The transistor is a semiconductor device used to amplify or interrupt power or electrical signals and is one of the basic building blocks of modern electronics.

By making it possible to create circuits that did not require the high anode voltages of thermionic valves, the device enabled the creation of portable battery-powered radios. As a result, the term “transistor radio” was often used in common parlance to refer to such devices, which reached the mass market in the 1950s.

transistor scaled

Types of transistors

There are two main types of transistors: the bipolar junction transistor and the field effect transistor, and it is possible to miniaturize devices of both categories within integrated circuits, making them key components in microelectronics.

Historical background

Physicist and engineer Julius Edgar Lilienfeld designed the first transistor in Canada in 1925, describing a device similar to today’s field-effect transistor. However, Lilienfeld did not publish his research, and in 1934 German inventor Oskar Heil patented a similar device.

The first transistor was made with two electrodes whose tips, very thin and a few hundredths of a millimeter apart, 127 to 50 microns to be exact, were pressed onto the surface of a wafer of very pure, polycrystalline, n-type germanium crystal. The point contact technique was already known and used in the construction of detector diodes. Since the transistor functioned similarly to a triode, it was provisionally called a solid-state triode: the final name was derived from the combination of the terms “TRANSCONDUCTOR” and “RESISTOR”. The first working prototype was made in December 1947 by two researchers at Bell Labs: Walter Brattain and John Bardeen of William Shockley’s research group.

This was the spike transistor, while it was William Shockley who came up with the concept in January 1948 and, in the spring of the following year, formulated the theory of the junction transistor, initially called the “sandwich transistor” by Shockley himself in his laboratory journal. In 1956, the three researchers were awarded the Nobel Prize in Physics “for their research on semiconductors and for the discovery of the transistor effect. By the late 1950s, transistor production had shifted to the use of silicon as the semiconductor element, and by the 1970s, the germanium transistor was obsolete.

Package types multiplied, and over the years materials such as ceramic, metal, plastic, or mixed assemblies were used. Glass was also used in the 1960s: European manufacturer Philips encased its low-power devices, such as the OC70 and OC71, in a cylindrical glass ampoule painted black and filled with silicone grease. If the device had a higher power dissipation, such as the OC72, the device was simply covered with an aluminum cap; with identical rheophores, the distributor was marked with a dot of dark red paint. Over time, many types of packages have fallen into disuse in favor of geometries that are more efficient at dissipating the heat generated. Today’s low-frequency power devices, including some types of diodes and ICs, are packaged in the standard TO-3 package, which has two drilled flanges suitable for mounting to the heatsink with a pair of screws. Made of steel, copper or aluminum, at an ambient temperature of 25°C, it is capable of transferring to the heatsink 300 watts of thermal power generated by the die.

In terms of the movement of electrical charges within the device, the transistors are called bipolar transistors, in which both electrons and gaps contribute to the passage of current. Both the point contact transistor and the junction transistor are bipolar transistors. The point-contact type, which is of historical importance because it was the first to be made and found practical application, albeit limited, soon became obsolete and was replaced by the junction type, which is more stable and less noisy. Later, other types of transistors were created in which current flowed through only one type of charge carrier: these devices are the field-effect transistors. Over time, both have given rise to many different types of transistors used for a wide variety of purposes. The measuring instrument used to verify and characterize the many parameters of transistors, as well as diodes, is the curve tracer (traciacurve), a term given to the instrument in relation to electrical signals displayed in the form of graphs resembling multiple “curves”, the appearance is similar to an oscilloscope: this type of instrument is historically produced by the company Tektronix.

Evolution of the transistor

In the last decades of the 20th century, the miniaturization of transistors made it possible to achieve a continuous increase in the operating speed of components in computers and other electronic devices. In fact, the size of a transistor has a significant impact on its speed of operation; for example, in a bipolar transistor, a limiting factor is the time it takes for the electrons to travel through the base: reducing the thickness of the base reduces the distance the electrons have to travel and consequently increases the speed with which the transistor can go from the on state to the off state. The ability to etch more and more transistors onto a single silicon wafer is the basis for the cost-effectiveness of this electronic component, because the cost of the many complicated steps required to manufacture it can be divided by an increasing number of transistors.

