Coaxing More And More Light Out Of High-brightness LEDs
Inserting microstructures within the epilayers of a nitride LED can boost dispersion and ultimately increase chip efficacy, says Optogan’s Lauri Knuuttila and Pekka Törmä.
The LED chipmaking industry is incredibly competitive. There are many, many companies operating in this sector, and every one of them is working really hard to keep its customers, while welcome new ones currently being served by rival firms.
One way to stand out from the crowd and win business is to offer superior technology. Several years ago, cuttingedge crystal quality with dislocation densities on the order of 1 x 107 cm-2 would have given a company a major head start over its competitors. But many chipmakers can now boast of material quality as good as this, along with high internal efficiencies. So today, arguably the best way for a company to differentiate itself from the competition is to manufacture LEDs delivering unprecedentedly high extraction efficiencies.
At Optogan – a European chipmaker with production facilities for LED assembly in St. Petersburg, Russia, and high-volume chip production in Landshut, Germany – this is our key objective.
Getting light out of the LED chip is not easy, given the huge difference in the refractive indices of air and GaN. However, efficient light extraction is essential for the realisation of LEDs with state-of-the art efficacy. This requires excellent device architectures that get very close to delivering their theoretical level of light extraction.
Numerous different approaches have been used by the industry to get more light out of the chip, and by far the biggest challenge is to build an LED that can successfully combine a handful of these approaches to yield a cuttingedge, cost effective chip.
Our epimaxx LEDs that we have developed for lighting applications have been built with this goal. One of their hallmarks is that they are free from packaging because this simplifies light extraction, thanks to a reduction in the number of interfaces associated with the luminaire. Chip design is focused on maximising radiation from the top surface while minimizing radiation from back and sidewalls.
Several of the approaches that we have considered to boost the light extraction of a basic LED chip are highlighted in Figure 1. These desirable features have to be incorporated into the device during its fabrication, either during chip processing or the epitaxial growth of the nitride film. It is worth noting, however, that chip processing can have its downsides. It can lead to absorption in contact layers and defective areas, stifling light extraction through an increase in light reflected back into the structure. This has an unwanted side effect: additional heating of the LED chip.
Figure 1. A basic InGaN-based LED chip can include many features for improving light extraction: a)bottom mirror, b) scribing area, c) ultra low dislocation density GaN buffer, d) light scattering epitaxial layer, e) chi sidewalls, g) metal contacts, h) internal light scattering p-layer, i) contact material and j) chip coating
Preventing substrate losses
One of the first steps that we took to increase LED light extraction was to introduce a mirror on the reverse face of the emission surface, which reflected light towards the emitting side of the structure (see Figure 1 a). We found that the technology used to create this reflective surface, and also the geometry of the chip, can impact reflection, heat dissipation and conductivity. Unfortunately, the mirror has to be placed a relatively long way from the light-generating region of the device, leading to significant losses during the trip that light has to take twice through the GaN-substrate interface. A far better approach is to introduce a scattering plane at the interface between the substrate and initial GaN layers (Figure 1 c and d). Turning to substrates that are ‘structured’ on a micrometer scale can create such an interface, delivering an additional benefit – lower dislocation densities, which stem from more favourable initial growth phases for GaN.
It is tempting to incorporate a reflective layer within the nitride epitaxial stack, underneath the quantum wells that generate light emission. Adding many pairs of AlGaN with vastly differing compositions to create a distributed Bragg reflector can create such a structure, but it is difficult to form a really good reflector because material quality issues hamper multi-layer AlGaN structures.
We have found that it is more effective to introduce microscale structures within the GaN layer, which also reduce tensions in this low-dislocation-density film. Images acquired with a scanning electron microscope reveal that it is possible to control the size and shape of these features by judicious choice of the growth regime (see Figure 2 a and b). The shape of these voids can be controlled from nearly vertical to fully inclined. Thanks to this versatility, it is possible to produce an optimised dispersion structure with excellent crystal quality through careful selection of growth modes and the thickness and composition of the layers.
Figure 2. Scanning electron microscopy images reveal: the microstructures to reduce tension in GaN layer (a, b); the tailoring of transparent contact materials to increase light extraction (c,d); and the structuring of the sapphire surface for flip-chip technologies (e,f)
Extracting light through the top
Internal reflection at the chip’s top surface, which reducesLED output, can be cut with either antireflective opticalcoatings (see Figure 1 j) or objects that are highlydispersive. According to theory, objects are most effectiveat dispersing light when the ratio of the wavelength of thisradiation inside the material is between one-tenth andtwice its physical dimension. Dispersion efficiency peakswhen this ratio is between one-third and one.
