Adsorption of short-chain alkanethiols on Ag(1 1 1) studied by direct recoiling spectroscopy

Surface Science 600 (2006) 2305–2316 www.elsevier.com/locate/susc

Adsorption of short-chain alkanethiols on Ag(1 1 1) studied by direct recoiling spectroscopy L.M. Rodrı´guez a, J.E. Gayone a,*, E.A. Sa´nchez a, H. Ascolani a, O. Grizzi a, M. Sa´nchez b, B. Blum c, G. Benitez c, R.C. Salvarezza c a b

Centro Ato´mico Bariloche, CNEA, Instituto Balseiro, UNC, CONICET, Bustillo 9500, 8400 S.C. de Bariloche, Rı´o Negro, Argentina Planta Piloto de Ing. Quı´mica, Camino La Carringada km 7, Departamento de Fı´sica, Universidad Nacional del Sur, Av. Alem 1253, 8000 Bahı´a Blanca, Argentina c INIFTA, Departamento de Quı´mica, Universidad Nacional de La Plata y CONICET, C.C. 16, Suc. 4, B1904DPI La Plata, Buenos Aires, Argentina Received 27 December 2005; accepted for publication 21 March 2006 Available online 18 April 2006

Abstract We use direct recoiling spectroscopy with time-of-flight analysis to study the adsorption of propanethiol on Ag(1 1 1) cleaned and polished in vacuum by cycles of grazing ion bombardment and annealing. We discuss the advantages and drawbacks of the technique to follow the growth of the organic film. In particular, the low damage imparted by the technique allows to follow in detail the evolution of the H, C and substrate recoiling peaks for a wide range of exposures ranging from 101 to 2 · 104 L. The shape of the TOF spectra and the evolution of the recoiling intensities are consistent with a growth process in three stages: an initial fast one related to the density of defects at the surface, a second one where the surface is covered with a thin layer of organic molecules, presumably associated with lyingdown molecules, and a final stage corresponding to a thicker layer that can be associated with a standing-up orientation of the molecules in the film. Annealing of the organic film to 250 C produces complete depletion of C and H, leaving a small amount of S. The final S coverage after annealing depends on the initial roughness, being higher for rougher surfaces. We also observe an increase in the surface roughness after desorption of the thiol layer. Re-adsorption on this post annealed surface presents a marked enhancement of the initial sticking.  2006 Elsevier B.V. All rights reserved. Keywords: Ion scattering spectroscopy; Adsorption kinetics; Self-assembly; Sticking; Surface roughness; Silver; Alkanethiols; Solid–gas interfaces

1. Introduction The study of sulphur-based organic molecule adsorption at surfaces, with traditional surface science techniques in UHV, has received considerable attention during the past 15 years, partly because of the foreseen applications [1,2], and also because its basic comprehension represents a challenge for a number of reasons: (1) the usually large number of atoms involved with each molecule, (2) the extremely delicate balance of forces (between substrate atoms and *

Corresponding author. E-mail address: [email protected] (J.E. Gayone).

0039-6028/$ - see front matter  2006 Elsevier B.V. All rights reserved. doi:10.1016/j.susc.2006.03.024

head-groups, and amongst the ad-molecules) that govern the transient and final equilibrium structures, (3) the large number of paths that the organic molecules can undergo upon impinging on the surface such as precursor states, physical and chemical adsorption, dissociation, self-assembly, (4) possible rearrangement of the substrate atoms and changes in topography [3,4] and (5) further modifications of the organic films arising from the high sensitivity of organic molecules to electron and ion irradiation, either from the source or from secondary processes [5–9]. For the past 20 years, and following the pioneering work of Nuzzo and Allara [10], a large amount of experimental and theoretical work has been devoted to the study

2306

L.M. Rodrı´guez et al. / Surface Science 600 (2006) 2305–2316

of n-alkanethiols (HSCnH2n+1, hereafter Cn) on metals, particularly on Au, Ag and Cu, stemming from their stability and ease of preparation. Many studies have addressed their saturation coverage equilibrium structures, i.e., the denser and more stable phases obtained either by immersion of the samples in alkanethiol-containing solutions, or by exposing the cleaned surfaces to the vapours of pure alkanethiols in vacuum [1,2,11–18]. The kinetics of adsorption of the different phases formed on Au during exposure in vacuum has been delineated [4,11,19,20] by applying a combination of STM, thermal desorption, atom scattering, diffraction techniques, and electron spectroscopies. On Ag, the number of works describing the alkanethiol adsorption kinetics [21–24] is far less than on Au, and a full picture, starting from very low coverages, lags far behind. The case of adsorption of short-chain molecules on Au and Ag from the vapour phase has received particular attention very recently. For this case, part of the interest is based on the fact that electrical conductivity through the films increases with the shortening of the alkyl chain. Another interesting aspect is that, even on Au, the self-assembly process of short alkanethiols can be incomplete [25]. Experiments at low temperatures [26,27] show that short-chain molecules do not dissociate their H–S bonding, except for adsorption at defects, and desorb well below room temperature. Theoretical results support these findings of non-dissociative adsorption on atop sites [27]. We have shown recently [28] that for sufficiently high exposures it is possible to form a propanethiol layer on a clean and smooth Ag(1 1 1) surface from the vapour phase at room temperature, and that more than one phase is detected before reaching the saturation-coverage phase. In the present work we extend these results and discuss the pros and cons of direct recoil (DR) spectroscopy with time of flight analysis (TOF–DRS) [29,30] to study the adsorption kinetics of alkanethiols on Ag(1 1 1) surfaces. To our knowledge, this technique has been applied previously to just a few cases [28,31,32], while ion scattering spectroscopy (ISS), where only the ions are detected, has been used more often [33–35]. Here, we follow the H, C, and substrate DR intensities as a function of alkanethiol exposure. We discuss the products remaining at the surface after annealing these films at 250 C, the effect of surface roughness on the adsorption kinetics, and the changes observed for exposures performed immediately after an adsorption/desorption cycle, without cleaning in between. 2. Experimental details The TOF–DRS measurements were performed in a UHV chamber connected to an ion accelerator working from 1 to 100 keV [36]. The base pressure of the chamber was 2–3 · 1010 Torr when the ion beam line was open. The ions were generated in a radio frequency source, accelerated, and then mass selected and collimated to 0.1 of angular divergence. Most of the spectra were acquired by

