Edge-On Bonding of Benzene Molecules in the Second Adsorbed

Abstract. Abstract Image. The bilayer of benzene on Cu(110) was studied with ... STM images of the β layer reveal closely spaced edge-on benzene mole...
0 downloads 0 Views 474KB Size
15645

2006, 110, 15645-15649 Published on Web 07/21/2006

Edge-On Bonding of Benzene Molecules in the Second Adsorbed Layer on Cu(110) Junseok Lee, Daniel B. Dougherty, and John T. Yates, Jr.* Department of Chemistry, Surface Science Center, UniVersity of Pittsburgh, Pittsburgh, PennsylVania 15260 ReceiVed: April 26, 2006; In Final Form: June 28, 2006

The bilayer of benzene on Cu(110) was studied with temperature-programmed desorption (TPD), time-offlight electron stimulated desorption ion angular distribution (TOF-ESDIAD), and scanning tunneling microscopy (STM). TPD spectra show that three well-defined adsorption states exist. The R layer corresponds to the first layer containing flat-lying benzene molecules. As coverage increases, the β layer forms on top of the R layer, and eventually, a multilayer, γ, forms. TPD measurements show that the number of benzene molecules in the β layer is equal to the number of benzene molecules in the R layer. ESDIAD measurements establish that the orientation of the benzene molecules in the β layer is edge-on, with two C-H bonds directed toward the surface. STM images of the β layer reveal closely spaced edge-on benzene molecules arranged in repeating hexagons, as well as loosely spaced benzene molecules with greater apparent height, which are also edge-on species. Correlation between the different measurements suggests a structural model for the benzene bilayer.

The determination of the structure of ultrathin organic films and the growth mechanism of the films, including the relative molecular orientation between layers, are important due to current interest in the optimization of organic molecular electronic devices based on these materials.1 A delicate balance between adsorbate-adsorbate and adsorbate-substrate interactions plays a crucial role in determining the geometry of such layers. When building thin films on a surface, the interaction between first- and second-layer molecules is important. The layer-by-layer growth of a thin organic layer can often be witnessed by surface measurement methods.2,3 However, determining the orientation of individual molecules in adjacent layers presents some difficulties. Many surface science methods do not directly provide orientational information about surface species, and those methods which are sensitive to molecular orientation often provide averages of molecular orientational behavior measured over adjacent layers. In molecular crystals of benzene, which is a prototype of more complex crystals composed of aromatic molecules, a key structural motif is the “T” shape, with benzene rings in different layers oriented perpendicular to each other. Various measurements and calculations of the structure of benzene in the solid4-6 and liquid7 show that T-shaped molecular assemblies are favored and that the main intermolecular forces responsible are dispersion forces. In the gas phase, parallel-displaced and T-shaped (π-hydrogen-bonded) structures are both found to be the most stable configurations with a very small difference in binding energy.8-12 The adsorption of benzene on the single-crystal Cu(110) surface has been studied by various methods.13-18 A common observation is that the benzene adsorbs in the first layer with * E-mail: [email protected].

10.1021/jp0625661 CCC: $33.50

its plane parallel to the surface and with slight bending of the C-H bonds away from the surface;16 there is also a report of a tilted first layer adsorption geometry.14 Studies dealing with higher-coverage behavior are rare despite the importance in understanding the mechanism of thin film growth. In this Letter, we report temperature-programmed desorption (TPD), time-of-flight electron stimulated desorption ion angular distribution (TOF-ESDIAD), and scanning tunneling microscopy (STM) experiments that probe the structure of the stable benzene bilayer on Cu(110). We provide evidence that benzene molecules in the second layer are oriented in an edge-on configuration and that the second layer molecules are bound with two adjacent C-H bonds directed toward the surface plane. This edge-on configuration may correspond to the earliest stages of growth of a supported molecular crystal of benzene and is notably different than bilayer structures proposed for benzene on Cu(111).19,20 We also provide evidence that there is a preferential azimuthal orientation of the edge-on benzene molecules in the second layer. Experiments were performed in two separate ultrahighvacuum (UHV) chambers.21 The C6D6 (99.5%, Cambridge Isotope Laboratories) was used for TPD and ESDIAD experiments after purification by freeze-pump-thaw cycles. The background hydrogen contribution to the ESDIAD signal could be eliminated by using fully deuterated benzene. A calibrated microcapillary array beam doser was used to achieve accurate exposure in the TPD-ESDIAD chamber, which operated with a typical background pressure of 1 × 10-10 mbar with a crystal temperature of 81 K. All ESDIAD measurements involve ninepoint smoothing and twofold symmetrization. STM experiments were carried out using a commercial instrument (Omicron LTSTM) housed in a second UHV chamber with base pressure below 1 × 10-11 mbar. The Cu(110) crystal was held at 5 K © 2006 American Chemical Society

