There are few things in the world as simple as sand, and perhaps none as complex as computer chips. Yet the simple element silicon in sand is the starting point for making the integrated circuits that power everything today, from supercomputers to cell phones to microwave ovens.
Turning sand into tiny devices with millions of components is an extraordinary feat of science and engineering that would have seemed impossible when the transistor was invented at Bell Labs in 1947.
More
Computerworld
QuickStudies
Silicon is a natural semiconductor. Under some conditions, it conducts electricity; under others, it acts as an insulator. Silicon's electrical properties can be altered by the addition of impurities, a process called doping. These characteristics make it an ideal material for making transistors, which are simple devices that amplify electrical signals. Transistors can also act as switches - on/off devices used in combination to represent the Boolean operators "and," "or" and "not."
Several types of microchips are made today. Microprocessors are logic chips that perform the computations inside most commercial computers. Memory chips store information. Digital signal processors convert between analog and digital signals (QuickLink: a2270). Application-specific integrated circuits are special-purpose chips used in things such as cars and appliances.
The Process
Chips are made in multibillion-dollar fabrication plants called fabs. Fabs melt and refine sand to produce 99.9999% pure single-crystal silicon ingots. Saws slice the ingots into wafers about as thick as a dime and several inches in diameter. The wafers are cleaned and polished, and each one is used to build multiple chips. These and subsequent steps are done in a "clean room" environment, where extensive precautions are taken to prevent contamination by dust and other foreign substances.
A nonconducting layer of silicon dioxide is grown or deposited on the surface of the silicon wafer, and that layer is covered with a photosensitive chemical called a photoresist.
The photoresist is exposed to ultraviolet light shined through a patterned plate, or "mask," which hardens the areas exposed to the light. Unexposed areas are then etched away by hot gasses to reveal the silicon dioxide base below. The base and the silicon layer below are further etched to varying depths.
The photoresist hardened by this process of photolithography is then stripped away, leaving a 3-D landscape on the chip that replicates the circuit design embodied in the mask. The electrical conductivity of certain parts of the chip can also be altered by doping them with chemicals under heat and pressure. Photolithography using different masks, followed by more etching and doping, can be repeated hundreds of times for the same chip, producing a more complex integrated circuit at each step.
To create conducting paths between the components etched into the chip, the entire chip is overlaid with a thin layer of metal - usually aluminum - and the lithography and etching process is used again to remove all but the thin conducting pathways. Sometimes several layers of conductors, separated by glass insulators, are laid down.
Each chip on the wafer is tested for correct performance and then separated from other chips on the wafer by a saw. Good chips are placed into the supporting packages that allow them to be plugged into circuit boards, and bad chips are marked and discarded.
See additional Computerworld QuickStudies