EUV’s Most Difficult Challenge

ASML and its partners had to overcome many challenges in order to make EUV lithography a reality.
For instance, in a previous video I talked about the EUV light source and its double-shot technique. But while challenging, that had not been considered one of EUV's dealbreaking issues.
A greater struggle was how to achieve a zero defect rate for the EUV photomask, or reticle. I will use the two terms interchangeably here.
The EUV reticle contains the chip design. Any defects on the reticle larger than a certain size will show up on the printed wafers themselves. So in order for it to work, it must be truly perfect.
In this video, we look at how ASML managed to overcome this ultra-critical aspect of the technology.
Links:
- The Asianometry Newsletter: asianometry.com
- Patreon: / asianometry
- The Podcast: anchor.fm/asianometry
- Twitter: / asianometry

[@herobrineharry7698]

11:22 “Graphene, a carbon-based nanomaterial that can do everything except leave the laboratory.” Nature and Science should put that up on their paper submitting system

[@me0101001000]

I've been making that joke for the last 10 years. It makes me laugh, but it also hurts my soul.

[@mmmmm49513]

It’s not quite true any more though. Graphene LiPo batteries are commercially produced. I’ve used several for high powered hybrid drones (they have a great recharge and discharge rate). I ordered them from normal RC drone battery manufacturers so it’s not a speciality thing either.

[@jrherita]

@@mmmmm49513 also graphene based USB power banks with extremely high output are available

[@danharold3087]

Was going to comment but others have made my point. 20 years is not too long to get something from discovery to first use, Most of the easy stuff has been done in the past. If you want something that is long in the tooth look at hydrogen cars. Nearly no advances and people keep championing it.

[@cerebralm]

I was astounded at his fearless savagery with that steaming-hot take.

[@claysweetser4106]

Mirror, mirror in the machine, which of you is perfect and clean?

[@maartenherrmann]

The Zeiss one ;)

[@trolleyking4002]

@@maartenherrmann but all the mirrors in the machine are from zeiss. ಠ_ಠ

[@elizabethwinsor5140]

Pre Zeiss ly !

[@vlodimyir]

Mirror in the Twinscan, defect free The fabroom's locked, just you and me

[@poilkjmnb3814]

1st time hearing 'ma fan' mentioned in an actual technical context

[@caiocc12]

"Graphene, a material that can do everything except leave the laboratory" Oof

[@danharold3087]

Graphene was discovered less than 20 years ago and researchers are making progress. We also see "graphene is finally leaving the lab". Neither of these are strictly true. Give it more time as it is worth the wait.

[@brodriguez11000]

Imagine they had to get fusion working first before proceeding to better semiconductors.

[@cv990a4]

@@brodriguez11000 Yeah, fusion is the gold standard of big, expensive, always-just-over-the-horizon technologies.

[@simontemplar404]

I used to work in a semiconductor FAB. The technology now is considerably more advanced than rocket science and has been for decades. But as pointed out here it has become a trudge from one generation to the next involving huge teams of scientists and engineers. If you want to be involved in some real breakthrough that changes the world and still use the most advanced technology possible I would go into the bio sciences. The last century was the one of the semiconductor chip and the computer, this century will be about modified biology.

[@simply6162]

Why tho ? What is emerging in bio science

[@Dan-hx6ni]

@@simply6162 He probably means, that in bio sciences the breakthroughs don't yet require gargantuan teams of engineers and scientists, and a relatively small group of bio scientists could still make a serious contribution.

[@J_X999]

Is it true working in a fab is stressful, low pay and simply not worth it?

[@saraofnorthwales]

I think the true scientific breakthroughs are in understanding Nature in the overview, rather than the microscopic. This leads to Wisdom rather than Knowledge. When we become a Wisdom Society we will have Transcended the Knowledge Economy. Then we may not need nanotubes at all. We might not even think about them. Or microchips, either. And many other things we think we need. Food for thought.

