The Causes of Color Shift in LED Devices

Abstract

Chromaticity stability is increasingly being applied as a reliability metric for solid-state lighting devices, and a myriad of factors can influence chromaticity at the LED, lamp, and luminaire level. At the LED level, changes in phosphor emission efficiency caused by heating or other local effects can produce either blue or yellow shifts, whereas changes in the phosphor emission peak shape can produce either green or red shifts. Changes in the die-silicone and silicone-phosphor interfaces can also have a significant impact on chromaticity by altering the refractive properties of materials along the optical path. These mechanisms occur in all LED package architectures (i.e., high-power LEDs, mid-power LEDs, and chip-on-board) although the relative impact depends on phosphor chemistry (e.g., warm white vs. cool white), binder composition, and operational conditions (e.g., current and temperature). In addition, aging of lenses, reflectors, and other optical materials often produces a filtering of blue photons that causes a yellow color shift at the lamp and luminaire level. In many instances, these shifts can be predicted and controlled provided a fundamental understanding of the physics of failure of the materials used in constructing the SSL device is known. This presentation will provide experimental data from major luminaire components that was gathered from accelerated life testing on commercial SSL luminaires. This data gives insights into color shift mechanisms occurring in commercial SSL products and also forms the basis for predictive models of lumen maintenance and chromaticity stability.

Full text

www.rti.org
RTI International is a registered trademark and a trade name of Research Triangle Institute.
The Causes of Color
Shift in LED Devices
Presentation to Phosphor Global Summit
March 8, 2016
Lynn Davis, Bobby Yaga,
Cortina Johnson, and Karmann Mills
Contact Info: ldavis@rti.org
919-316-3325
1

Outline
Basics of Color
LED-Level Color Shift
Luminaire-Level Color Shift
Modeling Color Shift
Conclusions
Acknowledgements
2

Factors That Might Contribute to Luminaire Color Shift
LED
–LED type (e.g., HPLED, MPLED,
COB)
–Phosphor chemistry (e.g., CCT)
–Junction temperature
–Drive current
Luminaire
–Lens materials
–Reflector materials
–Heat Dissipation
Electronics
–Stability
Lens
Reflector
MPLED
< 0.5 W
HPLED
> 2 W
COB
> 5 W

Typical Phosphor-Converted LED (pcLED)
Phosphor +
Binder Layer
Silicone lens
High-Power LED
Mid-Power LED
From Tuttle & McClear, LED Magazine Feb. 2014.
1976 Color Space
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70
v'
u'
Blue Emitter
Yellow Emitter

Typical Phosphor-Converted LED (pcLED)
1976 Color Space
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70
v'
u'
Blue Emitter
Yellow Emitter
Yellow Shift
For color shifts along the blue-yellow line, the
peak shapes and peak maxima are
unchanged, but the relative intensities change.
Blue Shift
•Possibly caused by a drop in yellow
emissions, especially if not at phosphor
saturation.
•Characterized by large drop in v

and modest
negative shifts in u

.
Yellow Shift
•Possibly caused by an increase in yellow
emissions (e.g., greater down conversion) or
a drop in blue emissions.
•Characterized by large increase in v

and
modest positive shifts in u

.

Typical Phosphor-Converted LED (pcLED)
1976 Color Space
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70
v'
u'
Blue Emitter
Yellow Emitter
For color shifts that deviate from the blue-
yellow line, the peak shapes and/or peak
maxima do change.
Green Shift
•Possibly caused by oxidation of a nitride
phosphor that produces a shift to lower lof
phosphor emissions.
•Characterized by a negative shift in u

