Self-Assembly of Alkanethiol Monolayers on Ag− Au (111) Alloy

21124

J. Phys. Chem. B 2006, 110, 21124-21130

Self-Assembly of Alkanethiol Monolayers on Ag-Au(111) Alloy Surfaces Mitsuo Kawasaki* and Masaki Iino Department of Molecular Engineering, Graduate School of Engineering, Kyoto UniVersity, Katsura, Kyoto 615-8510, Japan ReceiVed: June 7, 2006; In Final Form: August 14, 2006

The self-assembly of ethanethiol (C2) and 1-octanethiol (C8) on Ag-Au(111) alloy films was studied by X-ray photoelectron spectroscopy (XPS), cyclic voltammetry (CV), and scanning tunneling microscopy (STM), to illuminate how the monolayer structures and chemisorption-induced substrate defect structures depend on the alloy composition. The thiolate packing density at saturation increased approximately linearly with increasing Ag ratio. The CV data for reductive desorption of thiolates evidenced predominant or major contributions of Ag atoms to the substrate-sulfur interactions for the alloy surfaces. The STM study supported the lack of elemental periodicity on Ag-Au(111) and the consequent absence of periodicity in substrate-sulfur bonding. For C8-covered films, we observed systematic changes of substrate defect structures from elevated monatomic islands on Ag(111) to vacancy island structure on Au(111), in good correlation with the reductive desorption characteristics. The former type of defects can be explained best in terms of breakup of atomic terraces under excess thiolate packing density for Ag(111) and Ag-rich Ag-Au(111). As for the vacancy island formation, the present results are not agreeable with the chemical etching model but compatible with the lattice relaxation model.

Introduction Fabrication of well-ordered organic monolayers on metal surfaces has many useful applications such as in corrosion protection,1 lubrication,2 and surface modification for building up a variety of bio-3 or photosensitive4 thin films on metal surfaces. Self-assembly of n-alkanethiols [CH3(CH2)n-1SH, Cn in short] on coinage metals,5 as the process yielding prototypes of such ordered monolayers, referred to as self-assembled monolayers (SAMs), has long been the focus of much of the fundamental research. In particular, Cn SAMs on Au(111) have so extensively been studied by many groups in different disciplines that there seems little that is left unknown with respect to the basic chemisorption kinetics as well as the detailed SAM structures of various possible phases, (x3×x3)R30°,6 3×2x3,7-9 3×4,10-12 and p×x3,13-16 which are all commensurate with the Au(111) lattice. The self-assembly of Cn on Ag(111) has not so extensively been studied as that on Au(111), but currently some detailed information is also available. One of the particular interests here lies in the fact that despite only a 0.3% difference in the (111) nearest neighbor spacing between Au and Ag, Cn SAMs on Ag(111) have been found to exhibit markedly different packing order and density as compared to those on Au(111). It is known that the SAM structures on Ag(111) follow either (x7×x7)R19.1° for very short Cn (n e 2)17-19 or incommensurate, substantially distorted (x7×x7)R19.1° for relatively long Cn (n g 4).20-22 The former structure can be constructed by binding, for example, one-third of the sulfur atoms on on-top sites and the rest on alternate hpc and fcc hollow sites of the Ag(111) lattice. The resulting thiolate packing density, 5.93 × 1014 cm-2, with 4.4 Å nearest neighbor spacing is ∼30% greater than that (4.6 × 1014 cm-2) expected for, for example, (x3×x3)R30° * Corresponding author. Tel./fax: (+81)-75-383-2574. [email protected].

E-mail:

typical on Au(111). However, this excess density causes strong alkyl-chain packing constraint for long Cn, and the consequent repulsive forces between the tail alkyl chains are believed to play a critical role in causing the aforementioned distortion of a (x7×x7)R19.1° structure. Our recent observation of a surprisingly stable mixed thiolate-chloride monolayer on Ag(111) for n g 8, with a well-defined 2:1 S/Cl atomic ratio to yield a mixed x7×x7 adlayer ordering,23 could also be correlated intimately with this tight alkyl-chain packing in the (x7×x7)R19.1° structure. The large difference between the characters of Cn selfassembly on Au(111) and Ag(111) is thought as a clear manifestation of the fact that the ultimate SAM structures are likely determined by the two major forces: the substrate-sulfur interactions and the interchain van der Waals dispersion forces. Other relatively minor or subsidiary (by no means negligible) forces are associated with sulfur-sulfur interactions between the headgroups and the alkyl-substrate interactions. On Ag(111), the strong S-Ag bonds that are more ionic than S-Au bonds5,21,24 act in favor of the commensurate (x7×x7)R19.1° structure, but can significantly weaken at the same time the underlying Ag-Ag bonds, thereby enhancing the surface mobility of thiolates as thiol-silver complexes.21 Their consequent rearrangement aided by this increased mobility and driven by the interchain repulsive forces reasonably accounts for the substantial distortion from a commensurate (x7×x7)R19.1° structure. Importantly, Cn self-assembly on Ag(111) gives rise to another striking feature that is diametrically different from the monatomic depressions (pits) or vacancy islands that have been commonly observed for SAM-covered Au(111). In the case of Ag(111), the SAM-covered surface exhibits a large number of “elevated” monatomic silver islands, 20-180 Å across.21,23 As for the pit-like defects on Au(111), they were initially interpreted as the consequence of chemical etching of Au(111) in al-

