evolution of flat panel display

8/3/2019 Evolution of Flat Panel Display

1/11

Evolution of Flat-Panel DisplaysLAWRENCE E. TANNAS, JR., SENIOR MEMBER, IEEEInvi ted Paper

The evolution in performance and manufacturability ofliquid-crystal displays has created a new reality in the electronicinformation displays industry. LCDs now outnumber all otherjlat-panel displays in production volum e, by more than TW O ordersof magnitude. LCD s may equal CRTs in market sales by theyear 2000. However, LCDs are still an order of magnitudemore expensive than comparable performing CRTs. LCDsare enabling new products, such as personal digital assistants,moving map navigators, picture telephones, etc., which could notbe readily done with CRTs because of their size and sensitivityto ambient illumina tions. The LCD techn ology in various modes,from passive twisted nematic to active matrix, is being used inall product siz es, from the 0.7-in camcorder viewfinder to the14-in full-color display with XG A resolution. Full color is veryimportant in future products and any 3at-panel display technologywithout it will be relegated to niche markets.I. INTRODUCTION

The concept of a flat-panel display is a simple extensionof a printed picture with a time-varying dimension. Thetechnical requirements to achieve this, however, are im-mense. Just how does one alter an image so that it portraysa time-varying likeness to a real-life action scene?About 20 years ago flat-panel displays (FPDs) beganto be used in military and industrial applications. Theyhave been used commercially in the consumer market2 foronly ten years, and in high-volume consumer products forless than five years. In the last five years FPDs haveadvanced to the consumer product stage with the statusof a major electronic component, whereas in 1989 theyexceeded one billion dollars in annual sales [l]. Only inthe last three years have FPDs been available in productionquantities, with performance comparable to color CRTs inVGA resolution.This paper deals primarily with High Information Content(HIC) Displays which offer performance that is comparableto the Cathode Ray Tube (CRT).

Manuscript received July 2 7, 1993; revised December 3, 1993.The author is President of Tannas Electronics, Orange, CA 92666.IEEE Log Number 9400122. The first successful ITD technology was the ac gas-discharge matrix-addressed display invented at the University of Illinois in 1964. Pilot lineproduction was established by Owens Illinois cu. 1974.*The first FPD used successfully in a consumer product and, still inproduction today, is the Grid personal portable PC using an ac thin-filmelectroluminescent display manufactured by Sharp of Japan.

The picture element (pixel) count ranges from 50000to 1.5 million and greater in monochrome or full-colorversions. This corresponds to products from low-endvideogames, portable televisions, and personal computersup to full-color VGA, XGA, and HDTV.Lower pixel count displays have a very wide and diversevariety of markets and display technologies. The history ofthis class of display is often traced to the first productionby Burroughs of Nixie Tubes, ca . 1955, inventor unknown.Typical low-pixel-count displays include hand-held cal-culators, clocks and meters, marquees and billboards. Inaddition, low-pixel-count displays do not suffer from theperplexing matrix addressing cross-coupling problem (to bediscussed later) which is proportional o the pixel count, andwhich plagues all HIC FPDs.

11. N E W REALITYN DISPLAYSThe original market objective, starting c a . 1950, to re-place the CRT with FTDs, has not been achieved. Instead,a whole new family of products, which could not easilyuse CRTs, has been created. Examples of these first newproducts include briefcase- and notebook-style personalcomputers and hand-held color television displays.In the FPD technology spectra3,which include electrolu-minescent displays (ELDs), plasma display panels (PDPs),light-emitting diode (LED) displays, flat CRTs, and liquid-crystal displays (LCDs), most recently the LCD technologyhas emerged as the clear leader. The production volume ofHIC LCDs is now more than 100 times greater than allother HIC FPD technologies combined.There are many reasons for the acceleration of LCDtechnology, including:a) highest immunity to ambient illumination,b) thinnest profile,c) lightest weight,d) lowest power requirement,e) color performance comparable to CRT,f) lowest cost, compared to other FPD technologies.3An FPD is a display which is flat like a pancake, as opposed to flatlike flat iron.

0018-9219/94$04.00 0 1994 IEEEPROCEEDINGS OF THE IEEE. VOL. 82. NO. 4, APRIL 1994

~ _ _ _ ~1 __499


2/11

Fig. 1. Expanding applications for LCDs.Negative limitations are diminishing and becoming moreacceptable to consumers, including:

very high cost compared to CRTs (more than tentimes that of a comparable CRT),limited viewing angle (now greater than f 45 in oneaxis with 1O:l contrast ratio),slow speed of response (now less than 200 ms inpassive LCDs and 50 ms in active matrix LCDs),narrow temperature operating range (now as wide as-3OOC to +85OC).

The cost issue will prevail indefinitely due to the highelectronic content required for the row and column driversand buffering electronics. Active-matrix liquid-crystal dis-plays (AMLCDs), which are the highest performing FPDswith performance comparable to CRTs, are presently 10to 20 times more expensive than CRTs. The cost of a10-in color AMLC VGA computer display is presently$1400.00 each and, possibly, $1000.00 each in 100000 unitquantities. Japanese market analysts speculate that the high-volume price may be reduced to $600.00 each by 1996. Atthe present price, the industry is production-limited.A major price difference between AMLCDs and CRTswill prevail through the year 2000 and inhibits AMLCDsfrom replacing CRTs any time in the foreseeable future.A 14-in color CRT TV can be purchased in Asia, in highvolume by original equipment manufacturers, for $50.00VI.The speed of response of all LCDs is slower than aCRT. Basically, a CRT can create a new full-intensity,viewable image in one complete electron-beam scan of thephosphor screen. An AMLCD, the fastest responding LCDconfiguration, takes approximately three complete scansof the image (due to the viscosity and restoring forcesupon the liquid-crystal molecules themselves) which is fastenough for consumer video and games. The lower costLCD configurations, such as passive twisted nematic (TN)and super-twisted nematic (STN) LCD, take longer but, ingeneral, are suitable for nonvideo imagery such as PC wordprocessing, spreadsheets, graphics, etc.A tabulation of some of the new and anticipated productapplications is given in Fig. 1. The preferred LCD config-uration is shown beside each product, such as amorphous-silicon thin-film transistor (a-Si TFT), AMLCD, passive

1 2 3 45 10 14Diagonal Size In lnches (log rate)

