QuantumATK Forum

QuantumATK => General Questions and Answers => Topic started by: abhi005 on November 17, 2014, 15:55

Title: boundary condition
Post by: abhi005 on November 17, 2014, 15:55

i am making an cntfet.i dont know which boundary condition are best for my device.i have choosen neuman but the multigrid residual error is infinite. so i choose drichlet for B bottom and neumann for all others then error is less but quite large(e+000).so i choose drichlet for B bottom A front A back and neumann for B top.now error is quite small(e-010).so my question is

1.Does drichlet condition can be consider for device with gate?

Title: Re: boundary condition
Post by: Umberto Martinez on November 25, 2014, 10:30

The correct boundary condition is to use Neumann boundary conditions along the direction where the gate is defined.
However, Neumann boundary conditions are more difficult to converge, thus, a good strategy is to converge with Dirichlet boundary conditions and then use the converged state to start up the Neumann boundary condition calculation.

Title: Re: boundary condition
Post by: abhi005 on December 23, 2014, 14:29

thanks for your reply.i am not getting how it can be done. Can you please tell me how can I converged my geometry with one boundary condition and then how can I use this converge state for further calculations.

Title: Re: boundary condition
Post by: Anders Blom on January 4, 2015, 00:07

http://quantumwise.com/publications/tutorials/item/503-saving-time-by-initializing-a-calculation-from-the-converged-state-of-another