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

Подождите немного. Документ загружается.

994 P. Lin e t al.

(PECVD) [110, 184]. The gaseous precursor is decomposed using either a pulsed

microwave or a radio-frequency plasma; the reaction and deposition can then take

place at the surface [185, 186]. These processes can provide a partially-organic

coating with a hardness and elasticity gradient [187], thus improving adhesion and

reducing stress.

Although very popular for thin ﬁlm coatings on dielectric materials, sputtering

is not preferred for direct thin ﬁlm deposition on plastic materials because of the

high emissions of heat and UV radiation during the process. As a result, a protective

layer such as lacquer or a PECVD thin protection ﬁlm is usually required prior to

sputtering [188, 189].

Process Considerations for Vacuum Coatings

There are several intrinsic issues related to vacuum coatings on polymer materi-

als, including plasma/polymer interactions, types of polymer chosen, stresses in the

coatings, and testing methods [110].

As for the interactions between plasma and polymer, the basic problem comes

from high-energy ions, radicals, and short-wavelength radiation produced by plasma

ion sources, glow-discharge equipment, and electron beam evaporation. These high-

energy sources can break the chemical bonds in polymers and cause chemical and/or

physical modiﬁcations. On the other hand, the advantage of bond-breaking is that

it can increase the interaction forces between the substrate and the coating ﬁlm via

crosslinking, interfacial diffusion, or changes in surface wettability, thus improving

the adhesion [190, 191]. Additionally, polar groups formed on the polymer surface

can reduce the surface hydrophobicity of the polymer [110, 192], which is bene-

ﬁcial for moving liquids in microﬂuidic systems due to the high surface tension

[193–195]. However, the wettability of the polymer surfaces is not permanent, i.e.,

the wetting property disappears completely in an atmospheric environment over a

couple of months. This can be attributed to the orientation rearrangement of surface

polar groups and/or the recovery of polymer non-polar groups in a non-polar envi-

ronment such as atmosphere [191]. Hence, if the plasma treatment is not performed

properly, small fractures will form on the bulk substrate and thus produce a weak

boundary layer between the substrate and the thin ﬁlm. Therefore, shortening the

plasma treatment time or using a pulse treatment is necessary to minimize this issue

[110].

Aside from plasma/polymer interactions, material speciﬁc properties includ-

ing heat distortion, thermal expansion, UV absorption, and degassing, may cause

problems in thin ﬁlm coatings. Oxide layers deposited using plasma-ion-assisted

electron-beam evaporation ( PIA E-beam) can exhibit excellent adhesion on poly-

cyclooleﬁns, polyamides, and polyether sulfones [196]. On the other hand, PMMA

usually forms a weak adhesive layer because it degrades after contacting plasma

[197]. Therefore, deposition of a vacuum UV protective layer prior t o the plasma

coating process is required to protect the surface from degradation [151].

13 Surface Treatment and Planarization 995

In some cases, stresses induced during the deposition process, in the coating or

at the interface between thin ﬁlm and substrate, may cause cracking or delamination

of the coatings. Two major sources of this type of defect are related to stress build-

up. One is from the ion-bombardment on the substrate surface, which usually causes

compressive stresses in coatings [151]. The other is the thermal stress resulting from

the coefﬁcient of thermal expansion mismatch between the coating and the polymer

substrate [ 151]. The thermal expansion coefﬁcient of polymers is usually an order

of magnitude higher than that of metal oxides. For example, if a high-index oxide

is deposited onto a polymer surface, high temperature conditions are often encoun-

tered during the fabrication process, either in e-beam evaporation or in sputtering

systems. Therefore, fabrication parameters such as temperature, process time, and

thin ﬁlm thickness must be well controlled and characterized to achieve successful

coatings.

