An Electroluminescent Light-Amplifying Picture Panel

PROCEEDINGS OF TIHE IRE

Another advanitage of these evaporated CdS films istheir high dark resistance.Although for the particular photoconductor used in

this experimental storage light amplifier n = 0.4, it ispossible to produce photoconductors with either a loweror higher value for n, which may even exceed n= 1.A description of the processing of these evaporated

films was given by Arthur Bramley at the Chicagomeeting of the American Physical Society.6 It may beadded that the CdS was evaporated from a tantalum

6 A. Bramley, "Deposition of CdS in vactuuim," Phys. Rev., vol. 98,p. 246, abstract VI1; April 1, 1955.

boat onto the prepared NESA coating. This preparationiusually consisted in depositing a very thin film of Agon the NESA coating and the gap in it. At the high tem-perature, at which the NESA coated glass plate wasmaintained during the evaporation and which was stillfurther raised during the aging, the Ag diffuses into theCdS film evaporated over the Ag film. On the other handsince the temperature of the tantalum boat was notraised beyond 850°C, the temperature of evaporation,the amount of impurity Cd evaporated should be lessthan 1 part to 105 parts of evaporated CdS. As men-tioned in footnote 6, the evaporated CdS films wereaged by baking to 350°C in air for 10 minutes.

An Electroluminescent Light-Amplifying Picture Panel*B. KAZANt, SENIOR MEMBER, IRE, AND F. H. NICOLLt, SENIOR MEMBER, IRE

Summary-This paper describes an experimental amplifier forlight which uses a photoconductive layer electrically in series with anelectroluminescent phosphor layer. With the series combinationexcited by an alternating voltage of audio frequency; e.g., 400 cps,a low light level impinging on the photoconductor decreases its re-sistance sufficiently to cause a much larger light output from thephosphor. Large-area light-amplifying panels, 12 inches square, usinga newly developed photoconductive CdS powder have been built.These panels are capable of producing intensified half-tone imageswith high resolution. The response time of these panels, determinedbasically by the photoconductive material, varies from 0.1 secondto several seconds. New photoconductive materials offer promiseof greatly increased speed. Since the photoconductors can be madesensitive to X-rays and infrared, the principles of the device may beuseful for image conversion and intensification with these radiationsources.

INTRODUCTIONrlf HE POSSIBILITY of a panel type of light ampli-

fier was initially demonstrated by the operation ofan electroluminescent panel1 in series with a small

photoconductive crystal of CdS as shown in Fig. 1.In the dark the high impedance of the CdS crystalpermitted only a negligible fraction of the alternatingsupply voltage to be applied across the electrolumines-cent layer. As the incident light on the crystal was in-creased, its impedance decreased and greater voltageappeared across the phosphor layer. It was further

* Original manuscript received by the IRE, September 1, 1955.t RCA Laboratories, Princeton, N. J.I E. C. Payne, E. L. Mager, and C. W. Jerome, "Electrolumines-

cence, a new method of producing light," Ill. Eng., vol. 45, pp. 688-693; November, 1950.

demonstrated that, with equal areas, the lumensemitted by the phosphor could be many times greater(e.g., 50 times) than the lumens incident oni the crystal.

INCIDENT LIGHT

/ CdS ELECTROLUMINESCENT

CONDUCTING EMITTEDCOATING LIGHT

APPLI ED A-CVOLTAGCE

Fig. 1-Elemental light amplifier.

The experiment described used small single crystalsof CdS. Subsequently, however, a powder was developedwhich could he used for making large-area photoconduc-tive cells.2 This material enabled the practical construc-tion of a thin large-area light-intensifying panel capableof picture reproduction.

2 F. H. Nicoll and B. Kazan, "Large-area high-current photocon-ductive cells using CdS powder," Jour. Opt. Soc. Amer., vol. 45,pp. 647-650; August, 1955.

