ULTRAPURE MATERIALS—GASES

Examining the corrosion resistance of chromium-passivated stainless-steel tubes

Mohamed Saleem and Sowmya Krishnan, Ultra Clean Technology; and Dafna Beery and Aleks Kabansky, GaSonics International

Semiconductor manufacturing involves the use of many corrosive and reactive gases, including hydrogen chloride, hydrogen bromide, hydrogen fluoride, and tungsten hexafluoride. The transport of specialty gases in semiconductor processing poses several challenges that are unlike those encountered with bulk gas delivery. Electropolished stainless steel—the standard material of choice for semiconductor process equipment as well as gas delivery systems and components—undergoes severe corrosion and degradation caused by halogen gases at specific sites, such as weld bead and heat-affected zones (HAZs).¹ This results in metallic particle generation at susceptible points. Thus, corrosion-induced particles generated at the wetted surfaces of the electropolished tubes affect halogen's point-of-use purity. Recently, chromium-rich oxide passivation (CrP) was developed on stainless-steel tubes as a way to improve corrosion resistance.² Because the surface of the tubing has an enhanced chromium oxide film that inherently resists corrosive attack by halogen gases, CrP tubing can be employed to transport gases that normally attack gas distribution systems and cause microcontamination. The CrP tubing surface also exhibits excellent moisture drydown characteristics compared to electropolished tubing. In addition, the tubing can be used to transport silane, diborane, and other reactive gases, which undergo decomposition at low temperatures caused by reaction with adsorbed moisture on the wetted surfaces.³

CrP layers can be grown on both austenitic and ferritic stainless-steel surfaces. The formation of the CrP layer on a ferritic material is relatively easier than on an austenitic material because the former application requires fewer processing steps. Austenitic tubing requires the formation of a fine grain boundary structure on the surface in order to promote diffusion of chromium to the surface. Such a surface can be developed by electrochemical buffing. Ferritic stainless steel, on the other hand, requires no such surface preparation. The diffusion rates of Cr differ in ferritic and austenitic materials. The diffusion coefficient of Cr in ferritic stainless steel is 10³—10⁵ times greater than in austenitic tubes at 400°—700°C. In addition, the Cr concentration in bulk ferritic stainless steel is 25—29 wt%, compared to 12—17 wt% in austenitic stainless steel. Thus, chromium diffusion occurs more readily in ferritic than in austenitic material.

Welding leads to a loss of the passivation layer in the weld bead and HAZ. For ferritic stainless steel, techniques have been developed to repassivate the weld bead in situ during welding. Previous studies have characterized the CrP layer on stainless steel and reported on the corrosion resistance of austenitic CrP tubing delivering HCl gas to an oxidation furnace.¹ The study reported here compares the corrosion resistance of ferritic CrP tubing with electropolished tubing during the delivery of chlorine gas used for selective etching in a downstream plasma etcher (GaSonics International, San Jose).

Experimental Procedure

Ferritic CrP tubing and electropolished tubing weldments, each 20-in. long, were chosen for the study. Nine weldments were made on each section of tubing. Welding was carried out in a Class 1 cleanroom using ultra-high-purity gas mixtures. Techniques developed at Ultra Clean Technology (Menlo Park, CA) were used to weld the CrP tubing and to repassivate the weld beads in situ during welding. For the electropolished tubing, conventional welding techniques applicable for double-melt stainless-steel tubes were employed. The welds were spaced ~2 in. apart on each tube, except for the first two welds (on the inlet side of the chlorine), which were placed ~0.25 in. apart. This was done to produce a broad, common new HAZ for the first two weldments on each tube. Figure 1 shows a simplified schematic of the test setup.

Figure 1: Schematic of test setup used for corrosion testing of electropolished and CrP tubing.

