Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
Thermal management of the remote phosphor layer in LED systems Indika U. Perera and Nadarajah Narendran*
Lighting Research Center, Rensselaer Polytechnic Institute, 21 Union St., Troy, NY 12180 USA
ABSTRACT
Generally in a white light-emitting diode (LED), a phosphor slurry is placed around the semiconductor chip or the phosphor is conformally coated over the chip to covert the narrowband, short-wavelength radiation to a broadband white light. Over the past few years, the remote-phosphor method has provided significant improvement in overall system efficiency by reducing the photons absorbed by the LED chip and reducing the phosphor quenching effects. However, increased light output and smaller light engine requirements are causing high radiant energy density on the remote-phosphor plates, thus heating the phosphor layer. The phosphor layer temperature rise increases when the phosphor material conversion efficiency decreases. Phosphor layer heating can negatively affect performance in terms of luminous efficacy, color shift, and life. In such cases, the performance of remote-phosphor LED lighting systems can be improved by suitable thermal management to reduce the temperature of the phosphor layer. To verify this hypothesis and to understand the factors that influence the reduction in temperature, a phosphor layer was embedded in a perforated metal heatsink to remove the heat; the parameters that influence the effectiveness of heat extraction were then studied. These parameters included the heatsink-to-phosphor layer interface area and the thermal conductivity of the heatsink. The temperature of the remote-phosphor surface was measured using IR thermography. The results showed that when the heat conduction area of the heatsink increased, the phosphor layer temperature decreased, but at the same time the overall light output of the remote phosphor light engine used in this study decreased due to light absorption by the metal areas.
Keywords: light-emitting diode, remote phosphor, thermal management, solid-state lighting, down-conversion, IR thermography, extended-surface heat conduction
1. INTRODUCTION There are two main components in a phosphor-converted white LED package that convert energy: 1) the LED chip that converts electrical energy to a short-wavelength visible radiant energy, and 2) the phosphor material that down-converts the short-wavelength visible radiation to a broadband long-wavelength visible radiation [1],[2] The heat generation inside a phosphor-converted white LED package is mostly due to the inefficiencies in these conversion processes.
At present, the heat generation within an LED and the effect of this heat on LED performance is well understood [3]. With the industry moving toward higher lumen packages with smaller footprint light engines, there is significant research interest in understanding the heat production within the phosphor layer and its effects. Generally, phosphor conversion efficiency is negatively affected by an increase in temperature [4]. In addition, the temperature rise affects the binding material used in creating the phosphor layers and reduces the overall light output of the LED. The temperature rise caused by a reduced phosphor conversion efficiency and additional light absorption by the binding material accelerates lumen degradation and reduces system useful lifetime [5].
In a white LED system, the amount of phosphor layer heat buildup depends on a number factors that include: 1) the location of the phosphor, 2) the thickness and concentration of the phosphor layer, 3) the binding media used in creating the phosphor layer, 4) the LED chip and package structure, and 5) the phosphor conversion efficiency [1],[2],[6]-[8].
Past studies have observed the heat generated in the phosphor layer, caused by the conversion efficiency losses (quantum conversion losses and Stokes shift losses) and trapped photons in the phosphor layer (due to total internal reflection and Fresnel reflection), can be as high as 13% of the total input electrical power to the LED system [8]. Furthermore, studies have stated this generated heat in the phosphor layer can increase the operational temperature of the phosphor layer to greater than 150°C [2],[7].
