Hazardous gases emitted by ion implanters: Characterization and abatement
10/01/1998
Hazardous gases emitted by ion implanters: Characterization and abatement
Josep Arn?, Michael Hayes, William J. Mecouch, ATMI-EcoSys, Danbury, Connecticut
Wendell Boyd, Applied Materials, Austin, Texas, Michael J. Rendon*, Motorola/SEMATECH, Austin, Texas
Terry Romig, Motorola, Austin, Texas
Ion implantation processes use a number of hazardous and reactive materials to dope target substrates. While the use of SDS gas sources brings about significant safety improvements, there is insufficient information about the nature and volume of gaseous emissions discharged during ion implantation. Such information is necessary to assess post-pump environmental and safety hazards and to design abatement equipment specific to ion implanters. Our studies presented here provide a complete characterization of the effluent streams released by the roughing pump and cryopumps of an Applied Materials xR-LEAP ion implant tool, while running standard implantation recipes using SDS AsH3, PH3, BF3, and SiF4 sources. Contrary to previous assumptions, the concentrations of some source gases, particularly corrosive gases such as BF3 and SiF4, in the effluent streams are well in excess of accepted industrial hygiene and environmental standards for these toxic gases. Implanter exhaust analyses were performed in a quantitative, real-time, continuous, on-line mode using a calibrated Fourier transform infrared spectrophotometer to develop a customized dry scrubber for integration inside ion implanters.
The hazardous nature of dopants used in ion implant tools generates justified concerns to users. The relatively low flow rates of these source gases are outweighed by their high toxicity (especially arsine) and corrosiveness (BF3 and SiF4). To ensure safe operations, many IC manufacturers require special measures to ensure that the exposure to any of these materials is below threshold-limit values (TLV). Table 1 includes the TLV and immediate danger to life and health (IDLH) levels of representative gas phase sources used in ion implantation.
 |
The path to safe implantation depends on a thorough understanding of the issues involved and combined efforts in areas such as gas delivery systems, advanced scrubbing technologies, and dependable toxic monitoring systems. The use of safe delivery source (SDS) dopant gases brings about significant safety improvements in the handling and delivery of hazardous materials [1]. Optimum scrubber design requires comprehensive knowledge of the nature and concentration of species emitted, overall gas flow rates, and discharge frequencies exhausted by each specific pump. We will describe the complete characterization of effluent gases released by the xR-LEAP ion implanter using SDS gas sources, as well as examine industry trends in the safe abatement of ion implant tool emissions.
Experimental setup
We performed the work in the applications laboratory of Applied Materials located at SEMATECH (Austin, TX). Figure 1 depicts the setup used to measure the effluent streams discharged by the different pumps of an AMAT xR-LEAP ion implanter. The xR-LEAP tool includes a single roughing pump to back the turbo pumps evacuating the source and plenum chambers, and two separate cryopumps located at the rear and side of the wheel chamber. Sampling ports installed at the exhaust of the roughing and cryogenic pumps collected exhaust gas streams into a Fourier transform infrared (FTIR) spectrophotometer placed adjacent to the implanter. Pneumatically actuated 3-way valves at each sampling port collected either purging nitrogen or exhaust gas flowing into the gas cell of the spectrophotometer. We used dry nitrogen to generate background spectra prior to analyzing the exhaust streams and to purge the sampling lines between tests. A sampling pump drew, in a continuous mode, approximately 5 standard liters/min (slpm) of sample gas into the gas cell of the spectrophotometer. We also kept the temperature and pressure of the gas cells constant throughout calibration and testing procedures to provide accurate quantitative results.
 |
Figure 1. Schematic of analytical equipment and setup.
