RECENT ADVANCES IN EXPERIMENTAL PHASE DETERMINATION USING THREE-BEAM DIFFRACTION. E. Weckert, Institut für Kristallographie, Universitåt Karlsruhe (TH), Kaiserstr. 12, D-76128 Karlsruhe, Germany

The direct determination of triplet phases ([[Phi]]_[[Tau]]=[[phi]](-h)+[[phi]](g)+[[phi]](h-g)) can be achieved by three-beam interference experiments where two reflections with reciprocal lattice vector h and g are simultaneously excited. The intensity change of the h reflection during a [[Psi]]-scan depends on [[Phi]]_[[Tau]]. This method has been successfully applied to determine the absolute structure of non-centrosymmetric light atom compounds where anomalous dispersion effects can hardly be exploited. Using highly collimated synchrotron radiation (SR) of good stability experimental phase determination is feasible even for small protein structures. Three-beam interference effects have been observed for a number of protein crystals. Tetragonal lysozyme has been chosen as a test candidate for further analyses. About 630 triplet phases have been determined by the use of SR from a bending magnet (Swiss-Norwegian Beamline) at the ESRF in Grenoble, France. These triplets contain reflections up to 2.5 Å resolution. The mean triplet-phase error compared to the known structure is about 20deg.. Among the measured triplets are 16 [[Sigma]]₁-relations which directly give seminvariant phases. From these single phase and two origin fixing reflections further single phases can be obtained by measured triplets. Phases that could not be assigned were permuted using magic integers. For each permutation a maximum entropy map was calculated. The entropy of each maps as well as a likelihood function proved to be suitable 'figures of merits' for the indication of maps with small mean phase errors. The extrapolating features of the maximum entropy maps can be used to determine further phases. With this phases a first electron density map has been calculated.