Optimum design domain of LED-based solid state lighting considering cost, energy consumption and reliability Bong-Min Song a,1 , Bongtae Han a,⇑ , Joon-Hyun Lee b a Mechanical Engineering Department, University of Maryland, College Park, MD 20742, United States b School of Mechanical Engineering, Pusan National University, Busan, South Korea article info Article history: Received 11 May 2012 Received in revised form 9 September 2012 Accepted 23 October 2012 Available online xxxx abstract Design parameters of LED-based solid state lighting products are interdependent and the corresponding requirements are dictated by operating conditions. We propose a scheme to define optimum design domains of LED-based luminaires for a given light output requirement by taking cost, energy consump- tion and reliability into consideration. First three required data sets to define design domains are expressed as contour maps in terms of the forward current and the junction temperature (I f and T j ): (1) face lumen and cost requirement as lumen/LED; (2) power consumption and energy requirement as luminaire efficacy (LE); and (3) reliability requirement as L70 lifetime. Then, the available domain of design solutions is defined as a common area that satisfies all the requirements of a luminaire. The pro- posed scheme is implemented for a wall wash light and the optimum design solutions are presented. Ó 2012 Elsevier Ltd. All rights reserved. 1. Introduction Recently LED-based luminaires have emerged rapidly for com- mercial and residential applications. For a required light output, the optimum design of a LED-based luminaire can be achieved by considering cost, energy consumption and reliability. Design considerations for LED-based luminaires are unique in that many design solutions are possible for the same required light output unlike the conventional light sources (e.g., compact flores- cent light, incandescent light, etc.). This is due to the well-known fact that the lumen output that each LED produces is a function of the driving current, I f , as well as the junction temperature, T j . Fig. 1 illustrates the details. The luminous flux of LEDs increases as the driving current increases, but the corresponding luminous efficacy decreases at a higher driving current because the luminous flux is not linearly proportional to a driving current. More impor- tantly, when a higher driving current is used, heat flux becomes proportionally larger, resulting in a higher junction temperature. This increased junction temperature not only decreases the lumi- nous flux but also significantly affects the rate of lumen mainte- nance (i.e., life time). A lower driving current can be utilized if more LEDs are used in the light engine of a luminaire. Yet this is often not the most desired solution due to the high cost of LEDs. In an LED-based luminaire design, all design parameters are interdependent and the corresponding requirements are dictated by operating conditions (I f and T j ). This paper suggests a methodol- ogy to define the optimum design domains of LED-based lumi- naires considering cost, energy consumption and reliability. The required data sets to define design domains are described in Sec- tion 2. An application using the requirements of wall wash light is provided in Section 3. 2. Design considerations and required data sets Optimum design of LED-based luminaires can be achieved by considering cost, energy consumption and reliability. These design parameters can be expressed as lumen/LED, luminaire efficacy, and L70 lifetime, respectively. Since they are a function of operating conditions, the parameters can be quantified in the domain of junc- tion temperature (T j ) and forward current (I f ). This section describes the three interdependent design parame- ters and defines the data sets required to obtain them. Test data obtained from a commercial phosphor converted LED are used to illustrate the concept. 2.1. Cost (lumen/LED) The Department of Energy (DoE) of U.S. published the require- ments for solid state lighting luminaires [1]. To deliver a required light output, two different approaches can be considered depend- ing upon the nature of applications. For medical and military appli- cations where stringent reliability requirements must be met, the required light output should be delivered at a junction tempera- ture that is sufficiently low to satisfy the high reliability standards. For a given passively or actively-cooled luminaire, a low junction 0026-2714/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.microrel.2012.10.010 ⇑ Corresponding author. Tel.: +1 3014055255. E-mail address: [email protected](B. Han). 1 Participant in the University of Maryland/Pusan National University Joint Doctoral Program. Microelectronics Reliability xxx (2012) xxx–xxx Contents lists available at SciVerse ScienceDirect Microelectronics Reliability journal homepage: www.elsevier.com/locate/microrel Please cite this article in press as: Song B-M et al. Optimum design domain of LED-based solid state lighting considering cost, energy consumption and reliability. Microelectron Reliab (2012), http://dx.doi.org/10.1016/j.microrel.2012.10.010
Transcript
Microelectronics Reliability xxx (2012) xxx–xxx
Contents lists available at SciVerse ScienceDirect
Optimum design domain of LED-based solid state lighting considering cost,energy consumption and reliability
Bong-Min Song a,1, Bongtae Han a,⇑, Joon-Hyun Lee b
a Mechanical Engineering Department, University of Maryland, College Park, MD 20742, United Statesb School of Mechanical Engineering, Pusan National University, Busan, South Korea
a r t i c l e i n f o
Article history:Received 11 May 2012Received in revised form 9 September 2012Accepted 23 October 2012Available online xxxx
0026-2714/$ - see front matter � 2012 Elsevier Ltd. Ahttp://dx.doi.org/10.1016/j.microrel.2012.10.010