[@everry3357]

@@J_X999 depending on what's your job in the team is

[@i-aloofrawley2562]

Could you do a video on the filtration systems for the clean rooms, with info on the bunny suits (fabric materials, laundering protocols?) and the De-ionized pure water filtration, including the infrastructure within the Fab buildings, and the prerequisites to enter into clean rooms, etc... Some extra info on the mobile filtration systems during the transport of the EUV machines and reticles, and maybe even how and where the reticles are stored once they reach their destination and are not in use

[@Redsson56]

ASML’s achievements have been remarkable. But they did not do it alone. Many companies contributed. This includes Billions of $ that some of their eventual customers such as Intel invested in ASML.

[@omegapm4699]

Love these educational journeys

[@synergy021]

Love these educational presentations that are so straight forward and all the visuals are relevant. Not some guy in the camera half the time just waving their hands around pointlessly while reading a teleprompter with occasional graphics in the background.

[@davidgunther8428]

It's awesome that the pelicle is made from CNTs. That feels like a good use of their properties.

[@punditgi]

Another excellent video as always. Side note: Still better to use micrometre than the obsolete micron. Glad to see you switched later in the video. Thanks! 😃

[@danharold3087]

What makes it obsolete vs trendy ?

[@shibasurfing]

@@danharold3087 “micron” used to be accepted for use but in 1967 it was revoked by the SI. I think it sticks around because in the US we spell the unit name “micrometer” which is the same as a measuring device.

[@danharold3087]

@@shibasurfing With all respect to the people on the commission. WHY?

[@shibasurfing]

@@danharold3087 consistency with the other units, for sure. but I think everyone still understands what it means.

[@danharold3087]

@@shibasurfing Makes sense. It is unfortunate it overloads the term especially in that both are used in machining.

[@wmopp9100]

10:10 for the math to make sense, it has to be 81% of the radiation that is reflected (loosing 10% per pass-through) otherwise we would loose 90% on the first pass-through and another 9% on the second (making the output 1% of the incoming radiation)

[@Derpy1969]

Lose. To be 81% weaker, it has to lose 43% coming in and again coming out. Otherwise it’s 81% of the original intensity. Story problems are hard. Just don’t ever say something is 10x smaller because that’s impossible. It’s 1/10 as big.

[@PetsoKamagaya]

Lasertec stock has gone through the roof since 2008. Dang it I missed it. Anyway, I can tell you that shipping the blanks and masks without a pellicle was a total ma fa. I was at Entegris when ASML was doing all that testing at SEMATECH and Selete. But, it was a lot of fun.

[@brianfong5711]

12:13 Mafun (cantonese) = slang term for troublesome, hard, difficult So the equivalent english slang = pain in the ass Note: I do not speak mandarin, only a rare old dialect of cantonese that no hong konger would understand. I can't read or write in Chinese.

[@AnonymaxUK]

But it's written in Traditional Chinese, so it's not Cantonese. It would be máfan (which is how it sounded when he said it too).

[@NicolasChanCSY]

As a Hong Konger who speaks Cantonese natively, I would not say 麻煩 is a slang term. I do agree with you English translation though 😂

[@NicolasChanCSY]

@@AnonymaxUK Cantonese are written in Chinese words, preferably traditional Chinese if you ask any Taiwanese or Hongkonger.

[@AnonymaxUK]

@@NicolasChanCSY I know plenty. I go by what's most commonly used in each place.

[@ntabile]

Singlish have it too.

[@TheZeeray]

i love your videos so much, i look forward to each upload

[@barnmaddo]

I think 10:10 is a typo. Should be a 10% reduction in power each time, for a combined 1-0.9*0.9 = 19% reduction in power, or 81% efficiency.

[@nocomment3294]

watching this i am reminded how fragile our advanced soceities are. How many people have the working knowledge to build and use the machines our technology is based on. A large disruption might be enought to throw us back decades.

[@andersjjensen]

Stellar work as always Jon!

