This MOF E-Nose, Connected to a Machine Learning Brain, Can Sniff Out Various VOC Mixtures

Offering a 96.5 percent accuracy in distinguishing different xylene mixtures, this compact sensor can replace a gas chromatograph.

Researchers from the Karlsruhe Institute of Technology (KIT) and the University of Pittsburgh have developed an "electronic nose" sensor capable of distinguishing between different mixtures of harmful volatile organic compounds (VOCs) — a step up from the usual sensors that simply detect all VOCs generically.

"Detection and recognition of volatile organic compounds (VOCs) are crucial in many applications," the researchers explain in the paper's abstract. "While pure VOCs can be detected by various sensors, the discrimination of VOCs in mixtures, especially of similar molecules, is hindered by cross-sensitivities. Isomer identification in mixtures is even harder."

The team's solution: An "electronic nose" made up of six gravimetric, quartz crystal microbalance (QCM)-based sensors arranged in an array. Each sensor is coated in metal-organic framework (MOF) films, which are tuned to particular isomers — enabling the sensor to separate mixtures with impressive accuracy.

The e-nose focuses on the distinguishing of xylene isomers, a substance which is harmful in large quantities. The traditional approach to finding o-xylene, m-xylene and p-xylene is to use a gas chromatograph — but the sensor the team has created is smaller, cheaper, and faster to operate.

Fed into a machine learning algorithm, the sensor proved capable of detecting pure isomers as low as one part per million; at an increased concentration of 100 parts per million, the sensor could identify 16 different ternary o–p–m-xylene mixtures with an impressive 96.5 percent accuracy.

"We foresee that extending the e-nose with further MOFs with distinguished affinities and pore sizes will expand the range of VOC mixtures which can be precisely detected," the team writes. "There, tailored side pockets in rigid MOFs, for instance, in MOFs like HKUST-1 and differently functionalized UiO-type-MOFs, will allow us to tune the shape and size exclusion."

The team's work has been published in the journal ACS Sensors, and is available under open-access terms for a limited time.

Gareth Halfacree
Freelance journalist, technical author, hacker, tinkerer, erstwhile sysadmin. For hire: freelance@halfacree.co.uk.
Latest articles
Sponsored articles
Related articles
Latest articles
Read more
Related articles