(210i) Computational Design of MOF-Based Electronic Noses for Disease Detection By Breath | AIChE

(210i) Computational Design of MOF-Based Electronic Noses for Disease Detection By Breath


Day, B. A. - Presenter, University of Pittsburgh
Wilmer, C. E., University of Pittsburgh
Although it well-known that our breath contains a multitude of biomarkers for various diseases, breath-based screenings and diagnostics are almost non-existent in a clinical setting. This is largely due to the fact the current state-of the-art gas sensing technologies, namely gas chromatography-mass spectrometry (GCMS), are too bulky and too expensive for screening a large number of samples. Furthermore, due to the sheer number of compounds in breath, simple low-cost sensors such as mass-based and chemiresistive devices struggle to differentiate between the compound(s) of interest and other unknown gases. Thus, conventional diagnostics such as blood tests are still preferred. However, researchers have envisioned using gas sensing arrays, often called electronic noses, to achieve a level of sensitivity and selectivity comparable to GCMS, while maintaining the cost, speed, and size advantages of simple sensors, ultimately creating a device which would make breath-based screening and diagnostics competitive with conventional techniques.

Herein, we discuss the use of metal-organic frameworks (MOFs) in electronic noses of surface acoustic wave (SAW) devices for breath-based disease detection. Owing to the number of MOFs (>130,000) one could choose from, as well as the number of gases found in our breath (700+), the task of building an electronic nose is as much of a big data problem as it is an engineering challenge. Thus, to properly address this big data aspect of the problem, our efforts to design an electronic nose have been computational in nature. Using Grand Canonical Monte Carlo (GCMC) simulations, we can generate massive libraries of adsorption data which can be used to match against real sensor outputs, enabling quantification of a gas mixture. However, as we desire to include more gases and detect them in smaller concentrations, the number of gas mixtures which would need to be simulated grows exponentially. Thus, in order to maintain a reasonable computational demand, we pivoted to a new model in which we use GCMC simulations to generate a library of adsorption coefficients which enable us to determine the adsorbed mass of trace gas species without running a distinct simulation for each new set of concentrations. In turn, we created a flexible algorithm for the quantification of trace gas species. Since most of the gases of interest in breath are present in trace quantities, the above method works very well for simultaneously quantifying numerous compounds in breath, and using that information for disease screening and diagnostics.