Table of contents

The implications of 1-D interstitial migration for dissociative impurity diffusion are discussed. It is pointed out that, in contrast to the case of 3-D migration, with 1-D interstitial migration long-range impurity diffusion cannot take place without vacancy motion. Further it is shown that, in the direction of interstitial migration, a value of isotope effect parameter E of E approximately=0 may be expected. This is in contrast to the correlation factor f for which, under the same conditions, a value of f>0 would be expected.

The ¹²⁵Te Mossbauer transition following radioactive decay from the ¹²⁵Sb parent reveals a field of mod H mod =848+or-30 kOe at 4.2K, the largest field measured to date at a Te nucleus or in any Heusler alloy.

Experimental evidence for the existence of atomic and magnetic short-range order in 16.7 at.% MnCu is examined in relation to the magnetic susceptibility at low temperatures and its dependence on heat treatment. A neutron scattering experiment indicates the existence of magnetic clusters of size greater than 35 AA at 4.2K. A model for clusters is proposed, based on a lattice of second- and third-nearest-neighbour Mn spins, and which is centred on regions of atomic short-range order about 18 AA in size.

An analysis is made of the exact treatment for the orthogonalization hole potential, denoted by nu _OPW^*, in Harrison's first principle pseudopotential theory (1966, 1969). A corrected form of Harrison's expression for the approximation of nu _OPW^* is given and its effect on the calculation of phonon spectra for lithium, sodium and zinc are discussed. To demonstrate the effect of using the 'exact' formulation for the orthogonalization hole potential, the nu _OPW^* and an exact orthogonalization hole density are used in the calculation of a nonlocal pseudopotential and phonon dispersion curves for zinc.

The repulsive and electronic contributions to the phonon frequencies have been calculated for metallic lithium in the symmetry directions. It is shown that the former are truly negligible as is usually assumed. For the latter the energy wavenumber characteristic is calculated using the electron-ion matrix elements for energy conserving transitions on the Fermi surface and then extended, by fitting the elastic constants, to simulate, in turn, local and nonlocal approximations. It is demonstrated that the nonlocal approach is essential in a complete theory of transport properties and phonon frequencies. Also, the crossword in the (100) branches can be explained as a reflection of the electron-ion matrix element characteristic of lithium. There are indications that anisotropic effects in the electron-ion matrix element may be important and that perturbation theory may not be adequate in lithium metal.

Thick (>2000 AA) epitaxial films of nickel and copper grown on silver (111) have been annealed in ultra-high vacuum at temperatures in the range 500-600K. This annealing procedure appears to cause diffusion of silver to the outer surface to take place, with the formation of an 'equilibrium layer' of silver. The silver layer appears to be strained in each case to give an integral ratio matching with the nickel or copper, the matching being 6:7 for Ag:Ni and 8:9 for Ag:Cu. A comparative study was made for thin films of silver deposited directly on to copper (111). LEED, RHEED and Auger spectroscopy were used throughout the study. Extra non-primary satellite beams in both LEED and RHEED from thin layers of Ag on Cu or Ni are believed to result from a combination of multiple diffraction and periodic deformations in the overlayer.

An energy independent model pseudopotential formalism for calculating the electronic structure of metals is developed so that uncertainties in the determination of the model potential parameters, such as those occurring in the Heine-Abarenkov type model potential due to the energy dependence, can be avoided. A model potential, in general more suitable than the square-well potential, with parameters determined by fitting to the energy levels of the low-lying atomic valence states is presented and is used in the construction of the model pseudopotential for the applications to metals. An expression for the form factor, including the effective mass and exchange-correlation effects, is derived in a way consistent with the proposed model pseudopotential.

For pt.I see ibid., vol.5, p.1836 (1975). The energy independent model pseudopotential (EIMP) formalism, developed in the preceding paper, and Shaw's optimized model potential (OMP) formalism are applied to first principles calculations of phonon spectra of the alkali metals and Al and of Fermi surface distortions for the alkali metals. Further, when compared with existing APW calculations for Fermi surface distortions, the relativistic effects due to the deep-core electrons, which is taken into account implicitly through the model potential parameters in the EIMP calculation, are found to be rather significant for heavier metals.

Monochromatized XPS experiments show that the 5d level of Au diluted into Cd evolves from an atomic-like spin-orbit split doublet into energy bands as the number of nearest neighbour Au atoms increases. The rapid rise to higher energy of the antibonding-like portions of the 5d bands with increasing Au concentration is attributed to the nonorthogonality of the d orbitals on adjacent sites.

