Thesis (MSc)--University of Stellenbosch, 2006. / ENGLISH ABSTRACT: Quantum mechanics is the method of choice when it comes to the accurate modeling of single
molecules and clusters. The correlation energy is the single most important aspect when
studying clusters computationally, and reproducing the correlation energy accurately poses a
bigger challenge to the computational chemist than in the modeling of single molecules. Very
high levels of theory and large basis sets need to be used.
Nevertheless, since the calculation of large systems, such as crystals and biological
systems, is generally beyond the capacity of quantum mechanics, molecular mechanics is
generally used for these systems. Unfortunately due to its nature, molecular mechanics cannot
model important quantum effects, but this problem can be solved by a hybrid system in which
one part of the system is treated by quantum mechanics and the remaining part by molecular
mechanics.
In order to combine quantum mechanics with molecular mechanics one needs to optimize
the parameters for the molecular mechanics part to allow it to function with the quantum
mechanics. The research described in this work is based on the ONIOM-EE method, which is
such a hybrid method.
In this work we investigate the applicability of the ONIOM-EE method in modeling
hydrogen fluoride, carbon monoxide and CO/HF clusters. Most of the clusters’ geometries in
this work are not experimentally or computationally known. We therefore perform a
computational analysis of all of the clusters by using various methods including Atoms in
Molecules, Natural Bond Orbital analysis, Mulliken population analysis and the analysis of
delocalized molecular orbitals to obtain information for the development of hybrid systems.
During this process we look at different charge derivation schemes and at two different
methods of optimizing force field parameters for these clusters. We develop a method to
make force field optimization faster and better for specific hybrid systems. This method
showed that in all cases the optimized parameters were an improvement on those of the
Universal Force Field. We show the importance of an accurate description of the electrostatic
interactions in HF, CO and CO/HF clusters and that this is the Achilles heel when attempting
to optimize van der Waals parameters for force fields. We further show that atomic point
charges are not a good approximation of a molecules’ charge density in hybrid methods. In
addition, we make suggestions on how the present method for ONIOM-EE can be improved
to make the modeling of van der Waals clusters feasible. / AFRIKAANSE OPSOMMING: Kwantum meganika is die metode van keuse wanneer enkele molekule en molekulêre sisteme
op rekenaar gemodeleer moet word. Dit is egter bekend dat die modelering van molekulêre
sisteme ’n groter uitdaging stel aan die molekulêre modeleerder, aangesien baie hoë vlakke
van teorie en groot basisstelle gebruik moet word om die korrelasie-energie, rekenkundig te
produseer. Die akkurate herprodusering van die korrelasie-energie is seker die heel
belangrikste vereiste waaraan voldoen moet word as molekulêre sisteme d.m.v. ’n rekenaar
gemodeleer word.
Nietemin is dit onprakties om kwantum meganiese metodes te gebruik vir groot sisteme
soos kristalle of biologiese molekule en juis om dié rede word molekulêre meganika meestal
ingespan vir sulke gevalle. Molekulêre meganika is egter ondoeltreffend om belangrike
kwantumeffekte te modeleer. Tog is daar ’n oplossing vir hierdie probleem in die vorm van ’n
hibried sisteem waar een deel van die sisteem met kwantum meganika en die oorblywende
deel van die sisteem met molekulêre meganika behandel word.
Om dit moontlik te maak om molekulêre meganika met kwantum meganika te kombineer,
moet parameters vir die molekulêre meganika deel geoptimiseer word sodat dit saam met die
kwantum meganiese deel kan funksioneer. Die navorsing wat in hierdie studie beskryf word is
gebaseer op so ’n hibriedmetode wat bekend staan as ONIOM-EE.
In hierdie studie bestudeer ons die moontlikheid om ONIOM-EE te gebruik vir die
modelering van molekulêre sisteme van waterstoffluoried, koolstofmonoksied en CO/HF
sisteme. Die meeste van die sisteme, wat in hierdie studie behandel word, se strukture is
onbekend, beide in terme van eksperimentele gegewens en molekulêre modelering. Ons voer
dus ’n volledige analise van al die sisteme uit deur van verskeie metodes soos “Atoms in
Molecules”, “Natural Bond Orbital” analise, Mulliken populasie analise en die analise van
gedelokaliseerde molekulêre orbitale, gebruik te maak. Dit stel ons in staat om ’n
hibriedsisteem te ontwikkel vir die molekulêre sisteme. Gedurende die proses ondersoek ons
ook die gebruik van verskillende ladingsafleidings-sisteme en twee metodes word ondersoek
waarop ’n kragveld vir ’n hibriedsisteem geoptimiseer kan word. Ons toon aan dat die
geoptimiseerde parameters beter resultate lewer as die van die “Universal Force Field” en lig
ook die belangrikheid daarvan uit dat die elektrostatiese interaksies se beskrywing ’n
hibriedsisteem se Achilles hiel is indien van der Waals parameters geoptimiseer moet word.
Ons toon aan dat die gebruik van puntladings op atome om die ladingsdigtheid in molekulêre
sisteme te beskryf, ’n onakkurate benadering is. Sekere aanbevelings hoe om die ONIOM-EE
metode sodanig te verbeter, dat dit wel gebruik kan word om van der Waals sisteme suksesvol
te modeleer, word ook gemaak.
| Identifer | oai:union.ndltd.org:netd.ac.za/oai:union.ndltd.org:sun/oai:scholar.sun.ac.za:10019.1/21774 |
| Date | 12 1900 |
| Creators | Crous, Werner |
| Contributors | Dillen, J.L.M., Esterhuysen, C., Stellenbosch University. Faculty of Science. Dept. of Chemistry and Polymer Science. |
| Publisher | Stellenbosch : Stellenbosch University |
| Source Sets | South African National ETD Portal |
| Language | en_ZA |
| Detected Language | English |
| Type | Thesis |
| Format | xi, 216 leaves : ill. (some col.) |
| Rights | Stellenbosch University |
Page generated in 0.0195 seconds