SEL EXHIBIT NO. 2012
`INNOLUX CORP. v. PATENT OF SEMICONDUCTOR ENERGY
`LABORATORY CO., LTD.
`
`IPR2013-00064
`
`

`

`ActiveMati-ix
`Liquid Crystal
`Displays
`
`Applications
`
`Fundamentals and
`
`Willem den Boer @
`
`Newnes
`
`

`

`Copyrighted Materia!
`
`Hul‘nn Li H.“ Jlnpnnt ud [lac-rim
`1m 'ullhll'ullir 1 lmr. 51m;- iflfl. Ihlrlnurmi. MA L'lHL'L.
`[arm-n- I |u1|u-._Iu.n|.m Hill. L'rnlnrJ (‘IKJ NI‘I’. UK
`
`th'ia‘u
`
`Liql'lnuhl C 2W5. ELL-1w: llh'.
`
`fl.” nuhn- Ira-crud
`
`Nl' run Ill Elm [III1'IILIIhII any It rcflrthlnc'd. «run-.3 III .I Ierrirrdl 'Uc-‘IIEIII. -.-r Inuullltlrni IIL .II|.1. IIIILI
`u-r |:'II' an! l'llr.'||'Il. draw-nit. nh—tthwnl. [III-.I-u'um'Ir-u mum-Jung, nr m1u-rwI-cr. I'll l'hJIIII:
`l|1u.- prur-
`u'nrru-n [u-rnllmmn rd. rl'u; |'Ill|'|iI-|'||_'l.
`
`Pump-lune. I'I'I:|r§ h- mug-hr «.llru-nl'. hum Elm u.1'- Scum 6-. TL-n'hnnlnm' Eluhh IM-unlmm m
`iHh-nl. UK' [in-11" Li'H'l I 945:: “PH-“L".
`|:I-: “4-H IHI'Ji “-5 5 H 1. |."1l1.1l|'. F'L'IIII'IIMLIIII'IlI-ml‘n'i Inn-rill
`Tum "1:11.
`.‘ll'ill cmnrllc'lc 1-H" rl'qnnr tm-Ilm' 1-1.. t'h-r EII-fl'wr I'll-Impw{httrchlI-cu'ltr.i:|1lnl.H
`u'll'utu'm; "Em-hulur fitq'plrrl' IuL-J ll'lm "C'Halnlm; Pmnéglhuu."
`
`IL—cupunng :Ik ll'Ilp-Ill'l'ullht- d [hind-rung wlur I'm. hn-IL “Inn-m Elli-11E“? pnnrp l1:- i'. .- uh. uh m'LLhnn-
`plum-r I'Iu-m-Lrs [um-Mr
`
`Lillnn' I'II Caryn}! CIIIirIfim-hl-I‘ullllrfl'lnn ”In
`.-"|.p1'~|1-L'.|Iiuln .IIflTIIIIIL‘J:
`
`III-Irhh Lth-lrll.‘ Cm'lqpflnfialn-hlflinliun ”II:
`."'|.
`a. .Iuln-uur rn-nnl Inl ll1l-I bun-1 In. .wnIIJIIr Ilull'l ll|r I‘I'm-h l d-un.
`
`[HEN-IE: D-FfiW-Ffli 1-5
`[EFH- ll: WEI -E' J'SL'III-i'lil 1|
`
`En: LI1|LInLut||rL llll all Nfl'llfl- |1l."|.i.L'.ll|:I.=I|:|.
`I. iH'l mu 'fi-i-l'l pitr .II unwhq-Lh.rIn-rm;ml1
`
`I.“
`
`IN C'T
`
`C14
`
`IN III'
`
`II."
`
`'J
`
`:-
`
`3
`
`I'-
`
`a
`
`all
`
`.n.
`
`.u
`
`E'I'II'IIL'J |I'|. Ilfl' LEI1II1‘IJ filurL-HII Ami-mm
`
`fibrlning tngmhfr m gmw
`librarics in dmluplng muntrifi
`
`mthrvwrm I miwlumdmg i mflwn;
`
`Cnpyrlghted Material
`
`

