Building a Transistor (1905)

On another board we are discussing the practical side of a time traveller arriving in (say) 1905 from (say) 2005. He has with him some artifacts. Amongst these are some transistors and other solid-state components.

Could 1905 science replicate them? I figure that all you have to do is a spectral analysis of their chemical makeup and your in business. After all they have no moving parts.

Others say the purity of such chemical could not be reached (in large amounts anyway) in 1905.

Your thoughts?

“YOU’RE in business” darn it. (Where is the edit button?)

Pat# 2524035:
June 17, 1948

It’s a little more complicated than just knowing the chemical makeup of a transistor. Transistors aren’t made just by combining chemicals. Transistors are “grown”. You basically start with a puddle of molten silicon. It can’t have too many impurities in it (don’t ask me if this is possible in 1905). Now, you basically “grow” a silicon crystal out of it, but here’s the catch. You have to add certain types of impurities while the crystal is growing.

Transistors are made out of N and P type silicon. If your silicon crystals contain some bits of phosphorus, arsenic, or antimony (hint - these are all in the same column on the periodic table) you end up with N type silicon. If your impurities are boron, aluminum, gallium, or indium (again, all from the same column in the periodic table) you end up with P type silicon. A layer of N type silicon grown on top of a layer of P type silicon makes a diode. NPN or PNP makes a transistor.

I don’t know what was available in 1905, but just examining a modern transistor using spectral analysis and looking at it under a microscope is just going to baffle someone from 1905. If you can explain the basic physical and chemical properties, maybe you’ve got a shot at it, but if someone without any knowledge of semiconductors takes one back with them, the chances of a 1905 scientist being able to reverse engineer the thing probably aren’t good.

Single-crystal germanium could have been made in 1905, but it would have been far too impure to make useable transistors. All og this had to wait until the invention of “zone refining”-in the mid-1940s-by a guy named Pfund.

Thank you very much everyone. Bedtime here, I will be back in the morning.

Beyond the purity problem, which I agree with, I don’t think the state of physics in 1905 was adequate to explain how the thing worked.

Typically only one type of impurity is doped into the crystal as it is grown. The remaining layers are diffused in from the surface after the crystal is sawed and polished into wafers. All of the doping could be done by diffusion, but starting with a doped crystal saves several processing steps.

That may not be an issue. Edison didn’t understand how/why rectification occured when he discovered the vacuum tube diode, and it is pretty clear that Lee DeForest didn’t really understand regeneration when he made use of it.

The patent I reffed refers to this patent, 2402661 for the making of silicon. They start with 99.5% pure, and melt it slowly in a crucible to drive off impurities. As the melt cools, it spontaneously separates into an upper, P type, and a lower, N type layer.

Lyden jar

Bill Pfann, no?

(I only have the slightest recognition of the name because of Richard Hamming’s anecdote in here)

You’d need more than just the knowledge of how to make a transistor, you’d need to know what it does and how best to incorporate that into the state of electronics in 1905. You’d probably have better luck in advancing technology by getting someone to dust off Babbage’s difference engine and combining that with the primitive mechanical computing systems that existed then.

All in all, it seems it would have been right on the ragged edge of doable in 1905. Oddly, the goal would not have been closer in 1910 or even 1920. By 1940, chemical techniques would have made it easier. (Conversely, it would seem doing it in 1905 would have been no easier than say 1900 or 1890.)

Since we are talking about time travel, once you can swallow that idea, accepting a 1907 transistor becomes pretty easier.


Eh? A Leyden jar is a type of simple high-voltage capacitor. Yes, these would have been easily constructed in 1905, but what does this have to do with transistors?

I doubt that Og could help you with building a transistor, whatever the date, because OG SMASH!

Let’s skip ahead to 1910, or so just for a minute…

Crystal Radios, (Link - )were, technically semiconductors in that they used naturally occuring Galena (Lead Oxide) crystals as the “rectifier”. The Crystals were, as mentioned lead okide, and had impurities which occured naturally. A fine wire was traced over the surface of the crystal until it hit a “sweet spot” which contained the right balance of impurities to act as a rectifier.

Vacuum tube style radios came after these generally home built ones. I have in my collection of antique radios a crystal radio built by a great uncle around 1915. It has been restored to full working condition. One other interesting note; Crystal Radios required no outside source of power - they were powered by the energy in the radio waves themselves. It should be noted that no amplification was involved, so that the only way to listen to them was on high impedance headphones.


I meant understanding how amplification is achieved, and the right dopant level. They’d have no problem using them even in 1905.

hijack -

I’ve long considered that truly advanced tchnology taken far enough out of context will appear to be an inert lump. Show a 1905 scientist a transistor, even a large one, and he’ll see a blob with 3 wires coming off it. His best tools wil be able to destructively determine that it was made of silicon. Little or no insight will result.

Were somebody to show a warp drive to NASA tomorrow, they’d see what appears to be an inert lump of ???. We might be able to determine its gross makeup, but whatever the “secret sauce” is that makes it work would be utterly opaque to us.

The artifact embodies a tremendous amount of accumulated knowledge & effort, but does not expose it to be extracted.

John Campbell had a great editorial in Analog in the early '60s that made exactly this point. His example was a jet plane showing up around WW I. Even then, a lot of the control equipment would be unfathomable. If a processor turned up in 1905, I suspect that microscopes that could see the signal lines didn’t exist (you can see the basic structure) and assay techniques to determine the level of dopants didn’t exist. No to mention, who would be apply to apply a clock of sufficient speed.

Even today, when we strip away levels of material to do physical fault analysis, half the time we don’t see what we want to see, and we have all the advantages.

But a transistor with a time-traveller might help.