Topics on LED

Group members Sooseok Oh Xia Cao Lakshminarayanan Mukundagiri Fang Chu

Chapter 1
1.1 Background Information


The light emitting diode (LED) could be considered the ultimate general source of continuous light because of its high luminescence (EL) efficiency, quick response time, and long lifetime. For example, the electrical efficiency of a standard “white” light is almost halved if a color filter is employed to produce color, such as is the case for traffic signal lights. A more sensible approach would be to use colored LEDs. This would also reduce the amount of maintenance required, as standard traffic light signal lights need to be replaced every six months. The initial cost of replacing standard traffic signal lights with colored LEDs would be rewarded within ten years. In addition, the cost for white LEDs is reduced year by year with the advance of related technology. Therefore, the cost would not be a problem in the near future. Table1.1 is a comparison of light bulbs and LEDs for application in traffic lights and railway signals. Light Bulbs Red Yellow Green 70Watt 70Watt 70Watt LEDs Power consumption Red 18Watt Yellow 20Watt Blue ~ green 17Watt Replacement interval 6 ~ 12 months 5 ~ 10 years (estimated) Failure mode Sudden total failure Gradual intensity decrease Visibility Uses color filter, reflects sunlight Direct Cost Expensive at maintenance Cheap at maintenance Cheap at manufacturing Expensive at manufacturing Luminous Efficacy (lm/w) 10 ~ 85 lm/W 100 lm/W Table 1.1 For room light application, the power and stability requirement for LEDs is higher. Basically, 100 lm/W white LED is needed for meeting the requirement. Two announcements from Cree and Nichia make the room light application closer to reality and the Meijo University in Japan also announced high efficient white LED: (1) On Nov 16, 2005: A scientist, Prof. Satoshi Kamiyama, at the Meijo University in Japan has announced a white LED of 130 lm/W using a purple LED and a silicon carbide substrate. He will establish a

startup to manufacture and sell their white LED units using a purple LED on a silicon carbide substrate. (2) On June 20, 2006: Cree has demonstrated a cool-white LED that achieves an efficacy of 131 lm/W@20mA (3) On Nov 15, 2006: Nichia has developed white LEDs with indium tin oxide contacts that can deliver an efficacy of 138lm/W@20mA (4) On Dec 20, 2006: Nichia Achieved 150 lm/W White LED Development @ 20mA drive current 1.2 Introduction to Nitride-based LED Technology Nowadays, in industry, two major methods are to generate white LED. (see Fig1.1)

Fig1.1 White LED generation One is to mix RGB (red, green, blue) to generate white light. The other is to use blue LEDs with sufficiently high energy to excite phosphors or YAG phosphor powder to generate white light. With this method, only one blue LED chip is required. Most currently available white LEDs are based on the blue LED and phosphor approach. Therefore, the price is lower than that with the first method. From the information above, it is clearly known that the blue or purple LED is the key to white LED generation. The Fig 1.2 shows a comparison of white LED Technologies.

Fig 1.2 A comparison of white LED technologies [1.4] From semiconductor physics, light generation is due to the recombination of electrons in conduction band and holes in valence band. When an electron meets a hole then it will falls into a lower energy level and releases energy (Eph) in the form of a photon [1.5]. The emission wavelength is related to the energy bandgap (Eg) of material. For example, Eg (GaN) = 3.44eV and Eg(SiC) = 2.42eV. The related emission wavelength (λc) for GaN is around 360nm. That’s why GaN is called blue LED material. (See Fig. 1.3)

Fig. 1.3 Energy band diagram of LED

1.2.1 Properties of Nitride-based Semiconductor Compounds GaN and its related compounds can crystallize both in the zincblende as well as in the wurtzite structure. However, the wurtzite structure is more common. They are all direct bandgap materials where the minimum of the conduction band lies directly above the maximum of the valence band in momentum space, which is good for light generating. The advantages of nitride-based semiconductors are: excellent hardness, extremely large heterojunction offsets, high thermal conductivity, and high melting temperature. The basic GaN structure is shown in Fig1.4.

Fig 1.4 Structure of GaN For all commercial available products so far, they are fabricated on non-polar plane. Theoretically, growing GaN on polar plane could have better efficiency. However, it’s hard to grow GaN crystalline film due to higher dislocation density. In 2006, Nichia and research team in UCSB have announce that they can grow GaN crystalline film on semi-polar plan with breakthrough efficiency with low output power. 1.2.2 Challenges of Nitride-based LED Technology before 1993 The first commercial available blue InGaN LED was announced by Nichia in 1993 though Pankove (RCA Princeton Laboratory) demonstrated a metal-insulator-semiconductor GaN-based LED in 1971. Lots of fabrication difficulties were in the development process of GaN LED. A brief history from 1971 ~ 1999 is summarized in Table1.2. Time 1971 1974 1981 1986 1988 1990 Achievements Pankove demonstrates blue GaN MIS LED Pankove and Temple demonstrate cubic GaN Akasaki at Matsushita demonstrates the first flip-chip type MIS GaN LED with 10 milli-candela Akasaki grows high quality GaN using a-AlN buffer layers; Mizuta et al. grow cubic GaN Akasaki discovers p-type conducting GaN using low energy electron beam irradiation Nakamura develops new “two-flow” MOCVD equipment for growth of high quality single crystal GaN layers

Feb. 1991 March 1991 June 1991 1992 Sep. 1992 Nov. 1993 Nov. 1993 May 1994 Sept. 1995 Jan. 1996 1996 Nov. 1996 1999

Nakamura grows high quality p-type GaN Nakamura fabricates GaN pn-junction light emitting diode 3M reports ZnSe-CdZnSe based blue semiconductor laser Akasaki demonstrates GaN based blue pn-junction LED (light output: 1.5mW at room temperature, quantum efficiency: 1.5%) Nakamura fabricates InGaN double heterojunction LED Nakamura demonstrate 1 candela InGaN blue LED product Nichia announces commercial blue InGaN LEDs Nakamura demonstrates 2 candela InGaN blue LED product Nichia announces commercial green InGaN LEDs Nakamura reports pulsed blue InGaN injection laser at room temperature Nichia sells several million blue InGaN LEDs per month Nakamura announces the first CW blue GaN based injection laser at room temperature Nichia announces commercial violet InGaN laser diodes Table 1.2 [1.1]

In general, there were two main challenges of nitride-based LED technology before 1993. (1) Hard to grow high quality GaN crystalline films (2) Hard to have low-resistivity p-type doping GaN The first problem is due to the large lattice constant mismatch between GaN and Sapphire substrate. (see Fig1.3)


From Fig1.3, it shows that the lattice difference between all three nitride-based materials and Sapphire substrate is greater than 15%. Theoretically, only < 1% lattice mismatch is allowed to grow high quality film due to the stress and stored inner energy issues. This problem is solved by introducing buffer layer between substrate and high quality GaN film. (see Fig1.4)

Fig1.4: Interference micrographs of the surface of GaN films grown w/ and w/o the GaN buffer layer. The growth times of the buffer layer (a) 30s (b) 70s (c) 90s (d) 0s [1.2] High quality yet low cost GaN film growth on Silicon has also been reported through papers [1.6] [1.7] [1.8].

The second problem is about high quality low-resistivity p-type GaN firm. To obtain a higher external quantum efficiency of GaN LEDs, future improvement in crystal quality and p-type conductivity control of GaN films is required. Since GaN is III-V compound, elements in column II of periodic table can be used as p-type dopant while elements in column IV of periodic table can be used as n-type dopant. In reality, Mg is used to be the p-type dopant and Si (or Ge) is used as n-type dopant for GaN. For n-type doping, the doping efficiency of Ge is about one order of magnitude lower than that of Si. Therefore, Si is preferable for n-type GaN. How did Nakamura obtain low resistivity p-type GaN film? He used two patented inventions to solve high resistivity p-type GaN film problem. The first invention is two-gas flow MOCVD (TF-MOCVD). The setup is shown in Fig1.5.

Fig1.5 TF-MOCVD [1.3] Generally, for conventional MOCVD method, it uses a thin delivery tube to obtain a high gas velocity (~ 5m/s). So that it’s hard to obtain high quality film uniformly over the substrate. With TF-MOCVD method, a second gas jet consisting of nitrogen and hydrogen gas perpendicular to the substrate surface is introduced to push the reactants towards the growth surface leading to improved crystal growth. With this method, the main flow can still keep in high velocity and the uniformity of the grown film can be improved. In addition, thermal convection can be suppressed by the subflow. The second invention is about the thermal annealing. In 1992, Nakamura found that p-type GaN with low resistance could be obtained by thermal annealing of the GaN crystal after growth. This founding is a breakthrough step in obtaining p-type III-V nitride films because it is an easy, reliable, and mass-production technique. The resistivity change as a function of annealing temperature is shown in Fig 1.6. The thickness of p-type layer for this sample is 4µm and annealing time was performed for 20min in nitrogen atmosphere for each of the temperature regions. The resistivity of this sample was 1x106Ω-cm before annealing. Almost no change was observed when the sample was annealed at temperatures between room temperature and 400oC. However, when the annealing temperature was increased to 500oC, the resistivity began to decrease suddenly. At 700oC, it decreased to 5Ωcm. At 900oC, the resistivity, hole carrier concentration, and hole mobility became 2Ω-cm, 3x1017cm-3, and 10cm/(V-s).

Fig 1.6 Resistivity vs. Temperature With these two problems solved, Nichia announces commercial blue InGaN LEDs in 1993. This is double heterojunction (DH) type LED. The couple of standard blue LED structures are shown in Fig1.7.


(b) Fig 1.7 (a) InGaAlP Based LED (b) GaN based LED [1.9]

The In0.06Ga 0.94N layer is so-called active layer. The bandgap for InGaN can vary from 2.0eV to 3.4eV depending on the Indium mole fraction. In order to guide the generated photons, the refractive index for this layer should be lower than the surrounding layers. From the optical side, three layers, p-AlGaN, InGaN, and n-AlGaN, in Fig1.7(b) can be regarded as in Fig1.8. The efficiency improvement will be discussed further in chapter 3.

Fig1.8 1.2.3 Trend of LED structures Today, the structure for LEDs has moved from Double Heterojunction (DH) to single quantum well (SQW) and multiple quantum well (MQW) in order to obtain higher efficiency, better thermal performance, and freedom to design transition energies. The MQW has some advantages over the SQW because it has more than one well and more carriers can be accommodated. With the SQW structure, luminescence saturation occurs at high current densities due to band filling problem [1.10]. Two standard SQW and MQW LEDs are shown in Fig 1.9.

Fig 1.9 structure of SQW and MQW LED

1.2.4 Simulation of LED structure In order to reduce the cost, the simulation work is needed before doing the fabrication for these two types of structures. This means the correct and fast calculation simulation tools are on demand for MQW-LED design. The simulation part will be discussed in Chapter 2.

(1) Shuji Nakamura, Stephen Pearton, and Gerhard Fasol, (2002) “The blue light diode”, Springer. (2) S. Nakamura, (1991) “GaN growth using GaN buffer layer, Japanese Journal of Applied Physics, Part 2, 30, pp. L1705-L1707 (3) S. Nakamura, Y. Harada, M. Senoh, (1991) “Novel metalorganic chemical vapor deposition system for GaN growth”, Applied Physics Letters, 58, pp. 2021-2023 (4) Color Quality of White LEDs (5) HowStuffWorks - LED (6) Investigation of GaN crystal quality on silicon substrates using GaN/AlN superlattice structures Gwo-Mei Wu*, Chen-Wen Tsai, Nie-Chuan Chen, and Pen-Hsiu Chang (7) T. Lei, M. Fanciulli, R. J. Molnar, and T. D. Moustakas, “Epitaxial growth of zinc blende and wurtzitic allied nitride thin films on (001) silicon”, (8) Chunlan Mo, Wenqing Fang, Yong Pu, Hechu Liu and Fengyi Jiang , “Growth and characterization of InGaN blue LED structure on Si(1 1 1) by MOCVD”, (9) History of LEDs and LED Technology – MarkTech optoelectronics (10) Rensselaer Polytechnic Institute

Chapter 2

Main topics on nitride-based LEDs

After 2002, the main research topics for nitride-based LED technologies can be divided into six parts as following (1) Lifetime (2) Efficiency (3) Output power (4) Simulation and modeling (5) Ultraviolet-LED (UV-LED) (6) White LED The related researches’ results and achievements will be discussed in this chapter for the first five parts and the white LED will be covered in chapter 3. 2.1 Lifetime Lifetime definition for LED device or system is the operation time, in hours, for the light output to reach 70% of its initial value. Many factors could make the LED lifetime shorter. The main one is heat at the p-n junction. If heat could not be removed efficiently, the accumulated heat will degrade the performance of the devices. As a commercial product, the lifetime should be longer than 10,000 hours. Narendran and Gu [2.1] compared the lifetime of some commercial available high-power white LEDs in 2005. Two experiments were conducted: (1) What is the relation between heat generated in the p-n junction and the lifetime of LEDs. All the test samples for this experiment are from the same manufacturing batch (2) What is the difference of lifetime performance among different packaging types from different companies Before the results from these two experiments are discussed, the method to test p-n junction temperature needs to be mentioned. In general, people use T-point method to do the testing. The T-point is the location external to the package, easy to do the temperature measurement, and have good relationship to the junction temperature.

