You need to remove all phosphor around the die to get rid of the yellow corona.
That would turn it into a blue corona and reduce the lumen output.
I think this is how skylumen do it too… he shave dome on sst40 lately and it doesnt turn out green. Output is great too. Here is his sst40 shave dome led.
You can keep a tiny layer, but practically speaking removing most of it will imrpove the beam quality.
Maybe cutting it away at an edge, say 45 degrees, makes for the best beam? Nice work Yokiami!
So Lexel when are you going to start offering the dedomed LEDs for sale?
I would like to see this done to Samsung LH351D as well…
why selling those?
all you need are 2 slightly different washer thickness one for cut and one for sanding
get 1500 and 5000 grit paper for 7$ from Ebay
and a razor blade
Interesting idea. Did you sand leds under water or dry? Did you measure the throw of the xhp 50.2?
Why sell ’em? Because we’re lazy maybe? A bit unsure of ourselves? Prefer your proven results over trying it with our own precious emitters? Some of us drink coffee, you know. Caffeine is not great for steadiness of hands.
the amount of domed emitters which usually get sliced
which I have in my box I can count on my fingers 5 XPL W2-2B
a few XML2
Nichia I would not really make sense to do it on high CRI which have a wonderful tint
I would do it on orders, but shipping time from China for emitters is slow and Mouser does only have good bins of XHP35HDs
Can you tell us what washer thickness did you use?
1mm and 1.2mm for XML
on Flip chip Emitters you could get closer to the phosphor, as they have no bond wires
I have a question. How about slicing the dome and then applying a thin layer of transparent lacquer? Like some modders do with dedomed emitters for the phosphorus protection. That would make the silicone surface smooth even without polishing. My only concern is whether there is any such a kind of lacquer that would not destroy the silicone. Is that even possible?
I wonder if that’s actually a good thing.
I read that manufacturers rough up surfaces to get a wider range of incidence angles and therefore improve extraction efficiency.
Maybe it would be best to polish only up to fairly-rough grit?
extraction efficiency is for the blue diode, to get the photons out of the semiconductor
Most LEDs have a perfectly shiney surface like XML2, XHP and XPL are textured to reduce the tint shift, but photon extraction has nothing to do with dome texture
On my C8F build with XP-L2s:
I shaved them, then sanded using 1500 grit, then when about 1.5mm away from the phosphorus layer (checked under my microscope), I THEN mounted these three shaved XP-L2s onto my Sofirn C8F build’s provided MCPCB. With the three LED on the MCPCB, I finally used Tarnex (silver polish) on my larger knife sharpening stone (fine diamond dusted metal sharpening stone). It took about six passes to “polish” the leds down to about 1/2mm from the phosphorous.
As I got closer to a fully sanded surface, the sanding slowed down and I had to keep rinsing the stone as the buildup from the leds would get between the led’s and stone. Took about 30 minutes total time: and the longer I went at it, the slower my rotations went. AND… the last pass, the Tarnex was on very liberally and the passes were very slow.
For me, the trick was to not go any faster than the leds (on the star, in the hand) would let me. If they started “jumping” up or “jittering” (or squeaking even)… time for a full stone rinse and NEW Tarnex. I found the MOST downward pressure was needed at the LAST pass. But when done, they led faces were like mirrors- very happy results!!!
I have an SST-40 I am pulling and shaving and will do this same process. Perhaps another agent (besides Tarnex) would work, but I found it to be very good for my LED shaving process. FWIW…
Aaah! The fine grit led dome polishing thread! I have 2000 and 5000 grit sandpaper sheets on the slow boat from china now, ordered just a couple days ago. Plenty of time to decide which leds to try them over, we'll see. Maybe some high-CRI LH351B and LH351D. 
Cheers :-)
Next time… (especially for a single emitter), I’ll flow the leds onto the MCPCB FIRST- then shave and proceed. It’s a lot easier to get the razor cut to match the face angle of the MCPCB— AFTER flowing the led. I had to work the C8F MCBCB harder than I would have liked, because ONE of the THREE XP-L2s was at such an angle as to require my taking them ALL down right to the phosphorous layer (I finally got a perfect, clear face on all three— with zero room to grind down further). But I got lucky… and did get three very clear faces— and I’m happy.
When I shave my older XPLs… I’ll have to be extra careful to not grind into the tiny lead wires which those LEDs (and my SST-40) use— as they stand up taller than the phosphor layer (encased into the clear dome, which I will look through on my microscope BEFORE shaving).
As the old carpenter’s saying goes: “Measure twice, cut once.” 