Self-Assembly of 1, 2-Benzenedithiol on Gold and Silver: Fourier

The adsorption of 1,2-benzenedithiol (1,2-BDT) in benzene on gold and silver surfaces has been investigated by .... Caroline M. Whelan, Malcolm R. Smy...
0 downloads 0 Views 213KB Size
5830

Langmuir 1996, 12, 5830-5837

Self-Assembly of 1,2-Benzenedithiol on Gold and Silver: Fourier Transform Infrared Spectroscopy and Quartz Crystal Microbalance Study Young Joo Lee,† Il Cheol Jeon,‡ Woon-kie Paik,§ and Kwan Kim*,† Department of Chemistry and Center for Molecular Catalysis, Seoul National University, Seoul 151-742, Korea, Department of Chemistry, Jeonbuk National University, Jeonbuk 561-756, Korea, and Department of Chemistry, Sogang University, Seoul 121-742, Korea Received April 1, 1996. In Final Form: August 8, 1996X The adsorption of 1,2-benzenedithiol (1,2-BDT) in benzene on gold and silver surfaces has been investigated by infrared reflection-absorption (IRA) spectroscopy, quartz crystal microbalance (QCM), and ellipsometry. The IRA spectral data were found to be consistent with those of QCM and ellipsometry. On both metals, 1,2-BDT was chemisorbed by forming two metal-sulfur bonds after deprotonation. Although monolayered film was formed exclusively on gold, multilayered film could also be formed on silver. The orientation on gold changed little with respect to the surface coverage, but structural change from vertical to tilted orientation occurred on silver at the fractional surface coverage near 0.6. The tilt angles of benzene ring were estimated to be ca. 51° on gold and 38° on silver with respect to the surface normal. From the analysis of the IRA peak intensities, the free energy of adsorption, ∆Gao, under submonolayer coverage limit was estimated to be -27.3 kJ/mol on gold and -39.6 kJ/mol on silver, suggesting that adsorption on silver would be energetically more favorable than that on gold. The multilayered film formed on silver consisted of aggregates of randomly oriented silver thiolate.

Introduction Spectroscopic characterization of molecular adsorbates on metal surfaces is very important for the fundamental understanding of various phenomena such as heterogeneous catalysis1-3 and corrosion.3,4 Spectroscopic information is also invaluable for the preparation of various types of self-assembled organic films applicable to electronic and optical devices,5,6 chemical sensors,7 artificial membranes,8,9 and electron transfer barriers.10,11 Among the several spectroscopic techniques developed for this purpose, vibrational spectroscopic techniques such as infrared reflection absorption spectroscopy (IRAS),12-14 surface-enhanced Raman scattering (SERS),15-17 and electron energy loss spectroscopy (EELS)18,19 have provided * Author to whom all correspondence should be addressed: e-mail, [email protected]; tel, 82-2-880-6651; fax, 82-2889-1568. † Seoul National University. ‡ Jeonbuk National University. § Sogang University. X Abstract published in Advance ACS Abstracts, November 1, 1996. (1) Somorjai, G. A. Chemistry in Two Dimensions: Surfaces; Cornell University Press: Ithaca, New York, 1981. (2) Murray, R. W. In Electroanalytical Chemistry; Bard, A. J., Ed.; Marcel Dekker: New York, 1984; Vol. 13. (3) Adamson, A. W. Physical Chemistry of Surfaces; John Wiley & Sons: New York, 1990. (4) Notoya, T.; Poling, G. W. Corrosion 1979, 35, 193-200. (5) Prasad, P. N.; Williams, D. J. Introduction to Nonlinear Optical Effects in Organic Molecules and Polymers; John Wiley & Sons: New York, 1990. (6) Dalton, L. R.; Sapochak, L. S.; Yu, L. J. Phys. Chem. 1993, 97, 2871-2883. (7) Beswick, R. B.; Pitt, C. W. Chem. Phys. Lett. 1988, 143, 589-592. (8) Holmes-Farley, S. R.; Bain, C. D.; Whitesides, G. M. Langmuir 1988, 4, 921-937. (9) Lukes, P. J.; Petty, M. C.; Yarwood, J. Langmuir 1992, 8, 30433050. (10) Chidsey, C. E. D.; Bertozzi, C. R.; Putvinski, T. M.; Mujsce, A. M. J. Am. Chem. Soc. 1990, 112, 4301-4306. (11) Hickman, J. J.; Ofer, D.; Zou, C.; Wrighton, M. S.; Laibinis, P. E.; Whitesides, G. M. J. Am. Chem. Soc. 1991, 113, 1128-1132. (12) Troughton, E. B.; Bain, C. D.; Whitesides, G. M.; Nuzzo, R. G.; Allara, D. L.; Porter, M. D. Langmuir 1988, 4, 365-385. (13) Smith, E. L.; Porter, M. D. J. Phys. Chem. 1993, 97, 8032-8038. (14) Stranick, S. J.; Parikh, A. N.; Tao, Y.-T.; Allara, D. L.; Weiss, P. S. J. Phys. Chem. 1994, 98, 7636-7646.

