Nowadays, with the continuous improvement of the chip manufacturing process, there can be more than 10 billion transistors in the chip. How are so many transistors installed?
1
When the chip is continuously enlarged, it looks like a huge city inside.
This is a Top-down View SEM photo. You can clearly see the layered structure inside the CPU. The line width becomes narrower as you go down, closer to the device layer.
This is a cross-sectional view of the CPU. You can clearly see the layered CPU structure. The chip is arranged in layers. This CPU has about 10 layers. The lowest layer is the device layer, which is the MOSFET transistor.
When the Mos tube is enlarged in the chip, a three-dimensional structure like a "podium" can be seen. The transistor has no inductance, resistance or other devices that are prone to heat generation. The top layer is a low-resistance electrode, which is separated from the platform below by an insulator. It generally uses P-type or N-type polysilicon as the raw material for the gate, and the insulator below is silicon dioxide.
The two sides of the platform are the source and the drain by adding impurities, and their positions can be interchanged. The distance between the two is the channel, and it is this distance that determines the characteristics of the chip.
Of course, the transistors in the chip are not only Mos tubes, but also tri-gate transistors. The transistors are not installed, but engraved during chip manufacturing.
When designing a chip, the chip designer will use EDA tools to plan the layout of the chip, and then route and route.
If we zoom in on the designed gate circuit, the white dots are the substrate, and some green borders are the doped layers.
The wafer foundry is manufactured according to the physical layout designed by the chip designer.
There are two trends in chip manufacturing. One is that wafers are getting bigger and bigger, so that more chips can be cut out to save efficiency. The other is the chip manufacturing process. The concept of manufacturing process is actually the size of the gate, which can also be called In the transistor structure, the current flows from the Source to the Drain, and the Gate (Gate) is equivalent to a gate, which is mainly responsible for controlling the on-off of the source and drain at both ends.
The current will be lost, and the width of the gate determines the loss when the current passes, which is manifested in the common heat generation and power consumption of mobile phones. The narrower the width, the lower the power consumption. The minimum width (gate length) of the gate is the manufacturing process.
The purpose of shrinking the nanometer process is to pack more transistors into a smaller chip, so that the chip will not become larger due to technological improvement.
But if we make the gate smaller, the faster the current will flow between the source and the drain, the more difficult the process will be.
The chip manufacturing process is divided into seven major production areas, which are diffusion, photolithography, etching, ion implantation, film growth, polishing, and metallization. Photolithography and etching are the two core steps.
Transistors are engraved by lithography and etching, and lithography is to make the circuits and functional areas required for chip production.
The light emitted by the photolithography machine is used to expose the sheet coated with photoresist through a photomask with a pattern. The role of the graph.
This is the role of lithography, similar to taking pictures with a camera. The photo taken by the camera is printed on the negative, and the lithography does not print the photo, but the circuit diagram and other electronic components.
Etching is the process of selectively removing unwanted material from the surface of a silicon wafer using chemical or physical methods. In the usual wafer processing flow, the etching process is located after the photolithography process, and the patterned photoresist layer will not be significantly eroded by the corrosion source during the etching, so as to complete the process step of pattern transfer. The etching process is a key step in replicating mask patterns.
picture
Among them, the material involved is photoresist. We need to know that the circuit design is first written on the photomask by laser, and then the light source is irradiated through the mask to the surface of the silicon wafer with photoresist, causing the exposure area The photoresist has a chemical effect, and then the exposed or unexposed area is dissolved and removed by developing technology, so that the circuit pattern on the mask is transferred to the photoresist, and finally the pattern is transferred to the silicon wafer by etching technology.
Photolithography is divided into two basic processes, positive photolithography and negative photolithography, according to the difference between positive and negative photolithography. In positive photolithography, the structure of the exposed part of the positive resist is destroyed and washed away by the solvent, so that the pattern on the photoresist is the same as the pattern on the mask.
Conversely, in negative-tone lithography, the exposed portion of the negative resist hardens and becomes insoluble, and the mask portion is washed away by the solvent, making the pattern on the photoresist the opposite of the pattern on the mask.
We can simply explain this step from a micro level.
A pre-made photoresist plate is covered on the wafer (or silicon wafer) coated with photoresist, and then the wafer is irradiated with ultraviolet rays for a certain period of time through the photoresist plate. The principle is to use ultraviolet rays to degrade part of the photoresist and make it easy to corrode.
Dissolving photoresist: The photoresist exposed to ultraviolet light in the photolithography process is dissolved away, and the pattern left after removal is consistent with that on the mask.
"Etching" means that after photolithography, the deteriorated part of the photoresist (positive resist) is etched away with an etching solution, and the surface of the wafer shows the pattern of the semiconductor device and its connection. Then use another etching solution to etch the wafer to form semiconductor devices and their circuits.
Removal of photoresist: After the etching is completed, the mission of the photoresist is declared complete, and the designed circuit pattern can be seen after all removal.
More than 10 billion transistors have been carved in this way, and transistors are used in a wide variety of digital and analog functions, including amplification, switching, voltage regulation, signal modulation and oscillators.
More transistors can increase the computing efficiency of the processor; moreover, reducing the size can also reduce power consumption; finally, after the chip is reduced in size, it is easier to plug it into a mobile device to meet the needs of future thinning and lightening.
Image Chip Transistor Cross Section
After 3nm, the current transistors are no longer suitable, and the semiconductor industry is currently developing nanosheet FETs (GAA FETs) and nanowire FETs (MBCFETs), which are considered the way forward for today's finFETs.
Samsung is betting on the GAA gate-around transistor technology, which TSMC has yet to release specific process details. Samsung first announced the GAA surround gate transistor in 2019. According to Samsung's official statement, based on the new GAA transistor structure, Samsung manufactured MBCFET (Multi-Bridge-Channel FET, multi-bridge-channel field effect transistor) by using nanosheet devices. ), which can significantly enhance transistor performance and replace FinFET transistor technology.
picture
In addition, MBCFET technology is also compatible with existing FinFET manufacturing process technology and equipment, thereby accelerating process development and production.
2
