THE STATE OF OUR KNOWLEDGE OF SEMICONDUCTORS AND FUTURE TASKS

A. F. Ioffe

Semiconductors are a comparatively “young” science—it is not yet 30 years old. It bears the features of the new era. One cannot fail to note, for example, the close interaction of physics and technology throughout its entire history. If one does not count the photoconductivity of selenium, discovered as early as the last century, and attempts to use it in the first years of the twentieth century, then the beginning of the systematic study of semiconductors may be taken as 1927–1932, when alternating-current rectifiers and photoelements made of cuprous oxide appeared. These technical outcomes predetermined the development of semiconductor physics for the next decade. Cuprous oxide became the principal object of scientific investigations, whose interest was directed toward the properties of barrier layers, responsible both for rectification and for the photoelectric effect.

As new applications appeared—voltage stabilizers, time relays, semiconductor cathodes, electrolytic capacitors—the front of physical work broadened, and new semiconductor materials were studied.

Thousands of printed works have been devoted to selenium and cuprous oxide. During the war of 1939–1945, high-frequency rectifiers and amplifiers made of germanium and silicon appeared. Since then, the attention of physicists has shifted to these and to analogous semiconductors.

Another feature characterizing the progressive path of development of semiconductors is the close connection between experiment and theory. From the very beginning, the doctrine of semiconductors was based on quantum theory, which by that time had achieved considerable development in the physics of metals. If at first the similarities between the theories of metals and of semiconductors were emphasized and the entire technical apparatus of the theory of metals was transferred to semiconductors, then in recent years ever greater importance has been acquired by the properties that distinguish them from one another.

It may be considered a great stroke of luck, one that conditioned the success of semiconductors, that the sound path of the interconnection between physics and production, and of the combination of theory and experiment, became ever stronger as the scope and significance of the problem grew.

At the present time semiconductors are a central problem of solid-state physics, and their application in technology is becoming the principal route to solving problems in radio engineering, automation, and power engineering.

More than 10,000 works in this field have been published over the last 30 years, and their number is growing at an ever-increasing rate. The volume of production using semiconductors has already reached the scale of the most important branches of electrical engineering.

All this obliges us to consider attentively the strong and weak sides of the present state of the semiconductor problem and to outline paths for further progress.

I have no doubt that we shall spare no effort to justify the great hopes placed precisely in this field of knowledge. Semiconductors can reveal the nature of processes occurring in the solid body, and, having mastered them, we shall place them at the service of our main task—the building of communist society.

In five years significant shifts have taken place in our science. And nowhere, perhaps, have physics and technology been intertwined so closely as in the field of semiconductors.

The problem of \(p-n\) junctions has moved to the foreground, and on this basis there have arisen not only new radio devices—diodes, triodes, and generators of radio oscillations—but also new types of rectifiers of alternating current, photoelements, and sources of current that convert the energy of radioactive radiation.

New semiconductor materials have appeared, which have brought great clarity to the doctrine of semiconductors and have proved to be much more valuable objects of research than the prewar copper oxide and selenium. The attention of researchers has been drawn to nonequilibrium processes and to minority current carriers.

Liquid and glassy semiconductors have appeared, disrupting the seemingly inseparable connection of semiconductors with crystals.

Alongside ferromagnets, semiconductor ferrites have appeared; alongside insulators—barium titanates, which concentrate electrical energy and convert it into mechanical energy.

Semiconductor thermoelements are acquiring ever greater significance, both as generators of electrical energy and as means of producing cold and heat.

Semiconductor catalysts of chemical processes are coming onto the scene, and phosphors are coming ever closer to us.

In this stream of scientific and technical successes, one wishes to single out the participation of our Soviet science.

The idea and theory of polarons has already become a firm achievement of science. We owe it to the Kiev physicist S. I. Pekar.

The idea of excitons, proposed 24 years ago by the late Ya. I. Frenkel, has gone through a complex path of theoretical and experimental development and is represented at the present conference by a number of reports in several sections.

The phenomenon of paramagnetic resonance discovered by E. K. Zavoisky and the cyclotron resonance predicted by Ya. G. Dorfman opened new paths for penetrating into the nature of the effective mass of electrons and the structure of energy levels. True, these phenomena received their principal development in the USA, and not in our laboratories.

To the credit of Soviet physics one may also record the development of the doctrine of photoelectricity and photo-electromotive forces by V. E. Lashkarev and his school.

B. M. Vul’s barium titanates and the piezoelectrics made from them have had no small influence on the development of radio-engineering devices.

