Moore's Law "Continues".
For a long time, silicon has been an inexhaustible basic material for the semiconductor industry, and now we simply can't imagine life without silicon-based chips.
The iteration of the silicon-based semiconductor era basically follows the "Moore's Law", but with the passage of time, the failure of Moore's Law has become inevitable. In 1965, Gordon Moore, one of the founders of Intel, said in Electronics magazine that the number of transistors per square inch of silicon wafers would double every 12 months, which was the predecessor of Moore's Law; Ten years later, Moore adjusted the time for doubling integrated circuits from 12 months to 24 months; The last accurate timeline was set at "doubling every 18 months".
Over the past half century, Moore's Law has led the rapid development of the semiconductor industry, and TSMC is now even conquering the 2nm process. However, the semiconductor industry also needs to respect the laws of physics, that is, chips cannot shrink indefinitely, and Moore's Law will be forced to the end sooner or later.
The consensus in the industry is that when the silicon-based chip touches below 10nm, it will be limited by materials, devices, etc., and the chip will have problems such as leakage effect and short channel effect caused by quantum tunneling - in other words, the current is not easy to control, easy to error, and the current control error, so how to represent 0 and 1?
In order to continue and expand Moore's Law, and to meet the computing power needs of the new era, the industry has begun to explore the possibilities of new semiconductor materials. Carbon-based materials, which are also abundant, are getting attention.
Among all carbon-based materials, the research and application of graphene semiconductors may be more worthy of attention. Not long ago, the team of Professor Ma Lei of Tianjin University and the team of Professor Walter de Heer of Georgia Institute of Technology announced the development of the world's first functional semiconductor made of graphene materials. This means that 20 years after the birth of graphene materials, a breakthrough from 0 to 1 has finally been achieved in the field of semiconductors.
Why is graphene, as a "magic material", suitable for the integrated circuit industry? And what is hindering the application of graphene semiconductors? This starts with the material properties of graphene.
Achilles' heel of the magic material
In 2004, two physics teams from the University of Manchester and the Chernogolovka Institute of Microelectronics Technology worked together to create a graphene material in a crude but effective way – using tape that could be found in stationery stores, folding and pasting it on both sides of a graphite sheet, then tearing it apart, pasting it again, tearing it apart, repeating the process until it was made up of a completely new material made up of only one layer of carbon atoms.
This discovery not only brought the two scientists the Nobel Prize in physics, but also opened up a new direction in materials science: two-dimensional materials. Although graphite is "soft and weak", graphene has completely different qualities, such as being transparent, harder than steel, electrically conductive more than copper and silver, and thermal conductivity is the top echelon ...... known materials
Two-dimensional material graphene
These fascinating material qualities are why graphene is more suitable for transistors and wires. Zhang Yuanbo, a professor at the Department of Physics at Fudan University, once explained that silicon is like a sponge ball, because its surface always reacts with oxygen to form silicon oxide, so the transistor made cannot be infinitely small, which will affect the performance; Graphene, on the other hand, is a thin sheet of graphite composed of a single layer of carbon atoms, but despite its thinness, it remains stable because it does not react with the outside world.
Theoretically, if you make a transistor out of graphene, you can make electrons really go their own way at the chip level without interfering with each other. At the same time, as a very good thermoelectric material, it can turn waste heat into electrical energy, and can also better dissipate heat, thereby improving the stability and life of the chip. From a manufacturing point of view, the production process of carbon-based chips does not require high temperature and high pressure, not to mention smaller and lighter chips, and we can also get higher computing speeds than existing silicon-based chips.
However, theory is always an idealized formulation, and graphene, like any living thing in nature, will not be perfect. The most fatal drawback of graphene in semiconductors is that there is no "band gap".
From metal to semiconductor, hard fabrication?
The "band gap", also known as the band gap or energy gap, refers to the energy separation between the valence band (the band with lower electron energy) and the conduction band (the band with higher electron energy) in a solid material. If the band gap is small or non-existent, then the electrons can jump directly from the valence band to the conduction band at room temperature, and this solid will exhibit good electrical conductivity, which is the material we classify as metals; If the band gap is large and it is difficult for electrons to jump from the valence band to the conduction band, the solid exhibits insulation, and such a solid is called an "insulator"; If the band gap is moderate, it can conduct electricity under certain external conditions, but not under other conditions, which is what we often call "semiconductors".
Qualified semiconductors are stimulated by the outside world, and the excited electrons will jump from one energy band to another, which can effectively turn the current on and off, thus controlling the conductive switch, and creating a binary system using 0s and 1s in digital computers. Previous studies have found that graphene is a zero-bandgap material, and the folds, mounds and holes on the graphene sheet will have an unusual effect on the current, resulting in its conductivity at the Fermi level not falling to zero like ordinary semiconductors, but reaching a minimum value. This means that it cannot represent 0s and 1s by powering on and off like silicon, and it cannot be made into transistors.
The greatest value of the graphene functional semiconductors jointly developed by the above-mentioned Chinese and American research teams is that they open the energy gap of graphene.
From the team's ** published in Nature, we can roughly understand how this energy gap appears. The researchers took advantage of the high evaporation rate of silicon atoms compared to carbon atoms to heat silicon carbide wafers; After the silicon evaporates, a graphene multilayer is left on the surface of the crystal.
The first layer of this graphene polylayer is an insulating "epitaxial graphene" layer, which is equivalent to a layer of graphene that "grows" on a silicon carbide crystal, and it is bonded to the carbon atoms on the crystal. The interaction between atoms makes the energy band of this layer of epitaxial graphene produce **, and an energy band gap is created.
The researchers say the graphene semiconductor is 10 times more mobile than silicon carbide. The higher the mobility, the faster the transistor that controls the current switch, which can be switched faster and more efficiently, "it's like the difference between driving on a gravel road and driving on a highway."
Although the first step has been taken to practical use, from the experience of other new materials, it is still 10 to 20 years before graphene semiconductors can really enter thousands of households, after all, it has taken a hundred years for silicon to penetrate into various fields.