Analytical methodology
We used a MIDAC I-2000 FTIR spectrophotometer (MIDAC Corp.) for the measurements presented here. The instrument was equipped with pressure- and temperature-controlled high-concentration (10-cm pathlength) and low-concentration (10-m pathlength) nickel-coated gas cells with zinc selenide windows. We collected spectra averaging 16 scans at 30-sec intervals to provide continuous analysis. Each spectrum covered the spectral region between 600 and 4200 cm-1, taken at a 0.5 cm-1 resolution. The spectral regions were carefully selected to minimize interferences and maximize the detection limits to quantify the species of interest. The spectrophotometer was calibrated twice against changing concentrations of acid and hydride gases: first, in the EcoSys applications laboratory (Danbury, CT) using 100-ppm AsH3, PH3, and BF3 certified standards; and later, at the testing site by measuring the implanter effluent gas streams while delivering different flow rates of AsH3, PH3, SiF4, and BF3 into the implanter with the beam off. We also quantified additional gases using a calibrated spectral library compiled by Infrared Analysis Inc.
Roughing pump emissions
The concentrations of BF3, SiF4, AsH3, and PH3 released by the roughing pump during standard SDS implantation processes were measured. Ballast nitrogen provided by the pump considerably diluted the gases removed from the tool. The gas flow was determined indirectly to be 51.1 slpm by comparing outlet concentration to input gas flow without the ion beam turned on. The study included continuous, on-line characterization during the initial tuning stages, standard recipes with the beam on, and with the beam off using the full range of the mass flow controller. Figure 2 represents the complete analyses depicted as gas discharge from the pump as a function of time. Labels of experimental conditions provided an indication of the emission levels as a function of tool operation. Comparing the emissions with and without applying energy to the beam provided a measure of gas utilization inside of the tool (Table 2). In all cases investigated, tuning the beam resulted in discrete emissions of significant volumes of source gases.
 |
Figure 2. Roughing pump emissions of selected dopant gases.
 |
During standard implanter operation, roughing pump emissions varied significantly, depending on the nature of dopant gas. For instance, hydride releases were below 300 ppb and 2.3 ppm for PH3 and AsH3, respectively. In the case of arsine, 2.3 ppm is equivalent to only 6% of the initial source gas introduced. However, from a safety point of view, this concentration is 50 times higher than arsine`s TLV level. Conversely, measurements at the exhaust of the roughing pump showed 70-83% of the acid gases delivered to the implanter. BF3 and SiF4 emissions ranged between 15 and 37 ppm, well over their respective IDLH levels. Based on the limited range of experimental parameters investigated using a given source, roughing pump emissions varied with gas delivery rates and were not significantly affected by the current or energy beam conditions.
Cryogenic pump emissions
Regeneration of a cryogenic pump is the process of removing materials accumulated on the inner surfaces of the pump after extended operation. Thus, the nature and concentration of gases evolved are a function of the volume and nature of dopants used and implanter activity since the last regeneration. The emissions reported here included the continuous measurement of multiple-component releases during the regeneration of the pump as a function of time. We analyzed the rear cryopump effluent gases twice during two separate events and characterized the side cryopump once. Due to the high concentration of the species expected, the analyses used the short (10-cm) pathlength gas cell. Compared to the 10-m cell, the main disadvantage of using reduced pathlengths is an increase in minimum detection limits.
Both rear cryopump regeneration analyses detected the same gas phase species including CO, CO2, AsH3, CF4, and water. In addition, BF3 and SiF4 were detected during the second (2-week- load) study.
Figures 3 and 4 depict the concentrations of measured species with respect to time released during regeneration of the 2-week-load rear and side cryopumps respectively. As the temperature within the pump increased, each individual gas vented after reaching its specific boiling or melting point. Side and rear cryopumps released similar gas species although the concentrations varied significantly. The side cryopump discharged larger volumes of BF3 and AsH3 and considerably less water compared to the rear cryopump. The overall volume of gases (in cm3) and mass (in mg) released during the separate regenerations were determined using the measured N2 dilution factor and integrating the areas under each gas curve (Table 3).
 |
Figure 3. a) Low and b) high concentration of species released during regeneration of the rear cryopump.
 |
Figure 4. a) Low and b) high concentration of species released during regeneration of the side cryopump.
 |
Abatement considerations
Our analysis shows that ion implant tools present a very specific abatement challenge. The low toxic concentrations, combined with the high flow rates from standard dry roughing pumps, necessitate the need for high-efficiency scrubbers. Of the three traditional abatement methods (wet, dry chemical, and thermal), dry chemical scrubbing offers the best advantages for ion implant scrubbing.