1 Participant in the University of Maryland/PusaDoctoral Program.
Please cite this article in press as: Song B-M etreliability. Microelectron Reliab (2012), http://d
a b s t r a c t
Design parameters of LED-based solid state lighting products are interdependent and the correspondingrequirements are dictated by operating conditions. We propose a scheme to define optimum designdomains of LED-based luminaires for a given light output requirement by taking cost, energy consump-tion and reliability into consideration. First three required data sets to define design domains areexpressed as contour maps in terms of the forward current and the junction temperature (If and Tj):(1) face lumen and cost requirement as lumen/LED; (2) power consumption and energy requirementas luminaire efficacy (LE); and (3) reliability requirement as L70 lifetime. Then, the available domain ofdesign solutions is defined as a common area that satisfies all the requirements of a luminaire. The pro-posed scheme is implemented for a wall wash light and the optimum design solutions are presented.
� 2012 Elsevier Ltd. All rights reserved.
1. Introduction
Recently LED-based luminaires have emerged rapidly for com-mercial and residential applications. For a required light output,the optimum design of a LED-based luminaire can be achievedby considering cost, energy consumption and reliability.
Design considerations for LED-based luminaires are unique inthat many design solutions are possible for the same required lightoutput unlike the conventional light sources (e.g., compact flores-cent light, incandescent light, etc.). This is due to the well-knownfact that the lumen output that each LED produces is a functionof the driving current, If, as well as the junction temperature, Tj.
Fig. 1 illustrates the details. The luminous flux of LEDs increasesas the driving current increases, but the corresponding luminousefficacy decreases at a higher driving current because the luminousflux is not linearly proportional to a driving current. More impor-tantly, when a higher driving current is used, heat flux becomesproportionally larger, resulting in a higher junction temperature.This increased junction temperature not only decreases the lumi-nous flux but also significantly affects the rate of lumen mainte-nance (i.e., life time). A lower driving current can be utilized ifmore LEDs are used in the light engine of a luminaire. Yet this isoften not the most desired solution due to the high cost of LEDs.
In an LED-based luminaire design, all design parameters areinterdependent and the corresponding requirements are dictated
by operating conditions (If and Tj). This paper suggests a methodol-ogy to define the optimum design domains of LED-based lumi-naires considering cost, energy consumption and reliability. Therequired data sets to define design domains are described in Sec-tion 2. An application using the requirements of wall wash lightis provided in Section 3.
2. Design considerations and required data sets
Optimum design of LED-based luminaires can be achieved byconsidering cost, energy consumption and reliability. These designparameters can be expressed as lumen/LED, luminaire efficacy, andL70 lifetime, respectively. Since they are a function of operatingconditions, the parameters can be quantified in the domain of junc-tion temperature (Tj) and forward current (If).
This section describes the three interdependent design parame-ters and defines the data sets required to obtain them. Test dataobtained from a commercial phosphor converted LED are used toillustrate the concept.
