By growing two-dimensional crystals on a silicon wafer, MIT engineers may have laid the groundwork for a major paradigm shift.
In 1965, computer science was still a fledgling discipline. But that didn’t stop Gordon Earl Moore from formulating a very interesting hypothesis. He predicted that the number of transistors — the small subunits that are the basis for the operation of many electronic components — on a chip would increase exponentially on average every two years, with a significant increase in the power of key calculation. Over the years, this empirical observation has proven to be almost exact, and this prognosis has therefore passed to posterity under the name of Moore’s Law.
But Moore himself was well aware that this trend could not last forever. Specialists are now beginning to perceive its limits in a very concrete way, for several reasons.
The first concern is quite simply the manufacturing capacity. The miniaturization of components is a difficult exercise. At each stage, it becomes even more difficult to grab space and integrate these very small elements into the same architecture.
There are also thermodynamic limits; the more the size of the transistors is reduced to increase their number, the more difficult it becomes to cool them correctly. This is something that we already see with the latest generations of CPUs, which display insane thermal envelopes.
But the biggest pitfall is at the level of the architecture of the material itself. When the size of a transistor approaches that of an isolated atom, electrons stop behaving as expected. The silicon that composes them therefore tends to lose its electrical properties, and the chip can no longer function correctly.
Two sizes, endless possibilities
This is where the team of Jeehwan Kim and Ki Seok Kim, two laboratory managers at the prestigious Massachusetts Institute of Technology, come in. They are working on an approach that could overcome this fundamental limit: 2D materials.
All the objects you can lay your eyes on right now are made up of atoms packed into a three-dimensional lattice. A 2D material, on the other hand, is a kind of plate of atoms made up of a single layer. One can for example cite graphene, composed of a single stage of carbon atoms.
These materials have quite unique properties, and today they are the subject of particular interest in many cutting-edge scientific disciplines. One can for example cite twistronics. It is a young sub-discipline of optics that plays on the interactions of ultrafine materials to control light (see our article). THE property that interests MIT is their ability to conduct electricity with insane efficiency. They allow the passage of electrons with almost no resistance. It could therefore be a revolutionary alternative to current silicon transistors.
What push the industry beyond the limits of Moore’s law? On paper, yes. But that’s much harder said than done. ” It is traditionally considered almost impossible to grow 2D crystals on silicon “says Kim. Indeed, achieving a perfect layer of atoms is not easy.
Put things right
The most popular approach is to “peel” a layer of atoms from a raw material. But in the latter, the crystals sometimes grow in a somewhat anarchic way. The crystal structure then presents unacceptable imperfections in this context. It therefore takes a lot of time to achieve a workable result.
The researchers explain that to force the crystal to grow properly, they need to be grown on surfaces with very particular patterns of atoms — hexagons. However, if there are indeed some materials of this kind such as sapphire, this is not the case with silicon. A big problemknowing that all modern computing is built around this material.
To overcome this obstacle, they have developed a technique called ” non-monocrystalline growthepitaxial “. It allowed them to grow a 2D crystal on a standard silicon wafer, a medium commonly used by computer chip makers.
Concretely, they started by cultivating a crystal on a wafer, or wafer of silicon. They then covered it with a layer of silicon dioxide. At the atomic scale, this material presents sorts of pockets which make it possible to trap the branches of the crystal. This prevents it from pushing haphazardly in all directions; we therefore obtain a single layer of very clean and organized atoms.
A result as impressive as it was unexpected for the researchers. Kim even calls it “particularly shocking” – a choice of vocabulary that is anything but trivial in the academic world. “We observe the growth of a monocrystalline structure [très vulgairement, un cristal “parfait” où la structure est cohérente de bout en bout] whereas there is no structural relationship between the 2D material and the silicon sheet,” he explains.
An approach with immense potential
And above all, this method allowed them to produce a working transistor. The end result was incredibly fine, and featured performance equivalent to that of a conventional transistor. It is therefore a remarkable proof of concept which perhaps represents the beginning of a small revolution in electronics and computing.
“ Until now, we had no way to produce 2D materials in single crystal form on silicon sheets. The community had almost given up on the idea of using 2D materials for tomorrow’s processors “says Kim. ” Now we have completely solved this problem “, he rejoices before concluding on a sentence not piqued by the cockchafers. He believes that these works will changing the Moore’s Law paradigm ” – just that !
It will therefore be very interesting to follow the progress of this work. Obviously, this is only basic research for the moment, and developing a large-scale industrial process will be long, expensive and difficult. But if successful, the payoff could be phenomenal. And in any case, they show at least that the track of 2D materials remains very promising.