Introduction

Self-assembled monolayers (SAMs) have generated substantial interest recently.¹ The preparation procedure of such thin organic thin films is simple. More importantly, it is possible to control the physical and chemical properties and desired chemical functionality with a unique molecular architecture.²

SAMs have found many applications in many areas to date. One of the applications is as a friction reducing and anti-stiction layer in micro-electromechanical systems (MEMS), which has been extensively studied recently.^3-7 The performances of SAMs are closely related to their intrinsic chemical composition and structures. For example, terminal groups of SAMs determine the wettability, adhesion force, friction force, and shearing force, middle chains relate to their flexibility, elasticity, load-carrying capacity, and friction force, while head groups dominate their affinity, stability, integrity, and wear resistance.⁸

Besides, intermolecular interaction within SAMs can also play an important role. Introducing functional groups, such as diacetylene,⁹ peptide,^10-12 and sulfone,¹³ into straight hydrocarbon chains can construct SAMs with desirable properties. It is hypothesized that, within those SAMs, the functional groups interact laterally, taking the form of hydrogen bonding, dipole interaction, -stacking, or covalent attachment, which in turn enhance the mechanical integrity and stability.

Generally, there are two approaches to obtain these more complex structures on surfaces. One is to synthesize target precursors with functional group(s) and then assemble them onto surfaces by a one-step method,^14-18 but there are difficulties in purification during the synthesis of more complex molecules, especially alkylsilanes derivatives. In this case, stepwise formation of the films with desired structures based on surface chemical reactions is preferred.^12,19 Compared to the one-step method, the stepwise formation strategy is more suitable for formation of SAMs with more complex structures.

Several reports^14-16 have demonstrated that incorporation of amides into hydrocarbon backbones of precursors could improve the stability of the SAMs. The likely reason was that the amide underlayers were capable of being cross-linked by hydrogen bonding. In our previous reports^12,18 we built amide-containing SAMs on silicon surfaces and found very excellent wear-resistant properties of the films. In these works, one amide group was introduced, so only one layer of interlinked hydrogen bond was formed. Hutchison et al.^15-17,20 designed a series of alkylthiols containing one or more amide groups per precursor and assembled them on gold surfaces. In these monolayers, two or three amides contained in chains formed three-dimensional networks of hydrogen bonds, which greatly improved the stability of SAMs by cooperative interactions.²⁰ The enhanced stability clearly demonstrates the potential use of internally cross-linked SAMs in a wide range surface modification applications. However, these stability-enhanced structures are mostly realized in alkylthiol SAMs on gold. However, few alkylsilane derivatives, which are more suitable as lubrication layers in MEMS, have been prepared and studied; moreover, a tribological evaluation is seldom done.

In our previous study,¹² we self-assembled 3-aminopropyltriethoxylsiliane (APS) onto silicon and then modified the amino-terminated with stearic acid (STA) to form an amide-containing film. The amide-containing film has a less densely packed structure with a water contact angle of only 98. Thus, it is not effective enough for its anti-stiction use in MEMS. In the present study, we choose N-[3-(trimethoxylsilyl)propyl]ethylenediamine (DA) and stearoyl chloride (STC) instead of APS and STA. The reasons are follows: first, DA with a longer chain length than APS tends to form a more densely packed and well-ordered film; second, STC is more reactive than STA for the surface coupling reaction with the NH₂ group; last, the two -NH- groups contained in the DA chain can provide opportunities to form two-layer networks of hydrogen bonds, which will greatly enhance the stability of the SAMs by cooperative interactions.²⁰ The ideal structure of the amide-containing stratified film is shown in Figure 1.

Figure 1 Schematic view of the amide-containing stratified film on silicon.

Experimental Section

Materials. N-[3-(Trimethoxylsilyl)propyl]ethylenediamine (DA) and 3-aminopropyl-triethoxylsilane (APTES) were purchased from Aldrich. Stearic chloride was purchased from Fluka. Acetone, toluene, and triethylamine were all analytical reagent. All reagents were used as received. The P-type polished single-crystal Si(111) wafer used as the substrate was obtained from GRINM Semiconductor Materials Co., Beijing, China. Ultrapure water (>18 M) was used in this work.

Preparation of Films on Silicon Wafer. Silicon wafers were cleaned and hydroxylated in Piranha solution (mixture of 7:3 (v/v) 98% H₂SO₄ and 30% H₂O₂) at 90 C for 30 min. After that they were rinsed copiously by deionized water.

A 20 L amount of DA was dissolved in 20 mL of a mixture of acetone and water. The ratio of acetone and water was 10:1 (by volume). Cleaned wafers were immersed in fresh DA solutions. After a certain period of time, the wafers were removed from the solutions and sonicated for 5 min in acetone and cyclohexane.

After being ultrasonically cleaned in cyclohexane, the above-treated wafers were immersed in a 10 mM cyclohexane solution of lauroyl chloride with 5 L of triethylamine for 24 h, then ultrasonically cleaned in toluene and acetone, in succession, and rinsed by deionized water and blown dry with nitrogen.

Besides, APTES film was also prepared according to the procedures mentioned in our previous report¹² and then modified by STC. The film is coded as APSC18 film.

Contact Angle Measurement. A CA-A type contact angle meter (Kyowa Scientific Co. Ltd.) was used to measure the static water contact angle of the films. At least three points were measured for each specimen, and the measurement error was ±1. For each drop on the surfaces, both sides were measured to get the averages.

Film Thickness Measurement. The ellipsometric thickness measurements were performed on a L116-E ellipsometer (Gaertner, USA), which was equipped with a He-Ne laser (632.8 nm) set at an incident angle of 50. A real refractive index of 1.46 was set for the silica layer and 1.45 for organic layers. The data were collected from 10 different positions for each specimen to get the averages.

Attenuated Total Reflectance Fourier Transform Infrared (ATR-FTIR) Spectra. ATR-FTIR spectra were recorded on a Bruker IFS 66 V/S Fourier transformation infrared spectrometer. The spectra of the films were obtained using a Harrick Scientific horizontal reflection Ge-attenuated total reflection accessory (GATR, 65 incidence angle). The samples were placed in contact with the flat surface of a semispherical Ge crystal, which serves as the optical element. The spectra were collected for 32 scans with a resolution of 4 cm^-1. The background was collected using the accessory with no sample placed on it. In order to eliminate the effect of H₂O and CO₂, the pressure in the sample chamber and optical chamber was kept below 6.0 × 10^-4 MPa.

Atomic Force Microscopy. A Nanoscope IIIa Multimode atomic force microscope (AFM, Digital Instruments) was employed to observe the film morphology using tapping mode. The relationship between normal load and friction force was measured on a SPA 300HV scanning probe microscope (SII NanoTechnology Inc., Japan). Normal loads were applied on by varying force references in the units of nanoNewtons (nN). Friction forces were obtained from friction loops at five separate points on each surface with a scan velocity of 5 m/s. The output voltages were directly used as frictional forces. The triangle-shaped silicon nitridecantilevers with an announced spring constant of 2N/m were used. No attempt was made to calibrate the torsional force constant. For all measurements, the same cantilever was used. Experiments were carried out under ambient conditions of 20 C and 30%-40% relative humidity.

Macrotribology. Macrotribological properties were studied on a UMT-2MT tribometer (CETR, USA) using a ball-on-plate mode. The upper ball counterpart was fixed, while the lower sample plate was adhered on the flat base which reciprocated at a distance of 0.7 cm. The balls used here were commercially available ruby balls (Al₂O₃, 4 mm). Loads of 10, 30, and 50 g (corresponding to 0.1, 0.3, and 0.5 N, respectively) were applied. The friction coefficient-time plots were recorded automatically, and at least three repeated measurements were performed.

Results and Discussion

Formation of Amide-Containing Stratified SAMs via a Two-Step Method. The amide-containing stratified films were formed on silicon surfaces by a two-step method. DA SAMs with NH₂ groups exposed outside served as adhesive layers. STC molecules were chemically absorbed on NH₂ groups of DA to form a hydrophobic overlayer. For convenience, the film derived from DA with STC is coded as DAC18 film. 3-Aminopropyltriethoxylsilane (APTES) or 3-aminopropyltrimethoxylsilane (APTMS) is frequently used to fabricate amino-terminated surfaces for various applications. While these APTES or APTMS films are usually disordered and multilayered. With one more -CH₂CH₂NH- units, DA tends to form more densely packed and more ordered SAMs. Additionally, hydrogen bonding between adjacent molecules may occur in the buried -NH- layer(s), which will enhance the rigidities and stabilities of the aminosilane SAMs.

