Generally speaking, reducing the mix_charge_amp in your param file should stabilise the convergence. This can also be set in the Materials Studio GUI from the SCF tab of the "Electronic Options" dialogue box.

In this particular case, the calculation is generally well behaved, but it seems like the geometry for the last step is causing some numerical instability. This can happen if the trial geometry causes atoms to move too close to each other, for example. Have you tried restarting the calculation as a "continuation"? This will pick up where your calculation left off in terms of geometry, wavefunction and density, but will reset the density mixing model which might help it to converge.

I notice that you're using CASTEP 17, which is nearly 5 years old. I generally recommend updating to the newest version, which is free for academic research:

https://licenses.stfc.ac.uk/product/castep

Finally, I recommend you sign up to the CASTEP email list:

https://www.jiscmail.ac.uk/cgi-bin/webadmin?A0=CASTEP

All the best,

Phil Hasnip

(CASTEP developer)

Dear Prasad,

generally, CASTEP convergence is slow. However, convergence failure is usually due to mistakes in specifying the system. This includes:

• The atomic co-ordination are too close.
• Specifiying fix_occupancy=true when the system is really metallic.
• Specifying spin=0 if the system is really magnetic.

In addition, if you use the density mixing algorithm, there exist several parameters that you can use to improve convergence:

• mixing_scheme There are two main density mixing schemes, one due to Broyden and one to Pulay. The default is Broyden, If your system is not converging try Pulay.
• The default value is the plane-wave cut-off energy ; the maximum sensible value is given by cut_off_energy * (fine_grid_scale)2.mix_cut_off_energy The density mixing is actually performed in reciprocal space, and only plane-waves whose energy is less than mix_cut_off_energy are mixed.
• mix_charge_amp (and mix_spin_amp) The density mixing takes a fraction of the proposed new density and the old density, and this fraction is set by mix_charge_amp. The possible values of mix_charge_amp range from 0 to 1. Use values between 0.05 and 0.2 to improve convergence.

Thank you Philip James Hasnip sir for your nice suggestions. I tried the following things as per your advice:

1. I changed the Charge mixing amplitude from 0.5 to 0.1.

2. I restarted the calculations from the continuation.

It helped me a lot. Most of my models were converged successfully and quickly. However, one of the model calculation was failed to converge. When I changed the vacant position, the calculation was converged but we need to calculate energy at that particular vacant position for consistent results with my other models' results.

I am attaching the result files for more details. Kindly, let me know what are other things can be done?

Thank you Muhammad Hamza El-Saba sir for your advice. I have already used Pulay in the density mixing scheme and changed the charge mixing amplitude from 0.5 to 0.1. However, I will try to do it as per your suggestions and update you as early as possible. I have noticed that by changing the charge mixing amplitude, the final result is changed slightly from the previous calculations, even though results were converged in both cases for the same model. In that case, what would be the optimum charge mixing amplitude value, or which result is more preferable?

Manohar Prasad as long as the results are both within your convergence tolerance, there is no particular reason to choose one over the other. In other words, if the difference is important then you need stricter convergence tolerances!

Okay, thanks a lot Philip James Hasnip sir for the quick reply and your suggestions are really helpful for me!