Direct Measure of Electrode Spatial Heterogeneity: Influence of Processing Conditions on Anode Architecture and Performance

Abstract

In this work, the spatial (in)homogeneity of aqueous processed silicon electrodes using standard poly(acrylic acid)-based binders and slurry preparation conditions is demonstrated. X-ray nanotomography shows segregation of materials into submicron-thick layers depending on the mixing method and starting binder molecular weights. Using a dispersant, or in situ production of dispersant from the cleavage of the binder into smaller molecular weight species, increases the resulting lateral homogeneity while drastically decreasing the vertical homogeneity as a result of sedimentation and separation due to gravitational forces. This data explains some of the variability in the literature with respect to silicon electrode performance and demonstrates two potential ways to improve slurry-based electrode fabrications.

Influence of the Polyacrylic Acid Binder Neutralization Degree on the Initial Electrochemical Behavior of a Silicon/Graphite Electrode

“…34 Hays et al also reported a much higher adhesion for PAAHbased Si/Gr electrodes compared to PAALi-based ones. 31 Contradictions between different studies may result from differences in the electrode manufacturing process (mixing and drying conditions), 42 the characteristics of the active material (particle size, silicon surface chemistry, and silicon/graphite ratio), 33,43 or the molar mass of the binder, 44 as well as the cycling conditions and, in particular, the choice of the electrolyte. This great diversity makes the rationalization difficult and highlights the need to understand and rationalize the underlying phenomena.…”

Role of Low Molecular Weight Polymers on the Dynamics of Silicon Anodes During Casting

“…From the zeta potential data, the dispersant PAA molecules interact with the silicon powder (although it is important to note there is not carbon black present in these measurements). Furthermore, it has been shown through X‐ray nanotomography that the addition of a low molecular weight dispersant increases the lateral homogeneity of the electrode, [4] which may be why the capacity of the Si+PAA with dispersant is better on the second cycle. However, since dispersants cause poor vertical homogeneity through the thickness of the electrode, [4] this could be one reason why the Si+PAA with dispersant electrodes perform worse in subsequent cycling than their counterparts without dispersant.…”

“…Most composite high‐silicon content anodes are comprised of silicon, conductive additive, and polymeric binder. The surface chemistry of silicon is highly dynamic, and the silicon particles tend to flocculate in a slurry, resulting in an inhomogeneous composite electrode architecture and the resulting losses of capacity and cycle life [4–7] . The challenges in making “good” silicon electrodes begins at the very initial stages of electrode fabrication where the handling of reagents affects electrode production.…”

“…The challenges in making “good” silicon electrodes begins at the very initial stages of electrode fabrication where the handling of reagents affects electrode production. For example, using a ball milling (BM) or planetary centrifugal mixing (PCM) method to prepare stock solutions of polyacrylic acid (PAA) binders cleaves a 450,000 g/mol PAA into 56 K and 53 K average molecular weight with linear and branched structures, respectively [4] . The formation of the smaller molecular weight polymers with different structures leads to a number of fundamental questions including: how does the distribution of polymer molecular weights effect electrode architectures?…”

Investigating Multiscale Spatial Distribution of Sulfur in a CNT Scaffold and Its Impact on Li–S Cell Performance

“…34−36 We have demonstrated in a few studies that the structural heterogeneity can be distilled by the K-means clustering analysis on the Raman mapping. 37,38 The basic idea of K-means clustering algorithm 39 is to partition the total number of Raman spectra (denoted as m) (x 1 , x 2 , x 3 , ..., x n ) within the Raman mapping into K sets (K ≤ n), S = [S 1 , S 2 , S 3 , ..., S k ], to minimize the within-cluster sum of squares, defined by the objective function, J as where c i is the mean of points (or centroid). In this case, it means the cluster spectrum in the K set S i .…”