File: SECT01.WM of Tape: Various/ETH/s10-diss
(Source file text)
- 1 - @ka I. INTRODUCTION. ________________ @ke @ba This work will be primarily concerned with the anisotropy of the conduction electron relaxation times, in the context of conduction phenomena, i.e. resistivity and Hall effect. Our main effort will be to explain the influence of impurities introduced, substitutionally, in an otherwise ideal metal. We shall now briefly discuss the developments leading toward the concept of anisotropic relaxation times and the reasons why we only discuss the simple trivalent metals. @be @ka 1). Historical background. _________________________ @ke @ba The concept of relaxation times was introduced by Drude(1902) and Lorentz(1909) shortly after the discovery of the electron. They developed simple phenomenological theories based on the kinetic arguments of the theory of gases, where the relaxation time was a parameter describing the time between collisions of the conduction electrons. Amongst other problems, this relax- ation time, or in other words, the mean free path, turned out to be much longer than the interatomic distance when the theory was applied to low temperature, or even to room temperature phenomena. At that time it was difficult to explain how an electron could move over several thousand atoms before it was removed, by succesive scattering events, from the conduction process. @be - 2 - @ba The advent of quantum mechanics solved most of the problems. Sommerfeld(1928) used Fermi-Dirac statistics to show that it was necessary to consider only a small part of all the avail- able electrons in order to explain the conduction process. The ideas of Bloch(1928) about periodic potentials led to the fact that, in a first order theory, the electrons should experience no resistance whatsoever in a perfect lattice at zero tempera- ture. These advances in the theoretical description of the conduction mechanism reversed the situation: one had to show why metals have a resistance at all. (i.e. why they are not all "superconductors"!). This led to refinements of the models which now had to include imperfections of the perfect lattice. @be @ba The problem can be roughly separated into two types of dis- turbing effects of the ideal lattice; the first has to do with displacements of host atoms from their equilibrium positions, the second with the replacement of host atoms by other atoms. Naturally this division is rather artificial, but it seems that the two subjects were often treated separately. The first part was mainly concerned with the intrinsic resistance of the pure metal attributed to the thermal motion of lattice. The first theory adequate for a large temperature range and explaining experimental results rather well was proposed by Grueneisen (1933). The second aspect dealt with the deviations of the periodicity of the lattice due to the change in atomic poten- tials themselves and was mainly used to explain the resis- tivities due to foreign atoms, vacancies, etc. @be - 3 - @ba For a long time it was assumed that the relaxation times were isotropic, i.e. independent of the location in wave number space of the electronic wave packet being scattered from one place of the Fermi surface to another. At a relativily early date (see Wilson(1953)) two- band models were introduced, but these models had no direct connection with anisotropic relax- ation times; they were phenomenological theories used mainly to separate contributions, with different effective masses, from different parts of the Fermi surface. @be @ba In this same period of time the influence of the boundaries (size effects) was investigated. A review by Sondheimer(1952) showed that these size effects were usually applied in a free electron model and that they were assumed to modify the mean free paths by direct geometric effects rather than by an influence of the boundary on the intrinsic bulk properties of the crystal. @be @ba Anisotropic relaxation times definitely became fashionable when the ideas of Ziman (1960,1961) explained most of the conflict- ing experimental results in the bulk noble metals. He proposed a model incorporating the main features of the Fermi surface and the anisotropies of the electronic relaxation times, which explained many of the difficulties. After this, the subject of anisotropic relaxation times became an active research topic and the first conference dealing specifically with this topic took place in Zuerich. (Chambers,Olsen,Ziman(1968)). @be - 4 - @ba In considering the reasons for preferring investigations in trivalent simple metals, we may note that experimental work relating to conduction phenomena has been the most intensive in the noble metals group. (see for instance Alderson;Hurd (1970)) Although very many experimental facts are known about these metals, it remains that the theoretical treatment is complicated by the fact that their Fermi surfaces are rather strongly influenced by the vicinity of the d-bands and have to be described by augmented plane wave methods. @be @ba The advantage of the trivalent metals aluminum and indium is that they satisfy the small core approximation needed for the pseudopotential theories. Their Fermi surface is accurately described by the orthogonalized plane wave formalism which uses matrix elements derived from pseudopotentials. We think that such a description is much better suited for discussions in terms of the standard text book methods of electronic conduc- tion. We will treat these two metals from this view point and develop simple models to describe the Hall effect and the resistivity, in a way closely related to the work of Sorbello. (for a recent review see Sorbello (1975)) @be - 5 - @ka 2). Organization. _________________ @ke @ba In Sect.II we first give an introduction to the characteristics of these two metals, which we used in our calculations of the Fermi surface properties. As our main concern is in measuring relaxation time effects in alloys of these two metals, we further discuss the range of solutes available and the reasons why some solutes should be better suited than others for our kind of measurements. Some collected data is also presented concerning lattice distortions, which will come to use in later sections. @be @ba The next section III handles the problem of scattering centers. It starts off with a presentation of newly compiled free elctron data of the two metals in question and of their solute candidates. This data was also given for future reference, because we found that the constants used in low temperature calculations were often erroneous. @be @ba In Sect.III.2, two pseudopotential formalisms are discussed. The first is the simple Ashcroft empty core pseudopotential and we give our new values of it's single parameter. It was found that this potential was not realistic enough for scattering calculations and we also present the Shaw potential as modified by us for the case of solutes in solution in some other metal. We give our re-evaluated parameters for the pure metal case and for the solute case. @be - 6 - @ba Sect.III.3 shows how to construct a scattering potential from the last mentioned pseudo potentials and includes a discussion of lattice distortion effects. A new series of phase shifts, parametrizing the scattering properties of solutes in Al and In are the main result of Section III. @be @ba For dealing with the resistivity and the Hall effect, we recap- itulate the theory of Sorbello and add some extra features con- cerning the character of the wave functions of Al and In on certain parts of the Fermi surface, in Sect.IV.1. Our extension of this theory to the situation in a magnetic field is given in Sect.IV.2. The result is given by a general expression for the conductivity tensor in a magnetic field and also an expansion in powers of H is given for the treatment of low field effects. Finally a simple 3-band model is proposed which, we hope, should describe the main features of the Fermi surfaces. @be @ba In Sect.V our experimental results on the Hall effect and the resistivity in dilute Indium alloys is presented. We show that the important sign reversals in the Hall effect cannot be due to rigid band effects. At the end of this section our exper- imental results are discussed in a quantitative manner, using the theoretical results obtained in the earlier sections, concerning scattering potentials, phase shifts, character of the wave functions, etc., in the context of a 3-band model. @be @ba We also give some results for the resistivity of Al based alloys in connection with our de Haas- van Alphen (dHvA) experiments as discussed by Wejgaard (1975) and compare some experimental Hall effect data by Boening et al. (1975). @be