Automated Graph Neural Networks Accelerate the Screening of Optoelectronic Properties of Metal–Organic Frameworks

Abstract

The numerous organic and inorganic components of metal− organic framework (MOF) materials provide intriguing optoelectronic properties. Accurately predicting the electronic structural properties of MOFs has become the main focus. This work establishes two graph neural network models, crystal graph convolutional neural networks and a materials graph network, for predicting the band gaps of more than 10 000 MOF structures and promotes to improve the prediction accuracy through automatic hyperparameter tuning algorithms. Subsequently, for exploring machine learning-assisted screening of MOFs for the broader electronic properties, the screened copper-based MOFs are compared with lead-based MAPbI 3 solar cells with respect to the band gaps, densities of states, and charge density distributions, and the results have demonstrated that the overlap of the wave functions between the initial and final states of MOFs is weakened, which is conducive to the improvement of photoelectric performance. The chlorine doping strategy further enhances the advantage. The tuning of the machine learning model and hyperparameters and the doping strategy of halogen elements furnish empirical rules for the design of MOFs with excellent optoelectronic properties.

Revealing the Untapped Potential of Photocatalytic Overall Water Splitting in Metal Organic Frameworks

“…This model has consistently demonstrated reliable performance in predicting DFT-computed properties, surpassing other machinelearning models for MOFs. [19,22] As our objective needs to predict the E g of MOFs with Natoms >150, we initially assessed the transferability of CGCNN for predicting the DFT computed E g from the dataset with Natoms <150 to dataset with Natoms >150. Note that the QMOF database has computed the PBE computed E g for all 20375 MOFs, and the transferability can be validated with the PBE computed E g dataset.…”

Deep learning based on small sample dataset: prediction of dielectric properties of SrTiO 3 -type perovskite with doping modification

“…The network structure is shown in figure 2 [33], where not all upper and lower layer neurons are directly connected, but connected through 'convolution kernels' as intermediaries (partial connections). Owing to the characteristic of limiting the number of parameters and mining local structures, most research on CNN has focused on image recognition [34][35][36][37]. However, CNN's role in data regression problems is also extremely powerful.…”

Understanding ZIF particle chemical etching dynamics and morphology manipulation: in situ liquid phase electron microscopy and 3D electron tomography application

“…11,[20][21][22][23][24] In particular, there are limited methods to quantify MOF particles' etching speed, the product surface area, 3D volume and other structural properties, which are crucial to their applications, except for some limited computational model studies guided by machine learning. [25][26][27][28] This study aims to address this gap by revealing the etching mechanism of ZIF NPs using in situ liquid phase TEM in colloids. Firstly, different types of porous/hollow ZIF particles have been obtained by tuning the pH values of the etching solutions.…”

“…11,20–24 In particular, there are limited methods to quantify MOF particles’ etching speed, the product surface area, 3D volume and other structural properties, which are crucial to their applications, except for some limited computational model studies guided by machine learning. 25–28…”