SANDIA REPORT
`SAND2000-1612
`Unlimited Release
`Printed July 2000
`
`THE CASE FOR A NATIONAL
`RESEARCH PROGRAM ON
`SEMICONDUCTOR LIGHTING
`
`Roland Haitz, Fred Kish, Jeff Tsao, Jeff Nelson
`
`Prepared by
`Sandia National Laboratories
`Albuquerque, New Mexico 87185 and Livermore, California 94550
`
`Sandia is a multiprogram laboratory operated by Sandia Corporation,
`a Lockheed Martin Company, for the United States Department of Energy’s
`National Nuclear Security Administration under Contract DE-AC04-94AL85000.
`
`Approved for public release; further dissemination unlimited.
`
`Page 1 of 24
`
`SAMSUNG EXHIBIT 1035
`
`

`

`
`
`
`
`
`(865) 576-8401
`Telephone:
`(865) 576-5728
`Facsimile:
`reports@adonis.osti.gov
`E-Mail:
`Online ordering: http://www.osti.gov/bridge
`
`Issued by Sandia National Laboratories, operated for the United States Department of
`Energy by Sandia Corporation.
`
`NOTICE: 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, nor any of their contractors, subcontractors, or their
`employees, make any warranty, express or implied, or assume any legal liability or
`responsibility for the accuracy, completeness, or usefulness of any information,
`apparatus, product, or process disclosed, or represent 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, any agency thereof, or any of their contractors or subcontractors. The
`views and opinions expressed herein do not necessarily state or reflect those of the
`United States Government, any agency thereof, or any of their contractors.
`
`Printed in the United States of America. This report has been reproduced directly from
`the best available copy.
`
`Available to DOE and DOE contractors from
`
`U.S. Department of Energy
`
`Office of Scientific and Technical Information
`
`P.O. Box 62
`
`Oak Ridge, TN 37831
`
`
`
`
`
`
`Available to the public from
`
`U.S. Department of Commerce
`
`National Technical Information Service
`
`5285 Port Royal Rd.
`
`Springfield, VA 22161
`
`
`
`
`
`0#online
`
`
`
`Telephone:
`Facsimile:
`E-Mail:
`Online order:
`
`(800) 553-6847
`(703) 605-6900
`orders@ntis.fedworld.gov
`http://www ntis.gov/help/ordermethods.asp?loc=7-4-
`
`
`
`
`
`
`
`2
`
`Page 2 of 24
`
`

`

`TT HH EE CC AA SS EE FF OO RR AA NNAA TT II OO NNAA LL RR EE SS EE AA RRCC HH PP RROO GG RR AA MM OO NN
`SS EE MM IIC OO NN DD UUC TTOR LLI GGHT IING 11,, 22
`
`Roland Haitz and Fred Kish, Hewlett-Packard Company, Palo Alto, CA 94304
`Jeff Tsao and Jeff Nelson, Sandia National Laboratories, Albuquerque, NM 87185-0601
`
`EXECUTIVE SUMMARY
`
`Dramatic changes are unfolding in lighting technology. Semiconductor light emitting diodes
`(LEDs), until recently used mainly as simple indicator lamps in electronics and toys, have become as
`bright and efficient as incandescent bulbs, at nearly all visible wavelengths. They have already begun
`to displace incandescent bulbs in many applications, particularly those requiring durability,
`compactness, cool operation and/or directionality
`(e.g.,
`traffic, automotive, display, and
`architectural/directed-area lighting).
`Further major improvements in this technology are believed achievable. Recently, external
`electrical-to-optical energy conversion efficiencies exceeding 50% have been achieved in infrared light
`emitting devices. If similar efficiencies are achieved in the visible, the result would be the holy grail
`of lighting: a 200lm/W white light source two times more efficient than fluorescent lamps, and ten
`times more efficient than incandescent lamps.
`This new white light source would change the way we live, and the way we consume energy. The worldwide
`amount of electricity consumed by lighting would decrease by more than 50%, and total worldwide
`consumption of electricity would decrease by more than 10%. The global savings would be more
`than 1,000TWh/yr of electricity at a value of about US$100B/year, along with the approximately 200
`million tons of carbon emissions created during the generation of that electricity. Moreover, more
`than 125GW of electricity generating capacity would be freed for other uses or would not need to be
`created, a savings of over US$50B of construction cost.
`Bringing about such revolutionary improvements in performance will require a concerted
`national effort, of the order $0.5B over ten years, tackling a broad set of issues in semiconductor
`lighting technology. The effort would also require harnessing the most advanced high-technology
`companies, the best national laboratory resources, and the most creative university researchers in this
`area.
`
`Fire Candles and Lamps Bulbs and Tubes Semiconductors
`r
`e
` Lam
`bs a d
`e m c nd tor
`
`
`
`
`
`
`1 This white paper was first presented publicly at the 1999 Optoelectronics Industry Development Association (OIDA)
`forum in Washington DC on October 6, 1999.
`2 Revision B:03/30/1999
`
`3
`
`Page 3 of 24
`
`

