Researchers at Carnegie Mellon University, working with collaborators at Stanford University and Purdue University, have demonstrated that precisely engineered arrays of gold metamaterial structures can transfer up to four times more thermal energy across nanoscale gaps than conventional systems — without requiring exotic or hard-to-fabricate materials. The finding, published in Nature on May 27, 2026, opens a credible engineering path toward chips that shed heat without physical contact, and toward solid-state devices that harvest waste heat as electricity at efficiencies previously out of reach.
At the core of the result is a mechanism that had been predicted theoretically for years but never experimentally confirmed: when gold split-ring resonators patterned on silicon nitride membranes are placed face-to-face across a gap of a few hundred nanometers, the resonant electromagnetic modes of the gold structures couple strongly with naturally occurring surface phonon polaritons in the silicon nitride. That coupling amplifies thermal energy flow across the gap by as much as fourfold compared to unstructured gold plates on the same membrane, or bare silicon nitride alone.
"Unlike conventional materials, metamaterials are built with tiny, repeating patterns that interact with energy in precise ways," said Sheng Shen, a professor of mechanical engineering at Carnegie Mellon and the study's senior author. "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 Chips Cannot Simply Run Hotter
Modern chip thermal management relies on conduction and convection — moving heat through physical contact or through flowing fluid. As transistor geometries have crossed below 2 nanometers and power densities continue to rise, those approaches are running into physical limits: thermal paste can only conduct so much, and cooling fans or liquid loops add mechanical complexity, size, and failure risk. Near-field radiative heat transfer offers a fundamentally different path: transferring heat as electromagnetic energy across a gap, without touching the heat source at all.
The physics behind near-field radiative heat transfer departs sharply from everyday experience with thermal radiation. At large distances, objects radiate heat according to Planck's blackbody law — a theoretical upper bound on how much thermal radiation any object of a given temperature can emit. At gap distances smaller than the dominant thermal wavelength (roughly 10 micrometers at room temperature), that limit breaks down. Evanescent electromagnetic waves — modes that decay exponentially away from a surface and normally cannot carry energy across a gap — begin to tunnel quantum mechanically from one surface to the other, carrying heat with them. The result can be heat fluxes orders of magnitude above the classical blackbody limit.
The problem, historically, has been that most dramatic demonstrations of super-Planckian near-field radiative heat transfer relied on dielectric materials like silicon carbide or polar crystals that support strong surface phonon polaritons naturally but are not the workhorses of semiconductor manufacturing. The Carnegie Mellon result changes the constraint: it shows that gold — already a standard material in nanofabrication for its stable and tunable plasmonic behavior — can be engineered into split-ring resonator arrays that deliver comparable enhancement on a silicon nitride substrate.
Split-Ring Resonators: How the Physics Works
A split-ring resonator is a specific metamaterial geometry: typically a metallic ring or C-shaped structure with a gap, patterned at subwavelength dimensions so that it functions as a microscopic LC circuit. The structure supports resonant electromagnetic modes whose frequency is set by the inductance of the ring and the capacitance of the gap, and that frequency can be tuned by adjusting the physical dimensions. When an array of such structures is designed so its resonant frequency overlaps with the surface phonon polariton frequency of the adjacent substrate — in this case, silicon nitride — the two systems couple strongly, much as two resonant springs tuned to the same frequency will transfer energy efficiently between them.
That strong coupling is what the Carnegie Mellon team exploited. By fabricating gold split-ring resonators on suspended silicon nitride membranes and aligning two such membranes face-to-face across gaps ranging from 250 to 730 nanometers, the team measured heat transfer enhancements confirmed by both direct electromagnetic simulations and coupled-mode theory modeling. Zexiao Wang, a PhD student in Shen's research group and co-first author of the Nature paper, described the result in elemental terms: "Rather than simply adding more pathways for heat, the gold structures interact with naturally occurring energy waves in the material, known as surface phonon polaritons, creating a resonance effect. These coupled vibrations allow energy to move more freely and efficiently across the gap."
The fabrication process required only standard cleanroom steps: patterning platinum heater and sensor elements, depositing and patterning the gold split-ring resonator arrays, etching the silicon nitride, and releasing the suspended structure through a potassium hydroxide wet etch of the underlying silicon followed by supercritical drying to preserve the membranes' alignment. No materials outside the standard semiconductor and photonics toolkit were required.
SRR Geometry as Tunable Design Variable
One implication the paper establishes but does not foreground as its primary conclusion deserves emphasis for engineers considering applications. The enhancement is not a fixed number tied to one specific configuration. Because split-ring resonator resonant frequency is set by geometry — the ring's inductance and the gap's capacitance — and because surface phonon polariton frequencies are material-specific, the same engineering approach can in principle be reapplied to different substrate materials by redesigning the resonator dimensions. Silicon carbide, hexagonal boron nitride, and other polar dielectrics commonly used in photonics and semiconductor processing each have distinct surface phonon polariton frequencies; a split-ring resonator array tuned to any of those frequencies would, by the same mechanism, produce strong coupling and enhanced heat transfer on that substrate.
This makes the result a design methodology as much as a measurement. The gap distance, the resonator size, the substrate material, and the operating temperature are all parameters an engineer can vary. Prior demonstrations of enhanced near-field radiative heat transfer required choosing materials based on their intrinsic phonon polariton frequencies; this approach inverts that constraint, allowing the designer to match the enhancement mechanism to an existing material stack rather than the other way around.
