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

Подождите немного. Документ загружается.

214 E. Meng et al.

Fig. 4.17 (a) Layout with detail of a ﬂexible retinal prosthesis on a polyimide substrate. (b) Close-

up photograph of electrodes on the polyimide support freed by peeling off from a silicon carrier

wafer. (c) Fabrication process for the retinal prosthesis (PI 2611, HD Microsystems) (Original

• Si substrates (undercut using HF:HNO

(hydroﬂuoric acid:nitric acid) (1:1) etch)

[196]

• Oxide layers (dissolve in HF); if OH terminated (oxidized Si or Pyrex wafers),

then hot wafer followed by buffered HF [202, 235]

• Al (wet release with a mixture of phosphoric–acetic–nitric acids and water [215,

236, 237] or anodic dissolution in neutral sodium chloride [65, 238])

• Thick electroplated Cu (dissolve in ferric chloride) [218]

4 Additive Processes for Polymeric Materials 215

• Cr (HCl:H

O 1:1 etch) [207]

• Ti (removal in dilute HF) [239, 240]

• Thermal decomposition of polymers (polycarbonates [ 65] and polynorbonenes

[206]) (300

◦

C and 370–425

◦

C, respectively)

4.3.2.3 Bonding

Several bonding processes have been developed that utilize polyimide as an adhesive

layer:

• RF dielectric heating of spin-on polyimide ﬁlms to join Si pieces (1–4 bar

clamping pressure, RF level of 500 W at 14 MHz, and 165–180 V

rms

)[241]

• Electrostatic bonding of fully cured and chemical mechanical polished polyimide

(350

◦

C, 1 kg/cm

clamping pressure, and 100 V) (Fig. 4.18)[242]

Fig. 4.18 (a) Schematic views of a thermal inkjet printhead assembled by polyimide bonding. (b)

Photograph of assembled nozzle oriﬁce showing polyimide adhesive layer (Duramide 7520, Arch

4.3.3 Lessons Learned

As in many MEMS polymers, adequate adhesion of polyimide to various substrates

can be elusive. Although adhesion to Cr is excellent, adhesion to other substrates

(such as Si, Si-derivatives, Al, and Cu) requires silane adhesion promoters [243].

Reactive ion etching in an oxygen plasma roughens polyimide surfaces to enable

bonding to metals [244] and other polyimide layers [202].

Signiﬁcant thermally induced dimensional changes in polyimide structures

inevitably follow the imidization process. Shrinkage as high as 40–50% has been

reported [202]. Although some applications may resort to using uncured ﬁlms

to avoid thermal shrinkage, most processes must account for this effect in the

design. Below, a case study is presented that utilizes thermal shrinkage to fabricate

out-of-plane structures.

216 E. Meng et al.

4.3.4 Case Study

Thermal s hrinkage due to imidization can also be used to one’s advantage; hinge

structures take advantage of the shrinkage for out-of-plane and self-assembly

[245–247]. V-groove trenches were etched in Si and then ﬁlled with polyimide. This

forms a polyimide joint that undergoes thermal shrinkage upon curing and results in

the bending of attached structures (Fig. 4.19). By using multiple V-groove joints, it

is possible to create bends of up to 200

◦

. These polyimide joint actuators attached

to Si legs were assembled into microrobotic conveyors [247]. In contrast, out-of-

plane assembly of polyimide hinged structures was achieved manually [248, 249]

or electrostatically [250, 251].

Fig. 4.19 (a) Principle of forming a polyimide V-groove joint by virtue or thermal shrinkage.

(b) Out-of-plane assembly with a three V-groove joint (HTR-3 200, OCG) (Original ﬁgure from

[246], used with permission of Institute of Physics Publishing Ltd.)

4.4 Hydrogels

4.4.1 Gelatin

Gelatin is well known as a both an edible dessert and binder for photographic

ﬁlms. It is an organic polymer that is arises from thermal denaturation or phys-

ical and chemical degradation of collagen; but unlike collagen, it does not exist

naturally. The degradation process unfolds the triple helix structure characteristic of

collagen and results in dissociated, unraveled, and disordered polypeptide chains

[252, 253]. Cooling the gelatin allows partial recovery of the triple helix struc-

ture, albeit only in small regions, due to the random recombination of polypeptide

4 Additive Processes for Polymeric Materials 217

chains. At this point, gelatin becomes a thermoreversible gel [253–258]. Depending

on the nature of the solidiﬁcation process, gelatin can exhibit a wide range of

physicomechanical properties; this is linked to the conformation of its constituent

macromolecules (either collagen-like in helical format or having a coiled struc-

ture). Gelatin is less expensive than collagen but possesses s imilarities in its amino

acid composition, structure, and properties [259]. Gelatin also possesses unique

properties not found in collagen. It is edible, biodegradable, biocompatible, and

nonimmunogenic [253].