However, there are limits to the miniaturization trend. In the case of extremely miniaturized devices, new phenomena must be taken into account that do not occur at larger sizes; in order to deal with them analytically, numerical simulations of the movement of electrons inside the device must be used at the design stage. At relatively large lengths, the electrons, after rapid acceleration, dissipate an amount of energy in collisions that balances the kinetic energy provided by the field. The particles are therefore kept at a constant velocity for much of the time, and their behavior can be described by a linear equation. As the size decreases, the electrons are no longer able to maintain a constant velocity; they accelerate at each instant, and the associated equations must take this into account.

Femtosecond experiments are needed to refine the simulation models. It is essential to deepen knowledge of solid-state physics because as chips become more complex, the number of steps required to manufacture them increases, and each step can affect the previous ones. Currently, the most advanced commercial chips are made in the far ultraviolet, a technique complicated by the difficulty of making lasers that emit in this region of the spectrum.

For the next generation of devices, it will most likely be necessary to use even higher frequencies, in the X-ray band. Of course, to make a new generation of devices, you have to have the appropriate equipment, which is getting more and more expensive. So we are approaching natural limits in chip design, and many researchers have long focused their studies on the use of semiconductor materials other than silicon, such as gallium arsenide and similar compounds.

The fact remains, however, that the industry has invested enormous sums in equipment to manufacture silicon-based devices and would clearly like to find ways to increase performance without abandoning this element. Researchers at IBM decided to take a hard look at silicon technology to find ways to improve the performance of silicon-based logic chips while maintaining full compatibility with current manufacturing technologies. The results were more than flattering, showing that an alloy of silicon and germanium could be the basis for exceptionally fast transistors.

By the mid-1950s, it had been established in principle that heterojunctions could be a way to make a transistor switch faster without reducing its size, but by changing its fundamental electronic properties. The electric fields that naturally occur in the two materials that make up the heterojunction cause positive or negative charges to be confined to opposite sides of the interface. By gradually changing the composition of a junction from that of one material to that of the other, an extended electric field can be obtained in the transition region. The field generated in a gradual heterojunction can be used to induce electrons to rapidly cross the base of a bipolar device.

Further research indicated that of the possible pairs of semiconductor materials, silicon and germanium were the most promising. However, the time was not right for the realization of a working silicon-germanium heterojunction. The interatomic spacing in a germanium crystal is 4 percent larger than in a silicon crystal, so the lattices of the two elements, which have the same crystal structure, do not match perfectly. When germanium atoms are deposited on a much thicker silicon substrate, they initially closely follow the atomic arrangement of the underlying layer, but an enormous strain builds up that increases as more layers are deposited. Eventually, defects are created in the germanium structure that relieve the strain by ejecting entire rows of germanium atoms from the crystal lattice, allowing the other atoms to move away and regain their natural spacing. To relieve the stress, the ejection must affect about 4 percent of the deposited germanium atoms, but this rearrangement in an area the size of a chip would create thousands of defects that would prevent it from functioning.

After experimenting with various techniques, the best results were obtained by chemical vapor deposition at temperatures between 400 and 500 ºC under high vacuum (one millionth of an atmosphere). Using this technique, transistors were built that offered significantly better performance than conventional ones. In the bipolar transistor, for example, a standard measure of performance is the way in which the device gain (given by the ratio of the current emitted by the transistor to the current required to turn it on) depends on the switching frequency. A conventional bipolar transistor may have a gain of 100, but as the switching frequency is increased, the gain gradually decreases; when it becomes one, the transistor behaves like a simple electrical wire, in that the input current is equal to the output current.

The speed of a transistor is generally judged by the switching frequency it can reach before its gain drops to one. The first gradual heterojunction made by the IBM research group in 1989 had a speed of 75 GHz, about twice as fast as the best comparable silicon devices; later speeds of more than 110 GHz were achieved, and it was found that inserting such devices into real circuits was uneventful. Although silicon-germanium alloy technology is still in its infancy, commercialization of silicon-germanium integrated circuits has begun. To date, IBM is the only company to have demonstrated the ability to integrate significant numbers of high performance stepwise heterojunction bipolar transistors into circuits; researchers are working to achieve significant results with this technique for field-effect transistors as well.

Notify of

Inline Feedbacks
View all comments
Scroll to Top