Junghwa Kim, Colin Gilgenbach, Aaditya Bhat, James LeBeau
Ion implantation is widely used to dope semiconductors for electronic device fabrication, but techniques to quantify point defects and induced damage are limited. While several techniques can measure dopant concentration profiles with high accuracy, none allow for simultaneous atomic resolution structural analysis. Here, we use multislice electron ptychography to quantify the damage induced by erbium implantation in a wide band gap semiconductor 4H-SiC over a 1,000 nmtextsuperscript{3} volume region. This damage extends further into the sample than expected from implantation simulations that do not consider crystallography. Further, the technique's sensitivity to dopants and vacancies is evaluated as a function of damage. As each reconstructed analysis volume contains approximately 10$^5$ atoms, sensitivity of 10textsuperscript{18} cmtextsuperscript{-3} (in the order of 10 ppm) is demonstrated in the implantation tail region. After point defect identification, the local distortions surrounding ch{Er_{Si}} and ch{v_{Si}} defects are quantified. These results underscore the power of multislice electron ptychography to enable the investigation of point defects as a tool to guide the fabrication of future electronic devices.
{"title":"Quantifying Implantation Damage and Point Defects with Multislice Electron Ptychography","authors":"Junghwa Kim, Colin Gilgenbach, Aaditya Bhat, James LeBeau","doi":"arxiv-2409.06987","DOIUrl":"https://doi.org/arxiv-2409.06987","url":null,"abstract":"Ion implantation is widely used to dope semiconductors for electronic device\u0000fabrication, but techniques to quantify point defects and induced damage are\u0000limited. While several techniques can measure dopant concentration profiles\u0000with high accuracy, none allow for simultaneous atomic resolution structural\u0000analysis. Here, we use multislice electron ptychography to quantify the damage\u0000induced by erbium implantation in a wide band gap semiconductor 4H-SiC over a\u00001,000 nmtextsuperscript{3} volume region. This damage extends further into the\u0000sample than expected from implantation simulations that do not consider\u0000crystallography. Further, the technique's sensitivity to dopants and vacancies\u0000is evaluated as a function of damage. As each reconstructed analysis volume\u0000contains approximately 10$^5$ atoms, sensitivity of 10textsuperscript{18}\u0000cmtextsuperscript{-3} (in the order of 10 ppm) is demonstrated in the\u0000implantation tail region. After point defect identification, the local\u0000distortions surrounding ch{Er_{Si}} and ch{v_{Si}} defects are quantified.\u0000These results underscore the power of multislice electron ptychography to\u0000enable the investigation of point defects as a tool to guide the fabrication of\u0000future electronic devices.","PeriodicalId":501234,"journal":{"name":"arXiv - PHYS - Materials Science","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-09-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142187918","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
ZiJie Qiu, Luozhijie Jin, Zijian Du, Hongyu Chen, Yan Cen, Siqi Sun, Yongfeng Mei, Hao Zhang
Discovering functional crystalline materials through computational methods remains a formidable challenge in materials science. Here, we introduce VQCrystal, an innovative deep learning framework that leverages discrete latent representations to overcome key limitations in current approaches to crystal generation and inverse design. VQCrystal employs a hierarchical VQ-VAE architecture to encode global and atom-level crystal features, coupled with a machine learning-based inter-atomic potential(IAP) model and a genetic algorithm to realize property-targeted inverse design. Benchmark evaluations on diverse datasets demonstrate VQCrystal's advanced capabilities in representation learning and novel crystal discovery. Notably, VQCrystal achieves state-of-the-art performance with 91.93% force validity and a Fr'echet Distance of 0.152 on MP-20, indicating both strong validity and high diversity in the sampling process. To demonstrate real-world applicability, we apply VQCrystal for both 3D and 2D material design. For 3D materials, the density-functional theory validation confirmed that 63.04% of bandgaps and 99% of formation energies of the 56 filtered materials matched the target range. Moreover, 437 generated materials were validated as existing entries in the full database outside the training set. For the discovery of 2D materials, 73.91% of 23 filtered structures exhibited high stability with formation energies below -1 eV/atom. Our results highlight VQCrystal's potential to accelerate the discovery of novel materials with tailored properties.
