Physics of Computer Chips - Computerphile
Вставка
- Опубліковано 18 сер 2016
- You can't beat physics. Why the chip manufacturers are heading for a wall. We asked nano-scientist Phil Moriarty Professor of Physics in the School of Physics and Astronomy, University of Nottingham
EXTRA BITS: • EXTRA BITS: More Silic...
Teaching Physics in Ethiopia: • Teaching Physics in Et...
Is it the end for Moore's Law: • Is it the End for Moor...
Password Cracking: • Password Cracking - Co...
Gamer's Paradise: • Gamer's Paradise - Com...
/ computerphile
/ computer_phile
This video was filmed and edited by Sean Riley.
Computer Science at the University of Nottingham: bit.ly/nottscomputer
Computerphile is a sister project to Brady Haran's Numberphile. More at www.bradyharan.com
"You don't understand it, you get used to it" sooo relatable
Yep
It's not sad, it's just the way it is. No one can fully understand the implications of quantum mechanics (namely entanglement), which is why there are so many interpretations of the same observation. Look up the double-slit experiment and tell us that doesn't blow your mind.
females xD
Isn't it like that for everything we "know"? Classical physics is just as strange, we are just used to it.
Not really, quantum physics is very un-intuitive. The fact that an electron doesn't exist in any particular place or time (and is just a probability wave of being somewhere) until it is observed is something spooky that no one really understands. There's nothing like that in classical physics. Or the fact that two entangled protons can be a universe (or any length) apart and still affect each-other (instantaneously, breaking the laws of relativity).
"You don't understand it, you just get used to it"
Probably one of the best physics quotes ever
Yeah that is actually very profound , it keeps you going.
It’s sounds like the food that’s served in US prisons and jails!
John von Neumann was the first one to say it. Give him the credits
How do the individual atoms feel about being manipulated?
#METoo
electronized
Uncertain. Of course.
The electrons are TRIGGERED
Try to learn about transistor...
Phil, aren't you too large to call yourself a nanoscientist?
Dad, I told you not to make these jokes in UA-cam comments!
its okay, he identifies as a nanoscientist!
cissized pig
He was referring to little phil... downstairs.
If he's a nanoscientist, I hope I never meet even a microscientist, let alone a full scientist!
My favorite connection from physics to electronics, is the fact that quantum tunnelling effects are at the heart of how flash memory and EPROMS work.
This is one of my all time favorite videos on UA-cam! I have watched this video about 7 times now and I just absolutely love how well Mr. Moriarty explains the semi-transistor manufacturing.
I don't always watch computerphile (over my head) but you could put Phil Moriarty in a video about paint drying and I would watch it........oh wait.
It seemed like the video ended while he was still explaining something.
Prof. Moriarty actually never stops talking. The best you can do is turn off the camera just as he changes topics.
He was about to unify classical and quantum physics
He was about to show his pee pee
I want more of this guy. He could start his own channel and just talk and I think he'd have thousands of subscribers;)
He has his own channel. Called Phillip Moriarty.
He does have his own channel! Moriarty2112, or you could follow Sixty Symbols where he has many videos about physics!
You should watch Sixty Symbols, he features in literally several dozen videos there.
+Phi6er Aww really? I liked this guy. :/
seconded
Why don't they just download more RAM into the electron beam to make it go faster?
I'll create a GUI interface using Visual Basic to see if I can track down an IP address for the download.
CSI?
Usual Hollywood hacker nonsense, that particular excerpt is from CSI:NY.
Justin Bell I thought so
"we need to hack faster !!!"... 3 people typing at once... on the same keyboard
This was a really informative video. Some illustrations/animations to visualize what he said would have made the video even better, although I understand that they take quite some time to make.
Check out en.wikipedia.org/wiki/Photolithography
i agree. His experience as a professor really shines through.
I'm not sure if there's a definite correlation, but all the physics professors I had in college were the best teachers I ever had. Along with their ability to explain things they were passionate/excited about their field
I love how excited he gets to answer each question and you can tell it’s genuine too
More of this guy please.
