Friction and wear are part and parcel of all walks of life, and for interfaces that are in close or near contact, tribology and mechanics are supremely important. They can critically influence the efficient functioning of devices and components. Nanoscale friction force follows a complex nonlinear dependence on multiple, often interdependent, interfacial and material properties. Various studies indicate that nanoscale devices may behave in ways that cannot be predicted from their larger counterparts. Nanoscale friction and wear mapping can help identify some ‘sweet spots’ that would give ultralow friction and near-zero wear. Mapping nanoscale friction and wear as a function of operating conditions and interface properties is a valuable tool and has the potential to impact the very way in which we design and select materials for nanotechnology applications.
Nanotechnology, defined literally as any technology performed on a nanoscale that has applications in the real world (Feynman 1960), has spurred the development of innovative micro/nanosystems with the discovery of novel materials, processes and phenomena on the micro/nanoscale and led to the rapid advancement of micro/nanoelectromechanical systems (MEMS/NEMS) and their various biological and biomedical applications (BioMEMS). Recent years have seen a multitude of new emerging applications in this field. Commercial applications such as the microfluidic devices that can manipulate tiny amounts of fluids, ‘lab-on-chip’ sensors used for drug delivery, accelerometers used for automobile air bag deployment, and digital micromirror devices used in hi-definition TVs and video projectors in homes and theatres are just the tip of the iceberg (Bhushan 2007a). In fact, these MEMS/NEMS are now believed to be the next logical step in the ‘silicon revolution’. Visionaries and leading scientists and researchers, presenting at the National Nanotechnology Initiative Workshop on Nanotechnology in Space Exploration held in Palo Alto, CA (USA) in August 2004, have slated the emerging field of nanotechnology to be the next disruptive technology that will have a major impact on the next one to three decades. It is estimated that the annual global impact of products where nanotechnology will play a key role will exceed US$1 trillion by 2015 and would require approximately 2 million nanotechnology workers (Roco 2003).
Despite the increasing popularity and technological advances in nanotechnology applications, the severe tribological (friction and wear) problems tend to undermine their performance and reliability. In fact, several studies have shown that the tribology and mechanics of these devices are the limiting factors to the imminent broad-based impact of nanotechnology on our everyday lives (Maboudian & Howe 1997; Bhushan 1998, 2003, 2007a,b). Miniaturization and the subsequent development of devices for nanotechnology applications require better tribological performance of the system components and a fundamental understanding of the basic phenomena underlying friction, wear and lubrication on the micro- and nanoscale (Bhushan 1997, 1998, 1999a,b, 2001, 2007a,b). The components used in the micro/nanostructures are very light (of the order of a few micrograms) and operate under very light loads (of the order of a few micrograms to a few milligrams). Moving from the macro- to nanoscale, the surface area-to-volume ratio increases considerably and becomes a cause of serious concern from the tribological point of view. On the nanoscale, surface forces, such as friction, adhesion, meniscus forces, viscous drag and surface tension, which are proportional to area, significantly increase and can limit the life and reliability of nanotechnology applications.
2. Measurement technique
The emergence of the new field of nanotribology, which pertains to the experimental and theoretical investigations of interfacial processes occurring during adhesion, friction, wear and thin film lubrication of sliding surfaces on the scales ranging from the atomic and molecular scale to the microscale, and its associated techniques (Bhushan 1999a) have provided a viable means of addressing the tribological issues on the nanoscale. Nanotribological investigations can be performed using the surface force apparatus that was pioneered by David Tabor, R. H. S. Winterton and Jacob Israelachvili in the early 1970s at Cambridge University, and the atomic force microscope (AFM) that was developed by Binnig et al. (1986). These instruments have already provided valuable insights into the behaviour of materials on the nanoscale (Bhushan et al. 1995; Bhushan 1999a, 2007a).
