Integrating Electrochemical Analysis and Machine Learning for Enhanced Discovery of Conductive MOFs: A Database-Driven Approach

Abstract

Metal-organic frameworks (MOFs) are highly versatile structures composed of metal ions linked by organic molecules. They offer a broad range of applications in electronics and light-based technologies due to their flexibility. However, the connection between metal ions and organic components often hinders efficient charge movement, which is essential for electrical conduction. To address this challenge, researchers have been innovatively designing MOFs to enhance their conductivity, which is typically low. This process is complex and time-consuming. To overcome these barriers, and to save both time and costs, we integrated machine learning (ML) with electrochemical testing. Our aim was to pinpoint ion and electron conductive MOFs from the Computation-Ready, Experimental Metal–Organic Framework (CoREMOF) Database, which contains over 14,000 MOFs. The ML approach predicted 84 intrinsically conductive MOFs, which were then subjected to rigorous experimental validation through synthesis and conductivity performance testing. Further experimental data confirmed that only two conductive MOFs, containing copper and pyrazole structures with carbonyl groups, exhibited both proton and electron conductivity. The differences between them lie solely in the preparation techniques and solvents used. Sample 1 was prepared using hydrothermal methods and alcohol as the solvent, and NH4@1 was prepared with an ammonium solution at room temperature. To understand deeper their intrinsic conductivity and semiconductor properties, we employed temperature-dependent conductivity and electrochemical measurements, including Cyclic Voltammetry (CV), Electrochemical Impedance Spectroscopy (EIS), Linear Sweep Voltammetry (I-V), and Mott-Schottky analysis. The EIS data showed a smaller semicircular arc for synthesized NH4@1, indicating its lower charge transfer resistance due to the presence of two different positive charge carriers on its surface. We investigated the effect of increasing positive charge carriers by raising the humidity up to 98% RH. EIS data showed that increasing positive charge has a significantly reduced charge transfer resistance of sample 1 compared to NH4@1. Because NH4@1 consists of ammonium and H3O+ simultaneously, causing saturation of positive charges and increasing charge transfer resistance. Mott-Schottky analysis further elucidated the electrical properties of compounds 1 and NH4@1, revealing their p-type behavior with negative slopes and flat band potentials of 0.61 V and 0.9 V, respectively. The CV analysis of pyrazole carboxylic acid in a 0.1 M KOH solution, using three reference electrodes, revealed reduction and oxidation peaks vs. Ag/AgCl. These peaks highlight the redox behavior of the linker, generating a stable carboxylate anion (-COO−) that facilitates charge hopping. This reversible redox activity is crucial for the charge transport mechanism within the MOFs. CV analyses confirmed the presence of different copper oxidation states in both compounds, with cathodic peaks indicating the reduction from Cu(II) to Cu(I), suggesting active charge transfer. Extensive electrochemical studies, encompassing both Mott-Schottky and EIS analyses, provided deeper insights into charge carriers and transfer resistance, areas not extensively explored in previous research on intrinsically conductive MOFs. Our results underscore the critical role of experimental validation of ML predictions in accelerating the discovery and characterization of conductive MOFs for potential applications in batteries.