Handbook of Micro and Nano Tribology P16

Micro/Nanotribology and Micro/Nanomechanics of MEMS Devices Bharat Bhushan 16.1 Introduction Background • Tribological Issues16.2 Experimental TechniquesDescription of Apparatus and Test

Bhushan, B “Micro/Nanotribology and Micro/Nanomechanics of MEMS ”

Handbook of Micro/Nanotribology

Ed Bharat Bhushan

Boca Raton: CRC Press LLC, 1999

Micro/Nanotribology

and Micro/Nanomechanics

of MEMS Devices

Bharat Bhushan

16.1 Introduction Background • Tribological Issues16.2 Experimental TechniquesDescription of Apparatus and Test Procedures • Test Samples16.3 Results and Discussion

Micro/Nanotribological Studies of Virgin, Coated, and Treated Silicon Samples • Micro/Nanotribological Studies of Doped and Undoped Polysilicon Films, SiC Films, and Their Comparison to Single-Crystal Silicon • Macroscale Tribological Studies of Virgin, Coated, and Treated Samples • Boundary Lubrication Studies • Component Level Studies16.4 Closure

References

16.1 Introduction 16.1.1 Background

The advances in silicon photolithographic process technology since 1960s have led to the development

of microcomponents or microdevices, known as microelectromechanical systems (MEMS) Morerecently, lithographic processes have been developed to process nonsilicon materials These lithographicprocesses are being complemented with nonlithographic micromachining processes for fabrication ofmilliscale components or devices Using these fabrication processes, researchers have fabricated a widevariety of miniaturized devices, such as acceleration, pressure and chemical sensors, linear and rotaryactuators, electric motors, gear trains, gas turbines, nozzles, pumps, fluid valves, switches, grippers,tweezers, and optoelectronic devices with dimensions in the range of a couple to a few thousand microns(for an early review, see Peterson, 1982; for recent reviews, see Muller et al., 1990; Madou, 1997; Trimmer,1997; and Bhushan, 1998a) MEMS technology is still in its infancy and the emphasis to date has been

on the fabrication and laboratory demonstration of individual components MEMS devices have begun

to be commercially used, particularly in the automotive industry Silicon-based high-g accelerationsensors are used in airbag deployment (Bryzek et al., 1994) Acceleration sensor technology is slightlyless than a billion-dollar-a-year industry dominated by Lucas NovaSensor and Analog Devices TexasInstruments uses deformable mirror arrays on microflexures as part of airline-ticket laser printers andhigh-resolution projection devices.

Potential applications of MEMS devices include silicon-based acceleration sensors for anti-skid brakingsystems and four-wheel drives, silicon-based pressure sensors for monitoring pressure of cylinders inautomotive engines and of automotive tires, and various sensors, actuators, motors, pumps, and switches

in medical instrumentation, cockpit instrumentation, and many hydraulic, pneumatic, and other sumer products (Fujimasa, 1996) MEMS devices are also being pursued in magnetic storage systems(Bhushan, 1996a), where they are being developed for supercompact and ultrahigh-recording-densitymagnetic disk drives Horizontal thin-film heads with a single-crystal silicon substrate, referred to assilicon planar head (SPH) sliders are mass-produced using integrated-circuit technology (Lazarri andDeroux-Dauphin, 1989; Bhushan et al., 1992) Several integrated head/suspension microdevices havebeen fabricated for contact recording applications (Hamilton, 1991; Ohwe et al., 1993) High-bandwidthservo-controlled microactuators have been fabricated for ultrahigh-track-density applications whichserve as the fine-position control element of a two-stage, coarse/fine servo system, coupled with aconventional actuator (Miu and Tai, 1995; Fan et al., 1995b) Millimeter-sized wobble motors and actu-ators for tip-based recording schemes have also been fabricated (Fan and Woodman, 1995a) In somecases, MEMS devices are used primarily for their miniature size, while in others, as in the case of the airbags, because of their high reliability and low-cost manufacturing techniques This latter fact has beenpossible since semiconductor-processing costs have reduced drastically over the last decade, allowing theuse of MEMS in many previously impractical fields

