Quantitative visualization of large-variation phase objects

Thesis (Ph.D.)--University of Rochester. College of Engineering and Applied Science. Institute of Optics, 1971.Living cells, thermoplastics, bleached photographic emulsions, optical elements, and aerodynamic flow patterns are all phase objects. It is desirable in the study of these objects to have a technique by which they may be imaged as an intensity variation having a direct relationship between the image intensity and the object phase. The technique of phase contrast achieves this result, but is limited in application to objects having phase variations smaller than a small fraction of a wavelength. This thesis describes a technique which achieves the same result for objects with large phase changes. The phase object is placed in a coherent optical system containing a spatial filter whose transmittance varies as a linear slope plus a constant. The image, which is the derivative of the phase variation biassed by the phase variation itself, is recorded on photographic film with unit gamma and used as the input to a second coherent system. In the second system a filter is used whose transmittance varies as the inverse of spatial frequency, effectively performing an integration. If the constant term is also partially blocked, the modulus of the image amplitude is directly proportional to the object phase, with no approximation on the magnitude of that phase. If the bias amplitude is fairly large, the image intensity is also proportional to the object phase. To implement this system, the spatial filters are produced photographically on Kodak 649F Photographic Emulsion. For the differentiation filter, film non-linearities are compensated by using a sliding mask technique. For the integration filter, the film is exposed by the shadow of a knife edge produced with a wide source. Film nonlinearities are compensated by varying the source shape. The image in the differentiation system is recorded on 649F emulsion and developed at unit gamma with D165 developer. The system is evaluated by using it to image several sample phase objects. The technique is shown analytically and experimentally to produce the desired result for large-variation two-dimensional phase objects. It is shown to be limited in application to phase objects which do not vary too rapidly. This limitation is analyzed in detail and the system is determined to give the correct image if the phase function does not change by more than a wavelength across the total width of the system point spread function. A number of other possible techniques for achieving the same result are also discussed