MOLECULAR DYNAMICS in CATALYTIC SYSTEMS Richard Catlow, Alex O’Malley, Daniel Dervin, School of Chemistry, Cardiff University; Department of Chemistry, University College London; UK Catalysis Hub, Research Complex at Harwell THEMES: Application of computational and experimental techniques to - Hydrocarbon dynamics and reactivity in microporous catalysts - Dynamical restructuring of nano-particles MD MODELLING of MATERIALS with DL_POLY: Materials Chemistry Consortium Molecular Dynamics of Sorbed Molecules Nano-Particle Dynamics and Restructuring Radiation Damage and Radiation Cascades Ionic Migration in Superionic Conductors Crystal Nucleation and Growth Mechanisms THEME ONE: Application of computational and neutron techniques to: Hydrocarbon dynamics and reactivity in microporous and mesoporous catalysts MICROPOROUS MATERIALS: Framework structures materials with pores of molecular dimensions GAS SORPTION and CATALYSISMAPO-36 oxime + ION-EXCHANGE SEPARATION cyclohexane n-alkanes caprolactam ketone + epoxide, diol + benzaldehyde benzoic acid alkene + lactone + benzaldehyde benzoic acid cyclohexanone cyclohexanol, cyclohexanone, + NH3 adipic acid C3 + C2 functionalised Al P O M =Co,Mn products Sorbate Mobility in Zeolites • To improve the design/optimisation of microporous catalytic processes • Diffusion can control catalytic processes:modelling needed to complement experiment - Rate determining step? - Selectivity and separation - Preferred siting - Active site interactions - Mechanism/energetics • “Microscopic methods” - Molecular motion at equilibrium - PFG-NMR (μm and μs) - Quasielastic neutron scattering - Molecular dynamics simulations Quasielastic Neutron Scattering • Very small changes in neutron energy (μeV-meV) - Interact with translational motion. - Sensitive to 1H (useful for confined sorbates) • Change in spectral profile around the elastic line. - Broadening of a Lorentzian component - Fitting HWHM to models of jump/fickian diffusion - Matches scales of MD simulation Complementarity of QENS and current MD simulations • Increases in computational power have facilitated more sophisticated and accurate classical MD simulations. • Flexible frameworks • Removal of energy traps • Non-elastic sorbate/framework interactions • “Breathing” allowing access to areas • Simulation size/length • 10-100 ns with c. 4000 atoms • Allows new qualitative and quantitative phenomena to be observed • Channel switching/clustering 1. A. J. O’Malley, and C. R. A. Catlow. Phys. Chem. Chem. Phys. 15.43 (2013): 19024 2. 2. A. J. O’Malley., C. R. A. Catlow, M. Monkenbusch, and H. Jobic. J. Phys. Chem. C 119, 48 2015: 26999-27006.06 Longer n-alkanes in silicalite: Detailed MD models giving closer agreement with QENS • C8 – C20 in silicalite - directly compared with QENS studies by Jobic (2). • Used the same loadings as experiment e.g. 1.56 mol/uc • Use flexible frameworks and all- atom hydrocarbon potentials • Same temperature range to calculate diffusion coefficients and activation energy. (2) Hervé Jobic*,† and and Doros N. Theodorou‡ The Journal of Physical Chemistry B 2006 110 (5), 1964-1967 Diffusion coefficients: Compare simple MD models, hierarchical simulations, current study A plot of the calculated diffusion coefficients from the current MD simulations (О), MD simulations at infinite dilution (▲)1, MD simulations using simpler models (▲), Hierarchical simulations (∗) and QENS studies (□) . 3. O'Malley, Alexander J., and C. Richard A. Catlow. "Molecular dynamics simulations of longer n-alkanes in silicalite: a comparison of framework and hydrocarbon models." Phys. Chem. Chem. Phys. 15.43 (2013): 19024-19030. Activation energies: QENS, Hierarchical simulations, current MD simulations 4. O'Malley, Alexander J., and C. Richard A. Catlow. "Molecular dynamics simulations of longer n-alkanes in silicalite: Up-to-date models giving close agreement with experiment." Phys. Chem. Chem. Phys. – 2015 Isobutane in Silicalite: Combined NSE and MD study • Collaboration with H. Jobic IRCE-Lyon using the neutron spin echo technique. • Suited for slower moving (bulky) sorbates – very high resolution (spin relaxation) • Jump diffusion observed through experiment – 10 Å jump distance 550 K 500 K 450 K 3. A. J. O’Malley., C. R. A. Catlow, M. Monkenbusch, and H. Jobic. The Journal of Physical Chemistry C 119, 48 2015: 26999-27006.06. Isobutane in Silicalite: Combined NSE and MD study • MSD and Ds calculated giving agreement with experiment within a factor of 6. • Differences attributed to use of a perfect crystal for simulations. 