Lithium metal batteries have drawn enormous attention over the last month. Lithium metal is the holy grail in batteries as it has the potential to bring a substantial boost to the energy density over current Li-ion batteries. One of the central problems in using metallic lithium involves suppressing the formation of filament-like structures, popularly called dendrites. Many material classes have been explored as electrolytes to address this: ceramics, polymers, and composites. Now, we find that there is a new material class worth exploring: liquid crystals.
In 2017, my group was pursuing several approaches to suppressing dendrites. Around this time, I helped Steve LeVine, who wrote an excellent story on progress made by Ionic Materials. These discussions led us to an interesting new class of materials: liquid crystals. The most widespread example of the use of liquid crystals is in cinema displays based on liquid crystalline displays (LCDs). Liquid crystals have properties in between a liquid and a solid; the typically large molecules that constitute liquid crystals are ordered much like a solid but also flow similar to a liquid.
The initial thesis was the following: liquid crystals could offer a new strategy to suppress dendrite growth through its orientational order. Liquid crystals are made of large molecules that have directionality, i.e. they prefer to orient along certain directions. A growing dendrite is now forced to re-orient the molecules leading to an energy penalty. This energetic driving force could provide a new way to suppress a growing dendrite.
Our first approach to testing this thesis was to incorporate the effect of liquid crystals into the kinetics and perform a linear stability analysis similar to what we had done earlier to analyze ceramic and polymer electrolytes. This turned out to be more difficult than expected as the modifications needed for liquid crystals were more complicated and made the problem intractable with that approach. Around the same time, we searched over a very large number of ceramic solid electrolytes using machine learning. The machine-learning-based study did not yield many candidates, highlighting the challenge of suppressing dendrites. This led us back to think deeper about trying to find a new method with which this problem could be analyzed.
At this time, Dr. Zijian Hong, a postdoc in my group, had developed a phase-field model to simulate lithium metal anodes with liquid electrolytes. This turned out to be the method we needed to simulate the challenging moving boundary problem. We could incorporate the liquid crystal by introducing a new director field capturing the orientation of the liquid crystal molecules and include the effects of anchoring between the liquid crystal and the metallic lithium surface. Using this method, we showed that a liquid crystalline electrolyte can suppress dendrite growth provided the anchoring energy is large enough. Interestingly, the anchoring energy is also of importance in liquid crystal displays and lithography.
Liquid crystals are a material class that already has large volume manufacturing. Li-ion batteries manufacturing inherited the magnetic cassette tapes equipment as the sales started to dwindle for magnetic tapes in favor of compact discs (CDs). Manufacturing electrolytes (separators) at a competitive price-point of well-below $10 per square metre remains a challenge. LCDs are losing out in favor of LEDs and this could be an opportunity for liquid crystal display manufacturing to be re-purposed for making electrolytes.
Excited by the result, in July 2019, we uploaded the first version of the paper to arXiv and submitted it to a journal. The paper went through desk rejections at a couple of journals. This forced us to reflect on our paper: it had a gaping hole. We had shown that suppressing dendrites was possible, but we did not go the next step of identifying how to achieve this in practice. We needed to identify what kind of liquid crystals could achieve this suppression and more importantly, whether liquid crystals could be integrated successfully in high-voltage batteries. Such a burden of proof is not necessary for conventional polymer or ceramic electrolytes, as many successful demonstrations exist. When proposing an entirely new class of electrolytes with limited precedent, we felt this was a necessary burden of proof for the work.
The anchoring of liquid crystalline molecules to metallic lithium surface is akin to understanding the surface adsorption energetics, a routine problem in surface electrocatalysis. With the help of another Ph.D. student, Dilip Krishnamurthy, an expert on these methods, Zeeshan was able to quantify the anchoring energy and found that the reactive nature of metallic lithium enables almost an order of magnitude greater anchoring energy than those found on silicon surfaces. This put to rest the anchoring energy question. Next arose the question of whether these could be integrated into high-voltage batteries. A necessary but insufficient criterion is to calculate the occupied molecular levels of the molecule. This again turned out to be a fairly routine problem in electrolyte design. With the help of Dr. Vikram Pande, then a Ph.D. student in my group, Zeeshan was able to calculate the oxidative stability of liquid crystal molecules and identify design rules for molecules that could be compatible with high voltage batteries.
Following this, we submitted the paper to the Proceedings of the National Academy of Sciences and after a couple of rounds of tough and challenging reviews, we were able to convince the reviewers of the concept and the idea. Interestingly, one reviewer praised our open-sourcing of the code to enable better replication and further use of the work. We recently wrote on the importance of open sourcing of phase-field codes in a viewpoint in ACS Energy Letters.
It has been a long journey from the innovation trigger moment from my conversations with Steve LeVine, to a well-executed and thorough study by Zeeshan, ably supported by Zijian Hong. The next step is to realize a practical material demonstration of this concept, similar to our work on polymer-ceramic composites. Finally, thanks to ARPA-E for funding the work and providing feedback through the IONICS program.