Process Considerations for Wet Chemical Coatings

In addition to the deposition of inorganic materials onto polymers using the vac-

uum coating technique, polymers can also be deposited onto polymer substrates as

antireﬂection layers using wet processes. Two common processes, dip coating and

spin coating, are primarily adopted for the deposition of low aspect ratio micro-scale

structures over rigid ﬂat substrates. However, for complex or high aspect ratio micro-

structures, spraying processes are more suitable for providing uniform coatings. In

the wet coating process for polymer thin ﬁlms, it is important to limit both curing

temperature and the number of radiation-induced cross-linking reactions to protect

the substrate surface. Wet coatings applied to polymers for antireﬂective purposes

are generally restricted to one or two layers because of the increasing difﬁculty of

thickness control for multilayer thin ﬁlms. Several systems have been developed for

wet coating processes, including silicate-based inorganic-organic hybrid polymers

[198], sol-gel alkoxide polymeric materials, and colloidal indium tin oxide [151].

Silicate-based inorganic-organic hybrid polymers contain organic components that

can covalently bond to the polymer substrate as well as to the inorganic network.

Thus they offer both the desired optical properties and enhanced mechanical prop-

erties [198]. Sol-gel alkoxide polymer and colloidal indium tin oxide usually require

several cycles of deposition and hardening, making precise control of thickness very

difﬁcult. In order to obtain good quality and precisely controlled thickness ﬁlms in

the nanometer range, the sol-gel process requires a controlled environment such

as a clean room with well-controlled temperature and humidity. Some successful

applications of wet coating polymers have been commercialized as AR coatings on

PMMA by Nagase & Co. Ltd. and YTC America Inc.

Case Study 7: Surface Coatings for Interferometry Biosensors

To illustrate the optical utility of thin ﬁlm coatings, here we give an example in

which a thin ﬁlm coating was used as an interferometer for bio-sensing applications.

996 P. Lin e t al.

To enhance the interference signal for an immune sensing ﬁber, the tip of the opti-

cal ﬁber was coated with thin metal ﬁlms as well as gold nano particles (GNP)

to increase the optical path for detection (Gold NanoParticle enhanced Fiber Optic

Interferometry, GNPFOI, sensor) [199, 200]. The light source was generated by a

broadband edge-emitting light emitting diode (ELED, PD-LD Inc., USA) driven

by a current controller (LDC210, Proﬁle Optische System, Germany) with a peak

wavelength of 1550 nm, a spectrum half-width of 70 nm, and an emission power

of 15 μW. The light emitted from the ELED was transmitted by a single mode

optical ﬁber (FS-SC-7324, 3MTM, USA) with a 125 μm ﬁber and 8 μm core

diameter, respectively, and directed into the immune-sensing probe to carry the

interferometry signal through the coupling of the splice. A three-port optical circula-

tor (FOCI, Taiwan) was used to guide interfered light to an optical spectrum analyzer

(MS9710C, Anritsu, Japan) to record and analyze the interference spectrum.

The interference immune sensing probe was mainly composed of a resonant cav-

ity of transparent polymer, and two thin reﬂective layers of gold (Fig. 13.43a). To

prepare the probe, ﬁrst a cleaved ﬁber was cleaned with a 95% alcohol solution in

an ultrasound bath. A 3 nm gold ﬁlm was then deposited on the tip surface by e-

beam evaporation. The interfering resonant cavity was formed by dip-coating the

probe with a 30 μm thick layer of polydimethylsiloxane (PDMS, Sylgard 184, Dow

Fiber core

Au film 3nm

0.00E+00

2.00E-07

4.00E-07

6.00E-07

8.00E-07

1.00E-06

1.20E-06

1450 1500 1550 1600 1650

Wavelength (nm)

Intensity (mW)

Interference of R

and R

Fiber PDMS

Fiber Core

(8 μm)

17 μm

NA = 0.16

Δλ

(A)

(B)

(C)

Fig. 13.43 (a) Operating principle of the interference sensor on a ﬁber tip; (b) SEM side view

of this sensor with 125 μm outer diameter and 8 μmcore;(c) interference spectrum of the ﬁber

