Strange LED die surface (Microscope image, XP-G2)

matt304

Newly Enlightened
Joined
Sep 28, 2006
Messages
183
I suppose this might have just been detected as a "high K bin" :tinfoil:, but my eyes saw something strange immediately.

When I looked at my XP-G2 to prepare for a lens removal, I noticed its die surface was extremely "white". I always view the dies under a microscope to check for their consistency and any die errors before considering for a build (since I don't have a light test meter/software). This one lonely Cree XP-G2 was a very strange LED to me!

Only 25% area of the die surface looks like it has the "proper yellow" color. The rest almost looks silvery white. It looks like a circle set on a square, partially overlapping the square corner, with the rest of wafer circle chopped off. There is a notable difference in material thickness at the line where this color changes when rotating under a microscope.

The LED is still putting out quite the light amount it appears it should be, when compared to a cool white XP-G, running up to an amp in testing.

I used the small phone aperture to get a picture through the microscope. This is around 300X (darker yellow substrate is surrounding the bond wire in that corner of the die only, rest is silvery/yellow/white to the eye):



The pattern is much more apparent to the eye through the microscope than through the phone camera I was holding. Die was viewed while rotating, to rule out any shadow casting effect.

[Notice the "holes" can actually be seen that are inside LED chips centers, which follow a pegboard-like pattern all over the die if you look closely. Never really saw them before this very "white" die was looked at to know what they really were.]

Also, a comparison photo of a proper XP-G2 image in case you haven't seen one closely:



So, is this a classic wafer mishap that is common, a very poor chip, or just a high-K bin likely, with this being the reason? Etc... I'm just wondering what this visible pattern means, performance wise.
 

phantom23

Flashlight Enthusiast
Joined
Nov 3, 2007
Messages
2,044
I think it's a wafer/chip issue, it's definitely not normal. The emitter on the second picture is warm white version with different die (orange and without dark spots).
 

matt304

Newly Enlightened
Joined
Sep 28, 2006
Messages
183
Correct me if I'm wrong but, wouldn't all similar LEDs have "dark spots" underneath the top layer? I think you are being able to see inside the LED die material here, due to the fore light amount needed at that zoom.
 

Yoda4561

Flashlight Enthusiast
Joined
Jan 22, 2007
Messages
1,265
Location
Florida, U.S.A.
If you're saying that the die looks silvery/white instead of the light yellow typical of a cool white, I wonder if it might be a thin air bubble over the die surface.
 

matt304

Newly Enlightened
Joined
Sep 28, 2006
Messages
183
Could the lens possibly have begun to lift? The MCB was pulled from a P60 drop-in it was silicone set into. The reflector was not centered, or the LED was not centered to the reflector let's say. When I unscrewed it, it appeared it rub tightly on one side of the protective lens--I did notice this, but my end goal was a lens removal in the beginning anyways for an aspheric build. I want to perform an XP-G lens removal, as I have only removed large chip lenses so far which were more forgiving when they peeled; to work around the die. Is this generally hard to do without damaging the bond wires in such a small surface area emitter?

My normal procedure for big dies: Get a blade edge started under the side opposite of the bond wires, slide it into the silicone at the base, and then pull up on the blade lifting it. Will it easily peel off as long as I peel it up and off like this? Or do those bond wires ever get "caught" by the lens silicone when pulling the lens? Is there any trick to this to aid the procedure? Because now I really want to lift the lens, and take a manual photo comparison of this LED with and without lens at same drive current (450mA), while projected through an aspheric lens onto the wall.
 

Yoda4561

Flashlight Enthusiast
Joined
Jan 22, 2007
Messages
1,265
Location
Florida, U.S.A.
I'm to understand that soaking in high octane gasoline is the hot ticket for de-doming XPGs and XML emitters. Never done it myself, but after few hours/day or so it apparently either dissolves the dome or reduces its bond enough that it comes right off the die with no residue, it also doesn't seem to damage anything else on the LED or PCB it's mounted to.
 

matt304

Newly Enlightened
Joined
Sep 28, 2006
Messages
183
Gasoline will turn the self-vulcanizing rubber tape into two compounds. The tape "core", and the dissolved butyl rubber. Once acetone is added the butyl rubber will form out of solution as the bottom layer. So yes, I know that gas indeed has a property of debonding certain polymers. My question is residual junk left on the die. How could the die then be cleaned of the oils etc in gas? Does it all evaporate? Or would a blade be safest from a die preserving aspect? Decisions, decisions. :)
 

bshanahan14rulz

Flashlight Enthusiast
Joined
Jan 29, 2009
Messages
2,819
Location
Tennessee
Take two LEDs from the same bin, and see which method makes the dimmer LED.

Original problem was most likely caused by pressure on the dome.
 
Top