1 Experimental data management platform for data-driven investigation of combinatorial alloy thin films Jaeho Song Haechan Jo and Dongwoo Lee

2025-04-27 0 0 1.59MB 19 页 10玖币

侵权投诉

Experimental data management platform for data-driven investigation of

combinatorial alloy thin films

Jaeho Song, Haechan Jo, and Dongwoo Lee*

School of Mechanical Engineering, Sungkyunkwan University, Suwon, Gyeonggi-do, South Korea

*Corresponding author, E‐mail address: dongwoolee@skku.edu

ABSTRACT

Experimental materials data are heterogeneous and include a variety of metadata for processing

and characterization conditions, making the implementation of data-driven approaches for developing novel

materials difficult. In this paper, we introduce the Thin-Film Alloy Database (TFADB), a materials data

management platform, designed for combinatorially investigated thin-film alloys through various

experimental tools. Using TFADB, researchers can readily upload, edit, and retrieve multidimensional

experimental alloy data, such as composition, thickness, X-ray Diffraction, electrical resistivity,

nanoindentation, and image data. Furthermore, composition-dependent properties from the database can

easily be managed in a format adequate to be preprocessed for machine learning analyses. High flexibility

of the software allows management of new types of materials data that can be potentially acquired from

new combinatorial experiments.

Keywords: research data management, combinatorial materials science, multi-component alloys, thin

films, machine learning

1. INTRODUCTION

Search for new alloys with an enhanced set of properties can be benefited from having diverse and

large property datasets and advanced analysis tools such as machine learning [1-12]. The advent of high-

throughput computation and experimentation has accelerated the exploration of composition spaces,

especially for multicomponent systems, such as metallic glasses [13-18], High-entropy alloys [19-24], and

magnetocaloric materials [25-27]. For these alloys, the unexplored composition region is immense,

necessitating the construction of large materials datasets. Computational materials databases based on

density function theory (DFT), such as AFLOW [10, 28], Materials Project [29], and OQMD [30] provide

materials data of millions of different crystals, such as structures, enthalpies, electronic structures, etc.

These databases have been successfully used for data-driven approaches to predict the material properties

of crystals [31-33]. When it comes to multi-component alloys, however, limited information can be

acquired from the simulation-based materials databases. This is mainly because many important properties

of the alloys, such as strength, and electrical resistivity, are strongly dependent on the structural defects,

whose features are not readily available in the DFT-based databases.

Combinatorial experiments [3, 34-40] allow facile acquisition of the property data of multi-

component alloys. Thin-film alloys with composition spreads can readily be fabricated through physical

vapor deposition (PVD) processes, such as evaporation [41, 42], electrodeposition [43, 44], and magnetron

sputtering [34, 45-48]. Scanning high-throughput measurements on the combinatorial films can effectively

produce composition-dependent materials data, such as microstructure, micrographs, as well as mechanical,

electrical, thermal, and optical properties [34, 39, 48-53]. Materials data from the combinatorial

experiments can be useful not only for designing advanced small-scale materials (e.g., interconnects and

metal MEMS materials, etc. [52]) but also for predicting the properties of bulk materials [53].

Experimental materials data, however, are rather heterogeneous, multifaceted, and contain a variety

of processing and characterization parameters (metadata), making the construction of the databases

challenging.[54] Nevertheless, previous work demonstrated that a wide variety of metadata and data

generated from combinatorial and high-throughput experiments can be successfully managed [55-59].

Soedarmadji et al. [55] demonstrated systematic management of millions of thin-film data, including

structural, optical, and electrochemical properties as well as metadata from various characterization

experiments. Banko, L. et al. [58] utilized a commercial document management system to manage the

metadata from PVD processes and diverse characterization data, such as chemical, optical, structural, and

magnetic properties.

The computational and experimental materials databases introduced above have achieved

important advances in research data management (RDM) to study various classes of materials, although

implementation of RDM and utilization of data-driven alloy investigation are still challenging for the

research groups with a lack of related experiences. Here, we introduce the Thin-Film Alloy Database

(TFADB), a RDM platform that is designed specifically for thin-film alloys investigated by combinatorial

experiments. The database manages the metadata and the property data obtained from high-throughput

experiments. TFADB is a ready-to-use platform that is designed to be easily implemented by the research

groups to construct their own alloy databases, quickly overcoming the obstacles to the employment of

customized RDMs. The database is also readily managed, edited, and re-configured through the user-

friendly GUI (graphical user interface).

