Bhushan B. Nanotribology and Nanomechanics: An Introduction

Подождите немного. Документ загружается.

1274 Bharat Bhushan

Torsion hinge

Mirror

Mirror array

Yoke and

hinge array

Metal array

AFM images

of various

arrays

Yoke

Via 2 contact

to CMOS

Landing site

CMOS

memory

Spring tip

Mirror

address

electrode

Yoke

address

electrode

Bias reset bus

DMD exploded pixel view

14 μm

Fig. 22.53. Exploded view

of a DMD pixel and AFM

surface-height images of

various arrays. The DMD

layers were removed by an

ultrasonic method [192]

tion of micromirrors; these are also sometimes used for the construction of hinges,

spring tips, and landing sites. The aluminium alloy ﬁlms are overwhelmingly com-

prised of aluminium; trace elements (including Ti and Si) are present to suppress

contact spiking and electromigration, which may occur if current densities become

high during electrostatic operation. Multilayered sputtered SiO

TiN/Al alloy ﬁlms

are now generallyused for the landingsite structure to minimize refractionthrough-

out the visible region of the electromagnetic spectrum in order to increase the con-

trast ratio in projection display systems [193, 194]. These multilayered ﬁlms are

also generally used for hinges and spring tips. A low-surface-energy SAM is main-

tained on the surfaces of the DMD, which is packaged in a hermetic environment

to minimize stiction during contact between the spring tip and the landing site. An

SAM of perﬂuorinated n-alkanoic acid (C

2n−1

H) (e.g., perﬂurordecanoic acid

or PFDA, CF

(CF

)

COOH) applied by the vapor-phase deposition process is used.

A getter strip of PFDA is included inside the hermetically sealed enclosure contain-

22 Characterization of MEMS/NEMS and BioMEMS/BioNEMS 1275

ing the chip, which acts as a reservoir in order to maintain a PFDA vapor within

the package.

In order to identify a stuck mirror and characterize its nanotribological proper-

ties, the chip was scanned using an AFM [192]). It was found that it is hard to tilt

the stuck micromirror back to its normal position by adding a normal load at the

rotatablecorner of the micromirror;thus, this is called a hard stuck micromirror. An

exampleof a stuck micromirroris shownin Fig. 22.54a.Once the stuck micromirror

was found, the region was repeatedly scanned at a large normal load, up to 300nN.

After severalscans, thestuck micromirrorwasremoved.Once the stuckmicromirror

was removed, the surrounding micromirrors could also be removed by continuous

scanning under a large normal load (Fig. 22.54a bottom row). The adhesive force of

the landing site underneath the stuck micromirror and the normal micromirror are

presented in Fig. 22.54b, which clearly indicates that the landing site underneath

the stuck micromirrorhas much larger adhesion. A 1 µm×1 µm view of the landing

sites under stuck and normal micromirrors are also shown in Fig. 22.54b. The land-

Normal mirrors

40.0

μm

30.020.010.00

Stuck mirror (in the middle)

40.0

μm

30.020.010.00

0 2000 nm

Stuck mirror removed by AFM

40.0

μm

30.020.010.00

Highpass filtered image

40.0

μm

30.020.010.00

0 5000 nm

1.000.750.500.250

0 100 nm

1.000.750.500.250

0 50 nm

Adhesive force (nN)

Landing site

Normal II

200

150

100

Normal IStuck

a) b)

Fig. 22.54. (a)Thetop row shows AFM surface-height images of a stuck micromirror sur-

rounded by eight normal micromirrors. The left image in the bottom row shows the stuck

micromirror, which was removed by an AFM tip after repeated scanning at high normal

load. The right image in the bottom row presents a high-pass-ﬁltered image showing that the

residual hinge that sits underneath the removed micromirror is clearly observed. (b)AFM

surface-height images and adhesive forces of the landing sites underneath the two normal

micromirrors and the stuck micromirror [192]

1276 Bharat Bhushan

ing site under the stuck micromirror has an apparent U-shaped wear mark, which is

surrounded by a smeared area.