[@anonymous.youtuber]

The perseverance in itself is extremely fascinating. Let alone the technology of course. What a beautiful story. ❤️

[@Indrid__Cold]

"There's a lot of stuff in there I don't even understand. I'm really just a systems management guy working with people that are far more brilliant. I say build me a laser this. Design me a molecular analyzer that. They do, and I just stick'm together" Seth Brundle: The Fly.

[@EarthB00]

KLA's Teron 6xx tools are worth a mention here.

[@DrBunhead95]

Have you changed anything with audio recording for this video? It seems like some background noise got into your voice recodings, like a PC fan noise of some sorts. Besides that, great great work and very informative video as always, glad to have someone so passionate about such topics!

[@Asianometry]

It has been raining a lot lately and that's probably the sound of rain hitting the roof. Sorry about that.

[@johnclark8359]

Great video thanks a lot

[@craigcollings5568]

Can you layer on optical rectification layer over the top of deep defects?

[@insu_na]

There's a German saying: "Das Schwierige erledigen wir sofort, für das Unmögliche brauchen wir etwas länger." => "The difficult thing's we'll do immediately, for the impossible we need slightly longer" Edit: There's also a variant: "Unmögliches erledigen wir sofort, Wunder brauchen etwas länger." => "The impossible things we do immediately, miracles take a little longer."

[@nixietubes]

I guess billie holiday heard that saying

[@admiralspyro9722]

Never heard that in Austria tho

[@SaintFluffySnow]

Dutch ASML's EUV Lithography machine mirrors are Germany's Zeiss's proprietary mirror technology

[@andrewfrancis3591]

That is so. If you were to give Zeiss lenses to China would they then be able to do 7nm?

[@brodriguez11000]

So China would have to invade multiple countries to make their semiconductor industry a success.

[@andrewfrancis3591]

@@brodriguez11000 I wonder which country they would have to invade, to loose their sketchiness?

[@admiralspyro9722]

Nah, it s a symbiosis.

[@SaintFluffySnow]

Andrew Francis👈not "transparent lenses", but rather "reflective mirrors"

[@egalanos]

Any rough idea on how many times a pellicle is typically replaced over the lifetime of a EUV mask?

[@Knowbody42]

42

[@TSAlpha2933]

I wonder if they can use any sort of flat/bias subtraction* at any step in order to mitigate existing defects (similarly to how astrophotography does)

[@Matt33318]

Love your channel, great content again.👍🙂

[@stephanhart9941]

Now that's a MONOPOLY. ASML is like Tech God

[@covert0overt_810]

asml is more of a conglomerate that sources out super special processes. (ie ziess)

[@ADHD55]

It's a spin off from a American company, majority owned by US institutions

[@gregparrott]

A monopoly uses marketing, and often illegal tactics to exclude competitors. But if a company is one of just one with the technical prowess to create a product, that's not really a monopoly. The competitors just need to up their game.

[@ervie60]

@@ADHD55 LOL yes a spin off..from Philips Gloeilampen fabrieken N.V. Now I am curious about your sources, can you give them please?

[@ADHD55]

@@ervie60 it collaborated with the department of energy in 90s got funding there and Intel, it brought a SVG a US lithography company then started supplying Intel, its largest shareholders are Americans and relies heavily on US Tech, ASML is a great company pushing the boundaries in lithography, China cannot create a company like this

[@thesteambreaker9449]

Wouldnt it be possible to correct some degree of these bumb defects by changing temp of small parts of the structur surrounding the defect and use contraction and expansion to straighten the area ? Honestly no idea if this would be even possible on these small scales but still interested if someone got input on that

[@tombouie]

Thks & real star-trek stuff.

[@AmericanMinutemen]

I was just going to post a comment regarding your "graphene" quip, but the last posting already mentioned it.

[@pizzablender]

The square of 90% absorption is not 81% absorption. It is more that 90% transmissivity twice is 81% transmissivity. Also, "actinic" light is mostly light that is shortish in wavelength - starting with quartz bulbs and aquarium lighting.

