682
Langmuir 1990,6, 682-691
Chemical Functionality in Self -Assembled Monolayers: Structural and Electrochemical Properties Christopher E. D. Chidsey* and Dominic N. Loiacono AT&T Bell Laboratories, Murray Hill,New Jersey 07974 Received October 13, 1989 We examine the defectiveness and permeability of monolayers of long-chain organic molecules bound to gold electrodes via a sulfur attachment. The adsorbates, which were chosen to present a range of chemical groups at the outer surface of the monolayer, are a fluorinated alkanethiol (CF,(CF,),(CH,),SH), an unsubstituted alkanethiol (CH (CH,),SH), an alcohol-thiol (HOCH,(CH,)lgSH), a nitrilethiol (NC(CH,),,SH), and a carboxylic acib-thiol (HO,C(CH,),,SH). Both the defectiveness of the monolayers, as determined by electron transfer to acceptors in a contacting aqueous solution, and the permeability of the monolayers to aqueous ions, as determined by electrochemical capacitance measurements, correlate reasonably well with the quality of the packing of the carbon chains, which was determined by infrared spectroscopy, and with the relative cross sectional area of the chains compared to the other parts of the adsorbate. On the other hand, neither the permeability nor the defectiveness of the monolayers correlates well with the wettability of their surfaces. Thus, the fluorinated alkanethiol monolayer, with larger area chains, is the least permeable and the least defective, followed in order by the hydrophobic alkanethiol monolayer, the hydrophilic alcohol-thiol monolayer, the hydrophobic nitrilethiol monolayer, and finally by the hydrophilic acid-thiol monolayer.
I. Introduction Structurally well-defined organic monolayers on solid surfaces allow experimentalists to simplify and model a large variety of interfacial phenomena that are often difficult to study at "natural" interfaces due to heterogeneous or poorly defined structure. Organic disulfide^,'-^ thiols,5-" and sulfides" on gold surfaces and carboxylic acids13-15and ~ i l a n e s lon ~ -various ~~ oxide surfaces have been explored in recent years to find good models for such processes as wetting, chemical reaction at organic surfaces, and adhesion. In our laboratory, we are particularly interested in preparing well-defined models for studies of interfacial electron transfer between a metal electrode and a molecular electron donor or acceptor. (l)Nuzzo, R. G.: Allara, D. L. J . Am. Chem. SOC.1983, 105, 44814483. . (2) Nuzzo, R. G.; Zagarski, B. R.; Dubois, L. H. J . Am. Chem. SOC. 1987, 109, 733-740. (3) Nuzzo, R. G.: Fusco, F. A.; Allara, D. L. J . Am. Chem. SOC.1987, 109, 2358-2368. (4) Dubois, L. H.; Zegarski, B. R.; Nuzzo, R. G. Proc. Natl. Acad. Sei. U.S.A. 1987,84, 4739-4742. (5) Porter, M. D.; Bright, T. B.; Allara, D. L.; Chidsey, C. E. D. J . Am. Chem. SOC.1987,109, 3559-3568. (6) Finklea, H. 0.;Avery, S.; Lynch, M.; Furtsch, T. Langmuir 1987, 3,409-413. (7) Diem, T.; Czajka, B.; Weber, B.: Regen, S. L. J . Am. Chem. SOC. 1986,108,6094-6095. (8) Harris, A. L.; Chidsey, C. E. D.; Levinos, N. J.; Loiacono, D. N. Chem. Phys. Lett. 1987, 141, 350-356. (9) Bain, C. D.; Whitsides, G. M. Science 1988,240,62-63. Bain, C. D.; Whitesides, G. M. J . Am. Chem. SOC.1988, 110, 3665-3666. Bain, C. D.; Whitesides, G. M. J . Am. Chem. SOC.1988,110,5897-5898. Bain, C. D.; Whitesides, G. M. J . Am. Chem. SOC.1988,110, 6560-6561. (10) 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. (11) Nuzzo, R. G.; Dubois, L. H.; Allara, D. L. J . Am. Chem. Soc. 1990, 212, 558-569. (12) Troughton, E. B.; Bain, C. B.; Whitesides, G. M.; Nuzzo, R. G.; Allara, D. L.; Porter, M. D. Langmuir 1988, 4, 365-385. (13) Allara, D. L.; Nuzzo, R. G. Langmuir 1985,1, 45-52. (14) Allara, D. L.; Nuzzo, R. G. Langmuir 1985, 1, 52-66. (15) Moaz. R.: Saeiv. J. Lanemuir 1987.3,1034-1044. Moaz, R.: Saeiv, J. Langmuir 1987, d, 1045-1051. (16) Tillman, N.; Ulman, A.; Schildkraut. J. S.; Penner, T. L. J . Am. Chem. Soc. 1988, 110.6136-6144.
0743-7463/90/2406-0682$02.50/0
Alkanethiol monolayers adsorbed on gold make particularly convenient and versatile models for electrochemical s t ~ d i e s . ~ , ~ With ,'~,'~ functional groups a t their outer surface to which electron donors or acceptors could be attached, such monolayers would offer an ideal starting place in the construction of electron-transfer models. However, to be useful, the surface groups must not compromise the underlying structural integrity of the monolayer. In order to understand the role that various groups have in determining the structure of these monolayers, we report here a study of the structural and electrochemical properties of monolayers with very simple functional groups at their outer surface. Monolayers of unsubstituted n-alkanethiols are the natural starting point in this study because a significant amount of structural information is now available for these monolayers. From the frequencies of the peak positions in infrared spectra of the alkanethiol monolayers, Allara and co-workers have inferred that the polymethylene chains of thiols longer than about 10 carbons exist in a crystalline-like environment with fully extended chain^.^ From the intensity of the infrared absorption of the C-H stretching modes, the chain axes are calculated to be tilted 28-40' from the surface n~rrnal.~,"Strong and Whitesides have shown by electron diffraction that docosanethiol (CH,(CH,),,SH) forms a hexagonal array on the Au(ll1) surface with a lattice spacing of 4.97 A or an area of 21.4 A2 per chain,lg significantly more than the 18.4 A2 per chain in crystalline polyethylene." They also conclude that the docosanethiol lattice is not commensurate with the Au( 111) lattice. We have reexamined their electron diffraction pattern and conclude that the monolayer adopts the commensurate d 3 X d 3 R 3 0 " overlayer lattice, which places all adsorbates at identical, next-nearest-neighbor (17) Sabatani, E.; Rubinstein, I.; Moaz, R.; Sagiv, J. J . Electroanal.
Chem. 1987,219, 365-371.
(18)Sabatani, E.; Rubinstein, I. J. Phys. Chem. 1987,91,6663-6669. (19) Strong, L.; Whitesides, G. M. Langmuir 1988, 4, 546-558. (20) Wunderlich, B. Macromolecular Physics Vol. 1: Academic Press: New York, 1973; p 97.
0 1990 American Chemical Society
Chemical Functionality in Self-Assembled Monolayers (a) CH3(CH2),SH/Au f i x & R30'
(b) CF3(CF,h(CH2)2SH/AU 2x2
+
5.6 A
tC 5.76
A
t
cross-section
crass-sectian
side view
side view
Figure 1. Schematic illustration of possible chain packing i n self-assembled monolayers. See t e x t for details of models.
