Bhushan B. Handbook of Micro/Nano Tribology, Second Edition

Подождите немного. Документ загружается.

crystal tube mount supporting the upper silica disk of the basic apparatus shown in Figure 1.43 is replaced.

Lateral motion is initiated by a variable-speed motor-driven micrometer screw that presses against the

translation stage, which is connected via two horizontal double-cantilever strip springs to the rigid

mounting plate. The translation stage also supports two vertical double-cantilever springs (Figure 1.46)

that at their lower end are connected to a steel plate supporting the upper silica disk. One of the vertical

springs acts as a frictional force detector by having four resistance strain gauges attached to it, forming

the four arms of a Wheatstone bridge and electrically connected to a chart recorder. Thus, by rotating

the micrometer, the translation stage deﬂects, causing the upper surface to move horizontally and linearly

at a steady rate. If the upper mica surface experiences a transverse frictional or viscous shearing force,

this will cause the vertical springs to deﬂect, and this deﬂection can be measured by the strain gauges.

The main support, force-measuring double-cantilever spring, movable clamp, white light, etc. are all

parts of the original basic apparatus (Figure 1.43) whose functions are to control the surface separation,

vary the externally applied normal load, and measure the separation and normal force between the two

surfaces, as already described. Note that during sliding, the distance between the surfaces, their true

molecular contact area, their elastic deformation, and their lateral motion can all be simultaneously

monitored by recording the moving FECO fringe pattern using a video camera and recording it on a

tape (Gee et al., 1990).

The two surfaces can be sheared past each other at sliding speeds which can be varied continuously

from 0.1 to 20 µm/s while simultaneously measuring both the transverse (frictional) force and the normal

(compressive or tensile) force between them. The lateral distances traversed are on the order of several

hundreds of micrometers which correspond to several diameters of the contact zone.

With an oscillatory shear attachment, developed by Granick et al., viscous dissipation and elasticity

and dynamic viscosity of conﬁned liquids by applying periodic sinusoidal oscillations of one surface with

respect to the other can be studied (Van Alsten and Granick, 1988, 1990a,b; Peachey et al., 1991; Hu

et al., 1991). This attachment allows for the two surfaces to be sheared past each other at varying sliding

speeds or oscillating frequencies while simultaneously measuring both the transverse (friction or shear)

force and the normal load between them. The externally applied load can be varied continuously, and

both positive and negative loads can be applied. Finally, the distance between the surfaces, their true

molecular contact area, their elastic (or viscoelastic) deformation, and their lateral motion can all be

simultaneous by recording the moving interference fringe pattern using a video camera–recorder system.

To produce shear while maintaining constant ﬁlm thickness or constant separation of the surfaces, the

top mica surface is suspended from the upper portion of the apparatus by two piezoelectric bimorphs.

A schematic description of the SFA with the installed shearing device is shown in Figure 1.47 (Van Alsten

and Granick, 1988, 1990a,b; Peachey et al., 1991; Hu et al., 1991). Israelachvili (1989) and Luengo et al.

FIGURE 1.46 Schematic of the sliding attachment. The translation stage also supports two vertical double-cantilever

springs, which at their lower end are connected to a steel plate supporting the upper silica disk. (From Gee, M.L.

et al. (1990), J. Chem. Phys., 93, 1895–1906. With permission.)

(1997) have also presented similar designs. The lower mica surface, as in the steady-shear sliding attach-

ment, is stationary and sits at the tip of a double-cantilever spring attached at the other end to a stiff

support. The externally applied load can be varied continuously by displacing the lower surface vertically.

An AC voltage difference applied by a signal generator (driver) across one of the bimorphs tends to bend

it in an oscillatory fashion while the frictional force resists that motion. Any resistance to sliding induces

an output voltage across the other bimorphs (receiver), which can be easily measured by a digital

oscilloscope. The sensitivity in measuring force is on the order of a few micronewtons and the amplitudes

of measured lateral displacement can range from a few nanometers to 10 µm. The design is ﬂexible and

allows the inducement of time-varying stresses with different characteristic wave shapes simply by

changing the waveform of the input electrical signal. For example, when measuring the apparent viscosity,

a sine wave input is convenient to apply. Figure 1.48a shows an example of the raw data, obtained with

a hexadecane ﬁlm at a moderate pressure, when a sine wave was applied to one of the bimorphs (Van

