Sustainable development activities that satisfy the needs of both current and future generations, while also achieving a balance between environment, economy, and society, are the foundation of various efforts in solving global environmental issues.

Toshiba Electronic Devices & Storage Corporation has taken the initiative in developing a variety of semiconductor products in the fields of discrete, logic, and analog semiconductors, which are crucial for achieving a sustainable society, and is making efforts to promote performance and functional improvements of individual products.

KAWAI Hirofumi / YAMAMOTO Hiroshi

Construction of fifth-generation (5G) base stations is expected to increase in line with the proliferation and expanded scope of application of 5G mobile communications for greater convenience in society. To meet the needs of this growing network, the conversion efficiency of power management systems, DC-DC converters, and power metal-oxide-semiconductor field-effect transistors (MOSFETs) must be improved to reduce base station power consumption.

Toshiba Electronic Devices & Storage Corporation offers silicon (Si) power MOSFETs that help improve the conversion efficiency of DC-DC converters. We have launched a new environmentally-friendly 100V power MOSFET fabricated with U-MOS11-H process, our latest eleventh-generation process with trench gate structures. It provides a 62% reduction in R_DS(ON)×Q_gd, an important performance index affecting power loss derived from multiplying drain-source on-resistance (R_DS(ON)) and gate-drain charge (Q_gd), compared with our conventional products fabricated with the tenth-generation U-MOSX-H process.

TANAKA Katsuhisa / KUSUMOTO Yuji / SUZUKI Takuma

Concerted efforts are being made to develop low-loss silicon carbide (SiC) power devices to help achieve a carbon-neutral society.

Toshiba Electronic Devices & Storage Corporation has fabricated a prototype 1 200 V-class SiC trench metal-oxide-semiconductor field-effect transistor (MOSFET) to reduce on-resistance through cell-pitch scaling. Experiments have confirmed the prototype’s performance including (1) specific on-resistance (R_onA) of 1.50 mΩ·cm² at a temperature of 25°C, which is 30% lower than that of conventional planar SiC MOSFETs, (2) breakdown voltage of more than 1 500V, (3) short-circuit withstand time of 2.5 µs, (4) application of the trench shape rounding process to reduce the gate leakage current (I_g) to less than 1/100 that of samples without the process, reduction of the cell pitch by 29% by suppressing increases in the trench opening width, and (5) suppression of threshold voltage (V_th) shifts and improving reliability via an optimized manufacturing process to control gate switching instability (GSI) phenomena.

LENG Ke / KUGIYAMA Yuta

With the introduction of renewable energy systems to combat global warming, use of photovoltaic (PV) systems has been expanding rapidly from residential use to mega solar installations. To enhance power generation efficiency and reduce the cost of PV systems, voltage is being increased from the conventional 1 500V to 2 000V. In line with this trend, there is a need for better insulation performance to ensure the safety of 2 000V PV systems.

Toshiba Electronic Devices & Storage Corporation has developed a gate driver coupler contributing to improved PV system insulation performance both by increasing the creepage distance from 8 mm, the longest in its existing products, to 15 mm, and by using resin material with a high comparative tracking index (CTI) of more than 600. This device complies with the safety requirements for 2 000V PV systems specified in the International Electrotechnical Commission (IEC) 62109-1 and IEC 62477 standards.

ITO Yuichi / TAJIMA Satoshi / MATSUMOTO Hiroaki

In line with recent trends of electrifying equipment to improve usability and work efficiency in the consumer, industrial, and automotive sectors, demand is growing for highly efficient motor control technology combining brushless DC (BLDC) motors with vector control to suppress power consumption increases.

To meet such market needs, Toshiba Electronic Devices & Storage Corporation has developed a motor control software development kit corresponding to vector control called Motor Studio for its microcontroller units (MCUs) and SmartMCD series gate driver integrated circuits (ICs) with embedded MCUs. To streamline motor control software development, we have developed the following new technologies for enhancing Motor Studio performance: (1) a control parameter tuning support function, (2) real-time control information waveform data display and logging functions, and (3) a current detection function during the low-speed rotation. Motor Studio incorporating these technologies is expected to offer optimal motor control design for our MCUs and SmartMCD series ICs.

MURASE Terutaka / Ajay KADOLKAR / HISAMOTO Koji

With edge computing becoming more important due to advancements in Internet of Things (IoT) devices, demand is increasing for compact, low power consumption, and low cost microcontroller units (MCUs), which are a key platform of IoT systems.

Toshiba Electronic Devices & Storage Corporation has built an artificial intelligence (AI) development environment focusing on the usability of AI on MCU, which is capable thanks to open-source software (OSS). We have also developed and introduced new compression technology based on the trainable tensor-train decomposition network (TTD-Net) to reduce both the size of AI models including a large number of weight coefficients and the computational load while maintaining inference accuracy. Our approach contributes to making practical use of real-time AI inference on edge devices.

TEH Chen Kong / BI Te / ITO Shuichi

There is a growing need for large voltage step-down DC-DC converters with 48 V input and 1 V output to reduce conduction losses in power supply system due to rapid rises in load currents at data centers. However, this poses issues such as an insufficient voltage step-down ratio and poor conversion efficiency using conventional Buck topology. Isolated topologies that use transformers can solve such issues, but require a larger mounting volume.

