I'm trying to figure out whether lenses for thermal (long-wave infrared, i.e., 7-14µm) cameras are inherently more expensive or less capable than the lenses for visible cameras.

Production lenses for digital cameras often incorporate a dozen or more individual glass elements up to perhaps 50mm in diameter. With surface coatings these systems still transmit over 90% of light in the visible spectrum, and are able to nearly eliminate all forms of distortion while providing a wide focal (zoom) range. These systems can be mass produced for a few hundred dollars per unit.

In the limit (i.e., assuming the same economies of scale were provided), is similar performance and/or price possible with current uncooled thermal lens technology?

I am having trouble determining the state of the art in thermal lenses. I know that for production cameras at least the front element is typically made of coated Germanium, or Ge-As, and a single Ge element in real-world temperatures (i.e., below 50°C) transmits over 90% across the LWIR spectrum. (My understanding is that Zn-Se and Zn-S are also very good in the thermal spectrum, but they are too soft to be exposed, so they could work as interior elements, but not the front element.) But I don't know, for example:

  1. Is transmission through these materials so low that using more than a couple of elements is not practical?
  2. Are these materials inherently significantly more expensive to manufacture than the glasses used in visible light lenses?
  3. Are the physics of refraction in these materials in the LWIR spectrum too variable to build lenses with many elements, or with "zoom" capabilities?
  • $\begingroup$ as for price... the classic supply-demand-price set from Economy 101, with low demand the unit production costs are high. The demand for visible light cameras is vastly higher, so bulk-produced lenses are bound to be cheaper regardless of quality differences. $\endgroup$
    – SF.
    Commented May 31, 2017 at 12:56

2 Answers 2


I have worked with specifying optical materials for use as windows in visible and IR applications. Optical material selection will be determined by two constraints:

  1. The portion of the EM spectrum you need to observe. You specified LWIR, but in my limited experience only certain wavelength bands are observed. You will find that many materials used for IR imaging have uneven transparency ratings as a function of light wavelength. For IR applications ZN-S or Cleartran is the go to material for optical windows.
  2. Manufacturing costs. Optical materials must be grown and formed into what is called a blank. Then initially machined into your desired shape, and then polished. Depending on the size of your application, and how tight your optical tolerances are, the polishing process can take months. Other things to consider are homogeneity of index of refraction in the material, it can vary slightly in the material, number of occlusions allowed per volume (defects, bubbles, and other stuff that blocks transmission), and then finally coatings. If your lens is exposed to the elements, you will need one coating for abrasion (usually the one you mentioned), coatings for anti reflection may be required, and so on.

To sum up:

Your application will drive your selection to one or two materials and then the manufacturing process.

Material performance is one metric, manufacturing cost is your second.

Your best bet for a real answer is to contact one or three optical material vendors. The more information you can provide to them about what your use case is, the better their answer will be.


Got the following info from Edmunds:

These lenses can have comparable performance to visible spectrum lenses given the proper coatings and arrangements within the optical system. As far as I know, the greatest limiting factor for high precision IR imaging lenses is manufacturing costs and substrate durability. Many of the IR substrates cannot handle as wide a range of conditions, whether that be moisture/weather related or temperature related.

Now ... LWIR is such a wide range of wavelengths that a given imaging system cannot possibly work across all of it. If you compare to the visible spectrum, most systems are accounting for about 300nm of variation in optical properties. Some systems cannot really handle that, but are used in more specific situations at precise design wavelengths so it is not really a problem. The LWIR has a range of several microns which means that some regions of the LWIR transmit through materials in a very different fashion than others. If LWIR applications are more specified to a specific region that has a similar bandwidth to the visible spectrum, I think designing lenses with great performance would be quite feasible, albeit still very expensive.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.