4. Physisorption and Chemisorption of

Alkanethiols and Alkylsulfides on Au(111)


4.1. Introduction

The spontaneous ordering of adsorbates to form self-assembled monolayers (SAMs) is a phenomenon with great technological potential. Through the deposition of molecules with specific symmetry or functionality, surfaces can be custom-tailored for particular applications such as electrodes, chemical sensors, or for template patterning1-6. Although several adsorbate/substrate systems have been observed to self-assemble, the deposition of alkanethiols on single-crystal gold surfaces has been studied most frequently2. Due to the relative inertness of the gold surface to most potential contaminants, high-quality alkanethiol monolayers can be deposited from solution. Once prepared, SAMs composed of long­chain alkanethiols are air stable and do not desorb at room temperature7.

The principal ingredient for self-assembly is a relatively strong interfacial binding asymmetry of the molecular constituents. In the alkanethiol SAM case, this is provided by the affinity of the sulfur atom for gold and a comparably strong lateral interaction (4-8 kJ/mol per CH28) arising from the van der Waals forces between the chains. The magnitude of this lateral interaction can be controlled by changing the length of the hydrocarbon. While the lateral interactions are relatively well understood, the sulfur-gold interaction in the alkanethiol-gold SAMs has remained the subject of frequent debate.

Until recently, SAM formation was thought to proceed via simple Langmuir-type kinetics9. However, the recent discovery of the existence of a lower density phase (the so­called "striped" phase10) and measurements of the rate of thiol adsorption on gold surfaces (determined by LEED11, STM12, X-ray and atomic beam scattering13,14) have indicated that SAM growth from the vapor occurs in at least two steps. Atomic force microscopy (AFM)15 studies have recently shown that this is also true for SAM growth from solutions of alkanethiols in ethanol. Furthermore, during gas phase deposition, the growth rate shows a complex dependence on the impingement rate (pressure) with evidence of three regimes of linear, quadratic, and saturated growth16. To understand this complex behavior, an improved knowledge of the kinetics and energetics of adsorption is essential. However, only limited data on adsorption energies is available in the literature at present. Based on the adsorption of hexane, Nuzzo et al.17 estimated a methylene-surface interaction of 8 kJ/mol. Nuzzo has also reported an chemisorption energy of 117 kJ/mol for dimethyldisulfide adsorption.

To better understand the head group/substrate interactions responsible for the structure and growth kinetics of SAMs, a program was initiated to produce accurate measurements of the energetics of the interaction of alkanethiols, dialkylsulfides, and dialkyldisulfides with Au(111). The ability of alkanethiols to both physisorb through van der Waals interactions and to chemisorb through the sulfur atom provided an excellent opportunity to also study the role of the physisorbed precursor state in the chemisorption kinetics. By systematic variation of the chain length of the alkanethiol, the dependence of the physisorption and chemisorption energy on the identity of the alkanethiol was explored. By quantifying the rate of chemisorption from a physisorbed precursor state, the energy of the activation barrier to chemisorption was also determined. With this information a more unified picture of the adsorption process will be provided for this important class of molecules with an extremely widely used, but only partially understood, substrate.

4.2. Results and Discussion: Energetics

4.2.1. Energetics of Alkanethiol Adsorption

Temperature programmed desorption measurements were performed on various coverages of several linear 1-alkanethiols (ethanethiol, butanethiol, hexanethiol, octanethiol, and nonanethiol). All thiols were initially deposited on a Au(111) crystal at a surface temperature of 200 K. For each thiol, two thermal desorption peaks are observed (Figure 4-1). Although the peak desorption temperature of the lower temperature peak varied as a function of the chain length of the alkanethiol adsorbate, the higher temperature peak position remained relatively constant.



Figure 4-1. a) Temperature programmed desorption performed with a clean Au(111) crystal (- - -) and a hexanethiol-covered Au(111) surface (---). b) Differentiation of hexanethiol desorption curve shown with (- - -) and without (---) correction for Debye-Waller attenuation of the specular signal during the temperature ramp.

From the hydrocarbon desorption experiments presented in the previous chapter, it was found that the peak desorption temperature of physisorbed species increased linearly with increasing chain length. For comparison, the activation energies for desorption of the linear alkanes and alkanethiols are plotted in Figure 4-2 as a function of the bulk heat of vaporization of the adsorbate. Both series of data display a similar linear relationship between the adsorption energy and the heat of vaporization with a slope of 1.15. This relationship implies that the low­energy interaction with the surface is due to van der Waals forces which are responsible for the cohesive energy in the bulk phases.



Figure 4-2. Activation energy for desorption of both alkanethiols () and alkanes () from a Au(111) surface plotted as a function of bulk heat of vaporization. The alkanethiols shown are ethanethiol, butanethiol, hexanethiol, and nonanethiol. The alkanes are hexane, heptane, octane, nonane, decane, and dodecane. Linear fits to the alkanethiol (---) and the alkane (- - -) energy data are shown.

The robustness of this simple relationship of the physisorption enthalpy to the bulk heat of vaporization is displayed in Figure 4-3. Here, a wide variety of adsorbates ranging from simple alkanes to sulfur-containing molecules of variable structural complexity, as well as some aromatic compounds, show the same linear relationship between the bulk heat of vaporization of the molecule and its enthalpy of physisorption on the Au(111) surface.

Figure 4-3. Enthalpy of desorption of alkanes and various sulfur containing species versus their respective bulk heats of vaporization. The compounds displayed are (A) hexane, (B) heptane, (C) octane, (D) nonane, (E) decane, (F) dodecane, (G) benzene, (H) ethanethiol, (I) butanethiol, (J) hexanethiol, (K) octanethiol, (L) nonanethiol, (M) diethylsulfide, (N) dibutylsulfide, (O) 1,4-butanedithiol, (P) t-butanethiol, (Q) isopropylthiol, and (R) thiophene. A linear fit to the alkane data exclusively (- - -) and to all data points (---) are shown.