Liu and Bhushan[192] calculatedcontact stresses to examineif the stresses were

high enough to cause wear at the spring tip–landing site interface. The calculated

contact stress value was about 33MPa which is substantially lower than the hard-

ness, therefore much plastic deformation and consequently wear was not expected.

The wear mark was only found on a very few landing sites on the DMD, which

means that the SAM coating can generally endure such high contact stresses. Based

on data reported in the literature, the coverage of vapor-depositedSAMs is expected

to be about 97%. The bond strength of the molecules close to the boundary of the

uncovered sites is expected to be weak. Thus, the uncovered sites and the adjacent

molecules are referred to as defects in the SAM coating. Occasionally, if contact

occurs at the defect sites, the large cyclic stress may be close to the critical load,

and lead to the initial delamination of the SAM coating at the interface. The con-

tinuous contact leads to the formation of a high-surface-energysurface by exposure

of the fresh substrate and the formation of SAM fragments. This eventually leads

to an increase in stiction by the formation of large menisci. Once this happens, the

stress at the contact area is increased, which would accelerate the wear. Based on

this hypothesis, suggested mechanisms for the wear and stiction of the landing site

are summarized in Fig. 22.55. Wear initiates at the defect sites and consequent high

stiction can result in high wear. Improvingthe coverageand wear resistance of SAM

coatings could enhance the yield of DMD.

In some cases, the micromirrors are not fully stuck and can be moved by ap-

plying a load at the rotatable corner of the micromirror with a discontinuous mo-

tion, which is thus called soft stiction. Soft-stuck micromirrors studied by Liu and

Bhushan [195] were identiﬁed in quality inspection. These micromirrors encoun-

tered slow transition from to end to the other end (+1/ −1). Figure 22.56 shows the

AFM surface-height images of a location showing a stuck mirror (S) and surround-

ing normal micromirrors N

(i = 1, 2, and 3). Surprisingly, the images of the stuck

and normal micromirror array are almost the same. On the micromirrors of interest,

a tilting test was performed at the corner of the micromirrors, which are marked by

white dots in Fig. 22.56. The rotatable direction of the microarray is indicated by

an arrow bar in Fig. 22.56. The load–displacement curve for the stuck micromirror

Defects exist close

to the uncovered sites

Wear could initiate if

contact occurs at uncovered

or defects sites

SAM delaminated

from the interface

Formation of high energy surface

increases the water adsorption,

which in turn leads to large adhesion

Assembled

molecules

Defects

in SAM

Uncovered

sites

Residual

head groups

Molecular

fragments

Water

molecule

Fig. 22.55. Suggested mechanisms for wear and stiction [192]

22 Characterization of MEMS/NEMS and BioMEMS/BioNEMS 1277

40.030.020.010.00 5.02.50

0 1000 nm

N2 N3

Fig. 22.56. AFM surface-

height images of normal

micromirrors and a soft-stuck

micromirror. The soft-stuck

micromirror was labeled S,

and the normal micromirrors

studied are labeled as N1, N2,

and N3 [195]

is presented in Fig. 22.57; it is not smooth and appears serrated. It is clearly indi-

cated that, although the S micromirror can be rotated, it rotates with hesitation. In

regimes 1 and 2, as marked in Fig. 22.57, the slopes are much higher. In order to

understandthe mechanismsfor the occurrenceof stiction, stiction of landingsites of

normal and stuck mirrors were measured. Unlike hard-stuck mirrors, the adhesive

forces of soft-stuck and normal mirrors were comparable, which suggests that the

SAM coating is intact with stuck mirror. It was found that a high normalload (about

900nN) and on the order of a couple of hundred scans were required to remove

the soft-stuck micromirrorsby an AFM. However, only about 300nN and about ten

scans were required to remove a hard-stuck mirror. After careful examination of the

AFM images of the micromirror sidewalls in Fig. 22.56 (bottom left), it is noted

that there are contaminant particles attached to the sidewalls of the S mirror. It is,

therefore, believed that, during the tilting test, the S micromirror (see schematic

in Fig. 22.57) a sharper slope regime will occur in the displacement curve. Extra

force is required to overcome the resistance that is induced by the sidewall con-

tamination particles. This is believed to be the reason for the slow transition of the

micromirror during quality inspection.

