(292d) Computational Optimization of Metal-Organic Framework (MOF) Arrays for Chemical Sensing

Electronic noses are used in applications ranging from food quality assessment to environmental monitoring. However, current sensor technology is inadequate to properly address today’s challenges, such as low-concentration methane detection and disease detection via breath analysis. Metal-organic frameworks (MOFs), having high surface areas and tunable pore structures are ideal candidates as substrates for improved gas sensors. We use computational screening to predict the adsorption of gas mixtures of N₂, CH₄, and O₂ onto arrays of MOFs to identify the materials that maximize the signal-to-noise ratio for gas detection, obtaining sensor response signals from this adsorption data. Subsequently, an algorithm employs statistical methods to estimate the component concentrations in unknown gas mixtures probabilistically. Additionally, we quantify the information content of each possible MOF array, calculated via the Kullback-Liebler Divergence. Ultimately, we show how this computational approach is dramatically more efficient than choosing sensing materials for electronic noses by trial-and-error. For example, given 9 MOF materials to construct a 4-element array, rather than building and testing all 126 arrays, the best one can be computationally predicted in minutes. Such improvements will expedite the development of comprehensive electronic noses for hazardous materials sensing and disease detection.