|
Star Rizin
Senior Poster
106 Posts
|
View Profile
| Reply
|
Imprint Rides Photonic Crystal Wave
Light-emitting diodes (LEDs) may gain efficiencies by using photonic crystals defined by imprint lithography. A range of LED makers are developing photonic-crystal LEDs as a way of improving light extraction and beam shaping. Challenges include that the wafers are not very clean or flat, and tend to have surface nodules several microns in size.
Michael P.C. Watts, Consultant, Impattern Solutions -- Semiconductor International, 4/20/2009 8:27:00 AM
A series of recent announcements suggests that the use of photonic crystals in light-emitting diodes (LEDs) is about to transition from niche applications such as LED backlights to volume high-brightness LEDs.
Unilite announced plans this month to use photonic-crystal technology from its subsidiary Luxaltek for high-brightness lighting.
Also this month, Lumileds developed a thin (700 nm) photonic-crystal LED with a 73% extraction rate.
Early adopter Luminus Devices signed a cross license agreement in February with Nichia Chemical Industries.
Cree Research announced a new research benchmark late last year of 107 lm/W for a photonic-crystal LED.
Luxaltek acquired the intellectual property of Mesophotonics, and ordered $13M of imprint equipment from Sweden’s Obducat last year.
Photonic crystals have been talked about in the LED community for the past 10 years, as a way of improving light extraction and beam shaping in LEDs. The semiconductor in LEDs is a high-refractive-index flat slab of material that acts as a very effective waveguide, trapping most of the light created in the waveguide. A great deal of creative effort has been used to extract light from the LED waveguide such as roughening the surface and adding reflective layers.
A classic waveguide photonic crystal traps light, which is expelled out of the waveguide.
An alternative solution is to use a photonic crystal. A classic application of the use of photonic crystals is as an out coupler for waveguides. The idea is that if the waveguide has an array of subwavelength through holes of the correct dimensions, then no light is transmitted through the waveguide, and all the light is emitted perpendicular to the plane of the waveguide. In photonic crystal “speak,” there is an “optical bandgap” in the waveguide.
Unfortunately, in an LED, the quantum well that creates the photons cannot have through holes, and the holes also interfere with the passage of electrons getting to the quantum well. As a result, companies cannot use a “classic” photonic crystal. Research teams from the Massachusetts Institute of Technology (MIT), Cree, Luminus Devices, Lumileds, Mesophotonics, Nanonex, and many universities have published papers on how to use photonic crystals in a practical LED device. The trend to thinner LEDs will tend to make the photonic crystal in an LED a closer approximation to the classic photonic crystal, and more attractive. Luminus Devices Inc. was the first company to sell LEDs with photonic crystals, aimed at the projection display market.
Is this the time?
LEDs have made it past fluorescents in efficiency, so the next milestone is cost parity. If maintenance costs are high, such as in traffic lights, then LEDs pay off now. One of the biggest cost barriers today is the heat sink. Though LEDs use less power, they must be kept much cooler than conventional lights, hence the bulky heat sinks in LED luminaries.
If photonic crystals can deliver on less power through greater extraction efficiency, or directionality — at a small enough cost adder — then they will eventually be universal. With a wafer value of several hundred dollars, a lithography cost of tens of dollars would be acceptable, and easily achieved by imprint lithography. The question of timing is in the hands of the LED designers, whether the impact of photonic crystals is now the largest element in the pareto of possible LED improvements. Certainly, there is a lot of money and energy being committed by the biggest LED companies.
Who will benefit?
Given that the patents for photonic crystals in LEDs are held by a number of companies, the first answer is probably the lawyers involved in the inevitable rounds of lawsuits. Secondarily, the license holders and their licensees stand to gain. Nichia, Unilite, Cree and Lumileds are some of the heavy hitters in the LED market and they will be big winners if photonic crystals produce leading-edge products in the general high-brightness LED market.
The other winners will be companies in the supply chain for the technology needed to implement photonic crystals on LEDs. These technologies include lithography and etch. By today’s standards, the lithography requirements are modest — typically ë/4 or 200 nm. The challenges are that the wafers are not flat, tend to have surface nodules several microns in size, and are not very clean. The non-flatness of tens of microns is a direct result of the difference in the thermal expansion coefficient of the substrate, such as silicon carbide or sapphire, and the epitaxially grown semiconductors, such as those made of gallium nitride, grown at temperatures >900°C. The two layers act just like a bimetallic strip and form a potato chip-like structure. These thermal stresses prevent the use of larger wafers. The nodules are a by-product of the epitaxy when there is not a perfect lattice match between the substrate and the semiconductor. Finally, the other geometries in an LED are 50 µm contact pads, which typically are made in fabs that are not very clean.
Much of the research into photonic crystals has been done using electron-beam lithography and using early-edition manual imprint systems. The soluble mold film from Transfer Devices Inc. (TDI, Santa Clara, Calif.), which develops molecular transfer lithography, has been used by several research groups, particularly on manual imprint tools. Obducat has taken the first order for an automated production system as part of its deal with Luxaltek, which is also the largest multisystem order for imprint equipment from LED fabricators.
In the Obducat imprint system, a thermal imprint head is used to pattern an intermediate film mold that is then used to imprint the LED wafer. The film keeps dirty wafers away from the first thermal imprint head. (Source: Impattern Solutions)
The Obducat solution has a thermal imprint head used to pattern an intermediate film mold, which is then used to imprint the LED wafer. The advantage of the film mold is that it conforms to the potato-chipped wafer and any nodules, and damage to the film is not a concern because it is used only once. In addition, the film keeps the dirty wafers away from first thermal imprint head. Overlay is not a challenge in LEDs so the stretchable film mold is not a problem. It appears Obducat has used these advantages successfully in getting sales and orders from Epistar, Luminus Devices and Luxtalek.
The other competitors are not standing still. Molecular Imprints Inc. is customizing its disk imprint tool for the LED market, and has published photonic crystal process results. SUSS MicroTec has published examples of photonic crystals for its new flexible mold technology, called Substrate Conformal Imprint Lithography (SCIL). Also, EV Group and SUSS have a significant installed base for their proximity printers in the LED market.
Author
Mike Watts is an independent consultant leading Impattern, which helps companies match their product ideas to the most appropriate patterning technology. He spent 28 years working in all aspects of semiconductor patterning and materials, and was a member of the founding team at Molecular Imprints Inc.
|
|
Obducat in the news....
