Setting up the basis set

Energy cutoff

The basis set, as defined by the energy cutoff value, is one of the most important parameters that determine the accuracy (and the cost) of a CASTEP calculation. The easiest way to set this value is by using the Energy cutoff option on the Electronic tab of the Calculation dialog. If you use this method, the cutoff value which is optimal for a given level of the total energy convergence will be determined, based on the information in the pseudopotential data files.

Alternatively, you can specify a custom energy cutoff value on the Basis tab of the Electronic Options dialog. Custom energy cutoffs may be required in a number of cases, for instance, when modeling a molecule-surface interaction, for example CO on a Pt surface. In such situations, the protocol for the calculation of the binding energy includes separate CASTEP runs for the isolated molecule, the clean Pt surface and finally the CO-Pt complex. The basis sets for the systems containing oxygen will be determined by the convergence properties of the oxygen potential, the hardest element out of O, C, and Pt. In order to achieve maximum error cancellation, it is necessary to set the energy cutoff for the clean Pt surface calculation to the same value as in the other two runs. Therefore, an automatic selection of the energy cutoff is not totally appropriate in this case.

To change the energy cutoff quality

  1. Choose Modules | CASTEP | Calculation from the Materials Studio menu bar.
  2. Select the Electronic tab.
  3. Set the Energy cutoff quality using the dropdown list.
  4. The energy cutoff value that corresponds to the selected quality is displayed next to the dropdown list.

To set a specific energy cutoff value

  1. Choose Modules | CASTEP | Calculation from the Materials Studio menu bar.
  2. Select the Electronic tab.
  3. Click the More... button to access the Electronic Options dialog.
  4. Select the Basis tab.
  5. Check the Use custom energy cutoff option.
  6. Modify the energy cutoff value.

FFT grid

The size of the FFT grid which is used to represent wavefunctions, charge density and potentials is determined by the energy cutoff value and the reciprocal lattice vectors.

The ideal FFT value is determined by the requirement that there be no wrap-around error in the representation of wavefunctions and potentials (Payne et al., 1992).

This value corresponds to the maximum G-vector of the FFT grid, Gmax, which is exactly twice the radius of the reciprocal space sphere determined by the energy cutoff Ecut.

However, it is possible to use smaller FFT values and still obtain accurate answers even though some wrap-around error will be present. By default CASTEP uses a grid that corresponds to 3/4 of the ideal grid size. This setting can be changed using the Derived grid option on the Basis tab of the Electronic Options dialog. The actual values of the FFT grid are then displayed in the Divisions text boxes on the same dialog.

Ultrasoft potentials require a better FFT grid in order to represent the augmentation density accurately. The PW91 exchange-correlation functional also requires a better grid, to represent gradient-dependent terms in the potential.

Therefore, CASTEP derives a finer FFT grid in both of these cases, compared with other exchange-correlation functionals or norm-conserving potentials.

The table below shows the factor by which the actual grid differs from the ideal one depending on (i) Derived grid setting and (ii) the use of either the PW91 functional or ultrasoft pseudopotentials.

  Standard Fine Precise
No USP, no PW91 0.75 0.875 1.0
Either USP or PW91 0.875 1.0 1.15

Use the Fine setting (or better) for high-accuracy calculations, especially when the cell optimization is required.

To change the FFT grid quality

  1. Choose Modules | CASTEP | Calculation from the Materials Studio menu bar.
  2. Select the Electronic tab.
  3. Click the More... button to access the Electronic Options dialog.
  4. Select the Basis tab.
  5. Set the Derived grid quality from the dropdown list.
  6. The FFT grid values that corresponds to the selected quality are displayed below the dropdown list.

Finite basis set corrections

It is recommended that you apply a finite basis set correction to the total energy and to the stress tensor when stress is calculated, including geometry optimization runs with variable cell parameters and NPT/NPH molecular dynamics calculations. The Apply finite basis set correction setting corresponding to this is Smart.

Variable cell CASTEP calculations can be carried out in two modes, either fixed basis set quality (that is, fixed energy cutoff), or fixed basis set size (fixed number of plane waves). Finite basis set correction is required only in the former case, and this is accounted for automatically when the Smart setting is used.

CASTEP calculates the correction by determining the total energy at a number of cutoff energies, then evaluating the derivative required in the correction term, numerically. It is normally sufficient to use three reference points in such calculations.

In some cases the value of the derivative may be known from a previous calculation. In such circumstances, the value can be input directly, using the Manual Correction mode.

However, under normal circumstances, it is strongly recommended that the finite basis set correction settings are left at their default values. These are:

Apply correction: Smart
Correction mode: Automatic
Numerical differentiation: 3 points

See Also:

Setting up electronic options
Electronic tab - CASTEP Calculation dialog
Basis tab - CASTEP Electronic Options dialog