Page 307 - Phase Space Optics Fundamentals and Applications
P. 307

288   Chapter Nine


               becomes d. In contrast, the cross-terms superposed with self-terms at
                 m = m/d can only be associated with pairs of self-terms separated
               by integer multiples of 2/d, and as a consequence the modulation we
               observe has a period of d/2, that is, one-half of what we obtain for
               interlaced cross-terms.
                 Fresnel diffraction corresponds to a horizontal shear of the phase-
               space distribution, and we see without difficulty that the PSD does
               not change if the shear equals d at   = 1/(2d). All other discrete
               lines are automatically sheared by an integer multiple of the period
               d, and we obtain the original phase-space distribution. The line S in
               Fig. 9.4 corresponds to the tilt we would observe for a vertical line at
               the input and is equivalent to the WDF of the associated point-spread
               function of free space in Eq. (9.14). Using Eq. (9.15), the Talbot length
               can now be deduced from d =  z T /(2d), or

                                             2d 2
                                        z T =                       (9.18)

               It is immediately clear from the PSD in Fig. 9.4 that self-images can
               be found at any integer multiple of the Talbot distance. This is re-
               lated to the fact that the WDFs (and not only their projections) be-
               fore and after shearing are identical, which proves that self-imaging
               recovers not only the intensity distribution, but also the complex am-
               plitude of the input signal. Also note that once the WDF of a periodic
               function is known, the self-imaging condition can be deduced with a
               minimum of mathematical formalism. Furthermore, the quantitative
               result can only be obtained correctly by including the cross-terms,
               namely, the interlaced terms at half intervals, into our analysis. This
               is of significance as PSDs are often constructed as heuristic notions of
               phase space rather than from the results of a rigorous evaluation of
               the WDF.
                 ThisrigorousanalysisoftheWDFalsoprovidesaccesstotheFresnel
               diffraction amplitude at z T /2. The diffraction patter is often described
               as an additional grating image which is reversed in contrast. 30  To un-
               derstand this notion, we again turn to Fig. 9.4. If we consider the shear
               associated with line H, which is only one-half the shear necessary for
               Talbot self-imaging, we again recover the phase-space distribution of
               the input signal, however shifted in x by d/2 compared to the distri-
               bution in Fig. 9.4. By inspecting the points of intersection between line
               H and the horizontal delta lines, we can verify that the lateral shift at
                 m = m/d in fact is a multiple of d, while it is an odd multiple of d/2 for
               the interlaced frequencies. Thus the terms at even multiples of 1/(2d)
               and at odd multiples of 1/(2d) only register because the base period
               of the modulation at   m = m/d is d/2, that is, one-half of what we
               might expect intuitively for a phase-space distribution of a periodic
               signal. For typical grating profiles, including, for instance, the Ronchi
   302   303   304   305   306   307   308   309   310   311   312