Researchers Create a Simpler Chip Fabrication Process

AlphaAtlas

[H]ard|Gawd
Staff member
Joined
Mar 3, 2018
Messages
1,713
Researchers from MIT and other institutions have reportedly come up with a new 2D circuit fabrication method that could be significantly cheaper than traditional chip making processes. Instead of progressively depositing multiple layers of different materials onto a disposable substrate, the researchers' technique can grow a 3 atom thick layer of semiconducting molybdenum disulfide onto existing substrates. Using water, sulfer, and oxygen plasma to deposit the materials , the method is said to be so "clean" that it preserves the original substrate for subsequent reuse, while the single pass should significantly reduce the chance of surface defects. A test FET transistor made on the process has a "spatial resolution of about 2 microns," which is orders of magnitude larger than the nanometer-scale transistors used in the silicon chips that loaded this web page, but with continued research and better equipment, the researchers hope to "shrink the pattern size, and directly integrate complex circuits on 2-D materials using their fabrication technique." Thanks to cageymaru for the tip.

Due to the weak interaction between MoS2 and the growth substrate, however, the researchers found they could detach the MoS2 cleanly from the original substrate by submerging it in water. This process, called "delamination," eliminates the need for using any supporting layer and produces a clean break with the material from the substrate. "That's why we can recycle it," Guo says. "After it's transferred, because it is purely clean, our patterned substrate is recovered and we can use it for multiple growths."
 
Let's jump to 3nm chips plz :p just skip 5nm.

We cant. The width of the valence shell of an Silicon atom is just about 0.2 to 0.3nm. Electron transfer from atom to atom will be sketchy if not possible at less than 5nm or so. That's why 5nm is probably are theoretical limit in process without some serious new tech.

The problem is that a transistor needs several atoms to be constructed. Anything smaller and we start to get weird quantum side effects. Silicon may reach its end as far as process size. That's why chiplets and 3d stacking have to be the next big leap.
 
Last edited:
Guys check out this new chip MIT just made.
ldgon2msb3dx6tz5esbw.jpg
 
As AMD has proved not every single piece of silicone needs to be 7+ or smaller.

Even if only used for some parts or some chips this is exciting -- better, of course, if there is a significant shrink.
 
We cant. The width of the valence shell of an Silicon atom is just about 0.2 to 0.3nm. Electron transfer from atom to atom will be sketchy if not possible at less than 5nm or so. That's why 5nm is probably are theoretical limit in process without some serious new tech.

The problem is that a transistor needs several atoms to be constructed. Anything smaller and we start to get weird quantum side effects. Silicon may reach its end as far as process size. That's why chiplets and 3d stacking have to be the next big leap.
Well, molyb disulphide isn't si, but I don't know how much better or worse it is as a semiconductor. I would assume worse, but they may find a better semiconductor with similar properties in the future.
 
Well, molyb disulphide isn't si, but I don't know how much better or worse it is as a semiconductor. I would assume worse, but they may find a better semiconductor with similar properties in the future.

Molybdenum is a far heavier transition metal than Si and add in two sulphur atoms and you create a far larger molecule than silicon.

Molybdenum disulphide has a far heavier atomic mass as a compound thus it cant possibly transfer electrons any smaller than it can now and it has a significantly larger atomic radius than a single Si atom.
 
Molybdenum is a far heavier transition metal than Si and add in two sulphur atoms and you create a far larger molecule than silicon.

Molybdenum disulphide has a far heavier atomic mass as a compound thus it cant possibly transfer electrons any smaller than it can now and it has a significantly larger atomic radius than a single Si atom.
Right, the current materials are lacking, but the process is interesting.
 
Right, the current materials are lacking, but the process is interesting.

Oh I agree and I also admit besides having 2.5 years of college chemistry which gives me fundamental and maybe a wee bit extra understanding of chemistry and matter, I emphatically emphasize I am NOT an engineer and do not pretend to know more about something I have zero qualifications to assume something that just isn't so. There is way more to the tech of chip building than I'll ever comprehend so maybe they will be able to go sub nanometer and I'll just eat crow lol.

I've been known to have a delicious foot I tend to dine on frequently in my mouth.
 
Oh I agree and I also admit besides having 2.5 years of college chemistry which gives me fundamental and maybe a wee bit extra understanding of chemistry and matter, I emphatically emphasize I am NOT an engineer and do not pretend to know more about something I have zero qualifications to assume something that just isn't so. There is way more to the tech of chip building than I'll ever comprehend so maybe they will be able to go sub nanometer and I'll just eat crow lol.

I've been known to have a delicious foot I tend to dine on frequently in my mouth.
Same, but only one semester of college chem. lol
 
Same, but only one semester of college chem. lol

And with that semester you already know more than 99% of the general public about matter. I am not downtalking anyone out there in the general public but seriously go ask literally anyone on the street if they know what a proton is and they will look at you with a confused gaze.

Science and facts are not taught in school anymore. Now it is just gender studies and classes on how to be more offended than the next offended student less offended than the one behind them.



and to this ...
Guys check out this new chip MIT just made.
View attachment 146124

Oh my god ... I actually laughed out loud at that thing. I reckon that if Intel doesn't do something to get away from monolithic design that is going to be the 128 core 14+++++++++++++++++++++++++++++++ chip in 2030.
 
Back
Top