Protect DUV Processes with Real-Time Molecular Monitoring
Bobbie Demandante and Kevin Murray, Philips Semiconductors, San Antonio, Texas; Mike Alexander, Extraction Systems Inc., Franklin, Mass. -- Semiconductor International, 9/1/2000
 | At a Glance | | |
| | ||
| |||
The resist is particularly sensitive after the coated wafer is exposed in the exposure tool and transported back to the development tool, until the time it is baked. Typically the delay between last flash and post-exposure bake condition may be less than 3 minutes, depending on the tool loading; but on occasion it may be 10 minutes or longer. A 10-minute pause easily can cause a linewidth variation on a wafer that may have just finished exposure and is not yet baked. If this event happens when a wafer is in an area where the contamination levels could be as much as 5 ppb, it is possible this may cause a critical dimension (CD) shift of as much as 10-20 nm.2 If the controlled deep ultraviolet (DUV) tool environment allows the wafer to be exposed to high concentrations of airborne molecular contamination, the resulting difficulties may range from a few nanometers of variation in the device CD to catastrophic device failure due to T-topping. This issue becomes progressively more problematic as geometries and line/space dimensions decrease. The time the wafers in process (WIP) are exposed relative to the contamination concentration determines the potential risk to linewidth control.
|
|
|
|
In 1997, when the San Antonio facility was operating independently as VLSI Technologies, company officials decided to expand into DUV photolithography. The DUV exposure and development tools include integrated chemical air filters — a standard filter protection system that was at the time limited in the event of elevated levels of molecular base concentrations. To evaluate the conditions in its facility, Philips Semiconductor decided to conduct an initial survey of the facility for molecular bases that could threaten the resists.
Extractions Systems Inc. (Franklin, Mass.) was retained to conduct a survey of the total molecular base background in the facility and thereby ascertain the optimal method of ensuring airborne-contamination-free operation of the DUV equipment in the midst of active processing activities. The fab was measured using traditional impingement and carbon trap sampling methods, and analyzed by gas chromatography mass spectroscopy and ion chromatography. Samples in 10 locations showed a range of 12 to 23 ppb total molecular base concentrations. Contaminants found included ammonia and relatively lower concentrations of cyclohexylamine, diethanolamine, diethylamine, dimethylamine, monoethanolamine, monoethylamine, morphine, triethyanolamine, triethylamine and trimethylamine.
Surveying with real-time monitoring
In 1998, after the DUV tools were installed, Extraction Systems conducted a second survey to make certain stepper and track environments were free of harmful concentrations. This survey included air measurement of the room, filter condition and inside the DUV tools during normal operation. This facility consisted of four cleanroom areas arranged in separated square areas under the same roof. Real-time monitoring allowed for more economical survey of multiple locations in the fab and allowed on-the-spot adjustments to the original scope of work as measurements were taken. The real-time survey also offered an opportunity to gain experience in real-time monitoring specifically for DUV lithography applications.
The real-time study used a total molecular base real-time monitor manufactured by Extraction Systems Inc. Combining patented conditioning and conversion technology with chemiluminescent detection, this instrument detects, quantifies and adds the total molecular pollution concentration of ammonia, NMP, all other amines and all base gases that can cause DUV photoresist exposure problems. Sequentially collecting air samples at 10 locations, it achieves an ultralow detection limit of 0.6 parts per billion.
|
|
Of later significance in terms of understanding measurements and potential sources found in DUV environments, the instrument proved so sensitive it detected ammonia emissions from passing fab personnel — nominally 30 to 100 ppb, depending on the individual.4 In the event that elevated concentrations were present in the cleanroom, one approach to meeting this objective safeguards the DUV lithography tool cluster and associated environmental enclosures through a combination of real-time total molecular base monitoring and chemical air filtration.
The survey did find base concentrations that were moderately high and potentially outside the recommended room concentrations for the standard integrated chemical air filtration provided with the original equipment. This influenced Philips' decision on how to ensure clean air quality all the time and remove the possibility of premature filter breakthrough.
One option was to envelop the DUV cluster in a minienvironment with cabinet-style chemical air filtration. This would have required installing walls, limiting accessibility and workflow in the production fab. The second option was to install powerful cabinet-style chemical air filters ducted directly to the intake of the installed integrated filters on the exposure and development tool. This left the cluster open for uninterrupted accessibility and workflow.
It was decided to add to the existing integrated chemical air filter system with exterior cabinet-style filters (Fig. 1). The cleanroom air is filtered by the exterior filters and ducted to the integrated filters.
Besides installing the chemical air filtration for the DUV equipment DNS Dspin 200 and Canon ES3, Philips also invested in a total molecular base real-time monitor to continuously monitor the installation. Key Philips staff members developed outstanding expertise using the instrument. For example, if the monitor detected a spike in airborne molecular contamination inside the DUV tool, they could correlate the data to specific interventions, maintenance activities, open tool doors, ventilation fan shutdowns and the like.
Potentially serious contamination
In most instances, real-time monitoring provides users the reassurance and confidence that airborne molecular contamination within the DUV environment is within tolerable limits. Occasionally, however, the monitoring process discloses a potentially serious problem and serves as an indispensable troubleshooting tool.
For example, on one eventful day, staff members at the San Antonio facility noted that readings on the total molecular base real-time monitor were showing airborne contamination concentrations ranging from very low levels up to 200 ppb and more in the cleanroom where the DUV tool cluster was located. Since the DUV cleanroom was environmentally and spatially isolated from the rest of the facility, staff members were concerned that the instrument might be out of calibration and providing erroneous readings. So Extraction Systems Inc. dispatched a team of engineers who verified, first, that the tool was indeed properly calibrated and, second, that there was indeed airborne molecular contamination in the DUV cleanroom.
