kalpakjian 17

7/27/2019 kalpakjian 17

1/34

Fabrication ofMicroelectro-mechanical Devicesand Systems andNanoscaleManufacturing Many of the processes and materials used for manufacturing microelectronicdevices are also used for manufacturing micromechanical devices and micro-electromechanical systems; this chapter investigates topics in the production ofvery small mechanical and electromechanical products. The chapter beginswith considerations of micromachining and surface machining of mechanicalstructures from silicon. The LIGA process and its variations are then described, along with micromold-ing, EFAB, and various other techniques for replicating small-scale mechanicaldevices. Solid free-form fabrication processes are sometimes suitable for the productionof MEMS and MEMS devices. The chapter ends with a discussion of the emerging area of nanoscalemanufacturing.T ical arts made: Sensors actuators accelerometers o tical switches ink-`et9 9 9printing mechanisms, micromirrors,micromachines, and microdevices.Competing processes: Fine blanking, small scale machining, microforming.

29.| IntroductionThe preceding chapter dealt with the manufacture of integrated circuits and prod-ucts that operate purelyon electrical or electronic principles-called microelectronicdevices. These semiconductor-based devices often have the common characteristicof extreme miniaturization. A large number of devices exist that are mechanicalin nature and are of a similar size as microelectronic devices. A micromechanicaldevice is a product that is purely mechanical in nature and has dimensions betweena few millimeters and atomic length scales, such as some very small gears andhinges.

A microelectromechanical device is a product that combines mechanical andelectrical or electronic elements at these very small length scales. A microelectro-mechanical system MEMS) is a microelectromechanical device that also incorpo-rates an integrated electrical system into one product. Common examples ofmicromechanical devices are sensors of all types Fig. 29.1). Microelectromechanical

29.| Introduction 83|29.2 Micromachining of MEMS

Devices 83329.3 The LIGAMicrofabrication

Process 84429.4 Solid Free-form Fabrication

of Devices 85029.5 Nanoscale

Manufacturing 855EXAMPLES:29.| Surface Micromachining

of a Hinge 83629.2 Operation and Fabrication

Sequence for a ThermalInk-jet Printer 84329.3 Production of Rare-earthMagnets 847CASE STUDIES:29.| Digital Micromirror

Device 83729.2 Accelerometer forAutomotive Air Bags 85|

83|


2/34

832 Chapter 29 Fabrication of Microelectromechanical Devices and Systems and Nanoscale Manufacturing

"\\\._ffs,. ,MW 9 gg; saS ~ f;=e2 ~ cc_ccc *sf =='===f~ V Lii?is `.'fffffTi.' ,%21iii?;,n`;, "M l WW

%'3

~___=_ __=...,. __ I5 ,... Wc fflf. 'fffffifffffi i_ " " T ff? if "i=?f f?52i~ 'f

_' zzz; .1 fz2:>ff1=== EQ ~e 1, ~1


3/34

Section 29.2 Micromachlnmg of MEMS DEVICCS

industrial applications, only a few industries, such as the computer, medical, and au-tomotive industries, have exploited MEMS to date. Many of the processes describedin this chapter have not yet become widespread, but are of interest to researchersand practitioners in MEMS.

29.2 Micromachining of MEMS DevicesThe topics described in the preceding chapter dealt with the manufacture of inte-grated circuits and products that operate purely on electrical or electronic principles.These processes also are suitable for manufacturing devices that incorporatemechanical elements or features as well. The following four types of devices can bemade through the approach described in Fig. 28.21

a. Microelectronic devices are semiconductor-based devices that often have thecommon characteristics associated with extreme miniaturization and use elec-trical principles in their design.

b. Micromechanical devices are products that are purely mechanical in natureand have dimensions between atomic length scales and a few millimeters. Verysmall gears and hinges are examples.c. Microelectromechanical devices are products that combine mechanical andelectrical or electronic elements at very small length scales. Most sensors areexamples of microelectromechanical devices.d. Microelectromechanical systems are microelectromechanical devices that alsoincorporate an integrated electrical system in one product. Microelectro-mechanical systems are rare compared with microelectronic, micromechanical,

or microelectromechanical devices, typical examples being air-bag sensors anddigital micromirror devices.The production of features from micrometers to millimeters in size is calledmicromac/oining. MEMS devices have been constructed from polycrystalline siliconpolysilicon and single-crystal silicon because the technologies for integrated-circuitmanufacture, described in Chapter 28, are well developed and exploited for these

devices, and other, new processes have been developed that are compatible with theexisting processing steps. The use of anisotropic etching techniques allows the fabri-cation of devices with well-defined walls and high aspect ratios; for this reason,some MEMS devices have been fabricated out of single-crystal silicon.One of the difficulties associated with the use of silicon for MEMS devices isthe high adhesion encountered at small length scales and the associated rapid wear.Most commercial devices are designed to avoid friction by, for example, using flex-ing springs instead of bushings. However, this approach complicates designs andmakes some MEMS devices not feasible. Consequently, significant research is beingconducted to identify materials and lubricants that provide reasonable life andperformance.

Silicon carbide, diamond, and metals such as aluminum, tungsten, and nickelhave been investigated as potential MEMS materials. Lubricants also have beeninvestigated. It is known that surrounding the MEMS device, for example, in asilicone oil practically eliminates adhesive wear Section 33.5 , but it also limits theperformance of the device. Self-assembling layers of polymers also are being investi-gated, as well as novel and new materials with self-lubricating characteristics.However, the tribology of MEMS devices remains a main technological barrier toany further expansion of their already widespread use.


4/34

Chapter 29 Fabrication of Microelectromechanical Devices and Systems and Nanoscale ManufacturingSubstrate Diffused layer Nonetching mask Freestanding 111) planese Q n type Si) e.g., p-type Si) e.g., silicon nitride) cantilever I/ ,__,,_ ._ ,,,, .,

V j Ai1. 2. 3.FIGURE 29.2 Schematic illustration of bulk micromachining. 1. Diffuse dopant in desiredpattern. 2. Deposit and pattern-masking film. 3. Orientation-dependent etching ODE) leavesbehind a freestanding structure. Source: Courtesy of K.R. Williams, Agilent Laboratories.

29.2.l Bulk MicromachiningUntil the early 1980s, bulk micromachining was the most common method ofmachining at micrometer scales. This process uses orientation-dependent etches onsingle-crystal silicon see Fig. 28.15b), an approach that depends on etching downinto a surface and stopping on certain crystal faces, doped regions, and etchablefilms to form a desired structure. As an example of this process, consider the fabri-cation of the silicon cantilever shown in Fig. 29.2. Using the masking techniquesdescribed in Section 28.7, the process changes a rectangular patch of the 11-typesilicon substrate to p-type silicon through boron doping. Etchants such as potassiumhydroxide will not be able to remove heavily boron doped silicon; hence, this patchWill not be etched.

A mask is then produced-for example, with silicon nitride on silicon. Whenetched with potassium hydroxide, the undoped silicon will be removed rapidly,while the mask and the doped patch will be essentially unaffected. Etching pro-gresses until the 111) planes are exposed in the n-type silicon substrate; they under-cut the patch, leaving a suspended cantilever as shown.29.2.2 Surface MicromachiningBulk micromachining is useful for producing very simple shapes. It is restricted tosingle-crystal materials, since polycrystalline materials will not machine at differentrates in different directions when wet etchants are used. Many MEMS applicationsrequire the use of other materials; hence, alternatives to bulk micromachining areneeded. One such alternative is surface micromachining. The basic steps in surfacemicromachining are illustrated for silicon devices in Fig. 29.3. A spacer or sacrificiallayer is deposited onto a silicon substrate coated with a thin dielectric layer calledan isolation, or buffer, layer).Phosphosilicate glass deposited by chemical-vapor deposition is the most com-mon material for a spacer layer, because it etches very rapidly in hydrofluoric acid.Step 2 in Fig. 29.3 shows the spacer layer after the application of masking and etch-ing. At this stage, a structural thin film is deposited onto the spacer layer; the filmcan be polysilicon, metal, metal alloy, or a dielectric step 3 in Fig. 29.3). The struc-tural film is then patterned, usually through dry etching, in order to maintain verti-cal Walls and tight dimensional tolerances. Finally, wet etching of the sacrificial layerleaves a freestanding, three-dimensional structure, as shown in step 5 of Fig. 29.3.Note that the wafer must be annealed to remove the residual stresses in the depositedmetal before it is patterned; otherwise the structural film will severely Warp once thespacer layer is removed.


