Mask Repair Enters The Spotlight

By Mark LaPedus
For years, the biggest challenges in photomask manufacturing have revolved around the slow write times for electron-beam tools and soaring mask inspection costs. Now, photomask repair, a sometimes forgotten technology in the mask shop, is in the spotlight and turning into the clash of the titans.

Mask repair involves the process of finding defects on a photomask and repairing them on the fly. At advanced nodes, photomask shops primarily use two types of complementary mask repair technologies: electron-beam and nanomachining.

SII NanoTechnology (SIINT), a photomask repair tool supplier, is proposing a new and third mask repair option based on gas field ion source (GFIS) technology. SIINT also contends that e-beam repair, the workhorse technology in today’s mask shops, could soon hit the wall.

“The minimum repairable dimension of the current state-of-the-art e-beam repair systems is between 20nm and 30nm,” said Fumio Aramaki, manager of the engineering department within the Beam Technology Business Division at SIINT, a subsidiary of Japan’s Seiko Instruments. “GFIS supports only 9nm, which is less than half of that of e-beam.”

Carl Zeiss, the sole supplier of e-beam mask repair tools, dismissed the claims. “We cannot confirm Seiko’s statement,” said Markus Waiblinger, senior product manager for Germany’s Zeiss. “Our current system performance is significantly smaller than the claimed limit of 20nm.”

In any case, the challenge in repairing photomasks is growing at each node, said Thomas Faure, senior technical staff member for photomask development at IBM’s Mask House. “The tolerance of repair has to be really tight,” he said.

Moreover, SIINT, Zeiss, and the champion of nanomachining-based repair, Rave, also must contend with the new and complex masks using extreme ultraviolet (EUV) lithography.

Big changes ahead
There is a sea of change taking place in mask repair and the overall photomask business. In total, the photomask market is expected to grow by 2% in 2012, according to Sematech. And by SEMI’s forecast, it is expected to reach $3.35 billion in 2013.

Banqiu Wu, principal member of the technical staff and chief technology officer for the Mask and TSV Etch Division at Applied Materials, said the photomask equipment business can be divided into three parts: pattern generation, transfer and quality assurance. Pattern generation includes e-beam and resists. Transfer involves photomask etch. Quality includes inspection, cleaning, repair and pellicles.

“Every one of these has its own challenges,” Wu said. “For e-beam writing, the real issue is throughput. The write times are pretty long.”

For quality control, there are also several challenges. Photomask inspection remains the most expensive tool in a mask shop. “Repair is also challenging,” he said. “For the optical mask, we have a tighter spec on cleaning.”

In the mask supply chain, the number of captive and merchant photomask makers has dwindled over the years, causing a shakeout in the equipment base. In mask repair, for example, there were many vendors at one time. Today, there are three main suppliers: SIINT, Rave and Zeiss.

But the mask repair market remains in flux. In May, Seiko Instruments announced plans to sell SIINT to Hitachi High-Technologies. In July, the completion of the deal was postponed, as various entities are still reviewing the transaction.

And up until recently, mask shops relied on three basic mask repair technologies: laser, focused ion beam (FIB), and nanomachining. Laser-based repair tools hit the wall around the 100nm node. FIB scaled to about 50nm to 80nm.

In 2005, Zeiss acquired Nawotec, a supplier of e-beam mask repair tools. Then, e-beam mask repair became a mainstream technology. Rave has been able to scale nanomachining, which is in use today. And now, GFIS is entering the picture.

Each repair technology has its advantages and disadvantages. GFIS is still an unknown. Nanomachining is a solution for certain types of repairs. E-beam repair provides fine resolutions, but the technology is slower. The other knock on e-beam is the so-called charging effect. Secondary electrons with enough energy cannot escape from the deep regions in a structure, thereby causing a back scattering effect.

“Electron-beam is easier to handle and the optics are well developed. The problem is that the scattered signal is huge. I also don’t think we should underestimate the challenge that exists with a gas phase ion repair tool,” said Franklin Kalk, executive vice president and chief technology officer at Toppan Photomasks.

Repair also becomes more challenging amid the shift to more complex masks. There are repair challenges for traditional binary, optical proximity correction (OPC) and phase-shift photomasks. The newer opaque-MoSi-on-glass (OMOG) and EUV masks also present some problems.

OMOG masks provide finer resolutions. Using MoSi as the absorber layer, OMOG appears opaque at 193nm wavelengths. OMOG is also thin enough to reduce the electromagnetic field effects that plague immersion lithography. But in e-beam mask repair, OMOG is much more difficult to etch, according to some experts.

EUV masks are reflective structures, which include a multilayer Bragg mirror onto which an absorber pattern is defined. “I am sure that you already accept that defects on finished masks are unacceptable, because they are replicated so many times on wafer,” said Emily Gallagher, senior technical staff member within IBM’s photomask operations. “The problem is compounded for EUV masks, because the blanks will have multilayer phase defects into the foreseeable future. Non-actinic, including 193nm optical or e-beam, inspection cannot locate phase defects with the absorber in place. Consequently, repair of multilayer phase defects will be needed after the mask is completed and in the vicinity of absorber patterns.”

Repair to the rescue
For repairs, Zeiss says it’s ready to meet the challenges. E-beam tools can repair a mask using either an etch or deposition process. “We have certified repair processes for more than 30 mask types and many of them are for OMOG. So I see no problem here,” said Zeiss’ Waiblinger. “You use different chemicals for etching and deposition. You have back scattered electrons, which are required for imaging.”

Regarding the scaling limitations for e-beam, Waiblinger said: “I see no fundamental limits to use e-beam mask repair in production down to 10nm. I would say the weakness (for e-beam) is a consequence of the high resolution. Different than laser repair, you cannot increase the spot of your beam. This means that for large defects, the repair time increases.”

E-beam repair is also ready for EUV. “E-beam and ion-beam technologies have advantages and disadvantages,” he said. “For EUV mask repair, e-beam has the clear advantage that the multilayer or capping layer is not damaged during imaging or repair. With ions you must take great care not to reduce the reflectivity or damage the capping layer.”

Not surprisingly, SIINT has a different viewpoint. SIINT is developing a rival GFIS tool with a voltage of 15- to 30-kV. In GFIS, helium and hydrogen gases penetrate the shallow region of a structure and are scattered in the deep region. Gallium gases and electrons are scattered in the shallow region.

GFIS is still in R&D and may not be ready for production until late 2013. “GFIS technology is a promising candidate for the solution to repair a MoSi mask named ‘A6L2.’ The material selectivity between ‘A6L2’ and quartz under GFIS etching is 6:1,” SIINT’s Aramaki said.

The other mask repair technology is nanomachining, which is considered a complementary technology to e-beam and GFIS. Using atomic force microscopy (AFM) as a means to chisel and repair mask defects, nanomachining can be used at 20nm and beyond. The advantage with nanomachining is that the technology is “material independent,” said Mike Archuletta, director of marketing for Rave.

Rave and IBM recently described a breakthrough method for EUV mask repair using nanomachining. “Previously, EUV phase repairs were reported using only absorber compensation. This changes the amplitude of the defect, but does not work for the phase component,” said IBM’s Gallagher. “What is significant about using nanomachining is that multilayer material can be removed precisely to alter the phase locally. Using both amplitude compensation, by removing absorber pattern, and phase compensation, by removing multilayer, it is possible to achieve a complete repair of a phase bump.”