Ricochet is the best place on the internet to discuss the issues of the day, either through commenting on posts or writing your own for our active and dynamic community in a fully moderated environment. In addition, the Ricochet Audio Network offers over 50 original podcasts with new episodes released every day.
How to Build a Computer Part 2 of N: Crystallography
Last week we saw how to turn sand into silicon. This week I was planning on showing you how to turn silicon into a semiconductor. I mean more of one than it already is. Unfortunately my brief notes on crystallography went long. This week we’ll discuss crystals, next week we’ll do doping, and the week after that we’ll finally get to transistors. Unless I wax even more loquacious, which is the way the smart money is betting.
In a crystal every atom is slotted neatly into an ordered lattice, and every spot in the lattice has an atom in it. With some exceptions. Actually those exceptions are most of what we’re going to talk about today. Let’s assume this is a perfect silicon crystal:
Delicious Delicious Science!
Each cookie represents a silicon atom in it’s spot in the crystal lattice. Then this would be a crystal with a vacancy:
Om nom nom…
The thing about a crystal is that it really isn’t as static as we might assume. All those atoms in there are jiggle-jiving around. More so at higher temperatures, and never actually stopping above absolute zero. That means that even in our ‘perfect’ monocrystal we’re going to have vacancies where the random motion has jiggled the occasional silicon atom out of it’s spot. Let’s look at some more crystal defects:
Wish I had thought of this method while I was still in school.
Couple more things going on here. One, note the cookies that’re not aligned to the grid. It’s possible to get atoms to sit in spots in the crystal between the lattice points. Either a silicon atom or an impurity. That’s called an ‘interstitial’ defect. Some atoms dissolved in a silicon crystal will tend to stick to the places were atoms ought to be in the crystal, and some will tend to be interstitial defects. Oxygen, for example (recall how last week we mentioned getting oxygen in our boule. Or don’t recall; it was a thing either way.) Oxygen, for example, will tend to be an interstitial defect. We can also see a substitutionary defect where the silicon atom has been replaced with another. Looks like molasses.
Atoms can diffuse through a crystal too. Diffusion is the natural process by which one material mixes into another. It’s easy to see diffusion if you’re dropping food coloring in water (or smell if you’re sharing an elevator). It happens in solids too, only much slower. Two atoms can swap places in the crystal due to that same jiggling I was talking about a cookie ago. Minute. I was talking about it a minute ago. Two silicon atoms swapping places is boring (how can you tell it happened?), but a silicon atom swapping places with another element has potential.
Better do this right; someone’s going to go back and check if I actually moved all those cookies.
Atoms swapping lattice spots is one way to move them through the material. Interstitial atoms diffuse much more quickly; there’s nothing sitting in their stopping point. You can also think of the vacancy defect ‘diffusing’ through the crystal. An atom shifts over to fill it’s place leaving a new vacancy where that atom was. This hole can shift through the crystal like one of those slider puzzles, before you took all the pieces out to solve it the easy way.
All these defects involving one atom (or space) are called point defects. You can also have line defects (where a whole sequence of atoms is shifted out of the way) or plane defects (two dimensions), or volume defects, which generally means it’s a chunk of foreign matter in your crystal. If one defect diffuses into another one they have a tendency to stick together after that; lower energy requirement. While a larger defect can also diffuse it’s much less likely because you need to diffuse the whole thing, so they tend to stay in the same place.
There’s actually a benefit to having the larger defects in your crystal. So long as you’re not building a device right on top of them. If a point defect hits another crystal defect it’ll tend to stick there. The larger defect won’t be moving around much. It’ll act as a sort of vacuum cleaner sucking up the point defects. They call this “gettering”. That’s why you might want some oxygen in your wafer. Oxygen near the surface of your wafer will tend to diffuse out (remember that interstitial defects move relatively quick). Oxygen deeper down can getter any other defects which might mess up your device. As you manufacture smaller and smaller transistors this trade-off doesn’t pay off so well, and oxygen becomes less desirable.