Alsten and Granick, 1990a). By comparing the calibration curve with the response curve, which was

attenuated in amplitude and lagged in phase, an apparent dynamic viscosity can be inferred. On the

other hand, a triangular waveform is more suitable when studying the yield stress behavior of solidlike

ﬁlms as in Figure 1.48b. The triangular waveform, showing a linear increase and decrease of the applied

force with time, is proportional to the driving force acting on the upper surface. The response waveform,

which represents a resistance of the interface to shear, remains very small indicating that the surfaces are

in stationary contact with respect to each other until the applied stress reaches a yield point. At the yield

point the slope of the response curve increases dramatically, indicating the onset of sliding.

Homola (1994) compared the two approaches — steady-shear attachment and oscillating-shear attach-

ment. In experiments conducted by Israelachvili and his co-workers, the steady-shear attachment was

employed to focus on the dynamic frictional behavior of the ﬁlm after a sufﬁciently high shear stress was

applied to exceed the yield stress and to produce sliding at a constant velocity. In these measurements, the

ﬁlm was subjected to a constant shearing force for a time sufﬁciently long enough to allow them to reach

a dynamic equilibrium; i.e., the molecules, within the ﬁlm, had enough time to order and align with respect

to the surface, both normally and tangentially. Under these conditions, dynamic friction was observed to

be “quantized” according to the number of liquid layers between the solid surfaces and independent of the

shear rate (Israelachvili et al., 1988). Clearly, in this approach, the molecular ordering is optimized by a

steady shear which imposes a preferred orientation on the molecules in the direction of shear.

FIGURE 1.47 Schematic of the oscillatory shearing apparatus. (From Van Alsten, J. and Granick, S. (1988), Phys.

Rev. Lett., 61, 2570–2573. With permission.)

The above mode of sliding is particularly important when the sheared ﬁlm is made of long-chain

lubricant molecules requiring a signiﬁcantly long sliding time to order and align and even a longer time

to relax (disorder) when sliding stops. This suggests that a steady-state friction is realized only when the

duration of sliding exceeds the time required for an ensemble of the molecules to order fully in a speciﬁc.

shear ﬁeld. It also suggests that static friction should depend critically on the sliding time and the extent

of the shear-induced ordering (Homola, 1993).

In contrast, the oscillatory-shear method, which utilizes periodic sinusoidal oscillations over a range

of amplitudes and frequencies, addresses a response of the system to rapidly varying strain rates and

directions of sliding. Under these conditions, the molecules, especially those exhibiting a solid-like

behavior, cannot respond sufﬁciently fast enough to stress and are unable to order fully during duration

of a single pass i.e., their dynamic and static behavior reﬂects an oscillatory shear-induced ordering which

might or might not represent an equilibrium dynamic state. Thus, the response of the sheared ﬁlm will

depend critically on the conditions of shearing i.e., the strain, the pressure, and the sliding conditions

(amplitude and frequency of oscillations) which in turn will determine a degree of molecular ordering.

This may explain the fact that the layer structure and “quantization” of the dynamic and static friction

were not observed in these experiments in contrast to results obtained when velocity was kept constant.

FIGURE 1.48 (a) Two output signals induced by an applied sine wave (not shown) are displaced. The “calibration”

waveform is obtained with the mica sheets completely separated. The response waveform is obtained with a thin

liquid ﬁlm between the sheets, which causes it to lag the calibration waveform. (b) The oscilloscope trace of the drive

and response voltages used to determine critical shear stress. The drive waveform shows voltage proportional to

induced stress on the sheared ﬁlm and the response waveform shows voltage proportional to resulting velocity. Spikes

in the response curve correspond to the stick-slip event. (From Van Alsten, J. and Granick, S. (1990), Tribol. Trans.,

33, 436–446. With permission.)

Intuitively, this behavior is expected considering that the shear-ordering tendency of the system is

frequently disturbed by a shearing force of varying magnitude and direction. Nonetheless, the technique

is capable of providing an invaluable insight into the shear behavior of molecularly thin ﬁlms subjected

to nonlinear stresses as it is frequently encountered in practical applications. This is especially true under

conditions of boundary lubrication, where interacting surface asperities will be subjected to periodic

stresses of varying magnitudes and frequencies (Homola, 1993).