Table 4-1 presents the activation energies for desorption corresponding to the observed peak desorption temperatures of all the adsorbates studied. For most adsorbates, more than one TPD peak was observed. These multiple features have been classified by proposed origin and will be discussed individually in more detail in the following sections.

The column labeled "Phys" indicates energies corresponding to peaks assigned to desorption from the physisorbed layer, "Hindered" indicates a "lower than normal" chemisorbed energy state which is not observed for the simple alkanethiols, "Chem 1" refers to the first peak arising from chemisorption which has been identified as the "normal" chemisorption state of the thiols, and "Chem 2" are peaks from a higher energy chemisorption feature that can only be obtained at high exposures and that can be annealed away. The column "Phys(calc)" represents the physisorption energy calculated by an additive model outlined in Section 4.2.2. "Deviation" is the percent error between the prediction of the model and the observed experimental results.

Table 4-1. Activation Energies for Desorption (kJ/mol)
Molecule
Phys
Hindered
Chem 1
Chem 2
Phys(calc)
Deviation %
Methanethiol
48, 5818
49
2.0
Ethanethiol
57
127
55.1
-3.4
Butanethiol
68
127
67.3
-1.0
Hexanethiol
79
124
79.5
0.6
Octanethiol
87
125
147
91.7
5.4
Nonanethiol
103
127
152
97.8
-5.3
Decanethiol
126
146
Dodecanethiol
127
Tetradecanethiol
128
Hexadecanethiol
150
140.5
-6.7
Octadecanethiol
158 ± 10
152.7
-3.4
Docosanethiol
169 ± 10
177.1
4.8
Diethylsulfide
68
67.3
-1.0
Dibutylsulfide
86
91.7
6.6
Diethyldisulfide
124
t-Butanethiol
64
107
64.5
0.8
2-Propanethiol
64
107
64.5
0.8
Neopentanethiol
68
128
141
70.6
3.8
Thiophene
60
61.5
2.5
1,4 Butanedithiol
82
117
2,3 Butanedithiol
80
107
133
1,6 Hexanedithiol
129
150

4.2.2. Physisorption Enthalpies

Figure 4-4 displays the activation energies for desorption corresponding to all the desorption features found in the TPD spectra of the linear 1-alkanethiols plotted as a function of the chain length.

Figure 4-4. Plot of desorption enthalpy as a function of number of carbons for alkanes and alkanethiols. Desorption enthalpy of alkanethiols from the physisorbed state () and the chemisorbed state () are shown. Transient chemisorption features at higher energies are shown as open diamonds. The dotted and solid lines are the result of linear fits to the physisorbed and chemisorbed data, respectively. The dashed line is the linear fit to the alkane data of Figure 4-2.

The parameters of the linear fits to the alkane and alkanethiol data are shown in Equations 4-1 and 4-2, respectively. The incremental adsorption energy per carbon atom for the two series are similar in accordance with the linear relationship between the adsorption energy and the bulk heat of vaporization.

HtAu = (6.08 ± 0.74 kJ/mol) × C + (43.5 ± 5 kJ/mol) (4-1)

HaAu = (6.16 ± 0.16 kJ/mol) × C + (19.4 ± 1.4 kJ/mol) (4-2)

where C represents the number of carbon atoms, HtAu is the activation energy for desorption of the alkanethiols, and HaAu is the activation energy for desorption of the alkanes. The difference between the y-intercepts of the fits of the two series is 24.1 kJ/mol, due primarily to the presence of a highly-polarizable sulfur atom in the alkanethiols. The linear slopes of both energy relationships are indicative of the additive nature of the contribution of the methylene subunits to the desorption enthalpy, implying that the plane of the molecule (as defined by the carbon backbone) lies parallel to the metal surface; with each methylene unit contributing relatively equally via its polarizability.

The methylene contribution to the physisorption energy of the alkanethiols had previously been estimated as 7.9 kJ/mol6. In addition, Dubois et al.18 reported adsorption energies of 48.5 kJ/mol and 58 kJ/mol for methanethiol physisorbed on Au(111). (The presence of the lower energy peak was attributed to multilayering.) For methanethiol (C=1), Equation 4-1 predicts that the physisorption energy will be 49.6 kJ/mol, in good agreement with literature values. For comparison, a study by Sexton and Hughes19 which measured the adsorption of alcohols, ethers, and alkanes on Cu(100) and Pt(111) showed that the contribution of a methylene group to adsorption is 5.0­6.5 kJ/mol on both metals, and that the oxygen atom contributes 42 kJ/mol to the observed adsorption energy on platinum and 35 kJ/mol on copper. An alkane adsorption study by Teplyakov et al.20 has recently measured the methylene contribution to the activation energy for desorption from Cu(100), 6.3 kJ/mol.

The directional polarizability model of the previous chapter established that the contributions to the physisorption energy of hydrocarbons on Au(111) could be broken down into additive contributions of atoms or groups of atoms. In particular, a methylene group contributes 6.2 ± 0.1 kJ/mol in a long linear chain (but 8.1 kJ/mol in cyclic compounds) while a methyl group contributes 15.5 kJ/mol to the physisorption energy. In addition, the contribution of a double bond was found to amount to 6.1 kJ/mol. Because of the additive nature of the adsorption energies, these energy values were used with the linear alkanethiol adsorption energies reported in Table 4-1 to extract the contribution of the SH group. Simple algebra assigns the contribution of the SH group to the observed adsorption energy as:

HSH = HS + HCH3 - HCH2 (4-3)

HCH3 and HCH2 are the contributions of the methyl and methylene groups reported from the directional polarizability model and HS is the difference between the y­intercepts of the fits to the thiol and hydrocarbon data, 24.1 kJ/mol. The result of this calculation assigns the SH contribution to be 33.5 kJ/mol which generates good agreement between the calculated and measured values of H for the linear alkanethiols of Table 4-1.

