This enabled circuit operation below 3 V with an operating frequency of up to 25 kHz, which was constrained by parasitic capacitances
I would guess process improvements would help a lot towards lowering those parasitics. So I wouldn't take this initial attempt as a guide for ultimate speed.
Since this is 2D materials, a capacitor is a dielectric sandwiched by two conductors and capacitance scales linearly with area, I would assume just scaling things down would help immensely with parasitic capacitance. Changing materials or process could also change the dielectric constant which also affects the capacitance linearly.
Paper is sadly not open access, so I can't check if they mention this or have done some theoretical peak calculations or something. Would indeed be interesting to know.
> molybdenum disulfide for n-type transistors and tungsten diselenide for p-type transistors
Isn't this rather unusual?
NegativeK 12 days ago [-]
Yes? But it’s been in research for a decade or two, based on a quick search.
It’s confusing to me because moly d is a very common lubricant, even for home uses.
avmich 11 days ago [-]
Isn't it a good lubricant because it's easily split into 2D layers?
m-watson 11 days ago [-]
Something that is nice with MoS2 and the others are transition metal dichalcogenides and have some beneficial physical properties like a natural electronic bandgap, unlike silicon.
RayfromBoston 11 days ago [-]
I wonder how this compares in speed and capabilities to photonic computers
ConradKilroy 11 days ago [-]
I was wondering that too!
Valgrim 11 days ago [-]
Molybdenum and tungsten both have melting point much higher than silicon, Maybe these circuits could be a good candidate for Venus rovers?
kxndnddn 11 days ago [-]
I don't see how that would be relevant since the melting temperature of Silicon is already _significantly_ higher than temperatures on Venus can reach outside of reentry
znpy 11 days ago [-]
Isn’t tungsten much much more expensive than silicon and harder to work with?
IsTom 11 days ago [-]
Does its price really matter for amounts used in chips?
znpy 10 days ago [-]
i mean, can you imagine how many chips are built?
nowadays there's at least a chip in most physical objects...
IsTom 10 days ago [-]
And the part that is substrate, not the packaging is tiny and thin. It's much less than a gram of material in CPUs and even less in smaller chips. That's not going to be a significant part of final price. Also remember that silicon currently used is not some regular sand, but grown monocrystals that are a bit pricier.
11 days ago [-]
Razengan 12 days ago [-]
A small step towards Sophons
lowwave 12 days ago [-]
Well with all the sabre-rattling by Kratsios on space time control, Sophons is not that far fetched.
9dev 12 days ago [-]
Well—I, for one, welcome our new Trisolaran overlords!
l3x4ur1n 11 days ago [-]
Traitor
yodon 12 days ago [-]
WTF is up with that illustration at the top of the article?
DavidSJ 12 days ago [-]
Some attempt to visually represent molybdenum disulfide and tungsten diselenide with the keys of a QWERTY keyboard.
mjmas 12 days ago [-]
Which if it was done properly would have WSe2 and MoS2 rather than seemingly random keys
close04 11 days ago [-]
It shows just the symbols of the elements (W, Se, Mo) and the number 2, not the compounds. The "W", "S", "M", and "2" characters are in the correct place on a QWERTY keyboard, and they appended the necessary additional characters to complete the symbols as needed, even if the "e" in Se and "o" in Mo aren't in the correct spot on the layout.
gfody 12 days ago [-]
someone tries to explain cmos to the graphics dept
adastra22 12 days ago [-]
To an AI prompt more likely.
bobmcnamara 12 days ago [-]
If the frame is made of atoms what are the keys and display made out of? Quarks?
a3w 11 days ago [-]
Yupp, I stopped reading and closed the browser tab when I saw that. Then reconsidered, to find the original source.
TacticalCoder 12 days ago [-]
AI but it's kinda cool. Computers books in the old days used to have crazy representations of computers and all kinds of stuff. I don't mind this one.
https://en.wikipedia.org/wiki/One-instruction_set_computer
Modern microprocessor built from complementary carbon nanotube transistors https://www.nature.com/articles/s41586-019-1493-8
ie: https://sam.zeloof.xyz/category/semiconductor/
How high could this technique go?
This enabled circuit operation below 3 V with an operating frequency of up to 25 kHz, which was constrained by parasitic capacitances
I would guess process improvements would help a lot towards lowering those parasitics. So I wouldn't take this initial attempt as a guide for ultimate speed.
Since this is 2D materials, a capacitor is a dielectric sandwiched by two conductors and capacitance scales linearly with area, I would assume just scaling things down would help immensely with parasitic capacitance. Changing materials or process could also change the dielectric constant which also affects the capacitance linearly.
Paper is sadly not open access, so I can't check if they mention this or have done some theoretical peak calculations or something. Would indeed be interesting to know.
[1]: https://www.nature.com/articles/s41586-025-08963-7
It’s confusing to me because moly d is a very common lubricant, even for home uses.
nowadays there's at least a chip in most physical objects...