Also found in: Dictionary, Thesaurus, Acronyms.
Miniature electronic circuits produced within and upon a single semiconductor crystal, usually silicon. Integrated circuits range in complexity from simple logic circuits and amplifiers, about 1/20 in. (1.3 mm) square, to large-scale integrated circuits up to about 1/2 in. (12 mm) square. They can contain millions of transistors and other components that provide computer memory circuits and complex logic subsystems such as microcomputer central processor units. See Semiconductor
Integrated circuits consist of the combination of active electronic devices such as transistors and diodes with passive components such as resistors and capacitors, within and upon a single semiconductor crystal. The construction of these elements within the semiconductor is achieved through the introduction of electrically active impurities into well-defined regions of the semiconductor. The fabrication of integrated circuits thus involves such processes as vapor-phase deposition of semiconductors and insulators, oxidation, solid-state diffusion, ion implantation, vacuum deposition, and sputtering.
Generally, integrated circuits are not straightforward replacements of electronic circuits assembled from discrete components. They represent an extension of the technology by which silicon planar transistors are made. Because of this, transistors or modifications of transistor structures are the primary devices of integrated circuits. Methods of fabricating good-quality resistors and capacitors have been devised, but the third major type of passive component, inductors, must be simulated with complex circuitry or added to the integrated circuit as discrete components. See Transistor
Integrated circuits can be classified into two groups on the basis of the type of transistors which they employ: bipolar integrated circuits, in which the principal element is the bipolar junction transistor; and metal oxide semiconductor (MOS) integrated circuits, in which the principal element is the MOS transistor. Both depend upon the construction of a desired pattern of electrically active impurities within the semiconductor body, and upon the formation of an interconnection pattern of metal films on the surface of the semiconductor.
Bipolar circuits are generally used where highest logic speed is desired, and MOS for largest-scale integration or lowest power dissipation. High-performance bipolar transistors and complementary MOS (CMOS) transistors have been combined on the same chip (BiCMOS) to obtain circuits combining high speed and high density.
Bipolar integrated circuits
A simple bipolar inverter circuit using a diffused resistor and an npn transistor is shown in Fig. 1. The input voltage Vin is applied to the base of the transistor. When Vin is zero or negative with respect to the emitter, no current flows. As a result, no voltage drop exists across the resistor, and the output voltage Vout will be the same as the externally applied biasing voltage, +5 V in this example. When a positive input voltage is applied, the transistor becomes conducting. Current now flows through the transistor, hence through the resistor: as a result, the output voltage decreases. Thus, the change in input voltage appears inverted at the output.
The tendency toward increased complexity is dictated by the economics of integrated circuit manufacturing. Because of the nature of this manufacturing process, all circuits on a slice are fabricated together. Consequently, the more circuitry accommodated on a slice, the cheaper the circuitry becomes. Because testing and packaging costs depend on the number of chips, it is desirable, in order to keep costs down, to crowd more circuitry onto a given chip rather than to increase the number of chips on a wafer.
Integrated circuits based on amplifiers are called linear because amplifiers usually exhibit a linearly proportional response to input signal variations. However, the category includes memory sense amplifiers, combinations of analog and digital processing functions, and other circuits with nonlinear characteristics. Some digital and analog combinations include analog-to-digital converters, timing controls, and modems (data communications modulator-demodulator units). See Analog-to-digital conver-ter, Data communications
In the continuing effort to increase the complexity and speed of digital circuits, and the performance characteristics and versatility of linear circuits, a significant role has been played by the discovery and development of new types of active and passive semiconductor devices which are suitable for use in integrated circuits. Among these devices is the pnp transistor which, when used in conjunction with the standard npn transistors described above, lends added flexibility to the design of integrated circuits.
MOS integrated circuits
The other major class of integrated circuits is called MOS because its principal device is a metal oxide semiconductor field-effect transistor (MOSFET). It is more suitable for very large-scale integration (VLSI) than bipolar circuits because MOS transistors are self-isolating and can have an average size of less than 10-7 in.-2 (10-5 mm2). This has made it practical to use millions of transistors per circuit. Because of this high-density capability, MOS transistors are used for high-density random-access memories (RAMs), read-only memories (ROMs), and microprocessors. See Computer storage technology
Several major types of MOS device fabrication technologies have been developed since the mid-1960s. They are (1) metal-gate p-channel MOS (PMOS), which uses aluminum for electrodes and interconnections; (2) silicon-gate p-channel MOS, employing polycrystalline silicon for gate electrodes and the first interconnection layer; (3) n-channel MOS (NMOS), which is usually silicon gate; and (4) complementary MOS (CMOS), which employs both p-channel and n-channel devices.
Both conceptually and structurally the MOS transistor is a much simpler device than the bipolar transistor. In fact, its principle of operation has been known since the late 1930s, and the research effort that led to the discovery of the bipolar transistor was originally aimed at developing the MOS transistor. What kept this simple device from commercial utilization until 1964 is the fact that it depends on the properties of the semiconductor surface for its operation, while the bipolar transistor depends principally on the bulk properties of the semiconductor crystal. Hence MOS transistors became practical only when understanding and control of the properties of the oxidized silicon surface had been perfected to a very great degree.
A simple CMOS inverter circuit is shown in Fig. 2. The gates of the n-channel and p-channel transistors are connected together as are the drains. The common gate connection is the input node while the common drain connection is the output node. A capacitor is added to the output node to model the loading expected from the subsequent stages on typical circuits.
