Integrated Circuits

Type of physical science: Classical physics

Field of study: Electromagnetism

In an integrated circuit, a complete electronic circuit is accomplished on a single substrate. Such circuits are generally small, highly reliable, inexpensive, lightweight, and suited for large-scale production.

Overview

In an integrated circuit, all basic components of an electronic circuit, such as resistors, capacitors, transistors, diodes, and their interconnections, are generally accomplished on a single substrate. When the substrate is a single chip of a semiconductor material such as silicon crystal, it is called a monolithic integrated circuit. The other types of integrated circuits are thick and thin film integrated circuits, where generally passive components such as resistors and capacitors are deposited as films on a single substrate. In general, the overall size of a chip is less than a few thousandths of a centimeter. Until the 1950's, all electronic circuits were designed using vacuum tubes. With the invention of the transistor in the 1950's, almost all electronic circuits--especially those that can be operated at low power levels, such as those in a radio--are designed using transistors. When computers were developed for scientific computation and data processing, even with replacement of vacuum tubes by discrete transistors, the systems were complex and bulky.

A need existed to miniaturize the electronic circuits in order to simplify building large-scale computers. Thus, a desire to build these computers and reliable production of complex circuits in all applications gave scientists a strong motivation to develop integrated circuits. The advantages of developing integrated circuits, besides miniaturization, are their high reliability; light weight, especially for aerospace applications; and suitability for production of a large number of circuits at low cost.

To build an integrated circuit, it is necessary to know how to assemble many resistors, capacitors, transistors, and diodes on a small chip of silicon crystal with great accuracy. Building an integrated circuit is called integrated circuit processing and requires several carefully designed steps; some of these steps can be achieved in more than one way. Development of these processing techniques requires a good understanding of the basic science behind processing, especially chemistry. With this knowledge, it may be possible to develop a technology for a highly reliable, repeatable, and high-yielding integrated circuit process. The key steps in processing an integrated circuit are preparation of semiconductor wafers, epitaxial growth, masking and etching, diffusion of impurities, oxidation, and isolation of devices. It is also necessary to know how to interconnect all components and package them. Andrew S. Grove in his PHYSICS AND TECHNOLOGY OF SEMICONDUCTOR DEVICES (1967) details some of the key processing steps of silicon devices.

The role of dopants in semiconductors is important in integrated circuit processing technology. Pure silicon by itself is a poor conductor. To improve its conductivity, impurities are introduced into it in a controlled fashion, namely, dopant elements such as arsenic, boron, and phosphorus. Phosphorus has five electrons in its outermost orbit, and it can share four of these electrons with silicon to make the remaining fifth electron loosely bound to it. This fifth electron can become free easily at room temperature under thermal excitation. When a large number of phosphorus atoms are introduced into silicon, there will be a large number of free electrons (negative charges). In this case, silicon has become an "n-type" material after doping with phosphorus. These electrons can move and constitute a current in the presence of an applied voltage. On the other hand, boron has three electrons in its outermost orbit. When a large number of boron atoms are introduced into silicon, each boron atom shares its electrons with the silicon atoms, which have four electrons in their outermost orbits. Thus, for each boron atom, one unfilled position is left in its outermost orbit. These vacancies are called holes, and they drift in the presence of an applied electric field (actually, electrons move away). In this case, because of the presence of holes, silicon has become a "p-type" material after doping with boron. The elements such as boron, phosphorus, and other elements of groups III and V, which convert silicon from a low conductivity to a high conductivity (depending on impurity concentration) material, are called dopants.

For monolithic integrated circuits, commonly used substrates are silicon, germanium, and gallium arsenide. Silicon, by far, has been the most frequently used substrate, especially for digital circuits and computers, since silicon can be oxidized easily to form an insulating layer of silicon dioxide. This layer has two useful roles. Since hydrofluoric acid can dissolve silicon dioxide but not silicon, selected regions of silicon dioxide are etched away to expose silicon. To do so, the silicon dioxide layer in other regions is protected with a photosensitive polymer film exposed to photographic masks. Dopant atoms can be diffused into the etched-away regions by reactive gases or can be implanted into them by high-energy ion beams. Silicon dioxide protects the underlying silicon from being doped.