10.1021/jp0635213 CCC: $33.50 © 2006 American Chemical Society Published on Web 09/19/2006

Self-Assembly of Alkanethiol Monolayers kanethiol solution,25 but a more elaborate model was later proposed by Poirier26 in terms of a chemisorption-induced relaxation of the compressed herringbone reconstruction of the Au(111) surface. Nevertheless, it seems that this issue of chemisorption-induced substrate defects for SAM-covered Au(111) still remains to be clarified. This is also the case with the mechanism of the elevated silver island formation on Ag(111). Given that all of the remarkable differences between Au(111) and Ag(111) as the substrates for Cn self-assembly can be traced back primarily to the difference in the nature of substrate-sulfur bonding, it seems highly useful to study what the self-assembly is like on the (111) surface consisting of both Au and Ag atoms, that is, Ag-Au(111) alloy surfaces with the surface atomic composition systematically altered. To our knowledge, no such attempts seem to have been reported previously, probably because of the problem associated with the control of the surface atomic composition. In this paper, we first show that our DC sputtering method, previously used to grow atomically flat Au(111) and Ag(111) films,27,28 can also yield high-quality Ag-Au(111) alloy films by means of simple co-sputtering of Ag and Au targets. The resultant alloy films are supported to have homogeneous atomic mixing without any substantial surface segregation, despite that an Ag surface segregation is to be expected theoretically under equilibrium conditions.29,30 The main purpose of this study then is to clarify at what or in what range of alloy composition the SAM-covered surface exhibits transition from one way to another. This offers a unique opportunity to link the various observations previously reported separately for Au(111) and Ag(111), and to reilluminate some of the still arguable issues such as the mechanisms of defect island formation. The results further support the major contributions of Ag atoms to the substratesulfur interactions for the self-assembly of alkanethiols on AgAu(111) alloy surfaces. Experimental Section Atomically flat high-quality Au(111) films, ∼0.2 µm thick, were grown on freshly cleaved natural mica (heated to ∼300 °C) by using the simple DC glow-discharge sputtering method in an Ar atmosphere as described elsewhere.27,28 Ag(111) and Ag-Au(111) alloy films, ∼0.05 µm thick, were epitaxially overgrown on the thus prepared Au(111) film at somewhere between 180 and 300 °C depending on the desired alloy composition (the higher was the desired ratio of Ag, the lower was the temperature we chose). The choice of excess growth temperatures very often led to a foggy appearance of the films, indicating the onset of significant surface roughening. Control of the alloy composition was made by co-sputtering circular Ag and Au targets (with relative areas adjusted to match the desired alloy composition) stacked coaxially in the target position. The atomic compositions of Ag-Au(111) alloy films were determined by XPS (Shimadzu Corp., ESCA-750 spectrometer) with Mg KR radiation of 1253.6 eV for samples typically ∼6 × 6 mm2 in area, from the integrated intensities of Ag3d and Au4f signals by taking into account that the ratio of the overall sensitivity factors for these signals was 1.1 (Ag3d) to 1.0 (Au4f). The measurements were repeated at two different photoemission angles, 90° and 15°, where the average XPS analysis depth (inelastic mean free path of photoelectrons) is estimated to be ∼15 and ∼4 Å, respectively. These XPS data (Ag3d and Au4f signals) taken for largely different analysis depths allow one to infer the extent of surface segregation. n-Alkanethiols used in this study were limited to ethanethiol (C2) and 1-octhanethiol (C8). They were obtained from Wako

J. Phys. Chem. B, Vol. 110, No. 42, 2006 21125

Figure 1. Examples of 1.0 × 0.9 µm2 STM images showing extensively terraced surface morphologies of Ag-Au(111) films with various alloy compositions.