Fig2.1 Examples of T-points for two different types of LED (Left: Barracuda package; right: Hemispherical encapsulant package

) The results for the first experiment are shown in Fig2.2 & Fig2.3.

Fig 2.2 Light output as a function of time vs. different ambient temperature

Fig 2.3 Lifetime as a function of T-point temperature From the testing results, it shows clearly that the higher the ambient temperature, the shorter the LED lifetime. In addition, the correctness of T-point chosen is also verified. The results for the first experiment are shown in Fig2.4.

Fig 2.4 Different samples vs. Time Samples A to F represent different company’s products. In order to know the influence from packaging, the samples use the same type of LED chip. The output power specification and driving current for rated power (=350mA) is the same for all samples. Surprisingly, the results are much different from each other. This result told us that thermal management is an important issue for LED lifetime. 2.2 Efficiency Efficiency is a very important parameter for commercial products. Higher efficiency means lower cost. Basically, from the chip side, the efficiency can be increased by increasing the extraction number of photons that are trapped inside the LED chip. From the output light coupling side, a better method is needed for light collecting from the LED chip with loss power loss. Currently, many researches are on the former part. How could we increase the extraction number of photons inside the chip? There are four methods mentioned in the IEEE journal. (1) Surface-roughening (2) GaN growth on a patterned sapphire substrate (PSS)[2.2][2.3] (3) Integration of 2-D photonic crystal patterns [2.11] [2.12] (4) Forming V-shaped pits on surface that originate from low-temperature growth (LTG) conditions of topmost p-GaN contact layer (EQE~ 30%) [2.4] Conceptually, increase in the scattering and diffraction of the generated photons will increase the efficiency. These methods are trying to maximize this effect. One structure for 2nd method is shown in Fig 2.5 [2.2]. The performance improvement is shown in Fig2.6. The light emission intensity is improved by 63% and the output power and external quantum efficiency under 20mA operating current are 10.2mW and 14.1%, respectively.

Fig 2.5 Patterned sapphire substrate (PSS) on C-plane

Fig 2.6 Output power vs. current Another example using 2nd method is shown in Fig2.7. [2.3]

Fig 2.7 PSS on {1102}R-plane This group uses wet-etch method to create pattern on the {1102}R-plane while the previous example creates pattern on C-plane. The way they fabricated the PSS structure is described as follows: The SiO2 film with hole-patterns of 3µm diameter and 3µm spacing was deposited onto the sapphire substrate by PECVD method and defined by standard photolithography to serve as a wet-etching mask. The sapphire substrate was then etched using an H 3PO4-based solution at an etching temperature at 300oC. The sapphire etching rate is about 1µm/min. Based on this method, the 9mW output power and 16.4% external quantum efficiency under 20mA operating current are obtained, respectively. The 3rd method for improving efficiency is integration of 2-D photonic crystal patterns as shown in Fig 2.8 [2.12].

Fig 2.8 The schematic side-view diagram of experimented photonic crystal slab structures (a) InGaAsP–InP slab with a SiO (n = 1:46) bottom cladding (b) Free-standing InGaAsP slab (c) Top-view scanning electron micrograph of a fabricated free-standing slab. Lattice constant and air-hole radius are 1000 and 300 nm, respectively, in this case. The pattern size is about 20 μm [2.12] As seen in Fig 2.9, relative Photoluminescence (PL) is enhanced with a function of lattice constant.

Fig 2.9 Photoluminescence (PL) enhancement in the oxide-supported photonic crystal slab (a) PL spectra of two patterned samples and an as-grown sample. The y-axis scale is normalized by the peak intensity of the PL spectrum of the as-grown sample. (b) The relative PL enhancement normalized by the area of the active material is plotted as a function of lattice constant. The airhole radius is 0.3 times the lattice constant [2.12] The paper also claims that the free-standing photonic crystal slabs are expected to show the highest extraction efficiency compare to the other slab structures.

The 4th method is about creating some V-shaped pits on p-type GaN contact. Structure related to the 4th method is shown in Fig 2.10. [2.4]. This group claims they achieve 16mW output power and 30% external quantum efficiency under 20mA operating current for 465nm output wavelength.

Fig 2.10 Topview SEM image of LED 2.3 Output Power High efficiency means high output power. When driving LED to higher current, in most cases, the efficiency becomes lower. Using multi-chip method to increase the LED output power is a common way for high brightness LED product. But the fundamental way to increase the LED output power is from chip design. However, the high power package is always an issue no matter which method is used. At high power level, thermal management is always a big challenge. Currently, many research teams focus on new substrate and electrode materials in order to make the LEDs chip suitable for high power operation. For the new substrate material, Si and SiC are potential ones due to their better thermal conductivity. The disadvantage of SiC is about price. Not only about good thermal conductivity, the use of Si as a substrate offers many advantages, such as low cost, large-sized wafers, and the possibility of the integration of Si electronics on the same chip. Like growing GaN on Sapphire, in order to grow GaN on Si, some problems need to be overcome, such as the large lattice constant mismatch (16.9%) and large thermal expansion coefficients mismatch (57%). Dadgar et al. successfully grew InGaN LEDs on Si with a low-temperature thick AlN buffer layer (~ 750oC, 8-30nm) and an AlGaN-GaN strained-layer superlattice in 2001 [2.5]. The output power improvement for Si substrate is shown in Fig 2.11 [2.6].

Fig 2.11 Output power vs. Injected current with Si/Sapphire substrate This group grew large area green LED on Si and sapphire and conducted the testing. From this figure, LEDs on Si will have higher output power under very high injection current. (> 800mA) The reason for the output power from LED on Si is lower than that on sapphire below 120A/cm 2 is due to the absorption of the emitted light by Si substrate. However, because of high thermal conductivity (1.5W/cm-K), the LED output power for Si substrate case will not saturate under high injection current. For the electrode contact, the poor conductivity of p-GaN is the limit for LEDs performance due to current crowding effect. This problem can be solved by using a thin Ni-Au layer or a highly transparent (> 80%) indium-tin-oxide (ITO) layer as a current spreading layer [2.7]. Either of these methods is widely used in some research groups. High brightness LED is a trend for more application, especially for room bulb replacement. In general, very high brightness LED lamps use large LED chips, with the chip size of the order of several millimeters on one side. And larger chips are usually less efficient. High power packaging is another hot topic in LED industry right now [2.8]. A typical single-chip LED package for high output power is shown in Fig 2.10. For low power package (electrical power < 1W), molded plated copper lead frame is used for the connection. But for high power package, new heat sink and electrode material should be used in order to reduce the thermal expansion mismatch among them. Ceramic, laminates (Cu/Mo), and specialized alloy (Cu/W) are potential ones.

Fig 2.12 Package of typical high power LED In general, high power LEDs incorporating single high current LEDs or multiple standard current LEDs on highly conductive substrate. The substrate should provide lower resistance thermal paths from the LED die to the case or board. For examples, in current available devices, Lumileds Luxeon Emitters and Osram Golden Dragon [Fig 2.13] utilize a built-in heatsink which creates a link from the LED die to the PCB. Lamina Ceramics uses a metal clad ceramic substrate to directly mount the LED die to the heatsink.

Fig 2.13 Osram Golden Dragon Another example from Luxeon Power LED is shown in Fig 2.14. Thermal heatsink extracts heat from LED chip. These packaging methods all allow power level more than 5W per package.

Fig 2.14 Package of Luxeon K2 Power LED [2.13] 2.4 Simulation As is known, the LED fabrication process cost a lot of money. Therefore, correct models and suitable simulation tools can help the LED designer to verify and optimize their design before doing the fabrication. Right now, the commercial available simulation tools are fewer in LED field though there are lots of researches going on this field. There is a simulation tool called SimuLED package available on the market. The package include 1-D LED simulation tool called Simulator of Light Emitters based on Nitride Semiconductors (SiLENSe). This tool allows you to compute the gain spectrum and simulations for DH, SQW, and MQW structures. After putting all the design parameters of the target LED structure, I-V curve, emission spectrum, internal light emission efficiency, temperature effect, and many important device performances can be obtained by this simulation tool. The other tools called SpecLED, and RATRO provides 3-D simulation for current spreading, heat transfer, ray-tracing analyzer of light propagation in LED dice.

Fig 2.15 Applicability of SimuLED package for various LEDs [2.14]

Fig 2.16 Simulation approach and package structure [2.14] A simulation package called SimuLED and its flows are shown in Fig 2.15 and Fig 2.16. Of course, there are some assumptions for LED modeling in this software, the detailed information can be found in the following website ( Synopsys TCAD Sentaurus Device Optoelectronics flow also supports the simulation of lightemitting devices with advanced band structure and gain calculations. It includes a comprehensive set of models for carrier and heat transport, quantization effects, and heterostructures. The detailed information can also be found in their website ( One thing needs to be mentioned about nitride-based LED or Laser diode (LD) simulation. Most software can only be used to simulate standard structure under general operation because many effects are not included in the current simulation model, such as “many-body effects on InGaN AlInGaN QWs”[2.9], “boundary effects” on the optical properties of InGaN MQWs and “surface band-bending effects” on the optical properties of InGaN MQWs. But this situation would be improved in the near future because of growing market size of LEDs. 2.5 Ultraviolet-LED (UV-LED) The AlGaN is suitable material for UV-LEDs application because of its ultra-large bandgap property. The emission wavelength for AlGaN compound can be down to 250nm. Actually, the researches about UV-LED has been conducted for many years because of their promise for numerous applications, such as biological-agent sensing, air and water purification, and biomedical diagnosis. Now, after the first commercial available white LED announced in 2001, more and more researches are on this field because that UV-LED can be used to excite UV phosphor to generate white light (see Fig1.1) and this technology can have better conversion efficiency and color rendering index than current methods.

Compared to InGaN-based LEDs, AlGaN LEDs are subject to more severe self-heating due to higher operation voltage and lower radiative efficiency. The poor electrical conductivity of AlGaN materials was found to exacerbate the current crowding and localized heating problem. [2.10] In addition, the temperature dependence of AlGaN-based UV-LEDs is another serious issue [2.11]. Researchers at Sandia National Laboratory made major breakthrough in a short-wavelength and high output power UV-LEDs development. They announced two high power UV-LEDs. One emits at a wavelength of 290 nm with output power of 1.3mW and the other emits at a wavelength of 275 nm with output power of 0.4mW [2.15]. There are still lots of research works to do for UV-LED field. Information of UV-LEDs manufacturers and their comparison can be found in the website ( 2.6 White LED The researches about white LED are booming at this time because of large market potentials, especially for room light replacement. The concept of generating white light based on LED technologies is shown in Fig1.1. And the related technologies are discussed in detailed in chapter 3. 2.7 Other researches related to current LED industry In current LED industry, only few big companies can survive due to large cost associated with design and fabrication LED chips, plus the patent issues which will be discussed in more detail in Chapter 5. Many key patents are hold by Nichia and Osram. It’s hard to bypass their patents to fabricate an attractive LED product. Following are the challenges; Without buffer layer design, how do you fabricate a high efficiency LED? Without thermal annealing method, how do you fabricate a high quality p-type GaN film? ( Without phosphor powder, what kind of material can be used to achieve the same performance? ( ( All these methods are protected by patents. That’s why there are still many research teams working on other topics, such as packaging, light collecting, efficient LED driving circuit design and so on. Actually, many new companies started and are working on those challenging issues to hopefully optimize the device and system performance. Take a new startup company opened in 2006 for example. (Company name: YLX) [2.12] This company uses one simple figure (see Fig 2.17) to show their potential in the current market.

Fig 2.17 The company does not fabricate LED chips due to the cost. They buy LED chips from Nichia and do the back-end processing. They have three patented technology which are shown in Fig 2.18

. Fig 2.18 (a) YLX combiner (b) YLX LED brightness enhancement technique based on angleselective filter (c) YLX white light generation chip With the combiner technique, they can mix many light outputs with different wavelength, such as Red, Green, and Blue, from fiber. This is another way to generate high power white light. About the angle-selective filter, it is used to raise the coupling efficiency from the LED chip to transmission media, such as optical fiber. Obviously, the smaller the light output angle from the chip, the higher the coupling efficiency. The five common methods to couple the light from LED chip to optical fiber are shown in Fig 2.19.