`

`1 INTRODUCTION
`
`Energy is the lifeblood of our economy, and a critical building block for global peace and
`security. Its generation incurs huge costs: both direct economic costs as well as indirect
`environmental costs (smog and particulate emissions, acid rain, global warming, waste disposal, etc).
`And, the direct economic costs will only increase as concern heightens over how to reduce the
`indirect environmental costs.3 As a consequence, there is great benefit to enhancing the efficiency
`with which energy is used -- virtually all major energy consumers from transportation to heating to
`the various users of electricity are constantly being examined for energy saving opportunities.
`
`
`1000
`
`100
`
`10
`
`1
`
`Energy Consumption (Quads)
`
`400 Quads
`
`Energy
`
`130 Quads
`
`25 Quads
`
`Electricity
`Illumination
`(assuming 20% of electricity)
`
`WORLD
`
`1970
`
`1980
`
`Projected
`
`1998
`
`2000
`1990
`Year
`
`2010 2020
`
`1970 1980
`
`2000
`1990
`Year
`
`2010 2020
`
`Energy
`
`94 Quads
`
`34 Quads
`
`Electricity
`Illumination
` 60% commercial
` 26% residential
` 14% industrial
`1998
`
` 6.9 Quads
`
`Projected
`
`U.S.
`
`Figure 1. World (left) and
`U.S. (right) consumption of
`energy for use in all forms
`(blue), for use in electricity
`generation (pink), and for use
`in illumination (green).4 One
`Quad (one quadrillion
`BTUs) of primary energy
`consumed is roughly
`equivalent, after energy
`conversion and transmission
`losses, to 92TWh of
`electricity at the wall plug.
`
`Among the most widespread, important, and growing uses of energy is the electricity used for
`lighting. As illustrated in Figure 1, in the U.S., about 20% of all electricity consumed,5 and about
`7.2% of all energy consumed, can be estimated to be used for lighting. In 1998, the cost was about
`6.9 quads of primary fuel energy (with an associated 112 million tons of carbon emissions), and about
`637TWh of actual electricity consumed at a cost of about US$63B. Worldwide, about 3.4% of all
`energy consumed can be estimated to be used for lighting, a percentage that is expected to increase
`with standard of living. In 1998, the worldwide cost was about 25 quads of primary fuel energy (with
`an associated 410 million tons of carbon emissions), and about 2,350TWh of actual electricity
`consumed at a cost of about US$230B.
`Because of this large contribution of lighting to worldwide energy consumption, it is no wonder
`that the lighting industry receives its fair share of inquiries regarding energy reduction. In 1995, the
`three major US lighting manufacturers – GE Lighting, Osram/Sylvania and North American Philips
`
`
`3 In the Kyoto Protocol of 1997, e.g., the developed nations agreed to limit their greenhouse gas emissions, relative to the
`levels emitted in 1990. The United States agreed to reduce emissions from 1990 levels by 7% during the period 2008 to
`2012.
`4 World data taken from the International Energy Agency (http://www.iea.org), and assuming projected energy, electricity
`and illumination growth rates of 1.6%, 3.5% and 3.5%. U.S. data taken from the Energy Information Administration
`(http://www.eia.doe.gov), and assuming projected energy, electricity and illumination growth rates of 1.2%. We
`acknowledge Gerald Hendrickson and Arnold Baker at Sandia National Laboratories for assistance interpreting the data.
`5 According to a recent EPRI report (TR-106196), the four top electricity-consuming applications in the U.S. in 1995 were:
`electric motors (24%), cooling/refrigeration (18%), lighting (17%), and space/water heating (16%). These percentages
`include the three major market segments -- residential, commercial and industrial -- but not street lights, traffic signals, nor
`the use of electricity to remove the heat generated by lighting in air-conditioned buildings. The Industrial Lighting
`handbook estimates that it takes 1 kW of electricity in the air-conditioning system to remove 3 kW of heat generated by
`lighting. After including the above omissions, it is safe to say that, in the U.S., lighting consumes at least 20% of electricity
`and ranks a close second to the 24% consumed by electric motors.
`
`4
`
`Page 4 of 24
`
`

