JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 27, NO. 22, NOVEMBER 15, 2009 5145

Effect of Phosphor Particle Size on LuminousEfficacy of Phosphor-Converted White LED

Nguyen The Tran, Jiun Pyng You, and Frank G. Shi, Senior Member, IEEE

Abstract—In this paper, the influence of YAG:Ce phosphor par-ticle sizes on the lumen output and the conversion efficiency of bothin-cup phosphor and top remote phosphor LED packages are in-vestigated with 3-D ray-tracing simulations. The lumen output andthe conversion efficiency of both types of phosphor-converted (pc)white LED packages are dependent on the size of YAG:Ce par-ticles. The lumen output and conversion efficiency of both typesof pc-white LED packages are minimal at the phosphor particlesize with the size parameter of around one and are highest at theparticle size in micron size. The simulation results show that bothin-cup and top remote phosphor packages have the highest lumenoutput and the highest conversion efficiency at the particle size ofaround 20 m.

Index Terms—LED, phosphor, phosphor size, WLED.

I. INTRODUCTION

S OLID-STATE lighting (SSL) for generating light that isperceived as white color by the human eye can be done

with one of the following options: 1) discrete color-mixing:mixing different LEDs of different colors to generate broad vis-ible spectrum (e.g., blue (B) + green (G) + red (R) LEDs, orB + G + R + yellow (Y) LEDs); 2) phosphor conversion LED:combining wavelength conversion materials with a short-wave-length LED providing activation wavelength that excites thewavelength conversion material and is partially converted toa longer wavelength to create a perceived white spectrum asphosphor-converted (pc) wavelength and the remainder of LED-emitted light are combined (e.g., UV LED + RGB phosphors,or blue LED + RG phosphors, or blue LED + RGY phosphors,or blue LED + Y phosphor, or blue LED + YR phosphor) and;3) combination of options 1 and 2 (blue and red LEDs + greenor yellow phosphor). Generation of white LED light is usuallydone by combining color LEDs and wavelength conversion ma-terials instead of RGB LEDs because of the absence of an effi-cient emitter of green and yellow light and the complexity of aRGB LED package including electrical connections and sophis-ticated optics for blending the discrete colors.

Although inorganic phosphor materials such as YAG phos-phors usually have high quantum efficiency, the conventional

Manuscript received February 11, 2009; revised June 02, 2009. First pub-lished July 21, 2009; current version published October 02, 2009.

N. T. Tran and F. G. Shi are with the Optoelectronics Packaging andMaterials Labs, University of California, Irvine, CA 92697 USA (e-mail:ntran3000@yahoo.com; jpyou20004@yahoo.com; fgshi@uci.edu).

J. P. You is with Nepes LED Corporation, Singapore 569060 Singapore(e-mail: jpyou20004@yahoo.com).

Color versions of one or more of the figures in this paper are available onlineat http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/JLT.2009.2028087

pc-white LED still has low phosphor conversion efficiency. Theconventional pc-white LED suffers greatly from light absorp-tion loss inside the package. Light loss in the pc-LED deviceis dominated by scattering and reflection of excitation light byphosphor particles back into the LED chip and backward emis-sion and scattering of pc light being directed into the LED chip.There is a large amount of light energy reflected backward tothe package from the phosphor layer and circulated inside thepackage. Backscattering and circulation of light inside the LEDpackage result in a high absorption loss of the emitted light.The backward propagation light is due to scattering of lightby phosphor particles and backwardly emitted portion of phos-phor-emitted yellow light. It is showed that about 40% of lightis transmitted through the phosphor layer while about 60% oflight is reflected backward [1], [2].

