Prepared by

Next Generation Lighting Industry Alliance
LED Systems Reliability Consortium

Acknowledgments
The Next Generation Lighting Industry Alliance wishes to acknowledge the valuable contributions of the U.S. Department of Energy and its support for Dr. Lynn Davis of RTI International, whose reliability testing work contributed significantly to the knowledge base for this document. We are grateful to the
members of the LED Systems Reliability Consortium (LSRC), who offered their considerable time and expertise over the past three years to the development of this document.
Terry Clark, Finelite
Lynn Davis, RTI International
Mark Duffy, GE*
Jim Gaines, Philips*
Monica Hansen, LED Lighting Advisors**
Eric Haugaard, Cree*
Steve Paolini, Telelumen
Morgan Pattison, SSLS, Inc.**
Clark Robinson, National Energy Technology Laboratory**
Sanwal Sarraf, Lumentek Global
Wouter Soer, Philips*
Willem van Driel, Philips*

Table of Contents

• Introduction
• LED Lifetime and Lumen Maintenance
• Chromaticity Stability
• Traditional Lighting Technologies
• LED Lighting
• Chromaticity Consistency
• LED Packages
• Chromaticity Shift Mechanisms
• High-Power LEDs
• Mid-Power LEDs
• COB LEDs
• Phosphor Degradation
• Remote Phosphor LED Modules
• Optical System Impacts
• Optics and Diffusers
• White Reflective Materials
• Summary
• Path Forward
• List of Acronyms

The emergence of solid state lighting (SSL) with its high efficiencies and long lifetimes has led to the potential for significant energy and cost savings to our nation once wide-scale adoption occurs. Light emitting diodes (LEDs) are the heart of SSL lighting products and can provide long lifetimes that last well
beyond 50,000 hours of operation, much longer than most conventional light sources. The end of life for all lighting technologies is signaled by the loss of light, but this may be less evident for LED luminaires, where the light output may continuously fade or the color may slowly shift to the point where low light output or an unacceptably large color change constitutes practical failure.
As integrated lamps and luminaires appeared on the market, it was at first assumed that one could estimate the lumen depreciation of the LED packages to describe the degradation characteristics of the integrated lighting product. While the lifetime of an LED source is one important indicator of LED luminaire life, it would be misleading to rate the entire LED luminaire based solely on the LED source. Now, after further research, it is understood that electronics failures in the driver or degradation of optical components can often occur long before LED lumen depreciation causes failures. Lifetime claims should take into account the whole luminaire system, not just the LEDs. A system reliability model that integrates the failure mechanisms in the various luminaire subsystems would create a much more accurate lifetime claim from LED luminaire manufacturers.
To address the challenge of developing accurate lifetime claims, the SSL Program of the U. S. Department of Energy (DOE) together with the Next Generation Lighting Industry Alliance (NGLIA) formed the creation of an industry consortium, the LED Systems Reliability Consortium (LSRC), to coordinate activities and foster improved understanding. The LSRC has published three editions of the document LED Luminaire Lifetime: Recommendations for Testing and Reporting,
1 in which they reviewed studies intended to identify potential failure modes and provide additional understanding of product life. The resulting conclusions were that numerous other subsystems and components in a luminaire introduce other potential failure modes which will affect, and may actually dominate, the determination of system lifetime. Work by the LSRC and other funded R&D by the DOE SSL program is focused on understanding the various degradation mechanisms to enable the development of new models so that system reliability can be confidently understood, modeled, predicted, and communicated.