STN LCD, etc. Each LCD configuration is later describedin the section The Emergence of LCDs. The LCDconfigurations are listed above the size categories currentlyin production. Most of these FPD applications are presentlyemerging or are expected to emerge before the year 2000.The primary applications paying for the technological evo-lution have been LCDs for portable television sets andportable personal computers.111. THE CRT CHALLENGE

The concept of modem television was conceived by A.C. Campbell Swinton in London and reported in 1908and 1911 [3, pp. 2-31. Television was not demonstrated,however, until 1926 by J. L. Baird, also in England. For hisdemonstration, Baird used a CRT, previously developed byKarl Ferdinand Braun in 1897 in Germany as an instrumentto show the wave shape of alternating current. [3, pp.It probably occurred to the early inventors that there mustbe a simpler way to portray an image than with a full CRT.However, the CRT was and still is an elegant way to displayan image, particularly when used in conjunction with anelectronic imaging camera.The elegance of the CRT resides in the direct way a

serial stream of spatial amplitude data can be imaged onthe phosphor faceplate. A single amplification is used forelectron-beam amplitude control and two additional ampli-fiers for electrostatic or magnetic electron-beam horizontaland vertical deflection. The undesirable depth of the CRTis a consequence of the elegant electron-beam deflectiontechnique.For obvious reasons of ease of use and installation, theearly motivation was to replace the bulky CRT with anFPD. Also, it seemed intuitive that an FPD could be lessexpensive than a CRT without the complexity of the CRTstructure and the need for high-voltage and high-frequencyamplifiers.Serious research and development into FPD technologydid not begin in earnest, however, until the televisionindustry was well established. The most promising earlydevelopments were seen in flat CRTs such as the AikenTube (U.S.A., 1951) and the Gabor Tube (England, 1953)[3, pp. 180-1851. Versions of these innovations are stillbeing studied today, principally by Matsushita of Japan,but have not been successful due to increased complexity

140-141

50 0 PROCEEDlNGS OF TH E IEEE, VOL. 82, NO . 4, APRIL 1994


3/11

and reduced performance compared to conventional CRTsand FPD options.Needless to say, replacing the CRT with some sort ofFPD turned out to be much more difficult than anyone hadanticipated. Someone once postulated, in retrospect, that ifengineers and scientists had known how difficult it wouldbe, they would not have started. Certainly, management andinvestors would not have funded them.Even today the CRT is capable of higher overall per-formance than any FPD yet demonstrated. Furthermore, inalmost every category where FPDs are used, a CRT canperform as well or better at a lower price. FPDs are beingused only where CRTs cannot reasonably fit, for example,in the briefcase or pocket, where consumers want to carrytheir personal computer and TV .The idea of making an FPD got a gigantic boost whenRCA head David Sarnoff announced that his companyplanned to develop the TV picture on the Wall. Thissimple characterization dentifies both the virtues and a mar-ket for the FPD. Sarnoffs challenge created an interestingparadox to corporate management of the time: What willwe do with all those CRT factories after we develop anFPD TV? ven today, the paradox has never become anissue, as we still do not have a cost-effective TV pictureon the Wall.Ironically, it was not the TV industry that developedthe FPD component, but the wristwatch and computerindustries. Non-CRT consumer product manufacturers likeSeiko Epson, Canon, Casio, Sharp, and others made theearly hand-held television and computer FPDs.The exceptions, Toshiba and Hitachi, are now, after alate start, among the top manufacturers of FPDs; butwhere were GE, Philips, RCA, Sony, Thomson-CSF, Zenithduring the development phase?6Philips is now building a factory to manufacture AMLCFPDs in Eindhoven, The Netherlands, which is also notableas the first commercial AMLCD factory outside of Japan.

IBM, Yorktown Heights, NY; Litton Systems, CanadaLtd., Rexdale, Ont., Canada; 01s Optical Imaging Systemsin Troy, MI; Sagem, Paris, France; Thomson LCD inGrenoble, France; and Xerox PARC, Palo Alto, CA; allcurrently have thin-film transistor (TFT) AMLCD prototyp-ing capability in their respective local facilities. In Japan,approximately 30 R&D centers and 10 major corporationsare doing LCD research, development, and manufacturing[4]. Both Samsung and Goldstar have pilot line AMLCDcapabilities and are considering production in Korea.IV. THE EMERGENCEF LCDs

Since 1963, LCDs have been vigorously pursued as anHIC FPD, but only recently have they emerged as a major4 0 n e of the early contenders for a truly flat FPD was electrolumines-50 n e exception is readability in high ambient illumination.6Matsushita, RCA, and Sony did significant research and development

work in flat cathodoluminescence. GE developed an avionics AMLCD in1987 and then sold it to what is now Thomson LCD, Grenoble, France.Sharp buys CRTs for their TV product line from other companies.

cence.

electronic component industry surpassing one billion U.S.dollars in 1989. Before 1990 (a somewhat arbitrary pointin time) all FPDs were niche market components imple-mented by numerous technologies. The major technologieswere thin-film ELDs, PDPs, LCDs, LEDs, VFDs, flatCRTs, and numerous others that never got out of researchand development. Thus far, the TNLCDs, LEDs, VFDs,and gas-discharge tubes have been successful as low- andmedium-information content displays.Clearly, today the leading FPD technology is the LCtechnology. It is the only FPD in volume production withfull color in video and PC sizes7. Its emergence has beendue to two technical solutions to the complex matrixaddressing problem. One is the STN LCD [5]configuration,which has sufficient nonlinearity for large arrays to bematrix-addressed. This configuration falls into the passiveLCD classification, since there are no active componentsintemal to the display panel for addressing the array ofpixels. The other solution is the AMLCD [6] which utilizesan active component at each pixel or color subpixel toprovide sufficient nonlinearity for matrix addressing. Thusfar, the most widely used active element has been a TFT,in which the semiconductor is amorphous silicon.The LCD technology is still evolving. Several configura-tions are in high-volume production:

a) Twisted nematic LCD-Matrix addressability limitedto approximately64 ows, lowest cost, twist angle of90 limited viewing angle and response speed. TheLCD configuration n highest volume production. Dueto limited matrix addressability, it is used mostly insmall- to medium-size displays.b) Supertwisted nematic LCD-Matrix addressable upto 512 rows, ideal for low-end PCs, color nowavailable, lowest cost PC LCD, twist angle of 145to 20O0,limited in viewing angle and response speednot fast enough for quality video presentation. Largestcomputer LCD in volume production. Improved ma-trix addressability is achieved from the nonlinearityachieved from the higher twist angle.c) Multiple row addressed or Active AddressingTM8-Used in addressing an STN LCD to increase the re-sponse speed by addressing multiple rows in parallelwith a modest increase in cost for the extra elec-tronic processing needed to preprocess the signal toperform multiple row addressing. This configurationhas been demonstrated by In Focus and OptrexB, butnot yet commercialized. As always, the columns areaddressed in parallel. (Further discussion to follow.)

