Integrated-Circuit Design

Summary

The integrated circuit (IC) is the essential building block of modern electronics. Each IC consists of a chip of silicon upon which has been constructed a series of transistor structures, typically MOSFETs, or metal oxide semiconductor field effect transistors. The chip is encased in a protective outer package whose size facilitates use by humans and by automated machinery. Each chip is designed to perform specific electronic functions, and the package design allows electronics designers to work at the system level rather than with each individual circuit. The manufacture of silicon chips is a specialized industry, requiring the utmost care and quality control.

Definition and Basic Principles

An integrated circuit, or IC, is an interconnected series of transistor structures assembled on the surface of a silicon chip. The purpose of the transistor assemblages is to perform specific operations on electrical signals that are provided as inputs. All IC devices can be produced from a structure of four transistors that function to invert the value of the input signal, called NOT gates or inverters. All digital electronic circuits are constructed from a small number of different transistor assemblages, called gates, that are built into the circuitry of particular ICs. The individual ICs are used as the building blocks of the digital electronic circuitry that is the functional heart of modern digital electronic technology.

The earliest ICs were constructed from bipolar transistors that function as two-state systems, a system high and a system low. This is not the same as “on” and “off,” but is subject to the same logic. All digital systems function according to Boolean logic and binary mathematics.

Modern ICs are constructed using metal oxide semiconductor field effect transistor (MOSFET) technology, which has allowed for a reduction in size of the transistor structures to the point in which, at about 65 nanometers (nm) in size, literally millions of them can be constructed per square centimeter (cm²) of silicon chip surface. The transistors function by conducting electrical current when they are biased by an applied voltage. Current ICs, such as the central processing units (CPUs) of personal computers, can operate at gigahertz frequencies, changing that state of the transistors on the IC billions of times per second.

Background and History

Digital electronics got its start in 1906, when American inventor Lee de Forest constructed the triode vacuum tube. Large, slow, and power-hungry as they were, vacuum tubes were nevertheless used to construct the first analogue and digital electronic computers. In 1947, American physicist William Shockley and colleagues constructed the first semiconductor transistor junction, which quickly developed into more advanced silicon-germanium junction transistors.

Through various chemical and physical processes, methods were developed to construct small transistor structures on a substrate of pure crystalline silicon. In 1958, American physicist and Nobel laureate Jack Kilby first demonstrated the method by constructing germanium-based transistors as an IC chip, and American physicist Robert Noyce constructed the first silicon-based transistors as an IC chip. The transistor structures were planar bipolar in nature, until the CMOS (complementary metal oxide semiconductor) transistor was invented in 1963. Methods for producing CMOS chips efficiently were not developed for another twenty years.

Transistor structure took another developmental leap with the invention of the field effect transistor (FET), which was both a more efficient design than that of semiconductor junction transistors and easier to effectively manufacture. The MOSFET structure also is amenable to miniaturization and has allowed designers to engineer ICs that have one million or more transistor structures per centimeters squared.

How It Works

The electronic material silicon is the basis of all transistor structures. It is classed as a pure semiconductor. It is not a good conductor of electrical current or insulate well against electrical current. By adding a small amount of some impurity to the silicon, its electrical properties can be manipulated such that the application of a biasing voltage to the material allows it to conduct electrical current. When the biasing voltage is removed, the electrical properties of the material revert to their normal semiconductive state.

Silicon Manufacture. Integrated-circuit design begins with the growth of single crystals of pure silicon. A high-purity form of the material, known as polysilicon, is loaded into a furnace and heated to melt the material. At the proper stage of melt, a seed crystal is attached to a slowly turning rod and introduced to the melt. The single crystal begins to form around the seed crystal and the rotating crystal is “pulled” from the melt as it grows. This produces a relatively long, cylindrical, single crystal that is then allowed to cool and set.

Wafers. From this cylinder, thin wafers or slices are cut using a continuous wire saw that produces several uniform slices at the same time. The slices are subjected to numerous stages of polishing, cleaning, and quality checking, the end result of which is a consistent set of silicon wafers suitable for use as substrates for integrated circuits.

Circuitry. The integrated circuit itself begins as a complex design of electronic circuitry to be constructed from transistor structures and “wiring” on the surface of the silicon wafer. The circuits can be no more than a series of simple transistor gates (such as invertors, AND-gates, and OR-gates), up to and including the extremely complex transistor circuitry of advanced CPUs for computers.

Graphics technology is extremely important in this stage of the design and production of integrated circuits, because the entire layout of the required transistor circuitry must be imaged. The design software also is used to conduct virtual tests of the circuitry before any ICs are made. When the theoretical design is complete and imaged, the process of constructing ICs can begin.