The two types that are available, A and B, are differentiated by the method in

which they are produced. Type A is produced by acid-treatment of the precursor

whereas in contrast, Type B involves an alkaline treatment. These types are further

categorized by the Bloom number which is a measure of the gelatin’s mechanical

strength. In the solid state, gelatin can be brittle at low humidity and high temper-

atures [259]. This property arises from its rigid chain structure. Gelatin is typically

hydrophilic and can change properties depending on its water content. It conve-

niently has a low melting point (<100

◦

C) [260] and is readily dissolved in warm

water.

Gelatin is rendered photosensitive by the addition of a suitable photosensitizer.

Examples of photosensitizers include potassium dichromate (K

) and ammo-

nium dichromate ((NH

)

)[260, 261]. Photosensitized gelatin is a negative

tone resist that cross-links after exposure to UV; uncross-linked gelatin is simply

removed by dissolution in warm water (∼50

◦

C). Gelatin solutions are spin pro-

cessed up to concentrations of ∼10% after which point an infrared heater is required

[260]. Cross-linking imparts resistance to acids and bases. Gelatin is also cross-

linked by applying glutaraldehyde (Fig. 4.20)[262–264]. Glutaraldehyde, however,

is cytotoxic [258, 265, 266]. Other cross-linking agents, some less toxic, include

formaldehyde, carbodiimide [267], dextran dialdehyde [268], and genipin [269].

Enzymatically cross-linked gelatins are biocompatible [270, 271]; transglutaminase

is a naturally occurring enzyme capable of cross-linking gelatin [272]. A detailed

protocol of microchannel fabrication using transglutaminase cross-linked gelatin is

described in [273].

Fig. 4.20 Selective

cross-linking of gelatin with

glutaraldehyde (After [263])

218 E. Meng et al.

Photolithographically deﬁned gelatin has been used as a physical mask for

sandblasting with Al

particles [274] and its permeability enables grayscale

masking during a thru-gelatin Cu electroplating process [261]. Gelatin is also resis-

tant to plasma etching but develops surface roughness during the plasma exposure

(0.1 μm/min, 400 W, O

/SF

)[260]. Alternatively, gelatin can be removed by

enzyme digestion (proteinase K) [275]. Although the etch rate is slow (∼50 nm/min

in 1 mg/ml proteinase K solution), this offers new possibilities for the use of

unphotosensitized gelatin as a sacriﬁcial material. Here, photoresist was used

as a masking material but suffers from poor adhesion to the underlying gelatin

[260]. Fabrication of epoxy microchannels was performed using a gelatin sacriﬁcial

layer [276].

The thermoresponsive nature of gelatin allows its use as a “thermoresist.” Gelatin

ﬁlms were patterned after contact with a heated stamp (50

◦

C) and subsequently used

as a mask to allow selective conjugation of molecules to underlying chitosan [262].

Following conjugation, gelatin was removed either with warm water or by digestion

with a protease.

Immersion in water results in swelling and signiﬁcant dimensional changes (up to

twice the original geometry) [253]. Rinsing in ethanol or acetone extracts absorbed

water but excessive extraction may induce cracks or delamination [260].

Gelatin adheres reversibly with PDMS [272] and adheres well with glass [260].

Adhesion to Parylene C was promoted by a brief plasma roughening step (1 min,

100 W, O

/CF

Release of gelatin from patterned Si is achieved by soaking Si surfaces for several

hours in a 50% wt hexamethyldisilazane (HMDS) solution in hexane and appropri-

ate thermal processing (30% wt gelatin (type B, calf bone derived, 200 Bloom),

◦

C for 2 h followed by rapid air cooling at 5

◦

C and 12 h of settling time).

Although sealing to PDMS is reversible, PDMS molds treated with Pluronic



F127

(BASF Corporation) are nonadherent to gelatin [277, 278] and molded gelatin pat-

terns are removed by warming (25

◦

C) and ﬂushing with 1% bovine serum albumin

(BSA).