Pure Chemical Industry, Ltd. and used without further purification in the form of typically 1 mM solution in ethanol. C8 was chosen as the one with the longest chain length satisfying the requirement that the SAM structure could be molecularly resolved in our STM imaging conditions (see below). The asgrown metal substrate films were bathed in each alkanethiol solution typically for 10 min at room temperature, thoroughly rinsed with ethanol, and dried. The SAM-covered samples were subjected to combined surface analyses by XPS (see above), CV, and STM. The measurements were done both for multiple substrates with identical or very close alloy compositions and for different portions of one substrate. Some of the experimental data presented below include the results of such multiple measurements, while others include those that were representative of each alloy composition. The photoemission angle in the XPS analysis here was fixed at 90°. CV experiments were carried out in a standard three-electrode cell with a potentiostat (Hokuto Denko Corp., model HA-501) in combination with an external function generator (Hokuto Denko Corp., model HB-107A). STM images were taken in the ambient atmosphere by a Nanoscope I microscope (Digital Instruments Inc.) with a Pt/Ir tip. The microscope was operated under the constant current mode, with the sample bias of 400-600 mV (negative) and the tunneling current of ∼0.1 nA, which was the minimum setpoint current allowed to choose. C8 was about the longest-chain alkanethiol for which the SAM structure could be imaged under this instrumental limitation. Results and Discussion Characterization of Ag-Au(111) Films. The typical surface morphologies of the Ag-Au(111) films epitaxially grown on Au(111) are shown in Figure 1, where an extensively terraced surface texture, characteristic of the (111) surface of fcc metals, can be seen regardless of alloy composition. We found that the alloy compositions (hereafter expressed in terms of the Ag molar ratio; XAg), as determined from the low-emission-angle XPS data corresponding to ∼4 Å analysis depth (∼2 atomic layers), were indistinguishable from that based on the XPS measurement with ∼15 Å sampling depth. We believe that this comparison can reveal a deviation of the surface alloy composition, at least ∼0.05 in XAg, from the average for the first ∼10 atomic layers. It therefore seems that the alloy films prepared by the present method had a very low level (if any) of surface segregation. As was mentioned already, Ag surface segregation is expected theoretically under equilibrium conditions for Ag-Au alloy systems,29,30 but the growth of alloy films does not necessary proceed under equilibrium conditions. The lack of surface segregation in the present case is probably due to the relatively low growth temperatures for the alloy films and to the fact that the films were subjected to no deliberate annealing process in our sample preparation. Of course, the above result alone concerning the XPS-based average alloy composition does not necessarily guarantee the

21126 J. Phys. Chem. B, Vol. 110, No. 42, 2006

Kawasaki and Iino

Figure 2. Dependence of (a) valence band XPS spectra and (b) optical reflection spectra of Ag-Au(111) films on alloy composition. Figure 4. Sulfur/substrate XPS intensity ratio for C8-covered AgAu(111) versus alloy composition, suggesting an approximately linear increase of thiolate packing density with increasing Ag ratio.

Figure 3. (a) Series of O1s spectra taken for Ag-Au(111) films separated into two overlapping peaks from adventitious contaminants containing oxygen atoms (higher binding energy) and from native oxides (lower binding energy). (b) Relationship between the oxide signal intensity and the alloy composition.

complete alloying in the atomic level. To support good atomic mixing in the alloy films, we next measured the valence band XPS spectra and the optical reflection spectra of the films as functions of the XPS-based average alloy composition. As shown in Figure 2a, the 4d and 5s valence bands of Ag(111) (XAg ) 1.0) are clearly distinguishable from the 5d and 6s counterparts of Au(111) (XAg ) 0). The spectra for the alloy films then exhibited systematic and smooth changes between these two extremes according to the average alloy composition, indicating the formation of well-mixed valence bands. Of course, none of them could be described by composition-weighted average of the individual Ag and Au valence bands. A similar trend was confirmed also in the optical reflection spectra (Figure 2b) that are influenced by the alloy composition in the region far thicker than the XPS analysis depth underneath the film surface. A more convincing experimental support for not only the desired atomic mixing but also the minor surface segregation came from an XPS analysis of a native oxide (of Ag) monolayer on Ag-Au(111). The oxygen signal in our XPS measurement stemmed partly from inevitable hydrocarbon adsorbates containing oxygen atoms (from our laboratory atmosphere or from the residual gas in the XPS analysis chamber). However, because the corresponding binding energy was considerably higher than that associated with the surface metal oxide species, the two overlapping peaks could be easily separated from each other. Figure 3a shows some typical O1s spectra taken for films with different alloy compositions. Here, the peak separation into the above-noted two overlapping peaks is also demonstrated. As expected, the surface oxide signal steadily decreased with decreasing Ag ratio. Moreover, the plot of the oxide signal intensity (as separated in Figure 3a) as a function of alloy composition (Figure 3b) revealed a striking feature. The oxide

Figure 5. Collections of single-scan CV curves recorded at 1 V/s and in 1 M KOH for (a) C8-covered and (b) C2-covered Ag-Au(111) films, showing systematic changes of reductive desorption peaks with alloy composition. The curves were given vertical offsets for visual clarification.