S0743-7463(96)00313-7 CCC: $12.00

the most valuable information on the structural details of adsorbates. The sensitivity of SERS is remarkable, enabling routine investigation of adsorbates even at submonolayer coverages, but SERS has two noticeable disadvantages: that its applicability is limited to a few metals and that its unequivocal surface selection rule has not been established yet.16,20 Such limitations are not encountered when one employs IRAS. The intrinsic weakness of IRAS is its relatively low sensitivity compared with SERS. These characteristics suggest that the combination of SERS and IRAS should be very useful in investigating the molecular adsorption on metal surfaces. In this sense, we have recently investigated the adsorption of 1,2-benzenedithiol (1,2-BDT) on a silver surface by the combination of SERS and IRAS, attempting to use IRA spectral information to test the validity of variously proposed SER selection rules.21 During the course of that work, the IRA spectral pattern was observed to be very susceptible to adsorption conditions, such as the concentration and the length of time allowed for 1,2-BDT to contact the silver surface. At higher concentration and longer adsorption time, 1,2-BDT seemed to form multilayers on silver. We could not observe, however, any dramatic spectral change in the SER spectra. This may be understood by recalling that the SERS effect is only observed for the molecules in intimate contact with the metal substrate.15,16 As a continuation of our effort to understand the adsorption behavior of aromatic thiols on metal surfaces, a more thorough IRAS study than that performed previ(15) Chang, R. K., Furtak, T. E., Ed. Surface Enhanced Raman Scattering; Plenum: New York, 1982. (16) Otto, A.; Mrozek, I.; Grabhorn, H.; Akemann, W. J. Phys.: Condens. Matter 1992, 4, 1143-1212. (17) Lewis, M.; Tarlov, M.; Carron, K. J. Am. Chem. Soc. 1995, 117, 9574-9575. (18) Ibach, H.; Mills, D. L. Electron Energy Loss Spectroscopy and Surface Vibrations; Academic Press: New York, 1982. (19) Stern, D. A.; Wellner, E.; Salaita, G. N.; Laguren-Davidson, L.; Lu, F.; Batina, N.; Frank, D. G.; Zapien, D. C.; Walton, N.; Hubbard, A. T. J. Am. Chem. Soc. 1988, 110, 4885-4893. (20) Cho, S. H.; Han, H. S.; Jang, D.-J.; Kim, K.; Kim, M. S. J. Phys. Chem. 1995, 99, 10594-10599. (21) Cho, S. H.; Lee, Y. J.; Kim, M. S.; Kim, K. Vib. Spectrosc. 1996, 10, 261-270.

© 1996 American Chemical Society

Self-Assembly of 1,2-BDT on Au and Ag

Langmuir, Vol. 12, No. 24, 1996 5831

ously is made in this work on 1,2-BDT self-assembly on gold and silver surfaces. An attempt is made to check whether the IRA spectral feature can be correlated with the data obtainable from ellipsometry and quartz crystal microbalance studies. It is intended also to know whether the orientation of adsorbed aromatic thiols is dependent on the metal substrates in analogy with long chain aliphatic thiols.22,23 In addition, the kinetics of adsorption of 1,2-BDT on the gold and silver surfaces is examined by monitoring the IRA peak intensities as a function of selfassembling time at various concentrations. It is tested whether the transient Langmuir adsorption kinetic model24 can be applied to the present system. The possibility of order-disorder transition as a function of surface coverage is examined. Furthermore, in cases when multimolecular layers are formed, the aim of our investigation is to understand whether the metal-adsorbate complex is formed in the solution phase prior to aggregating onto the metal substrate. Experimental Section The metal substrates used for self-adsorption of 1,2-BDT were prepared by the resistive evaporation of chromium (Aldrich, 99.5%) and gold (Aldrich, 99.9%) or silver (Aldrich, 99.99%) at 10-5-10-6 Torr on batches of glass slides, cleaned previously with (NH4)2S2O8/H2SO4 solution and then sonicated. Deposition of chromium prior to that of gold or silver was performed to enhance adhesion to the substrate. After a deposition of approximately 200 nm of gold or silver, the evaporator was backfilled with nitrogen. Adsorbate solutions were prepared by dissolving weighed portions of 1,2-BDT in nitrogen-bubbled benzene to desired concentrations covering the range from 10-2 to 10-7 M. Precleaned glass vials were used as the self-adsorption cells. The metal substrates were immersed subsequently into the adsorbate solutions for a predetermined period of time. After the substrates were removed, they were rinsed with excess benzene and then subjected to a strong nitrogen gas jet to blow off any remaining liquid droplets on the surface or the edges of the substrates. The ellipsometric thickness of self-assembled 1,2-BDT was estimated using a Gaertner Model L116C optical ellipsometer. The measurement was performed with a 632.8 nm line of He/Ne laser incident upon the sample at 60°. The infrared spectra were obtained with a Bruker IFS 113v Fourier transform spectrometer equipped with a globar light source and a liquid N2-cooled mercury cadmium telluride detector. The method for obtaining the IRA has been reported previously.25,26 Each spectrum was obtained by averaging 1024 interferograms at 4 cm-1 resolution, with p-polarized light incident on the metal substrate at 83°. To reduce the effect of water vapor rotational lines, the sample and reference interferograms were recorded alternately after every 32 scans. The Happ-Genzel apodization function was used in Fourier transforming all the interferograms. The IRA spectra are reported as -log(R/Ro), where R and Ro are the reflectivities of the sample and the bare clean metal substrates, respectively. Separately, the self-adsorption of 1,2-BDT on gold and silver was examined in situ with a quartz crystal microbalance (QCM). The apparatus consisted of a frequency counter (in house customized, 1 Hz resolution), an oscillator (in house customized), and a 10 MHz Au- or Ag-coated quartz crystal (International Crystal). Just before the experiment, the quartz crystal was cleaned with a piranha solution, rinsed with ethanol, and dried with a N2 gas jet. The cleaned crystal was mounted in a Teflon cell with Viton O-rings. After a steady state was reached in the presence of pure benzene, an aliquot of 1,2-BDT solution was (22) Porter, M. D.; Bright, T. B.; Allara, D. L.; Chidsey, C. E. D. J. Am. Chem. Soc. 1987, 109, 3559-3568. (23) Laibinis, P. E.; Whitesides, G. M.; Allara, D. L.; Tao, Y.-T.; Parikh, A. N.; Nuzzo, R. G. J. Am. Chem. Soc. 1991, 113, 7152-7167. (24) Chen, S. H.; Frank, C. W. Langmuir 1989, 5, 978-987. (25) Son, D. H.; Ahn, S. J.; Lee, Y. J.; Kim, K. J. Phys. Chem. 1994, 98, 8488-8493. (26) Kim, S. H.; Ahn, S. J.; Kim, K. J. Phys. Chem. 1996, 100, 71747180.

