Development of a Vacuum Line Downstream Detector for Monitoring Nano/Micro-Particle Generation in Process Chambers

Principal Investigator: Hani E. Elsayed-Ali

Institution: Old Dominion University

PROPOSAL TITLE:
Development of a Vacuum Line Downstream Detector for Monitoring Nano/Micro-Particle Generation in Process Chambers

RESEARCH SUMMARY:

A. Objectives and expected significance
The goal of the proposed project is to develop a sensitive downstream nano/micro-particle detector to be used for diagnostics and quality control of the operation of semiconductor plasma processing reactors without the need for added diagnostics equipment on the processing chamber. The detector is to be installed on the vacuum line with minimal or no effect on the pumping speed.

Plasmas are extensively used in microelectronic fabrication steps, e.g., thin film deposition, etching, surface cleaning, and photoresist ashing. Various processes in microfabrication generate nano/micro-particles (particulates) [1, 2], and these particulates are responsible for about 75% of the total defects detected in semiconductor fabrication [3]. Formation of particulates in plasma processing reactors is a major source of defects in semiconductor fabrication. Particulates formed in these discharges range in size from the nanometer to the micrometer.

Most observations of particulate formation in plasma processing devices were conducted on capacitive RF discharges [4]. Particulates grow by homogeneous reactions in the gas phase or by heterogeneous interactions of the plasma with the electrode surfaces and chamber walls [4]. In a plasma, the particulates are negatively charged due to the electron flux (number/cm2 ) bombarding them being higher than the ion flux. However, in the plasma sheath near the electrodes, the particulates are positively charged due to the ions being the predominant species in that region. Charged particles are often observed suspended in the sheath and at the plasma-sheath interface [5]. These particles can be pumped out of the plasma reactor during discharge operation or after the discharge is turned off, or can fall on the wafer or the reactor surface. In addition to capacitive RF discharges, particulates are observed in plasma-enhanced chemical vapor deposition due homogeneous nucleation [6]. Magnetron sputtering also generate particulates near and on the sputter target. These particulates migrate, by ion drag, in the processing chamber and can end up on the substrate [5]. Nanoparticles were observed seconds after initiation of the magnetron discharge. The main mechanism suggested for their formation is inhomogeneity of the plasma at the target surface leading to filament formation and its vaporization [5].

Laser light scattering (LLS) was used to probe particulate formation by monitoring changes in the spatial distribution of a laser passing through the plasma [4]. However, LLS needs sufficient scatters to create a detectable change in the scattered laser beam. Using LLS, Roth et al. [7], identified the plasma/sheath interface as an area of the discharge where particulates form and grow. Other methods for detection of nanoparticles is mass-spectrometry direct line observation of the plasma [8]. Mass-spectrometers detect molecules, but commercial units are not designed to detect the relatively large masses of micro-particulates. They are limited to detecting mass-to-charge ratio m/q of up to ~10,000 and require line-of-sight observation. Laser-induced breakdown spectroscopy (LIBS) was also suggested as a real-time method of monitoring particles generated during semiconductor manufacturing [9]. For LIBS, a particle is ablated by the laser emitting the spectra of its composition. This requires the particle to interact with the laser near its focus, where the laser intensity is high. Characterization of the particulate shape formed in capacitive discharges was conducted by ex situ scanning electron microscopy (SEM) [10]. Since these earlier studies, many researchers have investigated the mechanisms of particulate formation in capacitive discharges [4]. The initial growth of sub-nanometer particles is mainly due to chemical processes. Coagulation of the negatively-charged particles and the positively-charged ones is the main mechanism for growth of the nanoparticles to the micrometer scale [4].

We propose to develop an online particulate detector based on a charging electron beam, acceleration of the charged particulates, and microchannel plate (MCP) detection. The particulate detector will be installed on the vacuum line downstream of the processing chamber and will detect the presence of particulates during discharge operation without characterizing its m/q.