Formation of DA SAMs was monitored by water contact angle measurement and ellipsometric thickness measurement during the preparation process. We assumed that one CH₂ unit (considering one NH or NH₂ equal to one CH₂) is 0.14 nm in length in the all-trans configuration chains,²² so a complete DA SAM has a theoretical thickness of about 1.00 nm. Every 10 or 20 min two specimens of each kind were picked out of the solutions, cleaned, and blown dry and then used for the thickness measurements. The measured thickness increased slowly with time; we immediately released all of them out of the solutions when the measured thickness reached the corresponding theoretical value. If the reaction time was prolonged, the thickness would be more than the theoretical values, indicating a multilayer was formed. Because the formation process was so sensitive to solvent, concentration, temperature, and water content, the time to form a completed monolayer slightly varied every time. The thickness of the SiO₂ layer on silicon wafer is about 1.7 nm after the cleaning process; the thickness of the DA film is 0.98 nm, which is close to the theoretical value b (Figure 2). Water contact angles changed little with the increase of thickness. At the early stage the contact angles already reached around 40. At the end, the contact angle remained at 39 on average, as shown in Figure 2. For amino-terminated SAMs, a wide range of contact angles have been reported²³ and high contact angles indicate disordered structures for the exposure of alkyl chains.²⁴ We obtained a contact angle of 44 on APTES SAMs surface.¹² Thus, DA formed more densely packed SAMs than APTES with many more polar NH₂ groups exposed outside.

Figure 2 Water contact angles and ellipsometric thicknesses of the various films.

In the second step, STC was used to modify the amine surfaces. Figure 2 shows the sharp increase in film thickness and water contact angles after modification, causing the water contact angle to reach 109 and film thickness to increase about 2.4 nm. The theoretical thickness of a fully covered STC layer is about 2.6 nm; a complete SAMs with an 18-carbon chain has a water contact angle more than 110.^22,25 Therefore, nearly full coverage was obtained on the amino surface, which is a great improvement over APTES in our previous study.¹² The density of self-assembled molecules on top of the aminosilanized layer was determined by the surface density of the amine group. Thus, DA formed a dense film with larger -NH₂ density on the surface. With the same amino tail, longer chain aminosilanes can form a more dense film. APTES, with a shorter chain, self-assembles onto the silicon surface randomly to form a film with fewer amine groups on top. Smirnov et al.²⁶ also found a higher adsorption ability of DA than APS for coumarin molecules in their recent study.

The FTIR spectra of the films on silicon wafers are shown in Figure 3. For DA film (Figure 3a), only the asymmetric and symmetric methylene vibrations are observed at 2928 and 2855 cm^-1, respectively. Upon addition of STC (Figure 3b), the vibrational signatures of methylene shifted to 2923 and 2855 cm^-1, respectively. The intensity of both increases obviously, indicating the increase of crystallinity of the methylene chains. Even then, methylene chains are not so closely packed.²⁷ The key bands at 1639 and 1542 cm^-1 can be assigned to amide I (predominantly C=O stretching) and amide II (involving torsional motions of both N-H and C-N) due to formation of a -NHC=O group, which is in good agreement with the literature.^10,14,28 Hydrogen bonding may cause a decrease in electron density of the C=O bond and increasing restriction in N-H bending. Whitesides et al. claimed the existence of hydrogen bonding in the SAMs based on FTIR results that the amide I band is at 1650 cm^-1 and amide II at 1550 cm^-1.¹⁴ Thus, the existence of hydrogen bonding can be indicated by the red shift of the amide I band and blue shift of the amide II band. Therefore, it is reasonable to conclude that hydrogen bonding exists in our DAC18 film.

Figure 3 Ge attenuated total reflective FTIR spectra of DA film (a) and DAC18 film (b) on silicon.

In order to obtain information on the surface characteristics, such as the uniformity, roughness, grain distributions, and defect formation, we observed the surface morphology of the prepared films using AFM.²⁹ The morphologies of DA and DAC18 films are shown in Figure 4. It can be seen that both surfaces are characterized by regular grains distributed on the surface. Nevertheless, the surfaces of the films are still rather smooth on the micrometer scale: DA film has a microroughness of root-mean-square (rms) about 0.15 ± 0.01 and 0.20 ± 0.02 nm for DAC18 film over a scanning range of 500 nm × 500 nm. The slight increase in microroughness after addition of STC may be caused by incomplete coverage of STC on DA surface. This is consistent with contact angle and FTIR measurements.

Figure 4 AFM morphologies of DA film (a) and DAC18 film (b) over a scan area of 500 nm × 500 nm. The Z range data scale is 2 nm.