A possible exception is the new 21-in PDP by Fujitsu scheduled tobe manufactured starting in 1994,with good color and video capability,exhibited at the 1992 and 1993 Japan Electronics Shows. At the JES93, NH K and Matsushita have demonstrated HDTV color PDPs at 40-indiagonal.No plans have been announced to manufacture this panel. Planardemonstrated a color IO-in (640x480) ELD at the SID 1993 Conferencein Seattle, WA. In the opinion of this author, these latter two examplesare not ready for production.

Trade mark of In Focus Systems, Tualatin, OR.501ANNAS: EVOLUTION OF FLAT-PANEL DISPLAYS


4/11

d) Metal-Insulator-Metal (MIM) AMLCD-Used inaddressing a TNLCD to increase pixel nonlinearityfor improving matrix addressing. Since the MIMthin films of typically Ta/TaO,/Cr are placed ateach9 addressable pixel internal to the display panel,an MIM-augmented display is called an AMLCD.However, the individual pixel is a two-terminalelement. (The addressable pixels or subpixels ofmost AMLCDs are three-terminal elements.) Thisconfiguration is in production by Seiko-Epson and isused primarily in hand-held LCD TVs. The price andperformance puts MIM LCD TV products betweenSTN LCD and AMLCD portable TVs.e) Split Electrode STN LCD-Used in addressing colorSTN LCDs to reduce the effects of cross couplingin matrix-addressed FPDs. The column electrodesare opened in the center of the display to renderthe display electrically two separate displays. Columndrivers are placed at the top and bottom of each visiblecolumn line. The matrix addressing requirements arereduced by a factor of two since the number of rowlines are reduced by 1/2 in each half. The increasein cost for the extra column drivers is justified bythe improved overall color performance as requiredby VGA and portable PCs-still not fast enough forvideo.

f) TFT AMLCDs-Use a full switch at each pixel orsubpixel to stop cross coupling. The TFT semicon-ductor is typically a-Si:H, but may be a poly-Si:H,or single-crystal Si for projector displays. CdSe isalso used as a TFT semiconductor, but is not yet inproduction. The full switch gives the fastest speed,widest viewing angle, and is best for gray scale of allthe LCD configurations.g) Others-There are numerous other versions of LCDsnot yet in production. Examples include FerroelectricLCDs (under development by Canon), Verticallyaligned LCDs (by Stanley), Plasma-addressed LCDs(by Tektronix), and others.

These LCD configurations cover a wide spectrum andall seem to be finding markets which best fit their in-dividual performance/cost ratio and features. It is impor-tant to note that the commercial market is utilizing allthese configurations in consumer products because of theprice/performance differences and unique features of each.Within the LCD community of display developers andusers, there has been an ongoing debate whether STNLCDs or a-Si TFT AMLCDs will dominate in the majormarkets for portable PCs. It now appears that, because ofclear price and performance differences, both will hold asignificant segment with their respective market advantages.Instead of a dominance by one, the spectrum is expandingwith MIM, split column, and multiple row addressing (Ac-9When spatial color is used, the pixel is made up of 3 or 4 subpixels,one for each primary color. Each subpixel must be addressed seperately.In R&D by Kopin, Taunton, MA .In R&D by Litton Systems Canada Ltd., Rexdale, Ont., Canada.

tive Addressing) with price and performance characteristicsbetween the two major configurations, STN LCDs and a-Si TFT AMLCDs. Japanese market analysts are predictingthat, by 1996, a-Si TFT AMLCDs will command 60%of the 10-in and larger VGA color display market. Thisprediction is predicated upon a significant price reductionin a-Si TFT AMLCDs, as previously discussed.The separation in price and performance between STNand a-Si TFT LCDs is fundamental to the technology of therespective approaches. The STN LCD is a passive displayutilizing single row and column electrodes for addressing.This approach depends upon the nonlinear response of theSTN mode to make the display matrix addressable. The a-Si TFT forms a transistor switch at each pixel or subpixeland, therefore, can use the more linear-respondingTN modeand still be matrix-addressable. The TN mode is faster inresponse and gives more uniform gray scales than the STNmode. Only the a-Si TFT LCD is fast enough for videoand has enough gray scales for full color. The a-Si TFT,however, greatly complicates the pixel or subpixel structureand makes it a three-terminal device. The manufacturingof a-Si TFT AMLCDs is highly machine-intensive andrequires several photolithographic steps to be performed ata 2-pm design rule.V. THE LCD ADVANTAGE

As a group, the LCD technologies have several uniquetechnical advantages that underscore why they have ad-vanced so far beyond the other FPD technologies:a) Use of low-voltage row and column CMOS electronicdrivers-The drive electronics of a matrix-addressedFPD constitute almost half the cost. Because of thelow voltage and power requirements, the LCD driverscan be fabricated in LSI with up to approximately 180drivers per chip. Of all FPD drivers, CMOS driverscost the least and can only be used on LCDs.b) Separation of luminous power from image sig-nal-The row and column drivers do not haveto deliver the luminous power to the panel. Theluminous power is applied when used as a separatebacklight module. To conserve power or enhance theviewability, such as in a high ambient illuminatedenvironment, the luminous powers intensity ismodulated independent of the image. LCDs havebeen made with a dimming ratio of over 3000:1,exceeding even a CRT in performance. Further,the luminous efficiency of the backlight can beoptimized without regard to the LCD. Also, thespectral color of the backlight can be selected andoptimized to enhance the color properties of the

LCD and backlight combination. LCDs have ahigher luminous efficiency than that of other FPDsdue to the availability of high-efficiency fluorescentbacklights.c) Color capability and flexibility-Color filters areadded in combination to the LCD panel and backlightto make possible a wide spectrum of highly saturated502 PROCEEDINGSOF THE IEEE, VOL. 82, NO. 4. APRIL 1994


5/11

I- - I I1 -I-- (-*- YF

Fig. 2. Characteristics of different addressing methods.

colors. The selection of the three-color primaries isalmost unlimited due to the wide variety of pigmentsand dyes available. The RGB color filters are typicallyadded in front of the individual LCD subpixels insidethe front glass substrate of the panel. Other FPDtechnologies, such as VFDs, ELDs, and PDPs,require a unique phosphor or gas emission to achievebright and efficient colors which, in general, have notyet been developed for all the colors.d) Immunity to ambient illumination-Through the useof polarizers, the LCD is a nonreflecting, or black,display. As a consequence, when optimized, it istransparent to ambient illumination. Further, the col-ors maintain their chromaticity coordinates in varyingambient lighting. To enhance this feature, an an-tireflecting coating is added to the first surface ofthe display and low reflecting black matrix is addedaround all the areas between the active display areaof each pixel or subpixel.