4.4.2 Chitosan

Like gelatin, chitosan i s also a naturally occurring polymer. Chitosan is an N-

deacetylated derivative of chitin which is commonly obtained from crustacean shells

(Fig. 4.21). The natural origins of both materials impart a host of biomedically

favorable properties such as nonimmunogenicity, biocompatibility, and nontoxic-

ity. However, in contrast to highly insoluble chitin, chitosan is also biodegradable.

In addition, chitosan has the ability to form gels, has optical clarity, is antimicrobial,

and has been found to be conducive to the wound healing process. Unlike most nat-

ural polysaccharides which are neutral or acidic, chitosan is highly basic. Chitosan

is a well-known biomedical material with applications in controlled drug delivery,

sutures, contact lenses, and medical coatings [279].

4 Additive Processes for Polymeric Materials 219

Fig. 4.21 Chemical structure

of chitosan (Modiﬁed from

[279])

Chitosan is produced by the deacetylation of chitin in 40% sodium hydroxide at

120

◦

C for 1–3 h. This process typically yields 70% deacetylated chitosan. In this

form, chitosan can be dissolved in dilute acids (such as acetic and formic); its solu-

bility is pH-dependent [262]. Chitosan ﬁlms were spin-casted from acidic solutions

(typically 1–2% w/w). For example, contact lenses were spin-casted from squid

pen chitosan [279]. Above a pH of 6.5 or in neutral conditions, chitosan is ren-

dered insoluble [262, 280, 281 ]. Exposure of chitosan ﬁlms to strong bases results

in delamination (e.g., 1 M NaOH for 30 min) [262]. Chitosan solutions, mixed with

hydroxyapatite (a bioceramic), can also be dispensed to create three-dimensional

scaffolds with a feature resolution of ∼200 μm[282].

In acidic solutions, chitosan is positively charged whereas it is negatively charged

in basic solutions. This property enables the electrodeposition of several μm thick

chitosan ﬁlms onto the negative electrode. Plating was performed at a pH of 5 using

a 1% w/v chitosan cationic polyelectrolyte solution [280, 283].

Chitosan can be selectively cross-linked by using naturally occurring phenol

reactions such as anodic catechol oxidation. Cross-linking occurs through electro-

chemical oxidation only at patterned anodes [284]. Other chitosan biofabrication

techniques are reviewed in [285]. Chitosan can also be rendered photosensitive

[286]; following UV exposure, uncross-linked chitosan is rinsed away in phosphate-

buffered saline (PBS). Although this process can achieve features down to 100 μm,

the process is carried out while chitosan is in the solution state [287]. By adding

a layer of PMMA between the photoresist and the underlying chitosan, the chi-

tosan is protected from the photoresist developer and photopatterning is possible

[288]. In the process shown below (Fig. 4.22), photolithography is performed and

then the PMMA/chitosan layer is etched using the photoresist mask by reactive ion

etching in an oxygen plasma (70 W, 50 sccm, 40 mtorr, 20

◦

C substrate cooling,

bias for vertical sidewalls, selectivity to photoresist of 1:1.65). Following pat-

terning, it is also possible to undercut and release the chitosan structure by gas

phase etching of the silicon substrate in XeF

. The mechanical properties of both

hydrated and dehydrated chitosan structures prepared using the process were also

characterized [288].

Thermal imprinting of structures in chitosan ﬁlms is accomplished by using soft

PDMS stamps. In earlier work, glycerol was added to chitosan as a plasticizer

to increase chain mobility and facilitate imprinting [289]. Glycerol addition low-

ered the modulus of the resulting ﬁlm and enhanced transfer of thick structures.

220 E. Meng et al.

Fig. 4.22 Fabrication of chitosan structures by photolithographic techniques (Original ﬁgure

Imprinting of 280 nm wide and 12 nm tall lines was achieved under low pressure

and temperature conditions (25 kPa and 80

◦

C with 1:1 glycerol-to-chitosan ratio

(1.0 wt%)). Later, this process was simpliﬁed; plasticizers were omitted by directly

imprinting wet chitosan ﬁlms [290]. PDMS stamps were used speciﬁcally to allow

for solvent evaporation through the stamp. Imprinting of nm features was performed

at 5–25 psi and 90

◦

C. PDMS stamps were easily released from chitosan structures

without any prior surface treatment.