Backscattering and back reflection of light by phosphor par-ticles can be minimized by optimizing phosphor particle sizes.It has been reported that photoluminescence (PL) of YAG:Cephosphor particles can be improved by manipulating the sizeof phosphor particles [3], [4]. In their reports, a smaller averagephosphor size has a lower PL intensity or PL quantum yield thatis defined as the ratio of the number of photons being radiatedby the photoluminescent materials to the number of photons oflight being absorbed by the luminescent materials. However, theauthors do not tell how they measured the PL intensity and onlythree phosphor sizes were studied. The measurements of thePL quantum yield are usually made on thin film samples, so-lutions, and powders, and thus do not represent a real pc-LEDdevice. Moreover, their studies focus on the internal quantumefficiency of YAG:Ce phosphor and do not tell the effects ofphosphor sizes on the fluorescence conversion efficiency of areal pc-LED device. Kuma [5] shows more details on the ef-fect of phosphor sizes on the fluorescence conversion efficiencyof a CdSe composite film. Both the effects of reflective indexof semiconductor nanocrystal and the effects of self-absorptionof CdSe material are taken into consideration in Kuma’s study.However, his study is for CdSe nanosize particles and limits toa flat CdSe-composite film.

In a conventional pc-LED device, phosphor particles canbe formed a thin layer on top of a LED chip, or distributedthroughout the cup, or dispensed in a layer separated from theLED chip by a clear encapsulation material [6]–[8] to increasethe luminous efficacy (lm/W). In any phosphor geometriesdescribed above, the phosphor size is expected to influence theluminous efficacy of the pc-LED device.

The goal of this paper is to study the effect of phosphor par-ticle sizes on the luminous efficacy of the pc-LED device andfind the optimal phosphor size based on 3-D ray-tracing simula-tions (LightTools). Two different pc-LED packages with in-cup

0733-8724/$26.00 © 2009 IEEE

5146 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 27, NO. 22, NOVEMBER 15, 2009

Fig. 1. Schematic cross-sectional view of pc-white LEDs with: (a) in-cup phos-phor; (b) top remote phosphor.

distributed phosphor and remote phosphor as shown in Fig. 1are used to examine the effects of YAG:Ce particle sizes on theluminous efficacy.

II. SIMULATION OF LIGHT OUTPUT OF PHOSPHOR-CONVERTED

WHITE LED

Effects of YAG:Ce phosphor particle sizes on the light outputof pc-white-LED are examined by using 3-D ray-tracing simu-lations with LightTools software. Two white LED package ge-ometries are used in this study: 1) in-cup phosphor package inwhich phosphor particles are mixed with an encapsulant and thephosphor-encapsulant mixture is dispensed throughout the cupas shown in Fig. 1(a); 2) remote phosphor package in whichthe phosphor-encapsulant mixture forms a thin phosphor layerbeing separated from the LED chip by a layer of a transparentmaterial as shown in Fig. 1(b).

To perform simulations, a 3-D model for each of two geome-tries as shown in Fig. 1 was built by using LightTools. An LEDchip with a square base of 1 mm and a height of 0.15 mm isbonded in a cavity of reflective cup. The LED chip is encap-sulated with a phosphor layer formed by an encapsulation ma-terial containing phosphor particles [as shown in Fig. 1(a)] ora clear encapsulation material residing below an encapsulationmaterial containing phosphor particles [as shown in Fig. 1(b)].A convex lens resides on the top surface of the phosphor layerof each of these two packages. The refractive indexes of the ma-terials for the convex lens, the transparent encapsulant, and theencapsulant mixing with the phosphor material are assumed tobe the same and are equal to 1.46 at the excitation wavelengthof 460 nm. Phosphor particles are assumed to have a sphericalshape and have refractive index of 1.8 at all wavelengths of theemitted light.