VI. TECHNICALHALLENGEThe inability to achieve a low-cost, general-purposeHIC FPD s due to several difficult technical issues. Thedisplay is the most complex of all electronic components orsubsystems. In addition to the usual functional parametersof electronic equipment, there are optical issues, luminousefficiency, spectral emission, and human factors includingissues of photometry, size, color, readability, and dimma-bility. Ambient light is added to the list of environmental

operating conditions of temperature, humidity, shock, vi-bration, etc.There have always been markets which have been willingto pay premium prices for FPDs. The niche industrial andmilitary computer markets supported the early develop-ments of HIC FPDs using ELD, PDP, and, to a lesserextent, LED technologies. Typically, these displays were

---. .--I-

monochrome, from 6 to 12 diagonal inches in size with512x 512 lines of resolution.Technical issues have limited the commercial develop-ment of FPDs. The primary issue has been the ability toaddress a large array of pixels at a suitable speed withappropriate optical contrast, luminance, resolution, powerefficiency, color and gray shades, all at an affordable cost.A . Matrix Addressing

The design evolution process starts with an electroopticaleffect of a technology, such as ELD, PDP, or LCD, andthen expands the array size until the display performancerequirements can no longer be achieved.As shown in Fig. 2, there are only five known techniquesfor addressing an array of pixels [3 , p. 1191. The success ofthe CRT is directly attributable to the simplicity of scan ad-dressing in combination with appropriate phosphors. Thusfar, scan addressing has not been successfully configuredfor a low-profile, flat configuration. Grid addressing, asapplied to flat cathodoluminescence, has had only limitedproduction success in VFDs.In the 1960s and 1970s shift addressing was verysuccessfully applied to the Burroughs Selfscan productline of PDPs. When Burroughs stopped making displays,however, no one else continued with this technology.Burroughs gas shift register was then surpassed by theCMOS column driver shift register now used in PDPs.Matrix addressing is used to save electronic driver cost,which becomes an issue at 30 or more pixels. A displaywith 30 pixels would require 30 drivers in direct drive, or1 1 drivers in a 5x 6 matrix addressing arrangement.The direct addressing of each pixel in a display is anobvious approach and is used extensively for the lowest oflow-information-contentdisplays where less than 30 pixelsare used. There are typically 7 pixels for each numericcharacter , plus 1 for decimal or colon. In a typical computer

TANNAS: EVOLUTION OF FLAT-PANEL DISPLAYS 503


6/11

display, with typically 480 rows of 640 pixels, or 307 200total pixels, it is technically impossible to electricallyconnect to each pixel individually.Matrix addressing has been the most fruitful way toaddress a large array of pixels. Each pixel is in a row anda column which can be addressed by common electrodes.In the computer example above, the 480 rows and 640columns are each connected by an electrode for matrixaddressing as shown in Fig. 2. In this technique, thereare 480 row signal drivers and 640 column signal driversconnected to the edges of the display panel.B . Cross Coupl ing

Matrix addressing an LCD is analogous to addressingan electronic core memory device. In a core memory, crosscoupling is preventable due to the hysteresis nonlinearity ofthe magnetic toroid at each memory location. In displays,the pixel is analogous to the memory toroid, but, for mostdisplay material, the pixel has the electrical properties ofa lossy capacitor. One plate of the capacitor is made upof the column electrode and the other, the row electrodewith the FPD material, such as LC organic compounds, ELphosphors, or PDP gases, in between. As a consequence,cross coupl ing is the most serious FP D problem in allmatrix-addressed FPDs, and the degree of cross couplingis in direct proportion to the number of rows. Furtherdiscussion to follow.Of all the addressing approaches summarized in Fig.2, matrix addressing is potentially the simplest overallwith the promise of the lowest cost. The structure of amatrix-addressed display panel is the simplest; however,the number of row and column drivers is the largest.Therefore, the cost is shifted from the panel structure tothe electronics. In PDP, LC or EL displays, nearly half thecost is attributable to the row and column drivers.Matrix addressing is done by parallel columns and se-quential rows in the same way that electrical engineersaddress memory and imaging arrays. Typically, the rowsare addressed as a function of time. The signal is shiftedin a serial-to-parallel register, then applied to the columnsone row at a time. The shift register, associated electronics,and column drivers are now made in LSI MO S chips insizes sufficient to drive 180 or more columns per chip atone time.Recently, there has been significant research to simul-taneously address all the rows and all the columns of anFPD. The most recent work has been published [7] andsuccessfully demonstrated by In Focus Systems and Motif,Inc. and was coined by In Focus as active addressing.Previous work on multiple row addressing was published byBell Labs [SI and Optrex [9]. The key to the concept is theuse of a set of orthogonal functions for the row signals. Thismethod is acceptable to LCDs because the LC molecularrotating movement is proportional to the rms value of thevoltage applied to the pixel capacitor. The rotation of theLC molecule affects the optical retardation properties ofthe display and, therefore, transmittance of light for imagecontrast.

