Halogen-Substituted Thiophenol Molecules on Cu(111)

10928 Langmuir 2004, 20, 10928-10934

K. L. Wong,† X. Lin,† K.-Y. Kwon,† G. Pawin,† B. V. Rao,† A. Liu,† L. Bartels,*,† S. Stolbov,‡ and T. S. Rahman*,‡

Pierce Hall, University of California, Riverside, California 92521, and Department of Physics, Kansas State University, Manhattan, Kansas 66506

Received July 16, 2004. In Final Form: September 19, 2004

Para-halosubstituted thiophenols (X-TPs, where X is Br, Cl, or F) form ordered islands and monolayers on Cu(111) at temperatures as low as 81 K. At incomplete coverages, all X-TPs adsorb with the dehydrogenated thiol group attached to the substrate and the substituted ring inclined toward the surface, as verified experimentally and theoretically. The structure of ordered islands has a pronounced dependence on the nature of the halogen substituent: while unsubstituted TP and pentafluoro-TP molecules do not self-assemble into extended ordered patterns at 81 K, X-TP molecules form a range of different structures which depend both on the size and electronegativity of the substituent, as well as on the coverage.

Introduction TP film forms a commensurate (x13 ×x13)R30° structure on Au(111).32 However, Dhirani et al. report that TP does The study of self-assembled monolayers (SAMs) is 15 motivated by a large array of potential applications that not form an ordered SAM in their STM study. In addition, - includes molecular electronic devices,1,2,3 chemical sen- a variety of results concerning the sulfur substrate sors,4 corrosion inhibitors,5 biocompatible coatings,6 etc. bonding, the precise orientation of the phenyl ring with respect to the surface, and the two-dimensional film While the majority of the studies on thiol-based SAMs 16-19 address dense coverages7,8 and the mechanism of SAM structure can be found in the literature. Biphenylic formation,9 a limited number of studies has investigated SAMs were found to lie close to the surface plane at lower coverage and to tilt toward the surface normal at higher the evolution of film structure with increasing coverage 20 and the interplay between intermolecular and molecule- coverage. substrate interactions.10 Poirier and Pylant11 proposed In ultrahigh vacuum (UHV), a clean Au(111) surface that SAMs form in a two-step mechanism driven by strong forms a herringbone reconstruction, which leads to dis- molecule-substrate interaction while being governed by similar surface sites and provides a complication for the a dynamical rather than an energetic equilibrium. investigation of the SAM formation process. Cu(111) The most popular SAM system is organothiols on shares many physical and chemical properties with 21 Au(111) due to their facile preparation and good stability Au(111) as far as thiol interactions are concerned, yet even in air.12 Experiments show a sequence of different it offers the advantages of the absence of a reconstruction. adsorbate patterns13,14 depending on coverage. Compared At the same time, it is structurally stable upon thiol to the alkanethiols, arenthiols have been studied less intensely. In the case of the simplest arenethiol, thiophenol (15) Dhirani, A. A.; Zehner, R. W.; Hsung, R. P.; Guyot-Sionnest, P.; Sita, L. R. J. Am. Chem. Soc. 