con-The fabrication techniques for MEMS devices employ photolithography and fall into three basiccategories: bulk micromachining, surface micromachining, and LIGA a German acronym (LithographieGalvanoformung Abformung) for lithography, electroforming, and plastic molding The first twoapproaches, bulk and surface micromachining, use planar photolithographic fabrication processes devel-oped for semiconductor devices in producing two-dimensional (2D) structures (Jaeger, 1988; Madou,1997; Bhushan, 1998a) Bulk micromachining employs anisotropic etching to remove sections throughthe thickness of a single-crystal silicon wafer, typically 250 to 500 µm thick Bulk micromachining is aproven high-volume production process and is routinely used to fabricate microstructures such asacceleration and pressure sensors and magnetic head sliders Surface micromachining is based on depos-iting and etching structural and sacrificial films to produce a free-standing structure These films aretypically made of low-pressure chemical vapor deposition (LPCVD) polysilicon film with 2 to 20 µmthickness Surface micromachining is used to produce surprisingly complex micromechanical devicessuch as motors, gears, and grippers LIGA is used to produce high-aspect ratio (HAR) MEMS devicesthat are up to 1 mm in height and only a few microns in width or length (Becker et al., 1986) The LIGAprocess yields very sturdy 3D structures due to their increased thickness The LIGA process is based onthe combined use of X-ray photolithography, electroforming, and molding processes One of the limi-tations of silicon microfabrication processes originally used for fabrication of MEMS devices is lack ofsuitable materials which can be processed With LIGA, a variety of nonsilicon materials such as metals,ceramics and polymers can be processed Nonlithographic micromachining processes, primarily inEurope and Japan, are also being used for fabrication of millimeter-scale devices using direct materialmicrocutting or micromechanical machining (such as micromilling, microdrilling, microturning) orremoval by energy beams (such as microspark erosion, focused ion beam, laser ablation, and machining,and laser polymerization) (Friedrich and Warrington, 1998; Madou, 1998) Hybrid technologies includingLIGA and high-precision micromachining techniques have been used to produce miniaturized motors,gears, actuators, and connectors (Lehr et al., 1996, 1997; Michel and Ehrfeld, 1998) These millimeter-scale devices may find more immediate applications

Silicon-based MEMS devices lack high-temperature capabilities with respect to both mechanical andelectrical properties Recently, researchers have been pursuing SiC as a material for high-temperaturemicrosensor and microactuator applications (Tong et al., 1992; Shor et al., 1993) SiC is a likely candidatefor such applications since it has long been used in high-temperature electronics, high-frequency andhigh-power devices, such as SiC metal–semiconductor field effect transistors (MESFETS) (Spencer et al.,1994) and inversion-mode metal-oxide-semiconductor field effect transistors (MOSFETS) Many otherSiC devices have also been fabricated including ultraviolet detectors, SiC memories, and SiC/Si solar cells.SiC has also been used in microstructures such as speaker diaphragms and X-ray masks For a summary

of SiC devices and applications, see Harris (1995) Table 16.1 compares selected bulk properties of SiCand Si(100) Because of the large band gap of SiC, almost all devices fabricated from SiC have good high-temperature properties This high-temperature capability of SiC combined with its excellent mechanicalproperties, thermal dissipative characteristics, chemical inertness, and optical transparency makes SiC

an ideal choice for complementing polysilicon (polysilicon melts at 1400°C) in MEMS devices SinceMEMS devices need to be of low cost to be viable in most applications, researchers have found low-costtechniques of producing single-crystal 3C-SiC (cubic or β-SiC) films via epitaxial growth on large areasilicon substrates (Zorman et al., 1995) This technique allows high-volume batch processing and has theadvantage of having silicon as the substrate, an inexpensive material for which microfabrication andmicromachining technologies are well established It is believed that these films will be well suited forMEMS devices

16.1.2 Tribological Issues

In MEMS 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 area decreases by a factor of a million and the volumedecreases by a factor of a billion The resistive forces such as friction, viscous drag, and surface tensionthat are proportional to the area, increase a thousand times more than the forces proportional to thevolume, such as inertial and electromagnetic forces The increase in resistive forces leads to tribologicalconcerns, which become critical because friction/stiction (static friction), wear and surface contaminationaffect device performance and in some cases, can even prevent devices from working

Examples of two micromotors using polysilicon as the structural material in surface machining — a variable capacitance side drive and a wobble (harmonic) side drive — are shown inFigures 16.1 and 16.2, which can rotate up to 100,000 rpm Microfabricated variable-capacitance side-drive micromotor with 12 stators and a 4-pole rotor shown in Figure 16.1 is produced using a three-layer polysilicon process and the rotor diameter is 120 µm and the air gap between the rotor and stator

micro-is 2 µm (Tai et al., 1989) It micro-is driven electrostatically to continuous rotation (by electrostatic attractionbetween positively and negatively charged surfaces) The intermittent contact at the rotor–stator interfaceand physical contact at the rotor–hub flange interface result in wear issues, and high stiction betweenthe contacting surfaces limits the repeatability of operation or may even prevent the operation altogether.Figure 16.2 shows the SEM micrograph of a microfabricated harmonic side-drive (wobble) micromotor

TABLE 16.1 Selected Bulk Properties a of 3C ( β - or cubic) SiC and Si(100)

Sample

Density

(kg/m 3 )

Hardness (GPa)

Elastic Modulus (GPa)

Fracture Toughness (MPa m 1/2 )

Thermal Conductivity b

(W/m K)

Coeff of Thermal Expansion b

( × 10 –6 /°C)

Melting Point (°C)

Band-Gap (eV)

a Unless stated otherwise, data shown were obtained from Bhushan and Gupta (1997).

b Obtained from Shackelford et al (1994).

