High-purity tank systems for liquid helium supply
12/01/2001
Daniel M. Buck, Joseph E. Yasi, Brian J. Barr, Andrew R. Homyak, Matthew A. Lalik, Air Products and Chemicals Inc., Allentown, Pennsylvania
 Helium tankers being filled at Air Products' Liberal, KS, facility |
overview Helium (He) is effective in numerous wafer-processing steps. It is used for backside-cooling in high-pressure chemical vapor deposition (HP-CVD) and etch processes, and as a carrier gas in CVD applications for anti-reflective, oxy-nitride and other films. In optical fiber processing, He is used to fabricate fiber preforms and to transfer heat efficiently from drawn fibers. The use of a liquid He (LHe) supply system has not been attractive for microelectronics applications because of the relatively high capital cost of vessels required to maintain cryogenic He temperatures. With increasing volumes of He being consumed at semiconductor and fiber optic facilities, however, LHe can now be an attractive supply alternative for consistent, high-purity He. Since He is liquid below -450°F, virtually all impurities are frozen as solids in the liquid phase. Theoretically, by engineering vapor-phase withdraw of gaseous He from a supply tank, a stream of UHP gas can be supplied without using downstream purification. We set out to investigate the actual purity capabilities of LHe supply systems to confirm that consistent He purity is attainable, and to understand the level of purity that can be expected. LHe supply The 3400 and 5000-gallon containers are most amenable to volumes of He now required for semiconductor industry applications. The 5000 and 11,000-gallon units are amenable to fiber optics applications. We chose a 5000-gallon tank because we wanted to test a tank most representative of the actual size that would be deployed to supply LHe to end-users. An important characteristic of the 5000-gallon tank is that it is transportable and can be replaced with a new, full container whenever LHe in the tank is low. Hauling in a full LHe container to replace a depleted tank is known as "drop-and-swap." By returning the tank to the He production facility for refilling, rather than filling the container at the user's site, He quality is enhanced in two ways. First, the heel of the tank can be completely drained to eliminate impurities that accumulate in the liquid phase. Second, the container and fill lines can be properly purged prior to filling to minimize contaminants that could be introduced during the fill process. Primary design features of our test tank included:
Tests have shown that a standard 5000-gallon liquid helium supply tank and system can deliver consistent <1 ppb impurity levels without additional purification. This is an effective means of switching from low-volume cylinder supply to larger bulk supply to support the growing application of helium in semiconductor and optical fiber processing.
We chose a standard-grade LHe tank manufactured by Gardner Cryogenics, Bethlehem, PA, as our test container. "Standard-grade" means that the tank's valves are packed and other components are non-UHP. This test container was representative of typical systems used worldwide. As a user's demands grow beyond the 160,000 scf capacity of gaseous tube trailers, liquid He supply system containers can be configured to 1800 gallons (equivalent to a 180,000 scf gaseous He capacity), 3400 gallons (343,000 scf), 5000 gallons (504,000 scf), and 11,000 gallons (1,109,000 scf).
LHe tanks are designed for minimal heat loss. A tank can be expected to build no more than 2-3 psig/day when idle. Also, these tanks can be equipped with a pressure-build coil that vaporizes liquid into the headspace of the tank to raise tank pressure. Tanks with pressure build capability allow users to continuously withdraw up to 800 standard-cubic-feet/hr (scfh) of vapor at up to 160 psig He gas delivery pressure.
 Figure 2. APIMS data showing the consistency of He supply from tank vapor. |
We designed our test apparatus (Fig. 1) to investigate purity capabilities for both liquid and vapor delivery tank installations. Our 5000-gallon tank was delivered on a 27-foot flatbed truck and placed near our analytical instruments. To permit analysis, sample ports were located at two places on the LHe tank: one for liquid phase and one for vapor phase. We analyzed each phase separately, detecting N2, O2, CO, CO2, CH4, and H2O using an atmospheric pressure ionization mass spectrometer (APIMS) with 10-pptv to ~10-ppbv detection capability. We calibrated the APIMS regularly during our 21-day test to verify the instrument's stability.