time of flight (TOF) methods with 4.2 keV Ar+ pulsed ion beams (frequency 30 kHz, time resolution better than 50 ns); some spectra were acquired with Ne+ and Kr+ projectiles. The detection of both ion and neutral forward recoils was performed at a scattering angle d = 45 with a channeltron located after a flight path of 96 cm. A deflector plate and grids located in front of the channeltron allows ion-fraction measurements [37]. Typical current densities in the continuous beam were 1 nA/mm2, which were then reduced by three orders of magnitude during pulsing. The number of spectra required to follow a full adsorption process was between 10 and 30, amounting to a maximum total bombarding dose in the range of 1012 incident ions per cm2. The damage generated by this total dose is typically not detectable [32]. The sample was an 8 mm in diameter and 1 mm thick Ag(1 1 1) single crystal (brand new from MaTecK GmbH, 99.99% purity, initial surface roughness < 0.03 lm, orientation accuracy < 0.5). The sample was mounted on a manipulator allowing continuous variation of the projectile incident and azimuthal angles, and cooling and heating of the sample. The sample temperature was monitored by a chromel–alumel thermocouple mechanically attached to the sample surface, and controlled to within 1 C. Cleaning and polishing of the sample were carried on by cycles of grazing sputtering and annealing [38], consisting on 20 keV Ar+ bombardment at 2 or 3 from the surface plane (for 30–40 min, with current densities around 30 nA/mm2 and continuous rotation of the sample azimuthal angle), followed by annealing at 500 C for several minutes. This procedure was very effective to smooth out the initial surface roughness and the roughness generated at each thiol adsorption/desorption cycle. For comparison purposes, some measurements were performed on another Ag(1 1 1) single crystal that showed a higher initial roughness. This crystal was mechanically polished with alumina grit down to 0.05 lm, followed by a mechanical–chemical polish with aqueous 0.14%H2O2 + 0.02%NH3. Final cleaning of this crystal was performed in vacuum by sputtering and annealing. Most of the shown spectra correspond to samples prepared by dosing with propanethiol (99% purity, Aldrich Chemical Company) of the surface sputter-cleaned and polished in UHV. The thiols were contained in a glass reservoir fixed to the vacuum chamber through a leak valve, and before each exposure the thiols were purified through freezing–pump–thaw cycles. The reported pressures are corrected by the gauge sensitivity, which for propanethiol should be around 3.7 [11]. For comparison purposes, some films were prepared with C6 and C12, either by exposure in vacuum to the pure thiols, or by immersion of the Ag(1 1 1) samples in 2 mM ethanolic solutions, with typical immersion times varied between 5 h and 20 h. Some AES spectra were acquired in separate chambers with a 3 keV electron beam, with some spectra taken in the pulse counting mode to minimize beam damage, and some in the derivative mode at different sample regions.

L.M. Rodrı´guez et al. / Surface Science 600 (2006) 2305–2316

3. Results and discussion In this section we present and discuss the growth of propanethiol and other short-chain alkanethiol films on Ag(1 1 1). We discuss the advantages and disadvantages of using TOF–DRS (Section 3.1) and show examples of characteristic spectra (Section 3.2), before delving into the adsorption kinetics (Section 3.3). We then discuss the effects of surface roughness (Section 3.4) and film annealing (Section 3.5) on the final film and surface qualities. 3.1. Advantages and disadvantages of TOF–DRS TOF–DRS [30] is a useful technique for the observation of organic molecules at surfaces given its high sensitivity to H atoms, normally not observed with electron spectroscopies; and the high efficiency of TOF techniques for detection of both ions and neutrals, making it suitable to follow adsorption and thermal desorption processes with little damage despite the inherent destructive character of each projectile [7,39]. Provided that an important fraction of the surface is well ordered, it is also possible to obtain information about the surface crystallography, as has been shown for hexadecanethiol on Au [32]. Additionally, the technique has enhanced sensitivity to the outermost layer of atoms, a feature that is desirable, for example, in the study of functionalized molecules [32,33]. However, in some cases, as we will see later, this enhanced surface sensitivity can be a disadvantage, since it precludes quantitative determination of S signals even for low coverages or for the smaller molecules studied in this work (C3), and only a fraction of the total amount of H and C atoms are sampled by the projectile, i.e., those in the outermost layers. This results in little (if any) information about the whole molecule composition or its possible fragmentation. The uncertainty in the determination of S lying below the hydrocarbon chain is perhaps the most important drawback because it becomes difficult, by using TOF– DRS alone, to ascertain the quality of the film grown at the surface, since after all, most contaminants in a typical surface experiment are composed of C and H atoms. Nonetheless, the information obtained is complementary both to that obtained in for example thermal desorption experiments, in the sense that TOF–DRS samples the molecules remaining at the surface instead of the desorbing products, and also to that obtained from AES and XPS, which can follow the S-quantity but not the H-quantity present at the surface, and are known to produce measurable radiation damage. It is therefore worthwhile to give a brief account of the main TOF–DRS features for these systems, before introducing the results for the adsorption kinetics. 3.2. TOF Spectra and recoiling intensities The process of forward recoiling in direct collisions with the projectile (DR) can be used for a first characterization of the surface cleanliness and its crystallography, to get

2307

some information about the surface roughness, and to normalize the intensity of the different recoiling peaks in spectra obtained after exposure to vapours [40]. The latter needs relatively high incident and observation angles to discard shadowing or focusing effects. Incident angles h around 20 (with respect to the surface plane) and scattering angles d around 45 (from the ion beam direction) are standard for Ne and Ar projectiles scattering at 4–5 keV from many surfaces, except for heaviest and densely packed faces. Ag(1 1 1) is an example of such cases and requires a careful choice of the azimuthal angle, otherwise shadowing effects will be present. Fig. 1(a) shows a TOF spectrum for 4.2 keV Ar+ scattering from clean Ag(1 1 1) at d = 45. The inset shows the scattering geometry and Fig. 1(b), defines the azimuthal angle / = 0 when the incident beam is aligned with the [1 0 1] direction. The sharp peak observed in Fig. 1(a) corresponds to the Ar projectiles scattered from the Ag surface atoms, and the broad peak observed at higher time-of-flight corresponds to Ag recoils. A multiple scattering feature is also observed at the left side of the Ar single scattering peak. The lack of peaks (due to H, C, O recoils) at the left side of the Ar scattering peak is consistent with a surface clean to better than 1% of a ML. The large variation of the Ag recoil intensity with the incident angle (Fig. 1(c)) and with azimuth (Fig. 1(d)) reflects the well ordered crystallography of the Ag(1 1 1) surface, and the need to work away from main crystallographic axes to be able to use the Ag recoil peak as a reference peak. Spectra taken at large scattering angles (d = 135, not shown) confirm the crystallography of the sample. Because of the high density of the surface, the observed Ag recoil peak is not coming from true direct recoiling processes: it appears somewhat shifted in energy and with a distorted shape due to focusing effects in the outgoing trajectory. The dependence of the Ar-scattering-from-Ag intensity with the incident angle also gives some information regarding the-surface smoothness. Figs. 2(a) and (b) show such dependence for the Ag sample from MaTeck after many cleaning and polishing cycles (representing a ‘‘smooth’’ surface) and for the rougher Ag(1 1 1) surface. The Arscattering-from-Ag intensity in the case of an atomically flat surface should be vanishing at h = 5 incidence (Fig. 2(b)). The corresponding spectrum for the rougher surface presents a much higher Ar scattered intensity and a ‘‘true’’ Ag direct recoil coming from surface defects. In the next section we will discuss how these different initial conditions affect the kinetics of adsorption. Fig. 3 shows examples of TOF spectra obtained with 4.2 keV Ar+ projectiles for Ag(1 1 1) exposed in UHV to the vapours of pure C3 (Figs. 3(a) and (b)) and C6 (Fig. 3(c)), and for C12 (Fig. 3(d)) prepared by immersion. The spectra were acquired at an incident angle h = 20 with respect to the surface. In all cases the TOF–DRS spectra show characteristic features: well-defined H and C DR peaks (narrower at lower coverages), absence of clear peaks due to S DR or Ar scattering from S atoms, and

