You might want to start your quest for knowledge here: http://hyperphysics.phy-astr.gsu.edu/hbase/electronic/led.html
And eventually your quest will come across Max Planck's quantum physics work which is covered here http://hyperphysics.phy-astr.gsu.edu/hbase/mod2.html#c4
After you start to understand the geometry of the LED and the quantum physics of the photo release mechanism (from the previous suggested links), your quest folds back to understanding the current density and geometry relationships in a semiconductor junction. A search might leads you to articles like http://gautier.moreau.free.fr/articles/asryan96.pdf
which involve so much higher math that it will take the nerdiest of engineering nerds a month or more to figure out if they are applicable to your quest and to more general college level text books like http://ecee.colorado.edu/~bart/book/book/contents.htm
where you might eventually learn enough from (after taking a detour into learning enough math to do the calculations required) to figure it out yourself. I don't recommend either of these approaches as they involve a LOT of complicated aspects of electrical engineering at the masters, doctorate, and post-doc level.
Why don't you try the Scientific Method? Come up with a hypothesis that you can test, run the experiments, evaluate your data, and see if you guesed right or wrong or if your experiment neither proved or disproved your hypothesis. That is what scientists do.
But first, you do need to try and understand the concepts from the first two links as you can't directly observe these aspects (but that is another area, the uncertainty priciple that Werner Heisenberg discovered).