Application of FTIR for monitoring cleanroom air and process emissions
05/01/1998
Application of FTIR for monitoring cleanroom air and process emissions
Larry Zazzera, William Reagen, 3M Company, St. Paul, Minnesota
The semiconductor industry is pursuing numerousemission reduction strategies to comply with the 1990 Clean Air Act amendments and Occupational Safetyand Health Administration (OSHA) standards. Feasibility testing and development of these strategies will require test methods that provide information on the composition of emission sources. Numerous technical protocols guide the use of Fourier transform infrared (FTIR) spectroscopy to generate verifiable data on perfluorocompounds (PFCs), hazardous air pollutants (HAPs), and volatile organic compounds (VOCs).
This article summarizes the various protocols, test methods, and reference libraries for the use of FTIR spectroscopy to generate verifiable data on hydrofluorocarbons (HFCs), PFCs, HAPs, and VOCs in process tool exhaust and cleanroom air. Monitoring methods based on FTIR are a powerful tool for environmental engineers at 3M, either as a stand-alone tool or a complement to other methods like gas chromatography and mass spectrometry. Air and exhaust monitoring using FTIR methods is valuable because it is a feasible, reliable, and cost effective way to support the environmental safety and health (ES&H) programs of a diverse manufacturing company. Our experience shows that by using commercially available FTIR equipment and published guidelines, we can monitor semiconductor process tool exhaust, indoor air, and other emission sources.
The United Nations conference on global warming held in Kyoto, Japan, last December asked nations to roll-back emissions of six greenhouse gases - carbon dioxide (CO2), methane (CH4), nitrous oxide (N2O), PFCs, HFCs and SF6 - to pre-1990 levels. In the US, the two-year-old memorandum of understanding (MOU) between the US Environmental Protection Agency (EPA) and the semiconductor industry guide proactive emission reduction objectives. PFCs (CF4, C2F6, C3F8, SF6, and NF3) and the HFC (CHF3) listed in the MOU are widely used in semiconductor manufacturing. The compounds have relatively high global warming potential (GWP) values relative to CO2 because they are strong infrared radiation absorbers and have long atmospheric lifetimes. In addition, the 1990 amendments to the Clean Air Act and individual state regulatory requirements call for controlling the emissions of hundreds of HAP and VOC listed compounds.
The 1997 National Technology Roadmap for Semiconductors (NTRS) includes an ES&H Roadmap to guide future semiconductor process technology [1]. It lists reduction of PFC emissions as one of the three most imminent ES&H "difficult technical challenges" while the industry moves towards 100-nm IC feature sizes. Guidelines are provided for significant reductions of PFC, HAP, and VOC emissions for 300-mm equipment. PFC and VOC emissions from 300-mm equipment will be reduced a minimum of 50% on a per-wafer basis compared to 200-mm equipment, even though a 300-mm wafer represents a 225% increase in silicon area. The ES&H Roadmap recommends that these significant emission reductions should be followed by the eventual elimination of PFC emissions.
The industry is evaluating a broad range of technology options to achieve emission reduction objectives. Pollution prevention techniques under consideration include process optimization, alternative clean and etch chemistries, exhaust capture and recycling, and abatement of emissions by various decomposition mechanisms. Feasibility testing, development, and verification of these emission reduction programs require gas sampling and test methods which provide information on the composition of various emission sources.
In recent years, extractive FTIR spectroscopy has developed into an effective technology in field analysis [2, 3]. US EPA reports confirm the capability of extractive FTIR to generate highly accurate emissions data for a wide range of VOCs in difficult test matrices [4, 5].
Developing FTIR methods to monitor emissions
Infrared spectroscopy has been widely used for qualitative and quantitative analysis for many years, and numerous publications provide detailed discussion on the technology [6]. Infrared spectroscopy involves the interaction of radiation in the middle infrared region (2.5-50 ?m) with molecular vibrations (6.0 ? 1012 - 1.2 ? 10 14 Hz).
In order for a molecule to be infrared-active, the molecular vibrations must be accompanied by a change in the dipole moment. Consequently, homonuclear diatomic molecules, such as H2, N2, F2, and Cl2, which do not have a dipole, do not absorb in the infrared. Larger molecules or molecules containing heteroatoms may have a large number of vibrations and rotations that are infrared active. Thus, their infrared spectrum is generally complex and has a unique fingerprint. Quantitative infrared spectroscopy is based on Beer`s Law:
A = e C _
where A is infrared absorbance, e is molar absorbance, C is molar concentration, and _ is the path length of the sample cell. These features of FTIR spectroscopy allow one to use FTIR test methods that can simultaneously identify and quantify components of complex gas mixtures.