Fig 2.19 (a) general diagram (b) Burrus SLED (c) microlens coupling (d) macrolens coupling (e) rounded-end and taper-ended fibers As for the white light generation chip, they use different materials to generate white light.

(1) Ja-Hao Chen, Shyh-Chyi Wong, and Yeong-Her Wang, (2005) “Life of LED- based white light source”, Electron Devices, IEEE Transactions on Vol. 48, Issue 7, pp. 1400-1405 (2) Song Zhao, Shaoping Tang, and Nandakumar, M, (2005) “Enhanced output power of near ultra-violet InGaN-GaN LED grown on patterned sapphire substrate”, Simulation of Semiconductor Processes and Devices, 2002. SISPAD 2002. International Conference on 4-6, pp. 43 - 46 (3) Hyun Sang Hwang, (2006) “Enhancing the output power of GaN-based LEDs grown on wetetched patterned sapphire substrate”, US Patent No. 5684317 Shrinivasan Chakravarthi, (2006) “High efficiency and improved ESD characteristics of GaNbased LEDs with naturally textured surface grown by MOCVD”, US Patent No. 6847089 (4) A. Dadgar, J. Christen, and T. Riemann, (2001) “Bright blue electroluminescence form an InGaN-GaN multiquantum-well diode on Si(111): Impact of an AlGaN/GaN multilayer”, Appl. Phys. Lett., vol. 78, no. 15, pp.2211-2213 (5) T. Egawa, B. Zhang, and H. Ishikawa, (2005) “High performance of InGaN LEDs on (111) silicon substrates grown by MOCVD”, Electron Device Letters, IEEE vol 26, issue 3, pp.169 171 (6) J. K. Sheu, Y. K. Su, and G. C. Chi, (1999) “Indium-tin oxide ohmic contact to highly doped n-GaN”, Solid-State Electron., vol. 43, pp. 2081-2084 Karlicek, R.F., (2005) “High power LED packaging”, Lasers and Electro-Optics, 2005. (CLEO). Conference on Volume 1, pp. 337 – 339 (7) Ahn, D.; Seoung-Hwan Park; Eun-Hyun Park; Tae-Kyung Yoo, (2005) “Non-Markovian gain and luminescence of an InGaN-AlInGaN quantum-well with many-body effects”, Quantum Electronics, IEEE Journal of Volume 41, Issue 10, pp. 1253 - 1259 (8) A. Chitnis, J. Sun, and V. Mandavilli, (2002) “Self-heating effects at high pump currents in deep ultraviolet light-emitting diodes at 324nm”, Appl. Phys. Lett., vol. 81, no 18, pp. 3491-3493 (9) X.A.Cao, S.F. LeBoeuf, and T.E. Stecher, (2006) “Temperature-dependent electroluminescence of AlGaN-based UV LEDs”, Electron Device Letters, IEEE vol 27, issue 5, pp.329 – 331 (10) Li Xu, (2006) ”YLX business summary”

(11) Kenji Orita, Shinichi Takigawa Enhanced light extraction efficiency of GaNbased blue LED using photonic crystal (12) Han-Youl Ryu, Jeong-Ki Hwang, Yong-Jae Lee, and Yong-Hee Lee “Enhancement of Light Extraction From Two-Dimensional Photonic Crystal Slab Structures” , IEEE Journal on selected topics in quantum electronics, vol. 8, vo. 2, March/April 2002, pp.231-237 (13) Philips LUXEON Thermal Capabilities (14) Engineering tool for LED and laser diode design and optimization – SimuLED , Semiconductor Technology Research, April 2007 (15) UV LED

Chapter 3
3.1 Introduction

White Light Technologies

This chapter begins with explaining the theory generated the white light (GWL) based on the chromaticity diagram. Color rendering index is introduced. Two methods to GWL base on LED are described in detail. One method to create white light is to mix multiple color lights such as red, green, and blue LED based to GWL. The other method is to use the phosphor based as the converter combining with the blue LED or UV LED to GWL. Current research phosphors works are summarized.

3.2 Chromaticity Diagram
The X, Y, Z values are weights applied to match a color. Remember, that Y correlates approximately with brightness. With X and Z not correlate, even approximately, with any perceptual attributes. However, important color attributes are related to the relative magnitudes of the X, Y, Z values, called chromaticity coordinates, which are calculated as follows: x = X / (X + Y + Z) y = Y / (X + Y + Z) z = Z / (X + Y + Z). Clearly, x+y+z=1 and if x and y are known, it is easy to calculate z = 1 - x - y. With only two variables, such as x and y, it is possible to construct a two- dimension diagram, or chromaticity diagram [3.17].

Fig 3.1 Cone of visible colors in XYZ color space [3.17]

This Fig 3.1 shows the cone of visible colors in XYZ color space. The X + Y + Z = 1 plane is shown as a triangle. The projection of the plane onto the (X, Y) plane forms the chromaticity diagram. Fig 3.2 shows the CIE 1931 chromaticity diagram for a 2-degree field "Standard Observer" [3.16].

Fig 3.2 CIE 1931 chromaticity diagram [3.16]

3.3.1 Combination of Red, Green and Blue Can GWL
The human eyes can sense the wavelength of visible light from 400 nm to 700 nm. In the chromaticity diagram, if the point at the wavelength of 700 nm (red color), the point at the wavelength of 555 nm (green color), and the point at the wavelength of 460 nm (blue color) are connected, it would form a triangle called the white region as show on Fig 3.3.

It is CIE 1931 (x, y) chromaticity diagram. Monochromatic colors are located on the perimeter. Color saturation decreases towards the center of the diagram. Also shown are the regions of distinct colors. The equal energy point is located at the center and has the coordinates (x, y) = (1/34, 1/3).

Fig 3.3 CIE 1931 (x, y) Chromaticity diagram [3.2]

The chromaticity diagram shown in Fig 3.4 is the Planckian locus. It relates to the color temperature of the object. As temperature increases, the color glows from red, yellow, and to white. For incandescent source, it has a temperature of 2856 K with (x, y) = (0.4476, 0.4074). For the daylight source, it has a temperature of 6500 K with (x, y) = (0.3128, 0.3292) [3.2].

Fig 3.4 the Planckian locus is in the chromaticity diagram [3.2]

3.3.2 Chromaticity Difference Result from Reference and Test Light Source
Chromaticity difference resulting from the illumination of an object with a reference and test light source as show figure 3.5. In u’, v’ uniform chromaticity diagram, the color difference is directly proportional to the geometric distance. The reference light source is located on the planckian locus at the correlated color temperature of the test light source.

Figure 3.5 Chromaticity difference result from reference and test light source [3.2]

3.4 Color Rendering Index (CRI)
CRI is a unit of measure that defines how well colors are rendered by different illumination conditions in comparison to a standard (i.e. a thermal radiator or daylight). CRI is calculated on a scale from 1-100 where a CRI of 100 would represent that all color samples illuminated by a light source in question, would appear to have the same color as those same samples illuminated by a reference source. To put it another way, low CRI causes colors to appear washed out and perhaps even take on a different hue, and high CRI makes all colors look natural and vibrant. Fig 3.6 shows CRI of some light sources. The fluorescent light has a CRI in between 60 ~ 85. The phosphor-based white LED has a CRI in between 60 ~ 90. The green monochromatic light has a CRI of less than 50. A CRI greater than 85 is most suitable for many applications [3.2].

Fig 3.6 CRI of some light sources [3.2] Figure 3.7 show vase illuminated with (a) high-CRI source and (b) low-CRI source ( picture from Auguste Renoir, French impressionist, 1841-1919)

Figure 3.7 (a) high-CRI source and (b) low-CRI source [3.3] Power reflectivities of CIE test color samples used to calculate the general color rendering index as shown on figure 3.8.

Figure 3.8 Show power reflectivities of CIE test color samples used to calculate CRI [3.3] Power reflectivities of CIE test color samples used to calculate the special color rendering indices as shown on figure 3.9.

Figure 3.9 Power reflectivities of CIE test color samples used to calculate the special CRI [3.3]

3.5 Color Temperature
Color Temperature describes certain color characteristics of light sources. A "blackbody" is a theoretical object which is a perfect radiator of visible light. As the actual temperature of this

blackbody is raised, it radiates energy in the visible range, first red, changing to orange, white, and finally bluish white. Color temperature describes the color of a light source by comparing it to the color of a blackbody radiator at a given temperature. For example, the color appearance of a halogen lamp is similar to a blackbody radiator heated to about 3000 degrees Kelvin. Therefore it is said that the halogen lamp has a color temperature of 3000 degrees K- which is considered to be a warm color temperature. The hotter the blackbody, the cooler the color temperature! (Note: The Kelvin temperature scale uses the same size degree as the centigrade scale, but its zero point is at absolute zero, or -273 degrees C). As Fig 3.10 sunlight can be "warm" or "cool" depending on the time of day and the ambient conditions. Though color temperature is not a measure of the physical temperature of the light source, it does correspond to the physical temperature of the blackbody radiator when the color appearance is the same as the source being tested [3.23].

Warm (2000-3000K)

Mid-range (3000-4000K)

Cool (4000K +)

Fig 3.10 show color temperature [3.23]

3.6 Color Comparisons
Using CRI in conjunction with Color Temperature is the best way to draw comparisons between different light sources as shown in Fig 3.11 [3.15].

Fig 3.11 Color Comparisons [3.15]

3.7 Two Approaches to Generate White LED Based on LED Technology
Fig 3.12 shows three types of white light LED technology. These types are based on two categories: the first type uses the combination of three discrete RGB LEDs method [3.2]. The second type and the third type employees the phosphor coating on the LED method.

Fig 3.12 Three Types of White Light LED [3.2]

3.7.1 Multi-color (RGB) LED Based
Light emitting diodes, as a new generation of solid-state light source, have received great development since they were invented in the 1960s. With the external quantum efficiency of III–V nitride, LEDs continuously improved, the emission wavelength region was widened and covered the area from the ultraviolet region to the infrared region. Thus, white LEDs can be created for the first time by combining blue, green, and red LEDs, bringing white LEDs toward general lighting, as shown in Fig 13 [3.13].

Fig 3.13 (a) Schematic of additive color mixing of three primary colors (b) Additive color mixing using LEDs [3.13]

3.7.2 Phosphor Based
As discussed earlier, multi color based white LEDs experience different light output degradation rates and will produce unstable white light over time. In order to provide a stable white light, a blue LED pre-coated with a yellow phosphor (YGd ) (AlGa ) O : Ce (YAG: Ce) was invented The yellow phosphor YAG: Ce emits yellow phosphorescence when excited by blue light; consequently, the mixing of the yellow light and the blue light can be perceived by human eyes as white light shown in Fig 3.14. It is most common and widely used technique to create white light [3.25].

Fig 3.14 Blue LED + Yellow Phosphor based white LED [3.25] Fig 3.15 shows the phosphor based white LED spectrum having two emission bands due to blue light and yellow phosphorescence.

Fig 3.15 YAG: Ce Phosphor-based White LED Spectrum [3.2] This blue/yellow white light LED, with characteristics of compact size, high efficiency, long lifetime, low power requirement, and energy savings, has been widely used in various applications such as liquid crystal display back lighting, full-color displays, cell phones, and traffic signals. However, because the yellow light emitting from the phosphor YAG: Ce lacks sufficient red emission, this white light has a low color rendering index For improving , redenhanced YAG: Ce and a red phosphor were used in this blue/yellow white light LED [3.25]. Figure 3.16 shows the arrangement of the phosphor. Figure 3.16a illustrates the first generation white phosphor based LED, which has the phosphor distribution centered on top of the die. Figure 3.16b is the second-generation phosphor deposition process developed by Lumileds, which has the phosphor uniformly distributed around the die. The conformal phosphor coating improves the CRI and also gives better color distribution[5].

