An economical solution for BEOL post-ash residue removal
06/01/2001
COVER FEATURE
Steve Loper, Weiping C. Ma, Luke Chang, FSI International, Chaska, Minnesota
KangHeon Lee, Dyana Peter-Kini, 1st Silicon, Malaysia
overview
Current backend-of-the-line processes require significant volumes of semiaqueous solvents for removing post-ash residues. A joint study on reducing chemical consumption while maintaining process capability by FSI International and 1st Silicon found that batch spray processing with built-in recirculation could reduce chemical costs by more than 80%. Taking advantage of a combination of centrifugal spray processing and recirculation features in a batch surface conditioning tool, comparisons were made to other residue removal methods in terms of particles, metal lines and vias, and electrical performance.
 |
Removing post-ash residues
Wet cleaning is widely used in backend-of-the-line (BEOL) processes to remove post-ash residues. Although organic solvents have been commonly used, semiaqueous solvents (SAS), a relatively new class of etch-residue-removing solutions, has been gaining momentum in recent years [1]. The drivers for the adoption of semiaqueous solvents include concerns over fab and personnel safety risks, as well as the high environmental impact and disposal costs associated with organic solvents. Semiaqueous solvents have lower flammability points, require less rinse water, and have no VOC emissions.
While there is a flurry of activity to develop dry cleaning methods to radically reduce the safety risks and the environmental impact associated with BEOL residue removal application, wet cleaning will likely remain the mainstay method due to its proven effectiveness [2]. One way to minimize the potential negative impact of wet cleaning is through reduced chemical consumption.
Weighing the alternatives
The solution to reducing chemical consumption centers around the chemical delivery system, whether it is immersion (bench) or spray processing, single wafer, or batch. While immersion systems dominate the wet cleaning market, they face particular hurdles in BEOL applications, notably in process control. Instead of the one-chamber approach in spray-processing systems where chemical dispensing, rinsing, and drying all occur in the same sealed, nitrogen-purged environment, immersion processing requires the transfer of wafers from one open tank to another. Variations in chemical concentration may occur as solutions evaporate from the open tanks. The wafers may also undergo an inordinate amount of agitation and extended wafer transfer time from tank to tank, potentially leading to corrosion [3].
 Figure 1. The schematic diagram illustrates the recirculation component in the ZETA 200BE surface conditioning system. |
In IC manufacturing, a wide variety of metals may be used on a single device, requiring different recipes and/or chemicals. If immersion processes are used, the addition of a new chemical can mean major hardware modifications or additional tools. In contrast, spray-processing systems can dispense several chemical sequences onto the wafer with no need for hardware modifications or footprint increases as would be required by immersion processing to achieve the same degree of flexibility. Fresh chemical dispensing and highly effective drainage methods on spray-processing systems can also dramatically reduce the dangers of cross-contamination.
Single wafer process tools in which each wafer goes through a separate chemical dispense, rinse, and dry step are similar to batch spray processing. The chemical is dispensed onto a spinning wafer, then rinsed.
 Figure 2. Particle performance test on a 1 kÅ thermal oxide wafer with 3mm edge exclusion using ELM-C30 chemical indicates DI water can effectively rinse ELM-C30. |
For applications that emphasize small batch sizes and high throughput, single wafer processing provides unique value. Chemical consumption, however, tends to be higher than with batch systems. For SAS chemicals, such as Mitsubishi's cleaner ELM-C30, this approach is quite costly.
 |
One promising strategy to reduce chemical consumption relies on the integration of a high-performance recirculation system.
Reduced chemical consumption
An internal study indicated that batch spray processing with built-in recirculation could reduce chemical costs by more than 80% relative to single wafer wet processing.
Centrifugal spray-processing systems have historically enabled low chemical and water consumption [4]. The latest surface-conditioning systems such as the ZETA 200BE surface conditioning system, extend this capability to BEOL processing by combining spray processing and recirculation.
 Figure 3. SEMs show metal lines a) before cleaning and at b) 1-min, c) 2-min, and d) 3-min dispense times. |
The system is a recirculated, centrifugal spray system for BEOL cleans providing dry-in, dry-out processing with dispense, rinse, and dry cycles contained in a single chamber. The dispense of chemicals, water, and nitrogen occurs through two spray posts located in the center of the chamber and on the side of the bowl. Processing cassettes are secured on a turntable, which is programmed to spin at controlled speeds. Spray-processing efficiency is enhanced with the presence of centrifugal force, leading to high mass transfer of chemicals to the surface and of contaminants away from the surface. The result is effective cleaning in a short time, with low chemical consumption.
The system also contains additional nozzles at each spray post to accommodate higher flow (Fig. 1). The SAS chemistry is fed directly to the recirculation tank, which holds roughly 20 liters. The chemical is then pumped through a filter and to the spray posts, delivering a total flow up to 12 liters/min. After the chemical is dispensed, it flows back to the recirculation tank to be recycled.
After the chemical dispense step, there is approximately 0.5-1 liter of chemical remaining in the chemical line, on the chamber surfaces, and in the drain block. At the start of the rinse cycle, this amount is rinsed down a separate solvent drain and does not reach the recirculation tank. Thus, approximately 0.5-1 liter of chemical is "consumed" per run while the rest is recirculated. As the process run continues, the tank automatically refills to replace the unrecovered chemical so it is ready for the next product run.