{"title":"VQCrystal: Leveraging Vector Quantization for Discovery of Stable Crystal Structures","authors":"ZiJie Qiu, Luozhijie Jin, Zijian Du, Hongyu Chen, Yan Cen, Siqi Sun, Yongfeng Mei, Hao Zhang","doi":"arxiv-2409.06191","DOIUrl":"https://doi.org/arxiv-2409.06191","url":null,"abstract":"Discovering functional crystalline materials through computational methods\u0000remains a formidable challenge in materials science. Here, we introduce\u0000VQCrystal, an innovative deep learning framework that leverages discrete latent\u0000representations to overcome key limitations in current approaches to crystal\u0000generation and inverse design. VQCrystal employs a hierarchical VQ-VAE\u0000architecture to encode global and atom-level crystal features, coupled with a\u0000machine learning-based inter-atomic potential(IAP) model and a genetic\u0000algorithm to realize property-targeted inverse design. Benchmark evaluations on\u0000diverse datasets demonstrate VQCrystal's advanced capabilities in\u0000representation learning and novel crystal discovery. Notably, VQCrystal\u0000achieves state-of-the-art performance with 91.93% force validity and a\u0000Fr'echet Distance of 0.152 on MP-20, indicating both strong validity and high\u0000diversity in the sampling process. To demonstrate real-world applicability, we\u0000apply VQCrystal for both 3D and 2D material design. For 3D materials, the\u0000density-functional theory validation confirmed that 63.04% of bandgaps and\u000099% of formation energies of the 56 filtered materials matched the target\u0000range. Moreover, 437 generated materials were validated as existing entries in\u0000the full database outside the training set. For the discovery of 2D materials,\u000073.91% of 23 filtered structures exhibited high stability with formation\u0000energies below -1 eV/atom. Our results highlight VQCrystal's potential to\u0000accelerate the discovery of novel materials with tailored properties.","PeriodicalId":501234,"journal":{"name":"arXiv - PHYS - Materials Science","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142187926","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Inverse materials design has proven successful in accelerating novel material discovery. Many inverse materials design methods use unsupervised learning where a latent space is learned to offer a compact description of materials representations. A latent space learned this way is likely to be entangled, in terms of the target property and other properties of the materials. This makes the inverse design process ambiguous. Here, we present a semi-supervised learning approach based on a disentangled variational autoencoder to learn a probabilistic relationship between features, latent variables and target properties. This approach is data efficient because it combines all labelled and unlabelled data in a coherent manner, and it uses expert-informed prior distributions to improve model robustness even with limited labelled data. It is in essence interpretable, as the learnable target property is disentangled out of the other properties of the materials, and an extra layer of interpretability can be provided by a post-hoc analysis of the classification head of the model. We demonstrate this new approach on an experimental high-entropy alloy dataset with chemical compositions as input and single-phase formation as the single target property. While single property is used in this work, the disentangled model can be extended to customize for inverse design of materials with multiple target properties.
