Session FT1: Plasma Etching I

Dynamics of Plasma Atomic Layer Etching: Molecular Dynamics Simulations and Optical Emission Spectroscopy

Atomic layer etching is intrinsically dynamic as it involves sequential and repeated exposures of a surface to be etched with different species at different energies. The composition and structure of the near surface region changes in both time and depth. Full understanding of this process requires resolving both temporal and spatial variations. In this work, we consider silicon (Si) atomic layer etching (ALE) by alternating exposure to chlorine gas (Cl2) and argon ions (Ar⁺). Molecular dynamics (MD) simulations are compared to experimental measurements with the aim of better understanding the dynamics of ALE and to test the simulation procedure. MD predictions of etch per cycle and species leaving the surface during the Ar⁺ step are compared to experimental measurements. Optical emission measured just above the surface being etched can be related to etch products and can therefore be directly compared to simulation predictions. The simulations capture the measured initial product distribution leaving the surface and match the measured etch per cycle reasonably well. Simulations demonstrate the importance of ion-induced surface damage and mixing into a layer below the surface, the depth of which depends mainly on ion energy. But the experiments also suggest there is more Cl mixed into the layer than the MD procedure predicts. The implications of these observations are discussed.   

Studies on the discharge characteristics and atomic layer etching of high aspect ratio patterned-wafer in radio frequency biased inductively coupled plasma with Ar/C4F6 gas mixture

Recently, as the critical dimension decreases and the aspect ratio of the pattern increases in semiconductor manufacturing industry, precise control of the plasma etching process is urgently required. This can be achieved through process design based on understanding of the discharge characteristics, which depends on the chamber configuration, plasma source, and type of processing gas. Among them, C₄F₆ (Hexafluoro-1,3-butadiene) is a next-generation etching gas with a low global warming potential and excellent etch selectivity. A radio-frequency (RF) biased inductively coupled plasma (ICP) has been actively used in the etching process because it can control ion energy with high plasma density. In this study, we comprehensively investigated the discharge characteristics and atomic layer etching of high aspect ratio patterned-wafer in RF biased ICP with Ar/C₄F₆ mixture¹. Our investigation includes discharge mode transition, instability, and electron density characteristics by C₄F₆. Considering these discharge characteristics, the etching thickness was controlled at the atomic-level in the high aspect ratio pattern etching.   

Peculiarities of Ion and Neutral Transport in Plasma Etch Applications

The continuous shrinkage of critical dimensions, introduction of new materials, and increase of integration complexity in semiconductor technology imposes stringent requirements to plasma processing techniques. The full utilization of existing process techniques and the development of new ones is impossible without a detailed understanding of the transport of charged and electrically neutral species within small etch features on a nanometer scale. At the same time, control over the transport of reactive ions and neutrals for surface interaction remains a vexing challenge in the industry and requires novel methods for further process improvement.

This presentation reviews a few important aspects of ionic and neutral species transport through microscopic dimension features in technological plasmas. We focus on methods for effective control over the key process parameters, including the ion and radical fluxes, ion energy, and ion angular distribution and consider advanced manipulation techniques for effective delivery of the reactive species to the targeted areas. Additionally, we describe various experimental methods to characterize ion and neutral transport and charge effects through the features with conductive and insulating sidewalls. Ion and neutral transport characteristics are monitored using commercially available and custom-made via structures, which are attached to the front of the sampling orifice of a mass spectrometer and a very sensitive electrostatic probe. Applying accelerating voltages of different tailored waveforms and frequencies to the structures enables a comprehensive charge effect study for conductive and insulating sidewalls. The relative radical densities for diatomic gas plasmas can be derived from the ionized mass intensity counts at electron energies below the dissociative ionization thresholds. The transport of atomic radicals (O* and N* in this study) is characterized by measuring their fluxes at the sample exit as a function of the aspect ratio.   

Atomic Layer Etching of Silicon in HBr-Containing Plasmas

Atomic layer etching (ALE) of Si(100) was investigated in an inductively-coupled  plasma (ICP), with a constant flow of He/Ar/O₂ and periodic 2s injections of HBr. The ICP and bias powers were on for 4s and 3s, respectively. Optical emissions above the radio frequency bias substrate surface were detected and monitored for Si, SiBr, OH, HBr⁺, Br, O and Ar. Etching rates were measured on SiO₂-masked substrates by laser interferometry. Optical emission spectra and etching rates were obtained over the 7s ALE cycle as a function of the time between HBr exposure and energetic ion bombardment, O₂ addition between 0 and 1%, addition of He, and DC self-bias voltage between -30 and -175 V_DC. At any bias power, below a threshold level of O₂ addition, steady etching of Si was accompanied by a periodic instantaneous rise and then decay in emissions from Si, SiBr and HBr⁺ during each of the energetic ion bombardment steps, with little OH emission. The presence of HBr⁺ emission after the flow of HBr had left the chamber suggests that HBr desorbed during the energetic ion bombardment step. As O₂ flow was increased at a constant bias power, etching stopped and emission from OH became the dominant feature in the UV. The amount of O₂ added before etching stopped was an increasing function of increasing bias power. For conditions near the transition between etching and no etching, the etching per ALE cycle decreased to immeasurably slow over multiple cycles as the surface becomes oxidized. The etching rate under conditions with minimal oxidation was 3 nm/cycle, even with minimal overlap between HBr injection and ICP power. ALE with HBr will also be compared with Cl₂ as the reactive gas.   

OES-based Monitoring Method of Non-maxwellian EEDF and Radical Density for Etch Process Control in Ar/SF6/O2 VHF-CCP

Fine control and monitoring of etch processes are essential due to the increasing difficulty and complexity. To achieve this goal, reliable real-time plasma information (PI) monitoring sensor technology and virtual metrology (VM) for process prediction must be developed. In conventional etch plasma monitoring technique using optical emission spectroscopy (OES) signal, the maxwellian electron energy distribution (EEDF) assumption and simple actinometry is used. However, etch process plasmas, which target sub-nanometer-scale high-aspect-ratio (HAR) processes with very-high frequency (VHF) power, the EEDF shows significant tail deviations due to electron heating mechanisms, such as the bounce resonance heating and plasma series resonance. This becomes a limiting factor in accurately monitoring the electron temperature and radical density of the etch plasma.

This study proposes a method to obtain non-max. EEDF and radical density using OES data in Ar/SF6/O2 VHF-CCP. Non-max. EEDF can be monitored using OES data and 1s state included collisional-radiative model (CRM). The actinometric coefficient, which is proportionality constant between the line intensity ratio and the density ratio, is determined using the obtained EEDF and excitation cross-section data. The proposed PI-OES (Non-max. EEDF and Radical Density) monitoring method can be effectively utilized as a real-time process monitoring technique for mass volume production (MVP) processes, ensuring reliability and accuracy.