Finally, the nanomechanical characterization of various layers used in the con-

struction of landing sites, hinge and micromirror materials have been measured by

Wei et al. [193, 194]. Bending and fatigue studies of the hinge have been carried

out by Liu and Bhushan [196] and Bhushan and Liu [197] to measure stiﬀness

and fatigue properties. For these studies, the micromirror was removed. During re-

moval, the micromirror/yoke structure was removed simultaneously, leaving the

hinge mounted on one end of the array; see Fig. 22.58. The stiﬀness of the Al hinge

was reported to be comparable to the stiﬀness of bulk Al. The Al hinge exhibited

a higher modulus than the SiO

hinge. The fatigue properties depended upon the

preparation of the hinge for testing.

1278 Bharat Bhushan

Normal load (nN)

Micromirror displacement (nm)

100

150

200 400 1000600 800

3D view

A-A view

Top view

N1 S

N2 N3

S rotatable corner

Regime 1 Regime 2

Fig. 22.57. Load–displace-

ment curve obtained on the

rotatable corner of the S mi-

cromirror and schematic to

illustrate the suggested mech-

anism for the occurrence of

soft stiction

AFM cantilever

Hinge

Residual hinge (yoke is removed)

Fig. 22.58. AFM surface-

height image of the residual

hinge and schematic diagram

of the relative position of the

hinge and AFM tip during

the nanoscale bending and

fatigue tests. The tip is lo-

cated at the free end of the

hinge [196]

22.7 Conclusion

The ﬁeld of MEMS/NEMS has expanded considerably over the last decade. The

length scale and large surface-to-volume ratio of the devices result in very high

retarding forces such as adhesion and friction, which seriously undermine the

performance and reliability of the devices. These tribological phenomena need

22 Characterization of MEMS/NEMS and BioMEMS/BioNEMS 1279

to be studied and understood at the micro- to nanoscales. In addition, materials

for MEMS/NEMS must exhibit good microscale tribological properties. There

is a need to develop lubricants and identify lubrication methods that are suit-

able for MEMS/NEMS. Using AFM-based techniques, researchers have conducted

micro/nanotribologicalstudiesof materialsand lubricantsfor usein MEMS/NEMS.

In addition, component-level testing has also been carried out to aid understanding

of the observed tribological phenomena in MEMS/NEMS.

Macroscale and microscale tribological studies of silicon and polysilicon ﬁlms

have been performed. The eﬀect of doping and oxide ﬁlms and environment on

the tribological properties of these popular MEMS/NEMS materials have also

been studied. SiC ﬁlm is found to be a good tribological material for use in high-

temperature MEMS/NEMS devices. Perﬂuoroalkylself-assembled monolayers and

bonded perﬂuoropolyether lubricants appear to be well suited for lubrication of mi-

crodevices under a range of environmental conditions. DLC coatings can also be

used for low friction and wear. Adhesion of biomolecules on Si surfaces can be

improved by nanopatterning and the chemical linker method. Roughness should

be optimized for superhydrophobicity low adhesion and friction. Surface rough-

ness measurements of micromachined polysilicon surfaces have been made using

an AFM. The roughness distribution on surfaces is strongly dependent upon the

fabrication process. Adhesion and friction of microstructures can be measured us-

ing a novel microtriboapparatus. Adhesion and friction measurements of silicon

on silicon conﬁrm AFM measurements that hexadecane thiol and bonded perﬂu-

oropolyether ﬁlms exhibit superior adhesion and friction properties. Static friction

forcemeasurementsof micromotorshavebeen performedusing anAFM. Theforces

are found to vary considerably with humidity. A bonded layer of perﬂuoropolyether

lubricant is found to satisfactorily reduce the friction forces in the micromotor. Tri-

bological failure modes of digital micromirror devices are either hard stiction or

soft stiction. In hard stiction, the tip on the yoke remains stuck to the landing site

underneath. The mechanism responsible for the hard stiction is localized damage to

the SAM on the landing site. However, in soft stiction, the mirror–yoke assembly

rotates with hesitation. The mechanism responsible for soft stiction is contaminant

particles present on the mirror sidewalls.