The primary motivation for active addressing, as op-posed to line-at-a-time matrix addressing, is to speed upthe response of the passive LCD at the cost of addedpre-processing of the display drive signal to make pas-sive LCDs more performance-competitivewith AMLCDs.Both Optrex and Motif13 have demonstrated approxi-mately a four-fold increase in speed for a 20% increasein cost, while viewing angle and contrast remain the same.The cross coupling in matrix addressing is difficult todescribe and compute without resorting to writing all theloop and node equations based on Kirchoffs Laws, andsolving the equations with matrix algebra. However, theequations are simplified by observing that all the pixels(loops) have identical impedance and all the row andcolumn electrodes (nodes) can be assumed to have zeroimpedance. If the pixels did not have identical impedance,and if the electrodes had a significant impedance, thedisplayed image would not be uniform,In FPDs cross coupling degrades the image directly. Theimage is seen by the viewer because there is a luminancecontrast difference between the pixels. This is usuallycharacterized as the contrast ratio of the luminance of thepixel commanded on to a neighboring pixel commandedoff. The maximum contrast ratio can be shown to beinversely proportional to the number of rows in the array,and independent of the number of columns, so long as thecolumns are addressed in parallel.Further it can be shown that there is nothing that can bedone in the external circuit to minimize the cross couplingbeyond making the voltage applied to all off pixels, one-third that of the on pixels. This can best be seen bydrawing a loop and node diagram of the entire matrix-addressed array and tying all the common nodes together.14The minimum effect of cross coupling is directly propor-tional to the responseof the pixel intended to be off, timesthe number of rows in the display panel when using line-at-a-time addressing and optimum voltages. The offpixel

looks more and more like an on pixel as the number ofrows increase; thus the contrast ratio becomes smaller andsmaller, which directly degrades the quality of the image.To minimize the impact of cross coupling, display materialis selected which has little or no response at the cross-coupled voltage. This is why thin-film EL, argon-neon gasmixture, and LEDs make good FPD materials.The response of a cross-coupled EL pixel i s down byas much as six orders of magnitude, gas mixes havea threshold voltage and do not respond at the cross-coupled voltage, and LEDs have minimal conduction whenreverse-biased. (At least one reverse-biased pixel occursin every cross-coupling current loop.) Further, TN LC,electrochromic, electrophoretic, powder EL, cathodolumi-nescent, incandescent, and many other technologies donot work well in matrix-addressed displays because these

I2JES93.l3 Flat Information Display Conference and Exhibition, December,14There is not sufficent space in this paper to develop the proof. A

1993, Santa Clara, CA.complete description of this analysis is given in [ 3, ch. 51.

50 4 PROCEEDINGS OF THE IEEE, VOL. 82, NO. 4, APRIL 1994


7/11

materials have a nearly linear response with applied voltage.A switch, like a TFT, or a nonlinearity, like MIM diode,is added at each pixel intemal to the panel to accomplishmatrix addressing with such technologies.Over the years, this cross-coupling problem has ledto many false promises in the display device industry.Typically, a display breadboard of a small number ofrows and columns is made and successfully operated withinsignificant cross coupling. The severity of cross couplingonly becomes apparent when the full-scale display with allrows operating is made and demonstrated.C . Du t y C y c l e

The second major issue with FFDs is the duty cycle orthe time spent tuming on a pixel or a row of pixels.In a CRT the duty cycle is the time that the electronbeam excites the area of the phosphor associated withone pixel. In a raster-scan CRT, the pixels are addressedsequentially. The duty cycle is then the reciprocal of theproduct of the number of pixels in each scan line, andthe number of scan lines in the raster. Thus for a CRTdisplaying VGA format (640x480) at 72 frames/s, the dutycycle is 1/(640x480) = 3 . 2 6 ~ 1 0 ~ ~nd the dwell time oneach pixel is (3 . 2 6 ~ / 72 = 46 ns. Fortunately, CRTphosphors can absorb sufficient energy during this shortdwell time to emit light. The resulting light is actuallyemitted long after the beam leaves the area during a periodof time called persistence. FPDs, in general, also respondwith a delayed optical effect.15The duty cycle of FPDs is made significantly greaterthan that of CRTs by addressing the columns in parallel. Asa consequence, the FPD dwell time for the VG A problemabove is 640 times longer, or 29 ps. This is fortunatebecause it takes a minimum of approximately 20 ps to tumon ELD and PDP pixels. This is purely a materials issueand cannot be altered significantly by the electronics.D. Luminous EficiencyThe display materials efficiency is a major issue. It hasbeen a continuous challenge to make displays as brightand efficient as possible. It tums out that, for FPDs, thereare only a few materials that qualify and are also matrix-addressable. The material must also have a high responsespeed because of the duty cycle consideration.Thus far, in the case of ELDs, zinc sulfide activatedwith manganese has been the only successful thin-film ELmaterial in production. It has a basic material efficiency ofapproximately four lumens per watt in monochrome andluminance in a display of about 90 candelas per squaremeter with a yellow-range color centered at 583 nm. Thereis promise of new materials since so many combinationsand types of hosts and activators exist. Planar demonstrateda prototype color ELD at SID93 held in Seattle, WA.However, this display is not ready for production due tolimited efficiency and brightness of the blue phosphor,among other things.

15Theone exception is LE D technology.

PDP materials are not as good in this category as ELDmaterials. The primary gas mixture used for monochromedisplays has been neon, typically combined with 0.1%argon, called the Penning Mixture. The efficiency isless than 1 lm/W and luminance in a display applicationis less than 100 cd/m2. The new color plasma displaydemonstrated by Fujitsu at JES93 has an efficacy of 0.7lm/W and luminance of 35 cd/m2. Photonics has alsodemonstrated a 19 diagonal inch color PDP with VG Aresolution with similar efficiency.The luminous efficiency of LCDs is much higher sincethey can be used in the reflective mode without a backlight.The color versions need a backlight due to the absorptionof the pixel color filters and polarizers, which reducesthe transmittance of an AMLCD to approximately 4%.Highly efficient fluorescent lamps that have a luminousefficiency of over 55 lm/W are typically used in consumerproducts. In both LCD cases, the brightness can be madeat any level, independent of the LCD panel, by simplyincreasing the intensity of the backlight. This gives a netluminous efficiency of 2.2 lm/W in color, which exceeds allother display technologies including CRTs under similarperformance conditions.E. Ambient Illumination

