As “efficiency” is often defined for technology, the efficiency of anything that changes energy into a usable form for some purpose is the fraction of the changed energy that actually winds up in the intended usable form. A heat engine, for instance, takes energy from some heat source and turns some fraction of it into mechanical energy. The maximum possible efficiency for a heat engine that operates in cycles between the heat source’s absolute temperature Ts and the lower temperature of its environment Te is always less than 100%, since at least the fraction Te/Ts of the heat input becomes heat output without doing any mechanical work; thus cyclic heat engines never have efficiencies greater than 1 - Te/Ts.[Wikipedia]
Many sources of light also produce a great deal of heat. If the light source is also meant to be a heat source, its efficiency is high since both the heat and the light are wanted. But if the point is to simply provide light, while heat is not wanted or even counterproductive, a hot light source can be quite inefficient; much of the energy put into the source does not produce light and is thus wasted. Fires and incandescent light bulbs both put out a lot more heat than light. Where only light is wanted, an ideal light source would turn all of its energy input into visible light and none into heat, noise, or anything else.
LEDs, or light-emitting diodes,[Wikipedia] are solid-state electronic circuit elements that (a) being diodes,[Wikipedia] normally allow much higher currents to pass in one direction than in the opposite direction, and (b) emit light when current passes through them. While LEDs are not 100%-efficient light sources, they can turn much more of their input energy into light than incandescent bulbs do, and in principle could even surpass the efficiency of fluorescent bulbs.[Wikipedia; Note] Much recent work sponsored by the Department of Energy has determined the usefulness of existing light-emitting diodes for lighting public spaces, while other research has led to various improvements in LEDs and associated devices as well as assessing the effects of increased LED use on our environment.
The newer any technology is, the less experience anyone has with it, and the more uncertain the results of adopting it are. The problem of gaining experience is exacerbated if the earliest results don’t become widely known, which dooms those who don’t know the history of those results to repeat it. And it’s not just bad experiences and mistakes of early adopters that get repeated unnecessarily. Others who might have been able to gain the new technology’s advantages will wait longer to do so, to avoid risking disadvantages larger than they can afford to deal with, until the size and likelihood of those disadvantages becomes clearer as the early adopters’ experiences become widely known. The sooner good and bad experiences become widely known, the faster they can be assessed, the problems solved, and the benefits widely realized.
To circulate early experiences with the use of light-emitting diodes in buildings and public spaces, the Energy Department’s GATEWAY Program reports on the plans and results of several demonstration projects.[DoE] Pacific Northwest National Laboratory manages the program and conducts evaluations on the Energy Department’s behalf.
As noted in one October 2012 report on actual experiences with LEDs activated by motion detectors at two parking structures and two parking lots, the energy savings possible from the motion detectors can be lost if the detectors aren’t used correctly. The report, “Use of Occupancy Sensors in LED Parking Lot and Garage Applications: Early Experiences”,[SciTech Connect] shows a marked range of outcomes—from energy savings of 76% just from the motion detectors, after the initial conversion to LED lighting, to virtually no additional savings. The authors find several issues that influenced savings in these early-stage installations:
As the authors further state,
“The experiences and observations described in this report are intended to bring these and related issues into focus for those considering the use of occupancy sensor based control systems in their own applications. Ultimately, care must be taken in the design, selection, and commissioning/optimization of a sensor-controlled lighting installation, or else the only guaranteed result may be its cost.”
Experience with one parking structure is described in the report “Demonstration Assessment of Light-Emitting Diode Parking Structure Lighting at U.S. Department of Labor Headquarters”.[SciTech Connect] In this project, the old high-pressure sodium luminaires were replaced with light-emitting diode luminaires and the new luminaires evaluated. This conversion alone showed an energy savings of 52%. These savings were further increased, to 88%, by using motion detectors to reduce power to 10% of high-state operation after a 2.5-minute time delay. The motion detectors’ default factory setting was for a 10-minute delay, but this was judged to be longer than necessary, since people need the brightest lighting in a parking structure at most while they enter the structure and park, plus a short time thereafter while they gather their things and leave their vehicle. The delay was found to significantly influence the lighting system’s energy use, as shown by 85 days’ worth of data with the factory-set delay and 42 more days’ worth of data with the reduced delay. The time required to pay back the initial cost of the LED luminaires (which was relatively high when they were bought in 2010) with energy savings was found to be 6.5 years and 4.9 years for retrofit and new-construction scenarios, respectively. The report also notes the quality of the new lighting: the chief users, Labor Department Headquaters staff, reported high satisfaction with its operation.