Figure 1. (a) IRA spectrum of 1,2-BDT self-assembled on gold in 1.0 × 10-3 M solution for 6 h. (b) Same as (a) but self-assembled on silver in 1.0 × 10-6 M solution. (c) Transmission infrared spectrum of neat 1,2-BDT in molten state. (d) Same as (b) but self-assembled in 10-6 M solution for 10 min. introduced into the cell to attain the desired concentration. The subsequent frequency changes due to self-adsorption were monitored with respect to the initial frequencies of the crystals in neat benzene. All chemicals otherwise specified were reagent grade. To prevent air-oxidation, 1,2-BDT purchased from Aldrich was always handled under dry N2 atmosphere.

Results and Discussion IRA Spectral Feature. Parts a and b of Figure 1 show the IRA spectra of 1,2-BDT self-assembled on gold and silver surfaces, respectively, by immersing the metal substrates in 1.0 × 10-3 (for gold) and 1.0 × 10-6 (for silver) M solutions for 6 h. As to be discussed later, they should correspond to near complete monolayers. For comparative purposes, the transmission (TR) infrared spectrum of neat 1,2-BDT in molten state is shown in Figure 1c. Positions of major peaks appearing in Figure 1 are summarized in Table 1 together with their vibrational assignments. It is seen that only a few peaks appear in the IRA spectra. Three peaks are identified clearly on gold while six peaks are seen for 1,2-BDT on silver. In the previous study,21 we could correlate the IRA peaks on silver with those in the TR spectra of neat 1,2-BDT and its silver salt. Namely, the IRA peaks at 3055, 1439, 1245, 1122, 1033, and 748 cm-1 in Figure 1b can be correlated, respectively, with the TR peaks at 3054 (ν20b, A1), 1452 (ν19b, A1), 1252 (ν14, A1), 1125 (ν9a, A1), 1041 (ν18b, A1), and 743 (ν11, B1) cm-1 in Figure 1c. At first glance, one might be tempted to correlate the IRA peak at 1439 cm-1 in Figure 1b with the TR peak at 1428 (ν19a, B2) cm-1 rather than the peak at 1452 cm-1 in Figure 1c. We prefer, however, the latter correlation. In the SERS study,21 the 1,2-BDT molecule was found to adsorb on silver by forming two Ag-S bonds after deprotonation. In so far as the adsorption takes place via the formation of two Ag-S bonds, only the x- and z-polarized vibrational bands should appear in the IRA spectrum when the molecular axes of 1,2-benzenedithiolate are defined such that the x-axis lies perpendicularly to the ring and the z-axis bisects the

5832 Langmuir, Vol. 12, No. 24, 1996

Lee et al.

Table 1. Infrared Spectral Data for 1,2-Benzenedithiol (1,2-BDT)a 1,2-BDTb 431 657 705 743 926 941 1041 1113 1125 1160 1252 1265 1428 1452 1561 1571 2536 2558 3054

s m m vs m w s s m vw sh m s vs m m m sh w

silver saltc

IRA on Agd

653

s

745

vs

748

vs

1030 1096 1121 1159 1244

s m m w m

1033

s

1122

w

1245

m

1421 1438 1549 1566

m s vw m

1439

vs

3045

w

3055

w

IRA on Aue

739

1435

3048

vs

s

w

assignmentf 16b 6a 6b 11 β(SH) 17b 18b 1 9a 18a 14 3 19a 19b 8a 8b ν(SH) ν(SH) 20b

B1 A1 B2 B1 B1 A1 A1 A1 B2 A1 B2 B2 A1 A1 B2 A1

a

Wavenumbers in cm-1. Intensities: vs, very strong; s, strong; m, medium; w, weak; vw, very weak; sh, shoulder. b Taken in molten state. See Figure 1c. c Silver benzenedithiolate salt formed by mixing AgNO3 and 1,2-BDT solutions in a 2:1 mole ratio. d See Figure 1b. e See Figure 1a. f Wilson notation. Symmetry types corresponding to C2v point group.

molecular plane symmetrically (see Figure 2a). This stems from the IRA surface selection rule27,28 that only the vibrational modes whose dipole moment derivatives have components normal to the metal surface are exclusively infrared active. Since the TR peak at 1428 cm-1 belongs to the B2 type (y-polarized band), the IRA peak at 1439 cm-1 has to be then correlated with the TR peak at 1452 cm-1 belonging to the A1 type.29 The unusually large frequency difference, 13 cm-1 ()1452-1439), may be understood by recalling that the 19b mode is very substituent sensitive. This is evidenced from the spectral feature of silver benzenedithiolate salt that in the salt spectrum the counterpart of 1452 cm-1 peak in the neat TR spectrum is observed at 1438 cm-1, being red-shifted by 14 cm-1. On the basis of similar reasoning, we could correlate the IRA peak at 1245 cm-1 in Figure 1b with the shoulder peak at 1252 (ν14, A1) cm-1 rather than the distinct one at 1265 (ν3, B2) cm-1 in Figure 1c. It is intriguing that only three peaks are observed in the IRA spectrum on gold. Nonetheless, those peaks can also be correlated with the TR peaks. Namely, the peaks at 3048, 1435, and 739 cm-1 in Figure 1a can be correlated, respectively, with the TR peaks at 3054 (ν20b, A1), 1452 (ν19b, A1), and 743 (ν11, B1) cm-1 in Figure 1c. The fact that the substituent sensitive 19b mode appears at 1435 cm-1 in the IRA spectrum suggests that the species responsible for Figure 1a is 1,2-benzenedithiolate rather than 1,2BDT. In this sense, there will be no doubt about the absence of S-H stretching peaks in Figure 1a. The absence of B2 type modes can be understood by presuming that the adsorbate is bound to gold symmetrically via two Au-S bonds. On the other hand, the appearance of both the A1 and B1 modes dictates that the benzene ring of adsorbed 1,2-BDT should take neither a perfectly flat nor a perfectly perpendicular orientation with respect to the gold surface. Furthermore, the disappearance of the A1type 9a, 14, and 18b modes in Figure 1a may be ascribed (27) Greenler, R. G. J. Chem. Phys. 1966, 44, 310-315. (28) Golden, W. G. In Fourier Transform Infrared Spectroscopy: Application to Chemical Systems; Ferraro, J. R., Basile, L. J., Ed.; Academic Press: New York, 1985; Vol. 4. (29) According to C2v point group, x-, y-, and z-polarized vibrational bands are classified into B1, B2, and A1 modes, respectively.

