Competitive Self-Assembly of Symmetrical, Difunctional Molecules on

5396

Langmuir 2005, 21, 5396-5404

Competitive Self-Assembly of Symmetrical, Difunctional Molecules on Ambient Copper Surfaces Lawrence Pranger,*,† Alex Goldstein,‡ and Rina Tannenbaum† School of Materials Science and Engineering, Georgia Institute of Technology, Atlanta, GA 30332, and Manufacturing Research Center (MARC) and Center for Board Assembly Research (CBAR), Georgia Institute of Technology, Atlanta, GA 30332 Received February 25, 2005. In Final Form: April 5, 2005 This paper describes competitive self-assembly from solutions of symmetric R,ω-difunctional molecules on Cu substrates briefly exposed (less than 5 min) to ambient conditions. XPS and PM-IRRAS were utilized as complimentary surface analytical techniques to characterize the resulting organized organic thin films (OOTFs) on these “ambient” Cu surfaces. The order of preferential adsorption was observed to be diisocyanide ≈ dithiol > dicarboxylic acid > dinitrile > diisothiocyanate, indicating that the isocyanide (-NC), and thiol (-SH) functions provide the strongest adhesion to ambient Cu. 1,4-Phenylene diisocyanide and 1,4-terephthalic acid were both observed to adopt a standing-up phase configuration, in which the difunctional molecules bond to the base substrate through only one terminal functional group, with the other terminal group disposed away from the substrate. This indicates the ability to utilize OOTFs to produce “sticky surfaces” on ambient Cu. All other molecules bonded to the substrate through both terminal groups, in either surface-parallel or arched “hairpin” configurations. On the basis of these findings, aromatic diisocyanides and diacids are the most suitable molecules for creating OOTFs with high packing density. Such films can be utilized as protective coatings in the assembly of printed circuit boards, where Cu is becoming an increasingly important substrate for interconnects. Moreover, the ability to create chemically sticky surfaces on ambient Cu substrates indicates exciting potential for the development of a new surfacemount technology operative at the nanometer scale.

Introduction Organized organic thin films (OOTFs) have been successfully prepared on a variety of substrates to study and to optimize surface properties such as wetting, friction, chemical resistance, and biocompatibility.1 Such films are spontaneously formed when amphiphilic molecules, e.g., short-chain alkane thiols or fatty acids, are allowed to self-assemble into a monolayer or multilayer structure on a substrate.2,3 When difunctional molecules bearing reactive functional groups at each end are selected in lieu of amphiphiles, “chemically sticky” OOTFs can be prepared.4 Such films provide a means of bonding adlayers to a target substrate, e.g., layers of metallic nanoclusters to metal or semiconductor surfaces.5-7 However, the self-assembly of difunctional molecules is complicated by the fact that both terminal functional groups are reactive and may have comparable affinity for the target substrate. To produce a chemically sticky OOTF, the constituent difunctional molecules must bond to the substrate through only one terminal and collectively form a chemically reactive surface with the terminal group disposed away from the * Author to whom correspondence should be addressed. E-mail: [email protected]. † School of Materials Science and Engineering, Georgia Institute of Technology. ‡ MARC and CBAR, Georgia Institute of Technology. (1) Schreiber, F. Prog. Surf. Sci. 2000, 65, 151-257. (2) Duwez, A. J. Electron Spectrosc. 2004, 134, 97-138. (3) Ulman, A. Chem. Rev. 1996, 96, 1533-1554. (4) Henderson, J.; Feng, S.; Ferrence, G.; Bein, T.; Kubiak, C. Inorg. Chim. Acta 1996, 242, 115-124. (5) Chen, S. J. Phys. Chem. B 2000, 104, 663-667. (6) Horswell, S.; O’Neill, I.; Schiffrin, D. J. Phys. Chem. B 2001, 100, 941-947. (7) Janes, D.; Batistuta, M.; Datta, S.; Melloch, M.; Andres, R.; Liu, J.; Chen, N.; Lee, T.; Reifenberger, R.; Chen, E.; Woodall, J. Superlattice. Microst. 2000, 27 (5/6), 555-563.

substrate.8-11 Kinetic trapping in a surface-parallel “lyingdown” phase may occur in the early stages of self-assembly, thus preventing the molecules from achieving the required surface-normal bonding configuration.1,12 The fate of a given molecule upon will depend on the relative magnitude of chemisorption (headgroup-substrate bonding) and intermolecular interactions, including solvent-solute interactions.1,3,10 Of the various metallic substrates selected for selfassembly studies, Au has been by far the most popular choice,2 while comparatively few studies have described the self-assembly on Cu. The tendency of Cu to rapidly form a layer of native oxide was early implicated as an impediment to forming high-quality, reproducible OOTFs on Cu.13 However, it has been shown that by minimizing exposure of freshly evaporated Cu surfaces to air or by following reasonable pretreatment procedures, such as polishing and acid cleaning, OOTFs can be produced on ambient Cu surfaces of sufficient quality to retard further oxidation.14-17 In this paper, the term “ambient Cu” is (8) Tour, J.; Jones, L.; Pearson, D.; Lambda, J.; Burgin, T.; Whitesides, G.; Allara, D.; Parikh, A.; Atre, S. J. Am. Chem. Soc. 1995, 117, 95299534. (9) Reed, M. P IEEE 1999, 87, 652-658. (10) Tour, J. Acc. Chem. Res. 2000, 33, 791-804. (11) Tour, J.; Rawlett, A.; Kozaki, M.; Yao, Y.; Jagessar, R.; Dirk, S.; Price, D.; Reed, M.; Zhou, C.; Chen, J.; Wang, W.; Campbell, I. Chem. Eur. J. 2001, 7, 5118-5134. (12) Henderson, J.; Feng, S.; Bein, T.; Kubiak, C. Langmuir 2000, 16, 6183-6187. (13) Laibinis, P.; Whitesides, G.; Allara, D.; Tao, Y.; Parikh, A.; Nuzzo, R. J. Am. Chem. Soc. 1991, 113, 7152-7169. (14) Laibinis, P.; Whitesides, G. J. Am. Chem. Soc. 1992, 114, 90229028. (15) Laffineur, F.; Delhalle, J.; Guittard, S.; Geribaldi, S.; Mekhalif, Z. Colloids Surf., A 2002, 198-200, 817-827. (16) Whelan, C.; Kinsella, M.; Carbonell, L.; Ho, H.; Maex, K. Microelectron. Eng. 2003, 70, 551-557. (17) Timmons, C. O.; Zisman, W. A. J. Phys. Chem. 1965, 69, 984990.