`

`Cepyrlghted Materia!
`
`WIIIIII-l-I-I-I-I-Illll-I-I-I-I-I-II-IlrlI-I-I-I-I-I-I-IllllI-I-IIlIIIIII-rIIIII-I-Ifi
`
`Contents
`
`Mm: WhitilIiiiiiii-I'i-i-Iiiiii-i-i-I'i-i-i-i-i-III'II'I'i-I'I'i-Il
`l.l
`HierIricalliersmiiw1
`LiquidCryetalepcrtite........
`.T
`PIIIarizatI'IIn, IIIIiIzliI-IILIIII-I1 and Birefringence ......................... 11
`1.1115 Tmstui Namath: LEII ..................................... 14
`
`LI
`l.3
`
`I.‘I
`I.‘i
`
`LimirIII'Ium uf Pmiw MIIIIIII Addmsing .......................... 1?
`References .................................................. II
`
`ChepterTwe: Opuefingfihmifluafflefiwflafielms......................i'3
`The LIISII III: I‘I'IETII'L Matrix .................................... 2.3
`
`Requirements Fur Attire Matrix Swilching IkI'icfi-s __________________ 14
`Till! Thin Film Transistm' ...................................... 29
`
`Thin Film Silicon FrIIpII-rries .................................... 32
`AmIIrpI'IIIIII: Silicnn TFTs .......................................34
`Pul'r'-SII'I4.‘J.II.‘I TFTII ......................................... .
`.
`. 3113
`Basic Pixel Circuit and .‘IIdIIreIIJ-Iing Madurai; ........................ 39
`Hindu-amend Displays ......................................... 43
`PIasm-Addreficd LClh ....................................... 4T
`RelizrcncL'I-I .................................................. ‘I-E
`
`Chapter Three. MW; erAMLCDII' ........................... 49
`I3IIriiI: SInIILIure III AMI.Ll'h .................................... 49
`
`3.1
`
`3.3
`
`3.4
`3.5
`
`3.6
`3.7
`
`3.3
`
`Thin Film Frucmsing .......................................... 'ifl
`Thin Film Frupmies .......................................... I5]
`.‘ImIIIpI'IIILIS Silicnn TFT Ian-my “Imam .......................... I35
`I'IIIIr'FSI Tf'T Army I‘ntesses .................................... IIiI‘iI
`[IIIIIIr FiEIIII Array Fnuxm _______________ _ ______________________ T3
`LC Cull fisIIEIIII‘If ............................................ T4
`MIHJIIII: Assembly ............................................ T?
`
`UH
`
`Cupyrlghted Mate rial
`
`

`

`Emma
`
`Cnpyrlghted Materia!
`
`5.9 Yield Impnwementa and Ctm'iidnratinm .......................... 1'?
`3.1:] Trendsinhflanufacturing......................................33
`
`ChpflrFauHM-HCDEIEMHH ............... ..............3?
`4I DriveMtthmls
`.
`..S?
`4.1 Rnwficluttarnlliilumn11in: Drivers
`.
`.
`.
`.
`. .. . .. .. .
`. 91
`4. 3 Timing Qintmllrrs. Display Uuntrullcrs. :iml Interim .
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`. 9'9
`4.4-
`lntflgmnnn n! Electrimics nn [11355 ............................. III!
`4.5 Hitclcligltts ................................................ 135
`4.15 aner L‘nnfiumptinn ________________________________________ 1W
`Refiemmufi ................................................ 110
`
`Chpflrfirl: WChmmrisnh ............. . ........ ....”3
`5.l Basics u!" I‘ltmunwrnr and CninrimrL-tn- ........................... 11 i
`5.2
`finghrnunn and CHI'II‘L'EH‘L' Ranin ................................. 1 IT
`5.} Viewing Hnglu Fichatrim' ..................................... 131
`5.4 Calm and Gray Scale Perfnrrnancc ............................. 123
`5.5
`fltspunm: Turn: and Hid-2111' ................................... 125'
`5.15 Hemlutinn and Sir: ......................................... 1 1|
`
`5.?
`
`Image Artifilcts ............................................ 154
`Reference: ................................................ 11".]r
`
`.. . ....1'39
`Chapflrfix: lmpmwmntnflmgegwflyinflflflfls .. .
`15.! Erlghtnm Impmwmmts .................................... 1W
`fill
`lncreuiiai Cultir Filter Tnlnsrniiir-inn ....................... 140
`15.1.3 High-Anemln: Harlin I‘lesignfi ____________________________ 14C]
`15. I ..5 Alternative Culnr FIll‘El’ Arrangements .................... 14-4
`6.1 .4
`Fanuhtnm: Enhunmncnt Films .......................... H'i
`15.1 Readability Under High Ambient Lighting Binditinm .............. 146
`6.1 Gil-[Jr Uumul lntpt'uwmenti .................................. 14-9
`15.4 Wide Viewing iIiiniglr: Technnlnflies .......................... _
`.
`. 150
`6.4.1 Cmnpflnsntiun Films .................................. 151
`6.4.1
`ln-I‘lanerfiwilching Mule- .............................. 151
`15.4.3 Vertical Alignment ................................... 151r
`fi.4_4
`FL {Lump-:Lrimn and (filter Viewing Angle:
`lmprnvemenr Merhmls ................................ 1612
`6.5 Enhancement of ‘Ir'idt'u Ftnlrrmanuu: ............................ 1%
`
`15.5.1 Emma: Time Cnmpcrnntinn ........................... 1IIEI'.|r
`15.5.1 Entulatifln iii-an linpulmvTi-Tm Elsplny .................. .
`.
`. 1:55
`
`W
`
`Eupyrlghted Mate rial
`
`