[@asmael666]

Could you, maybe in a future video, address the topic of how the wafer is positioned (or the light beam directed) as perfectly as it is required? I mean, double, triple, quadruple patterning of a single layer, 81 layers, and the wafer isn't exposed all at once but the mask only contains one copy of the chip layout (did i get this right?) - so how is the positioning done so perfectly with single/sub nanometer accuracy over and over again while stepping through the rows and columns of the partially finished wafer?

[@MisterTwister222]

i understood some of these words

[@ominollo]

Nice 👍

[@speedstor]

I don't know why but this reminds me of the rick and morty bit, morty mindblowers of Rick making a truly leveled ground.

[@piotrd.4850]

Well, I think graphene hype should have ended after MIT supposedly proving, that most of amazing properties are lost in macro-scale.

[@jerrywelch870]

Can you do a Cisco video?

[@ebx100]

You have a gift of making the mundane fascinating.

[@choilive]

Maybe it was never mundane in the first place :)

[@MostlyPennyCat]

That is a "princess and the pea" defect!

[@leponpon6935]

12:15 Super "May Fan" 😂

[@MultiMates7]

How do you design a chip ? like what are the reasoning for specific designs?

[@Megabean]

That's like asking how do you build a space shuttle

[@Gilberto90]

Chips/transistors are essentially collections of logic gates and certain combinations of logic gates are better suited/needed in order to perform various mathematical operations. Some kinds of applications will use certain operations more than others so often the hardware of the chip is tailored towards those applications. As a starting point I'd look at the differences between the kinds of calculations that a CPU performs versus those typically performed by a GPU. In many cases the CPU can perform the same calculations as the GPU but due to its less optimised architecture it can't complete those operations as fast as a dedicated GPU. I'm vastly simplifying things here but it's something to start with if you're interested in the subject.

[@Joker-no1fz]

you cut potatoes really thin and fry them then add salt. goes good with sour cream french onion dip.

[@end2end302]

@@Joker-no1fz heard that's your mothers fav dish

[@danharold3087]

Chips are the way to miniaturize electronics. The first step is understand what electronics are need for your design. Next we design an electronic circuit that does what we need. A computerized schematic is created for the circuit. He is thinking transistors and gates but only uses symbolic representations. The circuit design is most likely tested in a simulator that utilizes the electronic schematic. When satisfied a chip designer creates a chip layout that is the circuit. For the first time he is creating real transistors and gates Maybe more simulation. The chip design is printed on to a wafer. Chip samples go back to the lab where they are tested. If it works your done. If not figure out what is wrong and try again. These are broad strokes and certainly more accurate information can be found elsewhere.

[@ricardokowalski1579]

13:45 transparency I see what you did there.

[@law-abiding-criminal]

I wish ASML would be as transparent as their pellicle.

[@andytroo]

they are - a bit over 10% of the important stuff removed by 2 passes through the processing layer :D

[@hugoboyce9648]

13:32 Ah the good old days when Elon Musk was someone looked up to and not an internet farce.

[@danharold3087]

Elon has been labeled as a farce in the past. There is a standing pack of wolves in the wings waiting to pounce. What does that software geek know about making cars? Rockets Neural link Boering company He already has twitter making money. Yes he did it by firing people but it was the right move. Lots of companies are firing and should he replace these people they will not be mooning about how it used to be.

[@NielMalan]

Is there work being done to design circuits with redundancy? I imagine that it might be cheaper to produce a chip with double the circuits, which can be tested and selected during testing, than to produce a defect-free mirror and mask. In other words, if chips can be made more tolerant of defects, savings in mask production costs and lower reject rates could yield chips with a better price/performance ratio.

[@angryakita3870]

I think the issue might be that increasing the surface area of the design leads to more defects than just throwing out a couple parts of the wafer.