Other reports describe recent assessments of other demonstration projects: post-top lighting in New York City’s Central Park, street lighting in Kansas City, Missouri, and roadway lighting in Philadelphia. Unlike some demonstrations that compare one solid-state lighting product against what it replaces, these demonstrations all compared existing technology with multiple candidate replacements.
The Central Park demonstration[SciTech Connect] compared the existing lighting for paved walking trails and adjacent grass areas with five new products. The evaluation was further complicated by the fact that, unlike the usually even spacing of lights for relatively flat parking-lot surfaces, the Central Park pathway lights are unevenly spaced on one side of paths that meander around various landscaping features. In addition to the kind of horizontal measurements of the light’s adequacy for path navigation, vertical measurement of the illumination needed to identify approaching persons’ faces are also important. A system using both kinds of measurements to compare luminaire performance consistently across locations was devised by the authors, since current guidance from the Illuminating Engineering Society on appropriate measurement procedures for walkway evaluations was limited.
New York City’s Department of Transportation selected a total of five different light-emitting diode products representing a variety of energy use and lumen packages for evaluation against the metal halide baseline luminaire. Four of the products were complete new luminaires installed on top of the poles and the other was a retrofit insert kit installed in an existing housing unit. As tabulated in the report, energy savings of the different LED systems ranged from 50% to 83% relative to the incumbent metal halide luminaire. The life-cycle costs included in the same table are based on an 18.3-year analysis period, or 75,000 operating hours, which corresponds to the longest claimed lifetime among the products evaluated. Four of the LED products offer lower life-cycle costs than the incumbent metal halide luminaire—ranging from about $2,258 to $4,688.
Not all the energy and cost savings necessarily represent comparable or “suitable” replacement conditions; in fact, the quality of the different products was mixed, as some limited color quality data shows, though park users weren’t surveyed about the new fixtures’ subjective acceptability. Of the three LED fixtures that were cost effective and saved energy, two had correlated color temperature (CCT)[Wikipedia; Wikipedia] notably higher than the traditional metal-halide luminaire, which could prove objectionable to some users, while the other fixture’s CCT was roughly comparable to the traditional metal halide luminaire. The experience led to New York City’s Department of Transportation replacing Central Park’s lighting with an updated version of the third fixture.
The report on a study of replacements for street lights in Kansas City, Missouri[SciTech Connect] describes the results of a study in which nine different light-emitting diode products were installed and examined as replacements for high-pressure sodium street lights. Of possibly greater interest, though, are several points made in the report that illustrate how any evaluation of different lighting systems can be complicated, even if only one or a few products are considered:
For Philadelphia’s assessment, described in the report “Demonstration Assessment of LED Roadway Lighting: Philadelphia, PA”,[SciTech Connect] 10 different groups of LED luminaires were installed at three sites. Each site represented a different set of conditions, most importantly with regard to the incumbent high-pressure sodium luminaires, which were nominally 100 W, 150 W, and 250 W. Each product’s performance was evaluated based on manufacturer data, illuminance calculations, field measurements of illuminance, and the subjective impressions of both regular and expert observers.
Conclusions from the subjective impressions seem relatively simple. “In general, there were only small deviations in the perception of the luminaires, and it would be difficult to select any products that were clearly superior to the others in the same category. Nonetheless, both [lighting professionals and local residents/passersby] generally viewed the demonstration LED lighting favorably, especially compared to the typical [high-pressure sodium] luminaires used in Philadelphia.”
Large pole-spacing differences at each site limited the usefulness of field-illuminance measurements for comparing performance, but did reveal that differences in measured and calculated illuminance were up to approximately 40%, and that the high-pressure sodium luminaires were more likely to provide lower illuminance than predicted by calculations. From supplemental calculations using a single representative model of each of the three sites, two of the three sodium luminaires and seven of the ten LED luminaires were predicted not to meet at least one of the maintained illuminance design criteria. The report notes that the LED systems were not designed to meet illuminance criteria, but to perform similarly to the sodium luminaires.
