Inquiry about Parameter Settings for H2O Phase Diagram Calculations in Calphy

[stargolike]

Dear Calphy Development Team,

First of all, thank you for developing this valuable computational tool. I'm currently attempting to use Calphy for H2O phase diagram calculations, but I have some questions regarding the parameter settings.

I would like to ask if the development team has conducted any testing of Calphy for H2O phase diagram applications? If you have relevant testing experience, could you please provide some reference parameters or best practice recommendations?

Thank you!

[stargolike]

Actually, my primary current concern lies with the setting of n_switching_steps. Given that I am using a custom input structure, I would like to know how I can determine whether convergence has been achieved for this parameter (i.e., n_switching_steps), and how it should be properly configured.

[srmnitc]

@stargolike sorry for missing the previous question! H2O system has not been tested as far as I know.

For switching steps, I would simply calculated the free energy with different values of n_switching_steps. When you plot the calculated free energy against the switching steps, it should give something like Fig. 1(b) here: https://arxiv.org/pdf/2107.08980

[stargolike]

Thank you so much for your comprehensive reply and for taking the time to address my question in detail. I truly appreciate the guidance you provided on the phase transition issue.

Regarding the n_switching_steps parameter, I have encountered an issue—I am not sure if you have experienced it before—where setting it too long leads to a phase transition. In my case, the initial structure was ice, configured as follows. When I set n_switching_steps: 200000, melting occurred. How can I avoid this situation? Alternatively, should the initial temperature not cross phase boundaries? Below are my key parameters:

mode: ts  
temperature: [350, 200]
temperature_high: 350
pressure: 0
lattice: 2304_solid.data
reference_phase: solid
repeat: [1, 1, 1]
n_iterations: 1
n_equilibration_steps: 0
n_switching_steps: 200000
n_print_steps: 1000
tolerance:
  solid_fraction: 0

Additionally, when I set temperature: [200, 350], the two free energy curves do not intersect, which seems to be the reason behind my current parameter choice. I am seeking your insights on this matter—do you have any thoughts or suggestions regarding this behavior?

[srmnitc]

@stargolike , I would choose temperature: [t_lower, t_higher]; if they do not intersect like you mentioned, the first step would be to see if the solid or liquid is lower in free energy. If it is the solid, and the liquid and solid free energies are not parallel to each other, but coming together, then you are way below melting temperature. You could simply take the free energies, make a fit and extrapolate it to find the guess melting temperature, and then rerun the calculations.

[stargolike]

@stargolike , I would choose temperature: [t_lower, t_higher]; if they do not intersect like you mentioned, the first step would be to see if the solid or liquid is lower in free energy. If it is the solid, and the liquid and solid free energies are not parallel to each other, but coming together, then you are way below melting temperature. You could simply take the free energies, make a fit and extrapolate it to find the guess melting temperature, and then rerun the calculations.

Thank you so much for your detailed reply. If I have understood correctly, you are suggesting that we perform multiple fe mode calculations within the temperature range of [t_lower, t_higher] to replace the ts mode calculation. This is truly an excellent suggestion, and I will immediately give it a try.