Kutz M. Handbook of materials selection

Подождите немного. Документ загружается.

2 MATERIALS SELECTION FOR SELECT COMPONENTS 1321

Fig. 12 Soot deposition onto a rotating rod by the outside vapor deposition (OVD) process.

attention must be paid to eliminate contamination of the crucible and furnace

environment.

In the vapor-phase oxidation process, highly pure vapors of metal halides

(e.g., SiCl

and GeCl

) react with oxygen to form a white powder of SiO

par-

ticles. These particles are collected and are sintered by one of a variety of

techniques to form a clear glass rod or tube. This rod or tube is called a preform,

which is as much as 100 mm in diameter and 100 cm long, yielding hundreds

of kilometers of ﬁber. Fibers are drawn from preforms in a precision draw tower

furnace. In it, the bottom is softened so that a thin ﬁlament of glass (optical

ﬁber) can be drawn. The thickness of the ﬁber is determined by the drawing

speed.

The ﬁrst optical ﬁber was manufactured by outside vapor-phase oxidation

(OVPO) at Corning Works.

A layer of silica particles called soot is deposited

from a burner onto a bait rod, where a porous perform is made layer by layer

(as depicted in Fig. 12). When the desired thickness is attained, the bait rod is

removed, and the porous perform is then vitriﬁed and collapsed into a clear glass

rod in a dry and hot environment (above 1400

⬚C). Other methods exist and are

described in other texts.

31,34

One method, which is a new fabrication point of

view, known as sol–gel, starts from a liquid precursor [e.g. Si(OC

)

hydro-

lyzed in water], is dried and sintered into a glass perform.

One important type of vapor-phase fabrication is the modiﬁed chemical vapor

deposition (MCVD) created at Bell Labs.

The process starts with a very pure

tube of silica glass (SiO

), about 15 mm in diameter and 1 m long. Reactant

gases of metal halide gases and oxygen ﬂow through a revolving tube, as de-

picted in Fig. 13. As the SiO

particles are deposited, they are sintered into a

clear glass by a moving oxyhydrogen torch, which travels back and forth along

the tube. Successive layers of SiO

followed by GeO

doped SiO

are deposited.

When the total deposit reaches the desired thickness, the vapor ﬂow is stopped,

and higher heat is applied to collapse the solid tube. Glass ﬁbers are surprisingly

strong. The longitudinal tensile strengths of 5 GPa in a short length of glass

ﬁber is larger than the yield strength of steel wire (2 GPa). Unlike steel, glass

does not yield and does not deform plastically. Glass ﬁbers can elongate 1%

more than 10% before fracture occurs. These extraordinary mechanical proper-

1322 ADVANCED MATERIALS IN TELECOMMUNICATIONS

Fig. 13 A schematic of the modiﬁed chemical vapor deposition (MCVD) process.

ties of glass ﬁber derive from the pristine (ﬂaw free) surface of a freshly drawn

ﬁber. In the drawing process, the ﬁber is quickly coated with a polymer layer

to protect the surface from mechanical scratches and chemical reactions with

the atmosphere. Commercial optical ﬁbers are routinely proof tested at levels of

0.69 GPa or 1.38 GPa before sale.

Fiber-Optic Cables

Special cable designs are required because of the mechanical properties of glass.

These designs are greatly dependent on how the cable will be used. For example,

cable may be pulled underground, submerged underwater, or installed outdoors.

Generally, a ﬁber is coated with buffer material and placed loosely in a poly-

ethylene tube. The tube is surrounded by aramid yarn to provide strength and

encapsulation. Finally, an outer jacket of polyurethane, polyethylene, or nylon

binds the encapsulated ﬁbers together. Larger cables are fashioned from these

units by binding ﬁber bundles to a support with tape, then surrounding them

with an outer jacket. Undersea optical ﬁber cable, like copper cable, needs to

be protected from water vapor and its mechanical strength needs to be increased.

This is accomplished by forming cables with steel wire strands and polymer

jackets (see Fig. 14).

Fiber Devices

Advances in optical ﬁber have extended into device fabrication, such as ﬁber

Bragg gratings (FBG).