13 Surface Treatment and Planarization 997

Corning Corp., USA) and curing the probe at 90

◦

C for 10 min. A second gold ﬁlm,

3 nm thick, was deposited onto the PDMS surface to serve as a second reﬂection

surface and as the interface for bio-molecule immobilization. This ﬁlm actually

consisted of gold islands (20–30 nm wide and 4–5 nm high) on the surface. These

spreading islands made the gold thin ﬁlm partially transparent to electromagnetic

waves. Finally, the completed probe was annealed at 200

◦

C for 2 h to enhance the

adhesion between the gold ﬁlm and PDMS layer.

The appearance of the resonant cavity is shown in the SEM image of Fig. 13.43b.

The sensing area can be considered to be a ﬂat plane due to the very small incident

spot on the outermost surface (17 μm in diameter) compared to the ﬁber diameter

of 125 μm.

Immunological detection was conducted by immobilizing rabbit IgG on the

thiol self-assembly-monolayers (SAMs) surface as the antigen, to conjugate with

anti-rabbit IgG-Cy3 modiﬁed with 13 nm GNPs (Taiwan Advance Nanotech Inc.,

Taiwan). The anti-rabbit IgG-coated GNPs were used to demonstrate the effects of

signal enhancement on immune sensing. In this experiment, the stable interference

signal in PBS solution was ﬁrst recorded as a reference for the binding signal. After

obtaining a baseline in PBS buffer, a 17 nM anti-rabbit IgG-coated GNPs solution

was introduced and an interference fringe shift was observed (Fig. 13.43c).

As shown in Fig. 13.43a, there are two partially transparent gold ﬁlms with reﬂec-

tions R

and R

, respectively, on each interface. The ﬁlms are separated by a cavity

with length L

and refractive index n

. As the light (I

) passes through the m

ﬁlm,

light is partially reﬂected as R

and the transmitted light T

going through the res-

onant cavity is partially reﬂected back as R

by the m

ﬁlm. The phase difference

(φ) between the two reﬂections R

and R

yields the interference R, which can be

estimated from the reﬂection coefﬁcients r

and r

[199]:

R =

+ 2r

cos ϕ + r

)

(1 + 2r

cos ϕ + r

)

and R

= r



− n

+ n



; R

= r



− n

+ n



(13.49)

where the magnitude of R represents the interfering reﬂectance from the wave

coupling of R

and R

. The phase difference ϕ can be expressed as:

ϕ =

4πn

(13.50)

where λ is the wavelength of the incident light. After inserting values for the param-

eters, such as λ = 1550 nm, L

= 30 μm, n

= 1.45, n

bio

= 1.44, n

GNPs

= 2.56,

and L

bio

= 1 nm for thiol self-assembly-monolayers (SAMs) and a cysteine

molecule layer, Equation (13.50) can be rewritten as follows [199]:

λ = 0.0513 +0.0912 · k ·D

GNP

· N (13.51)

where D

GNP

is the diameter of the GNPs, N is the binding number of GNPs, and k is

a constant. An empirical equation of the interference fringe shift can be obtained to

998 P. Lin e t al.

show the linear relationship between the diameters of GNPs (D

GNP

) and the bind-

ing number of GNPs (N). This linear relationship, which was further veriﬁed by

experiments [199], shows t hat with GNPs on the sensor surface, the signal can be

enhanced by at least two orders of magnitude, depending on the size of GNP applied

[199, 200].

13.6.2.6 Applications for Light Absorption

The property of light absorption can play a very important role in optical com-

ponents, especially in photo patterning applications. Unlike antireﬂection coatings,

which are on top of the surface, light absorption thin ﬁlms or s tructures are typ-

ically applied beneath the photo-patternable ﬁlms. The purpose of these layers is

to prevent light reﬂection at the bottom surface, minimizing structure defects. Two

case studies in this section will illustrate the design and fabrication of light absorp-

tion layers. The ﬁrst case study is a light absorption coating for embedded ﬂuidic

channels [201, 202]. The second case study discusses the production of inclined

structures without defects using light reﬂection [125, 131].