Silicon required for production of integrated circuits must be very pure; there cannot be more than one impurity atom in about 1023 atoms of total silicon material. Such a pure silicon is obtained through several steps. In a very high-temperature oven, silicon dioxide and carbon interact to produce elemental silicon. This silicon is converted to trichlorosilane, and through distillation, pure trichlorosilane is separated from all other chlorine compounds. Using hydrogen, it is reduced to pure silicon consisting of randomly oriented crystals. As a result of random orientation of the crystals, it is called polycrystalline. Then, using either the Czochraski or the flat-zone method, these crystals can be reoriented into a regular arrangement to produce single crystalline silicon. In the Czochraski method, polycrystalline is melted in a crucible, and a seed crystal is lowered and then pulled out slowly from the melt. The melt adheres to the seed crystal and cools to form a large crystal. In the flat-zone method, a polycrystalline rod in contact with a seed crystal is held vertically and rotated while a melting zone traverses it slowly from the top to the bottom, part by part. The radio frequency heating in the melting zone segregates and evaporates the impurities, leaving pure silicon. The single crystal is now cut into thin circular slices called wafers, using a diamond saw, and polished on one side through etching and chemicals to be defect-free. The wafer can be broken into dice along specific orientations, which are marked on the wafer.

The next step is to deposit an insulating layer of silicon oxide on silicon, which can be achieved in several ways. Yet, the most common method is exposing the silicon surface to dry oxygen or water vapor in a thermal oven for several hours. The normal thickness of the insulating layer is 20 nanometers to 1 micrometer. (One nanometer is a billionth of a meter and a micrometer is a millionth of a meter.)

The silicon dioxide layer is now covered with a light-sensitive polymer resist. A photographic mask, which is transparent for the regions where silicon dioxide must be etched away in the subsequent steps, is placed on the wafer and exposed to ultraviolet or near-ultraviolet light. The polymer changes its structure in the exposed areas and can be dissolved away using trichloroethylene. That leaves a resist pattern, which is the same as the pattern on the mask. This process is called photolithography. There are other techniques, such as electron beam or X-ray lithography, to which one can resort to achieve the required minimum feature of the integrated circuit. The next step is to etch away the silicon dioxide in the exposed areas with hydrofluoric acid, leaving a silicon surface in those areas. This etching can be done by exposing the photoresist to gases containing reactive gases that contain fluorine.

The next step is adding dopants or impurities such as boron, phosphorus, and arsenic to the silicon. The dopants are first ionized and then accelerated in a high electric field to strike the exposed silicon surface and penetrate into the silicon to a depth of a few micrometers. The number of dopant atoms varies from 1011 to 1016 per square centimeter. Often, a specific type of variation of dopant atoms in the silicon is required. This distribution can be controlled through this initial implantation of dopants. If heated to high temperatures, the dopant atoms can be moved from regions of high concentration to low concentration through diffusion, a property that allows atoms from the regions of high density tend to move to regions of low density. Another method to introduce dopants is to expose the silicon surface to a gas containing the desired impurity and control the gas flow rate. Further diffusion can be achieved through heating.

A lightly doped layer can be grown over a heavily doped layer of silicon through a process called epitaxy. To grow an epitaxial layer, silane or silicon tetrachloride is passed over a silicon surface in a heated chamber. The gas decomposes, leaving a layer of silicon on the surface. When the wafer is heated, the silicon atoms are forced to form a covalent bond with the substrate. This epitaxially grown layer is less doped than the substrate. This process of using reactive gases is called chemical vapor deposition, and it can also be used for deposition of polycrystalline and insulating films such as silicon dioxide.