There are several events happened with propagating lightwithin an LED package including light absorption (such as byLED chip materials, encapsulant, reflective cup, and phosphorparticles), scattering by phosphor particles, and reflection andrefraction at interfaces. The scattering intensity distribution ofthe excitation light and phosphor-emitted light is an importantparameter in this simulation. It depends on the phase function,which is a function of phosphor size, wavelength, and refractiveindex of the phosphor particle and the medium [9]. In thisstudy, variable parameter is phosphor particle size. Phosphorparticles of different sizes scatter different amount of lightwith different angular distribution, and a smaller particle sizescatters more light than a larger particle size [9]. In order to

Fig. 2. Illustration of the interaction between the excitation light and a phos-phor particle: (a) scattering light distribution; (b) isotropic emission of yellowlight by a phosphor particle.

account for different scattering properties of different particlesizes, different scattering intensity distributions of light wave-length of 430, 460, 500, 550, and 600 nm for different particlesizes are calculated by using Mie theory. Scattering intensitydistribution of light at other wavelengths is interpolated orextrapolated during simulation.

Absorption loss by materials other than phosphor particlesis calculated by using Beer–Lambert law. Light absorption byphosphor particles is calculated based on absorption probabilityof a phosphor particle per incident event. Absorption probabilityof a phosphor particle is calculated by using Mie theory [9].

Phosphor concentration is varied in this study so that dif-ferent correlated color temperature (CCT) of output light can beachieved. Phosphor concentration is represented by light pathlength in the software. Higher phosphor concentration meanssmaller photon path length due to higher probability of hittingphosphor particles by light.

In these simulations, light is treated as photon particle. Eachphoton of light is traced until it is emitted to ambient from LEDpackage or its energy is vanished due to absorption by packagematerials.

Simulation process starts with the emission of blue light spec-trum from the LED chip in terms of power (watt). Upon inter-acting with phosphor particles, the excitation blue light is par-tially absorbed and scattered by phosphor particles. The absorp-tion of the excitation blue light spectrum by phosphor particlesis followed by fluorescence of isotropic radiation of yellow lightspectrum by phosphor particles. Fig. 2(a) and (b) illustrate theinteraction between the excitation blue light and a phosphor par-ticle. Fig. 2(a) is for scattering of the excitation blue light bya phosphor particle while Fig. 2(b) is for the isotropic emis-sion of yellow light by a phosphor particle. Unlike the bluelight, yellow light that interacts with phosphor particles doesnot result in the fluorescence of light. The interaction betweenyellow light and phosphor particles results in a partial absorptionand scattering of yellow light. The absorption of yellow lightby yellow-emitting phosphor particles is called self-absorption.Kuma [5] shows that phosphor self-absorption significantly in-fluences the outcome. The amount of yellow light absorbed byphosphor particles at each interaction site depends on optical

TRAN et al.: EFFECT OF PHOSPHOR PARTICLE SIZE ON LUMINOUS EFFICACY 5147

Fig. 3. Lumen output of in-cup phosphor LED packages with different particlesizes. The unit for the diameter D shown in this figure is in �m.

properties of phosphor particles that can be obtained from phos-phor manufacturers or measurements.

In simulation, absorption loss of a traced light ray is updatedat every interfaces and scattering sites to calculate the remainingenergy of a traced light ray. As light is incident on an interface,it undergoes reflection and/or refraction that are governed byFresnel equation of reflection. Once light propagates to ambientair through the interface between the convex lens and ambientair, it is collected by a receiver. The collected power of light isscored into memory so that lumen and CCT of light output arecalculated.

III. EFFECT OF PARTICLE SIZES

The influence of phosphor particle sizes on the lumen outputof in-cup phosphor package is shown in Fig. 3(a) and (b).Fig. 3(a) shows that as the particle size increases from nano-sizes, the lumen output of pc-white-LED decreases thenincreases. The lumen output decreases as the phosphor particlesize increases from the particle of nanosize to submicronsize (between 0.1 m and 0.5 m), and then increases as theparticle size continues to increase to the micron sizes. In themicron-size region, the lumen output reaches its maximumvalue at the particle size of around 20 m, and decreases at a

larger particle size. It is interesting that the minima point occursat the size parameter of around one.This observation can be explained by using the extent of lightscattering. In particle-encapsulant composite, the extent of lightscattering is complex. The factors for determining the extent oflight scattering include the phase function that depends on theparticle size, the degree of refractive index mismatch betweenthe particle and the encapsulant, the concentration of suspendedparticle, and the wavelength of light.