4.4.3 Polyethylene Glycol

Poly(ethylene glycol) (PEG) is a linear or branched polyether having the follow-

ing structure: HO-(CH

-OH [291]. This neutral hydrogel, also

referred to as carbowax, possesses hydroxyl groups at either end and is available

in many molecular weights. This Food and Drug Administration (FDA) approved

4 Additive Processes for Polymeric Materials 221

polymer is nontoxic, nonimmunogenic, and nonantigenic. Films of PEG act as bio-

logical passivation layers inasmuch as they can be applied to surfaces to reduce

protein adsorption [292].

PEG is soluble in water and most organic solvents, allowing it to be applied to

substrates from solution [293–297]. In surface modiﬁcation applications, PEG has

been applied by physical adsorption [298–303], covalent bonding (grafting or chem-

ical coupling) [293, 296, 304–309], and self-assembly [310]. However, solution-

based deposition techniques are not suitable for producing uniform, conformal, and

ultrathin PEG ﬁlms on surfaces. Dry vapor deposition approaches are available

using ethylene oxide with a gas catalyst (boron triﬂuoride) for deposition on surfaces

or within microchannels [291, 292, 311, 312]. Prior to deposition, surfaces (sili-

con or glass) are cleaned in piranha and silanized (3-aminopropyltrimethoxysilane

(3-APTMS)).

Photosensitive PEG (negative photoresist) is possible by addition of a photoini-

tiator. PEG, PEG-diacyrlate (PEG-DA), PEG-methacrylate, PEG-dimethacrylate

(PEG-DM), PEG-disilane, and PEG-tetraacrylate (PEG-TA) of varying molecular

weights where modiﬁed with 2,2-dimethoxy-2-phenyl acetophenone (DMPA), 2-

hydroxy-2-methyl propiophenone, and Irgacure 2959 photoinitiators. Solutions are

spin-coated on substrates and then UV-exposed to cross-link the material [161,

310, 313–317]. Development is performed in water, solvent, or supercritical CO

[313, 315, 318]. PEG-DA and PEG-DM are preferred for short exposure times

and low viscosity [317–319]. In addition, lower molecular weight formulations

exhibit improved mechanical strength and less swelling when immersed in aque-

ous solutions [314, 317]. Irgacure 2959 photoinitiator is somewhat cytocompatible

and has been used with PEG-DA (MW = 1000 Da) solutions to form a precursor

used in three-dimensional stereolithographically deﬁned tissue engineering scaf-

folds [316]. In this application, 20 or 30% (w/v) PEG-DA was required to form

a gel following cross-linking by laser exposure (HeCd laser, 325 nm, 40 mW,

250 μm spot diameter). When using 15% (w/v) or less, a mechanically weak paste

resulted.

Alternatively, PEG is either a masking material or a sacriﬁcial layer with nm

resolution when deposited by dip pen lithography [320]. Patterns were written on Au

surfaces from 5 mg/ml acetonitrile solutions of PEG (MW =2000 Da). When using

the PEG as a masking material, the Au was selectively etched in 20 mM thiourea

and 30 mM iron nitrate nonahydrate. To use the PEG patterns as a sacriﬁcial layer,

the masked surface was ﬁrst selectively passivated with 1 mM 1-octadecanethiol

(ODT). Following removal of the PEG by dichloromethane, Au features were etched

using the same chemistry previously mentioned.

PEG may be selectively removed by oxygen plasma etching [321] or selectively

deposited by screen printing [322]. For example, PEG (MW =8000 Da) was screen

printed by heating to 80

◦

C and squeegeeing (Teﬂon blade) through an Al screen

(2 kPa pressure and 8 ft/min speed). Deposited patterns were reﬂowed at 65

◦

Cto

smooth the surface. These patterns formed sites for thermopneumatic actuation fol-

lowing deposition of Parylene to complete the chamber. Used in this manner, PEG

may produce up to 30% volume change [322].

222 E. Meng et al.

Stamping of patterns using a PDMS stamp is possible; a variation on this

technique uses a PDMS mold and photosensitized PEG-DM in a capillary force

lithography method [323, 324] for creating PEG structures for cell culture appli-

cations [314, 318]. However, PDMS molds were too soft; the tendency of features

to deform, buckle, or collapse prevented creation of sub-100 nm features [325].