5/34

Section 29.2 Micromachining of MEMS Devices 835Phosphosilicate glass(spacer layer) Polysilicon

Silicon

if .ffxi' ,r

1. 2. 3.

Ti' Suspended 'ffe ,S cantilever4. 5.FIGURE 29.3 Schematic illustration of the steps in surface micromachining: 1. Depositionof a phosphosilicate glass (PSG) spacer layer; 2. Lithography and etching of spacer layer;3. Deposition of polysilicon; 4. Lithography and etching of polysilicon; 5. Selective wetetching of PSG, leaving the silicon substrate and deposited polysilicon unaffected.Figure 29.4 shows a microlamp that emits a white Film 2/.tm thicklight when current is passed through it; it has been pro- |

duced through a combination of surface and bulk micro-machining. The top patterned layer is a 2.2-/.tm-thick layerof plasma-etched tungsten, forming a meandering filamentand bond pad. The rectangular overhang is dry-etchedsilicon nitride. The steeply sloped layer is wet-HF-etchedphosphosilicate glass. The substrate is silicon, which isorientation-dependent etched.The etchant used to remove the spacer layer must beselected carefully. It must preferentially dissolve the spacerlayer while leaving the dielectric, silicon, and structuralfilm as intact as possible. With large features and narrowspacer layers, this becomes a very difficult task, andetching can take many hours. To reduce the etching time,additional etched holes can be designed into the mi-crostructures to increase access of the etchant to the spacer ,s ,,.,..,..y ,,_ ...... .,,.,,_,_, f,layer. Another difficulty that must be overcome is stiction |=|;u|1529_4 A microlamp produced from a Combiafter wet etching. Consider the situation illustrated in nation of bulk and surface micromachining processesFig. 29.5 _ After the spacer layer has been removed, the liq- Source: Courtesy of K.R. Williams, Agilent Technologiesuid etchant is dried from the wafer surface. A meniscusforms between the layers and results in capillary forces thatcan deform the film and cause it to contact the substrate as the liquid evaporates.Since adhesive forces are more significant at small length scales, it is possible thatthe film may stick permanently to the surface; thus, the desired three-dimensionalfeatures will not be produced.


6/34

836 Chapter 29 Fabrication of Microelectromechanical Devices and Systems and Nanoscale ManufacturingPolysilicon beam Spacer oxide Rinse Water

. 2.

3.

FIGURE 29.5 Stiction after wet etching: 1. Unreleased beam; 2. Released beam beforedrying; 3. Released beam pulled to the surface by capillary forces during drying. Once Contactis made, adhesive forces prevent the beam from returning to its original shape. Source: AfterB. Bhushan.

EXAMPLE 29.l SurfaceMicromachining of a HingeSurface micromachining is a widespread technol-ogy for the production of MEMS. Applicationsinclude accelerometers, pressure sensors, microp-umps, micromotors, actuators, and microscopiclocking mechanisms. Qften, these devices requirevery large vertical walls, which cannot be manufac-tured directly because the high vertical structure isdifficult to deposit. This obstacle is overcome by `thmachining large, flat structures horizontally andthen rotating or folding them into an upright posi- astion, as shown in Fig. 29.6.

alFIGURE 29.6 a) SEM image of a deployed micromirron b) Detail hinge.Source: Courtesy of Sandia National Laboratories.


7/34

Section 29.2 Micromachining of MEMS Devices

Spacer |ayer1 Polyi Spacer \ayer2

Silicon1. 2. 3.

Poly2

4. 5.FIGURE 29.7 Schematic illustration ofthe steps required to manufacture a hinge. 1. Depositionof a phosphosilicate glass PSG spacer layer and polysilicon layer see Fig. 293 . 2. Depositionof a second spacer layer. 3. Selective etching of the PSG. 4. Deposition of polysilicon to form astaple for the hinge. 5. After selective wet etching ofthe PSG, the hinge can rotate.

837

Figure 29.7 shows the cross section of a hingeduring its manufacture. The following steps are in-volved in the production of the hinges:I.

2.

3.

A 2-/im-thick layer of phosphosilicate glassPSG is first deposited onto the substratematerial.A 2-/.im-thick layer of polysilicon Polyl in step1 in Fig. 29.7 is deposited onto the PSG, pat-terned by photolithography, and dry etched toform the desired structural elements, includingthe hinge pins.A second layer of sacrificial PSG with a thick-ness of 0.5 /.im is deposited step 2 in Fig. 29.7 .

4. The connection locations are etched throughboth layers of PSG step 3 in Fig. 29.7 .5. A second layer of polysilicon Poly2 in step 4 in

Fig. 29.7 is deposited, patterned, and etched.6. The sacrificial layers of PSG are removed bywet etching.

Hinges such as these have very high friction.If mirrors as shown in Fig. 29.6 are manipulatedmanually and carefully with probe needles, they willremain in position. Often, such mirrors will be com-bined with linear actuators to control their deploy-ment precisely.

CASE STUDY 29.l Digital Micromirror DeviceAn example of a commercial MEMS-based product digital micromirror devices DMDTM to project ais the digital pixel technology DPTTM device, illus- digital image, as in computer-driven projection sys-trated in Fig. 29.8. This device uses an array of tems. The aluminum mirrors can be tilted so that


8/34


Mirror'Qr-10 Hinge Mirror +10

Mirror support mo-postLanding tips

625325553 QYoke AddresselectrodeTorsion hinge occ fElectrodeHIUQG 3 Q support postsupport postLandingsites|\/|eta| Yoke banding CMOSaddress pads P SU*-S eBias/Reset

US y yTo SRAM

(al lb)

t.=_ .f f' _r cc. , T ~t;;= 1|`e l .T_ * H M it', f -1' .-*rf V,,.=`;,_/ , gy). / _, y/, ; i ~ ,, V Lf/ ~ /i/ .V p 2 mmi s $3

C) (dlFIGURE 29.8 The Texas Instruments digital pixel technology (DPTTM) device. (a) Exploded view of a singledigital rnicromirror device (DMDTM). (b) View of two adjacent DMD pixels. (c) Images of DMD arrays withsome mirrors removed for clarity; each mirror measures approximately 17 /.tm on a side. (d) A typical DPTdevice used for digital projection systems, high-definition televisions, and other image display systems. Thedevice shown contains 1,310,720 micromirrors and measures less than 50 mm per side. Source: Courtesy ofTexas Instruments Corp.

light is directed into or away from the optics that focus time is about 15 pts (which is much faster than thelight onto a screen. That way, each mirror can repre- human eye can respond), the mirror will switch be-sent a pixel of an images resolution. The mirror al- tween the on and off states in order to reflect thelows light or dark pixels to be projected, but levels of proper dose of light to the optics.gray also can be accommodated. Since the switching


9/34

CMOS Sacrificial CMPmetal level spacer-1 oxidemi-E-if-E3E' ",1*i.Q.-::.E.E a2:-: E?fEE3e2il 3e1:a "$2ss.s.::.t.,.;z=, ,masse 'Silicon substrate with CMOS circuits1. Pattern spacer-1 layer

Yoke metal Oxide mask

3. Deposit yoke and patternyoke oxide maskMirror Mirrormask Spacer-2

5. Deposit spacer-2 and mirror

Section 29.2 Micromachining of MEMS Devices

Oxide hinge mask Hinge metal. Deposit hinge metal; depositand pattern oxide hinge mask

Hinge Yoke Hinge post

4. Etch yoke and strip oxide

Mirror Mirror postYoke Hinge

6. Pattern mirror and etchedsacrificial spacersFIGURE 29.9 Manufacturing sequence for the Texas Instruments DMD device.

The fabrication steps for producing the DMDdevice are shown in Fig. 29.9. This sequence is similarto that of other surface micromachining operations,but has the following important differences:

All micromachining steps take place at tempera-tures below 400C--low enough to ensure thatno damage occurs to the electronic circuit.A thick silicon-dioxide layer is deposited and ischemical-mechanical polished to provide anadequate foundation for the MEMS device.The landing pads and electrodes are producedfrom aluminum, which is deposited by sputtering.High reliability requires low stresses and highstrength in the torsional hinge, which is producedfrom a proprietary aluminum alloy.

