Chemical Effects on the Adhesion and Friction between Alkanethiol

Alkanethiol SAMs (S(CH2)nX) are investigated as a function of the terminal group .... J. Christopher Love, Lara A. Estroff, Jennah K. Kriebel, Ralph G...
0 downloads 0 Views 470KB Size
Download PDF
Langmuir 2003, 19, 9239-9245

9239

Chemical Effects on the Adhesion and Friction between Alkanethiol Monolayers: Molecular Dynamics Simulations Byeongwon Park,† Michael Chandross, Mark J. Stevens,* and Gary S. Grest Sandia National Laboratories, Albuquerque, New Mexico 87185 Received January 22, 2003. In Final Form: August 19, 2003 The chemical effects on the adhesion and friction between alkanethiol self-assembled monolayers (SAMs) on gold are studied using molecular dynamics simulations. Alkanethiol SAMs (S(CH2)nX) are investigated as a function of the terminal group (X ) CH3, OH, COOH) and chain length n. The adhesive force is calculated as a function of the separation between two SAMs with the same terminal group. The maximum attraction for methyl-terminated SAMs is much weaker than for the OH- and COOH-terminated SAMs which form interlayer hydrogen bonds. The COOH-terminated SAMs have the strongest attraction. For the COOH- and OH-terminated SAMs, the adhesion-separation curves depend strongly on n. In both cases the change in the position of the attractive minimum is large as a function of n. The magnitude of the attractive minimum varies little for OH-terminated SAMs, and the variation is also small for COOHterminated SAMs once the odd:even effect is considered. The source of the n dependence is structural changes within the SAMs. For separations within the attractive region of the adhesion curve, the interlayer hydrogen bond interactions pull the chains to a more upright tilt angle than the equilibrium, singlemonolayer tilt angle. Longer chains have more upright tilt angles for separations in the attractive region. For COOH SAMs with n e 13 the maximum adhesion is significantly larger for odd n in comparison to even n - 1. The difference between even and odd cases decreases with n, and for n > 13 the difference is smaller than our uncertainty. Friction simulations between two SAMs with the same n show that the shear stress is significantly larger for X ) COOH than for CH3.

I. Introduction Self-assembled monolayers (SAMs) of organic thin films on metals/metal oxides have been widely investigated recently,1-3 due to the ease of assembly, high packing density, and robust stability. Using SAMs the surface structure can be controlled at the atomic scale by modifying the terminal group of the SAM molecule. This ability to easily modify the surface structure and, consequently, the surface interactions has led to applications in molecular electronics, nanotribology, and nanooptics. SAMs also offer the opportunity to study fundamental interactions on the atomic scale.4,5 Much of this work has used developments in force microscopy to probe surfaces on the nanometer scale.6 An example application is chemical force microscopy in which surface functional groups are used to discriminate chemical inhomogeneities of the surface.7-10 Most of the research on SAMs has focused on alkanethiols on gold. Atomically flat gold substrates are * Corresponding author. † Present address: Korea Institute of S&T Evaluation and Planning (KISTEP), Dongwon Industry Building, 275, YangjaeDong, Seocho-Gu, Seoul 137-130, South Korea. (1) Ulman, A. An Introduction to Ultrathin Organic Films: From Langmuir-Blodgett to Self-Assembly; Academic Press: San Diego, 1991. (2) Ulman, A. Chem. Rev. 1996, 96, 1533. (3) Ulman, A. Thin Films: Self-assembled Monolayers of Thiols; Academic Press: San Diego, 1998. (4) Thomas, R. C.; Tangyunyong, P.; Houston, J. E.; Michakske, T. A.; Crook, P. M. J. Am. Chem. Soc. 1995, 98, 4493. (5) Houston, J. E.; Kim, H. I. Acc. Chem. Res. 2002, 35, 547. (6) Joyce, S. A.; Houston, J. E. Rev. Sci. Instrum. 1991, 62, 710. (7) Frisbie, C. D.; Rozsnyai, L. F.; Noy, A.; Wrighton, M. S.; Lieber, C. M. Science 1994, 265, 2071. (8) Brewer, N. J.; Beake, B. D.; Leggett, G. J. Langmuir 2001, 17, 1970. (9) Noy, A.; Vezenov, D. V.; Lieber, C. M. Annu. Rev. Mater. Sci. 1997, 27, 381. (10) Noy, A.; Frisbie, C. D.; Rozsnyyai, L. F.; Wrighton, M. S.; Lieber, C. M. J. Am. Chem. Soc. 1994, 116, 2071.

