Taiwan Semiconductor Will Have Volume Production of 3 Nanometer Chips in 2022

Taiwan Semiconductor Manufacturing Company (TSMC) plans to expand extreme ultraviolet (EUV) lithography tools for its next-generation 3 nm (N3) node process.

TSMC was the first company to use ASML’s EUV lithography machines for high-volume production and now has at least three processes that use EUV for select layers. They use EUV lithography for its N7+, N6, and N5 nodes.

TSMC’s 2nd generation 7 nm technology (N7+) uses EUV for up to four layers in order to reduce its use of multi-patterning techniques when building highly complex circuits. The 6 nm process (N6) is for customers to re-use IP designed for 1st generation 7 nm, per the report. TSMC’s 5 nm process (N5) can use EUV for up to 14 layers. 5nm has significant increase in transistor density and performance enhancements.

2nd generation 5 nm (N5P) and 4 nm (N4) fabrication processes are based 5 nm technology and will have performance and power benefits. N5P will be available in 2021. N4 chips will have volume production in 2022.

Next-generation 3 nm process (N3) will be a full node improvement over N5. N3 will have 70 percent logic density gain, 15-percent performance gain, and 30-percent power reduction over the 5-nm process. N3 will use EUV over 20 layers.

Nextbigfuture covered Taiwan Semiconductors plans from 2017. TSMC has kept on track with its Moore’s Law roadmap to reach 3-nanometer chips by 2022.

Intel 7nm process is four years behind TSMC.

TSMC’s 2nm process will use a different transistor design. This design is called Multi-Bridge Channel Field Effect (MBCFET) transistor. It adds on to the previous FinFET designs.

The 2 nanometer process will start trial production in mid-2023 and mass production in 2024.

TSMC expects yield rate for its 2nm process node to reach a staggering 90% in 2023. Samsung will use MBCFET for 3 nanometer. Samsung expects the 3nm transistors to reduce power consumption by 30% and 45% respectively and improve performance by 30% over 7nm designs.

SOURCES- Taiwan News, Seeking Alpha, WCCFtech
Written by Brian Wang, Nextbigfuture.com

26 thoughts on “Taiwan Semiconductor Will Have Volume Production of 3 Nanometer Chips in 2022”

  1. Military might doesn't seem to work that well. Just ask the US. As for China growing, if you do an half decent job of governing, you can grow your economy well until you reach middle income. To go from middle income to high income that is a bit harder. We will see if China can make the transition.

  2. America is far more vulnerable if its infrastructures are destroyed, and your dismissal of Russia is laughable. Hope you’re not serious about it. They have enough nukes to destroy the US completely.
    Three gorges dam is far inland from the populated coastal region. If breached there will be plenty of warnings and the water would be widely dispersed in the many flood plains along the way, sure one harvest may be lost, the rice China grows is three crops per year … China has food reserves for more than two years, that can be stretched in dire emergency, US is much less able to cope. Almost all cities in China has nuclear shelters built mostly in times of tensions with Soviet Union, and recent developments incorporate them in the designs.

  3. Russia is a has-been and everyone knows it: shrinking population, low fertility rate, highest rate of TB in the world, one of the shortest life expectancies for males, one of the highest rates of HIV in the world. It can no longer afford its road-mobile launchers and will phase them out by 2024 leaving them with just 286 land-based ICBMs and their sea-based deterrent. The air-based leg is not survivable. In effect, their 144 SLBMs are their only viable deterrent. No need to even go into their conventional forces. In short, Russia has no business tangling with Europe let alone the U.S. or the combined forces of NATO.

    The U.S. has some of the lowest population density cities in the world. The public perception of a nuclear strike on a U.S. city is based on an HPAC model with very questionable assumptions: the entire city is built out of wood and the entire population is out in the open at the time of the explosion. For New York City this results in ~1,600,000 deaths. In reality, the nature of steel and concrete construction would provide significant protection to most of the residents of the city and fatalities would be limited to those within the ~20psi overpressure radius (~200,000 deaths, less if the warhead doesn't detonate in the optimum location). That's the worst case for any city in the U.S.

    Contrary to popular belief, you don't kill people with nukes. You kill people by destroying infrastructure with nukes. The Three Gorges Dam is the ultimate vulnerability.

  4. Actually … while that might be possible, I think more sobering is that 100% of China's (like Taiwan) super-high-tech chipmaking works are in what, 7 facilities? Something like that.  