One reason dry chemical scrubbing is the best solution is due to the low volumes of toxic gases that render other abatement methods restrictive from a cost-of-ownership point of view. Additional issues, such as the need for corrosive additives in a water scrubber, toxic substances in the wastewater stream, and toxic by-products from a thermal scrubber, further reduce the attractiveness of these other abatement technologies. A dry chemical scrubber overcomes these restrictions. The scrubbing media is contained within a canister, efficiently trapping all incoming toxic gases. Using an integrated toxic gas monitor, the system warns when the scrubber canister is being depleted and should be replaced. The low volume of toxic gases, combined with a large scrubber canister, means that a dry scrubber can last years in a typical ion implant application, making cost of ownership very attractive. A dry scrubber`s low facilities requirements further reduce ownership costs.
As part of SEMATECH`s zero impact process (ZIP) initiative, EcoSys is collaborating with tool manufacturers to integrate dry scrubber canisters immediately after the pumps inside the ion implanter. These integrated scrubber canisters will be interlocked to the tool, ensuring that ion implantion becomes a ZIP. New scrubbing media, tested under conditions based on the preceding analytical work, have shown high capacity and efficiency, even at higher flow rates. By using a more efficient resin, a smaller canister can achieve the same level of scrubbing efficiency. These new resins have permitted a smaller canister, which maintains an acceptably long lifetime under normal implanter operation.
An ion implant tool fitted with integrated scrubbers will have many advantages. First, safety and maintenance are improved by confining the hazardous gas streams inside the implanter and by minimizing the exposure of exhaust lines to corrosive gases. Another advantage of this approach is that integrated scrubbers do not require additional footprint in the fabs, where floor space is at a premium. Customizing the scrubber design and scrubbing media to the discharge characteristics of each pump can further enhance the system. For example, if an implanter is going to be dedicated to BF3 processing, the scrubber media can be designed for maximum capacity for the gas.
Conclusion
The concentration and nature of gas phase species discharged by the AMAT xR-LEAP ion implanter varied depending on the pump and experimental conditions investigated. The roughing pump released irregular yet significant volumes of source materials during the unstable conditions of the beam-tuning stages. During standard implant processes, the same pump discharged steady volumes of source gases at concentrations varying widely, depending on the nature of the gas. Acid gas source discharges were >70%, exceeding accepted IDLH values by 80-125%. Arsine source concentrations accounted for <6% of the volume introduced into the process tool, yet this represents more than 20 times the accepted TLV standards for arsine discharges. In addition, dopant gases and other carbon-containing species also evolved from the cryopump during the regeneration procedure.
This characterization study provides substantial safety and environmental information specific to ion implantation. In addition, the study defines the critical parameters essential for the development of advanced abatement equipment to be integrated inside ion implanters.
Acknowledgment
SDS is a registered trademark of ATMI and Matheson Gas Products.
Reference
T. Roming, J. McManus, K. Olander, R. Kirk, "Advances in Ion Implanter Productivity and Safety," Solid State Technology, Vol. 39, pp. 69-74, Dec. 1996
For more information, contact Josep Arn? at ATMI-EcoSys, 7 Commerce Drive, Danbury, CT; ph 408/526-9400; Wendell Boyd at Applied Materials, 2706 Montopolis Drive, Austin, TX 78741; ph 512/ 356-7818, e-mail [email protected]; Michael J. Rendon at Motorola/SEMATECH, 2706 Montopolis Drive, Austin, TX 78741; ph 512/356-3430, e-mail [email protected].; or Terry Romig at Motorola, 3501 Ed Bluestein Blvd., Austin, TX 78721; ph 512/933-3010, e-mail [email protected].