Inelastic Spectroscopy Identification of STM-Induced Benzene Dehydrogenation

Inelastic electron tunneling spectroscopy (IETS) performed with the scanning tunneling microscope (STM) has been deemed as the ultimate tool for identifying chemicals on the atomic scale. However, IETS-based chemical analysis is error-prone due to the numerous degrees of freedom of chemisorbed molecular systems. First-principles simulations of IETS are presented that, by quantitative comparison with the experimental spectra, permit one to determine the final products of an STM-induced reaction on chemisorbed benzene. Our simulations reveal that IETS possesses an enhanced sensitivity to atomic structure as compared to topographic imaging due to both its energy and space resolution.

Figure 1

(Top) Sketch of the optimized adsorption geometry for benzene and derivatives on Cu(100) including key bonding distances in Å. Dark gray (dark blue), light gray (blue), black, and white spheres correspond to surface Cu, sublayer Cu, C, and H atoms, respectively. (Center) Large-scale top view of the adsorption configurations on a $p(4\times 4)$ unit cell. (Bottom) Associated STM simulations ($10.32\AA \times 10.32\AA$) at the same constant current that leads to an overall tip-surface distance of $\sim 9\AA$ as indicated by the scale bar. Positions of C, H atoms are indicated with letters in the plots.

Figure 2

(Top) Constant probability amplitude plots of the LUMO of the free fragments, phenyl (${\mathrm{C}}_6{\mathrm{H}}_5$) (left) and benzyne (${\mathrm{C}}_6{\mathrm{H}}_4$) (right). (Bottom) Difference maps of electron density $\Delta \rho$ (interacting minus noninteracting) for fragments on copper surface. Regions of maximum electron accumulation or depletion are displayed in dark gray (red)/gray (blue) with an isosurface value of $\pm 0.005e/{\AA }^3$.

Figure 3

Sketch of all C-H stretch modes (frequencies included in meV) for adsorbed (a) phenyl and (b) benzyne molecules and their theoretical inelastic efficiencies over the $p(4\times 4)$ unit cell of Cu(100) ($10.32\AA \times 10.32\AA$) for the same simulation parameters as for Fig. 1. The computed relative change in conductance at the center of the unit cell is included in percent for each vibration mode.

Figure 4

Simulated IETS spectra of (a) ${\mathrm{C}}_6{\mathrm{H}}_5$, (b) ${\mathrm{C}}_6{\mathrm{D}}_5$, (c) ${\mathrm{C}}_6{\mathrm{H}}_4$, and (d) ${\mathrm{C}}_6{\mathrm{D}}_4$: change of conductance as a function of the tip-surface bias voltage above the center of each fragment. The black solid line is the total spectra, and the dashed lines are the contributions of each mode to the total spectra. For (a) and (b) phenyl species, individual contributions display a doublet peak corresponding to the experimental one [11]. For (c) and (d) benzyne species, individual contributions are localized in energy, thus yielding a more intense single peak different from the experimental one.