{"title":"Data-efficient and Interpretable Inverse Materials Design using a Disentangled Variational Autoencoder","authors":"Cheng Zeng, Zulqarnain Khan, Nathan L. Post","doi":"arxiv-2409.06740","DOIUrl":"https://doi.org/arxiv-2409.06740","url":null,"abstract":"Inverse materials design has proven successful in accelerating novel material\u0000discovery. Many inverse materials design methods use unsupervised learning\u0000where a latent space is learned to offer a compact description of materials\u0000representations. A latent space learned this way is likely to be entangled, in\u0000terms of the target property and other properties of the materials. This makes\u0000the inverse design process ambiguous. Here, we present a semi-supervised\u0000learning approach based on a disentangled variational autoencoder to learn a\u0000probabilistic relationship between features, latent variables and target\u0000properties. This approach is data efficient because it combines all labelled\u0000and unlabelled data in a coherent manner, and it uses expert-informed prior\u0000distributions to improve model robustness even with limited labelled data. It\u0000is in essence interpretable, as the learnable target property is disentangled\u0000out of the other properties of the materials, and an extra layer of\u0000interpretability can be provided by a post-hoc analysis of the classification\u0000head of the model. We demonstrate this new approach on an experimental\u0000high-entropy alloy dataset with chemical compositions as input and single-phase\u0000formation as the single target property. While single property is used in this\u0000work, the disentangled model can be extended to customize for inverse design of\u0000materials with multiple target properties.","PeriodicalId":501234,"journal":{"name":"arXiv - PHYS - Materials Science","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142187921","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Mary Kathleen Caucci, Jacob T. Sivak, Saeed S. I. Almishal, Christina M. Rost, Ismaila Dabo, Jon-Paul Maria, Susan B. Sinnott
Rare-earth oxides (REOs) are an important class of materials owing to their unique properties, including high ionic conductivities, large dielectric constants, and elevated melting temperatures, making them relevant to several technological applications such as catalysis, ionic conduction, and sensing. The ability to predict these properties at moderate computational cost is essential to guiding materials discovery and optimizing materials performance. Although density-functional theory (DFT) is the favored approach for predicting electronic and atomic structures, its accuracy is limited in describing strong electron correlation and localization inherent to REOs. The newly developed strongly constrained and appropriately normed (SCAN) meta-generalized-gradient approximations (meta-GGAs) promise improved accuracy in modeling these strongly correlated systems. We assess the performance of these meta-GGAs on binary REOs by comparing the numerical accuracy of thirteen exchange-correlation approximations in predicting structural, magnetic, and electronic properties. Hubbard U corrections for self-interaction errors and spin-orbit coupling are systematically considered. Our comprehensive assessment offers insights into the physical properties and functional performance of REOs predicted by first-principles and provides valuable guidance for selecting optimal DFT functionals for exploring these materials.
{"title":"Performance of Exchange-Correlation Approximations to Density-Functional Theory for Rare-earth Oxides","authors":"Mary Kathleen Caucci, Jacob T. Sivak, Saeed S. I. Almishal, Christina M. Rost, Ismaila Dabo, Jon-Paul Maria, Susan B. Sinnott","doi":"arxiv-2409.06145","DOIUrl":"https://doi.org/arxiv-2409.06145","url":null,"abstract":"Rare-earth oxides (REOs) are an important class of materials owing to their\u0000unique properties, including high ionic conductivities, large dielectric\u0000constants, and elevated melting temperatures, making them relevant to several\u0000technological applications such as catalysis, ionic conduction, and sensing.\u0000The ability to predict these properties at moderate computational cost is\u0000essential to guiding materials discovery and optimizing materials performance.\u0000Although density-functional theory (DFT) is the favored approach for predicting\u0000electronic and atomic structures, its accuracy is limited in describing strong\u0000electron correlation and localization inherent to REOs. The newly developed\u0000strongly constrained and appropriately normed (SCAN) meta-generalized-gradient\u0000approximations (meta-GGAs) promise improved accuracy in modeling these strongly\u0000correlated systems. We assess the performance of these meta-GGAs on binary REOs\u0000by comparing the numerical accuracy of thirteen exchange-correlation\u0000approximations in predicting structural, magnetic, and electronic properties.\u0000Hubbard U corrections for self-interaction errors and spin-orbit coupling are\u0000systematically considered. Our comprehensive assessment offers insights into\u0000the physical properties and functional performance of REOs predicted by\u0000first-principles and provides valuable guidance for selecting optimal DFT\u0000functionals for exploring these materials.","PeriodicalId":501234,"journal":{"name":"arXiv - PHYS - Materials Science","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142224529","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
After the discovery of ferroelectricity in HfO$_2$ based thin films a decade ago, ferroelectric Hf$_{0.5}$Zr$_{0.5}$O$_2$ (HZO) thin films are frequently being utilized in the CMOS (Complementary Metal- Oxide Semiconductor) and logic devices, thanks to their large remnant polarization, high retention and endurance. A great deal of effort has been made towards understanding the origin of ferroelectricity in epitaxial HZO thin films and controlling the microstructure at the atomic level which governs the ferroelectric phase. Nevertheless, the HZO films still suffer from fundamental questions, such as (1) the vagueness of interfacial mechanisms between HZO, buffer layer and the substrate which controls the polar phase; (2) the nature of the metastable polar phase responsible for the ferroelectricity, be it orthorhombic or rhombohedral; which are poorly understood. Here, we have addressed these issues by employing the in-situ reflection high energy electron diffraction -- assisted pulsed laser deposition and mapping the asymmetrical polar maps on high quality HZO films grown on functional perovskite oxide substrates. The interface between La$_{0.7}$Sr$_{0.3}$MnO$_3$ (LSMO) and the substrate is shown to be quite important, and a slightly rougher interface of the former destabilizes the ferroelectric phase of HZO irrespective of well-controlled growth of the ferroelectric layers. A rhombohedral-like symmetry of HZO unit cell is extracted through the x-ray diffraction asymmetrical polar maps. The ferroelectric measurements on a nearly 7 nm HZO film on STO(001) substrate display a remnant polarization close to 8 uC/cm$^2$. These results highlight the complexities involved at the atomic scale interface in the binary oxides thin films and can be of importance to the HfO$_2$-based ferroelectric community which is still at its infancy.