Figure 6. Model lattices of circles for SAM-covered Ag-Au(111) alloy surfaces with two typical alloy compositions. White circles represent Au atoms, light gray circles represent Ag atoms, and thick open circles represent S atoms arranged orderly according to a (x3×x3)R30° structure.

formation became negligibly small not only for Au(111) but also for Au-rich alloy films in the range of composition with the upper bound of XAg very close to 1/3. The result shown in Figure 3b is quite reasonable, however, given that the surface atomic composition is indeed identical to the average value as determined by XPS (i.e., negligible surface segregation), and that Ag and Au atoms are uniformly mixed in the (111) surface lattice. By using a 2D lattice of circles (cf., Figure 6a), one can easily confirm that in these conditions the composition XAg ) 1/3 agrees with the upper limit of Ag ratio that allows every Ag atom on the surface to be totally surrounded by Au atoms in the nearest neighbor positions in

Self-Assembly of Alkanethiol Monolayers

J. Phys. Chem. B, Vol. 110, No. 42, 2006 21127

Figure 7. Examples of high-resolution (10 × 9 nm2) STM images of C8 (top) and C2 (bottom) SAMs on Ag-Au(111). SAMs on alloy films rarely produced molecularly resolved STM images with long-ranged ordered patterns.

the 2D lattice plane. This means that the surface Ag atoms are so isolated that stoichiometric silver oxide species, Ag2O, no longer form on the surface. It should be also stressed here that the native oxide monolayer on the relatively Ag-rich alloy films is easily and totally eliminated during the self-assembly of alkanethiols. As was recently suggested by Kondoh et al., the surface oxide layer may even increase the rate of thiol chemisorption.31 XPS and CV Studies of SAMs on Ag-Au(111). In the case of good planar Ag-Au(111) substrates, as shown in Figure 1, the intensity of S2p XPS signal relative to that of the sensitivitycorrected substrate signal provides convenient information about the molecular packing density of Cn SAMs.23,32 Figure 4 shows the plot of this relative intensity of S2p signal for C8 SAMs as a function of alloy composition. Figure 4 suggests that the thiolate packing density on Ag-Au(111) increased approximately linearly with increasing Ag ratio in the alloy film. A similar relationship was observed also for C2 SAMs with much shorter chain length, although in this case the XPS analysis gave not so much reliable data due to an X-ray induced partial desorption and to a local corrosive reaction (for Ag-rich films) beyond the simple chemisorption scheme.5,31 It should be also noted that, in Figure 4, while the S2p intensity on Au(111) at the left end of the plot was consistent with the thiolate packing density corresponding to a (x3×x3)R30° (or other equivalents in packing density) structure on Au(111), the one on Ag(111) at the right end had a tendency to exceed (by up to ∼6%) even the limiting high density corresponding to a commensurate (x7×x7)R19.1° structure. This is certainly not consistent with the C8 SAM structure (distorted (x7×x7)R19.1°) suggested by the STM image presented later (Figure 7). One possible reason for this behavior is that epitaxial (111) films grown with a higher Ag ratio may have had some rougher surface morphologies in part to increase the net surface area for thiolate chemisorption. Note that our XPS data represent the average over the considerably large sample projective area, ∼5 mm in diameter. Overall, Figure 4 proves that regardless of alloy composition our Cn SAMs grown on Ag-Au(111) had thiolate packing densities no less than that typically found on Au(111). Figure 5 shows the SAMs’ reductive desorption behaviors as studied by the CV method. The result uncovers a very unique dependence of the substrate-sulfur interactions on the alloy composition. In Figure 5, a series of single-scan CV curves recorded at 1 V/s in 1 M KOH supporting electrolyte are presented separately for (a) C8 and (b) C2 SAMs on various substrates with systematically varying alloy compositions. The choice of relatively fast scan and high concentration of KOH

