AVS1996 Session MS-WeA: New Sensors for Diagnostics, Fault Detection, and Process Controls

Wednesday, October 16, 1996 2:00 PM in Room 201A

Wednesday Afternoon

Time Period WeA Sessions | Abstract Timeline | Topic MS Sessions | Time Periods | Topics | AVS1996 Schedule

Start Invited? Item
2:00 PM MS-WeA-1 Sandia/Sematech Contamination Free Manufacturing Research Center Novel Sensor Development Activities for Enhanced Process Control
S. Cohen (IBM/Sematech); M. Kelly (Sandia National Laboratories)
The Sandia/SEMATECH Contamination Free Manufacturing Research Center was founded in 1992 with the goal of providing research and development support to the US semiconductor industry in the area of defect reduction in manufacturing equipment and processes. The program encompasses topics in equipment/process modeling, advanced wafer cleaning, water use reduction, organic contamination, wafer-map defect data analysis and contamination sensor development . The contamination sensor development activity focuses on producing advanced tools for the semiconductor industry by development and commercialization of in-line cost-effective sensors for measurement of contaminants in critical process tools. There are three phases to our sensor development activities. Initially our efforts focus on sensor feasibility testing whereby several potential sensors are evaluated for technical and business issues such as sensitivity, reproducibility, cost, size, etc. After this initial screening, subsequent refinement of one or more chosen sensors occurs through a beta-test in a manufacturing environment to ensure it's viability for manufacturing applications. Lastly, commercialization with an existing supplier is critical in ensuring availability of the sensors for the industry. In this talk we will give examples of sensors at all three stages in the evolution. The topics to be covered include: a. Detection of trace moisture in vacuum tools using the residual gas analyzer (RGA) and the porous silicon capacitor (PSC) sensor. b. Monitoring of wall residues and gas phase contaminants in deposition and etch chambers using optical emission spectroscopy (OES), quartz crystal microbalance (QCM) monitors and fiber optic reflectance (FOR) monitors. c. Trace moisture detection in corrosive gases using an FTIR spectrometer with a specialized gas cell.
2:40 PM MS-WeA-3 In Situ Mass Spectrometry for Real Time Process Sensing in Amorphous and Selective-area Silicon Plasma CVD
A. Chowdhury, W. Read, G. Parsons (North Carolina State University)
Demands for improved product yields, throughput, and reproducibility in semiconductor manufacturing requires real-time process sensors for reaction monitoring and equipment analysis. Plasma enhanced processes are widely used for device fabrication, and involve complex reactions that are difficult to observe, analyze, and quantify. We have used in-situ mass spectroscopy during plasma deposition to analyze chemical processes, and to detect faults and fluctuations in real time. We are using a double-differentially pumped quadrupole mass spectrometer system, designed for optimized flow, to sample the exhaust stream of a large area PECVD reactor which is used for hydrogenated amorphous silicon (a-Si:H), silicon nitride, and selective area microcrystalline silicon deposition for thin-film transistors. To analyze a-Si:H, we monitored reactants (SiH\sub 2\\super \+\\ and SiH\sub 3\\super \+\\) and products (H\sub 2\\super \+\\), and found that the reactant signal decreased and the product signal increased significantly when the plasma was initiated. The integrated differential area corresponding to silane loss correlates with H\sub 2\ production and measured film thickness, demonstrating real-time plasma process metrology. We have also analyzed a pulsed-gas process for selective area silicon PECVD, where SiH\sub 4\ flow is modulated into a H\sub 2\ plasma. In this process, we are able to quantify the silane mass balance during deposition, as well as observe etching during the hydrogen plasma exposure, confirming that the process proceeds through selective etching by atomic hydrogen. We will also discuss results of plasma process simulation, and its integration with real-time analysis.
3:00 PM MS-WeA-4 Endpoint Prediction for Polysilicon Plasma Etching via Optical Emission Interferometry
K. Wong, D. Boning, H. Sawin (Massachusetts Institute of Technology); S. Butler (Texas Instruments, Inc.); E. Sachs (Massachusetts Institute of Technology)
In semiconductor manufacturing, the completion or "endpoint" of a plasma etch is typically controlled as a timed process, or monitored by use of optical emission spectroscopy. As etch process requirements and complexity increase (sequences of discrete recipe changes during the etch), consistency in timing changes is becoming an important issue. Newly developed full wafer interferometry sensors enable real-time monitoring of the entire wafer surface during an etch. Previously, etch rate extraction and endpoint detection using these interferometric signals have been examined. In this paper, endpoint prediction (or film thickness estimation) is considered. For the etch of a specific film structure, the interferometric signal is cyclical with a known number of cycles. Similar to the phase angle of a cosine function, data points of an interferometric curve can be associated with a linear phase function. If the phase at endpoint is known, in-situ film thickness can be obtained by determining the phase of the incoming signal in real-time. Simulations have been performed to analyze the sensitivity of the proposed film thickness estimation algorithm with respect to etch rate drift, variation in film structure, and the selection of signal wavelengths. In particular, it is shown that the estimation is more robust if short wavelength signals are used. Finally, experimental data verifies that remaining film thickness can be estimated within reasonable accuracy. These methods provide the critical information needed to make decisions (e.g. switch to a more selective chemistry) based on reaching a desired known film thickness.
3:20 PM MS-WeA-5 Real-time Detection of Inadvertant Composition Changes in SiGe Alloys using Spectroscopic Ellipsometry