easily constructed, removing issues and difficulties of rough surfaces. Recent research has studied the modification of surface properties and functionality through changes in the terminal group.4,5,9 The terminal groups can be categorized by their surface energy. Polar terminal groups, OH and COOH, have a high surface energy and are hydrophilic, providing sites for hydrogen bonding. In contrast, groups such as CH3 or CF3 produce low-energy, hydrophobic surfaces. Different fundamental interactions (van der Waals, hydrogen bonding) have been investigated with force microscopes using tips coated with SAMs.4,5,9,11 By measuring the forces between two SAM-coated surfaces with various combinations of terminal groups, the fundamental, atomic interactions are measured. Recent measurements using the interfacial force microscope on alkanethiol SAMs (S(CH2)nCOOH) measured a difference between odd (n ) 15) and even (n ) 10) chain lengths.5,11 These results were explained in terms of the different orientations of the terminal groups depending on whether n is even or odd. The SAM with even n was measured to have a much larger attraction than the SAM with odd n. Differences in the end group geometries were claimed to affect the ability to form hydrogen bonds between the two monolayers. Interlayer hydrogen bonds were thought to form relatively easily for even n, while for odd n intralayer hydrogen bonds formed more easily. Here, we use molecular dynamics (MD) simulations to investigate the chemical effects on the adhesion between two alkanethiol SAMs. There have been many simulations of alkanethiol SAMs.12-19 Most of the work to date has (11) Kim, H. I.; Houston, J. E. J. Am. Chem. Soc. 2000, 122, 12045. (12) Landman, U.; Luedtke, W. D.; Nitzan, A. Surf. Sci. 1989, 210, L177. (13) Glosli, J. N.; McClelland, G. M. Phys. Rev. Lett. 1993, 70, 1960. (14) Siepmann, J. I.; McDonald, I. R. Phys. Rev. Lett. 1993, 70, 453. (15) Tupper, K. J.; Brenner, D. W. Langmuir 1994, 10, 2335. (16) Bonner, T.; Baratoff, A. Surf. Sci. 1997, 377, 1082. (17) Tutein, A. B.; Stuart, S. J.; Harrison, J. A. Langmuir 2000, 16, 291.

10.1021/la0341106 CCC: $25.00 © 2003 American Chemical Society Published on Web 10/01/2003

9240

Langmuir, Vol. 19, No. 22, 2003

Park et al.

treated only the case of methyl-terminated molecules, although one work has studied OH-terminated SAMs.19 We also consider the sliding friction between two SAMs. Using MD, we can simultaneously examine the interactions between the two surfaces as well as the structure of the SAM as a function of separation. Three different terminal groups (CH3, COOH, and OH) are studied for three pairs of even:odd chain lengths (n ) 8:9, 12:13, and 16:17), although for COOH SAMs adhesion curves were calculated for additional values of n (7, 10, and 11). In section II, details of computational procedures are discussed. The results for the adhesion simulations for the three different end groups are presented in section III. In section IV we present the frictional dynamics for CH3- and COOH-terminated SAMs. In section V we present our conclusions. II. Computational Procedure The initial monolayer configuration is composed of 100 alltrans alkanethiol molecules (S(CH2)nX) on a rectangular grid (43 Å × 50 Å) at a tilt angle of 30° with respect to the monolayer normal (z-axis). Recently, our group reported that larger system sizes have little effect on adhesion and friction simulations of alkylsiloxane SAMs on SiO2 beyond a reduction in noise,20 indicating that this system size is sufficient for quantitative comparisons. Three terminal groups (X ) CH3, OH, COOH) with different numbers of CH2 units in the carbon backbone (n ) 8, 9, 12, 13, 16, 17) are treated. For COOH we also treated n ) 7, 10, and 11. For adhesion and friction studies, a second layer mirroring the first was added with a 10 Å initial vacuum gap between the outermost hydrogens in the terminal group. We define D as the relative displacement of the two monolayers such that at this starting state D ) 10 Å. Figure 1a shows the system at this initial separation (after a compression-decompression cycle). Alternatively, we could define D as the separation between the gold surfaces, but this choice makes D dependent on the chain length. Our simulations were performed using the COMPASS force field.21 This is an all-atom force field parametrized against a wide range of experimental observables and quantum mechanical calculations. Partial charges for the OH (includes terminal C), and COOH terminal groups are given in Table 1. Sulfur atoms are kept frozen during the simulation. We expect that the gold surface and thiol groups have a negligible effect on adhesion and friction, with the majority of the inter-monolayer interaction coming from the interaction between the terminal groups. The thiol on the gold structure should not change under the conditions we study, because the bonding of the thiol group to gold has a strong bond strength of 40 kcal/mol.22 This justifies fixing the position of the sulfur atoms. There are at least three different SAM structures that have been reported as representing the ground state: (x3 × x3) R30°, herringbone, and c(4 × 2).23 Early experimental data showed alkanethiols formed a simple (x3 × x3)R30° ground state which is commensurate with the underlying gold layer.24-26 Recent experiments found that densely packed SAMs have a c(4 × 2) superlattice of the (x3 × x3)R30° lattice27-29 or a mixture (18) Jiang, S. Mol. Phys. 2002, 100, 2261. (19) Leng, Y.; Jiang, S. J. Am. Chem. Soc. 2002, 124, 11764. (20) Chandross, M.; Stevens, M. J.; Grest, G. S. Langmuir 2002, 18, 8392. (21) Sun, H. J. Phys. Chem. B 1998, 102, 7338. (22) Dubois, L.; Nuzzo, R. G. Ann. Phys. Chem. 1992, 43, 437. (23) Schreiber, F. Prog. Surf. Sci. 2000, 64, 151. (24) Porter, M. D.; Bright, T. B.; Allara, D. L.; Chidsey, C. E. D. J. Am. Chem. Soc. 1987, 109, 3559. (25) Chidsey, C. E. D.; Liu, G. Y.; Rowntree, Y. P.; Scoles, G. J. J. Chem. Phys. 1989, 91, 4421. (26) Alives, C. A.; Smith, E. L.; Porter, M. D. J. Am. Chem. Soc. 1992, 114, 1222. (27) Camillone, N.; Chidsey, C. E. D.; Liu, G.-Y.; Scoles, G. J. Chem. Phys. 1993, 98, 3053. (28) Fenter, P.; Eberhardt, A.; Eisenberger, P. Science 1994, 266, 1216. (29) Poirier, G. E.; Tarlov, M. J. Langmuir 1994, 10, 2853.

