From mean-field localized magnetism to itinerant spin fluctuations in the “nonmetallic metal” FeCrAs

FeCrAs displays an unusual electrical response that is neither metallic in character nor divergent at low temperatures, as expected for an insulating response, and therefore it has been termed a “nonmetal metal.” The anomalous resistivity occurs for temperatures below $\sim 900$ K. We have carried out neutron scattering experiments on powder and single crystal samples to study the magnetic dynamics and critical fluctuations in FeCrAs. Magnetic neutron diffraction measurements find ${\mathrm{Cr}}^{3+}$ magnetic order setting in at $T_N=115\mathrm{K}\sim 10$ meV with a mean-field critical exponent. Using neutron spectroscopy we observe gapless, high velocity, magnetic fluctuations emanating from magnetic positions with propagation wave vector ${\vec{q}}_0=(\frac{1}{3},\frac{1}{3})$, which persists up to at least 80 $\mathrm{meV}\sim 927$ K, an energy scale much larger than $T_N$. Despite the mean-field magnetic order at low temperatures, the magnetism in FeCrAs therefore displays a response which resembles that of itinerant magnets at high energy transfers. We suggest that the presence of stiff high-energy spin fluctuations extending up to a temperature scale of $\sim 900$ K is the origin of the unusual temperature dependence of the resistivity.

Figure 1

(a) Crystal structure of FeCrAs illustrating the ${\mathrm{CrAs}}_5$ pyramids and connectivity along the $c$ axis. (b) The structure projected onto the $ab$ plane.

Figure 2

(a) Integrated intensity of the (2/3,2/3,0) magnetic Bragg peak measured on 1T-1 (LLB); solid line is a fit to ${\left(1-T/T_N\right)}^{2\beta }$ with $T_N=115.5\pm 0.5$ K and $\beta =0.54\pm 0.05$. (b) Transverse scans through the magnetic Bragg peak at (2/3,2/3,0).

Figure 3

(a) Powder-averaged inelastic neutron spectrum in FeCrAs taken on MARI. The intensity between 75 meV and 100 meV is integrated and plotted as a momentum cut in panel (b). The blue dashed line is an estimate of the background from extrapolating from large momentum transfers as described in the text and the dark red curve is the scale magnetic ${\mathrm{Cr}}^{3+}$ form factor. (c) Illustrates the same data as in panel (a), but with the background removed. The solid line at $E=40$ meV shows where the background subtraction fails due to strong and highly structured in momentum phonon scattering. Constant energy cuts from this panel are plotted in (d) for the energy interval $E=[55,60]$ meV and (e) for the energy interval $E=[25,30]$ meV.

Figure 4

The powder averaged magnetic response at 5 K in FeCrAs measured with (a) $E_i=300$ meV (MARI, ISIS) and (b) $E_f=2.4$ meV (MACS, NIST). The variation in pixel size as a function of energy transfer in the MACS data, panel (b), is due to the difference in the way the data is collected. MARI data was collected in a time of flight configuration while MACS is a triple-axis which each energy transfer corresponding to a different constant energy scan. (c),(d) The powder averaged heuristic parametrization based on the single mode approximation (SMA) discussed in the text. The calculation was done assuming two-dimensional linear spin waves with a velocity of 200 meV Å.

Figure 5

The powder averaged magnetic response measured with $E_i=150$ meV at (a) $T=5$ K and (b) 150 K. (c),(d) Constant energy cuts illustrating the decay of magnetic intensity with momentum transfer. Panel (e) illustrates an energy cut at 25 meV where phonon scattering prevents a reliable subtraction of the background.