The thermal budget
One order of magnitude decides it.
At 15 gigawatts on a 16 square meter sail the irradiance is 9.4 × 10⁸ watts per square meter — roughly 690,000 times the solar constant at Earth. Whatever fraction the sail absorbs becomes heat, radiated from both faces.
| Absorptivity | Equilibrium temperature | Verdict |
|---|---|---|
| 1 × 10⁻⁴ | 1,134 K | Exceeds the survivable range for any known dielectric membrane |
| 1 × 10⁻⁵ | 638 K | Workable for silicon nitride and silicon carbide photonic crystals |
| 3 × 10⁻⁶ | 472 K | Comfortable margin; likely required for a reusable design |
So the specification is an absorptivity below one part in 100,000. That is achievable in a narrowband photonic crystal — a membrane patterned with sub-wavelength structure so that it reflects a chosen band almost perfectly and absorbs almost nothing in it.
The difficulty is that the band moves.
The band moves
A 22 percent Doppler shift during the burn.
The sail is receding from the beam, so the beam is redshifted in the sail’s frame by the square root of the ratio of one plus β to one minus β. At 0.2c that factor is 1.2247.
A 1.06-micron laser arrives as 1.06 microns at ignition and 1.30 microns at burnout. The photonic crystal must hold both high reflectivity and absorptivity below 10⁻⁵ across a 22 percent wavelength band, at a gigawatt per square meter, at several hundred kelvin, in a membrane on the order of 100 nanometers thick, with enough tensile strength to survive 10,000 g.
Narrowband is easy and broadband is not. A dielectric stack tuned to one wavelength can be made almost perfectly reflective; holding that performance across a fifth of the spectrum while keeping the membrane thin enough to accelerate is the whole problem.
Which discipline owns this
Semiconductor fabrication.
This is the hardest single specification in the interstellar argument, and it is not a propulsion problem or a power problem. It is a thin-film photonics problem, which is to say a semiconductor fabrication problem, which is to say it sits squarely inside the industrial base the rest of this site describes.
That matters for the same reason everything else here does. A propulsion problem needs a propulsion program. A thin-film patterning problem at 100 nanometers needs a fab, and by the 2070s the architecture has fabs off Earth for entirely unrelated reasons — orbital compute, thin-film collector production, and Mercury’s film line. The sail is a product those facilities could make rather than a program someone must fund.
One caution against over-reading that. Mercury’s film is a 10-gram-per-square-meter metallized reflector optimized for cost per square meter. The sail is a 100-nanometer patterned dielectric optimized for absorptivity at one part in 100,000. They share a fab and a materials tradition, not a product line.
Where the art stands
A pentagonal lattice with a billion holes.
In 2025 a Delft group fabricated a pentagonal photonic-crystal reflector 60 by 60 millimeters and 200 nanometers thick, with over a billion nanoscale holes placed by neural-topology optimization, achieving broadband response without added thickness and cutting modeled fabrication cost per square meter by a factor of 9,000. It is the highest-aspect-ratio nanophotonic element made to date.
Caltech’s light-sail group works the companion problem of Doppler-broadened, self-stabilizing metasurface sails — structures that not only tolerate the moving band but generate a restoring force that keeps the sail centered in the beam without active control, which matters because a gram-scale craft has no attitude system worth the name.
| Property | Demonstrated (2025) | Required | Gap |
|---|---|---|---|
| Area | 60 × 60 mm | 4 m diameter | ~10⁴ |
| Thickness | 200 nm | ~100 nm | Within reach |
| Absorptivity | Not characterized at this flux | <10⁻⁵ | The open item |
| Bandwidth | Broadband demonstrated | 22% band at <10⁻⁵ | The open item |
| Acceleration tolerance | Not tested | 10,000 g | Untested |
These are the beginning of that line rather than the end of it. The area gap is a manufacturing scale-up. The absorptivity and bandwidth gaps are the two places where nobody yet knows the answer.
This is one of two genuine open items in the beamed branch rather than a pure scale-up. The other is solar-pumped laser efficiency. Everything else in the architecture — the array, the collector, the phasing, the pointing — exists at small scale on the correct physics and needs to be made larger. These two need an answer nobody has.