AVS2001 Session PS2-MoM: Diagnostics I

Monday, October 29, 2001 9:40 AM in Room 104

Monday Morning

Time Period MoM Sessions | Abstract Timeline | Topic PS Sessions | Time Periods | Topics | AVS2001 Schedule

Start Invited? Item
9:40 AM PS2-MoM-1 Optical Emission Spectroscopy of Exhaust Stream Gases for Improved Process Control
G. Powell (Lightwind Inc.); C.T. Gabriel (Advanced Micro Devices)
Traditional optical emission spectroscopy uses the processing plasma as the light source, limiting the application of OES to when the plasma is on. Another drawback occurs when OES is used in a plasma with a varying magnetic field, which induces signal variations that must be filtered out. We introduce here a new OES technique, where a small secondary plasma is added just downstream of the process chamber to provide a continuous OES light source. The secondary plasma generates emissions characteristic of gases flowing through the chamber or being emitted by the chamber walls, enabling continuous monitoring of species in the chamber. The monitoring can be carried out whether the main chamber plasma is on or off, giving an indication of chamber condition prior to processing a wafer. Signal oscillations from a moving magnetic field are completely removed by monitoring gases downstream of the chamber, simplifying data collection and interpretation for endpointing. Data from using the technique in a MERIE dielectric etch tool are given.
10:00 AM PS2-MoM-2 Peak Wafer Temperature Measurements during Dielectric Etching in a MERIE Etcher
C.T. Gabriel (Advanced Micro Devices)
Using disposable instrumented SensArray APTI wafers, peak wafer temperatures were measured during plasma processing in a MERIE dielectric etch tool. An inorganic low-k dielectric etch process was studied, and many parameters were varied to determine their effect on wafer temperature. Wafer temperature rose rapidly when the plasma was turned on, approaching a stable temperature after about 30 sec. When the lower electrode temperature setpoint or the RF power setting was increased, wafer temperature increased linearly. Peak wafer decreased as backside He cooling pressure was increased. The dual-zone electrostatic chuck allows separate control of center and edge He pressure. These pressures were varied individually or together. Temperature measurements indicated that the zones give reasonably independent control of center and edge wafer temperature. The importance of monitoring and controlling wafer temperature during dielectric etching is also discussed.
10:20 AM PS2-MoM-3 Spatially (z)-Resolved Electron Temperatures and Species Concentrations in Inductively-Coupled Chlorine Plasmas, Measured by Trace-Rare Gases Optical Emission Spectroscopy
V.M. Donnelly (Agere Systems); M.J. Schabel (Bell Laboratories, Lucent Technologies)
Determining the spatial dependence of charged and neutral species concentrations and energies in inductively coupled plasmas (ICP) is important for understanding basic plasma chemistry and physics, as well as for optimizing the placement of the wafer with respect to the ICP source to maximize properties such as etching rate uniformity, while minimizing charging-induced damage and feature profile anomalies. We have determined the line-integrated electron temperature (Te) and Cl-atom number density (nCl) as a function of the distance (z) from the wafer in a chlorine ICP, using trace rare gases optical emission spectroscopy (TRG-OES). The gap between the wafer and the window adjacent to the flat coil inductive source was fixed at 15 cm. The pressure was 2, 10, or 20 mTorr (95% Cl2, 1% ea. of He, Ne, Ar, Kr, Xe) and the inductive mode power was 340 or 900 W. The % nCl (100% = full dissociation of Cl2) increased with power and was highest in the region between mid-gap and the ICP window, reaching nearly 100% at 900 W. Te measured by TRG-OES, characteristic mostly of the high-energy (>10 eV) part of the electron energy distribution function (EEDF), peaked near the source under all conditions except 2 mTorr and 900 W, where a maximum Te of 5.5 eV was observed at mid-gap. The fall-off in Te away from the power dissipation region is mainly due to a preferential loss of high-energy electrons, sensed at high Te -conditions by a relative reduction in the intensity of higher energy Ar emission. We can explain this by both local and non-local effects: Electrons lose kinetic energy in reaching the higher potential energy regions of lower electron density near the wafer (non-local effect). At higher pressures, the mean free path for inelastic scattering by high-energy electrons becomes comparable to the reactor dimensions, causing the EEDF to be relatively hot at the source and cool at the wafer (local effect).