The weldments were exposed to VLSI-grade Cl^₂ gas flowing to a plasma etcher. Prior to the experiment, valves V1 and V2 were closed, and the etch chamber was isolated from the rest of the test setup. Valve V1 was then opened and the CrP and electropolished weldments were filled with Cl^₂ for a predetermined time. After the Cl^₂ exposure step, V1 was closed, V2 was opened, and Cl^₂ was pumped out of the system. The etch chamber was subsequently vented to atmosphere. This caused back-diffusion of the chamber ambient to the test weldments. After a predetermined time, V2 was closed, the chamber was pumped down, V1 was opened, and the Cl^₂ exposure cycle was repeated. The weldments were exposed to ambient air for a total of 6 hours in 2-hour intervals in the 190-hour chlorine treatment cycle. (This would amount to moisture contamination in the >=300-ppm range.)

After the exposure treatment was completed, the tubes were cross-sectioned, and each weld site and HAZ was examined by using scanning electron microscopy (SEM), energy dispersive spectrometry (EDS), and x-ray photoelectron spectroscopy (XPS). XPS was performed with depth profiling in order to obtain chemical composition data about the surface and at various depths within the passivated film. The samples were stored in high-purity (100-ppt) nitrogen between measurements. This step ensured minimal reaction between the wetted surfaces and ambient air prior to surface analysis. For the purpose of comparison, unexposed weld beads and HAZs for electropolished and CrP tubing were also examined by SEM as control samples. The depth-profile data on electropolished and CrP weld beads and HAZ were collected at a sputtering rate of 55 Å/min. (The spot size of the x-ray beam used was ~1000 µm.)

The following labeling scheme was used to designate the weld sites: the two CrP tubes were labeled A and B, with welds designated A1, A2 . . . A9 and B1, B2 . . . B9. The electropolished tube was labeled E, with welds designated E1, E2 . . . E9. The welds A9, B9, and E9 were at the inlet side of the Cl^₂ gas. The control electropolished and CrP weld beads were designated as E0 and A0, respectively.

Results and Discussion

Figure 2: SEM micrograph of electropolished weld bead (E0) before chlorine exposure.

Figure 3: SEM micrograph of HAZ of electropolished weld bead (E0) before chlorine exposure.

Figure 4: SEM micrograph of electropolished weld bead (E9) after chlorine exposure.

Figure 5: SEM micrograph of HAZ of electropolished weld bead (E9) after chlorine exposure.

SEM/EDS Analysis. Figures 2 and 3 are SEM micrographs of the control sample of electropolished stainless-steel weld bead (E0) and HAZ before Cl^₂ exposure. The surface of the weld bead showed the chevron pattern typical of double-melt stainless-steel welds. The weld bead and HAZ were also free of particles and debris. The SEM micrographs in Figures 4 and 5 depict the electropolished weld bead (E9) and HAZ after exposure to Cl^₂. It can be seen that the weld bead and HAZ suffered severe corrosion after exposure, and both areas contained numerous particles generated by corrosion. An EDS spectra (Figure 6) of the particles found in the weld bead showed the presence of Cl^₂, indicating that these particles are corrosion induced. (The particles are broadly identified as chlorides of iron in this study.) These results mesh with those from prior studies, which described the effect of materials and welding conditions on the corrosion resistance of the HAZ in electropolished stainless-steel tubing.⁴ The EDS spectra also reveal elements from the regular stainless-steel matrix, although part of this signal can be attributed to the EDS technique's large interaction volume. Electropolished tubing is susceptible to corrosion in the weld and HAZ when exposed to VLSI-grade halogen gas. Such corrosion is caused by the redeposition of manganese in the HAZ and the loss of the passivation layer surface during welding.⁵ The manganese is not clearly visible in the EDS spectra since stainless steel has a background surface that obscures the manganese peak.

Figure 6: EDS spectra of electropolished weld bead after chlorine exposure.

Figure 7: SEM micrograph of CrP weld bead (A0) before chlorine exposure.

Figure 8: SEM micrograph of HAZ of CrP weld bead (A0) before chlorine exposure.