The MEMS portion of the DMD is very delicate,and special care must be taken in separatingthe dies. When completed, a wafer saw (seeFig. 28.6c) cuts a trench along the edges ofthe DMD, which allows the individual dice to bebroken apart at a later stage. A special step deposits a layer that prevents ad-hesion between the yoke and landing pads. The DMD is placed in a hermetically sealedceramic package (Fig. 2910) with an opticalwindow.

An array of such mirrors represents agrayscale screen. Using three mirrors (one each forred, green, and blue light) for each pixel results in acolor image with millions of discrete colors. Digital

839


10/34


Hermetic optical window DMD(Corning 7056)EE;';;;eta| Gold wire bondsS ,d K a' ameeam We Kovar seal ringge.; Z;EiiIFEIS I Z tt rCeramic header 9 9 99 9 SHeat sinkFIGURE 29.10 Ceramic flat-package construction used for the DMD device.

pixel technology is applied widely in digital projection in Fig. 29.8 requires much more than two-and-onesystems, high-definition television, and other optical half-dimensional features, so full three-dimensional,applications. However, to produce the device shown multipart assemblies have to be manufactured.

SCREAM. Another method for making very deep MEMS structures is theSCREAM (single-crystal silicon reactive etching and metallization) process, depictedin Fig. 29.11. In this technique, standard lithography and etching processes producetrenches 10 to 50 /,tm deep, which are then protected by a layer of chemically vapor

Photoresist Oxide . Deposit oxide and 2. Lithography andphotoresist oxide etching

4- Coat Sidel/\(8||S with 5. Remove oxide at bottom 6.PECVD OXIGG and etch siliconFIGURE 29.l I The SCREAM process. Source: After N. Maluf.

3. Silicon etching

Suspended Sharpbeam

VI

Plasma etching in SF6to release structures


11/34

Section 29.2 Micromachining of MEMS DevicesPhotoresist Oxide Suspended feature edlz -` _ n+P' Silicon 1. Deposit oxide and 2. Lithography and 3. Plasma etching of 4- |SffV0PlC et'-Qijlflg Ofphotoresist on oxide etching p doped silicon fl doped Sl||C0nlayered substrateFIGURE 29.l2 Schematic illustration of silicon micromachining by single-step plasma etching(SIMPLE) process.

deposited silicon oxide. An anisotropic~etching step removes the oxide only at thebottom of the trench, and the trench is then extended through dry etching. Anisotropic etching step (using sulfur hexafluoride, SF6) laterally etches the exposedsidewalls at the bottom of the trench. This undercut (when it overlaps adjacent un-dercuts) releases the machined structures.

SIMPLE. An alternative to SCREAM is the SIMPLE (silicon niicronmchining bysingle-step plasma etching) technique, as depicted in Fig. 29.12. This techniqueuses a chlorine-gas-based plasma-etching process that machines p-doped or lightlydoped silicon anisotropically, but heavily n-doped silicon isotropically. A sus-pended MEMS device can thus be produced in one plasma-etching device, asshown in the figure.

Some of the concerns with the SIMPLE process are as follows: The oxide mask is machined, although at a slower rate, by the chlorine-gasplasma. Therefore, relatively thick oxide masks are required. The isotropic etch rate is low, typically 50 nm/min. Consequently, this is a veryslow process. The layer beneath the structures will have developed deep trenches, which mayaffect the motion of free-hanging structures.

Etching Combined with Diffusion Bonding. Very tall structures can be producedin crystalline silicon through a combination of silicon-diffusion bonding and deepreactive-ion etching (SEB-DRIE), as illustrated in Fig. 29.13. First, a silicon wafer isprepared with an insulating oxide layer, with the deep trench areas defined by astandard lithography procedure. This step is followed by conventional wet or dryetching to form a large cavity. A second layer of silicon is fusion bonded to theoxide layer, and the second silicon layer can be ground and lapped to the desiredthickness if necessary. At this stage, integrated circuitry is manufactured through thesteps outlined in Fig. 28.2. A protective resist is applied and exposed, and the de-sired trenches are etched by deep reactive-ion etching to the cavity in the first layerof silicon.


12/34

ResistOxide

Silicon

I-.:.:~' I W....i.i.,iiii,i. .,,......_._...M.,,,,, a i 3,1 oxide silicon f ~iii sgtf_@i*ZF* I.-11 1 T>2Q`l'll,[J sigfmri Ie $iQ9$ffff1`ffl lfii weI'Embedded cavity1. Expose resist 2. Etch cavity 3. Silicon-diffusion bondingResist

CMOS circuits Suspended beaml 1 k 35:1 I 1 /,. __ __ V_______ f~f_~~~_~_ `_ ,V ., ggiipf , 122 23313I IIIIIII I A `ll. Fabricate CMOS 5. Expose resist 6. Etch (DRIE) beam(H)

100 Mm(D)

FIGURE 29.l3 (a) Schematic illustration of silicon-diffusion bonding combined with deepreactive-ion etching to produce large, suspended cantilevers. (b) A microfluid-flow devicemanufactured by DRIE etching two separate wafers and then aligning and silicon-fusionbonding them together. Aftervvard, a PyreX layer (not shown) is anodically bonded over thetop to provide a window to observe fluid flow. Source: (al After N. Maluf. (b) Courtesy ofK.R. Williams, Agilent Technologies.


13/34

Section 29.2 Micromachining of MEMS Devices 843

EXAMPLE 29.2 Operation and Fabrication Sequence for a Thermal Ink-jet PrinterThermal ink-jet printers are perhaps the most success-ful application of MEMS to date. These printers oper-ate by ejecting nano- or picoliters 1042 liter of inkfrom a nozzle towards paper. Ink-jet printers use avariety of designs, but silicon-machining technologyis most applicable to high-resolution printers. Notethat a resolution of 1200 dpi dots per inch requiresa nozzle spacing of approximately 20 nm.

The mode of operation of an ink-jet printer isshown in Fig. 29.14. When an ink droplet is to be gen-erated and expelled, a tantalum resistor below a noz-zle is heated. This heats a thin film of ink, so that abubble forms within 5 microseconds, with internalpressures reaching 1.4 MPa. The bubble then expandsrapidly, and as a result, fluid is forced rapidly out ofthe nozzle. Within 24 us, the tail of the ink-jet dropletseparates because of surface tension, the heat source isremoved turned off , and the bubble collapses insidethe nozzle. Within 50 pcs, sufficient ink has beendrawn into the nozzle from a reservoir to form the de-sired meniscus for the next droplet.Traditional ink-jet printer heads were made withelectroformed nickel nozzles, produced separately

lnk Bubble

Heating element1. Actuation

3. Droplet ejection

from the integrated circuitry, so a bonding operationwas required to attach these two components. Withincreasing printer resolution, it is more difficult tobond the components with a tolerance of less than afew micrometers. For this reason, singlecomponent,or monolithic, fabrication is of interest.The fabrication sequence for a monolithic ink-jet printer head is shown in Fig. 29.15. A siliconwafer is prepared and coated with a phosphosilicate-glass PSG pattern and a low-stress silicon-nitridecoating. The ink reservoir is obtained by isotropicallyetching the back side of the Wafer, followed byPSG removal and enlargement of the reservoir. Therequired CMOS complementary metal-oxide semi-conductor controlling circuitry is then produced, anda tantalum heater pad is deposited. The aluminuminterconnection between the tantalum pad and theCMOS circuit is formed, and the nozzle is producedthrough laser ablation. An array of such nozzles canbe placed inside an ink-jet printing head, and resolu-tions of 2400 dpi 95 dots per mm or higher can beachieved.

is2. Droplet formation

Satellite droplets

0->4. Liquid refillsFIGURE 29.l4 Sequence of operation of a thermal ink-jet printer. 1. Resistive heating elementis turned on, rapidly vaporizing ink and forming a bubble. 2. Within 5 /zm, the bubble hasexpanded and displaced liquid ink from the nozzle. 3. Surface tension breaks the ink stream intoa bubble, which is discharged at high velocity. The heating element is turned off at this time, sothat the bubble collapses as heat is transferred to the surrounding ink. 4. Within 24 /.ts, an inkdroplet and undesirable satellite droplets are ejected, and surface tension of the ink draws moreliquid from the reservoir. Source: After F.-G. Tseng.


