InGaN light-emitting diodes of stripe geometries have been demonstrated. The elongated geometry facilitates light spreading in the longitudinal direction. The chips are further shaped by laser-micromachining to have partially-inclined sidewalls. The light extraction efficiencies of such 3D chip geometries are enhanced by ~12% (~8% according to ray-trace simulations), leading to a reduction of junction temperatures. The effective emission area is also increased four times compared to a cubic chip. The stripe LEDs are thus more efficient emitters with reduced luminous exitance, making them more suitable for a wide range of lighting applications.
© 2015 Optical Society of America
Although light-emitting diodes (LED) have found their way into many lighting and display products such as energy-saving lamps and liquid-crystal display backlights, solid-state lighting is still competing fiercely with compact fluorescent lamps (CFLs) and, to a certain extent, incandescent lamps [1,2]. While the price tag may be one of the major deterring factors (although many would argue about the cost-of-ownership in the long run), the reality is that many of the existing LED-based lamps produce lighting experiences that are far from ideal. The common complaints of LED lighting include poor color-rendering, flickering , excessive glare and non-uniform “patchy” illumination. While the color perception and flickering effects are attributed to the choice of phosphors and the drivers respectively, many of the other shortcomings are due to the nature of the chip itself. With intense efforts to boost device efficiencies, LEDs are now capable of delivering >100 lumens of white light out of a 1 W millimeter chip, corresponding to luminous exitances exceeding 105 lm/cm2. Compare this with a typical 5W CFL emitting >200lm across >30 cm2 of emission surface area from its glass envelope, whose luminous exitance would be three orders of magnitude lower than that of the LED, thus the unpleasant experience when viewing direct. It has been reported that the damage threshold to the human eye due to incident visible light is ~10x10−3 W/cm2; as the output powers of LED become increasing high, our eyes are at increasing risks of harm [4,5].
The root cause of this problem lies with the geometry and dimensions of LED chips. The common approach of overcoming the annoying glare is through the use of external optical elements such as diffusers. That, unfortunately, will not only reduce the overall efficiency of the lighting system due to low transmission and absorption , but also increases the cost of the product, as well as its bulkiness. Organic LEDs (OLEDs) are doing better in this and other aspects, with advantages in form factor, large-area emission as well as color comfort, despite lagging behind on costs, efficiencies and lifetimes . In view of these limitations, a solution is proposed in this paper: re-design the LED geometry to reduce light crowding. Mere enlargement of the chip, with a proportional increment of current drive to maintain current density (thus light output) would keep the luminous exitance high. Instead, expansion of the aspect ratio of a chip would reduce its luminous exitance by spreading emission along the longitudinal axis. LEDs of stripe geometries aspect ratio exceeding 5:1 are demonstrated. Instead of cutting chips with vertical sidewalls, the stripe LEDs are further shaped to have partially-inclined sidewalls to further boost their light extraction efficiencies. The designs of the chips are optimized with the aid of ray-trace simulations; based on these designs, chips are fabricated, packaged, thoroughly characterized and compared with the predictions.
2. Experimental details
The epitaxial structure of the blue LEDs in this work are grown by metal-organic chemical vapor deposition (MOCVD) on c-plane sapphire, which consists of a 3-μm-thick undoped GaN layer, a 3-μm-thick Si-doped GaN layer, 10 periods of InGaN/ GaN quantum wells, capped with a 0.25-μm-thick Mg-doped GaN layer. A 200 nm-thick indium tin oxide (ITO) is deposited as a p-type current spreading layer. The mesas of the stripe LEDs with dimensions of 15 mm x 0.235 mm are patterned by photolithography and etched by Cl2-based inductively-coupled plasma etching. After definition of contact pad regions, Ti/Al/Ti/Au (40/150/20/40 nm) is deposited by electron beam evaporation followed by annealing. For comparisons, 1.9 x 1.9 mm2 cubic LEDs with identical mesa area are fabricated alongside. All devices emit at a peak wavelength of ~470 nm with spectral width of ~35 nm full-width at half-maximum (FWHM). The chips are diced by laser micro-machining with a diode-pumped solid-state (DPSS) ultraviolet (UV) pulsed laser at 349nm. Some of the stripe LEDs are shaped to have partially-inclined sidewalls using a modified laser micro-machining setup involving the use of a beam steering mirror; these are referred to as ∠-stripe LED. Details of the chip-shaping process can be found in . Cross-sectional profiles of the stripe and ∠-stripe chips are depicted in the ray-trace diagrams of Figs. 1(c) and 1(d) respectively. The separated chips are epoxy-mounted onto specially-designed ceramic substrates with copper conductor layers and are wire-bonded using Al wires. The electrical, optical and thermal performances of the LEDs are comprehensively characterized. Microphotographs of the cubic, stripe and ∠-stripe LEDs, driven at currents of 20 mA, are shown in Figs. 1(a)-1(c) respectively.