188 Decembher

Kazan and Nicoll: A Light-Amplifying Picture Panel

INCIDENT LIGHT IMAGE

AC VOLTAGESUPPLY

TRANSPARENTCONDUCTINGCOATING

GLASS PLATES

OBSERVED ELECTROLUMINESCENT PICTURE

Fig. 2-Sandwich type light amplifier.

The structure initially selected was sandwich-like,using a thin photoconductive layer and a layer of elec-troluminescent material held between two pieces ofglass as shown in Fig. 2. A similar sandwich-like struc-ture was also suggested by another laboratory.3 Exten-sive experiments with such structures showed numerousshortcomings, in particular the severe limitation ingain caused by the opacity of the thick photoconductivelayer required to control the electroluminescent layer.The development of new structures was thus required.

INTERDIGITAL-ELECTRODE TYPE AMPLIFIER

To illuminate the photoconductor more efficiently,structure of Fig. 3 (next page) was devised. A set of fineinterdigital transparent conducting electrodes on a glassplate was covered with a thin continuous layer of elec-troluminescent phosphor. Above phosphor, another thinlayer of photoconductive powder was deposited. Withac voltage applied between two sets of electrodes inthe dark, there is relatively little ac current flow be-tween them due to the high impedance of the ma-terials between the conducting lines. When portionsof the photoconductive layer are illuminated, however,relatively low impedance paths are provided at theseareas between the electrodes. Since the resulting in-crease in ac currents must flow through the phosphorlayer, which is between the photoconductor and theelectrodes, electroluminescent light is emitted adjacent

3R. K. Orthuber anid L. R. Ullery, "A solid-state image inteiisi-fier," Jour. Opt. Soc. Amer., vol. 44, pp. 297-299; April, 1954.

to the corresponding illuminated areas of the photo-conductor.

In operation, it was found that most of the photo-currents flowed only through the regions of the phosphorlayer adjacent to the edges of the electrodes. This re-duced the sensitivity and made it difficult to make theimpedance of the phosphor elements low enough withrespect to the illuminated photoconductive elements.Although the principles were sound, it appeared possibleto obtain greater light gains with other structures.

MESH-SUPPORTED-PHOTOCONDUCTOR TYPE AMPLIFIEROne such structure is in Fig. 4 (p. 1891). A transpar-

ent conducting coating on glass was sprayed with a thinphosphor layer. To prevent light feedback, a thin in-sulating opaque layer of lampblack in binder was thensprayed on the phosphor layer. A fine metallic mesh4 wasthen stretched approximately 15 mils above the surfaceof the opaque layer. The photoconductive powder in abinder was then spread over the mesh, forming a curvedsurface within each mesh hole, to improve light penetra-tion. The above device was relatively effcient, but wasdifficult to make uniform.

RIDGED-PHOTOCONDUCTOR TYPE AMPLIFIER

Another arrangement is in Fig. 5 (p. 1891). A groovedlucite plate, with many fine parallel "V" grooves, has afine conducting line of silver paint at the bottom of

4 "Lektromesh," 25 square holes to the linear inch, made of nickel,with about 50 per cent transmission.

19O55 1 889

PROCEEDINGS OF THE IRE

I NCIDENTLIGHT IMAGE

ELECTROLU MINESCENTLAYER

PHOTOCONDUCTI V ELAYER

A.C. VOLTAGE SOURCE

TRANSPARENTC ON DUCTINGLI NES

-- O-BOBSERVED_- ELECTROLUMINESCENT.-- PICTURE

-~~G LASS P LAT E

Fig. 3-Interdigital type light amplifier.