14/34

Chapter 29 Fabrication of Microelectromechanical Devices and Systems and Nanoscale Manufacturing

Phosphosilicate Siliconglass nitride ilicon dioxide

1. Silicon-nitride deposition 2. Wet-etch manifold, 3. Wet-etch, enlargeremove PSG chamberTantalum Aluminumheater interconnect

4. Heater and interconnection 5_ Laser mZZ|eformulationFIGURE 29.15 The manufacturing sequence for producing thermal ink-jet printer heads.Source: After E-G.Tseng.

29.3 The LIGA Microfabrication ProcessLIGA is a German acronym for the combined process of X-ray lithography, elec-trodeposition, and molding in German, X-ray lithographic, galvanoformung, undabformung). A schematic illustration of this process is given in Fig. 29.16.The LIGA process involves the following steps:l. A very thick up to hundreds of microns) resist layer of polymethylmethacry-late PMMA) is deposited onto a primary substrate.

2. The PMMA is exposed to columnated X-rays and is developed.3. Metal is electrodeposited onto the primary substrate.4. The PMMA is removed or stripped, resulting in a freestanding metal structure.5. Plastic injection molding takes place.Depending on the application, the final product from a LIGA process mayconsist of

A freestanding metal structure resulting from the electrodeposition process. A plastic injection-molded structure. An investment-cast metal part, using the injection-molded structure as ablank. A slip-cast ceramic part, produced with the injection-molded parts as themolds.The substrate used in LIGA is a conductor or a conductor-coated insulator.Examples of primary substrate materials include austenitic steel plate, silicon wafers


15/34

/ 1 _LSynchrotron radiationMask membraneAbsorber structurePMMA resistl l l l l,ifr y&. Irradiation'iii2. Developing3 Electroforming

Substrate

PMMA Structure

MetalPMMA Structure

Section 29.3 The LIGA Microfabrication Process 8

'? >~ \ );

1. Mold insert (from LIGA)

2. Mold filling

3. Mold removal? V;t;,. fi.7,;;.=; ; ;g;i @;; Q,,

4. Second electroforming

Mold cavity

PlasticElectricallyconductivesubstrate

Plastic structureas mold or finalproduct

Metal fromelectroforming orceramic from slipcasting

Metal or ceramicstructureFinal product Mold insert4. Resist removal 5. Final product

(2) (D)FIGURE 29.l6 The LIGA (lithography, electrodeposition and molding) technique. (a) Primaryproduction of a metal final product or mold insert. (b) Use of the primary part for secondaryoperations or replication. Source: Courtesy of IMM Institut fur Mikrotechnik, Mainz,Germany.

with a titanium layer, and copper plated with gold, titanium, or nickel. Metal-platedceramic and glass also have been used. The surface may be roughened by grit blast-ing to encourage good adhesion of the resist material.Resist materials must have high X-ray sensitivity, dry- and Wet-etching resist-ance when unexposed, and thermal stability. The most common resist material is poly-methylmethacrylate, which has a very high molecular Weight (more than 106 gramsper mole). The X-rays break the chemical bonds, leading to the production of free rad-icals and to a significantly reduced molecular Weight in the exposed region. Organicsolvents then preferentially dissolve the exposed PMMA in a Wet-etching process.After development, the remaining three-dimensional structure is rinsed and dried, or itis spun and blasted with dry nitrogen.

Two newer forms of LIGA are UV-LIGA and Silicon-LIGA. In UVLIGA, specialphotoresists are used instead of PMMA, and they are exposed through ultraviolet


16/34

8 Chapter 29 Fabncation of Microelectromechanical Devices and Systems and Nanoscale Manufacturinglithography Section 28.7 . Silicon-LIGA uses deep reactive-ion-etched silicon Sec-tion 28.8.2 as a preform for further operations. These processes, like the traditionalX-ray-based LIGA, are used to replicate MEMS devices, but, unlike LIGA, they do notrequire the expensive columnated X-ray source for developing their patterns.The electrodeposition of metal usually involves the electroplating of nickel.The nickel is deposited onto exposed areas of the substrate; it fills the PMMA struc-ture and can even coat the resist Fig. 29.16a . Nickel is the material of choice be-cause of the relative ease in electroplating with vvell-controlled deposition rates.Electroless plating of nickel also is possible, and the nickel can be deposited directlyonto electrically insulating substrates. However, because nickel displays high Wearrates in MEMS, significant research has been directed towards the use of other ma-terials or coatings.After the metal structure has been deposited, precision grinding removes eitherthe substrate material or a layer of the deposited nickel. The process is referred to asplanarization Section 28.10 . The need for planarization is obvious when it is rec-ognized that three-dimensional MEMS devices require micrometer tolerances onlayers many hundreds of micrometers thick. Planarization is difficult to achieve:Conventional lapping leads to preferential removal of the soft PMMA and smearingof the metal. Planarization usually is accomplished with a diamond-lapping proce-dure referred to as mmogrinding. Here, a diamond-slurry-loaded, soft-metal plate isused to remove material in order to maintain flatness Within 1 /.um over a 75-mmdiameter substrate.

If cross linked, the PMMA resist is then exposed to synchrotron X-ray radia-tion and removed by exposure to an oxygen plasma or through solvent extraction.The result is a metal structure, which may be used for further processing. Examplesof freestanding metal structures produced through the electrodeposition of nickelare shown in Fig. 29.17.The processing steps used to make freestanding metal structures are extremelytime consuming and expensive. The main advantage of LIGA is that these structuresserve as molds for the rapid replication of submicron features through molding op-erations. The processes that can be used for producing micromolds are shown andcompared in Table 29.1; it can be seen that LIGA provides some clear advantages.Reaction injection molding, injection molding, and compression molding describedin Chapter 19 also have been used to make these micromolds.

G D

FIGURE 29.l1 a Electroformed, 200-,um-tall nickel structures; b Detail of 5-/.tm-widenickel lines and spaces. Source: After T. Christenson, Sandia National Laboratories.


17/34

Section 29.3 The LIGA Microfabrication Process 847TABLE 29.|

Comparison of Mierumold Manufaetuting Techniques JProduction technique

Characteristic LIGA Laser machining EDMAspect ratio 10-50 10 up to 100Surface roughness


18/34


H D

FIGURE 29.19 SEM images of Nd2Fe14B permanentmagnets. The powder particlesize ranges from 1 to 5 um, and the binder is a methylene-chloride-resistant epoxy.Mild distortion is present in the image due to magnetic perturbation of the imagingelectrons. Maximum energy products of 9 MGOe have been obtained with thisprocess. Source: Courtesy of T. Christenson, Sandia National Laboratories.

mold is produced by exposure to X-ray radiation andsolvent extraction. The rare-earth powders are mixedwith a binder of epoxy and applied to the moldthrough a combination of calendering see Fig. 1922and pressing. After curing in a press at a pressurearound 70 MPa, the substrate is planarized. The sub-strate is then subjected to a magnetizing field of at

least 35 kilo-oersteds kOe in the desired orientation.Once the material has been magnetized, the PMMAsubstrate is dissolved, leaving behind the rare-earthmagnets, as shown in Fig. 29.19.Source: Courtesy of T. Christenson, Sandia NationalLaboratories.

