Figure 1

(a) Atomically resolved image of one-atomic-layer NaBr. Image size $7.80\times 7.80{\mathrm{nm}}^2$. (b) Schematic of the CuPc molecule. (c) Large-size image ($38.65\times 38.65{\mathrm{nm}}^2$) after dosing CuPc molecules. One-, two-, and three-atomic-layer NaBr are indicated as I, II, and III, respectively. All three layers exhibit mismatch-induced long-range modulation along the row direction of the underlying NiAl(110) substrate, which roughly runs diagonally from the lower-left corner to the upper-right corner. (d) One CuPc molecule adsorbed on one-atomic-layer NaBr. This configuration is denoted as $A$. Image size $4.70\times 4.70{\mathrm{nm}}^2$. (e) One CuPc on two-atomic-layer NaBr. This configuration is denoted as $B$. Image size $4.70\times 4.70{\mathrm{nm}}^2$. (f) Two CuPc on three-atomic-layer NaBr, with the lower one in configuration $A$ and the higher one in $B$. Image size $9.40\times 9.40{\mathrm{nm}}^2$. The left palette (color online) is used in image (c) while the right palette (grayscale) is used in images (a),(d),(e), and (f).Reuse & Permissions

Figure 2

$dI/dV$ spectra of CuPc measured on (a) bare NiAl(110) surface, (b) one-atomic-layer NaBr, (c) two-atomic-layer NaBr, and (d) three-atomic-layer NaBr. Solid lines represent CuPc spectra taken at the center of the molecule, dotted lines are background spectra taken at a position away from the molecule, and dashed lines indicate zero of $dI/dV$. In (a), the CuPc spectrum was taken with the tunneling gap set at (1.50 V, 120 pA), and the NiAl spectrum was taken with the set point (2.50 V, 60 pA). The bias voltage was not ramped above 1.5 V over the molecule because higher bias voltage might induce molecular motion. All the spectra in (b), (c), and (d) were taken with the set point (2.50 V, 100 pA), and all the corresponding CuPc molecules were adsorbed in configuration $A$.Reuse & Permissions

Figure 3

Eletronic energy level diagrams of the double-barrier tunnel junction. Dark colors represent filled states, light colors represent empty states, and white represents the absence of allowed states. $E_C$ ($E_V$) denotes the conduction (valence) band edge of NaBr, and $E_M$ denotes the LUMO of CuPc. The vacuum barrier width is denoted as $z$, and the NaBr thickness is denoted as $d$. (a) The zero-bias case. The Fermi levels ($E_F$) of the tip and NiAl line up and lie in the middle of NaBr band gap. For simplicity, the work functions of the tip, NaBr, and NiAl were assumed to be the same; therefore, the vacuum levels ($E_{\mathrm{vac}}$) also lined up. (b) The case with positive sample bias. Note the finite voltage drop $\eta V$ across NaBr.Reuse & Permissions

Figure 4

(a) Local density of states $D(\epsilon )$ used in the calculation. Simulated $dI/dV$ spectrum for CuPc adsorbed on (b) one-atomic-layer NaBr ($z=0.75\mathrm{nm}$, $d=0.40\mathrm{nm}$), (c) two-atomic-layer NaBr ($z=0.65\mathrm{nm}$, $d=0.70\mathrm{nm}$), and (d) three-atomic-layer NaBr ($z=0.55\mathrm{nm}$, $d=1.00\mathrm{nm}$) [26]. The tip-sample distance is usually estimated to be around 1 nm in STM. Here the vacuum barrier width $z$ was chosen so that $z+d$ was of the order 1 nm. In the calculation, we assumed $E_{\mathrm{vac}}-E_F=5.0\mathrm{eV}$ and $E_C-E_F=4.0\mathrm{eV}$ at zero bias and we used the bulk NaBr dielectric constant ($\epsilon =6.3$). These parameters were varied by 10% and the results remained qualitatively the same.Reuse & Permissions