Ghodssi R., Lin P., MEMS Materials and Processes Handbook

Подождите немного. Документ загружается.

714 D.R. Hines et al.

Fig. 9.39 Schematic illustration of LCVD process (Reprinted with permission from [136])

Fig. 9.40 Three mm tall replica of the Eiffel Tower fabricated using LCVD (Reprinted with

permission from [136])

the laser is moved. Figure 9.40 shows a 3 mm tall replica of the Eiffel Tower fabri-

cated using LCVD to deposit Al features onto a preformed polycarbonate structure.

9.5.6 Stereolithography and Microstereolithography

Stereolithography (SL), sometimes referred to as rapid prototyping, was ﬁrst intro-

duced in 1981 [137–139]. In its simplest form it employs a laser source, a vat of

photocurable polymer, a sample holder, and a means of moving the beam relative to

the liquid polymer. A CAD-generated design is used to control the relative motion

between the laser and the vat causing photopolymerization where the laser interacts

with the liquid polymer. This mapping/polymerization produces a solid structure

deﬁned by the CAD control and held in place by the sample holder. The process is

shown schematically in Fig. 9.41.

The process can also be performed with solutions loaded with photocurable

metallic or ceramic suspensions. In this way, metal or ceramic parts can be fab-

ricated. If the photocurable solution is replaced by a powder, 3-D structures can be

fabricated by laser-induced sintering.

9 MEMS Lithography and Micromachining Techniques 715

Fig. 9.41 Schematic illustration of stereolithography system (Reproduced with permission from

Microstereolithography (MSL) is an extension of stereolithography to the micron

scale. Instead of employing a laser spot size of around 200 µm, the laser spot is

focused down to 1–2 µm. Also the parts are small enough to be self-supporting and

so the support structure generally needed for SL-formed parts may not be needed for

MSL. Photopolymerization happens on the surface of the vat as the beam is scanned

in 2-D (x−y). Once the desired layer is formed, it is lowered to allow fresh liquid to

cover the previously formed structure and a new layer is polymerized on top of the

existing feature. In this way a 3-D structure is generated layer by layer. For SL the

polymerized layer thickness i s on the order of hundreds of µm’s thick whereas for

MSL it is on the order of microns thick.

Two main types of MSL exist, scanning and projection, of which the scanning

mode is most widely used. The size and shape of structures fabricated using a sin-

gle beam of UV light is limited by the physical size of the focused beam and the

fact that the polymerization occurs at the surface of the vat as the beam (or vat) is

scanned in 2-D. The scanning aspect of SL and MSL can introduce a speed lim-

itation for the fabrication of structures. Because polymerization takes place at the

surface of the vat, the UV light can be projected through a mask (as in projection

photolithography). In this case polymerization across the entire desired area of the

surface happens simultaneously thus removing the need for any scanning in the x−y

direction. Subsequent layers can be polymerized by lowering the substrate to allow

more uncured liquid to cover the surface and then repeat the exposure. The shape

of structures is limited by the number of different masks available and the ability

to align subsequent masks to previous polymerized layers. This limitation can be

overcome by using dynamic liquid crystal masks, or using a micromirror array to

project patterns. Figure 9.42 shows an example of features that can be fabricated

using MSL.

716 D.R. Hines et al.

Fig. 9.42 Illustration of various structures fabricated using MSL: (a) microchannels, (b) gear, and

A limitation of single-photon processing is that polymerization occurs at the sur-

face of the vat. This can be overcome by using a two-photon process. The process

employs two intersecting i nfrared (IR) beams. At the point of intersection, the liq-

uid polymer can polymerize only by simultaneous absorption of two photons. This

produces a spot size below the diffraction limit of a single beam and allows the

polymerization to take place anywhere within the vat, not just at the surface. This

also allows for small structures that are truly 3-D. The two-photon absorption pro-

cess can take place either as a resonant excitation or a simultaneous absorption. In

the resonant-excitation mode the ﬁrst photon is absorbed to create a real excited

state with a lifetime on the order of 10

−4

− 10

−9

s. Polymerization can occur only

if a second photon is absorbed by the monomer while in the excited state. In the

simultaneous absorption process, no intermediate excited state exists. Therefore

both photons must be absorbed within 10

−15

s of each other [142].