3. Results and discussions
Considering the lengths of the stripe chips and the resistivities of the n- and p-type GaN layers, current spreading cannot be ignored. Based on a simple calculation, the voltage would have dropped by 60% if the potential difference is applied between the cathode and anode end-to-end. Conversely, in order to maintain 95% of the applied voltage along the entire length of the chip, 6 cathodes/ anodes will be required . For this purpose, additional bond pads, interconnected via a narrow metal rail along the long edge, have been implemented, as illustrated in the schematic diagrams of Fig. 2. To illustrate their effectiveness, microphotographs of the stripe LEDs driven at 50 mA with 2, 4 and 6 pads connected are displayed in Figs. 3(a), 3(c) and 3(e) respectively. Cross-sectional emission intensity maps of the respective devices, obtained by confocal microscopy, are shown in Figs. 3(b), 3(d) and 3(f). With all pads connected, light emission uniformity along the strip becomes homogenous.
The light extraction efficiencies of the LEDs of cubic, stripe and ∠-stripe LEDs are found to be 9.39%, 9.73% and 10.14% according to simulations by optical ray-tracing. In the model the refractive indices of GaN and sapphire had been set to 2.44 and 1.78 respectively, while their absorption coefficients are 15 mm−1 and 0 mm−1 respectively . Ray-trace plots for the cubic, stripe and ∠-stripe LEDs are illustrated in Figs. 1(d)-1(f) respectively, pictorially showing that more light rays can be extracted from the tapered structured. Details of the ray-trace simulations have been reported in . Based on the simulated data, the stripe geometry offers enhanced light extraction due to increased surface are to volume ratios. The enhancement effect is further increased by introducing partially-inclined sidewalls, reducing the symmetry of the device so that light rays which would otherwise be confined between parallel sidewalls can now be extracted. The light output- injection current characteristics are measured using an LED characterization system by placing the packaged devices, driven at 50mA, within a 12-inch integrating sphere which is coupled to a radiometrically-calibrated spectrometer, results of which are shown in Fig. 4. The stripe and ∠-stripe LEDs emit 8.5% and 12.6% more light than the cubic LED at 250 mA respectively, which are generally in line with ray-trace predictions albeit with larger magnitudes, which can be attributed to scattering on chip sidewalls which are not taken into account in the simulations. Far-field angular emission profiles of the LEDs have also been generated from ray-trace simulations, as shown in Fig. 5(a). It is noted that the ∠-stripe LED emits with reduced divergence. As observed from Fig. 1(e), light rays emitted from the angled facet generally bend upwards due to refraction, accounting for the narrower divergence.
Light emission distributions are further analyzed with the aid of confocal microscopy, a powerful tool which has been demonstrated to be effective in characterizing emissions from LEDs through the acquisition of 3D intensity maps [11,12]. The measurements are conducted using a Carl Zeiss LSM700 confocal microscope with 150x objective lens (NA = 0.95). To avoid contact with the bond wire, the first (lowest) scanning plane is 250 µm from the surface of the chip. The vertical scanning range is set to 5000 µm in steps of 200 µm; a total of 26 confocal planes are obtained. From the collected data, 3D emission profiles, as well as cross-sectional light distribution diagrams can be reconstructed. Figure 6 (a)-(c) show normalized cross-sectional emission profiles from the cubic, stripe, and ∠-stripe LEDs respectively along the lateral axis of the chips, driven at 80 mA. As the measurements are taken at distances on par with the dimensions of the devices, the emission profiles are dependent on the lateral dimensions of the LEDs and thus cannot be treated as point sources as with far-field measurements (by photogoniometry for instance). For this reason, only comparisons between the stripe and ∠-stripe LEDs are meaningful. As observed, light emission from the ∠-stripe LED is stronger than the stripe LED, consistent with the observations from ray-tracing and L-I measurements.