1996, 118, 3319. (TP), Wan et al. concluded that the phenyl ring is tilted (16) Szafranski, C. A.; Tannaer, W.; Laibinis, P. E.; Garrell, R. about 30° from the surface normal and that a well-ordered Langmuir 1998, 14, 3570. (17) Carron, K. T.; Hurley, G. J. Phys. Chem. 1991, 95, 9979. * Authors to whom correspondence should be addressed. E- (18) Hasan, M.; Bethell, D.; Brust, M. J. Am. Chem. Soc. 2002, 124, 1132. mail: [email protected] (L.B.); [email protected] (19) Nuzzo, R. G.; Zegarski, B. R.; Dubois, L. H. J. Am. Chem. Soc. (T.S.R.). 1987, 109, 733. † University of California. (20) Leung, T. Y. B.; Schwartz, P.; Scoles, G.; Schreiber, F.; Ulman, ‡ Kansas State University. A. Surf. Sci. 2000, 458, 34. (1) Reed, M. A.; Zhou, C.; Muller, C. J.; Burgin, T. P.; Tour, J. M. (21) Ulman, A. Chem. Rev. 1996, 96, 1533. Science 1997, 278, 252. (22) Kariapper, M. S.; Grom, G. F.; Jackson, G. J.; McConville, C. F.; (2) Mantooth, B. A.; Weiss, P. S. Proc. IEEE 2003, 91, 1785. Woodruff, D. P. J. Phys.: Condens. Matter 1998, 10, 8661. (3) Collier, C. P.; Jeppesen, J. O.; Luo, Y.; Perkins, J.; Wong, E. W.; (23) Toomes, R. L.; Polcik, M.; Kittel, M.; Hoeft, J. T.; Sayago, D. I.; Heath, J. R.; Stoddart, J. F. J. Am. Chem. Soc. 2001, 123, 12632. Pascal, M.; Lamont, C. L. A.; Robinson, J.; Woodruff, D. P. Surf. Sci. (4) Flink, S.; van Veggel, F.; Reinhoudt, D. N. Adv. Mater. 2000, 12, 2002, 513, 437. 1325. (24) Jackson, G. J.; Woodruff, D. P.; Jones, R. G.; Singh, N. K.; Chan, (5) Scherer, J.; Vogt, M. R.; Magnussen, O. M.; Behm, R. J. Langmuir A. S. Y.; Cowie, B. C. C.; Formoso, V. Phys. Rev. Lett. 2000, 84, 119. 1997, 13, 7045. (25) Rieley, H.; Kendall, G. K.; Chan, A.; Jones, R. G.; Ludecke, J.; (6) Prime, K. L.; Whitesides, G. M. J. Am. Chem. Soc. 1993, 115, Woodruff, D. P.; Cowie, B. C. C. Surf. Sci. 1997, 392, 143. 10714. (26) Nara, J.; Kino, J.; Kobayashi, N.; Tsukada, M.; Ohno, T. Thin (7) Poirier, G. E. Chem. Rev. 1997, 97, 1117. Solid Films 2003, 438, 221. (8) Schreiber, F. Prog. Surf. Sci. 2000, 65, 151. (27) Vondrak, T.; Cramer, C. J.; Zhu, X. Y. J. Phys. Chem. B 1999, (9) Schwartz, D. K. Annu. Rev. Phys. Chem. 2001, 52, 107. 103, 8915. (10) Barrena, E.; Palacious-Lidon, E.; Munuera, C.; Torrelles, X.; (28) Madhavan, K.; Muralidharan, S.; Iyer, S. V. Anti-Corros. Methods Ferrer, S.; Jonas, U.; Salmeron, M.; Ocal, C. J. Am. Chem. Soc. 2004, Mater. 1998, 45, 227. 126, 385. (29) Frey, S.; Stadler, V.; Heister, K.; Eck, W.; Zharnikov, M.; Grunze, (11) Poirier, G. E.; Pylant, E. E. Science 1996, 272, 1145. M.; Zeysing, B.; Terfort, A. Langmuir 2001, 17, 2408. (12) Poirier, G.; Tarlov, M. M. Langmuir 1994, 10, 2853. (30) Whelan, C. M.; Barnes, C. J.; Gregoire, C.; Pireaux, J. J. Surf. (13) Azzam, W.; Cyganik, P.; Witte, G.; Buck, M.; Wo¨ll, C. Langmuir Sci. 2000, 454, 67. 2003, 19, 8262. (31) Sawaguchi, T.; Mizutani, F.; Yoshimoto, S.; Taniguchi, I. (14) Schreiber, F.; Eberhardt, A.; Leung, T. Y. B.; Schwartz, P.; Electrochim. Acta 2000, 45, 2861. Wetterer, S. M.; Larvic, D. J.; Berman, L.; Fenter, P.; Eisenberger, P.; (32) Wan, L. J.; Terashima, M.; Noda, H.; Osawa, M. J. Phys. Chem. Scoles, G. Phys. Rev. B 1998, 57, 12476. B 2000, 104, 3563.

