Continuous Flow Chemistry and Bayesian Optimization for Polymer-Functionalized Carbon Nanotube-Based Chemiresistive...

Area of Science:

• Materials Science
• Nanotechnology
• Chemical Sensing

Background:

• Chemiresistive sensors are crucial for detecting gases like methane (CH4).
• Humidity can significantly impair the performance and stability of carbon nanotube (CNT)-based sensors.
• Non-covalent functionalization offers a route to modify CNT properties without damaging their structure.

Purpose of the Study:

• To develop novel poly(ionic) polymer-wrapped single-walled carbon nanotube (SWCNT) dispersions for enhanced methane sensing.
• To improve the humidity tolerance of chemiresistive methane sensors.
• To explore the use of continuous flow chemistry, Bayesian optimization, and machine learning for optimizing polymer functionalization.

Main Methods:

• Non-covalent functionalization of SWCNTs with poly(4-vinylpyridine) (P4VP) modified with poly(ethylene glycol) (PEG) moieties (PyBrR1).
• Quaternization of P4VP using continuous flow chemistry and Bayesian optimization.
• Fabrication of chemiresistive sensors incorporating PyBrR1-SWCNT composites, a platinum catalyst, and a polyoxometalate (POM) redox mediator.

Main Results:

• Increased degree of quaternization in PyBrR1-SWCNT composites led to improved methane sensor response magnitude, with optimal performance at ~10% quaternization.
• Incorporation of PEG significantly enhanced sensor stability in high humidity conditions (57-90% relative humidity) compared to unfunctionalized P4VP.
• Devices fabricated with the optimized dispersions outperformed in-situ prepared sensors and demonstrated superior stability at elevated humidity levels.

Conclusions:

• Polymer-wrapped SWCNT dispersions, particularly those with PEG functionalization and optimized quaternization, offer a promising approach for robust methane sensing.
• The study highlights the effectiveness of continuous flow chemistry and Bayesian optimization in tailoring polymer-CNT composites for sensor applications.
• Further investigation into POM character and machine learning models can lead to even more advanced and humidity-resilient gas sensors.