An inorganic salt and a liquid organic solvent: That has been the traditional formulation for a lithium battery electrolyte. In the salt-in-solvent mixtures, lithium ions solvate/ligate with organic solvent in part or completely to form solvated/aggregated/contact ion-pairs. Though being simple and effective in concept, this way of manufacturing battery electrolytes imposes serious challenges – including fire-safety, voltage limitations, and low thermomechanical stability. Next generation batteries would rely on non-flammable solid electrolytes, such as inorganic ceramics, or organic polymers, but these materials do not supply all the needed physical and chemical properties, including high mechanical and thermal stability, passive electrode-electrolyte contacts, high conductivity, and large voltage window.
Zdilla and Wunder’s lab at Temple University have pioneered soft-solid “co-crystalline” electrolytes made of inorganic salts like LiCl, LiPF6 and NaClO4, and organic solvents like, N,N-dimethylformamide (DMF), and Adiponitrile (Adpn), which form solid solvates. This engineering approach stems from two important concepts of chemistry: First, a weakly Lewis basic solvent with sufficient polarity to dissociate ions of inorganic salt, and second, separation of Li+ or Na+ cations into a solvation sphere, leaving counter anion completely dissociated from the ion-pair. Prakash and Venkatnathan from IISER Pune joined this collaboration to understand materials properties at atomic and molecular levels where complex interplay of interactions governs the thermal and chemical stability of these crystals. The Temple collaboration (Zdilla and Wunder) earlier synthesized and characterized initial candidates like, LiCl.DMF, and NaClO4(DMF)3 (ref. 1,2) cocrystalline structures, which were good Lithium (or Sodium) conductors (ionic conductivity ~ 10−4 S/cm). Unfortunately, the cocrystals were unstable to high-energy anodes, resulting in no practical application in a high-voltage battery. The simulations on these materials provided an understanding of their structural and dynamic properties and explanations to electrochemical shortcomings. The Temple-IISER collaboration worked hand in hand to explain the experimental observations from simulations and then to design new characterization strategies based on simulations.3,4 The collaboration also lead to discovery of an exotic new materials class, termed “solvate sponge crystals” that underwent pressure- or temperature-dependent stoichiometric conversion and melt- and press-castability.5,6 Though these physical properties, are desirable for device fabrication, the chemical stability again fell short for use in high-voltage batteries.
In 2018, Birane Fall synthesized two new cocrystals with Adpn solvent using NaClO4, and LiPF6 salts. The cocrystals were highly thermally stable and also demonstrated stability toward high-energy electrodes, enabling assembly of full cells for the first time. These cocrystals exhibit an unforeseen best-of-both-worlds combination of liquid and solid electrolytes properties. For example, (Adpn)3NaClO4 (ref. 3) and (Adpn)2LiPF6 (this work) have high melting temperature (> 150 0C), and good interfacial contact with electrodes, like liquid electrolytes. The cocrystals, as seen from X-Ray diffraction (Fig. 1a), exist in ordered phase (leading to high structural stability), with inherent nanofluid grain-boundaries that provide excellent intergrain and electrode-electrolyte contact, as seen from scanning electron microscopy (Fig. 1b) and modeled from MD simulation (Fig. 1c). The (Adpn)2LiPF6 cocrystals exhibit moderately high ionic conduction (~ 10-4 S/cm) for an organic solid electrolyte, high transference number for Li+ ions (> 0.5), and excellent cycling (Fig. 1d) with high coulombic efficiency (> 99 %), and capacity in a Li0|LiFePO4 cell. An LTO/(Adpn)2LiPF6/NMC622 cell was cycled at C-rates of C/20 to 1C with Coulombic efficiencies > 96%, with no dendritic failure after 100 cycles.
Some outstanding questions about the underlying molecular behavior of this material were of our interest: e.g., i) how is it that (Adpn)2LiPF6 melts at a temperature higher than the decomposition temperature of LiPF6 and the melting temperature of Adpn? ii) how are the grain-boundaries so conductive, even more than the crystals? And iii) what is the probable mechanism of ion-conduction? Classical molecular dynamics and DFT simulations provided some answers from routine analysis, i) the non-bonded forces stabilize a Li+ in a tetrahedral solvation shell of soft-base N atoms of Adpn. The formation of Li+ channels in Adpn matrix prohibit undesirable Li---F ion-pair interactions (the same interaction that is root-cause of degradation of pure LiPF6. ii) The classical MD simulations showed facile formation of a liquid-like surface (composed of Adpn) which forms intergrain regions and also surrounds the crystals, which we confirmed with the Raman spectroscopy.