Figure 3.16 (a)Non-conformal coating vs. (b)conformal coating [3.5]

3.7.3 White sources Using Phosphors Excited by UV or Blue LEDs
A new interesting item of research in white light is three-band white LED. A blue LED is precoated with a green phosphor and a red phosphor, and the phosphors emit the green and red emissions when excited by the blue light from the blue LED, respectively. Not all the blue light is absorbed by the phosphors; the remnant blue light is mixed together with the green and red emissions as a three-band white light. This white light from the combination of the blue, green, and red emissions has an obvious advantage. It has higher color rendering index than blue/yellow white light, i.e., colors can be reproduced more vividly. Therefore, it can be more suitably used in museums, galleries, and the medical field [3.26] To solve the high color temperature and low CRI problem of the blue LED combined with the yellow phosphor technique, researchers looks for the alternate ways to create white light such as employing the UV LED technique. The LED industry also looks into this UV LED technique to prevent from the violation of intellectual patent against Nichia and Osram. UV LED is very similar to the blue LED combined with the yellow phosphor technique. The UV light is absorbed by the RGB phosphors and output the white light. Figure 3.17 shows the white source using phosphors that are optically excited by UV or blue LEDs. The challenge the UV LED researchers facing is on how to improve the quantum efficiency of the UV LED chip due to the low doping efficiency in AlGaN with a high Al content [3.13].

Fig 3.17 White source using phosphors that are optically excited by UV or blue LEDs [3.13]

3.8 Current Researches in Phosphors
As already discussed so far, blue LED combines with the phosphor technique is the most common method to generate the white light. However, using such method may violate Nichia’s intellectual patent and get sued. Alternate ways to create white light have been investing underway. One of the research areas is on the phosphor material. Newly developed oxynitride and nitride phosphors for this application are because they have suitable excitation and emission wavelengths and stable optical properties in a high temperature environment. High brightness warm-white LED lamps have been realized using a yellowish-orange α-SiAlON oxynitride phosphor. High color-rendering index white LED lamps have been also realized using three color oxynitride/nitride phosphors. Other research is consideration that phosphor-free white-light light-emitting diode of weakly carrier-density-dependent spectrum with pre-strained growth of InGaN/GaN quantum wells. It uses the color mixing for white light is implemented by adding a blue-emitting QW at the top of the yellow-emitting QWs. The blue shifts of the blue and yellow spectral peaks of the generated electroluminescence spectra are only 1.67 and 8 nm, respectively, when the injection current increases from 10 to 70mA. Such small blue shifts imply that the piezoelectric fields in their QWs are significantly weaker than those previously reported [3.27].

3.8.1 Some Facts about the Lighting Industry Today and Its Potential for Change in Future
1• Today, the domestic market for lamps (light bulbs), ballasts, lighting fixtures, and lighting controls is about $12 billion. Globally, this market is about $40 billion. 2• Today, about 22% of our nation’s electrical energy is used for lighting. • Today, the total electrical energy used for lighting equals the output of about 100 large power plants*. (More than 3X this amount is needed to produce the electricity.) The cost of this electricity is about $55 billion (2003). 3• Today, incandescent lamps, by far the least efficient of the common lamp types, consume electrical energy equal to the output of more than 40 large power plants. Solid state lighting (SSL) systems, using white LEDs (light emitting diodes) or white OLEDs (organic light emitting diodes), are new technologies that represent a huge break with current lighting paradigms. These technologies will result in the creation of new lighting fixtures, controls, and power supplies, in addition to new application and installation practices and codes. While there are many technical unknowns today, the results of the past few years suggest that there are no fundamental reasons why these light sources can’t achieve efficiencies 10-12X that of incandescent lamps, and 2X that of fluorescent lamps. The energy saving prospect is huge:

• When SSL reaches these efficiencies, the U.S. will save annually the output of about 30 large power plants, or about 6-7% of our country’s total electrical energy usage. This will result in a savings of $17 billion in annual electrical costs (at 2003 rates).** • Accompanying environmental benefits are large, and include a reduction in carbon dioxide emissions of 155 million tons, and about a million tons in combined nitrous oxides and sulfur dioxide. ** • The main attractive features of SSL are: o High efficiencies theoretically possible (both LEDs and OLEDs) o Much greater design freedom (color flexibility – including many “whites” – without the use of filters, fully dimmable, new form factors for fixture designs) o Rugged, long lifetimes, no catastrophic failures (LEDs) o Intrinsically diffuse (OLED) or “point” (LED) light sources [3.27]

Furthermore, current researcher in phosphors, there are two journals on this research field published in 2005 are listed: 1.“Organic/inorganic optical nano-composite with highly-doped rare-earth nanoclusters: Novel phosphors for white LEDs,” by H Mataki & T. Fukui. 2. “Direct White Light Phosphor: A Porous Zinc Gallophosphate with Tunable Yellow-to-White Luminescence,” by Liao, Y. C.; Lin, C. H.; Wang, S. L.

3.9 LED Basics & Ongoing Research & Applications 3.9.1 LED Basics: Efficiency
Some fundamental optical properties related to efficiency. There are three types of efficiency which are commonly used in LED device & its applications: internal efficiency, extraction efficiency and power efficiency [3.13]. • Internal Efficiency

Extraction Efficiency


Pint /( hv) I /e

External Efficiency


P /( hv) I /e

Power Efficiency

3.9.2 The Light Escape Cone

Fig 3.18 The Light Escape Cone [3.13] Total internal reflection occurs inside LED chip. Light escape cone defined by critical angle for total internal reflection. Fig 3.18 (a) Definition of the escape cone by the critical angle Фc (b) Show area element dA (c) Area of calotte-shaped section of the sphere defined by radius r and angle Фc.

Fig 3.19 Show derive the Lambertian emission pattern [3.13] Fig 3.19 shows geometrical model used to derive the Lambertian emission pattern. (a) The light emitted into angle dФ inside the semiiconductor is emitted into the angle dФ in air. (b) Illustration of the area element dA of the calotte-shaped section of the sphere.

3.9.3 Light Escape in Planar LEDs

Pescape Psource

≅ 1 / 2 1 − (1 − φC / 2) = 1 / 4φC




Where Фc is critical angle of total internal reflection.But it has only small fraction of light can escape from semiconductor. Follow equation gives <10% extraction efficiency for typical III-V semiconductors

Pescape Psource

1 nair = 4 ns 2


3.9.4 High Internal Efficiency LED Designs
Confinement of carriers in active region of double heterostructure (DH). High carrier concentration in active region. Fig 3.20 illustration of a double heterostructure consisting of a bulk or quantum well active region and two confinement layers. The confinement layters are frequently called cladding layers [3.13].

Fig 3.20 Illustration of a double heterostructure consisting of a bulk or quantum well [3.13]

3.9.5 High Carrier Concentration in Active Region Double Hetero-structure
Free carrier distribution in figure 3.21(a) a homo-junction and (b) a hetero-junction under forward bias conditions. In homo-junctions, carriers are distributed over the diffusion length. In hetero-junctions, carriers are confined to the well region. Compare (a) and (b), we can get high carrier concentration in active region of DH

Figure 3.21 Show free carrier distribution (a) a homo-junction [3.3]

Figure 3.22 (b) a hetero-junction under forward bias conditions[3.3]

3.9.6 High Extraction Efficency Stucture
Fig 3.23 show absorption coefficient of a semiconductor with bandgap Eg versus energy. The “Urbach tail” dominates absorption near but below the bandgap.Absorption further below the bandgap is dominated by free-carrier absorption.

Fig 3.23 High extraction efficiency structure [3.13]

Since double hetero-structures are optically transparent. All efficient LEDs designs uses a double hetero-structures. Figure3,24 show double hetero-structure with optically transparent confinement regions. Re-absorption in the active region is unlikely due to the high carrier concentration in the active region and resulting shift of absorption edge.

Figure 3.24 double hetero-structures [3.3]

Efficiency versus Active Layer Thickness

Fig 3.25 Efficiency vs. active layer thickness [3.13]

Figure 3.26 show dependence of luminous efficiency of an ALGaInP double heterosturcture LED emitting at 565 nm on the active layer doping concentration Al

Figure 3.26 Luminous efficiency vs. carrier concentration in active layer [3.4]

Light Extraction Efficiency
Thinning down the active layer reduces considerably re-absorption losses in the active layer, especially in material with low internal quantum efficiency (see Fig. 3.27). This can also shift the operating point of the device towards the high-level injection regime.

Figure 3.27 Light extraction efficiency vs. internal quantum efficiency [3.26]

Quality of the active layer material determines whether the preferred device design should be thick or thin. (See Fig. 3.28). For a high internal quantum efficiency device (>90%), one should minimize bulk absorption by making the device as thin as possible. (This requires that a light randomization mechanism such as nano-textu.ring be incorporated in the device, or that photon recycling be used for additional randomization of light.). On the other hand, if the active layer has a low (<90%) internal efficiency, it's better to make a thick substrate device which allows the photons to see the device edges where 4 additional escape cones are present. This increases the photon to escape probability from the semiconductor on the very first surface bounce.

Figure 3.28 Light extraction efficiency vs. height of the chip for LED with epoxy lens [3.36] The photon gas model is a good approximation when the average photon path consists of many bounces inside the LED. When low quality contacts are introduced on the top and bottom surfaces of the LED, a comparison of the photon gas model with the Monte-Carlo simulation shows a breakdown the former (Fig 3.29).

Figure 3.29 Scaling properties of light extraction efficiency (air interface) [3.36] Nonetheless the qualitative scaling properties of the LED's light extraction efficiency is calculated correctly. From the point of view of light extraction efficiency, the smaller device area is preferable, since light randomization happens only on the edges of the device. Even though contacts cover only about 15% of the area of the device under consideration, the dependence of light extraction efficiency on the contact reflectivity is strong (Fig 3.30). Indeed, for a 50% reflective sheet, contact, every probability of contact absorption causes 7% loss, which is comparable with 6x(lIn2)- 12%.

Figure 3.30 LED with one bottom contact [3.36]

We considered the optimal position of the active layer for the device with sheet contact on the bottom and small circular contact on the top surface. Figure 3.31. That bringing the active region closer to the smaller contacts results in up to 6% improvement in the extraction efficiency. Of course, we can not make the active layer too close to the top surface, since we use that upper layer for current spreading. Light output is higher near the edges of the surfaces adjacent to the roughened sides of the device. This is due to the fact that the texturing at the edges scatters some of the light directly into the escape cone. Obviously, better performance

Figure 3.31 Show top and side views of light escaping from an LED. The white areas have more light emission than the dark ones. [3.36]

Measures of Internal Quantum Efficiency
By investigating the quality of InGaA1P wafers with the following structure: n-GaAs substrate(2OOim) the top edge n-AHnP Lower confming layer (1,nn) undoped A1GaInP Active layer (— Side surface p-A1InP Upper confming layer (O.5tm) p-GaP Window Layer ( The samples with an absorbing substrate was optically pumped using the 568nm line of a continuous wave ArKr laser. The absolute external quantum efficiency is measured by referencing the measured photoluminescence (PL) from the sample the measured from laser off of the perfect white Lambertian reflector. Corrections for transmission through the optical setup and the relative detector the side edge efficiency at different wavelengths were made in the

An optical model is used to obtain the internal quantum efficiency of the quantum well (QW) from

the measured external quantum efficiency. We used an optical model similar to that used in the modeling of the light extraction efficiency which takes into account Fresnel transmission, absorption at the pump wavelength in cladding layers and in the active region and photon recycling. Measurements were performed on samples with active layers emitting at wavelengths 589, 604, 61 5 and 623 nanometers, yielding internal quantum efficiencies of 15, 23, 45 and 65 percent respectively, after correction for re-absorption.

3.9.7 Reasons: Extraction Inefficiency
The photons cannot reach the free space or may not be able to escape to be extracted from semiconductor material. That means photons emitted from active region may not make into free space, may not be able to escape to be extracted from semiconductor material so the light cannot get itself into free space. The extraction efficiency is the result of many internal reflections. Depending on LED device and final packing structures, its extraction efficiency can vary between 5% to 50 %. It is very difficult for the light generated within a high index material to propagate into the surrounding medium containing air which leads to total internal reflection resulting in light extraction inefficiency. In Fig 3.32, the red dashed line indicates that the incidence angle is large so, we will get reflection at every incidence at the interface; this is called the trapped light. The most undesirable effect is that the photons revolving around the crystal cannot escape. If these photons go into free space then they would be useful, but this is the main problem in LEDs, so we have to design a semiconductor to get better light extraction efficiency.

Fig 3.32 Illustration of “trapped light” that cannot escape from a cube-shaped semiconductor for emission angles larger than αc due to total internal reflection [3.12] Performance of light-emitting diodes is defined to the great extent by two figures of merit, namely internal quantum efficiency of the active region and light extraction efficiency. While the former quantity reflects the quality of an epitaxially grown structure and normally lies in the range 20-90%, the latter strongly depends on particular design and can be as low as 2% [3.12].

