A miniature electrostatic element has been designed to selectively apply a ninetydegree phase shift to the unscattered beam in the back focal plane of the objective lens, in order to realize Zernike-type, in-focus phase contrast in an electron microscope. The design involves a cylindrically shaped, biased-voltage electrode, which is surrounded by a concentric grounded electrode. Electrostatic calculations have been used to determine that the fringing fields in the region of the scattered electron beams will cause a negligible phase shift as long as the ratio of electrode length to the transverse feature-size is greater than 5:1. Unlike the planar, three-electrode einzel lens originally proposed by Boersch for the same purpose, this new design does not require insulating layers to separate the biased and grounded electrodes, and it can thus be produced by a very simple microfabrication process. Scanning electron microscope images confirm that mechanically robust devices with feature sizes of ~1 µm can be easily fabricated. Preliminary experimental images demonstrate that these devices do apply a 90-degree phase shift between the scattered and unscattered electrons, as expected. Powerful though modern electron microscopes have become, they still remain relatively poor instruments for imaging thin, unstained biological specimens. Such specimens are weakly-scattering "phase objects", which is to say that the intensity of the transmitted electron beam remains virtually equal to that of the incident electron beam. In this case, the effect of spatial variations in specimen structure is imprinted only on the phase of the transmitted wave. As was eloquently explained by Zernike, perfect images of such objects show no contrast [1]. As a result, weak phase objects must be intentionally viewed in an out-of-focus condition in order to partially convert the phase modulation into an intensity modulation.
of 23The use of objective-lens defocus (and, at higher resolution, spherical aberration as well) is an imperfect way to produce phase contrast in images of biological macromolecules. The problem with that "simple" approach is that such phase shifts vary continuously over the spectrum of spatial frequencies. While the contrast transfer function produced as the result of an optimal trade-off between the phase distortion due to defocus and that due to spherical aberration can approximate the desired value of 1 over a relatively broad range of spatial frequencies [2], this band of high contrast is limited to rather high-resolution features of the image. Since it is necessary to have substantial contrast at low resolution in order to see biological macromolecules, however, it is often necessary to use a much larger amount of defocus. Unfortunately, when a high enough amount of defocus is used to generate contrast on the size scale of individual macromolecules, the phase shifts produced at high resolution can be many times the desired value of /2. As a result, the contrast transfer function then oscillates multiple times in ...