Debate Over Shallow Junctions—How Low Can You Go?
-- Semiconductor International, 2/1/2001
Recent research and development in the area of shallow junction ion implantation and RTP "spike" annealing has led to recommendations on ways of continued scaling of ultra-shallow junctions to the 0.10 µm generation and perhaps beyond. Nobody in the industry argues a requirement for ultra-low energy implants to reduce junction depths by reducing transient enhanced diffusion. But the lowest practical implant energy, influenced by such issues as dopant self-sputtering and tool throughput, has yet to be determined.With the lowering of ion implantation energy, at some point the engineer must face a trade-off between forming shallower junctions and the higher junction resistivity that results. In fact, a comparison of sheet resistance and junction depth data from many different sources showed there is more than one way to arrive at the same junction properties. Though energies can be lowered to produce junctions as thin as 30-40 nm with existing implanters, the device manufacturer must determine whether the corresponding resistivity of 400-1000 W/sq provides acceptable performance.
Aside from performance issues, a lowering of ion implantation energy leads to an increasing penalty in throughput. New studies indicate that it may be preferable to reduce the implant dose at a sufficiently low energy, according to Aditya Agarwal of Axcelis Technologies (Beverly, Mass.). Indeed, Axcelis engineers found that equivalent boron implantation results could be achieved with a lower-dose 0.5 keV implant as with a higher-dose 0.2 keV implant. This approach lessens the painful throughput penalty associated with ultra-low energy implants, which can increase process time by >3×.
In a similar manner, the spike anneal — dubbed this because of its infinitesimally short time at peak temperature — dictates a high target temperature for sufficient dopant activation and damage removal. But the peak temperature and time near peak temperature must be minimized to reduce diffusion and render an abrupt junction.
|
|
Though commercial ion implanters have been designed to implant at energies as low as 0.1 keV, for several reasons Agarwal argues against reducing ion implantation energy too much. For one, self-sputtering — the sputtering of target material during ion implantation — can become more of an issue at lower implant energies because more atoms are sputtered off the target per incoming ion. In addition, as an increasingly significant number of dopant atoms come to rest near the surface, even more dopant is lost. Self-sputtering can lead to a 20% dose loss at 0.2 keV (Figure). Agarwal added that self-sputtering becomes even more of an issue for BF2 implants. As a result, 0.5 keV may be the lowest practical limit of implant energy.
— Laura Peters
