Phosphor conversion of photons in LEDs & photon recycling efficiency

[Genzod]

Re: Phosphor conversion of photons in LEDs

Photon in the German forum tried selfmade stainless steel collars a few years ago. They worked, but not very good. Why? Because stainless steel only reflects around 30% of 450nm light. Low gain wa salso caused by the larger hole on the top, by the fact that he polished them instead of putting on a real coating, and imperfect shape, small collar size, sub-optimal led tint, maybe imperfect focus etc..

I remember reading exactly what you just said here about stainless steel. It's a completely valid point that demonstrates albedo might be high but the important components of the light getting absorbed by the reflector render the albedo by itself misleading.

What is the ratio on blue photons to yellow photons in the white light headed to the wavien collar?

 

[The_Driver]

Re: Phosphor conversion of photons in LEDs

40% of the light hitting the collar is added to the opening. 40% * 75% = 30%

This is the 30% of total light you need to make the added 120% you mentioned, as 30% / 25% =120%

The conversion process is probably not a constant for all LEDs and setups. I wouldn't use this equation. A LED with a reddish tint (less yellow-green) might work better for example than one which already has a very yellow-greenish tint. Who knows...

What is the ratio on blue photons to yellow photons in the white light headed to the wavien collar?

This ratio cannot be easily stated. It depends on the specific LED that is used. Just take a look at the spectrum of any standard cool-white LED (in the datasheet). Chemically de-domed Cree LEDs have a little bit higher-yellow green content. But these are not needed, the Osram Black Flat (it comes from the factory without the dome, no greenish tint) also works nicely.

To do this you would need to have the measured watt output of the LED (not lumens!) and then put it in relation to the absorbance spectrum of YAG:Ce Phosphor. After this you would apply the conversion loss factor for this type of phosphor. Then you would get the watts that come out after conversion.

 

[Genzod]

Re: Phosphor conversion of photons in LEDs

The conversion process is probably not a constant for all LEDs and setups. I wouldn't use this equation. A LED with a reddish tint (less yllow green) might work better for example than one which already has a very yellow-greenish tint. Who knows...



This ratio cannot be easily stated. It depends on the specific LED that is used. Just take a look at the spectrum of any standard cool-white LED (in the datasheet). Chemically de-domed Cree LEDs have a little bit higher-yellow green content. But these are not needed, the Osram Black Flat (it comes form the factory without the dome, no greenish tint) also works nicely.

To do this you would need to have the measured watt output of the LED (not lumens!) and then put it in relation to the absorbance spectrum of YAG:Ce Phosphor. After this you would apply the conversion loss factor for this type of phosphor. Then you would get the watts that come out after conversion.

I'm only analyzing one set-up, the Wavien collar and XP-G2, so to clarify, I'm asking if anyone knows the yellow to blue ratio in the XP-G2. The equation is not being used as a general design tool, only to set up discussion and analysis of the dynamics in one set-up. I'm not trying to assert it is correct or useful. I already mentioned I see one thing that might be wrong with it.

The equation is accurately describing the portioning of light in the set up. It is sending a surviving 40% of the lumens involved in recycling/reflection part out the hole for a 2.2 gain.

The equation does not (yet) accurately describe the specific details of the dynamic between the reflector and the LED. That is what I'm trying to investigate. The (0.40)*(75) component can be expanded to depict the specifics of the dynamics, like average number of reflections between the LED and reflector, blue to yellow conversions and corresponding heat losses, etc. That is the reason I am asking what the blue to yellow photon conversion ratio is in the XP-G2, so I can start to modify the equation to account for losses.

 

[Enderman]

Re: Phosphor conversion of photons in LEDs

The 40% collar efficiency is pretty accurate.
Even between different LEDs most of the light is in the visible range and won't change the output much whether it is warmer or cooler colour temp.

What is more difficult to estimate is how much light is recycled by angle.
The areas of the collar closer to vertical will be reflecting light almost straight down, so more light would exit the collar compared to light being reflected from the side of the LED.
However, the majority of an LED's total output is emitted at 45 degrees.
Having a collar with a 30 degree opening rather than 60 would therefore give higher intensity, however it is impossible to know how much more without testing because of the non-linearity.

Of course this would only apply to custom made collars, since all wavien collars are 60 degrees only.
I guess they found this to be a good compromise between lumen output and intensity increase.

 

[Genzod]

Re: Phosphor conversion of photons in LEDs

What is more difficult to estimate is how much light is recycled by angle.