In this study, the refractive index mismatch is 0.34 at thewavelength of 460 nm and is the same for all sizes at the samewavelength of light because only one type of phosphor par-ticle (YAG:Ce) is considered in this study. Generally, the ef-fect of the refractive index mismatch on the extent of light scat-tering is less important for the composite with particles of a sizemuch smaller than the wavelength of light [10]. Visible lightpropagating in a composite with nanosize particles usually doesnot see particles as individual particles. It is well known that ananoparticle composite transmits higher amount of visible light(that is much larger than the size of particles) than a compositewith the particles having the size parameter of around one ora composite with the particle size of around the wavelength oflight. This means that the nanoparticle composite has lower trap-ping efficiency than the composite with the size parameter ofaround one or with the particle sizes of around the wavelengthof light. Therefore, as the particle size increases from a nano-size to submicron size, the lumen output decreases. As the par-ticle size continues to increase from the submicron size corre-sponding to the minimal lumen output, the composite becomesmore and more transparent to the visible light [9]. The increasein light transparency reduces the trapping efficiency caused bythe scattering of particles. The lumen output thus increases asthe particle size increases from the particle size correspondingto the minimal lumen output. However, an increase in the par-ticle size requires a higher phosphor concentration to achievethe light output of the same CCT or to absorb enough amountof the excitation blue light for yellow light emission.

As particle size becomes too large, propagation directionof light is slightly changed because a relatively large particleslightly scatters light. This increases the trapping efficiency ofphosphor-emitted light and LED-emitted light propagating inthe horizontal direction, which is parallel to the top surfaceof the phosphor layer or in a direction having a low anglerespective to the horizontal direction. In order to understandan increase in absorption loss as phosphor particle size islarger 20 m, a detailed analysis is performed to reveal lossmechanism. In the detailed analysis, absorption loss by clearencapsulant, reflective cup, and LED chip are revealed. Forperforming the detailed analysis, a light collector or receiveris set at each surface of the reflective cup to calculate amountof light being absorbed by the reflective cup, and two lightcollectors or receivers are set at each interface to calculatethe amount of light going in and out of each material layer.From this information, absorption loss in each layer can bedetermined. The analysis reveals that there is a higher amountof light being absorbed by the tilted surface of the cup and thereis a lesser amount of light being absorbed by the bottom surfaceand the LED chip in the LED package with a larger particle

5148 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 27, NO. 22, NOVEMBER 15, 2009

Fig. 4. Lumen output of top remote phosphor LED packages with differentparticle sizes. The unit for the diameter D shown in this figure is in �m.

size. This means there is more energy of light propagating inthe horizontal direction, which is parallel to the top surfaceof the phosphor layer or in a direction having a low anglerespective to the horizontal direction, and there is less energyof light propagating in backward direction as particle sizeincreases. Although there is a significant change in the amountof light absorption by LED chip and the tilted and bottomsurfaces of the reflective cup, the total amount of light beingabsorbed by LED chip and the surfaces of the reflective cupis insignificantly changed as phosphor particle size increasesfrom 20 m. However, Fig. 3(b) shows that the lumen outputof the LED device with the phosphor particle size of 20 m ishigher than the LED device with a phosphor particle havingits size larger than 20 m. This means there is an increase inabsorption loss by phosphor layer with phosphor particle sizelarger than 20 m. It is expected that self-absorption of light byphosphor materials accounts for a high portion of addition losswhen phosphor particle size increases.

Similar to the pc-white LED package with in-cup phosphor,the LED package with top remote phosphor has similar trendof the lumen output (Fig. 4) but different in magnitude. Fig. 4shows that the light output of the top remote phosphor packageis less dependent on the phosphor size than that of the in-cupphosphor. Separation of the phosphor layer and the LED chipin the top remote phosphor package lowers the amount of back-wardly emitted yellow light and backscattered light being di-rected into the LED chip. As a result, the conversion efficiencyof the top remote phosphor package is higher than that of thein-cup phosphor package and changes at a slower rate as thephosphor particle size is varied.