It is also possible to obtain angular emission profiles from the cross-sectional intensity maps, based on the method and rationale discussed in , as shown in Fig. 6(b). Once again, it is evident that the emission divergence of the ∠-stripe LED has been narrowed down compared to the stripe LED. It is worth pointing out that angular emission plots generated from the confocal microscopy data are not directly comparable to ray-trace simulated plots for two reasons: (1) the measurement distances are dissimilar and (2) light emitted from the vertical sidewalls of the LEDs cannot be captured by the objective of the confocal microscope.
High-resolution planar light intensity mapping, obtained by confocal microscopy, is carried out to evaluate emission uniformity of the chips. Confocal imaging is capable of recording point-by-point intensities over the measured area on a particular focal plane. In this case, a 5x objectives lens (NA = 0.13) is used while the measurement plane is set to be 4 mm away from the top surface of the chips, which are powered up at an injection current of 50 mA. The measured intensity maps are depicted in Fig. 7. Light emission from the cubic LED is non-uniformly distributed, with the highest intensities at the center of the chip, dropping to less than half of that value near the periphery. On the other hand, light emission is homogenous across the entire chip area. If the emission area of an LED is defined to have a boundary at which light intensities have dropped to 30% of its peak value, then the emission areas of the cubic, stripe and ∠-stripe LEDs can be evaluated as ~4.8, ~14.33 and ~16.06 mm2 respectively. In order words, the effective emission area of the ∠-stripe LED is four times that of the cubic LED, despite identical pn junction area of ~3.6 mm2. The enlarged emission area contributes to the reduced luminous exitance, signifying that the stripe geometry is capable of spreading light along the longitudinal axis.
The geometry of the LED chips affects not only their optical properties but also their thermal characteristics, which in turn affects their internal quantum efficiencies. The junction temperatures of the chips are measured using a T3ster Transient Thermal Characterization system, results of which are tabulated in Table 1. At 300 mA, the cubic LED attains the highest Tj of 150°C, nearly twice as high as the ∠-stripe LED. The observed trend may be explained in terms of light extraction efficiencies. For the ∠-stripe LED with the highest light extraction efficiency and light output, light re-absorption is reduced resulting in lower lattice temperatures, which in turn increases internal quantum efficiencies. The higher light extraction efficiencies of the ∠-stripe LEDs may also have promoted the emission of light at infrared wavelengths, which is essentially thermal radiation, thus lowering junction temperatures, particularly at higher driving currents .
The stripe design, apart from being more efficient emitters, represents a promising approach towards narrowing the gap between LEDs and OLEDs in several aspects, particularly in terms of emission areas. The stripe LEDs can be further arrayed to form longer strips, with applications for LED tubes as fluorescent lamp replacements, LED “filament” lamps as incandescent lamp replacements, as well as for edge-type liquid-crystal display (LCD) backlight units. They are also particularly suitable for thin light guide plates (LGPs) with thickness of 0.5 mm or even less which are typically used in mobile devices. In the article by C. Baum et al. , the limiting factor for the reduction of the LGP thickness is identified to be the LED. To illustrate the suitability of the stripe LED for this purpose, the coupling and distribution of light emitted from a linear array of 5 2.1 x 2.1 mm2 cubic LEDs and 15.6 x 0.405 mm2 stripe LEDs into 1000 x 1000 mm2 sapphire LGPs with thicknesses of 0.5 mm and 2.2 mm are investigated through ray-trace simulations. While ~90% of light from the stripe LEDs is coupled into the LGP, the coupling efficiency drops to ~66% for the cubic LEDs. After from coupling efficiency, the ray diagrams in Fig. 8 highlight another feature of the stripe LEDs that is emission uniformity, heavily reducing the need for scattering elements for achieving homogeneity.