1.3.4.2 Georges et al.’s Design

The SFA developed by Tonck et al. (1988) and Georges et al. (1993) to measure static and dynamic forces

in the normal direction, between surfaces in close proximity, is shown in Figure 1.49. In their apparatus,

a drop of liquid is introduced between a macroscopic spherical body and a plane. The sphere is moved

toward and away from a plane using the expansion and the vibration of a piezoelectric crystal. Piezo-

electric crystal is vibrated at low amplitude around an average separation for dynamic measurements to

provide dynamic function of the interface. The plane specimen is supported by a double-cantilever spring.

Capacitance sensor C

measures the elastic deformation of the cantilever and thus the force transmitted

through the liquid to the plane. Second capacitance sensor C

is designed to measure the relative

displacement between the supports of the two solids. The reference displacement signal is the sum of

two signals: ﬁrst, a ramp provides a constant normal speed from 50 to 0.01 nm/s, and, second, the

piezoelectric crystal is designed to provide a small sinusoidal motion in order to determine the dynamic

behavior of sphere–plane interactions. A third capacitance sensor C measures the electrical capacitance

between the sphere and the plane. In all cases, the capacitance is determined by incorporating the signal

of an oscillator in the inductive capacitance (L-C) resonant input stage of an oscillator to give a signal-

dependent frequency in the range of 5 to 12 MHz. The resulting ﬂuctuations in oscillation frequency are

detected using a low-noise-frequency discriminator. Simultaneous measurements of sphere–plane dis-

placement, surface force, and the damping of the interface allow an analysis of all regimes of the interface

(Georges et al., 1993). Loubet et al. (1993) used SFA in the crossed-cylinder geometry using two freshly

cleaved mica sheets similar to the manner used by Israelachvili and co-workers.

Georges et al. (1994) modiﬁed their original SFA to measure friction forces. In this apparatus, in

addition to having the sphere move normal to the plane, the sphere can be sheared at constant separation

FIGURE 1.49 Schematic of the SFA that employs a sphere–plane arrangement. (From Georges, J.M. et al. (1993),

J. Chem. Phys., 98, 7345–7360. With permission.)

from the plane. The shear force apparatus is shown in Figure 1.50. Three piezoelectric elements controlled

by three capacitance sensors permit accurate motion control and force measurement along three orthog-

onal axes with displacement sensitivity of 10

–3

nm and force sensitivity of 10

–8

N. Adhesion and normal

deformation experiments are conducted in the normal approach (Z-axis). Friction experiments are

conducted by introducing displacement in the X-direction at a constant normal force. In one of the

experiments, Georges et al. (1994) used a 2.95-mm-diameter sphere made of cobalt-coated fused boro-

silicate glass and a silicon wafer for the plane.

1.3.5 Vibration Isolation

STM, AFM, and SFA should be isolated from sources of vibration in the acoustic and subacoustic

frequencies, especially for atomic-scale measurements. Vibration isolation is generally provided by placing

the instrument on a vibration isolation air table. For further isolation, the instrument should be placed

on a pad of soft silicone rubber. A cheaper alternative consists of a large mass of 100 or more newtons

suspended from elastic “bungee” cords. The mass should stretch the cords at least 0.3 m, but not so much

that the cords reach their elastic limit. The instrument should be placed on the large mass. The system,

including the microscope, should have a natural frequency of about 1 Hz or less both vertically and

horizontally. Test this by gently pushing on the mass and measure the rate at which it swings or bounces

(Anonymous, 1992a).

1.4 Magnetic Storage and MEMS Components

1.4.1 Magnetic Storage Devices

Magnetic storage devices used for storage and retrieval are tape and ﬂexible and rigid disk drives. These

devices are used for audio, video, and data storage applications. The magnetic storage industry is a

$90 billion a year industry with $30 billion for audio and video recording (almost all tape drives/media)

and $60 billion for data storage. In the data storage industry, magnetic rigid disk drives/media, tape

drives/media, ﬂexible disk drives/media, and optical disk drive/media account for about $42, $9, $5, and

$4 billion, respectively. Magnetic recording and playback involves the relative motion between a magnetic

medium against a read–write magnetic head. Heads are designed so that they develop a (load-carrying)

hydrodynamic air ﬁlm under steady operating conditions to minimize head–medium contact. However,

FIGURE 1.50 Schematic of shear force apparatus. (From Georges, J.M. et al. (1994), Wear, 175, 59–62. With

permission.)

physical contact between the medium and head occurs during starts and stops (Bhushan, 1992, 1996a).

