Plasma Arc Implanter Features
This page summarizes the most important features of the Plasma Arc Implanter for ion implantation (doping) in semiconductor manufacturing. The emphasis is in the p-doping (with boron ions).
Boron doping in semiconductors by use of the boron cathodic arc
(by J.M. Williams, CTO, 13-Feb-2007)
Robustness: Very robust singly charged boron ion plasma source with up to 2 A of unfiltered and 400 mA of macroparticle-filtered, effective boron ion current at a nascent energy of 20 eV (based on Anders and Yushkov for C [1]) for boron ion energy, followed by acceleration as desired.
Delivery Schemes: Versatility in delivery to target is possible because of useful directional velocity and full space charge compensation of nascent plasma (or low energy ion beam). Plasma immersion (target bias) may be used. Also acceleration by conventional extraction is possible, or an accelerated beam may be formed by gradual acceleration with magnetic containment and maintenance of space charge neutrality. .
“Greenness:” Absolutely the safest and most environmentally friendly process because of no gas used whatever. That means no corrosive gas, no poison gas, no solid rocket fuel (boranes), no gallons of liquid chemical waste generated from cleaning up systems [2]. No solid waste upon system disposal requiring special attention. No gases exhausted in the process. Small footprints of systems, much smaller pumping systems than those needed for gas sources, etc.
Potential for Radiation Hardness: Simpler and more economical opportunity to continue providing for delivery of isotopic 11B in shallow junction doping than any of the other two emergent schemes of boron generation and delivery (PLAD or cluster ion). This process is needed to prevent the neutron fissionable isotope, 10B, from being implanted in circuits.
Charging: Superior intrinsic defenses against charging exist because of lower voltage for given ion ranges, only the ions needed are delivered, and fewer secondary electrons are emitted (in comparison with BF3 -PLAD) Nascent plasma with full space charge compensation may be deliverable for very low energy implants (20 eV). For keV energies, gradual acceleration is believed possible while preserving space charge compensation. For the wafer bias mode, the spark arrested d.c. power supply has been used with complete success in the intermediate voltage range (to 600 V), but has not been tried with devices with gate oxide yet. The present system was developed for and has been used extensively for depositing thick insulating layers of pure boron at energies of up to 550 eV without arcing. Pulse biasing could still be used if needed.
Macroparticle Filtering: Present filter design reduces particle transmission to about 5/coulomb of transmitted boron ions at a position just past the filter. That would be about one particle per 30-cm wafer at that position. However, the wafer will be considerably removed from that position, for various reasons. There will be ample opportunity for further improvement. Particles are captured. Overall cleanliness for this type of system will be excellent.
Energy Contamination: Energy contamination due to possible doubly charged B ions in the plume has been eliminated as a consideration by the present experiments. The plume energy of 20 eV has a variance of about +- 1 eV [1]. In the present experiments target biasing is virtually continuous d. c with periodic downspikes for arc control. The fraction of up time is overwhelming. In contrast to cracked BF3, which contains free B ions at the bias energy, in addition to the designed-for BF2 ions, the present plasma contains only B ions. As a result there is mono-energetic B ion entry and considerably less energy contamination than for PLAD with B ions in BF2 and free B ions.
Voltage: Minimum acceleration voltage for desired treatment depth aids processing in several ways: charging, energy contamination and operational advantages such as commercial power supply availability. A bias of 500 V is equivalent to 2300 V for BF3 – PLAD. A very high rate of delivery is possible at a true low voltage.
Secondary Electrons: Minimal secondary electrons are emitted because of minimal polarization or acceleration voltage for the desired result, and because of no unwanted ions in the flux.
Heat and Power: Minimal heat in the wafer for the desired result because of lower voltage and not accepting co-implanted cracked BF3 products. Also, less power is needed because of fewer secondary electrons.
Dosimetry: Dosimetry depends only on detecting actual boron ions in faraday devices, Langmuir probes or actual target. No neutral fraction.
Purity: Rutherford Backscattering (RBS) and non-Rutherford elastic (NRES) analyses [present work] of the orthopedic coatings indicate a maximum carbon impurity level of 3 at % C for the plume delivered with the present, nominally pure, cathodes. If more carbon is desired, perhaps for some advantage [3], implantation using boron-carbide cathodes within the composition range of the equilibrium phase field (9 to 21 at. % C) is even easier [4]. Metallic impurities are not detected, although the anode is of tungsten. [Present work and 5].
Vacuum: Excellent vacuum is possible in plasma generator and delivery system. System vacuum issues are limited only by target requirements. Only small pumping systems are needed for the source. Boron ion throughput has nothing to do with gas throughput.