(Mehregany et al., 1988) In this motor, the rotor wobbles around the center bearing post rather thanthe outer stator Again friction/stiction and wear of rotor-center bearing interface are of concern There

is a need for development of bearing/bushing materials that are both compatible with MEMS fabricationprocesses and which provide superior friction and wear performance Monolayer lubricant films are also

of interest Figure 16.3 shows the SEM micrograph of an air turbine with gear or blade rotors, 125 to

240 µm in diameter, fabricated using polysilicon as the structural material in surface micromachining.The two flow channels on the top are connected to the two independent input ports and the two flowchannels at the bottom are connected to the output port Wear at the contact of gear teeth is a concern

In microvalves used for flow control, the mating valve surfaces should be smooth enough to seal while

FIGURE 16.1 (a) SEM micrograph, and (b) schematic cross-section of a variable capacitance side-drive micromotor fabricated of polysilicon film (From Tai et al., 1989, Sensors Actuators A21–23, 180–83 With permission.)

FIGURE 16.2 SEM micrograph of a harmonic side-drive (wobble) micromotor (From Mehregany, M et al., 1990,

in Proc IEEE Micro Electromechanical Systems, pp 1–8, IEEE, New York With permission.)

maintaining a minimum roughness to ensure low friction/stiction (Bhushan, 1996a, 1998b) Studies havebeen conducted to measure the friction/stiction in micromotors (Tai and Muller, 1990), gear systems(Gabriel et al., 1990) and polysilicon microstructures (Lim et al., 1990) to understand friction mechanisms.Several studies have been conducted to develop solid and liquid lubricant and hard films to minimizefriction and wear (Bhushan et al., 1995b; Deng et al., 1995; Beerschwinger et al., 1995; Koinkar andBhushan, 1996a,b; Bhushan, 1996b; Henck, 1997).

In a silicon planar head slider for magnetic disk drives shown in Figure 16.4, wear and friction/stictionare an issue because of the close proximity between the slider and disk surfaces during steady operationand continuous contacts during start and stops (Lazzari and Deroux-Dauphin, 1989; Bhushan et al.,1992) Hard diamondlike carbon (DLC) coatings are used as an overcoat for protection against corrosionand wear Two electrostatically driven rotary and linear microactuaters (surface-micromachined, poly-silicon microstructure) for a magnetic disk drive shown in Figure 16.5, consist of a movable plateconnected only by springs to a substrate, on which there are two sets of mating interdigitated electrodeswhich activate motion of the plate in opposing directions Any unintended contacts may result in wearand stiction

Figure 16.6 shows an SEM micrograph of a micromechanical switch (Peterson, 1979) As the voltage

is applied between the deflection electrode and the p+ ground plane, the cantilever beam is deflected andthe switch closes, connecting the contact electrode and the fixed electrode; wear during contact is ofconcern Figure 16.7 shows an SEM micrograph of a pair of tongs (Mehregany et al., 1988) The jawsopen when the linearly sliding handle is pushed forward, demonstrating the linear slide and the linear-to-rotary motion conversion; for this pair of tongs, the jaws open up to 400 µm in width Wear at theteeth is of concern

As an example of nonsilicon components, Figure 16.8a shows a DC brushless permanent magnetmillimotor (diameter = 1.9 mm, length = 5.5 mm) with an integrated milligear box which is producedwith parts obtained by hybrid fabrication processes including the LIGA process, micromechanicalmachining, and microspark erosion techniques (Lehr et al., 1996, 1997; Michel and Ehrfeld, 1998) Themotor can rotate up to 100,000 rpm and deliver a maximum torque of 7.5 µNm The rotor, supported

on two ruby bearings, consists of a tiny steel shaft and a diametrically magnetized rare earth magnet.The rotational speed of the motor can be converted by the use of a milligear box to increase the torquefor a specific application Gears are made of metal (e.g., electroplated Ni–Fe) or injected polymer materials(e.g., POM) using the LIGA process, Figures 16.8b and c Optimum materials and liquid and solidlubrication approaches for bearings and gears are needed

FIGURE 16.3 SEM micrograph of a gear train with three meshed gears, in an air turbine (From Mehregany, M.

et al., 1988, IEEE Trans Electron Devices 35, 719–723 With permission.)