For studying surface interaction on the micro/nanoscale, the sharp tip of an AFM is ideally suited and has been successfully employed by a number of researchers for studying friction and wear properties of various materials, coatings and lubricants (Mate et al. 1987; Ruan & Bhushan 1994; Koinkar & Bhushan 1996; Bennewitz et al. 2001; Bhushan & Liu 2001; Liu & Bhushan 2003; Tambe & Bhushan 2004, 2005a–i; Tambe 2005; Gnecco et al. 2007; Tao & Bhushan 2007). Contrary to the classical friction laws postulated by Amontons (1699) and Coulomb (1785) centuries ago, nanoscale friction force is found to be strongly dependent on the normal load and sliding velocity. Many materials, coatings and lubricants that have wide applications show reversals in friction behaviour corresponding to the transitions between friction mechanisms (Bowden & Tabor 1950, 1964; Singer & Pollock 1992; Bhushan 1999b; Tambe 2005; Tambe & Bhushan 2005b,f,i; Gnecco et al. 2007; Tao & Bhushan 2007). Recently, Tambe & Bhushan (2005a) and Tao & Bhushan (2006) developed new AFM-based techniques for studying the effect of sliding velocity on nanoscale friction behaviour for velocities from a few μm s−1 to hundreds of mm s−1. This has enabled friction investigations to be conducted in velocity ranges that are of scientific as well as engineering significance.
The AFM can be used to investigate how surface materials can be moved or removed on the nanoscale, for example, in scratching and wear (Bhushan 1999a, 2005), where these things are undesirable and nanofabrication/nanomachining, where they are desirable (Bhushan 2007a). A ‘continuous microscratch’ technique developed for an AFM by Sundararajan & Bhushan (2001) gave a direct dependence of the scratch/wear depth on the applied normal load and has been used for understanding critical loads for ‘visible’ wear damage (Sundararajan & Bhushan 2001; Liu & Bhushan 2002; Tambe & Bhushan 2005g). Figure 1 shows a typical scanning electron microscopic image of such a wear mark and the associated wear particles. The mechanism of material removal during nanoscale wear under ultralow loads was studied by Koinkar & Bhushan (1997) and Zhao & Bhushan (1998) using both a scanning electron microscope and a transmission electron microscope (TEM). They reported an increase in the number and size of cutting-type particles with the normal load, thereby suggesting nanoscale wear by plastic deformation. They undertook a systematic study to analyse wear debris and concluded based on the evidence they found from the TEM images and the diffraction patterns that the strain fields arising inside the wear mark with no applied stress, the ribbon-like wear debris observed outside the wear mark, an absence of phase transformation (amorphization) and the existence of dislocation arrays all pointed to a process wherein the material was being removed by a cutting action via plastic deformation and with a small contribution from elastic fracture.
To understand the wear mechanisms on the nanoscale, Bhushan et al. (1994) studied the evolution of wear using an AFM. They observed that wear evolution was not uniform, but was initiated at the sites of nanoscratches where the surface defects (with high surface energy) acted as initiation sites for wear. Wear precursors (precursors to measurable wear) were studied by making surface potential measurements (DeVecchio & Bhushan 1998; Bhushan & Goldade 2000a,b). The contact potential difference, or simply the surface potential between two surfaces, depends on a variety of parameters such as electronic work function, adsorption and oxide layers. The surface potential map of an interface gives a measure of changes in the work function, which is sensitive to both physical and chemical conditions of the surfaces, including structural and chemical changes. Before the material is actually removed in a wear process, the surface experiences stresses that result in surface and subsurface changes of structure and/or chemistry. These can cause changes in the measured potential of a surface. An AFM tip allows mapping of the surface potential with nanoscale resolution. Surface height and change in the surface potential maps of a polished single-crystal aluminium (100) sample abraded using a diamond tip at loads of 1 and 9 μN are shown in figure 2a. (Note that the sign of the change in the surface potential is reversed here from that in DeVecchio & Bhushan (1998).) It is evident that both the abraded regions show a large potential contrast (approx. 0.17 V) with respect to the non-abraded area. The black region in the lower right-hand part of the topography scan shows a step that was created during the polishing phase. There is no potential contrast between the high region and the low region of the sample, indicating that the technique is independent of surface height. Figure 2b shows a close-up scan of the upper (low load) wear region shown in figure 2a. Note that while there is no detectable change in the surface topography, there is, nonetheless, a large change in the potential of the surface in the worn region. Indeed, the wear mark in figure 2b might not have been visible at all in the topography map, were it not for the noted absence of wear debris generated nearby and then swept off during the low load scan. Thus, even in the case of zero wear (no measurable deformation of the surface using the AFM), there can be a significant change in the surface potential inside the wear mark, which is useful for the study of wear precursors. It is believed that the removal of the thin contaminant layer including the natural oxide layer gives rise to the initial change in the surface potential. The structural changes, which precede the generation of wear debris and/or measurable wear scars, occur under ultralow loads in the top few nanometres of the sample and are primarily responsible for the subsequent changes in the surface potential.
