The integrated circuit is very complex, and requires a great deal of accuracy and cleanliness in order to produce a working chip the majority of the time. The proficiency of these processes have advanced greatly over the years because of increasingly clean working conditions, and new manufacturing processes. This is quite evident when comparing the clothes of the Intel workers in the late 1970's, and in the 1990's.
Although the clothes of the scientists manufacturing the chips has changed, the basic process used to manufacture these chips has stayed virtually the same. This manufacturing process is outlined below.
1. Pure silicon is extracted from sand, and formed into cylindrical crystals. These crystals are then grown in a vacuum. These cylinders are then cut into slices, or wafers, of approximately 1/2 of a millimetre thick, and ground absolutely smooth by a machine. Each wafer will yield several hundred chips.
2. Photomasks are then constructed. One photomask is created for each layer of the chip. These photomasks are approximately 10 cm2, and are extremely accurate pictures created by some photographic process, or by electrobeam lithography. Each photomask usually contains the design for several hundred chips side by side.
3. The wafers of silicon are then put in an extremely hot furnace, where they grow an insulating layer of a chemical known as silicon dioxide. These wafers are then coated with a soft, light-sensitive plastic known as photoresist. This process is repeated for every layer of the chip.
4. The circuit pattern is then formed by placing the photomask over the treated silicon, and then flooded with ultraviolet light. This process hardens the areas of the silicon that are not protected by the photomask, and leaves the areas that were covered relatively soft.
5. The soft photoresist is then stripped away by the use of various acids and solvents, leaving bare silicon in those areas.
6. "Dopents" are implanted into the bare silicon by the use of a machine that bombards the wafer with ions. These ions travel at a high speed, and hit the silicon with such impact that they embed themselves into the exposed areas of the silicon.
7. The components are then formed into the silicon, such as transistors.
8. After the components are formed, aluminium tracks are laid down. If the chip contains multiple layers, these tracks are insulated by a layer of silicon dioxide. These aluminium tracks are also defined by a mask. Now this wafer of chips is formed, and is ready for testing, and then packaging.
9. The chips are then tested for defects by a miniature probe, and the defective chips are marked with a red dot.
10. The wafer is then cut into the individual chips by the use of a diamond saw or by a laser.
11. The functional chips are then placed on a frame. These are then encased by a plastic cover, and then their pins are usually bent to be connected to the other components.
Because of this process, there are very few defects in the finished chips compared to manually connected components, and therefore are less prone to error. These chips can be mass-produced, making them affordable. Also, because the construction is done by precise robots, the chips can be microscopically small, thus requiring less power, and reducing the size of the computer drastically. This miniaturization is shown by the fact that a MPU such as a Pentium Pro 200 constructed in the fashion of the ENIAC would be about the size of a large office building, and would draw about 1/10 of the power that a major city consumes today.