Maskless E-Beam: Worth Another Look?

It’s interesting to hear the
same person who made such a push for immersion now making a similar
case for electron-beam direct-write (EBDW) lithography. Given his
success with immersion, not to mention his considerable experience
in the industry, it’s hard not to listen to what TSMC’s
Burn Lin now has to say about maskless technologies, especially
when he says himself that he has been an “optical
bigot” most of his life.

And he makes some awfully good
arguments: the ability to cluster tools to give a chipmaker just
the amount of throughput it needs or can afford; a much smaller
footprint than EUV; and the clincher: cost. There’s the
potential for a multiple e-beam tool to have about the same cost as
an immersion scanner, Lin contends, both in terms of cost of
ownership (CoO) and wall power (bunking initial conceptions that
e-beam would be a power-hungry solution). To make his cost
argument, Lin presented a chart comparing various lithography
possibilities.

To compare these costs, Lin detailed
some assumptions: Water immersion tools will cost 40 million euros
($62.58M), with ¥800 million ($7.64M) for the track; and EUV
tools will cost 50 million euros ($78.23M), plus ¥700 million
($6.69M) for a 100 wph track, and ¥300 million ($2.87M) for a
20 wph track. The final column shows what a 20 wph e-beam system
would have to cost in order to compete with each technology. The
$3.75M cost is unrealistic, so does not stand a chance against
single-exposure water immersion. But competition with
double-patterning water immersion and 100 wph EUV is feasible, and
with 20 wph EUV is easy.

Maskless lithography has been talked
about for years, in a variety of forms. Its leading downfall has
been throughput, but research has continued relatively quietly
nonetheless because it offers some real resolution advantages,
particularly for contact layers. In fact, one of the most critical
factors affecting maskless adoption at this point, according to
Lin, is that it has been around too long, and people have already
formed their conceptions. “One has to change their concepts
and look at the potential of a multiple e-beam system,” he
said. This is not the old e-beam, he added later. “This is a
new era for e-beam.”

Besides Lin’s presentation,
several other promising presentations were made at the Litho Forum
updating attendees on maskless e-beam progress, including talks
from Mapper Lithography (Delft, Netherlands), Multibeam Systems
(Santa Clara, Calif.), IMS Nanofabrication (Vienna, Austria) and
CEA-Leti (Grenoble, France). (By the way, Mapper
shows a great movie showcasing its technology.)

During the panel discussion that
wrapped up the Litho Forum talks, Freescale’s Will Conley
added, “I very much support Burn and the maskless team. I
think in the low-power device community, this kind of a solution is
quite nice. And we, along with similar companies to Freescale, we
have a lot of products that are very low runners, and we may build
50 wafers per mask set, or maybe 100 wafers per mask set, or
something like that. And it’s expensive for us to do it, but
it’s part of our business model, and we do make money. And if
this kind of technology exists for us, it’s possible it could
even expand that kind of a business, and provide more revenue for
us in that area. So there’s a lot of companies I think that
can benefit from it.”

However, when Conley posed a
question — “Would those companies participate in
funding it?” — he got no immediate reply, and the topic
moved on. But he did get a response, perhaps, in the form of Litho
Forum survey results. In the survey conducted at the close of the
two-day conference, one open-ended question attendees were asked
was which alternative technologies should be pursued more
aggressively to achieve ITRS goals. The clear winner was maskless,
with runners up including nanoimprint, interference, 3-D
integration and directed self-assembly.

Couched another way, when attendees
were asked which technologies warrant reconsideration, based on
information presented at the Litho Forum, 21% of the respondents
said that maskless was one to reconsider. The only technology
ranking higher was EUV, with 29%.

You can view the
full results of the Litho Forum surveys in a 20-minute webcast
produced by Semiconductor International and presented by Litho
Forum program chair Bernie Roman.

One company making news recently for
its use of e-beam direct write (EBDW) technology is e-Shuttle Inc.
(Kawasaki, Japan), a joint venture started up by Fujitsu Ltd. and
Advantest Corp. to provide prototyping services for leading-edge
ICs. E-Shuttle is
now delivering 65 nm logic ICs manufactured with the maskless
technology, lauded as the first full-scale application of
EBDW.

Presenting last week at EIPBN
(International Conference on Electron, Ion and Photon Beam
Technology and Nanofabrication) in Portland, Ore., Shinji Sugatani,
e-Shuttle’s general manager, showed the available markets for
maskless lithography, provided a given throughput. Although at 0.5
wph it’s difficult to define a market beyond R&D
applications, a throughput of 0.5-3 wph enables LSI prototyping
with a market size for EBDW tools of ~¥10 billion ($94.32M),
and throughput of 3-30 wph enables small-volume or mid-size
production of ASICs and/or MOS logic, for example, with a tool
market size of ¥30-200 billion ($282.94M-$1.89B). If, with a
multi-column system, throughput could reach 30-100 wph, Sugatani
could foresee maskless technology unseating EUV at mass production
levels, reaching a tool market size of ¥1 trillion ($9.44B).
That is not the level of production e-Shuttle aims to achieve,
however, since mass production is handled by parent company
Fujitsu.

So, suffice it to say that
throughput is a major consideration in improving the chances for
maskless e-beam, and it will require innovation, Sugatani said. As
Chris Mack quipped during the Q&A after Lin’s
presentation, maskless is the only technology that makes EUV
throughput seem high. According to Lin, though, clustering multiple
e-beam tools together looks like a good bet.