LED packages rarely fail abruptly (i.e., instantaneously stop emitting light), but rather experience parametric failures such as degradation or shifts in luminous flux, color point (chromaticity coordinates), color rendering index (CRI), or efficacy. Of these parametric shifts, lumen depreciation has received the most attention because it was previously thought that the degradation of lumen output of the LED source itself would be the prime determinant of lifetime for the completed product. While it is now understood that this is not the case, lumen maintenance is still used as a proxy for LED lamp or luminaire lifetime ratings, largely due to the availability of standardized methods for measuring and projecting LED package lumen depreciation. The useful life of an LED package is often cited as the point in time where the luminous flux output has declined to 70% of its starting value or L70. For products with lifetimes of many years or even decades, failures may be very slow to appear under normal operation. In 2008, the Illuminating Engineering Society (IES) published IES LM-80, which is an approved method for measuring the lumen maintenance of solid-state (LED) light sources, arrays, and modules.2
The LM-80 test method has been recently updated to reflect the experience and knowledge gained by the LED industry. The LM-80-08 procedure required measurements of lumen output and chromaticity for a representative sample of products to be taken at least every 1,000 hours, for a minimum of 6,000 hours. Luminous flux and chromaticity shifts are to be measured for three different LED case temperatures: 55o C, 85o C and a third temperature to be
selected by the manufacturer. A newer version, LM-80-15, has undergone changes to the testing method, which now requires only two different case temperature, one of which should be 55°C or 85°C (commonly used case temperatures for industry testing to support direct product comparisons of testing
results).
Many researchers have put a great deal of effort into devising a way to project the time at which L70 will be reached for an LED package in a luminaire, and IES has documented a forecasting procedure, IES TM21, 3 which uses the LM-80 test data for the lumen maintenance projections (a minimum of 6,000 hours
of test data is required). The LM-80 data (luminous flux vs. test hours) for the LEDs tested is averaged and an exponential curve fit is applied to the data; the results of the curve fit are used to calculate a lumen maintenance lifetime projection. This technical memorandum stipulates that any projection may not exceed a set multiple (depending on sample size statistics) of the actual hours of LM-80 testing data taken, which helps avoid exaggerated claims.
With the development of IES TM-21 for projecting lumen maintenance, experts agreed the projecting method should use the trend (over a sufficient period of time) of single case temperature testing data. There is a separate projection method in TM-21 based on using two tested case temperatures and interpolating data between the tested case temperatures. Thus, the requirement of testing three case temperatures is not completely necessary and leads to an unnecessary testing burden on LED manufacturers. (LM-80-15 reduced the number of test temperatures.)
It should be noted that LM-80 measurements are taken with the LED packages operating continuously in a temperature-controlled environment, where the solder point and ambient air temperature are at equilibrium. This does not necessarily reflect real-world operating conditions, so there may not be a perfect match between predictions based on laboratory test results and practical experiences with lamps and luminaires in the field. Nevertheless, lumen maintenance projections can help sophisticated users compare products, as long as their limitations are properly understood.

When LEDs are installed in a luminaire or system, there are many additional factors that can affect the rate of lumen depreciation or the likelihood of catastrophic failure. These include temperature extremes, humidity, chemical incursion, voltage or current fluctuations, failure of the driver or other electrical components, damage or degradation of the encapsulant material covering the LEDs, damage to the interconnections between the LEDs and the fixture, degradation of the phosphors, and yellowing of the optics. In addition, abrupt semi-random short-term failures may be observed due to assembly, material, or design defects. More information on system level lifetime can be found in LSRC’s LED Luminaire Lifetime: Recommendations for Testing and Reporting.

While lumen maintenance has dominated discussions about LED lifetime, the color stability (also known as chromaticity stability) is another important performance attribute that can be a barrier to purchase or lead to unmet expectations of LED lighting. Shifts in color and appearance are a regular part of our
lives, whether it is fading paint, fabric colors, or lighting. Color shift in lighting has always occurred in traditional lighting technology, but has gained more prominence with LED lighting due to its long operating life of 10 years or more in many applications. Traditional lighting technology, such as halogen, fluorescent, or metal halide technology, experiences color shifts. Frequent relamping every few years is required due to catastrophic failures or lumen depreciation and this mitigates the impact of the color shift of these lighting technologies.

The color of light can be represented using chromaticity coordinates to describe its hue and saturation. A pair of chromaticity coordinates corresponds to a unique color of light; two sources with the same chromaticity coordinates should theoretically appear the same. Chromaticity diagrams to represent the
different color space have been developed and standardized by the Commission Internationale de l’Eclairage (CIE). The most commonly used chromaticity diagrams are the CIE 1931 chromaticity diagram Yellow Blue Green Green Blue Red (a) (b) 7 using (x, y) coordinates to specify chromaticity and the CIE 1976 chromaticity diagram using (u’, v’) coordinates. Examples of the chromaticity diagrams are shown in Figure 1. The 1976 CIE diagram has been used more extensively by the LED industry in recent years to describe chromaticity changes with its advantage of describing a linear color space, which allows for a more intuitive determination of chromaticity differences or shifts (Δu’v’). Note: Δu’v’ describes the magnitude of a chromaticity change, but not the direction of shift; this is the value provided in the LM-80-08 reports. The newer LM-80-15 reports do require reporting of the individual chromaticity coordinates (u’ and v’) instead of the total shift (Δu’v’). LED package manufacturers have been shifting to this new LM-80-15 reporting requirement with newer LED product lines.
The importance of chromaticity stability varies by application. For example, a high degree of chromaticity stability is crucial for light sources in a museum or retail store, but less important for street lighting. Chromaticity stability of the lamp and luminaires is important where multiple lamps or luminaires are being used to wash a wall, or where objects are being evaluated based on color, such as in a hospital or factory. The chromaticity maintenance of LED lamps and luminaires varies among different products, and potentially for the same product used in different applications. Like many other metrics, there are no official standards limiting the amount of acceptable chromaticity shift.

