pubs.acs.org/Langmuir © 2009 American Chemical Society
Friction of Polyaromatic Thiol Monolayers in Adhesive and Nonadhesive Contacts Y. Yang and M. Ruths* Department of Chemistry, University of Massachusetts Lowell, 1 University Avenue, Lowell, Massachusetts 01854 Received May 11, 2009. Revised Manuscript Received July 25, 2009 We have used friction force microscopy to study the effects of adhesion on the boundary friction of self-assembled monolayers of the aromatic compounds thiophenol, p-phenylthiophenol, p-terphenyl thiol, 2-naphthalenethiol, and benzyl mercaptan on gold. To control the adhesion between the monolayer-covered tip and substrate, the friction measurements were made in dry N2 gas or in ethanol. At low loads, low adhesion (in ethanol) resulted in a linear dependence of the friction force on load (i.e., F = μL) whereas higher adhesion between the same monolayers (in N2) gave an apparent area-dependent friction. The friction in the adhesive systems was well described by F = ScA with the contact area, A, calculated for a thin, linearly elastic film confined between rigid substrates using the thin-coating contact mechanics (TCCM) model in a transition regime between its DMT- and JKR-like limits. With increasing packing density of the monolayers, a systematic decrease was found in the friction coefficient (μ) obtained in ethanol and the critical shear stress (Sc) obtained in N2. To describe these aromatic monolayers with the extended TCCM model, a higher Young’s modulus was neeeded than for fatty acid monolayers of similar packing density.
Introduction Self-assembled structures formed by aromatic compounds reflect the complex intermolecular and molecule-substrate interactions in such systems.1,2 Spectroscopic2-16 and electrochemical3,7,9,13,17 investigations, supported by ellipsometry2,7,13,14,16-19 and scanning tunneling microscopy,2,11,12,17,20 have provided information on the structure and optical and electronic properties *Corresponding author. E-mail:
[email protected]. (1) Ulman, A. Acc. Chem. Res. 2001, 34, 855–863. (2) Kang, J. F.; Ulman, A.; Liao, S.; Jordan, R.; Yang, G.; Liu, G.-y. Langmuir 2001, 17, 95–106. (3) Sabatani, E.; Cohen-Boulakia, J.; Bruening, M.; Rubinstein, I. Langmuir 1993, 9, 2974–2981. (4) Jaffey, D. M.; Madix, R. J. J. Am. Chem. Soc. 1994, 114, 3020–3027. (5) Szafranski, C. A.; Tanner, W.; Laibinis, P. E.; Garrell, R. L. Langmuir 1998, 14, 3570–3579. (6) Frey, S.; Stadler, V.; Heister, K.; Eck, W.; Zharnikov, M.; Grunze, M.; Zeysing, B.; Terfort, A. Langmuir 2001, 17, 2408–2415. (7) Kim, B.; Beebe, J. M.; Jun, Y.; Zhu, X.-Y.; Frisbie, C. D. J. Am. Chem. Soc. 2006, 128, 4970–4971. (8) Shaporenko, A.; Terfort, A.; Grunze, M.; Zharnikov, M. J. Electron. Spectrsc. Relat. Phenom. 2006, 151, 45–51. (9) Kolega, R. R.; Schlenoff, J. B. Langmuir 1998, 14, 5469–5478. (10) Ganesh, V.; Lakshminarayanan, V. J. Phys. Chem. B 2005, 109, 16372– 16381. (11) Hallmann, L.; Bashir, A.; Strunskus, T.; Adelung, R.; Staemmler, V.; W€oll, Ch.; Tuczek, F. Langmuir 2008, 24, 5726–5733. (12) Azzam, W.; Wehner, B. I.; Fischer, R. A.; Terfort, A.; W€oll, C. Langmuir 2002, 18, 7766–7769. (13) de Boer, B.; Meng, H.; Perepichka, D. F.; Zheng, J.; Frank, M. M.; Chabal, Y. J.; Bao, Z. Langmuir 2003, 19, 4272–4284. (14) Stoycheva, S.; Himmelhaus, M.; Fick, J.; Korniakov, A.; Grunze, M.; Ulman, A. Langmuir 2006, 22, 4170–4178. Stoycheva, S.; Himmelhaus, M.; Fick, J.; Korniakov, A.; Grunze, M.; Ulman, A. Langmuir 2008, 24, 2260. (15) Himmel, H.-J.; Terfort, A.; W€oll, C. J. Am. Chem. Soc. 1998, 120, 12069– 12074. (16) Shaporenko, A.; Brunnbauer, M.; Terfort, A.; Grunze, M.; Zharnikov, M. J. Phys. Chem. B 2004, 108, 14462–14469. (17) Tao, Y.-T.; Wu, C.-C.; Eu, J.-Y.; Lin, W.-L.; Wu, K.-C.; Chen, C.-h. Langmuir 1997, 13, 4018–4023. (18) McNally, H.; Janes, D. B.; Kasibhatla, B.; Kubiak, C. P. Superlattices Microstruct. 2002, 31, 239–245. (19) Barriet, D.; Yam, C. M.; Shmakova, O. E.; Jamison, A. C.; Lee, T. R. Langmuir 2007, 23, 8866–8875. (20) Jiang, P.; Nion, A.; Marchenko, A.; Piot, L.; Fichou, D. J. Am. Chem. Soc. 2006, 128, 12390–12391.