One of my favorite Computerphile videos. Great explanation of concepts I've always wanted to understand. Thank you!
Being a chemist, just having listened to a physicist, talking about mechanics, for the purpose of computing, I just realized that the electron couldn't care less about how it's manipulated and by whom.
Prof Moriarty is one of my favorite guests on any of the “phile” videos. Awesome guy and very good at breaking things down to a level I can understand.
I usually just understand 25% of these talks but just love this channel and will keep coiming back to it again and again. Thank you for this!
Very good speaker, very well explained, and engaging. This really helped me understand exactly what the subject was about. I'd love to have this guy as a teacher.
That was a brilliantly clear and energetic overview of modern chipmaking. Professor Moriarty explained how a Silicon transistor works, but didn't label it as such. The Silicon substrate is formed into the transistors by adding impurities (doping).
Wow, that was was barely within what I could follow all the way through. The few times I started to get lost he stopped and explained it a bit more. Very well done, and I even picked up a few new things.
This is brilliant! I would be really glad if you made more videos on this topic
these videos are like the best thing in my life sometimes. thanks for continuing to make them. :)
This material is way over my head, but this video was fascinating. Thanks for putting it together.
I really admire Phil Moriarty's ability to talk around naming concepts like quantum tunneling, and keep his talk on track even with tangential questions.
Really like how you explain things, hope you do more videos in the immediate future!
Great video; thanks Computerphile and Dr. Moriarty :)
Moriarty is one of my favorite science communicators
hi there camera man!
Hello!
Was the camera man sitting on a basketball?
not enough giggles for that to be the case.
And this why movies tend not to use real mirrors!
movies use real mirrors, they just don't have the camera face on with the mirror
this was incredible to watch. the passion and conviction he showed was amazing
This is a great effing interview.. awesome enthusiasm and passion.
I learned SO much from this. Thank you.
I also love his passion. It excited me to learn this.
This is probably my favorite video on the channel.
2019 update: 5nm in the works, 7nm in production (AMD Ryzen 3000 series, e.g.)
So if 14nm = 50 atoms, 7nm = 25 atoms, 5nm = 17 atoms. Getting there
3nm in lab
yet they are too inneficient
Thats not the case there. 5nm isn’t exactly 5nm. It does not represent a geometrical shape on transistor. It’s just the technology’s name. You can call it “marketing”. The real limitation for geometrical shape of transistor is 7nm. Nothing more than that.
For silicon ^^
This man is brilliant and captivating. Would definitely enjoy more videos featuring him.
please, more on the physics of computer hardware! there's been so many amazing inventions and discoveries through the years in the semiconductor industry so we can use computers as we know it...
Watching this in 2019, they are now manufacturing 7nm microprocessors, how things move on.
intel 14nm have 8nm wide fins in a finfet transistor, I think they are making features bit smaller than 7nm in the absolute sense.
And on what he call Extreme UV (EUV)
@SuperTanner But how can they go smaller than atoms
And tomorrow morning a 5nm machine’s getting delivered to my home.
Let’s come back in a year, see where we’re at
we are in 5nm stage now
"If you find that confusing.. Good" - soo funny and soo true
I can never get enough of Professor Moriarty. Such a fantastic and interesting person!
You should have seen him in Sherlock Holmes
God, I love how professor Moriarty explains stuff !
Nice to see you again Dr Phil!
this is great info, since I come from a physical science background this shows the application of what I learned, wish they had a class on the physics of computers
Physics AND computer science, all with Professor Moriarty. Great combo, I really enjoyed this episode. Thanks guys!
The physics and technology in making the 13.5nm light that is talked about here is some of the most advanced applied physics in industial RnD going on im the world today. Primarily this work is done close by to you, in The Netherlands. I would love to see you or Brady make a video on it. great job on videos. From another Irish physicist.
I love how you can see how passionate this man is about what he does.
This interview elevated my understanding of how we're able to manipulate atoms. Thank you.
This is my favorite video in a while.
