Silicon is the raw material most often used in integrated circuit (IC) fabrication. It is the second most abundant substance on the earth. It is extracted from rocks and common beach sand and put through an exhaustive purification process. In this form, silicon is the purist industrial substance that man produces, with impurities comprising less than one part in a billion. That is the equivalent of one tennis ball in a string of golf balls stretching from the earth to the moon. Semiconductors are usually materials which have energy-band gaps smaller than 2eV. An important property of semiconductors is the ability to change their resistivity over several orders of magnitude by doping. Semiconductors have electrical resistivity between 10-5 and 107 ohms. Semiconductors can be crystalline or amorphous. Elemental semiconductors are simple-element semiconductor materials such as silicon or germanium.
Silicon is the most common semiconductor material used today. It is used for diodes, transistors, integrated circuits, memories, infrared detection and lenses, light-emitting diodes (LED), photosensors, strain gages, solar cells, charge transfer devices, radiation detectors and a variety of other devices. Silicon belongs to the group IV in the periodic table. It is a grey brittle material with a diamond cubic structure. Silicon is conventionally doped with Phosphorus, Arsenic and Antimony and Boron, Aluminum, and Gallium acceptors. The energy gap of silicon is 1.1 eV. This value permits the operation of silicon semiconductors devices at higher temperatures than germanium. In the early 1900’s before integrated circuits and silicon chips were invented, computers and radios were made with vacuum tubes. The vacuum tube was invented in 1906 by Dr.Lee DeForest. Throughout the first half of the 20th century, vacuum tubes were used to conduct, modulate and amplify electrical signals. They made possible a variety of new products including the radio and the computer. However vacuum tubes had some inherent problems. They were bulky, delicate and expensive, consumed a great deal of power, took time to warm up, got very hot, and eventually burned out. The first digital computer contained 18,000 vacuum tubes, weighed 50 tins, and required 140 kilowatts of power. By the 1930’s, researchers at the Bell Telephone Laboratories were looking for a replacement for the vacuum tube. They began studying the electrical properties of semiconductors which are non-metallic substances, such as silicon, that are neither conductors of electricity, like metal, nor insulators like wood, but whose electrical properties lie between these extremes. By 1947 the transistor was invented. The Bell Labs research team sought a way of directly altering the electrical properties of semiconductor material. They learned they could change and control these properties by “doping” the semiconductor, or infusing it with selected elements, heated to a gaseous phase. When the semiconductor was also heated, atoms from the gases would seep into it and modify its pure, crystal structure by displacing some atoms. Because these dopant atoms had different amount of electrons than the semiconductor atoms, they formed conductive paths. If the dopant atoms had more electrons than the semiconductor atoms, the doped regions were called n-type to signify and excess of negative charge. Less electrons, or an excess of positive charge, created p-type regions. By allowing this dopant to take place in carefully delineated areas on the surface of the semiconductor, p-type regions could be created within n-type regions, and vice-versa. The transistor was much smaller than the vacuum tube, did not get very hot, and did not require a headed filament that would eventually burn out. Finally in 1958, integrated circuits were invented. By the mid 1950’s, the first commercial transistors were being shipped. However research continued. The scientist began to think that if one transistor could be built within one solid piece of semiconductor material, why not multiple transistors or even an entire circuit. Within a few years this speculation became one solid piece of material. These integrated circuits(ICs) reduced the number of electrical interconnections required in a piece of electronic equipment, thus increasing reliability and speed. In contrast, the first digital electronic computer built with 18,000 vacuum tubes and weighed 50 tons, cost about 1 million, required 140 kilowatts of power, and occupied an entire room.
Today, a complete computer, fabricated within a single piece of silicon the size of a child’s fingernail, cost only about $10.00. Before the IC is actually created a large scale drawing, about 400 times larger than the actual size is created. It takes approximately one year to create an integrated circuit. Then they have to make a mask. Depending on the level of complexity, an IC will require from 5 to 18 different glass masks, or “work plates” to create the layers of circuit patterns that must be transferred to the surface of a silicon wafer. Mask-making begins with an electron-beam exposure system called MEBES. MEBES translates the digitized data from the pattern generating tape into physical form by shooting an intense beam of electrons at a chemically coated glass plate. The result is a precise rendering, in its exact size, of a single circuit layer, often less than one-quarter inch square. Working with incredible precision , it can produce a line one- sixtieth the width of a human hair. After purification, molten silicon is doped, to give it a specific electrical characteristic. Then it is grown as a crystal into a cylindrical ingot. A diamond saw is used to slice the ingot into thin, circular wafers which are then polished to a perfect mirror finish mechanically and chemically. At this point IC fabrication is ready to begin. To begin the fabrication process, a silicon wafer (p-type, in this case) is loaded into a 1200 C furnace through which pure oxygen flows. The end result is an added layer of silicon dioxide (SiO2), “grown” on the surface of the wafer. The oxidized wafer is then coated with photoresist, a light-sensitive, honey-like emulsion. In this case we use a negative resist that hardens when exposed to ultra-violet light. To transfer the first layer of circuit patterns, the appropriate glass mask is placed directly over the wafer. In a machine much like a very precise photographic enlarger, an ultraviolet light is projected through the mask. The dark pattern on the mask conceals the wafer beneath it, allowing the photoresist to stay soft; but in all other areas, where light passes through the clear glass, the photoresist hardens. The wafer is then washed in a solvent that removes the soft photoresist, but leaves the hardened photoresist on the wafer. Where the photoresist was removed, the oxide layer is exposed. An etching bath removes this exposed oxide, as well as the remaining photoresist. What remains is a stencil of the mask pattern, in the form of minute channels of oxide and silicon.
The wafer is placed in a diffusion furnace which will be filled with gaseous compounds (all n- type dopants), for a process known as impurity doping. In the hot furnace, the dopant atoms enter the areas of exposed silicon, forming a pattern of n-type material. An etching bath removes the remaining oxide, and a new layer of silicon (n-) is deposited onto the wafer. The first layer of the chip is now complete, and the masking process begins again: a new layer of oxide is grown, the wafer is coated with photoresist, the second mask pattern is exposed to the wafer, and the oxide is etched away to reveal new diffusion areas. The process is repeated for every mask – as many as 18 – needed to create a particular IC. Of critical importance here is the precise alignment of each mask over the wafer surface. It is out of alignment more than a fraction of a micrometer (one-millionth of a meter), the entire wafer is useless. During the last diffusion a layer of oxide is again grown over the water. Most of this oxide layer is left on the wafer to serve as an electrical insulator, and only small openings are etched through the oxide to expose circuit contact areas. To interconnect these areas, a thin layer of metal (usually aluminum) is deposited over the entire surface.
The metal dips down into the circuit contact areas, touching the silicon. Most of the surface metal is then etched away, leaving an interconnection pattern between the circuit elements. The final layer is “vapox”, or vapour-deposited-oxide, a glass-like material that protects the IC from contamination and damage. It, too, is etched away, but only above the “bonding pads”, the square aluminum areas to which wires will later be attached.