In semiconductor and optoelectronics manufacturing, many hazardous gases are used, such as pyrophoric silane (SiH4) for thin-film deposition, corrosive chlorine (Cl2) for dry-etching, and toxic arsine (AsH3) for ion implantation. Also, global-warming perfluorocompounds (PFCs, e.g., SF6, NF3) are applied to etching products or cleaning the inner surfaces of a process chamber. The used gases and their by-products (e.g., HF, SiF4) are discharged to local and central scrubbers for treatment . However, during the gas transferring process, gas leaks could occur unexpectedly because of a cracked pipe, duct clogging (by SiO2 particles), seal degradation, or a loose fitting. As a gas leaks, it mixes with the circulation air and spreads around the cleanroom to become an AMC , which will then cause process tool damage, product corrosion , wafer defects [4-9], and potential worker injury [10-11]. Thus, quickly finding and terminating the source of the leaks is critical to fab operations.
Because of the high dilution factor (~105) of cleanroom circulation air , however, the AMC may not be detected by open-area gas sensors, which usually operate and sound an alarm when the gas concentration exceeds the permissible exposure limit (PEL) or 25% of the low explosion limit (LEL) . For example, if an SiH4 exhaust flow of 1000ppm leaks from the inlet connector of a local scrubber with a dilution factor of 105, the dispersed SiH4 will be only 10ppb (parts per billion), which is much less than its PEL (5ppm) and 25% LEL (3500 ppm) .
Although some real-time AMC sensors are commercially available [14-15], almost no gas sensor can detect the SiH4 leakage at this low concentration level (1~10 ppb). To resolve the problem, a GLDS was developed and tested on-site in 8˝ and 12˝ fabs. The results demonstrated its effectiveness in locating the leaking position and reducing potential fab losses.
Materials and methods
The GLDS is composed of two open-path Fourier transform infrared (OP-FTIR) spectrometers, a gas composition database, and diagnosis software to help determine the emitting origin. One OP-FTIR was installed at the make-up air unit (MAU) to continuously monitor the incoming pollutants; the other OP-FTIR was situated at the recirculation air unit (RAU) to sense airborne contaminants inside the cleanroom. Upon detection of the contaminants at the RAU, the diagnosis software compares the gas compositions measured by two OP-FTIRs to determine if the pollutant originates inside the cleanroom. If that’s the case, the software automatically searches the database to match the most probable leaking source and then sends a warning signal to the emergency response center (ERC), or to a responsible engineer for a follow-up mitigation action.
The OP-FTIR system comprises of an infrared (IR) source, an interferometer (model Work IR, ABB Bomem, Quebec, Canada), a transmitting/receiving telescope, and a retroreflector (PLX Inc. model AR-30-20). An OP-FTIR can instantly determine the gas contaminants present in the IR traveling path by detecting the IR absorbance spectra fingerprint of each gas and using advanced computer calculation. To accurately operate the OP-FTIR and analyze the collected IR spectra, the US-EPA published guidelines and related literature were followed [16-18]. A model of the gas pipeline system contains the following information: process and exhaust gas pipelines, gas compositions and concentrations in each pipe, process tool models, local scrubber types and abatement performances. The diagnosis software for the model was developed using Microsoft Visual Basic 6.0.
Etcher gas leak. Although no odor complaint nor gas-sensor alarm was issued, the GLDS developed in this study continuously sent a gas leaking warning to the responsible engineer. As Figs. 1a and 1b show, there is a CF4-SF6 emission source inside the fab because the measured CF4 and SF6 concentrations at the RAU (inside the fab) are significantly higher than those at the MAU (outside the fab). When CF4 and SF6 concentration profiles at the RAU are plotted in Fig. 1c, it appears that these two gases are simultaneously emitted from the same source. With the assistance of the computer software and the database, the suspected leaking source was identified as the exhaust pipelines of poly-etchers. By comparing the measured on-off times of CF4-SF6 (by OP-FTIR) with the operation records of poly etchers, the leaking spot was then determined to be a cracking duct behind a poly-etcher pump.
Another example of a gas leak from the exhaust pipeline of a new poly etcher is shown in Fig. 2a. As before, by comparing the measured on-off times of CF4-SiF4 with the operation records of poly etchers, the suspected poly etcher was identified. To speed up the search for leaks, the Teflon sampling tubing of an extractive FTIR was moved along the exhaust ducts of the process tool that was intentionally discharging a PFC gas (e.g., SF6).
Eventually, a loose pipeline connection situated between a poly-etcher pump and a local scrubber was located as the source of the leak. In this example, corrosive gases (Cl2, SiF4) were used in the process recipe. During the GLDS measurement, an odor complaint was issued by on-site workers. Thus, the immediate mitigation of the leaking scenario by the GLDS helped to relieve the workers’ odor concern and also greatly reduced the potential fab losses caused by wafer corrosion and production interruption.
Figure 2. Measured concentrations using an OP-FTIR at the RAU for a) CF4 and SiF4, and b) SiH4.
