First-principles study on the negative-
<mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline" overflow="scroll"><mml:mi>U</mml:mi></mml:math>
 behavior of 
<mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline" overflow="scroll"><mml:mi>K</mml:mi></mml:math>
 centers in amorphous 
<mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline" overflow="scroll"><mml:msub><mml:mi>Si</mml:mi><mml:mn>3</mml:mn></mml:msub><mml:msub><mml:mrow><mml:mrow><mml:mi ma

Kang, Gijae; Lee, Dong Hoon; Lee, Kyeongpung; Kim, Jeenu; Han, Seungwu

doi:10.1103/physrevapplied.10.064052

Cited by 8 publications

(7 citation statements)

References 79 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These results agree well with prior DFT reference data (indicated by circles), which were derived from averaging results across 40 smaller amorphous structures including 112 atoms. In particular, shoulder peaks at 90° are also well-reproduced in g (θ), which are indicative of square planar structures …”

Section: General-purpose Gnn-ipsupporting

confidence: 86%

“…In particular, shoulder peaks at 90°are also well-reproduced in g(θ), which are indicative of square planar structures. 52 We note that SevenNet-0 is exclusively trained on crystal structures, including polymorphs of Si 3 N 4 . Disordered phases like liquid or amorphous are not present in the training set.…”

Section: Model Trainingmentioning

confidence: 99%

“…(c) Angular distribution function [ g (θ)] of Si–N–Si and N–Si–N. Solid lines represent the results reproduced by SevenNet-0 and empty circles represent the DFT results (reproduced with permission from ref Copyright 2022 by the American Physical Society).…”

Section: General-purpose Gnn-ipmentioning

confidence: 99%

See 2 more Smart Citations

Scalable Parallel Algorithm for Graph Neural Network Interatomic Potentials in Molecular Dynamics Simulations

Park,

Kim,

Hwang

et al. 2024

J. Chem. Theory Comput.

Self Cite

View full text Add to dashboard Cite

Message-passing graph neural network interatomic potentials (GNN-IPs), particularly those with equivariant representations such as NequIP, are attracting significant attention due to their data efficiency and high accuracy. However, parallelizing GNN-IPs poses challenges because multiple message-passing layers complicate data communication within the spatial decomposition method, which is preferred by many molecular dynamics (MD) packages. In this article, we propose an efficient parallelization scheme compatible with GNN-IPs and develop a package, SevenNet (Scalable EquiVariance-Enabled Neural NETwork), based on the NequIP architecture. For MD simulations, SevenNet interfaces with the LAMMPS package. Through benchmark tests on a 32-GPU cluster with examples of SiO 2 , SevenNet achieves over 80% parallel efficiency in weak-scaling scenarios and exhibits nearly ideal strong-scaling performance as long as GPUs are fully utilized. However, the strong-scaling performance significantly declines with suboptimal GPU utilization, particularly affecting parallel efficiency in cases involving lightweight models or simulations with small numbers of atoms. We also pretrain SevenNet with a vast data set from the Materials Project (dubbed "SevenNet-0") and assess its performance on generating amorphous Si 3 N 4 containing more than 100,000 atoms. By developing scalable GNN-IPs, this work aims to bridge the gap between advanced machine-learning models and large-scale MD simulations, offering researchers a powerful tool to explore complex material systems with high accuracy and efficiency.

show abstract

Section: General-purpose Gnn-ipsupporting

confidence: 86%

Section: Model Trainingmentioning

confidence: 99%

See 1 more Smart Citation

Scalable Parallel Algorithm for Graph Neural Network Interatomic Potentials in Molecular Dynamics Simulations

Park,

Kim,

Hwang

et al. 2024

J. Chem. Theory Comput.

Self Cite

View full text Add to dashboard Cite

show abstract

“…Unique local orders of amorphous phases were also reported for other materials. [ 35,36,41 ] These I‐centered motifs produce two distinct CuICu bond angles around 60° and 100°, as found in the tetragonal phase. On the other hand, Cu ions with CN = 4 maintain the tetrahedral shape.…”

