by M. David Levenson, editor-in-chief, Microlithography World
July 16, 2008 – For several years photomask lifetimes in fabs have been reduced by photo-induced defects (PIDs), or haze, which grow on the mask surfaces, even under the protective pellicles. Residual surface contamination and airborne molecular impurities evidently combine under DUV illumination to form crystalline deposits on the reticle surface. Once seed crystals are formed, they can continue to grow even in storage, turning good reticles bad after relatively few wafer exposures. As the CDs have fallen below 90nm, smaller and smaller PIDs can cause yield loss and that has meant shorter and shorter lifetimes for more advanced reticles.
At SEMICON West, reticle repair company Rave LLC (Delray Beach, FL) has announced an innovative technology to remove 100% of the haze in the wafer fab. The Rhazer (“Rave Haze Removal”) system employs laser cleaning to vaporize the haze on a 6-in. reticle in eight hours, without removing the pellicle or employing fluids of any kind.
Fig. 1: The number of wafer exposure before reticle clean as a function of time. (Source: Rave)
“Haze forms when the contamination is at the parts-per-billion level. After the Rhazer process, the contamination is reduced to parts-per-trillion,” explained company president/CEO Barry Hopkins in a presentation on Tuesday. Haze is expected to reform eventually, he noted, but this product allows wafer fabs a way to manage it.
In the past, a reticle found to have printable haze defects would be shipped to a mask house to have its pellicle removed and undergo wet cleaning. At best, that has a three-day turnaround (and a cost of $6000), which means that a fab needs a second mask to stay in production (at a cost of $100,000), Hopkins said. Wet cleaning also can potentially damage the absorber materials and may add residue to the surface that helps haze form again, even more quickly.
The Rhazer process is entirely different — it’s in-house, completely dry with no waste, and no impact on absorber material. It takes eight hours for both mask surfaces and costs about $1500 per cleaning.”
The tool as described is fairly small, with a 1.25m × 2.34m footprint, and requires only 7kW of electrical power. It accepts both standard reticle SMIF pods and will cost about $4M, according to Hopkins. In a typical fab, he estimated, the tool would pay for itself in 6-12 months just by reducing the number of mask cleans. Hopkins reported that a prototype system is available now at Rave’s Delray Beach facility for user validation, with the first beta-test shipments planned in 1Q09. In one Rave demonstration, all the haze on a 6-in. reticle with >350 PIDs/mm2 was removed, leaving just two (hard) defects on the entire plate.
Brian Grenon, president of Grenon Consulting and world expert on mask haze, firmly believes that this is “the first real solution” to the problem of mask haze. “Both ammonium oxalate and sulfate material is removed with 100% efficiency, even on some really awful test masks” which he personally supplied to Rave. AIMS inspection shows the mask transmission and CDs are restored (see Fig. 2).
Fig. 2: KLA-Tencor Starlight and Zeiss AIMS images of a mask with haze defects before and after Rhazer cleaning. (Source: Rave)
Hopkins advocated using the Rhazer to pre-clean masks coming into the fab to increase the number of good exposures, to re-clean masks during litho runs to maintain process window, and to post-clean masks headed for storage to discourage haze formation in the reticle stocker (see Fig. 3).
Fig. 3: Wafer fab/reticle management system with Rhazer. (Source: Rave)
While the Rhazer is not intended to prevent haze formation, it seems the most practical way to manage that phenomenon to date. Other attempts have included employing very dry environments to limit residue adhesion and slow haze formation, and completely changing mask process chemistry, and neither of those schemes offers a way to deal with haze that has already appeared on a mask needed urgently in production. The quick in-house turnaround offered by the Rhazer may make it the most practical approach for most existing fabs. Once haze ceases to be the limiting factor for reticle lifetime, some unexpected new problem will appear for the photomask industry…but that’s a story for another day. — M.D.L.