Royal Society Publishing

Nanotribology and nanomechanics in nano/biotechnology

Bharat Bhushan


Owing to larger surface area in micro/nanoelectromechanical systems (MEMS/NEMS), surface forces such as adhesion, friction, and meniscus and viscous drag forces become large when compared with inertial and electromagnetic forces. There is a need to develop lubricants and identify lubrication methods that are suitable for MEMS/NEMS. For BioMEMS/BioNEMS, adhesion between biological molecular layers and the substrate, and friction and wear of biological layers may be important, and methods to enhance adhesion between biomolecules and the device surface need to be developed. There is a need for development of a fundamental understanding of adhesion, friction/stiction, wear, the role of surface contamination and environment, and lubrication. MEMS/NEMS materials need to exhibit good mechanical and tribological properties on the micro/nanoscale. Most mechanical properties are known to be scale dependent. Therefore, the properties of nanoscale structures need to be measured. Component-level studies are required to provide a better understanding of the tribological phenomena occurring in MEMS/NEMS. The emergence of micro/nanotribology and atomic force microscopy-based techniques has provided researchers with a viable approach to address these problems. This paper presents an overview of micro/nanoscale adhesion, friction, and wear studies of materials and lubrication studies for MEMS/NEMS and BioMEMS/BioNEMS. It also presents a review of scale-dependent mechanical properties, and stress and deformation analysis of nanostructures.


1. Introduction

Microelectromechanical systems (MEMS) refer to microscopic devices that have a characteristic length of less than 1 mm but more than 100 nm and combine electrical and mechanical components. Nanoelectromechanical systems (NEMS) refer to nanoscopic devices that have a characteristic length of less than 100 nm and combine electrical and mechanical components. In mesoscale devices, if the functional components are on micro- or nanoscale, they may be referred to as MEMS or NEMS, respectively. These are referred to as an intelligent miniaturized system comprising sensing, processing and/or actuating functions, and combine electrical and mechanical components. The acronym MEMS originated in the USA. The term commonly used in Europe is microsystem technology (MST) and in Japan is micromachines. Another term generally used is micro/nanodevices. MEMS/NEMS terms are also now used in a broad sense and include electrical, mechanical, fluidic, optical and/or biological functions. MEMS/NEMS for optical applications are referred to as micro/nanooptoelectromechanical systems (MOEMS/NOEMS). MEMS/NEMS for electronic applications are referred to as radiofrequency MEMS/NEMS or RF-MEMS/RF-NEMS. MEMS/NEMS for biological applications are referred to as BioMEMS/BioNEMS. These devices are produced by either top–down (lithographic, non-lithographic or micromachining) or bottom–up (nanochemistry) processes.

To put the dimensions of MEMS and NEMS in perspective, see table 1. Individual atoms are typically a fraction of one nanometre in diameter, DNA molecules are approximately 2.5 nm wide, biological cells are in the range of thousands of nanometres in diameter and human hair is approximately 75 μm in diameter. Three NEMS reported in the table range in size from 10 to 300 nm and one MEMS reported is 12 000 nm. The weight of a micromachined silicon structure can be as low as 1 nN, and NEMS can be built with weight as low as 10−20 N with cross sections of approximately 10 nm. In comparison, the weight of a drop of water is approximately 10 μN and that of an eyelash is approximately 100 nN.

View this table:
Table 1

Dimensions and masses in perspective.

A broad class of MEMS/NEMS are used in electromechanical, electronics, information/communication, chemical and biological applications (Bhushan 2007a). MEMS for mechanical applications include acceleration, pressure, flow and gas sensors, linear and rotary actuators, and other microstructures or microcomponents such as electric motors, gear chains, gas turbine engines, fluid pumps, fluid valves, switches, grippers and tweezers. MEMS for chemical applications include chemical sensors and various analytical instruments. MOEMS are the devices that include optical components, such as micromirror arrays for displays, infrared image sensors, spectrometers, bar code readers and optical switches. RF-MEMS include inductors, capacitors, antennas and RF switches. High aspect ratio MEMS (HARMEMS) have also been introduced. Examples of NEMS include microcantilevers with integrated sharp nanotips for scanning tunnelling microscopy (STM) and atomic force microscopy (AFM), quantum corral are formed using STM by placing atoms one by one, AFM cantilever array (Millipede) for data storage, STM and AFM tips for nanolithography, dip pen nanolithography for printing molecules, nanowires, carbon nanotubes, quantum wires (QWRs), quantum boxes (QBs), quantum transistors, nanotube-based sensors, biological (DNA) motors, molecular gears by attaching benzene molecules to the outer walls of carbon nanotubes, devices incorporating nanometre-thick films (e.g. in giant magnetoresistive or GMR read/write magnetic heads and magnetic media for magnetic rigid disk and magnetic tape drives), nanopatterned magnetic rigid disks and nanoparticles (e.g. nanoparticles in magnetic tape substrates and nanomagnetic particles in magnetic tape coatings).

BioMEMS/BioNEMS are increasingly used in commercial and defence applications (Bhushan 2007a). They are used for chemical and biochemical analyses (biosensors) in medical diagnostics (e.g. DNA, RNA, proteins, cells, blood pressure and assays and toxin identification), tissue engineering and implantable pharmaceutical drug delivery. Biosensors, also referred to as biochips, deal with liquids and gases. There are two types of biosensors. A large variety of biosensors are based on micro/nanofluidics. Micro/nanofluidic devices offer the ability to work with smaller reagent volumes and shorter reaction times, and perform analyses of multiple types at once. The second type of biosensors includes micro/nanoarrays that perform one type of analysis thousands of times.

(a) Examples of devices with tribological issues in MEMS/NEMS and BioMEMS/BioNEMS

Tribological issues are important in MEMS/NEMS and BioMEMS/BioNEMS requiring intended and/or unintended relative motion. In these devices, various forces associated with the device scale down with the size. When the length of the machine decreases from 1 mm to 1 μm, the surface area decreases by a factor of a million and the volume decreases by a factor of a billion. As a result, surface forces such as adhesion, friction, meniscus forces, viscous drag forces and surface tension that are proportional to surface area become a thousand times larger than the forces proportional to the volume, such as inertial and electromagnetic forces. In addition to the consequences of the large surface-to-volume ratios, the small tolerances that these devices are designed for make physical contacts more likely, thereby making them particularly vulnerable to adhesion between adjacent components. Slight particulate or chemical contamination present at the interface can be detrimental. Furthermore, the small start-up forces and the torques available to overcome retarding forces are small, and the increase in resistive forces such as adhesion and friction becomes a serious tribological concern, which limits the durability and reliability of MEMS/NEMS (Bhushan 1998). A large lateral force required to initiate relative motion between two surfaces, large static friction, is referred to as ‘stiction’, which has been studied extensively in tribology of magnetic storage systems (Bhushan 1996, 1999a,b, 2001, 2002, 2003, 2005a,b, 2007b). The source of stiction is generally liquid-mediated adhesion with the source of liquid being process fluid or capillary condensation of the water vapour from the environment. Adhesion, friction/stiction (static friction), wear and surface contamination affect MEMS/NEMS and BioMEMS/BioNEMS performances and in some cases can even prevent devices from working (Bhushan 1998, 1999a, 2005a, 2007a). Some examples of devices which experience tribological problems follow.

Figure 1a shows an integrated capacitive-type silicon accelerometer fabricated using surface micromachining, a couple of millimetres in dimension, which is used for deployment of airbags in vehicles, and more recently for various other consumer electronics market (Core et al. 1993; Sulouff 1998). The central beam has interdigitated, cantilevered electrode fingers on all four sides that alternate with those of the stationary electrode fingers as shown. Lateral motion of the central beam during use may result in stiction between the adjacent electrodes as well as stiction of the beam structure with the underlying substrate which is detrimental to the operation of the sensor (Core et al. 1993; Sulouff 1998). Wear during unintended contacts of these polysilicon fingers is also a problem. A molecularly thick diphenyl siloxane lubricant film, resistant to high temperatures and oxidation, is applied by a vapour deposition process on the electrodes to reduce stiction and wear (Martin & Zhao 1997).