3.9.8 Increase Efficiency Method: By Design Current-Spreading Layer
Since light is generated under top contact and top contact shadows light, it made current spreading layer spreads current to edges of LED die. Figure 3.33show effect of the current-spreading layer o LED output. (a) Top view without a current-spreading layer. Emission occurs only near the perimeter of the contact. ( b) Top view with a current-spreading layer

Fig 3.33 (a) Top view without a current-spreading layer (b) Top view with a current-spreading layer.[3.15]

Theory of Current Spreading
Figure3.34 show schematic illustration of current spreading in structures with different top contact geometries. ( a) Linear stripe contact geometry. (b) Circular contact geometry.

Figure3.34 (a) Linear stripe contact geometry. (b) Circular contact geometry.[3.15]

Figure3.35 show current-spreading structures in high-brightness AlGaInP LEDs. Illustration of the effect of a current-spreading layer for LEDs (a) without and (b) with a spreading layer on the light extraction efficiency. (c) GaP current-spreading structure (d) AlGaAs current-spreading structure.

Fig.3.35 (a) without and (b) with a spreading layer on the light extraction efficiency. (c) GaP current-spreading structure (d) AlGaAs current-spreading structure.[3.15] Figure3.36 show the effect of GaP window thickness on current spreading is illustrated by surface light emission intensity profiles for three different AlGaInP LED chips with window layer thickness of 2, 5, and 15 um. The profile is indicated by the dashed line in the inset. The dip in middles of the profiles is due to the opaque ohmic contact pad.

Figure3.36 show the effect of GaP window thickness on current spreading [3.15] Figure 3.37show chip external quantum efficiency and luminous efficiency versus for ward current for AlGaInP LEDs with GaP window layer thicknesses of 2, 5, and 15 um. Solid curves are under DC conditions. Dashed curve is under pulsed condition using 400 ns pulses and a 3% duty cycle. Heating is essentially eliminated in this case.

Fig 3.37 chip external quantum efficiency &luminous efficiency versus for ward current [3.16]

3.9.9 Current Optimizing Light Extraction Efficiency of LED: FDTD Solution
The layout editor shows all of objects contained in the simulation project. Objects can be moved and resized with simple mouse movements. The windows show the top, side, end and perspective views to assist with visualization as shown on figure 3.38

Figure 3.38 construct the FDTD solutions model of the LED multi-layer
    

construct the LED layer structure by using the object align functions add photonic crystal-like microstructure by using the cylinder primitive and the array function orange box shows the extent of the computation volume and the PML boundaries yellow planes show transmission monitors use a x-oriented dipole source to model the recombination of an electron-hole pair in the active layer; the blue arrow indicates the polarization direction

3.9.10 Watch Movies and Gain Operation Insight LED Light Extraction
To better understand the origin of the light extraction efficiency improvement, use the built-in movie monitor in FDTD Solutions to capture the field dynamics of the simulation as shown on figure 3.39.

 

Here FTD investigate the extraction gain for a vertically-oriented dipole Note the vertically-scattered radiation present in the LED with the textured layer compared to the un-textured structure

un-patterned LED – click to play movie

patterned LED – click to play movie

Fig. 3. 39 Watch Movies and Gain Operation Insight LED Light Extraction [3.27]

1 2 3 4

3.10 Ways: To Improve Efficiency Structures Encapsulation Packaging Thickness of Active Layer

3.10.1 Structures To shape a die is a good way to get better extraction of light. In the figure below there is LED which is a parallel epithet and 3 of 6 escape cones are shown below. Having a cylindrical shape of LED, the three side escape cones merge to a side escape ring as indicated in the Fig 3.40 and this will increase extraction efficiency.. In the lower part of the figure we have a spherical LED with a point like active region in middle of LED. In a spherical shape, and a cone shaped LED with an active region, in whichever direction the light is emitted there is normal incidence always at semiconductor interface as a result of that light can escape from a semiconductor. A similar principle applies to the structure in Fig 3.41. That is light that is being emitted upward can be reflected initially within the cone. The angle of incidence becomes more and more rectangular, so the light escapes from the semiconductor. These structures are not practical as they assume a light emitting point source. When we have a point source, the light emitted is very small, but the overall power of the device would not be high even though the device has a high extraction and overall efficiency, so it is not efficient. .

Fig 3.40 Different shapes for LEDs a) Rectangular shape b) Cylindrical shape

Fig 3.41 Geometric shapes for LEDs with perfect extraction efficiency  Spherical shaped LED b) Cone shaped LED

The devices with larger active region are practical and those devices are shown here, LED that looks like an inverted pyramid and a pedestal type of LED .These devices are very efficient, as the sharp edges of LED help to extract light out of the device and thus, higher extraction efficiency.

Fig 3.42 Die-shaped Devices: a) Blue GaInN emitter on SiC substrate with trade name “Aton”. b) Schematic ray traces illustrating enhanced light extraction c) Mipes cro-graph of truncated inverted pyramid (TIP) AlGaInP/GaP LED. d) Schematic diagram illustrating enhanced extraction

Fig 3.43 External efficiency vs. forward current for red-emitting (650nm) truncated inverted pyramid (TIP) LEDs and large junction (LJ) LEDs mounted in power-lamp packages. The TIP LED exhibits a 1.4 times improvement in extraction efficiency compared with the LJ device, resulting in a peak external quantum efficiency of 55% at 100mA (after Krames et al., 1999)

Fig 3.44 Different shapes of LED

Fig 3.45 Current-voltage characteristic, forward voltage, series resistance of absorbing-substrate (GaAs) and transparent-substrate (GaP) LEDs with AlGaInP active regions

• Effect of Encapsulation Encapsulation improves the performance of LED. A semiconductor die coated with a encapsulant and the encapsulant is typically Epoxy or PMMA (Polymethyl Methacrylate) or Silicone and the refractive index is reduced between the semiconductor and surrounding air. The epoxy acts as a buffer to change the refractive index from high value in the semiconductor to a intermediate value in the epoxy and gradually, to a low value in the air. This diagram shows the curve to depict the improvement in extraction efficiency by a factor of 2 to 3, by using an encapsulant, which typically has a refractive index of 1.6~1.7, which is significant.

Fig 3.46 a) LED without and b) with dome-shaped epoxy encapsulant. 3.10.2 Packaging a) Conventional Packages  Electrical path  Optical path b) Thermal Packages 1) Electrical path 2) Optical path 3) Thermal path Let’s talk little bit more about packaging. optical path length (OPL) is the product of the geometric length of the path light follows through the system, and the index of reflection the medium through which it propagates. A difference in optical path length between two paths is often called the optical path difference (OPD). Optical path length is important because it determines the phase of the light and govern interference and diffraction of light as it propagates.

The electrical path is define as the mechanism in which the device efficiency is increased within the defined boundaries of operating current/voltage which can increase the throughput of the device. Conventional packages There is need of both Electrical and Optical paths to get the light out which is there in Conventional Packaging and these LEDs are shown below. The encapsulant and the encapsulant dome provide an index contrast that is reduced which increases light extraction, hermeticity, mechanical stability is also provided by encapsulant. In the figure below, there are typical shapes with a reflector cup with semiconductor chip in this cup, bonded to bond wire and encapsulated. At times, with rectangular and cylindrical are required too.

Fig 3.47 Typical LED shapes a) LED with hemispherical encapsulant b) LEDs with cylindrical and rectangular encapsulant

Fig 3.48 Thermal resistance of LED packages: (a) 5mm (b) low-profile (c) low-profile with extended lead frame (d) heatsink slug (e) heatsink sluf mounted on printed circuit board (PCB). Trade name for these packages are “Piranha” (b and c, Hewlett Packard Corp.), “Barracuda” (d and e, Lumileds Corp.), and “Dragon” (d and e, Osram Opto Semiconductors Corp.) (atopted from Arik et al., 2002) [3.13] Thermal Packages The need of high power devices have introduced new generation power packages as illustrated in the example below. The thermal path has a heat sink that could be silver, Al, Cu which directs heat away to a large area PCB that will dissipate the heat and thus, LED will not reach a high temperature. The package showed below is called as a Baraccuda package, which is a newer generation package. As the LED heat escalates, several key characteristics may become apparent, which deminstrate the importance of LED thermal mangament. The forward voltage will begin to decrease. The decreasing voltage can impose an increased load on related LED driver components causing their temperature to increase as well. In resistor driven circuits, the forward current will increase. As the LED lights temperature continues to rise, the optical wavelength can shift. The increasing wavelength can cause orange LED lights to appear red or even white LED lights to appear bluish. This color shift typically intensifies with the AlInGaP technologies (red, orange, amber, and yellow). In addition, a thermally stressed LED lights will loose efficiency and light output will diminish. If the LED thermal management continues to race out of control, the LED junction may break down causing a state of complete thermal runaway. The result is typically catastrophic failure. Other affects of overstressed LEDs may include broken wire bonds, delaminating, internal solder joint detachment, damage to die-bond epoxy, and lens yellowing.

Fig 3.49 Thermal package

Fig 3.50 LED current of power across years [3.13] While the HB LED industry remains healthy and is projected to grow to as high as $7 billion in just three years, new applications are developing less quickly and focus is shifting towards the limitations of packaging. Without improved light output, greater efficiencies and better thermal management, LEDs may not reach the LED illumination market as anticipated. As with power transistors, the performance and longevity of high-brightness LEDs depends upon effective thermal management. Accordingly, LED manufacturers are adopting similar package design techniques to those now seen in surface-mounted power devices. However, thermal management between the boundaries of the package and the ambient environment is equally important, and systems integrators are also adopting power electronics techniques to achieve cost-effective solutions.

Thickness of Active Layer Since internal quantum efficiency is the number of photons emitted from active region per second divided by total no. of electrons injected into LED per second. Internal quantum efficiency is highly dependent upon the thickness of active layer. IQE normally lies in the range of 20-90%.

Fig 3.51 Dependence of the internal differential quantum efficiency (emitted photons per injected electron) on temperature for different p-type doping levels in the cladding layer (after Kazarinov and Pinto, 1994) [3.13] Total efficiency is the product of internal efficiency and extraction efficiency. However, light extraction efficiency is self-dependent on internal quantum efficiency due to inevitable reabsorption of some of the light. In thin LEDs reabsorption effect is less severe. Fig. shows extraction efficiency vs LED chip height. For high IQE, LED should be thin film, but for lower IQE a thick field is better because light escapes more readily from edges.

Fig 3.52 Light extraction efficiency vs. internal quantum efficiency [3.13]

3.10.3 Ongoing Research Areas Current Research efforts are focusing at applications of LEDs and basic research involving the enhancements of device structures, efficiency, packaging, material to gain maximum benefits for its technology. Total efficiency is the product of internal efficiency and extraction efficiency. However, light extraction efficiency is self-dependent on internal quantum efficiency due to inevitable reabsorption of some of the light. In thin LEDs reabsorption effect is less severe. Fig. shows extraction efficiency vs. LED chip height. For high IQE, LED should be thin film, but for lower IQE a thick field is better because light escapes more readily from edges. The company, headquartered in Carlsbad, California, produces high-brightness LED light sources and is best known for its light recycling technology. In this approach, multiple LEDs are combined in a “light-recycling cavity” to enhance their individual brightness output. The patent relates to color conversion using a solid, ceramic-based element, rather than the more conventional approach using phosphor powder. It encompasses ceramic processing techniques such as tape casting and sintering to form thin luminescent sheets for volume production. University of Texas at Dallas (UTD) nanotechnologists and an Australian colleague have produced transparent carbon nanotube sheets that are stronger than the same-weight steel sheets and have demonstrated applicability for organic light-emitting displays, low-noise electronic sensors, artificial muscles, conducting appliqués and broad-band polarized light sources that can be switched in one ten-thousandths of a second. Cree and Lawrence Berkeley also focused on demonstrating a compact LED with output in the