each groove. These grooves are filled with a bondedphotoconductive powder. An opaque current-diffusinglayer (consisting of a conducting form of CdS powder)is interposed between the grooved plate and the electro-luminescent phosphor which is bonded to the trans-parent conducting coating of a glass plate. The ridgesof the photoconductor provide continuous conductingpaths for the photocurrents from the electrode at theapex of the photoconductor to the bottom of the photo-conductor grooves. By means of the auxiliary current-diffusing layer, photocurrents which otherwise wouldenter the electroluminescent layer at restricted regionsnear the groove bottoms are caused to spread or dif-fuse slightly before entering the phosphor layer. Sincethe amount of diffusion is limited to about the width ofa single photoconductive groove, essentially all of the

phosphor layer can be excited by the photoconductivelayer and at the same time resolution of device, basicallydetermined by distance between grooves, is not signifi-cantly affected. Current-diffusing layer, being opaque,serves additional function of preventing feedback.

This arrangement enables efficient illumination of thephotoconductor, and efficient excitation of the phosphor.Devices up to 12 inchesX 12 inches in area operatedwith high gain and good resolution but conductivityvariations in the dry CdS powder used for the diffusinglayer were sometimes evident.

GROOVED-PHOTOCONDUCTOR TYPE AMPLIFIER

Structure of Fig. 6 (p. 1892) had the advantageousfeatures of previous structure, but did not require agrooved lucite plate and used a current-diffusing powder

1890 December



ELECTROFORMEDMESH

AC VOLTAGESOURCE-- .

ELE

OBSERVED ELECTROLUMINESCENT PICTURE

Fig. 4-Mesh-supported-photoconductor type light amplifier.

in bonded form. A transparent conducting coating on aglass plate is sprayed with a thin (about 1 mil thick)phosphor layer. After covering the phosphor layer withan opaque layer (lampblack in Araldite, a fraction of amil thick), a heavier current-diffusing layer of conduct-ing CdS powder (bonded with Araldite and initiallyabout 20 mils thick) is spread on the opaque layer andmachined flat (to about 10 mils in thickness). A heavylayer of bonded photoconductive CdS powder (bondedin Araldite approximately 20 mils thick) is spread on theconducting CdS and again machined flat (to about 14mils). After spraying the photoconductor surface withair-drying silver paint, fine "V" grooves (of about 60degrees included angle) are cut into the photoconductor,(15 mils deep, and 25 mils between centers). The bottomof the "V" grooves cuts slightly into the conducting CdSlayer and the tops of the grooves are left with narrowconducting silver lines, several mils wide, which, con-nected to a common terminal, act as one electrode forthe device.'

Structures such as described in Fig. 6 were made upto 12 inches X12inches in area. The gain and resolutionequalled those of the grooved lucite plate structure ofFig. 5. In terms of resolution the output pictures werecomparable in quality with commerical tv pictures andhad very good uniformity.

5 The structure of Fig. 6 required machinable plastic-bondedlayers of photoconductive powder which retained full photosensi-tivity. This was accomplished by using CN-502 Araldite suitablydiluted with diacetone alcohol. It was noted that the plastic binderhad marked effects on the response time of certain batches of CdSphotoconductive powder. Such changes in response time were pro-gressively acquired as the samples cured over a period of severaldays, with the photosensitivity also changing. These phenomena arenot yet completely understood.

A.C. VOLTAGE SOURCE

INCIDENT LIGHTCONDUCTINGIMAGE ~~~~~~LINES

I~ -OBSERVEDELECTRO-LUMINESCENT

ELECTRODE PICTUREFOR CONDUCT-

,,-GLASS PLATE

- _ / TRANSPARENTCONDUCTINGCOATING

GROOVED ELECTROLUMINESCENTLUCITE LAYERPLATE OPAQUE CURRENT-DIFFUSING

LAYER

PHOTOCONDUCTORFILLING GROOVES

Fig. 5-Ridged-photoconductor type light amplifier.

Fig. 7 (next page) is a photograph of an early 12-inchgrooved photoconductor-type amplifier with an ampli-fied output picture produced by projecting a low levelimage on photoconductor side. Small dark spots are at-tributed to imperfect bonding of the photoconductiveand current-diffusing layers. The few narrow dark linesvisible in the picture are caused by disconnected con-ducting lines.