In addition to sampling and analytical procedures for extractive emissions measurements using FTIR [7], the EPA has prepared a library of quantitative IR spectra that contains approximately 109 out of the 189 HAP listed compounds [8]. The quantitative IR spectra of HAP-listed compounds has helped make possible ambient air monitoring for occupational exposure assessments [9, 10]. Development of an extractive, FTIR-based analysis of the exhaust from semiconductor plasma tools has been largely carried out by SEMATECH [11]. All six PFCs listed in the MOU, and most of the HAP listed compounds, can be analyzed using US EPA-approved FTIR methods described in National Institute of Standards and Technology (NIST) protocols, US EPA test methods, and US EPA reference libraries.
The FTIR method development can be divided into laboratory-study and field-study phases. In practice, portions of these developmental phases may occur simultaneously, and the results of some later portions of the development effort may require adjustment or repetition of some activities appearing in both phases.
Laboratory-study phase
The laboratory-study phase begins with a proposed set of spectroscopic conditions for field studies and subsequent field applications. These conditions include the minimum instrumental linewidth, spectrometer wavenumber range, sample gas temperature, sample gas pressure, absorption path length, maximum sampling system volume (including the absorption cell), minimum sample flow rate, and maximum allowable time between consecutive infrared analyses of the exhaust.
The laboratory-study phase estimates the expected concentration range, which can be based on previous emissions test results or process knowledge. This range covers values between the maximum expected concentration and the minimum concentration of interest.
Establishment of reference spectra is a key objective of the laboratory-study phase. A minimum of four reference spectra, at different concentrations (ppm-meter), must be available for each analyte. When the set of spectra is sorted according to absorbance, the absorbance levels of adjacent reference spectra should not differ by more than a factor of six. Optimally, reference spectra for each analyte should be available at absorbance levels that bracket the analyte`s expected concentration range (Fig. 1). If reference spectral libraries meeting these criteria do not exist for all analytes and interferants, or cannot be accurately generated from existing libraries, the engineer must prepare the reference spectra.
 |
Figure 1. Plot of C3F8 infrared absorbance vs. the product of concentration and path length. Reference spectra should be available at absorbance levels that bracket the analyte`s expected concentration range.
 |
Figure 2. Plots of infrared absorbance vs. wave number for 5 MOU listed compounds generated using EPA methods.
Reference spectra are generated from samples prepared from neat forms of the analyte or from high-quality gas standards commonly available from commercial sources. Either barometric or volumetric methods may be used to dilute the reference samples to the required concentrations. Periodic independent calibration of the equipment ensures accuracy. Dynamic and static reference sample preparation methods are acceptable, but dynamic preparations are more likely than static methods to give consistent results for reactive analytes. Any well-characterized absorption path length may be used to record reference spectra, but the temperature and pressure of the reference samples should match as closely as possible those of the proposed spectroscopic conditions. Figure 2 shows typical FTIR reference spectra for five semiconductor process gases listed in the MOU.
The laboratory-study phase also includes preparation of a sampling system suitable for delivering the proposed sample flow rate from the exhaust source to the infrared absorption cell. For the compounds of interest, only stainless steel and Teflon surfaces of the system can be exposed to the exhaust stream. Because of the potential for generation of inorganic acid gases, glass surfaces within the sampling system and absorption cell should be Teflon coated. It is important to demonstrate that the system delivers a volume of sample at least four times the maximum sampling system volume (anticipated in field applications) faster than the proposed minimum time between consecutive infrared analyses.
Field-study phase
The field-study phase rigorously examines the performance of the proposed spectroscopic system, sampling system, and analytical methods developed in the laboratory-study phase. The field study should examine all the sampling and analytical procedures envisioned for future field applications. It should also include additional procedures not required during routine field applications, notably thorough dynamic spiking studies of the analyte gases. The field study only needs to be performed once if the results are acceptable and if the exhaust sources in future field applications prove suitably similar to those chosen for the field study. Significant changes in the exhaust sources in future applications will require substantial changes to the analytical equipment and/or procedures as well as a separate field study for the new set of source conditions. The field-study phase includes system installation and setup, gas cell calibration, response time checks, sampling system integrity checks, and analyte spiking.
