Starting from a phenomenological standard energy functional to describe the condensation of a dilute gas of bosonic quasiparticles in magnetic insulators, we find that the inclusion of a perturbation term that explicitly violates the axial symmetry significantly modifies the details of the resulting Bose Einstein condensation. Systems with an originally axial symmetry must show an intrinsic tendency to spontaneously violate this symmetry as soon as the condensation sets in, and the phase transition at the respective critical field may even become of first order. We can explain a number of features in the experimental data of various insulating spin systems, such as a slightly nonlinear magnetization near the critical field as well as hysteresis effects and peculiarities in the energy-level scheme of TlCuCl3. We also offer a consistent explanation for certain anomalies in the magnetocaloric effect and in the magnetization of BaCuSi2O6 by assuming a spontaneous violation of axial symmetry at the magnetic phase transition on an energy scale of about 1 micro-eV. The resulting anisotropy gap in the magnetic excitation spectrum, that inevitably forms at the critical field of all such systems, lifts the linear Goldstone mode and is therefore seriously limiting the lifetime of magnetic condensates to a few nanoseconds at most.
We describe the Bose-Einstein condensate of magnetic bosonic quasiparticles in insulating spin systems using a phenomenological standard functional method for
T=0. We show that results that are already known from advanced computational techniques immediately follow. The inclusion of a perturbative anisotropy term that violates the axial symmetry allows us to remarkably well explain a number of experimental features of the dimerized spin-1/2 system
TlCuCl3. Based on an energetic argument we predict a general intrinsic instability of an axially symmetric magnetic condensate toward a violation of this symmetry, which leads to the spontaneous formation of an anisotropy gap in the energy spectrum above the critical field. We, therefore, expect that a true Goldstone mode in insulating spin systems, i.e., a strictly linear energy-dispersion relation down to arbitrarily small excitations energies, cannot be observed in any real material.