and
modest changes in v

.
Red Shift
•Rare for pcLED systems.
•Characterized by a positive shift in u

and
modest changes in v

.
Red Shift

Lumileds Luxeon Cool White LEDs (Nominal CCT = ~4200 K)
•Cool white LEDs provide insights in the die-
level effects, due to higher stability on YAG
phosphor.
•Low ambient temperatures generally
produce a blue shift. Practically no
activation energy required for blue shift.
•In general, high ambient temp. produces a
small blue shift followed by a strong yellow
shift.
•Activation energy for the yellow shift was
calculated at 1.26 eV. Consistent with
ohmic contact degradation results from
Zanoni.
0.355
0.357
0.359
0.361
0.363
0.365
0.367
0.369
0.371
0.373
0.375
0.364 0.366 0.368 0.370 0.372 0.374 0.376
y
x
Lumileds Rebel HP LED, Cool White, 1000 mA
Planckian locus
55 C, 1000 mA
85 C, 1000 mA
105 C, 1000 mA
Color Shift
Direction
Du’v’ = < 0.001
Color Shift
Direction
Du’v’ = 0.003
7
From Lumileds Luxeon Rebel Dataset DR-05

Major Findings from CALiPER Studies
Constant operation @ 45oC in special
chamber. Photometrics measured weekly.
–7,500 hr test time for A-lamps
–13,925 hr test time for PAR 38 lamps
Lumen maintenance performance of LED
lamps generally exceeded conventional
lighting technologies.
Parametric failure was rare
–13% of A-lamps for L70
–17% of A-lamps for color shift
–0% of PAR38 lamps for L70
–8% of PAR38 lamps for color shift
Part-to-part variation found in some lamp
models, but not common.
8
15 different A-lamp models, 60W Eq.
32 different LED PAR38 models
32 different LED PAR38 models

PAR38 Lamp Models (CALiPER 20.4 and 20.5)
32 different LED models
Luminous Flux Range: 440 –1530
lm
Power: 8.6 –24.5 W
Luminous Efficacy Range: 47 - 99
L/W
Test started in March 2013.
Simple optical design with reliance on
clear secondary optics. Minimal use
of reflectors.
Lumen and chromaticity maintenance
dominated by LED behavior.
9
12-64 12-66

60 W Equivalent A-Lamp Models
15 different LED models
Rated Luminous Flux Range:
800 –850 lm
Rated Power Range: 9.5 –
13.5 W
Rated Luminous Efficacy
Range: 59 –86 LPW
Test started in January 2014.
Complex optical designs to
achieve isotropic radiation
pattern. Extensive use of
diffusers and opaque lenses.
Optical plastic degradation may
impact lumen and chromaticity
maintenance.
10

LED Packages Breakout in CALiPER Studies
11
60W Eq A Lamps PAR38 Lamps
CALiPER 20.5
HB-LED 7 18
COB LED 0 7
Plastic Leaded Chip
Carrier (PLCC) 6 6
Hybrid 1 1
Remote Phosphor 1 0
Total 15 32

Chromaticity Shift Modes (CSMs)
12
In the lamp samples, the initial chromaticity shift is
generally in the blue direction.
Some chips initially shift in a green direction, then blue.
After the initial blue shift, chromaticity often changed in a
systematic progression through difference chromaticity
shift modes.
CSM1 –continuation of blue shift
CSM2 –shift off of blue-yellow line in green direction
CSM3 –shift in yellow direction
CSM4 –double shift first in the yellow direction & then blue.

CSM-1 Behavior in PAR38 Lamps
13
Characterized by a persistent
shift in the blue direction.
Rate of shift is rapid at first but
slows down at time progresses.
LEDs often become more
efficiency when first turned on
producing more blue photons.
Possible causes of CSM-1
behavior:
–Drop in quantum efficiency of
phosphor
0.510
0.511
0.512
0.513
0.514
0.515
0.516
0.517
0.518
0.244 0.245 0.246 0.247 0.248 0.249 0.250 0.251 0.252
v'
u'
12-86 PAR38 Lamp in Extended 45 C Test (PNNL)
Planckian locus
Sample 1
Sample 2
Sample 3
Sample 4
Sample 5
Direction of shift