extended the negative end of potential window (not seriously affected by the solvent reduction) to -1.5 V vs Ag|AgCl|3 M NaCl. When one carefully and precisely controls the effective sample (working electrode) area in contact with the electrolyte, the integrated area of the reduction peak in such a CV curve may also provide useful information about the thiolate packing density. In our experiment, however, we did not take so much care of the working area, adjusted roughly to ∼2 × 5 mm2. Our main interest in Figure 5 is solely with respect to how the position and shape of the reduction peak are altered according to the alloy composition. It can be seen that the reductive desorption peaks for C8 SAMs (Figure 5a) are shifted as a whole to more negative potential by ∼0.2 V relative to the C2 counterparts (Figure 5b). This unanimous shift is not due to a change in the reversible reduction potential, but reflects the difference in the rate at which ions can move through the SAMs to the electrode surface to support the reduction reaction.33 Densely packed C8 SAMs naturally have a higher capability to block such an ion flux to the electrode surface than do C2 SAMs, so as to cause the reductive desorption at more negative potential. Figure 5a and b otherwise exhibits very similar dependence of the reduction peaks on the alloy composition. It is therefore believed that this compositional dependence primarily reflects some systematic changes of substrate-sulfur interactions. In Figure 5, when we first focus on the CV curves for Aurich alloy films, it can be seen that alloying of Au with a relatively minor ratio of Ag (XAg ≈ 0.2) noticeably shifted and significantly extended the desorption peak to more negative potential. In the region of XAg ) 0.2-0.4, then, a second additional peak clearly showed up at a considerably more negative potential. This is due most probably to the contribution of substrate-sulfur interactions mainly involving Ag atoms. Although the metal-sulfur binding energy may depend on the specific binding site of the S atom, the higher thiolate binding energy involving S-Ag bonds than that based on S-Au bonds can be justified also by, for example, a recent theoretical study by Cometto et al.34 Furthermore, the initial peak on the positive side, representing the contribution of Au atoms, faded out at XAg ≈ 0.5 already. These features testify to a very strong influence of Ag atoms on the substrate-sulfur interactions for the self-assembly on Ag-Au(111). This is further strengthened by looking at the desorption peaks for Ag-rich alloy films. As we go from XAg ) 1.0 to XAg ≈ 0.5, there seems only a minor effect of alloying of Ag with Au atoms, except that the reduction peak noticeably shifted to an even more negative potential (the opposite to what one would expect if the mixing with Au atoms

21128 J. Phys. Chem. B, Vol. 110, No. 42, 2006 tends to weaken the substrate-sulfur interactions on the average due to the above-noted smaller binding energy associated with S-Au bonds). Because this commonly occurred both for C8 and for C2 SAMs, the alkyl chains have probably no relevance. The origin of this unique shift will be discussed later. To explain the results shown in Figure 5 in more detail, it is helpful to refer to the simple model lattices of circles as shown in Figure 6a and b, which are for XAg ≈ 0.2 and for XAg ≈ 0.5, respectively. Here, Ag and Au atoms were arranged randomly in the (111) lattice, so as to remove the elemental periodicity on the surface. This is the most likely situation on the (111) surface of Ag-Au alloy that is an isomorphous binary system where metals are completely soluble in each other at an arbitrary composition. Figure 6a also helps one visualize the concept of “isolated” Ag atoms in the (111) lattice for the range of alloy composition, XAg e 1/3. We already suggested that surface oxide species no longer form once this condition for the surface Ag atoms is satisfied (Figure 3). In Figure 6, all of the S atoms (represented by large open circles) were arranged orderly on hollow sites with a packing density identical to that for a (x3×x3)R30° structure for simplicity. This does not properly represent, therefore, the increased thiolate packing density with increasing Ag ratio (Figure 4). Also, the choice of hollow sites for the sulfur binding sites on such an alloy surface has no particular justification. Figure 6a suggests that even in this Au-rich alloy phase a considerable fraction of the S atoms are forced to have partial coordination to Ag atoms. This accounts for the gradual shift and extension of the reductive desorption peak to more negative potential with increasing Ag ratio in the Au-rich regime (Figure 5). The point made in Figure 6b is that the contribution of S-Ag bonds may increase dramatically at XAg ≈ 0.5 already. This is consistent with the result (Figure 5) that the reductive desorption peak analogous in position to that for Ag(111) predominated in some wide rage of alloy composition down to XAg ≈ 0.5. We have also pointed out that the reduction peak for Agrich alloy films exhibited an unexpected shift to even more negative potential than that for Ag(111). The reason for this interesting behavior is not necessarily clear at present, however. It might be related to the excess thiolate packing density on Ag(111) that does not allow all of the S atoms to occupy equivalent lattice sites of highest binding energy. For Ag-rich alloy films, the thiolate packing becomes less dense than on Ag(111), so that more S atoms may look for stronger biding sites. It may be also possible that the presence of Au atoms (more electronegative than Ag atoms) makes adjacent S-Ag bonds more ionic and thus stronger. STM Studies of SAMs on Ag-Au(111). Figure 7 shows a collection of narrow-scan (10 nm across) STM images taken for C8 and C2 SAMs on various substrates with different alloy compositions. On Au(111) and Ag(111), both C8 and C2 SAMs were quite easy to image with good enough (often excellent) molecular resolution to infer in what possible packing patterns they are ordered. According to the corresponding images presented in Figure 7, the structures of C8-Au(111), C2-Au(111), C8-Ag(111), and C2-Ag(111) can be identified with 3×2x3, 3×4, distorted (x7×x7)R19.1°, and (x7×x7)R19.1°, respectively. On Ag-Au(111), however, the chance to acquire such molecularly resolved images was extremely reduced. In Figure 7, images that fortunately exhibited some periodic structures on the SAM-covered alloy films are selected, but even in such images and, besides, in such a small scan window, the fraction of ordered or quasi-ordered portion was still limited. This is not due to the relatively short SAM incubation time (10 min)

