Epistar's chips plug and play

The LED business is changing. This decade began with rocketing mobile phone sales that drove up the shipments of LEDs employed for keypad and screen backlighting, which helped to create a multibillion dollar device market with a compound annual growth rate of 45%, according to Strategies Unlimited. But revenue growth had plummeted to single digits by 2006, due to strong downward pressures on LED prices, and now remains at similar levels.

New markets are now needed to reinvigorate LED sales growth. Automotive, architectural and general lighting all offer exciting opportunities, and the latter promises the greatest potential – it s a market that could be worth $100 billion in 2010, in terms of lamps, lighting fixtures and components, according to Global Industry Analysts. However, taking a significant share of this market will require improvements in LED luminosity, alongside a reduction in the overall costs of these lighting systems.

The general public is not buying LED lighting systems today because they are too expensive. AC-to-DC converting systems are needed in addition to the LEDs, because these unidirectional devices cannot be powered directly from the mains.

AC-to-DC converter systems introduce other problems, such as a reduction in overall lighting efficiency. LED driver-IC companies claim that their product s efficiency is as high as 90%, but this is only the DC-to-DC conversion figure, and overall efficiency is much lower. Converters also diminish the lifetime of the LED lighting system, particularly if they are built with short-lifetime electronic devices. In addition, they negate one of the most attractive features of LED illumination – its compact size.

At times it seems that the LED is destined to be driven by DC power. But there is another way. At Epistar Corporation in Taiwan, we have developed an LED chip that can be plugged directly into the wall socket. Thanks to the elimination of complicated electronic power supplies, our approach promises to slash the costs of an LED lighting system, while making it more compact and reliable. On-chip rectification Although moving from an LED powered by a DC source to one driven by AC mains is a radical leap forwards, the technology required to make this jump is actually rather simple. It is centered on the construction of diode-based rectification circuits, and will be familiar to anyone with a basic grasp of standard circuit layouts.

We have focused our attention on the development of two types of AC-LED: those that employ an anti-parallel circuit for rectification and those that use a bridge circuit (an overview of both circuits is provided in the box "Rectifying circuits for LEDs"). A simpler chip design and layout are realized with the anti-parallel approach, but these benefits have to be weighed against a smaller illuminated area per cycle, compared with a bridge-type circuit. The challenge associated with the latter type is optimization of the ratio of rectifying area to the central ever-illuminated area (figure 1). The III-V diodes that are used to produce these rectifying circuits are very similar to silicon-based equivalents, but we have to modify our epitaxial process for LED production.

AC-LED manufacture can be divided into the production of monolithic chips and packages. Packaged AC-LEDs are an assembly of multiple, individual LEDs, which are united to form a single unit. In comparison, the monolithic approach involves fabrication of an array of cells that are combined within a single chip. This has advantages over the packaged approach: it is far less laborious and it produces smaller chips via a lower-cost process. Devices can also be customized to meet specific demands by tailoring the number of cells and the chip size.

The monolithic AC-LED production process is very similar to that used to make conventional LEDs, but it employs three additional, subsequent steps: isolation, insulation and interconnection. Each of these plays a key role in device operation.

Isolation is needed to electrically separate the cells from one another. The isolation trench has to be deep enough to reach the insulating substrate, but thin enough to prevent a substantial reduction in the size of the emitting area. So a high-aspect-ratio trench is needed because a higher ratio boosts device efficiency.

The next step – insulation – is the most critical. The material selected for this process has to have good insulating properties, and also enable excellent step coverage so that good-quality films can be deposited on the walls of each cell (figure 2). A metallic coating interconnects these cells, with a yield that is governed by the degree of planarization of the insulation layer.

One distinguishing feature of our AC-LEDs is their variation in efficiency with time, due to variations in the driving voltages for each cell. Our devices are predominantly operated in a high current regime and suffer from droop, a decline in efficiency at high drive currents. The cause of droop is not clear in the LED community, but it is attracting intense debate between researchers. Future developments in chip design may be able to address this weakness without significantly complicating the production process and this could lead to an increase in the efficiency of our AC-LEDs.

One downside of discarding the AC-to-DC converter is that it exposes the LED to variations in AC supply, including power surges. We have addressed this issue through the addition of an external resistor or capacitor, which prohibits the burn-out of our LEDs from a sudden power surge by diminishing the effects of voltage fluctuations (figure 3). However, this approach also cuts the overall lighting efficiency and it would be good to discover a better solution to improving the AC-LED s robustness to mains supplies.

The applications that we are targeting, such as solid-state lighting, require our AC-LEDs to operate at high powers. Heat dissipation is critical when LEDs are driven in this regime and our devices would benefit from a high thermal conductivity substrate. Sapphire, the substrate currently used for AC-LED production, has a relatively poor thermal conductivity, but improvements could result from the transfer of the epistructure to a silicon or metal platform.

Even in its current form, our AC-LED promises to play a major role in driving the transition to solid-state lighting because it addresses the need for a fully integrated system. It majors on its compactness and the opportunity that it provides to reduce the cost of lighting systems, through elimination of the AC-to-DC converter. However, the device is still in its infancy and its efficacy lags behind that of the leading conventional LEDs. But this gap will shrink through improvements to the device, such as better designs for the epitaxial structure and the cell, and this should propel the fortunes of our AC-LED in the lucrative lighting market.   

View pdf of article