9
JOURNAL OF THE OPTICAL SOCIETY OF AMERICA Color and Efficiency of Fluorescent and Fluorescent-Mercury Lamps HENRY F. IVEY Westinghouse Research Laboratories, Pittsburgh, Pennsylvania 15235 (Received 10 December 1971) Calculations are presented for the range of color and luminous efficiency to be expected when typical phosphors are employed with either low-pressure or high-pressure mercury-vapor discharges. In the case of binary phosphor blends corresponding to colors lying on the planckian locus, maximum efficiency is always obtained when the wavelengths of peak emission for the two components lie near 445 and 580-590 nm. INDEX HEADINGS: Lamps; Color; Fluorescence. In a previous publication 1 (hereafter referred to as I), the color and luminous efficiency of single phosphors and of binary blends of phosphors were discussed. In that case, all of the emission was considered to come from the luminescent material itself, as in the case of electroluminescent lamps, cathode-ray-tube screens, and fluoroscopic screens. However, another very important commercial application of phosphors is in connection with electric discharges in mercury vapor. In this case, ultraviolet radiant energy generated in the discharge is used to excite a phosphor and the useful emission con- sists of the phosphor output plus the visible light gene- rated in the discharge. Two different cases are of practical interest. The usual fluorescent (F) lamp consists of a phosphor or blend of phosphors surrounding a low-pressure Hg discharge 2 - 4 (vapor pressure about 6X 10-3 torr, corresponding to a wall temperature of about 40TC). In this case, the efficiency of generation of visible light in the discharge is relatively low and most of the mercury radiant energy is in the resonance lines at 1850 and 2537 A. In such lamps, most of the light therefore comes from the phos- phor and the visible light from the discharge is a second- ary factor in determining the color and efficiency of the lamp. Phosphors are also employed in high-pressure mercury-vapor (HPMV) discharges. 4 - 6 In this case, the vapor pressure is about 5 atm, the luminous efficiency of the discharge is fairly high, and most of the ultraviolet radiation is at wavelengths longer than 2537 A, particularly the group of lines near 3650 A. In such fluorescent-mercury lamps the situation is re- versed, as compared to low-pressure F lamps. The dis- charge itself is now the predominant factor in determin- ing the over-all color and luminous efficiency and the phosphor is a secondary, although important, factor. In this paper, the performance to be expected of phosphors in these two different types of lamps is dis- cussed along the same lines used in I for the simpler case in which no discharge is involved. COMBINATION OF LIGHT FROM DISCHARGE AND PHOSPHOR In the ideal case, the luminous output of a phosphor at a wavelength X, excited by an amount r of ultra- violet power at wavelength Xiis L(i/Xe) = 680g(Xe)(Xi/Xe)q(xi)r, (1) where g is the CIE visibility function for the normal human eye and q is the quantum efficiencyof the phos- phor (the ratio of the number of quanta emitted to the number absorbed). The factor (Xi/Xe) enters because of the different energy content of the absorbed and emitted quanta. Here it is assumed that all of the power r is actually absorbed in the phosphor and also that all of the emitted light is able to escape and be utilized. In practice, losses occur; these will be discussed further below. In practice, the light emitted by the phosphor has a spectral distribution that may be represented by E(X) and the exciting power another spectral distribution r(X). The total luminous output is then L=680 V(X)E(X) dXf Xr(X)q(X) dX/ IXE(X) dX]. (2) Here the first term is the normal expression for the luminous output corresponding to an energy distribu- tion E(X), whereas the second factor (ratio of two integrals) relates the number of incident and emitted quanta. It is convenient to introduce -q,the luminosity factor for the emitted light, X = f(X)E(X) dX JE(X) dX; (3) X,related to the average wavelength of the emitted light [see I, Eq. (5)], A= JXE(X) dX/400 E(X) dX; (4) A, the corresponding quantity for the exciting radiant power, A-=Xr(X) dX/400 r(X) dX; (5) 814 VOLUME 62, NUMBER 6 JUNE 1972

Color and Efficiency of Fluorescent and Fluorescent-Mercury Lamps

  • Upload
    henry-f

  • View
    215

  • Download
    2

Embed Size (px)

Citation preview

Page 1: Color and Efficiency of Fluorescent and Fluorescent-Mercury Lamps

JOURNAL OF THE OPTICAL SOCIETY OF AMERICA

Color and Efficiency of Fluorescent and Fluorescent-Mercury LampsHENRY F. IVEY

Westinghouse Research Laboratories, Pittsburgh, Pennsylvania 15235(Received 10 December 1971)

Calculations are presented for the range of color and luminous efficiency to be expected when typicalphosphors are employed with either low-pressure or high-pressure mercury-vapor discharges. In the caseof binary phosphor blends corresponding to colors lying on the planckian locus, maximum efficiency isalways obtained when the wavelengths of peak emission for the two components lie near 445 and 580-590 nm.INDEX HEADINGS: Lamps; Color; Fluorescence.

In a previous publication1 (hereafter referred to as I),the color and luminous efficiency of single phosphorsand of binary blends of phosphors were discussed. Inthat case, all of the emission was considered to comefrom the luminescent material itself, as in the case ofelectroluminescent lamps, cathode-ray-tube screens, andfluoroscopic screens. However, another very importantcommercial application of phosphors is in connectionwith electric discharges in mercury vapor. In this case,ultraviolet radiant energy generated in the discharge isused to excite a phosphor and the useful emission con-sists of the phosphor output plus the visible light gene-rated in the discharge.

