Taiwan Semiconductors 2 Nanometer Future and Beyond

Taiwan Semiconductor is realizing its major research projects. They will have volume production of their 3 nanometer chips in 2021 and will get beyond to 2 nanometers and toward 1 nanometer chips in 2023-2025.

2020-2026, Taiwan semiconductor is working specialty on SoC (system on a chip) technology (including new NVM, MEMS, RF, analog) and transistors for 8 – 10 year out horizon.

Beyond the 2nm node, they are working on more advanced 3D transistors, new memory and low-R interconnect.

For 3D IC advanced packaging, TSMC is developing innovations for energy-efficient sub-system integration and scaling to provide further augmentation to CMOS logic applications.

They have an intensified its focus on new specialty technologies such as RF and 3D intelligent sensors targeting 5G and smart IoT applications. They are still working on new materials, processes, devices, nanowires and memories for the long-term, beyond eight to ten years.

Taiwan Semiconductor has over 50% of the semiconductor market and number 2 is Samsung with 18%.

TSMC has a huge potential to capture and drive the smartphone move to 5G.

Written By Brian Wang, Nextbigfuture.com

12 thoughts on “Taiwan Semiconductors 2 Nanometer Future and Beyond”

  1. Electrons don't behave that way. They are neither waves nor particles; nor is any other "particle"; the search term you want is wave-particle duality.

    If electrons behaved like clasical point particles, orbiting a charged nucleus. They would just radiate brehmsstralung and spiral into the nucleus; that's how they minimize energy.

    Quantum mechanically, there is a wave function for momentum and a wave function for position (they are like probability density functions except they have a complex "phase", so they can interfere, even with themselves). They are related through a fourier transform. You can't have a definite frequency (a continuous tone) and an infinitely short duration; if it's just a blip it has to contain every frequency in equal amounts to be localized to just a point in amplitude space and vice versa. That's how you get Heisenbergs uncertainty principle; if an electron goes and sits exactly ontop of the nucleus, you know exactly where it is and you know nothing about its momentum. Since it's momentum is equally likely to be anything, it could be going 99.999999999% of the speed of light; it requires infinite energy to confine the electron to an infinitely small space. The electron in an atom is not whizzing around the nucleus really fast with a definite position; it really is behaving like a smeared out S-orbitals and P-orbitals and so on. Interactions with other matter is how it is not like a wave; interactions are definite and particle-like.

  2. The node name has been decoupled completely from the process name. If gate length is the process name as in the good old days, then TSMC 7 nm should be called 54 nm.

  3. Though atoms DO have a measurable size.
    And in crystals of silicone they have a regular spacing, which is 0.235 nm
    Once the components get to the size of a single atom you know they aren't getting any smaller without adopting completely new tech.

  4. "Do simicounductor manufactures believe the public is so uneducated as to believe a 3nm gate is possible?"
    Yes, I don't know all that much about the subject. I just follow the news. So you say here the limit is 5 nm. So what would the point be in spending money on smaller technology?
    I had been under the impression that an electron has no size. And a quick search says:
    "When it comes to electrons, we know
    of their mass and charge… and that's it [1]. As far as structure goes
    they are considered to have none. … It is thus more realistic to
    suggest that an electron does have a non-zero size, even though it be exceedingly small and irrelevant in most considerations."
    So why should the limit be in nm?

  5. We can find confident pronouncements with technical arguments proving that Moore's law is physically unable to go below 10 nm, 7 nm, 5 nm, 3 nm, and I suspect 15 and 20 nm, maybe even larger.

    At some point they will be correct. But note that the actual design of the transistors (and presumably other components) are being redesigned to work at smaller and smaller scales.

    My point being that it will stop when it actually stops, and only the bravest person would try to predict this point.

  6. Nope. The "size" of an electron in a chip is about 5nm. Higher density may come from better engineering but the gates are about as small as they will ever get in silicon.

    The hype is kind of insulting. Do simicounductor manufactures believe the public is so uneducated as to believe a 3nm gate is possible?


Leave a Comment