There are tribological issues in the fabrication processes as well For example, in surface chining, the suspended structures can sometimes collapse and permanently adhere to the underlyingsubstrate, Figure 16.9 (Guckel and Burns, 1989) The mechanism of such adhesion phenomena needs to

microma-be understood (Mastrangelo, 1997)

Friction/stiction and wear clearly limit the lifetimes and compromise the performance and reliability

of microdevices Since microdevices are designed to small tolerances, environmental factors, surfacecontamination, and environmental debris affect their reliability There is a need for development of afundamental understanding of friction/stiction, wear, and the role of surface contamination and envi-ronment in microdevices (Bhushan, 1998a) A few studies have been conducted on the tribology of bulksilicon and polysilicon films used in microdevices (Bhushan and Venkatesan, 1993a,b; Gupta et al., 1993;Venkatesan and Bhushan, 1993, 1994; Gupta and Bhushan, 1994; Bhushan and Koinkar, 1994; Bhushan,1996b) Mechanical properties of polysilicon films are not well characterized (Mehregany et al., 1987;Ericson and Schweitz, 1990; Schweitz, 1991; Guckel et al., 1992; Bhushan, 1995; Fang and Wickert, 1995).The advent of atomic force/friction force microscopy (AFM/FFM) (Bhushan, 1995, 1997; Bhushan et al.,1995a) has allowed the study of surface topography, adhesion, friction, wear, lubrication, and measure-ment of mechanical properties, all on a micro- to nanometer scale Recently, microtribological studies

FIGURE 16.4 Schematic (a) of a silicon planar head slider and (b) of cross section of the slider for magnetic disk drive applications (From Bhushan, B et al., 1992, IEEE Trans Magn. 28, 2874–2876 With permission.)

have been conducted using the AFM/FFM on undoped and doped silicon and polysilicon films and SiCfilms that are used in MEMS devices (Bhushan, 1996b, 1997, 1998; Bhushan et al., 1994, 1997a,b, 1998;

Li and Bhushan, 1998; Sundararajan and Bhushan, 1998)

This chapter presents a review of macro- and micro/nanotribological studies of single-crystal siliconand polysilicon, oxidized and implanted silicon, doped and undoped polysilicon films and SiC films Asummary of limited component-level tests is also presented

FIGURE 16.5 Schematics of (a) a microactuator in place with magnetic head slider, and (b) top view of two electrostatic, rotary and linear microactuators (electrode tree structure) (From Fan, L.S et al., 1995; IEEE Trans Ind Electron. 42, 222–233 With permission.)

~ 0.6 N/m) sliding over the sample surface orthogonal to the long axis of the cantilever at 25 µm/s Acoefficient of friction and conversion factors for converting the friction signal voltage to force units (nN)were obtained through the methods developed previously by Bhushan and co-workers (Bhushan, 1995).The normal loads used in the friction measurements varied between 50 to 300 nN The reported valuesare each an average of six separate measurements.

FIGURE 16.6 SEM micrograph of single-contact and tact (with two orientations of the fixed electrodes) designs of micro- mechanical switches (Peterson, 1979) (From Peterson, K.E., 1979,

double-con-IBM J Res Dev. 23, 376 With permission.)

FIGURE 16.7 SEM micrograph of a partially released pair of tongs (From Mehregany, M et al., 1988, IEEE Trans Electron Devices 35, 719–723 With permission.)

FIGURE 16.8 Schematics of (a) permanent magnet millimotor with integrated milligear box, (b) of wolfrom-type system made of Ni–Fe metal (Lehr et al., 1996), and (c) of multistage planetary gear system made with microinjected POM plastic showing a single gear and the gear system (From Thurigen, C et al., 1998, in Tribology Issues and Opportunities in MEMS, B Bhushan, ed., Kluwer Academic, Dordrecht With permission.)

For the scratch and wear tests, specially fabricated diamond microtips were used (Bhushan et al., 1997a;Sundararajan and Bhushan, 1998) These microtips consisted of single-crystal natural diamond, ground

to the shape of a three-sided pyramid, with an apex angle of 60° and tip radius of about 70 nm, mounted

on a platinum-coated stainless steel cantilever beam whose stiffness was 50 N/m Samples were scannedorthogonal to the long axis of the cantilever with loads ranging from 20 to 100 µN to generate scratch/wearmarks Scratch tests consisted of generating scratches in a reciprocating mode at a given load for 10 cyclesover a scan length (stroke length) of 5 µm at 10 µm/s Wear marks were generated over a scan area of

2 × 2 µm at 4 µm/s and the wear marks were observed by scanning a larger 4 × 4 µm area with the wearmark at the center Imaging scans of both scratch and wear tests were done at a low normal load of 0.5.The reported scratch/wear depths are an average of three runs at separate instances All measurementswere performed in an ambient environment (21 ± 1°C, 45 ± 5% RH)

Hardness and elastic modulus were calculated from load–displacement data obtained by tation using a commercially available nanoindenter (Bhushan, 1995; Bhushan et al., 1997b; Li and Bhus-han, 1998) The instrument monitored and recorded dynamic load and displacement of a three-sidedpyramidal diamond (Berkovich) indenter with a force resolution of about 75 nN and displacementresolution of about 0.1 nm Multiple loading and unloading were performed to examine reversibility ofthe deformation and thereby ensuring that the regime was elastic

nanoinden-The fracture toughness measurements were made using a microindentation technique A Vickersindenter (four-sided diamond pyramid) was used to indent samples in a microhardness tester at a normalload of 0.5 N The indentation impressions were examined in an optical microscope to measure thelength of median-radial cracks to calculate the fracture toughness (Li and Bhushan, 1998)