Kawasaki and Iino

Figure 8. Series of wide-scan (150 × 135 nm2) STM images of C8covered Ag-Au(111) films showing how defect island structures make the transition from elevated type on Ag(111) to vacancy type on Au(111) with alloy composition.

that we typically allowed in the present study, as it was sufficient for the self-assembly to complete on the Au(111) and Ag(111) surfaces, while a much longer incubation of, for example, 10 h did not yet improve the above situation on the alloy surfaces. The results thus strongly suggest that the structures of SAMs on Ag-Au(111) alloy surfaces lose at least the long-ranged periodicity necessary to observe well-ordered patterns in the STM image, despite the sufficiently high thiolate packing densities (see Figure 4) achieved regardless of the alloy composition. This can be rationalized by referring again to the models presented in Figure 6, where there is no perfect elemental periodicity with respect to the substrate atoms on (111) alloy surface. As a result, the S atoms, no matter how they are ordered, also lose the periodicity in terms of their binding sites. With the real thiolate packing density on the alloy films, significantly greater than that for a x3×x3 structure, the situation becomes even worse. Overall, one can thus hardly expect to observe wellordered patterns in the STM image. Next, Figure 8 presents a series of wide-scan STM images for C8-covered samples that uncover a unique way of transition of the defect island structure from the elevated type for Ag(111) to the vacancy type for Au(111). Looking at these images from the Ag-rich side, the image for Ag(111) reproduces a number of elevated monatomic islands as already reported in previous STM studies.21,23 Interestingly, alloying of Ag with a relatively minor ratio of Au (XAg ) 0.82) resulted in a marked increase in the number density of these elevated islands. The same type of defect island structure persisted for some wide range of XAg down to ∼0.4. Note that this range agrees well with that in Figure 5a where the reduction peaks were located at the position representing dominant contributions of Ag atoms to the substrate-sulfur interactions. Furthermore, in the image for XAg ) 0.4, near the lowest Ag ratio to give rise to the elevated island pattern, one can see some considerable increase in the island size, together with more irregular shapes of such bigger islands and also of terrace edges. This feature strongly suggests that the elevated islands most likely formed by breakup of atomic terraces. The elevated type of defect island structure then disappeared certainly at XAg ≈ 0.2 in Figure 8. Note that a number of depressed features here, producing somewhat fuzzy image contrast, were less than 0.2 nm deep, and thus represent not the vacancy islands but a large number of domain boundaries accompanying the C8 SAM for this alloy composition. The same features are still visible in the highly Au-rich regime of XAg ≈ 0.05, where the presence of real vacancy islands is now evident. However, the occurrence of vacancy islands at this alloy composition was not yet so frequent as in the case of C8-covered

