Surprising loss of three-dimensionality in low-energy spin correlations on approaching superconductivity in ${\mathrm{Fe}}_{1+y}{\mathrm{Te}}_{1-x}{\mathrm{Se}}_x$

We report inelastic neutron scattering measurements of low-energy ($\hslash \omega \lesssim 10$ meV) magnetic excitations in the “11” system ${\mathrm{Fe}}_{1+y}{\mathrm{Te}}_{1-x}{\mathrm{Se}}_x$. The spin correlations are two-dimensional (2D) in the superconducting samples at low temperature, but appear much more three-dimensional (3D) when the temperature rises well above $T_c\sim 15$ K, with a clear increase of the (dynamic) spin correlation length perpendicular to the Fe planes. This behavior is extremely unusual; typically, the suppression of thermal fluctuations at low temperature would favor the enhancement of 3D correlations, or even ordering, and the reversion to 2D cannot be naturally explained when only the spin degree of freedom is considered. Our results suggest that the low temperature physics in the 11 system, in particular the evolution of low-energy spin excitations towards superconducting pairing, intrinsically involves changes in orbital correlations.

Figure 1

Inelastic neutron scattering measurements in the $\left(HK0\right)$ plane measured on SEQUOIA at energy transfers $\hslash \omega =7$ meV from (a)–(c) SC40 and (d) Ni10 samples. The samples temperatures are (a) 5 K, (b) 20 K, (c) 200 K, and (d) 5 K. All slices were taken with an energy width of 2 meV. Intensity scale is the same in (a) to (c); data in (d) have been scaled to be directly comparable. Black regions of each panel represent gaps in the detector array.

Figure 2

Inelastic neutron scattering intensity in the $\left(HHL\right)$ plane measured on HYSPEC at energy transfer $\hslash \omega =6.5$ meV from the SC40 sample. Left column are 2D intensity slices, and right column are linear cuts along $(0.5,0.5,L)$, obtained at (a) and (b): 5 K; (c) and (d) 20 K; (e) and (f) 100 K; and (g) and (h) 300 K. The q-width of the linear cuts is 0.05 r.l.u. along [110] direction. The white line along $(0.5,0.5,L)$ in the left panels shows where the $L$ cuts in the right column were taken. The dashed lines in right panels are the estimated background obtained from fitting around $(0.65,0.65,0)$. The blue solid lines in (b) and (d) are magnetic form factors for a ${\mathrm{Fe}}^{2+}$ ion scaled to the data average. All slices were taken with an energy width of 2 meV. The error bars represent statistical error.

Figure 3

Inelastic neutron scattering intensities in the $\left(H0L\right)$ plane measured on HYSPEC at energy transfer $\hslash \omega =4$ meV on the SC40 and Ni10 samples. The intensities are scaled by the sample mass for better comparison. Left column are 2D intensity slices, and right column are linear intensity cuts along $(0.5,0,L)$. The q width of the linear cuts is 0.05 r.l.u. along [100] direction. The panels are (a) and (b): SC40 at 5 K; (c) and (d) SC40 at 300 K; (e) and (f) Ni10 at 5 K; and (g) and (h) Ni10 at 300 K. The white line in the left panels at $H=0.5$ shows where the $L$ cuts in the right column were taken. The dashed lines in right panels are estimated background values obtained from fitting around $(0.65,0,0)$. The blue solid lines in (d), (f), and (h) are magnetic form factors for a ${\mathrm{Fe}}^{2+}$ ion scaled to the data range, and the red solid lines are fits to the data using two symmetric Lorentzians peaked at $L=\pm 0.5$. All slices were taken with an energy width of 2 meV. The error bars represent statistical error.

Figure 4

Inelastic neutron scattering measurements on (a) Ni02 and (b) NSC70 samples along $(0.5,0,L)$ at energy transfer $\hslash \omega =4$ meV, performed on BT7. The solid lines are fits to the two symmetric Lorentzian peaked at $L=\pm 0.5$. The intensities at $(0.5,0,0.5)$ vs. $T$ for both samples are plotted in (c). The dashed lines are guides to the eye. The error bars represent statistical error.