Two different cases are of practical interest. The usualfluorescent (F) lamp consists of a phosphor or blend ofphosphors surrounding a low-pressure Hg discharge2 - 4

(vapor pressure about 6X 10-3 torr, corresponding to awall temperature of about 40TC). In this case, theefficiency of generation of visible light in the dischargeis relatively low and most of the mercury radiant energyis in the resonance lines at 1850 and 2537 A. In suchlamps, most of the light therefore comes from the phos-phor and the visible light from the discharge is a second-ary factor in determining the color and efficiency of thelamp.

Phosphors are also employed in high-pressuremercury-vapor (HPMV) discharges.4 -6 In this case,the vapor pressure is about 5 atm, the luminousefficiency of the discharge is fairly high, and most of theultraviolet radiation is at wavelengths longer than2537 A, particularly the group of lines near 3650 A. Insuch fluorescent-mercury lamps the situation is re-versed, as compared to low-pressure F lamps. The dis-charge itself is now the predominant factor in determin-ing the over-all color and luminous efficiency and thephosphor is a secondary, although important, factor.

In this paper, the performance to be expected ofphosphors in these two different types of lamps is dis-cussed along the same lines used in I for the simplercase in which no discharge is involved.

COMBINATION OF LIGHT FROMDISCHARGE AND PHOSPHOR

In the ideal case, the luminous output of a phosphorat a wavelength X, excited by an amount r of ultra-

violet power at wavelength Xi is

L(i/Xe) = 680g(Xe)(Xi/Xe)q(xi)r, (1)

where g is the CIE visibility function for the normalhuman eye and q is the quantum efficiency of the phos-phor (the ratio of the number of quanta emitted to thenumber absorbed). The factor (Xi/Xe) enters because ofthe different energy content of the absorbed and emittedquanta. Here it is assumed that all of the power r isactually absorbed in the phosphor and also that all ofthe emitted light is able to escape and be utilized. Inpractice, losses occur; these will be discussed furtherbelow.

In practice, the light emitted by the phosphor has aspectral distribution that may be represented by E(X)and the exciting power another spectral distributionr(X). The total luminous output is then

L=680 V(X)E(X) dXf Xr(X)q(X) dX/

IXE(X) dX]. (2)

Here the first term is the normal expression for theluminous output corresponding to an energy distribu-tion E(X), whereas the second factor (ratio of twointegrals) relates the number of incident and emittedquanta. It is convenient to introduce -q, the luminosityfactor for the emitted light,

X = f(X)E(X) dX JE(X) dX; (3)

X, related to the average wavelength of the emitted light[see I, Eq. (5)],

A= JXE(X) dX/400 E(X) dX; (4)

A, the corresponding quantity for the exciting radiantpower,

A-=Xr(X) dX/400 r(X) dX; (5)

814

VOLUME 62, NUMBER 6 JUNE 1972

Page 2: Color and Efficiency of Fluorescent and Fluorescent-Mercury Lamps

COLOR AND EFFICIENCY OF FLUORESCENT LAMPS

R, the total exciting radiant power,

R=fr(X) dX; (6)

and q, an effective quantum efficiency for the particularphosphor and distribution of exciting radiation con-cerned,

= fq(XA)r(X) dX/ fxr(X) dA. (7)

Equation (2) then assumes a form analogous to Eq. (1),

L = 680O7(A/X)qR. (8)

If Eu is the efficiency of generation of ultravioletradiant energy by the discharge, then R= euW, where Wis the energy input. Finally, if e,, is the luminous effi-ciency of the discharge, then the total luminousefficiency of the lamp (discharge plus phosphor), stillneglecting losses, is

e= e,+680 1 (A/)qeu. (9)

Several possible causes of loss in output must nowbe considered ." 7-9 First, all of the ultraviolet radiantenergy may not be absorbed by the phosphor. If thecoating is too thin, some will be transmitted and ab-sorbed by the glass on which the phosphor is coated.Furthermore, some will be reflected at the phosphorsurface. In most lamps, the phosphor forms an almostcompletely enclosed system so the probability of ulti-mate useful absorption is increased by interreflection.However, some loss of ultraviolet does occur at the endsof the lamp and, particularly for HPMV lamps, byabsorption at internal supports and electrodes. Further-more, particularly in F lamps, the reflected ultravioletmust pass through the discharge before it can reach thephosphor-coated wall again so that the transparency ofthe discharge for its own ultraviolet output is impor-tant.' 0 Since the phosphor is normally applied as apowder, the particle size has an effect on the reflectanceand transmittance of the coating and hence on theultraviolet losses.10 "'1 For the present purposes, acoefficient a,, which is the utilization factor for theultraviolet radiation, may be introduced.

Not all of the light emitted by the phosphor may suc-ceed in leaving the lamp. Some of the light is initiallydirected inward and is subject to interreflection beforefinally escaping. The reflectance and transmittance ofthe phosphor coating for visible light as well as lossesat internal parts and at the lamp ends are thereforeimportant and may be allowed for by a utilizationfactor a2 for the phosphor emission. If the phosphorlayer is too thick, egress of the light will be impeded,whereas if it is too thin, the ultraviolet absorption willbe low. There is thus an optimum coating thickness8

that depends on the particle size and other propertiesof the phosphor."