Self-Assembly of Alkanethiol Monolayers Au(111). In other words, Figure 8 suggests that the well-known vacancy island structure for SAM-covered Au(111) may be eliminated by alloying of Au with only a minor ratio of Ag. Here, the effect of chain length on the defect island formation should also be noted briefly. According to our previous work,23 the elevated monatomic islands on SAM-covered Ag(111) began to show up with C3, and an equally large number of islands formed for C4 and longer alkanethiols. Thus, in the case of C2 SAMs studied in the present work, such island structures were almost totally absent. This was also true for Ag-Au(111) alloy films. We also found that the vacancy island formation was a minor event for C2-covered Au(111). These observations suggest an important additional role of alkyl chains for the defect island formation regardless of the alloy composition. As was mentioned already, in the case Ag(111) and Ag-rich Ag-Au(111), the strongly ionic S-Ag bonds can significantly weaken the underlying and in-plane Ag-Ag bonds (or in the case of alloy films, Au-Ag bonds). In this situation, coupled with the interchain repulsive forces in the tight packing of relatively long (C8 in the present case) alkyl chains, some fraction of Ag atoms bound to the S atoms might be forced out of the surface to nucleate into elevated monatomic islands. However, this process should also leave vacancies at the same time, contrary to the experimental results. The alternative model that we already suggested from Figure 8, that is, breakup of atomic terraces into small islands, easily solves this problem, as it naturally leaves no such vacancy structures. It was also shown in Figure 8 that, at a considerably Ag-rich alloy composition (XAg ≈ 0.8), the number of elevated islands became particularly large. This suggests that the extent to which S-Ag bonds weaken the nearby metal-metal bonds may become somehow larger by mixing with Au atoms, so that atomic terraces more easily break apart. With further increase of Au ratio, however, the thiolate packing density decreases, and so do the interchain repulsive forces that drive the breakup of atomic terraces. This naturally tends to decrease the island density, and what then follows is partial terrace breakup leaving much bigger, irregularly shaped islands, as seen at XAg ) 0.4 in Figure 8. As for the vacancy islands, Figure 8 suggests that they tend to be eliminated by mixing of Au with a minor ratio of Ag (XAg < 0.1). There seems no rationale for this fact to be agreeable with the chemical etching model. In the alternative model proposed by Poirer,26 excess Au atoms associated with the compressed herringbone reconstruction of Au(111) are forced out of the surface by the relaxation caused by thiolate chemisorption. The resultant migrating adatoms are then absorbed at the terrace edges, while vacancies nucleate to stable islands. Although we have no sure evidence,35 the compressed herringbone reconstruction may also be relaxed easily by alloying of Au with a minor ratio of Ag. If so, our observation is indeed compatible with the lattice relaxation model. It may be also worth noting that atomic terraces in this model (sink for adatoms) play a role analogous to that in the model (breakup of atomic terraces) for elevated island formation, in the sense that they function to leave only one type (either vacancy or elevated) of defect islands. Summary and Conclusions Atomically flat, high-quality Ag-Au(111) films with arbitrary alloy compositions could be epitaxially grown on Au(111) by co-sputtering Ag and Au targets. The films exhibited no substantial surface segregation, and good atomic mixing was supported by the valence band XPS spectra, optical reflection

