United States Patent
`Hornbeck
`
`[1
`
`[54]
`
`SPATIAL LIGHT MODULATOR AND
`METHOD
`
`(75)
`
`Inventor:
`
`Larry J. Hornbeck, Van Alstyne,
`Tex.
`
`[73]
`
`Assignee: Texas Instruments Incorporated,
`Dallas, Tex.
`
`[21]
`
`Appl. No.: 582,804
`
`[22]
`
`Filed:
`
`Sep. 13, 1990
`
`[63]
`
`Related U.S. Application Data
`Continuation of Ser. No. 355,049, May 15, 1989, aban-
`doned, which is a continuation of Ser. No. 168,724,
`Mar. 16, 1988, abandoned, whichis a continuation-in-
`part of Ser. No. 159,466, Feb. 19, 1988, abandoned,
`which is a continuation of Ser. No. 636,180, Jul. 31,
`1984, abandoned, and Ser. No. 43,740, Apr. 29, 1987,
`abandoned, whichis a continuation-in-part of Ser. No.
`792,947, Oct. 30, 1985, Pat. No. 4,662,746, and Ser. No.
`129,353, Nov. 30, 1987, abandoned, whichis a continu-
`ation of Ser. No. 877,654, Jun. 23, 1986, abandoned,
`which is a continuation-in-part of Ser. No. 646,399,
`Aug. 31, 1984, Pat. No. 4,596,992.
`
`[51]
`
`[52]
`
`[58]
`
`Trt, CUS ceececsesenenseesseesenes G02F 1/27; B44C 1/22;
`HO4N 5/74; G02B 26/08
`U.S, CL, sesseqeasurengeosneeasresanentel 359/224; 358/206,
`359/213; 359/298; 359/847
`Field of Search .........-:ssesssecesesseerseresesners 350/6.9
`
`fut] Patent Number:
`(45) Date of Patent:
`
`5,061,049
`Oct. 29, 1991
`
`[56]
`
`References Cited
`U.S. PATENT DOCUMENTS
`2,993,403
`7/1961 Harries <...e+ssese-seesreeeeseceneees 350/6.5
`7/1975 Nathanson et al. .
`3,896,338
`. 315/373
`
`3/1982 Petersen .....-..essseneeesrerereneetes 350/6.6
`4,317,611
`
`wee 350/360
`4,441,791
`4/1984 Hornbeck ....
`
`4,592,628 6/1986 Altman et al.
`wees 350/486
`
`6/1986 Hornbeck ......--sssssesrecrsrecess 350/360
`4,596,992
`
`wee, 350/360
`4,638,309
`1/1987
`
`wwe 332/7.51
`4,662,746
`5/1987
`
`4,698,602 10/1987 Armtage.....
`one BaFESA
`
`4,710,732 12/1987 Hornbeck...
`w-. 332/7.51
`
`4,793,699 12/1988 Tokuhara....
`were: 350/6.6
`4,831,614 5/1989 Duerig et Al, secsssaeessesssesneee 369/101
`FOREIGN PATENT DOCUMENTS
`0232413 10/1986 Japat ...-.cssesssseeressrrerencestrees 350/6.5
`0021115
`1/1987 Japan......
`». 350/6.5
`0035321
`2/1987 Japan......
`350/486
`0035322
`2/1987 Japan ......
`350/6.5
`0100417
`5/1988 Japan ...seseeres
`- 350/6.5
`
`1441840 3/1974 United Kingdom ...........0 350/487
`Primary Examiner—Nelson Moskowitz
`Attorney, Agent, or Firm—James C. Kesterson; James T.
`Comfort; Melvin Sharp
`ABSTRACT
`[57]
`An electrostatically deflectable beam spatial light mod-
`ulator with the beams (30), address electrodes (42, 46),
`and landing electrodes (49, 41) to provide soft-landing
`of the beams on the landing electrodes (40, 41) which
`gives uniform large-angle deflection plus high reliabil-
`ity.
`
`
`
`28 Claims, 34 Drawing Sheets
`
`
`
`TTPTTdLMkdeaioooh
`
`VWGO0OA EX1039
`U.S. Patent No. 9,955,551
`
`VWGoA EX1039
`U.S. Patent No. 9,955,551
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`Sheet 1 of 34
`
`5,061,049
`
`
`
`

