of a device or system, compatibility with device materials and processes and align-ment accuracy in coupling to other micro-optic or electro-optic devices becomeimportant.31,32

Figure 1.9 shows a 200-mm-size, F/2 microlens array processed by the reflowtechnique and RIE. For more information about refractive microlens processingtechniques, refer to Chapter 3 of this book.Considering again the pros and cons of diffractive and refractive microlenses,

the major drawback of the refractive technique is the troublesome task of improv-ing the fill factor. It is difficult to achieve 100% packing of circular structures suchas those shown in Fig. 1.9. During the late 1990s, many investigations and reportspopped up in the literature with suggestions on how to improve the fill factor ofrefractive microlenses.31

The method most often reported is using grayscale HEPS photomasks. Thismethod is successfully used for enhanced quantum efficiency of both visible andIR imagers.32 The microlenses processed by this method have virtually 100%fill factor independent of pixel size or shape, and have spherical shape over theentire square or rectangular pixel area (Fig. 1.10).For more information about different methods of microlens fabrication for

micro-optics, refer to Chapter 3 of this book.Micro-optics has been used in many diverse applications, both for miniaturiza-

tion of conventional systems and for many novel and unique uses made possible bythe unique properties of micro-optic components. We will briefly review salientaspects of several applications and discuss three of these applications in detailin Chapter 5 of this book.

1.4 Micro-optics in MEMS: MOEMS overview

Recent industrial needs for commercial research and development (CR&D) inoptical systems, including telecom and optical communication, demand device

Figure 1.9: SEM microphotograph of an F/2 refractive microlens array in fused silica.31

Introduction 11

miniaturization that has led to the merge of two major technologies that we havediscussed: micro-optics and MEMS. A promising team-up of these technologiescombined with microelectronics creates the rich enabling technology ofMOEMS (see Fig. 1.11). All three constituent technologies in MOEMS allowfor batch processing, while micro-optics and MEMS—also involving both micro-machining and embossing—makes MOEMS highly interesting for commercialapplications.33

MOEMS technology is a new star, formed in the late 1990s. The technologyis by now so advanced that in the near future it will lead to the development ofa new family of miniaturized devices with enormous potential to revolutionizephotonic systems and bring a new dimension to optical networking and tele-communications.A recent report byNEXUS indicated that theMOEMSmarket is rapidly growing.

The growth rate of the emerging MOEMS market is reported by NEXUS tobe about 1.6 per year, with the market size reaching $4 billion in 2007.34

Figure 1.10: Full fill factor of a refractive microlens. Lens diameter is 60 mm.32

Figure 1.11: MOEMS—interactions of micro-optics and micromechanics with micro-

electronic integrated circuits.

12 M. E. Motamedi

(See Fig. 1.12.) Note that this is just the emerging market. The growth potential ofMOEMS technology is much greater than what has been forecast by NEXUS.In this section, we will introduce several attractive MOEMS devices that are

being developed at different laboratories.

1.4.1 New developments in optical switches

Current demands for telecom, wireless communication, and mobile phonesequipped with GPS and/or Internet access require large-bandwidth performancein optical communication and effective use of wavelength by advancingwavelength division multiplexing (WDM) and dense wavelength division multi-plexing (DWDM) techniques. Using fiber optics for optical transmission linesrequires light concentrators, beam shapers, optical transformers, beamsplitters,and beam scanners. All of these optical components require more or less opticalswitching. Many research centers and universities have worked on differenttypes of optical switches. We will discuss some of the recent development inthis area.Bishop et al.35 reported that MOEMS devices such as optical switches, variable

optical attenuators (VOAs), active equalizers, add/drop multiplexers, opticalcrossconnects, gain tilt equalizers, data transmitters, andmany others are beginning

Figure 1.12: Interim results: emerging existing MST products. (Courtesy of NEXUS,

after reference 34.)

