Prediction and Optimization of Ion Transport Characteristics in Nanoparticle-Based Electrolytes Using Convolutional Neural Networks

Abstract

<pre>We develop a convolutional neural network (CNN) model to predict the diffusivity of cations in nanoparticle-based electrolytes, and use it to identify the characteristics of morphologies which exhibit optimal transport properties. The ground truth data is obtained from kinetic Monte Carlo (kMC) simulations of cation transport parameterized using a multiscale modeling strategy. We implement deep learning approaches to quantitatively link the diffusivity of cations to the spatial arrangement of the nanoparticles. We then integrate the trained CNN model with a topology optimization algorithm for accelerated discovery of nanoparticle morphologies that exhibit optimal cation diffusivities at a specified nanoparticle loading, and we investigate the ability of the CNN model to quantitatively account for the influence of interparticle spatial correlations on cation diffusivity. Finally, using data-driven approaches, we explore how simple descriptors of nanoparticle morphology correlate with cation diffusivity, thus providing a physical rationale for the observed optimal microstructures. The results of this study highlight the capability of CNNs to serve as surrogate models for structure--property relationships in composites with monodisperse spherical particles, which can in turn be used with inverse methods to discover morphologies that produce optimal target properties.</pre>

Ion States Impact Charge Transport and Dielectric Constant for Poly(ethylene oxide)-Based Sulfonylimide Lithium Ionomers

“…Although such an ion conduction mechanism cannot be validated experimentally, MD simulations have predicted such a hopping mechanism for some ionomers with weak ionic interactions. 70,71 The proposed ion conducting mechanism ties in nicely with our experimental and DFT simulation work and suggests that delocalized ion chemistries could be the path forward to create soft single-ion conductors with superior ionic conductivity.…”

“…In addition to the interaction energy, the energy difference between the associated cluster and the dissociated cluster to form single Li + solvated by 4 DME (Li-DME) agrees with a recent MD simulation result, which suggests that Li + transport through clusters is more favorable than through the PEO matrix. 70 DFT computations suggest that Li + can be found in all the states due to the similar interaction energy for different ion states (Figure 4) and moderate dissociation energy between associated and dissociated clusters (Table 2). The interaction energy and dissociation energy without explicitly added DME are summarized in Figure S4 and Table S3, which highlights the solvation effect of DME in lowering the dissociation energy and the interaction energy for charged ion states (i.e., LiTLi + and (Li-DME) + ).…”

“…34,35,41 The smaller spacing indicates a lower average aggregation number (AN, the average number of Li + per cluster, see Table 1) which can facilitate Li + hopping along the aggregates as implied from a recent MD study for PEO-based ionomers. 70 Ion states also impact the dielectric and conductivity spectra from DRS, which further supports the ion hopping mechanism. Noticeably, MTLi52 shows significant relaxation strength which increases with temperature despite its most aggregated morphology (Figure 5).…”

“…The proposed ion conduction mechanism ties in nicely with the experimental results and is supported by a recent MD simulation. 70 The implication of this study is two-fold. First, it reveals a strong correlation between dielectric response and charge transport; both are impacted by the different ion states the conducting Li + can sample.…”

“…Our DFT results agree with an MD simulation for PEO-based sulfonylimide-lithium singleion conducting ionomers, where an easy reorganization of ion aggregates was reported. 70 The combination of experimental and simulation results implies that Li + can hop along the ion aggregates with small ion aggregation spacing and low cluster dissociation energy between the neutral and charged states. Such mechanism rationalizes the extremely high dielectric constant and low Haven ratio, which will be further discussed in this study.…”