Figure 1. Structure of alkanethiol SAMs ((CH2)8COOH) at various relative displacements: (a) at the separation D ) 10 Å, (b) at the maximum attraction, D ) 2.9 Å, and (c) at a highly compressed position D ) -4 Å. The H atoms are gray, the C atoms are cyan, the O atoms are red, and the S atoms are yellow. The green lines at the top and bottom denote the ghost atoms (see text). The distance s is the average separation between the terminal C on each SAM. While s and D are almost equal in a, they are different in b, s ) 3 Å, and in c, s = 2 Å. Table 1. Partial Charges Used in Force Field -OH

-COOH

atom

charge

atom

charge

C O H

-0.1060 -0.5371 +0.4241

C O (carbonyl) O H

+0.3994 -0.4271 -0.3964 +0.4241

of c(4 × 2) and (x3 × x3)R30° lattices.30 The minimum energy state calculated with the force field we used yields (x3 × x3)R30° as the ground state. We determined that the forceseparation curve does not depend on the choice of the starting SAM structure. Given the uncertainty in the experimental data, the fact that the (x3 × x3)R30° structure is the basis for the other structures as well as being the minimum energy state for our force field, we use it as our starting configuration. Experimentally alkanethiol chains tilt about 30° from the normal.23 To achieve this tilt angle, a harmonic angle force between the first carbon atom, the sulfur atom, and a ghost site directly below the sulfur is included to substitute for the neglected (30) Paum, J.; Bracco, G.; Schreiber, F., Jr.; Colorado, R.; Shmakova, O. E.; Lee, T. R.; Scoles, G.; Kahn, A. Surf. Sci. 2002, 498, 89.

Adhesion and Friction between Alkanethiol SAMs

Langmuir, Vol. 19, No. 22, 2003 9241

Figure 3. Tilt angle and interlayer separation s of (CH2)8X SAMs during decompression. The data are for a decompression rate of 8.3 m/s.

Figure 2. Compression/decompression data for alkanethiol SAMs of (CH2)8X, where X is (a) CH3, (b) COOH, and (c) OH. The compression started with the films having an initial vacuum gap of D ) 10 Å. The displacement velocities are 8.3 m/s (symbol) and 1.7 m/s (solid line). Squares (0) are for compression and open circles (O) for decompression. interactions with the gold substrate. An equilibrium angle of 119.65° with a force constant of 46.06 kcal/mol for the (x3 × x3)R30° structure was used. Without this term, the measured tilt angle was only 15°. For the other structures the values for the equilibrium angle are set to achieve the correct geometry. Temperature was constrained to 298 K using the Nose´-Hoover thermostat with a damping rate of 0.03 fs-1. Multitime step integration was performed using rRESPA31 with time steps of 0.30 fs for the bond forces, 0.6 fs for the other intramolecular forces, and 1.2 fs for van der Waals interactions. The cutoff value for Coulombic interactions and van der Waals interactions were set to 12 Å. The one to four van der Waals interactions are fully included. Adhesion simulations were carried out by compressing two SAMs at a constant velocity of 8.3 m/s, with additional simulations at 1.7 m/s for some systems to study the dependence of adhesion on the compression rate. We monitored the change in structure and energetics during the simulations as a function of separation. For the friction studies, two SAMs were sheared in opposite directions with a constant relative velocity of 2 m/s at a fixed separation corresponding to an applied pressure approximately 1 GPa. In the majority of our friction studies, the shear was in the tilt direction, but we have verified that shear in other directions eventually causes the chains to tilt along the shear direction.

