How Microchips Work | Hacker News Reader

How Microchips Work

exclusivearchitecture.com

270 pointspbamotra36 comments

Silicon is the perfect material for semiconductor. It has a low band gap energy between the valence band and conduction band. A small amount of energy, electricity applied to it, can knock its outermost valence electrons off and it becomes conductive. Withholding the energy, its valence electrons fall back in place and it becomes non-conductive. As if by luck, silicon is plentiful and cheap.

This is missing one of the most important reasons though: Silicon Oxide.

Silicon Oxide is almost perfectly lattice matched to silicon, but completely insulating. Which means it's incredibly easy to grow features onto polished silicon wafers because the oxidation product of the material is exactly what you need in order to build up insulating features - i.e. MOSFET junctions, capacitors and conductive paths.

Silicon oxide grown on Si is actually amorphous, so it is not lattice matched.

But you are complety right, the oxidation properties of Si are really fortunate and ICs would have taken decades longer if it were not for that. SiO2 is really the unsung hero of the silicon age.

- SiO2 has a high bandgap and a very good insulator.

- It is quite inert to many chemical and gasses. (e.g. germanium oxide is soluble in water, which is a headache)

- It can easily be grown on stoiciometric form by oxidizing silicon and will form an abrupt interface to Si.

- The formation proceeds by diffusion of oxygen to the Si interface. This is in contrast to other metal oxides, where the metal will diffuse to the surface and create a nonstoiciometric mixture.

There is no other semiconductor that forms as good an oxide. Very few metals form insulating oxides on their surface, one notable exception is Aluminum.

Edit: The famous paper that describes the SiO2 formation kinetics was actually co-authored by Andy Grove, from intel CEO fame.

https://en.wikipedia.org/wiki/Deal%E2%80%93Grove_model

Huh, it's as if the universe was built to make computers out of.

One time I was driving around, and I thought to myself "Boy, it sure is lucky we had all the products for asphalt laying around, or it would have been difficult to build all these roads."

Then I realized that if the products for building roads weren't around, then we wouldn't have had those roads in the first place, and I wouldn't have been reflecting on how lucky we were to have all this stuff.

The universe is such that the computers that do exist seem perfectly matched to its properties. How could it not be so? Other possible universes might have completely different computers and the kind we have here would be unimaginable there.

If silicon didn't exist, our computers wouldn't seem perfectly matched. It would be a struggle to make them and keep them working.

Not at all.

Gallium Arsenide works fine. So does Indium Phosphide. etc.

Silicon was much simpler to start and then path dependence kicked in. So, we poured R&D money at silicon because it was "better" than everything else. Then, because we poured R&D money at silicon, it was "better" than everything else. Lather, rinse, repeat.

For silicon not to exist, the universe would have to be quite different. Maybe making life possible and in some planet a life form might eventually find out how to build computers with whatever chemistry they'd end up with. And the chemistry of such a universe would seem uncannily suited for such computers.

Point being, you can't delete an element from the universe and expect everything else to be the same. Silicon exists because of the physics in this universe. So do silicon based computers.

It's as if someone created one element that is perfectly suited to build microelectronics. Sure, there are other materials that improve on one property or the other, but there is not a single other element which balances properties as well as silicon.

Not even mentioned yet:

- Excellent mechanical properties of the single crystal (think MEMS, or wafers that don't break all the time)

- Piezoresistive properties can be used to measure strain (also quite unique due to silicon band structure)

- Optical properties perfectly suited to detect visible light (think detectors, image sensors). Good combination of band gap and carrier lifetime to build solar cells.

It took ages from the theoretical invention of the MOS transistor to us being able to grow practically good enough oxides (which aren't riddled with interface traps)...

More like a decade, unless you refer to the Lilienfeld devices.

Yes, I was thinking of Lilienfeld. What starting point are you taking?

Bell labs tried to build a FET before the bipolar transistor. It's not so clear which theory the Lilienfeld devices are based on.

Imagine how long it would have taken if silicon wasn't this ideal.

A fully optical computer would be better still.

https://en.wikipedia.org/wiki/Optical_computing

There is the flaw that they don’t build themselves, though.

Clay and glass are also partially made of silicon, right?

I’ve always found of fascinating that silicon was right there at the beginning of material science, and has stuck around since. Same for copper.

I don’t actually think the universe has intentions, but copper, silicon, and dogs do sometimes make me question that belief, it is just a little suspicious that our species would have such loyal friends.

It’s fun to watch the Anthropic principle develop organically in this sub-thread.

Watch the movie Alpha some time :)

Low band gap is actually detrimental for transistors. The more spare band gap, the higher temperature you can operate them at.

And it's relatively easy to make large pure silicon single crystal.

Just before the section on Moore's Law, it says this about silicon purity:

> Electronic grade silicon (EG-Si): 99.9999999 pure ('nine nines pure') Thats one impurity atom in every 10.000.000 silicon atoms.

I believe that should be 1.000.000.000 (10^9 atoms) to correspond to nine nines pure. Just as one impurity atom in every 100 (10^2) atoms would be 99% (two nines) pure.

Hi everyone! I am the author of the website exclusivearchitecture.com.

Thanks everyone for all the positive feedback - so rewarding to see! I have already made some corrections to a detail where I gave an incorrect number with the "nine-nines-purity" and it should be 1 impurity in 1.000.000.000 silicon atoms. Thanks gshubert17 for pointing that out.

I have noticed that my website is currently down with a timeout - sorry for that. I hope this is going to be resolved asap.

Cheers! MK

> I have noticed that my website is currently down with a timeout - sorry for that. I hope this is going to be resolved asap.

HN hug of death strikes again

LOL - It should be back online again.

I loved the article! Spot on and just the right depth for the article size (I am a CPU architect).

I am not an expert but it seems like a great source to understand chips without getting top deep. It reminds me the classic « Nand 2 Tetris » course [1] with less involvement indeed. Thanks for the developer and thanks for sharing. Curious to know industry expert comments !

[1]:https://www.nand2tetris.org/

There is also the NandGame: https://nandgame.com/

From the "Overview" page:

> Microchips – also referred to as integrated circuits – are considered to be among the greatest technological achievements of the last century. Their invention has paved the way for a digital revolution that keeps changing the world to the present day.

...

> The ENIAC computer from 1946 had over 17.000 vacuum tubes and suffered a tube failure on average every two days, which was time-consuming to troubleshoot and repair. With the invention of the transistor in 1947 by Bell Labs, the components became significantly smaller, but the transistors were still wired together individually. This reduced power consumption of those computers and their overall size, but not their wiring complexity. It was not before the invention of integrated circuits before computers became way more efficient and easier to operate and maintain.

I find it on some level hilarious that one of the fundamental breakthroughs that allowed the technological revolution pick up speed and perception-wise cross the barrier from "sophisticated machinery" to "magic" was, in some sense, proper cable management.

It's really /elimination/ of cable management. Essentially the same thing that makes printed circuit boards superior to wire wrapping. Turned a manual labor process into a (photo)lithographic process. Not that different from the replacement of hand-lettered manuscripts with the output of a printing press!

In larger electronic and electromechanical systems, cables and connectors ("harnessing", collectively) are still major weak points.

I guess with scale, the tiniests of details become the bottleneck :)

I am really impressed by the clarity of the illustrations in the article.

https://archive.is/hYvUp for most contents.

I wonder where we would be if a printed copy of that post was delivered back to TI and Intel etc R&D labs 50 years ago.

Interesting, Turing Complete is also from NAND to microcomputer.

[1] - https://store.steampowered.com/app/1444480/Turing_Complete/

Reply on news.ycombinator.com