Showa Denko (SDK) (TOKYO: 4004) proposed its “Project to Develop SiC Wafers Technology for Next-generation Green Power Semiconductors*1” (hereinafter “the Project”) to New Energy and Industrial Technology Development Organization (NEDO) as a candidate for “Projects to Develop Wafers Technology for Next-generation Power Semiconductors” which was set as a research and development target of “Next-generation Digital Infrastructure Construction*2” in “Green Innovation Fund Projects” (hereinafter GI Fund Projects). And the Project has been selected for GI Fund Projects by NEDO.
In October 2020, the Japanese Government declared that it aims to achieve carbon neutrality by 2050. Aiming to significantly accelerate efforts toward structural changes in the energy and industrial sectors and undertake bold investment for innovation, which are necessary for achievement of the above national target, Ministry of Energy, Trade and Industry (METI) decided in October 2020 to develop a Green Innovation Fund at the level of 2 trillion yen as part of the NEDO. The plan of GI Fund Projects, which are based on the specific goals shared by public and private sectors, is to continuously support companies and other organizations, which show their commitment to challenge such ambitious goals as their business issues ranging from research and development (R&D) to demonstrations to social implementation of the outcomes.
SDK’s business to manufacture SiC epitaxial wafers for power semiconductors (hereinafter SiC epi-wafers) has conducted business transactions with leading power semiconductor manufacturers inside and outside Japan, and has the global-top share in the SiC epi-wafers market*3. In the Project, SDK plans to make the most of its resources including intelligent property portfolio and development know-how, develop SiC epi-wafers with a diameter of 8 inches, and reduce density of deficiencies by one digit or more, thereby reducing production cost of next-generation power semiconductors. This time, NEDO highly appreciated SDK’s aggressive proposal and its SiC epi-wafers business’s good performance. As a result, NEDO selected SDK’s proposal for GI Fund Projects. The Project’s implementation period will be 9 years, from fiscal 2022 to fiscal 2030. In the Project, SDK will develop technology to accelerate growth rate of SiC bulk single crystal in cooperation with the National Institute of Advanced Industrial Science and Technology (AIST).
The Showa Denko Group aims to be a “Co-Creative Chemical Company” and contribute to the sustainable development of global society. Under this vision, SDK positions its operation to produce SiC epitaxial wafers, which contributes to efficient use of energy, as a next-generation business, and will allocate much of our business resources. The Group will continue contributing to the spread of SiC power semiconductors by maintaining “Best in Class” as its motto and continuing provision of high-performance and highly-reliable products.
*1. Next-generation green power semiconductors are power semiconductors used in xEVs, power equipment for renewable energies, power modules for servers, etc., made from next-generation materials including SiC.
*2. URL for Projects to Next-generation Digital Infrastructure Construction is https://green-innovation.nedo.go.jp/en/project/building-next-generation-digital-infrastructure/
*3. SDK has the global-top share in the market as an independent supplier of SiC epitaxial wafers.
This news originally published on www.sdk.co.jp
Showa Denko K.K. (SDK) (TOKYO: 4004) (President and CEO: Hidehito Takahashi) has introduced MLOps* (Machine Learning Operations) for efficient management of machine learning models deployed into Artificial Intelligence (AI) systems for materials design ahead of its competitors. Machine learning models can predict material properties based on formulations and manufacturing-process conditions of materials. This time, we automated input of the latest data into computers that develop machine learning models and data processing in those computers. This automation has reduced the time required to build and operate machine learning models from five days to one day per month. In addition, the introduction of MLOps enabled us to accelerate materials development by predicting material properties based on the latest data.
SDK utilizes AI systems for efficient materials development, such as exploring the optimal material formulation. Machine learning models deployed into the AI systems predict material properties from formulations or suggest formulations that improve material properties. The machine learning process for managing the AI systems includes inputting the latest data, data processing, and continuous training of machine learning models. Previously, data scientists had to input and process the latest data for themselves. These steps accounted for about 80% of the time required for the entire machine learning process. In addition, machine learning models deployed into the AI systems are built specifically for each material. Therefore, before introducing MLOps, the development of machine learning models required a lot of time and effort due to the necessary work specialized for each material.
Aiming to address these issues caused by applying AI systems to the development of numerous materials in the Company and operating machine learning models efficiently, we have installed programs to automate the input of the latest data and data processing into our AI systems. Moreover, we have introduced technologies that enable data scientists responsible for building machine learning models and software engineers responsible for building AI systems to develop systems collaboratively even if there are differences in operating systems and programming languages they use. By introducing MLOps ahead of our competitors to manage machine learning models efficiently, we can reduce the time required to develop machine learning models and their operation, improve prediction accuracy, and stably operate dozens of AI systems. As a result, now we can propose ideal materials to our customers promptly.
The Showa Denko Group will apply the fruits of basic research in AI and computational science to materials development and quickly provide solutions that solve our customers’ problems, thereby contributing to the development of a sustainable society.
Machine learning process from model development to operation
* MLOps: The method and philosophy for integrating the development and operation of machine learning models. MLOps include continuous training of machine learning models, automating the machine learning process, and establishing tools and operational rules for collaborative development between data scientists and software engineers.
This news originally published on www.sdk.co.jp
Showa Denko K.K. (SDK) (TOKYO: 4004) has developed hard disk (HD) media for hard disk drives (HDDs) which support data recording with Microwave Assisted Switching-Microwave Assisted Magnetic Recording (MAS-MAMR) technology, which is a next-generation data-recording technology based on a new data recording principle suggested by Toshiba Corporate Research & Development Center and Toshiba Electronic Devices & Storage Corporation (Hereinafter collectively called Toshiba).
MAS-MAMR is a next generation data recording method that can realize further increase in data-storage capacity of HDDs. At present, Microwave Assisted Magnetic Recording (MAMR) is a leading edge data-recording technology, which has already been put into practical use. The newly developed MAS-MAMR technology realizes data-recording track on the surface of HD media drastically narrower than that of MAMR-technology-based HD media through utilization of the strong magnetic oscillation effect of Microwave Assisted Switching (MAS effect)*1, thereby increasing data-storage capacity of HDDs.
Aiming to put this new data-recording technology into practical use, SDK has been developing HD media supporting MAS-MAMR in collaboration with Toshiba and TDK Corporation (TDK) which is a manufacturer of read/write heads for HDDs. In this joint development program, SDK, Toshiba, and TDK have cooperatively proved for the first time in the world that HDD as a combination of read/write head equipped with dual spin-injection-layer, which has been developed by TDK, and HD media equipped with new-type magnetic layer, which has been developed by SDK, can substantially increase HDD’s data-storage capacity through manifestations of the MAS effect.
In this year, SDK has already started supplying Toshiba Electronic Devices & Storage Corporation with HD media supporting MAMR. These media are mainly installed in 18TB HDDs for near-line use in data centers. On the basis of the fruit of the technology development program mentioned above, and aiming to realize large-capacity near-line HDDs with storage capacity of more than 30TB, SDK will accelerate development of HD media supporting MAS-MAMR which Toshiba aims to put to practical use as the second generation MAMR.
The amount of data generated and communicated has been increasing rapidly due to progress in Digital Transformation (DX) including the spread of teleworking, 5th generation (5G) mobile communication services, and Internet of Things (IoT). As a result, it has become a more important task for HDD manufacturers to develop large-capacity near-line HDDs for use in data centers, which record and store a large amount of data. In order to respond to the demand for the increase in storage capacities of HDDs, SDK will accelerate two-way development of HD media supporting MAS-MAMR and HAMR (heat assisted magnetic recording) in accordance with its motto of “Best in Class,” thereby developing the best HD media in the world.
*1: MAS effect: MAS effect is an abbreviation of Microwave Assisted Switching effect. MAS effect is an effect of strong magnetic oscillation between Spin Torque Oscillator (STO) and magnetic recording media. This strong magnetic oscillation enables HDD manufacturer to record digital data on the surface of HD media with recording track narrower than those of HDDs equipped with conventional magnetic recording technologies.
This news originally published on www.sdk.co.jp
Showa Denko K.K. (SDK) (TOKYO: 4004) has developed neural network models*1 to predict mechanical properties of 2000-series aluminum alloys*2 from their design conditions with high accuracy in collaboration with National Institute for Material Science (NIMS) and The University of Tokyo (UTokyo). The developed models enable us to accelerate the process to explore optimal compositions and heat-treatment conditions for aluminum alloys that can maintain strength at high temperatures and shorten development time for aluminum alloys to about half to one-third of that with conventional development method, which was not easy in the past.