[@Dan-hx6ni]

I believe, that's how some of the low-end CPUs are made. High-end CPUs with defects in cores or memory have them disabled and sold at a smaller price

[@lonyo5377]

That already happens with lots of chips. When Intel or Nvidia it AMD sell cut down processors, they disable defective parts and sell the rest. On some GPUs at times the top part shipped with disabled units at the start and then an updated one got released later with full spec. No need to design redundancy when you can just sell the "bad" ones as cut down chips. This has been done for decades

[@brodriguez11000]

@@lonyo5377 The significance of chip-lets.

[@kajurn791]

That's what they call Binning. It's been done since forever, but yes. Also chiplets, like AMD's been doing to great success.

[@Gameboygenius]

"The amazing thing to me was just how much transparency they gave..." especially at EUV wavelengths.

[@danharold3087]

Industrial spies already know everything here. My guess.

[@rumpelschtutz]

Were you voice acting for Birdperson in Rick and Morty?

[@inesfi66166]

What is megafab?

[@WahlanSahlan1982]

12:12 Everything about semiconductor manufacturing is super ma fan 麻烦

[@leechjim8023]

Socket set ain't gonna help. You'll need a hammer and chisel!😂😂😂

[@BetamaxFlippy]

Cheers from Italy

[@quinncampbell9255]

Wow I'm way wrong with this video title lol. I thought this was about Paradox EU5 the next game of eu4. They made hoi4, ck3, vic3, next is eu5.

[@z3d12]

19%

[@Czeckie]

EUV masks are mirrors? how does that work, what video have I missed?

[@Clark-Mills]

Coffee!

[@defeatSpace]

oh I'd be furious over an absorber defect

[@alexdhall]

13:39...meanwhile at Twitter....it's not late....it's impossible....

[@AG-jj3lx]

Impressive feats if science and engineering.

[@0MoTheG]

It was nice to live in the late age of progress and abundance. I am willing to bet that we have now entered the age of algorithmic exploitation. Physics is technologically exhausted. New technologies are too expensive to scale.

[@repatch43]

"Graphene, which can do everything except leave the laboratory"!! ROTFLMAO!!!

[@tuxlovesyou]

It must be November.

[@jon_j__]

@10:10 I think you got the maths a bit wrong here. If 90% of the light is absorbed on each of the two passes through the pellicle then only 1% is left afterwards. (IE: You have to do (1-(10%*10*))=99% absorbed, not (90%*90%)=81% absorbed.) Your assertion also doesn't make sense based on a simple sense-check: in the first step, 90% of the light is absorbed (ie. 10% is transmitted), so it's a bit weird when adding another identical step magics 9% of the light into existence (81% absorption = 19% transmitted). I don't have enough background knowledge to fill in the gaps and guess what you were trying to say here, so, unfortunately, I basically have to disregard this section as unreliable. 😕 Otherwise, very interesting!

[@tolentarpay5464]

It's truly astonishing clever monkeys like us can do all this stuff, BUT it's also terrifying; think about it - OUR ENTIRE "MODERN" WAY OF LIFE IS TOTALLY DEPENDENT ON THIS LVL OF COMPLEXITY, & IT'S JUST GOING TO BECOME EVEN MORE SO! It's like the whole 21st century is hooked on some superdrug. There's so Many failure points in the process, what happens to us if that process stops working? I was born in 1969, so I remember a world without PC's, but how many people can say that today? For years now I've had a growing premonition that we've all had it too good for too long & very soon now we're going to be stuck with a bill we can't pay...& what happens then?

[@snjtgbjmmk]

funny lol im stealing that joke

[@lawrencebishton9071]

BT ???

[@alexlo7708]

I wonder why they are preoccupied by EUV approach. Indulgent in this way make them confuse in complex probelms on each step in process. While some others initiate other way such leap to intense invest in X-rays lithography. X-rays lithography reduces or eleminates many steps problem occuring in EUV such mirror lens energy loss and precision control of the focus. While left problem to solve only photo resist chemical that abides to the energy level while movement precision are not stricted required because the light is so thin than error coverage. I think American knew it well , but kept it secret and made people in industry waste their effort and lost its way and time. This will keep American in going ahead of others.