1955 1891


A. C. VOLTAGESOURCE

GLASS PLATE

OBSERVEDELECTROLUMINESCENT

PICTURE

TRANSPARENTCONDUCTING COATING

ELECTROLUMINESCENTLAYER

OPAOUE LAYECURRENT- DIFFUSING

LAYER


Fig. 6-Grooved-photoconductor type light amplifier.

er, photoconductive layer and current diffusing layer.The electroluminescent layer is a capacitor with

negligible losses at the audio frequencies. For layers 1mil thick as used, the capacitance is about 200 ,u,uf percm'. When excited with ac voltage, phosphor used emitslight whose average value is given approximately by:

Lo = k,V( 3)(,0(.7) (1)

Fig. 7-Photograph of grooved-photoconductor type light amplifiershowing amplified picture resulting from a projected image.

COMPONENT CHARACTERISTICSThree major components of ridged and grooved pho-

toconductor-type amplifiers are electroluminescent lay-

where k1 is approximately 3 X 10-10, V is the rms voltageapplied, w/27r is the frequency, and Lo is luminance infoot lamberts. Fig. 8 (next page) shows a curve of meas-ured values of electroluminescent light output as a func-tion of applied ac voltage at a fixed frequency of 400 cps.Decay time of phosphor upon removal of exciting volt-age is relatively short (approximately 1 milli-second).The phosphor response time, however, upon sudden ap-plication of voltage depends on its previous excitationand on the audio frequency used, being in the range ofone to 30 milliseconds.The photoconductive powder with a given dc field

across it has approximately linear conductivity withincident light intensity. On the other hand, for a givenlight intensity, the conductivity of a cell varies as arelatively high power of the field. The measured dccurrent may be approximately written:

J,,c = k2LV' + Id, (2)

1892 December


09

,06

.05

.074

.02;

.oil-10 2 3 4 56789100 2 3 45 6 789 1

.°00890HHH=Hi jj~~~~~~~100

A- C VOLTS(R.M.S.)Fig. 8-Curve of electroluminescent light vs applied ac voltage.

where k2 is a constant, Li is the input illuminance in footcandles, V is the dc voltage applied, n is approximately4 or greater, and Id is the dark current. The dark cur-rent, Id, varies as a high power of the applied voltage,but is usually small compared with the first term.

If an ac voltage is applied to the photoconductor,the instantaneous photocurrent may be represented bythe expression:

'ac = - Li(Vp sin cot) n - jwcCVP sin cot (3)10

where Vp sin cot is the applied ac voltage, C is the capaci-tance of the layer, and w/2ir is the frequency. In general,the dark current under ac conditions is very muchsmaller than the capacitive current, wCV, sin cot, and canbe neglected. The constant k2 for the dc expression (2)has been changed to k2/10 for the ac expression (3).This factor of 10, experimentally determined, is un-changed in the range of 300 to 4,000 cps.

If the measured dc photocurrents are plotted on alinear scale, a curve as shown in Fig. 9 is produced. Overthe range of dc voltage used, the photocurrent per cm2,neglecting the dark current can be approximately ex-pressed as follows instead of by (2):

D.C. VOLTS

Fig. 9-Curve of photocurrent vs applied dc voltage.

Idc = (1/R)(V - VT) when VI > VTand Idc = when VI < VT. (4)

In the above expression VT represents the thresholdvoltage; i.e., the point of intersection of the straight-line sloping portion of the curve with the voltage axisand R the incremental resistance of the photoconductoralong the sloping portion. Since 1/R of the photoconduc-tor varies linearly with the input light intensity, Li,the above expression can be rewritten:

Ide = k3Li(V - VT) when V| > VTand Idc = ° when I VI < VT. (4a)

For the case of ac applied voltage we obtain:

Iac = (k3/1O)Li(V, sin cot - VT) -jcoCV, sin cotwhen Vp sinCwtI > VT

and Iac = 0 when I VpsincotI < VT. (5)A calculated value of k3 =1.25 X 10-s is obtained using

the dc curve of Fig. 9. The photoconductive cell usedhad electrodes of 20 mil spacing, an effective area ofphotoconductor of 0.75 cm2, and was illuminated with0.035 foot candles (2,870°K Tungsten source).

