Statistical approaches to powder preparation for Si/C anodes in Li-ion Batteries

Summary:

INTRODUCTION
Ball milling is an effective and low-cost production technique to obtain composite materials. However, adjusting the various milling parameters is quite important to prevent contamination formation due to jar-balls reactions and agglomeration of particles. Design of experiment (DOE) technique is the best way to find the effect of inputs and optimize the parameters. In this study, the effect of ball milling parameters on powder properties was investigated statistically and regression models were created. Furthermore, the effect of powder properties on slurry viscosity, material loading, and porosity of the electrode were examined.

METHODOLOGY
4-factor 2-level full factorial experiment was designed (RPM, Time, Ball to powder ratio, medium to powder ratio) by Minitab software and milling was performed using 20:80 Si/Gr pristine powder. Powders were characterized by XRD, SEM and Particle Size Distribution analysis. Contamination level was determined by measuring weights of balls before and after milling. Slurries were prepared via selected three different Si/Gr powder, carbon black, and PVDF in NMP solution respectively with 60:30:10 as dry composition. Viscosity measurements were carried out using Physica MCR 501-Anton Paar device and PP25 measuring plate. The resulting ink was cast onto Cu-foil using Dr. Blade. Films were dried in vacuum dryer at 100 °C for 24 hours. Cell was assembled in Ar-filled Gloveboxand galvanostatic charge-discharge tests were performed at different current densities with PAT-CELL system.

EXPERIMENTAL RESULTS
1. Powder Preparation
Experiments performed to find main effects and interactions for crystallite size, span (D90-D50/D10) and contamination level. 20:80 Silicon/Graphite powders (5gr) were used as pristine material and milling were carried out in WC jar with ø2mm WC balls in ethanol medium.It was found that RPM(A), Time(B), and interaction of RPM/Time (AB) are the most effective parameters with ball to powder ratio (D). According to regression models, contour plots were drawn for weighted crystallite size and contamination level. By using regression model, optimized small crystallite size powder (C3) were produced without contamination (