Figure 1

Examples of (a) an accelerometer MEMS device (Sulouff 1998) and (b) a MOEMS device having commercial use that experiences tribological problems (Hornbeck 1999).

Figure 1b shows two digital micromirror device (DMD) pixels used in digital light processing (DLP) technology for digital projection displays in computer projectors, high-definition television (HDTV) sets and movie projectors (Hornbeck & Nelson 1988; Hornbeck 1999, 2001). The entire array (chip set) consists of half a million to more than two million oscillating aluminium alloy micromirrors as in digital light switches, each approximately 12 μm2 and with 13 μm pitch. For binary operation, the micromirror/yoke structure, mounted on torsional hinges, is oscillated to ±10° (with respect to the plane of the chipset) and is limited by a mechanical stop. Contact between cantilevered spring tips at the end of the yoke (four present on each yoke) with the underlying stationary landing sites results in stiction and wear during contact between aluminium alloy spring tips and landing sites (Henck 1997; Douglass 1998, 2003). Vapour phase-deposited self-assembled monolayer (SAM) of a fatty acid—perfluorodecanoic acid (PFDA)—on the surfaces of the tip and landing sites is used to reduce stiction and wear (Hornbeck 1997; Robbins & Jacobs 2001). To keep moisture out and create a background pressure of PFDA, hermetic packaging is used.

An example of a microarray-type biosensor (under development in our laboratory) is based on a field-effect transistor (FET) and is shown in figure 2a (Bhushan et al. 2005a; Lee et al. 2005a). In this sensor, the gate metal electrode of a metal oxide semiconductor field effect transistor (MOSFET) is removed and replaced with a protein (receptor layer) whose cognate is the analyte (e.g. virus or bacteria) that is meant to be sensed. The binding of the receptor layer with the analyte produces a change in the effective charge which creates a change in the electrical field. This electrical field change may produce a measurable change in the current flow through the device. Adhesion between the protein layer and silica substrate affects reliability of the biosensor. In the case of implanted biosensors, the biosensors come in contact with the exterior environment such as tissues and fluids, and any relative motion of the sensor surface with respect to the exterior environment of tissues or fluids may result in surface damage (figure 2b; Bhushan et al. 2006a).

Figure 2

(a) Schematic of MOSFET-based bioFET sensor (Bhushan et al. 2005a) and (b) schematic showing the generation of friction and wear points due to the interaction of an implanted biomolecule layer on a biosensor with living tissue (Bhushan et al. 2006a).

Polymer BioMEMS are designed to measure cellular surfaces. For two examples, see figure 3 (Wei et al. 2005). The device on the left shows cantilevers anchored at the periphery of the circular structure, while that on the right has cantilevers anchored at the two corners on the top and the bottom. The cell adheres to the centre of the structure, and the contractile forces generated in the cell's cytoskeleton cause the cantilever to deflect. The deflection of the compliant polymer cantilevers is measured optically and related to the magnitude of the forces generated by the cell. Adhesion between the cells and the polymer beam is desirable. In order to design the sensors, micro- and nanoscale mechanical properties of polymer structures are needed.

Figure 3

(a,b) Schematic of two designs for polymer bioMEMS structures to measure cellular forces (Wei et al. 2005).

Various MEMS/NEMS are designed to perform expected functions in the millisecond to picosecond range. The expected life of the devices for high-speed contacts can vary from a few hundred thousand to many billions of cycles, e.g. over a hundred billion cycles for DMDs, which puts stringent requirements on materials. Adhesion between a biological molecular layer and the substrate, referred to as ‘bioadhesion’, reduction of friction and wear of biological layers, biocompatibility and biofouling for BioMEMS/BioNEMS are important. There is a need for the development of a fundamental understanding of adhesion, friction/stiction, wear, the role of surface contamination and environment (Bhushan 1998). MEMS/NEMS materials need to possess good mechanical and tribological properties on the micro/nanoscale (Bhushan 1998, 1999a, 2001, 2005a, 2007b). There is a need to develop lubricants and identify lubrication methods that are suitable for MEMS/NEMS. Methods need to be developed to enhance adhesion between biomolecules and the device substrate, referred to as bioadhesion. Component-level studies are required to provide a better understanding of the tribological phenomena occurring in MEMS/NEMS. The emergence of micro/nanotribology and AFM-based techniques has provided researchers with a viable approach to address these problems (Bhushan et al. 1995a; Bhushan 1999a, 2005a,b).

(b) Scale dependence of mechanical properties

Most mechanical properties are scale dependent (Bhushan 1999a,b, 2002, 2005a). Few researchers have measured mechanical properties of nanoscale structures using AFM (Namazu et al. 2000; Sundararajan & Bhushan 2002) and nanoindentation (Li & Bhushan 2003; Li et al. 2003; Wei et al. 2005; Palacio et al. 2007). Finite-element method (FEM) analysis of nanostructures has been performed to analyse the effect of types of surface roughness and scratches on stresses in nanostructures (Bhushan & Agrawal 2002, 2003).

(c) Objective of the paper

This paper presents an overview of micro/nanoscale tribological studies of materials and lubrication studies for MEMS/NEMS, bioadhesion, friction and wear of BioMEMS/BioNEMS. It also presents a review of scale-dependent mechanical properties and stress and deformation analysis of nanostructures.

2. Tribological studies of silicon and related materials

Materials of most interest for planar fabrication processes using silicon as the structural material are undoped and boron-doped (p+-type) single-crystal silicon for bulk micromachining and phosphorus (n+-type) doped and undoped LPCVD polysilicon films for surface micromachining. Since silicon-based devices lack high-temperature capabilities with respect to both mechanical and electrical properties, SiC is being developed as a structural material for high-temperature microsensor and microactuator applications (Shor et al. 1993; Mehregany et al. 1998). SiC can also be desirable for high-frequency micromechanical resonators, in the GHz range, owing to its high modulus of elasticity to density ratio and consequently high resonance frequency.

As will be shown, bare silicon exhibits inadequate tribological performance and needs to be coated with a solid and/or liquid overcoat or be surface treated (e.g. oxidation and ion implantation commonly used in semiconductor manufacturing), which reduce friction and wear. SiC films exhibit good tribological performance. Friction and wear of single-crystal and polycrystalline silicon samples have been studied, and the effect of ion implantation with various doses of C+, B+, N2+ and Ar+ ion species at 200 keV energy to improve their friction and wear properties has been studied (Gupta et al. 1993, 1994; Gupta & Bhushan 1994). The coefficient of macroscale friction and wear factor of virgin single-crystal silicon and C+-implanted silicon samples as a function of ion dose are presented in figure 4 (Gupta et al. 1993). The coefficient of friction and the wear factor for bare silicon are very high and decrease drastically with ion dose. Silicon samples bombarded above the ion dose of 1017 C+ cm−2 exhibit extremely low values of coefficients of friction (typically 0.03–0.06 in air) and the wear factor reduced by as much as four orders of magnitude. Gupta et al. (1993) reported that a decrease in the coefficient of friction and wear factor of silicon as a result of C+ ion bombardment occurred owing to the formation of silicon carbide rather than the amorphization of silicon. Gupta et al. (1994) also reported an improvement in friction and wear with B+ ion implantation.

Figure 4

Influence of ion doses on the (a) coefficient of friction and (b) wear factor on C+ ion-bombarded single-crystal and polycrystalline silicon slid against an alumina ball. V corresponds to the virgin single-crystal silicon, while S and P denote tests that correspond to doped single and polycrystalline silicon, respectively. Each datum bar represents the average value of four to six measurements (Gupta et al. 1993).

