Chapter 8: Nanolithography

The MAPPER Multiple-Electron-Beam System. # 2009 IEEE
The MAPPER Multiple-Electron-Beam System. # 2009 IEEE

Nanolithography, the patterning of hundreds to thousands of trillions of nano structures on a silicon wafer, determines the direction of future nanoelectronics. The technical feasibility and the cost-of-ownership for technologies beyond the mark of 22 nm lines and spaces (half pitch) is evaluated for Extreme- Ultra-Violet (EUV = 13 nm) lithography and Multiple-Electron-Beam (MEB) direct-write-on-wafer technology. The 32 nm lithography-of-choice, Double- Patterning, 193 nm Liquid-Immersion Optical Technology (DPT) can possibly be extended to 22 nm with restrictive design patterns and rules, and it serves as a reference for the future candidates.
EUV lithography operates with reflective mirror optics and 4x multi-layer reflective masks. Due to problems with a sufficiently powerful and economical
13 nm source, and with the quality and lifetime of the masks, the introduction of
EUV into production has been shifted beyond 2012.
MEB direct-write lithography has demonstrated 16 nm half-pitch capability, and it is attractive, because it does not require product-specific masks, and electron optics as well as electron metrology inherently offer the highest resolution. However, an MEB system must operate with more than 10,000 beams in parallel to achieve a throughput of ten 300 mm wafers per hour. A data volume of Peta-Bytes per wafer is required at Gigabit rates per second per beam. MEB direct-write would shift the development bottleneck away from mask technology and infrastructure to data processing, which may be easier.
The R&D expenses for EUV have exceeded those for MEB by two orders of magnitude so far. As difficult as a cost model per wafer layer may be, it shows basically that, as compared with DPT immersion optical lithography, EUV would be more expensive by up to a factor of 3, and MEB could cost less than present optical nanolithography. From an operational point-of-view, power is a real concern. For a throughput of 100 wafers per hour, the optical system dissipates 200 kW, the MEB system is estimated at 170–370 kW, while the EUV system could dissipate between 2 and 16 MW.
This outlook shows that the issues of minimum feature size, throughput and cost of nanolithography require an optimisation at the system level from product definition through restrictive design and topology rules to wafer-layer rules in order to control the cost of manufacturing and to achieve the best product.
Optical patterning through masks onto photosensitive resists proceeded from minimum features of 20 µm in 1970 to 22 nm in 2010. The end of optical lithography was initially predicted for 1985 at feature sizes of 1 µm, and hundreds of millions of dollars were spent in the 1980s to prepare X-ray proximity printing as the technology of the future, which has not happened. The ultimate limit of optical lithography was moved in the 1990s to 100 nm, initiating next-generation- lithography (NGL) projects to develop lithography technology for sub-100-nm features and towards 10 nm. This chapter is focused on two of these NGL tech- nologies: EUV (extreme-ultraviolet lithography), and MEB (multiple-electron- beam lithography). These are compared with the ultimate optical lithography, also capable of delineating features of similar size: DPT (double-patterning, liquid-immersion lithography).


Burn J. Lin