16.2.1.2 Macroscale Tests

Macroscale studies were conducted using either a ball-on-flat tribometer under reciprocating motion or

a magnetic rigid disk drive In the ball-on-flat tribometer tests, a 5-mm diameter alumina ball (hardness

~ 21 GPa) was slid in a reciprocating mode (2 mm amplitude and 1 Hz frequency) under a normal load

of 1 N in the ambient environment (Gupta et al., 1993) The coefficient of friction was measured duringthe tests using a strain gauge ring Wear volume was measured by measuring the wear depth using astylus profiler

In the magnetic disk drive tests, a modified disk drive was used The silicon pins or magnetic headslider specimens to be tested were slid in a unidirectional sliding mode against a magnetic thin-film diskunder a normal load of 0.15 N and the rotational speed of 200 rpm The sliding speeds at track radiiranging from 45 to 55 mm varied from 0.9 to 1.2 m/s (Bhushan and Venkatesan, 1993) At these speeds,the pin or slider specimen remained in contact throughout the period of testing The coefficient of frictionwas measured during the tests using a strain gauge beam Samples were examined using scanning electronmicroscopy to detect any wear Chemical analyses of the samples were also carried out to study failuremechanisms

FIGURE 16.9 Schematics of microstructures during fabrication using surface micromachining before and after removal of sacrificial/spacer layer.

16.2.2 Test Samples

Materials of most interest for planar fabrication processes using silicon as the structural material areundoped and boron-doped (p+-type) single-crystal silicon and phosphorus-doped (n+-type) LPCVDpolysilicon films For tribological reasons, silicon needs to be coated with a solid and/or liquid overcoat

or be surface treated, which exhibits low friction and wear

Studies have been conducted on various types of virgin silicon samples: undoped (lightly doped)single-crystal Si(100), Si(111), and Si(110) and the following types of treated/coated silicon samples:PECVD-oxide-coated Si(111), dry-oxidized, wet-oxidized, and C+-implanted Si(111) (Bhushan and Ven-katesan, 1993; Bhushan and Koinkar, 1994) Studies have also been conducted on heavily doped (p+-type)single-crystal Si(100), undoped polysilicon film, heavily doped (n+-type) polysilicon film and 3C-SiC(cubic or β-SiC) film (Bhushan et al., 1997a,b, 1998; Li and Bhushan, 1998; Sundararajan and Bhushan,1998) A 10 × 10 mm coupon of each sample was ultrasonically cleaned in methanol for 20 min anddried with a blast of dry air prior to measurements The undoped Si(100) was a p-type material grown

by the CZ process It had a boron concentration of 1.7 × 1015 ions/cm3 from intrinsic doping during themanufacturing process The doped wafer (p+-type single-crystal silicon) was heavily doped with boronions (from a solid source of oxide of boron) with concentration of 7 × 1019 ions/cm3 down to a depth of5.5 µm using thermal diffusion The grain size of polysilicon wafer was about 5 mm The polysilicon filmwas produced as follows: (1) The substrate used was thermally oxidized Si(100) wafers with the oxidelayer grown using a standard wet oxidation recipe to a nominal thickness of about 100 nm; (2) thepolysilicon film was grown on the substrate using an LPCVD process (deposition temperature, 610°C;silane flow rate, 285 sccm; deposition pressure, 230 mtorr), using the thermal decomposition of silanevapor The films were about 3 µm thick, with columnar grains and a grain size of about 750 nm X-raydiffraction and transmission electron microscope characterization showed the film to be highly oriented(110) The n+-doped polysilicon film was obtained by doping the polysilicon film with phosphorus ionsfrom a solid source of P2O5 by thermal diffusion at 875°C for 90 min The 3C-SiC films were grownthrough an atmospheric pressure chemical vapor deposition (APCVD) process on an Si(100) substrate

To grow the SiC film on the wafer by carbonizing its surface, the wafer is placed on an SiC-coated graphitesusceptor, which is induction-heated by an RF-generator to the growth temperature of 1360°C in thepresence of propane and silane at 1 atm Prior to film growth, the wafer is heated to 1000°C in thepresence of hydrogen, which etches the native oxide from the wafer surface (Zorman et al., 1995) Thefilms obtained were about 2 µm thick Both as-deposited and polished versions of the undoped polysiliconand SiC films were studied The polysilicon film was chemomechanically polished in a Struers Planopol-

3 polishing machine using 100 ml colloidal silica dispersion (Rippey Corporation, particle size of 30 to

100 nm) mixed in 2000 ml deionized water at a force of 210 N for 15 min, with the pads running at

150 rpm in the same direction The SiC and doped polysilicon films were polished in a Buehler

