Next: Appendix C Up: TBMD user guide Previous: Appendix A

Appendix B: technical details about tight-binding molecular dynamics





TB basis set

In the tight-binding method the basis functions are atomic like orbitals. The absolute minimum basis set is usually composed of the states of the outer shell. For instance for Si one would take four orbitals (adopting the bra and ket notation) :

equation473

and for transition metals nine :

equation478

The number of basis functions per atoms will be noted tex2html_wrap_inline2010 . If the number of atoms in the system is tex2html_wrap_inline2012 then the dimension of the space spanned by the basis set is tex2html_wrap_inline2014 .

Any state is expressed as a linear combination of these atomic-like orbitals tex2html_wrap_inline2016 :

equation490

where the index tex2html_wrap_inline2018 runs over atoms and orbitals.

In all generality we will assume that the basis set is not orthonormal :

equation497



Empirical tight-binding method

In the empirical tight-binding method (ETBM) the basis functions are not know explicitly. Instead the matrix elements of the Hamiltonian and the overlap are given :

equation504

where the label tex2html_wrap_inline2020 refers to the type of the ion associated with which orbital tex2html_wrap_inline2022 .



Electronic-structure

The aim of any of the methods presented here is to calculate the electronic structure of the system or in other words to find the electronic ground state.

The most straightforward approach to this problem is to calculate the eigenvalues tex2html_wrap_inline2024 and eigenstates tex2html_wrap_inline2026 of the Hamiltonian operator :

  equation528

The eigenvalues and the eigenvectors of the Hamiltonian contain all the information we need to calculate the total-energy and the forces on the atoms.

The number of electrons is given by

  equation532

where tex2html_wrap_inline2028 is some broadening function, in our case we will use the Fermi function tex2html_wrap_inline2030 , and tex2html_wrap_inline2032 is the width of the broadening. The quantity tex2html_wrap_inline2034 is called the occupation of state i. For tex2html_wrap_inline2008 tex2html_wrap_inline2040 0 the Fermi function tends to the step function tex2html_wrap_inline2044 and tex2html_wrap_inline2034 =0 if tex2html_wrap_inline2048 > tex2html_wrap_inline2052 and tex2html_wrap_inline2034 =1 if tex2html_wrap_inline2048 < tex2html_wrap_inline2052 . The factor two accounts for the spin degeneracy (each state is occupied by two electrons). Equation (7) should be regarded as the definition of the Fermi energy tex2html_wrap_inline2052 which solves the equation tex2html_wrap_inline2064 .

The electronic energy is expressed as the sum of the eigenenergies of the Hamiltonian :

  equation543

where

equation553

Deriving the forces from the eigenstates is done via the Hellman-Feynman theorem. For simplicity we will consider an insulator for which the force acting on atom I is simply :

equation556



Molecular-dynamics

The molecular dynamics-method (MD) is simply a method for integrating the classical equations of motion of the atoms, which given by :

displaymath2068

The simulations proceeds in four steps :

  1. The positions of the atoms are given as input and a used to calculate the Hamiltonian of the system.
  2. The electronic ground state of the system is calculated by finding the eigenvalues and eigenvectors of the Hamiltonian.
  3. The eigenvectors are used to derive the forces on the atoms.
  4. Knowing the forces on the atoms we can integrate the EOM for the ions and find the new atomic positions.

picture574



Florian Kirchhoff
Tue Jun 9 16:34:36 EDT 1998