[@tommy2cents492]

Yawn... Even here there are conspiracy theories... Your reply clearly indicates little understanding of the whole process and its requirements.

[@alexlo7708]

@@tommy2cents492 Requirement in process are comply by your approach. If you chosse the complex way, yes, you face many problems. US has gone to growth diamond wafer and testing gamma photon light. Thats why they discover as they semi telling out that expose diamond to gamma rays ,will occuring electric current.

[@testboga5991]

“Graphene, a carbon-based nanomaterial that can do everything except leave the laboratory.” 😀

[@gustavludwig9719]

super 麻烦🤣

[@FritzSchober]

Videos like this show the span in human intelligence from people with galaxy brains designing these technical wonders to the idiots that still believe the earth is flat.

[@Phil-D83]

These mainland chinese will have "fun" with this part if they are not using euv machines for 7nm. Multi patterning is a disaster.

[@lawrencebishton9071]

UN ??

[@AndrewMellor-darkphoton]

hi

[@TheOtherSteel]

03:47 -- "...more sucky..." You can do better than that.

[@xco1113]

Challenge is $. Soon to be changed.

[@mateuszzimon8216]

Why? TSMC have monopol in 5 and sub5 nm. Intel still fight to go under 7nm, GloFo try 7nm but petrodolars are cut. OnSemi aquire fabs from GloFo.

[@godfreypoon5148]

Chips are too hard, just go back to vacuum tubes.

[@MichaelT_123]

... or vacuum of celestial space.😉

[@rich_in_paradise]

Reminiscent of Edison's search for a viable lamp filament material.

[@wildreams]

Super mafan? That pronunciation sounds weird tho.

[@truefan1367]

NANOTUBES SON!

[@kimchi_taco]

EUV seems the last lithography technique humans can manage the complexity. The next generation lithography will be developed by AI powered by semiconductors made by EUV.

[@karehaqt]

The only problem with your videos is that they are not long enough, 15-20 minutes is to short :(

[@mudhen24]

Why do you apologize for adding Elon musk at the end lmao. Very strange thing to say.

[@artysanmobile]

Perhaps because Elon Musk famously is a fascist and racist? Those are generally not favored characteristics regardless the technological advancements a person might bring. See ‘Henry Ford ‘if you wish to do a little research.

[@KarpucMotoring]

Why are you sorry for mentioning Elon Musk? Figured you'd be a fan considering your material

[@zorktxandnand3774]

Have you been living under a rock?

[@KarpucMotoring]

@@zorktxandnand3774 what the hell are you talking about?

[@zorktxandnand3774]

@@KarpucMotoring Lets say he has been saying and doing a lot of "not so smart" things lately, and it only seems to be getting worse. Among other things, the world richest man has just converted 44 billion dollars in to a big worthless joke at his own expense. He lost to someone who spend just $8.

[@KarpucMotoring]

@@zorktxandnand3774 oh boohoo so because his tweets hurt your feelings the fact that he is the frontrunner in the race to sustainable and renewable energy, is providing internet to the entire world (including Ukrainian soldiers) and oh yeah possibly saving the human race by colonizing Mars in case of a disaster doesn't mean much huh? Frankly I find fun entertaining and love that the richest man isn't some bore and isn't afraid to call ppl out. If he wants to buy a company for fun that's his business, he's just going to make it better anyway.

[@zorktxandnand3774]

Why do you say his tweets hurt my feelings? I didn't say anything remotely like that. In fact I am watching the shitshow of his latest purchase, popcorn in hand. But Musk is a bit of an ashole. If you think he does anything for the benefit of others you really have not been paying attention.

[@bassmechanic237]

Wtf are you apologizing for Elon for?

[@zizzyballuba4373]

take this video down because it helps the chinese to make their own

[@thefbat5847]

This is public knowledge lol

[@liutaurasbalnius]

I hope you get a potato for free one day, and i hope its a nice potato