OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
Structural isomerism in transition-metal clusters
T. R. WalshDept. of Chemistry and Centre for Scienti�c Computing
www.warwick.ac.uk/go/nanoclusters
March 2, 2006
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
1 Background
2 Methods
3 Results for Niobium
4 Rhodium reactivity
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
Structural isomerism: the problem
For a given size n, a cluster can adopt many di�erentstructures.
The number of structures grows exponentially with size!
In the case of transition-metal clusters, experiments indicatesome structures are more reactive than others ! catalysis
Despite recent advances, it is still di�cult to assign clusterstructures from experimental data alone
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
Structural isomerism: the problem
Calculations can complement experimental approaches byidentifying
which structures are observed and why
which structures are more reactive than others and why
The main ingredient used to answer these questions is thepotential energy landscape (PEL)
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
PEL and Structural Isomers
During experiments, observed clusters could be low-energystructures, or they could be trapped metastable structures, ora mix of both
Cannot address structural isomerism by studying minima alone
Instead, must determine the structure of the PEL to makepredictions about cluster relaxation dynamics
Predictions done using simulation techniques can tell us aboutequilibrium and transient populations of structural isomers.
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
Two Simulation Issues for Structural Isomers
1. The relaxation simulations used here must be appropriate:
Can't use straightforward molecular dynamics : can'tnecessarily surmount high barriers
Rare-event simulation methods show promise: thermally
activated dynamics
Can use kinetic models : master equation approach
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
Two Simulation Issues for Structural Isomers
2. The PEL must be accurately represented:
Energy/Forces evaluated thousands of times in a run!
Full electronic structure theory not practical forsimulations
Even density-functional theory (DFT) using Car-Parrinellotakes too long
Need a good model potential to describe metal-metalbonding
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
Interpreting results
Simulations are limited ! they suggest possible structuralisomers.
Must complement these results with other calculated data forcomparison with experiment.
Ionization potentials.IR spectra.Franck-Condon factors for simulated ZEKE or MATI spectra.Reaction kinetics (wrt small molecule adsorbates).
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
Master Equation approach
Potential energy landscape (PEL) information supplied viacalculating the rate of interconversion between isomers
Obviates need for expensive `on-the- y' dynamics
We only need relative rates here!RRKM theory is used to calculate these rates
Input for RRKM { energies and harmonic frequenciescalculated using DFT
These rates are used as input to the Master Equation
This approach yields the relative population of each isomer asa function of time (coupled di�erentials equations solvednumerically).
This simulates cluster relaxation in the beam
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
Master Equation approach
Potential energy landscape (PEL) information supplied viacalculating the rate of interconversion between isomers
Obviates need for expensive `on-the- y' dynamics
We only need relative rates here!RRKM theory is used to calculate these rates
Input for RRKM { energies and harmonic frequenciescalculated using DFT
These rates are used as input to the Master Equation
This approach yields the relative population of each isomer asa function of time (coupled di�erentials equations solvednumerically).
This simulates cluster relaxation in the beam
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
The catch:
We must pre-suppose the rearrangement mechanisms forinterconversion.How we do it: a `two-stage' procedure
Use Basin-Hopping with existing (unsuitable) potential to �ndcandidate structures
Re�ned these minima using DFT optimisations
Mapped out connectivities between minima using a knowninterconversion mechanism
Diamond-Square-Diamond (DSD) and Cap Migration (CM)mechanismConnectivity was mapped outwards from global minimum
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
Nb10 : Experimental motivation
Knickelbein and Yang claim observation of two isomers ofNb10 with practically same ionization potential
Bondybey and coworkers observe bi-exponential reactionkinetics of Nb+10 with ethene
Smalley and coworkers did not observe structural isomerism ofNb10 when reacting with H2.
No structural assignment to date
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
Example barrier heights for neutral and cationic systems
Connection Forward ReverseBarrier (eV) Barrier (eV)
1{2 1.42 (1.06) 0.43 (0.48)2{3 0.71 (0.79) 0.32 (0.33)2{4 0.97 (0.92) 0.38 (0.47)4{6 0.70 (0.77) 0.46 (0.13)5{6 0.62 (0.65) 0.59 (0.23)
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
Isomer VIP (eV)
1 6.47 (5.24)2 5.84 (4.88)4 6.00 (4.61)7 5.89 (4.92)
Relation to Knickelbein experiments
LDA and BLYP IP's are not always consistent
However, IP for structure 1 is consistently higher than allothers
Suggests they may have observed structure 2 and 4.
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
IR Spectra
Recently reported FIR-MPD experiments show great promise
(Fielicke, von Helden, Meijer and co-workers)
But occasional ambiguities reported for structural isomerism
IR spectra are easy to calculate, given the vibrationalfrequencies
Seek IR spectra of structures 1, 2, 4 and 7: are theydistinctive?
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
Niobium Conclusions
Structures 2 and 4 identi�ed as candidate isomers observed byKnickelbein and Yang
Structures 1, 2 and 7 identi�ed as candidate isomers forcationic system
Structures 4 and 7 have distinctive IR spectra - could beidenti�ed by FIR-MPD?
Greater chance of �nding more than one isomer signi�cantlypopulated in cation system
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
Possible structural isomerism in Rh+6
Taken from M. S. Ford et.al., PCCP, 7, 975 (2005)
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
Candidate minima for Rh6
No need to run simulations (?). Two isoenergetic structuresemerge from the `two-stage' procedure : the octahedron andthe trigonal prism.
Work done with Dan Harding
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
Path1 Energy/kJmol�1
Forward Barrier 156Reverse Barrier 99Products � Reactants 57
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
Path2 Energy/kJmol�1
Forward Barrier 107Reverse Barrier 137Products � Reactants �30
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
Path3 Energy/kJmol�1
Forward Barrier 149Reverse Barrier 52Products � Reactants 97
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
Summary for Rh+6 so far:
The second reaction shown is the clear favourite so far.
All pre-dissociation minima have binding energies between 250and 300 kJmol�1 (wrt separated fragments).
Many more possibilities { other pathways currently underinvestigation
Crucial for comparison is the behaviour of N2O { currentlyunderway
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
Outlook
Potential energy landscape is crucial for suggesting/predictingstructural isomerism.
Master equation is a plausible way forward: but rare-eventdynamics needed to supplement assumed mechanisms.
Need model potential designed for transition-metal clusters!
Complementary calculations (IP) are critical for assigningstructures.
Reaction pathways with small molecules appears promising.
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters
OverviewBackgroundMethodsResults for NiobiumRhodium reactivity
Acknowledgements
Computing facilities of the Centre for Scienti�c Computing,University of Warwick
Helpful discussions with Stuart Mackenzie
T. R. Walsh Dept. of Chemistry and Centre for Scienti�c Computing www.warwick.ac.uk/go/nanoclustersStructural isomerism in transition-metal clusters