Define a measurement standard to monitor local variabilities of immersion lithography baseline for photonic devices

The Line-Edge Roughness (LER) is critical for photonic devices, as it can induce significant optical losses in the data transmission of a PIC (Photonic Integrated Circuit) [1]. Indeed, due to SiO2/Si interface irregularities or nm-scale roughness of the sidewalls, optical waveguides are prone to propagation losses by light scattering. With the down-scaling of PIC dimensions, performances of narrower waveguides become more sensitive to width variations [2], as well as LER [3]. Therefore, the main goal is to minimize these optical losses by optimizing the lithography process at the most critical fabrication steps of the PIC. In this context, we investigate the impact of 193 nm immersion lithography process variations on LER by process monitoring in high volume manufacturing (HVM) environment. Those parameters can go from chucks or tools mismatches to the impact of reworks, as well as the slight drift of the tools over time. This allows us to establish a baseline for our lithography process, which will be more detailed in the article, before attempting on more consequent changes to reduce the LER and improve the optical loss of the PIC. Moreover, this study is supported by an appropriate measurement methodology within which the image acquisition conditions (e.g., pixel size and field of view), the measurement algorithm for contour extraction and the frequency analysis parameters are well defined, using a new generation CD-SEM tool and a dedicated off-line software. The methodology allows us to implement a LER frequency analysis by extracting power spectral density (PSD) curves for waveguides characterization.