Case Study 8: Light Absorption Layer for Microﬂuidic Channels

In embedded micro channel formation, we typically want to take advantage of the

dosage gradient of UV light along the vertical direction of the photoresist. When

the top part of the negative photoresist receives adequate exposure dosage while

the bottom part does not, an embedded micro channel can form after the bottom

portion of the photoresist has developed. However, the reﬂection of light from the

very ﬂat and smooth surface underneath the photoresist imposes greater exposure

on the bottom portion and thus prevents the partial exposure process from creating

embedded micro channels, as shown in Fig. 13.44a. As a result, channel height

is very hard to control by partial exposure in the presence of reﬂection from the

underlying substrate. The process window for channel formation is very narrow and

hard to reproduce. An antireﬂection layer on the photoresist-wafer interface can be

used to absorb reﬂected UV light, as shown in Fig. 13.44b. The antireﬂection layer

absorbs excess UV energy, allowing the exposure window to become much wider

and enabling accurate control of exposure depth.

Fabrication starts with spin coating of SU-8 resist on top of the FujiFilm CK-

6020L resist, the antireﬂection coating commonly used as a ﬁlter for UV light.

After t he ﬁrst exposure, which deﬁnes the through hole, a partial second exposure

is applied to deﬁne the channel top region. Multilayer channel structures can be

obtained simply by repeating the processes. It is noted that the antireﬂection layers

must be removed with solvents during the developing process to open up paths for

the lower channels to develop. This simple process has been routinely employed

to fabricate channels with accurate top wall thickness and channel structures [201].

Figure 13.45 shows four different channel top-wall thicknesses of 14, 23.2, 27.2,

and 31 μm, which were exposed to UV light (365 nm) for 122.5, 161.7, 176.4, and

191.1 mJ/cm

, respectively.

13 Surface Treatment and Planarization 999

CrossCross-linked region

Anti-reflection

coating

UV light

mask mask mask mask

SU-8

Channel fully filled

Channel formed

(a)

(b)

Fig. 13.44 UV exposure under (a) normal situation with reﬂection off the substrate; (b) with

14 μm

23.2 μm

27.2 μm

31 μm

Fig. 13.45 Fabrication

process of thickness control

by different UV dosage [ 201].

Reprinted with permission.

Science

Case Study 9: Inclined Structures with Light Absorption Layers

Inclined exposure technology [125, 131] can be used to fabricate optical microstruc-

tures using several-millimeter-thick polymer layers on glass substrates. A ﬁlm

1000 P. Lin e t al.

Top glass surface

Bottom glass surface

Desired Micro mirror

structure

Undesired reflection

structures

Top glass surface

Bottom glass surface

Desired Micro mirror

structure

Undesired reflection

structures

Fig. 13.46 Irregular

reﬂection structures of an

inclined SU-8 micro mirror

generated from multiple

reﬂection surfaces (1.4 mm

thick) [125]. Reprinted with

Institute of Physics

(e.g. SU-8) with a thickness of 1.4 mm requires an exposure energy of about

14,000 mJ/cm

. Reﬂection of such high energy light at the ﬁlm/substrate inter-

face causes unwanted exposure and partial photopolymerization in the reﬂective

direction, resulting in undesirable V-shaped structures inside the thick SU-8 ﬁlms,

asshowninFig.13.46. The reﬂection effect at each interface is calculated using

Fresnel equations. An absorption layer can be applied to absorb the undesired

reﬂection to eliminate this problem.

