SPECIMEN FLATNESS OF THIN CRYSTALLINE ARRAYS - INFLUENCE OF THE SUBSTRATE

The extreme degree of specimen flatness (i.e. planarity) required for high-resolution electron diffraction and electron microscopy at high tilt angles cannot be realized with thin, sheet-like crystals of biological macromolecules, just on the basis of the intrinsic stiffness of the specimen itself. In an effort to improve the rate of success at which suitably flat specimens are prepared, this paper analyzes several different factors that can either limit or enhance the specimen flatness. If specimens are adsorbed (by attractive forces) to a support film, such as evaporated carbon, which itself is not flat to atomic dimensions, quantitative calculations show that it is quite likely that the specimen will be too wrinkled to be used for high-resolution studies. Adsorption to an air-water interface is more likely to result in the necessary degree of flatness. Repulsive interactions, which might be used to "sandwich" a specimen between two interfaces, are estimated to be too "soft", i.e. too long-range in character, to be effective. Finally, if only one edge of a specimen sticks firmly to a substrate, then surface tension forces can pull the specimen taut over the surface of the substrate, so that the the specimen itself can be more flat than the surface of the substrate upon which it is deposited. A second, important consideration in many studies is the fact that cooling the specimen to low temperature can result in specimen wrinkling, because of the fact that the biological crystal has a much larger coefficient of thermal expansion than that of the evaporated carbon film. In this case one expects that cooling-induced wrinkling might be reduced by using a metal support grid which has a smaller thermal coefficient than that of the carbon film. The validity of this qualitative idea is supported by experiments which show that cooling-induced wrinkling of glucose-embedded purple membrane can be prevented if molybdenum grids are used rather than copper.