AFM/FFM-based techniques show the capability to study and evaluate micro/

nanoscale tribological phenomena related to MEMS/NEMS devices.

22.A Appendix Micro/Nanofabrication Methods

22.A.1 Top-Down Methods

The top-downfabrication methods used in the construction of MEMS include litho-

graphic and non-lithographictechniques to produce micro- and nanostructures. The

lithographic techniques fall into three basic categories: bulk micromachining, sur-

face micromachining, and LIGA (a German acronym for Lithographie Galvanofor-

mung Abformung), a German term for lithography, electroplating, and molding.

1280 Bharat Bhushan

Bulk micromachining

Deposition of silica layers on Si

Membrane

111 face

Patterning with mask and

etching of Si to produce cavity

Silicon

Silica

Surface micromachining

Deposition of sacrificial layer

Patterning with mask

Deposition of microstructure layer

Etching of sacrificial layer to produce freestanding structure

Silicon Polysilicon Sacrificial

material

Fig. 22.59. Schematic of the

process steps involved in bulk

micromachining and surface-

micromachining fabrication

of MEMS

The ﬁrst two approaches, bulk and surface micromachining, mostly use planar pho-

tolithographic fabrication processes developed for semiconductor devices in pro-

ducing two-dimensional (2D) structures [13, 19, 198, 199]. The various steps in-

volved in these two fabrication processes are shown schematically in Fig. 22.59.

Bulk micromachining employs anisotropic etching to remove sections through the

thickness of a single-crystal silicon wafer, typically 250–500µm thick. Bulk micro-

machining is a proven high-volume production process and is routinely used to fab-

ricate microstructures such as accelerometers,pressure sensors, and ﬂow sensors. In

surface micromachining, structural and sacriﬁcial ﬁlms are alternatively deposited,

22 Characterization of MEMS/NEMS and BioMEMS/BioNEMS 1281

patterned, and etched to produce a freestanding structure. These ﬁlms are typically

made of low-pressure chemical vapor deposition (LPCVD) polysilicon ﬁlm with

a thickness of 2–20µm. Surface micromachining is used to produce sensors, actua-

tors, micromirror arrays, motors, gears, and grippers. The resolution in photolithog-

raphy is dependent upon the wavelength of light. A commonly used light source

is an argon ﬂuoride excimer laser with a wavelength of 193nm (ultraviolet or UV)

used in patterning 90-nm lines and spaces. Deep-UV wavelengths, X-ray lithogra-

phy, electron beam (e-beam) lithography, focused ion-beam lithography, maskless

lithography, liquid-immersion lithography, and STM writing by removing material

atom by atom are some of the recent developments for sub-100-nm patterning.

The fabrication of nanostructures such as nanochannels with sub-10-nm reso-

lution can be accomplished through several routes: electron beam lithography and

sacriﬁcial-layer lithography (SLL). The process for e-beam lithographic technique

is a ﬁnely focused electron beam that is exposed over a resist surface, where the

exposure duration and location is controlled with the use of a computer [200,201].

When the resist is exposed to the electron beam, the electrons either break or join

the molecules in the resist, so the local characteristicsare changed in such a way that

further processes can either remove the exposed part (positive resist) or remove the

unexposedpart (negative resist). The resist material determinesif the molecules will

either break or join together and thus determines if a positive or negative image is

produced. E-beam lithography can either be used to create photolithographic masks

for replication or to create the devices directly. The masks that are created can be

used for either optical or X-ray lithography. One limitation of e-beam lithography is

that throughput is drastically reduced since a single electron beam is used to create

the entire exposurepattern on the resist. While this technique is slower than conven-

tional lithographic techniques, it is ideal for prototype fabrication because no masks

are required.

