Royal Society of Chemistry Theoretical Chemistry Group

Graduate Student Meeting

Imperial College, London - March 24th, 2004

The annual meeting for presentations by final year graduate students in theoretical chemistry will be held on the afternoon of Wednesday 24th March 2004 in Lecture Theatre C, RCS1 Building, Imperial College, London. (Entering the campus from Exhibition Road, the entrance to the RCS1 Building is a double door on your left, just after the Science Library and Post Office - a map can be found here (RCS1 is building number 8).)

This year, the Group is introducing an award for the student who is judged to have presented the best talk at the meeting.


13:30Prof. R. M. Lynden-BellIntroduction
13:35Darragh O'Neill (Nottingham)Hartree-Fock-Wigner Theory: a two-electron alternative to DFT
14:00Natalie Lambert (University College London)Computational and experimental studies of the bond-forming reactivity of multiply-charged cations with molecules in the gas phase
14:25Akyl Tulegenov (Nottingham)Alchemy with intermolecular potentials
14:50Oliver Lanning (Oxford)Screening at a charged surface by a molten salt
15:45Ben Morgan (Oxford)Pressure driven transformations between tetrahedrally and octahedrally coordinated nanocrystals
16:10Tim Watson (Nottingham)Calculating vibrational frequencies of amides - from formamide to flavodoxin
16:35Dominic Clare (University College London)Molecular dynamics simulation of peptide-protein interactions
17:00Scott Habershon (Birmingham)Biologically-inspired computational chemistry: application to crystal structure determination from powder diffraction data
17:30Award Announcement and Presentation 

All are welcome. There are no registration formalities.


Hartree-Fock-Wigner Theory: a two-electron alternative to DFT

Darragh O'Neill (Nottingham)

Hartree-Fock (HF) calculations account for approximately 99% of the total electronic energy in molecules. However, it is important to account for the final 1%, the electron correlation energy, for both qualitatively and quantitatively accurate results. Methods for treating electron correlation range from the computationally expensive wavefunction based methods (CI, CCSD, etc.) to the much cheaper density functional theories. We have recently introduced a method based on the Wigner intracule which gives a measure of the probability of finding two electrons at a point in phase space. In particular we use the HF intracule which results in a method which scales in an analogous way to HF calculations. We have parameterised our method using accurate calculations on the neon atom and this gives good estimates to the correlation energy of molecules. This can also be implemented in a self-consistent manner to allow geometry optimisation and frequency calculations.

Computational and experimental studies of the bond-forming reactivity of multiply-charged cations with molecules in the gas phase

Natalie Lambert (University College London)

Whilst the reactivity of multiply-charged ions is quite well understood in the context of aqueous media and metallic phases, relatively little is known of their chemical behaviour in the gas phase. It is only since the 1980's, due mostly to advances in experimental technology, that their role in the chemistry of extreme environments, such as the interstellar medium, planetary atmospheres and natural and man-made plasmas, has been revealed. In this talk I intend to present the results of our group's investigations into the bond-forming reactivity of a number of small atomic and molecular doubly-charged cations (dications) with molecules. These reactions are monitored experimentally by means of crossed-beam time-of-flight mass spectrometry. The potential energy surfaces of these reactions are then explored computationally in order to reveal the reaction mechanisms. Due to [dication + neutral] surfaces often existing as excited states of [monocation + monocation] surfaces, these calculations often require the use of sophisticated ab initio techniques such as CCSD(T) and QCISD(T). We have also found that for certain systems, the use of multireference methods such as MCSCF and MRCI are crucial in order to provide accurate descriptions of the potential energy surfaces.

Alchemy with intermolecular potentials

Akyl Tulegenov (Nottingham)

The SIMPER-P method to estimate intermolecular potentials for weakly bound systems is described. The potentials obtained using SIMPER-P are shown to be competitive in accuracy with those obtained from CCSD(T) ab-initio calculations. At the same time the computational expense is comparable to low-level methods (MP2). SIMPER-P is successfully tested on different systems (Ar-Ar,Ar-HF,Ar-H2,H2O-N2).