This phase begins by installing the FTIR monitoring system (Fig. 3). The FTIR spectrometer and sampling subsystem should be assembled according to manufacturers` recommendations. The length of sample lines used should exceed the maximum length envisioned for future field applications. The system should be given sufficient time to stabilize before testing begins.
Gas cell calibration is an important aspect of the field-study phase because it verifies the path length of the gas cell in the field equipment. The gas cell is calibrated at the beginning and end of the field test by introducing a calibration transfer standard (CTS) gas directly into the absorption cell at the expected sample pressure, and recording the initial field CTS absorbance spectrum. The comparison of the field CTS spectrum to the laboratory CTS spectra determines the effective absorption path length. The CTS gas standard should consist of an inert compound that exhibits a complex FTIR fingerprint.
Checking the response time of the FTIR sampling system is another important step in the field-study phase. The system check is performed by continuously recording absorbance spectra, while sampling ambient air and CTS gas introduced at the probe tip. Examine the absorbance spectra to determine whether the flow rate and sample volume allow the system to respond quickly to changes in the sampled gas. Substitute the field calibration standards for the CTS gas and repeat the process for each reactive analyte. Examine the subsequent spectra to ensure that the reactivity of the analytes with the exposed surfaces of the sampling system do not limit the time response of the analytical system.
 |
Figure 3. Schematic of a typical FTIR monitoring system, including the FTIR spectrometer and sampling subsystem.
Check the sampling system integrity while sampling ambient air and use a mass flow meter or controller to introduce a known flow rate of CTS gas into the sample stream as close as possible to the probe tip. Measure and monitor the total sample flow rate during this process. Verify from the observed CTS concentration and the two flow rates that the sampling system has no leaks and that flow rate measurements are accurate.
Analyte spiking allows identification and quantification of a given exhaust compound in the multicomponent gas mixture. The analyte spike introduces or "spikes" a known flow rate of CTS gas into the sample stream as close as possible to the probe tip. Measure and monitor the total sample flow rate during this process, and adjust the spike flow rate until it represents 10-20% of the total flow rate. After sampling at least four absorption cell volumes, record four spectra of the spiked exhaust, terminate the CTS spike flow, pause again until at least four cell volumes are sampled, and then record four unspiked spectra. Repeat this process and obtain 12 spiked and 12 unspiked spectra. During this process, monitor the absorption cell temperature and pressure; verify that the pressure remains within 2% of the pressure specified in the proposed system conditions. Calculate the expected CTS compound concentrations in the spectra and compare them to the values observed in the spectrum.
It is important to realize that amendments of analytical method may be necessary after the laboratory- and field-study phases. The presence of unanticipated compounds and/or the observation of compounds at concentrations outside their expected concentration ranges may require the repetition of portions of the method development. Such amendments are acceptable as long as they are documented. The required level of detail for the documentation allows an independent analyst to reproduce the results from the documentation and the stored interferometric data.
Applying the FTIR method in the field
When the required laboratory and field studies are complete, and if the results indicate a suitable degree of accuracy, the methods developed may be applied to practical field measurement. During field applications, we advise close adherence to field study procedures with the following exceptions:
Ideally, the sampling lines should be as short as possible and not longer than those used in the field study.
Variations from field-study conditions, equipment, and analytical techniques must be noted and included in reports of the field application results.
Analyte spiking and reactivity checks are required only immediately following the installation of, or major repair to, the sampling system.
 |
Figure 4. Process emissions concentration (ppmv) vs. time (sec) plot during a two-step PECVD chamber cleaning process using a C3F8/O2 plasma. The minimum detection limits for the plotted species ranged from 10-160 ppmv.
Recently FTIR emission-monitoring methods have been used for the analysis of selected exhaust gases during reactive ion etching [12]. Minimum detection limits, mass balance calculations, and comparison to mass spectrometry have been reported for the application of FTIR to plasma enhanced chemical vapor deposition (PECVD) chamber cleaning [13]. Figure 4 is a plot of process emissions concentrations vs. time, during a two-step PECVD chamber clean using a C3F8/O2 plasma, and following deposition of SiO2. Following EPA protocol requirements [14] and using the method-development procedures described above, 3M produced quantitative reference spectra, allowing identification and quantification of these process emissions.