In SLL process, the use of a sacriﬁcial layer allows the direct control of

nanochannel dimensions as long as there exists a method for removing the sacri-

ﬁcial layer with absolute selectivity to the structural layers. A materials system with

such selectivity is the silicon/silicon oxide system used widely in the microfabrica-

tion of MEMS devices. The use of sidewall deposition of the sacriﬁcial layer and

subsequent etching allows the fabrication of high-densitynanochannels for biomed-

ical applications. It is based on surface micromachining [77]. Figure 22.60 shows

a schematic of the process steps in sacriﬁcial-layer lithography based on Hansford

et al.’s [77] work on fabrication of polysilicon membranes with nanochannels. As

with all the membrane protocols, the ﬁrst step in the fabrication is the etching of the

support ridge structure into the bulk silicon substrate. A low-stress silicon nitride

(LSN or simply nitride), which functions as an etch stop layer, is then deposited

using LPCVD. The base structural polysilicon layer (base layer) is deposited on top

of the etch stop layer. The plasma etching of holes in the base layer is what deﬁnes

the shape of the pores. The buried nitride acts as an etch stop for the plasma etching

of a polysilicon base layer. After the pore holes are etched through the base layer,

the pore sacriﬁcial thermal oxide layeris grown on the base layer. The basic require-

1282 Bharat Bhushan

Polysilicon

Si (100)

Thermal silicon oxide

Silicon nitride

2nd Polysilicon

Silicon nanochannels

Fig. 22.60. Schematic of

process steps involved in

sacriﬁcial-layer lithography:

(a) Growth of silicon nitride

layer (etch stop) and base po-

lysilicon deposition, (b) hole

deﬁnition in base, (c)growth

of thin sacriﬁcial oxide and

patterning of anchor points,

(d) deposition of plug poly-

silicon, (e) planarization of

plug layer, and (f) deposition

and patterning of the pro-

tective nitride layer through

etch, followed by etching of

protective, sacriﬁcial and etch

layers before ﬁnal release of

the structure in HF [77]

ment of the sacriﬁcial layer is the ability to control the thickness with high precision

across the entire wafer. Anchor points are deﬁned in the sacriﬁcial oxide layer to

mechanically connect the base layer with the plug layer (necessary to maintain the

pore spacing between layers). This is accomplished by using the same mask shifted

from the pore holes. This produces anchors in one or two corners of each pore hole,

which provide the desired connection between the structural layers while opening

as much pore area as possible. After the anchor points are etched through the sac-

riﬁcial oxide, the plug polysilicon layer is deposited (using LPCVD) to ﬁll in the

holes. To open the pores at the surface, the plug layer is planarized using chemical

mechanical polishing (CMP) down to the base layer, leaving the ﬁnal structure with

22 Characterization of MEMS/NEMS and BioMEMS/BioNEMS 1283

the plug layer only in the pore hole openings. As the silicon wafer is ready for re-

lease, a protective nitride layer is deposited on the wafer (completely covering both

sides of the wafer). The back-side etch windows are etched in the protective layer,

exposing the silicon wafer in the desired areas, and the wafer is placed in a KOH

bath to etch. After the silicon wafer is completely removed up to the membrane

(as evidenced by the smooth buried etch-stop layer), the protective, sacriﬁcial, and

etch-stop layers are removed by etching in concentrated HF. Etching of sacriﬁcial

layer in polysilicon ﬁlm deﬁnes nanochannels.

The LIGA process is based on the combined use of X-ray lithography, electro-

plating, and molding processes. The steps involved in the LIGA process are shown

schematically in Fig. 22.61. LIGA is used to produce high-aspect-ratio MEMS

(HARMEMS) devicesthat are up to 1mm in height and only a few microns in width

or length [202]. The LIGA process yields very sturdy 3D structures due to their in-

creased thickness. One of the limitations of silicon microfabrication processes orig-

inally used for fabrication of MEMS devices is the lack of suitable materials that

can be processed. With LIGA, a variety of non-silicon materials such as metals,

ceramics and polymers can be processed. Non-lithographic micromachining pro-

cesses, primarily in Europe and Japan, are also being used for the fabrication of

millimeter-scale devices using direct material microcutting or micromechanicalma-

LIGA

Resist

structure

Base

plate

Lithography

Electroforming

Metal

structure

Gate plate

Mold

insert

Mold fabrication

Molding

mass

Mold filling

Plastic

structure

Unmolding

Fig. 22.61. Schematic of the

process steps involved in

LIGA fabrication of MEMS