In the dark, the capacitance per cm2 of a photocon-ductive layer in the grooved lucite structure of Fig. 5is about 8,u,uf measured at 1,000 cps. This capacitanceis measured between the set of conducting lines at thebottom of the grooves and an electrode covering thetop surface of the photoconductor, thus including thecapacitance of the lucite.The photoconductor used has a response time be-

tween 0.1 and several seconds with the shortest responsetime at the high light levels. Since the photoconductor ismuch slower in response than the electroluminescentmaterial, it controls the amplifier operation.

Current-diffusing layer, consisting of specially pre-pared CdS powder, is a nonphotoconductive resistivematerial. Unlike photoconductor its conductivity is notlowered by ac voltage. Fig. 10 (next page) shows a typi-cal curve of dc current vs dc voltage for a 1 cm2 bonded

1955 1893


sample of this material 10 mils thick, with electrodeson opposite surfaces. Since operation of the lightamplifier requires currents below one milliampere/cm2,the voltage drop going through the layer is relativelysmall (<30 volts).

D.C. VOLTS

Fig. 10-Curve of current vs applied voltagefor current-diffusing layer.

As shown by the curve, the powder has a rapidly in-creasing impedance with a lowering of the voltage be-low 30 volts, which is useful for maintaining pictureresolution and small-area contrast.

ANALYSIS OF THE COMPLETE AMPLIFIER

The dark current and the capacitive current throughthe photoconductor layer are small and can be neglected.The effects of the current-diffusing layer can also beneglected since, as previously indicated, a relativelysmall voltage builds up across this layer at any time.With these assumptions, the problem simplifies to aseries combination of a nonlinear resistor (controlledby the incident light) and a capacitor with ac voltageapplied across the two. Using (3) for the photoconductor,the following differential equation applies.

kddq/dt = 2 LIi V, sin Cot - l]n10 (6)

where C=capacity per cm2 of phosphor layer, q=in-stantaneous charge on phosphor layer, and V,=peakvalue of applied ac voltage. Because (6) is awkward tosolve, it is more convenient to use the approximation,(4). This leads to:

dq VU sin wt-VVT-q/C-= R |~~~~VVsinwt - qlC > VTdt R

dqd- 0 V, sin wt-q/ClC < VT (7)di

conductor (= 1ORdc). Solution of this equation6 givesfor the charge, q, on the condenser:

VpC(sin ct - RCw cos cot)q = -VTC + R2C21 + R2C2C,12

( ~~~VTCFVT2 _+1{TC+ ql1-R2C2w,2L[Vp +CVp

- RCOV1 (- + q 2]}

* {e--[t-- arcsin (2 + (8)

where qi = charge on condenser at the start of a con-duction cycle.

Of interest is the maximum value of q which deter-mines the peak voltage on the capacitor (phosphor).Tests have shown that the light output of the electro-luminescent layer with the nonsinusoidal wave shapesacross it in the light amplifier, is approximately the sameas that obtained with a sine wave of equal peak value.Since q reaches a maximum at the end of each conduc-tion period, a solution is desired for its value at this timeunder cyclic conditions. However, under cyclic condi-tions the charge at the end of a conduction cycle must beminus that at the beginning of the cycle. Suppose q1 isthe charge at the end of the last conducting cycle so thatat the end of the new conducting cycle q = -ql. The timet, corresponding to the end of the new cycle can be deter-mined from the relation given with (7) for the conduct-ing condition: Vp sin t - qlC > V17. This relation gives,at the end of the conducting cycle,

