首 页
滾動信息 更多 >>
npj Computational Materials 2019年影响因子达到9... (2020-07-04)
npj Computational Materials获得第一个SCI影响因... (2018-09-07)
英文刊《npj Computational Materials(计算材料学... (2017-05-15)
開放的數據庫 (2015-12-17)
上海矽酸鹽所與Nature出版集團聯合創辦的中國首本n... (2015-11-27)
快捷服務
最新文章 研究綜述
過刊浏覽 作者須知
期刊編輯 審稿須知
相關鏈接
· 在線投稿
會議信息
友情鏈接
  凤凰娱乐
  無機材料學報
  OQMD數據庫
mm.jpg
期刊介紹
  《npj 计算材料学》是在线出版、完全開放獲取的国际学术期刊。发表结合计算模拟与设计的材料学一流的研究成果。本刊由凤凰娱乐与英国自然出版集团(Nature Publishing Group,NPG)以伙伴关系合作出版。
  主編爲陳龍慶博士,美國賓州大學材料科學與工程系、工程科學與力學系、數學系的傑出教授。
  共同主编为陈立东研究员,凤凰娱乐研究员高性能陶瓷与超微结构国家重点实验室主任。
  辦刊目的與報道範圍
  《npj 计算材料学》是在线出版、完全開放獲取的国际...
【查看詳細】
近期文章 更多 >>

Ab initio molecular dynamics and materials design for embedded phase-change memory(面向嵌入式相变存储应用的材料设计与优化)
Liang Sun, Yu-Xing Zhou, Xu-Dong Wang, Yu-Han Chen, Volker L. Deringer, Riccardo Mazzarello, and Wei Zhang
npj Computational Materials 7:29(2021) doi:s41524-021-00496-7
Published online:21 Feb2021

Abstract| Full Text | PDF OPEN

摘要:基于鍺銻碲Ge2Sb2Te5研發的相變存儲器件兼具快速讀寫與非易失存儲特性。該材料成分並不適用于在高溫環境下工作的嵌入式存儲應用,例如汽車工業等。富鍺合金如Ge2Sb1Te2具有較高的結晶化溫度,具有被用于嵌入式相變存儲器件的潛力,但是其原子結構以及結構性能關系尚不清晰。本工作使用第一性原理分子動力學計算深入研究了十余種鍺銻碲合金非晶結構,使用原子位置重疊”SOAP方法定量化表征了鍺銻碲合金與單質鍺的非晶相似度,闡明了非晶富鍺合金熱穩定性機理以及相分離趨勢根源,爲嵌入式相變存儲應用提供了理論支撐並給出了材料設計的方案。 

Abstract:The Ge2Sb2Te5 alloy has served as the core material in phase-change memories with high switching speed and persistent storage capability at room temperature. However widely used, this composition is not suitable for embedded memories—for example, for automotive applications, which require very high working temperatures above 300°C. Ge–Sb–Te alloys with higher Ge content, most prominently Ge2Sb1Te2 (‘212’), have been studied as suitable alternatives, but their atomic structures and structure–property relationships have remained widely unexplored. Here, we report comprehensive first-principles simulations that give insight into those emerging materials, located on the compositional tie-line between Ge2Sb1Te2 and elemental Ge, allowing for a direct comparison with the established Ge2Sb2Te5 material. Electronic-structure computations and smooth overlap of atomic positions (SOAP) similarity analyses explain the role of excess Ge content in the amorphous phases. Together with energetic analyses, a compositional threshold is identified for the viability of a homogeneous amorphous phase (‘zero bit’), which is required for memory applications. Based on the acquired knowledge at the atomic scale, we provide a materials design strategy for high-performance embedded phase-change memories with balanced speed and stability, as well as potentially good cycling capability. 

編輯概述

嵌入式相變存儲器:材料的設計與優化

富鍺合金相變存儲器件可用于汽車微控制芯片的大規模集成,有著廣闊的市場前景,但過量的鍺元素極易引發相分離,進而導致器件失效。近日,西安交通大學CAID材料創新設計中心的張偉教授團隊利用第一性原理分子動力學方法模擬了十余種不同成分比例鍺銻碲合金的非晶結構,全面分析了非晶鍺銻碲的局部原子構型和化學成鍵機制,量化表征了鍺銻碲合金與單質鍺的非晶相似度,並類比于合金形成能的概念,計算了非晶鍺銻碲的相分離趨勢。結果表明,偏離經典GeTe-Sb2Te3二元平衡線,過量鍺可大幅提升鍺銻碲合金與單質鍺的非晶結構相似性,從而增強非晶鍺銻碲的熱穩定性。但鍺含量超過55%,接近Ge4Sb1Te2時,非晶鍺銻碲的結合機制發生根本性轉變,導致相分離形成單質鍺與傳統鍺銻碲合金的趨勢大幅提升。因此,該工作提出一個合理的合金成分選擇範圍,如鍺銻碲三元圖中綠色虛線所示。基于該範圍內的鍺銻碲合金,可通過少量碳氮摻雜或微縮器件尺寸引入納米尺寸效應進一步提升富鍺合金的非晶熱穩定性,從而滿足嵌入式相變存儲芯片在存儲穩定性與循環工作壽命上的需求。 

Editorial Summary

Embedded phase-change memory: Materials design and optimization

Ge-rich Ge-Sb-Te (GGST) phase-change alloys are to be exploited as embedded memory for the automotive industry. However, due to the excess of Ge, phase segregation becomes a critical issue in these alloys, leading to the failure of the embedded devices. Recently, Prof. Wei Zhang from the Center for Alloy Innovation and Design (CAID) of Xi’an Jiaotong University and his colleagues performed a comprehensive first-principles study on more than ten different GGST alloys, aiming at understanding the local atomic structural features and chemical bonding properties. The results show that the presence of the additional Ge content off the GeTe-Sb2Te3 pseudo-binary line increases the similarity between amorphous Ge-rich alloys and amorphous Ge, leading to an enhanced amorphous stability. However, the energy analysis reveals that the GGST alloys with more than 55% Ge, e.g., Ge4Sb1Te2, have a strong tendency towards phase segregation, forming an energetically more favorable phase of stoichiometric GST and amorphous Ge. Based on these ab initio simulations, this work suggests an optimal compositional pool, as marked by the green dash circle in the Ge–Sb–Te ternary diagram. Starting from compositional pool, slight doping of nitrogen or carbon or device miniaturization are suggested to be employed, which could potentially lead to balanced performance of stability, speed as well as cycling endurance.

