Metal-Organic Framework-Based Chemiresistive Array Paired with Machine Learning Algorithms for the Detection and...

Area of Science:

• Materials Science
• Chemical Sensing
• Nanotechnology

Background:

• Development of sensitive and selective gas sensors is crucial for environmental and safety monitoring.
• Existing sensors often lack the ability to differentiate multiple toxic gases effectively.
• Metal-organic frameworks (MOFs) offer tunable properties for gas sensing applications.

Purpose of the Study:

• To develop a chemiresistive sensor array using hexahydroxytriphenylene-based MOFs for detecting and differentiating toxic gases.
• To investigate the role of metal identity and framework packing in MOF sensing performance.
• To apply machine learning for analyzing sensor array responses and predicting gas compositions.

Main Methods:

• Fabrication of a sensor array using three conductive MOFs: Ni3(HHTP)2, Cu3(HHTP)2, and Zn3(HHTP)2.
• Testing the sensor array's response to various gases (CO, NH3, SO2, H2S, NO) and binary mixtures at room temperature.
• Utilizing Principal Component Analysis (PCA) and Random Forest classification for data analysis and gas discrimination.
• Employing feature importance methods to quantify individual sensor contributions.
• Conducting spectroscopic investigations to understand structure-property relationships.

Main Results:

• The MOF sensor array successfully detected and differentiated toxic gases (CO, NH3, SO2, H2S, NO) at parts-per-million (ppm) levels.
• The sensor array could also discriminate binary mixtures of SO2 and H2S.
• Machine learning algorithms (PCA and Random Forest) confirmed the array's ability to identify specific gases and mixtures.
• Feature importance analysis revealed the distinct roles of individual MOFs in gas discrimination.
• Spectroscopic data provided insights into MOF-analyte interactions governing selectivity.

Conclusions:

• The developed MOF-based sensor array demonstrates high potential for selective toxic gas detection and differentiation.
• Variations in metal centers and framework structures within the MOF array are key to achieving selective sensing.
• Machine learning techniques are effective tools for interpreting complex sensor array data and enabling accurate gas identification.
• This work contributes to the advancement of low-power, sensitive gas sensing technologies for safety and environmental applications.