OTHER TITLES IN THE SERIES ON SEMICONDUCTORS Vol. 2 Photo and Thermoelectric Effects in Semiconductors. JAN TAUC. Vol. 3 Semiconductor Statistics. J. S. BLAKEMORE. Vol. 4 Thermal Conduction in Semiconductors. J. R. DRABBLE and H. J. GOLDSMD. SEMICONDUCTING III-V COMPOUNDS by G. HILSUM and A. C. ROSE-INNES Services Electronics Research Laboratory Baldocky Hertfordshire England y PERGAMON PRESS OXFORD · LONDON · NEW YORK · PARIS 1961 PERGAMON PRESS LTD. Headington Hill Hall, Oxford. 4 and 5 Fit&oy Square, London W.l PERGAMON PRESS INC. 122 East 55th Street, New York 22, N.Y. Statler Center 640, 900 Wibhire Boulevard, Los Angeles, California. PERGAMON PRESS S.A.R.L. 24 Rue des Écoles, Paris V. PERGAMON PRESS G.m.b.H. 75 Kaiserstrasse, Frankfurt am Main. Copyright © 1961 Pergamon Press Ltd. Library of Congress Card Number 60-53574 Set in Baskerville irfispt. and printed in Gnat Britain by Spotnswoodt, Ballantyru & Co., Ltd., London and Colchester To our wives BETTT and BARBARA who spent so many hours helping with this book PREFACE IT is now ten years since the semiconducting properties of the III-V compounds were first reported. Since then several hundred papers and articles have been written about them, and a stage has now been reached where a broad study of these materials is possible. We have attempted such a study with two objects in mind: firstly, to describe the properties of III-V compounds as a family of semiconducting crystals and, secondly, to relate these compounds to the better known and better understood monatomic semiconductors, silicon and germanium. The rapid advance that has been made in semiconductor physics and devices is a notable example of how pure research and technology can stimulate each other, and for this reason we have thought it important to include a chapter on applications of III-V compounds. In this book we have assumed that the reader is acquainted with the fundamental ideas about semiconductors and that there is no need to explain basic concepts, such as "A-space", "hole" or "donor impurity". We have also assumed that the reader has some knowledge of the pro­ perties of germanium and silicon, and we often compare the properties of III-V compounds with those of these "prototype" covalent semi­ conductors. Even in so specialized a topic as semiconducting III-V compounds it is impossible to refer to all the published work, and it would not be desirable to include references simply for fear of leaving somebody out. We have chosen references either as sources of information given in the text or as suitable further reading and we are sorry if any injustice has arisen from this choice. It is a pleasure to acknowledge the many helpful discussions we have had with the members of the Solid State Division at the Services Elec­ tronics Research Laboratory, and we especially thank Mr. J. W. Allen, Dr. F. A. Cunnell, Dr. E. H. Rhoderick, Dr. O. Simpson and Mr. R. I. Walker for reading and commenting on the manuscript. We are grateful to the many authors who have given us permission to quote from their unpublished work. C.H. A. C. R.-L Services Electronics Research Laboratory Baldock, Herts August 1960 Chapter 1 INTRODUCTION ATOMS from group Illb of the periodic table combine with atoms from group Vb to form crystalline semiconducting compounds. We call these semiconductors "III-V compounds". They are not alloys but definite chemical compounds with a 1:1 atomic ratio between the III and V atoms, which occupy alternate sites in the crystal lattice. For example, if we melt a mixture of indium and antimony, the compound InSb is formed, and, on solidification, any excess of either constituent is thrown out as a second phase. Ill atoms have one valence electron less, and V atoms one electron more than those atoms from group IV of the periodic table which form the semiconducting crystals, diamond, germanium, silicon and grey tin. III-V compounds have, therefore, the same average number of electrons per atom as the group IV semiconductors, and it is found that the com­ pounds indeed have a crystal structure and electronic properties which are in many ways similar to those of the group IV semiconductors. Nevertheless, III-V compounds possess characteristic properties which distinguish them from the group IV semiconductors. These differences arise chiefly from the fact that the compound crystals have a lower symmetry than the group IV crystals and that, whereas the group IV semiconductors consist of covalently bound neutral atoms, the III-V compounds contain positive and negative ions at the lattice sites. The first of the III-V compounds to be reported was InP, prepared over fifty years ago by Thiel and Koelsch (Tl). Huggins (HI) pointed out in 1926 that binary compounds formed between group Illb and group Vb elements should crystallize in a form similar to the diamond structure taken up by the group IV semiconductors, and this was con­ firmed by the experiments of Goldschmidt (GÌ), who in 1929 investigated the crystal structures of InSb, GaSb, GaAs, GaP, AlSb, AlAs, AIP and AIN. The fact that one of the compounds, InSb, is a semiconductor akin to germanium and grey tin, was reported in 1950 by Blum, Mokrovski and Regel' (Bl) and by Goryunova and Obukhov (G2). Welker (Wl), i 1 2 INTRODUCTION however, appears to have been the first to appreciate the importance of the III-V compounds as a new family of semiconductors, when in 1952 he described the semiconducting properties of several of these com­ pounds and drew attention to some of their special properties, such as the very high electron mobility and small energy gap of indium antimonide. Binary IIIa-Vb compounds have been prepared (II) between the Ilia rare earths and the Vb elements, for example, CeAs, PrN, but these compounds have the sodium chloride crystal structure and are probably essentially ionic crystals. No compounds or alloys between group Ilia -and Va elements have been reported and it seems likely that they do not exist Lanthanum and niobium are