Unusual Relationship between Magnetism and Superconductivity in ${\mathrm{FeTe}}_{0.5}{\mathrm{Se}}_{0.5}$

We use neutron scattering to study magnetic excitations in crystals near the ideal superconducting composition of ${\mathrm{FeTe}}_{0.5}{\mathrm{Se}}_{0.5}$. Two types of excitations are found, a resonance at (0.5,0.5,0) and incommensurate fluctuations on either side of this position. We show that the two sets of magnetic excitations behave differently with doping, with the resonance being fixed in position while the incommensurate excitations move as the doping is changed. These unusual results show that a common behavior of the low energy magnetic excitations is not necessary for pairing in these materials.

Figure 1

(a) Measurement of the heat capacity of a piece of the ${\mathrm{FeTe}}_{0.5}{\mathrm{Se}}_{0.5}$ crystal B near and below $T_c=14\mathrm{K}$, clearly showing bulk superconductivity. The inset shows the magnetic susceptibility measured with $H=20\mathrm{Oe}$ using zfc and fc protocols. The diamagnetic susceptibility for the zfc data corresponds to complete diamagnetic screening. (b) Diagram of the scattering plane in reciprocal space showing the position of the incommensurate excitations (crosses), the (0.5,0.5,0) position as the circle, and the scan directions as dotted lines. The incommensurate excitations rise quickly in energy. (c) Constant $Q=(0.5,0.5,0)$ scans taken at 1.6 and 20 K. The low temperature curve rises above the high temperature curve the most at about 6.5 meV giving the resonance energy. (d) Temperature dependence of the scattering at the resonance position showing the rapid increase at $T_c$. The red solid line is a power law fit to the data.

Figure 2

(a) Scan along ($H$, $1\mathrm{\text{−}}H$, 0) at 13 meV for sample A. The scan shows the incommensurate peaks when the measurement was made above the resonance. The line is a Gaussian least squares fit. (b) Scan along ($H$, $1\mathrm{\text{−}}H$, 0) at 13 meV for sample B. The incommensurate peaks are much closer to (0.5,0.5,0) and overlap considerably. A least squares fit can still be made to two peaks rather than one with a much higher quality of fit parameter. (c) Scan at 6 meV through the resonance region in the ($H$, $1\mathrm{\text{−}}H$, 0) direction at 1.6 and 20 K for sample A. The resonance grows out of the 20 K peak as the temperature is lowered. However, it only stems from the middle third of the distribution suggesting that it does not come from the incommensurate excitations. (d) Scan through the resonance region along ($H$, $1\mathrm{\text{−}}H$, 0) for sample B. Similar to sample A, the resonance comes out of the middle of the 20 K distribution. The resonance is also larger compared to the 20 K distribution perhaps reflecting the fact that $T_c$ is slightly higher.

Figure 3

(a) Contour plot for sample B made from scans in the ($H$, $1\mathrm{\text{−}}H$, 0) direction at 1.6 K. It is wide at all energies as the incommensurate peaks are covered as well. The high intensity shown in red comes from the resonance. The map shows that the distribution spreads out slightly as the energy is increased. The lack of intensity at low energies stems from the superconducting gap. (b) Contour plot for ($H$, $H$, 0) scans. From 11 meV to higher energies, the left-hand part of the data could not be obtained due to scattering geometry constraints. The data were thus used from the right side giving a symmetrical pattern. The scattering is narrow at all energies as the incommensurate contribution is largely avoided. There appears to be extra intensity at the top of the plots around 13 meV. We have no explanation for this, but it could stem from phonon scattering.