Strain-induced resonances in the dynamical quadratic magnetoelectric response of multiferroics (多铁动力学二次磁电响应中的应变感应共振)
S. Omid Sayedaghaee, Charles Paillard, Sergey Prosandeev, Bin Xu & Laurent Bellaiche
npj Computational Materials 6:60(2020)
doi:s41524-020-0311-z
Published online:21 May2020

Abstract| Full Text | PDF OPEN

摘要:近年來,對磁電(ME)效應——這一多鐵材料中鐵電有序與磁有序間交叉耦合的研究興趣經曆了重大的複興。近期大量的工作不僅研究了利用磁場(或電場)對極化(或磁化)的交叉控制來設計傳感器,驅動器,換能器和存儲設備,更旨在清楚地理解ME響應的來源以及與之相關的新穎效應。在這裏,我們推導出解析模型用于理解多鐵體系中ME效應的驚人和新穎的動力學,並通過原子模擬進一步確認該現象的存在。具體而言,揭示了應變可以導致電聲磁振子的存在,其爲一種聲學和光學聲子與磁振子混合的新型准粒子。該粒子可導致共振,從而顯著增強了磁電響應。而且,在本工作之前,尚未有工作討論變頻磁場導致動態二次ME響應,此爲二次諧波過程。這些過程表明處理此類系統時應考慮非線性的是十分重要的 

Abstract:For the last few years, the research interest in magnetoelectric (ME) effect, which is the cross-coupling between ferroelectric and magnetic ordering in multiferroic materials, has experienced a significant revival. The extensive recent studies are not only conducted towards the design of sensors, actuators, transducers, and memory devices by taking advantage of the cross-control of polarization (or magnetization) by magnetic (or electric) fields, but also aim to create a clearer picture in understanding the sources of ME responses and the novel effects associated with them. Here we derive analytical models allowing to understand the striking and novel dynamics of ME effects in multiferroics and further confirm it with atomistic simulations. Specifically, the role of strain is revealed to lead to the existence of electroacoustic magnons, a new quasiparticle that mixes acoustic and optical phonons with magnons, which results in resonances and thus a dramatic enhancement of magnetoelectric responses. Moreover, a unique aspect of the dynamical quadratic ME response under a magnetic field with varying frequencies, which is the second harmonic generation (SHG), has not been discussed prior to the present work. These SHGs put emphasis on the fact that nonlinearities should be considered while dealing with such systems.

Editorial Summary

Strain matters: resonance enhanced dynamic quadratic magnetoelectric effect應變生新花:共振增強動態二次磁電效應

本研究發現了應變可以誘導多鐵體系中動態二次磁電耦合響應的共振增強。由美國阿肯色大學Bellaiche教授領導的國際團隊基于唯象模型和分子動力學模擬研究了應變對于多鐵體系中的動態二次磁電耦合響應的影響。他們首先構建了多鐵體系應變、磁和電多場耦合的唯象朗道模型,並基于該模型的推導得到了二次磁電系數隨外場頻率的變化關系。他們發現,當體系允許應變存在時,體系中會出現一種新的元激發准粒子,即電聲磁振子,其爲聲學/光學聲子與磁振子耦合態。該准粒子導致二次磁電系數在某些頻率會極大的增強,發生所謂共振現象。爲證實上述結果,他們基于典型多鐵體系BiFeO3的等效哈密頓量開展了分子動力學模擬。通過施加不同頻率的交變外場,他們發現在應變允許發生時的確觀察到磁電系數的共振增強,而當應變不能發生時,共振增強消失,從而證實了唯象模型的結果。該研究提出一種全新的強磁電響應的物理機制,且可以單相材料中實現,不僅豐富磁電耦合的物理圖像,而且有望用于實現新型磁電耦合器件 

Stain induced resonant enhancement of dynamic quadratic magnetoelectric (ME) response in multiferroics was discovered. An international team led by Professor L. Bellaiche from the University of Arkansas studied the role of strain on the dynamic quadratic ME coupling in multiferroic systems based on phenomenological model and molecular dynamics simulations. They first proposed a phenomenological Landau model to describe the coupling of strain, magnetic and electric variants in multiferroic system. By derivation of this model, the relationship between the dynamic quadratic ME coefficient and the frequency of external field was obtained. They found that when strain exists, a new kind of elementary excitation, namely electroacoustic magnon, will appear in the system, which is the coupling state of acoustic/optical phonon and magnon. This excitation results in a dramatic enhancement of the quadratic ME coefficient at certain frequencies, resulting in the so-called resonance phenomenon. To confirm the above results, they carried out molecular dynamics simulations based on the effective Hamiltonian of BiFeO3. By applying alternating external fields of different frequencies, they found that the resonance enhancement of ME coefficient can be observed when  strain was allowed, while it disappears when the strain is clamped, which confirms the results of the phenomenological model. This study proposes a new mechanism of ME response, which can be realized in single-phase materials. It not only enriches the physics of ME coupling, but is also expected to be used to realize new ME devices.

Simulating Raman spectra by combining first-principles and empirical potential approaches with application to defective MoS2(第一性原理結合經驗勢方法模擬拉曼光譜並應用于缺陷MoS2的研究)
Zhennan KouArsalan HashemiMartti J. PuskaArkady V. Krasheninnikov & Hannu-Pekka Komsa
npj Computational Materials 6:59(2020)
doi:s41524-020-0320-y
Published online:15 May 2020

Abstract| Full Text | PDF OPEN

摘要:二維過渡金屬雙硫屬化合物在光電、催化或傳感器件中的成功應用,很大程度上依賴于材料的質量,即厚度均勻性、晶界的存在以及點缺陷的類型和濃度。拉曼光譜是探測這些因素的一個強大而無損的工具,但光譜的解釋,特別是不同貢獻的區分並不簡單。與模擬光譜進行比較是有益的,但對于有缺陷的材料系統,由于所涉及的尺寸太大,第一性原理模擬通常在計算上過于昂貴。在此,本研究提出了一種第一性原理和經驗勢結合的方法來模擬缺陷材料的拉曼光譜,並將其應用于具有MoS空位隨機分布的單層MoS2中。我們研究了在何種程度上可以區分空穴類型,並提供隨缺陷濃度變化時拉曼光譜演化的起源分析。我們將模擬光譜應用于聲子局域模型(之前實驗中使用的模型)來評估缺陷濃度。結果顯示,該模型的最簡單形式不足以完全捕獲峰形,但是當聲子局域類型與完整的聲子譜聯系起來的時候,該模型獲得的結果與實驗數據有很好的一致性 