Electronic components can be fabricated in a substrate. For example, a resistor can be constructed by diffusing into a uniformly doped silicon wafer, an impurity of opposite conductivity over a precalculated area and depth into the substrate. This type of impurity also isolates the resistor from the substrate. The ends of the resistors are heavily doped and metallized on the surface for connection. The resistors may take on a sheet or serpentine pattern. To fabricate a capacitor, a thin film conducting material is deposited on the substrate as the first electrode. A dielectric film is placed on the first electrode and then a conductor material as the second electrode is deposited.

To fabricate a p-n junction, an n-type epitaxial layer is grown over a p-type substrate.

This n-type layer isolates different devices in the chip. As a result, a p-type region is created within the n-type epitaxial layer near the top. A heavily doped n-layer is diffused over a small central part of this p-layer near the surface. Together, the n-layer and the p-layer form a p-n junction. Metallic contacts are now deposited at proper places for connections. The first n-layer, the p-region, and the heavily doped n-region can be used to form an n-p-n transistor. To fabricate a metal-oxide-semiconductor field effect transistor (MOSFET), which is the most common device used in silicon chips, a p-type substrate is taken and two nontouching n-regions are diffused into it near the surface. Between these regions and under the surface, there is still p-type material. For this region, a silicon oxide layer is deposited on the surface. Metallic contacts are deposited over the silicon dioxide layer (gate) and the n-regions (source and drain) for connecting voltages. When suitable polarities for connecting voltages are used, the electrons can be allowed either to flow from source to drain in the surface under the gate or can be cut off.

The different components and devices deposited on the silicon should be connected through metallizations. The silicon dioxide is removed from the silicon surface where metallic connections are to be made. A metal is converted into vapor using electron bombardment or sputtering of solid metal, and the resulting vapor is then deposited on the surface. The metal in unwanted areas can be removed by using lithography or etching by phosphoric acid solutions.

Applications

Integrated circuits are used everwhere in day-to-day life. The main thrust for this technological development has been designing large-scale, superfast computers for scientific computation and control.

In digital computers, circuits such as inverters, NAND, NOR, and compound gate multiplexers and memory circuits--which perform digital logic (1 or 0)--are used. The subsystem designs in a digital computer include counters, random access memory (static RAM and dynamic RAM), adders, data flow paths, and logic arrays. Thus, much of the integrated circuit technology is oriented toward realizing these digital circuits in the most optimum fashion. Such optimization can be realized by using proper models for both passive and active devices after fabrication in a given technology. Typical characterization, modeling, and performance studies include estimation of capacitance; switching characteristics such as rise, fall, and delay times; power consumption; scaling; and voltage-current characteristics. Over the years, experience has shown that despite the increase in the complexity in the architecture and design of computer systems, the cost of realizing such computers has decreased by developing new integrated circuit technologies such as Bi-CMOS and CMOS to achieve Very Large Scale Integration (VLSI).

It should be noted that, generally, monolithic integrated circuits are used only when large-scale production is anticipated. In fact, development of personal computers useful for large-scale computation and programmable calculators is essentially a result of large-scale integration of electronic circuits. To enhance the speed of these systems, other types of semiconductor materials, such as gallium arsenide, are being explored.

In aerospace applications, the light weight of integrated circuits can be useful. They are cost-effective and provide high-speed response. Many aerospace systems use radio and microwave analog circuits beside digital circuits in their signal processing and communication instruments. Almost all digital systems on a commercial airplane contain integrated circuits, but the same cannot be said for analog circuits. A great need exists to develop analog microwave monolithic integrated circuits (MMIC) for communication and radar for aerospace applications.

