Cocaine by-product detection with metal oxide semiconductor sensor arrays

Area of Science:

• Materials Science
• Chemical Sensing
• Machine Learning Applications

Background:

• Detection dogs are commonly used to detect methyl benzoate, a cocaine byproduct.
• Development of electronic noses using metal oxide semiconductor (MOS) gas sensors is an area of interest for chemical detection.
• Modification of MOS sensors with zeolites can enhance their selectivity and sensitivity.

Purpose of the Study:

• To develop a gas sensor array capable of detecting methyl benzoate, a cocaine byproduct.
• To evaluate the effectiveness of machine learning algorithms for classifying various vapors using the sensor array.
• To assess the potential of zeolite-modified MOS sensors for rapid and inexpensive chemical detection.

Main Methods:

• Fabrication of n-type and p-type MOS gas sensors based on SnO2 and Cr2O3.
• Modification of sensors with zeolites (H-ZSM-5, Na-A, H-Y) to create a sensor array.
• Exposure of eleven sensors to various vapors, followed by data analysis and selection of four promising sensors.
• Assessment of the four-sensor array's discrimination capability against nine different vapors using machine learning (Weka software, polykernel function).

Main Results:

• The zeolite-modified sensor array demonstrated the ability to detect methyl benzoate.
• Machine learning models achieved 94.1% accuracy in classifying nine different vapors, including methyl benzoate, within five seconds of gas injection.
• The study identified four sensors with promising qualities for methyl benzoate detection.

Conclusions:

• Zeolite-modified metal oxide semiconductor sensors show significant potential for the rapid and inexpensive detection of methyl benzoate.
• Machine learning algorithms can effectively classify various vapors using data from the developed sensor array.
• Further research is needed to optimize the sensor performance for competitive detection capabilities compared to trained dogs.