We anticipated that vapor delivery would yield higher-purity He based on the thermodynamic properties of the LHe system as discussed above (i.e., where impurities other than hydrogen freeze below -400°F and remain trapped in LHe). We reasoned that with a LHe heel in the tank, He vaporizing off the liquid would be free of impurities. We also wanted to test the purity composition of the liquid to determine the typical purity that could be achieved, assuming that withdrawal of He was exclusively from the liquid phase.
In our analysis of the liquid-phase He sample, we performed two tests: the first analyzes the liquid sample directly and the second includes an in-line cryogenic filter. We used the cryogenic filter to determine whether mechanical filtration would result in a significant reduction of impurities in our LHe.
Analytical results
The APIMS was initially operated and stabilized using nitrogen flow through a getter purifier. The analyzer's background moisture concentration was below 1 ppbv, and all other contaminants were recorded near the analyzer's limits of detection (~10 ppt).
We directed He vapor sample flow to the test apparatus, flowing through the APIMS, and analyzed vapor sample flow for three days. The gaseous He sample flow initially displayed moisture readings in the 2-3 ppbv range (see table). This moisture level decreased gradually during the test period, which is characteristic of moisture dry-down of supply lines and components. Other contaminants were consistently near the analyzer's limits of detection, similar to purity levels we observed with getter-purified-nitrogen calibration gas.
 Figure 3. Liquid helium supply tank schematic, showing fill, vent, pressure build, liquid withdrawal, and vapor-phase withdrawal circuits. |
During He vapor phase testing, we prepared for subsequent analysis of LHe by purging through a separate line to facilitate moisture dry down. When the sample flow was switched from the vapor source to the liquid source, the value of the extended system dry-down time was evidenced by initial low moisture concentration readings (i.e., ~1 ppbv). Other impurity concentrations were near the analyzer's limits of detection. When we sent the liquid through the cryogenic filter, moisture levels increased slightly, indicating slight moisture contamination in the filter system itself. The flow downstream of the test apparatus was increased to assure liquid flow through the filter. The bypass flow downstream of the cryogenic filter was increased, but this change did not affect results. Again, contaminant levels remained near the analyzer's limits of detection.
APIMS analysis results indicated that a LHe tank can consistently supply He to meet UHP specifications. Evidence of the ability of the system to provide consistent quality He is apparent by the limited variability in impurity readings over the test period (Fig. 2); once stabilized, variability of purity was 1-20 pptv. The moisture concentration was found to be higher and inconsistent compared to the other contaminants. We attributed this to non-UHP-grade tubing and components on the He container. The cryogenic filter that we tested did not make a noticeable improvement in purity.
During our tests, we found some leaks in valve packings, which we corrected by tightening fittings. Despite the fact that tank valving and components were non-UHP grade (Fig. 3), impurity declined to low levels over time. A He tank that is designed using UHP components to prevent impurities from entering the system would probably result in improved gas purity. In addition, the clean-up time of the system would probably decrease.
Conclusion
Using a standard LHe supply system, we have demonstrated the ability to deliver consistently low <1 ppb impurity levels without additional purification. Direct replacement of non-UHP components with UHP components should improve such a system's consistency to provide high-purity He and reduce moisture. All large UHP bulk He users should evaluate a LHe supply system as a means to enhance the economy and consistency of UHP He supply.
Acknowledgments
We recognize Alex Varghese, Jim Bailey, Ralph Richardson, and Subhash Vaidya for their valuable assistance with this project.
Daniel M. Buck received his BS in metallurgical engineering and his MS in business administration from the University of Notre Dame. He is commercial manager of bulk gas systems for the electronics division at Air Products and Chemicals Inc., 7201 Hamilton Blvd., Allentown, PA 18195; ph 610/481-5687, fax 610/481-8647, e-mail [email protected].
Joseph E. Yasi is a principal product development engineer at Air Products.
Brian J. Barr is a design engineer at Air Products.
Andrew R. Homyak is supervisor of the Electronic Division's analytical technical services group at Air Products.
Matthew A. Lalik is a research technician at Air Products.