Table I. ThicknessaSband Contact Angles" of Thiol Monolayers on Gold
A compd
measured"
maximumb
B,(HD), deg
B,(H,O), deg
" Thickness measured ellipsometrically assuming a real refractive index of 1.33 for the fluorinated alkanethiol monolayer (measured bulk value) and a real refractive index of 1.45 for the other four monolayers.6 The standard deviations were less than f l 8, with from 3 to 10 samples. Maximum expected thickness of monolayer assuming the chain is fully extended and normal to the surface. Bond lengths and angles and Van der Waals radii from ref 44. Thiolate adsorption radius assumed to be 1.5 8, as in ref 10. e,, advancing contact angle; HD, n-hexadecane. Advancing contact angles less than 15' were irreproducible and decreased with time. Such samples were considered to be wet by the contacting liquid. The variation of the other contact angles averaged about f1' from sample t o sample with the exception of the advancing contact angle of water on the nitrile, which was quite variable.
*
sites on the Au(ll1) surface." Chidsey, in collaboration with Scoles and co-workers, has found by low-energy helium diffraction that the methyl groups in monolayers of both CH,(CH,),,SH and CH,(CH,),SH form ordered lattices with the expected 5-A spacing." From these results, we conclude that the structures of the nalkanethiol monolayers are similar from 10-carbon to 22carbon chains and that all probably adopt the v'3 X d3R30° overlayer structure. A simple model of the CH,(CH,),SH monolayer is illustrated in Figure la. The adsorbates are represented as hard cylinders bound to a surface. The adsorbate diameter is 4.6 A, which, a t dense packing, would give the same cross sectional area as found in crystalline polyethylene (18.4 A').'' The adsorbate length is from Table I of this paper. The lateral spacing of the adsorbates is 4.99 A, the next-nearest-neighbor Au(ll1) spacing. The cylinders are shown tilted 2 7 O , which is the most dense packing they can achieve with this simple hard cylinder model and with this lattice. (21) Strong and white side^'^ state that "... the principal reciprocal lattice vectors (of the monolayer) coincid(e) with the (110)8,(101), and (Oll), directions in the substrate ... Although the chain spacings are equal to the second nearest neighbor gold spacings ( d 3 ) , the monolayer lattice vectors lie in the same direction as the principal gold latThus the monolayer lattice is not strictly epitaxtice directions {llO}m. ial to that of gold lattice." This conclusion that the monolayer is not epitaxial is incorrect. Although the reciprocal space lattice vectors are coincident, the real space lattice vectors are 30' apart as required by the d 3 X d 3 R 3 O 0 structure. The confusion arises because the monolayer is indexed in a hexagonal lattice and the substrate in a cubic lattice. (22).Chidsey, C. E. D.; Liu, G.-Y.; Rowntree, P.; Scoles, G. J . Chem. Phys., In press.
Langmuir, Vol. 6, No. 3, 1990 683 Electrochemical measurements show that the longchain n-alkane monolayers are impermeable to ions in a contacting aqueous electrolyte solution for times of at least tens of second^.^ The films act like dielectrics of the expected thickness and dielectric constant. In addition, very low electron-transfer currents are observed with electron acceptors in the contacting solution. Although the monolayers may have thin regions where the electrontransfer rate is larger than average, the density of gross pinholes which can act as sites for electrochemicallyreversible electron transfer is very Due to the strength of the interaction of thiols with gold, many polar functional groups can be included in the adsorbate without changing the adsorption of the thiol group to a gold surface. w-Substituted alkanethiols have been used in several studies of wetting and adsorption at organic surfaces. The evidence to date suggests that, as expected, the terminal groups are located at the outer surface.*'' However, functional groups on the outer surface of the monolayer may perturb the structure of the monolayer in important ways. For instance, if the size of the functional groups sets the spacing of the chains, we could expect greater tilting than with the unsubstituted alkanethiols. The chains may even disorder if the spacing is too large. In contact with a condensed phase, functional groups on the outer surface of the monolayer may also affect the properties of the monolayer through their interaction with the other phase. In particular, we might naively expect monolayers that are wet by water to be more permeable to water and aqueous ions than hydrophobic monolayers. In the other extreme, we might imagine that some adsorbates would pack more densely, for instance, if their chain diameters more nearly matched a commensurate spacing of the substrate lattice. One of the thiol derivatives we examine in this paper is a highly fluorinated nalkanethiol. Figure l b shows our best guess for the structure of its monolayer on Au(ll1) by extrapolation from the structure of the n-alkanethiol monolayer. The chain diameter is taken as 5.6 A, which is the diameter of a perfluoropolymethylene chain in crystalline polytetrafluoroethylene." The spacing is assumed to be twice the gold spacing, which is the most dense commensurate overlayer possible for this chain diameter. The most dense packing in this lattice is achieved with a 16' chain tilt. Note that this tilt is significantly less and that the packing of the chains is significantly denser than in the n-alkanethiols on Au(ll1). With a slightly smaller, incommensurate lattice constant, the fluorinated chains would not need to tilt a t all, and hexagonal closest packing could be achieved. Such an alternate structure cannot be ruled out. In this paper, we explore the relationship of electrochemical permeability and defectiveness to monolayer structure and wettability. Simple functional groups are used to vary the cross sectional area of the adsorbate and the polarity of the surface. The five adsorbates examined are the unsubstituted alkanethiol (CH,(CH,),SH) and the fluorinated alkanethiol (CF,(CF,),(CH,),SH) (depicted in a and b of Figure 1, respectively), an alcohol-thiol (HOCH,(CH,),,SH), a nitrile-thiol (NC(CH,),,SH), and a carboxylic acid-thiol (H0,C(CH,),,SH). The chains are all 10 carbons long if we consider hydroxymethyl, cyano, and carboxy to be functional groups. We have chosen a 10-carbon chain for this study for two reasons. First, because 10 carbons is about the shortest chain length for which the high-quality packing is
684 Langmuir, Vol. 6, No. 3, 1990
Chidsey a n d Loiacono
Ag/AgCO/KCO REFERENCE ELECTRODE
ADJUST VOLUME ELECTRODE
Figure 2. Schematic diagram of drop cell used for electrochemical measurements on monolayer-covered electrodes.
mounted, with the gold surface up. The chamber and associated equipment were supplied with the Rame-Hart contact angle goniometer and modified for this use. After the air in the chamber is exchanged for argon, a glass tube with a tapered tip, containing the counter electrode and a modified Luggin capillary, is inserted from above through an O-ring seal. With argon escaping through the tube, the electrolyte solution and a reference electrode are inserted into the tube from above, and the top of the tube is sealed. The volume of the tube can be varied with an attached syringe to form a drop on the surface of the sample. With a narrow Luggin capillary that passes through the opening of the tip and into the drop, 10-pA currents in 0.1 M electrolytes cause only millivolt errors in the measured potential. The cells used for bare metals are of more conventional design with Viton or Teflon seals to define the area of the electrode. With either type of cell, current densities were calculated from the macroscopic electrode area, with no correction for surface roughness.
found in the n-alkanethiol~,~ we expect it to be the chain length at which the disruptive effects of surface functional groups are most prominent. Second, because 10 carbons is a reasonable length over which to expect electron tunneling,23 this length is relevant to our ultimate goal of exploring electron transfer between an electrode and a donor or acceptor.