`

`U.S. Patent
`
`Sheet 2 of 34
`
`Oct. 29, 1991
`
`5,061,049
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`Sheet 3 of 34
`
`5,061,049
`
`10.0
`
`8.0
`
`0(eG) °°
`
`4.0
`
`2.0 0.0
` /307
`
`40
`V (VOLTS)
`
`0
`
`20
`
`60
`
`80
`
`Fig.3a
`
`TORSION
`BEAM
`
`ADDRESS —
`
`/46 i
`
`Fig.4
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`Sheet 4 of 34
`
`5,061,049
`
`
`
`JONIHNOISYOLGNV
`
`‘G3AOW3Y
`
`
`
`WV38HLIM13XId
`
`O¢/
`
`SEI921
`
`Pé/
`
` cc]
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`Sheet 5 of 34
`
`5,061,049
`
`%A
`
`SS3NaaV NO!193430
`
`WV38p
`
`00'b
`
`8'0
`
`20
`
`
`
`
`
`(0p)39VLI0A3004193973
`
`964
`
`

`

`USS. Patent
`
`Oct. 29, 1991
`
`Sheet 6 of 34
`
`5,061,049
`
`TORQUE
`
`0.2
`
`0.4
`
`0.6
`
`0.8
`
`1.0
`
`Fig.5
`
`ZO,
`
`DIFFERENTIAL
`
`TORSION
`BEAM
`
`_
`
`T
`BIAS Vg
`
`40
`
`LANDING
`ELECTRODE
`
`ADDRESS
`ELECTRODE
`
`ADDRESS
`SIGNAL
`Pa
`fig.7
`
`

`

`U.S. Patent
`
`'
`
`2
`
`Oct. 29, 1991
`
`Sheet 7 of 34
`
`5,061,049
`
`+ oO
`
`
`
`\
`
`
`———sooy
`
`
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`Sheet 8 of 34
`
`5,061,049
`
`
`
`-41.0
`
`
`
`,+1.0
`|
`
`MONOSTABLE|
`
`(a=0)
`
`I | |
`
`Fig.10
`
`
`
` TRISTABLE(2=0,=4)
`
`Fig. 11
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`Sheet 9 of 34
`
`5,061,049
`
` Tq( Ve= Vo)”
`
`
`
`
`
`JUST BISTABLE(a= 2 4)
`
`
`
`fig.l2
`
`BISTABLE (a=)
`
`
`
`Fig. /F
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`Sheet 10 of 34
`
`5,061,049
`
`VpB=0
`
`Vez V4
`
`VB= Vo
`
`a
`
`+1.0
`
`.
`
`a aie
`
`|:
`
`-4.0|
`
`VB = Vo
`
`Fig./4
`
`
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`Sheet 11 of 34
`
`5,061,049
`
` 4
`
`Fig./6
`
`+7IVB|-- —
` ov——4—- — — — — —-OJ- - - - -
`
`
`
`Fig.I7
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`Sheet 12 of 34
`
`5,061,049
`
`0.14
`
`O42 0.10
`
`7 =799(7, 8)
`
`Tg=7 Agla,B)
`
`Fig. 19
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`Sheet 13 of 34
`
`5,061,049
`
`T(UNITSTo)
`
`I
`
`T(UNITS 79)
`[9(2,6)-g(-2,8)]
`
`(2)
`
`[i447 129 (a, 6)-9(-a,8)]
`
`(4)
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`Sheet 14 of 34
`
`5,061,049 -
`
`"Har
`(HOLDING TORQUE)
`
`+ (UNITS 19)
`
`0.4
`
` VB
`
`Vu
`BISTABLE HYSTERESIS
`CUR
`Fig. 23
`
`Vo
`
`

`

`U.S. Patent
`
`Oct. 29, 1991 |
`
`Sheet 15 of 34
`
`5,061,049
`
`+ (UNITS 79)
`
`ADDRESS
`
`ELECTRODE
`
`7a
`
`
`6 p
`
`+ +7VB
`
`
`
`ADORESS
`ELECTRODE
`
`
`
`
`fig.25
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`Sheet 16 of 34
`
`5,061,049
`
`T(UNITS TQ)
`
`
`
`
`Tal ie =o
`
`(2) €=€m
`
`(3)€>€m
`
`0.06 7
`
`Fig.26
`
`2eQ
`
`-lvel
`
`
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`Sheet 17 of 34
`
`5,061,049
`
`Tq+(€=0)
`7
`
`ee
`(Tae (€*4/2 &m)
`
`Tq- (€70)
`
`

