Congresses and Conferences

Third All-Union Conference on Semiconductors

L. Groshev, Leningrad

From May 27 to June 2, the Third All-Union Conference on Semiconductors was held in Odessa. Representatives of 15 research institutes and laboratories from Leningrad, Moscow, Odessa, Kiev, Tomsk, and a number of other cities took part in its work. It is by no means accidental that a physics conference is devoted entirely to semiconductors, which have recently become the subject of serious study in both Soviet and foreign physics.

Previously, solid-state physics dealt mainly with the study of metals and insulators, since both play an important role in technical questions. Recently, however, semiconductors, which by their properties lie between metals and insulators, have come increasingly into technology. This circumstance naturally compelled physicists to turn their attention to the problem of semiconductors. The intensive work being carried out in this field may be judged from the unusually increased number of experimental and theoretical papers published in Soviet and foreign journals.

In our Soviet Union, where more than in any other country the connection between science and technology is being realized, the problem of semiconductors is also becoming a subject of study for many physicists. There is hardly a single institute where, to one degree or another, semiconductors are not being studied. The fact that at a conference where only original papers are reported, and which does not include time for review presentations, about 50 papers were presented is an indication of the intensive work being conducted by Soviet physicists in this field.

The conference was opened by Academician A. F. Ioffe. In his introductory address he noted that for modern physics the problem of semiconductors is just as urgent as the problem of the atomic nucleus. Its importance is explained by the fact that many properties of semiconductors, for example electrical conductivity, change by hundreds and thousands of times when minute quantities of impurities, on the order of tenths of a percent, are introduced into them. Taking temperature effects into account, this circumstance makes it possible, in one and the same semiconductor, to study the properties of a solid body at all intermediate stages—from a good insulator to a poor metal. In addition, semiconductors possess another valuable property. They are capable of changing their electrical conductivity under the influence of illumination. In some cases light causes an increase in electrical conductivity by tens of percent; in others, by tens and hundreds of times, and sometimes by hundreds of thousands and even millions of times. These properties of semiconductors are widely used in modern technology. It is enough to mention only selenium and cuprous rectifiers, used in technology for rectifying alternating current, and solid photoelements (with a barrier layer), which convert the light energy falling on them into electrical energy without any additional supply of energy from outside.

The papers presented at the conference covered five main topics: a) the external photoeffect in metals, semiconductors, and insulators; b) the dark and light conductivity of semiconductors; c) the crystalline photoef-

d) the internal photoelectric effect in dielectrics, and e) the valve photoeffect and rectification.

The main report on the external photoelectric effect was delivered by P. I. Lukirsky. In the phenomenon of the external photoelectric effect there is one feature which until now could not be explained. We are speaking of the selective photoelectric emission observed on certain surfaces. Excellent work by Ives and his collaborators showed for thin layers of alkali metals that their photoelectric emission is determined by the density of the light flux (the square of the light vector) at this surface. Developing Ives’s propositions, the speaker applies them to pure metallic surfaces of alkali metals. The starting point here is the following fact. Light penetrates into metals to a depth comparable with the wavelength of light, while the electrons are emitted outward only from a layer of thickness approximately \(10^{-6}\)–\(10^{-7}\) cm. Therefore there is no justification for calculating the photocurrent per unit of energy absorbed in the metal, as was done previously. The photocurrent must be referred to the unit of that energy which is absorbed in the thin upper layer and is directly expended in pulling electrons out. Such a calculation was made for mirror-smooth surfaces of alkali metals (liquid state). In this calculation it was assumed that the energy density in the layer from which the electrons are emitted outward is determined by the square of the light vector in the layer lying directly beneath the surface of the metal. This assumption can be made because of the small thickness of this layer in comparison with the depth of penetration of light into the metal. As a result of the calculations it was shown that the photoelectric emission, referred to the unit of energy absorbed only in the upper layer, reveals no selectivity and increases continuously in going from long waves to short ones. Thus the selective photoeffect (spectral and vectorial) for pure metals, after its 30 years of existence, must be abolished and assigned to the category of purely optical phenomena.

On the question of the dark and light conductivity of semiconductors, the principal reports were made by A. F. Ioffe and Ya. I. Frenkel. Then, in a series of reports, staff members of the Leningrad Physico-Technical Institute presented concrete experimental material on the same question.

In Ya. I. Frenkel’s report a quantum-mechanical model of an electronic semiconductor was analyzed. In the crystal lattice of a semiconductor, a portion of the electrons, by thermal motion, is transferred from the normal continuous filled band of levels into the conduction band. These transferred electrons determine the conductivity of the semiconductor. However, the essential point in the scheme presented—the point especially emphasized by the speaker—is the circumstance that, in the conductivity of electronic semiconductors, in addition to the transferred electrons, so-called “holes” also take part (they should not be confused with Dirac’s “holes”). When an electron is removed from the filled band into the conduction band, free quantum levels appear in the former, which may be occupied by electrons from other levels of the same band. The levels thus vacated are in turn filled by new electrons, and so on. This process corresponds to the displacement of electric charges. The same thing can be represented more vividly if one passes from the energetic description to a kinematic one.

