QuantumATK Forum

QuantumATK => General Questions and Answers => Topic started by: leedongchan on January 19, 2026, 08:06

Title: Issue with relaxation convergence in Charged Point Defect calculation
Post by: leedongchan on January 19, 2026, 08:06

I am currently performing Charged Point Defect (CPD) calculations on Al-doped HfO2 supercells.

I have prepared optimized pristine cells for two compositions: Hf54Al10O123 (Hf-rich) and Hf32Al32O112 (Al-rich). While the calculation works fine for the Hf-rich composition, the relaxation for the Al-rich composition (Hf32Al32O112) fails to converge even after reaching 300 steps.

Here are the calculation details:

Calculator: GGA-PBESol

System Size: 176 atoms

K-point Sampling: 1x1x1

Electron Temperature (Broadening): 300 K

Optimization Settings: Max step length = 0.07 Å

Iteration Control Parameters:

Damping factor = 0.05

Number of history steps = 6

Effective rank fraction = 0.06

I have also verified that the forces are stable and do not fluctuate significantly when tested with 1x1x2 and 2x2x1 k-point grids.

Could you please provide any suggestions on how to resolve this convergence issue? Any help would be greatly appreciated.