2.1. Cost (lumen/LED)
The Department of Energy (DoE) of U.S. published the require-ments for solid state lighting luminaires [1]. To deliver a requiredlight output, two different approaches can be considered depend-ing upon the nature of applications. For medical and military appli-cations where stringent reliability requirements must be met, therequired light output should be delivered at a junction tempera-ture that is sufficiently low to satisfy the high reliability standards.For a given passively or actively-cooled luminaire, a low junction
LED-based solid state lighting considering cost, energy consumption and12.10.010
Fig. 1. Effect of If on performance of LED-based luminaire.
2 B.-M. Song et al. / Microelectronics Reliability xxx (2012) xxx–xxx
temperature can be achieved only by a low forward current level,which forces to employ as many LEDs as needed. For commercialand residential applications, however, LED-based luminairesshould use only a minimum number of LEDs to be cost-effectiveand thus to be competitive with the conventional light sources.Therefore, the light output of a single LED at a specified conditionis a very important parameter. Since the light output is a functionof forward current and junction temperature, the light output hasto be measured at various forward current and junction tempera-ture conditions.
Fig. 2 shows an experimental setup used to acquire lightoutput at different currents and junction temperatures. An LED(CREE XR-E) is mounted on a thermoelectric cooler (TEC) in anintegrating sphere. The LED has a conformally coated phosphorlayer on the chip and encapsulated with transparent silicone. Thetop part dome is glass. The package is surrounded with the alumi-num reflector. The TEC controls the solder point temperature2 of LED(Ts) and the integrating sphere (SMS-500: Labsphere) equipped witha spectroradiometer system measures the spectral power distribu-tion (SPD) of the LED at the various operating conditions. The sam-pling interval of the spectroradiometer is 1 nm. The measurementrange of wavelength is 360–1000 nm. The TEC is controlled by athermal controller (LB 320: Silicone Thermal. Co. Ltd.) using an inputprovided by a T-type thermocouple directly mounted on the solderpoint surface. The setup can adjust the solder point temperaturefrom �40 to 125 �C with a resolution of 0.1 �C.
The junction temperature can be estimated by the well-knownforward voltage method [2]. The method is valid only when apulsed current with a very short duration is used so that thejunction temperature does not change during forward voltagemeasurement. For the SPD measurement, however, the spectrora-diometer requires an integration time [3], which is usually muchlonger than the short pulse duration for the forward voltagemethod: a typical integration time for a full white light spectrumis 100–1000 ms. As a result, an additional increase of junctiontemperature is unavoidable during the SPD measurement.
In order to avoid this undesired error in junction temperaturemeasurement, an LED is subjected to a steady-state condition
2 The ‘‘solder point’’ is the point at which the solder meets with a printed circuitboard and LED leads. Some LED manufactures refer to ‘‘case temperature but referringto the solder point temperature is more precise.
Please cite this article in press as: Song B-M et al. Optimum design domain ofreliability. Microelectron Reliab (2012), http://dx.doi.org/10.1016/j.microrel.20
inside an integrating sphere for accurate junction temperaturemeasurement. With the 4p configuration in Fig. 2, a current is ap-plied to the LED continuously until the solder point temperaturereaches a preset value which is maintained by controlling theTEC. Once the solder point temperature is stabilized, the SPD andthe forward voltage of the LED are measured simultaneously.
The SPDs have been measured at various solder point tempera-tures and forward currents: Ts from 65 to 125 �C with a constantinterval of 15 �C and If from 300 to 1000 mA with a constant inter-val of 100 mA. The total number of measurements is 40. TypicalSPDs obtained at Ts = 65 and 125 �C at If = 300 and 1000 mA areshown in Fig. 3. The radiant flux and the corresponding luminousflux obtained from all 40 measurements are shown in Fig. 4a andb, respectively. As expected, the luminous flux increases as the sol-der point temperature decreases under a constant current.