Figure 2. (a) Molecular axes and directions of the transition dipoles of A1, B1, and B2 modes of 1,2-benzenedithiolate. (b) Plausible configuration of 1,2-benzenedithiolate adsorbed on gold and silver surfaces.

to the fact that 1,2-benzenedithiolate is inclined toward the gold surface far more than the silver surface. Previously, we estimated the tilt angle of 1,2-BDT on silver21 by approximating the angle β between the benzene ring and the surface normal to be given by30

tan2 β ) [A(A1)/A(B1)]TR [ν(A1)/ ν(B1)] [A(B1)/ A(A1)]IRA (1) in which A(A1) and A(B1) correspond, respectively, to the infrared band intensities of A1 and B1 modes with frequencies of ν(A1) and ν(B1) in TR and IRA spectra. Since the IRA peaks on silver were better correlated with the TR peaks of silver salt than those of neat 1,2-BDT, we used the spectral data of silver salt in the estimation of ring orientation. On the basis of the intensity ratios of ring 11 and 19b modes, the benzene ring of adsorbed 1,2BDT on silver was previously calculated to be tilted away at ca. 38° from the surface normal (see Figure 2b). We can get the same result from the IRA spectrum shown in Figure 1b. Upon adsorbing on the metal surface, vibrational modes of the adsorbate can couple with those of the adsorbent. The transition dipole moments of adsorbate will thus be dependent on the kind of substrate. This may be one of the reasons why the ring 9a, 14, and 18b modes of 1,2benzenedithiolate were barely seen on the gold surface. Nonetheless, assuming that the infrared spectral data of silver salt are not much different from those of gold salt, the tilt angle of 1,2-BDT on the gold surface is calculated to be ca. 51° from the surface normal. (See Figure 2b. The tilt angles of benzene ring, 38° and 51°, imply the S-C bonds of 1,2-benzenedithiolate to take 48° and 58°, respectively, with respect to the surface normal.) Although the present estimation may not be quantitative, it would be worth comparing the orientations of aromatic thiols on silver and gold with those of aliphatic thiols. It is wellknown that the axes of alkyl chains of aliphatic thiols are tilted by 7-15° on silver23,31,32 and by ∼30° on gold.22,23,33 Assuming that the carbon atoms in aliphatic thiols resume tetrahedral coordination, these values imply that the angles between the S-C bond and the surface normal are 35-50° on silver and ∼65° on gold, in fair agreement with those of 1,2-benzenedithiolate. (30) Fan, J.; Trenary, M. Langmuir 1994, 10, 3649-3657. (31) Walczak, M. M.; Chung, C.; Stole, S. M.; Widrig, C. A.; Porter, M. D. J. Am. Chem. Soc. 1991, 113, 2370-2378. (32) Sellers, H.; Ulman, A.; Shnidman, Y.; Eilers, J. E. J. Am. Chem. Soc. 1993, 115, 9389-9401. (33) Nuzzo, R. G.; Dubois, L. H.; Allara, D. L. J. Am. Chem. Soc. 1990, 112, 558-569.

Self-Assembly of 1,2-BDT on Au and Ag

Langmuir, Vol. 12, No. 24, 1996 5833

Table 2. QCM and Ellipsometry Data for 1,2-BDT Self-Assembled on Au and Ag QCM mass increase (ng) monolayer on Au monolayer on Ag multilayer on Ag

ellipsometry thickness (Å)

obsa

calcb

obsc

calcd

36 (1.5 × 1014) 43 (1.8 × 1014) 86e (3.6 × 1014)

37 (1.6 × 1014)

3.3 ( 0.5

3.3

4.9 ( 0.6

4.3

7.7 ( 0.9f

a Measured on gold- or silver-coated quartz. Values in parentheses correspond to the number of adsorbed 1,2-BDT molecules. b Estimated by assuming that the surface roughness factor is 1.2 and that each 1,2-BDT is bonded to two adjacent gold atoms separated by 3 Å. See text. c Measured for 1,2-BDT self-assembled on vacuum evaporated Au or Ag. d Estimated by taking account of the tilting angles, 38° on Ag and 51° on Au, determined from IRA spectra. e After self-assembly for 40 min in 1.0 × 10-3 M solution. See text. f After self-assembly for 70 min in 1.0 × 10-4 M solution. See text.

QCM Measurement. In the above discussion, we have regarded the IRA spectra shown in parts a and b of Figure 1 as corresponding to 1,2-BDT in a full-coverage limit. To check the surface coverage, a QCM study was performed. When gold-coated quartz was exposed to 1.0 × 10-3 M 1,2-BDT in benzene (the same as that used in obtaining the film corresponding to Figure 1a), the mass increase attained a plateau value, ca. 36 ng, within 20 min. In our experimental configuration, the latter value corresponds to the adsorption of 1.5 × 1014 molecules of 1,2-BDT onto the gold surface. The nearest Au atom spacing on a Au(111) plane is known to be ca. 3 Å.34 From a SCF/631G ab initio calculation,35 the distance of two sulfur atoms of 1,2-BDT is calculated to be 3.3 Å. Hence, assuming that the gold surface consists mainly of Au(111) planes, 1,2-benzenedithiolate can be viewed as binding with two adjacent gold atoms. According to the recent work of Schlenoff et al.,36 the roughness factors (R) of various gold substrates range from 1.1 to 2.1. Taking R ) 1.2, as assumed frequently for the evaporated gold film on mica,36-39 the number of adsorbed molecules should be 1.6 × 1014. The latter value is surprisingly close to the experimental value, 1.5 × 1014, which implies that roughness factor 1.2 is a reasonable value for gold (see Table 2). Although a number of assumptions are made, this is thought to indicate that 1,2-BDT should form complete monolayers on gold surface in at least a 10-3 M solution. A similar QCM experiment with silver-coated quartz demonstrated that a near complete monolayered film could be obtained on silver in ∼10-6 M solution while a multilayered film was formed eventually in a concentrated solution, i.e., at 10-3-10-5 M. Under a monolayer coverage limit, however, 1,2-benzenedithiolate appeared to be 15-20% more densely packed on the silver surface than on the gold surface. Taking into consideration that the basic spacings of the lattices of these two metal substrates are very similar, the silver substrate may be thought slightly more rough than the gold substrate. Overall, the QCM measurements seem to support that (34) Widrig, C. A.; Alves, C. A.; Porter, M. D. J. Am. Chem. Soc. 1991, 113, 2805-2810. (35) Ab initio quantum mechanical calculation was performed using Gaussian 94 program for Windows. (36) Schlenoff, J. B.; Li, M.; Ly, H. J. Am. Chem. Soc. 1995, 117, 12528-12536. (37) Alves, C. A.; Smith, E. L.; Porter, M. D. J. Am. Chem. Soc. 1992, 114, 1222-1227. (38) Rodriguez, J. F.; Mebrahtu, T.; Soriaga, M. P. J. Electroanal. Chem. 1987, 233, 283-289. (39) Deakin, M. R.; Melroy, O. J. Electroanal. Chem. 1988, 239, 321331.

