Realistic Modelling of the Effects of Solvent and Additives on Crystal Growth
The aim of this project is to investigate, at the molecular level, the interactions between the crystal surfaces and solvents/additives/foreign ions, using both interatomic potential and quantum mechanical techniques. The information from these studies will shed light on many important complex processes in crystal growth, from the role of solvent to the mode of action of growth modifiers. These processes are extremely important in many industrial processes, ranging from hydrometallurgy to the pharmaceuticals. The project requires enormous computing resources if accurate models are to be used due to the large number of atoms involved in solvent calculations and the complexity of quantum chemical calculations of additives with crystal surfaces.
The aim of this project is to investigate, at the molecular level, the interactions between the crystal surfaces and solvents/additives/foreign ions, using both interatomic potential and quantum mechanical techniques. The information from these studies will shed light on many important complex processes in crystal growth, from the role of solvent to the mode of action of growth modifiers. These processes are extremely important in many industrial processes, ranging from hydrometallurgy to the pharmaceuticals. The project requires enormous computing resources if accurate models are to be used due to the large number of atoms involved in solvent calculations and the complexity of quantum chemical calculations of additives with crystal surfaces.
|
Principal Investigator Andrew RohliVEC, The hub of advanced computing in Western Australia Curtin University of Technology |
Project d64, e20 |
|
Co-Investigators Terry RankineiVEC Damien Carter Physics University of Sydney Mark Ogden Applied Chemistry Curtin University of Technology |
RFCD Codes 250601, 250603 |
Significant Achievements, Anticipated Outcomes and Future Work
The work in this project has progressed exceedingly well in 2005. The PhD student Damien Carter has had his thesis
examined and passed. Details of his work can be found on the Curtin Library website
(http://opac.lis.curtin.edu.au/F?func=find-a&find_code=SYS&local_base=cur01www&request=000560419). His thesis was well
received by his examiners and he is now doing a postdoc at the University of Sydney, where he continues to be a big user
of APAC resources. A paper on his work on the surfaces of potassium hydrogen phosphate (KDP) is now in press.
The study of the performance of DFT in modeling condensed phases of molecules has almost finished. The simulations of
the crystal structures of the small organic molecules glucose and deoxygalactose are complete. Both systems have been
simulated using CASTEP, Gaussian 03 and SIESTA. In addition SIESTA simulations of the biologically important cholesterol
and lactose monohydrate crystals are nearing completion, and as predicted are each taking significant amounts of CPU
time e.g. optimization of the 8 cholesterol molecules within the unit cell takes nearly 10,000 hours on the AC. This
calculation keeps the cell fixed at the experimental values. Simulations allowing the unit cell to optimize too are
currently underway.
Some of work undertaken as part of this project has been incorporated into the paper 'Robert Boyle's Chiral Crystal
Chemistry: Computational Reevaluation of Enantioselective Adsorption on Quartz' which has been accepted by Chirality.
Another body of work has used quantum mechanics to simulate the electronic structure of methyl red in different
protonation states. This information was used to understand the behaviour of methyl red adsorbed inside growing crystals
of phthallic acid
The scene is set in 2006 to simulate complex surfaces, both inorganic and organic, using high quality density functional calculations. These simulations will be some of the largest quantum mechanical simulations to date and will demonstrate the power of the APAC AC system.
Computational Techniques Used
The DFT calculations on KDP were performed using the first principles program SIESTA (http://www.uam.es/departamentos/ciencias/fismateriac/siesta/). A particular strength of SIESTA is that it is capable of treating large molecular and periodic systems. Calculations with SIESTA employed periodic boundary conditions, in which the primary cell is surrounded by periodic images of itself. Simulations typically involved between approximately 170-340 atoms. Calculations are typically run on 8 processors and require approximately 2-4GB of RAM. For the parallelisation of SIESTA, a distributed memory approach has been taken with communication via MPI. When using matrix diagonalisation to solve the SCF procedure, then the Scalapack library has been utilised. The orbitals are distributed over the processors in a 1-D block cyclic fashion, while for evaluation of the Hartree and Exchange-Correlation potentials on the real space mesh a 2-D block-cyclic distribution of grid points is utilised.
The work on molecular crsytals has employed CASTEP, SIESTA and Gaussian 03 programs to see if indeed the long-range forces are being correctly modeled or if it is just BSSE that is holding these crystals together. Although all three programs are using DFT with the same functional, they handle the basis set representation quite differently and sothe simulations are essentailly probing how well these different techniques (Gaussian basis set vs numerical basis vs planewaves) handle molecular crystals.
The calculations docking organic molecules to quartz surfaces utilised the GULP program (http://www.ivec.org/gulp). GULP
uses interatomic potentials and as such ran the docking calculations in a few hours on a single processor. However, 424
starting configurations were trialled which is a perfect job for the LC.
Finally, the calculations on methyl red utilised Gaussian 03.
Publications, Awards and External Funding
External Funding and Awards
None.
Publications
B. Kahr, B. Chittenden and A.L. Rohl, Robert Boyle's Chiral Crystal Chemistry: Computational Reevaluation of
Enantioselective Adsorption on Quartz, Chirality, 18, 2006, 127-133.
J.B. Benedict, D.E. Cohen, S. Lovell, A.L. Rohl and B. Kahr, What is Syncrystallization? States of the pH Indicator
Methyl Red in Crystals of Phthalic Acid, J. Am. Chem. Soc., in press.
D.J. Carter, A.L. Rohl and J.D. Gale, Ab Initio Simulations of the (101) Surfaces of Potassium Dihydrogenphosphate (KDP), J. Chem. Theory Comput., in press.