The measurement data obtained for solder point temperaturescan be converted to the junction temperature domain throughthe relationship between junction temperature and solder pointtemperature, which can expressed as [4–6]:
Tj ¼ Ts þ Pð1� gPÞRjs ¼ Ts þ If V f ð1� gPÞRjs ð1Þ
where Ts is the solder point temperature, P is the total electricalpower consumption in W ( = If � Vf), gp is the LED power efficiency(the radiant flux divided by the total electrical power input), If isthe forward current, Vf is the forward voltage and Rjs is the thermalresistance between the chip and the solder point (8 �C/W for theLED tested in this study [7]). It is important to note that power effi-ciency in Eq. (1) was often ignored in the thermal analysis of lowpower LEDs due to low efficiency [8,9]. For high power LEDs withtypical power efficiency of 15–30%, the effect of the power effi-ciency must be considered in a thermal analysis.
The forward voltage is measured first as a function of solderpoint temperatures (Fig. 5a), from which the total electrical powerconsumption is determined. The results are shown in Fig. 5b. Thepower efficiency then can be determined simply by dividing the to-tal power consumption by the radiant flux amount shown inFig. 4a, and the results are shown in Fig. 6. The power efficiencyvaries significantly with the forward current.
The solder point temperatures in Fig. 6 can be converted to thecorresponding junction temperatures using Eq. (1). The convertedpower efficiency as a function of junction temperature is shownin Fig. 7. It is worth noting that the power efficiency decreases al-most linearly as the junction temperature increases and the rate ofthe power efficiency reduction remains the same regardless of thecurrent.
The luminous flux data of Fig. 4b can be also presented in thejunction temperature domain using Eq. (1) and the power effi-ciency data. The converted luminous flux data is shown in Fig. 8.From the data in Fig. 8, a contour plot of the luminous flux of a sin-gle LED in a domain of the forward current and the junction tem-perature can be determined (Fig. 9). The contour plot shows thedistinctive characteristic of LED luminous flux, which depends onthe junction temperature as well as the forward current.
2.2. Energy consumption (luminaire efficacy)
One of the key attributes of LED lighting is low power consump-tion. In other words, LED-based luminaire can deliver the sameamount of luminous flux with lower power consumption com-pared to the conventional light sources. This property is expressedas luminaire efficacy [10], which is defined as the total lumens pro-duced by a luminaire divided by the total wattage drawn by thepower supply/driver, expressed in lumens per watt (lm/W); i.e.,
jLEjluminaire ¼ jLEjLED � Ffixture ð2Þ
LED-based solid state lighting considering cost, energy consumption and12.10.010
B.-M. Song et al. / Microelectronics Reliability xxx (2012) xxx–xxx 3
where |LE|luminaire is the luminaire efficacy (lm/W), |LE|LED is theluminous efficacy of each LED (lm/W) and Ffixture is the luminairefixture efficiency (%). It is important to note that the luminaire effi-cacy remains the same regardless of the number of LEDs used in aluminaire as long as the forward current and the junction tempera-ture remain the same.
Luminaire efficacy is a function of luminous flux and powerconsumption. The luminous efficacy of an LED can be obtainedby dividing the lumen/LED (Fig. 9) by power consumption(Fig. 5b). The luminaire efficacy contour is shown in Fig. 10, wherethe fixture efficiency of 90% is assumed based on the values pro-vided in Ref. [11]. The light output does not increase linearly withthe increasing current due to the efficiency drooping [12–14] at afixed junction temperature and it is well known that LED light out-put decreases with an increasing junction temperature at a fixedcurrent. As expected, the luminaire efficacy increases with thedecreasing forward current and junction temperature.
2.3. Reliability (L70 lifetime)
All electrically powered light sources experience light reductionas they continuously illuminate light under normal operating con-ditions. This is known as lumen depreciation [15]. The lifetime ofan LED is characterized by its L70 lifetime, which is the time to70% of an initial light output. In order to estimate the LED lifetime,major LED manufacturers have adopted IESNA LM-80 [16]. Thedocument prescribes standard test methods to measure lumenmaintenance of LEDs under controlled conditions (constant junc-tion and ambient temperatures with a constant DC mode). SinceLEDs have very long expected lifetime (>30,000 h), a complete life-time test is not practical. The LM-80 recommends a minimum of
Fig. 3. SPDs at various conditions; Ts = 65 and 125 �C at If = 300 and 1000 mA.
