Most optical devices exhibit some vulnerability to temperature changes. However, the high thermo-optic coefficient of silicon (1.86×10-4 K-1) and the wavelength selectivity of microring resonators make them especially susceptible to fluctuations in temperature.

This requires a very tight range of thermal tuning for the microring and deviation from the selected temperature by less than 2 F degrees will makes the device inoperable (wavelength drift will be  outside of  the  operating range). A really good bench demonstration of this by the founders of Ayar Labs was presented some time ago on this forum.  Can’t find the demo video clip but here is the article: https://www.nature.com/articles/nature16454#Sec14   

Andrew Richman (Rockley) has presented a discussion on silicon photonics where he makes reference to bigger is better. One of the reasons for this is that the larger the waveguide is the larger the process deviation that can be tolerated. A 1nm variation in a 100nm wide waveguide is a deviation of .01. A variation 1nm in a 1000nm wide waveguide is a variation of .001 (arbitrary numbers for demonstration). The wavelength error is proportional to waveguide dimensional deviation. So now consider the expansion as a result of temperature variation and we come to one of the biggest challenges associated with silicon photonics. That is dimension change from temperature but we also have refractive index changes from temperature variation. We know that silicon goes through very large changes in temperature and the electronics are designed to tolerate this but not the photonics.

So to get back to the discussion on the microring modulator (or any microring device) solutions are focused on maintaining temperature stability of the microring (in a silicon photonics environment this means active control of the temperature) and/or solutions that reduce thermal dependency (athermal solution).

Silicon as stated has a high thermo-optic coefficient so the insertion of a negative thermo-optic (polymer) is a solution but is extremely difficult to implement due to high fabrication temperatures associated with silicon processing steps and the high operating temperature of  silicon. Introduction as a backend solution requires significant re-engineering and is not what you want in a zero-change silicon process,

POET’s waveguides are athermal and part of the secret sauce we think is the use of polymers which offsets changes in the refractive index caused by temperature rise.

The strict requirement on waveguide dimensions may make it difficult to consistently achieve the desired athermalization given the natural variations present in fabricating silicon photonic structures.

The other solution to produce an athermal microring is to insert it into one arm of an MZI to provide a negative thermo-optic coefficient to counter silicon’s high positive thermo-optic coefficient but now  you are messing with thhe footprint.

So to conclude we once again look to POET to provide the answer. A big reason for using microring modulators in silicon photonics is because they don’t take up much space and they are efficient.  But presents problems as discussed.  POET has choices as one of the benefits is the ability to mix and match devices according to application a customer requirements. Isolation and control of the temperature signature of each devise provides the POET platform with great flexibility that is not constrained and can be adapted for future devices. The reason why POET can now talk about optical integration into AI and application to the growing requirements 5G and the edge.

Simply stated…POET keeps it simple and industry needs platform that can adapt.