New Bismuth Transistor Runs 40% Faster And Uses 10% Less Power

Recently in material science news from China we hear that [Hailin Peng] and his team at Peking University just made the world’s fastest transistor and it’s not made of silicon. Before we tell you about this transistor made from bismuth here’s a whirlwind tour of the history of the transistor.

The Bipolar Junction Transistor (BJT, such as NPN and PNP) was developed soon after the point-contact transistor which was developed at Bell Labs in 1947. Then after Resistor-Transistor Logic (RTL) came Transistor-Transistor Logic (TTL) made with BJTs. The problem with TTL was too much power consumption.

Enter the energy-efficient Field-Effect Transistor (FET). The FET is better suited to processing information as it is voltage-controlled, unlike the BJT which is current-controlled. Advantages of FETs include high input impedance, low power consumption, fast switching speed, being well suited to Very-Large-Scale Integration (VLSI), etc.

The cornerstone of Complementary Metal-Oxide-Semiconductor (CMOS) technology which came to replace TTL was a type of FET known as the Metal-Oxide-Semiconductor Field-Effect Transistor (MOSFET). The type of MOSFET most commonly used in CMOS integrated circuits is the Enhancement-mode MOSFET which is normally off and needs gate voltage to conduct.

A transistor’s technology generation is given with the “process node”, in nanometers (nm). This used to mean the size of the smallest feature that could be fabricated, but these days it’s just a marketing term (smaller is “better”). Planar CMOS MOSFETs were initially dominant (through ~28nm), then came SOI MOSFETs (28nm to 16nm), then FinFETs (16nm to 5nm), and now finally Gate-All-Around FETs (GAAFETs, 3nm and beyond).

All of that in order to say that this new transistor from [Hailin Peng] and his team is a GAAFET. It’s made from bismuth oxyselenide (Bi₂O₂Se) for the channel, and bismuth selenite oxide (Bi₂SeO₅) as the gate material. See the article for further details.

Keep in mind that at this point in time we only have a prototype from a lab and the gory details about how to mass-produce these things, assuming that’s even possible, haven’t yet been worked out. We have previously discussed the difficulty of manufacturing state-of-the-art transistors. If you’re interested in bismuth be sure to check out how to use bismuth for desoldering.

31 thoughts on “New Bismuth Transistor Runs 40% Faster And Uses 10% Less Power

      1. I was just going to say that you can’t mention this kind of thing on YT since the primitives commenting there cling to their ‘China only copies’ propaganda lifebuoy.
        But I forgot you guys are here too. Doh.

        1. Doesn’t mean reasonable skepticism doesn’t apply, regardless who made a claim. You well know research like this often takes a decade to make it to production, and that’s if it’s actually useful outside of a lab. Plenty of US technology is the same.

  1. Behind a paywall. Any “alternative” links? Any reviews from someone who has read the paper?

    Bismuth is not used in any commercial transistors. If there were some in production there might be a route to advancing an existing fabrication facility to make these new 2D transistors. Even if it works well in the lab (I have my doubts) there is no path to making chips you can buy on AliExpress.

    I’m putting this headline in my “Bigfoot sightings” folder.

      1. AliExpress isn’t the goal, but it’s populated by both resellers and literal factories, so no that’s just incorrect. Wholesale purchases are usually on Alibaba for overseas sales anyway,

  2. “The Bipolar Junction Transistor (BJT, such as NPN and PNP) was invented by Bell Labs in 1947.”

    Not exactly. The transistor demonstrated by Bell Labs in ’47 was a point contact transistor. BJTs came a bit later.

    “Enter the energy-efficient Field-Effect Transistor (FET)….”

    Interestingly, Julius(?) Lilienfeld was issued a patent for (what amounts to) a FET two decades earlier, though I’ve never seen images of a practical device.

    1. Ah, you are quite right, thanks for picking me up on that one, it was the point-contact transistor developed at Bell Labs in 1947. I will edit the article to clarify. I have recently read about these topics in Principles of Electronic Circuits 2ed by Burns and Bond and I misremembered the detail about the type of transistor developed at Bell Labs (in my head my logic was “it wasn’t the FET therefore it was the BJT, but of course my “logic” was flawed). Chapter 1 of PoEC starts with a brief history of the transistor. They mention Lilienfield (1925), Heil (1935), and Pohl (1938) as all providing precursors to the FET.

  3. Is it just me or did this article bring out the worst of HaD comments? I thought we were all on a roll there being cordial and stuff.

    Anyway, I liked the article. I learned some things, maybe I’m a fool but I believe the research, and I thought this was pretty well written…

  4. Probably a silly question, but what is the substrate ? Is it silicon as is used by almost all fabs worldwide. Or is the substrate material bismuth oxyselenide (Bi₂O₂Se) or bismuth selenite oxide (Bi₂SeO₅), because if it is either I can imagine really heavy devices, and new fabs needing to be created. And for 40% faster and 10% less power, the economics are probably not worth it.

Leave a Reply to TGCancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.