The two-photon absorption process is a nonlinear effect with polymerization

dependent on the square of the photon density. This phenomenon is responsible for

a resolution better than the diffraction limit of the single-beam proﬁle. Figure 9.43

shows examples of features fabricated using two-photon MSL. Note that whereas

the features in Fig. 9.42 are in the hundreds of microns, these structures are on the

micron scale with submicron details!

9.6 Print/Imprint Lithography

Printing is an ancient art dating back to the eighth and ninth centuries A.D. An

example of such a printed work is the Diamond Sutra which was fabricated using

a technique referred to as block printing. Printing using movable type based on

clay [147] was invented in 1041 and based on metal [148] in 1232. Ushering in

a more modern style of printing, Gutenberg is credited with having invented the

9 MEMS Lithography and Micromachining Techniques 717

Fig. 9.43 3-D microstructures produced by multiphoton microfabrication: (a) microbull model

[143]; (b) Venus model [144]; (c) microturbine [145]; (d) nanotweezers [146] (Reproduced with

permission; copyright Wiley-VCH)

modern printing press in the 1450s. Lithography, which is a form of printing based

on a planographic process that relies on patterning inks onto a ﬂat surface using

hydrophilic and hydrophobic interactions, was invented in 1798.

The semiconductor industry has developed a form of pattern transfer for the fab-

rication of solid-state electronics based on Si called photolithography. As described

in Section 9.2 above, photoresist is applied to and patterned on Si wafer substrates

in the manufacturing of computer chips and on glass substrates in the manufacturing

of ﬂat-panel displays. Either way, the fabrication methods are limited to batch pro-

cessing of ﬁnite size substrates. Serious efforts are underway to develop larger-scale

manufacturing processes that can fabricate electronic devices in a continuous fash-

ion in a manner more similar to the fabrication of magazines or newspapers. In order

to accomplish this, the added requirement of ﬂexible substrates will be needed.

As a testimony to the importance of such efforts, Xia and Whitesides [20] list no

fewer than 18 different nonphotolithographic patterning methods being developed

to fabricate devices at the micrometer and nanometer length scales. The table is

reproduced here as Table 9.5. The earliest reference dates back to 1974, with a

majority of the references coming from the 1990s. Even though printing is an old

technique with a rich history in fabricating and/or reproducing images and writing,

it is being revisited as a method for fabricating active electronic devices, circuits,

and MEMS components.

718 D.R. Hines et al.

Table 9.5 Nonphotolithographic methods for micro- and nanofabrication

Method Resolution

References

Injection molding 10 nm [149–154]

Embossing (imprint) 25 nm [155–163]

Cast molding 50 nm [164–166]

Laser ablation 70 nm [167–172]

Micromachining with a sharp stylus 100 nm [173]

Laser-induced deposition 1 µm[174–176]

Electrochemical micromachining 1 µm[177]

Silver halide photography 5 µm[178–180]

Pad printing 20 µm[181]

Screen printing 20 µm[182]

Inkjet printing 50 µm[24, 183–186]

Electrophotography (xerography) 50 µm[187, 188]

Stereolithography 100 µm[189–192]

Soft lithography [18, 21, 193, 194]

Microcontact printing (µCP) 35 nm [195, 196]

Replica molding (REM) 30 nm [197]

Microtransfer molding (µTM) 1 µm[198]

Micromolding in capillaries (MIMIC) 1 µm[199]

Solvent-assisted micromolding (SAMIM) 60 nm [200]

These numbers do not represent ultimate limits

The lateral dimension of the smallest feature that has been generated

Various types of printing techniques are reviewed here as they pertain to the pat-

terning and/or assembly of components needed to fabricate active electronic devices

and MEMS components.

The printing industry has been primarily interested in reproducing pictures and/or

text on a page. Both are passive features composed of two components: an ink

and a substrate (the page). The ink is designed for passive color and high visual

contrast against the substrate. The substrate can be made from various materials

such as paper, plastic, metals, glass, and so on. For the printing of active devices

the substrate materials may not change much but the inks will have to be very

different.

Visual contrast and passive color will no longer be important properties.

Admittedly, transparency over certain optical ranges may be important for some

applications; however, active inks will primarily be chosen for their conductivity,

semiconducting properties, insulating properties, and mechanical properties among

others. Such inks will need to be prepared and processed in ways that optimize

these properties. This will most likely require techniques very different from those

used to prepare and process passive inks, therefore traditional printing techniques

are not discussed. Rather, the discussion is dedicated to introducing and review-

ing printing methods being developed that will meet the needs for printing active

materials.

