Electrolyte Solutions to Battery Problems (Part 1)

Battery chemistry is a complex beast, an intricate combination of materials science, organic and inorganic chemistry, electrochemistry, and surface science. The puzzle is part of the fun, but there’s a lot to learn. Recently, electrolyte chemist Dr. Jennifer Allen gathered a selection of helpful tips, nuanced points, and hot takes based on misconceptions and oversights she’s learned from throughout her decade as a battery researcher. In this first of a two-part series, she discusses interfaces, theory, & mechanisms.

---

Surface species: SEI vs degradation products

Here’s two concepts often described in lithium-ion batteries:

(a) Degradation products. Electrolyte solution components undergo unwanted parasitic reactions at electrode surfaces, with gaseous, liquid, or solid reaction products. Other cell degradation mechanisms can also lead to solid reaction products that may deposit on electrode surfaces. Degradation products are, in concept, bad.

(b) SEIs. A solid electrolyte interphase is a film formed by reaction of electrolyte components at an electrode surface. By Peled’s definition, it has the properties of a solid electrolyte: ionically conductive yet electronically insulating. It turns a reactive electrode surface into a passivated one, protecting electrolyte from the electrode surface, whose potential will drive parasitic reactions and sap cell capacity. SEIs are, in concept, great.

There’s a difference between these two, but where solid degradation products are concerned, it becomes murky.

Schematic of the polyhetero microphase SEI on Li or carbon.

The initial SEI is formed of reaction products that permit battery operation, and the maturing SEI may continue incorporating reaction products. But if the SEI is poor, decomposition products will continue to accumulate, now driving up impedance and consuming cyclable lithium. In some cases, species may accumulate on the electrode surface and incorporate into the SEI that wouldn’t be considered ‘the SEI’. For instance, if Mn dissolves from an NMC cathode, crosses over to a graphite anode, and becomes deposited in the SEI, the Mn is generally not referred to as an SEI component, but rather an SEI contaminant.

The question of ‘an SEI component’ and ‘a degradation species on the electrode surface’ becomes pertinent when the existence or quality of an SEI is unclear, as with SEIs on high-voltage anodes like LTO. However, it’s not mutually exclusive to suggest that (1) if the electrode potential is within the electrolyte’s stability window, we may not expect a passivating SEI, and (2) there will still be some number of (electro)chemical degradation products that accumulate on the electrode surface.

Exactly what is and isn’t an SEI component in today’s research landscape can be harder to pin down than you might think – is it everything on the electrode surface, no matter how it got there, including through mechanical deposition? Is it only electrochemical reaction products? Only initially-formed compounds? Only sufficiently protective species? Whether we call something an SEI component or not, there’s a variety of species on electrode surfaces, deposited from different mechanisms, that can have harmful or beneficial effects. Just because a species is present on the electrode surface and technically incorporated into the SEI doesn’t mean it’s helpful or necessary.

CEI is a horrible, horrible term

Now that I’ve confused you about what an SEI is, let’s make things worse. The cathode SEI is often referred to as the CEI, but I think it’s worth digging into this term. There’s a difference between the cathode/electrolyte interface and a cathode electrolyte interphase. Cathode Electrolyte Interphase indicates an ionically conductive, electronically insulating solid electrolyte film formed between the cathode and electrolyte solution… right?

I started writing this believing the E in CEI refers to the solid electrolyte film: CEI is a sister term to SEI, surely implying cathode (solid) electrolyte interphase. But in the literature, CEI is sometimes written as “cathode-electrolyte interphase” or “cathode/electrolyte interphase”, where the E seems to refer to liquid electrolyte. Having polled friends and colleagues, I can confirm the confusion here is real. It doesn’t make sense that we use SEI to explicitly refer to a solid electrolyte implicitly on the anode, and CEI to implicitly refer to a solid electrolyte explicitly on the cathode.

You could argue the CEI isn’t a passivating solid electrolyte and doesn’t serve the same role as an SEI on, say, graphite. But a terrible SEI on graphite in a short-lived cell is still referred to as an SEI, and the term’s resemblance to ‘SEI’ seems to imply it’s the same thing but on the cathode. Interestingly, before ‘CEI’ was widely adopted, Edström et al in 2004 studied the “cathode–electrolyte interface”, concluded that the film on LMO wasn’t passivating, and suggested the term SPI: “[we] choose the term solid permeable interface (SPI) rather than SEI to better depict the character of the layer formed on the cathode”. SPI has had some uptake, but isn’t common. In 2006, Abe et al tested conjugated additives that oxidised to form a film on LCO and proposed ‘ECM’ for this case, writing “We have named this resulting novel-type thin surface film as electro-conducting membrane (ECM) since it is different from solid electrolyte interphase (SEI) by the point of its electro-conductivity”.

