The impact of semiconductor laser reliability on silicon-photonic-based systems and paths for future improvement

The shift from fiber optic communication from enterprise to far larger-scale cloud computing has been beset with reliability challenges, and the desire to both reduce costs, and improve reliability, has been difficult to meet. We start with a case study from the rollout of an earlier generation of transceivers that had reliability field failures far above normal targets. After discussing many of the primary causes of field failure, we address proposed responses through more rigorous reliability qualification and ongoing screening requirements and attempts to formalize those as part of a broader standards revisions. The impact of scaling on system reliability requirements is investigated, and those allow us to calculate target reliability requirements for the next generation of lasers or light sources used in digital communication systems, as well as applications such as LIDAR PICs. Options for improving systems reliability using software-defined channels or laser sparing are investigated, leading to attractive options that can improve reliability with existing lasers, rather than relying on innovation in semiconductor laser reliability. Methods of validating improvement are a major challenge, and experimental methods of validation are proposed. For some applications a 10-1000x improvement in laser reliability is being called for: options such as new more robust materials (including quantum dot epitaxy) are discussed. As a final item, a proposal for improved device screening using novel high-speed inspection tools will be shared.