3. A. J. O’Malley., C. R. A. Catlow, M. Monkenbusch, and H. Jobic. The Journal of Physical Chemistry C 119, 48 2015: 26999-27006.06. Isobutane in Silicalite: Combined NSE and MD study • At 550 K two modes of motion are observed, long term diffusion in either channel system or free diffusion with fast switching between both. • Appears isobutane diffusion is dictated by temperature dependent trapping in small sinusoidal channel segments 3. A. J. O’Malley., C. R. A. Catlow, M. Monkenbusch, and H. Jobic. The Journal of Physical Chemistry C 119, 48 2015: 26999-27006.06. Studying the Methanol-to-Hydrocarbons Process • Began with studying methanol diffusion in H- ZSM-5 (catalytic) and HY (non-catalytic): • Lack of diffusion in ZSM-5 due to pore size..? Should be accessible to OSIRIS • INS (MAPS) used to look deeper at the interactions with the different frameworks Methoxylation and methanol diffusion 4. A. J. O'Malley; S. F. Parker.; A. Chutia; M. R. Farrow; I. P. Silverwood; V. Garcia-Sakai; C. R. A. Catlow, Chemical Communications 2016, 52, 2897-2900. Methanol Diffusion in HY • Previous work by Plant, Maurin and Bell4,5 studied methanol diffusion siliceous Y and NaY using MD simulations. • Slower diffusion by a factor of 22 + when Na present, Ea 4 x higher! (Si/Al = 2.4) – experimental data? • Strong interaction with Na+ observed. • How does HY compare both with Na-Y and with siliceous Y? 5. Plant, D. F.; Maurin, G.; Bell, R. G. The Journal of Physical Chemistry B 2006, 110, 15926. 6. Plant, D. F.; Maurin, G.; Bell, R. G. The Journal of Physical Chemistry B 2007, 111, 2836. Methanol diffusion in HY: QENS, OSIRIS -10 2 -1 • Ds of 2–5 x 10 m s -1 • Ea = 8.8 kJ mol • Faster than NaY (PFG–NMR) by factor of 7 at c.360 K. 7. A. J. O'Malley, V. García Sakai, I. P. Silverwood, N. Dimitratos, S. F. Parker, C. R. A. Catlow, Phys. Chem. Chem. Phys., 2016, 18, 17294-17302 MD Simulations • Diffusivities measured as 1.4–3 x 10-9 m2s-1 (factor of 6 higher than QENS), - defects? • An order of magnitude lower than MD in siliceous Y, significant H-bonding lowering mobility. • Almost 2 orders of magnitude higher than MD in NaY! (Far fewer sites Si/Al = 30) • Siliceous Y > HY> Na+Y 7. A. J. O'Malley, V. García Sakai, I. P. Silverwood, N. Dimitratos, S. F. Parker, C. R. A. Catlow, Phys. Chem. Chem. Phys., 2016, 18, 17294-17302 Comparing ammonia mobility in different NH3-SCR catalysts 8. A. J. O’Malley, I. Hitchcock, M. Sarwar, I. P. Silverwood, S. Hindocha, C. R. A. Catlow, A. P. E. York and P. J. Collier, Phys. Chem. Chem. Phys., 2016,18, 17159-17168 Cu-Chabazite: The SCR Process • Collaboration with Johnson Matthey using QENS to study the diffusion of ammonia in chabazite (CHA) and the industrial catalyst Cu-CHA, to determine the effect of counter ion presence on diffusion. • Small pore zeolite (3.8 x 3.8 Å) – counterion location variable… 8. A. J. O’Malley, I. Hitchcock, M. Sarwar, I. P. Silverwood, S. Hindocha, C. R. A. Catlow, A. P. E. York and P. J. Collier, Phys. Chem. Chem. Phys., 2016,18, 17159-17168 Ammonia Diffusion in Chabazite as a Function of Cu Counterion Presence • Ammonia seems to behave basically the same in both zeolites.. • Jump distance of 3 Å, similar residence times at each T (16-28 ps) • More likely through 8-ring than 6-ring? • Why is the copper ion not changing anything..? 8. A. J. O’Malley, I. Hitchcock, M. Sarwar, I. P. Silverwood, S. Hindocha, C. R. A. Catlow, A. P. E. York and P. J. Collier, Phys. Chem. Chem. Phys., 2016,18, 17159-17168 Ammonia Diffusion in Chabazite as a Function of Cu Counterion Presence • MD shows clusters of NH3 coordinated to the Cu2+ . • ‘Shell’ effectively shields the remaining NH3 from the potential sink of the Cu2+ . • The coordinated NH3 moved far too slowly to be detected by the OSIRIS instrument – so is not measured (NSE?) • Trajectory plots appear to confirm that diffusion is dominant through 8-rings • Cu2+ shell allows intercage diffusion through 8- ring to carry on unimpeded. 8. A. J. O’Malley, I. Hitchcock, M. Sarwar, I. P. Silverwood, S. Hindocha, C. R. A. Catlow, A. P. E. York and P. J. Collier, Phys. Chem. Chem. Phys., 2016,18, 17159-17168 Ammonia Diffusion in Levynite • Building on study in CHA, in the 2D framework LEV. • Same 8-ring windows present as in CHA, however stacking leads to the channels running in 2-dimensions in LEV, (3D in CHA). • Difference in overall diffusivity? • QENS experiments performed on OSIRIS and complementary MD simulations performed at JMTC. • 273, 323 and 373 K at ammonia saturation. Ammonia Diffusion in Levynite LEV • Again, ammonia seems to behave with the same jump diffusion mechanism • Jump distance of 3 Å, similar residence times at each T (16-25 ps) • Same characteristics suggest mobility through 8-ring windows dominant motion, • Similar Ds values obtained, same case for the MD simulations? Ammonia Diffusion in Levynite • While QENS experiments show similar diffusivity, the MD simulations suggest diffusion is faster in CHA by over a factor of 2. • 2-dimensional