Unable to replicate Graphene junction device tutorial in ATK 13.8

[cvnu]

Hello everybody,

Tried to replicate the graphene junction tutorial :http://quantumwise.com/documents/tutorials/latest/GrapheneDevice/index.html/ in ATK 13.8, there are no errors in execution but the gate voltage is having negligible effect on the device. At Vg=-1V Neither does the transmission spectrum shift, nor can we observe the expected ElectrostaticDifferencePotential (deltaVH range obtained is 1.3e-8 to 0.054 eV).  All steps mentioned in the tutorial have been followed explicitly. Upon further examination, if Vg=-10V, bands are shifted up by 2.4eV and ElectrostaticDifferencePotential deltaVH range obtained is (6.5e-7 to 0.37 eV) . I have attached the device configuration, and relevant plots. Can anyone please help me out? 

[kstokbro]

We are working on updating the tutorials to ATK 13.8,
Most likely the problem is that you are running the tight binding model non-selfconsistently, which is the default.

Select selfconsistent model in the scripter. Or set
iteration_control_parameters=IterationControlParameters(),
in the calculator.

[cvnu]

Thank you for the quick response. The self-consistent model is set. See line 357 in the previously attached device configuration. If the simulation was not self-consistent i would not have got band shifting at Vg=-10v. The problem is that gate voltage is not having the same effect as described in the tutorial, even when the procedure is followed exactly.  I actually came across this problem when trying to backgate a CNT, so i tried the graphene tutorial and even that failed as mentioned.

[kstokbro]

We are looking into it.

[kstokbro]

The problem is related to that we updated the SCF parameters for Cerda in the ATK13.8 release. The Cerda parameters are originally developed for Non-SCF use. In ATK 13.8 we implemented, that the Cerda parameters give the same results SCF and NON-scf for the reference structure that was used for fitting the parameters. In C2H4 there is a charge transfer from C to H, and the C2H¤ hydrogen has a reference charge of 1.2. When this parameter is mixed with the Graphene C parameter, which has a reference charge of 4, the structure become charged and the transmission spectrum gets wrong.
We will release a 13.8.1 where we will check for such inconsistencies in parameters and give an error.
To see the reference charges you can print the script out using show defaults.

The solution in this case is to use the Hoffmann Hydrogen parameters, which has a reference charge of 1. We are currently working on updating the tutorials for 13.8 and will make this change in the tutorial.

Thanks for you feedback.

[Dipankar Saha]

Dr Kstokbro

What does it mean??... Are you saying that any simulation done on ATK 13.8 (or, in RC version of it) for Graphene using cedra.carbon[graphite] in the basis set of elements (as is the case in ATK-SE Extnded Huckl) will be wrong??

[Dipankar Saha]

Sir, if I use ATK-SE  Slater Koster model with DFTB-CP2K basis set.... Then with the default calculator (ATK 13.8.0), can we do the calculations on a carbon based system having defects (and, where the dangling bonds are passivated with Hydrogen)?? Is it(default one) the "Pi-model"

[kstokbro]

You can use the Cerda Graphite paramters, they are perfectly fine. However, for the Hydrogen passivation use the Huckel Hofmann parameters.

You can also use the DFTB-CP2K parameters

[Dipankar Saha]

Thanks kstokbro ....

But what I specifically wanted to know is....whether a semi empirical Slater-Koster model (with DFTB,CP2K basis set) can be (rather, should be) used to find out the transmission property of any defected (where the periodicity has been perturbed ) carbon based system ??

[Anders Blom]

It's very difficult to answer such general questions, and often the only way to really know is to at least make one comparison for the particular system of interest. But the DFTB models are quite adept at handling non-ideal situations, and so are the extended Huckel models in ATK, at least if you use them selfconsistently. Tight-binding is less so, they are more hardcoded for the periodic situations with the correct number of bonds in the expected angles etc, but can still often be used to get a general idea.

Our general experience with the CP2K basis sets is relatively poor - if you want to use DFTB I would strongly suggest to get the basis sets from http://dftb.org -  they are free for academic users, you just have to register first.

[Dipankar Saha]

Dr. A. Blom,

Thank you sir for the information regarding the website....... I could be of a great use....

Thanks & Regards__

Dipankar Saha