The Royal Blue LEDs we use have a ~55% EQE (external quantum efficiency). This gives a μmoles per joule a fair amount above 2.0. We lose a few photons during the conversion process and that drops us down to about 2.0.
We have a ~30% efficiency improvement over standard white LEDs because the phosphor is remote. The gain is 100% due to to the recycling cavity and omnidirectional nature of phosphors.
For others who are interested, here's another interesting link on the remote phosphor approach:
http://www.ledsmagazine.com/articles/print/volume-9/issue-7/...
And a nice overview of the efficiencies of white LED's, and where the losses usually are: http://www.sandia.gov/~jytsao/SSL_Trajectories_Proc_IEEE.pdf
The exact figure in the last are already somewhat out of date, but if you plug in the specs for the newer LED's, presumably the underlying physics hasn't changed.
I don't think growing lights are full spec
In theory, you can get a significant gain in efficiency by choosing specific wavelengths. In practice, it's difficult figure out exactly what these wavelengths should be, and to match them to existing high efficiency emitters.
Each of our remote phosphor fixtures are rated to last for 96,000 hours to 90% output. That's over 10 years of 24/7 use.
How has this been tested? And if it is true, why is there a 5 year warranty rather than a 10 year? And what level of decline in output is covered by the 5 year warranty?
Efficiency: >2.0 μmoles per watt
Presuming you mean "per joule" rather than "per watt", this is very close to the claimed efficiency of the underlying Royal Blue LED, and (I think) much higher than any white LED. While I can believe there is some benefit from the lower temperature of the remote phosphor, this would be a remarkably high number for a full spectrum light. How has this been measured?