Integrated circuit: design, fabrication, types, and impact

Colloquially, microchip or chip is now sometimes equated with an IC, although the chip only represents the "inner workings" of a circuit, i.e. the semiconductor chip with the actual circuit on it.

Also to be distinguished from integrated circuits are double, triple or other multiple components (whether semiconductor or tube), each of which is contained in a common package or glass bulb, such as double diodes, triodes, pentodes, double or triple Darlington transistors, etc.

Predecessor

Before the development of integrated circuits, there were active electronic components that were hardwired together with several passive ones and shipped in one package. One example was the 3NF and 2HF electron or vacuum tubes developed in the mid-1920s - the former a triple tube similar to the later ECC83 double triode. Unlike the younger tube, however, both 1920s tubes had four or two resistors and two or one capacitor already incorporated inside the bulb of the tube, as in an IC. This resulted in ready-made radio circuits - the 3NF as basic circuit for local reception and amplification and the 2HF additionally for optional long-distance reception, where only the external components for station selection, power supplies and playback (loudspeakers or headphones) as well as the antenna had to be connected.

Until the end of the 1950s, electronic circuits were built with discrete components, i.e. with individual transistors, diodes, etc., which were assembled on a printed circuit board to form a circuit, cf. integration (technology). In terms of size and service life, this was already a significant breakthrough compared to the electron tubes that were competing at the time.

Even before the invention of the transistor, there were electronic components which integrated several functions in one component, in the form of multi-system tubes, composite tubes such as the 3NF, duodiodes or also multi-anode mercury vapour rectifiers which combined the function of several controlled or uncontrolled rectifiers (one cathode and several anodes) in one component. Transistors have significant advantages over vacuum tubes, such as lower power consumption and size. With the application of printed circuit boards or PCBs and the resulting downsizing of products, this newer technology began to displace the early tube-based integrated systems. This trend fully intensified with the development and massive use of integrated circuits beginning in the 1960s.

Jacobi patent

Little known is the "semiconductor amplifier", a circuit consisting of five transistors on a semiconductor serving as a carrier material, invented and patented by Werner Jacobi as early as 1949. These form a three-stage amplifier circuit in the form of an integrated circuit. Two transistors are connected "overhead" to effect impedance conversion between the transistor stages. Jacobi noted that this allows hearing aids, for example, to be made small, light and cheap.

An immediate commercial use of his patent is not known. The formulation of the integration idea in the patent specification published on May 15, 1952 reads: "Semiconductor amplifier, characterized in that several electrode systems acting in different switching or amplifier stages are placed on the semiconductor. Thus, for example, the integration of several light-emitting diodes in one housing goes back in its basic idea to Jacobi.

From Kilby and Noyce to today

The first integrated circuit (a flip-flop) was developed by Jack Kilby in September 1958. It consisted of two bipolar transistors mounted on a germanium substrate and connected by gold wires. This hybrid circuit is thus a first example of the implementation of the already known transistor-transistor logic (TTL) on a circuit. It was a preliminary stage for the further development of TTL circuits towards smaller designs.

Robert Noyce filed a patent application for the first "monolithic" integrated circuit, i.e. made from or in a single single-crystal substrate, in July 1959. The key feature of Noyce's invention was the complete fabrication of the components, including wiring, on a single substrate. The manufacturing process already used photolithographic techniques and diffusion processes that Fairchild Semiconductor had developed shortly before for the production of the first modern diffusion bipolar transistor. Based on these techniques, among others, the first microprocessors were introduced almost simultaneously by three companies in 1970/71: the Intel 4004, the Texas Instruments (TI) TMS 1000 and the Garrett AiResearch "Central Air Data Computer" (CADC).

The first integrated circuits in series production were created in the early 1960s (mainly by Texas Instruments and Fairchild Semiconductor). They consisted of up to a few dozen bipolar transistors (small-scale integration, SSI), typically in RTL technology. Over the years, however, the components became smaller and smaller, passive components such as resistors were integrated and the complexity of the integrated circuits increased. As a result, the number of transistors per chip or per unit area also increased; the number of transistors was the most important parameter of ICs.

A promoting factor for further development was the defense industry and space travel. Until the mid-1960s, the US government was the main customer for integrated circuits. The aim was to miniaturize the technology of both sectors. From 1965, the Gemini program was equipped with on-board computers based on ICs.

With medium-scale integration (MSI), a few hundred transistors found space on a chip, and with large-scale integration (LSI) at the beginning of the 1970s, a few thousand transistors. This made it possible for the first time to integrate an entire main processor (CPU) as a so-called microprocessor on a chip, which extremely reduced the costs for computers. At the beginning of the 1980s, this was followed by very-large-scale integration (VLSI) with several hundred thousand transistors, which soon made it possible to produce memory chips (RAM) with a capacity of 256 KiBit and 1 MiBit. This further development of manufacturing technology was accompanied by ever greater design automation (see chip design) of the design and the photomasks required for production, without which it was no longer possible to develop more complex circuits.

In 2010, graphics processors contained up to three billion transistors (see Nvidia Tesla), "normal" general-purpose CPUs up to 1.17 billion transistors (Intel Core i7-980X). The Itanium 2 Tukwila consists of 2.05 billion transistors. Meanwhile, graphics processors have reached transistor counts of over eight billion transistors (Nvidia GTX TitanX). Even larger numbers are reached with memory components, but with less complexity of the entire chip.