the IRA spectra shown in parts a and b of Figure 1 should correspond to near complete monolayers of 1,2-BDT on gold and silver. Ellipsometry Measurement. On the basis of the ab initio calculation and the standard bond lengths reported,40 the thickness of perpendicularly oriented 1,2-benzenedithiolate on metal surface is estimated to be ca. 5.4 Å. Referring to the tilt angles of 1,2-BDT approximated from the IRA spectral pattern (ca. 51° on gold and 38° on silver), the thickness of 1,2-BDT on gold is then predicted to be around 3.3 Å and that on silver to be around 4.3 Å. It would be desirable to compare these values with ellipsometric measurements. However, due to the various uncertainties in the ellipsometry measurement (i.e., oxide formation and ambient contamination), a direct comparison of the measured thicknesses with the predicted ones may be controversial. To reduce these uncertainties, the substrates were handled under dry N2 atmosphere, and the bare substrates were exposed to either UV radiation41 or piranha solution before determining their refractive indices. Nonetheless, it was unfortunate that the solutions of the ellipsometry equations, i.e., refractive index and thickness, could not be obtained simultaneously. Hence, for the estimation of thickness a refractive index of 1.5 was assumed, taking into account that the refractive indices of organic films range usually in the region of 1.451.55.22,31,42-47 From the three-phase model, the ellipsometric parameters corresponded then to the thicknesses of 3.3 ( 0.5 and 4.9 ( 0.6 Å , respectively, for 1,2-BDT self-assembled for 1 h on gold in 1.0 × 10-3 M solution and on silver in 1.0 × 10-6 M solution. (See Table 2. When the measurement was performed 1 day after the substrate preparation without any pretreatment, the thickness measured was ∼10 times more scattered than that provided here.) Although the above precautions could not rule out the growth of oxide overlayer and other ambient contamination, the experimental thickness values agreed surprisingly with the predicted values within a 10% error limit. In this respect, the IRA spectral data seem consistent not only with the QCM data but also with the ellipsometric data. Binding Scheme. The binding scheme of organic thiols to metal varies, depending on the kind of metal surface. For instance, methanethiol has been reported to adsorb on Ag(111) perpendicularly,48 but on Au(111) methanethiol adsorbs with a tilted orientation.49 However, the exact nature of different binding schemes has not yet been elucidated clearly. Walczak et al.31 reported that the electron-accepting capability of the metal substrate should play an important role in determining adsorbate geometry. On the basis of an ab initio calculation, Sellers et al.32 suggested on the other hand that the thiolate on Au(111) can have two modes of chemisorption, one with sp3 hybridization at sulfur, giving a surface-S-C bond angle (40) Handbook of Chemistry and Physics, 60th ed.; Weast, R. C., Ed.; Chemical Rubber Co.: Boca Raton, FL, 1981. (41) Huang, J.; Dahlgren, D. A.; Hemminger, J. C. Langmuir 1994, 10, 626-628. (42) Nuzzo, R. G.; Fusco, F. A.; Allara, D. L. J. Am. Chem. Soc. 1987, 109, 2358-2368. (43) Allara, D. L.; Nuzzo, R. G. Langmuir 1985, 1, 45-52. (44) Chang, S.-C.; Chao, I.; Tao, Y.-T. J. Am. Chem. Soc. 1994, 116, 6792-6805. (45) Bain, C. D.; Troughton, E. B.; Tao, Y.-T.; Evall, J.; Whitesides, G. M.; Nuzzo, R. G. J. Am. Chem. Soc. 1989, 111, 321-335. (46) Tillmann, N.; Ulman, A.; Penner, T. L. Langmuir 1989, 5, 101111. (47) Evans, S. D.; Urankar, E.; Ulman, A.; Ferris, N. J. Am. Chem. Soc. 1991, 113, 4121-4131. (48) Harris, A. L.; Rothberg, L.; Dubois, L. H.; Levinos, N. J.; Dhar, L. Phys. Rev. Lett. 1990, 64, 2086-2089. (49) Nuzzo, R. G.; Zegarski, B. R.; Dubois, L. H. J. Am. Chem. Soc. 1987, 109, 733-740.

