Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/index.php?threads/canon-ships-first-next-gen-chipmaking-equipment-to-u-s-consortium.21115/
        )

    [addOns] => Array
        (
            [DL6/MLTP] => 13
            [Hampel/TimeZoneDebug] => 1000070
            [SV/ChangePostDate] => 2010200
            [SemiWiki/Newsletter] => 1000010
            [SemiWiki/WPMenu] => 1000010
            [SemiWiki/XPressExtend] => 1000010
            [ThemeHouse/XLink] => 1000970
            [ThemeHouse/XPress] => 1010570
            [XF] => 2021370
            [XFI] => 1050270
        )

    [wordpress] => /var/www/html
)

Canon ships first next-gen chipmaking equipment to U.S. consortium

Barnsley

Well-known member
https://asia.nikkei.com/Business/Te...t-gen-chipmaking-equipment-to-U.S.-consortium

TOKYO -- Canon said Thursday it is delivering the first unit of its next-generation lithography equipment to a U.S. group developing advanced semiconductors, introducing a new production technology that could cut costs as well as electricity usage.

The nanoimprint lithography machine is going to the Texas Institute for Electronics, a consortium supported by the University of Texas at Austin that includes Intel and other chip companies as well as public-sector and academic organizations. The equipment will be used by chipmakers for research and development.

Semiconductor manufacturing usually involves forming circuits using photolithography, in which patterns are projected onto a resin-coated wafer through exposure to intense light. Instead, nanoimprint lithography equipment stamps patterns on a mold into the resin.

Canon has been developing the technology since 2014. Sales of the equipment, which was created in partnership with Kioxia and Dai Nippon Printing, began last October.

While photolithography gear is equipped with arrays of lenses or mirrors, Canon's nanoimprint equipment has a simpler design that consumes about a tenth as much power. It can also form complex three-dimensional circuit patterns with a single stamp, and can handle the extremely fine circuits used in state-of-the-art logic chips.

Kazunori Iwamoto, Canon's deputy chief executive for optical products, said the company aims to sell around 10 to 20 units annually within three to five years.

Some hurdles remain to bringing nanoimprint lithography into broader use, including the need for more advanced technology to keep out fine dust particles that can cause defects. Partnerships with other companies to develop manufacturing materials that work with the technology will also be key for Canon.

Something or nothing?
 
In a way, the pattern definition difficulty is moved upstream. Can the electron beams manage it? It was reported multi patterning (at least double patterning) was needed even there.
 
In a way, the pattern definition difficulty is moved upstream. Can the electron beams manage it? It was reported multi patterning (at least double patterning) was needed even there.

Am thinking it wont need a photomask , so would make me a bit sad if the method take off and proves economically viable
 
https://asia.nikkei.com/Business/Te...t-gen-chipmaking-equipment-to-U.S.-consortium

TOKYO -- Canon said Thursday it is delivering the first unit of its next-generation lithography equipment to a U.S. group developing advanced semiconductors, introducing a new production technology that could cut costs as well as electricity usage.

The nanoimprint lithography machine is going to the Texas Institute for Electronics, a consortium supported by the University of Texas at Austin that includes Intel and other chip companies as well as public-sector and academic organizations. The equipment will be used by chipmakers for research and development.

Semiconductor manufacturing usually involves forming circuits using photolithography, in which patterns are projected onto a resin-coated wafer through exposure to intense light. Instead, nanoimprint lithography equipment stamps patterns on a mold into the resin.

Canon has been developing the technology since 2014. Sales of the equipment, which was created in partnership with Kioxia and Dai Nippon Printing, began last October.

While photolithography gear is equipped with arrays of lenses or mirrors, Canon's nanoimprint equipment has a simpler design that consumes about a tenth as much power. It can also form complex three-dimensional circuit patterns with a single stamp, and can handle the extremely fine circuits used in state-of-the-art logic chips.

Kazunori Iwamoto, Canon's deputy chief executive for optical products, said the company aims to sell around 10 to 20 units annually within three to five years.