Furnace SiH As Fig. 2b shows, a case of SiH4 contamination was sensed by the GLDS at the RAU. The leaking spot was then found to be a loose connector in piping at the inlet of an electrical-heating local scrubber (CDO-858). Although the SiH4 concentration at the origin of the leak was as high as 4800ppm , the measured SiH4 concentration at the RAU using the GLDS was only 10ppb because of the massive amount of air circulation in the room. At such a low concentration level, very few gas sensors can accurately detect the occurrence of SiH4 exhaust leakage inside a cleanroom.
Figure 3. Measured exhaust CVD chamber-cleaning gases (C3F8, CF4, and SiF4) by an OP-FTIR at the RAU.
CVD chamber cleaning gas leak. Using the experience gained from the previous successful leak detection studies, the GLDS pinpointed within an hour the source of a leak in a portion of a flexible bellows pipeline behind a CVD pump. As shown in Fig. 3, the leaks comprised the CVD chamber-cleaning gas (C3F8) and its by-products (CF4 and SiF4). The molecular acids’ emissions and their adverse effects on wafers and facilities were greatly reduced.
Figure 4. Measured PGMEA (propylene-glycol-monomethyl-ether-acetate) emissions by an OP-FTIR at the RAU.
Emission of photo PGMEA. Figure 4 shows that a significant amount of PGMEA (propylene-glycol-monomethyl-ether-acetate) is being emitted into the cleanroom from the process tools themselves. During the measurement, replacing an empty PGMEA container was also found to increase the measured concentration. To improve the air quality in the photo region, a well-vented housing was built for the PGMEA container and a ventilation hood was installed around the process tool bottom. Based on the proven performance evaluation techniques , both the housing and the hood were measured to have contaminant capturing efficiencies of >85%.
The gas leak detection system developed in this study quickly and accurately detects leaking gases from different processes (e.g., dry-etching, CVD, diffusion, and photo). With the assistance of the GLDS, the fab engineers could efficiently locate the leaking spot among thousands of pipelines, which helped to reduce contaminant emissions and avoid their adverse effects on wafer defects, facilities, production, and personnel.
Additionally, due to the high dilution factor (105~106) of the cleanroom circulation air, the emitting contaminants might not be detected by open-area gas sensors. With the increasing requirement for AMC control , the real-time, low-detection-limit (1~10ppb) property of the GLDS appears to be an appropriate sensing system for detecting internally generated contaminants.
The authors thank Hui-Ya Shih and Kuang-Sheng Wang for analyzing the FTIR spectra. Teflon is a registered trademark of DuPont.
- .N. Li, J.N. Hsu, H.Y. Shih, S.J. Lin, J.L. Hong, “FTIR Spectrometers Measure Scrubber Abatement Efficiencies,” Solid State Technology, pp. 157-165, July 2002.
- S.C. Chen, C.J. Tsai, S.N. Li, H.Y. Shih, “Dispersion of Gas Pollutant in a Fan-filter-unit (FFU) Cleanroom,” Building and Environment, at press 2007.
- J.K. Higley, M.A. Joffe, “Airborne Molecular Contamination: Cleanroom Control Strategies,” Solid State Technology, pp. 211-214, July 1996.
- D. Ruede, M. Ercken, T. Borgers, “The Impact of Airborne Molecular Base on DUV Photoresists,” Solid State Technology, pp. 63-70, August 2001.
- S. Barzaghi, A. Pilenga, G. Vergani, S. Guadagnuolo, S. Getters, “Purged Gas Purification for Contamination Control of DUV Stepper Lenses,” Solid State Technology, pp. 99-104, September 2001.
- F.A. Stevie, E.P. Martin, P.M. Kahora, J.T. Cargo, A.K. Nanda, A.S. Harrus, et al., “Boron Contamination of Surfaces in Silicon Microelectronics Processing: Characterization and Causes,” J. Vac. Sci. Technol., A9 (5), pp. 2813-2816, Sept./Oct. 1991.
- Y. Ishihara, D. Nakajima, T. Ohmi, “Economical Clean Dry Air System for Closed Manufacturing System,” IEEE Trans. on Semiconductor Engineering, pp.16-23, Vol. 13, No.1, February 2000.
- S.R. Kasi, M. Liehr, P.A. Thiry, H. Dallaporta, M. Offenberg, “Hydrocarbon Reaction with HF-Cleaned Si(100) and Effects on Metal-Oxide-Semiconductor Device Quality,” Appl. Phys. Lett., pp. 108-110, 59(1), July 1991.
Shou-Nan Li is a department manager at Energy and Environment Research Laboratories, Industrial Technology Research Institute (ITRI), 11F, Bldg. 51, 195-10 Sec. 4 Chung-Hsing Rd., Chutung, Hsinchu, Taiwan 310, R.O.C.; e-mail SNLi@itri.org.tw.
At Energy and Environment Research Laboratories, ITRI, Hsinchu, Taiwan, author Gen-Hou Leu is a researcher; author Shaw-Yi Yen is a researcher; author Shin-Fu Chiou is an associated researcher; and author Sheng-Jen Yu is a researcher. Author Chang-Fu Hsu is a PhD candidate at National Chiao-Tung U.