Abstract:Successful application of two-dimensional transition metal dichalcogenides in optoelectronic, catalytic, or sensing devices heavily relies on the materials’ quality, that is, the thickness uniformity, presence of grain boundaries, and the types and concentrations of point defects. Raman spectroscopy is a powerful and nondestructive tool to probe these factors but the interpretation of the spectra, especially the separation of different contributions, is not straightforward. Comparison to simulated spectra is beneficial, but for defective systems first-principles simulations are often computationally too expensive due to the large sizes of the systems involved. Here, we present a combined first-principles and empirical potential method for simulating Raman spectra of defective materials and apply it to monolayer MoS2 with random distributions of Mo and S vacancies. We study to what extent the types of vacancies can be distinguished and provide insight into the origin of different evolutions of Raman spectra upon increasing defect concentration. We apply our simulated spectra to the phonon confinement model used in previous experiments to assess defect concentrations, and show that the simplest form of the model is insufficient to fully capture peak shapes, but a good match is obtained when the type of phonon confinement and the full phonon dispersion relation are accounted for.

Editorial Summary

Simulating Raman spectra in defective MoS2: by first-principles and empirical potential approaches模擬缺陷MoS2的拉曼光譜:第一性原理結合經驗勢方法

该研究展示了一种基于经验势和第一性原理计算的方法,用于模擬缺陷材料的拉曼光谱,其中经验势用于评估缺陷系统的振动模式,然后与第一性原理计算得到的拉曼张量进行结合。来自芬兰阿尔托大学应用物理系的Hannu-Pekka Komsa領導的團隊,構建了該組合方法,並研究了在何種程度上可以區分空位類型,最後探討了隨缺陷濃度增加時拉曼光譜不同演化的機理。這種方法不僅能可靠地模擬拉曼光譜,還可深入了解缺陷系統中振動模式的物理內涵,以及如何用拉曼光譜對它們進行探測。作者利用該方法研究了單層MoS2中的空位缺陷,捕獲了缺陷對突出峰位移和不對稱展寬的影響,其結果與實驗數據定性一致。此外,他們使用聲子局域模型來擬合其模擬的拉曼光譜,以評估該模型在缺陷材料中的適用性。結果發現,當同時考慮完整的聲子色散關系和局域類型時,該模型非常有效。通過本研究發現,只要有適當的經驗勢,就可以有效地評估缺陷系統的拉曼光譜 

A method for simulating Raman spectra of defective materials based on a combination of empirical potentials and first-principles calculations is demonstrated, in which the empirical potentials are used to evaluate the vibrational modes of the defective system, which are then combined with Raman tensors evaluated from the first-principles calculations. A team led by Hannu-Pekka Komsa from the Department of Applied Physics, Aalto University, Finland, had constructed this combined method and studied to what extent the types of vacancies can be distinguished, and finally provided insight into the origin of different evolutions of Raman spectra upon increasing defect concentration. This approach allows them to not only reliably simulate Raman spectra, but also gain insights into the physics of vibrational modes in defective systems and how they can be probed with Raman spectroscopy. The authors used this method to study vacancies in monolayer MoS2 and captured the effect of defects on the shifts and on the asymmetric broadening of the prominent peaks, with the results being in a qualitative agreement with experimental data. They then used the phonon confinement model to fit their simulated Raman spectra to assess the applicability of the model in the context of defective materials. They found it to work well when the full dispersion relation and the type of confinement are accounted for. The approach presented here allows for efficient evaluation of the Raman spectra of defective systems provided that an appropriate empirical potential is available.

Predicting synthesizable multi-functional edge reconstructions in two-dimensional transition metal dichalcogenides(预测二维过渡金属双卤化物中可合成的多功能边缘重构)
Guoxiang HuVictor FungXiahan SangRaymond R. Unocic & P. Ganesh
npj Computational Materials 6:120(2020)
doi:s41524-020-0327-4
Published online:13 August 2020

Abstract| Full Text | PDF OPEN

摘要:二維(2D)過渡金屬雙硫屬化合物(TMDC)由于其獨特的多樣性和可調性,尤其是其邊緣特性,已經引起了人們極大的興趣。除了常規六邊形2D材料常見的扶手椅邊和鋸齒形邊緣外,通過對合成條件的精細調控,可以實現更複雜的邊緣重構。然而目前缺乏對整個可合成的邊緣重構家族的研究。本研究開發了一種集成計算方法,整合了構型生成、力的弛豫以及電子結構計算等流程,以系統、有效地發現的重構邊緣和篩選其功能特性。以MoS2爲模型系統,對數百條重構邊緣進行篩選,發現超過160條重構邊緣比傳統邊緣更穩定。更令人興奮的是,我們發現了9個新的可合成的重構邊緣,具有熱力學穩定性,此外還成功地再現了3個最近合成的邊緣結構。我們還發現預測的重構邊緣具有多功能特性(與常規邊緣相比,還具有接近最佳的析氫活性),是析氫反應(HER)的理想選擇,並且具有半金屬性,且磁矩變化很大,使其特別適合納米自旋電子應用。我們的工作揭示了在2D TMDC中存在大量可合成的重構邊緣,並打開了2D材料“本征”邊緣工程多功能性的材料設計新範式 

Abstract:Two-dimensional (2D) transition metal dichalcogenides (TMDCs) have attracted tremendous interest as functional materials due to their exceptionally diverse and tunable properties, especially in their edges. In addition to the conventional armchair and zigzag edges common to hexagonal 2D materials, more complex edge reconstructions can be realized through careful control over the synthesis conditions. However, the whole family of synthesizable, reconstructed edges remains poorly studied. Here, we develop a computational approach integrating ensemble-generation, force-relaxation, and electronic-structure calculations to systematically and efficiently discover additional reconstructed edges and screen their functional properties. Using MoS2 as a model system, we screened hundreds of edge-reconstruction to discover over 160 reconstructed edges to be more stable than the conventional ones. More excitingly, we discovered nine new synthesizable reconstructred edges with record thermodynamic stability, in addition to successfully reproducing three recently synthesized edges. We also find our predicted reconstructed edges to have multi-functional properties—they show near optimal hydrogen evolution activity over the conventional edges, ideal for catalyzing hydrogen-evolution reaction (HER) and also exhibit half-metallicity with a broad variation in magnetic moments, making them uniquely suitable for nanospintronic applications. Our work reveals the existence of a wide family of synthesizable, reconstructed edges in 2D TMDCs and opens a new materials-by-design paradigm of ‘intrinsic’ edge engineering multifunctionality in 2D materials.