1000+ lumens range (1000 lumens is the approximate light output of a 60 watt incandescent bulb). To reach the 1000 lumens target, high radiance arrays were developed using multiple chips and custom LED packaging. Thermal management was another key area of research. Electrical energy not converted to light is converted to heat, the effective dissipation of which is challenging. An effective maximum temperature of only ~125 °C (due to available LED encapsulation materials) severely limited the allowable thermal compensation for the system. Osram Sylvania Product Inc. is attempting to develop a high efficient phosphor converting white LED product through increased extraction efficiency of the LED package. A multi-layer thin film coating is applied between LED chip and phosphors to reflect the inward yellow emission, increasing their probability of forward escape. Additionally, a transparent monolithic phosphor may replace the powdered phosphor to reduce the back scattering blue light caused by phosphor powders. LED lighting may be the answer to saving us from such hypothetical events as global warming. LED lighting is extremely efficient and illuminating due to the low power requirements it takes. If all the lights in the United States were changed to LED lighting the amount saved would mean the reduction of production of energy by nearly 30%. That is a huge savings and it also means less pollution would be generated as well. 3.10.4 White/Blue LED Manufacturers Nichia Corporation Nichia has grown in the field of manufacturing and sales of fine chemicals, particularly inorganic luminescent materials (phosphors). In the process of the challenging pursuit of brighter luminescent and light-emitting materials, we succeeded in developing and commercializing the super high brightness Blue LED in 1993. It was accepted with great amazement in the related industries throughout the world. Since the first announcement of the Blue LED in 1993, Nitride-based LEDs in different emission colors ranging from Ultraviolet to Yellow have been contributing to the diversification of LED application fields. Cree Cree, Inc. is a market-leading innovator and manufacturer of semiconductors that enhance the value of LED solid-state lighting, power and communications products by significantly increasing their energy performance. Key to Cree’s market advantage is unrivaled materials expertise in silicon carbide (SiC) with gallium nitride (GaN) to deliver chips and packaged devices that can handle more power in a smaller space while producing less heat than other available technologies, materials and products. Lumileds Philips Lumileds Lighting Company is a fully integrated manufacturer of LED dice, packaged LEDs, and high-brightness LEDs—LUXEON®—designed for integration into

general lighting products. The company markets LED solutions designed specifically for automotive applications, LCD displays, general lighting, portable applications, signage, traffic signals, and other segments. Philips Lumileds makes the world's brightest white, red, amber, blue and green LED light sources. Toyoda Gosei Toyodo-Gosei has accumulated unique galvanizing and thin-film technologies such as vacuum deposition and sputtering in order to meet demand for a decorative finish on interior and exterior products. In 1986 (61st year of Showa Era), TG launched LED development using gallium nitride (GaN) under the direction of Prof. Isamu Akasaki of the Faculty of Engineering, Nagoya University (Currently University Prof. of Nagoya University and Prof. of Meijo University) and Toyota Central R&D Labs., Inc. In the following year, TG was contracted by the Science and Technology Promotion Group to develop manufacturing technology for blue LEDs and achieved great success in its development. OSRAM Optosemiconductor OSRAM is one of the two leading lighting manufacturers in the world. Light from OSRAM shines in about 150 countries. With sales outside Germany accounting for 88 % of total turnover and a sales presence on every continent, they are a true global player. Seoul semiconductor in Korea Seoul Semiconductor is the biggest LED manufacturer in Korea. The Company's products include custom display LED products, displays with controllers, back lights, LED dot matrixes, LED lamps and arrays. During the year ended December 31, 2006, it had an annual production capacity of 2,835 million LED chips and its actual production output was 2,114 million chips. They announces new P4 LED product that emits 240 lumens at 1A and claims the industry’s highest efficacy: up to 100 lm/W@360mA. They have plans projected for 135lm/W by 2007 and 145 lm/W by Q1 2008. Shimei Semiconductor On Nov 16, 2007, Shimei Semiconductor Co. (Kyoto, Japan) has developed a blue LED grown on a silicon wafer that it plans to make available by next April. The company claims that using silicon wafers as a substrate for GaN epitaxy could drastically lower the cost, simplify LED structure, extend the lifetime and enable the integration of an optical device in CMOS circuits. The prototype LED emits blue light of the 450-nm wavelength and an output power of 10mW.

[1] Wikipedia, the free encyclopedia [Online]. Available

[2] Light-Emitting-Diodes-dot-org home page [Online]. Available [3] S. Muthu, F. J. Schuurmans, M. D. Pashley, “Red, green, and blue LED based white light generation: Issues and control,” IEEE 2002. [4] S. Muthu, J. Gaines, “Red, green and blue LED-based white light source: implementation challenges and control design,” IEEE 2003. [5] D. A. Steigerwalk and etc, “Illumination with solid state lighting technology,” IEEE Journal on Selected Topics in Quantum Electronics,” Vol. 8, NO. 2, March/April 2002. [6] J. K. Sheu and etc, “White-light emission from near UV InGaN-GaN LED chip precoated with blue/green/red phosphors,” IEEE Photonics Technology Letters, Vol. 15, NO. 1, January 2003. [7] K. M. Lee and etc, “Emission characteristics of white light phosphor,” IEEE 2003. [8] H. Wu and etc, “Three-band white light from InGaN-based blue LED chip precoated with green/red phosphors,” IEEE Photonics Technology Letter, Vol. 17, NO. 6, June 2005. [9] H. S. Chen and etc, “White light generation with CdSe-ZnS nanocrystals coated on an InGaN-GaN quantum-well blue/green two-wavelength light-emitting diode,” IEEE Photonics Technology Letter, Vol. 18, NO. 13, July 2006. [10] H. Mataki & T. Fukui, “Organic/inorganic optical nanocomposite with highly-doped rare-earth nanoclusters: Novel phosphors for white LEDs,” 5th IEEE Conference on Nanotechnology, 2005. [11] [12] [13] [14] [15]

[16] [17] [18] [19] [20] [21] [22] [23] [24] [25] IEEE photonics technology letters. Vol.15, No. 1, January 2003 [26] Subramanian Muthu dvance Transformer Co.10275 West Higgins Rd, Rosemont, IL 600185603 [27] Ieice trans. Electron. Vol. E88-C, No. 11Noverber 2005 [28] 0-7803-7420-7/02 2002 IEEE

Chapter 4: White LED Market Analysis and Patent related issue
4.1 Abstract In this chapter, a simple traffic signal case will be discussed. Reasons why LED is going to be a more and more important light source to replace traditional light source will be introduced. The bottleneck of LED development will also be discussed to realize why it still takes some time for LED to be ready to take over the market. Then the market analysis of LED will be showed to predict the future expectation of LED market. Finally, some patent related issue of white LED and the patent fight between big companies in the market will also be discussed. LED light sources, on the other hand, do not tend to fail catastrophically. Instead, the light output degrades gradually over time. The useful operating lifetime of a power LED is extremely long, and often longer than the lifetime of the product in which it is embedded. In a large installation, depending on the mission life of the fixture, the effect of this fading over time may be to reduce the overall light output to a level below the specified minimum. However, as revamping is so infrequent, the total cost of ownership is reduced. 4.2 Case Study - Traffic Signals 4.2.1 Efficiency Traffic signals is the first out door light source to be replace by LED because it prevents enormous energy waste from using traditional light bulb base traffic signal. Almost all energy conscious organizations in the world target replacing traffic signal with LED as the first thing that government should finish. The luminous efficiency of the LED system far exceeds the efficiency of light bulbs. Generally, LED traffic signals save around 80% and 90% of the electricity comparing with conventional incandescent signals.

Table 4.1 [4.1] In table 4.1, it describes the energy consumption of traffic signals in a simple intersection. From this table, about four thousand Watts could be saved by replacing the light source with LED system for each intersection. According to a conservative estimation2, there are 3 million traffic signals in the United States. If all of them were using LED signals, the energy savings would be significant: 3 billion kWh saved annually, reducing overall peak demand by 340 MW.

4.2.2 Life Time The second benefit brought from LED system is its amazing long endurance. Table 4.2 shows the rated life and lumens output of different light source.

Table 4.2 [4.2] Generally the life time of a traditional light source is defined by the ANSI/IES and IEC/CIE as the time when 50% of the test object population are burned out. However, LED light source does not follow the same way as traditional light sources do. Instead of burning out, their brightness decreases over the time as shown in Fig 4.1. Currently, the definition of LED life is the point at which the LEDs reach 50% of their rated lumen output. Because LEDs are solid-state devices they are not subject to catastrophic failure when operated within design parameters. DDP® LEDs are designed to operate upwards of 100,000 hours at 25°C ambient temperature. Operating life is characterized by the degradation of LED intensity over time. When the LED degrades to half of its original intensity after 100,000 hours it is at the end of its useful life although the LED will continue to operate as output diminishes. Unlike standard incandescent bulbs, DDP® LEDs resist shock and vibration and can be cycled on and off without excessive degradation.

Fig 4.1 [4.3]

Fig 4.2 A lamp package LED

Fig 4.3 A high power LED

Fig 5.3 Fig 5.2 Fig 5.3 Fig 5.2

In table 4.3, the blue and red curve present the life time of conventional 5 mm lamp package white LED and high power LED with special thermal concerned packaging respectively. Comparing Fig 4.1 and Table 4.2, we can see that even with traditional lamp package, LED still lasts at least five times longer than traditional signals (up to 50,000 hours). In a recently published technical paper, instead of decreasing, the output lumen of some high power increases after light up for certain hours. Fig 4.2 and Fig 4.3 shows a lamp package LED and a high power LED. 4.2.3 Cost Currently, the major barrier for LED system to replace traditional light is its initial cost. The cost of LED system is way much higher than traditional light source. For example, as shown in Table 4.3, a 12” red LED signal can range from $60 to $125, and a 12” green LED module costs $160 to $250. Amber LED signals typically cost about $75 each. 12” red LED signal 12” green LED signal Amber LED signal Total setup cost 60 – 120 USD 160 – 250 USD 75 USD 295 – 445 USD Table 4.3 [4.4] However, it only costs about $2 to $2.50 for a traditional light bulb. Considering a typical fourway intersection, it will cost around three thousand more on the setup fee to replace the light bulbs with LED light source. At this time, it is estimated that about 10% of the traffic signals in the nation are LED traffic signals. The market for high brightness light emitting diodes (LEDs) will grow from $2.5 billion in 2003 to $5.9 billion in 2008 driven by LED demand from automobile and cell phone manufacturers. The good news for buyers is that prices will continue to fall. Prices dropped 10%-20% over the past 12 months and will decline at a similar percentage over the next year for most LEDs. The

exception will be white LEDs. Tags for these devices will fall, but at a smaller rate of decline because there are fewer suppliers and less capacity than for blue, green and yellow LEDs.

4.2.4 Conclusion As a new generation light source, LED has a lot benefits that make it a much better choice for indoor and outdoor light source. The two most important advantages of LED are its energy consumption and its indurations. However, the cost is main factor to slow down the pace of LED to take over the whole light source market. 4.3 Market Analysis 4.3.1 Haitz Law Like Moore’s law to semiconductor industry, Haitz Law5 typically states that the light output and efficacy of LEDs roughly doubles every 18 to 24 months, and that the future LED performance will likely follow a trend similar to that of the past 30 years. So far, though the growth of LED efficiency seems a little bit faster than Haitz Law’s expectation, the development of LED is still follow this statement. Most of the market reports right now did their prediction based on Haitz Law. LED market ready for accelerated growth in lighting, display backlights and automotive applications. The LED market reached $4.2 billion in 2006 and is set to emerge from its current state of low growth, according to Strategies Unlimited. Emerging applications including illumination will drive the market towards $9 billion by 2011, as Tim Whitaker reports. Growth in the high-brightness LED market in the next 3-5 years will be driven by lighting, display backlighting and automotive applications, and the overall market will reach $9 billion by 2011, according to the market research firm Strategies Unlimited. Presenting his annual market review and forecast at the Strategies in Light conference in early February, Strategies Unlimited's Bob Steele said that the market will begin to move away from its current low-growth phase in the next 1-3 years, as new markets begin to emerge. 4.3.2 Current LED Market

Unit: Million USD






Fig 4.4 Global LED market from 2003 to 2007 (Unit: Million USD) [4.6] Fig 4.4 shows the global LED market form 2003 to 2007. At 2005 the LED market is around 6.5 billion USD. The visible LED takes about 68% of the market and infrared LED takes the least 32% of the market. The annual growth rate of infrared LED steadily maintains at around 5% and the visible LED is around 12%. The total annual growth rate of LED market is about 10%. Unit: Million USD
4700 5900

14 %

- 0.2% 5 %



Fig 4.5 Growth of LED market from 2003 to 2005

As shown in Fig 4.5, most of the contribution of LED market growth from 2003 to 2005 comes from the strong demand of HB LED (High Brightness LED). Generally HB LED means those LED whose efficiency is higher than 8 lumens per Watt. 2.5 out of 5 billion of global LED market at 2004 are contributed by HB LED. According to the average growth rate of HB LED, the market of HB LED at 2007 will be around 3.2 billion USD. Energy-efficient LEDs penetrate all corners of daily life in industrialized countries. From traffic lights to mobile phones, the versatile light emitting diode is an increasingly commoditized product even as it develops high-end specialist uses. According to a new study from ABI Research, by 2010 the global LED market may consist of two segments: a highly commoditized market in simple, inexpensive LEDs for mass-produced goods, and another sector dealing in high-end, high-tech innovations. Light-emitting diode (LED) chip makers hope to make big inroads within the next decade into the $12 billion conventional lighting market, now served by incandescent and fluorescent bulbs. But in the meantime, they are doing very well supplying lights for cell phones, autos and traffic signals. The demand from mobile equipment creates the strong growth of HB LED from 2003 to 2005. At this time, almost all cell phone manufactures replaced their cell phone back light of monitor and keypad with LED because its high energy efficiency and small size. However, the growth rate will slow down because of the saturation of demand from cell phone. Fewer and fewer LED chip is going to be needed in future for mobile equipment due to the increasing LED efficiency. For next few years, it is believed that auto industry would be the next strong market to support the continuous growth of LED demand. LED is expected to replace the current interior and exterior light of automobiles. More and more auto-designer is using LED as the tail light in their design because of safety concern. According to the report, by using LED as tail light, driver could have longer reaction time when the driving speed is higher than 80 miles per hour because of the shorter “light-up” time of LED. The demand from mobile equipment creates the strong growth of HB LED from 2003 to 2005. At this time, almost all cell phone manufactures replaced their cell phone back light of monitor and keypad with LED because its high energy efficiency and small size. However, the growth rate will slow down because of the saturation of demand from cell phone. Fewer and fewer LED chip is going to be needed in future for mobile equipment due to the increasing LED efficiency. For next few years, it is believed that auto industry would be the next strong market to support the continuous growth of LED demand. LED is expected to replace the current interior and exterior light of automobiles. More and more auto-designer is using LED as the tail light in their design because of safety concern. According to the report, by using LED as tail light, driver could have longer reaction time when the driving speed is higher than 80 miles per hour because of the shorter “light-up” time of LED.