in its own UHV chambers. All measurements described here proceeded at 80-85 K, as measured by silicon diodes attached to the STMs (with exception of the data of Figure 4, which was obtained at 15 K). The base pressure of the STM chambers is consistently below 6 × 10-11 Torr (measured with a nitrogen- calibrated hot-cathode ion gauge). The STM tips were made by electrochemical etching of a tungsten wire followed by hot (white glow) annealing in vacuo. The Cu(111) samples (Monocrystals and MaTecK GmbH) are prepared using cycles of standard argon ion sputtering (up to 3 kV) and annealing (600 K). Sample cleanliness is ascertained by STM both at room temperature and after cooling to cryogenic temperatures. The samples were considered clean if no more Figure 1. 1. Ball and stick models of the molecules of this than 1/10 000 of a monolayer of contamination was found on the study (clockwise from the top left corner): Thiophenol (C6SH6), surface (corresponding to one or two contaminants per image p-fluorothiophenol (C6FSH5), p-chlorothiophenol (C6ClSH5), frame). p-bromothiophenol (C6BrSH5), and pentafluorothiophenol (C6F5- Deposition of the thiophenols onto the sample for STM purpose SH). All STM images represent molecules, whose thiol (-SH) proceeded by backfilling of the turbo-pumped chamber to 1 × group is dehydrogenated, so that the S atom is not bound to a 10-9 Torr using a leak valve while the sample was protected hydrogen atom but rather to the substrate. behind the cryostat thermal shields. A mass spectrum was taken to confirm the purity of the reactant. Subsequently, all filaments adsorption22 at low temperatures. Alkanethiols adsorb on of the chamber were turned off, the sample was extracted from higher coordinated sites23-25 on Cu(111). Due to its the cryostat, and it was exposed to the backfilled chamber for up properties, Cu(111) is a well-suited model system for the to 30 s at a time. The short exposure time ensures constant cryogenic temperature of the sample, as ascertained by virtually investigation of the evolution of SAM films and their drift-free imaging immediately after sample exposure. Repeated structural properties with increasing coverage. exposure cycles were used to reach the coverages indicated here, This study focuses on the adsorption of thiophenols as necessary. (benzenthiols) (TP) which are either unsubstituted or carry Where mentioned, annealing of the adsorbate layer involved a halogen (F, Cl, Br) in the para position or in all positions removal of the sample from the cryostat for a prolonged period of the ring (Figure 1). TPs are one of the simplest model (>1 h), which allows it to warm gradually to room temperature. systems for future molectronic building blocks,26 as they During this period, all filaments in the chamber were switched contain the common thiol substrate linker and an aromatic off to avoid sample contamination. moiety, which in an extended form may offer charge TPD Measurements. Sample preparation for TPD experi- transport properties.1-3,27 In addition, thiophenols have ments proceeded similarly except that the sample was held 28 continuously on a temperature-controlled manipulator. The also been proposed as corrosion inhibitors. The adsorp- sample was prepared by several sputtering and annealing cycles 29-35 36,37 tion of thiophenols has been studied on gold, silver, between successive TPD experiments; the impact of previous and nickel.38,39 Investigation of the geometric properties exposures to successive TPD runs will be described elsewhere. of SAMs as a function of varying substituents offers insight Exposure of the sample used a line-of-sight dosing setup for into how adsorbate-adsorbate interactions determine the enhanced purity of the adsorbates. The TPD experiments utilize SAM formation process. Here we identify quadrupolar a Stanford Research Systems mass spectrometer operated at a intermolecular interactions as important for the geometry multiplier voltage of 1700 V. Mass scans ranging from 10 amu of incomplete films. to slightly above the respective molecular mass were acquired Theoretical work on SAMs has focused mainly on the at a ca. 10 s interval, while ramping the sample temperature at a constant 10 K/min. interaction of thiols with gold surfaces. Recent results Computational Procedure. Density functional theory cal- 40 from Nara et al. indicate adsorption of thiophenol on culations of the electronic and geometric structure and adsorption Au(111) in a manner that places the sulfur anchor between energy of TP on Cu(111) surface were carried out using the two substrate atoms (i.e., close to the bridge site) but generalized gradient approximation for the exchange-correlation slightly shifted toward the nearest hollow. Work by Ferral functional43 and the plane-wave pseudopotential method.44 et al.41 revealed significant transfer of charge from the Ultrasoft pseudopotentials45 were used for all atoms in the system. substrate to the sulfur anchor. The Cu(111) surface was modeled by a supercell consisting of a four-layer slab and 11 Å of vacuum. We used a 3 × 3 unit cell, Experimental Section which included nine Cu atoms per layer and one TP molecule. The entire supercell thus contains 49 atoms. To ensure inde- STM Measurements. This study uses two home-built cryo- pendence of the result on the number of layers used, we also genic scanning tunneling microscope (STM) systems and a performed calculations with two and three layers, which lead to separate vacuum system used for thermally programmed de- bond angles and lengths around the sulfur atom that vary by sorption (TPD) experiments. Both STMs are of the beetle type,42 less than 2° and 0.05 Å from our four-layer result, respectively. and each is housed inside successive gold-plated thermal shields The cutoff energy for the plane-wave expansion is taken to be 300 eV, and a 3 × 3 × 1 Monkhorst-Pack k-point mesh in the (33) Lin, P. H.; Guyot-Sionnest, P. Langmuir 1999, 15, 6825. Brillouin zone sampling46 is used to obtain converged results for (34) Whelan, C. M.; Barnes, C. J.; Walker, C. G. H.; Brown, N. M. the total energy of the system. During lattice relaxation, atoms D. Surf. Sci. 1999, 425, 195. were allowed to move along all directions. The structure obtained (35) Jin, Q.; Rodriguez, J. A.; Li, C. Z.; Darici, Y.; Tao, N. J. Surf. Sci. 1999, 425, 101. was relaxed until the forces acting on each atom converged to (36) Li, X. W.; Zheng, J. W.; Zhou, Y. G.; Ji, Y.; Zhuang, Y.; Lu, T. better than 0.02 eV/Å. H. Chin. J. Anal. Chem. 2003, 31, 1333. (37) Han, S. W.; Lee, S. J.; Kim, K. Langmuir 2001, 17, 6981. Results (38) Kane, S. M.; Huntley, D. R.; Gland, J. L. J. Phys. Chem. B 2001, 105, 9548. Figure 2a shows TPD traces acquired after exposure of (39) Kane, S. M.; Huntley, D. R.; Gland, J. L. J. Phys. Chem. B 1998, a Cu(111) sample at 110 K to low (grey) and high (black) 102, 10216. (40) Nara, J.; Higai, S.; Morikawa, Y.; Ohno, T. J. Chem. Phys. 2004, (43) Perdew, J. P.; Wang, Y. Phys. Rev. B 1992, 45, 13244. 120, 6705. (44) Payne, M. C.; Teter, M. P.; Allan, D. C.; Arias, T. A.; Joannopoulos, (41) Ferral, A.; Paredes-Olivera, P.; Macagno, V. A.; Patrito, E. M. J. D. Rev. Mod. Phys. 1992, 64, 1045. Surf. Sci. 2003, 525, 85. (45) Vanderbilt, D. Phys. Rev. B 1990, 41, 7892. (42) Besocke, K. Surf. Sci. 1987, 181, 145. (46) Monkhorst, H. J.; Pack, J. D. Phys. Rev. B 1976, 13, 5188. 10930 Langmuir, Vol. 20, No. 25, 2004 Wong et al.

