Thermoelectric generators play a crucial role in sustainable energy harvesting by converting heat gradients directly into electrical energy without moving parts, thus promising long-lasting, maintenance-free power sources. Despite immense potential, the optimization of TEGs involves complex interplay among material properties, geometric configurations, and operational conditions. Historically, these optimizations have relied heavily on numerical simulations, particularly finite element methods, which require significant computational resources and time to iterate over multiple design variations. This bottleneck has severely limited the pace at which high-performance, practical thermoelectric devices can be developed.

TEGNet fundamentally transforms this paradigm by integrating artificial intelligence into the design loop. Developed as a composable neural emulator, TEGNet leverages deep learning to predict key device performance metrics—including voltage outputs and heat flow dynamics—based on input parameters such as material characteristics and geometric dimensions. What sets TEGNet apart is its modular architecture: it consists of independently trained neural submodels optimized for specific materials, which can be combined like building blocks to simulate complex device assemblies. This composability is rooted in physical laws, ensuring that the neural network’s predictions remain physically consistent and reliable.

By inputting material data and design parameters into TEGNet, researchers can instantly estimate power generation capacity and conversion efficiency without the need for exhaustive finite element simulations. The substantial reduction in computational requirements accelerates the design process dramatically, enabling rapid exploration of a vast design space that was previously impractical due to time and resource constraints. This approach facilitates not only quick optimization but also the evaluation of novel device architectures encompassing heterogeneous material combinations, which are challenging to model accurately via traditional methods.

To validate TEGNet’s efficacy, the NIMS team focused on thermoelectric devices utilizing Mg-Sb (magnesium-antimony) based compounds, materials known for their promising thermoelectric properties at practical operating temperatures. The researchers applied TEGNet to optimize two distinct device configurations, subsequently fabricating and experimentally evaluating the prototypes. These devices exhibited outstanding performance, achieving conversion efficiencies of up to 9.3% and 8.7%, underscoring the model’s unparalleled ability to guide real-world device design and improve material-device synergy.

This advance arrives at a pivotal moment when the energy sector increasingly embraces AI and machine learning to tackle complex, multivariate optimization problems. While prior studies have predominantly concentrated on material-level optimization through AI, the NIMS effort uniquely targets device-level design. This systemic approach enables a holistic enhancement of thermoelectric systems, integrating material discoveries with architectural innovation. Consequently, TEGNet extends the frontier of what AI can achieve in clean energy technologies, paving the way for smarter, more efficient energy conversion systems.

The implications of this technology extend far beyond thermoelectricity alone. By validating a method to create high-fidelity, physics-informed AI emulators that can be recombined modularly, the research opens avenues for accelerated design in numerous domains involving complex multiphysics simulations. Fields such as battery development, photovoltaics, and other energy harvesting or conversion devices stand to benefit from similar AI-driven frameworks that circumvent computational bottlenecks.

TEGNet’s development was spearheaded by Professor Takao Mori and his team at NIMS’s Thermal Energy Materials Group, under the broader initiative funded by the Japan Science and Technology Agency’s Mirai Program. The initiative focuses on creating innovative thermoelectric conversion technologies suitable for stand-alone power supplies, especially tailored for sensor applications in the rapidly growing Internet of Things ecosystem. By delivering autonomous, maintenance-free power, these technologies have the potential to drastically reduce environmental impact and operational costs associated with sensor networks worldwide.

In addition to accelerating design cycles, TEGNet offers unprecedented flexibility in device engineering. Designers can simulate various scenarios, critically evaluate trade-offs, and iterate designs with significantly increased confidence and efficiency. This capability is vital in pushing thermoelectric devices from laboratory prototypes to scalable, real-world applications, facilitating a faster transition to commercial viability.

Looking forward, the team envisages that this AI-driven approach will serve as a foundation for the next generation of energy device design strategies. By combining AI-enhanced material science with device-level optimization, future research can achieve unprecedented performance benchmarks, unlocking new capabilities such as enhanced waste heat recovery from industrial processes or self-powered smart infrastructure. The integration of TEGNet-like neural emulators into industrial design pipelines will significantly shorten R&D timelines and reduce costs, accelerating the deployment of sustainable technologies at scale.

The publication of these findings in Nature marks a milestone for the thermoelectric research community and highlights the transformative potential of AI in sustainable energy solutions. As energy demands grow globally, and environmental concerns intensify, innovative solutions like TEGNet underscore the vital role of interdisciplinary research bringing together materials science, artificial intelligence, and device engineering.

In summary, TEGNet’s advent signals a paradigm shift that could redefine how thermoelectric generators are designed and optimized. By merging fast, accurate, physics-informed neural networks with modular composability, researchers and engineers now possess a powerful tool to navigate complex design landscapes at unprecedented speeds. This breakthrough not only boosts the prospects of thermoelectric technology but also paves the way for AI-driven innovation across the broader energy sector, promising a more sustainable and energy-efficient future.

Subject of Research: Not applicable

Article Title: Composable neural emulators accelerate thermoelectric generator design

News Publication Date: 15-Apr-2026

Web References: DOI link

References: Published in Nature at 11:00 U.S. Eastern Standard Time, April 15, 2026 (0:00 Japan Standard Time, April 16, 2026).