That tells you what it is and how it works, but why would we want to dope our silicon?
Unlike these cookies, do not actually eat boron. It’s bad for you. I mean, worse than overindulging in cookies.
A silicon atom wants eight electrons in it’s outer shell. It has four of it’s own. It shares one of each with each other silicon atom surrounding it, and everyone is happy. Now let’s add some impurities. Replace some of those silicon atoms with boron. The green M&Ms represent electrons, the red one is the space where one is suspiciously missing.
Boron only has three electrons to bring to the table. This isn’t enough to please the silicon atoms around it. Eventually one is going to add a second electron, which it grabbed from one of the silicon atoms on another side. Who are now unhappy, and grabbing another atom. Sound familiar? Effectively we’ve got a hole where one electron could fit that’s diffusing through your crystal. These holes can carry electricity. You’ve got yourself a bonafide semiconductor.
By fiat of the Department of Energy all schluffable electrons must be blue. Not the representations, the electrons themselves.
Boron gives you a positive carrier. Or rather, a hole where a negatively-charged electron ought to be. Because it gives you a positive carrier it’s a P-type dopant. Phosphorus, on the other hand, comes to the party with five electrons. A similar process happens where the silicon atoms try to schluff off that extra electron off like a hot potato, and you get a negative carrier. That makes phosphorus (which has the chemical symbol P) an N-type dopant. (Yes, that bugs all of us).
Great! What does that do for me? Good question. We’ll cover that two weeks from now when we talk transistors. Join us next week when we cover the hows of doping in “Anne Oakley on the Ion Gun” or “Wicked Dope Moves, Yo.”
This is part two of my ongoing series on building a computer, the Florin way. You may find the previous part here: 1 (silicon), or all of them under the tag How to Build a Computer. This week’s post has been brought to you by Miracle Max. Miracles provided at prices that can’t be beat, and that’s not blaiving. Miracle Max!
Published in Science & Technology
It’s all good until a Ravenous Balzer comes along and eats all the silicon.
Those boron bastards. You just can’t trust them.
I have a technical question – do you use M&Ms, Reeses Pieces or Skittles for electrons? Does it make a difference for the performance of your computer?
Hank Rhody’s one of the best science popularizers I’ve ever read. “How to Build a Computer” is shaping up as a great series. But Judge Mental mentioned food (well, sort of) and now I’m hungry.
So you could look at silicon as a cereal bowl of Cocoa Puffs mixed with some Cheerios. The electrons are like Rice Crispies that pass through the holes in the Cheerios. Subatomic particles are said to have charms, which are guarded by a forcefield within the nucleus, run by a lucky subatomic man wearing green, with a pot of gold.
Think I’m beginning to get it.
I use M&Ms generally. I tried Skittles a week and a half ago, now I need to buy a new power supply. I save the Resses Pieces for special occasions.
One day in sonar school I learned, “holes flow.” The next day, after some classmates vociferously protested the paradox (“How can a nothing flow!!??”) I learned that they didn’t really flow, they just appeared to.
Also, “getter:” A noun from the days of vacuum tubes which seems to have become a verb.
Is that the guy who breaks computers by eating the
cookiessilicon wafers? And your silicon appears to have some interstitial impurities.Correct.
No, they really flow. What you learned is wrong.
If they don’t flow, then nothing else does either.
In this sequence, did x flow to the right?
…xAAAA AxAAA AAxAA AAAxA AAAAx…
Yes.
What about in this one?
… AAAA A AAA AA AA AAA A AAAA …
Yes. We’ve just substituted the space character for “x”.
It doesn’t matter what x represents, as long as it represents something for which we can give an operational definition. An operational definition of a class of real object is a method of determining whether or not a given observable phenomenon is an instance of that class.
You can call the character “x” something and a space character “nothing” but these are arbitrary distinctions with no significance. As long as we can give an operational definition for a space, it is no less something than “x”.
Great writing, Hank.
Where are the rainbows?
Outstanding, Rhody.
Perhaps cooking spray would help.
Stay tuned for the photoelectric effect.