L.M. Rodrı´guez et al. / Surface Science 600 (2006) 2305–2316

2308

(a)

4.2 keV Ar→ Ag(111) (b)

o

φ =30 Counts (arb. units)

θ =20

Ag(111)

o

θ

o

δ =45

δ

φ

Ar

φ=30

o

φ=0o

Ag 5

10

15

20

TOF (μs)

Ag Recoiling Intensity (arb. units)

Ag Recoiling Intensity (arb.units)

(c)

o

φ=30

0

10

20

30

40

Incident Angle θ (deg.)

(d)

θ=20o

[-101] [-211]

0

20

40

60

80

100

Azimuthal Angle φ (deg.)

Fig. 1. (a) TOF spectrum for 4.2 keV Ar+ scattering off the clean Ag(1 1 1) surface. The inset shows the definition of azimuthal / and incident h angles. (b) Schematic diagram of the Ag(1 1 1) surface. (c) Ag recoiling intensity as a function of the incident angle measured along the [2 1 1] direction (/ = 30). (d) Ag recoiling intensity as a function of the azimuthal angle measured at h = 20.

Ar-scattering-from-Ag peaks that are dependent on coverage. In the high coverage spectra (Figs. 3(b), and (d)), the Ar-scattering-from-Ag peak is strongly reduced in intensity (even for C3) and the Ag recoils are completely attenuated. The absence of clear S features is rather surprising, in particular for short-molecules (C3) and at relatively low coverage (Fig. 3(a)), where one could expect a large fraction of the molecules to be lying down. We tried at different incident and azimuthal angles and also with Ne and Kr projectiles (not shown), but were not able to see S features clearly. Some S features seemed to be barely observed with Kr and Ar projectiles at low coverages, but were never very clear. The existence of S in our samples was verified by XPS and AES in separate set-ups and by TOF–DRS after annealing the samples (as we will discuss later). The inset

in Fig. 3(c) shows a typical AES spectrum, where we can observe the large S content. In this example, the film was grown in UHV, where the TOF spectrum of Fig. 3(c) was acquired in situ within 5 min of the C6 exposure, then the sample was transferred to the AES chamber in air (within an hour), where the shown AES spectrum was acquired in pulse counting mode. It is interesting to note that even though the covered sample was transferred in air, no oxygen was detected within the AES sensitivity, showing the good quality of the film. In the TOF–DRS experiments performed by Rabalais group with longer-chain alkanethiols adsorbed on Au [32], S was also undetectable; only for methanethiol adsorbed on Pt [31], this group was able to see S features together with C and H ones. This could be due to the shorter alkyl chain used and the higher

L.M. Rodrı´guez et al. / Surface Science 600 (2006) 2305–2316

(a)

2309

(b) 4.2 keV Ar → Ag(111)

Counts ( arb.units)

Ar Intensity (arb. units)

o

4.2 keV Ar→ Ag(111)

φ=30

θ=5 o φ=30

ο

smooth rough

Ar

smooth rough

0

10

Ag 5

20

Incident Angle θ (deg)

10

15

20

TOF (μs)

Fig. 2. (a) Ar scattering from Ag intensity as a function of the incident angle along the [2 1 1] direction (/ = 30) for a smooth (full symbols) and a rougher (empty symbols) Ag(1 1 1) surface. (b) Corresponding TOF spectra measured for a grazing incident angle of h = 5, along / = 30.

(b)

(a) C3/Ag(111) Low Coverage

H

Calculated TOF Spectra Clean Ag(111) Saturated Surface

C

5

H

5

10 15 TOF (μs)

20

20

25

(c)

5

10

15 TOF (μs)

20

C

C12/Ag(111) Inmersion

H C

S

25

(d)

H C

25

C3/Ag(111) High Coverage

Ag

C

10 15 TOF (μs)

Ag 100

200

300

400

500

KE (eV)

C6/Ag(111)

5

10

15 TOF (μs)

20

25

5

10

15 TOF (μs)

20

25

Fig. 3. TOF spectra for Ag(1 1 1) exposed in vacuum to (a) C3 at low dose, (b) C3 at saturation dose, (c) C6, and (d) for Ag(1 1 1) prepared by immersion in a 2 mM C12 + ethanol solution. The inset in (b) shows simulations for the Ar projectiles scattered (without recoil trajectories) from the clean and the C3 covered Ag surfaces performed with the SRIM2003 [42] code. The inset in (c) shows the corresponding Auger spectrum taken in a separate chamber.

probability for molecule dissociation at the Pt surface [41]. The absence of S DR peaks for low and intermediate coverages on Ag suggests that the alkyl chain should be always resting higher than the head group (even for molecules with lying-down chains). A high coverage is evidenced in the TOF spectra by the strong reduction of both the substrate recoils and the Ar scattering from Ag intensity, as is shown in Figs. 3(b) and (d) for C3 adsorbed on Ag by evaporation of pure

propanethiol and for C12 prepared by immersion, respectively. The almost complete absence of Ar scattering from Ag atoms (regardless of azimuth) and the broadening of the C and H DR peaks suggest standing-up adsorption. We performed ion-trajectory simulations with the SRIM2003 code [42] to get some insight into the scattering and recoiling processes. The inset of Fig. 3(b) shows the spectra simulated only for the Ar projectiles scattered off the clean Ag surface (dotted line) and the surface with a