Please cite this article in press as: Song B-M et al. Optimum design domain ofreliability. Microelectron Reliab (2012), http://dx.doi.org/10.1016/j.microrel.20
6000 h testing with 1000 h measurement interval. Then, the life-time of LED has to be extrapolated from the test data. Althoughthe empirical equation of LED lumen depreciation is known to takean exponential form [5,6], LED manufacturers use their ownextrapolation schemes [17,18].
The LED lifetime is dictated by operating conditions: longer life-times are expected at the lower junction temperature and the low-er forward current. Ambient temperature is also an importantfactor for the LED lifetime. When LEDs are exposed to high temper-ature ambient conditions, the package of LEDs is degraded [19,20].The well-known package degradation mechanisms are reflectivecoating degradation and optical lens yellowing. They not only de-crease the total light output but also change CCT [5,20,21].
Fig. 11 shows the L70 lifetime at different junction temperaturesand forward currents at two different ambient temperatures of65 �C and 75 �C [18]. The data in Fig. 11 is converted to contourplots in Fig. 12. The contour plot clearly shows that the lifetimeof LED is more sensitive to the junction temperature than the for-ward current under the same ambient temperature.
2.4. Extra consideration: minimum junction temperature
In the previous sections the data has been shown over thewhole junction temperature range. In practice, the junction tem-
60 70 80 90 100 110 120 13040
60
80
100
120
140
160
180
Lum
inou
s flu
x (lm
)
Solder Point Temperature (oC)
800 (mA) 900 (mA) 1000 (mA)
(b) Fig. 4. (a) Radiant flux and (b) luminous flux at different forward currents as afunction of solder point temperatures.
LED-based solid state lighting considering cost, energy consumption and12.10.010
Fig. 8. Luminous flux at different currents as a function of junction temperatures.
4 B.-M. Song et al. / Microelectronics Reliability xxx (2012) xxx–xxx
perature is always higher than the solder point temperature (and theambient temperature), and there should be a low bound of the junc-tion temperature, below which a design solution is not valid.
Fig. 6. Power efficiency of LED at different currents as a function of solder pointtemperatures.
Please cite this article in press as: Song B-M et al. Optimum design domain ofreliability. Microelectron Reliab (2012), http://dx.doi.org/10.1016/j.microrel.20
The thermal network model can be utilized to determine thelow bound. Fig. 13 illustrates a simplified thermal resistance net-work model of the LED considering only the downward paththrough the heat slug to the mold, substrate and solder point [4].
8090100
110
120
130
140150
160
7080 90 100 110 120 130 140
300
400
500
600
700
800
900
1000
Forw
ard
Cur
rent
(mA)
Junction temperature (oC)
Fig. 9. Lumen/LED as a function of If and Tj.
LED-based solid state lighting considering cost, energy consumption and12.10.010
Fig. 10. Luminaire Efficacy as a function of If and Tj.
80 90 100 110 120 130 140 1500
10k
20k
30k
40k
50k
60k
70k
80k
90k
100k
L70
Life
time
(hou
rs)
Junction Temperature (oC)
350 mA (Ta = 65oC) 700 mA (Ta = 65oC) 1000 mA (Ta = 65oC) 350 mA (Ta = 75oC) 700 mA (Ta = 75oC) 1000 mA (Ta = 75oC)
Fig. 11. L70 lifetime vs. Tj at different Ifs at Ta = 65 �C and 75 �C [18].
75k 70k 65k 60k 55k 50k 45k
40k 35k 30k
80 90 100 110 120 130 140
400
500
600
700
800
900
1000
Forw
ard
Cur
rent
(mA)
Junction Temperature (oC)
(a)
75k 65k 60k 55k
50k
45k40k
35k
30k 2
80 90 100 110 120 130 140
400
500
600
700
800
900
1000
Forw
ard
Cur
rent
(mA)
Junction Temperature (oC)
(b) Fig. 12. L70 lifetime as a function of If and Tj at (a) Ta = 65 �C and (b) Ta = 75 �C.
