Integrated silicon photonics, which consists of using manufacturing processes from the microelectronics industry to produce photonic components, is considered a critical future technology for very high-speed communications and computing applications.
The creation of silicon photonics devices requires the manipulation of fully curved designs (so-called non-Manhattan). This gives rise to numerous challenges during their manufacturing, and particularly during the design stage of advanced photolithography masks. In order to determine the optimal masks, optical effects compensation (OPC) algorithms are systematically applied. The latter are particularly difficult to implement in the particular case of curvilinear patterns.
The accuracy with which OPC models are able to anticipate pattern printing performance can be assessed using CD-SEM on specific, simple, small structures in a single orientation. However, Photonic devices (for example, waveguides) are continuous, up to a few millimeters long, and cover all orientations in space. A broad and precise metrology of such objects does not exist, making it impossible for OPC engineers to diagnose the quality of the devices produced on the product.
The objective of the thesis is to develop a method for precise, large-scale dimensional measurement of Photonic structures. In particular, we will seek to implement solutions for stitching SEM images and extracting contours, with the development of dedicated metrics. The thesis proposes in particular:
- the study of metrological and scripted solutions to enable characterization by CD-SEM on a large scale, typically by combining several images.

- the implementation of extraction of contours of curved patterns on recombined images.

- the development and implementation of innovative 2D metrics to allow the measurement of curved objects, then their comparison with each other (or with a reference).

- the inclusion of real large-scale contours in optical simulation software (Lumerical, FDTD) to characterize the real performance of the devices.

The thesis will take place for 3 years between the STMicroelectronics site (Crolles) and that of CEA-LETI (Grenoble), in a context of strong collaboration between the teams of the two organizations.

The doctoral student will have access to clean rooms and state-of-the-art industrial and/or research equipment, as well as commercial reference software. You will benefit from all the technical expertise of the supervisory teams at STMicroelectronics and CEA-LETI in photolithography, metrology, image processing and applied IT development (Python).
?