The high-brightness of certain LED chips — based on AlInGaP and InGaN compound semiconductor materials — has made these sources into a billion dollar market. As soon as they first became available in the late 1990s, these devices promised to turn lighting upside down.
Now this is indeed happening, faster than we were expecting just a few years ago, and in many niche markets where LED power was previously totally insufficient.
California has almost completely replaced the light bulbs in traffic lights with LEDs (European cities like Munich are doing likewise), large corporations now use LEDs instead of fluorescent tubes in their advertising signs throughout the world, and high-end automobiles now show off with LED-powered brake lights.
A couple of years ago, people speculated at conferences about LED headlamps for cars, and now such lamps are being developed for cars that will come on to the market in just a few years time. At recent European lighting trade shows, no major luminaire manufacturer dared to omit LED-based products. In the illumination business, people are starting to call LEDs the third major revolution in lighting, after the incandescent bulb and the fluorescent tube. We believe that, when we look back in perhaps 2015, this will have turned out to be true. However, this revolution is not going to happen all by itself.
Let's consider what will be necessary for LEDs to replace incandescent and fluorescent lamps on a large scale. In recent years manufacturers have made great progress in internal quantum efficiency and thermal engineering. Thus "high flux" LEDs have been created, which many companies have then used to create novel products. Demand has been high, and LED companies have sold whatever standard packages they could produce as long as the flux per LED was high. However, these manufacturers have mostly neglected optical considerations in their package design in the past by considering optics to be an add-on for incorporation later.
LED manufacturers were misled by their instinct telling them that sub-millimetre optics around the chip would not create problems. But, in this case, their instinct is wrong. The quality of the light is just as important as the quantity. Any decrease in quality of the emitted light at the chip level cannot be recovered later. High-quality light means uniformly bright light from a well-defined area, delivered into a well-defined, not-too-large angular range.
This is now making life difficult for the designers of advanced applications, because current optical packages make their systems unnecessarily clumsy and expensive, even if the necessary power is there. Due to their inherent spectral width, LEDs will have a hard time replacing lasers in the bandwidth-driven telecom market. But we believe LEDs will gain market share for low-bandwidth, short-reach datacoms: for cost reasons LEDs are becoming the first choice for communication in automobiles, but either integrated optics for better chip-to-fibre coupling or better free-space interconnects will be required.
Instead of letting current LED packages dictate what is possible in illumination design, the requirements of customers and fundamental physics should dictate optical design of chips and packages.
In fact, the first optical component of any LED-driven lamp is the chip itself. LED chips have high refractive index, trapping the generated light due to total internal reflection. Breaking the chip's rectangular symmetry greatly enhances output. However, the light from current inverse pyramidal shapes is not well matched to optical requirements and therefore not uniformly bright and hard to redirect. What is needed are chip shapes that are designed to increase the quality of the light they emit.
The second optical component is usually some kind of cup. Even new high-brightness LEDs use unnecessarily large cups, creating a light source whose average brightness is easily reduced by a factor of four, needlessly. This makes a complete car headlamp, for example, larger by a factor of four. LED cups must therefore maximise flux and minimise size at the same time.
The solution is to make two improvements: increase the light output from the chip and care just as much about what every photon is doing from the point where it is created. LED companies are just starting to face this challenge, working with illumination engineers to integrate good optics into the complete light path. For really great LED products, we need to see much more of this.