allen steven c
TRANSCRIPT
UNIVERSITY OF CINCINNATI Date:___________________
I, _________________________________________________________, hereby submit this work as part of the requirements for the degree of:
in:
It is entitled:
This work and its defense approved by:
Chair: _______________________________ _______________________________ _______________________________ _______________________________ _______________________________
Illumination for the 21st Century:
High Efficiency Phosphor-Converted Light-Emitting Diodes for Solid-State Lighting
A dissertation submitted to the Division of Research and Advanced Studies
of the University of Cincinnati
in partial fulfillment of the requirements for the degree of
DOCTORATE OF PHILOSOPHY (Ph. D.)
In the Department of Electrical & Computer Engineering of the College of Engineering
2007
by
Steven Christopher Allen
M.S.,Princeton University, 2003
B.S., Ohio University 2001
Dr. Andrew J. Steckl, Committee Chair
iii
Abstract
Phosphor-converted light-emitting diode (pcLED) luminaires have been designed
and built that address the loss mechanisms present in commercial and literature pcLEDs.
Under dc drive conditions, a white pcLED with an organic phosphor achieved 97 lm/W
luminous efficacy and a package efficiency of 0.99, a white pcLED with YAG:Ce
phosphor achieved 86 lm/W efficacy and a package efficiency of 0.96, and a greenish-
yellow pcLED with an organic phosphor achieved 151 lm/W efficacy and a package
efficiency of 0.96. The results were obtained with a blue pump LED chip having an
efficiency of 0.36.
Organic green and red color-conversion materials for use with a blue LED
backlight for full-color emissive display applications have been demonstrated. The color
converters closely match the NTSC color gamut, have quantum efficiencies exceeding
90%, and are sufficiently stable to withstand a minimum 7,000 hours of operation at 300
cd/m2.
A phosphor composite material proposed for future pcLEDs will increase the
effective quantum efficiency of the phosphor by eliminating diffuse scattering to allow
virtually 100% light extraction. The composite material may also be useful as a gain
medium in solid-state lasers and as a fluorescence collector in solar energy applications.
iv
v
Acknowledgments and Dedication
I would like to thank my parents, Juan and Susan Allen for their endless support,
encouragement, and understanding over my 23 years of academic training culminating in
this dissertation.
My wife, Keiko Ishikawa is appreciated for joining me in investing in years of
poor graduate student life in the hope of a better future.
My advisor, Dr. Andrew J. Steckl deserves credit for persuading me to enter the
Ph. D. program at the University of Cincinnati at a time when I thought my academic
career was finished. He provided me the right balance of guidance and freedom
necessary to produce this work that was unrelated to other group projects.
Dedicated to my loving grandfather, Benjamin Allen, who passed away as this
manuscript was in preparation. He was always so proud of whatever his two grandsons
accomplished. He will not be forgotten.
vi
Table of Contents
Abstract.......................................................................................................................... iii Copyright Notice............................................................................................................ iv Acknowledgments and Dedication ................................................................................. v Table of Contents........................................................................................................... vi List of Tables and Figures ........................................................................................... viii
Chapter 1. Lighting Technologies and Roadmap ......................................................... 1
1.1 Introduction............................................................................................................... 1 1.2 Definition of lighting terms ...................................................................................... 2 1.3 Solid-state lighting roadmap..................................................................................... 4 1.4 Generating white light from LEDs ........................................................................... 8
(a) Color mixing.......................................................................................................... 9 (b) Full wavelength conversion ................................................................................ 10 (c) Partial wavelength conversion............................................................................. 11 (d) Chip, phosphor, and package efficiency targets.................................................. 11
1.5 White phosphor-converted LED efficiencies ......................................................... 14 Chapter 2. Materials: LEDs, Dyes, Polymers, and Phosphors ................................. 18
2.1 Inorganic light sources: light emitting diodes......................................................... 18 2.2 Organic dyes ........................................................................................................... 20 2.3 Transparent polymers ............................................................................................. 23 2.4 Hybrid inorganic-organic........................................................................................ 25 2.5 Inorganic phosphors................................................................................................ 27
Chapter 3. Luminaires for Solid-State Lighting ......................................................... 29
3.1 Planar luminaire...................................................................................................... 29 3.2 Cylindrical luminaire .............................................................................................. 31 3.3 Spherical luminaire ................................................................................................. 34
(a) Whispering gallery modes ................................................................................... 41 3.4 Introduction to the ELiXIR Luminaire ................................................................... 44 3.5 Fabrication of the ELiXIR luminaire...................................................................... 49
(a) First generation ELiXIR luminaire...................................................................... 49 (b) Second generation ELiXIR luminaire ................................................................. 52
3.6 ELiXIR luminaire characterization........................................................................ 59 3.7 ELiXIR luminaire results........................................................................................ 59
(a) Blue LED chip ..................................................................................................... 59 (b) YAG:Ce white..................................................................................................... 61 (c) Organic greenish yellow...................................................................................... 62 (d) Organic white ...................................................................................................... 65
3.8 Comparing ELiXIR luminaires with literature examples ....................................... 67 3.9 Conclusion .............................................................................................................. 67
vii
Chapter 4. Color, Efficiency, and Stability of Fluorescent Dyes in Polymer Matrices........................................................................................................................................... 70
4.1 Fabrication of Phosphor Films................................................................................ 70 4.2 Color ....................................................................................................................... 71 4.3 Photostability .......................................................................................................... 76
(a) Background.......................................................................................................... 76 (b) Experiment .......................................................................................................... 78 (c) Detailed Theory ................................................................................................... 79
4.4 Photostability Results ............................................................................................. 84 4.5 Display Application ................................................................................................ 85 4.6 Conclusion .............................................................................................................. 86
Chapter 5. Nearly Index-Matched Luminescent Glass-Crystal Composites ........... 87
5.1 Motivation: pcLED application .............................................................................. 87 5.2 Solid-state laser application .................................................................................... 93 5.3 Fluorescence collection application........................................................................ 95
Chapter 6. Obtaining Accurate Spectra and Power Measurements ......................... 96 Chapter 7: Conclusion/Summary ................................................................................ 102
References................................................................................................................... 104
viii
List of Tables and Figures Table 1. Summary of results for the Cree EZR1000 LED chip at 30 mA dc. ................. 60 Table 2. Summary of results for the white luminaire with YAG:Ce phosphor. .............. 61 Table 3. Summary of results for the yellowish-green luminaire ..................................... 63 Table 4. Summary of results for the white luminaire with orange organic phosphor ..... 66 Table 5. Comparison of ELiXIR luminaire efficiency and flux values to similar literature
and best production LEDs at 400 mA DC drive current........................................... 67 Table 6. Summary of dye characteristics for display applications. From [73]. .............. 86 Figure 1. Normalized eye responsivity and lumens per watt conversion versus
wavelength across the visible region. ......................................................................... 4 Figure 2. Solid-state lighting roadmap targets for years 2002 to 2020. Included for
comparison are typical incandescent and fluorescent values. From [2]. ................... 6 Figure 3. Photographs of an incandescent lamp (top left) [8], fluorescent lamps (top
right) [9], the Cree 7090 XR-E LED (bottom left) [7], and Luxeon K2 emitter LED (bottom right) [6], [10]................................................................................................ 7
Figure 4. The three approaches to producing white light from LEDs. Modified from [11]...................................................................................................................................... 8
Figure 5. The maximum CRI and luminous efficiency of various numbers of LEDs for a color temperature of 4870 K. From [14].................................................................... 9
Figure 6. Breakdown of efficiency targets for generation of white light through three methods: wavelength conversion, color mixing, and hybrid (LED + phosphor). From [2]. ................................................................................................................... 12
Figure 7. Maximum luminous efficacy versus correlated color temperature for a combination of 3 LEDs. From [2]............................................................................ 14
Figure 8. Diagram of conventional phosphor-converted LED. ....................................... 15 Figure 9. Electron micrographs of conventional pcLEDs: LED chip with conformal
phosphor coating (top) and slurry-deposited phosphor coating (bottom). From [23]. The conformal phosphor coating has a more uniform spectrum versus angle. ........ 16
Figure 10. Diagram of scattered photon extraction white pcLED. From [25]................ 17 Figure 11. External quantum efficiency versus peak wavelength for state of the art LEDs
at 350 mA drive current at a junction temperature of 25° C. V(λ) is the sensitivity of the human eye. From reference [29]. ....................................................................... 19
Figure 12. Radiant flux and external quantum efficiency versus dominant wavelength for Cree InGaN multiple quantum well LEDs. From reference [28]. ........................... 20
Figure 13. Emission spectra of Keystone Yellow, Lumogen Yellow, Lumogen Orange, Rhodamine 590, and Lumogen Red. A blue LED was the excitation source.......... 22
Figure 14. Structure of perylene, Lumogen Yellow, Lumogen Orange and Lumogen Red, and Rhodamine 590 perchlorate. .............................................................................. 22
Figure 15. Diagram showing spin-allowed fluorescence and spin-disallowed phosphorescence. ...................................................................................................... 23
Figure 16. Structure of poly-methyl methacrylate (PMMA) (top) and poly-dimethylsiloxane (PDMS) (bottom). The monomer of each is in parentheses........ 25
Figure 17. Three semi-transparent organic phosphors consisting of small molecule dyes dispersed in a modified acrylic. ................................................................................ 26
ix
Figure 18. Semitransparent organic phosphor (left) and diffuse YAG:Ce in a transparent silicone matrix. The films have similar optical density in the blue. ........................ 28
Figure 19. Diagram of planar luminaire configuration. The phosphor consists of a fluorescent dye in a polymer matrix and the waveguide a transparent sheet of glass or acrylic. From [45]. ............................................................................................... 29
Figure 20. Schematic of the cylindrical lamp, showing the three primary components: pump LED, acrylic rod waveguide, and fluorescent coating.................................... 32
Figure 21. Photograph of rod luminaires in operation. From left to right: violet (no color conversion), blue (Lumogen Violet), green (Lumogen Yellow), red (Lumogen Red), and white (Lumogen Violet and Orange) luminaires. From [49]. ........................... 33
Figure 22. Spectrum of the white rod luminaire. A 7500 K blackbody spectrum is provided for comparison. From [49]........................................................................ 33
Figure 23. Structure of the spherical luminaire. .............................................................. 34 Figure 24. Photographs of the spherical luminaire in operation. A close-up of the device
under ambient lighting (left) and illumination of a CIE diagram in an otherwise dark room (right). .............................................................................................................. 35
Figure 25. Schematic of spherical luminaire power measurement experiment. .............. 36 Figure 26. Typical result of the spherical lamp power efficiency measurement. As the
measurement distance varied, error in the distance measurement resulted in less than 5% variation in lamp power. ..................................................................................... 37
Figure 27. Intensity uniformity of the spherical lamp. The four positions of the equatorial and meridional measurements were in 90 degree intervals. .................... 38
Figure 28. Series of spectra from a spherical luminaire. Luminous efficiencies, correlated color temperatures, and color rendering indices are listed. The initial spectrum is at the bottom; progressive addition of dye solutions to adjust color resulted in the upper spectra. .................................................................................... 40
Figure 29. Schematic of a whispering gallery mode. A ray is emitted from a point source at radius r, at a sufficiently shallow angle to the surface and suffers total internal reflection. Rays such as these have long path lengths inside the device and are eventually lost to re-absorption................................................................................. 41
Figure 30. Diagram of a dielectric sphere with refractive index 1.5 surrounded by air (top left). Transmittance at first encounter of external surface versus emission angle from various r/R emitter positions (top right). Table of first incidence transmittance (T) integrated over the 4π emission solid angle for select values of r/R (bottom). .. 43
Figure 31. Conventional phosphor converted LED (left) and ELiXIR phosphor converted LED (right). From [50]. ........................................................................................... 45
Figure 32. Ray trace diagram of phosphor emission in the ELiXIR luminaire. From [50].................................................................................................................................... 46
Figure 33. Two dimensional schematic of ELiXIR lens with index-matching interior phosphor coating, where r is the internal radius of the lens where the phosphor is located, and R is the external radius of the lens. The refractive indices are such that n2 > n1 and Ө is the largest angle at which phosphor emission strikes the external lens surface................................................................................................................ 47
Figure 34. Maximum r/R values for lens/phosphor refractive indices from 1.3 to 2.0 when n1 = 1. .............................................................................................................. 48
x
Figure 35. Photographs of ELiXIR lens at various stages of fabrication. First, the PMMA is cast around a flask in an aluminum mold (top left); flask and lens assembly after removal from mold (top right); sawing through the lens and flask to obtain hemispherical shell (middle left); separated parts: lens on left, waste material on right (middle right); finally, the hemispherical shell lens after polishing (bottom).................................................................................................................................... 51
Figure 36. Spectra and photograph of first ELiXIR lamp fabricated. Blue LED emission, phosphor absorption, and ELiXIR luminaire emission spectra. Photograph of the luminaire in operation (inset). This luminaire achieved a package efficiency of only 0.78, limited by a 30% absorbing reflector and imperfect LED to lamp coupling (note the blue emission escaping from the edges of the reflector sheet). From [50].................................................................................................................................... 52
Figure 37. Mechanical drawing of chip mount fabricated from aluminum..................... 56 Figure 38. Mechanical drawing of lens base. Note the hole in center and screw holes to
position and secure the chip mount........................................................................... 57 Figure 39. Schematic of entire luminaire assembly including the chip mount, lens base,
and lens. .................................................................................................................... 58 Figure 40. Photograph of two second-generation ELiXIR luminaires. A white-emitting
luminaire having a YAG:Ce in silicone phosphor coating is on the left, while a green-emitting luminaire having a Lumogen Yellow in Joncryl 587 polymer phosphor coating is on the right................................................................................ 58
Figure 41. Cree EZR1000 LED spectrum........................................................................ 60 Figure 42. Spectrum of white luminaire with YAG:Ce phosphor. .................................. 61 Figure 43. Yellowish green luminaire spectrum (Lumogen Yellow phosphor). Notice the
small amount of blue LED leakage around 460 nm. ................................................ 63 Figure 44. Spectrum for the white luminaire with organic orange phosphor. ................. 66 Figure 45. Photograph of the organic phosphor white luminaire illuminating a CIE color
chart and UC logo. .................................................................................................... 69 Figure 46. CIE diagram showing color coordinates of the phosphors: Keyplast yellow
(A), Lumogen Yellow (B), Lumogen Orange (C), Rhodamine 6G (D), Lumogen Red (E), and a hypothetical monochromatic 460 nm source (F). The NTSC color gamut is indicated by the dashed triangle. ................................................................ 72
Figure 47. Photograph of phosphor-coated microscope slides under illumination with a long-wave ultraviolet lamp. From left to right, Keystone Yellow, Lumogen Yellow, Lumogen Orange, Rhodamine 6G, and Lumogen Red............................................. 72
Figure 48. Experimental setup for quantum efficiency measurement. ............................ 74 Figure 49. Ray trace diagram of an isotropic emitter inside a planar waveguide. Only
rays falling inside the cone defined by the critical angle escape the waveguide. ..... 75 Figure 50. Experimental setup for determining dye photostability. ................................ 79 Figure 51. Calculated fluorescence intensity versus time for three beam profiles (top) and
beam profiles (bottom). The beam profiles are normalized such that the intensity integrated over a circular beam of radius R are equal. Note the decay rate for the Gaussian profile is more than double that of the other two because higher discrepancy between the maximum and minimum intensities results in a higher power density factor.................................................................................................. 82
xi
Figure 52. Photostability results for several different samples and measurement conditions for fluorescent dyes Keyplast Yellow (KY), Lumogen Yellow (LY), Lumogen Orange (LO), Rhodamine 6G (R590), and Lumogen Red (LR) versus peak emission wavelength................................................................................................. 84
Figure 53. Schematic of color-by-blue display. From [73]. ........................................... 86 Figure 54. Diagram of glass-crystal composite structure. The material consists of a
crystalline inorganic powder phosphor in a transparent glass matrix....................... 88 Figure 55. Photograph of YAG:Ce / borosilicate glass LGCC powder in a vial (left).
Formation of glassy LGCC globules after firing the powder at 900°C for one hour in air (right). .................................................................................................................. 89
Figure 56. Scanning electron micrograph of a GCC consisting of 30% La2Zr2O7 crystals in soda borosilicate glass formed by hot pressing at 620° C for 1 hour. From [76].89
Figure 57. Photoluminescence spectrum of YAG:Ce in borosilicate glass LGCC. Excitation was at 325 nm. Photograph showing LGCC with a spot under excitation.................................................................................................................................... 90
Figure 58. Microscopic (top) and macroscopic property (bottom) comparison of glass-composites where the refractive indices of the two components differs significantly (left), and are closely matched (right)....................................................................... 91
Figure 59. Refractive index mismatch between YAG crystal and a mixture of 80% LAH60 and 20% LAH65 Ohara high-index glasses................................................. 91
Figure 60. Abbe diagram for glasses with refractive index greater than 1.6 manufactured by Schott. From [77]. ............................................................................................... 92
Figure 61. Schematic comparison of the most common solid-state visible laser, frequency doubled Nd:YAG (top) and the proposed solid-state visible laser using an NIMLGCC as a gain medium (bottom). ................................................................... 93
Figure 62. Schematic of fluorescence collector operation............................................... 95 Figure 63. Comparison of known spectrum of calibrated Ocean Optics LS-1 CAL
incandescent lamp (L0(λ)) in arbitrary power units (dashed line) and lamp spectrum obtained with an Ocean Optics USB2000 fiber optic spectrometer before any calibration (solid line). .............................................................................................. 97
Figure 64. Response of Newport power meter and Ocean Optics fiber spectrometer to monochromatic light across the visible spectrum. The correction spectrum is the factor the spectrometer response must be amplified by in order to give a calibrated spectrum in units of optical power per unit area per unit wavelength interval. Power conversion correction necessary to convert spectrum from USB2000 to calibrated optical power units for data in Figure 63. ................................................................. 98
Figure 65. L0(λ)C(λ) versus wavelength.......................................................................... 99 Figure 66. Factory-calibrated intensity data for the LS-1-CAL incandescent lamp (open
squares) and the lamp spectrum obtained with the Ocean Optics spectrometer after calibration with the Newport power meter and variable monochromatic source (solid line). ........................................................................................................................ 100
1
Chapter 1. Lighting Technologies and Roadmap
Introduction
White light generation by light-emitting diodes (LEDs) is a rapidly advancing
technology with a goal of reaching 50% efficiency or 200 lumen per watt efficacy.
Currently, all commercial white LEDs are phosphor-converted LEDs (pcLEDs),
consisting of a blue LED partially converted to yellow by a phosphor because this
method is most efficient and cost-effective with current technology. However,
commercial pcLED packages and others in the literature suffer significant losses due to a
combination of the following factors: (1) close proximity of the LED and phosphor result
in reflection of LED-emitted light and emission of phosphor-converted light into the
lossy LED chip; (2) quantum conversion and re-absorption losses due to phosphor scatter
inside the phosphor layer; and (3) trapping phosphor-converted light inside the device by
total internal reflection. The sum of these losses can easily exceed 50%.
Solid-state lighting is can be defined as illumination by use of solid-state
materials, namely by using a combination of semiconductor chips as light generators and
phosphors as light converters. This is in contrast to evacuated bulbs used in incandescent
lamps and low-pressure gas tubes in fluorescent lamps.
