![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
We are also approaching the limits of physics.
Two aspects is starting to work against smaller gates:
1. Quantum Tunneling. 14/16nm you are talking 10's, perhaps 100's of atom spacing between the gates, which is getting close to the point where quantum tunneling becomes a significant contributor to leakage, and this is completely unavoidable.
2. Laser, and its associated diffractive limit. To make masks with such a small spacing, you will need lasers with EUV, going into X-ray's, because light is incapable of being focused onto a single point due to effects of diffraction, which instead would create a spot. The longer the wavelength of the laser, the bigger that spot becomes. I have heard of TSMC actually need to 'cheat' their way through the masking process precisely because of that limitation. More powerful lasers requires better cooling due to the heat it creates when it hits the silicon, so the wafer, these days, are immersed in a cooling fluid for better cooling, that becomes impossible when you approach the very short wavelength because the liquid will scatter the laser (this is why morning sky appears blue, and dawn/dusk appear orange, blue light gets scattered by dust/water particles in the air, but red/orange light gets scattered less), and thus destroy the beam.
So all in all, we may see die shrink a few more times before physics says no.
Two aspects is starting to work against smaller gates:
1. Quantum Tunneling. 14/16nm you are talking 10's, perhaps 100's of atom spacing between the gates, which is getting close to the point where quantum tunneling becomes a significant contributor to leakage, and this is completely unavoidable.
2. Laser, and its associated diffractive limit. To make masks with such a small spacing, you will need lasers with EUV, going into X-ray's, because light is incapable of being focused onto a single point due to effects of diffraction, which instead would create a spot. The longer the wavelength of the laser, the bigger that spot becomes. I have heard of TSMC actually need to 'cheat' their way through the masking process precisely because of that limitation. More powerful lasers requires better cooling due to the heat it creates when it hits the silicon, so the wafer, these days, are immersed in a cooling fluid for better cooling, that becomes impossible when you approach the very short wavelength because the liquid will scatter the laser (this is why morning sky appears blue, and dawn/dusk appear orange, blue light gets scattered by dust/water particles in the air, but red/orange light gets scattered less), and thus destroy the beam.
So all in all, we may see die shrink a few more times before physics says no.
![Zarathustra[H]](https://cdn.hardforum.com/data/avatars/m/9/9298.jpg?1707758471){kind=avatar}