`

`– sponsored a three-day workshop to identify promising research areas for improving the efficiency
`of white light sources. This workshop confirmed that "lighting consumes about 20% of the electric
`power production of the nation." One of the most revealing figures in the resulting EPRI report6 is a
`graph of luminous efficiency vs. time for the major "true" white light sources: incandescent, halogen,
`and fluorescent lamps. As illustrated in Figure 2, none of these workhorse technologies has shown
`any significant efficiency improvements during the preceding 20 years!
`200
`Figure 2. Condensed history and projection
`of efficiencies (in lm/W) of vacuum tube
`(incandescent, halogen and fluorescent) and
`semiconductor (LED) white lighting
`technologies.
`
`150
`
`Projected
`
`With
`accelerated
`effort
`
`Halogen
`Incandescent
`
`There is, however, one striking
`exception. Light emitting diodes
`(LEDs), a 40-year-old semiconductor
`technology, have steadily improved
`their efficiencies and power levels to
`the point where they are knocking
`1990
`incandescent and halogen lamps out
`Year
`of such
`traditional monochrome
`lighting applications sockets as traffic lights and automotive tail lights. And, a recent breakthrough in
`the green and blue makes LEDs a serious contender for conventional white lighting.
`It is the purpose of this white paper to call attention to this new lighting technology and to the potential impact of a
`concerted national effort to advance it further. Such an effort would fill a need identified by the U.S.
`Department of Energy for research in advanced lighting technologies.7 And, such an effort would
`target the technology we believe has the highest potential to create an ideal lighting source, both in
`quality and in cost. LEDs and their semiconductor variants are visually appealing, convenient and
`environmentally friendly, and it is our assessment that they have a realistic shot at reaching the
`industry nirvana of an efficiency of 200lm/W.
`If semiconductor lighting can achieve this goal through a concerted national effort, the lighting
`industry would be revolutionized. An efficiency of 200lm/W would be more than 2x better than that
`of fluorescent lamps (80lm/W), and more than 10x better than that of incandescent lamps (15lm/W).
`If current lighting, with an aggregate efficiency of roughly 50lm/W (in between the efficiencies of
`fluorescent and incandescent lamps), were replaced by semiconductor lighting with an aggregate
`efficiency of 150lm/W (somewhat less than the target), then the electricity currently used for
`illumination would decrease by a factor of three, from 2,350TWh to 780TWh. This would represent a
`decrease in global electricity use of about 13%, and a decrease in global energy use and associated
`carbon emissions of 2.3%.
`In some ways such a revolution in lighting could be compared to the revolution in electronics
`that began 50 years ago and is only now reaching maturity. Just as for electronics, glass bulbs and
`tubes would give way to semiconductors. And, just as for electronics, the increased integrability,
`density, performance, and mass manufacturability of semiconductors may drive an explosion of
`additional, not-yet-thought-of uses for lighting. One can even speculate on visionary concepts in
`
`6 The workshop is summarized in EPRI report TR-106022.
`7 This need has been identified in the Department of Energy's ongoing "Vision 2020" lighting technology roadmapping
`activity. It has also been identified separately by the Department of Energy's Office of Building Technology, State and
`Community Programs, whose program plan consists of three overall goals: (1) Accelerate the introduction of highly
`efficient technologies and practices through research and development; (2) Increase minimum efficiency of
`buildings/equipment through codes, standards and guidelines; and (3) Encourage use of energy efficient technology
`through technology transfer and financial assistance.
`
`Efficiency (lm/W)
`
`100
`
`50
`
`0
`1970
`
`Fluorescent
`
`Semi-
`conductor
`
`Without
`accelerated
`effort
`
`1980
`
`2000
`
`2010
`
`2020
`
`5
`
`Page 5 of 24
`
`

