Supporting Information for J. Am. Chem. Soc., 1991, 113(19), , DOI: /ja00019a011 LAIBINIS

Transcription

1 Supporting Information for J. Am. Chem. Soc., 1991, 113(19), , DOI: /ja00019a011 LAIBINIS I Terms& Conditions Electronic Supporting Information files are available without a subscription to ACS Web Editions. The American Chemical Society holds a copyright ownership interest in any copyrightable Supporting Information. Files available from the ACS website may be downloaded for personal use only. Users are not otherwise permitted to reproduce, republish, redistribute, or sell any Supporting Information from the ACS website, either in whole or in part, in either machine-readable form or any other form without permission from the American Chemical Society. For permission to reproduce, republish and redistribute this material, requesters must process their own requests via the RightsLink permission system. Information about how to use the RightsLink permission system can be found at Copyright 1991 American Chemical Society

2 J - 7/6> 7-/n/ R esults Effect of Solvent on M onolayers on Copper, Silver and Gold. Table I summarizes the wettabilities of surfaces of the three metals exposed to 1 mm octadecanethiol solutions in a range of solvents for 2 to 4 h. A general discussion of these values can be found in the Results section of the paper. U tility of Passivating Agents in Determ ining E llipsom etrk Thicknesses. In the paper, we describe our difficulty in obtaining reproducible values of thicknesses by ellipsometry of monolayers on silver and copper. As oxidation could cause the scatter we observed, we attempted a procedure that minimized exposure of the unfunctionalized substrates to oxygen and environmental contaminants. This problem has been noted previously in a study of monolayers derived from alkanoic acids on alumina.1 We attempted to adapt the solution to this problem used in this earlier study by adsorbing a passivating agent that was capable of slowing oxidation, was not easily displaced by airborne contaminants, but would be readily displaced by a longer chain adsorbate. Bain et al. have shown that a propanethiolate monolayer on gold exposed to a 1 mm ethanolic solution of octadecanethiol reached a limiting ellip some trie thickness after -1 h that corresponded to the difference in thicknesses found for monolayers prepared by exposure of clean gold to either alkanethiol.2 Because our functionalization procedure required manipulation of solutions outside a fume hood, our choice of a higher molecular weight alkanethiol hexanethiol ~ was a compromise between its volatility and the rate that is was displaced by longer alkanethiols. The hexanethiolate monolayers on silver and copper, assembled by transfer of the unfunctionalized metal surface under flowing argon to a solution containing hexanethiol, were characterized by ellipsometry, subsequently immersed in 1 mm solutions of isooctane containing various alkanethiols (C6-C22X and periodically characterized. The experiments on silver and copper did not result in films of a limiting thickness; the thickness of the films obtained were, like those whose exposure to air was not minimized, thicker than could be explained for a simple monolayer coverage. The wetting properties of the films were, however, consistent with formation of a densely packed, methyl surface; by XPS, no oxygen could be detected in the films.

3 J - 7/ i, 7 - / K 7 P - 3 An increased signal due to sulfur was, however., detected by XPS. This observations suggests that sulfiding of the metal surfaces had occurredand is responsible for the high values of ellipsometric thicknesses we obtained. Investigation of Film Structure and Molecular Orientation by IR Vibrational Spectroscopy 1. Spectra of High- and Low-Quality Monolayers. Measurements were carried out on monolayers made from a series of long chain /z-alkanethiols (C16-C20 and C22) on the surfaces of copper, silver, and gold. We have noted in the paper that the variability in the samples prepared on copper was large; Figure 1 shows representative limiting high- and low-quality spectra for a Cis thiolate monolayer. The two spectra differ both in terms of their relative intensities and in the shapes of the bands. Significant differences in the positions of the C-H stretches are also evident. We found that our samples tended to range unpredictably between these extremes. The high quality samples that is, those giving spectra similar to that shown in the lower portion of the figure were only obtained, albeit sometimes irreproducibly, by excluding oxygen during the backfill of the metal deposition chamber and minimizing the duration of the exposure of the sample to the laboratory ambient prior to its immersion in the adsorbate solution. The top spectrum in Figure 1 is typical for a phase that contains significant conformational disorder (see below), while the bottom spectrum is of a monolayer containing more well-defined molecular orientations and higher conformational order. The nature of this latter structure is discussed in detail in the paper. The difference in the chemical nature of these phases is reflected to some degree by a broadening of the S(2p) envelope in XPS. 2. Relation Between Intensity of Methylene Modes (d+ and d-) and Chain Length. In the paper, we examined in detail the intensity of the methyl modes of the monolayers on silver and gold as a function of chain length. These modes exhibited odd-even variations on gold but not on silver. Figure 2 displays the intensity of the methylene modes (d+ and d ) as a function of the number of methylenes in the n-alkanethiolate monolayers on gold and silver. No odd-even

