I work with SERS, where the Raman signal gets enhanced by the electric field of the nanoparticles. Now I like to simulate this kind of spectrum with Gaussian. I found a paper, where it looks very nice how they simulate it. They don't get a bond between the molecule and the silver nanoparticle. If I try it on my own, then the nanoparticles are thrown away by Gaussian. Has anyone of you an idea of how they simulate it? My idea is the following workflow: Frist, they optimize the attached molecule alone; after this, they made a scan with one variable. The variable is the distance between the part of the molecule where the nanoparticle should attach and the nanoparticle itself. And after this, they put the nanoparticle in the distance with the lowest energy and make a single point calculation with options to get the orbitals and the ESP mapping. It makes sense, or is it not an ordinary workflow for gaussian? I learned Gaussian myself self so I'm not sure at all.
 Gao, Y.; Xu, M.-L.; Xiong, J. J. Environ. Sci. Heal. Part B 2019, 54, 665-675
This is my route section:
#p opt=vtight int=ultrafine oniom(PW6B95D3/aug-cc-pvtz:b3lyp/lanl2dz) nosymm geom=connectivity
I used the OMNIC layer model and used the high layer for the molecule with
PW6B95D3/aug-cc-pvtz and the low layer for the gold atom with
b3lyp/lanl2dz. I have no idea how to fix the gold atom in a specific region.