Section: Resultsmentioning

confidence: 84%

Origin of p‐Type Conduction in Amorphous CuI: A First‐Principles Investigation

Lee

Youn

Jeong

et al. 2020

Physica Status Solidi (b)

Self Cite

View full text Add to dashboard Cite

Recently, Jun et al. (Adv. Mater. 2018, 30, 1706573) reported Sn‐doped amorphous phases of CuI (a‐CuI) as a new class of transparent p‐type semiconductors. It is surprising that high mobilities are found despite amorphous structures tend to localize carriers, particularly the directional p orbitals of anions. To reveal the microscopic origin of the new p‐type amorphous semiconductors, density functional theory calculations are performed, and structural and electrical properties of a‐CuI are investigated. Amorphous models from melt‐quench simulations consist of structural motifs that resemble local orders in crystalline polymorphs of CuI. Despite the absence of translational symmetry, states at valence band maximum (VBM) are substantially extended in linear ways due to the hybridization between I‐5p states, explaining the high hole mobilities in the experiment. Sn‐doped a‐CuI is also investigated and it is found that Sn atoms stabilize the amorphous structure without disturbing the hole transport. Furthermore, results on nonstoichiometric models show that Cu‐deficiency increases the hole concentration in a‐CuI and also enhances the delocalization of VBM states, which is consistent with the experimental observation that the hole mobility increases with doping concentrations. Herein, the origin of hole conduction in amorphous CuI is enlightened, which is believed to contribute in the development of high‐performance p‐type amorphous materials.

show abstract

“…The nature of the sites responsible for electron and hole trapping is still under discussion. In the past, the dangling Si bond or K -center has been the target of several experimental and theoretical studies [ 5 , 6 , 7 , 8 ], being suspected as the main contributor to the charge trapping mechanism in amorphous silicon nitrides. However, experimental investigations of the defect concentration (≈

–

[ 1 , 9 , 10 ]) show that these dangling bonds only make up a small portion of the total density of electrical active defects.…”

Section: Introductionmentioning

confidence: 99%

Over- and Undercoordinated Atoms as a Source of Electron and Hole Traps in Amorphous Silicon Nitride (a-Si3N4)

et al. 2023

View full text Add to dashboard Cite

Silicon nitride films are widely used as the charge storage layer of charge trap flash (CTF) devices due to their high charge trap densities. The nature of the charge trapping sites in these materials responsible for the memory effect in CTF devices is still unclear. Most prominently, the Si dangling bond or K-center has been identified as an amphoteric trap center. Nevertheless, experiments have shown that these dangling bonds only make up a small portion of the total density of electrical active defects, motivating the search for other charge trapping sites. Here, we use a machine-learned force field to create model structures of amorphous Si3N4 by simulating a melt-and-quench procedure with a molecular dynamics algorithm. Subsequently, we employ density functional theory in conjunction with a hybrid functional to investigate the structural properties and electronic states of our model structures. We show that electrons and holes can localize near over- and under-coordinated atoms, thereby introducing defect states in the band gap after structural relaxation. We analyze these trapping sites within a nonradiative multi-phonon model by calculating relaxation energies and thermodynamic charge transition levels. The resulting defect parameters are used to model the potential energy curves of the defect systems in different charge states and to extract the classical energy barrier for charge transfer. The high energy barriers for charge emission compared to the vanishing barriers for charge capture at the defect sites show that intrinsic electron traps can contribute to the memory effect in charge trap flash devices.

show abstract

First-principles study on the negative- U behavior of K centers in amorphous Si3

Cited by 8 publications

References 79 publications

Scalable Parallel Algorithm for Graph Neural Network Interatomic Potentials in Molecular Dynamics Simulations

Scalable Parallel Algorithm for Graph Neural Network Interatomic Potentials in Molecular Dynamics Simulations

Origin of p‐Type Conduction in Amorphous CuI: A First‐Principles Investigation

Over- and Undercoordinated Atoms as a Source of Electron and Hole Traps in Amorphous Silicon Nitride (a-Si3N4)

Contact Info

Product

Resources

About