*Corresponding author: +1 (518) 687-7100; [email protected]; http://www.lrc.rpi.edu/programs/solidstate/
LED-based Illumination Systems, edited by Jianzhong Jiao, Proc. of SPIE Vol. 8835, 883504 · © 2013 SPIE · CCC code: 0277-786X/13/$18 · doi: 10.1117/12.2023094
Proc. of SPIE Vol. 8835 883504-1
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 10/03/2013 Terms of Use: http://spiedl.org/terms
- - - -ITTtrrri-
iphosp]layerheatsiv
reflect
LEDs
heatsiv
horandnk
aluhe
epoxy-I
tor cup b
nninum
.atsink
I- phosphor
ayer
heatsink #1
1
1
L
heatsink #2
Ac7
As a solutiomedia, such reduce the pphosphor lay
The operatinand thereforphosphor layLED system
A metal heatsurrounding heatsink, theconduction atemperature. the phosphorphosphor lay
Table 1. H
HeatsinkHeatsinkHeatsink
Figure 1. and epoxy
A sketch of tFigure 1 (leftthe drive curthe LEDs to embedded he(right). The eareas throheatsinks (Ta
n to this heat as glass and a
phosphor operayer has been pu
g temperature e, investigatio
yer and analyzwere the goals
tsink, similar toambient. It wae temperature area (Table 1)It was also hy
r layer. Alumiyer temperature
Heatsink configu
Diamthe
(mk #1 1k #2 7k #3 4
Sketch of LED y layer (right)
the LED light ft). The short-wrrent of the LE
the phosphoreatsink configuemitting surfacough in heable 1). The rea
buildup in phalumina, for fabation temperat
ursued with the
of the phosphoon of a therming the effect s of this study.
o those shown as hypothesized
of the phosp were used to
ypothesized thainum and acrye.
urations
meter of e hole mm)
Nu
9.1 7.2 4.4
light engine set
engine assembwavelength irra
Ds and maintar layer-embeddurations had eqce area of heatsink #2, are ason for keepin
hosphor layersbricating the pture [2]. In ad
e use of a therm
or layer becomal managemenof the thermal
2. MEin Figure 1, w
d that by increahor layer wou
o investigate that heatsinks wylic heatsinks
Thickness of
umber of holes su
1 7
19
tup and heatsink
bly for irradiatadiance on the aining the sameded heatsink wqual emitting suatsink #1 and tequal. Similar
ng the emitting
s, some studiephosphor layer ddition to thes
mally conductin
mes a critical pant method thal management
ETHODOLOwas used to conasing the contauld decrease. he effect of in
with higher therwere used for
all heatsinks =Emitting
urface area (mm2)
285 286 285
k (left), and sket
ting the phosphphosphor layere estimated jun
was also mainturface areas, athe total emittirly, heatsink #g areas the sam
es have investi[1], while oth
se studies, dedng material atta
arameter in deat could reduc
method on th
OGY nduct the heat act surface areaThree heatsinncreased heat rmal conductivr investigating
= 1.5 mm Emitting
surface arearatio 1.0 1.0 1.0
ch of heatsinks
hor layer ember was maintainnction temperatained at a conas indicated in ing surface are#3 had the samme was to ensur
igated thermalhers have studidicated thermaached to the ph
etermining the ce the operatinhe overall optic
away from thea between the p
nk configuratioconduction ar
vity would redug the thermal
a Conductiarea (mm
90 238 391
#1 and #2 with
edded in the hened at a constanature of the LEnstant distanceTable 1 and as
ea, which is theme emitting surre the same vis
lly conductive ied pulsed opeal managemenhosphor layer [
useful system ng temperaturecal performanc
e phosphor layphosphor layerons with varyrea on phosphuce the temperconductivity e
ion m2)
Conduarea r
1.02.64.4
embedded phosp
eatsink is illusnt value by con
EDs. The distane. The three phs illustrated in e summation ofrface area as tsible radiant po
binding ration to
nt of the 9].
lifetime, e of the ce of the
yer to the r and the ing heat
hor layer rature of effect on
ction ratio
0 6 4
phor
strated in ntrolling nce from hosphor-Figure 1 f surface the other ower. On
Proc. of SPIE Vol. 8835 883504-2
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 10/03/2013 Terms of Use: http://spiedl.org/terms
T$21111c:0Ul:12
iuwagitiíii¡ }%lll
T$21711c:0Up12
iuwv.,uitiíiig }+íit
the other hanheatsink mat#1, , is smincrease the dissipating th
A number ogenerate a unwas maintaincooling fan a
Figure 2.