Image Credits: Takao Mori, National Institute for Materials Science

Keywords

Thermoelectric generators, Artificial intelligence, Neural networks, Device optimization, Mg-Sb materials, Power generation, Conversion efficiency, Sustainable energy, Waste heat recovery, Composable models, Computational acceleration, IoT sensors

Thermoelectric materials convert temperature differences directly into electrical voltage and vice versa, offering a pathway to recover waste heat and realize energy conversion with no moving parts. The performance of these materials is quantified by the dimensionless figure of merit, ZT, which depends intricately on the electrical conductivity, Seebeck coefficient, and thermal conductivity of the material. Magnesium-based thermoelectrics have long been regarded for their low toxicity and sustainable supply chains. However, enhancing their ZT values to reach practical levels necessitates a deep understanding of the intertwined physical phenomena governing their thermoelectric responses.

The recent breakthrough from the research team revolves around a comprehensive workflow that combines high-throughput density functional theory (DFT) calculations with cutting-edge machine learning models to systematically screen and predict candidate materials. At the core of their analysis lies an important but often overlooked factor: thermal expansion. This phenomenon, where crystal lattices undergo volumetric expansion upon heating, fundamentally alters the atomic spacing and lattice dynamics within materials. By carefully quantifying how thermal expansion influences lattice anharmonicity and electronic band structures, the team revealed a critical mechanism that boosts thermoelectric performance in magnesium-based compounds.

As materials heat up, their atoms vibrate more intensely, increasing the lattice anharmonicity—a measure of deviation from perfectly harmonic atomic vibrations. Enhanced anharmonicity can scatter phonons more effectively, suppressing lattice thermal conductivity, which is beneficial for thermoelectric performance as it minimizes parasitic heat conduction. In tandem, thermal expansion changes the electronic band structures by concentrating bandwidth and increasing the effective mass of charge carriers. This manifests as an augmentation of the Seebeck coefficient, which relates directly to the voltage generated from a given temperature gradient. The synergy of these thermal expansion-driven effects propels the ZT parameter upward, illuminating new paths for materials optimization.

The researchers embarked on an extensive data-driven journey by selecting magnesium-containing crystal structures from the Open Quantum Materials Database (OQMD), a vast repository of computationally evaluated materials properties. Their selection criteria prioritized thermodynamic stability and structural feasibility under realistic temperature and pressure conditions. Subsequently, they utilized density functional theory to calculate key material properties across hundreds of potential candidates, generating a robust dataset that encapsulates the intricate links between composition, crystal structure, and thermoelectric parameters.

Recognizing the challenges of exploring this multidimensional dataset manually, the team implemented an array of machine learning algorithms, including Light Gradient Boosting Machine (LGB) and Extreme Gradient Boosting (XGB). After rigorous model training and validation, XGBoost emerged as the superior predictive model, demonstrating remarkable accuracy and computational efficiency. This enabled rapid screening of thousands of hypothetical magnesium-based compounds, significantly narrowing the search for optimal thermoelectric materials without resorting to costly experimental trial-and-error.

The integration of DFT-driven data generation with XGBoost-powered prediction constitutes a paradigm shift in materials science research. It allows for the fine-grained quantification of complex physical phenomena and accelerates the identification of compositions exhibiting desired thermal and electronic characteristics. Additionally, this methodological framework provides a transparent window into the structure-property relationships governing thermoelectric behavior, offering researchers actionable insights for materials design.

Notably, this study elucidates the broader physics underpinning thermal expansion’s influence in low-dimensional systems. The enhancement of lattice anharmonicity and modulation of electronic density of states hold implications that transcend magnesium-based thermoelectrics alone. As the demand for high-performance thermoelectric devices mounts across sectors such as automotive waste heat recovery, aerospace, and microelectronics cooling, such fundamental insights pave the way for tailored material strategies spanning a wide chemical space.

Beyond its immediate scientific contributions, the published research embodies a successful demonstration of interdisciplinary synergy—uniting computational physics, materials informatics, and machine learning in an elegant, scalable workflow. The findings empower researchers worldwide to embrace data-centric methodologies while preserving physical interpretability. Furthermore, the accessibility of databases like OQMD combined with open-source machine learning tools democratizes advanced materials discovery, accelerating innovation in sustainable technologies.

In sum, this research offers a landmark advancement toward the rational design of next-generation magnesium-based thermoelectric materials. By demystifying the role of thermal expansion, quantifying its effects on key thermoelectric parameters, and harnessing state-of-the-art computational intelligence techniques, the study sets a new standard for high-throughput materials screening. As the global community seeks cleaner energy solutions and smarter thermal management, such impactful scientific advancements could resonate across industry and academia alike, catalyzing the transition to efficient, eco-friendly thermoelectric devices.

Published in the prestigious journal Science Bulletin, this study not only deepens fundamental understanding but also supplies a powerful computational toolkit for future explorations. The approach outlined has far-reaching potential—not merely as a blueprint for magnesium-based systems but as a universal scheme applicable across varied thermoelectric material families. It marks an exciting juncture where traditional materials science converges with modern data science, heralding a new era of predictive, accelerated innovation.

Subject of Research: Magnesium-based thermoelectric materials and thermal expansion effects on thermoelectric performance.

Article Title: Not specified.

News Publication Date: Not specified.

Web References: http://dx.doi.org/10.1016/j.scib.2025.07.041

References: Published in Science Bulletin, DOI: 10.1016/j.scib.2025.07.041

Image Credits: ©Science China Press

Keywords

Magnesium-based thermoelectrics, thermal expansion, machine learning, high-throughput screening, density functional theory, XGBoost, lattice anharmonicity, Seebeck coefficient, thermal conductivity, materials informatics, sustainable energy, computational materials science