Fig. 13. Simplified resistance network model of LED package.
B.-M. Song et al. / Microelectronics Reliability xxx (2012) xxx–xxx 5
The upward path through the encapsulant to the lens surface canbe ignored since the upward heat transfer is less than 1% of the to-tal heat transfer.
The network model can be further simplified by breaking thefull network into two parts: internal (conductive) and external(convective and radiative) resistances [4]. It is expressed as:
q ¼ Ts � Ta
Rcr¼ Tj � Ts
Rjsð3Þ
where q is the heat generated by the LED (W) at a given current; Rjs
is the conduction resistance in (�C/W); Rcr is the effective externalresistance (�C/W); Tj is the junction temperature (�C); Ts is the sol-der point temperature of the downward path (�C); and Ta is theambient temperature. The effective internal thermal resistance forconduction, Rjs, and the effective external thermal resistance(including convection and radiation), Rcr, can be defined as:
Rjs ¼t
kAcdand Rcr ¼
1Acrðhþ hrÞ
ð4Þ
where t is the thickness (m), k is the heat conductivity (W/mK), h isthe convection coefficient (W/m2), hr is the effective convectioncoefficient for radiation (W/m2), Acd and Acr are the effective areasof the conduction and the convection/radiation (m2), respectively.
Please cite this article in press as: Song B-M et al. Optimum design domain ofreliability. Microelectron Reliab (2012), http://dx.doi.org/10.1016/j.microrel.20
Since Ts is always higher than Ta, the lowest possible junctiontemperature Tmin
j can be defined as:
Tminj ¼ Rjs � qmin þ Tmin
s ð5Þ
where qmin is the minimum heat generation at a given forward cur-rent, which can be defined as:
qmin ¼ Pminð1� gmaxÞ ð6Þ
The minimum power and the maximum power efficiency at agiven forward current can be obtained from Figs. 5b and 6,respectively.
The power consumption and the power efficiency of LED aredictated by the LED junction temperature. The minimum powerconsumption for a given forward current can be achieved at the
LED-based solid state lighting considering cost, energy consumption and12.10.010
Fig. 14. Minimum junction temperatures for different ambient temperatureconditions.
6 B.-M. Song et al. / Microelectronics Reliability xxx (2012) xxx–xxx
highest junction temperature while the maximum power effi-ciency for a given forward current can be achieved at the lowestjunction temperature (Fig. 7). Fig. 14 shows the minimum junctiontemperatures as a function of ambient temperatures obtained from
Fig. 15. Example of wa
8 LED
7 LED
6 LED
5 LED
9 LED
80 90 100 110 120 130 140300
400
500
600
700
800
900
1000
Forw
ard
Cur
rent
(mA)
Junction temperature (oC)
Fig. 16. Domain of lumen/LED for various numbers of LED.
Please cite this article in press as: Song B-M et al. Optimum design domain ofreliability. Microelectron Reliab (2012), http://dx.doi.org/10.1016/j.microrel.20
Eqs. (5) and (6). The junction temperatures shown in Fig. 14 are thepractical lower bound under the ambient temperatures.
3. Implementation for wall wash light
In order to illustrate the proposed approach, wall wash lightwas selected as an example for implementation. Wall wash lights(Fig. 15) are used to illuminate walls in buildings, bridges and tow-ers for decoration and lighting purpose. The requirements of a wallwash light are specified as [1]:
Assuming the same LED (CREE XR-E) is used for the wall washlight, the lumen/LED (Fig. 9) can be used first for cost consideration.The lumen output that each LED has to deliver is:
Lumen output ¼ 575N� 1
Ffixtureð7Þ
where N is the number of LEDs used in the luminaire and Ffixture isthe luminaire fixture efficiency. Considering again the typical fix-ture efficiency of 90%, the above relationship can be used to convert
ll wash light [22].