D. Robbins, C. Pickering, J. Glasper, A. Kier, D. Hope (Defence Research Agency, United Kingdom); T. Walther (University of Cambridge, United Kingdom)
The surface of a SiGe alloy grown from hydrides contains substantial adsorbed hydrogen. This paper will show that real-time spectroscopic ellipsometry (SE) in a low pressure vapour phase epitaxial reactor provides a wafer-state sensor to detect changes in Ge concentration dependent on the partial pressure of molecular hydrogen, and on the time taken to reestablish dynamic equilibrium when reactive hydrides are switched. This in-situ technique has been used to measure the sharpness of Ge profiles during growth of multi-quantum well samples, and to detect unintended interfacial concentration gradients caused by temporary growth interruptions. The pseudomorphic SiGe layers (Ge<0.2) were grown on Si from silane/germane mixtures between 600-700C, with and without hydrogen buffer gas, at pressures between 2-20 Pa. In some cases the alloy growth was stopped and then restarted, to investigate the stability of the SiGe surface and its effect on bulk alloy composition. The real-time SE data were subsequently analysed using the pseudosubstrate approximation to obtain the Ge concentration gradients. Samples were analysed by SIMS, X-ray diffraction and high resolution EELS/STEM to provide independent measurements of Ge profiles. An important new observation is the role of the hydrogen buffer gas in contolling alloy composition. For growth in the absence of hydrogen, stopping and restarting growth after 10-20s produces a Ge spike at the interface corresponding to the interruption. However, a similar experiment with hydrogen buffer produces a dip in the Ge concentration at the interface if the hydrogen flow is maintained when the silane and germane are switched out of the reactor. The results show that in- situ SE provides a real- time sensor for changes in the dynamic equilibrium between the gas and surface phases, and could be an important tool for process control in growing multilayer heterostructure devices such as SiGe/Si HBTs.
3:40 PM MS-WeA-6 Wavelet Approach to Temperature Determination via Diffuse Reflectance Spectroscopy
P. Krishnaprasad, T. Kugarajah, W. Dayawansa (University of Maryland)
Progress made in optical techniques for monitoring of semiconductor film properties (thickness, temperature etc.) in MBE processing has made Diffuse Reflectance Sensors (DRS) an attractive possibility. DRS is based on the following: during measurement, white light is focused onto the semiconductor substrate, which is polished on the front surface and textured on the back. The measurement collects only light diffusely reflected from the back surface to obtain reflectance as a function of wavelength. Since at short wavelengths light does not penetrate the front surface, the onset of transparency corresponds to a sharp increase in diffuse reflectance at a knee wavelength. Though such a knee wavelength is not always clear-cut, recent research has focused on using physical principles to relate the temperature to band gap, film thickness etc., (their effect on the onset of transparency and hence the knee wavelength), and the temperature is obtained approximately as a function of knee wavelength and substrate thickness. Johnson et. al. (1993), obtain a function that is quadratic in knee wavelength and linear in substrate thickness. However the observation that the reflectance measurements are made available over an entire range of wavelengths led us to explore the possibility of finding the temperature solely from analysis of patterns contained in the data without relying on approximate physical principles. We extract such patterns, using 2-dimensional (2-D) spectra defined on an appropriate phase plane, that we construct from a wavelet analysis of the DRS spectra. Then such 2-D spectra (called DRS faces) are further analyzed (to find key features or patterns) using a technique known as Principal Component Analysis (PCA). We show, using data supplied by S.R. Johnson, that each DRS face can be represented as a linear combination of a small number of such patterns. The associated weights serve as a compact code for a DRS spectrum, which we use for temperature determination.
4:20 PM MS-WeA-8 Radio-Frequency Electrical Measurements of Ar/NF\sub 3\ Plasmas
M. Sobolewski (National Institute of Standards & Technology); J. Langan, B. Felker (Air Products & Chemicals, Inc.)
Fluorinated gas plasmas are widely used in semiconductor manufacturing. NF3 plasmas in particular can achieve fast etch rates or PECVD chamber clean times, but its high electronegativity may lead to difficulties in maintaining plasma stability and coupling power into the plasma. To investigate these issues, we have measured the electrical characteristics of Ar/NF3 plasmas at 50-2000 mTorr in a 13.56 MHz, capacitively-coupled, parallel-plate reactor. The current and voltage at the powered electrode, the current at the ground electrode, and the voltage on a probe inserted into the electrode gap were measured. As total pressure or NF3 fraction increased, the discharge underwent a transition from electropositive to electronegative behavior, consisting of an increase in the plasma impedance, a shift in its phase from capacitive to resistive, and an increase in the voltage drop across the plasma. This voltage drop was not distributed uniformly across the gap. A higher gradient of voltage was observed in a highly resistive region close to the powered electrode. The power coupling efficiency (plasma power/applied power) and the emission efficiency (Ar and F optical emission intensity/plasma power) peaked at a point in the middle of the transition from electropositive to electronegative behavior, at a plasma impedance phase of -45=B0. At this optimum, power losses in the electrical system and in the plasma are minimized, suggesting that optimal process results may be obtained at this operating point. These results illustrate that plasma power and plasma impedance phase are useful parameters for monitoring and optimizing plasma processes in highly electronegative gases.
Time Period WeA Sessions | Abstract Timeline | Topic MS Sessions | Time Periods | Topics | AVS1996 Schedule