Fused silica is the default window material for visible and near-infrared systems, and for good reason: it is hard, stable, inexpensive and transparent from the deep ultraviolet to roughly 2.5 µm. Push the operating band past that point and the default stops working. Absorption rises, transmission collapses, and the window that was a minor line item becomes a design decision.

SWIR cameras, thermal imagers, gas-sensing instruments and multispectral systems all live beyond that boundary, and each has to choose from a small set of materials with genuinely awkward trade-offs.
The candidates and what they cost you
Silicon transmits from about 1.2 µm to 7 µm, which covers the SWIR band and part of the MWIR. It has high refractive index — around 3.4 — which means a bare silicon window reflects roughly 30 percent per surface. That is not a coating nicety; it is a fundamental design constraint, and AR coatings on silicon are mandatory rather than optional. Silicon is also opaque in the visible, which complicates alignment during assembly: you cannot see through it with a standard borescope or camera.
Germanium transmits from about 2 µm to 14 µm and is the traditional choice for thermal imaging. Its refractive index is around 4, giving more than 35 percent reflection per uncoated surface, and its transmission is strongly temperature-dependent — absorption rises noticeably as the material heats, which matters in uncooled cameras where the window sits close to a warm housing. Germanium is also dense and expensive, and supply is concentrated.
Sapphire transmits from about 0.15 µm to 5.5 µm and is exceptionally hard — second only to diamond among common optical materials. That hardness is the reason it appears on windows that face abrasion: outer windows on airborne pods, industrial inspection heads, and any surface that gets cleaned roughly. Its refractive index is around 1.7, which is manageable, but its birefringence means sapphire is not the right choice where polarisation must be preserved.
Zinc selenide transmits from about 0.6 µm to 21 µm, covering the full MWIR and LWIR range with low absorption. It is the standard choice for CO2 laser optics at 10.6 µm and for broadband thermal windows. The trade-offs are mechanical: ZnSe is soft, scratches easily, and is toxic in dust form, which imposes handling and disposal procedures.
The trade-off table that matters
- Fused silica. Useful band 0.19–2.5 µm, high hardness, index 1.46, and it absorbs beyond 2.5 µm.
- Silicon. Useful band 1.2–7 µm, high hardness, index 3.4, but opaque in the visible and highly reflective.
- Germanium. Useful band 2–14 µm, moderate hardness, index 4.0; thermal absorption rise, dense and costly.
- Sapphire. Useful band 0.15–5.5 µm, very high hardness, index 1.7; birefringent, and expensive at large size.
- Zinc selenide. Useful band 0.6–21 µm, low hardness, index 2.4; soft, scratches easily, handling restrictions.
What the band edge really means
A transmission range quoted as "0.6 to 21 µm" describes where the material is usable, not where it is good. Near the band edges, absorption rises, and absorption in a window has two consequences: transmission loss and thermal load. In a high-power or high-fluence system, the second is the one that destroys parts.
The practical rule is to stay at least a few hundred nanometres inside the quoted band unless the application genuinely requires the edge, and to ask for absorption data — not just transmission — at the operating wavelength. A window that transmits 92 percent because of 6 percent absorption per centimetre is a different component from one that transmits 92 percent because of reflection losses.
Coating implications
High-index materials change the coating problem substantially. A single-layer AR design that works on fused silica will not work on germanium; the layer indices available are simply too far from the square root of 4. Germanium and silicon windows need multilayer designs, often with materials chosen for adhesion to the substrate rather than purely for index contrast.
Durability also varies. Coatings on ZnSe are soft by necessity, and cleaning procedures have to match. Coatings on sapphire can be made genuinely robust, which is one reason sapphire wins in harsh environments despite its cost.
There is also a thermal-expansion consideration: coating stress behaves differently on a material with a low CTE like sapphire than on germanium, and large-diameter windows can show figure change after coating. The same physics that affects precision mirrors applies here, just at a coarser tolerance.
Environmental and handling factors
Windows for fielded systems are judged on more than optics.
- Thermal shock resistance matters for windows that see rain or spray on a hot surface. Sapphire and germanium behave very differently here.
- Chemical resistance matters in industrial and marine environments. ZnSe is the weakest of the group.
- Weight matters in gimbals and airborne systems, where germanium's density becomes a system-level cost.
- Supply and cost matter at volume. Germanium and ZnSe are constrained materials; silicon and sapphire have broader supply bases.
The decision sequence
The sequence that avoids expensive mistakes is: fix the operating band first, then the environment, then the coating, and only then the price. Buyers who start from price often end up with a material that works at the design wavelength but fails at the band edge, or a window that meets the optical requirement and scratches in the first month of service.
Fused silica remains the right answer for a large share of systems. The other materials are not upgrades — they are different answers to different questions, and the question is always the same one: what does this window have to survive?