By Tom Whitfield

A research team from Carnegie Mellon University has reported a fourfold increase in nanoscale heat transfer using engineered gold structures, a result published in Nature that could reshape thermal management in semiconductors and energy devices.

The work, carried out with Stanford University and Purdue University, targets near-field radiative heat transfer. When two surfaces sit within a few hundred nanometers of each other, thermal energy tunnels across the gap as electromagnetic waves rather than radiating away. The team aimed to amplify that tunneling with metamaterials, materials patterned at scales smaller than the wavelengths they manipulate.

What did the team actually build?

The group patterned microscopic gold structures onto thin membranes and placed them face-to-face across a nanoscale gap. The setup raised heat transfer by as much as a factor of four compared with equivalent arrangements lacking the engineered patterns. The improvement far exceeds what classical heat-transfer theory predicts over comparable distances.

Sheng Shen, a professor of mechanical engineering at Carnegie Mellon and senior author of the study, described the approach in plain terms. "Unlike conventional materials, metamaterials are built with tiny, repeating patterns that interact with energy in precise ways," Shen said. "We patterned microscopic gold structures onto thin membranes and positioned them face-to-face across a nanoscale gap. This increased heat transfer by as much as four times compared to similar setups without metamaterials which is far beyond what traditional physics would predict at larger distances."

Why does gold patterning help?

The enhancement does not come from simply adding more physical pathways for heat. According to Zexiao Wang, a PhD student in Shen's group and a co-first author, the gold structures couple with naturally occurring energy waves in the material called surface phonon polaritons. The interaction creates a resonance that lets energy cross the gap more freely.

"It is a cooperative effect," Shen said. "The structures and the material amplify each other."

The mechanism mirrors what optical engineers use to boost light-matter interaction, here applied to infrared thermal wavelengths. Treating heat as something that can be engineered, rather than merely dissipated, is the conceptual shift the authors want to underline.

What could it mean for chips and energy systems?

Thermal bottlenecks now limit the performance of advanced semiconductors, particularly in stacked die and 3D packaging where heat has no constant path to a heat sink. A controlled, near-field thermal channel could in principle route heat from a hot spot to a remote cold surface without physical contact, easing the design constraints on power density.

The authors also point to thermophotovoltaics, which convert thermal radiation into electricity. A fourfold lift in radiative coupling could lift cell output at the same source temperature, although converting the laboratory result into a working module will require fabrication at wafer scale and durability testing under continuous operation.

Infrared sensing offers a third application. Stronger, more directional thermal signals would improve the sensitivity of detectors used in environmental monitoring and defense, both areas that funded the underlying research.

What are the limits today?

The experiments were conducted under tightly controlled laboratory conditions at the nanoscale. The authors stop short of claiming a deployable cooler or generator, and they note that manufacturing sub-100-nanometer gold features reliably across a 300 mm wafer has not yet been demonstrated.

Still, the paper moves a long-standing prediction from theory to measurement. The U.S. Defense Threat Reduction Agency, the National Science Foundation, and the Air Force Office of Scientific Research supported the work. Co-first authors Zexiao Wang, Renwen Yu, and Hakan Salihoglu contributed equally. The journal reference records the study as Nature, volume 654, issue 8117, page 64, DOI 10.1038/s41586-026-10595-4.

The takeaway: treating heat as a tunable engineering variable, rather than an unavoidable byproduct, is now backed by measured data at the scale where it matters most for next-generation chips and energy converters.