SILICONE-PHOSPHOR ENCAPSULATION FOR HIGH POWER WHITE LEDS
Horatio Quinones, Brian Sawatzky, and Alec Babiarz
ASYMTEK
Carlsbad, CA, USA
Key word: LED, jetting, dispensing, YAG, CIE, volumetric
accuracy.
with internal light sensitive layers which converts the blue
wavelength to a variety of multiple wavelengths resulting in
a white appearance. Typical optical radiation pattern for a
packaged LED emission tends to be within an angle of
about 200 from the direction of maximum intensity. Intensity
of light is another parameter of importance, and has
dependencies on the integrity of the packaging of the LED
as well as on the volumetric accuracy of the fluid and its
composition homogeneity. Large discrepancies (up to 50%)
of photometric measurements of LED’s are common in the
industry. Various experiments were performed to address
potential parameters responsible for LED performance
variation. The CIE standard was used to quantify
differences.
EXPERIMENTAL WORK
Test Objective
The main purpose of the present study is to isolate possible
sources of variability associated with LED encapsulation
with Phosphor-Silicone mixture. Specifically we expect to
investigate the influence of phosphor loading, fluid volume,
and time-based behaviors on the photometric properties of
the devices. To determine the comparative performance
differences of various dispensing processes including
jetting[2] and traditional needle dispensing both, qualitative
and quantitatively. The test strategy consisted in the initial
characterization of high power (about 1 Watt) blue LEDs,
see figure 1.
B a re c h ip sp e c tru m
9.00E -03
8.00E -03
7.00E -03
6.00E -03
intensity
ABSTRACT
High-brightness Light Emitter Diodes (HB-LED) market
has grown at an average rate of about 43% since the
introduction of blue light wavelength die. HB-LEDs provide
for a range of applications such as architectural lighting,
signaling, entertainment lighting, automotive lights and
LCD backlighting. The flexibility in color-changing
properties, long life time, robustness, energy efficiency
(lumens/watt) makes this technology very attractive. There
are issues and challenges including the need to improve
electronic drive circuit technology, total lumen output,
“white light problem”, color quality and reproducibility and
cost efficiency. HB-LEDs are not a simple die, they need to
be packaged in a complex structure as to maximize the
effective intensity and prevent optical aberrations. HBLEDs come in many packages, from single chip to
sophisticated multi-directional aspheric lens designs. They
include lenses, colored materials, and diffusers (one or two
part silicone materials filled with phosphors) all of which
can alter the spatial and spectral distribution relative to the
basic light emitter die. Packages may include chips of
different size, different types and different locations.
Packages and chip locations may have different mechanical
tolerances. Process for dispensing the fluids that constitute a
large part of both, manufacturing and packaging of HBLEDs need to be precise reliable and cost effective. Jetting
technology offers a five fold faster process, and higher
precision than traditional needle dispensing. The paper
addresses some of the challenges in the LED packaging in
particular jetting processes that lead to tighter distribution of
the HB-LED color spectra (CIE, XYZ map) which in turn
reduces the binning of LEDs and reduces the packaging cost
of ownership.
5.00E -03
4.00E -03
3.00E -03
2.00E -03
1.00E -03
0.00E +00
-1.00E -03 0
200
400
600
800
1000
w avelength(nm )
INTRODUCTION
The introduction of blue LEDs as the base to obtain white
light by means of packaging compounds, i.e., phosphors
Yttrium-aluminum-garnet, YAG [1] and silicones/epoxies
presents several dispensing challenges. No longer can a pure
clear silicone compound with RGB source be used. Precise
volumetric accuracy of the phosphors needs is required to
assure minimum variation in the color. The white LEDs
addressed in this paper are those built from a blue source
Figure1. HB-Blue die wavelength spectrum
The HB-LED package for this study consisting of the wire
bonded die and fluid dispensed (two-part silicone matrix
and phosphor, YAG), and not including the optical lens is
evaluated for various optical characteristics. The metrology
process for evaluation consisted mainly on standard
photometry.