Ecomet-3 polishing machine with diamond slurry (General Electric Company, particle size of 100 to 500 nm)for 30 min for SiC and 12 min for doped polysilicon film at a load of 50 N, with the pads running at

10 rpm in the same direction Doped polysilicon film was polished using Fuji film lapping tape, LT-2,the main lapping agent being 37-µm-sized Cr2O3 particles

Boundary lubrication studies have been conducted on silicon samples coated with perfluoropolyetherlubricants (Koinkar and Bhushan, 1996a,b) and Langmuir–Blodgett and chemically grafted self-assem-bled monolayer films (Bhushan et al., 1995b)

16.3 Results and Discussion

Reviews of five studies are presented in this section The first study compares micro/nanotribologicalproperties of various forms of virgin, coated, and treated silicon samples The second study is composed

of similar studies conducted on SiC film and compares this material to other materials currently used

in MEMS devices The third study compares the macroscale friction and wear data of virgin, coated, and

treated silicon samples The fourth study discusses various forms of boundary lubrication that may besuitable for MEMS devices Finally, the fifth study presents a review of component level studies.

16.3.1 Micro/nanotribological Studies of Virgin, Coated,

and Treated Silicon Samples

Table 16.2 summarizes the results of the studies conducted on various silicon samples (Bhushan andKoinkar, 1994) Coefficient of microscale friction values of all the samples are about the same Table 16.3compares macroscale and microscale friction values for two of the samples When measured for smallcontact areas and very low loads used in microscale studies, indentation hardness and elastic modulusare higher than that at the macroscale This reduces wear This, added to the effect of the small apparentarea of contact reducing the number of trapped particles on the interface, results in less plowing contri-bution in the case of microscale friction measurements Figure 16.10 and Table 16.2 show microscalescratch data for the various silicon samples (Bhushan and Koinkar, 1994) These samples could bescratched at 10 µN load Scratch depth increased with normal load Crystalline orientation of silicon haslittle influence on scratch resistance PECVD-oxide samples showed the best scratch resistance, followed

by dry-oxidized, wet-oxidized, and ion-implanted samples Ion implantation does not appear to improvescratch resistance Microscratching experiments just described can be used to study failure mechanisms

on the microscale and to evaluate mechanical integrity (scratch resistance) of ultrathin films at low loads.Wear data on the silicon samples are presented in Table 16.2 (Bhushan and Koinkar, 1994) PECVD-oxide samples showed superior wear resistance followed by dry-oxidized, wet-oxidized, and ion-implanted samples This agrees with the trends seen in scratch resistance In PECVD, ion bombardment

TABLE 16.2 rms, Microfriction, Microscratching/Microwear and Nanoindentation Hardness Data for Various Virgin, Coated, and Treated Silicon Samples

Material

rms Roughness a

(nm)

Microscale Coefficient

b Versus Si 3 N 4 ball, ball radius of 3 mm at a normal load of 0.1 N (0.3 GPa) at an average sliding speed of 0.8 mm/s.

c Measured using an AFM with a diamond tip of radius of 100 nm.

TABLE 16.3 Surface Roughness and Coefficients of Micro- and Macroscale Friction of Selected Samples

Material

rms Roughness (nm)

Coefficient of Microscale Friction a

Coefficient of Macroscale Friction b

during the deposition improves the coating properties such as suppression of columnar growth, freedomfrom pinhole, decrease in crystalline size, and increase in density, hardness and substrate–coating adhe-sion These effects may help in improving mechanical integrity of the sample surface.

The wear resistance of ion-implanted silicon samples was further studied, Figure 16.11 (Bhushan andKoinkar, 1994) For tests conducted at various loads on Si(111) and C+-implanted Si(111), it is notedthat wear resistance of implanted sample is slightly poorer than that of virgin silicon up to about 80 µN.Above 80 µN, the wear resistance of implanted Si improves As one continues to run tests at 40 µN for

a larger number of cycles, the implanted sample exhibits higher wear resistance than the unimplantedsample Damage from the implantation in the top layer results in poorer wear resistance; however, theimplanted zone at the subsurface is more wear resistant than the virgin silicon

Nanoindentation hardness values of all samples are presented in Table 16.2 Coatings and treatmentsimproved nanohardness of silicon Note that dry-oxidized and PECVD films are harder than wet-oxidizedfilms, as these films may be porous High hardness of oxidized films may be responsible for measuredlow wear on the microscale and macroscale (data to be presented later) Figure 16.12 shows the inden-tation marks generated on virgin and C+-implanted Si(111) at a normal load of 70 µN with a depth ofindentation about 3 nm and hardness values of 15.8 and 19.5 GPa, respectively (Bhushan and Koinkar,1994) Hardness values of virgin and C+-implanted Si(111) at various indentation depths (normal loads)are presented in Figure 16.13 (Bhushan and Koinkar, 1994) Note that the hardness at a small indentationdepth of 2.5 nm is 16.6 GPa and it drops to a value of 11.7 GPa at a depth of 7 nm and a normal load

