Non-convergence issue in phonon calculations of monolayer HfS₂ when increasing ecut and ngkpt

Running Abinit Many-body theory
1

Dear Abinit developers and users,

I am performing phonon calculations for a monolayer HfS₂ system using ABINIT, and I am facing a convergence issue that I cannot resolve.

When I use low (experimental) values for ecut and ngkpt, ABINIT completes both the ground-state calculation and the phonon (DFPT) calculation without any problem.

However, when I increase ecut and ngkpt to obtain more accurate results, the SCF cycle fails to converge, and the calculation stops with an energy non-convergence error. This happens consistently whenever I try to use higher, more reliable values of these parameters.

I am attaching my input file (.abi) and the log file showing the convergence failure.

I would be grateful for any advice on:

  1. Why the SCF becomes unstable when increasing ecut or ngkpt,

  2. Whether this behavior is expected for 2D materials such as monolayer HfS₂,

  3. Possible strategies or recommended parameters to stabilize the phonon calculation.

Thank you in advance for your help.

Best regards,
khatir

log.txt (203.6 KB)

trf2_1.in (4.9 KB)

2

Dear Khadir,

This section of the forum is more dedicated to problems concerning many-body theory’s problem. I am not an expert of DFPT. After a quick look at your input/output, you could try to change the SCF algorithm you are using and use the default ones (iscf 17 or 7) instead of the conjugate gradient. Also if your system is metallic, occopt = 3 could be better than 4.

Best regards,

Fred.