Experiment Cells
The bulk of the work reported here consisted of fluid
dispensing processes with various methods that included
traditional needle dispensing and jetting process. Dow
Corning two part silicone material doped with YAG
phosphor was used. The die used was a blue spectrum
emitter diode from ELITE, 1mm2 with mean wavelength of
460 nm (+/-2.5 nm); the die is wire bonded (1 mm high).
Figure 2 Depicts the blue HB-LED cavity and wire bonded
die prior to fluid dispensing prior to fluid dispensing and
dicing
Once the baseline, mass and silicone-to-phosphors ratio was
established, CIE XY and brightness were evaluated as
function of various other process parameters. The effect of
volumetric accuracy was examined using jet dispensing and
traditional time-pressure dispensing. Cell II experiments
were carried by dispensing with both needle and jet.
Phosphors content was kept constant at 9.1% per weight.
Mass was varied from 1.2mg to 2.5 mg and CIE
measurements were taken. Figure 4 depicts HB-LEDs with
various amounts of material dispensed.
2mg 10wt% phosphor
Figure2. HB-LED blue die showing the wire bonded die
and the cavity prior to silicone and YAG dispensing.
Optical and electrical measurements were performed using
the Labsphere System. Experiments were partitioned into
four cells. (I) To determine the appropriate mass mean value
dispensed on the LEDs and appropriate silicon to YAG ratio
as to meet specific CIE X and Y values by an iterative
approach. (II) To determine correlation of material volume
dispensed on the LED (Silicone and 9.1% by weight of
phosphor) to various optical properties of the packaged HBLED. (III) To determine correlation of phosphor/silicone
mix ratio to various optical characteristics of packaged HBLED. (IV) To determine correlation of material (silicone +
9.1% phosphor) pot life to various optical characteristics of
the packaged HB-LED. (V) To determine the spread for
time-pressure needle dispense and jetting for the optimum
volume and silicon-to-phosphors ratio.
2.25mg 10wt% phosphor
Figure4. HB-LED with various volume of silicone-YAG
dispensed.
As the volume dispensed was increased so did the CIE
X&Y values. For the case of jetting the correlation can be
better established since its variation was tighter as depicted
in figure 5. For the needle dispense (time-pressure), figure 6
the CIE XY to mass correlation although has similar trend
as jetting, its correlation is not as high. This can be
attributed to larger variation of the mass dispensed by
needle.
Je ttin g : C IE Sp e ctr a vs M as s
0.9
5 20
5 30
0.8
54 0
510
0.7
5 50
56 0
0.6
Experimental Results
For the first cell, optimization of volume and phosphors
content for the HB-LED, it was determined that a mass of
about 1.75mg with a 9.1% phosphors content will give the
best results, i.e., acceptable brightness (>30 lm/W) and
white color emission. We can see in figure 3 (a) the process
of volume given 9.1% phosphors mixture, (b) the process of
mixture ratio given a 2mg shot.
C IE v s M a s s fo r 9 . 1 % Ph o s p h o rs
0.5
C IE v a lu e
C IE V a lu e
0.3
0.2
0.1
1
1.5
2
2.5
3
0.5
0.4
0.3
0.2
0.1
0
10
20
30
Phosphor wt%
Mass (m g)
Cx
Cy
target
Cx
Cy
5 70
5 00
0.5
58 0
59 0
0.4
6 00
2.5 mg
2.0mg
17.5 mg
0.3 4 90
6 10
6 20
83 0
1.5 mg
0.2
4 80
0.1
47 0
46 0
0
0
0.1
3 60
0.2
0.3
0.4
0 .5
0 .6
0 .7
0 .8
Cx
Figure5. CIE X-Y values for various volumes dispensed,
by jetting process, 9.1% phosphors per weight.
CIE vs. Phosphor wt%
0.4
2. 5mg 10wt% phosphor
40
50
Target
(a)
(b)
Figure3. CIE target and results from silicone volume (a)
and doping ratio (b) optimization weight of phosphors.