111. Results and Discussion A. Ellipsometry, Maximum Expected Thickness, and Contact Angles. To establish that we have formed monolayers with the expected thickness and wetting properties, Table I lists the thicknesses measured by ellipsometry, the maximum expected thickness, and the advancing contact angles for water and hexadecane for the five thiol monolayers on gold. The thicknesses are near the 11. Experimental Section values expected for densely packed chains extended away Detailed descriptions of the reagents; syntheses; analyses; subfrom the surface,27as has been found pre~iously.~~'' strate preparation and characterization; monolayer preparaThe advancing contact angle data in Table I are all in tion; and ellipsometric, contact angle, and infrared measureclose agreement with previous measurements of the same ments are included in the supplementary material. nor similar The fluorinated alkane and Decanethiol was purchased from Aldrich, and the syntheses of unsubstituted alkanethiol monolayers have very low surthe other thiols were adapted from the methods of Bain et a1." Monolayers were formed on evaporated gold films on silicon face energies, showing high contact angles for water and wafers by adsorption from dilute solutions of the adsorbates in hexadecane. The alcohol and acid surfaces both wet with ethanol in a procedure similar to that reported earliere5 The both hexadecane and water, which can be attributed to substrates were characterized by Auger electron spectroscopy, both the polarity and hydrogen-bonding capability of the X-ray diffraction, and scanning tunneling m i c r o ~ c o p y(see ~ ~ * ~ ~ exposed alcohol and acid functional groups?"' The nitrile supplementary materials). Ellipsometry was performed with a surface is not wet by water despite the complete miscirotating-analyzer ellipsometer (Gaertner). Contact angle meability of acetonitrile with water and the high dipole moment surements were made with a Rame Hart goniometer. Infrared of the nitrile group. We have observed significant varispectra were acquired with a Digilab FX80 FTIR spectrometer ability in the contact angle of water on the nitrile surmodified for glancing angle reflection in a manner similar to face of tens of degrees, which may be due to slight changes that used by other^."^'^ A. Electrochemical Measurements. The electrochemiin the surface structure due to differences in sample hiscal measurements were made with a modified PAR 273 potentory and contaminants. tiostat interfaced to a personal computer. The digitally generB. Infrared Absorption. Infrared spectroscopy is a ated voltage sweeps were filtered before the control amplifier valuable probe of monolayer stru~ture.''~'~We focus here with a single-pole analog filter to provide smooth ramps for cyclic on the C-H and C-F stretching regions because they can voltammetry. The reference electrode was Ag/AgCl in satutell us about the packing and orientation of the adsorrated KCl solution (Ag/AgCl/KCl). The cell used for the monobate chains. Figures 3 and 4 show these two regions of layer samples was a "drop cell", which allows areas of the order the IR spectra of the five thiol monolayers and the five of 1 cm2 to be examined on a planar sample without the need bulk compounds from which the monolayers are formed. for mechanical contact that might otherwise damage the surface. It works well with both the hydrophobic and hydrophilic The left-hand scales in Figures 4 and 5 refer to the monomonolayer samples, because stable drops can be established.26 layer spectra (solid). The bulk spectra (dashed, rightI t does not work well for bare metals, on which the drop size hand scale) have been normalized by the sample thicktends to change as the potential is cycled. Figure 2 shows a ness (see supplementary material). The relative scaling schematic of the drop cell in use. I t consists of an evacuable of the monolayer and bulk spectra was chosen for ease sample chamber with glass windows in which t h e sample is first of visual comparison only; meaningful comparisons of the intensities of the two types of spectra require quantita(23) Oevering, H.; Paddon-Row, M. N.; Heppener, M.; Oliver, A. M.; Cotsaris, E.; Verhoeven, J. W.; Hush, N. S. J. Am. Chem. SOC.1987, 109,3258-3269. (24) X-ray data and STM images were acquired under conditions and with instruments similar to that used for the study of gold deposition on mica.26 (25) Chidsey, C. E. D.; Loiacono, D. N.; Sleator, T.; Nakahara, S. Surf. Sci. 1988, 200, 45-66. (26) The drop size is stable on hydrophilic monolayers despite the theoretical expectation that they should spread to cover the entire surface. We have not observed complete spreading of water on any dry surface. Drop areas are calculated from the drop diameter, which is measured by translating the cell past a fixed telescope with a micrometer drive. The electrochemical current densities are independent of drop size for drop diameters greater than about 0.5 cm.
(27) The acid-thiol monolayer often showed several angstroms greater film thickness than reported here unless the surface was rinsed with water immediately before measuring the thickness, probably due to contamination of the highly polar surface on standing in the laboratory air. To test for adsorbed water, the space around the sample in the ellipsometer was purged for several minutes. The acid-thiol monolayer showed a reversible increase in thickness of only 1A when water-saturated argon was substituted for dry argon. The thickness of the alcohol-thiol layer showed no dependence on the water content of the purge gas. In vacuum, water readily desorbs from similar surfaces well below room temperature.4 We conclude that, under ambient conditions, these surfaces are not coated with several angstroms of water. (28) Shafrin, E. G.; Zisman, W. A. J. Phys. Chem. 1960,64, 519-524.
Langmuir, Vol. 6, No. 3, 1990 685
Chemical Functionality in Self- Assembled Monolayers
2
\ P Y
lQ
X
0 v X
a
-v (cm-1) Figure 3. Infrared absorption spectra from 2700 to 3100 cm-' collected in reflection for the monolayers on gold (solid, lefthand axis) and collected in transmission and normalized by the effective sample thickness for the bulk compounds (dashed, righthand axis).
tive consideration of the optical and orientation effects a t a metal interface (see be lo^).'^,^^,^^ In the supplementary material, spectra of the 1800-3800-~m-~ region are reported along with a tabulation of the functional group modes in the 800-3800-cm-' region. The expected modes (such as the C=O, C=N, and 0-H stretches) are observed. 1. C-H Stretching Modes. The peaks near 2850 and 2920 cm-' in Figure 3 for the alkanethiol, alcohol-thiol, nitrile-thiol, and acid-thiol are the symmetric and asymmetric methylene stretches v,(CH,) and v,(CH,), respe~tively.~ These modes are apparently completely absent in the fluorinated alkanethiol, although, based on the two CH, groups in the compound, one might simplistically have expected the methylene peaks to be present a t about 20% the intensity that they have in the other compounds. As can be seen from the bulk spectrum of the fluorinated alkanethiol (dashed line), these modes are apparently broadened and shifted so much as to be unrecognizable even in the neat liquid. This effect may be due to the neighboring fluoro and thiol groups. The other peaks in the alkanethiol spectrum have previously been assigned to the symmetric methyl stretch, (29) Greenler, R. G. J. Chem. Phys. 1966,44, 31e315. (30)Allara, D. L.; Swalen, J. D. J. Phys. Chem. 1982, 86, 2700-2704.