1 inV(TiVpsinwt + q1/C = VT or t --arcsin

Substituting this value of t into (8) and using -ql forq leads to the condition:

-x+b---aVl-(b-x)2= [x+b+aV11-(b-x)2]*e [arc sin (b-x)-arc sin(b+x)l (9)

where

qix =

VpC1

aR=RwC

and

VTb-

VP

6 Solution of the equation and formtulation of expression (9) wereprovided by E. G. Ramberg.where VT= threshold voltage, R = ac resistance of photo-

1894 December


For very large a,

ax b-

Va2 + 1

for very small a,

-a(1 -b2)/1 - b2 + ab

In (9), x represents the fraction of the peak ac supplyvoltage which appears on the capacitor when it has itsmaximum charge. Eq. (9) can be solved by trial anderror giving a curve such as shown in Fig. 11 of x vs a.This particular curve was plotted with an assumedvalue of b= VT/VP of 0.65. Assuming a given peak acvoltage, Vp, actual values of peak voltage on capacitorcan be determined from Fig. 11 as factor, a, is varied.

.35

x.|T~~~~V 65 b

0 2 3 4 5 6

RwC

Fig. 11-Curve of fraction of peak voltage across electroluminescentlayer vs a(= 1/RwC).

Converting the values of x in Fig. 11 to light output,using the curve of Fig. 8, produces the calculated curveof Fig. 12, assuming a 1,000 volt peak ac supply. Thevalues of a are indicated at the top of Fig. 12.

Referring now to (5) for the photoconductor with k3=1.25 X 10<, the factor, a, can be written in terms ofthe input light on the photoconductor, [assuming C=200,u,f per cm2 and w = 27r(400)]:

Li = .04a.

The corresponding input light levels in lumens persquare foot are shown at the bottom edge of Fig. 12.Also shown in Fig. 12 is a curve of measured values ofoutput and input light using a grooved lucite plateamplifier, such as shown in Fig. 5, operated at 400 cycleswith an applied ac voltage of 1,000 volts peak.

Fig. 13 shows the curves of light gain as a functionof input light level, derived from the curves of Fig. 12.Although Figs. 12 and 13 indicate fairly good correla-tion between the measured and calculated values, suchcorrelation was obtained by using a value of b = 0.65.

This assumes a threshold voltage for the photoconduc-tor of 650 volts as compared to the threshold voltageof about 550 volts shown in Fig. 9 for the photoconduc-tor sample measured. The discrepancy is believed to bepartially due to the voltage drop across the current-diffusing layer being somewhat greater than assumed,since the current flow into it from the photoconductoris concentrated along narrow lines rather than spread

INPUT LIGHT (LUMENS PER SOUARE FT.)

Fig. 12-Curve of output lumens vs input lumens for light amplifier.

z404

C, 40 || CALCULATE

20

O .OS .15 .2INPUT LIGHT (LUMENS PER SQUARE FT.)

Fig. 13-Curve of gain vs input light for light amplifier.

over the surface as is the case when solid electrodes areused in measurements. In addition, correlation of photo-conductor resistance with light level is somewhat inac-curate due to the time constants and slow drift of thephotoconductor at low light levels. Although the calcu-lated operation of the light amplifier is approximate, themodel assumed for the photoconductor is reasonablyuseful.

1955 1895

PROCEEDINGS OF THIE IRE

Both the measured and calculated curves of Figs. 12and 13 are based on the use of a tungsten light sourceoperating at 2,87O)°K. Such a light source emits a con-siderable amount of infrared radiation to which thephotoconductor is sensitive, but which is not measuredby a foot candle meter. To measure the "intensity gain"of the light amplifier, i.e., the gain using input lightwhose spectral distribution is identical to the outputlight, a test was made using an electroluminescentsource panel to provide the input light. This source hada spectral distribution identical to that of the amplifieroutput, having been made with the same type of phos-phor. Under these conditions, the maximum energygain was about 14.

