SUMMARY:
UCLA researchers in the Department of Chemistry and Biochemistry have developed a machine-learning based algorithm that identifies reaction mechanisms from cyclic voltammetry data and offers accurate interpretations of complex electrochemical data.
BACKGROUND:
Electrochemistry is a readily accessible and valuable tool for researchers to study the redox behavior of discrete molecules and offer insights into their electrochemical systems, kinetics, and thermodynamics. Cyclic voltammetry (CV) is a relatively simple technique and is becoming commonplace in most chemical research laboratories. CV can provide a pictorial representation of the electrochemical mechanism a specific system or substrate undergoes. However, interpretation of the graphical response evolution can be difficult and nuanced, even for highly trained electrochemists. As a result, it is very susceptible to over-interpretation. There is a strong need for an automated, machine learning-based analysis of CV data as it would lead to more accurate analyses and increase the utility of advanced electrochemical characterization.
INNOVATION:
UCLA researchers, led by Professor Chong Liu, have developed a machine-learning based analysis of cyclic voltammetry graphs to determine the action mechanism of a chemical system. The inventors utilized neural networks, a subset of machine learning and deep learning algorithms, that have proven profoundly adept at extracting intrinsic features from high-dimensional data—both quantitatively and qualitatively. As a result, the algorithm can identify reaction mechanisms simply by evaluating the CV response of a particular compound. A wide range of simulation-derived training cases helped develop and verify the technology. Additionally, the algorithm showed high accuracy and correctly elucidated reaction mechanisms of experimentally collected data—with as little as four datasets per system. The presented machine-learning algorithm can serve as an efficient solution to quickly analyze electrochemical data—potentially aiding in electronics, battery, and pharmaceutical research. The algorithm may also increase the utility of advanced electrochemical characterizations and reduce the difficulty of its implementation.
POTENTIAL APPLICATIONS:
- Automated CV analysis
- Potentiostat software add-on
- Battery research
- Electroactive medicinal chemistry research
ADVANTAGES:
- Automated analysis
- Simple mechanistic determination
- Can evaluate experimental data with high accuracy
DEVELOPMENT-TO-DATE:
Successful demonstration of the invention completed. Algorithm has been demonstrated on simulated data and empirical data.
RELATED PAPERS:
Hoar, B. B. et al. Electrochemical Mechanistic Analysis from Cyclic Voltammograms Based on Deep Learning. ACS Measurement Science Au (2022) doi:10.1021/acsmeasuresciau.2c00045.