ESPCI web site
Jeudi 29 Mars 2018 à 15h15, Anphi Urbain, Bâtiment N RdC
Thermodynamics and thermal transport in spin ice
The spin-ice materials R2Ti2O7 with R = Ho, Dy crystallize in a pyrochlore lattice with the magnetic
R3+ ions forming a network of corner-sharing tetrahedra. The crystal electric field induces a strong
Ising anisotropy, which forces the magnetic moments to point either into or out of each tetrahedron.
This results in a geometric frustration with a groundstate degeneracy described by the so-called
"2in/2out" ice-rule, meaning that 2 spins point into and 2 out of each tetrahedron, which is equivalent
to the hydrogen displacement in water ice. Excited states are created by single spin flips resulting in
"3in/1out" and "3out/1in" pairs, which, in zero magnetic field, fractionalize and are described as
independently propagating magnetic (anti-)monopoles. In this talk, I will discuss the thermodynamic
signatures of a possible residual groundstate entropy and the influence of the monopole excitations on
the total heat transport. From the anisotropic magnetic-field dependence and by additional
measurements on reference compounds, we are able to separate the phononic and the magnetic
contributions to the total heat transport. The magnetic contribution scales with the degree of
groundstate degeneracy indicating that it results from the monopole mobility. At lowest temperatures,
however, ultra-slow relaxation effects make an unambiguous determination of the groundstate and its
degeneracy in spin ice shows very difficult. These relaxation effects rapidly vanish upon the partial
substitution of the magnetic Dy3+ ions by non-magnetic Y3+ in the dilution series (Dy1-xYx)2Ti2O7. With
increasing x, our specific heat data suggest a rapid crossover from spin-ice physics in pure Dy2Ti2O7
towards weakly interacting single-ion physics in (Dy1-xYx)2Ti2O7 at intermediate x. This experimental
result contradicts the theoretically predicted zero-temperature residual entropy for dilute spin ice,
which has been obtained from a generalization of Pauling’s theory but is supported by recent Monte
Carlo simulations.