One way of exploring the broader wear patterns is to construct wear mechanism maps (Tabor 1983) that summarize the data and models for wear, thereby showing not only how the mechanisms interface but also allowing the dominant mechanisms for any given set of conditions to be identified. On the macroscale, the approach followed by various researchers to map wear involves running a multitude of experiments that measure the wear rate at a given normal load (or normal pressure) and a relative sliding velocity and then plotting wear maps based on the failure data or the rate of removal of material during wear (Bowden & Tabor 1950, 1964; Singer & Pollock 1992; Bhushan 1999b). Lim & Ashby (1987) constructed wear maps using empirical data as well as theoretical analysis. They demonstrated the utility of the wear mechanism mapping method as a way of classifying and ordering wear data and of showing the relationships between competing wear mechanisms.
A novel AFM-based technique was developed by Tambe & Bhushan (2005h) to generate wear maps on the nanoscale by varying the sliding velocity, the number of sliding cycles and the normal load. For generating nanoscale wear maps, a raster scanning mode was used. The sample was oscillated using the piezo stage and simultaneously the AFM tip was dragged perpendicular to the direction of motion of the sample (figure 3). The sample oscillation frequency (scan speed), the AFM tip velocity and the normal load were controlled to achieve appropriate relative sliding velocities, normal loads and a specific number of sliding cycles. The relative sliding velocity was varied by changing the scan speed, while the number of sliding cycles was varied by varying the rate of movement of the AFM tip (i.e. the velocity of the AFM tip while sliding perpendicular to the direction of motion of the sample; this varying rate of movement results in the AFM tip residing for different time intervals on the sample surface and thereby the number of sliding cycles obtained at each location on the sample surface is different).
For studying the effect of increasing normal load and increasing number of sliding cycles at constant sliding velocity, the AFM tip was programmed to make controlled movements at desired normal loads and velocities. This controlled motion of the AFM tip was achieved using custom software code written in NanoScript1 (Anon. 1999). To achieve varying sliding velocities across the scan area, the input voltage pulse for the piezo stage was slightly modified. In normal operation, a triangular time-varying voltage pulse is provided to the piezo stage to achieve scanning operation and to obtain a constant sliding velocity. For the wear mapping experiments, a parabolic voltage pulse was used as an input to drive the piezo. This resulted in a steady increase in the sliding velocity across the scan area. The synchronized and controlled movement of the AFM tip and the sample is a novel approach for obtaining nanoscale wear maps and helps generate a visual representation of the sample surface wear as a function of sliding velocity, applied normal load and number of sliding cycles (Tambe & Bhushan 2005h). Using this approach, a wear map can be generated in one single experiment, as against the rather cumbersome approach where the researchers conduct multiple experiments at different normal loads, sliding velocities and for different number of sliding cycles and then generate a contour map for the sample wear based on individual data points obtained from each experiment.