{"title":"Complexities in the growth and stabilization of polar phase in the Hf$_{0.5}$Zr$_{0.5}$O$_2$ thin films grown by Pulsed Laser Deposition","authors":"Deepak Kumar","doi":"arxiv-2409.06549","DOIUrl":"https://doi.org/arxiv-2409.06549","url":null,"abstract":"After the discovery of ferroelectricity in HfO$_2$ based thin films a decade\u0000ago, ferroelectric Hf$_{0.5}$Zr$_{0.5}$O$_2$ (HZO) thin films are frequently\u0000being utilized in the CMOS (Complementary Metal- Oxide Semiconductor) and logic\u0000devices, thanks to their large remnant polarization, high retention and\u0000endurance. A great deal of effort has been made towards understanding the\u0000origin of ferroelectricity in epitaxial HZO thin films and controlling the\u0000microstructure at the atomic level which governs the ferroelectric phase.\u0000Nevertheless, the HZO films still suffer from fundamental questions, such as\u0000(1) the vagueness of interfacial mechanisms between HZO, buffer layer and the\u0000substrate which controls the polar phase; (2) the nature of the metastable\u0000polar phase responsible for the ferroelectricity, be it orthorhombic or\u0000rhombohedral; which are poorly understood. Here, we have addressed these issues\u0000by employing the in-situ reflection high energy electron diffraction --\u0000assisted pulsed laser deposition and mapping the asymmetrical polar maps on\u0000high quality HZO films grown on functional perovskite oxide substrates. The\u0000interface between La$_{0.7}$Sr$_{0.3}$MnO$_3$ (LSMO) and the substrate is shown\u0000to be quite important, and a slightly rougher interface of the former\u0000destabilizes the ferroelectric phase of HZO irrespective of well-controlled\u0000growth of the ferroelectric layers. A rhombohedral-like symmetry of HZO unit\u0000cell is extracted through the x-ray diffraction asymmetrical polar maps. The\u0000ferroelectric measurements on a nearly 7 nm HZO film on STO(001) substrate\u0000display a remnant polarization close to 8 uC/cm$^2$. These results highlight\u0000the complexities involved at the atomic scale interface in the binary oxides\u0000thin films and can be of importance to the HfO$_2$-based ferroelectric\u0000community which is still at its infancy.","PeriodicalId":501234,"journal":{"name":"arXiv - PHYS - Materials Science","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142224525","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Robin Miquel, Thomas Cabout, Olga Cueto, Benoît Sklénard, Mathis Plapp
One of the most widely used active materials for phase-change memories (PCM), the ternary stoichiometric compound Ge$_2$Sb$_2$Te$_5$ (GST), has a low crystallization temperature of around 150$^circ$C. One solution to achieve higher operating temperatures is to enrich GST with additional germanium (GGST). This alloy crystallizes into a polycrystalline mixture of two phases, GST and almost pure germanium. In a previous work [R. Bayle et al., J. Appl. Phys. 128, 185101 (2020)], this crystallization process was studied using a multi-phase field model (MPFM) with a simplified thermal field calculated by a separate solver. Here, we combine the MPFM and a phase-aware electro-thermal solver to achieve a consistent multi-physics model for device operations in PCM. Simulations of memory operations are performed to demonstrate its ability to reproduce experimental observations and the most important calibration curves that are used to assess the performance of a PCM cell.
