Optical lithography has provided the semiconductor industry with ever-finer features. Most recently, resolution-enhancement techniques for optical masks, especially proximity correction and phase shifters, allowed 193nm-wavelength optical lithography to meet the demands of the 65nm-half-pitch (hp) technology node in 2007.
For smaller features in the future, the International Technology Roadmap for Semiconductors predicts that 193nm water-immersion lithography—most likely combined with double patterning—will be used for 45nm-hp device production starting in 2010. But for the 32nm-hp node, to be in production in 2013, several methods are still candidates: 193nm water-immersion with double exposure and patterning, 193nm immersion lithography that uses high-index fluid (193i+), extreme ultraviolet lithography, nano-imprint lithography, and maskless lithography (ML2).
ML2 is of particular interest because of sharply increasing costs of masks for optical lithography. Currently, ML2 is needed most urgently for fast prototyping of novel devices and for low-volume chip production.1 It is predicted, however, that ML2 will have the lowest cost of ownership of all of the previously mentioned techniques at the 32nm-hp node if it can achieve a throughput of 15 wafers wafers per hour.2
IMS Nanofabrication, in Vienna, Austria, has developed a promising concept called projection maskless lithography, or PML2. This method uses a programmable aperture-plate system (APS) to generate thousands of micrometer-sized electron beams, which are guided through electrostatic- and magnetic-lens optics with 200× reduction. Thus, thousands of reduced-size electron beams are projected in parallel onto the resist-coated wafer.
The multiple beams are not steered, as in direct-write electron-beam lithography, but instead expose predefined spots while the wafer moves underneath them. The local dose can be adjusted in fine steps by controlling the number of exposures at every spot. The only moving mechanical part is the laser-interferometer-controlled wafer stage. Errors in wafer-stage movement can be corrected with no moving mechanical parts by electrostatic- and magnetic-multipole-field techniques, which can place the whole ensemble of electron beams with sub-nanometer precision.
The APS consists of two plates: an aperture plate, which forms the beams; and a blanking plate, which addressably deflects selected beams. As shown in Figure 1, all deflected beams are blocked at a stopping plate further down the electron-projection optics; only undeflected beams survive to the wafer surface. An aperture plate with 3.5μm openings produces beams at the wafer 200 times smaller, or 17.5nm. Proprietary gray-scale exposure techniques will meet even the demanding requirements of 22nm lithography.
Figure 1. (left) Principle of operation of projection mask-less lithography (PML2) and (right) a photo of the proof-of-concept system.
Within the 6th European framework program, IMS Nanofabrication is coordinating the RIMANA project (radical innovation maskless nanolithography).3 As part of this project, the team built the PML2 proof-of-concept system shown in Figure 1, using fixed-shape stencil templates instead of a programmable APS. The novel electron-optical column, consisting of electrostatic and magnetic lenses, successfully achieved 200× reduction and the predicted 22nm-hp resolution on 150mm Si wafers coated with 60nm-thick poly-methyl methacrylate (PMMA) resist and 50nm-thick hydrogen silsesquioxane (HSQ) resist, as shown in Figures 2 and 3, respectively.
Figure 2. The PML2 proof-of-concept system resolved lines with a half-pitc (hp) as small as 22nm, using poly-methyl-methacrylate (PMMA) resist.
Figure 3. The system printed lines in 50nm-thick hydrogen-silsesquioxane (HSQ) resis with a half-pitch as small as 16nm.
In the next step of the RIMANA project (Oct 2005–Sept 2008), a programmable APS, providing 2300 beams, will generate 32nm-hp demonstration patterns on resist-coated 150mm silicon wafers. The APS units are provided by the Fraunhofer Institute for Silicon Technology (ISIT), Itzehoe, Germany and the Institute for Microelectronics Stuttgart (ims-chips).4
Further development of PML2 is foreseen within the 7th European framework program as part of the integrating project MAGIC (maskless lithography for IC manufacturing), which started in Jan 2008.5 MAGIC targets include both a PML2 ‘pre-alpha’ tool and an alpha tool that can expose wafers up to 300mm in diameter.
Christof Klein, Elmar Platzgummer, Hans Loeschner, Gerhard Gross
Christof Klein is the PML2 project manager at IMS Nanofabrication. From 2004 to mid-2006, he was the Erwin Schroedinger Fellow at the National Center for Electron Microscopy/Lawrence Berkeley National Laboratory, in Berkeley, CA, where he fabricated and studied magnetic nanostructures. In 2003 he was post doctoral scientist at the Institute for General Physics, Vienna University of Technology, where he had previously received his PhD in physics.