Multilayer X-ray Lithography. The LIGA technique is very powerful for produc-ing MEMS devices with large aspect ratios and reproducible shapes. It is often use-ful to obtain a multilayer stepped structure that cannot be made directly throughLIGA. For nonoverhanging geometries, direct plating can be applied. In this tech-nique, a layer of electrodeposited metal with surrounding PMMA is produced aspreviously described. A second layer of PMMA resist is then bonded to this struc-ture and X-ray exposed with an aligned X-ray mask.Often, it is useful to have overhanging geometries Within complex MEMS de-vices. A batch diffusion-bonding and release procedure has been developed for thispurpose, as is schematically illustrated in Fig. 29.20a. This process involves thepreparation of two PMMA patterned and electroformed layers with the Pl\/IMAsubsequently removed. The wafers are then aligned face to face with guide pins thatpress-fit into complementary structures on the opposite surface. Finally, the sub-strates are joined in a hot press, and a sacrificial layer on one substrate is etchedaway, leaving one layer bonded to the other. Figure 29.20b shows an example ofsuch a structure.HEXSIL. The HEXSIL process, illustrated in Fig. 29.21, combines hexagonal hon-eycomb structures, silicon micromachining, and thin-film deposition to producehigh-aspect-ratio, freestanding structures. HEXSIL can produce tall structures witha shape definition that rivals that of structures produced by LIGA.

In HEXSIL, a deep trench first is produced in single~crystal silicon by dryetching, followed by shallow wet etching to make the trench walls smoother. Thedepth of the trench matches the desired structure height and is limited practically toaround 100 um. An oxide layer is then grown or deposited onto the silicon, fol-lowed by an undoped-polycrystalline silicon layer, which leads to good mold filling


19/34

Sacrificial ~ ~ as;??i43?l731e5U?3$T?`afeve.e.e Jr t


20/34


H D

FIGURE 29.22 a SEM image of microscale tweezers used in microassembly andmicrosurgery applications. b Detailed view of gripper. Source: Courtesy of MEMS PrecisionInstruments www.memspi.com .

and good shape definition. A doped-silicon layer follows, providing a resistive por-tion of the microdevice. Electroplated or electroless nickel plating is then deposited.Figure 29.21 shows various trench widths to demonstrate the different structuresthat can be produced in HEXSIL.Microscale tweezers produced through the HEXSIL process are shown inFig. 29.22. A thermally activated bar activates the tweezers, which have been used formicroassembly and microsurgery applications.

29.4 Solid Free-form Fabrication of DevicesSolid free-form fabrication is another term for rapid prototyping, as described inChapter 20. This method is unique in that complex three-dimensional structures areproduced through additive manufacturing, as opposed to material removal. Many ofthe advances in rapid prototyping also are applicable to MEMS manufacture for pro-cesses with sufficiently high resolution. Recall that stereolithography Section 20.3.2involves curing a liquid thermosetting polymer using a photoinitiator and a highlyfocused light source. Conventional stereolithography uses layers between 75 and500 /zm in thickness, with a laser dot focused to a 0.05-0.25-mm diameter.Microstereolithography. Microstereolit/vography uses the same basic approach asstereolithography. However, there are some important differences between the twoprocesses, including the following:

The laser is more highly focused to a diameter as small as 1 /rm, comparedwith 10 to over 100 ,um in stereolithography . Layer thicknesses are around 10 /rm, which is an order of magnitude smallerthan in stereolithography. The photopolymers used must have much lower viscosities to ensure the for-mation of uniform layers. Support structures are not needed in microstereolithography, since the smallerstructures can be supported by the fluid.0 Parts with significant metal and ceramic content can be produced by suspend-

ing nanoparticles in the liquid photopolymer.


21/34

Developed mask Elastomeric(elastomer) insulator oo, o

A 1BSubstrate Plating Substratebath

1. 2.

Section 29.4 Solid Free-form Fabrication of Devices 85|Seleotivelydeposited material

Substrate3.

FIGURE 29.23 The instant-masking process: 1. Bare substrate; 2. During deposition, with thesubstrate and instant mask in contact; 3. The resulting pattern deposited. Source: Courtesy ofA. Cohen, MEMGen Corporation.

The microstereolithography technique has a number of cost advantages, butthe MEMS devices made by this method are difficult to integrate with the control-ling circuitry.Electrochemical Fabrication. Instant masking is a technique for producing MEMSdevices (Fig. 29.23). The solid free-form fabrication of MEMS devices using instantmasking is known as electrochemical fabrication (EFAB). A mask of elastomericmaterial is first produced through conventional photolithography techniques,described in Section 28.7. The mask is pressed against the substrate in an electrode-position bath, so that the elastomer conforms to the substrate and excludes the plat-ing solution in contact areas. Electrodeposition takes place in areas that are notmasked, eventually producing a mirror image of the mask. By using a sacrificialfiller made of a second material, instant-masking technology can produce complexthree-dimensional shapes complete with overhangs, arches, and other features.

CASE STUDY 29.2 Accelerometer for Automotive Air BagsAccelerometers based on lateral resonators representthe largest commercial application of surface microma-chining today and are used widely as sensors forautomotive air-bag deployment systems. The sensorportion of such an accelerometer is shown inFig. 29.24. A central mass is suspended over thesubstrate, but anchored through four slender beams,which act as springs to center the mass under static-equilibrium conditions. An acceleration causes themass to deflect, reducing or increasing the clearance be-tween the fins on the mass and the stationary fingers onthe substrate. By measuring the electrical capacitancebetween the mass and fins, the deflection of the mass(and therefore the acceleration or deceleration of thesystem) can be directly measured. Figure 29.24 showsan arrangement for the measurement of acceleration inone direction, but commercial sensors employ several

masses so that accelerations can be measured in multi-ple directions simultaneously.Figure 29.25 shows the 50-g surface microma-chined accelerometer (ADXL-50) with onboard sig-nal conditioning and self-diagnostic electronics. Thepolysilicon sensing element (visible in the center ofthe die) occupies only 5 of the total die area, and thewhole chip measures 500 >< 625 /tm. The mass is ap-proximately 0.3 /ag, and the sensor has a measure-ment accuracy of 5 over the i50-g range.Fabrication of the accelerometer proved to be achallenge, since it required a complementary metal-oxide-semiconductor (CMOS) fabrication sequenceto be integrated closely with a surface microma-chining approach. Analog Devices, Inc., was able tomodify a CMOS production technique to directlyincorporate surface micromachining. In the sensor


22/34


Stationary1 ol s |`conf`n ers -51 ,as p Y H 9 Suspended inertial mass ,r.1 Q/ Direction of accelerationJ ~-we fe ...ff h f llll C1 C2Anchor tosubstrateFIGURE 29.24 Schematic illustration of a microacceleration sensor. Source: AfterN. Maluf.

FIGURE 29.25 Photograph of Analog DevicesADXL-50 accelerometer with a surface micro-machined capacitive sensor (center), on-chip excita-tion, and self-test and signal-conditioning circuitry.The entire chip measures 0.500 >< 0.625 mm.Source: From R.A. Core, et al., Solid State Technology,pp. 39-47, October 1993.

design, n+ underpasses connect the sensor area to right after a borosilicate-glass planarization.the electronic circuitry, replacing the usual heat- After the planarization, a designated sensor re-sensitive aluminum connect lines. Most of the sensor gion, or moat, is cleared in the center of the die (step 1processing is inserted into the fabrication process in Fig. 2926). A thin oxide is then deposited to passi-


23/34

Section 29.4 Solid Free-form Fabrication of Devices 853

THOX ari it wwwPSGbep- P1.LTOLPCVD nitride J, ,,.~f ' '~ ,-;,_asf W ll BpSGTHOX ` ` '`'Ql1;. _s-~ PP

2.