70
66
6260
56 54
52
50 48 46
44 42 40
80 90 100 110 120 130 140300
400
500
600
700
800
900
1000
Forw
ard
Cur
rent
(mA)
Junction temperature (oC)
Fig. 17. Domain of luminaire efficacy (lm/W) for wall wash light.
LED-based solid state lighting considering cost, energy consumption and12.10.010
Fig. 18. Domain of L70 lifetime for wall wash light under an ambient temperatureof 75 �C.
B.-M. Song et al. / Microelectronics Reliability xxx (2012) xxx–xxx 7
Fig. 9 to Fig. 16 where each contour line shows available operatingconditions for the corresponding number of LEDs used in the wallwash light. The next step is to define domains for the requiredluminaire efficacy of 40 lm/W using Fig. 10. The domain that satisfiesthe requirement of luminaire efficacy is shown in Fig. 17. Since LEDhas very high luminaire efficacy, most of the domain satisfies therequirement.
The lumen/LED can be converted to ‘‘Dollar/Lumen’’ if an LEDprice is given. Since wall wash lights illuminate walls, the irradi-ance parameter (lux) is more important than luminous flux inpractice. The luminous flux can be converted to lux if the wall con-dition and beam angle of the application are given.
Finally the required L70 lifetime of 35,000 h is considered. MostLED-based luminaires use secondary optics such as a reflector cupand a polycarbonate lens. The ambient temperature of LED is high-er than the air temperature surrounding the luminaire since theLED is confined in the secondary optics. The domain of the L70 life-time considering the ambient temperature of 75 �C is shown inFig. 18. It should be noted that the low bound of junction temper-atures defined in Section 2.4 is shown in Figs. 16–18 and the do-main below the minimum junction temperature is marked as ‘‘N/A’’.
8 LED
7 LED
6 LED
5 LED
9 LED
80 90 100 110 120 130 140300
400
500
600
700
800
900
1000
Forw
ard
Cur
rent
(mA)
Junction temperature (oC)
35k
40
Fig. 19. Design domain for wall wash light under an ambient temperature of 75 �C:red points represent the solutions for the maximum luminaire efficacy and lifetime;and blue points represent the solutions for the maximum allowable junctiontemperature. (For interpretation of the references to color in this figure legend, thereader is referred to the web version of this article.)
Please cite this article in press as: Song B-M et al. Optimum design domain ofreliability. Microelectron Reliab (2012), http://dx.doi.org/10.1016/j.microrel.20
Fig. 19 shows design solutions for a wall wash light under anambient temperature of 75 �C. Any operating conditions withinthe area satisfy the requirements for the corresponding numberof LEDs. Yet several points in the domain offer solutions that wouldcarry more importance in practice. The solutions indicated by reddots in Fig. 19 offer the maximum luminaire efficacy and lifetimewithin each LED-quantity based solution line: i.e., the lowest cur-rent as well as junction temperature. These solutions require amore expensive cooling solution to maintain the low junction tem-perature. As opposed to these, the solutions indicated by blue dotsoffer solutions with the most relaxed thermal requirement. The lu-men maintenance is obviously compromised in these solutions.
The optimized operating condition for a given number of LEDscan be chosen along the lumen/LED lines in Fig. 19. This is illus-trated for the solution with five LEDs. If the junction temperatureis to be 95 �C, the proper current level should be 645 mA (markedas ‘‘h’’ along the five LED solution line in Fig. 19). If the current lev-els are higher than the value along the dashed line (marked as ‘‘h’’in Fig. 19) while maintaining the same junction temperature(95 �C), the luminaire will produce the face lumen higher thanthe required value (575 lm): at the points marked as ‘‘x’’ the lu-men/LED will be 138 and 152 lm and the total luminous flux ofthe luminaire will be 621 and 684 lumen, respectively. This extraluminous flux will increase the L70 lifetime. It should be noted thatthis condition requires a better thermal management solution tomaintain the same junction temperature.
