Advances in ion implanter productivity and safety
12/01/1996
Advances in ion implanter productivity and safety
Terry Romig, National Semiconductor (SEMATECH), Austin, Texas
Jim McManus, Karl Olander, Advanced Technology Materials Inc., Danbury, Connecticut
Ralph Kirk, Matheson Gas Products, San Jose, California
This article examines the impact of advanced dopant delivery systems on ion implanter productivity and safety. Conventional dopant source materials are compared with a new atmospheric pressure SDS Gas source. The source reversibly adsorbs the dopant gas onto high-surface-area materials and uses pressure swing desorption to deliver the dopant to the implanter ion source.
Increasing fab productivity and overall equipment effectiveness (OEE) is a constant objective for the manufacturers of integrated circuits. Ion implanters average about 65-85% machine availability, and recent surveys have confirmed ion implantation as the constraining step in most fabs. Increasing ion source lifetime and reducing species turnaround time have a direct impact on implanter availability. Any improvement in the source area must balance the exposure risks of handling highly toxic materials, system cleanup, and source material disposal. Dopant source selection has the potential for significant productivity gains.
Ion implanters are large, complex, high-performance tools and are costly to own and operate [1]. Despite such drawbacks, ion implantation is the preferred doping process for advanced circuits. Typically a variety of ions including As+, P+, B11F2+, and higher mass-to-charge species are implanted. Individual wafers may go through ion implantation 13-15 times. Running multiple ion species on the same implanter adds manufacturing flexibility and minimizes capital expenditures, so dopant species changes typically occur many times a day.
Dopant source materials
Dilute high-pressure gases and solid materials are used to supply dopants in ion implantation. Gases offer ease of use and higher control, but safety concerns have limited how they are employed. The toxicity of the source materials arsine,
phosphine, and boron trifluoride, and the associated clean-up and maintenance of process residues are a continuing safety concern [2]. Exposure limits for the common implant dopant gases set by the National Institute of Safety and Health (NIOSH) are shown in Table 1 [3].
 |
High-pressure gas sources. The dopant source materials offering the highest productivity are undiluted high-pressure gases. With gases, species switching times are short and limited to the time required to purge the manifolds and re-establish stable beam currents.
As a safeguard, high-pressure hydride source gases are diluted with hydrogen and packaged in relatively small cylinders (0.4 l-2.2 l). The dopant gas box, located within the implanter, is exhausted at 500-600 ft3/min (cfm). Diluting with hydrogen reduces the risk associated with inadvertent gas releases. These safety measures, while reducing the risk of injury, require more frequent cylinder replacements. Replacing dopant cylinders takes 1-2 hr. The dilute hydrogen also causes implanters to run at higher system pressures which, in turn, can cause high-voltage arcing for high-energy beams, such as n-well and cap implants, and strains the vacuum pumps.
Solid sources. Solid sources were developed in the early 1980s to provide higher beam currents as a replacement option for gases. As the name implies, solid arsenic and phosphorus are vaporized at about 900?C and the volatile dopant atoms are conveyed into the ion source chamber. Vaporizers have no potential for catastrophic release. No dilute hydrogen is added to the process and higher maximum beam currents are obtained.
Two disadvantages of solid sources are the setup time and maintenance. The setup time, or tune time, can be 40-180 min, most of which is required to heat and stabilize the vaporizer. Also, maintenance is more frequent because of the build-up of sputtered materials on the beam-line components and the need to recharge the vaporizer at each source change. The longer gas-switching periods and higher maintenance needs reduce machine availability and raise cost of ownership.
SDS atmospheric pressure Gas Source. A new and promising dopant source, the SDS [4-6] combines the safety and implanter operating performance of solid sources with the fast species changeover and tune times characteristic of high-pressure gases [7] (Table 2). A simple technique for storing and delivering gaseous arsine and phosphine at < 0 psig, based on physical adsorption of the gases onto a microporous adsorbent, greatly reduces the hazards associated with their use [8, 9]. No diluent, such as hydrogen, is present and problems with arcing and vacuum pump operation are avoided [10]. Maintenance is also reduced.
 |
Even though the SDS Gas Source is at 0 psig, it still delivers significant amounts of dopant. For example, a high-pressure ion implant cylinder (400 psig, 15% arsine) delivers about 30 gm of arsine, whereas the same size SDS cylinder delivers 200 gm of arsine. The greater delivery capacity reduces the number of cylinder changes and improves both the productivity and safety of the ion implant process.*
 |
Figure 1: Schematic of an SDS Gas Source
SDS Gas Source - how does it work?
The new source consists of a standard compressed gas cylinder, cylinder valve, adsorbent media, and adsorbed gas (arsine, phosphine, or boron trifluoride) (Fig. 1). The cylinder is charged with the dopant gas to just under one atm (-1 psig).
Adsorption forces. The gas source technology is based upon the physisorption of a gas molecule onto a solid microporous adsorbent. Physisorption results from the forces between the individual molecules of gas and the atoms or molecules comprising the adsorbent. Adsorption forces [11] are typically classified as London dispersion and can also include electrostatic forces if either the gas or solid is polar in nature.
