Electronic Identification of the Parental Phases and Mesoscopic Phase Separation of ${\mathrm{K}}_x{\mathrm{Fe}}_{2-y}{\mathrm{Se}}_2$ Superconductors

The nature of the parent compound of a high-temperature superconductor (HTS) often plays a pivotal role in determining its superconductivity. The parent compounds of the cuprate HTSs are antiferromagnetically ordered Mott insulators, while those of the iron-pnictide HTSs are metals with spin-density-wave order. Here we report the electronic identification of two insulating parental phases and one semiconducting parental phase of the newly discovered family of ${\mathrm{K}}_x{\mathrm{Fe}}_{2-y}{\mathrm{Se}}_2$ superconductors. The two insulating phases exhibit Mott-insulator-like signatures, and one of the insulating phases is even present in the superconducting and semiconducting ${\mathrm{K}}_x{\mathrm{Fe}}_{2-y}{\mathrm{Se}}_2$ compounds. However, it is mesoscopically phase-separated from the superconducting or semiconducting phase. Moreover, we find that both the superconducting and semiconducting phases are free of the magnetic and vacancy orders present in the insulating phases, and that the electronic structure of the superconducting phase could be developed by doping the semiconducting phase with electrons. The rich electronic properties discovered in these parental phases of the ${\mathrm{K}}_x{\mathrm{Fe}}_{2-y}{\mathrm{Se}}_2$ superconductors provide the foundation for studying the anomalous behavior in this new class of iron-based superconductors.

Popular Summary

For living systems, many essential biological characteristics of offsprings can be traced back to their parents’ genetic makeup. It turns out that, for a high-temperature superconductor, the questions of what compound is its parent and what the parent’s nature is are also meaningful and important ones to ask for understanding the offspring’s superconductivity. The parent compounds of cuprate high-temperature superconductors are antiferromagnetically ordered (Mott) insulators, while those of a recently discovered family of iron-based high-temperature superconductors, called pnictides, are metals with a particular kind of magnetic order, the so-called spin-density-wave order. In 2010, another new family of iron-based superconductors was discovered among the $A_x\mathrm{F}{e}_{2-y}\mathrm{S}{\mathrm{e}}_2$ compounds ($A$ being either an alkali-metal element or thallium), and they show essential features, for example, in their electronic structures, different from the pnictides. What is their parent compound, what is the parental physical nature, and what is the parent-offspring relationship? In this paper, we answer these questions by investigating a series of four chemically different ${\mathrm{K}}_x\mathrm{F}{\mathrm{e}}_{2-y}\mathrm{S}{\mathrm{e}}_2$ compounds using an array of experimental techniques.

Kicking out electrons from each compound with photons, measuring how the energy of an emitted electron depends on the angle of its emission (angle-resolved-photoemission spectroscopy), and combining these measurements with the resistivity data and sample images obtained with transmission-electron microscopy, we are able to identify among the four compounds a superconductor, a semiconductor, and two Mott-like insulators. More interesting, we find that one of the insulator phases is even present in both the superconductor sample and the semiconductor sample, but it exists in domains of several nanometers separated from the superconducting-phase or the semiconducting-phase matrix—a mesoscopic phenomenon not seen before.

Equally remarkable is another finding that, not only are both the superconductor and the semiconductor free of any magnetic order or iron-vacancy order, but the semiconductor can be turned into the superconductor upon electron doping. This leads us to suggest that the semiconductor is a closer parent compound of the superconductor than the Mott-like insulators. This is the first time that such a unique parent compound is observed for a high-temperature superconductor. We believe that it indicates the possibility of a novel route to high-temperature superconductivity.

Figure 1

Resistivities and valence bands of various ${\mathrm{K}}_x{\mathrm{Fe}}_{2-y}{\mathrm{Se}}_2$ compounds. (a) In-plane resistivities as a function of temperature for the AFI1 and AFI2 insulating compounds, the semiconducting compound, and the superconducting compound with the superconducting-transition temperature ($T_{\mathrm{c}}$) of 31 K. The resistivity of the superconducting sample is multiplied by 100. (b) Valence-band photoemission spectra at the $\Gamma$ point for the insulators, semiconductor, and superconductor. The dashed vertical lines are guides to eyes for the peak positions of the feature near $-1\mathrm{eV}$. If not specified otherwise, data here and hereafter were taken with 21.2 eV photons from an in-house helium-discharge lamp. The data for superconductor, semiconductor, and insulators were measured at 35, 160, and 180 K, respectively.