CSM-1 Behavior in PAR38 Lamps
14
Samples displaying only CSM-1 also showed high lumen
maintenance (> 92%).
Chromaticity shift was small, generally 1-2 SDCM.
LED board temperatures were ~25 –35 C above ambient.
0.245
0.246
0.247
0.248
0.249
0.250
0 2,000 4,000 6,000 8,000 10,000 12,000 14,000
u' Value
Time (hours)
Time Variation of u' Chromacity for PAR38 Lamps with Blue Shift
12-64
12-86
0.511
0.512
0.513
0.514
0.515
0.516
0 2,000 4,000 6,000 8,000 10,000 12,000 14,000
v' Value
Time (hours)
Time Variation of v' Chromacity for PAR38 Lamps with Blue Shift
12-64
12-86

CSM-2 Behavior in PAR38 Lamps
15
Direction of shift deviates from
Blue-Yellow line in the green
direction.
Rate of shift is rapid at first but
slows down as time progresses.
Examination of the spectral
changes demonstrates that the
emission peak of the phosphor is
shifting to lower wavelength by <
5 nm.
0.511
0.512
0.513
0.514
0.515
0.516
0.517
0.518
0.519
0.241 0.242 0.243 0.244 0.245 0.246 0.247 0.248 0.249
v'
u'
12-92 PAR38 Lamp in Extended 45 C Test (PNNL)
Planckian locus
Sample 1
Sample 2
Sample 3
Sample 4
Sample 5
Direction of shift

CSM-2 Behavior in PAR38 Lamps (more examples)
0.512
0.513
0.514
0.515
0.516
0.517
0.518
0.519
0.520
0.242 0.243 0.244 0.245 0.246 0.247 0.248 0.249 0.250
v'
u'
12-99 PAR38 Lamp in Extended 45 C Test (PNNL)
Planckian locus
Sample 1
Sample 2
Sample 3
Sample 4
Sample 5
0.518
0.519
0.520
0.521
0.522
0.523
0.524
0.525
0.526
0.254 0.255 0.256 0.257 0.258 0.259 0.260 0.261 0.262
v'
u'
12-65 PAR38 Lamp in Extended 45 C Test (PNNL)
Planckian locus
Sample 1
Sample 2
Sample 3
Sample 4
Sample 5
16

CSM-3 Behavior in PAR38 Lamps
17
Initial shift is in the blue direction,
followed by a reversal to a yellow
shift.
Time of the reversal varies
depending on operation conditions
and LED.
0.517
0.518
0.519
0.520
0.521
0.522
0.523
0.524
0.525
0.526
0.527
0.246 0.248 0.250 0.252 0.254
v'
u'
12-100 PAR38 Lamp in Extended 45 C Test (PNNL)
Planckian locus
Sample 1
Sample 2
Sample 3
Sample 4
Sample 5
A
B
C
BB/BABC/BBYB/YAYC/YB
Lamp
model
Average
Average
Average
Average
12-75 1.026 0.974 0.978 0.983
12-81 1.010 0.963 0.923 0.977
12
-
100
1.019 0.779 1.001 0.928
B-Blue peak max Y-Yellow phosphor max

CSM-3 Behavior in PAR38 Lamps (more examples)
0.516
0.517
0.518
0.519
0.520
0.521
0.522
0.523
0.524
0.245 0.246 0.247 0.248 0.249 0.250 0.251 0.252 0.253
v'
u'
Planckian locus
Sample 1
Sample 2
Sample 3
Sample 4
Sample 5
12-75 PAR38 Lamp Extended 45 C Test (PNNL)
0.514
0.515
0.516
0.517
0.518
0.519
0.520
0.521
0.522
0.242 0.243 0.244 0.245 0.246 0.247 0.248 0.249 0.250
v'
u'
12-81 PAR38 Lamp in Extended 45 C Test (PNNL)
Planckian locus
Sample 1
Sample 2
Sample 3
Sample 4
Sample 5
18

Possible Cause of CSM-3 Behavior
19
High temperature & CTE
mismatches produces stress at
die-phosphor interface.
High temperature can also
degrade the mechanical
compliance of binder in phosphor
layer.
Result is cracking and
delamination in the phosphor
layer which changes the optical
path of blue photons.
Reference, DOE Webinar, LED
Color Stability –10 Important
Questions, 2014.