The adsorption energies of molecules with the sulfur atom at the center of the molecule such as dialkylsulfides and thiophene can also be predicted with this model. Since HS represents the contribution of a sulfur atom (when added to the chain without altering the number of methyl groups), its value (24.1kJ/mol) can be used with the appropriate hydrocarbon contribution to calculate H for these molecules. As shown in Table 4­1, the calculated adsorption energies ("Phys(calc)") agree with the experimental results.

A challenge for the model was the rationalization of the experimental adsorption energies of the three highly-branched thiols: t-butanethiol, 2-propanethiol, and neopentanethiol. If the contribution of all the atoms in t­butanethiol are considered, the additive model significantly overpredicts the value of the adsorption energy by the contribution of a methyl group. However, excellent agreement is obtained if only the contributions of the groups that are in proximity to the surface in the most favorable configuration (as shown in Figure 4-5) are included. For t­butyl thiol, it can be seen that only three of the four groups attached to the central carbon will be equidistant from the surface and able to fully interact.

Lastly, it should be pointed out that the physisorption energies for the alkyldithiols cannot be understood in the simple terms reported above. While 1,6-hexanedithiol appears to rapidly chemisorb so that no value for the physisorption energy could be determined, the two butanedithiols physisorb with energies that are smaller than expected. Although the butanedithiols also chemisorb less strongly than expected, the origin of the reduction in physisorption energy remains unexplained for lack of sufficient data.



Figure 4-5. Sketch of possible orientation of neopentanethiol, 2-propanethiol, and t­butanethiol physisorbed on Au(111) surface. Only those groups closest to the surface contribute to binding. The dark spheres represent carbon atoms, the lighter spheres represent hydrogen atoms, and the spheres with "S" inscribed are sulfur atoms.

4.2.3. Chemisorption Enthalpy

Figure 4-4 also displays the activation energy for desorption derived from the second feature in the TPD spectra of alkanethiols (plotted as , also listed under the heading "Chem 1" in Table 4-1). For this feature, the adsorption energy does not change with chain length and has an average value of 126 ± 2 kJ/mol. Since chemisorption to the surface probably occurs through the formation of a chemical bond with the sulfur atom21, the length of the chain would not be expected to play a large role in the desorption enthalpy of the chemisorbed molecules.

At high coverage, the adsorption of alkanethiols on Au(111) leads to the formation of a well-ordered SAM where the activation energy for desorption was found previously to be between 126 -146 kJ/mol22. This adsorption energy is approximately the same for both alkanethiols and dialkyldisulfides. Dimethyldisulfide has been reported to desorb from Au(111) as a disulfide with an energy of approximately 117 kJ/mol17. The average adsorption energy for the chain-length independent feature in Figure 4-4 is consistent with the literature.

In the present TPD measurements, it was observed that alkanethiols composed of eight carbon atoms or less exhibit both a physisorption and a chemisorption feature. For octanethiol and longer chain thiols, the physisorption feature becomes difficult to detect with TPD and a third, higher energy peak becomes apparent at 148 ± 3 kJ/mol, which is 21 kJ/mol greater than the desorption energy of the first chemisorption peak (this will be discussed further in Section 4.2.5.). The disappearance of the physisorption feature between 100 and 126 kJ/mol is due to the kinetics of chemisorption. As the surface temperature is increased, the rate of chemisorption from the physisorbed population also increases. For the short-chain thiols, the temperature of desorption of the physisorbed molecule is low, so these molecules will desorb from the surface before the activation barrier to chemisorption can be easily crossed. For longer-chain thiols, however, the chemisorption channel will be able to compete successfully with the desorption channel and will deplete the observed physisorbed population.

4.2.4. Physisorption vs. Chemisorption

As the chain length of the linear alkanethiols is increased, the physisorption energy is also observed to increase. As a result, the physisorption energy will become larger than the chemisorption energy at chain lengths greater than C14H30 as shown in Figure 4-4. To confirm this trend, the activation energy for desorption of gas-phase deposited hexadecanethiol was measured to be 150 ± 15 kJ/mol. For comparison to the literature, Nuzzo et al.21 determined that solution-deposited hexadecanethiol desorbs from gold with a desorption enthalpy of approximately 167 kJ/mol. This energy was significantly higher than the previously reported "normal thiolate" adsorption energy of 117 kJ/mol17 and was explained by Nuzzo as being caused by a stabilization through interchain interactions.

In the present context, interchain attraction does not need to be invoked to justify the increased value of the desorption energy since the continuation of the physisorption trend observed for the shorter chains predicts that physisorbed hexadecanethiol can still be bound to the surface at temperatures where the "chemical" S-Au bond is already broken. Furthermore, at the temperature where hexadecanethiol desorbs (>500K) it is highly unlikely that the molecules will continue to form islands. At these temperatures, the molecules are expected to desorb from an uncondensed two-dimensional gas state since the energy needed for the molecules to detach from the island borders is lower than that needed to desorb from the gold.

To extend this trend, additional experiments were performed with longer chain lengths. However, because of the low vapor pressure of octadecanethiol, and docosanethiol, the experiments were carried out with a modified procedure. The gold crystal was cleaned as usual. The UHV sample chamber was vented to nitrogen and a solution of one of the thiols in ethanol was placed on the gold surface. The machine was pumped down and TPD was carried out as usual. The TPD peaks were not as sharply defined but clearly showed an activation energy for desorption that continued to increase with chain length, shown in Figure 4-4.