Editorial Summary

2D transition metal dichalcogenides: Predicting synthesizable multi-functional edge reconstructions二維過渡金屬雙鹵化物:預測可合成的多功能邊緣重構

由于2D過渡金屬雙硫屬化合物(TMDC中整個可合成的重構邊緣家族仍然未知,該研究開發出了一種集成計算方法,可以快速有效地發現2D TMDC體系中更多可合成的功能性邊緣,爲計算篩選和發現其他功能重構的TMDC邊緣提供了獨特的機會。來自美國橡樹嶺國家實驗室納米材料科學中心的Guoxiang HuP. Ganesh共同領導的研究團隊,從構型文件生成開始,使用力場方法,篩選了材料的穩定邊緣。並使用基于DFT的電子結構計算,進一步細化了所獲得的穩定邊緣,以生成相圖並篩選其功能特性。以MoS2爲例,篩選出了2H1TMoS2625個邊緣構型,並預測了穩定的邊緣以指導實驗合成。隨後,他們研究了這些邊緣的功能特性,發現許多這些內在可調的邊緣重構,對于析氫反應來說是接近最佳的。因此,他們的研究爲預測??2D材料的可合成功能邊緣提供了一個全面而經濟的計算方案,並爲實驗研究人員提供了有用的指導。許多研究已經通過Edisonian方法中的外部摻雜來調控了TMDCs的催化、電子和磁性能,但這項研究的優點是發現了一系列“本征”(基于金屬/硫屬元素比)可調材料。該研究成功開啓了一種新的材料設計範式,可搜索和發現其他具有特定多功能的、有潛在邊緣多型性的2D“本征”邊緣重構家族,並適用于納米尺度的廣泛應用 

Due to the whole family of synthesizable reconstructed edges in 2D TMDCs remains largely unknown, a computational approach to rapidly and efficiently discover more synthesizable functional edges in the family of 2D TMDCs is developed, which presents a unique opportunity to computationally screen and discover additional functional reconstructed TMDC edges. A team co-led by Guoxiang Hu and P. Ganesh from the Center for Nanophase Materials Sciences, Oak Ridge National Laboratory, USA, starting with configuration ensemble generations, screened for stable edges using a computationally affordable force-field method. The obtained stable edges are then further refined with DFT based electronic-structure calculations to generate phase diagrams and screen for their functional properties. Using MoS2 as an example, the authors screened 625 edge configurations for 2H and 1T MoS2 phases, and predicted stable edges to guide the experimental synthesis. Subsequently they studied the functional properties of these edges and discovered many of these intrinsically tunable edge reconstructions to be near-optimal for hydrogen evolution reaction. Their study thus provides a comprehensive yet affordable computational scheme for predicting synthesizable functional edges of 2D materials and provides useful guidelines to experimental researchers. Many studies have investigated tuning the catalytic, electronic and magnetic properties of TMDCs by external doping in an Edisonian approach, but the merit of this study is to discover a family of ‘intrinsically’ (based on the metal/chalcogen ratio) tunable materials with widely varying functional properties. Success of this study opens a new materials-by-design paradigm to search and discover other families of 2D ‘intrinsic’ edge reconstructions with specific multi-functionalities, with potential edge polytypism, for a wide range of nanoscale applications.

Perfect short-range ordered alloy with line-compound-like properties in the ZnSnN2:ZnO system (ZnSnN2ZnO系統中具有線化合物性質的完美短程有序合金)
Jie PanJacob J. CordellGarritt J. TuckerAndriy ZakutayevAdele C. Tamboli & Stephan Lany
npj Computational Materials 6:120(2020)
doi:s41524-020-0331-8
Published online:13 August 2020

Abstract| Full Text | PDF OPEN

摘要:我們提出了一種新的固體材料相,它是一種無序的固溶體,但具有許多有序線化合物的特征。這些新的物理現象源于完美的短程序,從而保留了局域八隅律規則。我們采用第一性原理計算、模型哈密頓量的蒙特卡羅模擬和基于短程序擴展的固溶模型對雙亞晶格混合半導體合金(ZnSnN21-xZnO2x开展模拟。我们证明,这种独特的固溶体必须在“幻术”组分中出现,其电子特征没有无序引起的電荷局域化,因此具有类似有序相的优良载流子传输。有趣的是,该相具有传统固溶体模型(如规则溶体和带隙弯曲模型等)没有的奇异性。在热力学上,该合金相的形成焓急剧降低(类似线化合物),但仍需要长程无序带来的熵才能在实验温度下稳定该化合物 

Abstract:We present a new solid-state material phase which is a disordered solid solution but offers many ordered line-compound features. The emergent physical phenomena are rooted in the perfect short-range order which conserves the local octet rule. We model the dual-sublattice-mixed semiconductor alloy (ZnSnN2)1?x(ZnO)2x(ZnSnN2)1?x(ZnO)2x using first-principles calculations, Monte-Carlo simulations with a model Hamiltonian, and an extension of the regular solution model by incorporating short-range order. We demonstrate that this unique solid solution, occurring at a “magic” composition, can provide an electronically pristine character without disorder-induced charge localization and, therefore, a superior carrier transport similar to ordered phases. Interestingly, this phase shows singularities that are absent in the conventional solid-solution models, such as the regular solution and band-gap bowing model. Thermodynamically, this alloy phase has a sharply reduced enthalpy at its composition (like a line compound), but it still requires the entropy from long-range disorder to be stabilized at experimentally accessible temperatures.

Editorial Summary

Singularity in solid solution: disordered structure but ordered properties固溶體的奇異點:具有有序化合物性能的無序固溶體

本文通過計算預測了一種具有固溶體結構特征但類似有序化合物物理性能的新奇固體相。來自美國可再生能源實驗室(NREL)的團隊基于綜合利用多種計算模型,即固溶體模型、蒙特卡洛模擬和第一性原理,針對雙亞晶格混合半導體合金(ZnSnN21-xZnO2x的相結構開展了研究。該體系可以看成是由ON四面體組合而成。他們首先提出了一種用于描述短程有序的參量,基于該參量構建了描述該化合物形成焓的經驗表達式。進而基于該能量表達式開展蒙特卡洛模擬,由此獲得所有固溶體組分(0<x<0.5)能量最低的結構,並開展密度泛函計算精確計算這些結構的能量。有趣的是,他們發現x=0.25成为所谓“幻数”组分。在该组分,混合焓随组分变化的曲线出现明显奇异点。分析表明,该组分对应的结构具有完美的短程有序,即所有四面体内部都满足八隅律,就是说八面体内阴阳离子化合价之和为零。然而八面体间的连接仍然是无序的。因此该体系处于短程有序长程无序的特殊状态。更有意义的是,对该体系电子结构分析表明,该组分结构的带隙明显偏离正常固溶体模型预测结果,且能带边缘的電子态具有很强的离域性,这是有序化合物的典型特征,其电输运性能应明显优于无序固溶体。通过进一步理论计算,研究人员证明了上述新奇化合物存在的可能性并预测可能存在的温度区间。该工作的意义在于,提出了在多元固溶体中可能存在特别的短程有序长程无序结构,其具有类型有序化合物的物理性质。这一发现为新型功能材料设计提供新思路和更广阔的搜索空间 