10.1021/la050508l CCC: $30.25 © 2005 American Chemical Society Published on Web 05/04/2005

Self-Assembly of Symmetrical, Difunctional Molecules

Figure 1. Idealization of a smart molecular solder.

used to denote a freshly evaporated Cu surface, which has undergone brief exposure (e5 min) to air prior to immersion in adsorption solution, as opposed to a clean Cu surface maintained under UHV conditions. The ability to produce OOTFs on ambient Cu surfaces is significant in view of increasing importance of Cu in the microelectronics industry in the assembly of printed circuit boards (PCBs).16,18 The lower bulk resistivity and better electromigration properties of Cu make it an appealing alternative to Au and Al for wafer metallization and wire bonding applications. To prevent problems caused by oxidation of bare Cu pads, surface treatments are necessary, such as etching, followed by application of a thin protective coating.19 Coatings are also applied to promote adhesion of photoresist in Cu patterning.18 Self-assembled OOTFs are well suited for these applications. Moreover, chemically sticky OOTFs of difunctional molecules on Cu may offer a surface-mount technology solution suited for the continuing miniaturization of integrated circuits into the nanoscale regime. State-of-the-art board printing technology relies on the ability to solder components onto their designated pads using a viscous “solder brick” of variable volume and shape, which limits the accuracy of component placement.20-22 In contrast, the film dimensions of an OOTF are well defined and reproducible, determined only by the length and stereoregular tilt of the constituent molecules and the area of the substrate pad.1,23,24 If chemically sticky OOTFs can be produced on ambient Cu surfaces, this may provide the means of supporting the ultrafine pitch requirements anticipated for nanosized electronics components.20,22,25-29 A “smart molecular solder” capable of bonding the nanosized components to corresponding pads is idealized in Figure 1. Significantly, OOTFs possessing a π-conjugated backbone structure, i.e., “molecular wires”, are capable of providing electrical connectivity in addition to mechanical bonding between component and pad.7,30 In view of these (18) Rickerby, J.; Steinke, J. H. G. Chem. Rev. 2002, 102, 15251549. (19) Mekhalif, Z.; Sinapi, F.; Laffineur, F.; Delhalle, J. J. Electron Spectrosc. 2001, 121, 149-161. (20) Zou, L.; Dusek, M.; Wickham, M.; Hunt, C. Soldering Surf. Mount Technol. 2003, 15, 43-49. (21) Warwick, M.; Harpley, I. Soldering Surf. Mount Technol. 1997, 9, 29-32. (22) Yang, T.; Tsai, T. J. Intell. Manuf. 2004, 15, 711-721. (23) Whitesides, G. Sci. Am. 1995, 273, 146-149. (24) Tao, Y.; Lee, M.; Chang, S. J. Am. Chem. Soc. 1993, 115, 95479555. (25) Hendriksen, M.; Frimpong, F.; Ekere, N. Microelectron. Int. 1999, 16, 49-54. (26) Rooney, D.; Castello, N.; Cibulsky, M.; Abbott, D.; Xie, D. Microelectron. Int. 2003, 20, 34-42. (27) Jagt, J. IEEE Trans. Compon., Packag. Technol. 1998, 21, 215225. (28) Gagne, D.; Quaglia, M.; Shina, S. Solid Surf. Mount Technol. 1996, 8, 9-11. (29) McGrath, P.; Soutar, A. Circuit World 1997, 23, 42-45. (30) Seminario, J.; Zacarias, A.; Tour, J. J. Am. Chem. Soc. 1999, 121, 411-416.