I brought up the fact that Chemistry and Physics and ultimately everything that follows is a seamless whole just divided into digestible parts to a chemistry instructor once, and he almost flipped his lid. It was almost the same reaction from the physics department, yet they worked together constantly on things, though the chemists tend to be more reserved and the physicists tended to let their reservations go a bit, especially on things that went "boom". Nice to hear the Professor say nearly the same thing, about the relationship.
great stuff, hoping for more stuff about chip manufacturing and such :)
Love this guy and his energy. Great talk as well.
To elaborate on the "Layers" question, yes, it's very much done in layers. In fact, even the first layer wasn't fully described here. After the exposed (or unexposed, depending on the process) areas of the polymer are washed away, another layer of some material is applied. Then the places where the polymer remained are washed away, leaving the new material only in the gaps. The material can be dopants for the underlying silicon, metal layers to connect components, insulating layers to separate things, etc. It can even be exposed to etchants, rather than a new material, to remove whatever layer is showing.
I really liked this sort of unplanned interview
Really cool to see semiconductor fab processing explained here!
It's very interesting to dive into the physics and chemistry of electronic computing, it's not a subject I've explored much as a computer scientist.
I was wondering, what is the limiting factor in hard drive size, both physical and storage related? if you could do a video explaining, that would be appreciated.
Videos like this immunize me from the Kruger-Dunning Effect.
same here my friend. i hqd to look up the effect so im even more ignorant ;-)
putting you in your place is a simpler way of saying it
If you think it did that, you should be worried.
sugarfrosted Okay... it was just a booster.
Or even the Dunning-Kruger Effect.
Great video, I'd love to see more like this
This guy is awesome! It's very obvious he's gushing with knowledge and enthusiasm of his subject.
Watched extra bits, still want more.
I don't know why but he reminds me of Roy from The IT Crowd
Just of course more intelligent than, "Hello IT, have you tried turning it off and on again?"
Similar accent
He's also got the Roy-esque quality of talking about intricate stuff in a non-jargon way
Was about to comment this, brilliant.
...and gestures.
I was wondering who he reminded me of!
Welcome back professor. I had wondered about your absence.
I love how excited he got for the silicon question.
great video, we need more low level hardware videos like this
This was a really good video. Lots of detail.
Could you do a video about error checking/handling in chips (CPU's). I read a few years ago that now CPU's are becoming smaller, there are more errors in the calculations. I would love to know how manufacturers get around the errors.
I'd love to hear Prof Moriarty talk about spintronics and photonics if you have him on again.
This is awesome! More videos like this please!
This guy is magic, and clearly loves his stuff. Excellent.
Just a silly question: doesn't diffraction screw up with the lithographic process considering that light has to go through such tiny apertures?
Yes, which is why you use light of very small wavelengths.
The wavelengths they're using in mass production today are not that small. I believe 193nm is still standard, even though they're making features as little as ~14nm in size (actually many parts of a "14nm" process are not 14nm, but it's all pretty much to scale compared to older process nodes like 22nm).
The next step is (supposed to be) EUV, where they do drop to very short wavelengths and high energies, as discussed in the video. They are having a lot of issues getting that to work for mass production though.
The diffraction of light really is a limitinig factor when reducing the size of features. One of the ways semiconductor manufacturers get around this is phase shift masking, which Prof. Moriarty explained as two masks just slightly offset from each other.
i didnt get how shifting the two templates helps
+sewer renegade There are actually several ways of masking phase shift masks. I think Professor Moriarty is portraying the general concept of setting up having the edges of the photomask phase shift the light passing by so that when the light reaches the photoresist on the wafer, the edges of the patterns are being enhanced by constructive and destructive interference of the light waves, thus making features smaller than the wavelength of light possible.
It is so amazing to see Prof Moriarty talk about computer science. *beaming* :D
Great video... More of Professor Moriarty!
I thought that the wave interference of the photons (as in, interference you see with the double split experiment) would be come a problem. But instead you could create a mask that actually utilizes this phenomenon to create interference patterns that match the target pattern on your silicon sheet.
