Quasiparticle coherence in the nematic state of FeSe

Electronic nematicity is a ubiquitous phenomenon in iron-based superconductors but its origin is still debated. Most models consider either spin or orbital degrees of freedom as the driving force but typically do not take electronic correlations into account. However, mass enhancements, coherent-incoherent crossovers, and the strong orbital differentiation can only be understood using correlations in a Hund's metal framework. Here, we study the influence of nematicity on the quasiparticle coherence in detwinned FeSe using angle-resolved photoemission spectroscopy (ARPES). We compare photoemission spectral weight from $d_{xz}$ and $d_{yz}$ orbitals in the coherent quasiparticle peak and in the incoherent Hubbard band and find an anisotropy between the two orbitals. We interpret our observation in terms of a more coherent $d_{xz}$ orbital compared to the $d_{yz}$ orbital inside the nematic phase. This result is in contrast to earlier predictions of an incoherent $d_{xz}$ orbital and highlights the importance of electronic correlations in the description of nematicity.

Figure 1

Experimental geometry of the ARPES measurement. Left: Horizontally (LH) or vertically (LV) polarized light photoemits electrons from the sample. The ARPES analyzer slit selects only those electrons emitted along the detection plane. Azimuthal rotation $\omega$ of the sample holder then allows the separate measurement of spectra along $k_x$ and $k_y$. Right: Photograph of detwinning device.

Figure 2

ARPES spectra of FeSe at 12 K. (a1) Spectrum taken along $k_x$. (a2) Spectrum taken along $k_y$. (a3) EDCs obtained from (a1) and (a2) integrated between $k=(0,-1.2){\mathring{\mathrm{A}}}^{-1}$ as indicated by the bars on the top of the spectra. (b) same as (a) but zoomed into a smaller energy window close to $E_{\mathrm{F}}$. Sketches in (b1) and (b2) indicate the measurement directions in the BZ and the orientation of the LV light polarization. The dominantly probed Fe $3d$ orbital character ($xz,yz$) is indicated in each panel. Bars on top of the spectra show the momentum integration window for the EDCs discussed in (a3) and (b3) (EDC 1) and in Figs. 3 (EDC 2) and 4 (EDC 3).

Figure 3

Comparison of EDCs along $k_x$ and $k_y$ taken close to $\mathrm{\Gamma }$. The momentum integration window for the EDCs is indicated in Fig. 2 with the label EDC 2. (a) EDCs shown in the whole measured energy window. (b) Difference of intensity $I$ between the two EDCs shown in (a). We limit the displayed range on the vertical axis in order to highlight anisotropies around the Hubbard band located around $-1$ eV. The signature of the quasiparticle peak is therefore cut off in this plot. (c) Zoom of (a) into smaller energy scales close to the Fermi level to highlight anisotropies in the quasiparticle spectral weight of the hole bands.

Figure 4

Comparison of EDCs along $k_x$ and $k_y$ taken close to $M_X$ and $M_Y$. The momentum integration window for the EDCs is indicated in Fig. 2 with the label EDC 3. (a) EDCs shown in the whole measured energy window. (b) Difference of intensity $I$ between the two EDCs shown in (a). We limit the displayed range on the vertical axis in order to highlight anisotropies around the Hubbard band located around $-2$ eV. The signature of the quasiparticle peak is therefore cut off in this plot. (c) Zoom of (a) into smaller energy scales close to the Fermi level to highlight anisotropies in the quasiparticle spectral weight of the electron bands.