4.3.3 Illumination/ Light Source Market

Incandescent, 27% fluorescent, 42% Metal halide lamps, 17% HID, 14%

Unit: 2002 Million USD


Global Illumination/Light source Market





Fig 4.6 Actually, to replace the indoor light source is the final destination that all LED manufactures try to accomplish. As Fig 4.6 shown, the global market size of HB LED at 2005 is about 5 billion USD while the global illumination market is around 130 billion USD. The market of HB LED is very small right now but that also means how big the potential market is once the LED overcomes the barrier and occupies all the light source market. "This is market even without conventional lighting," says Bob Steele, director of optoelectronics for market research firm Strategies Unlimited. He expects worldwide sales of high-brightness LEDs, the market's largest and fastest-growing segment, to grow about 14 percent annually over the next five years, from $3.7 billion in 2004 to $7.2 billion in 2009. High-brightness LEDs use newer gallium-based technologies to convert energy into light more efficiently than previous LEDs, producing enough light even for daylight, outdoor uses a thriving. Efficiency From Fig 4.6, traditional light bulbs take around 27% of illumination market at 2005. That’s a big market with amount around forty billion USD. The efficiency of HB LED in the market right now is around twenty to thirty lumens per Watt. It’s already higher than the efficiency of traditional light bulbs, eight to ten lumens per Watts. Then why the LED doesn’t take over light bulb’s market? The reason is the efficiency to cost rate.

As the simple case discussed in Sec 5.2, due to the expensive setup fee, the government didn’t start to replace the traffic signal with LED base light source until the efficiency of LED base light source was ten times higher than the light bulb source traffic signal. The same logic can also be used to predict the future LED market.

Table 4.4 Today's best LEDs are twice as energy-efficient as incandescent lights and almost equal to fluorescent bulbs, and multiple LEDs can be combined to generate just as much light as conventional bulbs. Yet LEDs typically cost 100 times as much for the equivalent light output, although the gap is closing. Traffic signals are also converting to LEDs, which use much less energy and can last a decade, whereas conventional bulbs must usually be replaced yearly. A recent U.S. Energy Department study found that LED traffic signals can reduce energy use by 90 percent. For similar reasons, automakers are putting LEDs into rear brake and running lights as well as dashboard and interior lighting. The first LED headlights are expected to appear in 2007.

Fig 4.7 LED Illuminated Skyscrapers in the City of Hong Kong "All the TV manufacturers are looking at it now," says Jovani Torres, America’s regional product marketing manager for Agilent's optoelectronics division. LED backlighting produces brighter, truer colors and eliminates the mercury used in today's cold cathode fluorescent lights, which

raises environmental concerns. LED modules will improve safety on New York's subway, while also reducing energy and maintenance costs.Dialight Corporation has been awarded a $1.8 million contract to provide LED trackside signals for the New York City subway system.The cost is expected to keep going down to reach 0.02 USD/lm at 2007. However, it has to be further reduced to around 0.01 USD/lm to really be applied in illumination market. This probably will be accomplished at around 2008 or 2009. To achieve this low cost, the main difficulty is the LED package. Though the package technology is already mature now, it still needs to be refined to find the better material and process to further reduce the cost. 4.3.4 Conclusion White LED has a huge potential market in indoor light source market. However, it still limited by the insufficient efficiency, higher setup cost and energy cost. The development of device and package technology will decide when LED could really dominate the light source market. Around 440,000 Cree LEDs will be used to illuminate the Water Cube, home to swimming and other events at the Summer Olympics in Beijing. XLamp LEDs from Cree will be used to provide the effects lighting for the new Beijing National Aquatics Centre. The Aquatics Centre, commonly known as the Water Cube, will be the home for nearly all of the 2008 Summer Olympics aquatic events. As LED technology continues developing and shows advantages of low consumption, long lifespan and environmental protection, its application has shifted from initial indicator light to such more potential fields as display panels, illumination, backlight, automobile lights and traffic lights, etc. Now LED application is developing in a diversified way. Demand for normal brightness still ranks first in China's LED market; however, the market of high brightness and super high brightness LED has even greater potential. Traditional red and green LED are widely drawn on in China. AlGaInP LED is still the mainstream and GaN LED is increasing sharply. The year of 2005 witnessed a growth rate of 18.4% in China's LED market.

Patent issues
On November 13, Cree and Nichia Corporation, two of the leading developers of GaN-based LEDs and lasers, announced that they have entered into a patent cross-license agreement. The agreement ends all ongoing litigation between the companies, and is designed to prevent future litigation. Cree and Nichia say that the agreement will allow them to focus more resources on developing technology for next-generation high-brightness LEDs, laser diodes and LED-based lighting. OSRAM and Philips have signed a patent crosslicense agreement covering optoelectronic semiconductors. The agreement involves the mutual licensing of patents for all inorganic and organic LED. The agreement relates to patents held by Philips, including the US subsidiary Lumileds, and by OSRAM including its subsidiary OSRAM Opto Semiconductors.

Fig 4.8

Siemens subsidiary Osram GmbH and Japanese company Toyoda Gosei Ltd. have agreed to exchange patents for Laser and LED technologies. The cooperation with Toyoda Gosei relates to indium gallium nitride (InGaN) semiconductor technologies used in the design and manufacturing of semiconductor lasers and LEDs producing white, green or blue light. Toyoda Gosei has developed the world's first blue LED in 1991 and owns protected intellectual property related to this LED type. In the patent infringement lawsuit involving Nichia's four U.S. design patents against Seoul Semiconductor, the presiding Judge Chesney from the U.S. District Court for the Northern District of California issued her rulings on the parties' summary judgment motions on August 22, 2007. Nichia Corporation today announced that Nichia and Cree, Inc. (Nasdaq: CREE) have entered into an agreement that expands their cross license arrangements announced in 2002 and 2005 to include additional patents relating to white LED technology and certain Cree patents relating to nitride lasers. The company’s recent records for light output, efficacy and thermal management are direct results of the ongoing commitment to advancing solid-state lighting technology and enabling lighting solutions that are more environmentally friendly, help reduce CO2 emissions and reduce the need for power plant expansion. Philips Lumileds’ LUXEON LEDs are enabling never before possible applications in the automotive, camera flash, display, general lighting, and signage markets.

Fig 4.9 shows the complicated deals and disputes in white light LED industry. It could be easily observed that almost 70% of them are surrounded with Nichia. Actually Nichia was still suing Osram, Toyoda Gosei, Cree and Lumileds patent infringements until late 2005. The agreements between these four big companies were formed at around September, 2005. Citizen is still the only company which Nichia licenses to use their patent to produce white LED because of their advanced technique at LED packaging9.

Fig 4.9

[1] LEDs in Exterior Applications: An Emerging Market. E Source #ER-01-17 November 2001. [2] IESNA Lighting Handbook: Reference and Application, 9th edition, 2000, N. Narendran, J.D. Bullough, N. Maliyagoda, and A. Bierman [3] Lumen Maintenance of White Luxeon Power light sources, available at [4] LED Lighting Technologies and Potential for Near-Term Applications. E Source #E03-114 June, 2003 [5] A Market Diffusion and Energy Impact Model forSolid-State Lighting, SAND2001-2830J, August 2001. Drennen, Thomas, Roland Haitz, and Jeffrey Tsao [6] 封裝技術與材料推動LED發光效能, 盧慶儒, A482570C30040B424&zKeyword=LED [7] 高亮度LED技術與應用趨勢, 郭長祐, B3F4FFD71482571A300574E16&zKeyword=LED [8] Small companies fight for a foothold in white LED sector, October 2005, Andrew Philips [9] Patent protagonists head to court, April 2005, editors of LED magazine

Chapter 5

Competitors for white light technology

5.1 Organic Light Emitting Diode (OLED)
OLEDs are a flat display technology, made by placing a series of organic thin films between two conductors. When electrical current is applied, a bright light is emitted. Because OLEDs produce (emit) light they do not require a backlight. This means that OLEDs can be made very thin and very power efficient when compared to LCD (Which do require a white backlight), and also it might be possible to create efficient white lighting from OLEDs. 5.1.1 Basic Principle An OLED is a special type of light emitting diode wherein the emissive layer consists of a thin film of organic compounds.

Fig 5.1 Structure of OLED The basic cell structure consists of a stack of thin organic layers between a transparent anode and a metallic cathode. When an appropriate voltage (typically between 2 and 10 volts) is applied to the cell, positive charges are injected from the anode and negative charges are injected from the cathode and they recombine in the emissive layer to produce light. The organic material is a luminophore which is divided into fluorophores and phosphors. The anode is made up of Indium Tin Oxide (ITO) and the cathode is generally made up of

Aluminium (Al). A variety of substrate materials such as films, foils, textiles, fabrics or plastics can be used. 5.1.2 Making OLEDs The biggest part of manufacturing OLEDs is applying the organic layers to the substrate. This can be done in three ways: • Vacuum deposition or vacuum thermal evaporation (VTE) - In a vacuum chamber, the organic molecules are gently heated (evaporated) and allowed to condense as thin films onto cooled substrates. This process is expensive and inefficient. • Organic vapor phase deposition (OVPD) - In a low-pressure, hot-walled reactor chamber, a carrier gas transports evaporated organic molecules onto cooled substrates, where they condense into thin films. Using a carrier gas increases the efficiency and reduces the cost of making OLEDs. • Inkjet printing - With inkjet technology, OLEDs are sprayed onto substrates just like inks are sprayed onto paper during printing. Inkjet technology greatly reduces the cost of OLED manufacturing and allows OLEDs to be printed onto very large films for large displays like 80-inch TV screens or electronic billboards. 5.1.3 Types of OLEDs There are several types of OLEDs:

Passive-matrix OLED: PMOLEDs have strips of cathode, organic layers and strips of anode. The anode strips are arranged perpendicular to the cathode strips. The intersections of the cathode and anode make up the pixels where light is emitted. External circuitry applies current to selected strips of anode and cathode, determining which pixels get turned on and which pixels remain off. Again, the brightness of each pixel is proportional to the amount of applied current. PMOLEDs are easy to make, but they consume more power than other types of OLED, mainly due to the power needed for the external circuitry. PMOLEDs are most efficient for text and icons and are best suited for small screens (2- to 3-inch diagonal) such as those you find in cell phones, PDAs and MP3 players. Even with the external circuitry, passive-matrix OLEDs consume less battery power than the LCDs that currently power these devices.

Fig 5.2 OLED: Passive Matrix [5.7]

Active-matrix OLED: AMOLEDs have full layers of cathode, organic molecules and anode, but the anode layer overlays a thin film transistor (TFT) array that forms a matrix. The TFT array itself is the circuitry that determines which pixels get turned on to form an image. AMOLEDs consume less power than PMOLEDs because the TFT array requires less power than external circuitry, so they are efficient for large displays. AMOLEDs also have faster refresh rates suitable for video. The best uses for AMOLEDs are computer monitors, largescreen TVs and electronic signs or billboards.

