Nội dung

Bhushan, B. “Boundary Lubrication Studies Using Atomic Force/Friction ...” Handbook of Micro/Nanotribology. Ed. Bharat Bhushan Boca Raton: CRC Press LLC, 1999 © 1999 by CRC Press LLC 8 Boundary Lubrication Studies Using Atomic Force/Friction Force Microscopy Bharat Bhushan 8.1 8.2 8.3 Introduction Nanodeformation, Adhesive Forces, and Molecular Conformation Boundary Lubrication Studies Liquid Lubricants • LB and Self-Assembled Monolayers 8.4 Closure References 8.1 Introduction Boundary films are formed by physical adsorption, chemical adsorption, and chemical reaction. The physisorbed film can be either monomolecular or polymolecular thick. The chemisorbed films are monomolecular, but stoichiometric films formed by chemical reaction can have a large film thickness. In general, the stability and durability of surface films decrease in the following order: chemical reaction films, chemisorbed films, and physisorbed films. A good boundary lubricant should have a high degree of interaction between its molecules and the sliding surface. As a general rule, liquids are good lubricants when they are polar and thus able to grip solid surfaces (or be adsorbed). Polar lubricants contain reactive functional groups with low ionization potential or groups having high polarizability (Bhushan, 1993). Boundary lubrication properties of lubricants are also dependent upon the molecular conformation and lubricant spreading (Novotny et al., 1989; Novotny, 1990; Mate and Novotny, 1991; Mate, 1992a). This chapter presents an overview of lubrication studies of polar and nonpolar lubricants and Langmuir–Blodgett and chemically grafted films, using atomic force/friction force microscopy. 8.2 Nanodeformation, Adhesive Forces, and Molecular Conformation Nanodeformation behavior of the bonded lubricant was studied using atomic force microscopy (AFM) by Blackman et al. (1990a). They used Si(100) substrate with about 1.5 nm of native oxide. Just prior to © 1999 by CRC Press LLC FIGURE 8.1 Wire deflection (normal load) as a function of tip–sample separation distance curves comparing the behavior of clean Si(100) surface to a surface lubricated with free and unbonded PFPE lubricant, and a surface where the PFPE lubricant film was thermally bonded to the surface. (From Blackman, G. S. et al. (1990), Phys. Rev. Lett. 65, 3189–3198. With permission.) the application of lubricant, the surface was cleaned with methylene chloride, spun dried, and followed by exposure to ultraviolet-created ozone for several minutes to remove the remaining adsorbates. Liquid films of the perfluoropolyether Z-Dol of about 4-nm thickness were deposited by a dip-coating method. The lubricant molecules were bonded to the substrate via the reactive end groups by heating at 150°C for 1 h, followed by rinsing with a freon solvent to remove any unbonded molecules, leaving behind about 2-nm-thick film. Before bringing a tungsten tip into contact with a molecular overlayer, it was first brought into contact with a bare clean-silicon surface, Figure 8.1. As the sample approaches the tip, the force initially is zero, but at point A the force suddenly becomes attractive (top curve) which increases until at point B where the sample and tip come into intimate contact and the force becomes repulsive. As the sample is retracted, a pull-off force of 5 × 10–8 N (point D) is required to overcome adhesion between the tungsten tip and the silicon surface. The deformation is reversible (elastic) since the retracting (outgoing) portion of the curve (C to D) follows the extending (ingoing) portion. When an AFM tip is brought into contact with a molecularly thin film of a nonreactive lubricant, a sudden jump into adhesive contact is observed. The adhesion is initiated by the formation of a lubricant meniscus surrounding the tip pulling the surfaces together by Laplace pressure. However, when the tip was brought into contact with a lubricant film which was firmly bonded to the surface, the liquidlike behavior disappears. The initial attractive force (point A) is no longer sudden as with the liquid film, but, rather, gradually increases as the tip penetrates the film. Meniscus formation is suppressed because the polymer molecules are no longer free to move about on the surface as at least one end is attached. According to Blackman et al. (1990a), if the substrate and tip were infinitely hard with no compliance in the tip and sample supports, the line for B to C would be vertical with an infinite slope. The tangent to the force–distance curve at a given point is referred to as the stiffness at that point and