Introduction 13

to find ubiquitous application in advanced lightwave systems. Bishop et al.35 showexamples of these devices and describe some of the challenges to attracting large-volume demand in addressable markets for this technology. Lucent developed theWaveStarTM router, which helps carriers reduce cost while performing trans-mission functions with less space and increased efficiency. There are discussionsabout WaveStar routers and their structures in several chapters of this book:Chapters 4, 6, 7, and 10.The Institute for Microstructure Technology, Forschungszentrum, Karlsruhe

reported a 2 � 2 fiber optic switch matrix with electromechanical micromotors.36

This report describes a prototype of an all-optical switch, which has been devel-oped with LIGA technique. The switch array has been designed for a telecomwavelength of 1.55 mm and for single-mode applications.In the Karlsruhe switch, the optical signals are directed inside a micro-optical

bench by movable micromirrors. A LIGA structure guide passively aligns allmounts and stops with optical fibers inside a micro-optical bench. The mirrorsare controlled by electrostatic wobble motors. The 2 � 2 switch matrix has dimen-sions of 10 � 10 mm2 plus six micromotors 1.7 mm in diameter. All alignment ispassive, the switching speed is around 30 ms, the crosstalk between channels isbelow 290 dB, and the switch insertion loss is 7 dB.36

Goering et al.37 report a hybrid MOEMS approach for fiber optic switches andswitch matrices. This switch array uses movable microprisms in the paths of inputbeams to diffract each beam to its output fiber through an array of focusingmicrolenses. The switching speed is about 2 ms with insertion loss of 1.5 dB,and crosstalk of below 270 dB.Prof. Voges’s group at the University of Dormund has reported a latching-type

fiber optic switch array.38 The switch is based on bulk micromachining of f100gsilicon with high-precision anisotropic etching, which is suitable for fiber opticalignments. The authors describe two types of switches: a 1 � 4 and a 2 � 2.The 1 � 4 switch is based on moving the input fiber precisely in front of each offour output fibers. The output fibers are stationary in front of four V grooves,and the input fiber is moved by two thermal actuators to be positioned in anydesired V groove. Since all the V-grooves are produced by bulk micromachining,the input and output optical fibers are self-aligned. Readers should refer toChapter 6 for further information about optical switches used for fiber opticnetworking.

1.4.2 Tunable filters and WDMs

Other hot topic areas in MOEMS are tunable filters and WDMs. NASA/GoddardSpace Flight Center reported a simulation study of a large-aperture MOEMStunable filter.39 The Fabry-Perot filter is a critical component for high-throughput,wide-field imaging spectroscopy for lidar applications.Zhang et al.40 describe a polysilicon micromachined 3D mirror that is integrated

for frequency tuning in WDM systems. In this work, the frequency tuners use 3Dmicromirrors, which are integrated with single-mode Fabry-Perot laser diodes and

14 M. E. Motamedi

antireflection-coated optical fibers. The frequency difference between the twotuners is adjusted by changing the external cavity length. A movable 3D mirrordriven by a comb drive performs the adjustment of the external cavity length.For this WDM a wavelength tunablity of 16 nm is obtained using a drivevoltage of 3 V.

1.4.3 Digital mirror devices

The natural interface to digital video is a digital display, which accepts electricalbits at its input and converts them into optical bits at the output. The digital-to-analog processing function is performed in the mind of the observer. TexasInstruments has developed such a display with the invention of the Digital Micro-mirror DeviceTM.41,42 This digital mirror device (DMD) works as a light switchand is designed and fabricated by CMOS-like processes over a CMOS memorywafer. Each light switch has an aluminum mirror, 16 mm square, that can reflectlight in one of two directions depending on the state of the underlying memorycell. With the memory cell in the (1) state, the mirror switches to þ10 deg.With the memory cell in the (0) state, the mirror switches to210 deg. By combin-ing the DMD with a suitable light source and projection optics, the mirror reflectsincident light either into or out of the pupil of the projection lens. Thus, the (1) stateof the mirror appears bright and the (0) state of the mirror appears dark, and thegrayscale is achieved by binary pulsewidth modulation of the incident light.Color is achieved by using color filters, either stationary or rotating, in combi-nation with one, two, or three DMD chips. Figure 1.13 shows two DMD mirrorswhere one mirror is in the (0) state and another one is in the (1) state.Readers can find more discussions about the Texas Instruments DMD switch—

its processing details, its principle of actuation, its action as optical mirror, itsapplication to display, its simulation and design, and finally its techniques ofpackaging—in Chapters 2, 4, 7, 8, 10, and 11, respectively.