‘SEI’, a term coined in 1979 to describe the film on lithium or carbon, is now used almost half a century later for all kinds of materials at various potentials. Given the vast number of electrodes and electrolytes out there, it’s hard to say anything about the singular CEI and how it differs from the singular SEI. If we use ‘SEI’ today to mean ‘interphasial films on electrode surfaces’, and we accept these can be of variable quality, then we can apply ‘SEI’ to films on any electrode. In fact some researchers do use “anode SEI” and “cathode SEI”; others use “anode/electrolyte interphase” and “cathode/electrolyte interphase”. I like either pair as long as both are used for consistency, though it might be useful for the field to converge on standard terms with agreed definitions. For better or worse, I do use ‘CEI’ near-daily… but I still hate it.

The mechanism of an electrolyte additive

Given how complicated electrode interphases are, how do researchers even determine what electrolyte additives do? I sometimes see articles where authors try an electrolyte component that improves cell performance in some unknown way, then study the electrodes with surface analysis methods. If the species on the electrode surface are different from the baseline, authors may conclude the new electrolyte component is acting by ‘forming a robust SEI’ at the anode or cathode. And often – there just isn’t sufficient evidence provided in the work to conclude that. Electrolyte components can improve cell performance in myriad ways and might, as a consequence, alter the nature of degradation products appearing on the electrode surfaces – but this is not necessarily the same thing as forming a robust SEI.

We could imagine, for example, a researcher is testing a new additive and isn’t sure exactly how it’s improving cell performance. If it’s acting as an HF scavenger, the side effects of lowered HF in the electrolyte may include: less HF is reduced to H₂ and LiF; fewer transition metals dissolve from the cathode and deposit on the anode; there is less acid-driven corrosion of the basic SEI. The experimentalist might see that the SEI composition is different and conclude the additive acts by forming a unique and beneficial SEI – rather than by preventing the destruction of the SEI and/or accumulation of degradation products in the SEI through an entirely different mechanism.

Given just how many mechanisms in batteries are entangled in nightmarish feedback loops, it’s helpful to keep the bar for fully trusting a mechanistic claim fairly high. For any proposed function of an electrolyte component, it’s important to consider whether the function has actually been proven by the methods used, and whether another function could explain the same results.

The flaw of HOMO-LUMO calculations, and why prediction is hard

Whether beneficially film-forming or parasitically capacity-consuming, reduction and oxidation reactions at electrode surfaces are key determinants of battery performance. Because of this, computational methods for screening and understanding mechanisms of electrolyte components commonly rely on HOMO (highest occupied molecular orbital) and LUMO (lowest unoccupied molecular orbital) energies. If electrolyte components are reduced or oxidised, the logic goes, then we can understand these reactions through the energies of valence orbitals where electrons are added or removed.

However, reduction and oxidation potentials are related to the Gibbs free energy difference between reactants and products, not the energy levels of individual molecules. (While I criticise HOMO-LUMO energies here, other computational methods may be more helpful.)

Schematic illustrating the electrolyte’s HOMO and LUMO energies and the electrochemical stability window.

Additionally, there are many different chemical interactions baked into observed redox potentials. For instance, it’s long been known that molecular stability in battery electrolytes depends on which electrode material is used. The onset potential for ethylene carbonate (EC) oxidation even decreases with Ni content of NMC, which may be caused by reactive oxygen release driving chemical oxidation, and metal–oxide covalency driving EC adsorptive dehydrogenation. Coordination between anions, Li⁺ ions, and solvents can also significantly change the favourability of anion or solvent redox reactions; furthermore, species that undergo hydrogen abstraction or defluorination reactions will have a different oxidation potential than the pristine molecules. While more sophisticated modelling techniques can capture some of these mechanistic nuances, the context specificity means that general prediction becomes far more difficult.

Oxidative stability of LiPF₆/EMC and LiPF₆/DMC toward LMO and glassy carbon.

For electrolyte additives, only thinking about orbital energies and single-step reactions means that multifunctional capabilities aren’t considered. For example, cleavage of HF-scavenging trimethylsilyl groups from phosphite additives enhances film formation on the cathode from the resulting species. Some additives may work in synergistic reactions, so that two additives together produce a uniquely beneficial SEI compared to either on its own.

Beyond neglecting multistep reactivity, relying on HOMO-LUMO energies assumes all additives are designed to be reduced or oxidised, but there’s a third class of additives that instead reacts chemically. These are primarily scavenging additives, which react with or sequester harmful degradation products. Examples include HF scavengers, PF₅ scavengers, oxygen scavengers, and transition metal chelation agents; also radical quenchers, which are both scavengers and flame retardants. For these additives, HOMO-LUMO energies may offer little to no predictive value. This is, I believe, a common oversight of predictive models in the electrolyte research field.