`

`U.S. Patent
`
`Oct. 29,1991
`
`———
`
`Sheet 18 of 34
`
`5,061,049
`
`
`
` os
`
`MLPALLEPPPESPPpiosisit
`22
`
`Fig. 30b
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`Sheet 19 of 34
`
`5,061,049
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`Sheet 20 of 34
`
`5,061,049
`
`J,IQ,
`
`3/0
`
`
`
`360 ' 36)
`
`fig.32b
`
`
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`Sheet 21 of 34
`
`5,061,049
`
`ENABLE
`GATE
`
`420
`D
`
`Fig. 33
`
`422
` 436 '
`
`P Si
`
`466
`
`Se ardhiacrascanairmcnmnecidl
`siete
`|
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`Sheet 22 of 34
`
`5,061,049
`
`JF
`pepe
`ZITTITTLISILLILLITLLLL
`
` 344
`
`Fig. 35a
`
`
`
`WITTTIIIIIITLITITILIPTTH
`322
`
`Fig.35C
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`Sheet 23 of 34
`
`5,061,049
`
`DICING
`
`es
`
`
`
`Ohiiilirr
`
`32E
`
`o DEBRIS
`
`
`SAN
`
`
`
`348
`\ChhSCEaee
`Ie6
`
`344
`
`
`Fig. 35d
`
`PLLLLEPALISPALTLADALLELELPPLTPP
`
`FenianpinaaiegicagStiedinervoonegestas
`VALLIIIITTTiTIie
`
`
`
`
`Jee
`fig. J5e@
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`Sheet 24 of 34
`
`5,061,049
`
`EDGE OF
`
`524 ye
`
`
`
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`5,061,049
`
`348
`
`Sheet 25 of 34
`
`7“
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`Sheet 26 of 34
`
`5,061,049
`
`
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`EDGE
`
`Sheet 27 of 34
`
`5,061,049 SPACER
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`Sheet 28 of 34
`
`5,061,049
`
`gn
`
`BEAM METAL
`
`190
`
` SUBSTRATE
`
`FIg.40¢
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`Sheet 29 of 34
`
`5,061,049
`
`
`
`192
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`Sheet 30 of 34
`
`5,061,049
`
`-
`
`PHOTORESIST! :
`
`OPENINGtt
`
`fig.4/¢
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`Sheet 31 of 34
`
`5,061,049
`
`6/0
`AMPLIFIERS
`
`neOF
`
`DECODER
`
`fig.42b
`
`DATA
`|
`
`DRAIN
`BUFFER
`
`

`

`Oct. 29, 1991
`
`Sheet 32 of 34
`
`5,061,049
`
`U.S. Patent
`
`O99
`
`G99
`
`yNe
`
`/
`
`[-
`
`,999
`
`O¢pbly
`
`\OL9
`
`(eggU9}029
`
`f
`
`[93
`
`oo9
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`Sheet 33 of 34
`
`5,061,049
`
`

`

`U.S. Patent
`
`Oct. 29, 1991
`
`Sheet 34 of 34
`
`5,061,049
`
`eeepc 2ST
`Clelallel
`
`GFFOTE
`6Gla) EY ey ee eS
`670
`
`
`
`A fe
`aahahaanSSTI™\
`u
`VTTIILLA, XX MITTTTIZLLLLILSEL.
`ne) (ne)(n+\o+J
`
`
`
`
`670° 6b 82" ppp \gz966l Con,
`
`fig.44b
`
`
`
`Cnhlallahaahhaaah
`
`"626oe
`630
`(oes aa
`646
`642
`646
`
`642"
`
`SS > oo
`
`LAPLILADLELPIALILILIILIPIDIPLY
`
`