15646 J. Phys. Chem. B, Vol. 110, No. 32, 2006

Figure 1. TPD spectra of C6D6 adsorbed at 81 K on atomically clean Cu(110). The heating rate is 2 K/s. Spectra a: Low-coverage regime for exposures less than 3.25 × 1014 molecules‚cm-2. The exposures to C6D6 in units of 1014 molecules‚cm-2 are (a), 0.027; (b), 0.055; (c), 0.096; (d), 0.269; (e), 0.543; (f), 1.08; (g), 1.44; (h), 2.15; (i), 2.70; (j), 3.25. Spectra b: Higher-coverage regime. Benzene and its mass spectrometer cracking products are the only species detected. The exposures to C6D6 in units of 1014 molecules‚cm-2 are (k), 3.24; (l), 3.98; (m), 4.32; (n), 5.39; (o), 6.46; (p), 8.69; (q), 10.79.

for imaging, and benzene was dosed through a stainless steel tube onto the crystal held at a temperature between 20 and 100 K. After dosing, the crystal was annealed to approximately 100 K. Imaging was performed in the constant-current mode with tungsten tips. To compare benzene coverages in the STM chamber to coverages in the TPD-ESDIAD chamber, thermal desorption experiments were performed by warming the crystal + benzene layer from 10 to ∼250 K through contact with the wobble stick holder at 300 K. Here, an ionization gauge was used as a pressure sensor. Figure 1 shows a sequence of TPD spectra for benzene (C6D6) as a function of exposure to a calibrated effusive beam on an atomically clean Cu(110) surface at 81 K. Initially, during the

Letters filling of the first layer, a desorption feature labeled R is observed, as shown in Figure 1a. At the lowest coverage, the peak maximum at 279 K corresponds to a binding energy of 17.5 kcal/mol as determined by the fit (not shown) to firstorder desorption kinetics. The R peak shifts significantly to lower temperature, and the peak-width broadens as the monolayer is filled, consistent with previous results.22 This could be evidence for the buildup of repulsive intermolecular forces between benzene adsorbate molecules or an increase in the preexponential factor in the desorption rate constant as the spatial confinement of the mobile molecules decreases with increasing coverage,23 or a combination of these two effects. Figure 1b shows that, at an exposure of 3.98 × 1014 molecules‚cm-2, a second desorption process, labeled β, is observed at about 154 K. Increasing exposure causes extensive development of the β layer, producing a sharp desorption feature. The β layer desorbs with zero-order kinetics (not shown) and reaches saturation coverage. A careful analysis of relative TPD peak areas shows that the coverage ratio of the β layer to the R layer is 1:1, which is identical to the ratio in two adjacent layers in solid benzene.5,6 The same ratio has been obtained on the Cu(111) surface in a 2PPE study.20 We assign the β layer to the bilayer. Finally, at higher benzene exposures, a layer designated γ begins to develop, which corresponds to the multilayer of benzene, as seen in spectra p and q in Figure 1b. Figure 2 shows D+ ESDIAD measurements from C6D6 layers on Cu(110) at three representative coverages. Observation of a sharp D+ angular distribution corresponds to the presence of surface species with regular alignment of C-D chemical bonds. At coverages corresponding only to the R layer of C6D6, in Figure 2a, a low yield of D+ ions is observed with no measurable ion angular distribution. Such an observation is consistent with the fact that benzene molecules in the first layer adsorb with their molecular planes parallel to the surface. This results in C-D bonds which are parallel to the surface and “ESDIADinactive”. Slight upward bending of the C-D bonds away from the surface would also not result in a measurable D+ yield due to efficient neutralization of D+ ions ejected at glancing angles.