Langmuir, Vol. 21, No. 12, 2005 5397

potential applications, further study of self-assembly on ambient Cu is well motivated. This paper describes a systematic study of competitive self-assembly of symmetric R,ω-difunctional molecules on ambient Cu. Cu substrates utilized in the fabrication of PCBs are likely to undergo brief exposure to air between etching and immersion in adsorption solution16 or application of adhesion promoter.18 In consideration of this, freshly evaporated Cu surfaces were immersed in adsorption solution immediately after removal from the evaporation chamber, to avoid surface contamination and to minimize oxidation of Cu to Cu2O, but without attempting to completely eliminate ambient exposure.24,32-34 The main objectives were to determine which functional groups provide the strongest adhesion on ambient Cu, and second, to investigate the feasibility of producing chemically sticky OOTFs on ambient Cu. For each two-component adsorption system, π-conjugated molecules with identical phenylene backbones were chosen to minimize differences in intermolecular interactions and solvent-solute interactions and highlight chemisorption of the terminal group on the metal substrate as the dominant factor determining the outcome of preferential adsorption. These molecules are also interesting, as they represent model molecular wires for the preparation of conductive OOTFs, though it will require additional optimization of the electronic interactions between the terminal group and substrate to maximize conductivity.35-39 Sufficient adsorption time was allowed to ensure complete OOTF formation. Under these conditions, the molecules with the functional groups forming the strongest bonds with the substrate should exhibit preferential adsorption.40-43 Strong surfaceadsorbate interactions may arise from aromatic bonding due to the π-conjugated backbone structure of molecular wires.31 However, chemically sticky OOTFs cannot be formed from difunctional molecules, which remain in a surface-parallel orientation after adsorption. Therefore, it was also investigated which molecules adopt the required vertical bonding configuration. The functional groups investigated in this work were thiol (-SH), carboxylic acid (-COOH), nitrile (-CN), isocyanide (-NC), and isothiocyanate (-NCS). Of these, the thiol group is by far the most frequently employed in self-assembly studies.3,44 However, difunctional molecules with all the above functional groups are known to chemisorb onto various metallic substrates, with the exception of the isothiocyanate group.1,4,6,32,45-47 Phenyl (31) Tour, J. Chem. Rev. 1996, 96, 537-553. (32) Steiner, U.; Caseri, W.; Suter, U. Langmuir 1992, 8, 27712777. (33) Kudelski, A. Langmuir 2003, 19, 3805-3813. (34) Ron, H.; Cohen, H.; Matlis, S.; Rappaport, M.; Rubenstein, I. J. Phys. Chem. B, 1998, 102, 9861-9869. (35) Tian, W.; Datta, S.; Hong, S.; Reifenberger, R.; Henderson, J.; Kubiak, C. J. Chem. Phys. 1998, 109, 2874-2882. (36) Chen, J.; Calvet, L.; Reed, M.; Carr, D.; Grubisha, D.; Bennett; D. Chem. Phys. Lett. 1999, 313, 741-748. (37) Zhou, C.; Deshpande, R.; Reed, M.; Jones, L.; Tour, J. Appl. Phys. Lett. 1997, 71, 611-613. (38) Zareie, M.; Ma, H.; Reed, B.; Jen, A.; Sarikaya, M. Nano. Lett. 2003, 3, 139-142. (39) Hong, S.; Reifenberger, R.; Tian, W.; Datta, S.; Henderson, J.; Kubiak, C. Superlattices Microstuct. 2000, 28, 289-303. (40) Chang, S.; Chao, I.; Tao, Y. J. Am. Chem. Soc. 1994, 116, 67926805. (41) Tao, Y. J. Am. Chem. Soc. 1993, 115, 4350-4358. (42) Bain, C.; Evall, J.; Whitesides, G. J. Am. Chem. Soc. 1989, 111, 7155-7164. (43) Bain, C.; Whitesides, G. J. Am. Chem. Soc. 1989, 111, 71647175. (44) Bain, C.; Troughton, E.; Tao, Y.; Evall; J.; Whitesides, G.; Nuzzo, R. J. Am. Chem. Soc. 1989, 111, 321-335. (45) Allara, D.; Atre, S.; Elliger, C.; Snyder, R. J. Am. Chem. Soc. 1991, 113, 1852-1854.

5398

Langmuir, Vol. 21, No. 12, 2005

Pranger et al.

isothiocyanate is known to adsorb on Cu, where it has sufficient adsorption strength to act as a corrosion inhibitor.48 However, the mode of adsorption (chemisorption versus physisorption) is not clear. OOTFs were characterized by XPS (X-ray photoelectron spectroscopy) and PM-IRRAS (polarization modulation infrared reflection absorption spectroscopy). XPS analysis was performed for qualitative chemical analysis. PMIRRAS was useful in confirming the qualitative analysis of XPS. However, its main utility was in helping to identify OOTFs in which the molecules adopt a standing-up phase configuration.49 The final outcome of the work consisted of obtaining a ranking of the relative affinities of the series of functional groups used for adsorption on ambient Cu substrates. Experimental Section Materials. All chemicals and solvents were purchased from Sigma Aldrich or Alfa Aesar and used as received. Chemical purity was g98%, with the exception of 1,6-hexanedithiol (g97%) and 1,4-benzenedithiol (g95%). Metal coatings were made by e-beam evaporation of >99.99% pure metal targets of Cu (Kurt J. Lesker) onto fresh double-sided, polished 2 in. (100) silicon wafers from Virginia Semiconductor or Montco Silicon. Sample Preparation. All adsorption solutions were prepared in new glassware, which was first cleaned by rinsing in deionized water, acetone, and pure adsorption solvent. Each solution contained two different R,ω-difunctional molecules, which were usually symmetric, at equimolar concentrations in the range of 1.0-5.0 mM. Tetrahydrofuran (THF) was used, without degassing, as the solvent in all adsorption experiments, unless otherwise specified. Cu films of 150-250 nm thickness were produced by e-beam evaporation (CVC Products) at 3 × 10-6 Torr or less. Cu films were deposited onto adhesion layers of 10-50 nm Ti or Cr. Deposition rates were e6 Å/s to minimize heating of the substrates during deposition, which may otherwise cause high residual stress on the film and diffusion of Si or adhesion layer metals to the film surface.50,51 After each metal deposition, the deposition chamber was back-filled with nitrogen. Cu films produced under similar conditions have been shown to have a predominantly (111) crystalline orientation.14 Freshly prepared metal substrates were carefully removed from the tray holder of the e-beam evaporator and transferred directly in a class 10 cleanroom environment to their adsorption solutions. Exposure to ambient was limited to ∼5 min prior to immersion. After the specified immersion time, substrates were removed from adsorption solution and rinsed with disposable pipets in pure solvent to remove all loosely bound molecules. Samples emerged autophobic from all solutions. Spectroscopic Characterization. XPS spectra were collected partly on a Surface Science Instruments SSX-100 and partly on a Kratos Axis Ultra, located at the Center for Microanalysis of Materials at the University of Illinois. Both instruments used a monochromatized Al X-ray source, a base pressure e3 × 10-8 Torr, and charge neutralization. Each sample was focused individually. No differential charging was observed on any of the samples analyzed. For peak fitting, software from a commercial vendor (XPS International, Inc.) was used. The maximum allowed fwhm for any peak was normally constrained to 1.6 eV. Peak-fitting solutions were sought for χ2 < 2, Shirley background subtraction, and allowing up to 5% peak asymmetry and 20% Lorenzian contribution (80% Gaussian) as required to (46) Han, H.; Han, S.; Joo, S.; Kim, K. Langmuir 1999, 15, 68686874. (47) Solomun, T.; Christmann, K.; Baumga¨rtel, H. J. Phys. Chem. 1989, 93, 7199-7208. (48) Singh, M.; Rastogi, R.; Upadhyay, B.; Yadav, M. Mater. Chem. Phys. 2003, 80, 283-293. (49) Ulman, A. An Introduction to Ultrathin Films: From LangmuirBlodgett to Self-Assembly; Academic Press: San Diego, 1991. (50) Branger, V.; Pelosin, V.; Badawi, K.; Goudeau, P. Thin Solid Films 1996, 275, 22-24. (51) Kondo, I.; Yoneyama, T.; Takenaka, O.; Kinbara, A. J. Vac. Sci. Technol. A 1992, 10, 3456-3459.