The motivation for developing solid-state lighting is the promise of improved
efficiency over other lighting technologies, which have saturated at 1-25% efficiency [1].
Reaching the goal of 200 lm/W luminous efficacy or 50% power efficiency would have
significant benefits. In the United States, approximately 20% of all electricity generated
is consumed by lighting. LEDs have the potential to more than double current lighting
efficiency, thus reducing total power consumption by 10% for a dollar savings of $25
2
billion. Because of the reliance on coal and gas-fired power plants, carbon emissions
would be reduced by 25 megatons per year [2].
Definition of lighting terms
Color temperature is strictly defined for blackbody sources only, and is simply
the temperature of a perfect blackbody. The spectrum of a perfect blackbody is given by
the Planck radiation law and depends only on its temperature:
1e1hc2)T,(I kT/hc5
2
−= λλ
λ , Equation 1
where I is the emitted radiation intensity in power per unit area per unit wavelength
interval, λ is the emission wavelength, T is the blackbody temperature, h is Planck’s
constant, c the speed of light, and k is Boltzmann’s constant. Blackbodies having a
temperature around 2000 K appear red to the eye; an example of this is the filament
inside a toaster. Incandescent lamp filaments have a temperature of ~ 2700 K, and appear
white with a strong orange component. The sun has a surface temperature of 5500-6000
K, and closely resembles a 5500 K blackbody at mid-day on the surface of the earth.
Devices such as fluorescent lamps and LEDs that produce white light through
means other than blackbody emission do not in general have spectra identical to
blackbodies, so do not have a color temperature. However, correlated color temperature
(CCT) is an approximate measure of how a given spectrum is perceived by the eye. It is
defined as “the temperature of the Planckian radiator whose perceived color most closely
resembles that of a given stimulus at the same brightness and under specified viewing
conditions [3]. It can be calculated using software provided with the Commission
Internationale de L’Eclairage (CIE) technical report 13.3 [4].
3
The color rendering index (CRI) is 100 point scale which measures how well a
given light source portrays a standard set of colors compared to the blackbody spectrum
of the same color temperature. A CRI of 80 is generally considered excellent for general
lighting applications. The combination of a blue LED with the YAG:Ce phosphor has a
maximum CRI of approximately 75 at a ~5500 K CCT. An incandescent lamp, being a
real blackbody source, has a CRI of near 100. Fluorescent lamp CRI can vary based on
the specific phosphor combination, but a value of 75 is typical. CRI is calculated from a
spectrum using the same program as the CCT [4].
The lumen is a measure of the power of visible light normalized to human eye
sensitivity. It is defined as 1.464 mW at 555 nm, where the human eye has maximum
sensitivity [5]. Hence, at all other wavelengths, more than 1.464 mW is power is
required to produce one lumen of light. The eye responsivity is shown in Figure 1.
The luminous efficacy of a light source, given in lumens per watt (lm/W) of input
power, is the figure of merit for white light generation efficiency. Blackbody sources are
limited to relatively low luminous efficacies because their strong infrared emission is not
perceived as visible light by the eye. Fluorescent lamps and LEDs have higher luminous
efficacy potential because all of their emission can be confined to the visible region.
4
0
0.2
0.4
0.6
0.8
1
0
100
200
300
400
500
600
700
350 400 450 500 550 600 650 700 750
Nor
mal
ized
eye
resp
onsi
vity
Lumens per w
att
Wavelength (nm)
Figure 1. Normalized eye responsivity and lumens per watt conversion versus wavelength across the visible region.
Solid-state lighting roadmap
The Optoelectronics Industry Development Association (OIDA) has produced a
comprehensive roadmap which outlines advances in various technologies necessary to
reach superior LED-based lighting by 2020 [2]. It includes milestones for features such
as luminous efficiency, lifetime, and flux for years 2007, 2012, and 2020.
Figure 2 lists all the milestones in detail. Luminous efficiency, measured in units
of lumens per watt (lm/W) has a goal of 200 lm/W by 2020 for solid-state lighting (SSL).
Incandescent and fluorescent lighting typically have luminous efficiencies of 15-20, and
goal of 200 lm/W by 2020 for solid-state lighting. Incandescent and fluorescent lighting
typically have luminous efficiencies of 15-20, and 80-100 lm/W, respectively. The
lifetime goal for SSL is to exceed 100 kilohours (khr) at 50% of initial intensity. Unlike
incandescent and fluorescent lamps, LEDs generally do not “burn out,” but output
5
decreases gradually with time. Incandescent and fluorescent lamps have typical lifetimes
of 1 and 10 khr, respectively, at which point they cease emitting light entirely. The 2007
roadmap target for efficiency is 75 lm/W and >20 khr lifetime. Today (early 2007),
efficiencies fall well short of the target. For high flux single chip commercial white
LEDs, luminous efficacies of 50 lm/W from a Luxeon K2 [6] and 69 lm/W from a Cree
7090 XR-E [7] have been achieved at 350 mA and a junction temperature of 25° C.
However, pulsed drive currents with a small (1%) duty cycle is the only practical way to
keep the LED junction temperature at 25° C. In a more realistic application, the LEDs
would be driven at 350 mA dc drive current, resulting in much higher junction
temperatures (150° C is typical) and much lower efficiency. Therefore, the luminous
efficacy numbers listed in manufacturer data sheets greatly exaggerate the efficacy of the
installed lamp. Lifetime of these products is well in excess of the roadmap
recommendation, as both claim greater than 70% initial intensity after 50 khr of operation.
Typical light output, measured in lumens per lamp (lm/lamp) is typically 1000-
1500 lm for an incandescent bulb having an input power of 60-100 W. For large
fluorescent tubes, 3400 lumens and a 40 W input is common. Today, light output is 140
lm at 1500 mA from the Luxeon K2 [6], and 176 lm at 1000 mA from the Cree 7090 XR-
E [7]. Again, these flux numbers are also at an impracticably low 25° C junction
temperature. Even so, both fall short of the 200 lm/lamp roadmap target in 2007. By
2020, 1500 lumens are projected from each LED lamp in order to match light output from
a 100 W incandescent lamp. At the projected 200 lm/W efficiency, LED input power
would be 7.5 W. Therefore, the projected LED luminaire in 2020 will provide the same
amount of light as a 100 watt incandescent lamp, but with 92% less power!
6
Technology SSL-LED 2002
SSL-LED 2007
SSL-LED 2012
SSL-LED 2020
Incan-descent
Fluorescent
Luminous efficacy (lm/W)
25 75 150 200 16 85
Lifetime (khr) 20 >20 >100 >100 1 10 Flux (lm/lamp) 25 200 1000 1500 1200 3400 Input power (W/lamp)
1 2.7 6.7 7.5 75 40
Lumens cost ($/klm)
200 20 <5 <2 0.4 1.5
Lamp cost ($/lamp) 5 4 <5 <3 0.5 5 Color rendering (CRI)
75 80 >80 >80 95 75
Lighting markets penetrated
Low-flux Incandescent Fluorescent All
Figure 2. Solid-state lighting roadmap targets for years 2002 to 2020. Included for comparison are typical incandescent and fluorescent values. From [2].
In summary, an LED lamp reaching the 2020 roadmap recommendation of 200
lm/W efficiency will have ~12X the efficiency of incandescent lamps, and over 2X the
efficiency of fluorescent lamps, thus yielding significant power savings. Over the 100
khr LED lifetime, an incandescent bulb would be replaced 100 times and a fluorescent
lamp would be replaced 10 times, thus yielding savings in total lamp costs and
maintenance costs.
7
Figure 3. Photographs of an incandescent lamp (top left) [8], fluorescent lamps (top right) [9], the Cree 7090 XR-E LED (bottom left) [7], and Luxeon K2 emitter LED (bottom right) [6], [10].
8
Figure 4. The three approaches to producing white light from LEDs. Modified from [11].
Generating white light from LEDs
Because LEDs have emission bandwidths of ~ 20 nm and the visible wavelength
region spans approximately 300 nm, the production of good quality white light from
LEDs is not trivial [12]. There are essentially three approaches to solving the problem
[13], color mixing, complete phosphor conversion, and partial phosphor conversion,
outlined schematically in Figure 4.
Color Mixing
Partial Conversion
9
Figure 5. The maximum CRI and luminous efficiency of various numbers of LEDs for a color temperature of 4870 K. From [14].
Color mixing
Color mixing involves the mixing of different color LEDs to achieve white light.
The advantage of color mixing is there are no phosphor conversion problems, such as
inconsistent color versus emission angle, phosphor quantum efficiency losses, Stokes
conversion loss, and package efficiency losses from light being reflected back into the
LED or being total internal reflected and absorbed in the phosphor itself. Typically 3
LEDs are used, as this is the minimum number required to achieve a good CRI. Figure 5
10
shows that using 3 LEDs of 25 nm bandwidth, CRIs approaching 90 are possible,
however, if only two LEDs are used, the maximum achievable CRI is an unacceptably
low 20. Figure 7 shows the maximum luminous efficiencies versus color temperature
that were calculated assuming 3 LEDs having 20 nm bandwidths and applying a CRI
minimum of 80 as a constraint. For CCTs between 2500 and 6000 K, the ideal LED peak
wavelengths are approximately 460, 540, and 605 nm. References on creating good
quality white spectra from combinations of LEDs include [14], [15].
In principle, color mixing is the most efficient method to generate white light. In
practice, InGaN LEDs in the 460 nm range are capable of high efficiencies and long
lifetimes, but InGaN LEDs in the 540 nm range have low quantum efficiencies an will be
discussed later in the chapter. The InGaAlP technology used for LEDs in the 605 nm
range is operating far from its peak quantum efficiency near 650 nm. In addition,
InGaAlP is less thermodynamically stable than InGaN, resulting in shorter lifetime and/or
reduced drive current capability. Another concern is lamp cost: color mixing requires
three separate LED chips, so assuming chip cost is the dominant factor in lamp cost, the
cost of color mixing lamps may approach three times the cost of single chip lamps.
Full wavelength conversion
Full wavelength conversion [16], [17] uses a single LED with a phosphor mixture
that completely absorbs the LED emission and emits white. This approach requires a
short wavelength LED in the violet or ultraviolet region. Advantages of this approach
include uniform color versus emission angle, a single LED required, and high efficiency
InGaN LEDs which are available in the required wavelength range. Disadvantages of the
approach are caused by the short emission wavelength. A high Stokes loss (25-30%) is
11
incurred and the most damage of the three approaches is inflicted on the packaging
because of the highest energy LED emission. This damage can result in the yellowing of
epoxy and encapsulation, resulting in shortened lifespan [18]. Another loss mechanism is
scattering of light by the phosphor, which lengthens photon path lengths inside the device,
leading to increased absorption.
Partial wavelength conversion
Partial wavelength conversion is also called the hybrid approach since a mixture
of LED and phosphor emission is used to create white. A blue LED chip is coated with
broadband yellow emitting YAG:Ce phosphor coating to obtain white. To the best
knowledge of the author, all commercial white LEDs use this approach, though some use
an additional red-emitting phosphor to obtain a warmer or lower color temperature white.
Advantages of this method include use of a single LED chip, namely a high efficiency,
long-life blue InGaN LED, and reduced Stokes loss (~15%) compared to the full
wavelength conversion approach. Disadvantages due to phosphor conversion include
inconsistent color versus emission angle and phosphor scattering loss.
Chip, phosphor, and package efficiency targets
Figure 6 details the efficiency and wavelength targets necessary for each of the
three approaches to reach 200 lm/W. All approaches require a combination of LEDs and
phosphors to emit in three wavelength regions in order to obtain white: blue region, 440-
480 nm; green region, 520-560 nm, red region, 590-630 nm. The color mixing approach
requires a chip efficiency of 53%, the lowest chip efficiency of the three approaches, and
a package capable of mixing the 3 LED emission uniformly with 95% efficiency. The
full wavelength conversion approach has the highest chip efficiency requirement, 76%,
12
CHIP AND PHOSPHOR λ AND EFFICIENCY
TARGETS
SSL-LED 2002
SSL-LED 2007
SSL-LED 2012
SSL-LED 2020
Full Wavelength Conversion
LED λs (nm) UV (370-410)
UV (370-410)
UV (370-410)
Phosphor λs (nm) R (590-630) G (520-560) B (440-480)
R (590-630) G (520-560) B (440-480)
R (590-630) G (520-560) B (440-480)
Stokes efficiency 0.73 0.73 0.73
Phosphor quantum efficiency 0.75 0.85 0.95
Package efficiency 0.75 0.9 0.95
Chip efficiency 0.46 0.67 0.76
Color Mixing
LED λs (nm) R (590-630) G (520-560) B (440-480)
R (590-630) G (520-560) B (440-480)
R (590-630) G (520-560) B (440-480)
Package efficiency 0.75 0.9 0.95 Chip efficiency 0.25 0.42 0.53
Partial Wavelength Conversion
LED λ (nm) B(460) B(440-480) B(440-480) B(440-480) Phosphor λs (nm) Y(580) R (590-630)
G (520-560) R (590-630) G (520-560)
R (590-630) G (520-560)
Stokes efficiency 0.88 0.86 0.86 0.86 Phosphor quantum efficiency 0.6 0.7 0.8 0.9 Package efficiency 0.5 0.75 0.9 0.95 Chip efficiency 0.24 0.42 0.61 0.68
Figure 6. Breakdown of efficiency targets for generation of white light through three methods: wavelength conversion, color mixing, and hybrid (LED + phosphor). From [2].
primarily because of the Stokes efficiency of 73%, the lowest of the three approaches.
Furthermore, full wavelength conversion requires phosphors with an effective quantum
efficiency of 95%, and a packaging that extracts 95% of the phosphor-converted and
unconverted light from the device. It should be noted that before this work, a package of
13
this efficiency did not exist. Finally, the partial wavelength conversion approach requires
a blue LED having an efficiency of 68%. Stokes efficiency is estimated to be 86% and
phosphor quantum efficiency and package efficiency targets are 90%, and 95%,
respectively.
It is the opinion of the author that the color mixing approach is least likely to
achieve 200 lm/W efficacy primarily because of the problem with achieving high
efficiency in the green region. The higher cost of the multiple chip architecture would
make it less competitive with other lighting technologies. Full wavelength conversion is
more likely to achieve 200 lm/W efficacy before color mixing because only one violet-
emitting chip needs to be optimized and Lumileds InGaN LED technology achieves
higher efficiencies at shorter wavelengths. However, since the LED emission is
completely absorbed, a large Stokes loss hinders full wavelength conversion efficiency
by approximately 37% (1.0/0.73 – 1) compared to color mixing and by ~ 18% (0.86/0.73
– 1) compared to partial wavelength conversion. In the author’s opinion, partial
wavelength conversion, the method used by all currently available white LEDs will
achieve 200 lm/W efficacy first. The most efficient visible region LED available today is
the Cree EZR260 that achieves 30-33 mW of output at approximately 50% wall-plug
efficiency around 460 nm peak wavelength in its highest efficiency bins. This, is
however, a low-power chip, unsuitable for lighting applications. Of the three approaches,
this is closest to achieving LED efficiency (0.68) required by the OIDA roadmap.
However, package efficiency and possibly phosphor quantum efficiency need significant
improvements.
14
CCT (K)
CRI Maximum luminous efficacy (lm/W)
Blue λ (nm)
Green λ (nm)
Red λ (nm)
2500 80 419 463 547 610 3000 80 416 462 544 608 3500 80 408 462 543 607 4000 80 397 461 542 606 4500 80 387 460 540 605 5000 80 378 459 539 604 5500 80 368 459 539 604 6000 80 361 459 539 604
Figure 7. Maximum luminous efficacy versus correlated color temperature for a combination of 3 LEDs. From [2].
White phosphor-converted LED efficiencies
The efficiency of a pcLED can be expressed as [19]:
pqsledpcL ηηηηη =, Equation 2
where ηpcL is the pcLED wall-plug efficiency, ηled is the pump LED wall-plug efficiency,
ηs the Stokes conversion efficiency, ηq the phosphor quantum efficiency, and ηp the
package efficiency, which is defined here as the fraction of photons emitted by the drive
LED extracted from the device as phosphor converted light after accounting for ηq. The
Stokes efficiency (or quantum deficit) is given by the quantum ratio of the average
emission wavelengths of the LED and the phosphor. Loss of blue LED light before
reaching phosphor and absorption of phosphor converted light by the LED chip,
reflectors, and encapsulation reduces ηp.
It should be noted that the phosphor quantum efficiency and the package
efficiency are generally somewhat entangled in pcLEDs, so another figure of merit is the
15
Figure 8. Diagram of conventional phosphor-converted LED.
product ηqηp. The phosphor layer is usually an inorganic powder layer that not only
converts photons, but also scatters and diffusely reflects. A thicker phosphor layer will
tend to reduce the effective quantum efficiency of the phosphor, but also tend to trap light
inside of the device by diffuse reflections and scattering, thus lowering the package
efficiency.
A schematic of a conventional high power pcLED is shown in Figure 8. It
consists of a blue LED chip mounted to a heat sink with a solder or conductive epoxy
back contact and with a wire bond attached to a top contact. The mounted chip is coated
with a phosphor layer, most commonly YAG:Ce, and is encapsulated with a transparent
polymer.
An analysis of package efficiency, ηp, begins with blue light exiting the LED chip.
In conventional pcLEDs, the light immediately encounters the YAG:Ce phosphor layer as
shown in Figure 9. For a balanced white, the YAG:Ce phosphor layer density is
approximately 8 mg/cm2 [20-22]. A study of 470 nm blue LED light incident on a planar
YAG:Ce phosphor layer with a density of 8 mg/cm2 showed the following results [22]:
(1) 34% of the total photons (phosphor-converted + unconverted blue) are directed
phosphor
blue LED chiptransparent encapsulation
heat sink and electrical contacts
phosphor
blue LED chiptransparent encapsulation
heat sink and electrical contacts
16
Figure 9. Electron micrographs of conventional pcLEDs: LED chip with conformal phosphor coating (top) and slurry-deposited phosphor coating (bottom). From [23]. The conformal phosphor coating has a more uniform spectrum versus angle.
backward, toward the LED chip, where high losses occur; (2) 30% of the total photons
are directed forward, away from the LED chip, toward exiting the device; (3) the
remaining 36% of the total photons are lost, with 13% being lost to total internal
reflection inside the phosphor layer. The 23% (36% - 13%) YAG:Ce conversion loss is
also consistent with other literature [24].
Based on the results above, significant efficiency improvements over
conventional white pcLEDs are achievable by several methods: (1) separating the chip
from the phosphor layer to minimize the portion of backward-directed light that enters
the lossy LED chip; (2) reducing the conversion loss of the phosphor either by
engineering ways to get light in and out of the phosphor layer more effectively or using a
phosphor with a higher intrinsic quantum efficiency than YAG:Ce; (3) altering the planar
phosphor geometry that is good at trapping emitted light by total internal reflection.
17
Figure 10. Diagram of scattered photon extraction white pcLED. Modified from [25].
