Hello Everyone,
I am a beginner to DFT calculation using CASTEP Academic Version. I don't know how to choose cutoff energy and k points to unit cell, 221 supercell and 331 supercell for 2H MoS2 Monolayer. Please help me to proceed my work in a correct way.
You can do all the convergence tests in the unit cell, to save computational time. You need to focus on converging the properties you're interested in or, if those properties are very demanding computationally, a good proxy property. The stress on the unit cell is quite a good proxy for lots of properties, because it is quite sensitive to error.
Once you have decided on your convergence property, you need to test different cut-off energies and k-point sampling to determine which settings are converged enough. Provided your cut-off and k-point sampling are reasonable, you should be able to test these properties independently.
A reasonable starting point for the cut-off energy is the value for "fine" precision. You can select this in your param file:
basis_precision : fine
A reasonable k-point sampling for many situations is a k-point spacing of 0.05 (in units of 2pi/A). In the cell file, you'd put:
kpoint_MP_spacing 0.05
(MP = Monkhorst-Pack)
With those initial guesses, you can now try higher and higher cut-off energies and see at which point your property converges. Then you go back to the "fine" setting for your cut-off energy and improve your k-point sampling instead, to see how your property converges with k-point sampling.
The CCP-NC consortium have created a tool called "castepconv" to help with this process, which you might find useful:
https://github.com/CCP-NC/castepconv
Finally, you use the converged cut-off energy with the converged k-point sampling for your actual research calculations.
A few things to remember:
1. If you change your pseudopotentials, you need to re-check your cut-off energy convergence. This might be because you've added a new chemical element, or it might be because you wanted to change pseudopotential.
2. When you simulate a supercell, the cut-off energy is unchanged but the k-point grid changes due to the change in real-space simulation cell. If you tell CASTEP to use a k-point spacing, rather than telling it the actual grid, then CASTEP will generate the correct grid in each case.
3. If you change any of the bonding in the system, e.g. adding a surface, interstitial, interface, vacancy etc. then you need to re-check your k-point convergence.
- Badges
- Science topic