Figure 2

Photoemission charging effects in ${\mathrm{K}}_x{\mathrm{Fe}}_{2-y}{\mathrm{Se}}_2$. (a) Temperature dependence of the photoemission-energy-distribution curve (EDC) at the $\Gamma$ point of the superconductor. The inset is the symmetrized EDCs across $T_c$, illustrating a superconducting gap of about 10 meV at the electron-Fermi surface around the zone corner. (b) EDC at the $\Gamma$ point of the superconductor as a function of the photon intensity. Data were taken at 17 K. (c) The EDCs after deducting the background [diagonal short-dashed line in (b)] to highlight the peak position shift due to the charging. (d) Temperature dependence of the EDCs at the $\Gamma$ point for the AFI1 sample. (e) The low-energy portion of data in (d). (f) and (g) are the same as (d) and (e), respectively, but for the semiconducting sample. The arrows on the dashed curves in all panels indicate the band shift with enhanced charging effects. The shaded bars in (b), (c), and (g) indicate charging-free features.

Figure 3

Spectral-weight difference between the insulators and the superconductor. (a) and (b) compare EDCs of the AFI1 insulator and the superconductor along cuts #1 and #2 across the projected two-dimensional Brillouin-zone center and corner, respectively. (c) and (d) make the same comparison between AFI2 and the superconductor. Data for the insulators and superconductor were taken at 180 and 11 K, respectively.

Figure 4

Low-energy electronic structures of the superconducting and semiconducting ${\mathrm{K}}_x{\mathrm{Fe}}_{2-y}{\mathrm{Se}}_2$. (a) Left: the Fermi surface of the superconducting phase in the 3D Brillouin zone. Right: the two-momentum cut #1 and cut #2 sampled with 21 eV and 31 eV photons, respectively, are illustrated in the two-dimensional cross-sections of the 3D Brillouin zone. $c^{\prime }$ is the distance between neighboring FeAs layers, and $a$ is the Fe-Fe distance in the FeAs plane. (b) A sketch of the band-structure evolution from the semiconductor to the superconductor. (c) The photoemission intensities (upper panel) and their second derivatives with respect to energy (lower panel) along cut #1 across $\Gamma$ for the superconductor taken at 35 K. (d) is the same as (c) but for the semiconductor taken at 100 K. (e) and (f) are same as (c), (d), respectively, except the data were taken along cut #2 across Z.

Figure 5

(a) Cartoon for mesoscopic phase separation in superconducting ${\mathrm{K}}_x{\mathrm{Fe}}_{2-y}{\mathrm{Se}}_2$. Different regions exhibit different photoemission-spectroscopic signatures. Inset in the left panel of (a): The diffraction pattern of the $\sqrt{5}\times \sqrt{5}$ order was observed with TEM in both the superconductors and semiconductors, indicating a mixing of either the superconducting or the semiconducting phase with the AFI1 phase. The arrow in the diffraction pattern indicates the superlattice-modulation wave vector $q_1=(1/5,3/5,0)$. The TEM data was collected at room temperature. Inset in the right panel of (a): The symmetrized EDCs of the superconductor across $T_{\mathrm{c}}$, illustrating a superconducting gap. (b) and (c) The typical raw HRTEM image of the semiconducting sample and superconducting sample, respectively. (d) and (e) The effective-dark-field images constructed from the images in (b) and (c) by using the ($\pm 1/5$, $\pm 3/5$, 0) superstructure spots as described in the text. In such dark-field images, only the $\sqrt{5}\times \sqrt{5}$ vacancy order contributes to the image contrast, and the mesoscopic phase separation of vacancy-ordered (brighter) and vacancy-disordered (darker) regions is clearly illustrated.