Backtrack of today
I sat down and wrote down whatever came to mind. Of course I should be finalising photomask designs, but designs can always be improved, hence the exploration.
I sorted the photomask for the diffraction grating (at least I've generated 1/100th of the final image size, which is good enough to pass on to the supplier...)
Next photomask type was the square patch, hexagonally arranged design. I couldn't help wonder about my already fabricated version.
Could 'Volumetric' etching in principle tailor the lateral size into a range I could characterise? My estimate was 300 GHz to 700GHz, but I should know for sure. SEM measure for sure, and then apply accurate isotropic etching while the top is protected (is it protected?). The SEM would also tell me if the regions between the patches (and then very likely, the side walls) are free of Negative Photo Resist, and I already know the top is covered by the obvious colour difference.
What etch rate? And what would I target?
KOH for 50 minutes would remove 120um, and retain 60um patch resonators. What is the expected resonance of that size -- by x3.3 shift it would be -- 1 THz - 2.3 THz -- or 0.3mm to 0.13mm. Is that compatible with Claudio's lab. Just about. Potentially. His range is 750 GHz to 1.1 GHz. So I could opt for 37.5 minutes of etching.
The remaining question is whether the negative PR will hold up to the KOH solution. And if KOH is directional (anisotrophic) for Aluminum, as it is for Silicon. Barring that I could deposit a much thicker Aluminium layer via thermal evapouration (?) say 100050 nm or 100.05um (to leave 50nm of thickness... for that accuracy I would need Al etch A not KOH though). This may also dome the resonators, or at least chaffer the edges. Would likely need to work that into a future simulation.
So using one of my pattern generator to vary the spacing between patch resonators in lieu of wet etching post- submersion in developer solution (as this was empirically seen to perform the job on Aluminium without etchant..!), and then running that through the simulation, again with consideration for dimensions/scales in various directions.. Z is nm, while X-Y is um.
That's 1 or 2 'jobs'. To get accurate predictions for that, and then actually doing the KOH to then characterise something expectantly within the target range... difficult.
The other thing was correcting the unit cell of my 'meta stick' i.e. the HfO2-TiN trenches on TiN-SiO2 base. The base of my SiO2 was not considerate of the metasurface unit cell boundary conditions. Yet it did yield a resonance -- 850nm approx. Can that be tailored? So it gave me an idea for a unit cell design to test this design structure -- but the issue was also the validity of 1nm thickness for the assumption of the 'unit cell' extrapolating to the width of a microscope slide. And also the curiousity of 5nm yielding a convergence to 0.918 which did match the paper broadband somewhat, and my 3nm yielded 0.9532 odd.. does that make sense? So my idea to test a 3D structure that could be photolithography fabricated. I'll explain it.
on the TiN base instead there are SiO2 island, their width can cover the 60um and their spacing can be 12um, on the SiO2 islands are the TiN trenches (that can be capped with HfO2) and their width is 36um. I could argue that this isn't even a metasurface, although technically it is because the unit cell is that structure i just described. Most important is the SiO2 thickness. That must be precision.
I think this could be outside the scope of the current photomask order, but with certain variations it could be made to fit within the same ones we will have. With good alignment and a little thought, the 3 variants of existing trench widths could be used to create the SiO2 islands.