Startup rolls complementary litho
EE Times, Mark LaPedus
SAN JOSE, Calif. – At the SPIE Advanced Lithography conference here, maskless startup Multibeam Corp. will outline more details about its ongoing efforts to commercialize its so-called Complementary E-Beam Lithography (CEBL) technology in the market.
Multibeam (Santa Clara, Calif.) will describe the latest developments of its CEBL tool, a multi-column, maskless lithography system designed for patterning the most critical layers in a design–contact holes, vias and line cutting–at the 16-nm node and beyond. Throughput is said to be five wafers an hour–more than twice the speed of today’s single-beam e-beam tools.
David Lam, venture capitalist and chairman of Multibeam, said CEBL will not replace today’s optical lithography. But rather CEBL ”complements’’ or works in tandem with today’s 193-nm immersion tools, Lam told EE Times. Multibeam is looking for partnerships to enable CEBL, he added. (Lam is also noted for being the founder of Lam Research Corp. He is no longer associated with Lam Research.)
At SPIE, Multibeam, along with Tela Innovations (Los Gatos, Calif.), will also present a paper, entitled ”E-beam Litho to Complement Optical Lithography for 1D GDR.’’ 1-D GDR, or one-dimensional gridded design rule, is based on Tela’s 1-D layout optimization technology. The combination of CEBL and 1-D gridded design rule technology could provide a powerful one-two punch to enable next-generation IC designs at 22-nm and beyond, he said.
Today’s optical lithography could hit the wall, prompting the need for another next-generation lithography (NGL) solution. Direct-write or maskless, EUV and nanoimprint are among the next possible waves in lithography.
For years, the industry has used direct-write e-beam technology to pattern lines directly on a wafer. Today’s direct-write tools make use of a single-beam technology. E-beams are also used in mainstream photomask production.
Direct-write promises to give chip makers some relief from the soaring costs of photomasks. Direct-write prints tiny features on a wafer, but it is slow in terms of throughput, thereby relegating the technology to niche applications like compound semiconductors. For the most part, direct-write is too slow and expensive for mainstream IC production.
In the last decade, a number of companies began working on next-generation e-beam technology, dubbed maskless lithography or ML2. IMS, KLA-Tencor, Mapper, Multibeam and others are separately working on tools that utilize multiple beams in a single machine. In theory, ML2 is supposed to overcome the throughput issues associated with single-beam direct-write.
Formed in 2001, Multibeam emerged from stealth mode in 2005. At about that time, the company obtained approximately $2 million in funding from the National Institute of Standards and Technology (NIST). The startup also disclosed that it was developing a maskless, 50-KeV tool that utilized an array of 10 electron beams.
At the time, Multibeam was among several vendors racing to develop a tool in NGL. In NGL, there are several technologies aimed to replace today’s optical lithography, such as extreme ultraviolet (EUV), maskless and nanoimprint.
Then, in 2009, Multibeam entered into a joint development program with Japan’s Tokyo Electron Ltd. (TEL) for an undisclosed project. But in more recent times, Multibeam has narrowed its focus.
Instead of replacing optical lithography, Multibeam’s new strategy is to work in tandem with 193-nm immersion-or perhaps another technology-in mainstream fabs. ”We are not an NGL,’’ Lam said. ‘’We will not replace optical in high-volume manufacturing anytime soon.’
To beam or not to beam
Multibeam, according to Lam, is looking to address only ”5 percent of the pattern density’’ in the challenging and critical layers in chips. Multibeam is somewhat taking a page from Intel Corp., which has been pushing a ”complementary’’ lithography strategy. The chip giant hopes to push 193-nm immersion to 11-nm. Then, at 11-nm, Intel is pushing its concept of a ”complementary” or mix-and-match strategy, in which 193-nm immersion could work hand-in-hand with EUV or maskless lithography to enable advanced chip designs.
It’s unclear which technology–EUV or maskless–will get the nod at Intel. Neither maskless or EUV have proven they are ready for prime time. EUV is still lagging in terms of power sources, resists, metrology and defect-free masks. Maskless is still underfunded and the tools are behind.
Still in the R&D phase, Multibeam believes it has the right formula. Instead of the original 50-KeV approach, Multibeam’s tool is a 10- to 15-KeV system. The company’s CEBL tool makes use of multiple-column arrays. A typical system could have 100 columns for 300-mm production. Each column has its own electron source or beam.
The company is in the testing phase for its fourth-generation column. Every column array is identical and they can work independently or in parallel. The range of current is 15- to 26- nanoAmps per column.
”E-beam litho systems on the market today are either single-column or a few columns, all derived from the scanning electron microscope (SEM). In contrast, Multibeam’s innovative approach eliminates the magnetic field: the e-beam column is all-electrostatic,’’ according to Multibeam.
Multibeam believes the technology is ripe to help propel designs based on 1-D GDRs. For years, chip makers have made use of two-dimensional layouts in IC designs. But the problem is that there are many challenges to scale 2-D layouts beyond 45-nm.
As a result, IBM, Intel, TSMC and others have proposed 1-D GDR schemes. Intel, for example, makes use of unidirectional, gridded layouts at 45-nm. In 1-D GDR at 45- and 32-nm, chip makers can make do with 193-nm immersion lithography.
That could change at 20-nm. A 193-nm immersion scanner can handle the traditional steps. To handle the line-cut scheme, chip makers are looking at EUV, maskless and even 193-nm immersion.
Multibeam believes that CEBL makes the most sense for line cut applications. ”Of the three, the most viable and cost-effective solution is (CEBL)-if it is used to pattern critical layers, in a mix-and-match mode’’ with 193-nm immersion, Lam said in a recent paper. ”This is the only path to high-volume manufacturing.’’