was determined by fitting a least-squares line through the nearby data points. For bonded lubricant film, at the point where slope of the force changes gradually from attractive to repulsive, the stiffness changes gradually, indicating compression of the molecular film. As the load is increased, the slope of the repulsive force eventually approaches that of the bare surface. The bonded film was found to respond elastically up to the highest loads of 5 µN which could be applied. Thus, bonded lubricant behaves as a soft polymer solid. © 1999 by CRC Press LLC FIGURE 8.2 Tip deflection (normal load) as a function of Z (separation distance) curve for (a) unlubricated and (b) lubricated thin-film magnetic rigid disks. The pull-off force is 42 nN for the unlubricated disk and 64 nN for the lubricated disk calculated from the horizontal distance between points C and D and the cantilever spring constant of 0.4 N/m. (From Bhushan, B. and Ruan, J. (1994), ASME J. Tribol. 116, 389–396. With permission.) The attractive adhesive forces at different parts of the surface can be estimated by bringing the sample into contact with the tip and then measuring the maximum force needed to pull the tip and sample apart, Mate (1993) and Bhushan and Ruan (1994). Figure 8.2 shows typical normal load as a function of separation distance (Z) curves for unlubricated and lubricated (with about 2-nm thickness of a perfluoropolyether lubricant with an alcohol group, Z-Dol) magnetic disk samples. In these experiments, the disks are first brought into contact and then withdrawn from the tip. The presence of the water vapor for the unlubricated disk and along with the lubricant film for the lubricated disk causes a sudden attractive force to occur at point A due to a meniscus of liquid forming around the top of the liquid film and a long break-free distance out to point B where the liquid meniscus breaks as the sample is withdrawn. The major difference between the two curves is that the pull-off force is lower for the unlubricated surface compared with lubricated surface. Pull-off force is determined by multiplying the cantilever spring constant (0.4 N/m) by the horizontal distance between points C and D, which corresponds to the maximum cantilever deflection toward the disks before the tip is disengaged. The horizontal distance/pulloff force is 105 nm/42 nN for the unlubricated disk and 160 nm/64 nN for the lubricated disk. The higher value of the pull-off force in the case of lubricated disk arises from the larger meniscus contribution from the liquid films (Bhushan and Ruan, 1994). Figure 8.3 illustrates two extremes for the conformation on a surface of a linear liquid polymer without any reactive end groups and at submonolayer coverages (Novotny et al., 1989; Mate and Novotny, 1991). At one extreme, the molecules lie flat on the surface, reaching no more than their chain diameter δ above the surface. This would be the case if a strong attractive interaction exists between the molecules and the © 1999 by CRC Press LLC FIGURE 8.3 Schematic representation of two extreme liquid conformations at the surface of the solid for low and high molecular weights at low surface coverage. δ is the cross-sectional diameter of the liquid chain and Rg is the radius of gyration of the molecules in the bulk. (From Mate, C. M. and Novotny, V. J. (1991), J. Chem. Phys. 92, 3189–3196. With permission.) solid. On the other extreme, when a weak attraction exists between polymer segments and the solid, the molecules adopt conformation close to that of the molecules in the bulk, with the microscopic thickness equal to about the radius of gyration Rg. Mate and Novotny (1991) used AFM to study conformation of 0.5- to 13-nm-thick perfluoropolyether molecules on clean Si(100) surfaces. They found that the thickness measured by AFM is thicker than that measured by ellipsometry with the offset ranging from 3 to 5 nm. They found that the offset was the same for very thin submonolayer coverages. If the coverage is submonolayer and inadequate to make a liquid film, the relevant thickness is then the height (he) molecules extended above the solid surface. The offset should then equal 2he, assuming that the molecules extend the same height above both the tip and silicon surfaces. They thus concluded that the molecules do not extend more than 1.5 to 2.5 nm above a solid or liquid surface, much smaller than the radius of gyration of the lubricants ranging between 3.2 and 7.3 nm, and to the approximate cross-sectional diameter of 0.6 to 0.7 nm for the linear polymer chain. Consequently, the height that