Figure 3. 3. Detailed view (left panel) and top view (right panel) of the calculated geometric structure of TP adsorbed on the Cu(111) surface; a ) 0.59 Å, b ) 1.97 Å, c ) 1.80 Å, R) 25°. Figure 2. 2. TPD traces of X-TP molecules: (a) Desorption of low (grey) and high (black) coverages of F-TP from a Cu(111) In our DFT calculations of TP on Cu(111), we initially surface after deposition at 110 K. The peak at 370 K indicates placed TP on the surface such that (a) the S occupies a desorption from the monolayers in which the sulfur is left behind on the surface. The peak at 190 K reflects molecular desorption 3-fold fcc hollow site, (b) the entire molecule is parallel to of the multilayer, which is strongly reduced at lower coverage. the surface, and (c) the abstracted thiol hydrogen is located (b) Dependence of the molecular desorption (black) and the on the bottom side of the slab. During the structural dissociative desorption of the monolayer (gray) on the sub- optimization, the molecule shifts from this high-symmetry stituent: for TP, F-TP, and Cl-TP, we find desorption of the position and tilts away from the surface. In the optimized corresponding halobenzene at slightly increasing temperatures, structure shown in Figure 3, the sulfur atom takes a for Br-TP, we find halogen abstraction and formation of nonsymmetric position between the fcc and bridge sites. biphenyl. The angle between the molecule and the surface is found to be ≈25°. The C-C and C-H bond lengths are not dosages of F-TP. The desulfurized species (bottom) desorbs changed upon the adsorption, keeping the benzene ring from the surface at 370 K, and the peak size is independent almost as a rigid body. However, the adsorption causes of the initial coverage. Consequently, we regard this an increase in S-C bond length from 1.74 to 1.80 Å. By species as the monolayer, which desorbs by scission of the placement of the hydrogen atom on the bottom side of the - 47 48 C S bond similar to ethanethiol and other alkanethiols. copper slab, we can calculate the energetic difference The top curves show desorption of the intact molecule; between the unperturbed molecule and the copper slab their high-coverage peak at 190 K represents desorption on one hand and the dissociated and adsorbed molecule from the multilayer. This peak varies strongly with initial on the other hand while preventing repulsive interaction exposure. We obtained a series of various initial exposures of the abstracted hydrogen with the thiolate in the 3 × 3 both in the monolayer and the multiplayer regime for TP, unit cell. This way the adsorption energy of TP is found F-TP, Cl-TP, and Br-TP, all of which follow a similar to be 0.67 eV. We attribute the mismatch between the pattern of multilayer desorption at ca. 200 K and dis- calculated adsorption energy and the desorption temper- sociative desorption of the monolayer at ca. 400 K. Figure ature of the monolayer of ca. 400 K to the assumption of 2b shows for each molecule two traces obtained simul- atomic hydrogen on the back of the slab in the former taneously: the black trace indicates molecular desorption case, whereas in reality no unrecombined hydrogen is from the multilayer and the gray one indicates dissociative readily available for the desorption process. We should desorption of the main product of the monolayer (for TP, mention that our results are similar to those recently F-TP, and Cl-TP, the desulfurized species, for Br-TP, we reported for TP adsorbed on Au(111),40 which similarly find abstraction of the Br resulting in biphenyl formation 49 predict a nonsymmetric adsorption position of the S atom along the Ullmann reaction scheme ). STM images of the and a slightly larger tilt (29°) of the aromatic moiety with Cu(111) surface taken after desorption of thiophenol reveal respect to the surface. characteristic patterns that are identical to known 50,51 STM measurements of individual dehydrogenated structures of S/Cu(111). thiophenol molecules at 15 K reveal elongated features With increasing size of the halogen, the desorption shifts that have an indentation on one side and a protrusion on toward higher temperatures (for the monolayer peaks: the opposite side (Figure 4). With increasing size of the TP, 360 K; FTP, 370 K; CTP, 380 K; BTP, 440 K). According halogen, the protrusion becomes more and more prominent 52 to the formula of Redhead and assuming a prefactor of and extended, whereas the indentation becomes less and 13 10 Hz, the multilayer desorption temperatures of 180 less readily resolvable. We regard the indentation as K, 190 K, 220 K, and 240 K correspond to binding energies representing the sulfur atom53 and the protrusion as the of ∼0.52, 0.55, 0.63, and 0.69 eV for TP, F-TP, Cl-TP, and (substituted) phenyl moiety. These measurements agree Br-TP, respectively. with the theoretical results in indicating that the thiophenols adsorb with the phenyl ring almost parallel to the (47) Sardar, S. A.; Syed, J. A.; Ikenaga, E.; Yagi, S.; Sekitani, T.; substrate. Comparison of the position of the indentation Wada, S.; Taniguchi, M.; Tanaka, K. Nucl. Instrum. Methods B 2003, with coadsorbed CO molecules (which occupy low- 199, 240. (48) Lai, Y. H.; Yeh, C. T.; Cheng, S. H.; Liao, P.; Hung, W. H. J. coordinated on-top sites) ascertains that the sulfur atom Phys. Chem. B 2002, 106, 5438. is adsorbed at a higher coordinated site of the substrate. (49) Ullmann, F.; Meyer, G. M.; Loewenthal, O.; Gilli, O. Justus Increasing the temperature gradually to liquid nitrogen Liebigs Ann. Chem. 1904, 331, 38. (50) Ulin-Avila, E.; Pawin, G.; Lin, X.; Wong, K.; Kwon, K.-Y.; Liu, (LN2) temperatures, we find a reduction of the resolution A.; Bartels, L. (to be published). (51) Wahlstro¨m et al. Phys. Rev. B. 2001, 64, 155406. (53) Oxygen, the second period homologue of sulfur, is well-known (52) Redhead, P. A. Vacuum 1962, 12, 203. to appear as an indentation in STM. Halogen-Substituted Thiophenol Molecules on Cu(111) Langmuir, Vol. 20, No. 25, 2004 10931