L.M. Rodrı´guez et al. / Surface Science 600 (2006) 2305–2316

2310

S layer plus three hydrocarbon layers with the density of p p the thiol layer in the ( 7 · 7)R19.1 superstructure (solid line). The latter intends to simulate a surface saturated with thiol molecules adsorbed in a standing-up configuration. From this figure we see that the intense Ar scattering peak (from Ag) observed for the clean surface is shifted to lower kinetic energy (longer TOF), and strongly attenuated. The complete disappearance of the Ar scattering from Ag atoms was obtained (experimentally) in some exposure sequences, for the smoother surfaces, but not in all cases; sometimes saturation happened at a situation as that shown for the hexanethiol example of Fig. 3(c), where the substrate recoil disappeared (complete substrate shadowing) while some Ar multiple-scattering involving substrate atoms was still observed. In these cases, additional exposures did not change the spectra any more. This will be discussed further below (Section 3.3), together with the effect of surface roughness. In the case of ISS experiments performed on films of C16 on Ag(1 1 1) [34] and on Au(1 1 1) [35], where only ions are detected, the high neutralization of He projectiles in collisions with C atoms precludes observation of single scattering features. Only after bombardment with the same projectiles with doses above 1 · 1015 ions/cm2 the scattering off S and off substrate atoms became observable [34,35]. In the case of the TOF–DRS method, neutralization does not play any role, and it is possible to detect both neutrals plus ions (N + I), or just the neutrals or the ions alone, by placing a pair of deflecting plates in front of the detector. We expected that the high affinity level of S together with the decrease in the work function due to thiol adsorption [43] might allow observation of S ions (giving a higher sensitivity to detect it in an only ion spectrum) as

4.2 keV Ar → C3/Ag(111) Ions+Neutrals Neutrals Ions

H

Counts (arb. units)

C

is the case for F in fluorides, or for O in many samples. Fig. 4 shows the corresponding N + I, N and I spectra for 4.2 keV Ar+ bombardment of C3/Ag(1 1 1) at h = 20 incidence. Neither the expected S contribution nor C+ one are present. The only ions observed above noise level are some H+, with a typical ion fraction of 10%. Azimuthal dependence in the TOF–DRS spectra was looked for in the case of C3 deposited in vacuum on Ag(1 1 1). The C and H recoil intensities varied very little, typically less than 10%, in agreement with the observations of methanethiol on Pt measured at room temperature with the same technique [31]. Reduction of the sample temperature to 25 C after adsorption of C3 at room temperature gave similar results. Two independent but concurrent effects could preclude the observation of crystallographic order in these samples: the presence of many domains of different sizes and with different orientations, as in the case of C3/Au(1 1 1) [25], and large vibrations of the molecules. It should be noted that in longer-chain systems where these effects might be less important, such as hexadecanethiol on Au(1 1 1), order has been observed with this technique [32]. Studying the effects of incident angle variations at a fixed azimuthal angle can be of interest for the cases where the end groups differ from the chain composition [32,33], however, it also provides complementary information on surface order. The H and C DR signals for the C3 sample were followed as a function of the incident angle, at a fixed azimuth (Fig. 5). The coverage in Fig. 5(a) corresponds to about one half of saturation. We observe that H DR are focused at lower angles than C DR and are less blocked at grazing outgoing angles (inset of Fig. 5(a)), suggesting that H atoms are protruding into vacuum more than C atoms. This effect is the same at different azimuths, again showing the lack of order seen by TOF–DRS. There is however a difference when these measurements are performed at saturation coverage (Fig. 5(b)), here, focusing effects are not so strong and blocking effects starts at lower angles. These effects, together with the fact that the recoil peaks become broader at higher coverage are consistent with C and H atoms leaving the surface from different heights, as would be expected from standing-up molecules. 3.3. Adsorption kinetics

+

H

4

6

8

10

12

14

TOF (μs) Fig. 4. TOF spectra for neutrals plus ions, neutrals, and ions, acquired with 4.2 keV Ar+ for the Ag(1 1 1) surface saturated with C3.

The average coverage in kinetic studies are often followed step by step with XPS or other quantitative techniques such as AES. In these cases one has to evaluate the possible damage introduced in the film during irradiation, which could eventually lead to changes in the nucleation and growth processes by generating sites with exposed S atoms which might have undergone S–C bond cleavage, or by generating C@C double bonds that could hinder self-assembly. There is then a compromise between the number of points to be measured to follow in detail the growth process and the damage imparted to the film. Ion scattering techniques also generate damage by collision processes and by electronic excitations [7]. However, by

L.M. Rodrı´guez et al. / Surface Science 600 (2006) 2305–2316

10

(a)

H to C recoiling intensity Ratio Intermidiate Coverage High Coverage

Direct Recoiling Intensity (Arb. units)

8 6 4 2 0 0

10

20

30

40

Incident angle (deg)

Intermidiate Coverage H C

0 1

(b)

Direct Recoiling Intensity (Arb. units)

4.2 keV Ar → Ag(111) o φ =0 High Coverage H C

p p p p cules/cm2 (( 3 · 3)R30 versus ( 7 · 7)R19.1). The kinetics of adsorption depends on the length of the hydrocarbon chain [1,48]. The ordered structures for short-chain thiols (C3) can be full of defects [25,44] and may never fully attain the ideal coverage. Moreover, it has been shown [23,26,27] that at low temperatures short-chain molecules adsorb non-dissociatively on Au(1 1 1) and Ag(1 1 0) and desorb below 200 K, except at defects where dissociation may be taking place. Recently, we have reported [28] the formation of a layer of C3 on Ag(1 1 1), grown at room temperature from the vapour phase. It is interesting to breach the gap between short- and long-chain alkanethiols, and different sample preparation methods. Here we exposed the Ag(1 1 1) surface to the vapours of C3 in UHV, and followed the scattering and recoiling intensities by TOF–DRS. We also studied the effect of surface roughness on the adsorption kinetics; an issue that could help to understand why in certain experiments under vacuum the final SAM phase is not reached. Fig. 6(a) shows the evolution of the recoiling intensities for Ag, C and H versus C3 exposure (in the inset the same curves are presented with the exposure-axis in a logarithmic scale, as they are usually put out in the literature). Recording of successive TOF spectra at a particular exposure gives very reproducible results, showing that the

Normalized Recoiling Intensity

1.0

0 0

10

20

30

40

Incident Angle θ (deg.) Fig. 5. H and C direct recoiling intensity as a function of incidence angle for C3/Ag(1 1 1) at (a) intermediate, and (b) high surface coverage. The inset shows the ratio of the H to C direct recoiling intensity for both coverages.