4.2.5. Higher Energy Chemisorption

The appearance of an additional high energy TPD peak at 148 ± 3 kJ/mol (shown as in Figure 4-4, listed in the column "Chem 2" in Table 4-1) was unexpected. This peak appears under non-equilibrium growth conditions when the surface has been saturated with adsorbates at high fluxes and low surface temperatures. To clarify its origin, several experiments were conducted by dosing the gold surface with decanethiol at 298 K and then waiting for several minutes before performing the TPD. Figure 4-6 shows the results of three experiments with anneal times ranging from 0 to 3600 sec. In this series of spectra, the area of the higher energy peak is largest with no anneal and consistently becomes smaller with longer anneal time. However, the temperature locations of the two peaks do not shift significantly.

Figure 4-6. Temperature programmed desorption spectra of decanethiol annealing experiments. A layer of decanethiol on Au(111) is annealed at 343 K for a) 0 seconds, b) 900 seconds, and c) 3600 seconds. The higher energy peak (143 kJ/mol) anneals into the lower energy peak (124 kJ/mol).

After possibly going through an intermediate state, the normal chemisorption feature becomes larger at the expense of the higher energy peak, implying that the higher energy population anneals into the lower energy state. This is surprising since it is the reverse of what is expected and commonly observed.

A possible explanation that would account for this apparent contradiction is that the higher energy peak is indicative of adsorbates desorbing as thiols while the lower energy peak corresponds to desorption of disulfides. Both thiol and disulfide desorption from Au(111) surfaces have been observed by Nishida et al.23. A comparison of Seller's24 calculation of methanethiol adsorption on Au(111) with a Born-Haber calculation of desorption as a dimethyldisulfide indicates that desorption as a disulfide is energetically favorable. Under a high­flux dosing condition, there may be enough hydrogen present on the surface to facilitate the desorption of thiol species. After annealing (or under slower growth conditions) the hydrogen would desorb, leaving only the disulfide desorption channel available.

A second, and perhaps more likely explanation, can be derived on the basis of a recent STM study of methanethiol adsorption from the gas phase onto reconstructed Au(111) by Feher et al.25 With STM images, the authors show that "the dosing rate affects the final structure of the monolayer." At high dosing rates, a very large number of regularly spaced islands of gold atom vacancies are formed which only upon annealing merge by Ostwald ripening into large depressions while the monolayer (both in and out of the depressions) assumes the normal c(4x2) structure. If one makes the reasonable assumption that before and after the annealing (which involves motion of the gold atoms on the surface) the binding energy of the sulfur atoms to the gold is different, the present results can then be explained.

The fact that the higher energy desorption peak is observed only for chains with more than eight carbons can also be explained in the context outlined above. As the annealing rate of the chemisorbed phase must be a function of the chain length, both dimerization and gold reorganization rates for chains shorter than eight carbons may be fast enough to allow the annealing to be completed during the temperature ramping of the TPD procedure and before the desorption temperature is reached.

4.2.6. Steric Hindrance Effects

In an attempt to gain further information on the bonding of thiols to the gold surface, other related sulfur-containing molecules were studied. To address the question of disulfide formation, sterically-hindered thiol groups were chosen to test the hypothesis that the measured adsorption energy would be greater if dimer formation was not possible due to steric constraints. The TPD spectrum of t-butanethiol (with a thiol group attached to a tertiary carbon) showed a low energy physisorption peak which was consistent with the additive model (see section 4.2.2.) and a second feature which, contrary to the stated hypothesis, corresponded to an adsorption energy of only 107 kJ/mol. No desorption feature around 124 kJ/mol was observed even after annealing.

Neopentanethiol (a thiol group attached to a primary carbon with a t-butyl end) showed instead a "normal" chemisorption feature at 128 kJ/mol. Although S-H bond strengths differ slightly between the two molecules (368 kJ/mol for ethanethiol and 364 kJ/mol for t­butanethiol26), it appears that the sterically hindered thiols are not able to bind as closely to the surface. Steric hindrance quickly ceases to be a factor when an additional methylene unit (as in neopentanethiol) allows the sulfur atom to interact more strongly with the metal surface without pulling the rest of the molecule into the repulsive wall of the physisorption potential well. This extra repulsive energy due to the presence of sterically hindering groups is 20 kJ/mol (or approximately 16% of the desorption energy) and will produce a relatively small change in the distance of the sulfur atom from the gold surface.

4.2.7. Dialkyldisulfides and Dialkylsulfides

Another set of experiments directly explored the adsorption behavior of diethyldisulfide on Au(111). TPD experiments indicated the presence of only one desorption peak for diethyldisulfide at 124 kJ/mol, comparable in energy to the chemisorption peak for the alkanethiols. Since the physisorption energy for diethyldisulfide was expected to be comparable to either ethanethiol (if it cleaved upon adsorption) or butanethiol (if it remained intact), it was surprising that no physisorption peak was observed. However, the kinetics of chemisorption of the disulfide species are different than that of the thiols. Experiments by Dubois et. al.6 found that the sticking probability for dimethyldisulfide was larger than for methanethiol by several orders of magnitude. They suggested that for the shorter chain species, the disulfide bond was more efficiently cleaved and that disulfides did not exhibit the same activation barrier for adsorption as the thiols. This is not surprising since the overall thermodynamics of chemisorption of thiols on Au(111) involves the breakage of the S­H bond as well as the formation of H-Au bonds, H-H bonds, and other reaction pathways. Instead, disulfide chemisorption as thiolates could occur by a process which would only cleave the disulfide bond and form two sulfur-gold bonds. However, according to Fenter et al.27, disulfides could potentially chemisorb intact without dissociation.