Microscale friction, scratching and wear measurements were performed using an atomic force/friction force microscope (AFM/FFM) on single-crystal silicon, polysilicon films and single-crystal silicon with ion implantation and oxide coatings deposited by thermal oxidation in dry and wet environments and by plasma-enhanced chemical vapour deposition (PECVD; Bhushan 1999a, 2005a,b). Bhushan & Koinkar (1994) reported that there is a decrease in the coefficients of micro- and macroscale friction values as a result of ion implantation and silicon dioxide coatings. PECVD-oxide samples showed superior scratch (figure 5) and wear resistance followed by dry-oxidized, wet-oxidized and ion-implanted samples.

Figure 5

Scratch depth as a function of normal load after 10 cycles for various silicon samples, virgin, treated and coated (Bhushan & Koinkar 1994). Open circle, undoped Si(100); closed circle, PECVD-oxide Si(111); open triangle, dry-oxidized Si(111); closed triangle, wet-oxidized Si(111); open square, C+-implanted Si(111).

Studies have also been conducted on undoped polysilicon film, heavily doped (n+-type) polysilicon film, heavily doped (p+-type) single-crystal Si(100) and 3C-SiC (cubic or β-SiC) film (Bhushan et al. 1998; Sundararajan & Bhushan 1998; Li & Bhushan 1999). Figure 6 shows AFM three-dimensional maps and averaged two-dimensional profiles of the scratch marks on the various samples. It is observed that scratch depth increases almost linearly with the normal load. Si(100) and the doped and undoped polysilicon film show similar scratch resistance. From the data, it is clear that the SiC film is much more scratch resistant than the other samples. Wear experiments showed the same trends. These studies indicate that SiC film exhibits desirable tribological properties for use in MEMS devices.

Figure 6

AFM three-dimensional maps and averaged two-dimensional profiles of scratch marks on various samples (Bhushan et al. 1998). (a) Undoped Si(100), (b) undoped polysilicon film, (c) n+-type polysilicon film and (d) SiC film.

3. Lubrication studies for MEMS/NEMS

Several studies of liquid perfluoropolyether (PFPE) lubricant films and SAMs have been carried out for the purpose of minimizing adhesion, friction and wear (Bhushan et al. 1995b, 2005b, 2006b,c; Koinkar & Bhushan 1996a,b; Henck 1997; Bhushan 1999a,c, 2005a,b; Bhushan & Liu 2001; Liu & Bhushan 2002, 2003; Kasai et al. 2005; Lee et al. 2005b; Tambe & Bhushan 2005; Tao & Bhushan 2005a,b). Many variations of these films are hydrophobic (low surface tension and high contact angle) and have low shear strength which provide low adhesion, friction and wear.

The classical approach to lubrication uses freely supported multimolecular layers of liquid lubricants (Bhushan 1999a,b, 2002, 2005a). The liquid lubricants are sometimes chemically bonded to improve their wear resistance. Partially chemically bonded, molecularly thick PFPE lubricants are widely used for lubrication of magnetic storage media, owing to their thermal stability and extremely low vapour pressure (Bhushan 1996), and are found to be suitable for MEMS/NEMS devices. Adhesion, friction and durability experiments have been performed on virgin Si(100) and silicon surfaces lubricated with two commonly used PFPE lubricants—Z-15 (with –CF3 non-polar end groups) and Z-DOL (with –OH polar end groups; Koinkar & Bhushan 1996a,b; Bhushan 1999a, 2005a; Liu & Bhushan 2003; Tao & Bhushan 2005a; Bhushan et al. 2006b,c). The adhesive forces of Si(100), Z-15 and Z-DOL (BW) measured by two methods are summarized in figure 7. The figure shows that the presence of mobile Z-15 lubricant film increases the adhesive force when compared with that of Si(100) by meniscus formation (Bhushan 1999b, 2002), whereas the presence of solid-phase Z-DOL (BW) film reduces the adhesive force when compared with that of Si(100) owing to the absence of mobile liquid. Figure 7b shows the relative size and sources of meniscus. It is well known that the native oxide layer (SiO2) on the top of the Si(100) wafer exhibits hydrophilic properties, and some water molecules can be adsorbed on this surface. The condensed water will form a meniscus as the tip approaches the sample surface. The larger adhesive force in Z-15 is not just caused by the Z-15 meniscus, and the non-polarized Z-15 liquid does not have good wettability and strong bonding with Si(100). Consequently, in the ambient environment, the condensed water molecules from the environment permeate through the liquid Z-15 lubricant film and compete with the lubricant molecules presented on the substrate. The interaction of the liquid lubricant with the substrate is weakened, and a boundary layer of the liquid lubricant forms puddles (Koinkar & Bhushan 1996a,b). This dewetting allows water molecules to be adsorbed on the Si(100) surface as aggregates along with Z-15 molecules. And both of them can form a meniscus while the tip approaches the surface. Thus, the dewetting of liquid Z-15 film results in higher adhesive force and poorer lubrication performance.

Figure 7

(a) Summary of the adhesive forces of Si(100) and Z-15 (2.8 nm film thickness) and Z-DOL (bonded washed, BW; 2.3 nm film thickness) films measured by force calibration plots and friction force versus normal load plots in ambient air. Open bars, force calibration plot; slashed bars, friction force plot. (b) The schematic showing the effect of meniscus, formed between the AFM tip and the surface sample, on the adhesive and friction forces (Liu & Bhushan 2003).

For lubrication of MEMS/NEMS, another effective approach involves the deposition of organized and dense molecular layers of long-chain molecules, referred to as SAMs, by chemical grafting of molecules (Bhushan et al. 1995b, 2005b, 2006b,c; Bhushan & Liu 2001; Liu et al. 2001; Liu & Bhushan 2002; Kasai et al. 2005; Lee et al. 2005b; Tambe & Bhushan 2005; Tao & Bhushan 2005b). Bhushan et al. (2005b), Kasai et al. (2005), Tambe & Bhushan (2005) and Bhushan et al. (2006b) studied perfluorodecyltrichlorosilane (PFTS), n-octyldimethyl (dimethylamino) silane (ODMS; n=7) and n-octadecylmethyl (dimethylamino) silane (ODDMS) (n=17) vapour phase deposited on Si substrate, and decylphosphonate (DP), octadecylphosphonate (ODP) and perfluorodecylphosphonate (PFDP) on Al substrate (figure 8). Figure 9a presents the contact angle, adhesive force, friction force and coefficient of friction of two substrates and with various SAMs. Based on the data, PFTS/Si exhibits a higher contact angle and lower adhesive force when compared with that of ODMS/Si and ODDMS/Si. Data of various films on Si substrate are similar to those of the corresponding various films on Al substrate. Thus substrate had little effect. The coefficient of friction of various SAMs was comparable. For wear performance studies, experiments were conducted on various films. Figure 9b shows the relationship between the decrease in surface height and the normal load for various SAMs and corresponding substrates (Kasai et al. 2005; Tambe & Bhushan 2005; Bhushan et al. 2006b). As shown in the figure, the SAMs exhibit a critical normal load beyond which point the surface height drastically decreases. Unlike SAMs, the substrates show a monotonic decrease in surface height with increasing normal load with wear initiating from the very beginning, i.e. even for low normal loads. The critical loads corresponding to the sudden failure are shown in figure 9c. Out of the two alkyl SAMs, ODDMS/Si and ODP/Al show a better wear resistance than ODMS/Si and DP/Al due to the chain length effect. Wear behaviour of the SAMs is reported to be mostly determined by the molecule–substrate bond strengths.

Figure 8

Schematics of structures of perfluoroalkylsilane and alkylsilane SAMs on Si with native oxide substrates, and alkylphosphonate and perfluoroalkylphosphonate SAMs on Al with native oxide.

Figure 9

(a(i)) Contact angle, (ii) adhesive force, (iii) friction force and (iv) coefficient of friction of Si with native oxide and Al with native oxide substrates, (b) decrease in surface height as a function of normal load after one scan cycle and (c) comparison of critical loads for failure during wear tests for various SAMs on Si and Al substrates (Kasai et al. 2005; Tambe & Bhushan 2005; Bhushan et al. 2006b).

