Search for: All records

Insights from data-driven surrogate models reveal that the Sherwood number characterizes mass transport conditions independent of the convection method, offering design guidelines for scaling up organic electrosynthesis.

 

Abstract Thin-film deposition from chemically reactive multi-component plasmas is complex, and the lack of electron collision cross-sections for even the most common metalorganic precursors and their fragments complicates their modeling based on fundamental plasma physics. This study focuses on understanding the plasma physics and chemistry in argon (Ar) plasmas containing lithium bis (trimethylsilyl) amide used to deposit Li x Si y thin films. These films are emerging as potential solid electrolytes for lithium-ion batteries, and the Li-to-Si ratio is a crucial parameter to enhance their ionic conductivity. We deposited Li x Si y films in an axial flow-through plasma reactor and studied the factors that determine the variation of the Li-to-Si ratio in films deposited at various points on a substrate spanning the entire reactor axis. While the Li-to-Si ratio is 1:2 in the precursor, the Li-to-Si ratio is as high as 3:1 in films deposited near the plasma entrance and decreases to 1:1 for films deposited downstream. Optical emission from the plasma is dominated by Li emission near the entrance, but Li emission disappears downstream, which we attribute to the complete consumption of the precursor. We hypothesized that the axially decreasing precursor concentration affects the electron energy distribution function in a way that causes different dissociation efficiencies for the production of Li and Si. We used Li line intensities to estimate the local precursor concentration and Ar line ratios to estimate the local reduced electric field to test this hypothesis. This analysis suggests that the mean electron energy increases along the reactor axis with decreasing precursor concentration. The decreasing Li-to-Si ratio with axially decreasing precursor concentration may be explained by Li release from the precursor having lower threshold energy than Si release. 

Autonomous chemical process development and optimization methods use algorithms to explore the operating parameter space based on feedback from experimentally determined exit stream compositions. Measuring the compositions of multicomponent streams is challenging, requiring multiple analytical techniques to differentiate between similar chemical components in the mixture and determine their concentration. Herein, we describe a universal analytical methodology based on multitarget regression machine learning (ML) models to rapidly determine chemical mixtures' compositions from Fourier transform infrared (FTIR) absorption spectra. Specifically, we used simulated FTIR spectra for up to 6 components in water and tested seven different ML algorithms to develop the methodology. All algorithms resulted in regression models with mean absolute errors (MAE) between 0–0.27 wt%. We validated the methodology with experimental data obtained on mixtures prepared using a network of programmable pumps in line with an FTIR transmission flow cell. ML models were trained using experimental data and evaluated for mixtures of up to 4-components with similar chemical structures, including alcohols ( i.e. , glycerol, isopropanol, and 1-butanol) and nitriles ( i.e. , acrylonitrile, adiponitrile, and propionitrile). Linear regression models predicted concentrations with coefficients of determination, R 2 , between 0.955 and 0.986, while artificial neural network models showed a slightly lower accuracy, with R 2 between 0.854 and 0.977. These R 2 correspond to MAEs of 0.28–0.52 wt% for mixtures with component concentrations between 4–10 wt%. Thus, we demonstrate that ML models can accurately determine the compositions of multicomponent mixtures of similar species, enhancing spectroscopic chemical quantification for use in autonomous, fast process development and optimization.