No abstract

We performed inelastic neutron scattering on the 2D Shastry-Sutherland system SrCu 2 ( 11 BO 3 ) 2 with an exact dimer ground state. Three energy levels at around 3, 5 and 9 meV were observed at 1.7 K. The lowest excitation at 3.0 meV is almost dispersionless with a bandwidth of 0.2 meV at most, showing a significant constraint on a single-triplet hopping owing to the orthogonality of the neighboring dimers. In contrast, the correlated two-triplets excitations at 5 meV exhibit a more dispersive behavior.-1-

The Dzyaloshinski-Moriya interaction partially lifts the magnetic frustration of the spin-1/2 oxide SrCu2(BO3)2. It explains the fine structure of the excited triplet state and its unusual magnetic field dependence, as observed in previous ESR and new neutron inelastic scattering experiments. We claim that it is mainly responsible for the dispersion. We propose also a new mechanism for the observed ESR transitions forbidden by standard selection rules, that relies on an instantaneous Dzyaloshinski-Moriya interaction induced by spin-phonon couplings. PACS numbers:Strontium Copper Borate (SrCu 2 (BO 3 ) 2 ) is a new example of a magnetic oxide with a spin gap [1], with a ground state well described as simply a product of magnetic dimers in two dimensions on the bonds giving the strongest magnetic exchange [2]. The weaker exchanges are frustrated by the geometry and, as shown by Shastry and Sutherland [3], the ground state of the isotropic Hamiltonian is independent of the value of the weaker exchange, up to a critical value. The excitations, however, are not purely local and cannot be explicitly given. Recent experiments by ESR [4] and neutron inelastic scattering presented here show how in fact there are spin anisotropies needed for an accurate description of the dynamics. We shall show the corrections to the ground state are needed that, while small, will be necessary to understand many physical properties. For example at finite external magnetic field SrCu 2 (BO 3 ) 2 appears to exhibit a number of finite magnetization plateaux [1,5], and the anisotropies will determine the observability of plateaux in different field directions. Furthermore SrCu 2 (BO 3 ) 2 is believed to be close in parameter space to a quantum critical point whose nature is somewhat controversial, and while the anisotropies are small they may be essential to its nature.For spin 1 2 the leading anisotropic terms are of form Dzyaloshinski-Moriya [6] and exchange anisotropy. The former is particularly relevant, since it may not be frustrated even if the isotropic exchange is. While a small Dzyaloshinski-Moriya interaction should not destroy a gap generated by larger isotropic interactions it modifies the pure locality of the ground state correlations and delocalizes the first triplet excitation. This is because it appears in lower order in perturbation theory than the frustrated isotropic interactions. In this paper we predict the Dzyaloshinski-Moriya interactions that should be expected in SrCu 2 (BO 3 ) 2 from the structure, and show that they do indeed explain new features of the excitations observed with ESR and neutron inelastic scattering experiments. Miyahara and Ueda [2] have introduced the frustrated Shastry-Sutherland modelfor SrCu 2 (BO 3 ) 2 , with S = 1/2 and where nn stands for nearest neighbor spins and nnn for next nearest neighbors. The lattice is shown in fig. 1. J = 85K and J ′ = 54K are antiferromagnetic interactions estimated from the susceptibility and the gap [2]. The spectrum of spin excitations has several interesting f...

No abstract

No abstract