Design rules applied to silver nanoparticles synthesis: A practical example of machine learning application.

Area of Science:

• Nanotechnology
• Materials Science
• Computational Chemistry

Background:

• Controlled synthesis of silver nanoparticles (AgNPs) is crucial for their function and safety.
• Numerous synthesis parameters influence AgNP properties, posing a challenge for predictable outcomes.

Purpose of the Study:

• To develop predictive models for AgNP properties based on synthesis parameters.
• To leverage machine learning for rational design and optimization of AgNP synthesis.

Main Methods:

• Manual extraction of synthesis data (parameters, properties, efficacy, toxicity) from scientific literature.
• Application of regression algorithms and cross-validation for model training and validation.
• Utilizing Shapley values to determine the impact of synthesis features on predictions.

Main Results:

• Developed predictive models for AgNP core size and antibacterial efficiency.
• Identified synthesis duration, scale, and capping agents as key predictors.
• Demonstrated the utility of machine learning in understanding structure-property relationships.

Conclusions:

• Machine learning can guide the rational design of AgNP synthesis processes.
• This approach supports the development of safe-by-design principles in nanotechnology.
• A valuable, curated dataset for computational nanotechnology research was created.