B. Detector design concept
The detector will be tested on an ATC Orion-5 sputter coater from AJA International equipped with RF and DC magnetrons. The elements of the particulate detector design are:

1. A down-flow vacuum line connected to the vacuum chamber, as shown in Fig. 1. The particulate detector will be useable on different processing chambers. For chambers operating above 10-3 Torr, the detector beam-line can be differentially pumped.

2. As the particulates are pumped with the processing gas, they are charged negatively by an electron beam directed between the grounded grid and the –V grid in Fig. 1 (–100 to –500 V applied to it). The electron gun produces 25 keV, ~1 mA beam. For 25 keV, 10 mA electrons, a 1- μm Al2O3 particle passing through it in 0.5 ms charges to 106 e, where e is the electron charge [11]. For detection, we only need a charge of 1 e on the particulate. Therefore, we will limit the electron current to get ≲103 e/particulate. Electron ionization is used in mass-spectrometry. The electron ionization cross-section for most materials peaks at a few tens of eV then drops quickly with the electron energy [12]. Electrons with 25 keV energy are highly effective for particulate charging but have a very low cross-section for gas ionization. The use of electrons to charge particulates with very limited ionization of gas is different from their use in mass-spectrometers where the electron energy is <1 keV for ionization. This is a new approach for particulate detection. Secondary electrons will be bent away from the axis by the Lorenz force of a permanent magnet, while F has little effect on the particulates due to their low velocity. [caption id="attachment_2382" align="aligncenter" width="3368"] Fig. 1. Schematic of the particulate detector attached to sputter coater (not to scale). The detector can be reduced in size to fit in a 2.75-in Conflat flange with a length of about 30 cm.[/caption]

3. The accelerated charged particulates will drift till they hit the MCP. The MCP detection efficiency depends on the particulate energy and charge, and is ~10% for negatively charged 0.1- 2 μm particles having a few electron charges e per particle [13]. Although a 10% MCP detection efficiency is sufficient for our application, we expect a much higher detection efficiency since the charge will be ≲10^3 e/particulate. MCPs are used in commercial secondary ion mass-spectrometers (SIMS) and operate for thousands of hours. There are highly developed methods for MCP protection from dust particle damage when used for space exploration, but we plan to use unprotected MCPs, which should provide thousands of hours of operation. If needed, the front MCP can be replaced at a cost of ~$350 [14].

C. Study plan
1. Assemble the electron source using a tungsten filament that is heated with alternating current and kept floating at a potential of –25 kV. A simple electron lens will be used to get an electron beam of ~1 cm diameter. A phosphor screen, placed as shown in Fig. 1, will be used to monitor the beam diameter, while a Faraday cup (not shown) will be used to measure the beam current.

2. After assembling the particulate detector, as shown in Fig. 1, we will operate the magnetrons in DC and RF modes for sputtering different targets (e.g., Si, Cu, C) and detect negatively-charged particles accelerated by the –V potential, without and with electron beam charging.

3. Conduct a preliminary study of particulate generation with sputter time and magnetron power.

4. Insert flat Si pieces with Au conductive layer and carbon-coated TEM grids in the path of the particulates and operate the sputter coater and the particulate detector. The Au-coated Si and TEM grids will be used to observe particulate hit density using high-resolution SEM and TEM. Characterize particulates and correlate their density and size with MCP signal.

D. Expected Impact/ Risk Associated With Effort
A downstream detector for monitoring particulates in process chambers is a valuable tool for thin film deposition and plasma etching optimization and yield improvement. This detector will be used to identify processing conditions minimizing particulate formation and to alert the operator of their presence, allowing for corrective action during processing. At processing pressures >10-3 Torr, differential pumping will be needed. Further development of the detector can yield m/q by pulsing the charging electron beam and detecting the time-of-flight of the particulates.

E. Anticipated expenditures
GRA $8,000, faculty summer salary $3,000, microchannel plate imaging detector $2,800, Conflat flange with high-voltage feedthrough $1,200.

F. Relation to existing and other proposed work
The work will utilize facilities established with ODU funds and previous grants. The proposed work has no relation to any of the PI’s current proposals or grants/contracts.