4. Tribological studies of biological molecules on silicon-based and polymer surfaces

Proteins on silicon-based surfaces are of extreme importance in various applications including silicon microimplants, various bioMEMS, such as biosensors, and therapeutics. Silicon is a commonly used substrate in microimplants, but it can have undesired interactions with the human immune system. Therefore, to mimic a biological surface, protein coatings are used on silicon-based surfaces as passivation layers, so that these implants are compatible in the body and avoid rejection. Whether this surface treatment is applied to a large implant or to a bioMEMS, the function of the protein passivation is obtained from the nanoscale three-dimensional structural conformation of the protein. Proteins are also used in bioMEMS owing to their function specificity. For biosensor applications, the extensive array of protein activities provides a rich supply of operations that may be performed at the nanoscale. Many antibodies (proteins) have an affinity to specific protein antigens. For example, pathogens (disease-causing agents, e.g. virus or bacteria) trigger production of antigens which can be detected when bound to a specific antibody on the biosensor. The specific binding behaviour of proteins, which has been applied to the laboratory assays, may also be redesigned for in vivo use as sensing elements of a bioMEMS. The epitope-specific binding properties of proteins to various antigens are useful in therapeutics. Adhesion between the protein and substrate affects the reliability of an application. Among other things, morphology of the substrate affects the adhesion. Furthermore, for in vivo environments, the proteins on the biosensor surface should exhibit high wear resistance during the direct contact with the tissue and circulatory blood flow without washing off.

Bhushan et al. (2005a) studied the step-by-step morphological changes and the adhesion of a model protein—streptavidin (STA)—on silicon-based surfaces. In addition to physical adsorption, they also used nanopatterning and chemical linker methods to improve adhesion. A nanopatterned surface contains large edge surface area leading to high surface energy, which results in high adhesion. In the chemical linker method, sulpho-NHS-biotin was used as a cross linker because the bonds between the STA and the biotin molecule are one of the strongest non-covalent bonds known. It was connected to the silica surface through a silane linker, 3-aminopropyltriethoxysilane (3-APTES). In order to make a bond between the silane linker and the silica surface, the silica surface was hydroxylated. Bovine serum albumin (BSA) was used before STA in order to block non-specific binding sites of the STA protein with silica surface. Figure 10 shows the adhesion values of various surfaces. The adhesion value between biotin and STA was higher than that for other samples, which is expected. Edges of patterned silica also exhibited high adhesion values. It appears that both nanopatterned surfaces and the chemical linker method increase adhesion with STA.

Figure 10

Adhesion measurements of silica, patterned silicon, silica boiled in deionized water (DI) water and sulpho-NHS-biotin using functionalized (with streptavidin) tips obtained from force–distance curves, captured in phosphate-buffered saline (PBS) (Bhushan et al. 2005a).

Bhushan et al. (2006a) studied friction and wear of STA deposited by physical adsorption and the chemical linker method. Figure 11 shows the coefficient of friction between the Si3N4 tip and various samples. The coefficient of friction is less for STA-coated silica samples than for uncoated samples. The STA coating acts as a lubricant film. The coefficient of friction is found to be dependent upon the concentration of STA, and it decreases with an increase in the concentration. Bhushan et al. (2005a) have reported that the density and distribution of the biomolecules vary with the concentration. At higher concentration of the solution, the coated layer is more uniform and the silica substrate surface is more highly covered with biomolecules than at lower concentration. This means that the surface forms a continuous lubricant film at higher concentration.

Figure 11

Coefficient of friction for various surfaces with and without biomolecules (Bhushan et al. 2006a).

In the case of samples prepared by the chemical linker method, the coefficient of friction increases with an increase in the biomolecular chain length due to increased compliance. When normal load is applied on the surface, the surface gets compressed resulting in a larger contact area between the AFM tip and the biomolecules. Besides that, the size of STA is much larger than that of APTES and biotin. This results in a tightly packed surface with the biomolecules, which results in very little lateral deflection of the linker in the case of STA-coated biotin. Owing to this high contact area and low lateral deflection, the friction force increases for the same applied normal load compared with the directly adsorbed surface. These tests reveal that the surfaces coated with biomolecules reduce the friction, but if the biomolecular coating of the surface is too thick or the surface has some cushioning effect as seen in the chemical linker method it would increase the coefficient of friction.

Figure 12 shows the wear maps of STA deposited by physical adsorption at three normal loads. The wear depth increases with the increasing normal load. An increase in normal load causes partial damage to the folding structure of the STA molecules. It is unlikely that the chemical (covalent) bonds within the STA molecule are broken; instead, the folding structure is damaged leading to wear marks. When the load is high such as 30% of free amplitude (approx. 8 nN), the molecules may have been removed by the AFM tip due to indentation effect. Owing to this, there is a significant increase in the wear depth from 50% of free amplitude (approx. 6 nN) to 30% of free amplitude (approx. 8 nN). The data show that biomolecules will get damaged during sliding.

Figure 12

Wear maps and cross-sectional profiles of pre-cycle cleaned silica coated with streptavidin by physical adsorption at three normal loads (increasing from (a) 75% of free amplitude, (b) 50% of free amplitude and (c) 30% of free amplitude; Bhushan et al. 2006a). Streptavidin absorbed on pre-cycle cleaned silica at 10 μg ml−1.

5. Nanopatterned surfaces

One of the crucial surface properties for various surfaces and interfaces in wet environments is non-wetting or hydrophobicity. Wetting is characterized by the contact angle, which is the angle between the solid and liquid surfaces. If the liquid wets the surface (referred to as wetting liquid or hydrophilic surface), the value of the contact angle is 0≤θ≤90°; whereas if the liquid does not wet the surface (referred to as non-wetting liquid or hydrophobic surface), the value of the contact angle is 90°<θ≤180°. A surface is considered superhydrophobic if θ is greater than 150°. These surfaces are water repellent. The surfaces with low contact angle hysteresis (difference between advancing and receding contact angles) also have a self-cleaning effect called the ‘lotus effect’. Water droplets roll off the surface and take contaminants with them (Kijlstra et al. 2002; Extrand 2004; Jung & Bhushan 2006). They have low drag for fluid flow and a low tilt angle. The self-cleaning surfaces are of interest in various applications, including self-cleaning windows, windshields, exterior paints for buildings and navigation ships, utensils, roof tiles, textiles and reduction of drag in fluid flow, e.g. in micro/nanochannels. When two hydrophilic surfaces come into contact, condensation of water vapour from the environment forms meniscus bridges at asperity contacts which lead to an intrinsic attractive force (Adamson 1990; Israelachvili 1992; Bhushan 1999b, 2002, 2003, 2005a). This may lead to high adhesion and stiction. Therefore, hydrophobic surfaces are desirable. Hydrophobic surfaces can be constructed by using low surface energy material coatings such as polytetrafluoroethylene or wax, by increasing surface area by introducing surface roughness and/or the creation of air pockets. Air trapped in the cavities of a rough surface results in a composite solid–air–liquid interface, as opposed to the homogeneous solid–liquid interface (Wenzel 1936; Cassie & Baxter 1944; Nosonovsky & Bhushan 2005, 2006; Jung & Bhushan 2006).

Examples of such surfaces are found in nature, such as Nelumbo nucifera (lotus) and Colocasia esculenta (Neinhuis & Barthlott 1997; Wagner et al. 2003), which have high contact angles with water and show strong self-cleaning properties known as the lotus effect (figure 13; Bhushan & Jung 2006). Lotus is known to be self-cleaning to prevent pathogens from bounding to the leaf surface. Many pathogenic organisms—spores and conidia of most fungi—require water for germination and can infect leaves in the presence of water (Neinhuis & Barthlott 1997). Recent studies have been carried out to fully characterize the hydrophobic leaf surfaces at the micro- and nanoscale while separating out the effects of the micro- and the nanobumps, and the hydrophobic compounds, called waxes on the hydrophobicity (Burton & Bhushan 2006; Bhushan & Jung 2006). The wax is present in crystalline tubules, composed of a mixture of aliphatic compounds, principally nonacosanol and nonacosanediols (Koch et al. 2006). By learning from what is found in nature, one can create roughness on various materials and study their surface properties, leading to successful implementation in applications where water repellency, fluid flow and lower adhesion forces are important.