Langmuir 2009, 25(20), 12151–12159
of several aromatic self-assembled monolayers (SAMs). The bonding and molecular orientation of aromatic thiols to a substrate have also been modeled.15,21,22 Because of the stronger intermolecular interactions between polyaromatic molecules, many of them pack more densely in a monolayer than simple aromatic ones, and their rigidity is useful in forming monolayers with exposed functional terminal groups.1,2 Mineral-oil-based fuels and lubricants contain small amounts of many different aromatic molecules that are known to protect metal surfaces from oxidation and wear.23-25 These include sulfur-containing ones such as thiophenes and thiols.25 Empirically, it is known that the natural lubricity of jet fuel and diesel fuel 25,26 is significantly diminished when their sulfur and aromatics content is reduced through hydrogenation. To compensate for this loss of lubricity, various aromatic, polyaromatic, and heteroaromatic molecules with polar or aliphatic substituents are included in additive formulations to improve the performance of refined fuels, biodiesel, and oil-based lubricants.23,25-27 Similar compounds also find use as friction-modifying additives in aluminum-on-steel sliding.24 Not only the formation of hard protective layers (sulfides, in the case of sulfur-containing additives) but also the structure of the hydrocarbon part of the molecules plays a role in the wear protection, especially at low loads.23-25 Despite these current applications, and the need for more specific additives to meet increasingly stringent emissions standards,28 only limited information is available on the (21) Jung, H. H.; Won, Y. D.; Shin, S.; Kim, K. Langmuir 1999, 15, 1147–1154. (22) Dirama, T. E.; Johnson, J. A. Langmuir 2007, 23, 12208–12216. (23) Forbes, E. S. Wear 1970, 15, 87–96. (24) Heenan, D. F.; Januszkiewicz, K. R.; Sulek, H. H. Wear 1988, 123, 257–268. (25) Wei, D.; Spikes, H. A. Wear 1986, 111, 217-235 and references therein. (26) Kenesey, E.; Ecker, A. In Tribology Science and Application; Herman, M. A., Franek, F., Kajdas, C., Eds.; Scientific Centre of the Polish Academy of Sciences (PAN): Vienna, Austria, 2004; pp 378-394. (27) McCormick, R. L.; Alvarez, J. R.; Graboski, M. S.; Tyson, K. S.; Vertin, K. PT-111 (Alternative Diesel Fuels); Society of Automotive Engineers, 2004, pp 51-60 . (28) Bardasz, E. A.; Antoon, F. A.; Schieferl, E. A.; Wang, J. C. ; Totten, W. SP1894 (Oils, Rheology, Tribology, and Driveline); Society of Automotive Engineers, 2004, pp 151-158.
Published on Web 09/28/2009
DOI: 10.1021/la901678c
12151
Article
Yang and Ruths Table 1. Aromatic Thiol Monolayer Properties; Tip Radius and Friction Coefficient in Ethanol θtilt (deg)a
μc
Tav (nm)
mol area (nm2)
0.4–0.76-8,18.19
0.5
0.4–0.74,17
31–765,6,8,21,22
52 52
1.20 ( 0.05 1.44 ( 0.07
0.01
0.85–1.42,3,6,8,12,14,17,18,29,30
1.0
0.32617
14–322,6,8,18,30
62 62 97
0.91 ( 0.03 1.12 ( 0.04 0.89 ( 0.02
TPT
0.01
1.05–1.83,6,8,13,15,16,18
1.5
0.21671
6–363,6,8,15,16,18
53 53
0.59 ( 0.03 0.57 ( 0.04 0.61 ( 0.02
NT
0.01
0.65–0.87,10
0.7
0.35–0.429,10,20
30–4410,20
134 186
1.51 ( 0.07 1.47 ( 0.03
BM
1
0.611
0.6
0.205–0.21611,17
8–1911,21,22
106
0.76 ( 0.03 0.82 ( 0.02
system
c (mM)
TP
1
PTP
a
T (nm)
R (nm)b
From the surface normal. b ΔR = 3 nm (R20%,68 it is unlikely that the monolayers are linearly elastic; therefore, we do not expect these regions to be described by the TCCM model. (63) Fox, H. W.; Hare, E. F.; Zisman, W. A. J. Colloid Sci. 1953, 8, 194-203.