Applications in Chip Cooling and Thermophotovoltaic Energy Harvesting
The most immediate application domain the paper identifies is chip cooling. As transistors shrink and power densities climb, the ability to remove heat from a chip surface without mechanical contact — and without the thermal resistance of bonding materials — becomes increasingly attractive. A nanoscale radiative heat transfer layer positioned between a processor die and a heat sink could, in principle, move heat more efficiently than conventional interfaces at the relevant scales. The Carnegie Mellon experiments were conducted under carefully controlled laboratory conditions and remain limited to nanoscale systems, but the use of gold and silicon nitride — both standard in the semiconductor industry — means the fabrication barrier to integration is lower than it would be for exotic material systems.
A second major application is thermophotovoltaics, a class of solid-state devices that convert thermal radiation directly to electricity by photon absorption in a photovoltaic cell. MIT and NREL demonstrated a thermophotovoltaic system with over 40 percent efficiency in 2022, but the power density of far-field thermophotovoltaic devices is constrained by the same blackbody limit that near-field radiative heat transfer transcends. A fourfold improvement in near-field radiative flux would translate directly into a fourfold increase in the photon flux available to a near-field thermophotovoltaic cell at the same device footprint, substantially increasing power output per unit area.
A third application, noted in the paper's abstract, is infrared sensing. Stronger and more precisely controlled near-field thermal signals could improve sensitivity in applications ranging from thermal imaging to national security infrared detection. Shen tied all three directions together in a summary statement reported by ScienceDaily on June 8: "If heat can be engineered with the same precision as electricity or light, it may open the door to a new class of technologies built not just to withstand heat, but to harness it."
Constraints and Next Steps
The paper is explicit about what the result does not yet demonstrate. The experiments were performed under high-vacuum conditions using suspended silicon nitride membranes at controlled temperatures, with gap alignment achieved through microfabrication to within 50 nanometers of coplanarity. Maintaining sub-micron gaps between thermally loaded surfaces in a real device environment — where thermal expansion and mechanical stress introduce drift — remains a significant engineering challenge.
The team measured heat transfer at three gap distances (730, 450, and 250 nanometers), confirming that the enhancement grows as the gap narrows. Future characterization across a wider range of temperatures and gaps will be needed to map the performance envelope for device design. The team has not yet demonstrated integration with an actual chip or thermophotovoltaic cell; those steps would require not just the near-field radiative heat transfer enhancement but a full thermal circuit design connecting the enhanced transfer to usable heat removal or electricity generation.
The funding sources — the Defense Threat Reduction Agency, the National Science Foundation, and the U.S. Air Force Office of Scientific Research — reflect the finding's relevance to both defense-relevant sensing applications and the fundamental physics of engineered heat flow. The fabrication work was performed at Carnegie Mellon's Bertucci Nanotechnology Laboratory and the scanning electron microscopy characterization at the university's Material Characterization Facility — established university infrastructure, underscoring that the result does not depend on unusual fabrication capabilities.
Frequently Asked Questions
What is near-field radiative heat transfer, and why does it matter for chip cooling?
Near-field radiative heat transfer occurs when two surfaces are separated by a distance smaller than the dominant wavelength of thermal radiation — roughly a few hundred nanometers to a few micrometers at chip operating temperatures. At these distances, evanescent electromagnetic waves that normally cannot carry energy across a gap begin to tunnel quantum mechanically between surfaces, producing heat fluxes that can far exceed the classical blackbody limit. For chip cooling, this matters because it offers a non-contact, solid-state mechanism to move heat away from a processor at the nanoscale — a potential complement to conventional conduction and convection cooling as transistors continue to shrink.
How do split-ring resonators enhance nanoscale heat transfer?
A split-ring resonator is a metallic ring or C-shaped structure with a gap, patterned at subwavelength dimensions so it behaves as a resonant electromagnetic circuit at specific frequencies. When the resonator's resonant frequency is engineered to match the surface phonon polariton frequency of the adjacent substrate material, the two systems couple strongly, dramatically increasing the density of electromagnetic modes that carry thermal energy across the gap. The Carnegie Mellon team demonstrated this coupling between gold split-ring resonators and the surface phonon polaritons of silicon nitride membranes, achieving up to a fourfold enhancement in heat transfer compared to unstructured gold or bare silicon nitride.
Can this metamaterial heat transfer approach work with materials already used in semiconductor manufacturing?
Yes — that is one of the result's key practical implications. Gold and silicon nitride are both standard materials in nanofabrication and photonics. Because the split-ring resonator geometry can be redesigned to match the surface phonon polariton frequencies of different substrate materials, the approach is not locked to any single exotic material. Researchers can in principle apply the same coupling mechanism to silicon carbide, hexagonal boron nitride, or other polar dielectrics already present in the semiconductor process stack.
What engineering challenges remain before this technology can reach chip-scale applications?
The result is a laboratory proof-of-concept under high-vacuum conditions with nanoscale suspended membranes. Commercial deployment in chips or thermophotovoltaic devices would require solving the challenge of maintaining sub-micron gaps under real operating conditions — including thermal expansion, mechanical vibration, and contamination. The research does not provide a product roadmap; it provides the experimental foundation that makes engineering development tractable and narrows the gap between theoretical prediction and demonstrated hardware.
ⓒ 2026 TECHTIMES.com All rights reserved. Do not reproduce without permission.