For this experiment the target was X=Y=0.3nm on the CIE
map. This target was met by using the mixture of 1.75gm of
silicone two part material and 9.1% YAG for each HB-LED.
Results from Cell II clearly show trends that are directly
related to the spread of the CIE X-Y values and that
influence the number of bins for a given LEDs class. Cell
III experiments were performed to quantitatively determine
the influence of silicone-to-phosphors ratio and CIE
parameters. Blue light emitter die depend on the phosphors
conglomerates to produce white light spectrum.
path of the fluid flow will dictate the phosphors-to-silicone
ratio being dispensed and thereby determining the CIE XY
values. Viscosity of the silicone has a direct correlation with
the time for this settling to occur, lower viscosities causes
faster settling. At the early stages of dispensing a good
mixture ratio yields good white color, however over time
the phosphors settling in various places (depending on the
fluid path geometry,) yields corresponding variations in the
color map. Eventually, the phosphors settling, either
dispenses (yielding yellow rich spectra), or will permanently
reside on stagnation zones and low phosphors-to-silicon
ratio will be dispensed resulting in a blue spectra LED .This
time dependence can be depicted in figure 8.
Ne e e d le : C IE Sp e ctr a vs M as s
0.9
520
530
0.8
540
510
0.7
550
560
0.6
570
500
0.5
2.5 mg
2.0mg
580
590
17.5 mg
0.4
600
610
620
0.3 490
830
1.5 mg
0.2
480
0.1
470
460
0
0
360
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Cx
Figure6. CIE X-Y values for various volumes dispensed,
with needle, time-pressure process, 9.1% phosphors per
weight.
Je tting : CIE v s Tim e
2.0m g, 9.1%P hosphpors
0.4
0.35
Hence, the distribution and amount of phosphors in the
silicone mixture is a very important parameter to control.
Scarcity of the phosphors results in blue wavelength
emission, excessive amounts of the phosphors although may
induce higher brightness, it also carries a yellowish color
emission. Figure 7 shows direct correlation of phosphors
content to the CIE chromaticity. The range for phosphorsto-silicone mix varied from 5% to 90% per weight.
530
540
510
0.7
0.1
0.05
0
0
0.1
0.2
0.3
0.4
Cx
1s t
2nd
3rd
4th
5th
6th
7th
8th
9th
10th
11th
Above plot can be incorporated in the CIE map and thereby
obtaining the corresponding variation and bin distribution.
Figure 9 shows this mapping.
520
0.8
0.2
0.15
Figure8. CX, CY map for different time after initial mixing
of YAG and silicone.
Jetting 2.0mg,5,10, 15,20,25,30,40,50,60,70,%Phosphors:
CIE Spectra
0.9
0.3
0.25
550
560
0.6
500
50%
40%
30%
25%
20%
0.5
0.4
570
70%
Jetting 2.0mg,9.1%Phosphorsvs Time: CIE Spectra
580
590
600
15%
10%
0.3 490
610
520
0.9
620
830
530
0.8
0.2
5%
480
540
510
0.1
0.7
470
460
0
0
0.1
550
560
360
0.6
0.2
0.3
0.4
0.5
0.6
0.7
0.8
570
500
Cx
0.5
580
590
0.4
Figure7. Plot of the CIE vs phosphors content in the
silicone matrix base, total mass dispensed was 2.0mg.
600
610
620
0.3 490
830
0.2
Clearly phosphor content ratio to silicone cause a large
variation in the CIE XY values, thereby increasing the
number of bins for the same die.
Cell IV experiments consisted of examining the
phenomenon of phosphors settling in the silicone mixture
over time and its influence on HB-LEDs optical
characteristics. Given the properties of the mix, the
phosphors conglomerates have negative buoyancy in the
silicone fluid and settling occurs. For a dispensing system
the presence of stagnation zones and their locations in the
480
0.1
470
460
0
0
0.1
360
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Cx
Figure9. CIE diagram of chromaticity as function of
phosphors settling in the silicone over time.
