Hi. I'm new here. I discovered these forums a few weeks ago and found some of the discussions regarding future LED developments to be rather interesting. As an electronics engineer I became interested in lighting in general and LEDs in particular recently as a new hobby, and have found the concept of solid state lighting to be fascinating.
I think this whole heat issue problem for future LED developments is already solved with the introduction of the Luxeon III. I believe with the Luxeon III Lumileds did some things to keep the phosphors from degrading at higher power levels, as well as enhancing the die to heat sink thermal resistance slightly(from 20°C/W to 17°/C). As a result, they can now claim a lifetime(30% lumen depreciation) of 50,000 hours at a power level of ~2.6 watts.
The ability to deal with ~2.5W of heat is good enough for the future, and let me explain why I don't think there is any more need to deal with higher heat dissipation levels. Right now LEDs are about 12% efficient, so of the 2.6W power inputted, the package has to deal with removing (1-0.12)*2.6W or ~2.3W of heat. In maybe two years LEDs will be about 35% efficient, meaning 65% of the power inputted will show up as heat rather than the 88% now. Remember that the Luxeon III can deal with 2.3W of heat without any changes in lumen maintainence. This means when LEDs are at a quantum efficiency of 35% the die can be driven at a power level of 2.3W/0.65 or 3.5W. 1.2W of this will leave the die as visible light, the rest will show up as heat. A 35% quantum efficiency is ~75 lm/W, so this hypothetical Luxeon IV using the exact same package will produce 3.5*75 or 262.5 lumens.
Fast forward to the day when quantum efficiency jumps to 80%, or ~170 lm/W. I don't think we'll easily or quickly get much higher than this. The same package can now be driven at a power level of 2.3/(1-0.8) or 11.5 W. No more heat will need to be removed than the 2.3W that the Luxeon III package is known to be capable of because of the higher quantum efficiency. 11.5W at 170 lm/W is 1955 lumens. I believe the stated goal of the solid state lighting initiative is 1000 lumens per package and 150 lm/W by 2012. I think I've demonstrated that we already have the package to do that. I would rather that all future developments should focus on improving quantum efficiency, and to a lesser extent lumen depreciation. I tend to think we'll eventually have LEDs with 30% lumen depreciation at 200,000 or 300,000 hours once the mechanisms for degradation are fully understood.
N.B. I based my lm/W conversions for quantum efficiency on the fact that today's LEDs of 25 lm/W (calculated from the Luxeon III datasheet) are about 12% efficient.