1.0

0.8 0.5

0.6

0.15

0.0 10

-1

10

0

10

2

10

3

10

4

C3/Ag(111) Ag H C

0.2

0

2

4

6 8 Dose (L)

(b)

0

1

0.4

H C

0.00

10

Dose (L)

0.0

Normalized Recoiling Intensity

using TOF techniques at forward angles the sensitivity of the method becomes very high, and fluences of 1012 ions/ cm2 are sufficient to follow a full adsorption sequence with little damage [32] and considerable detail, making observations of changes in the sticking coefficient possible. Moreover, C, H and substrate recoil signals can be followed simultaneously, providing additional information about the film growth. A key feature for the kinetics of alkanethiol adsorption is the film growth in two or more steps, with characteristic phase transitions driven mainly by coverage [1,11,19, 20,44]. For film growth from the vapour phase, this behaviour has been followed on a number of systems, mainly on Au [4,19,20], Ag [22,45], and Cu [46,47]. On Au(1 1 1), the saturation of the SAM phase corresponds to 4.6 · 1014 molecules/cm2, and on Ag(1 1 1) to 5.9 · 1014 mole-

(a)

Normalized Recoiling Intensity

1

2311

5

10 15 Dose (L)

20

25

1

0 0

10

(c)

12x103

Ag H C

200 400 Dose (L)

600

Fig. 6. Ag, H and C recoiling intensity versus C3 dose for Ag(1 1 1). The Ag and H intensities are normalized to 1. Expanded views of panel (a) are shown in (b) and (c). The inset in (a) shows the adsorption curves with the dose in logarithmic scale.

L.M. Rodrı´guez et al. / Surface Science 600 (2006) 2305–2316

2312

1.0

C3/Ag(111) 8.0

7.0

6.0

10

-1

10

0

1

10 10 Dose (L)

2

10

3

10

4

Fig. 7. Width of the H recoil peak as a function of exposure. The bars represent statistical errors.

damage introduced by the probe is essentially negligible. A multi stepped adsorption sequence becomes clear. First we observe a small, but steep increase of H and C intensities below about 2 L accompanied by no appreciable decrease of the Ag recoil intensity. This initial film growth takes place at a very high sticking coefficient (Fig. 6(b)). A large region follows up to near 300 L where average C and H intensities start to increase more slowly, accompanied by a decrease in the Ag recoil intensity (Figs. 6(b) and (c)). This process proceeds with a sticking coefficient that is, on average, about 30 times smaller than the initial one. At about 300 L a new change in the slope (Fig. 6(c)) and in the width (Fig. 7) of the spectra is observed, corresponding to another decrease in the sticking coefficient by a factor of nearly ten, and almost complete disappearance of the Ag recoil. Around 2 · 103 L the coverage reaches saturation and beyond this point, increased exposures (for more than an order of magnitude) produce changes neither in the shape nor in the intensity of the H and C DR signals. During the film-growth process, the Ag recoil signal is seen up to the second change in the sticking, near 300 L, meaning that the substrate is completely shadowed (or covered) at this point. Plots of the H versus C recoil intensity (Fig. 8) show that their ratio is maintained approximately constant during the film-growth process up to 300 L, showing a small change around that point. A full interpretation of all these features is difficult because the information about the phases formed during vacuum exposures to C3 is very scarce. However, comparison with different measurements helps to interpret some of them. From a study of XPS and NEXAFS for C20 adsorption from solution on polycrystalline Ag, Himmelhaus et al. [49] proposed SAM island formation even for relatively low coverage. From an angle resolved UPS study, Miller et al. [22] measured non-dispersive LUMO states at low coverage, and dispersive ones at high coverages, and suggested a phase transition in which the molecules undergo a change from a non-interacting to an interacting phase at relatively

Normalized Recoiling Intensity

H Peak Width /arb.units

9.0

0.8

0.6

310 L

0.4

0.2

C3/Ag(111) Ag H

0.0 0.0

0.1 0.2 0.3 0.4 Normalized C Recoiling Intensity

0.5

Fig. 8. H and Ag recoiling intensities versus C recoiling intensity for the adsorption curve of Fig. 6.

high coverages, while at intermediate coverages both phases coexist. Electrochemical results [24] combined with Raman techniques show that for short-alkanethiols (n 6 6) on roughened Ag electrodes there is a reversible two-phase electroadsorption process involving a strongly bound phase with disordered alkyl chains (gauche conformation) up till completion of approximately the first 1/3 of a ML, followed by a slightly weaker bound phase with molecules in the all transpconformation, up to completion of 1 ML, assuming p the ( 7 · 7)R19.1 structure. Coming back to Figs. 6 and 7, the initial sharp rise in the coverage, below 2 L, could be related to a nucleation process favouring the defect sites on the Ag(1 1 1) surface, as occurs on Au(1 1 1) [26]. This will be discussed in more detail in Section 3.4. Up till 300 L the height of the H and C recoil peaks increases fast, coming probably from a film-growth process involving molecules randomly dispersed on the surface, i.e., the non-interacting phase [22], until most of the substrate becomes shadowed (disappearance of Ag recoils). The narrow and constant shape of the H and C peak suggest that this layer should be extremely thin (lying-down molecules). From 300 L to saturation, the C and H recoil peaks become broader (Fig. 7), and the intensity below them, i.e., the surface recoils coming from multiple collisions, also increases. This would be consistent with C and H recoils coming from different ‘layers’ (different depths within the molecule) as would be the case both for molecules standing up and for molecules with gauche defects, and could correspond to the interacting [22] or weakly bound phase [24]. To understand better the different ‘‘saturation’’ coverages obtained on different samples, and by different preparation methods (Fig. 3), we have investigated in more detail both the effects of initial surface roughness, and readsorption on surfaces priorly annealed to desorb the alkanethiol film, but without sputter–cleaning or polishing in between.

