This PR is to implement 3D device geometry. Currently, we consider multi-material stacks which varies only along z (thickness of the films). This PR will bring us a step closer to simulating entire NCFETs instead of just the gate stack.
Following new features have been added:
Make FE/DE/SC lo and hi a three component array instead of scalar
Add diagnostics for epsilon to verify/visualize device geometry
Options for periodic, free (dP_i/dn = 0), and surface BC (dP_i/dn = P_i/lambda)for polarization was implemented along the lateral dimensions for the TDGL equation.
Material permittivity is parsed at cell centers now and it is averaged to cell faces in the code to be used by the Poisson solver.
Verifications:
Previous results can be reproduced with non EB simulations:
A voltage sweep from -5V to 5V was performed in an MFIM stack simulation
Simulations with EB were done for MFIM and MFIS stacks, where the high z Dirichlet BC was replaced by an EB object of box type with the surface potential on EB as the Dirichlet BC phi value.
Co-authored with @PeculiarOvertones.
This PR is to implement 3D device geometry. Currently, we consider multi-material stacks which varies only along z (thickness of the films). This PR will bring us a step closer to simulating entire NCFETs instead of just the gate stack.
Following new features have been added:
Verifications: Previous results can be reproduced with non EB simulations: