parameter settings about GNR on hexagonal boron nitride floating gate devices

[emma]

Dear QuantumWise Staff,

I am trying to simulate a GNR on hexagonal boron nitride floating gate device, but have several confusions about configuration, optimization and IV curve calculation settings. Would you please offer me some suggestions or relevant quantumwise tutorial links?

1.  configuration.  The simulation structure is set as follows. I have set the distance between GNR and hBN 3.3A as publications mentioned, but have no idea what value should be set for the distance between hBN and the gate. Personally, I would like to make them contact, which means the distance will be zero or 0.5A. But is it recommended? Or is there any publication has offered theory to guide us choose the value?


2.   optimization and IV curve calculation. DFT is the most reliable method but takes extremely long time. I have tried Extended Huckel for both optimization and IV calculation for GNR floating gate devices and got good results. But I am not sure if Extended Huckel is suitable for devices which contain boron and nitride atoms. It will be great if quantumwise tutorials can be posted for it.

Thank you for help!

[Jess Wellendorff]

1. Full contact between hBN and te gate+dielectric is probably not a good idea, but 0.5 Ang seems reasonable. What you need to remember is that the dielectric and gate, as implemented in ATK using ideal boxes ("regions" ), simply modify (or fix) the electrostatic potential inside those boxes. You should therefore avoid explicit overlaps between the boxes and any atoms in the configuration.

2. Which degrees of freedom are you optimizing? GNR and hBN lattice constants? GRN-hBN distance? If the former, that should be done before setting up the device configuration, using minimal unit cells (bulks). If the latter, perhaps minimizing the DFT total energy wrt. the distance is more efficient than GeometryOptimization?

[emma]

Hi Jess

Thank you for your reply.

No problem for the 1st question, but for the 2rd one, I would like to get the most stable system and then calculate its IV curve. I ran optimization for GNR on hBN devices before, with DFT, which costed me almost three weeks and offered similar IV curves compared to the non-optimized one. This time I have added a gate in, but based on previous experience, I am hesitating on optimization with DFT which is really time-consuming, even though accurate. That is why I am wondering if I can skip the optimization step or run optimization with Extended Huckel. And I also prefer Extended Huckel for IV curve calculation. Now I am trying to find relevant publications to support my idea.

[Jess Wellendorff]

OK, I see. My suggestion is then to first establish if Extended Huckel geometry optimization is useful in your case. Create a much smaller version of your system ( a "toy" system) such that DFT optimization is feasible. Then compare that to Externded Huckel optimization to see if the latter is reasonable.

[Anders Blom]

Extended Huckel geometry optimization does not work, since there is no repulsive pair potential in our Hucke models.