Transferability of Parameters for Al and O in the Adsorption of Alkanes on Alumina

1
Transferability of Parameters for Al and O in the
Adsorption of Alkanes on Alumina

• Damien A. Bernard-Brunel
• Department of Chemical Engineering and Materials
  Science
• Wayne State University
• CHM-7440 Winter 2005

2
Overview

• Introduction
• Methodology
• Job, Theory, and Basis set
• Counterpoise calculations
• Results / Discussion
• Genetic Algorithm
• Testing other compounds
• Conclusion
• Future work

3
Introduction

• Problems with simulation of adsorption on metal
  oxides
• Too costly for Gaussian, despite great accuracy
• Some missing parameters in Molecular Dynamics
  programs
• Proposed solution
• Run small calculations with Gaussian (methane on
  a small cluster)
• Extract the parameters necessary for the MD
• Test the transferability on other compounds

4
Job, Theory, and Basis sets

• Optimize both structures
• rhf/6-31G
• rmp2/6-31G(d,p)
• MP2 with restricted spin
• Default for Gaussian 98
• 6-31G(d,p)
• Use diffuse functions, mostly for O
• Use polarization
• Scan
• Vary the distance between Al and C
• Calculate energy at each step

5
Counterpoise calculations

• If alkane close to surface, interaction between
  the basis sets of the 2 groups gt use
  counterpoise calculations
• Perform the calculations with only one molecule
  present, but with both basis sets

• rmp2/6-31g(d,p) scan massage
• Sample Counterpoise calculation
• 0 1
• Z-Matrix connectivity...
• 1 Nuc 0.00
• 2 Nuc 0.00

6
Results

• Results look like a Lennard-Jones
• Parameters for fit extracted from Genetic
  Algorithm
• Uses analogy with replication of DNA
• At random mutation, crossover, survival of
  fittest

7
Results (contd)
sAl Å eAl kcal/mol sO Å eO kcal/mol
CHARMM 1.6116 0.118 10-3 1.7061 6.910
OPLS 1.9077 0. 031 10-3 2.2186 2.626
TraPPE 1.6139 0. 390 10-3 1.6438 7.523
8
Results (contd)
9
Discussion

• Check if new parameters yield a good fit for
  Propane and Methanol

10
Conclusion

• Problem
• Large scale adsorption on a surface too costly
  for Gaussian
• Missing parameters for Molecular Dynamics
• Perform small calculation with Gaussian
• Extract parameters accurately with Genetic
  Algorithm
• Test parameters with other chemicals to check
  Transferability

11
Future work

• ONIOM
• 1st layer alkane and Al Os on surface
• 2nd layer rest of the cluster
• Optimization of orientation of alkane
• Use multi-step jobs with the --Link1-- command
• Test other chemicals (alkenes, alcohols)

12
Acknowledgements

• Dr. J. J. Potoff
• Dr. H. B. Schlegel
• D. L. Carroll
• G. Kamath, B. Munk, J. P. Wang
• Thanks for your attention
• and enjoy the summer