An illustration of how different cities’ situations are similar in some ways and significantly different in others is given in the report’s discussion of cost savings:
“Despite lower input power for the LED luminaires, energy cost savings are not currently possible for Philadelphia because they have yet to reach an agreement for a new tariff for LED street lighting with the local utility, a situation that is common throughout the country. Even with a new tariff, energy savings alone are unlikely to result in a reasonable payback period given current LED efficacy levels. However, additional cost savings from reduced maintenance could make widespread luminaire replacement cost effective. In particular, the LED luminaires installed at one of the three sites—where the luminaires were mounted on an elevated rail track—were less susceptible [to] premature failure from strong vibrations.”
Preparation for a roadway demonstration project, rather than results of one, is described in the report “Technical Feasibility Assessment of LED Roadway Lighting on the Golden Gate Bridge”.[SciTech Connect] This preparatory study evaluated the feasibility of replacing the Golden Gate Bridge’s existing high-pressure and low-pressure sodium lamps with lamps using light-emitting diodes, electrodeless fluorescent light sources, and ceramic metal halide. A key assumption was that existing light levels are to be maintained. If reduced illumination were to be found acceptable, lower-power products could be used, greater energy savings would result, and light-emitting diode technology would be more feasible; but since reduced illumination limits constitute a different design criterion, a new study would be required.
For the Golden Gate Bridge, electrodeless fluorescent light sources were found to save energy only by reducing light levels. Ceramic metal halide luminaires were found that could offer 28% energy savings, but also were rated to last less time than current luminaires, so their energy savings could be more than offset by increased maintenance costs. While no suitable retrofit kits using light-emitting diodes were found, several complete light-emitting diode luminaires were found to offer 6%-18% energy savings, suggesting that custom retrofit kits could be developed to equal or outperform the existing bridge luminaires.
Alternatives to the current sodium luminaires’ yellow light would be whiter. The report authors recommend using relatively inexpensive ceramic metal halide mockups to see if whiter illumination would be appropriate for the Golden Gate Bridge. If it is, design criteria for light-emitting diode retrofit kits would need to be developed with the following issues in mind:
Unlike other sites, the Golden Gate Bridge seems that it would not yet benefit from the replacement of its existing light sources with light-emitting diodes, at least according to the design criteria already examined. The authors note that other options might be preferable, such as reformulating the current luminaires’ amber lenses, or replacing the luminaires entirely instead of their internal components. Using different lenses could keep white light sources from looking green, thus allowing use of off-the-shelf light-emitting diode or ceramic metal halide products, while complete luminaire replacement would, among other benefits, allow reputable manufacturers to offer standard warranties.
While maintaining a constant color of light from sets of light-emitting diodes is important enough on pathways and roadways, for quite different reasons it’s particularly important in museums. The report “Color Maintenance of LEDs in Laboratory and Field Applications”[SciTech Connect] describes how many LED lamps in museums have changed color beyond a reasonable tolerance well before reaching their rated lifetimes—in some cases in as few as a few thousand hours—while other LED lamps “can have exemplary color stability that is unmatched by traditional light sources.” The report offers useful color-maintenance information, such as the metrics used for communicating color shift, how to monitor chromaticity, what to look for in manufacturer warranties, and physical changes that have been shown to lead to color shift in some light-emitting diode packages.
While the aforementioned reports focus on how existing LEDs prove to work out in actual use, recent patents and other technical reports show continuing improvements in LED technology.
The patent “High efficiency III-nitride light-emitting diodes”,[DOepatents] assigned to Sandia Corporation,[Wikipedia] addresses the droop in existing LED’s luminous efficacy when high injection currents make them shine brighter. The problem is particularly significant for blue LEDs designed to form white light for general lighting by either mixing their light with that of red and green LEDs or by stimulating phosphors of other colors. The patent describes a way to mitigate the efficiency droop by tailoring a diode’s doping[Wikipedia] so that electrons don’t leak from the quantum wells[Wikipedia; Wikipedia] that comprise the diode’s active region.