Because the circuitry is so small, a great many copies can be produced on a single silicon wafer. The actual chips that are housed within the final polymer or ceramic package range in size from two to five cm². The actual dimensions of the circa 1986 Samsung KS74AHCT240 chip, for example, are just 1 cm × 2 cm. The transistor gate sizes used in this chip are 2 micrometers (um) (2 × 10⁻⁶ meters [m]), and each chip contains the circuitry for two octal buffers, constructed from hundreds of transistor gate structures. Transistor construction methods have become much more efficient, and transistor gate sizes are now measured in nanometers (10⁻⁹ m) rather than um, so that actual chip sizes have also become much smaller, in accord with Moore's law. The gate structures are connected through the formation of aluminum “wires” using the same chemical vapor deposition methodology used to form the silicon oxide and other layers needed.

Photochemical Etching. The transistor structures of the chip are built up on the silicon wafer substrate through a series of steps in which the substrate is photochemically etched and layers of the necessary materials are deposited. Absolute precision and an ultraclean environment are required at each step. The processes are so sensitive that any errant speck of dust or other contaminant that finds its way to the wafer's surface renders that part of the structure useless.

Accounting for losses of functional chips at each stage of the multistep process, it is commonly the case that as little as 10 percent of the chips produced from any particular wafer will prove viable at the end of their construction. If, for example, the procedure requires one hundred individual steps, not including quality-testing steps, to produce the final product, and a mere 2 percent of the chips are lost at each step, then the number of viable chips at the end of the procedure will be 0.98¹⁰⁰, or 13.26 percent of the number of chips that could ideally have been produced.

Each step in the formation of the IC chip must be tested to identify the functionality of the circuitry as it is formed. Each such step and test procedure adds significantly to the final cost of an IC. When the ICs are completed, the viable ones are identified, cut from the wafer, and enclosed within a protective casing of resin or ceramic material. A series of leads are also built into this “package” so that the circuitry of the IC chip can be connected into an electronic system.

Applications and Products

Bipolar Transistors and MOSFETS. Transistors are commonly pictured as functioning as electronic on-off switches. This view is not entirely correct. Transistors function by switching between states according to the biasing voltages that are applied. Bipolar switching transistors have a cut-off state in which the applied biasing voltage is too low to make the transistor function. The normal operating condition of the transistor is called the linear state. The saturation state is achieved when the biasing voltage is applied to both poles of the transistor, preventing them from functioning. MOSFET transistors use a somewhat different means, relying on the extent of an electric field within the transistor substrate, but the resulting functions are essentially the same.

The transistor structures that form the electronic circuitry of an IC chip are designed to perform specific functions when an electrical signal is introduced. For simple IC circuits, each chip is packaged to perform just one function. An inverter chip, for example, contains only transistor circuitry that inverts the input signal from high to low or from low to high. Typically, six inverter circuits are provided in each package through twelve contact points. Two more contact points are provided for connection to the biasing voltage and ground of the external circuit. It is possible to construct all other transistor logic gates using just inverter gates. All ICs use this same general package format, varying only in their size and the number of contact points that must be provided.

MOSFETS have typical state switching times of something less than 100 nanoseconds, and are the transistor structures of choice in designing ICs, even though bipolar transistors can switch states faster. Unlike bipolars, however, MOSFETS can be constructed and wired to function as resistors and can be made to a much smaller scale than true resistors in the normal production process. MOSEFTS are easier to manufacture than are bipolars as they can be made much smaller in VLSI (very large scale integration) designs. MOSFETS also cost much less to produce.

NOTs, ANDs, ORs, and Other Gates. All digital electronic devices comprise just a few basic types of circuitry—called logic elements—of which there are two basic types: decision-making elements and storage elements. All logic elements function according to the Boolean logic of a two-state (binary) system. The only two states that are allowed are “high” and “low,” representing an applied biasing voltage that either does or does not drive the transistor circuitry to function. All input to the circuitry is binary, as is all output from the circuitry.

Decision-making functions are carried out by logic gates (the AND, OR, NOT gates and their constructs) and memory functions are carried out by combination circuitry (flip-flops) that maintains certain input and output states until it is required to change states. All gates are made up of a circuit of interconnected transistors that produces a specific output according to the input that it receives.

A NOT gate, or inverter, outputs a signal that is the opposite of the input signal. A high input produces a low output, and vice versa. The AND gate outputs a high signal only when all input signals are high, and a low output signal only if any of the input signals are low. The OR gate functions in the opposite sense, producing a high output signal if any of the input signals are high and a low output signal only when all of the input signals are low. The NAND gate, which can be constructed from either four transistors and one diode or five diodes, is a combination of the AND and NOT gates. The NOR gate is a combination of the OR and NOT gates. These, and other gates, can have any number of inputs, limited only by the fan-out limits of the transistor structures.

