Battle heats up for growth market in implant
03/01/2002
The ion implantation market appears set for powerful growth, projected to almost triple over the next five years from an estimated US$819 million in 2001 to US$2.433 billion by 2006, with a CAGR of some 24%, according to VLSI Research (Table 1). Three vendors, Axcelis Technologies (including the company's joint venture, Sumitomo/Eaton Nova), Applied Materials, and Varian Semiconductor Equipment Associates (VSEA), split most of the market, with leadership bouncing from one to the other in the major categories: high-current, medium-current, and high-energy implanters (Table 2). There is an intense battle for market share, with each vendor claiming supremacy for features in contending systems, benefiting users as a result. While all the toolmakers have made great strides in meeting ITRS roadmap challenges, much more remains to be done.
 |
"The ion implant equipment suppliers have done an excellent job with new tool sets that can do ultra-low energy implants," says Larry Larson, associate director and gate stack/USJs/device compatibility program manager at International Sematech. "There are no clear weaknesses that need to be addressed from a process technology standpoint, but productivity issues and cost-competitivness need continued emphasis." While praising the equipment suppliers' process technology efforts, he thinks there remains a major challenge in getting RTP equipment up to speed for USJ technology.
 |
Applied Materials, a newcomer to the medium-current and high-energy segments, seems to have left the older technology market to others. The company's technology roadmap for transistor doping starts at 0.13μm (see figure on p. 34), the reasoning behind the company's combining medium-current and high-energy capabilities into a single platform.
 |
Axcelis retains leadership in the separate high-energy market segment. VSEA says its products have the range to cover roadmap requirements for low-energy, source/drain extensions, well-to-well spacing, and halo and pocket implants
The big three: Accuracy, flexibility, productivity
Specific areas that many agree have attained pre-eminence in equipment selection are doping accuracy, flexibility, and productivity. As with most complicated subjects, the devil is in the details. Take doping accuracy, for example. Three factors controlling it are energy, dose, and angle of implant. As the industry moves toward narrower channels and smaller linewidths, dopant placement becomes crucial to achieving device performance targets, including higher speed and lower power consumption, as well as process repeatability.
For controlling angle accuracy, Applied employs a mechanical setup through the use of an air-bearing vertical scan technology, beam parallelism control, and calibration techniques. Axcelis credits several factors: its use of closed-loop dose control, features such as an angular energy filter in its single-wafer platform, and compensation for ion charge neutralization (or stripping) in its multiwafer tools. VSEA says it uses real-time, closed-loop dose control and a parallel scan system for precision doping.
Applied is attempting to change the way ion implantation is categorized — the company likes to use conductive and parametric, instead of the traditional segmentation by current and energy, especially with its multirange tools. But to many users, production flexibility is really the issue. Here, the lines are drawn between a single-wafer scheme vs. a batch tool (or multiwafer tool), and that leads to the productivity debate. Indeed, single-wafer vs. multiwafer (batch) processing is a hotly contested topic.
Axcelis offers both multiwafer and single-wafer systems, saying that users can attain reduced cycle times by chaining on multiwafer systems and using single-wafer tools for special applications, e.g., high-tilt implants. (Chaining is running different implants on the same wafer at the same mask level without unloading the wafer from the tool.) "Multiwafer implanters always tune faster than single-wafer systems due to simpler beam lines," states Tom Parrill, Axcelis' product line manager. "But this strategy allows one to select the minimum number of single-wafer systems to cover special applications and run the majority of implants in chained mode on multiwafer systems."
According to Babak Adibi, Applied's senior director of group strategic marketing and technology, the company tackled productivity by making heavy use of queuing theory to stake its claims. It used data gathered from its customer base to develop a list of WIP management issues as a starting point. Applied believes its two-pronged approach — a single-wafer tool for medium-current/high-energy implants, and a batch tool for high-current implants — will be the winning combination. The company contends that the traditional approach of having process-dedicated systems is less efficient from a fab management standpoint.
Problems have been reported with batch implant tools, however [1]. The beam incident angle does not remain at recipe-prescribed values, resulting in a variation in device parameters that is becoming more noticeable at advanced technology nodes.
VSEA considers its "bay-level" approach a key productivity factor. "The opportunity to control and optimize the flow of product within the implant by the fab's host computer enhances manufacturing flexibility and increases utilization within the bay at any point during the fab's manufacturing ramp curve," comments David Hacker, strategic marketing director at VSEA.
Is there a clear winner?
So which approach will win out? It would be nice to have an independent evaluation of the major approaches being promulgated, but, as Larson notes, "The relative benefits of the approaches depend strongly on the fab lines' process/product mix and the rate of changes between recipes." Since the processes necessary to do a full investigation can't be cleanly clustered, International Sematech does not have data to fairly compare the single-wafer vs. batch approach.
As far back as 1999, Eaton (now Axcelis) articulated the belief that single-wafer, medium-current implant would become a niche, while batch high-energy and high-current segments would grow as device scaling, integration, and productivity issues came to the fore [2].
Kiril Pandelisev, founder/president/chief scientist at Single Crystal Technologies, basically agrees with this premise. "High-current, high-yield, low-cost device fabrication process tailoring will determine whether batch-type processing or single-wafer processing is better," says Pandelisev. "Ion implant processes for ultra-shallow junction formation that enable smaller transistors (90nm, 70nm, or smaller technology nodes enabled via 193nm and 157nm technology applications) and higher device speeds will be the major driver."
Complementary product offerings play a role as competition in ion implant spills over into other product segments, as Tom Seidel, executive VP/CTO at Genus, notes. Applied Materials has high-end RTP solutions and Axcelis offers RTP, as well as resist strip and clean, for example. Last year, VSEA announced the joint creation — along with Ultratech Stepper and TEL — of an advanced technology center dedicated to USJ formation for the sub-70nm node, in which each company could contribute its piece of an integrated solution. "The announcement was a good initiative to compete broadly," comments Seidel.
Are contending claims much ado about nothing? "The big three compete with one another by looking for process advantages. They look for process niches," states Robert Kaim, an industry consultant, formerly with both Eaton and Varian and a co-founder of ASM Ion Implant. "My contention is that processing niches have never yielded a knockout blow because the implant process hasn't been changing fast enough." But fabs, facing one processing challenge after another as films get thinner and devices shrink, probably hope that the battle will go on for a long time, making a variety of options available to them.
References
1. U. Jeong, J.Y. Jin, S. Mehta, "Devices Dictate Control of Implant-beam Incident Angle," Solid State Technology, 10/2001.
2. T. Parrill, M. Namaroff, "Productivity for Advanced Applications," European Semiconductor, 7/1999.
Debra Vogler is a senior technical editor at Solid State Technology magazine, ph 408/774-9283, e-mail [email protected].