III. Results A. Adhesion. In Figure 2 we show the normal pressure as a function of displacement during compression and decompression for (CH2)8X SAMs with X ) CH3, COOH, and OH at two compression rates: 8.3 and 1.7 m/s. The starting distance between the two monolayers is a D ) 10 Å gap between the outermost hydrogens in the terminal (31) Plimpton, S. J.; Pollock, R.; Stevens, M. J. Particle-Mesh Ewald and rRESPA for Parallel Molecular Dynamics Simulations. In Proceedings of the Eighth SIAM Conference on Parallel Processing for Scientific Computing; Heath, M., et al., Eds.; SIAM: Philadelphia, 1997.

group. The SAMs are compressed to a final relative displacement of 15 Å, corresponding to a 5 Å overlap between the two monolayers with respect to the starting structure or D ) -5 Å. In compression the two monolayers move from D ) 10 Å to D ) -5 Å, and in decompression it is the reverse. The agreement in the force-separation curves for the two rates, along with equilibration simulations at selected D, shows that we are in a regime that is independent of the compression rate. While we cannot perform simulations near the much lower experimental rates, our rate range is sufficient for molecular relaxations to occur (e.g., rearrangement of terminal groups). For our ideal systems this is all that is necessary. While we cannot treat the much slower dynamics due to large domains in systems with defects, such dynamics is not the source of the chemical interactions that are the focus of this work. The normal pressure is zero at separations greater than the cutoff distance for the nonbonded interactions (12 Å). As the monolayers are brought together, the interaction becomes attractive due to the Coulomb and van der Waals interactions between the two SAMs which are mainly due to interactions between terminal groups. An attractive minimum in the normal pressure-separation curves occurs at small separations in all cases. The magnitude of the attractive minimum is much larger for (CH2)8COOH, 1.3 GPa, and (CH2)8OH, 1.1 GPa, than for (CH2)8CH3, 0.28 GPa. The uncertainty for each of these is (0.06 GPa. Near the attractive minimum hysteresis occurs in the compression/decompression cycle, since at these displacement rates the films are unable to relax completely. As we show below the SAM tilt angle collectively changes with displacement. Such a collective motion relaxes on time scales orders of magnitude slower than these compression rates allow. Thus, the hysteresis is essentially unchanged for the range of displacement rates we could treat. To determine the equilibrium adhesion curve, several configurations at different separations from both the compression and decompression runs were equilibrated for 120 ps. Equilibrations of configurations from the decompression runs showed almost no relaxation of the normal pressure, while for equilibration of compression configurations the normal pressure relaxed to the decompression curve value. We therefore use states from the decompression runs as starting configurations for the shear studies. The changes in the tilt angle and the average terminal group separation s during decompression are shown in Figure 3. The interlayer separation, s, is defined to be the

9242

Langmuir, Vol. 19, No. 22, 2003

Park et al.

Figure 4. Number of H-bonds for the (CH2)8COOH system as a function of relative displacement, D. The total number of H-bonds is given by solid circles, the interlayer H-bonds by open circles, and the intralayer H-bonds by open triangles.

