QuantumATK Forum
QuantumATK => General Questions and Answers => Topic started by: yang su on May 16, 2013, 20:48
-
Hi,
I am currently working on the calculation of a 2D MoS2 device. As Ander's recommendation, in my script I first do a geometry optimization before calculating the lddos. And I tried to use GGA for the calculation. The program aborted and I checked the log and I see a non convergence in the simulation and the error message is shown below. I suspect that is because I am using quasi newton for the optimization and I am wondering how I should change the restricts. I am also thinking that for a device simulation if it is proper to use 11 k-point sampling and 60 Hartree as pseudo potential since that will be very time-consuming. The settings worked pretty well for the lattice simulation with just the basis of A moly atom and a sulfur atom but it is not working for the device. Can I know what can I do to modify the script to have the simulation well set? Thank you.
File "/home/koesters/suxxx172/suxxx172//ATKSIMS/shellscripts/20130505_lddos.py", line 488, in <module>
optimizer_method=QuasiNewton(),
File "./zipdir/NL/Dynamics/Optimization/OptimizeGeometry.py", line 229, in OptimizeGeometry
File "./zipdir/NL/Dynamics/Optimization/OptimizeGeometry.py", line 338, in runSimultaneousRelaxation
File "./zipdir/NL/Dynamics/Optimization/OptimizeGeometry.py", line 263, in runRelaxation
File "./zipdir/NL/Dynamics/ASEAtomsInterface.py", line 173, in get_forces
File "./zipdir/NL/Dynamics/ASEAtomsInterface.py", line 97, in _update
File "./zipdir/NL/Calculators/LCAOCalculator/Analysis/FastEnergyForcesStress.py", line 28, in energyFunction
File "./zipdir/NL/Calculators/LCAOCalculator/Analysis/FastEnergyForcesStress.py", line 65, in _checkAndUpdate
File "./zipdir/NL/Calculators/LCAOCalculator/Analysis/FastEnergyForcesStress.py", line 90, in _update
File "./zipdir/NL/Calculators/LCAOCalculator/Analysis/EnergyForcesStress.py", line 72, in calculateLCAOEnergyForcesStress
File "./zipdir/NL/Calculators/LCAOCalculator/Analysis/EnergyForcesStressUtilities.py", line 19, in setupLCAOEnergyForcesStressCalculator
File "./zipdir/NL/Calculators/LCAOCalculator/Analysis/EnergyForcesStressUtilities.py", line 344, in setupBackengineConstraints
NL.ComputerScienceUtilities.Exceptions.NLValueError: Stress can only be calculated in the z-direction for devices.
-
Update: I don't understand the error message, I have to check it.
However, this part remain: I would NOT recommend optimizing the device in this way, either for stress or force. It will take very long time, for little gain. The point I was making in our earlier communication was that you should optimize the basic MoS2 monolayer before constructing the device. That's a quick calculation.
It is not expected that the geometry is influenced in any noticeable way by the gates.
-
Ok, you need to roll back a little bit. The script you attached is wrong, it will fail, but not with the error message you showed (I think, at least).
In the script you attached, you attempt to optimize a "bulk_configuration", but no such configuration is defined in the script.
Nevertheless, you should go back to my original point about pre-optimizing the bulk configuration. This is a completely different script/calculation, which produces an output - the optimized bulk cell. (That's the script I sent by email.) The result of that calculation tells you the MoS2 geometry - and you use that to build the device. There is no additional need to optimize the device.
It could be interesting to figure out where the error message comes from, but probably it's from a simple mistake caused by an erroneous manual change to an automatically generated script.
-
Thank you so much Anders..I got it and I will try to build the device that way. However should I use some specific calculator like GGA or MGA, LCAO or huckel for the device simulation?
-
The Huckel basis sets for MoS2 are no good, although I have seen people developed new parameters for the material. We can look into that a bit later - for now I would recommend DFT (LCAO) with LDA which gives a slightly better band gap than GGA.