E LECT ROLU MI N ESC ENTl

OUTPUT ~~PHOTOCONDUCTOR

S00 600 700 800 900

MIl LUI MRICRONS

Fig. 14-Spectral curves for photoconductor andelectroluminescent phosphor.

Fig. 14 shows in relative values the spectral responsecurve of the photoconductor and the spectral distribu-tion curve for the electroluminescent phosphor usedin the light amplifier. These curves indicate that theoperation of the light amplifier with the yellow electro-luminescent input light reduces the gain to about 50per cent of the gain expected with an input light sourcehaving its peak coinciding with that of the photocon-ductor. With such matching conditions, the maximumenergy gain is about 30.

GAIN AND LIGHT OUTPUT AS AFFECTEDBY SUPPLY FREQUENCY

Both the measured and calculated gains are basedon the use of an ac voltage of 400 cycles. The effectsof varying the frequency can be seen by reference toFig. 12. Increasing Xo and decreasing R by a factor, K,causes no change in a (= 1/RcoC) and in the peak voltageapplied to the phosphor layer. Such an increase in thevalues of 1/R for the photoconductor, however, requiresthat the input light levels be multiplied by this samefactor. Since the voltage on the phosphor layer ischanged in frequency by a factor, K, the light outputis increased by the factor K( 7) as indicated by (1). Ifan audio frequency, f, other than 400 cycles is used, the

amplifier will have new gain curves of the same shapeas in Fig. 14, buit chatnged in level by the factor givenbelow:

(400).i4y o .3 (10)

Also, the input light levels will be multiplied by the fac-tor f/400 for corresponding points on the curves.As an example of the effects of a frequency change,

assume that the operating frequency is 10,000 cyclesinstead of 400. From (10) the gain at all points of thecurve, including the maximum gain will be reducedto 38 per cent of the gain at 400 cycles. At the point ofmaximum gain, the input light will be 1.1 foot candlesand the output light 2 7.5 foot lamberts. Similar consider-ations show that operation below 400 cps will increasethe gain and lower the levels of input and output lightat the point of maximum gain. At the high audio fre-quencies, considerable heat is generated in the photo-conductor, for example, at 10,000 cps the power dissipa-tion is about 1 watt per cm2.

THE USE OF FEEDBACK

The ridged photoconductor panel (Fig. 5) and thegrooved photoconductor panel (Fig. 6) amplifiers havebeen designed to operate without feedback. However,a fraction of the output light may be permitted to excitethe photoconductor, for example, by eliminating theopaque layer and making the current-diffusing layerthin. In determining the gain with feedback, the factthat the feedback light from the phosphor is of a differ-ent spectral distribution than the input light must beconsidered (factor k) and also the fact that the photo-conductive layer may not be illuminated from the phos-phor side with the same efficiency as from the inputside when, for example, the photoconductive layer isgrooved (factor B). With these considerations, the gainwith feedfack, Gf, is given by:

GGf -

1 - kBG

where G is the gain without feedback, k is the ratio ofphotoconductor responses to output and input lightrespectively due to spectral differences, and B is theeffective fraction of the output light capable of excitingthe photoconductor. Since the gain, G, of the amplifiervaries with the total input (or output) light (11) isvalid only if the output is maintained constant.The effect of limited amounts of feedback is thus to

increase the gain as well as to make the gain curve morepeaked. If the factor kBG is made greater than unity,a picture element, once excited by an input signal, willremain on regeneratively. Two of the primary problemswith such storage devices are: (1) providing efficientfeedback illumination of the photoconductor, and (2)preventing adjacent elements from exciting each other.