Depression in _,A spacer oxideSpacer LTO , - *~..rir,, IBPSG ; HOX i~.,.... n+ runner mmP pe PAnchor opening

3.

FIGURE 29.26 Preparation of IC chip for polysilicon. 1. Sensor areapost-borophosphosilicate glass (BPSG) planarization and moat mask.2. Blanket deposition of thin oxide and thin nitride layer. 3. Bumpsand anchors made in low-temperature oxide (LTO) spacer layer.Source: After R.A. Core.

vate the 11+ underpass connects and is followed by athin, low-pressure chemical-vapor deposited (LPCVD)nitride to act as an etch stop for the final polysiliconreleased etching (step 2 in Fig. 29.26). The spacer orsacrificial oxide used is a 1.6-pm densified low-tem-perature oxide (LTO) deposited over the whole die(step 3 in Fig. 29.26).

In a first etching, small depressions that willform bumps or dimples on the underside of the poly-silicon sensor are created in the LTO layer. Thesebumps will limit adhesive forces and sticking in casethe sensor comes in contact with the substrate. A sub-sequent etching cuts anchors into the spacer layer toprovide regions of electrical and mechanical contact

(step 3 in Fig. 29.26). The 2-,um thick sensor of poly-silicon layer is deposited, implanted, annealed, andpatterned (step 1 in Fig. 29.27).Metallization follows, starting with the removal

of the sacrificial spacer oxide from the circuit areaalong with the LPCVD nitride and LTO layer. A low-temperature oxide is deposited on the polysilicon-sensor part, and contact openings appear in the ICpart of the die, where platinum is deposited to formplatinum silicide (step 2 in Fig. 29.27). The trimmablethin-film material, TiW barrier metal, and Al-Cu in-terconnect metal are sputtered on and patterned inthe IC area.

The circuit area is then passivated in two sepa-


24/34

854 Chapter 29

BPSGSpacer LTO

,/ .._,f __,__,,_,,_,_

Sensorpolysilicon

Fabrication of Microelectromechanical Devices and Systems and Nanoscale Manufacturing

L..-.......... ,..r__s,_sr__r__.__.THOX e s................W n+ runnerP p~ D1.

Contact opening Sensor(Pt silicide forms here) polysiliconHOX ,_ + ...W ..._we - n runnerP p- P2.

Plasma oxide Sensorpolysilicon

THOX /5 SSSSQSSS ff~~...- n+ runnerP p~ P3.

FIGURE 29.27 Polysilicon deposition and IC metallization. 1. Cross-sectional View after polysilicon deposition, implanting, annealing, andpatterning. 2. Sensor area after removal of dielectrics from circuit area,contact mask, and platinum silicide. 3. Metallization scheme and plasma-oxide passivation and patterning. Source: After R.A. Core.

rate deposition steps. First, plasma oxide is depositedand patterned (step 3 in Fig. 29.27), followed by aplasma nitride (step 1 in Fig. 29.28), to form a sealwith the previously deposited LCVD nitride. The ni-tride acts as a hydrofluoric~acid barrier in the subse-quent etch release in surface micromachining. Theplasma oxide left on the sensor acts as an etch stop forthe removal of the plasma nitride (step 1 in Fig. 29.28).The sensor area is then prepared for the final release

etch. The dielectrics are removed from the sensor, andthe final protective resist mask is applied. The photore-sist protects the circuit area from the long-termbuffered oxide etch (step 2 in Fig. 29.28). The finaldevice cross section is shown in step 3 in Fig. 29.28.Source: Adapted from M. Madou, Fundamentals ofMicrofabrication, 2nd ed., CRC Press, 2002.


25/34

Section 29.5 Nanoscale Manufacturing

MET Plasma Plasmanitride oxide LTOBPSG ..,_ Sensor polysilicon t rip i ly ,, ;, _ =.2f;THOX l Spae'LTO........~f n+ runner MP p p1.MET PhotoresistHOX .s- n runnerP p- p2.METBPSGTHOX

fum

n+ runner

it Sensor polysilicon ..,,,,P p- p

3.FIGURE 29.28 Prerelease preparation, and release, 1. Post-plasma nitride passivationand patterning. 2. Photoresist protection of the IC. 3. Freestanding, released polysiliconbeam. Source: After R.A. Core.

29.5 Nanoscale ManufacturingIn nanomanufzzcturing, parts are produced at nanometer length scales. The termusually refers to manufacturing strategies below the micrometer scale, or between10% and 10`9 m in length. Many of the features in integrated circuits are at thislength scale, but very little else with significant manufacturing relevance is.Molecularly engineered medicines and other forms of biomanufacturing are the onlycommercial applications at present. However, it has been recognized that manyphysical and biological processes act at this length scale; consequently, the approachholds much promise for future innovation.Nanoscale manufacturing techniques are outlined in Table 29.3. Nano-manufacturing takes two basic approaches: top down and bottom up. Top-downapproaches use large building blocks (such as a silicon wafer; see Fig. 282) andvarious manufacturing processes (such as lithography, and wet and plasma etching)to construct ever smaller features and products (microprocessors, sensors, andprobes). At the other extreme, bottom-up approaches use small building blocks


26/34

856 Chapter 29 Fabrication of Microelectromechanical Devices and Systems and Nanoscale ManufacturingTABLE 29.3Comparison of NanoscaleManufacturing Techniques

Bothtop-down andCharacteristic Top down bottom-up Bottom up

Nanopatterning Photolitho- Electron beam Nanoimprint Dip pen nano- Microcontact Scanningtechnique graphy lithography lithography lithography printing tunnelingmicroscopyMaterial No No No Yes Yes LimitedflexibilityResolution ~35 nm ~15 nm ~ 10 nm 14 nm ~ 100 nm AtomicRegistration High High High Extremely high Low Extremely highaccuracySpeed Very fast Moderate Fast Slower, but Fast Very slowscalableCycle Time Weeks Days Days-week Hours Days-weeks DaysCostPurchase > 10 M > 1 M > 5O0 K 250KOperation High High Moderate Low Moderate LowSource: Courtesy of Nanolnk, Inc.

(such as atoms, molecules, or clusters of atoms and molecules) to build up a struc-ture. In theory, bottom-up approaches are similar to the additive manufacturingtechnologies described in Section 20.3. However, when placed in the context ofnanomanufacturing, bottom-up approaches suggest the manipulation and construc-tion of products on an atomic or molecular scale.

Bottom-up approaches are widely used in nature (e.g., building cells is a fun-damentally bottom-up approach), whereas conventionalmanufacturing has, for themost part, consisted of top-down approaches. In fact, there are presently nonanomanufactured products (excluding medicines and drugs manufactured bybacteria) that have demonstrated commercial viability.Bottom-up approaches in various research applications use mainly atomic forcemicroscopy (AFM) for the manipulation of materials on the nanoscale. Figure 29.29is an illustration of an atomic-force microscope. A probe (Fig. 29.29b) is mountedinto the microscope, and a laser is reflected from a mirror on the back side of theprobe so that it reflects onto a set of photosensors. Any vertical or torsional deflec-tion of the cantilever is registered as a change in voltage on the photosensors. Atomicforce microscopes can have true atomic resolution (< 1040 m).

Atomic-force microscopes are widely used to measure the surface profile ofvery smooth surfaces (Section 333). Several approaches have been developed toallow nanoscale manufacturing processes to be performed on these microscopes.Some top-down approaches are as follows: Photolithography, electron-beam lithography, and nanoimprint lithography,

all using soft lithography techniques, are capable of top-down manufacture ofstructures with resolution under 100 nm, as discussed in Section 28.7. Nanolithography. The probes used in atomic force microscopy vary greatly in

size, materials, and capabilities. The diamond-tipped stainless-steel cantilevershown in Fig. 29.29b has a tip radius around 10 nm. By contacting and plowingacross a surface, it can produce grooves up to a few microns thick. The spacingbetween lines depends on the groove depth needed.