Increasing dispersion by tailoring the chip’s surface is awidely adopted approach for boosting light extraction.There are numerous highly sophisticated, very robustmethods that can be adopted, but traditional photo-resisttechnologies are far from ideal because it is challenging toscale this approach to dimensions comparable to the veryshort emission wavelengths of GaN-based materials.
Quite often mask-less approaches are more suitable, from both a cost and yield perspective. A well known, very efficient method for increasing dispersion involves altering the growth conditions for the last few layers of the epistructures so that they form a rough surface (Figure 1 i). The downside of this approach is that it can compromise electronic and optical performance, and we believe that it is better to insert a crystalline scattering layer inside the p-type GaN layer (Figure 1 h). Take this route and a flat surface can be formed on the top of the chip, simplifying subsequent processing steps.
Gains are also possible by tailoring the transparent contact material by chemical treatments to create scattering objects on the contact surface (see Figure 2 c and d). In addition, it is possible to use a similar technology with chips employing a flip-chip geometry, with light extracted from the sapphire side of the device. In this case, scattering objects are formed on the sapphire surface (see Figure 2 e and f).
Traditional scribing technology for chip separation tends to create visible damage on our substrates and their epilayers near the scribing area. Low damage scribing techniques combined with post scribing chemical treatment is an effective way to solve this problem.
Although the sidewalls of the chip account for a very small proportion of its surface area, they play a pivotal role in determining the LED’s extraction efficiency. That’s because emission from the active region transgresses equally in all directions, and due to the high degree of total internal reflection within the device, a significant portion of this light is guided towards the sidewalls. We have found that extraction efficiency can be improved with various etching methods that either taper the sidewalls and guide the light, or remove sidewall scribing damage and increase light dispersion by roughening the sidewalls. Absorption by metal contacts is another issue, which we address with some very simple approaches. We limit the area of this contact and suppress current injection into the active region under the contacts with current blocking layers (see Figure 1 g).
Forming a good p-type contact to any visible nitride LED is challenging. The p-GaN surface must be predominantly covered with contact material to ensure uniform current spreading and minimised contact resistance. Consequently, if the LED is to exhibit high emission efficiency, either highly reflective or highly transparent p-contact materials are mandatory, depending on chip geometry. This sets challenges for the correct contact material and structuring of the selected material.
Putting it all together
Making LEDs that excel in light extraction requiresoptimisation of the chip architecture on several fronts. Atremendous amount of research related to this is ongoing,and a search of the literature reveals that more than 2000papers were published on this topic last year. Given this highlevel of research activity, it is not surprising that there area multitude of schemes for extracting more light from the LED.The portfolio of light extraction technologies is actually amixed blessing, because not all the schemes are suitablefor LED manufacturing. Selecting those that are mostappropriate is of paramount importance – one trap for theunwary is the approaches that promise incredibly highlevels of light extraction, but are impractical, complex, anddifficult to integrate into the LED.
We are devoting a great deal of time and effort to selecting a handful of technologies for light extraction that can work together to create LEDs with cutting-edge performance. This effort is already paying dividends, with our in-situ epitaxial and ex-situ mask-less approach (see Figure 2) yielding a 187 percent improvement over our previous generation of LED chips.
However, we know that we still have a long way to go on the road to the production of LEDs with incredibly high values of extraction efficiency.
Left: Optogan’s chip assembly within the newly installed LED component and module factory in St. Petersburg
Right: High-brightness chip technology currently ramping up at the new Optogan facility in Landshut, Germany
© 2011 Angel Business Communications. Permission required.
FURTHER READING V.E. Bougrov et al. LED professional Review 18 42 (2010) T.-X. Lee at al. Optics Express 15 6670 (2007)
The Optogan Group manufacturers innovative, competitively priced, high- brightness chips, LED components, LED lamps and LED luminaires. Three Russian scientists and entrepreneurs in Helsinki, Finland founded the company in 2004, and the following year this start-up began developing chip technologies in Dortmund, Germany. Currently the new production facilities for LED assembly in St. Petersburg, Russia, and high volume chip production in Landshut, Germany, are ramping up.
The manufacturing plant in St. Petersburg was opened on 29 November 2010 by Deputy Prime Minister of the Russian Federation, Sergey Ivanov. With an overall investment of 3.35 billion rubles (80 million euros), it is the largest LED component and module factory in both Eastern Europe and the Commonwealth of Independent States. The factory, which will employ up to 800 people, covers 15,000 m2 of floor space, 5,000 m2 of which is taken up by a clean room environment. The first production line has an annual production capacity of 360 million LEDs and further capacity extensions are scheduled.