CSM-4 Behavior in PAR38 Lamps
20
Short, initial shift in the blue
direction, followed by a reversal
to a yellow shift, followed by a
second blue shift.
Time of the reversal varies
depending on operation
conditions and LED.
CSM-4 behavior was only
observed in lamps with PLCC
LED packages suggesting that
it is associated with some
plastic molding resins.

CSM-4 Behavior in PAR38 Lamps (another example)
0.237
0.238
0.239
0.240
0.241
0.242
02000 4000 6000 8000 10000 12000 14000
u

Time (hours)
12-74 PAR38 LED Lamp in Extended 45C (PNNL)
0.495
0.497
0.499
0.501
0.503
0.505
0.507
0.509
0.511
02000 4000 6000 8000 10000 12000 14000
v

Time (hours)
12-74 PAR38 LED Lamp in Extended 45C (PNNL)
21

Summary of Color Shift Behavior of PAR38 Lamps
22
The major CSM for HB-LEDs
is CSM-3.
The major CSM for PLCC
packages is CSM-4.
CSM-1 and CSM-2 is found
in some HP-LED & COBs.
Possible that CSM-3
behavior will occur with
longer test time or more
aggressive conditions.

Summary of Color Shift Behavior of Retail A Lamps
23
Much less reversal in color
shift direction.
–Shorter test duration
–Power per LED is lower
Small green shift is more
evident in the first 24 hr than
in PAR38s.
Major CSM for HPLED is
CSM-1.
Three instances of CSM-4 in
a PLCC package.
0
1
2
3
4
5
6
CSM-1 CSM-2 CSM-3 CSM-4 Complex
Number of LED Lamp Models
Color Shift Mode (CSM)
Color Shift Modes for CALiPER Retail A Lamps
COB
HPLED
PLCC
Hybrid
Remote

Summary of Color Shift in pcLEDs
Color shift at the LED package level
depends on many factors including:
–Package type
–Materials of construction
–Operating conditions
In many cases, initial color shift is a
small blue shift. This may be the only
shift observed under very mild
conditions or short times.
At higher operating conditions or longer
times, a yellow shift occurs and will
continue for some LED packages
designs.
24

Approach
System-level approach consisting of both accelerated life tests
(ALT) and modeling of both entire luminaires and key system
components such as LEDs, drivers, and optical elements.
ALT TestingALT Testing
Modeling
Lifetime Perf
Modeling
Lifetime Perf
Validation in
Real World
Validation in
Real World
Background
Literature
Background
Literature
6” downlights have been chosen as
representative luminaires because they
combine several desirable attributes:
•Low cost
•Readily available and widely used
•Multi-generational products
•Incorporate many design features
HBLEDs, mid-power, and hybrid LEDs
Remote phosphor and proximate phosphor
Physics of
Failure
Degradation
Mechanisms
Use
Environment

Accelerated Testing of Luminaires
RTI has studied the aging characteristics of more 500 SSL devices
–Individual LEDs
–Lamps
–Luminaires (mostly 6” downlights)
Teardown analysis has been performed on many parts and the aging
characteristics of different components studied.
In accelerated aging tests (such as temperature & humidity), changes in different
system components can be studied throughout the test.
26

27
0
20
40
60
80
100
120
0 500 1000 1500 2000 2500 3000 3500
Lumen Maintenance (%)
75/75 Exposure Time (hours)
6" Downlight-Warm White
138-Control
148
149
150
151
152
0.40
0.41
0.42
0.43
0.44
0.45
0.44 0.45 0.46 0.47
y
x
6" Warm White Downlight in 7575
Planckian locus
Luminaire #149
Luminaire #150
Luminaire #151
Luminaire #152
Direction of
Color Shift
Temperature & Humidity Testing of Warm White LEDs
•Aging of the warm white (2900 K) HPLEDs in temperature and humidity (T-H) causes a drop in
luminous flux after a short delay.
•Fast decrease in luminous flux for warm white phosphors.
•A yellow shift occurs initially followed by a prolonged green shift (Du’v’ ~ 0.010 @ 3000 hr in
75/75).