9 MEMS Lithography and Micromachining Techniques 719

9.6.1 Inkjet Printing

Inkjet printing [22] is a digital form of printing that can be performed either in

a continuous or a drop-on-demand mode. For the continuous mode, a stream of

droplets passes through a set of signal drive electrodes. An electrical pulse to the

electrodes causes a droplet to be deﬂected to the substrate. Undeﬂected droplets are

collected and recirculated back to the ink reservoir. For the drop-on-demand modes,

a piezoelectric pulse is used to create a droplet by pushing ink through a nozzle.

Nozzles are typically 20–30 µm in diameter and create ink droplets of about 10–20

pL. The droplet-on-demand mode results in better printing quality when compared

to continuous mode inkjet printing with feature resolution on the order of 20–50 µm

for standard inkjet printing.

Viscosity and surface tension are important parameters of the ink. Ink must ﬁll the

nozzle in approximately 100 µs and not drip out between demands. This typically

requires a viscosity range of 2–100 cp and a minimum surface tension of about 35

mN/m. To prevent the ink from drying and clogging the nozzle, a liquid such as

ethylene glycol can be added at a level of 10–20%. If an ink contains particulates,

the particulates must remain uniformly suspended in the ink and not contribute to

clogging in the nozzle. This usually requires a particulate size less than 1 µm.

High-molecular-weight polymers tend to be too viscous to print easily using

inkjet printing. However, inkjet printing of waxy and low-molecular-weight poly-

mers is possible. Printed droplets can exhibit pinholes and can have a tendency to

bead up when printed onto nonabsorbent substrates (such as plastics). Also, edge

roughness and printing resolution can be affected by droplet-spreading character-

istics and can be difﬁcult to control. Some materials’ printing problems can be

overcome by incorporating additives into the ink that can be postprocessed, such as

the addition of UV curable polymers. Also precursor materials can be used directly

as inks that are postprocessed, such as precursor metal solutions.

A wide range of materials has been successfully inkjet printed. Calvert [22]

lists nanoparticle suspensions, sol-gels, conducting polymers, ceramic powders, sol-

der, DNA, and proteins as materials that have been inkjet printed, all of which

must either be suspended or dissolved. Some organic semiconductor materials

tend to be insoluble (pentacene, e.g.) and, therefore, soluble forms of these mate-

rials have been investigated for use in speciﬁc inkjet applications [201]. Other

organic semiconductor materials tend to be very sensitive to drying conditions such

as poly(3-hexylthiophene) (P3HT) [202]. In either case, inkjet printing of these

materials has, to date, typically resulted in low-quality ﬁlms.

There have been attempts to improve the resolution associated with inkjet print-

ing. One method used an undisclosed print head technology as part of a superﬁne

inkjet printing system [203]. This system was reported to have achieved a line width

of 3 µm. Also, Sirringhaus et al. have developed a droplet-on-droplet technique in

which the second droplet does not wet the ﬁrst droplet [204]. When printed, the

second droplet slides off the ﬁrst droplet, producing a sub-100 nm gap between

the printed features, which are then used as source/drain electrodes for an OTFT

device.

720 D.R. Hines et al.

9.6.2 Soft Lithography

Photolithography has been used for decades in the fabrication of inorganic elec-

tronics; however, it has several disadvantages that limit its usage for nontraditional

materials: it allows for very little variation in the chemistry of either the resist layer

or the substrate and it can only be performed on ﬂat surfaces in a batch-processing

mode.

In an attempt to go beyond the limitations of photolithography, soft lithographic

methods have been developed, as listed i n Table 9.5. Soft lithography encompasses

techniques such as microcontact printing [195, 196](µCP), replica molding [197]

(REM), microtransfer molding [198](µTM), micromolding in capillaries [199]

(MIMIC), and solvent-assisted micromolding [200] (SAMIM). These t echniques all

have in common a patterned elastomeric mold or stamp (typically made from poly-

dimethylsiloxane (PDMS)). Because the stamp (mold) is ﬂexible, these techniques

are compatible with nonﬂat surfaces.

