Plasma enhanced atomic layer deposition (PEALD) is one of the most widely adopted deposition methods used in the semiconductor industry. It is chosen largely due to its superior ability to produce ultra-thin high-k dielectric films, which are needed for the further miniaturization of microelectronic devices with the pace of Moore's Law. In contrast to the traditional thermal atomic layer deposition (ALD) method, PEALD allows for high deposition growth per cycle (GPC) under low operating temperature with the help of high energy plasma species. Despite the experimental effort in finding new precursors and plasmas, the detailed surface structures and reaction mechanisms in various PEALD processes remain hard to understand because of the limitation of current in-situ monitoring techniques and the deficiency of the first-principles based analysis. Therefore, in this work, an accurate, yet efficient kinetic Monte Carlo (kMC) model and an associated machine learning (ML) analysis are proposed to capture the surface deposition mechanism and to propose optimal operating conditions of the HfO₂ thin-film PEALD using tetrakis-dimethylamino-Hafnium (TDMAHf) and oxygen plasma. Density Functional Theory (DFT) calculations are performed to obtain the key kinetic parameters and the structural details, subsequently employed in the kMC model. After the kMC model is validated by experimental data, a database is generated to explore a variety of precursor partial pressure and substrate temperature combinations using the kMC algorithm. A feed-forward Bayesian regularized artificial neural network (BRANN) is then constructed to characterize the input–output relationship and to investigate the optimal operating conditions.

等离子体增强原子层沉积（PEALD）是半导体工业中使用最广泛的沉积方法之一。之所以选择它，是因为它具有生产超薄高k介电膜的卓越能力，而随着摩尔定律的发展，微电子器件的进一步小型化是必需的。与传统的热原子层沉积（ALD）方法相比，PEALD借助高能等离子体种类，可以在较低的工作温度下实现高的每周期沉积生长（GPC）。尽管在寻找新的前体和等离子体方面进行了实验性的努力，但是由于当前的现场监测技术的局限性以及基于第一原理的分析的缺陷，各种PEALD工艺中详细的表面结构和反应机理仍然难以理解。₂使用四-二甲基氨基-（（TDMAHf）和氧等离子体的薄膜PEALD。执行密度泛函理论（DFT）计算以获得关键动力学参数和结构细节，随后在kMC模型中使用。在通过实验数据验证了kMC模型后，将使用kMC算法生成数据库以探索各种前体分压和底物温度组合。然后构建前馈贝叶斯正则化人工神经网络（BRANN）来表征输入输出关系并研究最佳操作条件。