1. Natalie Fey CombeDay, 8 January 2004 @ University of Southampton Development of a Ligand Knowledge Base for Phosphorus Ligands

2. Overview Introduction Computational Approach Statistical Analysis Results Challenges Outlook

3. Introduction Ligand Knowledge Base –mine CSD and other databases geometry of metal complexes (bond lengths, angles, conformations) supramolecular interactions experimental data –supplement by calculated data geometry, conformational freedom electronic structure transition states complexes not structurally characterised

4. Introduction Phosphorus Ligands, PX 3 (X = R, Hal, Ar, OR, OAr, NR 2, mixed) –widespread use as ligands in transition metal complexes tune steric and electronic properties –importance in homogeneous catalysis –established measures of steric and electronic properties steric: Tolman’s cone angle, solid angle, Brown’s steric parameter, Orpen’s S4’ parameter electronic: Tolman’s electronic parameter ( CO ), pK a, PA, IE, E B, C B,  CO Tolman, Brown, QALE (Prock, Giering)

5. Computational Approach Problems with TM Complexes –treatment of large numbers of electrons, electron correlation –geometrical effects of partially filled d-orbitals (spin states, Jahn-Teller effects) –variable coordination numbers and modes –suitable data for verification Density Functional Theory –Jaguar, BP86/6-31G* on ligands, LACV3P on metal

6. Computational Approach Phosphorus Ligands –alkyl phosphines, PR 3 (R = H, Me, Et, Pr, i Pr, Bu, t Bu, Cy, CF 3, asymmetric: 1, 2, 3) –aryl phosphines, PAr 3 (Ar = Ph, o-tolyl, p-tolyl, p-F-Ph, p-(MeO)- Ph, p-Cl-Ph, p-(CF 3 )-Ph, p-(Me 2 N)-Ph, C 6 F 5, 3,5-(CF 3 ) 2 -Ph), model: CH=CH 2 –phosphine halides, PHal 3 (Hal = F, Cl) –phosphites, POR 3 (R = Me, Et, Ph, 4) –amino phosphines, PNR 2 (R = H, Me; cyclic: NC 4 H 4, NC 4 H 8, NC 5 H 10 ) –mixed halides (PH 2 Hal, PHHal 2, PMe 2 Hal, PMeHal 2, Hal = F, Cl, CF 3 (Me only))

7. Computational Approach Complexes –free ligand (PX 3 ) –phosphorus ligand cation ([HPX 3 ]+) –H 3 B(PX 3 ) –OPX 3 –[(PH 3 ) 5 Mo(PX 3 )] –[Cl 3 Pd(PX 3 )] - –[(PH 3 ) 3 Pt(PX 3 )] Variables –energetic: E HOMO, E LUMO, PA, BDE, He (steric) –NBO charges of ML n fragments coordinated to PX 3 –geometrical:  (P-X),  (X-P-X), d(P-M), geometry of M-L fragment (cis, trans effects, L- M-L)

8. Statistical Analysis Bivariate Correlations –linear, non-linear Hierarchical Clustering –identify groups by measuring distance in multi- dimensional space Principal Component Analysis –reduce number of variables by formulation of principal components (linear combinations of variables which account for maximum of variation in original variables) –chemical interpretation of PCs? (steric, electronic ( ,  ))

9. Results Pearson Correlations –identify linearly correlated variables –use to reduce number of variables fewer complexes to optimise simplify interpretation of PCs e.g. [Cl 3 Pd(PX 3 )] - and [(PH 3 ) 3 Pt(PX 3 )]:

10. Results Hierarchical Cluster (Pearson Correlation, STD=1, B & Pt data)

11. Principal Component Analysis (excl. mixed Halides)

13. Principal Component Analysis

15. Principal Component Analysis (excl. mixed Halides)

16. Principal Component Analysis

17. Challenges selection of complexes and variables –treatment of bidentate phosphorus ligands –expansion to other ligand sets chemical interpretation of principal components –steric and electronic effects contribute to variables –reliability of established measures (cone angles) robustness of analysis –variation in ligand set and variables (high correlation) exploration of conformational space –treatment of multiple minima automation of calculations, data analysis, statistical analysis –eliminate data transfer mistakes –reliable error behaviour

18. Outlook started expansion of ligand sets explore model building –predict experimental and calculated data from subset of variables –linear, non-linear explore measures of quantum similarity (Fukui function, HSAB)

19. Acknowledgements Guy Orpen, Jeremy Harvey Athanassios Tsipis, Stephanie Harris Funding: