Prospects of a polymer-waveguide-based board-level optical interconnect technology

Abstract

In the long-distance telecom, local-area, and rack-to-rack link classes, optical interconnects have gradually replaced electrical interconnects. We believe that this trend will be continued in the short-distance card-backplane-card datacom link class. Convincing arguments for the predicted transition from electrical to optical interconnects are bandwidth*length advantages, density benefits [1-3], crosstalk reduction, and finally cost considerations. Based on this forecast, we currently develop a board-level optical interconnect technology facing several challenges, such as I) the manufacturing of reliable polymer waveguides, II) the elaboration of simple light-coupling concepts, III) the development of high-speed electro-optical modules, and IV) the application of cost- efficient packaging approaches. The successful mastering of all these tasks is a prerequisite for convincing high-speed system designers and porting optical interconnect technology into future product development plans. In this paper, we will present different achievements of our optical interconnect technology, e.g.: 10 Gb/s per channel over 1 m link length, optical link propagation loss below 0.05 dB/cm at 850 nm, linear link densities up to 16 channels/mm, feasibility of 2D channel arrays (e.g. 4 * 12), a fully passive, low-cost alignment concept with a position accuracy of ≤ 5 μm, enabling coupling losses ≤ 0.5 dB, and electro-optical transmitter and receiver modules operating at 10 Gb/s per channel. Finally, we will report on the successful realization of a 12 × 10 Gb/s card-to-card optical link demonstrator.