Pd-Modified Microneedle Array Sensor Integration with Deep Learning for Predicting Silica Aerogel Properties in Real...

Area of Science:

• Materials Science
• Chemical Engineering
• Artificial Intelligence

Background:

• Silica aerogel is a lightweight insulator with excellent properties, but its production is hindered by lengthy aging processes.
• Predicting material properties in real-time during synthesis can optimize manufacturing.
• Current AI approaches for material property prediction primarily use stoichiometry or structure data.

Purpose of the Study:

• To develop a system for real-time prediction of silica aerogel physical properties (pore diameter, pore volume, surface area).
• To utilize electrochemical impedance data, frequency, and time parameters obtained during processing.
• To enable process optimization and monitoring for silica aerogel production.

Main Methods:

• A 3x3 array Pd/Au sensor was used to collect real-time electrochemical impedance, frequency, and time data during aerogel synthesis.
• The sensor system demonstrated high sensitivity to pH variations during synthesis.
• A deep neural network algorithm processed the collected data for property prediction.

Main Results:

• The developed system accurately predicted silica aerogel properties in real-time.
• The system achieved a mean absolute percentage error of approximately 0.9% for property predictions.
• Optimal alignment was observed between the true and predicted values of silica aerogel properties.

Conclusions:

• Real-time prediction of silica aerogel properties using electrochemical impedance data is feasible.
• This AI-driven approach significantly enhances process optimization and monitoring in aerogel production.
• The method promises to improve the efficiency and effectiveness of silica aerogel manufacturing.