Screening at a charged surface by a molten salt

Oliver J. Lanning (Oxford)

The screening of the electrical potential at a charged solid surface in a molten salt, KCl, has been investigated in a Molecular Dynamics simulation study. In the study the molten salt was confined between two rigid walls of equal and opposite charge. The relaxation time associated with the screening of the charged walls by the molten salt is found to be very short, and not dependent on diffusion. We also find that despite pronounced oscillatory structure in the charge density, the structure and dynamics of the ions close to the interface are very similar to those in the bulk. Click here for a pdf version, including figure.

Pressure driven transformations between tetrahedrally and octahedrally coordinated nanocrystals

Ben Morgan (Oxford)

The pressure-driven transformation of ionic nanocrystals, from the four coordinate zinc-blende to the six coordinate rocksalt crystal structures, have been studied using constant stress molecular dynamics, for system sizes ranging between 1000 and 4000 ions. A rigid-ion potential model was employed for the ion-ion interactions in the nanoparticles, and a binary Lennard-Jones system was used as a surrounding pressurising medium. Calculated diffraction patterns confirm the changes in the underlying crystalline structure, and the transformation mechanism is identified by following individual trajectories. This mechanism is found to be the same as has been observed in earlier simulations of bulk material, and provides a microscopic explanation for observed changes in the shape of the nanoparticle, and for the formation of grain boundaries, as well as the internal energy profile during the transformation.

Calculating vibrational frequencies of amides - from formamide to flavodoxin

Tim Watson (Nottingham)

The infrared (IR) is an information rich region of molecular spectra. From characteristic absorptions it is possible to determine much structural information about molecules. This has been used to a large degree in the study of protein structure as a complementary technique to circular dichroism, X-ray crystallography and NMR. However, the current understanding of protein IR spectra is predicated mainly on empirical structure-spectra relationships that are not infallible. Providing a theoretical basis for protein spectra will help to reduce these problems. This talk will outline the accurate methods used for small molecule calculations, their relationship to simpler methods for protein calculations and our current results.

Molecular dynamics simulation of peptide-protein interactions

Dominic Clare (University College London)

Protein-peptide interactions in aqueous solution have been probed using a molecular dynamics procedure with explicit solvent. An implicit solvent model has been used in post-processing with systematic mutation of each peptide residue to calculate binding free energies. Entropic contributions to the binding free energy have been estimated using classical ideal gas thermodynamics. The interaction between the oncoprotein Mdm2 and the tumour suppressor peptide p53 has been used as a test system. The method has been extended to study the interaction of IQN17, an HIV-1 gp41 mimic, with a potential fusion inhibitor D10-p1.

Biologically-inspired computational chemistry: application to crystal structure determination from powder diffraction data

Scott Habershon (Birmingham)

Many materials of interest, including pharmaceuticals, biomolecules, pigments and zeolites, cannot be easily prepared in the form of single crystals of sufficient size and quality for analysis by single-crystal diffraction techniques. In such cases, powder diffraction methods offer the only alternative route to complete structure determination. Consequently, recent years have witnessed an increasing interest in the development and application of new computational approaches aimed at aiding crystal structure determination from powder diffraction data. This presentation will highlight two recent aspects of work within this area. First, a new approach for determining unit cell parameters from powder diffraction data (commonly referred to as indexing) will be presented. Here, the indexing process is treated as a pattern recognition problem in which an Artificial Neural Network (ANN) is used to predict unit cell parameters from experimental powder diffraction data after analysis of the Bragg peak positions in powder diffraction patterns produced by known unit cells. Illustrative examples for orthorhombic systems will demonstrate the potential of this approach, and several aspects of the performance of the ANN will also be discussed. Secondly, recent developments of a Genetic Algorithm (GA) approach for direct-space crystal structure solution will be presented, with the focus being on improving the calculation speed and efficiency. Examples illustrating the current scope of this methodology will also be presented.