The open version of TFADB (see data availability) can manage the composition, electrical

resistance, thickness, X-ray diffraction, nanoindentation, and sample image data (optical microscope,

scanning electron microscope, and transition electron microscope images) of combinatorially investigated

thin films. New types of data from other experimental tools can also be uploaded and managed through

TFADB. Furthermore, TFADB can output the experimental datasets in a machine-readable format,

facilitating the implementation of data-driven approaches to develop new alloys.

2. RESULTS AND DISCUSSION

Figure 1. Alloy development process using TFADB (a) The target composition range for a combinatorial study is determined

by the domain knowledge or the property prediction model that is based on previously deposited materials data available in TFADB.

(b) Combinatorial synthesis of the target composition is carried out. Deposition histories available in TFADB can be used to

determine the processing parameters of the new deposition. (c) Scanning property measurements (high-throughput experiments)

measure position-dependent properties. Since the composition of the combinatorial specimens varies with respect to the position,

position-dependent properties can be converted to composition-dependent properties. (d) The data are uploaded to TFADB. The

uploaded data can be managed, displayed, and downloaded by multiple users. (e) Systematic data management of TFADB allows

implementation of data-driven approaches for new alloy design.

Fig. 1 schematically illustrates the data-driven novel alloy design procedure that utilizes TFADB.

The alloy candidates with target compositions and properties for the synthesis are selected based on the

domain knowledge of the researcher or the property prediction model that was built using previously

deposited materials data in TFADB (Fig. 1a, e). Combinatorial thin film alloys with the target composition

range are then synthesized by reviewing the processing parameters (metadata) of similar alloy systems that

may be available in TFADB (Fig. 1b). Combinatorial magnetron sputtering without a substrate rotation is

one of the common choices to produce combinatorial thin films: films with different compositions at

different positions on a substrate [3, 34, 40, 45-48, 52, 60]. Typically, hundreds of alloys with different

compositions can be synthesized from a single deposition. Processing parameters of the PVD process, such

as Ar pressure, substrate temperature, sputter gun power, and gun angle affect the composition ranges and

properties of the combinatorial specimens. Therefore, the metadata, recipe of the synthesis process and the

information related to the thin film sample such as project name, date, and researcher, are systematically

managed in TFADB (Fig.1d).

The combinatorially synthesized films are then subjected to scanning property measurements and

position-dependent properties are acquired and uploaded to TFADB (Fig. 1c). Different kinds of materials

data can be acquired from conventional experiments that support the scanning measurement modes, such

as XRD (X-ray diffraction), SEM (scanning electron microscope), and nanoindentation. Custom-designed

combinatorial techniques can also be used for thermal [39, 61-64], electrical [3, 34, 40, 52, 60], and

magnetic [49, 65, 66] properties. Acquisition of the materials data from different sets of experiments and

tools results in heterogeneity in the data format. Therefore, the establishment of the standard format of the

characterization data is required as will be discussed later. The users of TFADB are encouraged to upload

the metadata and property data of combinatorial alloy films from unsuccessful experiments, such as poor

materials properties (e.g, low hardness) or the processing parameters of a problematic synthesis (e.g,

peeling-off of thin films). This will not only help future researchers find adequate processing parameters

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

1Experimentaldatamanagementplatformfordata-driveninvestigationofcombinatorialalloythinfilmsJaehoSong,HaechanJo,andDongwooLee*SchoolofMechanicalEngineering,SungkyunkwanUniversity,Suwon,Gyeonggi-do,SouthKorea*Correspondingauthor,E‐mailaddress:dongwoolee@skku.eduABSTRACTExperimentalmaterialsdataarehete...

展开>> 收起<<

1 Experimental data management platform for data-driven investigation of combinatorial alloy thin films Jaeho Song Haechan Jo and Dongwoo Lee.pdf

共19页,预览4页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

1 Experimental data management platform for data-driven investigation of combinatorial alloy thin films Jaeho Song Haechan Jo and Dongwoo Lee

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: