Complexes and aggregates of chlorophylls

Chlorophyll (Chl) molecules can form complexes in two important ways: by ligation at the magnesium atom and/or by hydrogen bonding at the keto- carbonyl group. Under certain conditions these processes may give rise to dimer formation. This thesis describes some properties of complexes and dimers of Chl a , Chl b and pheophytin (Ph) a .

Using optically detected magnetic resonance and fluorescence methods, the positions of the lowest excited singlet state and the zero field splitting parameters D and E of the lowest excited triplet state have been determined for complexes of these molecules. The combination of these parameters opens the possibility to analyze a complicated fluorescence spectrum. Different fluorescence bands were assigned to various complexes, such as singly ligated, doubly ligated, and hydrogen bonded Chl. It is found that for Chl a both the lowest excited singlet state and the D value decreases in energy as the number of molecules bonded to Chl increases, even more in the case of ligation than in the case of hydrogen bonding. Probably, the aldehyde group in Chl b forms hydrogen bonds more easily than the keto group.

For dimers the following was found: two different Chl a dimers could be distinguished, the "pure" dimer (PD) and the "special pair" (SP). In the PD both monomer parts are bonded directly to each other, whereas in the SP two small interstitial "glue" molecules are present. Both the zero field splitting values and the positions of the lowest excited singlet state are different for the two dimers, whereas for the PD a triplet spin polarization pattern was found different from that of the monomer. Oh the basis of these data we suppose that the SP consists of two approximately plane-parallel monomer parts, as proposed earlier, whereas both monomer parts in the PD are not plane-parallel. For Chl b the dimerization behaviour is much more complicated than for Chl a , due to the presence of an aldehyde group in Chl b . Possibly, Chl b dimers form by interactions of the aldehyde groups, contrarily to those of Chl a , in which keto groups are most important. Ph a dimers form by a different mechanism, due to the absence of a magnesium atom. In this case dimers form by pi-piinteractions between the monomer parts; the binding energy is smaller than in Chl dimers.

In order to investigate the geometry of these dimers, we also performed high resolution proton nuclear magnetic resonance measurements. These experiments yielded two sets of mutually independent data, i.e. for each proton a nuclear spin lattice relaxation time and a chemical shift. Both sets can provide geometrical information on the dimers which were studied:.

(a) A calculation method was developed to determine the dimer geometry with the help of the experimentally determined spin lattice relaxation times and the monomeric geometry. It is possible to give a quantitative description of spin lattice relaxation times in terms of a rotational diffusion tensor and a geometrical structure. The results of these calculations are:.

- the PD consists of two approximately perpendicular molecules;.

- for the SP we found a structure corresponding to that previously postulated by Shipman & Katz.

(b) A model previously introduced by others, and describing proton chemical shifts in a porphyrin molecule, has been extended to Chl dimers. Application of this model to the abovementioned Chl a dimers resulted in similar geometries as those found in the spin lattice relaxation calculations.

We also investigated dimer geometries utilizing experimental data on the excited states. To this end we applied a simple model including both "exciton" and "charge resonance" contributions to the excited state singlet and triplet wave functions. Again, the same geometries for the PD and for the SP were found. The Ph a dimer is found to consist of two plane-parallel Monomers, somewhat shifted relative to each other.

An additional, more general conclusion is, that ignoring "charge resonance" contributions to the triplet state wave function may result in unreliable geometrical statements, even if these contributions are small (≈5%)..