Indeed. I would need a plot of the incidence verses %reflection for a bare, rough, phosphor/silicone surface for that. While I've seen that for glass, I'm doubtful one exists for a hobbyist's dedomed LED.

 

[The_Driver]

Re: Phosphor conversion of photons in LEDs

Indeed. I would need a plot of the incidence verses %reflection for a bare, rough, phosphor/silicone surface for that. While I've seen that for glass, I'm doubtful one exists for a hobbyist's dedomed LED.

I have not found exactly what you are looking for, but I am getting closer:

Investigation of the Optical Properties of YAG:Ce Phosphor

Measurement and Numerical Studies of Optical Properties of YAG:Ce Phosphor for White Light-Emitting Diode Packaging

EDIT:
Page 130 of this book is interesting in this regard.

 

[Genzod]

Re: Phosphor conversion of photons in LEDs

I have not found exactly what you are looking for, but I am getting closer:

Investigation of the Optical Properties of YAG:Ce Phosphor

Measurement and Numerical Studies of Optical Properties of YAG:Ce Phosphor for White Light-Emitting Diode Packaging

EDIT:
Page 130 of this book is interesting in this regard.

I appreciate your efforts to help me refine my work, Driver.

The first paper I found a few days ago and I haven't yet fully digested it. The second is new. The Mie Theory calculations look interesting, but since that is new, it will take me a while to digest it as well.

You added the edit for the book while I was looking the second paper over. I'll give that a look as well. Thank you.

 

[Genzod]

Re: Phosphor conversion of photons in LEDs

Page 130 of this book is interesting in this regard.

There was no p 130 in the preview, but I went to the English version on Google and found on p 104 figure 3.36 which provides coefficients that I'm interested in verses volume fraction of phosphor in LEDs. I believe there is an optimal fraction for LEDs like the XP-G2. Do you know what that tends to be?

 

[The_Driver]

Re: Phosphor conversion of photons in LEDs

No, and I think Cree would really prefer to keep that a secret . Maybe we will find out at some point though.

 

[Genzod]

Re: Phosphor conversion of photons in LEDs

No, and I think Cree would really prefer to keep that a secret . Maybe we will find out at some point though.

That's interesting. I thought I saw a paper recently that determined the optimal fraction, around 20% if I'm not mistaken. I can't remember where I saw that, but I remember they produced two plots that they made that conclusion from.

 

[The_Driver]

Re: Phosphor conversion of photons in LEDs

BTW: I just noticed something interesting. The Osram Black Flat, which doesn't have a dome, is very smooth on top of the die. It seems to be much more reflective compared to to chemically de-domed LEDs.

 

[Genzod]

Re: Phosphor conversion of photons in LEDs

BTW: I just noticed something interesting. The Osram Black Flat, which doesn't have a dome, is very smooth on top of the die. It seems to be much more reflective compared to to chemically de-domed LEDs.

Yes, and I believe the XP-G2 and XP-G3 surfaces are now intentionally roughened to change the angle of incidence at the surface for internal photons trying to get out. That works to our advantage with a wavien collar.

One might use a silicone sealer in discrete bursts to create a roughened surface. Be interesting to see if it would improve the performance of a collar and Oslon Black Flat.

 

[Genzod]

Re: Phosphor conversion of photons in LEDs

I know that in the original waiven collar, there is a specific reflection efficiency--95 or 99% depending on who you ask (not important for the moment). It looks like this light is getting transmitted through the collar rather than absorbed even though I'd venture a guess the collar is getting very warm under a barrage of nearly 1000 lumens. What is important to me though is, it is lost and does not return to the die.

When the light goes back to the die, it enters the silicone/phosphor matrix through refraction, even to some extent at a horizontal angle because there is still refraction at that angle of approach. The blue photons are said to undergo conversion to yellow photons to some extent and the energy difference between the two wavelengths gets converted to energy (heat) according to the Planck-Einstein relation E=h*c/λ, and the photons bounce around inside the matrix until they get "recycled" and find an angle out. Chance of direction is 25% out and 75% back to the collar.

I'm wondering of what photons that go into the die's matrix, what percentage comes back out?

I realize each phosphor/silicone die has a different concentration ratio, making one exact figure impossible for so many different compositions, but that doesn't preclude a generalized range in the optimal ratio that dies tend to be made at around 10-20%. For example, we may have a coated reflector with unknown efficiency, but if we know it was electrically coated by a manufacturer, we can say it is most likely in the range of 85-95%.