`

`which information and illumination technologies combine to create ultra-fast wireless local-area
`networks that are mediated through building lights!
`We begin this white paper in Section 2 with a brief history of LED technology, and compare its
`current and projected performance and cost with those of conventional technology. In Section 3, we
`discuss its penetration (and replacement of conventional technology) in signaling and lighting
`applications. We expect LED penetration into signaling applications, currently dominated by
`inefficient filtered incandescent lamps, to be rapid, and to drive continued improvements in
`performance and cost. These improvements will, in turn, enable gradual penetration of LEDs into
`lighting applications, currently dominated by a mix of incandescent and fluorescent lamps. Although
`the penetration will be gradual, its global impact will already be very significant, since lighting
`represents such a large fraction of global energy consumption. In Section 4, we describe an
`economic model for that global impact.
`We believe much more dramatic improvements to be possible. In Section 5 we discuss such
`improvements, the resulting acceleration of the penetration of semiconductor lamps into lighting
`applications, and the resulting huge impact on global energy consumption. Finally, in Section 6 we
`discuss in general terms the daunting technical challenges, and the magnitude and nature of a national
`research program that might enable these challenges to be overcome.
`
`2 HISTORY AND PROJECTION OF LED PERFORMANCE AND
`COMPARISON WITH CONVENTIONAL LAMPS
`
`LEDs have had a "colorful" history, alternately pushed by technology advances and pulled by key
`applications. The first demonstration was in 1962 by General Electric. The first products were
`introduced in 1968: indicator lamps by Monsanto and the first truly electronic display (a successor to
`the awkward Nixie tube) by Hewlett-Packard. The initial performance of these products was poor,
`around 1mlm at 20mA, and the only color available was a deep red.8 Steady progress in efficiency
`made LEDs viewable in bright ambient light, even in sunlight, and the color range was extended to
`orange, yellow and yellow/green. Within a few years, LEDs replaced incandescent bulbs for
`indicator lamps, and LED displays killed the Nixie tube.
`Small Signal
`(monochrome)
`
`1 lamp/function
`
`Power Signal
`(monochrome)
`
`Lighting
`(white)
`
` 1-100 lamps/function
`
`10-... lamps/function
`
`Figure 3. Flux and numbers of lamps
`required for various classes of LED
`applications: low-medium-flux "signaling"
`applications, in which lamps are viewed
`directly, and medium-high-flux "lighting"
`applications, in which lamps are used to
`illuminate objects. Current LED lamps
`emit 0.01-10lm of light.
`
`0.001
`
`0.01
`
`0.1
`
`100
`
`1000
`
`10000
`
`Until 1985, LEDs were limited
`to small-signal applications requiring
`1
`10
`Flux (lm)
`than 100mlm of
`less
`flux per
`indicator function or display pixel. Around 1985, LEDs started to step beyond these low-flux small
`signal applications and to enter the medium-flux power signaling applications with flux requirements
`of 1-100lm (see Figure 3). The first application was the newly required center high-mount stop light
`(CHMSL) in automobiles. The first solutions were crude and brute-force: 75 indicator lamps in a
`row or in a two-dimensional array. It did not take long to realize that more powerful lamps could
`reduce the lamp count, a significant cost advantage. This was the first situation where efficiency became an
`
`
`8 For comparison, a 60W incandescent lamp emits 6 orders of magnitude higher light flux (about 900lm).
`
`6
`
`Page 6 of 24
`
`

