by Franklin Kalk, Toppan Photomasks Inc.
March 3, 2010 – The SPIE advanced lithography symposia remind me of baseball spring training. The pervasive "collaboration speak" in lithography circles has transformed the various camps into ersatz teams, which suffer no shortage of illustrious players, heroes and villains alike. And, like baseball fans in February, it’s fun to see which players have burst onto the scene, retired, defected to another camp, signed for ridiculous money, or appealed to the commissioner for leniency for past transgressions.
193nm/ArF, like the Boston Red Sox, just keeps thriving, and its progeny immersion is beginning to blossom. Admittedly, source-mask optimization and double patterning aren’t production-worthy yet — but the optical proximity correction (OPC) and resist benches are deep, with years of successful production behind them. So, it’s just a matter of refining, not rebuilding.
Extreme ultraviolet lithography (EUV) is more like the defending champ New York Yankees: loaded with talent, colossal bankroll, favored to win, and every seat is a luxury box. It had a solid week at SPIE, with progress reported in scanner optics and computational lithography (yes, Virginia, EUV will need complithity). Those who have ordered pre-production scanners finally seem to understand that they should fund the mask infrastructure that will benefit only them for the next five or six years. Source performance is progressing deliberately — but expect appeals to a higher power, as shot noise is one transgression that the commish will not ignore.
Like the Philadelphia Phillies (2008 champs and 2009 runners-up), nanoimprint lithography (NIL) is cleverly managed, adding new pitches to its repertoire just before we wonder why we’re cheering. This year, it’s the adaptation of a conventional mask-inspection strategy to NIL template fabrication for improved cost and cycle time. It is common in mask manufacturing to first inspect one (reference) die of a multi-die reticle in die-to-database mode, and then to inspect the remaining dice against that reference die in die-to-die mode. Similarly, a slow but sensitive e-beam inspection can be used to inspect the NIL master template, and then photon-based inspection can be used to find the process defects (particles) common on replica templates.
E-beam direct write (EBDW) resembles the Cubs: in the mix, nearly breaking through on occasion, but often lacking one key component to succeed. Over the past couple of years, however, both multi-beam and character projection architectures have enjoyed revivals. Still, EBDW’s flexibility could be its greatest challenge — maybe we can afford one EBDW approach, but which one? Smart money is on the one best adapted to mask writers.
Finally, self-assembly, in a well-organized set of talks, came off as a two-sport star in the making. Will it only be used for hard disk media patterning, or could it become the next Anaheim Angels and succeed in Si? Now that would be a gift from heaven!
So which, if any, of the contenders will succeed ArF-i? For the next several years, with the Curse of the Bambino a distant memory, look for the BoSox to bat around and for occasional singles from the challengers.