{"title":"Multi-Physics Modeling Of Phase Change Memory Operations in Ge-rich Ge$_2$Sb$_2$Te$_5$ Alloys","authors":"Robin Miquel, Thomas Cabout, Olga Cueto, Benoît Sklénard, Mathis Plapp","doi":"arxiv-2409.06463","DOIUrl":"https://doi.org/arxiv-2409.06463","url":null,"abstract":"One of the most widely used active materials for phase-change memories (PCM),\u0000the ternary stoichiometric compound Ge$_2$Sb$_2$Te$_5$ (GST), has a low\u0000crystallization temperature of around 150$^circ$C. One solution to achieve\u0000higher operating temperatures is to enrich GST with additional germanium\u0000(GGST). This alloy crystallizes into a polycrystalline mixture of two phases,\u0000GST and almost pure germanium. In a previous work [R. Bayle et al., J. Appl.\u0000Phys. 128, 185101 (2020)], this crystallization process was studied using a\u0000multi-phase field model (MPFM) with a simplified thermal field calculated by a\u0000separate solver. Here, we combine the MPFM and a phase-aware electro-thermal\u0000solver to achieve a consistent multi-physics model for device operations in\u0000PCM. Simulations of memory operations are performed to demonstrate its ability\u0000to reproduce experimental observations and the most important calibration\u0000curves that are used to assess the performance of a PCM cell.","PeriodicalId":501234,"journal":{"name":"arXiv - PHYS - Materials Science","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142187923","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Sherry Yang, Simon Batzner, Ruiqi Gao, Muratahan Aykol, Alexander L. Gaunt, Brendan McMorrow, Danilo J. Rezende, Dale Schuurmans, Igor Mordatch, Ekin D. Cubuk
Generative models trained at scale can now produce text, video, and more recently, scientific data such as crystal structures. In applications of generative approaches to materials science, and in particular to crystal structures, the guidance from the domain expert in the form of high-level instructions can be essential for an automated system to output candidate crystals that are viable for downstream research. In this work, we formulate end-to-end language-to-structure generation as a multi-objective optimization problem, and propose Generative Hierarchical Materials Search (GenMS) for controllable generation of crystal structures. GenMS consists of (1) a language model that takes high-level natural language as input and generates intermediate textual information about a crystal (e.g., chemical formulae), and (2) a diffusion model that takes intermediate information as input and generates low-level continuous value crystal structures. GenMS additionally uses a graph neural network to predict properties (e.g., formation energy) from the generated crystal structures. During inference, GenMS leverages all three components to conduct a forward tree search over the space of possible structures. Experiments show that GenMS outperforms other alternatives of directly using language models to generate structures both in satisfying user request and in generating low-energy structures. We confirm that GenMS is able to generate common crystal structures such as double perovskites, or spinels, solely from natural language input, and hence can form the foundation for more complex structure generation in near future.