A novel solid phase with the characteristics of solid solution like structure but ordered compound like physical properties was predicted. A team from the National Renewable Energy Laboratory (NREL) utilized a set of calculation models, namely solid solution model, Monte Carlo simulation and density functional theory calculation to study the phase structure of (ZnSnN2)1-x(ZnO)2x, a dual-sublattice-mixed semiconductor alloy. This system can be regarded as a combination of O and N tetrahedrons. The authors first proposed an order parameter to describe the short-range order, based on which an empirical expression describing the formation enthalpy of the compound was constructed. Based on the energy expression, Monte Carlo simulations were carried out to obtain the structures for each composition (0 < x < 0.5). Then density functional calculations were carried out to determine the energy of these structures accurately. Interestingly, they found that x = 0.25 becomes a so-called "magic number" composition. In this component, a singularity appears in the curve of the enthalpy of mixing with the component. The analysis shows that the corresponding structure of the component has perfect short-range order, that is, all tetrahedrons satisfy the local octet rule, i.e. the sum of the valence of cation and anion within one tetrahedron is zero. While the connection between octahedrons is still disordered. Therefore, the system is in a special state with short-range order but long-range disorder. More importantly, the analysis of the electronic structure shows that the band gap of the structures with the “magic” composition deviate from the predicted results of the normal solid solution model, and the electronic states at the edge of the energy band exhibit strong delocalization, which is a typical characteristic of ordered compounds and could lead to superior electrical transport properties over that of disordered solid solutions. By further theoretical calculation, the researchers demonstrated the possibility of the existence of these novel compounds and predicted the possible temperature range. The significance of this work lies in the prediction that special short-range ordered but long-range disordered structures may exist in multicomponent solid solutions, which could exhibit physical properties of typical ordered compounds. This study provides a broader space for new material design and discovery.

Machine learning for accelerating the discovery of high-performance donor/acceptor pairs in non-fullerene organic solar cells (機器學習助力高性能非富勒烯有機太陽能電池供體/受體材料的開發)
Yao Wu, Jie Guo, Rui Sun & Jie Min
npj Computational Materials 6:120(2020)
doi:s41524-020-00388-2
Published online:13 August 2020

Abstract| Full Text | PDF OPEN

摘要:通過人工智能、計算機科學和材料的合成與優化有機結合,可以大幅促進高性能有機光伏材料的開發。在這個過程中,機器學習模型與算法的選擇發揮著至關重要的作用。本研究以565組供體/受體對的數據爲訓練集通過五種常見算法構建了機器學習模型,並評估了這些模型應用于指導材料設計和供體/受體配對物篩選的可靠性,結果顯示基于隨機森林(RF)和提升回歸樹(BRT)算法的模型表現優異。因此本研究進一步利用RF和BRT模型對3200萬組供體/受體對進行性能預測和篩選,並從該數據庫中選出六組供體/受體對進行合成與器件表征,從而獲得它們的實驗光電轉化效率。實驗驗證結果顯示,基于RF的機器學習模型更適合用于有機光伏材料的高通量篩選。這爲材料的設計和供體/受體配對物的選擇提供了新的思路,從而加速有機太陽能電池的發展 

Abstract:Integrating artificial intelligence (AI) and computer science together with current approaches in material synthesis and optimization will act as an effective approach for speeding up the discovery of high-performance photoactive materials in organic solar cells (OSCs). Yet, like model selection in statistics, the choice of appropriate machine learning (ML) algorithms plays a vital role in the process of new material discovery in databases. In this study, we constructed five common algorithms, and introduced 565 donor/acceptor (D/A) combinations as training data sets to evaluate the practicalities of these ML algorithms and their application potential when guiding material design and D/A pairs screening. Thus, the best predictive capabilities are provided by using the random forest (RF) and boosted regression trees (BRT) approaches beyond other ML algorithms in the data set. Furthermore, >32 million D/A pairs were screened and calculated by RF and BRT models, respectively. Among them, six photovoltaic D/A pairs are selected and synthesized to compare their predicted and experimental power conversion efficiencies. The outcome of ML and experiment verification demonstrates that the RF approach can be effectively applied to high-throughput virtual screening for opening new perspectives to design of materials and D/A pairs, thereby accelerating the development of OSCs.

Editorial Summary

Donor/acceptor pairs screening:in organic solar cells有機太陽能電池供體-受體材料的配對:需要“紅娘”!

傳統有機光伏材料研究方法包括對化學合成、供體/受體材料匹配和器件制備進行精細控制及優化,需要大量的資源投入和較長的研究周期,限制了有機光伏産業的發展與實際商業應用。近日,武漢大學闵傑研究員課題組以被文獻報道過的565組基于非富勒烯小分子受體材料和聚合物給體材料的供體/受體對數據庫,采用ASCII碼字符串的表達方式將供體/受體材料的化學結構進行轉化成二進制機器語言,並與其相關光伏參數一起作爲訓練集和驗證集,分別采用線性回歸(LR)、多類邏輯回歸(MLR)、提升回歸樹(BRT)、人工神經網絡(ANN)和隨機森林(RF)算法構建機器學習模型(如圖1所示),可對供體、受體材料以及活性層供體/受體對的適配性進行快速的評估和篩選。研究人員對五種典型的算法模型進行評估發現,基于RF和BRT模型的預測結果與測試集中真實值的皮爾森相關系數(r)均超過了0.7,說明該兩種模型是進行這類機器學習的最佳表達方式。進一步,研究人員通過原有數據集並結合RF和BRT模型,分別篩選和計算出了3200萬個給受體對。爲了驗證上述模型是否能夠有效地指導設計新的有機光伏體系,研究人員從該數據庫中選出六組易于合成且具有高效率的給受體對,並進行了材料合成、器件制備與表征。研究結果表明,相較于BRT,RF機器學習模型預測的結果和實驗結果之間具有更高的一致性,從而驗證了RF模型的高通量虛擬篩選與預測能力。這體現了機器學習方法在解決有機光伏材料問題方面強大的能力,將大大加快高性能有機光伏材料及其供體/受體對的探索過程 