In a final set of experiments, the dialkylsulfides (diethylsulfide and dibutylsulfide) were studied to investigate the probability of cleavage of the C-S bond. In TPD experiments, both compounds showed physisorption enthalpies consistent with the correlation with the bulk heat of vaporization. However, neither molecule showed a chemisorption feature. These findings are in apparent contradiction to those of Porter28 who demonstrated C-S bond cleavage in several organosulfides under electro-chemical conditions. Other studies, however, support the present findings. In an earlier study utilizing dialkylsulfides, Troughton et al.29 found that dialkylsulfides formed a poorly organized layer from solution. They found that the poor quality of the layer was not due to the decomposition of the dialkylsulfide, but rather to the fact that the adsorbates were only weakly bonded through the sulfur group. A recent study by Beulen et al.30 has also shown that a variety of dialkylsulfides remain intact on the surface and experience no C-S bond cleavage.

Thiophene, a heterocyclic molecule containing four carbon atoms and a sulfur atom, also showed only a physisorption feature for adsorption on gold as in the case of the dialkylsulfides. The single desorption peak occurs at 60 kJ/mol which is consistent with desorption from a physisorbed layer. No other desorption peaks indicative of chemisorption were found, even for samples prepared under high dosing and/or long annealing procedures. The thiophene measurements were stimulated by a recent paper of Dishner et al.31 in which the formation of well­ordered thiophene monolayers on Au(111) was observed. The present work confirms the theoretical results of Elfeninat et al.32 who concluded that thiophene could not chemisorb on Au(111) surfaces.

4.3. Results and Discussion: Kinetics

4.3.1. Precursor mediated chemisorption: a brief introduction

Physisorbed precursor states have been investigated with respect to their effect on chemisorption kinetics33,34, and the dynamics of adsorption and desorption35. Typically, for a direct chemisorption process, sticking coefficients increase with increased surface temperature. Systems that show a decrease in sticking coefficient with increased surface temperature are thought to involve a precursor state36-39. The precursor is typically a more weakly bound state such as a physisorbed state. The decrease of sticking coefficient into the chemisorbed state arises from a kinetic competition between desorption from the precursor state and the crossing of an activation barrier to chemisorption. Precursor chemisorption has been found, for example, for such systems as O2 on Pt(111)40, CO and CO2 on Ni(100)41,42, and N2 on W(100)43. These systems have been explored using molecular beam deposition, studying the dependence of the sticking on the angle of incidence of the molecular beam with the surface, the kinetic energy of the beam, and the surface temperature. With the exception of alkanes on transition metals44-49, larger molecules have not been as extensively studied. In these systems, a decrease in initial sticking coefficient has been observed with increasing incident energy and increasing surface temperature.

The gold/alkanethiol system with its relatively low chemisorption energy and physisorption dependence on chain length presents the opportunity to probe the phenomenon of precursor-mediated chemisorption in a way not possible with the diatomic systems. Since only the sulfur head-group chemisorbs to the gold surface with an energy that does not depend on chain length, the chemisorption well remains constant for different chain lengths. Increasing the alkane chain length, however, increases the physisorption (precursor) well depth. In this way, the activation barrier to chemisorption from the physisorbed state can be systematically varied and the process of precursor mediated chemisorption explored. Whether the barrier between the two wells is independent of the chain length or is decreased by the deepening of the physisorption well had to be determined experimentally. Since the physisorption interaction is distributed throughout the molecule while the chemisorption well is localized, the barrier was predicted to be independent of the chain length.

4.3.2 Chemisorption from a physisorbed layer

Helium atom reflectivity is unable to readily distinguish between chemisorbed and physisorbed adsorbates. As a result, the rate of chemisorption during dosing cannot be easily determined at surface temperatures significantly below the physisorption desorption temperature. However, by performing TPD after deposition, it is possible to distinguish between the two adsorbed species to determine if there is chemisorption of molecules from a physisorbed precursor state.

To perform these experiments, the Au(111) crystal was held at a surface temperature low enough to accumulate a physisorbed monolayer. By increasing the amount of wait time between the deposition and the TPD, the rate of chemisorption from the physisorbed population was determined. As the chemisorbed population increases by depletion of the physisorbed population, the relative populations of adsorbates will change. In this series of experiments, it is assumed that the contribution of each molecule to the total area of the TPD peaks is the same irrespective of the physisorbed or chemisorbed status of the molecules that produce the specularity drop. This assumption is probably not strictly correct, but for the purpose of comparison of similar conditions this should not represent a serious problem.

Figure 4-7 shows a series of TPD experiments for butanethiol deposited onto a Au(111) crystal at 208 K. Curve a) shows the presence of mostly physisorbed species when the waiting time is zero. The series of curves b), c), and d) represent progressively longer waiting times showing conversion of the physisorbed species into chemisorbed species. The final curve d), the result of annealing the sample for 3600 seconds, exhibits mostly chemisorbed species. Figure 4-8 displays the ratio of chemisorbed peak area to the total peak area derived from each of the TPD profiles in Figure 4-7. The data clearly shows that conversion into chemisorbed species increases with waiting time. The slope of the first three points at shorter waiting times indicates that, at the surface temperature of 208 K, the rate of chemisorption of physisorbed species is on the order of 4 × 10-4 sec-1.

The TPD profiles in Figure 4-7 show also that the physisorption peak location does not change with decreasing coverage but that the chemisorption peak location shifts to higher temperatures with increasing coverage. Annealing at higher temperatures would be needed to ensure that the gold surface has also equilibrated after chemisorption has occurred. Therefore no conclusions can be drawn from the presence of these small shifts but they are noted for the sake of completeness.

Figure 4-7. TPD spectra of a butanethiol layer on Au(111) deposited and "annealed" at 208 K for various times. The higher energy peak (chemisorbed) becomes more intense over time at the expense of the lower energy peak (physisorbed).