distance between the average postions of the terminal carbon atoms in each monolayer. Unlike D, the value of s depends on the SAM structure. In particular the tilt angle directly influences s as can be seen in Figure 3. Roughly speaking, s as a function of D is composed of two straight segments for all cases. Decreasing from D ) 10 Å where the monolayers are well-separated, s ) D plus a constant equal to the distance between terminal H and C atoms. In this range of D the tilt angle is the equilibrium, single-monolayer value of 30°. After the monolayers contact the tilt angle increases, since the dense monolayers cannot interpenetrate (see Figure 1c). Consequently, s decreases very slowly as the terminal groups are packing together under the increasing normal pressure. The minimum s at D ) -5 Å is in the range 1-2 Å and depends on the terminal group, since how well these groups pack determines how close the terminal C atoms can get. In the contact region, Figure 3 shows that the tilt angle increases linearly with decreasing D. A structural origin of the large difference in adhesion strength among the different terminal groups can be seen in the tilt angle in the range D ) 0-5 Å, where the attractive minimum occurs in the force-separation curves. In contrast to the methyl case where there is little variation, the COOH- and OH-terminated SAMs have a significant decrease in the tilt angle in this separation range. In other words, the chains become more upright, due to the strong H-bond interactions as we will show. The image (Figure 1b) shows the more upright structure that occurs at the attractive minimum for (CH2)8(COOH). The number of hydrogen bonds can be determined quantitatively for the OH and COOH terminal groups and is shown in Figure 4 for (CH2)8COOH. A hydrogen bond is defined to exist if the distance between an acceptor (oxygen in COOH or OH) atom and a donor (hydrogen in COOH or OH) is less than 2.6 Å and the angle between acceptor oxygen and the OH is greater than 90°.32 The distribution of H-bonds is qualitatively the same for COOH and OH SAMs. When the monolayers are in contact or close to contact, most hydrogen bonds are interlayer bonds, as shown in Figure 4. In the fully compressed state there (32) Lommerse, J. P. M.; Price, S. L.; Taylor, R. J. Comput. Chem. 1997, 18 (6 (Apr)), 757.

Figure 5. Chain length and terminal group dependence of normal pressure vs displacement for (CH2)nX, where X ) (a) CH3, (b) COOH, and (c) OH (n ) 8 (2), 9 (4), 12 (9), 13 (0) 16 (b), 17 (O)). The data are decompression curves at 8.3 m/s that have been smoothed by taking 10 point running averages.

are about 200 interlayer H-bonds which is one H-bond per molecule. The COOH group can form two H-bonds which when divided between the top and bottom layers comes out to one H-bond per molecule. The total number of H-bonds at the smallest values of D is greater than 200. This is an overcount of the number of H-bonds due to our criterion being distance based, and to the fact at these separations (and correspondingly large loads) the H and O atoms are close enough to have more than one candidate H-bond. Only one of the candidate H-bonds can actually occur. At D = 2.6 Å the number of interlayer H-bonds goes to zero as it must once the separation is greater than the maximum H-bond bond length. For larger D all hydrogen bonds are from intralayer interactions and their number remains constant. Clearly, the strong H-bond interaction is able to pull the chains upright and thus maintain adhesion between the monolayers beyond the separation at which two single monolayers at a 30° tilt could no longer possess interlayer H-bonds. In Figure 5 we show the chain length dependence of the adhesive interaction for (CH2)nX. The pressure curves for the methyl-terminated SAMs do not change much with n. One advantage of our definition of the relative displacement, D, is that the minimum in the pressure curves should occur at the same D for different n, if there is no structural change in the SAMs. This is what we find for the methyl-terminated SAMs. The repulsive part of the adhesion curve becomes weaker as n increases, similar to our previous results on alkylsiloxanes on silicon dioxide.20 However, the data for COOH- and OH-terminated SAMs shows significant n-dependence, particularly with respect to the position of the attractive minimum. The larger n have their minimum position at larger D. For the OH SAMs the depth of the attractive well does not vary much with n. The OH SAM adhesion curves for the 12:13 and 16:17 pairs overlap in the attractive region, while the 8:9

Adhesion and Friction between Alkanethiol SAMs

Langmuir, Vol. 19, No. 22, 2003 9243

Figure 6. Chain length dependence of maximum attractive normal pressure for (CH2)nX, X ) CH3 (b), OH (2), COOH (9). Pressures were determined from equilibrium simulations at the minimum position. Solid lines connect even:odd pairs for CH3 and OH. For COOH separate least-squares fit (dashed) lines to the even and odd n are given. Error bars are shown for the COOH; the CH3 and OH data have similar values.