With regard to the use of phosphors with blue LEDs, the Lightscape Materials, Inc. patent “Oxycarbidonitride based phosphors and LED lighting devices”[DOepatents] notes that “color tuning of the blue LED chip and the included phosphor(s) is critical for the effectiveness and optimization of [phosphor-converted LED, or] pcLED devices. Accordingly, there is a continuing need for phosphor development to provide pcLED device manufactures with enhanced color tuning capabilities.” The patent further notes that since conventional phosphor-converted LEDs are close to the blue LED, the phosphors reach temperatures in the range of 100-150° C and can emit heat instead of light, rendering them inadequate as light sources.
Nitride-based phosphors perform well at such high temperatures, and have thus been proposed for use in phosphor-converted LEDs. The host crystals of metal silicon nitride based phosphors consist of stable silicon-nitrogen, aluminum-nitrogen, and hybrid bonds. Silicon-carbon bonds are even more stable, and many metal-carbon bonds are also stable, but some metal carbides can absorb or quench emitted light, and residual or unreacted carbon in a phosphor preparation can reduce the phosphor’s emission intensity. However, the carbon in certain carbidonitride phosphors can enhance the phosphor’s luminescence instead of quenching it. “Oxycarbidonitride based phosphors and LED lighting devices” describes a particular class of red phosphors and white-light sources that incorporate them.
The General Electric patent “Phosphors for LED lamps”[DOepatents] addresses problems with two phosphor-converted LED-based methods of producing light that has a high color rendering index[Wikipedia]. The patent notes that the deep red phosphors used in one in one actual method may reabsorb light of other colors emitted from other phosphors, and describes orange/red phosphors intended to overcome this problem. Another potential method would involve enhancing the emission of blue-green light by using an appropriate phosphor; the patent describes blue/green and green/yellow phosphors for this purpose.
A slide presentation from Los Alamos National Laboratory entitled “Giant Nanocrystal Quantum Dots as Stable and Efficient Down-Conversion Phosphor for LED based Solid State Lighting”[SciTech Connect] contrasts the advantages and disadvantages of phosphors that include rare-earth dopants, like the ones described in the patent “Phosphors for LED lamps”, with those of phosphors that consist of nanocrystal quantum dots. The presentation then discusses how a nanocrystal quantum dot, encapsulated in a thick inorganic shell to form a “giant nanocrystal quantum dot”, overcomes ordinary nanocrystal quantum dots’ problems of photostability, sensitivity to surface-related effects, losses of emitted light by self-reabsorption.
“LEDs can now generate white light nearly as efficiently as a compact fluorescent lamp, and efficiencies are expected to increase. To fully realize the energy savings of the LEDs, the electronics that drive them must also be efficient.” So state two patents assigned to Koninklijke Philips Electronics: “LED lamp color control system and method”[DOepatents] and “LED lamp power management system and method”[DOepatents].
The first patent addresses color control in an LED device that uses diodes that emit different colors directly instead of using single-color diodes augmented with phosphors. By changing the currents supplied to diodes of different colors, the color of light output by the device as a whole can be changed. But if the currents that produce a given color are controlled by feedback from the device’s output, the color control is limited by several factors. For one example, if a certain color is produced by having one or more of the LEDs flicker on and off in repeated pulses of a certain fraction of a second each, and if it takes longer than that fraction of a second for the color control system’s light-measurement devices to stabilize, the control system won’t be able to accurately determine that color, or use the feedback to keep the lamp’s output steady at that color. “If no measures are taken to release control when approaching problem regions,” the patent says, “the LED lamp is likely to exhibit unpredictable behavior and instability in color and intensity.”
The invention described avoids this problem by not having its feedback system rely on measurements it can’t accurately make. When it can make those measurements, the system will go ahead and store the measurement data for comparison with the intended output, and adjust the lamp’s input currents accordingly. But whenever any of the diodes operate outside their feedback-controllable range, the system quits trying to measure the lamp’s output and falls back on the last valid measurement instead. This is expected to work well since “[o]ptical flux feedback is used primarily to correct for LED source performance degradation over extended times and degradation is most likely for LED sources driven at full output, so temporarily using a stored value for measured optical flux for LED sources driven at low output has a minimal effect on LED lamp performance.”
The second patent addresses efficient power consumption when a multicolor LED lamp has its color setting changed and consequently ends up consuming different amounts of power than before in the parts of the circuitry that govern the different-color LEDs. Here, the problem addressed is not control of the color, but optimizing power use for any chosen color. The power consumed for each color may increase whether the light output for that color increases or decreases, since more power is consumed by both the brightened LEDs and the shunt switches that divert power from the LEDs that are being dimmed. The patent describes control circuitry that reduces power to the parts of the circuit whose power consumption exceeds given thresholds and then recalculates the fraction of light to be produced by each color of LED, as well as the LEDs’ duty cycles that correspond to the reduced power level.
