Wikipedia offers an excellent list of semiconductor materials but does not include which can be used to create integrated circuits or even logic transistors (PNP or NPN). The Integrated Circuit page mentions "transistors built from other materials: graphene transistors, molybdenite transistors, carbon nanotube field-effect transistor, gallium nitride transistor, transistor-like nanowire electronic devices, organic field-effect transistor, etc", which does not lay out other possibilities.

If you took away the requirement of being smaller/faster/cheaper than modern Silicon transistors, which semiconductors besides Si, GaAs and GaN could be used to construct logic circuits? I am particularly interested in abundant non-toxic compounds that could be purified and crystallized with relatively few resources. In particular, are any of the following possible candidates?

  • 5
    \$\begingroup\$ Silicon not Silicone. \$\endgroup\$ – Spehro Pefhany Oct 23 at 20:43
  • 2
    \$\begingroup\$ Silicon is the second most abundant element in the Earth's crust. As a raw material it is about as cheap as cheap can get. What makes the silicon for semiconductor electronic devices expensive is purifying the silicon to extreme levels. Presumably any other semiconductor you find will need to be similarly purified to be useful in electronics. So there is not likely to be any real cost advantage in the raw material, relative to silicon. \$\endgroup\$ – The Photon Oct 23 at 20:55
  • 1
    \$\begingroup\$ Silicon carbide? \$\endgroup\$ – pjc50 Oct 23 at 21:15
  • \$\begingroup\$ @ThePhoton actually purification is relatively simple compared to forming pure crystals. Purification is a several-step chemical processes with distillation. It is difficult to form pure crystals because of the high melting point of silicon and the fact that it is difficult (impossible?) to create a pure crystal using deposition -- hence my question of other material choices \$\endgroup\$ – vitiral Oct 23 at 21:39
  • 2
    \$\begingroup\$ @vitiral, The process is well known but that doesn't make it simple. It is certainly more expensive than taking an end loader to a sand pit and loading up a dump truck with the raw material. I don't get why you think purifying your other candidate materials would be simpler? \$\endgroup\$ – The Photon Oct 23 at 22:01

You can in theory use any semiconductor junction to build nonlinear behaviour, and that's enough to implement logic, and that's enough to implement a CPU.

Of course, since you might want to still be able to power them, a composition with a reliable and still low band gap is necessary.

You'll find it's no coincidence that Si is the usual semiconductor, and that the available adjacent groups in the periodic systems limit your choice of elements to use as dopant.

Realisitically, if any other simple anorganic materials would be easier to handle or cheaper to use to produce working semiconductors, even bad ones, then they would already be in use. You'll find, for example, that Copper oxide- and Selenium-based diodes used to be produced, but they were so bad that the harder-to-make silicon diodes have completely replaced them. Same for Germanium, and a few other materials. You can investigate these obsolete technologies, but you'll quickly find out that people tend to care more about materials not being toxic than they did 50 years ago...

A field that might still be yielding new materials that can be used in semiconductor junctions are organic semiconductors, like you already find them in organic light-emitting diodes. That's not saying they are abundant, or can be processed "with relatively few resources", but really, what technical material is?

In the end, I think you should be looking into the homebrew semiconductor community. There's people building ICs in their garages! They still have to use rather nasty chemicals, and have to buy raw wafer materials, mostly, but the truth is that a monocrystal of an any significant size is hard to produce without industrial style processes, and that applies to semiconductors as much as to e.g. artificial sapphire or diamonds for technical applications.

All in all, I think as broad as your question is, you underestimate the complexity of all this. Building a CPU will take at least hundreds to thousands of transistors, and you need them to be reliable, and you want them to be reasonably effective at what they do, lest you get in real heat and power supply trouble.

By the way, aside from maybe the iron(II) oxide, I don't think your substances could be described as non-toxic at all.

| improve this answer | |
  • \$\begingroup\$ I understand the complexity somewhat well. I took a graduate course in my EE BS where I made my own transistor. I'm just questioning everything right now because I would like to reduce technological complexity as much as possible to build a CivBoot.org tech stack. This was good feedback, but I'm still hopeful there can be an easier material to work with than Silicone! I'm guessing that the main reason Si is best for ICs is because pure crystals can be manufactured with very few defects (hence reliability). Is that correct or are there other reasons it is preferred? \$\endgroup\$ – vitiral Oct 23 at 21:35
  • \$\begingroup\$ no, I don't expect there is an easier material: the material isn't the challenge. The purity is, and the hard part is the process of purification, and silicon is as far as I can tell not especially hard to purify. \$\endgroup\$ – Marcus Müller Oct 23 at 21:37
  • \$\begingroup\$ by the way, I consider civboot.org to be wrong in basically all axioms, but that's irrelevant to your question. \$\endgroup\$ – Marcus Müller Oct 23 at 21:41
  • \$\begingroup\$ Would love to hear your thoughts! Feel free to post an issue :) \$\endgroup\$ – vitiral Oct 23 at 21:43
  • \$\begingroup\$ (and to add to the purity aspect above: of course, the growth of monocrystals requires well extensively precise extremely high temperature-controlled vacuum chambers. Like, you're missing but 180 years of mechanical engineering required to get there.) \$\endgroup\$ – Marcus Müller Oct 23 at 21:44

which semiconductors besides Si, GaAs and GaN could be used to construct logic circuits?

Indium phosphide (InP) is used commercially to make very fast logic circuits. It has even higher electron mobility than GaAs, but also the same disadvantages as GaAs (expensive, lacking a native oxide) relative to silicon.

| improve this answer | |

Compared to silicon, germanium requires much lower technology and is itself non-toxic. And by low-technology, I mean it is possible to make your very own germanium transistors in your garage with quite a moderate investment (both in money and learning).

Then again, the "non-toxic" part of your requirements is tricky - the technology is not only the final product. From the mining to the final electronic device you have a lot of processes using wide variety of chemicals - some of them highly active and dangerous to handle. Even if you have the bulk purified semiconductor material, you still have to deal with dopants, etching solutions, flux solutions, probably some polymers for the package, etc, etc... quite a few of them non-FDA approved.

And there is quite a leap between a transistor and a CPU.

But you probably know this.

| improve this answer | |
  • \$\begingroup\$ Germanium is a good suggestion, but I believe it's fairly rare. Was just wondering if there are alternatives \$\endgroup\$ – vitiral Oct 23 at 22:34
  • \$\begingroup\$ And "non toxic" it's clearly a relative statement :) \$\endgroup\$ – vitiral Oct 23 at 22:34
  • \$\begingroup\$ In terms of abundance, nothing beats silicon with its ~25% of the Earth's crust. Of those above 1%, only Iron gets in your list and I am not sure you can even make P-type FeO. At 1.5ppm and easy extraction from common ore byproducts Ge is not that a bad deal. The quite ubiquitous Tungsten, Iodine, Bromine and Arsenic are in the same league with their 0.5..2.0 ppm. \$\endgroup\$ – fraxinus Oct 24 at 7:26

Not the answer you're looking for? Browse other questions tagged or ask your own question.