Figure 3.
nd, the heat cterial, increase
maller than the amount of hea
he heat to the a
f LEDs (peakniform irradianned at a fixed and a heating el
Experimental se
Experimental se
conduction ares as the numbetotal area in h
at transferred fambient.
k wavelength ~nce on the pho
temperature alement control
etup for surface
etup for visible r
ea, which is ther of holes inc
heatsink #2 in Ffrom the phosp
3. EX~448 nm) wersphor plate at
at the referencelled by a tempe
temperature mea
radiant power wi
he interfacial crease (Table 1Figure 1 (rightphor layer to t
XPERIMENre arranged ona fixed distance location specerature control
asurement includ
ith sphere-spectr
surface area b1). As an examt). This increasthe heatsink, g
NT n an LED heace from the LEcified by the Lsystem, as illu
ding the IR imag
rometer system
between the phmple, the conduse in the heat c
given that the h
atsink and refleEDs. The LEDLED manufactustrated in Figu
ging camera
hosphor layer uction area in conduction areheatsink is effe
ector cup asse operation temturer with the ure 2.
and the heatsink
ea would fective in
embly to mperature
use of a
Proc. of SPIE Vol. 8835 883504-3
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 10/03/2013 Terms of Use: http://spiedl.org/terms
The machine(1:10 ratio bsimilar surfafrom the pho3, respectivel
An IR cameembedded hecenter of the conducted atheatsink. A spower was m
Figure 4.
Due to the plocation of thresults for theach heatsinkreduced the increase in hloss of one-theatsink #1 rincrease did reduced when
Table 2. G
GeometrtemperatRelativeradiant
The results ioperation temlight engine u
ed aluminum aby weight). Thace properties. osphor layer anly.
ra (spectral seeatsink with anphosphor layet steady-state sphere-spectrom
measured at stea
Geometric cente
physical symmhe phosphor la
he aluminum hk configuratiogeometric cen
heat conductiontenth of the opreduced the mnot provide mn the heat cond
Geometric cente
ric center ture (°C) e visible t power
in Table 2 supmperature of thused in this stu
nd acrylic heahe aluminum a
Surface tempend heatsink we
ensitivity: 7.5–n estimated tarer-embedded he
operation of meter system wady-state opera
er temperature m
metry of the hayer-embedded
heatsink configon relative to hnter temperaturn area by 2.6 tptical power. F
maximum tempemuch benefit wduction area ra
er temperature an
HeatsinkAvg. 110.8
100%
pport the hypohe phosphor layudy decreased.
tsinks were emand acrylic heaerature of the re alternatively
–14 μm) was uget surface emeatsink assembthe LEDs an
was used to mation of the LE
measurement loc
4.heatsink configd heatsink was
gurations for thheatsink #1. Inre by 25%, butimes reduced Further increaserature by 29%
with respect toatio was increas
nd relative visibl
k #1 Std. Dev.
1.5
0%
othesis that anyer. However,
mbedded with tatsink surfacesphosphor laye
y measured wi
used to measumissivity. The sbly, as illustratend steady-stead
measure the visiEDs and the pho
cation on heatsin
RESULTSgurations, the s used for anahe geometric cncreasing the ut the relative the phosphor
sing the heat c% while the reo thermal manased greater than
le radiant power
HeatsAvg.
83.5 (-25%)
89% (-11%)
n increased heaat the same tim
the same phosps facing the Ler and heatsinkith the equipme
ure the surfacesurface tempered in Figure 4.dy temperaturible radiant poosphor layer-em
nk configurations
steady-state talysis of heatsienter temperatconduction arvisible radian
operation maxonduction area
elative radiant agement whilen 2.6 times the
results for alum
sink #2 Std. Dev.
0.2
2%
at conduction me, the overall
phor-to-epoxy LEDs were paik, and visible ent illustrated
temperature orature was mea. All temperature of the phoower. The steadmbedded heats
s
temperature atink effectiveneture and the virea from heatsnt power also ximum temperaa of heatsink #power reduced
e the optical pe heat conducti
minum heatsink c
HAvg.