pair is distinct. For the COOH SAMs there is significant variation among the different n, and there is a clear distinction within the even:odd pairs. For all the COOH pairs, the odd n has the lower minimum, and the value of D at the attractive minimum monotonically increases with n. For both the OH and COOH SAMs the variation in the position of the minimum is related to changes in the molecular tilt angle, as will be discussed below (see Figure 7). Configurations at the minimum position in the decompression adhesion curve were run for 120 ps to provide equilibrated values of the maximum attractive pressure Pmax. In addition a few simulations were run for nearby separations to check that separation corresponding to the maximum attraction had indeed been located. In Figure 6 Pmax is shown for three sets (8:9, 12:13, 16:17) of even: odd chain length pairs, as well as n ) 7, 10, and 11 for X ) COOH. The values of Pmax for SAMs with CH3 terminal groups are the smallest for all chain lengths and are approximately constant at 0.25 GPa, which is similar to the value for alkylsiloxane SAMs.20 For the OH (and COOH) terminal group, a large even:odd difference in Pmax occurs for the 8:9 pair. However, the data at 12:13 and 16:17 for the OH SAMs have only a small difference that is within our uncertainty. Ignoring the even:odd dependence and just focusing on the general n-dependence, we find for the OH SAMS Pmax is relatively constant about an average of 1.4 GPa. For the COOH SAMs there is a general trend of increasing Pmax with n. Thus, it appears that the COOH SAMs can adjust their structure to better optimize Pmax with increasing n. The most interesting results are for the COOH SAMs. For the COOH SAMs the even:odd difference is significant up to about n ) 13. In every case the odd n have larger Pmax than for the even n ( 1. However, for n g 13 Pmax is equal within the uncertainty of the data. The general trends are that the even:odd effect weakens with increasing n and that Pmax increases with increasing n. We now examine the structure of the SAMs as a function of n, paying particular attention to the even:odd effect for COOH systems. The tilt angle as a function of interlayer distance for the six chain lengths is shown in Figure 7. At values of D where the attractive minimum occurs in

Figure 7. Chain length dependence for COOH SAMs of the terminal group separation (s), tilt angle, and number of interlayer H-bonds vs displacement for n ) 8 (2), 9 (4), 12 (9), 13 (0), 16 (b), and 17 (O) with lines connecting the data. The vertical lines in the bottom panel denote the positions of the attractive minima in Figure 5 for n corresponding to the symbols at the end points.

the normal pressure curves, the tilt angle for COOHterminated chains shows significant n-dependence. The tilt angle decreases to 13-20° compared to the singlemonolayer equilibrium tilt angle (30°). The lowest minimum angle is for n ) 17, while the highest is for n ) 8. With the exception of n ) 13 the minimum angle increases monotonically with n. This variation in tilt angle is the main source of the chain length dependence of the position of the attractive minima. The relative displacement where the minimum tilt angle occurs is close to the position of the corresponding attractive minimum at each n. The smaller tilt angle allows the longer chains to maintain a smaller interlayer separation (s) and more H-bonds at the larger D. Figure 7 shows that the change in tilt angle is reflected in the interlayer separation between two SAMs. All the curves for s have the same shape for different n. The primary difference is in the position of the jump which increases monotonically with n. The sharp increase in s is correlated with the sharp increase in the tilt angle, both occurring at larger D than the position of the attractive minimum (see also Figure 5). The small values of s at larger D for larger n are a consequence of the smaller tilt angles for the longer chains. The more upright chains can maintain a separation such that H-bonds can occur at the larger values of D, where the steric repulsion is smaller. Thus, the attractive minimum is stronger with increasing n, because the counterbalancing steric repulsion is smaller. The value of s at the positions of the minima is about 4 Å in all cases. At this separation, the terminal C atoms on different monolayers do not repel each other. Evidently, this separation distance maximizes the attraction due to the H-bonds and minimizes the van der Waals repulsion. Figure 7 shows that the number of interlayer H-bonds is maximum where the interactions are most repulsive

9244

Langmuir, Vol. 19, No. 22, 2003

Park et al.