For both types of the LED package with the micron particlesizes, the lumen output increases and then decreases as the CCTincreases. This is because the lumen output is a function of thetotal power output and the power distribution of a white lightspectrum. In the dichromatic white light spectrum with a lowerCCT, the power ratio of yellow to blue light is higher. Sinceyellow light has a higher luminous efficacy than blue light, theluminous efficacy increases as the CCT decreases. In order tolower the CCT, the concentration of phosphor needs to be in-creased to increase the absorption of blue light and the emission

of yellow light by phosphor particles. An increase in the phos-phor concentration leads to an increase in the trapping efficiencyand thus increases the absorption loss [11]. As the concentrationincreases up to a certain value, the negative effect of the totalpower loss on the lumen output is superior to the positive effectof the yellow light on the lumen output. The lumen output thusdecreases as the CCT becomes relatively low.

Fig. 5 shows relative conversion efficiency that is the ratioof conversion efficiency to maximum conversion efficiency ofthe pc-white LED as a function of phosphor size and CCT. Theconversion efficiency is defined as the ratio of the emitted powerof light by the LED package to the power of light emitted by theLED chip while the maximum conversion efficiency is definedas

where is the power of light (blue) emitted by the LED chip,is the power of blue light in the dichromatic white light

output of a pc-white LED, and is the conversion efficiencyof phosphor material. The power of blue light in the dichromaticwhite light output can be calculated by dividing the dichromaticwhite spectrum into two spectra: one spectrum for blue light andone spectrum for yellow light as shown in Fig. 6.

The LED package with the particle sizes corresponding tohigher lumen output has higher relative conversion efficiencies.In contrast to the lumen output that increases then decreaseswith the CCT varying in the range between 4000 K and 8000 Kas seen in Figs. 3(b) and 4, the relative conversion efficiency asshown in Fig. 5 has the lowest value at the CCT of 4000 K andincreases with an increase in the CCT for all particle sizes. Thepackage with a lower CCT has a higher trapping efficiency dueto higher concentration of phosphor particles [11]. The highertrapping efficiency results in a higher power loss or a lowerpower output, and thus, it lowers the relative conversion effi-ciency. At the light output with the CCT of 6500 K, the top re-mote phosphor package with the optimal phosphor size has rela-tively high conversion efficiency, around 89%. Therefore, thereis no need for an improvement of the conversion efficiency ofphosphor layer if the cost is high. However, at a relatively lowCCT such as 4000–5500 K, the conversion efficiency of bothtypes of the pc-white LED packages is low. Therefore, a highefficient package such as PSE package [2] or other type of re-mote phosphor packages is needed.

Key experiments were performed to verify the simulation re-sults. Commercial InGaN LED chips with its size of mm

mm mm and its emission light peaked at 460 nmwere used in the experiments. The LED packages used in theseexperiments had approximately same size, shape, and materialproperties as those of the simulated package shown in Fig. 1(a).Commercial YAG:Ce phosphors with two mean sizes of 6–8 mand 15 m, according to the phosphor manufacturer, were used.The experimental results show that the lumen output with CCTof around 5200 K from the package with a mean phosphor sizeof 15 m is about 3.2% higher than that from the package withmean phosphor size of 6–8 m. This is consistent with simula-tion results, which is 1.9% (for the particle size of 15 m versus8 m) and 7.8% (for the particle size of 15 m versus 6 m).

TRAN et al.: EFFECT OF PHOSPHOR PARTICLE SIZE ON LUMINOUS EFFICACY 5149

Fig. 5. Relative conversion efficiency of the pc-white LED as a function ofCCT and the particle size: (a) & (b) for in-cup phosphor; (c) remote phosphor.The unit for the diameter D shown in this figure is in �m.

