No abstract

The modal method is well adapted for the modeling of deep-groove, high-contrast gratings of short period, possibly involving metal parts. Yet problems remain in the case of the TM polarization in the presence of metal parts in the corrugations: whereas most of the diffraction features are explained by the interplay of an astonishingly small number of true propagating and low-order evanescent modes, the exact solution of the diffraction problem requires the contribution of two types of evanescent modes that are usually overlooked. We investigate the nature and the role of these modes and show that metal gratings can be treated exactly by the modal method.

No abstract

The grating thickness limit l FP between the Raman-Nath and the Bragg diffraction regimes is calculated for an index grating placed in an asymmetric Fabry-Perot resonator with a totally reflecting back mirror and compared with that which was obtained for the same grating with no cavity l M . Owing to the increase of the effective interaction length inside the Fabry-Perot cavity, the stronger the front mirror reflectivity R 1 of the cavity, the smaller the thickness above which the whole diffracted intensity can be concentrated into one unique diffracted beam: l FP ϭ ͓(1 Ϫ ͱR 1 )/(1 ϩ ͱR 1 )͔l M /2.

The diffraction of Gaussian beams on intracavity Bragg gratings is analyzed theoretically. For reasonable waists the associated beam divergence does not significantly influence the diffraction efficiency of such devices. Nevertheless, the tilt angle of the incident beam, imposed by the Bragg resonance condition, strongly reduces the diffraction efficiency at short grating periods. However, the angular selectivity can be maintained if the Fabry-Perot cavity is tuned to the incident beam direction, which allows the use of small-volume holograms together with a dense angular multiplex. This theoretical analysis can be applied to the optimization of the diffraction properties of Gaussian beams on any intracavity Bragg grating, which could then be used for freespace parallel signal processing.