10:40 AM PS2-MoM-4 Diagnostic of Silicon Etch Plasmas by Optical and Mass Spectrometry, Correlation with XPS Surface Diagnostics
N. Sadeghi (University Joseph Fourier-Grenoble and CNRS, France); G. Cunge, R.L. Inglebert, L. Vallier (LTM/CNRS (CEA-LETI), France); O. Joubert (CNRS, France)
As device dimensions are continuously decreasing, the precise critical dimension control during the etch processes becomes a key issue in device fabrication. A better understanding of both plasma composition and plasma-surface interaction during the process are required to obtain a deeper insight on etch mechanisms involved in plasma etching. This work is centered on polysilicon gate etch processes using HBr/Cl2/O2 and HBr/Cl2/O2/CF4 chemistries in an inductively coupled DPS-5200 Applied Materials reactor. The chemical composition of the layers deposited on the trench sidewalls is analyzed by quasi-insitu XPS and is correlated to the relative densities of the different neutral species present in the gas phase and to the composition of the ions impinging on the wafer surface and on the reactor walls. The following conclusions can be drawn: - Contrary to the expectations, halogen ions are not the dominant ionic species impacting on the wafer surface. Mass spectrometry measurements show that in standard etch conditions, Si+ and other silicon containing ions can account by more than 50% of the total ion flux. - In HBr/Cl2/O2 gas mixtures, the passivation layer is formed by redeposition of silicon etch products fragmented in the plasma and then redeposited on the feature sidewalls where they get oxidized by the oxygen present in the gas phase. When CF4 is added into the gas mixture, the passivation layer is formed by condensation of CFx species on the feature sidewalls. - Mass spectra experiments also give some interesting information on the influence of CF4 addition in keeping the chamber walls clean. Consequences on real processes are discussed.
11:20 AM PS2-MoM-6 Electrical and Plasma Property Measurements of a DRIE Bosch Process
I.C. Abraham, P.A. Miller, J.R. Woodworth, C.G. Willison, R.J. Shul (Sandia National Laboratories)
We measured electrical and plasma properties of a DRIE (Deep Reactive Ion Etching) Bosch process used for micromachining bulk silicon. The DRIE process enables the patterning of high-aspect-ratio deep Si features using an iterative inductively coupled plasma (ICP) deposition/etch cycle in which a polymer etch inhibitor is conformally deposited over the wafer during the deposition cycle. The polymer deposits over the resist mask, the exposed Si field, and along the sidewall. During the ensuing etch cycle, the polymer film is preferentially sputtered from the Si trenches and the top of the resist mask due to the acceleration of ions perpendicular to the surface of the wafer. Provided that the ion scattering is relatively low, the polymer film on the sidewall is removed at a much slower rate, thus minimizing lateral etching of the Si. Both the 2 MHz ICP source and the 13.56 MHz cathode components of the floating-potential oscillations (and therefore the plasma-potential oscillations) were measured by a glass-enclosed capacitive probe immersed in the plasma. We used rf-potential and current sensors installed at the output of the chuck's matching network and a calibrated equivalent circuit model to compute the chuck potential waveform. The plasma density and electron temperature were measured using a floating double Langmuir probe. Measurements were made throughout the etch and deposition cycles of the Bosch process. Estimates of the ion energy distribution are presented. This data is expected to increase understanding of the DRIE Bosch process and ultimately improve process control. Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the Department of Energy under Contract DE-AC04-94AL85000.
Time Period MoM Sessions | Abstract Timeline | Topic PS Sessions | Time Periods | Topics | AVS2001 Schedule