The visible light emitted by the discharge must alsousually pass through the phosphor coating. Again,interreflections will occur and the transparency of thedischarge for its own visible output is important."Another utilization factor a, may be introduced for thevisible light from the discharge. In the case of HPMVlamps, in which the luminous efficiency of the dischargeis high, the optimum coating thickness is a compromisebetween high ultraviolet absorption and low visible-light absorption; such layers are usually thinner thanin F lamps." Reflection of light from the discharge canalso play a role in F lamps. Ouweltjes'4 states, "We havesupposed that the mercury spectrum is always the sameas long as no absorption (by the phosphor) is involved.This is known to be not quite correct. By a mechanismnot yet understood the phosphors, even when white, dohave a certain effect on the relative intensities of themercury lines." It is suggested here that this mechanismis reflection at the phosphor coating, which is influencedby particle size and other factors, followed by differen-tial absorption"2 and re-emission in the discharge. Wehave observed that not only does the phosphor coatingaffect the relative intensities of the mercury lines, butthat the total visible output from the discharge may behigher in the presence of a coating than in its absence(effect of coating on e,) rather than lower, as would beexpected from simple absorption. This is presumably dueto a change of electron temperature and voltage gra-dient in the discharge as a result of re-absorption ofradiant energy (ultraviolet and visible) reflected at thephosphor."

Finally, absorption by the phosphor of visible lightfrom the discharge may result in visible output at alonger wavelength for some phosphors. This is inten-tionally employed in some double-coated Europeanfluorescent lamps that are designed for superior colorrendition by absorption of the excessive violet mercuryemission and reradiation in the red region.'4"6 It alsooccurs accidentally in HPMV lamps using the magne-sium fluorogermanate phosphor.'7 This is effect meansa decrease of e, and an increase of eu in Eq. (9). Thiscomplication will not be considered further here.

With the utilization coefficients introduced in theabove discussion, Eq. (9) becomes

6 = asi,+680n(A,/)qa1a2E, = a3E,+ Ep. (10)

Before calculations can be made, the various factorsmust be evaluated individually for typical low- andhigh-pressure mercury discharges.

Phosphors in Low-Pressure F Lamps

Although higher-wattage F lamps are now commer-cially available,'8 the most common size is still the40-W lamp (diameter, 38 mm; length, 1.2 m; current,420 mA). Data on this and similar discharges have been

8l5June 1972

Page 3: Color and Efficiency of Fluorescent and Fluorescent-Mercury Lamps

HENRY F. IVEY

given by a number of workers.2' 13 Although there

are some differences between individual reports, it isbelieved that representative values (measured andcalculated) are input: 40 W; output, 1850 A: 3 W;output, 2537 A: 20 W; output, total visible: 1.11 W;e'= 6 lm/W, e =0.575, A=0.612, xd=0. 2 1 6 , yd=0. 2 1 6 ,where Xd and yd are the trichromatic coefficients of thedischarge. Both the color and efficiency of the dischargeare functions of the ambient temperature (effect onvapor pressure)2' and the current density.2 0 Thus, inhighly loaded lamps'8 the total output is greater, but theefficiency is lower than for the standard 40-W lampconsidered here.

A point of obvious importance is the value to beassigned to the effective quantum efficiency q. In theideal case, this should be unity,24-27 but the phosphorsuffers some damage during lamp manufacture. Further-more, there is a certain amount of deterioration (3-4%)in output of commercial lamps early in life; for thisreason, lamps are conventionally rated after 100 h ofoperation and the decrease of output during thisperiod appears as a reduced effective quantum effi-ciency. Experimental determination of quantum efficien-ciencies is a very complicated procedure and the mea-surements that are available do not always agree.2 8

In particular, the question of the relative efficiency ofthe radiant energy at 1850 and 2537 A for phosphorexcitation is open to debate. 26'2 7 For the present calcu-lations a value of 0.82 has been selected for the com-bined factor ala2 q and a3= 1. Equation (10) thenbecomes

e=6+196(X7/A)=6+196n'. (11)

Similar empirical determinations made in 1939 and1953 by Thayer and Barnes' and by Jerome," respec-tively, led to 5+ 1 6 8 n' (after the maximum efficiencyof light generation was changed from 621 to 680 lm/W)and 5+ 1 7 8n'. The present calculation is therefore moreoptimistic. However, it is in accord with the gradualimprovement that has accompanied continued develop-ment work on F lamps. Thus there are now commerciallyavailable warm-white 40-W lamps rated at 3250 lm;with q'=0.382 as calculated for this color, the valuepredicted by Eq. (11) is 3240 Im. Although absolutevalues are of interest, another important question isthe relative performance of various phosphors; for thispurpose absolute accuracy of the constants in Eq. (11)is not essential.

Equation (11) applies only to the 1.2-m 40-W lampconsidered here. The data used include the effect of thevoltage drop at the electrodes (18 out of 103 V or 7 outof 40 W), which is a loss as far as production of lightis concerned. Thus, for 2.4-m lamps, the efficiencyshould be higher by a ratio of 73/(2)(33) or 10.6% andfor lamps of infinite length by 40/33 or 21.2%. Shorterlamps will, of course, have lower efficiencies.' 9

The color of a lamp (x',y') containing a phosphorwhose trichromatic coefficients are (xy) is also of inter-

est. The ordinary laws of color mixture apply, giving

x=Xd(a3fE vlYd) +X(6,1Y)

a3E/1Yd+-EY

I a3et,+,E,My =_

a3e,E/Yd+ E/Y'

(12a)

(12b)

where e, and e, are the luminous contributions of thedischarge and of the phosphor, respectively [see Eq.(10)]. With the values given earlier and Eq. (11), thisbecomes

6+196y'x/y

27.8+196y'/y

6+196v'Y2 /

27.8+196n'/y

(13a)

(13b)

Equations (11) and (13) give the predicted perfor-mance (color and efficiency) of a given phosphor in a40-W low-pressure F lamp.

