Ab initio prediction of the Li5Ge2 Zintl compound

Tipton, William W.; Matulis, Catherine A.; Hennig, Richard G.

doi:10.1016/j.commatsci.2014.06.014

Cited by 11 publications

(3 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Table summarizes selected efforts on the development and applications of nonempirical structure determination. − A majority of previous efforts were devoted to the methodological development and demonstration for molecular clusters or crystals using empirically built potentials. Since 2005, the focus of the nonempirical structure determination has shifted to practical applications as exemplified by noble metal clusters as well as the utilization of DFT calculations for “local” geometry optimization and energy assessment.…”

Section: Introductionmentioning

confidence: 99%

Machine Learning-Aided Structure Determination for TiCl₄–Capped MgCl₂ Nanoplate of Heterogeneous Ziegler–Natta Catalyst

et al. 2019

View full text Add to dashboard Cite

Dynamic reconstruction under a physicochemical environment is an intrinsic property of solid surfaces, in particular when associated with catalysis on nanosized or nanostructured materials. Here, we report nonempirical structure determination of TiCl4-capped MgCl2 nanoplates that is based on the combination of a genetic algorithm and density functional calculations. The methodology for the nonempirical structure determination was developed, and its application was demonstrated for 7MgCl2, 15MgCl2, and 15MgCl2/4TiCl4 in relation to the hidden identity of primary particles of the Ziegler–Natta catalyst. Bare MgCl2 nanoplates dominantly exposed {100} surfaces at their lateral cuts, but the chemisorption of TiCl4 induced reconstruction by stabilizing {110} surfaces. The most important finding of the present research is that TiCl4 exhibited distributed adsorption states as consequences of chemisorption on nonideal finite surfaces and the diversity of thermodynamically accessible structures. The assessment of the Ti distribution is essential for the distribution of primary structures of produced polymer, and in this study, we made these determinations.

show abstract

Section: Introductionmentioning

confidence: 99%

Machine Learning-Aided Structure Determination for TiCl₄–Capped MgCl₂ Nanoplate of Heterogeneous Ziegler–Natta Catalyst

et al. 2019

View full text Add to dashboard Cite

show abstract

“…In particular, GAs have proven their utility to identify thermodynamically stable phases efficiently; successfully identifying previously unknown materials for applications such as Li-Ge batteries 17 and solar cells 18 . Unfortunately, finding thermodynamically stable phases in ternary and quaternary systems is notoriously difficult due in part to the computationally expensive ab initio calculations required to relax and calculate the energies of GA produced structures, accounting for 99% of the algorithm's computational cost 19 .…”

Section: Introductionmentioning

confidence: 99%

Data-augmentation for graph neural network learning of the relaxed energies of unrelaxed structures

2022

Self Cite

View full text Add to dashboard Cite

Computational materials discovery has grown in utility over the past decade due to advances in computing power and crystal structure prediction algorithms (CSPA). However, the computational cost of the ab initio calculations required by CSPA limits its utility to small unit cells, reducing the compositional and structural space the algorithms can explore. Past studies have bypassed unneeded ab initio calculations by utilizing machine learning to predict the stability of a material. Specifically, graph neural networks trained on large datasets of relaxed structures display high fidelity in predicting formation energy. Unfortunately, the geometries of structures produced by CSPA deviate from the relaxed state, which leads to poor predictions, hindering the model’s ability to filter unstable material. To remedy this behavior, we propose a simple, physically motivated, computationally efficient perturbation technique that augments training data, improving predictions on unrelaxed structures by 66%. Finally, we show how this error reduction can accelerate CSPA.

show abstract

“…In particular, GAs have proven their utility to identify thermodynamically stable phases efficiently; successfully identifying novel materials for applications such as Li-Ge batteries [17] and solar cells [18]. Unfortunately, finding thermodynamically stable phases in ternary and quaternary systems is notoriously difficult due in part to the computationally expensive ab initio calculations required to relax and calculate the energies of GA produced structures, accounting for 99% of the algorithm's computational cost.…”

Section: Introductionmentioning

confidence: 99%

Data-Augmentation for Graph Neural Network Learning of the Relaxed Energies of Unrelaxed Structures

Gibson¹,

Hire²,

Hennig³

2022

Preprint

Self Cite

View full text Add to dashboard Cite

Computational materials discovery has continually grown in utility over the past decade due to advances in computing power and crystal structure prediction algorithms (CSPA). However, the computational cost of the ab initio calculations required by CSPA limits its utility to small unit cells, reducing the compositional and structural space the algorithms can explore. Past studies have bypassed many unneeded ab initio calculations by utilizing machine learning methods to predict formation energy and determine the stability of a material. Specifically, graph neural networks display high fidelity in predicting formation energy. Traditionally graph neural networks are trained on large data sets of relaxed structures. Unfortunately, the geometries of unrelaxed candidate structures produced by CSPA often deviate from the relaxed state, which leads to poor predictions hindering the model's ability to filter energetically unfavorable prior to ab initio evaluation. This work shows that the prediction error on relaxed structures reduces as training progresses, while the prediction error on unrelaxed structures increases, suggesting an inverse correlation between relaxed and unrelaxed structure prediction accuracy. To remedy this behavior, we propose a simple, physically motivated, computationally cheap perturbation technique that augments training data to improve predictions on unrelaxed structures dramatically. On our test set consisting of 623 Nb-Sr-H hydride structures, we found that training a crystal graph convolutional neural networks, utilizing our augmentation method, reduced the MAE of formation energy prediction by 66% compared to training with only relaxed structures. We then show how this error reduction can accelerates CSPA by improving the model's ability to filter out energetically unfavorable structures accurately.

show abstract

Ab initio prediction of the Li5Ge2 Zintl compound

Cited by 11 publications

References 25 publications

Machine Learning-Aided Structure Determination for TiCl₄–Capped MgCl₂ Nanoplate of Heterogeneous Ziegler–Natta Catalyst

Machine Learning-Aided Structure Determination for TiCl₄–Capped MgCl₂ Nanoplate of Heterogeneous Ziegler–Natta Catalyst

Data-augmentation for graph neural network learning of the relaxed energies of unrelaxed structures

Data-Augmentation for Graph Neural Network Learning of the Relaxed Energies of Unrelaxed Structures

Contact Info

Product

Resources

About

Ab initio prediction of the Li5Ge2 Zintl compound

Cited by 11 publications

References 25 publications

Machine Learning-Aided Structure Determination for TiCl4–Capped MgCl2 Nanoplate of Heterogeneous Ziegler–Natta Catalyst

Machine Learning-Aided Structure Determination for TiCl4–Capped MgCl2 Nanoplate of Heterogeneous Ziegler–Natta Catalyst

Data-augmentation for graph neural network learning of the relaxed energies of unrelaxed structures

Data-Augmentation for Graph Neural Network Learning of the Relaxed Energies of Unrelaxed Structures

Contact Info

Product

Resources

About

Machine Learning-Aided Structure Determination for TiCl₄–Capped MgCl₂ Nanoplate of Heterogeneous Ziegler–Natta Catalyst

Machine Learning-Aided Structure Determination for TiCl₄–Capped MgCl₂ Nanoplate of Heterogeneous Ziegler–Natta Catalyst