When an electron is removed from a given region of the crystal, a positive charge remains in it. This formation Ya. I. Frenkel calls a “hole.” The “hole” that has arisen at this point is filled by an electron displaced from another region of the crystal, where, in turn, a “hole” is formed, which again is filled by an electron from another neighboring region, and so on. If such a crystal is placed in an electric field, then the replacement occurs predominantly in the direction of the electron flow. Thus the region with the missing electron—the “hole”—moves in the elec-

tric field in the direction opposite to the motion of the electrons. The transfer of charges occurs exactly as if positive charges were moving, although in reality the positive ions do not shift from their places in the crystal lattice at all, and only the electrons move.

By the motion of “holes” Ya. I. Frenkel explains a number of anomalous phenomena observed in semiconductors, for example the negative Hall effect, negative thermoelectromotive forces, and a number of others.

For the electrical conductivity of electronic semiconductors, impurities play an essential role. Some impurities are a source of additional conduction electrons, while others capture electrons and fasten them to themselves (acceptors).

Acad. A. F. Ioffe, in his extensive report, dwelt in detail on the thermal and light conductivity of semiconductors, relying mainly on experimental material obtained by him jointly with A. F. Ioffe on cuprite crystals. In this report, above all, the mechanism of electrical conductivity was emphasized—by electrons and by “holes.” It is pointed out that for semiconductors with a small amount of impurities at low temperatures, the greater part of the conductivity is due to electrons, which primarily come from impurities. The remaining positive charges in this case are localized on the impurities and do not participate in current transport. Conversely, at high temperatures, when electrons are taken mainly from the zone of occupied levels, the “holes” participate to a significant degree in the transfer of electric charges. True, it must be noted that not all impurities act in the same way.

To explain the photoconductivity of cuprite, in addition to the principal absorption band and the conductivity band, one must allow for the existence of intermediate levels connected, in all probability, with inhomogeneities of the lattice. The ejection of electrons from these additional levels into the conductivity band corresponds to additional absorption in the infrared region.

Among the phenomena observed in cuprite, let us note the unipolarity of the photocurrent and the negative photocurrent. When cuprite is illuminated through a semitransparent gold electrode, a greater or lesser photocurrent is observed in the crystal depending on whether the illuminated electrode is the cathode or the anode. In this case, a rather sharply expressed unipolarity of the photocurrent appears only for strongly absorbed light. It can be explained by the fact that in the second case the electrons are rapidly removed from that layer of the crystal in which appreciable absorption of light occurs; in the first case, however, the photoelectrons create an increased concentration throughout the entire thickness of the crystal. Negative photocurrent—a decrease in conductivity upon illumination—is observed in strong electric fields if the number of photoelectrons is relatively small compared with the number of thermal ones. This phenomenon can be explained in the following way. Electrons raised from the principal band into the conductivity band are carried out of the crystal by the strong field. As a result of the rarefaction of charges that has occurred, the recombination of thermal electrons increases. If the number of thermal electrons put out of action is greater than the number of photoelectrons that have arisen, then a negative photocurrent will occur.

In his next report A. F. Ioffe communicated research on the crystalline photoeffect in cuprite crystals. When a cuprite crystal is illuminated (through a semitransparent electrode) by light absorbed in it, a certain potential difference arises between the electrodes. It may be obtained from Teykhman’s formula if one assumes that this electromotive force is a consequence of different concentrations of photoelectrons at the one and the other electrode. The speaker showed that, qualitatively, Teykhman’s formula describes the crystalline photoeffect correctly. In this case, in one limiting case—of a small number of photoelectrons compared with the number of thermal ones—a linear dependence is obtained

of the photodiffusion electromotive force on \(\frac{T}{\sigma}\), where \(\sigma\) is the conductivity corresponding to the temperature \(T\). In the other limiting case, of a large number of photoelectrons in comparison with the number of thermal electrons, it increases proportionally to \(T\). As regards the dependence of the photoelectromotive force on the intensity of the incident light, theory and experiment show linearity for the first limiting case and a more complicated dependence, with saturation, for the second (for example, at the temperature of liquid air). The basic assumption that has to be made in order to explain the facts observed in cuprite is that the photoelectrons do not remain in the place in the crystal where they are produced, but are scattered over its dark regions. Consequently, there must exist in the crystal some electronic currents that produce this mixing.

Starting from this proposition, one could try to replace the ratio of free electrons at one and the other electrode by the ratio of conductivities in the corresponding places. However, even in this case one has to state that, for quantitative calculations, Teichmann’s formula is of little use. The observed photoelectromotive forces always exceed the calculated ones.