FBGs are made by doping the core of an optical ﬁber

with germanium, which increases the index of refraction allowing the guiding

of light. This ﬁber is masked according to the desired modulation pattern and

exposed to ultraviolet light, which increases the index of refraction in the areas

exposed due to the photorefractive effect. Thus, the core of ﬁber posseses an

index modulation along the propagation direction, thereby demonstrating Bragg

grating characteristics. These gratings can be used as simple ﬁlters where the

reﬂected wavelength is removed from the transmitted signal. However, many

functionalities can be added to the FBG to monitor network operation and to

permit rapid reconﬁguration. For instance, thin-ﬁlm heaters can be deposited on

2 MATERIALS SELECTION FOR SELECT COMPONENTS 1323

Fig. 14 Cross section of optical ﬁber cable for undersea deployment.

the ﬁber, which dynamically compensate for chromatic dispersion during prop-

agation of light in long-haul ﬁber by adjusting the chirp of the FBG.

2.5 Electro-Optical Materials

The area of optoelectronics is demonstrating exploding growth. In the past few

decades, optical communication used ﬁber as the transmission medium but relied

on electronics for ampliﬁcation, signal processing, and switching. The conver-

sion of information from optical to electronic to optical represents a bottleneck

in the optical network. It is obvious that faster data transfer will only be possible

if the electronic components are replaced by all-optical counterparts. To manip-

ulate light, materials demonstrating nonlinear optical (NLO) phenomena are nec-

essary. Materials with large electro-optical coefﬁcients also demonstrate

signiﬁcant NLO behavior. One requirement for all electro-optical materials is a

lack of inversion symmetry in the crystallographic space group, also known as

acentric.

Piezoelectric materials are acentric with a crystallographically deﬁned polar-

ization direction. This direction must be oriented (textured) as deposited if in

thin-ﬁlm form. For nonferroelectric materials, such as ZnO and AlN, poling is

not required and fatigue is not observed. However, usually nonferroelectric ma-

terials have low electro-optical coefﬁcients and only demonstrate a linear electro-

optical coefﬁcient. These materials are not applicable in nonlinear devices that

utilize the Kerr effect.

As mentioned earlier, lithium niobate (LiNbO

), is an electro-optical material

that demonstrates a change in index of refraction (n) with applied electric ﬁeld

(E). High-speed modulators with rates of 40 Gb/s are commercially available.

Waveguides are deﬁned by Ti diffusion, and Au is used as an electrode material.

LiNbO

is a ferroelectric material with a moderate electro-optical coefﬁcient,

1324 ADVANCED MATERIALS IN TELECOMMUNICATIONS

and it does not display fatigue. However, due to the photorefractive effect,

LiNbO

has a low optical damage threshold limiting it from nonlinear applica-

tions that require high-intensity light.

Another ferroelectric, BaTiO

, demonstrates a large electro-optical coefﬁcient

⫽ 100 pm/V), but the thermal stability is low due to a Curie temperature

of 120

⬚C, which is too close to ambient temperature. The Pb-based ferroelectrics

(PT

⫽ PbTiO

, PZT ⫽ PbZr

⫺

, and PLZT ⫽ Pb

⫺

3y /2

⫺

)have

very large electro-optical coefﬁcients, but their properties exhibit fatigue (i.e.,

the electro-optical coefﬁcient decreases with time). In addition, optical attenua-

tion is high since large single crystals have not been grown and epitaxial ﬁlms

cannot be grown to thicknesses required for optical waveguides (

⬃5–10

␮

m).

One reason to investigate the Pb-based ferroelectrics is their high electro-optical

coefﬁcients (r

), which in turn lead to a large change of the index of refraction

(

⌬n) in an applied electric ﬁeld (E) through the following relationship:

–

⌬n ⫽⫺ nrE

2 c

where r

is approximately r

⫺ r

for tetragonal symmetry.

Reported values

for r

are the following (units of pm/V): BaTiO

80, LiNbO

18,

0.88

0.08

(Zr

0.40

0.60

100, Pb

0.88

0.08

(Zr

0.65

0.35

520.

A new direction pursued to ﬁnd high-efﬁciency elecro-optical materials is in

the area of organic materials.