Phosphors in HPMV Lamps

HPMV lamps are made in a variety of sizes but themost common is the 400-W lamp (type H33-1; arclength 7.0 cm; arc-tube diameter 1.8 cm; current, 3.2 A).In such lamps, the fused-silica arc tube, which operatesat a quite-high temperature (650-700 0C), is surroundedby a glass envelope that also operates at an elevatedtemperature (350-400 0 C); the thermal and otherproperties of the lamps are strongly influenced by thepressure of the gas in this outer envelope.2 The pres-ence of this outer envelope has two other effects of im-portance, one of practical consequence and the otheraffecting only the present calculation. The glass useddoes not transmit all of the ultraviolet flux falling on it;that at shorter wavelengths is absorbed. Hence,measurements of the ultraviolet output of such lampsas ordinarily made do not give the total amount orspectral distribution of radiant energy available forextinction of a phosphor coated on the inner wall ofthe second envelope. The required measurements mustbe made on bare arc tubes but care must also be exer-cised that the operating temperature in this case istypical of that in jacketed lamps. Second, to operateefficiently in HPMV lamps, any phosphor must absorba wide range of ultraviolet wavelengths and also havegood quantum efficiency at the high operating tempera-ture. Because of these limitations, only five practicalphosphors for such applications are known today. 7

30-33

Measurements on discharges of the type concernedhere have been made by various workers,4 ' 6 34-37 butin general no single worker has measured all of thequantities of interest. Representative figures arebelieved to be input: 400 W; output, ultraviolet: 75.4 W;output, total visible: 59.8 W; e = 53.5 lm/W,

816 Vol. 62

Page 4: Color and Efficiency of Fluorescent and Fluorescent-Mercury Lamps

FLUORESCENT LAMPS 817

eu =0.1885, A=0.766, Xd=0.3 10 , yd=0.3 9 0. The visiblespectrum is complicated by the presence of a smallcontinuum. The color of the discharge given here is ingood agreement with that given by Jerome. 8 The lu-minous efficiency is based on the present commercialrating for clear lamps (21 500 lm). The value for thevisible output is based on the data of Weitz,37 multipliedby a factor corresponding to the increase of ratings(largely due to improved electrodes) since his measure-ments. Only data measured in this laboratory by C. H.Haake were available for the total ultraviolet output,including short wavelengths not transmitted by theouter envelope. His results give A =0.7663, which isquite close to the value of 0.7655 calculated from thedata of R6ssler6'3 4 for a slightly different discharge.Finally, the total ultraviolet output was calculated onthe basis of the visible output (discussed above) andHaake's result that 55.8% of the radiated energy is inthe ultraviolet region.

Substitution of these values in Eq. (10) gives

e=53.5a3+98.2aja 2 q('q/X)* (14)

The determination of the remaining parameters is some-what uncertain. In view of the internal structure ofHPMV lamps, it is to be expected that a2 and a3 shouldbe smaller than in low-pressure lamps. Furthermore,because the coating is normally thinner than in F lamps(so as not to make a3 too small), a, is also expected tobe smaller. Here a3 =0.859 and ala2q=0.692 werechosen. The result for the luminous efficiency is then

e=4 6.0+68.07i', (15)

and for the color,

x= 36.53 +68.'x/y (16a)117.8+68.0-q'/y

46.0+68.0X7'ye = (16b)

117.8+68.077'/y

For the commercial phosphor, tin-activated strontium/zinc orthophosphate3 2 (' '=0.206), used in high-outputwhite lamps, the calculated and experimental3 6 resultsare

Calc.: x=0.363, y=0.399, e=59.98 lm/W;

Obs.: x=0.350, y=0. 4 00, e=60.0 lm/W.

The difference of color is no larger than the differencebetween results given by various workers for uncoatedlamps,36' 37 whereas the agreement of efficiency is exactbecause of the way in which values for a3 and ala2qwere selected. For these reasons and because the se-lected parameters seem reasonable, we believe thatEqs. (15) and (6) give a realistic estimate of the per-formance to be expected of a phosphor in a 400-WHPMV lamp.

Comparison of Eq. (15) with the result for an un-coated HPMV lamp (53.5 lm/W) shows that, owing toreflection by the phosphor and subsequent absorptionof light from the discharge, no gain of luminous efficiencycan be achieved unless 7'>0. 10. Otherwise, the outputof the lamp with phosphor will actually be less thanthat of a clear lamp, although the color rendition may beimproved as in the case of standard white lamps.Comparison of Eqs. (11) and (15) also shows that agiven phosphor emission spectrum will give higheroutput in a F lamp than in a HPMV lamp if -'>0.312.Of course, in these two cases the phosphor-excitationspectrum and the temperature dependence of quantumefficiency must be quite different, so that in practice aphosphor cannot normally be used interchangeably inthe two types of lamp.

CALCULATED RESULTS

In I, the emission spectra of many common phosphors(exclusive of those activated with rare-earth ions ortransition-metal ions, in which the responsible electrontransitions occur in inner shielded levels) were shownto be represented fairly accurately as a single bandaccording to the expression

E(X) = exp -[ (17)

where Xo is the peak wavelength and Av is the emission-bandwidth parameter. Such curves have been calledasymmetrical gaussian distributions. Values of Xo andAvP for several real phosphors of interest were given inI, Table I, and some typical spectra, showing reason-able fit to curves of this type, were shown in I, Fig. 2.

Calculations have been made for the expected per-formance in F and fluorescent-mercury (HPMV) lamps,on the basis of the preceding sections, of the same seriesof hypothetical phosphors, given by Eq. (17), which westudied in I (Table II). The calculations were per-formed on an IBM 1401 computer programmed byD. W. Morgan.