A white pcLED using a method called scattered photon extraction (SPE) was used
to demonstrate a 60% improvement in light extraction over conventional pcLEDs [25],
[20]. The luminaire, shown in Figure 10 addressed the largest loss mechanism in
conventional pcLEDs by moving the YAG:Ce phosphor layer away from the LED chip.
It should be noted, however, that further package improvements over the SPE
pcLED are possible and, indeed, necessary to achieve the ultimate OIDA roadmap goals.
The SPE pcLED and other remote phosphor LEDs [26] still retains a planar YAG:Ce
phosphor layer with its conversion and total internal reflection losses of 36%. In addition,
a smaller but still significant loss occurs as all backward-directed light from the phosphor
will have to contact a presumably metallic reflector with 95% reflectance at least once
before exiting the device.
95% reflective surface
blue LED chip
~ 60% reflected light
phosphor
18
Chapter 2. Materials: LEDs, Dyes, Polymers, and Phosphors Inorganic light sources: light emitting diodes
Light emitting diodes (LEDs) are among the most efficient light generators in the
visible region and are capable of lifetimes of tens of kilohours [27], [18]. The InGaN
material system covers the wavelength range from near-ultraviolet to green, while the
InGaAlP material system extends from yellow to near-infrared. However, the InGaN
material system tends to have highest efficiencies at short wavelengths, while the
InGaAlP system has the highest efficiencies at long wavelengths.
External quantum efficiency versus wavelength for Cree InGaN LEDs in 2004 is
shown in Figure 12 [28]. The peak EQE exceeds 45% near 455 nm, with efficiency
dropping below 30% and 25% at wavelengths less than 400 nm and greater than 530 nm,
respectively. This efficiency peak is fortuitous for solid-state lighting applications, where
dominant wavelengths of 450-460 nm are required.
External quantum efficiencies for state of the art LEDs at 350 mA in 2006 are
shown in Figure 11 [29]. Peak EQE for InGaN LEDs is near 45% at ~ 370 nm,
decreasing to just below 40% at ~ 450 nm, and less than 20% at ~ 525 nm. Peak EQE for
AlGaInP LEDs exceeds 50% at 650 nm, but drops rapidly with wavelength to less than
20% at 600 nm.
Because both InGaN and AlGaInP LEDs decrease in efficiency toward the center
of the spectrum, a “green gap,” or window of relatively low efficiency between the two
material systems exists. Despite the low efficiency in the green and yellow regions of the
spectrum, green and yellow LEDs have already been used in traffic lights for several
19
Figure 11. External quantum efficiency versus peak wavelength for state of the art LEDs at 350 mA drive current and junction temperature of 25° C. Modified from reference [29].
years to save energy and maintenance costs when compared to filtered incandescent
lamps. An opportunity exists to raise the EQE of green LEDs to that of blue LEDs by
phosphor conversion. For a sufficiently efficient phosphor and package, phosphor
converted green LEDs can outperform direct-gap emitting green LEDs. Suppose the ~
450 nm LED having EQE of ~ 40% from Figure 11 is converted by a phosphor to 530 nm
and extracted from the device at 90% efficiency. The green phosphor-converted LED
(pcLED) would have an EQE of 36%, which nearly doubles the value of the direct-gap
device. Even when the approximately 18% higher voltage required (530/450 = 1.18) to
20
Figure 12. Radiant flux and external quantum efficiency versus dominant wavelength for Cree InGaN multiple quantum well LEDs. Ovals represent many data points. Modified from reference [28].
drive a blue LED is considered, the green pcLED will outperform the direct-gap green in
electrical to optical power conversion efficiency by nearly 70%. Such an LED would be
superior in traffic lights, display backlighting, and 3-color (RGB) solid-state lighting.
The power output of single LEDs has been increasing at a rate of 20X per decade,
while the cost per lumen has been falling at a rate of 10X per decade [30]. This trend is
known as Haitz’s Law, and is analogous to Moore’s Law for silicon technology.
Organic dyes
Organic fluorescent dyes possess qualities such as high quantum efficiency and
low cost, making them an attractive material for photonic applications. Fluorescent dyes
have been used for decades in dye lasers and more recently in organic light-emitting
21
diodes (OLEDs) and organic photovoltaics. Unfortunately, their limited stability requires
constant replenishment of active dye in lasers and limits OLEDs to low intensities to
achieve required lifetimes in the thousand hour range.
Upon excitation by photon absorption or electrical stimulation, fluorescent dyes
form higher energy states called excitons. Excitons are excited-state molecules having a
bound electron-hole pair. Each electron has a distinct spin (up or down). Photon
excitation results in 100% excited-state singlets, so dyes are capable of nearly 100%
quantum efficiency when excited by photons. Singlets are antiparallel spin states that are
allowed to recombine by the Pauli Principle. The Pauli Principle states that no two
electrons can have the same quantum numbers, which here means that electrons in the
same state must have opposite or antiparallel spins. Singlets are thus short-lived states,
having a high radiative efficiency. Electrical excitation is less efficient due to quantum
mechanical spin statistics. Electrical excitation generally results in exciton formation.
Electrically generated excitons are 25% singlet and 75% triplet states because of the spin
statistics. Consequently, OLEDs based on fluorescent dye molecules are limited to 25%
internal quantum efficiency.
So, electrically pumped fluorescence, as found in OLEDs has a quantum
efficiency limit of 25%, while photon pumped fluorescence has a quantum efficiency
limit of 100%.
Fluorescent dyes studied in this work include Lumogen F series dyes from BASF,
dyes from Keystone Anilene, and Rhodamine 590 or 6G [31], a well-known laser dye.
Lumogen F dyes having a perylene-based structure included Lumogen Yellow [32],
Lumogen Orange [33], and Lumogen Red [34]. Lumogen Violet [35] has a
22
0
20
40
60
80
100
120
450 500 550 600 650 700 750 800
Keystone YellowLumogen YellowLumogen OrangeRhodamine 590Lumogen Red
Nor
mal
ized
Inte
nsity
(a.u
.)
Wavelength (nm)
Figure 13. Emission spectra of Keystone Yellow, Lumogen Yellow, Lumogen Orange,
Rhodamine 590, and Lumogen Red. A blue LED was the excitation source.
napthalimide-based structure. Dyes from Keystone Anilene that were studied were
Keystone White (actually a broadband blue), and Keystone Yellow [36].
+
ClO4-
+
ClO4-
Figure 14. Structure of perylene, Lumogen Yellow, Lumogen Orange and Lumogen Red, and Rhodamine 590 perchlorate.
23
Ground state (s = 0)
Excited singlet state (s = 0)
Excited triplet state (s = 1)
+ hν
fluorescence+ hν+ hν
fluorescence + hν
phos
phore
scen
ce
+ hν+ hν
phos
phore
scen
ce
Ener
gy
Ground state (s = 0)
Excited singlet state (s = 0)
Excited triplet state (s = 1)
+ hν
fluorescence+ hν+ hν
fluorescence+ hν
fluorescence+ hν+ hν
fluorescence + hν
phos
phore
scen
ce
+ hν+ hν
phos
phore
scen
ce
+ hν
phos
phore
scen
ce
+ hν+ hν
phos
phore
scen
ce
Ener
gy
Figure 15. Diagram showing spin-allowed fluorescence and spin-disallowed phosphorescence.
Transparent polymers
Transparent polymers have two important functions in devices explored in this
work. Most importantly, transparent polymers serve as a host material for the dispersal
of fluorescent dyes. The other application is as a transparent encapsulation material for
the dye dispersed in polymer phosphor. Important properties for these polymers are
optical clarity, high solid-state solubility of the fluorescent dyes of interest, solubility in
solvent for solvent-casting, and sufficient stability to withstand thermal treatment
necessary to drive off solvent.
Polymethyl methacrylate (PMMA), or acrylic, is a polymer having high optical
clarity, often used as a glass substitute in such applications as eyeglasses, windows
(Plexiglass), and automotive headlights. PMMA has a glass transition temperature of
24
107°C, making it suitable for relatively high temperature applications. Unfortunately,
PMMA is poorly soluble in most solvents, especially at high molecular weights, which
can lead to poor quality films when cast from solvent. A modified styrene acrylic
polymer, J587 from Johnson Polymer [37], [38] was found to have the high optical clarity
of PMMA, but a better solubility in solvent. Most results concerning dyes in a polymer
matrix in this work used J587 as the polymer. Solid-state solubilities of the Lumogen
series of fluorescent dyes were found to exceed 1% by weight in J587. The J587 polymer
was found to be highly soluble in acetone, butyl acetate, and benzyl alcohol. The J587
raw material was in the form of chopped extruded cylinders, having dimensions on the
order of a few millimeters. Solutions containing 25% of J587 by volume in solvent were
generally used for casting of thin and thick films. The polymer was completely dissolved
after 1-2 hours of stirring with a magnetic stirring hotplate.
A transparent silicone, General Electric RTV615 [39] was also tested for use as a
host material for fluorescent dyes Silicones have the advantage of being more easily
molded into complex shapes than PMMA when there is no access to injection molding.
In addition some silicones have been shown to have greater stability than PMMA at high
temperatures and light intensities, and are used as encapsulation materials in high power
LEDs. Unfortunately, the fluorescent dyes used here were not sufficiently soluble in the
silicone, resulting in precipitation of dye clumps, which give poor optical properties.
25
CH3
--C--—C—--C--—C--—C----C--
H
H
HH
H
C=OCH3
C=OCH3
C=OCH3
CH3
H
CH3CH3CH3
--C--—C—--C--—C--—C----C--
HH
HH
HHHH
HH
C=OCH3
C=OCH3
C=OCH3
C=OCH3
C=OCH3
C=OCH3
CH3CH3
HH
CH3CH3
-Si—O—Si—O—Si—O—Si—O-
CH3
CH3
CH3 CH3 CH3
CH3 CH3 CH3
-Si—O—Si—O—Si—O—Si—O-
CH3
CH3
CH3 CH3 CH3
CH3 CH3 CH3
-Si—O—Si—O—Si—O—Si—O-
CH3CH3
CH3CH3
CH3CH3 CH3CH3 CH3CH3
CH3CH3 CH3CH3 CH3CH3
Figure 16. Structure of poly-methyl methacrylate (PMMA) (top) and poly-dimethylsiloxane (PDMS) (bottom). The monomer of each is in parentheses.
Hybrid inorganic-organic
Hybrid inorganic-organic (HIO) is defined in this work as the combination of
inorganic light emitting diodes with organic color conversion materials. The motivation
for the combination is generation of light covering the visible spectrum at maximum
efficiency. Such a system has potential application to displays, lighting, and lasers.
Inorganic sources will be predominantly InGaN-based light emitting diodes with
emission in the violet and blue regions of the spectrum. Organic color converters consist
of small molecule fluorescent dyes dispersed in transparent polymer hosts. Inorganic
violet-blue LED technology and organic color converters have complementary strengths
that allow highly efficient generation of light across the visible spectrum. Combining
26
blue LEDs with organic fluorescent dyes was first achieved in 1997 [40], with a patent
issued in 1999 [41].
This dispersal of fluorescent dyes in a polymer matrix is important in preventing
aggregation of dye molecules. Aggregation reduces quantum efficiency because of the
nonradiative energy transfer mechanisms between nearby molecules. The net effect of
this transfer of energy is extending the lifetime of the excited state, thereby increasing
probabilities of non-radiative decay. Non-radiative decay is undesirable, not only
reducing quantum efficiency of the material, but can result in chemical destruction of the
dye molecules, shortening the useful lifetime of the material.
In this work, the organic dyes discussed previously were dispersed in Joncryl 587
modified styrene acrylic polymer at concentrations of 0.2 to 1% by weight. A
photograph of three of these samples is shown in Figure 17.
Figure 17. Three semi-transparent organic phosphors consisting of small molecule dyes dispersed in a modified acrylic.
27
Inorganic phosphors
The primary inorganic phosphor involved in this work is the cerium-doped
yttrium aluminum garnet (YAG). Because of its broad yellow emission, the phosphor is
ideal in partially converting a blue LED to create white. Nearly, if not, all commercial
white LEDs use the YAG:Ce phosphor at least in combination with other phosphors. The
emission of YAG:Ce can be tuned somewhat by substitution of Y and Al with Gd and Ga
atoms, respectively [42]. Other inorganic phosphors used for white LEDs are rare-earth
doped SiAlON [43], nitridosilicates [44], CaS:Eu2+ [24]
The biggest advantage of inorganic phosphors over organic phosphors is their
stability. Inorganic phosphors can be used in long-lived high-intensity applications such
as in lighting and lasers. The biggest disadvantage of inorganic phosphors in maximum
efficiency applications is their high refractive index and prohibitively high cost for large
single crystals. The high refractive index means they cannot, in general, be encapsulated
in a polymer of matching refractive index to eliminate light scattering. The scattering
results in trapping of emitted light, effectively lowering the quantum efficiency. As
mentioned before, there is a 36% conversion loss when using YAG:Ce to create white
from a blue LED, thus giving an effective quantum efficiency of 64%. This is far below
the Year 2020 OIDA roadmap requirement of 95% to achieve 200 lm/W luminous
efficacy. So either both scattering and the consequent light trapping by the phosphor
must be reduced or another phosphor must be found to reach the ultimate solid-state
lighting efficiency goals.
28
Figure 18. Semitransparent organic phosphor (left) and diffuse YAG:Ce in a transparent silicone matrix. The films have similar optical density in the blue.
29
Chapter 3. Luminaires for Solid-State Lighting
Planar luminaire
The white planar emitter constructed consists of three components: (i) an edge-
mounted violet pump LED; (ii) a glass or acrylic waveguide to transmit pump light; (iii) a
semitransparent organic thick film color conversion material (CCM). Light exits the
violet LED and is coupled to the waveguide. The pump light travels through the
waveguide via total internal reflection until it encounters the index-matched organic
CCM. Here, the violet light is absorbed and white light is emitted from a combination or
blue and orange fluorescent dyes.
To fabricate the planar luminaire, a modified styrene acrylic polymer, Joncryl 587
with 1% by weight fluorescent dye was dissolved in an organic solvent. Two solutions
with different dyes were required to produce white light; one with Lumogen Violet,
which emitted in the blue region; the other with Lumogen Orange, which emitted a
broadband orange. Microscope slides were used for the waveguide. They were placed
on a hotplate at 150°C and different ratios of the dye solutions were pipetted onto the
slide surface. The solvent gradually evaporated, leaving a solid polymer film with the
dyes uniformly dispersed throughout.
Figure 19. Diagram of planar luminaire configuration. The phosphor consists of a fluorescent dye in a polymer matrix and the waveguide a transparent sheet of glass or acrylic. From [45].
waveguide
CCM
400 nm LED n = 1.5
n = 1.5
n = 1
waveguide
phosphor
n = 1.5 405 nm LED
30
The planar luminaire is attractive because of its thin profile and large emission
area. It consists of a planar PMMA or glass waveguide with an index-matched
lumophore coating pumped from the edge by LEDs. Violet LEDs edge-pumping
microscope slides coated with a lumophore coating containing blue and orange dyes were
demonstrated. The color uniformity over the emitting area is very good for two reasons.
Because the violet light is completely converted to other wavelengths by dye-based
phosphors, all visible light originates from the same place, inside the phosphor coating.
Also, any violet pump light that does escape the waveguide does not shift the color
because of the low sensitivity of the eye at short wavelengths.
The planar geometry is desirable for many applications. For general lighting it
can cover large areas of ceilings or walls. Additional fixturing is unnecessary because
the emission is at a uniform low intensity for large coverage areas and the radiation
pattern is lambertian. Another application for the planar geometry is backlighting for
LCDs.
The primary drawback to planar light emitting structures is the trapping of color
converted light by total internal reflection. For an emitter inside a planar structure of
refractive index 1.5, only 12.7% of the light escapes into air from both the top and bottom.
So 74.6% of the light is trapped or escapes from the edges, where it does not contribute to
useful device output. The efficiency of organic light emitting diodes (OLEDs) is also
much less than 100% for the same reason. External quantum efficiencies (EQE) of
planar OLED structures have approximately 20% external quantum efficiency [46] and
can be improved somewhat by adding external lenses. A truncated pyramid OLED
luminaire showed a 1.8-2.1X improvement in EQE [47], and addition of hemispherical
31
polymer lenses showed a 2.5X improvement [48]. With the 2.5X improvement, planar
OLED EQE is currently limited to ~ 50%. So, despite internal quantum efficiencies
approaching 100% for phosphorescent OLEDs, poor EQE likely prohibits such designs
from achieving 50% power efficiency or 200 lm/W efficacy.
Cylindrical luminaire
The cylindrical luminaire consists of a transparent acrylic rod waveguide coated
by organic fluorescent material with a violet pump LED coupled to one end of the rod as
in Figure 20. This configuration has the same form factor as fluorescent tubes, thus
allowing a relatively low-cost retrofit of existing fluorescent fixtures.
Fabrication involved cutting 3” lengths from acrylic rod stock. A 1/4” hole was
drilled in the end of the rod for LED mounting, approximately ¼” deep. The 5 mm
package violet LEDs were secured with a transparent epoxy. Test tubes filled halfway
with the polymer and dye solutions described above were prepared for dip-coating of the
rods. Blue, green, and red colored lamps were produced by dipping in the polymer
solutions containing Lumogen Violet, Lumogen Yellow, and Lumogen Red, respectively.
The white lamp was made by first coating in Lumogen Violet solution, then Lumogen
Orange solution. A photograph of all the lamps in operation is in Figure 21.
32
Figure 20. Schematic of the cylindrical lamp, showing the three primary components: pump LED, acrylic rod waveguide, and fluorescent coating.
Measuring power and luminous outputs of the cylindrical white lamp was done by taking
several spectra along the length and ends of the rod. Power measurements were made
with a calibrated silicon photodetector, partially covered by a reflector with ~2 mm
diameter aperture. The aperture allows measuring the power from a precise spot on the
lamp surface while excluding emission contributions from other places on the lamp
surface.
The spectrum of the white lamp (Figure 22) clearly shows strong emission in both
the 420-500 nm range from Lumogen Violet and 530-650 nm range from Lumogen
Orange. The correlated color temperature was approximately 7500 K. Luminous
efficacy was estimated at 7 lumens per watt when emission from the rod ends was
included. In practice, however, in a relatively long fixtured rod luminaire, emission from
the rod ends would likely be lost.
405 nm LED
acrylic rod
fluorescent coating
33
Figure 21. Photograph of rod luminaires in operation. From left to right: violet (no color conversion), blue (Lumogen Violet), green (Lumogen Yellow), red (Lumogen Red), and white (Lumogen Violet and Orange) luminaires. From [49].
400 450 500 550 600 650 700
7500 K blackbodylamp spectrum
Inte
nsity
(a.u
.)
Wavelength (nm)
Figure 22. Spectrum of the white rod luminaire. A 7500 K blackbody spectrum is provided for comparison. From [49].
34
Spherical luminaire
The objective of the spherical luminaire was to increase lamp efficiencies by
elimination of the strong internal reflection found in both the planar and rod luminaire
configurations. Emission from the lamp center, where the LED is located would travel
radially outwards before encountering the glass/air interface at near normal incidence,
resulting in high transmission and eliminating internal reflection.
Coincidentally, the form of the spherical luminaire is reminiscent of a typical
incandescent bulb. This fact would allow a lower-cost retrofit of existing fixtures
designed for incandescent lamps rather than design of new fixtures from the beginning.