J. Phys. Chem. B, Vol. 110, No. 42, 2006 21129 spectra, and in particular the extent of monolayer oxide formation examined as a function of alloy composition. We prepared C2 and C8 SAMs on these alloy surfaces and studied how systematically the SAM structures and SAM-induced substrate defect structures change with alloy composition. The thiolate packing density increased approximately linearly with increasing Ag ratio. The reductive desorption behaviors of the SAMs studied by the CV method uncovered major contributions of Ag atoms to the substrate-sulfur interactions for the self-assembly on Ag-Au(111) alloy surfaces. The highresolution STM imaging of the SAM-covered Ag-Au(111) seldom yielded well-ordered patterns, supporting that Ag-Au(111) surfaces have no elemental periodicity and the sulfur atoms bound to such surfaces also lose periodicity at least with respect to their binding sites. The SAM-induced substrate defect structures, as studied also by STM, changed systematically with alloy composition, in good correlation with the reductive desorption characteristics. Elevated monatomic islands predominated for films in some wide range of alloy composition from XAg ) 1 to ∼0.4, whereas vacancy islands were observed only for Au(111) and for highly Au-rich alloy films. The formation of elevated islands is explained best in terms of breakup of atomic terraces aided by the weakening of nearby metal-metal bonds due to the strongly ionic S-Ag bonds, and by the interchain repulsive forces associated with the excess thiolate packing density for Ag(111) and Ag-rich Ag-Au(111). As for the mechanism of vacancy island formation, the present result is not agreeable with the chemical etching model, but compatible with the model of chemisorption-induced relaxation of the compressed herringbone reconstruction of Au(111) if the alloying of Au with a minor ratio of Ag can also relax it. References and Notes (1) Hutt, D. V.; Liu, C. Appl. Surf. Sci. 2005, 252, 400. (2) Zuo, L.; Xiong, Y.; Xie, X. C.; Xiao, X. D. J. Phys. Chem. B 2005, 109, 22971. (3) Maxwell, D. J.; Taylor, J. R.; Nie, S. J. Am. Chem. Soc. 2002, 124, 9606. (4) Kawasaki, M.; Sato, T.; Yoshimoto, T. Langmuir 2000, 16, 5409. (5) 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. (6) Dubois, L. H.; Zegarski, B. R.; Nuzzo, R. G. J. Chem. Phys. 1993, 98, 678. (7) Camillone, N.; Chidsey, C. E.; D.; Liu, G.-Y.; Scoles, G. J. Chem. Phys. 1993, 98, 3503. (8) Delamarche, E.; Michel, B.; Cerber, Ch.; Anselmetti, D.; Gu¨ntherodt, H. H.; Wolf, H.; Ringsdorf, H. Langmuir 1994, 10, 2869. (9) Poirier, G. E.; Tarlov, M. J. Langmuir 1994, 10, 2853. (10) Kondoh, H.; Nozoye, H. J. Phys. Chem. B 1999, 103, 2585. (11) Hagenstro¨m, H.; Schneeweiss, M. A.; Kolb, D. M. Langmuir 1999, 15, 2435. (12) Kawasaki, M.; Nagayama, H. Chem. Lett. 2001, 942. (13) Camillone, N.; Eisenberger, P.; Leung, T. Y. B.; Schwartz, P.; Scoles, G.; Poirier, G. E.; Tarlov, M. J. J. Chem. Phys. 1994, 101, 11031. (14) Poirier, G. E.; Tarlov, M. J.; Rushmeier, H. E. Langmuir 1994, 10, 3383. (15) Kang, J.; Rowntree, P. A. Langmuir 1996, 12, 2813. (16) Truong, K. D.; Rowntree, P. A. J. Phys. Chem. 1996, 100, 19917. (17) Schwaha, K.; Spencer, N. D.; Lambert, R. M. Surf. Sci. 1979, 81, 273. (18) Rovida, G.; Pratesi, F. Surf. Sci. 1981, 104, 609. (19) Heinz, R.; Rabe, J. P. Langmuir 1995, 11, 506. (20) Fenter, P.; Eisenberger, P.; Li, J.; Camillone, N., III; Bernasek, S.; Scoles, G.; Ramanarayanan, T. A.; Liang, K. S. Langmuir 1991, 7, 2013. (21) Dhirani, A.; Hines, M. A.; Fisher, A. J.; Ismail, O.; Guyot-Sionnest, P. Langmuir 1995, 11, 2609. (22) Mohtat, N.; Byloos, M.; Soucy, M.; Morin, S.; Morin, M. J. Electroanal. Chem. 2000, 484, 120. (23) Kawasaki, M.; Nagayama, H. Surf. Sci. 2004, 549, 237. (24) Sellers, H.; Ulman, A.; Shnidman, Y.; Eilers, J. E. J. Am. Chem. Soc. 1993, 115, 9389.

21130 J. Phys. Chem. B, Vol. 110, No. 42, 2006 (25) Sondag-Huethorst, J. A. M.; Schonenberger, C.; Fokkink, L. G. J. J. Phys. Chem. 1994, 98, 6826. (26) Poirier, G. E. Langmuir 1997, 13, 2019. (27) Kawasaki, M.; Uchiki, H. Surf. Sci. 1997, 388, L1121. (28) Kawasaki, M. Appl. Surf. Sci. 1998, 135, 115. (29) Meinel, K.; Klaua, M.; Bethge, H. Phys. Status Solidi A 1988, 106, 133. (30) Wang, H. Y.; Najafabadi, R.; Srolovitz, D. J. Interface Sci. 1993, 1, 7. (31) Kondoh, H.; Nambu, A.; Ehara, Y.; Matusi, F.; Yokoyama, T.; Ohta, T. J. Phys. Chem. B 2004, 108, 12946.

Kawasaki and Iino (32) Kawasaki, M.; Sato, T.; Tanaka, T.; Takao, K. Langmuir 2000, 16, 1719. (33) Widrig, C. A.; Chung, C.; Poter, D. J. Electroanal. Chem. 1991, 310, 335. (34) Cometto, F. P.; Paredes-Olivera, P.; Macagno, V. A.; Patrito, E. M. J. Phys. Chem. B 2005, 109, 21737. (35) We have not been able to resolve any reconstruction patterns for Ag-Au(111) alloy surfaces by our in-air STM imaging. Nevertheless, this does not yet definitely prove the absence of such reconstructions because the chance to resolve the herringbone patterns in air was not so high (although certainly imaged occasionally) even for the Au(111) surface.