Unrestricted fragments

The fragments from which the molecule is built must be spin-restricted, that is: the fragment files must be result files of spin-restricted calculations. For purposes of analysis, however, it may be desirable in some applications to build your molecule from unrestricted fragments. This can be simulated as follows.

You tell adf that you want to treat the fragments as if they were unrestricted; this causes the program to duplicate the one-electron orbitals of the fragment: one set for spin-α and one set for spin-β. You can then specify occupation numbers for these spin-unrestricted fragments, and occupy spin-α orbitals differently from spin-β orbitals.

Of course, the unrestricted fragments that you use in this way, are not self-consistent: different numbers of spin-α and spin-β electrons usually result in different spatial orbitals and different energy eigenvalues for spin-α and spin-β when you go to self-consistency, while here you have spatially identical fragment orbitals. Nevertheless it is often a fair approximation which gives you a considerable extension of analysis possibilities.

FRAGOCCUPATIONS
 fragtype
   irrep spin-a // spin-b
   irrep spin-a // spin-b
   ...
 subend
 fragtype
   irrep spin-a // spin-b
   ...
 subend
 ...
end

fragtype

One of the fragment types and functions as a (block type) subkey. The data block for the subkey ends with the standard end code for block type subkeys (subend).

irrep

One of the irreducible representations (irreps) for the point group symmetry that was used in the computation of that fragment.

spin-a // spin-b

Two sequences of occupation numbers, which will be applied to the spin-α and spin-β versions of the Fragment Orbitals. The sequences must be separated by a double slash (//). See for comparison the specification of occupation numbers for the overall system (key charge).

The sum of spin-α and spin-β occupations must, for each fragment orbital in each irrep separately, be equal to the total (restricted) occupation of that orbital as it is stored on the fragment file. In other words: you can only change the distribution over spin-α and spin-β electrons within one orbital.

(Without this restriction the spatial distribution of the total (sum over spins) fragment charge density would be changed, leading to an incorrect bonding energy analysis after the calculation).

The data block of fragoccupations is not parsed for expressions and constants or functions defined under define. Any such items will not be recognized and not be replaced by their values.

Be aware that in more-dimensional irreps (e, t) the number of electrons in a fully occupied orbital is input as the dimension of the irrep times the one-electron orbital occupation. Compare the key occupations.

For irreps that are not mentioned in this input block, and hence for all irreps of fragment(type)s that are not mentioned at all, the spin-α and spin-β occupations will be set equal, which is of course what they in fact are on the (restricted) fragment file.

For an example of applying this option see [112].