DOI: 10.1021/la901678c
12157
Article
Monolayer Elastic Modulus. The application of the TCCM model requires an estimate of the monolayer elastic modulus. The spherical indenter (probe) and flat substrate are assumed to be rigid. In our experiments, both of these carry a metal layer (ca. 100 nm Au on the flat surface and 20 nm Cr + 20 nm Au on the tip). The Young’s modulus of bulk gold is 78.5 GPa.64 The modulus of a self-assembled organic film with many defects is likely to be significantly lower than that of the tip and substrate. The value of the uniaxial strain modulus of the monolayer, Eu = E(1 - v)/[(1 + v)(1 - 2v)], affects W (and thus γTCCM) slightly and has a strong effect on the contact radius a at a chosen L. The elastic moduli of self-assembled monolayers are not well known, and a wide range of values is found in the literature. Local compliance measurements with AFM65 suggested a Young’s modulus of E = 0.2-0.4 GPa for closely packed monolayers, and thickness changes on compression of fatty acid monolayers formed by Langmuir-Blodgett deposition gave E = 1-5 GPa.66 Other measurements with AFM and interfacial force microscopybased techniques on closely packed alkanethiol and alkylsilane monolayers suggested67,68 composite moduli of around 10 GPa (including some substrate deformation; cf. discussion in ref 68). Computer simulations of alkanethiol monolayers indicated moduli of around 20 GPa69 and 36 GPa64. These higher values are possibly the result of modeling ideal systems with few defects. In molecular dynamics simulations of a flat plate compressing an alkylsilane monolayer, the monolayer was found to have a uniaxial strain modulus of 3 GPa at 10% nominal strain.70 Contact areas calculated with the extended TCCM model using this modulus were successfully compared to molecular dynamics simulations of the contact between an AFM tip and an alkylsilane monolayer.60,70 When applying the extended TCCM model to the data in Figures 3 and 4, it became apparent that neither the JKR-like limit nor very low values of E described our data well. The experimental friction force increased more rapidly at low L than the calculated area in the JKR limit, and the pull-off occurred at a lower F than the position of the apex of JKR-like curves scaled to fit the experimental data between the lowest loads and the monolayer transition regime. This has been seen before for simple aromatic systems33 and also noted in systems of very loosely packed fatty acid monolayers self-assembled from organic solvent, although for the fatty acids a reasonable fit could still be obtained with a JKR-like model.43,56 This is illustrated in Figure 5, where data from an experiment on TP is compared to a self-assembled fatty acid monolayer43 of similar low packing density. In Figure 5a, curves calculated with the extended TCCM model and different values of the Young’s modulus (E = 0.1, 0.7, and 7 GPa, ν = 0.4) were multiplied by constants (Sc) to bring the curves as close to the experimental data as possible. (The curve for E = 7 GPa, with Sc = 770 MPa, is also plotted in Figure 3a.) As E was increased, the lower branch of the parabolic curve was reduced, and the apex of the curve (64) Lio, A.; Morant, C.; Ogletree, D. F.; Salmeron, M. J. Phys. Chem. B 1997, 101, 4767–4773. (65) Overney, R. M.; Meyer, E.; Frommer, J.; G€untherodt, H.-J.; Fujihira, M.; Takano, H.; Gotoh, Y. Langmuir 1994, 10, 1281–1286. (66) Tsukruk, V. V.; Bliznyuk, V. B.; Hazel, J.; Visser, D.; Everson, M. Langmuir 1996, 12, 4840–4849. (67) Burns, A. R.; Houston, J. E.; Carpick, R. W.; Michalske, T. A. Langmuir 1999, 15, 2922-2930. (68) Vezenov, D. V.; Noy, A.; Lieber, C. M. J. Adhes. Sci. Technol. 2003, 17, 1385-1401. (69) Leng, Y.; Jiang, S. J. Chem. Phys. 2000, 113, 8800–8806. (70) Chandross, M. Personal communication. (71) Azzam, W. Self-Assembled Monolayers on Gold Made from Organothiols Containing an Oligophenyl-Backbone. Ph.D. Thesis, Ruhr-University Bochum, Bochum, Germany, 2003, p 108.