Figure 2. XPS analysis after competitive adsorption between terephthalonitrile and 4-cyanophenylisothiocyanate. optimize the peak fit on the low-binding-energy side. All peaks were internally referenced to carbon as aromatic hydrocarbon at 284.8 eV. PM-IRRAS spectra were collected on a Nexus 870 (Thermo Electron) with a PM-IRRAS attachment including a photoelectric modulator (Hinds) and a nitrogen-cooled MCT detector at near grazing angle. An average of 1000 scans per analysis was collected, with a resolution of 2-4 cm-1. Reflectance spectra are expressed ∆I/I ) 100 × (Is - Ip)/(Is + Ip), converted to arbitrary units. Ip and Is designate the spectra collected parallel and perpendicular to the plane of incidence, as usual.52

Results and Discussion Nitrile versus Isothiocyanate. Figure 2 shows the XPS survey spectrum after competitive adsorption between terephthalonitrile (NC-φ-CN) and 4-cyanophenyl isothiocyanate (NC-φ-NCS) at a concentration of 1 mM for 24 h. The presence of carbon and nitrogen peaks at ∼285 and 400 eV, respectively, are consistent with adsorption of terephthalonitrile. No sulfur peaks were detected in the XPS spectrum, which demonstrates that terephthalonitrile preferentially adsorbs over 4-cyanophenyl isothiocyanate. This indicates that the nitrile group has a higher affinity for ambient Cu than the isothiocyanate group. Hence, the order of preferential adsorption is -CN > -NCS. As expected, a small oxygen peak was observed at 531 eV, consistent with a minor degree of oxidation of Cu to Cu2O.32,34 The adsorption characteristics of nitrile-substituted benzenes varies depending on the metal substrate. For example, benzonitrile adsorbs on Cu(111)53 and Au(111)47 with both nitrile and ring plane axes parallel to the surface, while on Ni(111) and Rh(111), the conjugation between the nitrile group and phenyl ring is broken to accommodate η2 (N,C) bonding through the nitrile group alone.47,54 In the present case, the PM-IRRAS spectrum (not shown), revealed no peaks which could be assigned to the nitrile group or to the phenyl backbone. Adsorption in either the η1 (N) configuration, as expected for metal-nitrile compounds in bulk form,46 or the η2 (N,C) configuration, as proposed for some alkane dinitriles on Cu,32 would result in the molecule having a free nitrile group protruding away from the surface, with an IR-visible vibration somewhere in the vicinity of the gas-phase frequency of 2270 cm-1.47 The lack of peak intensity in this region demonstrates that both nitrile groups, and hence the phenyl ring, are adsorbed flat and parallel to the surface. (52) Golden, W.; Saperstein, D. J. Electron. Spectrosc. 1983, 30, 4350. (53) Kordesch, M.; Feng, W.; Stenzel, W.; Weaver, M.; Conrad, H. J. Electron. Spectrosc. 1987, 44, 149-162. (54) Murphy, K.; Azad, S.; Bennett, D.; Tysoe, W. Surf. Sci. 2000, 467, 1-9.


Figure 3. XPS analysis after competitive adsorption between terephthalic acid and terephthalonitrile. The absence of nitrogen shows that no terephthalonitrile has adsorbed.

Figure 4. Carbon peak after competitive adsorption between terephthalic acid and terephthalonitrile.

Carboxylic Acid versus Nitrile. Figure 3 shows the XPS spectrum after competitive adsorption between terephthalic acid (HOOC-φ-COOH) and terephthalonitrile at a concentration of 1 mM each, for 18 h. Carbon and oxygen peaks were detected at ∼285 and 532 eV, respectively, while no nitrogen peak was detected in the XPS spectrum. This is consistent with preferential adsorption of terephthalic acid over terephthalonitrile. The high-resolution carbon spectrum confirms the assignment of the carbon peak to carboxylic acid. In the peak fit for C 1s, shown in Figure 4, there are three peaks at 284.8, 286.0, and 288.6 eV, corresponding to aromatic carbon, carbon adjacent to either carboxylic acid group, and carbon in carboxylic acids groups, respectively.55 This indicates that the carboxylic acid group has a higher affinity for ambient Cu than the nitrile group. Since the nitrile group has a stronger affinity for ambient Cu than isothiocyanate, terephthalic acid should also preferentially adsorb over isothiocyanate, and this was confirmed with XPS. Hence, the order of preferential adsorption is -COOH > -CN > -NCS. A similar preference for carboxylic acid over nitrile has been observed for cyanobenzoic acid on Ag, which adsorbs through the COOH group.56 Figure 5 shows the PM-IRRAS spectra in the 12001800 cm-1 region after competitive adsorption between terephthalic acid and terephthalonitrile (bottom) and between terephthalic acid and p-phenylene diisothiocyanate (top) at a concentration of 1 mM for 18 h. Both (55) Briggs, D. In Practical Surface Analysis; Briggs, D., Seah, M., Eds.; Wiley: Chichester, 1990. (56) Han, S.; Han, H.; Kim, K. Vib. Spectrosc. 1999, 21, 131-142.