IV. CONCLUSION

The lumen output and the conversion efficiency of a pc-whiteLED device with in-cup phosphor and remote phosphor geome-tries are dependent on the size of YAG:Ce particles. They have

Fig. 6. Spectrum of dichromatic white LED.

a lowest value at the particle size with the size parameter ofaround one and they have a highest value at the particle size inmicron size. Simulations show that both types of packages haveoptimal phosphor size of around 20 m. Under same conditionssuch as CCT and phosphor size, lumen output of the remotephosphor package is less sensitive to phosphor size and higherthan that of the in-cup phosphor package.

ACKNOWLEDGMENT

The authors thank Dr. Y. He for technical support in fabri-cating the devices.

REFERENCES

[1] K. Yamada, Y. Imai, and K. Ishii, “Optical simulation of light sourcedevices composed of blue LEDs and YAG phosphor,” J. Light Vis. En-viron., vol. 27, no. 2, pp. 70–74, 2003.

[2] N. Narendran, Y. Gu, J. P. Freyssinier-Nova, and Y. Zhu, “Extractingphosphor-scattered photons to improve white LED efficiency,” Phys.Stat. Solidi (a), vol. 202, no. 6, pp. R60–R62, 2005.

[3] Y. C. Kang, I. W. Lenggoro, S. B. Park, and K. Okuyama, “YAG:Cephosphor particles prepared by ultrasonic spray pyrolysis,” Mater. Res.Bull., vol. 35, pp. 789–798, 2000.

[4] F. Yuan and H. Ryu, “Ce-doped YAG phosphor powders prepared byco-precipitation and heterogeneous precipitation,” Mater. Sci. Eng.,vol. B107, pp. 14–18, 2004.

[5] H. Kuma, “Fluorescent conversion medium and color light emittingdevice,” U.S. Patent 2007/0165661 A1, Jul. 19, 2007.

[6] H. Chen, “LED structure with ultraviolet-light emission chip andmultilayered resins to generate various colored lights,” U.S. Patent5,962,971, Oct. 5, 1999.

[7] A. Duggal, “Phosphors for white light generation from UV emittingdiodes,” U.S. Patent 6,294,800 B1, Sep. 25, 2001, et al..

[8] T. Taguchi, Y. Uchida, and K. Kobashi, “Efficient white LED lightingand its application to medical fields,” Phys. Stat. Solidi (a), vol. 202,no. 12, pp. 2730–2735, 2004.

[9] N. T. Tran, C. G. Campbell, and F. G. Shi, “Study of particle size ef-fects on an optical fiber sensor response examined with Monte Carlosimulation,” Appl. Opt., vol. 45, no. 29, pp. 7557–7566.

[10] M. J. Manning, “CMP pad with composite transparent window,” U.S.Patent 6,832,947 B2, Dec. 21, 2004.

[11] N. T. Tran and F. G. Shi, “Study of phosphor concentration andthickness for phosphor-based white light-emitting-diodes,” J. Lightw.Technol., vol. 26, no. 21, pp. 3556–3559, Nov. 2008.

Nguyen The Tran received the Ph.D. degree in chemical and biochemical en-gineering from University of California (UCI), Irvine, in 2007.

Before coming to UCI, he worked on renewable energy, specifically self-sus-tained Hynol-Gaso-Fischer-Tropsch at Ce-cert. Currently, he is with NEPESLED and works on developing LED lighting systems.

5150 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 27, NO. 22, NOVEMBER 15, 2009

Jiun Pyng You received the Ph.D. degree from the University of California,Irvine, in 2008.

He is working for Nepes LED Corp. to develop new packaging materials forhigh-power LED. Before joining LED industry, he worked for IC manufacturingcompany, TSMC in Taiwan.

Frank G. Shi (M’01–SM’02) is a Professor at the University of Cali-fornia, Irvine, where he directs the Optoelectronics Packaging and MaterialsLaboratory.