How to solve? Unexpected Behaviour of IV Analyzer Results With the same context.

[techenthusiast]

Dear experts,

Good morning, I have attached my script for a  GAA device structure ।

Just for your kind information -I have used a poison solver: Conjugate gradient

Top and button -Neuman boundary conditions
Front and back -periodic
Side by side(Left and Right)  -Dirichlet

Slater koster model ।

I have doped using -The miscellaneous section -Doping just above the spatial region of the current version of the software।
It does not provide suitable IV Analyzer Results such as -https://www.sciencedirect.com/science/article/abs/pii/S143484112030892X

Thank you once again।

[techenthusiast]

Dear experts,

Good morning, I have attached my script for a  GAA device structure ।

Just for your kind information -I have used a poison solver: Conjugate gradient

Top and button -Neuman boundary conditions
Front and back -periodic
Side by side(Left and Right)  -Dirichlet

Slater koster model ।

I have doped using -The miscellaneous section -Doping just above the spatial region of the current version of the software।
It does not provide suitable IV Analyzer Results such as -https://www.sciencedirect.com/science/article/abs/pii/S143484112030892X

Thank you once again।

Enclosed is unexpected behavior with conventional IV Analyzer procedures. Valuable input would be highly appreciated

[techenthusiast]

Dear experts,

Good morning, I have attached my script for a  GAA device structure ।

Just for your kind information -I have used a poison solver: Conjugate gradient

Top and button -Neuman boundary conditions
Front and back -periodic
Side by side(Left and Right)  -Dirichlet

Slater koster model ।

I have doped using -The miscellaneous section -Doping just above the spatial region of the current version of the software।
It does not provide suitable IV Analyzer Results such as -https://www.sciencedirect.com/science/article/abs/pii/S143484112030892X

Thank you once again।

Enclosed is unexpected behavior with conventional IV Analyzer procedures. Valuable input would be highly appreciated

Also , It is taking long time to simulate the device which is unusual.  It is stuck at Execution 1 for long time.

[Anders Blom]

Looks like there might be a convergence problem, but impossible to say without log files.
I would make the XY box larger, esp. since your spatial regions extend outside (which is a mistake, those parts get cut off) but also to avoid periodic images interacting. You do not need Neumann in top/button, since you have a metal gate around, just use periodic (FFT2D).

[techenthusiast]

Thanks, sir, for your prompt response. Enclosing herewith the file and other associated files for the same.   

the main focus is to get the Dip of ambipolar behavior at the negative x axis (Voltage) of the given device. Maybe it is called Dirac point.  People are taking that point as off current for this kind of scenario. Even attached journal link is getting like that only, even for "N-I-N" Doping profile of a Conventional CNTFET.

So, what will be the approach for exact workflow for efficient calculation of these kind of devices such as Cylindrical GAA and to speed up the simulation process with 16 MPI (multiprocessor parallel). Only IV Analysis might be taken place.

Thanks and highly obliged.

[Anders Blom]

I already outlined the changes needed to the geometry, then one needs to monitor convergence by looking at the log files.
16 MPIs should provide good speedup for this system, and it will scale to much higher number of processes too.
With Slater-Koster it should not be too time-consuming, at least if it converges reasonably fast.

[techenthusiast]

Thanks for your kind reply. Highly appreciated.  Also, what is the meaning periodic image interacting. Ref-
 
https://sci-hub.st/10.1016/j.aeue.2020.153354. I didn't able to replicate the result from curiosity point of view with the same kind of device structure. 

Thank You.   

[sukhito teh]

I never use the Slater Koster Model, so I can only provide limited suggestion. As the other comment had pointed out, you may need to increase vector_a and vector_b to reduce interaction between neighboring cells. I also don't see any calculation regarding the device hamiltonian, so it is hard to tell whether convergence happens.

[techenthusiast]

I never use the Slater Koster Model, so I can only provide limited suggestion. As the other comment had pointed out, you may need to increase vector_a and vector_b to reduce interaction between neighboring cells. I also don't see any calculation regarding the device hamiltonian, so it is hard to tell whether convergence happens.

Thanks for your kind reply. Highly appreciated.   Ref-
https://sci-hub.st/10.1016/j.aeue.2020.153354. I didn't able to replicate the result from curiosity point of view with the same kind of device structure. ..

Eventually oxide thickness is 1.5nm they have mentioned. Which is showing peculiar at VNL for 7nm technology nodes and 10nm Source /Drain.

Can you please share some suggestions regarding this?