Toshiba Electronic Devices & Storage Corporation has developed a proprietary star-delta switching topology for DC-DC converters to achieve a large voltage step-down ratio and high conversion efficiency without using any transformers. The new topology reduces switching losses by expanding switching-on times fivefold compared with the Buck topology. We have also developed a bootstrap topology for downsizing integrated circuits (ICs) by lowering the capacitance of capacitors in ICs, thereby reducing the layout area by up to 61%. The mounting volume has been shrunk by approximately one tenth by replacing an external transformer with three flying capacitors. Experiments on a prototype IC using the new topology have confirmed basic performance, including a maximum conversion efficiency of up to 88% equivalent to that of conventional ICs using an external transformer and the industry's highest current density of 790 mA/mm².

HIRAKI Motohiro / TAKEDA Shun / BABA Emiru

With product development in the field of power electronics growing increasingly complex in recent years, more companies are introducing model-based development (MBD). As semiconductors tend to affect the performance of final products, it is necessary to provide customers with a simulation program with integrated circuit emphasis (SPICE) model which simulates the electrical properties of each device in response to MBD development environment early in the design process. To shorten product development periods and reduce the number of prototypes, demand continues to grow for rapid provision of SPICE models for circuit- and system-level verifications in parallel with device manufacturing planning.

Toshiba Electronic Devices & Storage Corporation has developed technology which generates Conceptual SPICE Model capable of predicting the electrical characteristics of each power semiconductor by combining electrical characteristics extracted using device simulation tools with conceptual design data prior to device manufacturing. Evaluations of a conventional SPICE model based on measured electrical characteristics of the existing low-voltage metal-oxide-semiconductor field-effect transistor (LVMOSFET) and our Conceptual SPICE Model of the same LVMOSFET have verified that the differences in drain-source voltage (V_ds), gate-source voltage (V_gs) and drain current (I_d) between two models are less than 15% in static characteristics and less than 20% in dynamic characteristics. Conceptual SPICE Model makes it possible to predict electrical characteristics with a sufficient degree of accuracy in the conceptual design phase.

Feature Articles

ISHIDA Tomomi / GODA Koji

Machine learning operations (MLOps) play a critical role in all processes from the development to the operation of artificial intelligence (AI) models. However, an increase in AI models tends to result in complicated update states and deployment destinations of individual models, leading to heavy operational workloads.

To rectify this situation, Toshiba Corporation has implemented a new AI model release management function that supports large-scale AI model operation in its MLOps platform. It helps shorten the time required for processes from relearning to release and reduces the workload for model releases by supporting a series of operational workflows including approval processes. User evaluation tests show that there is still room for optimization while confirming function effectiveness. We are working to further expand AI model release management function to a wide range of business activities by enhancing functionality and improving operability.

ENDO Kotaro / TOYAMA Haruhiko

With the progress of digital transformation (DX), demand is growing for blockchain technology that accelerates data linkage between companies. However, social implementation of blockchain systems is hindered by their complexity, the need for expertise in this field, and the difficulty of ensuring reliability when collaborating with multiple companies in operating such systems.

To overcome these problems, the Toshiba Group launched DNCWARE Blockchain+ (hereafter referred to as BC+), our proprietary blockchain for enterprise use, in 2022. We made further improvements to usability and released version 3.0 in 2025, which ensures Byzantine fault tolerance (BFT) for four blockchain nodes and provides better flexibility for network configurations. We are actively engaged in applying BC+ to a broad range of use cases, and several projects are already in full-scale operation.

NISHIO Naoki / SAITO Takasumi / NAGASAKA Masanori

With the goal of achieving carbon neutrality by 2050, electrification of small- and medium-sized vehicles continues to accelerate. With the electrification of rolling stock and marine vessels expected to contribute to reducing carbon dioxide (CO₂) emissions, there is demand for the development of batteries with high energy density, high power output, high durability, and a high level of safety.

Toshiba Corporation has responded by developing the SCiB™ lithium-ion battery, which meets all of these requirements, and providing it for a number of projects in various fields. We have developed new technology to achieve a balance between higher energy density and longer lifetime SCiB™ by adopting appropriate cathode materials, structures, and electrolytes. Experiments on a prototype battery cell have verified that it can retain 95% of its initial capacity after approximately 6 000 charge-discharge cycles at a high temperature of 45°C.

MUKAI Takero / HASHIDATE Tadayuki / NAKAZONO Masahiko

Integration of renewable energy systems to make power grids more flexible is garnering more attention than ever before. Hydroelectric power systems play a key role in adjusting electricity supply and demand as they enable quick startup and shutdown and can also quickly control output fluctuations. In recent years, demand for faster startup times and more frequent startup and shutdown operations has grown, however, such operations could significantly impact the lifetime of turbine runners, as the mechanisms behind fatigue damage remain unknown.

Toshiba Energy Systems & Solutions Corporation has succeeded in identifying the mechanisms behind turbine runner damage and established a method for evaluating turbine runner strength during the startup process. Simulations have confirmed that applying reinforcement learning to shorten startup time and suppress fatigue life consumption shortens the startup time by approximately 8% compared with on-site test data regardless of the same degree of fatigue life consumption. Our approach to developing turbine runners capable of high-speed startup while maintaining hydraulic performance is expected to contribute to enhancing the value of hydroelectric power in the supply and demand adjustment market.

Frontiers of Research & Development

Method Combining Additive Manufacturing with Topology Optimization for Design of Lightweight BWR Control Rod Velocity Limiters

*Company, product, and service names appearing in each paper include those that are trademarks or registered trademarks of their respective companies.

Vol. 81, No. 1, January 2026

Index