of 100 µN Higher hardness values obtained in low-load indentation may arise from the observed sure-induced phase transformation during the nanoindentation (Pharr, 1991; Callahan and Morris,1992) Additional increase in the hardness at an even lower indentation depth of 2.5 nm reported heremay arise from the contribution by complex chemical films (not from native oxide films) present on thesilicon surface At small volumes there is a high probability that indentation would be made into a regionthat was initially dislocation free Furthermore, at small volumes, it is believed that there is an increase

pres-in the stress necessary to operate dislocation sources (Gane and Cox, 1970; Sargent, 1986) These aresome of the plausible explanations for an increase in hardness at smaller volumes If the silicon material

is to be used at very light loads such as in microsystems, the high hardness of surface films would protectthe surface until it is worn

From Figure 16.13, hardness values of C+-implanted Si(111) at a normal load of 50 µN is 20.0 GPawith an indentation depth of about 2 nm which is comparable to the hardness value of 19.5 GPa at 70 µN,whereas measured hardness value for virgin silicon at an indentation depth of about 7 nm (normal load

of 100 µN) is only about 11.7 GPa Thus, ion implantation results in an increase in hardness Note thatthe surface layer of the implanted zone is much harder compared with the subsurface, and may be brittleleading to higher wear on the surface The subsurface of the implanted zone is harder than the virginsilicon, resulting in higher wear resistance, which will be shown later in macroscale tests conducted athigh loads

FIGURE 16.10 Scratch depth as a function of normal force after 10 cycles for various silicon samples, virgin, treated, and coated (From Bhushan, B and Koinkar, V.N., 1994, J Appl Phys. 75, 5741–5746 With permission.)

16.3.2 Micro/Nanotribological Studies of Doped and Undoped

Polysilicon Films, SiC Films, and Their Comparison

to Single-Crystal Silicon

16.3.2.1 Surface Roughness and Friction

The surface roughness of various samples obtained with the AFM is compared in Figure 16.14a(Sundararajan and Bhushan, 1998) Polishing of the as-deposited polysilicon and SiC films drasticallyaffect the roughness as the values reduce by two orders of magnitude Si(100) appears to be the smoothestfollowed by polished undoped polysilicon and SiC films, which have comparable roughness The dopedpolysilicon film shows higher roughness than the undoped sample, which is attributed to the dopingprocess The coefficients of microscale friction of the various samples are shown in Figure 16.14b (Bhus-han et al., 1998; Sundararajan and Bhushan, 1998) Despite being the smoothest sample, Si(100) showshigher friction than the other samples Doped polysilicon also shows high friction, which is due to thepresence of grain boundaries and high surface roughness In the case of the undoped polysilicon andSiC films, polishing did not affect the friction values by much From the data, the polished SiC filmshows the lowest friction followed by polished, undoped polysilicon film, which strongly supports thecandidacy of SiC films for use in MEMS devices Gray-scale top-view images of surface height and

FIGURE 16.11 Wear depth as a function of (a) load (after one cycle), and (b) cycles (normal load = 40 µN) for Si(111) and C + -implanted Si(111) (From Bhushan, B and Koinkar, V.N., 1994, J Appl Phys. 75, 5741–5746 With permission.)

corresponding friction force of selected samples obtained with the AFM/FFM are shown in Figure 16.15.

Brighter regions indicate higher height than darker regions in the case of the surface height images, while

brighter regions indicate higher friction force experienced by the tip than the darker regions The doped

polysilicon film shows a large number of grain boundaries It appears that regions of high and low surface

height do not necessarily correspond to regions of high and low friction

The low microscale friction exhibited by SiC compared to the other materials agrees with the fact that

many ceramic–ceramic interfaces generally show low friction on the macroscale In this study, the Si3N4

tip–SiC film interface also shows the same trend In the case of ceramic materials, formation of

tribo-chemical films on the surface due to sliding results in low values of friction SiC can react with water

vapor to form silicon hydroxide and oxide, which improve the frictional behavior (Xu and Bhushan,

1997) From this discussion, it can be seen that an obvious drawback with SiC is its readiness to form

tribochemical films, which is environment dependent Another factor that affects friction in ceramic

materials is fracture, which leads to high friction because of the plowing contribution But in the case

of SiC film, fracture toughness is believed to follow the same trend as that of the bulk material, which

is higher than that of silicon (see Table 16.1), thereby resulting in lower plowing contribution and hence

lower friction Macroscale friction measurements indicate that SiC film exhibits one of the lowest friction

values as compared to the other samples Doped polysilicon sample shows low friction on the macroscale

as compared to the undoped polysilicon sample possibly due to the doping effect

16.3.2.2 Scratch/Wear Tests

As explained earlier, the scratch tests consisted of making scratches for ten cycles with varying loads