In the modern magnetic storage devices, the ﬂying heights (head-to-medium separation) are on the order

of 50 nm and roughnesses of head and medium surfaces are on the order of 2 nm rms. The need for

ever-increasing recording densities requires that surfaces be as smooth as possible and the ﬂying heights

be as low as possible. In fact, several contact or near-contact recording devices are at various stages of

development. High static friction (stiction) at the head–medium interface (HMI) after storage and

increase in kinetic friction after use as a result of medium wear and head wear from usage are the major

impediments to the commercialization of the contact recording.

Magnetic media fall into two categories: (1) particulate media, where magnetic particles (γ-Fe

, Co-

γFe

, CrO

, Fe or metal, barium ferrite) are dispersed in a polymeric matrix and coated onto a polymeric

substrate for ﬂexible media (tape and ﬂoppy disks) or onto a rigid substrate (typically aluminum and

more recently glass or glass ceramic) and (2) thin-ﬁlm media, where continuous ﬁlms of magnetic

materials are deposited onto the substrate by vacuum techniques. The most commonly used thin magnetic

ﬁlms for tapes are evaporated Co–Ni (82–18 at %). Typical magnetic ﬁlms for rigid disks are metal ﬁlms

of cobalt-based alloys (such as sputtered Co–Pt–Ni, Co–Ni, Co–Pt–Cr, Co–Cr, and Co–NiCr). For high

recording densities, trends have been to use thin-ﬁlm media. Magnetic heads used to date are either

conventional inductive or thin-ﬁlm inductive and magnetoresistive (MR) heads. The air-bearing surfaces

(ABS) of tape heads are cylindrical in shape. For dual-sided ﬂoppy-disk heads, two heads are either

spherically contoured and slightly offset (to reduce normal pressure) or are ﬂat and loaded against each

other. The ABS of rigid-disk heads are a two- or three-rail taper ﬂat design supported by a leaf-spring

(ﬂexure) suspension. The ABS of magnetic heads are almost exclusively made of Mn–Zn ferrite, Ni–Zn

ferrite, Al

–TiC, and calcium titanate. The ABS of some conventional heads are made of plasma-sprayed

coatings of hard materials such as Al

–TiO

and ZrO

(Bhushan, 1992, 1996a).

Figure 1.51 shows the schematic of a data-processing linear tape drive (IBM 3490) which uses a

rectangular tape cartridge (100 × 125 × 25 mm). Figure 1.52A shows the sectional views of particulate

and thin-ﬁlm magnetic tapes. Almost exclusively, the base ﬁlm is made of semicrystalline, biaxially

oriented poly (ethylene terephthalate) or PET. The particulate coating formulation consists of binder

(typically polyester polyurethane), submicron magnetic particles, submicron head-cleaning agents (typ-

ically alumina), and lubricant (typically fatty acid ester). The thin-ﬁlm tape is topically lubricated typically

with a perﬂuoropolyether lubricant. Figure 1.52B shows the schematic of a 36-track, thin-ﬁlm read–write

head (with a radius of cylindrical contour of about 20 mm) whose ABS is made of Ni–Zn ferrite with

multilayered thin-ﬁlm head structures. A tape tension of about 2 N over a 12.7-mm-wide tape (normal

pressure ~14 kPa) is used during use. The rms roughnesses of ABS of the heads and tape surfaces typically

are 1 to 1.5 nm and 5 to 10 nm, respectively.

