Emissiv’s transparent radiative-cooling coating reflects 95% of sunlight by day and vents heat to deep space by night, through a window in the atmosphere. This page is the physics, with every number’s provenance stated.
By day the coating works as a mirror: 95% of incoming sunlight never lands as heat. By night it works as a radiator, shedding warmth as infrared straight through the 8–13 µm atmospheric window, the one band where the sky is transparent all the way to the cold of space. Scrub the time slider below and this scene follows.
A clear UK summer day. The dark roof today peaks near 66 °C. The coated roof stays at or below the air temperature for all 24 hours. Plain white paint is shown too, in the interest of honesty.
Surface temperature, °C. The coated roof is drawn as a band, not a line, because the out-of-window emittance εout is estimated (0.10–0.20); a lab spectrum of the cured film will narrow it to a single curve.
Coated and plain white sit within ~1 °C at the day’s peak. By day, reflectance does the work: any high-reflectance roof wins, the coating is not special yet.
After sunset the engines separate. The coating keeps shedding heat through the window while white paint, radiating into a partly-closed sky, lags behind.
The coated roof stays at or below air temperature all 24 hours. On the same day the dark roof peaks 43 °C above the air.
The Solar Reflectance Index is the industry shorthand for “how cool a roof runs”. We choose not to print one for this coating, and the reason is the honest part of the story.
SRI assumes a surface radiates heat the same at every wavelength. This coating deliberately does not: it emits strongly (ε ≈ 0.95) inside the 8–13 µm sky window and weakly outside it. Feed its computed broadband average (≈ 0.42, modelled from the measured window value and an estimated out-of-window emittance) into the standard formula and the maths reads it as a poor radiator: it lands at 48.4 °C, hotter than plain white paint. That is physically wrong, an artefact of the single-number model, not the coating.
A two-band model that credits the window emission puts it at 40.3 °C, level with a good white roof at the same albedo by day. Both figures are the same albedo under E1980 standard sun (1000 W/m², 37 °C air), so only the emittance treatment differs. The selective design’s real win is elsewhere: the night, and still air.
Nothing on this page is asserted. Each number is one of three things, and we say which.
Window emittance ε ≈ 0.95 in the 8–13 µm band, the coating’s defining property: the engine that vents heat to the sky.
95% solar reflectance. Modelled at 0.95 and now supported by measured reference-paint spectra (94–98% across 300–1400 nm, ~89% of the solar energy, and flat through the near-infrared where ordinary white paint fails). A full spectrum to 3000 nm on our own cured film is in progress to confirm it.
These curves. A clear-sky UK summer day, dew point 12 °C, a humidity-aware Berdahl & Martin sky, and ASTM E1980-style convection. Quasi-steady surface temperatures, half-hourly.
Out-of-window emittance, 0.1–0.2, which is why the coated curve is drawn as a band rather than a line. A lab spectrum of the cured film will narrow it to a single value.
The still-air (hc 2) numbers are upper bounds. The model is radiative-only, with no coupling to the roof’s thermal mass, so it over-states cooling when convection is weak. Trust the breezy case for deployment.
A reflective roof also rejects weak winter sun that was giving the building free heat. We bounded that tradeoff with each city’s own monthly sunlight (PVGIS): at most ~11% of the cooling saving under gas heating, ~12% with a heat pump. An upper bound, stated rather than hand-waved.
The same engine that draws these curves prices them, per roof, against the real grid.
For every dark roof the engine computes the solar heat a 95% coating would bounce away each year: pure physics, reflectance times the roof’s own measured sunlight (PVGIS, per location) times its area. A single large dark deck can reject thousands of MWh over a coating’s life.
On top of the sunlight it reflects, the coating actively sheds heat to deep space through the 8–13 µm sky window. The two-band model puts this at ≈ 68 W/m² net at ambient under a clear UK sky; annualised at a deliberately conservative 0.18 clear-sky duty (cloud closes the window) it lands at ≈ 13% of the reflected solar load in MWh/yr. This is separate from and additional to the heat rejected above, and is not priced in the £ below: it shows up as passive cooling, sub-ambient nights and comfort, not a bill. Modelled, clear-sky only (cloud closes the window); in winter it adds to the heating tradeoff already bounded. Full physics in the radiative-cooling simulator.
Where a building is air-conditioned, rejected heat becomes electricity not bought. The engine prices it at the live domestic cap and shows an honest low-to-high range, because cooling intensity varies. Air-conditioned stock is the minority in the UK; for the rest the gain is fewer overheated days, not a bill. The band is anchored against the best available UK evidence (no UK metered cool-roof study exists yet). For a real building we replace this assumption with your metered cooling load, so the estimate becomes your own arithmetic.
Avoided electricity is converted to CO₂ using the National Grid’s live carbon intensity, fetched at the moment you look, not a stale yearly average. Greener grid hour, smaller number: the figure is honest by construction.
See it priced on a real roof: open the map, click any dark roof, and press “Apply 95% coating”. The full chain from pixel to pound is on How it works.
The coating is the answer. The map shows where the question is loudest: the dark roofs running 40 degrees over the air.