Many types of light sources have some chromaticity instability over time. The most pronounced is metal halide, but fluorescent and halogen lamps can also shift. The problem with this chromaticity instability is illustrated in Figure 2, where the lamps are creating a different visual appearance for the wall wash. A DOE study on the chromaticity maintenance of LED PAR38 lamps was performed to understand the current performance of LED products on the market. As part of this broader study on the chromaticity

Figure 2. A room lit with ceramic metal halide lamps shows the impact of poor chromaticity stability on the appearance of the wall wash. The varying color appearance of neighboring lamps would necessitate relamping.
8 shift in LED lamps, the chromaticity shift of various conventional lamp technologies has been measured.4 The average change in chromaticity for PAR38 lamp samples with different lighting source technology was measured and the results are shown in Figure 3. A high-level analysis shows that on average, the
LED PAR38 lamps had better chromaticity stability than any of the comparable conventional lighting benchmarks. Of the traditional lighting technologies, ceramic metal halide had the poorest chromaticity stability compared to halogen and compact fluorescent lamps, whereas the latter two are generally considered to have acceptably small levels of color stability.

Chromaticity stability can vary based on LED lamp or luminaire product design with several factors affecting the resulting performance. Ambient air temperature, drive current, and the design of the lamp or luminaire’s thermal management system can influence the junction temperature of the LED, which in
turn, can affect its output characteristics. Of greater concern for long-term chromaticity stability is the effect that high operating temperatures can have on certain package and optical materials. Depending on the design of the LED package, the phosphor layers may settle, curl, delaminate, or otherwise change
the number of photons that are converted to white. This behavior can occur even in the absence of high ambient temperatures. Likewise, other materials in the optical path, such as silicones or plastics may discolor over time. In addition, materials such as glues or chemicals may diffuse into the LED package and affect chromaticity stability. Temperature fluctuations during operation may also intensify degradation mechanisms for some LED products.

There are no official standards limiting the amount of acceptable chromaticity shift in LED lighting products, but different certifications have established requirements. For example, to qualify for the ENERGY STAR® label, nine out of 10 samples of an LED lamp must have a measured chromaticity shift
(Δu’v’) of less than 0.007 over the first 6,000 hours of operation. For applications that require high chromaticity stability, a specification may be established on a project-by-project basis.
Beyond the lack of agreement on acceptable levels of chromaticity shift, there is currently no standard methodology for projecting future chromaticity maintenance using standard test procedures like there is for projecting LED package lumen maintenance. Furthermore, there are no established methods for accelerated testing, leaving each manufacturer to develop their own testing methodologies and predictive modeling approaches. A consensus methodology for predicting chromaticity shift will be a challenge as different materials of construction and manufacturing processes can affect the results; however, an IES committee is working to come to accord on this pressing issue (TM-31).

Chromaticity stability should not be confused with chromaticity consistency — also referred to as color consistency. Chromaticity stability refers to the ability of a product to maintain a constant chromaticity point over its lifetime, whereas chromaticity consistency refers to the product-to-product variation within a lamp or luminaire type. This lamp-to-lamp consistency is important to provide uniform lighting within a room and building. In LED lighting, the chromaticity consistency from lamp to lamp depends on the consistency of the phosphor-converted LEDs. To counter variability that is inherent in the manufacturing process, white LEDs are binned based on chromaticity, lumen output, and forward voltage. This allows the manufacturers of LED lamps and luminaires to provide a more consistent
product.

Figure 4 illustrates the problem of poor chromaticity consistency from lamp to lamp upon installation. Although this case represents a chromaticity consistency challenge, similar effects can be seen with varying chromaticity maintenance over time. Both factors, chromaticity consistency (at time = 0) and chromaticity stability (after thousands of hours in operation), are crucial for the customer

The LED package construction often drives the performance and long term behavior of the LED light source. The impact of LED package design and materials of construction on performance, color quality, lumen maintenance and chromaticity shift, have been investigated for a variety of LED packages under the DOE SSL Core Technology Research Project awarded to RTI International.6 One of the goals of this project is to determine failure modes for LED packages and develop software approaches to model failure rates in an effort to correlate package behavior to system reliability results.
Four main LED package platforms have emerged as light sources for LED luminaires:
 High-power packages (1 to 5 W) typically used in products requiring small optical source size (e.g., directional lamps) or high reliability (e.g., street lights)
 Mid-power packages (0.1 to 0.5 W) typically used in products requiring multiple light sources for diffuse emission (e.g., troffers, A-type lamps)
 Chip-on-board (COB) packages typically used in products needing high luminous fluxes from a small optical source or extremely high luminous flux density (e.g., high-bay lighting)
 Chip scale packages (CSPs), also called package-free LEDs or white chips, which have gained attention as a compact, low cost alternative to the high-power and mid-power platforms.
Representative packages from these major LED package platforms are illustrated in Figure 5