vs(CH,), a t 2878 cm-', the Fermi resonance-enhanced overtone of the symmetric methyl bending mode, v,(CH,), a t 2938 cm-', and the asymmetric methyl stretch in the plane of the trans zig-zag carbon chain, v,(CH,,ip) a t 2965 ~ m - ' . ~ The alcohol-thiol monolayer shows a peak a t 2880 cm-', which may be due to a C-H stretching mode of the hydroxymethyl group. There is a shoulder in that vicinity in the bulk spectrum. Nuzzo and co-workers have seen a similar feature in some spectra of 16-mercaptohexadecanol on gold and speculate it is a C-H stretching mode of the methylene adjacent to the oxygen." Such shifts are known to occur for CH, groups adjacent to heteroatoms with localized nonbonding electron pairs.,' Table I1 summarizes the peak positions, peak widths, and peak areas for the two methylene modes of the four monolayers and bulk compounds that show those modes. We use this data to learn about the chain packing. In the bulk compounds, the peak positions and widths correlate with the physical state of the compounds. The liquids (the alkanethiol and the nitrile-thiol) have about 5-cm-l-higher peak positions and noticeably broader peaks than the solids (the alcohol-thiol and the acid-thiol). This correlation is well established for the polymethylene chains of numerous compounds32and has been turned to advantage in the interpretation of monolayer ~ p e c t r a .Follow~ ing that approach, we conclude that the alkanethiol, the alcohol-thiol, and the nitrile-thiol monolayers have nearly crystalline-like packing and the acid-thiol monolayer nearly liquid-like packing. The peak areas offer another important trend. The alcohol, nitrile, and acid monolayers all show the same peak areas to within the experimental error, but the alkane monolayer shows considerably smaller peak areas even allowing for its 10% fewer methylene groups. The simplest explanation for this difference is that the chains in the alkanethiol monolayer are tilted less than are the chains in the other three monolayers, and so the methylene stretching modes have smaller projections on the infrared electric field (which is perpendicular to the gold surface14s2') and thus show less absorption. 2. C-F Stretching Modes. We focus here on the peaks in the 900-1400-cm-' region of the fluorinated alkanethiol spectra (Figure 4), which can give us some insight into the packing and orientation of the fluorinated alkane chains. Note the substantial differences in the relative peak intensities between the monolayer and bulk spectra. Rabolt and co-workers have seen similar effects in the infrared spectra of Langmuir-Blodgett films of fluorinated fatty including the relative loss of intensity in the feature near 1210 cm-' and the appearance of two strong bands between 1300 and 1400 cm-'. Primarily on the basis of the assignment of the 1210-cm-l feature to a mode polarized along the chain axis,34they interpreted their spectra to indicate that the perfluorinated section of the alkyl chain in their Langmuir-Blodgett monolayer is substantially tilted with respect to the surface normal. We find this interpretation physically implausible. Furthermore, the assignment of the 1210-cm-' feature to a parallel mode is highly suspect. Although normal mode calculations suggest the parallel assignment, IR dichroism of stretched polytetrafluoroethylene films (31) Bellamy, L. J. TheZnfra-redSpectra of Complex Molecules; Wiley: New York. 1975. (32) Snyder, R. G.; Strauss, H. L.; Elliger, C. A. J. Phys. Chem. 1982, 86, 5145-5150. (33) Naselli, C.; Swalen, J. D.; Rabolt, J. F. J. Chem. Phys. 1989, 90, 3855-3860. (34) Masetti, G.; Cabassi, F.; Morelli, G.; Zerbi, G. Macromolecules 1973,6, 700-707.
686 Langmuir, Vol. 6, No. 3, 1990
Chidsey a n d Loiacono Table 11. Methylene Modes0Sb monolayer
i, cm-'
compd
4ifWhm, cm-'
bulk
peak area X lo2, cm-'
CH,(CHJ,SH HOCH2(CH,),oSH NC(CH,)i$H HO&(CHJ1oSH
2850.5 2850.5 2850.5 2853.0
12 18 18 21
v,(CH,): Symmetric Methylene 0.8 f 0.2 1.3 f 0.1 1.3 f 0.2 1.4 f 0.1
CH,(CHz)gSH HOCH,(CHJlOSH NC(CHJ1oSH HO,C(CH,)1oSH
2920.5 2920.0 2921.0 2925.0
19 22 24 26
v,(CH,): Asymmetric Methylene 3.2 f 0.3 4.8 f 0.8 4.3 f 0.5 4.0 f 0.1
i, cm-'
cm-'
A;,,,,
peak area X lo4, cm-'/8,
Stretch 2854.5 2850.5 2854.5 2850.0
17 12 19 10
2.4 f 0.1 2.4 f 0.1 2.1 1.8 f 0.1
Stretch 2925.0 2919.5 2927.0 2920.0
24 21 32 19
6.5 f 0.1 6.8 f 0.3 5.9 4.8 f 0.1
'Peak widths are full width a t half-maximum. Where necessary single peak profiles were estimated graphically. Peak areas are t h e width times the peak height. T h e errors are 1.0 cm-' or less in peak positions and 2 cm-' or less in peak widths. All errors are standard deviations of from two to four samples.
*
indicates that all three strong modes in this region are polarized perpendicular to the chain axis.34 An alternate explanation of both our spectra and those of Rabolt and co-workers is that the 1210-cm-l mode is polarized perpendicular, not parallel, to the chain and that the chain is only slightly tilted. To distinguish between these two possible orientations, we have calculated the tilt, 9, with respect to the surface normal, of the transition dipole moment of each of the modes a t 1150, 1210, and 1240 ern-' in our spectra. We use the dichroic r e l a t i ~ n , ~ ~ - , ~
3 cos29 =
Am
(1)
Eopdm(Ab/db)
where A , and A,, are the absorbance of the monolayer and bulk (isotropic) samples, respectively; d , and d , are the thickness of the monolayer and bulk samples, respectively; and E,, is the optical enhancement for the monolayer on a gold surface (17.1 for an angle of incidence of 85' in this region of the spectrum). 9 is calculated to be 76' and 74' for the modes at 1150 and 1240 cm-', respectively, and greater than 84' for the mode at 1210 cm-l. If the 1210-cm-' feature were assigned to a parallel mode, the chain tilt would be more than 84'. In addition to the geometrical implausibility of this structure, it is inconsistent with the spectroscopy. A t least one of the other modes would be expected to have a large projection on the surface normal, and neither do. On the other hand, if the 1210-cm-l as well as the 1150- and 1240-cm-' modes are polarized perpendicular to the chain, the chain tilt is only 15-16°.38,39 Using a similar analysis for the n-al(35) The ratio of the absorbance of the mode in the monolayer to the absorbance in a hypothetical, isotropic monolayer is 3 cos2 0, where 0 is the angle between the transition dipole moment and the electric field (which lies along the surface normal for glancing incidence with ppolarization)." The absorbance of the hypothetical, isotropic monolayer is taken to be the product of the monolayer thickness, the absorbance of the bulk sample per unit thickness, and the optical enhancement of the reflection experiment over the transmission experiment. The enhancement is obtained from the calculated normal incidence transmissitivity of free-standing films and the calculated grazing incidence reflectivity of substrate-supported thin films by using the matrix method.36 The following values were used in the calculations. For all modes: film thickness, 10 A; imaginary film index, 0.0 (transparent reference films) and 0.1 (absorbing films). For C-F modes: optical frequency, 1200 cm-'; real film index, 1.33; real and imaginary substrate indices, 9.0 and 48.0 re~pectively.~'For C-H modes: optical frequency, 3000 cm-'; real film index, 1.45; real and imaginary substrate indices, 2.0 and 21.0 re~pectively.~'The enhancement was averaged over grazing angles of incidence from 82.4' to 87.6' (the experimental range). (36) Allara, D. L.; Baca, A,; Pryde, C. A. Macromolecules 1978, 21, 1215-1220. (37) Lynch, D. W.; Hunter, W. R. In Handbook of Optical Constants of Solids; Palik, E. D., Ed.; Academic Press: New York, 1985.