Figure 13

SEM micrographs of two hydrophobic leaves, (a) Nelumbo nucifera (lotus) and (b) Colocasia esculenta.

Consider a rough solid surface with a typical size of roughness details smaller than the size of the droplet. For a droplet in contact with a rough surface without air pockets, referred to as a homogeneous interface, the contact angle is given as (Wenzel 1936)Embedded Image(5.1)where θ is the contact angle for rough surface; θ0 is the contact angle for smooth surface; and Rf is a roughness factor defined as a ratio of the solid–liquid area to its projection on a flat plane. The dependence of the contact angle on the roughness factor is presented in figure 14a for different values of θ0 based on equation (5.1). The model predicts that a hydrophobic surface (θ0>90°) becomes more hydrophobic with an increase in Rf and a hydrophilic surface (θ0<90°) becomes more hydrophilic with an increase in Rf (Jung & Bhushan 2006).

Figure 14

(a) Contact angles for rough surfaces (θ) as a function of the roughness factor (Rf) for various contact angles of the smooth surface (θ0). (b(i)) Formation of a composite solid–liquid–air interface for a rough surface and (ii) fLA requirement for a hydrophilic surface to be hydrophobic as a function of the roughness factor (Rf) and θ0. (c) Tilted surface profile (the tilt angle is α) with a liquid droplet (if θH(=θadvθrec) is low, a droplet can move easily at a small tilt angle).

For a rough surface, a wetting liquid will be completely absorbed by the rough surface cavities while a non-wetting liquid may not penetrate into surface cavities, resulting in the formation of air pockets, leading to a composite solid–liquid–air interface as shown in figure 14b. Cassie & Baxter (1944) extended the Wenzel equation for the composite interface, which was originally developed for the homogeneous solid–liquid interface,Embedded Image(5.2)where fLA is the fractional flat geometrical area of the liquid–air interface under the droplet. This model shows that, for a hydrophilic surface, the contact angle on a smooth surface increases with an increase of fLA. When the roughness factor increases, the contact angle decreases but at a slower rate, due to formation of the composite interface. At a high value of fLA, the surface can become hydrophobic; however, the value required may be unachievable or formation of air pockets may become unstable. For the hydrophobic surface, the contact angle increases with an increase in fLA for both smooth and rough surfaces. Using equation (5.2), the fLA requirement for a hydrophilic surface to be hydrophobic can be found as (Jung & Bhushan 2006)Embedded Image(5.3)Figure 14b shows the value of the fLA requirement as a function of Rf for four surfaces with different contact angles, θ0. Hydrophobic surfaces can be achieved above certain fLA values as predicted by equation (5.3). The upper part of each contact angle line is the hydrophobic region. When Rf increases, the fLA requirement also increases.

Another important characteristic of a solid–liquid interface is the contact angle hysteresis (θH), which is the difference between the contact angle at the increased droplet volume (advancing contact angle, θadv) and the contact angle at the decreased droplet volume (receding contact angle, θrec) for a droplet on the solid surface. The contact angle hysteresis occurs due to surface roughness and heterogeneity. Low contact angle hysteresis results in a very low water roll-off angle, which denotes the angle to which a surface may be tilted for roll off of water drops (Kijlstra et al. 2002; Extrand 2004). Low water roll-off angle is important in liquid flow applications such as in micro/nanochannels and surfaces with self-cleaning ability.

There is no simple expression for the contact angle hysteresis as a function of roughness; however, certain conclusions about the relation of the contact angle hysteresis to roughness can be made. In the limiting case of very small solid–liquid fractional contact area under the droplet, when the contact angle is large and the contact angle hysteresis is small, equation (5.2) is reduced toEmbedded Image(5.4)

For the homogeneous interface, fLA=0, whereas for the composite interface fLA is not zero. It is observed from equation (5.4) that, for the homogeneous interface, increasing roughness (high Rf) leads to increasing the contact angle hysteresis, while, for the composite interface, an approach to unity of fLA provides both high contact angle and small contact angle hysteresis (Jung & Bhushan 2006; Bhushan et al. 2007). Therefore, the composite interface is desirable for superhydrophobicity.

Formation of a composite interface is also a multiscale phenomenon, which depends upon the relative sizes of the liquid droplet and roughness details. A stable composite interface is essential for the successful design of superhydrophobic surfaces. However, the composite interface is fragile and it may transform into the homogeneous interface. Nosonovsky & Bhushan (2007a) have studied destabilizing factors for the composite interface and found that the sign of the surface curvature is important, especially in the case of multiscale (hierarchical) roughness. A convex surface (with bumps) leads to a stable interface and high contact angle. Also, they have suggested that the effects of a droplet's weight and curvature are among the factors which affect the transition.

Jung & Bhushan (2008) developed the model to predict the transition from the Cassie & Baxter regime to the Wenzel regime based on the factors discussed above. First, they considered a small water droplet suspended on a superhydrophobic surface consisting of a regular array of circular pillars with diameter D, height H and pitch P. The local deformation for small droplets is governed by surface effects rather than gravity. The curvature of a droplet is governed by the Laplace equation that relates the pressure inside the droplet to its curvature (Adamson 1990). The curvature is the same at the top and at the bottom of the droplet (Lafuma & Quéré 2003; Nosonovsky & Bhushan 2007b). For the patterned surface considered here, the maximum droop of the droplet occurs in the centre of the square formed by the four pillars as shown in figure 15a. Therefore, the maximum droop of the droplet (δ) in the recessed region can be found in the middle of two pillars that are diagonally across as shown in figure 15a, which is Embedded Image. If the droop is much greater than the depth of the cavity,Embedded Image(5.5)then the droplet will just contact the bottom of the cavities between pillars, resulting in the transition from the Cassie & Baxter regime to the Wenzel regime. Furthermore, in the case of large distances between the pillars, the liquid–air interface can easily be destabilized due to dynamic effects, such as surface waves, which are formed at the liquid–air interface due to the gravitational or capillary force. This leads to the formation of the homogeneous solid–liquid interface.

Figure 15

(a) A small water droplet suspended on a superhydrophobic surface consisting of a regular array of circular pillars. The maximum droop of the droplet occurs in the centre of the square formed by four pillars. The maximum droop of the droplet (δ) can be found in the middle of two pillars which are diagonally across. (b) Static contact angle (dotted line, the transition criteria range obtained using the model; solid line, droplet with 1 mm radius) and (c) hysteresis and tilt angles as a function of geometric parameters for two series of the patterned surfaces with different pitch values for a droplet with 1 mm in radius (5 μl volume; open circle, hysteresis; open triangle, tilt angle). Data at zero pitch correspond to a flat sample (Bhushan & Jung 2007; Jung & Bhushan 2008).

To validate the model, contact angle measurements on micropatterned samples with a range of pitch values were made using droplets of 1 mm in radius (5 μl volume; Bhushan & Jung 2007; Jung & Bhushan 2008). The contact angles on the prepared surfaces are plotted as a function of pitch between the pillars in figure 15b. A dotted line represents the transition criteria range obtained using equation (5.5). The flat Si coated with PF3 showed the static contact angle of 109°. As the pitch increases up to 45 μm of the first set and 126 μm of the second set, the static contact angle first increases gradually from 152° to 170°. Then, the contact angle starts decreasing sharply. The initial increase with an increase of pitch has to do with more open air space present which increases the propensity of air pocket formation. As predicted from the transition criteria (equation (5.5)), the decrease in contact angle at higher pitch values results due to the transition from composite interface to solid–liquid interface.