The familiar shape would also likely ease consumer acceptance for the new technology.
Fabrication of the spherical luminaire would have been difficult with a solid
polymer light-conversion layer, so a liquid color converter was used for fabrication and
experimental simplification. The liquid color converter consisted of the same fluorescent
dyes used in the solid color converters in a solution of acetone. The spherical shell of the
25 mL flask
405 nm LED
solvent:dye
wire leads
rubber stopper
solution
Figure 23. Structure of the spherical luminaire.
35
lamp was a common 25 mL round-bottom boiling flask available from the chemistry
stockroom. The flask was filled with liquid color converter. A 5 mm package violet
pump LED needed to be suspended in the center of the sphere, so copper wires were
soldered to the leads and extended out of the mouth of the flask. Insulation needed to be
stripped from the wires to avoid being attacked by the solvent. The LED was held in
place by inserting a rubber stopper into the flask, which firmly held the copper wires
against the inside of the flask mouth. A schematic of the luminaire assembly is shown in
Figure 23.
Figure 24. Photographs of the spherical luminaire in operation. A close-up of the device under ambient lighting (left) and illumination of a CIE diagram in an otherwise dark room (right).
36
Figure 25. Schematic of spherical luminaire power measurement experiment.
Because of the spherical symmetry, lamp characterization was relatively simple.
Since the violet LED light is completely absorbed and re-emitted by the color converter,
emission should be uniform over the surface of the lamp and decrease with the inverse
square of the distance from the lamp.
First, power measurements were made with a silicon photodetector (Figure 25).
The distance, d, was varied to verify the inverse square law and to estimate the
uncertainty of the power value due to error in d. The distance was varied from 5 to 30
inches. The result of these measurements is plotted in Figure 26. There was less than 5%
variation in the power calculated from 14 different distances. Thus, the error in
measurement distance plus any deviation from the inverse square law affect the overall
power measurement by under 5%.
Distance, d
Si photodetector
37
Figure 26. Typical result of the spherical lamp power efficiency measurement. As the measurement distance varied, error in the distance measurement resulted in less than 5% variation in lamp power.
Second, uniformity of the emission from around the spherical emitting surface
had to be confirmed. Four spectra were taken with the fiber spectrometer around both the
equator and meridian of the lamp. The spectra were integrated to obtain the total output
intensity at each position. Results of this measurement are shown in Figure 27. The
standard deviation for the eight measurements combined was approximately 10%. It was
concluded that the spherical symmetry greatly simplified power measurement and spectra
by requiring measurements only be taken from a single direction at a fixed distance in
order to characterize the entire lamp’s spectrum and output power within ~ 10%.
0.001.002.003.004.005.006.007.008.009.00
10.00
5 10 15 20 25 30
distance (in.)
Pow
er E
ffic
ienc
y (%
)
< 5% variation
38
Figure 27. Intensity uniformity of the spherical lamp. The four positions of the equatorial and meridional measurements were in 90 degree intervals.
Spectra and results for a series of spherical lamps is shown in Figure 28. Initially,
the lamp contained only Keystone White, a blue fluorescent dye, and the result is shown
in the bottom spectrum. Note that there is no significant emission at 405 nm, indicating
that the pump LED light has been nearly completely absorbed. The spectrum and power
reading were taken at a distance of 15 inches. After measurement, the rubber stopper is
removed, and a precise amount of additional dye was added. The LED is re-centered, the
stopper replaced, and another power measurement and spectrum is collected. The lamp
reaches a maximum luminous efficiency of 14.5 lm/W with a CCT of 2408 K and CRI of
82 in the final (top) spectrum. Notice the wide range of color temperatures useful for
lighting applications ranging from a warm white of 2408 K to cool white at 6554 K,
roughly the temperature of the solar spectrum at noon. Color temperature decreased
monotonically with increasing dye amounts because the spectrum was increasingly red-
shifted by the additional dye. Luminous efficiencies increased as more dye was added
0
0.05
0.1
0.15
0.2position 1
position 2
position 3
position 4
equatorialmeridional
~ 10% variation
39
because the average wavelengths of the spectra increased, shifting the spectrum closer to
the maximum eye responsivity. The spectrum efficiency increase was enough to
overcome the monotonic decrease in package and dye quantum efficiencies, which
occurred for two reasons. First, adding more dye results in more radiative and non-
radiative energy-transfer events between dye molecules; each one of these transfers has a
probability of loss equal to 1 - ηq for the specific dye. Second, each wavelength
conversion occurs, on average, at a farther average distance from the center of the
luminaire as dyes are added. The farther from the center a light emitter is located, the
more chance for reflection or worse, total internal reflection. These reflected rays have
longer path lengths inside the luminaire before escaping, a greater chance of being lost,
thus reducing the package efficiency, ηq.
40
Figure 28. Series of spectra from a spherical luminaire. Luminous efficiencies, correlated color temperatures, and color rendering indices are listed. The initial spectrum is at the bottom; progressive addition of dye solutions to adjust color resulted in the upper spectra.
400 450 500 550 600 650 700
Spe
ctra
l Irr
adia
nce
(a. u
.)
Wavelength (nm)
4.6 lm/W, 34367 K, CRI = 9
10.2 lm/W, 34463 K, CRI = 44
12.2 lm/W, 19562 K, CRI = 44
12.3 lm/W, 12673 K, CRI = 74
11.3 lm/W, 8250 K, CRI = 48
11.8 lm/W, 6554 K, CRI = 46
12.8 lm/W, 4680 K, CRI = 60
13.2 lm/W, 3492 K, CRI = 72
13.7 lm/W, 3025 K, CRI = 80
14.5 lm/W, 2408 K, CRI = 82
41
Figure 29. Schematic of a whispering gallery mode. A ray is emitted from a point source at radius r, at a sufficiently shallow angle to the surface and suffers total internal reflection. Rays such as these have long path lengths inside the device and are eventually lost to re-absorption.
Whispering gallery modes
Whispering gallery modes can occur in a device having radial symmetry. They
are rays traveling perpendicular to the radius that collide with the outer surface at a
sufficiently shallow angle to undergo total internal reflection, hence will never escape the
device. Consider the sphere in Figure 29 of radius R and refractive index n1, surrounded
by a medium of refractive index n2, and having an internal point emitter at radius r. For
the case of an external phosphor coating, r = R and a significant portion of the emission is
trapped in whispering gallery modes. Thus external phosphor coatings will always be
lossy. For the case of a phosphor emitting from the center of the device, r = 0, all rays
travel parallel to the radius, strike the outer surface at normal incidence, and are either
r
R
n1
n2
42
transmitted out of the device, or are reflected back through center, striking the opposite
surface at normal incidence with another chance to exit the device. External or package
efficiencies for this configuration can approach 100% because of the absence of
whispering gallery modes.
In the case of a maximum efficiency phosphor converted LED having radial
symmetry, it is undesirable to have the phosphor at r = 0, because that is the location of
the LED. A large portion of the phosphor emission will enter the LED chip and be
subject to high losses. So, a pcLED with maximum efficiency will have the phosphor
located at some intermediate position 0 < r < R. The optimum position will be far
enough from the LED such that little phosphor emission is directed back into the LED
chip, yet also far enough from the outer surface to avoid total internal reflection or
whispering gallery modes.
Figure 30 approximately models the dye emission inside the spherical luminaire.
It consists of a sphere with a homogeneous refractive index of 1.5 surrounded by air with
a single point source at r = 0.6. In the actual luminaire, the internal refractive index of
the lamp (acetone) is 1.36, surrounded by a glass shell having a refractive index of
approximately 1.5. The lamp is filled with many dye molecules interacting with each
other instead of a single point source, but the model is still valuable in illustrating the
paths various rays take inside the luminaire. Polarization of the rays was assumed to be
random when calculating transmittance values. That is, half strike the surface with plane-
parallel polarization and half strike the surface at plane-perpendicular polarization. The
transmittance of the ray at the dielectric/air interface is
43
( )
+
=+=ii
tt
2
ttii
iilarperpendicuparallel cosn
cosncosncosn
cosn2TT21T
θθ
θθθ
, Equation 3
where i subscripts denote the dielectric lens material and t subscripts denote the
surrounding medium, air. The relation between Өt and Өi is given by Snell’s law:
ttii sinnsinn θθ = Equation 4
Rays emitted from near the center of the lamp (r/R ≤ 0.5) have a greater than 90%
chance of exiting at the first incidence on the surface regardless of emission direction,
resulting in high external efficiencies. Rays emitted from intermediate positions (0.5 ≤
Figure 30. Diagram of a dielectric sphere with refractive index 1.5 surrounded by air (top left). Transmittance at first encounter of external surface versus emission angle from various r/R emitter positions (top right). Table of first incidence transmittance (T) integrated over the 4π emission solid angle for select values of r/R (bottom).
r/R 0 0.3 0.5 0.6 0.7 0.9
T 0.96 0.95 0.93 0.88 0.58 0.28
r/R ≤ 0.6) have a greater than 75% chance of exiting at first incidence. The
transmittance of rays emitted in certain directions from r/R = 0.7 drops to zero. The
physical interpretation of this is that these rays suffer total internal reflection and never
00.10.20.30.40.5
0.60.70.80.9
1
0 45 90 135 180 225 270 315 360angle (deg.)
Tran
smitt
ance
0.3
0.5
0.6
0.7
0.9
point emitter
n = 1.5 n = 1
180°
90° 270°
T
R
0°
r/R
0.3
0.5
0.7
0.9
0.0
r/R =
44
escape the device. For r/R = 0.9, if isotropic emission is assumed, nearly half of all
emitted rays will be lost to total internal reflection.
The conclusion reached by building and characterizing the spherical luminaire
and analyzing the loss mechanisms is that the phosphor must be confined to some radius r
< R so that total internal reflection is avoided and package losses are reduced.
Introduction to the ELiXIR luminaire
As discussed earlier, conventional pcLED approach to solid state lighting utilizes
a blue LED chip coated with a phosphor layer, as shown in Figure 31(a). The
prototypical example of this is the InGaN blue/YAG:Ce white LEDs available
commercially. YAG:Ce is a broadband yellow emitting phosphor which when combined
with the partially transmitted blue LED light yields an acceptable white spectrum.
Advantages of this configuration include compact device size, minimal phosphor mass,
and consistent color vs. angle. The primary drawback of this configuration is poor
package efficiency. Conventional pcLED losses are dominated by reflection of pump
light back into the LED chip and the large fraction of phosphor emission directed into the
LED chip. Secondary losses are encountered at imperfect mirrors used to reflect light out
of the device. Because of these limitations, conventional pcLEDs may never be capable
of the high efficiencies demanded by DOE and OIDA. For the case of InGaN/YAG:Ce
white, ~ 40% of the blue LED flux is transmitted into the phosphor, while the remaining
~ 60% is directed back into the chip [20]. In addition, by simple geometric reasoning,
half of the
45
PMMA
phosphor
air
glass
reflector
7.6 cm
blue LED chip
heat sink
~ 2 cm
(a) Conventional pcLED (b) ELiXIR pcLED
Figure 31. Conventional phosphor converted LED (left) and ELiXIR phosphor converted LED (right). From [50].
phosphor converted light is emitted directly back into the chip, resulting in further loss.
The OIDA roadmap [2] indicates that currently the range for ηp is 40-60%. In the case of
pcLEDs with 100% phosphor conversion, lower ηp values are normally expected because
of increased losses from the re-absorption of additional converted light. The device
addressed here takes its name from one of its working principles: enhanced light
extraction by internal reflection (ELiXIR). The ELiXIR device, shown schematically in
Figure 31(b), seeks to maximize ηp by: (1) separating the chip and phosphor to nearly
eliminate phosphor emission and LED reflection back into the LED chip; (2) the use of
internal reflection at the air/phosphor layer interface to reduce the number of mirror
reflections.
Separation of the chip and phosphor has been shown to produce higher
efficiencies. Scattered photon extraction (SPE) [20], for example, was used to
demonstrate a 61% increase in both light output and luminous efficiency compared to
conventional white InGaN/YAG:Ce pcLEDs. If the efficiency can be raised 61% over
conventional pcLEDs, an upper bound of conventional pcLED package efficiency is set
at 1/1.61 or 0.62 assuming 100% package efficiency for the SPE pcLED. Since the SPE
46
Figure 32. Ray trace diagram of phosphor emission in the ELiXIR luminaire. From [50].
pcLED has a planar YAG:Ce phosphor layer, it should suffer a 13% total internal
reflection loss [22], thus reducing the maximum package efficiency to 87%. The 0.62
value is consistent with the upper end of the ηp range in the OIDA estimate above. The
package efficiency as defined in Equation 2 here is not determinable.
The concept of combining blue or violet LEDs with organic dyes in a polymer
matrix, such as luminescence conversion LEDs (LUCOLEDs) [51], [52] or hybrid
organic-inorganic LEDs [11], [40], [53], [54] have been previously reported. Efficiencies
of such devices have been less than optimal, however, due to locating the dye in close
proximity to the LED chip as in pcLEDs [40], [51], [52], [53] and limited extraction or
package efficiency due to the planar [40], cylindrical [11], or external coating [54]
organic converter geometry.
The ray tracing diagram of Figure 32 is used to illustrate the various paths
phosphor-emitted photons take in the ELiXIR structure. Ray 1 exits the device without
encountering any reflections and comprises ~ 35% of the phosphor emission. Ray 2
(representing ~ 35% of the phosphor emission) demonstrates the advantage of the
n = 1.5n = 1
n = 1
1 (35%) 2 (35%)
3 (17%)
4 (13%)5 (0.1%)
n = 1.5n = 1
n = 1
1 (35%) 2 (35%)
3 (17%)
4 (13%)5 (0.1%)
47
Figure 33. Two dimensional schematic of ELiXIR lens with index-matching interior phosphor coating, where r is the internal radius of the lens where the phosphor is located, and R is the external radius of the lens. The refractive indices are such that n2 > n1 and Ө is the largest angle at which phosphor emission strikes the external lens surface.
ELiXIR concept. The ray is headed toward the planar reflector, where significant loss
normally occurs, but is instead internally reflected at the phosphor/air interface. The ray
can now exit the device and may avoid the reflector entirely. Since internal reflection is a
lossless process, it is the most efficient way to steer light out of the device. Ray 3,
comprising ~ 17% of phosphor emission, heads directly to the reflector before exiting the
device and never encounters the phosphor/air interface. Ray 4 is transmitted across the
phosphor/air interface but avoids the LED chip and re-crosses the air/phosphor interface
before exiting the device (~ 13%). Finally, ray 5 is transmitted across the phosphor/air
interface and enters the LED packaging or chip where the highest losses are incurred. In
the ELiXIR structures fabricated with a separation between chip and phosphor of 1.9 cm,
ray 5 comprises less than 0.1% of the total phosphor converted light.
Note the absence of internal reflection at the PMMA/air interface, which had
limited external efficiency in previous similar designs [11], [40], [54]. In the ELiXIR
device presented here, the ratio of the phosphor layer radius to that of the outer lens
(rphosphor/rlens) is ~ 0.5.
rRӨ
n1n2
rRӨ
n1n2
48
Figure 33 shows the general case for phosphor emission from a hemispherical
shell lens having internal radius r and external radius R. The ELiXIR luminaire contains
this hemispherical shell with an interior index-matched phosphor coating. A simple
calculation based on the figure can determine the geometric constraints to eliminate total
internal reflection losses. Consider a ray emitted perpendicular to the internal lens
surface by a point on the phosphor. This ray will strike the outer lens surface with the
largest angle Ө. By keeping Ө less than the critical angle, Өc, internal reflection can be
avoided. From the law of sines
r90sinrsinR =°=θ .
At the external lens surface, the definition of critical angle is
2
1c n
nsin =θ . Equation 5
By substitution (when Ө = Өc), the condition to avoid total internal reflection inside the
device becomes
2
1
nn
Rr< . Equation 6
So for the current case of n1 = 1 and n2 = 1.5, r/R must be kept less than 2/3 to eliminate
total internal reflection. Increasing the refractive index of the lens and phosphor requires
a smaller r/R ratio. As a consequence, when LED-phosphor distance (r) remains constant,
the minimum device size increases for larger phosphor/lens refractive index. A list of
maximum r/R values for common phosphor/lens refractive indices is shown in Figure 34.
Figure 34. Maximum r/R values for lens/phosphor refractive indices from 1.3 to 2.0 when n1 = 1. n2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0
r/R 0.769 0.714 0.667 0.625 0.588 0.556 0.526 0.500
49
Although the ELiXIR implementation will achieve maximum package efficiency
when a scatter-free phosphor is utilized (e.g. dye-doped polymer phosphor with uniform
refractive index), use of internal reflection at the air/phosphor layer interface to steer
phosphor emission is also possible with conventional inorganic powder phosphors.
However, powder phosphors strongly scatter both LED and phosphor-emitted light,
which will lead to additional loss. For maximum efficiency, the phosphor must have a
homogeneous refractive index that is equal to or less than that of the lens material so that
phosphor layer traps no emitted light by internal reflection.
Fabrication of the ELiXIR luminaire
(a) First generation ELiXIR luminaire
Luminaire fabrication required casting and finishing of a hemispherical shell lens,
application of phosphor, and attaching the planar reflector to the lens base. The lens
consisted of a thin glass inner shell surrounded by a thick polymethyl methacrylate
(PMMA) outer shell, and was fabricated by polymerization of methyl methacrylate
monomer around a 25 mL round bottom flask with an outside diameter of 3.8 cm. The
outer surface of the lens was shaped by a 7.6 cm diameter spherical aluminum mold,
while the inner surface was defined by the flask.
Preparation of the methyl methacrylate monomer involved washing the monomer
with a solution of sodium hydroxide to remove the hydroquinone inhibitor, rinsing with
deionized water, drying with anhydrous magnesium sulfate, and filtering. The
polymerization was initiated by the addition of 0.1% benzoyl peroxide and heating to 90°
C in a water bath until a viscous syrup was obtained. The flask was positioned in the
mold and the cooled syrup was poured into the mold. The entire assembly was placed in
50
an oven at 35° C for curing. The assembly was removed from the oven every 24 hours
for the first 3-4 days to top off the mold with additional PMMA syrup to replace what
had been lost to vaporization of the monomer. After 3-4 days, the PMMA had cured to
sufficient viscosity that vaporization was no longer fast enough to require additional
topping off. When 7-10 days had elapsed, oven temperature was increased to 60° C for
two hours of final curing. For some lens assemblies, cracking of the inner glass walls
occurred during this high temperature step. After curing, the lens was separated from the
mold after being placed in a freezer for one hour. This step also sometimes resulted in
cracking of the glass interior of the lens. Despite the cracking, however, there was still
good adhesion and optical contact between the inner glass shell and lens bulk. The
performance of the finished lenses was unaffected, so the cracking is only a cosmetic
problem.
The green phosphor consisted of Joncryl 587 modified styrene acrylic [38] with
0.2% BASF Lumogen F Yellow 083 fluorescent dye [32], and was applied to the inner
surface of the lens from a solution in acetone. Phosphor layer thickness was on the order
of 100 µm. The reflector consisted of aluminized Mylar attached to an acrylic sheet.