The traditional research lifecycle of organic photovoltaic (OPV) materials is tedious and laborious process which contains materials design and synthesis, device characterization and optimization and performance evaluation, hampering the development of organic photovoltaic. Recently, Prof. Min’s group collected data of 565 donor/acceptor (D/A) pairs with nonfullerene small molecule acceptors and polymer donors as training and testing set to construct five machine learning (ML) methods. These methods which were based on five common algorithms, linear regression (LR), multinomial logistic regression (MLR), boosted regression trees (BRT), artificial neural network (ANN) and random forest (RF), can be used to fast screening and evaluation of new promising materials and donor or acceptor counterparts. According to the predicted results of testing set, the researchers found that ML models based on RF and BRT algorithms performed well with high Pearson’s coefficient of over 0.7. What’s more, the RF and BRT models were used to screen 3.2 million D/A pairs automatically generated by the original dataset, among which six D/A pairs were selected, synthesized and characterized to further evaluate the applicability of these ML methods in OPV. The experiment results correlated better to the predicted results of RF methods compared to that of BRT methods, which indicated superiority of RF method in high throughput virtual screening of OPV materials. This work demonstrates machine learning as a powerful tool to solve problems in OSCs, which will accelerate the discovery of high-performance D/A combinations to a large extent.

Fundamental electronic structure and multiatomic bonding in 13 biocompatible high-entropy alloys (13种生物相容性高熵合金的基本电子结构和多原子键合)
Wai-Yim ChingSaro SanJamieson BrechtlRidwan SakidjaMiqin Zhang & Peter K. Liaw
npj Computational Materials 6:45(2020)
doi:s41524-020-0321-x
Published online:06 May 2020

Abstract| Full Text | PDF OPEN

摘要:高熵合金(HEA)由于其諸多獨特性能和潛在應用而備受關注。在此獨特的複雜多組分合金類別中,原子間相互作用的性質尚未得到充分認識或開發。本研究報告了一種理論建模技術,可以對其電子結構和原子間鍵合進行深入分析,並根據量子力學指標,即總鍵序密度(TBOD)和部分鍵序密度(PBOD),的使用來預測HEA性能。將該理論建模技術應用于13種生物相容性多組分HEA的研究,得到了許多新穎而有價值的結果,包括使用價電子數不足、對大晶格畸變進行量化、利用實驗數據驗證機械性能、對孔隙率進行建模以降低楊氏模量等。這項研究概述了應用HEA的路線圖作生物醫學用材料的合理設計方法 

Abstract:High-entropy alloys (HEAs) have attracted great attention due to their many unique properties and potential applications. The nature of interatomic interactions in this unique class of complex multicomponent alloys is not fully developed or understood. We report a theoretical modeling technique to enable in-depth analysis of their electronic structures and interatomic bonding, and predict HEA properties based on the use of the quantum mechanical metrics, the total bond order density (TBOD) and the partial bond order density (PBOD). Application to 13 biocompatible multicomponent HEAs yields many new and insightful results, including the inadequacy of using the valence electron count, quantification of large lattice distortion, validation of mechanical properties with experiment data, modeling porosity to reduce Young’s modulus. This work outlines a road map for the rational design of HEAs for biomedical applications.

Editorial Summary

Fundamental electronic structure and multiatomic bonding:biocompatible high-entropy alloys生物相容性高熵合金:基本電子結構和多原子鍵合

該研究通過使用先進的大型超胞建模方法研究了13種受生物啓發的HEA的電子结构、原子间键合和机械性能,得到了许多对开发和应用生物相容性高熵合金(HEA)至關重要的新認識。來自美國密蘇裏大學堪薩斯城分校物理與天文學系的陳慧妍領導的團隊,報道了他們針對HEA的形成理論及其潛在應用方面所面臨的挑戰,所作的有關電子結構、原子間鍵合以及總鍵序密度(TBOD)和部分鍵序密度(PBOD)的研究結果。他們指出,使用TBODPBOD作爲評估多組分合金基本性能的關鍵指標時,具有特別的優點:無論HEA的原子種類、組成或大小如何,都可以直接將它們進行相互比較。而且,該方法還可應用于其他材料系統,只需每對原子間的所有原子間鍵合,再通過單胞的體積作標准化即可。此特性與基于焓評估的方法中所使用的基態能有很大不同,後者在評估不同組成的多組分HEA性能時計算繁重且耗時 

The electronic structures, interatomic bonding, and mechanical properties of the 13 bioinspired HEAs are investigated through advanced modeling using large supercells yielding many new and insightful results critical to the development and application of biocompatible HEAs. A team led by Wai-Yim Ching from the Department of Physics and Astronomy, University of Missouri Kansas City, USA, presented the electronic structure, interatomic bonding, and the application of total bond order density (TBOD) and partial bond order density (PBOD) in addressing the challenges for fundamental understanding on the theory of formation of HEAs and its potential applications. They pointed out the special merits of using TBOD and PBOD as key metrics for assessing the fundamental properties of multicomponent alloys. They can be directly compared with each other irrespective of their atomic species, composition, or size. Moreover, they can be applied to other materials systems as long as all interatomic bonding between every pair of atoms are included and normalized by the volume of the cell. This characteristic is very different from other techniques based on ground state energies used in the enthalpy evaluation, which can be quite onerous and time consuming for multi-component HEAs with different compositions.

