Surface Stress in the Self-Assembly of Alkanethiols on Gold

See allHide authors and affiliations

Science  27 Jun 1997:
Vol. 276, Issue 5321, pp. 2021-2024
DOI: 10.1126/science.276.5321.2021


Surface stress changes and kinetics were measured in situ during the self-assembly of alkanethiols on gold by means of a micromechanical sensor. Self-assembly caused compressive surface stress that closely followed Langmuir-type adsorption kinetics up to monolayer coverage. The surface stress at monolayer coverage increased linearly with the length of the alkyl chain of the molecule. These observations were interpreted in terms of differences in surface potential. This highly sensitive sensor technique has a broad range of applicability to specific chemical and biological interactions.

Molecular and biomolecular layers are scientifically appealing for a wide range of potential applications (1, 2). Alkanethiols, which are known to self-organize into well-ordered, densely packed films, represent a model molecular system for controlling surface properties (3,4). These self-assembled monolayers (SAMs) are used in applications such as microcontact printing (5) and voltametric microsensors (6), and they have recently been applied to molecular host-guest recognition (7).

There is little information available on the mechanical properties of SAMs, particularly concerning the nature of surface stress in films during the formation process, because it is difficult to follow the structural evolution of monolayer self-assembly. One recent approach (8) used scanning tunneling microscopy to infer the growth kinetics of alkanethiol SAMs indirectly from snapshot images obtained at various coverages. Here, we used micromechanical sensors to gather quantitative data on surface stress changes that develop during the self-assembly process of HS-(CH2)n−1-CH3 forn = 4, 6, 8, 12, and 14, where n is the number of carbon atoms in the alkyl chain (Fig.1A). Micromechanical sensors are attracting increasing interest in science and technology for in situ process control (9-11) because they feature high sensitivity, small size, and compatibility with Si microelectronic fabrication (12).

Figure 1

(A) Alkanethiols assemble spontaneously from solution or vapor onto gold. The specific monolayer component is drawn in a space-filling model for butanethiol (n = 4), octanethiol (n = 8), and dodecanethiol (n = 12). (B) Scanning electron micrograph of the Au-coated SiNx cantilever showing the sensor's receptor surface. (C) The sketch shows a cut perpendicular to the surface through the sensor layers as used for the self-assembly process. Coverage of chemisorbed alkanethiols on the sensor θt = 0: The sensor is covered by adsorbates in air (average thickness ∼1 nm), which causes a bending,C 1, of the sensor. (D) θt = 1: When the sensor is exposed to alkanethiol vapor, a SAM is formed. This causes a bending,C 2, which compresses the underlying substrate.

V-shaped micromechanical silicon nitride (SiNx) cantilevers with a 20-nm gold receptor layer evaporated on one side were used as sensors for gas-phase adsorption of alkanethiols (Fig. 1B). Scanning force microscopy techniques (13) were used to detect sensor deflections down to the picometer scale. A laser beam reflected off the cantilever's apex onto a quadrant photodiode used as a position-sensitive detector (PSD) indicated sensor displacement. Alkanethiol vapors were generated by placing a few microliters of alkanethiol in the center of a glass beaker, which was then closed by a shutter. After thermal equilibrium was reached, the shutter was opened, thereby exposing the sensor to alkanethiol vapor (14).

The sensor deflection, derived from the PSD voltage, was measured as a function of time for experiments with alkanethiols of various chain lengths (Fig. 2A). In all cases, we observed a strong response in deflection, which saturated at a permanent value (between ∼50 and 200 nm) corresponding to the expansion of the receptor side of the sensor (Fig. 1D). This saturation developed on a time scale similar to that previously reported for monolayer formation by chemisorption on gold (15). In contrast, reference experiments performed with octane vapor showed no comparable response.

Figure 2

(A) Deflection, Δz, and changes in surface stress, Δσ, of the sensors are plotted as a function of time for exposure to alkanethiol and a reference vapor. Reference experiments consisted of exposing the sensor to vapors of alkanes—molecules that do not chemisorb on gold—and showing the background noise. The reference curve used octane and is representative of all controls. In the reference experiment, no deflection was observed, except for a small signal attributed to the removal of the shutter. In contrast, the sensors started bending immediately after exposure to alkanethiol vapors. We estimate the exposure of the sensor to butanethiol molecules to be ∼1.5 × 1020 molecules cm−1s−1 under our experimental conditions. A faster response for lighter alkanethiols, corresponding to their higher vapor pressures, was observed. Each alkanethiol curve was fitted by a LM adsorption isotherm, which determines the zero point of the stressograms. In our experiments with butanethiol and octanethiol, LM fits the entire stress curve. For dodecanethiol, we found a deviation from the LM beginning above ∼80% coverage, indicating a decrease of the sticking coefficient. From our data, we calculate the average tension σt = σsat t f −1 in a monolayer of thickness t f to be 0.16 ± 0.03 GPa for all five SAMs studied. Before exposure, the sensor response was recorded in air for ∼1 min, which reflected the initial amount of adsorbates of the gold layer from our laboratory environment. Initial values of deflection before exposure do not influence the LM fitting procedure. (B) The change in surface stress at saturation coverage obtained for n = 4, 6, 8, 12, and 14 is plotted as a function of alkyl chain length. The value of Δσsat atn = 0 reflects the constant surface stress contribution from sulfur chemisorption, the formation of depressions in the gold, or both.