Reflectance of the aluminized Mylar reflector with adhesive layer was greater than 70%
51
Figure 35. Photographs of ELiXIR lens at various stages of fabrication. First, the PMMA is cast around a flask in an aluminum mold (top left); flask and lens assembly after removal from mold (top right); sawing through the lens and flask to obtain hemispherical shell (middle left); separated parts: lens on left, waste material on right (middle right); finally, the hemispherical shell lens after polishing (bottom).
52
0.0
0.2
0.4
0.6
0.8
1.0
1.2
400 450 500 550 600 650 700 750
Inte
nsity
(a.u
.)
Wavelength (nm)
Luminaire emission
LED emission
Phosphor absorption
Figure 36. Spectra and photograph of first ELiXIR lamp fabricated. Blue LED emission, phosphor absorption, and ELiXIR luminaire emission spectra. Photograph of the luminaire in operation (inset). This luminaire achieved a package efficiency of only 0.78, limited by a 30% absorbing reflector and imperfect LED to lamp coupling (note the blue emission escaping from the edges of the reflector sheet). From [50].
across the visible spectrum. The commercial blue power LED had a peak wavelength of
455 nm and 1000 mA d.c. drive capability. The complete structure of the luminaire is
shown in Figure 31(b).
(b) Second generation ELiXIR luminaire
A more advanced second generation ELiXIR luminaire was fabricated to
demonstrate the highest efficiencies possible. A photograph of two completed luminaires
is shown in Figure 40. The lens and phosphor coating for a green luminaire remained the
same as the first generation, but additional lenses were made using other phosphors as
well. A white luminaire with a Lumogen Orange (0.1% in J587) phosphor layer was
made to demonstrate maximum package efficiency. Another white luminaire with the
inorganic YAG:Ce phosphor was also fabricated in order to determine the effectiveness
of the ELiXIR luminaire with an diffuse phosphor. The YAG:Ce powder phosphor
53
obtained from Phosphor Technology Ltd. [55] had an average particle diameter of 6 µm.
The phosphor size was small enough that it could be uniformly suspended by vigorous
stirring in a solvent and polymer solution similar to that used for the Lumogen Yellow
phosphor. Applying the solution to the inner surface of the lens as before was adequate
for thin, low optical density films. However, the inorganic phosphor absorption is much
weaker than the organic, thus requiring a much thicker film to obtain proper color
balance. A series of thin films cast on top of each other was attempted to reach the
needed thickness. Unfortunately, after the first layer, subsequent addition of polymer-
phosphor solution would attack the cured layer, forming scattered clumps as the solvent
evaporated instead of a uniform film. Therefore, thick films of the polymer-phosphor
cast from solution were of poor quality.
To remedy this problem, a transparent curable silicone, GE RTV-615 [56] was
used to suspend the phosphor particles and form better quality thick films. The silicone
was observed to cure to a non-flowing viscosity in about 24 hours at room temperature.
The cure can be accelerated by additional heating. Phosphor was added at a
concentration of 20% by weight to the uncured silicone, and stirred with a metal spatula
until the phosphor clumps were broken up and particles were distributed evenly. The
silicone was then applied to the inner surface and manipulated by hand to obtain a
uniform coating. A heat gun was used to warm the silicone and accelerate curing, since
manipulation by hand for 24 hours is impractical, even for a graduate student. The film
cured sufficiently to resist further movement in about 2 hours.
High efficiency EZR1000 LED chips [57] were obtained from Cree. The chips
needed to be packaged for handling and characterization. The entire bottom surface of
54
the chips was covered by metallization, so they needed to be mounted to a conducting
base since the bottom of the chip is the anode. The small gold contact pads on top of the
chip require wire bond connections to a wire insulated from the conducting base. A
further consideration of the package design is to properly secure the mounted chips to the
phosphor coated lens, but have the mounted chip removable from the lens assembly for
independent testing of the chip.
A mechanical drawing for the LED base is shown in Figure 37. The base is large
enough to be easily handled by hand for characterizing the LED chip alone, but small
enough to allow inserting into a lens base for phosphor-converted luminaire
characterization. The base has cylindrical symmetry except for the internal hole for an
insulated wire to make contact with the top electrode of the LED chip, thus allowing most
of the fabrication steps to be performed quickly on a lathe. Aluminum was chosen as the
material for the base because of its high thermal and electrical conductivity, high
reflectivity, and easy machining characteristics. The UC College of Engineering machine
shop was hired for the fabrication.
After fabrication of the LED bases, LED chips needed to be attached. Aluminum
is relatively difficult to solder because of its stable oxide layer. The LED base was placed
on a hotplate to increase its temperature; a small drop of flux was applied to the Al
surface, followed by a small chunk of solder, precut from wire. A hot soldering iron tip
was placed on the Al surface near the solder to melt it, firmly attaching the solder to the
Al surface. The iron was removed and the solder solidified before a LED chip was
carefully placed on the solder drop. The hot iron was again placed on the Al surface near
the solder drop to melt it. Unfortunately, no solder tested adhered well to the LED
55
metallization. Various solder and flux combinations were tried to securely mount the
chips to the base, but all failed to securely attach the chips. Sometimes the chips seemed
secure, but the chips would always delaminate during a few seconds of sonication,
necessary for cleaning the flux.
Flux soldering was also attempted, but also failed. Flux soldering involves only
placing a drop of flux between the chip and the Al base. Raising the temperature to
300°C reflows the Au/Sn eutectic on the back of the chip, allowing it to bond to other
metallic surfaces. Use of two different fluxes, including one especially recommended for
Al, and raising temperatures well above 300°C all failed to secure the chip to the base.
Finally, a silver epoxy was found to attach the chips securely with good
conductivity. The two-part epoxy consists of ~ 70% silver by weight when cured and
was rated for temperatures up to 260°C. Since the maximum recommended junction
temperatures for the LEDs are only 150°C, the epoxy easily exceeds this requirement.
The two parts of the epoxy were mixed, and then a small amount was placed on the Al
base. The Al base was then scratched with a steel spatula to break up the surface oxide
and allow good Al to epoxy contact. The epoxy was pressed down into a thin, uniform
layer with the spatula before carefully placing a chip on the epoxy. High power chips,
having a side length of approximately 1 mm, could be placed onto the epoxy layer with
tweezers. Low power chips, having side lengths as small as 300 µm, could not be
handled with tweezers without damage. The small chips were light enough, however,
that they were found to weakly adhere to a wood stick, which was used to pick up the
chips and drop them onto the epoxy. Success rates for mounting the small chips were
above 50%, due to three possibilities when landing onto the epoxy: face up, face down,
56
0.046 dia0.125 dia
0.010
1.000 0.125
0.250
0.750
0.100 0.125
0.125
and on the side. Face down chips were discarded since epoxy adhered to the emitting
surface, face up chips were good, and chips landing on their sides could be pushed over
to face up with a probe tip under a microscope.
Figure 37. Mechanical drawing of chip mount fabricated from aluminum.
Once the chips were face up, they were pressed down into the epoxy with probe
tips under a microscope to ensure epoxy contacted the entire backside of the chip. Then,
the mounted chip assemblies were placed in an oven at 120°C for 1-2 hours for curing.
After curing, the assemblies were sonicated in isopropanol for 15 minutes for cleaning.
In contrast to the soldered chips, epoxied chips remained securely attached during
sonication.
A mechanical drawing of the lens base is shown in Figure 37. The LED base fits
into the center of the lens base, positioning the LED at the center of the luminaire. A
0.18750.500
57
0.5000.125
0.5000.1250.125
sheet of 3M Vikuiti ESR reflector film [58] was glued to the top of the lens base with a
spray adhesive. The film is a multi-layer polymer Bragg reflector with greater than 98%
reflectance across the visible spectrum [59]. Excess reflector film around the outer edges
of the lens base was trimmed with a razor blade. A hole in the center of the reflector film
3.0002.000
0.250
0.250+
3.0002.000
0.250
0.250+0.250+
Figure 38. Mechanical drawing of lens base. Note the hole in center and screw holes to position and secure the chip mount.
to expose the LED and wire was made with a sharp scalpel. Finally, the lens was
attached to the reflector film using a transparent, curable adhesive. A schematic of the
0.750 dia. 6-32 screw hole
slot for power cord
0.6406
58
Figure 39. Schematic of entire luminaire assembly including the chip mount, lens base, and lens.
Figure 40. Photograph of two second-generation ELiXIR luminaires. A white-emitting luminaire having a YAG:Ce in silicone phosphor coating is on the left, while a green-emitting luminaire having a Lumogen Yellow in Joncryl 587 polymer phosphor coating is on the right.
entire assembly is shown in Figure 39. In addition, a photograph of two completed
luminaires is shown in Figure 40.
It should be noted that the ELiXIR luminaire design is well suited to low cost
mass production, so the luminaire cost would be dominated by the pump LED. The lens
could be produced with high quality at low cost by injection molding of the PMMA
rather than polymerization of the monomer. In addition, though the luminaire here is
large (7.6 cm diameter lens) by LED standards, the design can be easily scaled to a size
59
approaching that of conventional LEDs. The high package efficiency is maintained as
long as the proportions of the luminaire, namely the rphosphor/rlens ratio, are preserved. The
package size is ultimately limited by the chip size. The phosphor distance from the chip
must be sufficiently far so that only a small fraction of converted light re-enters the chip,
where high losses occur. For example, a typical power LED chip has an area of ~ 1 mm2.
If we specify that less than 1% of phosphor light emitted from any point on the phosphor
may re-enter the chip, a minimum chip-phosphor separation of approximately
))01.0(4/(1 π , or ~ 2.8 mm is obtained. The minimum lamp diameter would be four
times this value, ~ 1.1 cm, which is approaching the size of the transparent lens
encapsulation and smaller than the heat sink diameter on a typical power LED.
ELiXIR luminaire characterization
The ELiXIR luminaires and the blue LEDs were characterized for color and total
power output. The total power output was measured by placing the device on a rotation
stage and simultaneously measuring the power and spectrum as a function of angle from
a distance of 38 cm. Care was taken to minimize reflections and subtract contributions
not coming directly from the lamp. The output power was measured with a calibrated
silicon photodiode. Output spectra were collected with a CCD spectrometer with fiber
optic cable input, which had been calibrated with the photodiode and variable
monochromatic source over the visible region. The spectra shown for the luminaires are
weighted averages over all angles.
ELiXIR luminaire results
(a) Blue LED chip
60
The LED chip with highest efficiency was found to be the Cree EZR1000 high
power chip. At 30 mA dc drive current, the wall-plug efficiency peaked at 0.360 at a
voltage of 2.809 V. Power output was 30.34 mW, average wavelength was 460.2 nm,
and the luminous efficacy was 18.9 lm/W. A summary of these results and measurement
uncertainties is shown in Table 1 and Figure 41. This chip was used to characterize the
yellowish green luminaire, the white luminaire with YAG:Ce phosphor, and the white
luminaire with orange organic phosphor. At 350 mA dc current, output power was 313
mW, but the wall-plug efficiency had dropped to 0.255.
Table 1. Summary of results for the Cree EZR1000 LED chip at 30 mA dc.
Input: 30.0 mA at 2.809 V Output: 30.34 +/- 0.08 mW Flux: 0.568 +/- 0.004 lm Efficiency: 0.360 +/- 0.001 Avg. λ: 460.2 +/- 0.1 nm Spectrum efficiency: 52.4 +/- 0.4 lm/W Luminous efficacy: 18.9 +/- 0.1 lm/W
Effective LED spectrum
0.000
5.000
10.000
15.000
20.000
25.000
400 450 500 550 600 650 700Wavelength (nm)
Opt
ical
pow
er (a
.u.)
Figure 41. Cree EZR1000 LED spectrum.
61
(b) YAG:Ce white
The white luminaire with YAG:Ce phosphor achieved 87 lm/W efficacy with a
correlated color temperature of 6805 K at 30 mA. This level of efficacy is comparable to
fluorescent lamps and the color temperature is suitable for general lighting. At 350 mA,
luminous efficacy dropped to 61 lm/W, but flux increased to 75 lumens.
Table 2. Summary of results for the white luminaire with YAG:Ce phosphor.
Input: 30.0 mA at 2.813 V Output: 19.24 +/- 0.04 mW Flux: 2.596 +/- 0.005 lm Efficiency: 0.228 +/- 0.001 Avg. λ: 535.0 +/- 0.0 nm Stokes efficiency: 0.860 +/- 0.007 Phosphor QE: 0.77 +/- 0.03 ηpηq: 0.737 +/- 0.011 Package efficiency: 0.957 +/- 0.032 Spectrum efficiency: 380.6 +/- 0.0 lm/W Luminous efficacy: 86.5 +/- 0.2 lm/W CCT: 6805 +/- 29 K CRI (Ra): 59 +/- 0
Effective lamp spectrum
0.000
2.000
4.000
6.000
8.000
10.000
12.000
400 450 500 550 600 650 700Wavelength (nm)
Opt
ical
pow
er (a
.u.)
Figure 42. Spectrum of white luminaire with YAG:Ce phosphor.
62
Since YAG:Ce conversion of blue to obtain white results in 23% conversion loss, the
effective phosphor quantum efficiency was 0.77. The package efficiency is 0.96.
Considering that conventional white YAG:Ce pcLEDs have ηp ranging from ~ 0.4 to 0.6,
implementation of the ELiXIR approach would immediately increase efficiency and flux
of conventional white LEDs by a factor of 2.4 to 1.6, respectively. The great majority of
all losses are now due to the YAG:Ce phosphor conversion of blue light. As discussed
earlier, literature values for losses in YAG:Ce conversion (23%) and total internal
reflection (13%) of blue light to obtain white total 36% [22]. Since this luminaire
achieved a package efficiency and phosphor quantum efficiency total loss of only 26%, it
is concluded that conversion loss remained constant at 23%, since the phosphor thickness
remained the same compared to the literature, but total internal reflection losses were
reduced from 13% to 3% by the hemispherical shell phosphor layer geometry. The
improvement is primarily due to the separation of chip and phosphor, because less than
0.1% of phosphor-emitted light re-enters the LED chip, compared to nearly 50% in the
conventional pcLED case. To meet OIDA roadmap targets, phosphor conversion losses
must be reduced, requiring a phosphor layer of homogeneous refractive index to
eliminate scattering losses and/or a different phosphor with higher quantum efficiency.
(c) Organic greenish yellow
As discussed earlier no material system provides high efficiency LEDs in the
green-yellow regions of the spectrum. The InGaN material system performs best in the
short wavelength violet/blue regions, while AlGaInP system performs best in the
orange/red regions.
63
A green pcLED using a blue InGaN pump LED coated with a SrGa2S4:Eu
phosphor has been demonstrated in the literature [60]. The pcLED achieved a flux of 50
lumens and efficiency of 37 lm/W at 400 mA. After accounting for ηs and ηp, blue to
green conversion efficiency (ηpcL/ηled in this work) was given as ~ 50%. The results were
still favorable compared to InGaN green LEDs.
Here, a homogeneous refractive index phosphor was used to produce a high
efficiency greenish yellow luminaire. A summary of the luminaire results appear in
Table 3. Summary of results for the greenish yellow luminaire
Input: 30.0 mA at 2.813 V Output: 23.75 +/- 0.11 mW Flux: 4.516 +/- 0.022 lm Efficiency: 0.281 +/- 0.001 Avg. λ: 547.5 +/- 0.0 nm Stokes efficiency: 0.841 +/- 0.007 Phosphor QE: 0.97 +/- 0.03 ηpηq: 0.931 +/- 0.012 Package efficiency: 0.960 +/- 0.033 Spectrum efficiency: 535.2 +/- 0.3 lm/W Luminous efficacy: 150.5 +/- 0.7 lm/W
Effective lamp spectrum
0.000
5.000
10.000
15.000
20.000
25.000
400 450 500 550 600 650 700Wavelength (nm)
Opt
ical
pow
er (a
.u.)
Figure 43. Greenish yellow luminaire spectrum (Lumogen Yellow phosphor). Notice the small amount of blue LED leakage around 460 nm.
64
Table 3 and Figure 43. The purpose of this luminaire was to demonstrate the highest
luminous efficacy possible by: (1) maximizing luminous efficacy potential by having
light emitted in the green-yellow region, where the eye is most sensitive; (2) use of a
phosphor with a high phosphor refractive index; (3) use of a high efficiency package.
The phosphor layer consisted of 0.1% Lumogen Yellow fluorescent dye dispersed in
Joncryl 587.
The luminaire achieved a luminous efficacy of 151 lm/W at 30 mA, which to the
best knowledge of the author, is a record for a solid-state device under dc conditions. In
addition, the package efficiency was 0.960, which is also believed to be a record for any
pcLED. Package losses were believed to be dominated by absorption by imperfections in
the phosphor layer. Before the device was characterized, the phosphor layer developed a
roadmap of tiny cracks to relieve stresses that developed as the last portion of solvent
gradually evaporated from the layer. These cracks provide places where total internal
reflection can occur, resulting in losses. The luminaire produced 4.52 lumens with a wall
plug efficiency of 0.281. The luminaire spectrum is capable of a luminous efficacy of
535 lm/W in the limit of 100% wall-plug efficiency. At 350 mA, luminous efficacy
dropped to 107 lm/W, but flux increased to 131 lumens.
The ELiXIR greenish yellow pcLED outperforms the reported green pcLED [60]
in efficiency and flux by nearly a factor of three. At 400 mA, this ELiXIR luminaire
achieves 145 lm at 101 lm/W, compared with 50 lm at 37 lm/W for the conventional
pcLED. Approximate efficiencies of ηpcL/ηLED (~ 0.5), ηs (~ 0.9), ηq (~ 0.9), are given for
this literature pcLED. Insertion of these values into Equation 1 gives an estimate for ηp
of 0.6. Based on package efficiencies alone, the ELiXIR luminaire should outperform the
65
literature pcLED by ~ 26%. A summary of all efficiency values discussed here along
with literature values is presented in Table 5.
(d) Organic white
The purpose of creating a white luminaire with an organic phosphor was to
demonstrate a phosphor converted white capable of maximum efficiency by eliminating
phosphor scattering effects and use of a high efficiency package. A summary of the
results is seen in Table 4 and Figure 44. For simplicity, a single phosphor, 0.1%
Lumogen Orange [33] in Joncryl 587 was used to obtain white. Lumogen Orange has the
additional advantage of having extremely high quantum efficiency, which was
determined to be virtually 100% by experiment. This helps maximize device efficiency,
but also reduces uncertainty in the calculation of package efficiency.
Package efficiency was a nearly perfect 0.990, exceeding the year 2020 OIDA roadmap
target by 4%. The package efficiency is greater than for the greenish-yellow luminaire
mainly because the phosphor layer was freshly prepared and lacked the roadmap of
cracking imperfections, and also because a smaller fraction of LED light was converted
by the phosphor.
At 30 mA, luminous efficacy was 97 lm/W and CCT was 5358 K. The luminous
efficacy was somewhat lower than anticipated, surpassing the value of the YAG:Ce white
luminaire by only 11 lm/W. The primary reason for this is that the organic phosphor
white spectrum is only capable of 320 lm/W at 100% wall-plug efficiency, while the
YAG:Ce white spectrum is capable of 381 lm/W. Wall-plug efficiency of this luminaire
is 0.305, compared to 0.360 for the blue LED alone. Losses were dominated by the
Stokes efficiency since the quantum efficiency was 1.00 and the package efficiency was
66
0.990. At 350 mA, luminous efficacy decreased to 69 lm/W and flux increased to 84
lumens.
