Science-based, data-driven developments in plasma processing for material synthesis and device-integration technologies

Abstract

Low-temperature plasma processing technologies is essential for material synthesis, device fabrication, and surface treatment. The development of plasma-related products and services requires an understanding of the multiscale complex behaviors of plasma and the hierarchical integration of plasma generation, energy and mass transports through sheath region, surface reactions, and other processes. The importance of science-based and data-driven approaches to controlling systems is argued. The state-of-the-art of deep learning, machine learning, and artificial intelligence in low-temperature plasma science and technology is reviewed. In this review, the requirements and challenges for plasma parameter prediction and processing recipe discovery are asserted by researchers in the fields of material science and plasma processing. It also outlined a science-based, data-driven approach for development of virtual metrology in plasma processes.

Data-driven science and machine learning methods in laser–plasma physics

“…For detailed descriptions and strict definitions of methods, the reader may kindly refer to the references given throughout the text. Furthermore, we would like to draw the reader's attention to some recent reviews on the application of machine learning techniques in the related fields of plasma physics [80][81][82] , ultra-fast optics [83] and HEDP [84] .…”

Review and future perspective of feature scale profile modeling for high-performance semiconductor devices

“…However, in the future, the simulation technologies will be used for the developing field of "process informatics" in which results of real-time prediction combined with equipped engineering system (EES) data, fault detection and classification (FDC) data, and various plasma monitoring data are feedback to process correction that satisfies the etched profile spec for a certain apparatus or a semiconductor device fabrication line. 99,100 Figure 47 shows an example of a process informatics system performing real-time and automatic process correction for Si gate etching using cloud computing, as proposed by Kuboi et al 99 To realize process informatics, two factors are necessary. The first one is improvement of the simulation model reflecting knowledge originating from accurate measurements for basic properties (electron temperature, electron density, gas density distribution, and plasma-wall interaction) of plasma, gas transportation in the pattern, and the surface condition.…”

“…2 and 3, numerical simulation technologies have been developed to model the mechanism of physical and chemical phenomena during etching and deposition processes to predict feature scale profiles and determine important factors for the process and pattern layout designs. However, in the future, the simulation technologies will be used for the developing field of “process informatics” in which results of real-time prediction combined with equipped engineering system (EES) data, fault detection and classification (FDC) data, and various plasma monitoring data are feedback to process correction that satisfies the etched profile spec for a certain apparatus or a semiconductor device fabrication line 99 , 100 Figure 47. shows an example of a process informatics system performing real-time and automatic process correction for Si gate etching using cloud computing, as proposed by Kuboi et al 99 …”

“…After this improvement, fast calculation comparable to process time is the next issue as the second factor. Currently, machine learning, 100 – 104 fusion physical model with machine learning, 105 , 106 and surrogate models 107 , 108 are adopted to solve this issue. Figure 48 shows an example of a fusion model (or hybrid model) in which incident gas fluxes derived from machine learning using real-time monitoring of plasma and EES were used as an input and feature scale profiles, and damage distributions were simulated by a physical model.…”

Review: Machine learning for advancing low-temperature plasma modeling and simulation

“…The past and present interest in the context of plasma processing may be attributed to the potential for hidden pattern detection in correlated data and, particularly, the efficiency in related optimization tasks 4 – 12 …”