제 18 호 Semiconductors
Kicker: SCIENCE
Semiconductors
By Jun-Seo Park, Cub-reporter
Recently, there has been an AI* craze due to chatGPT. It is clear that AI is a key technology that will lead the 4th industrial revolution. But there is a core technology that is no less important than AI, which is semiconductors. This is because semiconductors are essential to implementing AI. As semiconductors are important along with AI in the future, it will be very helpful to know basic knowledge about semiconductors. Therefore, in this article, let’s take a look at some basic principles of semiconductors and the relationship between semiconductors and AI.
* AI: Artificial Intelligence
What Is a Semiconductor?
Semiconductor is a compound word of ‘semi’ and ‘conductor’. Simply put, it refers to a material that is half-conductor. So, what is a conductor? The materials around us can be largely divided into insulators, conductors, and semiconductors. The biggest criterion for distinguishing these materials is electrical conductivity. First, insulators are materials with very low electrical conductivity and little electricity flows. These include glass, plastic, wood, etc. Next, since conductors have high electrical conductivity, electricity flows well. Materials corresponding to conductors include copper, iron, aluminum, gold, silver, etc.
Semiconductors
Finally, semiconductors, which are our concern, are materials whose electrical conductivity is intermediate between conductors and insulators. The most representative material used as semiconductor is silicon. To control the electrical conductivity of semiconductors, they go through an injecting process of impurity atoms called ‘doping’. It is difficult to artificially control the conditions because conductors’ current flow too easily and insulators’ current do not flow well. For this reason, semiconductors are important.
The Most Important Principle of Semiconductors: Energy-Band-Gap
Previously, it was described that the electrical conductivity of semiconductors is intermediate between conductors and insulators. Therefore, the most important characteristic of semiconductors is that semiconductors’ electrical conductivity can be controlled by users. Then, it is necessary to take a closer look at the principle of electrical conductivity of semiconductors, and here, the term of Energy-Band-Gap will be introduced. In order for the current to flow, the movement of electrons must occur. Insulators, conductors, and semiconductors all have energy levels at which electrons can exist inside the material, which are not continuous but discontinuous. It is said that having such a discontinuous energy level is ‘quantized’.
Energy Band Gap
Current flows only when electrons move from a relatively low energy band called a ‘valance band’ to a high energy band called a ‘conduction band’. In order for such movement of electrons to occur, light or thermal energy must be applied from outside. The distance between the valence band and the conduction band of conductor is very close. So, the movement of electrons easily occurs. Therefore, the current flows well. On the other hand, since insulators have a long distance between the valence band and the conduction band, even if energy is applied from the outside, electrons do not move well. So, current does not flow well. The semiconductors have an appropriate energy band gap and can be controlled by a user.
Types of Semiconductors: Why Are Semiconductors Essential to AI?
Semiconductors are used largely for two things. The first is processing of electrical signals, and the second is processing of data. The role of semiconductors related to AI is the processing of data. Data processing methods can be largely divided into storage and control. Semiconductors play a role in storing information, and semiconductors used to store information are called ‘memory semiconductors. Also, semiconductors used to perform numerical operations or control information are called ‘system semiconductors’. AI learns a lot of data and derives its inference from it. In order to receive and process data to be learned in a short time, computation and control must be performed. Therefore, system semiconductors are essential. The CPU** we have often heard of is a kind of system semiconductor.
** CPU: Central Processing Unit
The Reason Why NVIDIA Stock Prices Are Surging Recently: GPU’s Era
NVIDIA
NVIDIA’s stock price has been soaring recently. There is a good reason for this. In fact, GPU*** is much more important than CPU to implement AI such as chatGPT, and NVIDIA is a company that produces GPU almost exclusively. Therefore, while many companies struggle to dominate the AI market, NVIDIA has to supply GPU as their customers. Then let’s figure out why GPU is important to implement AI. Since CPU, the brain of a computer that we commonly use, processes data in a series one by one, it takes a long time to handle large scale data. In contrast, GPU is a parallel structure that processes multiple data at the same time. Therefore, it is appropriate to use GPU rather than CPU to implement AI.
*** GPU: Graphic Processing Unit
In the AI era, AI itself is important, but semiconductors are also essential to implement it. As you might already know, South Korea is globally competitive in semiconductors. To be exact, however, it is strong in the memory semiconductor field. In the field of system semiconductors, companies of the United States are superior to South Korean companies. Therefore, in order to be ‘really’ competitive in semiconductors, South Korea should strive to cultivate methods and manpower to be competitive in the field of system semiconductors.
Source:
https://www.aitimes.com/news/articleView.html?idxno=133691
Image sources:
https://www.etnews.com/20220513000100
https://www.quora.com/What-is-the-range-of-an-energy-band-gap-in-a-semiconductor
https://www.pcgamer.com/nvidias-fastest-next-gen-gpu-reportedly-enters-testing-phase/