`

`Cnpyrlghted Materia!
`
`u—I—
`amt-net:
`
`ELEI
`
`IJIrgLISI'zI:In
`Refirmnce5-..-.... I'I'IEI
`
`Chapuritwn: SpedaIAMLCDConfgumhbns.I?!
`UlrrII- High RLH'IIIIIIUI'I Mnnimm and Impnwwj {3155' Scale . ........ .
`.
`.
`IT"‘EI1
`I] ReflucnwnndTmmfiuttiwI‘Iifiplawu
`.Ifll
`H FicIdaSqul‘ItiaIQIIurLCl'h ..-...............................IEI‘3
`T31 SlemtmuplcfiMLCDs.......................................I3?
`1'3
`TLILII:I1 Smut" TecI‘InIIIL'IHI'L‘s .................................... I'II'EI
`RHErenm-I ................................................. I'Qfi
`
`Chaptcrfi'gfit WWfitPflfldDflflfljTflhflkyfl ............... 1'9?
`SJ FIustispIafs....... ..
`.199
`E. 2
`EILEII‘LIIun'Iinmrcnt Displays ................................... 1‘31
`5. 3.. I
`TFEL D'LIIPIIIIFII ........................................ 132
`21.1.2 Organic LEI'J DEM-51,5 .................................. 253
`3.2.3
`Plifiik'L‘MflIfiI UrganicLElHllfiplays EDS
`3-1.4
`fictiveMatrixUmani:LEDDI'IIpInII's....-..................1flEI
`5.3 Electmniu:Paperam'lFlexihlcflimlays 209
`5.4 DrgzlnicThI'nFIImTerIfitIIrs..................................2H
`15.5 ant and RI.Ir Projectinn Displays ................... . .......... 114
`References.
`. III
`
`CharterNI'nn: MmMflmHann-ndlmgsfimsm..n.n...........223
`l-I.I FIHtI'IInEIInIIIgESme.......
`.12“-
`9.2
`IIIirccli'.Imvcrs'mn Ikrecrurs. ..... . ............... . ............. 225
`9.3
`Indirect Cnnversinn Detectum ................................. 323
`
`. 233
`.
`.
`. .. .
`.
`.
`.
`.
`.
`.
`.
`'14 Applicatium u’I'FIIIL Pancl III-Ra}- Scmurr- .......... .
`ReferencesEH
`
`tmII-llllI-lIIIII-IIIIIII-l-l-l-l-I-lll-l-ll-ll-Ill-IIl-IIl-IIIIIl-IIl-ll-I-ESE
`
`III
`
`Cupyrlghted Mate rial
`
`

