4.4 Charge density, charge differences, and ELF
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.
4.4.1 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.
Original schematic. Curves explain concepts; blank data areas await verified learner results. No calculation is claimed.
4.4.2 Worked case: procedure, interpretation and checks
Three questions that need three quantities
Electron density asks where electrons reside. A density difference asks how a density changes relative to a defined reference. The electron localization function (ELF) is a dimensionless localization indicator constructed from orbital information; it is not another name for electron density. Bright regions in these three images answer different questions. None by itself is a direct map of bond strength.
For a first charge-difference exercise, choose an adsorbate A on a surface B, with a converged combined geometry AB. Define \(\Delta n(\mathbf r)=n_{AB}(\mathbf r)-n_A(\mathbf r)-n_B(\mathbf r)\). In this definition, both fragments retain the positions they have in AB. The result isolates electronic rearrangement for that frozen geometry. If you separately relax A and B first, you answer a different question that includes geometry changes.
A reproducible three-run procedure
- Copy the relaxed AB geometry to a static directory and converge its density. Keep the lattice, vacuum, FFT grid, cutoff, functional, PAW choices, k sampling, and relevant electrostatic corrections fixed for all three runs.
- Create fragment A by removing B atoms without changing the cell or remaining coordinates. Create fragment B analogously. Construct each licensed POTCAR in its actual remaining species order. Preserve the same PAW dataset for every retained species.
- Use explicitly matched fine-grid dimensions NGXF, NGYF, NGZF in all calculations. Read the grid actually written; never assume equal ENCUT alone guarantees identical grids. Let each fragment converge electronically in its intended charge and spin state.
- Write CHGCAR for each static run. Parse the first density block correctly, check its grid and normalization, then subtract pointwise. Treat extra magnetization and PAW augmentation blocks separately; blind subtraction of every number in three text files is invalid.
The ordinary CHGCAR grid stores a positive electron-number density in a volume-scaled form. With raw grid data d and cell volume Ω in ų, use \(n=d/\Omega\); its discrete electron count is sum(d)/Ngrid. Many libraries already normalize the data, so inspect their convention before dividing again. This integral is the most useful protection against a visually plausible normalization error. CHGCAR format, fine FFT grid
Quantitative interpretation
For neutral fragments whose electron counts sum to AB, the integral of Δn should be approximately zero. Positive Δn means electron accumulation, not positive electric charge; the electronic charge-density change has the opposite sign. If you intentionally chose charged fragments, write their electron-count balance and expected nonzero integral before interpreting the image. A compensating background in a periodic charged calculation is another reason to avoid casual claims about isolated charge transfer.
For a slab with in-plane area A and orthogonal surface-normal coordinate z, calculate the planar average \(\overline{\Delta n}(z)=\frac1A\int_A\Delta n(x,y,z)\,dx\,dy\). Its unit is electrons/ų. The cumulative redistribution \(\Delta N(z)=A\int_{z_0}^{z}\overline{\Delta n}(z\prime)\,dz\prime\) is an electron count. A plateau may support a stated spatial partition, but a division plane through a bond is still a choice. Clearly define the starting plane and periodic wrap. For a nonorthogonal cell, transform coordinates and area/volume elements correctly before calling a plot a z average.
ELF branch, caveats, and exercise
ELF is a separate static calculation on the converged structure. Add LELF=.TRUE. and, following the documented implementation requirement, explicitly use NPAR=1; remove conflicting manual band-parallel settings and verify compatibility with the installed build. Read ELFCAR with its dimensionless scale. In the usual convention ELF near one indicates strong localization, while one-half is the homogeneous-electron-gas reference; a high value may describe a lone pair, a bond region, or a localized shell. Do not assign a universal bonding threshold. LELF, ELFCAR
Choose a plane through chemically relevant atoms, show atomic positions, and keep the same color scale when comparing systems. Plot positive and negative density-difference isosurfaces at equal magnitude and state that magnitude. Test a second isovalue: an apparent bridge that vanishes immediately should not be used as sole proof of bonding. Exercise: produce a difference isosurface, a planar-average curve with its integral check, and an ELF section. Write one sentence describing what each quantity can establish and one describing what it cannot.
4.4.3 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.
4.4.3.1 Input block 1
# Charge-map static delta, applied consistently to AB, A, and B
NSW = 0
IBRION = -1
LCHARG = .TRUE.
LWAVE = .FALSE.
PREC = Accurate
# NGXF = <identical positive integer in all three runs>
# NGYF = <identical positive integer in all three runs>
# NGZF = <identical positive integer in all three runs>
4.4.4 Related learning paths
- 1.1 Four input files, one physical question
- 1.2 A convergence laboratory with an error budget
- 4.3 DOS and PDOS
- 4.5 Bader populations and charge conventions