An interesting report was given by I. K. Kikoin on a new photomagnetoelectric effect, recently discovered by him jointly with M. M. Noskov. The new effect is as follows: if a plate of pure polycrystalline cuprous oxide is placed in a horizontal magnetic field directed along the large surface of the plate, and illuminated with light perpendicular to the magnetic field, then between electrodes deposited on the upper and lower faces there appears a large potential difference, reaching, at the temperature of liquid air, 20 V. At room temperature the effect is very small. The most interesting point is that the transverse potential difference is observed in the absence of a macroscopic current in the plate. The photomagnetic electromotive force that arises increases linearly with the magnitude of the magnetic field and changes its sign when the direction of the field is changed. It depends only weakly on the light intensity, growing strongly only at low intensities. The photocurrents observed with closed transverse electrodes, however, increase linearly with increasing intensity of the incident light.

Red light, in contrast to other light, causes a decrease of the effect; this explains the increase of the effect when an aqueous filter is placed in the path of the white light beam falling on cuprous oxide. To explain this effect it is necessary to admit the existence of two compensating electric currents, which are deflected differently in a magnetic field. Two explanations were proposed. In the first explanation (Frenkel’s), the role of the compensating currents is played by photoelectrons and “holes” moving in one and the same direction and deflected in the magnetic field in different directions. In this case the observed effect is the total effect. In the second explanation (A. F. Ioffe’s), the role of the compensating currents is played by the stream of photoelectrons and the compensating stream of thermal electrons (in the opposite direction). The streams are deflected in the magnetic field in different directions. In this case the different degree of deflection of these streams is based on the fact that the excess photoelectrons diffusing from the illuminated place consist on average of faster electrons in comparison with the thermal ones. In this case the effect should be of opposite sign.

In the reports devoted to the internal photoeffect in dielectrics, the chief question was that of establishing the energy levels of the electrons in the crystal lattice. The principal report on this question was given by P. S. Tartakovskii. He set forth a scheme of electronic levels for rock salt, constructed by him on the basis of certain literature data and of work carried out at the Siberian Physico-Technical Institute (Tomsk).

In addition to the main energy band, from which electrons are taken when rock salt is colored by ultraviolet or X-rays,

and the conduction band, two intermediate levels are included in the scheme. One of them corresponds to the sodium atom (the atomic color center). The other appears in the depolarization current (experiments of Kalabukhov and Fishelev) and in the passage of current through the salt when it is bombarded with electrons (Fishelev). From studies of the internal photoeffect in X-rayed rock salt and from studies of its ultraviolet fluorescence under illumination by visible light, the relative position of these levels has been established. On the basis of other experiments by the speaker, the position of these levels relative to the top of the potential barrier has also been established.

On the last question—the barrier-layer photoeffect and rectification—Acad. A. F. Ioffe presented the proposed and generalized theory of solid photoelements and rectifiers, representing a further development of the views of Auers and Kerschbaum. The speaker proceeded from the dependence of the resistance of the barrier layer on voltage. Introducing four constants, the author derives the basic regularities of the phenomena occurring in solid photoelements. Experimental material illustrating the theory was reported by a number of his collaborators. However, in order to construct a generalized theory it is necessary to take the dependence of the resistance of the barrier layer not only on voltage, but also on the intensity of light, as was noted by A. F. Ioffe.

Of substantial importance for understanding the processes occurring in solid photoelements and rectifiers is the work of V. P. Juzé, reported at the conference by A. F. Ioffe. Juzé deposited on a cuprous-oxide plate, on one side, artificial barrier layers made of materials of high specific resistance, and on the other, metallic electrodes without contact resistance. To investigate the dependence on thickness, the barrier layer was applied in the form of a wedge. A series of small gold electrodes was deposited on the insulating wedge (by evaporation in vacuum).

Juzé’s experiments established that, when a barrier layer was deposited from materials not possessing photoconductivity (quartz, \(B_2O_3\), Bakelite), elements with rectifying action were obtained in the complete absence of the barrier-layer photoeffect. The magnitude of the rectification coefficient depends both on the nature of the barrier layer and on its thickness, reaching a certain maximum value at thicknesses on the order of several units of \(10^{-6}\) cm. When a barrier layer of thallium sulfide, which possesses photoconductivity, was deposited, elements with photoelectric and rectifying properties were obtained. Hence one may conclude that, in the barrier-layer photoeffect, the photoelectric properties of the barrier layer itself play an essential role.

In his report, M. P. Bronstein showed how it is possible, at least qualitatively, to explain the experiments of V. P. Juzé on the basis of the quantum-mechanical model of a semiconductor. V. N. Lepeshinskaya reported on work being carried out at the Central Radio Laboratory on selenium photoelements with a barrier layer.

At the closing session A. F. Ioffe, in his final remarks, emphasized the most essential points in the work of the conference and noted the most important questions that must be addressed first of all. At the same session organizational questions were considered. A bureau of the All-Union Brigade for Semiconductors was elected, and separate areas of work on semiconductors were distributed among institutes and laboratories.

It should be noted with great gratitude that an enormous amount of work was carried out by the staff of the Odessa Physical Institute to provide the conditions for the fruitful work of the conference.