the molecules extend above the surface is considerably less than the diameter of gyration of the molecules and only a few molecular diameters in height, implying that the physisorbed molecules on a solid surface have an extended, flat conformation. They also determined the disjoining pressure of these liquid films from AFM measurements of the distance needed to break the liquid meniscus that forms between the solid surface and the AFM tip. (Also see Mate, 1992a). For a monolayer thickness of about 0.7 nm, the disjoining pressure is about 5 MPa, indicating strong attractive interaction between the liquid molecules and the solid surface. The disjoining pressure decreases with increasing film thickness in a manner consistent with a strong attractive van der Waals interaction between the liquid molecules and the solid surface. © 1999 by CRC Press LLC FIGURE 8.4 (a) Friction force and (b) normal load over an oscillation of the X-sample position during sliding of the tungsten tip on an Si(100) surface coated with perfluoropolyether lubricant with alcohol end group. (From Mate, C. M. (1992), Phys. Rev. Lett. 68, 3323–3326. With permission.) Attempts to measure mechanical properties of self-assembled monolayer films on Au(111) films have been made by Salmeron et al. (1993). They have used AFM in the tapping mode. This technique has the potential of measuring local viscoelastic properties of lubricant films. 8.3 Boundary Lubrication Studies 8.3.1 Liquid Lubricants Mate (1992b), O’Shea et al. (1992), Bhushan et al. (1995a–c), and Koinkar and Bhushan (1996a,b) used AFM to provide insight into how lubricants function at the molecular level. Mate (1992b) conducted friction experiments on Si(100) surface lubricated with a lubricant with alcohol end group (Z-Dol). In these experiments, the sample was moved rapidly back and forth in the X-direction at a velocity of 1 µm/s, while the normal load on the tip was slowly increased to some maximum value, then decreased back to zero by moving the sample in the Z-direction. Figure 8.4 shows an example of the friction force on the tip during one complete X oscillation of the sample. Initially, the tip moves with the sample, until, at point A, the cantilever wire exerts enough force to overcome the static frictional force and the tip starts to slide across the surface. When the X sample direction is reversed at point B, the tip again moves with the sample until it starts to slide at point C. The upward shift in the normal load over the cycle comes with the slight increase in load as the sample is slowly pushed up against the tip. The slight variation in load during the cycle correspond to a surface roughness of about 0.1 nm. Figure 8.5 shows the average normal load and friction forces during sliding as a function of the Z-sample position for the 3-nm-thick films of different types of lubricants. Each data point in the figure represents the average over the sliding portion of a cycle in the X-direction like the one shown in Figure 8.4. For the liquid film in Figure 8.5a and c, Mate (1992b) noted that, just before the hard wall contact is made, the normal load during sliding becomes more attractive for nonalcohol end group (Z03) than that for alcohol end group (Z-Dol) in the same manner as when no sliding occurs. When the sample is withdrawn, the friction force returns to zero when the hard wall contact is broken. This regime is called full-film lubrication, where shearing of a liquid film takes place resulting in a negligible friction © 1999 by CRC Press LLC FIGURE 8.5 The friction force and normal load as a function of the Z sample position for 3-nm-thick films on Si(100) of (a), (b) unbonded lubricant with unreactive end groups, (c), (d) unbonded lubricant with alcohol end groups and (e), (f) bonded lubricant. The open circles in (f) show the friction force when the experiment is repeated in the same spot on the bonded lubricant. (From Mate, C. M. (1992), Phys. Rev. Lett. 68, 3323–3326. With permission.) (Bowdon and Tabor, 1950). After hard wall contact, one is in boundary lubrication regime, where solid–solid shearing takes place. The transition between the two regimes is very sharp, requiring a change in separation distance less than a chain diameter. In the boundary lubrication regime, the friction force for the liquid films initially rises quickly, but soon increases linearly with load (coefficient of friction in the range of 0.5 to 0.8). Similar values of the coefficient of friction are observed for unlubricated silicon surfaces, indicating that most of the unbonded liquid lubricant is squeezed out from