Magnetic and superconducting phases of heavy fermion compounds

Superconductivity in metals arises from the tendency of conduction electrons to move as bound pairs and not as essentially independent particles. In the conventional model, the ''glue'' which holds the pairs together is the fluctuating ions in the lattice or, more precisely, the phonons. These act as messenger quanta in some degree analogous to the photons in quantum electrodynamics that bind electrons and protons to form atoms, or to the gluons in quantum chromodynamics that hold quarks together to produce protons and other nuclear particles. The aim of this dissertation is to investigate the possible existence of a very different type of glue connected with quanta of a field associated with the magnetisation density. This type of glue is expected to be most important in pure systems on the border of magnetic long-range order, i.e., in homogeneous conducting materials in which the magnetic transition temperature tends to be vanishingly small. In an effort to search for such magnetic pairing of electrons we have identified suitable stoichiometric compounds that have low transition temperatures at ambient pressure, and then we have depressed the transition towards absolute zero by fine tuning via the application of hydrostatic pressure. The thesis will describe (i) the selection of appropriate stoichiometric compounds based on the elements cerium and uranium, (ii) the growth of pure specimens via induction melting techniques in ultra high vacuum or in ultra-pure argon gas, (iii) resistivity and magnetic susceptibility measurements at high pressures and very low temperatures, and (iv) experimental results in the nearly antiferromagnetic metal CeNi2Ge2 and in the ferromagnetic compounds CeIr2B2 and UGe2. Our findings will be discussed in terms of current models of unconventional superconductivity and of novel 'normal' states expected to arise on the border of magnetic long-range order. (author)