The design of organic molecules to crystallize

in acentric space groups with large electro-optical coefﬁcients has been discussed

based on geometrical arguments of rodlike dipolar molecules.

In addition,

inorganic-organic hybrid nanocomposites are being developed for electro-optical

applications.

2.6 Antireﬂection (AR) Coatings

Attenuation of optical signals through lenses can be signiﬁcant. For example,

pure SiO

with an index of refraction (n) of 1.5 will reﬂect 4% of the light from

a single air–glass interface.

This attenuation can be reduced to 1% using an

antireﬂection coating of MgF

. The coating consists of a quarter wave ﬁlm,

meaning the thickness of the ﬁlm is of the wavelength of light that is desired

–

to be transmitted. The optimal index of refraction for the antireﬂection coating

is n

1/2

. Therefore, SiO

would beneﬁt most from a ﬁlm with n equal to 1.22.

Such a low index is difﬁcult to achieve; however, MgF

has an index of 1.37 at

1550 nm, making it a good candidate as a coating material. In addition, it is

easy to fabricate using electron beam evaporation. AR coatings are used on

almost every optical element in a device.

2.7 Optical Filters

Selecting an appropriate bandgap and index of refraction, multilayers of inor-

ganic thin ﬁlms can be designed for use as optical ﬁlters. A variety of applica-

tions are available: antireﬂection coatings, reﬂective coatings (dielectric mirrors),

neutral density ﬁlters, beam splitters, short-wave pass ﬁlters, long-wave pass

ﬁlters, bandpass ﬁlters (monochromatic), minus ﬁlters, and shaping ﬁlters.

2 MATERIALS SELECTION FOR SELECT COMPONENTS 1325

2.8 Isolators Incorporating Faraday Rotators

An isolator prevents light propagating down a optical ﬁber to reenter a laser

cavity, which disrupts the mechanism of stimulated emission. The isolator com-

bines two polarizers with a Faraday rotator. Light emitted from the laser cavity

is sent through a polarizer to achieve plane polarized light. Then, this light passes

through a Faraday rotator that exactly rotates the polarized light by 45

⬚, which

passes through another polarizer tipped at 45

⬚ to allow the light to pass out of

the isolator. Any light that propagates down the ﬁber and is reﬂected back would

undergo an additional 45

⬚ rotation at which point the initial polarizer would

deﬂect the 90

⬚ out-of-phase light away from the laser cavity, thereby isolating

the cavity.

An iron garnet, (Bi, Eu, Ho)

(Fe, Ga)

, has been developed as the material

for Faraday rotators for telecommunications at 1.55

␮

This material was

selected for its isotropic optical properties, which arise from cubic symmetry.

The garnet structure can be described by C

, where the cations have a

⫹3 oxidation state and reside in dodecahedral (C), octahedral (A), and tetrahe-

dral (D) coordination, and the oxygen anions (X) having a

⫺2 charge reside in

tetrahedral coordination. To increase the speciﬁc rotation, bismuth was added

into the dodecahedral site. However, this also increases the temperature depend-

ence of the speciﬁc rotation, an undesirable consequence. The reason for in-

creasing the speciﬁc rotation is to reduce the thickness of the device. For

instance, yttrium iron garnet, Y

, would require a thickness of 2.7 mm in

order to rotate light 45

⬚, whereas the optimized material is only 0.4 mm thick

to achieve similar rotation. Gallium doping is used to reduce the saturation

magnetization such that the material can only switch between saturated single-

domain states. This permits the use of Faraday rotators in applications were an

external magnet are prohibited, since the rotator will remain in a single-domain

state in the absence of an external magnetic ﬁeld. In addition, all cations must

retain the

⫹3 oxidation state to maintain minimal optical absorption at 1.55

␮

The polarizers can be replaced by a highly birefringent material to make a

polarization independent device. Rutile, TiO

, is one possible material for this

application. In this manner, both components of the refracted light (ordinary and

extraordinary) can propagate through the device without signiﬁcant attenuation.

Future developments would include thin-ﬁlm fabrication for integrated optics.