1.4.4 MOEMS scanners

Since the 1990s, many institutes, research organizations, and national labs haveworked on the development of optical scanning chips, using different methodsof actuation, and different sizes, and targeting different applications. Recentadvances in MOEMS, supported by developments of both micro-optics andMEMS, have opened new ways for the miniaturization of scan engines. Manyoptical scanning devices have been developed, and some currently are under inves-tigation.43–50 Among them is a MOEMS scanning engine, which has been undercontinuous development since 1998.44 This MOEM scanner chip was initiallydeveloped at the Rockwell Science Center. The device design is currently in thefinal stage of commercialization at Revoltech Microsystems.Issues under consideration for this scan engine are the scan angle, driving

voltage, structural stability, device size, and mirror flatness. The fabrication and

Introduction 15

assembly processes are CMOS-compatible. The flatness of the mirror is on theorder of a fraction of a wavelength. These issues bear on the manufacturabilityof a device meeting many of the commercial specifications given for differentapplications.Figure 1.14 shows a processed 4-in. wafer containing 60 scanning chips of two

different designs. This is an example of a multichip module with only two chips inthe module. The process is applicable to a scan module with several chipsdesigned for different applications. The chip can be separated and packaged inde-pendently, or the module can be used as one integrated system. For more detailsabout this scanner and other developed types, we direct readers to Chapter 7 of thisbook.

Figure 1.13: Two DMD pixels, where the mirror is shown transparent and rotated. (Image

courtesy of Texas Instruments.)

Figure 1.14: A 4-in. scanner wafer previously processed at Rockwell Science Center.

16 M. E. Motamedi

1.4.5 MOEMS technology applied to telecom

The new advances in fiber optics technology, Internet, and data networks have ledto enormous growth in broadband networking, whose large bandwidth requirementcan be filled only through optical networking.51 Present optical networks are con-fined to point-to-point static links using WDM. MOEMS technology, carryingrecent advances in micro-optics and MEMS, has the potential to revolutionizethe telecom industry by changing the point-to-point static links to an all-opticalmesh network where light paths can be quickly changed on demand.This potential breakthrough in optical networking is driven by the needs of the

carriers for new services, and by trade-offs in cost, bit rate, power consumption,and protocol transparency. To bring all-optical networking to commercialization,a vast demand for inexpensive, yet flexible components is emerging. The routingfunction of optical networking could be achieved using MOEMS components. Inparticular, there is demand for VOAs for power leveling, tunable lasers for wave-length conversion, and all sizes of optical switches including mirror switches,which are the only devices that might lead to the realization of large matrix cross-connects. MOEMS, in most cases, is the only way to develop all of these requiredcomponents. If MOEMS technologies did not exist, it would be almost impossibleto advance today’s telecom to all-optical networking, especially where demand inthe network is not uniform.During 1999–2000, the telecom market evolved significantly, with a vision of a

$110 billion market.52 This prospective market size stimulated many companies inthis field. The recent crash in the economy dramatically changed this prospect,although there are still large investment efforts by venture capitalists in thisfield. It is obvious that fiber optic telecom will become a “killer application” formicrosystems technology in the near future.The optical telecom market is still growing at a rapid rate. NEXUS forecast that

the fiber optic component market (about $5.5 billion in 2000) would grow to about$20 billion in 2005.52 This forecast growth is a result of the huge demand forpicture data traffic on the Internet. The main market driver is the increasingdemand for bandwidth. The market for optical bandwidth management systemswas only $543 million in 2000, but is forecast to reach $15 billion in 2007.In summary, the telecom business is forecast to be extraordinarily active in the

first decade of the twenty-first century. This bright prospect is due to recentadvanced developments in fiber optics, the growth of metropolitan and accessnetworks, and wide implementation of WDM and DWDM technologies todramatically increase the bandwidth. Even though some of these demands havenot been achieved yet, MOEMS may be the only technology bridge to this future.

1.5 Microsystems: Terms and visions

The terms “integrated circuits,” “micromachining,” and “microtechnology” havebeen used worldwide. The term “microsystem” represents a microelectronicsystem that has been enhanced with micromechanics and possibly one or more

Introduction 17