`

`CHAPTER 2 Operating Principles of Active Matrix L CDs 2.1 The Case for Active Matrix Passive matrix LCDs are relatively easy to manufacture and do not require very large investments. An active matrix LCD, on the other hand, consists of thin film semiconductor circuitry on a glass substrate and shares some of the manufacturing complexity of integrated circuits on semiconductor wafers. The efficient production of active matrix LCDs requires large capital investments at the same level of wafer fabs for IC manufacturing. The incentive for an LCD producer to make this investment rather than sticking with passive matrix LCDs is simple: image quality, including resolution, maximum size, contrast ratio, viewing angle, gray scale, and video performance is much better in AMLCDs than in passive matrix LCDs. The resolution limitation of the passive matrix was addressed in the previous chapter. Another drawback of passive matrix addressing is that the capacitance of the select and data buslines increases with the square of the diagonal size of the LCD. Since the buslines in passive matrix LCDs consist of transparent conductors with relatively high resistance, the RC delays and power consumption in large passive LCDs become unmanageable. By adding a semiconductor switch at each pixel, the drawbacks of multiplexing in a passive matrix are eliminated. In the active matrix configuration, the voltage at each pixel and therefore its transmission and gray level can be accurately controlled. A conventional semiconductor process on crystalline Si wafers could, in principle, be used for AMLCDs. In fact, this has been successful in reflective microdisplays for projection and personal viewers and is commonly referred to as liquid crystal on silicon (LCOS). For larger displays, however, several factors preclude the use of crystalline silicon. First, it is opaque for visible light and is therefore not compatible with transmissive, backlit displays such as used in notebooks or desktop monitors. Secondly, processed silicon wafers 23
`
`

`

`Active Matrix Liquid Crystal Displays have a limited size (not exceeding 12 in.) and are too expensive. In the semiconductor industry there is a constant push for smaller and more compact circuitry by reducing design rules, which is not needed for AMLCD fabrication. In an AMLCD most of the viewing area consists of the transparent pixel electrodes. The semiconductor switch and address lines cover only a small fraction of the pixel area. There is no strong incentive to move to advanced submicron design rules since, for large displays with larger pixels, most features actually become larger. In other words, apart from the need for transparent substrates, the use of crystalline wafers would be overkill for the simple circuitry required in most AMLCDs. Thin film processing has come to the rescue; it allows simple circuitry to be manufactured on large glass substrates with relatively modest design rules and an acceptable, reduced number of process steps. In terms of operation, the AMLCD is comparable to dynamic random access memory (DRAM). Like the DRAM, the AMLCD consists of arrays of cells in which a voltage is stored. In the case of most LCDs this is not a digital voltage, but an analog voltage to represent different gray levels. The business model for AMLCD manufacturers also shows some resemblance to that of memory manufacturers. To a large degree, both products have become commodities with constant price pressure and their markets are cyclical in the sense that there are periods of excess supply and excess demand. 2.2 Requirements for Active Matrix Switching Devices The LC pixel in an LCD can, for all practical purposes, be treated as a low-leakage capacitor. This capacitance needs to be charged from the data voltage at one polarity to the data voltage of opposite polarity during each refresh cycle to obtain an AC voltage without a DC component across the LC pixel. In static images the amplitude of the data voltage across the LC pixels remains constant; only their polarity will change in every frame to prevent degradation of the LC cell. When the displayed information changes (as in video displays with moving images), the amplitude may change as well, so that the gray level can be varied according to the image data. In Fig. 2.1 the geometry of the LC capacitance CLc the equation for the pixel is shown. It is given by CLc ~o~cwl = d ' (2.1) 24
`
`

`

`Operating Principles of Active Matrix LCDs Figure 2.1" Geometry of the LC pixel capacitance. where e 0 is the permittivity constant in vacuum, eLc is the dielectric constant of the LC material, w and l are the side dimensions of the pixel electrode, and d is the LC layer thickness (also called the LC cell gap). Since the LC dielectric constant depends on the orientation of the molecules, the LC pixel capacitance varies significantly with applied voltage. For example, in a TN cell that uses LC fluid with a positive dielectric anisotropy, the LC director will be perpendicular to the electric field between the pixel electrodes for applied voltages below the threshold. Then, e(cid:127) should be substituted for eLc in Eq. 2.1. When 5-10 V AC is applied so that the LC molecules will line up along the electric field direction,
`should be substituted in Eq. 2.1. As a result, the capacitance can increase by a factor of two to three. For intermediate applied voltages there is a continuous variation of the effective dielectric constant of the LC layer from e(cid:127) to ell. LC pixel capacitances in practical LCDs are typically 1 pF or less. Figure 2.2 shows the circuit of four pixels in a passive matrix LCD. The solid lines represent the circuitry on one substrate, while the dotted lines represent the circuitry on the opposite substrate. In the passive matrix LCD, the LC pixel capacitors are simply formed where the striped transparent electrodes intersect (see also Fig. 1.13 in Chapter 1). The select and data lines are on opposite substrates. In a TFT-based AMLCD (Fig. 2.3), both select and data lines are on the active matrix substrate, along with the pixel electrode and storage capacitor. The counter-electrode of each LC pixel capacitor, indicated by the dotted line, is a single common transparent 25
`
`ell
`
`