5834 Langmuir, Vol. 12, No. 24, 1996

of ∼104°, and another with sp hybridization at sulfur, giving a surface-S-C bond angle of ∼180°. The energy difference for binding at the 3-fold hollow and on-top sites was calculated to be 25.1 kJ/mol on Au(111) while 13.8 kJ/mol on Ag(111). In this respect, they proposed that n-alkanethiolate could have two binding modes, one with sp and the other with sp3 hybridization, on Au(111) but involving only 3-fold hollow sites. On Ag(111), however, n-alkanethiolate was proposed to adopt a covering scheme in which on-top sites were also involved so as to have higher packing density, smaller S‚‚‚S distance, and near perpendicular aliphatic chain orientation. Recently, Chang et al.44 reported that the compounds, 4′-alkoxybiphenyl-4-methanethiol (I), (6-alkoxynaphth-2-yl-methanethiol (II), and 4′-alkoxylbiphenyl-4-thiol (III), all formed well-ordered monolayers on Au and Ag with nearly identical hydrocarbon tail structures. This was ascribed to the same herringbone packing of the aromatic chromophores on nearly identical lattices of Au and Ag substrates. Since the compounds I and III appeared to have the same structure, the binding geometry of sulfur was supposed to be either sp3 or sp hybridized, depending on packing interactions in other parts of the molecules. At this moment, we are not certain whether the benzene ring of 1,2-benzenedithiolate is also packed into a herringbone-shape on gold and silver (In this respect, we plan to perform a molecular mechanics calculation as well as a STM/AFM measurement.) However, if such a packing is indeed very favorable even without any hydrocarbon chain, 1,2-BDT may have to take nearly identical orientations on both the gold and silver surfaces. Since the IRA spectral pattern is in fact susceptible to the kind of metal, we suppose that the orientation of adsorbed 1,2-benzenedithiolate is governed exclusively by the metal-sulfur bond(s). On the other hand, recalling that the orientation of a S-C bond of adsorbed 1,2-benzenedithiolate was closer to that of n-alkanethiolate, 1,2-BDT might be supposed to bind mainly at the hollow sites when adsorbing on gold, even though the on-top sites could also be involved when adsorbing on the silver surface. Since 1,2-BDT should bind to the gold and silver surfaces by forming two metalsulfur bonds simultaneously, the sulfur atoms of 1,2benzenedithiolate may adopt the sp3 hybridization rather than the sp hybridization. In order to get more detailed information on the adsorption behavior of 1,2-BDT, a series of IRA spectra were recorded after self-assembling for a predetermined period of time. On the gold surface, the IRA peak positions changed little as a function of immersion time. These results may imply that the intermolecular interaction on gold is not substantial. The peak intensities were observed to increase as a function of immersion time, but their relative intensities were kept nearly constant, suggesting that the orientation of adsorbed 1,2-BDT is barely affected by the surface coverage. In contrast with the case of gold, the IRA spectral pattern on silver varied as a function of immersion time. At an early stage of self-assembly, the ν 11 band was not observed at all while the ν19b band appeared distinctly, as shown in Figure 1d. After a certain period of self-assembly, the ν11 band appeared instantly with its intensity relative to that of the ν19b band being little different from what was observable at a full coverage limit; e.g., see Figure 1b. We could also notice the frequency changes for the ν19b and ν20 bands from 1437 and 3049 cm-1 to 1439 and 3055 cm-1, respectively. The present observations suggest that the adsorbate, being perpendicular at an early stage of adsorption, becomes tilted instantly at a certain surface coverage. Langmuir Isotherm and ∆Gao. Monolayer adsorption can be thought of as a surface site filling procedure,

Lee et al.

with the adsorption and desorption steps counteracting each other. Since the surface concentration of adsorbate is usually much smaller than the corresponding bulk concentration, the adsorption step can be thought to be rate-controlling. On this ground, we have confirmed previously26 that the self-assembly of small organics like 4-cyanobenzoic acid on silver could be described in terms of the Langmuir adsorption model. To differentiate the adsorption behavior of 1,2-BDT on two different surfaces, gold and silver, a similar analysis has thus been made in this work. If the model is applicable to the present system, the following equation should hold:24

dθ/dt ) (ka/No)c(1 - θ) - (kd/No)θ

(2)

where θ is the fractional surface coverage calculated from the normalized infrared peak intensity, t is the adsorption time, ka and kd are the adsorption and desorption rate constants, No is the surface adsorbate concentration at full coverage, and c is the solution concentration of adsorbate. Integrating the above equation with the initial condition θ ) 0 at t ) 0 results in

θ ) [kac/(kac + kd)]{1 - (exp[-(ka/No) (c + kd/ka)t])} (3) This equation reduces to the Langmuir adsorption isotherm at equilibrium, i.e., as t f ∞

θeq ) kac/(kac + kd) ) c/(c + κ)

(4)

where

κ ≡ kd/ka ∝ exp(∆Ga°/RT) and ∆Ga° is the free energy of adsorption at infinite dilution. Taking the inverse of eq 4, one can get a linear equation with a slope κ

1/θeq ) 1 + κ/c

(5)

To test the applicability of the above model to the adsorption of 1,2-BDT on gold, we have plotted the IRA intensity variation of the ν11 mode as a function of immersion time, as shown in Figure 3a. The IRA peak intensity (Imax) observable after 20 min of self-assembly at 2 × 10-3 M was hardly different from the observed value after 6 h at the same concentration. In agreement with the QCM data, the observed plateau at 2 × 10-3 M can be regarded as corresponding to a full-coverage limit, i.e., θ ) 1. The θeq at other concentrations was evaluated by taking the intensity ratio, i.e., I/Imax, observable after a 6 h self-assembly (the latter immersion time was chosen simply to make sure that equilibrium has been fully attained). Then, 1/θeq is plotted vs 1/c as shown in Figure 3b. The plot exhibits a linear relationship between the two values in accordance with eq 5. This reflects that the Langmuir adsorption isotherm is applicable to the present system. From the slope in Figure 3b, the value of κ is estimated to be 1.5 × 10-5 M. Using the latter value, an attempt was made to fit the data shown in Figure 3a. As a check for fitting, the obtained exponential constants (kobs) corresponding to (ka/No)(c + kd/ka) were plotted against c. The plot shown in Figure 3c also exhibits a linear relationship. The value of κ estimated from the ratio of intercept and slope is 1.7 × 10-5 M, in fair agreement with that obtained above. The latter value gives the free energy of adsorption (∆Ga°) to be -27.3 ( 0.4 kJ/mol. It is difficult to estimate the value of No without knowing the exact degree of surface roughness, but

