SoylentNews is people
SoylentNews
With the development of carbon nanotubes and graphene, scientists were given an entirely new collection of materials to work with: sheets and tubes that could be consistently made with thicknesses roughly those of individual atoms. These materials hold the promise of building electronic devices with dimensions smaller than is currently possible through any other process and with properties that can be tuned by using different starting materials.
So far, most of the attention has gone to re-creating new versions of familiar devices. But a new paper by a group of researchers in Shanghai looks into what can be done if you're not constrained by the sorts of devices we currently make in silicon. The result is a device that can perform basic logic in half the transistors silicon needs, can be switched between different logical operations using light, and can store the output of the operation in the device itself.
Small footprint transistor architecture for photoswitching logic and in situ memory (DOI: 10.1038/s41565-019-0462-6) (DX)
(Score: 5, Informative) by vux984 on Wednesday May 29 2019, @04:44AM (4 children)
You can kind of think of a transister much like an electromagnetic switch.
Suppose you had a single wire, with power applied to one end and an iron connector in the middle on a spring that keeps the circuit open; so electricity can't flow.
+ __________/ _________ -
Then you place an electromagnet just below the spring on a secondary circuit. When you turn the electromagnet on, it pulls the switch closed.
Apply power to the main circuit but not the electromagnet circuit, and the switch is open, and there's nothing getting through.
Apply power to just the electromagnet circuit, the switch closes, but there's nothing coming through.
But when you apply power to both circuits, the 2nd closes the switch on the main circuit, and the power from the first flows through.
A modern transister is the same idea, except using the properties of semiconductors to open and close the switch instead of a spring and a magnet. It's even the same construct with a main power 'emitter', connected to a 'drain', and then another lead is called 'gate'. You power the 'gate' lead and the electricity can flow from the source to the drain. The essential bit is that the application of current to the gate changes the conductivity of the material allowing the current to flow. hence the term "semi-conductors".
(Score: 0) by Anonymous Coward on Wednesday May 29 2019, @03:06PM
That's a PNP transistor. An NPN sets a ground path, then there's PN and 4 lead transistors.
(Score: 3, Informative) by RS3 on Wednesday May 29 2019, @03:49PM (2 children)
That's a great analogy. Silicon is called a "semiconductor" because just like the electromagnetic switch (relay) you described, we can control the flow of electrons through it by introducing an electric field somewhere in the middle of the flow. The physics is far too much to go into here.
There are many other semiconducting materials. Germanium was very popular in the 1950s and 60s, and is still used in a very few applications. Gallium-Arsenide is another popular one.
In your last paragraph you referred to some terms and they're a little mixed. It's not going to matter unless someone tries to design or build something.
Emitter, Collector, and Base are the parts of a BJT (Bipolar Junction Transistor) and they're generally NPN or PNP. The terms are somewhat descriptive, called: collector and emitter due to electron flow in an NPN, and base because it was the physical base plate of the original construction. Current between the base and emitter controls the current between the collector and emitter.
Source, Drain, and Gate are the parts of an FET (field-effect transistor), which was the first type invented. Main current flows through the source and drain, and the gate, well, gates that current, like the electromagnet in your analogy.
Simplified: the source and drain are connected to a (tiny) bar of either P or N doped silicon.
There are JFETs, where the gate is a PN junction (or NP), and IGFETs, or MOSFET as they're generally called, where the gate is an insulator and tiny metal plate. In either case, the volts applied to the gate (electron "charge") produces an electric field that causes more or less electron flow in the main semiconductor material, thus controlling the flow of electrons between the source and drain.
Electronic circuit-wise, FETs behave very much like vacuum tubes, or "electron valves" as the Brits descriptively call them. Electrons flow from the cathode to the plate, and are controlled by a voltage on the grid.
(Score: 2) by vux984 on Saturday June 01 2019, @03:21AM (1 child)
Argh. Thanks for the clarification. I'd meant to just stick with FET, source/drain/gate. Not sure why i said emitter. I was originally going to describe BJT as well but thought it would be simpler to stick to just one, and FET is, to my mind at least, the most accessible.
I'm a CS grad, I got some exposure to microarchitecture, bus design, clock signals, and pipeline design; and developed finite state machines with gates... but never got any formal education lower down than that. From what I've read even an "antique" FET transistor is too difficult to easily produce as a "science project". I'd have loved to have made one with the kids.
(Score: 2) by RS3 on Saturday June 01 2019, @06:43AM
You're quite welcome and please never feel badly about terminology. Emitter, source, electron spray nozzle, tomato, tomatto, potato, potatto, let's call the whole thing off! I mix up terminology all the time. People's names too.
Great ASCII sketch, btw!
It's awesome and impressive that you've learned so much about hardware in a CS program. Lots of great analogies- a valve, relay, floodgate. Somewhere in one of these discussions (maybe on another board?) I recently wrote about digital logic being largely CMOS- FETs, but BJTs are used a lot too. As an EE I should know why you'd choose one over the other but I'm not 100% sure. Over the years we've seen both types used for digital, analog, high-speed/frequency, high power, high voltage, etc. Now we have FETs that have extremely low ON resistance so they're being used for power circuits, but so are IGBTs.
Yeah, it would be super-cool to build an FET. You need very pure silicon, and I'm not sure how it's done but it's easy enough to look up. There was (and still is) a silicon lab at my college. I don't think I ever went in there- wasn't interested in device physics, but it turns out to be very lucrative. Hmmmm. Anyway, in my days they weren't able to produce usable silicon- too much impurity, but now that lab makes chips, it's all lit with amber lighting (not sure why), some low class number clean room, air locks / air showers, clean suits, etc. My hunch is the silicon is purified with a lot of heat in a vacuum chamber, and it's grown as a big crystal- need the lattice structure for electron flow, and then infusing ("doping") boron or arsenic or whatever. I love the physics but the actual manufacturing bored me, and otherwise I'm into manufacturing. All the hard work we EEs do- we even breath arsenic vapors to give you guys something to play with. :-}
Your homework assignment: read about 4-layer silicon devices (SCRs, TRIACs) (an SCR stays "on" as long as there's current flowing through it, so it's a 1-transistor memory cell, rather than a 2-transistor "flip-flop"). I'm full of useless information...