Figure 1.53 shows the schematic of a data-processing rigid-disk drive with 48-, 65-, 95-, or 130-mm

form factor. A nonremovable stack of multiple disks mounted on a ball-bearing spindle is rotated by an

electric motor at constant angular speed ranging from about 3000 to 10,000 rpm, dependent upon the

disk size. Head slider–suspension assembly (allowing one slider for each disk surface) is actuated by a

stepper motor or a voice coil motor using a linear or rotary actuator. Generally a rotary actuator is used

to save space, as shown in Figure 1.53. Figure 1.54A shows the sectional views of particulate and thin-

ﬁlm rigid disks. The substrate for rigid disks is generally the nonheat-treatable aluminum–magnesium

alloy 5086. For metal thin-ﬁlm disks, the substrate is coated with a hard coating to improve its surface

hardness to Knoop 600 to 800 kg/mm

and smoothness. The particulate coating formulation consists of

binder (typically hard epoxy), submicron magnetic particles, and submicron head-cleaning agent (typ-

ically alumina). The protective overcoat commonly used for thin-ﬁlm disks is generally either sputtered

diamondlike carbon (DLC) or silicon dioxide. Both types of disks are topically lubricated with perﬂuo-

ropolyether types of lubricants. Lubricants with polar-end groups are generally used for thin-ﬁlm disks

in order to provide chemical bonding to the overcoat surface. Figure 1.54B shows the schematic of a

thin-ﬁlm head nanoslider with a three-rail, taper-ﬂat design (MR-read, inductive-write). These sliders

generally use Al

–TiC (70–30 wt%) as the substrate material with multilayered thin-ﬁlm head structure.

In the MR head design, the ABS and the MR structure are generally coated with about 10-nm-thick DLC

coating. A nanoslider commonly used today has the dimensions of 2 × 1.6 × 0.425 mm with a mass of

about 7.7 mg, Table 1.5. A normal load of about 3.5 g is applied during use. Future trends are to use

smaller drives with ultrasmall sliders and an ultrasmooth (1 to 2 nm rms) and ﬂat disk substrate (made

of glass or glass ceramic) with about 5- to 10-nm-thick DLC overcoat and 0.5- to 2-nm-thick bonded

perﬂuoropolyether lubricant.

Horizontal thin-ﬁlm heads with a single-crystal silicon substrate referred to as silicon planar head

(SPH) sliders are under development which can be mass produced inexpensively and miniaturized using

silicon integrated-circuit technology, Figure 1.55 (Lazzari and Deroux-Dauphin, 1989; Bhushan et al.,

1992). Several integrated suspension/head devices have been proposed. For example, Fujitsu’s integrated

suspension with a so-called GUPPY nanoslider attached to an MR head, Figure 1.56 (Ohwe et al., 1993).

Suspension design is a miniaturized monolithic ﬂexible body with a load beam and a gimbal. The

integrated suspension is 25-µm thick, 10-mm long, 4-mm wide, and the beam width of the gimbal is

about 200 µm.

Various contact recording devices for ultimate recording densities are at various stages of development.

Censtor Corporation in San Jose, CA is developing an integrated head/ﬂexure/conductor structure called

a microﬂex head which is substantially smaller than the conventional head (Hamilton et al., 1991). The

microﬂex head is supposed to be in continuous sliding contact with the disk surface during use. A single-

pole perpendicular-probe-type transducer is used so as to tolerate several microns of pole-tip wear

without signiﬁcant signal degradation. In order to achieve low stiction and low head wear, the slider

suspension mass, the footprint of the contact pad, and the normal load are reduced by two to three

orders of magnitude from current designs. In the microﬂex head design, a read–write transducer is

fabricated at one end of a rectangular (30-µm-thick) sputtered amorphous alumina ﬂexure beam con-

taining two parallel copper leads connected to the read–write element at one end and bonding pads at

FIGURE 1.51 Tape path in an IBM 3490 data-processing tape drive.

the other end, Figure 1.57. The length, width, and thickness of the initial conﬁgurations of the ﬂexure

beam are 10 mm, 500 µm, and 30 µm, respectively, the spring constant is about 0.6 N/m, and the mass

is about 600 µg. A perpendicular-probe-type head with a layered pancake coil design and ﬂux return

pole is fabricated in the end of the beam. The main pole and yoke reside on the vertical end of the beam.

The pole tip (about 7 µm in height) is encapsulated in the wear-resistant amorphous hydrogenated carbon

material, which forms a contact pad whose nominal length, width, and depth (height) are 40, 20, and

7 µm, respectively. This conﬁguration provides a large tolerance for head wear and negligible air-bearing

lift of the ﬂexure beam. The pole material is an ion-beam-sputtered Co–Nb–Zr alloy of about 275-nm

thickness, surrounded on both sides by about 200 nm of rf-sputtered alumina. There is also an rf-

sputtered SiC adhesion layer of about 70-nm thickness between the rf-PECVD-deposited amorphous

hydrogenated carbon wear pad and the alumina. On the side nearest the free end of the beam, there is

an rf-sputtered Si adhesion layer of about 100 nm between the alumina and the carbon. The progression

FIGURE 1.52 (A) Sectional views of particulate and thin-ﬁlm magnetic tapes and (B) schematic of a 36-track

magnetic thin-ﬁlm read–write head (with a radius of cylindrical contour of about 20 mm) for an IBM 3490 tape drive.