Soft lithography techniques have also been used to print patterned layers from

PDMS stamps onto substrates. For example, Au evaporated onto the patterned sur-

face of a PDMS stamp and then treated with a thiol- terminated self-assembled

monolayer (SAM) has been printed onto a GaAs substrate [205]. Also PDMS fea-

tures have been printed onto Si substrates by treating the patterned surface of a

PDMS stamp with a ﬂuorinated SAM and then adding an extra layer of PDMS

[206]. When contacted to the Si substrate, the top layer of PDMS can be transferred

to the Si surface. Childs and Nuzzo [206] provide detailed procedures for preparing

PDMS ﬁlms and bonding such ﬁlms to Si substrates to produce both closed and

open patterns on the substrate. Below, a case study is provided for the preparation

of PDMS ﬁlms.

Unlike photolithography, which transfers a pattern using optical techniques, soft

lithography is a direct printing method. Also, the stamp (mold) can be replicated

from a reusable master template. Therefore, soft lithography has the potential to be

a less complex and less expensive method of patterning compared to photolithogra-

phy. In addition, soft lithographic techniques can be used to directly pattern a wider

range of surfaces comprised of polymers, sol-gels, biological materials, organic

thin-ﬁlms, and colloidal materials, in addition to inorganic surfaces.

9.6.3 Nanoimprint Lithography (NIL)

In addition to the limitations discussed above, pattern sizes produced by pho-

tolithography are limited by the resolution of the optical system used to transmit

light through the mask. Over decades of research, this resolution limit has been

steadily reduced at the expense of increased complexity and cost of the fabri-

cation equipment. Nanoimprint lithography (NIL) [156] has been developed as a

patterning technique that can simultaneously improve pattern resolution and reduce

both equipment complexity and cost. It is based on a rigid template containing a

9 MEMS Lithography and Micromachining Techniques 721

Fig. 9.44 (a) Nanoimprint lithography (NIL) and (b) the transfer printing method (Reprinted with

patterned surface. The template is pressed into a resist layer that has been coated

onto a substrate. Two types of NIL have been developed and are referred to as

hot and cold embossing. Hot embossing uses a thermoplastic resist layer and cold

embossing uses a UV-curable liquid layer. The general process is illustrated in

Fig. 9.44a.

In hot embossing, the patterned template surface is placed in contact with the

thermoplastic resist layer. The resist layer is heated up above its glass transition

temperature and the template is pressed into the resist layer. After a speciﬁed time

the resist layer is cooled down below its glass transition temperature and the pressure

is released. The template is then removed to reveal a replica of the template surface

contained in the surface of the resist layer.

In cold embossing, sometimes referred to as step-and-ﬂash imprint lithography

[207] (SFIL), the patterned template surface is placed in contact with a UV-curable

liquid layer. The resist layer is then cured by exposure to UV light. The template

722 D.R. Hines et al.

is then removed to reveal a replica of the template surface contained in the surface

of the resist layer. In this case, the template (or possibly the substrate) must be

transparent to UV light. This is similar to replica molding [197] which typically

uses a ﬂexible template, however, SFIL has been developed using a rigid t ransparent

template.

Several variations of hot embossing have been reported [208, 209] as a means

of reducing the operating pressure and/or temperature. One technique eliminates

the resist layer entirely [210]. NIL is also being expanded to include imprinting

into a variety of active materials as opposed to sacriﬁcial resists [211]. The reso-

lution of NIL has been demonstrated to be at the 10 nm level [212]. Most of the

developmental work associated with NIL is directed toward inorganic substrates

and is targeted as a disruptive technology competing against photolithography and

e-beam lithography. Schift [213] provides a review article that discusses a wide vari-

ety of processing issues and materials parameters associated with conventional NIL.

NIL techniques have begun to enter the realm of MEMS fabrication largely asso-

ciated with microﬂuidics applications incorporating small-diameter ﬂow channels.

Nanochannel fabrication for applications involving DNA stretching and sequencing,

protein separation, and drug delivery are discussed by Kim et al. within Chapter 13

in the book Unconventional Nanopatterning Techniques and Applications [15].

