 forms the expected products LiF, POF 3 , LiPF 2 O 2 , and potentially PF 2 OOH via entirely chemical reactions with inorganic carbonates (especially Li 2 CO 3 ).PF 5 and POF 3 show a strong affinity to react with highly anionic oxygens and Lewis bases, suggesting that that efforts to control the reactivity of LiPF 6 should focus on limiting the exposure of PF5 to oxyanion and other basic species, including and especially inorganic carbonates like Li 2 CO 3 , in the SEI as well as on the surface of positive electrodes.This consideration may include morphological control, such as reducing porosity and/or abundance of inorganic species in the outer regions of the SEI.In the future, theoretical studies should be combined with experimental spectroscopy to validate the reported here.It should be possible to compare rate laws obtained by experiment (via e.g.time-resolved spectroscopy with varying amounts of inorganic carbonates and LiPF 6 ) and theory (via kinetic simulations, e.g.kinetic Monte Carlo).More challenging, but no less worthwhile, would be to confirm if the decomposition of LiPF 6 in a battery is primarily chemical or electrochemical in nature.This could be accomplished by tracking the rate of decomposition of LiPF 6 in the presence of inorganic carbonate species in a reductively stable solvent under varying applied potentials.While we have focused here primarily on LiPF 6 decomposition in EC-based electrolytes, we suspect that LiPF 6 could chemically react in a range of solvents via mechanisms similar to what we have described, provided that those solvents reduce and decompose to form oxyanions with highly charged reactive oxygens or sufficiently strong Lewis bases.The extent of LiPF6 decomposition will depend on the availability of these basic and oxyanion species.Additional investigations into solvent degradation and SEI formation in EC-free (and especially carbonate-free) electrolytes should be conducted to assess if the mechanism that we have described here is general or specific to carbonate-based solvents.Detailed study of the elementary reaction mechanisms between LiPF 6 decomposition products (especially PF 2 O 2 R species) and other SEI species (e.g.organic carbonates), as well as the formation mechanisms of organophosphorus compounds and phosphate polymers in the SEI, should also be conducted.