When the input node is in the “low state,” at 0 V, the n-channel gate to source voltage is 0 V while the p-channel gate to source voltage is -5 V. The n-channel transistor requires a positive gate-to-source voltage, which is greater than the transistor threshold voltage (typically 0.5–1 V), before it will start conducting current between the drain and source. Thus, with a 0-V gate-to-source voltage it will be off and no current will flow through the drain and source regions. The p-channel transistor, however, requires a negative voltage between the gate and source which is less than its threshold voltage (typically -0.5 to -1.5 V). The -5-V gate-to-source potential is clearly less than the threshold voltage, and the p-channel will be turned on, conducting current from the source to the drain, and thereby charging up the loading capacitor. Once the capacitor is charged to the “high state” at 5 V, the transistor will no longer conduct because there will no longer be a potential difference between the source and drain regions.
When the input is now put to the “high state” at 5 V, just the opposite occurs. The n-channel transistor will be turned on while the p-channel will be off. This will allow the load capacitor to discharge through the n-channel transistor resulting in the output voltage dropping from a “high state” at 5 V to a “low state” at 0 V. Again, once there is no potential difference between the drain and source (capacitor discharged to 0 V), the current flow will stop, and the circuit will be stable.
This simple circuit illustrates a very important feature of CMOS circuits. Once the loading capacitor has been either charged to 5 V or discharged back to 0 V, there is no current flow, and the standby power is very low. This is the reason for the high popularity of CMOS for battery-based systems. None of the other MOS technologies offers this feature without complex circuit techniques, and even then will typically not match the low standby power of CMOS. The bipolar circuits discussed above require even more power than these other MOS technologies. The price for CMOS's lower power are the additional fabrication steps required (10–20% more) when compared to NMOS.
BiCMOS integrated circuits
There is a strong interest in combining high-performance bipolar transistors and high-density CMOS transistors on the same chip (BiCMOS). This concept originated with work on bipolar circuits when power limitations became important as more functionality (and thus more transistors) was added to the chip. It is possible to continue adding more circuits on a chip without increasing the power by combining the low-power CMOS circuits with the bipolar circuits. This is done with both memory circuits and logic circuits, resulting in speeds somewhere between those of typical CMOS and bipolar-only circuits, but with the functional density of CMOS. The disadvantage of BiCMOS is its additional cost over plain CMOS or bipolar circuits, because the number of processing steps increases 20–30%. However, this increased complexity is expected to be used when either the additional functionality over bipolar circuits or the increased speed over CMOS circuits justifies the cost.
Integrated-circuit fabrication begins with a thin, polished slice of high-purity, single-crystal semiconductor (usually silicon) and employs a combination of physical and chemical processes to create the integrated-circuit structures described above. Junctions are formed in the silicon slice by the processes of thermal diffusion or high-energy ion implantation. Electrical isolation between devices on the integrated circuit is achieved with insulating layers grown by thermal oxidation or deposited by chemical deposition. Conductor layers to provide the necessary electrical connections on the integrated circuit are obtained by a variety of deposition techniques. Precision lithographic processes are used throughout the fabrication sequence to define the geometric features required.
VLSI chips containing 106 transistors and operating at tens of megahertz have been designed and fabricated and are commercially available. Projections indicate that silicon chips containing as many as 108 transistors may be feasible for digital applications and that perhaps even a 109 transistor chip is feasible for dynamic random access memories (DRAMs) before fundamental limits constrain the growth of complexity. (The limits beyond which the size of a transistor cannot be reduced are thought to depend on the degradation of its material properties when it is operated at high-field conditions and the general degradation of its performance and reliability.) Computer-aided engineering (CAE) systems provide the environment, specific computer tools, data management, and other services that are intended to support the design of these very complex, high-performance products. In many cases, the design of complex chips requires the cooperative endeavors of large design teams; thus the CAE system must also manage the design process to ensure that proper documentation has occurred, needed changes in the design database are made, and a chosen design methodology is enforced. The design process must be adapted to the very short design cycle times from product conception to production of a salable product that are characteristic of the semiconductor industry.
Gallium arsenide circuits
Integrated circuits based on gallium arsenide (GaAs) have come into increasing use since the late 1970s. The major advantage of these circuits is their fast switching speed.
The gallium arsenide field-effect transistor (GaAs FET) is a majority carrier device in which the cross-sectional area of the conducting path of the carriers is varied by the potential applied to the gate. Unlike the MOSFET, the gate of the GaAs FET is a Schottky barrier composed of metal and gallium arsenide. Because of the difference in work functions of the two materials, a junction is formed. The depletion region associated with the junction is a function of the difference in voltage of the gate and the conducting channel, and the doping density of the channel. By applying a negative voltage to the gate, the electrons under the gate in the channel are repelled, extending the depletion region across the conducting channel. The variation in the height of the conducting portion of the channel caused by the change in the extent of the depletion region alters the resistance between the drain and source. Thus the negative voltage on the gate modulates the current flowing between the drain and the source. As the height of the conducting channel is decreased by the gate voltage or as the drain voltage is increased, the velocity of charge carriers (electrons for n-type gallium arsenide) under the gate increases (similar to water in a hose when its path is constricted by passing through the nozzle). The velocity of the carriers continues to increase with increasing drain voltage, as does the current, until their saturated velocity is obtained (about 107 cm/s or 3 × 105 ft/s for gallium arsenide). At that point the device is in the saturated region of operation; that is, the current is independent of the drain voltage.