kanethiol monolayer, we get a chain tilt of 27" from the absorption of the symmetric and asymmetric methylene stretches (E,? = 11.5).35-38 The surprising agreement of these chain tilt angles with those obtained by using the simple hard cylinder models (see Figure la,b) is certainly fortuitous. Even the more accurate approach of Allara and c o - w ~ r k e r s ~ does , ~ ' , ~not ~ give 1' accuracy. However, the qualitative result that the fluorocarbon chain is significantly less tilted than the hydrocarbon chain is u n a r n b i g u o u ~ . ~ ~ C. Electron Transfer to Electron Acceptors and Donors in Solution. The ability of a monolayer to block electron transfer between the gold surface and an electron donor or acceptor in solution is a useful measure of its defectivene~s.~A freely diffusing electron acceptor or donor in the contacting electrolyte solution will react exclusively at defect sites if electron transfer is blocked across the majority of the monolayer. Figure 5 shows cyclic voltammograms of monolayer-covered and "bare" gold electrodes in a solution containing 1mM Flu(",),Cl, a t pH 7 (solid) or in a solution containing 1mM K,Fe(CN), at pH 7 (dashed). The voltammograms of bare gold show the reduction and reoxidation peaks t,ypically seen for these redox couples at metal electrode^.^^ All of the monolayers inhibit these electron-transfer reactions to some extent; the monolayers block access to the electrode surface and so reduce the electron-transfer rates. As a result, more negative potentials are required to drive the reduction process to the diffusion At the (38) This result is obtained by assuming that the fluorinated chain has the same nearly planar, all-trans zig-zag carbon backbone as perf l u o r o o ~ t a n e(very ~ ~ long chains are helical"). In the planar case (and neglecting end groups), there are two seta of modes which are polarized perpendicular both to each other and to the chain axis. Assigning the 1150- and 1240-cm-' modes to one set and the 1210-cm-' mode to the other, we have cos2eehaic+ COS, e,,,, + COS^ e,,,, = 1 and can solve for the chain tilt, Beh.in. (39) Campos-Vallette, M.; ReyHLafon, M. J. Mol. Struct. 1983, 101,
-.,
'?.%A5_-.
(40) There remain two major unresolved aspects of the fluorocarbon spectra. The source of the degeneracy that gives two modes with apparently the same polarization (1150 and 1240 cm-') could be mixing of CF, bending and stretching modes. The origin of the intense modes in the monolayer spectrum between 1300 and 1400 cm-' could be C-F stretches that are IR allowed due only to the finite length of the chain as suggested by Rabolt and co-workers,8s but this assignment does not explain their absence in the bulk spectra. An alternate explanation is CF, modes,39which could be very intense in the monolayer due to interfacial enhancement of the oscillator strength." (41) The peak-to-peak separation for R u ( N H ~ ) ~ +indicates /~+ that the "bare" Au has some adventitious contamination from the ambient environment. No particular care was taken to maintain a clean gold surface.
Langmuir, Vol. 6, No. 3, 1990 687
Chemical Functionality in Self- Assembled Monolayers
0
-2
0 -50 0
0 n
m
J
-NC( CH,) 1oSH/Au 1 , I , , , 3
i2 1
2t
-200-,
,
I
I
0
0
l
l
1
0 14.
-200
0 1000
v (crn-l)
-----
1500
Figure 4. Infrared absorption spectra from 800 to 1800 cm-I collected in reflection for the monolayers on gold (solid, lefthand axis) and collected in transmission and normalized by the effective sample thickness for the bulk compounds (dashed,righthand axis). fluorinated alkane surface, electron transfer to ruthenium hexammine is blocked very effectively (note current scale). At this surface, the current for the reduction of ruthenium hexammine at its redox potential (about -0.12 V vs Ag/AgCl/KCl) is negligible compared with the capacitive charging current; the reduction current is at least 3 orders of magnitude lower than the current a t the same potential at bare gold. Because there is no detectable reduction current at Ellz,we conclude that the monolayer is essentially free of exposed pinholes.' The reduction current a t the fluorinated surface rises as the potential is made more negative because t h e increased exothermicity of the reaction increases the electrontransfer rate.5 However, the reduction currents at high exothermicity are about 1 order of magnitude higher a t the alkane and alcohol surfaces than at the fluorocarbon surface, indicating that nonpinhole defects are more numerous in these monolayers. A t the nitrile surface, ruthenium hexammine reduction is more facile still, and, a t (42) The blocking behavior of a sample does not change with repeated cycling within the region where low capacitances are measured (see section on capacitance), but the blocking behavior of different samples does show some variability, suggesting that inhomogeneities in the substrate surface can lead to defects in the monolayers.
