You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
In a highly secured lab in the Netherlands, ASML spent a decade developing a $400 million machine that’s transforming how microchips are made. High NA is the latest generation of EUV, the only machines in the world that can etch nanoscopic blueprints on advanced chips for giants like Intel, TSMC and Samsung. The size of a double-decker bus, only five of these machines have been shipped so far. It’s unclear how Trump’s tariffs will impact the complex global supply chain for the lithography machines. But without them, no advanced chips can be made by the likes of Nvidia, Apple and AMD. No filming of High NA has ever been allowed until CNBC went to the Netherlands for this exclusive first look.
Technology's getting harder and harder. Now we might need 2x masks per layer + stitching just to manufacture single-large chip. In good old days, people simply waited for new light sources, but now I'm not really sure what comes next after High-NA.
Technology's getting harder and harder. Now we might need 2x masks per layer + stitching just to manufacture single-large chip. In good old days, people simply waited for new light sources, but now I'm not really sure what comes next after High-NA.
X-Ray? Gamma Ray? I remember IBM investigating X-Ray lithography decades ago. There's probably too much at stake economically for the industry to hit a brick wall and just call it quits.
Technology's getting harder and harder. Now we might need 2x masks per layer + stitching just to manufacture single-large chip. In good old days, people simply waited for new light sources, but now I'm not really sure what comes next after High-NA.
X-Ray? Gamma Ray? I remember IBM investigating X-Ray lithography decades ago. There's probably too much at stake economically for the industry to hit a brick wall and just call it quits.
A large part of it is also the NA made a big jump this time (67%), which is different from prior generations (10-20%). Depth of focus dropped a lot. Resist is supposed to be thinner than it, so it will be too thin (<30 nm).
The wavelength jump can also be said to be too drastic, becoming ionizing for the first time. That leads to electron blur and stochastics.
Need to get out of the wavelength spiral. Maybe it's time for bottom-up like DSA.
Who is CNBC and ASML trying to convince with this video? The actual customers have run their financials and won’t jump to this until they get their moneys worth out of standard EUV. Makes no sense for tsmc to not multipattern on their existing machines considering how much they’ve already paid for them.
I don't know that it has been proven in volumes anywhere close to production yet. Of course, a limiting factor is that it only fits regular grids of features, so maybe some DRAM layers may be immediately relevant for now. Actually, I had the impression Intel had dabbled with it, not sure where there are with it now.
Who is CNBC and ASML trying to convince with this video? The actual customers have run their financials and won’t jump to this until they get their moneys worth out of standard EUV. Makes no sense for tsmc to not multipattern on their existing machines considering how much they’ve already paid for them.
It’s to keep governments and policy makers happy. Besides, if you were a CNBC reporter, wouldn’t you jump at the chance to get to see the high NA tool in person?
Most likely no one from our business. But the "ordinary man" who first heard of semiconductors when he could not buy his <insert any device> because of missing chips might find this fascinating.
Despite all attention semiconductors got in recent times, I fear we are still a niche business when it comes down to overall understanding of whatever is going on in our bubble.