Luminaire Level: Stable Optical Materials
28
0
0.002
0.004
0.006
0.008
0.01
0.012
0.014
0.016
0.018
400 450 500 550 600 650 700 750 800
Spectral Power Distribution (W/nm)
Wavelength (nm)
Luminaire #145
0 hr
500 hr
1000 hr
2000 hr
3000 hr
4000 hr
0
0.002
0.004
0.006
0.008
0.01
0.012
0.014
0.016
0.018
400 450 500 550 600 650 700 750 800
Luminaire #159
0 hr
500 hr
1000 hr
1500 hr
2000 hr
2500 hr
Luminaire #145
•Lumen maintenance @ 2K hr: 60%
•Color shift @ 2K hr: 0.014
•Blue peak intensity decreases some but
yellow peak intensity decreases sharply.
BLUE SHIFT.
•Small increase in intensity around 520 nm
Luminaire #159
•Lumen maintenance @ 2K hr: 90%
•Color shift @ 2K hr: 0.006
•Blue peak intensity decreases more than
the yellow peak. YELLOW SHIFT.
•Minimal change in phosphor emission
peak shape.
75oC and 75% RH Testing

0.360
0.362
0.364
0.366
0.368
0.370
0.372
0.374
0.376
0.378
0.380
0.372 0.374 0.376 0.378 0.380 0.382 0.384
y
x
6" Neutral White Downlight in 7575
Planckian locus
Luminaire #158
Luminaire #159
Luminaire #160
Luminaire #161
Neutral White: Lumen Maintenance & Chromaticity
29
Yellow Shift
•Aging of the neutral white (4000 K) HPLEDs in temperature and
humidity (T-H) causes a drop in luminous flux after a short delay.
•A yellow shift also occurs (Du’v’ ~ 0.006 @ 3000 hr in 75/75).
0
20
40
60
80
100
120
0 500 1000 1500 2000 2500 3000 3500
Lumen Maintenance (%)
75/75 Exposure Time (hours)
6" Downlight-Natural White
129-Control
158
159
160
161
162

0
20
40
60
80
100
120
0 500 1000 1500 2000 2500 3000 3500
Lumen Maintenance (%)
75/75 Exposure Time (hours)
6" Downlight-Natural White
129-Control
158
159
160
161
162
0.01
0.011
0.012
0.013
0.014
0.015
0.016
0.017
400 450 500 550 600 650 700 750 800
Spectral Radiant Flux (W/nm)
Wavelength (nm)
Luminaire #159
0 hr
500 hr
Spectral Changes, Incubation Period
•Small increase in
luminous flux occurs
initially.
•Phosphor emissions
increase slightly while
blue emissions drop
slightly.
•Net result is a yellow shift.
30

0
20
40
60
80
100
120
0 500 1000 1500 2000 2500 3000 3500
Lumen Maintenance (%)
75/75 Exposure Time (hours)
6" Downlight-Natural White
129-Control
158
159
160
161
162
0
0.002
0.004
0.006
0.008
0.01
0.012
0.014
0.016
0.018
400 450 500 550 600 650 700 750 800
Spectral Radiant Flux (W/nm)
Wavelength (nm)
Luminaire #159
0 hr
500 hr
1500 hr
2500 hr
Spectral Changes, Growth of Microvoids and Loss of Luminous Flux
31
•Luminous flux decreases
after an initial delay.
•Phosphor emissions
decrease at a slower rate
than blue decreases.
•Net result is a yellow shift.
•No change in spectral
positions and peak shapes.