{"title":"Generative Hierarchical Materials Search","authors":"Sherry Yang, Simon Batzner, Ruiqi Gao, Muratahan Aykol, Alexander L. Gaunt, Brendan McMorrow, Danilo J. Rezende, Dale Schuurmans, Igor Mordatch, Ekin D. Cubuk","doi":"arxiv-2409.06762","DOIUrl":"https://doi.org/arxiv-2409.06762","url":null,"abstract":"Generative models trained at scale can now produce text, video, and more\u0000recently, scientific data such as crystal structures. In applications of\u0000generative approaches to materials science, and in particular to crystal\u0000structures, the guidance from the domain expert in the form of high-level\u0000instructions can be essential for an automated system to output candidate\u0000crystals that are viable for downstream research. In this work, we formulate\u0000end-to-end language-to-structure generation as a multi-objective optimization\u0000problem, and propose Generative Hierarchical Materials Search (GenMS) for\u0000controllable generation of crystal structures. GenMS consists of (1) a language\u0000model that takes high-level natural language as input and generates\u0000intermediate textual information about a crystal (e.g., chemical formulae), and\u0000(2) a diffusion model that takes intermediate information as input and\u0000generates low-level continuous value crystal structures. GenMS additionally\u0000uses a graph neural network to predict properties (e.g., formation energy) from\u0000the generated crystal structures. During inference, GenMS leverages all three\u0000components to conduct a forward tree search over the space of possible\u0000structures. Experiments show that GenMS outperforms other alternatives of\u0000directly using language models to generate structures both in satisfying user\u0000request and in generating low-energy structures. We confirm that GenMS is able\u0000to generate common crystal structures such as double perovskites, or spinels,\u0000solely from natural language input, and hence can form the foundation for more\u0000complex structure generation in near future.","PeriodicalId":501234,"journal":{"name":"arXiv - PHYS - Materials Science","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142187920","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Aqshat Seth, Rutvij Pankaj Kulkarni, Gopalakrishnan Sai Gautam
Investigating Li$^+$ transport within the amorphous lithium phosphorous oxynitride (LiPON) framework, especially across a Li||LiPON interface, has proven challenging due to its amorphous nature and varying stoichiometry, necessitating large supercells and long timescales for computational models. Notably, machine learned interatomic potentials (MLIPs) can combine the computational speed of classical force fields with the accuracy of density functional theory (DFT), making them the ideal tool for modelling such amorphous materials. Thus, in this work, we train and validate the neural equivariant Interatomic potential (NequIP) framework on a comprehensive DFT-based dataset consisting of 13,454 chemically relevant structures to describe LiPON. With an optimized training (validation) energy and force mean absolute errors of 5.5 (6.1) meV/atom and 13.6 (13.2) meV/{AA}, respectively, we employ the trained potential in model Li-transport in both bulk LiPON and across a Li||LiPON interface. Amorphous LiPON structures generated by the optimized potential do resemble those generated by ab initio molecular dynamics, with N being incorporated on non-bridging apical and bridging sites. Subsequent analysis of Li$^+$ diffusivity in the bulk LiPON structures indicates broad agreement with computational and experimental literature so far. Further, we investigate the anisotropy in Li$^+$ transport across the Li(110)||LiPON interface, where we observe Li-transport across the interface to be one order-of-magnitude slower than Li-motion within the bulk Li and LiPON phases. Nevertheless, we note that this anisotropy of Li-transport across the interface is minor and do not expect it to cause any significant impedance buildup. Finally, our work highlights the efficiency of MLIPs in enabling high-fidelity modelling of complex non-crystalline systems over large length and time scales.