A Ruderman-Kittel interaction between magnetic ions is assumed and the defect is represented by a change in the hopping integral of the conduction electrons. If the hopping integral across the defect is substantially larger than the perfect crystal value, the direction of magnetization reverses across the defect. This is a possible explanation of the observed pinning of domain walls to antiphase boundaries in Cu-Mn-Al Heusler alloys and in MnAl.

The electronic contribution to the energy of formation of transitional alloys A_cB_1-c is determined from their band structure. Preliminary numerical results are presented for nickel and iron paramagnetic alloys and it is shown that the variation of the energy of formation with concentration is not very sensitive to the detailed local environment, but is mainly determined by the average surrounding of A and B atoms.

When an anisotropic relaxation time is rigorously defined through the Boltzmann equation, it depends on the applied fields. Therefore, the relaxation times to be employed when using the path integral to evaluate the magnetoconductivity at high magnetic fields are different from those used for the electrical conductivity. The relationship between these two sets of scattering times is discussed proceeding from the path integral. It is argued that for the longitudinal magnetoconductivity it is a reasonable first approximation to use the same relaxation times for any magnetic field strength. Calculations are presented of the temperature and orientation dependence of the longitudinal magnetoconductivity of pure potassium, based on relaxation times previously calculated by Hayman and Carbotte (1972).

Electrical ( rho ) and thermal (W) resistivity measurements in very dilute In alloys between 1K and 4K show a temperature dependence expressed by rho = rho ₀+ beta T⁵ and WT=(WT)₀+AT³+BT⁵. The BT⁵ term in the thermal resistivity is due to a contribution from a 'diffusion-like process' of electrons over the Fermi surface, as is the one that describes the temperature dependence of the electrical resistivity. This claim is supported by the experimental fact that the Lorenz number defined as L= beta /B coincides with the value given by L= rho ₀/(WT)₀ which in turn agrees with the free electron value within 15%.

The scattering formalism of Blaker and Harris (J. Phys. C., vol.4, p.569, 1971) has been extended to hexagonal systems with magnetic impurities and applied to the behaviour of the Dingle temperature in ZnMn and ZnCr alloys. The calculations represent a substantial improvement over previous analyses and confirm the necessity of including band structure information in the description of the scattering.

The superconducting transition temperature T_c has been determined for a number of technetium-based h.c.p. solid solution alloys. Additions of V, Nb and Mo raise T_c whereas Co, Ni, Ru, and, in particular, Cr and Fe lower T_c relative to pure technetium. The general trend of T_c with the electron per atom ratio is consistent with that expected for a simple dependence on a rigid-band density of states, with the exception of that for the Cr and Fe alloys. An analysis of the data for these latter alloys in the context of the theory due to Kaiser (J. Phys. C., vol.3, p.410, 1970) would suggest that the enhanced depression of T_c is due to the formation of a nonmagnetic resonance state. The hydrostatic pressure dependence of T_c has been investigated up to 25 kbar with particular attention given to the h.c.p. region of the Mo-Tc-Ru system. The observed variation of delta T_c/ delta P is discussed in relation to current theoretical models.

The present study establishes hyperfine field systematics at the non-magnetic sites in Heusler alloys and emphasizes the effect of local screening on the hyperfine field value. A range of Heusler alloy hosts were doped with 1 at.% of the isotope ¹¹⁹Sn for examination using the Mossbauer effect. The results reported indicate that the hyperfine field at a given site is strongly dependent in magnitude and sign on the magnetic character of the local screening charge. The fields at Sb sites in several alloys have been measured using the Mossbauer isotope ¹²¹Sb and at a Cd impurity site in Ni₂MnIn using the method of time-differential perturbed angular correlations. The Sn site fields have been measured in the Co based alloys Co₂TiSn, Co₂ZrSn and Co₂HfSn. The similarity of the values obtained is interpreted in terms of the similar outer electron structure of these alloys.

The observation of well defined spin waves and Stoner excitations above T_C in nickel by Mook et al. (Phys. Rev. Lett., vol.30, p.556, 1973) is explained by assuming that the spin density in nickel varies in direction only and that it does so slowly over distances small compared with the correlation length. This picture is assumed to be also true in the itinerant model of magnetism, and its consequences are developed. The magnetic excitations near T_C resemble those at much lower temperatures. The intensity of the polarized neutron inelastic scattering by spin waves is predicted to be a linear function of magnetization, the slope of which gives the magnitude of the local spin density. Also, possible explanations are given as to how it is possible for a magnetic phase transition to take place without the observed Stoner band splitting vanishing at T_C.

