How an ASML Lithography Machine Moves a Wafer

Holding ~1 nm alignment tolerance over meter-scale distances is hard enough; holding that tolerance dynamically while the waver and mask are moving at a relative speed of several meters per sec is insane. In the time it takes the light to get from the mask to the wafer, the wafer will have moved something like 5 nm. So somewhere in those billion+ lines of code controlling the system there is a correction for the finite speed of light.

I work in a cleanroom assembling these machines for ASML. We hold tolerances in microns and this video gives a good explanation of just why we do that lol

You realize how precise this operation is when he says "... refractive index of the air, which changed as the stage moved around, it threw off things by as much as ONE NANOmeter, which was UNACCEPTABLE..." that is really crazy!

It always blows my mind how complex these machines are, how many technologies and innovations that has devoloped over decades to reach this point. Probably the most complex thing humans have ever constructed.

The hydraulic backstory of early ASML is really fascinating. Little historical lessons like the chip manufactures rejecting ASML equipment make for powerful stories to understand and explain decisions you'll need to make in the future. Great video!

The technology behind this is (to say at least) insane. My dad worked at RIZ semiconductors in Yugoslavia, they made transistors and ICs. He regularly talks about his days working there and how interesting and good it was. He was very sad when company shut down before the war and he never recovered from it. I can imagine how happy he would be if they had something like this in factory.

Fun fact: I grew up in the city of Eindhoven and the Philips Natlab lays in the cities district called “Strijp” when Philips moved out of Eindhoven with all their manufacturing facilities this part of town became like a ghost town. The municipality of Eindhoven invested heavily in the district for the past 10 years or so and it now houses many restaurants, cafes and foodhalls. The old Philips factories have been converted into apartments, offices spaces, start-ups and the district really became a new hotspot for the city.

They do all this so I can watch my NETFLIX. Thanks for your efforts ASML.

I worked with a former ASML mechanical engineer for a while. He was of the opinion that the ASML stage is a masterpiece that has held up for decades and generations of machines, and he obsessed over specifications I have never heard of. Fantastic stuff

So fun to see the ASML videos. It is interesting to see what has and has not come passed the walls of our company. As someone that works in their learning department these videos are great inspiration. Keep it up!

Fifteen kilograms. For an easy point of reference, that's the same weight as twenty-six Cornish piebald hissing marmots.

Fantastic video containing a lot of details I wasn't really aware of. Thanks for making this!

I love your litho videos bro I’ve seen everyone at least 5 times. Your videos are the only ones I have found that go into great detail about every process and very easy to understand. This semiconductor field just fascinating and boggles my mind

I worked on these tools for 15-16 years at various sites. From. 2500s, 5500s all the way to NXTs. Left ASML in 2018 to live a more peaceful life in the Philippines. Working under pressure to deadlines, to customers watching your every move, I have not regretted my decision. I am out of the rat race. As they say, money isn’t everything

YES! more videos on how the ASML machines actually function please!

The thing that really makes my head 🤯is that the wafer and reticle are in motion during the exposure. They move opposite directions with a tiny slit between them such that only a narrow line of the image is projected onto the wafer at a time. The relative motion of the two needs to match exactly so the reticle image lines up with the correct part of the wafer. So that crazy 1nm overlay spec that you were probably thinking is a "line up and hold it still within 1nm" is actually "drive the wafer and reticle in opposite directions while accelerating at 20Gs and do not allow the speed of either the reticle or wafer to deviate by more than 1nm from it's predicted x or y location at any point along the path."

Correction on your throughput statement at 10:06. 125 wafers per hour is about two wafers per minute, not one wafer every two minutes. This isn't the only time you make this error in this video.

Always wondered how they could realign the wafer with insane precision...

Just this last year we got rid of two new-to-us pas 5500 ASML steppers. They were used as initial process development demonstrators and have been replaced by much newer double-sided alignment capable machines. Good luck to the next folks who find those 5500s as new-to-them.

Great job documenting ASML technical history, old Philips guy here. mounted the lasers on sharks, less error when you reverse that, lol