Langmuir, Vol. 21, No. 12, 2005 5399

Figure 5. PM-IRRAS analysis after competitive adsorption between terephthalic acid and terephthalonitrile (bottom) and p-phenylene diisothiocyanate (top) for 18 h.

spectra exhibit similar features, consisting of absorption peaks at 1407, 1433, and 1585 cm-1, corresponding to the symmetric stretch νs of free COO-, the symmetric stretch of COO- bonded to the surface, and the asymmetric stretch of bonded COO-, respectively.24,57-58 The asymmetric stretch, νa, of free carboxylate groups is presumed hidden in the low-frequency shoulder of the peak at 1585 cm-1. The peak at 1600 cm-1 is assigned to the in-plane vibration mode of the phenyl ring, and the peak at 1507 cm-1 is identified as the vibration along the long (4,4′) axis of the phenyl ring.24,57 On the basis of the presence of νs absorption peaks and the in-plane stretching vibrations of the phenyl ring, terephthalic acid must be adsorbed in a upright, but tilted configuration, through one COOgroup, with the other oriented away from the surface.24,59 This configuration, in which the carbons of the carboxylic acid groups bonded to the surface60 experience slightly different chemical environments compared to the carbons of the pendant groups, is consistent with the broad (2 eV) XPS carbon peak at 288.6 eV. Some of the PM-IRRAS scans revealed an additional peak around 1700 cm-1, which is in the range expected for the CdO stretch of side-by-side, or “polymeric” hydrogen-bonded carboxylic acid groups.59,61 Several vibration peaks associated with ring stretching along the 4,4′ axis of the phenyl ring were also observed. As shown in Figure 6, samples immersed for an extended adsorption time of 3 weeks prior to analysis exhibited higher intensity of the peaks for the COO- stretching and the phenyl ring stretching vibrations, while the peak intensity in the range of 1700-1740 cm-1 completely vanished. Taken together, these observations are consistent with formation of a single-layer OOTF adopting an increasingly surface-perpendicular molecular orientation over time. Thiol versus Carboxylic Acid. Figure 7 shows the XPS analysis after competitive adsorption between 1,4benzenedithiol (HS-φ-SH) and terephthalic acid at a concentration of 5 mM for 24 h. The presence of carbon and sulfur peaks at ∼285, 164, and 228 eV is consistent with adsorption of thiol on the surface. The peak fit for the C1s peak, shown in Figure 8a, exhibits two peaks, at 284.8 and 286.2 eV, corresponding to aromatic carbon and carbon adjacent to thiol, respectively.62,63 Significantly, there is no intensity in the region (57) Varsanyi, G. Assignments for Vibrational Spectra of Seven Hundred Benzene Derivatives; Wiley: New York, 1974. (58) Boerio, F.; Chen, S. J. Colloid Interface Sci. 1980, 73, 176-185. (59) Smith, E.; Alves, C.; Anderegg, J.; Porter, M. Langmuir 1992, 8, 2707-2714. (60) Allara, D.; Nuzzo, R. Langmuir 1985, 1, 45-52. (61) Duevel, R.; Corn, R. Anal. Chem. 1992, 64, 337-342. (62) Vogt, A.; Taejoon, H.; Beebe, T. Langmuir 1997, 13, 3397-3403.

5400


Pranger et al.

Figure 6. PM-IRRAS analysis after competitive adsorption between terephthalic acid and terephthalonitrile (bottom) and p-phenylene diisothiocyanate (top) for 3 weeks. Note vibration peaks associated with phenyl ring stretching at 1160, 1030, 985, and 840 cm-1.

Figure 8. (a) Peak fitting of carbon peak of Figure 7. Note that no peak intensity is observed in the range of 288-289 eV. This shows that no carbon is present as CdO; hence, no terephthalic acid has adsorbed. (b) Sulfur peak of Figure 7, showing that significant oxidation of free thiol groups to sulfonate occurred after 24 h adsorption time. Figure 7. XPS analysis after competitive adsorption between 1,4-benzenedithiol and terephthalic acid.

of 288-289 eV, which indicates that no adsorption of carboxylic acid has occurred. This demonstrates that 1,4benzenedithiol preferentially adsorbs over terephthalic acid. Hence, the order of preferential adsorption is -SH > -COOH > -CN > -NCS. This is consistent with previous work showing that COOH-terminated alkanethiols chemisorb on Cu through the thiol group rather than through the carboxylic acid group64 and that alkanethiols end-functionalized with the nitrile group adsorb through the thiol group.44,59,65 Figure 8b shows the sulfur peak after 24 h of adsorption. In addition to the main 2p3/2/2p1/2 doublet, with 2p3/2 centered at 162.9 eV corresponding to thiolate, there is also a second doublet with 2p3/2 centered at 168.0 eV. Since aromatic thiols are known to oxidize,14,66 even under conditions where the oxygen content of the adsorption system is minimized (i.e., by N2 flushing8 and deoxygenation of solvent prior to sample immersion13), the peak in the high-binding-energy region is assigned to sulfur in sulfonate groups. The oxidation of thiol to sulfonate increased dramatically when the concentration was increased to 5 mM and the adsorption time extended to 1 week. In parallel, the ambient Cu surface was observed to undergo further oxidization to Cu(II), as evidenced by the appearance of a broad satellite (shake-up) peak (63) It is possible that the thiol peak cannot actually be resolved due to conjugation effects. The carbon peak can alternately be fitted with a single peak with fwhm ) 1.37 eV for χ2 < 2.0. (64) Evans, S.; Ulman, A. J. Am. Chem. Soc. 1991, 113, 5866-5868. (65) Chidsey, C.; Loiacono, D. Langmuir 1990, 6, 682-691. (66) Weckenmann, U.; Mittler, S.; Naumann, K.; Fischer, R. Langmuir 2002, 18, 5479-5486.

Figure 9. Cu2p region after competitive adsorption between 1,4-benzenedithiol and terephthalic acid at a concentration of 5 mM for 24 h (bottom) and 1 week (top). Note satellite peak in top spectrum.

between the Cu 2p3/2 and 2p1/2 peaks, shown in Figure 9. It has been suggested on the basis of previous findings that the oxidation of thiol to sulfonate and the oxidation of Cu(I) to Cu(II) may be related processes.14 OOTFs formed from COOH-terminated alkanethiols chemisorb on Cu have been noted for surface pitting.13 This was also observed is the present study for an OOTF of mercaptoundecanoic acid on ambient Cu. Defects are thought to be introduced by inversion, whereby an occasional COOH group bonds to the Cu in lieu of thiol, and by the formation of hairpin bonds, or “loops”, where both terminal functionalities group have bonded to the surface.67 (67) Laibinis, P.; Whitesides, G. J. Am. Chem. Soc. 1992, 114, 19901995.