Figure 15b shows hysteresis and tilt angle as a function of pitch between the pillars (Bhushan & Jung 2007). The flat Si coated with PF3 showed a hysteresis angle of 34° and tilt angle of 37°. The patterned surfaces with low pitch increase the hysteresis and tilt angles compared with the flat surface due to the effect of sharp edges on the pillars, resulting in pinning (Nosonovsky & Bhushan 2005). For a droplet moving down on the inclined patterned surfaces, the line of contact of the solid, liquid and air will be pinned at the edge point until it will be able to move, resulting in increased hysteresis and tilt angles. For various pitch values, hysteresis and tilt angles show the same trends with varying pitch between the pillars. After an initial increase as discussed above, they gradually decrease with increasing pitch (due to the reduced number of sharp edges) and show an abrupt minimum in the value which has the highest contact angle. The lowest hysteresis and tilt angles are 5° and 3°, respectively, which were observed on the patterned Si with 45 μm of the first set and 126 μm of the second set. As discussed earlier, an increase in the pitch value allows the formation of the composite interface. At higher pitch values, it is difficult to form the composite interface. The decrease in hysteresis and tilt angles occurs due to the formation of the composite interface at pitch values ranging from 7 to 45 μm in the first set and from 21 to 126 μm in the second set. The hysteresis and tilt angles start to increase again due to the lack of formation of air pockets at pitch values ranging from 60 to 75 μm in the first set and from 168 to 210 μm in the second set. These results suggest that the air pocket formation and the reduction of pinning in the patterned surface play an important role for a surface with both low hysteresis and tilt angle.

6. Component-level studies

In MEMS devices involving parts in relative motion to each other, such as micromotors, large friction forces become the limiting factor to the successful operation and reliability of the device. It is generally known that most micromotors cannot be rotated as manufactured and require some form of lubrication. Continuous physical contact occurs during rotor movement (rotation) in the micromotors between the rotor and lower hub flange. In addition, contact occurs at other locations between the rotor and the hub surfaces and between the rotor and the stator. Friction forces will be present at these contact regions during motor operation. It is critical to determine the friction forces present in such MEMS devices. A novel technique to measure the static friction force (stiction) encountered in surface micromachined polysilicon micromotors using an AFM has been developed by Sundararajan & Bhushan (2001). An AFM tip was pressed against a rotor arm in a direction perpendicular to the long axis of the cantilever beam (the rotor arm edge closest to the tip is parallel to the long axis of the cantilever beam) and the static friction force was measured.

In one of the lubrication studies, micromotors were lubricated with two types of lubricants. Figure 16 summarizes static friction force data for two motors, M1 and M2, along with schematics of the meniscus effects for the unlubricated and lubricated surfaces. Capillary condensation of water vapour from the environment results in the formation of meniscus bridges between contacting and near-contacting asperities of two surfaces in close proximity to each other as shown in the figure. For unlubricated surfaces, more menisci are formed at higher humidity resulting in higher friction force between the surfaces. The formation of meniscus bridges is supported by the fact that the static friction force for unlubricated motors increases at high humidity (Sundararajan & Bhushan 2001). Solid bridging may occur near the rotor–hub interface due to silica residues after the first etching process. In addition, the drying process after the final etch can result in liquid bridging formed by the drying liquid due to meniscus force at these areas (Mastrangelo & Hsu 1993; Maboudian & Howe 1997; Bhushan 1999b, 2002, 2003). Therefore, the initial static friction force will be quite high as evidenced by the solid data points in figure 16. Once the first movement of the rotor permanently breaks these solid and liquid bridges, the static friction force of the motors will drop (as seen in figure 16) to a value dictated predominantly by the adhesive energies of the rotor and hub surfaces, the real area of contact between these surfaces and meniscus forces due to water vapour in the air, at which point the effect of lubricant films can be observed. Lubrication with a mobile layer, even a thin one, results in very high static friction forces due to meniscus effects of the lubricant liquid itself at and near the contact regions. It should be noted that a motor submerged in a liquid lubricant would result in a fully flooded lubrication regime. In this case, there is no meniscus contribution and only the viscous contribution to the friction forces would be relevant. However, submerging the device in a lubricant may not be a practical method. A solid-like hydrophobic lubricant layer (such as bonded Z-DOL) results in favourable friction characteristics of the motor. The hydrophobic nature of the lubricant inhibits meniscus formation between the contact surfaces and maintains low friction even at high humidity (Sundararajan & Bhushan 2002). This suggests that solid-like hydrophobic lubricants are ideal for lubrication of MEMS while mobile lubricants result in increased values of static friction force.

Figure 16

Summary of the effect of liquid and solid lubricants on the static friction force of micromotors. Despite the hydrophobicity of the lubricant used (Z-DOL), a mobile liquid lubricant (Z-DOL as is) leads to a very high static friction force due to increased meniscus forces, whereas a solid-like lubricant (bonded Z-DOL, BW) appears to provide some amount of reduction in the static friction force (Sundararajan & Bhushan 2001). Open circle, unlubricated; open triangle, Z-DOL, 2 nm (as is); open square, Z-DOL, 1 nm (BW).

7. Bending tests of nanostructures using an AFM

Quasi-static bending tests of fixed nanobeam arrays have been carried out using an AFM (Sundararajan & Bhushan 2002; Sundararajan et al. 2002). The wafer with nanobeam array with trapezoidal cross section is fixed onto a flat sample chuck using double-sided tape (Sundararajan & Bhushan 2002). For the bending test, the tip is brought over the nanobeam array with the help of the sample stage of the AFM (figure 17). Elastic modulus and bending strength (fracture stress) of the beams can be estimated from the load–displacement curves based on the assumption that the beams are made of an isotropic material and follow linear elastic theory.

Figure 17

Schematic showing the details of a nanoscale bending test using an AFM. The AFM tip is brought to the centre of the nanobeam and the piezo is extended over a known distance. By measuring the tip displacement, a load–displacement curve of the nanobeam can be obtained (Sundararajan & Bhushan 2002).

Fracture toughness is another important parameter for brittle materials such as silicon. In the case of the nanobeam arrays, these are not best suited for fracture toughness measurements because they do not possess regions of uniform stress during bending. Sundararajan & Bhushan (2002) developed a methodology. First, a crack of known geometry is introduced in the region of maximum tensile bending stress, i.e. on the top surface near the ends of the beam. This is achieved by generating a scratch at high normal load across the width of the beam using a sharp diamond tip (radius <100 nm). By bending the beam, a stress concentration will be formed under the scratch. This will lead to failure of the beam under the scratch once a critical load (fracture load) is attained. The fracture load and relevant dimensions of the scratch are input into the FEM model that is used to generate the fracture stress plots. If we assume that the scratch tip acts as a crack tip, a bending stress will tend to open the crack in mode I. In this case, the stress field around the crack tip can be described by the stress intensity parameter KI (for mode I) for linear elastic materials (Hertzberg 1989). In particular, the stresses corresponding to the bending stresses were used to calculate fracture toughness.

In addition to the properties mentioned so far that can be evaluated from quasi-static bending tests, the fatigue properties of nanostructures are also of interest. This is especially true for MEMS/NEMS involving vibrating structures such as oscillators and comb drives (Nguyen & Howe 1999) and hinges in DMDs (Hornbeck 1999). To study the fatigue properties of the nanobeams, Sundararajan & Bhushan (2002) applied monotonic cyclic stresses using an AFM.

Figure 18 shows typical load–displacement curves for Si and SiO2 beams that were bent to failure (Sundararajan & Bhushan 2002; Sundararajan et al. 2002). The upper width (w1) of the beams is indicated in the figure. Also indicated in the figure are the elastic modulus values obtained from the slope of the load–displacement curve. All the beams tested showed linear elastic behaviour followed by abrupt failure, which is suggestive of brittle fracture. Previously reported numbers of strengths range from 1 to 6 GPa for silicon (Johansson et al. 1988; Ericson & Schweitz 1990; Wilson & Beck 1996; Wilson et al. 1996; Sharpe et al. 1997; Sato et al. 1998; Tsuchiya et al. 1998; Greek et al. 1999; Mazza & Dual 1999; Yi et al. 2000) and approximately 1 GPa for SiO2 (Tsuchiya et al. 2000) microscale specimens. This clearly indicates that bending strength shows a specimen size dependence. Strength of brittle materials is dependent on pre-existing flaws in the material. Since for nanoscale specimens, the volume is smaller than for micro- and macroscale specimens, the flaw population will be smaller as well, resulting in higher values of strength. Fracture toughness is considered to be a material property and is believed to be independent of specimen size. The values obtained in this study, given its limitations, appear to show that fracture toughness is comparable, if not a little higher on the nanoscale.

