While simulations are fairly accurate for simple Photonic Crystal Structures, we had a hard time creating the exact designs in Stanford's CIS. The problem is that at almost every step of fabrication there is some uncertainty, and the uncertainties add up. Our process is very simple. We use the Raith E-Beam system and a reactive ion etch to create a pattern of holes in PMMA resist and transfer it to a GaAs membrane. The important parameters are hole circularity, diameter and spacing. In order to get the right operation wavelength and cavity properties we found that the best solution is to start with a theoretical design and explore the fabrication parameter space.

The following procedure has allowed us to get the best from the Raith (note that this work was done before the new proximity correction system, and things may change):

1. Get VERY good at stigmation and focusing - the burn dot should be as round as possible.
2. Adjust radii of holes to be ~ 10 nm less than target and write a dose array to find the appropriate result
3. Write an array of cavities and PC structures with varying periodicities:
1. For our work we wrote an array spanning the period from 234 nm to 256 nm in 2 nm increments to get a final periodicity of 246 nm for the target wavelength
2. Write this array with the smallest radius and increase the dose to get the target radius. For our work we wrote a radius of 60 nm and increased the dose to span the space up to 85 nm. 




4. Holes which are next to edges of the PC lattice or defects have a lower dose. The outer most holes o the PC lattice should be adjusted to a dose of 1.1 or 1.2 the inner most holes or holes along the waveguide should be adjusted anywhere from 1.08 to 1.2 (1.1 being the usual dose) to hit the target design.