12158 DOI: 10.1021/la901678c
Yang and Ruths
corresponded better to the pull-off region (lowest data points) in the data. The rise of the curve also better approximated the shape of the data. At moduli well above E = 7 GPa (not shown), the distinct pull-off at a finite friction force F was lost (i.e., the apex of the curve was at F = 0), which corresponded to the DMTlike limit. We found that the data in all our five aromatic systems was best described by a modulus 6-7 GPa e E < ca. 15 GPa, except one data set in Figure 3c (R = 300 nm) that was better reproduced with the DMT-like limit of the model (curve marked TCCM-DMT). The transition parameter ζ was in the range of 0.01-0.03, with the higher values obtained for the thicker monolayers (Table 2). It has been noted60 that with a larger tip radius the monolayer appears to be stiffer (less penetration occurs), which is what we are seeing for the two larger radii in Figure 3c (R = 104 and 300 nm), where the pressure was not high enough to induce a detectable monolayer transition. In the loosely packed fatty acid system in Figure 5b, the opposite trend was seen. High values of E gave too steep a rise of the calculated curve and did not reproduce the pull-off region (lowest loads) well. Reasonable fits were found with 0.1 GPa < E < 0.7 GPa in this system and also in other fatty acid systems (oleic, linoleic) from ref 43 (not shown), with values of ζ of around 0.1. This is consistent with the concept that aromatic molecules are stiffer than alkanes and might form stiffer monolayers if sufficiently closely packed. Adhesive versus Nonadhesive Systems. Our observations of the different functional form of the F versus L data at low L in nonadhesive and adhesive systems are consistent with results on other surfaces. A linear increase in F with L has been observed in single-asperity contact between mica sheets experiencing repulsive hydration forces in aqueous electrolyte solution37,40 and in selfassembled monolayer31,33,38-40,42,43 and polymer41 systems under conditions where these surfaces adhered very weakly to one another. In adhesive systems, F commonly shows a nonlinear increase with L that is ascribed to an area dependence.36,40-43,59 Models for the dependence of the friction force on load and contact area have been put forward on the basis of empirical observations. It has been proposed that work has to be done (i) against the external load and (ii) against the adhesion forces (if present) to enable the surfaces to slide past one another (cf. ref 40 and references therein) (i.e., the surfaces must dilate for sliding to occur). This is often expressed as F = μL þ ScA, where one of the terms may dominate the friction response under certain conditions. In one of the simplest models, only the areadependent term depends on the adhesion (interfacial free energy γ), which is incorporated into Sc. This implies that the area dependence of the friction could be largely reduced if the adhesion was decreased and only the μL term would remain. In such cases, the friction would not depend on R, and data obtained with different probe sizes could be directly compared with one another; this appears to be the case in Figures 1 and 2 and has been demonstrated in work on simpler aromatic systems31 and with probe sizes differing by 5 to 6 orders of magnitude.38 Other models suggest that the friction force always depends on the contact area. The linear dependence of F on L is explained as a nonconstant (pressure-dependent) shear stress. Changes in the shear stress at high pressure are certainly possible, especially in view of the monolayer transition we observe that may be preceded by smaller changes in the orientation of the molecules. However, it is unlikely to be the explanation for the linear behavior that we see in these and other systems at low adhesion, in particular, because investigations of the same monolayers with the same tips but at higher adhesion give a different (nonlinear) F versus L. Within the accuracy of our experiments, Sc appears to be Langmuir 2009, 25(20), 12151–12159
Yang and Ruths
a constant when determined from the data at low L (Figures 3 and 4, Table 2). Similar observations have been made in fatty acid monolayer systems using a JKR-like model for layered systems.43 Differences between adhesive and nonadhesive systems are also seen in computer simulations, albeit not as two additive terms. Molecular dynamics simulations of lubricated surfaces (n-hexadecane confined between slightly rough gold surfaces) showed a linear dependence of F on L with a different slope and a shift along the L axis as adhesion was introduced.39 In a recent molecular dynamics simulation of unlubricated (dry) contacts, sublinear and linear F versus L were obtained with and without adhesion, respectively.44 These friction responses were described as F being proportional to the real, atomic-scale contact area of multiasperity contacts with roughness on the atomic scale.44 Since most of our systems have a packing density lower than the maximum close-packing of aromatic rings, there may be slight differences in their molecular-scale roughness even in the dry state (in N2). However, an effect of this was not discernible in our systems