Figure 18

Typical load–displacement curves of (a) silicon and (b) SiO2 nanobeams. The curves are linear until sudden failure, indicative of brittle fracture of the beams. The elastic modulus (E) values calculated from the curves are shown. The dimensions of the Si beam were w1=295 nm, w2=484 nm and t=255 nm, while those of the SiO2 beam were w1=250 nm, w2=560 nm and t=425 nm (Sundararajan et al. 2002).

Fatigue strength measurements of Si nanobeams have been carried out by Sundararajan & Bhushan (2002) using an AFM and various stress levels. Figure 19 shows a nanoscale SN curve, with bending stress (S) as a function of fatigue in cycles (N) with an apparent endurance life at lower stress. In general, the fatigue life decreased with increasing mean stress as well as increasing stress amplitude. When the stress amplitude was less than 15% of the bending strength, the fatigue life was greater than 30 000 cycles for both Si and SiO2. However, the mean stress had to be less than 30% of the bending strength for a life of greater than 30 000 cycles for Si, whereas, even at a mean stress of 43% of the bending strength, SiO2 beams showed a life greater than 30 000 cycles. This study clearly demonstrates that fatigue properties of nanoscale specimens can be studied.

Figure 19

Fatigue test data showing applied bending stress as a function of number of cycles for (a) Si and (b) SiO2. A single load–unload sequence is considered as 1 cycle. The bending strength data points are therefore associated with one-half cycle, since failure occurs upon loading (Sundararajan & Bhushan 2002).

Table 2 summarizes the various properties measured via quasi-static bending in this study (Sundararajan & Bhushan 2002). Also shown are bulk values of the parameters along with values reported on larger scale specimens by other researchers. Elastic modulus and fracture toughness values appear to be comparable to bulk values and show no dependence on specimen size. However, bending strength shows a clear specimen size dependence with nanoscale numbers being twice as large as numbers reported for larger scale specimens.

View this table:
Table 2

Summary of measured parameters from quasi-static bending tests.

8. Bending tests of polymeric microbeams using a nanoindenter

Hardness, elastic modulus, creep and scratch resistance as well as quasi-static bending tests of fixed and cantilevered polymeric microbeams have been carried out using a nanoindenter (Wei et al. 2005; Palacio et al. 2007). (For nanoindenter details, see Bhushan & Li (2003).) Figure 20 shows the SEM images of suspended microbeams and cantilever microbeams for beam bending experiments. Selected structural materials, of interest for BioMEMS/BioNEMS, were poly(propyl methacrylate) (PPMA), poly(methyl methacrylate) (PMMA), polystyrene (PS) and a polystyrene–nanoclay composite (PS/clay). Experiments were conducted on samples that were soaked in deionized (DI) water to assess the effect of the aqueous medium. In addition, experiments were carried out at human body temperature (37.5°C).

Figure 20

SEM images of polystyrene (a) double-clamped and (b) cantilever beams (Palacio et al. 2007).

Figure 21a is a summary of the effects of beam soaking and elevated temperature on the hardness (H) and elastic modulus (E) as obtained by nanoindentation on the supported part of the beam. Among the four materials studied, soaking had the greatest effect on PPMA. On the other hand, the H and E of the other materials (PMMA, PS, and PS/clay) were not significantly affected. The properties of the PPMA beam were adversely affected when the indentation temperature was increased to 37.5°C. This is because the glass transition temperature (Tg) of PPMA is within 35–43°C, and the glassy to rubbery transition is being observed here. On the other hand, no significant decrease in H and E was observed on the other three materials since their Tg values are much higher, such that the indent was performed well within the glassy regime.

Figure 21

(a) Effect of soaking and elevated temperatures on (i) hardness and (ii) elastic modulus and (b) effect of (i) soaking (filled circle, soaked; open square, unsoaked) and (ii) elevated temperatures (filled circle, 37.5°C; open square, 22°C) on the beam bending (Palacio et al. 2007).

In figure 21b, beam bending data are shown for each of the four supported beams investigated. The PPMA beam is affected by soaking in DI water, as shown by the appearance of a large amount of hysteresis in the load–displacement data. Human body temperature has an adverse effect on the stiffness of the PPMA beam. These are consistent with results from indentation. From a material selection standpoint, these are important findings as they imply that the performance of PPMA structures will be compromised if they were in a device subjected to either soaking in aqueous medium or above ambient temperature.

9. Finite-element analysis of nanostructures with roughness and scratches

Micro/nanostructures have some surface topography and local scratches dependent upon the manufacturing process. Surface roughness and local scratches may compromise the reliability of the devices and their effect needs to be studied. Finite-element modelling is used to perform parametric analysis to study the effect of surface roughness and scratches in different well-defined forms on tensile stresses which are responsible for crack propagation (Bhushan & Agrawal 2002, 2003). The analysis has been carried out on trapezoidal beams supported at the bottom whose data (on Si and SiO2 nanobeams) have been presented earlier.

The finite-element analysis was carried out by using the static analysis of ANSYS, which calculates the deflections and stresses produced by applied loading. The type of element selected for the study was SOLID95 type, which allows the use of different shapes without much loss of accuracy. The mesh is kept finer near the asperities and the scratches in order to take into account variation in the bending stresses. Based on bending experiments presented earlier, the beam materials can be assumed to be linearly elastic isotropic materials. A point load applied at the centre of the beam is simulated with the load being applied at three closely located central nodes on the beam used. It has been observed from the experimental results that the Si nanobeam breaks at approximately 80 μN. Therefore, in this analysis, a nominal load of 70 μN is selected. At this load, deformations are large and a large displacement option is used.

To study the effect of surface roughness and scratches on the maximum bending stresses, the following cases were studied. First, the semicircular and grooved asperities in the longitudinal direction with defined geometrical parameters are analysed (figure 22a). Next, semicircular asperities and scratches placed along the transverse direction separated by pitch p from each other are analysed (figure 22b). Lastly, the beam material is assumed to be either purely elastic, elastic-plastic or elastic-perfectly plastic. In the following, we begin with the stress distribution in smooth nanobeams followed by the effect of surface roughness in the longitudinal and transverse directions and scratches in the transverse direction.

Figure 22

(a) Plots showing the geometries of modelled roughness, (i) semicircular and (ii) grooved asperities along the nanobeam length with defined geometrical parameters. (b) Schematic showing (i) semicircular asperities and (ii) scratches in the transverse direction. The illustration of the mesh created on the beam with fine mesh near the asperities and the scratches.

The roughness in the form of semicircular and grooved asperities in the longitudinal direction on the maximum bending stresses is analysed (Bhushan & Agrawal 2003). The radius R and depth L are kept fixed at 25 nm, while the number of asperities is varied and their effect is observed on the maximum bending stresses. Figure 23a shows the variation of maximum bending stresses as a function of asperity shape and the number of asperities. The maximum bending stresses increase as the asperity number increases for both semicircular and grooved asperities. This can be attributed to the fact that, as asperity number increases, the moment of inertia decreases for that cross section. Also the distance from the neutral axis increases because the neutral axis shifts downwards. Both these factors lead to the increase in the maximum bending stresses and this effect is more pronounced in the case of semicircular asperity as it exhibits a higher value of maximum bending stress than that in grooved asperity.