Calculations for F Lamps

The results of the chromaticity and luminous-effi-ciency calculations for single phosphors for this case[Eqs. (11) and (13)] are given in Figs. 1 and 2, re-spectively. These figures may be compared with Figs.3 and 4 of I for the case in which all of the emissioncomes from the phosphor. In F lamps, the range ofavailable colors is more limited; although the visiblelight from the discharge has only a relatively smalleffect on the luminous efficiency, it has an appreciableeffect on the color.3 8

Figure 1 shows one feature that is lacking in the corre-sponding case of cathodoluminescent or electrolumines-cent sources (I). This is that light whose color corre-sponds to a point lying on the planckian locus may beobtained with a single phosphor. In this case, a yellow

June 1972 COLOR AND EFFICIENCY OF

Page 5: Color and Efficiency of Fluorescent and Fluorescent-Mercury Lamps

HENRY F. IVEY

0.3 0.4VALUES OF x

FIG. 1. Chromaticity plot for phosphors with asymmetricalgaussian spectral distributions in F lamps. The solid lines corre-spond to constant bandwidth [Av in Eq. (17)] and the dashedlines to constant peak wavelength (Xo). D represents the color ofthe low-pressure mercury discharge and E is equal-energy white.Temperatures on superimposed planckian locus indicated inhundreds of kelvins.

phosphor is combined with the predominantly bluelight of the discharge. In practice, such lamps would, ofcourse, have very poor color rendition. The character-istics of these single-component white lamps are shownin Fig. 3. The luminous efficiency falls quite rapidly asthe emission bandwidth is increased.

Calculations were also made for binary blends. Inthis case the luminous efficiency of the blend is

eo = 6+ Y 1 + y2,

t 160

El)z 140

~ 100

ul

2

o 80

a6-J

2

`100

-O

20LL 6w r

400 500 600PEAK WAVELENGTH, (nm)

700

(18)

FIG. 2. Luminous efficiency for phosphors with asymmetricalgaussian spectral distributions in F lamps (chromaticity shownin Fig. 1).

10000 6500 4500 3500 3000 2500 2000 1800 I t K I

z

°, 40 / TUNGSTEN A= 0.35PMr~10- (

2 -1

-PAN0A

WC)20 PRAcTuCLftwo UNoS TEN cn ef LestMqan s

I 2 324 5

104/T (K-'

FIG. 3. Maximum luminous efficiency as a function of colortemperature for F lamps with blends of phosphors with asym-metrical gaussian spectral distributions. The efficiency of thermal(blackbody) radiators is included for comparison.

where Y, and Y2 are the luminous contributions of thetwo phosphor components. These can be calculatedfrom the set of equations [see Eqs. (13)]

6 +x1(Yl/y 1 ) +x 2 (Y2/y 2)Xo =

27.8+ Yi/yl+ Y2/y2

6+ Y1+ Y2yo'=-27.8+ Yi/yi+ Y2/y2

196 = yll?7'1+ Y2/77'2.

(19a)

(19b)

(19c)

0.3 0.4VALUES OF x

FIG. 4. Contours of maximum luminous efficiency (Im/W) onCIE chromaticity diagram for F lamps with phosphors withasymmetrical gaussian spectral distributions [Eq. (17)] andAv= O.1 om-'.

- I

~0I'I N0v.2

I \

I I

I

Vol. 62

Page 6: Color and Efficiency of Fluorescent and Fluorescent-Mercury Lamps

COLOR AND EFFICIENCY OF FLUORESCENT LAMPS

VALUES OF x VALUES OF x

FIG. 5. Contours of maximum luminous efficiency (lm/W) onCIE chromaticity diagram for F lamps with phosphors withasymmetrical gaussian spectral distributions [Eq. (17)] andAP 0.2 Am-'.

The last equation expresses the fact that the totalnumber of quanta emitted from the phosphor blend isconstant (under the assumption made here that q isthe same for all components). For each point in theavailable region of the color diagram some particularcombination of X10 and X2 0 (the peak wavelengths of thetwo components) will give the highest efficiency.Figures 4-6 present contour plots of the maximumefficiency39 obtained in this way; these curves may becompared with I, Figs. 5-8.

Blends whose trichromatic coefficients fall on theplanckian locus are of special interest. The results forthese cases are summarized in Table I. Here Y5 , E2 ,and Q2 represent the fractional contribution of thesecond (long wavelength) component of the blend interms of luminous output, energy, and quanta, re-spectively. The performance of these optimum planck-ian blends is shown in Figs. 3 and 7. In the former case,the data for thermal (blackbody) radiators40 are alsoincluded. The superior efficiency of F lamps overpractical incandescent sources is evident from thisfigure.

Table I shows that for these optimum planck-ian blends, except for very-wide emission bands(AP = 0.3 um'l), the peak wavelength of the short-wave-length component is always near 445 nm. Table IIfurther shows that if the color temperature is selected togive the maximum output in each case, the peak wave-length of the other component is always near 480 nm.This is consistent with the demonstration in I that thereare "preferred wavelengths for high output in blends" inthese regions. The presence of the visible radiation

FiG. 6. Contours of maximum luminous efficiency (Im/W) onCIE chromaticity diagram for F lamps with phosphors withasymmetrical gaussian spectral distributions [Eq. (17)] andAv=0.3 pumV.

TABLE I. Results for optimum planckian blends of asymmetricalgaussian emitters in F lamps.

