Se p 20 01 Slow relaxation , confinement , and solitons

Millisecond crystal relaxation has been used to explain anomalous decay in doped alkali halides. We attribute this slowness to Fermi-Pasta-Ulam solitons. Our model exhibits confinement of mechanical energy released by excitation. Extending the model to long times is justified by its relation to solitons, excitations previously proposed to occur in alkali halides. Soliton damping and observation are also discussed. 1 Crystals respond in picoseconds. This is the characteristic time for all sorts of lattice adjustments, for example passage to a " relaxed excited state. " Nevertheless, an anomaly in the decay of luminescence in doped alkali halides has led us to conclude [1,2] that relaxation can sometimes take place on a scale of milliseconds. It's as if the tides moved like tectonic plates. In this article we propose that the formation and slow decay of Fermi-Pasta-Ulam (FPU) solitons [3–5] can account for this dramatic slowdown. The presence of FPU solitons in alkali halides has been previously proposed [6,7] and we believe that particular features of our system [1,2] enable their production and observation. We begin with an overview. KBr (say) is dilutely doped with Pb 2+. One can think of the Pb and its 6 surrounding Br's as a quasimolecule. This quasimolecule is excited by a flash of UV light and because of the Jahn-Teller effect distorts considerably. In the absence of the constraining lattice one would expect expansion along one of the axes of perhaps 15%. The distortion is asymmetric (there is shrinking along other axes) and the Jahn-Teller interaction causes well-understood degeneracy-breaking effects. It is the subsequent return to the ground state that provides evidence for slow lattice relaxation. The excited quasimolecule can go to a radiative level with a 25 ns lifetime or to a metastable level with an 8 ms lifetime. The luminescence decay however is not the sum of two exponentials, but rather, after the initial burst from the fast level, molecules in the metastable level exhibit an enhanced decay rate, which we have explained to be due to lattice-induced coupling to the fast level. This continues for milliseconds and implies that the lattice itself takes ms to accommodate to the distortion. This is the slow relaxation, and our ability to use this assumption to fit a variety of systems in a large range of temperatures is the evidence for this phenomenon. Because the distortion of the quasimolecule is asymmetric, a reasonable model is a …