Figure 4-8. The chemisorbed fraction of total peak area from the data of Figure 4-7.

The final point in Figure 4-8 at 3600 sec indicates the limit of full conversion into a chemisorbed layer. The y-intercept of 0.35 corresponding to zero wait time indicates the presence of partial conversion of the physisorbed population as the surface temperature is ramped during the TPD cycle. As the surface temperature increases there is an increase in the rate of chemisorption from the physisorbed population, therefore some fraction of physisorbed molecules will be converted before desorption.

Hexanethiol was also deposited on Au(111) at 208 K in a similar set of experiments. Once again, the TPD curve obtained after zero waiting time (the solid curve of Figure 4-9) shows the presence of both physisorbed and chemisorbed species. Unlike butanethiol, there appears to be no further conversion to the chemisorbed state even after waiting for 2460 sec.

Figure 4-9. Temperature programmed desorption spectra of hexanethiol layer deposited at 208 K after a waiting time of a) 0 seconds and b) 2460 seconds. The relative intensity of lower energy peak (physisorbed) and higher energy peak (chemisorbed) remain relatively unchanged in this range of waiting time.

A close examination of the increase in peak area of the chemisorbed peak from no anneal to 2460 sec indicates a much slower growth rate of about 110-5 sec-1. This reduction in chemisorption rate compared to butanethiol could be due either to a change in the activation barrier from the physisorbed to the chemisorbed state or a change in the pre-exponential factor. This question is considered in the next section.

4.3.3. Chemisorption from a steady-state physisorbed population

A more convenient experiment is to study the growth of the chemisorbed layer in real­time at a constant surface temperature and with a small, constant physisorbed population on the surface. By conducting growth experiments in a temperature regime high enough that the desorption rate of the physisorbed molecules is significant (approximately 20 K below the peak physisorption desorption temperature), a steady state of physisorbed molecules can be achieved. Additional decreases of specular intensity after this physisorbed population has been established can then be assigned to an increasing population of chemisorbed species on the surface. However, since under these conditions there is a chance that direct chemisorption may contribute to the chemisorption rate, this effect must be taken into account.

A steady-state approximation can be applied to the simple Langmuir growth equation (Equation 2-1) to determine the coverage expected when the rates of adsorption and desorption are balanced:

ss = ka / ( ka + kd ) (4-4)

where ss is the steady state coverage of adsorbates, ka is the adsorption rate (controlled in part by the partial pressure of the adsorbing molecules), and kd is the desorption rate (controlled by the surface temperature).

Figure 4-10. Specular decay of Au(111) caused by adsorption of ethanethiol from the gas phase at a surface temperature of 223 K. The solid curve is the raw helium specular signal. The dotted line depicts the partial pressure of ethanethiol in the UHV sample chamber.

As shown in Figure 4-10, the specularity initially drops to a finite value (0.75) when exposed to a constant flux of molecules which corresponds to a steady state coverage of physisorbed molecules. However, even though the partial pressure of thiols remains constant for 1600 sec, a slower decrease in specularity is observed. This is attributed to conversion of the physisorbed molecules into chemisorbed species, thereby "permanently" occupying a fraction of the surface sites and decreasing the fraction of surface available to the physisorption equilibrium. At the end of the growth experiment, the thiols are evacuated from the chamber and a rapid partial recovery of specularity is found. The rate of recovery is consistent with desorption of the physisorbed species. The chemisorption rate is identified as the rate of the slower specular decay and is determined by differentiating this portion of the decay curve and normalizing to the coverage. By conducting this same type of experiment over a range of temperatures, an Arrhenius plot for the physisorption to chemisorption conversion process can be constructed.

Figure 4-11 shows these Arrhenius plots for ethanethiol, butanethiol, hexanethiol, and decanethiol. For all four species, the same slope is found with different y-intercepts, as shown in Table 4-2. Since the rates found from the TPD "annealing" experiment in the previous section fall within the data on this Arrhenius plot, any direct process contribution to the rate of chemisorption must be small compared to the rate of precursor mediated chemisorption.

Figure 4-11. Arrhenius plot of the chemisorption rate from a steady-state physisorbed population for ethanethiol (), butanethiol (), hexanethiol (), and decanethiol ().


Table 4-2. Fit Parameters Derived from Figure 4-11.
Ethanethiol
Butanethiol
Hexanethiol
Decanethiol
Slope (kJ/mol)
31.0 4.7
29.0 4.7
29.6 4.4
27.3 5.8
y-intercept ln(s-1)
10.5 2.5
6.0 2.2
5.3 1.8
5.9 1.6

While the slopes of the Arrhenius plots have sizable errors, they all correspond to 29 kJ/mol within 5%, independent of chain length. Nuzzo6 had estimated the barrier to chemisorption of methanethiol from the gas phase as 25 kJ/mol. However, his model predicts that the activation barrier would decrease with chain length. This is due to the fact the chemisorption well for all alkanethiol chain lengths remains fixed and the physisorbed well becomes deeper with increased chain length. In his model, the transition state is viewed as being stabilized by attractive dispersion forces to a degree comparable to that reflected by the increased heats of adsorption of the molecular precursors6. Therefore, increased chain length should decrease the activation barrier. Within the limits imposed by the errors, the present data suggest that while the physisorbed well depth increases the location of the curve crossing with the chemisorbed state remains the same, and therefore the activation barrier to chemisorption remains constant with increasing chain length.

In this model of fixed activation barrier, the rate of chemisorption is controlled in part by the lifetime of the precursor state. Therefore, the overall rate of chemisorption is a branching ratio between chemisorption of the physisorbed molecules and their desorption. With a fixed activation barrier, the rate for chemisorption increases overall with increased surface temperature. The fact that physisorbed TPD features become difficult to detect above chain lengths greater than octanethiol is consistent with this model, since the longer chains remain on the surface at higher temperatures resulting in faster chemisorption.