`

`Active Matrix Liquid Crystal Displays Data line Select line : - : A
`
`it"
`
`"
`
`i."
`
`"
`
`LC pixel ~,~ u (cid:12)9 u (cid:12)9 ! | (cid:12)9 , Passive matrix Figure 2.2" Circuit diagram for four pixels in a passive matrix LCD. Data line Storage capacitor Select line TFT ._1_ i" i ' -i i _.12 i"i ~r-_ : "3__ i --I-- : --- LC pixel ...................... Common Figure 2.3: Circuit diagram for four pixels in an active matrix LCD. electrode on the opposite substrate (which usually also has color filters). This common electrode is shared by all pixels in the array. After the LC pixel capacitance is charged up to the data voltage, the TFT is switched OFF and the pixel electrode on the active array is floating. The pixel capacitor should retain its data voltage for the remainder of the frame time until the next address time. As discussed in Sec. 1.2 in Chapter 1, the LC fluid needs to have a high resistivity to prevent leakage currents. The capacity to retain charge is often expressed in the voltage holding ratio (VHR), which is defined as V=s VHR - Vpeak , (2.2)
`
`1
`
`V 2
`
`(2.3)
`
`0 where V is the RMS voltage, V o~c is the peak voltage of the LC pixel waveform, and Tf is the frame time, as shown in Fig. 2.4. 26
`
`V s- (oat,
`

`

`Operating Principles of Active Matrix LCDs In addition to the LC pixel capacitance, most AMLCDs also have an auxiliary storage capacitance connected in parallel to the LC capacitance (as shown in Fig. 2.3). The storage capacitor uses a thin film dielectric with negligible leakage current and its capacitance value is voltage-independent. Therefore, the storage capacitor acts as a buffer to suppress the undesirable voltage dependence and potential leakage current in the LC capacitance. With a storage capacitor at each pixel it is easier to control the RMS voltage on the pixel and therefore its gray level. In Fig. 2.5 the generic, basic circuit for the switch at each pixel is shown. When one row is selected, all the switches on that row are turned ON and the data voltages are transferred to the pixel electrodes. Ideally, the ON resistance of the switch is zero and the OFF resistance is infinite. In practice, the requirements can be much relaxed. In a display with N rows and a refresh rate of 60 Hz, the frame time Tf~m e to refresh the entire display will be 16.6 msec and the line time T~n e to address one row will be T~me/N or less. In the case of an XGA display with 768 rows, this translates into a maximum line time of 21 btsec to select one row. The ON current Ion of the switch should be sufficient to fully discharge the LC pixel capacitance and charge it to the opposite polarity voltage Von during the line time. This leads to 2 (Cst"~- CLC ) Fog Ion > Wlin e (cid:12)9 (2.4) ::: :::: VLc(t) Vpeak Vrms Tframe ~t Figure 2.4: Voltage across LC versus time. RoB Roff Cic Cst I Figure 2.5: Basic switch configuration in an AMLCD pixel. 27
`
`

`

`Active Matrix Liquid Crystal Displays The OFF current Iof f should be low enough to retain the charge on the pixel during the frame time Io~< (Cs,+ C~) AV Wframe , (2.5) where AV is the maximum tolerable voltage loss from the pixel capacitance during the frame time. When realistic values for a 15-in. XGA display are substituted in Eqs. 2.4 and 2.5, Cs~ = 0.3 pF, CLc = 0.3 pF, and AV = 20 mV, we get Ion > 1 gA and Io~< 0.5 pA. (2.6) In other words, the ON/OFF current ratio has to exceed six orders of magnitude. The current levels are quite low and can be achieved by transistors using thin film semiconductors such as amorphous silicon (a-Si). Since the pixels do not draw a current after being charged up during each refresh cycle, the power consumption in the LCD itself is low. Most of the power consumption in transmissive LCDs is usually in the backlight. Active matrix LCDs can be classified into direct-view displays and light valves (Fig. 2.6). Light valves or microdisplays are used for projection applications and as personal viewers and viewfinders. Microdisplays can be transmissive or reflective. Transmissive types use high-temperature poly-Si TFTs, while reflective types employ crystalline Si circuitry. Direct-view displays can be further categorized according to the type of switch used at each pixel. The vast majority of AMLCDs on the market use a-Si or poly-Si thin film Transistors (TFTs). Table 2.1 lists some of the pertinent properties of the various Figure 2.6: Classification of active matrix LCDs. 28
`
`

