Enabling concepts for low-energy ion implantation
04/01/2005
Significant improvements in the mid-1990s, including advances in ion source technology and the introduction of reduced path-length beamlines, enabled production-worthy beam currents as low as the 5-10keV range of energies. It was not until one generation later, in the late 1990s, that further design innovations, including the introduction of electron injection and confinement technologies, made large advances in the 2keV range of energies. Since then, additional beamline optimization around fundamentally stable architectures continues to deliver steady incremental improvements at all energies.
Beam transport model
Some of the important beamline design considerations that must be addressed for high-efficiency beam transport at low energies can be uncovered using a simple beam-transport model [1]. This model describes the size and shape of the beam at all points in the beamline and takes into account the dominant space-charge forces that cause beam expansion and lead to the loss of usable beam current.
Figure 1 shows the calculated distance along the beamline over which the beam would double in size as a result of beam space-charge expansion alone. The data are for a beam of B+ ions of 10mA total current drifting through a beamline at an energy of 2keV. At one extreme, where beam neutralization is perfect (0% nonneutralized) and there is no space charge, an ideal parallel beam would remain parallel over very long distances. At the other extreme, where a beam is fully stripped of all electrons and space charge is at its maximum (100% nonneutralized), the distance it takes for a beam to double in size is quite short, on the order of a few tens of millimeters. This is problematic, considering that even the highest-output low-energy beamlines are still ~1500mm long. Any small region of high space charge could lead to beam expansion and significant beam loss.
Implications for beamline design
While one extreme of beam neutralization (a perfectly neutralized beam with no space charge) is difficult to achieve in practice, recent innovations have made significant advances toward this goal. Passive electron confinement technologies that use specially shaped magnetic fields to slow the loss of neutralizing electrons to the wall of the beamline, as well as active electron injection from plasma discharge sources that increase electron populations in the beamline, have enhanced usable beam current and have improved beam size and shape control [1].
The other extreme of beam neutralization (a fully stripped beam with maximum space charge) is unfortunately quite common in any region in the beamline where there is an externally generated electric field, such as in the vicinity of an electrostatic lens. Simple electrostatic lenses typically are designed into beamlines to focus the beam or act as a potential barrier to restrict the backflow of electrons to regions of higher potential.
Beamlines using deceleration or differential modes of operation, in which the beam is transported through much of the beamline at some higher energy and then decelerated to the final desired energy just before reaching the wafer, also utilize simple electrostatic lenses. Occasionally, even beamlines operating in drift mode, where in principle the beam is transported through the entire length of the beamline at its desired energy, make use of electrostatic lenses for beam focusing.
One drawback of this lens configuration is that it generates a region of the beamline that is completely stripped of all electrons that would normally neutralize some of the space charge in the beam (i.e., regions of 100% nonneutralized beam). The longer the path length is through the lens, the greater the fully stripped region and space-charge expansion of the beam. As seen in Fig. 2, even relatively short lengths where the beam is fully stripped can affect beam size. The lens focusing strength will mitigate the space-charge expansion somewhat, but will require a larger applied voltage than would otherwise be necessary without the strong space-charge forces.
 Figure 2. Path length to double beam size vs. % nonneutralization of the beam. |
The low-energy beam loss and refocusing produced by a lens is demonstrated by the data shown in Fig. 1, where the usable delivered beam current for a 2keV beam of B+ ions is plotted as a function of the voltage applied to a simple electrostatic lens ~500mm upstream of the wafer. When no voltage is applied, the delivered beam current is at its maximum, because no regions of high space charge exist in the beamline. Once some voltage is applied to the lens, a fully stripped region is created in the beamline, causing significant space-charge expansion of the beam, leading to significant loss of usable beam current. When the applied voltage is increased significantly, some refocusing takes place and some of the usable beam current is recovered.
Another drawback of the requirement for a large focusing voltage is creating a region of the beamline (typically in direct line-of-sight of the wafer) where the beam is no longer at its desired final energy. In this respect, operation with a strong focusing lens, even in drift mode, presents some of the same risks of energy contamination as operation in decel or differential mode. Namely, for the time that the beam is being accelerated and then decelerated in the lens, any charge exchange reaction occurs will produce energetic neutrals at an energy higher than the desired beam energy. The quantity of energetic neutrals produced will be proportional to the pressure in the lens region and also to the physical length of the lens.
The energy at which the energetic neutrals arrive at the wafer will depend on exactly where in the lens the neutrals are created, and thus a range of energy contaminants is generated. The focusing voltages applied in such a lens to overcome the space-charge expansion forces can be large compared to the transport energies in deceleration mode. This introduces energy contamination concerns somewhat beyond even those posed by conventional deceleration schemes.
Particular choices of beamline architecture and technology mitigate these concerns. The need for focusing lenses in drift mode operation is practically eliminated once sufficiently high levels of beam neutralization enable efficient beam transport at the lowest energies required in high-volume manufacturing. If focusing lenses are still present, minimizing their physical length reduces both the region over which space-charge expansion occurs, and the likelihood of introducing high-energy contamination due to charge exchange reactions in that region. Reducing line-of-sight from the lens to the wafer also may filter out any energetic neutrals created in the lens region or anywhere upstream. Filtering concepts with electrostatic deflection of the beam have long been used to mitigate this effect successfully on medium-current platforms [2] and are now finding their way into high-current beamlines as well.
Conclusion
High-productivity solutions for low-energy ion implantation are enabled by beamline designs that improve beam transport by controlling space-charge forces on the beam through the entire length of the beamline. Important architectural choices that can fully realize the potential of these designs without introducing new process risk need to be considered.
References
- M. Graf, B. Vanderberg, et al., “Low Energy Ion Beam Transport,” Proc. Intl. Conf. on Ion Implant, pp. 359-364, 2002.
- D. Kamenitsa, R. Rathmell, “Beam Energy Purity in the Eaton NV-8200P Ion Implanter,” Nucl. Instr. And Meth. B96, pp. 13-17, 1995.
Michael Graf is product technology manager at Axcelis Technologies Inc., 108 Cherry Hill Dr., Beverly, MA 01915; ph 978/787-9179, fax 978/787-3000, e-mail [email protected]. Brian Freer, Axcelis Technologies Inc., Beverly, Massachusetts