Orbital Selectivity in Scanning Tunneling Microscopy: Distance-Dependent Tunneling Process Observed in Iron Nitride

In scanning tunneling microscopy, orbital selectivity of the tunneling process can make the topographic image dependent on a tip-surface distance. We have found reproducible dependence of the images on the distance for a monatomic layer of iron nitride formed on a Cu(001) surface. Observed atomic images systematically change between a regular dot array and a dimerized structure depending on the tip-surface distance, which turns out to be the only relevant parameter in the image variation. An accompanied change in the weight of $\mathrm{Fe}\text{−}3d$ local density of states to a tunneling background was detected in $dI/dV$ spectra. These have been attributed to a shift in surface orbitals detected by the tip from the $d$ states to the $s/p$ states with increasing the tip-surface distance, consistent with an orbital assignment from first-principles calculations.

Figure 1

(a) Large scale topographic image at $V_s=100\mathrm{mV}$, $I=5\mathrm{nA}$. (b) LEED pattern obtained with an incident electron energy of 100 eV. (c) Close view of a topographic image for the ${\mathrm{Fe}}_2\mathrm{N}$ islands revealing a dimerization of Fe atoms ($V_{\mathrm{s}}=50\mathrm{mV}$, $I=5\mathrm{nA}$). The dimerization is indicated by encirclement. (d) Crystal structure model of ${\mathrm{Fe}}_4\mathrm{N}$. A dotted parallelogram represents an ${\mathrm{Fe}}_2\mathrm{N}$ plane. (e) Schema of $p4g(2\times 2)$ surface reconstruction corresponding to the LEED pattern. From an unreconstructed coordination (dotted circles), each two of Fe atoms dimerizes in two perpendicular directions indicated by arrows.

Figure 2

Current dependence of STM images. (a) Topographic images taken at $V_s=0.25\mathrm{V}$ with varying $I$ from 0.1, 3.0, 10 to 45 nA. (b) Line profiles at $V_s=0.25\mathrm{V}$, measured along lines indicated in (a). From the top to the bottom, $I$ varies as follows: 45, 40, 35, 30, 28, 25, 22, 20, 18, 15, 12, 11, 10, 9.0, 8.0, 7.0, 5.0, 3.0, 2.0, 1.0, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, and 0.1 nA, respectively. Empty circles indicate peak positions extracted from one Fe dimer.

Figure 3

(a) Spin-resolved LDOS of single-layer ${\mathrm{Fe}}_2\mathrm{N}$ on Cu(001). Fe and N states are separately shown. (b) $dI/dV$ spectra of ${\mathrm{Fe}}_2\mathrm{N}$ (solid) and bare Cu (dotted). The STM tip was stabilized at $I=30\mathrm{nA}$ and $V_s=1\mathrm{V}$. (c) Threshold $d_c$ for the image change. Empty circles indicate parameter sets when the shift in the image from the dimerized to the dot structure occurred. Lines fitted freely to the experimental data with assuming the constant $d$ (solid), $E$ (dot-dashed), $V_{\mathrm{s}}$ (fine-dashed), $P$ (rough-dashed), or $I$ (dashed) in Eq. (1) are also indicated. (d) Distant-dependent $dI/dV$ spectra measured at $d=4.2$, 3.4, and 2.9 Å from the bottom to the top. Dashed curves indicate a tunneling background obtained by a Tersoff-Hamann approximation.

Figure 4

Total and orbital-resolved LDOS of (a) surface and (b) vacuum (4 Å above the surface) layers. The states with negligible intensity in this energy range are not shown here. In the vacuum layer, LDOS cannot be separated into each atomic component due to a strong orbital mixing between Fe and N.

Figure 5

Charge distributions calculated at different distances. (a) Unreconstructed $c(2\times 2)$ surface model used for the calculation. Large (small) spheres correspond to the surface Fe (N) atoms. (b,c) Charge distribution at $d=1.8$ (b) and $d=6.5\AA$ (c).