Self-Assembly of 1,2-BDT on Au and Ag

Langmuir, Vol. 12, No. 24, 1996 5835

Figure 3. (a) Variation of peak intensity of ring 11 mode at 739 cm-1 in IRA spectrum of 1,2-BDT on gold as a function of self-assembling time. (b) Langmuir-adsorption isotherm plot of 1,2-BDT on gold; reciprocal of the fractional coverage vs reciprocal of the concentration of self-assembling solution. See text. (c) Exponential constant of eq 3 vs concentration of selfassembling solution for the data shown in (a). Table 3. ∆Gads, ka, and kd Values for 1,2-BDT/Benzene, 1-Octanethiol/n-Hexane, 1-Octadecanethiol/n-Hexane Systems adsorbate/adsorbent

∆Gads (kJ/mol)

ka (cm/s)

kd (mol/ (cm2 s))

1,2-BDT/Aga 1,2-BDT/Aua 1-octadecanethiol/Aub 1-octanethiol/Aub

-39.6 ( 0.2 -27.3 ( 0.4 -23.0 ( 1.7 -18.4 ( 0.8

1.2 × 10-1 2.2 × 10-4 2.2 × 10-3 7.4 × 10-4

1.5 × 10-11 3.6 × 10-12 1.5 × 10-10 3.8 × 10-10

a This work. Determined from concentration dependence of IRA peak intensities. See text. b Reference 50. Determined from QCM measurement. To present the ka and kd values in the above units, the No value was taken to be 9.1 × 10-10 mol/cm2, referring to ref 60.

referring to the QCM data (No ) 1.07 × 10-9 mol/cm2), the adsorption and desorption rate constants are estimated to be ka ) 2.2 × 10-4 cm/s and kd ) 3.6 × 10-12 mol/(cm2 s) (see Table 3). Recently, Karpovich and Blanchard50 reported QCM data on the adsorption of n-alkanethiols, i.e., n-octade-

canethiol and n-octanethiol, onto a gold surface. Over a limited concentration range, the adsorption behavior of those systems could be described by the Langmuir adsorption isotherm. On the basis of the data in hexane, they derived the free energy of adsorption, ∆Ga°, to be -23.0 ( 1.7 and -18.4 ( 0.8 kJ/mol, respectively, for n-C18H37SH and n-C8H17SH adsorbing onto gold. They noticed also that these values were virtually the same as the estimates of the aliphatic interchain interaction energies. Nonetheless, since the spontaneous formation of alkanethiol molecular assemblies was not observed in solution, the quantity of ∆Ga° they measured was rendered to be characteristic of only the thiolate-gold bond. On these grounds, the adsorption/desorption equlibrium behavior of n-alkanethiol on gold was proposed to proceed predominantly at the edges of monolayer islands and/or at step edges and defect sites. However, if one supposes that the difference in the values of ∆Ga°, -4.6 kJ/mol, between n-C18H37SH and n-C8H17SH arises solely from their different chain lengths, the free energy of formation of a Au-S bond is approximated to be -14.7 kJ/mol. Twice this value, -29.4 kJ/mol, is then noticed very close to the free energy of adsorption of 1,2-BDT on gold, -27.3 kJ/ mol. This may reflect that the ∆Ga° of 1,2-BDT we measure on gold is largely unaffected by the interaction between aromatic rings. Also noteworthy is that the formation of 1,2-BDT monolayers on gold seems to proceed over a much longer time scale in comparison with that of n-alkanethiol. As summarized in Table 3, the adsorption rate constant of 1,2-BDT adsorbing onto gold is 3-10 times smaller than those of n-alkanethiols adsorbing on gold. The desorption rate constant of 1,2-BDT is, on the other hand, more than 2 orders of magnitude smaller than those of n-alkanethiol. These results suggest that the activation barrier for the adsorption/desorption of 1,2-BDT onto/from gold should be substantially higher than that of nalkanethiol. To test the applicability of Langmuir adsorption model to self-assembly onto silver, we measured the intensity of the ν19b band as a function of self-assembling time in 10-610-7 M 1,2-BDT solution. In the case in which the outof-plane ν11 band was absent in the IRA spectrum, the intensity of the ν19b band we measured was scaled as much as by the tilt angle, 38°, of the adsorbate at a full-coverage limit. Figure 4a shows the peak intensity variation of the ν19b mode. We analyzed those data similarly as on the gold surface referring to the QCM data. Figure 4b represents a plot drawn for 1/θeq vs 1/c. The plot exhibits a linear relationship between the two values. This reflects that the Langmuir adsorption isotherm is also applicable to the self-assembly of 1,2-BDT on silver. From the slope in Figure 4b, the value of κ is estimated to be 1.1 × 10-7 M. Figure 4c shows a plot drawn for the exponential constants (kobs) corresponding to (ka/No)(c + kd/ka) vs c, obtained by fitting the data shown in Figure 4a using the above κ value. The plot in fact exhibits a linear relationship. The value of κ estimated from the ratio of intercept and slope is 1.3 × 10-7 M. The latter value gives the free energy of adsorption, ∆Ga° ) -39.6 ( 0.2 kJ/mol. The ∆Ga° value on silver is thus different from that on gold as much as by 12.3 kJ/mol. Referring to the QCM data on No, the adsorption and desorption rate constants are calculated to be ka ) 1.2 × 10-1 cm/s and kd ) 1.5 × 10-11 mol/(cm2 s) (see Table 3). From the ν19b band intensities, the structural change from vertical to tilted orientation was deemed to occur at a fractional surface coverage near 0.6. (50) Karpovich, D. S.; Blanchard, G. J. Langmuir 1994, 10, 33153322.

