This review summarizes experimental evidence for the freezing of reorienting moments in solids. The moments may be of dipolar or quadrupolar nature, or both; they belong to one of the constituents of a mixed-crystal solid. Extensive results are reported for the following systems: KC1 doped with hydroxyl, potassium tantalate doped with Li, Na and Nb, alkali halide cyanides and alkali-alkali cyanides, rubidium ammonium dihydrogen phosphate, solid ortho-para hydrogen and argon-nitrogen mixtures. These have clearly glass-like properties. In other systems, results are limited to one or two methods hinting at glass formation; some of those are also reported. Clustering phenomena and the slow-down of reorientations at the freezing temperature are observed in susceptibility measurements and by local probing on nuclear spins. The modulation of the structure by cluster formation is revealed by diffraction experiments. These phenomena are confronted with model predictions and numerical simulations. l , 2.
IntroductionWe apply the term 'orientational glass' to a solid consisting of a regular lattice some of whose sites are occupied by constituents containing a dipole or quadrupole moment. These moments have orientational degrees o f freedom; they interact with one another and, below some freezing temperature Tf, their motion slows and they freeze into a configuration devoid of long-range order. A typical example is KC1 doped with OH molecules that replace C1 at random sites. At ! 0 K the dipoles associated with the OH molecules reorient rapidly, whereas at 0-1 K their orientations are static on experimental time scales, yet there is no macroscopic (spontaneous) polarization. Early investigations of this orientational-glass prototype (of the subspecies 'dipole glass') were performed by Kfinzig et al. (1964). Brout (1965)considered its short-range order and pointed out the analogy with what was to become the spin-glass prototype, Cu doped with Mn, where Mn has a spin of ~ and therefore contains magnetic degrees of freedom.The use of the word 'glass' suggests a similarity with what we call ~canonical glasses' (the prototype of which is fused silica, SiO2), in order to distinguish them from orientational and spin glasses. There are indeed orientational degrees of freedom in SiO2, in addition to rapid freezing of the moments and a low-temperature configuration devoid of order. Owing to the lack of an underlying lattice, however, it is impossible to associate moments at site i with interactions between moments at sites i and j, which is a pre-requisite for modelling a glass in terms of a Hamiltonian.Spin-glass models are based almost entirely on the existence of a Hamiltonian containing interactions between spins and their coupling to a magnetic field. The interactions chosen to model spin glasses are, in general, totally unspecific for spins. They are usually probabilistic distributions o f interaction strengths with a variance Var~j much larger than their average ~j and with both quantities independent of i j. The picture of spi...