----e-
-200
-0.5
"Bare" Au
0.0 +0.5 E(V) vs. Ag/AgCl/KCl
Figure 5. Cyclic voltammograms of the monolayer-covered and bare electrodes in 1 mM Ru(NH,),Cl, (solid) and 1 mM K,Fe(CN), (dashed). Supporting electrolyte is 0.1 M NaC10, buffered at pH 7 with 10 mM phosphate, scan rate 100 mV/s. the acid surface, it is almost diffusion-limited. The ferricyanide reduction currents show similar behavior, except that larger overpotentials are required to drive the same current with ferricyanide as the acceptor rather than ruthenium hexammine, indicating that the electron-transfer rates are intrinsically lower for this acceptor.' We attribute the difference seen above between the fluorinated alkanethiol monolayer and the alkanethiol and alcohol-thiol monolayers to the difference in the cross sectional area of their respective chains (see Figure 1). The polymethylene chain, which is tilted 28-40°,5J1 is likely to form defects at which the monolayer is notably thinner. As an example, let us assume that the tilted chains form domains, then there will be domain boundaries, and at some of those boundaries the chains will be tilted to either side resulting in a locally thin region of the monolayer and much higher electron-transfer rates. Finklea and co-workers have suggested this type of mechanism to explain the low slopes of Tafel plots for the reduction of hexaquo iron(II1) a t hexadecanethiolcoated gold surfaces., The perfluoropolymethylene chain,
688 Langmuir, Vol. 6, No. 3, 1990 on the other hand, is less tilted (see Figure l b and the discussion of the fluorocarbon infrared spectra) and so would form less defective domain boundaries. D. Cyclic Voltammetry in Simple Electrolytes and Electrochemical Capacitance. The permeation of aqueous ions into a self-assembled monolayer is another valuable measure of its integrity. Previous work has shown that n-alkanethiols with at least 10 carbon atoms in the chain are impermeable to a variety of aqueous ions at moderate electrochemical potentials on the time scale of tens of second^.^ The group of adsorbates considered in this paper gives us an opportunity to test the generality of this barrier behavior for monolayers with varied functional groups on the outer surface. Naively, one might expect that monolayers with hydrophilic outer surfaces, which clearly interact strongly with the aqueous electrolyte solutions, would be more permeable to aqueous ions than hydrophobic monolayers. Cyclic voltammetry gives ready access to information on ion permeation into the monolayers. If the monolayer is impermeable (acting as a simple dielectric), it behaves like an ideal capacitor with the capacitance given by the geometric relation
c = fddeff
(2) where C is the capacitance per unit area, t is the dielectric constant of the film, to is the permittivity of free space, and de, is the monolayer film t h i c k n e ~ s .There~ fore, during the linear potential scans of a cyclic voltammetry experiment, the charging current is independent of potential, proportional to scan rate, and inversely proportional to film thickness. If the monolayer is not impermeable, the capacitance and charging current will be larger. Cyclic voltammetry is a relatively slow technique for measuring capacitance and allows us to observe ion permeation even if it occurs over tens of seconds. At extreme electrochemical potentials, which result in large electric fields across the monolayers, ion permeation followed by electron-transfer chemistry at the metalmonolayer interface is likely. In this section, we report the capacitance of the monolayers at a moderate potential (0.0 V vs Ag/AgCl/KCl) and the negative and positive limits beyond which substantial ion permeation and subsequent chemistry occur. Figure 6 shows cyclic voltammograms for the alcohol monolayer on gold in contact with perchlorate electrolyte solutions of three different pH values. Initially, the potential was scanned from 0.0 V vs Ag/AgCl/KCl to -0.2 V to +0.2 V and back to 0.0 V at 100 mV/s (dotted, X10) and at 10 mV/s (dashed, X100). A t the positive end of these scans, the current is independent of potential and scales linearly with scan rate, as expected for an impermeable dielectric film. After the scans about zero, a scan to negative potentials was made to determine at what potential the barrier properties of the monolayer fail. At pH 10, for instance, a substantial increase in current is not noted until the potential is scanned down to -0.8 V. On the other hand, at pH 1,a substantial increase is observed at about -0.3 V. The onset of permeation at negative potentials and the pH dependence suggest that, in this case, the permeating ion is H+. The huge increase in the current at very negative potentials is presumably due to proton reduction. We have arbitrarily identified the potential at which a current of 10 pA/cm2 flows as the onset of proton reduction and the point at which the film no longer acts as a barrier. Similar experiments have been performed on all five monolayers and the results summarized in Figure 7 ,
Chidsey and Loiacono
,,,......
I.'
-5 -1.0
-0.5 0.0 E(V) vs. Ag/AgCl/KCI
Figure 6. Cyclic voltammograms of HOCH,(CH,),,SH/Au at moderate and negative potentials in perchlorate electrolytesolutions at three different pH values. The buffer for pH 7 was 10 mM phosphate, and for pH 10 it was 5 mM borate. The curves were acquired in the following order: the potential was cycled from 0.0 to -0.2 to +0.2 and back to 0.0 V vs Ag/AgCl/KCl at 100 mV/s (dotted, X10 vertical expansion);then the same cycle was repeated at 10 mV/s (dashed, X l O O vertical expansion). Finally, the potential was scanned negative from 0.0 V to observe barrier failure (solid; dotted, x10 vertical expansion).
I
l
,
,
t
,
L4
l
which shows the potential a t which a current of 10 PA/ cm2 is observed as a function of pH. The solid lines are plots of Nernstian behavior a t 25 "C and provide a visual guide showing that for each monolayer this potential displays a roughly Nernstian dependence on pH as expected for proton reduction. However, there are significant differences in the potentials at which the protons are reduced in the different monolayers, reflecting either different rates of proton permeation through the monolayers or different kinetics of hydrogen discharge. Proton reduction is most difficult a t the fluorinated alkane surface followed in approximate order by the alkane, then the alcohol, then the nitrile, and finally the acid. Figure 8 shows the current vs potential plots for cyclic potential scans to positive potentials and back to moderate potentials in 0.1 M HC10,. The scan for "bare" Au shows two gold oxide formation and one gold oxide reduction features, which are broader but at approximately the same potentials as observed for single-crystalAu(111).& This is expected for these gold films as X-ray diffraction
Langmuir, Vol. 6, No. 3, 1990 689
Chemical Functionality in Self- Assembled Monolayers
150 mV
120OmV
3-
1“
-200
200
scan rate (mV/s)
0
-200 -400 0.5
1.0
1.5
E(V) vs. Ag/AgCl/KCl
Figure 8. Cyclic voltammograms of the monolayer-covered and bare electrodes in 0.1 M HClO, at positive potentials (solid): scan rate 100 mV/s; vertical expansions as labeled (dotted).
shows them to have (111)texture (see supplementary material). These gold oxidation and rereduction features are suppressed by all of the monolayers to some extent. The fluorinated alkanethiol monolayer provides the best barrier to oxidation, followed by the alkane, the alcohol, the acid, and finally the nitrile monolayers. Unlike the failure of the monolayers as barriers a t negative potentials, the differences between the alkane, alcohol, nitrile, and acid a t positive potentials are fairly minor. The fluorinated alkane, however, is a much better barrier a t positive potentials than the other four monolayers. We know that the monolayers are damaged by these potential cycles because subsequent cycles to +1.5 V show less inhibition of the gold oxide features. The damage increases with both higher potential and longer time at high potentials. The onset of the anodic current increase is a t +1.2 to +1.3 V for all the monolayers, about 200 mV more positive than the onset of oxidation a t bare gold. It seems likely that this is the potential range in which the gold-sulfur interaction can be rapidly disrupted if aqueous ions can get to that interface. Because of the damage that cycles to positive potential cause, we do not believe that the area under the gold oxide reduction peak is a direct measure of the area occupied by pinholes before the oxidative cycle, although it may be a good measure of the area occupied immediately afterward. This conclusion is different than those reached by Finklea and co-workers6 for hexadecanethiol on gold in sulfuric acid and Rubinstein et al.17*18for mixtures of octadecyltrichlorosilane and octadecanethiol in gold in sulfuric acid. We speculate that the shorter chains used here desorb more readily a t positive potentials leading t o rapid monolayer degradation. We return now to the capacitance values obtained at moderate potentials. Figure 9 shows the capacitance values obtained from cyclic voltammetry measurements on the five monolayers in contact with 0.1 M NaF. Capacitance values were obtained at scan rates of 100 and 10 mV/s, using scans of A50 and A200 mV about 0.0 V vs (43) Angeratein-Kozlowska, H.; Conway, B. E.; Hamelin, A.; Stoicoviciu, L. J. ElectroanaL Chem. 1987, 228, 429-453.
