Silicon exists widely in nature in the form of silicates or silicon dioxide in rocks and gravel. The manufacturing of silicon wafers can be summarized into three basic steps: silicon refining and purification, single crystal silicon growth, and wafer forming.
The first step is silicon purification. The raw sand and gravel are placed in an electric arc furnace at approximately 2000°C with a carbon source. At this high temperature, the carbon and silicon dioxide in the sand and gravel undergo a chemical reaction (carbon combines with oxygen, leaving silicon), yielding pure silicon with a purity of approximately 98%, also known as metallurgical-grade silicon. However, this is not pure enough for microelectronic devices because the electrical properties of semiconductor materials are highly sensitive to impurity concentrations. Therefore, the metallurgical-grade silicon undergoes further purification: the pulverized metallurgical-grade silicon is chlorinated with gaseous hydrogen chloride to generate liquid silane. Then, through distillation and chemical reduction processes, high-purity polycrystalline silicon is obtained, with a purity as high as 99.999999999%, becoming electronic-grade silicon.
Next is the growth of single-crystal silicon, the most commonly used method being the Czochralski method. As shown in the diagram, high-purity polycrystalline silicon is placed in a quartz crucible and continuously heated by a surrounding graphite heater, maintaining the temperature at approximately 1400°C. The air in the furnace is usually an inert gas, which melts the polycrystalline silicon without causing unwanted chemical reactions. To form single-crystal silicon, the crystal orientation needs to be controlled: the crucible, carrying the molten polycrystalline silicon, is rotated, a seed crystal is immersed in it, and a pulling rod pulls the seed crystal in the opposite direction while slowly and vertically pulling it upwards from the molten silicon. The molten polycrystalline silicon adheres to the bottom of the seed crystal and grows upwards according to the orientation of the seed crystal lattice. Therefore, the orientation of the grown crystal is determined by the seed crystal, and after being pulled out and cooled, it grows into a single-crystal silicon rod with the same lattice orientation as the seed crystal. After growth using the Czochralski method, the single crystal rod is cut to an appropriate size, then ground to remove the uneven cut marks, and finally polished using a chemical mechanical polishing process to make at least one side as smooth as a mirror, thus completing the wafer manufacturing process.
The diameter of a single-crystal silicon rod is determined by the pulling speed and rotation speed of the seed crystal. Generally, the slower the pulling speed, the larger the diameter of the grown single-crystal silicon rod. The thickness of the cut wafer is related to the diameter. Although the fabrication of semiconductor devices is only completed within a few micrometers of the top of the wafer, the wafer thickness generally needs to reach 1 mm to ensure sufficient mechanical stress support. Therefore, the wafer thickness increases with the increase of the diameter.
Wafer manufacturers melt polycrystalline silicon, plant seed crystals in the molten liquid, and then slowly pull them out to form cylindrical single-crystal silicon ingots. Since the silicon ingot is gradually formed from a seed crystal with a defined crystal orientation in molten silicon, this process is called "crystal growth." After being cut, tumbled, sliced, chamfered, polished, laser-etched, and packaged, the silicon ingot becomes the basic raw material for integrated circuit factories—silicon wafers, or "wafers."
