Improving Silicon-photonics inverse-design printability by leveraging SEM contours for advanced Optical Proximity Correction techniques

Abstract

On a Silicon Photonics integrated circuit, information is carried by the light that propagates within silicon waveguides. The waveguide's geometry determines the functionality. The curvilinearity of Silicon Photonics designs would raise challenges for the manufacturability. However, so far Silicon Photonics design dimensions are considered relaxed by the industry. Also, the type of shapes that are drawn would generally use rather simple geometry objects. This allows the usage of conventional techniques in the different phases of the manufacturability.

Recently has emerged a type of Silicon Photonics design, referred to as "inverse design". This new technique produces designs that are very exotic and quite unpredictable. It shows complex geometries which critical dimensions require innovative Resolution Enhancement Techniques at the different stages of the Optical Proximity Correction flow. The success of these "inverse designs" relies on a very accuratepattern fidelity.

This presentation will demonstrate a flow going from modelling, to OPC and metrology and verification of the manufactured wafer data. This flow permits to tackle the challenges brought newly to the Silicon Photonicsenvironment.

1- Modelling: EP-based modeling leveraging MXP tool developed by ASML. It extracts EP gauges from top-down SEM images. Complex geometries and elaborate 2D-non-Manhattan geometries are captured and feed the modeling engine.

2- OPC is performed after the staircasing of the input design. Model compensation is applied, and manufacturable mask is output.

3- Metrology and verification: After the fabrication, MXP tool is used again. It compares the wafer extracted contours to the original design, leading to quantitative assessment of the pattern fidelity.