3. Nanoscale friction maps and mechanisms
Most of the analytical models developed for explaining nanoscale friction behaviour have remained limited in their focus and have left investigators short-handed when trying to explain friction behaviour scaling multiple regimes. Nanoscale friction maps provide fundamental insights into friction behaviour. They help identify and classify the dominant friction mechanisms, as well as determine the critical operating parameters that influence transitions between different mechanisms (Tambe & Bhushan 2005i). Figure 4 shows the nanoscale friction maps obtained by varying the normal load and the sliding velocity (details of the samples used are given in table 1). The contours represent constant friction force lines and are marked by the value of the friction force in nN. Horizontal contour lines indicate the velocity-independent nature of the friction force. This behaviour is found at high velocities for highly oriented pyrolytic graphite (HOPG), at moderately high velocities for Al and at low velocities for polymethylmethacrylate (PMMA). Studies on the nanoscale friction force dependence on velocity (Riedo et al. 2003; Gnecco et al. 2007) indicate that friction force becomes constant relative to sliding velocity when the atomic scale stick–slip occurring at low sliding velocities loses its dominance. A constant friction force with respect to sliding velocity would appear as horizontal contour on a friction map and is the reason, for example, for the horizontal contours observed for HOPG at high velocities. Vertical contour lines indicate a normal load independence of friction force. In all the samples studied, this behaviour is not observed, although the steep contour lines for Si(100), HOPG and Al indicate that there is a very small normal load dependence on friction at low velocities. For all practical instances, it would be impossible to find a material that shows normal load independence of friction force.
Some other characteristic contours are those with slanting lines (with either a positive or negative slope). These friction contours arise from the microscale stick–slip-related contributions or from the formation of meniscus bridges by preferential condensation of liquid films at the sliding interface, particularly for hydrophilic interfaces. Stick–slip can originate from the atomic scale stick–slip leading to an increase in friction force with velocity (Riedo et al. 2003; Gnecco et al. 2007) and thus appearing as slanted contours with a positive slope. HOPG, diamond-like carbon (DLC) and Al showed this behaviour. Stick–slip originating as a result of some other mechanisms can result in a decrease in friction force with an increase in sliding velocity (Ruths & Israelachvili 2007). This behaviour would result in slanted contours with a negative slope on the friction map such as that seen for PMMA at high sliding velocities and for polydimethylsiloxane (PDMS). The friction force arising from meniscus contributions, found for the hydrophilic surfaces such as Si(100), results in the drop in friction force with an increase in sliding velocity. A minimum threshold equilibrium time is necessary for the formation of stable meniscus bridges at contacting and near-contacting asperities for a sliding interface (Bouquet et al. 1998; Tambe & Bhushan 2005b). With increasing velocity, fewer meniscus bridges build up at the interface, and thus the overall contribution to friction force drops with an increase in velocity. This behaviour would manifest itself in the form of slanted contour lines with a negative slope on the friction map.
Contour maps can also consist of concentric contour lines suggesting a peak friction force for a particular critical normal load–sliding velocity combination. Any change in the normal load or the sliding velocity around this critical value would result in a decrease in the friction force. This kind of behaviour typically would imply localized melting at the contact zone or a phase transformation by the formation of a low friction phase at the interface. Localized melting would arise from very high frictional energy dissipation and is expected particularly in the case of polymer materials. PMMA appeared to show concentric lines at moderately high velocities; however for the given range of normal load and sliding velocity, the experimental evidence is not sufficient to support this hypothesis. Phase transformation has been known to occur for DLC resulting in a low friction graphite-like layer by an sp3 to sp2 phase transition (Grill 1997; Tambe & Bhushan 2005e). The sharp Si3N4 tip used in the AFM studies (30–50 nm radius) gives rise to contact pressures in the range of 1.8–3.8 GPa for DLC films corresponding to the normal loads of 10–100 nN (assuming Hertzian contact analysis for single-asperity elastic contact), and Voevodin et al. (1996) have reported the formation of debris with polycrystalline graphite-like structure for contact pressures in the range of 0.8–1.1 GPa.
Another interesting facet in the friction maps is the region where the contour lines change direction. This implies a change in the dominant friction mechanism, which can be abrupt as seen for Si(100) and Al or gradual as for DLC and PDMS. Generally speaking, the contours for each material are characteristic of the friction behaviour exhibited by that particular material. There are some features that can be classified as universal irrespective of the material though. Each characteristic contour suggests a specific friction mechanism. Based on the specific arrangement of contour lines, the dominant friction mechanisms can be identified. In figure 4b, the most commonly observed features in the contours (or the characteristic contours) as found from the study are summarized and the significance of each is stated. It is evident that different mechanisms are dominant for different operating conditions and the factors that influence them include not only the normal load and the sliding velocity but also other factors such as the operating environment (humidity and temperature), surface roughness and mechanical properties of the interface.