Figure 2. D+ ESDIAD patterns from C6D6 layers on Cu(110) at 81 K in different coverage regimes. Three-dimensional ESDIAD patterns are also shown for ease of understanding. Scale bars are shown on upper patterns for ion yield comparison and ion yields per electron, YD+, are indicated at the bottom part of the figure. (a) Monolayer R-C6D6 species. (b) Bilayer (R- and β-C6D6 species). (c) Multilayer C6D6.

Letters

Figure 3. (a) STM image of the edge of a β-phase island on top of an R-phase surface (26.5 nm × 26.5 nm, -0.4 V, 0.1 nA). (b) Total pressure measurement showing two distinct desorption peaks observed during warming of the STM sample holder when the β phase was present on the surface (compare with Figure 1).

The ESDIAD pattern in Figure 2b, corresponding to the bilayer or β phase, has several important features. First, there is a significant increase in the yield of D+ ions desorbing from the surface when compared to the yield in the R phase. This indicates that C-D bonds must be directed away from the surface plane in the β phase. In addition, two sharp D+ beams emerge in the ion-angular distribution oriented along the 〈11h0〉 azimuth of the surface. The crystal bias dependence of the position of the two distinct beams was used to determine the angle of ejection of the D+ ions from the surface normal. When corrected for ion-optical24 and final state effects,25,26 this angle is 29.5 ( 1°, corresponding to C-D bond directions expected from edge-bound benzene molecules whose molecular planes are oriented parallel to the 〈11h0〉 azimuth of the Cu(110) substrate. Another significant ESDIAD feature in Figure 2b is evident. Accompanying the pair of sharp D+ beams is a volcanoshaped D+ pattern exhibiting a region of reduced D+ yield at its center. The two sharp D+ beams are superimposed on the volcano rim, which is clear in the three-dimensional ESDIAD pattern in Figure 2b. The observation of a circularly symmetrical volcano-shaped D+ pattern, with the volcano rim at a D+ polar angle near 30° from the normal, indicates the presence of edgebound C6D6 molecules which do not have a discernible preference for azimuthal orientation. At coverages involving multilayers of C6D6, a broad normally oriented D+ ion angular distribution is observed, characteristic of multilayers of randomly oriented molecules as shown in Figure 2c. An STM image showing coexisting domains of the R and β phases is shown in Figure 3a. When the phase assigned as the β (or bilayer) structure was present on the surface, a sharp thermal desorption peak could always be observed in addition