`

`Operating Principles of Active Matrix LCDs Table 2.1: Different types of switches for AMLCDs and their main applications Switching device a-Si TFT Mobility (cm2/Vsec) * 0.3-1 Highest processing temperature~ - 300~ (glass) Major applications Notebooks, flat panel monitors, LCD TVs High-T poly-Si TFT 100-300 -- 1000~ (quartz) Projection light valves, viewfinders Low-T poly-Si TFT 10-200 -- 500~ (glass) PDAs, notebooks, projection light valves, viewfinders Crystalline Si MOSFET 400 -- 1100~ (C-Si) Projection light valves, viewfinders Thin film diode < 300~ (glass) Handheld devices * Mobility determines the potential to integrate peripheral electronics. , Highest processing temperature determines substrate choice. switches, along with display applications in which they are common. The field effect mobility of the TFT (mostly a semiconductor material property) determines the feasibility of designing row and column drivers directly on the glass for a more integrated solution. The highest process temperature determines the type of substrate. For processing below 600~ low-cost glass substrates can be used. If there is any process step exceeding 1000~ as in high-temperature poly-Si LCDs, the substrate will be expensive quartz. High-temperature poly-Si TFTs are therefore used only in small displays for projection systems. Amorphous silicon TFTs have become dominant in large-area displays such as those used in notebook computers, flat panel desktop monitors, and LCD televisions. Low- temperature poly-Si-based LCDs are somewhat less common and are mostly applied in smaller devices, including mobile phones, PDAs, and projection light valves, where their higher processing cost is offset by integration of some of the peripheral electronics on the display glass. Thin film diodes with ON and OFF currents satisfying Eqs. 2.2 and 2.3 are suitable for LCDs as well. Thus far they have mostly been applied in displays with a limited number of gray levels, such as cell phones. 2.3 The Thin Film Transistor Thin film transistors (TFTs) are cousins of the transistors used in semiconductor chips. In terms of switching speed and operating voltage, they are inferior to state-of-the-art MOS transistors in crystalline Si. However, they are quite adequate as a simple ON/OFF pixel 29
`
`

`

`Active Matrix Liquid Crystal Displays switch in AMLCDs with a 60-Hz refresh rate. In Fig. 2.7 the top view and cross section of a basic TFT are depicted. The TFT has three terminals: the gate, source, and drain. The gate is separated from the semiconductor film by a gate insulator layer, and the source and drain make contact to the semiconductor. One type of TFT, the N-channel TFT, operates similarly to a field-effect N-channel metal oxide semiconductor (NMOS) transistor in crystalline silicon. It is switched ON by applying a positive voltage on the gate. The insulator acts like a capacitor dielectric, so that in the semiconductor channel an opposing negative charge is induced (see Fig. 2.8). This negative charge creates a conductive channel for electrons flowing from the source to the drain. The magnitude of the current depends on the dimensions of the conductive channel (the channel length L and the channel width W), on the capacitance of the gate insulator per unit area Cg, on the properties of the semiconductor film, and on the applied gate voltage. When a negative voltage is applied to the gate, the channel is depleted of electrons and negligible current flows. Figure 2.8: Conductive channel formation and current flow in TFT. 30
`
`

`

`Operating Principles of Active Matrix LCDs The ON current Ias of the TFT can be described, in a first approximation, by the same equations used for traditional NMOS transistors in crystalline Si [1]" I&- ~.~Cg w Vds(Vg-- Vth- 0.5Vds ) for VdsVg-- Vth (2.7) Ias- I.tCg-~ Vas(Vg- V,~) 2 for V~> Vg- V,h. (2.8) Here lit is the field effect mobility of the thin film semiconductor, a material property, and V,~ is the TFT threshold voltage, which depends on both the semiconductor and the gate insulator, and on their interface. Equations (2.7) and (2.8) describe, respectively, the linear and saturation regimes of operation for the TFT, as illustrated in Fig. 2.9. In the ON state of the TFT, when a data voltage is applied to the source, the drain with the LC load capacitance will charge up to the same voltage, thereby transferring the data Figure 2.9: Current-voltage characteristics of a-Si TFT. 31
`
`

