A thought…
I’m just combining some information I gathered from being here for a while, I do not know the exact physics behind it (I’m a biologist, in led physics just a BLF-amateur, i.e. DrJones knows better than me what happens to photons).
*The Cree leds used to have a very smooth phosfor layer and showed a lot of tint shift and gained a lot of illuminance upon dedoming
*I have had Luxeon Q leds, Oslon Square leds, Nichia leds that all have rougher phosfor layers. I dedomed or sliced them all and the illuminance never became anything impressive. The tint shift was also less (qualitative observation).
*The latest XP-G2 and XP-G3 have a rough phosfor layer and thusfar the throw after dedoming them seems less than impressive.
The source of improved illuminance upon dedoming (following DrJones’ explanation from back in 2012) is that after passing the phosfor layer, remaining blue photons that are leaving the phosfor layer at high angles are reflected back into the die and have another chance of being absorbed by the phosfor and then emitted at increased wavelength (and thus luxvalue) and at a perhaps lower angle that will escape the die, thus adding to the illuminance and causing a tint shift away from blue.
What if the outer edge of rougher phosfor without silicon dome, that has less surface area parallel to the die surface than a smooth layer of phosfor (this is entirely my own assumption), shows much less internal reflection in the first place, so even without dome the light at high angle can escape anyway? Then less photons are ‘recycled’ (is this word also patented by Wavien? 
), and illuminance and tint shift are less increased. So less throw.