Sequential logic circuits are used for timing, sequencing, and storage functions. The flip-flops are the main elements of these circuits, and memory functions are their primary uses. Counters consist of a series of flip-flops and are used to count the number of applied input pulses. They can be constructed to count up or down, as adders, subtracters, or both. Another set of devices called shift registers maintains a memory of the order of the applied input pulses, shifting over one place with each input pulse. These can be constructed to function as series devices, accepting one pulse (one data bit) at a time in a linear fashion, or as parallel devices, accepting several data bits, as bytes and words, with each input pulse. The devices also provide the corresponding output pulses.

Operational amplifiers, or OP-AMPs, represent a special class of ICs. Each OP-AMP IC contains a self-contained transistor-based amplifying circuit that provides a high voltage gain (typically 100,000 times or more), a very high input impedance and low output impedance, and good rejection of common-mode signals (that is, the presence of the same signal on both input leads of the OP-AMP).

Combinations of all of the gates and other devices constructed from them provide all of the computing power of all ICs, up to and including the most cutting-edge CPU chips. Their manufacturing process begins with the theoretical design of the desired functions of the circuitry. When this has been determined, the designer minimizes the detailed transistor circuitry that will be required and develops the corresponding mask and circuitry images that will be required for the IC production process. The resulting ICs can then be used to build the electronic circuitry of all electronic devices.

Social Context and Future Prospects

IC technology is on the verge of extreme change, as new electronic materials such as graphene and carbon nanotubes are developed. Research with these materials indicates that they will be the basic materials of molecular-scale transistor structures, which will be thousands of times smaller and more energy-efficient than VLSI technology based on MOSFETs. The computational capabilities of computers and other electronic devices are expected to become correspondingly greater as well.

Such devices will utilize what will be an entirely new type of IC technology, in which the structural features are actual molecules and atoms, rather than what are by comparison mass quantities of semiconductor materials and metals. As such, future IC designers will require a comprehensive understanding of both the chemical nature of the materials and of quantum physics to make the most effective use of the new concepts.

The scale of the material structures as well will have extraordinary application in society. It is possible, given the molecular scale of the components, that the technology could even be used to print ultrahigh resolution displays and computer circuitry that would make even the lightest and thinnest of present-day appliances look like the ENIAC (Electronic Numerical Integrator and Computer) of 1947, which was the first electronic computer.

The social implications of such miniaturized technology are far-reaching. The RFID (radio-frequency identification) tag is now becoming an important means of embedding identification markers directly into materials. RFID tags are tiny enough to be included as a component of paints, fuels, explosives, and other materials, allowing identification of the exact source of the material, a useful implication for forensic investigations and other purposes. Even the RFID tag, however, would be immense compared with the molecular scale of graphene and nanotube-based devices that could carry much more information on each tiny particle. This would make computers smaller, faster, and more powerful. Graphene transistors were successfully created in a laboratory setting at the University of Florida in 2017, but use in commercial products is well off. Five years later, scientists in Shanghai, China, announced they had succeeded in building a graphene transistor gate just 0.34 nanometers in length, essentially the size of one carbon atom. Researchers in Sweden developed a simple hydrocarbon molecule with a logic gate function.

The ultimate goal of electronic development, in current thought, is the quantum computer, a device that would use single electrons, or their absence, as data bits. The speed of such a computer would be unfathomable, taking seconds to carry out calculations that would take present-day supercomputers billions of years to complete. The ICs used for such a device would bear little resemblance, if any, to the MOSFET-based ICs of the early twenty-first century.

Bibliography

Afacan, Engin, Nuno Lourenco, Ricardo Martins, and Gunhan Dundar. "Review: Machine Learning Techniques in Analog/RF Integrated Circuit Design, Synthesis, Layout, and Test." Integration, vol. 77, Mar. 2021, pp. 113-130, DOI: 10.1016/j.vlsi.2020.11.006. Accessed 7 June 2022.

Brown, Julian R. Minds, Machines, and the Multiverse: The Quest for the Quantum Computer. New York: Simon & Schuster, 2000.

"Graphene Transistor Could Mean Computers That Are 1,000 Times Faster." Science Daily, 13 June 2017, www.sciencedaily.com/releases/2017/06/170613145144.htm. Accessed 31 Aug. 2018.

Kurzweil, Ray. The Age of Spiritual Machine: When Computers Exceed Human Intelligence. New York: Penguin, 2000.

Marks, Myles H. Basic Integrated Circuits. Blue Ridge Summit, Pa.: Tab Books, 1986.

Pavlidis, Vasilis F. Three-Dimensional Integrated Circuit Design. 2nd ed., Morgan Kaufman, 2017.

Tasic, Magdalena, et al. "Electro-Mechanically Switchable Hydrocarbons Based on [8]Annulenes." Nature, vol. 13, 2022, DOI: 10.1038/s41467-022-28384-8. Accessed 7 June 2022.

Zheng, Y., et al. “Graphene Field Effect Transistors with Ferroelectric Gating.” Physical Review Letters 105 (2010).