Skip to content

7.2 Independent-particle optical spectra with LOPTICS

These are unexecuted teaching inputs and starting settings to test. Original figures are schematics, not computed results. Use licensed VASP and PAW data, replace every placeholder, record the executable version and validate convergence.

7.2.1 Model, units and provenance

Use eV for energy, Å for length and eV/Å for force; 1 kbar = 0.1 GPa. State normalization per atom, molecule, primitive cell or simulation cell. Record PAW identifiers, release, ZVAL, ENMAX and permitted hashes; never redistribute POTCAR. SCF convergence addresses the chosen electronic problem; convergence of the target property requires separate tests.

 Independent-particle optical spectra with LOPTICS — original conceptual schematic

Original schematic. Curves explain concepts; blank data areas await verified learner results. No calculation is claimed.

7.2.2 Worked case: procedure, interpretation and checks

What the approximation contains

Light can induce vertical transitions between occupied and empty states. “Vertical” means that photon momentum is neglected relative to crystal momentum, so the initial and final Bloch states share k. The spectrum depends both on transition energies and on optical matrix elements; a large joint density of states does not ensure a strong allowed optical transition. LOPTICS gives a practical independent-particle electronic spectrum. It does not turn a ground-state DFT calculation into a full experiment including bound excitons, phonon-assisted indirect absorption, disorder, and temperature broadening.

Begin with a converged, small bulk semiconductor. A polar semiconductor can have an additional infrared lattice response, which this electronic interband workflow does not supply. For metals, intraband/free-carrier effects need separate attention and should not be inferred from a semiconductor-style interband plot. For strongly anisotropic or magneto-optical systems, keep the full tensor and solve the appropriate electrodynamics rather than averaging entries without justification.

A staged input strategy

First converge the ground state on a regular mesh. Then compute sufficiently many empty states and the optical response on a denser, property-converged regular mesh. A semilocal insulating starting delta is:

SIGMA and CSHIFT here are in eV. The smaller Gaussian width and larger Lorentzian shift are illustrative, not an experimental linewidth assignment. Increase NBANDS systematically and retain enough high-energy transitions for the real part to stabilize. The real and imaginary dielectric parts are connected by a frequency integral, so high-energy truncation can affect the low-energy response. LOPTICS, CSHIFT, OMEGAMAX

A useful order is to establish an energy window, converge empty states, converge k sampling, refine the frequency grid, then test line broadening. Do not change all four at once. A narrow broadening on a sparse k mesh reveals discrete transitions, not necessarily physical fine structure. Conversely, a broad smooth spectrum can hide poor sampling. LOPTICS with fixed occupations ISMEAR=-2 is unsupported in the documentation checked here; tetrahedron combinations require VASP 6.3 or newer. The Gaussian branch above avoids that version dependency. LOPTICS compatibility

From dielectric tensor to observables

Inspect the frequency-dependent real and imaginary dielectric tensors in the documented output blocks of OUTCAR/vasprun.xml or supported HDF5 reader. Keep the component ordering supplied by the file; do not assume every parser uses the same order. The dielectric tensor is dimensionless, and the horizontal energy convention is normally photon energy in eV, not frequency in Hz.

For an isotropic, nonmagnetic medium, or an appropriately justified principal optical response, write \(\varepsilon=\varepsilon_1+i\varepsilon_2\) and \(n+i\kappa=\sqrt{\varepsilon}\) with the passive-medium branch. Then

  • \(n=\sqrt{\frac{\sqrt{\varepsilon_1^2+\varepsilon_2^2}+\varepsilon_1}{2}}\);
  • \(\kappa=\sqrt{\frac{\sqrt{\varepsilon_1^2+\varepsilon_2^2}-\varepsilon_1}{2}}\);
  • \(\alpha=2\omega\kappa/c\), with ω in rad/s, yielding α in m⁻¹;
  • for photon energy E in eV, use \(\omega=E/\hbar\) with ℏ in eV·s;
  • normal-incidence reflectance from vacuum is \(R=\frac{(n-1)^2+\kappa^2}{(n+1)^2+\kappa^2}\) for this scalar setting.

Divide m⁻¹ by 100 to obtain cm⁻¹. These formulas require their stated optical geometry; a general anisotropic tensor cannot be replaced by the square root of an arbitrary component. An absorption coefficient is not the same as a finite sample's measured absorbance or absorptance, which also depend on thickness, interfaces, and interference. Optical-response definitions

Validation and exercise

For a semilocal insulator, compare the zero-frequency no-local-field electronic limit against the matching DFPT approximation, not against a relaxed-ion dielectric constant. This helps diagnose insufficient empty states. Check that equivalent crystal directions agree, and that passive diagonal absorption does not develop unexplained negative weight. A shifted optical onset may reflect a DFT gap error, but applying a rigid scissors correction is an additional model assumption, not a repair of missing excitons or matrix-element errors.

Slab ε values depend on the simulation volume and vacuum height. Report a justified two-dimensional polarizability or sheet response when appropriate instead of comparing raw vacuum-diluted ε to a bulk experimental number. Exercise: produce three spectra with increasing NBANDS at fixed mesh, then refine the mesh at fixed broadening. Report stability of a chosen feature and ε1(0), and only afterward compare two broadening values. Keep computed curves distinguishable from schematic teaching curves.

7.2.3 Unexecuted inputs and analysis scaffolds

These are unexecuted teaching inputs and starting settings to test. Original figures are schematics, not computed results. Use licensed VASP and PAW data, replace every placeholder, record the executable version and validate convergence.

7.2.3.1 Input block 1

NSW      = 0
IBRION   = -1
EDIFF    = 1E-8
LREAL    = .FALSE.
LOPTICS  = .TRUE.
ISMEAR   = 0
SIGMA    = 0.01
CSHIFT   = 0.10
NEDOS    = 3001
# NBANDS = <integer chosen and converged for the spectral range>
# OMEGAMAX = <tested maximum energy in eV, if explicitly set>

7.2.5 Technical sources