L.M. Rodrı´guez et al. / Surface Science 600 (2006) 2305–2316

3.4. Effect of surface roughness Comparison of the exposure-values in different works is a difficult task: these values might be affected both by the not well known gauge sensitivity to the different alkanethiols, and by the different geometry of the set-ups, i.e., if the molecules approach the surface through a thin tube or fill the chamber uniformly. Besides these basic problems, there may be effects related to the alkanethiol pressure [11] (sometimes we have observed different coverages for equal exposure doses performed at different rates), to small amounts of S residing at the surface from previous adsorptions [21,23], to surface roughness, and to sample temperature [4]. Because of these effects (or a combination of them) we have observed variations in some aspects of the adsorption kinetics, namely in the maximum value of the coverage for the dense phase (i.e., in the final ratio of the Ar scattering from Ag to recoiling peaks), and in the characteristic exposure values. However, the general shape of the curves was well reproduced, with well defined initial rises in the coverage, and clear changes in the sticking coefficients. As an example of the effect of the surface roughness we show, in Fig. 9, the initial film-growth mode on the rougher Ag(1 1 1) surface, together with the one discussed above. We already saw in Fig. 2(b) that for this rougher Ag sample the intensity of Ar ions scattered at d = 45 for h = 5 incidence was much higher (20 times) than for the in situ polished sample described above. The root mean square roughness for the smooth surface was less than 0.7 nm in 1 lm · 1 lm AFM scans, while that of the rougher surface was at least 5 times higher. An increase in the surface roughness results in a higher initial H and C content, followed by a quasi saturation at low exposures. In the example shown, alkanethiol film growth stops after about 100 L.

0.4

C3/Ag(111) rough surface

H Recoiling Intensity

0.3

Smooth Surface

0.2

0.1

0.0 1

10

100 Dose (L)

Fig. 9. H recoil intensity as a function of dose for the adsorption of C3 on a smooth (full symbols) and a rough (open symbols) Ag(1 1 1) surface.

2313

This quasi saturation for the rougher surface may then explain the different degree of coverage obtained in different adsorption sequences and for different experimental setups at different laboratories. Note the crossing of the curves corresponding to the smooth and the rough surfaces, near 60 L. More than one factor could be acting on the filmgrowth process to inhibit attainment of the high saturation phase. Firstly, a rougher surface would contain more initial nucleation sites, thereby resulting in a populated mosaic structure. Secondly, it might reduce the mobility of both adsorbed and surface atoms, thus hampering the formation of good quality SAMs. Himmelhaus et al. [49] showed that the presence of oxides on the Ag surface has similar effects. They proposed that Ag2O and AgOH accelerate alkanethiol film growth by facilitating cleavage of the H-S bond. However, the final quality of the film is lesser than in the absence of silver oxide. 3.5. Effect of film annealing Annealing to 250 C of the alkanethiol film deposited on Ag leaves a surface free from H and C, but retaining a small amount of S, barely noticed at 20 incidence (not shown) and seen better at 5 incidence. Fig. 10 shows the resulting spectra for three different initial conditions, the counts in the three figures are normalized with those in the initial clean surface recoil, so the vertical scales can be compared. The spectra of Fig. 10(a) correspond to the clean smooth surface (thick line) and to the annealed surface after adsorbing a saturation phase of C3 (light line, higher-count spectrum). The spectrum at 20 incidence taken before annealing is presented in the inset, which shows the high degree of initial coverage (absence of Ar scattering from Ag atoms). Figs. 10(b) and (c) show similar spectra for a low initial coverage for the smooth and the rough Ag surfaces, respectively. We can clearly distinguish the presence of S through both the scattering of Ar from S, and the S recoiling peak. The amount of the remaining sulphur changes with the initial coverage and also with the initial surface topography, being much higher at the rougher surface. A crude estimate suggests that the S observed along the different adsorption/desorption sequences in the smooth Ag surface is between 5% and 20% of the initial alkanethiol coverage, suggesting that a major part of the alkanethiolate molecules desorb either intact, or through by-products containing S. This is not the case for the rough surface, where a larger fraction of S remains at the surface. For the smooth surface, the amount of final S is so small and disperse, that we were not able to detect clear crystallographic effects in its intensity, i.e., measurements at different azimuths showed essentially the same coverage; only the clean surface crystallography showed up in both the recoiling Ag and in the Ar scattering from Ag. This dependence on surface roughness and the lack of azimuthal dependence in the S recoil peak suggest that the S remaining after annealing might be that produced during the adsorption near defects, and that molecules leaving from

L.M. Rodrı´guez et al. / Surface Science 600 (2006) 2305–2316

2314

2000 10

Annealed interface Clean Surface

1000

5

θ=20

0 5

ο

500

4.2 keV Ar → Ag(111) Smooth Surface High Coverage ο

3

1500

x10

Normalized Counts

(a)

θ=5

S recoil

10

TOF (μs)

15

20

Ar scattered from S 0 5

400

10 10

5

ο

ο

200

20

Smooth Surface Low Coverage

3

300

x10

Normalized Counts

(b)

15

θ=20

θ=5

0 5

10

TOF (μs)

15

20

100

0 5

8000

10

20

10

Rough Surface Low Coverage 3

6000 ο

θ=5 4000

x10

Normalized Counts

(c)

15

ο

5

θ=20

0 5

10

TOF (μs)

15

20

2000

gions and the true Ag direct recoil increase to a large extent (the first by a factor near 10) for the post annealed surface, suggesting an important change in the overall surface topography (Fig. 10(a)). This increase in the scattering intensity cannot be explained by the presence of S at the surface, since S should contribute to a decrease in the scattering intensity of Ar from clean Ag regions (because of shadowing), and an increase at the TOF corresponding to scattering of Ar from S, which is well separated in the TOF scale. Therefore, the changes observed in the surface topography could be related to the formation of Ag ad-islands [15,16,50,51], other defects, or a surface reconstruction [45], which cannot be discerned with TOF–DRS, but that seems to be more important than that observed in Au [4]. This roughening effect was very reproducible in all the adsorption/desorption sequences performed on flat surfaces (i.e., where the remaining S content is smallest). The final roughening was much smaller when the initial thiol coverage was low (Fig. 10(b)), pointing to a thiolinduced roughening effect. The Ag(1 1 1) surface modified in this way (i.e., by a complete adsorption/desorption cycle) has an important effect on subsequent adsorption/desorption sequences. In the particular case of Ag(1 1 0) [21,23] and Cu(1 1 0) [23], the number of adsorption/desorption cycles performed on the same surface (without cleaning or polishing in between) has a strong effect on the adsorption processes of C1 [23] and C2 [21]. In the case of Ag(1 1 0) it was proposed that the presence of S generates a reaction mechanism by which the C1 molecules dissociate, thus enhancing the adsorption probability of the resulting products (SH2 and CH4) [23]. Desorption of by-products from C2/S/Ag(1 1 0) [21] support these conclusions. Fig. 11 shows two spectra obtained for 300 L of C3 on Ag(1 1 1). One of them on Ag(1 1 1) cleaned by cycles of grazing sputtering and annealing, and the other one obtained in the following manner: first

0 5

10

15

20

TOF (arb. units)

flat regions carry with them the S atoms. The behaviour in Ag is completely different to that observed for methanethiol on Pt [31], where as a result of annealing most of the C and S atoms remain on the surface, and apparently only H desorbs. The S coverage in this latter case presented a clear crystallographic dependence. In addition to the appearance of the S features described above, both the forward scattering of Ar from clean Ag re-

4.2 keV Ar → C3/Ag(111) 300 L 300 L on the postannealed 300 L C3/Ag(111) interface.

