I. Introduction

Microprocessor, electronic circuit that functions as the central processing unit (CPU) of a computer, providing computational control. Microprocessors are also used in other advanced electronic systems, such as computer printers, automobiles, and jet airliners.

The microprocessor is one type of ultra-large-scale integrated circuit. Integrated circuits, also known as microchips or chips, are complex electronic circuits consisting of extremely tiny components formed on a single, thin, flat piece of material known as a semiconductor. Modern microprocessors incorporate as many as 10 million transistors (which act as electronic amplifiers, oscillators, or, most commonly, switches), in addition to other components such as resistors, diodes, capacitors, and wires, all packed into an area about the size of a postage stamp.

A microprocessor consists of several different sections: The arithmetic/logic unit (ALU) performs calculations on numbers and makes logical decisions; the registers are special memory locations for storing temporary information much as a scratch pad does; the control unit deciphers programs; buses carry digital information throughout the chip and computer; and local memory supports on-chip computation. More complex microprocessors often contain other sections–such as sections of specialized memory, called cache memory, to speed up access to external data-storage devices. Modern microprocessors operate with bus widths of 64 bits (binary digits, or units of information represented as 1s and 0s), meaning that 64 bits of data can be transferred at the same time.

A crystal oscillator in the computer provides a clock signal to coordinate all activities of the microprocessor. The clock speed of the most advanced microprocessors is 1 gigahertz (GHz)–about 1 billion cycles per second–allowing billions of computer instructions to be executed every second.

II. Computer Memory

Because the microprocessor alone cannot accommodate the large amount of memory required to store program instructions and data, such as the text in a word-processing program, transistors can be used as memory elements in combination with the microprocessor. Separate integrated circuits, called random-access memory (RAM) chips, which contain large numbers of transistors, are used in conjunction with the microprocessor to provide the needed memory. There are different kinds of random-access memory. Static RAM (SRAM) holds information as long as power is turned on and is usually used as cache memory because it operates very quickly. Another type of memory, dynamic RAM (DRAM), is slower than SRAM and must be periodically refreshed with electricity or the information it holds is lost. DRAM is more economical than SRAM and serves as the main memory element in most computers.

III. Microcontroller

A microprocessor is not a complete computer. It does not contain large amounts of memory or have the ability to communicate with input devices–such as keyboards, joysticks, and mice–or with output devices, such as monitors and printers. A different kind of integrated circuit, a microcontroller, is a complete computer on a chip, containing all of the elements of the basic microprocessor along with other specialized functions. Microcontrollers are used in video games, videocassette recorders (VCRs), automobiles, and other machines.

IV. Semiconductors

All integrated circuits are fabricated from semiconductors, substances whose ability to conduct electricity ranks between that of a conductor and that of a nonconductor, or insulator. Silicon is the most common semiconductor material. Because the electrical conductivity of a semiconductor can change according to the voltage applied to it, transistors made from semiconductors act like tiny switches that turn electrical current on and off in just a few nanoseconds (billionths of a second). This capability enables a computer to perform many billions of simple instructions each second and to complete complex tasks quickly.

The basic building block of most semiconductor devices is the diode, a junction, or union, of negative-type (n-type) and positive-type (p-type) materials. The terms n-type and p-type refer to semiconducting materials that have been doped–that is, have had their electrical properties altered by the controlled addition of very small quantities of impurities such as boron or phosphorus. In a diode, current flows in only one direction: across the junction from the p- to n-type material, and then only when the p-type material is at a higher voltage than the n-type. The voltage applied to the diode to create this condition is called the forward bias. The opposite voltage, for which current will not flow, is called the reverse bias. An integrated circuit contains millions of p-n junctions, each serving a specific purpose within the millions of electronic circuit elements. Proper placement and biasing of p- and n-type regions restrict the electrical current to the correct paths and ensure the proper operation of the entire chip.

V. Transistors

The transistor used most commonly in the microelectronics industry is called a metal-oxide semiconductor field-effect transistor (MOSFET). It contains two n-type regions, called the source and the drain, with a p-type region in between them, called the channel. Over the channel is a thin layer of nonconductive silicon dioxide topped by another layer, called the gate. For electrons to flow from the source to the drain, a voltage (forward bias) must be applied to the gate. This causes the gate to act like a control switch, turning the MOSFET on and off and creating a logic gate that transmits digital 1s and 0s throughout the microprocessor.

VI. Construction of Microprocessors

Microprocessors are fabricated using techniques similar to those used for other integrated circuits, such as memory chips. Microprocessors generally have a more complex structure than do other chips, and their manufacture requires extremely precise techniques.