Current efforts in MMIC work are using gallium arsenide as the substrate material. Gallium arsenide performs better at higher frequencies than silicon, since the speed with which electrons can move in gallium arsenide is much higher than that in silicon. Thus, many high-frequency active devices are being designed using gallium arsenide. Also, computers using gallium arsenide chips run faster than those with silicon chips. Successful design of MMICs will lead to miniaturization of circuits used in a variety of systems, from television to low-cost decoy missiles. These may also replace bulky wave guide technology of radars with some simple integrated technology, with the help of planar transmission lines. This area of research is important and is being pursued actively.

Integrated circuit technology is also widely used in analog circuit design, though not in a spectacular way. Analog integrated circuits tend to be thin and thick film integrated circuits.

These circuits are fabricated on an insulating substrate, such as rutile and alumina, but one of the major disadvantages of film technology is its inability to make active devices (such as transistors) on the film. Analog circuits are useful for developing a wide variety of resistors and capacitors. To incorporate active devices, a hybrid approach, whereby active device chips can be connected to the film circuit, is adopted. In one approach, the back of a chip is soldered to a metallized pad on the substrate and fine wires are used for connecting proper points on the top of the chip. Sometimes, beam leads that protrude beyond the edges of the chip are provided and can be connected to proper points on the thin or thick film. Another method uses solder-coated metal balls to mount the chip in an inverted position and provide connections to all needed points.

Almost all electronic measurement instruments, such as spectrum analyzers and oscilloscopes, have thin film and hybrid analog circuits. There are several other types of analog integrated circuits, such as operational amplifiers, which have a wide array of applications.

Automobile technology becomes increasingly electronic through use of equipment such as electronic fuel injection and collision avoidance that involve integrated circuits. They, too, are cost-effective. Integrated circuits continue to be a major part of communications and signal processing technology, which are being used in telephony for speech recognition and analysis and medical diagnostics. A promising future exists for integrated circuit design when fiber-optic communication links and integrated circuit technology are combined to develop modular optical multiplexers, switches, amplifiers, filters, and other optical communication components.

Context

The development of integrated circuit technology has been an evolving process. Each step in its development has depended on the success of previous steps. Transistors whose overall sizes are on the order of a few micrometers have been developed successfully, and integrated circuits of less than a centimeter with millions of devices and circuits have been processed and tested. Despite its great complexity, the technology has progressed very well.

The success of any electronics industry of any country in international competition depends on its ability to develop reliable semiconductor processing facilities. Electronics sales were expected to reach $400 billion in 1991, of which integrated circuit sales account for $50 billion. In the 1960's, integrated circuits were mostly bipolar, but since the 1970's, they are mostly digital metal-oxide-semiconductor integrated circuits. Chips that have about a million devices are available.

An important issue in the technology of integrated circuits is the type of device used in the design. This type of device classifies the processing technology into bipolar device technology, FET technology, MOSFET technology, and the like. Also, efforts are under way to replace silicon with gallium arsenide, as it can provide faster circuits. Since 1960, device dimensions in silicon technology have been reduced, on the average of 13 percent per year. Such a reduction will lead to faster switching times and less cost per chip.

Another closely related issue is the enormous effort needed to model and characterize active devices and passive components for a given technology. It is also important to model the process itself. There are computer modeling programs to simulate the integrated circuit processing; these programs will also give estimates for yield of process. Significant efforts have been made to develop computer-amenable models for passive and active devices. There are several computer-aided design programs, such as SPICE, which use these computer models of individual elements to analyze an extremely large circuit for its performance. Writing such computer programs, especially when the device behavior is nonlinear, is extremely difficult.

Nevertheless, good programs have been written and designs using them have matched the practical design behavior. Layout is another issue in integrated circuit technology. No draftsman can draw a figure showing the millions of circuit elements and their interconnections for processing engineers. To circumvent this difficulty, layout computer programs have been developed that help draftsmen draw the circuit on a paper for use in processing.