研究非晶态磷氧化锂（LiPON）框架内的锂$^+$输运，特别是跨Li||LiPON界面的输运，由于其非晶态性质和不同的化学计量，需要大型超单元和长时间尺度的计算模型，因此具有挑战性。值得注意的是，机器学习原子间势（MLIPs）可以将经典力场的计算速度与密度函数理论（DFT）的精确性结合起来，使其成为此类非晶材料建模的理想工具。因此，在这项工作中，我们在一个基于 DFT 的综合数据集上训练和验证了神经权变原子间势（NequIP）框架，该数据集由 13,454 个化学相关结构组成，用于描述 LiPON。经过优化的训练（验证）能量和作用力平均绝对误差分别为 5.5 (6.1) meV/atom 和 13.6 (13.2) meV/{AA}，我们将训练好的势用于模拟块状 LiPON 和跨 Li||LiPON 界面的锂传输。由优化势生成的无定形 LiPON 结构与由 ab initio 分子动力学生成的无定形 LiPON 结构非常相似，N 被结合在非桥接顶端位点和桥接位点上。此外，我们还研究了 Li$^+$ 在 Li(110)||LiPON 界面上传输的各向异性，我们观察到 Li 在界面上的传输要比 Li 在块体 Li 和 LiPON 相中的运动慢一个数量级。尽管如此，我们注意到这种跨界面锂传输的各向异性是微小的，预计不会造成任何显著的阻抗增大。最后，我们的工作凸显了 MLIPs 在大长度和时间尺度上对复杂非晶系统进行高保真建模的效率。
{"title":"Investigating Ionic Diffusivity in Amorphous Solid Electrolytes using Machine Learned Interatomic Potentials","authors":"Aqshat Seth, Rutvij Pankaj Kulkarni, Gopalakrishnan Sai Gautam","doi":"arxiv-2409.06242","DOIUrl":"https://doi.org/arxiv-2409.06242","url":null,"abstract":"Investigating Li$^+$ transport within the amorphous lithium phosphorous\u0000oxynitride (LiPON) framework, especially across a Li||LiPON interface, has\u0000proven challenging due to its amorphous nature and varying stoichiometry,\u0000necessitating large supercells and long timescales for computational models.\u0000Notably, machine learned interatomic potentials (MLIPs) can combine the\u0000computational speed of classical force fields with the accuracy of density\u0000functional theory (DFT), making them the ideal tool for modelling such\u0000amorphous materials. Thus, in this work, we train and validate the neural\u0000equivariant Interatomic potential (NequIP) framework on a comprehensive\u0000DFT-based dataset consisting of 13,454 chemically relevant structures to\u0000describe LiPON. With an optimized training (validation) energy and force mean\u0000absolute errors of 5.5 (6.1) meV/atom and 13.6 (13.2) meV/{AA}, respectively,\u0000we employ the trained potential in model Li-transport in both bulk LiPON and\u0000across a Li||LiPON interface. Amorphous LiPON structures generated by the\u0000optimized potential do resemble those generated by ab initio molecular\u0000dynamics, with N being incorporated on non-bridging apical and bridging sites.\u0000Subsequent analysis of Li$^+$ diffusivity in the bulk LiPON structures\u0000indicates broad agreement with computational and experimental literature so\u0000far. Further, we investigate the anisotropy in Li$^+$ transport across the\u0000Li(110)||LiPON interface, where we observe Li-transport across the interface to\u0000be one order-of-magnitude slower than Li-motion within the bulk Li and LiPON\u0000phases. Nevertheless, we note that this anisotropy of Li-transport across the\u0000interface is minor and do not expect it to cause any significant impedance\u0000buildup. Finally, our work highlights the efficiency of MLIPs in enabling\u0000high-fidelity modelling of complex non-crystalline systems over large length\u0000and time scales.","PeriodicalId":501234,"journal":{"name":"arXiv - PHYS - Materials Science","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142187928","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In this work we study the temperature independent resistivity of rare-earth magnetic (Gd, Tb, Dy) and non-magnetic (Lu) impurities diluted in dhcp Lanthanum. We considered a two-band system where the conduction is entirely due to $s$-electrons while the screening of the charge difference induced by the impurity is made by the $d$-electrons. We obtain an expression of the resistivity using the $T$-matrix formalism from the Dyson equation. As the electronic properties depend strongly on the band structure, we have considered two types of bands structure, a "parabolic" band and a more realistic one calculated by first principles with VASP. We verify that the exchange parameters appearing as cross products strongly affect the magnitude of the spin resistivity term; And that the role of the band structure in resonant scattering or virtual bound states, depends on the band structure. Our study, also includes the influence of the translational symmetry breaking and the excess charge introduced by the {it rare-earth} impurity on the resitivity.