2.9 Microelectromechanical Systems

Microelectromechanical systems (MEMS) is a burgeoning technology that in-

tegrates many of the lessons of integrated circuit technology into small-scale

systems. The thrust for this technology combines the high volumes of production

with low cost. In addition to the economic potential, this approach allows unique

opportunities in the full integration of devices at very small scales. A device

composed of mechanical and electrical elements built on silicon can now sense,

actuate,

produce power,

50,51

guide light,

and maintain chemical reactions.

The success of this technology, however, lies in the understanding of the ma-

terials issues associated with the design and fabrication of the tiny devices.

Although the materials set borrows heavily from silicon technology, there are

a few new players in the MEMS arena. MEMS technology extends the materials

1326 ADVANCED MATERIALS IN TELECOMMUNICATIONS

set beyond those of microelectronics to include diamondlike carbon,

54,55

silicon

carbide,

metals and polymers,

and piezoelectric materials.

Silicon process-

ing has also changed its character to create high-aspect ratio devices. For

example, deep reactive ion etch (RIE).

Bosch techniques,

60,61

and other

techniques are now used to create deep trenches and other unique geometries

that were not compatible with etching anisotropies of conventional materials.

A host of fabrication techniques

are used, including surface machining,

bulk machining,

and molding processes (or LIGA).

Surface machining builds

on top of the silicon substrate, bulk machining relies on etching into the wafer,

and molding produces pieces by microfabricated molds. The basis of micro-

machining is photolithography, which is the means of deﬁning regions on a

wafer. Etching (or removal of silicon), doping, and deposition can all be con-

trolled by photolithography and make up the key ingredients to building three-

dimensional forms. Surface micromachining is comparable to the process to

build VLSI technology. Molding is based on LIGA, an acronym based on the

German phrase for Lithographie, Galvanik (electroplating) and Abformung

(molding). This technique can create high-aspect ratio photoresist regions by

illumination it with high-intensity X-ray radiation (often done with a synchro-

tron). The resist is exposed to form open regions in which electroplated metal

is deposited. The resulting device consists of photoresist and electroplated metals

agglomerations. Other techniques on the horizon for fabrication include conven-

tional techniques such as cutting, shaping, and drilling on a small scale.

66,67

Sand

blasting, laser ablation,

and spark erosion may become viable paths to micro-

fabrication as well.

MEMS Materials Set

The pathways to fabrication described above provide designers a means of de-

veloping a host of devices. However, achieving reliable low-cost devices is

strongly dependent on the materials used. The materials set for MEMS consists

of all the ﬂavors of conventional integrated circuit technology, plus a few new

ones. The original materials include single-crystal wafer substrates, polysilicon

resistive layers, and aluminum and copper conductive paths. In addition, silicon

oxide, silicon nitride, and titanium nitride are used for insulation, passivation,

and passivation/protection, respectively.

In the MEMS world, these materials

now serve additional functions. Silicon, polysilicon, and silicon nitride are me-

chanical elements, aluminum is used as a reﬂective material, and silicon oxide

is used sacriﬁcially. Other materials are employed as well. Gold is used as a

reﬂective material and as a conductive path. Titanium and chromium are used

as adhesion layers, and silicon-on-insulators (SOI) wafers serve as substrate

material. In addition, photoresist was once used and removed in processing; now

it remains as a structural material in MEMS.

For materials in MEMS, every-

thing that was once old is now new.

In MEMS, mechanical elements for displacement-based sensors include ro-

tating parts and torsional ﬂexures. Silicon is an attractive material due to its high

stiffness and high strength. Silicon carbide, diamond, and alumina (Al

) are

also good choices for similar applications. For applications in which high me-

chanical forces and power levels are required, metals are a more obvious choice.

Electroplated metals such as copper, nickel, and their alloys are used, as are

2 MATERIALS SELECTION FOR SELECT COMPONENTS 1327

Fig. 15 An optical cross-connect MEMS mirror used to route light from

one optical ﬁber to another.

other metals such as chromium, iron, cobalt, and strengthened versions of each.