1 896 December

(11)

Loebner: Opto-Electronic Devices and Networks

OPERATION WITH INFRARED AND X-RAYINPUT PICTURES

Although much of the work on the light amplifier asdescribed above has been concerned with its operationusing input pictures of visible light, some tests havebeen made using both infrared and X-ray input images.Inspection of the spectral sensitivity curve (Fig. 14)for the photoconductor shows that in the near infra-red (700-900 millimicrons) the powder has sensitivitiescomparable with those for wave-lengths in the visibleregion. For wavelengths further in the infrared theamplifier requires new photoconductive materials.

Tests with X rays indicate that some amplificationcan be achieved with direct X-ray excitation. It isalso possible to operate the amplifier with a fluoroscopescreen close to the input side so that the visible imageformed on the fluoroscope screen will be intensified bythe amplifier.

FUTURE IMPROVEMENTS

Further development of electroluminescent phos-phors will directly benefit the light amplifier. The sensi-tivity of the photoconductive powder is considerablylower when operated with ac voltages as compared to

dc. Since this effect is not inherent in photoconductorsbut a property of the particular powder, substantialincreases in light gain can be expected with improvedphotoconductors. In addition, the nonlinearity of thephotoconductive powder further limits the effectivenessof the present amplifier. Increase of the photo-conductorbreakdown voltage should serve to produce substantialincreases in gain because of the rapidly increasing lightoutput of the phosphor with voltage. The response timeof the amplifier is also subject to improvement with newphotoconductive materials.

ACKNOWLEDGMENT

We wish to thank Dr. V. K. Zworykin for his activesupport of the work and to acknowledge the interestshown by E. W. Herold in the project. Discussions withD. W. Epstein, E. G. Ramberg, and A. Rose during thecourse of the work are appreciated. The project couldnot have succeeded without the contributions of S. M.Thomsen in developing the photoconductive powderand S. Larach in developing the electroluminescentphosphor. The assistance of P. J. Messineo is also ac-knowledged in the fabrication of the electroluminesceintlayers.

Opto-Electronic Devices and Networks*E. E. LOEBNERt, SENIOR MEMBER, IRE

Summary-The transducing properties of electroluminescent andphoto-responsive cells are described. The light amplifying and spec-

trum-converting characteristics of a circuit consisting of an electricpower supply and a series combination of the two types of cells whoseimpedances have been matched are discussed. The construction andoperation of an opto-electronic bistable device-the "optron"-em-ploying positive radiation feedback, is reported. The optron is both a

storage cell and a switch with dual (optical and electrical) signal in-put and output. Numerous logic networks, composed of electricseries and parallel combinations of electroluminescent and photo-conductive cells with selected optical couplings have been designed,constructed, and operated.

INTRODUCTION

HE DISCOVERIES of electroluminescent phe-nomena and recent remarkable advancements inthe manufacture of electroluminescent materials

and cells [1], together with the progress currently beingmade in the photoconductive branch of solid statescience [2-5], have opened up practical possibilities for

* Original inanuscript received by the IRE, October 5, 1955.t David Sarnoff Research Center, RCA Laboratories Division,

Princeton, N. J.

a new class of solid state devices. The solid state ana-logs of cathode ray imaging and pick-up tubes, andlight amplifying [6, 7] as well as color and radiationconverting [8] screens are only some of the simpler em-bodiments of possible opto-electronic (optronic) devicesand systems.

Proper matching of electrical and optical character-istics of electroluminescent and photoconductive cellsand their incorporation into circuits with electricaland optical branches can give rise to active optronicdevices, such as radiation amplifiers, bistable elementsfor switching and storage, optronic flip-flops, counters,shift-registers, etc. A remarkable similarity to mag-netic amplifier circuits is exhibited by these systems,especially when their operation is based on ac electro-luminescent and photoconductive phenomena. Theoptical branches of the systems enable a complete isola-tion of various electrical subsystems, as well as a si-nmultaneous iniput and output to or from a very greatniumber of elements. The most attractive features seemto be the possible microminiaturization (which, in thepresent state of the art, is the ultimate in miniaturiza-

18971955

Documents

An Electroluminescent Light-Amplifying Picture Panel