The interdependence between friction and material properties on the nanoscale is of significant interest for selecting materials that would be ideal from the tribology point of view, i.e. materials with low friction and adhesion. Scientific studies indicate that mechanical properties can strongly affect the tribological performance (Bhushan 1999b). Efforts to explicitly characterize the nanoscale friction and adhesion of various materials on the basis of their mechanical properties remain limited though. Recently, Tambe & Bhushan (2005f) have established a link between Young's modulus of materials and their coefficient of friction and adhesive force over a range of sliding velocities.
Table 2 lists the materials used in this study along with their Young's modulus (E) and the corresponding coefficient of friction values. Also listed are the materials studied by previous researchers under identical experimental and environmental conditions. In the case of the materials for which a range of values has been reported for the coefficient of friction, the average values are listed. A clear trend was observed for the coefficient of friction dependence on E (figure 5). Low-E materials show higher coefficient of friction when compared with the high-E materials. (It has to be noted that the sliding interface will only undergo elastic deformations under the very low normal loads used in the experiments.) This result can be intuitively inferred from the classical theories of friction (Bowden & Tabor 1964; Bhushan 1999b; Persson 2000). An approximate relation can be developed between the coefficient of friction and Young's modulus by assuming a single-asperity elastic contact and using Hertzian contact analysis. For most sliding interfaces though, the contact is often a multi-asperity contact and no closed-form analytical solutions exist. Numerical methods have to be employed for solving problems dealing with multi-asperity contact (Bhushan 1999b). Moreover, nanoscale friction and adhesion are largely dependent on the sample surface roughness and the shear strength of the sliding interface (Bhushan 1999b; Persson 2000). Table 2 indicates that the roughness values of the samples are not the same, although they are comparable. In light of this limitation on the analytical formulations and the inherent complexity involved in relating the nanoscale friction and adhesion to the material properties, the near logarithmic dependence of the coefficient of friction on Young's modulus shown by a wide variety of materials in the experimental data is extremely interesting. Only HOPG stands out as an anomaly in this trend and this is owing to its extremely low shear strength.
Figure 6 shows the velocity dependence of adhesion and the coefficient of friction over a range of velocities for different materials. The contour maps (figure 6) give Young's modulus dependence of these two quantities for the same materials. Some very interesting trends are revealed by the contour maps. Adhesion is high for low-E materials and at low sliding velocities, and it gradually decreases with an increase in velocity. The velocity dependence of friction and adhesion in the case of viscoelastic polymers (such as PDMS used in this study) is well known, and is defined by a definite peak occurring at a specific sliding velocity (Moore 1972). Similar behaviour was found even in the case of materials with higher Young's modulus where the adhesive force increased with the velocity and reached a peak. Moreover, the peak was attained at higher velocities for materials with higher E value. The contour map for coefficient of friction also reveals peculiar trends. At low velocities, the coefficient of friction decreases with increasing Young's modulus, and this decrease is found to be nearly logarithmic in nature (figure 5). However, at high velocities, this trend is reversed and the coefficient of friction is found to increase with increasing Young's modulus. At high velocities, friction is primarily governed by impact deformations and ploughing effect (Bhushan 1999b; Tambe & Bhushan 2005b). Thus, while low-E materials are able to absorb most of the impacts during sliding, for the high-E materials, the impacts during sliding result in high friction. DLC is the only high-E material that shows a decrease in the coefficient of friction at very high sliding velocities. The reason for this anomaly is the phase transformation of the amorphous DLC to a low shear strength graphite-like phase (Voevodin et al. 1996; Grill 1997; Tambe & Bhushan 2005e).
The contour map reveals a small central zone of very low friction. The corresponding adhesive force values for this zone are also moderate. This can be considered as a ‘sweet spot’ and corresponds to an ideal material that a tribologist would like to choose. HOPG falls in this zone. Various other zones of interest are shown in figure 7. They indicate the dependence of Young's modulus on the operating parameters and thus can be used as a guide for material selection for various nanotechnology applications. We find that, from the tribology point of view, low-E materials show promise for high sliding velocity applications, while high-E materials are more suitable for relatively lower sliding velocities.