Silicon carbide offers a high stiffness, hardness, and wear-resistant option in the

MEMS material palette, yet in comparison to other materials in the VLSI col-

lection, its development is not as mature. Promising materials for the future

include sapphire, amorphous alumina, fused silica, and diamond, which are

available in wafer form to create MEMS devices.

Transducer materials used to sense, actuate, or convert mechanical stimuli

into electrical responses are also important to note. Actuation can occur electro-

statically, magnetically, piezoelectrically, and thermally. Electrostatic and ther-

mal actuation materials are generally made from conventional thin-ﬁlm metals.

More esoteric piezoelectric and magnetic materials are used in the associated

form of actuation. The most common piezoelectric materials include lead zir-

conate titanate (PZT),

, zinc oxide (ZnO).

and aluminum nitride (AlN).

These materials are typically deposited by sputtering; however, a special case of

sol–gel deposition onto micromachined parts has also been reported.

Magnetic

materials such as Permalloy

and shape memory alloys such as NiTi

are means

of actuation.

One example of a MEMS-based device for telecommunications is the optical

cross connect, developed at Bell Laboratories.

74,75

This device redirects light

between ﬁbers by micromachined mirrors (see Fig. 15). The silicon mirrors are

supported by gimbals of the same material. Two sets of torsional silicon bars

allow the mirror to tilt in two directions. The mirrors are moved by electrostatic

actuation and require passivation layers for protection. Electrical signals for ac-

1328 ADVANCED MATERIALS IN TELECOMMUNICATIONS

tuation traverse gold conductive paths. Gold is also used as the reﬂective layer

on the mirrors. This example illuminates the plethora of materials needed to

make a device and underscores the need for understanding the materials in order

to create reliable devices in this fast-moving technology.

From this discussion we have seen that MEMS devices will certainly be an

enabling technology for telecommunications for several decades. Its success is

due largely to understanding, creating, and designing of materials. Understanding

material properties provide the means to create reliable, low-cost and fast de-

vices. In the light-wave network, a host of possible MEMS applications remain

undiscovered. The new capabilities of this technology and the lessons learned

from integrated circuit technology are rich fodder for growth in the telecom-

munication commercial arena.

2.10 Microwave Radio-Frequency (RF) Resonators

A review of materials for microwave applications has been written by Lavergh-

etta highlighting dielectric laminates and metals.

The most important properties

for microwave dielectrics is the magnitude of the dielectric constant and the

dissipation factor.

Bulk Resonators

Most wireless base stations use a bulk dielectric material, Ba

, for RF

resonators. This material uniquely displays a high dielectric constant and low-

temperature coefﬁcient of the dielectric constant (TCK). Most dielectric mate-

rials demonstrate a nearly linear relationship of the dielectric constant and TCK

known as the Harrop correlation (Fig. 16). However, Ba

displays an

anomalous low TCK allowing for the development of a high-quality factor (Q

factor) resonator with temperature insensitivity. Also, Ba

is able to with-

stand the high power requirements of base station transmitters.

Recent advances in high-temperature superconductors have yielded fre-

quency-tunable resonators using copper-oxide-based superconductors for wire-

less communications. Base station ﬁlters and ampliﬁers are currently made by

Superconductor Technologies from thin ﬁlms of TBCCO (thallium barium cal-

cium copper oxide).

Thin-Film Resonators (TFRs)

TFRs are already emerging in wireless handsets and have several advantages

over the conventional bulk dielectric resonator. First, the dielectric resonator is

the largest single component is a wireless telephone. In addition, due to the

many wireless network standards around the world, a global traveler would need

several separate phones. The advent of the TFR brings a device with multiple

ﬁlters all within an area of a square centimeter and as thin as any integrated

circuit. Also, with further development, TFRs will be integrated with electronics

on the same wafer leading to reduced cost and even smaller phones.

Optimal materials for these devices are materials having the wurtzite structure,

such as ZnO and AlN. To be implemented in 2G and 3G wireless networks, the

material must be piezoelectric and have a large electromechanical coupling co-

efﬁcient that is proportional to the bandwidth of the ﬁlter. TFR devices operate

2 MATERIALS SELECTION FOR SELECT COMPONENTS 1329

-1000

-800

-600

-400

-200

200

400

0 102030405060708090100

dielectric constant, K (unitless)

temperature coefficient of the dielectric constant, TCK

(ppm/°C)

organics

inor

anics

Fig. 16 Harrop correlation for dielectric materials.

based on the propagation of acoustic waves either as a bulk acoustic wave or a

surface acoustic wave. Research toward the fabrication of reliable TFRs from

ferroelectric materials is underway.