 

[JacksonXI]

normally the phosphorus layer absorbs specific wavelength and then re-emit the light. if the LED chip is poorly design, leakage may occur

 

[Genzod]

Re: Phosphor conversion of photons in LEDs

The Osram Black Flat's luminance has been measured by multiple people and several record braking throwers have been built with it. The highest value ever measured (that I have seen) is 260cd/mm^2 at maximum output (4.5-5A). In the light linked in my sig we got up to around 240-250cd/mm^2.

I'm back after a family related medical emergency. Been a month now.

The one case I've seen of a surface brightness test (on TLF) resulting in 250cd/mm2​ was tested at 4.5amps. Unfortunately the charts presented as images are no longer available. 4.5 amps corresponds to 891 lm on a scale leading to 937 lm max at max 5.5 amps from another source on TLF. If you increase the lumens about 5% to max, the surface brightness would increase from 250 to 263, had it been tested at peak. That's fairly close to the calculated 266cd/mm2​, where you take the Lumens and simply divide by Pi and the emitter surface area (1.122mm2​ in this case).

 

[The_Driver]

Re: Phosphor conversion of photons in LEDs

Sorry to hear that!
I currently prefer Köf3's tests on BLF. He uses a new method for measuring. He measures the actual luminous intensity with a reflector and lens (always the same one) to get more realistic values (of single-die LEDs) by calculating back from that. It turns out that the LE UW Q8WP actually manages the same luminance (if you get a resonably good sample), but with a bigger die and thus more lumens. Also just recently Osram announced new LEDs which are basically updated versions of these models which are more practical (center solder pad is neutral, no missing corner on the die, die not offset etc.).

 

[Genzod]

Re: Phosphor conversion of photons in LEDs

Sorry to hear that!
I currently prefer Köf3's tests on BLF. He uses a new method for measuring. He measures the actual luminous intensity with a reflector and lens (always the same one) to get more realistic values (of single-die LEDs) by calculating back from that. It turns out that the LE UW Q8WP actually manages the same luminance (if you get a resonably good sample), but with a bigger die and thus more lumens. Also just recently Osram announced new LEDs which are basically updated versions of these models which are more practical (center solder pad is neutral, no missing corner on the die, die not offset etc.).

That's great to hear.

Post #19 at BLF is where I get the surface brightness is 250 cd/mm2 ​@4.5A. I get my 891 lm @4.5A, 937 max lumens @ 5.6A and 1.122mm^2 die size figures from Köf3's tests. The π divisor I use to obtain surface brightness from lumens and die area comes from the limit analysis I explained here.

My point isn't which of the many measured surface brightnesses is more correct than the others, but rather that the method of determining surface brightness with that limit is. You can see using the numbers the result isn't far removed from these measurements when they are compared at the same amperage.

 

[The_Driver]

Re: Phosphor conversion of photons in LEDs

Yes, dividing the Lumens/mm^2 by Pi is the the formula for the lambertion emitter which a de-domed LED basically is. It gives a very good ballpark value! I have tried it before and is always close, but never 100% precise.

 

[Genzod]

Re: Phosphor conversion of photons in LEDs

Yes, dividing the Lumens/mm^2 by Pi is the the formula for the lambertion emitter which a de-domed LED basically is. It gives a very good ballpark value! I have tried it before and is always close, but never 100% precise.

Exactly the same problem you have with hobbyist measurements. .

 

[Enderman]

Re: Phosphor conversion of photons in LEDs

Hey The_Driver can you do two simulations of a thick vs thin lens of the same diameter in your 3d ray tracing program?

Genzod claims that the reason the UF-T20 from vinh gets 260kcd is because the equation of lens area * led intensity is incorrect for thicker lenses.
With 250cd/mm^2 and 32mm diameter the result is 200kcd without taking into account transmission losses.

He says that the correct equation is

I(cd)= T*sin2(A)*L*(d0/s0)2

T=transmittance
A=half the internal beam angle or arctan[Di/2/(BFL+x0
L=total raw lumens available in hemisphere.
d0=object distance
X0=EFL2/X
Xi=di-EFL
EFL=BFL+CT/n
S0=die dimension

However this gives incorrect lux values for pretty much everything else I've tried that isn't the L20, and also implies that using a thicker lens will give more lux than a thinner lens at the same diameter.
This doesn't make any sense to me because the brightness of the LED would need to be amplified when looking at the lens from a distance, more than simply reflecting/refracting using a typical reflector or lens.
Also the maximum candela has nothing to do with the size of the LED or the distance that the measurement is taken, so I don't know why the equation he is talking about even uses those values.

Please let me know what you think, thanks.