As a result of mass production of any given integrated circuit design, reliability and yield have become important criteria, since they affect production costs directly. Extensive research needs to be done in the area of packaging to reduce signal propagation delays, power dissipation, and parasitics on the chip. The trend in general electronics is to develop systems that have communication, computation, and control. Such systems tend to have both hardware and software components. To develop this technology, the National Research Council has recommended more research in specific areas, such as bipolar silicon logic, design and fabrication of submicron devices, three-dimensional devices, interconnect technology, and new magnetic and optical storage for computers and sensors.

Principal terms

ANALOG CIRCUITS: electronic circuits that amplify, generate, or operate on an input signal, consisting of sinusoidal waves

CRYSTAL: atoms arranged in a regular repeated pattern in many solids; crystals tend to exhibit morphology planes externally

DIGITAL CIRCUITS: electronic circuits, that can perform digital logic; examples are circuits such as OR, AND, NAND, multiplexers, and memory circuits; the basic building block is a circuit whose input and output states are characterized either as one or zero

DOPANTS: elements such as boron, phosphorus, and arsenic added to a pure semiconductor, such as silicon, to improve its electrical conductivity

MONOLITHIC INTEGRATED CIRCUIT: an electronic circuit that is fabricated entirely on a single large crystal

SEMICONDUCTORS: materials such as silicon, germanium, and gallium arsenide, whose electrical conductivity is not as good as that of metals such as copper but not as poor as that of insulators such as rubber

Bibliography

Elliott, David J. INTEGRATED CIRCUIT FABRICATION TECHNOLOGY. New York: McGraw-Hill, 1982. Provides a complete description of technology used to fabricate integrated circuits. Guidelines for the practical use of integrated circuit imaging equipment and detailed discussions on processing parameters and quality control are key features. Geared for both engineers and technicians.

Fink, Donald G., ed. ELECTRONICS ENGINEERS' HANDBOOK. New York: McGraw-Hill, 1975. Covers a variety of topics in electronics engineering. Of particular interest to integrated circuits are sections on properties of materials and discrete circuit components. Section 8 is on integrated circuits and provides an exhaustive review of all aspects of integrated circuits, including packaging and assembly.

Fogiel, Max. MODERN MICROELECTRONICS. New York: Research and Education Association, 1972. An exhaustive treatment of almost all topics of integrated circuit technology. Although somewhat dated, it gives several details in processing not found elsewhere in a single volume. Presentation is very clear, with numerous figures.

Grove, Andrew S. PHYSICS AND TECHNOLOGY OF SEMICONDUCTOR DEVICES. New York: John Wiley & Sons, 1967. An introductory book on semiconductor devices. Grove emphasizes the physics principle in great detail, while maintaining clarity. In the first three chapters, vapor-phase growth, thermal oxidation and solid-state diffusion, and useful design information for integrated circuit processing are discussed.

Howes, M. J., and D. V. Morgan. LARGE SCALE INTEGRATION. New York: John Wiley & Sons, 1981. A collection of lectures on a variety of aspects of large-scale integration. After a review of the progress to date and future developments, the authors present bipolar and FET device technology. Good emphasis on applications through chapters on design and testing of large-scale integration circuits, memory design, and custom design.

Muller, Richard S., and Theodore I. Kamins. DEVICE ELECTRONICS FOR INTEGRATED CIRCUITS. New York: John Wiley & Sons, 1977. Geared for a first course on electron devices for serious students of integrated circuits. After providing physics of semiconductor materials, the authors discuss silicon processing technology in detail. The remainder of the book is on modeling and physics of related device topics, such as p-n junctions, bipolar transistors, MOSFETs, and other devices. A popular book for undergraduates in physics and engineering. Chapter 2 is especially relevant.

Sze, S. M. VLSI TECHNOLOGY. New York: McGraw-Hill, 1983. A good book covering developments on processing techniques. Emphasizes VLSI technology and discusses process simulation and diagnostic techniques as well.

Conductors and Resistors

Insulators and Dielectrics