Track, Scanner Push for Litho Productivity
Aaron Hand, Executive Editor, Electronic Media -- Semiconductor International, 7/19/2007
“Speed is the key.” So go the famous last words of many a daredevil stunt rider — and a lithocell productivity manager. This was the focal point of the keynote address by Michael Lercel, director of lithography at Sematech/IBM, at yesterday’s 14th Annual Sokudo Lithography Breakfast Forum. The theme of the breakfast was “Lithocell Productivity: Scanner vs. Track.” But while speed and throughput are paramount for keeping productivity high, it means nothing without maintaining yield. As Lercel noted, “Litho cluster productivity is an important driver of cost per good wafer-level exposure.”
| Michael Lercel, Director of Lithography, Sematech/IBM |
With this point in mind, Lercel detailed the issues facing lithocell productivity today and in the future, including resolution enhancements (mask complexity), illumination optimization, and film stack improvements. Complexity is further compounded with increasing numerical apertures (NAs), and the increasing need for reflectivity control; the introduction of immersion lithography; and the prospect of double patterning, which would, for obvious reasons, cause a significant hit in throughput.
In addition to Lercel, presenting at the forum were Charles Pieczulewski of track supplier Sokudo; and Mark Slezak, technical manager of the microlithography group for materials supplier JSR Micro. The presentations were then followed by a panel discussion with the toolmakers, including Pieczulewski, Skip Miller of ASML, Phil Ware of Canon, and Gene Fuller of Nikon Precision.
As Pieczulewski pointed out, a general rule of lithography has been that you should always buy 10-20% more track throughput in case you underestimate your scanner throughput. The low-cost track system should never jeopardize the throughput of the high-cost scanner. But several factors are increasing the complexity of the track system, particularly with the introduction of immersion lithography. This means that more and more modules must be added to the track to help it keep up.
Because of defectivity concerns, as well as hyper-NA reflectivity, immersion lithography brings with it the need for more layers within the film stack. With a single-layer bottom antireflective coating (BARC), it is impossible to control reflectivity for multiple pitches at hyper-NAs. Therefore, multilayer BARCs are required to achieve the necessary reflectivity and, in turn, multiple coat steps are required within the track system.
| Sokudo Lithography Breakfast Forum Panelists (left to right): Skip Miller of ASML, Phil Ware of Canon, Gene Fuller of Nikon Precision and Charles Pieczulewski of Sokudo. |
Also contributing to the growing film stack is the current need for a topcoat to cover the resist layer in immersion lithography schemes. Slezak made a push for topcoats in immersion lithography, noting that the increased process steps were worthwhile considering the benefits they provide in minimizing leaching and greatly reducing defectivity. He argued that the “simpler” solution the industry is pushing for (i.e., a topcoat-less resist) may not actually make the lithography process simpler. Although removing the topcoat from the film stack would shorten the process cycle time, a topcoat would enable lithographers to make use of “the know-how that we’ve established over the last seven years on ArF resist design,” Slezak said.
During the panel discussion, Pieczulewski agreed, noting that it was relatively simple to compensate for these kinds of productivity losses within the track system itself. On the other hand, it’s not so easy to compensate for the defects on process variability that can be more common without a topcoat. Fuller, however, pointed out the improvements that have been made recently with topcoat-less resists, and that their performance was such that they would be preferable to the added film stack complexity.
A significant question addressed by the panel was the idea of delinking the lithography scanner and track system. As Pieczulewski pointed out, this was common practice some 20-30 years ago, and it may be time to go there again to realize productivity benefits through increased flexibility. Addressing the concern that delinked tools would still require comparable throughputs so that wafers were not lying around waiting to be produced, both Ware and Miller noted the abilities of onboard metrology to decrease downtime, and increasingly available software capabilities to better align productivity between various tools.
In general, the question remained open for debate. However, whether linked or delinked, all agreed about the need for better coordination between the two systems, and that the entire lithocell would have to work well together to achieve the highest productivity possible.
Webcast: Immersion Lithography
03/10/2004Extending Immersion Lithography to 32 nm
07/12/2006Dual-Track Sokudo Duo Boosts Throughput
04/21/2009