Philips LED Systems’ development of standalone LED-driving electronics, managed with the National Energy Technology Laboratory, is reported further in “High Efficiency Driving Electronics For General Illumination LED Luminaires”.[SciTech Connect] The report describes in detail two arrangements of circuit elements, one suitable for LED drivers with powers up to 50 watts and the other suitable for 40-watt to 300-watt drivers, and three new drivers for rugged outdoor lighting applications, with powers of 40 watts, 75 watts, and 150 watts. The new driver platforms have greater efficiencies of more than 90%, smaller sizes of 2.5 cm3/watt, lower costs of 12¢/watt, and have become the bases of new, successfully introduced products.
While the inventions just described relate to the production of light or to the electronics that drive light production, other inventions involve the directing of light from LEDs so it can be seen.
“Organic light emitting diodes with structured electrodes”,[DOepatents] a University of California patent, relates to diodes made of organic molecules. Organic light-emitting diodes require relatively little power for the light they produce and offer many advantages over liquid crystals for the manufacture of information displays like television screens and computer monitors. However, there are still several obstacles to making organic LED displays commercially feasible. One obstacle is that typical electrodes for organic LEDs, which are made of materials that easily transfer electrons between themselves and the diodes, trap much of the light that the diodes emit so that nobody sees it. The invention described in the patent avoids this problem by using nanostructures for the cathodes. Visible light waves are hundreds of nanometers long; even the slightly shorter waves of near-ultraviolet light are a few hundred nanometers long. This is much longer than the few nanometers width of the cathode components. Any kind of wave that is so much longer than the width of an object in its path will scarcely be disturbed by the object, let alone blocked by it. Light waves that would be blocked by a larger cathode would diffract[Wikipedia] right around a structure of nanometer-sized elements. An additional benefit of nanostructure electrodes is that their large surface-area-to volume ratio reduces heating effects where the electrode touches the light-emitting diode, thus increasing the light source’s reliability.
A diode’s light can be trapped in other ways besides direct obstruction. Light generally travels at different speeds within different materials. Suppose a light wave traveling through a diode reaches its surface, and the speed of such waves through material surrounding the diode is higher. If the wave is traveling perpendicular to the interface, part of the wave will be reflected while the rest will cross over into the substrate and keep going, but faster. If the wave is not quite perpendicular to the interface, the component of the transmitted wave’s velocity that’s parallel to the interface will stay the same as it crosses over, while the velocity component perpendicular to the interface will be increased so that the wave’s speed overall is raised by the same amount as in the first case. With the lower perpendicular velocity, the transmitted portion of the wave will make a shallower angle with the interface than the incident wave, thus forming a refracted wave. The transmitted portion will also be a smaller fraction of the incident wave, with more of the wave being reflected back into the diode.
If the approaching wave’s angle with the interface is shallower still, its velocity component parallel to the interface will be even larger while its perpendicular component is even further reduced, the overall speed reduction again being the same, resulting in an even shallower angle of refraction. But for a particular critical angle of approach, the velocity component parallel to the interface will exactly match the speed of the wave outside the diode, while the perpendicular component is zero. In this case, none of the wave gets past the interface, but runs along it instead. For angles of approach even shallower than the critical angle, the entire wave is reflected back into the diode.[Wikipedia]
Since light produced by the diode is propagated in all directions, some waves will reach its surface at every possible angle, meaning that much of the light will hit the interface at angles too shallow to escape from the diode. How shallow is too shallow depends on the ratio of light speeds in the diode and its surroundings. The higher this ratio, the less a light wave can deviate from a perpendicular angle and still have any fraction of it escape from the diode.
If a light-emitting diode’s surface is flat, only a certain fraction of the light waves it produces will strike the surface within the critical angle and thus be actually emitted. But the rougher the diode’s surface is, the more of the light waves within it will travel within critical angles of some portion of the surface, and the more light will escape. This use of rough diode surfaces, along with ways to form them, is the concept behind the 2013 Cree, Inc. patent “Light emitting diode with high aspect ratio submicron roughness for light extraction and methods of forming”.[DOepatents] “Submicron roughness” means that no portion of the diode’s emitting surface is wider than one micron (one millionth of a meter).
