[Note: I've struck out some the more obvious "capering" that my editors sprayed this article with in the name of making it smell like house style back when it went up.]
Tuesday, April 21, 1998
by Rob Landley
Austin, TX (Apr. 21, 1998) -- Brace yourselves, Fools. Today we're going to talk about Intel (Nasdaq: INTC).
I know, I know. I probably should be writing all about Viagra and Pfizer's
$8 rise yesterday (and $2+ rise today!). But, c'mon, we own a bunch of other
stocks. And there's no guarantee that our beloved Intel won't march into
the lead in the Cash-King portfolio over the next five years! In fact, this
may well be the most attractive C-K business to buy right now (I think so).
So let's talk about its business, aiming for eventual mastery of it.
How many investors in America own some Intel shares without really understanding
the company's products? It reminds me of the many of us with relatives who
have no real understanding of what we do for a living, but are extremely
impressed by it nonetheless. Such is the life of the programmer... and the
engineer, the accountant, the geneticist, yea, even The Fool.
And yes, strangely enough, that's the way a lot of us feel about Intel, too.
She's such a nice company, flosses regularly, always eats her vegetables,
and is very polite -- at least when she's not hanging around with that riff-raff,
Microsoft (sorry, I couldn't contain myself). On the whole, she's a model
citizen. And for investors, she's been a dream -- with profit margins ascending
into rising global demand. An absolute dream. (I tried to contain myself
there, but I escaped.)
But, we ask (peering across the cocktail party at her), what does Intel actually
do for a living? Apparently, she goes into the office every day and makes
microprocessors? Fine. Lovely. Brilliant! But what in the heck are
microprocessors? How do they work? Why do they matter?
Intel puts it simply: "Microprocessors are the brains of your personal computer." On this tiny silicon chip, millions of switches and pathways help your computer make important decisions and perform helpful tasks.
And how do they get made?
Well, manufacturing microchips basically involves a very large, expensive photocopier. Massively oversimplifying things, you make a really really REALLY big drawing of all the wires you want on the chip, and then you photograph that drawing to make a big slide. Then you shine a light through the picture on the slide and project it onto a specially prepared piece of not-exactly-glass with gunk painted on it, which is called a "wafer."
Wafer? For starters, think of the
delightful, light cookie and you're not
far off. But actually, it looks rather more like a pizza. The suckers are
usually at least a foot across. How do I know? They had a few etched wafers
on display outside the library at IBM when I worked there.
Ok, back to our chip development. To restate, you shine your photo of the chip onto a thin wafer.
Anyone who has ever used sunlight and a magnifying glass to set fire to something
understands the next part. Shining the light through the slide onto the wafer
ends up burning some of the painted gunk off the wafer. The burnmarks perfectly
line up with the outline of wires on the slide above.
If you feel lost, just
re-read the line above again -- this is simple.
The final step is that you dip the glass-like wafer into some truly noxious chemicals, and it deposits metal (generally aluminum) on any exposed glass not protected by gunk. The outlines burned onto the wafer from the slide above provide the grooves for the formation of aluminum wires on the wafer.
Eh?! Routine. I promise you that if you re-read the last few paragraphs,
you'll forever understand the basics of how microchips get made. Now we're
going to get a little more technical and make you work a bit, Fool.
Ok, during this whole burning-and-wiring process, transistors can be formed. All a transistor is, basically, is a little gap in the wire with some highly toxic chemicals tainting the glass in between. For example, you can stick some arsenic at one end of the gap and some gallium at the other end. Chemically -- and you don't have to follow this too closely for now, but enjoy reading it -- the arsenic is happy to loan out one of its own electrons, but won't welcome anything that increases its electrons. Gallium, on the other hand, is happy to gain a spare, but unconditionally refuses to give any of its away.
Now, let me see if I can make this clear... ponder this diagram and read
on for an explanation:
<--- Wire : Silicon+Arsenic : Silicon+Gallium : Wire <----
This arrangement of chemicals works on a wire like a one-way valve on a pipe. The electricity can flow down the wire in only one direction. Here's how it works:
The arsenic is happy to lose an electron, so it sends one of its own into its wire (at the left of the diagram). The gallium is happy to gain an extra electron, so it sucks an extra one out of its wire (at the right in the diagram). Then the gallium gives the extra to the arsenic so they balance out back to where they started. The end result is a single electron getting shot out through the wire on the left side.
Electricity simply won't flow back in the other direction. The wire can't give an electron back to arsenic (it won't take it unless it's already lost one first), and gallium is similarly reluctant to give one up to its wire (or anything else) unless it already has an extra from somewhere.
By the way, some guys at Bell Labs (now Lucent) won a Nobel Prize for discovering
all this. And they deserved it. Without it, the Motley Fool would still be
a hand-printed newsletter distributed to a few dozen people in Alexandria,
I don't know about you, but my financial situation would look very
different (notably weaker) without the Fool. Without that flow of electrons
down the wire, without the transistor, you wouldn't be here either.
What's the point of all this? Well, transistors can be used to make on/off
switches, by sticking a third wire into the side of the gap in a way that
I can just about understand when I have a diagram in front of me.
that I don't.
Suffice it to say that making more wires in a chip is just a matter of drawing more lines on the picture that gets xeroxed onto the wafer. You can then paint a NEW layer of gunk on top of this mess and do the whole process over again to create more than one layer of wires and transistors. It gets increasingly elegant. Layer upon layer of circuitry sits on a single chip, after light has burnt in the pattern and chemical baths have laid down wires, transistors, and insulation.
Now, don't be too impressed yet. Why? Because the above process as described
is to the Pentium manufacturing process what a Volkswagen Bug's air-cooled
engine is to the modern, fuel-injected, computer-controlled, turbo-charged
high performance Ferrari engine. But, you've got the general idea now of
how microprocessors get made -- Intel's core business.
Tomorrow I'm going to tell you generally how Intel makes its chips faster than the competition -- faster and faster every year. In the meantime, here's a link to some of Intel's web pages on this subject: How Microchips Are Made.