Counts (arb. units)

Fig. 10. TOF spectra acquired at a grazing incident angle (h = 5) for the clean Ag(1 1 1) surface (thick line) and the post-annealed C3/Ag(1 1 1) surface (thin line) for different coverages and different initial roughness: (a) smooth surface with high C3 coverage, (b) smooth surface and low C3 coverage and (c) rough surface and low coverage. The inset in each figure shows the TOF spectra acquired at h = 20 for the corresponding C3/ Ag(1 1 1) surface before annealing.

5

10

15

20

TOF (μs) Fig. 11. TOF spectra resulting after an exposure of 300 L of C3 on the clean Ag(1 1 1) surface (thick line) and on the 300 L C3/Ag(1 1 1) surface annealed to 250 C for 10 min (see the text for details).

L.M. Rodrı´guez et al. / Surface Science 600 (2006) 2305–2316

the clean surface is exposed to 300 L of C3, then the surface is annealed to 250 C, and finally this surface is exposed again to 300 L of C3 (at the same rate). We observe that the H and C DR peaks increase considerably more than for the virgin surface. The thermal desorption of the films grown in this way leaves again some S at the surface but both H and C go away from the surface. With TOF– DRS we cannot determine if this increase in the sticking coefficient for Ag(1 1 1) is due to an enhanced probability for S–H bond cleavage to form the thiolate, followed by S–C bond cleavage with temperature, or if it is related to another dissociation mechanism. To what extent this effect is related to the remaining S, the increased roughness, or both is still under investigation.

2315

Desorption induced by annealing of the samples to 250 C produces depletion of C and H, leaving a small amount of S. The S coverage after annealing depends on the initial roughness, being higher for rougher surfaces. Crude estimations indicate that the final S content in the Ag(1 1 1) smooth surface is between 5% and 20% of the ultimate thiol coverage attained prior to annealing. A complete adsorption–anneal cycle of the thiol layer generates a rougher Ag surface, as evidenced by the increased intensity for 45 Ar scattering from Ag at grazing incidence. The increase in the surface roughness is only observed for initially flat surfaces exposed to high thiol doses. Readsorption on this post annealed Ag(1 1 1) surface presents a marked enhancement of the initial uptake, but inhibits attainment of full coverage at saturation.

4. Summary We have applied TOF–DRS to study the adsorption of propanethiol on Ag(1 1 1) cleaned and polished by cycles of grazing ion bombardment and annealing. The shape of recoiled and scattered peaks change with exposure. Near 300 L, the peak corresponding to Ar scattering from Ag is still seen and related to multiple collision sequences with substrate atoms, and the Ag recoil peak vanishes, indicating full coverage. From 300 L to saturation the peaks coming from H and C recoils become broader and the scattering of Ar from Ag is strongly reduced. For both C3 and longer-chain molecules we were not able to see the S-associated structures clearly. Only at very low exposures some shoulders at the positions corresponding to S recoil and to Ar scattering from S are observed, meaning that even in the low coverage phase, S atoms reside below the hydrocarbon chain. The C and H recoil peaks have a strong dependence on the projectile incident angle, consistent with a top layer terminated mainly by H atoms. This dependence is different at low and at high coverages, suggesting a change in relative H and C populations at the outermost layers, possibly related with a change in the tilt of the molecule. No clear azimuthal dependence was found in the TOF spectra. Measurements of ion fractions in the recoiled and scattered particles show some H+, typically 10%; other particles leave the surface as neutrals, having ion fractions below our detection limit. The evolution of C3 coverage on smooth Ag(1 1 1) proceeds with clear changes in the sticking coefficient near 2 L, near 300 L, and near 2000 L, being consistent with film growth in three stages, two of them on terraces, preceded by an initial stage associated with surface roughness. The coverage attained during this initial adsorption stage, observed below 2 L, is associated with high sticking at defect sites. To attain the final phase, large exposures (>103 L) are required, showing the low efficiency of the chemisorption process. The overall evolution of coverage on smooth Ag(1 1 1) can therefore be assimilated to the growth of short-chain alkanethiol films on Au(1 1 1), with an initial high uptake at defects, followed by slower organic film growth.

Acknowledgements We acknowledge fruitful discussions with Drs. M.L. Martiarena, P. Lustemberg and W.M. Lau, and financial support from FONCYT (PICT03 14452, PICT02 11111, PICT03 17492, PICT04 25959, PME 118), Fundacio´n Antorchas (Physics at the Nanoscale), CONICET (PIP 5248, PIP 5945, CIAM), Universidad Nacional de Cuyo (Project No. 06/C202), Universidad Nacional de La Plata, and Universidad Nacional del Sur (PGI 24/F027). This research was performed within the framework of the Argentine network of ‘‘Nanociencia y nanotecnologı´a molecular, supramolecular e interfaces’’. References [1] J.C. Love, L.A. Estroff, J.K. Kriebel, R.G. Nuzzo, G.M. Whitesides, Chem. Rev. 105 (2005) 1103. [2] A. Ulman, Chem. Rev. 96 (1996) 1533. [3] G.C. Jackson, D.P. Woodruff, R.G. Jones, N.K. Singh, A.S.Y. Chan, B.C.C. Cowie, V. Formoso, Phys. Rev. Lett. 84 (2000) 119. [4] G.E. Poirier, W.P. Fitts, J.M. White, Langmuir 17 (2001) 1176. [5] S.S. Duewez, J. Electron Spectrosc. 134 (2004) 97. [6] M. Zharnikov, S.F. Frey, K. Heister, M. Grunze, Langmuir 16 (2000) 2697. [7] D.E. Riederer, R. Chatterjee, S.W. Rosencrance, Z. Postawa, T.D. Dunbar, D.L. Allara, N. Winograd, J. Am. Chem. Soc. 119 (1997) 8089. [8] H. Kondoh, H. Nozoye, J. Phys. Chem. B 102 (1998) 2367. [9] L.P. Ratliff, R. Minniti, A. Bard, E.W. Bell, J.D. Gillapsy, D. Parks, A.J. Black, G.M. Whitesides, Appl. Phys. Lett. 75 (1999) 590. [10] R.G. Nuzzo, D.L. Allara, J. Am. Chem. Soc. 105 (1983) 4481. [11] F. Schreiber, Prog. Surf. Sci. 65 (2000) 151. [12] T. Pradeep, N. Sandhyarani, Pure Appl. Chem. 74 (2002) 1593. [13] G.E. Poirier, Chem. Rev. 97 (1997) 1117. [14] G.E. Poirier, E.D. Pylant, Science 272 (1996) 1145. [15] P. Fenter, P. Eisenberger, J. Li, N. Cammillone III, S. Bernasek, G. Scoles, T.A. Ramanarayanan, K.S. Liang, Langmuir 7 (1991) 2013. [16] A. Dhirani, M.A. Hines, A.J. Fisher, O. Ismail, P. Guyot-Sionnest, Langmuir 11 (1995) 2609. [17] R. Heinz, J.P. Rabe, Langmuir 11 (1995) 506. [18] S.M. Driver, D.P. Woodruff, Surf. Sci. 457 (2000) 11. [19] H. Kondoh, C. Kodama, H. Sumida, H. Nozoye, J. Chem. Phys. 111 (1999) 1175.