Langmuir, Vol. 21, No. 12, 2005 5401

Figure 10. XPS analysis after competitive adsorption between 1,4-phenylene diisocyanide and terephthalic acid. The oxygen peak is attributed to oxides of Cu.

If present, a standing-up phase OOTF of 1,4-benzenedithiol should be identifiable in PM-IRRAS by the absorption peaks of the phenyl ring, which occur at approximately 1012, 1550 and 1600 cm-1.8 In addition, when oxidized sulfur species are present, a distinct sulfonate peak is expected in PM-IRRAS at ∼1169 cm-1.8,13 However, no peaks corresponding to the phenyl ring, thiol, or sulfonate groups were observed in the PM-IRRAS analysis after competitive adsorption between 1,4-benzenedithiol and terephthalic acid for 24 h. Extending the adsorption time to 1 week had no effect on the PM-IRRAS results. This demonstrates that, when 1,4-benzenedithiol adsorbs on ambient Cu, it forms a lying-down phase OOTF bonding through both terminal groups. Difunctional molecules may become kinetically trapped in the lyingdown configuration if the interaction between their terminal functional groups and the substrate is strong enough, even though the standing-up phase would be energetically more favorable. This appears to be the case for benzene dithiol on ambient Cu. To investigate whether the failure to obtain a standing-up phase is due to a strong interaction between the phenyl ring and the ambient Cu surface, a concentrated (50 mM) solution of 1,6 hexanedithiol was allowed to self-assemble on ambient Cu for 72 h for comparison. Also in this case, XPS confirmed that dithiol was present, with incipient oxidization of thiol to sulfonate. However, no peaks were observed in the methylene or sulfonate stretch regions, ruling out the presence of both standing-up and hairpin-bonded species. This demonstrates that 1,6-hexanedithiol adopts a flat configuration on ambient Cu. A lying-down configuration has also been described for 1,8-octanedithiol on Au and Ag substrates.1 Isocyanide versus Carboxylic Acid. Figure 10 shows the XPS analysis after competitive adsorption between 1,4-phenylene diisocyanide (CN-φ-NC) and terephthalic acid at a concentration of 5 mM for 24 h. The carbon and nitrogen peaks detected at ∼285 and 400 eV, respectively, are consistent with adsorption of phenylene diisocyanide. Both carbon and nitrogen peaks are highly asymmetric and require three peaks to obtain χ2 < 2. The peak fit for the carbon C1s peak, shown in Figure 11a, has significant asymmetry on the high-binding-energy side and can be fitted with three peaks at 284.8, 286.1, and 287.6 eV. Carbon atoms in isocyanide groups experience a relatively strong upward chemical shift. For example, in methyl isocyanide (gas phase), the isocyanide group causes a high chemical shift of 2.5 eV on the carbon of the adjacent methyl group, while the carbon of the isocyanide

Figure 11. (a) Peak fitting of carbon C1s peak of Figure 10. (b) Peak fitting of nitrogen N1s peak of Figure 10.

group itself is shifted an additional 0.7 eV.68 In the present case of chemisorbed, π-conjugated diisocyanide, chemical shifts are affected by the donation of σ electrons from the bonded isocyanide group to surface Cu atoms, backbonding of electrons from Cu to the antibonding π* orbitals of the bonded isocyanide group, and conjugation effects.55 The broad binding energy peak at 287.6 eV is tentatively assigned to carbon in both bound and free isocyanide groups, the middle binding energy peak at 286.1 to carbon atoms adjacent to isocyanide groups, and the main peak at 284.8 eV to aromatic carbon, as usual. The broad, high-binding-energy peak has some intensity in the region 288-289 eV, which might indicate the presence of carboxylic carbon. However, no absorption peaks for COO- or COOH groups were detected in the PM-IRRAS spectrum. This demonstrates that coadsorption of terephthalic acid in the OOTF is negligible and that 1,4-phenylene diisocyanide preferentially adsorbs over terephthalic acid. This indicates that the isocyanide group has a higher affinity for ambient Cu than the carboxylic acid group. Hence, the order of preferential adsorption is -NC > -COOH > -CN > -NCS. The N1s peak, shown in Figure 11b can be fitted with three peaks at 400.8, 399.7, and 398.6 eV. The peaks at 400.8 and 399.7 eV are tentatively assigned to bonded and free isocyanide group, respectively. The peak at 398.6 eV may indicate that some isocyanide groups have converted to imine groups as a result of rehybridization.32,69-70 Possible explanations for this are oligomerization of terminal isocyanide groups to polyisocyanide, which has already been described on Au and Pt,55,71-72 or (68) Clark, D.; Harrison, A. J. Polym. Sci. Pol. Chem. 1981, 19, 19451955. (69) Sexton, B.; Avery, N. Surf. Sci. 1983, 129, 21-36. (70) Ellison, M.; Hamers, R. J. Phys. Chem. B 1999, 103, 62436251.

5402


Figure 12. XPS spectra after competitive adsorption between 1,4-phenylene diisocyanide and terephthalic acid at a concentration of 5 mM, 24 h (bottom); 5 mM, 1 week (middle); and saturated solution, 72 h (top).

Figure 13. PM-IRRAS spectrum after competitive adsorption between 1,4-phenylene diisocyanide and terephthalic acid.

