A LiDAR aperture is a piece of optical glass until the first cold morning, when it becomes a dew trap. A sensor that has been cold-soaked overnight at minus ten degrees and is then driven into a humid loading bay will fog within seconds, and a fogged aperture does not degrade gracefully. It loses range in the centre of the field and loses it completely at the edges, which is exactly where the safety-critical detections live. Fleets discovered this the hard way over the past two winters, and the fix that survived procurement review was not a wiper, a hydrophobic dip or a compressed-air jet. It was a heater laminated into the window itself.

That shift matters well beyond automotive. Autonomous mobile robots in cold stores, drone delivery in coastal climates, and fixed LiDAR on port and yard equipment all run into the same physics, and all of them are now writing a heated aperture into the optical specification rather than treating it as an accessory.
Why passive solutions stop working
Hydrophobic coatings raise the contact angle and help water sheet off, but they do nothing about condensation that forms from the air side when the optic is below dew point. Wipers need space, motors and maintenance intervals that fleet operators will not accept. Air knives consume power continuously and stir up dust. A transparent heater addresses the actual mechanism: it holds the outer surface a few kelvin above dew point, so the water never nucleates.
The power involved is smaller than most designers expect. A 30 by 40 mm aperture held 5 K above ambient in still air typically needs 1 to 3 W; wind at 20 m/s can triple that. On a robot with a 48 V pack this is affordable, but it is not free, and it forces a real conversation about warm-up time, duty cycle and where the energy comes from when the platform is idle.
Three ways to make a window conductive
Indium tin oxide is the incumbent. A 100 to 300 nm film gives sheet resistance in the 10 to 100 ohm/square range at 85 to 90 percent luminous transmission, and it is deposited by the same vacuum houses that already coat the window. Its weakness is near-infrared absorption: at 905 nm the film loses more than the visible transmission figure suggests, and at 1550 nm the loss and the heating of the film itself both climb. For time-of-flight sensors working at 905 nm this is manageable. For 1550 nm FMCW systems the heater and the AR stack have to be co-designed, not stacked independently.
Fine metal mesh is the second route. Lines 3 to 5 µm wide at 20 to 50 ohm/square reach 88 to 92 percent transmission and add electromagnetic shielding as a free by-product, which matters for sensors mounted close to switching electronics. The cost is pattern visibility: a mesh can print a faint grid into an imaging channel, and the mesh period must be kept far from the pixel pitch to avoid moiré.
Carbon nanotube and graphene films sit at the research-to-product boundary. They promise higher transmission at a given resistance, but sheet resistance uniformity across a 50 mm part and long-term stability under thermal cycling are still the questions that decide whether a design goes to volume.
What heating does to the optical specification
A heater turns the window into a thermally inhomogeneous element. Watt density is rarely uniform, because busbars concentrate current at the edges, so the centre runs cooler than the perimeter. The resulting refractive index gradient does not blur a ToF channel much, but it does bend the wavefront, and on a coaxial architecture with a narrow receive aperture that bend shows up as range bias rather than as an obvious image defect. Thermal bow is the second effect: a laminate of glass, heater and AR coating with mismatched expansion coefficients will curve by a few micrometres over 40 mm, which is enough to shift a collimated return off a small detector.
Both effects are controllable, but only if the drawing states them. A heater specified purely by resistance and wattage leaves the coating house free to optimise transmission and ignore figure.
Busbars, delamination and the tests nobody runs
Nearly every field failure of a heated window traces back to the busbar. Silver frit printed at the edge, a soldered tab, a flex tail bonded with anisotropic adhesive: each is a CTE mismatch against glass, and each sees the full thermal excursion of the application. Current crowding at the tab creates a hot spot that can exceed the film's stability limit, and repeated cycling grows a delamination front that starts at the edge and walks inward.
The tests that catch this are not exotic, they are simply not in most optical datasheets. Thermal shock from minus 40 to plus 85 degrees C for 100 cycles, damp heat at 85/85 for 1,000 hours, 10,000 power cycles, and an adhesion check after each. Add salt fog for anything that lives outdoors, and an IP69K jet test for equipment that gets washed down. A supplier who can show those results per lot is a different kind of supplier from one who can only show a transmission curve.
Writing it into the RFQ
The requirement that gets a usable quote reads roughly like this: heater resistance 12 ohm plus or minus 10 percent, watt density between 0.5 and 2 W/cm², surface reaches 5 K above dew point within 90 seconds from minus 20 degrees C at 20 m/s airflow, transmission at 905 nm above 88 percent, haze below 0.5 percent, transmitted wavefront error below lambda/2 over the clear aperture, AR coating on the outside face, and the full cycling test suite reported per lot.
That is a longer paragraph than most optical buyers are used to writing. It is also the paragraph that separates a window from a sensor that works in February.
Product page: optical windows and heat-resistant glass — specifications, coating options and lead times.