The total interaction energy of the adsorbent with the gas molecule is:
((z) = (D + (R + (elec
where
z = height of the molecule above the adsorbent
fD = energy due to dispersion forces
fR = energy due to repulsive forces
felec. = energy due to electrostatic forces (dipole and quadrupole interactions)
The magnitude of these forces is relatively small (5-25 kcals/mole) [12], which explains why adsorption processes are reversible, but are strong enough to limit the mobility of the adsorbed gas. Polar adsorbents such as zeolites tend to adsorb more strongly due to the presence of the electrostatic forces, whereas nonpolar sorbents, such as activated carbon, adsorb less strongly.
 |
Adsorbents. Some of the common microporous solids include zeolites, aluminas, silica gels, and activated carbon (Table 3). All exhibit relatively large surface areas and micropore volumes.
The properties listed in Table 3 largely determine the adsorption isotherm for a particular gas/sorbent combination. Generally a gas molecule will adsorb more strongly on an adsorbent that has a pore opening 1-2? the molecular diameter of the gas because dispersion forces are maximized in this condition. For the SDS Gas Source technology, it is important that the adsorbent take up large amounts of gas, but not so tightly as to limit desorption.
 |
Figure 2. Isotherms for arsine adsorbed on different adsorbents at 20?C.
The interaction of these properties and their effect on the arsine adsorption isotherm can be seen in Fig. 2. The isotherm represents the amount of gas adsorbed/l of adsorbent as a function of equilibrium gas pressure at constant temperature.
 |
Of the three isotherms, activated carbon yields the most favorable adsorption characteristics [13]. This result is a combination of the very large micropore volume and an idealized pore-size distribution. The zeolite adsorbent adsorbs arsine very tightly with a large fraction adsorbed below 100 torr due to the very small pore size (5 ?) and the ionic character which tends to increase the adsorption interaction energy. The relatively poor adsorption capacity of the silica gel is attributed to its large pore size.
Since the adsorption process is reversible, the amount of arsine stored between 650 torr and 20 torr (working capacity) is approximately the quantity that can be extracted by the ion implanter (Table 4). Phosphine isotherms on these adsorbents exhibit a similar trend.
Delivery methodology
Since the SDS Gas Source is at 0 psig, special delivery techniques are required. Gas is generated using the differential pressure between the ion source chamber (10-6 torr) and the cylinder (650 to 20 torr). This technique is commonly referred to as vacuum desorption [14].
 |
Figure 3. SDS Gas Source delivery scheme.
High-conductance metering valves, thermal mass flow, and pressure-based flow controllers are employed to control flow. The scheme used must be capable of delivering ~5 sccm with pressure differentials as low as 10 torr in order to achieve maximum cylinder utilization. A typical delivery system using a pressure-based flow controller is shown in Fig. 3.
Since the flow rates for ion implant are relatively small, 0.1-5 sccm, the mass transfer and mechanical pressure drop limitations are usually negligible until cylinder pressures of 5-20 torr are reached.
 |
Figure 4. Overall equipment effectiveness for an implanter using a vaporizer.
 |
Figure 5. SDS Gas Source productivity benefit vs. solid source.
Productivity and overall equipment effectiveness
The productivity advantage of the SDS Gas Source is approximately the same as that of high-pressure gas systems over a vaporizer (Fig. 4). Species tuning times account for about 15-20% of the total implanter availability [15]. The differential time savings with an SDS Gas Source is nominally 10-25 min/species change. Implanter availability increases of an additional 1-2 hr/day are obtainable.
The productivity payback is largely a function of the number of species or recipe changes made/day [16]. Using SEMATECH`s cost of ownership criteria, the net productivity benefit is shown in Fig. 5.
Gas release studies
The opportunity for a gas release is significantly reduced over high-pressure cylinders as there is no internal pressure to drive the release and the dopant is "held" on the adsorbent. Roy F. Weston Inc. conducted a series of gas-release studies using arsine, phosphine, and BF3 SDS Gas Source cylinders [17]. The study was designed to simulate the release rate that would occur in an ion implant gas box if the cylinder valve were left open during handling. The release tests were conducted over a one-hr period to produce a time-weighted average. The results of the tests are summarized in Table 5. The release profile for a 600-torr arsine cylinder is shown in Fig. 6.
The time-weighted average release rates for all SDS Gas Source cylinders resulted in exposure concentrations below the threshold limit value (TLV, a concentration usually expressed in ppb). The arsine cylinder exceeded TLV momentarily after opening. In all cases, the exposure concentrations were below limits set by NIOSH (see Table 1). Since the Weston results were produced under specific conditions, release rates will vary depending on a number of variables including pressure and temperature.
 |
Figure 6. Study of arsine release rate at 20?C.
Secondary benefits accrued from safer gas sources
In addition to the safety and productivity aspects described above, the following ancillary benefits are readily identified.