Fig 5.3 OLED: Active Matrix [5.7]

Transparent OLED: Transparent OLEDs have only transparent components (substrate, cathode and anode) and, when turned off, are up to 85 percent as transparent as their substrate. When a transparent OLED display is turned on, it allows light to pass in both directions. A transparent OLED display can be either active- or passive-matrix. This technology can be used for heads-up displays.

Fig 5.4 OLED: Transparent Structure [5.7]

Top-emitting OLED: Top-emitting OLEDs have a substrate that is either opaque or reflective. They are best suited to active-matrix design. Manufacturers may use top-emitting OLED displays in smart cards.

Fig 5.5 OLED: Top-Emitting Structure [5.7]

Foldable OLED: Foldable OLEDs have substrates made of very flexible metallic foils or plastics. Foldable OLEDs are very lightweight and durable. Their use in devices such as cell phones and PDAs can reduce breakage, a major cause for return or repair. Potentially, foldable OLED displays can be attached to fabrics to create "smart" clothing, such as outdoor survival clothing with an integrated computer chip, cell phone, GPS receiver and OLED display sewn into it.

Fig 5.6 OLED: Folded [5.5]

White OLED: White OLEDs emit white light that is brighter, more uniform and more energy efficient than that emitted by fluorescent lights. White OLEDs also have the true-color qualities of incandescent lighting. Because OLEDs can be made in large sheets, they can replace fluorescent lights that are currently used in homes and buildings. Their use could potentially reduce energy costs for lighting.

5.1.4 Comparison of light emission efficiency and light emission life between OLED and existing forms of illumination OLED white-colored light Incandescent light Fluorescent device (Konica Minolta) *2 bulb lamp Light emission efficiency (lm/W)*1 Light emission life (hours) 64 10 – 20 60 - 100 LED 50 - 70 Up to 40,000


Up to 3,000

5,000 - 10,000

*1 Lamp efficiency  *2 Evaluated at initial luminance of 1000cd/m2, life means 50% lumen maintenance Data from “White OLED Combining High Efficiency With Long Life”

5.1.5 Advantages of OLED (1) A large advantage of OLED is they can be made at low temperature. This allows for plastics, which in turn allow for flexible and thinner displays. It is also a reduction in weight. (2) As OLED pixels in a display can be made very small, they allow for high resolution displays. (3) Their response time is much faster then LCD pixel. (4) As they generate light themselves, it eliminates the need for a backlight. This means that they draw far less power and when powered from a battery can operate longer on the same charge. (5) They are brighter and more efficient than LEDs. (6) They offer great potential for lighting applications ranging from general purpose illumination to small flat panel displays found in mobile phones and digital music players. (7) They have the ability to tune the light emission to any desired color. (8) They are current-driven devices, where brightness can be varied over a very wide dynamic range and they operate uniformly without flicker. (9) They can be deposited on any substrate such as glass, ceramics, metal, thin plastic sheets, fabrics and therefore, can be fabricated in any shape and design. (10) Since OLED can me printed onto any suitable substrate, they can have a significantly lower cost than LCD or plasma displays. (11) The viewing angle possible with OLED is greater because OLED pixels directly emit light. 5.1.6 Disadvantages of OLED (1) The biggest problem is the limited lifetime of the organic materials. Particularly, blue OLEDs typically have lifetimes of around 5000 hours. (2) The intrusion of water into displays can damage or destroy the organic materials. Therefore, improved sealing processes are important for practical manufacturing and may limit the duration of more flexible displays. (3) Commercial development of the technology is also restrained by patents held by firms. (4) As they are made with organic material, they are susceptible to heat. All the energy which is not emitted in the form of light is converted to heat, degrading the organic layer. (5) Because efficiency of OLED is not yet very high, higher current is needed to make the OLED emit the desired amount of light. This also results in more heat, which slowly destroys the LED. (6) When large displays are made, the lifetime also drops.

5.2 Quantum Dots based Light Emitting Diode
Quantum dots are nanoscale semiconductor particles composed of periodic element groups of IIVI, III-V, or IV-VI that can be tuned to brightly fluoresce at virtually any wavelength in the visible and infrared portions of the spectrum when illuminated by a shorter wavelength source. Quantum yields are in excess of 80% and the emission spectrum is bell shaped with a full-widthhalf-max (FWHM) of 35nm or 70nm depending on the quantum dot composition. Quantum dots have a combination of performance attributes not achievable with other lumophores. The ideal OLED lumophore would have high photoluminescence quantum yield, be

capable of emitting light from 100% of electrically generated excitons, be solution processable, and have extremely high stability and differential stability (to eliminate image sticking and white-point drift effects). Polymers, dendrimers, fluorescent, and phosphorescent small molecules have all undergone extensive development, but have yet to yield a single material set that meets the complete set of industry needs. Quantum dots are a new class of lumophores that have the promise to meet all these needs simultaneously. Quantum-dot LEDs have theoretical performance limits that meet or exceed all other display technologies. Today, phosphorescent OLEDs have the best demonstrated efficiency of any nonreflective display technology, but QD-LEDs have the potential to exceed OLED luminous efficiency by more than 20%, reducing power consumption significantly. The color saturation of red, green, and blue QD-LEDs is represented by their position on the CIE diagram relative to the high-definition-television (HDTV) standard color triangle. The red and green devices exceed the HDTV standard, while the blue QD-LED CIE color coordinates lie just inside the standard. All three color QD-LEDs have reproducible, stable current-voltage characteristics, with turn-on voltages of 2 to 4 V and operating voltages of 5 to 9 V. 5.2.1 Quantum Dot Research A QD (quantum dot) is also called a semiconductor nanocrystal or an “artificial atom” because they have an energy level similar to an atom. These energy levels yield the color you can see in a nanocrystal display. Research into QDs/nanocrystals began in 1960, according to the Quantum Dot History Project from Evident Technologies, and it has grown substantially during the past 25 years. Researchers at Sandia National Labs, MIT, and the U.S. Department of Energy have made several breakthroughs using QDs in lighting in the past several years. In year 2005, a discovery at Vanderbilt University has combined LEDs and QDs to create a more pure white light. LED lights usually have a bluish tinge, but they became whiter and warmer when researchers combined them with QDs. And just last year Nanosys received two patents related to using nanocrystals in displays. Research into QDs goes far beyond lighting and displays, though, including blue lasers for HDDVD and Blu-ray, quantum computation, development of memory, biological research, and some medical applications.

5.2.2 Basic Principle

Fig 5.7

Fig 5.8 Consider a bulk semiconductor. The electrons in the bulk semiconductor material may have different energy levels. Electrons occupying energy levels below the bandgap are described as being in the valence band. The electrons that have been raised into the conduction band will stay there only momentarily before falling back to the valence band. As the electron falls back, an electromagnetic radiation is emitted whose wavelength corresponds to the energy lost in the trasition. When the electron jumps, it travels from one edge of the bandgap to the other. Because the bandgap of the bulk is fixed, this transition results in fixed emission frequencies.

5.2.3 Principle of Quantum Confinement The electrons and holes have an average physical separation referred to as exciton bohr radius. In bulk, the semiconductor crystals are much larger than the bohr radius. If the size of the semiconductor crystal becomes small enough hat it approaches the size of the material’s bohr radius, then the electron energy levels can no longer be treated as continuous. They must be treated as discrete. This situation of discrete energy levels is called quantum confinement. Under these conditions, the semiconductor crystal is called a quantum dot. Thus, quantum dots are unique class of semiconductors because they are so small ranging from 2-10 nanometers in diameter. Quantum dots offer the ability to tune the bandgap and hence the emission wavelength.

Fig 5.9 Quantum Dots - New paths to Solid State Lighting 5.2.4 Size Dependent Control of Bandgap in Quantum Dots

Fig 5.10

As with bulk semiconductor material, electrons tend to make transitions near the edges of the bandgap. However, with quantum dots, the size of the bandgap is controlled simply by adjusting the size of the dot. Because the emission frequency of a dot is dependent on the bandgap, it is therefore possible to control the output wavelength of a dot with extreme precision. In effect, it is possible to tune the bandgap of a dot, and therefore specify its "color" output depending on the needs of the customer. 5.2.5 Emissions from Various Light Sources

Fig 5.11 Emission spectrum of QD-LEDs vs. Sunlight As the graphs shown in fig 5.11, QD-LED’s visible emissions spectrum is closer to sunlight than other light sources. As an added benefit, QD-LED would provide its strong visible light emission by using less power than the other options shown here. For example, some of the energy the 100W light bulb uses is lost through heat generation, as shown by the infrared emissions near and beyond the 800nm wavelength range in its graph. Most humans can see visible light ranging from 400 to 700nm wavelengths, although some people have the ability to view light in wavelengths ranging from 380 to 780nm. (Beyond 700 to 780nm wavelengths is infrared.)
Sources: Vanderbilt University, Wikipedia

5.2.5 Ohmic Contacts A layer of Ti/Al or Ti/Au is deposited on the n-GaN layer. The n-contact is formed on an etched surface. Ti/Al or Ti/Au can be replaced by V/Al or V/Au as the contact resistance is reduced by using V/Al or V/Au. A layer of Pd/Au or Ni/Au is deposited on the p-GaN layer.

5.2.6 Quantum Dot Material Systems and Emission Ranges Quantum dots are tuned by selecting a semiconductor composition and chemically growing it to the size that yields the desired properties. Quantum dots are available in the following nanomaterials systems: Quantum Emission Dot Range Material System CdSe 465nm 640nm CdSe/Zns 490nm 620nm CdTe/CdS 620nm 680nm PbS 850nm 2100nm PbSe 1200nm 2340nm Quantu m Dot Diameter Range 1.9nm 6.7nm 2.9nm 6.1 nm 3.7nm 4.8nm 2.3nm 9.8nm 4.5nm 9nm Quantum Standard Dot Type Solvents Core CoreShell CoreShell Core Core Toluene Toluene Toluene Toluene Toluene Qauntum Dot Example Applications Research, Solar Cells, LEDs Visible Fluorescence Applications, Electroluminescence, LEDs Deep Red Fluorescence Apps. Near Infrared Applications, Security Inks, Solar Cells, IR LEDs Opto-electronics, Optical Switching, Non-linear Applications, Photonics, Telecommunications

Fig 5.12 5.2.7 Advantages of Quantum Dots (1) The very first advantage is their small size. Due to this, they can be tuned to emit at any visible or infrared wavelength. (2) The small size also allows for incredible flexibility. (3) Extremely small size allows them to be inserted into any medium necessary. (4) Bandgap can be altered with the addition or subtraction of just one atom. (5) Predetermining the size of QLEDs dots would fix the emitted photon wavelength at the appropriate customer- specified color even if it is not naturally occurring. (6) They provide with high stability.

(7) The final advantage of quantum dot phosphors is the ability to produce high quality, low-cost white light relative to the standard white light producible in traditional semiconductor systems. This can be done by intermixing red, green, and blue emitting dots homogenously within the phosphor, a feat of great difficulty with traditional phosphors. Thus, quantum dots need only a single excitation source for multiple emission colors, even to the point of producing industry quality white light. In addition, the unique Core-Shell Technology allows quantum dot phosphors to exhibit great stability, retaining high quantum yields and luminosities over long lifetimes, in a variety of different environments. In contrast, organic dyes and other common phosphors/fluorophores have poor photo-stability and short lifetimes relative to quantum dots. 5.2.8 Disadvantages of Quantum Dots Quantum dots are generally made up of Cd, Se or Pb. These materials are toxic in nature. They cause environmental harm. They are more expensive. 5.2.9 Quantum Dot Global Market BCC Research has projected explosive growth for the QD market worldwide as new uses for the technology become available. 2003 Total: $5 million (all biology and medical sectors) 2005 Total: $13 million (all biology and medical sectors) 2007 Total: $279.5 million ($122 million biology and medical sectors) Memory sector: $157.5 million 2009 Total: $522.5 million ($193.2 million biology and medical sectors) Display sector: $18.8 million Memory sector: $190.5 million Solar energy sector: $120 million Source: BCC Research

References [1] [2] Wikipedia, the free encyclopedia [Online]. Available [3] H. C. Yu, J. S. Wang, Y. K. Su, S. J. Chang, F. I. Lai, Y H. Chang, H. C. Kuo, C. . P. Sung, H. P. D. Yang, K. F. Lin, J. M. Wang, J. Y Chi, R. S. Hsiao, and S. . Mikhrin, (2006) “1.3-_m InAs–InGaAs Quantum-Dot Vertical-Cavity SurfaceEmitting Laser With Fully Doped DBRs Grown by MBE” [4] “X-Ray Vision: Nanocrystal Displays”, Hard Hat Area , September 2006 • Vol.6 Issue 9 [5] [6] [7]

Sign up to vote on this title
UsefulNot useful