iii) Despite a firm qualitative understanding of the system, the usual simulation approaches initially failed to provide any consistent mechanism of ion condition. Before discussing how we finally modeled it with our unconventional models, it is important to understand how difficult it is to model ion conduction mechanism in these complex, contiguous assemblies of cocrystals, in comparison to liquid (polymer or salt in solvent) and inorganic solid electrolytes. For polymer electrolytes, above Tg, the diffusion of the Li+ ions can be easily associated with the backbone dynamics of the polymer chain. In organic solvents also, the mechanism is a simple vehicular diffusion of solvated Li+ ions. In both cases, the fluidity of the matrices often results in linear diffusion, and long enough MD trajectories become useful for the calculation of the diffusion coefficient, D. In the cases where polymer electrolytes solidify, a high temperature MD simulation is often used to estimate ion mobility precisely. Inorganic solid electrolytes (e.g., LISICON, ceramics) usually possess a strongly bonded, thermally stable sublattice of the anionic component of the electrolyte, and a mobile sublattice of Li+ ions. The fixed anionic sublattice in LISICONs enables modeling of Li+ ion hopping. MD simulations can be performed at sufficiently high temperature without destroying the sublattice matrix leading to good jump statistics even at small timescales. In contrast to these two major families, the solvate co-crystal family of electrolytes does not possess a stable anionic sublattice and/or fluidity at high temperatures, which pose serious limitations on the implementation of simulation methods; our NVT simulations, even 0.5 μs long, did not show ionic jumps. However, we had an intuition that the ion conduction is more facile in the grain-boundaries owing to their liquid like nature and the disordered regions at the co-crystal surface, which leads to the formation of defects at the crystal surface that may perpetuate into the interior of the bulk cocrystal. Hence the net ion conduction can have at least these two contributions – at the grain-boundaries of the cocrystals and inside the cocrystal where vacancies form. Thus, we created a state of art model of (Adpn)2LiPF6, where two grains of the cocrystals are solvated in excess Adpn (more than 140000 atoms, Fig. 1e). Long equilibration of this model predicted two diffusive patterns of Li+ ions (Fig. 1f), highly diffusive in the boundary layers and intergrain regions, and subdiffusive in the crystalline region. This suggests that the Li+ ions in these cocrystals conduct in various combinations of two extremes – a) highly conductive at the grain-boundary regions, where the ions are nano-confined, and b) moderate conduction in the bulk, where ions can only jump with a high barrier from one occupied site to a vacant site. This explanation predicted an activation energy consistent with the slope of the ionic conductivity plots from impedance spectroscopy, and provided a molecular-level view of the liquid grain boundaries seen in the SEM images.
The complete understanding of mechanism will require study of more control models in future. However, it should be noted that the experimentally determined solubility of LiPF6 in Adpn (whether directly dissolved or by dissolution of pre-existing co-crystals) is only 0.04M. Therefore, diffusion in the grain boundaries must involve a fluid boundary layer on the co-crystals (as observed experimentally) and a stabilized nanoconfined LiPF6 supersaturated Adpn solution.
More details on this study can be found in our recent article "A soft co-crystalline solid electrolyte for lithium-ion batteries" published in Nature Materials journal (https://doi.org/10.1038/s41563-023-01508-1)
- Chinnam, P. R., Clymer, R. N., Jalil, A. A., Wunder, S. L. & Zdilla, M. J. Bulk-Phase Ion Conduction in Cocrystalline LiCl·N,N-Dimethylformamide: A New Paradigm for Solid Electrolytes Based upon the Pearson Hard–Soft Acid–Base Concept. Chem. Mater. 27, 5479–5482 (2015).
- Chinnam, P. R. et al. A Self-Binding, Melt-Castable, Crystalline Organic Electrolyte for Sodium Ion Conduction. Angew. Chemie Int. Ed. 55, 15254–15257 (2016).
- Prakash, P. et al. Unravelling the structural and dynamical complexity of the equilibrium liquid grain-binding layer in highly conductive organic crystalline electrolytes. J. Mater. Chem. A 6, 4394–4404 (2018).
- Fall, B. et al. Experimental and Theoretical Investigation of the Ion Conduction Mechanism of Tris(adiponitrile)perchloratosodium, a Self-Binding, Melt-Castable Crystalline Sodium Electrolyte. Chem. Mater. 31, 8850–8863 (2019).
- Prakash, P. et al. Solvate sponge crystals of (DMF)3NaClO4: reversible pressure/temperature controlled juicing in a melt/press-castable sodium-ion conductor. Chem. Sci. 12, 5574–5581 (2021).
- Prakash, P. et al. Mechanism of Ion Conduction and Dynamics in Tris( N , N -dimethylformamide) Perchloratosodium Solid Electrolytes. J. Phys. Chem. C 126, 4744–4750 (2022).