Hi,
I want to calibrate the TCAD Sentaurus simulator in order to match the measured AlN-GaN HEMT DC IV characteristics and the simulated one. I'm using a simple drift diffusion model. For the mobility, I would like to use either constant mobility model or doping dependence (Arora Model). In the literature, there are no arora model parameters available for the GaN, AlN and AlGaN materials.
For the High field saturation, Caughey Thomas model is used.
mu_highfield = mu_lowfield / ( 1 + (mu_lowfield E / vsat)^beta )1/beta
mu_lowfield^(-1) = mu_dop(mu_max)^(-1) + mu_Enorm^(-1) + mu_cc^(-1)
I have seen in some of the thesis, mu_lowfield, vsat, beta are used as a fitting parameters to match with the expt. characteristics. But in TCAD, I think, mu_lowfield is calculated automatically using inbuilt function.
Please let me know, how can I use these as a fitting parameters. Any other useful information related to the TCAD calibration ? Any good documents ?
Please do reply. Thanks in advance.
The simulators are "calculators". Therefore, simulators cannot be calibrated.
The models can be calibrated by either changing equations or values of parameters in these equations, or both. In TCAD simulators, the parameters are material properties and geometry, while the biasing goes in boundary conditions.
Use literature data for initial values of parameters. Vary the values of parameters in the model that you use in the simulation, until the curves from simulation match reasonably the experimental curves.
In general, the model calibration is optimization problem. In TCAD and circuit simulation, the optimization usually is curve fitting by constraints for model equations and ranges for values of model parameters. However, the diversity of models is large. Therefore, there is no universal rule for calibration of TCAD model. Gradient and Monte-Carlo methods can be used for variation of the parameters, if available in your simulator. If so, read the manuals and application notes for your simulator. Journal publications usually report final values, but do not describe the steps in obtaining of optimum values for model parameters.
The bottom line is that the model calibration is sole due to experience of the researcher. So, do not expect much useful advice from newsgroup asking and searching. Nobody has your device and data.
Many thanks for your reply Marinov. It seems it is difficult to calibrate the TCAD model for my device. TCAD model has got of lot of parameters to adjust. Sometimes, non-physical values yield the good match with the experiment.
Many thanks for your reply Anurag.
I know the drift diffusion model does not account for the velocity saturation and impact ionization effects. Is that what you mean .?? But if use hydrodynamic models, convergence is a major issue.
Thanks Anurag..
Yes. In my structure, there is no doping at all. So I use the Constant Mobility model and High Field saturation model (Caughey Thomas) for my calibration.
In my humble opinion, Devices may be caliberated depending upon the tools capabilities.. How much efforts have been put by the tool developers in providing various models for a particular device? You will have to understand all the parameters provided in the tool and their modeling context for your chosen device (HEMT in your case). However, modelling in MATLAB is the best option as it gives whole lot of control for caliberating the devices.
I have gone through thesis "Physics-Based TCAD Device Simulations and Measurements of GaN HEMT Technology for RF Power Amplifier Applications".
I am unable to solve same issue. Please guide me how to call "mu_lowfield" value for GaN device..??
- Badges
- Science topic
- Similar topics
- Geoscience
- Asia
- India