I wish to know the fabrication difficulties limiting the manufacture of very low surface sidewall roughness in III-V optical waveguides?
Number one is the etch-mask you use. If you go to high-res Deep-UV stepper, you'll typically get the lowest sidewall roughness. Some tricks can be played with Contact (otherwise the highest-roughness lithography) eg. small address unit, photoresist reflowing.
Additionally, the transfer of this mask into the InP plays a large role. Here users found that the (rather complicated) process of Photoresist:Chrome:SiO2:InP ridge yielded the best loss, but then all of these etches need to be developed (Chrome in particular can be tricky). The InP etch itself can have some smoothing capabilities, eg. addition of N2 or H2 as a process gas, with some tradeoff with verticality.
There are, as you've probably found, many papers on InP ridge waveguide sidewall roughness. For example:
J. W. Bae, W. Zhao, J. H. Jang, I. Adesida, A. Lepore, M. Kwakernaak, and J. H. Abeles, “Characterization of sidewall roughness of InP/InGaAsP etched using inductively coupled plasma for low loss optical waveguide applications,” Journal of Vacuum Science & Technology B, vol. 21, no. 6, pp. 2888–2891, Nov. 2003.
C. Carlstrom, E. Van Der Drift, R. Van Der Heijden, R. Notzel, R. van Veldhoven, F. Karouta, H. Salemink, and A. Talneau, “Cl2/O2- and Cl2/N2-based inductively coupled plasma etching of photonic crystals in InP: sidewall passivation,” Indium Phosphide and Related Materials, 2005. International Conference on, pp. 315–318, 2005.
Note the inspection methods: SEM sometimes looks "smooth", only because the SEM's resolution (or charging) isn't good enough! AFM is a great way to get RMS roughness, as long as you don't have an undercut (check your SEMs for this). AFM would be particularly applicable for checking the line-edge roughness of your masks (PR, SiO2 etc.)
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Optics