The transmission window is why one crystal shows up in a cutting head and in a thermal imager. The RFQ still has to pick a job. 10.6 µm lasers and 8–12 µm cameras punish different defects.
CO2 (10.6 µm)
You need laser-grade absorption, a coating that actually hits 10.6 µm, and usually a visible HeNe path. Thermal lensing (dn/dT ~ 61 ppm/K, modest conductivity) is the design constraint once the AR is honest. This is where KM’s laser-grade sheet, cores and windows earn the freight from Gwangju.
Thermal / LWIR (8–12 µm)
You need scatter control, cosmetic quality, and often a different AR. Optical grade is the default. A camera window that “worked on the FTIR” can still be a bad laser lens. The reverse is also true: a laser V-coat can starve a broadband imager.
Why not Ge, ZnS, Si
| Material | Does well | Gives up |
|---|---|---|
| ZnSe | 0.6–16 µm, HeNe + CO2, lower absorption than many ZnS lots at 10.6 | Soft, toxic, dn/dT |
| Ge | LWIR index, toughness vs ZnSe | Opaque in the visible, thermal runaway |
| ZnS (CVD / MS) | Harder, some visible, harsh environments | Often more scatter / absorption at 10.6 than laser ZnSe |
| Si | MWIR, cheap, hard | Cuts off before LWIR / 10.6 |
We quote ZnSe. If the print is germanium, you are on the wrong desk — say so and we will not pretend. If the print is “IR window” with no wavelength, we will ask 10.6 or 8–12 before a number leaves Florida.