the rubbing surfaces. Some liquid molecules may still be trapped among the microasperities of the tip and contribute to the solid-solid shearing. Figure 8.5e and f show the results from sliding on the bonded lubricant film prepared by conveniently bonding of the alcohol-ended lubricant (by heating Z-Dol film at 150°C for 1 h in order to react the end groups with hydroxyl groups on the silicon surface with native oxide and washing off nonreactive portion with freon). As was the case for the unbonded lubricant, no significant friction is observed until the hard wall contact was made. So, even though the ends of the polymer are rigidly attached to the substrate, the backbone of the polymer apparently has enough flexibility to offer little resistance to the sliding tip except when rigidly compressed between the two surfaces. For the bonded lubricant, the initial coefficient of friction is about 0.3, which is about half that for the unbonded liquid films. The lower coefficient of friction indicates that significantly more molecules are trapped between the rubbing surfaces than for the unbonded lubricant. With repeated traversals of the sliding tip, these attached molecules eventually wear away and the coefficient of friction increases with increasing load. Mate (1992b) concluded that the liquid films have negligible shear stress to applied shear strains until the molecules are completely compressed or squeezed out from between the sliding surfaces, showing that hydrodynamic lubrication can occur for surfaces separated by only a few cross-sectional diameters of the polymer backbone. (Also see O’Shea et al., 1992, 1994). The addition of alcohol end groups greatly improves the resistance of the molecules to being squeezed out from between the sliding surfaces. Koinkar and Bhushan (1996a,b) studied the friction and wear performance of Si(100) sample lubricated with about 2-nm-thick Z-15 and Z-Dol perfluoropolyether (PFPE) lubricants. Z-Dol film was thermally bonded at 150°C for 30 min and washed off with a solvent to provide a chemically bonded layer of the lubricant film. Data showing the effect of environment on the unlubricated and lubricated samples are summarized in Figure 8.6. Note that lubricated silicon samples show a lower value of © 1999 by CRC Press LLC FIGURE 8.6 Coefficient of friction for unlubricated and lubricated samples in ambient (~50% RH), dry nitrogen (~5% RH), and dry air (~5% RH). (From Koinkar, V. N. and Bhushan, B. (1996), J. Vac. Sci. Technol. A 14, 2378–2391. With permission.) coefficient of friction than that of unlubricated sample. Furthermore, sample lubricated with Z-Dol FIGURE 8.7 Friction force as a function of scanning velocity for unlubricated and lubricated samples in ambient and dry nitrogen environments. Normal loads used are given in the figure. (From Koinkar, V. N. and Bhushan, B. (1996), J. Vac. Sci. Technol. A 14, 2378–2391. With permission.) exhibits a lower value of the coefficient of friction than that of the Z-15 lubricated sample. For the unlubricated and lubricated samples, the coefficient of friction in a dry environment is lower than at ambient of about 50% relative himidity. At high humidity, the condensed water film from the environment results in dewetting of the lubricant film (or water film for the unlubricated sample) resulting in poorer lubrication performance. Figure 8.7 shows the effect of scanning velocity on the coefficient of © 1999 by CRC Press LLC FIGURE 8.8 Friction force as a function of number of cycles using an Si3N4 tip at a normal load of 300 nN for unlubricated and lubricated samples in ambient environment. Arrows in the figure indicate significant changes in the friction force because of removal of surface or lubricant film. (From Koinkar, V. N. and Bhushan, B. (1996), J. Vac. Sci. Technol. A 14, 2378–2391. With permission.) fricition in ambient and dry nitrogen environments. The coefficient of friction of unlubricated silicon sample and lubricated with Z-15 decreases with an increase of the scanning velocity in the ambient environment, whereas the sample lubricated with Z-15 in the dry nitrogen and the sample lubricated with Z-Dol in both ambient and dry nitrogen environments do not show any velocity dependence. We believe that alignment of free liquid molecules at higher scanning velocities results in lower values of the coefficient of friction. To study lubricant depletion during microscale measurements, Koinkar and Bhushan (1996a,b) conducted nanowear studies using Si3N4 tips. They measured friction as a function of the number of cycles for virgin silicon and silicon