The transmitted and reﬂected optical energy at each interface in accordance with

the layer order and refractive indices are shown in Fig. 13.47, where E

is the

reﬂected lightwave energy at the interface of n1 and n2, and E

is the transmitted

lightwave energy at the interface of n1 and n2. Because irregularities on the SU-8

photoresist surface can cause an air gap to form between mask and photoresist,

an inclined exposure will lead to a strong reﬂection at the air/SU-8 interface [125,

131]. As a result, glycerol was employed to compensate for the index mismatch

between the mask and SU-8 resist [131]. To allow for a large incident angle for 45

◦

mirror fabrication, the whole setup was also immersed in a glycerol container to

compensate for the refraction effect [131].

The Fresnel equations that describe the reﬂection and transmission of electro-

magnetic waves at an interface were derived in Equations (13.22)to(13.28). The

results of these calculations are shown in Table 13.8 [125]. It can be seen from the

results in Table 13.8 that the most severe reﬂection occurred at the Glycerol/SU-8

(n2’/n3) and SU-8 photoresist/glass (n3/n4) interfaces. Energy reﬂection between

the glycerol/SU-8 interface reduces the exposure of SU-8, but does not affect the

structure shape.

AsshowninTable13.8, although the total reﬂection at the glycerol/SU-8 inter-

face has been eliminated, there is not much improvement at the SU-8/glass substrate

interface – as much as 2.15 and 1.49% of the perpendicular component of incident

light energy are reﬂected at the SU-8/glycerol interface and SU-8/glass substrate

interface, respectively. Since the reﬂection at the glass mask/SU-8 interface does not

13 Surface Treatment and Planarization 1001

2,2

glycerol

quartz mask

glycerol

SU-8 resist

glass substrate

glycerol

antireflection

coating

Fig. 13.47 Schematic light paths of UV exposure on SU-8 thick resist supported by a glass sub-

strate through a contact quartz mask inside an index matching medium with or without a removable

anti-reﬂection layer (between the n3 and n4 layers) [125]. Reprinted with permission. Copyright

2008 Institute of Physics

Table 13.8 Reﬂectance and transmittance on each interface layer for Fig. 13.47. The parameters

are deﬁned and calculated in Equations (13.22), (13.23), (13.24), (13.25), (13.26), (13.27), and

(13.28)

Interface θ



(%) T



(%) R

⊥

(%) T

⊥

(%)

n1/n2 53.45 50.66 0.015 99.985 0.252 99.748

n2/n2



50.66 53.45 0.015 99.985 0.252 99.748



/n3 53.45 45.12 0.049 99.951 2.147 97.853

n3/n4 45.12 52.08 0.024 99.976 1.492 98.508

n4/n1 52.08 53.45 0.004 99.996 0.062 99.938

n1: glycerol refractive index 1.473 at 598 nm, n2: quartz mask index 1.53, n2



: glycerol

refractive index 1.473 at 598 nm, n3: SU-8 refractive index 1.67 at 365 nm, n4: glass

substrate index 1.5 [125]

affect the SU-8 structure, the r eﬂection at the SU-8/glass substrate interface should

be eliminated. Therefore, a removable anti-reﬂective layer (optical absorption layer,

as shown in Fig. 13.47) is employed to absorb reﬂected light at the SU-8/glass sub-

strate interface (between n3 and n4). Once the SU-8 has developed, the antireﬂection

layer in the open area can be removed with solvents.

After applying an antireﬂection layer at the interface between the SU-8 resist

and the glass substrate, as shown in Fig. 13.47, the reﬂection of light was almost

completely eliminated, and clean SU-8 dove prisms were obtained, as illustrated in

Fig. 13.48. The slanted surface on the SU-8 structures was very smooth and ﬂat,

1002 P. Lin e t al.

Fig. 13.48 Optical

micrograph of SU-8 dove

prisms fabricated in a batch.

Thesizeoftheprismis

approximately 3 nm (l) ×

2mm(w)×0.5 mm (t). The

surface roughness as

determined by the WYKO is

approximately 20 nm [125].