27/34

Section 29.5 Nanoscale ManufacturingMirroredprism 5 Drode laser

and |f`1SGoin

Mirror-~ '* fi l Vr r ,_ ,,:, ,r,r,;, Cantileversp|ar-diode Sami and Substratephotodetector xyz piezo-crystaltube scanner

(H) (D)

FIGURE 29.29 (a) Schematic illustration of an atomic-force microscope. A probe is mountedon a cylinder containing piezoelectric material; this arrangement allows translation of theprobe in three dimensions. A laser reflected from a mirror on the back of the probe onto a setof photosensors allows measurement of the probes location and monitoring of interactionswith a sample surface. (b) Scanning-electron microscope image of a diamond-tipped stainless-steel cantilever suitable for nanolithography.

Ink-coated DPN pen lo5ndividua| ink mo|eu|eater meniscus ZfY2`~l 'fn ' 1 ' _ anopatterned ink QQi $Substrate iii* s-i` --


28/34


KEY TERMSBulk micromachiningDiffusion bondingDip pen nanolithographyEFABElectrochemical fabrication

BIBLIOGRAPHY

Bottom-up approaches include the following: Dip pen nanolithography can also be a bottom-up approach, wherein the ink

contains the material used to build the structure. Microcontact printing uses soft-lithography approaches to deposit material onsurfaces from which nanoscale structures can be produced. Scanning tunneling microscopy can be used to manipulate an atom on anatomically smooth surface usually cleaved mica or quartz .

SUMMARYThe MEMS field is relatively new and developing rapidly. Most successful com-mercial MEMS applications are in the optics, printing, and sensor industries. Thepossibilities for new device concepts and circuit designs appear to he endless.MEMS devices are manufactured through techniques and with materials that forthe most part have been pioneered in the microelectronics industry. Bulk andsurface micromachining are processes that are well developed for single-crystalsilicon.Specialized processes for MEMS include variations of machining, such as DRIE,SIMPLE, and SCREAM. These processes produce freestanding mechanical struc-tures in silicon.Polymer MEMS can be manufactured through LIGA or microstereolithography.LIGA combines X-ray lithography and electroforming to produce three-dimensional structures. Related processes include multilayer X-ray lithographyand HEXSIL.Nanoscale manufacturing is a new area that has significant potential. Nanoscalemanufacturing processes are typically bottom up, whereas conventional manu-facturing is top down. Some lithography processes extend to the nanoscale, asdoes dip pen lithography. Materials such as carbon nanotubes have great poten-tial for nanoscale devices.

HEXSIL Multilayer X-ray Silicon-LIGALIGA lithography SIMPLEMEMS Planarization StictionMicromachining Sacrificial layer Surface micromachiningMicrostereolithography SCREAM UV-LIGA

Allen, ].]., Microelectromechanical System Design, CRC Elwenspoek, M., and Jansen, H., Silicon Micromachining,Press, 2006.

Berger, L.I., Semiconductor Materials, CRC Press, 1997.Campbell, S.A., The Science and Engineering ofCambridge University Press, 1998.

Elwenspoek, M., and Wiegerink, R., Mechanical Microsensors,Springer, 2001.

Microelectronic Fabrication, 2nd ed., Oxford, 2001. Gad-el-Hak, M. ed. , The MEMS Handbook, 2nd ed.Electronic Materials Handbook, Vol. 1: Packaging, ASMInternational, 1989. 3 vols. , CRC Press, 2006.


29/34

Harper, C.A. ed. , Electronic Packaging and InterconnectionHandbook, 3rd ed., McGraw-Hill, 2000.-, High-Performance Printed Circuit Boards, McGraw-Hill, 2000.Hwang, ].S., Modern Solder Technology for CompetitiveElectronics Manufacturing, McGraw-Hill, 1996.Kovacs, G.T.A., Micromachined Transducers Sourcebook,McGraw-Hill, 1998.Leondes, C.T., MEMS/NEMS Handbook 5 vols. , Springer,2007.Liu, C., Foundations of MEMS, Prentice-Hall, 2006.Madou, M.].,Manufacturing Techniques forMicrofabrication

and Nanotechnology, CRC Press, 2009.-l, Applications of Microfabrication and Nanotechno-logy, CRC Press, 2009.Mahajan, S., and Harsha, K.S.S., Principles of Growth andProcessing of Semiconductors, McGraw-Hill, 1998.

Maluf, N., An Introduction to MicroelectromechanicalSystems Engineering, 2nd ed., Artech House, 2004.Nishi, Y, and Doering, R. eds. , Handbook of SemiconductorManufacturing Technologies, Marcel Dekker, 2000.

REVIEW QUESTIONS29.1.29.229.329.4.

Define MEMS, SIMPLE, SCREAM, and I-IEXSIL.Why is silicon often used with MEMS devices?Describe bulk and surface micromachining.What is the purpose ofa spacer layer in surface micro-machining?

29.5. What is the main limitation to successful applicationof MEMS?

QUALITATIVE PROBLEMS

Qualitative Problems 859Pierret, R.F., and Neudeck, G.W., Semiconductor DeviceFundamentals, Addison-Wesley, 1996.Quirk, M., and Serda, J., Semiconductor Manufacturing

Technology, Prentice Hall, 2000.Rockett, A., The Materials Science of Semiconductors,Springer, 2008.Sze, S.M. ed. , Modern Semiconductor Devices Physics,

Wiley, 1997.-l, Semiconductor Devices: Physics and Technology,Wiley, 2001.

Taur, Y., and Ning, T.H., Fundamentals of Modern VLSIDevices, Cambridge, 1998.

van Zant, P., Microchip Fabrication: A Practical Guide toSemiconductor Processing, 4th ed., McGraw-Hill,2000.Varadan, VK., Jiang, X., and Varadan, V, Microstereo-lithography and other Fabrication Techniques for 3D

MEMS, Wiley, 2001.Wise, K.D. ed. , Special Issue on Integrated Sensors, Micro-actuators and Microsystems MEMS , Proceedings ofthe IEEE, 1998.

29.6. What are common applications for MEMS andMEMS devices?29.7. What is LIGA? What are its advantages?29.8. What is a sacrificial layer?29.9. Explain the differences between stereolithographyand microstereolithography.

29.10. Describe the difference between isotropic etchingand anisotropic etching.29.1 1. Lithography produces projected shapes, so true three-dimensional shapes are more difficult to produce. What lithog-raphy processes are best able to produce three-dimensionalshapes, such as lenses? Explain.29.12. Which process or processes in this chapter allow thefabrication of products from polymers?29.13. What is the difference between chemically reactiveion etching and dry-plasma etching?29.14. The MEMS devices discussed in this chapter are ap-plicable to macroscale machine elements, such as spur gears,hinges, and beams. Which of the following machine elements

can or cannot be applied to MEMS, and why? a ball bear-ings, b bevel gears, c worm gears, d cams, e helicalsprings, f rivets, and g bolts.29.15. Explain how you would produce a spur gear if itsthickness was one-tenth of its diameter and its diameter wasa 10 um, b 100 um, c 1 mm, d 10 mm, and e 100 mm.29.16. List the advantages and disadvantages of surface mi-cromachining compared with bulk micromachining.29.17. What are the main limitations to the LIGA process?Explain.29.18. Other than HEXSIL, what process can be used tomake the microtweezers shown in Fig. 29.22? Explain.


30/34

860 Chapter 29 Fabrication of Microelectromechanica

QUANTITATIVE PROBLEMSDevices and Systems and Nanoscale Manufacturing

|29.l9. The atomic-force microscope probe shown inFig. 29.29 has a stainless steel cantilever that is450 >< 40 >< 2 /J.m. Using equations from solid mechanics,estimate the stiffness of the cantilever, and the force requiredto deflect the end of the cantilever by 1 /tm.|]29.20. Estimate the natural frequency of the cantilever inProblem 29.19. Hint: See Problem 3.21.|]29.2l. Tapping-mode probes for the atomic-force micro-scope are produced from etched silicon and have typical

SYNTHESIS, DESIGN, AND PROIECTS

dimensions of 125 um in length, 30 /.tm in width, and 3 umin thickness. Estimate the stiffness and natural frequency ofsuch probes.|l29.22. Using data from Chapter 28, derive the time neededto etch the hinge shown in Fig. 29.7 as a function of the hingethickness.