4 kj- 7/6 7~/y?3 variations are evident, suggesting, within the experimental uncertainty of the measurement, that there exists a constant chain orientation on the surface for these films, one that differs in magnitude for silver and gold. A very important feature of Figure 2 is that extrapolation of the intensity of the methylene modes suggests that carbon chain lengths significantly greater than zero exhibit zero spectral intensity (~Cs-Cn based on least-squares extrapolation). This is obviously a non-physical extrapolation in that zero intensity must occur at exactly zero CH2 groups in the alkyl chain. This behavior provides a graphic illustration of the fact that the vibrations of the CH2 groups are not independent (localized) and that the intensity arises from a coupled (delocalized) vibration of all CH2 groups. As a consequence, one cannot predict the spectrum for a Cn monolayer by simply scaling the intensity of a Cm chain by n/m. It also follows that scaling the intensities seen in spectra of different monolayers and derivatives could lead in principle to erroneous conclusions as to their structure. We note that the magnitude of the error bars shown in Figure 2 implicitly suggests that the variations in coverage that occur from sample to sample must be small (less than ~5%) for monolayers on silver and gold. Larger variations are seen on copper and could well reflect in part the variable coverages attained in different sample preparations. The largest experimental scatter for the measured intensities of these modes is for the d' (CH2 asym. str.) mode absorptions on gold. This scatter reflects the difficulty of separating the differing contributions of the overlapping Fermi resonance d+ and r+ bands. The calculations of chain geometries presented in the paper are based on complete spectral simulations and do not depend on measured peak heights. 3. Determination of Optical Functions. In this study, the determination of quantitative values of the average orientations of molecular groups on the surface involves analyses based on spectral simulations* To be valid, these analyses require both an accurate understanding of selected vibrational modes and reliable values of the optical functions of these modes as manifested in some physical state of the alkanethiols that serves as a close representation of the physical state of the alkyl chains as they are found in the monolayer. The reference materials selected as being most suitable for measurements of such optical functions are derivatives of the thiols: that is, the

5 J- 7/& 7-/ny corresponding polycrystalline dialkyl disulhdes. Since the C-S and S-H vibrations are not analyzed in this report, this substitution has no direct bearing on the quantitative analyses presented below. The higher melting points of the disulfide also have proven unexpectedly to be most important in minimizing non-uniform mixing in KBr dispersions, a factor we now regard as being the most serious source of error in analyses of these samples. The calculations presume that a correspondence exists between the optical functions in the bulk and monolayer phases, and imply a similarity in the molecular environments and packing densities in these phases. The character of the data (see below) suggest that a bulk crystalline reference phase provides a good first-order approximation to the phase state found in the monolayers. The method we have used to determine optical functions in this and earlier works3*4 involves quantitative measurement of transmission spectra using pressed KBr pellets of both known thickness and sample dispersion. We have found that, with low molecular weight compounds (such as some of the shorter thiol adsorbates used in this study) and materials with low melting points (<100 C), great care must be exercised in exactly how these KBr pellets are prepared. The experimental scatter of the chain tilt angles reported previously by us3 directly reflects the inherent scatter of this measurement protocol. In the current study we have greatly minimized this contribution by making numerous replications of the sample preparations and measurements for all the adsorbates studied. In fact, in excess of 100 data points were taken to determine the statistics of this scatter thoroughly. The optical functions used in the calculations presented immediately below are based on such averages. Based on our error analysis, the reported values of the absorption index k are accurate to ±5% or better. The assignment of the C-H stretching mode absorptions to specific frequencies was presented in the previous section. The quantity of interest for each of these modes is the complex optical function spectrum, n, defined in equation 1 n(v) = n(v) + ik(v) (1) where n(v) is the real refractive index spectrum, k(v) is the imaginary spectrum or the absorption index, and v is frequency. The quantities n and k are intrinsic to a given material and k represents the ability of the material to absorb light as a function of frequency. The k spectrum in the C-H

6 v j - stretching mode region for a bulk isotropic polycrystalline sample (determined from a KBr dispersion) o f dioctadecyl disulfide ((n-cigh37s)2) is given in Figure 3. The spectral envelope has been fit as a summation o f individual components, each assigned to a specific mode. The resolution of components in this spectrum serves as an example of those obtained for each o f the compounds involved in the present study.5 Although the underlying theoretical formalisms and accompanying assumptions are unchanged, we now use the matrix element o f the transition dipole to define the magnitude of the transition dipole and the direction for each mode to generate an n, k data set (eq 1) in the form o f a 3x3 tensor having numerical values of the tensor elements specific for a selected uniform orientation of the molecules on the surface plane. The reflection spectra are calculated using boundary value electromagnetic theory in a 4x4 matrix formalism capable o f handling full biaxial symmetry. A complete description o f this procedure will be published elsewhere.6 The chief advantage of this method to the current study is that the molecular geometries can be modified until the simulated and experimental spectra converge acceptably. It does not require that selected comparisons be made to specific band intensities. For reflectance spectra from a metallic substrate at infrared frequencies, reaching a convergence is relatively simple since only the electric field component perpendicular to the surface (the z direction) has any significant intensity. In this particular instance, therefore, only the zz elements o f the optical function tensor are o f interest and the optical tensor reduces to a simple scalar value, kzz.7 Such simplifications do not result for non-metallic or weakly metallic substrates and are not applicable at optical and UV frequencies.6