Reprinted with permission.

Physics

with less than 20 nm variation in ﬂatness over a 100 μm length, and less than 20 nm

in average roughness, making it suitable for optical applications.

13.7 Chemical Mechanical Planarization

13.7.1 Overview

Chemical Mechanical Planarization (CMP) has been successful as an enabling tech-

nology in integrated circuit (IC) manufacturing, particularly when the minimum

feature size is much below 0.5 μm, in which case the ultra smooth surface is required

to match the sharply reduced depth of focus in photolithography. Outside of the IC

industry, CMP also has important applications in the fabrication of microelectrome-

chanical systems (MEMS) such as sensors, actuators, R/W heads for hard drive

disks, and inkjet print heads. CMP was ﬁrst used to planarize interlayer dielectrics

in microelectronics manufacturing in the 1980s. Later, the technology was adapted

for the manufacture of more complicated structures, such as damascene for cop-

per metallization. Owing to its excellent performance, CMP is now widely used for

planarization in a variety of materials including polymers, metals, dielectrics, and

ceramics.

13.7.1.1 Chemistry of CMP

CMP applications can be generally classiﬁed into two categories: metal CMP and

non-metal CMP. In this section, we will demonstrate the role of chemistry in s ilicon

dioxide and copper CMP, as representative of these two main categories.

Oxide polishing techniques have been studied in the ﬁeld of optical glass for

more than 40 years (see Holland [203] or Izumitani [204] for a review of major

developments). Izumitani examined glasses of various hardnesses and chemical

durabilities that were polished with different media, and found that oxide polishing

13 Surface Treatment and Planarization 1003

occurs as a combination of mechanical abrasion and chemical reactions. Izumitani

also proved that water is important in the oxide polishing process, a ﬁnding that was

conﬁrmed by others [205, 206]. Cook [205] suggested that during oxide polishing,

loads imposed by abrasive particles enable water to enter the oxide network via the

general reaction:

Si-O-Si+ H

O ↔ Si - OH (13.52)

During the polishing process, hydration of the oxide softens the s urface; the softened

layer can be easily removed by mechanical abrasion. Hydration of all four Si-O

bonds on a given silicon atom produces Si(OH)

, which is highly soluble in water

at high pH (>10). Thus, hydration causes dissolution of the network on the surface.

Among abrasives commonly used in glass polishing (such as TiO

,ZrO

,Cr

,SiO

, and others), ceria (CeO

) has the fastest polishing rate [205]. Tetssuya

and co-workers [207] studied the polishing mechanism in SiO

by CeO

particles.

The proposed model is illustrated by Fig. 13.49. First, chemical tooth bonds are

formed between the CeO

particles and the SiO

ﬁlm. Then, the CeO

particle with

attached SiO

lump is removed from the SiO

ﬁlm. The Ce-O-Si bonding and the

cleaving of Si-O-Si bonds are believed to be the polishing rate-limiting steps. In

this process, mechanical tearing has a signiﬁcant effect on the cleaving of Si-O-Si

bonds.

Metal CMP requires a more complex chemical system to maintain stable material

removal and planarization. It is generally understood that metal slurries chemically

modify the surface to be polished, yielding a softer, porous complex layer. This

layer is easily removed through the application of mechanical force in the contact

region. Meanwhile, the protection imparted on the recess region is sufﬁcient to pre-

vent dissolution. The combination of these effects results in planarization of the

surface. Typical metal CMP slurries contain an oxidant, a chelating agent, a passiva-

tion agent, abrasive particles, and a surfactant or other chemical additives. Various

models have been proposed for the chemical reaction in metal CMP. Steigerwald

CeO

particle

SiO

film

SiO

film

Lump of SiO

Dispersion of

SiO

lump

Pad

Direction of pad

CeO

particle

Fig. 13.49 The polishing mechanism in SiO

by CeO

particles. Reprinted with permission from