J. Phys. Chem. B, Vol. 110, No. 32, 2006 15647 to the broad, higher-temperature peak associated with the R phase. This is illustrated by the thermal desorption measurements in the STM chamber shown in Figure 3b. The R phase, which is completed at a benzene coverage of ∼2.5 × 1014 molecules/cm2, displays a local c(4 × 2) superlattice that is interrupted by numerous extended vacancy lines as shown in the upper-right part of Figure 3a. At higher benzene exposures, the β phase nucleates in islands on top of the R phase. This island formation in the bilayer (R + β) coverage regime is consistent with previous results on Cu(111).20 STM images of the β-phase islands reveal substantial complexity. Two differently imaged topographic features were always visible in the bilayer regions. In the β-phase region of Figure 3a, the darkest regions are spaces between molecules where the tip cannot penetrate to image the underlying R phase. There are molecular features arranged in a honeycomb lattice of repeating hexagons as well as molecular features with greater apparent heights that are imaged as irregularly shaped protrusions (marked “Bright” in Figure 3a). The six molecular features in the hexagons (as indicated by the arrow in Figure 3a) are spaced apart by about 0.5 nm, and the different hexagons are spaced apart by about 1.4 nm. The unit cell of these features is illustrated more clearly in Figure 4a. Whether the hexagon features or the greater apparent height bright features were imaged depended on the condition of the STM tip and the chosen tunneling conditions. While we do not understand the imaging mechanism that leads to the preferential imaging of one set of features or the other, both features were always visible in the β phase. Figure 4b shows an STM image obtained under tunneling conditions where the brightest bilayer features were imaged more clearly than in Figures 3a and 4a. The densest arrangement of these features was observed to form a locally centered rectangular unit cell structure approximately 1.45 nm × 0.85 nm, with the long dimension along the 〈11h0〉 azimuth as indicated by the rectangle in the figure. Despite these regular structures, the bright features were usually very loosely packed, and their appearance was significantly variable depending on the condition of the STM tip. The superlattice of bright features in the β phase is less dense than the superlattice of benzene molecules in the R phase. Thus, to explain the 1:1 ratio of desorption peak areas of the R and β phases as measured by TPD, it is necessary to attribute both of the different topographic features (i.e., lower honeycomb features and higher bright features) in Figure 3 to β-phase molecules on top of an intact R phase. The frequent observation of centered rectangular arrangements of the bright topographic features with long edges along the 〈11h0〉 azimuth suggests that the C6D6 molecules in this phase are responsible for the regular C-D bond directions that give rise to the sharp 〈11h0〉-oriented two-beam ESDIAD pattern shown in Figure 2b. The packing of the less-bright β-phase features in repeated hexagons is dense enough (∼0.5 nm spacing between centers of the features) that these molecules must also be upright on the surface. The lack of correlation to the 〈11h0〉 azimuthal direction makes it unlikely that the C6D6 molecules comprising the honeycomb lattice contribute to the two-beam D+-ESDIAD pattern. Instead, the C6D6 molecules in these regions probably correlate with the volcano D+-ESDIAD pattern which exhibits no discernible preferential azimuthal features. Figure 4c,d shows tentative surface structures correlating with the two unit cells observed by STM with different imaging conditions. It may be noted that the only difference between the two unit cells is the presence of two extra edge-on benzene

15648 J. Phys. Chem. B, Vol. 110, No. 32, 2006

Letters

Figure 4. (a) STM image obtained on a bilayer island showing coexisting bright and dark topographic protrusions (13.1 nm × 13.1 nm, -0.4 V, 0.1 nA). (b) STM image where bright protrusions are more clearly resolved (16.6 nm × 16.6 nm, +0.9 V, 0.25 nA). (c,d) Tentative structural models of the β or bilayer phase of benzene on Cu(110). Orange molecules arranged in tightly packed hexagons represent the darker features observed by STM and light-green features represent the loosely packed, brighter features.

molecules (light green) for the rectangular unit cell which are aligned with their molecular planes parallel to the 〈11h0〉 azimuth. We postulate that the edge-on C6D6 molecules lying parallel to the 〈11h0〉 azimuth are rather rigidly held on the surface, giving the two-beam D+ ESDIAD pattern. The other edge-on C6D6 molecules, not aligned parallel to the 〈11h0〉 azimuth, contribute to the volcano pattern and do not yield sharp D+ beams. This may be due to thermal vibration of these second-layer molecules at the ESDIAD imaging temperature of 81 K. The registry of the β phase with the underlying R phase is not known at present, and this is an important issue in understanding the full structure of the benzene bilayer on Cu(110). Nevertheless, the arrangements illustrated in Figure 4c,d point out the unique edge-on benzene bonding resulting from two superimposed second-layer benzene superlattices on the Cu(110) template. To summarize our findings, ESDIAD and STM studies have clearly observed flat-lying benzene molecules in the R layer on Cu(110). As one enters the (R + β)-bilayer regime of coverage, edge-bound benzene molecules are observed by ESDIAD and by STM to fall into two classes: (1) 〈11h0〉 azimuthally oriented molecules with two C-D bonds oriented upward (and downward) at a 30° angle to the normal, and (2) random azimuthally oriented edge-on benzene molecules, also projecting C-D bonds at a 30° angle to the normal. The multilayer forming on top of the (R + β) bilayer exhibits completely random C-D bond orientation.