T AV X1o X20 co(K) (jzm') (nm) (nm) Y2 E 2 92 (lm/W)

0o 0 445 561 0.9634 0.442 0.499 78.1Do 0.1 450 558 0.9312 0.424 0.478 71.3XC 0.2 445 556 0.8477 0.396 0.457 57.7

0 0.3 435 563 0.7179 0.374 0.441 47.9

10 000 0 445 566 0.9764 0.558 0.617 91.210 000 0.1 447.5 563 0.9610 0.552 0.607 82.910 000 0.2 445 563 0.9059 0.538 0.601 67.410 000 0.3 435 569 0.8347 0.549 0.620 54.7

6500 0 445 569 0.9829 0.641 0.696 99.86500 0.1 445 566.5 0.9750 0.637 0.692 90.56500 0.2 442.5 567 0.9413 0.639 0.698 72.76500 0.3 432.5 572.5 0.8979 0.665 0.742 57.8

4500 0 445 573.3 0.9890 0.743 0.788 107.64500 0.1 445 571.7 0.9847 0.750 0.795 98.04500 0.2 442.5 575.5 0.9650 0.762 0.809 76.54500 0.3 425 585 0.9584 0.806 0.857 57.2

3500 0 445 576.7 0.9930 0.824 0.859 111.93500 0.1 445 576.7 0.9910 0.843 0.875 101.53500 0.2 442.5 584.2 0.9808 0.865 0.896 76.23500 0.3 415 598.5 0.9879 0.917 0.941 53.9

3000 0 445 579.5 0.9956 0.885 0.909 113.73000 0.1 445 580.2 0.9939 0.892 0.915 102.13000 0.2 440 591 0.9906 0.926 0.945 73.3

2500 0 445 583 0.9980 0.947 0.959 114.32500 0.1 445 586 0.9995 0.991 0.994 101.6

2200 0 445 585.9 0.9997 0.991 0.993 113.1

June 1972 819

Page 7: Color and Efficiency of Fluorescent and Fluorescent-Mercury Lamps

HENRY F. IVEY

140

U-

120

(-4

U-UW

EMISSION BANDWIDTH, Av (amLn)

F;IG. 7. Maximum luminous efficiency of F lamps with planckianblends of phosphors with asymmetrical gaussian spectral dis-tributions of various emission bandwidths. Color temperature inkelvins.

from the low-pressure mercury discharge does notaffect these preferred wavelengths.

The calcium halophosphate phosphor,22 ' 4' currentlyused in almost all commercial F lamps, is doubly acti-vated and has two emission bands, one (due to Sb)characterized by X0=478 nm and Av=0.368 gnr'and the other (due to Mn) by X0= 588 nm andA~v=0.178 tim-. The favorable location and relativelynarrow emission band of the yellow Mn emission ac-counts for the improvement in performance of thisphosphor over the blends customarily used before itsintroduction. However, it appears that even higherefficiency could be achieved if the blue Sb emission werereplaced by a band with peak nearer the preferred valueof 445 nm and with a narrower width, with the additionof a small amount of a third component to give thedesired color.

Calculations for Fluorescent-Mercury Lamps

The results of the calculations for single phosphors inHPMV lamps [Eqs. (15) and (16)] are shown in Figs.

TABLE II. Performance of optimum white F lamps with blendsof asymmetric gaussian emitters.

PeakBandwidth wavelengths Color Luminous(Av1= Av2) X10 X20 temperature efficiency

(Jm'1) (nm) (nm) T (K) E(lm/W)

0 445 582 2600 114.50.1 445 581 2900 102.50.2 442.5 579 4000 77.00.3 430 579 5300 58.0

x

FIG. 8. Chromaticity plot for phosphors with asymmetricalgaussian spectral distributions in fluorescent-mercury (HPMV)lamps. The solid lines correspond to constant bandwidth [Av inEq. (17)] and the dashed lines to constant peak wavelength (Xo).D represents the color of the high-pressure discharge and E isequal energy white. Temperatures on superimposed planckianlocus indicated in hundreds of kelvins.

8 and 9; the chromaticities are plotted in Fig. 8 and theluminous efficiencies in Fig. 9. The curves of Figs. 2and 9 and Fig. 4 of I are similar in shape but theirabsolute values differ. The highest efficiency for aphosphor with a relatively narrow emission band(zP=0.l um-1) is 90.3 lm/W, an increase of 69% overthat (53.5 lm/W) for a lamp without phosphor. Aphosphor with the same spectral distribution in a low-pressure lamp would have an efficiency of 134 lm/W(Fig. 2).

As is to be expected from the large amount of visiblelight from the discharge in HPMV lamps, the phosphorhas a comparatively small influence on the color andthe available range is fairly limited (compare Figs. 1and 8). For phosphors with even moderate emissionbandwidth (Av=0.2 tzm'l), the planckian locus liesalmost entirely outside the attainable region. As in thecase of low-pressure lamps, white emission can in somecases be achieved with a single phosphor (see TableIII). This is true, however, only for fairly narrow emis-sion bands.

In the case of binary blends, formulas similar toEqs. (18) and (19) apply, but with the appropriatenumerical constants [see Eqs. (15) and (16)]. Theresults of such calculations are plotted in Fig. 10, whichmay be compared with Fig. 3 for low-pressure lamps.For an emission bandwidth as small as can be achievedin practice (except for a phosphor activated by a rare-earth ion or some other ion, like Mn4 +, with shieldedtransitions) or Av=0.1 um'l, the maximum efficiencythat can be expected for white emission is about

Vol. 62-

Page 8: Color and Efficiency of Fluorescent and Fluorescent-Mercury Lamps

COLOR AND EFFICIENCY OF FLUORESCENT LAMPS

500 600PEAK WAVELENGTH (nm)

FIG. 9. Luminous efficiency for phosphors with asymmetricalgaussian spectral distributions in fluorescent-mercury (HPMV)lamps (chromaticity shown in Fig. 8).