The identification of the "bottleneck" to chemisorption is difficult. Typically, for diffusion controlled processes on surfaces, the activation barrier increases with chain length while the pre-exponential remains relatively constant50-52. The present data show that the pre­exponential factor is greatest for ethanethiol and smallest for the longer chains. Indeed, hexanethiol, octanethiol, and decanethiol have approximately the same pre-exponential factor. While the pre-exponential factor should not be too greatly emphasized since it is notoriously difficult to measure, these data lead to the tentative conclusion that diffusion of the molecules across the surface is not likely to be responsible for this "bottleneck" to chemisorption. It seems more likely that orientation of the sulfur group with respect to the surface is important, and that shorter chain lengths such as ethanethiol have an advantage over the longer ones. However, this advantage is bound to saturate with increased chain length because of the increased chain flexibility. In this way the effect would become, for the longer chains, independent of chain length.

4.4. Conclusions53

Alkanethiols physisorb to the surface of gold through van der Waals interactions that generate adsorption energies on the order of their bulk heats of vaporization. The molecules tend to bind more strongly to the gold surface than to each other in the bulk by a factor of about 1.15. This holds true for a wide variety of sulfur containing alkanes of differing degrees of branching as well as thiophene. The physisorption enthalpy per methylene group for alkanethiols is on the order of 6.1 kJ/mol, a value very similar to that observed for n-alkanes and 1-alkenes. The physisorption energy contributed by the thiol group (SH) is on the order of 33 kJ/mol while the sulfur atom alone contributes ~24 kJ/mol.

Alkanethiols show a chemisorption enthalpy of 126 kJ/mol which is independent of chain length. This implies that for chain lengths greater than 14 carbons, the physisorption enthalpy should be higher than the chemisorption energy. Indeed, longer thiols exhibit a single TPD feature at temperatures which are consistent with the projection of the physisorption energy with chain length. The TPD spectra for thiols shorter than octanethiol show both physisorption and chemisorption desorption features. Octanethiol and longer chain thiols exhibit both the normal chemisorption feature and a higher energy peak at 148 kJ/mol. This higher energy peak can be annealed into the lower energy chemisorption peak. This is likely due to a rearrangement of the gold surface atoms which takes some time to occur and has an influence on the value of the sulfur­gold surface bond. The lack of physisorption features for chain lengths longer than octanethiol is attributed to the fact that the chemisorption rate is large enough around 350 K that as the TPD temperature ramps through this region, the molecules will chemisorb instead of desorbing from the physisorbed state.

Sterically hindered thiol groups show less binding energy with the surface. Both t­butanethiol and 2-propanethiol show physisorption enthalpies consistent with predictions based on an additive contribution to binding per methylene unit, and a lower-than-normal chemisorption enthalpy of 107 kJ/mol. Adding one methylene group between the thiol and the t-butyl group forms neopentanethiol. This species behaves "normally" in that it has a predicted physisorption enthalpy and a chemisorption enthalpy at 124 kJ/mol. The extra methylene group removes the steric hindrance of the t-butyl group.

Diethylsulfide and dibutylsulfide both show only physisorption peaks. This is consistent with earlier studies of dialkylsulfides on Au(111) that showed them to be poorly organized, physisorbed monolayers. Diethyldisulfide, however, showed only a chemisorption feature. This is consistent with Nuzzo's finding that the disulfide species tend to chemisorb more readily than the comparable thiols. Disulfides probably have a lower barrier to chemisorption due to the fact that they can chemisorb without the need to eliminate hydrogen.

By probing the rate of conversion from physisorbed layer to chemisorbed layer at 208 K for butanethiol and hexanethiol, it is found that the rate decreases from 4 × 10-4 sec-1 for butanethiol to 3.3 × 10-5 sec-1 for hexanethiol. These rates are similar to those found in constant exposure experiments that create a steady state coverage of physisorbed molecules from which chemisorption can occur. Arrhenius plots generated with the chemisorption rate for ethanethiol, butanethiol, hexanethiol, and decanethiol indicate that the activation barrier to chemisorption for these molecules is about 29 kJ/mol irrespective of the chain length. This is consistent with a model involving a constant chemisorption well depth as the physisorption well depth increases with increased chain length. In this case, the increased physisorption energy does not affect the height of the barrier but does increase the residence time of the molecules on the surface at temperatures where the chemisorption occurs more easily. The systematic study described in this paper has clarified several issues related to the adsorption of alkylsulfides and other sulfur-containing molecules on Au(111) and has provided new clear challenges for those interested in the theoretical simulation of the behavior of this very popular system

References

(1) See references in Atre, S. V.; Lieberg, B.; Allara, D. L.; Langmuir. 1995, 11, 3882.

(2) Dubois, L. H.; Nuzzo R. G. Annu. Rev. Phys. Chem. 1992, 43, 437.

(3) Swalen, J. D.; et al. Langmuir, 1987, 3, 932.

(4) Ulman, A. An Introduction to Ultrathin Organic Films Academic Press: New York, 1991.

(5) Ford, J. F.; Vickers, T. J.; Mann, C. K.; Schlenoff, J. B.; Langmuir. 1996, 12, 1944.

(6) Dubois, L. H.; Zegarski, B. R.; Nuzzo, R. G. J. Chem. Phys. 1993, 98, 678.

(7) Walczak, M. M.; Alves, C. A.; Lamp, B. D.; Porter, M. D. J. Elec. Chem. 1995, 396, 103.

(8) Salem, L. J. Chem. Phys. 1962, 37, 2100.

(9) Thomas, R. C.; Sun, L.; Crooks, R. M.; Ricco, A. J. Langmuir 1991, 7, 620.

