Figure 1

SQUID magnetization measurements on a bulk ${\mathrm{YbAl}}_3$ sample. The temperature dependence exhibits two maxima: one at $126\mathrm{K}$ which corresponds to the energy scale of single impurity screening, and one at $16\mathrm{K}$ which may be related to the onset of coherence.Reuse & Permissions

Figure 2

(a) Overview topography obtained on the surface of the ${\mathrm{YbAl}}_3$ sample after preparation as described in the main text ($26\times 26{\mathrm{nm}}^2$). Different crystallographic facets are observed ($V=0.35\mathrm{V}$, $I=3\mathit{nA}$). (b) image of a (110) facet ($V=-0.9\mathrm{V}$, $I=1.4\mathit{nA}$), (c) image of a (210) facet ($V=0.1\mathrm{V}$, $I=1.4\mathit{nA}$). The dashed line drawn on top of the bright lines on the upper terrace shows that the lines on the upper terrace are shifted with respect to those of the lower terrace as expected for a (210) facet. Insets in (b) and (c) show zoomed-in topographies of the facets; for the (210) facet [in (c)] a simple ball model of the surface is also shown. (The model can be drawn on top of the topography in two equivalent ways, only one of them is shown; the atoms shown in light gray correspond to Al atoms, the ones in purple to Yb atoms.) Tunneling parameters for the insets: (b) $V=1.8\mathrm{V}$, $I=3\mathit{nA}$, (c) $V=0.5\mathrm{V}$, $I=3\mathit{nA}$. (d) and (e) Line cuts across a step edge on a (110) and (210) facet, respectively.Reuse & Permissions

Figure 3

(a) and (b) Model of the (110) and (210) facets without reconstruction (left) and with a proposed model for the reconstruction (right); atomically resolved STM topographs of the same surfaces can be seen in Figs. 2b and 2c. (c)–(e) Closeups of the facets considered. Shown are topographies and ball models for (c) a (611) facet ($V=0.346\mathrm{V}$, $I=3\mathit{nA}$), (d) a (522) facet ($V=0.349\mathrm{V}$, $I=3\mathit{nA}$), and (e) a (931) facet ($V=-22\mathit{mV}$, $I=0.31\mathit{nA}$). Ytterbium atoms are shown as purple balls, aluminum atoms as gray balls.Reuse & Permissions

Figure 4

(a) Same topography as shown in Fig. 2a overlayed with a colored map, where the color encodes the local angle relative to a (522) facet. The pseudo-3D projection shows the surface with isotropic scale (lateral and vertical dimensions are scaled by the same factor). (b) Histogram of the image in (a) showing peaks at the angles at which a facet is seen in the topography, black vertical lines indicate the expected angles (see also Table  for a direct comparison). There is a systematic deviation between expected and measured angles which increases with increasing angle, which is likely due to uncertainty in the piezocalibration and drift.Reuse & Permissions

Figure 5

(a) Overview tunneling spectrum acquired on a (110) facet at 77 K, the spectrum is composed of two spectra taken with different stabilization conditions before switching off the feedback loop ($V=0.4\mathrm{V},I=0.8\mathit{nA}$; $V=2.546\mathrm{V},I=3\mathit{nA}$) (Ref. 26). Besides a peak close to the Fermi energy (marked as K), a broad maximum around $-0.5\mathrm{V}$ (marked as S) and two satellites (SO1 and SO2) can be seen. (b) STS on different crystallographic facets of ${\mathrm{YbAl}}_3$; spectra are acquired at 6 K except for the (110) facet (77 K). Spectra are shifted vertically for clarity.Reuse & Permissions

Figure 6

(a) STS on different crystallographic facets of ${\mathrm{YbAl}}_3$ close to the Fermi energy, spectra are acquired at 6.7 K except for the (110) facet (77 K); the shaded regions tentatively assign the resonances to bulk (I) and surface (II) related features [solid lines show fits of Eq. (2)]. (b) Spectra on a (210) facet taken at 6.7 K in different locations on the surface. The red solid lines show fits of the sum of two Fano functions to the 6 K spectra to extract the peak positions and widths. Vertical lines are shown as a guide to the eye to allow for a comparison of the peak positions. Shading of regions as in (a), spectra in (a) and (b) are shifted vertically for clarity. Inset in (b) shows the two positions where the spectra have been taken on the (210) facet. (c) Temperature dependence of spectra on the (210) facet; shown are a spectrum acquired at 6.7 K and at 77 K. In addition, shown as a solid red line, the 6.7 K spectrum is broadened as if it would have been taken at 77 K by accounting for the broadening of the Fermi distribution. The comparison shows that the changes in spectra cannot be explained due to only the temperature broadening.Reuse & Permissions