A photograph of the luminaire in operation is shown in Figure 45. Despite the
low color rendering index (45), the different color regions of a CIE color chart are clearly
distinguishable.
Table 4. Summary of results for the white luminaire with orange organic phosphor
Input: 30.0 mA at 2.826 V Output: 25.75 +/- 0.02 mW Flux: 2.915 +/- 0.162 lm Efficiency: 0.305 +/- 0.000 Avg. λ: 536.8 +/- 4.7 nm Stokes efficiency: 0.857 +/- 0.009 Phosphor QE: 1.00 +/- 0.03 ηpηq: 0.990 +/- 0.011 Package efficiency: 0.990 +/- 0.032 Spectrum efficiency: 320.0 +/- 18.1 lm/W Luminous efficacy: 97.2 +/- 5.4 lm/W CCT: 5358 +/- K CRI (Ra): 45 +/-
Lamp spectrum
0
5
10
15
20
25
30
35
400 450 500 550 600 650 700 750wavelength (nm)
Irra
dian
ce (a
.u.)
Figure 44. Spectrum for the white luminaire with organic orange phosphor.
67
Comparison of ELiXIR luminaires with literature examples
Table 5 compares the ELiXIR luminaire efficiencies, luminous efficacies, and
luminous fluxes with those in the literature and the current best production white LED.
The conditions of 400 mA dc drive current were chosen because luminous efficacies and
fluxes were given or could be calculated for all the LEDs. In addition, this is a
Table 5. Comparison of ELiXIR luminaire efficiency and flux values to similar literature and best production LEDs at 400 mA DC drive current
pcLED
ηpcL
ηled
ηs
ηq
ηp
LE
(lm/W)
flux
(lm)
SrGa2S4 green [60] * * ~0.9 ~0.9 ~0.6 37 50
ELiXIR greenish-yellow 0.188 0.241 0.841 0.97 0.96 101 145
conventional YAG:Ce white [20] - - ~0.85 0.77 0.4-0.6 ~30 ~43
SPE YAG:Ce white [20] - - ~0.85 0.77 0.64-0.87 ~47 ~63
ELiXIR YAG:Ce white 0.153 0.241 0.860 0.77 0.96 58 77
Cree XR-E YAG:Ce white [7] - - ~0.85 0.77 - ~65** ~75**
ELiXIR organic white 0.204 0.241 0.857 1.0 0.99 65 94
dashed values (-) were not given or calculable
* ηpcL / ηled was given as ~0.5;
**assumed junction temperature of 80° C
reasonable condition for actual use, as opposed to values at 25° C and/or pulsed current
measurements often reported in literature and product data sheets.
Conclusion
A pcLED luminaire design with a package efficiency of 0.99 and quantum
conversion efficiency of 1.0 has been demonstrated. The package significantly
outperforms other pcLEDs in the literature and in commercial production. The luminaire
68
had a luminous efficiency of 69 lm/W and a luminous flux of 84 lumens at 350 mA dc
drive current. An otherwise identical device with a YAG:Ce phosphor layer had a
package efficiency of 0.96. This is the first explicit calculation of this value for this type
of LED. It exceeds the values of other white YAG:Ce pcLEDs because the
hemispherical shell shaped phosphor reduces total internal reflection losses from the
planar phosphor geometry. The YAG:Ce white luminaire had a luminous efficacy of 61
lm/W and a flux of 75 lumens at 350 mA dc drive current. This compares favorably with
the best commercial white YAG:Ce LED which achieves an artificially high 70 lm/W at
350 mA by using pulsed drive conditions to maintain an impractically low junction
temperature of 25° C to increase the chip wall-plug efficiency. The luminaires utilized a
blue pump LED with a peak wavelength near 460 nm and a wall-plug efficiency of 0.255
at 350 mA dc drive current.
Features of the organic white ELiXIR luminaire that lead to its near maximal
conversion efficiency efficiency are (in order of importance): (1) spatial separation of the
chip and phosphor; (2) high quantum efficiency phosphor with homogeneous phosphor-
layer refractive index that minimizes diffuse reflections and scattering; (3) phosphor layer
refractive index less than or equal to the encapsulating lens, located far enough from the
encapsulation exterior to eliminate whispering gallery modes; (4) use of internal
reflection to steer phosphor-converted light away from the LED chip and mirror surfaces;
(5) use of specular, high reflectivity (>98%) mirror surfaces to minimize losses for those
rays that do encounter the reflector.
69
Figure 45. Photograph of the organic phosphor white luminaire illuminating a CIE color chart and UC logo.
70
Chapter 4. Color, Efficiency, and Stability of Fluorescent Dyes in Polymer Matrices
Fabrication of Phosphor Films
Organic fluorescent dyes were dispersed into a transparent host polymer. This
offers important advantages over a neat film or dye powder alone. Quantum efficiency is
increased over a neat film because the dye concentration is reduced. Films of good
optical quality can be obtained more easily with a polymer via techniques such as spin
coating, screen printing, ink jet printing, or injection molding. The polymer can be more
easily molded into various geometries to improve light extraction.
Perylene-based fluorescent dyes are known for high quantum efficiency and
excellent stability [61], making them attractive for light emitting applications. The
perylene-based dyes studied here include Lumogen Yellow F083 [32], Lumogen Orange
F240 [33], and Lumogen Red F300 [34] available from BASF. Also studied for
comparison are Rhodamine 6G [31], a well-known efficient and stable laser dye, and
Keyplast FL Yellow 10G [36], an economical Coumarin-based fluorescent dye from
Keystone Aniline Corporation. The overlapping absorption and emission spectra
characteristic of these dyes results in a relatively small Stokes shift. This feature allows
use of relatively long wavelength blue excitation sources.
Polymethyl methacrylate (PMMA) or acrylic has nearly ideal properties for this
application. PMMA (n = 1.49) features superior optical clarity, dye solubility as high as
a few percent by weight, and low cost. Perhaps the only disadvantage of PMMA is
limited solubility and slow dissolution in solvent. Joncryl 587 [37], [38], a modified
styrene acrylic polymer, was used to retain the good properties of PMMA with improved
71
solubility in solvent. Solutions of dye and J587 polymer were produced by weighing the
appropriate amount of polymer, dye, and benzyl alcohol. The solutions contained
polymer and dye in a 100:1 ratio and were approximately ten percent solids by weight.
Thick films were used for the efficiency and color determinations. These had
thicknesses ranging from 20 to 100 µm and were fabricated by pipetting the solutions
onto microscope slide substrates on a hotplate heated to 150°C. The films were left on
the hotplate until tacky, then moved to a vacuum oven at 150°C to drive off the
remaining solvent.
Color
The phosphor films were pumped by a 455 nm LED coupled into an optical fiber.
Fluorescence spectra were collected with a calibrated Ocean Optics USB2000 fiber optic
CCD spectrometer. Fluorescence spectra are shown in Figure 13. CIE coordinates were
calculated from the dye fluorescence spectra collected for the efficiency measurement.
The coordinates of Lumogen Yellow and Lumogen Red were determined to be (0.21,
0.70) and (0.67, 0.31), respectively. These closely match the NTSC standard green and
red coordinates of (0.21, 0.71) and (0.67, 0.33), respectively as shown in Figure 46. A
monochromatic 460 nm blue, approximating a blue LED, has coordinates of (0.14, 0.06),
closely matching the NTSC standard blue (0.14, 0.08).
Keyplast Yellow, Lumogen Orange, and Rhodamine 6G, while not ideal for RGB
phosphor applications, had CIE coordinates of (0.08, 0.65), (0.57, 0.42), and (0.60, 0.39),
respectively.
A photograph of the phosphors under UV illumination is shown in Figure 47.
72
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
NTSC color gamut
AB
CD
E
F
Y
X
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
NTSC color gamut
AB
CD
E
F
Y
X
Figure 46. CIE diagram showing color coordinates of the phosphors: Keyplast yellow (A), Lumogen Yellow (B), Lumogen Orange (C), Rhodamine 6G (D), Lumogen Red (E), and a hypothetical monochromatic 460 nm source (F). The NTSC color gamut is indicated by the dashed triangle.
A B C D E
Figure 47. Photograph of phosphor-coated microscope slides under illumination with a long-wave ultraviolet lamp. From left to right, Keystone Yellow, Lumogen Yellow, Lumogen Orange, Rhodamine 6G, and Lumogen Red.
Efficiency
Films for the efficiency measurement varied in thickness, but were thick enough
to absorb >99.5% of the pump LED spectrum. The pump LED was coupled into an
73
optical fiber and the fiber output was used as the excitation source. Films were excited
from the top with a LED best matching the absorption spectrum of the dye to minimize
the amount of dye required. Fluorescence power was measured from the bottom with an
integrating sphere and calibrated silicon photodetector to collect all the emission from the
bottom surface. A diagram of the experiment setup is shown in Figure 48. Both of the
yellow dyes were pumped with an LED having a peak wavelength of 470 nm, Lumogen
Orange at 505 nm, Rhodamine 6G at 520 nm, and Lumogen Red at 530 nm. The
fluorescence spectrum was measured with a calibrated Ocean Optics S2000 fiber optic
spectrometer.
The determination of quantum efficiency is relatively simple due to the physics of
planar waveguides. It can be shown that the fraction of light emitted from inside a planar
waveguide that exits out each face is
2)cos1( c
extθ
η−
= , Equation 7
where Өc is the critical angle, defined as
1
2c n
nsin =θ , Equation 8
where n1 is the index of refraction of the waveguide and n2 is the surrounding medium,
usually air. For a planar waveguide with refractive index 1.5 surrounded by air, the
critical angle is ~ 41.8° and internal isotropic light emission has a 12.7% probability of
escaping from each the top and bottom surfaces (see Figure 49). The integrating sphere
74
Film sample < 0.5% T
LED pump source
Optical fiberIntegrating
sphere
Si photodetector PLED Pf
Figure 48. Experimental setup for quantum efficiency measurement.
captures the 12.7% emitted from the bottom. The remaining 74.6% undergoes total
internal reflection and can only escape at the edges of the waveguide.
The quantum efficiency of the phosphor can be found through the following
relation:
extSLED
fdye P
Pηη
η =, Equation 9
where Pf is the fluorescence power from the phosphor collected by the integrating sphere,
PLED is the excitation power, ηs is the Stokes efficiency as before, and ηext is the external
quantum efficiency of the planar phosphor.
The results of the quantum efficiency measurement are: Keyplast yellow 0.98,
Lumogen Yellow 0.97, Lumogen Orange 1.01, Rhodamine 6G 0.38, and Lumogen Red
0.92. The anomalously high value obtained for Lumogen Orange can be explained in two
75
ways: spectrometer calibration error or self-absorption and re-emission, thus giving total
internal reflected rays another chance to escape. The low value obtained for Rhodamine
6G is due to the incompatibility of the dye with the host polymer. Apparently 1%
concentration was too high resulting in dye molecule aggregation not visible to the eye.
Aggregation results in non-radiative energy transfer between neighboring molecules,
extending the excited state lifetime and decreasing the quantum efficiency.
The high optical density of the films complicates the experiment due to two
opposing effects: (1) self-absorption of dye emission, which reduces perceived quantum
efficiency; (2) re-emission of self-absorbed emission, which increases the perceived
quantum efficiency due to the second chance to escape the waveguide for re-emitted
photon. Because of the above factors, quantum efficiency obtained is only approximate.
In an emissive display application, however, high optical density is required to absorb
most pump light and maintain good color coordinates. This arrangement using optically
thick films accurately simulates the results likely to be encountered in an actual color-by-
blue device application.
n = 1.5
n = 1
Өc = 42°
dye molecule
n = 1.5
n = 1
Өc = 42°
dye molecule
Figure 49. Ray trace diagram of an isotropic emitter inside a planar waveguide. Only rays falling inside the cone defined by the critical angle escape the waveguide.
76
Photostability of the Dyes
Background
Photostability is the amount of pump energy absorbed per mole of dye molecules
required to reduce fluorescence by 50 percent of its initial value. In the literature, units
are usually expressed in units of gigajoules per mole (GJ/mol). Since the molecular
formulas of some commercially available dyes are proprietary, I suggest using units of
energy per unit mass rather than energy per mole. Units of MJ/g or megajoules per gram
are in general of the same order of magnitude as the traditional GJ/mol and are equivalent
to GJ/mol when the dye molecular weight is 1000 g/mol. Additionally these units make
more practical sense as dyes are generally purchased and implemented into devices by
mass, not by moles.
The photostability of organic dyes and degradation mechanisms have been
extensively researched for application to dye lasers [62], [63], [64], [65], [66], [67], [68],
[69]. Relatively few studies have studied photostabilities at milder excitation intensities
[70]. Essentially, organic dyes in excited states can react with oxygen, water, the host
material, or impurities. After chemical reaction, the dye molecules no longer fluoresce
efficiently and may or may not absorb pump excitation as efficiently.
Measured photostability values depend on many factors including the excitation
wavelength, excitation intensity, and the host matrix, which can be solvent, polymer, or
sol-gel glass, dye concentration, and presence of oxygen, water, or other impurities.
Solid-state photostabilities for a given dye are generally higher than liquid-state
photostabilities, probably because the dye molecules have less freedom and reactive
impurities have slower diffusion rates. Photostabilities in the liquid state can be
77
improved significantly by additives that quench singlet oxygen, inhibit free radicals, or
by adding dendrimers to the dye molecules to make it more difficult for reactive
impurities to reach and destroy the dye [71].
More recently, some longer lived dyes have been suggested for application to
solid state dye lasers (SSDL). SSDLs have much greater demands for photostability
since the supply of dye is not constantly replenished as in a solvent-dye laser system.
Rhodamine 590, perylene orange, and perylene red were reported to have photostabilities
of 18, 21, and 80 GJ/mol respectively [67]. In that work, dyes were cast in PMMA slabs
and pumped with 35 mJ, 6 ns pulses from a Nd:YAG laser at 532 nm focused into a 2
mm dot. This gives a pump intensity of 190 MW/cm2. Pump intensities for the proposed
solid state lamp are on the order of 1 W/cm2, or a factor of 108 less than the literature case,
so a greater photostability is expected in the solid state lamp if two photon processes are
the dominant degradation mechanism in the SSDL. Therefore, photostability values for
low intensity pumping as in displays and solid-state lighting may be significantly greater
than those under pulsed laser conditions due to the relative absence of two photon
processes. Pump intensities necessary for display applications are of the order 10-3
W/cm2, or a factor of 1011 less than the SSDL.
For solid-state lighting applications, the requisite photostability can be estimated.
Assuming a 1 W/cm2 LED excitation intensity incident on the phosphor, full conversion
of the LED light by the phosphor, a half-life of 50,000 hours, and a phosphor mass
density of 10-4 g/cm2, the required photostability is ~ 2 × 106 MJ/g.
78
Experiment
Since literature values for photostability under mild pump intensities (~1 W/cm2)
are rare, and photostabilities are very sensitive to environmental and excitation, an
experiment is necessary. To observe decay in fluorescence intensity as quickly as
possible, the effective dye mass should be minimized and the pump intensity should be
maximized. Phosphor films were spin-coated to obtain film thicknesses on the order of
100 nm. The resulting dye density is on the order of 10-7 g/cm2. Typical optical
absorption of the films was on the order of a few percent. An Omnichrome Ar ion laser
with a wavelength of 488 nm, output power of ~ 35 mW, and a beam diameter of 0.2 cm
was used to excite the phosphor films. Without a focusing lens, the beam intensity is ~1
W/cm2, but with a focusing lens, beam diameters down to ~ 140 µm were achieved,
yielding intensities up to 200 W/cm2. Beam diameters were measured using a digital
caliper with a 10 µm resolution. Illumination of a ~100 nm 1% dye doped PMMA film
that absorbs 1% at 488 nm with an Ar laser will result in a dose rate of 105 W/g, or 0.1
MW/g. If the low intensity photostability of the dye is less than the ~ 100 MJ/g literature
values as in the high intensity pulsed laser case, 50% fluorescence degradation would be
observed in less than 1000 seconds. Fluorescence intensity was monitored with a
spectrometer collecting spectra at several second intervals, while the Ar laser power was
monitored with a silicon photodetector attached to an integrating sphere. The
experimental setup is shown in Figure 50.
79
Polymer:dye thin film
Integrating sphere
Si photodetector
Ar ion laser
35 mW 488 nm
CCD spectrometer
Convex lens
Optical fiber
Figure 50. Experimental setup for determining dye photostability.
Detailed Theory
The following is a derivation of the theoretical fluorescence intensity versus time
observed for a laser beam striking a partially transparent dye-doped film. In this
treatment, a radially symmetric, constant intensity laser beam, zero dye diffusion, and
perfectly transparent host material is assumed.
From the definition of photostability, the pump dose in joules/cm2 required for a
50% fluorescence reduction is
mD 2/1 Π= , Equation 10
where Π is the dye photostability in J/g and m is the dye mass in g/cm2. Hence the dose
required for a 1/e fluorescence reduction is
2lnmD0
Π= . Equation 11
80
The dose received for a given position and time is proportional to the laser pump
intensity, IL(r), in W/cm2, the optical absorption of the dye at the pump wavelength (1-T),
and the exposure time in seconds (t).
t)T1)(r(I)t,r(D L −= Equation 12
In general, the laser pump beam intensity profile is not constant, so the dye degrades
unevenly. Areas of high pump intensity obviously decay faster. So, the fluorescence
intensity, IF, as a function of time and position is given by
)(e)0,r(I)t,r(I 0D
)t,r(D
fF
−
= , Equation 13
where
)r(I)T1()0,r(I LsqF −= ηη . Equation 14
Integration over the entire beam area gives the total laser power, PL,
∫∫∫ ==R
0LLL rdr)r(I2dA)r(IP π . Equation 15
The fluorescence intensity emitted by the dye as a function of position and time is:
)(e)r(I)t,r(I 0D
)t,r(D
LsqF
−
= ηη . Equation 16
The power density factor, ρ, is a dimensionless constant that takes into account the beam
profile’s effect on the observed fluorescence decay. The power density factor essentially
allows an arbitrary radially symmetric laser beam or fluorescence intensity profile I(r)
over area A to be replaced by a constant intensity distribution having a value of ρP/A.
Calculation of the power density factor allows simplification of the fluorescence decay
81
equations. Calculation of ρ from an intensity distribution involves evaluating the
following:
)(eRI
dr)(
re)r(I2
dr)(
reI
dr)(
re)r(I
mt)T1(I2ln
2L
R
0
mt)T1)(r(I2ln
L
R
0
mt)T1(I2ln
L
R
0
mt)T1)(r(I2ln
L
L
L
L
L
Π−
Π−
Π−
Π−
−
∫−
=
∫−
∫−
=ρ . Equation 17
The power density factor is limited to the following values:
∞<ρ≤1
For a constant or rectangular beam profile, ρ = 1. The more extreme the intensity
variation over the beam area, the larger the power density factor becomes. At the other
extreme a delta function beam profile gives ρ = ∞ because the beam intensity approaches
infinity and degradation is instantaneous. The effect of different beam profiles on the
fluorescence decay rate is shown in Figure 51.