MicroTribological Study. Friction force measurements were performed in air using an FFM. Here, the friction force is given in voltage. The voltage signal should be proportional to the real friction force.²⁷ Therefore, the friction forces on various film surfaces can be compared with one another when the same AFM tip is used. Figure 5a shows linear relationships between the lateral deflection and load for the different surfaces. The linearity of the friction load relationship suggests that generalized Amonton's Law can be applied in which the lateral force (F_L) is given by

	Figure 5 (a) Friction versus load curve for the surfaces of bare silicon, film DA, and film DAC18. (b) Relative friction coefficients directly derived from Figure 5a.
	Figure 6 Variation in friction coefficient with time for DA and DAC18 film at different applied loads and a sliding frequency of 1 Hz: (a) film DA at 0.1 N; (b) film DAC18 at 0.1 N; (c) film DAC18 at 0.3 N; (d) film DAC18 at 0.5 N.

Macrotribological Study. The wear-resistance ability of SAMs is very important for their potential use as a lubricant layer. The wear-resistance property of the films mentioned above was tested on a ball-on-plate macrotribometer. Poor wear resistance has been found for DA film (Figure 6a). It was worn out as soon as the counterpart ball began to slide on it. The antiwear ability of DA film was enhanced greatly after being modified by STC (Figure 6b,c). Film DAC18 can remain as an effective lubricant layer for more than 18 000 s at a load of 0.1 N and a sliding rate of 60 rev/min. The average friction coefficient is about 0.08 and remains below 0.1 over a period of 5 h as the load increased to 0.3 N (Figure 6c). However, when a load of 0.5 N was applied, the film was worn out after several seconds, as witnessed by the sudden increase of the friction coefficients (Figure 6d). In our previous study,¹² we prepared films using APTES and STA and found very excellent anti-wear-resistant properties on them. However, these were tested on a different tribometer and at a lower sliding velocity (90 mm/min). Using similar procedures, we prepared films with the same structure, e.g., APSC18 film. Then the same testing conditions and counterpart ball for DAC18 were used to test APSC18 films. We found that APSC18 film was stable under a load of 0.1 N at a sliding frequency of 1 Hz (equal to a sliding velocity of 840 mm/min) but was easily destroyed under a load of 0.2 N (Figure 7). Besides, the friction coefficient of APSC18 film is higher and grows more rapidly. The enhanced friction reduction and load affording ability of DAC18 film should rely on its intrinsic structure. The densely packed and rigid underlayer of DA enhanced the stability and load carrying capacity of the entire film. STC with a long and flexible chain forms an ordered and a hydrophobic overlayer from which low friction and long life were obtained. A network of lateral cross-links within the film was formed due to two layers of internal hydrogen bonds, which largely improve the stability of the DAC18 film. From the above we can find ways to design films with good frictional performances. Both a rigid part to withstand load and a hydrophobic and flexible part to reduce surface adhesion and friction force should be contained in the structure.

Figure 7 Variation in friction coefficient with time for APSC18 film at different applied loads and sliding frequency of 1 Hz: (a) 0.1 N and (b) 0.2 N.

Additionally, we found that frictional coefficient could be affected by sliding speed. The average coefficients for the DAC18 film during the first 5 min for each sliding frequency were collected. As shown in Figure 8, the friction coefficients increase with the increase of sliding velocity. We contribute this phenomenon to the higher degree of oscillation and distortion of the molecules at higher shear velocity, which accelerates dissipation of the accumulated energy.³¹

Figure 8 Variation in friction coefficient with sliding velocity for DAC18 film at a normal load of 0.1 N.

Conclusions

An amide-containing stratified thin film was formed on a silicon surface by a simple two-step method. DA was first assembled on the substrates to get a rigid and dense underlayer, and STC was grafted onto it to achieve a hydrophobic surface with a water contact angle of nearly 110. We confirmed the structure by GATR FTIR spectra and concluded the existence of hydrogen bonding in the film. A tribological study performed on AFM showed that the friction force was reduced by formation of DAC18 film. Better friction reducing and anti-wear performances of the DAC18 film than APSC18 film were also confirmed from the results of the macrotribology test. We contribute these improvements to the two layers of internal hydrogen bonds within the film. The resulting network of lateral cross-links by internal hydrogen bonds can enhance both the load affording ability and the stability of the film. Thus, it is an efficient way to design a reliable lubricating layer applied in microelectromechanical systems on the molecular level.

Acknowledgment

The authors thank the Natural Science Foundation of China (grant nos. 50572107 and 20673131) and "Hundreds Talent Program" of the Chinese Academy of Sciences for financial support. We also thank Professor Xin Shao (College of Material Science and Engineering of Liaocheng University) for assistance in the use of the AFM facility.