(c) 2017, Stanislav Markov, The University of Hong Kong
All rights reserved

This work is licensed under the Creative Commons Attribution-ShareAlike 4.0
International License. To view a copy of this license, consult the LICENSE file
or visit http://creativecommons.org/licenses/by-sa/4.0/ .

NOTE: The rights holder(s) for this work explicitly require that the attribution
conditions of this license are enforced. Use in part or in whole of this data is
permitted only under the condition that the scientific background of the
Licensed Material will be CITED IN ANY PUBLICATIONS ARISING FROM ITS USE. The
required references are specified in this file and must be included in resulting works.


========================================
SKF for Si, O, and H
========================================

The SKFs here contain the electronic parameterisation of DFTB for Si and O
and H. They are verified for electronic structure and permittivity of bulk
Si and SiO2 and of extremely thin Si films (oxidised and hydrogenated),
both for amorphous SiO2 and for alpha-quartz SiO2.

The set is intended to accurately describe electronic and dielectric
properties of natively oxidised Si nanostructures, and for calculations of
electron current in atomistic models of field-effect transistors.
The key aspects of that are: low pressure diamond Si crystal and 
low pressure thermally grown, amorphous SiO2.

The set is unlikely to perform well for high-pressure phases of Si or SiO2,
or at least -- it has not been verified for that.

The following publications report results obtained with these files; 
Required references are [1] and [2]:

   [1] Stanislav Markov, Balint Aradi, Chi-Yung Yam, Hang Xie, 
       Thomas Frauenheim, Guanhua Chen,
       "Atomic Level Modeling of Extremely Thin Silicon-on-Insulator
       MOSFETs Including the Silicon Dioxide: Electronic Structure,"
       IEEE Trans. Elec. Dev., vol. 62/3, pp.696-704, 2015, 
       DOI: 10.1109/TED.2014.2387288.
   [2] Stanislav Markov, Gabriele Penazzi, YanHo Kwok, Alessando Pecchia, 
       Bálint Aradi, Thomas Frauenheim, GuanHua Chen,
       "Permittivity of oxidized ultra-thin silicon films from atomistic 
       simulations," 
       IEEE Elec. Dev. Lett., vol. 36/10, pp.1076-1078, 2015,
       DOI: 10.1109/LED.2015.2465850.
   [3] Stanislav Markov, Balint Aradi, Gabriele Penazzi, ChiYoung Yam, 
       Thomas Frauenheim, and GuanHua Chen, 
       "Towards Atomic Level Simulation of Electron Devices Including 
       the Semiconductor-Oxide Interface"
       SISPAD 2014, 9-11 Sept. 2014, Yokohama, Japan 
       DOI: 10.1109/SISPAD.2014.6931564.

Parameterisation details are reported in [1] above.
SKOPT, SKGEN, slateratom, twocnt, and DFTB+ (1.2.2) were used throughout the parametrisation.

DFTB calculations:
=======================================

The following maximum angular momenta must be used:

  Si: d, O: p, H: s

For calculations including spin-orbit coupling, the dftb_in.hsd
must have the following part added to the Hamiltonian section:

  SpinPolarisation = {}
  SpinOrbit = { Si [eV] = {0.0 +0.037 0.0} }

The parameters are obtained without orbital resolved Hubbard U values.
Therefore, in the Hamiltonian section of dftb_in.hsd:

  OrbitalResolvedSCC = No

SKF optimisation of electronic parameters
=======================================

Definitions for Si were optimised with SKOPT.
https://bitbucket.org/stanmarkov/skopt.

Definitions for H and O were adapted from Wahiduzzaman et al JCTC 9 
(2013 4006--4017), accounting for the integer-only potential-powers of skgen,
and adapting the basis for O and H to reproduce the energy levels 
reported in the paper.

Although optimisation of O was performed, it did not yield substantial 
improvement in the description of crystalline or amorphous SiO2.


Repulsive energy:
========================================

NOTABENE: NO REPULSIVE potential is provided; there is a dummy spline only. 
          The SKFs CANNOT be used to do structural relaxation.


Additional remarks:
========================================

Si-Si interaction:
    Reproduces the band-structure of bulk Si with reasonable accuracy,
    but the direct gap and CB min at L are underestimated largely.
    The effective masses at VB maximum and CB minimum (along Delta-line)
    are close to experimental too.  For the relevant errors consult 
    S. Markov et al, IEEE TED-62(3), 696, 2015 -- [1] above.

Si-O interaction:
    Reproduces the valence band-structure of bulk alpha quartz 
    SiO2 with reasonable accuracy, and yields a large band-gap (overestimates
    experimental values by ~20%). But the conduction band is poorly 
    expressed. Specifically, there is only a very shallow CB minimum at Gamma,
    compared to DFT calculations, and that is why BG is overestimated.
    Works well in confining Si by SiO2 (quartz or amorphous), and this is
    its main purpose: to enable exploration of oxidised nano-sized Si 
    films, wires, particles in terms of electronic, dielectric and 
    transport properties.

Si-H interaction:
    Reproduces electronic/deielectric properties of 
    H-passivated Si(100) with reasonable accuracy, compared to DFT.

O-H interaction:
    This is fictitious, and have not been a target of optimisation.
    It is here to only help in cases where outer SiO2 surface of the
    nanostructure must be passivated and it may have O-H bonds.
    If the SiO2 is thick enough (e.g. 1nm), the O-H interaction will
    not affect the properties of the Si channel.