`

`Active Matrix LiquM Crystal Displays signal from the data line to the pixel electrode. The TFT is turned OFF when applying a negative voltage to the gate. Since in the N-channel TFT the source and drain contacts block the injection of positive charge carriers (holes), there will be negligible current flow in the OFF state. Similar equations and considerations apply to P-channel TFTs, in which charge transport is by holes. They are switched ON by a negative gate voltage and switched off by a positive gate voltage. Many TFTs are of the staggered type, in which the source and the drain have an overlap region with the gate (as shown in Fig. 2.7). The overlap causes a parasitic capacitance that can have an impact on display performance unless the pixel is addressed by an appropriate drive scheme. The earliest TFTs for LCDs used CdSe semiconductor thin films. Since CdSe is not a standard material in the semiconductor industry, special deposition and etching processes needed to be developed to build CdSe-based TFT LCDs. Although CdSe TFTs showed good performance in AMLCDS, they have never been able to overcome the drawbacks of not being mainstream. They are now relegated to small research efforts. 2.4 Thin Film Silicon Properties Silicon is a very familiar material in the semiconductor industry and is therefore the preferred choice for use in AMLCDs. Thin films of silicon on glass are, however, inferior in quality to crystalline silicon. This is illustrated in Fig. 2.10, where the atomic structure of c-Si, p-Si and a-Si are schematically compared. In c-Si, the Si atoms are all four-fold coordinated and constitute a perfect crystal lattice with long-range order of the Si atom o = Si atom (cid:12)9 = H atom Grain boundary Mobility > 500 Mobility = 10-400 Mobility = 0.3-1 (A) (g) (C) Figure 2.10: Atomic structure of crystalline silicon (A), poly-crystalline silicon (B) and amorphous silicon (C). 32
`
`

`

`Operating Principles of Active Matrix LCDs positions. As a result, electrons and holes can flow with high velocities V e and vp an electric field E is applied: when Ve= ~e E and vp- lapE. (2.9) The values of the mobilities }.l e for electrons and gp exceed 500 cm2/Vsec in crystalline silicon. Low-temperature poly-Si thin films are first deposited as amorphous films, and then crystallized by laser annealing or thermal annealing. The annealing temperature cannot exceed 600~ when low-cost glass substrates are used. Sometimes seed materials such as Ni or preferential crystal growth from certain locations are used. The quality of poly-Si varies greatly depending on deposition and crystallization methods, but invariably it has grain boundaries that reduce the mobility by creating barriers for the flow of electrons and holes (see Fig. 2.10B). Some improvement can be obtained by passivating the grain boundaries with hydrogen in a post-hydrogenation step. The grain size is important in determining TFT characteristics. Preferably, to maximize the mobility, there are no grain boundaries in the channel area of the TFT. A number of techniques have been developed to crystallize the silicon films, including excimer laser annealing (ELA), solid phase crystallization (SPC), sequential lateral solidification (SLS), and metal- induced lateral crystallization (MILC). They result in TFTs with mobilities ranging from 10-400 cm2/Vsec, the electron mobility often being higher than the hole mobility. The concept of hydrogenation is important in amorphous silicon as well. The structure of amorphous silicon is more irregular, so that there is no long-range order and no crystal lattice. Nonetheless, since amorphous silicon films are deposited from the decomposition of Sill 4 gas, 5-10% hydrogen is automatically built in during film growth. The hydrogen acts to terminate dangling bonds in the loose a-Si network (Fig. 2.10C) and is essential to improve its electronic properties. The type of a-Si used in TFT LCDs is therefore often referred to as hydrogenated amorphous silicon or a-Si:H. The electron field effect mobility is only 0.3-1 cm2/Vsec in a-Si, but is adequate for pixel switches in TFT LCDs. The hole mobility is much lower, so that practical p-type TFTs are impossible in a-Si. Hydrogenated amorphous silicon, as used in LCDs, is a reddish-colored thin film with a thickness between 30 and 200 nm. It can be readily deposited on large glass surfaces by plasma-enhanced chemical vapor deposition. Some of its properties are listed in Table 2.2. It has a band gap of 1.7 eV, (considerably larger than crystalline and poly-crystalline Si [1.1 eV]), a very low dark conductivity, and high photoconductivity. The low dark conductivity makes it relatively easy to obtain a low OFF current in the TFT. The high photoconductivity is an undesirable property of a-Si, since it can cause unwanted photo-leakage currents in the TFT. To prevent photo-leakage, the a-Si channel in TFTs
`
`33
`
`