    Conventional 2,000 pound ("910 kg") bombs, high-altitude dropped from stealth craft … each … could take out a chip-plant.  Remember, chip plants work by being hyper-clean and having extraordinary machine-to-machine precision. One big ol' bomb … well: like throwing buckets of sand into a transmission and engine. No good comes from it. 

    Even if a bombed facility is 'reconstructed', there are mountains of dust to deal with. I think such a target is rendered permanently inoperable.  

    FOR ONE, I do not advocate the Great Powers dirking their opponents' über-high-tech this way, unless there is a seriously believable near-term escalation toward a full-on nuclear war.  ONLY in that case would a SECOND STRIKE (a 'deëscalation strike') of the offending nations' high-tech manufacturing facilities would seem prudent. Technologically, "getting knee-capped" would have a very-low non-combatant casualty rate (moral high ground), but at the same time, would basically ruin the offending nations' ability to create high-tech devices.  

    Like the Allies bombing Germany's ball-bearing plants. Can't build airplanes, engines-of-war without ball bearings.  Same, it is, for refineries, chemical plants, electricity distribution substations, rail-yard switching terminals, and bridges n roads.

  5. If you have a big stick you do not have to use it
    Just show your opponent that you have it

    Just saying

    Lukkha

  6. True.
    But the Americans are notoriously cowards and sissies, this is why they will never attack first a country with a nuclear triad.
    MAD my dear kid.
    So no one will start a conflict
    And China will win as they are growing 3x times as fast as the USA.
    Now get out of the bed and go to school otherwise mama will punish you kid

    Just saying

    Lukkha

  7. I don’t think China is anxious to fight Taiwan. The national service in Taiwan is only four months and that is hardly a professional soldier made. China can just sits and wait, time is on her side.

  8. Over 95% of Americans live in cities while only half of Chinese live in cities. It’s rubbish that hundred thousands people will be affected from collapse of three gorges dam. Most of the populated regions are further down streams and there will be plenty of warnings for evacuation. One crop may be lost but that is not disastrous like the recent floods this year, the water would spread out much on the flat plains. Also, Russia will be the winner of any nuclear exchanges between China and the United States. Unless you want to have a go at them too. That will be the end of the world you know.

  9. I know they can play a lot of games with calling something "7nm" but if we're getting the energy savings and the performance gains there must be something to it.

    5-4-3-2 nm is still impressive even if 50% of it is marketing spin.

  10. Nuking Taiwan eliminates the reason for trying to acquire it. Ever wonder why nations don't nuke each other. Watch "War Games". The only winning move is not to play.

  11. A single, super-fused W86 detonated below the water line of the Three Gorges Dam in June or July would blow the dam and approximately twelve dams down stream killing 150-200 million people in the first 24 hours. About half of China's remaining population would starve to death in the subsequent year due to the collapse in agricultural output.

    If China were to commit all of its nuclear weapons to a strike on U.S. cities it would kill, at most, 10% of its population with no effect on agricultural output.

  12. You cant eliminate China' s retaliatory capability

    China was the first Asian country and the 5th globally to successfully design, build and commission a nuclear-powered submarine.
    The PLAN currently has two types of nuclear-powered attack submarines in service:
    Type 093 submarine (NATO designation Shang-class) – In service.
    Type 091 submarine (NATO designation Han-class) – In service.

  13. China has the nuclear triad (ICBMs, nuclear bombers and nuclear armed submarines) and second strike capability, kid.
    You strike China and half an hour later the United States of America cease to exist as a functioning country, kid.

    Just saying

    Lukkha

  14. The Chinese haven't fought a major conflict in about 70 yrs and they have not done an amphibious assault in their entire history. And this first would be against a nation where every single adult has had military training. The Chinese don't have the required throw weight to pulverized Taiwan's shore fortifications. It would be suicide for them to try.

  15. It would take a lot more nukes to try to eliminate China's retaliatory capacity. Probably wouldn't get it all.

  16. The US could take out 50% of China's industrial capacity with just 2 nukes.

    Just saying… watch out what you wish for.

  17. Density, power consumption, performance. These all matter and they are vaguely tied to "x nm" process name like iphone 12 is better than iphone 11. It is just a marketing name. Very much unlike how 180 nm was better than 250 nm; where this single measurement told you a lot about density, performance and power consumption.

Comments are closed.