72 lm/W; the corresponding figure for a low-pressurelamp is 102 lm/W. For an emission band as wide as

P= 0.2 pm-', the range of color temperatures availableis restricted to 8000-38 000 K and the efficiency showslittle or no improvement over an uncoated lamp. It istherefore necessary to use a yellow phosphor with quitenarrow bandwidth, if fluorescent-mercury lamps withcolor temperatures in the usual range of 2500-6500 Kare to be made.

The highest output for white phosphor blends inHPMV lamps (Av=0, T=3500 K; Pv=0.1 pnm',T=4000-4500 K) is obtained when the two phosphorcomponents have their emission peaks near 445 and 590nm, once again confirming the existence (even in thepresence of large amounts of external radiation) ofthese two preferred wavelengths.

CONCLUSION

Calculations have been made for the color and lu-minous efficiency to be expected if a series of hypo-

TABLE III. Performance of single asymmetrical gaussianphosphors that produce white light in HPMV lamps.

AP X0 T eaum-l) (nm) (K) (lm/W)

0 420 14 000 46.3a0 593 3 200 78.80 645 4 150 52.0a0.1 417 14 000 46.4a0.1 605 3 420 70.00.1 641 4 000 55.2a0.2 430 38 000 48.4

a The same color temperature may also be obtained with a binary phos-phor blend (same value of Ab) at higher efficiency (see Fig. 8).

3 5 NO ' as6'V=Ol

70

I 30-z

20-z

i 1 T( K)

a-

tJ20000 10000 6500 5000 4000 3500 3000

0 ++ t +++ + +t f t t . 80 1 2 3

14T ( K )

FIG. 10. Maximum luminous efficiency as a function of colortemperature for fluorescent-mercury (HPMV) lamps with blendsof phosphors with asymmetrical gaussian spectral distributions.

thetical phosphors, with emission spectra typical ofreal phosphors, were incorporated in 40-W low-pressure fluorescent lamps or in 400-W high-pressurefluorescent-mercury lamps. In the latter case, theavailable range of colors is rather limited because of thepredominance of visible light from the mercury-vapordischarge.

In both cases, white-emitting lamps of maximumluminous efficiency are obtained when a binary blendof phosphors with peak emission near 445 and 580-590nm are employed, confirming the existence of these twopreferred wavelengths in these cases as well as for thesimpler case, discussed in an earlier paper,' in which allof the light comes from the phosphor.

This manuscript was completed (in 1963) while theauthor was employed at the Westinghouse LampDivisions in Bloomfield, N. J. Since that time, Thorn-ton4 2 has extended the considerations to blends of morethan two phosphor components and also to include thecolor-rendering properties of such light sources.

REFERENCES

H. F. Ivey, J. Opt. Soc. Am. 53, 1185 (1963).2 W. E. Forsythe and E. Q. Adams, Fluorescent and Otiler

Gaseous Discharge Lamps (Murray Hill Books, New York, 1948).3 Fluorescent Lamnps and Lighting, edited by W. Elenbaas

(Macmillan, New York, 1959).4 J. Funke and P. J. Oranje, Gas Discharge Lamps (Elsevier,

New York, 1951).5 H. K. Bourne, Discharge Lamnps for Photography and Projec-

tion (Chapman and Hill, London, 1948).6 W. Elenbaas, The High Pressure Mercury Vapor Discharge

(Wiley-Interscience, New York, 1951).7 R. N. Thayer and B. T. Barnes, J. Opt. Soc. Am. 29, 131

(1939).8 R. N. Thayer, Trans. Electrochem. Soc. 87, 413 (1945).9 C. W. Jerome, J. Electrochem. Soc. 100, 586 (1953).'OJ. L. Ouweltjes, Elektrizitatsverwertung 33, 293 (1958).11 K. H. Butler and H. H. Homer, Illum. Engr. 55, 396 (1960);

R. G. Young and E. G. F. Arnott, J. Electrochem. Soc. 112, 982(1965).

June 1972 821

Page 9: Color and Efficiency of Fluorescent and Fluorescent-Mercury Lamps

HENRY F. IVEY

12 C. Kenty, J. Appl. Phys. 21, 1309 (1950); M. Koedam andA. A. Kruithof, Physica 28, 80 (1962).

13 C. H. Haake, J. Electrochem. Soc. 106, 866 (1959); 108, 78(1961).

14 J. L. Ouweltjes, Farbe 9, 207 (1960).16 The writer is indebted to D. A. Larson for discussion on this

topic.16 A. A. Kruithof and J. L. Ouweltjes, Philips Tech. Rev. 18,

249 (1957).17 L. Thorington, J. Opt. Soc. Am. 40, 579 (1950); G. Kemeny

and C. H. Haake, J. Chem. Phys. 33, 783 (1960).18 A. C. Barr and W. J. Karash, Illum. Engr. 50, 5 (1955);

J. 0. Aicher and E. Lemmers, Illum. Engr. 52, 579 (1957); 55,39 (1960); J. F. Waymouth, F. Bitter, and E. Lowry, Illum. Engr.52, 257 (1957); C. J. Bernier, Illum. Engr. 55, 282 (1960); L.E. Vrenken, Ilium. Engr. 57, 683 (1962); L. Thorington and J.Shurgan, Illum. Engr. 57, 127 (1962).