The line profiles of emissions vs. time provide valuable information for process optimization efforts because the profiles indicate how process efficiency and process time change with operating conditions. The concentrations of SiF4 and CF4 exhausts change significantly during the chamber clean, and are inversely related to each other. Emission profiles (Fig. 4) give valuable information on chamber clean optimization. For example, the SiF4 profile can indicate the rate of SiO2 removal from the chamber as well as the start and finish of both the high- and low-pressure cleans.
Mass balance verification provides a check on the quantitative accuracy of emissions analysis by confirming that the amount of exhaust species coming out of the plasma reactor accounts for the amount of etch gas going into the plasma reactor. The mass balance is based on the mass of fluorine going into the reactor as etch gas (Fin) and the mass of fluorine exhausted (Fout). For example, in a C3F8/O2 plasma clean, C3F8 flows in as the etch gas, with CF4, C2F6, C3F8, SiF4, and COF2 coming out as exhaust. Fout was calculated to be 96% of Fin.
A <100% fluorine mass balance is likely because the exhaust profiles here do not account for HF and fluorine (F2) emissions. F2 is not detectable in the infrared because it is a homonuclear diatomic molecule with no dipole moment and is therefore infrared inactive. Although quantitative emission assessment of hydrofluoric acid (HF) is possible, method validation was limited to the targeted PFC species in this example. Despite the exclusion of quantitative HF and F2 data, the 96% mass balance in the example given here is typical.
FTIR spectroscopy has recently been used to assist the optimization of PECVD chamber cleaning processes with lower net greenhouse gas emissions [15]. Plotting the SiF4 emissions and the cumulative million metric tons of carbon equivalent (MMTCE) emissions vs. time provides a view of cleaning performance based on cleaning time and greenhouse gas emissions, respectively. The MMTCE expresses the carbon component of net greenhouse gas emissions in a million metric tons of carbon equivalent. Figure 5 is a plot of the SiF4 concentration and the cumulative MMTCE emissions vs. time during two different C3F8/O2 plasma cleans of silicon dioxide films of the same thickness. The results from the PECVD chamber clean study shows that a C3F8/O2 plasma clean, modified to use approximately 27% less etch gas, reduced MMTCE emissions approximately 30% compared to the C3F8/O2 plasma clean using more etch gas. The SiF4 profile shows that the cleaning time remained approximately the same despite the decreased flow rate.
The line profiles of emissions vs. time also provide valuable information to ES&H engineers because the profiles allow measurement and subsequent control of HAP emissions. Quantitative information on HAP emissions allows them to develop strategies to minimize emissions of, and exposure to, these compounds. FTIR monitoring systems based on long (100 m) path length gas cells have been used as an industrial-hygiene air monitor to screen cleanroom air during preventative maintenance of PECVD systems [16]. Changing the path length of the FTIR system changes the concentration range available for analysis, so FTIR methods can be developed to measure low (ppt or ppb) concentrations in cleanroom air or to measure higher (0.1%) concentrations in process emissions. This broad concentration range makes FTIR useful for environmental and process monitoring.
 |
Figure 5. Plot of SiF4 concentration, and the cumulative MMTCE emissions vs. time during two different C3F8/O2 plasma cleans. The cleans followed deposition of silicon dioxide films with the same thickness, and show that the clean could be modified to use less etch gas, reduce MMTCE emissions, and not significantly increase the cleaning time.
Conclusion
Using FTIR monitoring methods, HAP, VOC, PFC, and HFC emissions can be identified and quantified in process exhaust and in cleanroom air. Emissions monitoring based on FTIR is practical because the guidelines, reference spectra libraries, and equipment needed to develop and validate these test methods already exist. FTIR methods have been used by industry and government agencies to verify emissions reductions governed by the Clean Air Act, and to support the semiconductor industry`s own emissions reductions efforts. This experience demonstrates that emissions monitoring methods based on FTIR will provide valuable support to the technical activities needed to meet the improved ES&H and manufacturing standards outlined for the semiconductor industry, as published in the NTRS.