surfaces lubricated with Z-15 and Z-Dol lubricants, Figure 8.8. An area of 1 × 1 µm was scanned at a normal force of 300 nN. Note that the friction force in a virgin silicon surface decreases in a few cycles after the natural oxide film present on silicon surface gets removed. In the case of Z-15-coated silicon sample, the friction force starts out to be low and then approaches that of an unlubricated silicon surface after a few cycles. The increase in friction of the lubricated sample suggests that the lubricant film gets worn and the silicon underneath is exposed. In the case of the Z-Dol-coated silicon sample, the friction force starts out to be low and remains low during the 100-cycles test. It suggests that Z-Dol does not get displaced/depleted as readily as Z-15. (Also see Bhushan et al., 1995a.) Microwear studies were also conducted using the diamond tip at various loads. Figure 8.9 shows the plots of wear depth as a function of normal force and Figure 8.10 shows the wear profiles of worn samples at 40 µN normal force. The Z-Dol-lubricated sample exhibits better wear resistance than the unlubricated and Z-15-lubricated silicon samples. Wear resistance of the Z-15-lubricated sample is little better than that of the unlubricated sample. The Z-15-lubricated sample shows debris inside the wear track. Since Z-15 is a liquid lubricant, debris generated is held by the lubricant and it becomes sticky, which moves inside the wear track and does damage, Figure 8.10. 8.3.2 LB and Self-Assembled Monolayers Organized and dense molecular-scale layers of, preferably, long-chain molecules have shown to be superior lubricants on both macro- and microscales as compared with freely supported multimolecular layers (Bhushan et al., 1995b,c). Common methods to produce monolayers and thin films (Ulman, 1991) are the Langmuir–Blodgett (LB) deposition (Roberts, 1990) and self-assembled films by chemical grafting of molecules (Jaffrezic-Renault and Martelet, 1992). The LB films are bonded to the substrate by weak © 1999 by CRC Press LLC FIGURE 8.9 Wear depth as a function of normal force using a diamond tip for unlubricated and lubricated samples after one cycle. (From Koinkar, V. N. and Bhushan, B. (1996), J. Vac. Sci. Technol. A 14, 2378–2391. With permission.) van der Waals attraction; whereas the grafting process forms directly covalently bonded dense films of long-chain molecules. Bhushan et al. (1995b,c) conducted micro- and macroscale friction and wear studies on single-grafted and double-grafted octodecyl (C18) and LB films and their subsurfaces. C18 films were produced by grafting of long-chain organic molecules onto an Si(100) wafer covered with a thermally grown SiO2 layer, Figure 8.11. The structure of the LB film consisted of an octadecylthiol (ODT) coated gold sample (gold films thermally evaporated onto single-crystal silicon) on top of which a single, upper inverted bilayer of zinc arachidate (C20, ZnA) was deposited by the LB technique, Figure 8.12. Macro- and microscale friction and wear data are summerized in Table 8.1 and Figure 8.13. We note that C18 doublegrafted film exhibits the lowest coefficient of friction of 0.018 as compared with other samples measured in this study (the coefficient of friction of ZnA films is ~0.03). The coefficient of friction for LB film is comparable to that of the ODT layer and lower than the Au film. The wear resistance of C18 doublegrafted film is much better than C18 single-grafted, LB, and Au films and is comparable with that of SiO2 film. The C18 double-grafted films can withstand much higher normal force of 40 µN as compared with 200 nN for the case of LB films. Surface profiles showing the worn regions after one scan cycle for C18 double-grafted and ZnA films are shown in Figure 8.14. Nanoindentation studies conducted by Bhushan et al. (1995c) showed that the C18 double-grafted films are more rigid than the LB films which may be responsible for the high wear resistance of C18 films. Flexibility of choosing the alkyl chain length, functional terminal group, and cross-linking enables the adaptability of the grafting process for lubrication of microcomponents. 8.4 Closure In nanodeformation experiments, bonded lubricants behave as soft polymer solids. AFM/friction force microscopy friction experiments show that lubricants with polar (reactive) end groups dramatically increase the load or contact pressure that a liquid film can support before solid–solid contact and these exhibit long durability. Chemically grafted films may be suitable for lubrication of microcomponents. © 1999 by CRC Press LLC