Further investigation disclosed that a chemical spill, involving a wetbench leak of photoresist stripper that included ethanolamine and a range of volatile bases, had occurred in an adjacent cleanroom within the San Antonio facility, remote from the DUV area. Previously it was thought a release of this type in an adjacent room would not significantly migrate to this walled section of the fab. It was revealing to learn an airborne molecular release could diffuse in the fab air and migrate to cause a 200 ppb concentration in the adjacent lithography area (Fig. 2).
|
|
Also, the high-capacity chemical air filters supplied by Extraction Systems, installed to protect the minienvironment surrounding the DUV tool cluster, had withstood the enormous airborne contamination challenge, even though the spike had lasted several hours. A comparison of the spike in the ambient cleanroom with the air quality plots of the internal exposure and development tool environments showed no increase in molecular base. This clearly demonstrates the tool filter protection system is an effective barrier to contamination events.
Several months later, there was another increase in cleanroom molecular base level when a door to the CMP cleanroom area was open for an extended time. CMP slurry commonly contains trimethylamine and was readily detected by the monitor in the cleanroom containing lithography equipment. The monitor detected an event that, without real-time monitoring, might have gone undetected; and, again, the incident demonstrated effective tool filtration at levels exceeding 100 ppb total molecular base (Figs. 3 and 4).
Excursions inside the stepper and track were in many cases correlated to events in work logs. As shown in Figure 5 some of the elevations in tool enclosures were:
1. Laser recharge.
2. Alignment lamp door open.
3. Power outage (recirculation air stopped).
4. Maintenance on stepper and laser, doors open.
5. Wafer feeder locked up, doors opened.
6. Holiday shutdown.
Contamination events in filtered minienvironments usually are not caused by filter failure but may be traced to various maintenance and operation activities. At the San Antonio fab, not all tool events were correlated with known tool access. None of the elevated levels detected inside the tools was related to filter degradation. In most cases tools were not in production, and no wafers normally would have been exposed. Knowledge of the impact of gaining access to the inside of tools allows for planning and procedure to protect WIP.
|
|
Conclusion
Historically, when a contamination problem arises, the prevalent practice has been to associate the problem with an air filtration failure and therefore to replace the filters immediately. In reality, filter failures are rare; and as the incident described in this article proved, contamination events are not necessarily filter-related. Nor is the physical isolation of one cleanroom area from another a sufficient guarantee that the airborne molecular effects of a spill will not somehow migrate to an area perceived as safe.
If the San Antonio fab had not been using a real-time monitor, the staff might not have noted the existence of an airborne molecular contamination problem until there had been a loss of critical dimension in devices that was significant enough to be attributable to AMC. By the time the device linewidths were affected, they might have shut down production and replaced the air filters — a potentially very expensive, time-consuming act that, as the facts demonstrated, was wholly unnecessary.
The logical conclusion, then, is that in production DUV lithography environments, continuous total molecular base real-time monitoring is an effective early warning system. The 1999 edition of the International Technology Roadmap for Semiconductors prescribes a CD control budget of 15 nm, post-etch.5 As decreasing CD budgets and the cost of production time lost identifying problems and solutions mounts, in situ real-time monitoring of molecular bases will remain a useful form of process protection. •
Lisa Napolitano is engineering manager of the photolithography area at Philips Semiconductors in San Antonio, Texas. She has been at Philips, formerly VLSI, for five years and was project manager for the airborne molecular contamination measurement and control of the DUV lithography cluster. She has a B.S. in microelectronics from Rochester Institute of Technology and will receive an M.S. in management of technology from the University of Texas at San Antonio in May 2001.Phone: 1-210-522-7056
Fax: 1-210-522-7100
e-mail: Lisa.Napolitano@Philips.com
Kevin Murray is senior equipment engineer at Philips Semiconductors, formerly VLSI. He has 10 years experience there in etch, photo and CMP processes, and 17 years in the semiconductor industry. He received a B.S.E.E. from Missouri Institute of Technology in 1983.
Phone: 1-210-522-7449
e-mail: kevin.murray@philips.com
Michael Alexander is regional manager of DUV measurement and control systems for Extraction Systems Inc. He has 13 years experience in the semiconductor industry and previously was national sales manager for electronics filtration products at Memtek and co-owner of AU Associates (Tempe, Ariz.), specializing in semiconductor capital equipment. He holds a B.S. from University of Central Florida.
Phone: 1-602-483-1151
Fax: 1-602-483-3864
e-mail: malexander@extractionsystemsinc.com
REFERENCES
- S.A. MacDonald et al., "Airborne Chemical Contamination of a Chemically Amplified Resist," SPIE, Advances in Resist Technology and Processing VIII, vol.1446, (1991), 2-12.
- D.Kinkead and M. Ercken, "Progress in Qualifying and Quantifying the Airborne Base Sensitivity of Modern Chemically Amplified DUV Photoresists," SPIE, Advances in Resist Technology and Processing, Paper 3999-81 (2000).
- Kishkovich and Dean, "Environmental Stability of Chemically Amplified Resists: Proposing an Industry Standard Methodology for Testing," SPIE, Advances in Resist Technology and Processing, Paper 3999-140, (2000).
- Readers interested in this phenomenon may learn more by reading the paper "Amine Control for DUV Lithography: Identifying Hidden Sources," by Carl Larson of IBM and Oleg Kishkovich, Ph.D., of Extraction Systems Inc., presented at the SPIE conference in March 2000. (See www.extractionsystemsinc.com/hidden.pdf)
- SIA, International Technology Roadmap for Semiconductors, 1999 Edition, Table 39A, Lithography Technology Requirements - Near Term, p. 147.