Reactions at electrode surfaces can be incredibly difficult to predict, especially as the degree of complexity (number of electrolyte components or number of reaction steps) increases. As a result, while computational methods can be valuable, sometimes the frustrating answer is that for now, we must still rely on trying many electrolyte candidates in many combinations (or the ‘buy it and try it’ method).

Electrode-electrolyte interactions are not stochastic

Setting aside the HOMO-LUMO debate, a simple element I think is sometimes overlooked is this: a species needs to be in contact with the electrode in order to react. There is evidence that the electrolyte components don’t make contact with the surface of the electrode in a totally random way. Thus, factors like preferential solvation and preferential adsorption are important to consider.

Preferential solvation can dictate electrolyte-electrolyte interactions because the molecules solvating the lithium are closest to the electrode surface during lithium intercalation. This is thought to play a role in the dominance of lithium ethylene dicarbonate (LEDC) in the graphite SEI when the electrolyte contains a mix of EC and linear carbonates: EC preferentially solvates lithium, so it’s more likely to be reduced to form LEDC. This effect may also be due to factors like preferential EC solvation at the interphase under electric field, and preferential Li⁺ desolvation from linear carbonates. High salt concentrations can further manipulate Li⁺ solvation, forcing anions in the solvation shell to permit otherwise unsuitable solvents like PC and acetonitrile.

Similarly, preferential adsorption can manipulate interfacial reactivity because the molecules that are more likely to adsorb on the electrode surface are more likely to react there, and even block other species from reacting. This is the proposed mechanism of some electrolyte additives and co-salts thought to form cathode SEI films, like TPFPB and LiDFOB. Anionic and electron-rich species such as borates, nitriles, or fluorinated species are of particular interest, owing to their disproportionate presence in the inner Helmholtz layer.

Schematic of a lithium-ion cell, showing a greater anion concentration at the cathode surface.

Preferential solvation and adsorption help explain why modifying concentrations of species in the bulk electrolyte may not lead to proportionally different cell performance. On the bright side, through the addition of components more likely to interact with electrode surfaces, these phenomena can be leveraged to incorporate otherwise unstable species into the electrolyte solution.

Manganese dissolution and how it occurs

One problematic interfacial degradation mechanism sometimes targeted through electrolyte design is transition metal dissolution. Transition metal oxide cathodes are known to react with the electrolyte solution, releasing metal ions. There’s a fair bit of interest in transition metal dissolution – particularly manganese dissolution – owing to the harmful SEI-disrupting reactions of transition metal ions that are well-documented for graphite anodes. The mechanism of manganese dissolution from cathode materials is often described in the literature as occurring via disproportionation. This is Hunter’s mechanism, first proposed to describe LiMn₂O₄: Mn³⁺ in the cathode disproportionates into soluble Mn²⁺, which dissolves in the electrolyte, and insoluble Mn⁴⁺, which remains in the lattice.

Hunter’s mechanism for Mn²⁺ dissolution from LiMn₂O₄.

There’s only one problem with this. When do we see the most Mn dissolution?

State-of-charge dependence of Ni and Mn dissolution from LNMO.

If we think about materials containing Mn³⁺ in the lithiated state, like LMO, we see that at top of charge, all Mn³⁺ is converted to Mn⁴⁺. And yet, top of charge is when the most transition metal dissolution occurs. The case is even more clear in materials like NMC, where all Mn should be tetravalent even when lithiated (we can make some arguments about Mn³⁺ impurities, but I’m not convinced all dissolution only occurs via impurities).

While Mn³⁺ disproportionation is a reasonable argument to explain why lithiated, discharged LMO undergoes Mn dissolution, it’s not a sufficient explanation to explain why delithiated, charged NMC undergoes Mn dissolution. The reduction and subsequent dissolution of transition metals at high voltages is more likely related to, and coupled with, electrolyte oxidation reactions.

Many of us have a habit of reaching for disproportionation when explaining Mn dissolution, even when we know transition metal dissolution is enhanced at higher states of charge. That there’s likely at least an additional mechanism at play is the sort of thing that may feel obvious in hindsight once pointed out, but it’s not immediately apparent.

Closing thoughts (for now)

If you’ve made it this far, congrats on wrapping your head around the nuances of electrolyte chemistry – the next round of paracetamol is on me. Come back next week for Part 2, where we’ll be covering cell design & experimental considerations!

---

🌞 Thanks for reading!

📧 For tips, feedback, or inquiries - reach out

📣 For newsletter sponsorships - click here

🌐 Follow us on Twitter, LinkedIn, and our website