You're a gifted explainer of these topics.
I could listen to this guy for hours
I freaking love this video, man.
Three years ago they were talking about 14 nm, today we're talking about 5 nm; so this video becomes history in less time than it would take to study Electronics.
Unfortunately the 5nm and 7nm are really brand names than actual sizes of features - they are usually much more refined 10nm process which are capable of higher densities by eliminating issues with previous 10nm and 14nm processes.
Essentially we are right up at the limits of manipulatable sizes when it comes to computing - most every improvement now comes from extremely complex and well designed architecture - cache for instance which is what amd rides on for their superior processing of late with 5000 and coming 6000 series chips
@@aravindpallippara1577 You need to search before replying, and address what was written.
Plants are being constructed in Arizona, Tainan, etc. with 2 nm coming in 2025 - call it creative naming or fudging on the numbers - each new plant builds a smaller process.
People don't invest and they don't spend over a hundred billion to convince you, you are not convinced.
Instead the money is spent to place billions of transistors in the space previously occupied by one transistor, decades ago.
They really are getting smaller.
Amazing video. Learned a lot.
8:43 ish - How do they make the masks with features that small?
The energy of his explanations.... HE should be duplicated in the kindergarten, school, high school... NOBODY cannot be energized and curious the way he is talk and explain things! His energy in language would make me a damn poet!!
I loved this! So informative :)
You said the wavelength is a limit but the nobel prize last year was for the invention of microscopes which overcome this wavelength barrier. I think they used the light emitted by proteins and blocked the light emitted by neighbouring proteins so that the resolution was down to one protein. There might be a way to use that for making smaller chips.
Big man working on small stuff, respect
Great video, thanks!
Nice to see Phil again. :3
My fiancé worked for Intel, he was sent around to various clean rooms and such to work on the computers that were running the electron scanning microscopes that they were using for debugging chips.
The whole process is pretty cool to me, as I just stopped learning the abstractions of the CPU at the logic gates, and VHDL design.
A curious wonder, what kind of feature size would a hobbyist be able to achieve? I mean, there's that guy who built a macro-computer by using full chips for his transistors, and I know most of us are better off using FPGAs anyways. But say someone wanted to get into etching their own silicon, what do you think would be the range of quality that they could get to?
Please do an updated one of these about the current physics of the newest chips
Fascinating stuff!
Amazing video! I have a question about Professor Moriarty's explanation of how the semiconductor industry is able to create such precise patterns on transistors. When the two offset masks are placed over the silicon wafer and light is shown, how is it that the light is able to deterministically etch a pattern? Why would it not behave like a wave/particle in the double slit experiment and defract into a probabilistic wave pattern on the wafer?
I have been searching for this video. Thanks! UA-cam! For putting it into Recommendation.
I can listen to this man for hours ...
Will you please carry on with this it was brilliant !!,
so.. a moving electron creates a magnetic wave/wake in it's passing that expands radially. can you take advantage of these wakes and layer the silicone wafers and their circuitry in a way that layer two could be providing power to layers 1 and 3 as well of course with less amps/volts but still enough to open or close gates... perhaps even have currents passing through the wafers laterally connecting circuits on all layers at differing voltages and the gates be three way and open differently due to the differing currents being direct or eddie currents... etc.
would these methods improve or be inovative to the industry or are they already in use?
i love this guy he explains so well.get him on the show more often plz
best video i'm seen on computer chips
"Quantum Mindset" #bandname
+MaxPower ^ rofl
But damn if you try to find the location and time of any given concert in particular.
+scabbynacker "Are you thinking with quanta yet?" #tagline
We missed you!
Hm, I heard that there are directions that light moves in with its waves, and some crystals can eliminate some of those directions. Would it be possible to use properties like that to carve chips?
This makes me so excited/happy!! :D
I'll follow you anywhere Phil
So is it possible to make computer what uses light insted of copperwires or is this a thing allredy?
This guy won't shut up. I love it!!!!
Congratulations! A very successful comback.