2316

L.M. Rodrı´guez et al. / Surface Science 600 (2006) 2305–2316

[20] F. Schreiber, A. Eberhardt, T.Y.B. Leung, P. Schwartz, S.M. Wetterer, D.J. Lavrich, L. Berman, P. Fenter, P. Eisenberger, G. Scoles, Phys. Rev. B 57 (1998) 12476. [21] D.M. Jaffey, R.J. Madix, Surf. Sci. 311 (1994) 159. [22] A.D. Miller, K.J. Gaffney, S.H. Liu, P. Szymanski, S. Garrett-Roe, C.M. Wong, C.B. Harris, J. Phys. Chem. A 106 (2002) 7636. [23] J.G. Lee, J. Lee, J.T. Yates Jr, J. Phys. Chem. B 108 (2004) 1686. [24] D.W. Hatchett, R.H. Uibel, J. Stevenson, J.M. Harris, H.S. White, J. Am. Chem. Soc. 120 (1998) 1062. [25] C. Vericat, M.E. Vela, R.C. Salvarezza, Phys. Chem. Chem. Phys. 7 (2005) 3258. [26] I.I. Rzeznicka, J. Lee, P. Maksymovych, J.T. Yates Jr, J. Phys. Chem. B 109 (2005) 15992–15996. [27] P. Maksymovych, D.C. Sorescu, D. Dougherty, J.T. Yates Jr., J. Phys. Chem. B 109 (2005) 15992. [28] L.M. Rodrı´guez, J.E. Gayone, E.A. Sa´nchez, O. Grizzi, B. Blum, R.C. Salvarezza, J. Phys. Chem. B 110 (2006) 7095. [29] H. Niehus, W. Heiland, E. Taglauer, Surf. Sci. Rep. 17 (1993) 213. [30] J.W. Rabalais, Principles and Applications of Ion Scattering Spectrometry, Surface Chemical and Structural analysis, WileyInterscience, 2003. [31] S.S. Kim, Y. Kim, H.I. Kim, S.H. Lee, T.R. Lee, S.S. Perry, J.W. Rabalais, J. Chem. Phys. 109 (1998) 9574. [32] L. Houssiau, M. Graupe, R. Colorado Jr, H.I. Kim, T.R. Lee, S.S. Perry, J.W. Rabalais, J. Chem. Phys 109 (1998) 9134. [33] M.W.G. Ponjee, M.A. Reijme, B.M.W. Langeveld-Voss, A.W. Denier van der Gon, H.H. Brongersma, Surf. Sci. 512 (2002) 194. [34] S.P. Chenakin, B. Heinz, H. Morgner, Surf. Sci. 397 (1998) 84.

[35] S.P. Chenakin, B. Heinz, H. Morgner, Surf. Sci. 421 (1999) 337. [36] L. De Ferrariis, F. Tutzauer, E.A. Sa´nchez, R.A. Baragiola, Nucl. Instrum. Meth. 281 (1989) 43. [37] J.E. Gayone, E.A. Sanchez, O. Grizzi, L.I. Vergara, M.C.G. Passeggi Jr, R. Vidal, J. Ferron, Nucl. Instrum. Meth. B 193 (2002) 440. [38] J.E. Gayone, R.G. Pregliasco, G.R. Go´mez, E.A. Sa´nchez, O. Grizzi, Phys. Rev. B 56 (1997) 4186. [39] B. Arezki, A. Delcorte, P. Bertrand, Nucl. Instrum. Meth. B 193 (2002) 755. [40] J.E. Gayone, E.A. Sa´nchez, O. Grizzi, Surf. Sci. 419 (1999) 188. [41] T.H. Lin, T.P. Huang, Y.L. Liu, C.C. Yeh, Y.H. Lai, W.H. Hung, Surf. Sci. 578 (2005) 27. [42] Available from: . [43] V. De Renzi, R. Rousseau, D. Marchetto, R. Biagi, S. Scandolo, U. Del Pennino, Phys. Rev. Lett. 95 (2005) 46804. [44] E. Barrena, E. Palacios-Lido´n, C. Munuera, X. Torrelles, S. Ferrer, U. Jonas, M. Salmeron, C. Ocal, J. Am. Chem. Soc. 126 (2004) 385. [45] M. Yu, S.M. Driver, D.P. Woodruff, Langmuir 21 (2005) 7285. [46] G. Loepp, S. Vollmer, G. Witte, Ch. Wo¨ll, Langmuir 15 (1999) 3767. [47] A. Ku¨nle, S. Vollmer, T.R. Linderoth, G. Witte, C. Wo¨ll, F. Besenbacher, Langmuir 18 (2002) 5558. [48] L.H. Dubois, B.R. Zegarski, R.G. Nuzzo, J. Chem. Phys. 98 (1) (1993) 678. [49] M. Himmelhaus, I. Gauss, M. Buck, F. Eisert, Ch. Wo¨ll, M. Grunze, J. Electron Spectrosc. 92 (1998) 139. [50] M. Kawasaki, H. Nagayama, Surf. Sci. 549 (2004) 237. [51] J.-P. Bucher, L. Santesson, K. Kern, Langmuir 10 (1994) 979.