29.23. List similarities and differences between IC technolo-gies described in Chapter 28 and miniaturization technologiespresented in this chapter.29.24. Figure 1.8 in the General Introduction shows a mirrorthat is suspended on a torsional beam and can be inclinedthrough electrostatic attraction by applying a voltage on eitherside of the micromirror at the bottom of the trench. Make aflowchart of the manufacturing operations required to pro-duce this device.

29.25. Referring to Fig. 29.5, design an experiment to findthe critical dimensions of an overhanging cantilever that willnot stick to the substrate.29.26. Design an accelerometer by using la the SCREAMprocess and b the HEXSIL process.29.27. Design a micromachine or device that allows the directmeasurement of the mechanical properties of a thin film.


31/34

joining Processesand EquipmentWhen inspecting various common products, note that some products, such as paperclips, nails, steel balls for bearings, staples, screws and bolts, are made of only onecomponent. Almost all products, however, are assembled from components thathave been manufactured as individual parts. Even relatively simple products consistof at least two different components joined by various means. For example, a somekitchen knives have wooden or plastic handles that are attached to the metal bladewith fasteners; b cooking pots and pans have metal, plastic, or wooden handlesand knobs that are attached to the pot by various methods; and lc the eraser of anordinary pencil is attached with a brass sleeve.On a much larger scale, observe power tools, washing machines, motorcycles,ships, and airplanes and how their numerous components are assembled and joined sothat they not only can function reliably, but also are economical to produce. As shownin Table 1.1 in the General Introduction, a rotary lawn mower has about 300 partsand a grand piano has 12,000 parts. A C-SA transport plane has more than 4 millionparts, and a Boeing 747-400 aircraft has more than 6 million parts. A typical automo-bile consists of 15,000 components, some of which are shown in Fig. VI.1.joining is an all-inclusive term covering processes such as welding, brazing, sol-dering, adhesive bonding, and mechanical fastening. These processes are an essential

Bonding of windshield Spot-weldedto car body car bodyFastenersi- Mechanical fasteningBolted engine Of b0dY trimassembly Welded pipes forSoldered exhaust systemelectricalTCircuitry Seamed bpdyBrazed joint for Componen S

bonded fabric

FIGURE Vl.| Various parts in a typical automobile that are assembled by the processesdescribed in Part VI.86|


32/34

Part VI joining Processes and Equipment

Tubes

i _in1 -

and important aspect of manufacturing and assembly operations, for one or more ofthe following reasons:

Even a relatively simple product may be impossible to manufacture as a singlepiece. Consider, for example, the tubular construction shown in Fig. VI.2a.Assume that each of the arms of this product is 5 m long, the tubes are 100mm in diameter, and their wall thickness is 1 mm. After reviewing all of themanufacturing processes described in the preceding chapters, one would con-clude that manufacturing this product in one piece would be impossible or un-economical.

The product, such as a cooking pot with a handle, is easier and rnore economicalto manufacture as individual components, which are then assembled into aproduct.

Products such as hair dryers, appliances, and automobile engines need to bedesigned to be able to be taken apart for maintenance or replacement of theirparts. Different properties may be desirable for functional purposes of the product.For example, surfaces subjected to friction, wear, corrosion, or environmentalattack generally require characteristics that differ significantly from those of thecomponents bulk. Examples are (a) masonry drills with carbide cutting tipsbrazed to the shank of a drill (Fig. VI.2b), (b) automotive brake shoes, and(c) grinding wheels bonded to a metal backing (Section 262).

Transporting the product in individual components and assembling them latermay be easier and less costly than transporting the completed item. Metal orwood shelving, large toys, and machinery are assembled after the components orsubassemblies have been transported to the appropriate site.Although there are different ways of categorizing the wide variety of available

joining processes, we will follow the classification by the American Welding Society

Carbideinsert /gi;g gBraze Drill body ei (low-alloy is ~my (5S 99 l .

(H) (D) (C)

FIGURE V1.2 Examples of parts utilizing joining processes. (a) A tubular part fabricated byjoining individual components. This product cannot be manufactured in one piece by any ofthe methods described in the previous chapters if it consists of thin~walled, large-diameter,tubular-shaped long arms. (b) A drill bit with a carbide cutting insert brazed to a steelshank-an example of a part in which two materials need to be joined for performancereasons. (c) Spot welding of automobile bodies. Source: (c) Courtesy of Ford Motor Co.


33/34

Part VI Joining Processes and Equipment

1J6i`nln processesreeee eeee eeeeee _ss c e 'ee ee ss eee sChapter 30)eee

Chapter32) FasteningSeaming

. s,s2 s, a M Xssi s .s sssf s s, ,ss,,fssffqs Crimping,esfs,ee= stitching(Chapter 32) (Chapter 32)eeee sss sss r sss,la ...tCerQif=aP_cf t.E' 9t t=@L K . E ...E'PF*=?' Chemie' ss,sE cc M @n?a'.. 5Oxyfuel gas Arc Resistance Diffusion ColdThermit Resistance Explosion FrictionElectron beam Ultrasonic

Laser beam(Chapter 30) (Chapter 31)

FIGURE Vl.3 Outline of topics described in Part VI.

(AWS). Accordingly, joining processes fall into three major categories (see Figs. V1.3and I.7f): Welding Adhesive bonding Mechanical fastening.

Table V1.1 lists the general relative characteristics of various joiningprocesses. Welding processes, in turn, are generally classified into three basiccategories:

Fusion welding Solid-state welding Brazing and soldering.

As will be seen, some types of Welding processes can be classified into both thefusion and the solid-state categories.Fusion Welding is defined as the melting together and coalescing of materials bymeans of heat, usually supplied by chemical or electrical means; filler metals may ormay not be used. Fusion welding is composed of consumable- and nonconsumable-electrode arc Welding and high-energy-beam Welding processes. The Welded joint un-dergoes important metallurgical and physical changes, which, in turn, have a major


34/34

864 Part VI joining Processes and EquipmentTABLE Vl.lComparison ofVarious joining Methods

Characteristicsm in EG t 3 5 ._. i:

..i= W 'J : .22 SQ Q* 5 S S E _5 E % *S #3 3 E % 8Method Q an -I I-1 D5 L1-I E Lu .E U

Arc welding 2 3 1 3 1 2 2 2Resistance welding 2 1 1 3 3 3 3 1Brazing 1 1 1 3 1 3 2 3Bolts and nuts 2 3 1 2 1 1 1 3Riveting 2 3 1 1 1 3 1 2Fasteners 3 3 1 2 2 2 1 3Seaming and crimping 2 1 3 3 1 3 1 1Adhesive bonding 1 1 2 3 2 3 3 2Note: 1 = very good; 2 = good; 3 = poor. For cost, 1 is the lowest. '

(a) Butt joint (b) Cornerjoint (c) Tjoint (d) Lap joint (e) Edge jointFIGURE Vl.4 Examples of joints that can be made through the various joining processesdescribed in Chapters 30 through 32.

effect on the properties and performance of the welded component or structure. Somesimple welded joints are shown in Fig. VI.4.

In solid-state welding, joining takes place without fusion; consequently, there isno liquid (molten) phase in the joint. The basic processes in this category are diffusionbonding and cold, ultrasonic, friction, resistance, and explosion welding. Brazing usesfiller metals and involves lower temperatures than welding. Soldering uses similarfiller metals (solders) and involves even lower temperatures.

Adhesive bonding has unique applications that require strength, sealing, thermaland electrical insulating, vibration damping, and resistance to corrosion between dis-similar metals.Mechanical fastening involves traditionalmethods of using various fas-teners, especially bolts, nuts, and rivets. The joining of plastics can be accomplished byadhesive bonding, fusion by various external or internal heat sources, and mechanicalfastening.

kalpakjian 17

Documents