Temporal response of an excitation engineered 1/spl times/16 splitter

The temporal response of a funnel-type 1/spl times/16 splitter is investigated. Improper input alignment is shown to generate temporal satellite pulses which could negatively impact excitation-engineered splitters for future very-high repetition rate pulse train generators.

Techniques for generating very-high repetition rate bursts (500 GHz to 1 THz) of short optical pulses using an arrayed waveguide grating (AWG) [1] have recently been reported [2], [3]. The earliest work [2] demonstrated, the concept of generating a-high repetition rate pulse burst from each femtosecond input pulse, with a Gaussian temporal roll-off. This-roll-off is caused in part by the Gaussian excitation profile of the various waveguides in the AWG array. In later work [3], we loss-engineered the various waveguides in the array, resulting in a flat-topped pulse burst. While effective, the loss-engineering obviously leads to increased device insertion loss which could significantly impact potential system use of the technique. In the current work, we investigate the femtosecond response of a 1×16 excitation engineered splitter, which in the future could play the same role as loss-engineering but without the excess insertion loss. While excitation-engineered splitters have been previously reported [4], here for the first time to our knowledge we report the short-pulse response of this design methodology and illuminate critical device alignment issues that impact the utility of the excitation-engineering scheme.