1. Fab layout. The requirement to isolate the implanters because they use high-pressure toxics is removed, paving the way for the "ballroom configuration." Using the ballroom layout, the implanters may be located toward the center of the fab, providing a more efficient work flow. The improved logistics are expected to reduce chip-manufacturing time 3-5%. Lucent Technologies, National Semiconductor, and Rockwell have adopted the ballroom layout.
2. Reduced exhaust air. Eliminating the possibility of an emergency gas release allows the ventilation rate through the implanter gas box to be reduced significantly.
3. Emergency scrubbers. The need for catastrophic release scrubbers is eliminated, since release rates are intrinsically < 1/2 IDLH (immediately dangerous to life or health) values.
4. Cryopump regenerations may be reduced.
5. Duct fires associated with phosphorus build up in the exhaust are reduced.
Summary
The SDS Gas Source promises to be a major enabling technology in the manufacture of integrated circuits. The safety of dopant use/ handling and implanter productivity are both enhanced.
Acknowledgment
SDS Gas Source is a trademark of Advanced Technology Materials Inc.
References
1. Ion Implantation Science and Technology, edited by James Ziegler, Ion Implantation Technology Corp., 1966.
2. P. Pei, H.R. Kirby, "Next Generation Processes and Equipment that Lead to Positive Environmental, Safety and Health Impacts, Semiconductor Fabtech, pp. 25-29, ICG Publishing, 1996.
3. Pocket Guide to Chemical Hazards, NIOSH Publications, Cincinnati, OH, June 1994.
4. J.V. McManus, G.M. Tom, R. Kirk, "A Zeolite-based Atmospheric Pressure Hydride Gas Source for Ion Implantation," Proc. of the X Int. Ion Implant Tech. Conf., pp. 523-526, June 1994.
5. S.R. Walther, S. Mehta, R.L. Brown, R. Kaim, "Ion Implant, Results for Zeolite-based 100% PH3 and AsH3 Gas Bottles on the Varian E500 Implanter," Proc. of the X Int. Ion Implant Tech. Conf., pp. 519-522, June 1994.
6. K. Brown, S. Walther, J. Jillson, "Ion Implanter Testing and Process Results Using Low-pressure, Zeolite-based Gas Bottles in Medium and High Current Ion Implanters," Proc. of the XI Int. Ion Implant Tech. Conf., to be published (IEEE).
7. R.L. Brown, "SDS Gas Source Feed Material Systems for Ion Implantation," Proc. of the XI Int. Ion Implant Tech. Conf., to be published (IEEE).
8. US Patent No. 4,744,221 entitled "Zeolite-based Arsine Storage and Delivery System," May 17, 1988.
9. US Patent No. 5,518,528 entitled "Storage and Delivery System for Gaseous Hydride, Halide and Organometallic Group V Compounds," May 21, 1996.
10. T. Marin, W.G. Boyd Jr., J.V. McManus, "Performance of the SDS, High-pressure Hydrides and Solid Vaporizer Feed Materials on a 9500xR Implanter," Proc. of the XI Int. Ion Implant Tech. Conf., to be published (IEEE).
11. S.J. Gregg, K.S.W. Sing, Adsorption, Surface Area and Porosity, Academic Press, 1982.
12. D.P.Valenzuela, A.L. Myers, Adsorption Equilibrium Handbook, Prentice Hall, Englewood Cliffs, NJ, 1989.
13. J.V. McManus, D. Edwards, R. Kirk, "New Developments in SDS Technology for Ion Implantation," Proc. of the XI Int. Ion Implant Tech. Conf., to be published (IEEE).
14. R.T. Yang, Gas Separation by Adsorption Processes, Butterworth, New York, 1987.
15. OEE model courtesy of James Neroda, Eaton Corp., Beverly, MA.
16. B.E.P. Beeston, "Ion Implant Modeling: Does It Give Us What We Want?" Proc. of the XI Int. Ion Implant Tech. Conf., to be published (IEEE).
17. The complete report is available from Advanced Technology Materials Inc. upon request.
TERRY ROMIG received his BA degree in chemistry from Texas A&M University, and worked with cyclotrons for six years before joining National Semiconductor in 1989, as an implant engineer. He is currently the project manager for installed-base implanter development at SEMATECH, as an assignee.
JIM MCMANUS received his BS degree in chemistry from the State University of New York at Fredonia. He is Technology Development Manager for SDS products at Advanced Technology Materials, and holds eight US patents in the areas of electronic gas purification, semiconductor effluent waste treatment, and low-pressure gas delivery systems. Advanced Technology Materials Inc., 7 Commerce Dr., Danbury, CT 06810; ph 203/794-1100, fax 203/792-8040, e-mail SDS @ATMI.com.
KARL OLANDER received his PhD degree from the University of Illinois. A founder of Advanced Technology Materials, he is currently VP of its NovaSource Division. He was previously VP of new business development and president of the Novapure Corp., a joint venture with Millipore Corp. He holds more than 30 US patents.
RALPH KIRK received his PhD degree in high-temperature boron, fluorine, and silicon chemistry from the University of California at Berkeley. He joined Matheson Gas Products in 1992, where he is currently manager of new technologies in the Electronics Products Group.