Figure 4. 4. STM images of individual molecules adsorbed on Cu(111) at 15 K. (a) TP, (b) F-TP, (c) Cl-TP, and (d) Br-TP. Image size ) 16 Å × 15 Å, U )-0.2 V, I ) 32 pA. The black bar indicates the length of the TP footprint of ∼8 Å for Figure 6. 6. (a) STM image of a Cl-TP island on Cu(111) formed comparison. With larger size of the substituent, the apparent after an exposure to 0.015L of Cl-TP (image size ) 128 Å × 128 length of the molecule increases. Å, U )-0.6 V, I ) 30 pA). (b) Complete film of Cl-TP after 0.15 L of exposure (image size ) 64 Å × 64 Å, U )-1.3 V, I ) 27 pA).

Figure 5. 5. (a) STM image of islands on Cu(111) formed after exposure of the surface to 0.008L of Br-TP molecules (image size 134 Å × 112 Å, U )-0.8V, I ) 26 pA). The insert shows Figure 7. 7. (a) Isolated islands of a honeycomb structure a step edge at which Br-TP islands nucleate in an ordered form after exposure of Cu(111) to 0.01 L of F-TP (image size ) fashion (image size ) 34 Å × 30 Å, U )-0.8 V, I ) 26 pA). (b) 144 Å × 105 Å, U )-1.4 V, I ) 27 pA) and coexist on the same STM image of Cu(111) terraces after exposure to 0.05 L of Br- terrace with (b) islands of a (3 × 3) structure (image size ) 116 TP. The terraces are completely covered by a Br-TP film (image Å × 89 Å, U )-1.4 V, I ) 29 pA). (c) Magnified view (image size ) 93 Å × 90 Å, U ) 1.1 V, I ) 32 pA). The insert shows size ) 94 Å × 94 Å, U )-1.3 V, I ) 37 pA) of the honeycomb a magnification of the film structure (image size ) 39 Å × 37 structure of (d) (image size ) 365 Å × 512 Å, U )-1.3 V, I ) Å, U )-1.1 V, I ) 57 pA). 37 pA) which was generated by annealing 0.11 ML of F-TP to 125 K. Further increase of the coverage leads to a complete ) × of the STM image but no change of the shape of the coverage of the substrate (e) (image size 116 Å 116 Å, U )-1.3 V, I ) 39 pA.) Notice the protrusions at the domain adsorbate. Correspondingly, we assume that X-TP mol- boundaries (arrows). ecules attain the same surface orientation at LN2 temperatures as at 15 K. For the remainder of this manuscript, Figure 5b shows Cu(111) terraces after exposures to we will assume that each of the protrusions in the STM 0.05 L of Br-TP. The entire surface is covered with images will represent one X-TP molecule. adsorbates. For the remainder of the manuscript, we will Figure 5a shows several isolated islands formed after refer to such coverages as complete, which shall imply exposure of the sample to 0.008 L of Br-TP. The islands that no unoccupied terrace area can be found anymore, cover only a small fraction of the otherwise clean Cu(111) while not ruling out that further compression of the film terraces, and they nucleate both adjacent to step edges can create more dense structures. By FT of the STM image 3 0 (inset) and on the terraces. The islands exhibit a variety (Table 1), we identify this structure as [-1 3], which of simple lattice structures, which we analyze by Fourier corresponds to an adsorbate density of 0.11 ML. For all Transformation (FT) of the corresponding portion of the film patterns described in this manuscript, we ascertained STM images (Table 1). This results in a pattern that can the assignment of the film structure by determining the be interpreted similarly to low-energy electron diffraction surface coverage independently. To this end, we calibrate (LEED). The advantage of this approach over measure- the piezo response (which depends on the precise piezo ments in real space lies in the precise measurement of the temperature) using the well-known Cu(111) step height. lattice vectors. In contrast to LEED, the FT of STM images By counting several thousand molecules (manually), we can be chosen to originate exclusively from one island of then determined the actual coverage at high fidelity. All a specific orientation and symmetry, so that rotational measured coverages fall well within the statistical (n1/2) domains, etc. do not affect the pattern. Further, the error of the densities associated with the patterns derived availability of a real space image in STM offers direct by FT; the match between density measurements and the information about the number of molecules in each unit theoretical value associated with the inferred surface cell. Also, electron damage, as common with conventional pattern is typically better than 1% of the actual coverages (non channel-plate) LEED systems, is avoided. The most- or better than 1/1000 of a complete ML. × 3 -2 common pattern of Br-TP islands are (3 4) and [0 4] Figure 6a shows a Cu(111) surface after exposure to (Table 1), both of which correspond to a coverage of 0.083 0.015 L of Cl-TP at 81 K. Similar to Br-TP, Cl-TP arranges monolayers (ML). On the basis of the determination of itself in islands of distinctive periodicity. Its island the molecular adpositions at low temperatures, these structure is more complicated than that of Br-TP: the lattice vectors correspond to a film geometry as that shown dominating motif involves rows of pairs of molecules. We 7 -1 in Table 1. The primitive vectors of the mesh are also identify it as a [-1 3 ] pattern, which is composed of two indicated in the FT patterns in Table 1. Cl-TP molecules adsorbed in two different (hcp vs fcc) 10932 Langmuir, Vol. 20, No. 25, 2004 Wong et al.