Figure 9. Plot of capacitance of the monolayer-covered electrodes in 0.1 M NaF vs potential scan rate used to measure the capacitance for two different scan ranges (*50 and f200 mV). Top axis is the time required to scan from the initial potential to one extreme of the potential range. Right-hand axis is the effective dielectric thickness from eq 2 obtained by using the dielectric constant of polyethylene. The monolayers are as follows: (0) CF (CF,),(CH,),SH/Au; (0)CH (CH,),SH/Au; (A) HOCH,(CH$,,SH/A~; ( 0 ) N C ( C H , ) , , S ~ / A ~ ;(HI HO,C(CHJ,$H/Au*
Ag/AgCl/KCl. The fluorinated alkane shows the smallest capacitance, followed in turn by the alkane, the alcohol, the nitrile, and finally the acid. There is no time dependence in any of the data except for the data on the acid, which shows larger capacitances for both slower scan rates and larger scans, suggesting greater permeation in the scans requiring longer times. One indication of the time available for permeation is the time to scan from the central potential (0.0 V vs Ag/AgCl/KCl) to one of the extremes of the scan (k50 or A200 mV). We have called that time the “time-to-charge” in Figure 9. As can be seen, with the exception of the acid, there is no time dependence of the capacitance of these monolayers for times from 0.5 to 2 0 s. Several aqueous electrolytes were used to measure the capacitance because we anticipated that the permeability of some of the monolayers might be electrolyte dependent, as was found for the shorter n-alkanethiol monolayers on gold.5 In addition to 0.1 M NaF, capacitance measurements were made in 0.1 M NaC1,O.l M HClO,, 0.1 M NaC10, buffered a t pH 7 with 10 mM phosphate, and 0.1 M NaC10, buffered at pH 10 with 10 mM borate. The capacitance of the fluorinated alkanethiol, the alkanethiol, and the alcohol-thiol monolayers did not show any electrolyte dependence. The nitrile-thiol monolayer showed quite variable capacitance values (from 2.7 to 4.0 pC/cm2) but showed no consistent electrolyte dependence. The carboxylic acid-thiol monolayer had a noticeably higher capacitance in the neutral and basic perchlorate electrolytes, and the highest capacitance for the acidthiol monolayer was with the unbuffered sodium chloride electrolyte in which a value of 6.7 pC/cm2 was obtained for scans of f200 mV about 0 V vs Ag/AgCl/KCl a t 10 mV/s. E. Measures of Thickness. Using eq 2, we have calculated the effective thickness of dielectric, deff,that is required to obtain the capacitance values in Figure 9, assuming the dielectric constant to be the same as polyethylene-2.26.44 The right-hand axis of Figure 9 shows this thickness scale. Note that the fluorinated alkane(44) Handbook of Chemistry and Physics; Weast, R. C . , Ed.; CRC: Cleveland, OH, 1973.
690 Langmuir, Vol. 6, No. 3, 1990 thiol monolayer appears to be the thickest, with a thickness of 13 A. The thickness of the alkane and alcohol appears to be about 1 2 and 11 A, respectively, and the nitrile and acid appear to be at least 3 8, thinner yet. The difference between the values for the fluorinated alkane and the alkane can be roughly accounted for by the expected difference in dielectric constants for perfluoropolymethylene chains and polymethylene chains; the thickness of the fluorinated alkanethiol monolayer calculated with the dielectric constant of polytetrafluoroethylene (2.1)44 is 12 A. We can compare both the ellipsometric and dielectric measures of the thickness of the monolayers with our expectations. For instance, the actual physical thickness of the alkanethiol monolayer is probably about 14.2 A. This value is obtained from the length of the adsorbate (16.4 A from Table I) and the chain tilt (about 30°).5'" Thus the ellipsometric value is high. Similar problems with obtaining accurate thicknesses by ellipsometry have been noted Also, the dielectric thickness of 12 A, determined by the capacitance measurement, is low. This discrepancy is surely partly due to assuming the entire adsorbate molecule is well approximated by the dielectric constant of polyethylene. The sulfur head group of the molecule could have a very large polarizability and could effectively decrease the dielectric thickness of the monolayer by a few angstroms (the thiolate group represents about 20% of the expected molecular length in Table I). The neglect of surface roughness would also lead to lower dielectric thicknesses. The important point is that the dielectric, ellipsometric, and expected thicknesses are in rough agreement. As a result, it is reasonable to interpret differences in the dielectric thicknesses of the monolayers in terms of partial permeation by the electrolyte. Like the alkane and fluorinated alkanethiol monolayers, the alcohol-thiol adsorbate forms a monolayer that is free of pinholes, but its dielectric thickness is about 1A thinner, despite its extra hydroxymethyl group. It is reasonable to ascribe this result to hydration of the outer few angstroms of this hydrophilic monolayer, which "shorts out" the outer portion of the monolayer. In the case of the nitrile-thiol and the acid-thiol monolayers, ions or water penetrate many angstroms into the monolayer. For the acid-thiol monolayer, the penetration is deeper the longer the charging time and is also strongly electrolyte dependent. F. Barrier Failure. The barrier properties of all of the monolayers fail at very negative and very positive potentials in simple electrolytes. As an example, the lower limit is about -0.5 V vs Ag/AgCl/KCl for the alcoholthiol monolayer in 0.1 M HClO,, and the positive limit is about +1.3 V in the same electrolyte. A potential drop of at least half the difference, 0.9 V, exists across the monolayer at one or the other limit. From the dielectric thickness of about 11 A, we conclude that the alcoholthiol monolayer can support a field of the order of lo7 V/cm before failing. G. Defectiveness, Ionic Permeation, Wettability, and Chain Packing. There is not a good correlation between defectiveness or ionic permeation and wettability for these monolayers. In particular, the fluorinated alkanethiol, the alkanethiol, and the alcohol-thiol monolayers are pinhole-free and impermeable to aqueous ions for a t least tens of seconds, even though two are very hydrophobic and one is hydrophilic. Further, the nitrilethiol and the acid-thiol monolayers are both defective, even though one is hydrophobic and the other hydrophilic. The alcohol-thiol monolayer shows that it is pos-
Chidsey and Loiacono sible to obtain impermeable monolayers with hydrophilic surfaces, and the nitrile-thiol monolayer shows that hydrophobicity does not, in itself, lead to good barrier properties. What then is the origin of good barrier properties? The quality of the packing as judged by the infrared spectra correlates much better, though not perfectly, with the barrier properties. The fluorinated alkanethiol is the best barrier-preventing electron transfer to electron acceptors in a contacting electrolyte out to large overpotentials. This correlates with the infrared spectra, which indicate that the fluorocarbon chains are tilted very little, qualitatively in accord with the commensurate packing model in Figure lb. (Incommensurate packing could allow even denser packing, with no tilting a t all.) The alkanethiol and alcohol-thiol monolayers are also good barriers but require lower overpotentials for the reduction of electron acceptors. From the infrared spectra, the packing appears to be nearly crystalline but with significant tilt of the chains. This tilt appearently leads to thin regions and electron-transfer rates higher than with the fluorinated alkanethiol. The nitrile-thiol monolayer is somewhat problematic. The infrared spectrum indicate nearly crystalline packing, but the electrochemical properties are poor. It may be that the chains rearrange in contact with water. This monolayer also has the most variable wetting and permeation properties, suggesting that the conformation of the chains may be very variable. Finally, in agreement with the correlation of packing and barrier properties, the acid-thiol monolayer shows liquid-like packing in the infrared spectrum and exceedingly poor barrier properties. The length of the chain is also an important factor but has not been explored here. From other work, a comparison of the packing of two acid-terminated thiol monolayers with different chain lengths is possible. Nuzzo and co-workers report that the asymmetric methylene stretch in a monolayer of 16-mercaptohexadecanoic acid (H0,C(CH,),,SH) on gold occurs a t 1918 cm-', a crystallinelike value." In contrast, we have observed a liquid-like peak position for the 11-carbon homologue. This comparison suggests that the presence of the carboxylic acid group increases the minimum chain length at which crystalline-like packing can occur but that, with long enough chains, an impermeable acid-terminated surface might be obtained.
