EPREVAEngineer Paper
ENERGYNewsMaterialsEnergyComputing & AI for engineering

Simulation shows tungsten metasurface absorber reaching 98.3% average absorptivity

Simulations show the optimised structure absorbs 98.3% on average across 350–1,000 nm and reaches 99.4% calculated spectral conversion efficiency at 2,000 K.

Interior of a modern workshop featuring detailed automotive machinery and equipment.
Photo: Mike van Schoonderwalt / Pexels

A tungsten-based metasurface designed with machine learning absorbs 98.3% of light on average across the 350–1,000-nanometre wavelength range, according to simulations. Researchers at the Changchun Institute of Optics, Fine Mechanics and Physics, part of the Chinese Academy of Sciences, with collaborators, report the result in a peer-reviewed study in Applied Thermal Engineering. The metasurface works as both a selective absorber and a thermal emitter.

The structure uses a metal–insulator–metal design with arranged nanoscale resonators. The team chose tungsten for its optical absorption characteristics and its suitability for high-temperature applications, according to the press release. They combined finite-difference time-domain simulations with machine learning to study how structural parameters affect light absorption. They compared six algorithms and identified XGBoost as the most effective for predicting absorption and guiding optimisation.

The reported performance comes from simulations, not from a fabricated device or a field test. In the simulations, the optimised metasurface reached an average absorptivity of 98.3% across 350–1,000 nm, with absorption of 99.8% in the 450–700-nanometre region. It maintained more than 90% absorption at incident angles up to 60° under both major polarisation conditions.

As a selective thermal emitter, the structure achieved a calculated spectral conversion efficiency of 99.4% at 2,000 K. The researchers say this indicates potential for directing thermal radiation toward wavelengths useful for silicon photovoltaic cells. The press release frames this as a promising route to improving solar thermophotovoltaic energy conversion, in which absorbed sunlight becomes heat and is re-emitted as radiation that photovoltaic cells convert into electricity.

The release notes that conventional absorbers often have limited spectral selectivity, are sensitive to incident angles and show reduced performance under demanding operating conditions, and that optimising nanostructures usually requires extensive numerical simulation. The authors suggest the approach could also contribute to photodetectors, optical sensors and other devices requiring precise light control.

Key facts

  • 98.3 % average absorptivity across 350–1,000 nm, from simulation
  • 99.8 % absorption in the 450–700 nm region
  • More than 90 % absorption at incident angles up to 60° under both major polarisations
  • 99.4 % calculated spectral conversion efficiency at 2,000 K as a selective thermal emitter
  • Six machine learning algorithms compared; XGBoost identified as most effective

Sources

  1. Machine learning-assisted ultra-selective tungsten metasurface absorber/emitter for high-efficiency Vis-NIR photothermal conversion — Applied Thermal Engineering — Riaz Ali, Wei Su, Iqra Mamoon et al.
  2. Machine learning informs highly efficient light absorption in tungsten metasurfaces — Phys.org (Technology)

Discussion

Technical questions and corrections are welcome. Comments are signed and checked before they appear.

No comments yet.

Sign in to add a typed review comment. Reviews are signed with your name.

More news