Machine-Learning-Guided Morphology Engineering of Nanoscale Metal-Organic Frameworks

Chen, Peican; Tang, Zeyu; Zeng, Zhongming; Hu, Xuefu; Xiao, Liangping; Liu, Yi; Qian, Xiang‐Yang; Deng, Chunyu; Huang, Ruiyun; Zhang, Jingzheng; Bi, Yilong; Lin, Rongkun; Yang, Zhou; Liao, Hong‐Gang; Zhou, Da; Wang, Cheng; Lin, Wenbin

doi:10.1016/j.matt.2020.04.021

Cited by 50 publications

(43 citation statements)

References 53 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…At this point we highlight the emerging power of data science, machine learning and artificial intelligence computational tools and concepts, which may boost the field in the future. Based on current knowledge, machine learning is already applicable for the optimization of synthesis procedures [30] . Although so far it has not been implemented in MOF catalysis, this will be of high importance in the future for the realization of catalyst design strategies [31] …”

Section: Discussionmentioning

confidence: 99%

Exploitation of Intrinsic Confinement Effects of MOFs in Catalysis

2021

View full text Add to dashboard Cite

In catalysis research the design of bio‐inspired,'artificial enzymes‘ is a field of huge interest. These catalysts are distinguished by their high catalytic efficiency resulting from a close proximity of several active sites and secondary substrate‐catalyst interactions enabled by functional groups in the catalytic pocket. A class of materials which meets these requirements are metal‐organic frameworks (MOFs). Here, the pores confine a reaction environment where several functionalities can be incorporated and spatially positioned in a tunable fashion. Recently, a number of reports revealed the importance of such confinement effects for the control of catalytic activity and selectivity by exploiting the intrinsic properties like pore size and neighboring group effects and the alignment and distance of different active sites within one MOF pore. Thus, this concept aims to accentuate the potential of the exploitation of those effects in MOFs for the design of sophisticated catalysts.

show abstract

Section: Discussionmentioning

confidence: 99%

Exploitation of Intrinsic Confinement Effects of MOFs in Catalysis

2021

View full text Add to dashboard Cite

show abstract

“…As mentioned above, the using of ACSFs has an advantage in representing the local neighboring environment of an atom by using a fingerprint, which is composed of the output of several two-body and three-body functions that can be customized to detect specific structural features. Surprisingly, we found that its prediction error is closely sensitive to the number of atoms, Natom, and the square of the number of atoms, (Natom) 2 , with the Pearson correlation coefficients of 0.93 and 0.87, respectively ( Figure S9). In the process of message passing, every added atom will affect other atoms and increase prediction errors through many-body interactions, which should be considered in our future work.…”

Section: Factors That Influence the Performance Of Deep Learning Modelmentioning

confidence: 97%

“…Drug-like molecules 95 All the test molecules were re-optimized with B3LYP/6-31G(2df,p) 2 24/17 MAE, MAE MPCONF196 96 2 192/11 MAE, MAE Singlet fission molecules 37 2 262/9 MAE, MAE oligomers 37 2 12 MAE, MAE protein 37 2 2/1 MAE, MAE…”

Section: Transferability Test Setsmentioning

confidence: 99%

Transferable Multi-level Attention Neural Network for Accurate Prediction of Quantum Chemistry Properties via Multi-task Learning

Liu¹,

Lin²,

Jia³

et al. 2021

Preprint

View full text Add to dashboard Cite

The development of efficient models for predicting specific properties through machine learning is of great importance for the innovation of chemistry and material science. However, predicting global electronic structure properties like frontier molecular orbital HOMO and LUMO energy levels and their HOMO-LUMO gaps from the small-sized molecule data to larger molecules remains a challenge. Here we develop a multi-level attention neural network, named DeepMoleNet, to enable chemical interpretable insights being fused into multi-task learning through (1) weighting contributions from various atoms and (2) taking the atom-centered symmetry functions (ACSFs) as the teacher descriptor. The efficient prediction of 12 properties including dipole moment, HOMO, and Gibbs free energy within chemical accuracy is achieved by using multiple benchmarks, both at the equilibrium and non-equilibrium geometries, including up to 110,000 records of data in QM9, 400,000 records in MD17 and 280,000 records in ANI-1ccx for random split evaluation. The good transferability for predicting larger molecules outside the training set is demonstrated in both equilibrium QM9 and Alchemy datasets at density functional theory (DFT) level. Additional tests on non-equilibrium molecular conformations from DFT-based MD17 dataset and ANI-1ccx dataset with coupled cluster accuracy as well as the public test sets of singlet fission molecules, biomolecules, long oligomers, and protein with up to 140 atoms show reasonable predictions for thermodynamics and electronic structure properties. The proposed multi-level attention neural network is applicable to high-throughput screening of numerous chemical species in both <a></a><a>equilibrium</a> and <a></a><a></a><a>non-equilibrium</a> molecular spaces to accelerate rational designs of drug-like molecules, material candidates, and chemical reactions.