To begin the modeling work of the lumen maintenance and chromaticity shift of LEDs, a methodology was developed to analyze LM-80 data across multiple LED manufacturers to provide new insights into LED-level factors impacting lifetime. Data from more than 200 different LED data sets was analyzed using this methodology combined with TM-21 projections and supplemented with experimental data. This process calculates a decay rate constant () that provides a measure of the rate of luminous flux change. Higher  values indicate faster lumen depreciation whereas small  values indicate longer lumen maintenance times. The analysis provided a detailed look at lumen maintenance and chromaticity shift behavior for a range of LED packages with different designs and materials of construction from multiple manufacturers and found that the materials of construction have a direct impact on long-term performance of LEDs.

Figure 6. Summary of the LM-80 report records by year and LED platform type.
The different LED package platforms have different intrinsic characteristics based on materials of construction and manufacturing processes, which impact their lumen depreciation and chromaticity point stability. Figure 7 shows the decay rate constants as a function of LED junction temperature for different package platforms. Mid-power LEDs can often exhibit more rapid lumen degradation than high-power LEDs or chip on board LEDs; this faster decay of luminous flux is largely due to degradation of the plastic resin body used in the mid-power LED compared to the more stable ceramic substrate used in the high-power LED. The plastic material most commonly employed in mid-power LED packages is polyphthalamide (PPA), a thermoplastic resin. At high temperatures and long operating times, the materials in the package can discolor, crack, or delaminate, leading to lumen depreciation and chromaticity shift.
Different types of plastic resin, however, have different lumen degradation behavior. Improved plastic resins such as epoxy molding compound (EMC) can reduce the thermal constraints associated with conventional mid-power commodity packages. Mid-power LEDs based on EMC resin are more resistant to degradation than PPA and compatible with higher operating temperatures. Figure 8 compares the

Figure 7. Decay rate constants for high-power, mid-power, and COB LEDs as a function of junction temperature. These were calculated using LM-80 and TM-21 projections combined with the new analysis methodology described.

Figure 8. Lumen degradation performance of mid-power packages (PPA and EMC plastic resins) operating at 150 mA and a high-power package (ceramic substrates) operating at 1 A drive current.
lumen degradation performance of mid-power packages using PPA and EMC plastic resins to high power packages using ceramic substrates. While the quality of mid-power packages can vary between LED manufacturers, one commonly seen trend is that EMC-based LED packages can achieve high lumen maintenance at higher temperatures and drive currents than PPA-based LED packages. In addition, it is commonly observed that the ceramic substrates in high power LEDs provide improved heat dissipation and thus result in higher lumen maintenance behavior, especially at high currents and temperatures.
Though the various package materials of construction have different lumen maintenance performance,

both PPA and EMC mid-power packages can achieve excellent lumen maintenance performance (e.g., better than 50,000 hours in some cases), as long as their drive currents are kept low enough to stay in a ‘safe operating zone’ (below where that particular resin material discolors and breaks down over time). While lumen maintenance is important, other forms of parametric failure for LED packages must not be overlooked. Chromaticity shift, for example, may be more detrimental than lumen depreciation for some applications; however, this is sometimes difficult to know in advance. To date, the industry generally quantifies chromaticity shift using Δu’v’, which describes the magnitude of chromaticity shift, but it does not capture the direction of the shift. (The actual chromaticity coordinates u’ and v’ are required to know the direction of the chromaticity shift.) The point at which a chromaticity shift becomes noticeable and results in parametric failure will depend on the lighting application. If the chromaticity change occurs slowly over a very long period (e.g., 25,000 hours), it may not be objectionable in the case where the light sources shift by the same magnitude and in the same direction (unlikely in practice).
Factors impacting chromaticity point stability in LEDs include aging-induced changes in the emitter, phosphor, encapsulant materials, and plastic resin. Emitters can exhibit decreases in radiant flux over time; phosphors can experience decreases in quantum efficiency or shifts in emission spectrum due to oxidation; encapsulants can exhibit cracking, oxidation and yellowing, or changes in index of refraction; and resins can discolor and absorb photons. Higher temperatures will accelerate these degradation mechanisms leading to greater color shift, though the magnitude of the color shift as a function of temperature will vary with packaging materials and manufacturing processes. As with lumen maintenance behavior, if the LEDs are operated at low drive currents and lower than normal operating temperatures, these materials changes leading to chromaticity shift will be very slow to develop, if they occur at all

Figure 9. 1976 CIE chromaticity diagram (u’, v’) illustration the white chromaticity region (denoted by the black circle) and the common directions of chromaticity shift in LED packages. The right figure is an enlargement of the black circle, showing the white chromaticity bins.