Acknowledgments
The authors thank Jess Eldridge and Lew Tousignant of the 3M Co. for assistance with FTIR data collection and processing; and Stephanie Grelle of AMD, and Rick Kachmarick and Bob Ypsilanti of Motorola for their assistance during field-study and method-application phases of this work.
References
1. The National Technology Roadmap for Semiconductors, Technology Needs, 1997 edition, published by the Semiconductor Industry Association, San Jose, CA, 95110, ph 408/436-6600.
2. P.L. Hanst, S.T. Hanst, "Gas Measurement in the Fundamental Infrared Region," Vol. 1, Infrared Analysis, Anaheim, CA.
3. A.L. Cone, S.K Farhat, L. Todd, "Development of QA/QC Performance Standards for Field Use of Open Path FTIR Spectrometers," International Symposium on Optical Sensing for Environmental and Process Monitoring, SPIE 2365:334-338, 1994.
4. G.M. Plummer, T.A. Dunder, T.J. Geyer, L.L. Kinner, "Field Applications of FTIR Spectroscopy for Detection of Hazardous Air Pollutants at Industrial Sources," Proceedings of the AWMA 87th Annual Meeting and Exhibition, 94-RP-129.05, Cincinnati, pp. 1-20, 1994.
5. Appendix A to part 63-test methods, Method 318-Extractive FTIR Method for the Measurement of Emissions from the Mineral Wool and Wool Fiber Glass Industries, USEPA EMTIC Bulletin Board, 1996.
6. A. L. Smith, "Applied Infrared Spectroscopy," Wiley, New York, 1979.
7. Draft EPA Test Method 320, "Measurement of Vapor Phase Organic and Inorganic Emissions by Extractive FTIR Spectroscopy."
8. USEPA Emission Measurement Center bulletin board at http://134.67.104.12/html/emticwww/FTIR.html.
9. P.W. Logan, W.K. Reagen, "Near Real-Time Ambient Air Monitoring in a Chemical Plant Using Extractive FTIR," American Conference of Governmental Industrial Hygienists, Applied Workshop on Occupational & Environmental Exposure Assessment, Chapel Hill, NC, Feb. 23-25, 1998.
10. A. Samanta, L.A. Todd, "Mapping Air Toxics Indoors using a Prototype Computed Tomography System," Ann. Occup. Hyg. 40 (6): 675-691, 1996.
11. M.D. Tucker, G. D. Marbury, J. S. Stanley, J. E. O`Donnel, L. M. Henning, "S68 Final Report: Template Methodology and Lessons Learned for Sampling and Analyzing Tool Effluents," Technology Transfer # 95123039A-TR SEMATECH, Dec. 31, 1995.
12. A.E. Gubner, U. Kohler, J. Molecular Structure, 348, 209, 1995.
13. L.A. Zazzera, W.J. Reagen, A. Cheng, J. Electro. Chem. Soc. Vol 144, No. 10, 3597, 1997.
14. Protocol For The Use of FTIR Spectrometry to Perform Extractive Emissions Testing at Industrial Sources, EPA Contract No. 68-D2-0165, Work Assignment 3-12, EMTIC Bulletin Board 919-541-5742, Sept., pp. 1-50, 1996.
15. Sey-Ping Sun, et al., "PFC Reduction and Process Impact Using C3F8 Clean Chemistry," SEMI Technical Program Proceedings, PFC Technical Update, July 1997.
16. Screening experiments have been conducted which use a portable FTIR with a 100-m optical path length to monitor ambient cleanroom air during the maintenance of PECVD equipment.
LARRY ZAZZERA received his PhD degree in materials chemistry from the University of Minnesota in 1994. He is a product development specialist for 3M, where he par-ticipates in the technical developments needed to attain the emission reduction objectives of the semiconductor industry. Zazzera has 13 years of experience in semiconductor process and materials R&D programs. 3M Company, 3M Center 236-2B-01, St. Paul, MN 55144; ph 612/737-5462, fax 612/733-4335, [email protected].
WILLIAM REAGEN received his PhD degree in inorganic chemistry from the University of North Dakota in 1987. He works as a specialist in the 3M Environmental Laboratory, where he has developed and applied extractive FTIR test methods to a wide range of 3M process and environmental needs. He is currently developing FTIR hardware and test methods for indoor air measurements. 3M Company, 3M Center 03-3E-09, St. Paul, MN 55144; ph 612/778-6565, [email protected].