because they all showed a nonlinear friction response in adhesive contact. At this point, we do not have information on how the roughness of the monolayers might be altered in contact with a liquid. Previous experiments on BM and closely packed alkanethiol monolayers have suggested (on the basis of measured vs calculated adhesion strengths and the possibility to obtain identical “atomic resolution” images of closely packed monolayers in dry N2 gas and in ethanol) that sliding occurs without an intervening full monolayer of solvent,31,33 but some penetration of small solvent molecules into the monolayers is possible. Dependence of the Friction Force on Packing Density. Although some previous friction force microscopy studies of aromatic monolayers were done under ambient conditions or using nominal spring constants, relative values indicated that their friction typically was higher than that of alkanethiols: The friction coefficient μ of PTP in contact with a Si3N4 tip was ca. 3 times higher than that of hexadecanethiol,29,30 and μ of terphenyl methanethiol was 7-8 times larger than that of octadecanethiol.32 In ethanol, μ of TP probed with a TP-covered tip was 13-15 times larger than that of octadecanethiol.31,33 Practical applications of aromatic friction modifiers do not necessarily rely on a particularly low sliding friction but on their wear resistance, which is likely to increase with increasing stability of the self-assembled monolayer. In pin-on-plate experiments,32 it has been found that terphenyl methanethiol monolayers had a higher wear resistance than octadecanethiol. It is also known from macroscopic experiments23 that dibenzyl disulfide, which forms a BM-like monolayer, protected steel surfaces better from wear than diphenyl disulfide, which forms a TP-like monolayer. A “spring” model has been proposed29,30 in which the high friction of PTP was ascribed to the high stiffness of individual aromatic molecules combined with their stronger lateral interactions, which made it difficult to orient and compress the monolayer under the probe tip. In our five systems, the individual aromatic molecules are believed to be stiffer than alkane chains of similar length and thus more difficult to bend or compress. However, as shown in Figure 6, we found decreasing friction (decreasing μ and Sc) with increasing lateral interactions (i.e., with improved packing and thus increased stiffness of the monolayer structure). The uncertainties in μ are from the linear fits in Figures 1 and 2 (Table 1), and that in Sc (Table 2) is 20% (15% for TPT), as discussed in the Supporting Information. Possible systematic errors in the measurements of F would affect μ and Sc in the same direction.
Langmuir 2009, 25(20), 12151–12159
Article
Figure 6. μ (b) and Sc (O) as a function of average molecular area
(cf. Table 1). The uncertainties in μ are from the linear fits in Figures 1 and 2, and the uncertainty in Sc is 20%, except for in the TPT system, where it is 15%. Both parameters indicate a lower friction in the more closely packed systems.
Even in our most closely packed systems (TPT and BM), the friction did not reach the low values seen for closely packed alkanethiol monolayers.33,35 This overall higher magnitude of friction, taken together with the absence of a distinct velocity dependence (cf. Materials and Methods, data not shown), might be the result of the inherent stiffness of the constituent molecules. This is consistent with the need for a higher Young’s modulus to describe the deformations of the confined aromatic monolayers.
Summary Atomic force microscopy was used to study the effects of adhesion strength and probe size on the friction forces in five aromatic self-assembled monolayer systems with different packing densities. Low adhesion (measurements performed in ethanol, Figures 1 and 2) resulted in a linear increase in the friction force with load (i.e., F = μL) whereas higher adhesion (in N2 gas, Figures 3 and 4) gave an apparent area dependence of the form F = ScA, where Sc is the critical shear stress. By using the extended TCCM model to calculate A versus L, we obtained Sc values at low loads in the adhesive systems. μ and Sc were found to decrease with increasing packing density of the monolayers. A larger value of the Young’s modulus, E g 7 GPa, was needed to reproduce the F versus L curves in these systems compared to those in fatty acid systems of similar packing density, where a good decription was obtained with E < 0.7 GPa. Acknowledgment. We thank J. Mead for access to the contact angle goniometer and T. Petterson for software to analyze the AFM data. Acknowledgment is made to the Donors of the American Chemical Society Petroleum Research Fund for support of this research through grant no. 45101-G5. This work was also supported by start-up funding through the NSF-funded Nanoscale Science and Engineering Center—Center for HighRate Nanomanufacturing (CHN) (award no. NSF-0425826). Supporting Information Available: Determination of tip radius, R, from reverse images. Uncertainty in the critical shear stress, Sc, and in the transition parameter, ζ. This material is available free of charge via the Internet at http:// pubs.acs.org.
DOI: 10.1021/la901678c
12159