Table 1. Overview of Experimental Results

hollow sites. This arrangement corresponds to an adsor- honeycomb pattern (Figure 7d, enlargement in 7c). This bate density of 0.10 ML. suggests that the honeycomb structure is energetically It requires ∼0.15 L exposure to cover the substrate with more favorable than the simple 3 × 3 pattern. a complete Cl-TP film. The most-common film motif of If the density of F-TP molecules on the surface is this film is shown in Figure 6b. It corresponds to a increased beyond the coverage associated with the hon- 3 -1 [-1 3 ] structure with an adsorbate density of 0.125 ML. eycomb pattern (0.109ML), a denser film structure is Similar to Br-TP, all pattern assignments used FT of observed (Figure 7e). The dominant motif of this film is corresponding sections of STM images (Table 1). 2 -1 a [1 3 ] pattern (Table 1) which corresponds to a coverage In contrast to Br-TP and Cl-TP, which form relatively of 0.14 ML. simple structures at both low and high coverage, F-TP can form a low-coverage structure of significant complex- Figure 8a shows (unsubstituted) TP molecules on Cu- ity: Figure 7a,b shows two strikingly different types of (111) after adsorption at 81 K. The surface is scattered F-TP islands that form on the same terrace in close vicinity with small clusters of molecules, which coalesce if the after exposure of a Cu(111) substrate to 0.01 L of F-TP at coverage is high enough, but do not aggregate to a 81 K. The simple structure of Figure 7b is a 3 × 3 pattern continuous ordered island. Annealing of the surface to (Table 1), which corresponds to an adsorbate surface various temperatures up to room temperature does not density of 0.111 ML. Figure 7a shows a honeycomb-like lead to the formation of an ordered film. This suggests structure involving seven molecules per unit cell, which that the lack of order is not due to insufficient mobility can be identified as an (8 × 8)R19° (Table 1) pattern with of the adsorbates, i.e., kinetically determined, but rather a density of 0.109 ML. Upon gradual increase of the due to the lack of intermolecular interaction that sustains coverage, the islands develop a distinct propensity toward an ordering process. The absence of extended ordered the honeycomb structure. Deposition of ∼0.11 ML of F-TP regions persists at increasing coverage up to the point on Cu(111) at 81 K and slight annealing to ∼120 K results where the film appears as complete. Figure 8b shows an in complete conversion of the 3 × 3 structure to the apparently complete layer of TP molecules with a density Halogen-Substituted Thiophenol Molecules on Cu(111) Langmuir, Vol. 20, No. 25, 2004 10933

