Another mechanism for the insulator-metal transition observed in Mott insulators

The two widely accepted mechanisms of the insulator-metal Mott–Hubbard transitions which have been considered up until now are driven by the band-filling or bandwidth effects. We found a different mechanism of the Mott–Hubbard insulator-metal transition, which is controlled instead by the changes in the Mott–Hubbard energy $U$. In contrast to the changes in the bandwidth $W$ in the “bandwidth control” scenario or to the variations of the band-filling $n$ parameter in the “band-filling” scenario, a dramatic decrease in the Mott–Hubbard energy $U$ plays the key role in this mechanism. We have experimentally observed this type of the insulator metal transition in the transition metal oxide $\mathrm{Bi}\mathrm{Fe}{\mathrm{O}}_3$. The decrease in the Mott–Hubbard energy is caused by the high-spin–low-spin crossover in the electronic $d$ shell of $3d$ transition metal ion ${\mathrm{Fe}}^{3+}$ with $d^5$ configuration under high pressure. The pressure-induced spin crossover in $\mathrm{Bi}\mathrm{Fe}{\mathrm{O}}_3$ was investigated and confirmed by synchrotron x-ray diffraction, nuclear forward scattering, and x-ray emission methods. The insulator-metal transition at the same pressures was found by the optical absorption and dc resistivity measurements.

Figure 1

(a) The unit cell structure and (b) local crystal structure of $\mathrm{Bi}\mathrm{Fe}{\mathrm{O}}_3$ at normal conditions could be considered as distorted perovskite with strongly distorted $\mathrm{Fe}{\mathrm{O}}_6$ octahedrons. (c) The evolution of x-ray patterns of $\mathrm{Bi}\mathrm{Fe}{\mathrm{O}}_3$ measured in helium pressure medium does not show any change of crystal symmetry up to $70\mathrm{GPa}$. (d) Solid lines $V_1$ and $V_2$ are the Bürch–Murnaghan approximations, indicating a structural transition between 45 and $55\mathrm{GPa}$. This transition is caused by the spin crossover of ${\mathrm{Fe}}^{3+}$ ion. We expect a first-order-type structural transition at $T\to 0\mathrm{K}$ due to the difference in volume of ${\mathrm{Fe}}^{3+}$ ions in HS $(S=5∕2)$ and LS $(S=1∕2)$ states. However, due to thermal (HS↔LS) fluctuations in the region of the spin crossover at $T=300\mathrm{K}$, the transition is broadened to about $10\mathrm{GPa}$ pressure range. Thus, we observe a broad anomaly in $V\text{−}P$ relation indicating the transition from the high-spin to the low-spin state.

Figure 2

[(a) and (b)] The complex scenario of electronic transition in $\mathrm{Bi}\mathrm{Fe}{\mathrm{O}}_3$ at high pressure and low temperature. The evolution of XES spectra at room temperature (a) demonstrates the decrease in the satellite peak intensity in the region of $45–55\mathrm{GPa}$ (b) indicating the HS-LS spin crossover (Ref. 24). The ${\mathrm{I}}_{\mathrm{sat}}$ is the integrated intensity of satellite peak in arbitrary units (located on the left side from the main peak), being the measure of the high-spin value. (c) We show different types of NFS spectra observed in high-pressure–low-temperature measurements. The spectra of type I correspond to magnetic phase of ${\mathrm{Fe}}^{3+}$ ions in HS state. The spectra of type II belong to an intermediate state in the region of HS-LS spin crossover close to the HS state, type III belongs to an intermediate state in the region of HS-LS spin crossover close to the LS state, type IV is pure LS state, type V is an intermediate state at the boundary of transition from the LS (HS) state into a metal, and type VI belongs to the pure metallic phase. (d) The complex $P\text{−}T$ diagram of electronic states which has the region with intermediate electronic states due to HS-LS fluctuations, the regions with pure HS (insulator) and LS (semiconductor) states, and finally, the region with LS metallic phase. At room temperature, the start and the end of the transition are bracketed by XES (gray squares) and resistance (gray circles) data. The width of the intermediate region (hatched) decreases at low temperature. The transition is narrow at $100\mathrm{K}$ [see also Fig. 4c]. Further experiments are required to constrain the transition width and position below $100\mathrm{K}$.

Figure 3

The evidence for the insulator-metal transition from optical absorption and resistance measurements. The evolution of optical absorption spectra at (a) compression and (b) decompression reveals a drop of the optical gap to nearly zero in the (c) pressure range from $45\text{to}55\mathrm{GPa}$. This decrease in the optical gap correlates well with the drastic drop of resistance of the sample (almost by 7 orders of magnitude) in the (d) same pressure region. The transition is continuous, and its width could be explained by the thermal HS-LS fluctuation model (Ref. 16).

Figure 4

(Color online) The insulator-metal transition invoked from the evolution of temperature dependence of the resistance under pressure. (a) The temperature dependence of $\log \left(R\right)$ is fitted well to the Mott dependence $\sim 1∕T^{1∕4}$. At pressures higher than $46\mathrm{GPa}$, the temperature dependence becomes more complex due to the mixing of the HS and LS states in the intermediate region of the spin crossover where thermal fluctuations between the HS and LS states play an important role. (b) At about $55\mathrm{GPa}$, the temperature dependence of resistance is changing to metallic behavior (the upturn in the resistance curves at low temperatures may be due to the nonuniform pressure distribution, when some parts of the sample remain insulating at the pressures close to IMT). The width of the insulator-metal transition at $100\mathrm{K}$ becomes substantially lower at $295\mathrm{K}$ (c) thus supporting the thermal HS-LS fluctuation model. Images of the sample at (d) $7.2\mathrm{GPa}$ and at (e) $55\mathrm{GPa}$ show the change in optical transmission of $\mathrm{Bi}\mathrm{Fe}{\mathrm{O}}_3$ at the IMT. The four Pt leads are connected to the $\mathrm{Bi}\mathrm{Fe}{\mathrm{O}}_3$ sample in the central hole of the $c\mathrm{B}\mathrm{N}$ gasket. (d) Reddish crystal is in the center of the image; a transparent part of NaCl pressure medium is visible on the top of the sample. (e) In the image, $\mathrm{Bi}\mathrm{Fe}{\mathrm{O}}_3$ sample becomes opaque because of the transition into the metallic state. At these pressures, the $c\mathrm{B}\mathrm{N}$ gasket becomes nearly transparent.