One of the hottest questions in condensed matter physics is: What causes superconductivity with high exceptionally high T_c (the temperature above which the material becomes a normal conductor)? A precise answer to this question was obtained by collaborative efforts of the University of Geneva with Stanford and Tsukuba [1] and with Caltech [2], using frequency- and time-resolved optical techniques.

Super-currents are carried by bound pairs of electrons. Binding of two electrons can occur if an electron polarises its surrounding medium, and a second electron is trapped inside this polarisation cloud. This induced polarisation, often indicated as “the glue” depends in general on the motional energy of an electron. A hotly debated issue is, whether aforementioned polarisation is an excitation of the crystal lattice (in which case pairing is mediated by phonons) or a polarisation of the magnetic moments of the electrons (which corresponds to spin-fluctuation mediated pairing).

Van Heumen et al. [1] and Carbone et al. [2] succeeded in determining the nature of the aforementioned glue using optical experiments. They attacked the problem from opposite directions: Van Heumen’s experiment probes the electrons, and measures how they are influenced by their coupling to the glue. Carbone’s experiment probes the excitations of the crystal lattice, and measures how this excited state decays as a function of time due to the coupling to the electrons. The experiments provide complementary information: Van Heumen’s method allows reconstructing the complete glue-spectrum, but it is insensitive to the nature of the glue. Carbone measured the relaxation of lattice vibrations in real time using time-domain electron diffraction, but his experiment is “blind” for the spin-fluctuations. Van Heumen observed a peak in the glue spectrum for all temperatures from 10 Kelvin up to room temperature, with a characteristic frequency of 55 meV. Carbone observed that lattice vibrations with precisely this frequency have a fast relaxation due to electron-phonon coupling. Both experiments indicate moderate electron-phonon coupling, i.e. λ < 1.

In addition van Heumen observed a second component in the glue spectrum up to 5 times the maximum frequency of the lattice, which is 100 meV for the cuprates. Experimental tools for fast time-domain imaging of spin-fluctuations are presently non-existing, but since this high frequency part of the glue-spectrum can not be due to lattice vibrations, it is overwhelmingly natural to attribute it to spin-fluctuations. The main novelty is the demonstration of internal consistency and the possibility to induce experimentally verifiable quantities once the glue-spectrum has been deduced: Firstly, if for a given

sample the glue-spectrum is determined from the optical data at, say, room temperature, the optical spectra at all other temperatures can be predicted exactly. Secondly, T_c can be predicted from the glue-spectrum. The predicted T_c’s for 10 crystals from strongly underdoped to highly overdoped are 'only' two times higher than the optimal T_c's. This is quite close, especially considering that, among other things, all of the measured ‘glue’ was taken to mediate d-wave pairing, which is certainly too optimistic

The part of the glue-spectrum below 100 meV represents all of the lattice vibrations and part of the spin-fluctuations. For overdoped samples this part of the glue-spectrum turns out to be much too weak to account for their high T_c of about 70 Kelvin. Phonon-mediated pairing can therefore be eliminated as “the mechanism of high T_c” in favour of the other possibility, namely that pairing in the high T_c cuprates is mediated by spin-fluctuations.