Cell V addresses various dispensing processes namely
needle and jetting. Determination of correlations often
requires precise dispensing systems capable of resolving
small variations. This was accomplished by jetting
processes since it was determined that the volume accuracy
was better than that of needle. A targeted of 2.0mg of
silicone-YAG mixed (9.1% phosphors per weight was
dispensed on HB-LEDs. For the needle a time-pressure
valve was used. For the jetting a DJ-9000 DispenseJet from
Asymtek was used. Figure 10 shows the results; there one
can observe a larger spread in the needle dispensing process.
This is inherently a problem facing time-pressure dispensing
today, and will most likely worsen for small volumes.
Jetting CIE for 2.0mg 9.1%Phosphors
0.9
Needle CIE for 2mg 9.1%Phosphors
0.9
520
530
0.8
520
530
0.8
540
510
540
510
0.7
0.7
550
560
0.6
550
560
0.6
570
500
0.5
590
0.4
600
610
620
0.3 490
580
Cy
Cy
590
0.4
570
500
0.5
580
600
610
620
0.3 490
830
830
0.2
0.2
480
480
0.1
0.1
470
460
360
0
0
0.1
470
460
360
0
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0
0.1
0.2
0.3
Cx
0.4
0.5
0.6
0.7
0.8
Cx
(a)
(b)
Figure10. (a) CIE map for jetting dispense of silicone and
YAG mix. (b) CIE map corresponding to traditional timepressure needle dispense.
The brightness of the HB-LED and its wavelength Spectrum
are plotted in figure 11. Jetting yielded high brightness.
Jetting (PK2) spectrum
lm/W=33.06
1.60E-03
1.40E-03
intensity
1.20E-03
1.00E-03
8.00E-04
6.00E-04
4.00E-04
2.00E-04
0.00E+00
0
200
400
600
jetting process exceeded target. Jetting materials for LED
packages provides a dispensing method contact free and
volumetrically consistent. Jetting
being exercised at
dispense gaps noticeably higher than those used by the
traditional needle dispensing, makes it surface topology free
as well as damage proof on the parts. Asymtek jetting valve
DJ9000 is a proven technology: used extensively in the
underfill process and optoelectronics including several LED
fluid dispensing. Jetting shows quality improvement over
the traditional needle dispensing. Jetting process throughput
about 3X to 8X higher than the needle process.
800
1000
wavelength(nm)
Figure11. LED wavelength spectrum after encapsulation
During the HB-LED dispensing process it was observed that
the needed z-motion retraction for traditional needle
dispensing resulted in longer dispensing times compare to
jetting. Experiments simulating production runs showed a
difference of about 5X, the jet having higher throughput.
CONCLUSIONS
The CIE chromaticity spread is very much dependent on the
quality of the dispensing process. The volumetric accuracy
is directly related to the spread of the CIE map, i.e., the
binning of LEDs. Phosphors doping variation yields spread
of the CIE XY values and increased binning. Phosphor
settling occurs over time. This settling results in variations
of the silicone-to-phosphors ratio and thereby increasing
binning as well. Jetting dispensing will result in less binning
compared to that of the needle. The HB-LED brightness for
ACKNOWLEDGMENT
The authors would like to thank the ITRI center for their
metrology and suggestions on this work. We also would like
to extend our thanks to Mr. G. Gibbs for his support and
guidance in completing this work. Last we would like to
thank the various personnel from the Asymtek Taiwan Lab
that performed the experimental work, Mr. C. Chen, B. Lin,
J. Lai and P. Lee.
REFERENCES
[1] Japanese Patent Application No. 10-194156 (194156,
1998) filed on Jul. 9, 1998, Japanese Patent Application No.
10-316169(316169/1998) filed on Nov. 6, 1998 and
Japanese Patent Application No. 10-321605(321605/1998)
filed on Nov. 12, 1998.
[2] H. Quinones, A. Babiarz, L. Fang (EMAP, Taiwan,
October 2002) Jetting Technology: A Way of the Future in
Dispensing.