Fifth-degree elastic energy for predictive continuum stress-strain relations and elastic instabilities under large strain and complex loading in silicon (大变形任意载荷下可预测材料不同失稳条件的五阶连续介质模型)
Hao ChenNikolai A. Zarkevich, Valery I. Levitas, Duane D. Johnson & Xiancheng Zhang
npj Computational Materials 6:115(2020)
doi:s41524-020-00382-8
Published online:04 August 2020

Abstract| Full Text | PDF OPEN

摘要:材料在複雜載荷下會有大變形,並經常伴有彈性失穩的相變過程。這種過程在簡單體系和複雜體系內都被觀察到。這裏,基于對大量DFT計算結果的擬合,五階連續介質力學模型被發展來擬合任意載荷下材料失穩條件。該模型的柯西應力-拉格朗日應變曲線可以很好重現第一性原理計算結果。並且該模型准確的預測了任意載荷下材料的臨界失穩應力,包括多軸正應力和剪切應力下的失穩應力。這個模型將爲連續介質力學模擬材料在任意載荷下大變形失穩提供了理論基礎 

Abstract:Materials under complex loading develop large strains and often phase transformation via an elastic instability, as observed in both simple and complex systems. Here, we represent a material (exemplified for Si I) under large Lagrangian strains within a continuum description by a -order elastic energy found by minimizing error relative to density functional theory (DFT) results. The Cauchy stress-Lagrangian strain curves for arbitrary complex loadings are in excellent correspondence with DFT results, including the elastic instability driving the Si III phase transformation (PT) and the shear instabilities. PT conditions for Si I II under action of cubic axial stresses are linear in Cauchy stresses in agreement with DFT predictions. Such continuum elastic energy permits study of elastic instabilities and orientational dependence leading to different PTs, slip, twinning, or fracture, providing a fundamental basis for continuum physics simulations of crystal behavior under extreme loading.

Editorial Summary

Fifth-degree elastic energy for predictive continuum stress-strain relations and elastic instabilities under large strain and complex loading in silicon准確預測材料在任意載荷下失效的大變形彈性理論

任意载荷下,材料失效的临界应力具有很大的不同,例如静水压下,材料失稳的压强可以到100GPa,而在多方向剪切应力下材料失稳可能只需100-200MPa,可以达到3个量级差。然而目前针对材料失稳的连续介质模型大多基于能量或者最大剪切应力,并不能完全覆盖材料任意并不完善任意载荷。该研究基于连续介质理论提出了基于拉格朗日应变的五阶大变形模型,该模型能够准确获得硅在任意载荷下的材料失效应力。来自中国华东理工大学的陈浩讲师和其博士导师美国爱荷华州立大学航空航天工程和机械工程系的Valery I. Levitas教授团队,以及爱荷华州立大学材料学院的Duane D. Johnson教授团队合作,采用第一性原理计算得到了单晶硅材料在任意载荷下的失稳应力,拟合了提出的大变形弹性理论,发现该弹性理论可以精确给出硅材料任意载荷下的失稳应力。该研究为在连续介质框架下研究精确模拟材料在任意载荷下的失稳条件提供了理论基础,由于不同载荷可以导致不同的失稳模式,比如剪切应力下发生塑性变形,而在正应力下发生相变。因此该模型为连续介质力学提供了模拟任意载荷下导致不同失效模式的可能性 

Materials under complex loading develop large strains and often phase transformation via an elastic instability, as observed in both simple and complex systems. Here, we represent a material (exemplified for Si I) under large Lagrangian strains within a continuum description by a 5th-order elastic energy found by minimizing error relative to density functional theory (DFT) results. The Cauchy stress-Lagrangian strain curves for arbitrary complex loadings are in excellent correspondence with DFT results, including the elastic instability driving the Si I to Si II phase transformation and the shear instabilities. Phase transformation conditions for Si I to Si II under action of cubic axial stresses are linear in Cauchy stresses in agreement with DFT predictions. Such continuum elastic energy permits study of elastic instabilities and orientational dependence leading to different phase transformations, slip, twinning, or fracture, providing a fundamental basis for continuum physics simulations of crystal behavior under extreme loading.

EPIC STAR: a reliable and efficient approach for phonon- and impurity-limited charge transport calculations (EPIC STAR:一种可靠且高效的方法,用于声子和雜質限制的電荷输运计算)
Tianqi DengGang WuMichael B. SullivanZicong Marvin WongKedar HippalgaonkarJian-Sheng Wang & Shuo-Wang Yang
npj Computational Materials 6:46(2020)
doi:s41524-020-0316-7
Published online:7 May 2020

Abstract| Full Text | PDF OPEN

摘要:本研究提出了一種計算效率高的第一性原理方法,以預測半導體本征電荷輸運性質。利用短程電子-聲子散射的廣義Eliashberg函數和長程電子-聲子和電子-雜質散射的解析表達式,實現了不需要經驗參數即可快速可靠地預測載流子遷移率和電子熱電性能。該方法被命名爲“能量依賴性聲子-和雜質-限制的載流子散射近似(EPIC STAR)”方法。通過對幾種代表性半導體的實驗測量和其他理論方法的比較,驗證了該方法的有效性,得到了極性和非極性、各向同性和各向異性材料的定量一致性。該方法的效率和魯棒性有助于實現自動化預測和無監督預測,從而可對半導體材料進行高通量篩選和新材料發現,以作導電、熱電和其他電子學方面的應用 

Abstract:A computationally efficient first-principles approach to predict intrinsic semiconductor charge transport properties is proposed. By using a generalized Eliashberg function for short-range electron–phonon scattering and analytical expressions for long-range electron–phonon and electron–impurity scattering, fast and reliable prediction of carrier mobility and electronic thermoelectric properties is realized without empirical parameters. This method, which is christened “Energy-dependent Phonon- and Impurity-limited Carrier Scattering Time AppRoximation (EPIC STAR)” approach, is validated by comparing with experimental measurements and other theoretical approaches for several representative semiconductors, from which quantitative agreement for both polar and non-polar, isotropic and anisotropic materials is achieved. The efficiency and robustness of this approach facilitate automated and unsupervised predictions, allowing high-throughput screening and materials discovery of semiconductor materials for conducting, thermoelectric, and other electronic applications.

Editorial Summary

EPIC STAR: a reliable and efficient approach for phonon- and impurity-limited charge transport calculationsEPIC STAR:一种可靠且高效的方法,用于声子和雜質限制的電荷输运计算

  該研究提出了一種計算效率高的第一性原理方法,以預測半導體本征電荷輸運性質。來自新加坡科學技術研究局高性能計算研究所的Gang WuShuo-Wang Yang共同領導的團隊,通過引入廣義Eliashberg函數並加入光學聲子極化、雜質散射和自由載流子屏蔽等過程,经过密度泛函微扰理论的计算,使该方法在计算量很小的情况下,尤其对非极性和极性半导体都能实现高保真度。该研究论证了极性光学声子散射的重要性,这表明在研究极性半导体的電子性质时,在没有考虑极性光学声子散射时,需格外小心。通过与SiGaAsMg2SiNbFeSb的實驗和理論結果進行的比較,作者驗證了這一方法的可靠性,並揭示了NaInSe2是一種潛在的新型熱電材料。隨著近年來高通量DFPT計算發展,該方法提出的方法可廣泛應用于高遷移率半導體和高性能熱電光伏材料的高通量篩選。 

