Q: Tell me, whatever happened to the transistor?
Q: You know, the transistor. That little thing about the size of a garden pea with three wires coming out of it.
A: Oh, now I'm with you. It still exists, but in a different form. What you're talking about is the discrete transistor. That little package held one transistor, and nothing else.
Q: So, what's happening now?
A: The transistor exists on an integrated circuit: a little sliver of silicon on which is mounted a lot of components – almost always several transistors. Also capacitors, diodes, and other things.
Q: So what makes that so good?
A: Several things. First, it is small. But more important, that integrated circuit is made as a single electronic device. Nobody has to mount discrete components on it; they are made as part of it. That saves lots of labor.
Q: You mean they make an integrated circuit, then pull it off the production line and make another integrated circuit?
A: No. What they do is they take a large silicon substrate, and using photographic techniques put a lot of the integrated circuits on it. Then they make it, but they are in fact making hundreds of those identical circuits. Then they separate them, and use them in some electronic device.
Q: Who figured this out?
A: Two people are credited for the integrated circuit. Jack Kilby an engineer with Texas Instruments, and Robert Noyce, with Fairchild Semiconductor Corporation. Kilby used a germanium substrate, and Noyce a silicon one. They both filed for patents in 1959. After a protracted court fight, they cross licensed their technologies. Incidentally, the original IC had only one transistor, three resistors, and one capacitor. It was the size of your little finger.
Q: Now they hold more?
A: Today an integrated circuit smaller than a penny can hold 125,000,000 transistors.
Q: I'd say that's enough! Right?
A: Wrong. The number of transistors and other components that can be placed on a single chip has been doubling every 18 months. We've talked about that before. It's called Moore's Law. And it's still happening.
Q: Does that mean that the transistors will be getting smaller still?
A: They will indeed. But we should mention first why that is important. Clearly the product into which these things are built will be smaller – and everybody likes smaller. Think of the cell phone.
Also, our switching circuits operate so fast that the speed with which signals are sent from one chip to another becomes important. So if the signal has to go half as far, that is important.
However, there's a flip side to this smallness. And that's the old bugaboo: power. In this case when a chip with thousands of transistors gets smaller, then there is more leakage of electricity from one component to another. And leakage translates into heat - thousands of watts in some applications. That heat is hard to handle. If it remains captured in a cabinet it will burn everything up. So the heat has to be pumped out, and the whole room has to be cooled. Now we have the added expense of buying more air conditioning equipment.
Q: Is there an answer?
A: Intel and IBM are working on one. In their designs, silicon is not used as a substrate, the element hafnium is. With hafnium there is much less leakage, so the substrate can be thinner. In fact, they can get approximately twice as many transistors on a chip with the new element. Also, because of the reduced leakage, there is less heat produced. And if the device is being used in a battery-powered appliance, then batteries last longer. Finally, switching speed is a lot faster.
Q: Now wait a minute. You say it is smaller and faster. Be specific.
A: 30,000,000 fit on the head of a pin. It can switch on and off approximately 300 billion times a second.