Table 2. H-O Distances for Figure 8 distance (Å) label

even

odd

1 2 3 4 5

2.1 2.8 3.5 5.4 2.1

1.9 2.3 3.1 1.9 2.3

and decreases rapidly in the vicinity of where the interactions are attractive. This would appear to counter the idea that the attractive interaction is due to the interlayer H-bonds. Even though the H-bond interaction is large and attractive for small D, the repulsive van der Waals interaction, which rises sharply upon contact, dominates the total interaction in this region. While the number of H-bonds decreases for displacements within the attractive minimum, so also does the amount of repulsive van der Waals interactions. Figure 7 shows the drop in the number of H-bonds occurs at larger D for larger n just as expected given the tilt angle and separation distance data. Furthermore, the number of H-bonds is always larger at the position of the minimum for the odd case. The expectation of different behavior for odd:even chain lengths comes from the difference in the orientation of the terminal group and its effect on the ability to form hydrogen bonds. Since we do not find a significant odd: even effect for large n and since Pmax is larger for n ) 9 than for n ) 8, the terminal group orientation is worth further examination. We note that the experiment5,11 studied the odd:even pair of n ) 10 and n ) 15 which is not a n:n + 1 pair. This raises the possibility that part of the measured result is due to a chain length dependence. However, interpolating from Figure 6, we do not expect a large difference for Pmax between n ) 10 and 15 as was observed in the experiment.11 The results above show that the tilt angle is an important factor and the most relevant tilt angle is less than the single-monolayer equilibrium value. Another factor that has not been considered is the threedimensional geometry. In particular we are interested in whether the geometry is different for even or odd n and whether this significantly affects H-bond formation between monolayers. We examined the orientation of the terminal COOH group for chains in the all-trans configuration with different tilt angles. The H-O separation distances were determined for possible H-bonds (Table 2). Figure 8 shows images for n ) 8 and 9 at a tilt angle of 15° for a four-chain cell. The separation distance between the terminal C atoms is s ) 4 Å, which approximately is the separation at Pmax. Another factor influencing the H-bond geometry is the lateral translation of one monolayer over the other. For a chain on the top monolayer, the closest H-bond partners on the bottom monolayer are often with chains to the left or right. A top layer that is the mirror image of the bottom will have its terminal group directly above the corresponding bottom terminal group, preventing H-bonds. Only a small lateral displacement is necessary to make H-bonds possible. The number of H-bonds can be optimized by adjusting the lateral displacement. For n ) 9 almost all chains can form H-bonds with both oxygens in COOH. The O with label 3 in Figure 8 is the one exception. In contrast only 2 out of the 5 bonds in Figure 8 can make H-bonds for n ) 8. This difference is not quite as strong as it appears. Only a small bend is needed to make the H-O pair labeled 2 close enough to form a H-bond. The H in 3, 4, and 5 is the same and can only form one H-bond (5). The carbonyl oxygen does not

Figure 8. Four chain cells showing oxygen-hydrogen separation distances for (a) even (n ) 8) and (b) odd (n ) 9) systems. The distances labeled 1-5 are given in Table 2. Coloring is the same as in Figure 1 except ghost atoms are blue.

have a formable interlayer H-bond and instead forms an intralayer H-bond (not labeled). This implies that in the all-trans configuration Pmax for n ) 9 should be larger (i.e., more attractive) than for n ) 8, as seen in Figure 8. This is consistent with the result shown in Figure 6 that has Pmax larger for odd n in general and specifically for the 8:9 pair. Allowing deformation of the chain, i.e., bending, of only tenths of an angstrom can significantly increase the number of H-bonds for the even case. For larger n the cost of such bending is lowered, explaining the small difference in Pmax for even:odd pairs at large n seen in Figure 6. In summary, from the molecular geometry we expect odd n to have a larger attractive interaction and that the difference between even and odd should decrease with increasing n. IV. Frictional Dynamics Friction simulations were performed at constant separation such that the average normal pressure P⊥ = 1 GPa. Results of friction simulations are shown in Figure 9. Only part of the shear dynamics is shown in the figure; the total displacement was 90 Å for (CH2)8COOH and about twice that for (CH2)8CH3. There is an initial transient period not shown in the figure of approximately 1.5 ns or 30 Å in which the system adjusts to being sheared. Thereafter, the dynamics remains the same as pictured in Figure 9. The fluctuations in both the normal and shear stresses show a strong dependence on the chemical nature of the terminal group. The average shear stress in the COOH SAM is about 5 times larger than for the CH3 SAM. The difference occurs for the same reason that the adhesion is much larger for the COOH SAMs. Hydrogen bond interactions dominate in the COOH case, while weak van der Waals interactions occur in the CH3 case. The stronger H-bond interactions in the COOH SAMs yield larger peak-peak shear stresses, since as an individual chain slides from one site to another on the opposite SAM, it breaks and re-forms H-bonds. The magnitudes in the shear

Adhesion and Friction between Alkanethiol SAMs

Langmuir, Vol. 19, No. 22, 2003 9245

V. Conclusion

Figure 9. Friction between alkanethiol films: (a) normal stress P⊥ and (b) shear stress P| for (CH2)8CH3; (c) P⊥ and (d) P| for (CH2)8COOH. The shear rate is 2 m/s.