Some hurdles remain to bringing nanoimprint lithography into broader use, including the need for more advanced technology to keep out fine dust particles that can cause defects. Partnerships with other companies to develop manufacturing materials that work with the technology will also be key for Canon.

Something or nothing?
who is this US consortium???? The New york group? NY CREATES’ Albany NanoTechComplex.

I have not heard of any company planning on this technology
 
Am thinking it wont need a photomask , so would make me a bit sad if the method take off and proves economically viable

You can still litho the imprint mold, which will be way faster than mask writer for the number of molds needed, given their wear rate.

Otherwise you will need a gazillion mask writers to make imprint molds which only last for 1 lot.
 
You can still litho the imprint mold, which will be way faster than mask writer for the number of molds needed, given their wear rate.

Otherwise you will need a gazillion mask writers to make imprint molds which only last for 1 lot.

Will need to do some study.

Hope the management is on it too
 
Nanoimprint lithography (NIL) is a simple mechanical lithography technique involving a stamp, or a template, pressed against a deformable imprint resist layer deposited on a substrate to make the contours of the template.

Sounds like how to make a copy photomask in this method , can that really be so accurate?

Another method!

Nanoimprint can be performed in a way similar to the step-and-repeat optical lithography. The imprint field (die) is typically much smaller than the full-wafer nanoimprint field. The die is repeatedly imprinted to the substrate with certain step size. This scheme is good for nanoimprint mold creation.

We used to use this step and repeat method for mask making 25 yrs ago.
 
who is this US consortium???? The New york group? NY CREATES’ Albany NanoTechComplex.

I have not heard of any company planning on this technology
Not in NY but at the old Sematech site in Austin - owned by UT Austin.
Group is called Texas Institute for Electronics (TIE): https://www.txie.org/
Focus is on Advanced Packaging and Heterogenous Integration technologies - definitely not targeted at front end device making.
It's part of DARPA's aim to move some advanced packaging capability back to the US; not for the HVM logic/memory makers.
UT (Grant Wilson) developed this technology, and Canon bought the company (Molecular Imprints) some years back.
 
Not in NY but at the old Sematech site in Austin - owned by UT Austin.
Group is called Texas Institute for Electronics (TIE): https://www.txie.org/
Focus is on Advanced Packaging and Heterogenous Integration technologies - definitely not targeted at front end device making.
It's part of DARPA's aim to move some advanced packaging capability back to the US; not for the HVM logic/memory makers.
UT (Grant Wilson) developed this technology, and Canon bought the company (Molecular Imprints) some years back.
thanks! That explains why no one plans to use Canon technology.
 
NIL makes most sense for a fab starting up in India, or even a company like Rapidus. The idea being they don't have a fleet of legacy ArF/immersion systems to keep supporting with multipatterning. In fact, there is multipatterning even for 5nm EUV and below. But the manufacture of the mold or template needs to be thought out carefully. That could be the breaker. Maybe Nikon's Digital Scanner would be necessary: https://www.spiedigitallibrary.org/...-manufacturing/10.1117/1.JMM.22.4.041402.full.
 
NIL makes most sense for a fab starting up in India, or even a company like Rapidus.

Powerchip is in trailing node business. They only recently setup a 40nm fab. They are the only relatively big fab making PMICs on 300mm, as far as I know.

But the manufacture of the mold or template needs to be thought out carefully. That could be the breaker. Maybe Nikon's Digital Scanner would be necessary: https://www.spiedigitallibrary.org/...-manufacturing/10.1117/1.JMM.22.4.041402.full.

There are different methods with regards to original pattern "multiplication". Almost the most basic, and seemingly the one to be commercialised by Canon is metal pattern->mold->resist pattern. I read from decade old papers that the mold is only good for 1 lot. And the pattern is only good for <100 molds. Things, of course, could've changed in 10 years.

To increase the life of the expensive pattern, there are processes which make the pattern inverted, and make copy patterns from it, which then used to make molds.
 
Back
Top