Acknowledgment. The authors wish to thank the W. M. Keck Foundation through the W. M. Keck Center for Molecular Electronics and NEDO (Japan) for support of this work. References and Notes (1) Forrest, S. R. Nature (London) 2004, 428, 911. (2) Casalis, L.; Danisman, M. F.; Nickel, B.; Bracco, G.; Toccoli, T.; Iannotta, S.; Scoles, G. Phys. ReV. Lett. 2003, 90, 206101. (3) Lukas, S.; So¨hnchen, S.; Witte, G.; Wo¨ll, C. ChemPhysChem 2004, 5, 266. (4) Cox, E. G. ReV. Mod. Phys. 1958, 30, 159. (5) Cox, E. G.; Cruickshank, D. W. J.; Smith, J. A. S. Proc. R. Soc. London, Ser. A 1958, 247, 1. (6) Craven, C. J.; Hatton, P. D.; Howard, C. J.; Pawley, G. S. J. Chem. Phys. 1993, 98, 8236. (7) Narten, A. H. J. Chem. Phys. 1968, 48, 1630. (8) Hobza, P.; Selzle, H. L.; Schlag, E. W. J. Am. Chem. Soc. 1994, 116, 3500. (9) Tsuzuki, S.; Honda, K.; Uchimaru, T.; Mikami, M.; Tanabe, K. J. Am. Chem. Soc. 2002, 124, 104. (10) Tsuzuki, S.; Uchimaru, T.; Sugawara, K.; Mikami, M. J. Chem. Phys. 2002, 117, 11216. (11) Sinnokrot, M. O.; Valeev, E. F.; Sherrill, C. D. J. Am. Chem. Soc. 2002, 124, 10887. (12) Sinnokrot, M. O.; Sherrill, C. D. J. Phys. Chem. A 2004, 108, 10200. (13) Haq, S.; King, D. A. J. Phys. Chem. 1996, 100, 16957. (14) Lomas, J. R.; Baddeley, C. J.; Tikhov, M. S.; Lambert, R. M. Chem. Phys. Lett. 1996, 263, 591. (15) Doering, M.; Rust, H.-P.; Briner, B. G.; Bradshaw, A. M. Surf. Sci. 1998, 410, L736. (16) Weinelt, M.; Wassdahl, N.; Wiell, T.; Karis, O.; Hasselstro¨m, J.; Bennich, P.; Nilsson, A.; Sto¨hr, J.; Samat, M. Phys. ReV. B 1998, 58, 7351.

Letters (17) Triguero, L.; Luo, Y.; Pettersson, L. G. M.; A° gren, H.; Va¨terlein, P.; Weinelt, M.; Fo¨hlisch, A.; Hasselstro¨m, J.; Karis, O.; Nilsson, A. Phys. ReV. B 1999, 59, 5189. (18) Rogers, B. L.; Shapter, J. G.; Ford, M. J. Surf. Sci. 2004, 548, 29. (19) Xi, M.; Yang, M. X.; Jo, S. K.; Bent, B. E.; Stevens, P. J. Chem. Phys. 1994, 101, 9122. (20) Velic, D.; Hotzel, A.; Wolf, M.; Ertl, G. J. Chem. Phys. 1998, 109, 9155. (21) Lee, J.; Dougherty, D. B.; John T. Yates, J. J. Phys. Chem. B 2006, 110, 9939.

J. Phys. Chem. B, Vol. 110, No. 32, 2006 15649 (22) Lomas, J. R.; Baddeley, C. J.; Tikhov, M. S.; Lambert, R. M. Langmuir 1995, 11, 3048. (23) Frechard, F.; van Santen, R. A.; Siokou, A.; Niemantsverdriet, J. W.; Hafner, J. J. Chem. Phys. 1999, 111, 8124. (24) Gao, Q.; Cheng, C. C.; Chen, P. J.; Choyke, W. J.; Yates, J. T., Jr. J. Chem. Phys. 1993, 98, 8308. (25) Misˇkovic´, Z.; Vukanic´, J.; Madey, T. E. Surf. Sci. 1984, 141, 285. (26) Misˇkovic´, Z.; Vukanic´, J.; Madey, T. E. Surf. Sci. 1986, 169, 405.