To the electronic displays engineer, the most perplexinginstallation problem is the impact of the ambient illu-mination reflecting off the display surface. The ambientillumination can be very high when compared to the emittedluminance. At the display surface, the reflected ambientillumination is added to the emitted luminance whichinevitably reduces the contrast ratio. In equation formLo, + ReflectionsContrast Ratio = Lotf +Reflections

where Lo, is the displays emitted luminance of the onpixel and Lotf is the luminance of the off pixel. Theluminance of the off pixel is due to cross coupling andintemal light scattering and light piping. The reflections arethe same, regardless of whether a pixel is on or off.The first surface reflections are typically 4% due to themismatch of the indices of refraction between air and glasswhich can be minimized with antireflection coatings usingan index-tapered sequence of thin films. However, this isonly the first of many surfaces in a typical display. In CRTsthe major problem is phosphor itself, which is an excellentLambertian reflector with typical reflectivity of 70%. This isthe principal reason why CRTs cannot be used in the brightoutdoors without special filters and additional power.The classical displays engineers technique to counteracthigh ambient illumination is to use antireflective coat-ings for the first surface and neutral density filters forintemal reflections. The neutral density filter always helpsthe contrast ratio, as the ambient illumination must passthrough the filter twice-once going in and once whenreflected back-whereas the emitted luminance need onlypass through the neutral density filter once going out. Theproblem with this approach is that the display now gets

TANNAS: EVOLUTION OF FLAT-PANEL DISPLAYS

1505


8/11

dimmer. The classical solution is to increase the emittedluminance. The consequence is a larger power requirement,shorter life, etc., and the solution works well up to the pointwhere the display cannot produce enough luminance to bereadable at the highest illumination.The second technique to get further contrast ratio im-provements is to use narrow-band-emission phosphors forthe display and a notch filter to match the emission ofthe display. This technique, using P43 phosphor, was abreakthrough necessary to make avionic CRTs readablein direct sunlight.A third technique used often in LED, VF, EL, and PDPFPDs is to use a circular polarizer that traps much of thereflected ambient illumination due to a phase shift of 180of the incoming light at the reflecting surface.In general, a combination of antireflective coatings, neu-tral density filtering, notch filtering, and circular polarizersis used. These techniques improve a display after it has beenoptically cleaned up, i.e., had all reflections stopped to thegreatest extent possible. One of the most effective cleanupsis a black matrix used to blacken all the nonemittingareas between the pixels. A black matrix is now used onmost displays. Most of the other cleanup techniques onecan imagine, such as using a black layer behind the ELDtransparent phosphors, black phosphors for any displayusing phosphors, black dielectrics, black electrodes, etc.,have never been fully realized due to fundamental materialsissues of dielectrics and conductors.When compared to all other emitting displays, back-lit LCDs are unique. Liquid-crystal displays are black-absorbing displays because they use polarizers front andback. As such, they cannot benefit from neutral densityfiltering, notch filtering, or circular polarizers. The contrastratio is achieved through the difference between absorbedand transmitted light. The light can be transmitted from acontinuously emitting backlight, reflected ambient illumi-nation, or both. Each of the color filters of an LCD actslike a switchable notch filter and renders an LCD highlyimmune to ambient illumination.Liquid-crystal displays act like printed inks on a contrast-ing background in the reflecting mode. In the transmissivemode, the backlights luminance can be increased inde-pendent of the image on the LCD to an almost unlimiteddegree.This immunity to ambient illumination is the single mostimportant performance advantage of LCDs over all otheremitting displays.F . Color and Gray Shades

It is the consensus of the display user community, that afull-color display must have three highly saturated primariesof red, green, and blue and that each primary color musthave 256 shades (8 b), for 16 million colors (24 b).Full color is easily achieved with modem CRTs, but itis extremely difficult in FPD technologies. The reasons aretwo-fold: 1) the basic materials necessary to emit all threesaturated primaries satisfactorily are not available in anyof the FPD technologies, and 2) mamx addressing requires

nonlinear pixel response, which is the exact opposite of theresponse required to achieve good gray shades.Liquid-crystal displays respond differently than otherlight-emitting FFDs. Good-quality saturated colors inLCDs are achieved through the combination of narrowbandemitting backlights and color filters. Several shades ofgray can be achieved through a combination of amplitude,pulsewidth, and frequency modulation to overcome thenonlinear pixel response of passive LCDs.Full color has only been achieved recently by one modeof FPD, the a-Si TFT AMLCD. The active matrix usesa TF T at each subpixel primary color (red, green, andblue). Several manufacturers used this mode of LCD todemonstrate a full 16 million colors, for the first time, atJES93. The AMLCD can display full color because theTFT at each subpixel acts as a switch to provide sufficientnonlinearity for matrix addressing. When the switch istumed on by a row line, the pixel capacitance is chargedlinearly, and the TN mode of the LCD is used and respondsin proportion to the rms voltage. A gamma correction curveis used in the buffering electronics to optimally linearizethe LC pixel response.VII. SUCCESSES OF LIQUID-CRYSTALISPLAYSOver the years, LCDs have evolved as the leadingFPD technology. They are used at all market levels, atall but the very largest sizes, at all but the very highestresolutions, and at all performance levels achievable byany FPD. The chronology is as follows: 1972-hand-heldcalculators; 1973-clocks; 197Lmult imeters , gas pumps,fish finders; 1977-wrist watches; 1978-avionic numer-ics; 1980-small monochrome portable TVs; 1985-smallcolor portable TVs; 1987-monochrome personal comput-ers; 199 -color portable personal computers; 1993-HICcolor avionics. Only in the last five years have they becomethe dominant HIC FPD.Initially, LCDs were used for their sunlight readabilityin reflective mode operating at extremely low power. It wasnot until the late 1980s that the industry made a culturalshift and added backlighting for a significant improvementin brightness at the cost of significantly greater power. Thiswas a major tuming point for LCDs. The combination ofimproved matrix addressing and the addition of backlightsmeans that LCDs can be produced with brightness andcontrast equal to any other display, including the CRT.At the same time, LCDs maintain their immunity to highambient illumination.At first, the use of the backlight was a great impediment.In addition to causing LCDs to use just as much power asother displays, it added weight and thickness. However,a new lighting segment to the industry has developedto counteract these negative aspects. Backlights are nowmanufactured less than 3 mm thick and have over 55-lm/Wefficacy, using flat fluorescent lamps and flat waveguides touniformly distribute the light over the area of the display.Further, the use of tri-band phosphors in fluorescent lighthas the ideal spectral emissions matching the color filtersof LCDs.