4. Conclusion
Design parameters of LED-based luminaires have been analyzedto suggest a design domain that optimizes cost, energy consump-tion and reliability. The required data sets were lumen/LED, lumi-naire efficacy, and L70 lifetime. The data sets were obtained bymeasuring the SPD of an LED as a function of forward current aswell as junction temperature (If and Tj). The minimum junctiontemperature limit was also defined as a low bound of a valid designdomain. The proposed scheme was implemented for a wall washlight and the optimum design solutions were presented. Thescheme is general and can be applied to LED-based luminaires witheither passive or active cooling solutions.
Acknowledgments
This work was partially supported by the University of Mary-land/Pusan National University Joint Doctoral Program funded inpart by the BK21 Program in Korea.
References
[1] ENERGY STAR� program requirements for solid state lighting luminaireseligibility criteria – version 1.1. U.S. D. o. Energy; 2008.
[2] Xi Y, Schubert EF. Junction temperature measurement in GaN ultraviolet light-emitting diodes using diode forward voltage method. Appl Phys Lett 2004;85:2163–5.
[3] Gaigalas AK et al. Procedures for wavelength calibration and spectral responsecorrection of CCD array spectrometers. J Res Natl Inst Stan Technol2009;114(July–August):215–28.
[4] Han B et al. Coupled thermal and thermo-mechanical design assessment ofhigh power light emitting diode. IEEE Trans Compon Pack Technol 2010;33:688–97.
[5] Ishizaki S et al. Lifetime estimation of high power white LEDs. J Light VisualEnviron 2007;31:11–8.
[6] Narendran N et al. Solid-state lighting: failure analysis of white LEDs. J CrystGrowth 2004;268:449–56.
[7] Cree� XLamp� XR-E LED. CLD DS-05 REV11. CREE. Inc.; 2009.[8] Narendran N et al. Estimating junction temperature of high-flux white LEDs.
Light-Emitting Diodes: Research, Manufacturing, And Applications VII.Proceeding of SPIE, vol. 5366. Society of Photo-Optical InstrumentationEngineers; 2004. p. 158–60.
LED-based solid state lighting considering cost, energy consumption and12.10.010
8 B.-M. Song et al. / Microelectronics Reliability xxx (2012) xxx–xxx
[9] Jayasinghe L et al. Characterization of thermal resistance coefficient of high-power LEDs, Presented at the sixth international conference on solid statelighting. Proceedings of SPIE 6337; 2006.
[10] Luminaire efficacy. Energy efficiency and renewable energy. led applicationseries, U.S. Department of Energy; 2008.
[11] Neary M, Quijano M. Solid state lighting for industrial locations. In: Petroleumand chemical industry conference, 2009. PCIC 2009. 2009 record of conferencepapers – Industry Applications Society 56th annual; 2009, p. 1–7.
[12] Li YL et al. Efficiency droop behaviors of InGaN/GaN multiple-quantum-welllight-emitting diodes with varying quantum well thickness. Appl Phys Lett2007;91:181113–181113-3.
[13] Kim M-H et al. Origin of efficiency droop in GaN-based light-emitting diodes.Appl Phys Lett 2007;91:183507–183507-3.
[14] Chao C-L et al. Reduction of efficiency droop in InGaN light-emitting diodegrown on self-separated freestanding GaN substrates. IEEE Photon TechnolLett 2011;23:798–800.
Please cite this article in press as: Song B-M et al. Optimum design domain ofreliability. Microelectron Reliab (2012), http://dx.doi.org/10.1016/j.microrel.20
[15] Lifetime of White LEDs. U.S. D. o. Energy; 2006.[16] Subcommittee on Solid State Lighting of the IESNA Testing Procedures
Committee. Approved method: measuring lumen maintenance of LED lightsources, LM-80-08. New York: Illuminating Engineering Society of NorthAmerica; 2008.
[17] Evaluating the Lifetime Behavior of LED Systems. Philips, Lumileds;2010.