FIGURE 1.53 Schematic of a data-processing rigid-disk

drive.

of layers is carbon, SiC, alumina, Co–Nb–Zr, alumina, Si, and carbon, going from the ﬁxed to free end

of the beam. Typical normal load applied is about 40 mg. We note that the microﬂex head consists of

nongimbaled stiff suspension, which requires appropriate mounting.

These small structures are fabricated by photolithography and etching on a host substrate. In a

departure from convention, thin-ﬁlm fabrication is accomplished on two orthogonal surfaces. The

magnetic core, coil, conductors, and ﬂux return pole are constructed within the dielectric ﬂexure on the

host wafer. The wafer is then sliced into bars and the pole/yoke structures are fabricated on the cut surface

of the bar. The ﬁnished structures are chemically released from the host substrate, the entire process

FIGURE 1.54 (A) Sectional views of particulate and thin-ﬁlm rigid disks and (B) schematic of a thin-ﬁlm slider

with a three-rail, taper-ﬂat design (arrow indicates the disk rotation).

TABLE 1.5 Physical Dimensions of Rigid-Disk Head Sliders

Typical Dimensions

L × W × H Rail Width Volume Mass Load

Slider Type (mm) (mm) (mm

) (mg) (g)

Regular slider 4.00 × 3.20 × 0.850 420 10.88 50 9.5

Microslider (70%) 2.80 × 2.24 × 0.595 330 3.73 17 6.0

Nanoslider (50%) 2.00 × 1.60 × 0.425 or 250 1.36 6 3.5

(small slider) 2.50 × 1.70 × 0.425 1.81 8

Under development

Picoslider (35%) L = 1.25 mm (ﬂying slider) 1

L = 1.00 mm (contact slider)

Femtoslider (30%)

Attoslider (10%)

Note: L = length, W = width, and H = height.

holding the potential to all but eliminate the machining process. The integrated ﬂexure/head almost

looks like an AFM cantilever, except the dimensions are much larger.

1.4.2 MEMS

In recent years, the emerging ﬁeld of MEMS has received increasing attention. MEMS are fabricated

using the technology developed for integrated circuits. The total size of the component can be as small

as 100 µm or even smaller. Microsensors, ﬁne-polishing microactuators, microgrippers, micromotors,

micropumps, gear trains, cranks, manipulators, nozzles, and valves a fraction of a millimeter in size are

being fabricated in laboratories around the world (Howe, 1988; Tai and Muller, 1989; Fan et al., 1989;

Tang et al., 1990; Bhushan and Venkatesan, 1993; Miu et al., 1993; Bhushan, 1996b, 1998b; Bhushan et al.,.

1997a,b; Sundararajan and Bhushan, 1998). Figure 1.58 shows an example of a surface micromachined

structure — a micromotor-driven electrostatically (by electrostatic attraction between positive and neg-

atively charged surfaces) which has been fabricated at MIT and UC Berkeley. The UC Berkeley micro-

motor consists of 12 stators and a four-pole rotor of 120-µm diameter with an air gap between the rotor

and stator of 2 µm. Structural material for the hub, stator, and rotor is polysilicon ﬁlm deposited by

LPCVD. The vertical walls of the stator and rotor contain 340-nm-thick LPCVD silicon nitride ﬁlm for

low friction and wear at high operating speeds, Figure 1.44. The motor built at MIT delivered a torque

of 12 pNm and synchronous motor speeds of up to 2500 rpm. The wear resistance of rotor/stator interface

and reduction of friction at the hub/rotor interface is critical for high-speed interfaces.

FIGURE 1.55 (a) Isometric view (not drawn to scale) and (b) cross section of the silicon planar head slider. (From

Bhushan, B. et al. (1992), IEEE Trans. Magn., 28, 2874–2876. With permission.)