Luminaire Level: Optical Materials Change (6” Downlight)
32
0.350
0.354
0.358
0.362
0.366
0.370
0.374
0.378
0.340 0.344 0.348 0.352 0.356 0.360 0.364 0.368 0.372 0.376 0.380
y
x
Cree CR6 (CW, New Design) in 8585
Planckian locus
Luminaire #103
0 hr
Yellow 1000 hr
500 hr •Sharp drop in emission intensity at
all wavelengths due to absorption
from the lens and reflectors.
•Peak maximum for both blue and red
emissions is shifting.
85oC and 85% RH Test
0.000
0.002
0.004
0.006
0.008
0.010
0.012
0.014
0.016
400 450 500 550 600 650 700
Spectral Radiant Flux (W/nm)
Wavelength (nm)
Luminaire #103
0 hr
500 hr
1000 hr
Polycarbonate Lens

Luminaire Level: Optical Materials Change (6” Downlight)
33
•Replacing lens and reflectors restores
most of emission intensity, although
there is a still a small color shift.
•Blue peak returns to original position
but red peak is still shifted indicating a
change in red LED during test.
85oC and 85% RH Test
0.000
0.002
0.004
0.006
0.008
0.010
0.012
0.014
0.016
400 450 500 550 600 650 700
Spectral Radiant Flux (W/nm)
Wavelength (nm)
Luminaire #103
0 hr
500 hr
1000 hr
1500 hr
0.350
0.354
0.358
0.362
0.366
0.370
0.374
0.378
0.340 0.344 0.348 0.352 0.356 0.360 0.364 0.368 0.372 0.376 0.380
y
x
Cree CR6 (CW, New Design) in 8585
Planckian locus
Luminaire #103
0 hr
Yellow 1000 hr
500 hr
1500 hr
New Optics

Modeling Chromaticity Shift
34
•Chromaticity shift at the LED level is complex and depends on many factors. These
factors generally produce color shifts smaller than 0.003.
•Larger color shifts are often irreversible and may be predictable.
•HPLEDs: yellow shift
•MPLEDs: blue shift
•COBs: yellow shift (assumed)
•Junction temperature and drive current play a significant role.
• Maybe possible to define “safe” operational zones using LM-80 data.

Representative Examples of Safe Operational Zones
35
HPLEDs
0
50,000
100,000
150,000
200,000
250,000
300,000
020 40 60 80 100 120 140 160
Estimate Time to 7-Step Color Shift (hr)
Junction Temperature ( C)
0
20,000
40,000
60,000
80,000
100,000
120,000
140,000
160,000
020 40 60 80 100 120 140 160
Estimate Time to 7-Step Color Shift (hr)
Junction Temperature ( C)
MPLEDs
•Temperature and current are dominant contributors to color shift.
•Blue photon flux (i.e., current) may have a secondary impact for some
MPLEDs.

Conclusions
36
In many instances, chromaticity shifts in PAR38 and A lamps
often proceeds in a systematic progression with the LED
having a significant impact. Initial shift is often in the blue
direction.
After initial blue shift, chromaticity shift can proceed along
different chromaticity shift modes
–Continued shift along the blue direction (CSM-1)
–Shift off the blue-yellow line in the green direction (CSM-2)
–Reversal of chromaticity shift toward the yellow direction (CSM-3)
–Complex shift involving a yellow shift followed by second blue shift in PLCCs
(CSM-4)

Timing and occurrence of different CSMs depends on
–LED package and materials
–LED operational conditions (Tj, current, operational time, on/off cycles)
–Ambient conditions (Temperature, humidity, contaminants)
Lens, reflectors, and other materials in the optical path
can have an impact.
Knowing where your product lies on this chromaticity shift
progression is critical to achieving desired chromaticity
stability.
37

Acknowledgements
Data and samples of the PAR38 lamps was obtained from Dr. Michael Royer at Pacific
Northwest National Laboratory.
This material is based upon work supported by the Department of Energy under Award
Number DE-EE0005124.
Disclaimer: This report was prepared as an account of work sponsored by an agency of
the United States Government. Neither the United States Government nor any agency
thereof, nor any of their employees, makes any warranty, express or implied, or
assumes any legal liability or responsibility for the accuracy, completeness, or
usefulness of any information, apparatus, product, or process disclosed, or represents
that its use would not infringe privately owned rights. Reference herein to any specific
commercial product, process, or service by trade name, trademark, manufacturer, or
otherwise does not necessarily constitute or imply its endorsement, recommendation,
or favoring by the United States Government or any agency thereof. The views and
opinions of authors expressed herein do not necessarily state or reflect those of the
United States Government or any agency thereof.