(10) Camillone III, N.; Leung, T. Y. B.; Scoles, G. SPIE, OE/LASE Proceedings 1994, 2125.

(11) Gerlach, R.; Polanski, G.; Rubahn, H.-G. submitted to Appl. Phys. A., 1997.

(12) Poirier, G. E.; Pylant, E. D. Science 1996, 272, 1145.

(13) Li, J.; Liang, K. S.; Camillone III, N.; Leung, T. Y. B.; Scoles, G. J. Chem. Phys 1995, 102, 5019.

(14) Schwartz, P. manuscript in preparation.

(15) Liu, G.-Y.; Song, X. Langmuir, 1997, 13, 127.

(16) Eberhardt, A.; Schwartz, P. private communication.

(17) Nuzzo, R. G.; Zegarski, B. R.; Dubois, L. H. L. Am. Chem. Soc. 1987, 109, 733.

(18) Dubois, L. H.; Zegarski, B. R.; Nuzzo, R. G. J. Am. Chem. Soc. 1990, 112, 570.

(19) Sexton, B. A.; Hughes, A. E. Surf. Sci. 1984, 140, 227.

(20) Teplyakov, A. V.; Gurevich, A. B.; Yang, M. X.; Bent, B. E.; Chen, J. G., preprint.

(21) Nuzzo, R. G.; Dubois, L. H.; Allara, D. L. J. Am. Chem. Soc. 1990, 112, 558.

(22) Nuzzo, R. G.; Fusco, F. A.;. Allara, D. L. J. Am. Chem. Soc. 1987, 109, 2358.

(23) Nishida, N.; Hara, M.; Sasabe, H.; Knoll, W. Jpn. J. Appl. Phys., 1996, 35, 5866.

(24) Sellers, H. Surf. Sci. 1993, 294, 99.

(25) Feher, F. J.; Hemminger, J. C.; Dishner, M. H. Langmuir, 1997, 13, 2318.

(26) Lide, D. R. Handbook of Chemistry and Physics 73rd ed. (CRC Press, London, 1992).

(27) Fenter, P.; Eberhardt, A.; Eisenberger, P. Science 1994, 266, 1216.

(28) Porter, M. D.; Zhong, C. J. J. Am. Chem. Soc., 1994, 116, 11616.

(29) Troughton, E. B.; Bain, C. D.; Whitesides, G. M.; Nuzzo, R. G.; Allara, D. L.; Porter, M. D. Langmuir, 1988, 4, 365.

(30) Beulen, M. W. J.; Huisman, B.-H.; van der Heijden, P. A.; van Veggel, F. C. J. M.; Simons, M. G.; Biemond, E. M. E. F.; de Lange, P. J.; Reinhoudt, D. N. Langmuir, 1996, 12, 6170.

(31) Dishner, M. H.; Hemminger, J. C.; Feher, F. J. Langmuir 1996, 12, 6176.

(32) Elfeninat, F.; Fredricksson, C.; Sacher, E.; Selmani, A. J. Chem. Phys. 1995, 102, 6153.

(33) Izawa, M.; Kumihashi, T. J. Chem. Phys. 1995, 103, 9418.

(34) Cassuto, A.; King, D. A.; Surf. Sci. 1981, 102, 388.

(35) Brivio, G. D.; Grimley, T. B. Surf. Sci. Rep. 1993, 17, 1.

(36) Sullivan, D. J.; Flaum, H. C.; Kummel, A. . J. Phys. Chem. 1993, 97, 12051.

(37) Doren, D. J.; Tully, J. C. J. Chem. Phys. 1991, 94, 8428.

(38) Rettner, C. T.; Mullins, C. B. J. Chem. Phys. 1991, 94, 1626.

(39) Lee, C-Y; DePristo, A. E. J. Chem. Phys., 1986, 85, 4161.

(40) Luntz, A. C.; Williams, M. D.; Bethune, D. S. J. Chem. Phys., 1988, 89, 4381.

(41) D'Evelyn, M. P.; Steinruck, H. -P.; Madix, R. J. Surf. Sci. 1987, 180, 47.

(42) D'Evelyn, M. P.; Hamza, A. V.; Gdowski, G. E.; Madix, R. J. Surf. Sci., 1986, 167, 451.

(43) Rettner, C. T.; Scheizer, E. K.; Stein, H.; Auerbach, D. J. J. Vac. Sci. Tech. A, 1989, 7, 1863.

(44) Soulen, S. A.; Madix, R. J. Surf. Sci 1995, 323, 1.

(45) Hamza, A.V.; Steinruck, H.-P.; Madix, R. J. 1987, 86, 6506.

(46) Kelley, D.; Weinberg, W. H. J. Chem. Phys. 1996, 105, 3789.

(47) Mullins, C. B.; Weinberg, W. H. J. Chem. Phys.1990, 92, 4508.

(48) Mullins, C. B.; Weinberg, W. H. J. Vac. Sci. Tech. A 1990, 8, 2458.

(49) Mullins, C. B.; Weinberg, W. H. Springer Series in Surface Sciences, Vol 34 Surface Reactions edited by Madix, R.J.; Springer-Verlag: Berlin, 1994.

(50) Huang, D.; Chen, Y.; Fichthorn, K. A. J. Chem. Phys. 1994, 101, 11021.

(51) Arena, M. V.; Deckert, A. A.; Brand, J. L.; George, S. M. J. Phys. Chem. 1990, 94, 6792.

(52) Brand, J. L.; Arena, M. V.; Deckert, A. A.; George, S. M. J. Chem. Phys. 1990, 92, 5136.

(53) The results of this chapter were published in J. Phys Chem. B. 1998, 102, 3456.


Next Chapter
Return to the Table of Contents
Return to the Theses Page
Return to the Scoles Group Homepage