Figure 10. 10. Dependence of the footprint of a molecule on Figure 8. 8. Thiophenol molecules form disordered structures its size as represented by the intermolecular distance between on Cu(111) both at low coverages (a) (image size ) 40 Å × 70 the halogen and the sulfur atom assuming standard bond length. Å, U )-3V,I ) 25 pA) and at high coverages (b) (image size The superlinear dependence may suggest an impact of increased ) 128 Å × 128 Å, U )-3.3 V, I ) 12 pA), as confirmed by FT width of the molecule. Different symbols denote different (c) of panel b. coverages (triangles/squares ) incomplete; circles ) complete). A compression of approximately 1/4 is found before the molecules start to tilt upright.

determines the film pattern. However, we find that even for closely related values of the coverage, there is more than one surface pattern available (e.g., Br-TP forms two different structures at a density of 1/12 ML). The molecular densities of the disordered films of TP and 5F-TP also scale with the molecular size, yet the lack of order permits only smaller absolute densities than found for para-substituted molecules. This is consistent with the behavior of hard-sphere packing, where random arrangement in a container will lead to a lower density Figure 9. 9. PentaF-TP molecules form disordered patterns than ordered hcp or fcc packing.54 ) × on Cu(111) both at low coverages (a) (image size 64 Å 128 The higher molecular density of the complete films at Å, U )-1.3 V, I ) 220 pA) and at high coverages (b) (image size ) 256 Å × 256 Å, U )-1.3 V, I ) 220 pA), as confirmed no significant change to the molecular shape as observed by FT (c) of panel b. in STM suggests that in order to accommodate a maximum number of molecules on the terrace in an orientation of 0.103 ML. The FT (Figure 8c) confirms the lack of lateral almost parallel to the substrate, the film is exposed to periodicity of the molecular arrangement on the surface. compressive strain. The difference in coverage between Figure 9 shows different coverages of 5F-TP on Cu(111) the low- and the high-coverage patterns for each of the substituted molecules directly indicates a compression after exposure at 81 K. Similar to TP, 5F-TP forms ordered ∼ structures neither at incomplete nor at complete coverages. by 25% for all films investigated. At low coverage, 5F-TP appears more mobile than the At domain boundaries of complete coverages, we find previously discussed molecules; diffusing molecules give individual protrusions embedded in the film (see Figure rise to the streaky noise on the bottom of the STM image 7e). We regard these protrusions as molecules that are of Figure 9a. At high coverage, the molecules form a forced in an upright position by compressive strain in the complete layer with a density of 0.079 ML. The FT confirms film. Further increase of the coverage leads to surfaces the random orientation of molecules on the surfaces. which can only be imaged reliably after annealing to room temperature. Figure 11 shows a complete Cl-TP film Discussion (green) with the periodicity described above. In the vicinity of step edges and in well-defined islands, we find areas The ability of self-organization present for para- that appear elevated (inset). Here the film corrugation is substituted TP molecules but absent for species that carry reduced and the molecules appear to be closer to each identical atoms at all but the thiol site of the ring (TP and other; however, no reproducible film structure could be 5F-TP) indicates that the identity and the site of substi- discerned. We interpret these areas as being covered by - tution have a strong impact on the adsorbate adsorbate molecules whose phenyl moiety is tilted substantially interaction. In the case of Br-, Cl-, and F-TP, the molecules toward the surface normal. self-organize into distinctively different patterns depend- In addition to the molecular size, intermolecular ing on the nature of the halogen substituent. To be able interactions may determine the film pattern: here we to understand this difference, it is important to establish wish to focus on the electrostatic interactions between what properties of the molecules are changed when neighboring molecules, which may be understood in terms different halogens are introduced. of the polarization of the aryl-halogen bond. In first - Table 1 indicates the intramolecular sulfur halogen approximation, the polar nature of the Aryl-X bond is distance based on textbook bond lengths (which neither determined by the electronegativity (EN) difference reflect intramolecular interactions specific to thiophenols between the halogen and the supporting carbon atom. nor any effects of the substrate). Going from F via Cl to Due to the large EN of the halogens (Br, 2.8; Cl, 3; F, 4), - Br, the X S length of the molecule increases by almost the negative side of the bond dipole is at the halogen atom. 10%. Figure 10 shows the corresponding increase of area Calculations by Ferral et al.41 indicate that, in thiol occupied by each of the molecules at different coverages. linkages to copper, the sulfur atom becomes partially The densities scale with the molecular size at much smaller negatively charged by donation from the substrate. For increments than suggested by the surface lattice constant para-halosubstituted thiophenols, this results in ac- of Cu(111) of 2.55 Å. This requires surface unit meshes of different symmetry/aspect ratio for different thiophe- (54) Donev, A.; Cisse, I.; Sachs, D.; Variano, E. A.; Stillinger, F. H.; nols. Consequently, the size of the molecule partially Connelly, R.; Torquato, S.; Chaikin, P. M. Science 2004, 303, 990. 10934 Langmuir, Vol. 20, No. 25, 2004 Wong et al.