`

`1
`
`5,061,049
`
`SPATIAL LIGHT MODULATOR AND METHOD
`
`2
`duced by the modulated electron beam are converted
`into bright spots of light at the screen by the schlieren
`projector. In spite of numerous technical difficulties
`RELATED APPLICATIONS
`associated with oil polymerization by electron bom-
`This application is a continuation of application Ser.
`bardmentand organic vapor contamination ofthe cath-
`No. 07/355,049, filed May 15, 1989 now abandoned
`ode, this type of oil-film system has been successfully
`which is a continuation of Ser. No. 07/168,724, filed
`developed to the point that it is the almost universally
`Mar. 16, 1988, abandoned; which is a continuation in
`used system fora total light requirement of thousands of
`lumensat the screen. However, such systems are expen-
`part of Ser. No. 07/159,466, filed Feb. 19, 1988, aban-
`doned; which is a continuation of Ser. No. 06/636,180,
`sive, bulky, and have short-lived components.
`filed Jul. 31, 1984, abandoned; and a continuation of
`A numberof non-oil-film SLMshavealso been devel-
`Ser. No. 07/043,740, filed Apr. 29, 1987, abandoned:
`oped and include deflectable element types, rotation of
`whichis a continuation in part of Ser. No. 06/792,947,
`plane of polarization types, and light scattering types.
`filed Oct. 30, 1985, now U.S. Pat. No. 4,662,746; and a
`These SLM types employ variouseffects such as defor-
`continuation in part of Ser. No. 07/129,353 filed Nov.
`mation of reflective layers of metal, elastomer, or elas-
`30, 1987, abandoned; which is a continuation of Ser.
`tomer-photoconductor, and polarization and scattering
`No. 06/877,654 filed June 23, 1986, abandoned; which
`of ferroelectrics, PLZT ceramics, and liquid crystals.
`is a continuation in part of Ser. No. 06/646,399 filed
`For example, R. Sprague et al, Linear total internal
`Aug. 31, 1984, now U.S, Pat. No. 4,596,992.
`reflection spatial light modulatorfor laser printing, 299
`BACKGROUND OF THE INVENTION
`Proc. SPIE 68 (1981) and W. Turner and R. Sprague,
`The present invention relates to spatial light modula-
`Integrated total internal reflection (TIR)spatial light
`tors (light valves), and, more particularly,
`to spatial
`modulator forlaser printing, 299 Proc. SPIE 76 (1982)
`light modulators with pixels formed of electronically
`and U.S. Pat. No. 4,380,373 describe a system for non-
`addressable deflectable beams.
`impact printing on a photosensitive medium in which
`Spatial light modulators (SLM)are transducers that
`laser light
`is formed into a line of illumination and
`modulate incident light in a spatial pattern correspond-
`passed throughalinear array of light modulators and
`ing to an electrical or optical input. The incidentlight
`then imaged onto the photosensitive medium. The arra
`may be modulatedinits phase,intensity, polarization,or
`is implementedasatotal internal reflection spatial light
`direction, and the light modulation may achieved by a
`modulator with the electrodes and drive electronics
`variety of materials exhibiting various electrooptic or
`fabricated on an integrated drive element which is
`magnetoopotic effects and by materials that modulate
`placed against thetotal internal reflection surface of an
`light by surface deformation. SLMs have found numer-
`electrooptic crystal such as lithium niobate. The local-
`ousapplications in the areas ofoptical information pro-
`ized change in index of refraction produced by the
`cessing, projection displays, and electrostatic printing.
`fringing field between each two electrodes is read out
`See references cited in L. Hornbeck, 128 X 128 Deform-
`with schlieren readout optics which image the TIR
`able Mirror Device, 30 IEEE Tran. Elec. Dev. 539
`interface onto the photosensitive medium. This is a one
`(1983).
`dimensional image, and the photosensitive medium is
`A well known SLM used for large bright electronic
`rotated on a drum beneath the imageofthe linear array
`displays is the Eidophor, a system which uses an elec-
`to generate the two dimensional image (e.g., a page of
`trostatically dimpled oil film as the active optical ele-
`text) for printing applications. However,
`the SLM
`ment. See, E. Baumann, The Fischer large-screen pro-
`(light valve) is highly susceptible to fabrication prob-
`jection system (Eidophor), 20 J. SMPTE 351 (1953). In
`lems due to its hybrid nature. The fringing field
`this system a continuous oil film is scanned in raster
`strength, and hence the amount oflight diffracted from
`fashion with an electron beam that is modulated so as to
`modulated pixels, is sensitive to changes in the air gap
`create a spatially periodic distribution of deposited
`thickness between the address electrodes and the elec-
`charge within each resolvable pixel area on the oil film.
`trooptic crystal surface of less than one tenth micron.
`This charge distribution results in the creation of a
`Thus, even very small particles trapped between the
`phase grating within each pixel by virtueof the electro-
`crystal and electrode structure could cause illumination
`static attraction between the oil film surface and the
`nonuniformity problems at the photosensitive medium.
`supporting substrate, which is maintained at constant
`The system optical response for pixels located at the
`potential. This attractive force causes the surface of the
`boundary between modulated and unmodulatedareas of
`film to deform by an amount proportional to the quan-
`the light valve is also significantly lower than the re-
`tity of deposited charge. The modulated oil film is illu-
`sponse for pixels near the middle of a modulated region
`minated with spatially coherent light from a xenon arc
`due to the nature of the addressing technique. A com-
`lamp. Light incident to modulated pixels on the oil film
`mercially available printer based on this technology has
`is diffracted by the local phase gratings into a discrete
`not been introduced to date.
`set of regularly spaced orders which are madeto fall on
`M. Little et al., CCD-Addressed Liquid Crystal
`a schlieren stop consisting of a periodic array ofalter-
`Light Valve, Proc. SID Symp. 250 (April 1982) de-
`nating clear and opaquebars bypart of the optical sys-
`scribes a SLM with a CCD area array on the front side
`tem. The spacing of the schlieren stop bars is chosen to
`match the spacing of the diffracted signal orders at the
`ofa silicon chip andaliquid crystal array on the back-
`stop plane so that high optical throughput efficiency is
`side of the chip. Charge is input into the CCD until a
`achieved. Light thatis incident to unmodulated regions
`complete frame of analog charge data has been loaded;
`of the light valve is blocked from reaching the projec-
`the charge is then dumped to the backside of the chip
`tion lens by the opaque bars of the schlieren stop. Im-
`where it modulates the liquid crystal. This device suf-
`ages formed of unmodulated areas on the light valve by
`fers from severe fixed pattern noise as well as resolution
`the schlieren imaging system on the projection screen
`degradation due to the charge spreading from thefront-
`to-back transfer.
`are therefore dark, while the phase perturbations intro-
`
`10
`
`20
`
`25
`
`35
`
`"55
`
`65
`
`

