PhD - Leuven | More than two weeks ago
Can you help our ions to see better?
Measuring the composition of a material, i.e. the concentration of each of its constituting elements, has been pivotal to proper manufacturing of nanoelectronics devices. In the old days of planar technology, transistors were essentially Si-based and compositional analysis simply meant measuring e.g. the doping concentration in their source and drain (S/D) regions. As technology evolved, the S/D regions eventually became ultra-shallow junctions and it became necessary to follow the variations in composition along the z, i.e. depth, direction. But this was pure evolution. The introduction of three-dimensional fin field-effect transistors (finFETs) in the 14-nm technology node, however, was a revolution for compositional analysis since the latter was no longer only needed along the depth but also along the lateral x and y directions. Ever since, the need for 3D compositional information has only kept rising. This is certainly even more true today, as the industry is about to replace finFETs with gate-all-around (GAA) devices, the architecture of which is reaching unprecedented complexity.
A key compositional depth-profiling technique in nanoelectronics is Secondary Ion Mass Spectrometry (SIMS). In SIMS, a primary ion beam (Cs+, O2+, Ar+, Xe+,...) sputters the sample layer by layer. Among the sputtered material, the ionized species constitute the secondary ions which are accelerated towards a mass-sensitive detector to allow for compositional analysis. As this process is repeated over time, deeper layers are attained and analyzed. Depth-dependent information is therefore naturally acquired as the measurement proceeds.
Thanks to its critically low limit of detection and inherent depth profiling capabilities, this technique has successfully passed almost all the hurdles of the race towards more efficient devices, providing a 1D characterization of the bulk composition and dopant levels of nanoelectronic devices. Recently, novel measurement concepts such as Self Focusing – SIMS (SF-SIMS) have even been introduced to allow for depth profiling in 3D finFET devices. However, to date, SIMS remains a one-dimensional technique for nanoelectronics applications, i.e. it offers no information about the compositional variations in the x and y directions. The main reason for this limitation is the large size of the primary SIMS beam (~ 1 mm) compared to the critical device dimensions (<~ 10 nm).
This project aims at further unlocking the use of SIMS for full 3D profiling, i.e. for compositional profiling not only along z but also along the lateral x and y directions. For this purpose, we propose to use a combination of physical modeling and Machine Learning (ML). Physical modeling will be used to solve the forward problem, i.e. to generate SIMS profiles of synthetic samples. Once the accuracy of the model is validated against real experimental profiles, massive data can be generated and used to feed a ML algorithm. ML is indeed a proven solution to solve the backward problem, i.e. to reconstruct the full 3D device geometry and composition based on SIMS profiles.
This approach will be evaluated on a variety of relevant structures and devices with increasing complexity. The project will start with multilayer (1D, vertical) and line/space (1D, lateral) structures with various geometries, i.e. different thicknesses, widths and pitches. Both front- and back-end of line structures will be tested to ensure the versatility of the approach on a wide range of materials. Two-dimensional arrays of holes or vertical nanowires (2D, lateral) as well as GAA transistors with a simplified stack (2D, lateral and vertical) will then be considered. To reach the ultimate goal of this project, full 3D GAA transistors, including their metallized gates, will be looked into in the final stage of the project.
This work will be done in a collaboration between the advanced patterning (AP) metrology group and materials characterization (Materials and Components Analysis, MCA) department of imec offering expertise in a multitude of characterization techniques in support of this project. The very close collaboration with the process engineers of imec and its partners warrants direct industrial impact.
Required background: physics (solid-state, semiconductors), materials science, engineering
Type of work: 50% experimental, 50% theoretical
Supervisor: Claudia Fleischmann
Daily advisor: Janusz Bogdanowicz, Alexis Franquet
The reference code for this position is 2022-053. Mention this reference code on your application form.