Why "screened"? Because in a metal, the Coulomb interaction dies off quickly. HSE06 introduces a screening parameter ($\omega$) to cut off the long-range HF exchange, making it computationally feasible for periodic systems.
In this post, I’ll break down what HSE06 is, how to set it up in VASP, and when it’s actually worth the pain. HSE06 is a screened hybrid functional. It mixes 25% exact (Hartree-Fock) exchange with 75% PBE exchange at short range, while keeping PBE correlation. hse06 vasp
KPOINTS: Automatic generation 0 Gamma 4 4 2 0 0 0 Why "screened"
If your system has less than 50 atoms and you care about the band gap to 0.1 eV accuracy, pay the cost. If you're studying a metal or a giant interface, stick with PBE+U. Have you had success (or nightmares) running HSE06? Let me know in the comments below. And yes, your SCF will oscillate on the first try—check your mixing parameters. In this post, I’ll break down what HSE06
Enter (Heyd-Scuseria-Ernzerhof). This hybrid functional has become the gold standard for "affordable accuracy" in solid-state physics. But let’s be real—it comes at a computational cost.
ALGO = Damped # Damped algorithm (often more stable than Normal) TIME = 0.4 # Mixing time (increase from default 0.1) BMIX = 0.0001 # Small mixing parameter AMIN = 0.01 # Avoid Pulay collapse If still failing, try ALGO = All (fast but memory hungry) or IALGO = 53 (very stable but slow). | System | PBE wall time | HSE06 wall time | Memory | | :--- | :--- | :--- | :--- | | Si (8 atoms, 6x6x6 kpoints) | 2 min | 20 min | 2x | | TiO₂ (12 atoms, 4x4x4) | 5 min | 1.5 hours | 3x | | NiO (8 atoms, 8x8x8) | 3 min | Fails / 4 hours | 5x |