19 E. F. Lowry, W. C. Gungle, and C. W. Jerome, Illum. Engr.49, 545 (1954).

20 S. T. Henderson and M. B. Halstead, Brit. J. Appl. Phys.3, 255 (1952).

21 C. W. Jerome, Illum. Engr. 49, 237 (1954); 51, 205 (1956).22 M. Doherty and W. Harrison, Brit. J. Appl. Phys. Suppl. 4,

11 (1955).23 B. T. Barnes, J. Appl. Phys. 31, 852 (1960).24 It is, of course, energetically possible for one 1850-A quantum

to produce more than one visible quantum (Ref. 25). Experi-mental evidence for this in lamp phosphors has been claimed(Ref. 26), but this was not supported by later work (Ref. 27).

21 D. L. Dexter, Phys. Rev. 108, 630 (1957); 126, 1962 (1962);F. Varsanyi and G. H. Dieke, Phys. Rev. Letters 7, 442 (1961).

26 F. A. Butaeva and V. A. Fabrikant, Izv. Akad. Nauk SSSR,Ser. Fiz. 21, 537 (1957); Proceedings CIE Meeting, Brussels,1959, Paper 59.4 (Bureau Central CIE, 25 rue de PIpiniere, Paris,1960).

27 P. D. Johnson, J. Opt. Soc. Am. 51, 1235 (1961).28 J. Tregellas-Williams, J. Electrochem. Soc. 105, 173 (1958);

A. Bril and W. Hoekstra, Philips Res. Rept. 16, 356 (1961);F. J. Studer, J. Opt. Soc. Am. 52, 605 (1962).

29 J. M. Harris, Illum. Engr. 52, 363 (1957).30 J. L. Ouweltjes, W. Elenbaas, and K. R. Labbert6, Philips

Tech. Rev. 13, 109 (1951); H. A. Klasens, Philips Res. Rept.9, 377 (1954).

31 H. G. Jenkins and A. H. McKeag, Trans. Illum. Engr. Soc.(London) 17, 159 (1952); A. H. McKeag and E. G. Steward,Brit. J. Appl. Phys. Suppl. 4, 26 (1955); A. H. McKeag, J. Elec-trochem. Soc. 105, 78 (1958).

32 M. J. B. Thomas, K. H. Butler, and J. M. Harris, Illum.Engr. 52, 279 (1957); R. W. Mooney, J. Electrochem. Soc. 105,456 (1958).

33 L. G. van Uitert, R. R. Soden, and R. C. Linares, J. Chem.Phys. 36, 702 (1962); 36, 1793 (1962); F. C. Palilla and A. K.Levine, Appl. Opt. 5, 1467 (1966); A. K. Levine and F. C. Palilla,in Proceedings of International Conference on Luminescence,Budapest, 1966 (Akademiai Kiado, Budapest, 1968), Vol. 2,p. 2055.

34 H. Krefft, F. Rossler, and A. RUttenauer, Z. Tech. Phys.18, 20 (1937); F. Rossler, Ann. Physik 34, 1 (1939).

35 C. E. Weitz, G. E. Appl. Bull. LD-1 (General Electric Co.,Cleveland, Ohio, 1956).

36 C. W. Jerome, Illum. Engr. 56, 209 (1961).37 G. Dejardin, R. Falgon, and J. Janin, Lux (Paris), Jan.,

1961.38 With the data used here, the intensities of the Hg lines are

4047 A, 0.211 W; 4358 A, 0.544 W; 5461 A, 0.289 W; 5791 A,0.066 W. Almost all (81 %) of the luminous output of the dis-charge is contributed by the green (5461 A) line and essentiallyall the rest by the yellow line. The blue lines, particularly that at4358 A, however, have a predominant effect on the color.

39 The efficiency obtained with the optimum binary blend willalways be greater than that with a multicomponent blend.

40 M. Pivovonsky and M. R. Nagel, Tables of Blackbody Radia-tion Functions (Macmillan, New York, 1961), Tables VI andVII.

41 H. G. Jenkins, A. H. McKeag, and P. W. Ranby, J. Elec-trochem. Soc. 96,1 (1949); R. Nagy, R. W. Wollentin, and C. K.Lui, J. Electrochem. Soc. 95, 187 (1949); K. H. Butler and C. W.Jerome, J. Electrochem. Soc. 97, 265 (1950); G. R. Fonda, Brit.J. Appl. Phys. Suppl. 4, 69 (1955); J. L. Ouweltjes, PhilipsTech. Rev. 13, 344 (1952), J. Phys. Radium 17, 641 (1956); J.L. Ouweltjes and W. L. Wanmaker, J. Electrochem. Soc. 103,160 (1956); W. L. Wanmaker, J. Phys. Radium 17, 636 (1956);W. L. Wanmaker, A. H. Hoekstra, M. G. A. Tak, and M. L.Verheyke, Philips Res. Rept. 10, 11 (1955); 11, 1 (1956).

42 W. A. Thornton, J. Opt. Soc. Am. 61, 1155 (1971).

COPYRIGHT AND PERMISSION

This Journal is fully copyrighted, for the protection of the authors andtheir sponsors. Permission is hereby granted to any other authors to quotefrom this journal, provided that they make acknowledgment, includingthe authors' names, the Journal name, volume, page, and year. Reproduc-tion of figures and tables is likewise permitted in other articles and books,provided that the same information is printed with them. The best andmost economical way for the author and his sponsor to obtain copies is toorder the full number of reprints needed, at the time the article is printed,before the type is destroyed. However, the author, his organization, orhis government sponsor are hereby granted permission to reproduce partor all of his material. Other reproduction of this Journal in whole or inpart, or copying in commercially published books, periodicals, or leafletsrequires permission of the editor.

Vol. 62822