Rochester scientists increase solar technology performance 15-fold without modifying semiconductor material

Solar thermoelectric generators have promised cheap, versatile electricity for decades — yet they still convert less than 1% of sunlight into power, while a standard rooftop panel manages around 20%. The gap has stubbornly resisted a long parade of research efforts, nearly all of them aimed at the same target.

A team at the University of Rochester decided to look somewhere else entirely. Rather than following the field’s dominant focus, they left the core semiconductor materials completely untouched — and found a different lever to pull.

A stubborn efficiency ceiling — and a field stuck in one lane

Solar thermoelectric generators work on a beautifully simple principle. Heat one side of a semiconductor-based device, keep the other side cool, and the temperature difference drives an electric current — a phenomenon called the Seebeck effect. No moving parts, no exotic fuels, no complex chemistry. In theory, STEGs should be a natural fit for cheap, distributed power generation.

Those structures improve heat dissipation through both radiation and convection, effectively doubling the cooling performance of a standard aluminum heat dissipator.

In practice, they’ve been stuck. Current devices convert less than 1% of incoming sunlight into electricity — compare that to the roughly 20% efficiency of a standard residential solar panel, and the gap is hard to ignore.

The research community hasn’t ignored it. For decades, scientists have worked to close that gap almost exclusively by engineering better semiconductor materials. The gains have been real but modest, and the ceiling has barely moved. For practical energy applications — rural electrification, off-grid sensors, low-cost wearables — “modest” simply isn’t enough.

The unconventional fix: leave the semiconductors alone

Professor Chunlei Guo and his colleagues at the University of Rochester’s Institute of Optics looked at that ceiling and asked a different question: what if the semiconductors aren’t actually the bottleneck?

Their answer, published in Light: Science and Applications, reframes the entire engineering challenge. Instead of focusing on what sits between the hot and cold sides of the device, Guo’s team focused on the hot and cold sides themselves — improving how heat is absorbed on one end and shed on the other. This strategic shift separates the work from virtually everything that came before it.

“For decades, the research community has been focusing on improving the semiconductor materials used in STEGs and has made modest gains in overall efficiency,” Guo said. “In this study, we don’t even touch the semiconductor materials.”

That dual-sided thermal engineering approach produced a device that generates 15 times more power than previous state-of-the-art STEGs.

Black metal, mini greenhouses, and laser-etched heat sinks

Four distinct engineering techniques were applied, each targeting a specific weakness in how conventional STEGs handle heat.

On the hot side, the team used a “black metal” technology developed in Guo’s own lab. Using intense femtosecond laser pulses — extremely short bursts of light — they etched nanoscale structures into ordinary tungsten, transforming the metal’s surface so it selectively absorbs light at solar wavelengths while reducing heat loss at others. More sunlight captured, less energy wasted.

A piece of plastic placed over the black metal surface then traps heat by minimizing convection and conduction — the same logic that makes a glass greenhouse warm on a cold day. Guo describes it simply as a mini greenhouse. This keeps the hot side hotter, widening the temperature gap that drives the Seebeck effect.

On the cold side, the same femtosecond laser technique was turned on regular aluminum to create a heat sink covered in tiny surface structures. Those structures improve heat dissipation through both radiation and convection, effectively doubling the cooling performance of a standard aluminum heat dissipator. A colder cold side means a larger temperature differential — which means more electricity. Each technique addresses a different part of the thermal equation, and together they compound in ways no single improvement could have achieved alone.

From lab bench to LED — what 15x efficiency looks like in practice

The researchers didn’t stop at measuring efficiency gains on paper. They demonstrated their improved STEGs powering LEDs far more effectively than current STEG methods allow — a concrete proof-of-concept that the gains translate into real, usable electricity.

The potential applications extend well beyond light bulbs. Guo points to wireless sensors for the Internet of Things, wearable electronics, and off-grid renewable energy systems in rural areas as near-term targets. STEGs can also harvest any thermal energy source, not just sunlight, which broadens their potential role considerably. The research received support from the National Science Foundation, FuzeHub, and the Goergen Institute for Data Science and Artificial Intelligence.

What comes next for solar thermoelectric power

A 15-fold improvement is striking, but it doesn’t close the full gap with photovoltaic panels. STEGs still have ground to cover before they compete directly with conventional solar on efficiency alone.

What the Rochester work opens up is a new axis of progress. Thermal management alone produced this kind of jump — combining it with better semiconductors could push efficiency further still. The two approaches aren’t mutually exclusive; they’ve simply never been pursued together seriously.

There’s also a manufacturing argument worth watching. The techniques Guo’s team used rely on tungsten and aluminum, both common and widely available, which could make scaling up considerably easier than approaches dependent on exotic or expensive compounds.

For the billions of people living without reliable grid access, the more important question isn’t whether STEGs can match rooftop solar — it’s whether they can become cheap and robust enough to fill gaps that photovoltaics can’t easily reach. This research suggests that question deserves fresh attention.

The full version of the study is available here: Tianshu Xu, Ran Wei, Subhash C. Singh, Chunlei Guo. 15-Fold increase in solar thermoelectric generator performance through femtosecond-laser spectral engineering and thermal management. Light: Science, 2025; 14 (1) DOI: 10.1038/s41377-025-01916-9