Researchers are Creating a Brighter Future for Solar Cells — Without the Sun
Electrical and computer engineers at the University of California, Davis, are homing in on better solar cells and a brighter future for renewable energy by ditching the sun.
In new research published in ACS Applied Materials & Interfaces, Professors Jeremy Munday and Junshan Zhang have theoretically demonstrated a thermophotovoltaic system that can achieve a power conversion efficiency rate of 50%, more than double that of commercially available solar cells.
Thermophotovoltaic systems use photovoltaic cells, also known as solar cells, to create electricity through heat energy rather than sunlight.
The Heat Is On
Thermophotovoltaic systems turn heat made by an extremely hot emitter (operating around 2,500 degrees Fahrenheit or hotter) into electromagnetic radiation, or light. This light is then fed to a solar cell and becomes electricity.
Compared to traditional solar cell technology, thermophotovoltaic systems are unique for their tunability, meaning the energy source can be modified or changed. This also means they work independently of the sun and can operate 24/7 even in places with limited daylight, such as Alaska in the winter.
While promising, there’s still a lot to optimize about thermophotovoltaics, which is where Munday’s research comes in.
Emission Source
In a previous project funded by the Defense Advanced Research Projects Agency, Munday worked to advance thermophotovoltaic systems by focusing on the emitter.
He explored different emitter designs in collaboration with Professor of Materials Science and Engineering Marina Leite. They were looking for a material that could sustain high temperatures and create the right optical energy for the solar cells.
“We found some solutions that worked pretty well,” Munday said, “but then we started thinking, ‘Well, there's a big limitation here with having to heat materials to such high temperatures.’ That’s when I started thinking about using a filter instead.”
Let the Right Light In
The idea behind the filter solves multiple problems in a thermophotovoltaic system.
For one, it would only allow usable light to reach the solar cell. Solar cells are capable of efficiently producing electricity from a small band of the visible spectrum, and this filter would only let that part of the light reach the solar cell.
The filter would also make use of the unneeded light by reflecting it back to the emitter, making the system work less hard to maintain its emitter’s high temperature.
Additionally, from a design perspective, the filter provides engineers with much more flexibility. An emitter can only be made from a small pool of materials with high melting points, whereas a filter can be made from any material stable at room temperature.
“Now, by focusing on a filter, you have many more materials you can choose from,” Munday said. “That's good from the point of view of having a lot of options, but then it becomes bad from the point of view of trying to figure out what the best materials are. That's where machine learning comes in.”
Iterative Design
Munday connected with Professor of Electrical and Computer Engineering Junshan Zhang, an expert in edge intelligence and reinforcement learning, to expedite the design process with machine learning.
Following their discussion, Zhang’s graduate student Hang Wang teamed up with Munday’s graduate student Paulina Escobar, who had been working on the filter problem with Munday. Together, Wang and Escobar created a model that could learn from the input simulations of filters in a thermophotovoltaic system. They trained their model by grading the quality of its results, both in how much light was reflected by a filter and how much usable light made it to the solar cell.
“With the model,” Zhang said, “we can get feedback very quickly on many different material and design permutations through simulation, rather than plodding through real-world experiments.”
Hot on the Tracks
Thanks to the AI model, the team discovered a filter design that achieved a 50% energy conversion rate in the simulations, which far surpasses what is currently achievable.
Outside of the filter, a key feature of the design is that the photovoltaic cell is made of silicon, the most common material for traditional solar cells. Most thermophotovoltaic systems avoid silicon because the spectrum emitted from the hot emitter doesn’t match the spectrum that silicon can efficiently convert to electricity.
The next step of this research is to physically test it.
“What we [aim to demonstrate] with this design is that, because of this filter, we're able to filter the light’s spectrum so that it can still work well with silicon,” Munday said. “That was one of our goals: to use a very common material, like silicon, so it'd be potentially more impactful and easier to commercialize.”
The team hopes to refine the design further and explore other technologies where the filter might be useful.