The SEM micrographs in Figures 7 and 8 illustrate the control CrP weld bead (A0) and HAZ before Cl^₂ exposure. The weld bead, which was found to be free of particle contamination, had a columnar grain morphology typical of ferritic stainless-steel welds. Figures 9 and 10 are SEM micrographs of a CrP weld bead (A9) and HAZ after Cl^₂ exposure. The weld bead was as clean as the control weld bead, and the HAZ contained very few corrosion-generated particles. An EDS spectra on the CrP weld bead (Figure 11) reveals no Cl^₂, unlike the electropolished weld bead spectra shown in Figure 6, which indicates that no Cl^₂ penetration had occurred in the weld. In addition, the spectra show elements from the ferritic stainless-steel matrix. Also of note is the strong Cr signal at ~5.4 keV, compared to the electropolished weld. This is because ferritic stainless steel contains 25—29 wt% Cr, while standard EP SUS316L stainless steel contains only 12—17 wt%.

Figure 9: SEM micrograph of CrP weld bead (A9) after chlorine exposure.

Figure 10: SEM micrograph of HAZ of CrP weld bead (A9) after chlorine exposure.

XPS Analysis. The XPS surface analysis on an electropolished weld bead (E8) and HAZ showed that after Cl^₂ exposure, the weld and HAZ contained ~20 and ~24 at% Cl^₂, respectively. In addition to Cl^₂, the electropolished HAZ also showed manganese as well as elements from the regular stainless-steel matrix. As discussed previously, the presence of manganese stems from its evaporation and redeposition in the HAZ by a crude vapor deposition mechanism. The XPS analysis of a CrP weld bead (A8) and HAZ revealed that the amounts of Cl^₂ in the weld bead and HAZ were ~2 and ~4 at%, respectively. No manganese was detected in the heat-affected zone of the CrP.

Figure 11: EDS spectra of CrP weld bead (A9) after chlorine exposure.

Figure 12: XPS depth profile of electropolished weld bead (E8) after chlorine exposure.

Figure 13: XPS depth profile of HAZ of electropolished weld bead (E8) after chlorine exposure.

Figures 12 and 13 display depth-profile data for the electropolished weld bead (E8) and HAZ. A high concentration of Cl^₂ can be seen, both on the surface and in the bulk of the material (~20 at%). The depth profile of the electropolished weld bead exposed to Cl^₂ shows ~18 at% Cl^₂ on the surface, which increases to 24 at% into the bulk of the material (at a 500 Å depth) and indicates Cl^₂ penetration into the bulk. The depth profile of the HAZ of CrP shows 24 at% Cl^₂ on the surface and in the bulk material. Figure 14, a depth profile for the CrP weld bead (A8), shows that the chromium oxide (Cr^₂O^₃) layer thickness is ~275 Å, and the CrP layer consists of nearly 100% Cr^₂O^₃. Only 1.8 at% Cl^₂ remains on the surface, and it disappears within 45 Å of the CrP layer. This finding is consistent with the EDS spectra found in Figure 11, which showed that Cl₂ is absent in the bulk of the material. A depth profile for the HAZ of the CrP weld after chlorine exposure is illustrated in Figure 15. Up to ~200 Å, the surface is rich in Cr, and the Fe-Cr crossover does not occur until ~300 Å. There is ~4 at% Cl^₂ on the surface, and the concentration decreases to 2% as the CrP layer deepens, finally disappearing at ~190 Å into the CrP layer. There is no Cl^₂ in the bulk of the material. This finding demonstrates how the CrP film provides excellent resistance against chlorine penetration and subsequent corrosion of the tubing.

Figure 14: XPS depth profile of CrP weld bead (A8) after chlorine exposure.

Figure 15: XPS depth profile of HAZ of CrP weld bead (A8) after chlorine exposure.

Conclusion

Sections of electropolished stainless-steel tubing and ferritic CrP stainless-steel tubing were exposed to chlorine gas under identical conditions. As the study's results show, CrP tubing exhibits superior corrosion resistance. Both the weld bead and HAZ areas of the CrP tubing remained impervious to chlorine attack. The weld site and HAZ areas of the electropolished tubing, however, suffered severe corrosion that led to the formation of metallic particles. There was also considerable chlorine penetration into the bulk of the electropolished tubing, which resulted in a continuously corroding system. These findings reveal how CrP tubing offers better corrosion resistance to chlorine than the electropolished tubing under accelerated corrosion testing, which translates to extended lifetimes for weldments, interconnections, and gas delivery system components in applications involving corrosive gas transport. The integrity of the CrP layer in a corrosive application results in improved delivery system uptime, excellent process repeatability, and processes free of metallic particle defects. These features lead to lower cost of ownership of process gas panels and higher wafer yields.

