New clue toward paramagnons as the glue

Certainly lots of doubts are over the magnetic fluctuations as the paring source of carriers in cuprate superconductors. A central issues concerns if there is sufficient paramagnons in the Sc region, since experiments have so far detected only a limited volume of such stuff. Now this gets changed due to this work [Nature Phys. 7, 725–730 (2011). ] reviewed below [http://www.nature.com/nphys/journal/v7/n9/full/nphys2077.html?WT.ec_id=NPHYS-201109]:

However, the observed excitations were restricted to a narrow window in both energy and momentum and furthermore carried relatively little spectral weight, posing a challenge to theoretical ideas about magnetic fluctuations being the source of Cooper pairing in these superconductors. Some researchers have suggested that the experimental limitations inherent in neutron scattering were partially responsible for this state of affairs — and only now has a breakthrough occurred.

In Nature Physics, Le Tacon and colleagues² report the application to various copper oxides of an alternative technique to map magnetic excitations: resonant inelastic X-ray scattering (RIXS)³. Here, an electron is transferred, by a high-energy photon, from a deep core level into an unoccupied low-energy state; subsequently, an electron from a different low-energy state fills the core hole and emits a high-energy photon. Thus, a net excitation is generated in a low-energy band, the energy and momentum of which can be measured by examining the scattered photon.

Among the advantages of RIXS, compared with neutron scattering, is the large cross-section for the scattering of photons (which eliminates the need for large samples) and the possibility to probe essentially the entire Brillouin zone. There are disadvantages as well: in contrast to neutron scattering, the cross-section is not simply related to a dynamic susceptibility, which complicates the data analysis, and the energy resolution is at present limited to about 100 meV (it's far below 1 meV in state-of-the-art neutron-scattering experiments). Despite these limitations, the past decade has seen exciting progress in RIXS³ such that investigations of elementary spin excitations have now become feasible.

Le Tacon et al.² have investigated magnetic excitations using RIXS in a family of copper-oxide materials, covering a range of hole dopings from the undoped insulator to the slightly overdoped superconductor. In all doped materials, they identified damped spin excitations with high intensity over a large part of momentum space. These excitations, in both their overall dispersion and their intensity, seem to show surprisingly little variation with doping.

These findings are important for a number of reasons. First, together with similar recent experiments^{3, 4, 5}, they establish RIXS as a powerful tool for the investigation of complex correlated-electron materials. Second, they show that previous neutron-scattering studies have indeed missed a significant part of the spectral weight of spin fluctuations in copper oxides. This implies that theories of electron pairing based on the exchange of magnetic fluctuations can be considered on safer ground. In fact, Le Tacon et al. provide a sample calculation of a superconducting critical temperature (T_c), in which they use the measured spin-fluctuation spectrum and electronic bands as input and obtain a T_c value comparable to the experimental one.

Third, and perhaps most importantly, their data indicate that key features of the spin fluctuations in doped copper oxides are strikingly similar to that of their undoped counterparts (Fig. 1): at the elevated energies probed by RIXS, the only significant effect of doping is an energy broadening of the excitations, probably arising from damping due to electron-hole excitations. (One should note that the present energy resolution of RIXS is insufficient to resolve fine structures on scales below 100 meV; therefore the similarity of doped and undoped spectra refers to gross features, and the details may well differ.)