A far-field measurement of the fluorescence power will include contributions
from the entire beam area, so integration over the beam area is required:
dr)(
re)r(I)T1(2rdr)t,r(I2dA)t,r(I)t(PR
0
mt)T1)(r(I2ln
Lsq
R
0fff
L
∫−
−ηπη=∫π∫∫ == Π−
. Equation 18
In the case of a constant beam profile,
0r
)r(IL =∂
∂ ,
Pf(t) decay is purely exponential and simplifies to
)(eI)T1(R)t(P m
t)T1)(r(I2ln
Lsq2
f
L
Π−−
−ηηπ= . Equation 19
82
y = 1.86E-03e-4.37E-05x
R2 = 9.95E-01y = 1.94E-03e-2.15E-05x
R2 = 1.00E+00
y = 1.94E-03e-1.99E-05x
R2 = 1.00E+00
0.0E+00
5.0E-04
1.0E-03
1.5E-03
2.0E-03
2.5E-03
0 5000 10000 15000 20000 25000Time (min.)
Inte
nsity
(a. u
.)
Constant
Gaussian
Sinusoidal
0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
0 0.2 0.4 0.6 0.8 1Normalized position (r/R)
Inte
nsity
(W/c
m2 )
ConstantGaussianSinusoidal
Figure 51. Calculated fluorescence intensity versus time for three beam profiles (top) and beam profiles (bottom). The beam profiles are normalized such that the intensity integrated over a circular beam of radius R are equal. Note the decay rate for the Gaussian profile is more than double that of the other two because higher discrepancy between the maximum and minimum intensities results in a higher power density factor. What is measured in the experiment, however is only a fraction of the fluorescence power
and its decrease with time, so the exponential component is what is being observed, and
the pre-factor is unimportant. For real beam profiles, IL(r) is not generally constant, so
83
decay is not purely exponential, but an exponential fit can yield a very high correlation
coefficient.
The fluorescence power decay as a function of time can be described by replacing
the laser beam intensity with the beam power and power density factor in Equation 19
and normalizing the fluorescence power to the initial fluorescence power:
)(e
)0(P)t(P mA
t)T1(P)2(ln
f
fL
Π−
−=
ρ
. Equation 20
So, the fluorescence half-life is:
L2/1 P)T1(
mAtρ−
Π= . Equation 21
It is not necessary to record data for a full half life however, because the time in seconds
to reach an arbitrary decay value, δ, can be calculated with
2lnP)T1()ln(m
tL
11
1 ρ−Π
= δ−δ− , 10 << δ . Equation 22
For example, for δ = 0.5, t1- δ simplifies to the equation for the half life; for 5% decay, δ =
0.05; and for 90% decay, δ = 0.9.
The lifetime of the dye in a device application can be determined by the following
equation:
BLEtI
ItI
t extsqLLdevice
LLdevice π
ηηη== , Equation 23
where I denotes excitation intensities, t denotes the lifetimes (half-lives), ηext is the
external quantum efficiency which depends on the device geometry, LE is the luminous
efficacy of the dye spectrum in lm/W, and B is the required device brightness in cd/m2.
84
10
100
1000
10000
100000
1000000
450 500 550 600 650 700 750Peak emission wavelength (nm)
Phot
osta
bilit
y (M
J/g)
LY
LO
R590
LR
KY
Figure 52. Photostability results for several different samples and measurement conditions for fluorescent dyes Keyplast Yellow (KY), Lumogen Yellow (LY), Lumogen Orange (LO), Rhodamine 6G (R590), and Lumogen Red (LR) versus peak emission wavelength.
Photostability Results
A chart of the photostability results is seen in Figure 52. The average
photostabilities for the dyes in MJ/g are as follows: Keyplast Yellow 150, Lumogen
Yellow 3000, Lumogen Orange 10,000, Rhodamine 6G 900, and Lumogen Red 30,000.
These values fall 2-3 orders of magnitude below the necessary photostability for
application in solid-state lighting.
There is a clear trend of increasing photostability with longer emission
wavelength. The only exception is Rhodamine 6G which may have a deceptively low
photostability in this experiment since dye aggregation is suspected based on the quantum
efficiency results. As discussed previously, dye aggregation can have the effect of
lengthening the excited state lifetime, thus increasing the probability of photochemical
reactions that can destroy dye molecules.
85
Display Application
Organic dyes in a polymer matrix have been suggested for use as color converters
for display applications [72].
A summary of the results of the effective quantum efficiency and accelerated
aging experiments is shown in Table 6. The spectrum efficiency is the luminous
efficiency of the dye spectrum in the case of 100% power conversion efficiency. Values
are highest in the yellow where the eye is most sensitive and lower at longer red and
shorter blue wavelengths. The effective quantum efficiency is greater than 0.9 for all
dyes except Rhodamine 6G. Stokes efficiency is the average efficiency of converting a
460 nm blue photon to a lower energy phosphor photon. The Stokes efficiency decreases
as emission wavelength increases. The required excitation intensity is the 460 nm pump
intensity necessary to achieve a luminance of 300 cd/m2, a common display brightness. It
assumes a 12.7% external quantum efficiency, which is the emission from one side of a
planar phosphor having a refractive index of 1.5. The highest intensity is required for red,
since the spectrum has the lowest luminous and Stokes efficiencies. Lifetimes are
calculated from Equation 23. They range from ~1,000 hours for Rhodamine 6G to
47,000 hours for Lumogen Red. Lumogen Yellow would limit display life to 7,000 hours
under these conditions.
86
Table 6. Summary of dye characteristics for display applications. From [73].
spectrum effective required luminous quantum Stokes excitation estimated
Dye efficacy efficiencya efficiency intensityb lifetimeb (lm/W) (mW/cm2) (khr)
Keyplast Yellow 285 0.98 0.91 2.9 2 Lumogen Yellow 445 0.97 0.87 2.0 7 Lumogen Orange 401 1.01 0.76 2.4 11 Rhodamine 6G 429 0.38 0.77 5.9 1 Lumogen Red 110 0.92 0.71 10.3 47 460 nm bluec 41 - - 2.3 -
a) potentially higher than actual quantum efficiency due to self absorption and subsequent re-emission b) based on monochromatic 460 nm excitation, 300 cd/m2 brightness, and external efficiency of 12.7%
c) hypothetical monochromatic blue source
The combination of a blue backlight with a green phosphor based on Lumogen
Yellow and a red phosphor based on Lumogen Red as shown in Figure 53 is an attractive
option. The phosphors provide an excellent match to the NTSC color coordinates, have
quantum efficiencies above 90%, and would have a display half-life limited to 7,000
hours by the green phosphor.
Conclusion
Though organic fluorescent dyes in a polymer matrix have been found to be
insufficiently stable for use in high intensity solid-state applications, their high quantum
efficiency and excellent color matching to NTSC coordinates make them attractive for
use in color-by-blue display applications.
RGB emission
Patterned phosphors
Blue pump light
Blue backlight
RGB emission
Patterned phosphors
Blue pump light
Blue backlight
Figure 53. Schematic of color-by-blue display. From [73].
87
Chapter 5. Nearly Index-Matched Luminescent
Glass-Crystal Composites
Motivation: pcLED application
Because phosphors consisting of an organic dye in a polymer matrix have a
homogeneous refractive index and easily achieve quantum efficiencies greater than 90%,
they are capable of greater efficiency pcLEDs than the inorganic powder phosphors
shown earlier. Their limited photostability, however, limits their application to relatively
low luminance applications such as in displays. Use of these materials for photo-pumped
solid-state lasers or even compact illumination sources is impractical because device
lifetimes would be unacceptably short. A question arises: can the otherwise excellent
optical properties of organic dyes in a polymer matrix be replicated in a stable, long-lived
inorganic material system? This chapter proposes a new material system, nearly index-
matched luminescent glass-crystal composites (NIMLGCC), as a solution to this problem.
The glass-crystal composite (GCC) proposed here consists of a high index glass
and a crystalline inorganic powder phosphor. There are many motivations for combining
these two materials. Crystalline phosphor powders have long been used as in
illumination, display, and scintillation applications. Like organic fluorescent dyes, they
are capable of quantum efficiencies approaching unity. Powder phosphors are much
more economical than bulk single crystals of the same material. Currently, optical
glasses with a refractive index in excess of 2.1 in the visible region are commercially
available from Sumita. The primary applications of high index glass are eyeglass lenses
88
~ 10 µm
transparent glass host
crystalline powder phosphor
Figure 54. Diagram of glass-crystal composite structure. The material consists of a crystalline inorganic powder phosphor in a transparent glass matrix.
and “crystal,” used for wine glasses, vases, and jewelry. High percentages of heavy
metal ions such as those of lanthanum or lead are used to raise the index above the typical
values around 1.5 for more common glasses such as pure silicon dioxide and borosilicate
glasses. Yttrium aluminum oxide (YAG) melts at 1900°C, roughly 1000°C higher than
typical glass processing temperatures, so no thermal degradation of the phosphor would
be expected.
An initial demonstration of a LGCC used YAG:Ce as the phosphor with
borosilicate glass obtained from microscope slides essentially reproduced the material
found in [74] and [75]. The YAG:Ce is of special significance because of its suitability
in producing white light for lighting applications. A glass microscope slide was ground
into a fine powder with a mortar and pestle and added to a portion of YAG:Ce phosphor
powder having an average size of 6 µm, then mixed thoroughly. A vial of the powder is
shown in Figure 55. A portion of this powder was placed on a 2 inch fused silica wafer
and pressed into a flat layer approximately 1 mm thick with a metal spatula.
89
Figure 55. Photograph of YAG:Ce / borosilicate glass LGCC powder in a vial (left). Formation of glassy LGCC globules after firing the powder at 900°C for one hour in air (right).
The wafer was placed into a tube furnace at 600°C for an hour in air before being taken
out for observation. Subsequently, this process was repeated at 700, 800, and 900°C.
For temperatures less than 900°C, the result remained mostly powdery in nature,
with granules adhering slightly to each other, but cracking easily under light pressure
applied with a metal spatula. After firing at 900°C, the result was qualitatively different.
Figure 56. Scanning electron micrograph of a GCC consisting of 30% La2Zr2O7 crystals in soda borosilicate glass formed by hot pressing at 620° C for 1 hour. From [76].
90
Figure 57. Photoluminescence spectrum of YAG:Ce in borosilicate glass LGCC. Excitation was at 325 nm. Photograph showing LGCC with a spot under excitation.
The rough-textured powdery cake of material was transformed into densely packed
globules with a smooth, glassy surface seen in Figure 55. The result looked promising,
but the retention of strong luminescence from the YAG:Ce needed to be confirmed.
Results of sample excitation with a HeCd laser at 325 nm are shown in Figure 57.
The main peak at 540 nm with a bandwidth of ~100 nm is due to the YAG:Ce phosphor,
with the smaller peak near 410 nm due to the borosilicate glass. A photograph in the
same figure shows intense yellow emission despite the ambient room light. So, efficient
luminescence from the YAG:Ce phosphor has been maintained, but absorption and
luminescence from the glass must be managed to maximize the quantum efficiency of the
composite. Borosilicate glass absorbs strongly below ~350 nm, so significant absorption
of the HeCd laser light is not surprising. For lighting applications, borosilicate glass
absorption would be negligible because of the longer excitation wavelength peaked near
450 nm.
0
50000
100000
150000
200000
250000
300000
350000
350 450 550 650 750Wavelength (nm)
Inte
nsity
(cou
nts)
91
microscopic scale
macroscopic scale
index-matched glass and crystal
glass and crystal of different refractive index
Figure 58. Microscopic (top) and macroscopic property (bottom) comparison of glass-composites where the refractive indices of the two components differs significantly (left), and are closely matched (right).
The refractive index for YAG ranges from approximately 1.85 at 450 nm to
approximately 1.83 at 650 nm. This wavelength interval is of particular interest to
lighting applications using blue LEDs peaked near 450 nm with the YAG:Ce phosphor,
with emission peaked at ~ 540 nm with a fwhm of ~ 100 nm. As an example (Figure 59),
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
400 450 500 550 600 650 700Wavelength (nm)
inde
x m
ism
atch
(%)
Figure 59. Refractive index mismatch between YAG crystal and a mixture of 80% LAH60 and 20% LAH65 Ohara high-index glasses.
92
Figure 60. Refractive index versus dispersion for glasses with refractive index greater than 1.6 manufactured by Schott. From [77].
a mixture of two high index glasses from Ohara can match the YAG:Ce refractive index
within 0.4% across the visible spectrum. As a demonstration of the large number of high
index glasses available, see Figure 60, an Abbe diagram for glasses available from
Schott.
Schott alone offers more than 10 glasses having refractive index above 1.8, with three of
those above 1.9.
93
Solid-state laser application
In addition to solid-state lighting, NIMLGCCs may find application in efficient
solid-state lasers, especially in the visible region. The homogeneous refractive index and
high density would result in minimal scattering loss. The transparency of the glass would
yield low absorption loss. High gain is achieved through high phosphor densities.
Optical pumping is possible with commercially available 405 nm diode lasers. Single
crystal gain medium performance could be achieved for the price of glass.
A schematic of the proposed solid-state NIMLGCC laser structure is shown in
Figure 61 and consists of the following components: a 405 nm InGaN technology pump
laser, input mirror, NIMLGCC gain medium, and output mirror. The 405 nm InGaN
technology pump laser beam is incident on the input mirror having high transmission at
visible output @ 532 nm
808 nm fiber-coupled AlGaAs laser
input mirror
output mirror YAG:Nd
KTP crystal
405 nm fiber-coupled InGaN laser
input mirror
output mirror NIMLGCC
visible output
Figure 61. Schematic comparison of the most common solid-state visible laser, frequency doubled Nd:YAG (top) and the proposed solid-state visible laser using an NIMLGCC as a gain medium (bottom).
94
405 nm, but high reflectivity at the longer lasing wavelength of the specific NIMLGCC
gain medium. The pump beam continues into the cavity, where it encounters strong
absorption in the NIMLGCC gain medium. The NIMLGCC then emits radiation
characteristic of the specific phosphor with high quantum efficiency. At sufficient pump
intensity, a population inversion in the NIMLGCC will be achieved, and converted laser
light can exit the output mirror having high reflectivity at 405 nm and moderate
reflectivity at the lasing wavelength.
Currently, wavelengths of visible solid-state lasers are somewhat restricted.
InGaN technology semiconductor lasers are limited to the violet and blue, with 405 nm
peak wavelength the most common. Frequency-doubled Nd:YAG are relatively cheap
and efficient, but are restricted to a single wavelength, 532 nm. AlGaInP lasers are
confined to the deep red.
Fluorescence collection application
The strong tendency of planar phosphors to trap light by total internal reflection
provides a challenge to efficient light extraction in pcLEDs, but this property can be
utilized for fluorescence collection. Fluorescence collectors are sheets of a fluorescent
material such as dyes in a polymer matrix or NIMLGCCs that are exposed to a light
source such as the sun. The incident light is absorbed inside the sheet and re-emitted as
fluorescence. Much of this fluorescence is trapped by total internal reflection and is
efficiently guided to the edges, where it can be harvested by photovoltaic cells to produce
electrical power. The appeal of fluorescence collectors is that light can be collected over
a large area and a significant fraction of this can be recovered over a relatively small area
at the edges. This can greatly reduces the area and cost of photovoltaic cells necessary to
95
n = 1.8
n = 1
Өc = 33.7°
phosphor particle
n = 1.8
n = 1
Өc = 33.7°
phosphor particle
incident light
solar cell
transmitted light
NIMLGCC sheet
n = 1.8
n = 1
Өc = 33.7°
phosphor particle
n = 1.8
n = 1
Өc = 33.7°
phosphor particle
incident light
solar cell
transmitted light
NIMLGCC sheet
Figure 62. Schematic of fluorescence collector operation.
generate a given amount of power. A schematic of the application of NIMLGCC to
fluorescence collection is shown in Figure 62.
96
Chapter 6. Obtaining Accurate Spectra and Power Measurements
Determination of lamp power outputs, spectra, and phosphor quantum efficiencies
require well-calibrated measurement instruments to avoid unreasonable results.
Unreasonable results include quantum efficiencies greater than 1.0 and white lamp power
efficiencies greater than that of their blue or violet LED pump source. Accurate,
repeatable measurements of device power outputs and spectra require a calibrated
photodetector, a calibrated spectrometer consistent, and a calibrated reference lamp
source. Maintenance of the calibration requires recalibration of the photodetector every
year, while the spectrometer is recalibrated at every use by taking a spectrum of the
reference lamp.
The photodetector used was a model 1830C Newport optical power meter with a
1.00 cm2 silicon detector. Initially, the detector was found to be out of calibration
through inconsistencies in experimental results, so it was returned to Newport for
recalibration.
An Ocean Optics LS-1 CAL tungsten-halogen lamp was returned to the factory
for recalibration. Calibration data consisted of intensity (µW/cm2/nm) values every 10 to
50 nm. The LS-1 CAL lamp includes a fiber optic connector for repeatable coupling to
the optical fiber attached to the spectrometer. The lamp spectrum, which was found to
approximate the spectrum of a 2250 K blackbody, is shown in Figure 63.
Figure 63 provides a dramatic demonstration of why spectrometer calibration is
critical. The spectrum of the LS-1 CAL was taken with an Ocean Optics USB2000 fiber
97
0
1000
2000
3000
4000
5000
6000
300 400 500 600 700 800
Wavelength (nm)
Inte
nsity
(cou
nts/
pow
er) spectrometer
lamp spectrum
Figure 63. Comparison of known spectrum of calibrated Ocean Optics LS-1 CAL incandescent lamp (L0(λ)) in arbitrary power units (dashed line) and lamp spectrum obtained with an Ocean Optics USB2000 fiber optic spectrometer before any calibration (solid line).
spectrometer and compared with the known lamp spectrum. The spectra were adjusted
such that good agreement between the lamp spectrum and spectrometer reading was
obtained for wavelengths from 300 to ~ 460 nm. From approximately 460 to 620 nm, the
spectrometer far over-represents the amount of light received, while far under-
representing the amount of light received at wavelengths greater than 650 nm. This is
clearly unacceptable for accurate broadband power measurements.
98
0
10
20
30
40
50
60
70
80
90
100
350 450 550 650 750
Wavelength (nm)
Inte
nsity
(pow
er/c
m2 /n
m)
power meterspectrometercorrection
Figure 64. Response of Newport power meter and Ocean Optics fiber spectrometer to monochromatic light across the visible spectrum. The correction spectrum is the factor the spectrometer response must be amplified by in order to give a calibrated spectrum in units of optical power per unit area per unit wavelength interval. Power conversion correction necessary to convert spectrum from USB2000 to calibrated optical power units for data in Figure 63.
Initial calibration of the fiber spectrometer was accomplished using the calibrated
Newport power meter with silicon detector and a variable source of monochromatic light.