Acknowledgments

The authors wish to thank Philip Cruickshank for his assistance in welding, Elizabeth Lawrence for her help with SEM, and Mary Ann Galindo for help with this work. The authors would also like to thank Peter Sorbel and Ed Principe of Charles Evans and Associates (Redwood City, CA) for carrying out the XPS analysis on the tubing samples.

References

1. Krishnan S, Grube S, Laparra O, et al., "Investigating Corrosion Resistance of Heat-Affected Zones in CrP Tubing," MICRO, 14(5):37—41, 1996.

2. Takahashi S, Miyoshi S, Kojima T, et al., "Corrosion Resistance and Noncatalytic Properties of Cr^₂O^₃ Surface Treatment for Specialty Gases", in Proceedings of the 1993 Microcontamination Conference, Santa Monica, CA, Canon Communications, pp 596—605, 1993.

3. Ohmi T, "Corrosion-Free Cr^₂O^₃ Passivated Gas Tubing System for Specialty Gases," supplement to Solid State Technology, S18—S22, October 1995.

4. Krishnan S, Grube S, Laparra O, et al., "Site-Specific Corrosion in Gas Delivery Tubing Exposed to Semiconductor Grade HCl," supplement to Solid State Technology, S11—S15, October 1995.

5. Miyoshi S, Kojima T, Suenaga T, et al., "Metal Fume-Free Welding Technology," in Proceedings of the 1993 Microcontamination Conference, Santa Monica, CA, Canon Communications, pp 606—615, 1993.

Mohamed Saleem, PhD, is an applications engineer in the technology development group of Ultra Clean Technology (Menlo Park, CA), a manufacturer of gas delivery systems for semiconductor equipment and device manufacturers. He received an MS in chemical engineering from Tufts University (Medford, MA) and a PhD in materials science and engineering from the University of Florida (Gainesville). (Saleem can be reached at 650/323-4100.)

Sowmya Krishnan, PhD, is Ultra Clean Technology's director of technology and applications development. She is responsible for research in gas delivery technology and overseeing the development of the company's new products. An active participant in SEMI standards, she chairs the gas panel task force (contamination in gas distribution systems subcommittee) and has chaired sessions on contamination-free manufacturing and gas delivery at Semicon West, Semicon Southwest, and CleanRooms West. She has authored several technical papers and editorials in the field. Krishnan holds MS and PhD degrees in chemical engineering from Clarkson University (Potsdam, NY). (Krishnan can be reached at 650/323-4100 or krish@ix.netcom. com.)

Dafna Beery, PhD, is a process development engineer in the R&D department of GaSonics International (San Jose). She is mainly responsible for developing novel wafer cleaning technology for the removal of residues, particles, and metallic and organic contaminants. Before joining the company, she was a postdoctoral researcher in the chemical engineering department at the University of California at Berkeley, where she studied the weld-line characteristics of liquid crystalline polymers. She has a BS in chemical engineering and an MS and PhD in material engineering from Technion—Israel Institute of Technology (Haifa). Beery has published eight papers in the field of materials science, some of which have been presented at major conferences.

Aleks Kabansky, PhD, is also a process development engineer in GaSonics's R&D group. He is responsible for developing novel cleaning and etching processes and diagnostic techniques for the company's new tools. Before joining GaSonics, he was a research scientist at Mariupol Polytechnic University (Ukraine), where he specialized in plasma-surface interactions, processing, and diagnostics. Kabansky holds a PhD in physics and materials science from the University of Tomsk (Russia).

MicroHome | Search | Current Issue | MicroArchives
Buyers Guide | Media Kit