`

`issue and for which the market was willing to pay a premium.9 So, in the late 1980’s, we saw the first horse
`race for efficiency improvements. By 1990, efficiencies reached 10lm/W in the GaAlAs materials
`system, for the first time exceeding that of equivalent red filtered incandescent lamps. Nevertheless,
`even higher efficiencies were desired to continue to decrease the number of lamps required per
`vehicle. Plus, the GaAlAs system was limited in color to a deep red, above 640nm.
`This horse race triggered the exploration of new materials system with still higher efficiency and a
`wider color range. First emerged GaAlInP materials, covering the range of red to yellow/green and
`quickly exceeded 20lm/W in the 620nm red/orange part of the spectrum. In 1995, Hewlett-Packard
`projected a room-temperature efficiency of 50lm/W by the Year 2000, with a theoretically possible
`efficiency of 150lm/W that could challenge that of even the most efficient conventional light source,
`the yellow low-pressure sodium lamp. This projection spawned a joint venture with Philips, and
`accelerated the use of LEDs in power signaling applications.
`In 1993, there was another breakthrough in LED technology. Based on work at several
`universities, both in the US and Japan, Nichia Chemical Corporation in Japan announced a fairly
`efficient blue material, GaN. Efficiency improvements followed quickly, together with an extension
`of the color range from blue to green (430-530nm). Now, LEDs could cover practically the entire
`visible spectrum, enabling their entry into additional power signaling applications such as traffic
`lights.
`
`103103
`
`103
`
`Figure 4. Historical and projected
`evolution of the performance (lm/package)
`and cost ($/lm) for commercially
`available red LEDs. This data was
`compiled by R. Haitz from HP historical
`records.
`
`Cost / Lumen ($/lm)
`
`~30 X Increase / Decade
`
`
`
`100100
`
`100
`
`~10X Reduction / Decade
`
`Flux / Package (lm)
`Flux / Package (lm)
`
`Before going on, we want to
`emphasize here the importance of
`the power signaling market on
`LED evolution. The penetration
`of LEDs
`into
`this market
`depended
`(and
`continues
`to
`1988
`depend) critically on performance
`Year
`and cost. Solutions based on large
`numbers of small-signal lamps are too expensive, thus demanding the development of higher-power
`LEDs. This evolution is illustrated in Figure 4 covering the period from first LED sales in 1968,
`projected to 2008. In a Moore's-law-like fashion, flux per unit has been increasing 30x per decade,
`and crossed the 10lm level in 1998. Similarly, the cost per unit flux – the price charged by the LED
`supplier to OEM manufacturers – has been decreasing 10x per decade and will reach 6cents/lm in
`2000. At this price, the LEDs in a typical 20-30-lm CHMSL contribute only $1.50 to the cost of the
`complete unit!10 In other words, the power signaling market drove, and continues to drive,
`improvements in the design and manufacturing infrastructure of the compound semiconductor
`materials and devices on which LEDs are based.
`These improvements have led to the LED efficiencies summarized in Figure 5 for the visible
`wavelength range 450-650nm. Because the efficiencies vary with temperature, the data shown refer to
`
`
`10-310-3
`1968
`
`1978
`
`1998
`
`10-3
`2008
`
`
`9 Back in the small signal days where one lamp was used per function, a 2x improvement in efficiency did not allow
`customers to use half a lamp. And, to reduce the drive current of an indicator lamp from 20mA to 10mA did not matter
`very much in an instrument that used 10-100W for other electronic functions.
`10 Although this cost is higher than that of an incandescent light bulb, it is low enough that other factors, such as
`compactness, styling freedom and absence of warranty cost, easily make up the difference.
`
`7
`
`Page 7 of 24
`
`

`

`a junction temperature of 85°C. For the GaAlInP material system (red to yellow), we show efficiency
`data for: (a) the expected Year 2000 production capability of the industry, (b) the expected Year 2005
`production capability, and (c) the best results reported as of 1999, shown to substantiate our
`confidence in the Year 2005 forecast. For the GaInN material system (green to blue), we show
`efficiency data for: (a) current average production performance of the industry leader, Nichia, and (b)
`a curve that is 50% higher. According to Nichia, their best results seem to be 50% above their
`average, and we assume that these best results will become average industry production within five-
`six years (by the Year 2005).
`80
`InGaN
`
`GaAlInP
`
`Figure 5. LED efficiency at an 85°C junction
`temperature as a function of wavelength. For the
`two dominant materials systems (GaAlInP and
`GaInN) we show current production data and
`our best estimate for Year 2005 production.
`
`Best Reported
`Results
`
`650
`
`60
`
`2005
`Production
`
`2000
`Production
`
`40
`
`20
`
`0
`
`450
`
`Luminous Efficiency (lm/W)
`
`At this point, LEDs of reasonable
`efficiency span virtually the entire visible
`2005
`wavelength range (with the exception of
`Production
`a narrow window in the yellow-green),
`2000
`and it is possible to create white light
`Production
`sources. One approach, which gives
`600
`550
`500
`white light sources with excellent color
`Peak Wavelength (nm)
`rendering properties, involves combining
`3-6 LEDs of different colors. Another approach involves combining a blue LED with down-
`conversion phosphors in a relatively inexpensive package. Both of these approaches involve some
`losses (color mixing in the former and photon down-conversion in the latter), but nevertheless can
`achieve good overall efficiencies. In fact, assuming the efficiencies of Figure 5, and a color mixing
`loss of 15%, semiconductor white light sources made with red, yellow, green and blue LEDs will
`already exceed that of standard 60-100W incandescent lamps in the Year 2000.
`
`Lamp Type
`
`Table 1. Efficiencies and lifetimes of various conventional
`and semiconductor white light sources. Similar to Figure
`5, the semiconductor white light sources refer to a junction
`temperature of 85°C.
`
`Standard Incandescent
`Standard Incandescent
`Long Life Incandescent
`Halogen
`Halogen
`Compact Halogen
`Compact Fluorescent
`Standard Fluorescent
`White LED 2000
`White LED 2002
`White LED 2005
`White LED 2010
`
`Power Efficiency Lifetime
`(W)
`(lm/W)
`(hrs)
`15
`8
`1,000
`100
`15
`1,000
`135
`12
`5,000
`20
`12
`3,000
`300
`24
`3,000
`50
`12
`2,500
`11
`50
`10,000
`30
`80
`20,000
`Any
`20 100,000
`Any
`30 100,000
`Any
`40 100,000
`Any
`50 100,000
`
`in Table 1, which
`illustrated
`is
`This
`compares current and projected efficiencies of
`white LED-based lamps with those of the most
`widely used conventional white light lamps. The
`most popular incandescent lamps with a power
`rating of 60-100W have an efficiency of around
`15lm/W and a rated life of 1,000 hours. The
`efficiency of incandescent lamps drops off at
`lower power ratings or for lamps with a longer 3,000-6,000 hour rated life. Halogen lamps show a
`similar pattern covering the range of 12-24lm/W. Fluorescent lamps at 80lm/W are the most efficient
`white light sources and dominate commercial and industrial lighting applications.
`In comparison, using the projections shown in Figure 5, LED-based white light sources will have
`efficiencies of 20lm/W in the Year 2000, should reach 40lm/W in the Year 2005, eventually leveling
`off in the 40-60lm/W range by the Year 2010. These efficiencies exceed significantly those of
`standard 60-100W incandescent lamps.
`
`8
`
`Page 8 of 24
`
`