506 PROCEEDINGS OF THE IEEE, VOL. 82, NO . 4, APRIL 1994


9/11

The second major issue with LCDs is the viewing angle.This is a consequence of the optics of the phenomenoninvolved. Ten to one contrast ratio performance is typically45 to the right and left in the horizontal, +30 to -10 inthe vertical, quite adequate to the single viewer, but limitedfor multiple viewers. This has been a definite problemin aircraft panels where the pilot must see the copilotsdisplays and visa versa and in automotive displays whereall passengers may want to see the display.The AMLCD manufacturers are exploring new tech-niques for increasing the viewing angle. The extraordinarymethods include half-tone pixel, molecular arrangement formulti-domain pixels, and multiple rubbing for dual-domainpixels. These techniques can potentially double the viewingcone.The third area which could use some improvement isthe response speed. The speed of STN LCD is presentlytoo slow for video and games, but fast enough for graph-ics, word processing, and general computer-generated im-ages. The speed and performance of STN LCDs is betterfor smaller row counts because of less cross coupling.AMLCDs are fast enough for video games, thus there is asignificant performance and market application separationbetween STN and AMLCDs, with a commensurate costdifference. Displays engineers and researchers are chal-lenged to improve the speed and performance of STNLCDs to approach that of AMLCDs without significantlyincreasing the cost. So far, this is done principally bymultiple row addressing and electronically operating thedisplay as a split screen to reduce the number of effectiverows.However, the personal computer industry, which is, atpresent, the main customer and benefactor for these newdisplays, is quick to point out that the speed or viewingangle should not be improved if it increases cost.Cost is the single most important issue today. It willnot improve as long as the industry is production-limited.However, in Japan, the consensus is that, by 1996, thecolor 10-in VGA AMLCD will cost $600.00 each, in largevolume.VIII. CONSUMER RODUCTS

In spite of their problems, the principal impact of FPDshas been to enable the creation of new products onlypreviously dreamed ofa hand-held calculator more powerful than a main-frame computer of the 1950s, that can fit into onespocket;a wristwatch that can run for years on a small battery;portable televisions and VCRs;a personal computer that can fit into the palm of oneshand;electronic games, translators, personal digital assis-tants, etc. (see Fig. 3) .

The evolution of the color AMLCD is now at a pointwhere it can display a sufficiently detailed color mapat a cost commensurate for use in aircraft, automotive,

Fig. 3. A powerful information tool for the new age of comput-ing, the Expert Pad PI-7000 from Sharp Electronics Corporation isan intelligent assistant which makes information management andcommunications easier than ever. The STN-type L CD is used in thereflective mod e, making it highly sunlight-readable and extremelylow-power for maximum portability.

Fig. 4. An advanced portable global positioning system (GPS).Sony Electronics has introduced the Pyxis IPS-760 intelligent posi-tioning system fo r marine and aviation applications. A low-powerSTN-type LCD is used in optimizing portability and sunlightreadability.

and recreational applications. The navigational aspects arefacilitated by global positioning satellites. The map pageswill be facilitated by a laser disc. Such a system has beenmade in prototype form and is expected to be in high-volume production within a year. In Japan, productionsystems are now in use and available to consumers inAkihabara for $2000.00, as shown in Fig. 4.AMLCDs have also impacted the avionics industry. Inthe cockpit of a typical aircraft, there are many differentinstruments (engine instruments, radar displays, altimeters,directional gyros, etc.) made by many different manu-facturers, using many different technologies. All theseinstruments may be replaced with an AMLCD-based in-strument which, in conjunction with software, will lookand function like the old altimeter, radar, or directionalgyro. The utility of this is that now all the instruments arefunctionally and mechanically interchangeable, reducing

TANNAS: EVOLUTION OF FLAT-PANEL DISPLAYS 507


10/11

Fig. 5. Sharp Corporation has developed the worlds firstwide-vision multimedia 16.5-in color (16.7 million tones) TFTLCD with an RGB pixel resolution of 480x853 ( 1 228320addressable dots). This is the first direct-view LCD in the HDTVaspect ratio to be demonstrated.No date has been announced foravailability of samples.

part numbers, increasing cockpit instrument redundancy,and improving reliability. Over the next ten years the entireaviation avionics industry will be reshaped by this newAMLC FPD technological application.The new personal communicator will be changing ourwhole way of communicating and accessing data. Theintegration of the cellular telephone, personal computer,and satellite datalink with improved FPDs and inpudoutputdevices will have incalculable influence on our lives.IX. S U M M Y AN D HDTV

The state of the art of the most advanced FPD is,perhaps, exemplified by the general-purpose AMLCD asshown by Sharp at JES93 (Fig. 5) . It is targeted for thepersonal portable computer market. At the same show,Sharp demonstrated a 17-in diagonal color AMLCD with1024x 1280 lines of resolution targeted for the personalportable workstation with multimedia capability. Fujitsushowed a 21-in color PDP with 640x480 lines of res-olution, targeted for large workstations and productionscheduled for 1994. Other companies have shown similardisplays. Sharp, now the leader in the LCD market, an-nounced they were investing another $750 M to upgradeand expand their production facility in Tenri and startconstruction on a new AMLCD plant in Mie Prefecture.By March of 1994, Sharp expects to yield 200OOO AMLCDcolor 10-in displays per month.A . HDTV Display

The next challenge, the HDTV display, presents anotherlevel of technical complexity beyond the personal portablecomputer display. Today, we do not yet have a consumerpriced NTSC Picture on the Wall, much less an HDTVversion. Sharp marketed a TV Picture on the Wall, usingan 8.4-in color AMLCD, for approximately $8.5 K in1992; but, the viewing angle is limited and the cost is

