ABSTRACT
Graph neural networks (GNNs) have shown remarkable performance in predicting the retention time (RT) for small molecules. However, the training data set for a particular target chromatographic system tends to exhibit scarcity, which poses a challenge because the experimental process for measuring RT is costly. To address this challenge, transfer learning has been used to leverage an abundant training data set from a related source task. In this study, we present an improved transfer learning method to better predict the RT of molecules for a target chromatographic system by learning from a small training data set with a pretrained GNN. We use a graph isomorphism network as the architecture of the GNN. The GNN is pretrained on the METLIN-SMRT data set and is then fine-tuned on the target training data set for a fixed number of training iterations using the limited-memory Broyden-Fletcher-Goldfarb-Shanno optimizer with a learning rate decay. We demonstrate that the proposed method achieves superior predictive performance on various chromatographic systems compared with that of the existing transfer learning methods, especially when only a small training data set is available for use. A potential avenue for future research is to leverage multiple small training data sets from different chromatographic systems to further enhance the generalization performance.
ABSTRACT
One of the most critical issues in preparing high-performance transparent supercapacitors (TSCs) is to overcome the trade-off between areal capacitance and optical transmittance as well as that between areal capacitance and rate capability. Herein, we introduce a TSC with high areal capacitance, fast rate capability, and good optical transparency by minimizing the charge transfer resistance between pseudocapacitive nanoparticles (NPs) using molecular linker- and conductive NP-mediated layer-by-layer (LbL) assembly. For this study, bulky ligand-stabilized manganese oxide (MnO) and indium tin oxide (ITO) NP multilayers are LbL-assembled through a ligand exchange reaction between native ligands and small multidentate linkers (tricarballylic acid). The introduced molecular linker substantially decreases the separation distance between neighboring NPs, thereby reducing the contact resistance of electrodes. Moreover, the periodic insertion of ITO NPs into the MnO NP-based electrodes can lower the charge transfer resistance without a meaningful loss of transmittance, which can significantly improve the areal capacitance. The areal capacitances of the ITO NP-free electrode and the ITO NP-incorporated electrode are 24.6 mF cm-2 (at 61.6% transmittance) and 40.5 mF cm-2 (at 60.8%), respectively, which outperforms state of the art TSCs. Furthermore, we demonstrate a flexible symmetric solid-state TSC that exhibits scalable areal capacitance and optical transmittance.
ABSTRACT
Electrical communication between an enzyme and an electrode is one of the most important factors in determining the performance of biofuel cells. Here, we introduce a glucose oxidase-coated metallic cotton fiber-based hybrid biofuel cell with efficient electrical communication between the anodic enzyme and the conductive support. Gold nanoparticles are layer-by-layer assembled with small organic linkers onto cotton fibers to form metallic cotton fibers with extremely high conductivity (>2.1×104 S cm-1), and are used as an enzyme-free cathode as well as a conductive support for the enzymatic anode. For preparation of the anode, the glucose oxidase is sequentially layer-by-layer-assembled with the same linkers onto the metallic cotton fibers. The resulting biofuel cells exhibit a remarkable power density of 3.7 mW cm-2, significantly outperforming conventional biofuel cells. Our strategy to promote charge transfer through electrodes can provide an important tool to improve the performance of biofuel cells.