2.11 Repeaters and Switches

Repeaters, also known as regenerators, perform ﬁltering and ampliﬁcation to

electrical or optical signals. For optical communication, it would be advanta-

geous to have an all-optical network including repeaters and switches. For the

time being, conventional electronic circuits dominate these systems. Most of

these challenges are similar to those experienced in the microelectronics market

in order to achieve higher speed: high thermal conductivity materials, such as

SiC,

high dielectric constants for logic gates and memory capacitors, and low

dielectric constant materials that minimize RC delay. These topics will not be

covered in this review.

2.12 Multilayered Ceramics

The multilayer ceramic integrated circuit (MCIC) has enabled the production of

low-cost RF wireless devices.

The devices are fabricated with low-temperature

co-ﬁred ceramics (LTCC) to incorporate components such as capacitors, resis-

tors, transmission lines, and inductors into the substrate equivalent of a printed

circuit board. The following materials are used in this process: dielectrics (crys-

talline and glassy phases containing MgO, SiO

,Al

, PbO, and CaO)

and conductors (Au, Ag, Cu, Pd/Ag).

Power systems also take advantage of multilayer ceramics for magnetic com-

ponents with Mn–Zn ferrite cores, such as transformers and inductors.

This

1330 ADVANCED MATERIALS IN TELECOMMUNICATIONS

work led to the innovation of a power supply for a microprocessor nearly the

same size as the integrated circuit itself.

Alumina-based multilayer ceramic packages are ubiquitous in optoelectronic

devices due to superior RF characteristics and high reliability, as previously

described.

3 COMMUNICATION SYSTEM COMPONENTS

Table 2 provides information on various communication system components.

The table is not inclusive, and with advances in research and development, the

number of new materials in telecommunications is increasing dramatically.

4 SYNTHETIC METHODS

4.1 Introduction

Remarkable progress in the ﬁeld of telecommunication materials have taken

place due to the advances in thin-ﬁlm processing for deposition and in material

fabrication. In this section, a review of such processes needed to create tech-

nologically relevant materials will be discussed. Emphasis is placed on offering

a working knowledge of each technique and providing a guide to the literature.

4.2 Sputtering

A common technique for deposition is sputtering.

It consists of the bombard-

ment of the material of interest, known as the target, with energetic ions. In the

sputtering process, an inert gas, such as argon, ﬂows to provide a medium in

which the sputtering process can occur. Gas pressures usually range from several

to 100 mTorr.

Sputtering is induced when a high negative bias is applied to

the target, as depicted in Fig. 17. The bias initiates the ionization of the gas,

which turns into a plasma consisting of ions, electrons, and neutral species. The

positive argon ions accelerate toward the target. When the ions strike the target,

they remove target atoms by momentum transfer. The target erodes as atoms or

clusters of atoms are removed, condensing on the facing substrate and forming

successive layers.

Sputtering is advantageous because the use of energetic sputtered species

gives rise to highly adherent dense ﬁlms of uniform composition and thickness.

Commonly used for metallization in telecommunications, sputtering generally

preserves the stoichiometry of the source materials. These beneﬁts are often

offset, however, by the technique’s low deposition rate and the inclusion of

sputtering gas.

4.3 Evaporation

The simplest evaporator consists of a resistively heated source. In this conﬁgu-

ration, a ﬁlament wire, refractory metal sheet, or crucible source is heated. Com-

mon materials for the source are tungsten, molybdenum, and tantalum. In all

cases, these materials should have a high melting point and low vapor pressure.

The source ﬁlaments support the materials to be evaporated and during heating

are wetted by them before evaporation takes place. Wafers suspended above the

assembly are then coated with a thin ﬁlm of the evaporating material.

Although

the resistive method of evaporation is easy to set up, it is plagued with contam-