在这项工作中，我们研究了稀土磁性（Gd、Tb、Dy）和非磁性（Lu）杂质稀释在 dhcpLanthanum 中与温度无关的电阻率。我们考虑了一个双带系统，其中传导完全由 s 电子完成，而杂质引起的电荷差的屏蔽则由 d 电子完成。我们利用戴森方程中的 $T$ 矩阵形式得到了电阻率的表达式。由于电子特性在很大程度上取决于能带结构，我们考虑了两种能带结构，一种是 "抛物线 "能带，另一种是用 VASP 根据第一性原理计算的更现实的能带。我们验证了作为交叉积累出现的交换参数会强烈影响自旋电阻率项的大小；而且带状结构在共振散射或虚拟束缚态中的作用取决于带状结构。我们的研究还包括平移对称性破缺和{it稀土}杂质引入的额外电荷对电阻率的影响。
{"title":"The resistivity of rare earth impurities diluted in Lanthanum (Part I)","authors":"Viviana P. Ramunni","doi":"arxiv-2409.06400","DOIUrl":"https://doi.org/arxiv-2409.06400","url":null,"abstract":"In this work we study the temperature independent resistivity of rare-earth\u0000magnetic (Gd, Tb, Dy) and non-magnetic (Lu) impurities diluted in dhcp\u0000Lanthanum. We considered a two-band system where the conduction is entirely due\u0000to $s$-electrons while the screening of the charge difference induced by the\u0000impurity is made by the $d$-electrons. We obtain an expression of the\u0000resistivity using the $T$-matrix formalism from the Dyson equation. As the\u0000electronic properties depend strongly on the band structure, we have considered\u0000two types of bands structure, a \"parabolic\" band and a more realistic one\u0000calculated by first principles with VASP. We verify that the exchange\u0000parameters appearing as cross products strongly affect the magnitude of the\u0000spin resistivity term; And that the role of the band structure in resonant\u0000scattering or virtual bound states, depends on the band structure. Our study,\u0000also includes the influence of the translational symmetry breaking and the\u0000excess charge introduced by the {it rare-earth} impurity on the resitivity.","PeriodicalId":501234,"journal":{"name":"arXiv - PHYS - Materials Science","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142224526","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Quanliang Liu, Maciej P. Polak, So Yeon Kim, MD Al Amin Shuvo, Hrishikesh Shridhar Deodhar, Jeongsoo Han, Dane Morgan, Hyunseok Oh
Materials design often relies on human-generated hypotheses, a process inherently limited by cognitive constraints such as knowledge gaps and limited ability to integrate and extract knowledge implications, particularly when multidisciplinary expertise is required. This work demonstrates that large language models (LLMs), coupled with prompt engineering, can effectively generate non-trivial materials hypotheses by integrating scientific principles from diverse sources without explicit design guidance by human experts. These include design ideas for high-entropy alloys with superior cryogenic properties and halide solid electrolytes with enhanced ionic conductivity and formability. These design ideas have been experimentally validated in high-impact publications in 2023 not available in the LLM training data, demonstrating the LLM's ability to generate highly valuable and realizable innovative ideas not established in the literature. Our approach primarily leverages materials system charts encoding processing-structure-property relationships, enabling more effective data integration by condensing key information from numerous papers, and evaluation and categorization of numerous hypotheses for human cognition, both through the LLM. This LLM-driven approach opens the door to new avenues of artificial intelligence-driven materials discovery by accelerating design, democratizing innovation, and expanding capabilities beyond the designer's direct knowledge.
{"title":"Beyond designer's knowledge: Generating materials design hypotheses via large language models","authors":"Quanliang Liu, Maciej P. Polak, So Yeon Kim, MD Al Amin Shuvo, Hrishikesh Shridhar Deodhar, Jeongsoo Han, Dane Morgan, Hyunseok Oh","doi":"arxiv-2409.06756","DOIUrl":"https://doi.org/arxiv-2409.06756","url":null,"abstract":"Materials design often relies on human-generated hypotheses, a process\u0000inherently limited by cognitive constraints such as knowledge gaps and limited\u0000ability to integrate and extract knowledge implications, particularly when\u0000multidisciplinary expertise is required. This work demonstrates that large\u0000language models (LLMs), coupled with prompt engineering, can effectively\u0000generate non-trivial materials hypotheses by integrating scientific principles\u0000from diverse sources without explicit design guidance by human experts. These\u0000include design ideas for high-entropy alloys with superior cryogenic properties\u0000and halide solid electrolytes with enhanced ionic conductivity and formability.\u0000These design ideas have been experimentally validated in high-impact\u0000publications in 2023 not available in the LLM training data, demonstrating the\u0000LLM's ability to generate highly valuable and realizable innovative ideas not\u0000established in the literature. Our approach primarily leverages materials\u0000system charts encoding processing-structure-property relationships, enabling\u0000more effective data integration by condensing key information from numerous\u0000papers, and evaluation and categorization of numerous hypotheses for human\u0000cognition, both through the LLM. This LLM-driven approach opens the door to new\u0000avenues of artificial intelligence-driven materials discovery by accelerating\u0000design, democratizing innovation, and expanding capabilities beyond the\u0000designer's direct knowledge.","PeriodicalId":501234,"journal":{"name":"arXiv - PHYS - Materials Science","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142187922","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：