Material maps created by plotting the coefficient of friction and adhesion as a function of Young's modulus reveal various interesting facets of the behaviour of nanoscale friction. For example, the coefficient of friction decreased with increasing velocity for materials with low Young's modulus, but the reverse was true for materials with high Young's modulus. The map shows that if the sliding velocities are high, then a compliant material would perform better than a stiffer material.
4. Nanoscale wear maps and mechanisms
Similar to friction mapping, one way of exploring the broader wear patterns is to construct wear mechanism maps that summarize data and models for wear, thereby showing mechanisms for any given set of conditions. Figure 8a shows the wear maps obtained for Si(100) by varying different operating parameters. The wear mark is roughly located at the centre of each image. The arrow marks on the sides of the AFM images indicate the beginning and the end of the wear marks. A larger area was imaged after the wear mapping tests to enable comparison of the worn surface with respect to the virgin surface in its vicinity. The AFM images reveal the dependence of wear on the operating parameters. Drastic failure was observed for high normal loads. This is evident from the large amount of debris found for the experiment conducted by keeping the sliding velocity and the number of sliding cycles constant at 2.5 mm s−1 and 500, respectively, and varying normal load from 0 to 5000 nN. For low loads, no visible wear debris is found. Wear edges start becoming visible approximately for loads over 2000 nN and ultimately catastrophic failure is seen at and above approximately 4000 nN. A similar experiment was conducted by maintaining a constant sliding velocity and varying the number of sliding cycles and the normal load, but the number of cycles was increased to 2500 and the normal load was varied from 0 to 1000 nN. In this case, no catastrophic failure was observed. The wear mark was visible in the form of piled-up debris at the edges. The pile-up was higher at higher normal loads. This indicates that for the given set of test parameters, the effect of normal load is more pronounced. The third AFM image in figure 8a corresponds to the wear map obtained by keeping the sliding velocity and the normal load constant at 2.5 mm s−1 and 1000 nN, respectively, and varying the number of sliding cycles from 0 to 250 across the scan area. In this case, the amount of debris pile-up was minimal and the wear mark edges were barely visible.
Figure 8b shows the results obtained for DLC. Wear maps were obtained by varying the normal load from 0 to 1000 nN and keeping the number of sliding cycles and the sliding velocity constant. For experiments conducted at 200 μm s−1, the wear mark edges were barely visible. However, considerable wear was visible for a sliding velocity of 2.5 mm s−1. The wear mark generated suggests that the effect of sliding velocity is more profound than that of the normal load. The larger concentration of debris particles, towards the end of the wear region, indicates that in general higher wear occurs for higher normal loads as expected. The effect of the number of sliding cycles on wear behaviour was investigated by keeping the normal load constant at 500 nN and the sliding velocity constant at 2.5 mm s−1. The wear mark generated from these experiment shows larger accumulation of wear debris for a larger number of sliding cycles. The wear marks for DLC appear ‘fuzzy’ as the loose debris easily moves during imaging. In comparison with Si(100), DLC sample shows larger amount of wear debris for lower normal loads and number of sliding cycles. Also, the effect of sliding velocity is found to be more profound on the generation of wear particles. The mechanisms of wear in both Si(100) and DLC are completely different. While Si(100) is a brittle material and wear occurs by two- and three-body abrasion, for DLC wear is the result of phase transformation as discussed above.
The wear maps in figure 8 indicate that wear debris particles are generated only for certain combinations of sliding velocities, normal loads and number of sliding cycles. In these wear maps, only one operating parameter was varied at a time. To obtain true wear maps that can reveal different wear mechanisms simultaneously, it is necessary to vary both normal load and sliding velocity and investigate the resulting wear. Such a nanowear map obtained for DLC for a normal load range of 0–1000 nN and sliding velocity range of 0–2.5 mm s−1 is shown in figure 9. Wear debris was seen to form only for particular sliding velocities and normal loads, i.e. beyond certain threshold frictional energy dissipation. Hence, the wear area was curved indicating that for lower velocities and lower normal loads, there is no phase transformation. For clarity, the wear mark corners are indicated by white dots in the AFM image and the various zones of interest over the entire wear mark are illustrated in figure 9.