A light-emitting diode shines in many directions, but what one often wants is light directed to cover a certain large or small area—not in a way that would form an image of the light source, as a mirror would, but diffusively. The surface of a perfect mirror would reflect every ray that reached it from one direction into some other unique direction. On the other hand, a perfectly diffuse reflector’s surface would reflect some fraction of every ray into every possible direction outside the reflector. A research agreement between WhiteOptics LLC and the National Energy Technology Laboratory describes work that demonstrated a reflective material that significantly improves the efficiency of light-emitting diode luminaires. “Recovery Act: Low-Cost, Highly Lambertian Reflector Composite For Improved LED Fixture Efficiency and Lifetime”[SciTech Connect] describes the work and its result: a composite coating material for light-emitting diode luminaires that costs less than 80¢/ft2, diffusively reflects 97% of the light that strikes it, and loses no more than a few percent of its reflectivity after hundreds of hours of exposure to ultraviolet light, freeze/thaw cycles, and water immersion—conditions simulating expected operating extremes.
Sometimes we want light directed to a very specific spot. This can be accomplished with many kinds of light source, but the ones normally used all have drawbacks, as the Philips Electronics patent “Integrated LED-based luminare for general lighting”[DOepatents] points out. Fluorescent lights are often cheap and efficient, but are too diffuse for spot lighting, and the light they emit doesn’t have an appropriate color mix for some applications. Halogen lamps overcome fluorescent lights’ deficiencies and are also cheap, but have low output efficiencies; incandescent lights have similar advantages and disadvantages, also tending to be too inefficient for spot lighting. And ceramic metal halide lamps can be nondiffuse and energy efficient, but are typically expensive upfront and can be so bright (and non-dimmable) that they make areas next to the illuminated spots look dark by comparison. The patent describes designs for LED-based luminaires that can serve as replacements for other light sources, even having the same key parameters so they can fit the same sockets and be used in existing lighting systems. In particular, some implementations of the LED luminaires are designed to overcome the aforementioned deficiencies of non-LED spot lights.
The user-experience reports from Pacific Northwest National Laboratory described at the beginning of this article each focused on the cost and benefits of light-emitting diodes in particular locations. A more general question is what happens when the luminaires are burned out or are otherwise in need of replacement and disposal. This is dealt with in the recent report “Life-Cycle Assessment of Energy and Environmental Impacts of LED Lighting Products, Part 3: LED Environmental Testing”.[SciTech Connect] Parts 1[EERE] and 2[SciTech Connect] of the assessment dealt with lamps’ energy consumption (LED, incandescent, and compact fluorescent alike) and with LED manufacturing and performance respectively. The conclusion of both reports is that the lamps’ operation energy dominates their environmental impact. On the other hand, since more and more light-emitting diode lamps are now being installed, understanding the implications of their disposal has become more important.
Standard chemical analyses of 22 samples representing 11 different models of light-emitting diode, compact fluorescent, and incandescent lamps were made to see whether any of 17 elements were present in excess of California or Federal regulatory thresholds for hazardous waste. (The authors note that this type of testing doesn’t provide an indication of product safety during use.) Among the findings:
The report notes that light-emitting diode lamps have lesser energy and environmental impacts than incandescent and compact fluorescent lamps, but that recycling will likely become more important as more light-emitting diode lamps come into use.
[Note] ^ Another advantage of light-emitting diodes over incandescent bulbs is that they take a fraction of a second less time to reach their full brightness when turned on—not a crucial difference for most uses, but quite significant for their use in automobile brake lights. The faster a brake light comes on, the sooner the driver in the car behind can see and react to it. Cars traveling 55 miles per hour (~ 88.5 km/h) cover just over 8 feet (or almost 2.5 meters) every tenth of a second, so every tenth of a second sooner that someone driving at that speed can see a brake light ahead of him can put another 8 feet of stopping distance between him and the driver in front. See also “Light-emitting diode: Applications: Indicators and signs” in Wikipedia.
Prepared by Dr. William N. Watson, Physicist
DoE Office of Scientific and Technical Information
Last updated on Friday 13 December 2013