ACCURATE CELL DIMENSIONS FROM AREA DETECTOR ROTATION IMAGES

Albert J.M. Duisenberg^l, Rob W.Hooft², Antoine M.M.Schreurs³

l,3Department of Crystal and S'tructural Chemistry, BIJVOET Centre for Biomolecular Research, Utrecht University, Padualaan 8, 3584 CH Utrecht, The Netherlands
2Nonius Delft B. V., Röntgenweg 1, 2524 BD Delft, The Netherlands
3e-mail: schreurs@chem.ruu.nl

For 'big molecules' (proteins) cell dimensions are usually and from sheer necessity derived from one picture (spindle rotation Dw= 0-1^o) showing many active reflections. Precise determination of reflection detector coordinates (x,y) is possible, but the third dimension w 'perpendicular' to the detector is cursed with an uncertainty Dw plus the lifetime of a reflection during w rotation: a few tenths of a degree for equatorial reflections to several degrees for reflections far off the equator. This procedure therefore results in a rather imprecise cell.

With 'small molecules' one such a picture yields too few reflections to obtain a cell from. Therefore multiple ~ 1^o rotation images are made over a total angle Dw typically 5^o-10^o. Here also the reflection coordinates (x,y) are found accurately, but the uncertainty in w still exists. The remedy is to use the detector as a strip chart recorder', like this:

One 'reference image' is made by rotation over the full Dw range, up to 5^o-10^o or even more for small molecules. Then a second image is produced over exactly the same Dw range, but an extra crystal or detector movement is executed simultaneously in one of the following ways:

1. The detector is rotated about the w-axis
2. An extra crystal rotation is applied about the primary beam. On a 3-circle eulerian goniostat this corresponds to simultaneous rotations of j and c with the c-circle perpendicular to the primary beam ( ' j/cscan'), on a kappa goniostat ajudicious combination of w- k- and j -rotations is chosen.
3. The detector is translated.
4. The detector is rotated about the primary beam. This could be achieved in principle with some image plate instruments.

In the second image a reflection impact (x',y') will be farther apart from the original (x,y) in the reference image as it occurs later in the Dw range. From the distance between corresponding (x,y)-(x',y') pairs the exact w value at which the reflection appeared can be found, with comparable accuracy as (x,y) coordinates. Option 2. is to be preferred because it involves only 'natural' movements of the goniostat.

To eliminate small instrumental and crystal centering errors the procedure is executed at four w starting values 90^o apart and reflection coordinates of Friedel pairs are averaged in an appropriate way.

These investigations were supported in part by the Netherlands Foundation for Chemical Research (SON) with financial aid from the Netherlands Technology Foundation (STW grant: 349-4431).