PPT Purification elimination process variables
04/01/1998
PPT purification eliminates process variables
B. Gotlinsky, J. O`Sullivan, Pall Corporation, Port Washington, New York
F. Shadman, University of Arizona, Dept. of Chemical Engineering, Tucson, Arizona
M. Horikoshi, S. Babasaki, Nihon Pall Ltd., Tsukuba, Japan
A composite of intercalate reactive metal compounds forms an integral part of the existing fibrous filter matrix to create a POU purification surface. The reactive sites are located on the stainless steel filter medium to ensure contact and subsequent removal of homogeneous impurities.
POU purification techniques are necessary to meet the demand for the distribution and delivery of ultrahigh-purity process gases within the semiconductor industry. To satisfy increasingly stringent particulate and cleanliness requirements, typical purification equipment such as getters, resins, and molecular sieves achieve single-digit part per billion (ppb) levels in process gases. The latter techniques rely on the removal of homogeneous impurities by a purifier and the subsequent removal of heterogeneous impurities by a filter.
An alternative approach is necessary to achieve gas purity at the low single ppt level. For example, the University of Arizona (SEMATECH Center of Excellence) and Pall Corp. developed a reactive filter that is a truly integrated purifier/filter, with reactive sites located on the stainless steel filter medium (Fig. 1).
 |
Figure 1. The Gaskleen PPT purifier
In general terms, the chemistry can be described as an integral composite of intercalate reactive metal compounds (IRMC) in reduced form. These inserted compounds become an integral part of the existing fibrous filter matrix, creating a true POU purification surface. The use of the high-surface-area stainless steel filter medium as the support structure promotes the removal of homogeneous impurities by ensuring contact of the impurity with the reactive sites.
The reactive filter is ideally suited to establishing a zero (or baseline) gas for atmospheric pressure ionization mass spectrometry (APIMS) and gas chromatography-mass spectroscopy (GC-MS). The reactive filter demonstrates the ability to maintain baseline levels for moisture, O2, and CO2 of <20 ppt for APIMS. In addition to the purification of inert gases, the reactive filter allows for the purification of specialty hydride gases. The efficiency of the reactive filter in purifying silane gas is evidenced by the quantitative reduction in siloxane peaks (at mass-to-charge [m/z] ratios of 77 and 109) in the APIMS spectrum of silane. The reduction in the thickness of SiO2 films required for the successful development of 256-Mbit and 1-Gbit DRAM processes dictates the need for high-purity silane gas.
The purification of the argon gas in a PVD process reduced chamber downtime by ensuring faster pumpdown to base pressure. The reactive filter reduced the pumpdown time of a200-mm PVD chamber by 25% over that observed with conven tional purification techniques.
Zero gas for APIMS
The use of APIMS for the evaluation of ultrapure semiconductor process gases arose from the need to ensure compliance with stringent purity requirements. APIMS has greater sensitivity relative to conventional quadrupole mass spectrometers due to a secondary ionization that increases the quantity of impurity ions available for detection by the quadrupole mass analyzer. The resulting detection limit of an APIMS can be in the very low ppt range. The actual sensitivity of the instrument will be dependent on the instrument design, gas supply system, and sampling methods [1].
As instrumentation and gas sampling systems evolve, calibration becomes more important. Lower contamination levels drive the need for lower detection capabilities. The zero gas, or relative baseline, must be established at the limits of instrument detection. Conventional purification technologies have been used to generate zero gases in the low ppb to high ppt range. However, with the advent of APIMS capable of low ppt detection limits, new technology purification, capable of producing low ppt zero gas, becomes necessary.
Hitachi recently introduced the model UG-410-P APIMS, which it claimed has the lowest detection limit of 15 ppt [2]. The company produced a zero gas to establish this sensitivity using a Pall PPT purifier/filter (Gaskleen PPT Purifier). In conjunction with the use of the PPT purifier, researchers carefully baked out the sampling line and APIMS chamber at 200?C. The sampling line consisted of 0.125-in. CrP tubing.
Figure 2 shows an example of an Ar analysis, with a moisture level of 40 ppt and an oxygen level of 140 ppt. As determined from the calibration curves, the signal-to-noise ratios (S/N) for these peaks were 300 and 50, respectively, from a calculated noise level of 0.1 ppt for moisture and 3 ppt for oxygen.
A standard moisture generating system (MG -10: Hitachi Tokyo Electronics Co., Ltd.) generated a constant concentration of moisture in a ppm range to allow accurate verification by weight of the evaporation rate. The moisture generating unit consists of a Pyrex water bottle and a mass flow controller (MFC). The oxygen calibration was performed by diluting the reference gas using a MFC in a single step.
While Hitachi obtained a level of 40 ppt moisture in oxygen, additional studies in nitrogen achieved measurable 15 ppt moisture and oxygen levels in nitrogen (Fig. 3). One of the only methods to reach this zero gas level is the PPT purification technology and the appropriate gas sampling design and test methods.
The S/N for these peaks were 500 and 50, respectively, using a calculated noise level of 0.03 ppt for moisture and 0.3 ppt for oxygen in the calibration curves. A standard moisture generating system added moisture for the calibration. The oxygen calibration was performed by diluting the reference gas using a MFC, again in a single step.
