eli5 How are transistors made so crazy small?
In: 7
Very carefully, perhaps these can help
Interesting piece of technology
Okay, basic electronic construction and physics. Some substances are able to easily gain or lose electrons. These allow electricity to flow well, as electrical current is simply electrons moving around. These are “conductors.” Other substances are highly resistant to gaining or losing electrons, which means they do not allow electricity to flow well. These are called “insulators.”
There is a third kind of substance that falls in between them, that holds on to its electrons harder than conductors but not as hard as insulators. They are called “semiconductors,” of which silicon is the most important one.
Since everything that happens here is on the atomic level, it is very easy to make transistors on the small scale. A mechanical switch with copper contacts would have to be much larger than a transistor. Copper is a conductor, one of the best ones we have, so electrons can jump from one contact to another over a “decent distance.” A gap of a couple millimeters is enough to break the circuit, but compared to transistors, that’s a massive gulf. Plus, you need something to mechanically move the contacts. Usually an electromagnet is used. Put an electromagnet in a formation that it will cause contacts to open or close when the magnet is energized, and you have a “relay.” That’s what we used before transistors, and are often used today, though we no longer use them for “thinking” in electronics.
But with semiconductors, they can change from being a conductor to being an insulator very easily. The trick is to add just the right amount of impurities in just the right structure. This is called “doping,” and in the world of electronics, it’s a good thing. All it takes is a single atom to switch a properly doped piece of silicon from an insulator to conductor and back again. Plus the process is purely electronic. There are no moving parts, so no mechanical components are needed. All you need to do is apply an electrical current to the third leg of a transistor, and the other two legs will go from “open” to “closed.” Once the current on the third leg stops, the transistor “opens” again and electricity can’t pass through.
Since the conductor itself changes to an insulator, you don’t need an air gap for the contacts like you do in a relay. And since you don’t need to physically move the contacts, you don’t need a magnet. So a transistor is able to do the same job at 1/100 the scale of a relay.
Hope this helps.
Silicon + lithography.
Silicon’s important because it can be doped to change its properties, so we can create semi-conductors that change conductivity based on a control signal.
Lithography basically means using light and chemicals to print different layers of doping/metal traces/etc onto the silicon.
So we start with a thin disk of silicon and cover it in a thin photo resist which forms a protective coating. Then we project a stencil onto the photoresist as light, which damages it where it’s exposed to the light allowing a acid bath to wash off the resist in those places (or in some cases they use a resist that hardens under light and the unlit areas wash off).
Now you can use another chemical to dope the exposed silicon, cut grooves, deposit metals, etc.
Because the light pattern can be extremely exact, you can create incredibly tiny details. Allowing for insanely small transistors to be formed.
One key thing is that because lithography uses light you can use lenses to focus a bigger stencil in a much smaller area.
So while the features you are creating are only nanometers big, the stencil from which they are made is in a much easier to make size.
You know how a microscope can make something really tiny look big. Now, instead of an eye to the microscope, put a film (or LCD display) and light source. And instead of the sample you’d observe, place a slice of silicon crystal – that had been put into a mist of vapor of a photo-sensitive substance.
Your film (or the display) has a blueprint of spots where all the transistors are to be on the chip. The blueprint might be pretty big actually, with the microscope’s ocular big enough. You’re now projecting the image – like cinema projector does – except instead of making it big, like in cinema, you’re making it very, very tiny. The photo-sensitive layer hardens under the light – everywhere except where the transistors need to go. Wash the crystal removing the compound from where it wasn’t hardened, and put it in vapor of a substance that penetrates into silicon turning it into the right type of semiconductor for half of the transistors. Repeat the process with a different substrate to complete the transistors. Then coat it in a metal vapor, then the photo-compound, and shine the map of all electric paths onto it. Remove compound, apply acid to remove the metal where it’s unwanted, then apply a solvent that dissolves the hardened photo-compound, and you have a chip, with the silicone crystal forming transistors where needed, and metal paths connecting them.
Others answers have lithography down. The other thing to keep in mind is that transistors are extremely simple devices. They’re not like an oil refinery with pipes running all over the place going to different storage and reaction chambers–they’re essentially just three (or four if you include the substrate of FETs) regions of differently doped semiconductors abutting each other with a couple of metallic contacts.
It should also be noted that we are nearing the minimum viable size of transistors without new technological improvements. Modern IC transistors are so small that leakage currents are becoming a formidable challenge.
* We’ve developed a process where we can create transistors out of a single piece of material that we “inject” with different stuff to change it’s electrical properties.
* We’ve also figured out that we can choose where to inject the stuff by applying a layer of paint to the transistor material.
* If we create patterns in the paint, we can control which areas get the injection and which don’t.
* Here is the really fancy part.
* We figured out that we can make paint that is sensitive to light.
* The light “bakes” it and makes it stay.
* The areas that don’t get baked can have the paint washed away.
* Now we use very powerful projectors that can project very very precise images on a super tiny scale.
* These projectors shine those intricate patterns onto super tiny areas of the transistor material and bake the paint into those patterns.
* Then we wash away the paint that didn’t get baked.
* That exposes the material so we can inject the stuff that changes the electrical properties of just that part of the transistor.
Latest Answers