`

`Moreover, the comparison between LED and incandescent lamp efficiencies favors LEDs even
`more in the case of monochrome applications. For these applications, there are no color-mixing
`losses for the LEDs, but there are additional filtering losses for incandescent lamps. 11 As shown in
`Table 2, LED efficiencies exceed those of filtered incandescent lamps by a large margin over the
`entire visible wavelength range except for yellow, where the two technologies are close to parity.
`
`
`Color Filtered Long-Life
`Incandescent
`Efficiency (lm/W)
`1-6
`4-8
`3-10
`1-4
`12
`
`Red
`Yellow
`Green
`Blue
`White
`
`Year 2000
`LED Production
`(lm/W)
`16
`10
`48
`13
`20
`
`Table 2. Current (Year 2000) LED efficiencies in broad color
`ranges as compared to those of filtered long-life incandescent lamps.
`The LED efficiencies refer to a junction temperature of 85°C.
`
`
`
`3 LED PENETRATION INTO POWER SIGNALING AND LIGHTING
`APPLICATIONS
`
`The penetration of LEDs into the signaling and lighting markets is a complex issue. Like in any new
`technology, in the early years LED solutions will be considerably more expensive than conventional
`solutions. To justify their selection, the higher initial cost has to be compensated with lower
`operating costs or other tangible benefits.
`With the dramatic progress that has been made in LED performance and cost over the past
`decades, however, LEDs have already begun to penetrate a number of monochrome signaling
`applications. We describe several of these applications in Appendix A, which include traffic and
`automotive lights, and large-screen outdoor TVs. Energy savings are the driving force for traffic
`lights; ruggedness, long life and styling are important factors in automotive tail lights; and lamp
`density and integrability are the key factors in TV screens with 3,000,000 pixels over an area of 600m2.
`The penetration of LEDs into white light applications will be much more difficult. A
`comparison between Table 2 (monochrome efficiencies) and Table 1 (white light efficiencies) shows
`why. At Year 2000-2005 performance levels, an LED-based red traffic light consumes 10x less
`power than its filtered incandescent alternative, while an LED-based white light consumes only 2x
`less power than its standard incandescent alternative, and about 2-3x more power than its fluorescent
`alternative.
`As a consequence, in the very near term, the white light applications that can realistically be
`attacked will be lower-flux "specialty" lighting applications in the 50-500lm range, currently
`dominated by incandescent and compact halogen lamps with relatively modest efficiencies in the
`range of 8-12lm/W. We describe several of these applications in Appendix A, which include accent
`and landscape lights, and flashlights.
`General lighting of residential, office, retail or industrial buildings, which consumes much more
`total energy than either signaling or specialty white lighting, will be much more difficult to penetrate
`for several reasons, the foremost being cost. Lamp cost: a 100 W incandescent lamp delivering a
`
`
`11 Note that this comparison does require some caution, due to the variability in efficiency of the filters used to produce
`various colors. For instance, the filter used in a red traffic light absorbs 90% of the white light and results in a deep red
`color. The red filter of an automobile taillight has a wider transmission band and yields an orange-red color. Yellow and
`green filters are fairly efficient and transmit a large fraction of the white spectrum. Blue filters are comparable to the
`transmission of red filters. Nevertheless, filtered incandescent color sources will always be less efficient than unfiltered
`white sources, while LEDs are inherently monochrome and do not suffer filtering losses.
`
`9
`
`Page 9 of 24
`
`