Figure 16.16a shows a plot of scratch depth vs normal load for various samples (Bhushan et al., 1998;

Sundararajan and Bhushan, 1998) Error bars are given for the Si(100) data The variation was typically

about 12% Scratch depth increases with increasing 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 The increase in scratch depth with normal load is very small

and all depths are less than 30 nm, while the Si(100) and polysilicon films reach depths in excess of

150 nm Figure 16.17 shows 3D images of the scratch marks

Wear tests were conducted on the samples for one cycle at normal loads ranging from 20 to 80 µN

The resulting wear depths are plotted against the normal loads in Figure 16.16b (Bhushan et al., 1998;

Sundararajan and Bhushan, 1998) Again, the wear depth increases with increasing normal load Similar

to the scratch resistance data, Si(100) and the polysilicon samples also behave similarly in terms of wear

resistance The SiC film starts out showing comparable wear depth at 20 µN to all the other samples, but

at higher loads SiC film shows superior wear resistance Also there is hardly an increase in wear depth

with increasing normal load in the case of SiC film The evolution of wear on the various samples was

also studied This test was performed by wearing the same region for 20 cycles at a normal load of 20 µN,

while observing wear depths at different intervals (1, 2, 10, 15, and 20 cycles), Figure 16.18 This would

give information as to the progression of wear of the material The wear depths observed are plotted

against the number of cycles in Figure 16.16b For all the materials, the wear depth increases almost

linearly with increasing number of cycles This suggests that the material is removed layer by layer in all

the materials Here also, SiC film exhibits lower wear depths than the other samples Doped polysilicon

film wears less than the undoped film Not many debris particles are seen in the figures This is due to

the fact that the debris particles created are loose and are pushed outside the imaging area by the sliding

tip during scanning It can also be seen from the figure that the wear marks are uniform at the bottom,

indicating that uniform wear has occurred with particle pile up on the edges of the wear mark

The wear tests clearly show that SiC film possesses an extremely wear-resistant surface compared to

the other samples and, together with the scratch tests results, indicate that SiC film has better surface

mechanical properties than Si(100) and the polysilicon films Table 16.4 shows a summary of the various

properties of the samples measured in this study (Bhushan et al., 1998; Sundararajan and Bhushan 1998)

The superior scratch/wear resistance of the SiC film is consistent with the hardness values near the surface

measured with the nanoindenter Higher hardness of SiC film is one of the factors responsible for its

better scratch/wear resistance Wear in ceramic materials in the case of asperity contacts occurs due to

brittle fracture As the AFM tip slides over the material surface, Hertzian cone cracks can occur when

the normal stress exceeds a critical value (Hutchings, 1992) Friction (tangential) forces during sliding

reduce this critical value High fracture toughness and low coefficient of friction of SiC film help in

reducing the chances of brittle fracture, which results in low wear Therefore, abrasive wear due to plastic

deformation and fracture govern the wear process In the case of all samples, surface reactions result in

the formation of tribochemical films that are different in nature than the underlying material Therefore,

at low loads and at the onset of wear, all samples would show similar wear depths as they all have

comparable interfacial shear strengths and attack angles (attack angle is the included angle between the

leading face of the asperity and the contact plane at the point of contact) This is seen in the case of the

reported wear depths at 20 µN for 1 cycle (Figure 16.16b) In the case of the scratch test data

(Figure 16.16a), SiC shows slightly lower scratch depth at 20 µN since the data is for ten cycles, during

FIGURE 16.12

which time it is possible that the bulk properties of the material start coming into play in addition to

that of the tribochemical films This is consistent with the wear depths at 20 µN for ten cycles and more

(Figure 16.16c) As the normal load increases, Si(100) and polysilicon films show a sharp increase in

degree of penetration (ratio of groove depth to radius of contact) and attack angle, which leads to higher

wear (Hokkirigawa and Kato, 1988; Koinkar and Bhushan, 1997) Higher fracture toughness and higher

hardness of SiC as compared to Si(100) is responsible for its lower wear Also the higher thermal

conductivity of SiC (see Table 16.1) as compared to the other materials leads to lower interface

temper-atures which generally results in less degradation of the surface (Bhushan, 1996a) Doping of the

poly-silicon does not affect the scratch/wear resistance and hardness much The measurements made on the

doped sample are affected by the presence of grain boundaries

FIGURE 16.12 Gray-scale plot and line plot of the inverted nanoindentation mark on (a) Si(111) at 70 µN (hardness

~ 15.8 GPa), and (b) gray-scale plot of indentation mark on C + -implanted Si(111) at 70 µN (hardness ~ 19.5 GPa).

The indentation depth of indent was about 3 nm.

FIGURE 16.13 Nanohardness and normal load as function of indentation depth for virgin and C + -implanted

Si(111) (From Bhushan, B and Koinkar, V.N., 1994, J Appl Phys. 75, 5741–5746 With permission.)