`

`3
`Another SLM type which may befabricated in both
`one and two dimensional arrays is the deformable mir-
`ror. Deformable mirrors may be subdivided into three
`classes: elastomers, membranes, and cantilever beams.
`In the elastomer approach a metallized elastomer is
`addressed by a spatially varying voltage that produces
`surface deformation through compression ofthe elasto-
`mer. Because of the address voltage requirements in the
`order of one or two hundredvolts, the elastomer is not
`a good candidate for integration with a high-density
`silicon address circuit. See, generally, A. Lakatos and
`R. Bergen, TV projection display using an amorphous
`Se-type RUTICONlight valve, 24 IEEE Tran. Elec.
`Dev. 930 (1977).
`Membrane deformable mirrors come in a variety of
`types. One type is essentially a substitute for the oil film
`of the Eidophor system discussed above. In this system
`a thin reflective membrane is mounted to the faceplate
`of a cathode ray tube (CRT) by meansofa support grid
`structure. Addressing is by a raster scanned electron
`beam as with the Eidophor. The charge deposited on
`the glass faceplate of the CRT by the electron beam
`electrostatically attracts the membrane whichisheld at
`a constant voltage. This attractive force causes the
`membraneto sag into the well formed by the grid struc-
`ture, thereby forming a miniature spherical mirror at
`each modulated pixel
`location. The light diffracted
`from this type of modulated pixel is concentrated into a
`relatively narrow conethat is rotationally symmetric
`about the specularly reflected beam. This type oflight
`valve is thus used with a schlieren stop that consists of
`a single central obscuration positioned andsized so as to
`block the imageofthe light source that is formed by the
`optical system after specular reflection from unmodu-
`lated areas ofthe light valve. Modulated pixels giverise
`to a circular patch oflight at the schlieren stop plane
`that is larger than the central obscuration, but centered
`on it. The stop efficency, or fraction of the modulated
`pixel energy that clears the schlieren stop, is generally
`somewhat lower for projectors based on deformable
`membranes thanitis for the oil film Eidophorprojector.
`Further, such membrane deformable mirror systems
`have at least two major problems. High voltages are
`required for addressing the relatively stiff reflective
`membrane, and slight misalignments between the elec-
`tron beam raster and the pixel support grid structure
`lead to addressing problems. Such misalignments would
`cause image blurring and nonuniformity in display
`brightness.
`Another type of membrane deformable mirror is
`described in L. Hornbeck, 30 IEEE Tran. Elec. Dev.
`539 (1983) and U.S. Pat. No. 4,441,791 and is a hybrid
`integrated circuit consisting of an array of metallized
`polymer mirrors bonded to a silicon address circuit.
`The underlying analog address circuit, which is sepa-
`rated by an air gap from the mirror elements, causes the
`array of mirrors to be displaced in selected pixels by
`electrostatic attraction. The resultant two-dimensional
`displacement pattern yields a corresponding phase mod-
`ulation pattern for reflected light. This pattern may be
`converted into analog intensity variations by schlieren
`projection techniques or used as the input transducer
`for an optical
`information processor. However,
`the
`membrane deformable mirror has manufacturability
`problems due to the susceptibility to defects that result
`when even small, micron sized particles are trapped
`between the membrane and the underlying support
`structure. The membrane would form a tent over these
`
`40
`
`45
`
`60
`
`65
`
`5,061,049
`
`4
`trapped particles, and the lateral extent of such tents is
`muchlarger than thesize ofthe particleitself, and these
`tents would in turn be imaged as bright spots by a
`schlieren imaging system.
`A cantilever beam deformable mirror is a microme-
`chanical array of deformable cantilever beams which
`can be electrostatically and individually deformed by
`some address means to modulate incident
`light
`in a
`linear or areal pattern. Used in conjunction with the
`proper projection optics, a cantilever beam deformable
`mirror can be employed for displays, optical informa-
`tion processing, and electrophotographic printing. An
`early version with metal cantilever beams fabricated on
`glass by vacuum evaporation appears in U.S. Pat. No.
`3,600,798. This device has fabrication problems which
`include the alignment of the front and back glass sub-
`strates arising from the device’s nonintegrated architec-
`ture.
`A cantilever beam deformable mirror device is de-
`scribed in R. Thomas et al, The Mirror-Matrix Tube: A
`Novel Light Valve for Projection Displays, 22 IEEE
`Tran. Elec. Dev. 765 (1975) and U.S. Pat. Nos.
`3,886,310 and 3,896,338. This device is fabricated as
`follows: a thermalsilicon dioxide layer is grown on a
`silicon on sapphire substrate; the oxide is patterned in a
`cloverleaf array of four cantilever beams joined in the
`middle. Thesilicon is isotropically wet etched until the
`oxide is undercut, leaving within each pixel four oxide
`cantilever beams supported by a central silicon support
`post. Thecloverleafarray is then metallized with alumi-
`num for reflectivity. The aluminum which is deposited
`on the sapphire substrate forms a reference grid elec-
`trode which is held at a DC bias. The device is ad-
`dressed by a scanning electron beam which deposits a
`charge pattern on the cloverleaf beams causing the
`beams
`to be deformed by electrostatic attraction
`towards the reference grid. Erasure is achieved by neg-
`atively biasing a closely spaced external grid and flood-
`ing the device with low-energy electrons. A schlieren
`projector is used to convert the beam deformation into
`brightness variations at the projection screen. A signifi-
`cant feature of this device is the cloverleaf geometry
`which leads to beam deflection in a direction rotated
`forty-five degrees from the openings between the
`beams; this permits use of a simple cross shaped schlie-
`ren stop to block out the fixed diffraction background
`signal without attenuating the modulated diffraction
`signal. The device was fabricated with a pixel density of
`five hundred pixels per inch with beamsdeflectable up
`to four degrees. The optics employed a 150 watt xenon
`arc lamp,reflective schlieren optics and a 2.5 by 3.5 foot
`screen with a gain of five. Four hundred TV lines of
`resolution were demonstrated with a screen brightness
`of thirty-five foot-lumens, a contrast ratio of fifteen to
`one, and a beam diffraction efficiency of forty-eight
`percent. Write times of less than 1/30 second were
`achieved and erase times were as short as 1/10 of the
`write time. However, the device has problems, includ-
`ing degradation of resolution from scanning errors,
`poor manufacturing yield, and no advantage over con-
`ventional projection cathode ray tubes. That
`is,
`the
`scan-to-scan positioning accuracyis not high enough to
`reproducibly write on individual pixels. The resulting
`loss of resolution forces at least a four fold increase in
`the number of pixels required to maintain the same
`resolution compared to comparably written phosphor.
`Also, the device yield is limited by the lack of an etch
`stop for the cloverleaf support post, the wet etching of
`
`