Safety: There is intrinsic safety, since no gases whatever are required for boron. That means no decaborane, no BF3, etc.
Sheath: In plasma doping mode, the technique provides the thinnest sheath of the plasma doping techniques [6]. High plasma ion energy (20 to 30 eV eV [1, 7]), very low electron temperature, and high plasma density imply thin sheath. Ions are not isotropic but somewhat pre-directed before entering acceleration through the sheath near the wafer surface. At the incident velocity and arrival rate, a steady state sheath is set up, which does not expand due to depletion [6]. Again, no gas is in sheath, unless introduced to perform some function, such as scattering to sidewalls. (When is thin sheath an advantage and when a disadvantage?)
Radiation Damage: Minimal total radiation damage because of no unwanted or unneeded atomic constituents in plasma. PLAD with BF3 produces a factor of about 8 times as many vacancies and interstitials for the given injection of B at an effective B energy of 500 eV or at 100 eV for example (TRIM calculation). If amorphization occurs, that might be an advantage for BF3-PLAD, but otherwise more transient enhanced diffusion (TED) presumably occurs for BF3. For the cathodic arc method, damage is confined to implant region because no products of cracked molecules with higher ranges at the bias energy are involved.
Process Design Flexibility: Desired gases, particularly inert, could be added to the process by intent to produce advantages as desired. This could be done at the arc to produce a fraction of heavier ions for early amorphization. Gas could be introduced at the wafer surface to provide for scattering for treatment of trenches. Any gas process introduced could be optimal for the purpose, rather than having to respect ion generation as part of the process optimization.
Target Sputtering: Process has minimum possible target sputtering because of no extra atoms beyond what are objectively needed, no extra energy, and no corrosion sputtering or etching. Sputter limited incorporation of B will be highest in comparison with other techniques. Sputtering due to the combined effects of the constituents in BF3 for a given B energy is approximately a factor of 5 greater than for the present pure B implantation. With no unwanted ions in the bombardment, the sputtering rate of Si-B decreases with increasing B over a wide range [5]. Therefore, saturation does not readily occur, and as much B as desired can be added.
System Wall Sputtering: Minimal system sputtering will occur because of only B ions at desired energies, and no corrosion sputtering, or etching. ALSO, non-isotropic streaming boron plasma will be screened off of system walls by magnetic ducting or other magnetic containment in viable systems designs, which also involve minimal wall area.
High Availability: A small volume of solid B cathode contains more B atoms than a huge volume of gas. One gram of cathode (delivered) will dose 60,000 wafers passes. Fraction of cathode delivered will be small because of macroparticle losses and further plasma losses in filters. Still there is plenty for semiconductors, bearing in mind that the process is strong enough to be a coater. Therefore, despite macroparticles, availability is still expected to be higher than for other systems in use for semiconductors. This type of hardware is used for coating many products such as plumbing fixtures, door knobs and cutting tools. Much is known about operation, and adaptation to the special needs of the semiconductor industry is possible. There is a companion program in boron orthopedic coatings, which will have more in common with semiconductor industry needs than do some other applications of cathodic arc.
Cleanliness: No hygroscopic corrosion product, other corrosion product, or gas adsorption in walls, to affect turn-around in wafer cycling or to make particles.
Systems: Simple, compact, and elegant tool designs are possible for high compatibility with other tools and for use in the most basic of infrastructures. No venting, etc.
References:
1. Andre Anders and George Yu. Yushkov, J. Appl. Phys., 96 (2004) 970
2. Frank Dimeo, et. al., The development of novel in situ cleaning processes fro ion implanters, paper O703, 16th International Conference on Ion Implantation Technology, June 11-16, 2006, Marseille, France.
3. Wade Krull, et. al., Simplifying the 45 nm SDE process with ClusterBoron TM and ClusterCarbon TM Ion Implantation, paper O1204, 16th International Conference on Ion Implantation Technology, June 11-16, 2006, Marseille, France.
4.. O. R. Monteiro, M.-P. Delplancke-Ogletree, C. C. Klepper, J. Materials Sci., 38 (2003) 3117.
5. J.M. Williams, C. C. Klepper, and R, C, Hazelton, Nuclear Inst. Methods in Physics Res., B237 (2005) 278.
6. Andre Anders, Surface and Coatings Technology, 93 (1997) 158.
7. Frank Richter, Siegfried Peter, Volodymer B. Filippov, Gert Flemming, and Michael Kuhn, IEEE Transactions on Plasma Science, 27 (1999) 1079.