7 Footnotes and R eferences 1. Allara, D. L.; Nuzzo, R. G. Langmuir, ,1, Bain, C. D.; Troughton, E. B.; Tao, Y.- T.; Evall, J.; Whitesides, G. M.; Nuzzo, R. G. J. Am. Chem. Soc. 1989, 111, Nuzzo, R. G.; Dubois, L. H.; Allara, D. L. J. Am. Chem. Soc ,112, Allara, D. L.; Nuzzo, R. G. Langmuir, ,1, Tables of the optical functions for all the alkanethiols reported in this study (C16-C20 and C22) are available on request (D.L. A.). 6. Allara, D. L.; Parikh, A. N. Manuscript in preparation. 7. In this context kzz can be calculated from the relationship: n kzz(v) = ^ k ^ ( v ) (2) i = l k ^ (v ) = G i(a,(3)kj(v) (3) i where k^cv) is the k value for the ith vibrational mode for ni modes total contributing to the spectrum at frequency v, kj (v) is the k value intrinsic to the ith mode and associated with the direction of the transition moment vector, and G^o^f}) is the specific directional factor which relates the molecular coordinates of the ith mode to surface coordinates for a given orientation of the molecule. On a non-metallic surface, in the most general case, nine tensor elements need to be determined and G is additionally defined by an azimuthal coordinate. The components resolved in Figure 3 represent values o f k*/3 = kj(isob opic) since they are averaged over three directions. Details will be published elsewhere.6 It should also be noted that the theoretical plots in Figure 5 of the paper are essentially those of the geometric factor G^oCjp) vs. carbon number for the case where i refers to the CH3 C-K stretching modes. In these cases G is more complex than for the main chain CH2 modes since it also contains the geometrical relationships between the C-CH3 group and the CCC backbone plane.

8 TABLE I. Wetting Properties (0a, 0r) of Surfaces of Copper, Silver, and Gold after 2-4 Hour Exposure to 1 mm Octadecanethiol in Various Solvents [ - C11 1 r 1 r A p- Aii s 1 Solvent h 2o HD* h 2o HD h 2o HD Isooctane 119, , , , , , 39 CHCI3 117, , , , , , 38 THF 116, , , , ,, , 37 CH3 CN 119, 99 45, , , , , 39 Acetone 119, ,30 114, , , , 38 EtOH 120, 96 b 115, , , , 40 flhd = hexadecane. ^Reproducible values could not be obtained. The drop edge routinely became pinned and did not move smoothly across the surface.

9 ^ J - 7/6 7 - / n t f Figure Captions Figure 1. Extreme examples of IR spectra obtained from copper samples after exposure to octadecanethiol solutions. The upper spectrum is of a monolayer containing a significant density of gauche conformations; the lower spectrum is of a monolayer in which the chains adopt welldefined molecular orientations of extended trans- segments. The only difference in their preparation was the operating pressure of the evaporator, significant differences in hysteresis were noted: cosg^0 - cos0^2 = ~0.7 (for the top sample) vs. 0.3 (for the bottom). Figure 2. Absolute intensities of the methylene stretching modes, d+ and d, of various tt-alkanethiolate monolayers on gold (upper) and silver (lower). The size of the error bars are largely due to differences from samples prepared from three different laboratories. A least squares fit to each of the sets of data does not intersect zero intensity at zero methylenes (n = 0). The lines are provided not to imply that zero intensity occurs at n = 9, but rather to demonstrate that the intensities of the d+ and d modes exhibit a more complex, non-linear relationship to the number of methylenes (see text). Figure 3. The absorption index (k) spectrum of a representative adsorbate model phase, dioctadecyldisulfide, determined using a KBr dispersion. The spectral envelope has been fit as a summation of individual components (see text) and is used in the calculation of simulated spectra.

10 J- 7/G> 7- fin. f Figure o q: qc cn o WAVENUMBERS (cm"' 1 )

11 V/Co 7-/71/0 CH 3 (CH 2 ) n S/Au Fi igure 2 CO o X / T i.i'l 1.'1 $ -o* t ' / T' i 1 O DC CE 1»-1 r u> o CH 3 (CH 2 ) n S/Ag J' 1 ft T r 1 V 1 / / 0.0 -f 8 /. A -'-'f / ^ t 1 r n

12 if J- 7-sr)// Figure 3 WAVENUMBER (cm" 1 )