By replacing the PDMS mold with poly(urethane acrylate) (PUA), features down

to 50 nm were reported [326–328]. It is necessary to treat PUA molds with 1%

amorphous ﬂuropolymer (Dupont Teﬂon AF2400) to facilitate mold release.

4.4.4 Case Studies

PEG may be used as either a sacriﬁcial or structural material in conjunction with

microﬂuidic channels. For example, although PEG is relatively soft, it is an excellent

sacriﬁcial material for temporarily increasing the stiffness of thin-walled Parylene

microchannels [329]. This microﬂuidic neural probe must sustain 1 mN of force

during the insertion process into the brain. As fabricated, the hollow probes would

buckle below 1 mN (Fig. 4.23). By pulling PEG (MW = 2700 − 3500 Da) into

the channel by suction and then cooling, the PEG-stiffened probes were success-

fully inserted (buckled at >12 mN). Following insertion, the PEG was dissolved

upon contact with the physiological saline (∼200 s to dissolve 7 × 10

μm

in the

microchannel).

Fig. 4.23 PEG used as a sacriﬁcial material to temporarily stiffen a ﬂuidic probe for insertion into

biological tissue. (a) The ﬂuidic probe and (b) comparison of a hollow probe and PEG-ﬁlled probe

during insertion into biological tissue (Reproduced from [329] by permission of The Royal Society

of Chemistry)

4 Additive Processes for Polymeric Materials 223

PEG microchannels were directly formed without using a sacriﬁcial layer by

combining photosensitized PEG and irreversible UV sealing [317]. Two halves

of the channel were formed from either low molecular weight PEG-DA or PEG-

DM with 1 wt% UV initiator (DMPA). PEG was dispensed onto a silicon master

mold and covered with a thin polyethylene terephthalate (PET) ﬁlm for support.

Following UV exposure, cross-linked ﬁlms were demolded and ﬂuidic access holes

were drilled. Then the molded half was brought into contact with another supported

PEG ﬁlm (10

Pa pressure) and exposed to UV for a few minutes to irreversibly

cross-link the contacted surfaces.

4.5 Parylene

Poly (p-xylylene) is the chemical name for the trademarked Parylene family of poly-

mers offered by Specialty Coating Systems (Indianapolis, IL), the sole licensee of

the original Union Carbide invention. Several alternatives to Parylene exist commer-

cially that form in poly (p-xylylene) but differ in the synthesis path of the starting

dimer (di-para-xylylene). However, only the Specialty Coating Systems version is

currently approved by the FDA. Parylene is a USP (United States Pharmacopeia)

Class VI material suitable for long-term implant and thus is well known histori-

cally in the medical industry as a biocompatible coating [330]. Many have examined

its biocompatibility from different perspectives [42, 331–338]; its biostability, low

cytotoxicity, and resistance against hydrolytic degradation are crucial attributes in

Parylene’s popularity as a biomedical material [42, 339, 340]. Poly (p-xylylene)s

were reviewed in [341] and a brief review of their use in MEMS applications is

found in [342].

Swarc synthesized the ﬁrst Parylene (type N) in 1947 but development of a suit-

able deposition process delayed commercial introduction of the material until 1965;

key enablers include the development of a stable dimer precursor and vapor deposi-

tion polymerization process by Gorham at Union Carbide [343–345]. Over 20 types

of poly (p-xylylene) were synthesized, but only 4 are available as Parylene (types

N, C, D, and HT) and 2 additional types are available as diX A and AM (Kishimoto

Sangyo Co., Ltd., Japan). The chemical structure and deﬁning features of each type

are given below (Fig. 4.24 and Table 4.3). The type predominantly used in MEMS

is Parylene C which is the focus of this section.

Parylene physical vapor deposition is performed in a vacuum and starts with the

granular dimer which is loaded into a vaporizer (Fig. 4.25). The process param-

eters described here are standard for Parylene C. The temperature is ramped to

150

◦

C, at which point the dimer vaporizes. Next, the dimer is pyrolyzed to cleave

the dimer into the monomer radical para-xylylene (650

◦

C). In the deposition cham-

ber (25

◦

C), the monomer adsorbs to all exposed surfaces and polymerizes at the rate

of ∼5 μm/h for Parylene C. Recently, a process was developed to directly deposit

Parylene as a nanostructured thin-ﬁlm by modifying t he sample mounting apparatus

in the deposition chamber [347–349].