Several transduction mechanisms can contribute to these observations. Thermal effects are known to produce bendings as a consequence of the bimetallic effect (10). Self-assembly of alkanethiols on gold is exothermic, with an enthalpy of adsorption ΔE≈ −150 kJ mol−1 (16). Hence, chemisorption of a monolayer of alkanethiols on the sensor's receptor surface (∼1010 molecules) can produce ∼25 nJ of heat. The sensor's calculated transient bending caused by the reaction heat has the observed sign of deflection but is on the order of only 0.5 nm (10); thereafter, thermal effects are negligible. Nor can the gravimetric deflection resulting from the molecular loading (calculated to be ∼5 pm) account for the observed bending. On the basis of both the permanent nature of the deflection and its magnitude, we attribute the response to surface stress (17-19). Stoney's formula (20) relates the sensor curvature radiusR to the surface stress σ acting on the sensor,Embedded Image (1)where R 1= 3Δz/2L 2, L is length, E is Young's modulus, ν is Poisson's ratio of the sensor material, Δz is deflection, andt s is sensor thickness.

The alkyl chains in the monolayer exhibit a tilting away from the surface normal to reduce chain-to-chain separation and to optimize the chains' attractive intermolecular van der Waals interaction. We intuitively expected to observe a tensile surface stress, corresponding to a bending toward the SAM, from this effect. Surprisingly, all the chemisorbed alkanethiols we investigated caused compressive surface stress during self-assembly. Except for the initial few seconds of exposure, a Langmuir adsorption isotherm model (LM), for which θ ∝ 1 − exp(−κt), where θ is the coverage,t is the time, and κ is the reaction rate, fits the stress curves (21). LM describes the coverage dependence of alkanethiol adsorption both in solution (22) and from the vapor phase (15). Because the stress curves follow LM characteristics, we can conclude that the surface stress is proportional to the number of molecules adsorbed.

The saturated surface stress σsat generated by SAMs of alkanethiols increased linearly with chain length (Fig. 2B). From these data, we conclude that the compressive surface stress change is directly proportional to alkyl chain length. The molecular weight of linear alkanethiols is the principal determinant of the degree of structural order of SAMs on gold (4). In particular, short-chain monolayers of butanethiol have pronounced disorder and have been described as liquid-like films at room temperature (23). In contrast, SAMs with longer chains such as dodecanethiol form monolayers with a high degree of order. Consequently, our observations indicate that Δσ is insensitive to structural parameters.

In terms of electrostatic interactions, the apparent dipole moment of the SAM is considered to contain a contribution from the Au+-S head group and from the S-alkyl+ chain. Even at low coverage, the sulfurs are bound to the gold, and the −CH3 tail groups tend to emerge at the air-monolayer interface (4), providing an average apparent dipole moment. This apparent dipole moment increases linearly with n (24), resulting in a linear increase of electrostatic repulsion. Such dipolar repulsive forces in adsorbate-adsorbant systems are generally expected to produce surface stresses on the order of 10−3 N m−1(25), which is consistent with the magnitude of our measurements. In particular, the linear relations between Δσ and both θ and n are consistent with an electrostatic model.

Upon careful inspection, the stressograms display small, monotonic, step-like variations observable at higher magnification (Fig. 3). These variations are clearly above the noise level given by the reference experiment. The steps are typical for all stress curves performed withn ≥ 8, and we tentatively associate them directly with the self-assembly process. They may result from local changes of concentration in the immediate environment of the sensor (for instance, as a result of turbulence), which can lead to inhomogeneous chemisorption rates. In general, inhomogeneities in the self-assembly process can be detected in situ and in real time with a surface stress resolution of 10−7 N m−1 by means of standard-size sensors. This corresponds to a change of zepto (10−21) molar quantities in our experiments on SAMs.

Figure 3

The stressogram for dodecanethiol self-assembly at higher magnification displays step-like variations of surface stress. The reference, plotted for comparison, displays the background noise. The small variation, indicated by the arrows, is caused by an attomolar quantity of molecules.

The kinetics of SAM formation display clearly resolved minima at the beginning of each chemisorption process. X-ray photoelectron spectroscopy and second-harmonic generation studies of self-assembly of alkanethiols in solutions have led to the proposal that the initial phase of the reaction includes the replacement of residual adsorbates on the gold surface by chemisorbed alkanethiols (26). This process was confirmed for our samples by means of ellipsometry (27). All surface stress curves except the reference curve in Fig. 2A reveal the replacement process as a release of 11 × 10−3 to 19 × 10−3 N m−1 of residual surface stress (28) during the first ∼10 s of exposure to alkanethiols (Fig. 1C).

This replacement of one molecular layer by another can be extended to the specific binding of a molecule to a receptor layer. To demonstrate this concept, we used SAMs of ω-functionalized alkanethiols as acceptors for molecular recognition. We studied the influence of gas-phase hexylamine on mercaptohexadecanoic acid SAMs (Fig. 4). A clear decrease in Δσ was observed when hexylamine molecules “docked” onto the SAMs' carboxylic end groups. These observations are in qualitative agreement with a preliminary report of nonspecific binding of albumin on SAMs (29).

Figure 4

Model experiment of a sensor coated with mercaptohexadecanoic acid SAM (i) as a functionalized surface. Mercaptohexadecanoic acid acts as a specific receptor to bind hexylamine (ii) as an acceptor molecule. The resulting salt bridge formation (iii) is detected by a change in surface stress, which was normalized to that of the mercaptohexadecanoic monolayer. The arrow indicates the beginning of the exposure of the sensor to hexylamine vapors.


Stay Connected to Science

Navigate This Article