Letters
02/01/2002
Stop giving US secrets away
Your December editorial "Note to US media: Use your heads!" was well said. My frustration level is high every time I hear "secrets" being given away on the news. I wish there were more in the media industry taking the same stance. — Bob Greenberg, Texas Instruments
I wish to compliment you on your editorial. I plan to show it to several others. I would also hope that at least a portion of the mainstream media receives it. — Charles S. Thompson, Texas Instruments
Thank you for such a well-timed and extremely important editorial. This particular topic hits home with me. While I am now the marketing communications manager at MEMC Electronic Materials, I am also a West Point graduate, a combat veteran of Operation Desert Storm, and a former intelligence officer. I really appreciate someone in the industry taking a stand on the issue of "too much information" during an era of easy-to-access mass communications. You are right; we are at war. Although media personnel have an obligation to accurately report the news, they should respect the fact that such reporting might cause needless casualties, both civilian and military. — Brett W. Avants, MEMC Electronic Materials Inc.
_________________________________________
Debate on October ion implant article
The article, "Devices dictate control of implant-beam incident angle" (October, p. 79), contains factual errors. It refers to monolithic "batch implanters" as if all have identical beam angle variations regardless of end station design or selected tilt angle, which is not so. Beam angle variations depend in a complex way on the selected tilt angle, angle between wafer pedestal and disk's plane, and number of wafer sites on the disk. So, any analysis must include more specific information.
In Fig. 1, the red oval is asserted to be tilt and twist angle variations for a "batch implanter" set to 7µ tilt and 22µ twist. The authors misplaced the oval's center at 22µ twist from the <004> planes. The notch or major flat on a silicon wafer (from which 0µ twist is always counted) indicates the <022> direction, not the <004>. This mistake introduces a 45µ twist error, invalidating their discussion of planar channeling. Also, the oval's center tilt angle is 8µ. The correct polar coordinates on the underlying data for a tilt, twist angle of 7µ, 22µ are 7µ tilt from the origin point and 67µ twist from the <004> planes. Finally, the tilt and twist angle seen by a wafer in a batch implanter is an arc, not an oval. The precise position and length of this arc depends on the factors listed above, but for an optimized implanter setting would be about 1/3 the size of the red oval in the figure.
Fig. 2 purports to show simulated doping profiles for a phosphorus implant at 3µ tilt, but no information is given about design or settings of the batch end station. Our calculations show it roughly describes the situation of a 3µ implant on the Varian Kestrel high-energy implanter, which has the highest beam angle variations of any implanter. On a properly designed end station, this implant's effective tilt angle would be 3.0-3.1µ across an entire 200mm or 300mm wafer, with a max. twist variation of ±8.5µ, not the ±13µ cited.
In the source/drain extension-doping section, a space charge argument implies significant beam divergence angles in batch implanters, especially at low energies, ignoring the effects of space charge neutralization provided by electrons generated by beam collisions with residual gas molecules and by plasma flood guns. For the cases mentioned (5keV and 2keV As+), the beam divergence has been measured on a high-current batch ion implanter to be approximately ±2µ for state-of-the-art beam currents (10mA and 3mA). It is inaccurate to imply that angles as large as those mentioned are typical in low-energy implant tools. It can also be misleading to state that reducing beam divergence increases drive current in simulated devices. Drive currents can be increased by changing the dose or energy of the extension or halo implants, but typically with increased off-state leakage current. Since there is no mention of leakage currents, we assume simulations showed no net improvements in device performance.
The article ignores important issues by neglecting any mention of across-wafer beam angle variations in single-wafer implanters, particularly those using magnetic angle correction. In a poorly tuned serial implanter, angle variations can be larger than on most batch implanters with similar effects on devices. There is also no mention of wafer repositioning ("quad implants") as a production-proven solution to implant-shadowing problems. —Leonard Rubin, Michael Graf, Axcelis Technologies, Beverly, MA
- - - - - - -
We affirm claims and key points in our article. Beam angle variations are inherent to spinning-disk type batch implanters. The magnitude of tilt and twist angle variation is a function of pedestal (cone) angle and the distance from the disk center to the wafer center. End station design can moderate effective angular variation for certain tilt and twist angles, but the resulting control is inadequate for 0µ tilt implants that are critical for advanced device fabrication.
The trace of tilt and twist angle variation in a spinning-disk batch implanter follows a true arc only for a point-sized ion beam spot. In reality, any beam has a non-zero spot size that increases as beam energy drops. This imposes a range of tilt and twist angles on the wafer that is best represented by an area in Fig. 1 instead of an arc. Regrettably, the location of the oval was misplaced, but note that several channeling axes and planes still exist in the region of interest.
Incident angle spread's effect on NMOS transistor performance was investigated using TSUPREM4 and MEDICI simulations showing comparable off-state current performance for single-wafer vs. spinning-disk batch. This equivalence confirms the net increase in transistor performance with single-wafer implantation. Also, the simulation results of transconductance demonstrated performance gain was due to reduced series resistance at the SDE edge provided by the shadow-free single-wafer implant. Advanced devices require precise dopant placement independent of tilt or twist angle across a wafer. Only single-wafer ion implanters with parallel ion beams and total in situ control of the beam incident angle can satisfy these requirements. — Ukyo Jeong, Sandeep Mehta, Varian Semiconductor Equipment Assoc., Gloucester, MA
_________________________________________
Clarification
In January's special ITRS Roadmap report, the web site for viewing the roadmap was given incorrectly in one instance. The web site is http://public.itrs.net.