Economical manufacturing of microprocessors requires mass production. Several hundred dies, or circuit patterns, are created on the surface of a silicon wafer simultaneously. Microprocessors are constructed by a process of deposition and removal of conducting, insulating, and semiconducting materials one thin layer at a time until, after hundreds of separate steps, a complex sandwich is constructed that contains all the interconnected circuitry of the microprocessor. Only the outer surface of the silicon wafer–a layer about 10 microns (about 0.01 mm/0.0004 in) thick, or about one-tenth the thickness of a human hair–is used for the electronic circuit. The processing steps include substrate creation, oxidation, lithography, etching, ion implantation, and film deposition.

The first step in producing a microprocessor is the creation of an ultrapure silicon substrate, a silicon slice in the shape of a round wafer that is polished to a mirror-like smoothness. At present, the largest wafers used in industry are 300 mm (12 in) in diameter.

In the oxidation step, an electrically nonconducting layer, called a dielectric, is placed between each conductive layer on the wafer. The most important type of dielectric is silicon dioxide, which is "grown" by exposing the silicon wafer to oxygen in a furnace at about 1000°C (about 1800°F). The oxygen combines with the silicon to form a thin layer of oxide about 75 angstroms deep (an angstrom is one ten-billionth of a meter).

Nearly every layer that is deposited on the wafer must be patterned accurately into the shape of the transistors and other electronic elements. Usually this is done in a process known as photolithography, which is analogous to transforming the wafer into a piece of photographic film and projecting a picture of the circuit on it. A coating on the surface of the wafer, called the photoresist or resist, changes when exposed to light, making it easy to dissolve in a developing solution. These patterns are as small as 0.13 microns in size. Because the shortest wavelength of visible light is about 0.5 microns, short-wavelength ultraviolet light must be used to resolve the tiny details of the patterns. After photolithography, the wafer is etched–that is, the resist is removed from the wafer either by chemicals, in a process known as wet etching, or by exposure to a corrosive gas, called a plasma, in a special vacuum chamber.

In the next step of the process, ion implantation, also called doping, impurities such as boron and phosphorus are introduced into the silicon to alter its conductivity. This is accomplished by ionizing the boron or phosphorus atoms (stripping off one or two electrons) and propelling them at the wafer with an ion implanter at very high energies. The ions become embedded in the surface of the wafer.

The thin layers used to build up a microprocessor are referred to as films. In the final step of the process, the films are deposited using sputterers in which thin films are grown in a plasma; by means of evaporation, whereby the material is melted and then evaporated coating the wafer; or by means of chemical-vapor deposition, whereby the material condenses from a gas at low or atmospheric pressure. In each case, the film must be of high purity and its thickness must be controlled within a small fraction of a micron.

Microprocessor features are so small and precise that a single speck of dust can destroy an entire die. The rooms used for microprocessor creation are called clean rooms because the air in them is extremely well filtered and virtually free of dust. The purest of today's clean rooms are referred to as class 1, indicating that there is no more than one speck of dust per cubic foot of air. (For comparison, a typical home is class one million or so.)

VII. History of the Microprocessor

The first microprocessor was the Intel 4004, produced in 1971. Originally developed for a calculator, and revolutionary for its time, it contained 2,300 transistors on a 4-bit microprocessor that could perform only 60,000 operations per second. The first 8-bit microprocessor was the Intel 8008, developed in 1972 to run computer terminals. The Intel 8008 contained 3,300 transistors. The first truly general-purpose microprocessor, developed in 1974, was the 8-bit Intel 8080 (see Microprocessor, 8080), which contained 4,500 transistors and could execute 200,000 instructions per second. By 1989, 32-bit microprocessors containing 1.2 million transistors and capable of executing 20 million instructions per second had been introduced.

In the 1990s the number of transistors on microprocessors continued to double nearly every 18 months. The rate of change followed an early prediction made by American semiconductor pioneer Gordon Moore. In 1965 Moore predicted that the number of transistors on a computer chip would double every year, a prediction that has come to be known as Moore's Law. In the mid-1990s chips included the Intel Pentium Pro, containing 5.5 million transistors; the UltraSparc-II, by Sun Microsystems, containing 5.4 million transistors; the PowerPC620, developed jointly by Apple, IBM, and Motorola, containing 7 million transistors; and the Digital Equipment Corporation's Alpha 21164A, containing 9.3 million transistors. By the end of the decade microprocessors contained many millions of transistors, transferred 64 bits of data at once, and performed billions of instructions per second.

In 2000 chip manufacturer Advanced Micro Devices debuted a 1 GHz microprocessor, the fastest microprocessor ever mass-produced for personal computers. The high-speed processor contains approximately 22 million transistors.

Contributed By:

Gary H. Bernstein, B.S.E.E., M.S.E.E., Ph.D.

Associate Professor and Director of the Microelectronics Laboratory, Department of Electrical Engineering, University of Notre Dame.

See an outline for this article.


"Microprocessor," Microsoft® Encarta® Online Encyclopedia 2000

http://encarta.msn.com © 1997-2000 Microsoft Corporation. All rights reserved.


© 1993-2000 Microsoft Corporation.

All rights reserved.