Substrate-mediated band-dispersion of adsorbate molecular states - Supplementary Information

Transcription

1 Substrate-mediated band-dispersion of adsorbate molecular states - Supplementary Information M. Wießner, 1, 2 J. Ziroff, 1, 2 F. Forster, 1, 2 M. Arita, 3 K. Shimada, 3 P. Puschnig, 4 A. Schöll*, 1, 2, a) and F. Reinert 1, 2 1) Universität Würzburg, Experimentelle Physik VII & Röntgen Research Center for Complex Material Systems RCCM, Würzburg, Germany 2) Karlsruher Institut für Technologie KIT, Gemeinschaftslabor für Nanoanalytik, Karlsruhe, Germany 3) Hiroshima Synchrotron Radiation Center, Hiroshima University, Hiroshima , Japan 4) Institut für Physik, Karl-Franzens Universität Graz, Austria (Dated: 29 January 2013) a) Electronic mail: achim.schoell@physik.uni-wuerzburg.de

2 2 2 parallel momentum k 001 [Å -1 ] b 1a parallel momentum k 1-10 [Å -1 ] FIG. S1. Angle dependent photoelectron intensity distribution: Calculated angular intensity distribution of the lowest unoccupied molecular orbital of PTCDA from 27. The white arrows show the directions of the angle dependent measurements in Figure 1a and b

3 3 FIG. S2. Photoemission intensity normalization: Photoelectron spectroscopy (PES) data along the [ 110 ] -direction before (a) and after intensity normalization (b). The low energy regime ev (i.e. above the black horizontal line in (b)) was normalized to an angle dependent intensity distribution derived by integrating over the binding energy regime of the lowest unoccupied molecular orbital (LUMO) indicated by the solid horizontal lines in a. The high energy regime ev was normalized accordingly to the highest occupied molecular orbital region (see dashed lines in a). The evolution of the Ag sp-bands is clarified by dashed lines in a and b. (c) waterfall graph of individual energy distribution curves of the normalized PES data in b. (d), (e): Simulated influence of Ag-sp-bands by a superposition of the photoemission intensities from molecule and substrate for (d) a flat and (e) a dispersing LUMO band.the intensity normalized was identical to b.

4 4 FIG. S3. Three-dimensional real space model of the PTCDA/Ag(110) brickwall layer: The iso-surfaces show the partial charge densities of the lowest unoccupied molecular orbital (LUMO) (a), integrated from 0.45 ev to 1.0 ev binding energy and of the highest occupied molecular orbital (HOMO) (b), integrated from 1.35 ev to 1.9 ev. The LUMO shows a significant delocalization within the top-most Ag-plane, while this effect is very small for the HOMO.

5 5 FIG. S4. Electron diffraction patterns of PTCDA/Ag(110): low energy electron diffractograms of 1 monolayer PTCDA on Ag(110) recorded with and electron energy of 25 ev at (a) T < 60 K and (b) room temperature. The lateral order is not effected by the temperatures variation.

6 6 I. SUPPLEMENTARY NOTE 1 - INTENSITY NORMALIZATION To even out the strong photoelectron intensity effects shown in Supporting Figure S1 and to enable access to the dispersion, the intensity has to be normalized. For the low binding energy regime of 3,4,9,10-perylene-tetracarboxylicdianhydride (PTCDA) (0-1.5 ev, i.e. above the solid line in Supporting Figure S2b) an angle dependent intensity distribution derived by integrating the intensities in the energy regime of the lowest unoccupied molecular orbital (LUMO) was used as a reference (see solid lines in Supporting Figure S2a). The high binding energy regime ( ev) was normalized accordingly on a reference derived by integrating the energy regime of the highest occupied molecular orbital (HOMO) (see dashed lines Supporting Figure S2a). To give the reader a better impression of the LUMO band dispersion in comparison to its energetic broadening, Supporting Figure S2c displays the energy distribution curves for different k-values in a waterfall graph.