stress between two successive terminations are consistent with the adhesion curves. The shear stress exhibits stick-slip motion as expected for sliding betwen well-ordered surfaces.33 The stick-slip motion will have a periodicity related to the 5 Å periodicity in the SAM lattice. The shear stress for the COOH system in Figure 9 exhibits such a periodicity. In this case, sticking is due to hydrogen bonds and the slip event is the transfer of the hydrogen bond from one chain to the next. Therefore, the periodicity matches the chain lattice. The peak shear stresses are large for the COOH system due to the strong hydrogen bond interactions. The CH3 system has a smaller periodicity in P| of about 3 Å, which is what was found for alkylsiloxane SAMs.20 This smaller periodicity occurs because, for n ) 8, the final C-C bond of the methylene C to the terminal methyl C is almost parallel to the surface. Consequently, the H atoms in this methylene group can be in the top layer of atoms, and this alters the structure of interacting layers has a 3 Å periodicity which is in the shear stress. The geometry is different for n ) 9, putting the H atoms of the topmost methylene group well below the surface and outside the steric interaction range. Thus, for n ) 9 the shear stress should not exhibit some 3 Å periods, but it does. In fact, there is a double periodic behavior with consecutive periods of about 3 and 5 Å. In this case, the 3 Å period is not due to a structure with a 3 Å period, but due to a dynamical time that corresponds to a 3 Å distance. The trajectory of a methyl group of a chain in the top SAM sliding over the bottom SAM is a zigzag path. A straight trajectory would take the methyl group directly over a row of methyl groups on the bottom SAM. Images of the dynamics show that the methyl group (top SAM) slides around such methyl groups on the bottom SAM. Once past one of these methyl groups on the bottom SAM, the next two methyl groups are offset from the straight path and the top methyl group can slide between those two. These two events require different times. Sliding around the methyl groups in the direct path takes longer (and more stress) than sliding between the two offset methyls. These different times correspond to different shear distances. We expect that this behavior is shear velocity dependent. At the much slower shear velocities in AFM measurements, the resident time may be much slower than the passage time for either case, yielding only a single period in the shear stress. (33) Thompson, P. A.; Robbins, M. O. Science 1990, 250, 792.

In agreement with experiment,4 we find that the methylterminated SAMs have a small attractive adhesion in comparison to the COOH- and OH-terminated SAMs. Calculations have shown that the magnitude of the difference is consistent with experimentally measured values.34 In the methyl SAMs all SAM-SAM interactions are relatively weak van der Waals or Coulomb. For COOH and OH SAMs, interlayer H-bonds occur as explicit calculations show. H-bonding is directional and requires the end groups of the two SAMs to have the appropriate geometry. Many of the results in this paper are related to the H-bonding geometry and interaction strength, such as the significant chain length dependence in the COOH and OH data. In contrast, the weak and nondirectional van der Waals interactions result in no significant chain length dependence for the methyl-terminated SAMs. The molecular structure of COOH- and OH-terminated SAMs varies significantly as a function of both separation and chain length. This is because the hydrogen bond interaction is sufficiently strong to alter the tilt angle of the alkanethiol molecules. There is a significant ndependence in this change in tilt angle. The varying tilt angle changes the separation at which “contact” occurs, i.e., when the van der Waals repulsion begins to be appreciable. The tilt angle affects the geometry of hydrogen bonding and makes the difference in hydrogen bonds between even and odd chain lengths smaller. Moreover, the strength of the interaction between two monolayers is determined by a competition between H-bond attraction and van der Waals repulsion. The adjustment of the tilt angle allows optimization of the amount of hydrogen bonding and a stronger net attraction between the monolayers. In contrast to recent experimental measurements for X ) COOH,5,11 we only find an even:odd dependence for our shortest pair and odd chain length has the larger attraction. We find that the odd chain length has the larger attraction, while the measurement has the even chain length with the larger attraction. The experiment treated the even:odd pair 15:10 which is not successive and allows the possibility of significant chain length effects.35 In addition, the n ) 10 SAM will be less ordered than the n ) 15 SAM.35,36 If this results in a less tightly packed SAM, then the n ) 10 SAM would be able to relax its structure to obtain more hydrogen bonds than is possible in a tightly packed, ordered structure that we have treated. Interpolating from our data (Figure 6), we expect that for the same coverage, the difference in Pmax for n ) 10 and 15 to be small, certainly less than a factor of 2. We find that hydrogen bonds are possible for both even and odd chain lengths with more hydrogen bonds possible for odd in the all-trans configuration. Therefore, the even:odd effect should not be large without some significant structural change. Further experiments would be helpful in clarifying the effect. Acknowledgment. Sandia is a multiprogram laboratory operated by Sandia Corp., a Lockheed Martin Company, for the United States Department of Energy’s National Nuclear Security Administration under Contract DE-AC04-94AL85000. We also thank Jack Houston for many helpful discussions of our work. LA0341106 (34) Stevens, M. J. Macromolecules 2001, 34, 2710. (35) Lio, A.; Charych, D. H.; Salmeron, M. J. Phys. Chem. B 1997, 101, 3800. (36) Sun, L.; Kepley, L. J.; Crooks, R. M. Langmuir 1992, 8, 2101.