InP goes better with GaAs not silicon, Full story at http://www.physorg.com/news77804050.html

[guiding_light]

http://www.physorg.com/news77804050.html

The wider bandgap of GaAs (1.4 eV) allows shorter wavelengths to be used in waveguides than Si (1.1 eV).

If you"re going to use III-V, do it right.

[Ron]

Hi G L,
It seems to me the biggest advantage to using silicon in this application is cost , and with the bandwidths they're claiming with silicon, do you really think they'd gain that much even if they could use GaAs?
I've worked with UCSB on GaN projects (whose band gap blows silicon away) so I'd think they've at least thought about these things. Also, InP can be really difficult to lattice match to GaAs, and it doesn't seem like this application would accommodate a buffer layer. I'm only familiar with RF applications, so, I could be way off. Any thoughts?
Later,
Ron

[guiding_light]

The low silicon cost is overshadowed by using the InP wafer and processing which is even less mature than GaAs.

InGaAsP is already developed for several years for telecommunications, so I figured compatibility between InP and GaAs will not be such a big issue.

[Ron]

Hi again GL,
That makes sense to me. I'll buy it.
Thanks,
Ron