79.1 (-29
67% (-33
area of the hl light output o
concentrationinted white to radiant powerin Figure 2 an
of the phosphoasured at the geure measuremesphor layer-emdy-state visiblesink.
t the geometriess. Table 2 shisible radiant psink #1 to heareduced by 1ature considera#3 to 4.4 timed by 33%. The
performance drion area of heat
configurations
Heatsink #3 Std. D
%) 1.2
%) 1%
eatsink leads tof the remote p
n mixture achieve
r emitted nd Figure
or layer-eometric
ents were mbedded e radiant
ic center hows the power of atsink #2 1%. The ably at a s that of e further rastically tsink #1.
Dev.
2
%
to lower phosphor
Proc. of SPIE Vol. 8835 883504-4
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 10/03/2013 Terms of Use: http://spiedl.org/terms
LED mount surface
Heatsink surface
Cavity wall surface
Heat conduction surface
Table 3. Geometric center temperature measurements from aluminum and acrylic heatsinks
Heatsink material
Geometric center temperature (°C) Heatsink #1 Heatsink #2 Heatsink #3
Avg. Std. Dev. Avg. Std. Dev. Avg. Std. Dev. Aluminum 110.8 1.5 83.5 0.2 79.1 1.2
Acrylic 125.8 2.1 126.3 0.2 127.3 0.3
Table 3 shows the effect of thermal conductivity of the heatsink material on reducing phosphor layer temperature. The aluminum heatsink reduced the geometric center temperature as the heat conduction area was increased while no reduction in the geometric center temperature was observed with acrylic heatsinks when the heat conduction area was increased. In addition, the acrylic heatsinks displayed higher phosphor operational temperature for all tested heatsink conduction area configurations. These results illustrate that the geometrical aspects of the heatsink and thermal properties of the heatsink material are important in this thermal management method.
LightTools®, a commercial ray-tracing simulation software, was used to systematically analyze the reduction of visible radiant power when the heat conduction area was increased. The phosphor layer was modeled as a non-scattering medium with a constant refractive index for these numerical simulations while the internal surface reflectivities were changed systematically. These surfaces were considered as optical absorbers when the surface reflectivities were not equal to unity. Figure 5 shows the different internal surfaces of the LED light engine setup where the surface reflectance was varied.
Figure 5. Sketch of LED light engine setup internal surfaces
Reduction in surface reflectivity decreased the visible radiant power output from the LED light engine due to the surface absorption of photons. The visible radiant power output reduction was higher for the LED mount surface, the heatsink underside surface, and the cavity wall (reflector cup) surface reflectivity reduction. The visible radiant power output reduction was lower for the heatsink conduction surface reflectivity reduction compared to the other surfaces stated above.
The visible radiant power output change was small (<5%) among the heatsink configurations for all internal LED light engine surfaces except for the heatsink conduction surface. The change in heatsink configurations was ~10-15% with respect to visible radiant output power for the heatsink conduction surface as the heatsink conduction area surface reflectivity was reduced.
These findings provide a plausible explanation for the reduced visible radiant power when the heat conduction area was increased. The increased conduction area improved heat transfer but it also increased the surface absorption of photons, thus reducing the visible radiant power. A similar trend was observed with the reflectivity of the cavity surface and the LED mount surface.
Proc. of SPIE Vol. 8835 883504-5
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 10/03/2013 Terms of Use: http://spiedl.org/terms
Figure 6.
Figure 6 shodirection is itemperature phosphor laytemperature rof the phosph
The use of mtransferring ithermal condreduced the engine config
The authors wInstitute, Hensupport of Rat the Lightin
[1] Wiesmasimulat
[2] Wenzl, impact convert10.1364
[3] Arik, M5187, 6
[4] Lin, C. 1277 (2
[5] Narendgeneral
Surface tempera
ows the surfaceillustrated in tcompared to
yer. The surfacreduces the lumhor compared t
metal heatsinksit to the ambienduction propertvisible radiantgurations to im
would like to ankel, and the B
Russell Leslie, Ang Research Ce
ann, Ch., Linktions,” Proc. SPF. -P., Sommeof the non-lin
ted LEDs und4/OE.21.00A4
M., Becker, C., 64-75 (2004). d
-C., and Liu, 2011). doi: 10.1dran, N., Maliyl illumination a
ature profile of h
e temperature the inset diagrheatsink #1,
ce temperaturemen depreciatito phosphor an
s reduced heat nt. The effectivties. The heatst power. As pa
mprove overall
acknowledge tBesal Lighting Andrew Biermenter.