show abstract

“…There have been other attempts in materials chemistry to guide synthesis using automated equipment and algorithms capable of learning, such as the Cronin group's "chemputer" approach or the work of Raccuglia et al [296,297] More recently, Chen et al used post hoc machine learning to analyze synthesis and characterization data from experiments on UiO-67, from which they were able to infer the relative importance of reaction parameters for crystal growth. [298] Nevertheless, Moosavi et al's work remains exemplary, having elegantly combined a model which learns on-the-fly with high throughput techniques for exploring and optimizing MOF synthesis. Given that this approach can determine synthetic optima without prior chemical knowledge and with minimal human involvement, it may be a powerful potential partner to the computational methods discussed in Section 5; screening identifies target frameworks and the synthetic GA realizes them in an almost wholly automated and integrated process.…”

Section: Exploiting Feedback Loopsmentioning

confidence: 99%

High Throughput Methods in the Synthesis, Characterization, and Optimization of Porous Materials

et al. 2020

View full text Add to dashboard Cite

Notwithstanding these impediments, in many fields of materials science, solutions are being designed to mitigate hindrances to the efficient sampling of chemical space and improving the robustness of computational screening models through a combination of high throughput synthesis (HTS), characterization, and machine learning. In this review, we highlight key developments in high throughput approaches pertinent to porous materials. Specifically, we focus on developments in the field of zeolitic materials, metal-organic frameworks (MOFs), and touch upon covalent organic frameworks (COFs). Although superficially these classes of materials have little in common, there are structural links such as topology which allow for the consideration of how high throughput methods contrast with one another. By contrasting developments in different fields, we identify some underutilized approaches which could be leveraged in the future. We target structurally ordered materials because the exploration of chemical and topological space is more straightforward to enumerate, though similar approaches could be applied to disordered materials. The scale of the problem faced in materials science of targeted structure and function [1] is by no means unique; in drug discovery, HTS and chemoinformatic approaches have been used for more than 40 years in an attempt to accelerate the discovery of druggable molecules. [2,3] Unlike the drug discovery process, where Lipinski's "rule of 5" [4] provides a reliable top level sift for viable targets, identifying the most promising material for a target application requires very particular properties that may be intertwined in complex and contradictory ways. For example, thermoelectrics require high electrical conductivity and low thermal conductivity and yet these properties are typically correlated unless they can be separated. [5] Given the large scope of potential applications, the advent of the Materials Genome Initiative (MGI) [6,7] in the United States has undoubtedly helped to promote ways to solve the aforementioned obstacles and numerous others. Similarly there are major initiatives within the EU (e.g., NOMAD [8] and BIGmax [9]) and Switzerland (MARVEL [10]). In the MGI, nanoporous materials, including zeolites and MOFs, have been specifically targeted [11] and in this review, we seek to highlight particular challenges in the field of porous materials and how researchers have sought to overcome these challenges. We focus primarily on the methods used in HTS and rapid throughput/screening computational approaches. Aspects of high throughput computation applied to porous materials have been reviewed before, [12-14] as has high throughput experimentation (HTE) [15,16] but here we focus on their combination within an integrated workflow. Porous materials are widely employed in a large range of applications, in particular, for storage, separation, and catalysis of fine chemicals. Synthesis, characterization, and pre-and post-synthetic computer simulations are mostly carried out in a piecemeal a...

show abstract

Machine-Learning-Guided Morphology Engineering of Nanoscale Metal-Organic Frameworks

Cited by 50 publications

References 53 publications

Exploitation of Intrinsic Confinement Effects of MOFs in Catalysis

Exploitation of Intrinsic Confinement Effects of MOFs in Catalysis

Transferable Multi-level Attention Neural Network for Accurate Prediction of Quantum Chemistry Properties via Multi-task Learning

High Throughput Methods in the Synthesis, Characterization, and Optimization of Porous Materials

Contact Info

Product

Resources

About