µ2-η1 bridge-bonding, which has been observed on Pt at high coverage.73 In the PM-IRRAS spectrum, there is no peak intensity in the range of 1580-1800 cm-1 which could be assigned to an imine stretch, and hence, the presence of a µ2-η1 bridge-bonded configuration is unlikely.71-74 The low-binding-energy peak at 398.6 eV is therefore presumed to be due to the presence of polyisocyanide. On the basis of the absence of peak intensity in the imine region of PM-IRRAS, its most likely source is polyisocyanide adsorbed on top of the OOTF. This has previously been suggested for 1,4-phenylene diisocyanide on Au on the basis of angle-resolved XPS and UPS data.75 When the competitive adsorption between 1,4-phenylene diisocyanide (CN-φ-NC) and terephthalic acid was conducted with higher initial concentrations and longer adsorption periods, the Cu peak intensity was attenuated, consistent with increased adsorption of polyisocyanide on the surface of the OOTF, as shown in Figure 12. Figure 13 shows the PM-IRRAS spectrum after competitive adsorption between 1,4-phenylene diisocyanide (CN-φ-NC) and terephthalic acid. At a concentration of 5 mM for 24 h (bottom), very weak, but distinguishable peaks were observed at 1500 and 1600 cm-1. These peaks are identified as vibration modes of the phenyl ring.8,46 In addition, a single, broad isocyanide peak is seen around 2160 cm-1. In the PM-IRRAS spectrum after competitive adsorption from a 50 mM (saturated) solution for 72 h (71) Robertson, M.; Angelici, R. Langmuir 1994, 10, 1488-1492. (72) Lin, S.; McCarley, R. Langmuir 1999, 15, 151-159. (73) Avery, N.; Matheson, T. Surf. Sci. 1984, 143, 110-124. (74) Kang, D.; Trenary, M. J. Phys. Chem. B 2002, 106, 5710-5718. (75) Huc, V.; Bourgoin, J.; Bureau, C.; Valin, F.; Zalczer, G.; Palacin, S. J. Phys. Chem. B 1999, 103, 10489-10495.

Pranger et al.

(top), the ring stretching peaks have increased significantly in intensity, demonstrating that the phenylene diisocyanide has adopted a more surface-perpendicular configuration on the ambient Cu surface. In Figure 13b, the isocyanide peak is resolved as two peaks, at 2165 and 2125 cm-1, corresponding to surface-bonded and free isocyanide groups, respectively. This spectrum is comparable to those previously obtained for 1,4-phenylene diisocyanide on Ag and Au.46,76 The observation of only a single, broad isocyanide peak in Figure 13a (bottom) at 2160 cm-1 may be attributed to an overlap between free and bonded isocyanide stretching vibrations for the following reasons: (1) With flexible backbone diisocyanides, a single isocyanide stretching vibration may be observed, if the diisocyanide is adsorbed through both terminal groups in an η1 bonding configuration (i.e., “hairpin” bonding).72 This would be consistent with the single broad peak at 2165 cm-1. However, a hairpin bonding configuration is not possible with phenylene diisocyanide due to the rigidity of the conjugated structure.71 (2) The peak position observed in PM-IRRAS cannot be reconciled with a flat, η2(C,N) bonding configuration through a single terminal group. In this configuration, the dipole moment for the η2(C,N) bond is parallel to the surface, hence IR-silent, and a single peak would be observed for the pendant isocyanide group at 2125 cm-1.75 However, this is inconsistent with the PM-IRRAS peak at 2165 cm-1. (3) A significant overlap of free and bonded isocyanide peaks has previously been observed for 1,4phenylene diisocyanide on Ag and Au due to high asymmetry on the low-frequency side of the bonded isocyanide stretch.46,76 Hence, the resolution of the single peak of Figure 13a into two peaks in Figure 13b (top) is probably the result of an increased packing order due to increased concentration and time of adsorption. Interestingly, in Figure 13b, the free isocyanide stretching peak (2125 cm-1) is much smaller than the bonded isocyanide peak (2165 cm-1), although these peaks should be comparable in intensity. A similar disparity between peaks has also been observed for 1,4-phenylene diisocyanide on Au.55,71 The position of ν(CN) for the bonded isocyanide group is blue-shifted to 2165 cm-1 compared to 2128 cm-1 for 1,4-phenylene diisocyanide in dichloromethane, while the free group is slightly red-shifted to 2125 cm-1. These shifts are close to those of phenylene diisocyanide on Au and Ag, reported as 2180 and 2120 cm-1.46,71 The fact that the pendant isocyanide group exhibits a slight red-shift at 2120 cm-1 indicates that it experiences the increased electron density in the antibonding π* orbital of the bonded isocyanide group due to π-conjugation.6,54,72,73,76-80 Isocyanide versus Thiol. High chemical reactivity between 1,4-phenylene diisocyanide and 1,4-benzenedithiol interfered with competitive adsorption of this twocomponent system. A 5 mM THF solution of 1,4-phenylene diisocyanide and 1,4-benzenedithiol became cloudy during overnight agitation for 24 h, changing from clear to a dark orange color, with a significant amount of particulate matter precipitating from solution. This is presumed mainly due to oligomerization of phenylene diisocyanide to polyisocyanide. The PM-IRRAS spectrum, shown in Figure 14, exhibited significant peak intensity in the range of 1650-1720 cm-1, while the intensity in the isocyanide stretch region was practically negligible, with a barely discernible peak at ∼2160 cm-1. Strong and distinct peaks were also observed at 1510, 1610, and 1032 cm-1 for the (76) Kim, H.; Lee, S.; Kim, N.; Yoon, J.; Park, H.; Kim, K. Langmuir 2003, 19, 6701-6710.


Langmuir, Vol. 21, No. 12, 2005 5403

Figure 14. PM-IRRAS scan after competitive adsorption between 1,4-phenylene diisocyanide, and 1,4-benzenedithiol.

