I was thinking, fiber optic communication uses baseband signaling and even 400 gigabit Ethernet only needs ~100 GHz with the modulation that they use. So you still have 3 orders of magnitude before you hit the theoretical limit of the wavelength, and of course you can add many many more wavelengths, so it's practically limitless.

So my question is then, why use near-visible light (I know it's all IR now)? Why not use, say 1 THz waves Or 10 THz. Or 500 GHz. Does it have to do with the amount of energy that the waves carry and how quickly they attenuate?

I get confused about wavelength vs power because you can transmit visible light with 0.1W or 2 GHz waves with 100W, so I don't quite know what that relationship is.

Please keep it somewhat noob-friendly, I am not a scientist or an engineer, just curious.

  • $\begingroup$ I think the reason is that visible and NIR light sources and photodiodes are easier, technologically. But the characteristics of the fiber too. $\endgroup$
    – Pete W
    Commented Aug 23, 2021 at 19:54
  • 2
    $\begingroup$ I'm no expert but wouldn't the fibre have to be highly "translucent" at that wavelength? $\endgroup$
    – Transistor
    Commented Aug 23, 2021 at 20:00
  • $\begingroup$ Also, once you get approach 10um (30THz, at least in air), that is around the peak of blackbody radiation at room temperature. Don't know if that is a serious issue for typical communications applications, curious if someone can comment $\endgroup$
    – Pete W
    Commented Aug 23, 2021 at 22:54
  • $\begingroup$ So waveguides ? $\endgroup$ Commented Aug 24, 2021 at 10:48

2 Answers 2


You want both a laser that is easy to make and a material that can transmit that laser with low losses (transparent) and is also easy to make.

Also, for a gut feeling, 1THz is starting to become like a radio wave. Think about how radio waves seem to like to spill everywhere rather than stay confined inside a material.

Also, shorter wavelengths make the equipment required to focus them smaller. Compare a optical lens to a satellite dish. That probably isn't a primary concern though.


Shorter wavelengths get you into the microwave regime, i.e. use air or vacuum for the "fiber core."

Back to optical comms, then: Much of the design parameter space does depend on available material for the fiber and the driving lasers (absorption loss vs. wavelength is a major concern). And there are "holey fibers" which in fact have an array of open-air "cores" which match various TEM modes of the driving laser. These are difficult (so far) to manufacture but have far less absorption than any glass core.

Next, think about noise immunity. For one thing, the more wavelengths per pulse, the cleaner the square wave pulses will be -- and shorter wavelength light sources will put more wavelengths into a given carrier frequency than longer wavelengths. Typical protocols also use quadrature signalling (repeated data at different phase angles) to improve the bit error rate, so there's a lot more carriers and data pulses than the actual communications rate.


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.