`

`5,061,049
`
`_0
`
`5
`
`50
`
`6
`5
`duced substantially together with the diffraction effi-
`the beams leading to beam breakage, and the need to
`ciency. This means more lamp poweris required for the
`evaporate normally tensile aluminum in a state of zero
`stress on the oxide beams. Further, the device offers no
`same screen brightness. Because the address circuitry
`requires additional area, the pixel size is increased far
`apparent cost or performance advantage over conven-
`beyond the flap area with a resulting decrease in achiev-
`tional projection CRTs.
`able resolution. The wet etching required to form the
`Cantilever beam deformable mirrors integrated on
`wells leads to low electrical and mechanical yield; in-
`silicon with addressing circuitry, thus eliminating the
`deed, wet cleanups, such as after dicing into chips, de-
`electron beam addressing with its high voltagecircuitry
`stroy flaps.and diving boards because during the spin-
`and vacuum envelopes of the previously described can-
`tinse/dry cycle the water trapped under the beam
`tilever device, appear in K. Petersen, Micromechanical
`breaks the beam as it is spun from the surface. If the
`Light Modulator Array Fabricated on Silicon, 31 Appl.
`water is instead evaporated from the surface it leaves
`Phys. Lett. 521 (1977) and U.S. Pat. No. 4,229,732. The
`behind surface residues which can increase surface leak-
`first of these references describes a 16 by | array of
`age currents contributing to erratic device operation.
`diving board-shaped cantilever beams fabricated as
`Also, the addressing circuitry being on thesilicon sur-
`follows: an epitaxial layer of (100)-orientedsilicon (ei-
`face is exposed to theincidentlight to be modulated and
`ther p or n) of thickness of about 12 microns is grown on
`creates unwanted diffraction effects from the transistor
`a p+-substrate (or buried layer); the epilayeris oxidized
`to a thickness of about 0.5 micron and covered with a
`gates plus lowers the contrast ratio. In addition, light
`Cr-Au film of thickness about 500 A. The Cr-Au is
`leakage into the address structure produces photogene-
`rated charge and reduces storage time. Lastly, the ox-
`etched away to form contact pads and addresslines and
`ide/metal flap has the insulating side facing the well and
`to define the diving board metallization. The oxide is
`will charge up due to the intense electric fields which
`etched away in a comb pattern aroundthe metallization
`exist across the well; this produces a residual (“burn-
`in a second masking step. Finally, the silicon itself is
`in’) image. The AC drive required to eliminate this
`etched in a solution of ethylenediamine and pyrocate-
`residual
`image problem cannot be supplied by the
`chol at 120 degrees C. If the proper orientation of the
`NMOSdrivecircuitry described. Further, if the flap is
`mask with respect to the crystalline axes is maintained,
`deflected past the maximum stable deflection, then it
`the metal-coated oxide diving boards will be undercut
`will collapse and stick to the bottom of the well. Thus,
`by the etch and freed from thesilicon. Since the etch is
`voltages over the collapse voltage must be absolutely
`anisotropic, further lateral etching will be stopped by
`avoided.
`the (111) planes defining the rectangular envelope of the
`A variation of the cantilever beam approach appears
`comb pattern. In addition, the etchant is inhibited by
`in K. Petersen, Silicon Torsional Scanning Mirror, 24
`p-+material, so the depth of the well beneath the diving
`IBM J. Res. Devp. 631 (1980) and M. Cadmanetal,
`boards is defined by the thickness of the epilayer. When
`New Micromechanical Display Using Thin Metallic
`a de voltage is applied between the substrate and the