5836 Langmuir, Vol. 12, No. 24, 1996

Figure 4. (a) Variation of peak intensity of ring 19b mode near 1440 cm-1 in IRA spectrum of 1,2-BDT self-assembled on silver in dilute solutions (10-5 M), multilayers form readily on a silver surface. The IRA peaks grow continuously for up to 15 h. The Langmuir adsorption model can no longer be applied. When a QCM experiment was performed in 1.0 × 10-3 M solution, the mass was observed to increase by as much as 86 ng within 40 min. The latter value is more than twice the value required for the monolayer formation. When the ellipsometry was applied to a sample prepared in 1.0 × 10-4 M for 70 min, the thickness was measured to be 7.7 ( 0.9 Å. This value is also greater than the monolayer thickness. When a silver substrate was immersed in a 1.0 × 10-3 M solution for 10 days, the IRA spectral pattern was observed to be very similar to the TR spectrum of silver salt. This suggests that the multilayers on silver consist of aggregates of silver 1,2-benzenedithiolate. Monolayer vs multilayer film formation has been previously studied for a few systems such as xanthates, 3-mercaptopropionic acid, L-cysteine, etc. adsorbing onto gold, silver, and copper substrates.49,54-60 Multilayered film has been noticed to form readily on copper, but rarely on silver and gold substrates. When multilayered thiolate films are formed on gold, it has been supposed that the gold atoms are dissolved into the thiol-containing solution to ultimately form a complex on the gold surface or that the gold-containing thiol complex dissolved from the gold surface redeposits onto the gold surface.61,62 Similarly, (53) Tarlov, M. J. Langmuir 1992, 8, 80-89. (54) Ihs. A.; Uvdal, K.; Liedberg, B. Langmuir 1993, 9, 733-739. (55) Mielczarski, J. A.; Yoon, R. H. J. Phys. Chem. 1989, 93, 20342038. (56) Mielczarski, J.; Leppinen, J. Surf. Sci. 1987, 187, 526-538. (57) Ihs, A.; Liedberg, B. J. Colloid Interface Sci. 1991, 144, 282292. (58) Thomas, R. C.; Sun, Li. Crooks, R. M. Langmuir 1991, 7, 620622. (59) Kim, Y.-T.; McCarley, R. L.; Bard, A. J. Langmuir 1993, 9, 19411944. (60) Schneider, T. W.; Buttry, D. A. J. Am. Chem. Soc. 1993, 115, 12391-12397. (61) Edinger, K.; Go¨lzha¨user, A.; Demota, K.; W ll, Ch.; Grunze, M. Langmuir 1993, 9, 4-8. (62) McCarley, R. L.; Kim. Y.-T.; Bard, A. J. J. Phys. Chem. 1993, 97, 211-215.

Self-Assembly of 1,2-BDT on Au and Ag

Keller et al.63 recently proposed that the formation of multilayered thiolates on copper proceeded via dissolution of copper ions under the influence of hydrophobic thiols followed by a recrystallization process on the surface. In this respect, we wanted to know whether silver atoms are dissolved into the self-assembling medium prior to forming multilayers on the silver substrate. For this purpose, gold and silver substrates were immersed simultaneously into a concentrated 1,2-BDT solution for a prolonged time. It was expected that aggregates of Ag-thiolate should also be deposited on the gold surface if the species were present in the solution before anchoring to the silver substrate. Although multilayers were formed on the silver surface, however, only a monolayered film seemed to form on the gold surface. The IRA spectral pattern on gold was barely affected by the presence of silver substrate, nor could we observe any traces of silver atoms (ions) on the gold surfaces using X-ray photoelectron spectroscopy. The selfassembling medium was analyzed as not containing any silver atoms or ions from ICP measurement. All of these observations suggest that dissolution of silver ions into the solution may not be prerequisite to the build-up of multilayers on the silver surface. Hence, it is conjectured that Ag-thiolate complexes are formed only near the silver substrate and that, once formed, they aggregate immediately on the silver substrate without dissolution. Stability of 1,2-BDT Film. Finally, it should be mentioned that the 1,2-BDT films formed on both the gold and silver surfaces were very stable. Although Tarlov and Newman64 claimed that alkanethiolate self-assembled on a silver surface readily turned into sulfonate in the ambient condition, no IRA peak due to sulfonate was observed even after a self-assembled 1,2-BDT film was left for 1 month in the laboratory. The sulfonate peak was identified only when self-assembly was performed for several days under ambient conditions. It can be conjectured that the oxidation of 1,2-BDT occurs readily (63) Keller, H.; Simak, P.; Schrepp, W.; Dembowski, J. Thin Solid Films 1994, 244, 799-805. (64) Tarlov, M. J.; Newman, J. G. Langmuir 1992, 8, 1398-1405.

Langmuir, Vol. 12, No. 24, 1996 5837

in an air-containing solution but that the adsorbed 1,2BDT species is resistant to oxidation. Aromatic thiols would thus be beneficial in forming thin organic films on noble metals such as gold and silver. Conclusions The adsorption behavior of 1,2-BDT on gold and silver surfaces revealed by infrared spectroscopy was found to be consistent with quartz crystal microbalance and ellipsometry data. Self-adsorption of 1,2-BDT in benzene onto gold resulted in monolayered film, but either monolayer or multilayered film was formed on silver depending on self-assembling conditions. The molecule was adsorbed on both metals by forming two metal-sulfur bonds after deprotonation. On the gold surface the orientation of the adsorbate molecules was nearly invariant with respect to the surface coverage, while on the silver surface a structural change from vertical to tilted orientation occurred at a fractional surface coverage near 0.6. The tilt angles of aromatic thiolates were estimated to be close to those of aliphatic thiols. The self-assembling process forming monolayers could be described in terms of the Langmuir adsorption model. The multilayered film on silver appeared to consist of aggregates of randomly oriented silver thiolate. Its detailed formation mechanism could not be clarified, but the usual proposition that the dissolution of metal ions forms a metal-adsorbate complex in self-assembling medium followed by condensation on the solid substrate seemed not to be applicable to the present system. Acknowledgment. This collaborative work was supported in part by the Non Directed Research Fund of Korea Research Foundation (1995), the Specified Basic Research Fund (95-0501-09) and the Science Research Center Fund of Korea Science and Engineering Foundation, the Basic Science Research Fund of Ministry of Education, Republic of Korea (1996), and the Research Institute of Molecular Science, Seoul National University. Y.J.L. thanks Mr. Wansik Cha for ellipsometry measurement. LA9603131