# How do non-mechanical solid-state optical switches work?

I am currently looking for a fiber-optical switch (FOS) in order to be able to change the light source of a spectrometer. As this will be used in harsh conditions, I was hoping to find a FOS with no moving parts.

Today I came across this, which appears to be an optical fiber switch with no moving parts. Although I will not be able to use this switch due to the wavelength range not being a good fit, I was wondering how these kinds of switches work. So far, I have only seen optomechanical ones which I assume to use some sort of actuated mirror assembly to redirect the light from the correct input fibers to the output fiber. On a smaller scale, I have seen piezoelectric devices which just move the physical output fiber by a tiny amount left and right.

I have read somewhere about a magneto-optic crystals, which appear to be a crystal whose transmission is based on the voltage applied to it. Is this how such a switch could be implemented? Are there other ways?

• you could also take a look at optical switches from thorlabs. I do not know how those switches are made. Feb 2, 2021 at 23:04
• This question is the subject of a Meta discussion.
– rob
Feb 9, 2021 at 1:47
• In general I would not find such a device complicated to understand, but the switch says 2 things that are a bit incompatible for me: it latches with no power and it does not have movable parts. I would imagine that changing something to do with the faraday rotator without later on needing power means for example a rotation of a magnet. I could not yet find the patent for such a device (but I am looking). Feb 12, 2021 at 11:48
• @huhmonster Ok, I am sorry for the spamming, but found a paper which seems to point in the direction of this kind of switches: quasi-passive latching switches Feb 12, 2021 at 12:10
• I may have time to write an answer later, but two ways this can be done is with magnetically controlled polarization changes and with bistable MEMS mirrors (image). Lots of information by looking at the patents for Agiltron. Feb 16, 2021 at 1:55

# Bistable MEMS

MEMS (Microelectromechanical systems) are very small structures, with features from $$1-100\mu m$$ in size, generally made with semiconductor-like processes. They can be made so they are bistable, if you apply a pulse of power, their state will change and remain stable until another pulse of power is applied to change it back. Patent US8203775B1 describes one of these structures used by Agiltron for at least some of their 1xN and NxN optical switches.

Here is the patent drawing of the two states the MEMS mirror can be in:

Here is a diagram (from Agiltron) showing how they are used to form a 1xN switch:

# Magneto-optic effect

This can be achieved with materials that change the polarization of light like bismuth-substituted rare earth iron garnet single crystal that depending on the polarity of the applied magnetic field, change the polarization by plus or minus 45 deg. The light will then either be blocked or transmitted by a static polarizer. A magnetic field applied to the rare earth iron garnet will magnetize it, so a continuous field is not required. You only need a pulse of power to change the polarity of the magnetization. Below is a diagram from patent US6577430B1 that shows the stack up inside a bidirectional 1x2 switch. I believe this technique is used in Agiltron's CrystaLatch 1xN and NxN switches.

• Super interesting, I had no idea that "switchable" magnets existed. I just assumed they would have to move them mechanically to change the field. Although I didn't ask this question, I was definitely very curious about it, so, many thanks for this great answer. Feb 17, 2021 at 19:54
• Do bistable MEMS count as "no moving parts" as far as engineering goes? Feb 18, 2021 at 13:32
• Generally yes, they are small enough that "normal" issues with mechanical systems don't apply. People would say thay the accelerometers in your phone are solid state and not mechanical, even though they contain MEMS. Feb 18, 2021 at 14:10