Development of new strategies for Area Selective Deposition Processes

The continuous scaling of the electronic technologies and the development of nonplanar device architectures requires to pattern materials more precisely for the creation of specific features. This is becoming extremely challenging, causing the development of self-alignment bottom-up nanoscale fabrication techniques, such as Atomic Layer Deposition (ALD). It is a thin film deposition technique that involves the sequential exposure of a substrate to two self-limiting half reactions. It allows for precise control of the film thickness, through a layer-by-layer growth, and of the film composition, through an accurate choice of reactants and co-reactants. The growth can be engineered to deposit a vapor phase material just on a predetermined surface in a selective way. This is known as Area Selective Deposition (ASD). To make the process selective a blocking agent is required. The choice of the inhibitor depends on the specific materials and surface chemistry involved. Providing a complete understanding of inhibitor structure/properties relationships is one of the main goal to make advancements in the ASD research. The experimental section of my project focuses on the specific class of organic compounds of primary amines. The aim is to study their inhibiting properties and understand the possible advantages of using this specific class of organic compounds as inhibitors with respect to others. Different classes of materials can already be exploited as inhibitors, resulting in a broaden repertoire of ALD materials that can be selectively deposited. Nevertheless, some issues can impact the effectiveness and the reproducibility of the process, such as adhesion issues or insufficient selectivity. Moreover, the reaction geometry, the precursor flow conditions, fluctuations in temperature, pressure and deposition rate, need to be optimized to guarantee a successful and reproducible experiment. My personal work on Accelerated Discovery for ASD, consists in leveraging an optimised machine learning (ML) model that works as a guide in the decision making process, to choose the best combinations of experimental parameters, including the inhibitor properties required, that give rise to a successful experiment. In this way, new sets of experimental conditions and new classes of inhibitors can be defined to enable selective area deposition. The ML algorithm, based on classification, can identify relationships between process parameters and properties of the deposited material, allowing for the prediction of the behavior of precursor gases, their diffusion and their distribution on the substrate surface. This can compensate any experimental deviation, aiding in identifying potential experimental solutions. The ML model input data are gathered from Scanning Electron Microscopy (SEM) images, after the deposition of both the inhibitor layer and the thin film material. A combination of looking at them, and looking up their corresponding experimental notebooks, allows to gather all the parameters that can influence the ASD process and to visualize the effective growth rate of the thin film material.