`

`flux of 1.5klm costs only $0.50, or $0.33/klm, while a comparable LED-based light source would cost
`over $150, or roughly $100/klm. Efficiency: Incandescent lamps with a rating of 60-150W have an
`efficiency of 14-16lm/W. To recover the initial difference in lamp cost in a reasonable time, today's
`white LED efficiency of 20lm/W is insufficient. White LEDs will not cross the critical threshold of
`30lm/W before 2002. Maintenance labor cost: The majority of incandescent lamps are used in
`residential buildings where the cost of maintenance labor is not an issue.
`Penetration into these higher-flux general lighting markets thus depends on continued efficiency
`improvements to the point where the energy savings pay back the initial cost penalty in a reasonable
`time, i.e. in six years or less. To quantify this, we define a "breakeven" time, which is the period over
`which energy savings equal the difference in initial lamp costs. A simple calculation of breakeven
`times is given in Appendix B for a standard 100 W incandescent lamp and an LED lamp of
`equivalent flux. For example, in the Year 2002, when LED lamp retail prices are expected to be of
`the order 100$/klm with an efficiency of 30lm/W, the breakeven time for a daily operating time of 12
`hours is just about six years. This is a marginal payback situation and penetration will be quite
`limited. But continued improvements in LED cost and efficiency should gradually expand the
`penetration.
`
`Signaling
`
`Lighting
`
`High Flux
`White Lighting
`(2010’s)
`
`Low-Medium
`Flux
`White Lighting
`(2000’s)
`
`General
`Lighting
`
`Specialty
`Lighting
`
`Figure 6: The stepping stones from
`LED indicators to LED illumination
`over half a century from 1970 to 2020.
`Signaling applications are mostly
`monochrome; lighting applications are
`mostly white. Specialty lighting includes
`monochrome and low/medium flux white
`lighting and is dominated by incandescent
`lamps. General lighting includes high
`flux white lighting and is dominated by a
`combination of incandescent and
`fluorescent lamps.
`
`10,000
`
` 1,000
`
` 100
`
`Traffic lights (2000’s)
`
`Automotive (1990’s)
`
` 10
`
`Flux
`
` 1
`
`Outdoor TVs (1990’s)
`
` 0.1
`
`Outdoor displays (1980’s)
`
` 0.01
`
`Indicators (1970’s)
`
` 0.001
`
`Red Yellow Green Blue White
`Colors
`
`It is helpful at this point to
`remind ourselves
`that
`these
`improvements
`will
`almost
`certainly continue at a rapid rate,
`due to the pressure that has been,
`and will continue to be, supplied by the power signaling market. To emphasize this, we show in
`Figure 6 the key stepping stones in the cost evolution of LEDs. Large outdoor displays with
`thousands of LED lamps made sense only after the growing volume for indicator lamps had reached
`hundreds of millions of units per month at a price of 10 cents or less per unit. LEDs in automotive
`rear combination lamps will not make economic sense until the cost/lumen approaches 5cents/lm.
`Replacing a red traffic light with 12-18 LEDs has created LED power packages that can handle a heat
`dissipation of several Watts at a reasonable cost. In turn, such a capability is needed for the front
`turn indicators which are mounted close to the head lamps of the car. The cost sensitive and
`potentially huge automotive market will force the industry along a steep cost learning curve. And, it
`is this cost pressure that will enable white LEDs to cross the critical threshold of 100$/klm and
`30lm/W that we estimate will be achieved in the Year 2002.
`When this critical threshold is achieved, LED-based white lamps will begin to replace
`incandescent and compact halogen lamps in the following