Abstract

In the past ten years it was proved both theoretically and experimentally that cubic spatial solitons can be exploited to achieve fast all-optical processing of optical data. Taking benefit of the quasi-instantaneous response of Kerr type nonlinearity transient waveguides were generated by spatial solitons into homogeneous nonlinear materials[1]. It was also possible to modify at ultrafast speed the geometry of the waveguides thanks to the various interactions that may occur between soliton beams: couplers for signal splitting or combining and switches were realized using parallel solitons coupling [2] or solitons collision and fusion [3]. Spatial soliton based optical processing look attractive for routing high data rate streams. In the case of Kerr type nonlinearity however, the envisaged devices need a planar waveguide structure to avoid two-dimensional transverse instability and the peak powers required to reach self-guiding are close to 1 kW, even with the use of material of high nonlinearity. It was recently shown theoretically first and then experimentaly [4] that spatial solitary waves propagation can take place also in crystals exhibiting a substantial quadratic nonlinearity. This opened up new hopes on the use of soliton in optical processing. It was demonstrated that quadratic spatial solitary waves (QSSW) are stable even in the case of bulk propagation (2D-QSSW) on the contrary to their cubic counterparts. Moreover the advent of new materials, some of them of organic type, with high nonlinear coefficient lets expect that the self-guiding power threshold could be decreased below the above mentionned value for cubic spatial solitons. As it happened for interactions and collisions between kerr-type solitons, those between QSSWs are expected to offer the same promising field of applications. The problem of QSSWs collisions has been in fact recently addressed; as long as one refers to type I interactions, the problem has been treated numerically for both the 1D and 2D cases [5], analytically and experimentally only for the 1D case [6]. On the other hand, if type II (or vectorial) processes are concerned, the problem of QSSWs interactions has been faced only numerically or analytically both in the 1D and 2D cases [7].

© 1998 Optical Society of America