The magnetic specific heat C_M and susceptibility chi are calculated for dilute magnetic alloys based on theoretical results previously obtained by Riess and Mavroyannia (Physica, vol.75, p.483, 1974). Those results indicated that a broad maximum in C_M and chi was not associated with internal magnetic fields but rather magnetic 'fluctuations' about zero field. A computer simulation of a dilute alloy gives the distribution function, n(v), of those 'fluctuations' from which C_M and chi are obtained. A nearest neighbour (NN) approximation for n(v) yields better results for C_M and chi than the full computer-generated expression. In particular in the NN approximation, C_M varies as T and chi is finite for T to 0. In both cases chi has a rounded maximum which is not associated with antiferromagnetism but is due to the 'fluctuations'.

Direct measurements of the latent heat reveal that the transition between the paramagnetic and the incommensurate spin density wave structure is of the first order while the transition becomes of the second order when it is between the paramagnetic and the commensurate spin density wave structure.

The isomer shifts of the 86.5 keV gamma ray of ¹⁵⁵Gd in caesium chloride type pseudobinary compounds of Gd with Ag, In, Cu and Ga have been measured. It is proposed that as group IIIB elements replace IB elements the density of s-like electrons in the conduction band decreases.

A study of the nuclear magnetic resonance spectrum of ⁷¹Ga in beta-gallium has been made in the temperature range 77-242K. The results of this experiment combined with the temperature dependence of the pure quadrupole resonance of ⁶⁹Ga also reported here allows one to determine the value and temperature dependence of the following parameters: isotropic Knight shift (K_iso=+0.381%+or-0.005 at 77K), anisotropic Knight shift (K'=+0.027%+or-0.005 at 77K), asymmetry parameter in the electric field gradient ( eta =0.376+or-0.005 at 77K), and the quadrupole resonance frequency, nu _Q⁰, a parameter independent of the asymmetry parameter ( nu _Q⁰(⁷¹Ga)=3.830+or-0.003 MHz at 77K). Relaxation time measurements indicate that the longitudinal relaxation time, T_1Q, is almost entirely due to magnetic dipole relaxation and not quadrupole relaxation.

The ratio of the nuclear electric quadrupole frequencies of the two gallium isotopes has been measured in two forms of metallic gallium. The experimental and theoretical values of the deduced quadrupole anomaly, as defined by Pyykko (1970), imply that the electric field gradient in alpha-gallium is negative whereas in beta-gallium it is positive. The quadrupole moment ratio of the gallium-69 and gallium-71 isotopes has been determined by a precision measurement of the quadrupole frequencies in GaCl₃ where the pseudoquadrupole interaction is presumed to be small.

The satellites to the main aluminium resonance have been used to obtain information for up to fourth neighbours of the impurity. The principal axis directions have been determined. An anisotropy has been observed in the Knight shift at the nearest neighbours in AlMn which cannot be ascribed to a second rank tensor. The nearest neighbour Knight shift in AlCr was found to be isotropic and essentially temperature independent.

The optical and dielectric functions alpha and epsilon ₂ of Ag, Pd and Ag rich Ag-Pd alloys are deduced from reflectivity measurements in the range 0.1-80 eV. Starting from these experimental results and from band structure calculations, an experimental model of the conduction band density of states for both pure metals and alloys is proposed. The variation with alloy composition of the conduction band density of states for Ag rich alloys seems very similar to that of the valence band.

The L₃M₄₅M₄₅ Auger spectrum and the 2p_3/2 core level spectrum of pure Ni are found to be considerably wider than those of pure Cu. When Ni is diluted into Zn, the main peaks in the L₃M₄₅M₄₅ Auger spectrum and the 2p_3/2 core level spectrum become as sharp as those of pure Cu. High resolution UV photoemission experiments show that the Ni d-like density of states narrows and retreats from the Fermi level upon dilution of Ni in Zn. These results suggest that final state interactions, which broaden the core level spectra, are also broadening the Auger spectra. It is also found that the L₃M₂₃M₄₅ Auger spectrum of Ni does not narrow when Ni is diluted into Zn, indicating that it does not represent the density of d-like states.