Figure 13. 13. At the edge of (8 × 8)R19° FTP islands, the center positions of the adsorbate hexagons frequently remain vacant. The figure shows a superposition of an atomic model of the superstructure and an STM image acquired at U ) 1.7 Figure 11. 11. Increasing the Cl-TP coverage on Cu(111) ) 3 -1 V, I 32 pA. The hexagons indicate adsorbate unit cells with beyond the [-1 3 ] pattern at 0.125 ML (green) leads to the occupied (red) and vacant (blue) center positions. formation of islands (red) on the terraces and at the step edges, which we interpret as molecules whose aromatic moiety is tilted away from the horizontal (image size ) 150 Å × 150 Å, U ) S-X axes of the six circumferential molecules are angled -1.5 V, I ) 32 pA). The insert shows a linescan along the white with respected to a radial line. In this orientation, the F line starting at a lower terrace (blue) and crossing an island apex of the S-X axis of each of the molecules approaches of protruding molecules. the neighboring molecule mesial, i.e., toward the center of the benzene ring. This is consistent with the assumption of a strong molecular quadrupole moment, which predicts partial positive charge toward the meta and ortho positions of the ring (inset in Figure 12). The seventh, center molecule is geometrically prevented from partaking in this quadrupolar interaction. Correspondingly, we find that the center position frequently remains unoccupied, especially at island boundaries, as shown in Figure 13. The patterns of Cl-TP and Br-TP are not quite as Figure 12. 12. Comparison between the number of molecules indicative of quadrupolar interaction: these patterns in the low-coverage adsorbate unit cell (as an indicator of the consist of rows of molecules in which the row direction film complexity) and the difference in electronegativity between - the atoms at the meta and para positions (as an indicator of the forms an angle with the S X axes of the molecules. We intramolecular quadrupole moment). The good correlation note that, with decreasing substituent size and closer suggests that quadrupolar interactions affect the film pattern. spacing of the rows, the sequence of molecules in neigh- cumulation of negative charge both on the side of the boring rows becomes offset, so that the S or X group of a halogen and the on the side of sulfur atom. Correspond- molecule of one row falls closer to the benzene ring of a ingly, we expect the dipole moment of the para-substituted molecule in the neighboring row, i.e., at higher ∆EN; the species to lie between that of thiophenol, where no halogen negatively charged groups of molecules of one row are pulls charge to the para site of the aryl moiety, and better in phase with the positively charged benzene ring pentafluoro-thiophenol, where five fluorine atoms pull of the neighboring row. charge toward the ring and away from the thiol group. In an earlier publication, Wong et al.55 confirmed these expectations by density functional calculations using Summary 56 methods developed by Kong et al. Given that neither We have shown that (halo-substituted) thiophenols thiophenol nor pentafluoro-thiophenol form ordered films, adsorb inclined on Cu(111). They possess sufficient we look beyond the dipolar term in search of a molecular property that correlates with the pattern formation. mobility at 81 K to form ordered adislands/layers, whose The difference between the EN of the ortho/meta and patterns intricately depend on the substituent-dependent the para substituent may be regarded as an indicator of molecular size and quadrupole moment. Increase of the the molecular quadrupole moment. In Figure 12, we plot coverage leads to compressed films of horizontally ad- this ∆EN as a function of the substituent (solid columns). sorbed molecules; upon further increase of the coverage, If we regard the number of molecules per superstructure islands of vertically adsorbed molecules are generated. unit cell as an indicator of the film’s complexity (open columns), we find a remarkable correlation between these Acknowledgment. We wish to acknowledge crucial indicators. In the following paragraph, we wish to validate support by a Career grant (CHE 0132996) from the the importance of the molecular quadrupole moment for National Science Foundation (NSF) and by joined Basic the film geometry by closely analyzing the pattern found at maximal ∆EN. Energy Science grants (DE-FGO2-03ER15464/03ER- The F-TP (8 × 8)R19° structure consists of six-membered 154645) from the Department of Energy. The equipment rings of molecules, which enclose a seventh molecule. The was provided by an NSF Major Research Instrumentation grant (DMR 0116339) and by DOD/DARPA/DMEA under (55) Wong, K.; Kwon, K. Y.; Rao, B. V.; Liu, A.; Bartels, L. J. Am. Award No. DMEA90-02-2-0216. Chem. Soc. 2004, 126, 7762. (56) Kong, J. et al. J. Comput. Chem. 2000, 21, 1532. LA048208B