Consensus input from the recent US Department of Energy solid-state-lighting manufacturing workshop was that last year’s aggressive roadmap for 10X cost reduction by 2020 wasn’t aggressive enough - the industry now needs to cut cost per lumen by 8X by 2015 to enable a viable volume LED lighting market that competes with fluorescents. That means current costs of about $18/klm need to drop to $2.20 klm, according to the preliminary figures. DOE will release the final version of its LED roadmap sometime this summer.
Best options to target for reducing these costs going forward are likely systems integration, metrology and test, and phosphors, concluded the gathered industry experts. Help is also coming from ongoing efforts to develop production equipment designed specifically for making LEDs, not ICs, and to develop some consensus on basic requirements for tools and materials so those from different suppliers work together.
DOE research funding has jump-started development of lower-cost projection lithography tools from Ultratech designed specifically to handle warped, transparent sapphire wafers; inspection tools from KLA-Tencor designed to map the micropits that cause LED defects; and pyrometers from Sandia National Labs and Veeco Instruments that can measure the critical temperature directly on the transparent wafer to better control the quality of epitaxial deposition.
Users and suppliers are also making progress towards consensus in SEMI standards committees on such basic issues as common placement of flats and notches for marking 6-inch sapphire wafers, and common cassettes and software and hardware interfaces to enable automation.
Progress on reducing the LED area-per-lumen
Recent improvements in making LEDs mean we’re probably already closer than most people think to a decent LED light bulb at an acceptable $10 price, argues Paul Scheidt, Cree LED Components product marketing manager. Key is the progress in continuing to reduce the active LED area required for the lumens, efficacy and thermal constraints of any particular application. He cites the trend towards multiple die arrays per package that reduce the cost of packaging and optics needed per die, allow balancing for consistent color, and provide a familiar one-die-per-fixture system that’s easy for users to integrate. Die on metal substrates allow users without reflow solder equipment to manually assemble systems by screwing die to heat sinks.
Matching each blue die to the right mix of phosphors can get consistent white light from the full distribution of die. And there’s another 100 lm/W of efficiency already achieved in the lab to be brought into mass production. “The highest lab efficiencies are ~230 lm/W, while commercial products are now running 130 to 140 lm/W,” says Scheidt. “That future improvement will allow use of a whole lot fewer LEDs.”
He notes that regulators’ labeling and testing requirements have done a good job at getting the products off the market that would disappoint consumers by not performing as advertised, and big box stores now have some low-cost bulbs of at least acceptable quality. “The quality is far better than anyone expected even a year ago,” he says. “The consumer market will happen faster than people expect. And cost will matter less once the industry develops and new styles become the driver. Look what happened with the thin LED TV.”
Simpler assembly of fewer parts per system
Another area key area of progress is simpler LED system design, with fewer, more integrated parts for easier assembly. One proposal for simplifying the system is Intematix’s remote phosphor, which combines the phosphor and the optics in one part. “We see more efficiencies coming from more integrated designs, and we think we’ll supply the phosphor and optics part of that, for fewer parts, and better thermal management,” says Intematix VP of development Chuck Edwards.
The commercial availability of remote phosphors allows easy matching of a consistent phosphor to the die, and moving the phosphor away from the die disperses the heat to improve reliability. Using a remote phosphor also simplifies the packaging problem down to just getting the blue light out as effectively as possible, providing more freedom to use simpler chip-on-board (COB) packaging. “Many general lighting solutions will move to COB arrays,” argues Edwards. “And with a dozen die for a 100W-equivalent bulb, averaging out wavelength over multiple die to one consistent average number for each array will be easy.”
First step was phosphors printed on plastic sheets that could also serve as diffusers for downlights. Now the company has introduced phosphors in injection-molded plastic domes, designed to sit on top of an array of blue LEDs on an aluminum heat sink, and to also distribute the light in the desired broadcast pattern, such as 270° for an A-19 bulb.
More brightness perhaps possible from quantum dots
Progress on quantum dots may also help get more lumens out of LEDs, as this real potential application has spurred lots of recent work on scaling manufacture of these nanoparticles, and in developing practical solutions for printing or coating their polymer dispersions for lighting. These nanoscale particles can be tuned to emit a desired color with less lumen loss than phosphors, and can be mixed with fewer issues of differential degradation when used in remote geometry. “To mimic natural daylight, the penalty from phosphors can be 50%-60%,” argues Suresh Sunderrajan, president of NNCrystal.
Acuity Lighting and other suppliers are coating NNCrystal’s quantum dot solution on the secondary optics of their lamps to create the warm white light. Sunderrajan says NNCrystal’s particular technology reduces the typical re-absorption losses by separating the emission and absorption characteristics between the core and the shell, to be able to tune down the absorption to improve efficiency. The company says it has developed a stable, uniform, optically clear dispersion in polymer and a 3D precision coating process. The quantum dot materials are being manufactured in kilograms at its plant in China.
Intelligent lighting increases the value
With the maturing technology, users are also finding semiconductor-based lighting can do more than just provide light. Combine LEDs’ fast response and smooth dimming with some simple sensors and controls, and they can smartly adjust to deliver exactly the right amount of light in the right place at the right time, significantly reducing energy usage, without annoying people as current cruder systems are apt to do. Redwood Systems VP of building solutions Jeremy Steiglitz says users have seen ROIs in under two years from re-lighting projects that replace fluorescents with LEDs and sensors on a low-voltage DC network.
“Lighting is like a Trojan horse — it’s everywhere, in every room, and it’s on a grid,” he notes. “You couldn’t ask for a better place for a building’s eyes and ears.” And users are finding other uses for that network once it’s installed. One surprising application has turned out to be management of conference room space. The occupancy sensors can detect which rooms are in use, and by how many people, so users looking for an open conference room can find one, and facilities managers can track how many meeting rooms of what size the company’s workers need.
These issues and solutions are among those to be discussed at the all-day session on progress on manufacturing technology to bring down cost per lumen at SEMICON West, July 12 in San Francisco, which features updates on technology developments across all major parts of the manufacturing value chain - see http://semiconwest.org/SessionsEvents/ctr_043486.