`

`Active Matrix Liquid Crystal Displays Table 2,2: Some properties of a-Si films a-Si film property Typical value Unit Hydrogen content 5-10 % Dark conductivity intrinsic film 10-1~ -9 (.Qcm) -1 Dark conductivity n + doped film 10-2-10 -1 (Q.cm) -1 Photoconductivity in sunlight 10 -4 (.Qcm) -1 Energy gap 1.75 eV Field effect mobility 0.3-1.0 cm2/Vsec is sandwiched between opaque layers to shield it as completely as possible from both ambient light and backlight in the LCD. Amorphous silicon can be easily doped in the gas phase during deposition, to obtain n- type or p-type films with many orders of magnitude higher conductivity than undoped films. N-type films are used as contact layers to the source and drain of the TFT. The field effect mobility, which is important for the ON current of the TFT (see Eqs. 2.5 and 2.6), is only 0.3-1 cmZ/Vsec and much lower than that of c-Si (500-1000 cmZ/Vsec). Amorphous silicon is therefore used primarily in circuits with relatively low switching speeds (i.e., as the pixel switches in AMLCDs). 2.5 Amorphous Silicon TFTs Amorphous silicon TFTs have been dominant in notebook and desktop monitor LCDs and in LCD televisions, as a result of their relatively easy processing on very large glass substrates and a limited number of process steps. Amorphous silicon TFTs are N-channel enhancement-type field-effect transistors. As mentioned in the previous section, the electronic properties of a-Si do not allow the fabrication of high-quality p-type TFTs. P-channel a-Si TFTs cannot be made sufficiently conductive to get acceptable ON current for application in displays. Three different types of a-Si TFTs are shown in Fig. 2.11. The back-channel-etched (BCE) inverted staggered TFT is the most commonly used; it has good performance and a relatively straightforward manufacturing process. Another type is the trilayer (or etch- stopper) TFT. Both the BCE and the trilayer TFT have a bottom gate. The intrinsic a-Si layer on top of the silicon-nitride gate insulator forms the channel. The source and drain electrodes have an overlap region with the gate and make contact to the a-Si layer via a heavily doped thin n-type a-Si interface layer. The purpose of this a-Si n + layer is twofold: it acts as a low-resistance ohmic contact for electrons to
`
`34
`
`

`

`Operating Principles of Active Matrix LCDs Figure 2.11" Three types of a-Si TFTs: (A) back-channel-etched TFT, (B) etch- stopper or trilayer TFT, and (C) top-gate TFT. maximize the ON current, and in the OFF state it blocks the injection of holes into the intrinsic i-layer to minimize the leakage current. Since the n + layer is conductive, it needs to be removed from the channel area to obtain a low OFF current. This is done by a back-etch after the source and drain electrodes are patterned--hence, the name back-channel-etched TFT. The back-etch process is quite critical and needs to be uniform across the entire manufacturing substrate. Since it is difficult to obtain significantly different etch rates for the a-Si n + layer and the undoped intrinsic layer (i-layer), good process control is required to completely remove the n + layer from the channel without etching through the i-layer. In the trilayer or etch-stopper TFT (Fig. 2.11B), the n + back-etch is less critical, since there is an extra silicon-nitride etch stopper layer on top of the channel. The etch stopper is deposited and patterned before the n + contact layer. Amorphous silicon and silicon-nitride can be selectively etched in dry and wet etch systems. During the back- etch of the n + layer there is much better etch selectivity possible against the silicon- nitride etch stopper layer, enhancing the process latitude. The trilayer TFT also has a 35
`
`

`

`Active Matrix Liquid Crystal Displays thinner i-layer, which can suppress photo-leakage currents in the TFT. The major drawback of the trilayer TFT is that it requires an extra deposition and patterning step. The top-gate a-Si TFT (Fig. 2.11C) is less common. In this configuration the source and drain electrodes are deposited and patterned first, followed by the a-Si layer. The gate insulator (usually silicon-nitride) needs to be deposited on top of the a-Si layer, which usually leads to somewhat lower-performance TFTs in terms of ON current. To suppress photo-leakage currents, a light shield needs to be deposited and patterned prior to the actual TFT process. It is separated from the TFT by an extra insulator layer. The extra processin