By simply re-shaping the geometry of an LED chip, its efficiencies and functionalities can be enhanced. While a ~10% increase of optical output makes it a more energy-saving emitter, the elongated geometry stretches the light emission cone along the longitudinal axis, giving an overall increase of emission area by fourfold. Light emission along the LED stripe becomes much more uniform compared to the conventional cubic chip, all achieved without the usage of external optics. The stripe chips also sustain lower junction temperatures. LED chips adopting the stripe geometry are demonstrated to be well-suited for lighting applications due to their reduced luminous exitance compared to cubic chips.
This work is supported by the Theme-based Research Scheme (T22-715/12-N) of the Research Grant Council of Hong Kong.
References and links
1. S. Pimputkar, J. S. Speck, S. P. DenBaars, and S. Nakamura, “Prospects for LED Lighting,” Nat. Photonics 3(4), 180–182 (2009). [CrossRef]
2. S. T. Tan, X. W. Sun, H. V. Demir, and S. P. DenBaars, “Advances in the LED materials and architectures for energy-saving solid-state lighting toward “lighting revolution”,” IEEE Photonics J. 4(2), 613–619 (2012). [CrossRef]
3. A. Wilkins, J. Veitch, and B. Lehman, “LED lighting flicker and potential health concerns: IEEE standard PAR1789 update,” in IEEE Energy Conversion Congress and Exposition (IEEE, 2010), pp. 171–178.
4. D. H. Sliney, “Laser and LED eye hazards: safety standards,” Opt. Photonics News 8(9), 31–37 (1997). [CrossRef]
5. International Commission on Non-Ionizing Radiation Protection, “ICNIRP statement on light-emitting diodes (LEDS) and laser diodes: implications for hazard assessment,” Health Phys. 78(6), 744–752 (2000). [CrossRef] [PubMed]
6. C.-C. Sun, W.-T. Chien, I. Moreno, C.-T. Hsieh, M.-C. Lin, S.-L. Hsiao, and X.-H. Lee, “Calculating model of light transmission efficiency of diffusers attached to a lighting cavity,” Opt. Express 18(6), 6137–6148 (2010)
7. K. Ghaffarzadeh, “OLED Lighting: The Differentiation Challenge,” Inf. Disp. 30(2), 36–38 (2014).
8. Z.-H. Zhang, S. T. Tan, W. Liu, Z. Ju, K. Zheng, Z. Kyaw, Y. Ji, N. Hasanov, X. W. Sun, and H. V. Demir, “Improved InGaN/GaN light-emitting diodes with a p-GaN/n-GaN/p-GaN/n-GaN/p-GaN current-spreading layer,” Opt. Express 21(4), 4958–4969 (2013). [CrossRef]
9. W. Y. Fu, K. N. Hui, X. H. Wang, K. K. Y. Wong, P. T. Lai, and H. W. Choi, “Geometrical shaping of InGaN light-emitting diodes by laser micromachining,” IEEE Photon. Technol. Lett. 21(15), 1078 (2009). [CrossRef] [PubMed]
10. G. Yu, G. Wang, H. Ishikawa, M. Umeno, T. Soga, T. Egawa, J. Watanabe, and T. Jimbo, “Optical properties of wurtzite structure GaN on sapphire around fundamental absorption edge (0.78–4.77 eV) by spectroscopic ellipsometry and the optical transmission method,” Appl. Phys. Lett. 70(24), 3209 (1997). [CrossRef]
11. C. Griffin, E. Gu, H. W. Choi, C. W. Jeon, J. M. Girkin, M. D. Dawson, and G. McConnell, “Beam divergence measurements of InGaN/GaN micro-arrayed light-emitting diodes using confocal microscopy,” Appl. Phys. Lett. 86(4), 041111 (2005). [CrossRef]
12. Y. F. Cheung, K. H. Li, R. S. Y. Hui, and H. W. Choi, “Observation of enhanced visible and infrared emissions in photonic crystal thin-film light-emitting diodes,” Appl. Phys. Lett. 105(7), 071104 (2014). [CrossRef]
13. K. H. Li, C. Feng, and H. W. Choi, “Analysis of Micro-lens Integrated Flip-chip InGaN Light-emitting Diodes by Confocal Microscopy,” Appl. Phys. Lett. 104(5), 051107 (2014). [CrossRef]
14. C. Baum and C. Brecher, “Light guide films with LEDs as OLED alternatives”, Proc. LED Professional Symposium 2014, 208–213, 2014.