The University of Arizona tested the Gaskleen PPT purifier to determine its ability to provide the required zero gas levels for APIMS analysis of semiconductor process gases. The university`s researchers used a VG/Air Liquid Trace APIMS, which was modified to provide optimum analysis. Prior to the use of the ppt purifier, researchers measured calibration moisture levels of 100 ppt. With the use of the ppt purifier in conjunction with sampling techniques, including bake out and shortened tubing, moisture levels dropped to < 20 ppt, yielding an extremely low-impurity zero gas.
Purification of silane gas
Since semiconductor fabrication processes use silane gas for thin-film formation, it is imperative to purify the gas to eliminate defect-causing contaminants. One of the most difficult contaminants to control is moisture in silane gas. The mixing of moisture in the silane can cause particle generation in tubing and chambers, and can lower the quality of thin films of ever-decreasing dimensions.
Recent development of a modified APIMS to measure impurities in silane [3, 4] confirms the purity level. Laser particle counters can examine particle levels in silane gas.
 |
Figure 2. Ar gas analysis by APIMS.
 |
Figure 3. Zero gas baseline in nitrogen.
 |
Figure 4. Wafer counts with and without purification, 0.1 to 0.2 micron.
Filtration is a standard practice to maintain particulate cleanliness of process gases. For anhydrous silane, classical methods remove particles generated upstream in the gas delivery lines. If the silane gas becomes contaminated with moisture, however, particles will form downstream of the filter. The moisture contamination forms SiO2 nuclei, which condense to become particles. Particle formation can also result from the clustering of siloxane molecules.
A study [5] that tested filters in nitrogen and silane without purification found particles downstream of filters. For example, a filter used in nitrogen that yielded less than 1 particle/100 milliliter gas larger than 0.1 ?m, yielded 22,000 particles/100 milliliter over 0.1 ?m when the gas used was switched to silane. The gas was switched upstream of the filter without disturbing any connections to the filter.
Purification of the silane gas can alleviate this contamination by removing both the siloxane and the moisture before it can react to form siloxanes. A KLA-Tencor particle counter measured 0.1- to 0.2-?m particulates on wafers before and after silane purification (Fig. 4). The wafers were from a 4-Mbit DRAM LP-CVD process at a major semiconductor manufacturer. The counts on the wafers dramatically dropped after purification, supporting the theory of particle generation through the reaction of silane and moisture.
A study on silane purification using APIMS showed that the ppt purifier could deliver silane gas without siloxane impurities [6]. The test setup (Fig. 5) consisted of a silane gas cylinder, moisture injection line, and Ar carrier gas. The APIMS was modified to analyze silane by initially ionizing Ar in the first chamber, then ionizing silane and its impurities in the second chamber. The separate ionizations eliminate interference by the Ar, oxygen, nitrogen, hydrogen, and CH4 due to their high ionization potential.
Figure 6a shows the impurities in the silane without purification and with 10-ppb water injection. Silane peaks are at m/z of 31 and 63 (SiH3+ and SiH4-SiH3+, respectively). Moisture peaks are at m/z of 49, 67, and 79 (H2O-SiH3+, 2H2O-SiH3+, and H2O-Si2H3+, respectively). Siloxane peaks are at m/z 77 and 109 (SiH3OSiH2+ and [SiH3]2O-SiH3+, respectively). Disilane peaks are at 61 and 93 (Si2H5+ and SiH4-Si2H5+, respectively).
When the silane gas with the 10-ppb water injection passed through a chemically reactive resin based purifier, the water and siloxane peaks showed reduced intensities. A comparison of Fig. 6a and 6b shows the decrease in the siloxane and water relative intensities.
When the same silane gas with 10-ppb moisture was further purified with the PPT purifier technology after the resin-based purifier, the siloxane and water peaks were essentially eliminated. Therefore, to assure that particulate contamination and film defects were reduced to the levels required for advanced devices, the additional purification of the gas by the ppt purifier is necessary. The absence of the siloxane and water relative ion intensity peaks in Fig. 6c shows the benefits of the additional purification.
Pumpdown of PVD chamber
The PPT purifier quantitatively removes gaseous contaminants (H2O, CO2, and O2) to low- and single-digit ppt levels at a design flow rate of 1 standard liter/min (slpm). While a number of POU inert gas applications require an average gas flow of 1 slpm, the duty cycle in a majority of cases will consist of a vent or purge step, which will have a gas flow beyond the 1 slpm design rate of the purifier.
 |
Figure 5. APIMS analysis of impurities in silane.
 |
Figure 6. a) APIMS of silane + 10 ppb water without purification; b) APIMS of silane + 10 ppb water after resin-based purification; and c) APIMS of silane + 10 ppb water after resin and PPT purification.