IV. Conclusions Using the monolayers formed on gold by five different thiols, each with 10-carbon chains, we have illustrated the importance of chain packing in determining the electrochemical barrier properties of self-assembled monolayers. In contrast, wettability is a relatively unimportant factor in determining permeability because wetting measures the interaction of water with only the top few angstroms of an organic s ~ r f a c e Thus, .~ substituting larger area perfluoropolymethylene chains for normal polymethylene chains allows dense packing of the chains without tilting and reduces the occurrence of defects at which the monolayer is a thinner barrier than average. Conversely, appending the polar carboxylic acid group to the end of normal polymethylene chains leads to less well packed chains and extensive permeation of the monolayer by water or aqueous ions. This effect is not due to the expected increase in wettability from adding polar surface functionality. For example, appending the hydroxymethyl group leads to a hydrophilic surface with well-packed chains and nearly as good barrier properties as the unsubstituted and hydrophobic alkanethiol mono-
Langmuir 1990,6, 691-694 layer. Furthermore, the nitrile surface is not hydrophilic, but is permeable. These results suggest the importance of steric and electrostatic interactions between appended functional groups in determining the packing of the chains. Recognizing the sensitivity of barrier properties to such effects, we now intend to take advantage of the synthetic scope of organic chemistry to control the placement of electroactive species a t the electrochemical interface and simultaneously to control the packing and hence barrier properties of the monolayer.
Acknowledgment. We thank Ralph Nuzzo for the use of his equipment, extensive technical advice, and valu-
691
able scientific discussions; Dennis Trevor for the use of his STM; and Tony Mujsce for acquiring GC-MS data. C.C. thanks Dave Allara for introducing him to reflection spectroscopy and Ian Robinson for discussions of electron diffraction.
Supplementary Material Available: Text giving details of the syntheses and analyses of the thiols and the procedures for the preparation and analysis of the gold substrates and the
monolayer films; Table S1, vibrational assignments, frequencies, and absorption intensities; Figure S1,an STM image of Au on Ti on Si; and Figure S2,infrared absorption spectra from 1800 to 3800 cm-' (14 pages). Ordering information is given on any current masthead page.
General Equations for Describing Diffusion on the Heterogeneous Surface at Finite Coverages V. Pereyra and G. Zgrablich Instituto de Investigaciones en Tecnologia Quimica, Universidad Nacional de Sun Luis, CONICET, Casilla de Correo 290, 5700 Sun Luis, Argentina
V. P. Zhdanov* Institute of Catalysis, Novosibirsk 630090, U S S R Received May 1, 1989. In Final Form: August 16, 1989 General equations are derived for describing diffusion on the heterogeneous surface in the framework of the lattice-gas model. The effect of the surface heterogeneity and lateral interactions between adsorbed particles on the coverage dependence of the chemical diffusion coefficient is discussed.
Introduction Surface diffusion is of considerable intrinsic interest and is also important for understanding the mechanism of surface reactions, ranging from the simplest, such as recombination of adsorbed particles, to the complex processes encountered in heterogeneous Its intrinsic interest arises from the dynamical and statistical features of particles in adsorbed overlayers. For this reason, surface diffusion has received the attention of many researchers in the last decade. In particular, the effect of lateral interactions between adsorbed particles on diffusion over the uniform surface a t finite coverages is considered in detail.2 Diffusion on the heterogeneous surface a t low coverages (the random-walk models) is also well ~ t u d i e d . Our ~ objective is to investigate diffusion on the heterogeneous surface at finite coverages. In this case, diffusion is simultaneously affected by the surface heterogeneity and by lateral interactions. Our attention is centered on periodic lattices with random transition rates. Transport processes in topologically disordered systems are not considered. (1)Ehrlich, G.; Stolt, K. Ann. Reo. Phys. Chem. 1980,31,603. Gomer, R. Vacuum 1983, 33, 537. Naumovets, A. G.; Vedula, Yu. S. Surface Sci. Rep. 1985,4, 365. Doll, J. D.; Voter, A. F. Ann. Reu. Phys. Chem. 1987,38,413. (2) Zhadanov, V. P.; Zamaraev, K. I. Usp. Fir.Nauk 1986, 149, 635 (English translation: Sou. Phys. Usp.1986, 29, 755). (3) Haus, J. W.; Kehr, K. W. Phys. Rep. 1987,150, 264.
0743-7463/90/2406-0691$02.50/0
A similar problem has been recently studied by using Monte Carlo simulations of self-diffusion4 and collective diffusion5 of particles without lateral interactions. In our paper, we discuss the coverage dependence of the chemical diffusion coefficient taking into account lateral interactions. Besides, in comparison with ref 4 and 5 , we use a more general expression connecting a saddle point energy with energies of nearest-neighbor sites (see eq 7). Our main assumptions are as follows. Adsorbed particles A are located in a two-dimensional array of surface sites. For clarity, we consider a square lattice. A single type of sites is used. A given site is either vacant or occupied by a single adsorbed particle. The surface heterogeneity is caused by the energy distribution of sites. Diffusion occurs via activated jumps of particles to nearestneighbor empty sites. Activated particles A* are located on the boundary between two nearest-neighbor sites. Two neighbor sites, occupied by an activated particle (i.e., by an activated complex), cannot be occupied by other particles. Lateral interactions between particles are taken into consideration. It is also assumed that activated complexes interact with neighboring adsorbed particles. Repulsive interactions are assigned positive values. More(4) Kirchheim, R.; Stolz, U. Acta Metall. 1987, 35, 281 and references therein. (5) Mak, C. H.; Anderson, H. C.; George, S. M. J. Chem. Phys. 1988,
88, 4052.
0 1990 American Chemical Society
