Showa Denko Materials Develops Its Own MI Platform for Functional Materials to Offer Speedy Materials Recommendations

December 07, 2021

Showa Denko Materials Co., Ltd.

Showa Denko Materials Co., Ltd. (President and CEO: Hisashi Maruyama) announces the development of a materials informatics (MI)^※1 platform equipped with a database infrastructure integrating a series of experimental data, including material formulations, sample preparation information, and evaluated properties, along with its original artificial intelligence (AI) engine.^{※2 ※3} Showa Denko Materials introduced its own original MI platform ahead of its competitors for developing functional materials that simultaneously satisfy the requirements for multiple properties. This platform enables on-site developers to efficiently develop functional materials, thereby providing customers with speedy product proposal as well as new values and services.

In the past, skilled developers created functional materials based on their knowledge, experience, and abilities through the repeated process of exploring an enormous number of combinations of composition ratios and raw materials. Now, MI technologies based on information science are said to help greatly reduce the number of repeated process, development time, and cost.

Showa Denko Materials has collaborated with QUATRE-i SCIENCE Co. Ltd., Hitachi, Ltd., and Hitachi Document Solutions Co., Ltd.^※4 to build a materials database infrastructure capable of consolidating experimental data from various research segments in a reusable way and develop a unique platform equipped with a graphical user interface (GUI)^※5 that can be used without information technology knowledge.
In parallel with MI platform development, Showa Denko Materials has stepped up efforts since 2017 to train materials developers able to promote data utilization for more effective use of MI technologies,^※6 for example, by requiring new employees to learn statistical experimental methods. In 2020, Showa Denko Materials established a new system to certify workers who understand statistical methods and can provide guidance on MI utilization as in-house data scientists, with about 20 highly skilled workers trained annually.

Furthermore, ahead of its competitors, Showa Denko Materials has recently created and incorporated its original AI engine designed for functional materials development into its platform, and successfully built an infrastructure allowing on-site developers to utilize MI technologies for a wide range of development themes. During AI performance tests in experiments for developing transparent adhesive films that meet the property requirements for film transparency and adhesion to copper foil, AI was able to meet the target with about 30 percent fewer experiments than skilled developers who can roughly predict properties from molecular structures. Moreover, in a test with the target values for film transparency and adhesion to copper foil raised to a higher level, AI outperformed skilled developers by achieving a target yet to be reached by the developers.^※3

Showa Denko Materials’ MI platform helps even inexperienced developers create optimal designs on par with experts and speedily offer customers materials proposals, which is further promoted by in-house data scientists.

Showa Denko Materials will develop its MI platform into a more sophisticated functional materials development infrastructure as a common platform for the newly integrated company of Showa Denko and Showa Denko Materials. In addition to developing more advanced AI technologies,^※3 Showa Denko Materials aims to create a platform that can help prevent rework (redesign) by linking the platform to the databases of other divisions, including sales, manufacturing, and quality assurance, and analyzing data in the materials design stage to identify defects during mass production, quality deficiencies, and procurement risks, before determining the materials design conditions.

Showa Denko Materials will take advantage of these pioneering MI technologies to develop materials more efficiently and offer customers speedy product proposals as well as new values and services, thereby contributing to society.

The outcomes of the above developments are presented at the 2021 MRS (Materials Research Society) Fall Meeting & Exhibit,^※7 which takes place in the United States on December 6 through 8.

The logo combines the images of chemistry and crystal – “C” being the first letter of each word – representing the accumulation of data and knowledge, like a growing crystal. The logo will be displayed on the MI platform screen and also be used for external briefings.

 

• ※1 Materials informatics (MI) refers to the initiative to develop materials more efficiently through the use of informatics (information science) methods. MI has been applied more broadly in the field of materials, with advanced information processing technology allowing MI to predict materials properties and recommend new materials based on experimental and simulation results.
• ※2 For details, please refer to the following paper. Hanaoka, K., “Bayesian Optimization for Goal-Oriented Multi-Objective Inverse Material Design,” iScience, 2021, 102781 (2021).
• ※3 Be presented at the 2021 MRS (Materials Research Society) Fall Meeting & Exhibit (December 6-8, 2021).
• ※4 QUATRE-i SCIENCE Co. Ltd. (https://www.i4s.co.jp/) is responsible for system development, while Hitachi, Ltd. and Hitachi Document Solutions Co., Ltd. are in charge of GUI design based on user experience.
• ※5 The graphical user interface (GUI) allows users to operate computers through icons, buttons, and other elements displayed on the screen.
• ※6 To be presented at Discovery Summit Japan (December 16-17, 2021).
• ※7 Founded in 1973, the Materials Research Society is an academic organization for materials science in the United States. MRS Meetings are held twice a year, in spring and fall.

 

• To Japanese