`Films, 4 IEEE Elec. Dev. Lett. 3 (1983). This approach
`diving board metallization, the thin oxide diving board
`forms metallic-coated silicon flaps or metallic flaps
`will be electrostatically deflected downward into the
`which are connected to the surroundingreflective sur-
`etched well. Diving boards of length 106 microns and
`face at two hinges and operate by twisting the flaps
`width 25 microns showed a threshold voltage of about
`66 volts.
`along the axes formed by the hinges. The flaps are not
`formed monolithically with the underlying addressing
`The second reference (Hartstein and Petersen, U.S.
`substrate, but are glued to it in a manner analogous to
`Pat. No. 4,229,732) describes devices fabricated in a
`the deformable membrane devices mentioned above.
`manner similar to the diving board device (a buried
`Cade, U.S. Pat. No. 4,356,730 combines aspects of the
`p-+layer as an etch stop for forming the wells under-
`foregoing and hasasilicon substrate with metal coated
`neath metallized silicon dioxide cantilever beams) but
`45
`silicon dioxide cantilever diving boards, corner-hinged
`has a different architecture; namely,
`the cantilever
`flaps, and torsion hinged flaps. The addressing elec-
`beams are in the shape of square flaps hinged at one
`trodes (two per diving board orflap for x-y addressing
`corner, the flaps form a two dimensional array instead
`as in a memory array) are on the surface, and the diving
`of the one dimensional row of diving boards, and the
`board or flap may be operated as a switch or memory
`wells underneath the flaps are not connected so that
`bit and collapsed to the bottom ofthe pit etched in the
`addressing lines for the flaps may be formed on the top
`surface of the silicon between the rows and columns of
`silicon substrate by application of a threshold voltage.
`The diving board orflap can then be held at the bottom
`flaps. Of course, the corner hinging of the flaps derives
`from the cloverleaf architecture of U.S. Pat. Nos.
`of the pit by a smaller standby voltage.
`- The cantilever beam references discussed above sug-
`3,886,310 and 3,896,338, but the full cloverleaf architec-
`gest that schlieren projection optical systems be used.
`ture could not be used because this would preclude the
`with the cantilever beam devices. But such systems
`surface addressinglines since cloverleaf flaps are hinged
`have limitations in terms of attainable optical perfor-
`to a central post isolated from the silicon surface. Fur-
`mance. First, the aperture diameter of the imaging lens
`ther, these devices have problems including poor reso-
`must be larger than is necessary to pass the signal en-
`lution and low efficiency due to density limitations and
`the small
`fractional active area,
`low manufacturing
`ergy alone. Hence the speed of the lens must be rela-
`tively high (or, equivalently, its f-number must be rela-
`yield, degradation of contrast ratio due to diffraction
`tively low) to pass all
`the signal energy around the
`effects from the adress circuitry, and residual image
`central schlieren stop obscuration. In addition, the sig-
`due to the charging effects of the oxide flap. More par-
`nal passes through the outer portion of the lens pupil in
`ticularly, the addressing circuitry is squeezed around
`this imaging configuration. Rays of light emanating
`the active area(flaps) because no option exists for plac-
`from any given point on the SLM andpassing through
`ing the address circuitry under the active area due to
`the outermost areas of an imager lens pupil are the most
`the wells being formed by etching away the epilayer
`difficult ones to bring to a well-corrected focus during
`down to the p+etch stop. Thus the active area is re-
`
`40
`
`60
`
`65
`
`