7 7 II. SUPPLEMENTARY NOTE 2 - EXCLUSION OF POSSIBLE ARTEFACTS IN THE DETERMINATION OF BANDWIDTH There are a some effects which can in principle lead to experimental observations similar to the dispersing molecular states reported in this work and which thus have to be considered. First, a periodic variation of spectral weight of two (or more) signals not resolved in the spectra can lead to a periodic energy shift of the resulting sum feature. However, in case of PTCDA on Ag(110) the unit cell contains only one molecule and only one non-degenerate state exists in the binding energy region of the LUMO 29,30,32, Another issue is a possible superposition of signals from the substrate. Since the Ag 4d-states are at binding energies above E B > 3 ev and the Shockley state, which appears at E B 60 mev at the Y -point of the clean substrate 56,57, is shifted above the Fermi level upon adsorption of organic molecules 19 22,24,25 only the sp-bands appear in the LUMO binding energy region. To figure out if the crossing sp-bands can cause an apparent LUMO dispersion, we performed a simulation of the molecule-metal system. The backscattered sp-states 58 were modeled by parabolas with a similar effective mass and energetic broadening as observed in the experiment. These bands were superimposed to a flat (Supporting Figure S2d) and a dispersing LUMO band (Supporting Figure S2e), and subsequently normalised as described above. The simulation with the flat band shows substantial deviations from our experimental observation, thus providing strong evidence that the periodic binding energy variation in Figures 1a and b are not caused by crossing sp-bands. Other effects due to the angle-resolved measurements can be excluded due to the fact that the HOMO at higher binding energies does not show such a characteristic behavior.

8 8 III. SUPPLEMENTARY NOTE 3 - TEMPERATURE DEPENDENCE OF THE PTCDA MONOLAYER GEOMETRY To exclude a possible change of the lateral order of the PTCDA layer upon varying the sample temperature, we performed low energy electron diffraction experiments at different temperatures. Supporting Figure S4 shows the corresponding diffractograms at a substrate temperature of T < 60 K and at room temperature. Both images show the identical, well-known ( ) superstructure of the brickwall phase of the PTCDA monolayer on Ag(110), which can be 2 3 described by a superstructure matrix, resulting in a length of the unit vectors of b = b 2 = 11.9 Å 59.

9 9 SUPPLEMENTARY REFERENCES 52 Romaner, L., Nabok, D., Puschnig, P., Zojer, E. & Ambrosch-Draxl, C. Theoretical study of PTCDA adsorbed on the coinage metal surfaces, Ag(111), Au(111) and Cu(111). New J. Phys. 11 (2009). 53 Greuling, A., Rohlfing, M., Temirov, R., Tautz, F. S. & Anders, F. B. Ab initio study of a mechanically gated molecule: From weak to strong correlation. Phys. Rev. B 84, (2011). 54 Refaely-Abramson, S. et al. Quasiparticle spectra from a non-empirical optimally-tuned range-separated hybrid density functional. ArXiv e-prints (2012) Braun, D., Schirmeisen, A. & Fuchs, H. Molecular growth and sub-molecular resolution of a thin multilayer of PTCDA on Ag(110) observed by scanning tunneling microscopy. Surf. Sci. 575, 3 11 (2005). 56 Bartynski, R. A. & Gustafsson, T. Experimental study of surface states on the (110) faces of the noble metals. Phys. Rev. B 33, (1986). 57 Hayoz, J., Pillo, T., Naumovic, D., Aebi, P. & Schlapbach, L. Growth of Au on Ag(110): electronic structure by photoemission. Surf. Sci , (1999). 58 Bocquet, F. C. et al. Final-state diffraction effects in angle-resolved photoemission at an organic-metal interface. Phys. Rev. B 84, (2011). 59 Glöckler, K. et al. Highly ordered structures and submolecular scanning tunnelling microscopy contrast of PTCDA and DM-PBDCI monolayers on Ag(111) and Ag(110). Surf. Sci. 405, 1 20 (1998).