A swift and automation-friendly approach for intrinsic and impurity-limited charge transport property prediction from first-principles is proposed. A team co-led by Gang Wu and Shuo-Wang Yang from the Institute of High Performance Computing, Agency for Science, Technology and Research, Singapore, by introducing generalized Eliashberg function and adding polar optical phonon contribution, impurity scattering, and free carrier screening, enabled the new approach to achieve high fidelity especially for both non-polar and polar semiconductors with very small computational cost after calculations by density functional perturbation theory (DFPT). They demonstrated the importance of polar optical phonon scattering, which suggests that care should be taken when the electronic properties of polar semiconductors are studied without polar optical phonon scattering. The authors verified this approach by comparing with previous experimental and theoretical results for Si, GaAs, Mg2Si, and NbFeSb, and also revealed NaInSe2 as a potential new thermoelectric material. As high-throughput DFPT computations have been demonstrated recently, this methodology can be widely applied for reliable high-throughput screening of high mobility semiconductors and high-performance thermoelectric and photovoltaic materials.

Efficient training of ANN potentials by including atomic forces via Taylor expansion and application to water and a transition-metal oxide (通过包括通过泰勒展开产生的原子力并应用于水和过渡金属氧化物中,有效地训练ANN势)
April M. CooperJohannes KastnerAlexander Urban & Nongnuch Artrith
npj Computational Materials 6:54(2020)
doi:s41524-020-0323-8
Published online:13 May 2020

Abstract| Full Text | PDF OPEN

摘要:基于人工神經網絡(ANN)的經驗勢函數可以實現針對複雜材料的高精度(接近第一原理)大規模的原子模擬。對于分子動力學模擬,准確的能量和原子間作用力是先決條件,然而這需要同時基于電子結構計算得到的能量和力進行ANN訓練。本工作,我們基于總能量的泰勒展開,提出了一種同時基于能量和力信息來訓練ANN勢的有效替代方法。通過將力信息轉換爲近似能量,可以避免傳統力訓練方法中計算量隨原子數量二次方增長的關系,從而可以利用包含複雜原子結構的參考數據集進行訓練。以不同系統爲例,如水分子團簇、液態水和锂過渡金屬氧化物,我們證明了所提出的力訓練方法相對于僅依靠能量訓練的方案具有顯著提升。在訓練中包含力信息可減少構建ANN勢所需的參考數據集的大小,增加勢函數的可移植性,並整體提升力預測的精度。對于水團簇,與所有力分量的顯式訓練相比,泰勒展開方法可降低約50?%的誤差,而計算成本卻要低得多。因此,這樣的力訓練方法,簡化了用于模擬複雜材料能量和力的ANN勢的構造過程,正如本研究在水和過渡金屬氧化物中證明的情形那樣 

Abstract:Artificial neural network (ANN) potentials enable the efficient large-scale atomistic modeling of complex materials with near first-principles accuracy. For molecular dynamics simulations, accurate energies and interatomic forces are a prerequisite, but training ANN potentials simultaneously on energies and forces from electronic structure calculations is computationally demanding. Here, we introduce an efficient alternative method for the training of ANN potentials on energy and force information, based on an extrapolation of the total energy via a Taylor expansion. By translating the force information to approximate energies, the quadratic scaling with the number of atoms exhibited by conventional force-training methods can be avoided, which enables the training on reference datasets containing complex atomic structures. We demonstrate for different materials systems, clusters of water molecules, bulk liquid water, and a lithium transition-metal oxide that the proposed force-training approach provides substantial improvements over schemes that train on energies only. Including force information for training reduces the size of the reference datasets required for ANN potential construction, increases the transferability of the potential, and generally improves the force prediction accuracy. For a set of water clusters, the Taylor-expansion approach achieves around 50% of the force error improvement compared to the explicit training on all force components, at a much smaller computational cost. The alternative force-training approach thus simplifies the construction of general ANN potentials for the prediction of accurate energies and interatomic forces for diverse types of materials, as demonstrated here for water and a transition-metal oxide.

Editorial Summary

Transferring force into energy: accelerating construction of the machine learning potential變力爲能量:加速機器學習勢函數構建

  該研究提出一種基于原子間作用力信息來高效訓練高精度神經網絡經驗勢函數的方法。來自美國、德國和英國的聯合研究團隊,提出將力信息轉換爲近似能量,由此構建機器學習經驗勢的新方法。基于第一性原理集合機器學習,訓練經驗勢對于大規模材料模擬來說十分重要。准確的經驗勢需要同時擬合體系能量和原子間作用力。而作用力的作爲能量一階導數,擬合比較複雜,其計算量與原子數目成二次方關系。 

  作者將作用力轉化爲能量巧妙的繞開了上述問題,由此可以采用較大體系的數據集開展訓練。計算結果表明,與直接采用作用力訓練勢函數的方法相比,該方法不僅將勢函數的精度提升了50%,同時計算效率明顯提升。爲進一步驗證該方法的有效性,作者選取了三個具體的模型體系,即水分子團簇、液態水和複雜金屬氧化物開展勢函數訓練。他們發現,該方法可以顯著降低訓練所需的數據集大小;具有很好的可移植性,可以准確預測數據集之外的新體系;可以同時提升作用力預測的准確性。簡而言之,該方法簡化了神經網絡經驗勢的構造過程,有望推廣應用于任意類型的材料中。

An efficient training method for high-precision artificial neural network (ANN) empirical potential has been proposed based on the information of interatomic forces. 

The joint research team from the United States, Germany and the United Kingdom proposed this new method to by Taylor extrapolation of the total energy with inter-atomic forces. Empirical interatomic potential trained by machine learning based on first principles is important for large-scale material simulation. Accurate empirical potential requires simultaneous fitting of the total energy and interatomic forces. The force, as the first derivative of energy, is more complex to fit, as the computational cost is quadratic with the number of atoms. By translating the force information into energy, the authors bypass this problem. They found that compared with the conventional training method directly by force, the accuracy of potential can be improved by 50% by the newly proposed method with the calculation efficiency significantly improved. To further validify this method, three specific model systems, namely water molecular clusters, liquid water and complex metal oxides, are selected to train the potential. The results showed that this method can significantly reduce the size of the data set needed for training; it has good transferability as could accurately predict the new structure outside the data set; it can also improve the accuracy of force prediction. In short, this method simplifies the construction process of ANN potential, and is expected to be applied to any kind of materials.

版权所有 © 凤凰娱乐  滬ICP備05005480號-1
地址:上海市長甯區定西路1295號 郵政編碼:200050