Many variations of NIL are being developed in order to address speciﬁc

applications that are not so easily accomplished by conventional techniques. For

example, rather than competing with photolithography, NIL can be combined with

photolithography to simultaneously pattern large and small features. Work on com-

bining nanoimprint and photolithography (CNP) [214, 215] uses a hybrid mold that

is at once both a photomask and an NIL template. In addition, large volume roll-

to-roll NIL (R2RNIL) has also been envisioned for high-speed manufacturing. A

body of work designed to integrate NIL fabrication methodologies into MEMS-

based systems is beginning to emerge in the literature. S. Park and H. Schift have

contributed an entire chapter discussing NIL technology for biological application

in a book devoted to bio-MEMS [216]. A variety of topics is presented such as

nanoﬂuidic devices, engineered nanopores, and nanopatterns deﬁned by variations

in surface chemistry or protein coverage. Applications such as battery and fuel

cell technologies can greatly beneﬁt from increased surface area coatings. Zhang

et al. have incorporated NIL processing methods into the design and fabrication of

MEMS-based fuel cells as a means for increasing reaction surface areas among fuel,

catalyst, and electrolytes [217]. X. Fan et al. has utilized NIL as a low-cost method

for the fabrication of PMMA microlens arrays on a Si substrate [218].

9.6.4 Transfer Printing

As is illustrated in Fig. 9.44b, the same processing method used for NIL can be

used to transfer a patterned layer (printable layer) from one substrate (the transfer

substrate) to a second substrate (the device substrate). The only requirement is that

9 MEMS Lithography and Micromachining Techniques 723

the printable layer adheres more strongly to the device substrate than to the transfer

substrate. This process is referred to as transfer printing. In general, the transfer

printing process does not rely on temperature, but only on contact of the printable

layer to the surface of the device substrate where an appropriate differential adhesion

exists between the two interfaces containing the surfaces of the printable layer. This

is the same governing principle seen in decals for detailing model cars and airplanes

or for a child’s temporary tattoos.

It is with such a process that we wish to develop the ability to fabricate ﬂexi-

ble electronics and MEMS devices. Plastic substrates present an obvious choice of

materials for this application, not only because they are ﬂexible, but also because

of their low density, optical clarity, low cost, compatibility with roll-to-roll process-

ing, and so on. The process is simple in concept but has the potential to allow a

wide variety of dissimilar materials to be combined onto a single substrate in ways

difﬁcult if not impossible to achieve with conventional methods. Figure 9.45 shows

examples of metallic, organic, and carbon-based materials that have been printed

onto a plastic substrate.

Figure 9.45a illustrates a 50 µm wide serpentine Au electrode printed onto a

Kapton substrate. The Au feature was fabricated on a Si wafer transfer substrate

using standard photolithography. Prior to printing, the transfer substrate was spin

coated with a l ayer of polyimide. The Au/polyimide bilayer was then printed onto

the Kapton substrate. The proﬁlometer scan shows that the printed Au feature pro-

trudes above the Kapton surface. In general, Au features from 200 nm wide lines up

to full 3 in. diameter ﬁlms have also been successfully printed onto PET substrates.

Figure 9.45b illustrates a 50 nm thick pentacene ﬁlm printed onto a PET sub-

strate. In t his example, a Pn ﬁlm was thermally evaporated onto a SiO

/Si wafer

transfer substrate. Prior to printing, an Au electrode was evaporated onto the transfer

substrate such that it also partially covered the Pn ﬁlm. The three resulting r egions

(Au, Au/Pn, and Pn) were all successfully transfer printed onto the PET substrate.

Pn has also been printed onto Au, PC, latex, nitrile, PVC, and PMMA surfaces, to

name a few examples. In addition, Pn ﬁlms can be patterned by printing against a

patterned photoresist ﬁlm.

Figure 9.45c shows an example of a CNT network printed onto a PET substrate

that contains previously printed Au electrodes. The CNT network was grown by

CVD onto a SiO

/Si wafer transfer substrate. Similarly, graphene has also been

printed onto Au and polymer surfaces.

The last example shown in Fig. 9.46 illustrates the ability to print inorganic fea-

tures directly onto a prepatterned plastic or elastomer substrate. Here a Si grid is

shown printed onto a patterned PC substrate. Others have used PDMS as an inter-

mediate substrate for the printing of Si features using a kinetic printing process

where the adhesion is controlled by the speed at which the intermediate substrate

is removed. For all these illustrations, adhesion to the ﬁrst substrate must be less

than adhesion to the ﬁnal substrate. Efforts to establish printing parameters for a

wide variety of materials have relied on the engineering of surface energies. Self-

assembled monolayers (SAMS) can provide an easy effective tool for establishing

the desired surface energy of a given substrate [222]. For example, the adhesion of