insoluble in each other, both in the liquid and solid state (SI), and lanthanum and tantalum also appear to be mutually insoluble in both states. Though III-V crystals have a structure similar to that of diamond, there are two kinds of atom in the unit cell, and so they form a more general system than the group IV semiconductors. There are a large number of possible combinations and these compounds cover a wide range of semiconducting properties, useful both for a study of semi- conduction and for the manufacture of devices. In spite of the fact that the compounds have two different atoms in the unit cell, they sometimes show simpler behaviour than the monatomic group IV semiconductors. Indium antimonide and indium arsenide, for example, approximate rather closely to the simplest model of a semiconductor in which a spherical conduction band and a spherical valence band both he at the centre of the zone. On the other hand, III-V compounds are usually more difficult to prepare in a pure form than the group IV semiconductors, and, for many of the compounds, the measured properties are those governed by impurities rather than those of the pure material, even for the purest samples yet prepared. In Table 1.1 we list some of the important properties of the III-V compounds. Indium antimonide has probably been the most intensively studied because it has remarkable properties resulting from its small energy gap and very high electron mobility. Furthermore, it is relatively easy to grow pure single crystals of this material. In several of the com­ pounds, the mobility of the electrons is much greater than that of the holes. As a result, the familiar semiconductor terms, "n-type" and "/>- type", must be used with some caution. For instance, indium antimo­ nide shows properties characteristic of negatively charged carriers, such as a negative Hall coefficient, even when there are a hundred times more INTRODUCTION 3 Table l.I. Some properties of III-V compounds and group IV semiconductors Hall mobility at room Energy gap Melting temperature* Crystal at room point (cm*sec-W-1) Remarks structure temperature (°K) (eV) Electrons Holes BN hexagonal, also ZB ~10-0 A1N Würz >2400 GaN Würz 3-3 InN Würz BP ZB 2000-3000 5-9 100 AIP ZB 30 unstable in air GaP ZB 1450-1500 2-25 110 75 InP ZB 1062 1-29 4600 150 BAs ZB AlAs ZB >1600 2-16 unstable in air GaAi ZB 1237 1-4 8500 420 InAs ZB 942 0*36 33,000 460 BSb ZB? AlSb ZB 1050 1-62 200 420 unstable in moist air GaSb ZB 712 0-67 4000 1400 InSb ZB 525 0-17 78,000 750 InBi tetrag 103 metallic TIBi CsQ? 230 metallic C diamond -5-2 Si diamond 1420 1-08 1450 500 Ge diamond 958 0-66 4500 3500 a-Sn diamond (150) 0-08 Würz a wurzite; ZB » zinc-blende; tetrag » tetragonal. * Best experimental value reported. 4 INTRODUCTION holes than electrons. We shall use the term " n-type " to describe material in which there are more conduction electrons than holes, and "/>-type" to describe material with more holes than electrons, irrespective of the sign of the measured electrical properties. Many physical processes which are of importance in deciding the behaviour of III-V compounds are of little importance, or do not occur, in monatomic semiconductors such as germanium and sihcon. In this book we have tried to concentrate on these processes which are peculiar to III-V compounds, and not to dwell on those aspects which are com­ mon to semiconductors in general. REFERENCES Bl. BLUM, A. N., MOKROVSKI, N. P. and REGEL', A. R., Seventh All'Union Conference on the Properties of Semiconductors, Kiev (1950). Gl. GOLDSCHMIDT, V. M., Trans. Faraday Soc. 25, 253 (1929). G2. GORYUNOVA, N. A. and OBUKHOV, A. P., Seventh All-Onion Conference on the Properties of Semiconductors, Kiev (1950). HI. HUOGINS, M. L., Phys. Rev. 27, 286 (1926). II. IANDELLI, A. and BOTTI, E., Atti accad. nazi. Lincei 25, 129, 498, 638 (1937). SI. SAVITSDI, E. M., TEREKHOVA, V. F. and BUROV, I. V.,Zhur.Neorg. KhimA, 1462 (1959). Tl. THIEL, A. and KOELSCH, H„ Z. anorg. Chem. 65-66, 288 (1910). Wl. WELKER, H., Z. Naturforsch, 11, 744 (1952). Chapter 2 CRYSTAL STRUCTURE AND BINDING 2.1 CRYSTAL STRUCTURE NEARLY all the III-V compounds crystallize in an arrangement where each atom is at the centre of a regular tetrahedron, at the four corners of which he atoms of the other kind. These tetrahedra can be arranged into two forms of crystal structure: " zinc-blende'' which is cubic, and " wurzite" which is hexagonal. The cubic zinc-blende (or "sphalerite'') structure is the same as the diamond form except that the two different kinds of atom occupy alternate positions in the lattice (Plate I). The III and V atoms each lie on a face-centred-cubic sub-lattice, the two sub-lattices being displaced relative to each other by one quarter of the body diagonal of the cube. The wurzite structure is similar to zinc-blende except that alternate (111) layers are rotated through 180° about the [111] axis, giving the structure hexagonal symmetry. The twelve compounds that boron, aluminium, gallium and indium make with phosphorus, arsenic and antimony all have the zinc-blende structure. The nitrides of aluminium, gallium and indium have the wurzite form. Boron nitride normally has a hexagonal structure similar to graphite, though tiny crystals of the zinc-blende form ("Borazon") have been prepared (§ 5.3). The bismuthides appear to be metallic compounds without tetrahedral structure. The forms and constants of the crystal structures are collected in Table 2.1. We can assign a characteristic radius to each III and V atom, such that the separation between the nuclei of neighbouring atoms in the tetrahedrally bonded compounds is the sum of the two atomic radii. This "tetrahedral radius" is virtually independent of the compound in which the atom finds itself. In Fig. 2.1 are given the values of the tetra­ hedral radii for III and V atoms. As an example, the sum o fthe radii of indium and antimony is 2-80 Â, which is the separation o fthe nuclei of these atoms in indium antimonide. 5