The PPT purifier/filter design was challenged with research grade nitrogen at 2? ( 2 slpm) and 4? (4 slpm) the design rated flow to obtain effluent impurity levels against a total impurity challenge of 2.2 ppm. Concentrations of moisture, O2, and CO2 in the effluent gas were monitored as the gas flow rate switched from 2 slpm (0-150 min) to 4 slpm (150-180 min) and finally back to 2 slpm. The concentration of the three measured impurities was less than 20 ppt in the purifier effluent. The results illustrate the ability of the purifier to maintain low ppt gas purity levels at high purge flow rates.
 |
Figure 7. Effect of gas purity on pumpdown.
The effect of the argon gas purity on the time required to reach a base pressure of 10-8 torr during pumpdown of a 200-mm PVD chamber was determined with a getter purifier and a Gaskleen PPT purifier. The pumpdown curves (Fig. 7) are a result of venting the chamber to 700 torr of argon. In the case of getter purifier and the 99.999% argon from a gas bottle, the pumpdown required eight hours. With the Gaskleen PPT purifier installed immediately upstream of the chamber, the pumpdown took only 5.75 hours. The decrease in the time required to pump down to base pressure with the Gaskleen PPT purifier will reduce the downtime of the chamber.
The need for a heat sink typically results in a getter purifier being relatively far from the process chamber. The purity of the gas entering the process tool may be much higher than the purity of the gas exiting the getter due to outgassing from the delivery line. The delivery line between the purifier and the tool is typically baked out extensively at an elevated temperature to minimize the effect of outgassing on the purity of gas entering the tool. The Gaskleen PPT purifier operates at ambient temperature and can be installed immediately upstream of the process chamber and thereby function as a truly "point-of-use" purifier.
Conclusion
Conventional purification techniques continue to be employed to generate low ppb to high ppt purity gases. However, the continuing demand for higher-purity process gases results in the need for purification to low ppt levels. The ability of the PPT purifier/filter to establish low-digit ppt purity gas has enhanced the ability of APIMS to detect gas purities of <20 ppt for inerts. In addition, purification to low ppt levels will ensure that film defects are reduced to the levels required for advanced devices and can significantly impact downtime of process equipment. n
Acknowledgment
Gaskleen PPT Purifier is a registered trademark of Pall Corporation.
BARRY GOTLINSKY received his PhD degree in chemistry from the City University of New York. He is VP of Microelectronics Support in Pall Corporation`s Scientific and Laboratory Services department. He has been involved in technical support of the microelectronics market for more than 13 years, publishing numerous articles on filtration and contamination control issues. Gotlinsky is a senior member of IEST, and a member of ASM and ACS. Pall Corp., 25 Harbor Park Dr., Port Washington, NY 11050; ph 516/484-3600, fax 516/484-3637.
JOSEPH O`SULLIVAN received his BS degree in chemistry and PhD degree in physical chemistry from University College Cork, Ireland. He is a senior staff scientist at the scientific and laboratory services department at Pall Corp. O`Sullivan is responsible for technical support to the semiconductor industry. He has published papers on filtration and contamination control. He is a member of the SEMI Purifier Task Force and co-chair of the Liquid Chemical Distribution Systems subcomponent task force.
FARHANG SHADMAN received his PhD degree in chemical engineering from the University of California, Berkeley in 1972. Since 1979, Shadman has been with the University of Arizona, where he is a professor of chemical engineering and the director of NSF/SRC Engineering Research Center for Environmentally Benign Semiconductor Manufacturing. Shadman has extensive industrial and consulting experience in the design and operation of chemical reactors and purification systems. He is the author of more than 80 papers and 10 patents.
MOTONOBU HORIKOSHI received his BS degree in chemical engineering from Iwate University in 1975. He is a marketing manager for gas applications of microelectronics at Nihon Pall Ltd. in Tokyo, Japan. From 1989 to 1992, Horikoshi was a visiting researcher in the Electronics Department at Tohoku University. He is a member of the SEMI Filter Task Force and Purifier Task Force, as well as the Ultra Clean Society and a senior member of the Institute of Environmental Sciences.
SHINICHI BABASAKI received his BS degree from Meiji University in 1991. He works as a staff scientist for Nihon Pall Ltd.`s Scientific and Laboratory Services in the gas application division in Tokyo, Japan. He is also a member of the SEMI Filter Task Force.
References
1. Y. Mitsui, T. Irie, K. Mizokami, "Mass Spectrometer for PPT Trace Analysis," Ultra Clean Technology, 1, 1, pp. 3-12, 1990.
2. K. Kuriyama, K. Hasumi, J. Koike, "Development of New APIMS," UG-410, Clean Technology, pp. 51, 52, Sept. 1997.
3. A. Ohki, T. Ohmi, J. Date, T. Kijima, "High Purified Silane Gas for Advanced Silicon Semiconductor Devices."
4. Y. Mitsui, et al., "Development of a New APIMS for the Detection of Trace
Impurities in Special Gases," Proceedings of 40th Annual Technical Meeting
of the IES, pp. 246-253, May 1994.
5. K. Ichijo, K. Kondo, T. Abe, K. Higashi, "Particle Measurement in Processing Gases and CVD Apparatus," Proc. 28th Ultraclean Technology Workshop,
pp. 47-59, 1997.
6. T. Hashimoto, A. Ohki, M. Horikoshi, T. Ohmi, "SiH4 Purification with In-Line High Purity Gas Purifier," private communication.