too high. There are only three candidate direct-view FPDapproaches for HDTV under development today, which aresummarized below. There are several FPD projector styleHDTVs developed by Sharp, Sanyo, Seiko-Epson, TexasInstruments, and others, which are discussed in anotherarticle in this issue.In Japan, there are two major direct-view FPD thrustsfor flat-panel HDTV displays: 1) the NHK plasma panelapproach and 2) the GEC (Giant Electronics Company)Consortium approach using polysilicon TFTs in AMLCDs.The NHK plasma approach uses dc gas discharge togenerate UV which then excites color-emitting phosphors.Three phosphors are used for each pixel-ne each forred, green, and blue-using spatial color.16 NHK has madeseveral large panels for demonstration purposes. Matsushitahas packaged the NHK panels and attached the driveelectronics so they can display HDTV imagery. The plasmaapproach has the advantage that the large (40- o 60-in diagonal) display can be made using screen printingtechnology. However, the panels presently have a lumi-nous efficiency of significantly less than 1 lm/W and lowluminance. (Researchers are concentrating on improvingthe luminous efficiency and luminance, and some progresshas been demonstrated by Fujitsu and Photonics.) Also,the phosphor and phosphor protective shield are potentiallydegraded by the gas discharge reactants and UV radiation.Matsushita demonstrated a 40-in color HDTV with nearlyfull resolution at JES93.The most detrimental attribute of all color plasma panelsis the low luminous efficiency. There need to be significantmaterials advances before plasma can be advanced. Theplasma technology has been stalled here for a prolongedperiod and most avenues of relief appear exhausted.The Hitachi-led GEC consortium, along with MITI sup-port, uses a precision printing approach to make the TFTsfor the AMLCD. The initial concept was to make a 40-in panel using polysilicon as the semiconductor of theAMLCD TFTs. The ultimate hope would be to print theperipheral row and column drivers at the same time theTFTs are printed.The printing of the TFTs to the precision required over alarge area is difficult and beyond the present state of the art.While researchers have successfully printed small arrays ofpixels, the original schedule to make a 40-in panel by 1994has been reduced to making a 20-in section of the full panelfor demonstration and evaluation.A third approach, by Tektronix, called plasma-addressedliquid-crystal (PALC) display, has great promise as adirect-view HDTV FPD. Tektronix made an excellentdemonstration of a 16-in version of their concept at SID93.The display used a non-light-emittinggas discharge array ofrow lines in a back plane that is the source of electrons forenabling a row at a time in a conventional matrix addressingtechnique. The LC and color filters are located in the frontplane and addressed through column electrodes. The gas

I6Spatial color is achieved by dividing the pixel area into typically threesubpixels, one each for red, green, and blue. The subpixels are so smallthat, at the nominal view ing distance, they are not resolvable.508

71 --PROCEEDINGS OF THE IEEE, VOL. 82, NO.4,APRIL 1994


11/11

discharge generates a row line of available charges whichis attracted to a common glass plate when any column isaddressed to tum on an LCD pixel. The charge collects onthe plate and keeps the LC pixel on until the scan retums.The entire display is backlit with a fluorescent lamp.The PALC display has several characteristics makingit appropriate for HDTV. There is no conventional crosscoupling since the row addressing and column addressingare not electrically interconnected. All the circuitry of thepanel can be screen-printed to keep manufacturing costsdown. There is minimal technical risk as the gas dischargeand the T N LC phenomena are well understood. On theother hand, the intermediate glass plane is difficult to makethin enough to keep the voltage down, and the backlighttransmittance is lower than desired.B . HDTV Summary

The path to the development of a consumer-pricedHDTVdisplay by any means has not been identified. In Japan,where HDTV is operational, the 33-in direct-view CR THDTV displays are now being sold for $8 K with MUSEdecoders.For HDTV to be a consumer success the marketingconsensus is that the TV set must cost less the $4 K. Also, to

be successful in the U.S. (based on room size and geometry)the image should be 60-in on diagonal. The display shouldhave a resolution of approximately 1000 rows by 1500columns, with an aspect ratio of 9:16, which correspondsto the nominal standards in vogue today.At present, it can clearly be said that the next identifiablefrontier in displays is the HDTV display. There is afavored solution to HDTV-that it be flat and that ithang on the wall. However, the most feasible technicalapproach is in projection both rear and front and the leadingtechnology is the CRT. Significant research is being appliedto polysilicon TFT AMLCD. Sharp made the first everpublic demonstration of polysilicon TFT AMLCD (1.3-in substrate) HDTV projector at JES93. No details wereavailable. Stay tuned.

REFERENCESMITI data as published by Dempa Publications Inc., in JapanElectronics Almanac9394.Private conversation with K. Odawaro, Hitachi Ltd.Flat Panel Displays and CRTs, L . E. Tannas, Jr., Ed. Ne wYork Van Nostrand Reinhold, 1985.JTEC Report on Display Technology in Japan, 1992;W Z SRep.T. J. Scheffer and J . Nehring, Appl. Phys. Lett., vol. 45, p.1021, 1984.S. Morozumi et al., B/S and color LC video display addressedby poly-Si TITS, in SID 83 D ig.T. Scheffer and B. Clifton, Active addressing method forhigh-contrast video-rate STN display, in SID 92 Dig.J. Nehring and A. Kme tz, Ultimate limits for matrix addressingof rms responding liquid crystal displays, IEEE Trans. ElectronDevices, vol. ED-2 6, pp. 795-802, 1979.93 Japan Electronics Show Demonstration and Reports, Re -search Lab., Asahi Glass Co., Ltd., vol. 43, no. 1, 1993.Ibid., pp. 1 0 6 1 1 2 .J. Hirate et al., Viewing angle evaluation method for L CD swith gray scale images, in SID93 Dig. , pp . 561-564, 1993.

PB92- 100247.

Lawrence E. Tannas, Jr. (Senior Member,IEEE) received the B.S.E.E. degree in 1959and the M.S.E.E. degree in 1960, bothfrom the University of Califomia in LosAngeles.Prior to 1983, he worked as an individualcontributor and engineering manager at theGE Research Laboratories, Honeywell, MartinMarie tta, Rockwell Intema tional, and AerojetElectroSystems. Since 1983 he has beenPresident of Tannas Electron ics,an intemationalindependent consultant and lecturer on electronic information displays,consulting on technology, market studies, and designs. For the past tenyears, he has consulted and given seminars for numerous Fortune 50 0companies, govemment agencies, and universities. He recently served as amember and Co-chairman ofUS. ational Science Foundation Committeeto study Japans displays industry and as an expert witness before the U.S.Intem ational Trade Commission. For over ten years he has organized andtaught a series of Displays Engineering short courses at UCLA. He hasbeen awarded six patents, a NASA Disclosure, and a NASA Certificateof Recognition. In addition to the publication of numerous technicalpapers in various proceedings, digests, and joumals, he is the edito r of thebook Flar-Panel Displays and CRTs, publishe d in 1985 by Van NostrandReinhold, New York. Mr. Tannas is a fellow and past president of theSociety for Information Display (SID). Additionally, he is a Member ofAIAA, SPIE , IS&T, AVS, Human Factors Socie ty, and others.

TANNAS: EVOLUTION OF FLAT-PANEL DISPLAYS- 1 ~ 509

evolution of flat panel display

Documents