19 March 2026 | Ama Konadu Agyemang, Radboud University, Netherlands | Blog
“A scientist in his laboratory is not a mere technician: he is also a child confronting natural phenomena that impress him as though they were fairy tales.” – Marie Skłodowska Curie
Science often begins with curiosity
Long before equations, simulations, and sophisticated instruments, there is simply a question: why does nature behave the way it does?
Marie Curie captured this beautifully. A scientist, she suggested, is not just someone who performs experiments or analyzes data. At heart, a scientist is still a child; curious, fascinated, and constantly surprised by the hidden stories of the natural world. In modern materials science, those stories are written in atoms.
Every material carries information hidden deep within its atomic structure. This information tells us how its electrons move, how its atoms vibrate, how it responds to heat or magnetic fields. If we can decode those signals, we can discover entirely new technologies.
One of the most exciting applications of this is the search for ideal thermomagnetic materials. Materials that can convert waste heat into electricity.
Interestingly, a combination of powerful quantum mechanical tools like Density Functional Theory and machine learning models are helping to drive this search.
The next big energy resource is Waste Heat!
Every machine generates heat. Factories, computers, vehicles, and even the devices in our pockets release energy in the form of heat. Most of this energy simply escapes into the environment, unused.
But what if it could be captured?
Thermomagnetic materials offer a remarkable possibility. The magnetic properties of these materials change with temperature, especially near a critical point known as the Curie temperature. Around this temperature, even small thermal changes can produce significant magnetic responses. This behavior can be harnessed to convert low-grade waste heat into usable electrical energy.
The goal of the HEAT4ENERGY project is to develop thermomagnetic systems that can capture waste heat from industrial processes and convert it into electricity.
It almost sounds like something out of a fairy tale (one that would excite any physicist of course). Low-grade waste heat turning into electricity through magnetism? Yet the physics behind it is real. The difficulty lies in finding the right materials.
Most of the known materials have limitations, such as high cost, environmental concerns, unsuitable Curie temperatures and limited efficiency. Some of these materials exhibit strong magnetic responses but rely on rare or expensive elements. Others are more affordable but lack the necessary thermal or magnetic characteristics. No single material perfectly satisfies all the requirements needed for efficient thermomagnetic energy harvesting.
The challenge becomes even more daunting when we consider the sheer number of possible materials. There are millions of potential compounds. Exploring the enormous number of possible alloys and compounds experimentally would require enormous time and resources. In other words, testing every possible material in the lab is practically impossible.
And that is the challenge my work aims to address.
Who Am I?
I am Ama Konadu Agyemang, a doctoral candidate in theoretical and computational chemistry at Radboud University working on the development of renewable energy materials using first principles. As part of the HEAT4ENERGY project, my research focuses on discovering, designing and developing thermomagnetic materials that can convert low-grade waste heat into electricity. I use quantum mechanical tools like density functional theory to predict which materials might work best before they are synthesized in the laboratory.
Why Density Functional Theory (DFT)?
Quantum mechanics is the fundamental theory that governs the microscopic world. It describes how electrons move, how atoms bond, and how matter behaves at the smallest scales.
In 1998, the Nobel Prize in Chemistry was awarded to Walter Kohn and John Pople for their pioneering contributions to computational quantum chemistry. Walter Kohn was recognized for developing Density Functional Theory (DFT), a method that made it possible to calculate the electronic structure of complex materials far more efficiently than earlier quantum mechanical approaches. John Pople was awarded the prize for developing computational methods and software that made these quantum chemical calculations practical for real chemical systems. Together, their work transformed theoretical chemistry and enabled scientists to study realistic systems using computers rather than relying solely on experiments.
Since then, DFT has become one of the most widely used tools in materials science. It allows researchers to understand the fundamental properties of materials, such as magnetic moments, electronic structure, and thermodynamic stability.
One of the greatest advantages of DFT is that it allows scientists like me to explore materials before they are synthesized. Instead of creating one thermomagnetic material at a time in the laboratory, we can computationally evaluate thousands of materials quickly and at relatively low cost, while also reducing experimental risks and resource use and guiding experiments toward the most promising candidates.
However, this approach is not without limitations.
Standard DFT struggles to accurately calculate key properties such as exchange interactions and, ultimately, the Curie temperature. More advanced methods can improve accuracy, but they require significantly greater computational resources. In addition, standard DFT calculations are typically performed at absolute zero temperature, making it nontrivial to predict the behavior of real-world systems at finite temperatures.
This is where machine learning plays an important role
Machine learning does not replace physics. Instead, it learns patterns from data generated through physics-based methods such as DFT and from experimental measurements. By analyzing large datasets of known materials, machine learning models can identify relationships between composition, structure, and physical properties.
Once trained, these models can rapidly screen thousands or even millions of potential compounds, highlighting the most promising candidates for further investigation. In other words, while DFT provides accurate physical insights, machine learning helps navigate the enormous search space of possible materials much more efficiently.
Together with our consortium partners at the Danube University Krems, we are working on developing machine learning models that complement our DFT calculations and help accelerate the discovery of new thermomagnetic materials.
Conclusion
In many ways, the scientist Marie Curie described still exists today; curious, fascinated, and eager to uncover nature’s hidden stories. The difference is that today, our laboratories extend beyond test tubes and instruments into powerful computers that allow us to explore the behavior of matter at the atomic scale.
The combination of quantum mechanics, high-performance computing, and machine learning is transforming the discovery of thermomagnetic materials into something far more powerful.
And somewhere within those simulations may lie the next generation of thermomagnetic materials capable of turning low-grade waste heat into usable energy.
If you are interested in our research and want to see how we are paving the way for the next generation of energy, please follow our updates in this Blog, YouTube and LinkedIn.
Attribution: Originally published by HEAT4ENERGY. Reposted with permission. Original article: Blog #8: How Quantum Mechanics and Machine Learning are Accelerating Thermomagnetic Materials Discovery – HEAT4ENERGY
