A multimodal deep learning model, DeepNCI, is proposed for improving noncovalent interactions (NCIs) calculated via density functional theory (DFT). DeepNCI is composed of a three-dimensional convolutional neural network (3D CNN) for abstracting critical and comprehensive features from 3D electron density, and a neural network for modeling one-dimensional quantum chemical properties. By merging features from two networks, DeepNCI is able to reduce the root-mean-square error of DFT-calculated NCI from 1.19 kcal/mol to ∼0.2 kcal/mol for a NCI molecular database (>1000 molecules). The representativeness of the joint features can be visualized by t-distributed stochastic neighbor embedding (t-SNE), where they can distinguish categorized NCI systems quite well. Therefore, the fused model performs better than its component networks. In addition, the 3D CNN takes electron density as inputs that are in the same range, despite the size of molecular systems, so it can promote model applicability and transferability. To clarify the applicability of DeepNCI, an application domain (AD) has been defined with merged features using the K-nearest-neighbor method. The calculations for external test sets are shown that AD can properly monitor the reliability for a prediction. The model transferability is tested with a small database of homolysis bond dissociation energy including only dozens of samples. With NCI database pretrained parameters, the same or better performance than the reported results is achieved by transfer learning. This suggests that the DeepNCI model is transferable and it may transfer to other relative tasks, which possibly can resolve some small sampling problems. The source code of DeepNCI can be freely accessed at .

Theoretical predictions of macroscopic performance (power conversion efficiencies [PCEs]) and experimental analyses for microscopic material (conformation) have always urged for organic photovoltaics. A series of acceptors based on multi‐conformation bistricyclic aromatic enes core have been designed. The results suggested that A4‐2, A5‐2, and T4‐2 show the full folded conformation, fitting, and exhibiting advantageous properties of various parts for acceptors effectively, thus getting high VOC and JSC (kCS/kCR exceeds 1012) as well. Their PCEs of devices matching different donors were predicted through machine learning (ML). In traditional device structures and crude environments, a maximum PCE is about seven times higher than original. Herein, a comprehensive investigation, ranging for conformations → donor/acceptor interfaces → morphology → PCEs, is carried out by pure theoretical methods. Therefore, this quantitative micro‐analysis combined with the ML intelligent prediction leads to a new approach in the development of the next generation of nonfullerene acceptors.

No abstract