The variable monochromatic source was from a Perkin Elmer Lambda 900 UV/VIS/NIR
spectrometer with bandwidth set to 2.0 nm. The source was varied from 380 to 750 nm
with power measurements taken at various wavelengths with results indicated in Figure
64. The power meter was then removed and the optical fiber for the spectrometer placed
in the beam path. Spectra were collected at the same wavelength values as before, with
99
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
350 450 550 650 750Wavelength (nm)
L0(λ
)*C
(λ)
(Ene
rgy
units
)
Figure 65. L0(λ)C(λ) versus wavelength.
the results also shown. At this point, a correction function C(λ), could be constructed,
which when multiplied by the spectrometer reading D(λ), gives the correct spectrum in
power per unit wavelength interval units. C(λ) was calculated by first dividing the power
meter reading by the spectrometer response D(λ) at each wavelength. These data were
piecewise divided into six different wavelength ranges which were fit with a polynomial
up to sixth order. A continuous correction function C(λ) was generated by patching
together the polynomial fits from each wavelength range.
The procedure to collect a calibrated spectrum on a given day is as follows. First,
the LS-1-CAL lamp is turned on and allowed to warm up for 20 minutes. Second, a
spectrum of the LS-1-CAL lamp, L(λ), is collected. Next, the spectrum of the device to
be characterized, D(λ), is collected. The corrected device spectrum, Dc(λ), can now be
found from the relation:
100
0
2000
4000
6000
8000
10000
12000
350 450 550 650 750Wavelength (nm)
Inte
nsity
(pow
er/c
m2 /n
m)
Figure 66. Factory-calibrated intensity data for the LS-1-CAL incandescent lamp (open squares) and the lamp spectrum obtained with the Ocean Optics spectrometer after calibration with the Newport power meter and variable monochromatic source (solid line).
)()()()()( 0 λλ
λλλ DC
LLDc = Equation 24
The product L0(λ)C(λ) was determined on the day the spectrometer was calibrated with
the power meter and variable monochromatic source and remains constant Figure 65.
L(λ) is collected on the same day as the spectrum to be characterized, D(λ).
The factory calibration method also involves normalizing the spectrometer
response to the lamp source with each use. The number of data points (18) used for
calibration is insufficient, however, using only the factory lamp calibration data plotted in
Figure 66. The spectrometer response varies too quickly to be accurately reconstructed
by data often with intervals of 20 to 50 nm between points. This wide spacing of data
points allows fitting of the response with a lesser order polynomial fit, but results in much
larger response errors at certain wavelengths.
101
In contrast to the factory recommended calibration software and method, this
calibration method has given consistent, reasonable results. It corrects for day to day
variation in the spectrometer response by comparison with a known spectrum each use.
Because sufficient data points are used in correcting spectrometer response, errors at
many wavelength intervals are reduced. Its accuracy relies only on the initial calibration
of the silicon photodetector and consistency of the lamp spectrum since the initial
calibration.
102
Chapter 7. Conclusion/Summary
This dissertation demonstrated the following: a record efficiency phosphor-
converted LED package for solid-state lighting, a full color LED-phosphor system for
emissive displays, and proposed nearly index-matched glass-crystal composite material
system with applications to solid-state lighting, solid-state lasers, and sunlight
concentration.
A phosphor-converted LED package for solid-state lighting with record light
extraction efficiency was demonstrated. The high efficiency was made possible by
separating the phosphor from the LED chip, elimination of phosphor scatter, index-
matching the encapsulation to the phosphor, and steering phosphor emission out of the
device with internal reflection. The package efficiency exceeds the OIDA roadmap goal
for the year 2020, necessary for reaching 200 lm/W efficacy.
The three color LED-phosphor system for emissive displays consisted of
combining a blue LED backlight with green and red phosphors consisting of fluorescent
dyes in a polymer matrix. The phosphors had quantum efficiencies greater than 90%,
closely match NTSC standards and have a minimum lifetime of 7,000 hours at 300 cd/m2.
Most notably, when combined with increasingly efficient blue LEDs, this technology can
lead to improved efficiencies for LCDs, thus extending battery life for portable devices.
A borosilicate glass-YAG:Ce composite phosphor was demonstrated and nearly
index-matched glass-crystal composites, were proposed: specifically, YAG:Ce phosphor
powder in a high index glass matrix that had a refractive index mismatch of less than
0.4% across the visible spectrum. The most obvious application of this material is
implementation in a phosphor-converted LED package described above for improved
103
solid-state lighting efficiency. This composite and others of this type may find
applications in solid-state lighting, solid-state lasers, and sunlight concentration.
104
References
[1] J. Y. Tsao, "Solid-state lighting lamps, chips, and materials for tomorrow," in IEEE Circuits & Dev. Mag. vol. May/June, 2004, pp. 28-37.
[2] J. Y. Tsao, "An OIDA Technology Roadmap Update 2002," Optoelectronics Industry Development Association, 2002.
[3] CIE, "CIE Publ. No. 17.4/IEC Pub. 50(845) International Lighting Vocabulary," 1989.
[4] CIE, "CIE 13.3-1995 Method of measuring and specifying colour rendering properties of light sources," 1995.
[5] A. Ryer, "Light Measurement Handbook," 1998. [6] "Luxeon K2 Emitter Technical Datasheet DS51," Lumileds 2006. [7] "Cree XLamp XR-E data sheet," Cree 2007. [8]
http://upload.wikimedia.org/wikipedia/commons/thumb/1/10/Gluehbirne_2_db.jpg/180px-Gluehbirne_2_db.jpg.
[9] http://content.answers.com/main/content/wp/en-commons/thumb/0/0e/300px-Leuchtstofflampen-chtaube050409.jpg.
[10] Lumileds, "Luxeon K2 Brochure," 2006. [11] A. J. Steckl, J. Heikenfeld, and S. C. Allen, "Light wave coupled flat panel
displays and solid-state lighting using hybrid inorganic/organic materials," IEEE/OSA J. Disp. Tech., vol. 1, pp. 157-166, 2005.
[12] G. O. Mueller and R. Mueller-Mach, "Illumination grade white LEDs," Proc. SPIE, vol. 4776, pp. 122-130, 2002.
[13] D. A. Steigerwald, J. C. Bhat, D. Collins, R. M. Fletcher, M. O. Holcomb, M. J. Ludowise, P. S. Martin, and S. L. Rudaz, "Illumination with solid state lighting technology," IEEE J. Sel. Top. Quantum Electron., vol. 8, pp. 310-320, 2002.
[14] A. Zukauskas, R. Vaicekauskas, F. Ivanauskas, R. Gaska, and M. S. Shur, "Optimization of white polychromatic semiconductor lamps," Appl. Phys. Lett., vol. 80, pp. 234-236, 2002.
[15] H. Ries, I. Leiko, and J. Muschaweck, "Optimized additive mixing of colored light-emitting diode sources," Opt. Eng., vol. 43, pp. 1531-1536, 2004.
[16] T. Taguchi, Y. Uchida, and K. Kobashi, "Efficient white LED lighting and its application to medical fields," Phys. Stat. Sol., vol. 201, pp. 2730-2735, 2004.
[17] J. S. Kim, P. E. Jeon, Y. H. Park, J. C. Choi, H. L. Park, G. C. Kim, and T. W. Kim, "White-light generation through ultraviolet-emitting diode and white-emitting phosphor," Appl. Phys. Lett., vol. 85, pp. 3696-3698, 2004.
[18] N. Narendran and Y. Gu, "Life of LED-based white light sources," IEEE/OSA J. Display Tech., vol. 1, pp. 167-171, 2005.
[19] R. Mueller-Mach, G. O. Mueller, M. R. Krames, and T. Trottier, "High power phosphor-converted light-emitting diodes based on III-nitrides," IEEE J. Sel. Top. Quantum Electron., vol. 8, pp. 339-345, 2002.
[20] N. Narendran, Y. Gu, J. P. Freyssinier-Nova, and Y. Zhu, "Extracting phosphor-scattered photons to improve white LED efficiency," Phys. Stat. Sol., vol. 202, pp. R60-R62, 2005.
105
[21] K. Yamada, Y. Imai, and K. Ishii, "Optical simulation of light source devices composed of blue LEDs and YAG phosphor," J. Light Vis. Environ., vol. 27, pp. 70-74, 2003.
[22] Y. Zhu, N. Narendran, and Y. Gu, "Investigation of the optical properties of YAG:Ce phosphor," Proc. SPIE, vol. 6337, pp. 63370S-1-63370S-8, 2006.
[23] W. Goetz, "White Lighting (Illumination) with LEDs," Lumileds Lighting 2004. [24] R. Mueller-Mach, G. Mueller, and M. Krames, "Phosphor materials and
combinations for illumination grade white pcLEDs," Proc. SPIE, vol. 5187, pp. 115-122, 2004.
[25] N. Narendran, "Improved performance white LED," Proc. SPIE, vol. 5941, pp. 5941108-1-5941108-6, 2005.
[26] H. Luo, J. K. Kim, E. F. Schubert, J. Cho, C. Sone, and Y. Park, "Analysis of high-power packages for phosphor-based white-light-emitting diodes," Appl. Phys. Lett., vol. 86, pp. 243505-1-243505-3, 2005.
[27] "Luxeon star power light source technical datasheet DS23," Lumileds. [28] J. Edmond, A. Abare, M. Bergman, J. Bharathan, K. Lee Bunker, D. Emerson, K.
Haberern, J. Ibbetson, M. Leung, P. Russel, and D. Slater, "High efficiency GaN-based LEDs and lasers on SiC," J. Crystal Growth, vol. 272, pp. 242-250, 2004.
[29] M. R. Krames, O. B. Shchekin, R. Mueller-Mach, G. O. Mueller, G. Harbers, and M. G. Craford, "Status and future of high-power light-emitting diodes for solid-state lighting," IEEE J. Sel. Top. Quantum Electron., 2007.
[30] R. V. Steele, "The story of a new light source," Nature Photonics, vol. 1, pp. 25-26, 2007.
[31] U. Brackmann, "Lamdachrome Laser Dyes, 3rd Edition," Lambda Physik 2000. [32] "Lumogen F Yellow 083," BASF Corporation 2005. [33] "Lumogen F Orange 240," BASF Corporation 2005. [34] "Lumogen F Red 300," BASF Corporation 2005. [35] "Lumogen F Violet 570," BASF Corporation 2005. [36]
http://www.dyes.com/ProductLine/Plastics+and+Polymers/Fluorescents+for+Plastics+and+Polymers/ColorChart/802-082-70.html.: Keystone Corporation, 2006.
[37] "Joncryl 587 solid flake acrylic polyol for conventional solids polyurethane coatings," Johnson Polymer, LLC 2003.
[38] "Coatings Joncryl Selection Guide," Johnson Polymer 2005. [39] G. Electric, "RTV615, RTV655, RTV656 High Strength Transparent Silicone
Rubber Compounds Data Sheet," 2007. [40] S. Guha, R. A. Haight, N. A. Bojarczuk, and D. W. Kisker, "Hybrid organic-
inorganic semiconductor-based light-emitting diodes," J. Appl. Phys., vol. 82, pp. 4126-4128, 1997.
[41] N. A. Bojarczuk, S. Guha, and R. A. Haight, "Hybrid organic-inorganic semiconductor light emitting diodes," U. S. P. T. O., Ed. U. S. A., 1999.
[42] M. S. Shur and A. Zukauskas, "Solid-state lighting: toward superior illumination," Proc. IEEE, vol. 93, pp. 1691-1703, 2005.
[43] K. Sakuma, K. Omichi, N. Kimura, M. Ohashi, D. Tanaka, N. Hirosaki, Y. Yamamato, R.-J. Xie, and T. Suehiro, "Warm-white light-emitting diode with
106
yellowish orange SiAlON ceramic phosphor," Opt. Lett., vol. 29, pp. 2001-2003, 2004.
[44] R. Mueller-Mach, G. O. Mueller, P. J. Schmidt, D. U. Wiechert, and J. Meyer, "Nitridosilicates, a new family of phosphors for color conversion of LEDs," Proc. SPIE, vol. 5941, pp. 59410Z-1-59410Z-8, 2005.
[45] J. Heikenfeld, S. C. Allen, and A. J. Steckl, "A novel fluorescent display using light wave coupling technology," Soc. Info. Disp. Symp. Dig., vol. 35, pp. 470-473, 2004.
[46] N. K. Patel, S. Cina, and J. H. Burroughes, "High-efficiency organic light-emitting diodes," IEEE J. Sel. Topics Quantum Electron., vol. 8, pp. 346-361, 2002.
[47] B. W. D'Andrade and J. J. Brown, "Organic light-emitting device luminaire for illumination applications," Appl. Phys. Lett., vol. 88, pp. 192908-1-192908-3, 2006.
[48] W. X. Li, R. Jones, S. C. Allen, J. Heikenfeld, and A. J. Steckl, "Maximizing Alq3 OLED internal and external efficiencies: charge balanced device structure and color conversion outcoupling lenses," IEEE/OSA J. Disp. Tech., vol. 2, pp. 143-152, 2006.
[49] S. C. Allen, J. Heikenfeld, and A. J. Steckl, "Hybrid inorganic/organic devices for solid state white lighting applications," in 12th International Workshop on Inorganic and Organic Electroluminescence, Toronto, Canada, 2004, pp. 53-55.
[50] S. C. Allen and A. J. Steckl, "ELiXIR: Solid-state luminaire with enhanced light extraction by internal reflection," IEEE/OSA J. Disp. Tech., 2007.
[51] P. Schlotter, R. Schmidt, and J. Schneider, "Luminescence coversion of blue light emitting diodes," Appl. Phys. A, vol. 64, pp. 417-418, 1997.
[52] P. Schlotter, J. Baur, C. Hielscher, M. Kunzer, H. Obloh, R. Schmidt, and J. Schneider, "Fabrication and characterization of GaN/InGaN/AlGaN double heterostructure LEDs and their application in luminescence conversion LEDs (LUCOLEDs)," Mat. Sci. Eng., vol. B59, pp. 390-394, 1999.
[53] O. N. Ermakov, M. G. Kaplunov, O. N. Efimov, I. K. Yakushchenko, M. Y. Belov, and M. F. Budyka, "Hybrid organic-inorganic light-emitting diodes," Microelec. Eng., vol. 69, 2003.
[54] H.-F. Xiang, S.-C. Yu, C.-M. Che, and P. T. Lai, "Efficient white and red light emission from GaN/tris-(8-hydroxyquinolato) aluminum/platinum(II) meso-tetrakis(pentafluorophenyl) porphyrin hybrid light-emitting diodes," Appl. Phys. Lett., vol. 83, pp. 1518-1520, 2003.
[55] http://www.phosphor-technology.com/ [56] "RTV615, RTV655, RTV656 High Strength Transparent Silicone Rubber
Compounds Data Sheet," General Electric 2007. [57] "EZR1000 LEDs (data sheet)," Cree 2006. [58] "Vikuiti Enhanced Specular Reflector (ESR)," 3M 2003. [59] J. Anderson, C. Schardt, J. Wheatley, P. Watson, C. Leatherdale, A. Ouderkirk, J.
Schultz, M. Meis, V. Savvateev, and W. J. Bryan, "LED backlight design factors for large format LCDs," in 26th International Display Research Conference, Kent, Ohio, 2006, pp. 93-95.
107
[60] R. Mueller-Mach, G. O. Mueller, T. Trottier, M. R. Krames, A. Kim, and D. Steigerwald, "Green phosphor-converted LED," Proc. SPIE, vol. 4776, pp. 131-136, 2002.
[61] G. Seybold and G. Wagonblast, "New perylene and violanthrone dyestuffs for fluorescent collectors," Dyes and Pigments, vol. 11, pp. 303-317, 1989.
[62] J. Widengren and R. Rigler, "Mechanisms of photobleaching investigated by fluorescence correlation spectroscopy," Bioimaging, vol. 4, pp. 149-157, 1996.
[63] A. K. Sheridan, A. R. Buckley, A. M. Fox, A. Bacher, D. D. C. Bradley, and I. D. W. Samuel, "Efficient energy transfer in organic thin films--implications for organic lasers," J. Appl. Phys., vol. 92, pp. 6367-6371, 2002.
[64] R. Reisfeld, R. Gvishi, and Z. Burshtein, "Photostability and loss mechanism of solid-state red perylimide dye lasers," J. Sol-Gel Sci. Tech., vol. 4, pp. 49-55, 1995.
[65] M. D. Rahn, T. A. King, A. A. Gorman, and I. Hamblett, "Photostability enhancement of pyrromethene 567 and perylene orange in oxygen-free liquid and solid dye lasers," Appl. Opt., vol. 36, pp. 5862-5871, 1997.
[66] M. D. Rahn and T. A. King, "High-performance solid-state dye laser based on perylene-orange-doped polycom glass," J. Mod. Opt., vol. 45, pp. 1259-1257, 1998.
[67] M. D. Rahn and T. A. King, "Comparison of laser performance of dye molecules in sol-gel, polycom, ormosil, and poly(methyl methacrylate) host media," Appl. Opt., vol. 34, pp. 8260-8271, 1995.
[68] D. P. Pacheco, W. H. Russell, and H. R. Aldag, "Solid-state dye lasers pumped directly by diode lasers," SPIE Proc. 5332, pp. 180-188, 2004.
[69] M. J. Holmes and C. E. Mungan, "Photobleaching of cresyl violet in poly(methyl methacrylate)," J. Young Investigators, vol. 10, 2004 2004.
[70] A. Dubois, M. Canva, A. Brun, F. Chaput, and J.-P. Boilot, "Photostability of dye molecules trapped in solid matrices," Appl. Opt., vol. 35, pp. 3193-3199, 1996.
[71] M. Matsui, M. Wang, K. Funabiki, Y. Hayakawa, and T. Kitaguchi, "Properties of novel perylene-3,4:9,10-tetracarboxidiimide-centred dendrimers and their application as emitters in organic electroluminescence devices," Dyes and Pigments, vol. 74, pp. 169-175, 2006.
[72] S. A. Swanson, G. M. Wallraff, J. P. Chen, W. Zhang, L. D. Bozano, K. R. Carter, J. R. Salem, R. Villa, and J. C. Scott, "Stable and efficient fluorescent red and green dyes for external and internal conversion of blue OLED emission," Chem. Mater., vol. 15, pp. 2305-2312, 2003.
[73] S. C. Allen and A. J. Steckl, "Efficiency and stability of perylene-based dyes for emissive displays," in 26th International Display Research Conference, Kent, Ohio, 2006, pp. 59-62.
[74] S. Fujita, S. Yoshihara, A. Sakamoto, S. Yamamoto, and S. Tanabe, "YAG glass-ceramic phosphor for white LED (I): background and development," Proc. SPIE, vol. 5941111-1-594111-7, 2005.
[75] S. Tanabe, S. Fujita, S. Yoshihara, A. Sakamoto, and S. Yamamoto, "YAG glass-ceramic phosphor for white LED (II): luminescence characteristics," Proc. SPIE, vol. 5941, pp. 594112-1-594-112-6, 2005.
108
[76] S. Pace, V. Cannillo, J. Wu, D. N. Boccaccini, S. Seglem, and A. R. Boccaccini, "Processing glass-pyrochlore composites for nuclear waste encapsulation," J. Nuc. Mat., vol. 341, pp. 12-18, 2005.
[77] "Abbe diagram of Schott glasses," Schott, www.schott.com/optics_devices, 2006.