Aluminum has various applications because it is lighter than iron and easy to work. However, it is usually used as an aluminum alloy containing copper, magnesium and other additive elements because pure aluminum has low strength. The development of aluminum alloys that can maintain sufficient strength for a particular use at high temperatures is desired because conventional aluminum alloys lose strength when their temperature rises to 100℃ or higher. However, the mechanical properties of aluminum alloys depend on many process factors, including many kinds of additive elements and heat-treatment conditions. Developing high-performance aluminum alloys usually takes time because designing aluminum alloys requires developers’ knowledge-rich experience and repetition of analysis and evaluation.
Aiming to solve these problems, SDK has been taking part in a project under Council for Science, Technology and Innovation (CSTI), Cross-ministerial Strategic Innovation Promotion Program (SIP), “Materials Integration” for Revolutionary Design System of Structural Materials. In this Development, SDK, NIMS, and UTokyo have collaboratively developed a computer system using neural networks, an artificial intelligence (AI) algorithm, to accelerate the development of materials and explore globally for aluminum-alloy designing conditions that realize optimal mechanical properties.
In this development, we focused on 2000-series aluminum alloys, utilized design data of 410 records of aluminum alloys listed in public databases, including the Japan Aluminum Association, and developed neural network models that accurately predict the strength of aluminum alloys at various temperatures ranging from room temperature to high temperature. In addition, we optimized the architecture and parameters of the neural network with Bayesian inference*3 by applying the replica-exchange Monte Carlo Method*4. As a result, it became possible for us to evaluate aluminum alloy strength and its prediction uncertainty. Moreover, this neural network model can estimate the strengths of aluminum alloys under 10,000 different conditions within 2 seconds. Thus it became possible to evaluate aluminum alloys with various design factors comprehensively in a short time.
Furthermore, we successfully developed “an inverse design tool,” which suggests a set of aluminum-alloy design conditions that maximizes the probability of satisfying the desired strength at arbitrary temperature. Thus it enables us to design high-strength aluminum alloys at high temperatures above 200℃.
In its “Long-term Vision for Newly Integrated Company,” the Showa Denko Group has announced that it will continue committing itself to make the most of artificial intelligence and computational science, which is the core of its fundamental research activities. We will accelerate our material development programs by applying the results of this Development to our activities to develop various new materials, and provide our customers with solutions for their problems, thereby contributing to the prosperity of society.
Detail of the results of this Development will be presented at the virtual session of the 2021 Materials Research Society*5 Fall Meeting, which will be held from December 6 to 8 in the United States and broadcasted to the world via the Internet.
*1. Neural network model: Neural network model is a machine learning algorithm that imitates the human brain’s neural network. A typical neural network model has input, hidden, and output layers. The existence of a hidden layer enables a neural network model to learn and estimate relationships between the input and output of complicated events. Statistical machine learning with neural network models with many hidden layers is called “deep learning.”
*2. 2000-series aluminum alloys: 2000-series aluminum alloys contain copper and magnesium as additive elements, and have high mechanical strength. Duralumin and super duralumin are well-known 2000-series aluminum alloys. 2000-series aluminum alloys are used as materials for the bodies of aircrafts and industrial parts (screws, gears and rivets, etc.).
*3. Bayesian inference: Bayesian inference is a method of statistical inference that statistically infers causes from observed facts based on Bayes’ theorem. In this Development, we constructed neural network models that reproduce correlation between design conditions and mechanical properties of aluminum alloys with Bayesian inference.
*4. replica-exchange Monte Carlo Method: Replica-exchange Monte Carlo Method is one of the computational methods to simulate Bayesian inference with a computer. It is known to converge to a global minimum solution faster than other methods when solving multimodal problems with many local minimum solutions. This enables us to explore a wide range of parameter space efficiently to find the optimal solution.
*5. Materials Research Society: Materials Research Society is an academic society focusing on material science, established in the United Stated in 1973. It convenes general meetings twice a year, in spring and fall.
This news originally published on www.sdk.co.jp