Figure 15. XPS analysis after competitive adsorption between diisocyanohexane and hexanedithiol at a concentration of 1 mM for 3 weeks.

in-plane vibrations of the phenyl ring and in the sulfonate stretch region 1130-1230 cm-1. This is consistent with the formation of a mixed OOTF. Since the reactivity of the two-component system of phenylene diisocyanide and benzenedithiol interfered with the self-assembly process, competitive adsorption was also studied between corresponding aliphatic chain adsorbents, 1,6-hexanedithiol and 1,6-diisocyanohexane. The XPS spectrum of this system shown in Figure 15 is consistent with the conclusion that co-adsorption of dithiol and diisocyanide has taken place to form a mixed OOTF. Both nitrogen and sulfur peaks were observed, with comparable intensity. This is consistent with a mixed OOTF. The peak fit for the C1s peak, shown in Figure 16a, exhibits three peaks at 284.8, 286.0, and 286.8 eV, corresponding to aliphatic carbon, carbon adjacent to isocyanide, thiol or sulfonate groups, and carbon in isocyanide groups, respectively. The N1s peak, shown in Figure 16b, can be fit with three peaks at 400.3, 399.2, and 398.3 eV, corresponding to nitrogen in bonded isocyanide groups, free isocyanide groups, and imine groups, respectively. The peak fit for the sulfur 2p peak is shown in Figure 16c. Significantly, the main 2p3/2 peak of the 2p3/2/2p1/2 doublet is centered at 162.4 eV, which is indicative of thiol bonded directly to the ambient Cu substrate as thiolate. The doublet with 2p3/2 centered at 167.2 eV indicates oxidation of thiol to sulfonate. The PM-IRRAS spectrum for this system is shown in Figure 17a (bottom). There is a strong isocyanide peak at 2189 cm-1. This peak position is higher than that of bulk 1,6-diisocyanohexane at 2146 cm-1; hence, it is assigned to η1-bonded isocyanide groups.12 Peaks were observed at

Figure 16. (a) Peak fitting of carbon C1s peak of Figure 15. (b) Peak fitting of nitrogen N1s peak of Figure 15. (c) Peak fitting of sulfur S2p peak of Figure 15.

1352 and 1460 cm-1 for the methylene wagging and scissors vibrations and at 2866 and 2935 cm-1, corresponding to the symmetric and asymmetric stretching vibrations of the methylene groups of the hexane backbone.45,61,72,82 The positions of the latter peaks indicate a relatively disordered OOTF.61,72 This is consistent with hairpin bonding configuration. The formation of some polyisocyanide is evident from the weak and broad, but distinguishable, peak centered at 1650 cm-1 which is in the range expected for an imine stretch. A similar imine peak has been observed between 1580 and 1680 cm-1 after self-assembly of diisocyanohexane on Au.72 The broad peak at 1207 cm-1 is assigned to the stretching vibrations of sulfonate groups, the presence of which was concluded from the XPS sulfur peak. These peak assignments are (77) Kang, J.; Ulman, A.; Jordan, R. Langmuir 1999, 15, 5555-5559. (78) Horswell, S.; Kiely, C.; O’Neill, I.; Schiffrin, D. J. Am. Chem. Soc. 1999, 121, 5573-5574. (79) Shih, K.; Angelici, R. Langmuir 1995, 11, 2539-2546. (80) Ontko, A.; Angelici, R. Langmuir 1998, 14, 1684-1691. (81) Guy, M.; Coffer, J.; Rommel, J.; Bennett, D. Inorg. Chem. 1988, 27, 2942-2945. (82) Nuzzo, R.; Dubois, L.; Allara, D. J. Am. Chem. Soc. 1990, 112, 558-569.

5404


Pranger et al.

groups have similar affinity for ambient Cu. Hence, the final order of preferential adsorption is -NC ≈ -SH > -COOH > -CN > -NCS.

Figure 17. PM-IRRAS analysis after competitive adsorption between diisocyanohexane and hexanedithiol at a concentration of 1 mM for 3 weeks (bottom) and adsorption from a solution containing only concentrated 1,6-hexanedithiol at a concentration of 50 mM for 72 h (top).

supported by the PMIRRAS spectra obtained after adsorption from a pure solution of 1,6-diisocyanide at a concentration of 50 mM for 72 h, shown in Figure 17b (top). Comparing parts a and b of Figure 17, the spectra are very similar, with the exception of the peaks at 1036 and 1207 cm-1. This supports the assignment of these peaks to coadsorbed 1,6-hexanedithiol, which has presumably been induced to adopt a standing-up configuration by neighboring diisocyanide molecules. Taken together, the presence of both roughly equal amounts of sulfur and nitrogen in XPS, in conjunction with the identification of thiolate bonds in XPS and surface-bonded isocyanide groups in PM-IRRAS, demonstrates that competitive self-assembly of these two molecules leads to coadsorption of dithiol and diisocyanide as distinct species. This indicates that the isocyanide and thiol functional

Conclusions In summary, of the difunctional molecules studied in this work, diisocyanide and dithiol have the strongest affinity for ambient Cu. Dithiols tend to become kinetically trapped in a flat configuration on the substrate. In contrast, 1,4-phenylene diisocyanide and 1,4-terephthalic acid both adopt a standing-up phase configuration in which the molecules bonds to the ambient Cu substrate through one terminal functional group, with the other terminal group disposed away from the substrate. Of the difunctional molecules studies in this work, these are the most suitable precursors for further development of ultrathin protective coatings, adhesion promoters, and ultimately, smart molecular solder for ambient Cu substrates. Further studies are required to characterize what appears to be a topochemical polymerization of free isocyanide functions of the diisocyanide OOTF. Acknowledgment. This work was supported by a seed grant from the Center for Board Assembly Research (CBAR) in the Manufacturing Research Center (MaRC) at the Georgia Institute of Technology, and by the Petroleum Research Fund of the American Chemical Society (PRF 40014-AC5M). We wish to acknowledge Professors Taylor and Danyluk from the Georgia Institute of Technology for their support and encouragement and Dr. Rick Haasch from the University of Illinois at UrbanaChampaign for his kind assistance in collecting XPS data. XPS data from the Kratos Axis Ultra instrument was carried out in the Center for Microanalysis of Materials at the University of Illinois, which is partially supported by the U.S. Department of Energy under Grant No. DEFG02-91-ER45439. LA050508L

Competitive Self-Assembly of Symmetrical, Difunctional Molecules on

Recommend Documents