Figure 23

(a) Effect of longitudinal semicircular and grooved asperities in different numbers on maximum bending stresses after loading trapezoidal Si nanobeams (w1=200 nm, w2=370 nm, t=255 nm, Embedded Image, E=169 GPa, ν=0.28, load=70 μN). (b) Effect of transverse semicircular asperities located at different pitch values on the maximum bending stresses after loading trapezoidal Si nanobeams (w1=200 nm, w2=370 nm, t=255 nm, Embedded Image, E=169 GPa, ν=0.28, load=70 μN). (c) Effect of the number of scratches along with the variation in the pitch on the maximum bending stresses after loading trapezoidal Si nanobeams (w1=200 nm, w2=370 nm, t=255 nm, Embedded Image, E=169 GPa, ν=0.28, load=70 μN; Bhushan & Agrawal 2003).

We analyse semicircular asperities when placed along the transverse direction followed by the effect of scratches on the maximum bending stresses in varying numbers and different pitch (Bhushan & Agrawal 2003). In the analysis of semicircular transverse asperities, three cases were considered which included a single asperity, asperities throughout the nanobeam surface separated by pitch equal to 50 nm and pitch equal to 100 nm. Figure 23b shows that the value of maximum tensile stress is 42 GPa, which is much larger than the maximum tensile stress value with no asperity of 16 GPa or when the semicircular asperity is present in the longitudinal direction. It is also observed that the maximum tensile stress does not vary with the number of asperities or the pitch, while the maximum compressive stress does increase dramatically for the asperities present throughout the beam surface from its value when a single asperity is present. Maximum tensile stress occurs at the ends and an increase in p does not add any asperities at the ends whereas asperities are added in the central region where compressive stresses are maximum. The semicircular asperities present at the centre cause the local perturbation in the stress distribution at the centre of the asperity where load is being applied leading to a high value of maximum compressive stress (Timoshenko & Goodier 1970).

In the study pertaining to scratches, the number of scratches is varied. Furthermore, the load is applied at the centre of the beam and at the centre of the scratch near the end as well for all the cases. In all of these cases, the pitch value was equal to 100 nm, and the L value was equal to 100 nm with the h value being 20 nm. Figure 23c shows that the value of maximum tensile stress remains almost the same with the number of scratches for both types of loading, i.e. when load is applied at the centre of the beam and at the centre of the scratch near the end. This is because the maximum tensile stress occurs at the beam ends, no matter where the load gets applied. But the presence of scratch does increase the maximum tensile stress when compared with its value for a smooth nanobeam, although the number of scratches no longer matters because the maximum tensile stress occurring at the nanobeam end is unaffected by the presence of more scratches beyond the first scratch in the direction towards the centre. The value of the tensile stress is much lower when the load is applied at the centre of the scratch and it can be explained as follows. The negative bending moment at the end near the load applied decreases with load offset after two-thirds of the length of the beam (Shigley & Mitchell 1993). Since this negative bending moment is responsible for tensile stresses, their behaviour with the load offset is the same as the negative bending moment. Also the value of maximum compressive stress when load is applied at the centre of the nanobeam remains almost the same as the centre geometry is unchanged due to the number of scratches and hence the maximum compressive stress occurring below the load at the centre being the same. On the other hand, when the load is applied at the centre of the scratch we observe that the maximum compressive stress increases dramatically because the local perturbation in the stress distribution at the centre of the scratch where the load is being applied leads to a high value of maximum compressive stress (Timoshenko & Goodier 1970). It increases further with the number of scratches and then levels off. This can be attributed to the fact that, when there is another scratch present close to the scratch near the end, the stress concentration is more as the effect of local perturbation in the stress distribution is more significant. However, this effect is insignificant when more than two scratches are present.

Now we address the effect of pitch on the maximum compressive stress when the load is applied at the centre of the scratch near the end. When the pitch is up to a value of 200 nm, the maximum compressive stress increases with the number of scratches as discussed earlier. On the other hand, when the pitch value goes beyond 225 nm, this effect is reversed. This is because the presence of another scratch no longer affects the local perturbation in the stress distribution at the scratch near the end. Instead more scratches at a fair distance distribute the maximum compressive stress at the scratch near the end and the stress starts going down. Such observations of maximum bending stresses can help in identifying the number of asperities and scratches allowed separated by an optimum distance from each other.

10. Closure

In MEMS/NEMS, the length-scale and large surface-to-volume ratio of the devices result in very high retarding forces such as adhesion and friction that seriously undermine the performance and reliability of the devices. These tribological phenomena need to be studied and understood at the micro- to nanoscales. In addition, materials for MEMS/NEMS must exhibit good micro/nanoscale tribological properties.

Macro- and microscale tribological studies of silicon and polysilicon films have been performed. The doping and oxide films improve tribological properties of these popular MEMS/NEMS materials. SiC film is found to be a good candidate material for use in high-temperature MEMS/NEMS devices. Perfluoroalkyl SAMs and bonded PFPE lubricants appear to be well suited for lubrication of microdevices under a range of environmental conditions. Adhesion of biomolecules on Si surfaces can be improved by nanopatterning and chemical linker method. Roughness should be optimized for superhydrophobicity, low adhesion and friction. Static friction force measurements of micromotors have been performed using an AFM. A bonded layer of PFPE lubricant is found to satisfactorily reduce the friction forces in the micromotor.

Mechanical properties of nanostructures are necessary in designing realistic MEMS/NEMS and BioMEMS/BioNEMS devices. Most mechanical properties are scale dependent. Bending tests have been performed on the Si and SiO2 nanobeams using an AFM. The bending tests were used to evaluate elastic modulus, bending strength (fracture stress), fracture toughness (KIC) and fatigue strength of the beam materials. The Si and SiO2 nanobeams exhibited elastic linear response with sudden brittle fracture. Elastic modulus values of 182±11 GPa for Si(110) and 85±3 GPa for SiO2 were obtained, which are comparable to bulk values. Bending strength values of 18±3 GPa for Si and 7.6±2 GPa for SiO2 were obtained, which are twice as large as values reported on larger scale specimens. This indicates that bending strength shows a specimen size dependence. Fracture toughness value estimates obtained were Embedded Image for Si and Embedded Image for SiO2, which are also comparable with values obtained on larger specimens. At stress amplitudes less than 15% of their bending strength and at mean stresses of less than 30% of the bending strength, Si and SiO2 displayed an apparent endurance life of greater than 30 000 cycles.

Bending tests were also performed on the polymer microbeams, made of PPMA, PMMA, PS and PS/clay nanocomposite. The test environment affects the mechanical properties. After 36 hours of soaking in DI water, the hardness, elastic modulus and bending properties of PPMA were affected, while the properties of the other three polymers were unaffected due to their low water absorption properties. Similarly, for tests conducted at human body temperature (37.5°C), only PPMA exhibited a decrease in the hardness and modulus as the other materials did not exhibit any change because this temperature is within the glass transition temperature range of the former but well below the Tg of the other polymers. Adding nanoclay filler improves bending strength.

The AFM and nanoindenters used in this study can be satisfactorily used to evaluate the micro/nanoscale tribological properties and mechanical properties of micro/nanoscale structures for use in MEMS/NEMS.

FEM simulations have been used to analyse the effect of type of surface roughness and scratches on stresses and deformation of nanostructures. We find that roughness affects the maximum bending stresses. The maximum bending stresses increase as the asperity number increases for both semicircular and grooved asperities in longitudinal direction. When the semicircular asperity is present in the transverse direction, the maximum tensile stress is much larger than the maximum tensile stress value with no asperity or when the semicircular asperity is present in the longitudinal direction. This observation suggests that the asperity in the transverse direction is more detrimental. The presence of scratches increases the maximum tensile stress. The maximum tensile stress remains almost the same with the number of scratches for two types of loading, i.e. when load is applied at the centre of the beam or at the centre of the scratch near the end, although the value of the tensile stress is much lower when the load is applied at the centre of the scratch. This means that the load applied at the ends is less damaging. This analysis shows that FEM simulations can be useful to designers to develop the most suitable geometry for nanostructures.


  • One contribution of 7 to a Theme Issue ‘Nanotribology, nanomechanics and applications to nanotechnology II’.


View Abstract