ov, A., and MuPIE 8550, 8550er, C., Hartmannearity of the rder different c39 Weaver, S., an
doi: 10.1117/12R. -S., “Advan1021/jz200245yagoda, N., Bapplications,” P
heatsinks #1 and
profile along arams. Heatsinkit also mainta
e profile of heaion due to bindnd binding mate
5. Dbuildup in the
veness of heat tinks used in thart of an ongolight output wh
ACKNOWthe financial suEducation Fun
man, Jean Paul
REFuschaweck, J., 02G-1-5 (2012nn, P., Pachlerradiant flux oncurrent drivin
nd Petroski, J., 2.512731 nces in phosph52 ierman, A., P
Proc. SPIE 393
d #2
a radial directik #2 not only ained a lower atsink #3 was ding material derial operating
DISCUSSIONe phosphor layetransfer of theshe present studoing study, thehile reducing th
WLEDGMEupport of the Lnd provided byFreyssinier, Y
FERENCES“Estimating th
2). doi: 10.1117r, P., Hoschopfn the thermal lng schemes,”
“Thermal man
hors for light-em
ysar, R., and 38, 240-248 (20
ion of the phohad a lower uniform opersimilar to hea
degradation and at a higher tem
N er by effectivese metal heatsidy reduced the authors are inhe phosphor la
ENTS Lighting Researy Acuity Brand
Yi-wei Liu, Ma
he performance7/12.2001607f, H., Langer, Gload of the colOpt. Express
nagement of LE
mitting diodes
Overington, M000). doi:10.11
osphor layer anmaximum pho
rating temperaatsink #2. Thed increases themperature.
ely extracting tnks depends onphosphor laye
nvestigating otayer temperatur
rch Center at Rds Lighting; anartin Overingto
e of remote pho
G., Fulmek, P.,lor conversion21 (S3), A4
EDs: Package
s,” J. Phys. Che
M., “Character117/12.382836
nd heatsink. Thosphor layer oature across the lower phosphe conversion ef
the heat and efn the geometryer temperature ther heatsink are.
Rensselaer Polnd the contribuon, and Jennife
osphor SSL de
, and Nicolics,n elements in p39-A449 (201
to system,” Pro
em. Lett. 2 (11
rizing white L6
he radial operating he entire hor layer fficiency
fficiently y and the but also
and light
lytechnic ution and er Taylor
evices by
J., “The phosphor 13). doi:
oc. SPIE
1), 1268-
LEDs for
Proc. of SPIE Vol. 8835 883504-6
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 10/03/2013 Terms of Use: http://spiedl.org/terms
[6] Hu, R., Luo, X., and Zheng, Z., “Hotspot location shift in the high-power phosphor-converted white light-emitting diode packages,” Jpn. J. Apply. Phys. 51, 09MK05-1-4 (2012). doi: 10.1143/JJAP.51.09MK05
[7] Luo, X., Fu, X., Chen, F., and Zheng, H., “Phosphor self-heating in phosphor converted light emitting diode packaging,” Int. J. Heat Mass Tran. 58 (1-2), 276-281 (2013). doi.org/10.1016/j.ijheatmasstransfer.2012.11.056
[8] Huang M., and Yang, L., “Heat generation by the phosphor layer of high-power white LED emitters,” IEEE Photonics Technol. Lett. 25 (4), 1317-1320 (2013). doi: 10.1109/LPT.2013.2263375
[9] Lowery, C., “Methods and apparatuses for enhancing heat dissipation from a light emitting device,” U.S. Patent 8 076 833 B2, Dec. 13, 2011.
Proc. of SPIE Vol. 8835 883504-7
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 10/03/2013 Terms of Use: http://spiedl.org/terms