`

`5,061,049
`
`8
`
`15
`
`20
`
`25
`
`7
`the optical design of any imaging lens. When the outer
`rays are brought under good control, the rays passing
`through the center of the imager lens are automatically
`well-corrected. Hence, a greater level of optical design
`complexity is required of the imagining lens. Second,
`the field angle over which the imaging lens can form
`well-corrected images ofoff-axis pixels on a cantilever
`beam SLM isalso restricted. Any lens design task in-
`volves a compromise between the speed of the lens and
`the field angle it can cover with good image quality.
`Fast lenses tend to work over small fields, while wide
`angle lenses tend to be relatively slow. Since the schlie-
`ren imager must be well-corrected overits entire aper-
`ture, and since this aperture is larger in diameter than is
`required to pass the image forming light, the field angle
`that can be covered by thelensis smaller than it could
`be if a different imaging configuration could be devised
`in which the signal was passed throughthecenterof an
`unobscured, smaller diameterlens. Lastly, for an imager
`lens having a given finite speed, the use of the schlieren
`stop configuration also limits the size of the light source
`that can be utilized. This in turn limits the irradiance
`level that can be delivered to a projection screen or a
`photoreceptor at the image of a deflected pixel. This
`irradiance level, or the delivered power per unit area,
`depends on the product of the radiance of the light
`source, the transmittance of the optical system, and the
`solid angle of the cone of image forming rays oflight.
`The source radiance is determined only by the particu-
`lar lamp that is used. The optics transmittance depends
`on the stop efficiency for the particular SLM/schlieren
`stop configuration and surface transmission losses. But
`the solid angle of the image forming cone oflight is
`directly proportionalto the area ofthe imagerlens pupil
`thatis filled with signal energy. The use of a schlieren
`stop that obscures the central area of the imager lens
`pupil limits the usable pupil area and thus the image
`plane irradiance level that can be obtained for a lens of 40
`a given speed and a source of a given radiance;thisis in
`addition to the fundamental irradiance limitation that
`the maximum usable coneoflight has an opening angle
`equal to the beam deflection angle.
`The known beam SLMshave problems including
`beam insulator charging effects,
`lack of overvoltage
`protection against beam collapse, small-angle and nonu-
`niform beam deflection leading td optical inefficiency
`and nonuniformity, and high voltage addressing of the
`pixels.
`
`BRIEF DESCRIPTION OF THE DRAWINGS
`
`FIGS. 1A-C illustrate in perspective, cross sectional
`elevation, and plan viewsa first preferred embodiment
`pixel;
`FIG. 2 illustrates deflection of the pixel beam of the
`first preferred embodiment;
`FIGS. 3A-B show the dependence of beam deflec-
`tion on applied voltage for a simplified version of the
`first preferred embodimentpixel and a plan view of the
`simplified version pixel;
`FIG.4 is a cross sectional elevation schematic view
`of the simplified version pixel defining analysis terms;
`FIG.5 illustrates the torques on the beam in the sim-
`plified version pixel;
`FIG.6 illustrates the beam deflection as a function of
`control voltage for the simiplified version pixel;
`FIG.7 is a schematic cross sectional elevation view
`of the first preferred embodiment;
`FIG. 8 illustrates the torques on the beam in thefirst
`preferred embodiment;
`FIG. 9 shows definition of terms in the analysis of the
`first preferred embodiment;
`FIGS. 10-13 illustrate the torques on the beam for
`different modes of operation of the first preferred em-
`bodiment;
`FIG. 14 illustrates the potential energy functions
`involved in the different modes of operation ofthe first
`preferred embodiment;
`FIG. 15 showsthe differential bias on the first pre-
`ferred embodiment;
`FIG.16illustrates the torques for applied differential
`bias in the first preferred embodiment;
`FIG. 17 is a timing diagram forbistable operation of
`the first preferred embodiment;
`FIGS. 18-23 show the torques and deflection for
`analysis of the operation