Analogous to nanowear mapping, the nanoscale friction maps can also be generated by extending the same technique to monitoring of the friction force during scanning for wear. In §3, we discussed the nanoscale friction mapping as a function of two operating parameters: sliding velocity and normal load. In addition, friction force can be plotted as a function of the number of sliding cycles, thereby giving the time dependence as well. The nanofriction mapping in conjunction with the nanowear mapping can provide valuable information regarding the operating parameter dependence of nanoscale friction and wear. Tambe & Bhushan (2005d) have demonstrated the effectiveness and utility of these techniques when used in tandem while studying the phase transformation-related reduction in friction force for DLC.
5. Friction and wear mechanisms on nanoscale and comparison with macroscale
Friction and wear are part and parcel of all walks of life, and for interfaces that are in close or near contact, tribology and mechanics are supremely important. They can critically influence the efficient functioning of devices and components. Friction and wear at a sliding interface depend on the operating conditions such as normal load and sliding velocity; material properties such as Young's modulus and hardness; environmental conditions such as humidity and the medium to which the interface is exposed, such as air, a specific gas or simply water; and interface properties such as surface roughness and surface energy. Many of the commonly observed friction and wear mechanisms are shown in figure 10. A review of classical literature involving the pin-on-disc type of set-up as well as recent investigations with AFMs (Tambe 2005) would show that the order in which the mechanisms are illustrated in the figure from left to right follows from an increasing order of dominance with the increasing sliding velocity and/or normal load, with the ones on the left found to dominate at low sliding velocities and/or normal loads. It should of course be noted that this precedence order is an observation found from studying many materials, coatings and lubricants and could be different for some cases. On the macroscale, most of these mechanisms exist but they may not all have a role to play simultaneously. On the nanoscale, however, this is not the case. The nanoscale friction force follows a complex nonlinear dependence on multiple, often interdependent, interfacial and material properties.
Studies have shown that the fundamental laws of friction, as stated by Amontons and Coulomb, no longer hold on the nanoscale and tribological properties such as coefficient of friction and wear rates can be different on the nanoscale than on the macroscale (Bhushan et al. 1995; Bhushan & Kulkarni 1996; Bhushan 1999a,c, 2007a). Many studies have shown a strong size or scale dependence for mechanical properties such as indentation hardness (Bhushan & Koinkar 1994; Bhushan & Kulkarni 1996; Bhushan et al. 1996; Bhushan 1998, 1999a–,c, 2007a; Nix & Gao 1998; Hutchinson 2000), tensile strength (Hutchinson 2000) and bending strength (Sundararajan & Bhushan 2002), indicating that the bulk properties of many materials differ from those on the micro/nanoscale. The scale invariance of the theory of linear elasticity and the conventional plasticity theories has lead to the formulation of the strain-gradient plasticity theory (Fleck et al. 1994; Nix & Gao 1998; Gao et al. 1999; Huang et al. 2000; Hutchinson 2000). The theory, developed for microscale deformation, predicts a dependence of mechanical properties on the strain gradient, which is scale dependent. Recently, the strain-gradient plasticity theory has been used for modelling the scale effects in friction and wear (Bhushan & Nosonovsky 2003, 2004a,b; Nosonovsky & Bhushan 2005).
These studies indicate that micromechanical devices may behave in ways that cannot be predicted from their larger counterparts. It is encouraging in this regard to find that materials' properties at small scales can be superior. Nanoscale friction and wear mapping can help identify some sweet spots that would give ultralow friction and near-zero wear. Mapping nanoscale friction and wear as a function of operating conditions and interface properties is a valuable tool and has the potential to impact the very way in which we design and select materials for nanotechnology applications.
One contribution of 8 to a Theme Issue ‘Nanotribology, nanomechanics and applications to nanotechnology I’.
↵NanoScript is a trademark of Veeco Metrology.
- © 2007 The Royal Society