Experimental setup and results
Comparisons of batch cleans using the ZETA 200BE system to single wafer wet processing were performed at 1st Silicon. ELM-C30, a fluorine-based SAS chemical was used to remove post-ash residues.
 Figure 4. SEMs show vias a) before cleaning and at b) 2-min, c) 3-min, and d)5-min dispense times. |
Particle performance (Fig. 2) was determined by measuring the particles present on thermal oxide wafers treated with ELM-C30, using a recipe consisting of a 1-min chemical dispense and 8-min deionized (DI) water rinse. Pre-counts from initial measurements on unprocessed wafers ranged from 50-100 particles. Wafers in cassette slots 1, 13, and 25 were measured for particle diameters of 0.16µm. These results indicate that DI water can effectively rinse ELM-C30. Further, on average, <2 particles were added to the wafer following the cleaning process, a negligible contribution. Therefore, the spray processing system is considered particle neutral.
Patterned metal lines and vias were processed and scanning electron microscopy (SEM) was used to verify removal of the polymer residues. To confirm the complete removal and optimize the process time, dispense times were varied from 1-5 min. SEM images of the metal lines are shown in Fig. 3, and vias in Fig. 4. Complete polymer removal is clearly indicated after a 1-min dispense. Figure 3 also shows that there was no damage or corrosion in the metal lines for dispense times of 3 min. This is confirmed on the vias as well, where no damage or broadening of the structures was visible.
The final test compared the electrical performance of the new spray-cleaning sequence to conventional wet processes. This test provides further evidence confirming the complete removal of the polymer residual. Via resistance measurement results are shown in the table, comparing the system to a current single wafer process in a split lot experiment. For 1-3 min dispense times, the performance of the recirculated batch spray system tracks closely to that of the current process, indicating once again that the polymer was effectively removed.
Conclusion
We have shown that the recirculated batch spray system can effectively remove post-ash polymer residues. The system also can provide competitive throughput of >200 wafers/hr and uses approximately 1/6 of the chemicals used in the single wafer system (ELM-C30 consumption/wafer is 10 ml/wafer vs. 60 ml/wafer). The waste disposal costs are thus substantially lower. The system has also demonstrated low water consumption of <1.2 liters/wafer.
Continuing work at FSI International will further enhance the capabilities of this cleaning technique in BEOL processing. The latest results have shown similar performance with a wide range of SAS chemicals, including formulations from EKC Technology, J.T. Baker, Ashland-ACT, Tokyo Ohka Kogyo, and ACSI. Additional work in progress will determine the effect of bath life vs. residue removal for this application, possibly leading to further reduction in chemical consumption. FSI International is also actively collaborating with chemical suppliers to develop processes for future challenges, such as new materials and smaller IC design geometries.
Acknowledgments
The authors would like to thank Kevin Bresnehan of FSI International, and Boon Eng Lee, Steven Looi, and Chee Weng Ng of Metron Technology for their help and ongoing support. A very special thank you goes to V. Ramakrishnan of 1st Silicon for his contributions toward making this paper possible. ZETA is a registered trademark of FSI International.
References
- J. Zahka et al., "Optimizing Filtration for Solvent Photoresist Removal Processes," MICRO, p. 85, March 1999.
- M. Heyns et al., "Advanced Wet and Dry Cleaning Coming Together for Next Generation," Solid State Technology, pp. 37-47, March 1999.
- J. Rosato et al., "Characterization of Metal Corrosion in BEOL Process," Future Fab International, 6th Ed., pp.167-176, 1999.
- E. Olson, "Lowering Environmental Impact of Centrifugal Spray Acid Processors," FSI International Technical Report, 1145-TRS-0899, 1999.
Steve Loper received his BS in electrical engineering from the University of Minnesota in 1989 and has more than 12 years of process development experience. He worked for three years in the Field Applications Group at FSI International, and is currently a senior process engineer in the Surface Conditioning Division. FSI International, 322 Lake Hazeltine Dr., Chaska, MN 55318-1096; ph 952/361-7314, e-mail [email protected].
Weiping C. Ma received her BS in ceramic engineering from the University of Missouri-Rolla and an MS in materials science and engineering from the University of Wisconsin-Madison. She is a senior product marketing engineering for FSI International in the Surface Conditioning Division. Prior to joining FSI, she was a process engineer at two major semiconductor manufacturers.
Luke Chang received his PhD in chemistry from Iowa State University. He joined FSI International in 1995, where he worked on vapor phase processes. Chang is the manager of the Field Applications Engineering group of FSI's Surface Conditioning Division.
KangHeon Lee received his BS degree in chemical engineering from the University of KyungHee, Seoul, Korea, in 1989, and has more than 11 years of wet cleaning development experience at a number of major Asian semiconductor manufacturers. He is a principal engineer at 1st Silicon, where he leads the development of polymer removal and salicide processes for deep submicron CMOS technologies.
Dyana Peter-Kini received her BS in chemical engineering from the University of Manchester, Institute of Science and Technology, Manchester, UK. She is a process engineer at 1st Silicon, where she has been working on wet processing for two years.