An entire section is devoted to semiconductor catalysts.

The study and application of semiconductor thermoelements is the work of the hands (and not only of the hands, but also of the heads) of Soviet physicists.

I shall not enumerate other quite important successes of semiconductor science and the striking technical innovations to which it has led.

It is more interesting and more important to note the weak points and to outline ways of eliminating them. Here I shall express my own opinion, with which perhaps not everyone will agree.

The theory of semiconductors was created as a development of the theory of metals, as its extrapolation. It seems to me that in this extrapolation the theory has gone far beyond the limits of its applicability. Passing freely from one collision to another, an electron moves with a certain mean velocity and acceleration, which characterizes its effective mass. But what is to be done if the mean free path turns out to be shorter than the wavelength, shorter than the interatomic distances; of what velocity can one speak if the uncertainty in the magnitude of the energy exceeds the energy itself by tens of times? Yet such are precisely the conditions in many semiconductors.

Here all the customary tools are inapplicable—not only the analogy with the kinetic theory of gases, but also the kinetic equation, the velocity distribution function, and the relaxation time. Cyclotron and paramagnetic resonances are likewise inapplicable to semiconductors with a short mean free path.

Should we not pass to the probability of transition from one state to another, from one region to a neighboring one?

The quantum theory of semiconductors, the concept of effective mass, is built on the periodic properties of the field in a crystal. But what about liquid, amorphous, glassy semiconductors? They are without theory, although they are no worse than other semiconductors.

Experiment connects the properties of semiconductors with short-range order, theory—with long-range order. Evidently this is only a part of the theory, a special case of it, and not the whole theory. Should we not try another approach, one that does not presuppose a periodic field—for example, take the theory of plasma as the starting point?

The question of the limits within which the many-electron problem may be replaced by a one-electron one cannot be considered clarified. There is no theory of semiconductors with different types of bonding, with different structures of the periodic field.

The idea of excitons was advanced by Ya. I. Frenkel to explain the absorption of light without photoconductivity. But the data of E. F. Gross rather indicate a direct connection of photosensitivity with excitons.

There are also quite a few other shortcomings in the theory of semiconductors: thermoelectric forces at low temperatures, the course of the mobility with temperature agree poorly with theory, especially where the mobility is small. This is criticism of the theory from the consumer’s side. I turn to self-criticism of the experimentalist.

There are still more failures in the experimental data.

The basis of a semiconductor is the bonding forces acting in it, which manifest themselves above all in strength, in shear; yet physicists have forgotten the mechanical properties of semiconductors, and at the conference there is not a single report on this subject. We have handed over the whole problem of strength to the metallurgists!

The source that liberates charges, and the obstacle that scatters them, is the thermal motion of atoms. But who is interested in it? Only my report on thermal conductivity is directly connected with thermal motion.

What do we know about molecular and dipole lattices; how many types of semiconductors have still not been studied? It is astonishing that such a promising field as refractory semiconductors has not been investigated at all.

How little we still know about the connection of physical properties with short-range order, with the character of chemical forces, with structure.

There is too little work in the field of low temperatures, when the thermal background recedes into the background and the most important quantum properties of matter come to the fore. The best example of the significance of such studies is cyclotron resonance, which for the first time gave a complete quantitative theory of effective masses.

Many general words have been said, but too little concrete research is devoted to the various types of distortions and impurities.

The existing theoretical and experimental bases are insufficient for the all-round development of semiconductor technology.

The prospects for applications of semiconductors are truly grandiose. In the field of power engineering, we face the task of using the enormous quantities of heat that are wasted in fireboxes, in heat-

machines, in metallurgical furnaces, and under natural conditions, despite the presence of a considerable temperature difference.

The simplicity of semiconductor thermoelements brings to the fore the use of all kinds of fuel and, above all, of the sun’s rays, which deliver to the earth in a single day as much energy as is contained in all reserves of coal and oil combined.

Great prospects are opened by photoelements, ferroelectrics as energy converters, and ferrites as a means of concentrating high-frequency magnetic energy.

There is no doubt that semiconductors will occupy a prominent place in solving the problem of heat and cold.

No one any longer doubts the progressive role of semiconductors in radio engineering, in automation, remote control, and television, despite the fact that semiconductor diodes, triodes, and generators are not yet even 10 years old.

The quantitative advantages in dimensions, strength, energy consumption, low cost, and mass producibility here grow into qualitatively new possibilities.

Still greater and more concerted work by physicists, chemists, geologists, and engineers is needed in order to turn into real achievements the richest possibilities inherent in semiconductors.